Embed Size (px)
Citation preview
Assessing the Metastatic Potential of Cancer Cells with Microdevices
by
Brenda June Green
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy
Institute of Biomaterials and Biomedical Engineering
University of Toronto
© Copyright by Brenda Green 2018
ii
Assessing the Metastatic Potential of Cancer Cells with
Microdevices
Brenda June Green
Doctor of Philosophy
Institute of Biomaterials and Biomedical Engineering
University of Toronto
2018
Abstract
Cancer is a leading cause of death worldwide, and tumor heterogeneity presents a challenge for
the clinical management of the disease. The central aim of this thesis is to present new strategies
that can be applied towards cancer diagnostics and shed light on the metastatic process. Significant
progress has been made over the last decade towards the development of approaches that enable
the capture of rare circulating tumor cells (CTCs) from the blood of cancer patients and provide
detailed proteomic, genetic, and phenotypic analysis. During cancer metastasis, cells may
transition from an epithelial to a mesenchymal phenotype. We designed a microfluidic device that
sorts CTCs into subpopulations representing different phases of epithelial to mesenchymal
transition (EMT). This allows us to detect more invasive subpopulations and monitor their
abundance during treatment regimes. We hypothesize that identification of CTC subpopulations
can provide enhanced diagnostic information. CTCs are labeled with magnetic nanoparticles
conjugated to an antibody against a cancer cell surface marker (such as epithelial cell adhesion
molecule; EpCAM). Magnetically tagged CTCs are sorted into four zones of the device based on
their surface level of EpCAM. The microfluidic device was used to profile low- EpCAM CTCs
from metastatic castrate resistant prostate cancer patients over a defined treatment period, and
towards identification of invasive cell subpopulations.
iii
Metastasis begins with the invasion of tumor cells into the extracellular environment and migration
towards the blood stream. Elucidating factors which govern successful tumor cell dissemination
may allow the development of therapies that can halt metastasis. We hypothesized that invasive
cancer cells will selectively engage with pores of specific geometries when presented with porous
micro-structures. We created 3D 20µm- tall pores of varying cross section and aspect ratios to
mimic the extracellular space and to monitor the migration of breast cancer cells. We identified
that breast cancer cells exhibit a pathfinding behavior when navigating through a complex
arrangement of pores. Overall, the designed microdevices provide a means to identify features of
invasive cancer cells using nanoparticle- assisted sorting, and migration analysis through porous
structures. These approaches provide further insight into the metastatic process.
iv
Acknowledgements
I would like to acknowledge and thank my PhD supervisor Dr. Shana Kelley for encouraging
me to think critically and independently. Dr. Kelley provided me with multiple opportunities to
grow, expand my professional network, and she provided constant support towards my
endeavors. This experience is invaluable.
I thank Professor Dimos Poulikakos for his supervision and suggestions for the project
conducted at ETH Zurich. I would like to thank my ETH Zurich supervisor Dr. Aldo Ferrari for
providing excellent guidance during my foreign studies. His enthusiasm and continual feedback
guided me towards achieving success.
I would like to thank my committee member and Master’s supervisor Dr. Jonathan Rocheleau
for his in-depth questions, insight and brainstorming. Dr. Rocheleau provided inspiration and
encouraged me to consider multiple aspects when studying metabolism and microfluidics. I
would like to thank my committee members Dr. Chris Yip, Dr. Yu Sun, Dr. Craig Simmons and
Dr. Carolyn Ren for their suggestions and guidance.
I thank Dr. Anthony Joshua for the collaboration with Princess Margaret Hospital and for
providing me with such an excellent opportunity, and Dr. Marcelo Cypel at Latner Thoracic
Research Laboratories for his suggestions and input during the IVLP study. I acknowledge and
thank Jin Sakamoto and Nuria Prenafeta for their collaboration with the IVLP study.
I thank Dr. Mahmoud Labib for his mentorship and collaboration for multiple projects during
my PhD. His enthusiasm for research, motivation and thoroughness guided me through many
projects. I thank Dr. Reza Mohamadi, for his training and guidance during my PhD. His
suggestions, critical analysis and collaboration was essential for my progression through my
PhD. I thank Leyla Kermanshah for her teamwork and contribution to the ACS applied materials
and interfaces paper, and Mahla Poudineh for her guidance and inspiration. I would like to thank
Bill Duong for his excellent collaboration, thoroughness and confidence in our research projects.
I thank Barbara Alexander for providing endless coordination of the projects in the Kelley group.
I would like to thank all past and current members of the Kelley group, including Mark Pereira,
Peter Aldridge, Carine Nemr, David Philpott, Eric Lei, Fan Xia, Sharif Ahmed, Libing Zhang,
Tanja Sack, Sarah Smith, Surath Gomis, Wendi Zhou, Wenhan Liu, Justin Besant, Jagotamoy
Das, Zongjie Wang, Xiaolong Yang, Laili Mahmoudian, Alexander Zaragoza, Tina Saberi
v
Safaei, Adam Mepham, Sara Mahshid, Sahar Mahshid, Vivian Nguyen, Philip Weeber, Punithan
Thiagalingam and Carmen Tu for their help and support.
I thank Magdalini Panagiotakopolou, who integrated me into Switzerland and the LTNT group
and who worked collaboratively through the entire pathfinding project. I thank Thomas
Vasileiou for providing support and humor throughout my time at ETH. I thank Georgios
Stefopolous for his infectious enthusiasm towards anything bio-engineering related and his
guidance and support, and Francesca Pramotton for her explanations, and for creating such a
positive experience. I thank Costanza Giampietro for her confidence and direction for the
pathfinding project. In addition, I thank ETH LTNT members including Sandra Schneider, Tobi
Lendenmann, Bjorn Johann, Milan Jovic, Christian Holler, Francesco Robotti, Athanasios
Milionis, Gustav Graeber, Julia Gerber, Jovo Vidic and Hadi Eghlidi.
I thank my sisters Maryanne Parkinson and Lilli Green who provided enormous support. I have
extreme thanks towards my brothers, sisters, and their families, and my nephews, Bruce Green,
Jessica Sun, Wallace Sun Green, Chris Green, Sarah Newell, Emmett Parkinson, Keegan
Parkinson and Logan Parkinson.
I have special thanks towards my mother, Donna Fisher, who has always seen my full potential
and during difficult times during my PhD, has supported me with pride and has been there to
remind me of my capabilities. I thank my grandparents Lorna and Wallace John Fisher who
provided me with a strong foundation.
I thank my father, Larry Green, who constantly supported me and helped me through my thesis.
I thank my grandfather Bill Green who provided constant support. I recognize my grandmother
Margaret Green, who was a great companion and was very proud of my choices.
I would like to acknowledge my friends Arti Dubey, Noura Abou-Saleh, Andrea Tirone, Sang-
Mi Suh, Sandra Pallas, Halima Mao, Pamuditha Silva, Aschraf Danun, Katia Baynova and
Andrew Struthers, whose support and confidence in my decisions have made me keep going
through my graduate pursuit.
Finally, I wish to acknowledge the University of Toronto, the Institute of Biomaterials and
Biomedical Engineering, Ontario Graduate Scholarship, and Natural Sciences and Engineering
Research Counsil (NSERC) for their financial support.
vi
Table of Contents
Acknowledgements ................................................................................................................. .….ivi
Table of Contents ........................................................................................................................... vi
List of Tables ............................................................................................................................... viii
List of Equations .......................................................................................................................... viii
List of Figures ................................................................................................................................ ix
List of Abbreviations ................................................................................................................... xiii
1 Introduction ...............................................................................................................................1
1.1 Cancer Metastasis and Early Diagnosis ...............................................................................1
1.2 Circulating Tumor Cells ......................................................................................................3
1.3 Migration Analysis of Cancer Cells .....................................................................................9
1.4 Thesis Overview ................................................................................................................10
2 Nanoparticle- Mediated Capture and Sorting of Circulating Tumor Cells in a
Microfluidic Device .................................................................................................................13
2.1 Introduction ........................................................................................................................14
2.2 Results and Discussion ......................................................................................................15
2.3 Conclusion .........................................................................................................................28
2.4 Methods..............................................................................................................................29
3 Isolation of Phenotypically-Distinct Cancer Cells Using Nanoparticle-Mediated
Sorting ......................................................................................................................................36
3.1 Introduction ........................................................................................................................37
3.2 Results and Discussion ......................................................................................................38
3.3 Conclusion .........................................................................................................................45
3.4 Methods..............................................................................................................................46
4 Metastatic Cancer Cell Pathfinding through Porous Micro-structures ............................54
4.1 Introduction ........................................................................................................................55
4.2 Results and Discussion ......................................................................................................57
4.3 Conclusions ........................................................................................................................67
4.4 Methods..............................................................................................................................68
.iv
vii
5 Analysis of Circulating Tumor Cells from Metastatic Castrate Resistant Prostate
Cancer Patients Receiving Enzalutamide or Abiraterone ..................................................76
5.1 Study Design ......................................................................................................................77
5.2 Introduction ........................................................................................................................79
5.3 Results and Discussion ......................................................................................................82
5.4 Conclusion .........................................................................................................................94
5.5 Methods..............................................................................................................................95
6 Conclusions and Future Outlook ...........................................................................................98
7 References ..............................................................................................................................102
8 Appendix A- Cluster Migration in a Microfluidic Device .................................................124
8.1 Introduction ......................................................................................................................125
8.2 Results and Discussion ....................................................................................................126
8.3 Conclusions ......................................................................................................................138
8.4 Methods............................................................................................................................138
9 Appendix B – Effect of In-vivo Lung Perfusion on Lung Metastases and Circulating
Tumor Cells in Rat Sarcoma and Colorectal Cancer Models...........................................144
9.1 Introduction ......................................................................................................................145
9.2 Results and Discussion ....................................................................................................147
9.3 Conclusions ......................................................................................................................153
9.4 Methods............................................................................................................................153
10 Appendix C – Supporting Information ...............................................................................158
10.1 Supporting Information for Chapter 3 .............................................................................158
10.2 Supporting information for Chapter 4..............................................................................162
10.3 Supporting information for Chapter 5..............................................................................173
10.4 Referred Journal Publications ..........................................................................................178
viii
List of Tables
Table 2.1 Sequence of the nucleic acids utilized in the experimental setup……………….........34
Table 5.1 Study timetable: Assessment and procedures…………..………………………….…78
Table 5.2 Patient demographics……………………….…………..………………………….…83
Table 8.1 Fluid parameters used to determine the pressure drop across the cluster capture site.
…………..………………………………………………………………………………………132
Table 10.2.1 3D micro-structure pore width and heights for various cross sections and aspect
ratios…………..……………………………………………………………………...…………163
Table 10.3.1 Prior drug treatment for mCRPC patients…………..….……………..………....176
List of Equations
Equation (1) The magnetic force acting on a magnetic nanobead ……………….………..........18
Equation (2) The magnetic force acting on a cell ………………................................................18
Equation (3) Stokes’ drag force …………………………………………………..……….........18
Equation (4) Linear velocity relation in the velocity valley device …………………….............20
Equation (5) The pathfinding index (PI) through the pore ………………………………..........72
Equation (6) Capture efficiency of CTCs in velocity valley device ………………....................96
Equation (7) The resistance of a rectangular microchannel ………………...............................132
Equation (8) Capture efficiency of clusters ………………………….……..............................142
ix
List of Figures
Figure 1.1 The metastatic cascade showing escape of cancer cells from the primary tumor,
intravasation, survival in the circulation, extravasation and formation of a secondary tumor..…...1
Figure 1.2 Circulating tumor cell capture from a prostate cancer patient…………..…………….4
Figure 1.3. Immunostaining identification of a sarcoma Ck- positive CTC …...……………..…5
Figure 1.4 In situ phenotypic analysis of CTCs……...…..…………………………………….....9
Figure 2.1 Microfluidic blood sample preparation…………..……………………...……….…..16
Figure 2.2 Microfluidic CTC capture in the velocity valley chip…………..………...…………17
Figure 2.3 Flow pattern of magnetic beads in the velocity valley device….………..………….17
Figure 2.4 Flow simulations of the velocity valley device…………..…………………...….…..19
Figure 2.5 Comsol simulations of trapping structures…………..………………………...….….21
Figure 2.6 Schematic representation of the Ap2D-CTC approach and chip………….………...23
Figure 2.7 Performance of the aptamer-mediated capture and release approach in buffer and lysed
& leukocytes-depleted blood…………..……………………..……………………..…………....25
Figure 2.8. Validation of the Ap2D-CTC approach. …….…………..……………………….…26
Figure 2.9. Flow cytometric analysis of the collagen content of isolated cell sub-populations….27
Figure 2.10 Isolated CTC subpopulations from clinical samples…………..………………....…28
Figure 3.1. Phenotypic profiling of cancer cell subpopulations…………..………………….….39
Figure 3.2 Microfluidic profiling of breast cancer cells…………..……………………………..41
Figure 3.3 Collagen uptake assay…………..………………...…………………………….…....43
Figure 3.4 NAD(P)H response of breast cancer cells…………..…………………………….….45
Figure 4.1. Experimental pore micro-structure design…………..……………………………....58
Figure 4.2. Effect of pore shape and geometry on cell penetration dynamics…………..………60
Figure 4.3. Cell polarization during pore engagement…………..………………………………64
Figure 4.4. Cell navigation in complex porous environments…………..…………………….…66
Figure 5.1 Androgen Receptor signaling pathways…………..……………………………..…..80
x
Figure 5.2 Androgen receptor exon full-length and splice variant domains…………..……….....81
Figure 5.3 Capture and analysis of mCRPC CTCs receiving enzalutamide or abiraterone……..85
Figure 5.4 Cytokeratin CTC profile for enzaluamide and abiraterone treated patients………....87
Figure 5.5 Zone profiling of low- EpCAM CTCs over treatment period…………………….…90
Figure 5.6 EpCAM- capture versus NCadherin- capture of mCRPC CTCs……………….…..92
Figure 5.7 Androgen receptor variant 7 profiling of mCRPC CTCs…………..………………...93
Figure 8.1 Flow cytometry analysis of epithelial, mesenchymal and migration markers in PC3 and
PC3M cells…………..…………………………………………………………………..……...127
Figure 8.2 Fluorescent- collagen uptake in PC3 and PC3M cells…………..………………....128
Figure 8.3 Prostate cancer cluster characterization…………..………………………………...129
Figure 8.4 Schematic of cluster capture.. …………..……………………………………….….130
Figure 8.5 Gradient distribution through micro-channels…………..………………………….131
Figure 8.6 Schematic of the cluster capture device showing the cluster capture site and the
nozzles. …………..…………………………………..………………………………………... 132
Figure 8.7 Single cell migration and quantification in Ibidi chemotaxis devices…………….. 134
Figure 8.8 PC3M cells aligned along collagen fibers. …………..……………………………...135
Figure 8.9 Prostate cancer cluster migration through micro-channels…………..………….....137
Figure 9.1 Characterization of Sarcoma (MCA) cells……….………..…………………….…147
Figure 9.2 Circulating tumor cells captured from the blood of rats with Sarcoma- induced lung
cancer…...…………………………………..…………………………………..………………147
Figure 9.3 Sarcoma CTC zone distribution in the velocity valley microfluidic chip………......148
Figure 9.4 Characterization of RCN-9 cells…………..……………………………………..…149
Figure 9.5 Circulating tumor cells captured from the blood of rats with RCN- induced lung
cancer.…..…………………………………..…………………………………..………………150
Figure 9.6 RCN CTC zone distribution in the velocity valley microfluidic chip…………….....151
Figure 9.7 Immunostaining of rat cancer cells……………………………………………...…..152
Figure 10.1.1. SKBR3 cells grown on FITC type I collagen matrix……………………………158
xi
Figure 10.1.2. Collagen uptake assay of SKBR3 and SKBR3- EMT Cells…………………….158
Figure 10.1.3. Folate receptor protein levels of SKBR3 cells…………..…………………......159
Figure 10.1.4. NAD(P)H metabolic response of breast cancer cells…………………………..159
Figure 10.1.5. Collagen uptake in metastatic prostate cancer CTCs…………………………...160
Figure 10.1.6. Surface marker expression analysis of SKBR3 cells after isolation from the
microfluidic device. …………..…………………………………..………………………….…161
Figure 10.2.1. Scanning electron microscope images of walls in the square array configuration.
……...…………..…………………………………..……………………………………...……162
Figure 10.2.2. Scanning electron microscope images of MCF10CA1a.cl1 cells interacting with
basal pores…………..…………………………………..………………………………………164
Figure 10.2.3 Characterization of MCF10A and MCF10CA1a.cl1 cells………...…...…….….164
Figure 10.2.4 Topographic contact guidance of MCF10A and MCF10CA1a.cl1 cells..…....…165
Figure 10.2.5 Immunofluorescence and flow cytometry quantification of HRas in MCF10A and
MCF10CA1a.cl1 cells. …………..…………………………………..…………………………165
Figure 10.2.6 Flow cytometry analysis of migration markers in MCF10A and MCF10CA1a.cl1
cells…………..…………………………………..…………………………………..…………166
Figure 10.2.7 Effect of pore shape and orientation on cell penetration dynamics………….….167
Figure 10.2.8 Engagement of events of A) MCF10A and B) MCF10CA1a.cl1 cells for various
cell densities along the pore walls…………..…………………………………………………..168
Figure 10.2.9 Polarization of cells during penetration and disengagement of pores with cross
section of 36 µm2 and aspect ratio of 0.1…………..…………………………………………...169
Figure 10.2.10 Polarization of cells during penetration and disengagement for cross section 36
µm2 and aspect ratio 0.3…………..………………………………………………………….…169
Figure 10.2.11 Polarization of cells during penetration and disengagement for cross section 36
µm2 and aspect ratio .…………..……………………………………………………………….170
Figure 10.2.12 Cell polarization of A) MCF10A and B) MCF10CA1a.cl1 cells for various cell
densities in the absence of directional signals. …………..………………………………….….170
Figure 10.2.13 Flow cytometry analysis of Rac1 and RhoA levels in MCF10A and
MCF10CA1a.cl1 cells. …………..…………………………………..…………………………171
Figure 10.2.14 Correlation function and length for the collective migration of MCF10A and
MCF10CA1a.cl1 cells…………..…………………………………..…………………………..172
xii
Figure 10.2.15 Representative immunofluorescence confocal sections along the apical, equatorial,
and basal surfaces of MCF10CA1a.cll cells stained for nucleus (green) and actin (red) on substrate
without (A) and with (B) constrictions…………..……………………………...........................172
Figure 10.3.1 Metastatic castrate resistant prostate cancer patient profiles………………..……173
Figure 10.3.2 Number of metastases for progressive and responsive patients receiving
enzalutamide or abiraterone. …………..………………………………………………….…….174
Figure 10.3.3. PSA waterfall plots for progressive and responsive patients receiving enzalutamide
or abiraterone ………....…………………………………..………………………………….....175
Figure 10.3.4 Healthy donor CTCs captured in the velocity valley device…………......……....176
Figure 10.3.5 NCadherin capture efficiency ………………………………..………….…..….176
Figure 10.3.6 CellSearch counts…………..……………………………………………....…....177
xiii
List of Abbreviations
ADT – Androgen Deprivation Therapy
ALP – Alkaline Phosphatase
AR – Androgen Receptor
ARV7 – Androgen Receptor Variant 7
a.r. – Aspect Ratio (width/height)
AS – Antisense
BSA – Bovine serum albumin
CK – Cytokeratin
CoCl2 - Cobalt chloride
CTC – Circulating tumor cells
CT – Computed tomography
CXLC16 – Chemokine ligand 16
ECM – Extracellular matrix
ECOG – Eastern cooperative oncology group
EGFR – Epidermal growth factor receptor
EMT – Epithelial to mesenchymal transition
EpCAM – Epithelial cell adhesion molecule
FACS – Fluorescence-activated cell sorting
FISH – fluorescence in situ hybridization
GFP – Green fluorescent protein
Hb – Hemoglobin
HIF-1α – hypoxia-inducible factor 1α
IVLP – In vivo lung perfusion
ISET – Isolation by size of epithelial tumor cells
LDH – Lactate Dehydrogenase
LHRH – Luteinizing hormone-releasing hormone
mCRPC – Metastatic castrate resistant prostate cancer
MIC – Metastasis initiating cells
MMP – Matrix metalloproteinases
MNP – Magnetic nanoparticles
xiv
MRI – Magnetic resonance imaging
NAD(P)H – nicotinamide adenine dinucleotide phosphate
OS – Overall Survival
PBS – Phosphate buffered saline
PCa – Prostate Cancer
PCWG3 – Prostate Cancer Working Group 3
PDH – Prolyl hydroxylase enzymes
PDMS – Polydimethylsiloxane
PET – Positron emission tomography
PFS – Progression Free Survival
PSA – Prostate Specific Antigen
qPCR – quantitative polymerase chain reaction
RFP – Red fluorescent protein
TPP – two-photon polymerization
VHL – Von Hippel−Lindau
WBC – White blood cell
2D – 2 dimensional
3D – 3 dimensional
e
1
1 Introduction
1.1 Cancer Metastasis and Early Diagnosis
Cancer remains a leading cause of death worldwide. In 2012, an estimated 14.1 million new cancer
cases and 8.2 million cancer deaths occurred globally.1 Female breast cancer incidence rates are
the highest in Western Europe and the United States, while prostate cancer in men is commonly
diagnosed in North and South America, North/ West and Southern Europe. Lung cancer incidence
rates in both genders are high in North America, Eastern and Northern Europe.
Cancer may begin as a primary tumor, and spread through the process of metastasis to secondary
sites (Figure 1.1.). The metastatic process contributes to 90% of cancer related deaths, and involves
the detachment of cells from the primary tumor, intravasation into nearby blood vessels, survival
in the circulation, extravasation into a secondary environment and formation of distant metastases.2
Figure 1.1 The metastatic cascade showing escape of cancer cells from the primary tumor, intravasation, survival in
the circulation, extravasation and formation of a secondary tumor. Reprinted with permission from 2. Copyright ©
2018 by Elsevier Inc.
2
Circulating tumor cells (CTCs) are implicated in the metastatic cascade, and represent the tumor
cells released from the primary tumor into the bloodstream. CTCs experience stressful conditions
in circulation, due to attack from immune cells, hostile non-adherent conditions and shear stress.
These cells may avoid programmed cell death by fusion with immune cells, such as platelets and
lymphocytes.3 Cancer stem cell properties contribute to the survival of CTCs in circulation and
their resistance to conventional therapy.4 Upon reaching the target organ, cancer cells may remain
dormant or begin to divide and develop into a secondary tumor.2
Early diagnostic methods are essential for prolonging survival of cancer patients, including tissue
biopsies, screenings, magnetic resonance imaging (MRI), functional imaging and biomarker
analysis.5
Prostate cancer diagnostic methods include tissue biopsies, prostate specific antigen (PSA)
screening, MRI and functional imaging.6 Novel molecular biomarkers which classify tumor
aggressiveness are increasingly available. These genetic and proteomic assays can predict
biochemical recurrence and the formation of metastases.5 Radiotracers used with positron emission
tomography (PET) provide early diagnostic information, particular in patients with low PSA levels
and for detection of lymph node metastases.
Breast cancer diagnosis includes molecular imaging and genomic expression profiles.
Mammography screening can lead to 19% overall reduction in breast cancer mortality.7 A
diagnostic challenge for pathologists is the distinction between closely related breast
complications, including atypical ductal hyperplasia, or ductal cancer and lobular cancer. Clinical
treatment decisions are made based on protein and genetic analysis. Gene assays can predict the
risk of distant metastases in early- stage breast cancer. For suspected advanced stage breast cancer,
positron emission tomography/ computed tomography (PET/CT) scans are conducted.
Lung cancer entails two subtypes; non-small cell lung cancer (NSCLC; approximately 85% of all
lung cancers) and small cell lung cancer (SCLC; approximately 15% of all lung cancers).8 Lung
cancer is frequently associated with late diagnosis; therefore, early detection can significantly
improve survival.9 Tissue samples of the lung through bronchoscopy or surgical biopsy can show
morphological features of adenocarcinoma or squamous- cell carcinoma. The category of tumors
can be classified using immunocytochemical, immunohistochemical, or molecular analysis. PET-
CT and MRI remain powerful tools for determining the stage of lung cancer.8
3
Advances in early cancer diagnostics are essential for preventing metastasis and development of
effective treatment regimes. Biomarker analysis involving circulating tumor cells (CTCs) are
highly advantageous as they are non-invasive.
1.2 Circulating Tumor Cells
Circulating tumor cells (CTCs) are cells that are shed from the primary tumor, and released into
the bloodstream through intravasation. In order to survive in the blood stream, cancer cells avoid
lethal signals from reactive connective tissue, and upregulate cell survival and anti-apoptotic
pathways. Following circulation, these rare cells can migrate through capillaries and metastasize
to form a secondary tumor.10 During cancer progression, CTCs initially released into the
bloodstream may possess an epithelial phenotype and express surface proteins such as cytokeratin
and epithelial cell adhesion molecule (EpCAM). As the primary tumor progresses, the CTCs can
lose their epithelial markers and gain mesenchymal markers (N-Cadherin, vimentin, fibronectin).11
CTC are obtained through simply drawing blood, and can provide information relating to the stage
of the cancer without requiring expensive or invasive diagnostic techniques such as tissue biopsy
or medical imaging. Most new therapies for cancer are molecularly- targeted. Therefore,
characterizing CTCs can influence treatment options and patient-care outcome. CTC analysis
consists of two steps: enrichment of rare cells from the blood, followed by the confirmation and
characterization of CTCs in the purified sample (Figure 1.2).12 Enrichment steps are essential, as
blood cells outnumber CTCs by 5 billion: 1. Current microfluidic capture strategies rely on size,
immunoaffinity, immunomagnetic and impedance properties of CTCs for separation from other
cells in the bloodstream.13-17 The most common microfluidic capture approaches use magnetic
nanoparticles (MNPs) that are conjugated to antibodies against epithelial cell adhesion molecule
(EpCAM). The MNP- coated CTCs are isolated from the blood cells in the presence of a magnetic
field.18
The capture and analysis of CTCs is very challenging due to the low levels of these cells in blood.
19, 20 Hence, effective capture requires a high level of specificity for CTCs and the ability to handle
very low cell numbers.
4
Figure 1.2 Circulating tumor cell capture from a prostate cancer patient. CTCs are isolated from whole blood
using different techniques such as size-based separation, density gradient centrifugation, immunoselection or
microfluidics. Target cells are then detected with immunology- based or nucleic acid- based methods. Reprinted with
permission from 21. Copyright © Spandidos Publications 2017.
1.2.1 CTCs and EMT
Epithelial to mesenchymal transition (EMT) is a biological process associated with metastasis.
During EMT, polarized epithelial cells to undergo multiple biochemical alterations to adopt a
mesenchymal cell phenotype, which includes enhanced migratory capacity, invasiveness, and
increased production of extracellular matrix components. Transitioning cells may decrease
expression of epithelial genes such as E-Cadherin, EpCAM, cytokeratin, ZO-1 and entactin and
upregulate expression of mesenchymal genes such as vimentin, N-Cadherin, Twist, Snail,
fibronectin and β-catenin.11, 22, 23
The activation of EMT programs in epithelial cells may correlate with the appearance of stemness.
Partial EMT programs, whereby epithelial characteristics are maintained, can lead to collective
migration of tumor cells.11 Once cells reach a distant site, they may undergo mesenchymal to
epithelial transition (MET) to allow original epithelial phenotype to be regained. The ability of
cells to undergo EMT- MET conversion is known as epithelial plasticity that is associated with
stemness. Cancer stem cells adopt different phenotypes depending on micro-environmental cues.24
1.2.2 CTC Identification
CTCs are characterized using immunofluorescence as cells that display a DAPI-stained nucleus
and co-express EpCAM and cytokeratins (8, 18, 19) while also not expressing the pan-leukocyte
marker CD45 (Figure 1.3).25 Several groups use cytokeratin as the primary means of identifying
5
CTCs through proteomic and genetic analysis; however, they also include a panel of markers
which may be associated with cancer cells (such as E-Cadherin, vimentin, fibronectin 1, androgen-
receptor variant 7, N-Cadherin and SERPINE/PAI1). Given the highly heterogeneous nature of
CTCs, it is important to include multiple markers for accurate identification.
1.2.3 CellSearch
Developed in 1999, CellSearch is an immuno-magnetic enrichment method that relies on targeting
a marker specific to epithelial cells, the epithelial cell adhesion molecule (EpCAM). CellSearch
represents the most widely used technique in the clinical setting and is still the only CTC detection
method with FDA clearance.
The CellSearch approach labels CTCs with magnetic particles coated with anti-EpCAM
antibodies, captures the cells from whole blood, and automates their imaging.26 Numerous clinical
studies of CTC levels have been conducted using the CellSearch system 27-33 and have
demonstrated that monitoring these cells can provide powerful prognostic information for a subset
of cancers.
While CellSearch has allowed more thorough studies of the clinical relevance of CTCs, it has a
number of limitations. Several studies have indicated that this approach has an inherent lack of
sensitivity that reduces its applicability to the analysis of cells with high EpCAM levels.34, 35 An
20µm
Figure 1.3. Immunostaining
identification of a sarcoma Ck- positive
CTC as DAPI+/CK-FITC+/CD45-APC-.
CTC
WBC
6
additional constraint is the inability to access cellular material after cells are enumerated.
Therefore, there is a critical need for the development of improved CTC capture approaches.
1.2.4 Affinity Based Isolation of CTCs
Next-generation affinity capture approaches offer significant advancements in the sensitivity and
specificity of CTC capture and analysis. Antibody-modified microdevices and nanomaterials have
enhanced the capture efficiencies of CTCs from patient samples and allowed detailed molecular-
level characterization of CTCs to be performed. While EpCAM remains the predominant capture
target for affinity-based approaches, antibodies are interchangeable as capture agents, which
broadens the applicability of these systems to non-epithelial tumors or low EpCAM CTCs.36
One of the first microfluidic CTC affinity capture systems to be reported demonstrated the
remarkable gains in performance that could be achieved with a microscale approach.16 This device
featured microposts etched in a silicon substrate that were functionalized with anti-EpCAM. The
microposts were positioned to promote maximal contact with the cells flowing through the device.
After blood processing, immunostaining was used to identify CTCs that were positive for
cytokeratin and negative for CD45. High capture efficiencies were attained with a variety of cell
lines, and detectable levels of CTCs were observed in 115 out of 116 patient samples analyzed.
This system was the first that could process whole blood directly, and it was proposed that the lack
of pre-processing steps was a factor in increasing the sensitivity of CTC detection. Promoting
interactions between CTCs and an antibody-modified surface using microfluidic flow also likely
played a role in the improved performance.
Many other affinity-based microfluidic capture systems followed this groundbreaking work.37, 38
Devices featuring micropatterned surfaces that promote turbulence and high levels of collisions
between CTCs and immobilized antibodies have been engineered,39 as well as integrated systems
with electrical detectors for CTC counting.40 In addition, the use of microfluidic sorting systems
that can separate CTCs labeled with fluorescent antibodies based on partitioning into nanoliter
aliquots allowed the isolation of these cells without any need for detachment from the device.41
Progress in developing devices that permit recovery of CTCs after capture has also been made. In
particular, the MagSweeper, a rod-like device that can collect CTCs from clinical samples, has
provided a solution for isolation of patient CTCs for detailed characterization.42-44 Nanomaterials
7
have been shown to further enhance the sensitivity of CTC capture. A NanoVelcro chip, based on
an array of nanoscale silicon needles functionalized with antibodies against CTC surface markers,
has been shown to provide an optimal environment to promote the adhesion of the CTCs to the
capture substrate.45-47 Other nanomaterials, such as conducting polymer nanodots48 and
electrospun TiO2 nanofibers, 49 have been tested and optimized for CTC capture, and the optimal
nanoscale roughness for efficient cell binding has been estimated.
1.2.5 Capture and Release of CTCs
The first methods developed for CTC analysis were designed to facilitate the identification of these
cells via immunostaining, but the use of destructive characterization was quickly recognized as a
constraint that would limit downstream analysis of the genetics and proteomics of these cells.
Releasing viable cells allows for further analysis such as quantitative PCR, whole genome
sequencing, and xenograft studies50, 51, which are essential for full understanding of cancer
metastasis. This has prompted a search for gentle conditions that could be used to release fragile
CTCs from capture devices. Over the last several years, a variety of systems have permitted
efficient recovery of cancer cells after capture using chemical,52 enzymatic,53, 54 self-assembly,55
mechanosensitive,56 and thermal release57, 58 mechanisms. High levels of cellular viability have
been achieved for cancer cells isolated with low levels of contaminating cells. This is a new
capability that will enhance our understanding of the biological properties of CTCs and their
medical relevance.36
Recovering viable cancer cells after antibody-based capture is a challenge because of the high
affinity to surface antigens. Digestion of cell surface proteins has been pursued as a means to
unlink antibody/antigen complexes, but low recovery efficiencies were obtained.40 Recent work
on alternative methods has included the use of labile metal ion linkers between nanoparticles and
antibodies that can be displaced with EDTA,52 and gelatin-based nanocoatings that can be
denatured upon heating above 30˚C.56 The latter approach can also be used to release single CTCs
with mechanical force. Another thermoresponsive technique relies on the use of immobilized
polymer brushes that internalize the attached antibodies at low temperatures,57, 58 an effect that can
be used to release CTCs upon cooling of a brush-modified substrate. This approach permitted the
isolation of CTCs from patient samples and sequencing of tumor-related mutations.
8
Using aptamers instead of antibodies as capture agents presents an alternative capture approach
conducive to a variety of options for the release of viable cells. Aptamers immobilized within
large DNA networks53 or on silicon nanowires (SiNWs)54 have been used for cell capture, and then
treated with nucleases to allow the cells to be recovered. Aptamer-modified SiNWs achieved a
capture efficiency of 95% and recovery rate of 94% for lung cancer cells.54 Alternatively, a nucleic
acid with a sequence complementary to that of the capture aptamer can be used to trigger cell
release.55 Aptamer-based methods therefore allow cell release using mild conditions that appear
to facilitate the recovery of viable cells.
1.2.6 Identification of CTC Subpopulations and Visualizing Heterogeneity
Tumors are intrinsically heterogeneous, with cells that possess divergent phenotypes according to
exposure to different microenvironments and therapeutics.59, 60 Cells that undergo extravasation
from a tumor into the circulation may continue to evolve different properties as they persist in the
bloodstream, and several studies have elucidated heterogeneous transcriptional levels and surface
expression in CTCs.61
EMT is a source of dynamic heterogeneity in CTCs. The identification of specific subpopulations
of CTCs with pronounced metastatic potential further illustrates that these cells should not be
considered as phenotypically identical to their counterparts within a solid tumor.62 Morphological
heterogeneity can also be indicative of metastatic potential due to changes in pro-metastatic cell
signaling pathways. Recently, very small nuclear CTC counts were shown to be elevated in
prostate cancer patients with visceral metastatic disease.63, 64
Sources of dynamic and static heterogeneity present a challenge for CTC capture and
characterization. Microfabricated devices engineered to disrupt cell-cell interactions have been
used to study cultured cells as they diverge into different phases of EMT.65 These authors
demonstrated that cells in different phases of this transition exhibit differing levels of susceptibility
to chemotherapeutics, yet this approach remains untested on patients CTCs. Fluorescence-
activated cell sorting has been used to sort CTC subpopulations in patient samples.66, 67 However,
it is not effective with all subpopulations and requires large samples of blood that are difficult to
obtain in routine clinical trials or CTC culture. In addition, microfluidic devices have been used
to isolate a bulk fraction of CTCs that could then be characterized on a single cell basis.11 Overall,
there is a critical need for technologies that can identify subpopulations of metastatic cancer cells.
9
1.3 Migration Analysis of Cancer Cells
While the development of integrated devices will facilitate the investigation of known CTC
biomarkers, the complete range of metastasis-initiating factors remains uncharacterized. Thus,
methods that monitor cellular phenotypes are valuable sources of information and are progressing
in parallel to biomarker assays.36,68 Motility is a critical aspect of cellular behavior that is thought
to contribute to the aggressiveness of cancer cells. This behavior appears to be dependent on cell
density and local environment, which makes single cell approaches advantageous.
Recent advances in this area include the development of microfluidic devices that can measure the
migration of a specific mesenchymal phenotype of cells with single cell resolution.68 Using an
array of over 3000 miniaturized chambers, migration patterns and velocities can be monitored for
single cells (Figure 1.4A). Cultured cells treated to induce EMT were shown to have more
aggressive migration phenotypes, and cells that exhibited significant levels of drug resistance
possessed the highest velocities. A 3D version of the microfluidic chip also permitted this behavior
to be studied as a function of cell density (Figure 1.4B).69 The continued advancement of these
techniques along with molecular profiling may help elucidate the factors that enhance the
invasiveness of CTCs.12, 70
10
Figure 1.4 In situ phenotypic analysis of CTCs. A) The M-Chip is used for monitoring mesenchymal mode cell
migration. Cells are plated on a basement membrane located one side of the device and migrate along micro-channels
towards a chemotactic agent (FBS). The migration distance is shown to be dependent on the number of cells per well
(density). B) The MI-Chip represents a 3-dimensional cell migration assay. Cells are placed on top of a collagen gel
inside miniaturized wells. Nutrients are added on top of the collagen layer. Cells move towards the nutrients (FBS)
and are tracked using green fluorescent protein. Images adapted from 68, 69. Copyright © 1999-2018 John Wiley &
Sons, Inc. and Copyright © 2014 American Chemical Society.
1.4 Thesis Overview
The objective of this thesis is to apply microfluidic devices and micro-structured arrays to examine
aspects of cancer metastasis. Microfluidics enables the capture of rare cancer cells from whole
blood and provides phenotypic analysis of the isolated cells. We hypothesized that detection of
circulating tumor cell subpopulations will provide enhanced diagnostics. We designed a
microfluidic device to capture and sort CTCs based on the expression level of a surface marker.
This device was applied towards examining CTCs from metastatic castrate resistant prostate
cancer patients and from rat lung cancer models.
We further extended cancer cell profiling to examine migration of breast cancer cells through
porous micro-structured arrays. We hypothesized that invasive cancer cells will selectively engage
with pores of specific geometries when presented with porous micro-structures. This technique
was applied towards identifying pathfinding capabilities of invasive breast cancer cells.
Together these technologies can be applied towards advancing cancer diagnostics. The remainder
of the thesis will be organized as follows:
1.4.1 Chapter 2 – Nanoparticle- Mediated Capture and Sorting of Circulating Tumor Cells in a Microfluidic Device
CTCs provide a non-invasive liquid biopsy that can enable the early detection of cancer
biomarkers. We designed a microfluidic device (velocity valley device) for high efficiency capture
and sorting of CTCs. This device traps CTCs in different zones based on the expression level of a
surface marker, epithelial cell adhesion molecule (EpCAM). We begin with four zone separation,
and we validated this device with high- and- low EpCAM expressing cell lines. A two-dimensional
approach is designed, which profiles cancer cells into 16 different subpopulations based on
aptamer capture and antisense release. Breast cancer CTCs were sorted in the first zone with
11
EpCAM, released and then sorted in the second zone with HER2. These rare cell capture platforms
enable us to identify invasive cancer populations from patient samples.
1.4.2 Chapter 3 –Isolation of Phenotypically-Distinct Cancer Cells Using Nanoparticle-Mediated Sorting
The live-cell functional analysis of isolated CTC subpopulation provides additional clarification
of invasive cancer cell behavior. We designed downstream functional assays for identification of
subpopulations of CTCs isolated from the zones of the velocity valley device. Metastatic cancer
cells have the ability to digest their surrounding matrix to create paths of migration. During this
process, cancer cells ingest collagen through collagenolysis. Cells which are more metabolically
active have increased tumorigenic capacity. The metabolic readout of the cell is recorded with
NAD(P)H autofluorescence levels. We demonstrate that cancer cells isolated from low- EpCAM
zones uptake increased quantities of fluorescent collagen, and have higher folate- induced
NAD(P)H levels relative to high-EpCAM expressing cells. This phenotypic characterization is
applied towards analyzing patient CTCs.
1.4.3 Chapter 4 – Cancer Cell Migration Platforms
Cancer metastasis involves dissemination from the primary tumor, intravasation into blood vessels,
survival in circulation, extravasation in a secondary site, and formation of distant metastases.
During these processes, cancer cells migrate through dense extracellular matrix (ECM) towards
growth factors or oxygen gradients. We created a micro-structured platform to examine pore-
engagement dynamics of breast cancer cells, and identified their ability to pathfind through
favorable pores. Tall and narrow rectangular openings facilitate cancer migration in complex
architectures involving large deformations of cells.
1.4.4 Chapter 5 – Applications of Microfluidic CTC Sorting Approaches
CTCs from metastatic castrate resistant prostate cancer (mCRPC) patients are monitored over
multiple times points over the course of 148 weeks (37 months) using the velocity valley device.
CTCs are profiled with magnetic nanoparticles conjugated to EpCAM or to NCadherin. We
observe that mCRPC CTCs are reduced during enzalutamide or abiraterone treatment, and exhibit
a shift towards low-EpCAM zones over the course of treatment. Androgen receptor (AR) variants
are associated with poor prognosis in mCRPC patients. CTCs are immunostained with full-length
12
AR and ARV7; and we observe a reduction in the number of AR+ and ARV7+ CTCs over the
treatment period. The relative expression levels of AR and ARV7 did not change during treatment.
This study provides novel insight into mCRPC CTC biomarker analysis over an extended period.
1.4.5 – Appendix A: Cluster Migration in a Microfluidic Device
Migration of clusters of cancer cells may provide relevant diagnostic information. We developed
a microfluidic device to trap cancer cell clusters and monitor their migration through a collagen
matrix towards a chemokine gradient. This method enabled us to identify migration dynamics of
tumorigenic prostate cancer cells through 40-µm wide micro-channels.
1.4.6 – Appendix B: Effect of In-vivo Lung Perfusion on Lung Metastases and Circulating Tumor Cells in Rat Sarcoma and Colorectal Cancer Models
CTCs from rat sarcoma and colorectal cancer models are examined using the velocity valley
device. Treatment with in vivo lung perfusion (IVLP)- administered chemotherapy for lung
metastases caused a significant reduction in CTCs. In addition, in vivo CTC profiles shifted
towards lower-EpCAM phenotype over the course of disease progression.
In conclusion, we present several microdevice approaches for identification of invasive cancer cell
phenotypes and behaviors. These methods can be applied towards the clinical management of
cancer.
13
2 Nanoparticle- Mediated Capture and Sorting of Circulating Tumor Cells in a Microfluidic Device
Identifying heterogeneous subpopulations of cancer cells can significantly enhance diagnostic
capabilities. Here, we present a microfluidic device (velocity valley device) for CTC spatial sorting
and profiling using magnetic nanoparticles conjugated to antibodies against EpCAM. Cells with a
high level of EpCAM are trapped in zone 1 and 2, and cells with low levels of EpCAM are trapped
in zone 3 and 4. This binning approach enables us to identify subpopulations of CTCs that have
varying expression levels of EpCAM on their surface.
The velocity valley device is applied towards separation of cancer cells into 16 different
subpopulations based on aptamer- mediated capture and antisense- triggered release strategies.
This method sorts cells based on levels of two different surface makers, and results in further
flexibility for identification of invasive subpopulations.
This chapter has been submitted as two journal publications:
#1. Reprinted with permission from R. M. Mohamadi, J. D. Besant, A. Mepham, B. Green, L.
Mahmoudian, T. Gibbs, I. Ivanov, A. Malvea, J. Stojcic, A. L. Allan, L. E. Lowes, E. H. Sargent,
R. K. Nam, and S. O. Kelley, "Nanoparticle-mediated binning and profiling of heterogeneous
circulating tumor cell subpopulations," Angew Chem Int Ed Engl, vol. 54, pp. 139-43, Jan 2 2015.
Copyright 2015 John Wiley & Sons Inc.
Link to publication online: https://onlinelibrary.wiley.com/doi/abs/10.1002/anie.201409376
R.M.M. designed microfluidic device, performed experiments, analysis and aided in manuscript
writing. J.D.B. aided in device design, performed experiments and analysis. A.M., B.G., L.M, T.G.
and I.I. aided in device validation and experimental design. A.M., J.S., A.L.A., L.E.L. provided
CTC samples and aided in project coordination. E.H.S., R.K.N. and S.O.K. supervised the study.
#2. Reprinted with permission from M. Labib, B. Green, R. M. Mohamadi, A. Mepham, S. U.
Ahmed, L. Mahmoudian, I. H. Chang, E. H. Sargent, and S. O. Kelley, "Aptamer and Antisense-
Mediated Two-Dimensional Isolation of Specific Cancer Cell Subpopulations," J Am Chem Soc,
vol. 138, pp. 2476-9, Mar 2 2016. Copyright 2016 American Chemical Society.
Link to publication online: https://pubs.acs.org/doi/abs/10.1021/jacs.5b10939
14
M.L. designed experiments, performed analysis and wrote manuscript. B.J.G performed
experiments and analysis and aided in manuscript preparation. R.M.M. aided in project
coordination. A.M., S.U.A., L.M., and I.H.C. performed experiments and analysis. E.H.S. and
S.O.K supervised the study.
2.1 Introduction
Circulating tumor cells (CTCs) are rare tumor cells shed from primary and metastatic tumor sites
into the circulation as viable or apoptotic cells. Their presence in blood correlates with increased
metastatic burden and reduced time to relapse. As a result, their isolation and analysis as liquid
biopsies presents a powerful means to monitor tumors noninvasively.20
A single tumor can contain subclones with numerous phenotypes, hence a given patient’s CTCs
can possess heterogeneous subpopulations with varying relevance to the development of the
metastatic disease.11 Furthermore, CTCs have evolving phenotypes that may lead to additional
complexity. Isolation of CTC subpopulations, particularly metastasis-initiating cells (MICs),
remains challenging due to their low abundance in the circulation. Fluorescence-activated cell
sorting has been used to isolate CTC subpopulations and establish increased metastatic potential
of specific cell types,71 but this method does not possess sufficient sensitivity to be used with the
low numbers of CTCs typically found in patient samples. Therefore, it is critically important to
develop new, high-sensitivity approaches for CTC subpopulation isolation. Proving the successful
collection of CTC subpopulations in a minimally invasive fashion will represent a significant step
toward elucidating their cellular biology, identifying MICs and treatment-resistant clones, and
facilitating downstream molecular and functional analyses.
Several techniques have been used to isolate bulk CTCs,72 including gradient centrifugation,73
dielectrophoresis,17 size-based exclusion,74 mRNA tagging,75 and affinity-based enrichment.16, 40,
42, 70, 76, 77
Here, we describe a microfluidic approach to capture and sort CTCs using either antibodies or
aptamers conjugated to magnetic nanoparticles (MNPs). Microfluidic devices are systems that can
process small (micro liters) amounts of fluids, using channels with dimensions of tends to hundreds
of micrometers.78 The effects that become dominant in microfluidics include laminar flow,
diffusion, fluidic resistant and surface area to volume ratio.79 The laminar flow through
15
microfluidics is a result of low Reynolds number by design (< 2,100) and enables the controlled
application of shear stress and the delivery of multiple laminar streams in the absence of mixing.
This system provides an important new tool for identifying CTC subtypes with high clinical
relevance.
2.2 Results and Discussion
2.2.1 Velocity Valley Device
The velocity valley device is a microfluidic device that captures and sorts CTCs from patient
samples.80 Blood is incubated with magnetic nanoparticles conjugated to EpCAM antibodies. The
magnetic beads bind to cancer cells that express EpCAM on the cell surface. In the device, micro-
structures are used to increase the CTC capture efficiency in the presence of a magnetic field
(Figure 2.1). These X-shaped structures create regions of low flow which allow for localized
capture of cells while maintaining a high overall flow rate. Additionally, as cells move from the
inlet to the outlet of the device, they encounter an increasing number of chambers leading to a
concurrent decrease in overall flow velocity.
The magnetic force applied to a cell is proportional to the number of magnetic nanoparticles bound,
and this is turn is proportional to the number of EpCAM molecules on the cell surface. Therefore,
cells with higher EpCAM expression experience a higher magnetic force.
Cells with high EpCAM expression can thus be captured near the device entrance, where overall
flow velocity is high, whereas cells with lower expression will be caught in later zones, where flow
velocity is diminished (Figure 2.2, Figure 2.3). This allows for the spatial sorting of cells on the
basis of EpCAM expression, which is crucial for CTC characterizing the cells as they progress
through epithelial to mesenchymal transition. The CTC velocity valley device can be adjusted for
different CTC capture agents, including HER2 and NCadherin (NCad) antibodies in order to
optimize the capture efficiency. This system achieves a capture efficiency of more than 90%.
16
Figure 2.1 Microfluidic blood sample preparation. Blood samples are incubated with magnetic nanoparticles
conjugated to EpCAM or NCadherin antibodies. The blood is then introduced into the velocity valley device, and cells
are captured at the apex of X-shaped structures, in regions of low flow. In these low-flow regions, the magnetic force
overcomes the drag force, and CTCs are captured.
Three cell lines are used to validate the device, high EpCAM-expressing prostate cancer cells
VCaP, medium EpCAM- expressing breast cells SKBR3, and low EpCAM- expressing breast cells
MDA-MB-231 (Figure 2.2B).
Inlet Outlet
(A)
17
Figure 2.2 Microfluidic CTC capture in the velocity valley chip. (A) Blood is introduced into the microfluidic
device, and enters zone 1. Cells with high levels of surface marker expression (EpCAM) are captured in zone 1
(VCaP), cells with intermediate EpCAM expression are captured in the middle zone 2 (SKBR3) and cells with lower
levels of EpCAM are captured in zones 3 and 4 (MDA-MB-231). This allows sorting of heterogeneous populations
of CTCs. (B) Distribution of VCaP (red), SKBR3 (green), and MDA-MB-231 (blue) cells in the velocity valley device.
The linear velocity is reduced in a stepwise manner in each zone, depicted below the x-axis. This data was prepared
by R.M. Mohamadi, J. D. Besant, A. Mepham and B.J.Green.
Figure 2.3 Flow pattern of magnetic beads in the velocity valley device. FITC magnetic beads (7.5µm) are
introduced through the device at 600µl/h. Beads are suspended in 70% glycerol at a concentration of 10µl beads/ml.
Beads are introduced into the device in the absence of a magnetic field to illustrate the flow profile.
2.2.2 Rationale for CTC Projects
Circulating tumor cell research in the clinical setting relies heavily on the commercially available
CellSearch technology.81 However, several limitations have been reported with CellSearch,
including an inherent lack of sensitivity due to its primary use for the analysis of cells with high
EpCAM.82, 83 Additional constraints include lack of single cell analysis and the inability to access
cellular material after cells are enumerated. The velocity valley device presents an alternative for
(B) P
erc
en
tag
e o
f
cells
cap
ture
d (
%)
40
80
60
100
20
0 High EpCAM
(1x)
VCaP
SKBR3
MDA-MB-231
High-Med
EpCAM
(0.5x)
Med-Low
EpCAM
(0.25x)
Low
EpCAM
(0.125x)
100µm
I II III IV
18
CTC capture, and overcomes several limitations of the CellSearch system, including higher
capture efficiency, capture of low-EpCAM expressing cells and live-cell single cell analysis
capabilities. Importantly, the device can sort CTCs into different zones, which enables us to study
heterogeneous populations of cancer cells. CTC studies are expected to provide early diagnostic
information, and the ability to monitor disease recurrence. With the rapid development of new
CTC technologies, it may be possible to monitor heterogeneous CTCs at earlier stages of cancer
and gain new insight into the utility of CTC analysis.
2.2.3 Design of Velocity Valley Device
The velocity valley device was designed to capture CTCs, considering the magnetic force and
drag force acting on the cell. The magnetic force acting on the magnetic nanobeads is:
�⃗⃗� 𝑚_𝑏𝑒𝑎𝑑 = 𝑉𝑚Δ𝜒𝑏𝑒𝑎𝑑
𝜇0(�⃗⃗� ∙ ∇)�⃗⃗� (1) 84
Vm [m3] is the bead volume
Δχbead [unitless] is the difference between the magnetic susceptibility of the bead and the
medium
μ0 [H/m] is the permeability of free space (4π×10−7 H/m)
∇B⃗⃗ [T] is the applied magnetic field gradient
B⃗⃗ [T/m] is the applied magnetic field.
The magnetic bead volume was determined by scanning electron micrograph and also by dynamic
light scattering. The magnetic force acting on a cell is provided by multiplying the magnetic force
on an individual bead by the number of beads per cell (Nb):
�⃗⃗� 𝑚 = 𝑁𝑏𝑉𝑚Δ𝜒𝑏𝑒𝑎𝑑
𝜇0(�⃗⃗� ∙ ∇)�⃗⃗� (2)
The value of the VmΔχbead were experimentally determined for MACS magnetic beads as 2.3 to
2.5 x 10-16 mm3.85 When the microbeads are moving through the device, the Stokes’ drag force
(�⃗⃗� 𝑑) is generated against the opposite direction of the moving microbeads. The Stokes’ drag force
is:
�⃗⃗� 𝑑 = −6𝜋𝜂𝑟𝑣 (3)
�⃗⃗� 𝑑 [N] is the drag force
r [m] is the cell radius (5 µm)
19
η [Pa × s] is the dynamic viscosity of the medium (0.001 Pa × s)
v [m
s] is the velocity of the cell.
The average number of beads per cell Nb, is estimated as 4x104, based on the literature values for
the number of EpCAM sites per cell.86, 87 The number of beads per cell depends on the
concentration of the surface antigen, the affinity of the antibody- antigen interaction and the cell
radius.
The applied magnetic field (B⃗⃗ ) for one cell is estimated using Comsol simulations as 8.2 x10-7
T2/m, which results in a magnetic force of 6pN. Cells are captured in the device when the magnetic
force (�⃗⃗� 𝑚) is equal to or greater than the drag force (�⃗⃗� 𝑑). The dependency of the capture efficiency
on linear velocity is shown in Figure 2.4A.
Figure 2.4 Flow simulations of the velocity valley device. A) The dependency of drag force on linear velocity. The
cells are captured in the device when the drag force is less than 6pN, which occurs at velocities under 60µm/s. (B)
Flow profile around PDMS X-shaped structures. (C) Zoomed in image of X-structure showing low-flow regions in
the apex of the X (arrow). (D) Velocity profile along arrow depicted in (C).
0
20
40
60
80
100
120
0 1 2 3 4 5 6 7 8 8 9Lin
ear
velo
city (
µm
/s)
Drag Force Fd (pN)
Linear velocity µm/s
A
C
C
410 (µm)
B
0
20
40
60
80
100
120
140
0 20 40 60 80 100
Distance along arrow (µm)
C
Ve
locity (
µm
/s)
D
20
The linear flow velocity resulting in 6 pN of drag was calculated as 60 µm/s. Cell capture will
occur when the flow velocity in the chip is less than or equal to 60 µm/s.
Experimentally, it is determined that 100% of epithelial prostate cancer cells are captured when
the inlet flow rate is 600 µl/h using anti-EpCAM nanobeads. The cross-sectional area of the inlet
channel is calculated using the width (900µm) x height (100µm). Thus, an inlet flow rate of
600µl/h corresponds to an inlet linear velocity of approx. 1mm/s.
Fluid modeling with Comsol demonstrates that the flow velocity in the channels varies from 100-
900µm/s (Figure 2.4B). The flow rate profile in the apex of the X-shaped structure varies from 0-
100 µm/s (Figure 2.4B-D).
Magnetically- tagged cancer cells are captured in the apex of the X-structure, where the flow rates
are less than 60µm/s, and the magnetic force applied to one cell is greater than the drag force on
that cell. The velocity valley chip is designed where the linear velocity varies along the length of
the chip. In the following geometry, the velocity (V) is dependent on the distance from the origin
(W), assuming a constant flow rate through the device.
Using this approach, it is possible to capture low numbers of CTCs based on varying expression
levels of a target antigen. Cells with high numbers of beads could be captured at higher linear
velocities (near the inlet) and cells with lower numbers of beads would be captured at lower linear
towards the outlet.
Different capture structures were considered before the optimal X-shape was chosen. To model
the efficacy of each design, we compared the strength of the drag and magnetic forces acting on
cells in each design. Using Comsol, we simulated the distribution of linear velocities inside the
Vx =W0
Wx× V0 (4)
The velocity decreases as the channel width (Wx) increases.
The linear velocity in the device decreases stepwise in each
zone of the chip. The chip is designed such that the channel
width expands to twice the initially width (Figure 2.2A), and
as a result, the linear velocity is halved at each zone. The
height is constant.
21
chip for the 3 different trapping structure designs using an average linear velocity of 600 μm/s
which corresponds to a 600 µl/h flow rate (Figure 2.5).
There is a higher probability of cell capture if the magnetic force is much greater than the drag
force which opposes capture (�⃗⃗� 𝑑 << �⃗⃗� 𝑚). Therefore capture structure designs which create greater
regions of low linear velocity, and thus lower drag force, are expected to have higher capture
efficiency. To compare these chip designs we calculated the percentage area of the chip in which
the drag force, �⃗⃗� 𝑑, acting on a cell is much less than maximum magnetic force �⃗⃗� 𝑚. We expressed
this mathematically as �⃗⃗� 𝑑 < �⃗⃗� 𝑚. The percentage area for different trapping structures is plotted in
Figure 2.5B. The large X-structures were most efficient and used for the remainder of the
experiments.
Figure 2.5 Comsol simulations of trapping structures. A)
Simulation of the spatial distribution of linear velocities for
different capture structure designs. B) For the various trapping
structure designs, we simulated the percentage area of the chip in
which the drag force (�⃗⃗� 𝒅) is less than the maximum magnetic
force (�⃗⃗� 𝒎-max). The effect of trapping structure geometry on
capture efficiency. The large-X structures were most efficient and
used for the remainder of the experiments. The optimization was
performed by R.M.M Mohamadi.
A
B
Linear velocity
x103 µm/s
22
2.2.4 Aptamer Mediated Two Dimensional Sorting of CTCs
CTCs are highly hetereogeneous and multi-marker capture may enable more precise diagnostic
abilities. Thus, we leveraged the velocity valley device to present an aptamer- mediated, two-
dimensional approach (Ap2D-CTC) that isolates cells into 16 different subpopulations (Figure
2.6A). The capture-and-release strategy is performed using two different aptamers specific for
EpCAM and HER2, and allows the isolation of discrete subpopulations with differing surface
expression profiles. Furthermore, we show that the isolated subpopulations exhibit varying levels
of invasiveness using a collagen uptake assay.
While most affinity-based methods use antibodies against surface antigens for capture, the use of
aptamers may be advantageous for several reasons. The small size (2–3 nm in diameter) of
aptamers compared to antibodies (12–15 nm in diameter) could allow for more accurate
quantification of the cell surface markers and enhanced resolution in identifying distinct
subpopulations.45 In addition, cells captured using aptamers can be released gently using nucleases
or the aptamer's complementary strand,53, 88, 89 whereas antibody-based capture requires a harsh
proteolytic digestion for release, which can damage the extracellular domains of membrane
antigens and subsequently confound immunocytochemical analysis.90 Several microfluidic
devices were developed for isolation of CTCs using aptamers specific to PTK7,53, 91-93 EGFR,55, 94
PSMA,95 and EpCAM.54,96, 97
A modified version of the velocity valley device is applied for the 2D sorting approach (Figure
2.6A,B). In this design, the cross- section area of the zones are incremented by increasing the
channel height rather than the width.80, 98 As a result, the surface area is minimized to reduce
capture of non-specific white blood cells.
CTCs with high EpCAM levels and subsequently higher magnetic susceptibility to be trapped in
the first zone, whereas cells with a lower expression level of EpCAM become trapped only in later
zones based on the abundance of their surface EpCAM. After binning the subpopulations into four
sequential zones, we release the cells using the antisense DNA strand complementary to the
capturing aptamer. Cells released from the first, second, third, and fourth zone are denoted as E4,
E3, E2, E1, respectively; where E denotes EpCAM and the number represents abundance (Figure
2.6C).
23
Figure 2.6 Schematic representation of the Ap2D-CTC approach and chip. (A) Aptamer-mediated isolation of
CTC subpopulations. Cells are first tagged with magnetic nanoparticles labeled with an aptamer specific to the first
surface marker, and sorted into four subpopulations using the fluidic device. The four subpopulations are then released
using a complementary antisense DNA strand and subsequently tagged with magnetic nanoparticles labeled with an
aptamer specific to the second surface marker. After sorting the captured cells into sixteen subpopulations, cells are
released using the complementary DNA strand to the second aptamer. (B) Design of each four sequential zones that
features four different average linear velocities (1x, 0.5x, 0.25x, and 0.125x) that facilitate the capture of differentially
labeled cells. (C) Schematic of the fluidic capture and subpopulation sorting strategy. Cells are first sorted according
to EpCAM levels (E4 = high EpCAM, E1 = low EpCAM) and then HER2 levels (H4 = high HER2, H1 = low HER2).
24
To initiate the 2nd dimension of separation, we tag the four subpopulations using magnetic
nanoparticles labeled with aptamers specific for HER2. Each subpopulation is binned in four
sequential zones based on the HER2 expression level. Thereafter, sixteen different subpopulations
are released from the respective zones using a DNA strand complementary to the HER2 specific
aptamer. The subpopulations are labeled according to the expression of the two markers; for
instance, E1H1 denotes subpopulations showing a low expression level of both EpCAM and
HER2.
The efficiency of cancer cell release and capture using the aptamer-mediated approach was
investigated and optimized (Figure 2.7). Magnetic nanoparticles functionalized with streptavidin
were conjugated to biotinylated aptamers and the overall capture in the fluidic device was
monitored. Comparable capture efficiencies were obtained with optimized EpCAM, HER2, and
EGFR-targeted aptamers, and the levels of capture achieved with the aptamers were similar to
what was observed with antibody-functionalized magnetic particles. Variations in sequence, linker
chemistry, and length were tested to maximize capture efficiency.
The optimization of antisense-triggered release, included studies of antisense oligonucleotide
concentration, incubation time, and flow rate, and release efficiencies approaching 75% were
achieved under optimized conditions. The release of cells triggered by incubation with an
exonuclease that would digest the aptamers was also tested. Antisense (AS)-triggered release and
exonuclease-mediated release achieved similar rates of release, and we therefore conclude that the
cells that could not be liberated were irreversibly adsorbed to the chip surface.
We then proceeded to show that performance was retained when the assay was performed using
blood samples. Because aptamers are rapidly degraded in whole blood even in the presence of
nuclease inhibitors, we were required to develop modified aptamers. EpCAM1, HER2-1 or EGFR1
aptamers modified at the 3' end with an inverted nucleotide (InT) performed well in lysed blood.
These improved aptamers were tested for capture and release, and yielded performance levels that
approached what was attained in buffered solutions (Figure 2.7A and 2.7B).
25
Figure 2.7 Performance of the aptamer-mediated capture and release approach in buffer and lysed &
leukocytes-depleted blood. (A) Capture efficiency. The device was loaded with either 1:1 mixture of target (SKBR3
or VCaP) cells and nontarget U937 cells (200 cells each) in buffer or 200 target cells spiked in blood. The EpCAM1and
HER2 aptamers and antibodies were tested using SKBR3 cells, whereas the EGFR1 aptamer and antibody were tested
against VCaP cells. (B) Release efficiency. Release of captured cells was carried out using the corresponding antisense
(AS) strand. The post-release cell count was calculated after cells were released, stained, and counted. All aptamers
utilized in the blood experiments were modified with an inverted T at the 3′ terminus. Control experiments were
carried out using anti-EpCAM, anti-HER2, and anti-EGFR antibodies. SKBR3 cells were captured with EpCAM1 or
HER2-1 while VCaP cells were captured and released with EGFR1. P.R.; post-release. (C) Immunostaining approach
used to identify cancer cells. Only CK+/DAPI+/CD45– cells were counted when determining efficiencies. This data
was prepared by M. Labib.
To demonstrate proof-of-concept for aptamer/antisense-mediated sorting of sixteen cancer cell
subpopulations, we used SKBR3 and MDA-MB-361 cells. SKBR3 has significantly higher levels
26
of HER2 compared to MDA-MB-361, as shown using flow cytometry (Figure 2.8A,B). The two-
dimensional sorting profiles of the two cell lines as shown in Figure 2.8C and 2.8D reflect the
lower HER2 expression on MDA-MB-361 cells and support the feasibility of using this approach
to isolate subpopulations based on a dual-marker approach.
Flow cytometric analysis of EpCAM levels for the isolated subpopulations confirmed that the cells
captured at the first zone exhibited the highest EpCAM level, whereas a lower EpCAM expression
was observed among cells collected from the following zones. The viability of the retrieved
SKBR3 cell subpopulations was also tested by culturing the cells in plates coated with collagen.
The number of cells increased significantly after 48 h incubation at 37°C and 5% CO2.
EpCAM HER2
SKBR3 MDA-MB-361
27
Figure 2.8. Validation of the Ap2D-CTC approach. Flow cytometric analysis of (A) EpCAM and (B) HER2 levels
in SKBR3 and MDA-MB-361 cells. (C) Aptamer mediated 2D isolation of sixteen cell subpopulations from SKBR3
and (D) MDA-MB-361 cells. 1000 cells were tagged with magnetic nanoparticles labeled with the EpCAM1 aptamer
and captured in the fluidic device according to their EpCAM expression level. Subsequently, the E4, E3, E2, and E1
subpopulations were captured in 1D-zones 1, 2, 3 and 4, respectively. After releasing the cells using AS-EPCAM1,
the cells were tagged with magnetic nanoparticles labeled with the HER2-1 aptamer. H4, H3, H2, and H1
subpopulations were captured in 2D-zones 1, 2, 3 and 4. The sixteen different subpopulations isolated were removed
from the device for further characterization. This data was prepared by M. Labib and B.J.Green.
To determine whether the isolated subpopulations had detectable differences in phenotype, we
characterized the ability of the cells to ingest fluorescent collagen. This assay is used to measure
invasiveness of cancer cells, since collagen uptake has previously been recorded with the ability
of cells to invade the extracellular matrix.99 As shown in Figure 2.9A, flow cytometric analysis of
the collagen content for the subpopulations shows a marked difference in the behavior of different
subpopulations. Cells that exhibited low EpCAM and HER2 levels had much higher levels of
collagen ingestion relative to cells with high or moderate levels. Fluorescence microscope images
of the cellular content of collagen are provided in Figure 2.9B.
Figure 2.9. Flow cytometric analysis of the collagen content of isolated cell sub-populations. (A) Sixteen cell
subpopulations isolated from the SKBR3 cell line were cultured on 12-well plates previously coated with 1 mL of 100
µg/mL FITC-collagen, in the presence of 1 mL of McCoy's medium containing 10% FBS and 1% penicillin-
streptomycin for 48 h at 37 °C and 5% CO2. Samples were analyzed with flow cytometry and the fluorescent intensity
values were normalized to the unstained control. (B) Fluorescence microscope images of a DAPI+/collagen+/CK+
SKBR3 cell.
SKBR3
Rela
tive
flu
ore
sc
en
t
inte
ns
ity
28
Figure 2.10 Isolated CTC subpopulations from clinical samples. Three blood samples collected from prostate
cancer patients we sorted with anti-EpCAM and anti-EGFR aptamers and separated into sixteen subpopulations. The
shaded regions within the table indicated positive subpopulations. E denotes EpCAM and G denotes EGFR. This data
was prepared by M. Labib.
Collagen uptake is an invasive feature of tumor cells. Our results agree with previous studies
showing that low levels of EpCAM and HER2 may be correlated with invasiveness.100, 101
Finally, we analyzed a set of patient samples, using the developed method to determine if this
approach would be effective in clinical samples and whether patients would exhibit different
subpopulations profiles. As shown in Figure 2.10, we were able to isolate CTC subpopulations
exhibiting varying expression levels of EpCAM and EGFR from the blood of several prostate
cancer patients.
2.3 Conclusion
In summary, we report a new means of analyzing CTCs and describe a solution that allows the
characterization of specific subpopulations found in patient samples. Using the velocity valley
device that bins cells into compartments according to their levels of an epithelial marker, discrete
CTC subpopulations can be spatially sorted. The approach provides a powerful means to study
EMT in patient CTCs, and the sensitivity of the approach exceeds that obtained with the gold
standard CellSearch method. The highly tunable nature of this approach permits the construction
of a multizone chip that uses regions of varying linear velocity to trap cells with different levels of
antigen-targeted nanoparticles. The 2D capture and release approach enables isolation of CTCs
with minimal contamination from the surrounding WBCs, and paves the way towards molecular
and functional analysis of CTCs. This method will allow an improved understanding of cancer
progression, metastasis monitoring, and assessment of resistance to therapy in real-time to improve
the clinical outcome.
29
2.4 Methods
2.4.1 Flow Simulations
Numerical simulations were calculated by using Comsol Multiphysics software.
2.4.2 Cell Culture
All cell lines, media, and cell detachment buffer (0.25%w/v trypsin/0.53mM EDTA) were
purchased from Sigma, US. U937 cells (ATCC catalog number CRL-1593.2) are lymphocytes
derived from histocytic lymphoma. They were cultured in suspension in T-75 flasks. The cells
were culture in RPMI-1640 medium (ATCC catalog number 30-2001), supplemented with 10%
FBS, at 37°C and 5% CO2. Cells were collected for experiments or split for further subculturing
at concentration of approximately 106 cells/ml (recommended concentration by ATCC).
VCaP cells (ATCC catalog number CRL-2876) are a prostate cancer cell line. They have epithelial
morphology and they are adherent cells. The cells were cultured in DMEM medium (ATCC
catalog number 30-2002) supplemented with 10% FBS in T-75 flasks, at 37°C and atmosphere
containing 5% CO2. Cells were collected for experiments or split for further subculturing at
confluence of approximately 105 cells/cm2 (recommended confluence by ATCC).
SKBR3 cells (ATCC catalog number HTB-30) are a breast adenocarcinoma cell line. They have
epithelial morphology and they are adherent cells. The cells were cultured in McCoy’s Medium
Modified (ATCC catalog number 30-2007) supplemented with 10% FBS in T-75 flasks, at 37°C
and 5% CO2. Cells were collected for experiments or split for further subculturing at confluence
of approximately 105 cells/cm2 (recommended confluence by ATCC).
MDA-MB-231 cells (ATCC catalog number HTB-26) are a breast adenocarcinoma cell line. They
have epithelial morphology and they are adherent cells. The cells were cultured in Leibovitz’s L-
15 medium (ATCC catalog number 30-2008) supplemented with 10% FBS in T-75 flasks, at 37°C
and 5% CO2.
MDA-MB-361 were cultured in DMEM medium (ATCC 30-2002). All media were supplemented
with 10% FBS and cells were cultured at 37°C and 5% CO2 in T75 flasks. Cells were harvested
when they reached more than 70-80% confluency. Cell detachment from the culture dishes was
30
performed using 0.25% of Trypsin/EDTA (Sigma, US). In the case of aptamer experiments, cells
were released from culture dishes using 5 mL of a non-enzymatic cell dissociation reagent (1X;
MP Biomedicals, Solon, US) 37°C for 10 min. The non-enzymatic solution was used instead of
trypsin to avoid any damage to the extracellular domains of the membrane antigens. Prior to
introduction into the fluidic device, the cells were filtered using a 40 µm BD falcon cell strainer
(Becton, Dickinson and Company, Franklin Lakes, US).
2.4.3 Velocity Valley Device Fabrication
Masters were fabricated on silicon substrates and were patterned in SU-8 3050 (Microchem, US)
using photolithography. PDMS (Dow Chemical, US) replicas were poured on masters and baked
at 67 °C for 45 minutes. PDMS replicas were attached to no. 1 glass coverslips using a 30 s plasma
treatment and left to bond overnight. After peeling the replica, holes were pierced for tubing
connections. The replica was permanently sealed with a PDMS-coated glass slide. Bonding was
enhanced and made irreversible by oxidizing both the replica and the coverslip in a plasma
discharge for 1 min prior to bonding. Silicone tubing was then added at the inlet and the outlet.
The channel depth was 100 μm. The chip was sandwiched between arrays of N52 NdFeB magnets
(K&J Magnetics, US, 1.5 mm by 8 mm) with alternating polarity during the cell capture step.
Devices were treated with 0.1% Pluronic in PBS for 1 hour to reduce non-specific adsorption. A
syringe pump (Chemyx, US) was used for the duration of the cell capture process.
The dimensions of the velocity valley zones are as follows: zone 1- 3.5 x 5.9 mm, zone 2- 7.3 x
5.9 mm, zone 3- 14.8 x 5.9 mm, zone 4- 29.8 x 5.9 mm.
2.4.4 Apt2D Chip Fabrication
The Apt2D fabrication was conducted as described in the velocity valley device fabrication
protocol. Importantly, the sixteen zones used in the 2nd dimension capture were connected with
external tubes. The cell subpopulations were released gently by pipetting from each zone after
disconnecting each zone from the former and latter zone.
2.4.5 Cell tagging with magnetic nanoparticles labeled antibodies
Cells were tagged with magnetic nanoparticles labelled with anti-EpCAM antibodies (Miltenyi
Biotec, US). These superparamagnetic nanoparticles were composed of iron oxide and dextran and
31
were 50 nm in diameter. Cells were incubated with 10µl of EpCAM MNPs for 30 minutes at room
temperature. 100 cells were introduced into the velocity valley device in 1%BSA in PBS.
2.4.6 Cell tagging with magnetic nanoparticles labeled aptamers.
Briefly, 100 μL of 20 μM of the apatmer solution in Dulbecco's phosphate-buffered saline (DPBS,
Sigma-Aldrich, US) was first denatured for 5 min at 95°C then renatured on ice for 10 min.
Afterward, the aptamer solution was added to the wells of the microtitre plate and incubated with
1 μL of 10 mg mL–1 of streptavidin coated magnetic nanoparticles (100μm, Chemicell, US) for 1
h at room temp. Subsequently, the nanoparticles were deposited using a magnetic stand
(Thermofisher, US) and washed twice with DPBS. Prior to loading into the fluidic device, the
aptamer-labeled magnetic nanoparticles were incubated with the cells either in 1% BSA in DPBS
or in blood for 1 h at room temp. 1000 cells were tagged with magnetic nanoparticles and captured
in the fluidic device.
Control experiments were carried out using 20 μL of magnetic nanobeads labeled anti-EpCAM
antibody (Miltenyi Biotec Inc., US), magnetic nanobeads labeled anti-HER2 antibody (Miltenyi
Biotec Inc., US) and biotin labeled anti-EGFR antibody (Abcam, US).
Post- capture, 200 µL of 200 µM antisense DNA in PBS (preheated at 60°C) were loaded into the
chip and incubated with cells for 30 minutes, in order to release the on-chip captured CTCs.
2.4.7 Velocity Valley Cell Capture in Patient Sample or Spiked Blood
Patient blood samples were collected with consent. All blood samples were analyzed within a few
hours from sample collection. 10μl of anti-EpCAM Nano-Beads (MACS) were added to 1 ml of
blood and immediately withdrawn into the velocity valley chip at flow rate of 600 µl/h using a
syringe pump. Next 200 μl PBS-EDTA at 600µl/h (6 min) was introduced to remove non-target
cells followed by 2 wash steps with PBS EDTA (200 μl, 600µl/h, 6 min). After this step, chips
were immunostained as detailed below. For experiments where cells were spiked into blood from
healthy donors, the same protocol was employed.
2.4.8 Apt2D Cell Capture in Patient Sample or Spiked Blood
Patient blood samples were collected with consent. All blood samples were analyzed within few
hours of sample collection. 1 mL of the blood sample was centrifuged in a Ficoll tube to isolate
32
the mononuclear cells (CTCs and WBCs). Afterward, the cells were incubated with 50 μL of anti-
CD15 (Miltenyi Biotec Inc., US), for 30 min at room temp. Subsequently, the magnetic
nanoparticles were separated using a magnetic stand and the supernatant was collected. Second,
the supernatant was mixed with 75 μM of β-mercaptoethanol and incubated with 100 μL of the
EpCAM1 aptamer- magnetic nanobead solution in DPBS for 1 h at room temp. The mixture was
then loaded into the Ap2D-CTC chip. Importantly, the second dimension capture and release was
carried out using magnetic nanoparticles-labeled EGFR specific aptamer (EGFR1-InT) and AS-
EGFR1, respectively. Post- capture, 200 µL of 200 µM antisense DNA in PBS (preheated at 60°C)
were loaded into the chip and incubated with cells for 30 minutes, in order to release the on-chip
captured CTCs.
2.4.9 Velocity Valley and Apt2D Immunostaining
Captured cells were counted using fluorescence microscopy. Prior to staining, captured cells were
fixed inside the chip using 100 µL of 4% formaldehyde solution (Sigma-Aldrich, US) followed by
100 µL of 0.2% Triton X-100 (Sigma-Aldrich, US) for permeabilization. For staining, we used
100 µL of the following reagents: allophycocyanin-labeled anti-cytokeratin antibody (APC-CK,
Genetex GTX80205, US) and alexafluor 488-labeled anti-CD45 antibody (AF488-CD45,
Invitrogen MHCD4520, US), and the nuclear stain, 4,6-diamidino-2-phenylindole (DAPI Prolong
Gold reagent, Invitrogen, US). Antibodies (1:50 dilution) were prepared in 100 µL PBS containing
1% BSA and 0.1% Tween20. DAPI was prepared in 1 mL of 1% BSA solution in PBS. All chips
were stained for 60 min at a flow rate of 0.1 mL h–1. After staining, the chips were washed with
0.1% Tween20 in PBS and stored at 4°C. Chips were imaged using a fluorescent microscope
(Nikon) with an automated stage controller and cooled CCD camera (Hamamatsu, Japan) and
images were acquired with NIS Elements (Nikon). Cells were counted by overlaying the bright
field, red, blue, and green fluorescent images.
2.4.10 Image Scanning and Analysis
Immunostained cells were imaged using a fluorescent Nikon TiE eclipse microscope with an
automated stage controller and an Andor camera and images were acquired with NIS Elements
(Nikon) using a 10X and 50X objective.
33
2.4.11 Culture of isolated cancer cell subpopulations.
SKBR3 cells populations retrieved from the sixteen zones were cultured in 12-well plates
previously coated with 1 mL of 100 µg/mL fluorescein isothiocyanate labeled collagen (FITC-
collagen, Exalpha, US) over night. After adding the cells to the wells, 1 mL of McCoy's Medium
Modified (ATCC 30-2007), containing 10% FBS and 1% penicillin-streptomycin, was added to
each well and the plates were incubated for 48 h at 37°C and 5% CO2. Afterward, cells were
released using 1 mg mL–1 collagenase enzyme (Sigma-Aldrich, US) for 15 min at 37°C.
Immunostaing was performed as described in Section 2.4.9.
2.4.12 Flow Cytometry Based Analysis
Cells were incubated with primary anti-EpCAM or anti-HER2 antibody and fluorescent secondary
antibody for 30 min at room temperature. After incubation, samples were injected into a BD
FACSCanto flow cytometer and measurements were plotted as histograms of fluorescence
intensity.
Enriched cell subpopulations were fixed with 4% formaldehyde solution and incubated with Alexa
Fluor 647-labeled anti-EpCAM antibody (Biolegend, US) or FITC- labeled anti-HER2 antibody
(Biolegend, US), for 30 min at room temperature. Subsequently, samples were injected into a BD
FACS Canto flow cytometer (BD Biosciences, US) and measurements were plotted as median
intensities for each fluorophore (AF647 and FITC). Fluorescent intensity values were normalized
to an unstained control.
34
Table 2.1 Sequence of the nucleic acids (Integrated DNA Technologies, US), utilized in the experimental setup
Nucleic acid Sequence
T10–EpCAM1 96 Biotin–(T10)–5' TGA AGG TTC GTT GTT TCG GTG GGT GTA GAC TCT TTA GAA GAG ATA
CAG ATT TTG GGA ATG 3'
TEG–EpCAM1 96 Biotin–(TEG)–5' TGA AGG TTC GTT GTT TCG GTG GGT GTA GAC TCT TTA GAA GAG ATA
CAG ATT TTG GGA ATG 3'
T10–EpCAM2 102 Biotin–(T10)–5' AAC AGA GGG ACA AAC GGG GGA AGA TTT GAC GTC GAC GAC A 3'
TEG–EpCAM2 102 Biotin–(TEG)–5' AAC AGA GGG ACA AAC GGG GGA AGA TTT GAC GTC GAC GAC A 3'
T10–EpCAM3 97 Biotin–(T10)–5' CAC TAC AGA GGT TGC GTC TGT CCC ACG TTG TCA TGG GGG GTT GGC
CTG 3'
TEG–EpCAM3 97 Biotin–(TEG)–5' CAC TAC AGA GGT TGC GTC TGT CCC ACG TTG TCA TGG GGG GTT GGC
CTG 3'
Cy5-EpCAM1-TEG Cy5–5' TGA AGG TTC GTT GTT TCG GTG GGT GTA GAC TCT TTA GAA GAG ATA
CAG ATT TTG GGA ATG 3'–(TEG)-Biotin
EpCAM1 Biotin–5' TGA AGG TTC GTT GTT TCG GTG GGT GTA GAC TCT TTA GAA GAG ATA CAG
ATT TTG GGA ATG 3'
TEG–EpCAM1–TEG Biotin–(TEG)–5' TGA AGG TTC GTT GTT TCG GTG GGT GTA GAC TCT TTA GAA GAG ATA
CAG ATT TTG GGA ATG 3'–(TEG)–Biotin
TEG–HER2-1 103 Biotin–(TEG)–5' AAC CGC CCA AAT CCC TAA GAG TCT GCA CTT GTC ATT TTG TAT ATG
TAT TTG GTT TTT GGC TCT CAC AGA CAC ACT ACA CAC GCA CA 3'
TEG–HER2-2 104 Biotin–(TEG)–5' GGG CCG TCG AAC ACG AGC ATG GTG CGT GGA CCT AGG ATG ACC
TGA GTA CTG TCC 3'
TEG–HER2-3 105 Biotin–(TEG)–5' GCA GCG GTG TGG GGG CAG CGG TGT GGG GGC AGC GGT GTG GGG 3'
AS–EpCAM1 5' CAT TCC CAA AAT CTG TAT CTC TTC TAA AGA GTC TAC ACC CAC CGA AAC AAC
CAA CCT TCA 3'
EpCAM1–PO4 Biotin–(TEG) 5' TGA AGG TTC GTT GTT TCG GTG GGT GTA GAC TCT TTA GAA GAG ATA
CAG ATT TTG GGA ATG 3'–(PO4)
EpCAM1–InT Biotin–(TEG) 5' TGA AGG TTC GTT GTT TCG GTG GGT GTA GAC TCT TTA GAA GAG ATA
CAG ATT TTG GGA ATG (InT) 3'
35
HER2-1–InT Biotin–(TEG)–5' AAC CGC CCA AAT CCC TAA GAG TCT GCA CTT GTC ATT TTG TAT ATG
TAT TTG GTT TTT GGC TCT CAC AGA CAC ACT ACA CAC GCA CA (InT) 3'
AS-HER2-1 5' TGT GCG TGT GTA GTG TGT CTG TGA GAG CCA AAA ACC AAA TAC ATA TAC AAA
ATG ACA AGT GCA GAC TCT TAG GGA TTT GGG CGG TT 3'
EGFR1-InT97 Biotin–(TEG)–5' TAC CAG TGC GAT GCT CAG TGC CGT TTC TTC TCT TTC GCT TTT TTT
GCT TTT GAG CAT GCT GAC GCA TTC GGT TGA C (InT) 3'
AS-EGFR1 5' G TCA ACC GAA TGC GTC AGC ATG CTC AAA AGC AAA AAA AGC GAA AGA GAA
GAA ACG GCA CTG AGC ATC GCA CTG GTA 3'
36
3 Isolation of Phenotypically-Distinct Cancer Cells Using Nanoparticle-Mediated Sorting
Isolating subpopulations of heterogeneous cancer cells is an important capability for the
meaningful characterization of circulating tumor cells at different stages of tumor progression and
during the epithelial to mesenchymal transition (EMT). We present the velocity valley device for
phenotypic sorting of subpopulations of cancer cells. Magnetic nanoparticles coated with
antibodies against the epithelial cell adhesion molecule (EpCAM) are used to separate breast
cancer cells in the microfluidic platform. Cells are sorted into different zones based on the levels
of EpCAM expression, which enables the detection of cells that are losing epithelial character and
becoming more mesenchymal. The phenotypic properties of the isolated cells with low and high
EpCAM are then assessed using matrix-coated surfaces for collagen uptake analysis, and an
NAD(P)H assay that assesses metabolic activity. This work demonstrates that nanoparticle-
mediated binning facilitates the isolation of functionally-distinct cell subpopulations and allows
surface marker expression to be associated with invasiveness, including collagen uptake and
metabolic activity.
Reprinted with permission from: B. J. Green1, L. Kermanshah1, M. Labib, S. U. Ahmed, P. N.
Silva, L. Mahmoudian, I. H. Chang, R. M. Mohamadi, J. V. Rocheleau, and S. O. Kelley, "Isolation
of Phenotypically Distinct Cancer Cells Using Nanoparticle-Mediated Sorting," ACS Appl Mater
Interfaces, vol. 9, pp. 20435-20443, Jun 21 2017. Copyright 2017 American Chemical Society.
Link to publication online: https://pubs.acs.org/doi/10.1021/acsami.7b05253
1 Equal contribution to work.
B.J.G and L.K. designed experiments and assays, performed analysis and wrote manuscript.
M.L., S.A., P.N.S., L.M., I.H.C. performed experiments and data analysis. R.M.M aided in data
analysis and project coordination. J.V.R. and S.O.K supervised study.
37
3.1 Introduction
Circulating tumor cells (CTCs) have heterogeneous phenotypes, which may arise from epigenetic
changes, environmental cues or differentiation.106 Aggressive tumors release thousands of cancer
cells into the circulation each day; however, less than 3% of these disseminated cancer cells
metastasize.23,107 Studies have suggested that subpopulations of CTCs have a more aggressive
phenotype and a greater capacity to seed metastatic tumors.108, 109 Cancer cells may acquire
metastatic potential by undergoing epithelial to mesenchymal transition (EMT).11, 22, 23 Metastasis
may introduce metabolic changes in the cell; by increasing the capacity to withstand oxidative
stress caused by hostile environments.110
Molecular diversity among cancer cells has been recognized as a major driving force for the
evolution of the disease, and can occur outside of the primary tumor.111 Cytometry techniques
such as PCR-activated cell sorting, flow cytometry and deformability assays can be applied to
identify cell subtypes; however, they do not provide a measure of invasiveness.112-114
Currently, the characterization of CTCs can involve immunostaining, flow cytometry, quantitative
PCR (qPCR), fluorescence in situ hybridization (FISH), whole genome amplification, RNA-
sequencing or xenograft studies.50, 90, 111, 115-118 Live-cell functional assays are a relatively
unexplored area for CTCs, but could advance our understanding of cellular phenotypes correlated
with invasiveness. Existing functional assays that have been applied to CTCs include detection of
specific proteins secreted during the in vitro culture of CTCs, collagen adhesion assays to detect
invasive cancer cells, and in vivo transplantation of patient-derived CTCs into immunodeficient
mice.50, 99, 119-121 These approaches are limited by the low yield of CTCs from patients, but have
the ability to detect metastases-initiating cells.
Microfluidic cell sorting technologies have the potential to enhance characterization of
heterogeneous cell populations, and examples include aptamer-mediated separation 122, basement-
membrane coated chips 123, lateral displacement microarrays 124, and single-cell chemotaxis
chips.125 These methods enable the separation of cells based on their surface marker expression
level, adhesion capacity, transportability and migration potential, respectively.
38
A high- aspect ratio version of the velocity valley device is used, along with magnetic
nanoparticles, to not only capture CTCs, but also divide them into subpopulations according to
levels of protein surface expression.80, 98
This technology has been used to detect heterogeneous populations of cancer cells from cancer
patients 80, 126, and to profile CTCs from tumor-bearing animal models.127
We show that this technology is capable of isolating subpopulations of cells that exhibit different
biochemical and functional phenotypes. Subpopulations of cells are analysed with functional
assays to monitor collagen uptake and NAD(P)H metabolism (Figure 3.1). SKBR3 breast cancer
cells are sorted based on EpCAM expression levels, released from the zones of the microfluidic
device, and then subjected to functional assays. We demonstrate phenotypic differences in an EMT
model, and show that low-EpCAM expressing cells have properties that correlate with invasive
cell behaviour. Altogether, separating subpopulations of cells on the basis of surface marker
expression levels yields groups of cells with distinct functional phenotypes.
3.2 Results and Discussion
Invasive cancer cells are thought to be a rare subpopulation of the bulk group of circulating tumor
cells, but it remains unknown how best to identify them.107, 128 Evidence supports that cancer cell
plasticity is dependent on both epithelial and mesenchymal properties, and can contribute to the
invasion/metastatic cascade.129 The separation of cancer cells into distinct zones of a microfluidic
device using nanoparticles enhances the ability to identify cells with varying levels of EMT
markers.
3.2.1 A Hypoxia-Driven Model of EMT
In order to generate cells that could be analyzed to visualize varied phenotypic properties, a
SKBR3 cell model of EMT (SKBR3-EMT) was created in order to explore the phenotypic
differences of cells undergoing EMT in the microfluidic device. The model was created based on
a previously-described method that relies on the chemical induction of hypoxia in cell culture.130-
133 During tumor progression, cancer cells grow rapidly in an avascular environment; therefore,
oxygen becomes scarce in the inner layers of the cells. A substantial body of evidence indicates
that the hypoxic tumor microenvironment plays a pivotal role in the induction of EMT and
consequently the emergence of CTCs.130, 134-136
39
Figure 3.1. Phenotypic profiling of cancer cell subpopulations. (A) Schematic showing the separation of cancer
cells into 4 zones of a microfluidic device that processes a sample in the presence of an external magnetic field. Cells
are incubated with magnetic nanoparticles (MNPs) labelled with EpCAM. Cells that have high levels of EpCAM and
subsequently high number of MNPs are captured in zone 1 and 2, whereas cells with low levels of EpCAM, and low
number of MNPs, are captured in zone 3 and 4. The linear velocity in the device decreases in a stepwise manner in
each zone, to increase the probability of cell capture in the apex of the X-structures. Viable cells can be released from
each zone. (B) Phenotypic analysis of isolated tumor cells. Viable cells are assessed using a fluorescent collagen
uptake assay and a metabolic NAD(P)H assay. Low-EpCAM cells have increased collagen uptake, and increased
NAD(P)H response relative to high-EpCAM cells. Scale bars are 5µm.
Cancer cells adapt to hypoxic conditions by regulation of a transcription factor called hypoxia-
inducible factor 1-alpha (HIF-1α). In the presence of oxygen, HIF-1α is constantly synthesized
and rapidly degraded through a multistep process catalyzed by prolyl hydroxylase enzymes
(PHDs) and the Von Hippel–Lindau (VHL) tumor suppressor proteins, while in the absence of
oxygen, HIF-1α accumulates in the cell. The accumulation of HIF-1α results in the regulation and
transcription of genes involved in EMT and metastases, including HIF-1, VEGF, vimentin, MMP2,
MMP9, µPAR, PAI-1, c-Met, TWIST and CCR7.130, 134-136
40
SKBR3 cells were chosen to model EMT, as they are a non-aggressive cell line with high levels
of EpCAM, enabling us to monitor changes in epithelial status. Several methods have been
reported for induction of hypoxia in cell cultures, including incubation with cobalt chloride
(CoCl2).130-133 Cobalt chloride mimics the hypoxic microenvironment of tumor cells by
interfering with the degradation of HIF-1α by inhibition of VHL and therefore stabilizing HIF-1α.
The successful induction of EMT was confirmed by monitoring mRNA and protein levels using
qPCR and flow cytometry, respectively. Gene expression data demonstrated that epithelial genes
(EpCAM, cytokeratin 7, cytokeratin 8) were downregulated and mesenchymal genes (Snail1, Slug
and Vimentin) were upregulated after 24, 48 and 72 hours of treatment with CoCl2. On the protein
level, epithelial markers (EpCAM, E-Cadherin, and cytokeratin) were downregulated while the
mesenchymal marker N-Cadherin was upregulated after 72 hour treatment with CoCl2. These
results confirm the induction of EMT in SKBR3 cells (SKBR3-EMT).
3.2.2 Nanoparticle-Mediated Separation of Cell Subpopulations
Devices that can sort heterogeneous cancer cells based on phenotypic differences will advance our
understanding of the metastatic cascade.125 In our previous work, we demonstrated the principle
behind a nanoparticle-mediated sorting strategy for cancer cell subpopulations. Cancer cells can
be sorted on the basis of different surface marker expression levels, simply by using magnetic
nanoparticles functionalized with antibody- based capture agents such as anti-EpCAM. Cells are
captured using X-shaped structures made from PDMS that create localized regions of low flow.
Cells are captured in the apex of the X when the magnetic force applied on the cell is greater than
the drag force the cell experiences as it flows through the device.80
Here, we explored differing phenotypes of the EpCAM-sorted cells. We used four different cell
lines in this study: MCF-7, MDA-MB-231, SKBR3 and SKBR3-EMT (Figure 3.2). Cells are
captured in a microfluidic device using anti-EpCAM magnetic nanoparticles (MNPs) and binned
into 4 different zones based on their EpCAM expression. Highly epithelial MCF-7 cells are
captured in the early zones of the microfluidic device, due to a high concentration of magnetic
nanoparticles on the surface of the cells.98 MDA-MB-231 cells which express low levels of
EpCAM and display aggressive metastatic behaviour in vivo 137, are located in the later zones of
the device (Figure 3.2A). SKBR3 cells are captured in the earlier zones of the device (zone 1 and
zone 2), while SKBR3-EMT cells shift to zone 2 and zone 3 (Figure 3.2B). This shift is a clear
41
representation of a reduction in EpCAM expression and the epithelial-to-mesenchymal transition
in SKBR3 cells after treatment with CoCl2. EpCAM profiling in the microfluidic device is
compared to flow cytometric analysis of the four cells (Figure 3.2C). The microfluidic device is
capable of capturing low numbers of cancer cells spiked in blood (Figure 3.2D), allowing for the
functional analysis of rare CTCs.
Figure 3.2 Microfluidic profiling of breast cancer cells. Cells are labelled with anti-EpCAM magnetic nanoparticles
and captured in the microfluidic device. (A) Cell sorting profile of MCF-7 and MDA-MB-231 cells. (B) Cell sorting
profile of SKBR3 and SKBR3-EMT cells. SKBR3-EMT cells are treated with CoCl2 for 72 hours. (C) Flow cytometric
analysis of EpCAM levels in MDA-MB-231, SKBR3, SKBR3-EMT and MCF-7 cells. (D) Cell sorting profiles of
low numbers of MCF-7 and MDA-MB-231 cells spiked in whole blood. Cells are captured and then immunostained
with cytokeratin-APC, DAPI and CD45-FITC. Cancer cells are identified as CK+/DAPI+/CD45-. Experiments are
repeated in triplicate. Standard errors of the mean are shown. Statistics are performed with one-way ANOVA followed
by the Tukey multiple comparisons (p<0.05). This data was prepared by B.J.Green and L.Kermanshah.
42
3.2.3 Collagen Uptake as a Measure of Invasiveness
A collagen uptake assay was utilized to identify cancer cells that ingest fluorescently-labelled
collagen as a measure of invasiveness (Figure 3.3A). Cancer cells have the ability to extend
invadopodia into the extracellular matrix (ECM) and to secrete proteases that digest collagen
fragments, leading to extravasation.99, 120, 138, 139 We hypothesized that the phenotypic changes
cells experience during EMT can be monitored using the microfluidic approach.
Fibrillary type I collagen was chosen to mimic the ECM of invading breast tumor cells.140 Collagen
fragments are internalized into the cell via collagenolysis through the Endo180 receptor, which is
expressed in basal breast tumors, and thus promotes tumor growth.139
MCF-7, MDA-MB-231, SKBR3 and SKBR3-EMT cells were analyzed using the collagen uptake
assay. SKBR3-EMT cells were treated for 72 hours with CoCl2 prior to collagen uptake analysis.
This treatment time was chosen as an increase in protein levels of HIF-1α, and upregulation of
mesenchymal markers Snail, Slug, Vimentin, N-Cadherin and ZO-1 have been observed after 72
hours of treatment. Snail family members including Snail and Slug increase matrix
metalloproteinases (MMP) expression and activity, likely mediating EMT and invasion.141, 142
Cells were cultured on the matrices and collagen levels were measured using flow cytometry. The
inclusion of folate with collagen enhanced the collagen uptake of aggressive MDA-MB-231 cells,
with corresponding low uptake by benign MCF-7 cells (Figure 3.3A,B).
Additionally, the fluorescent collagen uptake of SKBR3-EMT cells was higher than SKBR3 cells
(Figure 3.3C). Hypoxic cells have increased expression levels of MMPs-1, -2, -9, 10 and -13,
which are implicated in ECM degradation and tumor cell migration via the protein kinase C
pathway; 134, 143 an effect that may contribute to enhanced collagen uptake.
To apply the collagen assay to the analysis of cancer cell subpopulations, SKBR3 cells were
captured in the microfluidic device with anti-EpCAM MNPs and isolated from each of the 4 zones.
Cells isolated from the later zones of the device (zones 3 and 4) had increased collagen uptake
compared to cells isolated from zones 1 and 2 (Figure 3.3D, Figure 10.1.2). SKBR3-EMT cells
exhibited higher collagen uptake in each zone relative to the SKBR3 cells. Additionally, SKBR3
cells isolated from low-EpCAM zones expressed higher levels of folate receptor (Figure 10.1.3),
which might have contributed to the enhanced metabolic activity. Notably, we also demonstrate
43
microfluidic enrichment and collagen uptake analysis of rare CTCs from four metastatic prostate
cancer patients (Figure 10.1.5).
Figure 3.3 Collagen uptake assay. (A) Representative images of breast cancer cells that have ingested collagen. Cells
are stained with DAPI, cytokeratin-APC, and FITC collagen. Scale bar represents 5 µm. (B and C) Collagen uptake
in MCF-7, MDA-MB-231, SKBR3 and SKBR3-EMT cells. Flow cytometry median relative fluorescent intensities
are shown normalized to the unstained control. (D) Collagen uptake in SKBR3 and SKBR3-EMT cell subpopulations.
Cells are released from the microfluidic device and analyzed using flow cytometry for ingested collagen. Median
fluorescent intensities are shown relative to the unstained control. Experiments are repeated in triplicate. Standard
errors of the mean are shown. Statistics are performed with two-tailed t-test (p<0.05)..
3.2.4 NAD(P)H Metabolism in Cell Subpopulations
We also applied an assay recording the NAD(P)H metabolism of the cancer cells to provide an
indication of invasive phenotype. Metastatic CTCs are proposed to have increased metabolic
activity; therefore, we hypothesized that low-EpCAM cells would have an elevated NAD(P)H
response to folate.110, 129
The main metabolic route for folate-dependent NADPH production involves transfer of a one-
carbon unit from serine to tetrahydrofolate, leading to the production of DNA and RNA.144 The
aggregate signal of NAD(P)H is therefore a readout of the metabolic state of the cell. Changes in
NAD(P)H levels are determined using live cell two-photon microscopy of endogenous NADH and
NADPH, which are both autofluorescent with an emission spectra of 380-550 nm (Figure
44
3.4A).145-147 These images provide sufficient spatial resolution to independently measure the
cytoplasmic and mitochondrial NAD(P)H responses.
The NAD(P)H response was measured for MCF-7, MDA-MB-231, SKBR3 and SKBR3-EMT
cells following treatment with folate (Figure 3.4B-E). SKBR3-EMT cells were treated for 24 hours
with CoCl2, as this time is typically sufficient to increase metabolites such as lactate
dehydrogenase activity, reactive oxygen species content and NAD(P)H oxidase (Nox1)
expression.148
The NAD(P)H response of SKBR3 cells was examined after the cells were isolated from each zone
within the microfluidic device. Cells isolated from zone 3 had a higher mitochondrial NAD(P)H
and cytoplasmic NAD(P)H response relative to cells isolated from zone 1 (Figure 3.4F,G, Figure
10.1.4). This zone effect was not observed in SKBR3-EMT cells, as the baseline autofluorescence
was already elevated due to hypoxia. Overall, NAD(P)H responses were greater in the
mitochondria than in the cytoplasm, when examining the total cell populations (Figure 3.4F,G),
likely due to the concentration of folate induced one-carbon metabolism in the mitochondria.144
45
Figure 3.4 NAD(P)H response of breast cancer cells. To assess the NAD(P)H response, all cells are treated with
2.8 mg/L folate. (A) Representative mitochondrial NAD(P)H images from MCF-7, MDA-MB-231 and SKBR3 cells.
Scale bar represents 5µm. (B and C) Relative NAD(P)H intensities in response to folate for MCF-7 cells and MDA-
MB-231 cells in the mitochondria and cytoplasm, respectively. (D and E) NAD(P)H relative intensities of SKBR3
and SKBR3-EMT cells in the mitochondria and cytoplasm, respectively. (F and G) NAD(P)H intensities of zone
populations of SKBR3 cells and SKBR3-EMT cells in the mitochondria and cytoplasm, respectively. Cells are serum-
starved for 30 minutes in folate-free media before incubation with folate. NAD(P)H intensities are reported relative
to the baseline autofluorescence. Zone 1 was chosen to represent low-EpCAM expressing cells and zone 3 was chosen
to represent high-EpCAM expressing cells. The average autofluorescence from each group was reported. Experiments
are repeated in triplicate, and standard errors of the mean are shown. Statistics are performed with two-tailed t-test
(p<0.05).
3.3 Conclusion
The identification and characterization of cancer cell subpopulations will enable more precise
monitoring of phenotypic changes occurring during EMT. Understanding the heterogeneity of
cancer cells is important for accurate diagnosis and effective treatment of the disease.149 Using
microfluidics and magnetic nanoparticles for cell separation, we demonstrate the ability to detect
subpopulation changes occurring during EMT. Overall, SKBR3-EMT cells had higher collagen
uptake and folate-induced NAD(P)H metabolism relative to SKBR3 cells. The unique feature of
46
the microfluidic device enables us to isolate subpopulations of cells. Invasive CTCs are reported
to retain some of their epithelial properties.121 Consistent with this result, we show that low-
EpCAM expressing SKBR3 cells had increased collagen uptake relative to high-EpCAM cell
populations. This effect was enhanced in cells undergoing EMT.
Rare cancer cells are often enriched and quantified as a total population,74, 77, 91, 150 and the work
described here highlights the importance of studying subpopulations. Our platform presents a
unique strategy to characterize circulating tumor cells for specific functional analysis.
3.4 Methods
3.4.1 Cell Culture
SKBR3 cells (ATCC HTB-30) were cultured in McCoy's medium (ATCC 30-2007). MDA-MB-
231 cells (ATCC HTB-26) were cultured in Gibco DMEM F12 (11330-032). MCF-7 cells (ATCC
HTB-22) were cultured in EMEM Medium with 10µg/ml insulin (ATCC 30-2003). All media
were supplemented with 10% FBS, 1% penstrep and cells were cultured at 37°C and 5% CO2.
Cells were harvested when they reached more than 70-80% confluency. Cell detachment from the
culture dishes was performed using 0.5 ml of 0.25% trypsin/EDTA (Sigma Aldrich, US) for 10
minutes at 37oC.
3.4.2 Hypoxic Induction of SKBR3 Cells
Hypoxia was mimicked in SKBR3 cells by seeding the cells in 6-well plates (2×105 cells/well) and
treating them with cobalt chloride (CoCl2) solution (Sigma Aldrich, US) at the concentration of
150 μM for 24, 48 and 72 hours. After the treatment, samples were subjected to genotypic and
phenotypic characterization.
SKBR3 cells were treated with 150µm of cobalt chloride (CoCl2) to create a hypoxia model of
EMT. After 72 hours of treatment with CoCl2, a significant morphological change was observed
in SKBR3 cells. The intercellular space between cells was increased due to the loss of cell
adherence. Also, the cell morphology became more spindle-shaped with pseudopodia being
extended to enhance motility.
Accumulation of HIF-1α was confirmed by Western blot analysis after 24 hours of treatment. To
determine the cell migration ability, an in vitro wound healing assay was carried out. Treated cells
47
had increased scratch closure rate compared to the control sample, exhibiting an invasive
mesenchymal phenotype.
Gene expression was confirmed using RT-qPCR, demonstrating that epithelial genes (EpCAM,
cytokeratin) were downregulated and mesenchymal genes (Snail1, Slug and Vimentin) were
upregulated after 24 hour, 48 hour and 72 hour treatment with CoCl2. Protein expression was
confirmed using flow cytometry, where epithelial markers (EpCAM, E-Cadherin, and cytokeratin)
were downregulated and mesenchymal marker N-Cadherin was upregulated after 72 hour
treatment with CoCl2.151
Protein expression was also confirmed using mass cytometry, where it was observed that CoCl2-
treated cells had reduced expression of epithelial markers EpCAM, E-Cadherin, cytokeratin and
ZO-1, and increased expression of mesenchymal marker β-catenin after 72 hour treatment with
CoCl2.
3.4.3 Chip Fabrication
Chips were fabricated using Poly(dimethoxysilane) (PDMS, Dow Chemical, US) soft-lithography.
Masters were fabricated on silicon substrates and patterned in SU-8 3050 (Microchem, US). PDMS
replicas were poured on masters and baked at 67°C for 45 minutes. After peeling the replicas,
holes were pierced to connect the tubing. PDMS replicas were attached to no. 1 glass cover slips
using a 30 second plasma treatment and left to bond overnight. Afterward, the silicon tubing was
attached to the inlet and outlet of the device. Prior to use, devices were conditioned with 1%
Pluoronic F68 (Sigma-Aldrich, US) in phosphate-buffered saline (PBS) for 1 h, to reduce
nonspecific adsorption. Each device was sandwiched between two arrays of N52 Nd FeB magnets
(K&J Magnetics, US, 1.5 mm by 8 mm) with alternating polarity. The arrays of magnets placed
against the microfluidic device created a high magnetic field gradient inside the channels. A
syringe pump (Chemyx, US) was used for the duration of the cell capture process. The height of
the microfluidic channels is 420µm. The velocity valley device was scaled up to enhance
throughput. The diameter of the PDMS X-structures is 834µm, which are equally spaced in the
device. The dimensions of the zones are as follows: Zone1- 4.3 x 5.2 mm, zone 2- 7.5 x 5.2 mm,
zone 3- 13.9 x 5.2 mm, zone 4- 28.4 x 5.2 mm.
48
3.4.4 Cell Enrichment using Anti-EpCAM Magnetic Nanoparticles
Cells were initially tagged with magnetic nanoparticles labelled with anti-EpCAM antibodies
(EpCAM MNPs, Miltenyi Biotec, US). These superparamagnetic nanoparticles were composed of
iron oxide and dextran, and were 50 nm in diameter. For on-chip cell capture, a cell concentration
of 5x104 cells were prepared in a 1ml solution of PBS with 1% bovine serum albumin (BSA). Cells
were incubated with 10µl of EpCAM MNPs for 30 minutes.
After the EpCAM MNPs were attached to the cells, the cells were captured by applying an external
magnetic field in the chip. Cells were captured in separate devices. Cells were introduced into the
chip using a syringe pump at constant flow rate of 18 ml/h. Cells were trapped in the apex of the
X- structures when the magnetic force exceeds the drag force. The magnetic force on the cell is
proportional the number of bound magnetic nanoparticles; therefore, cells with high EpCAM
expression will be captured in the first zone. The linear velocity decreases in a stepwise manner in
each zone, to increase the threshold area for capture. Consequently, cells with low EpCAM
expression (and a low number of magnetic nanoparticles) are captured in the later zones.
Captured cells were then extracted from the zones. The tygon tubing connecting each zone of the
device were cut, and the magnets were removed. Cells were gently pipetted out of each zone for
downstream analysis. Typically 8 devices were run in parallel and cells isolated from each zone
were combined. Multiple devices were run in parallel in order to obtain high cell concentrations
in low-EpCAM zones 3 and 4, for flow cytometry and metabolic analysis. A maximum of 50,000
cells were loaded per chip. The distribution for SKBR3 was approximately 30% in zone 1, 55% in
zone 2, 10% in zone 3 and 5% in zone 4. The viability of extracted cells was determined by staining
cells with Trypan blue (Gibco, US) and manually counting the live cells with a haemocytometer.
The cell viability was >80% from each zone.
49
3.4.5 Microfluidic Profiling of Breast Cancer Cells Spiked in Blood
Whole blood was obtained with consent from healthy donors. Low numbers of MCF-7 cells and
MDA-MB-231 cells were spiked into 1ml of whole blood, and incubated with anti-EpCAM
nanobeads for 30 minutes. Cells were then introduced into the microfluidic device and captured in
the presence of an external magnetic field. To determine capture efficiency, cells were
immunostained in the device and counted manually. The capture efficiency was 85 ± 5%.
After the blood was processed through the device, PBS-EDTA was added to wash away non-
specific cells. Cells were then fixed with 4% formaldehyde solution (Sigma-Aldrich, US) followed
by 0.2% Triton X-100 (Sigma-Aldrich, US) for permeabilization. Cells were stained with
cytokeratin- APC clone C-11 (Genetex GTX80205, US) and CD45-FITC clone 5B1 (Miltenyi
Biotec) for 1 hour in PBS containing 1% BSA and 0.1% Tween20. Cells were washed with 1%
BSA in PBS and stained with the nuclear stain, 4,6-diamidino-2-phenylindole (DAPI Prolong Gold
reagent, Invitrogen, US). Cells were imaged using a fluorescent Nikon TiE eclipse microscope and
images were acquired with NIS Elements (Nikon) using a 10X and 50X objective. Cancer cells
were identified as CK+/DAPI+/CD45-.
3.4.6 Collagen Uptake Assay
Fluorescein isothiocyanate (FITC)-labelled collagen I (1mg/ml) (US Biologicals, US) was
combined with folate (1mg/ml) (Sigma Aldrich, US) in a ratio of 70% collagen: 30% folate. The
resulting mixture was incubated in 24- well dishes for 24 hours at 4oC, achieving even coating of
the surface. Excess matrix solution was removed prior to cell loading onto the surface. Cells were
serum-starved with 0.5% FBS in their respective media for 8 hours, then released using 0.25%
trypsin/EDTA (Sigma Aldrich, US) and plated on the surface of the collagen matrix. Cells were
cultured on the matrix for 24 hours at 37ºC and 5% CO2 in their respective media with 10% FBS.
Post-culture, cells were released from the collagen matrix subsequent to incubation with 10mg/ml
collagenase (Sigma Aldrich, US) for 10 minutes. The ingested collagen is fluorescently labelled;
thus, the cells can be identified by immunofluorescence based methods, such as flow cytometry
and fluorescence microscopy.
50
3.4.7 Immunocytochemistry
Released cells were fixed with 4% formaldehyde solution (Sigma-Aldrich, US) followed by 0.2%
Triton X-100 (Sigma-Aldrich, US) for permeabilization. Cells were stained with cytokeratin- APC
clone C-11 (Genetex GTX80205, US) for 1 hour in PBS containing 1% BSA and 0.1% Tween20.
Cells were washed with 1% BSA in PBS and stained with the nuclear stain, 4,6-diamidino-2-
phenylindole (DAPI Prolong Gold reagent, Invitrogen, US). Cells were washed and introduced
into a 24-well dish for imaging. Immunostained cells were imaged using a fluorescent Nikon TiE
eclipse microscope with an automated stage controller and an Andor camera and images were
acquired with NIS Elements (Nikon) using a 10X and 50X objective. Cancer cells were identified
as CK+DAPI+FITC+ or CK+DAPI+FITC- depending on the levels of collagen uptake. Zoomed in
images of the cells were obtained using the 50X objective.
The average FITC intensity within the cells was measured using ImageJ. The relative fluorescence
intensity was obtained by subtracting the background (from the cell intensity), then dividing by
the background intensity.
3.4.8 Flow Cytometry
Cells were harvested from tissue culture using 0.25% trypsin/EDTA (Sigma-Aldrich, US) and
incubated with blocking buffer (PBS + 1% BSA) for 30 minutes. For each cell line, 5×105 cancer
cells were fixed and permeabilized using 4% paraformaldehyde (Sigma-Aldrich, US) and 0.2%
Triton X (Sigma-Aldrich, US) in PBS, respectively. Cells were then washed with PBS and
suspended in PBS containing 1% BSA and 0.1% Tween20, and incubated with anti-E-Cadherin-
APC clone 67A4 (BioLegend, US), anti-EpCAM Alexa Fluor 647 clone 9C4 (BioLegend, US),
anti-PAN CK-APC clone C-11 (Genetex, US), anti-N-Cadherin Alexa488 clone 8C11
(BioLegend, US), anti-vimentin-Alexa Fluor 488 clone RV202 (BD biosciences, US), or anti-
folate receptor alpha Alexa Fluor 647 clone 548908 (Novus Biologicals, US) at 1:50 dilution and
stained at room temperature for 30 minutes. Mouse IgG (Abcam, US) was used as a negative
control at the corresponding assay-specific concentrations. Samples were washed with PBS and
re-suspended in 1% BSA in PBS. Samples were injected into a BD FACS Canto flow cytometer
(BD Biosciences, US) and measurements were plotted as histograms or median fluorescence for
each fluorophore (AF647 and FITC). Median fluorescence values were normalized to an unstained
control. A total of 5,000- 10,000 cells were analyzed per cell line.
51
3.4.9 NAD(P)H Dose-Response
Cells were plated on glass- bottom Mattek dishes (Mattek, US) for 24 hours. Cells isolated from
each zone were plated on separate dishes. Cells were serum-starved in 0.1% BSA in imaging media
(125 mM NaCl, 5.7 mM KCl, 2.5 mM CaCl2, 1.2 mM MgCl2, 10 mM HEPES, pH 7.4) for 30
minutes at 37oC to achieve baseline autofluorescence. Cells were then incubated with increasing
concentrations of folate (1.7 mg/L and then 2.8 mg/L) for 10 minutes each. The NAD(P)H response
after treatment with 1.7 mg/L folate is shown in the supporting information Figure 10.1.4, and the
response after 2.8 mg/L folate is shown in Figure 3.4. NAD(P)H images were collected on a Zeiss
LSM710 microscope using the 40x 1.3 NA objective lens, and two-photon excitation (710 nm at
~3.5 mW). NAD(P)H fluorescence was collected through a custom 385-550 nm emission filter
using a non-descanned BiG (GaAsP) detector. Images were collected at baseline and with each
increase in folate concentration. The same cell population was imaged at each subsequent
treatment.
The total number of cells assayed was 20, 20, 19 and 14 for MCF-7, MDA-MB-231, SKBR3 and
SKBR3-EMT, respectively. After capture, cells were analyzed using ImageJ software. A custom-
built macro was used to apply an Otsu threshold on the NAD(P)H autofluorescence signal within
a single cell region of interest (ROI), to identify mitochondrial ROIs and the mitochondrial
NAD(P)H signal. Cytoplasmic NAD(P)H level was determined manually by measuring the
intensity within the nuclear region of the cells. The resulting NAD(P)H intensities were normalized
to the baseline to obtain the relative intensity.
3.4.10 Patient Sample Collection
Metastatic castration-resistant prostate cancer patients were recruited from the Princess Margaret
Hospital according to the University’s Research Ethics Board approved protocol. All patients were
enrolled following informed consent. 10 ml of peripheral blood samples from castration-resistant
prostate cancer patients (n=4) were used to validate the collagen uptake assay. Blood samples were
collected in a CellSearch tube containing the anticoagulant EDTA (Johnson and Johnson). All
samples were analyzed within a 24-hour window after blood collection.
52
3.4.11 CTC Capture from Patient Samples, Isolation and Analysis
CTCs were captured in the microfluidic device using EpCAM- specific aptamers conjugated to
magnetic nanoparticles. The modification of EpCAM specific aptamers with magnetic
nanoparticles was carried according to the following procedure. Briefly, 100 µL of 20 µM EpCAM
specific aptamer in Dulbecco's phosphate-buffered saline (DPBS, Sigma-Aldrich, US) were
denatured for 5 minutes at 95°C then it was renatured on ice for 10 minutes. Afterward, the aptamer
solution was incubated with 1 µL of 10 mg/mL of streptavidin coated magnetic nanoparticles
(100µm, Chemicell, US) in a microtitre plate for 30 minutes at room temperature. Subsequently,
the modified nanoparticles were deposited using a magnetic stand (Thermofisher, US) and washed
twice with DPBS.
Two millilitres of blood per patient were depleted of RBCs and WBCs prior to analysis. The blood
samples were depleted of RBCs by incubating with 2 ml of RBC lysis buffer (Sigma-Aldrich, US)
for 5 min at room temperature. The mixture was subsequently centrifuged for 10 minutes at 4,000g.
The supernatant was discarded and the cells were incubated with 50 µL of anti-CD15 antibody
(Miltenyi Biotec Inc., US) for 30 minutes at room temperature for WBCs depletion. Afterward,
the magnetized WBCs were deposited using a magnetic stand and the supernatant was collected.
The supernatant was then mixed with 75 µM mercaptoethanol (MCE) and incubated with 100 µL
of the EpCAM aptamer-conjugated nanoparticle solution in DPBS for 1 hour at room temperature.
The mixture was loaded into the microfluidic device and the cells were captured at a flow rate of
8 mL/h. Afterward, 200 µL of 200 µM antisense DNA in PBS (preheated at 60°C) were loaded
into the chip and incubated with cells for 30 minutes, in order to release the on-chip captured
CTCs.
Tygon tubing connecting each zone were cut, and the cells were gently pipetted out of the device
for downstream analysis. The cells retrieved from the four zones were cultured in 24-well plates
previously coated with 70% 1 mg/mL FITC-collagen (US Biological, US) and 30% 1 mg/ml folic
acid (Sigma Aldrich, US). After adding the cells to the wells, 1 mL of DMEM medium (ATCC
30-2002), containing 10% FBS and 1% penicillin-streptomycin, was added to each well and the
plates were incubated for 24 hours at 37 °C and 5% CO2. Cells were released after incubation with
10 mg/mL collagenase enzyme (Sigma-Aldrich, US) for 10 minutes at 37°C.
53
After this step, the released cells were mounted into a glass slide and immunostained as detailed
in Section 3.4.7. An additional antibody, CD45- Alexa Fluor 555 (Bioss Antibodies) was included
to stain non-target white blood cells. The FITC collagen relative intensity was measured using NIS
imaging software. The relative fluorescence intensity was obtained by subtracting the background
(from the cell intensity), then dividing by the background intensity. The collagen assay described
above was also performed using blood collected from healthy donors (n=2) as a control, and 0
CTCs were reported.
Sequence of the EpCAM aptamer
(PO4)-5' TGA AGG TTC GTT TCG GTG GGT GTA GAC TCT TTA GAA GAG ATA CAG
ATT TTG GGA ATG 3'–(TEG)-Biotin
Sequence of the antisense DNA
5' CAT TCC CAA AAT CTG TAT CTC TTC TAA AGA GTC TAC ACC CAC CGA AAC
CAA CCT TCA 3'
The aptamer and antisense DNA was purchased from Integrated DNA Technologies (IDT, US).
3.4.12 Statistics
For experiments with multiple data points, one-way ANOVA followed by the Tukey multiple
comparisons test was used to assess statistical significance. For experiments with two data points,
two-tailed paired t-test was performed. Pairings with p values < 0.05 were accepted as statistically
significant.
54
4 Metastatic Cancer Cell Pathfinding through Porous Micro-structures
Invasion of dense tissues by cancer cells involves the interplay between the penetration resistance
offered by interstitial pores and the deformability of cells. Metastatic cancer cells find optimal
paths of minimal resistance through an adaptive pathfinding process, which leads to successful
dissemination. The physical limits of nuclear deformation is related to the minimal cross section
of pores that can be successfully penetrated. However, this single biophysical parameter does not
fully describe the architectural complexity of tissues featuring pores of variable area and shape.
Here, employing laser nanolithography, we fabricate pore microenvironment models with well-
controlled pore shapes, through which human breast cells (MCF10A) and their metastatic offspring
(MCF10CA1a.cl1) could pervade. In these experimental settings we demonstrate that the pore
actual shape, and not only the cross section, is a major and independent determinant of cancer
penetration efficiency. In complex architectures containing pores demanding large deformations
from invading cells, tall and narrow rectangular openings facilitate cancer migration. In addition,
we highlight the characteristic traits of the explorative behavior enabling metastatic cells to
identify and select such pore shapes in a complex multi-shape pore environment, pinpointing paths
of least resistance to invasion.
Reprinted with permission from B.J. Green1, M. Panagiotakopoulou1, F.M. Pramotton, G.
Stefopoulos, S.O. Kelley, D. Poulikakos, A. Ferrari. “Pore shape and orientation define paths of
metastatic migration”, Nanoletters, 2018 Mar 14;18(3):2140-2147. Copyright 2018 American
Chemical Society.
1 Equal contribution
Link to publication online: https://pubs.acs.org/doi/10.1021/acs.nanolett.8b00431
B.J.G. conducted experiments, contributed to the device design, and aided with manuscript
writing. M.P. designed and prepared the samples, conducted experiments, analyzed the data and
aided with manuscript writing. B.J.G. and M.P. contributed equally to the study. F.M.P. aided with
experimental analysis. G.S. aided with experimental analysis, project guidance and manuscript
writing. S.O.K. supervised the study, aided with data interpretation and revised the manuscript.
55
D.P. supervised the study, aided with data interpretation and revised the manuscript. A.F. designed
and supervised the study, aided in data interpretation, and wrote the manuscript.
4.1 Introduction
Interstitial migration of tumor cells constitutes the pathological link between a primary lesion and
its metastatic progression in a distant body location.152 The dissemination of cancer seeds occurs
through interstitial microenvironments with complex architectures generated by extracellular
matrix fibers, adhesion proteins, proteoglycans, and stromal cells.153 Beyond biological signaling,
the physical resistance offered by inherent obstacles encountered along the migration path is
therefore a critical determinant of the tissue infiltration performance.154 Highly invasive cells can
optimize their migration strategy to select paths of least resistance in the interstitium.155
Invasive cancer cells embedded within a dense 3D extracellular matrix (ECM) must overcome the
surrounding physical constraints to allow tumor expansion.156, 157 Metastatic cells detect matrix
elasticity and adapt their migration mode to enable efficient dissemination158. For protease-
dependent advancement they recruit proteolytic systems to sever collagen fibrils and enlarge
matrix pores to a comfortable size.156 Alternatively, protease- independent migration modes can
be adopted. In this case the cells exploit their deformability to fit through narrow openings while
the ECM is not remodelled. Adaptive interconversion of these two migratory phenotypes is a
hallmark of invasive tumor cells.
Directional migration is the result of cell polarization, which generates a propulsive front-to-rear
imbalance of cellular tractions.159 In addition, the cell spreading sustains an apico-basal
distribution of the components establishing adhesion to the substrate.160 The resulting asymmetric
distribution of the cytoskeleton contributes to shape the cell and the nucleus.161 Lateral
compressive forces generated by actin filaments are the main actuator of nuclear deformations
typical of anisotropic spreading and polarized migration. Normal compressive forces contribute to
reshape the nucleus to a much lesser extent.162
Penetration of a cell through dense tissues and narrow openings is connected to the nuclear
stiffness.163 Nuclear deformability thus defines the physical limits of interstitial pore penetration
164, 165 and subtends to different performances in distinct cell types and cell cycle phases.166 It is
therefore logical to hypothesize that the penetration efficiency of a cell when interfacing an
56
obstacle that requires high nuclear deformation, is influenced not only by the pore area but also by
the pore shape confronted by the cell.167, 168 Based on this assumption, the pore geometry may
significantly contribute to define paths of least resistance that can be hijacked by metastatic cells.
The influence of pore cross-section on interstitial penetration has been established by detailed
studies, which exploited reconstituted collagen matrixes164, artificial filters169, or microfluidic
channels 165, 170, 171,172 to constrain the advancement of migrating cells. However, while interstitial
fissures of human tissues feature heterogeneous cross-sections and geometries,153, 173 these
methods do not adequately provide the flexibility to explore the mutual interplay between the size
and shape of constrictions challenging cell invasion.
In this work, 3D laser nanolithography of a host of basic rectangular pore designs was exploited
to distinguish the effects of pore cross-section and shape on interstitial cancer migration.174
Insurmountable vertical walls were fabricated on optically transparent substrates to obstruct the
progression of migrating cells and to force them to explore pore gates imposing large nuclear
deformations. The flexibility of the nanofabrication method allowed complete freedom in the
spatial arrangement and geometry of the openings, whose well-defined perimeter corresponds to
the values reported for interstitial pores of human tissues.173 This platform therefore provides a
model of protease- independent pore penetration during interstitial migration. The direct
observation of cell interaction with openings displaying variable shape allowed the establishment
of these topographic parameters as independent determinants of interstitial migration. Based on
this paradigm, complex spatial arrangements of pores with identical area and variable shape were
generated to reveal the pathfinding capability and polarization versatility of migrating metastatic
human breast cancer cells.
57
4.2 Results and Discussion
Vertical walls with basal pores of defined area and shape (Figure 4.1A) were designed to halt the
broad migration of human breast cells, and enable the focusing of their passage through the basal
openings. The 3D nanolithography structures were fabricated on glass coverslips, in block arrays
(Figure 10.2.1). The unit of the array comprised a 22 µm tall, 112 µm in length and 2.5 µm thick
vertical wall. These dimensions were comfortably realizable by laser nanolithography. The unit
wall height was selected to generate an impassable obstacle to cell migration while the wall length
ensured its structural stability. A single unit contained 3 pores of a given cross-sectional area and
aspect ratio (i.e. the width to height ratio; Figure 4.1A, Table 10.2.1). The openings were spaced
out by 20 µm to enable the cells to explore multiple passages at once (Figure 10.2.2). The block
array consisted of 2 units per side; and 8 units total per block. The structures were printed in block
fashion to allow cells to interact with 24 pores over a period of 24 h. The large number of openings
per block increased the probability of pore engagement.
The cross-sectional area of the openings was selected in a range encompassing the values reported
for pores in the human dermis.173 Specifically, four distinct values ranging from 16 to 49 µm2 were
included in order to impose large nuclear deformations on penetrating cells (60% or more 170).
Pores with the same cross-sectional area were designed in four aspect ratio (a.r.) variations,
yielding either squared (a.r. = 1), tall and narrow (a.r. = 0.1, 0.3) or flat and wide rectangular (a.r.
= 3) openings. Therefore, the resulting parametric matrix included 16 pores of variable shape
and/or size which could be freely arranged in the migration study (Figure 4.1A-C).
Human breast epithelial cells (MCF10A) and corresponding highly metastatic offspring
(MCF10CA1a.cl1) expressing histone-2B-eGFP were selected as their migratory behavior and
metastatic potential is well established.175, 176 The two cell lines feature similar nuclear size,
therefore setting the same nuclear deformation to penetrate identical pores (Figure 10.2.3A).
Migration of MCF10CA1a.cl1 cells on gratings (2 µm period, 50% duty cycle, 1 µm grove depth;
166) showed comparable alignment and slightly increased persistence, indicating a similar behavior
in response to topographic contact guidance (Figure 10.2.4).
58
Figure 4.1. Experimental pore micro-structure design. (A) Schematic of cells engaging with printed pores and (B)
Characteristic scanning electron microscopy images (45° tilt) of the structure design with pores featuring variable
aspect ratio (a.r.). (C) Overview of the pore geometries, with varying cross section area (µm2) and a.r. (D) Fluorescence
nuclear images extracted from a time lapse of the nucleus of an MCF10CA1a.cl1 cell disengaging (upper row) from
a 36µm2 pore with a.r.=0.3, an MCF10A cell penetrating (middle row) through a 16µm2 pore with a.r.=1, and an
MCF10A cell at an impasse (lower row) through a 36µm2 pore with a.r.=0.3. The engagement times are 1.7 h, 4.7 h
and 9.7 h, respectively.
59
As expected, transformed cells were characterized by a shorter cell cycle, faster migration velocity,
and higher invasiveness in matrigel invasion assays (Figure 10.2.3B-D). Protein expression
profiles further confirmed the different phenotype of MCF10CA1a.cl1 consistent with their
metastatic transition 177 (Figure 10.2.5, Figure 10.2.6). MCF10CA1a.cl1 cells showed elevated
protein levels of HRas, vimentin, Talin1 and neural cell adhesion molecule relative to MCF10A
cells.
In penetration experiments, cells were seeded on substrates featuring multiple topographic
elements (Figure 4.1C, Figure 10.2.2). Cells migrating along the 3D structures required nuclear
deformation for successful penetration through the basal openings. This penetration process was
monitored through live cell microscopy over 24 h. The nuclear deformation and position relative
to the pore was used to define three different outcomes of a penetration attempt (Figure 4.1D).
After introducing the nucleus under the pore, the cell may either complete its translocation on the
opposite side (successful penetration), remain blocked (impasse), or disengage on the same side
(disengagement). Engagement events encompass any of penetration, disengagement or impasse.
Long-term observation allowed the capture of multiple engagement events. Each penetration
attempt was fully resolved for the corresponding size and a.r. of the engaged pore, as well as for
its temporal dynamics. This set of data provided a quantitative fingerprint for the behavior of the
two cell lines under investigation. Neither cell line could successfully penetrate pores featuring the
smallest cross-section (i.e. 16 µm2) and very few attempts were recorded for these openings
(Figure 4.2A, 4.2C, and Figure 10.2.7A, B). This result is in line with previous reports setting the
limit of nuclear deformation to 80-90%.164, 166 Attempts to penetrate larger passages (27 µm2) were
recorded for tall and narrow (vertically-oriented) pores (a.r. = 0.1 and 0.3) but rarely for isotropic
(a.r. = 1) or wide and flat (horizontally oriented) ones (a.r. = 3; Figure 4.2A, 4.2C, Figure
10.2.7C,D). Few penetration attempts were observed for flat pores of all tested cross-sections,
indicating that the necessary nuclear deformations are highly disfavored (Figure 10.2.7). These
results indicate that the pore shape and orientation are major determinants of the outcome for
penetrations requiring a large deformation of the cell nucleus (in the range between 70-90%). In
addition, they demonstrate that penetration attempts requiring lateral nuclear compression can be
accomplished more efficiently that those requiring a normal (to the cell surface) compression.
60
61
Figure 4.2. Effect of pore shape and geometry on cell penetration dynamics. Pore penetration and disengagement
of MCF10A cells (A, B) and MCF10CA1a.cl1 cells (C, D), as a function of cross sections and a.r. The color-coding
corresponds to the pathfinding index (PI), a descriptor of the cell-pore interaction outcome (see Methods). Each dot
in B and D represents 2 events. (E, F) Representative immunofluorescence confocal sections along the apical,
equatorial, and basal surfaces of MCF10A cells stained for nucleus (green) and actin (red) on substrate without (E)
and with (F) constrictions. Printed vertical barriers forming pores are reported in blue. (G) Three-dimensional
reconstruction of the confocal image of panel F which reveals the basal localization of actin fibers and the lack of an
actin cap. Typical disengagement (H) and penetration time (I) of MCF10A and MCF10CA1a.cl1 cells as a function
of a.r. for pores featuring a cross section of 36 µm2. Error bars correspond to the standard error of the mean. The
number of events (n) includes penetration, disengagement, or impasse for either cell type. Engagement events were
recorded over 24 h. * p<0.05. These experiments and data analysis were performed by B.J. Green and M.
Panagiotakopoulou.
The ability of cells to better penetrate pores featuring low a.r. may be related to the architecture of
the actin cytoskeleton which actuates the required nuclear deformation.162 Lateral compressions
enabling the penetration of tall and narrow pores (low a.r.) are generated by actin filaments
flanking the nucleus. Normal deformations, necessary for the penetration of flat and wide openings
(high a.r.), entail a contractile structure overhanging the nucleus (i.e. the actin cap; 161, 178). Both
MCF10A and MCF10CA1a.cl1 cells displayed prominent actin fibers on their basal and dorsal
sides (Figure 4.2E, Figure 10.2.15A). However, no organized actin structure was detected at the
apical side above the nucleus. The absence of an actin cap is typical of transformed cells and has
been associated to increased nuclear deformability upon migration through narrow constrictions.
178, 179 In our experimental settings, pervading MCF10A and MCF10CA1a.cl1 cells displayed a
characteristic actin meshwork organized at the lateral sides of the nucleus in correspondence to the
region withstanding a large deformation to fit the opening;162 (Figure 4.2F, G, 10.2.15B). Based
on these observations, we speculate that low a.r. (tall and narrow) pores, which display greater
lateral surface area, may offer an easier access to the required actin-mediated nuclear deformation,
leading to faster and more successful penetration.
MCF10A cells could engage all vertically-oriented (tall and narrow) or square pores. Most
penetration attempts led to either a successful penetration or to an impasse (~75% and ~15% of
the events; respectively. Figure 4.2A and 4.2B). Therefore, only very rare disengagement events
(~10%) ensued an initial nuclear engagement (Figure 4.2A-B, Figure 10.2.7). When exposed to
corresponding pores MCF10CA1a.cl1 cells showed a surprisingly different behavior characterized
by an almost complete absence of impasse (~4%) and a high disengagement frequency (~50%;
Figure 4.2C-D, Figure 10.2.7). This was further supported by the dynamics of cell disengagement,
which was accomplished three times faster by these metastatic cells (Figure 4.2H). Penetration
times showed lower differences between the two cell types (Figure 4.2I). These results indicate
62
that, while their non-transformed counterpart mostly remain committed to the engagement of a
pore, metastatic cells have the ability to temporarily explore an opening and quickly retract to
continue searching. This exploratory behavior exhibited with specific pore shapes defines a typical
migration strategy of MCF10CA1a.cl1 cells, which we defined as ‘pathfinding’ (see Methods;
Figure 4.2A and 4.2C).
Over the course of an experiment, cells underwent division or migrated to a sufficient extent to
form clusters. Therefore, the cell density increased both globally (as a result of proliferation) and
locally (as a result of cell clustering). MCF10CA1a.cl1 cells did not show any notable change in
the pore penetration performance as a function of the local density. The same frequency of
penetrations, disengagements, or impasse was recorded for a given pore geometry at low (i.e.
individual cells engaging the pore) or high (cells in contact with one or more neighbors) densities.
Differently, MCF10A cells exhibited fewer disengagement events and increased impasse at higher
cell densities. The majority (80%) of engagement events for MCF10A cells took place at low cell
density, yet these results suggest that the establishment of cell-to-cell junctions between epithelial
cells restricts migration and thus demotes the pervasion of pores. Mesenchymal transition in
metastatic cancer relaxes contact inhibition and may thus allow pore penetration despite local
crowding.180
Directional migration requires key intracellular structures, including the actin cytoskeleton, the
mitochondria, the Golgi apparatus (Golgi), the microtubule organizing center, and the plasma
membrane, to assume a typical front-to-rear position relative to the nucleus in a process sustained
by the coordinated activity of small GTPases.181, 182 The relocation of the Golgi at the front edge
provides membrane and associated proteins required for the generation of cell protrusions.183
Instability of this polarization mechanism is associated with cancer cell plasticity and results from
the dysregulation of controlling factors such as Cdc42.180 To decipher the role of directional
migration in the navigation through complex environments, we analyzed the polarity of MCF10A
and MCF10CA1a.cl1 cells attempting to penetrate the 3D micro-structures (Figure 4.3A). The
front-to-rear polarity of cells interacting with pores was visualized by a double fluorescent
staining, reporting the mutual position of the Golgi (Cell Light Golgi RFP) and the cell nucleus
(histone 2B GFP; Figure 4.3C-F).
63
In the absence of directional signals, neither cell type displayed a preferential positioning of the
Golgi, indicative of a random, unpolarized migration modality (Figure 4.3A). The formation of
small clusters (less than 10 cells) did not influence cell polarity (Figure 10.2.12). During the
experiments, nuclear engagement of the basal pores (Figure 4.2) was accomplished by the cells
either in a polarized or unpolarized manner. In the majority of cases for polarized engagement,
(65% for MCF10A and 66% for MCF10CA1a.cl1 cells; respectively) the Golgi was the first
compartment to be inserted in the opening (Figure 4.3B, Figure 10.2.9A,B, Figure 10.2.10A,B,
Figure 10.2.11A,B). The Golgi’s rapid penetration was followed by the nuclear engagement of the
same passage (Figure 4.3C, D). Such configuration, resulted in successful nuclear penetration in
the majority of cases (p = 0.75 and p = 0.81; for MCF10A and MCF10CA1a.cl1 cells;
respectively). After penetration, the cell continued to migrate away from the pore.
A consistent fraction of polarized MCF10A cells that entered the opening remained blocked into
a non-evolving nuclear engagement, yielding a significant impasse probability (p = 0.17). Nuclear
disengagement from the pore was rare, and was only observed with very small frequency (p =
0.03; Figure 4.3B). On the contrary, MCF10CA1a.cl1cells attempting a penetration through a
polarized engagement did not linger at impasses, but instead showed a high disengagement
frequency (p = 0.19), which allowed them to release from the pore.
During unpolarized engagement events (35% for MCF10A and 34% for MCF10CA1a.cl1 cells;
respectively, Figure 4.3B), nuclear engagement was accomplished by the cell when the Golgi was
lagging behind (Figure 4.3E, F). These engagements were generally unproductive, yielding a low
penetration rate (p = 0.31 and p = 0.23 for MCF10A and MCF10CA1a.cl1 cells; respectively) and
mostly resulting in disengagement (p = 0.62 and p = 0.65 for MCF10A and MCF10CA1a.cl1 cells;
respectively, Figures 4.3B, Figure 10.2.9C,D, Figures 10.2.10C,D; 10.2.11C,D). Impasse events
were not observed for unpolarized cells (Figure 4.3B). Finally, relocation of the Golgi from the
rear to the front was possible upon nuclear engagement (p = 0.08 and p = 0.12 for MCF10A and
MCF10CA1a.cl1 cells; respectively). For such transition the Golgi had to squeeze and penetrate a
passage engulfed by the nucleus. In summary, the frequency of disengagement events recorded for
MCF10CA1a.cl1 cells, combined with their faster dynamics (Figure 4.2E) demonstrate a higher
versatility during penetrative migration of these cells.
64
Figure 4.3. Cell polarization during pore engagement. (A) Relative Golgi-to-nucleus orientation in control
MCF10A and MCF10CA1a.cl1 cells migrating in the absence of directional signals. The cartoon in the graph inset
defines possible positions of the Golgi (red) relative to the nucleus (green) and the direction of migration (Back, Front,
or Lateral). (B) Markov chains reporting the probability (i.e. the frequency) of penetration, impasse, and
disengagement upon a polarized or an unpolarized engagement of a printed pore by a MCF10A (p values reported in
blue) or a MCF10CA1a.cl1 cell (p values reported in red). (C) Schematic of polarized engagement reporting the
relative position of Golgi (red) and nucleus (green) upon pore penetration. (D) Corresponding examples of MCF10A
and MCF10CA1a.cl1 cells penetrating a pore after a polarized engagement. Pore cross-section areas are 36 µm2 and
a.r. = 0.3. (E) Schematic of unpolarized engagement reporting the relative position of Golgi (red) and nucleus (green)
upon pore disengagement (F) Corresponding examples of MCF10A and MCF10CA1a.cl1 cells disengaging from a
pore after an unpolarized engagement. Pore cross-section areas are 36 µm2 and a.r. = 0.3.(G) Immunocytochemistry
quantification of Cdc42 activity in MCF10A and MCF10CA1a.cl1 cells. These experiments and data analysis were
performed by B.J. Green and M. Panagiotakopoulou.
65
This result is supported by the observation that activity levels of Cdc42 are upregulated in
MCF10CA1a.cl1 cells as compared to MCF10A (Figure 4.3G). The Rac1 and RhoA levels
between the two cells lines are similar (Figure 10.2.13); highlighting the observation that the
polarization of the cells is a function mainly regulated by Cdc42.
We next investigated whether the engagement instability of metastatic cells is a key driving force
for selective navigation in a complex porous environment. A dedicated experiment was designed
to test whether MCF10CA1a.cl1 cells can find a preferential path (Figure 4.4). Four barriers, each
composed of a repetition of vertical walls for a total width of 1.5 mm, were printed in a parallel
arrangement (Figure 4.4A,B). Individual vertical walls featured 5 basal pores with identical cross
section (36 µm2) but variable a.r. While the central pore was vertically-oriented (tall and narrow,
a.r.= 0.3) to offer a low-resistance passage, the remaining 4 pores had a horizontal orientation (flat
and wide, a.r. = 3) thus producing a majority of high-resistance openings (Figure 4.2 and Figure
4.4A). Finally, the periodic assembly of vertical walls was out of phase between subsequent
parallel barriers, yielding a staggered distribution of low-resistance pores (Figure 4.4B).
MCF10A or MCF10CA1a.cl1 cells were seeded on one side of the array and were allowed to
migrate towards the obstacles. While approaching the parallel barriers, cells tended to proliferate
and aggregate forming multicellular clusters progressing in the same direction.184 A cell image
velocimetry (CIV)185 analysis of motility in these clusters revealed that MCF10A cells moved with
high correlation typical of connected epithelial cells, prior to making contact with the barriers
(Figure 10.2.14).186-188 The correlation length indicates that groups of up to 10 cells tended to move
coherently within these clusters.
The vertical barriers hindered the advancement of MCF10A cells, and the movement lost
coherence displaying a correlation length close to a single cell diameter (Figure 4.4C, D). In
contrast, MCF10CA1a.cl1 collectives showed an initial low coordination (2-3 cells; Figure
10.2.14), which was however not affected by the interaction with the array (Figure 4.4D).
Specifically, the correlation lengths indicates that small cell clusters (Figure 10.2.14) were able to
navigate coherently in the complex porous environment. These results suggest that the ability of
metastatic cells to retain some degree of coordination in the presence of physical barriers may
contribute to pathfinding and increase their pervasion efficiency.
66
Figure 4.4. Cell navigation in complex porous environments. (A) Scanning electron microscopy images (45° tilt)
of the vertical wall units featuring 5 openings with cross section of 36 µm2 and a.r. of 0.3 (central pore) or a.r. of 3
(lateral pores). (B) Scanning electron microscopy images (45° tilt) of the staggered array of parallel barriers. (C)
Correlation function and (D) correlation length obtained from a CIV analysis189
of MCF10A and MCF10CA1a.cl1
cell clusters interacting with the parallel barriers. Error bars correspond to the standard error of the mean (E) Graph
reporting the measured selectivity index (preferential penetration of central pores) for MCF10A and MCF10CA1a.cl1
cells pervading the parallel barriers. A dashed green line defines the selectivity index value imposed by the
experimental design, corresponding to no preferential penetration of the central opening. (F) Component of MCF10A
and MCF10CA1a.cl1 migration along the parallel barriers. Error bars in boxplots correspond to the standard deviation.
* p<0.05. *** p< 0.001. These experiments and data analysis were performed by M. Panagiotakopoulou and F.M.
Pramotton.
67
The migration of MCF10A and MCF10CA1a.cl1 cells interacting with basal pores in the barrier
array was monitored to capture multiple penetration events. Selective navigation of cells was
evaluated by measuring the overall frequency of penetration through low resistance pores. A
random pore selection (identified by the selection index in Fig 4.4E) was expected to approach
0.2, a value imposed by the design (Figure 4.4E). Any preferential pervasion of more favorable
openings would render higher values with a maximum of 1 if only central pores are penetrated.
The analysis of MCF10A cells navigating through the barrier array showed that for these cells, the
penetration of high-resistance pores was almost 4 times more frequent than for the low-resistance
counterparts. The resulting selection index of 0.29 ± 0.05 (Figure 4.4E) is indicative of a poorly
selective advancement. Pervading MCF10CA1a.cl1 cells showed a markedly different behavior,
and were able to preferentially penetrate low-resistance pores. The selection index = 0.52 ± 0.06
was correspondingly much higher (Figure 4.4E). This adaptive pathfinding scheme was further
supported by an increased explorative movement in the direction parallel to the barriers (Figure
4.4F), which was less pronounced in MCF10A cells. These results demonstrate that
MCF10CA1a.cl1 cells are able to navigate through a complex arrangement of pores and select
openings with a vertical orientation.
Cell migration in interstitial tissues is an adaptive process resulting from the interplay between
advancing cells and the surrounding mechanical and molecular extracellular environment. This
mutual exchange or reciprocity38 involves ECM stiffness and dimension as well as cell
deformability. In dense matrixes obstructing cell pervasion, the shape of pores and their spatial
distribution along the direction of movement may constitute an independent physical parameter
contributing to define both the path and the outcome of cell migration (Figure 4.4). In this scenario,
metastatic cells may tailor their migration mode to successfully navigate across the interstitium.
4.3 Conclusions
In summary, this work exploited an on-demand nanolithography approach to generate complex
arrangements of topographic openings offering a passage with well-defined cross section and
shape to monitor migrating cells. The resulting reductionist model of the interstitium was used to
decouple the role of pore shape during the penetrative migration of normal and cancer cells (Figure
4.1). The data obtained from live cell observation clearly demonstrate that the pervasion of pores
demanding large nuclear deformation is pore shape dependent. In particular, elongated pore shapes
68
oriented along the apico-basal axis of cells offer a more favorable passage as compared to identical
pores with horizontal orientation (Figure 4.2). Furthermore, the comparative analysis of breast
epithelial cells and their metastatic offspring showed a higher versatility in the penetrative behavior
of the latter, which was linked to a rapid and frequent reversal of migration directionality and to a
versatile front-to-rear polarization (Figure 4.3). Upon the pervasion of complex porous
environments, the pathfinding behavior typical of metastatic cells is depicted as a functional
preference for low-resistance pore shapes (Figure 4.4). Therefore, while non-transformed cells
remain committed to the pores encountered along the path of directional migration, regardless of
the offered resistance, the quick engagement/ disengagement turnover allows metastatic cells to
dislodge from pores demanding a long-lasting interaction and only commit to penetrate favorable
passages.
4.4 Methods
4.4.1 Cell Culture
MCF10A, a non-tumorigenic epithelial cell line 190 and MCF10CA1a.cl1 cells, an invasive human
breast cancer cell line from the MCF10A xenograft model 2, 191 were a kind gift of Prof. Giorgio
Scita (IFOM, Milan, Italy). MCF10A cells were cultured in DMEM/F-12, GlutaMAX (Sigma)
supplemented with 5% horse serum, 1% penicillin/streptomycin, 10 µg/ml insulin, 0.5 µg/ml
hydrocortisone, 100 ng/ml cholera toxin and 20 ng/ml epidermal growth factor. MCF10CA1a.cl1
cells were cultured in DMEM/F-12, GlutaMAX (Sigma) supplemented with 5% horse serum, 1%
penicillin/ streptomycin and 10 mM HEPES. Pheonix kidney cells were cultured in DMEM media
(Sigma) supplemented with 10% FBS.
4.4.2 Nuclear Transfection
MCF10A and MCF10CA1a.cl1 cells were stably transfected with histone-2B−eGFP for
visualization of the nucleus. The plasmid histone-2B−eGFP was obtained as a kind gift from Prof.
Giorgio Scita (IFOM, Milan, Italy). Retroviral phoenix cells (ATCC CRL-3213) were used for the
transfection.
69
Briefly, the DNA was prepared for application to phoenix cells by combining 10 µg of DNA with
0.24 M CaCl2 to a total volume of 500 µl in ddH2O. The mixture was added dropwise to 500 µl of
2xHank’s buffered salt solution (Sigma), and left to incubate at room temperature for 5 min.
Chloroquinone was added to each cell culture plate at a concentration of 20 µM. The
H2O/DNA/CaCl2/HBS solution was added gently to the phoenix cells, and incubated for 8 h in a
37oC incubator. At 8 h post-transfection, the media was exchanged for fresh media. 24 h post
transfection, the phoenix cell media was replaced with 5.5 ml of fresh DMEM media, to
concentrate the viral supernatant. 48 h post transfection, the phoenix cell supernatant was collected
and filtered (0.45 µm). Polybrene 8 µg/ml was added to the supernatant, and then this mixture was
added to the target cells. Cells were then incubated for 3 h in a 37oC incubator, and the viral
supernatant collection step was repeated. Cells were left in normal culture media overnight, and
the 48 h step was repeated the next day to enhance transfection efficiency. Cells were then
incubated with 2 µg/ml puromycin to select for transfected cells.
4.4.3 Golgi Transfection
MCF10A and MCF10CA1a.cl1 cells were transiently transfected with Golgi-RFP (CellLight
Golgi RFP, BacMam, Invitrogen) for visualization of the Golgi apparatus during live cell imaging.
Cells were prepared in 6-well dishes. The transfection was conducted in a low volume (~500 µl)
of media to increase transfection efficiency. Cells were treated with BacMam enhancer kit
(Invitrogen) prior to addition of Golgi transfection agent. The Golgi transfection agent was added
at a volume of 4 µl per 500 µl media and incubated with cells for 24 h. Post-incubation, the media
was exchanged for fresh media and cells were imaged.
4.4.4 Device Fabrication
TPP tool (Photonic Professional GT) from Nanoscribe GmbH was used to produce 3D micro-
structures. A CAD model of the desired structures was created and exported for analysis with the
tool native software, Describe.
The structures were fabricated on circular glass cover slides (Thermo Scientific) with dimensions
of 0.13- 0.16 mm thickness and 30 mm diameter. The fabrication was done using 60x Carl Zeiss
optics with oil dipping. OrmoComp resist (Micro Resist Technology) was used. The resist on the
substrate was pre-processed before fabrication by heating to 80oC for 2 min. Following the
70
fabrication, the resist was post-processed by heating to 130oC for 10 min and developed with
OrmoDev (resist specific developer) for another 10 min. The samples were then cleaned by dipping
in 2-propanol. Samples were stored dry at room temperature.
The 3D pores were designed with cross section areas of 16 µm2, 27 µm2, 36 µm2 and 49 µm2 and
aspect ratios (width to height) of 0.1, 0.3, 1 and 3 for each cross section, creating the full 4x4 array
with 16 different pore geometries. The width of the structures was 2.5 µm, and the height of the
structures was 22 µm.
Gratings for the directional migration analysis were produced on cyclic olefin copolymer (COC)
through hot embossing as described in.192 Gratings with grove depth and width, and ridge with of
1 µm were used to maximize contact guidance.
4.4.5 Sample Preparation
3D samples on glass coverslips were prepared for live cell imaging by gluing the samples to the
base of the wells in a 6-well plate. Holes were drilled in the base of the wells to accommodate the
glass sample. The plate and sample were left for 24 h to dry at room temperature, then sterilized
with 100% ethanol the following day. Subsequently, the samples were rinsed with PBS and coated
with 0.1% poly-L-lysine (Sigma) before the addition of media and cells. 30,000 cells/ml were
added to the wells. Cells were cultured in a 37oC incubator overnight, and imaged the next day, or
transfected with Golgi-RFP (Cell Light, Invitrogen) for imaging the subsequent day.
4.4.6 Live Cell Microscopy
Pore penetration videos were acquired using an inverted Nikon-Ti wide-field microscope (Nikon,
Japan) equipped with an Orca R-2 CCD camera (Hamamatsu Photonics, Japan) and an incubation
chamber (Life Imaging Services, Switzerland) to control temperature, CO2, and humidity. Images
were collected using a 20×, 0.45 NA long-distance objective (Plan Fluor, Nikon).
Videos were obtained over a period of 24 h, and images were captured at 20 min intervals. At each
time of measurement, a transmission and a fluorescent image of the nuclei of MCF10A or
MCF10CA1a.cl1 cells were acquired using brightfield and a FITC filter set, respectively. For
Golgi analysis, an additional red fluorescence channel was acquired using the TRITC filter set.
Focal drift during the experiments was avoided using the autofocus system of the microscope.
71
For the 3D rendering of cell nuclei and actin filaments, MCF10A cells or MCF10CA1a.cl1 cells
were cultured on the samples, fixed and stained for actin according to the protocol below (see
“immunostaining”). Fluorescent Z-stacks of the green signal emitted by the cell nuclei (δZ = 0.2
μm) and red signal emitted from fluorescent phalloidin (actin) were collected. A Nikon-Ti spinning
disk confocal microscope (Nikon, Japan) equipped with an Andor DU-888 camera (Oxford
Instruments, United Kingdom) were used. Images were collected with a 60X objective (Apo 60x
Oil λS DIC N2). The resulting stacks were loaded in Imaris 8.0.1 and a scene was created in the
program. Rendering of individual cell nuclei was obtained applying an automatically-detected
image intensity threshold based on the algorithm developed by Ridler and Calvard.193
4.4.7 Fixation of Cells for Scanning Electron Microscopy
For the SEM analysis of pore penetration, samples were washed 3 times with PBS and then
incubated for 1 h with 2.5% glutaraldehyde in 0.15 M pH 7.2 sodium cacodilate at room temp.
The samples were then washed 3 times with 0.15 M pH 7.2 sodium cacodilate and progressively
dehydrated in 50%, 70% and 100% ethanol. The specimens were stored in ethanol, until they were
dried using a critical point dryer (Automegasamdri 915 B, Tousimis). The specimens were sputter-
coated with gold/palladium using a BAL-TEC SCD-050 sputter coater and imaged with a Zeiss
ULTRA 55 scanning electron microscope.
4.4.8 Scanning Electron Microscopy
The geometrical characteristics of the 3D micro-structures were examined by SEM with a Zeiss
ULTRA 55 scanning electron microscope (FIRST Cleanroom Platform, ETH Zurich). First, an
overview picture (top view) of the entire sample was acquired. Afterwards, the magnification was
increased and the SEM stage was tilted in order to visualize the geometry of individual pores. The
cross section of the pores was subsequently measured using the freehand selection tool of ImageJ
(National Institute of Health, USA). The imaging angle of the microscope was used to convert the
measured areas of the titled images to the corresponding actual pore cross section through
geometric projection.
72
4.4.9 Pore Engagement Dynamics
Pore engagement dynamics (penetration, disengagement and impasse events) were recorded
manually using Nikon Instruments Software. The relevant events during pore penetration
experiments were defined as follows:
A penetration event was recorded when the cell migrated through the pore in a unidirectional
manner. A disengagement event was recorded when the cell attempted to pass through a pore (as
visualized as the cell nucleus partially entering the pore), and then released on the same side in a
reversible manner. An impasse event was recorded when the cell nucleus engaged into a pore but
required > 6 h to penetrate or did not disengage during the time series.
The pathfinding index (PI) through the pore is represented as
number of disengagement events (D)
number of penetration events (P)+ number of disengagement events (D)+number of impassse events (I). …....(5)
The engagement time was recorded as the time the cell initiates engagement with the pore, until
the time the cell releases from the pore.
The approximate initial cell confluency, at the beginning of the experiment, within the square
(Figure 10.2.1) for MCF10A is 30 ± 5% and MCF10CA1a.cl1 is 28 ± 3%.
The number of events represents the normalized cell count. The cell count was normalized to the
cell density. Each engagement event was corrected to represent the number of events relative to
the average cell density for both cell lines. This allowed us to account for differences in the cell
concentration along the pore walls.
The cell engagement events were recorded per micro-structure unit (800 µm length, with 24 pores).
The percentage of cells engaging with pores was calculated by dividing the engagement event by
the total number of cells in proximity to the pore walls per micro-structure unit.
4.4.10 Golgi Quantification
Golgi quantification analysis were performed manually using Nikon Instruments Software. The
polarization of the cells was determined by tracking the position of the Golgi relative to the nucleus
over the time lapse images. The position of the Golgi was recorded as back, front or lateral relative
73
to the direction of migration. Cells that were far away from the micro-structures were chosen as
controls as they do not sense the pores. The cell polarization probabilities were determined for
each engagement event as the # of polarized or unpolarized cells divided by the total # of cells
engaging with a particular pore.
Cell polarization was recorded for cells before pore (representing the time point the cell has not
initiated pore engagement), in pore/ engagement (the mid-point time that the cell engages with the
pore) and after pore (the time point after the cell has completed pore engagement).
4.4.11 Pathfinding and Directional Migration on Gratings Quantification
Pathfinding measures were performed using cell tracking software Imaris. The percentage of cells
engaging with low- and high- a.r. pores, and the y-displacement was obtained by tracking the
migration of individual cells over 24 h. The y-displacement represents the migration distance of a
cell in the direction parallel to the rows of micro-structures. The expected number of penetration
events was calculated from the proportion of pores in the pathfinding design. The percentage of
expected number of penetration events through a.r. 0.3 pores was 1/5 = 0.2 (20%).
The correlation function and correlation length were determined using CIV analysis with Matlab
software, based on previous work by Milde et al.189
Directional migration analysis of cells on gratings was analyzed using Nikon Instruments
Software. Cells were tracked over 20 h of imaging and the migration distance and angle between
the migration vector and grating were quantified.
4.4.12 Cell Cycle Duration, Nucleus Diameter Quantification, and Velocity
The cell cycle duration was quantified as the time between birth and division of the same randomly
chosen cell. The cell nucleus diameter was measured using Nikon Instruments Software. The
velocity of the cells was quantified using the particle tracking algorithm of Imaris (Bitplane
Scientific Software, Switzerland). Time-lapse NIS videos were uploaded into Imaris, and the voxel
size and time interval were adjusted before particle tracking.
74
4.4.13 Flow Cytometry
Cells were released from tissue culture dishes using 0.25% trypsin/EDTA (Sigma-Aldrich, US)
and incubated with blocking buffer (PBS + 1% BSA) for 30 min. For each cell line, 2×105 cancer
cells were fixed and permeabilized using 4% paraformaldehyde (Sigma-Aldrich, US) and 0.2%
Triton X (Sigma-Aldrich, US) in PBS, respectively. Cells were then washed with PBS and
suspended in PBS containing 1% BSA and 0.1% Tween20, and incubated with rabbit anti-human
HRas antibody (1:50 dilution; Abcam, US) for 1 h at RT. Cells were washed with 1% BSA in PBS
and stained with a secondary goat anti-rabbit Alex Fluor 647 (1:200 dilution; Abcam, US) for 30
minutes at RT. Cells were washed and resuspended in 1% BSA in PBS. Samples were then injected
into a BD FACS Canto flow cytometer (BD Biosciences, US) and measurements were plotted as
histograms for AF647. A total of 10,000 cells were analyzed per cell line.
Primary antibodies used also include mouse monoclonal neural cell adhesion molecule (Abcam,
US), mouse monoclonal active Cdc42 (New East Biosciences, US), rabbit monoclonal total Cdc42
(Abcam, US), mouse monoclonal Rac1 (Abcam, US), mouse monoclonal RhoA (Abcam, US),
mouse monoclonal active Rac1 (New East Biosciences, US), mouse monoclonal active RhoA
(New East Biosciences, US), rabbit monoclonal vimentin (Abcam, US) and mouse monoclonal
Talin1 (Abcam, US). Secondary antibodies used also include goat anti- mouse Alexa Fluor 647
(Invitrogen, US) and goat anti-rabbit Alexa Fluor 647 (Invitrogen, US).
4.4.14 Immunocytochemistry
Cells were washed with 1% BSA in PBS and fixed with 4% paraformaldehyde solution (Sigma-
Aldrich, US) followed by 0.2% Triton X-100 (Sigma-Aldrich, US) for permeabilization. Cells
were stained with rabbit anti-human HRas antibody (1:50 dilution; Abcam, US) for 1 h at RT in
PBS containing 1% BSA and 0.1% Tween20. Cells were then washed with 1% BSA in PBS and
stained with a secondary goat anti-rabbit Alex Fluor 647 (1:200 dilution; Abcam, US) for 30 min
at RT. Cells were finally washed with 1% BSA in PBS and imaged using a fluorescent Nikon TiE
eclipse microscope. Images were acquired using a 60X objective (Apo 60x Oil λS DIC N2).
Primary antibody used also include mouse monoclonal active Cdc42 (New East Biosciences, US),
and secondary antibody used also includes goat anti- mouse Alexa Fluor 647 (Invitrogen, US).
75
For filamentous actin staining, after fixation the samples were incubated with TRITC-phalloidin
(Sigma, U.S.A.) overnight at 4°C. Subsequently, the samples were rinsed four times for 1 h each
with 5% BSA in PBS and then washed three times (1 h each) in PBS, mounted with Fluoroshield
histology mounting medium (Sigma, USA) and imaged immediately. Α Nikon-Ti spinning disk
confocal microscope (Nikon, Japan) equipped with an Andor DU-888 camera (Oxford
Instruments, United Kingdom) were used. Images were collected with a 60X objective (Apo 60x
Oil λS DIC N2).
4.4.15 Matrigel Invasion Assay
The invasive potential of the cells was assessed using a Matrigel invasion assay (Corning BioCoat,
VWR) with an 8 µm pore-diameter PET membrane. The assay was performed in 6-well dishes.
The bottom side of the membrane was initially coated with 100 µg/ml fibronectin and incubated
for 1 h at RT. The top side of the membrane was coated with 3 mg/ml matrigel (Corning) for
coating overnight at RT. The following day, serum-free media was added to the top of the
membrane for 2 h to re-constitute the matrigel. 1 day prior to the invasion assay, the cells were
serum starved in their respective media with 0.5% horse serum. Cells were added to the top of the
membrane (at a concentration of 2 x 105 cells/ml), and cell culture media (5% horse serum) was
introduced to the bottom well. Cells were cultured in the invasion assay at 37oC for 24 h.
Subsequently, the cells above the membrane were removed with a cotton swab, and the PET
membrane was cut out of the migration chamber. The migrated cells on the bottom side were fixed
with 4% paraformaldehyde, permeabilized with 0.2% Triton X, immunostained with 1:1000
Hoechst dye, and washed with 1% BSA in PBS. The PET membranes were mounted on glass
coverslips for imaging with the Nikon-Ti wide-field microscope (Nikon, Japan). The number of
cells per field of view (0.4 mm2) were recorded.
4.4.16 Data Representation and Statistical Analysis
Boxes in all box plots extend from the 25th to the 75th percentiles, with a line at the median and a
square representing the mean. Error bars associated with box plots represent standard deviation.
The total number of recorded events from three or more independent experiments is shown in the
upper or lower right hand corner of the presented graphs. Statistical comparison of population
means was performed using a nonparametric Smirnov-Kolmogorov test (α = 0.05). All quantitative
measurements reported are expressed as mean ± s.e.m.
76
5 Analysis of Circulating Tumor Cells from Metastatic Castrate Resistant Prostate Cancer Patients Receiving Enzalutamide or Abiraterone
Prostate cancer affects 1.1 million men worldwide and early detection methods can significantly
prolong survival.194 We profiled circulating tumor cells (CTCs) from the blood of metastatic
castrate resistant prostate cancer patients (mCRPC) using the four- zone velocity valley device.
CTCs are captured with magnetic nanoparticles conjugated to EpCAM or NCadherin, and
subsequently immunostained with cytokeratin, NCadherin, androgen receptor and androgen
receptor variant 7 in separate devices. Patients (n=37) were profiled over the course of 148 weeks
while receiving abiraterone or enzalutamide. We demonstrated that cytokeratin positive and
NCadherin positive CTCs are reduced over 148 weeks of treatment. In addition, CTCs shift to later
zones during treatment, representing a phenotypic transition. We observe that androgen receptor
and androgen receptor variant 7 levels remain relatively constant throughout treatment. Overall,
this study enables us to track mCRPC CTCs and classify their metastatic phenotype using various
biomarkers.
This chapter is currently under preparation as a journal publication:
Brenda J. Green, Vivian Nguyen, Eshetu Atenafu, Philip Weeber, Punithan Thiagalingam, Carmen
Tu, Mahmoud Labib, Reza Mohamadi, Aaron Hansen, Anthony M. Joshua, Shana O. Kelley.
Experiments and data analysis were supervised and performed by B.J. G. Sample processing and
analysis was performed by V. N., E. A., P. W., P. T. and C. T. M.L. and R.M.M aided in project
coordination. A.H., A.M.J. and S.O.K. supervised the study and coordinated the clinical blood
collection.
Sponsor: University Health Network
Primary Scientific Investigator(s):
Dr. Shana Kelley, University of Toronto
Dr. Anthony Joshua, Princess Margaret Cancer Centre
77
Prior studies
Previously, Dr. A.M. Joshua demonstrated the ability to sample patients progressing through
abiraterone alone in a similar cohort (Clinicaltrials.gov NCT01857908).
5.1 Study Design
5.1.1 Study Design and Duration
This is a single centre scientific study of the pathophysiology of circulating tumor cells during and
following enzalutamide and/ or abiraterone treatment determined by analysis of peripheral blood
and CTCs. This is not a therapeutic intervention study. This protocol does not determine eligibility
to receive abiraterone or enzalutamide. The study duration is determined by the required patient
numbers and the availability of enzalutamide and abiraterone.
5.1.2 Inclusion Criteria
For entry into the study, the following criteria must be met.
1) Signed Written Informed Consent. Before any study procedures are performed, subjects (or
their legally acceptable representatives) will have the details of the study described to them, and
they will be given a written informed consent document to read. If subjects consent to participate
in the study, they will indicate that consent by signing and dating the informed consent document
in the presence of study personnel.
2) Be suitable for receiving treatment with enzalutamide and/or abiraterone
3) Naïve to other systemic agents, with the exception of bicalutamide, flutamide, ketoconazole,
prednisone and LHRH agents or other therapies at the investigator’s discretion
4) Patients must have histologically or cytologically confirmed adenocarcinoma of the prostate
AND a clinical presentation consistent with metastatic prostate cancer
5) Patients may not receive any other investigational agent during study participation.
6) Patient consents to comply to treatment with enzalutamide or abiraterone as directed by their
physician
5.1.3 Exclusion Criteria
1) Lack of compliance to daily dosing of abiraterone and prednisone OR enzalutamide
78
2) Patient of childbearing potential refusing to use a double barrier method on contraception
5.1.4 Study Discontinuation
Listed below are situations where participants must come off study
• Withdrawal of consent
5.1.5 Study Treatment/ Procedure
Patients were approached who are planned for treatment with enzalutamide and/or abiraterone.
Following consent, baseline peripheral blood was collected for analysis by study staff as outlined
in the study calendar (Table 5.1). Patients will discontinue treatment as clinically warranted and
will undergo biomarker investigation as per protocol. Biomarker evaluations will not be used to
drive clinical decision-making. Patients will be followed and have specimens collected at one
subsequent visit after abiraterone or enzalutamide discontinuation wherever possible (regardless
of reason for discontinuation, which may include disease progression by PCWG2 195 criteria or
undue toxicity).
Table 5.1 Study timetable: Assessment and procedures
Procedure STUDY
VISIT 1
Baseline
visit prior
to
treatment
STUDY
VISIT 2
W12 (+/-
2W)
STUDY VISIT 3
At first and/or
definitive PSA rise
(+/- 2W) (can be
prior to W12)2
STUDY
VISIT 4
At clinical
progression/
planned
change of
treatment
STUDY VISIT 5
(OPTIONAL)
After abiraterone
/Enzalutamide
discontinuation
Informed
Consent X
Inclusion/
Exclusion
Criteria
X
ECOG X X X X X
Study Bloods1 X X X X X Demographics X
1 Study bloods include 2 X 10ml cell save tubes for CTC analysis.
2 For Visit 3, a second sample set can also be taken at the first prostate specific antigen (PSA) rise by PCWG2 criteria
(25% above nadir), in which case, two Visit 3s would be completed.
79
5.1.6 Research Hypothesis
In this study, we propose that CTCs will serve as biomarkers for castrate- resistant prostate cancer.
The CTC profiles can be used to predict enzalutamide/ abiraterone resistance as well as determine
aggressiveness of the prostate cancer. In addition, evaluation of CTCs may guide treatment
titration of abiraterone and enzalutamide in future studies and clinical practice.
5.2 Introduction
Prostate cancer (PCa) affects 1 in 9 men in America; and may be treatable if discovered in the
early stages.196 Androgen deprivation therapy (ADT) is the main treatment approach for advanced
prostate cancer, and this treatment leads to prostate specific antigen (PSA) responses and clinical
improvements in more than 90% of patients.197 However, this therapeutic approach is not
guaranteed to provide a cure, and the majority of patients eventually become castrate resistant.
Metastatic castrate resistant prostate cancer (mCRPC) is defined by disease progression despite
castrate testosterone level changes that occur during ADT. There is usually continued androgen
receptor expression and signaling, thus leading to biochemical, radiological or symptomatic
disease progression.197
The androgen receptor (AR) is a master regulator transcription factor in normal and cancerous
prostate cells. Canonical AR activation requires binding of androgen ligand (testosterone) to the
AR ligand binding domain, translocation to the nucleus, and transcriptional activation of AR target
genes to promote survival of prostate cells.198 Indeed, disease recurrence in mCRPC may be due
to functional adaptations, which allow prostate cells to survive despite low levels of circulating
androgens. Different mechanisms of resistance have been identified, such as AR splice variant
expression, AR gene overexpression, increased expression of proteins acting as transcriptional
coactivators, and upregulation of enzymes involved in AR synthesis. Thus, despite castration
levels of androgen, in CRPC, the AR signaling pathway remains active 197.
Between 2011 and 2012, two new hormonal agents, abiraterone acetate and enzalutamide
demonstrated further overall survival (OS) improvements as second- line therapies for mCRPC
patients (Figure 5.1).197 Abiraterone is a class of androgen- deprivation therapy that selectively
and irreversibly inhibits the CYP17A1 enzyme, which is necessary for the synthesis of testosterone
80
precursors in the adrenal gland.196 Enzalutamide is an androgen receptor (AR) antagonist.
Enzalutamide binds AR with a high affinity and prevents AR translocation and DNA binding.199
Figure 5.1 Androgen Receptor signaling pathways. A) The hypothalamic- pituitary- testicular axis involving
gonadotropin- releasing hormone (GnRH) and luteinizing hormone (LH). Both GnRH and LH result in testosterone
secretion from Leydig cells of the testes. B) Adrenal androgen de novo steroidogenesis. CYP17A1 is inhibited by
abiraterone. C) Prostate conversion of adrenal androgens to dihydrotestosterone (DHT). DHT binds to the androgen
receptor in the cytoplasm, triggering a conformational change that leads to its nuclear translocation. DHT- bound AR
homodimerizes and with co-activators, binds to DNA at cis androgen response elements to activate or repress AR
target genes. Enzalutamide inhibits AR by competing with DHT for binding, blocking nuclear translocation, and
inhibiting DNA and cofactor binding. Reprinted with permission from 196. Copyright © 2015, Springer Nature.
Resistance to both agents is associated with truncated AR-variants lacking the ligand binding
domain (LBD), and are constitutively active receptors that continue to translocate to the nucleus
in the absence of the ligand. AR-variants have been detected in PCa cell lines as well as clinical
samples, including benign, malignant and metastatic tissue, and results from alternative splicing
or non-sense mutations of the human AR gene (Figure 5.2).198, 200
81
Figure 5.2 Androgen receptor exon full-length and splice variant domains. NTD; N-terminal domain, DBD;
DNA- binding domain, LBD; ligand-binding domain, CE; cryptic exons. Reprinted with permission from 200.
Copyright © 1996-2018 MDPI.
The analysis of CTCs is an important capability that may lead to new approaches for early cancer
diagnosis and treatment monitoring.19, 26, 201 PCa CTC counts in the blood are reported in low
concentrations of 5-50 CTCs/7.5ml, which poses a significant challenge for their isolation and
detection.81 Even at this low range, CTCs remain prognostic with the FDA approved EpCAM-
based CellSearch technology.202
In a study involving enzalutamide and abiraterone, multivariate Cox regression showed that prior
chemotherapy, a high baseline CTC count and increasing CTCs at follow-up were independent
predictors of progression- free survival (PFS). The authors found that measuring CTC changes
during treatment is associated with PSA response and can demonstrate therapy effectiveness.203
Likewise, Lorente et al 204 demonstrated in a study with abiraterone after chemotherapy that ≥ 30%
CTC decrease at 4, 8 and 12 weeks was associated with increased survival.
A significant portion of mCRPC patients do not respond to and develop resistance to the novel
anti-androgen agents. Multi-marker CTC platforms may improve our understanding of clonality,
molecular subtypes and drug resistance mechanisms.196, 21 Indeed, in patients with mCRPC, post-
treatment CTC counts were strongly and independently associated with survival, following
abiraterone or enzalutamide. PSA responses (≥ 30% and ≥ 50%) were less frequent in patients with
increasing CTCs at 10-12 weeks.203 It is important to note that a number of patients have low CTCs
despite widespread disease, indicating disease heterogeneity in CTC phenotype or detection.21.
82
During ADT, PCa cells can activate AR splice variants (such as ARV7), and induce EMT through
increased levels of TWIST, N-Cadherin, and vimentin.205 In mCRPC patients, ARV7 is expressed
in bone metastases and can predict poor prognosis.205 ARV7 status was determined previously in
CRPC patients using the mRNA- based AdnaTest. CTC+/ARV7+ patients were more likely to have
Gleason scores ≥ 8, metastatic disease at diagnosis, higher PSA, higher ALP, prior abiraterone/
enzalutamide treatment, prior taxane use or presence of pain.206
By comparison, Miyamoto et al. examined CTCs from mCRPC patients using the CTC iChip.118
Enzalutamide- receiving patients had significant enrichment for non-canonical Wnt signaling
(involved in cytoskeleton remodeling and cell migration) relative to treatment naïve patients.
However, AR abnormalities were not significantly increased among patients receiving
enzalutamide relative to enzalutamide- naïve patients.
In this study, we capture and profile CTCs from a cohort of 37 mCRPC patients using the velocity
valley microfluidic device 80 with whole blood obtained from patients at 0, 9-22, 23-44 and 45-
148 week time intervals (Figure 5.3). Patients received either abiraterone or enzalutamide, and had
no prior treatment with either therapy.
5.3 Results and Discussion
Metastatic castrate resistant prostate cancer patients (n=37) were profiled over the course of 148
weeks (Table 5.2). Primary treatment included prostetctomy (38%), radiation therapy (52%),
brachytherapy (6%) or focal therapy (3%). All patients received prior androgen therapy, including
LHRH agonists, anti-androgens, steroid, or immune therapy. Within this study, patients received
enzalutamide (70%) or abiraterone (30%) (Figure 10.3.1). At the onset of the study, patients
exhibited metastatic disease in the bone (73%) or lymph nodes (40%) (Figure 10.3.2).
83
Table 5.2. Patient demographics
Patient Characteristics All Patients
Unique patients, No. 37
Age, median (range), y 72 (55- 93)
Gleason score, median (range) 7 (6- 9)
Median (range) follow up (weeks)
59 (9- 148)
Primary treatment, No. (%)
Prostatectomy 6 (16)
Radiation 9 (24)
Prostatectomy and radiation 8 (22)
Radiation and focal therapy 1 (3)
Radiation and brachytherapy 1 (3)
Brachytherapy 1 (3)
None 11 (30)
Prior exposure to life-prolonging therapies, No. (%)
LHRH agonists 2 (5)
LHRH agonists and anti- androgens
26 (70)
LHRH agonists and steroid 1 (3)
LHRH agonists and anti- androgens and steroid and/or immune therapy
8 (22)
AR therapy, No. (%)
Enzalutamide 26 (70)
Abiraterone acetate 11 (30)
Metastatic disease, No. (%)
Bone only 18 (49)
LN only 9 (24)
Visceral only 1 (3)
Bone and LN 6 (16)
Bone and visceral and/or LN 3 (8)
Laboratory measures pre-therapy, median (range)
PSA, ug/L 16.41 (0.16- 305.23)
Hgb, g/L 131.5 (107- 155)
ALP, U/L 79 (12- 381)
LDH, U/L 222 (131- 317)
Abbreviations: LHRH agonists, Luteinising Hormone Releasing Hormone, includes Triptorelin, Leuprolide, Goserelin or Degarelix. AA, anti-androgens, includes Bicalutamide, Nilutamide or ARN509. Steriod includes Prednisone. Immune therapy includes Prostvac. LN, lymph node; PSA, prostate specific antigen; Hgb, hemoglobin; ALP, alkaline phosphatase; LDH, lactic dehydrogenase.
84
5.3.1 Velocity valley CTC capture
The velocity valley device is employed to detect heterogeneous subpopulations of CTCs and sorts
cells based on the expression level of an extracellular marker. We used EpCAM- magnetic
nanoparticles to capture mCRPC CTCs. CTCs with high expression of EpCAM are trapped in
early zones 1 and 2, whereas CTCs with low expression of EpCAM are trapped in later zones 3
and 4 (Figure 5.3A). CTCs captured in the velocity valley device are identified using
immunostaining as DAPI+/CK+/AR+/-CD45- or DAPI+/NCad+/CD45- (Figure 5.3D).
High- EpCAM LnCAP cells and low-EpCAM PC3 cells are profiled in the velocity valley device
(Figure 5.3B). LnCAP cells are non-tumorigenic epithelial prostate cancer cells derived from
lymph node metastases. PC3 cells are tumorigenic mesenchymal cells derived from bone
metastases.207 We observe that LnCAP cells are captured in zone 1 and 2 of the device, while PC3
are captured in zone 3 and 4 (Figure 5.3B).
The capture efficiency of the velocity valley device is compared to commercially available
CellSearch technology. The velocity valley device successfully captures LnCAP and PC3 cells
spiked in blood with high efficiencies of 90-95%. In comparison, CellSearch exhibits lower
capture efficiency for LnCAP of 80%, and significantly lower capture efficiency of 40% for
mesenchymal PC3 cells (Figure 5.3C).
Previously, it has been reported that CellSearch cannot detect approximately 30% of CTCs from
metastatic PCa patients.202 It was proposed that this lack of detection is due to EMT; which results
in down-regulation of epithelial markers necessary for CTC capture and enumeration. Thus, the
standard CellSearch capture definition may be missing the most invasive and highly metastatic
cells driving disease progression. In contrast, the velocity valley device sorting approach captures
low-EpCAM cells with high efficiency.
85
D DAPI CK AR CD45 Combined
10µm PC
a C
TC
W
BC
E F
86
Figure 5.3. Capture and Analysis of mCRPC CTCs receiving Enzalutamide or Abiraterone. A) Schematic of
patient sample collection and processing through the velocity valley device. Briefly, whole blood is incubated with
magnetic nanoparticles (MNPs) conjugated to EpCAM and introduced into the device at a flow rate of 600µl/h in the
presence of an external magnetic field. CTCs are trapped in different zones based on their expression level of EpCAM.
B) LnCAP and PC3 cells were profiled with the velocity valley device. High- EpCAM LnCAP cells are trapped in
zone 1, whereas low- EpCAM expressing PC3 cells are trapped in zones 3 and 4. C) Capture efficiency of LnCAP
and PC3 cells in the velocity valley device compared to CellSearch. Statistics are performed with two-tailed t-test,
*p<0.05. D) Immunostaining images of a white blood cell and a prostate cancer CTC. CTCs are identified as
DAPI+/CK+/AR+/-CD45- or DAPI+/NCad+/CD45-. E) NCadherin- positive CTCs and PSA profile for a progressive
patient receiving abiraterone. CTC counts per ml of blood are shown in the inset for all zones. F) NCadherin-CTC and
PSA profile for a responsive patient receiving enzalutamide. CTC counts per ml of blood are shown in the inset for
all zones. CTC trends are shown as a red line, whereas the PSA level is shown as a black line.
CTC detection from mCRPC patients was compared with the commercially available CellSearch
technology over the treatment period (Figure 10.3.6). We observed that progressive patients
exhibited elevated CTCs/ml (2.4 ± 0.7 CTCs/ 7.5 ml) relative to responsive patients (0.8 ± 0.2
CTCs/ 7.5 ml). However, the majority (91%) of CTCs fell below the clinically relevant threshold
of 5 CTCs/ 7.5 ml.208
Patient CTCs are profiled over the treatment period for PSA- progressive (PSA increase that is ≥
25% and ≥ 2 ng/mL above the nadir sustained for 3 weeks) versus PSA- responsive patients (>50%
decline from baseline measured twice 3 to 4 weeks apart) (Figure 5.3E-F).208
5.3.2 Enzalutamide and Abiraterone Treatment Reduce CTCs
The dynamic change of CTC counts over disease progression is associated with a significant
prognostic effect.209, 210
Cytokeratin- positive and NCadherin- positive mCRPC CTCs were captured in later zones 3 and
4 of the velocity valley device, representing low- EpCAM phenotypes (Figure 5.4A-D).
Previously, it has been reported that CTCs are prevalent in 69% of mCRPC patients at the initiation
of a new line of endocrine therapy, using CellSearch.203
Consistent with these results, we observe high baseline CTC counts of 10.3 ± 2.1 cytokeratin+
CTCs/ml, and 11.5 ± 3.1 NCadherin+ CTCs/ml for patients receiving enzalutamide or abiraterone.
Treatment subsequently caused a reduction in CTCs over the course of 148 weeks (5.2 ± 2.4
cytokeratin+ CTCs/ml and 5.9 ± 3.2 NCadherin+ CTCs/ml) (Figure 5.4A-D).
87
88
High baseline CTC counts were observed in 64% of patients receiving enzalutamide or
abiraterone. Elevated baseline counts are defined as counts greater than the false positive detection
rate of 2 cells/ml (Figure 10.3.4).
Decreasing CTC counts during therapy remains an important and independent biomarker of
survival in context of endocrine therapy. Patients were grouped according to CTC reduction <50%,
CTC stagnation (no change), and CTC increase >50% relative to the baseline count, for both
cytokeratin- positive and NCadherin- positive CTCs (Figure 5.4E) over 148 weeks of enzalutamide
or abiraterone treatment. The majority of patients exhibited CTC reduction (62.1% and 51.4% for
cytokeratin- positive and NCadherin- positive CTCs, respectively). A lower proportion of patients
exhibited CTC increase (21.6% and 24.3% for cytokeratin- positive and NCadherin- positive
CTCs, respectively) and the remaining patients did not show any change in CTC counts (16.2%
and 24.3% for cytokeratin- positive and NCadherin- positive CTCs, respectively).
5.3.3 Zone Distributions of Patient CTCs
The zone profile of CTCs can provide indication whether the cells undergo phenotypic change
over the course of treatment. Thus, CTC distributions from patients receiving abiraterone or
enzalutamide were recorded over 148 weeks (Figure 5.5). The percentage of low- EpCAM (zone
4) cytokeratin- positive CTCs increased significantly over the treatment period (from 57.6% to
70.5% between baseline and 9-22 weeks, respectively and from 57.6% to 69.1% between baseline
and 23-148 weeks, respectively) (Figure 5.5A,B). This suggests that CTCs are shifting towards a
lower- EpCAM phenotype over the treatment period.
Figure 5.4. Cytokeratin CTC profiles for Enzalutamide and Abiraterone- treated Patients. A)- B) Cytokeratin-
positive CTC profiles for patients receiving enzalutamide or abiraterone (data grouped together). CTCs are separated
based on Zone1+Zone2 and Zone3+Zone4 populations. CTC profiles are shown for 0, 9-22, 23-44 and 45-148 weeks
post- treatment. Box plots represent standard error of the mean. The mean is shown as the central square, with the
median depicted as a line. Each dot represents a patient. C)- D) Cytokeratin- positive CTC profiles for patients
receiving enzalutamide or abiraterone (data grouped together). CTCs are separated based on Zone1+Zone2 and
Zone3+Zone4 populations. CTC profiles are shown for 0, 9-22, 23-44 and 45-148 weeks post- treatment. Box plots
represent 25th and 75th percentile. Error bars represent standard deviation. Statistics were performed using two-tailed
t-test *p<0.05. E) Number of patients with CTC reductions from baseline over the course of 148 weeks. Data is shown
for cytokeratin- positive and NCadherin- positive CTCs. CTCs were identified as DAPI+/CK+/CD45- or
DAPI+/NCad+/CD45-. CTCs were captured with EpCAM- MNPs.
89
Responsive patients exhibited an enhanced shift relative to progressive patients (Figure 5.5C). This
observation potential provides a correlation between PSA responsiveness and reduction of high-
EpCAM CTCs.
Enzalutamide can promote EMT in prostate cancer by enhancing Snail expression through
inhibition of the AR signaling pathway. In a patient- derived xenograft model, a study
demonstrated that both N-Cadherin and vimentin are elevated after ADT.211, 205 Consistent with
these results, we observe that mCRPC CTCs undergo a phenotypic shift towards low-EpCAM
zones during treatment (Figure 5.5 A,B).
The proportion of androgen receptor (AR) positive CTCs relative to cytokeratin- positive CTCs
remains relatively constant during treatment (Figure 5.5D), and was on average 84.8 ± 1.8%. These
expression levels correspond with literature, which reports AR transcript levels in 78% of CTCs
from mCRPC patients.118
The ARs in tumors exposed to ADT undergo selective alterations leading to aberrant AR activation
which allows the AR pathway to remain active.205, 197 AR overexpression was detected in
approximately 30% of CRPCs but was not observed in treatment- naïve prostate cancers.205
Patients received first-line hormone therapy prior to enzalutamide and abiraterone, that is
composed of anti-androgens and leutenizing hormone - releasing hormone (LHRH) agonists.
Patients received anti-androgens (Triptorelin, Leuprolide, Goserelin, Degarelix) for an average
period of 1.69 ± 0.35 years (n=37) prior to initiation of enzalutamide or abiraterone. Concurrently,
patients received LHRH agonists for a period of 5.84 ± 0.58 years (n=37) prior to initiation of
enzalutamide or abiraterone (Table 10.3.1). Given that these patients have received significant
prior first- line ADT, we hypothesize that enzalutamide or abiraterone do not cause a further
amplification of AR.
Altogether, examining zone profiling of mCRPC CTCs over 148 weeks of treatment provides
evidence of a phenotypic shift towards reduced epithelial protein levels while AR levels do not
change significantly.
90
Figure 5.5. Zone profiling of low- EpCAM CTCs over treatment period.
A) Percentage of zone 4 cytokeratin positive CTCs relative to total CTCs. CTCs are DAPI+/CK+/CD45-. Box plots
represent 25th and 75th percentile. The mean is shown as the central square, with the median depicted as a line. Error
bars represent the standard deviation. Statistics are performed with two-sample t-test. *p<0.05 is significant. B)
Normal distribution of CTCs over the treatment period of 148 weeks. C) Percentage of zone 4 cytokeratin positive
CTCs relative to total CTCs for progressive versus responsive patients. CTCs are DAPI+/CK+/CD45-. Error bars
represent standard error of the mean. D) Percentage of androgen- receptor positive CTCs in zone 4. CTCs are
DAPI+/CK+/AR+/CD45-. Percentage of AR+ CTCs are reported relative to total CK+ CTCs. Distributions are obtained
at 0, 9-22 and 23-148 weeks. CTCs were captured with EpCAM- MNPs. Error bars represent standard error of the
mean. Statistics are performed with two-tailed t-test, *p<0.05. CTCs count >2 CTCs/ml were considered for zone
profiling analysis. Patients receive enzalutamide or abiraterone.
91
5.3.4 EpCAM Capture versus NCadherin Capture
CTCs are known to be highly heterogeneous; therefore, multi-plexed capture strategies are
advantageous.59, 60 We examined the capture profiles of mCRPC CTCs by comparing capture with
EpCAM- magnetic nanoparticles (MNPs) and NCadherin- MNPs. Capture profiles are validated
with prostate cancer cells LnCAP and PC3 (Figure 5.6A-C, Figure 10.3.5). The velocity valley
device captures high-EpCAM LnCAP cells in zone 1, and low-EpCAM PC3 cells in zone 3 and 4
with EpCAM- MNPs (Figure 5.3B). In comparison, LnCAP and PC3 cells are captured in zone 3
and 4 with NCadherin- MNPs (Figure 5.6C).
mCRPC CTCs were captured in zones 3 and 4 for both EpCAM- capture and NCadherin capture
approaches, suggesting that the cells express low levels of EpCAM and NCadherin (Figure 5.6D).
The mCRPC CTC zone profiles resemble that of PC3 cells, exhibiting low-EpCAM and low-
NCadherin profiles. Patient CTCs captured with NCadherin- MNPs were tracked over 148 weeks
(Figure 5.6E).
92
Figure 5.6. EpCAM- capture versus NCadherin- capture of mCRPC CTCs. A,B) Flow cytometry analysis of
EpCAM and NCadherin in LnCAP and PC3 prostate cancer cells. 10,000 cells were analyzed per sample. C) LnCAP
and PC3 cells profiled with the velocity valley device and captured with NCadherin- MNPs. D) Cytokeratin positive
CTCs profiled with EpCAM- MNPs and NCadherin- MNPs. Patient received enzalutamide or abiraterone, and data
was plotted together. Enzalutamide- and abiraterone- treated patient CTCs profiled with EpCAM- MNPs and
NCadherin- MNPs. CTCs are identified as DAPI+/CK+/CD45-.
5.3.5 Analysis of Androgen Receptor Variant 7
AR variant expression has been associated with ADT resistance.212, 213 Here, we valitaed ARV7
protein expression with prostate cancer cells DU145 and VCaP. VCaP cells are prostate cancer
cells isolated from bone metastases and have high protein levels of both AR and ARV7, whereas
androgen- independent DU145 cells express low levels of both AR and ARV7 214, 215 (Figure
5.7A,B).
Next, we examined ARV7 CTC expression on a set of patients (n=21) receiving abiraterone or
enzalutamide. ARV7 CTC counts are compared to AR- full length CTC counts obtained using
separate devices. CTCs were captured with EpCAM MNPs and co-stained with cytokeratin and
AR or ARV7 (Figure 5.7C). Previously it has been demonstrated that PCa CTCs obtained from
patients who had received prior ADT or chemotherapy had a high incidence (43%) of AR splice
variant expression (ARV7, ARV12, ARV1, ARV3, ARV4).118 In accordance with this data, we
93
observed that mCRPC patient CTCs have an ARV7 incidence of 52.4% (11/21 CTCs profiled) at
baseline.
AR- positive CTCs and corresponding ARV7- positive CTCs significantly decreased over
treatment period (from 11.7 AR+CTCs/ml to 4.4 AR+CTCs/ml and from 2.5 ARV7+CTCs/ml to
1.0 ARV7+ CTCs/ml between baseline and 22-148 weeks, respectively) (Figure 5.7D). The ratio
of ARV7 CTCs to AR-full length CTCs does not change significantly over ADT, and is
approximately 17.5 (Figure 5.7E). These results are consistent with AR expression levels, which
did not vary during ADT (Figure 5.5D).
Overall, ARV7+ CTCs are reported in baseline mCRPC patient samples, and we observe a
significant reduction in AR+ and ARV7+ CTCs over the treatment period.
DU
145
ARV7 DAPI Combined
VC
aP
10µm 10µm
DU
145
AR DAPI Combined V
Ca
P
10µm
DAPI CK ARV7 CD45 Combined C
A B
PC
a C
TC
94
Figure 5.7. Androgen Receptor Variant 7 Profiling of mCRPC CTCs A,B) Immunostaining of DU145 and VCaP
cells with antibodies against Androgen Receptor (AR) and Androgen Receptor Variant 7 (ARV7). C) An
immunostained image of a mCRPC patient CTC. CTC is DAPI+/CK+/ARV7+CD45-. D) Cytokeratin- positive CTCs
profiled at 0 and 22-148 weeks post treatment with abiraterone or enzalutamide. AR+ CTCs are compared to ARV7+
CTCs, and are co-stained with cytokeratin in separate devices. Dots represent individual patient CTCs. E) Ratio of
ARV7+ to AR+ CTCs at 0 and 22-148 weeks post treatment with abiraterone or enzalutamide. CTCs are co-stained
with cytokeratin. Statistics were performed with Mann Whitney test. *p<0.05 is considered significant.
5.4 Conclusion
CTCs were examined from mCRPC patients receiving enzalutamide or abiraterone at multiple
time points over 148 weeks. CTCs were sorted into four zones of the velocity valley device, based
on their surface level expression of epithelial or mesenchymal markers. We discovered that
mCRPC CTCs expressed low levels of EpCAM and NCadherin, consistent with capture in later
zones. High baseline levels of CTCs were observed in 64% of patients receiving enzalutamide or
abiraterone. Treatment caused a >50% reduction in cytokeratin positive CTCs in 62.1% of patients.
The velocity valley device stratifies CTCs expressing high- and low- levels of a surface marker.
We observe that the percentage of low-EpCAM CTCs increased during treatment. This phenotypic
shift suggests that the tumor cells remaining in circulation after enzalutamide or abiraterone
treatment experience a reduction in epithelial properties relative to baseline. Despite phenotypic
changes in CTCs, 78% of patients receiving enzalutamide or abiraterone had either a >50%
reduction or no change in CTC counts over the treatment period.
Studies involving CRPC patients have demonstrated that AR variants are often expressed in
metastases. High levels of these variants are associated with faster disease progression.212 We
observe that CTCs profiled over 148 weeks maintained a relatively constant proportion of AR and
ARV7 protein levels on cytokeratin- positive CTCs. In addition, AR+ and ARV7+ CTCs are
significantly reduced over the treatment period.
Overall, we apply the capture and sorting microfluidic approach for mCRPC patients in a
longitudinal study, and importantly demonstrate the ability to profile heterogeneous CTCs over
the treatment period.
95
5.5 Methods
5.5.1 Cell Culture
VCaP cells (ATCC catalog number CRL-2876). VCaP cells are a prostate cancer cell line. They
have epithelial morphology and they are adherent cells. The cells were cultured in DMEM medium
(ATCC catalog number 30-2002) supplemented with 10% FBS in T-75 flasks, at 37°C and
atmosphere containing 5% CO2. Human prostate cancer cells, PC3M and LnCAP were obtained
from Dr. Alison Allan, London Health Sciences Centre, London, ON. PC3 cells were cultured in
F12K media (ATCC) supplemented with 10% FBS. LnCAP cells were cultured in RPMI media
(Gibco) supplemented with 10% FBS. Cells were grown at 37°C and 5% CO2.
5.5.2 Device Fabrication
Microchips were fabricated by rapid prototyping using poly(dimethylsiloxane) (PDMS) soft-
lithography starting with an SU-8 master on a silicon wafer (University Wafer, MA). A PDMS
(Dow Chemical, MI) replica of the master was formed. After peeling the replica, holes were
pierced for tubing connections. The replica was permanently sealed with a PDMS-coated glass
slide. Bonding was enhanced and made irreversible by oxidizing both the replica and the cover in
a plasma discharge for 1 min prior to bonding. Silicone tubing was then added at the inlet and the
outlet. The channel depth was 100 μm. PDMS chips were conditioned with Pluronic F68 Sigma
(St. Louis, MO) to reduce sample adsorption and washed with PBS pH=7.4 before use using a
syringe pump (Chemyx, TX). Two arrays of NdFeB N52 magnets (KJ Magnetics, PA),1.5 mm
diameter and 8 mm long, were placed on both the bottom and top surfaces of the capture zones in
the chip for the duration of the cell capture process.
5.5.3 Cell Capture and Preparation
CTC analysis was performed using cells in PBS buffer, cells spiked in blood, and CTCs obtained
from patient samples. 100 cells were introduced into 1% BSA in PBS buffer for analysis in the
velocity valley device.
Patients (n=37) were recruited from the Princess Margaret Hospital according to the University’s
Research Ethics Board approved protocol. 20ml of blood were collected in two CellSearch tubes
that contained anticoagulant EDTA (Johnson and Johnson). One tube of blood was shipped to the
London Regional Cancer Program at the London Health Sciences Centre for CellSearch analysis,
96
and the second tube was analysed using the velocity valley device. All blood samples were
analyzed within 24 hours from sample collection. 10μl of anti-EpCAM Nano-Beads (MACS) were
added to 1 ml of blood/ or cell-suspension and incubated and mixed for 30 minutes at room
temperature. Nanobeads attach to EpCAM- expressing cells. NCadherin- conjugated nanobeads
are prepared by incubating NCadherin (0.5 mg/ml) (Abcam) and anti-biotin nanobeads (MACS)
with cells or 1ml of blood for 30 minutes at room temperature. During this incubation time, the
magnetic nanobeads are attached to NCadherin-expressing cells. Microfluidic devices perfused
with pluronic acid were prepared, and washed with PBS. In the case where we analyzed cell lines,
100 cells per chip were prepared in 1% BSA in PBS. Samples are introduced into the velocity
valley device at 600µl/h using a syringe pump. Next 200 μl PBS-EDTA at 600µl/h was introduced
to remove non-target cells. After this step, chips were immunostained as detailed below.
5.5.4 Cell Capture Efficiency
CTC capture efficiency is quantified for cells spiked in whole blood and captured in the velocity
valley device or using CellSearch. 20 cells were spiked into 1ml of healthy blood.
𝐶𝑎𝑝𝑡𝑢𝑟𝑒 𝐸𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑐𝑦 = Number of Cells Counted
Number of Cells Spiked into Blood ……………………………………..(6)
5.5.5 Progressive vs. Responsive Categorization
mCRPC patients were categorized as PSA progression or PSA responsive according to PCWG3
criteria 195. (Figure 10.3.3).
PSA response is defined as a >50% decline from baseline measured twice 3 to 4 weeks apart.
PSA progression: After decline from baseline: record time from initiation of therapy to first PSA
increase that is ≥ 25% and ≥ 2 ng/mL above the nadir, and which is confirmed by a second value
3 or more weeks later (ie, a confirmed rising trend). If there is no decline from baseline: PSA
progression is defined as ≥ 25% and ≥ 2ng/mL after 12 weeks.
5.5.6 Velocity Valley Immunostaining
After processing the blood, cells were fixed with 4% paraformaldehyde, and subsequently
permeabilized with 0.2% Triton X-100 (Sigma-Aldrich) in PBS. Cells were immunostained with
primary antibodies, biotin monoclonal Anti-Cytokeratin 18 (Lifespan), biotin monoclonal Anti-
97
NCadherin (Abcam), Androgen Receptor Alexa Fluor 555 (Cell Signaling), ARV7 Alexa Fluor
555 (Precision Antibody), CD45- APC (MACS) followed by secondary antibody Yellow-nanoB-
Avidin (Invitrogen) (1:500) to visualize the CTCs. All of the antibodies were prepared in 100 μl
PBS plus 1% BSA and chips were stained for 30 minutes at a flow rate of 200µl/h. Chips were
washed between each staining step using 200 μl 0.1% Triton X-100 in PBS, at 600µl/h for 10min.
Nuclei were stained with 100 μl DAPI ProLong Gold reagent (Invitrogen, CA) at 600µl/h. After
completion of staining, all devices were washed with PBS and stored at 4 °C before scanning.
Immunostained cells were imaged using a fluorescent Nikon TiE eclipse microscope with an
automated stage controller and an Andor camera and images were acquired with NIS Elements
(Nikon) using a 10X and 50X objective.
After immunostaining, devices were scanned using a 10X and 50X objective and a fluorescent
Nikon TiE eclipse microscope with an automated stage controller and an Andor camera. Bright
field as well as DAPI, FITC, TRTIC, and Cy5 channels were acquired with NIS Elements (Nikon),
and target cells were counted.
5.5.7 Flow Cytometry
Cells were harvested from tissue culture using 0.25% trypsin/EDTA (Sigma-Aldrich, US) and
incubated with blocking buffer (PBS + 1% BSA) for 30 minutes. For each cell line, 5×105 cancer
cells were fixed and permeabilized using 4% paraformaldehyde (Sigma-Aldrich, US) and 0.2%
Triton X (Sigma-Aldrich, US) in PBS, respectively. Cells were then washed with PBS and
suspended in PBS containing 1% BSA and 0.1% Tween20, and incubated with anti-EpCAM Alexa
Fluor 647 (BioLegend, US), NCadherin- FITC (Cell Signaling), Androgen Receptor Alexa Fluor
647 (Cell Signaling) and ARV7 Alexa Fluor 647 (Precision Antibody) at 1:50 dilution and stained
at room temperature for 30 minutes. Samples were washed with PBS and re-suspended in 1% BSA
in PBS. Samples were injected into a BD FACS Canto flow cytometer (BD Biosciences, US) and
measurements were plotted as histograms for each fluorophore (AF647 and FITC). A total of
10,000 cells were analyzed per cell line.
5.5.8 Statistics
Statistics were performed with two-sampled t-test for normally distributed populations and Mann
Whitney test for non-parametric populations. p<0.05 is considered statistically significant.
98
6 Conclusions and Future Outlook
6.1 Conclusions
Analysis of tumor cells that enter the circulation may allow tumors to be characterized non-
invasively and profiled in real-time as treatment is administered. The rarity and heterogeneous
biology of these cells present a significant challenge to their isolation, identification and
characterization. New devices and materials that have emerged recently provide valuable tools that
will allow more information to be extracted from these cells.
We presented a microfluidic device, which traps CTCs in zones based on their surface expression
of EpCAM, NCadherin or HER2. This approach enabled us to stratify low- and high- risk CTCs
from the blood, providing greater diagnostic potential compared to simple CTC enumeration.
We profiled CTCs from 37 mCRPC patients receiving abiraterone or enzalutamide over 148
weeks. The majority of CTCs were trapped in low- EpCAM and low- NCadherin zones of the
microfluidic device; indicating that these cells in vivo express low levels of EpCAM and
NCadherin. Androgen deprivation therapy caused a shift towards lower EpCAM expression over
treatment period, in both cytokeratin- CTC populations. Despite this phenotypic change, 78% of
patients receiving ADT had either a >50% reduction or no change in CTC counts over treatment.
Androgen receptor (AR) and AR variant 7 (ARV7) expression levels on cytokeratin positive CTCs
did not change over treatment period, suggesting that the phenotypic changes occur in a
mechanism independent of AR overexpression or AR variant overexpression.
It has become increasingly clear that simply capturing and counting tumor cells in the bloodstream
may not provide the information required to advance our understanding of the biology of these
rare cells, or to allow us to better exploit them in medicine. Thus, we applied the microfluidic
sorting device to examine functional properties of breast cancer cells. Cells isolated from low-
EpCAM zones ingested more collagen and had higher NAD(P)H autofluorescence levels relative
to cells isolated from high- EpCAM zones. This effect was mimicked in a CoCl2- induced EMT
model of breast cancer, where we observed a phenotypic shift in CTC profiles as the cells
transitioned to mesenchymal state. This study highlights that cells transitioning through EMT may
adopt more invasive behaviors, contributing to metastasis.
99
Cancer cell migration is a key component during metastasis. Methods that extend beyond
biomarker analysis of CTCs may reveal key tendencies of invasive cells. Cell invasion
incorporates navigation through a complex 3D matrix, either as single cells or collective units.
Cells can digest their matrix through proteolytic cleavage to create tracks of least resistance for
follower cells, or they can deform and compress through narrow openings.
We fabricated 3D silicon- oxide pores using laser nanolithography to mimic the extracellular
matrix, and monitor migration of breast cancer cells in the absence of chemical cues. Pore cross
sections ranged from 16- 49 µm2, with aspect ratios (width/ height) of 0.1- 3. Cancer cells preferred
to pass through tall, narrow openings rather than flat openings, due to lateral compressive forces
generated by actin filaments. We found that invasive breast cancer cells were able to disengage
from unfavorable pores at a higher frequency compared to non- invasive cells. Invasive cells had
higher levels of cell polarization initiator, Cdc42, and this could lead to more efficient pathfinding
strategies.
To summarize, we have presented microdevices for application of CTC biomarker and phenotypic
profiling, and for elucidating the complex migratory behaviors of single tumor cells and clusters.
Continued progress will advance our understanding of mechanisms of cancer metastasis, which is
fundamentally important for combating this disease.
100
6.2 Future Outook
This work demonstrates the applicability of microdevices for advancing our understanding of
metastasis. The emphasis of this thesis is on the design and application of microdevices devices to
analyze the heterogeneity in cancer cell subpopulations and to investigate cancer cell migration.
The velocity valley microfluidic device was used to profile CTCs from metastatic castrate resistant
prostate cancer patients and from rat lung metastases models. This analysis has largely focused on
detection of single tumor cells; however, clusters of tumor cells have 20- to 50-fold increased
capacity for metastasis.216 Clusters of tumor cells may contain extracellular matrix, stromal cells,
epithelial cells and myeloid cells. Components within a cluster, such as platelets, provide
protective factors for the tumor cells.216 Clusters also prevent tumor cells from programmed cell
death, promoting their survival in circulation. In a clinical trial, we detected prostate cancer clusters
in the range of 1-5 clusters per ml in the velocity valley device. Clusters were often composed of
a single tumor cell attached to 1-2 white blood cells. Capture of CTC clusters presents significant
challenges, as they may break apart during capture. Thus, future work could involve development
of microfluidic devices that capture CTC clusters with high efficiency. External magnetic
nanoparticle capture approaches for clusters may lack sensitivity due to the low surface area to
volume ratio of clusters compared to single cells. As such, intracellular CTC capture strategies
have been developed based on hybridized magnetic nanoparticles.217 This approach would enable
selective targeting of cluster mRNA sequences within fluidic environments.
CTCs examined from metastatic castrate resistant prostate cancer patients are captured in low-
EpCAM zones of the velocity valley device. Highly persistent subsets of prostate cancer cells are
reported to have stem cell and neuroendocrine properties.4 Future work could involve applying the
velocity valley device towards profiling cancer cells based on stem cell properties, to identify
resistant subtypes.
During microfluidic capture of CTCs, white blood cells (WBCs) are trapped non-specifically
within the velocity valley device. Despite coating the microfluidic channels with 0.1% pluronic,
approximately 1000- 4000 WBCs are trapped per ml of blood. This represents a significant
depletion, as they are reduced from an original concentration of approximately 1,000,000 WBCs
per ml. However, non-specific WBCs may interfere with downstream CTC assays. Future work
101
could involve modification of the inner surface of the microfluidic device with a coating that repels
non-specific cell interaction.
CTCs represent the visible cells in the circulation during metastasis.26 A subsequent aspect of
metastasis includes cancer cell migration through ECM.152 Here we applied micro-structured
porous arrays for migration analysis of breast cancer cells. The OrmoComp silicon oxide micro-
structures are biocompatible, stiff substrates which provide the basis for the migration assay.
Future experiments could involve coating the micro-structures with type I collagen such that the
cells are required to digest their micro-environment to successfully penetrate through a pore. This
setup would enable the detection of mesenchymal migration.
Through the migration analysis, we identified pathfinding tendencies of invasive MCF10CA1a.cl1
cells relative to non-invasive MCF10A cells. Invasive breast cancer cells disengaged from pores
at a higher rates and frequency compared to their non- invasive counterpart. In addition, invasive
cells were able to migrate and move through low aspect ratio pores when presented with a barrier
array. Experiments that focus on disrupting the pathfinding capability of invasive tumor cells
would be invaluable towards cancer therapeutics. Nocodazole is a microtubule- active drug that
interferes with apical protein delivery in cells.241 Targeting compounds such as nocodazole
towards invasive cancer cells may reduce the incidence of metastasis.
Overall, the presented microdevices represent valuable means to identify subpopulations of CTCs
and explore invasive potential of tumor cells.
102
7 References
[1] L. A. Torre, R. L. Siegel, E. M. Ward, and A. Jemal, "Global Cancer Incidence and
Mortality Rates and Trends--An Update," Cancer Epidemiol Biomarkers Prev, vol. 25, pp.
16-27, Jan 2016.
[2] A. Kowalik, M. Kowalewska, and S. Gozdz, "Current approaches for avoiding the
limitations of circulating tumor cells detection methods-implications for diagnosis and
treatment of patients with solid tumors," Transl Res, vol. 185, pp. 58-84 e15, Jul 2017.
[3] J. M. Pawelek and A. K. Chakraborty, "Fusion of tumour cells with bone marrow-derived
cells: a unifying explanation for metastasis," Nat Rev Cancer, vol. 8, pp. 377-86, May 2008.
[4] B. G. Hollier, K. Evans, and S. A. Mani, "The epithelial-to-mesenchymal transition and
cancer stem cells: a coalition against cancer therapies," J Mammary Gland Biol Neoplasia,
vol. 14, pp. 29-43, Mar 2009.
[5] E. A. Klein, M. R. Cooperberg, C. Magi-Galluzzi, J. P. Simko, S. M. Falzarano, T.
Maddala, J. M. Chan, J. Li, J. E. Cowan, A. C. Tsiatis, D. B. Cherbavaz, R. J. Pelham, I.
Tenggara-Hunter, F. L. Baehner, D. Knezevic, P. G. Febbo, S. Shak, M. W. Kattan, M.
Lee, and P. R. Carroll, "A 17-gene assay to predict prostate cancer aggressiveness in the
context of Gleason grade heterogeneity, tumor multifocality, and biopsy undersampling,"
Eur Urol, vol. 66, pp. 550-60, Sep 2014.
[6] M. S. Litwin and H. J. Tan, "The Diagnosis and Treatment of Prostate Cancer: A Review,"
JAMA, vol. 317, pp. 2532-2542, Jun 27 2017.
[7] E. S. McDonald, A. S. Clark, J. Tchou, P. Zhang, and G. M. Freedman, "Clinical Diagnosis
and Management of Breast Cancer," J Nucl Med, vol. 57 Suppl 1, pp. 9S-16S, Feb 2016.
[8] M. Reck and K. F. Rabe, "Precision Diagnosis and Treatment for Advanced Non-Small-
Cell Lung Cancer," N Engl J Med, vol. 377, pp. 849-861, Aug 31 2017.
[9] R. Gasparri, R. Romano, G. Sedda, A. Borri, F. Petrella, D. Galetta, M. Casiraghi, and L.
Spaggiari, "Diagnostic biomarkers for lung cancer prevention," J Breath Res, vol. 12, p.
027111, Feb 6 2018.
[10] E. Racila, D. Euhus, A. J. Weiss, C. Rao, J. McConnell, L. W. Terstappen, and J. W. Uhr,
"Detection and characterization of carcinoma cells in the blood," Proc Natl Acad Sci U S
A, vol. 95, pp. 4589-94, Apr 14 1998.
[11] M. Yu, A. Bardia, B. S. Wittner, S. L. Stott, M. E. Smas, D. T. Ting, S. J. Isakoff, J. C.
Ciciliano, M. N. Wells, A. M. Shah, K. F. Concannon, M. C. Donaldson, L. V. Sequist, E.
Brachtel, D. Sgroi, J. Baselga, S. Ramaswamy, M. Toner, D. A. Haber, and S. Maheswaran,
"Circulating breast tumor cells exhibit dynamic changes in epithelial and mesenchymal
composition," Science, vol. 339, pp. 580-4, Feb 1 2013.
[12] W. Gao and O. C. Farokhzad, "Self-propelled microrockets to capture and isolate
circulating tumor cells," Angew Chem Int Ed Engl, vol. 50, pp. 7220-1, Aug 1 2011.
103
[13] Y. T. Lu, L. Zhao, Q. Shen, M. A. Garcia, D. Wu, S. Hou, M. Song, X. Xu, W. H. Ouyang,
W. W. Ouyang, J. Lichterman, Z. Luo, X. Xuan, J. Huang, L. W. Chung, M. Rettig, H. R.
Tseng, C. Shao, and E. M. Posadas, "NanoVelcro Chip for CTC enumeration in prostate
cancer patients," Methods, vol. 64, pp. 144-52, Dec 1 2013.
[14] M. Hosokawa, T. Hayata, Y. Fukuda, A. Arakaki, T. Yoshino, T. Tanaka, and T.
Matsunaga, "Size-selective microcavity array for rapid and efficient detection of
circulating tumor cells," Anal Chem, vol. 82, pp. 6629-35, Aug 1 2010.
[15] J. S. Kuo, Y. Zhao, P. G. Schiro, L. Ng, D. S. Lim, J. P. Shelby, and D. T. Chiu,
"Deformability considerations in filtration of biological cells," Lab Chip, vol. 10, pp. 837-
42, Apr 7 2010.
[16] S. Nagrath, L. V. Sequist, S. Maheswaran, D. W. Bell, D. Irimia, L. Ulkus, M. R. Smith,
E. L. Kwak, S. Digumarthy, A. Muzikansky, P. Ryan, U. J. Balis, R. G. Tompkins, D. A.
Haber, and M. Toner, "Isolation of rare circulating tumour cells in cancer patients by
microchip technology," Nature, vol. 450, pp. 1235-9, Dec 20 2007.
[17] P. R. Gascoyne, J. Noshari, T. J. Anderson, and F. F. Becker, "Isolation of rare cells from
cell mixtures by dielectrophoresis," Electrophoresis, vol. 30, pp. 1388-98, Apr 2009.
[18] E. Ozkumur, A. M. Shah, J. C. Ciciliano, B. L. Emmink, D. T. Miyamoto, E. Brachtel, M.
Yu, P. I. Chen, B. Morgan, J. Trautwein, A. Kimura, S. Sengupta, S. L. Stott, N. M.
Karabacak, T. A. Barber, J. R. Walsh, K. Smith, P. S. Spuhler, J. P. Sullivan, R. J. Lee, D.
T. Ting, X. Luo, A. T. Shaw, A. Bardia, L. V. Sequist, D. N. Louis, S. Maheswaran, R.
Kapur, D. A. Haber, and M. Toner, "Inertial focusing for tumor antigen-dependent and -
independent sorting of rare circulating tumor cells," Sci Transl Med, vol. 5, p. 179ra47,
Apr 3 2013.
[19] F. A. Coumans, S. T. Ligthart, J. W. Uhr, and L. W. Terstappen, "Challenges in the
enumeration and phenotyping of CTC," Clin Cancer Res, vol. 18, pp. 5711-8, Oct 15 2012.
[20] C. Alix-Panabieres and K. Pantel, "Challenges in circulating tumour cell research," Nat
Rev Cancer, vol. 14, pp. 623-31, Sep 2014.
[21] W. Liu, B. Yin, X. Wang, P. Yu, X. Duan, C. Liu, B. Wang, and Z. Tao, "Circulating tumor
cells in prostate cancer: Precision diagnosis and therapy," Oncol Lett, vol. 14, pp. 1223-
1232, Aug 2017.
[22] P. Mehlen and A. Puisieux, "Metastasis: a question of life or death," Nat Rev Cancer, vol.
6, pp. 449-58, Jun 2006.
[23] X. Jin and P. Mu, "Targeting Breast Cancer Metastasis," Breast Cancer (Auckl), vol. 9, pp.
23-34, 2015.
[24] L. Ombrato and I. Malanchi, "The EMT universe: space between cancer cell dissemination
and metastasis initiation," Crit Rev Oncog, vol. 19, pp. 349-61, 2014.
104
[25] A. G. Tibbe, B. G. de Grooth, J. Greve, P. A. Liberti, G. J. Dolan, and L. W. Terstappen,
"Optical tracking and detection of immunomagnetically selected and aligned cells," Nat
Biotechnol, vol. 17, pp. 1210-3, Dec 1999.
[26] K. Pantel, R. H. Brakenhoff, and B. Brandt, "Detection, clinical relevance and specific
biological properties of disseminating tumour cells," Nat Rev Cancer, vol. 8, pp. 329-40,
May 2008.
[27] M. Cristofanilli, G. T. Budd, M. J. Ellis, A. Stopeck, J. Matera, M. C. Miller, J. M. Reuben,
G. V. Doyle, W. J. Allard, L. W. Terstappen, and D. F. Hayes, "Circulating tumor cells,
disease progression, and survival in metastatic breast cancer," N Engl J Med, vol. 351, pp.
781-91, Aug 19 2004.
[28] S. J. Cohen, C. J. Punt, N. Iannotti, B. H. Saidman, K. D. Sabbath, N. Y. Gabrail, J. Picus,
M. Morse, E. Mitchell, M. C. Miller, G. V. Doyle, H. Tissing, L. W. Terstappen, and N. J.
Meropol, "Relationship of circulating tumor cells to tumor response, progression-free
survival, and overall survival in patients with metastatic colorectal cancer," J Clin Oncol,
vol. 26, pp. 3213-21, Jul 1 2008.
[29] J. S. de Bono, H. I. Scher, R. B. Montgomery, C. Parker, M. C. Miller, H. Tissing, G. V.
Doyle, L. W. Terstappen, K. J. Pienta, and D. Raghavan, "Circulating tumor cells predict
survival benefit from treatment in metastatic castration-resistant prostate cancer," Clin
Cancer Res, vol. 14, pp. 6302-9, Oct 1 2008.
[30] M. G. Krebs, R. Sloane, L. Priest, L. Lancashire, J. M. Hou, A. Greystoke, T. H. Ward, R.
Ferraldeschi, A. Hughes, G. Clack, M. Ranson, C. Dive, and F. H. Blackhall, "Evaluation
and prognostic significance of circulating tumor cells in patients with non-small-cell lung
cancer," J Clin Oncol, vol. 29, pp. 1556-63, Apr 20 2011.
[31] J. M. Hou, M. G. Krebs, L. Lancashire, R. Sloane, A. Backen, R. K. Swain, L. J. Priest, A.
Greystoke, C. Zhou, K. Morris, T. Ward, F. H. Blackhall, and C. Dive, "Clinical
significance and molecular characteristics of circulating tumor cells and circulating tumor
microemboli in patients with small-cell lung cancer," J Clin Oncol, vol. 30, pp. 525-32,
Feb 10 2012.
[32] M. S. Khan, A. Kirkwood, T. Tsigani, J. Garcia-Hernandez, J. A. Hartley, M. E. Caplin,
and T. Meyer, "Circulating tumor cells as prognostic markers in neuroendocrine tumors,"
J Clin Oncol, vol. 31, pp. 365-72, Jan 20 2013.
[33] L. Khoja, P. Lorigan, C. Zhou, M. Lancashire, J. Booth, J. Cummings, R. Califano, G.
Clack, A. Hughes, and C. Dive, "Biomarker utility of circulating tumor cells in metastatic
cutaneous melanoma," J Invest Dermatol, vol. 133, pp. 1582-90, Jun 2013.
[34] A. M. Sieuwerts, J. Kraan, J. Bolt, P. van der Spoel, F. Elstrodt, M. Schutte, J. W. Martens,
J. W. Gratama, S. Sleijfer, and J. A. Foekens, "Anti-epithelial cell adhesion molecule
antibodies and the detection of circulating normal-like breast tumor cells," J. Natl. Cancer
Inst., vol. 101, pp. 61-6, Jan 7 2009.
105
[35] F. Farace, C. Massard, N. Vimond, F. Drusch, N. Jacques, F. Billiot, A. Laplanche, A.
Chauchereau, L. Lacroix, D. Planchard, S. Le Moulec, F. Andre, K. Fizazi, J. C. Soria, and
P. Vielh, "A direct comparison of CellSearch and ISET for circulating tumour-cell
detection in patients with metastatic carcinomas," Br. J. Cancer, vol. 105, pp. 847-53, Sep
6 2011.
[36] B. J. Green, T. Saberi Safaei, A. Mepham, M. Labib, R. M. Mohamadi, and S. O. Kelley,
"Beyond the Capture of Circulating Tumor Cells: Next-Generation Devices and Materials,"
Angew Chem Int Ed Engl, vol. 55, pp. 1252-65, Jan 22 2016.
[37] A. E. Saliba, L. Saias, E. Psychari, N. Minc, D. Simon, F. C. Bidard, C. Mathiot, J. Y.
Pierga, V. Fraisier, J. Salamero, V. Saada, F. Farace, P. Vielh, L. Malaquin, and J. L. Viovy,
"Microfluidic sorting and multimodal typing of cancer cells in self-assembled magnetic
arrays," Proc Natl Acad Sci U S A, vol. 107, pp. 14524-9, Aug 17 2010.
[38] J. Autebert, B. Coudert, J. Champ, L. Saias, E. T. Guneri, R. Lebofsky, F. C. Bidard, J. Y.
Pierga, F. Farace, S. Descroix, L. Malaquin, and J. L. Viovy, "High purity microfluidic
sorting and analysis of circulating tumor cells: towards routine mutation detection," Lab
Chip, Mar 27 2015.
[39] S. L. Stott, C. H. Hsu, D. I. Tsukrov, M. Yu, D. T. Miyamoto, B. A. Waltman, S. M.
Rothenberg, A. M. Shah, M. E. Smas, G. K. Korir, F. P. Floyd, Jr., A. J. Gilman, J. B. Lord,
D. Winokur, S. Springer, D. Irimia, S. Nagrath, L. V. Sequist, R. J. Lee, K. J. Isselbacher,
S. Maheswaran, D. A. Haber, and M. Toner, "Isolation of circulating tumor cells using a
microvortex-generating herringbone-chip," Proc. Natl. Acad. Sci., U.S.A., vol. 107, pp.
18392-7, Oct 26 2010.
[40] A. A. Adams, P. I. Okagbare, J. Feng, M. L. Hupert, D. Patterson, J. Gottert, R. L.
McCarley, D. Nikitopoulos, M. C. Murphy, and S. A. Soper, "Highly efficient circulating
tumor cell isolation from whole blood and label-free enumeration using polymer-based
microfluidics with an integrated conductivity sensor," J Am Chem Soc, vol. 130, pp. 8633-
41, Jul 9 2008.
[41] P. G. Schiro, M. Zhao, J. S. Kuo, K. M. Koehler, D. E. Sabath, and D. T. Chiu, "Sensitive
and high-throughput isolation of rare cells from peripheral blood with ensemble-decision
aliquot ranking," Angew. Chem. Intl. Ed., vol. 51, pp. 4618-22, May 7 2012.
[42] A. H. Talasaz, A. A. Powell, D. E. Huber, J. G. Berbee, K. H. Roh, W. Yu, W. Xiao, M.
M. Davis, R. F. Pease, M. N. Mindrinos, S. S. Jeffrey, and R. W. Davis, "Isolating highly
enriched populations of circulating epithelial cells and other rare cells from blood using a
magnetic sweeper device," Proc Natl Acad Sci U S A, vol. 106, pp. 3970-5, Mar 10 2009.
[43] A. A. Powell, A. H. Talasaz, H. Zhang, M. A. Coram, A. Reddy, G. Deng, M. L. Telli, R.
H. Advani, R. W. Carlson, J. A. Mollick, S. Sheth, A. W. Kurian, J. M. Ford, F. E.
Stockdale, S. R. Quake, R. F. Pease, M. N. Mindrinos, G. Bhanot, S. H. Dairkee, R. W.
Davis, and S. S. Jeffrey, "Single cell profiling of circulating tumor cells: transcriptional
heterogeneity and diversity from breast cancer cell lines," PLoS One, vol. 7, p. e33788,
2012.
106
[44] G. Deng, S. Krishnakumar, A. A. Powell, H. Zhang, M. N. Mindrinos, M. L. Telli, R. W.
Davis, and S. S. Jeffrey, "Single cell mutational analysis of PIK3CA in circulating tumor
cells and metastases in breast cancer reveals heterogeneity, discordance, and mutation
persistence in cultured disseminated tumor cells from bone marrow," BMC Cancer, vol.
14, p. 456, 2014.
[45] S. Wang, H. Wang, J. Jiao, K. J. Chen, G. E. Owens, K. Kamei, J. Sun, D. J. Sherman, C.
P. Behrenbruch, H. Wu, and H. R. Tseng, "Three-dimensional nanostructured substrates
toward efficient capture of circulating tumor cells," Angew Chem Int Ed Engl, vol. 48, pp.
8970-3, 2009.
[46] S. Wang, K. Liu, J. Liu, Z. T. Yu, X. Xu, L. Zhao, T. Lee, E. K. Lee, J. Reiss, Y. K. Lee,
L. W. Chung, J. Huang, M. Rettig, D. Seligson, K. N. Duraiswamy, C. K. Shen, and H. R.
Tseng, "Highly efficient capture of circulating tumor cells by using nanostructured silicon
substrates with integrated chaotic micromixers," Angew Chem Int Ed Engl, vol. 50, pp.
3084-8, Mar 21 2011.
[47] M. Lin, J. F. Chen, Y. T. Lu, Y. Zhang, J. Song, S. Hou, Z. Ke, and H. R. Tseng,
"Nanostructure embedded microchips for detection, isolation, and characterization of
circulating tumor cells," Acc Chem Res, vol. 47, pp. 2941-50, Oct 21 2014.
[48] J. Sekine, S. C. Luo, S. Wang, B. Zhu, H. R. Tseng, and H. H. Yu, "Functionalized
conducting polymer nanodots for enhanced cell capturing: the synergistic effect of capture
agents and nanostructures," Adv. Mater., vol. 23, pp. 4788-92, Nov 2 2011.
[49] N. Zhang, Y. Deng, Q. Tai, B. Cheng, L. Zhao, Q. Shen, R. He, L. Hong, W. Liu, S. Guo,
K. Liu, H. R. Tseng, B. Xiong, and X. Z. Zhao, "Electrospun TiO2 nanofiber-based cell
capture assay for detecting circulating tumor cells from colorectal and gastric cancer
patients," Adv Mater, vol. 24, pp. 2756-60, May 22 2012.
[50] I. Baccelli, A. Schneeweiss, S. Riethdorf, A. Stenzinger, A. Schillert, V. Vogel, C. Klein,
M. Saini, T. Bauerle, M. Wallwiener, T. Holland-Letz, T. Hofner, M. Sprick, M. Scharpff,
F. Marme, H. P. Sinn, K. Pantel, W. Weichert, and A. Trumpp, "Identification of a
population of blood circulating tumor cells from breast cancer patients that initiates
metastasis in a xenograft assay," Nat Biotechnol, vol. 31, pp. 539-44, Jun 2013.
[51] J. G. Lohr, V. A. Adalsteinsson, K. Cibulskis, A. D. Choudhury, M. Rosenberg, P. Cruz-
Gordillo, J. M. Francis, C. Z. Zhang, A. K. Shalek, R. Satija, J. J. Trombetta, D. Lu, N.
Tallapragada, N. Tahirova, S. Kim, B. Blumenstiel, C. Sougnez, A. Lowe, B. Wong, D.
Auclair, E. M. Van Allen, M. Nakabayashi, R. T. Lis, G. S. Lee, T. Li, M. S. Chabot, A.
Ly, M. E. Taplin, T. E. Clancy, M. Loda, A. Regev, M. Meyerson, W. C. Hahn, P. W.
Kantoff, T. R. Golub, G. Getz, J. S. Boehm, and J. C. Love, "Whole-exome sequencing of
circulating tumor cells provides a window into metastatic prostate cancer," Nat Biotechnol,
vol. 32, pp. 479-84, May 2014.
[52] M. Xie, N. N. Lu, S. B. Cheng, X. Y. Wang, M. Wang, S. Guo, C. Y. Wen, J. Hu, D. W.
Pang, and W. H. Huang, "Engineered decomposable multifunctional nanobioprobes for
capture and release of rare cancer cells," Anal Chem, vol. 86, pp. 4618-26, May 6 2014.
107
[53] W. Zhao, C. H. Cui, S. Bose, D. Guo, C. Shen, W. P. Wong, K. Halvorsen, O. C.
Farokhzad, G. S. Teo, J. A. Phillips, D. M. Dorfman, R. Karnik, and J. M. Karp,
"Bioinspired multivalent DNA network for capture and release of cells," Proc Natl Acad
Sci U S A, vol. 109, pp. 19626-31, Nov 27 2012.
[54] Q. Shen, L. Xu, L. Zhao, D. Wu, Y. Fan, Y. Zhou, W. H. Ouyang, X. Xu, Z. Zhang, M.
Song, T. Lee, M. A. Garcia, B. Xiong, S. Hou, H. R. Tseng, and X. Fang, "Specific capture
and release of circulating tumor cells using aptamer-modified nanosubstrates," Adv Mater,
vol. 25, pp. 2368-73, Apr 24 2013.
[55] Y. Wan, Y. Liu, P. B. Allen, W. Asghar, M. A. Mahmood, J. Tan, H. Duhon, Y. T. Kim,
A. D. Ellington, and S. M. Iqbal, "Capture, isolation and release of cancer cells with
aptamer-functionalized glass bead array," Lab Chip, vol. 12, pp. 4693-701, Nov 21 2012.
[56] E. Reategui, N. Aceto, E. J. Lim, J. P. Sullivan, A. E. Jensen, M. Zeinali, J. M. Martel, A.
J. Aranyosi, W. Li, S. Castleberry, A. Bardia, L. V. Sequist, D. A. Haber, S. Maheswaran,
P. T. Hammond, M. Toner, and S. L. Stott, "Tunable nanostructured coating for the capture
and selective release of viable circulating tumor cells," Adv Mater, vol. 27, pp. 1593-9,
Mar 2015.
[57] S. Hou, H. Zhao, L. Zhao, Q. Shen, K. S. Wei, D. Y. Suh, A. Nakao, M. A. Garcia, M.
Song, T. Lee, B. Xiong, S. C. Luo, H. R. Tseng, and H. H. Yu, "Capture and stimulated
release of circulating tumor cells on polymer-grafted silicon nanostructures," Adv Mater,
vol. 25, pp. 1547-51, Mar 20 2013.
[58] Z. Ke, M. Lin, J. F. Chen, J. S. Choi, Y. Zhang, A. Fong, A. J. Liang, S. F. Chen, Q. Li,
W. Fang, P. Zhang, M. A. Garcia, T. Lee, M. Song, H. A. Lin, H. Zhao, S. C. Luo, S. Hou,
H. H. Yu, and H. R. Tseng, "Programming thermoresponsiveness of NanoVelcro substrates
enables effective purification of circulating tumor cells in lung cancer patients," ACS Nano,
vol. 9, pp. 62-70, Jan 27 2015.
[59] P. L. Bedard, A. R. Hansen, M. J. Ratain, and L. L. Siu, "Tumour heterogeneity in the
clinic," Nature, vol. 501, pp. 355-64, Sep 19 2013.
[60] C. E. Meacham and S. J. Morrison, "Tumour heterogeneity and cancer cell plasticity,"
Nature, vol. 501, pp. 328-37, Sep 19 2013.
[61] A. Bonnomet, L. Syne, A. Brysse, E. Feyereisen, E. W. Thompson, A. Noel, J. M. Foidart,
P. Birembaut, M. Polette, and C. Gilles, "A dynamic in vivo model of epithelial-to-
mesenchymal transitions in circulating tumor cells and metastases of breast cancer,"
Oncogene, vol. 31, pp. 3741-53, Aug 16 2012.
[62] B. Polzer, G. Medoro, S. Pasch, F. Fontana, L. Zorzino, A. Pestka, U. Andergassen, F.
Meier-Stiegen, Z. T. Czyz, B. Alberter, S. Treitschke, T. Schamberger, M. Sergio, G.
Bregola, A. Doffini, S. Gianni, A. Calanca, G. Signorini, C. Bolognesi, A. Hartmann, P.
A. Fasching, M. T. Sandri, B. Rack, T. Fehm, G. Giorgini, N. Manaresi, and C. A. Klein,
"Molecular profiling of single circulating tumor cells with diagnostic intention," EMBO
Mol. Med., vol. 6, pp. 1371-86, Nov 2014.
108
[63] M. G. Krebs, R. L. Metcalf, L. Carter, G. Brady, F. H. Blackhall, and C. Dive, "Molecular
analysis of circulating tumour cells-biology and biomarkers," Nat Rev Clin Oncol, vol. 11,
pp. 129-44, Mar 2014.
[64] J. F. Swennenhuis, J. Reumers, K. Thys, J. Aerssens, and L. W. Terstappen, "Efficiency of
whole genome amplification of single circulating tumor cells enriched by CellSearch and
sorted by FACS," Genome Med, vol. 5, p. 106, 2013.
[65] I. Y. Wong, S. Javaid, E. A. Wong, S. Perk, D. A. Haber, M. Toner, and D. Irimia,
"Collective and individual migration following the epithelial-mesenchymal transition,"
Nat. Mater., vol. 13, pp. 1063-71, Nov 2014.
[66] I. Baccelli, A. Schneeweiss, S. Riethdorf, A. Stenzinger, A. Schillert, V. Vogel, C. Klein,
M. Saini, T. Bauerle, M. Wallwiener, T. Holland-Letz, T. Hofner, M. Sprick, M. Scharpff,
F. Marme, H. P. Sinn, K. Pantel, W. Weichert, and A. Trumpp, "Identification of a
population of blood circulating tumor cells from breast cancer patients that initiates
metastasis in a xenograft assay," Nat. Biotechnol., vol. 31, pp. 539-44, Jun 2013.
[67] L. Zhang, L. D. Ridgway, M. D. Wetzel, J. Ngo, W. Yin, D. Kumar, J. C. Goodman, M.
D. Groves, and D. Marchetti, "The identification and characterization of breast cancer
CTCs competent for brain metastasis," Sci. Transl. Med., vol. 5, p. 180ra48, Apr 10 2013.
[68] Y. Zhang, W. Zhang, and L. Qin, "Mesenchymal-mode migration assay and antimetastatic
drug screening with high-throughput microfluidic channel networks," Angew Chem Int Ed
Engl, vol. 53, pp. 2344-8, Feb 24 2014.
[69] Y. Zhang, L. Zhou, and L. Qin, "High-throughput 3D cell invasion chip enables accurate
cancer metastatic assays," J Am Chem Soc, vol. 136, pp. 15257-62, Oct 29 2014.
[70] S. Balasubramanian, D. Kagan, C. M. Hu, S. Campuzano, M. J. Lobo-Castanon, N. Lim,
D. Y. Kang, M. Zimmerman, L. Zhang, and J. Wang, "Micromachine-enabled capture and
isolation of cancer cells in complex media," Angew Chem Int Ed Engl, vol. 50, pp. 4161-
4, Apr 26 2011.
[71] L. Zhang, L. D. Ridgway, M. D. Wetzel, J. Ngo, W. Yin, D. Kumar, J. C. Goodman, M.
D. Groves, and D. Marchetti, "The identification and characterization of breast cancer
CTCs competent for brain metastasis," Sci Transl Med, vol. 5, p. 180ra48, Apr 10 2013.
[72] H. J. Yoon, M. Kozminsky, and S. Nagrath, "Emerging role of nanomaterials in circulating
tumor cell isolation and analysis," ACS Nano, vol. 8, pp. 1995-2017, Mar 25 2014.
[73] A. Markiewicz, M. Ksiazkiewicz, M. Welnicka-Jaskiewicz, B. Seroczynska, J. Skokowski,
J. Szade, and A. J. Zaczek, "Mesenchymal phenotype of CTC-enriched blood fraction and
lymph node metastasis formation potential," PLoS One, vol. 9, p. e93901, 2014.
[74] G. Vona, A. Sabile, M. Louha, V. Sitruk, S. Romana, K. Schutze, F. Capron, D. Franco,
M. Pazzagli, M. Vekemans, B. Lacour, C. Brechot, and P. Paterlini-Brechot, "Isolation by
size of epithelial tumor cells : a new method for the immunomorphological and molecular
characterization of circulatingtumor cells," Am J Pathol, vol. 156, pp. 57-63, Jan 2000.
109
[75] T. L. Halo, K. M. McMahon, N. L. Angeloni, Y. Xu, W. Wang, A. B. Chinen, D. Malin,
E. Strekalova, V. L. Cryns, C. Cheng, C. A. Mirkin, and C. S. Thaxton, "NanoFlares for
the detection, isolation, and culture of live tumor cells from human blood," Proc Natl Acad
Sci U S A, vol. 111, pp. 17104-9, Dec 2 2014.
[76] P. G. Schiro, M. Zhao, J. S. Kuo, K. M. Koehler, D. E. Sabath, and D. T. Chiu, "Sensitive
and high-throughput isolation of rare cells from peripheral blood with ensemble-decision
aliquot ranking," Angew Chem Int Ed Engl, vol. 51, pp. 4618-22, May 7 2012.
[77] H. J. Yoon, T. H. Kim, Z. Zhang, E. Azizi, T. M. Pham, C. Paoletti, J. Lin, N. Ramnath,
M. S. Wicha, D. F. Hayes, D. M. Simeone, and S. Nagrath, "Sensitive capture of circulating
tumour cells by functionalized graphene oxide nanosheets," Nat Nanotechnol, vol. 8, pp.
735-41, Oct 2013.
[78] G. M. Whitesides, "The origins and the future of microfluidics," Nature, vol. 442, pp. 368-
73, Jul 27 2006.
[79] D. J. Beebe, G. A. Mensing, and G. M. Walker, "Physics and applications of microfluidics
in biology," Annu Rev Biomed Eng, vol. 4, pp. 261-86, 2002.
[80] R. M. Mohamadi, J. D. Besant, A. Mepham, B. Green, L. Mahmoudian, T. Gibbs, I. Ivanov,
A. Malvea, J. Stojcic, A. L. Allan, L. E. Lowes, E. H. Sargent, R. K. Nam, and S. O. Kelley,
"Nanoparticle-mediated binning and profiling of heterogeneous circulating tumor cell
subpopulations," Angew Chem Int Ed Engl, vol. 54, pp. 139-43, Jan 2 2015.
[81] W. J. Allard, J. Matera, M. C. Miller, M. Repollet, M. C. Connelly, C. Rao, A. G. Tibbe,
J. W. Uhr, and L. W. Terstappen, "Tumor cells circulate in the peripheral blood of all major
carcinomas but not in healthy subjects or patients with nonmalignant diseases," Clin
Cancer Res, vol. 10, pp. 6897-904, Oct 15 2004.
[82] F. Farace, C. Massard, N. Vimond, F. Drusch, N. Jacques, F. Billiot, A. Laplanche, A.
Chauchereau, L. Lacroix, D. Planchard, S. Le Moulec, F. Andre, K. Fizazi, J. C. Soria, and
P. Vielh, "A direct comparison of CellSearch and ISET for circulating tumour-cell
detection in patients with metastatic carcinomas," Br J Cancer, vol. 105, pp. 847-53, Sep
6 2011.
[83] A. M. Sieuwerts, B. Mostert, J. Bolt-de Vries, D. Peeters, F. E. de Jongh, J. M. Stouthard,
L. Y. Dirix, P. A. van Dam, A. Van Galen, V. de Weerd, J. Kraan, P. van der Spoel, R.
Ramirez-Moreno, C. H. van Deurzen, M. Smid, J. X. Yu, J. Jiang, Y. Wang, J. W. Gratama,
S. Sleijfer, J. A. Foekens, and J. W. Martens, "mRNA and microRNA expression profiles
in circulating tumor cells and primary tumors of metastatic breast cancer patients," Clin
Cancer Res, vol. 17, pp. 3600-18, Jun 1 2011.
[84] S. S. Shevkoplyas, A. C. Siegel, R. M. Westervelt, M. G. Prentiss, and G. M. Whitesides,
"The force acting on a superparamagnetic bead due to an applied magnetic field," Lab
Chip, vol. 7, pp. 1294-302, Oct 2007.
[85] H. Zhang, L. R. Moore, M. Zborowski, P. S. Williams, S. Margel, and J. J. Chalmers,
"Establishment and implications of a characterization method for magnetic nanoparticle
110
using cell tracking velocimetry and magnetic susceptibility modified solutions," Analyst,
vol. 130, pp. 514-27, Apr 2005.
[86] N. Prang, S. Preithner, K. Brischwein, P. Goster, A. Woppel, J. Muller, C. Steiger, M.
Peters, P. A. Baeuerle, and A. J. da Silva, "Cellular and complement-dependent
cytotoxicity of Ep-CAM-specific monoclonal antibody MT201 against breast cancer cell
lines," Br J Cancer, vol. 92, pp. 342-9, Jan 31 2005.
[87] K. S. Kim and J. K. Park, "Magnetic force-based multiplexed immunoassay using
superparamagnetic nanoparticles in microfluidic channel," Lab Chip, vol. 5, pp. 657-64,
Jun 2005.
[88] G. S. Zamay, O. S. Kolovskaya, T. N. Zamay, Y. E. Glazyrin, A. V. Krat, O. Zubkova, E.
Spivak, M. Wehbe, A. Gargaun, D. Muharemagic, M. Komarova, V. Grigorieva, A.
Savchenko, A. A. Modestov, M. V. Berezovski, and A. S. Zamay, "Aptamers Selected to
Postoperative Lung Adenocarcinoma Detect Circulating Tumor Cells in Human Blood,"
Mol Ther, vol. 23, pp. 1486-96, Sep 2015.
[89] H. Ma, J. Liu, M. M. Ali, M. A. Mahmood, L. Labanieh, M. Lu, S. M. Iqbal, Q. Zhang, W.
Zhao, and Y. Wan, "Nucleic acid aptamers in cancer research, diagnosis and therapy,"
Chem Soc Rev, vol. 44, pp. 1240-56, Mar 7 2015.
[90] U. Dharmasiri, S. K. Njoroge, M. A. Witek, M. G. Adebiyi, J. W. Kamande, M. L. Hupert,
F. Barany, and S. A. Soper, "High-throughput selection, enumeration, electrokinetic
manipulation, and molecular profiling of low-abundance circulating tumor cells using a
microfluidic system," Anal Chem, vol. 83, pp. 2301-9, Mar 15 2011.
[91] W. Sheng, T. Chen, W. Tan, and Z. H. Fan, "Multivalent DNA nanospheres for enhanced
capture of cancer cells in microfluidic devices," ACS Nano, vol. 7, pp. 7067-76, Aug 27
2013.
[92] W. Sheng, T. Chen, R. Kamath, X. Xiong, W. Tan, and Z. H. Fan, "Aptamer-enabled
efficient isolation of cancer cells from whole blood using a microfluidic device," Anal
Chem, vol. 84, pp. 4199-206, May 1 2012.
[93] Y. Xu, J. A. Phillips, J. Yan, Q. Li, Z. H. Fan, and W. Tan, "Aptamer-based microfluidic
device for enrichment, sorting, and detection of multiple cancer cells," Anal Chem, vol. 81,
pp. 7436-42, Sep 1 2009.
[94] Y. Wan, J. Tan, W. Asghar, Y. T. Kim, Y. Liu, and S. M. Iqbal, "Velocity effect on
aptamer-based circulating tumor cell isolation in microfluidic devices," J Phys Chem B,
vol. 115, pp. 13891-6, Dec 1 2011.
[95] U. Dharmasiri, S. Balamurugan, A. A. Adams, P. I. Okagbare, A. Obubuafo, and S. A.
Soper, "Highly efficient capture and enumeration of low abundance prostate cancer cells
using prostate-specific membrane antigen aptamers immobilized to a polymeric
microfluidic device," Electrophoresis, vol. 30, pp. 3289-300, Sep 2009.
111
[96] Y. K. Jung, M. A. Woo, H. T. Soh, and H. G. Park, "Aptamer-based cell imaging reagents
capable of fluorescence switching," Chem Commun (Camb), vol. 50, pp. 12329-32, Oct 21
2014.
[97] Y. Song, Z. Zhu, Y. An, W. Zhang, H. Zhang, D. Liu, C. Yu, W. Duan, and C. J. Yang,
"Selection of DNA aptamers against epithelial cell adhesion molecule for cancer cell
imaging and circulating tumor cell capture," Anal Chem, vol. 85, pp. 4141-9, Apr 16 2013.
[98] J. D. Besant, R. M. Mohamadi, P. M. Aldridge, Y. Li, E. H. Sargent, and S. O. Kelley,
"Velocity valleys enable efficient capture and spatial sorting of nanoparticle-bound cancer
cells," Nanoscale, vol. 7, pp. 6278-85, Apr 14 2015.
[99] T. W. Friedlander, V. T. Ngo, H. Dong, G. Premasekharan, V. Weinberg, S. Doty, Q. Zhao,
E. G. Gilbert, C. J. Ryan, W. T. Chen, and P. L. Paris, "Detection and characterization of
invasive circulating tumor cells derived from men with metastatic castration-resistant
prostate cancer," Int J Cancer, vol. 134, pp. 2284-93, May 15 2014.
[100] G. Spizzo, D. Fong, M. Wurm, C. Ensinger, P. Obrist, C. Hofer, G. Mazzoleni, G. Gastl,
and P. Went, "EpCAM expression in primary tumour tissues and metastases: an
immunohistochemical analysis," J Clin Pathol, vol. 64, pp. 415-20, May 2011.
[101] M. Z. Gilcrease, W. A. Woodward, M. M. Nicolas, L. J. Corley, G. N. Fuller, F. J. Esteva,
S. L. Tucker, and T. A. Buchholz, "Even low-level HER2 expression may be associated
with worse outcome in node-positive breast cancer," Am J Surg Pathol, vol. 33, pp. 759-
67, May 2009.
[102] J. W. Kim, E. Y. Kim, S. Y. Kim, S. K. Byun, D. Lee, K. J. Oh, W. K. Kim, B. S. Han, S.
W. Chi, S. C. Lee, and K. H. Bae, "Identification of DNA aptamers toward epithelial cell
adhesion molecule via cell-SELEX," Mol Cells, vol. 37, pp. 742-6, Oct 31 2014.
[103] Z. Liu, J. H. Duan, Y. M. Song, J. Ma, F. D. Wang, X. Lu, and X. D. Yang, "Novel HER2
aptamer selectively delivers cytotoxic drug to HER2-positive breast cancer cells in vitro,"
J Transl Med, vol. 10, p. 148, 2012.
[104] J. H. Niazi, S. K. Verma, S. Niazi, and A. Qureshi, "In vitro HER2 protein-induced affinity
dissociation of carbon nanotube-wrapped anti-HER2 aptamers for HER2 protein
detection," Analyst, vol. 140, pp. 243-9, Jan 7 2015.
[105] L. T. M. Phuc, P. V. Phuc, and L. Q. Huan, "Using Electrochemical Technique to Detect
HER2 Antigen," Indian Journal of Applied Research, vol. 4, pp. 64-66, 2014.
[106] J. A. Magee, E. Piskounova, and S. J. Morrison, "Cancer stem cells: impact, heterogeneity,
and uncertainty," Cancer Cell, vol. 21, pp. 283-96, Mar 20 2012.
[107] K. Pantel and M. R. Speicher, "The biology of circulating tumor cells," Oncogene, vol. 35,
pp. 1216-24, Mar 10 2016.
[108] P. C. Hermann, S. L. Huber, T. Herrler, A. Aicher, J. W. Ellwart, M. Guba, C. J. Bruns,
and C. Heeschen, "Distinct populations of cancer stem cells determine tumor growth and
112
metastatic activity in human pancreatic cancer," Cell Stem Cell, vol. 1, pp. 313-23, Sep 13
2007.
[109] R. Pang, W. L. Law, A. C. Chu, J. T. Poon, C. S. Lam, A. K. Chow, L. Ng, L. W. Cheung,
X. R. Lan, H. Y. Lan, V. P. Tan, T. C. Yau, R. T. Poon, and B. C. Wong, "A subpopulation
of CD26+ cancer stem cells with metastatic capacity in human colorectal cancer," Cell
Stem Cell, vol. 6, pp. 603-15, Jun 4 2010.
[110] E. Piskounova, M. Agathocleous, M. M. Murphy, Z. Hu, S. E. Huddlestun, Z. Zhao, A. M.
Leitch, T. M. Johnson, R. J. DeBerardinis, and S. J. Morrison, "Oxidative stress inhibits
distant metastasis by human melanoma cells," Nature, vol. 527, pp. 186-91, Nov 12 2015.
[111] B. Polzer, G. Medoro, S. Pasch, F. Fontana, L. Zorzino, A. Pestka, U. Andergassen, F.
Meier-Stiegen, Z. T. Czyz, B. Alberter, S. Treitschke, T. Schamberger, M. Sergio, G.
Bregola, A. Doffini, S. Gianni, A. Calanca, G. Signorini, C. Bolognesi, A. Hartmann, P.
A. Fasching, M. T. Sandri, B. Rack, T. Fehm, G. Giorgini, N. Manaresi, and C. A. Klein,
"Molecular profiling of single circulating tumor cells with diagnostic intention," EMBO
Mol Med, vol. 6, pp. 1371-86, Nov 2014.
[112] S. C. Hur, N. K. Henderson-MacLennan, E. R. McCabe, and D. Di Carlo, "Deformability-
based cell classification and enrichment using inertial microfluidics," Lab Chip, vol. 11,
pp. 912-20, Mar 7 2011.
[113] M. Pellegrino, A. Sciambi, J. L. Yates, J. D. Mast, C. Silver, and D. J. Eastburn, "RNA-
Seq following PCR-based sorting reveals rare cell transcriptional signatures," BMC
Genomics, vol. 17, p. 361, 2016.
[114] G. Hamilton, M. Hochmair, B. Rath, L. Klameth, and R. Zeillinger, "Small cell lung
cancer: Circulating tumor cells of extended stage patients express a mesenchymal-
epithelial transition phenotype," Cell Adh Migr, pp. 1-8, Feb 26 2016.
[115] W. Luo, M. Pla-Roca, and D. Juncker, "Taguchi design-based optimization of sandwich
immunoassay microarrays for detecting breast cancer biomarkers," Anal Chem, vol. 83, pp.
5767-74, Jul 15 2011.
[116] M. E. Warkiani, G. Guan, K. B. Luan, W. C. Lee, A. A. Bhagat, P. K. Chaudhuri, D. S.
Tan, W. T. Lim, S. C. Lee, P. C. Chen, C. T. Lim, and J. Han, "Slanted spiral microfluidics
for the ultra-fast, label-free isolation of circulating tumor cells," Lab Chip, vol. 14, pp. 128-
37, Jan 7 2014.
[117] C. Alix-Panabieres and K. Pantel, "Technologies for detection of circulating tumor cells:
facts and vision," Lab Chip, vol. 14, pp. 57-62, Jan 7 2014.
[118] D. T. Miyamoto, Y. Zheng, B. S. Wittner, R. J. Lee, H. Zhu, K. T. Broderick, R. Desai, D.
B. Fox, B. W. Brannigan, J. Trautwein, K. S. Arora, N. Desai, D. M. Dahl, L. V. Sequist,
M. R. Smith, R. Kapur, C. L. Wu, T. Shioda, S. Ramaswamy, D. T. Ting, M. Toner, S.
Maheswaran, and D. A. Haber, "RNA-Seq of single prostate CTCs implicates
noncanonical Wnt signaling in antiandrogen resistance," Science, vol. 349, pp. 1351-6, Sep
18 2015.
113
[119] J. M. Ramirez, T. Fehm, M. Orsini, L. Cayrefourcq, T. Maudelonde, K. Pantel, and C.
Alix-Panabieres, "Prognostic relevance of viable circulating tumor cells detected by
EPISPOT in metastatic breast cancer patients," Clin Chem, vol. 60, pp. 214-21, Jan 2014.
[120] M. L. Pearl, Q. Zhao, J. Yang, H. Dong, S. Tulley, Q. Zhang, M. Golightly, S. Zucker, and
W. T. Chen, "Prognostic analysis of invasive circulating tumor cells (iCTCs) in epithelial
ovarian cancer," Gynecol Oncol, vol. 134, pp. 581-90, Sep 2014.
[121] X. Yao, A. D. Choudhury, Y. J. Yamanaka, V. A. Adalsteinsson, T. M. Gierahn, C. A.
Williamson, C. R. Lamb, M. E. Taplin, M. Nakabayashi, M. S. Chabot, T. Li, G. S. Lee, J.
S. Boehm, P. W. Kantoff, W. C. Hahn, K. D. Wittrup, and J. C. Love, "Functional analysis
of single cells identifies a rare subset of circulating tumor cells with malignant traits,"
Integr Biol (Camb), vol. 6, pp. 388-98, Apr 2014.
[122] M. Labib, B. Green, R. M. Mohamadi, A. Mepham, S. U. Ahmed, L. Mahmoudian, I. H.
Chang, E. H. Sargent, and S. O. Kelley, "Aptamer and Antisense-Mediated Two-
Dimensional Isolation of Specific Cancer Cell Subpopulations," J Am Chem Soc, vol. 138,
pp. 2476-9, Mar 2 2016.
[123] Y. Zhang, M. Wu, X. Han, P. Wang, and L. Qin, "High-Throughput, Label-Free Isolation
of Cancer Stem Cells on the Basis of Cell Adhesion Capacity," Angew Chem Int Ed Engl,
vol. 54, pp. 10838-42, Sep 7 2015.
[124] Z. Liu, Y. Lee, J. Jang, Y. Li, X. Han, K. Yokoi, M. Ferrari, L. Zhou, and L. Qin,
"Microfluidic cytometric analysis of cancer cell transportability and invasiveness," Sci
Rep, vol. 5, p. 14272, 2015.
[125] Y. C. Chen, S. G. Allen, P. N. Ingram, R. Buckanovich, S. D. Merajver, and E. Yoon,
"Single-cell Migration Chip for Chemotaxis-based Microfluidic Selection of
Heterogeneous Cell Populations," Sci Rep, vol. 5, p. 9980, 2015.
[126] R. M. Mohamadi, I. Ivanov, J. Stojcic, R. K. Nam, E. H. Sargent, and S. O. Kelley,
"Sample-to-Answer Isolation and mRNA Profiling of Circulating Tumor Cells," Anal
Chem, vol. 87, pp. 6258-64, Jun 16 2015.
[127] N. Muhanna, A. Mepham, R. M. Mohamadi, H. Chan, T. Khan, M. Akens, J. D. Besant, J.
Irish, and S. O. Kelley, "Nanoparticle-based sorting of circulating tumor cells by epithelial
antigen expression during disease progression in an animal model," Nanomedicine, vol.
11, pp. 1613-20, Oct 2015.
[128] R. Vaidyanathan, S. Rauf, E. Dray, M. J. Shiddiky, and M. Trau, "Alternating current
electrohydrodynamics induced nanoshearing and fluid micromixing for specific capture of
cancer cells," Chemistry, vol. 20, pp. 3724-9, Mar 24 2014.
[129] M. Ruscetti, B. Quach, E. L. Dadashian, D. J. Mulholland, and H. Wu, "Tracking and
Functional Characterization of Epithelial-Mesenchymal Transition and Mesenchymal
Tumor Cells during Prostate Cancer Metastasis," Cancer Res, vol. 75, pp. 2749-59, Jul 1
2015.
114
[130] X. Lu and Y. Kang, "Hypoxia and hypoxia-inducible factors: master regulators of
metastasis," Clin Cancer Res, vol. 16, pp. 5928-35, Dec 15 2010.
[131] A. G. Recchia, E. M. De Francesco, A. Vivacqua, D. Sisci, M. L. Panno, S. Ando, and M.
Maggiolini, "The G protein-coupled receptor 30 is up-regulated by hypoxia-inducible
factor-1alpha (HIF-1alpha) in breast cancer cells and cardiomyocytes," J Biol Chem, vol.
286, pp. 10773-82, Mar 25 2011.
[132] Y. B. Zhang, X. Wang, E. A. Meister, K. R. Gong, S. C. Yan, G. W. Lu, X. M. Ji, and G.
Shao, "The effects of CoCl2 on HIF-1alpha protein under experimental conditions of
autoprogressive hypoxia using mouse models," Int J Mol Sci, vol. 15, pp. 10999-1012,
2014.
[133] D. Kong, F. Zhang, J. Shao, L. Wu, X. Zhang, L. Chen, Y. Lu, and S. Zheng, "Curcumin
inhibits cobalt chloride-induced epithelial-to-mesenchymal transition associated with
interference with TGF-beta/Smad signaling in hepatocytes," Lab Invest, vol. 95, pp. 1234-
45, Nov 2015.
[134] W. Liu, S. M. Shen, X. Y. Zhao, and G. Q. Chen, "Targeted genes and interacting proteins
of hypoxia inducible factor-1," Int J Biochem Mol Biol, vol. 3, pp. 165-78, 2012.
[135] M. Z. Noman, Y. Messai, J. Muret, M. Hasmim, and S. Chouaib, "Crosstalk between CTC,
Immune System and Hypoxic Tumor Microenvironment," Cancer Microenviron, vol. 7,
pp. 153-60, Dec 2014.
[136] J. V. Joseph, S. Conroy, K. Pavlov, P. Sontakke, T. Tomar, E. Eggens-Meijer, V.
Balasubramaniyan, M. Wagemakers, W. F. den Dunnen, and F. A. Kruyt, "Hypoxia
enhances migration and invasion in glioblastoma by promoting a mesenchymal shift
mediated by the HIF1alpha-ZEB1 axis," Cancer Lett, vol. 359, pp. 107-16, Apr 1 2015.
[137] M. Walter, S. Liang, S. Ghosh, P. J. Hornsby, and R. Li, "Interleukin 6 secreted from
adipose stromal cells promotes migration and invasion of breast cancer cells," Oncogene,
vol. 28, pp. 2745-55, Jul 30 2009.
[138] A. C. Curino, L. H. Engelholm, S. S. Yamada, K. Holmbeck, L. R. Lund, A. A. Molinolo,
N. Behrendt, B. S. Nielsen, and T. H. Bugge, "Intracellular collagen degradation mediated
by uPARAP/Endo180 is a major pathway of extracellular matrix turnover during
malignancy," J Cell Biol, vol. 169, pp. 977-85, Jun 20 2005.
[139] D. Wienke, G. C. Davies, D. A. Johnson, J. Sturge, M. B. Lambros, K. Savage, S. E.
Elsheikh, A. R. Green, I. O. Ellis, D. Robertson, J. S. Reis-Filho, and C. M. Isacke, "The
collagen receptor Endo180 (CD280) Is expressed on basal-like breast tumor cells and
promotes tumor growth in vivo," Cancer Res, vol. 67, pp. 10230-40, Nov 1 2007.
[140] E. M. Grzincic and C. J. Murphy, "Gold Nanorods Indirectly Promote Migration of
Metastatic Human Breast Cancer Cells in Three-Dimensional Cultures," ACS Nano, vol.
9, pp. 6801-16, Jul 28 2015.
115
[141] A. Miyoshi, Y. Kitajima, K. Sumi, K. Sato, A. Hagiwara, Y. Koga, and K. Miyazaki, "Snail
and SIP1 increase cancer invasion by upregulating MMP family in hepatocellular
carcinoma cells," Br J Cancer, vol. 90, pp. 1265-73, Mar 22 2004.
[142] S. Elloul, M. B. Elstrand, J. M. Nesland, C. G. Trope, G. Kvalheim, I. Goldberg, R. Reich,
and B. Davidson, "Snail, Slug, and Smad-interacting protein 1 as novel parameters of
disease aggressiveness in metastatic ovarian and breast carcinoma," Cancer, vol. 103, pp.
1631-43, Apr 15 2005.
[143] O. Y. Fu, M. F. Hou, S. F. Yang, S. C. Huang, and W. Y. Lee, "Cobalt chloride-induced
hypoxia modulates the invasive potential and matrix metalloproteinases of primary and
metastatic breast cancer cells," Anticancer Res, vol. 29, pp. 3131-8, Aug 2009.
[144] J. Ye, J. Fan, S. Venneti, Y. W. Wan, B. R. Pawel, J. Zhang, L. W. Finley, C. Lu, T.
Lindsten, J. R. Cross, G. Qing, Z. Liu, M. C. Simon, J. D. Rabinowitz, and C. B. Thompson,
"Serine catabolism regulates mitochondrial redox control during hypoxia," Cancer Discov,
vol. 4, pp. 1406-17, Dec 2014.
[145] J. V. Rocheleau, W. S. Head, and D. W. Piston, "Quantitative NAD(P)H/flavoprotein
autofluorescence imaging reveals metabolic mechanisms of pancreatic islet pyruvate
response," J Biol Chem, vol. 279, pp. 31780-7, Jul 23 2004.
[146] M. Y. Sun, E. Yoo, B. J. Green, S. M. Altamentova, D. M. Kilkenny, and J. V. Rocheleau,
"Autofluorescence imaging of living pancreatic islets reveals fibroblast growth factor-21
(FGF21)-induced metabolism," Biophys J, vol. 103, pp. 2379-88, Dec 5 2012.
[147] A. V. Meleshina, V. V. Dudenkova, M. V. Shirmanova, V. I. Shcheslavskiy, W. Becker,
A. S. Bystrova, E. I. Cherkasova, and E. V. Zagaynova, "Probing metabolic states of
differentiating stem cells using two-photon FLIM," Sci Rep, vol. 6, p. 21853, 2016.
[148] X. X. Tong, D. Wu, X. Wang, H. L. Chen, J. X. Chen, X. X. Wang, X. L. Wang, L. Gan,
Z. Y. Guo, G. X. Shi, Y. Z. Zhang, and W. Jiang, "Ghrelin protects against cobalt chloride-
induced hypoxic injury in cardiac H9c2 cells by inhibiting oxidative stress and inducing
autophagy," Peptides, vol. 38, pp. 217-27, Dec 2012.
[149] C. Alix-Panabieres and K. Pantel, "Circulating tumor cells: liquid biopsy of cancer," Clin
Chem, vol. 59, pp. 110-8, Jan 2013.
[150] N. M. Karabacak, P. S. Spuhler, F. Fachin, E. J. Lim, V. Pai, E. Ozkumur, J. M. Martel, N.
Kojic, K. Smith, P. I. Chen, J. Yang, H. Hwang, B. Morgan, J. Trautwein, T. A. Barber, S.
L. Stott, S. Maheswaran, R. Kapur, D. A. Haber, and M. Toner, "Microfluidic, marker-free
isolation of circulating tumor cells from blood samples," Nat Protoc, vol. 9, pp. 694-710,
Mar 2014.
[151] B. J. Green, L. Kermanshah, M. Labib, S. U. Ahmed, P. N. Silva, L. Mahmoudian, I. H.
Chang, R. M. Mohamadi, J. V. Rocheleau, and S. O. Kelley, "Isolation of Phenotypically
Distinct Cancer Cells Using Nanoparticle-Mediated Sorting," ACS Appl Mater Interfaces,
vol. 9, pp. 20435-20443, Jun 21 2017.
116
[152] W. G. Jiang, A. J. Sanders, M. Katoh, H. Ungefroren, F. Gieseler, M. Prince, S. K.
Thompson, M. Zollo, D. Spano, P. Dhawan, D. Sliva, P. R. Subbarayan, M. Sarkar, K.
Honoki, H. Fujii, A. G. Georgakilas, A. Amedei, E. Niccolai, A. Amin, S. S. Ashraf, L.
Ye, W. G. Helferich, X. Yang, C. S. Boosani, G. Guha, M. R. Ciriolo, K. Aquilano, S.
Chen, A. S. Azmi, W. N. Keith, A. Bilsland, D. Bhakta, D. Halicka, S. Nowsheen, F.
Pantano, and D. Santini, "Tissue invasion and metastasis: Molecular, biological and
clinical perspectives," Semin Cancer Biol, vol. 35 Suppl, pp. S244-75, Dec 2015.
[153] C. D. Paul, P. Mistriotis, and K. Konstantopoulos, "Cancer cell motility: lessons from
migration in confined spaces," Nat Rev Cancer, vol. 17, pp. 131-140, Feb 2017.
[154] P. Friedl, K. Wolf, and J. Lammerding, "Nuclear mechanics during cell migration," Curr
Opin Cell Biol, vol. 23, pp. 55-64, Feb 2011.
[155] E. K. Paluch, I. M. Aspalter, and M. Sixt, "Focal Adhesion-Independent Cell Migration,"
Annu Rev Cell Dev Biol, vol. 32, pp. 469-490, Oct 06 2016.
[156] R. G. Rowe and S. J. Weiss, "Navigating ECM barriers at the invasive front: the cancer
cell-stroma interface," Annu Rev Cell Dev Biol, vol. 25, pp. 567-95, 2009.
[157] A. W. Holle, J. L. Young, and J. P. Spatz, "In vitro cancer cell-ECM interactions inform in
vivo cancer treatment," Adv Drug Deliv Rev, vol. 97, pp. 270-9, Feb 1 2016.
[158] P. Friedl, E. Sahai, S. Weiss, and K. M. Yamada, "New dimensions in cell migration," Nat
Rev Mol Cell Biol, vol. 13, pp. 743-7, Nov 2012.
[159] A. Ferrari, M. Cecchini, A. Dhawan, S. Micera, I. Tonazzini, R. Stabile, D. Pisignano, and
F. Beltram, "Nanotopographic Control of Neuronal Polarity," Nano Letters, vol. 11, pp.
505-511, 2011/02/09 2011.
[160] W. J. Nelson, "Remodeling Epithelial Cell Organization: Transitions Between Front–Rear
and Apical–Basal Polarity," Cold Spring Harbor Perspectives in Biology, vol. 1, p.
a000513, 2009.
[161] S. B. Khatau, C. M. Hale, P. J. Stewart-Hutchinson, M. S. Patel, C. L. Stewart, P. C.
Searson, D. Hodzic, and D. Wirtz, "A perinuclear actin cap regulates nuclear shape," Proc
Natl Acad Sci U S A, vol. 106, pp. 19017-22, Nov 10 2009.
[162] H.-R. Thiam, P. Vargas, N. Carpi, C. L. Crespo, M. Raab, E. Terriac, M. C. King, J.
Jacobelli, A. S. Alberts, T. Stradal, A.-M. Lennon-Dumenil, and M. Piel, "Perinuclear
Arp2/3-driven actin polymerization enables nuclear deformation to facilitate cell migration
through complex environments," Nature Communications, vol. 7, p. 10997, 03/15
[163] J. Lammerding, "Mechanics of the nucleus," Compr Physiol, vol. 1, pp. 783-807, Apr 2011.
[164] K. Wolf, M. Te Lindert, M. Krause, S. Alexander, J. Te Riet, A. L. Willis, R. M. Hoffman,
C. G. Figdor, S. J. Weiss, and P. Friedl, "Physical limits of cell migration: control by ECM
space and nuclear deformation and tuning by proteolysis and traction force," J Cell Biol,
vol. 201, pp. 1069-84, Jun 24 2013.
117
[165] P. M. Davidson, C. Denais, M. C. Bakshi, and J. Lammerding, "Nuclear Deformability
Constitutes a Rate-Limiting Step During Cell Migration in 3-D Environments," Cellular
and Molecular Bioengineering, vol. 7, pp. 293-306, September 01 2014.
[166] M. Panagiotakopoulou, M. Bergert, A. Taubenberger, J. Guck, D. Poulikakos, and A.
Ferrari, "A Nanoprinted Model of Interstitial Cancer Migration Reveals a Link between
Cell Deformability and Proliferation," ACS Nano, vol. 10, pp. 6437-48, Jul 26 2016.
[167] M. Mak, C. A. Reinhart-King, and D. Erickson, "Microfabricated physical spatial gradients
for investigating cell migration and invasion dynamics," PLoS One, vol. 6, p. e20825, 2011.
[168] M. T. Breckenridge, T. T. Egelhoff, and H. Baskaran, "A microfluidic imaging chamber
for the direct observation of chemotactic transmigration," Biomed Microdevices, vol. 12,
pp. 543-53, Jun 2010.
[169] T. Harada, J. Swift, J. Irianto, J.-W. Shin, K. R. Spinler, A. Athirasala, R. Diegmiller, P.
C. D. P. Dingal, I. L. Ivanovska, and D. E. Discher, "Nuclear lamin stiffness is a barrier to
3D migration, but softness can limit survival," The Journal of Cell Biology, vol. 204, pp.
669-682, 2014.
[170] C. M. Denais, R. M. Gilbert, P. Isermann, A. L. McGregor, M. te Lindert, B. Weigelin, P.
M. Davidson, P. Friedl, K. Wolf, and J. Lammerding, "Nuclear envelope rupture and repair
during cancer cell migration," Science, vol. 352, pp. 353-8, Apr 15 2016.
[171] M. Raab, M. Gentili, H. de Belly, H.-R. Thiam, P. Vargas, A. J. Jimenez, F.
Lautenschlaeger, R. Voituriez, A.-M. Lennon-Duménil, N. Manel, and M. Piel, "ESCRT
III repairs nuclear envelope ruptures during cell migration to limit DNA damage and cell
death," Science, vol. 352, pp. 359-362, 2016.
[172] D. Irimia and M. Toner, "Spontaneous migration of cancer cells under conditions of
mechanical confinement," Integr Biol (Camb), vol. 1, pp. 506-12, Sep 2009.
[173] K. Wolf, S. Alexander, V. Schacht, L. M. Coussens, U. H. von Andrian, J. van Rheenen,
E. Deryugina, and P. Friedl, "Collagen-based cell migration models in vitro and in vivo,"
Semin Cell Dev Biol, vol. 20, pp. 931-41, Oct 2009.
[174] F. Klein, T. Striebel, J. Fischer, Z. Jiang, C. M. Franz, G. von Freymann, M. Wegener, and
M. Bastmeyer, "Elastic Fully Three-dimensional Microstructure Scaffolds for Cell Force
Measurements," Advanced Materials, vol. 22, pp. 868-871, 2010.
[175] I. Matsuura, C. Y. Lai, and K. N. Chiang, "Functional interaction between Smad3 and
S100A4 (metastatin-1) for TGF-beta-mediated cancer cell invasiveness," Biochem J, vol.
426, pp. 327-35, Feb 24 2010.
[176] Z. Tang, M. Yu, F. Miller, R. S. Berk, G. Tromp, and M. A. Kosir, "Increased invasion
through basement membrane by CXCL7-transfected breast cells," Am J Surg, vol. 196, pp.
690-6, Nov 2008.
118
[177] S. Etienne-Manneville, "Cdc42--the centre of polarity," J Cell Sci, vol. 117, pp. 1291-300,
Mar 15 2004.
[178] D.-H. Kim and D. Wirtz, "Cytoskeletal tension induces the polarized architecture of the
nucleus," Biomaterials, vol. 48, pp. 161-172, 2015/04/01/ 2015.
[179] J.-K. Kim, A. Louhghalam, G. Lee, B. W. Schafer, D. Wirtz, and D.-H. Kim, "Nuclear
lamin A/C harnesses the perinuclear apical actin cables to protect nuclear morphology,"
Nature Communications, vol. 8, p. 2123, 2017/12/14 2017.
[180] S. Lamouille, J. Xu, and R. Derynck, "Molecular mechanisms of epithelial-mesenchymal
transition," Nat Rev Mol Cell Biol, vol. 15, pp. 178-96, Mar 2014.
[181] R. Mayor and S. Etienne-Manneville, "The front and rear of collective cell migration," Nat
Rev Mol Cell Biol, vol. 17, pp. 97-109, Feb 2016.
[182] B. Bisel, M. Calamai, F. Vanzi, and F. S. Pavone, "Decoupling polarization of the Golgi
apparatus and GM1 in the plasma membrane," PLoS One, vol. 8, p. e80446, 2013.
[183] B. Burke and K. J. Roux, "Nuclei take a position: managing nuclear location," Dev Cell,
vol. 17, pp. 587-97, Nov 2009.
[184] P. J. Albert and U. S. Schwarz, "Dynamics of Cell Ensembles on Adhesive Micropatterns:
Bridging the Gap between Single Cell Spreading and Collective Cell Migration," PLoS
Comput Biol, vol. 12, p. e1004863, Apr 2016.
[185] A. V. West, L. Wullkopf, A. Christensen, N. Leijnse, J. M. Tarp, J. Mathiesen, J. T. Erler,
and L. B. Oddershede, "Dynamics of cancerous tissue correlates with invasiveness," Sci
Rep, vol. 7, p. 43800, Mar 06 2017.
[186] B. Ladoux and R. M. Mege, "Mechanobiology of collective cell behaviours," Nat Rev Mol
Cell Biol, Nov 08 2017.
[187] T. E. Angelini, E. Hannezo, X. Trepat, M. Marquez, J. J. Fredberg, and D. A. Weitz,
"Glass-like dynamics of collective cell migration," Proc Natl Acad Sci U S A, vol. 108, pp.
4714-9, Mar 22 2011.
[188] C. Malinverno, S. Corallino, F. Giavazzi, M. Bergert, Q. Li, M. Leoni, A. Disanza, E.
Frittoli, A. Oldani, E. Martini, T. Lendenmann, G. Deflorian, G. V. Beznoussenko, D.
Poulikakos, O. K. Haur, M. Uroz, X. Trepat, D. Parazzoli, P. Maiuri, W. Yu, A. Ferrari, R.
Cerbino, and G. Scita, "Endocytic reawakening of motility in jammed epithelia," Nat
Mater, vol. 16, pp. 587-596, May 2017.
[189] F. Milde, D. Franco, A. Ferrari, V. Kurtcuoglu, D. Poulikakos, and P. Koumoutsakos, "Cell
Image Velocimetry (CIV): boosting the automated quantification of cell migration in
wound healing assays," Integrative Biology, vol. 4, pp. 1437-1447, 2012.
119
[190] A. D. Richardson, C. Yang, A. Osterman, and J. W. Smith, "Central carbon metabolism in
the progression of mammary carcinoma," Breast Cancer Res Treat, vol. 110, pp. 297-307,
Jul 2008.
[191] F. R. Miller, H. D. Soule, L. Tait, R. J. Pauley, S. R. Wolman, P. J. Dawson, and G. H.
Heppner, "Xenograft model of progressive human proliferative breast disease," J Natl
Cancer Inst, vol. 85, pp. 1725-32, Nov 03 1993.
[192] G. Stefopoulos, F. Robotti, V. Falk, D. Poulikakos, and A. Ferrari, "Endothelialization of
Rationally Microtextured Surfaces with Minimal Cell Seeding Under Flow," Small, vol.
12, pp. 4113-4126, 2016.
[193] T. Ridler and S. Calvard, "Picture thresholding using an iterative selection method," IEEE
transactions on Systems, Man and Cybernetics, vol. 8, pp. 630-632, 1978.
[194] G. P. Haas, N. Delongchamps, O. W. Brawley, C. Y. Wang, and G. de la Roza, "The
worldwide epidemiology of prostate cancer: perspectives from autopsy studies," Can J
Urol, vol. 15, pp. 3866-71, Feb 2008.
[195] H. I. Scher, S. Halabi, I. Tannock, M. Morris, C. N. Sternberg, M. A. Carducci, M. A.
Eisenberger, C. Higano, G. J. Bubley, R. Dreicer, D. Petrylak, P. Kantoff, E. Basch, W. K.
Kelly, W. D. Figg, E. J. Small, T. M. Beer, G. Wilding, A. Martin, and M. Hussain, "Design
and end points of clinical trials for patients with progressive prostate cancer and castrate
levels of testosterone: recommendations of the Prostate Cancer Clinical Trials Working
Group," J Clin Oncol, vol. 26, pp. 1148-59, Mar 1 2008.
[196] P. A. Watson, V. K. Arora, and C. L. Sawyers, "Emerging mechanisms of resistance to
androgen receptor inhibitors in prostate cancer," Nat Rev Cancer, vol. 15, pp. 701-11, Dec
2015.
[197] M. Tucci, G. V. Scagliotti, and F. Vignani, "Metastatic castration-resistant prostate cancer:
time for innovation," Future Oncol, vol. 11, pp. 91-106, 2015.
[198] L. J. Brand and S. M. Dehm, "Androgen receptor gene rearrangements: new perspectives
on prostate cancer progression," Curr Drug Targets, vol. 14, pp. 441-9, Apr 2013.
[199] H. I. Scher, K. Fizazi, F. Saad, M. E. Taplin, C. N. Sternberg, K. Miller, R. de Wit, P.
Mulders, K. N. Chi, N. D. Shore, A. J. Armstrong, T. W. Flaig, A. Flechon, P. Mainwaring,
M. Fleming, J. D. Hainsworth, M. Hirmand, B. Selby, L. Seely, and J. S. de Bono,
"Increased survival with enzalutamide in prostate cancer after chemotherapy," N Engl J
Med, vol. 367, pp. 1187-97, Sep 27 2012.
[200] T. van der Steen, D. J. Tindall, and H. Huang, "Posttranslational modification of the
androgen receptor in prostate cancer," Int J Mol Sci, vol. 14, pp. 14833-59, Jul 16 2013.
[201] S. T. Ligthart, F. A. Coumans, G. Attard, A. M. Cassidy, J. S. de Bono, and L. W.
Terstappen, "Unbiased and automated identification of a circulating tumour cell definition
that associates with overall survival," PLoS One, vol. 6, p. e27419, 2011.
120
[202] L. E. Lowes, D. Goodale, Y. Xia, C. Postenka, M. M. Piaseczny, F. Paczkowski, and A. L.
Allan, "Epithelial-to-mesenchymal transition leads to disease-stage differences in
circulating tumor cell detection and metastasis in pre-clinical models of prostate cancer,"
Oncotarget, vol. 7, pp. 76125-76139, Nov 15 2016.
[203] B. De Laere, S. Oeyen, P. Van Oyen, C. Ghysel, J. Ampe, P. Ost, W. Demey, L. Hoekx,
D. Schrijvers, B. Brouwers, W. Lybaert, E. Everaert, P. Van Kerckhove, D. De Maeseneer,
M. Strijbos, A. Bols, K. Fransis, N. Beije, I. de Kruijff, V. van Dam, A. Brouwer, P. J. van
Dam, G. Van den Eynden, A. Rutten, S. Sleijfer, J. Vandebroek, S. Van Laere, and L. Dirix,
"Circulating tumor cells and survival in abiraterone- and enzalutamide-treated patients with
castration-resistant prostate cancer," Prostate, vol. 78, pp. 435-445, May 2018.
[204] D. Lorente, D. Olmos, J. Mateo, D. Bianchini, G. Seed, M. Fleisher, D. C. Danila, P. Flohr,
M. Crespo, I. Figueiredo, S. Miranda, K. Baeten, A. Molina, T. Kheoh, R. McCormack, L.
W. Terstappen, H. I. Scher, and J. S. de Bono, "Decline in Circulating Tumor Cell Count
and Treatment Outcome in Advanced Prostate Cancer," Eur Urol, vol. 70, pp. 985-992,
Dec 2016.
[205] B. Kahn, J. Collazo, and N. Kyprianou, "Androgen receptor as a driver of therapeutic
resistance in advanced prostate cancer," Int J Biol Sci, vol. 10, pp. 588-95, 2014.
[206] E. S. Antonarakis, C. Lu, B. Luber, H. Wang, Y. Chen, Y. Zhu, J. L. Silberstein, M. N.
Taylor, B. L. Maughan, S. R. Denmeade, K. J. Pienta, C. J. Paller, M. A. Carducci, M. A.
Eisenberger, and J. Luo, "Clinical Significance of Androgen Receptor Splice Variant-7
mRNA Detection in Circulating Tumor Cells of Men With Metastatic Castration-Resistant
Prostate Cancer Treated With First- and Second-Line Abiraterone and Enzalutamide," J
Clin Oncol, vol. 35, pp. 2149-2156, Jul 1 2017.
[207] E. Sariisik, D. Docheva, D. Padula, C. Popov, J. Opfer, M. Schieker, H. Clausen-
Schaumann, and M. Benoit, "Probing the interaction forces of prostate cancer cells with
collagen I and bone marrow derived stem cells on the single cell level," PLoS One, vol. 8,
p. e57706, 2013.
[208] H. I. Scher, M. J. Morris, W. M. Stadler, C. Higano, E. Basch, K. Fizazi, E. S. Antonarakis,
T. M. Beer, M. A. Carducci, K. N. Chi, P. G. Corn, J. S. de Bono, R. Dreicer, D. J. George,
E. I. Heath, M. Hussain, W. K. Kelly, G. Liu, C. Logothetis, D. Nanus, M. N. Stein, D. E.
Rathkopf, S. F. Slovin, C. J. Ryan, O. Sartor, E. J. Small, M. R. Smith, C. N. Sternberg, M.
E. Taplin, G. Wilding, P. S. Nelson, L. H. Schwartz, S. Halabi, P. W. Kantoff, and A. J.
Armstrong, "Trial Design and Objectives for Castration-Resistant Prostate Cancer:
Updated Recommendations From the Prostate Cancer Clinical Trials Working Group 3," J
Clin Oncol, vol. 34, pp. 1402-18, Apr 20 2016.
[209] A. McGuire, J. A. Brown, and M. J. Kerin, "Metastatic breast cancer: the potential of
miRNA for diagnosis and treatment monitoring," Cancer Metastasis Rev, vol. 34, pp. 145-
55, Mar 2015.
[210] M. Wilbaux, M. Tod, J. De Bono, D. Lorente, J. Mateo, G. Freyer, B. You, and E. Henin,
"A Joint Model for the Kinetics of CTC Count and PSA Concentration During Treatment
121
in Metastatic Castration-Resistant Prostate Cancer," CPT Pharmacometrics Syst
Pharmacol, vol. 4, pp. 277-85, May 2015.
[211] U. G. Lo, C. F. Lee, M. S. Lee, and J. T. Hsieh, "The Role and Mechanism of Epithelial-
to-Mesenchymal Transition in Prostate Cancer Progression," Int J Mol Sci, vol. 18, Sep 30
2017.
[212] E. S. Antonarakis, C. Lu, H. Wang, B. Luber, M. Nakazawa, J. C. Roeser, Y. Chen, T. A.
Mohammad, H. L. Fedor, T. L. Lotan, Q. Zheng, A. M. De Marzo, J. T. Isaacs, W. B.
Isaacs, R. Nadal, C. J. Paller, S. R. Denmeade, M. A. Carducci, M. A. Eisenberger, and J.
Luo, "AR-V7 and resistance to enzalutamide and abiraterone in prostate cancer," N Engl J
Med, vol. 371, pp. 1028-38, Sep 11 2014.
[213] E. S. Antonarakis, C. Lu, B. Luber, H. Wang, Y. Chen, M. Nakazawa, R. Nadal, C. J.
Paller, S. R. Denmeade, M. A. Carducci, M. A. Eisenberger, and J. Luo, "Androgen
Receptor Splice Variant 7 and Efficacy of Taxane Chemotherapy in Patients With
Metastatic Castration-Resistant Prostate Cancer," JAMA Oncol, vol. 1, pp. 582-91, Aug
2015.
[214] L. L. Liu, N. Xie, S. Sun, S. Plymate, E. Mostaghel, and X. Dong, "Mechanisms of the
androgen receptor splicing in prostate cancer cells," Oncogene, vol. 33, pp. 3140-50, Jun
12 2014.
[215] E. Jernberg, A. Bergh, and P. Wikstrom, "Clinical relevance of androgen receptor
alterations in prostate cancer," Endocr Connect, vol. 6, pp. R146-R161, Nov 2017.
[216] Y. Hong, F. Fang, and Q. Zhang, "Circulating tumor cell clusters: What we know and what
we expect (Review)," Int J Oncol, vol. 49, pp. 2206-2216, Dec 2016.
[217] M. Labib, R. M. Mohamadi, M. Poudineh, S. U. Ahmed, I. Ivanov, C. L. Huang, M.
Moosavi, E. H. Sargent, and S. O. Kelley, "Single-cell mRNA cytometry via sequence-
specific nanoparticle clustering and trapping," Nat Chem, vol. 10, pp. 489-495, May 2018.
[218] A. F. Sarioglu, N. Aceto, N. Kojic, M. C. Donaldson, M. Zeinali, B. Hamza, A. Engstrom,
H. Zhu, T. K. Sundaresan, D. T. Miyamoto, X. Luo, A. Bardia, B. S. Wittner, S.
Ramaswamy, T. Shioda, D. T. Ting, S. L. Stott, R. Kapur, S. Maheswaran, D. A. Haber,
and M. Toner, "A microfluidic device for label-free, physical capture of circulating tumor
cell clusters," Nat Methods, vol. 12, pp. 685-91, Jul 2015.
[219] A. Ferrari, M. Cecchini, A. Dhawan, S. Micera, I. Tonazzini, R. Stabile, D. Pisignano, and
F. Beltram, "Nanotopographic control of neuronal polarity," Nano Lett, vol. 11, pp. 505-
11, Feb 9 2011.
[220] P. Friedl, P. B. Noble, P. A. Walton, D. W. Laird, P. J. Chauvin, R. J. Tabah, M. Black,
and K. S. Zanker, "Migration of coordinated cell clusters in mesenchymal and epithelial
cancer explants in vitro," Cancer Res, vol. 55, pp. 4557-60, Oct 15 1995.
122
[221] N. Gjorevski, A. S. Piotrowski, V. D. Varner, and C. M. Nelson, "Dynamic tensile forces
drive collective cell migration through three-dimensional extracellular matrices," Sci Rep,
vol. 5, p. 11458, Jul 13 2015.
[222] C. Wiltshire, B. L. Singh, J. Stockley, J. Fleming, B. Doyle, R. Barnetson, C. N. Robson,
F. Kozielski, and H. Y. Leung, "Docetaxel-resistant prostate cancer cells remain sensitive
to S-trityl-L-cysteine-mediated Eg5 inhibition," Mol Cancer Ther, vol. 9, pp. 1730-9, Jun
2010.
[223] J. Zhurinsky, M. Shtutman, and A. Ben-Ze'ev, "Plakoglobin and beta-catenin: protein
interactions, regulation and biological roles," J Cell Sci, vol. 113 ( Pt 18), pp. 3127-39, Sep
2000.
[224] M. Poudineh, M. Labib, S. Ahmed, L. N. Nguyen, L. Kermanshah, R. M. Mohamadi, E.
H. Sargent, and S. O. Kelley, "Profiling Functional and Biochemical Phenotypes of
Circulating Tumor Cells Using a Two-Dimensional Sorting Device," Angew Chem Int Ed
Engl, vol. 56, pp. 163-168, Jan 2 2017.
[225] F. Valderrama, S. Thevapala, and A. J. Ridley, "Radixin regulates cell migration and cell-
cell adhesion through Rac1," J Cell Sci, vol. 125, pp. 3310-9, Jul 15 2012.
[226] R. Li, J. D. Hebert, T. A. Lee, H. Xing, A. Boussommier-Calleja, R. O. Hynes, D. A.
Lauffenburger, and R. D. Kamm, "Macrophage-Secreted TNFalpha and TGFbeta1
Influence Migration Speed and Persistence of Cancer Cells in 3D Tissue Culture via
Independent Pathways," Cancer Res, vol. 77, pp. 279-290, Jan 15 2017.
[227] P. Osmulski, D. Mahalingam, M. E. Gaczynska, J. Liu, S. Huang, A. M. Horning, C. M.
Wang, I. M. Thompson, T. H. Huang, and C. L. Chen, "Nanomechanical biomarkers of
single circulating tumor cells for detection of castration resistant prostate cancer," Prostate,
vol. 74, pp. 1297-307, Sep 2014.
[228] N. Yamaguchi, T. Mizutani, K. Kawabata, and H. Haga, "Leader cells regulate collective
cell migration via Rac activation in the downstream signaling of integrin beta1 and PI3K,"
Sci Rep, vol. 5, p. 7656, Jan 7 2015.
[229] S. P. Desai, S. N. Bhatia, M. Toner, and D. Irimia, "Mitochondrial localization and the
persistent migration of epithelial cancer cells," Biophys J, vol. 104, pp. 2077-88, May 7
2013.
[230] K. M. Aw Yong, Z. Li, S. D. Merajver, and J. Fu, "Tracking the tumor invasion front using
long-term fluidic tumoroid culture," Sci Rep, vol. 7, p. 10784, Sep 7 2017.
[231] B. Weksler, J. Lenert, B. Ng, and M. Burt, "Isolated single lung perfusion with doxorubicin
is effective in eradicating soft tissue sarcoma lung metastases in a rat model," J Thorac
Cardiovasc Surg, vol. 107, pp. 50-4, Jan 1994.
[232] P. J. Villeneuve and R. S. Sundaresan, "Surgical management of colorectal lung
metastasis," Clin Colon Rectal Surg, vol. 22, pp. 233-41, Nov 2009.
123
[233] R. Fontao-Wendel, P. M. Hoff, A. Lazar, D. Freitas, Y. Novis, P. Patah, M. Tsujita, A.
Balthazar, M. Pierroti, and S. Wendel, "Immune-mediated pancytopenia induced by
oxaliplatin: a case report," Transfusion, vol. 50, pp. 1453-9, Jul 2010.
[234] P. R. dos Santos, I. Iskender, T. Machuca, D. Hwang, M. dePerrot, M. Liu, S. Keshavjee,
T. K. Waddell, and M. Cypel, "Modified in vivo lung perfusion allows for prolonged
perfusion without acute lung injury," J Thorac Cardiovasc Surg, vol. 147, pp. 774-81:
discussion 781-2, Feb 2014.
[235] H. Y. Wang, J. L. Port, S. N. Hochwald, and M. E. Burt, "Revised technique of isolated
lung perfusion in the rat," Ann Thorac Surg, vol. 60, pp. 211-2, Jul 1995.
[236] C. L. Chen, D. Mahalingam, P. Osmulski, R. R. Jadhav, C. M. Wang, R. J. Leach, T. C.
Chang, S. D. Weitman, A. P. Kumar, L. Sun, M. E. Gaczynska, I. M. Thompson, and T.
H. Huang, "Single-cell analysis of circulating tumor cells identifies cumulative expression
patterns of EMT-related genes in metastatic prostate cancer," Prostate, vol. 73, pp. 813-
26, Jun 2013.
[237] M. Tascilar, W. J. Loos, C. Seynaeve, J. Verweij, and S. Sleijfer, "The pharmacologic basis
of ifosfamide use in adult patients with advanced soft tissue sarcomas," Oncologist, vol.
12, pp. 1351-60, Nov 2007.
[238] D. Rathkopf, M. A. Dickson, D. R. Feldman, R. D. Carvajal, M. A. Shah, N. Wu, R.
Lefkowitz, M. Gonen, L. M. Cane, H. J. Dials, J. L. Winkelmann, G. J. Bosl, and G. K.
Schwartz, "Phase I study of flavopiridol with oxaliplatin and fluorouracil/leucovorin in
advanced solid tumors," Clin Cancer Res, vol. 15, pp. 7405-11, Dec 1 2009.
[239] T. Andre, C. Boni, L. Mounedji-Boudiaf, M. Navarro, J. Tabernero, T. Hickish, C.
Topham, M. Zaninelli, P. Clingan, J. Bridgewater, I. Tabah-Fisch, and A. de Gramont,
"Oxaliplatin, fluorouracil, and leucovorin as adjuvant treatment for colon cancer," N Engl
J Med, vol. 350, pp. 2343-51, Jun 3 2004.
[240] L. Kermanshah, M. Poudineh, S. Ahmed, L.M. Nguyen, S. Srikant, R. Makonnen, F.
Penacantu, M. Corrigan, S.O. Kelley, " Dynamic CTC phenotypes in metastatic prostate
cancer models visualized using magnetic ranking cytometry" Lab Chip, 2018,18, 2055-
2064. Mar 3 2018.
[241] H.J. Choi, M. Fukui, B.T. Zhu, "Role of cyclin B1/Cdc2 up-regulation in the development
of mitotic prometaphase arrest in human breast cancer cells treated with nocodazole" PLoS
ONE, 2011, 6, e24312. Aug. 30 2011.
124
8 Appendix A- Cluster Migration in a Microfluidic Device
CTC advances have largely focused on single cell capture and analysis; however, further
characterization of clusters may provide enhanced diagnostic information. CTC clusters pose an
increased tumorigenic potential relative to single circulating cells. Thus, we designed a
microfluidic device to facilitate high efficiency capture of small tumor clusters (2- 7 cells). The
device is designed with long serpentine channels to create a pressure drop across a nozzle capture
region. PC3 and PC3M clusters are trapped in the nozzles with 75% and 72% capture efficiency,
respectively. Post- capture, clusters adhere to the matrix and migrate through 3D collagen-filled
micro-channels towards a chemotactic gradient. Highly tumorigenic PC3M cells exhibit faster
velocity and displacement within collagen relative to PC3 cells. These cells display prominent
directional migration within the micro-channels.
This chapter is presented:
B.J. Green1, B.T.V. Duong1, S.O. Kelley. Microfluidic Capture and Migration Analysis of
Prostate Cancer Clusters.
1 Equal contribution
B.J.G. conducted experiments, designed the device, analyzed data and wrote the manuscript.
B.T.V.D. conducted experiments, fabricated and designed the device and analyzed data. S.O.K.
supervised the study.
125
8.1 Introduction
CTCs may consist of single tumor cells or clusters of cells. Recent investigations have shown that
CTC clusters have potentially high capacity for metastasis.216 The clusters are defined groups of
tumor cells that travel together in the bloodstream. Clusters may originate from the primary tumor,
or consist of single cells that divide and aggregate during metastasis. This aggregation and division
is unlikely to occur in the circulation, but rather in host niche environments or within narrow
vessels.216 Clusters are not simply aggregated tumor cells, but are often composed of stromal cells
(fibroblasts), endothelial cells and myeloid cells.216
Despite great advances in CTC isolation and detection, little progress has been achieved for CTC
cluster isolation and characterization. CTC cluster capture presents significant challenges, as they
may break apart during blood processing steps or during fluidic trapping. Thus, the limited data of
CTC clusters in patients vary greatly according to tumor type, disease stage, and detection
platform.216
Antibody based methods are the most widely used capture technique for CTC clusters, where
clusters are targeted using EpCAM, but this method does not provide adequate capture
efficiency.216 Compared to single CTCs, clusters have smaller surface are to volume ratios, which
reduces their capture efficiency.
Size-based methods have also been used to separate CTC clusters from single cells, such as the
isolation by size of epithelial tumor cells (ISET) platforms. As an unbiased method, ISET is
believed to be more sensitive than antibody- based capture methods. ISET retains the natural status
of CTC clusters isolated from whole blood without antibody selection.216 The Cluster Chip
developed by Sarioglu et al. is based on microfluidic and antigen-independent capture techniques
to isolate clusters through specialized bifurcating traps under low shear stress conditions.218
Following capture of CTC clusters, it is relevant to design down-stream functional assays to
examine the migratory behavior of these cells through extracellular matrix. Cells typically migrate
due to directional cues from their environment, such as mechanical constrictions, chemotactic cues
or durotaxis gradients. Cells polarize along the direction of migration to generate actin-driven cell
traction forces.219 This process involves the relocation of the mitochondrial organizing center and
Golgi apparatus at the front edge of the cell, coordinated by small GTPases such as Cdc42.181 The
126
relocation of the Golgi to the leading edge of the cell provides membrane and associated proteins
necessary for proteolytic cleavage of the ECM or chemokine sensing.180
Commonly used migration assays include the Boyden chamber or gel invasion assays created from
isolated cells or tumor spheroids. These assays are end-point based and do not provide details on
the cell movement. Thus, methods which enable live-cell imaging and tracking are preferred.172
A study of the migration of mesenchymal and epithelial cancer explants was examined in vitro.
Invasion and migration of oral squamous cell carcinomas and ductal breast carcinomas were
monitored in a 3D bovine type I collagen matrix (1.5 mg/ml) using time-lapse video and cell
tracking.220 During the migration of cancer cells through the ECM, cells proteolytically digest the
collagen fibers or move in an amoeboid manner.221 These authors demonstrated that clusters of
tumor cells may move in a more organized and efficient manner compared to single cells.
There is a critical need for the design of CTC cluster devices that not only capture the cells, but
provide live cell analysis post- capture. The migration of small CTC clusters (2- 7 cells) in 3D has
not been studied in detail. Therefore, we designed a microfluidic device to capture small clusters
of cells and monitor their migration through collagen towards a chemokine.
We hypothesize that highly tumorigenic clusters will migrate in a more efficient manner compared
to less- tumorigenic clusters, and will exhibit faster velocities and displacement towards the
chemokine.
8.2 Results and Discussion
8.2.1 Prostate Cancer Cells and Clusters
PC3 and PC3M cells are chosen to monitor cluster migration. PC3 cells are metastatic prostate
cancer cell isolated from bone metastases, and represent a mesenchymal cell line. PC3M cells have
dual epithelial and mesenchymal properties (Figure 8.1).222 These cells are isolated from PC3 liver
metastases and are significantly more tumorigenic in mouse xenograft models compared to PC3
cells.240 In an in-vitro invasion assay, we demonstrated that PC3M cells ingest greater quantities
of fluorescently labeled- collagen relative to PC3 cells (Figure 8.2).
127
Figure 8.1 Flow cytometry analysis of epithelial, mesenchymal and migration markers in PC3 and PC3M cells. Vimentin, E-
Cadherin, Talin1, CXCR6, Active RhoA, Total RhoA, Active Cdc42, Total Cdc42, Active Rac1 and Total Rac1 were analyzed.
10,000 cells were analyzed per cell line.
Fluorescent intensity
Fluorescent intensity
Fluorescent intensity
Fluorescent intensity
Fluorescent intensity
Fluorescent intensity
Fluorescent intensity
Fluorescent intensity
Fluorescent intensity
Fluorescent intensity
Co
un
t
Co
un
t
Co
un
t
Co
un
t
Cou
nt
Cou
nt
Cou
nt
Cou
nt
Cou
nt
Cou
nt
128
Figure 8.2 Fluorescent- collagen uptake in PC3 and PC3M cells. Cells are plated on 1mg/ml FITC- conjugated
type I collagen. After 24 hours of culture, cells are released from the surface and analyzed with flow cytometry for
the ingested FITC collagen. 10,000 cells are analyzed per cell line.
Clusters of cells were obtained from cell culture (Figure 8.3A). Cell lines contained primarily small
clusters of 2-3 cell (20%), and lower percentages of clusters greater than 4-cells (3%). The
diameter of PC3 and PC3M non-adhered single cells were measured on average as 19.7 ± 0.3 and
17.5 ± 0.4 µm, respectively; whereas cluster diameters were 43 ± 1.6 and 42 ± 1.6 µm, respectively
(Figure 8.3 B-C). Clusters of cells express elevated levels of plakoglobin (γ-catenin) and
NCadherin (Figure 8.3 D-F).223 Plakoglobin and NCadherin are key components in cell-cell
adhesions.
8.2.2 Cluster Capture Device Design and Setup
We designed a microfluidic device to examine prostate cancer cluster migration through collagen
within micro-channels over 24 hours. The device was designed with serpentine channels and
nozzle capture sites to induce high flow (Figure 8.4). The capture sites are joined to a migration
channel leading to a chemokine reservoir. To facilitate monitoring of cluster migration, cells are
transfected with nucleus- GFP and Golgi-RFP insect virus baculovirus.
Initially, 1mg/ml collagen type I is introduced through the collagen inlet port to fill the migration
channels, and incubated at 37°C overnight to cause gelation.
Co
un
t
Fluorescent Intensity
129
Figure 8.3 Prostate cancer cluster characterization. A) Percent of clusters in PC3 and PC3M cells. Clusters range
in sizes from 2- 10 cells. n=821 cells. B) Diameter of PC3 and PC3M single cells. n=284 cells. C) Diameter of PC3
and PC3M 2-cell and 3-cell clusters. n=141 clusters. D) Immunofluorescence analysis of plakoglobin and NCadherin
in PC3M single cells and clusters. Cells are fixed and stained using plakoglobin- Alexa Fluor 647 and NCadherin-
FITC. n=52 cells. E) Relative fluorescence intensity of Plakoglobin and NCadherin in PC3M single cells versus
clusters. Fluorescent intensity is corrected for background.
Cells are serum-starved overnight in normal media with 0.5% FBS. Subsequently, media
containing 10% FBS and 500 ng/ml chemokine ligand 16 (CXCL16) is introduced through the
chemotaxis inlets to initiate a chemokine gradient. Prostate cancer cells are previously reported to
migrate towards CXCL16 gradients.224 We demonstrate that this gradient is maintained for 24
hours (Figure 8.5). Next, cells are introduced through the cell- loading inlet at 50µl/h in starvation
media. At this low flow rate, we achieved cluster capture efficiencies of 74.9 ± 17.4 % and 71.9 ±
3.1 % for PC3 and PC3M clusters, respectively (Figure 8.4G). Cells adhere in the capture sites for
5 hours in a temperature- and CO2- controlled environment on a live-cell imaging microscope
mg/L
D
=
E F
130
setup. Live-cell imaging commences after cells adhere, and we monitor migration of the cells
through the collagen substrate towards the chemokine (Figure 8.4).
Figure 8.4 Schematic of cluster capture. A) Overview of cluster device, with collagen inlet and outlet, CXCL16
chemokine solution inlet and outlet, and cell loading inlet and outlet. Briefly, type I collagen is introduced into the
collagen inlet 1-day in advance. Collagen travels through the migration channels and forms a stable gel when placed
20µm
E
Velo
city (
m/s
)
40µm
C
D
A
B
F
0
20
40
60
80
100
PC3 PC3M
Cap
ture
eff
icie
ncy (
%)
G
131
in an incubator at 37oC overnight. Cells are serum starved for 24 hours. Subsequently, the chemokine solution (500
ng/ml of CXCL16 in normal media) is introduced into the device at a flow rate of 600µl/h for 10 minutes. Cells are
loaded in capture sites, and adhere for 5 hours. During this time, the chemokine gradient is established across the
migration channels. Cells are then imaged using FITC, TRITC and brightfield channels for 24 hours as they travel
through the migration channels towards the chemokine gradient.
B) Zoomed in feature of the cluster device. Serpentine channels 75µm-wide create a pressure drop across the 15µm-
nozzles. This pressure drop results in efficient capture of clusters. Migration channels are filled with collagen (shown
in green). Clusters are shown in red.
C) Comsol modeling of fluid flow through the cell capture region with a flow rate of 50µl/h. D) Zoomed in nozzle
feature showing the high-velocity field through the center of the nozzle which provides a suction to capture the cells.
E) Brightfield image of a PC3M cluster trapped in the nozzle.
F) Schematic illustrating tumor clusters captured in the nozzle of the cluster device. Post- capture, cells migrate
through the micro-channels filled with collagen towards a CXCL16 gradient.
G) Capture efficiency of PC3 and PC3M cells in the cluster device at a flow rate of 50µl/h. This data was prepared by
B.J.Green and B.T.V. Duong.
Figure 8.5 Gradient distribution through micro-channels. Channels are filled with 1 mg/ml type I collagen.
Gradient is visualized using 2 mg/ml FITC dextran. A) FITC dextran intensity profiles along 500µm of the migration
channels. Intensity values include background subtraction. B) Fluorescent and brightfield images of FITC dextran
diffusing through collagen. PC3M cells are shown initiating migration through the channel towards a CXCL16
gradient. Cells are transfected with nucleus-GFP and Golgi-RFP.
0
80
160
240
320
400
480
560
640
63 125 188 250 313 375 438
40µm
B
A
1 hour
24 hours
Distance (µm)
Flu
ore
sce
nt
Inte
nsity
132
The cluster device is designed with long serpentine channels to create a pressure drop across the
nozzle. The flow rate within a micro-channel is given by Q = ΔP/R, where Q is the flow rate, ΔP
is the pressure drop across the channel, and R is the channel resistance.79 The resistance of a
rectangular micro-channel is described as:
Table 8.1 Fluid parameters used to determine the pressure drop across the cluster capture site.
Units Cluster
device
µ Viscosity (media
viscosity is
approximated to equal
that of water)
kg/m.s 0.001002
L Length of channel µm 2302
w Width of channel µm 75
h Height of channel µm 50
R Resistance kg/m4.s 5.03E+12
Q Flow rate µl/h 50
ΔP Pressure drop across
cluster capture site
Pa
(N/m2) 70
The flow profiles are confirmed with Comsol Multiphysics modeling (Figure 8.4). The pressure
drop and narrowing constriction created across the nozzle creates high flow velocities (Figure
8.4C,D).
The optimal nozzle width is optimized as 15µm. With nozzle dimensions of 15µm x 35µm x 50µm
(width x length x height), single cells pass through as their volume is less than the nozzle volume.
(3,590 vs. 26,250 µm3, respectively). In comparison, clusters are captured as their volume exceeds
the nozzle volume (38,792 vs. 26,250 µm3, respectively). The volume of single cells and clusters
are calculated from the average diameters (Figure 8.3 B-C). The pressure drop of 70 Pa across the
nozzle, combined with the restricted area enables for efficient capture of clusters in the device.
The migration channels are designed with dimensions of 40µm x 500µm x 10µm (width, length,
height) (Figure 8.4F). The wide micro-channel dimensions are chosen to allow cluster re-
positioning during migration.
(7)
Figure 8.6. Schematic of the
cluster capture device showing
the cluster capture site and the
nozzles. The red line indicates
the length of the serpentine
channel. The resistance in the
channel creates a pressure drop
ΔP across the nozzle.
Cluster capture site
133
Cluster migration experiments demonstrate that PC3M cluster size is on average greater than
PC3 clusters (Figure 8.9C). This is consistent with their higher expression of cell-cell adhesion
protein E-Cadherin (Figure 8.1).
8.2.3 Prostate Cancer Cell Migration Results
The migration of prostate cancer clusters can provide relavent diagnostic information, and initiate
the development of therapeutics which prevent metastasis. PC3 and PC3M cell migration is
compared between the cluster capture device and the commercially available Ibidi chemotaxis
chambers. The comparison enables us to examine the differences between migration in a micro-
channel versus an open area. The dimensions of the Ibidi migration areas are 1mm x 2mm x 70µm
(width, length, height). The polarization of the cells was tracked by recording the position of the
Golgi relative to the nucleus along the direction of migration (Figure 8.7C). On serum- coated Ibidi
chemotaxis devices, we observe that PC3 single cells migrate rapidly towards the chemokine
relative to PC3M single cells, and displace greater distances (Figure 8.7A,D,H). Faster velocity is
correlated with majority of front polarization. Rapid single cell velocities of PC3 cells have been
previously reported.225
On collagen- coated Ibidi chemotaxis devices, both cells significantly decrease their velocities.
This migration data corresponds with literature, where PC3 cells migrate at speeds of 10µm/h
through type I collagen towards a growth factor gradient.226
PC3M cells exhibit faster velocities, front polarization and greater displacement relative to PC3
cells on collagen surfaces. This data suggests that the inputs received from the collagen fibers
greatly influence the migration tendencies of prostate cancer cells. Collagen fibers present
obstacles on the surface, and cells decrease their speed as they either digest the fragments through
collagenolysis, or deform and move via amoeboid migration.221 In addition, the randomly- oriented
fibers provide mechanical cues for migration. As shown in Figure 8.8, cells tend to align along
collagen fibers. These factors may contribute to reduction in cell migration speed.
The ability for PC3M cells to move faster through a collagen substrate may be due to their
enhanced collagen uptake relative to PC3 cells (Figure 8.2). In addtion, it has been previously
reported that invasive prostate cancer cells have increased deformability227, thus, PC3M cells may
also have the ability to switch between mesenchymal and amoeboid modes of migration.
134
cc
0
20
40
60
80
100
PC3 PC3M0
20
40
60
80
100
PC3 PC3M
0
10
20
30
40
50
PC3 PC3M
0
10
20
30
40
50
PC3 PC3M
*
0
5
10
15
20
25 *
*
*
PC3 PC3M PC3 PC3M
-6
0
6
12
18
24
Dis
pla
cem
ent
(µm
) P
erc
enta
ge o
f C
ells
V
elo
city (
µm
/h)
Front polarized cells
Back/lateral polarized cells
0
20µm 20µm
F G
A B
D E
H I
Front
Lateral
Back
Direction of Migration
Nucleus Golgi
Lateral
C
Collagen Substrate Serum- coated Substrate
Collagen Substrate Serum- coated Substrate
Collagen Substrate Serum- coated Substrate
135
Figure 8.7 Single cell migration and quantification in Ibidi chemotaxis devices. A) and B) Velocity of PC3 and
PC3M cells on a serum-coated substrate versus a substrate coated with 1 mg/ml type I collagen, respectively. The
average velocity is obtained from 1200 and 1120 cells, respectively over 24 hours. C) Golgi polarization of a cell
quantified along the direction of migration. Golgi (red) is monitored during migration relative to the position of the
nucleus (green) and characterized as front, back or lateral.
D) and E) Percentage of PC3 and PC3M cells which are front, or back/lateral polarized while migrating on a serum-
coated substrate versus a substrate coated with 1 mg/ml type I collagen, respectively. Data is obtained from 245 and
51 cells, respectively.
F) and G) Immunofluorescence and brightfield images of representative front- polarized PC3M cell and back-
polarized PC3 cell, respectively. Cells are transfected with Golgi-RFP and nucleus-GFP for migration and polarization
tracking.
H) and I) Displacement of cells towards a 500ng/ml CXCL16 gradient over 24 hours, on a serum-coated substrate
versus a collagen coated substrate. The average displacement is obtained from 1200 and 1120 cells, respectively.
Box plots represent 25 and 75 percentiles, squares represent the mean and error bars are the standard error of the mean.
Statistics are performed with two-sample t-test, *p<0.05.
Cells are serum starved for 24 hours. Chambers are coated with 1mg/ml type I collagen prior to cell seeding. Cells are
seeded in the left reservoir. The following day, the chemokine solution containing 500ng/ml CXCL16 and 10% FBS
is added to the right chemotaxis reservoir. Cell migration is recorded over 24 hours in a live-cell temperature controlled
unit.
Figure 8.8 PC3M cells aligned along collagen fibers. Cells are transfected with nucleus-GFP and Golgi-RFP.
Cells are plated on the surface of collagen inside the Ibidi chemotaxis chambers.
Next, we examined the migration patterns of prostate cancer cells through the micro-channels of
the cluster device (Figure 8.9). PC3 or PC3M clusters typically break apart during migration, while
maintaing some contact with neighbors (Figure 8.9E). Similar to the Ibidi chemotaxis results for
collagen coated surfaces, PC3M cells migrate at faster velocities and displace greater distances
inside collagen, relative to PC3 cells.
40µm
136
The displaced distances in the cluster device, for PC3M cells migrating on collagen, are
significantly higher than the Ibidi chambers (122.9 ± 10.1 µm versus -0.50 ± 1.08 µm,
respectively). Likewise, the velocities in the cluster device, for PC3M cells migrating on collagen,
are higher than the Ibidi chambers (24.0 ± 2.3 µm/h versus 16.2 ± 1.5 µm/h). Cells migrating in
the Ibidi chambers experience cues from multiple neightboring cells, which provide competing
signals for directional migration. In contrast, the cluster device enables us to monitor individual
cluster migration in the absence of competing paracrine signals, likely leading to greater
directional displacement.
Several intracellular signals are activated at the leading and trailing edges of migrating cells. At
the leading edge, signals drive preferential activation of small GTPases Rac, RhoA and Cdc42.228,
229 PC3M cells have higher levels of migration and polarization markers (Figure 8.1), such as
RhoA, Cdc42 and Rac1, relative to PC3 cells, possibly contributing to their enhanced migration.
137
Figure 8.9 Prostate cancer cluster migration through micro-channels. A) Individual cell velocities are recorded
in the clusters. B) Displacement of PC3 and PC3M cells towards chemokine CXCL16. C) Number of cells per cluster
for PC3 and PC3M clusters migrating through the cluster device. n=64 cells (24 clusters). D) FITC and brightfield
image of PC3 cell migrating through collagen micro-channel. Cell extends membrane protrusions during migration.
E) Brightfield image of PC3 2-cell cluster migrating through collagen. F) PC3 6-cell cluster migrating through
collagen. Front cell breaking off from cluster. Cells were transfected with nucleus- GFP and Golgi- RFP. Scale bars
are 20µm. Arrow indicates direction of migration. Box plots represent 25 and 75 percentiles, squares represent the
mean and error bars are the standard error of the mean. Statistics are performed with two-sample t-test, *p<0.05.
D
A
A
0
10
20
30
40
50
PC3 PC3M
Ve
locity (
µm
/h)
*
B
E
A
0
1
2
3
4
5
6
PC3 PC3M
C
Nu
mb
er
of
Ce
lls p
er
Clu
ste
r *
F
A
A
D
138
8.3 Conclusions
In conclusion, we have presented a microfluidic cluster device that allows for high capture
efficiency of prostate cancer clusters. Post- capture, clusters migrate through collagen- filled
micro-channels towards a CXCL16 gradient. Using live-cell imaging over 24 hours, we quantify
the migration, displacement and polarization of the cells. The cluster device is compared to Ibidi
chemotaxis chambers used for single cell migration analysis. We demonstrate faster velocities and
displacements in the cluster device micro-channels compared to the Ibidi environment. In addition,
we observe that PC3M clusters migrate more efficiently through collagen relative to PC3 clusters,
and that they have greater cluster sizes, and higher levels of Cdc42, Rac1 and RhoA. Together
these factors may correlate with their enhanced tumorigenic capacity in vivo.
This study provides a unique approach to examining small cluster migration through micro-
channels, and can be subsequently applied to CTC clusters.
The analysis of the tumor invasion front in a collectively migrating unit of cells is of particular
interest, because these cells could be directed for therapy.230 Future work could examine the
dynamic interplay between cells within a cluster, how they interact with eachother over the
migration period, and whether they secrete chemokines to encourage directional migration. The
plasticity of the migrating cancer cells could be examined by inhibiting matrix metalloproteinase
activity, and observing whether cells can readily switch from proteolytic to amoeboid migration.
Overall, efficient capture and downstream analysis of cluster migration can be directed towards
enhancing cancer diagnostics.
8.4 Methods
8.4.1 Cell Culture
Human prostate cancer cells, PC3 and PC3M were obtained from Dr. Alison Allan, London Health
Sciences Centre, London, ON. PC3 cells were cultured in F12K media (ATCC) supplemented with
10% FBS and 1% penstrep while PC3M were cultured in RPMI-1640 (ATCC) supplemented with
10% FBS and 1% penstrep. Cells were cultured at 37°C and 5% CO2.
139
8.4.2 Flow Cytometry
Cells were released from tissue culture dishes using 0.25% trypsin/EDTA (Sigma-Aldrich, US)
and incubated with blocking buffer (PBS + 1% BSA) for 30 min. For each cell line, 2×105 cancer
cells were fixed and permeabilized using 4% paraformaldehyde (Sigma-Aldrich, US) and 0.2%
Triton X (Sigma-Aldrich, US) in PBS, respectively. Cells were then washed with PBS and
suspended in PBS containing 1% BSA and 0.1% Tween20, and incubated with E-Cadherin- Alexa
Fluor 647 (Abcam), Vimentin Alexa Fluor- 647 (Abcam), CXCR6 Alexa Fluor 647 (BioLegend),
active Cdc42 (New East Biosciences, US), total Cdc42 (Abcam, US), total Rac1 (Abcam, US),
total RhoA (Abcam, US), active Rac1 (New East Biosciences, US), active RhoA (New East
Biosciences, US), and mouse monoclonal Talin1 Alexa Fluor 647 (Abcam, US) for 1 h at RT.
Cells were washed with 1% BSA in PBS and stained with goat anti- mouse Alexa Fluor 647
(Invitrogen, US) and goat anti-rabbit Alexa Fluor 647 (Invitrogen, US) for 30 min at RT.
Cells were washed and resuspended in 1% BSA in PBS. Samples were then injected into a BD
FACS Canto flow cytometer (BD Biosciences, US) and measurements were plotted as histograms
for AF647. A total of 10,000 cells were analyzed per cell line.
8.4.3 Immunocytochemistry
Cells were washed with 1% BSA in PBS and fixed with 4% paraformaldehyde solution (Sigma-
Aldrich, US) followed by 0.2% Triton X-100 (Sigma-Aldrich, US) for permeabilization. Cells
were stained with Plakoglobin-Alexa Fluor 647 (Novus Biological) and NCadherin- FITC
(BioLegend) (1:50 dilution) for 1 hr at RT in PBS containing 1% BSA and 0.1% Tween20. Cells
were then washed with 1% BSA in PBS and imaged using a fluorescent Nikon TiE eclipse
microscope. Images were acquired using a 50X objective.
8.4.4 Device Fabrication
Chips were fabricated using Poly(dimethoxysilane) (PDMS, Dow Chemical, US) soft-lithography.
Masters were fabricated on silicon substrates and patterned in SU-8 3050 (Microchem, US) to
create 50µm height channels. Briefly, SU8-3050 was spun on silicon wafer at 500 rpm for 10s ,
then 3000 rpm for 30s. The wafers are heated at 65°C for 20 seconds then 95°C for 20 minutes.
With a chromium mask, the wafers are UV exposed for 20 seconds (using mask aligner) with
“flood exposure setting". This step prints the cell loading channel and chemotaxis reservoir. The
140
wafers are post- baked for 5 minutes at 95°C. The wafers are then developed for 4 minutes using
SU-8 developer followed by a quick wash with IPA and ddH2O. They are then heated at 95°C for
1 minute to evaporate all excess solvent.
The migration channel height is 10µm, therefore we used a second spin- step. The wafer is spin-
coated with SU8 3010, 500rpm for 10s, 3000rpm for 30s, then baked for 5 minutes at 95°C. With
a chromium mask, we exposed the wafers again but for 12s using the "Hard contact setting”. This
step requires mask alignment. The wafers are post baked for 2 minutes at 95°C, and developed for
2 minutes using SU-8 developer, then IPA and ddH2O wash. The last step involves a hard bake at
150°C for 5-7 minutes.
PDMS replicas were poured on masters and baked at 67°C for 2 hours. After peeling the replicas,
holes were pierced to connect the tubing. PDMS replicas were attached to no. 1 glass cover slips
using a 30 second plasma treatment and left to bond overnight. Afterward, the silicon tubing was
attached to the inlet and outlet of the device. Prior to use, devices were conditioned with 0.1%
Pluronic F68 (Sigma-Aldrich, US) in phosphate-buffered saline (PBS) for 1 h, to reduce
nonspecific adsorption.
8.4.5 Device Setup
Type I collagen (1 mg/ml) (Gibco) is prepared in PBS to achieve a pH of 7.5. Collagen is infused
into the device at a flow rate of 200µl/h for 1 hour, in a cold room. This step fills the migration
channels with collagen. Next, 1% BSA in PBS is withdrawn from the cell loading channels at
200µl/h for 30 minutes to remove collagen from the capture sites. The devices are incubated inside
a water bath at 37°C overnight to allow the collagen to gel inside the migration channels. Cells are
serum-starved overnight in regular media with 0.5% FBS. The following day, 500 ng/ml CXCL16
(ProSpec) is combined with normal cell culture media to create the chemokine solution. The
chemokine solution is introduced into the chemokine reservoirs at a flow rate of 600µl/h for 20
minutes. Cells are released from 12-well dishes at a concentration of 4 x 105 cells/ml, with trypsin/
EDTA (Sigma). 100µl of the cell suspension is used for cluster capture. The device is placed in a
temperature- and CO2- controlled live cell imaging platform (Axio Observer, Zeiss). Cells and
clusters are introduced into the cell loading sites at a flow rate of 50µl/h until capture sites are
filled. Clusters are left to adhere to the matrix for 5 hours in the live-cell unit. Following adhesion,
live-cell imaging is initiated.
141
8.4.6 Golgi Transfection
PC3 and PC3M cells were transiently transfected with Nucleus- GFP (CellLight, Nucleus-GFP,
BacMam, Invitrogen) and Golgi-RFP (CellLight Golgi RFP, BacMam, Invitrogen) for
visualization of the nucleus and the Golgi apparatus during live cell imaging. Cells were prepared
in 6-well dishes. The transfection was conducted in a low volume (~500 µl) of media to increase
transfection efficiency. Cells were treated with BacMam enhancer kit (Invitrogen) prior to addition
of Golgi transfection agent. The Golgi transfection agent was added at a volume of 4 µl per 500
µl media and incubated with cells for 24 h. Post-incubation, the media was exchanged for fresh
media and cells were imaged.
8.4.7 Live Cell Microscopy
Migration analysis was acquired using an inverted Zeiss wide-field microscope (Axio Observer,
Zeiss) equipped with a CCD camera (Axiocam 506 mono, Zeiss) and an incubation chamber to
control temperature and CO2. Images were collected using an EC Plan-Neofluar 20x objective.
(Zeiss). Videos were obtained for a period of 24 h, and images were captured at 20 minute
intervals. At each time of measurement, a transmission and fluorescent images of the nuclei and
Golgi of the cells were acquired using a brightfield, FITC, and TRITC filter set. Focal drift during
the experiments was avoided by using the autofocus system of the microscope.
8.4.8 Collagen Uptake Assay
Fluorescein isothiocyanate (FITC)-labelled type I collagen (1mg/ml) (US Biologicals, US) was
incubated in 24- well dishes for 24 hours at 4oC, achieving even coating of the surface. Excess
matrix solution was removed prior to cell loading onto the surface. Cells were serum-starved with
0.5% FBS in their respective media for 8 hours, then released using 0.25% trypsin/EDTA (Sigma
Aldrich, US) and plated on the surface of the collagen matrix. Cells were cultured on the matrix
for 24 hours at 37ºC and 5% CO2 in their respective media with 10% FBS.
Post-culture, cells were released from the collagen matrix subsequent to incubation with 10mg/ml
collagenase (Sigma Aldrich, US) for 10 minutes. The ingested collagen is fluorescently labelled;
thus, the cells can be identified by immunofluorescence based methods, such as flow cytometry
and fluorescence microscopy.
142
8.4.9 Dextran Gradient
The cluster device was prepared with collagen, as mentioned in the Device Setup section. FITC
Dextran (40kDa) (Sigma Aldrich) was prepared at a concentration of 2mg/ml in PBS. The dextran
solution was introduced into the chemotaxis reservoirs at a flow rate of 600µl/h for 20 minutes.
The device was placed in the 37°C incubator in a humid environment. At regular time intervals at
1 h and 24 h, we imaged the migration channels using a fluorescent Nikon TiE eclipse microscope
with an automated stage controller and an Andor camera. FITC and brightfield channels were used
to record the dextran diffusion over the channels. The captured images were analyzed using ImageJ
software. The fluorescent intensity was measured at regular intervals along the migration channel.
Background intensity was subtracted for quantification.
8.4.10 Golgi Quantification
Golgi quantification analysis were performed manually using Nikon Instruments Software. The
polarization of the cells was determined by tracking the position of the Golgi relative to the nucleus
over the time lapse images. The position of the Golgi was recorded as back, front or lateral relative
to the direction of migration.
8.4.11 Capture Efficiency
The capture efficiency is quantified as the number of trapped clusters in a 20- nozzle cluster device.
𝐶𝑎𝑝𝑡𝑢𝑟𝑒 𝐸𝑓𝑓𝑖𝑐𝑖𝑒𝑛𝑐𝑦 𝑜𝑓 𝐶𝑙𝑢𝑠𝑡𝑒𝑟𝑠 = Number of Clusters Captured in Capture Sites
Number of Total Capture Sites ……………...(8)
8.4.12 Ibidi Chemotaxis Chamber Migration Analysis
Ibidi chemotaxis migration chambers (Ibidi, Germany) were used to examine the migration of
single prostate cancer cells. 50,000 cells were introduced into one side of the chamber in normal
media, and adhered overnight at 37°C and 5% CO2. Starvation media with 0.5% FBS is introduced
on the second day, and cells are left overnight. The chemokine solution of 500ng/ml CXCL16
(ProSpec) in normal 10% media is introduced into the opposite reservoir. Devices are placed on
the live-cell imagine platform (Axio Observer, Zeiss) and imaged overnight in a temperature- and
CO2- controlled environment.
143
In the collagen experiments, the cell-loading side of the device is pre-coated with type I collagen
(1 mg/ml) (Gibco) and gelled at 37°C in a humid environment overnight. The following day,
serum-starved cells are plated on the surface of the collagen matrix and adhered overnight. Cells
are then imaged as described above.
8.4.13 Cell Diameter, Velocity and Displacement Measurements
The cell and cluster diameters were measured using Nikon Instruments Software and ImageJ. Non-
adhered cells were plated on a glass coverslip at low confluency. The velocity and displacement
of cells within the Ibidi chemotaxis chambers were quantified using particle tracking algorithm of
Imaris (Bitplane, US). The track mean speed and the displacement of cells was obtained for each
video over 24 hour period. Time-lapse NIS videos were uploaded into Imaris, and the voxel size
and time interval were adjusted before particle tracking.
8.4.14 Data Representation and Statistical Analysis
Boxes in all box plots extend from the 25th to the 75th percentiles, with a line at the median and a
square representing the mean. Error bars associated with box plots represent standard error of the
mean. Statistical comparison of population means were performed using t-test for normally
distributed populations and the nonparametric Mann Whitney test. *p <0.05 is considered as
significant.
144
9 Appendix B – Effect of In-vivo Lung Perfusion on Lung Metastases and Circulating Tumor Cells in Rat Sarcoma and Colorectal Cancer Models
Lung cancer represents a significant problem worldwide that is commonly associated with late
diagnosis.9 Thus, circulating tumor cell (CTC) detection and characterization can contribute to the
development of effective therapies. We captured and analyzed CTCs from the blood of rat sarcoma
and colorectal cancer models over the course of 25 days using the four zone velocity valley device.
Rats receive chemotherapy either systemically or locally through in- vivo lung perfusion (IVLP)
at day 4. We observe a reduction in CTCs, lung weight and pulmonary nodules in IVLP- treated
rats relative to the control.
This chapter is currently under preparation as two manuscripts:
#1 Jin Sakamoto, MD1, Brenda J. Green, MASc1, Pierre-Benoit Pagès, MD, PhD, Pedro Reck dos
Santos, MD1, Ilker Iskender, MD , Manyin Chen, MD, Neesha Dhani, MD, Thomas K. Waddell,
MD, PhD, Shaf Keshavjee, MD, MSc, Mingyao Liu, MD, Shana O. Kelley, PhD, Marcelo Cypel,
MD, MSc. In-vivo Lung Perfusion on Lung Metastases and Circulating Tumor Cells in a Rat
Sarcoma Model.
and
#2 Jin Sakamoto, MD1, Brenda J. Green1, MASc, Pierre-Benoit Pagès, MD, PhD1, Pedro Reck
dos Santos, MD, Ilker Iskender, MD, Manyin Chen, MD, Minyao Liu, MD, Shaf Keshavjee, MD,
MSc, Thomas K. Waddell, MD, PhD, Shana O. Kelley, PhD and Marcelo Cypel, MD, MSc. In-
vivo Lung Perfusion with Oxaliplatin and 5-Fluorouracil for the Treatment of Colorectal Cancer
Pulmonary Micrometastases.
1 Equal contribution
Nature of collaboration: J.S. performed animal experiments at the University Health Network
facility, including injections, blood withdraw, sacrificing animals, lung weight measurements and
histology analysis. B.J.G. analyzed circulating tumor cells from the blood of rats using a
microfluidic device fabricated for this purpose, performed data analysis and summarized the study.
P-B.P aided in manuscript writing and data analysis. P.R.S., I.I., M.C., N.D., T.K.W, S.K., M.L,
S.O.K, M.C aided in study design and supervision.
145
Collaborators:
Collaborators include Jin Sakamoto, MD, PhD and Principal Researcher Marcelo Cypel, MD,
Msc, Latner Thoracic Research Laboratories, University Health Network.
9.1 Introduction
Soft tissue sarcoma is a malignant tumor that begins in connective and supporting tissue. Distant
metastases of sarcomas, including the lung, are the most common cause of death from these
mesenchymal tumors.231 In comparison, colon cancer is a systemic disease in 19% of patients and
metastasizes most often to the lung and liver.232
Current treatment of lung metastases involves surgical resection and chemotherapy. Previous
reports demonstrate that 20% of metastatic nodules are not detectable pre-operation.232 This has
led to the development of minimally invasive approaches.231
Systemic exposure to chemotherapy can cause adverse effects, including the depletion of red blood
cells (anemia) and white blood cells (leukopenia) due to immune targeted destruction of blood
cells in circulation or bone marrow suppression.233 In addition, systemic chemotherapy may not
target all metastatic sites, due to low concentrations in lungs.
In vivo lung perfusion (IVLP) is a minimally invasive surgical technique that can potentially
overcome the limitations of systemic chemotherapy. Through IVLP, high doses of chemotherapy
are administered exclusively to the lungs without systemic exposure.234 This technique is reported
to deliver 25 times higher drug concentration relative to intravenous (IV) delivery. The IVLP
technique has been previously demonstrated on six Yorkshire pigs.234, 235 Animals were subjected
to a 4-hour period of left lung IVLP followed by 4 hours of reperfusion. This procedure was also
used in a Phase I clinical study with patients with lung metastases. There was no measurable lung
injury due to the IVLP technique. In this study, we examine the effectiveness of IVLP-
administered chemotherapy in rats by measuring CTCs, lung weight and lung histology from
whole blood.
9.1.1 Proposed Research
The objective of this study is to evaluate the anti-cancer efficacy of different drug regimens using
a novel IVLP rodent model. Fisher rats are injected with colon cancer cells (epithelial RCN-9 cells)
146
or sarcoma cells (mesenchymal MCA cells), where both could metastasize to the lung. The
effectiveness of the therapy is determined by measuring CTCs, lung weight and lung histology.
The timeline of the study is illustrated below.
Day 0 Day 4 Day 12 Day 25
Inject rat cells
(colorectal cancer or
sarcoma) into rats
2-5 million cells
IVLP or systemic
chemotherapy treatment.
At this point the CTCs
spread to lungs
Blood Drawa (0.5-
1ml)
Sacrifice and Autopsy. Collect
blood after 25 days (2-3ml).
a Not all rats had blood drawn at Day 12 time point. This midpoint was included later in the
study.
Specific Objectives:
1. Examine whether the number of CTCs/ml increases during cancer progression
2. Examine whether chemotherapy treatment reduces the number of CTCs/ml
3. Determine whether IVLP (local) chemo is as effective as systemic chemo at reducing the
number of CTCs/ml
4. Determine if the epithelial and mesenchymal properties of CTCs change over the course of
treatment.
Rats with sarcoma- induced lung cancer and RCN-induced lung cancer are studied over a period
of 25 days. Rats are treated with chemotherapy 4 days after cancer cells are injected, which is
administered systemically or locally through IVLP. Blood samples are drawn at day 0, day 12 and
day 25. Circulating tumor cells are captured using the velocity valley CTC capture chip. CTCs
obtained from rats with sarcoma- induced lung cancer are captured with a combination of EpCAM-
and NCadherin- nanoparticles and stained with vimentin (mesenchymal marker found in sarcoma
cells). CTCs obtained from rats with RCN-induced lung cancer are captured with EpCAM
nanoparticles and stained with cytokeratin (epithelial marker).
Blood drawn
147
9.2 Results and Discussion
9.2.1 Study I – CTC Analysis from a Rat Sarcoma Model
Sarcoma cancer cells are initially characterized using flow cytometry (Figure 9.1). Results show
that sarcoma cells have relatively low levels of EpCAM, and higher expression levels of
cytokeratin and vimentin.
Sarcoma Cells
EpCAM
ECad C
K
NCad Vim
1
10
100
1000
10000
Med
ian
A
bso
rban
ce
Figure 9.1 Characterization of Sarcoma (MCA) cells. Cells are tagged with EpCAM-647, E-Cad-647, CK-APC,
Ncad-488 or Vim-488. Cells are analyzed with Flow Cytometry with 10,000 cells counted per group. Median
fluorescence of the corresponding fluorophore are recorded. Fluorescence intensities are normalized to the unstained
control. n=3.
Rats that have sarcoma- induced lung cancer are treated with chemotherapy through localized
IVLP technique and systemic administration. CTCs were examined over the 25-day time course
of the experiment (Figure 9.2). Tumor-free control rats had a false positive CTC count of 2.5 ± 2.5
Vim+ cells/ ml.
Rel
ativ
e fl
uo
resc
ence
in
ten
sity
Figure 9.2 Circulating tumor cells captured from
the blood of rats with Sarcoma- induced lung
cancer. Chemotherapy is administered 4 days after
injection via the IVLP surgical technique or via
systemic administration. Chemotherapy consists of
doxorubicin plus ifosfamide (Doxifos) or
doxorubicin alone (Dox). Rats are sacrificed after 25
days and blood samples are processed through the
velocity valley CTC capture chip. CTCs are captured
with a combination of EpCAM and NCadherin
nanoparticles and stained with Vimentin-488 with a
capture flow rate of 600 µl/h. CTCs are identified as
Vim+CD45-DAPI+. (Two-sampled t-test *p<0.05).
n=4. Dots represent individual rat samples.
148
We observed high numbers of CTCs in rats with sarcoma- induced lung cancer (No Treatment
group). These levels were significantly reduced with IVLP doxorubicin treatment; and this
reduction was not observed with systemic Doxifos treatment after 25 days (Figure 9.2).
In addition to the CTC data, we observed that IVLP administration of doxorubicin reduced the left
lung weight significantly compared to the No Treatment control after 25 days of treatment (649.0
± 193.2 mg vs. 335.8 ± 34.85 for No Treatment control versus IVLP treated rats, respectively)
(data not shown). Metastatic nodules (diameter > 200µm) were observed in the control. However,
we found essentially no tumor nodules in the Systemic Doxifos and IVLP Dox groups.
Next, we examined the distribution of sarcoma cells in the velocity valley CTC capture device.
The zone distribution can provide information relating to the epithelial and mesenchymal
properties of the cancer cells.
Figure 9.3 Sarcoma CTC zone distribution in the velocity valley microfluidic chip. (A) Sarcoma cells captured
in the 4 zones. Cells are stained with DAPI. The capture efficiency is 62%. (B) CTCs obtained from rats with sarcoma-
induced lung cancer after day 25. (C) CTCs obtained from rats with sarcoma- induced lung cancer treated with
systemic doxorubicin plus ifosfamide (Doxifos) after 25 days. (D) CTCs obtained from rats with Sarcoma-induced
lung cancer treated with IVLP doxorubicin (Dox) after 25 days. All CTCs are captured in the velocity valley CTC
chip with a combination of EpCAM nanoparticles and NCadherin nanoparticles at 600µl/h. CTCs are stained with
vimentin-488.
0
10
20
30
1 2 3 4
In Vitro DAPI Stained
In Vivo Vimentin Day 25
Number of CTCs
A B
In Vivo Systemic Doxifos Day 25
Number of CTCs
C
0
10
20
30
1 2 3 4
In Vivo IVLP Dox Day 25
D
Zone
Zone Zone
Zone
149
In vitro, sarcoma cells were distributed in each zone of the velocity valley chip. However, CTCs
captured from rats with sarcoma-induced lung cancer after 25 days are distributed mostly in zone
3 and zone 4 (Figure 9.3). The shift to later zones observed in vivo may be due to reduced epithelial
properties, relative to in vitro cells. IVLP chemotherapy causes a reduction in CTCs in all zones,
whereas systemic chemotherapy does not significantly reduce the zone- 4 population.
9.2.2 Study I – CTC Analysis from a Rat Colorectal Cancer Model
RCN colorectal cancer cells are initially characterized using flow cytometry (Figure 9.4). Flow
cytometry results show that RCN-9 cells also have relatively low levels of EpCAM, and higher
expression levels of cytokeratin. In order to ensure high capture efficiency of RCN-9 cells, we
reduced the capture flow rate in the velocity valley device to 300µl/h.
RCN-9 Cells
EpC
AM
ECad C
K
NCad
Vim
1
10
100
1000
Med
ian
A
bso
rban
ce
Rats with RCN- induced lung cancer were treated with chemotherapy through localized IVLP
technique, and systemic administration. Circulating tumor cells were examined over the 25-day
time course of the experiment (Figure 9.5). Tumor-free control rats had a false positive level of
2.5 ± 1 Ck+ cells/ ml.
Figure 9.4 Characterization of RCN-9 cells. Cells
are tagged with EpCAM-647, E-Cad-647, CK-APC,
Ncad-488 or Vim-488. Cells are analyzed with Flow
Cytometry with 10,000 cells counted per group.
Median fluorescence of the corresponding
fluorophore are recorded. Fluorescence intensities
are normalized to the un-stained control. n=3.
Rel
ativ
e fl
uo
resc
ence
in
ten
sity
150
Elevated numbers of CTCs were observed in rats with RCN- induced lung cancer. Similar to the
sarcoma- study, we noticed that the IVLP oxaliplatin treatment causes a reduction in CTCs that
was not observed with systemic FOLFOX (Figure 9.5).
IVLP administration of oxaliplatin caused a reduction in lung weight relative to the control after
25 days of treatment (656.1 ± 236.8mg vs. 259 ± 44.8 mg for No treatment control vs. IVLP
oxaliplatin treatment, respectively). Metastatic nodules (diameter > 200µm) were observed in the
lung parenchyma and on its surface for the control and systemic FOLFOX group, but there were
no significant tumors in the IVLP oxaliplatin group (data not shown).
The distribution of RCN cells in the velocity valley CTC capture device is assessed in Figure 9.6.
Figure 9.5 Circulating tumor cells captured from
the blood of rats with RCN- induced lung cancer. Chemotherapy (FOLFOX or oxaliplatin) is
administered through IVLP 4 days after injection or
via systemic administration. Rats are sacrificed after
25 days and 1ml blood samples are processed
through the velocity valley CTC capture chip,
captured with EpCAM nanoparticles with a flow rate
of 300 µl/h. CTCs are stained with Cytokeratin-488
and identified as CK+CD45-DAPI+. n=3. (Two-
sampled t-test *p<0.05). Dots represent individual
rat samples.
151
Figure 9.6 RCN CTC zone distribution in the velocity valley microfluidic chip. (A) RCN cells captured in the 4
zones. Cells were stained with DAPI. The capture efficiency is 72%. (B) CTCs obtained from rats with RCN-induced
lung cancer at day 25 (C) CTCs obtained from rats with RCN-induced lung cancer at day 25. Rats were treated with
systemic FOLFOX. (D) CTCs obtained from rats with RCN-induced lung cancer at day 25. Rats were treated with
oxaliplatin administered through IVLP. CTCs were stained with cytokeratin-488.All CTCs are captured in the velocity
valley CTC chip with EpCAM nanoparticles at 300µl/h.
RCN cells in vitro distribute evenly in all 4 zones of the CTC microfluidic chip. This suggests that
the cell population is a heterogeneous mixture of cells, which express low, medium and high levels
of EpCAM. CTCs captured from rats with RCN-induced lung cancer after day 25 were found
mostly in zone 3 and zone 4 (Figure 9.6 B and C). This suggests that the cells adopt a more
mesenchymal phenotype in vivo. IVLP treatment caused a reduction in zone 3 and zone 4 CTCs,
that was not observed with systemic chemotherapy. Fluorescence images of sarcoma cells, RCN
cells and rat CTCs are illustrated in Figure 9.7.
0
10
20
30
1 2 3 4
0
10
20
30
1 2 3 4
In Vitro DAPI Stained
In Vivo Cytokeratin Day 25
Number of CTCs
A B
In Vivo Systemic FOLFOX Day 25
Number of CTCs
C In Vivo IVLP Oxaliplatin Day 25
D
Zone
Zone Zone
Zone
152
Figure 9.7 Immunostaining of rat cancer cells.
(A-B) Sarcoma and RCN cancer cells.
Fluorescence microscopy images of sarcoma and
RCN cells. (A) Sarcoma cells stained with DAPI,
Vimentin-488. (B) RCN cells stained with DAPI
and Cytokeratin-488.
(C-H) CTCs and white blood cell
immunofluorescence staining. Fluorescence
microscopy images of RCN and sarcoma CTCs.
(C) RCN CTC stained positive for DAPI and CK-
488 and negative for CD45-647. (D) RCN CTC
stained positive for DAPI and Vim-488 and
negative for CD45-647. (E) Sarcoma CTC stained
positive for DAPI and Vim-488 and negative for
CD45-647. (F) White blood cell stained positive for
DAPI and CD45-647, and negative for CK-488.
(G-H) Apoptotic cancer cells. (G) RCN CTC
stained positive for DAPI, M30-Orange and CK-
488 and negative for CD45-647. (H) RCN CTC
stained positive for DAPI, M30-Orange and NCad-
488 and negative for CD45-647.
Scale bar is 10µm.
DAPI Vim-488 Combined A
DAPI CK18-488 Combined B
DAPI CD45 CK18-488 Combined
DAPI CD45 Vim-488 Combined
DAPI CD45 CK18-488 Combined
DAPI CD45 CK18-488 Combined M30
C
D
E
F
G
H DAPI CD45 NCad-488 Combined M30
153
9.3 Conclusions
Overall, in the RCN and sarcoma studies, we observe reduced CTCs and lung weight after 25 days
of IVLP treatment. The reduction in CTCs does not occur with systemic chemotherapy. Using the
IVLP strategy, high doses of anti-cancer drugs could be administered to the lungs. This localized
treatment also led to significant prevention of lung metastases from sarcoma or colorectal cancer.
Circulating tumor cells commonly undergo EMT during metastasis.236 In both RCN and sarcoma
rat models, we observe a shift in CTC phenotype towards a lower-EpCAM phenotype. This shift
is likely due to the low probability of survival of epithelial CTCs in circulation.135 In sarcoma rat
models, IVLP administration of doxorubicin caused a significant reduction in zone 3 and 4 CTCs.
In colorectal cancer rat models, IVLP administration of oxaliplatin caused the reduction of zone 3
CTCs; however, zone 4 CTCs remained in circulation. The zone 4 CTCs may represent a drug-
resistant phenotype that requires longer treatment periods to eradicate.
IVLP is a promising technique for the treatment of pulmonary metastases. Our study demonstrates
that IVLP could provide an effective therapeutic method, as determined by a reduction in CTCs,
lung weight and nodules.
9.4 Methods
9.4.1 Cell Culture
RCN and sarcoma cells are cultured in RPMI, 10% FBS media at 37°C and 5% CO2. For the rat
sarcoma cell line, methylcholanthrene-induced sarcoma (MCA) cell line was used. This cell line
has subcutaneous origin, and is used as mesenchymal cancer cells for rat experiments. For the rat
colorectal cancer cell line: RCB0511 (RCN-9) was used as an epithelial cancer cell line. Both
cell lines were obtained from Dr. Marcelo Cypel, at the Latner Thoracic Research Laboratories,
Toronto, Canada.
9.4.2 Rat Treatments
Male Fisher344 rats (250-350g) were used in the IVLP study. Rats were injected with rat sarcoma
cells (MCA) or rat RCN-9 cells (2.5 x 106) via the jugular vein on day 0. Rats received
chemotherapy on day 4 after the establishment of micrometastatic disease. IVLP of left lung was
154
performed with Steen solution; at the flow rate of 0.25 ml/ min (15 ml/h) for 60 min. Systemic
chemotherapy was administered through the jugular vein at Day 4.
Rats with sarcoma-induced lung cancer are treated with doxorubicin and ifosfamide. Doxorubicin
is part of a group of chemotherapy drugs known as anthracycline antibiotics. It slows or stops the
growth of cancer cells and induces apoptosis. Ifosfamide is an alkylating agent involved in cross-
linking DNA strands, thus inhibiting cell cycle and replication. The cytotoxic action is primarily
due to cross- linking of strands of DNA and RNA, as well as inhibition of protein synthesis.237
Rats with RCN-induced lung cancer are treated with FOLFOX. FOLFOX treatment has been
widely used as a treatment of colorectal cancer. Cancer chemotherapy is typically administered for
the primary tumor (colorectal cancer cells). FOLFOX is the combination of oxaliplatin, folinc acid
and fluorouracil. In the clinic, oxaliplatin has shown antitumor activity as a single agent in a variety
of solid tumors, and also in combination with leucovorin (folinic acid) and 5FU (Fluorouracil) as
part of the FOLFOX regimen for the treatment of metastatic colon cancer.238 Oxaliplatin is a
platinum containing antineoplastic agent. It is thought to exert its cytotoxic action in a similar
manner to alkylating agents by causing inter- and intrastrand cross links in DNA, inhibiting DNA
synthesis and inducing apoptotic cell death. The addition of oxaliplatin to 5-fluorouracil and folinic
acid significantly improves the overall response rate in patients with previously untreated
colorectal cancer.239
FOLFOX was administered with concentrations of (oxaliplatin 85 mg/m2 + 5-Fluorouracil (300
mg/m2) Oxaliplatin alone was administered at a concentration of 85 mg/m2. Doxorubicin was
administered at concentrations of 30 mg/m2, and Ifosfamide was administered at concentrations of
1.5 g/m2.
9.4.3 In Vivo Lung Perfusion
The IVLP technique was derived from the single-pass isolated lung perfusion previously described
by Wang et al.234, 235 Briefly, anaesthesia was induced with isoflurane (4%) in a mixture of nitrous
oxide (N2O) and oxygen (O2). After 5min, intubation was performed with 14-gauge tube
orotracheally by translaryngeal illumination and was ventilated with a volume-controlled
ventilator (Harvard apparatus, 683 Small Animal Ventilator, St. Laurent, Quebec, Canada). Then
Buprenorphine (Temgesic®, Reckitt Benckriser, Berkshire, United Kingdom) was injected at the
155
dose of 0.05 mg/kg intraperitoneally for analgesia. Isoflurane was adjusted between 1.5-3.0% and
ventilation was performed at a rate of 75 /min and a tidal volume of 8 ml/kg. Then, the left chest
was shaved and prepared with a 70% alcohol solution. Left thoracotomy was performed in the 4th
intercostal space and the hilum was dissected free. The pulmonary artery (PA) and vein (PV) were
clamped with microclips, and 12 gauge angiocatheter was inserted through the chest wall. A PE-
10 catheter (Clay Adams, Boston, MA, USA) was inserted in the PA through the angiocatheter
and secured by 7-0 prolene (Covidien, Saint-Laurent, Québec, Canada). Perfusate was delivered
through the catheter and drained at pulmonary venotomy with a gauze compress. Then, PA and
PV were repaired with 9-0 prolene (Covidien, Saint-Laurent, Québec, Canada) and the vascular
clamps were removed. Then the left lung returned in anatomic position. Through the thoracotomy,
two PE-50 catheters connected to 5ml syringes were inserted in the chest cavity to facilitate re-
expansion. The left thoracotomy was closed in three layers with vicryl 3.0 and the animals were
awaken. When they had recovered and breathed spontaneously, the endotracheal tube and chest
tube were removed. For all experiments, we used Steen solution (XVIVO Perfusion, Göteburg,
Sweden), which was delivered at flow rate of 0.25 ml/min and for 60 min. For safety and survival
study, a 10 min washout was performed after the chemotherapy perfusion.
9.4.4 Intravenous Perfusion
Anaesthesia and intubation were performed as describe above. Briefly, using a 5-mm skin incision
in the right anterior neck, a polyethylene catheter (PE-10; Clay Adams, Boston, MA) was inserted
into the right jugular vein. Then chemotherapy was injected in 6ml of saline for 60 min. The
incision was closed, and the animals were awaken.
9.4.5 Blood Preparation with Nanobeads
All blood samples were analyzed within a few hours from the sample collection time. EpCAM
conjugated nanobeads (MACS) are added to blood (10µl of nanobeads added to 1ml of blood) and
incubated and mixed for 30 minutes at room temperature. Nanobeads attach to EpCAM-
expressing cells. NCadherin- conjugated nanobeads are prepared by incubating NCadherin (0.5
mg/ml) (Abcam) and anti-biotin nanobeads (MACS) with 1ml of blood for 30 minutes at room
temperature. During this incubation time, the magnetic nanobeads were attached to NCadherin-
expressing cells. Microfluidic devices perfused with Pluronic F68 Sigma (Sigma Aldridge) were
prepared, and washed with PBS. The blood was introduced into the microfluidic device and
156
introduced at 300 µl/h or 600 µl/h. Circulating tumor cells are captured in the apex of X-shaped
structures in the presence of a magnetic field.80
9.4.6 Immunostaining
After the blood has been processed through the velocity valley chip, non-specific white blood
cells are washed away using 300µl of PBS-EDTA. Cells are then fixed with 4%
paraformaldehyde, and subsequently permeabilized with 0.2% Triton X-100 (Sigma-Aldrich) in
PBS.
Cells were immunostained with primary antibodies, biotin-Mouse monoclonal Anti-Cytokeratin
18 (Lifespan), Anti-Vimentin (Abcam) and Anti-N Cadherin (Abcam) used separately in different
chips, and Rabbit polyclonal Anti-CD45 (Abcam), followed by secondary antibodies Alexa 647-
Goat Anti-Rabbit (Abcam) to visualize the WBCs and Yellow-nanoB-Avidin (Invitrogen)
(1:2500) to visualize the CTCs. M30 Orange Cytodeath (Peviva) was used to visualize apoptotic
CTCs. All of the primary antibodies were prepared in 100 µl PBS plus 1% BSA plus 0.1% Tween
20 and chips were stained for 60 minutes at a flow rate of 0.1 ml/h. The secondary antibodies are
prepared in 100 µl PBS plus 1% BSA plus 0.1% Tween 20 and chips were stained for 30 minutes
at a flow rate of 0.1 ml/h. Chips were washed between each staining step using 200 µl 0.1% Tween
20 in PBS, at 0.6 ml/h for 10min. Nuclei were stained with 100 µl DAPI ProLong Gold reagent
(Invitrogen, CA) at 0.6 ml/h. After completion of staining, all devices were washed with PBS and
stored at 4 °C before scanning.
9.4.7 Image Scanning and Analysis
Chips were scanned using a 10X objective and a Nikon Ti-E Eclipse microscope with an automated
stage controller and a CMOS Camera (Andor Neo). Images were acquired with NIS software.
DAPI, FITC, TRITC and Cy5 channels were recorded. Target cells were manually counted.
9.4.8 Flow Cytometry
Flow cytometry is performed for various cancer cells to determine the expression profile of
intracellular and cell-surface proteins. Cells were fixed with 4% paraformaldehyde and
permeabilized with 0.2% Triton X, and then incubated with primary antibodies for 30 minutes.
EpCAM-647 (BioLegend), E-Cadherin-647 (BioLegend), CK-APC (LifeSpan), NCad-488 (Bioss)
157
and Vimentin-488 (BD Pharmagin) were analyzed using a BD FACSCanto flow cytometer.
Measurements are plotted as median fluorescent intensities for each marker. Cells were analyzed
with Flow Cytometry with 10,000 cells counted per group. Median fluorescent intensities of the
corresponding fluorophore were recorded. Fluorescent intensities were normalized to the un-
stained control.
9.4.9 Chip Fabrication
Microchips were fabricated using poly(dimethylsiloxane) (PDMS) soft-lithography starting with
an SU-8 master on a silicon wafer (University Wafer, MA). A PDMS (Dow Chemical, MI) replica
of the master was formed. After peeling the replica, holes were pierced for tubing connections.
The replica was permanently sealed with a PDMS-coated glass slide. The PDMS was adhered to
the glass slide using a plasma discharge for 1 minute prior to bonding. Silicone tubing was then
added at the inlet and the outlet. The channel depth was 100 μm. PDMS chips were conditioned
with Pluronic F68 Sigma (Sigma Aldridge) to reduce sample adsorption and washed with PBS
pH=7.4 before use using a syringe pump (Chemyx, TX). Two arrays of NdFeB N52 magnets (KJ
Magnetics, PA),1.5 mm diameter and 8 mm long, were placed on both the bottom and top surfaces
of the capture zones in the chip for the duration of the cell capture process.80
9.4.10 Lung weights and Histology
Lung weight is a surrogate marker of tumor burden in several literature studies.234 Immediately
after euthanasia, the left lung weights were measured. Tissues were fixed in 10% formalin for
48–72 hr and then transferred to 70% ethanol. Samples were paraffin embedded, sectioned and
stained with hematoxylin and eosin. The number of tumor nodules were counted and compared
between groups. We defined nodules greater than 200 micrometer in diameter as a significantly
grown metastases.
9.4.11 Statistical Analysis
Two-sampled t-test were performed on populations. P values < 0.05 were accepted as statistically
significant. Box plots represent standard error of the mean. The mean is shown as the central
square, with the median depicted as a line. Each dot represents an individual rat sample.
158
10 Appendix C – Supporting Information
10.1 Supporting Information for Chapter 3
10µm
Figure 10.1.1 SKBR3 cells grown on FITC Type I collagen matrix SKBR3 cells are plated on the FITC collagen matrix and grown for 24 hours at 37oC in an
incubator. Cells are then fixed and stained with cytokeratin-APC and DAPI, and imaged
using a 10X objective.
FITC Collagen Matrix
SKBR3 cells stained
with cytokeratin- APC
and DAPI
Figure 10.1.2 Collagen uptake assay of SKBR3 and SKBR3- EMT Cells. Collagen type I uptake in SKBR3 and SKBR3-
EMT cells. SKBR3 cells are isolated from the 4 zones of the microfluidic chip, and plated on the fluorescent collagen matrix.
Cells are analyzed with flow cytometry for ingested FITC collagen. SKBR3- EMT cells are treated with CoCl2 for 72
hours. The number of SKBR3 cells analyzed for FITC collagen uptake using flow cytometry is 4000 ± 1257 from zone 1, 4500
± 900 from zone 2, 1200 ± 500 from zone 3 and 300 ± 150 from zone 4. Standard errors of the mean are shown. Median relative
fluorescent intensities are shown relative to the unstained control. Statistics were performed with two sample tailed t-test
(p<0.05).
Rel
ativ
e fl
uo
resc
ence
inte
nsi
ty o
f FI
TC C
olla
gen
1
10
100
1000
Zone 1 Zone 2 Zone 3 Zone 4
Collagen Uptake
SKBR3 cells
SKBR3- EMT cells
*
* * *
159
Zone1
Zone2
Zone3
Zone4
Figure 10.1.3 Folate receptor protein levels of SKBR3 cells SKBR3 cells are isolated from the 4 zones of the microfluidic
chip and grown on the FITC collagen matrix for 24 hours. Flow
cytometry analysis of folate receptor protein levels are shown.
Fluorescence intensity
0
50
100
150
Zone 1Zone 3
0.8
1
1.2
1.4
MCF-7 MDA-MB-231
NA
D(P
)H R
elat
ive
Inte
nsi
ty A C E
MIT
OC
HO
ND
RIA
NA
D(P
)H R
elat
ive
Inte
nsi
ty
CY
TOP
LASM
B D
SKBR3 SKBR3- EMT
NA
D(P
)H In
ten
sity
N
AD
(P)H
Inte
nsi
ty
F
0.8
1
1.2
1.4
1.6
1.8
2
MCF-7 MDA-MB-2310.8
1
1.2
1.4
1.6
1.8
2
SKBR3 SKBR3- EMT
0
10
20
30
40
Zone 1Zone 3
0.8
1
1.2
1.4
SKBR3 SKBR3- EMT
*
2.8 mg/L folate
NA
D(P
)H R
elat
ive
Inte
nsi
ty
0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
Zone 1 Zone 2 Zone 3
0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
Zone 1 Zone 2 Zone 3
Mitochondria Cytoplasm
1.7 mg/L folate
G H
* * *
* 2.8 mg/L folate
1.7 mg/L folate
NA
D(P
)H R
elat
ive
Inte
nsi
ty
160
Figure 10.1.4 NAD(P)H metabolic response of breast cancer cells. To assess the NAD(P)H response, all cells are treated
with 1.7 mg/L folate. (A and B) NAD(P)H relative intensities of MCF-7 cells and MDA-MB-231 cells in the mitochondria
and cytoplasm, respectively. (C and D) NAD(P)H relative intensities of SKBR3 and SKBR3- EMT cells in the mitochondria
and cytoplasm, respectively. (E and F) NAD(P)H intensities of zone populations of SKBR3 cells and SKBR3- EMT cells
in the mitochondria and cytoplasm, respectively. SKBR3- EMT cells are treated with CoCl2 for 24 hours. (G and H)
NAD(P)H relative intensities of zone populations of SKBR3 cells in the mitochondria and cytoplasm, respectively. Zone 4
was not included as the cell density was too low for image analysis. Cells are serum-starved for 30 minutes in folate- free
media before incubation with folate. NAD(P)H is reported relative to the baseline autofluorescence or as the absolute
intensity. Standard errors of the mean are shown. Statistics are performed with two-tailed t-test (p<0.05) for paired data
points or one-way ANOVA followed by the Tukey multiple comparisons (p<0.05) for multiple data points.
Figure 10.1.5 Collagen uptake in metastatic prostate cancer CTCs. (A) Representative images of prostate cancer CTCs which have ingested
collagen. Cells are stained with DAPI, FITC- Collagen, Cytokeratin and
CD45. (B) Collagen uptake in CTCs isolated from prostate cancer patients
determined using immunocytochemistry. Mean fluorescent intensities of
FITC collagen are shown relative to the background intensity. Markers
denote individual CTCs from four prostate cancer patients (P1- P4). The
collagen assay is performed with healthy donors (n=2), and 0 CTCs are
reported.
CTCs are captured with EpCAM-aptamers conjugated to magnetic beads
in the microfluidic device, released from the beads using antisense DNA,
and cultured on the collagen matrix for 24 hours. CTCs are then released
from the matrix, immunostained and analyzed for fluorescent collagen
uptake. CTCs which ingested collagen are identified as
DAPI+/FITC+/Cytokeratin+/CD45-. All CTCs stain negative for CD45
Alexa Fluor 555 (image not included in panel).
10µm
Prostate
cancer CTCs
(individual
CTCs and
clusters)
DAPI FITC- Collagen Cytokeratin Combined
Prostate Cancer Patients
A
B
161
Figure 10.1.6 Surface marker expression analysis of SKBR3 cells after isolation from the microfluidic device.
(a) Folate receptor α expression levels in control (n=6000 cells) and cells isolated from the microfluidic device
(n=3000 cells). Cells are captured in the microfluidic device with anti-EpCAM MNPs (b) EpCAM expression levels
in control (n=9000 cells) and cells isolated from the microfluidic device (n=1750 cells). Cells are captured in the
microfluidic device using EpCAM aptamers conjugated to MNPs. Prior to flow cytometry EpCAM analysis, cells are
released from the beads using antisense DNA. Control cells represent cells not introduced into the microfluidic device.
Control
Cells isolated from the microfluidic device
Folate receptor α
Fluorescence intensity
EpCAM
Fluorescence intensity
A B
162
10.2 Supporting information for Chapter 4
Figure 10.2.1 Scanning electron microscope images of walls in the square array configuration. Image is depicting
the micro-structures from top view.
163
Table 10.2.1 3D micro-structure pore width and heights for various cross sections and aspect ratios.
16µm2
a.r. 0.1 0.3 1 3
Width (µm) 1.3 2.2 4.0 6.9
Height (µm) 12.6 7.3 4.0 2.3
27µm2
a.r. 0.1 0.3 1 3
Width (µm) 1.6 2.8 5.2 9.0
Height (µm) 16.4 9.5 5.2 3.0
36µm2
a.r. 0.1 0.3 1 3
Width (µm) 1.9 3.3 6.0 10.4
Height (µm) 19.0 11.0 6.0 3.5
49µm2
a.r. 0.1 0.3 1 3
Width (µm) 2.2 3.8 7.0 12.1
Height (µm) 22.1 12.8 7.0 4.0
164
Figure 10.2.2 Scanning electron microscope images of MCF10CA1a.cl1 cells interacting with basal pores. Pore
dimensions are as follows: (A and D) cross section 27 µm2 and aspect ratio 0.1, (B and E) cross section 36 µm2 and
aspect ratio 1 and (C) cross section 49 µm2 and aspect ratio 1. Cell membranes are artificially colored in green in
panels A-C.
Figure 10.2.3. Characterization of MCF10A and MCF10CA1a.cl1 cells. Boxplots reporting the (A) nucleus
diameter, (B) division time, (C) velocity of cells on flat substrate and (D) number of migrating cells through a matrigel
invasion assay. n: total number of analyzed cells. n’: total number of independent experiments. Error bars represent
standard deviations. * p<0.05
165
Figure 10.2.4 Topographic contact guidance of MCF10A and MCF10CA1a.cl1 cells. (A) Migration of MCF10A
and MCF10CA1a.cl1 cells on gratings recorded for a period of 20 h. Arrow represents direction of guidance (B)
Persistence of directional migration on gratins corresponding to the travelled distance before the direction is reversed.
(C) The migration angle represents the angle of alignment between migration vector and the direction of the gratings.
Error bars represent standard deviation. * p<0.05.
Figure 10.2.5 Immunofluorescence and flow cytometry quantification of HRas in MCF10A and
MCF10CA1a.cl1 cells. (A) Fluorescent images of cells transfected with H2B-eGFP and immunostained with HRas-
Alexa Fluor 647 antibody. (B) Flow cytometry analysis of total HRas. 10,000 cells were analyzed per cell line.
Experiments were performed in triplicate.
166
Figure 10.2.6 Flow cytometry analysis of migration markers in MCF10A and MCF10CA1a.cl1 cells. (A-C)
Flow cytometry analysis of Vimentin, Talin1 and neural cell adhesion molecule (NCAM), respectively. 10,000 cells
were analyzed per cell line. Experiments were performed in triplicate.
167
168
Figure 10.2.7 Effect of pore shape and orientation on cell penetration dynamics. Pores with cross sections of 16,
27, 36 and 49 µm2 and aspect ratios 0.1, 0.3, 1 and 3 are examined. Pore penetration and disengagement of MCF10A
cells (A, C, E, G) and MCF10CA1a.cl1 cells (B, D, F, H), as a function of pore aspect ratio for a cross section of 16,
27, 36 and 49 µm2 expressed in terms of cell percentage. Error bars represent standard error of the mean. * p<0.05.
Figure 10.2.8 Engagement of events of A) MCF10A and B) MCF10CA1a.cl1 cells for various cell densities along
the pore walls. Percent of engagement events (penetrate, disengage or impasse) is shown for low and high cell
densities. The percentage is calculated as the proportion of engagement event relative to the total number of
engagements at a given cell density. The number of engagement events are recorded for a.r. 0.1, 0.3, 1 and 3 and cross-
sections 16, 27, 36 and 49µm2 over 24 hours along the pore wall. Low density represents 330- 1100 cells/mm2 whereas
high density represents 1300- 1800 cells/mm2. MCF10A n=253 cells, n’=3. MCF10CA1a.cal1 n=306 cells, n’=3.
169
Figure 10.2.9 Polarization of cells during penetration and disengagement of pores with cross section of 36 µm2
and aspect ratio of 0.1. (A-D). Position of the Golgi relative to the nucleus for MCF10A and MCF10CA1a.cl1 cells
during engagement events.
Figure 10.2.10 Polarization of cells during penetration and disengagement for cross section 36 µm2 and aspect
ratio 0.3. (A-D) Position of the Golgi relative to the nucleus for MCF10A and MCF10CA1a.cl1 cells during
engagement events
170
Figure 10.2.11 Polarization of cells during penetration and disengagement for cross section 36 µm2 and aspect
ratio 1. (A-D) Position of the Golgi relative to the nucleus for MCF10A and MCF10CA1a.cl1 cells during engagement
events.
Figure 10.2.12 Cell polarization of A) MCF10A and B) MCF10CA1a.cl1 cells for various cell densities in the
absence of directional signals. The polarization of the cells was determined by tracking the position of the Golgi
relative to the nucleus over the time lapse images, as described in the Experimental Section. Low density represents
330- 1100 cells/mm2 whereas high density represents 1300- 1800 cells/mm2. Error bars represent the standard error
of the mean.
171
Figure 10.2.13 Flow cytometry analysis of Rac1 and RhoA levels in MCF10A and MCF10CA1a.cl1 cells. (A-D)
Flow cytometry analysis of Active Rac1, Total Rac1, Active RhoA and Total RhoA, respectively. 10,000 cells are
analyzed per cell line. Experiments were performed in triplicate.
172
Figure 10.2.14 Correlation function and length for the collective migration of MCF10A and MCF10CA1a.cl1
cells. (A) Angular velocity of MCF10A and MCF10CA1a.cll cells calculated with CIV189 analysis. The color map
indicates the direction of migration. Areas with homogenous colors indicate higher migration coherence. (B)
Correlation function and (C) correlation length of MCF10A and MCF10CA1a.cll cells. These experiments and data
analysis were performed by F.M. Pramotton.
Figure 10.2.15 Representative immunofluorescence confocal sections along the apical, equatorial, and basal
surfaces of MCF10CA1a.cll cells stained for nucleus (green) and actin (red) on substrate without (A) and with
(B) constrictions. These experiments and data analysis were performed by M. Panagiotakopoulou.
173
10.3 Supporting information for Chapter 5
Figure 10.3.1 Metastatic castrate resistant prostate cancer patient profiles. A) Number of patients which are
currently ongoing, completed, deceased and withdrawn. Patients receiving enzalutamide or abiraterone are shown.
B) Number of progressive and responsive patients for patients receiving enzalutamide or abiraterone.
11
15
1
3
7
00
5
10
15
20
Ongoing Completed (switchtreatment)
Deceased
Enzalutamide
Abiraterone
Num
be
r o
f pa
tien
ts
A
B
Num
be
r o
f pa
tien
ts
15
11
65
0
5
10
15
20
25
Progressive Responsive
Enzalutamide
Abiraterone
174
Figure 10.3.2 Number of metastases for progressive and responsive patients receiving enzalutamide or
abiraterone. The number of metastatic sites are recorded on average 1.5 ± 0.4 years prior to baseline using bone
scan, biopsies or CT scan. Metastatic sites represent bone or lymph nodes. Box plots represent standard error of the
mean. The mean is show as the central square, with the median depicted as a line. Each dot represents a patient.
175
Figure 10.3.3 PSA waterfall plots for progressive and responsive patients receiving enzalutamide or
abiraterone. A) Progressive patients. n=21 B) Responsive patients. n=16.
-100
0
100
200
45-148 weeks
-100
0
100
200
23-44 weeks
-100
0
100
200
9-22 weeks
-100
0
100
200
23-44 weeks
-100
0
100
200
45-148 weeks
-100
0
100
200
9-22 weeks
Progressive Responsive A B
760
Pe
rcen
t ch
an
ge
fro
m b
ase
line
(%
)
176
Table 10.3.1 Prior drug treatment for mCRPC patients. Percentages indicate the proportion of treatments given
relative to the total number of treatments recorded per patient.
LHRH agonists (46%) Anti-Androgens (46%) Steroids (5%) Immune Therapy (2%)
Triptorelin (Trelstar) Bicalutamide (Casodex) Prednisone Prostvac
Leuprolide (Eligard) Nilutamide
Goserelin (Zoladex) ARN509 V/s Placebo
Degarelix (Firmagon)
Figure 10.3.4 Healthy donor false positive cells captured in the velocity valley device. A) Target cells are captured
with EpCAM-MNPs and identified as DAPI+/CK+/CD45- or DAPI+/NCad+/CD45-. The false positive counts for
EpCAM capture with the velocity valley device is approx. 2 cells/ml. B) Cells are captured with NCadherin- MNPs
and identified as DAPI+/CK+/CD45-. The false positive counts for NCadherin capture with the velocity valley device
is approximately 2 cells/ml.
Cells
/ml
Figure 10.3.5. NCadherin capture efficiency. LnCAP and
PC3 cells are captured with NCad- MNPs in the velocity
valley device. 100 cells are loaded into the device.
Cells
/ml
177
Figure 10.3.6. CellSearch counts A)- D) CellSearch cytokeratin CTC counts per 7.5ml of blood for progressive
versus responsive patients receiving enzalutamide or abiraterone. The red line depicts the clinically relevant 5
CTCs/7.5 ml cutoff. Below 5 CTCs/7.5ml is considered favorable whereas ≥ 5 CTCs/7.5ml is unfavorable. CTCs are
captured with EpCAM and identified as DAPI+/CK+/CD45-.
178
10.4 Referred Journal Publications
1. B. J. Greenǂ, M. Panagiotakopoulouǂ, F.M. Pramotton, G. Stefopoulos, S.O. Kelley, D.
Poulikakos*, A. Ferrari*. “Pore shape and orientation define paths of metastatic migration”,
Nanoletters, 2018 Mar 14;18(3):2140-2147.
2. B. J. Greenǂ, L. Kermanshahǂ, M. Labib, S. U. Ahmed, P. N. Silva, L. Mahmoudian, I. H. Chang, R. M. Mohamadi, J. V. Rocheleau, and S. O. Kelley*, "Isolation of phenotypically distinct cancer cells using nanoparticle-mediated sorting," ACS Appl Mater Interfaces, vol. 9, pp. 20435-20443, Jun 21 2017.
3. M. Poudineh, P. M. Aldridge, S. Ahmed, B. J. Green, L. Kermanshah, V. Nguyen, C. Tu, R. M. Mohamadi, R. K. Nam, A. Hansen, S. S. Sridhar, A. Finelli, N. E. Fleshner, A. M. Joshua, E. H. Sargent*, and S. O. Kelley*, "Tracking the dynamics of circulating tumour cell phenotypes using nanoparticle-mediated magnetic ranking," Nat Nanotechnol, vol. 12, pp. 274-281, Mar 2017.
4. M. Labib, B. Green, R. M. Mohamadi, A. Mepham, S. U. Ahmed, L. Mahmoudian, I. H. Chang, E. H. Sargent*, and S. O. Kelley*, "Aptamer and antisense-mediated two-dimensional isolation of specific cancer cell subpopulations," J Am Chem Soc, vol. 138, pp. 2476-9, Mar 2 2016.
5. B. J. Green, T. Saberi Safaei, A. Mepham, M. Labib, R. M. Mohamadi, and S. O. Kelley*, "Beyond the capture of circulating tumor cells: Next-generation devices and materials," Angew Chem Int Ed Engl, vol. 55, pp. 1252-65, Jan 22 2016.
6. R. M. Mohamadi, J. D. Besant, A. Mepham, B. Green, L. Mahmoudian, T. Gibbs, I. Ivanov, A. Malvea, J. Stojcic, A. L. Allan, L. E. Lowes, E. H. Sargent, R. K. Nam, and S. O. Kelley*, "Nanoparticle-mediated binning and profiling of heterogeneous circulating tumor cell subpopulations," Angew Chem Int Ed Engl, vol. 54, pp. 139-43, Jan 2 2015.
7. P. N. Silvaǂ, B. J. Greenǂ, S. M. Altamentova, and J. V. Rocheleau*, "A microfluidic device designed to induce media flow throughout pancreatic islets while limiting shear-induced damage," Lab Chip, vol. 13, pp. 4374-84, Nov 21 2013.
8. M. Y. Sun, E. Yoo, B. J. Green, S. M. Altamentova, D. M. Kilkenny, and J. V. Rocheleau*, "Autofluorescence imaging of living pancreatic islets reveals fibroblast growth factor-21 (FGF21)-induced metabolism," Biophys J, vol. 103, pp. 2379-88, Dec 5 2012.
9. K. S. Sankarǂ, B. J. Greenǂ, A. R. Crocker, J. E. Verity, S. M. Altamentova, and J. V. Rocheleau*, "Culturing pancreatic islets in microfluidic flow enhances morphology of the associated endothelial cells," PLoS One, vol. 6, p. e24904, 2011.
10. W. Hallett*, B. Green, T. Machula and Y. Yang. "Packed bed combustion of non-uniformly sized char particles". 2013. Chemical Engineering Science. vol. 96, pp.1-9, June 2013.
11. P. Renton, B. Green, S. Maddaford, S. Rakhit, and J. S. Andrews*, "NOpiates: Novel dual action neuronal nitric oxide synthase inhibitors with mu-opioid agonist activity," ACS Med Chem Lett, vol. 3, pp. 227-31, Mar 8 2012.
12. S. J. Copeland, B. J. Green, S. Burchat, G. A. Papalia, D. Banner, and J. W. Copeland*, "The diaphanous inhibitory domain/diaphanous autoregulatory domain interaction is able to mediate heterodimerization between mDia1 and mDia2," J Biol Chem, vol. 282, pp. 30120-30, Oct 12 2007.
ǂ equal contribution *corresponding author
![Metastatic Heterogeneity of Cells from Lewis Lung …...(CANCER RESEARCH 43, 5314-5320, November 1983] Metastatic Heterogeneity of Cells from Lewis Lung Carcinoma1 Anne-Louise van](https://img.pdfslide.us/doc/110x75/5f901b8b1c8ad53dfd4f12f7/metastatic-heterogeneity-of-cells-from-lewis-lung-cancer-research-43-5314-5320.jpg)
