ISSN 1473-0197

Lab on a ChipMiniaturisation for chemistry, physics, biology, materials science and bioengineering

www.rsc.org/loc Volume 12 | Number 18 | 21 September 2012 | Pages 3199–3522

www.rsc.org/crystal_dataRegistered Charity Number 207890

New process for crystal data files

Do you submit crystallographic data with your articles?

Confused by what you need to do?

Then you’ll be glad to hear that we are making it easier for you:

from September 1st, all crystal data files (CIFs) must be accompanied by the CCDC

reference number(s) when you submit your article.

As this change also aligns us with preferred CCDC practices, it means a more

streamlined process for everyone involved.

Full details are available on our website.

Image courtesy of Professor Gang Chen and Dr Rencheng Jin DOI: 10.1039/C2CE06417K

PAPERChang-Soo Lee et al.Profi ling surface glycans on live cells and tissues using quantum dot-lectin nanoconjugates

RSC cover012018_LITHO.indd 2RSC cover012018_LITHO.indd 2 8/10/2012 7:17:19 AM8/10/2012 7:17:19 AM

Publ

ishe

d on

30

May

201

2. D

ownl

oade

d by

Uni

vers

ity o

f V

irgi

nia

on 1

0/11

/201

3 21

:23:

19.

View Article Online / Journal Homepage / Table of Contents for this issue

Profiling surface glycans on live cells and tissues using quantum dot-lectinnanoconjugates{

Heon-Ho Jeong,{a Yun-Gon Kim,{b Sung-Chan Jang,a Hyunmin Yic and Chang-Soo Lee*a

Received 12th March 2012, Accepted 24th May 2012

DOI: 10.1039/c2lc40248c

The surface of mammalian cells is densely coated with complex glycans, which are directly involved in

cell–cell or cell–protein interactions that trigger various biological responses. Here, we present a novel

glycomics approach that uses quantum dot (Qdot)-lectin nanoconjugates to interrogate the surface

glycans of tissues and patterned cells. Our approach allows highly sensitive in situ monitoring of

specific lectin–glycan interactions and quantitative information on surface glycans for each examined

cell line and tissue. The results clearly show significant changes in glycosylation for each cell line and

tissue sample. We expect that these results will be applicable in cancer diagnostics and promote the

development of new analytical tools for glycomics.

Introduction

Glycosylation plays an important role in physiological and

biological events associated with intra- and extracellular signal-

ing and interactions in cancer metastasis.1–4 A striking example is

the abnormal changes in glycosylation (over-expressed or highly

branched glycans) observed in cancerous cells that have a direct

influence on cancer cell activities, including proliferation,

adhesion, differentiation, invasion, angiogenesis, and immu-

nity.5–9 Therefore, the alteration of cancer cell surface glycans

could be a promising biomarker for early diagnosis, prognosis,

and targeted treatment in cancer.

There have been tremendous efforts to develop comprehensive

profiling methods for the cellular glycomes via mass spectro-

metry (MS), high-performance liquid chromatography (HPLC)

and frontal affinity chromatography (FAC).10–12 Although the

analytical technologies provide detailed quantitative and func-

tional information about the glycans, critical challenges in the

separation of cell surface glycans remain, such as time-

consuming and cumbersome procedures as well as the need for

a large number of cells.

As an alternative approach, glycan (N-/O-glycans and glyco-

sphingolipids) microarrays have been proposed for the structural

characterization of cell surface glycans.13 The microarray

strategies have successfully provided their functional informa-

tion using specific antibodies and glycan binding proteins

(GBPs).14–17 However, purifying and chemically modifying the

cell surface glycans to immobilize them onto the microarray

surfaces remains a formidable challenge. Moreover, synthetic

N-/O-glycans and neo-glycolipids are often insufficient to cover

all of the natural glycans expressed on cell surfaces.

More recent efforts have been directed towards the use of

lectin microarrays that enable direct analysis of the mammalian

cell surface glycans.18,19 Notably, the lectin-based approaches

are advantageous in discriminating between sugar isomers,

which is otherwise difficult via MS and high-throughput

analyses. In the early stages of development, the lectin

microarrays were used for rapid and sensitive analysis of

complex glycan structures from cell surface glycoproteins with-

out glycan liberation.20 Hirabayashi et al. successfully demon-

strated an analysis of intact cell surface glycans using lectin

microarrays by applying the fluorescence-labeled live cells

directly to the lectin microarrays and analyzing the cell surface

using an evanescent-field fluorescence scanner.21 However, there

are still inherent problems with this method, including the need

for cell staining, limited accessibility of cells to the lectin,

nonuniformity of the lectin binding sites on the patterns, and

challenges in handling tissue specimens obtained from an

excisional biopsy. Thus, immobilized cell microarray formats

are more promising due to the absence of cell staining, the

accessibility of the binding sites to lectins, and ability to extend

this method to tissue sample formats.

Here, we present a novel surface glycan profiling method of

live cells (cancerous vs. normal cells) using nanoparticle–lectin

nanoconjugates and a simple cell patterning technology, as

shown in Fig. 1a. The target cells are specifically patterned on

the culture plates containing cell-attracting or cell-repelling

regions, fabricated by a simple soft-lithography.22–25 To inves-

tigate glycan-related biomarker epitopes on mammalian cell

surfaces, we have employed Qdot–lectin nanoconjugates for

facile imaging with a fluorescence microscope.26–28 First, we

aDepartment of Chemical Engineering, Chungnam National University,Yuseong-gu, Deajeon, 305-764, South Korea. E-mail: rhadum @cnu.ac.krbDepartment of Chemical Engineering, Soongsil University, 369 Sangdo-Ro, Dongjak-gu, Seoul, 156-743, South KoreacDepartment of Chemical Engineering, Tufts University, Medford,Messachusetts 02155, USA{ Electronic Supplementary Information (ESI) available. See DOI:10.1039/c2lc40248c{ These authors have equally contributed and should be considered firstauthor.

Lab on a Chip Dynamic Article Links

Cite this: Lab Chip, 2012, 12, 3290–3295

www.rsc.org/loc PAPER

3290 | Lab Chip, 2012, 12, 3290–3295 This journal is � The Royal Society of Chemistry 2012

Publ

ishe

d on

30

May

201

2. D

ownl

oade

d by

Uni

vers

ity o

f V

irgi

nia

on 1

0/11

/201

3 21

:23:

19.

View Article Online

demonstrate facile cell patterning that leads to readily addres-

sable, quantitative and high fluorescent signals when compared

to 96-well plate assays. Next, we profile cell surface glycans of

various cancer cell lines and directly compare the glycan

expression level to that of normal cells. Finally, we demonstrate

that our novel method can be used to quantitatively examine

exicisional biopsy tissue samples. We believe that our approach

provides a robust platform to systematically investigate the

complex ‘‘glycan codes’’ for biological events and to integrate

glycomic information toward a deep understanding of how the

codes are controlled by cells.

Materials and methods

Fabrication of PDMS structure

Poly(dimethyl siloxane) (PDMS) microstamps were fabricated

with polydimethylsiloxane (Sylgard 184, Dow-Corning) through

replica micro-molding. Briefly, silicon masters were fabricated

with a SU-8 photoresist by conventional photolithography. A

mixture of PDMS prepolymer and curing agent in 10 : 1 (w/w)

ratio was thoroughly stirred, degassed in a vacuum oven, poured

onto the silicon master, and cured at 65 uC oven overnight. The

cured PDMS was then peeled away from the silicon master. To

make thin layers of PDMS, the 10 : 3 (w/w) ratio of PDMS

prepolymer and curing agent was dissolved in chloroform

(80 mg ml21). The PDMS mixture was poured onto a glass

substrate and was spin-coated at 2000 rpm for 20 s. The

chloroform was easily evaporated during spin coating. The

PDMS stamps were then put on the spin-coated PDMS thin

layer by microcontact printing (mCP). The PDMS stamp with

PDMS mixture was immediately transferred to the culture plates

and then underwent pre-curing at 65 uC for 15 min. After peeling

off the PDMS stamp, the PDMS patterns on culture plates were

post-cured at 65 uC for 4 h.

Preparation of cell patterning and tissue specimens

MCF-7 cells were cultivated in mammary epithelial basal

medium (MEBM) containing bovine pituitary factor (BPE),

hydrocortisone, human epidermal growth factor (hEGF),

insulin, and gentamicin/amphotericin-B, and other cell lines

(HCT116, HDF, ACHN, HEK293, HepG2, HeLa, and A549)

were cultivated in Dulbecco’s modified eagle medium (DMEM)

including the 10% fetal bovine serum and 1% penicillin-

streptomycin. First, the cell suspension (50 ml, 4.0 6 104 cells)

was loaded onto the PDMS pattern (1 cm 6 1 cm) and

incubated at 37 uC in a 5% CO2 environment. The unbound cells

were washed with fresh medium three times and then re-

cultivated for 12 h.

For the application of tissue samples (the detailed information

of breast cancer and colon cancer tissues is listed in ESI, Table

S5 and S6{) commercially available tissue arrays were obtained

from SuperBiochips (Seoul, Korea). The procedures of tissue

array were performed according to the recommended protocols

from the manufacturer. The tissue array were collected and

arranged by pathologists. The tissues were fixed in 4% buffered

neutral formalin for 24 h, sequentially dehydrated by gradient

ethanol (70%, 90%, 95%, 99%, 100%) three times for 1 h, cleaned

three times by xylene for 1 h and embedded in paraffin. Cores of

paraffin containing the tissue are transferred to the recipient

array and were then microtomed to a thickness of 4 mm. The

tissue samples are immobilized on silane coating slide.

For deparaffinization and hydration of the tissue specimen,

formalin-fixed paraffin-embedded tissue arrays were incubated

at 65 uC in an oven for 1 h and then immersed in xylene solution

for 4 min 5 times. The tissue arrays were sequentially hydrated in

100%, 95%, and 75% ethanol for 3 min with 2 times each.

Finally, the tissue arrays were immersed in water for 5 min. To

minimize the background signal from nonspecific binding of

streptavidin–Qdots, a streptavidin/biotin blocking solution

(Vector Laboratories Inc, CA, USA) was applied for 15 min

and were then washed with PBS buffer. All experiments with live

cells, animals, and human tissue are performed in compliance

with guidelines approved by the animal ethics committee

regulation of Chungnam National University.

Cell and tissue surface glycan assays

First, biotinylated lectins (Vector Laboratories Inc, CA, USA;

50 nM to the cell patterns and 500 nM to the tissue array) were

added onto patterned live cells and tissue arrays for 3 h at 37 uC.

Fig. 1 A glycan profiling approach using a combination of cell

patterning and Qdot–lectin nanoconjugates. (a) Schematic diagram of

glycan profiling. (b) Optical images of cell patterning onto the

regioselectively PDMS-patterned TCPS. In this experiment, the PDMS

microstamp, having 100 mm 6 100 mm square patterns, 50 mm spacing,

and 50 mm height, is used. Initial cell attachment (first row) and cell

attachment following 12 h incubation (second row). Scale bars indicate

100 mm.

This journal is � The Royal Society of Chemistry 2012 Lab Chip, 2012, 12, 3290–3295 | 3291

Publ

ishe

d on

30

May

201

2. D

ownl

oade

d by

Uni

vers

ity o

f V

irgi

nia

on 1

0/11

/201

3 21

:23:

19.

View Article Online

The cell patterns were rigorously washed three times each with

fresh medium to remove unbound lectins. Second, streptavidin

conjugated Qdots (Invitrogen, CA, USA) were added and

reacted for 1 h at 37 uC in a 5% CO2 incubator. The cell

patterns were also washed three times with fresh medium and

dried. As a negative control, streptavidin-conjugated quantum

dots (SA–Qdots) have also been investigated to define non-

specific binding. The signal intensity of non-specific binding is

subtracted from total binding signal of Qdot–lectin conjugates

(Fig. S1{).

The cell patterns were monitored by a fluorescence microscope

(TE-2000, Nikon, Japan) equipped with a high resolution CCD

camera (Coolsnap, Roper Science, USA). The software

ImagePro (Mediacybernetics, MD, USA) was used for the

fluorescence analysis of each obtained image.

Inhibition assay of WGA lectin

Inhibition study of WGA was performed by addition of

N-acetylglucosamine (GlcNAc) (0–500 mM) on the MCF-7

cells. First, biotinylated WGA (50 nM) and GlcNAc were

dissolved in medium and incubated for 1 h at 37 uC. This

solution was added to the patterned cells and then remained for

3 h at 37 uC in a 5% CO2 environment. Finally, the cell patterns

were washed three times with fresh medium and then reacted

with SA–Qdots for 1 h at 37 uC in 5% CO2 environment. The

percentage of inhibition (%) can be calculated as follows:

inhibition 0=0ð Þ~no inhibition F:I:½ �{ inhibition F:I:½ �

no inhibition F:I:½ � |100

� F:I: : fluorescence intensity

(1)

96-well plate analysis

For the cultivation of MCF-7 and HDF cells in 96-well plate, we

loaded 20 000 of the cells in tissue culture treated 96-well plates

(BD bioscience, USA) and incubated at 37 uC in 5% CO2

environment. It is optimized condition for fully coverage of well

surface (0.32 cm2) within 12 h culture time. The MCF-7 cells

were sequentially reacted with biotinylated WGA (50 nM) for

3 h and SA–Qdots for 1 h. The plates were read on the HTS

Multi-Label Reader (Perkin Elmer LAS model, MA, USA).

Results and discussion

Fig. 1 shows the overall schematic diagram of the fabrication of

cell patterns with two distinctive regions: the PDMS region,

prepared by microcontact printing, acts to repel cell adhe-

sion29,30 and the exposed cell culture regions are expected to

immobilize cells. As shown in Fig. 1b, we first demonstrate

consistent attachment (first rows) and proliferation (second

rows) of various cell lines on patterned surfaces. After 12 h of cell

seeding, cell proliferation is strictly limited to the cell adhesive

area (100 mm 6 100 mm) regardless of the cell type. The bright

field images of the cell patterns clearly show that the

hydrophobic PDMS layer selectively prevents cell attachment,

enabling a highly reproducible and quantitative analysis of living

cell surface glycans (Fig. 1b). Furthermore, the number of cells

per patterned area is well maintained (0 h incubation: 14 ¡ 3.7

cells/pattern; 12 h incubation: 25 ¡ 4.6 cells/pattern) over the

12 h incubation period with all six examined cell types. This cell

patterning provides 140 patterns per stamp with few mispat-

terned or inactive cell regions (less than 5%).

The success of this proposed method hinges upon the

development of a facile assay for lectin–cell surface glycan

interactions. We first investigated the difference in fluorescent

signals from wheat germ agglutinin (WGA), a lectin displaying

affinity with the terminal N-acetyl-glucosamine (GlcNAc) and

sialic acid epitopes, against either MCF-7 (breast cancer) and

human dermis normal fibroblasts (HDF) cells (Fig. 2a). As

expected, the fluorescence signal following incubation with

MCF-7 cells is significantly higher than that of the HDF cells

due to the breast cancer cell lines exhibiting a high expression of

terminal GlcNAc and sialic acid residues.31,32

To prove the specificity of the lectin binding to cell surface

glycans, we have also performed the inhibition assay with

GlcNAc monomers (Fig. 2b). GlcNAc monomers, which act as

WGA lectin inhibitors, are pre-incubated with the WGA lectin

prior to their addition to the patterned MCF-7 cells. As

expected, the fluorescence intensity gradually decreased with

increasing GlcNAc concentration. This result clearly confirms

that the observed fluorescence results from lectins specifically

binding to the target glycan epitopes on the cell surfaces

(Fig. 2b). The observed dose response data (IC50 value,

0.92 mM) are also comparable to the values reported in the

literature.33,34 Thus, we have confirmed that this is a feasible

method to analyze cell surface glycans. In addition, this

approach highlights the relevance of cell patterning technology

combined with the use of Qdot–lectin nanoconjugate systems for

identifying abnormal glycomics and profiling glycans on the

surface of cancer cells.35–37

Despite their routine use for screening, conventional 96- and

384-well plate formats with 2D cell monolayers suffer from

inefficient liquid handling, such as the removal of reagents from

the wells and the difficulty of subsequently washing the cell

monolayers. Other disadvantages include the need for relatively

large volumes of reagents and the incompatibility with image

analysis. These issues become more complicated when the

screening of cellular targets is coupled with the addition of

multiple reagents.38 In addition, Bovin et al. found that the

sensitivity of microarray methods was lower than that of

enzyme-linked immunosorbent assay (ELISA) methods for

detecting carbohydrate-protein interactions.39 These results lead

us to compare our glycan analysis approach with a routine 96-

well plate-based assay (Fig. 2c, ESI Fig. S2{).

The cells were cultured in 96-well plates for 12 h and

sequentially incubated with biotin–WGA lectin and SA–Qdot

(Fig. 2c). As expected, the signal intensity of a cancer cell (MCF-

7) for WGA is significantly higher than that of a normal cell

(HDF) in both methods. The signal intensity of MCF-7 for

WGA (50 nM) is approximately 3550, which is 17-fold higher

than the normal cell signal, further confirming the highly reliable

sensitivity of our approach. While the difference between normal

cells and cancer cells is not clear and remains statistically

unreliable (p . 0.6) for the 96-well plate method, data obtained

from our cell patterns show statistically significant values

(**p , 0.001). This result indicates that our simple cell

3292 | Lab Chip, 2012, 12, 3290–3295 This journal is � The Royal Society of Chemistry 2012

Publ

ishe

d on

30

May

201

2. D

ownl

oade

d by

Uni

vers

ity o

f V

irgi

nia

on 1

0/11

/201

3 21

:23:

19.

View Article Online

patterning method is reliable and quantitative. Additionally,

given that the working concentration of WGA lectin is 2 orders

of magnitude lower in our approach than in the 96-well plate

method, the overall sensitivity is 4 orders of magnitude higher

than the routine 96-well plate format (ESI Fig. S1{).

Furthermore, our method requires a smaller quantity of samples,

reagents and lectins when compared with the 96-well plate

method. These results indicate that the well-defined cell

patterning method readily distinguishes the difference in surface

glycan profiles between control and cancer cells, providing more

reliable and quantitative data than the 96-well plate method.

Meanwhile, our cell patterning method also provides less

variation in the number of cells per pattern than the 96-well

plate method (data not shown).

To evaluate the ability of this system to differentiate cells by

profiling the glycan content, we compare human dermis

fibroblasts (HDF) and breast cancer cells (MCF-7) using 6

different lectins (Fig. 3a). We have consecutively evaluated

120 patterns for each cell line in 3 replicate experiments to

compensate for intra- and inter-cell line variations. Because

WGA has an affinity for sialic acid and GlcNAc tightly binds

MCF-7, the difference in the obtained fluorescent signal (2-fold)

is significant (p , 0.05) and sufficient to discriminate normal

cells from cancer cells. Moreover, soybean agglutinin (SBA)

lectin shows strong binding only to MCF-7 (p , 0.01, 16-fold

difference with HDF), which confirms that the terminal

N-acetylgalactosamines (GalNAc) or galactoses (Gal) are highly

expressed on the breast cancer cells. These results are consistent

with previous reports.40,41 Although Lotus-tetragonolobus

(LTL, affinity for fucose, Sialyl Lewisx (SLeX) and Lewisx

(LeX)) lectin show relatively lower fluorescence than WGA and

SBA, the quantitative statistical analysis from at least 20

replicates reveals a significant difference (18-fold) between breast

cancer cells and normal cells. This result suggests that the

expression of fucose-containing glycans in MCF-7 is higher than

that of HDF, and also correlates well with previous results on

the high expression of fucosylated glycans, such as SLeX and

LeX, on the breast cancer cells.42,43

We then examined if our system could discriminate more

complex and heterogeneous features of glycans derived from

cells with different origins (Fig. 3b). We examined 7 human

cancer lines using 6 different lectins that display varying levels of

specificity and compared the degree of glycan expression levels

on each cell surface. HDF was used as a control cell line because

it is commonly found in normal connective tissue and is well

characterized.44 Fig. 3b presents a heat map of the cancer lines

with clustering via quantitative analysis of fluorescence intensity.

The heat map shows the expression level and diversity of the

cell surface glycans as well as the diversity of glycan expression

on each cell. WGA shows consistently higher fluorescence than

other lectins for all cancer lines, confirming that cancer-

associated cells highly express terminal GlcNAc and sialic acid

epitopes.45,46 While the binding levels of concanavalin-A (ConA,

affinity for mannose) lectin47 are relatively lower than WGA for

each cancer line, the carbohydrate epitope is highly expressed in

kidney (ACHN) and colon (HCT116) cancer lines. We have also

examined another normal kidney cell line (HEK 293) against

ACHN to verify that HDF could act as a control in this assay,

and the overall quantitative results correspond accurately with

HDF (ESI, Fig. S3{). Next, the Maackia amurensis (MAA,

affinity for sialic acid a-2,3 linkage) lectin shows high levels of

fluorescence following binding to either HCT116 (4-fold

difference with HDF) or HeLa cells (3-fold difference with

HDF). The majority of the binding signal of MAA is attributed

to the sialic acid a-2,3 linkages on HCT116 and HeLa cells.

These results suggest that the alterations in cell surface glycans

Fig. 2 Characterization of the glycan profiling approach. (a) The

comparative result of WGA (50 nM) binding to MCF-7 and HDF cells.

Scale bars indicate 50 mm. (b) The comparison of our approach and the

96-well plate method for monitoring lectin binding (**p , 0.001). (c)

Inhibition assay of MCF-7 binding with GlcNAc. An inhibitor (GlcNAc)

at various concentrations was preincubated with WGA lectin (50 nM),

and each fluorescent signal was detected.

This journal is � The Royal Society of Chemistry 2012 Lab Chip, 2012, 12, 3290–3295 | 3293

Publ

ishe

d on

30

May

201

2. D

ownl

oade

d by

Uni

vers

ity o

f V

irgi

nia

on 1

0/11

/201

3 21

:23:

19.

View Article Online

could be a promising biomarker in response to cell types, status

of cellular physiology, or disease.

The results shown in Fig. 2 and 3 indicate that our facile

Qdot–lectin nanoconjugate-based technique could be readily

enlisted to examine surface glycan profiles of tissue sections

obtained from excisional biopsy. Furthermore, the method for

glycan analysis should also be validated in actual tissue samples

and cell lines because the glycan expression of a cell line may not

always correspond with that of their tissues.48 We have evaluated

glycan binding of breast and colon tissue with SBA and WGA,

respectively, (Fig. 4) because our previous results showed

relatively strong signals of breast cancer cell line (MCF-7) and

colon cancer cell line (HCT116) against SBA and WGA,

respectively (Fig. 3). In the breast cancer tissue, the binding

levels of SBA strongly correlate with primary tumor stages

rather than regional metastasis stages (Fig. 4a first row, ESI

Table S3, and S4{). The glycan epitope, GalNAc/Gal, in the

tissue sample of patient #3 (primary tumor size . 50 mm in the

greatest dimension) is 20-fold higher than in normal breast

tissues (Fig. 4b). These results clearly show consistency in the

high expression of the GlcNAc/Gal epitope between the breast

cancer tissues and the breast cancer cells (MCF-7, Fig. 3).

Additionally, the colon cancer tissue shows a higher signal

intensity for WGA binding than does normal colon tissue

(Fig. 4a, second row), indicating that the sialic acid a-2,3 linkage

or the terminal-GlcNAc is highly expressed in colon cancer

tissues. The signal intensity strongly depends on the level of

tumor differentiation, regardless of the location in the colon,

such as the rectum and sigmoid colon (Fig. 4b, ESI Table S5, and

S6{). These results clearly indicate that our glycan profiling

method can be used to analyze genuine human tissue specimens

for clinical diagnostic assays and to develop new analytical tools

for glycomics.

These results reveal, for the first time, that the expression of

specific carbohydrate epitopes on cancer tissues is closely related

with the primary tumor size and its stage of differentiation. The

abnormal glycosylation changes are precisely identified using our

quantitative glycan profiling method, which leads to a reliable

and quantitative comparison between cancer cell lines and

patient tissues. Thus, this work clearly demonstrates the utility of

our lectin-based assay for rapid, yet precise, profiling of complex

glycan motifs on cells or tissues.

Conclusions

In summary, we demonstrated a simple and powerful method

that combined cell patterning with Qdot–lectin nanoconjugates.

Our approach enabled systematic analysis of glycan expression

on live cell and tissue samples related to cancer proliferation and

Fig. 3 Glycan profiling of cell lines with 6 lectins displaying varying levels of specificity. (a) Representative optical and fluorescence images of MCF-7

and HDF cell lines with 6 lectins. Scale bars indicate 40 mm. (b) Heat map with hierarchical clustering for 7 types of cell surface glycans using 6 lectins.

Heat map is generated with Cluster 3.0 and Java TreeView.

Fig. 4 Specific glycan analysis of normal and cancer patient tissues. (a)

Representative optical and fluorescence images of breast (SBA lectin)

and colon (WGA lectin) tissue pairs. Scale bars indicate 300 mm. (b) Heat

map with hierarchical clustering for examined tissue. (**p , 0.001)

3294 | Lab Chip, 2012, 12, 3290–3295 This journal is � The Royal Society of Chemistry 2012

Publ

ishe

d on

30

May

201

2. D

ownl

oade

d by

Uni

vers

ity o

f V

irgi

nia

on 1

0/11

/201

3 21

:23:

19.

View Article Online

differentiation. Cell patterning provided consistent attachment,

well-confined cell spreading and proliferation, and high relia-

bility of quantitative cell surface glycan profiles. In addition,

Qdot–lectin nanoconjugates as glycan probes produced highly

stable fluorescent signals and exhibited high sensitivity.

The results showed reliable and consistent glycan profiling for

both cells and tissues, enabling facile analysis of cancer glycan

expression. The MCF-7 (breast cell line) highly expressed the

GalNAc/Gal and GlcNAc/sialic acid a-2,6 linkage rather than

the sialic acid a-2,3 linkage. On the other hand, GlcNAc/sialic

acid a-2,3 linkages on HCT116 (colon cancer cell line) are

dominant glycan epitopes. Importantly, their corresponding

tissues also displayed the abnormal glycosylation patterns. Thus,

the method clearly exhibited the potential utility for cancer

diagnostics using patient cell or tissue specimens.

We believe that our method could be readily employed to

select the lectin probes to identify target glycans that differentiate

between diseased and normal cell or tissue specimens. The

availability of this glycan profiling tool may be exploited to solve

even more challenging research questions in the field of

glycomics and to accurate analyze glycan heterogeneity on cell

or tissue surfaces.

Acknowledgements

The research was supported by the Converging Research Center

Program through the Ministry of Education, Science and

Technology (2011K000709) and the National Research

Foundation of Korea (NRF) grant funded by the Korea

government (MEST) (no. 2011-0017322).

References

1 A. Varki, Glycobiology, 1993, 3, 97.2 K. S. Lau, E. A. Partridge, A. Grigorian, C. I. Silvescu, V. N.

Reinhold, M. Demetriou and J. W. Dennis, Cell, 2007, 129, 123.3 J. D. Marth and P. K. Grewal, Nat. Rev. Immunol., 2008, 8, 874.4 S. T. Laughlin, J. M. Baskin, S. L. Amacher and C. R. Bertozzi,

Science, 2008, 320, 664.5 M. M. Fuster and J. D. Esko, Nat. Rev. Cancer, 2005, 5, 526.6 D. H. Dube and C. R. Bertozzi, Nat. Rev. Drug Discovery, 2005, 4,

477.7 K. Ohtsubo and J. D. Marth, Cell, 2006, 126, 855.8 J. W. Dennis, M. Granovsky and C. E. Warren, BioEssays, 1999, 21, 412.9 Y. Y. Zhao, M. Takahashi, J. G. Gu, E. Miyoshi, A. Matsumoto, S.

Kitazume and N. Taniguchi, Cancer Sci., 2008, 99, 1304.10 Y. G. Kim, G. C. Gil, K. S. Jang, S. Lee, H. I. Kim, J. S. Kim, J.

Chung, C. G. Park, D. J. Harvey and B. G. Kim, J. Mass Spectrom.,2009, 44, 1087.

11 R. Naka, S. Kamoda, A. Ishizuka, M. Kinoshita and K. Kakehi,J. Proteome Res., 2006, 5, 88.

12 S. Nakamura-Tsuruta, J. Kominami, M. Kamei, Y. Koyama, T.Suzuki, M. Isemura and J. Hirabayashi, J. Biochem., 2006, 140, 285.

13 X. Song, Y. Lasanajak, B. Xia, J. Heimburg-Molinaro, J. M. Rhea,H. Ju, C. Zhao, R. J. Molinaro, R. D. Cummings and D. F. Smith,Nat. Methods, 2011, 8, 85.

14 L. Nimrichter, A. Gargir, M. Gortler, R. T. Altstock, A. Shtevi, O.Weisshaus, E. Fire, N. Dotan and R. L. Schnaar, Glycobiology, 2004,14, 197.

15 X. Song, B. Xia, S. R. Stowell, Y. Lasanajak, D. F. Smith and R. D.Cummings, Chem. Biol., 2009, 16, 36.

16 D. Peelen, V. Kodoyianni, J. Lee, T. Zheng, M. R. Shortreed andL. M. Smith, J. Proteome Res., 2006, 5, 1580.

17 W. Cheng, L. Ding, S. Ding, Y. Yin and H. Ju, Angew. Chem., Int.Ed., 2009, 48, 6465.

18 K. L. Hsu, K. T. Pilobello and L. K. Mahal, Nat. Chem. Biol., 2006,2, 153.

19 K. T. Pilobello, D. E. Slawek and L. K. Mahal, Proc. Natl. Acad. Sci.U. S. A., 2007, 104, 11534.

20 S. Chen, T. Zheng, M. R. Shortreed, C. Alexander and L. M. Smith,Anal. Chem., 2007, 79, 5698.

21 H. Tateno, N. Uchiyama, A. Kuno, A. Togayachi, T. Sato, H.Narimatsu and J. Hirabayashi, Glycobiology, 2007, 17, 1138.

22 H. H. Jeong, H. M. Song and C. S. Lee, Clean Technology, 2011, 17,25.

23 C. H. Choi, J. H. Lee, Y. G. Kim, Y. H. Yang, K. M. Huh, T. S.Hwang and C. S. Lee, Macromol. Res., 2010, 18, 254.

24 G. Kang, J. H. Lee, C. S. Lee and Y. Nam, Lab Chip, 2009, 9, 3236.25 H. W. Shim, J. H. Lee, T. S. Hwang, Y. W. Rhee, Y. M. Bae, J. S.

Choi, J. Han and C. S. Lee, Biosens. Bioelectron., 2007, 22, 3188.26 F. Erogbogbo, K. T. Yong, I. Roy, G. Xu, P. N. Prasad and M. T.

Swihart, ACS Nano, 2008, 2, 873.27 C. Tekle, B. Deurs, K. Sandvig and T. G. Iversen, Nano Lett., 2008,

8, 1858.28 A. Cambi, D. S. Lidke, D. J. Arndt-Jovin, C. G. Figdor and T. M.

Jovin, Nano Lett., 2007, 7, 970.29 P. Roach, D. Farrar and C. C. Perry, J. Am. Chem. Soc., 2005, 127,

8168.30 T. Ishizaki, N. Saito and O. Takai, Langmuir, 2010, 26, 8147.31 S. Muller and F. G. Hanisch, J. Biol. Chem., 2002, 277, 26103.32 S. C. Tao, Y. Li, J. Zhou, J. Qian, R. L. Schnaar, Y. Zhang, I. J.

Goldstein, H. Zhu and J. P. Schneck, Glycobiology, 2008, 18, 761.33 V. Wittmann and S. Seeberger, Angew. Chem., Int. Ed., 2004, 43, 900.34 D. Zanini and R. Roy, Bioconjugate Chem., 1997, 8, 187.35 Z. Zhelev, H. Ohba, R. Bakalova, R. Jose, S. Fukuoka, T. Nagase,

M. Ishikawa and Y. Baba, Chem. Commun., 2005, 1980.36 H. Y. Xie, M. Xie, Z. L. Zhang, Y. M. Long, X. Liu, M. L. Tang,

D. W. Pang, Z. Tan, C. Dickinson and W. Zhou, Bioconjugate Chem.,2007, 18, 1749.

37 M. Xie, J. Hu, Y. M. Long, Z. L. Zhang, H. Y. Xie and D. W. Pang,Biosens. Bioelectron., 2009, 24, 1311.

38 M. Y. Lee, R. A. Kumar, S. M. Sukumaran, M. G. Hogg, D. S. Clarkand J. S. Dordick, Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 59.

39 O. E. Galanina, M. Mecklenburg, N. E. Nifantiev, G. V. Pazyninaand N. V. Bovin, Lab Chip, 2003, 3, 260.

40 A. S. Powlesland, P. G. Hitchen, S. Parry, S. A. Graham, M. M.Barrio, M. T. Elola, J. Mordoh, A. Dell, K. Drickamer and M. E.Taylor, Glycobiology, 2009, 19, 899.

41 T. Freire, S. Bay, S. von Mensdorff-Pouilly and E. Osinaga, CancerRes., 2005, 65, 7880.

42 K. Yuan, C. M. Listinsky, R. K. Singh, J. J. Listinsky and G. P.Siegal, Pathol. Oncol. Res., 2008, 14, 145.

43 M. T. Elola, M. I. Capurro, M. M. Barrio, P. J. Coombs, M. E.Taylor, K. Drickamer and J. Mordoh, Breast Cancer Res. Treat.,2007, 101, 161.

44 K. Hedman, M. Kurkinen, K. Alitalo, A. Vaheri, S. Johansson andM. Hook, J. Cell Biol., 1979, 81, 83.

45 T. Satomaa, A. Heiskanen, I. Leonardsson, J. Angstrom, A. Olonen,M. Blomqvist, N. Salovuori, C. Haglund, S. Teneberg, J. Natunen,O. Carpen and J. Saarinen, Cancer Res., 2009, 69, 5811.

46 K. S. Lau and J. W. Dennis, Glycobiology, 2008, 18, 750.47 G. Alton, M. Hasilik, R. Niehues, K. Panneerselvam, J. R. Etchison,

F. Fana and H. H. Freeze, Glycobiology, 1998, 8, 285.48 R. Sandberg and I. Ernberg, Proc. Natl. Acad. Sci. U. S. A., 2005,

102, 2052.

This journal is � The Royal Society of Chemistry 2012 Lab Chip, 2012, 12, 3290–3295 | 3295

Publ

ishe

d on

30

May

201

2. D

ownl

oade

d by

Uni

vers

ity o

f V

irgi

nia

on 1

0/11

/201

3 21

:23:

19.

View Article Online