Towards Examining FGF21 Secretion from Pancreatic Islets in a Microfluidic Device

by

Brenda Green

A thesis submitted in conformity with the requirements for the degree of Master of Applied Science

Graduate Department of

Institute of Biomaterials and Biomedical Engineering

University of Toronto

© Copyright by Brenda Green (2012)

ii

Towards Examining FGF21 Secretion from Pancreatic Islets in a

Microfluidic Device

Brenda Green

Master of Applied Science

Institute of Biomaterials and Biomedical Engineering University of Toronto

2012

Abstract

Type II diabetes is characterized by impaired insulin-secretion from pancreatic β- cells. A

novel endocrine factor, fibroblast growth factor 21 (FGF21) is produced by the liver and

improves pancreatic β- cell function and survival. However, a state of FGF21- resistance has

been described in obese mice and humans.

This thesis examines (1) whether pancreatic islets temporally express FGF21 and (2) if

high glucose and fat reduce transcript expression of FGF21 receptors in islets. This thesis further

characterizes a novel microfluidic device that will be used to collect and assay FGF21.

Our data indicate that palmitate induces maximal production of FGF21 mRNA in β- cells

after 16 hours of stimulation through PPARγ. Secondly, we report that FGF21, glucose and

palmitate do not cause FGF21-resistance by down-regulating receptor expression. Overall, we

present a mechanism for the induction of FGF21 from islets, and describe a microfluidic device

for culture and secretion studies.

iii

Acknowledgments

I would like to thank my supervisor Professor Rocheleau for his guidance and

encouragement throughout my Master’s thesis. He has taught me valuable research approaches,

and thorough experimental techniques. I would also like to thank my committee members

Professor Volchuk and Professor Sefton for their helpful comments and directions.

In addition, I want to thank Svetlana Altamentova for all of the animal work, and all the

Rocheleau lab members for being there to lend a helping hand. I would like to acknowledge the

Simmons and Guenther labs for their assistance with microfluidic modeling and fabrication.

Lastly, I want to thank my friends and family for their support and understanding while I

have been studying. I could not have accomplished this without their encouragement.

iv

Table of Contents

Acknowledgments .......................................................................................................................... iii 

Table of Contents ........................................................................................................................... iv 

List of Abbreviations ..................................................................................................................... vi 

List of Tables ............................................................................................................................... viii 

List of Figures ................................................................................................................................ ix 

List of Appendices .......................................................................................................................... x 

Chapter 1 - Introduction .................................................................................................................. 1 1.1  Type II Diabetes .................................................................................................................. 1 1.1.1  Glucolipoxicity Model ................................................................................................... 2 1.2  FGFs and FGF21 ................................................................................................................. 4 1.2.1  Fibroblast Growth Factors.............................................................................................. 4 1.2.2  Fibroblast Growth Factor Receptors .............................................................................. 5 1.2.3  Klotho Co-Receptors ...................................................................................................... 7 1.2.4  Paracrine and Endocrine FGFs....................................................................................... 7 1.2.5  FGF Receptor Signaling Pathway .................................................................................. 9 1.2.6  FGF21 Expression and Function .................................................................................. 11 

Chapter 2 – Related Literature ...................................................................................................... 14 2.1 FGF21 Secretion ................................................................................................................ 14 2.2 FGF21 Resistance .............................................................................................................. 17 

Chapter 3 – Research Rationale .................................................................................................... 21 3.1 Rationale ............................................................................................................................ 21 3.2 Hypothesis and Objectives ................................................................................................. 22 

Chapter 4 – Materials and Methods .............................................................................................. 23 4.1 Pancreatic Islet Isolation, Tissue Culture and Cell Culture ............................................... 23 4.2 Fatty Acid Preparation ....................................................................................................... 23 4.3 RNA Isolation, reverse transcription PCR and quantitative PCR ...................................... 23 4.4 Immunoflourescence .......................................................................................................... 24 4.5 Microfluidics ...................................................................................................................... 25 4.5.1 Device Fabrication and Design ....................................................................................... 25 4.5.2 Microfluidic COMSOL modeling ................................................................................... 26 4.5.3 Microfluidic device long- term islet culture .................................................................... 27 4.5.4 Microfluidic device image application ............................................................................ 27 4.6 NAD(P)H Imaging ............................................................................................................. 28 4.7 Spherotech Velocity Profiles and Flou4 Ca2+ Oscillations ................................................ 29 4.8 Statistical Analysis ............................................................................................................. 30 

v

Chapter 5 – FGF21 Secretion and Resistance ............................................................................... 31 5.1 Introduction ........................................................................................................................ 31 5.2 Results ................................................................................................................................ 32 5.2.1 Verifying FGF21 Responsiveness ................................................................................... 32 5.2.2. FGF21 expression is higher in mouse islets compared to liver and WAT ..................... 33 5.2.3 Palmitate, Bezafibrate and Rosiglitazone induce FGF21 expression in Min6 β- cells

and in islets ....................................................................................................................... 34 5.2.4. Palmitate- induces FGF21 expression in a temporal manner in Min6 β- cells .............. 36 5.3 Discussion .......................................................................................................................... 39 

Chapter 6 – Microfluidic Device for Islet Culture ........................................................................ 41 6.1 Introduction ........................................................................................................................ 41 6.2 Results ................................................................................................................................ 42 6.2.1 The bypass microfluidic device is designed to hold islets in series while allowing media exchange through the tissue ............................................................................... 42 6.2.3 Pressure drop induces intercellular flow through the islet ............................................ 45 6.2.4 Islets cultured in the bypass microfluidic device have enhanced EC morphology relative to islets cultured in a dish ................................................................................. 48 6.2.5 Islets cultured in the bypass microfluidic device for 48 hours show glucose- stimulated Ca2+ oscillations ........................................................................................... 49 6.3 Discussion .......................................................................................................................... 53 

Chapter 7 – Conclusion ................................................................................................................. 55 

Appendix A – FGF21 Circadian Response ................................................................................... 56 A.1 Introduction ....................................................................................................................... 56 A.2 Results ............................................................................................................................... 57 

Appendix B – Microfluidic Calcium Staining with Fluo4 and β-Cyclodextrin ............................ 59 B.1 Introduction ....................................................................................................................... 59 B.2 Results ............................................................................................................................... 59 

Appendix C – Metabolic Gene Analysis and Insulin Secretion ................................................... 61 C.1 Introduction ....................................................................................................................... 61 C.2 Results ............................................................................................................................... 62 

Chapter 8 – References ................................................................................................................. 65 

vi

List of Abbreviations

Acronyms

ACC Acetyl-CoA Carboxylase Akt Serine/ threonine protein kinase B ATF6 Activating transcription factor 6 ATP Adenosine Triphosphate BMHH Imaging Buffer BSA Bovine Serum Albumin Ca2+ Intracellular calcium CD31 Cluster of differentiation 31 CDCA Chenodeoxycholic Acid CPT1 Carnitine Palmitoyl Transferase I CT Cycle Threshold for qPCR DAG Diacylglycerol DIO Diet induced obese mice EC Endothelial cell ECM Extracellular matrix ELISA Enzyme-linked Immunosorbent Assay ER Endoplasmic reticulum ERK1/2 Extracellular signal-regulated kinase ERO1 ER oxidoreductin 1 FACoAs Free fatty acid long- chain acyl coenzyme A esters FBS Fetal bovine serum FFA Free Fatty Acids FFT Fast fourier transform FGF21 Fibroblast Growth Factor 21 FGFR Fibroblast Growth Factor Receptor FXR Farnesoid X receptor HEPES 4(2-hydroxyethyl)-1-piperazineethanesulfonic acid HSPG Heparin Sulfate Proteogylcans Idh1 Isocitrate dehydrogenase 1 IgG Immunoglobulin G IRE1α Inositol-requiring 1α JNK JUN N-terminal kinase KATP ATP- sensitive potassium channel KLB Klotho- β MAPK Mitogen-activated Protein Kinase NAD(P)H Aggregate of NADH and NADPH NADH Nicotinamide Adenine Dinucleotide NADPH Nicotinamide Adenine Dinucleotide Phosphate

vii

NF-κB Nuclear transcription factor κB NNT Nicotinamide nucleotide transhydrogenase iNOS Inducible nitric oxide synthase PECAM Platelet endothelial cell adhesion molecule PERK Protein kinase- like ER kinase PC Pyruvate carboxylase PDI Protein disulphide isomerase PDMS Polydimethylsiloxane Pdk4 Pyruvate dehydrogenase lipoamide kinase isozyme 4 PI3K Phosphoinositide 3-kinase PLC Phospholipase C PPAR Peroxisome Proliferator-activated Receptor ROS Reactive oxygen species RPMI Cell and tissue culture media RT-PCR Reverse-Transcription Polymerase Chain Reaction RZD Rosiglitazone Spry Sprouty proteins STAT Signal Transducer and Activator of Transcription TCA Tricarboxylic acid cycle TZD Thiazolidinedione UCP2 Uncoupling protein 2 UPR Unfolded protein response VEGF Vascular endothelial growth factor WAT White Adipose Tissue ZT Zeitgeber time

viii

List of Tables

Table 1.1 – Fibroblast growth factor subfamilies ………………………...……………..…. 4

Table 1.2 – Endocrine FGFs and Klotho ……………….……………………………….… 7

Table 5.1 – In vivo and in vitro concentrations of FGF21 ………………...…………….… 32 Table 6.2.1 – Fluid parameters used to determine the pressure drop across the islet holding area in the direct- flow and bypass microfluidic devices…………..……………… 48 Table 6.2.2 – Shear stress calculated around the periphery of an islet in the direct flow and bypass microfluidic devices ………………………………………..……………………… 51

ix

List of Figures

Figure 1.1. Pancreatic islet of Langerhan………………………………….……….......................1 Figure 1.2. Fibroblast growth factor subtypes .............................................................................. 5 Figure 1.3. FGFR structure and dimerization ............................................................................... 6 Figure 1.4. Endocrine FGFs ........................................................................................................... 8 Figure 1.5. FGFR signaling network ........................................................................................... 10 Figure 1.6. Anatomic FGF21 qPCR expression profiles in mouse tissues from C57BL6 mice. . 11 Figure 1.7. Relative FGF21 mRNA expression levels in metabolic tissue of C57BL6 lean,  diet induced obese and ob/ob mice ............................................................................................... 12 Figure 1.8. Pancreatic β- cell metabolic pathways ....................................................................... 13 Figure 2.1. Protein levels of FGF21 in the supernatant of liver cells .......................................... 14 Figure 2.2. CDCA bile acids induce hepatic FGF21 secretion from HepG2 cells ...................... 15 Figure 2.3. Model of the PPARγ - FGF21 regulatory pathway in WAT tissue .......................... 16 Figure 2.4. FGF receptor and KLB mRNA expression in WAT of lean and diet- induced obese mice ....................................................................................................................... 17 Figure 2.5. Temporal mRNA expression profile of FGF21 in the liver ...................................... 18 Figure 2.6. Simplified model of a pancreatic islet consisting of clustered pancreatic β- cells…..20 Figure 4.1. Representative microfluidic device used for image analysis ..................................... 28 Figure 4.2. Oscillations and power spectral density plot of glucose- stimulated Ca2+ responses in pancreatic islets ....................................................................................................... 30 Figure. 5.2.1. Mitochondrial NAD(P)H dose response for islets treated with FGF21 ................ 32 Figure. 5.2.2. FGF21 expression is higher in mouse islets compared to liver and WAT ............ 33 Figure. 5.2.3. Palmitate induces FGF21 expression in β- cells, α- cell and islets ....................... 35 Figure. 5.2.4. Palmitate induces FGF21- expression in a temporal manner ................................ 36 Figure. 5.2.5. FGF21, glucose and palmitate treatment do not change FGF21, FGF receptor and co-receptor expression levels. ............................................................................................... 38 Figure 6.2.1. Direct- flow and bypass microfluidic device prototypes ....................................... 46 Figure 6.2.2. Comsol CFD velocity and pressure profiles through the microfluidic devices ...... 44 Figure 6.2.3. Media exchange through islets converges at 30 seconds as the flow rate increases ........................................................................................................................................ 46 Figure 6.2.4. Islets cultured in the short- pass device have enhanced EC morphology relative to islets cultured in a dish ................................................................................................. 49 Figure 6.2.5. Spherotech beads flowing in the short- pass microfluidic device………...............50 Figure 6.2.6. Periods of Ca2+ oscillations in islets cultured for 48 hours in a dish and short-pass device ............................................................................................................................................ 52 Figure A.1.1. The experimental design for examining the FGF21- circadian response .............. 56 Figure A.2.1. Islet mitochondrial and cytoplasmic NAD(P)H responses to glucose ................... 58 Figure B.2.1. β-Cyclodextrin increases the penetration depth of Fluo4 into the islet……..........60 Figure C.2.1. FGF21 does not alter the mRNA expression of UCP2, NNT, PC, Pdk4 or Idh1 in mouse islets………………………………………………………………………………………63 Figure C.2.2. Palmitate and glucose- stimulated insulin secretion……………………………...64

x

List of Appendices

Appendix A – FGF21- Circadian Response …………………………………….…...….56

Appendix B – Microfluidic Calcium Staining with Fluo4 and β-Cyclodextrin…..……..59

Appendix C – Metabolic Gene Analysis and Insulin Secretion…………………………61

1

Chapter 1 - Introduction

1.1 Type II Diabetes Diabetes mellitus is a group of metabolic diseases, which are characterized by high blood

sugar. This is a complex disease that involves the endocrine pancreas, an important organ

composed of the islets of Langerhans (Figure 1.1). Islets contain β- cells, the cells responsible for

the production and secretion of the hormone insulin.

Figure 1.1. Pancreatic islet of Langerhan. Islets are cell aggregates composed of insulin- secreting β- cells, α-, δ- , ε- , PP- cells, endothelial cells, nerves and extracellular matrix. The endocrine cells (α, δ, ε, PP) secrete glucagon, somatostatin, ghrelin and pancreatic polypeptide, respectively. Typically, β- cells consist of 75% of the islet, α- cells 15%, δ- cells 5%, PP cells 4% and ε- cells

2

We chose to examine fibroblast growth factor 21 (FGF21) due to its therapeutic potential

for type II diabetes. FGF21 is shown to have positive effects on fatty acid metabolism in liver,

white adipose tissue and pancreatic islets [5, 14].

1.1.1 Glucolipoxicity Model

The clinical state of type II diabetes is often accompanied by elevated blood glucose and

fatty acid levels, creating a glucolipotoxic environment in the β- cell. Excess carbohydrates and

fat in the circulation cause hyperinsulinemia and chronic insulin resistance, ultimately resulting

in β- cell failure [8]. Several pathways have been implicated in glucolipotoxicity including

endoplasmic reticulum (ER) stress, oxidative stress and the activation of inflammatory pathways

in pancreatic islets [6, 7]. Prentki et al. (2002) describe a role for malonyl- CoA, peroxisome

proliferator-activated receptors (PPARα and -γ) in the development of glucolipotoxicity.

The ER is responsible for protein synthesis, folding and assembly of many proteins, and

also plays an important role in sensing cellular stress [6]. Chronic glucose and lipid exposure

increase the demand for protein folding, creating an imbalance between the protein- folding

capacity of the ER. Unfolded or misfolded proteins accumulate in the ER lumen, creating a

condition called ER stress. Eukaryotic cells have developed the unfolded- protein response

(UPR) to cope with stressful conditions and the protein- folding defect. ER sensors inositol-

requiring 1α (IRE1α), protein kinase- like ER kinase (PERK) and activating transcription factor

6 (ATF6) sense unfolded proteins and trigger a pathway to inhibit transcription or translation

processes. If the cell fails to prevent ER stress, the UPR will initiate apoptosis through the CHOP

transcription factor [6].

The signaling pathways in the UPR and inflammation are connected to the production of

reactive oxygen species (ROS) and the activation of nuclear transcription factor (NF-κB) and

JUN N-terminal kinase (JNK). ROS are small molecules that are highly reactive due to the

presence of unpaired electrons and the accumulation of these molecules results in a state of

oxidative stress [6]. Protein folding in the ER consumes energy for the formation of intra- and

intermolecular disulphide bonds. The electron transport during disulphide bond formation is

driven by ER proteins: protein disulphide isomerase (PDI) and ER oxidoreductin 1 (ERO1).

ERO1 creates an oxidizing environment involving electron transfer from PDI to molecular

oxygen. The transfer of electrons to molecular oxygen results in the production of ROS. The

3

buildup of ROS is particularly detrimental to the pancreatic islet, due to low levels of intrinsic

antioxidant defenses. Excess ROS generation may initiate an inflammatory response [6].

The inflammatory response occurs when cells of the immune system sense cellular

damage, and trigger the release of inflammatory substances, including cytokines, free radicals

and hormones. Several inflammatory cytokines (TNF-α, IL-6) cause ER stress and activate the

UPR [6, 9]. The PERK pathway of the UPR can relieve oxidative stress by triggering the

induction of oxidant- detoxifying enzymes, including glutathione and NAD(P)H- quinone

oxidoreductase. Altogether, β- cell glucolipotoxicity triggers the unfolded- protein response in

the ER which is linked to oxidative stress and inflammation.

Prentki et al. (2002) describe a glucolipotoxicity model that implicates malonyl- CoA and

PPARα and-γ. Malonyl-CoA is a metabolic signaling molecule that controls the relative flux of

fatty acid oxidation and esterification. When nutrient levels are high, glucose is metabolized to

citrate, which is converted to malonyl-CoA in the cytoplasm. This is the first committed step of

fatty acid synthesis. Malonyl-CoA also acts to inhibit the transport of fatty acids into the

mitochondria, thus preventing fatty acid oxidation [8]. The combination of elevated free fatty

acids (FFAs) and glucose increase the concentration of FFA- derived long- chain acyl-CoA

esters (FACoAs) in the cytoplasm, which cause lipoapoptosis through nitric oxide cytotoxicity

[8]. Nitric oxide is known to form complexes in islets that result in an inhibitory effect on the

tricarboxylic acid cycle (TCA) and ATP production [31].

PPARs are nuclear receptor transcription factors that regulate diverse aspects of lipid

metabolism [10]. PPARα stimulates lipid catabolism mainly in the liver by upregulating genes

involved in fatty acid oxidation [10]. On the other hand, PPARγ causes increased lipid uptake

and lipogenic processes in adipocytes [11]. Circulating FFAs are natural agonists to PPARs.

Prentki et al. (2002) show that elevated glucose decreases the expression of PPARα, and induces

the expression of PPARγ; thus, favoring the lipid synthesis pathway. This leaves only one

option for FFAs, to undergo esterification, which then causes progressive lipid accumulation and

β- cell failure [8].

The lipid partitioning model of β- cell failure in type II diabetes is considered to be

complementary to the oxidative stress hypothesis. Saturated fatty acids such as palmitate

increase the production of ROS in rat islets [8]. Both models reveal the biochemical basis of β-

4

cell adaptation and failure in the progression of type II diabetes. It is therefore critical to study

the cell mechanisms that regulate β- cell activity during high nutrient loading and develop a

therapeutic to maintain mass and function.

1.2 FGFs and FGF21

1.2.1 Fibroblast Growth Factors

Fibroblast growth factors (FGFs) are members of the largest family of growth factor

ligands, comprising 22 members (Table 1.1, Figure 1.2). FGFs are structurally related

polypeptides that are necessary for embryonic development, proliferation, wound healing, and

angiogenesis; additionally they can function as hormones that regulate metabolism [12]. Human

FGFs are 150- 300 amino acid polypeptides with a conserved core of 120 amino acids exhibiting

30- 60% homology [12].

Most FGFs function in a paracrine fashion. These FGFs are secreted glycoproteins that

are generally sequestered to the extracellular matrix (ECM), as well at the cell surface by

heparan sulphate proteoglylcans (HSPGs). The cell surface HSPGs stabilize FGF ligand-

receptor interaction, forming a ternary complex with FGF receptors (FGFR) [13].

Endocrine FGFs, such as FGF21, are of recent interest due to their homeostatic and

metabolic effects [14]. They are able to act over a distance due to their low affinity for HSPGs;

however, they require Klotho- β (KLB) as a co-receptor for signaling.

Table 1.1 – Fibroblast growth factor subfamilies (Figure 1.2)

Intracrine Intracellular signaling proteins that do not bind FGF receptors and regulate voltage gated sodium channels.

Paracrine Canonical FGFs that mediate biological responses as extracellular proteins by binding and activating FGFRs with heparan sulphate as a cofactor. Paracrine FGFs act as growth and differentiation factors during development.

Endocrine Hormone like- FGFs that mediate biological responses in an FGFR- dependent manner. These FGFs bind to heparan sulphate with very low affinity, which enables them to function in an endocrine manner [15].

5

Figure 1.2. Fibroblast growth factor subtypes. The subtypes consist of intracrine, paracrine and endocrine members. Phylogenetic analysis suggests that FGF genes can be arranged into seven subfamilies, containing two to four members each. Branch lengths are proportional to the evolutionary distance between each gene [12].

1.2.2 Fibroblast Growth Factor Receptors

The FGF receptors consist of an extracellular ligand domain composed of two or three

immunoglobulin- like domains (IgG), a single transmembrane helix domain, and an intracellular

domain with tyrosine kinase activity (Figure 1.3) [16]. The IgG domain is an essential

determinant of ligand- binding specificity. FGF receptors signal as dimers, where ligand-

dependent dimerization leads to a conformational shift in receptor structure that activates the

6

intracellular kinase domain [13]. The FGF receptors are encoded by four genes (Fgfr-1 to Fgfr-

4), but because of alternative splicing of Fgfr-1, Fgfr-2 and Fgfr-3, seven prototype receptors are

generated (1b, 1c, 2b, 2c, 3b, 3c and 4) [16]. The specificity of the FGF-FGFR interaction is

established by the different ligand-binding capacities of the receptor isoforms and tissue- specific

expression of ligands, receptors and other transmembrane proteins (such as KLB) [13].

Figure 1.3. FGFR structure and dimerization. (A) FGF binds and is sequestered to the extracellular matrix. (B) FGF signal transduction is initiated upon binding of the FGF ligand in conjunction with heparan sulphate to form a ternary complex. The result is trans- autophosphorylation and activation of the tyrosine kinase, which facilitates second messenger signaling. (C) FGF receptor complex consists of extracellular Ig domains, single transmembrane domain, and a cytoplasmic tyrosine kinase domain [16].

A B C

7

1.2.3 Klotho Co-Receptors

The Klotho gene encodes a single-pass transmembrane protein. Klotho forms complexes

with FGFRs and increases their affinity to endocrine FGFs. These receptor complex formations

are required to induce phosphorylation and signal transduction. The expression of Klotho

proteins and FGFR splice variants determines the specificity of endocrine FGF signaling [15]. It

has been shown that FGF21 binds Klotho- β (KLB) and FGFR1c and 2c in liver, white adipose

tissue (WAT) and pancreas. KLB has a relatively large extracellular domain with 11 putative N-

glycosylation sites, two inactive glycosyl hydrolase-1 domains, and a relatively short C-terminal

tail [17].

1.2.4 Paracrine and Endocrine FGFs

Paracrine FGFs function in development, cell proliferation and migration by influencing

the intracellular signaling events of neighboring cells. The range of FGF signaling is regulated in

part by its affinity for the extracellular matrix and in part by dimerization of some FGFs [12].

Examples of paracrine FGFs include FGF4 and FGF8, which have essential roles in blastocyst

formation and gastrulation.

Endocrine FGFs function through FGFRs and act over long distances as hormone- like

factors (Table 1.2). These hormones have been recently appreciated as endocrine factors (in the

past 5 years), in contrast to paracrine FGFs, which have been a focus of research over the last 40

years [18]. There is a growing interest in understanding the effects of endocrine FGFs.

Table 1.2 – Endocrine FGFs and Klotho (Figure 1.4)

FGF15/ 19 FGF15/19 are orthologous genes in vertebrates where FGF15 is found in mice and FGF19 is found in humans. FGF15/19 is secreted from the ileum upon feeding. These FGFs act on hepatocytes to reduce bile acid synthesis through suppression of transcription factors. They are also shown to be involved in cell proliferation. Binds Klotho- β and FGFR4 found in the liver and gallbladder.

FGF21 FGF21 is secreted by the liver and adipocytes and is shown to regulate energy homeostasis (glucose and lipid metabolism). Binds Klotho- β and FGFR1c and 2c in liver, WAT and pancreas.

FGF23 FGF23 is secreted by bone, and is involved in phosphate and vitamin D homeostasis in the kidney. Binds Klotho- α and FGFR1c in the kidney [15].

8

Figure 1.4. Endocrine FGFs. FGF signaling is regulated by the expression of Klotho co- receptors. FGF15/19 is secreted from ileum upon feeding and acts on the liver and gallbladder via Klotho- β and FGFR4. FGF21 is secreted from the liver upon fasting and acts on WAT through Klotho- β and FGFR1c/2c. FGF23 is secreted from the bone in response to vitamin D and acts on the kidney though Klotho- α and FGFR1c [15].

Ileum  Liver Bone 

FGF15/19  FGF21 FGF23

FGFR4 βKlotho  βKlotho αKlotho FGFR1cor 2c

FGFR1c

Bile acid homeostasis Bile acid synthesis (Liver) ↓ Bile acid transport (Ileum) ↓ 

Energy homeostasis Glucose uptake (WAT) ↑

Phosphate & Vitamin D homeostasis 

Phosphate excretion (Kidney) ↑Vitamin D synthesis (Kidney) ↓

9

1.2.5 FGF Receptor Signaling Pathway

The FGF- FGFR ternary complex formation results in trans- autophosphorylation of the

receptor tyrosine kinase domains. The phosphorylated tyrosine residues function as docking sites

for adaptor proteins that may also be directly phosphorylated by FGFR, leading to the activation

of multiple signaling pathways (Figure 1.5) [13].

FGF binding to its receptor and co-receptor results in the activation of four key downstream

signaling pathways: (1) RAS-RAF-MAPK, (2) PI3K-AKT, (3) STAT and (4) PLCγ/ Ca2+ [12].

There are several mechanisms that attenuate FGFR signaling which are only partly understood

[13]. Sprouty (Spry) proteins, SEF family members and Stat transcription factors can modulate

receptor signaling at several points.

Spry proteins are thought to function in a dominant negative manner, by competing for

adaptor protein GRB2 binding, and preventing SOS-mediated RAS activation, or by directly

binding to RAF and blocking subsequent MAPK signaling [13].

SEF may function at multiple levels including a transmembrane form that can directly

interact with FGFRs and an intracellular form capable of inhibiting ERK1/2 phosphorylation.

STAT signaling transducer can translocate to the nucleus and affect transcription of target

genes (such as growth inhibitor, p21).

10

Figure 1.5. FGFR signaling network. The signal transduction network downstream of fibroblast growth factor receptors (FGFRs), along with negative regulators. Following ligand binding and receptor dimerization, the kinase domains transphosphorylate each other, leading to the docking of adaptor proteins and the activation of four key downstream pathways:

(1) RAS–RAF–MAPK [cell growth, differentiation, survival] (2) PI3K–AKT [proliferation, apoptosis] (3) Signal transducer and activator of transcription (STAT) [transcription activity] (4) Phospholipase Cγ (PLCγ) [intracellular calcium concentrations]

DAG, diacylglycerol; FRS2α, FGFR substrate 2α; GRB2, growth factor receptor-bound 2; IP3, inositol triphosphate; P, phosphorylation; PIP2, phosphatidylinositol-4,5-biphosphate; PKC, protein kinase C; Sos, son of sevenless [13].

FGF signaling pathways

11

1.2.6 FGF21 Expression and Function

Fibroblast growth factor 21 is an endocrine polypeptide that is a promising clinical

candidate for the treatment of type II diabetes [19, 20, 21]. FGF21 is a significant hormonal

regulator of glucose, lipids and energy balance. Endogenous FGF21 facilitates insulin-

independent uptake of glucose in adipocytes and promotes lipolysis [22, 23]. Our lab has also

demonstrated FGF21- responsiveness in pancreatic islets [5]. The relative mRNA tissue

expression in C57BL6 mice is presented in Figure 1.6 [24]. FGF21 is present is most tissues,

with elevated levels in the pancreas, testis, and liver. This thesis specifically explores FGF21

expression in the pancreatic islet α- and β- cells.

Figure 1.6. Anatomic FGF21 qPCR expression profiles in mouse tissues from C57BL6 mice. mRNA expressed relative to 18S rRNA [24].

The mRNA expression of FGF21 is examined in diet- induced obese (DIO) and ob/ob

mice. DIO mice are fed a high- fat diet for 8- 12 weeks. They become obese, moderately

hyperglycemic and develop impaired glucose tolerance. Typical serum glucose concentrations

and free fatty acids levels are ~9 mM and 0.5 mM, respectively [25]. Ob/ob mice are a mutant

strain that lack the hormone leptin, which controls appetite. These mice develop high blood sugar

(15- 20 mM) and increased levels of insulin. Both models are used in studies of type II diabetes.

Relative

 mRN

A expression 

   FGF21 

Eye

Bra

in S

tem

Cer

ebel

lum

Cer

ebru

mC

orpu

s St

riatu

mO

lfact

ory

Bul

bSp

inal

Cor

dH

ypot

hala

mus

Pitu

itary

Adr

enal

Thyr

oid

Panc

reas

Tong

ueSt

omac

hD

uode

num

Jeju

num

Ileum

Col

onG

all B

ladd

e rLi

ver

Kid

ney

BA

TW

AT

Sple

enTh

ymus

Ova

ryU

teru

sEp

idid

ymus

Prep

utia

l Gla

ndPr

osta

teSe

min

al V

esic

les

Test

isVa

s D

efer

ens

Aor

taH

eart

Lung

Bon

eM

uscl

eSk

in

12

FGF21 expression levels are significantly induced in the liver of ob/ob mice compared to

adipose and pancreatic tissue (Figure 1.7) [26].

Figure 1.7. Relative FGF21 mRNA expression levels in metabolic tissue of C57BL6 lean, diet induced obese and ob/ob mice. Tissues were collected from ad libitum fed lean (open bars, 8 and 16 wk old), DIO (gray bars, 16 wk old), or ob/ob mice (black bars, 8 wk old). Data were normalized to the cyclophilin control gene and are expressed as fold change in FGF21 mRNA levels relative to the level seen in the liver of lean mice [26].

The expression level of FGF21 in the pancreas of lean mice is affected by the diet. This

led us to consider FGF21 signaling in pancreatic islets treated with excess nutrients. In islets,

FGF21 causes a reduction in acetyl CoA carboxylase (ACC) levels, which we postulate leads to

improved fatty acid transport and oxidation [5]. As shown in Figure 1.8, high glucose and fat

cause a build up of citrate which is generated from the tri-carboxylic acid cycle (TCA). Citrate is

converted to malonyl CoA, in the presence of ACC, for fatty acid synthesis. Malonyl CoA

inhibits fatty acid transport into the mitochondria by blocking carnitine palmityl transferase

(CPT1), thereby blocking mitochondrial fatty acid oxidation [27]. Overall, ACC regulates the

levels of malonyl CoA, to ultimately regulate the levels of fatty acid synthesis and fatty acid

oxidation in the cell.

Relative

 mRN

A expression 

13

The TCA cycle produces NADH as a key energy intermediate. NADH is ultimately

converted to cellular energy in the form of ATP. NADPH is consumed during fatty acid

synthesis. The aggregate signal of NADH and NADPH (NAD(P)H) is a readout of the metabolic

state of the cell.

Figure 1.8. Pancreatic β- cell metabolic pathways

FGF21 improves pancreatic β- cell function and survival by activation of ERK1/2 and

Akt signaling pathways [20]. Wente et al. (2006) showed that in vitro FGF21 treatment of islets

isolated from healthy rats enhanced insulin mRNA and protein levels compared to the control.

FGF21 treatment also caused rapid phosphorylation of ERK1/2 and Akt. It has been previously

demonstrated that FGF21 is mitogenically inactive in vitro when tested with FGF- sensitive cell

lines [20]. Thus, FGF21 promotes β- cell survival and function, without inducing mitogenicity.

Fatty AcylCoA 

Pyruvate 

Malonyl CoA 

Citrate Palmitate

ACC

Acetyl CoA

Citrate

Glucose Metabolism

Fatty Acid Oxidation

Fatty Acid Synthesis 

Cytoplasm  Mitochondria

TCA Cycle 

↑NADH

↑ NADH 

↑ NADH

↓ NADPH 

CPT1 

14

Chapter 2 – Related Literature

2.1 FGF21 Secretion

FGF21 is known to be mainly secreted from the liver via PPARα and acts as an endocrine

factor [28]. Free fatty acids (FFA) induce the expression of PPARα, resulting in the downstream

production of FGF21 (Figure 2.1) [29]. This thesis explores PPAR- induced FGF21 transcript

expression in islets and β- cells.

Figure 2.1. Protein levels of FGF21 in the supernatant of liver cells. Human liver hepatocellular carcinoma cells (HepG2) are stimulated with an 0.16 mM FFA mixture (palmitic, oleic and linoleic acid) for 1, 2, 4, 8 and 24 h. Results are expressed as means (± sem) *P

15

Figure 2.2. CDCA bile acids induce hepatic FGF21 secretion from HepG2 cells. Cells were incubated with 100 μM CDCA for 6 or 12 hours. The level of FGF21 and albumin in the culture medium was measured by Western analysis [30]

FGF21 induction in adipocytes occurs via PPARγ agonists during the fed-state. PPARγ

induction has beneficial effects for lipotoxic human islets, through antioxidant properties and

reduction of iNOS [31]. Dutchak et al. (2012) decided to investigate the mechanisms of FGF21

secretion from adipocytes.

PPARγ agonists include thiazolidinediones (TZDs): rosiglitazone and pioglitazone.

PPARγ activity is regulated by posttranslational modification, including phosphorylation and

sumoylation. Small Ubiquitin-like Modifier or SUMO proteins are a family of small proteins that

are covalently attached to and detached from other proteins in cells to modify their function.

Sumoylation of PPARγ at K107 blocks its transcriptional activity. Dutchak et al. (2012) report

that FGF21 inhibits the sumoylation of PPARγ; thus, sustaining its own activity in a feed-

forward manner (Figure 2.3). PPARγ - induced FGF21 enhances adipocyte differentiation,

insulin sensitization, promotes proper lipid distribution and stimulates PPARγ transcriptional

activity.

Relative

 FGF21 

Protein Secretion 

6 h            12 h 

Human HepG2 

FGF21 

Albumin 

16

Figure 2.3. Model of the PPARγ - FGF21 regulatory pathway in WAT tissue [32].

FGF21 mRNA and protein levels were induced by rosiglitazone in WAT but not in liver

of wild type mice. Conversely, FGF21 mRNA and protein levels were increased by the PPARα

agonist in the liver, but not WAT. The PPARα agonist, GW7647, significantly increased plasma

FGF21 concentrations; however, rosiglitazone had no effect [32]. This suggests that FGF21 is

secreted from WAT but acts locally in an autocrine or paracrine fashion instead of entering the

circulation. Kharitonenkov et al. (2011) describe FGF21 as an endocrine factor secreted from the

liver; however, Dutchak et al. (2012) hypothesize that FGF21 is secreted from WAT but is

unable to enter the circulation due to the interactions with the WAT extracellular matrix. Thus,

FGF21 was postulated to act as an autocrine or paracrine factor, similar to conventional FGFs.

Johnson et al. (2009) also describes an autocrine role for FGF21 in exocrine pancreas and liver.

Acinar cell FGF21 mRNA expression markedly increased during pancreatitis in vitro. The

activation of FGF21 was an immediate cell autonomous result to pancreatic damage, leading to

an autocrine signal that reduced the expression of early response factor, Erg1 [64].

This led us to hypothesize that pancreatic islets express and secrete FGF21 to signal in an

autocrine manner. We further hypothesize that this signaling pathway may be diminished during

diabetes due to FGF21 resistance.

TZDs

Insulin sensitization, side effects 

 FGF21

SUMO‐PPARγ (inactive) 

PPARγ(active)

17

2.2 FGF21 Resistance

FGF21 improves β- cell function and survival by stimulating peripheral glucose and lipid

disposal [33]. However, a state of FGF21 resistance has been described in obese mouse models

and in obese humans [21, 34].

Fisher et al. report that the expected beneficial effects of endogenous FGF21 to increase

glucose tolerance and reduce circulating triglycerides are absent in obesity. Expression of

FGFR1 and KLB were significantly downregulated in WAT of obese mice (Figure 2.4), most

likely contributing to impaired FGF21 signaling [21].

Figure 2.4. FGF receptor and KLB mRNA expression in WAT of lean and diet- induced obese mice. Data for each gene are calculated using 36B4 as a reference gene [21].

FGF21 is secreted from the liver via PPARα in circadian fashion [28]. Figure 2.5

illustrates the FGF21 circadian mRNA levels for mice fed with a fasting diet containing PPARα

agonist, compared to mice fed with a high-fat ketogenic diet. Mice were maintained under light

cycle of 12 hrs lights on: 12 hrs lights off (lights on at ZT0) and fed ad libitum 5 days. The fold

change in FGF21 mRNA level is blunted in mice fed a high-fat diet. The peak to baseline ratio of

the mice fed a high-fat diet is reduced compared to the fasting mice. The baseline FGF21 levels

are also increased significantly in the high-fat case.

   FGFr1    FGFr2    FGFr3    FGFr4       KLB

FGFr/36B

4 mRN

A 

18

Figure 2.5. Temporal mRNA expression profile of FGF21 in the liver. Mice are fed a fasting diet (left) and high-fat ketogenic diet (right). Open circles represent control mice, and closed circles represent treated mice. The maximal value for control mice is expressed as 100% [10, 70].

Yu et al. (2011) recently published results showing circadian rhythm of FGF21 in 36 lean

and obese volunteers. FGF21 levels increase during periods of fasting (overnight) and are

reduced during feeding (daytime). The magnitude of the nocturnal rise was blunted in obese

individuals (peak to baseline ratios were 2 vs. 4 in lean individuals). The baseline FGF21

concentrations were much higher in obese than lean people.

Coskun et al. (2008) detected a prominent effect of exogenous FGF21 treatment on

endogenous FGF21 mRNA expression in liver, adipose and pancreatic tissue of DIO mice.

FGF21 transcript levels were substantially lowered in each of these tissues after FGF21

administration. These studies demonstrate that FGF21 and FGFR expression levels, and

circadian rhythm of FGF21 are diminished during obesity suggesting an FGF21- resistant state.

Fisher et al. (2010) measured ERK1/2 phosphorylation as a reporter of FGF signaling. In

obese mice, they found that there was a severe impairment in the ability of FGF21 to activate

signaling pathways through induction of phospho- ERK1/2 in liver and WAT. Therefore, we

hypothesized that FGF21- resistance may occur in pancreatic islets and interfere with FGF21

expression, secretion and autocrine signaling. Despite the evidence for FGF21- resistance, Hale

et al. (2012) report that the phospho- ERK1/2 response was only partly attenuated in liver and

WAT of obese mice. This area remains controversial and further studies are required to elucidate

FGF21 resistance.

0 6 12 18 24 0

2000

4000

6000 mRN

A level (%) 

mRN

A level (%) 

19

2.3 Microfluidic Devices for Pancreatic Tissue Culture

Microfluidic devices are systems that process small (10-4 to 10-3 liters) amounts of fluids,

using channels with dimensions of tens to hundreds of micrometers [35]. These low cost systems

offer a variety of useful features including the ability to use very small quantities of reagents,

short time for analysis and small footprints for the analytical devices. The effects that become

dominant in microfluidics include laminar flow, diffusion, fluidic resistance and surface area to

volume ratio [36]. The laminar flow through microfluidics is a result of low Reynolds number by

design (< 2,300) and enables the controlled application of shear stress and the delivery of

multiple laminar streams in the absence of mixing. When two fluids come together in a

microchannel, they do not mix convectively; instead they flow in parallel, without eddies or

turbulence. The only mixing that occurs is the result of the diffusion of molecules across the

interface of the fluids [35]. We have also shown that microfluidic devices used for islet culture

and analysis provide benefits such as enhanced nutrient delivery to the center of the tissue and

control of flow conditions. Microfluidic devices are also used for periodic effluent collection

(such as insulin) from the cultured tissue [37]. Due to small volumes, the effluent becomes

concentrated in the exit stream.

The devices are commonly made of polydimethylsiloxane (PDMS), an organic silicon

based compound. PDMS is chosen for tissue culture due to its inert, non-toxic, and optically

clear properties. PDMS is also inherently permeable to gas, and therefore fully compatible with

living samples.

Johansson et al. (2006) showed that β- cell proliferation is closely associated with

vascular growth. Pancreatic islets are cell aggregates that largely consist of endocrine cells and

vascular endothelial cells (ECs) [38]. β- cells form the core of the islet however every β- cell is

adjacent to several β- cells as well as capillary ECs (Figure 2.6).

20

Figure 2.6. Simplified model of a pancreatic islet consisting of clustered pancreatic β- cells (green) and endothlial cells (red). Glucose enters the capillary blood vessels, and diffuses into the cells of the islet. Islet β- cells secrete insulin in response to glucose, where it travels through the blood vessels to the rest of the body [39].

Sankar et al. (2011) examined EC morphology in pancreatic islets cultured in a

microfluidic flow environment. Islets cultured in this device had preserved EC- density and -

connected length throughout 24 and 48 hours in culture. The devices were designed to mimic in

vivo hemodynamics using flow rates between 3 and 6 ml/24 hours. Microfluidic tissue culture

presents several advantages over conventional culture in a dish, such as improved EC-

morphology and enhanced nutrient delivery to the center of the islet.

β- cells respond to glucose through closure of the ATP- sensitive potassium channel

(KATP), membrane depolarization, Ca2+ influx through voltage gated calcium channels, and

subsequent insulin secretion [40]. Sankar et al. (2011) observed a mutated glucose- stimulated

Ca2+ response with the Flou4 AM dye in device- cultured islets. The peripheral β- cells of the

islet experienced flow- induced damage due to shear stress, preventing the detection of Ca2+

oscillations. We aim to design a microfluidic device for islet culture that reduces shear stress

around the islet, allowing detection of Ca2+ oscillations, while maintaining enhanced delivery of

media to the centre of the tissue and EC morphology. Furthermore, the device will be custom-

designed for islet secretion studies.

      Glucose 

      Glucose 

Insulin

Insulin 

Blood Vessel 

Pancreatic Islet

21

Chapter 3 – Research Rationale

3.1 Rationale

The glucolipotoxicity model of type II diabetes describes that β-cell dysfunction develops

when both glucose and circulating fatty acids are high. Elevated glucose and fatty acids cause

toxicity in islets, which may be associated with type II diabetes. The mechanism leading to β-cell

glucolipotoxicity is described by oxidative stress and lipid partitioning models [7, 8]. We aim to

determine the regulation of FGF21- signaling in β- cells during glucolipotoxicity.

FGF21 is produced by the liver through induction of PPARα. It is secreted into the

circulation by hepatocytes, due to its low affinity for heparin sulfate proteoglycans in the

extracellular matrix, where it stimulates glucose- uptake and promotes lipolysis [13]. FGF21 is

induced in WAT via PPARγ during the fed state and acts in a feed-forward autocrine loop to

sustain its own activity [32]. Johnson et al. (2009) show that pancreatic acinar cell FGF21

expression immediately increases in response to cerulein- induced pancreatitis in mice. Thus, it

is possible that pancreatic islets respond to glucolipotoxicity by generating FGF21. This action

could preserve β- cell survival and function in an autocrine manner.

Excess circulating nutrients appear to diminish the beneficial effects of FGF21, leading to

reduced circadian rhythms, phospho- ERK1/2 levels and FGF21 and FGF receptor expression

levels [21, 34]. Given that FGF resistance could affect mRNA expression and secretion

pathways, we examined whether excess nutrients alter FGF21 and FGF receptor expression

levels in pancreatic islets.

Pancreatic islets secrete low quantities of hormones (5.5 - 15 ng/ml of insulin collected in

a 500 μl volume over 40 minutes). Our goal was to use a microfluidic device for the long- term

culture and assay of FGF21 secretion from pancreatic islets. Previous devices presented several

problems, such as non- uniform flow around islets, challenges identifying individual islets and

exposing the islets to high shear stress [37]. We designed a device that provided enhanced flow

through the islets and reduced shear stress around the islet periphery.

22

3.2 Hypothesis and Objectives

FGF21 is a metabolic hormone that is controlled by the nutritional state and influences

glucose and lipid metabolism. In WAT, FGF21 is an inducible, fed- state autocrine factor that

functions in a feed- forward loop [32]. Despite the beneficial effects of FGF21, Fisher et al.

(2010) and Yu et al. (2011) describe a state of FGF21 resistance in obese mice and humans.

1. We hypothesized that fatty acids induce FGF21 transcript expression (mRNA) in β- cells

in a temporal manner.

Objectives

1. Determine if palmitate, oleate, linoleate, PPARα agonist (bezafibrate) or PPARγ

agonist (rosiglitazone) induce the expression of FGF21 in β- cells and islets

2. Compare FGF21 expression in mouse islets, liver, WAT, β- cells and α- cells

3. Examine the temporal induction of FGF21 in β- cells

4. Design a microfluidic device for pancreatic islet culture and effluent collection

2. We hypothesized that high glucose and high fat cause FGF21 resistance in pancreatic

islets by down-regulating the mRNA expression of FGF21, FGFR1c and KLB.

Objectives

1. Determine if elevated levels of FGF21, glucose or palmitate reduce the expression

levels of FGF21, FGF receptor and co-receptor in islets

23

Chapter 4 – Materials and Methods

4.1 Pancreatic Islet Isolation, Tissue Culture and Cell Culture

Ex vivo culturing of pancreatic mouse islets were performed for quantitative PCR

(qPCR), ELISA and microfluidic analysis. Animal procedures were approved by the Animal

Care Committee of the University Health Network, Toronto, Ontario, Canada in accordance with

the policies and guidelines of the Canadian Council on Animal Care (Animal Use Protocol

#1531). The C57BL6 mice strain is a common inbred strain widely used for models of human

disease due to its availability and robustness. Pancreatic islets were isolated from 8- to 12- week-

old C57BL6 male mice by using collagenase digestion (Roche) [41]. Islets were subsequently

cultured in the islet media (full RPMI 1640 medium supplemented with 11 mM glucose, 10%

FBS, and 5 U/ml penicillin-streptomycin).

Mouse insulinoma 6 (Min6) β- cells are a cell- line commonly used as a model for

pancreatic β- cells [42]. The next most abundant cell type in the islet is the glucagon- secreting

α- cells. The adenoma- derived α- cell line αTC1 is used for diabetes research due to its glucagon

biosynthesis and cytokine sensitivity [43]. Min6 β- cells and αTC1 cells were cultured in RPMI

1640 media supplemented with 5 mM glucose, 1% FBS and 5 U/ml penicillin-streptomycin.

Cells are treated with 100 ng/ml of recombinant human FGF21 (R&D Systems).

4.2 Fatty Acid Preparation

Palmitate is prepared by initially dissolving 0.128 g of palmitic acid (Sigma Aldrich) in

0.1 M NaOH at 70oC to make 100 mM solution. Next, the solution was diluted to 2 mM in 2%

fatty acid free BSA (BioShop), and incubated for 15 minutes at 60oC. Finally, the 2 mM solution

was diluted to 0.2 mM in cell culture or islet media. Oleate and linoleate were purchased as

BSA- conjugated stock solutions (Sigma Aldrich) and added directly to the cell culture media.

4.3 RNA Isolation, reverse transcription PCR and quantitative PCR

RNA isolation, reverse transcription (RT) PCR and quantitative (q) PCR were performed

for mouse islets, liver, WAT and α- and β- cell lines. Islets were hand- picked from RPMI 1640

culture media into an Eppendorf tube where they were centrifuged and washed with 1x PBS. A

small piece of liver and WAT was excised from the mouse, minced and homogenized. Cells

24

were grown to 70- 80% confluency and then washed with 1x PBS. The tissues and cells were

lysed and the nucleic acid was extracted using TRIzol reagent and chloroform. RNA was

extracted using the Qiagen RNA extraction kit, and the quantity and quality were determined by

measuring the absorbance at 260 and 280 nm. Reverse transcription to create the cDNA template

was carried out using a PCR thermal cycler (10 minutes at 25oC, 2 hours at 37oC followed by 5

minutes at 85oC). The forward and reverse primers used for PCR amplification of mouse FGF21

are 5’-AGATCAGGGAGGATGGAACA-3’ and 5’-TCAAAGTGAGGCGATCCATA-3’. PCR

amplification was performed with Taq DNA polymerase (Sigma Aldrich) with the following

reaction steps: 95oC for 1 minute, 95oC for 30 seconds, 51oC for 30 seconds, 72oC for 30

seconds, 34 cycles, 72oC for 10 minutes. The cDNA was run on a 1.7% agarose gel.

Quantitative PCR was set up in duplicate with primers specific to FGF21, FGFR1c,

FGFR2c and KLB (Applied Biosystems: FGF21 Mm00840165_g1, FGFR1c Mm00438930_m1,

FGFR2c Mm01269938_m1, KLB Mm00473122_m1). The mRNA expression level was

determined relative to actin (Applied Biosystems 4352933E) by measuring the cycle threshold

(CT) value. The CT represents the cycle at which there is the first detectable increase in cDNA

fluorescence. Reactions with a CT greater than 38 cycles were determined to be below the limit

of detection. CT values typically varied between 33-36 cycles required for DNA amplification

(out of a total of 40 cycles).

4.4 Immunoflourescence

Islets were stained with PECAM-1 (platelet endothelial cell adhesion molecule 1)

antibodies after 48- hour culture in a dish and in the short-pass device. Islets cultured in the dish

were loaded into a device for PECAM-1 staining. Islets were fixed with 2% paraformaldehyde

for 1 hour at 100 μl/h in the device and then blocked with 0.1% Triton X, 10% normal goat

serum in 1x PBS on ice for 4 hours at 25 μl/h. Islets were stained with primary antibody (1:150

dilution of rat anti- mouse CD31, BD Pharmigen) for at least 12 hours at 2 μl/h and then blocked

for 1 hour at 100 μl/h. Islets were stained with secondary antibody (1:500 dilution of Alexa Fluor

633 goat anti-rat IgG, Invitrogen) for 3 hours at 25 μl/h and then rinsed with 1x PBS for ½ hour

at 100 μl/h. Islets were then imaged using confocal microscopy using 633 nm excitation. Islet

endothelial cell (EC) length was analyzed using the Simple Neurite tracer macro for Fiji. The

total EC connected length was normalized to the islet periphery and plotted as EC length/

25

perimeter. Islet EC fractional area was detected by adjusting the threshold for 5- 10 z-slices

through each islet and measuring the average fractional intensity.

4.5 Microfluidics

4.5.1 Device Fabrication and Design

Microfluidic channels were designed using Adobe Illustrator CS4. Channels were drawn

to actual size using Adobe Illustrator in white color, and the background was kept in black color.

Designs were printed to acetate film. Device masters were fabricated at the University of

Toronto Microfluidics Foundry. Corning glass slides were used as the master substrate. The

slides were washed sequentially using isopropanol, acetone, and isopropanol and then

immediately dried using an air gun. The cleaned glass slides were placed on a 65oC hot plate for

15 minutes to dry. SU8 photo-resist was used to coat the glass substrate slides. The 125 μm layer

was spun using SU8-2100 at settings, step 1 at 500 rpm, ramp 5 seconds, dwell 10 seconds, and

step 2 at 2,500 rpm, ramp 5 seconds, dwell 30 seconds. The subsequent SU8-2100 coat was soft

baked at 65oC for 5 minutes, 95oC for 30 minutes, 65oC for 5 minutes, and cooled at room

temperature for 10 minutes. The 125 μm features were added by applying the corresponding

master film negative, with channel features exposed to UV for 16 seconds with 13 mW of power.

The post-UV exposure bake was done at 65oC for 5 minutes, 95oC for 12 minutes, 65oC for 5

minutes, and cooled at room temperature for 30 minutes.

The UV exposed channel features of the SU8 coating were thermally cross linked, and

rendered insoluble to the MicroChem SU8 liquid developers. The masters were finally developed

by soaking in the SU8 developer solution in glass trays agitated on an orbital shaker for 15

minutes. The masters were left to dry at room temperature overnight.

Microfluidic devices were fabricated using elastomer polydimethylsiloxane (PDMS)

(Dow Corning Sylgard) [44]. The PDMS silicone elastomer base and the curing reagent were

poured into P100 dishes at a 10:1 ratio (35 g to 3.5 g), mixed in a fume hood, and vacuum

desiccated for removal of bubbles. The mixtures were subsequently poured over top of the

master slides containing positive moulds of channels. The dishes were desiccated again to

remove bubbles and baked at 80oC for 3 hours in a vacuum oven for curing. The PDMS devices

were peeled off from the master moulds gently, and cut into appropriate sizes for bonding onto

26

No. 1 thickness 24 × 50-mm coverslips (VWR Scientific). Prior to bonding, cover slips were

cleaned using methanol, and dried using β wipes. PDMS devices were first hole-punched at the

ends of channels. The hole-punched devices were placed channel side up on a glass slide, and

clear tape was used to clean the channel side. The cleaned PDMS device and dried cover slip

were simultaneously oxygen plasma treated (Harrick Plasma Cleaner) at high power for 1

minute. The irreversible bonding was performed immediately following oxygen plasma

treatment, and the bonded device was left on an 80oC hotplate for 30 minutes. Circular wells

were bonded as a second layer to the PDMS device to allow a reservoir for media during image

analysis. Finally, tygon tubing was inserted into the channel ends at punch holes.

The direct- flow device is designed with main channel dimensions: 300 μm in width, and

125 μm in height. The width expands to 1800 μm in the islet-holding area. The average islet

diameter is 150 μm; thus, the holding area was designed to accommodate this size. The bypass

device main channel dimensions are 300 μm in width and 125 μm in height. The islet holding

area is 300 width x 300 μm length to accommodate only 1 islet. The diameter of the nozzle is 50

μm. Two prototypes are considered with different bypass lengths: short-pass (0.5 cm) and long-

pass (1 cm). The islets were loaded into the device using gravity flow and imaged using confocal

microscopy.

4.5.2 Microfluidic COMSOL modeling

The microfluidic device was drawn in AutoCAD, and imported into COMSOL

Multiphysics as a DXF file. The 2- dimensional device was extruded to 3- dimensions by

defining the channel height. The islet was positioned in the holding area and extruded as a 3-

dimensional sphere. Laminar flow of water was chosen as the fluid type, as an approximation of

the islet media. The wall boundaries, inlet and outlet were defined and the finite element mesh

was set to free tetrahedral. The velocity and pressure profiles were created in the results section.

We modeled the device in COMSOL assuming that the nozzle is fully blocked, in order to

provide a visual representation of the system. This assumption is based on the observation that

the flow rate through the islet is significantly lower than the flow rate through the main channel

(Table 6.2.2).

27

4.5.3 Microfluidic device long- term islet culture

Device-treated islets were loaded into the microfluidic devices using gravity flow shortly

after isolation and incubated in a custom-built desk-top incubator [37]. Briefly, the microfluidic

device was submerged in a stirred water bath at 37oC with flow driven by a syringe pump. The

reservoir media was also submerged in a separate water-bath maintained just above 37oC to

reduce formation of air bubbles in the microfluidic channel. A cap of mineral oil was placed on

top of the media in the reservoir to reduce evaporation and drift in pH. A bubble- trap was also

used to minimize bubble formation in the channels. To create the bubble trap, we hole-punched a

second hole in the inlet channel, and inserted Tygon tubing. During islet culturing, the bubble-

trap tubing was raised above the device. The bubbles preferentially entered the bubble- trap tube

and allowed media to continually flow through the main channel. Islets were cultured in flow for

48 hours. Post- culture the islets were stained with Flou4 for Ca2+ oscillations and PECAM-1 for

endothelial cell detection.

4.5.4 Microfluidic device image application

The PDMS microfluidic devices were used in all imaging experiments to hold islets

stationary and allow subsequent controlled flow of stimulation media. Prior to imaging, islets

were loaded into the microfluidic device, subsequently the device was mounted into the 37oC

temperature controller chamber and taped in place. The inlet and outlet tubing was fed through

the exit holes on the side of the chamber, and the chamber was mounted onto the microscope

stage over the 20x objective or the 40x oil immersion objective. The channel was brought into

focus using the transmitted light. Prior to loading islets, the channel was washed using 2 mM

glucose BMHH media (BMHH imaging media: 125 mM NaCl, 5.7 mM KCl, 2.5 mM CaCl2, 1.2

mM MgCl2, 10 mM HEPES, pH 7.4) from the outlet towards the inlet using a 22 mm gauge

blunt end needle- tipped syringe.

Air bubbles were removed by applying pressure while washing the channel with BMHH.

After three washes, the presence of bubbles in the channel was checked by visual examination

through the eyepiece. Using the inlet tubing, the islets were loaded into the microfluidic device.

It was important not to overload the channel with islets, or to allow gravity flow to continue for

an extended period of time as loaded islets could be pushed underneath the drop wall or against

the channel wall. The outlet tube was connected to a syringe on the digital syringe pump. The

28

inlet tube was clamped to allow media to flow through the well (Figure 4.1). Loaded islets were

subjected to a 5 minute reset flow to allow reorientation in the channel.

Outlet 

Inlet 

Islet holding area 

Well

Figure 4.1. Representative microfluidic device used for image analysis. The device was placed over the 20x or 40x objective lens and then washed with BMHH media (stained red in this image), and bubbles were removed from the channels by applying pressure through the inlet. Islets were flown in through the inlet tube by gravity flow. When the islets reached the holding area, the outlet tube was connected to a syringe pump and the inlet tube was clamped. The pump pulled media through the well at a constant rate.

4.6 NAD(P)H Imaging

Islets were loaded into a microfluidic device and stimulated with increasing glucose

concentration while measuring changes in the levels of NAD(P)H. Changes in NAD(P)H levels

were determined using live cell 2-photon microscopy. Endogenous NADH and NADPH are both

autofluorescent at 705 nm, with an emission spectra of 380- 550 nm, giving off an aggregate

signal of NAD(P)H. Stimulating the islets with a higher glucose concentration (20 mM) caused a

rise in NAD(P)H signal with changes in intensity directly proportional to the level of glucose

metabolism. The fold NAD(P)H response was expressed relative to the 2 mM baseline.

NAD(P)H images are collected for islets treated with increasing glucose concentrations.

Syringe Pump Outlet

Well Inlet

Islet holding area

29

ImageJ was used to quantify the mean intensity of NAD(P)H images by selecting and

measuring the intensity of 20 small circular regions of interest on each islet. Regions were

selected at random while avoiding nuclei and saturated pixels. Five regions were also measured

at a distance from each islet to obtain average background intensity. The mean NAD(P)H

intensity of each islet was calculated by averaging the 20 intra-islet measurements and

subtracting the average background.

4.7 Spherotech Velocity Profiles and Flou4 Ca2+ Oscillations

Islets were loaded into the short- pass microfluidic device, and Spherotech beads

(carboxyl fluorescent pink, 5.3 μm in diameter, 13x dilution, Spherotech Inc) were introduced

into the device at a flow rate of 9.6 ml/ 24 hours. The beads were imaged using confocal

microscopy with the 543 nm excitation wavelength. Time- series imaging was used to capture

the flow of beads around the islet. The distance of the streaks was measured using Zeiss imaging

software, and divided by the scan time in order to calculate the velocity.

Islets were cultured in RPMI 1640 media (11 mM glucose, 10% FBS) in a dish (control)

or short- pass device for 48 hours. Post- culture, islets were stained with 4 μM Fluo4 AM Ca2+

dye (Invitrogen) for 2 hours in either the dish or short- pass device. For the β- cyclodextrin Flou4

experiments, the islets were stained with 4 µM Fluo4 for 2 hours on the same day as isolation in

the dish or short- pass device. The 4 µM Fluo4 was dissolved with 90 mM 2-hydoxypropyl β –

cyclodextrin (Sigma Aldrich). Islets were incubated with 10 mM glucose for 15 minutes and then

imaged using the 488 nm excitation wavelength. The penetration depth was analyzed by

measuring the intensity profile across the islet periphery, and excluding the data that were less

than 10% of the maximum intensity. This allowed us to uniformly assess the penetration depth in

different islets.

The islet Ca2+ oscillations were analyzed with ImageJ (Figure 4.2 A). The z-axis intensity

profile was plotted through the image time-series. The periods of oscillations were determined

with Matlab fast fourier transform (FFT). The FFT estimates the frequencies in the oscillation

data from a discrete set of values sampled at a fixed rate. The sample frequency was ~1 Hertz (1

sample/ second). A plot of power spectral density versus frequency (periodogram) was generated

(Figure 4.2 B). The FFT generates a symmetrical graph; where any frequency on the left side

was mirrored by the frequency on the right side. The period was then calculated as the inverse of

30

the frequency. One representative region was chosen in a single islet to obtain an estimate of the

islet Ca2+ frequencies.

Figure 4.2. Oscillations and power spectral density plot of glucose- stimulated Ca2+ responses in pancreatic islets. (A) Pancreatic islets were stained with 4 µM Fluo4 dye for 2 hours, and introduced into a microfluidic device for imaging. Islets were treated with 10 mM glucose for 15 minutes and the intensity changes were recorded using time- series imaging over 93 seconds. (B) Fast fourier transform plot of the data in (A) generated in Matlab. One frequency peak was detected at 0.07813 s-1. This corresponds to a period of 12.8 seconds.

4.8 Statistical Analysis

For experiments with two treatments, one-way and two- way student’s t-test were

performed to determine significance. Data with p values < 0.05 were considered significant.

For experiments with multiple treatments, a one-way ANOVA followed by the Tukey

multiple comparisons test was used to assess statistical significance. Parings with p values <

0.05 were accepted as statistically significant.

0        16      31      47       62      78       93 

A B

Mean Intensity 

Power Spe

ctral D

ensity 

Sampling Time (s) Frequency (Hz)0           0.2         0.4          0.6         0.8          1.0

55 

65 

75 

85 

200

400

600

31

Chapter 5 – FGF21 Secretion and Resistance

5.1 Introduction

FGF21 is secreted from hepatocytes due to induction of PPARα and FXR whereas

secretion from adipocytes results from activation of the PPARγ [29, 30]. Circulating free fatty

acids, such as palmitate, oleate and linoleate, are natural agonists to PPARα and -γ in hepatic and

colon tissue, respectively [29, 45]. Dutchak et al. (2012) also demonstrated that FGF21 acts in a

feed- forward loop in adipocytes to sustain the activity of PPARγ.

We examined whether islets express FGF21 in response to excess nutrients and if this

expression is temporally regulated. Mai et al. (2009) indicate that the relative FGF21 mRNA

expression levels in liver cells treated with fatty acids increases over a period of 8 hours. This

corresponded with increased FGF21 protein levels. Concurrently, Badman et al. (2007) reported

that isolated hepatocytes treated with fatty acids show significant FGF21- induction after 24

hours. Thus, we determined the palmitate- induced FGF21 mRNA expression after 4, 8, 16 and

24 hour treatment to find the maximal induction.

The elevated circulating levels of FGF21 in obese subjects may cause desensitization or

resistance of this hormone, thereby diminishing its metabolic and autocrine signaling effects

[34]. FGF21 causes a change in expression level of itself and its receptors in WAT and liver in

obese models [21, 46]. To address FGF21 resistance, we cultured islets with excess nutrients and

measured the change in FGF21, FGF receptor and co- receptor mRNA expression levels.

In the future, we aim to measure the temporal secretion of FGF21 from islets. Sankar et

al. (2011) demonstrated effluent collection from islets using a microfluidic device. Thus, we

custom- designed a microfluidic device for our application. This new device has the ability to

hold islets stationary in individual traps and reduce shear stress around the periphery of the islets.

The device was designed to concentrate the effluent from islets after long- term culture.

32

5.2 Results

5.2.1 Verifying FGF21 Responsiveness

Initially, the islet responsiveness to FGF21 was assessed by creating a dose response

curve. Typical FGF21 plasma concentrations vary in different studies, as seen in Table 5.1 [20,

47, 48, 49, 50, 51], thus we aimed to establish an effective concentration of FGF21. Islets were

treated with increasing levels of FGF21, and the autofluorescent NAD(P)H levels were

measured. The NAD(P)H response provides a readout of the metabolic state of the cell,

indicating the level of the TCA cycle activity. These results indicate that islets remain responsive

to FGF21 as the concentration increases (Figure 5.2.1). The NAD(P)H response is reduced at a

concentration of 100 ng/ml, therefore this concentration was chosen for future experiments.

Table 5.1 – In vivo and in vitro concentrations of FGF21 FGF21 conc (ng/ml) In vivo Mice [48, 50] 0.2- 13 Human [51] 0.1- 5 In vitro β- cells [20] 1000 Adipocytes [47] 4000 BaF3 cells [49] 400- 800

Figure. 5.2.1. Mitochondrial NAD(P)H dose response for islets treated with FGF21. Pancreatic islets were treated with 0, 1, 10, 100 and 1000 ng/ml of recombinant human FGF21 in standard RPMI 1640 media supplemented with 11 mM glucose and 10% FBS for 48 hours. Post- culture, islets were stimulated with 20 mM glucose and the NAD(P)H responses were recorded. The fold NAD(P)H was expressed relative to the 2 mM glucose baseline.

FGF21 concentration (ng/ml)0         1       10      100    1000

1.3 

1.4 

1.5 

1.6 

Fold NAD(P)H 

33

5.2.2. FGF21 expression is higher in mouse islets compared to liver and WAT

FGF21 transcript is predominately found in the pancreas and hepatic tissues of mice

(Figure 1.6) [24]. To determine FGF21 transcript expression in endocrine pancreas, we

compared the expression level of FGF21 by reverse transcriptase PCR (RT-PCR) in islets

relative to liver, WAT, Min6 β- cells and αTC1 cells. We observed more defined bands in islets

and liver compared to WAT, α- and Min6 β- cells (Figure 5.2.2 A). We further examined

transcript expression by quantitative PCR (qPCR). The qPCR data in Figure 5.2.2 B shows that

the highest expression of FGF21 occurs in islets. This result corresponds with literature results

which show elevated FGF21 expression levels in the pancreas [24].

Islets1      2     3     4 

Liver1      2     3    

WAT1      2     3    

Gapdh

αTC1 cells1       2     

Min6 β‐cells1      2

FGF21

Figure. 5.2.2. FGF21 expression is higher in mouse islets compared to liver and WAT. (A) RNA was isolated from pancreatic islets, liver, WAT of C57BL6 mice and αTC1 and Min6 β- cells and reverse transcribed. The RT-PCR reaction was set up with FGF21 and Gapdh primers. The numbers above each gel indicate the samples assayed. (B) qPCR data representing expression levels of FGF21 relative to actin for 3 independently assayed mice. Exception: WAT qPCR was performed with only 2 mice (One-way ANOVA *P

34

5.2.3 Palmitate, Bezafibrate and Rosiglitazone induce FGF21 expression in Min6 β- cells and in islets

Elevated circulating levels of FGF21 are present in models of obese mice [21]. We

therefore investigated whether the saturated fatty acid (palmitate) induces the expression of

FGF21 in Min6 β- cells and αTC1 cells. As seen in Figure 5.2.3. A, palmitate appears to increase

FGF21 transcript expression in α- and β- cells after 16 hours of treatment.

Subsequently, we determined if palmitate, unsaturated fatty acids (oleate or linoleate),

and PPARα/γ agonists induced FGF21 expression in the Min6 β- cell line (Figure 5.2.3 B).

Saturated fatty acids such as palmitate were previously shown to cause lipoapoptosis in human β-

cells; however, unsaturated fatty acids (oleate and linoleate) actually prevent stearate- induced

lipoapoptosis [52, 53]. FGF21 is produced in the liver via PPARα, and in WAT through PPARγ

induction [10]; thus, we included bezafibrate (PPARα agonist) and rosiglitazone (PPARγ

agonist) as positive controls.

We observed a quantitative induction of FGF21 transcript in Min6 β- cells stimulated

with palmitate, bezafibrate and rosiglitazone for 24 hours (Figure 5.2.3. B). However, no

induction was observed with oleate or linoleate. We similarly observed a more moderate

increase in pancreatic islets in response to palmitate and robust responses to both PPAR agonists

(Figure 5.2.3 C). Our data suggest that α- and β- endocrine cells respond to saturated fatty acids

by producing FGF21 in a PPAR- dependent manner. We postulate that elevated levels of FGF21

could then act in a protective manner to diminish glucolipotoxicity.

35

Figure 5.2.3. Palmitate induces FGF21 expression in β- cells, α- cells and islets. (A) Min6 β- cells and αTC1 cells were treated with 0.2 mM palmitate for 16 hours in RPMI 1640 media containing 5 mM glucose and 1% FBS. RNA isolation was performed and the RNA was reverse transcribed. The cDNA PCR reaction was set up with FGF21 and Gapdh primers. The numbers above each gel indicate the samples assayed. (B & C) qPCR data for Min6 β- cells and pancreatic islets. Min6 β- cells and islets were cultured for 24 hours in RPMI 1640 media with 5 mM glucose, 1% FBS and 0.2 mM palmitate, oleate or linoleate, 250 μM bezafibrate or 5 μM rosiglitazone. The qPCR data represents expression levels relative to actin for 3 independently assayed and treated β- cell samples and mice (One-way ANOVA *P

36

5.2.4. Palmitate- induces FGF21 expression in a temporal manner in Min6 β- cells

To determine the temporal effect of saturated fatty acid- induction of FGF21, we treated

Min6 β- cells for 4, 8, 16 and 24 hours with 0.2 mM palmitate. The data indicate that maximum

induction of FGF21 occurs after 16 hours of treatment, followed by decreased expression after

24 hours of treatment (Figure 5.2.4). FGF21 expression may be attenuated after 16 hours due to

translocation of STAT (signal transducer and activator of transcription) to the nucleus to affect

transcription [13].

Figure. 5.2.4. Palmitate induces FGF21- expression in a temporal manner. Min6 β- cells were cultured for 4, 8, 16 and 24 hours in RPMI 1640 media containing 5 mM glucose and 1% FBS with 0.2 mM palmitate, 250 μM bezafibrate or 5 μM rosiglitazone. The qPCR data represents expression levels relative to actin for 3 independently assayed β- cell samples (One way ANOVA *P

37

5.2.5 Elevated levels of FGF21, glucose and palmitate do not change FGF21, FGF receptor and co-receptor mRNA expression levels in islets

The ability of FGF21 to increase glucose tolerance and reduce circulating triglycerides is

absent in models of obesity, suggesting a state of FGF21 resistance [21]. FGF21 resistance can

occur at many different levels such as:

Enzymatic inactivation of FGF21

Alterations in the FGF21 receptor mRNA expression

Alterations in FGF21 protein trafficking to the cell membrane

Modifications to the FGF21 receptor

Alterations in the metabolic pathways down-stream of FGF21 receptors [54]

Fisher et al. (2010) and Coskun et al. (2008) report that FGF21 and FGF receptor

expression levels were significantly down- regulated in the WAT of obese mice. Hence, we

examined whether excess nutrients reduce receptor and co- receptor expression in pancreatic

islets. A reduction in FGF receptor expression levels would hypothetically reduce FGF21

autocrine signaling [21].

Pancreatic islets were cultured in the presence of elevated levels of FGF21, palmitate or

glucose. Concentrations were chosen to represent serum levels typically found in diabetic obese

mice [55, 56, 57] (100 ng/ml FGF21, 20 mM glucose and 0.2 mM palmitate). The results show

that elevated nutrients do not cause FGF21, FGF receptor or co- receptor down- regulation

(Figure 5.2.5), suggesting that FGF21 resistance does not occur in islets through reduced

receptor expression. Future studies are needed to confirm alternative mechanisms of resistance.

38

Figure. 5.2.5. FGF21, glucose and palmitate treatment do not change FGF21, FGF receptor and co-receptor expression levels. (A) Islets were cultured for 24 hours in RPMI 1640 media containing 11 mM glucose and 10% FBS the absence (control) and presence of 100 ng/ml FGF21. (B & C) Islets were cultured for 24 hours in RPMI 1640 media containing 10% FBS with 5, 11 and 20 mM glucose. (D & E) Islets were cultured for 24 hours in RPMI media containing 5 mM glucose and 10% FBS in the absence (control) and presence of 0.2 mM palmitate. The qPCR data represents expression levels relative to actin for the islets from 3 independently assayed and treated mice. Exception: (B & C) results were taken from 2 independently assayed mice.

A

B C D E

        R

elative mRN

A expression 

        R

elative mRN

A expression 

FGF21    FGFR1c   FGFR2c       KLB

ControlFGF21

0 

1 

2 

FGFR1c    KLB   FGFR1c    KLB   

0 

 1 

2 

0 

 1 

2 

0

1

2

0 

 1 

2 

Glucose (mM)  Glucose (mM) Palmitate (mM)  Palmitate (mM) 5      11      20  5      11      20 0      0.2  0      0.2

39

5.3 Discussion

We report that FGF21 mRNA expression levels are higher in islets compared to liver and

WAT of lean C57BL6 mice (Figure 5.2.2). These data imply that pancreatic islets could play an

important role in whole body FGF21 production and/or act as an autocrine factor.

Elevated circulating levels of fat are accompanied by excess FGF21 in the serum [21,

34]. This led us to examine the effect of fatty- acid induction of FGF21 expression. Our results

show that saturated fatty acids (palmitate), but not unsaturated fatty acids (oleate or linoleate),

induce the expression of FGF21 in Min6 β- cells and in islets (Figure 5.2.3). This fits with our

hypothesis that elevated levels of fat induce FGF21 expression and secretion in islets. Palmitate

increases the production of ROS in rat islets [8], thus FGF21 may lower oxidative and ER stress

by reducing the accumulation of fatty acid intermediates. The maximum induction of FGF21

occurs after 16 hours of palmitate treatment (Figure 5.2.4), followed by a decli