Citation
Glucokinase regulation of multicellular islet electrical activity

Material Information

Title:
Glucokinase regulation of multicellular islet electrical activity
Creator:
Ludin, Nurin ( author )
Place of Publication:
Denver, Colo.
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file (90 pages) : ;

Subjects

Subjects / Keywords:
Islands of Langerhans ( lcsh )
Diabetes ( lcsh )
Pancreatic beta cells ( lcsh )
Diabetes ( fast )
Islands of Langerhans ( fast )
Pancreatic beta cells ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references.
System Details:
System requirements: Adobe Reader.
Statement of Responsibility:
by Nurin Ludin.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Embargo Date:
8/24/2018
Resource Identifier:
on10121 ( NOTIS )
1012118815 ( OCLC )
on1012118815
Classification:
LD1193.E56 2017m L94 ( lcc )

Downloads

This item has the following downloads:


Full Text
GLUCOKINASE REGULATION OF MUTLICELLULAR
ISLET ELECTRICAL ACTIVITY by
NURIN LUDIN B.S. University of Denver, 2014
A thesis submitted to the faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Science Bioengineering
2017


2017
NURIN LUDIN
ALL RIGHTS RESERVED


This thesis for the Master of Science degree by Nurin Ludin
has been approved for the Bioengineering program by
Richard Benninger, Chair Emily Gibson Vitaly Kheyfets
Date: December 15, 2017


Ludin, Nurin
Glucokinase Regulation of Multicellular Islet Electrical Activity Thesis directed by Associate Professor Richard Benninger
ABSTRACT
Diabetes is caused by dysfunction or death of beta (P) cells in the islet of Langerhans, multicellular micro-organs in the pancreas that are important for glucose regulation. This results in poorly regulated glucose homeostasis due to defective insulin secretion, p cells do not act autonomously, they are electrically coupled together by gap junctions. As a result all the cells in the electrically continuous network are quiescent in the absence of glucose and secrete insulin in response to elevated glucose. Previously, others have experimentally shown and simulated a subpopulation of dysfunctional P cells rendering them inexcitable by introducing an inducible KAtp mutation modeling neonatal diabetes mellitus. It has been shown that in the presence of coupling if the number of inexcitable cells in the network exceeds 15% there is an emergence of critical behavior where all the cells in the islet transition rapidly from being globally active to inactive and there occurs a nearly complete suppression of insulin secretion. In the absence of coupling it was shown that the critical behavior was diminished, glucose homeostasis was rescued, and electrical activity in the islet withstood despite being subjected to silencing conditions. This suggested uncoupling the islet as a novel therapeutic option for treating neonatal diabetes mellitus and potentially other monogenic forms of diabetes. Glucokinase is a more upstream regulator of electrical activity and mutations to glucokinase results in diabetes. We explore emergent critical behavior further by generating an inducible
IV


mutation to glucokinase. We will examine how mutations to glucokinase decrease average excitability and have a disproportionately suppressive effect when coupled. Our simulations predict that decreases in glucokinase activity results in an emergence of critical behavior and the critical behavior is alleviated when gap junction coupling is removed. The experimental findings conclude that there is potential to recover islet function despite severe glucokinase mutations and inhibition. The results of this investigation contribute to the characterization of furthering our understanding of emerging critical behavior in cellular networks such as the islet in the hopes of discovering more patient specified and effective treatment methods for glucokinase dysfunction mediated neonatal diabetes mellitus and maturity onset diabetes of the young type 2.
The form and content of this abstract are approved. I recommend its publication.
Approved: Richard Benninger
v


ACKNOWLEDGEMENTS
This work was possible with the help of a number of exceptional people. I thank Richard Benninger for his guidance and support during the completion of this work and committee members Emily Gibson and Vitaly Kheyfets for providing valuable feedback in finishing this work. Additionally, I would like to thank all the current and previous members of the Benninger Lab. Especially, Matt Westacott, Aleena Notary, Nikki Farnsworth, Audrey Heintz, and Jenn Dwulet for their assistance on both the computational and experimental aspects of my thesis. Finally, I would like to thank my family and friends for their support over the course of the program. A special thanks to my parents Farah and Rafaat Ludin without whom this would not be possible.
Thank you to the Barbra Davis Center islet isolation core, Advanced Light Microscopy Core at the University of Colorado Denver, and University of Colorado at Boulder for access to the JANUS supercomputer cluster.
Funding for this research was providing by the grant Emergent Multicellular Properties Regulating Pancreatic Islet Function (R01 DK106412-01). Mouse studies were performed in according to the guidelines of IACUC protocol: B 95814(07)1D.
VI


TABLE OF CONTENTS
CHAPTER
I. INTRODUCTION...........................................................1
Diabetes: Definitions and Motivations..................................1
Beta Cell Function in Islet of Langerhans..............................4
Glucokinase Mutations in Diabetes.....................................11
Gap Junction Coupling.................................................19
Mathematical Model of the Islet.......................................25
Specific Aim..........................................................26
II. MATERIALS AM) METHODS................................................28
Mouse Studies............................................................28
Coupled Oscillator Model Simulations..................................31
Data Analysis.........................................................35
III. RESULTS.............................................................42
GK Deletion Modeled, Shows Critical Behavior..........................42
GK(-/-) Mice Generation and Physiology................................44
NAD(P)H Imaging of Islets Isolated from GK(-/-) Mice
to Characterize Metabolic Activity....................................47
Calcium Imaging of Islets Isolated from GK(-/-) Mice
to Investigate Calcium Dynamics and Electrical Activity...............48
tdTomato Quantification Identifies Percent Mutation Higher Than Expected.49
vii


Decreased Rate of Glycolysis Modeled, Shows Critical Behavior........51
NAD(P)H Imaging of Cx36(+/+) and Cx36(-/-) Islets treated with GK
inhibitor to Characterize Metabolic Activity..........................53
Calcium Imaging of Cx36(+/+) and Cx36(-/-) Islets treated with GK
inhibitor to Characterize Calcium Dynamics and Electrical Activity...54
Clinical Mutations in Glucokinase Examined............................55
IV. DISCUSSION AND FUTURE WORK.............................................62
Global Behavior and GK Heterogeneity..................................63
Effect of Parameters Altering Glucokinase Sensitivities...............70
Calcium Dynamics of GK Mutation.......................................71
Implications for Diabetes.............................................73
REFERENCES...................................................................76
viii


CHAPTERI
INTRODUCTION AND BACKGROUND Diabetes: Definition and Motivation
Diabetes is a complex metabolic disorder characterized by higher than normal blood glucose levels or hyperglycemia resulting from defects in insulin secretion, insulin action, or a combination of both. A diagnosis of diabetes is made if the glycated hemoglobin (HbAlc) is 6.5 percent and higher, fasting plasma glucose level is 126 mg/dL and higher, or a random or two-hour plasma glucose level is 200 mg/dL and higher (Holt and Hanley 2013).
Diabetes is a global epidemic affecting over 400 million people worldwide and the number is only projected to grow, rising more rapidly in middle- and low-income countries (World Health Organization 2016, Holt and Hanley 2013). Modern medical care uses lifestyle and pharmaceutical interventions to prevent and control hyperglycemia, so that there is an adequate delivery of glucose to the tissues of the body and the tissues are not harmed by hyperglycemia (Fowler 2008). Hyperglycemia occurs as a result of defects in insulin secretion, insulin action, or a combination of both (Diabetes Care 2010, Diabetes Care 2009, Diabetes Care 2004). The complications associated with being in a state of chronic hyperglycemia include microvascular complications (such as retinopathy, peripheral neuropathy, and diabetic nephropathy) and macrovascular complications (such as coronary artery disease, peripheral arterial disease, and stroke).
1


The total estimated cost of diagnosed diabetes in 2012 was $254 billion consisting of $176 billion in medical costs and $69 billion in reduced national productivity in 2007 (Diabetes Care 2012). The significant components of medical expenditures include hospital inpatient care (43 percent of total medical cost), prescription medications to treat the complications associated with diabetes (18 percent), anti diabetic agents and supplies (12 percent), physician office visits (9 percent), and facility stays (8 percent) (Diabetes Care 2012). And of the cost categories, care for people with diagnosed diabetes accounts for more than 1 in 5 health care dollars in the United States (Diabetes Care 2012, Diabetes Care 2007). The estimated total economic cost of diagnosed diabetes in 2012 increased a total of 41 percent from the previous estimate made in 2007. Therefore the burden that diabetes imposes on society is substantial, but it is projected to grow even more rapidly (Diabetes Care 2012, Diabetes Care 2007). Furthermore, there are additional burdens that cannot be quantified and they include pain and suffering experienced by the patients, resources supplied by nonpaid caregivers, and the burdens associated with undiagnosed diabetes.
Type I and Type II Diabetes, Permanent Neonatal Diabetes Mellitus, and Maturity-Onset Diabetes of the Young
The vast majority of cases of diabetes fall into two broad categories identified as type I and type II diabetes. Typically type I diabetes accounts for between 5 to 10 percent of total patients with diabetes, whereas type II diabetes accounts for about 90 to 95 percent of total cases. Type I diabetes occurs as a result of absolute deficiency of insulin secretion due to autoimmune destruction of Beta-cell destruction and type II occurs as
2


result of resistance to insulin action and inadequate compensatory insulin secretion response (Diabetes Care 2007). Loss of Beta-cell number and function underlies much of the pathology of diabetes. In type I diabetes, the immune system recognizes the beta-cell as foreign, probably due to a combination of genetic and environmental factors. The insulin resistance that occurs in type II diabetes involves an inability of cells in for example liver, muscle, and adipose tissue to respond to the normal actions of insulin and in order to compensate for this resistance, the pancreatic Beta cells increase their production of insulin while declining in function and insulin production eventually becomes inadequate (Amedeo Vetere et al. 2014).
Approximately between 2 to 5 percent of cases of diabetes are associated with monogenetic defects in Beta-cell function that induce an onset of hyperglycemia at an early age are associated with impaired insulin secretion (Holt and Hanley 2013). The most common types of monogenic forms of diabetes include neonatal diabetes and maturity-onset diabetes of the young (National Institute of Diabetes and Digestive and Kidney Diseases 2014). Neonatal diabetes can either be transient where patients resolve in the first 18 months of life or permanent where they have a life-long insulin dependence to regulate blood glucose levels. Roughly 50 percent of NDM cases occur as a result of mutations to the KCNJ11 or ABCC8 genes which encode for the Kir6.2 and SUR1 subunits on the KAtp channel (Gloyn et al. 2006, Hattersley 2005). PNDM can also result from a complete deficiency of the P cell transcription factor insulin promoter factor 1 (IPF1 also known as PDX1) and the enzyme glucokinase (Matschinsky, F.M. and Magnuson, M.A. 2004, National Institute of Diabetes and Digestive and Kidney Diseases 2014). Mutations in six genes cause most of the MODY cases. This includes mutations to
3


genes encoding the enzyme glucokinase (MODY-2) and the transcription factors: hepatocyte nuclear factor 4-a (MODY-1), HNF-la (MODY-3), insulin promoter factor 1 (MODY-4), HNF-ip (MODY-5), and neurogenic differentiation factor 1 (MODY-6) (Matschinsky, F.M. and Magnuson 2004, National Institute of Diabetes and Digestive and Kidney Diseases 2014).
Beta Cell Function in the Islet of Langerhans
Physiology of the pancreas
The pancreas has two main functions: to produce enzymes for digestion (exocrine) and make hormones (endocrine). Most of the pancreatic volume consists of exocrine tissue (MacDonald and Rorsman 2006). The stomach empties partially digested food into the intestine and the exocrine component of the pancreas releases digestive enzymes such as tryspin and chymotrypsin into the contents to digest proteins, amalyse to digest the carbohydrates, and lipase to break down the fats (Columbia University Department of Surgery). The remaining tissue consists of endocrine cells identified as the islets of Langerhans, which are multi-functional micro-organs that are dispersed throughout the pancreatic exocrine tissue and are highly vascularized (MacDonald and Rorsman 2006).
4


Structure and Function of the Islets of Langerhans
The islets of Langerhans are multi-cellular micro-organs located in the pancreas that regulate blood glucose by the P cells secreting hormones such as insulin. The three main cell types of the pancreatic islet include alpha cells, delta cells, and beta cells. The islets of Langerhans make up only 2 percent of the pancreatic tissue and secrete the hormones insulin via beta cells (constitutes 70 percent of total islet cells), glucagon via alpha cells (constitutes 20 percent of total islet cells), and somatostatin via delta cells (constitutes less than 10 percent of total islet cells).
It has been recognized that the cytoarchitecture of pancreatic islets vary between different species. The organization of the different hormone-secreting cell types are distinctly different between human and mouse islets. However, despite the differences in overall body and pancreas size as well as total Beta-cell mass among the species, the distribution of their islet sizes closely overlap and they share common architectural features that may correlate to similar demands for insulin (Abraham Kim et al. 2009, Suckale and Solimena 2008).
Figure 1: The cytoarchitecture of the islet and the relative abundance of the various hormones secreted by the islet in rat models (top) and human models (below).
5


It is well recognized that there is a close correlation between body and pancreas weight. It has been shown that total Beta-cell mass increases proportionately in order to compensate for the demand for insulin in the body within a certain limit of islet size where the overall changes in structure reflects the adaptation of the islet in response to increased body demands rather than differences amongst species (Abraham Kim et al. 2009).
The islets of Langerhans function to secrete the following hormones: beta cells secrete insulin, alpha cells secrete glucagon, delta cells secrete somatostatin, PP cells secrete pancreatic polypeptide, and epsilon cells secrete ghrelin. Insulin and glucagon are critical to regulation of blood glucose. The two hormones have antagonistic actions in order to maintain homeostasis of the blood glucose concentration, where insulin mediates the cellular uptake of blood glucose into skeletal muscle and the live after a meal and glucagon mediates the hydrolysis of liver glycogen between meals so that the liver can secrete glucose into the blood (Ishihara et al. 2003, Matschinsky and Magnuson 2004). It has also been showed that while both alpha and beta cells possess the capacity to respond to nutrients, secretions from alpha-cells is normally suppressed by simultaneous activation of beta-cells (Ishihara et al. 2003). Therefore, insulin lowers blood glucose and glucagon raises blood glucose levels.
6


After a meal
Between meals
Figure 2: The antagonistic actions of insulin and glucagon in order to reestablish blood glucose homeostasis.
Insulin Action
Insulin acts on a target cell by binding to the insulin receptor on the cell surface. The insulin receptor is characterized as a heterotetramer with two a-glycoprotein and two P-glycoprotein subunits linked together by disulphide bonds. Insulin binds to the extracellular a subunits of the insulin receptor inducing in conformational change that allows ATP to bind to the intracellular P subunit of the receptor resulting in autophosphorylation of the P subunits by tyrosine kinase activity. This is followed by tyrosine phosphorylation of insulin responsive substrates, which then bind to other signaling molecules that mediate subsequent cellular actions of insulin (Holt and Hanley 2013).
Insulin is the hormone that directs anabolic processes in intermediate metabolism. Insulin action has major effects on glucose, lipid, and protein metabolism. The tissues that are most sensitive to insulin effects include the liver, skeletal muscle, and adipose tissue. After the secretion of insulin, 60 percent of the insulin content is removed by the
7


liver where it plays an important in regulating hepatic glucose output by inhibiting gluconeogenesis and promoting the storage of glucose-6-phosphate to glycogen. In muscle cells, insulin-mediate glucose uptake promotes glycogen storage and for carbohydrates to be used as the primary energy source for muscle contraction instead of fatty acids or amino acids. In adipose tissue, fat break down is inhibited and the synthesis of fat is promoted through the formation and storage of triglycerides (Holt and Hanley 2013).
Glucose Stimulated Insulin Secretion and Beta Cell Electrophysiology
P cells respond to changes in glucose levels by secreting insulin directly into the circulatory system where it travels to target cells in the body (MacDonald and Rorsman 2006, Suckale and Solimena 2008). The process of insulin secretion is more specifically known as glucose stimulated insulin secretion or GSIS (MacDonald and Rorsman 2006).
Following the ingestion of a meal, there is a rise in blood glucose concentrations and the P cells sense the increase in glucose. Glucose enters P cells through the GLUT2 glucose transporter. Glucose is then phosphorylated by glucokinase. The phosphorylated glucose is then consumed by glycolysis. The products of glycolysis then enter the tricarboxylic acid (TCA) cycle and generate additional ATP. The metabolic pathways result in a net increase in intracellular [ATP] to [ADP] ratio, which then closes the ATP-sensitive K+ (KAtp) channel resulting in membrane depolarization and subsequent
opening of the voltage-gated Ca channels that result in an increase in intracellular Ca
2_|_
concentration ([Ca ],) which then acts on the exocytotic machinery to stimulate fusion of
8


insulin containing granules with the plasma membrane for secretion into the bloodstream (Robert Kennedy et al. 2007).
Figure 3: The GSIS pathway starting with the uptake of glucose via the GLUT2 transporter leading to the fusion of the vesicles consisting of insulin with the plasma membrane.
It is well known that [Ca ]; oscillates in islets in response to increases in glucose (Robert Kennedy et al. 2007, MacDonald and Rorsman 2006). And removal of extracellular Ca prevents the firing of an action potential and stops the subsequent insulin secretion (Curry et al. 1968, MacDonald and Rorsman 2006). Furthermore, the metabolism of glucose is essential for insulin secretion and inhibition of mitochondrial metabolism halts insulin secretion (Aschcroft et al. 1980, MacDonald et al. 2005). The breakdown of glucose results in the generation of ATP and an increased [ATP] to [ADP] ratio is a critical link between mitochondrial metabolism and electrical conductivity (ultimately resulting in insulin secretion) because of its ability to close the Katp channel and depolarization of the cellular membrane.
9


Electrical activity of the plasma membrane is an important component of GSIS.
At basal glucose levels of around 5mM, the plasma membrane has a resting membrane potential of about -70mV. This potential is governed by the Nemst equation relating extracellular and intracellular ion concentrations as follows:
where R is the ideal gas constant, T is temperature in Kelvin, z is charge of ion, and F is Faradays constant. P cells are very sensitive to small ion concentration changes due to the fact that the thickness of the plasma membrane is characterized as 3.5nm resulting in a large electric field amounting to 20x106 V/m.
The secretion of insulin is dependent on electrical activity and calcium entry. P-cells have channels embedded in their membranes that allow for flow of ions such as Ca2+ and K+ across the membrane. Potassium (K+) contributes greatest to the P cells overall membrane potential. At basal glucose levels when the P cell is electrically quiescent, K+ ions flow outward through the ATP-sensitive K+ (Katp) channel hyperpolarizing the membrane and setting the membrane potential at -70mV (Holt and Hanley 2013). And because these ions are electrically charged, their flux across the membrane will induce the generation of action potentials or sharp changes in voltage (MacDonald and Rorsman 2006, Holt and Hanley 2013).
Glucose Metabolism
P cells are an example of a glucose sensing cell type existing in a complex glucose sensing network responsible for maintaining the tightly regulated glucose
10


homeostasis. Glucose sensitivity is an essential component of the homeostatic feedback loops which serve to maintain blood glucose concentrations at a safe and tolerable range.
Glucose is transported into the Beta-cell via glucose transporters, where it is phosphorylated to glucose-6-phosphate by the enzyme glucokinase (GK), which is the rate determining step of glycolysis and is considered as the glucose sensor in the pancreatic Beta-cell. The product of glycolysis, pyruvate, is then a substrate for the TCA cycle in the mitochondria. In addition, cytosolic NADH also enters the mitochondria where both cytosolic and mitochondrial sources of NADH stimulate the electron transport chain to pump H+ ions out of the mitochondrial matrix causing a hyperpolarization of the inner mitochondrial membrane (Duchen et al. 1993, Maechler et al. 1997, MacDonald 2005). The dissipation of the H+ gradient induces the generation of
ATP via ATP synthase activity. The hyperpolarization of the inner mitochondrial
2+
membrane stimulates the mitochondrial inner membrane potential depended Ca uniporter to increase mitochondrial Ca2+. The increase in mitochondrial Ca2+ mediates in ATP transport into the cytosol, thereby increasing cytosolic ATP concentrations and the [ATP] to [ADP] ratio (Moreno-Sanchez 1985). The process of mitochondrial oxidative metabolism has been estimated to produce about 98 percent of the total P cell generated ATP (Erecinska et al. 1992).
Glucokinase and Glucokinase Mutations in Diabetes
Glucokinase (GK) is the glucose sensor of the P cell. GK is the rate-limiting step in glucose metabolism, triggering shifts in the metabolic pathway for varying levels of glucose.
11


Glucokinase and Glucose Metabolism
The role of the hexokinase GK as a sensor component in the glucose homeostatic feedback loop has become widely accepted. The discovery of the hexokinases occurred over 40 years ago and this led to the recognition that the enzymes were crucial for glucose metabolism (Walker et al. 1964, Sols et al. 1964, Sharma et al. 1964). Additionally, it was recognized that the one of the enzymes from the hexokinase family, hexokinase IV or GK, had unique functional and structural features that made it functionally and structurally different from other members of the hexokinase family (Grossbard et al. 1966). The unique features included having a mass of half that of hexokinase types I-III, a lower affinity for glucose, and a lack of significant feedback inhibition by its end product glucose-6-phosphate or G6P (Pilkis et al. 1968, Pilkis 1968). First GK was thought to be expressed only in the liver where it was responsible for glucose uptake into the liver (Pilkis et al. 1968), but then it was found to be expressed in mouse pancreatic P cells where it plays a key role in GSIS (Matschinsky et al. 1968). Additionally, it was found that the intracellular concentration of glucose in p cells was approximately equivalent to the plasma glucose concentration and the glucose transport across the membrane occurred at a very high capacity suggesting that GK could serve as a pacemaker for glycolysis allowing it to control insulin secretion (Matschinsky et al. 1968). These GK enzymatic capacities were extended to man, when they were demonstrated in human islet tissue using quantitative histochemical methods (Matschinsky et al. 1968). At first, GK was considered a biochemical glucose sensor based solely on the kinetic capacities of the enzyme (Matschinsky et al. 1968). However,
12


the most compelling argument for the central role of GK in glucose homeostasis is provided by the discovery that mutations in GK cause glycemic disorders in humans. The first report of hyperglycemia as a result of mutations to the GK gene was published in 1992 and since then there has been nearly 200 mutants of this gene reported that causes hyperglycemia or hypoglycemia depending on the nature of the mutation (Grimsby et al. 2003).
GK as the glucose sensor of the P cell asserts that the enzyme with its kinetics and its strength in regulating glycolysis is the determining factor in glucose metabolism which is understood to dictate insulin secretion. The sensing of glucose requires the enzyme to change its conformation or function with the plasma glucose concentration in the physiological range of 4mM to lOmM. The GK glucose sensor is integrated with the threshold for glucose-stimulated insulin release with a threshold close to 5mM glucose (for both rodents and humans), further establishing glucose homeostasis. The threshold has been established by the ATP-dependent K channels and voltage dependent Ca channels in the plasma membrane of the P cells, as well as neuroendocrine regulators establishing intracellular concentrations of Ca and cAMP. In this way, the capacity of GK as a glucose sensor is coupled to complex multilevel signaling pathways involved in regulated insulin release when the P cells are exposed to glucose concentrations exceeding 5mM (Matschinsky 2002 and Davis et al. 1999).
The biochemical kinetics of GK enables it to serve as the glucose sensor of the P cells. The point at which GK is most sensitive to changes in the concentration of glucose is 4mmol/l. This occurs as a result of the dynamics between the enzymes S0.5 and its hill coefficient when binding to glucose. The enzymes S0.5, the concentration of glucose at
13


which the reaction is half of the maximum rate of the system, is about 8mmol/l. The enzymes glucose hill coefficient (nH), cooperativeness for binding glucose, is approximately 1.7. The cooperativity of the enzyme binding to glucose indicated by its hill coefficient (nH) of 1.7 contributes significantly to glucose phosphorylation in the physiological range (Matschinsky and Magnuson 2004). This generates an inflection point for the catalysis of glucose to be around 4mmol/l and this is within the range necessary for the catalytic flux to be generated by the concentration of glucose (Matschinsky and Magnuson 2004). Furthermore, the metabolic flux doesnt depend on the concentration of the enzymes other substrate MgATP. The affinity of its other substrate, MgATP, is in the range of 0.3mM to 0.4mM which is significantly below the intracellular concentration of MgATP at 2.5mM. And so, the metabolic flux depends solely on the concentration of glucose and amount of GK in the cell (Matschinsky and Magnuson 2004). Other factors that are important for GK to be considered as the glucose sensor of the P cell includes that the entry of glucose must not be rate limiting and the fact that there must be a lack of significant end product inhibition by G6P so that catalysis is not diminished as glucose concentration increases. This does not seem to be the case (Matschinsky and Magnuson 2004). Finally, the last necessary prerequisite for GK to serve as the glucose sensor for the cell is that the initial phosphorylation of glucose by GK to G6P must be coupled to the final secretory event indicated by the exocytosis of insulin granules. This requires that the change in extracellular glucose is delivered to the plasma membrane and the exocytotic machinery is able to respond to the stimulus via a GK-mediated mechanism. Our current understanding fully supports the validity of this (Matschinsky and Magnuson 2004).
14


Glucokinase Mutations in Diabetes
Due to the vital role of GK in glucose stimulated insulin secretion, even small changes in the expression or activity of GK can result in significant effects in insulin secretion. And so it follows, mutations to GK cause diabetes. Mutations alter the protein sequence by one of the following methods: changing one amino acid, transform sequence of a site of RNA splicing at exon or intron resulting in abnormal messenger RNAs, or creation of a premature termination codon by point mutation or by deletion/insertion of basepairs. While heterozygous mutations in GK result in MODY-2, double heterozygous mutations or homozygous mutations result in PNDM.
The clinical consequences of the mutations can be put into perspective when remembering the concept of GK as the glucose sensor. The enzymatic nature of GK states that the capacity of GK in P cells is limited by its maximal activity or enzyme turnover (Vmax) and the affinities for its substrate glucose (S0.5,gi) and ATP (Km. atp) which then in turn determine the threshold for glucose stimulated insulin secretion. In man and rodents, the physiological P cell threshold is held at 5mmol/l which the K and Ca ion channels define this threshold. In patients with GK-mediated MODY or MODY-2, where inactivating mutations in one allele results in decreased enzyme turnover or affinity of enzyme to glucose generating an increased threshold of 7mmol/l. The heterozygous mutations result in haploinsufficiency and elevated glucose levels. In patients with GK-mediated PNDM or PNDM2, where inactivating mutations in both alleles result in total inactivation of GK there is an infinitely higher reduction in enzyme turnover and/or
15


affinities. The homozygous mutations resulting in truncation of protein and abolish enzyme activity (Matschinsky and Magnuson 2004).
MODY-2
Maturity onset diabetes of the young (MODY) is a monogenic disorder characterized by familial hyperglycemia with autosomal dominant inheritance. The hyperglycemia occurs due to decrease glycolytic flux leading to insulin secretion defects and develops at any time in childhood, adolescence, or young adulthood. MODY is caused by mutations in six genes, each kind presenting a different kind of MODY identified as MODY-1 to MODY-6. Mutations to genes encoding the enzyme GK is associated with MODY-2. Heterozygous mutations in the GK gene are the common cause of MODY. The key clinical feature of patients with GK mutations is that they present with a lifelong mild and stable fasting hyperglycemia typically ranging from 5.5mmol/l to 8.0mmol/l and their glucose is regulated to this elevated concentration (Stride et al. 2002). Because of its mild clinical presentation and its stability, complications for patients with MODY-2 is unusual and treatment is rarely needed with the more severe cases requiring oral agents or insulin (Fajans et al. 2001, Pearson et al. 2001).
In patients with MODY-2, P cell dysfunction is characterized by an increase in the threshold of blood glucose required for glucose stimulated insulin release from 4-5mM to 6-7mM as a result of a right-shifted dose response curves relating insulin secretion rate with glucose levels. Furthermore, MODY-2 patients demonstrated a 60% reduction in insulin secretion for a particular glucose level (Byrne et al. 1994).
16


Figure 4: (a) Dose response curves indicating relationship between average glucose concentrations and insulin secretion rates for control subjects (dashed line) and patients with MODY-2 (solid line), (b) P cell responsiveness as a function of changes in plasma glucose concentrations.
While patients do demonstrate hyperglycemia, investigators have found that interestingly insulin levels in MODY-2 subjects are usually normal throughout the day. This is believed to be as a result of the secretory defect relating solely to the glucose blindness of the P cell due to the fact that the release of insulin in response to other secretagogues such as arginine is anywhere from being unaffected to moderately decreased (Pueyo et al. 1994). Additionally, there is the contribution of physiologic adaptation of the P cells limiting the effects of the defect in insulin secretion. Furthermore, in subjects with more severe GK mutations resulting in more severe decreases in enzymatic activity, the insulin secretion defect is also less severe suggesting that there must be compensatory mechanisms that recover the insulin secretion response of the P cell (Pueyo et al. 1994). Some studies suggest that the compensatory mechanism is the upregulation of the single wild-type GK-gene allele (Sreenan et al. 1998).
17


PNDM
Neonatal diabetes mellitus is another monogenic disorder defined as an insulin-requiring hyperglycemia, which presents within the first months of life. Neonatal diabetes can be either transient or permanent. Permanent neonatal diabetes mellitus (PNDM) represents about 50% of the cases and is most commonly characterized by requiring lifelong insulin therapy. PNDM can occur as a result from a complete deficiency in the glycolytic enzyme GK. Infants with GK-mediated PNDM have presented with low birth weight, significant hyperglycemia, nearly undetectable c-peptide levels, and requiring insulin treatment within few days of birth (Njolstad et al. 2001, Njolstad et al. 2003, Sarici et al. 2001, Shehadeh et al. 1996). Theoretically, these patients should respond to GK-independent insulin secretagogues such as arginine and sulfonylureas but studies have not been able to detect any response.
Mouse Models with Decreased GK Activity
To test the role of decreased GK activity, investigators developed mouse models with heterozygous and homozygous GK knockout expression profiles. Previously, it was found that 50% reduction in GK gene expression lead to impairment in glucose stimulated insulin secretion. In mice with global knockout of GK, it was found that the mice die shortly after birth as the knockout of a second gene results in significant complications as the second gene is necessary in early embryogenesis. This lead to the development of transgenic mice that express Cre recombinase under control of P cell specific promoters. Postic et al. developed a Cre-loxP gene targeting strategy allowing the deletion of GK to be restricted using a conditional (or loxed) GK allele generated then
18


interbred with animals that expressed Cre under the control of the insulin 2-promoter (Postic et al. 1999, Niswender et al. 1997). However, it was found that the animals that lack GK only in the pancreatic P cells exhibit a similar phenotype to animals with global knockouts of GK by containing severe hyperglycemia as a result of significant decreases in insulin production. Suggesting that down regulation of GK in a global and specific form both are lethal to the species (Grupe et al. 1995).
Gap Junction Coupling
It is well established that P cells in the islet do not act autonomously, they are coupled together by connexin36 (Cx36) gap junctions. In the islet, gap junctions composed of Cx36 generate electrical and metabolic coupling between cells which then regulate electrical activity and insulin secretion. The gap junctions coordinate the islet response so that in basal glucose levels there is a complete suppression of insulin secretion and upon increased glucose concentrations the gap junctions allow for coordinated release of insulin from all the P cells in the islet. Gap junctions are intercellular channels made up of two connexon hemi-channels that are made up of six connexin subunits. The gap junctions allow for the transfer and exchange of ions (such as Ca and K ), metabolites, and other small molecules between neighboring cells. Studies have increasingly established the role of Cx36 gap junctions in the islet and have begun to look at its implications in the development of diabetes. Investigators have shown decreased gap junction coupling contributing to disease development, further looking at the role of Cx36 in Type 1 and Type 2 diabetes (Farnsworth and Benninger 2014). Previously, other investigators have looked at the role of Cx36 gap junctions in
19


exacerbating the effects of mutations that cause diabetes specifically mutations in the KAtp channel leading to decreased ATP sensitivity and closure of the KAtp channel (Notary et al. 2016, Hraha et al. 2014, Nguyen et al, Ashcroft and Rorsman 1989).
Role of Cx36 Under High Glucose and Basal Glucose
Recall that in p cells, insulin secretion occurs as a result of a series of metabolic and electrical events in response to elevations in blood glucose concentrations. The glucose stimulated insulin secretion pathway begins with glucose taken up by the P cells and metabolized resulting in an increase in [ATP]/[ADP] which then promotes the closure of the KAtp channels and subsequent membrane depolarization and activation of the voltage gated Ca2+ channels and elevation of [Ca2], The Ca2+ then acts on the exocytotic machinery to simulate the fusion of the insulin containing granules with the plasma membrane. Glucose stimulated insulin secretion is biphasic with first phase characterized by a burst release and a second phase characterized by a pulsatile release (Ashcroft and Rorsman 1989). Under stimulatory levels of glucose, electrical coupling provided by Cx36 gap junctions mediates in the transfer of a depolarizing current and the synchronization of KATp channel mediated membrane depolarization resulting in coordinated [Ca ], oscillations across the islet. Due to the coordination in coordinated [Ca ]; oscillations, there are coordinated insulin secretion oscillations (Benninger et al. 2008, Ravier et al. 2005, Calabrese et al. 2003). In islets isolated from Cx36 deficient mice, there is a loss of glucose stimulated [Ca ]; oscillations and coordinated insulin release with individual P cells show irregular and heterogeneous oscillations in response to elevated glucose (Benninger 2008 and Ravier et al. 2005).
20


At basal glucose, Cx36 gap junctions coordinate KAtp driven hyperpolarization across the P cells with the exchange of K+ between neighboring cells via the junctions (Benninger et al. 2011). In islets isolated from Cx36 deficient mice, there are spontaneous bursts of [Ca ]; at basal glucose as a result of the heterogeneity in the glucose sensitivity of individual P cells in the islet (Zhang et al. 2003 and Benninger et al. 2011).
Potential Roles of Cx36 in Type 1 and Type 2 Diabetes
Diminished expression of Cx36 is thought to contribute to the development of type 1 and type 2 diabetes so that the biphasic nature of insulin secretion is lost (Stamper et al. 2014 and Rocheleau et al. 2006).
In Type 1 diabetes, the insulin producing Beta cells are killed by infiltrating immune cells. During the progression of disease, the infiltrating immune cells produce large amounts of pro-inflammatory cytokines that cause oxidative stress in the islet and lead to Beta cell apoptosis leading to the loss of insulin and hyperglycemia. Studies have suggested a role for gap junctions in modulating cytokine-induced apoptosis showing that Beta cells are more susceptible to death under pro-inflammatory cytokines in a perfectly coupled islet and that overexpression of Cx36 gap junctions protects islets from apoptosis induced by pro-inflammatory cytokines, ER and oxidative stress. A possible mechanism by which this can occur is by Ca2+ regulation dysfunction. Studies have correlated cytokine-induced ER stress to reduced uptake of Ca2+ into the intracellular stores via SERCA pump. Cx36 gap junctions regulate intracellular [Ca+] and have been shown to affect Ca2+ uptake into the ER. It has been suggested that the loss of gap junction
21


coupling following pro-inflammatory cytokine induced ER stress in order to increase the concentration of residual Ca2+.
Type 2 diabetes is characterized by chronic hyperglycemia, hyperlipidemia, and insulin resistance that is developed as a result of genetic and lifestyle factors. Inflammation as a result of high levels glucose and circulating free fatty acids leads to glucotoxicity and lipotoxicity, which has also been attributed to disease development and progression. During the progression of disease, there is a decline in islet function and insulin resistance. Which one precedes the other? We dont know. We do know that the region of the gene encoding for Cx36 is located on the susceptibility locus for Type 2 diabetes, suggesting that there is some connection. Also, in mice models it has been shown that prolonged hyperglycemia lead to decreases in Beta cell coupling and Cx36 protein in isolated islets. Mice on a high far diet with high levels of FFA also showed decreased Cx36 gap junction coupling. Decreased coupling, lead to a loss of synchronization of electrical activity decreasing first phase and disrupting pulsatile second phase insulin secretion. Similar to what is observed in humans with Type 2 diabetes. This further suggests increasing Cx36 gap junction coupling as a therapeutic option to regain insulin action. Studies have shown disruption in Ca2+ similar to disruption in Cx36 gap junction coupling occurs in humans with higher BMI (potentially suggest pre-diabetes) suggesting Cx36 gap junction coupling may delay onset of disease.
Decreasing Cx36 Gap Junction Coupling in Katp Mutations
Previously, investigators have shown that Cx36 gap junction coupling can exacerbate the effects of mutations by specifically looking at mutations to the KAtp
22


channel making it insensitive to ATP resulting in a reduction in insulin secretion and hyperglycemia. Hypothesizing that removing Cx36 coupling could prevent the detrimental effects of subpopulation of unhealthy cells and restore electrical activity and insulin secretion, whereas the effects of full coupling would suppress the activity of individual and independent heterogeneous cells with a range of sensitivities in an islet characterized as a network of fully coupled cells.
Mice were generated with tamoxifen inducible overactive KAtp channels modeling NDM with wild type Cx36 expression had blood glucose levels that rapidly rose to over 500 mg/dL in the first two weeks after induction by tamoxifen injection and remained high. The mice also display significantly suppressed GSIS levels. Then in the mice expressing inducible over-active KAtp channels, a genetic knockout of Cx36 was introduced and it was found that normal blood glucose levels were maintained and there was a rescued GSIS response. These findings suggested that electrical coupling is incriminated in the progression of disease by exacerbating the effects of inactivating mutations in the progression of disease (Ngyuen et al. 2014).
Other investigators looked at how the inactivations (induced again by overactive KATp channels) in subpopulation of cells affect islet activity as a whole in a fully coupled network with wild type Cx36 expression. It was found that as cellular excitability approaches a critical threshold value, there is an emergence of critical behavior such that if over 20 percent of the cells in an islet express over active KATp channels then in the presence of Cx36 gap junction coupling there occurs a complete suppression of [Ca ], oscillations and insulin secretion. To further look at this, Aleena Notary et al. aimed to understand how the islet functions as a multicellular system and the role of gap junction
23


coupling in amplifying the effects of decreased cellular excitability in the islets with over active KAtp channels expressing different levels of Cx36 gap junction coupling. It was found that upon low levels of mutations there were elevations in Ca across the islet and frequent oscillations with and without electrical coupling, however at high levels of mutations there was transient elevations in Ca but the elevations were more prominent in the absence of coupling. In mice with high overactive KAtp percent mutation and no Cx36 expression, it was found that there were rescued islet wide activity and plasma insulin concentrations that were significantly diminished in the wild type Cx36 mice with high KATp mutation expression. Further suggesting coupling plays a critical role in amplifying the effects of mutations by rendering all the cells in an islet inactive as a result of mutations to a subpopulation of cells (Notary et al. 2016).
Figure 5: (a) Activity in the islet as measured by calcium activity for islets expressing different levels of Kir6.2[AN30 K185Q] mutation and gap junction coupling, (b) Insulin secretion data for a. (c) Plasma insulin for a. (d) Blood glucose from day 27 to day 29 for a.
24


Mathematical Model of the Islet
There are many computational models for the P cell and previously a model for the islet was generated as an extension of a model for the P cell. Here we will examine the model for p cell activity and how it was extended to a model of the islet. We will conclude the section with an in-depth analysis of the metabolic regulators in the model.
Models for ft Cell Activity
There have many models for P cell activity (Sherman 1996). The evolution of the many mathematical models started with modeling the primary ion currents of Ca and K+ (Chay and Keizer 1983, Chay 1997, Henquin 1990), to models including metabolism and ion current changes (Ainscow and Rutter 2002, Merrins et al. 2016), to models of Ca -cAMP-PKA dynamics in an oscillatory and regulatory circuit (Ni et al. 2011). The model used by investigators in our lab has been based off a model developed by Cha et al. that included essential components such as Ca dynamics and glucose dependence, as well as metabolism and ER dynamics (Cha et al. 2011).
While the understanding of P cell dynamics has been greatly improved as a result of these models, they were models only accounted for a single cell. And so the biology and information that comes from looking at the mechanism of a network of cells was unaccounted for and lost. This warranted the need of an extension of the islet model, allowing one to look at how P cells operate individually and in a network with varying levels of connections and communication capacities between neighboring cells.
25


Extension of the ft Model to the Islet Model
The previously introduced Cha-Noma model of the P cell was previously modified to account for cell-cell coupling by including a coupling current between neighboring cells to represent gap junction channels (Cha et al. 2011). Each current contributes to the membrane potential governed by the ordinary differential equation:
Gap junction coupling was then simulated by generating a heterogeneous coupling current gC0Up between neighboring cells assigned a center (i, j) governed by:
The islet model allows one to understand not just individual and independent cell behavior, but how a cell operates in a network of cells and induce cellular heterogeneities making the model more representative of physiological behavior.
Specific Aim
Diabetes is a severe disease that affects many people globally and is only expected to grow. In the progression of disease, the P cells are very amenable to changes in the environment and have the capacity to respond to treatments administered. But operating in an electrically coupled islet exacerbates the effects of a subpopulation of unhealthy cells and suppressive conditions by inactivating the whole islet. This thesis aims to further understanding of how the islet functions as a network of coupled P cells under electrically suppressive conditions by testing the effects of reducing coupling in metabolically inactivating conditions by inducing mutations to subpopulation of cells and
26


treating cells to induce uniform deactivation across all cells. This thesis explores the use of a the computational coupled oscillator model for the islet, microscopy techniques to examine calcium and NAD(P)H levels, transgenic mice, and experimental treatments to test islet dynamics and to what level electrical activity can be rescued upon glucokinase inactivation with modulated cell-cell coupling. The broader impact of this aim is to assess whether it is possible to recover electrical activity for specific mutations that occur in humans and lead to the development of diabetes.
27


CHAPTER II
METHODS AND MATERIALS
Mouse Studies
Ethics Statement
Experiments were performed in accordance with the relevant guidelines and laws, and approved by both the University of Colorado Institutional Biosafety Committee and the Institutional Animal Care and Use Committee.
Mouse Lines
A conditional GK homozygous knockout (GK -/-) or heterozygous knockout (GK +/-) was generated through crossing either GKlox/lox or GKlox/wt using a P cell specific mouse insulin promoter CreER with a variable deletion induced by tamoxifen. The Cre-loxP system of genetic recombination has been a powerful tool in determining tissue specific roles of GK in the P cell. Furthermore, the recombination can be P cell specific by generation of a MIP-CreER which allows for cell specific recombination dependent on the promoter driving Cre expression. The Cre functions to excise DNA regions of interest that are flanked by two loxP sites (34 nucleotide long DNA sequences with two symmetrical sides each having 13 basepairs and an asymmetric 8 basepair sequence in the middle). CreER is a fusion protein that consists of Cre and a mutated tamoxifen responsive estrogen receptor. In the absence of the inducing agent tamoxifen, the CreER remains in the cytoplasm. Upon tamoxifen binding, CreER is activated and enters the
28


nucleus and mediates recombination at the loxP sites (Carboneau et al. 2016). In our mouse model, once the CreER enters the nucleus of the P cell it cuts out the sequence.
After the generation of the GK knockout mice, the GK knockout mice were then crossed with mice expressing graded variations in Connexin36 coupling by having the following genotypes: Cx36 wild type (Cx36 +/+), Cx36 homozygous knockout (Cx36 -/-), and/or Cx36 heterozygous knockout (Cx36 +/-). Daily injections of tamoxifen (1-5 daily doses) were administered in 8-16 week old mice (at 50 mg/g normalized to body weight) to induce variable expression of GK knockout. Mice that lacked MIP-CreER were used as controls.
Blood Glucose and Plasma Insulin Measurements
Blood glucose was monitored and measured daily using a glucometer and then averaged from first couple days to determine baseline values and on 26th day to the 29th day after tamoxifen induction to determine resulting values.
Plasma insulin was measured on the 29th day after tamoxifen induction from blood samples collected from the submandibular vein were centrifuged for 15 minutes at 13,900RCF and assayed using the mouse ultrasensitive insulin ELISA (Crystal Chem).
Islet Isolation
Islets were isolated from by injecting collagenase in the pancreatic duct, harvesting the pancreas, digesting the pancreas, and handpicking islets. The islets were then cultured in RPMI (Invitrogen) with 10% FBS, 100 U/ml penicillin, and 100 pg/mL
29


streptomyocin and held at 37C humidified with 5% CO2 for 24 to 48 hours prior to experimentation.
Insulin Secretion
After isolation, the five islets per batch in duplicate were incubated in 2mM glucose with Krebs-Ringer Buffer (128.8mM NaCl, 5mM NaHCCE, 5.8mM KC1, 1.2mM KH2P04, 2.5mM CaCl2, 1.2mM MgS04, 10mM HEPES, 0.1% BSA, pH 7.4) and then incubated for 60 minutes in Krebs-Ringer Buffer with 20mM glucose. The medium was then lysed with 1% TritonX-100, frozen overnight at -20C, and then assayed for insulin secretion and islets measured for insulin content using a mouse ultrasensitive ELISA.
Calcium Imaging
Isolated islets were loaded with 4pM Fluo-4 (Invitrogen) in imaging medium (125NaCl, 5.7mM KC1, 2.5mM CaCl2, 1.2mM MgCl2, lOmM HEPES, 2mM glucose, 0.1% BSA, pH 7.4) for 75 minutes at room temperature, and were held in MatTek glass bottom microwell dishes. Fluorescence was imaged on spectral unmixing confocals LSM 780 (Zeiss) and LSM 800 (Zeiss) excited at 493nm using a diode laser with 506-545nm long-pass filter for emission holding temperature at 37C. The rate of image acquisition was an image per second and acquired 15 minutes after changing glucose concentrations from 2mM to 1 ImM and/or 20mM to record steady-state behavior.
30


NAD(P)H Imaging
Isolated islets were held in dishes with imaging medium (125NaCl, 5.7mM KC1, 2.5mM CaCl2, 1.2mM MgCl2, 10mM HEPES, 2mM glucose, 0.1% BSA, pH 7.4) for 60 minutes at 37C. Islets were placed in MatTek glass bottom microwell and imaged on LSM 780 (Zeiss) with 40x 1.2NA water-immersion objective. NAD(P)H
autofluorescence was imaged using two photon-excitation at 710nm with short-pass emission filter with collection peaking at 460nm and a six stack image was acquired with a width of 12pm.
Coupled Oscillator Model Simulations
As previously mentioned, this model has been expanded from the Cha-Noma P model to account for cell-cell coupling. In this thesis, we aim to introduce some altered glycolytic activity. Recall, each current contributing to the membrane potential is governed by the ODE:
= ICav + ITRPM + Isoc + hnsc + ^KDr + ^KCaiSKi + ^KCa(SK') + ^K(ATP) + L\aK
h ^NaCa + IpMCA
Gap junction coupling was then simulated by generating a heterogeneous coupling current gC0Up between neighboring cells assigned a center (i, j) assembled into a cluster by a sphere packing algorithm. The gC0Up follows a distribution with SD/mean=70% derived from previously published data (Farnsworth 2014}.
-c*vi = ii + 2j9cLv(yi-vj)
31


Previous Work: Katp Channel Current and Changes to the Katp Channel The Katp channel current in the model is described as:
^K(ATP) = 3 k iATP) X P OK (ATP) X (Y ~ )
where gK(ATP) describes the conductance of the KAtp channel, V represents the membrane voltage, and p0k(atp) describes the open channel probability of the channel. The open channel probability is described as:
Vk{atp}
.08 1 +

[ADP]\2( A5[ADP] .01 } \ ~r .026
Previously, Notary et al. modelled Kir6.2[AN30'K185Q] expression by modifying the
open channel probability of a fraction of cells (Pexc) to:
VK(ATP)Mut ~ y x (Pk(atp)) + C1 y)
where the modulation y is equal to 0.5 so that the KAtp channel has higher open channel probability and the opening of the channel prevents membrane depolarization and ultimately resulting in insulin secretion. Simulating islets (solid line) with increasingly higher Pexc showed an agreement with the experimental data (points).
Figure 6: Fraction activity as a function of percent mutation shown here as percent GFP. (a) Points indicate calcium imaging data from Cx36(+/+) mice with increasingly higher Kir6.2' XX30-Ivl85(-)l mutation expression shown by an increase in %GFP. And solid line indicates with simulation results for wild type coupling with gC0Up =120pS. (b) Distributions for islets isolated from Cx36(-/-) mice in points and simulation results for no coupling with gCOup=0pS.
32


We see the emergence of critical behavior displayed once again where if 20 percent of cells in the islet express ATP insensitive K-channels in the presence of gap junction coupling, the coupling will amplifying the effects of the subpopulation of cells by essentially shutting off the whole islet. In the absence of coupling, there is a more gradual decline with the uncoupled islets being more active than the coupled islets in the high percent mutation region.
While the model for percent mutation rendering KAtp channels ATP insensitive in a subpopulation of cells is a good model for NDM, mosaic expression of mutant KAtp is unlikely in human disease However, mosaic mutation expression is unlikely in human disease. And so, the investigators simulated uniform KATp activation (modeling diazoxide treatment) across all the cells in the islet. Diazoxide treatment was modelled by further modifying the open channel probability in all the cells to:
PK(ATP)Mut = + (1 a) X Vk(atp)
where a represents a residual current or the fraction of current remaining at saturating ATP concentrations. After modulating a by increasing the amount of residual current, the simulation results indicated a similar rapid onset of suppression occurs upon increasing diazoxide activation across all P cells in a fully coupled islet and a more gradual decline in activity in the absence of coupling. Furthermore, there is once again a region in the distribution where activity is fully suppressed in the presence of coupling and is rescued in the absence of coupling.
33


9oxf = 12QpS
9axp=sPs
9(0^= OpS
L
els w cmo
OOfl D.02 0.04 0.00 0.00 0.10
Figure 7: Fraction cells active as a function of current increase uniformly across all cells simulating diazoxide application for modulated coupling conductances such as gCOup=120pS, 50pS, and OpS.
Notary et al. (2016) showed different mechanisms in which inexcitability could be introduced to the islets resulting in KAtp over-activity and they all exhibited similar behavior being that there was a sharp decline observed in the presence of coupling and a gradual decline in the absence of coupling. Furthermore, the findings suggested that there was potential to regain function under the suppressive regions of the distribution by uncoupling the islet.
Metabolic Regulators of the Model
When there is an increase in the concentration of plasma glucose, the P cells take up glucose via a glucose transporter such as Glut2. The uptake of glucose by the glucose transporter occurs very quickly and works to establish equal intra- and extra- cellular concentrations of glucose (Johnson et al. 1990). Once transported into the cell, GK phosphorylates glucose to G6P. The rate of the GK reaction is governed by the model:
34


kgic is the rate of the glucokinase reaction and follows a normal distribution centered around the rate constant with 10% standard deviation. GK has two primary substrates glucose and ATP. For glucose, [Glu], is the cytoplasmic glucose concentration, Kmgi is Michaelis-Menten constant or the glucose concentration at which the reaction is half maximum, and hgl is the Hill coefficient or the cooperativity between ligand and enzyme. Similarly, [ATP]; is the cytoplasmic ATP concentration, KmAxp is the Michealis-Menten constant for ATP, and hATP is the Hill coefficient for ATP.
Changes to the Rate of Glucokinase Reaction
Mutations nullifying GK activity in different percentages of subpopulation of cells (PMut) was simulated by assigning Jgic=0 to the number of cells (where number of cells with the mutations equaled 1000 times PMut)Uniform graded decreases in the rate of glucokinase activity across all the cells in the islet was modelled by modifying Jgic so that:
J'sic = C1 percent decrease) X Qglc)
Uniform increase in the half maximal concentration of glucose was also modified across all the cells in the islet such that:
K'mgl = (1 + percent increase) X Kmgl
A uniform increase in the half maximal concentration of ATP was modified similarly to that of glucose shown above.
35


Changes to the Rate of Glucokinase Heterogeneity
As previously mentioned, the rate of the glucokinase reaction is governed by the variable kgic which follows a normal distribution centered at 0.000126ms'1 with a 10% standard deviation. To test the effects of glucokinase heterogeneity under different suppressive conditions, glucokinase heterogeneity was modified by expanding the standard deviation to 20% of the mean, 30% of the mean, and 50% of the mean.
Data Analysis
Calcium Imaging Data Analysis
MATLAB scripts were generated to analyze all the images acquired from
fluorescence imaging. After loading images, islet mask was selected and further analysis
was confined to the selected area. The images were binned in a 3x3 bin to sharpen image.
Photobleaching artifacts from the intensity fluctuations characterized by the time course
of the bin were removed by detrending the time course (MATLAB built in function). The
time course was then analyzed by a peak finder algorithm to detect peaks and troughs
with area of selection defined by ^maxtTcj-mmCrc), A peak amplitude map was
/&
generated by subtracting the mean peak amplitude by the mean trough amplitude for the bin. Then a silent cell was selected manually from the area with low peak amplitude value in the peak amplitude map and a silent cell threshold was determined by locating the maximum peak amplitude value in the area selected. Then all the pixels in the peak amplitude map were compared to the silent cell threshold such that if a pixel in the islet
36


had a peak amplitude of two fold greater than the silent cell threshold then the cell was considered active.
Figure 8: Activity map superimposed on islet image acquired by confocal imaging. Scale bars indicate degree of correlation of activity across the islet.
Background area in the islet was calculated by identifying pixels with peak amplitude values lower than 80 percent of the silent cell threshold. This percentage was identified after doing extensive testing. The fraction active area was calculated based on the number of active cells in the enclosed area divided by the enclosed islet area subtracted by the background area.
Figure 9: Background area correction superimposed on gray scale peak amplitude map where background area is defined by thresholding
the peak amplitude map. 37


Additionally, the active area of the images was further tested to analyze for duty cycle by defining time on as counts of intensity fluctuations above 25% of the mean and calculating duty cycle as time on divided by total time. Finally, correlation of activity was also calculated by randomly selected three cell areas on the peak amplitude map and comparing the time courses for the three areas against every 5x5 bin time course characterizing a cell time course in the islet using a correlation function (MATLAB built in function). The pixels were assigned a correlation from 0 to 1 in a correlation map with respect to each cell for the three cells selected and fraction correlated area was selected from the highest correlated area calculated from the three correlation maps.
NAD(P)H Imaging Data Analysis
MATLAB scripts were once again generated to analyze all the images acquired from fluorescence imaging. After loading the six stack image (example images shown below), an islet mask was selected for each individual stack and the mean intensity was calculated in that two dimensional area. This was repeated for the remaining images in the stack. Then a mean of the means were calculated to quantify mean NAD(P)H intensity for the islet. After all the images of islets were analyzed for the day, the intensities were normalized to the wild type control held at 2mM glucose to calculate fold change. Images of representative islets are included below for wild type islets held at 2mM glucose and then at llmM glucose. Notice the increase in net intensity which is quantified in NAD(P)H fold change.
38


Figure 10: (a) NAD(P)H autofluorescence collected by two-photon microscopy at low glucose (2mM). (b) Same as in a for islet held at high glucose (20mM).
Simulation Data Analysis
MATLAB scripts were once again generated to load and analyze text files outputted from the simulations including Ca levels, NADH, and membrane potential. For all time courses, the first 200 time points were excluded so as to analyze the behavior of only the steady-state solution.
First, the calcium text file is loaded and the first 200 time points were cut off. Previously published data by Notary et al. (2016) indicated that intracellular calcium fluctuates anywhere from 0.9pM to 0.45pM choosing a silent cell threshold of 0.165pM. To calculate the number of active cells, the script identifies activity based on the silent cell threshold. For every cell, it looks for the maximum value or the peak of the cells time course and identifies a cell as active if the maximum value exceeds 0.165pM. So the percent of active cells is quantified as the percent cells with maximum value over 0.165 pM divided by the total number of cells. Additionally, the script calculates duty cycle by
39


identifying time on as time where the calcium levels exceed 50% of the amplitude of the llmM calcium control time courses. The duty cycle for the cell is then the time on divided by the total time.
Then, the potential text file is loaded and the first 200 time points are cut off and the potential time courses for the cells are used in an alternative mode to calculate percent active cells. For every cell, the script looks at the maximum value of the cells membrane potential time course and if the amplitude exceeds -45mV the membrane is not at resting potential then the cell is considered to be active. Percent activity is then characterized as the percent active cell divided by the total number of cells.
Finally, the NADH text file is loaded and the first 200 time points are cut off again so that the analysis is for steady-state NADH solution of the differential equation. There two quantifications from the oscillations are calculating mean NADH and NADH amplitude. To calculate mean NADH, a mean of the time course is obtained (using a built in MATLAB function). Then to calculate the NADH amplitude the NADH oscillatory time courses for each cell is then analyzed for its maximum and minimum and the amplitude of the cell is quantified as the maximum subtracted by the minimum.
40


a
b
Figure 11: (a) Representative time courses of simulation results for changes in [Ca2+]i, NADH, and Vm for 100 cells with wild type gap junction coupling (gCOup=120pS) simulated at llmM glucose, (b) Same as in a for simulated islet with no gap junction coupling (gCOUp=0pS).
41


CHAPTER III
RESULTS
GK Deletion Modeled, Shows Critical Behavior
Previously, other investigators have shown that if you inactivate an increasingly higher subpopulation of cells in an islet with wild type or full electrical coupling there is an emergence of critical behavior. Simulated data showed that if 20 percent of the cells in a fully coupled (gCOup=120pS) network of cells expressed over-active KAtp channels by being presented with a Kir6.2[AN30,K185QI mutation, then the coupling will amplify the effects of that subpopulation of cells such that all the cells in the islet will transition from being globally active to inactive shutting off essentially all electrical activity and plasma insulin levels (T.H. Hraha and M.J. Westacott, et. al., (2014). In this section of the thesis, we will examine how coupling acts to enhance the action of a proximal metabolic regulator of electrical activity upstream of the KAtp channel by looking at GK.
To assess how deletion of GK in islets affects electrically and metabolic response with and without electrical coupling, we simulated islets with increasing percentage of cells expressing GK knockout (GK -/-) characterized as Pmup We found that if the subpopulation of non-metabolically active cells in the fully coupled islet exceeds 40 percent, then all the cells in the islet transition from being active to inactive. In the absence of coupling (gCOUp=0pS), we see a more gradual decline in activity. To assure that the behavior is as a result of decreased metabolic activity, we looked at the mean parameters of the coupled/uncoupled cells at PMut of 70 percent and found that GK KO cells indeed presented lower metabolic activity with significant decreases in ATP and
42


NADH but no change in gK_ATP- The distributions compared to one another indicate that before reaching the critical threshold the fully coupled is more active than the uncoupled islet and after the threshold the uncoupled islet is more active than the coupled islet. The findings suggest that upon high levels of percent mutation or PMut, in the absence of coupling the heterogeneity of cells in the islet become more explicit such that the more easily excitable cells are electrically active whereas perhaps the cells that have been inactivated are less electrically excitable.
a
i/i
"a3
u
+>
u
<
o
+>
ro
cc
p
Mut
ATP NADH gkATP
Figure 12: (a) Islet activity for increasing GK deletion in simulated islets with wild type gap junction coupling (gcouP=120pS) in black and no gap junction coupling (gcouP=0pS) in red. (b) Fold change in metabolic and Katp parameters for cells with 70% GK deletions (dark red) and with wild type GK (light red) for simulated islet with no coupling.
43


GK(-/-) Mice Generation and Physiology
To test the predictions of the computation model, GK conditional homozygous knockout(GK -/-) and heterozygous (GK +/-) mice were generated by crossing either GKiox/iox or GKlox/wt mice using P cell specific mouse insulin promoter (MIP) CreER inducible with tamoxifen, so that the Cre is only function when the inducing agent is administered. As previously mentioned, one to five tamoxifen injections were administered to the mice to generate a graded effective recombination or an increasing level of GK mutations present. Additionally, the GK knockout mice were crossed with Cx36 knockout mice to achieve graded variations in electrical coupling.
The GK knockout mice were monitored for blood glucose during the course of the tamoxifen injections and plasma insulin immediately prior to isolation. The blood glucose time courses for the GK knockout mice with and without electrical coupling a similar progression to diabetes between the different injection values. The blood glucose time courses indicated that the different tamoxifen injections made the mice diabetic (commonly defined as having blood glucose of over 200 mg/dL) after the first week of injections and maxing out the glucometers at 600mg/dL prior to isolation. Furthermore, the blood glucose time courses were the same for the mice expressing wildtype or no Cx36.
44


a
b
Figure 13: Blood glucose time courses during tamoxifen injection period to monitor disease progression for (a) 1 daily injection, (b) 2 daily injections (c) 3 daily injections, and (d) 5 daily injections in mice expressing wild type Cx36 gap junction coupling (black) and no Cx36 gap junction coupling (red).
When looking at fold change in blood glucose for the GK knockout islets measured as the
ratio of final values to initial values, we see that the data correlates with the other
findings suggesting significantly large increases in blood glucose and upon uncoupling
there is no relevant changes in blood glucose.
45


GK (-/-), Cx36(+/+) GK (-/-), Cx36(-/-)
Figure 14: Fold change in blood glucose compared to wild type controls for mice with wild type Cx36 gap junction coupling expression (black) and no Cx36 gap junction expression (black) and 1, 2, 3, and 5 tamoxifen injections administered for gradation in GK knockout.
The plasma insulin levels of the GK knockout mice indicate that there has been a significant reduction in circulating insulin content. Upon uncoupling the cells in the islet, the trend indicates that there is still a reduction in plasma insulin, which correlates with the blood glucose time courses that indicate these mice are diabetic.
GK (-/-), Cx36(+/+)
09 GK (-/-), Cx36(-/-)
p=0.075
Control GK KO
5x
Figure 15: Plasma insulin measured prior to islet
isolation from mice with GK knockout activated by
tamoxifen injection and wild type (black) and no
Cx36 gap junction coupling (red). 46


NAD(P)H Imaging of Islets Isolated from GK(-/-) Mice to Characterize Metabolic Activity
Isolated islets from GK knockout mice were imaged for NAD(P)H levels using two photon microscopy and analyzed by MATLAB scripts (for details of microscopy and analysis refer to the methods chapter) to affirm decline in metabolic activity. Change in NAD(P)H content indicated that there was a decline in metabolic activity in the GK knockout islets when compared to the controls.
a
b
GK (-/-), Cx36(+/+)
GK (-/-), Cx36(+/+)
GK (-/-), Cx36(-/-)
c
GK (-/-), Cx36(+/+)
GK (-/ ), Cx36(-/-)
Figure 16: Change in NAD(P)H content measured via two photon microscopy for islets isolated from (a) GK knockout mice with wild type Cx36 (GK Cx36 +/+) (b) GK knockout mice with no Cx36 (GK CX36 -/-) (c) and results from the two mice superimposed to compare responsiveness.
47


Isolated islets from GK knockout mice with Cx36 knockouts (GK-/-, Cx36-/-) and no coupling also showed a decline in metabolic activity compared to controls for all injection values except the islets from mice treated with two injections. When comparing the islets exhibiting GK knockout with coupling and without coupling, there was no significant change in metabolic activity between the two conditions suggesting that even after reducing cell-cell communication the effects of the mutations induced by the tamoxifen injections are still considerable and the islet is metabolically silenced.
Calcium Imaging of Islets Isolated from GK(-/-) Mice to Investigate Calcium Dynamics and Electrical Activity
Previously, Notary et al. showed that mosaic inactivations by mutating the Katp channels to become over-active resulting in a dramatic decline in activity, which was recovered to an extent by uncoupling the islet (refer to Chapter I for specific details). To test this for GK, the other more upstream regulator of electrical activity in the model, we imaged GK knockout islets for changes in the [Ca ], content shown by changes in intensities of fluo-4 and analyzed the time courses using MATLAB scripts previously mentioned. Findings suggest a statistically significant reduction in activity for one, two, and five tamoxifen injections and a reduced trend for three tamoxifen injections.
48


GK (-/-), Cx36(+/+)
GK (-/-), Cx36(-/-)
GK KO GK KO GK KO GK KO lx 2x 3x 5x
Figure 17: Islet activity measured by calcium dynamics as a function of different levels of GK knockout induced by different quantities of tamoxifen injections for mice with (a) wildtype levels of Cx36 gap junction coupling (black) and (b) no Cx36 gap junction coupling (red).
When we look at islets expressing GK knockout and Cx36 knockout, we see that there is no significant recovery in activity upon uncoupling the islet. Although the trends do seem to indicate that upon increased tamoxifen injections there is a trend of higher activity for the Cx36 knockout islets. To conclude this for certain, we would need to perform additional experiments capturing the distribution of activity of the behavior prior to the decline in activity.
TdTomato Quantification Identifies Percent Mutation Higher Than Expected
To quantify percent mutation in order to conclude how much mutation is actually expressed, we quantified percent area of tdTomato expression in respect to islet area. We quantified islets isolated from one, two, and three daily injection tamoxifen treated mice similar to previous experiments. Our control for this experiment was quantifying islets isolated from mice with no CreER and injected with five tamoxifen injections. Our results
49


indicated that we had significantly higher percent mutation expression than what others had previously seen with the same dosage to induce changes in gene expression in the electrical regulator KAtp.
, BGK (-/-), Cx36(+/+)
0.9
Control GK KO GK KO GK KO
lx 2x 3x
Figure 18: Quantification of tdTomato area by measuring ratio of tdTom+ area to total area as indicated by live cell staining using islets isolated from GK knockout mice treated with 1, 2, and 3 tamoxifen injections.
The tdTomato expression results addressed why the GK knockout mice physiology data, NAD(P)H data, and fraction active area data were so dispersed and unresponsive. And because the percent mutation was the same across all the tamoxifen injections, we were interested in seeing if perhaps we could recover electrical function if we summed up the fraction activity for all the tamoxifen injections. And we discovered that we were not able to recover function. It showed us exactly how potent a homozygous mutation to the GK gene can be.
50


GK (-/-), Cx36(+/+)
GK (-/-), Cx36(-/-)
Control GK KO
lx, 2x, 3x
Figure 19: Fraction activity for control and GK knockout islets where knockout was induced by 1, 2, and 3 tamoxifen injections and expressing wild type Cx36 gap junction coupling (black) and no coupling (red).
Decreased Rate of Glycolysis Modeled, Shows Critical Behavior
Mosaic expression of GK knockout is unlikely to occur in human disease and so we modelled and experimentally tested uniform deactivation of GK across all cells in the islet. In this section, we hope to assess how inhibition of GK in islets affects islet electrical activity dynamics with and without coupling using simulations. Instead of inducing no GK activity in a randomly selected subpopulation cells making them electrically inexcitable, we induce graded decreases in GK activity across all cells in the islet with wild type electrical coupling and no electrical coupling. When we plot ratio of active cells as a function of percent decrease in the rate of GK, the model predicted the following curves:
51


1
0.9
0.8
IS)
0.7
01
u
01 0.6
>
+- 0.5
u
<
o 0.4
+-
ra cc 0.3
0.2
0.1
0
10 20 30 40 50 60 70 80 90 100
% decrease in kgk
Figure 20: Islet activity as a function of percent inhibition of GK activity across all cells in simulated islets with electrical coupling in black (gCOup=120pS ) and no electrical coupling (gcouP=0pS) in red.
We see the emergence of critical behavior similar to what we have seen before. The results indicate that in a fully coupled islet, a 65 percent decrease in GK activity is tolerated above which all the cells in the islet are inactivated. In the absence of coupling, we see a more gradual decline so that prior to the critical threshold the coupled islet is more electrically active than the uncoupled islet and after the critical threshold the uncoupled islet is more electrically active than the coupled islet. This suggests that in islets subjected to large decreases in metabolic activity (consistent with the right region of the distribution) there is a possibility of recovery of activity by uncoupling the islet.
52


NAD(P)H Imaging of Cx36(+/+) and Cx36(-/-) Islets Treated with GK
Inhibitor to Characterize Metabolic Activity
To test the predictions generated by the computational model, we treated islets isolated from Cx36 wild type and knockout mice with increasing concentrations of the GK inhibitor Mannoheptulose (MH). First, we wanted to test the decline in metabolic activity by acquiring NAD(P)H levels for islets expressing wild type Cx36 and no Cx36 at 2mM Glucose, llmM Glucose and llmM Glucose + Treatment. We found that as we increase the concentration of MH, we get a graded decline in mean NAD(P)H content. Furthermore, the Cx36 knockout islets show higher metabolic activity than the Cx36 wild type islets. The increase in NAD(P)H content after adding high glucose and subsequent MH treatments shown below as the change in the NAD(P)H content suggests that the Cx36 knockouts have a lower fold change in NAD(P)H content from basal values when compared to Cx36 wild types.
Figure 21: Metabolic activity for islets isolated from Cx36 wildtype mice (black) and Cx36 knockout mice (red) treated with varying concentrations such as 3mM, 5mM, and lOmM of GK inhibitor MH. Figure (a) indicates mean NAD(P)H and (b) indicates fold change in NAD(P)H content.
53


Calcium Imaging of Cx36(+/+) and Cx36(-/-) Islets Treated with GK Inhibitor to
Characterize Calcium Dynamics and Electrical Activity
2+
To test electrical activity dynamics, we measured changes in the [Ca ]; levels
using confocal microscopy techniques and analyzed the time courses using MATLAB
scripts that look at the oscillatory dynamics across the islet. In the figure we show
fraction active area and duty cycle. The fraction active area plots indicate fraction activity
as a function of treatment for islets with wild type coupling and no coupling. For islets
with full electrical coupling, the activity at llmM glucose is characterized as baseline c
activity and islets treated with increasing levels of MH show decreasing levels of activity that are statistically significant. The trend of decreasing activity shows that the Cx36
knockout islets have a more gradual decline in activity than the Cx36 wild type islets.
Cx36(+/+)
Cx36(-/-)
11G 11G 11G 11G
+ 3MH +5MH +10MH
Figure 22: In islets from mice expressing wild type levels of Cx36 (black) and no Cx36 (red), results indicate (a) islet activity as a function of different MH treatments to modulate decreasing metabolic activity, (b) the distribution of the heterogeneity in islet results, and (c) duty cycle of the fraction active area.
54


Clinical Mutations in Glucokinase Examined
To test how the principles and findings could be extended to humans, we simulated GK mutations that occur in humans leading to the development of diabetes using kinetics reported in literature and investigated whether activity could be recovered by uncoupling the islet. Furthermore, we looked for literature citing patient data and GK mutation kinetics to see at which point in the development of disease could we recovery the most activity.
Simulating Additional Metabolic Parameters That Affect the Rate of GK Activity
Literature citing the kinetics of mutations to GK would not only cite changes to the rate of GK activity but would also report shifts in the half maximal concentration of glucose and the half maximal concentration of ATP with their respective hill coefficients. And so, before simulating mutations to GK that lead to the development of diabetes in humans it was important to characterize the different parameters that contribute to the activity of GK. This would allow us to better predict and understand how a mutation would behave by looking at where it would lie on the activity curves of the different parameters. And so, we simulated increases in the half maximal concentration of glucose (Kmgi), increases in the half maximal concentration of ATP (KmAxp), and the hill coefficient for glucose (hgl).
f [ATP]*iTP \f [Git,]"'1 y
s!c\[ATP]^tp A-K^J\lGlu]^ + K^\rNADH,] [NADHV
55


Figure 23: Islet simulations with wild type gap junction conductance in black (gCOup=120pS) and no gap junction conductance in red (gCOup=0pS) for modulated percentages in (a) the half maximal concentration of glucose, (b) the half maximal concentration of ATP, and the glucose hill coefficient with simulated (c) coupling (d) and no coupling.
Simulating percent increase in the half maximal concentration of glucose or increasing the glucose threshold after which GK would be activated exhibited a distribution similar to the ones we have seen before with the exception that it is more right shifted. Again the coupled islet shows a more rapid decline in activity and the uncoupled islet has a more gradual decline in activity. Then, simulating percent increases in the half maximal concentration of ATP did not contribute much to the decline in activity. In fact, it
56


contributed to the decline only after uncoupling the islet and a two-fold increase in the ATP threshold. Finally, simulating changes in the degree of cooperativity between the ligand binding to the enzyme by modulating the hill coefficient also did not contribute much to the decline. We see that a 50% reduction in cooperativity shifted the distribution only 10 percent to the right. To conclude, we ran these simulations and others for five random number seeds and we discovered that the most important parameters in determining how a GK will impact islet behavior include the rate of GK and the half maximal concentration of glucose.
Modeling Mutations Occurring in Humans
So we looked to the literature citing the kinetics of mutations in GK that lead to diabetes and we found mutations that lead to the two most common monogenic forms of diabetesPNDM and MODY-2 (refer to Chapter I for details). We were interested in the rate of glycolysis or the rate of GK, the half maximal concentration of glucose, the half maximal concentration of ATP, and their respective hill coefficients. Then we simulated the mutations with and without electrical coupling to determine if we could recover electrical activity by eliminating electrical coupling in an islet expressing the GK mutation simulated at low and high levels of glucose.
57


Fraction Cells Active
1
WT 78V LO s < :c o 2 00 WT 35V § CO r>. 1 3 QJ 1
ro < 8 i- 2 (N rM 2 o ro cc oc z
I g 3
e? c/>
01 ft
ro ^
0)
>
4-
U
<
CD
U
c
o
u
05
1 <
ro
2 3
coor<^^ora: oo o o m
> < CO 00 (N
o o
o ^ cn o
ro O rH
5 5
lO <
^ ^ m n in co
M *t (N (N (5
oi oi T1 f (Tl (Ji 00 <71 OC Is* T1 > rsi LO CL ro CD CD ro CL 3 >64Y T65I cn CO lO CO LO 2 8 > < ft < LO u < LO LO CD LL LO CD CD O
3 2 < 3 2 CL CO CO CO > > > > (N >- :> >- (N > 2 * CM 1
Figure 24: (a) Calcium dynamics results from simulations of PNDM mutations with wild type gap junction coupling in black (gCOUp=120pS) and no gap junction coupling in red (gcouP=0pS). (b) As in a for MODY-2 mutations.
58


And we found that we were able to recover electrical activity in a subset of mutations by uncoupling the cells in the islet. Most of the mutations we found reported the kinetics of MODY-2 mutations that are heterozygous and so a lot of the results of the simulations show full activity at high glucose, however in the subpopulation of MODY-2 mutations that are inactivating we were able to recover function in two cases. There were four PNDM cases reported from which three were inactivating and there was no recovery of activity for any of those cases upon uncoupling the cells in the simulated islet.
Ranking Mutations by Clinical Severity
The literature did not report much about the clinical severity of the mutations. For MODY-2 mutations, the treatment is just diet and exercise. But when you look at the electrical responsiveness of the islets, you see there is still a significant number of cases we identified with no activity. We thought it would be important to better characterize the clinical severity of the mutations to have a more complete understanding of the effects and clinical manifestations of those inactivating mutations, and so we looked for literature citing patient data.
When we rank mutations by the results of their oral glucose tolerance tests (OGTT), we find that the distribution of results could be split into three primary stages characterized as mild, mild to severe, and severe. Simulating islets expressing those mutations indicated that the mild mutations were generally active, the mild to severe being more inactivating, and the severe being generally inactive. Interestingly, our results also indicated that we were able to recover activity in the most sever mutation by uncoupling the islet.
59


a
250
b
c
oj
>
t;
<
J£
a
£
o
2
Figure 25: Mutations in GK presented clinically (a) ranked by oral glucose tolerance test (OGTT) scores and (b) simulated with coupling in black (gCOup=120pS) and no coupling in red (gCOup=0pS). (c) Fraction cells active grouped under categories of OGTT scoring of mild, mild-severe, and severe.
When we ranked mutations Dy no Ate percentages, we see a similar trend in the three stages. Then after simulating islets expressing the individual mutations with and without coupling, our results indicated recovery of activity in islets with milder HbAlc scoring. But when we look at the mean behavior as a function of severity, we see slight recovery in all three scores in the islets with the simulated mutations and no coupling.
60


a
10
9 I 8
7 -]
6 1
5 -I
4
2
1
£>v£)Ln^l^-iriroo^^fr>j
Figure 26: Mutations in GK presented clinically (a) ranked by oral glycosylated hemoglobin (HbAlc) percentages and (b) simulated with coupling in black (gCOup=120pS) and no coupling in red (gCOup=0pS). Fraction cells active grouped under categories of HbAlc scoring of mild, mild-severe, and severe.
61


CHAPTER IV
DISCUSSION AND FUTURE WORK
The islet is a very complex and dynamic system composed of multiple cell types that are important for establishing and maintaining glucose homeostasis. The insulin producing p cells possess a heterogeneous distribution of sensitivities. The P cells operate within a network and the strength of the connections within the network determine how the cells in the islet respond to a stimulus. In a network with strong connections manifested by the presence of gap junction coupling, the cells will be recruited to respond similarly to the environment they are in. In the absence of a stimulus or at basal glucose concentrations of less than 5mM, all the cells in the islet are quiescent. In the presence of a stimulus or at high levels of glucose, all the cells in the islet are actively responding (Benninger et al. 2011, Speier et al. 2008). This manifests itself in no electrical activity and insulin secretion at basal levels of glucose and synchronous oscillations of membrane potential, change in intracellular calcium concentration, and insulin secretion at high levels of glucose. In a poorly connected network with low to no levels of gap junction coupling, there is a heterogeneous distribution of responses depending on the sensitivity of the cells to extracellular changes. This results in a subpopulation of more excitable cells exhibiting spontaneous oscillations at low levels of glucose which become more frequent and in an increasingly number of cells upon the addition of high glucose (Benninger et al. 2014). These effects were removed in the presence of coupling, as cells were fated to respond to the stimulus globally (Benninger et al. 2011). Any individual P cells capacity for electrical excitability is tightly coupled to its glucose responsiveness
62


which is regulated via the action of two primary regulators of electrical activityGK and the KAtp channel. However, p cells do not act autonomously and so the responses of the individual P cells are shared through the cellular interactions established by Cx36 gap junction coupling. This work has been inspired by previous studies investigating the effects of a subpopulation of cells expressing over-active KAtp channels induced by a Kir6.2[AN30,K185Q] mutation and global decreases in excitability by increasing the amount of residual current in the KAtp channels within islets operating as networks with and without cell-cell interactions via electrical coupling. Previously, investigators have used mouse models, experimental tools, and predictive mathematical modeling to investigate these effects on islet dynamics and electrical excitability. This thesis aims to look at a metabolic regulator of electrical activity by applying previously generated principles to understand islet dynamics as a result of decreased metabolic activity characterized by mutations in GK-inducing decreased cellular excitability in the islet.
Global Behavior and GK Heterogeneity
It has been previously shown that if the percent of decreased cellular excitability exceeds a critical threshold value, then the intrinsically established connections between P cells by gap junction coupling amplifies the effects generated by the percent of decreased cellular excitability by shutting off essentially all electrical activity in the islet (Hraha et al. 2014). It has also been shown by Notary et al. that uncoupling a heterogeneous system of cells by decreasing gap junction coupling can allow heterogeneity to dominate the islets capacity to conduct electrical activity in the post-critical phase providing a means by which to recover electrical function and insulin
63


secretion (Notary et al. 2016). Our computational model results suggest that the window to recover electrical function exists for GK-mediated decreases in cellular excitability, but our experimental findings conclude that we havent identified it yet.
Mosaic GK Deletion Effects on Global Behavior
Using the computational model to simulate islets with increasingly greater number of cells with GK deletions indicated that GK regulates islet function to behave in a biphasic manner in the presence of electrical coupling, so that in the presence of electrical coupling, the islet functions in a globally active or inactive state. This mechanism of regulation of islet function could prove to be both helpful and detrimental to the islet. In low percent deletions, the electrical coupling allows for the cells with no deletions to compensate for the loss of function in the cells with GK deletions. But as the number of GK deletions increases past a critical threshold determined by the capacity of the unaffected cells to compensate, then the resulting population of cells will no longer be enough to compensate and cellular interactions will amplify these effects by shutting off activity globally across all the cells in the islet. Our results indicated that 45 percent of cells with GK deletions in an islet with full electrical coupling (gCOUp=120pS) is tolerated, above which the islet shows a dramatic decline in ratio of active cells (Figure 12a). Upon eliminating these cell-cell interactions in the network by computationally simulating an islet with no gap junction conductance (gCOup=0pS), this phenomenon of binary function is removed. Our findings indicate that increasing the percent of GK deletions in cells in an uncoupled islet results in a linear decline in activity (Figure 12a). In the absence of coupling, our results indicated that GK deletions resulted in a loss of global behavior with
64


a more gradual decline in activity so that that the heterogeneity in an individuals cells capacity for responsiveness was not reduced. To better understand the mechanism behind this, we looked at the functional characteristics of the GK knockout cells compared to the GK wild type cells in the phase after the critical threshold. We found that the knockout cells had significantly high metabolic activity shown by increases in the NADH and ATP content, regardless of the reduction in metabolic activity present in the GK knockout cells (Figure 12b). Since the simulations predicted that uncoupling the islet at high levels of mutation by GK deletion could result in higher electrical and metabolic excitability, this also suggests higher insulin secretion rates despite the presence of GK deletions.
To test the predictions generated by the computational model, we generated a mouse model expressing tamoxifen inducible GK deletions by the use of a MIP driven CreER and the mice also expressed different levels of gap junction coupling by having either wild type or no Cx36. We administered graded levels of tamoxifen injections to induce gradations in effective Cre recombination or percent GK deletions. However, our results indicated that despite the number of tamoxifen injections our mice had similar progression to disease (Figure 13). All the GK knockout mice rapidly became diabetic and remained diabetic at the end of the first two weeks of injections regardless of whether the mice expressed wildtype coupling or no coupling (Figure 13). When measuring fold change in blood glucose (Figure 14) and plasma insulin levels (Figure 15), we confirmed that there was really no decrease in blood glucose and increase in plasma insulin upon eliminating gap junction coupling. The physiological data accumulated from the mice over the experiments were not promising, but we persisted with the experiments hoping to extract meaningful data about islet function and dynamics. And we discovered just
65


how potent a GK deletion can be in its effects on global behavior. NAD(P)H imaging confirmed that there was in fact a decline in metabolic activity between the control and the different injection values, however the decline was not graded as the changes between the injection values did not generate statistically significant changes in metabolic activity (Figure 16a). Upon also introducing a Cx36 knockout in the GK knockout mice, we saw that the mice also generally showed decreases in the metabolic function in the islets (Figure 16b). And the decreases in metabolic activity are not statistically different than the metabolic activity of Cx36 wild type islets as we had hoped. Confocal imaging of calcium dynamics agreed with the data indicating metabolic function. The fraction of electrical activity was also decreasing but there was no gradation in the decreases as we increased tamoxifen injections (Figure 17). And after introducing a Cx36 knockout, we saw no significant recovery of electrical activity by measuring calcium dynamics. The lack of gradations in our metabolic and electrical data suggested that perhaps we are at higher levels of percent mutation than we had anticipated. Our td Tomato quantification allowed us to understand just how much percent mutation we are inducing by tamoxifen injections and we discovered it was high. Our results indicated anywhere between 70 percent to 85 percent mutations were induced by one to three tamoxifen injections (Figure 18). This was a lot more than what had previously been seen. However, since the one to three injection values appeared to have similar percent mutations, we thought perhaps we could investigate whether we could recover function by averaging the activity over the injections values. And we found that we could not. There was no recovery in activity upon uncoupling the islet for the GK knockout mouse model we generated
66


(Figure 19). This showed us the range of potency of GK mutations in affecting islet dynamics globally with and without electrical coupling.
We tried optimizing our experimental techniques by isolating islets sooner so that we could get a better representation of islets isolated from diabetic mice and not mice that are close to dying. Additionally, we tried reducing the time we place the pancreas of the mice in collagenase solution to increase viability. And while the modifications may have generated improvements in the quality of the islets, they did not improve average function. And so, we were not able to experimentally capture the capacity of a heterogeneous group of cells to locally rescue activity in an environment of decreased cellular excitability. However, we believe this is as a result of too much decreased cellular excitability. Our experimental data is indicative of the part of the prediction where there is essentially too much GK deletion and we think that perhaps the phenomenon we are trying to capture is more subtle in the GK model compared to the KAtp model previously published. Future work will consist of administering these mice with lower dosages of the inducing agent tamoxifen. The immediate next steps will be to do a variety of tdTomato experiments with modulated dosages of tamoxifen administered to the mice in one, two, three, and five daily injection cycles and then quantify tdTom+ area in isolated islets. We will repeat and modify the experiments to get a broad more elaborate distribution of percent mutations so that when experimenting on those mice we will capture different stages of metabolic and electrical activity to have a more complete understanding of global behavior and islet dynamics.
67


Uniform GK Inhibition Effects on Global Behavior
While the GK deletion mouse model is generally a good model for GK mutations that occur in human disease, mosaic expression of GK knockout is unlikely to occur in humans. As a result, we modelled and experimentally investigated the effects of uniform GK deactivation on islet dynamics.
To computationally model uniform GK inhibition in islets, we modulated decreases in the rate of glycolysis or the rate of GK in all the cells. We then analyzed electrical activity by testing calcium dynamics in simulated islets with and without electrical coupling in the presence of the decreases in the GK rate. The computational model predicted a similar binary rapid decrease in activity in the presence of coupling (gcouP=120pS) and a more gradual decline in activity in the absence of coupling (gcouP=0pS) as had been seen in the GK deletion model (Figure 20). Such that in the presence of coupling, 65 percent reduction in GK activity is tolerated above which all the cells in the islet are shut off. In the absence of coupling, there isnt a linear decline in activity as had been seen before but there is a more gradual decline in activity compared to the dynamics in islets with wildtype coupling. Furthermore, see that prior to the critical threshold the coupled islet is actually more active than the uncoupled islet but after the threshold the coupled islet is actually less active than the uncoupled islet. And so, the cell-cell interactions with coupling compensates for the decreases in the GK rate at first but all activity is shut off after reaching the critical threshold. In the absence of coupling, there is a gradual decline starting at over 50 percent which just gradually declines until all the cells are shut off at 80 percent decrease in GK rate. The computational model
68


suggests a smaller window of recovery of about 15 percent, however it is there and can be taken advantage of.
To experimentally test this prediction, we induced uniform GK inhibition by treating the islets expressing wild type and knockout Cx36 with the GK inhibitor Mannoheptulose (MH) and characterizing metabolic and calcium dynamics in the islet. Calculating mean NAD(P)H content indicated that Cx36 knockout islets have higher NAD(P)H content at every treatment and fold change in NAD(P)H content concluded that they are less glucose responsive but the trend of decline in activity suggested that the knockouts are still somewhat responsive (Figure 21). Then we tested calcium dynamics of Cx36 wild type and knockout islets treated with the varying levels of MH and we found a trend of slight recovery in fraction active area (Figure 22). While there is no explicit and statistically significant difference between the Cx36 wild type and knockout for each treatment, the decreasing trend is more rapid for the Cx36 wild type islets whereas we see a more gradual trend of declining activity in the knockouts. These experiments showed us just how small the window of recovery really is, as islets are so heterogeneous in their responsiveness to the treatments which is exhibited in the figure by the large error bars especially pronounced at the predicted critical threshold that occurs between 3mM MH and 5mM MH treatments. Ideally, the removal of coupling would cause an increase in the glucose sensitivity of the heterogeneous population of P cells that has a higher excitability and is able respond more fully to the stimulus at treatments such as 5mM MH, however, we see that there is no significant average change between islets expressing gap junction coupling to islets expressing no coupling.
69


Future work will consist of repeating these experiments to further tighten the error bars especially at the 3mM and 5mM treatments so that the differences in function between batches can also be averaged out.
Effects of Parameters Altering GK Sensitivities
While the rate of GK function is most impacted by the rate of glycolysis, some of the additional parameters that affect the GK rate include the half maximal concentration of glucose and the half maximal concentration of ATP with their respective hill coefficients.
GK mutations have shown to have a variety of effects to GK sensitivities by primarily altering the activating threshold which can be achieved by modulating the half maximal concentration of glucose with its hill coefficient and the half maximal concentration of ATP. The results for simulating increases in the half maximal concentration of glucose indicated a similar binary behavior for a coupled islet where upon reaching a critical threshold of an over 1.75-fold increase in the concentration of glucose, the islet transitioned from a globally active to a globally inactive state (Figure 23a). On the other hand, in the absence of coupling there is a more gradual decline in activity and even after a two fold increase in the glucose concentration there is still about 10 percent of cells active in the islet. The half maximal concentration of ATP has a less pronounced effect on GK sensitivities. Simulating increases in ATP showed no effect on the distribution (data not shown). So we investigated how it contributed to the decline in activity in conjunction with an increase in the glucose threshold to 60 and 70 percent of its original value (Figure 23b). The results indicated that it did not contribute much, but it
70


did contribute to the decline in cellular excitability and only in the absence of coupling. This furthered our hypothesis that changes in cellular excitability can only be introduced to an uncoupled islet because it capitalizes on the excitability of a heterogeneous population of cells. Lastly, we simulated changes in the degree of cooperativity between glucose and GK by altering the hill coefficient and examined islet dynamics. Our results concluded that in the absence of coupling, there was really no change in the decline in cellular excitability (Figure 23d). Whereas in a coupled islet, a 50 percent reduction in the hill coefficient correlated with a ten percent shift to the right (Figure 23c). And so there was a change, but not a significant one. These simulations exploring parameters affecting GK sensitivities indicated that the two most important parameters in determining the effects of GK activity on islet function include the rate of glycolysis or GK reaction rate and the half maximal concentration of glucose.
Calcium Dynamics of GK Mutations
Having a thorough understanding of the parameters important in determining GK and islet function is critical to comprehending the calcium dynamics of GK mutations occurring in humans. The kinetics of GK mutations reported in literature indicated a range of values. This is primarily because we found only a few homozygous GK mutations resulting in PNDM and a large portion of the literature found reported heterozygous GK mutations resulting in MODY-2. Simulating the mutations in islets with electrical coupling, we found that most of the mutations did not alter calcium dynamics when compared to a wild type control (Figure 24). However, a fairly significant number of mutations predicted the islet was completely electrically inactive.
71


To test how these dynamics would change by removing cell-cell interactions or uncoupling the islet, we found that there was generally no change in activity with a few islets predicting lower number of fraction cells active and recovered activity in two cells that were otherwise completely inactive at high levels of glucose.
The literature indicated that naturally occurring mutations in the human GK gene resulted in primarily MODY-2 or PNDM. The MODY-2 mutations most commonly resulted in a phenotype of mild diabetes, moderate hyperglycemia, and impaired insulin secretion. However, when you look at the calcium dynamics of the mutations, notice that there are still a significant number of mutations with no calcium activity (Figure 24). And so, we thought it was important to better understand the effects of GK mutations by putting them into the perspective of their clinical severity. When looking for literature citing both the kinetics of GK mutations and clinical data, we found we could rank mutations by the results of the patients oral glucose tolerance tests (OGTT) and the percentage of glycated hemoglobin (HbAlc). When we ranked by OGTT results, we found that the distribution can be split into three primary stages identified as mild, mild to severe, and severe (Figure 25). The results of the simulations of the mutations indicated that the mild mutations had mostly unaltered calcium dynamics, the mild to severe having more mutations that are inactivating, and the severe cases exhibiting mostly no calcium dynamics or are inactive. Upon simulating no cell-cell interaction by removing gap junction conductance, we found that we were able to recover function in the most severely ranked GK mutation. When we ranked by HbAlc percentages, we saw a similar three phase separation of the distribution (Figure 26). Simulating the mutations however indicated that upon removing gap junction coupling, activity could only be rescued in the
72


mutations with more mild HbAlcs with slight recoveries in the other scorings. These findings further supported the hypothesis that there is a window of recovery of electrical activity by removing coupling, however, the window is very narrow in space as it can only be recovered in a small population of mutations and are time-sensitive as the range of the effects of the mutation cant be too vast.
Implications for Diabetes
Islet studies is the heart of diabetes research. And this thesis has been aimed at understanding how the islet functions as a network of P cells in non-stimulatory conditions by examining decreases in GK activity by percent deletion and inhibition. We explored different mechanisms by which changes in the net activity in GK can perturb glucose homeostasis and how that manifests itself in metabolic activity by measuring NAD(P)H content and electrical activity by measuring and simulating calcium dynamics.
The ability of cells to communicate with one another by gap junction coupling in the islet is critical for the islet to respond robustly to a stimulus in healthy conditions, however, we have shown that it has the capacity to be detrimental in the presence of suppressive conditions. Our simulations of the effects of a subpopulation of inactive cells exhibiting GK mutations on islet dynamics showed an emergence of critical behavior which was alleviated with a more gradual decline in activity upon reducing gap junction coupling. The experimental results provided us with the understanding that if a GK-mediated disease progressed too far, uncoupling the islet would not recover further electrical activity as the cells will not have a capacity to be electrically excitable. Then, we looked at the effects of decreased cellular excitability on islet dynamics by simulating
73


and measuring calcium dynamics. GK inhibition in islets with wild type gap junction conductance resulted in a more rapid decline in activity, which was more gradual in the islets with no coupling. Our results allowed us to conclude that the window of recovery of electrical activity by uncoupling the islet for GK-mediated disease was very small and time sensitive. These principals were consistent with the findings from simulating naturally occurring GK mutations and testing whether recovery could be achieved by uncoupling the islet. We found we could recover activity in mutations with high OGTT scoring and mild HbAlc.
Diabetic patients do exhibit significant reductions in glucose stimulated insulin secretion and insulin therapy has been revolutionary in treating patients with diabetes. But perhaps it isnt necessary to substitute insulin delivery completely in all patients, eliminating the need for the islet to generate endogenous insulin by compensatory mechanisms. Perhaps we can provide a more patient specific treatment that addresses the specific limitations of the capacity of the patients body in secreting insulin in response to elevated glucose. Based on the conclusions of this thesis, we propose exploring treatment options that include reducing gap junction coupling in practice along with sulfonylureas. Our results have indicated that the window of recovery of islet function after GK deletion or inhibition by uncoupling the islet is very small and limited to when the disease has not severely damaged islet function, and so it will not be a sufficient therapeutic option on its own. Therefore, we propose continuing the use of current therapeutics such as sulfonylureas. But the efficacy of administering sulfonylureas has the capacity to be greater by uncoupling the islet and allowing the heterogeneity in the population of cells in the islet to drive islet function. By removing cell-cell interactions
74


by gap junction coupling, the intrinsic heterogeneity in the responsiveness of the cells in the islet will dominate allowing the range of treatment options to be expanded and more effective. Our results suggest that uncoupling the system of tightly coupled cells could recover heterogeneity and reaffirm the capacity of a subpopulation of cells to respond to a stimulus.
The principals discussed in this thesis dont just apply to islets, but can be applied to any stimulus-secretion coupled cells operating in a network. The pancreatic P cell is an outstanding example for a glucose-sensing cell and can be considered as a model by which other glucose-responsive cell types including hepatocytes, gut enterocytes, and certain hypothalamic neurons or other regions of the brain can be investigated. The P cell attributes its glucose-sensing capacity in large part to glucokinase, which acts as the glucose sensor of the cell and has also been found in the network of glucose sensing cell types. The findings from this research can be used to uncover novel therapeutic options not only for diabetes, but other glucose sensing deficient mechanisms of disease progression.
75


REFERENCES
Ainscow, E. and Rutter, G. (2002). Glucose-Stimulated Oscillations in Free Cytosolic ATP Concentration Imaged in Single Islet P-Cells: Evidence for a Ca -Dependent Mechanism. Diabetes 51, S162-S170.
American Diabetes Association. Microvascular and Macrovascular complications of Diabetes.
American Diabetes Association. (2010). Diagnosis and classification of diabetes mellitus. Diabetes care 33.Supplement 1, S62-S69.
American Diabetes Association. (2009). Diagnosis and classification of diabetes mellitus. Diabetes care 32.Supplement 1, S62-S67.
American Diabetes Association. (2005). Diagnosis and classification of diabetes mellitus. Diabetes care 28, S37.
American Diabetes Association. (2007). Diagnosis and classification of diabetes mellitus. Diabetes care 28, S37.
Ashcroft FM, Rorsman P. Electrophysiology of the pancreatic P-cell. Progress in Biophysics and Molecular Biology. 1989; 54:87-143.
Ashcroft SJ, Sugden MC, Williams IH (1980). Carbohydrate metabolism and the glucoreceptor mechanism. Horm Metab Res (Suppl 10): 1-7.
Benninger RKP, Head WS, Zhang M, Satin LS, Piston DW. (2011). Gap junctions and other mechanisms of cell-cell communication regulate basal insulin secretion in the pancreatic islet. The Journal of Physiology 589, 5453-5466.
Benninger RKP, Zhang M, Head WS, Satin LS, Piston DW. (2008). Gap Junction Coupling and Calcium Waves in the Pancreatic Islet. Biophysical Journal 95, 5048-5061.
Byrne, M.M., Sturis, J., Clement, K., Vionnet, N., Pueyo, M.E., Stoffel, M., Takeda, J., Passa, P., Cohen, D., Bell, G.I. (1994). Insulin Secretory Abnormalities in Subjects with Hyperglycemia due to Glucokinase Mutations. Journal of Clinical Investigation 93.3, 1120-1130.
Calabrese A, Zhang M, Serre-Beinier V, Caton D, Mas C, Satin LS, Meda P. (2003). Connexin 36 Controls Synchronization of Ca2+ Oscillations and Insulin Secretion in MIN6 Cells. Diabetes, 52:417-424.
Carboneau, B. A., Le, T. D. V., Dunn, J. C., & Gannon, M. (2016). Unexpected effects of
76


the MIP- CreER transgene and tamoxifen on ft- cell growth in C57B16/J male mice. Physiological Reports, 7(18), el2863.
Cha, C. Y., Nakamura, Y., Himeno, Y., Wang, J., Fujimoto, S., Inagaki, N., ... Noma, A.
2+
(2011). Ionic mechanisms and Ca dynamics underlying the glucose response of pancreatic P cells: a simulation study. The Journal of General Physiology, 755(1), 21-37.
Chay, T. R., & Keizer, J. (1983). Minimal model for membrane oscillations in the pancreatic beta-cell. Biophysical Journal, 72(2), 181-190.
Chay,T.R. (1997). Effects of extracellular calcium on electrical bursting and intracellular and luminal calcium oscillations in insulin secreting pancreatic P-cells. Biophys. J. 73 1673-1688.
Columbia University Department of Surgery. The Pancreas and its Functions.
Curry DL, Bennett LL, Grodsky GM. (1968). Requirement for calcium ion in insulin secretion by the perfused rat pancreas. Am J Physiol 214: 174-178.
Davis, E.A., Cuesta-Munoz, A., Raoul, M, Buettger, C., Sweet, I., Moates, M., Magnuson, M.A., Matschinsky, F.M. (1999) Mutants of glucokinase cause hypoglycaemia- and hyperglycaemia syndromes and their analysis illuminates fundamental quantitative concepts of glucose homeostasis. Diabetes 42, 1175-1186.
Duchen MR, Smith PA, Ashcroft FM. (1993) Substrate-dependent changes in mitochondrial function, intracellular free calcium concentration and membrane channels in pancreatic beta-cells. Biochem J. 294: 35-42.
Eddlestone, G. T., Goncalves, A., Bangham, J. A., & Rojas, E. (1984). Electrical coupling between cells in islets of Langerhans from mouse. The Journal of membrane biology, 77(1), 1-14.
Erecinska, M., Bryla, J., Michalik, M., Meglasson, M. D., Nelson, D. (1992). Energy metabolism in islets of Langerhans. Biochimica et Biophysica Acta 1101, 273-295.
Fajans, S.S., Bell, G.I., Polonsky, K.S. (2001). Molecular mechanisms and clinical pathophysiology of maturity-onset diabetes of the young. N Eng J Med 345, 971-980.
Farnsworth, Nikki L., and Richard K.P. Benninger. New Insights into the Role of Connexins in Pancreatic Islet Function and Diabetes. FEBS letters 588.8 (2014): 1278-1287. PMC. Web. 5 July 2017.
Gepts, W. (1965). Pathologic anatomy of the pancreas in juvenile diabetes mellitus. Diabetes, 14(10), 619-633.
77


Gilon, P., & Henquin, J. C. (1992). Influence of membrane potential changes on cytoplasmic Ca2+ concentration in an electrically excitable cell, the insulin-secreting pancreatic B-cell. Journal of Biological Chemistry, 267(29), 20713-20720.
Gloyn, A. L., Siddiqui, J. and Ellard, S. (2006). Mutations in the genes encoding the pancreatic beta-cell KATP channel subunits Kir6.2 (KCNJ11) and SUR1 (ABCC8) in diabetes mellitus and hyperinsulinism. Hum. Mutat., 27: 220-231.
Grimsby J, Sarabu R, Corbett WL, Haynes NE, Bizzarro FT, Coffey JW, Guertin KR, Hilliard DW, Kester RF, Mahaney PE, Marcus L, Qi L, Spence CL, Tengi J, Magnuson MA, Chu CA, Dvorozniak MT, Matschinsky FM, Grippo JF. (2003). Allosteric activators of glucokinase: potential role in diabetes therapy. Science 301, 370-373.
Grossbard L., Schimke R.T. (1966). Multiple hexokinases of rat tissues. Purification and comparison of soluble forms. J Biol Chem 241, 3546-3560.
Grupe, A., Hultgren, B., Ryan, A., Ma, Y.H. Bauer, M., Stewart, T.A. (1995). Transgenic knockouts reveal a critical requirement for pancreatic beta cell glucokinase in maintaining glucose homeostasis. Cell 83, 69-78.
Halban, P. A., Wollheim, C. B., Blondel, B., Meda, P., NIESOR, E. N., & MINTZ, D. H. (1982). The possible importance of contact between pancreatic islet cells for the control of insulin release. Endocrinology, 777(1), 86-94.
Hattersley, A. T., and Ashcroft, F. M. (2005). Activating Mutations in Kir6.2 and Neonatal Diabetes: New Clinical Syndromes, New Scientific Insights, and New Therapy. Diabetes 54(9): 2503-2513.
Henquin, J.-C. 1990. Glucose-induced electrical activity in 83-cells. Diabetes. 39:1457-1460.
Holt, R. and Hanley, N. (2013). Endocrinology and Diabetes. West Sussex: John Wiley & Sons.
Hraha, T. H., Westacott, M. J., Pozzoli, M., Notary, A. M., McClatchey, P. M., & Benninger, R. K. P. (2014). Phase Transitions in the Multi-cellular Regulatory Behavior of Pancreatic Islet Excitability. PLoS Computational Biology, 10(9), el003819.
Johnson JH, Newgard CB, Milburn JL, Lodish HF, Thorens B. The high Km glucose transporter of islets of Langerhans is functionally similar to the low affinity transporter of liver and has an identical primary sequence. J Biol Chem. 1990;265:6548-6551.
Kanno T, Rorsman P, Gopel SO (2002) Glucose-dependent regulation of rhythmic action potential firing in pancreatic P cells by KATP-channel modulation. J Physiol, 545, 501 507.
78


Kennedy R, Kauri L, Dahlgren G, Jung S. (2007) Metabolic Oscillations in Beta-cells. Diabetes 51, S263-S161.
Kim, Abraham et al. Islet Architecture: A Comparative Study. Islets 1.2 (2009): 129-136.
LeRoith, D. (2002). P-cell dysfunction and insulin resistance in type 2 diabetes: role of metabolic and genetic abnormalities. The American journal of medicine ,713(6), 3-11.
MacDonald, P. E., Joseph, J. W., and Rorsman, P. (2005) Glucose-Sensing Mechanisms in Pancreatic B-Cells. Philosophical Transactions of the Royal Society B: Biological Sciences. 360.1464, 2211-2225.
MacDonald PE, Rorsman P (2006) Oscillations, Intercellular Coupling, and Insulin Secretion in Pancreatic P Cells. PLoS Biol 4(2): e49.
Maechler P, Kennedy ED, Pozzan T, Wollheim CB. (1997). Mitochondrial activation directly triggers the exocytosis of insulin in permeabilized pancreatic beta-cells. Embo J 16(13), 3833-41.
Matschinsky, F.M. (2002). Regulation of pancreatic beta-cell glucokinase: from basics to therapeutics. Diabetes 51, S394-S404.
Matschinsky, F.M. and Magnuson, M.A. (2004). Glucokinase and Glycemic Disease: From Basics to Novel Therapeutics. Front Diabetes. Basel, Karger, 2004, vol 16.
Matschinsky, F.M., Ellerman, J.E. (1968). Metabolism of glucose in the islets of Langerhans. J Biol Chem 243, 2730-2736.
Merrins, M. J., Poudel, C., McKenna, J. P., Ha, J., Sherman, A., Bertram, R., & Satin, L. S. (2016). Phase Analysis of Metabolic Oscillations and Membrane Potential in Pancreatic Islet /?-Cells. Biophysical Journal, 770(3), 691-699.
Moreno-Sanchez, R. (1985). Regulation of oxidative phosphorylation in mitochondria by external free Ca2+ concentrations. J Biol Chem. 260(7), 4028-34.
Nguyen, Linda M. et al. Decreasing Cx36 Gap Junction Coupling Compensates for Overactive KAtp Channels to Restore Insulin Secretion and Prevent Hyperglycemia in a Mouse Model of Neonatal Diabetes. Diabetes 63.5 (2014): 1685-1697. EMC. Web. 5 July 2017.
Ni Q., Ganesan A., Aye-Han N. N., Gao X., Allen M. D., Levchenko A., et al. (2011). Signaling diversity of PKA achieved via a Ca2+-cAMP-PKA oscillatory circuit. Nat. Chem. Biol. 7 34-40.
79


Niswender, K. D., Shiota, M., Postic, C., Cherrington, A. D., and Magnuson, M. A. (1997). Cell-specific Expression and Regulation of a Glucokinase Gene Locus Transgene. J. Biol. Chem. 272, 22604-22609
Niswender, K.D., Postic C., Jetton, T.L., Bennett, B.D., Piston, D.W. Efrat S., Magnuson, M.A. (1997). Cell- specific expression and regulation of a glucokinase gene locus transgene. J Biol Chem 272, 22570-22575.
Njolstad PR, Sovik O, Cuesta-Munoz A, Bjorkhaug L, Massa O, Barbetti F, Undlien DE, Shiota C, Magnuson MA, Molven A, Matschinsky FM, Bell GI. (2001). Neonatal diabetes mellitus due to complete glucokinase deficiency. N Eng J Med 344, 1588-1592.
Njolstad PR, Sagen JV, Bjorkhaug L, Odili S, Shehadeh N, Bakry D, Sarici SU, Alpay F, Moines J, Molven A, Sovik O, Matschinsky FM. (2003). Permanent neonatal diabetes caused by glucokinase deficiency: inborn error of the glucose-insulin signaling pathway. Diabetes 52(11), 2854-2860.
Notary, A. M., Westacott, M. J., Hraha, T. FL, Pozzoli, M., & Benninger, R. K. P. (2016). Decreases in Gap Junction Coupling Recovers Ca2+ and Insulin Secretion in Neonatal Diabetes Mellitus, Dependent on Beta Cell Heterogeneity and Noise. PLoS Computational Biology, 12(9), el005116.
Patel, N., & Gold, M. G. (2015). The genetically encoded tool set for investigating cAMP: more than the sum of its parts. Frontiers in Pharmacology, 6, 164.
Pearson, E.R., Velho, G., Clark, P., Stride, A., Shepherd, M., Frayling, T.M., Bulman, M.P., Ellard, S., Froguel, P., Hattersley, A.T.(2001). Beta-cell genes and diabetes: quantitative and qualitative differences in the pathophysiology of hepatic nuclear factor -1 alpha and glucokinase mutations. Diabetes 50, S101-S107.
Pilkis, S.J., Hansen, R.J., Krahl, M.E. (1968). Hexose-ATP phosphor transferases: Comparative aspects. Comp Biochem Phys 25, 681-684.
Pilkis, S.J. (1968). Identification of human hepatic glucokinase and some properties of the enzyme. Proc Soc Exp Biol Med 129, 681-684.
Pilkis, S.J., Hansen, R.J., Krahl, M.E. (1968). Hormonal control of hexokinase in animal tissues. Diabetes 17, 300-301.
Porksen, N. (2002). The in vivo regulation of pulsatile insulin secretion.
Diabetologia, 45(1), 3-20.
80


Pueyo, M.E., Clement, K., Vaxillaire, M., Passa P., Froguel P., Robert, J.J., Velho, G. (1994). Arginine-induced insulin release in glucokinase-deficient subjects. Diab Care 14, 1015-1021.
Ravier MA, Guldenagel M, Charollais A, Gjinovci A, Caille D, Sohl G, Wollheim CB, Willecke K, Henquin J-C, Meda P. Loss of Connexin36 Channels Alters P-Cell Coupling, Islet Synchronization of Glucose-Induced Ca2+ and Insulin Oscillations, and Basal Insulin Release. Diabetes. 2005; 54:1798-1807.
Rocheleau, J. V., Remedi, M. S., Granada, B., Head, W. S., Koster, J. C., Nichols, C. G., & Piston, D. W. (2006). Critical Role of Gap Junction Coupled KAtp Channel Activity for Regulated Insulin Secretion. PLoSBiology, 4(2), e26.
Sarici, S.U., Alpay, F., Diindaryz, M.R., Gyknay, E. 2001. Neonatal diabetes mellitus: patient report and review of the literature. J Pediatr Endocrinol Metab 14, 451-454.
Sharma, C, Manjeshwar, R., Weinhouse S. (1964). Hormonal and dietary regulation of hepatic glucokinase. Adv Enzyme Regul 2, 189-200.
Shehadeh N1, Gershoni-Baruch R, Mandel H, Nutenko I, Etzioni A. (1996). Congenital permanent diabetes: a different type of diabetes? Acta Paediatr 85(12), 1415-1417.
Sherman, A. (1996). Contributions of modeling to understanding stimulus-secretion coupling in pancreatic beta-cells. American Journal of Physiology Endocrinology and Metabolism 271, E362-E372.
Sherman, A., and Rinzel, J. (1991). Model for synchronization of pancreatic beta-cells by gap junction coupling. Biophys J 59, 547-559.
Sols A., Salas M., Vineula E. 1964. Induced Biosynthesis of liver glucokinase. Adv Enzyme Regul 2, 177-178.
Sreenan, S.K., Cockburn, B.N., Baldwin, A.C., Ostrega, D.M., Levisetti, M., Grupe, A., Bell, G.I., Stewart, T.A., Roe, M.W., Polonsky, K.S. (1998). Adaptation to hyperglycemia enhances insulin secretion in glucokinase mutant mice. Diabetes 47(12), 1881-1888.
Stamper, I.J., Jackson, E., Wang, X. (2014). Phase transitions in pancreatic islet cellular networks and implications for type-1 diabetes. Phys Rev E 89, 012719
Stride, A., Vaxillaire, M., Tuomi, T., Barbetti, F., Njolstad, P.R., Hansen, T., Costa, A, Conget, I., Pedersen, O., Sovik, O., Lorini, R., Groop, L., Froguel, P., Hattersley, A.T. (2002). The genetic abnormality in the beta cell determines the response to an oral glucose load. Diabetologia 45(3), 427-435.
81


Suckale, J. and Solimena, M. (2008). Pancreas islets in metabolic signaling focus on the P-cell. Frontiers in Bioscience 13, 7156-717.
Vetere, A., Choudhary, A., Burns, S.M., Wagner, B.K. (2014). Targeting the pancreatic P-cell to treat diabetes. Nature Reviews Drug Discovery 13, 278-289.
Walker, D.G, Rao S. (1964). The role of glucokinase in the phosphorylation of glucose by rat liver. Biochem J. 90, 360-368.
World Health Organization. Global Report on Diabetes. 6 April 2016.
Zhang M, Goforth P, Bertram R, Sherman A, Satin L. The Ca2+ dynamics of isolated mouse betacells and islets: implications for mathematical models. Biophysical Journal. 84, 2852-2870.
82


Full Text

PAGE 1

GLUCOKINASE REGULATION OF MUTLICELLULAR ISLET ELECTRICAL ACTIVITY by NURIN LUDIN B.S. University of Denver, 2014 A thesis submitted to the faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements fo r the degree of Master of Science Bioengineering 2017

PAGE 2

ii 2017 NURIN LUDIN ALL RIGHTS RESERVED

PAGE 3

iii This thesis for the Master of Science degree by Nurin Ludin h as been approved for the Bioengineering program by Richard Benninger, Chai r Emily Gibson Vitaly Kheyfets Date: December 15 2017

PAGE 4

iv Ludin, Nurin Glucokinase Regulation of Multicellular Islet Electrical Activity Thesis directed by Associate Professor Richard Benninger ABSTRACT Diabetes is caused by dysfunction or death of ) cells in the islet of Langerhans multicellular micro organs in the pancreas that are important for glucose regulation This result s in poorly regulated glucose homeostasis due to defective do not act autonomously, they are ele ctrically coupled together by gap junction s As a result all the cells in the electrically continuous network are quiescent in the absence of glucose and secrete insulin in response to elevated glucose. Previously, others have experimentally shown and simu them inexcitable by introducing an inducible K ATP mutation modeling neonatal diabetes mellitus. It has been shown that in the presence of coupling if the number of inexcitable cells in the network ex ceeds 15% there is an emergence of critical behavior where all the cells in the islet transition rapidly from being globally active to inactive and there occurs a nearly complete suppression of insulin secretion. In the absence of coupling it was shown th at the critical behavior was diminished, glucose homeostasis was rescued, and electrical activity in the islet withstood despite being subjected to silencing conditions. This suggested uncoupling the islet as a novel therapeutic option for treating neonata l diabetes mellitus and potentially other monogenic forms of diabetes. Glucokinase is a more upstream regulator of electrical activity and mutations to glucokinase results in diabetes. We explore emergent critical behavior further by generating an inducibl e

PAGE 5

v mutation to glucokinase We will examine how mutations to glucokinase decrease average excitability and have a disproportionately suppressive effect when coupled. Our simulations predict that decreases in glucokinase activity results in an emergence of c ritical behavior and the critical behavior is alleviated when gap junction coupling is removed. The experimental findings conclude that there is potential to recover islet function despite severe glucokinase mutations and inhibition. The results of this in vestigation contribute to the characterization of furthering our understanding of emerging critical behavior in cellular networks such as the islet in the hopes of discovering more patient specified and effective treatment methods for glucokinase dysfuncti on mediated neonatal diabetes mellitus and m aturity o nset d iabetes of the y oung type 2. The form and content of this abstract are approved. I recommend its publication. Approved: Richard Benninger

PAGE 6

vi ACKNOWLEDGEMENTS This work was possible with the help of a number of exceptional people. I thank Richard Benninger for his guidance and support during the completion of this work and committee members Emily Gibson and Vitaly Kheyfets for providing valuable feedback in finishing this work. Addit ionally, I would like to thank all the current and previous members of the Benninger Lab. Especially, Matt Westacott, Aleena Notary, Nikki Farnsworth, Audrey Heintz, and Jenn Dwulet for their assistance on both the computational and experimental aspects of my thesis. Finally, I would like to thank my family and friends for their support over the course of the program. A special thanks to my parents Farah and Rafaat Ludin without whom this would not be possible. Thank you to the Barbra Davis Center islet is olation core, Advanced Light Microscopy Core at the University of Colorado Denver, and University of Colorado at Boulder for access to the JANUS supercomputer cluster. Funding for this rese arch was providing by the grant Emergent Multicell ul ar Properties Regulating Pancreatic Islet Function (R01 DK106412 01). Mouse studies were performed in according to the guidelines of IACUC protocol: B 95814(07)1D.

PAGE 7

vii TABLE OF CONTENTS CHAPTER I. INTRODUCTION .. Diabetes: Definitions and Motivat ....1 .. ..... ..11 ... ... ..25 .. ..26 II. ..28 ...28 .35 III. GK( / NAD(P)H Imaging of Islets Isolated from GK( / ) Mice to Characterize Metabolic Activity ... ..... ........................................ 47 Calcium Imaging of Islets Isolated from GK( / ) Mice tdTomato Quantification Identifies Percent Mutation Higher Than Expected

PAGE 8

viii Decrea ...51 NAD(P)H Imaging of Cx36(+/+) and Cx36( / ) Islets treated with GK i nhibitor to Characterize Metabolic Activity ..53 Calcium Imaging of Cx 36(+/+) and Cx36( / ) Islets treate d with GK i nhibitor to Characterize Calcium D 54 Clinical Mutations in Glucokinase Examined ...55 IV. DISCUSSION AND FUTURE WORK ..63 Effe ...70 ..71 ...73 .... 7 6

PAGE 9

1 CHAPTER I INTRODU CTION AND BACKGROUND Diabetes: Definition and Motivation Diabetes is a complex metabolic disorder characterized by higher than normal blood glucose levels or hyperglycemia resulting from defects in insulin secretion, insulin action, or a combination of b oth. A diagnosis of diabetes is made if the glycated hemoglobin (Hb A1c ) is 6.5 percent and higher, fasting plasma glucose level is 126 mg/dL and higher or a random or two hour plasma glucose level is 200 mg/dL and higher ( Holt and Hanley 2013). Diabetes i s a global epidemic affecting over 400 million people worldwide and the number is only projected to grow, rising more rapidly in middle and low income countries (World Health Organization 2016 Holt and Hanley 2013 ). Modern medical care uses lifestyle an d pharmaceutical interventions to prevent and control hyperglycemia, so that there is an adequate delivery of glucose to the tissues of the body and the tissues are not harmed by hyperglycemia (Fowler 2008). Hyperglycemia occurs as a result of defects in insulin secretion, insulin action, or a combination of both (Diabetes Care 2010, Diabetes Care 2009, Diabetes Care 2004). The complications associated with being in a state of chronic hyperglycemia include microvascular complications (such as retinopathy, peripheral neuropathy, and diabetic nephropathy) and macrovascular complications (such as coronary artery disease, peripheral arterial disease, and stroke).

PAGE 10

2 The total estimated cost of diagnosed diabetes in 2012 was $254 billion consisting of $176 billion in medical costs and $69 billion in reduced national productivity in 2007 (Diabetes Care 2012). The significant components of medical expenditures include hospital inpatient care (43 percent of total medical cost), prescription medications to treat the co mplications associated with diabetes (18 percent), antidiabetic agents and supplies (12 percent), physician office visits (9 percent), and facility stays (8 percent) (Diabetes Care 2012). And of the cost categories, care for people with diagnosed diabetes accounts for more than 1 in 5 health care dollars in the United States (Diabetes Care 2012, Diabetes Care 2007). The estimated total economic cost of diagnosed diabetes in 2012 increased a total of 41 percent from the previous estimate made in 2007. Theref ore the burden that diabetes imposes on society is substantial, but it is projected to grow even more rapidly (Diabetes Care 2012, Diabetes Care 2007). Furthermore, there are additional burdens that can not be quantified and they include pain and suffering experienced by the patients resources supplied by nonpaid caregivers, and the burdens associated with undiagnosed diabetes. Type I and Type II Diabetes Permanent Neonatal Diabetes Mellitus, and Maturity Onset Diabetes of the Young The vast majority of cases of diabetes fall into two broad categories identified as type I and type II diabetes. Typically type I diabetes accounts for between 5 to 10 percent of total patients with diabetes, whereas type II diabetes accounts for about 90 to 95 percent of tot al cases. Type I diabetes occurs as a result of absolute deficiency of insulin secretion due to autoimmune destruction of Beta cell destruction and type II occurs as

PAGE 11

3 result of resistance to insulin action and inadequate compensatory insulin secretion resp onse (Diabetes Care 2007). Loss of Beta cell number and function underlies much of the pathology of diabetes. In type I diabetes, the immune system recognizes the beta cell as foreign, probably due to a combination of genetic and environmental factors. Th e insulin resistance that occurs in type II diabetes involves an inability of cells in for example liver, muscle, and adipose tissue to respond to the normal actions of insulin and in order to compensate for this resistance, the pancreatic Beta cells incre ase their production of insulin while declining in function and insulin production eventually be comes inadequate (Amedeo Vetere et al. 2014). Approximately between 2 to 5 percent of cases of diabetes are associated with monogenetic defects in Beta cell f unction that induce an onset of hyperglycemia at an early age are associated with impaired insulin secretion (Holt and Hanley 2013) The most common types of monogenic forms of diabetes include neonatal diabetes and maturity onset diabetes of the young (Na tional Institute of Diabetes and Digestive and Kidney Diseases 2014). Neonatal diabetes can either be transient where patients resolve in the first 18 months of life or permanent where they have a life long insulin dependence to regulate blood glucose leve ls Roughly 50 percent of NDM cases occur as a result of mutations to the KCNJ11 or ABCC8 genes which encode for the Kir6.2 and SUR1 subunits on the K ATP channel ( Gloyn et al. 2006 Hattersley 2005). PNDM can also result from a complete deficiency of the cell transcription factor insulin promoter factor 1 (IPF1 also known as PDX1) and the enzyme glucokinase (Matschinsky, F.M. and Magnuson, M.A. 2004 National Institute of Diabetes and Digestive and Kidney Diseases 2014 ) Mutations in six genes cause most of the MODY cases. This includes mutations to

PAGE 12

4 genes encoding the enzyme glucokinase (MODY 2) and the transcription factors: hepatocyte nuclear factor 4 (MODY 1), HNF 1 (MODY 3), insulin promoter factor 1 (MODY 4), HNF 1 (MODY 5), and neurogenic differ entiation factor 1 (MODY 6) (Matschinsky, F.M. and Magnuson 2004 National Institute of Diabetes and Digestive and Kidney Diseases 2014 ) Beta Cell Function in the Islet of Langerhans Physiology of the pancreas The pancreas has two main functions: to pr oduce enzymes for digestion (exocrine) and make hormones (endocrine). Most of the pancreatic volume consists of exocrine tissue (MacDonald and Rorsman 2006). The stomach empties partially digested food into the intestine and the exocrine component of the p ancreas releases digestive enzymes such as tryspin and chymotrypsin into the contents to digest proteins, amalyse to digest the carbohydrates, and lipase to break down the fats (Columbia University Department of Surgery). The remaining tissue consists of e ndocrine cells identified as the islets of Langerhans, which are multi functional micro organs that are dispersed throughout the pancreatic exocrine tissue and are highly vascularized (MacDonald and Rorsman 2006).

PAGE 13

5 Structure and Function of the Islets of L angerhans The islets of Langerhans are multi cellular micro organs located in the pancreas The three main cell types of the pancreatic islet include alpha cells, delta cells, an d beta cells. T he islets of Langerhans make up only 2 percent of the pancreatic tissue and secrete the hormones insulin via beta cells (constitutes 70 percent of total islet cells), glucagon via alpha cells (constitutes 20 percent of total islet cells), an d somatostatin via delta cells (constitutes less than 10 percent of total islet cells). It has been recognized that the cytoarchitecture of pancreatic islets vary between different species. The organization of the different hormone secreting cell types a re distinctly different between human and mouse islets. However, despite the differences in overall body and pancreas size as well as total Beta cell mass among the species, the distribution of their islet sizes closely overlap and they share common archit ectural features that may correlate to similar demands for insulin (Abraham Kim et al 2009, Suckale and Solimena 2008 ). Figure 1 : The cytoarchitecture of the islet and the relative abundance of the various hormones secreted by the islet in rat models (top) and human models (below).

PAGE 14

6 It is well recognized that there is a close correlation between body and pancreas weight. It has been shown that total Beta cell mass increases proportionately in order to compensate for the demand for insulin in the body within a certain limit of islet size where the overall changes in structure reflects the adaptation of the islet in response to increase d body demands rather than difference s amongst species (Abraham Kim et al. 2009). The islets of Langerhans function to secrete the following hormones: beta cells secrete insulin, alpha cells secrete glucagon, delta cells secrete somatostatin, PP cells sec rete pancreatic polypeptide, and epsilon cells secrete ghrelin. Insulin and glucagon are critical to regulation of blood glucose. The two hormones have antagonistic actions in order to maintain homeostasis of the blood glucose concentration, where insulin mediates the cellular uptake of blood glucose into skeletal muscle and the live after a meal and glucagon mediates the hydrolysis of liver glycogen between meals so that the liver can secrete glucose into the blood (Ishihara et al. 2003, Matschinsky and Ma gnuson 2004 ). It has also been showed that while both alpha and beta cells possess the capacity to respond to nutrients, secretions from alpha cells is normally suppressed by simultaneous activation of beta cells (Ishihara et al. 2003). Therefore, insulin lowers blood glucose and glucagon raises blood glucose levels.

PAGE 15

7 Insulin Action Insulin acts on a target cell by binding to the insulin receptor on the cell surface. The insulin receptor is characterized as a heterotetramer wit h two glycoprotein and two glycoprotein subunits linked together by dis ulphide bonds. Insulin binds to the extracellular receptor resulting in auto phosphorylation of insulin responsive substrates, which then bind to other signaling molecules that mediate subsequent cellular actions o f insulin (Holt and Hanley 2013). Insulin is the hormone that directs anabolic processes in intermediate metabolism. Insulin action has major effects on glucose, lipid, and protein metabolism. The tissues that are most sensitive to insulin effects include the liver, skeletal muscle, and adipose tissue. After the secretion of insulin, 60 percent of the insulin content is removed by the Figure 2 : The antagonistic act ions of insulin and glucagon in order to reestablish blood glucose homeostasis.

PAGE 16

8 liver where it plays an important in regulating hepatic glucose output by inhibiting gluconeogenesis and promoting the stora ge of glucose 6 phosphate to glycogen. In muscle cells, insulin mediate glucose uptake promotes glycogen storage and for carbohydrates to be used as the primary energy source for muscle contraction instead of fatty acids or amino acids. In adipose tissue, fat break down is inhibited and the synthesis of fat is promoted through the formation and storage of triglycerides (Holt and Hanley 2013) Glucose Stimulated Insulin Secretion and Beta Cell Electrophysiology cells respond to changes in glucose levels by secreting insulin directly into the circulatory system where it travels to target cells in the body (MacDonald and Rorsman 2006 Suckale and Solimena 2008 ). The process of insulin secretion is more specifical ly known as glucose stimulated insulin secretion or GSIS (MacDonald and Rorsman 2006). Following the ingestion of a meal, there is a rise in blood glucose concentrations and the cells sense the increase in glucose. Glucose enters cells through the GLUT 2 glucose transporter. Glucose is then phosphorylated by glucokinase. The phosphorylated glucose is then consumed by glycolysis. The products of glycolysis then enter the tricarboxylic acid (TCA) cycle and generate additional ATP. The metabolic pathways re sult in a net increase in intracellular [ATP] to [ADP] ratio, which then closes the ATP sensitive K + ( K ATP ) channel resulting in membrane depolarization and subsequent opening of the voltage gated Ca 2+ channels that result in an increase in intracellular C a 2+ concentration ([Ca 2+ ] i ) which then acts on the exocytotic machinery to stimulate fusion of

PAGE 17

9 insulin containing granules with the plasma membrane for secretion into the bloodstream (Robert Kennedy et al 2007 ). It is well known that [Ca 2+ ] i oscillates in islets in response to incre ases in glucose (Robert Kennedy et al 2007, MacDonald and Rorsman 2006). And removal of extracellular Ca 2+ prevents the firing of an action potential and stops the subsequ ent insulin secretion ( Curry et al. 1968, MacDonald and Rorsman 2006). Furthermore, the metabolism of glucose is essential for insulin secretion and inhibition of mitochondrial metabolism halts i nsulin secretion (Aschcroft et al 1980, MacDonald et al. 2005). The breakdown of gl ucose results in the generation of ATP and an increased [ATP] to [ADP] ratio is a critical link between mitochondrial metabolism and electrical conductivity (ultimately resulting in insulin secretion) because of its ability to close the K ATP channel and de polariza tion of the cellular membrane. Figure 3: The GSIS pathway starting with the uptake of glucose via the GLUT2 transporter leading to the fusion of the vesicles consisting of insulin with the plasma membrane.

PAGE 18

10 Electrical activity of the plasma membrane is an important component of GSIS. At basal glucose levels of around 5mM, the plasma membrane has a resting membrane potential of about 70mV. This potential is governed by the Nernst equation relating extracellular and intracellular ion concentrations as follows: where R is the ideal gas constant, T is temperature in Kelvin, z is charge of ion, and F is cells are very sensitive to small ion concentration changes due to the fact that the thickness of the plasma membrane is characterized as 3.5nm resulting in a large electric field amounting to 20x10 6 V/m. The secretion of insulin is dependent on electri cells have channels embedded in their membranes that allow for flow of ions such as Ca 2+ and K + across the membrane. Potassium (K + ) contributes greatest to the overall membrane potential. At basal glucose levels w hen the cell is electrically quiescent, K + ions flow outward through the ATP sensitive K + ( K ATP ) channel hyperpolarizing the membrane and setting the membrane potential at 70mV (Holt and Hanley 2013). And because these ions are electrically charged, the ir flux across the membrane will induce the generation of action potentials or sharp changes in voltage (MacDonald and Rorsman 2006 Holt and Hanley 2013 ). Glucose M etabolism cells are an example of a glucose sensing cell type existing in a complex gluc ose sensing network responsible for maintaining the tightly regulated glucose

PAGE 19

11 homeostasis. Glucose sensitivity is an essential component of the homeostatic feedback loops which serve to maintain blood glucose concentrations at a safe and tolerable range. G lucose is transported into the Beta cell via glucose transporters, where it is phosphorylated to glucose 6 phosphate by the enzyme glucokinase (GK), which is the rate determining step of glycolysis and is considered as the glucose sensor in the pancreatic Beta cell. The product of glycolysis, pyruvate, is then a substrate for the TCA cycle in the mitochondria. In addition, cytosolic NADH also enters the mitochondria where both cytosolic and mitochondrial sources of NADH stimulate the electron transport chai n to pump H + ions out of the mitochondrial matrix causing a hyperpolarization of the inner mitochondrial membrane (Duchen et al. 1993 Maechler et al. 1997, MacDonald 2005). The dissipation of the H + gradient induces the generation of ATP via ATP synthase activity. The hyperpolarization of the inner mitochondrial membrane stimulates the mitochondrial inner membrane potential depended Ca 2+ uniporter to increase mitochondrial Ca 2+ T he increase in mitochondrial Ca 2+ mediates in ATP transport into the cytosol t hereby increasing cytosolic ATP concentrations and the [ATP] to [ADP] ratio (Moreno Sanchez 1985). The process of mitochondrial oxidative metabolism has been estimated to produce about 98 percent of the total cell generated ATP (Erecinska et al. 1992). Glucokinase and Glucokinase Mutations in Diabetes Glucokinase (GK) is the glucose sensor of the cell. GK is the rate limiting step in glucose metabolism, triggering shifts in the metabolic pathway for varyi ng levels of glucose.

PAGE 20

12 Glucokinase and Glucose Metabolism The role of the hexokinase GK as a sensor component in the glucose homeostatic feedback loop has become widely accepted. The discovery of the hexokinases occurred over 40 years ago and this led to the recognition that the enzymes were crucial for glucose metabolism (Walker et al. 1964, Sols et al. 1964, Sharma et al. 1964 ). Additionally, it was recognized that the one of the enzymes from the hexokinase family, hexokinase IV or GK, had unique functi onal and structural features that made it functionally and structurally different from other members of the hexokinase family (Grossbard et al. 1966) The unique features included having a mass of half that of hexokinase types I III, a lower affinity for g lucose, and a lack of significant feedback inhibition by its e nd product glucose 6 phosphate or G6P (Pilkis et al. 1968, Pilkis 1968) First GK was thought to be expressed only in the liver where it was responsible for glucose uptake into the liver (Pilkis et al. 1968) but then it was found to be expressed in mouse pancreatic cells where it plays a key role in GSIS (Matschinsky et al. 1968) Additionally, it was found that the intracellular concentration of glucose in cells was approximately equivalen t to the plasma glucose concentration and the glucose transport across the membrane occurred at a very high capacity suggesting that GK could serve as a pacemaker for glycolysis allowing it to control insulin secretion (Matschinsky et al. 1968) These GK e nzymatic capacities were extended to man, when they were demonstrated in human islet tissue using quantitative histochemical methods (Matschinsky et al. 1968) At first, GK was considered a biochemical glucose sensor based solely on the kinetic capacities of the enzyme (Matschinsky et al. 1968) However,

PAGE 21

13 the most compelling argument for the central role of GK in glucose homeostasis is provided by the discovery that mutations in GK cause glycemic disorders in humans. The first report of hyperglycemia as a re sult of mutations to the GK gene was published in 1992 and since then there has been nearly 200 mutants of this gene reported that causes hyperglycemia or hypoglycemia depending on the nature of the mutation (Grimsby et al. 2003) GK as the glucose sensor of the cell asserts that the enzyme with its kinetics and its strength in regulating glycolysis is the determining factor in glucose metabolism which is understood to dictate insulin secretion. The sensing of glucose requires the enzyme to change its con formation or function with the plasma glucose concentration in the physiological range of 4mM to 10mM. The GK glucose sensor is integrated with the threshold for glucose stimulated insulin release with a threshold close to 5mM glucose (for both rodents and humans), further establishing glucose homeostasis. The threshold has been established by the ATP dependent K + channels and voltage dependent Ca 2+ channels in the plasma membrane of the cells, as well as neuroendocrine regulators establishing intracellular concentrations of Ca 2+ and cAMP. In this way, the capacity of GK as a glucose sensor is coupled to complex multilevel signaling pathways involved in regulated insulin release when t he cells are exposed to glucose concentrations exceeding 5mM ( Matschinsky 2002 and Davis et al. 1999 ) The biochemical kinetics of GK enables it to serve as the glucose sensor of the cells. The point at which GK is most sensitive to changes in the con centration of glucose 0.5 and its hill coefficient when binding to glucose. 0.5 the concentration of glucose at

PAGE 22

14 which the reaction is half of the maximum rate of the system, is about 8mmol/l. The approximately 1.7. The cooperativity of the enzyme binding to glucose indicated by its hill coefficient ( nH ) of 1.7 contributes significantly to g lucose phosphorylation in the physiological range (Matschinsky and Magnuson 2004 ) This generates an inflection point for the catalysis of glucose to be around 4mmol/l and this is within the range necessary for the catalytic flux to be generated by the con centration of glucose (Matschinsky and Magnuson 2004 ) substrate, MgATP, is in the range of 0.3mM to 0.4mM which is signifi cantly below the intracellular concentration of MgATP at 2.5mM. And so, the metabolic flux depends solely on the concentration of glucose and amount of GK in the cell (Matschinsky and Magnuson 2004 ) Other factors that are important for GK to be considere d as the glucose sensor of the cell includes that the entry of glucose must not be rate limiting and the fact that there must be a lack of significant end product inhibition by G6P so that catalysis is not diminished as glucose concentration increases. This does not seem to be the cas e (Matschinsky and Magnuson 2004 ). Finally, the last necessary prerequisite for GK to serve as the glucose sensor for the cell is that the initial phosphorylation of glucose by GK to G6P must be coupled to the final secretory event indicated by the exocyto sis of insulin granules. This requires that the change in extracellular glucose is delivered to the plasma membrane and the exocytotic machinery is able to respond to the stimulus via a GK mediated mechanism. Our current understanding fully supports the va lidity of this (Matschinsky and Magnuson 2004 )

PAGE 23

15 Glucokinase Mutations in Diabetes Due to the vital role of GK in glucose stimulated insulin secretion, even small changes in the expression or activity of GK can result in significant effects in insulin secr etion. And so it follows, mutations to GK cause diabetes. Mutations alter the protein sequence by one of the following methods: changing one amino acid, transform sequence of a site of RNA splicing at exon or intron resulting in abnormal messenger RNAs, or creation of a premature termination codon by point mutation or by deletion/insertion of basepairs. While heterozygous mutations in GK result in MODY 2, double heterozygous mutations or homozygous mutations result in PNDM. The clinical consequences of th e mutations can be put into perspective when remembering the concept of GK as the glucose sensor. The enzymatic nature of GK states that the capacity of GK in cells is limited by its maximal activity or enzyme turnover (V max ) and the affinities for its substrate glucose (S o.5,gl ) and ATP (K m,ATP ) which then in turn determine the threshold for glucose stimulated insulin secretion. In man and rodents, the physio logical cell threshold is held at 5mmol/l which the K + and Ca 2+ ion channels define this threshold. In patients with GK mediated MODY or MODY 2, where inactivating mutations in one allele results in decreased enzyme turnover or affinity of enzyme to gluc ose generating an increased threshold of 7mmol/l. The heterozygous mutations result in haploinsufficiency and elevated glucose levels. In patients with GK mediated PNDM or PNDM2, where inactivating mutations in both alleles result in total inactivation of GK there is an infinitely higher reduction in enzyme turnover and/or

PAGE 24

16 affinities. The homozygous mutations resulting in truncation of protein and abolish enzyme activity (Matschinsky and Magnuson 2004 ) MODY 2 Maturity onset diabetes of the young (MODY) is a monogenic disorder characterized by familial hyperglycemia with autosomal dominant inheritance. The hyperglycemia occurs due to decrease glycolytic flux leading to insulin secretion defects and develops at any time in childhood, adolescence, or young adulthood. MODY is caused by mutations in six genes, each kind presenting a different kind of MODY identified as MODY 1 to MODY 6. Mutations to genes encoding the enzyme GK is associated with MODY 2. Heterozygous mutations in the GK gene are the common cau se of MODY. The key clinical feature of patients with GK mutations is that they present with a lifelong mild and stable fasting hyperglycemia typically ranging from 5.5mmol/l to 8.0mmol/l and their glucose is regulated to this elevated concentration (Stri de et al. 2002) Because of its mild clinical presentation and its stability, complications for patients with MODY 2 is unusual and treatment is rarely needed with the more severe cases requiring oral agents or insulin (Fajans et al. 2001, Pearson et al. 2 001) In patients with MODY threshold of blood glucose required for glucose stimulated insulin release from 4 5mM to 6 7mM as a result of a right shifted dose response curves relating insulin sec retion rate with glucose levels. Furthermore, MODY 2 patients demonstrated a 60% reduction in insulin secretion for a particular glucose level ( Byrne et al. 1994 )

PAGE 25

17 While patients do demonstrate hyperglycemia, investigators have found that inte restingly insulin levels in MODY 2 subjects are usually normal throughout the day. This is believed to be as a result of the secretory defect relating solely to the glucose blindness of the due to the fact that the release of insulin in response to other secretagogues such as arginine is anywhere from being unaffected to moderately decreased (Pueyo et al. 1994) Additionally, there is the contribution of physiologic adaptation of the limiting the effects of the defect in insulin secretion. Furthermore, in subjects with more severe GK mutations resulting in more severe decreases in enzymatic activity, the insulin secretion defect is also less severe suggesting that there must be compensatory (Pueyo et al. 1994) Some studies suggest that the compensatory mechanism is the upregulation of the single wild type GK gene allele (Sreenan et al. 1998) Figure 4: (a) Dose response curves indicating relationship between average glucose concentrations and insulin secretion rates for con trol subjects (dashed line) and patients with MODY responsiveness as a function of changes in plasma glucose concentrations. a b

PAGE 26

18 PNDM Neonatal diabetes mellitus is another monogenic disorder defined as an insulin requiring hyperglycemia which presents within the first months of life. Neonatal diabetes can be either transient or permanent. Permanent neonatal diabetes mellitus (PNDM) represents ab out 50% of the cases and is most commonly characterized by requiring lifelong insulin therapy. PNDM can occur as a result from a complete deficiency in the glycolytic enzyme GK. Infants with GK mediated PNDM have presented with low birth weight, significan t hyperglycemia, nearly undetectable c peptide levels, and requiring insulin treatment within few days of birth ( Njlstad et al. 2001, Njlstad et al. 2003, Sarici et al. 2001, Shehadeh et al. 1996 ) Theoretically, these patients should respond to GK indep endent insulin secretagogues such as arginine and sulfonylureas but studies have not been able to detect any response. Mouse Models with Decreased GK Activity To test the role of decreased GK activity, investigators developed mouse models with heterozygo us and homozygous GK knockout expression profiles. Pre viously, it was found that 50% reduction in GK gene expression lead to impairment in glucose stimulated insulin secretion. In mice with global knockout of GK, it was found that the mice die shortly afte r birth as the knockout of a second gene results in significant complications as the second gene is necessary in early embryogenesis. This lead to the development of transgenic mice that express Cre recombinase under control of cel l specific promoters. Postic et al developed a Cre loxP gene targeting strategy allowing the deletion of GK to be restricted using a conditional (or loxed) GK allele generated then

PAGE 27

19 interbred with animals that expressed Cre under the control of the in sulin 2 promoter (Postic et al. 1999, Niswender et al. 1997) However, it was found that the animals that lack GK only in the pancreatic cells exhibit a similar phenotype to animals with global knockouts of GK by containing severe hyperglycemia as a resu lt of significant decreases in insulin production. Suggesting that down regulation of GK in a global and specific form both are lethal to the species (Grupe et al. 1995) Gap Junction Coupling utonomously, they are coupled together by connexin36 (Cx36) gap junctions. In the islet, gap junctions composed of Cx36 generate electrical and metabolic coupling between cells which then regulate electrical activity and insulin secretion. The gap junction s coordinate the islet response so that in basal glucose levels there is a complete suppression of insulin secretion and upon increased glucose concentrations the gap junctions allow for junctions are intercellular channels made up of two connexon hemi channels that are made up of six connexin subunits. The gap junctions allow for the transfer and exchange of ions (such as Ca 2+ and K + ), metabolites, and other small molecules between neigh boring cells. Studies have increasingly established the role of Cx36 gap junctions in the islet and have begun to look at its implications in the development of diabetes. Investigators have shown decreased gap junction coupling contributing to disease deve lopment, further looking at the role of Cx36 in Type 1 and Type 2 diabetes (Farnsworth and Benninger 2014) Previously, other investigators have looked at the role of Cx36 gap junctions in

PAGE 28

20 exacerbating the effects of mutations that cause diabetes specifica lly mutations in the K ATP channel leading to decreased ATP sensitivity and closure of the K ATP channel (Notary et al. 2016, Hraha et al. 2014, Nguyen et al, Ashcroft and Rorsman 1989) Role of Cx36 Under High Glucose and Basal Glucose and electrical events in response to elevations in blood glucose concentrations. The lls and metabolized resulting in an increase in [ATP]/[ADP] which then promotes the closure of the K ATP channels and subsequent membrane depolarization and activation of the voltage gated Ca 2+ channels and elevation of [Ca 2+ ] i. The Ca 2+ then acts on the ex o c ytotic machinery to simulate the fusion of the insulin containing granules with the plasma membrane. Glucose stimulated insulin secretion is biphasic with first phase characterized by a burst release and a second phase characterized by a pulsatile releas e (Ashcroft and Rorsman 1989) Under stimulatory levels of glucose, electrical coupling provided by Cx36 gap junctions mediates in the transfer of a depolarizing current and the synchronization of K ATP channel mediated membrane depolarization resulting in coordinated [Ca 2+ ] i oscillations across the islet. Due to the coordination in coordinated [Ca 2+ ] i oscillations, there are coordinated insulin secretion oscillations (Benninger et al. 2008, Ravier et al. 2005, Calabrese et al. 2003) In islets isolated fr om Cx36 deficient mice, there is a loss of glucose stimulated [Ca 2+ ] i oscillations and coordinated insulin release with individual to elevated glucose (Benninger 2008 and Ravier et al. 2005)

PAGE 29

21 At basal glucose, Cx36 gap junctions coordinate K ATP driven hyperpolarization across the + between neighboring cells via the junctions (Benninger et al. 2011) In islets isolated from Cx36 deficient mice, there are spontaneo us bursts of [Ca 2+ ] i at basal glucose as a result of the heterogeneity in the glucose sensitivity (Zhang et al. 2003 and Benninger et al. 2011) Potential Roles of Cx36 in Type 1 and Type 2 Diabetes Diminished expression of Cx36 is thought to contribute to the development of type 1 and type 2 diabetes so that the biphasic nature of insulin secretion is lost (Stamper et al. 2014 and Rocheleau et al. 2006) In Type 1 diabetes, the insulin producing Beta cells are killed by infiltrating immune cells. During the progression of disease, the infiltrating immune cells produce large amounts of pro inflammatory cytokines that cause oxidative stress in the islet and lead to Beta cell apoptosis leading to the loss of insulin and hyp erglycemia. Studies have suggested a role for gap junctions in modulating cytokine induced apoptosis showing that Beta cells are more susceptible to death under pro inflammatory cytokines in a perfectly coupled islet and that overexpression of Cx36 gap jun ctions protects islets from apoptosis induced by pro inflammatory cytokines, ER and oxidative stress. A possible mechanism by which this can occur is by Ca2+ regulation dysfunction. Studies have correlated cytokine induced ER stress to reduced uptake of Ca 2+ into the intracellular stores via SERCA pump. Cx36 gap junctions regulate intracellular [Ca+] and have been shown to affect Ca2+ uptake into the ER. It has been suggested that the loss of gap junction

PAGE 30

22 coupling following pro inflammatory cytokine induced ER stress in order to increase the concentration of residual Ca2+. Type 2 diabetes is characterized by chronic hyperglycemia, hyperlipidemia, and insulin resistance that is developed as a result of genetic and lifestyle factors. Inflammation as a result o f high levels glucose and circulating free fatty acids leads to glucotoxicity and lipotoxicity which has also been attributed to disease development and progression. During the progression of disease, there is a decline in islet function and insulin resis region of the gene encoding for Cx36 is located on the susceptibility locus for Type 2 diabetes, suggesting that there is some connection. Also, in mice models it has been shown that p rolonged hyperglycemia lead to decreases in Beta cell coupling and Cx36 protein in isolated islets. Mice on a high far diet with high levels of FFA also showed decreased Cx36 gap junction coupling. Decreased coupling, lead to a loss of synchronization of e lectrical activity decreasing first phase and disrupting pulsatile second phase insulin secretion. Similar to what is observed in humans with Type 2 diabetes. This further s u ggests increasing Cx36 gap junction coupling as a therapeutic option to regain ins ulin action. Studies have shown disruption in Ca2+ similar to disruption in Cx36 gap junction coupling occurs in humans with higher BMI (potentially suggest pre diabetes) suggesting Cx36 gap junction coupling may delay onset of disease. Decreasing Cx36 Ga p Junction Coupling in K ATP Mutations Previously, investigators have shown that Cx36 gap junction coupling can exacerbate the effects of mutations by specifically looking at mutations to the K ATP

PAGE 31

23 channel making it insensitive to ATP resulting in a reduct ion in insulin secretion and hyperglycemia. Hypothesizing that removing Cx36 coupling could prevent the detrimental effects of subpopulation of unhealthy cells and restore electrical activity and insulin secretion, whereas the effects of full coupling woul d suppress the activity of individual and independent heterogeneous cells with a range of sensitivities in an islet characterized as a network of fully coupled cells. Mice were generated with tamoxifen inducible overactive K ATP channels modeling NDM with wild type Cx36 expression had blood glucose levels that rapidly rose to over 500 mg/dL in the first two weeks after induction by tamoxifen injection and remained high. The mice also display significantly suppressed GSIS levels. Then in the mice expressing inducible over active K ATP channels, a genetic knockout of Cx36 was introduced and it was found that normal blood glucose levels were maintained and there was a rescued GSIS response. These findings suggested that electrical coupling is incriminated in the progression of disease by exacerbating the effects of inactivating mutations in the progression of disease (Ngyuen et al. 2014) Other investigators looked at how the inactivations (induced again by overactive K ATP channels) in subpopulation of cells af fect islet activity as a whole in a fully coupled network with wild type Cx36 expression. It was found that as cellular excitability approaches a critical threshold value, there is an emergence of critical behavior such that if over 20 percent of the cells in an islet express over active K ATP channels then in the presence of Cx36 gap junction coupling there occurs a complete suppression of [Ca 2+ ] i oscillations and insulin secretion. To further look at this, Aleena Notary et al. aimed to understand how the i slet functions as a multicellular system and the role of gap junction

PAGE 32

24 coupling in amplifying the effects of decreased cellular excitability in the islets with over active K ATP channels expressing different levels of Cx36 gap junction coupling. It was found that upon low levels of mutations there were elevations in Ca 2+ across the islet and frequent oscillations with and without electrical coupling, however at high levels of mutations there was transient elevations in Ca 2+ but the elevations were more promin ent in the absence of coupling. In mice with high overactive K ATP percent mutation and no Cx36 expression, it was found that there were rescued islet wide activity and plasma insulin concentrations that were significantly diminished in the wild type Cx36 m ice with high K ATP mutation expression. Further suggesting coupling plays a critical role in amplifying the effects of mutations by rendering all the cells in an islet inactive as a result of mutations to a subpopulation of cells ( Notary et al 2016) Figure 5: (a) Activity in the islet as measured by calcium activity for islets expressing different levels of Kir6.2 85Q] mutation and gap junction coupling. (b) Insulin secretion data for a. (c) Plasma insulin for a. (d) Blood glucose from day 27 to day 29 for a. a b c d

PAGE 33

25 Mathematical Model of the Islet co nclude the section with an in depth analysis of the metabolic regulators in the model. Models for ell Activity (Sherman 1996) The evolution of the many mathematical models started with modeling the primary io n currents of Ca 2+ and K + ( Chay and Keizer 1983, Chay 1997, Henquin 1990), to models including meta bolism and ion current changes ( Ainsco w and Rutter 2002, Merrins et al. 2016 ) to models of Ca 2+ cAMP PKA dynamics in an oscillatory and regulatory circuit ( Ni et al 2011 ) The model used by investigators in our lab has been based off a model developed by Cha et al. that included essential components such as Ca 2+ dynamics and glucose dependence, as well as metabolism and ER dynamics (Cha et al 2011) Whil of these models, they were models only accounted for a single cell. And so the biology and information that comes from looking at the mechanism of a network of cells was unaccount ed for and lost. This warranted the need of an extension of the islet model, levels of connections and communication capacities between neighboring cells.

PAGE 34

26 Extension of The previously introduced Cha Noma model of the was previously modified to account for cell cell coupling by including a coupling current between neighboring cells to represent gap junction channels (Cha et al. 2011) Each current contributes to the membrane potential governed by the ordinary differential equation: Gap junction coupling was then simulated by generating a heterogeneous coupling current g coup between neighboring cells assigned a center (i, j) governed by: The islet model allows one to understand not just individual and independent cell behavior, but how a cell operates in a network of cells and induce cellular heterogeneities making the model more representative of physiological behavior. Specific Aim Diabetes is a severe disease that affects many people globally and is only in the environment and have the capacity to respond to treatments administered. But op erating in an electrically coupled islet exacerbates the effects of a subpopulation of unhealthy cells and suppressive conditions by inactivating the whole islet. This thesis aims to further understanding of how the islet functions as a network of coupled under electrically suppressive conditions by test ing the effects of reducing coupling in metabolically inactivating conditions by inducing mutations to subpopulation of cells and

PAGE 35

27 treating cells to induce uniform deactivation across all cells. This thesis explores the use of a the computational coupled oscillator model for the islet, microscopy techniques to examine calcium and NAD(P)H levels, transgenic mice, and experimental treatments to test islet dynamics and to what level electrica l activity ca n be rescued upon g lucokinase inactivation with modulated cell cell coupling. The broader impact of this aim is to assess whether it is possible to recover electrical activity for specific mutations that occur in humans and lead to the development of diab etes.

PAGE 36

28 CHAPTER II METHODS AND MATERIALS Mouse Studies Ethics Statement Experiments were performed in accordance with the relevant guidelines and laws, and approved by both the University of Colorado Institutional Biosafety Committee and t he Institutional Animal Care and Use Committee. Mouse Lines A conditional GK homozygous knockout (GK / ) or heterozygous knockout (GK +/ ) was generated through crossing either GK lox/lox or GK lox/wt mouse insulin promoter Cre ER with a variable deletion induced by tamoxifen. The Cre loxP system of genetic recombination has been a powerful tool in determining tissue Furthermore, the recombinati on can be by generation of a MIP Cre ER which allows for cell specific recombination dependent on the promoter driving Cre expression. The Cre functions to excise DNA regions of interest that are flanked by two loxP sites ( 34 nucleotide long DNA sequences with two symmetrical side s each having 13 basepairs and a n asymmetric 8 basepair sequence in the middle ) Cre ER is a fusion protein that consists of Cre and a mutated tamoxifen responsive estrogen receptor. In the absence of the inducing age nt tamoxifen, the Cre ER remains in the cytoplasm. Upon tamoxifen binding, Cre ER is activated and enters the

PAGE 37

29 nucleus and mediates recombination at the loxP sites (Carboneau et al. 2016). In our mouse model, once the Cre ER enters the nuc After the generation of the GK knockout mice, t he GK k n ockout mice were then crossed with mice expressing graded variations in Connexin36 coupling by having the following genotypes: Cx36 wild type (Cx36 +/+), Cx36 homozygous knockout (Cx36 / ), and/or Cx36 heterozygous knockout (Cx36 +/ ). Daily injections of tamoxifen (1 5 daily doses) were administered in 8 16 week old mice (at 50 mg/g normalized to body weight) to induce variable expression of GK knockout. Mice that lacked MIP Cre ER were used as controls. Blood Glucose and Plasma Insulin Measurements Blood glucose was monitored and measured daily using a glucometer and then averaged from first couple days to determine baseline values and on 26 th day to t he 29 th day after tamoxifen induction to determine resulting values Plasma insulin was measured on the 29 th day after tamoxifen induction from blood samples collected from the submandibular vein were centrifuged for 15 minutes at 13,900RCF and assayed u sing the mouse ultrasensitive insulin ELISA ( Crystal Chem ). Islet Isolation Islets were isolated from by injecting collagenase in the pancreatic duct, harvesting the pancreas, digesting the pancreas, and handpicking islets. The islets were then cultured in RPMI (Invitrogen) with 10% FBS, 100 U/ml penicillin, and 100 g/mL

PAGE 38

30 streptomyocin and held at 37 o C humidified with 5% CO 2 for 24 to 48 hours prior to experimentation. Insulin Secretion After isolation, the five islets per batch in duplicate were incub ated in 2mM glucose with Krebs Ringer Buffer (128.8mM NaCl, 5mM NaHCO 3 5.8mM KCl, 1.2mM KH 2 PO 4 2.5mM CaCl 2 1.2mM MgSO 4 10mM HEPES, 0.1% BSA, pH 7.4) and then incubated for 60 minutes in Krebs Ringer Buffer with 20mM glucose. The medium was then lysed with 1% TritonX 100, frozen overnight at 20 o C, and then assayed for insulin secretion and islets measured for insulin content using a mouse ultrasensitive ELISA. Calcium Imaging Isolated islets were loaded with 4M Fluo 4 (Invitrogen) in imaging medium (125NaCl, 5.7mM KCl, 2.5mM CaCl 2 1.2mM MgCl 2 10mM HEPES, 2mM glucose, 0.1% BSA, pH 7.4) for 75 minutes at room temperature, and were held in MatTek glass bottom microwell dishes. Fluorescence was imaged on spectral unmixing confocals LSM 780 (Zeiss) an d LSM 800 (Zeiss) excited at 493nm using a diode laser with 506 545nm long pass filter for emission holding temperature at 37 o C. The rate of image acquisition was an image per second and acquired 15 minutes after changing glucose concentrations from 2mM to 11mM and/or 20mM to record steady state behavior.

PAGE 39

31 NAD(P)H Imaging Isolated islets were held in dishes with imaging medium (125NaCl, 5.7mM KCl, 2.5mM CaCl 2 1.2mM MgCl 2 10mM HEPES, 2mM glucose, 0.1% BSA, pH 7.4) for 60 minutes at 37 o C Islets were plac ed in MatTek glass bottom microwell and imaged on LSM 780 (Zeiss) with 40x 1.2NA water immersion objective. NAD(P)H autofluorescence was imaged using two photon excitation at 710nm with short pass emission filter with collection peaking at 460nm and a six stack image was acquired with a width of 12m. Coupled Oscillator Model Simulations As previously mentioned, this model has been expanded from the Cha model to account for cell cell coupling. In this thesis, we aim to introduce some altered glycolytic activity. Recall, each current contributing to the membrane potential is governed by the ODE: Gap junction coupling was then simulated by generating a heterogeneous coupling current g coup between neighboring cells assigned a center (i, j) assembled into a cluster by a sphere packing algorithm. The g coup follows a distribution with SD/mean=70% derived from previously published data {Farnsworth 2014}.

PAGE 40

32 Previous Work: K ATP Channel Current and Changes to the K ATP Channel The K ATP channel current in the model is described as: where g K(ATP) describes the conductance of the K ATP channe l, V represents the membrane voltage, and p oK(ATP) describes the open channel probability of the channel. The open channel probability is described as: Previously, Notary et al. modelled Kir6.2 expression by modifying the open channel probab ility of a fraction of cells (P exc ) to: ATP channel has higher open channel probability and the opening of the channel prevents membrane depolarization and ultimately resulting in insulin secretion. Si mulating islets (solid line) with increasingly higher P exc showed an agreement with t he experimental data (points). Figure 6: Fraction activity as a function of percent mutation shown here as percent GFP. (a) Poin ts indicate calcium imaging data from Cx36(+/+) mice with increasingly higher Kir6.2 mutation expression shown by an increase in %GFP. And solid line indicates with simulation results for wild type coupling with g coup =120pS. (b) Distributions for islets isolated from Cx36( / ) mice in points and simulation results for no cou pling with g coup =0pS. a b

PAGE 41

33 We see the emergence of critical behavior displayed once again where if 20 percent of cells in the islet express ATP insen sitive K channels in the presence of gap junction coupling, the coupling will amplifying the effects of the subpopulation of cells by essentially shutting off the whole islet. In the absence of coupling, there is a more gradual decline with the uncoupled i slets being more active than the coupled islets in the high percent mutation region. While the model for percent mutation rendering K ATP channels ATP insensitive in a subpopulation of cells is a good model for NDM, mosaic expression of mutant K ATP is unl ikely in human disease However, mosaic mutation expression is unlikely in human disease. And so, the investigators simulated uniform K ATP activation (modeling diazoxide treatment) across all the cells in the islet. Diazoxide treatment was modelled by furth er modifying the open channel probability in all the cells to: simulation re sults indicated a similar rapid onset of suppression occurs upon increasing in activity in the absence of coupling. Furthermore, there is once again a region in the distribution where activity is fully suppressed in the presence of coupling and is rescued in the absence of coupling.

PAGE 42

34 Notary et al. (2016) showed different mechanisms in which inexcitability could be introduced to the islets resulting in K ATP over activity and they all exhibited similar behavior being that there was a sharp decline observed in the presence of coupling and a gradual decline in the absence of coupling. Furthermore, the findings suggested that there was potential to regain function un der the suppressive regions of the distribution by uncoupling the islet. Metabolic Regulators of the Model up glucose via a glucose transporter such as Glut2. The uptake of glucose by the glucose transporter occurs very quickly and works to establish equal intra and extra cell ular concentrations of glucose (Johnson et al. 1990) Once transported into the cell, GK phosphorylates glucose to G6P. The rate of the GK reaction is governed by the model: Figure 7: Fraction cells active as a function of current increase uniformly across all cells simulating diazoxide application for modulated coupling conductances such as g coup =120pS, 50pS, and 0pS.

PAGE 43

35 k glc is the rate of the glucokinase reaction and follows a normal distribution centered around the rate constant with 10% standard deviation. GK has two primary substrates glucose and ATP. For glucose, [Glu] i is the cytoplasmic glucose concentration, K mgl is Michaelis Menten constant or the glucose concentration at which the reactio n is half maximum, and hgl is the Hill coefficient or the cooperativity between ligand and enzyme. Similarly, [ATP] i is the cytoplasmic ATP concentration, K mATP is the Michealis Menten constant for ATP, and hATP is the Hill coefficient for ATP. Changes to the Rate of Glucokinase Reaction Mutations nullifying GK activity in different percentages of subpopulation of cells (P Mut ) was simulated by assigning J glc =0 to the number of cells (where number of cells with the mutations equaled 1000 times P Mut ).Unifo rm graded decreases in the rate of glucokinase activity across all the cells in the islet was modelled by modifying J glc so that: Uniform increase in the half maximal concentration of glucose was also modified across all the cells in the islet such that: A uniform increase in the half maximal concentration of ATP was modified similarly to that of glucose shown above.

PAGE 44

36 Changes to the Rate of Glucokinase Heterogeneity As previously mentioned, the rate of the glucokinase reaction is governed by the varia ble k glc which follows a normal distribution centered at 0.000126ms 1 with a 10% standard deviation. To test the effects of glucokinase heterogeneity under different suppressive conditions, glucokinase heterogeneity was modified by expanding the standard d eviation to 20% of the mean, 30% of the mean, and 50% of the mean. Data Analysis Calcium Imaging Data Analysis MATLAB scripts were generated to analyze all the images acquired from fluorescence imaging. After loading images, islet mask was selected and further analysis was confined to the selected area. The images were binned in a 3x3 bin to sharpen image. Photobleaching artifacts from the intensity fluctuations characterized by the time course of the bin were removed by detrending the time course (MATLA B built in function). The time course was then analyzed by a peak finder algorithm to detect peaks and troughs with area of selection defined by A peak amplitude map was generated by subtracting the mean peak amplitude by the mean trough amp litude for the bin. Then a silent cell was selected manually from the area with low peak amplitude value in the peak amplitude map and a silent cell threshold was determined by locating the maximum peak amplitude value in the area selected. Then all the pi xels in the peak amplitude map were compared to the silent cell threshold such that if a pixel in the islet

PAGE 45

37 had a peak amplitude of two fold greater than the silent cell threshold then the cell was considered active. Background area in the islet was calculated by identifying pixels with peak amplitude values lower than 80 percent of the silent cell threshold. This percentage was identified after doing extensive testing. The fraction active area was calculated based on the number of active cells in the enclosed area divided by the enclosed islet area subt racted by the background area. Figure 8: Activity map superimp osed on islet image acquired by confocal imaging. Scale bars indicate degree of correlation of activity across the islet. Figure 9: Background area correction superimposed on gray scale peak amplitude map where background area is defined by thresholding t he peak amplitude map.

PAGE 46

38 Additionally, the active area of the images was further tested to analyze for duty cycle by defining time on as counts of intensity fluctuations above 25% of the mean and calculating duty cycle as time on divided by total time. Finally, correlation of activity was also calculated by randomly selected three cell areas on the peak amplitude map and comparing the time courses for the three areas against every 5x5 bin time course characterizing a cell time course in the islet using a correlation function (MATLAB built in function). The pixels were assigned a correlation from 0 to 1 in a correlation map with respect to each cell for the three cells selected and fraction correlated area was selected from the highest correlated area calculated fr om the three correlation maps. NAD(P)H Imaging Data Analysis MATLAB scripts were once again generated to analyze all the images acquired from fluorescence imaging. After loading the six stack image (example images shown below) an islet mask was selected for each individual stack and the mean intensity was calculated in that two dimensional area. This was repeated for the remaining images in the stack. Then a mean of the means wer e calculated to quantify mean NAD(P)H intensity for the islet. After all the images of islets were analyzed for the day, the intensities were normalized to the wild type control held at 2mM glucose to calculate fold change. Images of representative islets are included below for wild type islets held at 2mM glucose and then at 11mM glucose. Notice the increase in net intensity which is quantified in NAD(P)H fold change.

PAGE 47

39 Simulation Data Analysis MATLAB scripts were once again generated to load a nd analyze text files outputted from the simulations including Ca 2+ levels, NADH, and membrane potential. For all time courses, the first 200 time points were excluded so as to analyze the behavior of only the steady state solution. First, the calcium tex t file is loaded and the first 200 time points were cut off. Previou sly published data by Notary et al. ( 2016 ) indicated that intracellular calcium fluctuates anywhere from 0.9M to 0.45M choosing a silent cell threshold of 0.165M. To calculate the numbe r of active cells, the script identifies activity based on the silent time course and identifies a cell as active if the maximum value exceeds 0.165M. So the percent of active cells is quantified as the percent cells with maximum value over 0.165 M divided by the total number of cells. Additionally, the script calculates duty cycle by Figure 10: (a) NAD(P)H autofluorescence collected by two photon microscopy at low glucose (2mM). (b) Same as in a for islet held at high glucose ( 20 mM). a b

PAGE 48

40 f the 11mM calcium control time courses. The duty cycle for the cell is then the time on divided by the total time. Then, the potential text file is loaded and the first 200 time points are cut off and the potential time courses for the cells are used in an alternative mode to calculate percent potential time course and if the a mplitude exceeds 45mV the membrane is not at resting potential the n the cell is considere d to be active. Percent activity is then characterized as the percent active cell divided by the total number of cells. Finally, the NADH text file is loaded and the first 200 time points are cut off again so that the analysis is for steady state NADH sol ution of the differential equation. There two quantifications from the oscillations are calculating mean NADH and NADH amplitude. To calculate mean NADH, a mean of the time course is obtained (using a built in MATLAB function). Then to calculate the NADH amplitude the NADH oscillatory time courses for each cell is then analyzed for its maximum and minimum and the amplitude of the cell is quantified as the maximum subtracted by the minimum.

PAGE 49

41 a b Figure 11 : (a) Representative time courses of simulation results f or changes in [Ca 2+ ] I NADH, and V m for 100 cells with wild type gap junction coupling (g coup =120pS) simulated at 11mM glucose. (b) Same as in a for simulated islet with no gap junction coupling (g coup =0pS).

PAGE 50

42 CHAP TER III RESULTS GK Deletion Modeled, Shows Critical Behavior Previously, other investigators have shown that if you inactivate an increasingly higher subpopulation of cells in an islet with wild type or full electrical coupling there is an emergence of c ritical behavior. Simulated data showed that if 20 percent of the cells in a fully coupled (g coup =120pS) network of cells expressed over active K ATP channels by being presented with a Kir6.2 mutation, then the coupling will amplify the effects of that subpopulation of cells such that all the cells in the islet will transition from being globally active to inactive shutting off essentially all electrical activity and plasma insulin levels ( T.H. Hraha and M.J. Westacott, et. al., (2014 ) In this section of the thesis, we will examine how coupling acts to enhance the action of a proximal metabolic regulator of electrical activity upstream of the K ATP channel by looking at GK. To assess how deletion of GK in islets affects electrically and metab olic response with and without electrical coupling, we simulated islets with increasing percentage of cells expressing GK knockout (GK / ) characterized as P Mut We found that if the subpopulation of non metabolically active cells in the fully coupled isl et exceeds 40 percent, then all the cells in the islet transition from being active to inactive. In the absence of coupling (g coup = 0pS) we see a more gradual decline in activity. To assure that the behavior is as a result of decreased metabolic activity, we looked at the mean parameters of the coupled/uncoupled cells at P Mut of 70 percent and found that GK KO cells indeed presented lower metabolic activity with significant decreases in ATP and

PAGE 51

43 NADH but no change in g K,ATP The distributions compared to one another indicate that before reaching the critical threshold the fully coupled is more active than the uncoupled islet and after the threshold the uncoupled islet is more active than the coupled islet. The findings suggest that upon high levels of percent mutation or P Mut in the absence of coupling the heterogeneity of cells in the islet become more explicit such that the more easily excitable cells are electrically active whereas perhaps the cells that have been inactivated are less electrically excitabl e. Figure 12 : (a) Islet activity for increas ing GK deletion in simulated islets with wild type gap junction coupling (g coup =120pS) in black and no gap junction coupling (g coup =0pS) in red. (b) Fold change in metabolic and K ATP parameters for cells with 70% GK deletions (dark red) and with wild type GK (light red) for simulated islet with no coupling. a b

PAGE 52

44 GK( / ) Mice Generation and Physiology To test the predictions of the computation m odel, GK conditional homozygous knockout(GK / ) and heterozygous (GK +/ ) mice were generated by crossing either GK lox/lox or GK lox/wt ER inducible with tamoxifen, so that the Cre is only function when the inducing agent is administered. As previously mentioned, one to five tamoxifen injections were administered to the mice to g enerate a graded effective recombination or an increasing level of GK mutations present. Additionally, the GK knockout mice were crossed with Cx36 knockout mice to achieve graded variations in electrical coupling. The GK knockout mice were monitored for bl ood glucose during the course of the tamoxifen injections and plasma insulin immediately prior to isolation. The blood glucose time courses for the GK knockout mice with and without electrical coupling a similar progression to diabetes between the differen t injection values. The blood glucose time courses indicated that the different tamoxifen injections made the mice diabetic (commonly defined as having blood glucose of over 200 mg/dL) after the first week of injections and maxing out the glucometers at 60 0mg/dL prior to isolation. Furthermore, the blood glucose time courses were the same for the mice expressing wildtype or no Cx36.

PAGE 53

4 5 When looking at fold change in blood glucose for the GK knockout islets measured as the ratio of final values to initial values, we see that the data correlates with the other findings suggesting significantly large increases in blood glucose and upon uncoupling there is no relevant changes in blood glucose. a b c d Figure 13 : Blood glucose time courses during tamoxifen injection period to monitor disease progression for (a)1 daily injection, (b) 2 daily injections (c) 3 daily injections, and (d) 5 daily injections in mice expressing wild type Cx36 gap junction coupling (black) and no Cx36 gap junction coupling (red).

PAGE 54

46 The plasma insulin levels of the GK knockout mice indicate that there has been a significant reduction in circulating insulin content. Upon uncoupling the cells in the islet, the trend indicates that there is still a reduction in plasma insulin which correlates with the blo od glucose time courses that indicate these mice are diabetic. Figure 14 : Fold change in blood glucose compared to wild type controls for mice with wild type Cx36 gap junction coupling expression (black) and no Cx36 gap junction expression (black) and 1, 2, 3, and 5 tamoxifen injections administered for gradation in GK knockout. Figure 15 : Plasma insulin measured prior to islet isolation from mice with GK knockout activated by tamoxifen injection and wild type (black) and no Cx36 gap junction coupling (red).

PAGE 55

47 NAD(P)H Imaging of Islets Isolated from GK( / ) Mice to Characterize Metabolic Activity Isolated islets from GK knockout mice were imaged for NAD(P)H levels using two photon microscopy and analyzed by MATLAB scripts (for details of microscopy and analysis refer to the methods chapter) to affirm decline in metabolic activity. Change in NAD(P)H content indicated that there was a decline in metabolic activity in the GK knockout islets when com pared to the controls. Figure 16: Change in NAD(P)H content measured via two photon microscopy for islets isolated from (a) GK knockout mice with wild type Cx36 (GK / Cx36 +/+) (b) GK knockout mice wit h no Cx36 (GK / CX36 / ) (c) and results from the two mice superimposed to compare responsiveness. a b c

PAGE 56

48 Isolated islets from GK knockout mice with Cx36 knockouts (GK / Cx36 / ) and no coupling also showed a decline in metabolic activity compared to controls for all injection values except the islets from mi ce tr eated with two injections. When comparing the islets exhibiting GK knockout with coupling and without coupling, there was no significant change in metabolic activity between the two conditions suggesting that even after reducing cell cell communication the effects of the mutations induced by the tamoxifen injections are still considerable and the islet is metabolically silenced. Calcium Imaging of Islets Isolated from GK( / ) Mice to Investigate Calcium Dynamics and Electrical Activity Previously, Notary et al. showed that mosaic inactivations by mutating the K ATP channels to become over active resulting in a dramatic decline in activity, which was recovered to an extent by uncoupling the islet (refer to Chapter I for specific details). To test this for G K, the other more upstream regulator of electrical activity in the model, we imaged GK knockout islets for changes in the [Ca 2+ ] i content shown by changes in intensities of fluo 4 and analyzed the time courses using MATLAB scripts previously mentioned. Fin dings suggest a statistically significant reduction in activity for one, two, and five tamoxifen injections and a reduced trend for three tamoxifen injections.

PAGE 57

49 When we look at islets expressing GK knockout and Cx36 knockout, we see that there is no significant recovery in activity upon uncoupling the islet. Although the trends do seem to indicate that upon increased tamoxifen injections there is a trend of higher activity for the Cx36 knockout islets. To conclude this for certain, we would nee d to perform additional experiments capturing the distribution of activity of the behavior prior to the decline in activity. TdTomato Quantification Identifies Percent Mutation Higher Than Expected To quantify percent mutation in order to conclude how mu ch mutation is actually expressed, we quantified percent area of tdTomato expression in respect to islet area. We quantified islets isolated from one, two, and three daily injection tamoxifen treated mice similar to previous experiments. Our control for th is experiment was quantifying islets isolated from mice with no Cre ER and injected with five tamoxifen injections. Our results Figure 17 : Islet activity measured by calcium dynamics as a function of different levels of GK knockout induced by different quantities of tamoxi fen injections for mice with (a) wildtype levels of Cx36 gap junction coupling (black) and (b) no Cx36 gap junction coupling (red). a b

PAGE 58

50 indicated that we had significantly higher percent mutation expression than what others had previously seen with the same dosag e to induce changes in gene expression in the electrical regulator K ATP The tdTomato expression results addressed why the GK knockout mice physiology data, NAD(P)H data, and fraction active area data were so dispersed and unresponsive. And because the percent mutation was the same across all the tamoxifen injections, we were interested in seeing if perhaps we could recover electrical function if we summed up the fraction activity for all the tamoxifen injections. And we discovered that we were not able to recover function. It showed us exactly how potent a homozygous mutation to the GK gene can be. Figure 18: Quantification of tdTomato area by measuring ratio of tdTom+ area to total area as indicated by live cell staini ng using islets isolated from GK knockout mice treated with 1, 2, and 3 tamoxifen injections.

PAGE 59

51 Decreased Rate of Glycolysis Modeled, Shows Critical Behavior Mosaic expression of GK knockout is unlikely to occur in human disease and so we mo delled and experimentally tested uniform deactivation of GK across all cells in the islet. In this section, we hope to assess how inhibition of GK in islets affects islet electrical activity dynamics with and without coupling using simulations. Instead of inducing no GK activity in a randomly selected subpopulation cells making them electrically inexcitable, we induce graded decreases in GK activity across all cells in the islet with wild type electrical coupling and no electrical coupling. When we plot rat io of active cells as a function of percent decrease in the rate of GK, the model predicted the following curves: Figure 19 : Fraction activity for control and GK knockout islets where knockout was induced by 1, 2, and 3 tamoxifen injections and expressing wild type Cx36 gap junction coupling (black) and no coupling (red).

PAGE 60

52 We see the emergence of critical behavior similar to what we have seen before. The results indicate that in a fully coupled islet, a 65 percent decrease in GK activity is tolerated above which all the cells in the islet are inactivated. In the absence of coupling, we see a more gradual decline so that prior to the critical threshold the coupled islet is more electrically active than the u ncoupled islet and after the critical threshold the uncoupled islet is more electrically active than the coupled islet. This suggests that in islets subjected to large decreases in metabolic activity (consistent with the right region of the distribution) t here is a possibility of recovery of activity by uncoupling the islet. Figure 20 : Islet activity as a function of percent inh ibition of GK activity across all cells in simulated islets with electrical coupling in black ( g coup =120pS ) and no electrical coupling (g coup =0pS) in red.

PAGE 61

53 NAD(P)H Imaging of Cx36(+/+) and Cx36( / ) Islets Treated with GK Inhibitor to Characterize Metabolic Activity To test the predictions generated by the computa tional model, we trea ted islets i solated from Cx36 wild type and knockout mice with increasing concentrations of the GK inhibitor Mannoheptulose (MH). First, we wanted to test the decline in metabolic activity by acquiring NAD(P)H levels for islets expressing wild type Cx36 an d no Cx36 at 2mM Glucose, 11mM Glucose and 11mM Glucose + Treatment. We found that as we increase the concentration of MH, we get a graded decline in mean NAD(P)H content. Furthermore, the Cx36 knockout islets show higher metabolic activity than the Cx36 w ild type islets. The increase in NAD(P)H content after adding high glucose and subsequent MH treatments shown below as the change in the NAD(P)H content suggests that the Cx36 knockouts have a lower fold change in NAD(P)H content from basal values wh en com pared to Cx36 wild types. Figure 21 : Metabolic activity for islets isolated from Cx36 wildtype mice (black) and Cx36 k nockout mice (red) treated with varying concentrations such as 3mM, 5mM, and 10mM of GK inhibitor MH. Figure (a) indicates mean NAD(P)H and (b) indicates fold change in NAD(P)H content. a b

PAGE 62

54 Calcium Imaging of Cx36(+/+) and Cx36( / ) Islets Treated with GK Inhibitor to Characterize Calcium Dynamics and Electrical Activity To test electrical activity dynamics, we measured changes in the [Ca 2+ ] i levels usi ng confocal microscopy techniques and analyzed the time courses using MATLAB scripts that look at the oscillatory dynamics across the islet. In the figure we show fraction active area and duty cycle. The fraction active area plots indicate fraction activit y as a function of treatment for islets with wild type coupling and no coupling. For islets with full electrical coupling, the activity at 11mM glucose is characterized as baseline activity and islets treated with increasing levels of MH show decreasing le vels of activity that are statistically significant. The trend of decreasing activity shows that the Cx36 knockout islets have a more gradual decline in activity than the Cx36 wild type islets. a b c Figure 22 : In islets from mice expressing wild type levels of Cx36 (black) and no Cx36 (red), results indicate (a) islet activity as a function of different MH treatments to modulate decreasing metabolic activity, (b) the distribution of the heterogeneity in islet results, and (c) duty cycle of the fraction activ e area.

PAGE 63

55 Clinical Mutations in Glucokinase Examined To test h ow the principles and findings could be extended to humans, we simulated GK mutations that occur in humans leading to the development of diabetes using kinetics reported in literature and investigated whether activity could be recovered by uncoupling the i slet. Furthermore, we looked for literature citing patient data and GK mutation kinetics to see at which point in the development of disease could we recovery the most activity. Simulating Additional Metabolic Parameters That Affect the Rate of GK Activi ty Literature citing the kinetics of mutations to GK would not only cite changes to the rate of GK activity but would also report shifts in the half maximal concentration of glucose and the half maximal concentration of ATP with their respective hill coef ficients. And so, before simulating mutations to GK that lead to the development of diabetes in humans it was important to characterize the different parameters that contribute to the activity of GK. This would allow us to better predict and understand how a mutation would behave by looking at where it would lie on the activity curves of the different parameters. And so, we simulated increases in the half maximal concentration of glucose (K m g l ) increases in the half maximal concentration of ATP (K mATP) an d the hill coefficient for glucose (hgl)

PAGE 64

56 Simulating percent increase in the half maximal concentration of glucose or increasing the glucose threshold after which GK would be activated exhibited a distribution similar to the one s we have seen before with the exception that it is more right shifted. Again the coupled islet shows a more rapid decline in activity and the uncoupled islet has a more gradual decline in activity. Then, simulating percent increases in the half maximal co ncentration of ATP did not contribute much to the decline in activity. In fact, it Figure 23: Islet simulations with wild type gap junction conductance in black (g coup =120pS) and no gap junction conductance in red (g coup =0pS) for modulated percentages in (a) the half maximal concentration of glucose, (b) the half ma ximal concentration of ATP, and the glucose hill coefficient with simulated (c) coupling (d) and no coupling. a b c d

PAGE 65

57 contributed to the decline only after uncoupling the islet and a two fold increase in the ATP threshold. Finally, simulating changes in the degree of cooperativity between t he ligand binding to the enzyme by modulating the hill coefficient also did not contribute much to the decline. We see that a 50% reduction in cooperativity shifted the distribution only 10 percent to the right. To conclude, we ran these simulations and ot hers for five random number seeds and we discovered that the most important parameters in determining how a GK will impact islet behavior include the rate of GK and the half maximal concentration of glucose. Modeling Mutations Occurring in Humans So we l ooked to the literature citing the kinetics of mutations in GK that lead to diabetes and we found mutations that lead to the two most common monogenic forms of diabetes PNDM and MODY 2 (refer to Chapter I for details). We were interested in the rate of gly colysis or the rate of GK, the half maximal concentration of glucose, the half maximal concentration of ATP, and their respective hill coefficients. Then we simulated the mutations with and without electrical coupling to determine if we could recover elect rical activity by eliminating electrical coupling in an islet expressing the GK mutation simulated at low and high levels of glucose.

PAGE 66

58 Figure 24: (a) Calcium dynamics results from simulations of PNDM mutations with wild type gap junction coupling in black (g coup =120pS) and no gap junction coupling in red (g coup =0pS). (b) As in a for MODY 2 mutations. a b

PAGE 67

59 And we found that we were able to recover electrical acti vity in a subset of mutations by uncoupling the cells in the islet. Most of the mutations we found reported the kinetics of MODY 2 mutations that are heterozygous and so a lot of the results of the simulations show full activity at high glucose, however in the subpopulation of MODY 2 mutations that are inactivating we were able to recover function in two cases. There were four PNDM cases reported from which three were inactivating and there was no recovery of activity for any of those cases upon uncoupling the cells in the simulated islet. Ranking Mutations by Clinical Severity The literature did not report much about the clinical severity of the mutations. For MODY 2 mutations, the treatment is just diet and exercise. But when you look at the electrical responsiveness of the islets, you see there is still a significant number of cases we identified with no activity. We thought it would be important to better characterize the clinical severity of the mutations to have a more complete understanding of the e ffects and clinical manifestations of those inactivating mutations, and so we looked for literature citing patient data. When we rank mutations by the results of their oral glucose tolerance tests (OGTT), we find that the distribution of results could be split into three primary stages characterized as mild, mild to severe, and severe. Simulating islets expressing those mutations indicated that the mild mutations were generally active, the mild to severe being more inactivating, and the severe being gener ally inactive. Interestingly, our results also indicated that we were able to recover activity in the most sever mutation by uncoupling the islet.

PAGE 68

60 When we ranked mutations by HbA1c percentages, we see a similar trend in the three stages. The n after simulating islets expressing the individual mutations with and without coupling our results indicated recovery of activity in islets with milder HbA1c scoring. But when we look at the mean behavior as a function of severity, we see slight recovery in all three scores in the islets with the simulated mutations and no coupling. Figure 25 : Mutations in GK presented clinically (a) ranked by oral glucose tolerance test (OGTT) scores and (b) simulated with coupling in black (g coup =120pS) and no coupling in red (g coup =0pS) (c) Fraction cells active grouped under categories of OGTT scoring of mild, mild severe, and severe. a c a b

PAGE 69

61 Figure 26: Mutations in GK presented clinically (a) ranked by oral glycosylated hemoglobin (HbA1c) percentages a nd (b) simulated with coupling in black (g coup =120pS) and no coupling in red (g coup =0pS). Fraction cells active grouped under categories of HbA1c scoring of mild, mild severe, and severe. a b c

PAGE 70

62 CHAPTER IV DISCUSSION AND FUTURE WORK The islet is a very complex and dynamic system composed of multiple cell types that are important for establishing and maintaining glucose homeostasis. The insulin within a network and the strength of the connections within the network determine how the cells in the islet respond to a stimulus. In a network with strong connections manifested by the presence of gap junction coupling, the cells will be recruited to respond similarly to the environment they are in In the absence of a stimulus or at basal glucose concentrations of les s than 5mM all the cells in the islet are quiescent I n the presence of a stimulus or at high levels of glucose all the cells in the islet are actively responding (Benninger et al. 2011, Speier et al. 2008). This manifests itself in no electrical activi ty and insulin secre tion at basal levels of glucose and synchronous oscillations of membrane potential, change in intracellular calcium concentration, and insulin secretion at high levels of glucose. In a poorly connected network with low to no levels of g ap junction coupling, there is a heterogeneous distribution of responses depending on the sensitivity of the cells to extracellular changes. This results in a subpopulation of more excitable cells exhibiting spontaneous oscillations at low levels of glucos e which become more frequent and in an increasingly number of cells upon the addition of high glucose (Benninger et al. 2014). These effects were removed in the presence of coupling, as cells were fated to respond to the stimulus globally (Benninger et al.

PAGE 71

63 which is regulated via the action of two primary regulators of electrical activity GK and the K ATP utonomously and so the responses of the shared through the cellular interactions established by Cx36 gap junction coupling. This work has been inspired by previous studies investigating the effects of a subpopulation of cells express ing over active K ATP channels induced by a amount of residual current in the K ATP channels within islets operating as networks with and without cell cell interactions via e lectrical coupling. Previously, investigators have used mouse models, experimental tools, and predictive mathematical modeling to investigate these effects on islet dynamics and electrical excitability. This thesis aims to look at a metabolic regulator of electrical activity by applying previously generated principles to understand islet dynamics as a result of decreased metabolic activity charac terized by mutations in GK inducing decreased cellular excitability in the islet. Global Behavior and GK Heterog eneity It has been previously shown that if the percent of decreased cellular excitability exceeds a critical threshold value, then the intrinsically established connections between percent of decreased cellular excitability by shutting off essentially all electrical activity in the islet (Hraha et al. 2014). It has also been shown by Notary et al. that uncoupling a heterogeneous system of cells by decreasing gap junction coupling can allow critical phase providing a means by which to recover electrical function and insulin

PAGE 72

64 secretion (Notary et al. 2016). Our computational model results suggest th at the window to recover electrical function exists for GK mediated decreases in cellular excitability, Mosaic GK Deletion Effects on Global Behavior Using the computational model to simulate islets with increasingly greater number of cells with GK deletions indicated that GK regulates islet function to behave in a biphasic manner in the presence of electrical coupling, so that in the presence of electrical coupling, the islet func tions in a globally active or inactive state. This mechanism of regulation of islet function could prove to be both helpful and detrimental to the islet. In low percent deletions, the electrical coupling allows for the cells with no deletions to compensate for the loss of function in the cells with GK deletions. But as the number of GK deletions increases past a critical threshold determined by the capacity of the unaffected cells to compensate, then the resulting population of cells will no longer be enoug h to compensate and cellular interactions will amplify these effects by shutting off activity globally across all the cells in the islet. Our results indicated that 45 percent of cells with GK deletions in an islet with full electrical coupling (g coup =120p S) is tolerated, above which the islet shows a dramatic decline in ratio of active cells (Figure 12a). Upon eliminating these cell cell interactions in the network by computationally simulating an islet with no gap junction conductance (g coup =0pS), this ph enomenon of binary function is removed. Our findings indicate that increasing the percent of GK deletions in cells in an uncoupled islet results in a linear decline in activity (Figure 12a). In the absence of coupling, our results indicated that GK deletio ns resulted in a loss of global behavior with

PAGE 73

65 capacity for responsiveness was not reduced. To better understand the mechanism behind this, we looked at the functiona l characteristics of the GK knockout cells compared to the GK wild type cells in the phase after the critical threshold W e found that the knockout cells had significantly high metabolic activity shown by increases in the NADH and ATP content regardless o f the reduction in metabolic activity present in the GK knockout cells (Figure 12b). Since the simulations predicted that uncoupling the islet at high levels of mutation by GK deletion could result in higher electrical and metabolic excitability, this also suggest s higher insulin secretion rates despite the presence of GK deletions. To test the predictions generated by the computational model, we generated a mouse model expressing tamoxifen inducible GK deletions by the use of a MIP driven Cre ER and the m ice also expressed different levels of gap junction coupling by having either wild type or no Cx36. We administered graded levels of tamoxifen injections to induce gradations in effective Cre recombination or percent GK deletions. However, our results indi cated that despite the number of tamoxifen injections our mice had similar progression to disease (Figure 13). A ll the GK knockout mice rapidly became diabetic and remained diabetic at the end of the first two weeks of injections regardless of whether the mice expressed wildtype coupling or no coupling (Figure 13). When measuring fold change in blood glucose (Figure 14) and plasma insulin levels (Figure 15), we confirmed that there was really no decrease in blood glucose and increase in plasma insulin upon eliminating gap junction coupling. The physiological data accumulated from the mice over the experiments were not promising, but we persisted with the experiments hoping to extract meaningful data about islet function and dynamics. And we discovered just

PAGE 74

66 h ow potent a GK deletion can be in its effects on global behavior. NAD(P)H imaging confirmed that there was in fact a decline in metabolic activity between the control and the different injection values, however the decline was not graded as the changes bet ween the injection values did not generate statistically significant changes in metabolic activity (Figure 16a). Upon also introducing a Cx36 knockout in the GK knockout mice, we saw that the mice also generally showed decreases in the metabolic function i n the islets (Figure 16b). And the decreases in metabolic activity are not statistically different than the metabolic activity of Cx36 wild type islets as we had hoped. Confocal imaging of calcium dynamics agreed with the data indicating metabolic function The fraction of electrical activity was also decreasing but there was no gradation in the decreases as we increased tamoxifen injections (Figure 17). And after introducing a Cx36 knockout, we saw no significant recovery of electrical activity by measurin g calcium dynamics. The lack of gradations in our metabolic and electrical data suggested that perhaps we are at higher levels of percent mutation than we had anticipated. Our td Tomato quantification allowed us to understand just how much percent mutation we are inducing by tamoxifen injections and we discovered it was high. Our results indicated anywhere between 70 percent to 85 percent mutations were induced by one to three tamoxifen injections (Figure 18). This was a lot more than what had previously be en seen. However, since the one to three injection values appeared to have similar percent mutations, we thought perhaps we could investigate whether we could recover function by averaging the activity over the injections values. And we found that we could not. There was no recovery in activity upon uncoupling the islet for the GK knockout mouse model we generated

PAGE 75

67 (Figure 19). This showed us the range of potency of GK mutations in affecting islet dynamics globally with and without electrical coupling. We t ried optimizing our experimental techniques by isolating islets sooner so that we could get a better representation of islets isolated from diabetic mice and not mice that are close to dying. Additionally, we tried reducing the time we place the pancreas o f the mice in collagenase solution to increase viability. And while the modifications may have generated improvements in the quality of the islets, they did not improve average function. And so, we were not able to experimentally capture the capacity of a heterogeneous group of cells to locally rescue activity in an environment of decreased cellular excitability. However, we believe this is as a result of too much decreased cellular excitability. Our experimental data is indicative of the part of the predi ction where there is essentially too much GK deletion and we think that perhaps the phenomenon we are trying to capture is more subtle in the GK model compared to the K ATP model previously published. Future work will consist of administering these mice wit h lower dosages of the inducing agent tamoxifen. The immediate next steps will be to do a variety of tdTomato experiments with modulated dosages of tamoxifen administered to the mice in one, two, three, and five daily injection cycles and then quantify tdT om+ area in isolated islets. We will repeat and modify the experiments to get a broad more elaborate distribution of percent mutations so that when experimenting on those mice we will capture different stages of metabolic and electrical activity to have a more complete understanding of global behavior and islet dynamics.

PAGE 76

68 Uniform GK Inhibition Effects on Global Behavior While the GK deletion mouse model is generally a good model for GK mutations that occur in human disease, mosaic expression of GK knockou t is unlikely to occur in humans. As a result, we modelled and experimentally investigated the effects of uniform GK deactivation on islet dynamics. To computationally model uniform GK inhibition in islets, we modulated decreases in the rate of glycolysi s or the rate of GK in all the cells. We then analyzed electrical activity by testing calcium dynamics in simulated islets with and without electrical coupling in the presence of the decreases in the GK rate. The computational model predicted a similar bin ary rapid decrease in activity in the presence of coupling (g coup =120pS) and a more gradual decline in activity in the absence of coupling (g coup =0pS) as had been seen in the GK deletion model (Figure 20). Such that in the presence of coupling, 65 percent reduction in GK activity is tolerated above which all the activity as had been seen before but there is a more gradual decline in activity compared to the dynamics in islets with wildtype coupling. Furthermore, see that prior to the critical threshold the coupled islet is actually more active than the uncoupled islet but after the threshold the coupled islet is actually less active than the uncoupled islet. And so, the cell cell interactions with coupling compensates for the decreases in the GK rate at first but all activity is shut off after reaching the critical threshold. In the absence of coupling, there is a gradual decline starting at over 50 percent which just gradually declines until all the cells are shut off at 80 percent decrease in GK rate. The computational model

PAGE 77

69 suggests a smaller window of recovery of about 15 percent, however it is there and can be taken advantage of. To experimentally test this predi ction, we induced uniform GK inhibition by treating the islets expressing wild type and knockout Cx36 with the GK inhibitor Mannoheptulose (MH) and characterizing metabolic and calcium dynamics in the islet. Calculating mean NAD(P)H content indicated that Cx36 knockout islets have higher NAD(P)H content at every treatment and fold change in NAD(P)H content concluded that they are less glucose responsive but the trend of decline in activity suggested that the knockouts are still somewhat responsive (Figure 2 1). Then we tested calcium dynamics of Cx36 wild type and knockout islets treated with the varying levels of MH and we found a trend of slight recovery in fraction active area (Figure 22). While there is no explicit and statistically significant difference between the Cx36 wild type and knockout for each treatment, the decreasing trend is more rapid for the Cx36 wild type islets whereas we see a more gradual trend of declining activity in the knockouts. These experiments showed us just how small the window of recovery really is, as islets are so heterogeneous in their responsiveness to the treatments which is exhibited in the figure by the large error bars especially pronounced at the predicted critical threshold that occurs between 3mM MH and 5mM MH treatme nts. Ideally, the removal of coupling would cause an increase in cells that has a higher excitability and is able respond more fully to the stimulus at treatments such as 5mM MH, however, we see that there is no significant average change between islets expressing gap junction coupling to islets expressing no coupling.

PAGE 78

70 Future work will consist of repeating these experiments to further tighten the error bars especially at the 3mM and 5mM treatmen ts so that the differences in function between batches can also be averaged out. Effects of Parameters Altering GK Sensitivities While the rate of GK function is most impacted by the rate of glycolysis, some of the additional parameters that affect the G K rate include the half maximal concentration of glucose and the half maximal concentration of ATP with their respective hill coefficients. GK mutations have shown to have a variety of effects to GK sensitivities by primarily altering the activating thr eshold which can be achieved by modulating the half maximal concentration of glucose with its hill coefficient and the half maximal concentration of ATP The results for simulating increases in the half maximal concentration of glucose indicated a similar binary behavior for a coupled islet where upon reaching a critical threshold of an over 1.75 fold increase i n the concentration of glucose, the islet transition ed from a globally active to a globally inactive state (Figure 23a). On the other hand, in the a bsence of coupling there is a more gradual decline in activity and even after a two fold increase in the glucose concentration there is still about 10 percent of cells active in the islet. The half maximal concentration of ATP has a less pronounced effect on GK sensitivities. Simulating increases in ATP showed no effect on the distribution (data not shown). So we investigated how it contributed to the decline in activity in conjunction with an increase in the glucose threshold to 60 and 70 percent of its o riginal value (Figure 23b). The results indicated that it did not contribute much, but it

PAGE 79

71 did contribute to the decline in cellular excitability and only in the absence of coupling. This furthered our hypothesis that changes in cellular excitability can on ly be introduced to an uncoupled islet because it capitalizes on the excitability of a heterogeneous population of cells. Lastly, we simulated changes in the degree of cooperativity between glucose and GK by altering the hill coefficient and examined isle t dynamics. Our results concluded that in the absence of coupling, there was really no change in the decline in cellular excitability (Figure 23d). Whereas in a coupled islet, a 50 percent reduction in the hill coefficient correlated with a ten percent shi ft to the right (Figure 23c). And so there was a change, but not a significant one. These simulations exploring parameters affecting GK sensitivities indicated that the two most important parameters in determining the effects of GK activity on islet functi on include the rate of glycolysis or GK reaction rate and the half maximal concentration of glucose. Calcium Dynamics of GK Mutations Having a thorough understanding of the parameters important in determining GK and islet function is critical to comprehe nding the calcium dynamics of GK mutations occurring in humans. The kinetics of GK mutations reported in literature indicated a range of values. This is primarily because we found only a few homozygous GK mutations resulting in PNDM and a large portion of the literature found reported heterozygous GK mutations resulting in MODY 2. Simulating the mutations in islets with electrical coupling, we found that most of the mutations did not alter calcium dynamics when compared to a wild type control (Figure 24). H owever, a fairly significant number of mutations predicted the islet was completely electrically inactive.

PAGE 80

72 To test how these dynamics would change by removing cell cell interactions or uncoupling the islet, we found that there was generally no change in ac tivity with a few islets predicting lower number of fraction cells active and recovered activity in two cells that were otherwise completely inactive at high levels of glucose. The literature indicated that naturally occurring mutations in the human GK g ene resulted in primarily MODY 2 or PNDM. The MODY 2 mutations most commonly resulted in a phenotype of mild diabetes, moderate hyperglycemia, and impaired insulin secretion. However, when you look at the calcium dynamics of the mutations, notice that ther e are still a significant number of mutations with no calcium activity (Figure 24). And so, we thought it was important to better understand the effects of GK mutations by putting them into the perspective of their clinical severity. When looking for liter ature citing both the kinetics of GK mutations and clinical data, we found we could rank percentage of glycated hemoglobin (HbA1c). When we ranked by OGTT results, we fou nd that the distribution can be split into three primary stages identified as mild, mild to severe, and severe (Figure 25). The results of the simulations of the mutations indicated that the mild mutations had mostly unaltered calcium dynamics, the mild to severe having more mutations that are inactivating, and the severe cases exhibiting mostly no calcium dynamics or are inactive. Upon simulating no cell cell interaction by removing gap junction conductance, we found that we were able to recover function i n the most severely ranked GK mutation. When we ranked by HbA1c percentages, we saw a similar three phase separation of the distribution (Figure 26). Simulating the mutations however indicated that upon removing gap junction coupling, activity could only b e re scued in the

PAGE 81

73 mutations with more mild HbA1cs with slight recoveries in the other scorings These findings further supported the hypothesis that there is a window of recovery of electrical activity by removing coupling, however, the window is very narro w in space as it can only be recovered in a small population of mutations and are time sensitive as the range Implications for Diabetes Islet studies is the heart of diabetes research. And this thesis has been aimed at stimulatory conditions by examining decreases in GK activity by percent deletion and inhibition. We explored different mechanisms by which changes in the net activity in GK can perturb glucose homeostasis and how that manifests itself in metabolic activity by measuring NAD(P)H content and electrical activity by measuring and simulating calcium dynamics. The ability of cells to communicate with one another by gap junction c oupling in the islet is critical for the islet to respond robustly to a stimulus in healthy conditions, however, we have shown that it has the capacity to be detrimental in the presence of suppressive conditions. Our simulations of the effects of a subpopu lation of inactive cells exhibiting GK mutations on islet dynamics showed an emergence of critical behavior which was alleviated with a more gradual decline in activity upon reducing gap junction coupling. The experimental results provided us with the unde rstanding that if a GK mediated disease progressed too far, uncoupling the islet would not recover further electrical activity as the cells will not have a capacity to be electrically excitable. Then, we looked at the effects of decreased cellular excitabi lity on islet dynamics by simulating

PAGE 82

74 and measuring calcium dynamics. GK inhibition in islets with wild type gap junction conductance result ed in a more rapid decline in activity which was more gradual in the islets with no coupling. Our results allowed us to conclude that the window of recovery of electrical activity by uncoupling the islet for GK mediated disease was very small and time sensitive. These principals were consistent with the findings from simulating naturally occurring GK mutations and testi ng whether recovery could be achieved by uncoupling the islet. We found we could recover activity in mutations with high OGTT scoring and mild HbA1c. Diabetic patients do exhibit significant reductions in glucose stimulated insulin secretion and insulin therapy has been revolutionary in treating patients with diabetes. very completely in all patients, eliminating the need for the islet to generate endogenous insulin by compensatory mechanisms. Perha ps we can provide a more patient specific treatment that addresses the elevated glucose. Based on the conclusions of this thesis, we propose exploring treatment options that include reducing gap junction coupling in practice along with sulfonylureas. Our results have indicated that the window of recovery of islet function after GK deletion or inhibition by uncoupling the islet is very small and limited to when th e disease has not severely damaged islet function, and so it will not be a sufficient therapeutic option on its own. Therefore, we propose continuing the use of current therapeutics such as sulfonylureas. But the efficacy of administering sulfonylureas has the capacity to be greater by uncoupling the islet and allowing the heterogeneity in the population of cells in the islet to drive islet function. By removing cell cell interactions

PAGE 83

75 by gap junction coupling, the intrinsic heterogeneity in the responsivene ss of the cells in the islet will dominate allowing the range of treatment options to be expanded and more effecti ve. Our results suggest that uncoupling the system of tightly coupled cells could recover heterogeneity and re affirm the capacity of a subpopu lation of cells to respond to a stimulus to any stimulus outstanding example for a glucose sensing cell and can be considered as a mo del by which other glucose responsive cell types including hepatocytes, gut enterocytes, and attributes its glucose sensing capacity in large part to glucokinase w hich acts as the glucose sensor of the cell and has also been found in the network of glucose sensing cell types. The findings from this research can be used to uncover novel therapeutic options not only for diab etes, but other glucose sensing deficient me chanism s of disease progression.

PAGE 84

76 REFERENCES Ainscow, E. and Rutter, G. (2002). Glucose Stimulated Oscillations in Free Cytosolic ATP Concentration Imaged in Single Isl Cells: Evidence for a Ca 2+ Dependent Mechanism Diabetes 51, S162 S170. American Diabetes Association. Microvascular and Macrovascular complications of Diabetes. American Diabetes Association. (2010). Diagnosis and classification of diabetes mellit us. Diabetes care 33.Supplement 1, S62 S69. American Diabetes Association. (2009). Diagnosis and classification of diabetes mellitus. Diabetes care 32.Supplement 1, S62 S67. American Diabetes Association. (2005). Diagnosis and classification of diabetes mellitus. Diabetes care 28, S37. American Diabetes Association. (2007). Diagnosis and classification of diabetes mellitus. Diabetes care 28, S37. Ashcroft FM, Rorsma cell. Progress in Biophysics and Molecular Biology. 1989; 54:87 143. Ashcroft SJ, Sugden MC, Williams IH (1980). Carbohydrate metabolism and the glucoreceptor mechanism. Horm Metab Res (Suppl 10): 1 7. Benninger RKP, Head WS, Zhang M, Satin LS, Piston DW. (2011). Gap junctions and other mechanisms of cell cell communication regulate basal insulin secretion in the pancreatic islet. The Journal of Physiology 589, 5453 5466. Benninger RKP, Zhang M, Head WS, Satin L S, Piston DW. (2008). Gap Junction Coupling and Calcium Waves in the Pancreatic Islet. Biophysical Journal 95, 5048 5061. Byrne M.M. Sturis J. Clm ent K. Vionnet N. Pue yo M.E. Stoffel M. Ta keda J. Passa P. Cohen D. Bell G.I. (1994). Insulin Secretory Abnormalities in Subjects with Hyperglycemia due to Glucokinase Mutations. Journal of Clinical Inves tigation 93.3, 1120 1130. Calabrese A, Zhang M, Serre Beinier V, Caton D, Mas C, Satin LS, Meda P. (2003). Connexin 36 Controls Synchronization of Ca2+ Oscillations and Insulin Secretion in MIN6 Cells. Diabetes, 52:417 424. Carboneau, B. A., Le, T. D. V., Dunn, J. C., & Gannon, M. (2016). Unexpected effects of

PAGE 85

77 the MIP CreER transgene and tamoxifen on cell growth in C57Bl6/J male mice. Physiological Reports 4 (18), e12863. ma, A. (2011). Ionic mechanisms and Ca 2+ dynamics underlying the glucose response of The Journal of General Physiology 138 (1), 21 37. Chay, T. R., & Keizer, J. (1983). Minimal model for membrane oscillations in th e pancreatic beta cell. Biophysical Journal 42 (2), 181 190. C hay,T.R. (1997 ). Effects of extracellular calcium on electrical bursting and intracellular and lum inal calcium oscillations in insulin secre cells Biophys. J. 73 1673 1688. Columbia University Department of Surgery. The Pancreas and its Functions. Curry DL, Bennett LL, Grodsky GM. (1968). Requirement for calcium ion in insulin secretion by the perfused rat pancreas. Am J P hysiol 214: 174 178. Davis, E.A Cuesta Muoz, A Raoul, M Buettger, C Sweet, I Moates, M Magnuson, M.A Matschinsky, F.M (1999) Mutants of glucokinase cause hypoglycaemia and hyperglycaemia syndromes and their analysis illuminates fundamental quantitative concepts of glucose homeostasis. Diabetes 42, 1175 1186. Duch en MR, Smith PA, Ashcroft FM. (1993) Substrate dependent changes in mitochondrial function, intracellular free calcium concentration and membrane channels in pancreatic beta cells. Biochem J. 294: 35 42. Eddlestone, G. T., Goncalves, A., Bangham, J. A., & Rojas, E. (1984). Electrical coupling between cells in islets of Langerhans from mouse. The Journal of membrane biology 77 (1), 1 14. Erecinska, M., Bryla, J., Michalik, M., Meglasson, M. D., Nelson, D. (1992). Energy metabolism in islets of Langerhans. Biochimica et Biophysica Acta 1101, 273 295. Fajans, S.S., Bell, G.I., Polonsky, K.S. (2001). Molecular mechanisms and clinical pathophysiology of maturity onset diabetes of the young. N Eng J Med 345, 971 980. Farnsworth, Nikki L., and Richard K.P. Ben FEBS letters 588.8 (2014): 1278 1287. PMC Web. 5 July 2017. Gepts, W. (1965). Pathologic anatomy of the pancreas in juvenile diabetes mellitus. Diabetes 14(10), 619 633.

PAGE 86

78 Gilon, P., & Henquin, J. C. (1992). Influence of membrane potential changes on cytoplasmic Ca2+ concentration in an electrically excitable cell, the insulin secreting pancreatic B cell. Journal of Biological Chemistry 267 (29), 20713 20720. Glo yn, A. L., Siddiqui, J. and Ellard, S. (2006). Mutations in the genes encoding the pancreatic beta cell KATP channel subunits Kir6.2 (KCNJ11) and SUR1 (ABCC8) in diabetes mellitus and hyperinsulinism. Hum. Mutat., 27: 220 231. Grimsby J Sarabu R Corbett WL Haynes NE Bizzarro FT Coffey JW Guertin KR Hilliard DW Kester RF Mahaney PE Marcus L Qi L Spence CL Tengi J Magnuson MA Chu CA Dvorozniak MT Matschinsky FM Grippo JF (2003). Allosteric activators of glucokinase: potential role in diabetes therapy. Science 301, 370 373. Grossbard L., Schimke R.T. (1966). Mul tiple hexokinases of rat tissues. Purification and comparison of soluble forms. J Biol Chem 241, 3546 3560. Grupe, A., Hultgren, B., Ryan, A., Ma, Y.H. Bauer, M., Stewart, T.A. (1995). Transgenic knockouts reveal a critical requirement for pancreatic bet a cell glucokinase in maintaining glucose homeostasis. Cell 83, 69 78. Halban, P. A., Wollheim, C. B., Blondel, B., Meda, P., NIESOR, E. N., & MINTZ, D. H. (1982). The possible importance of contact between pancreatic islet cells for the control of insul in release. Endocrinology 111 (1), 86 94. Hattersley, A. T., and Ashcroft, F. M. (2005). Activating Mutations in Kir6.2 and Neonatal Diabetes: New Clinical Syndromes, New Scientific Insights, and New Therapy. Diabetes 54(9): 2503 2513. Henquin, J. C. 199 0. Glucose induced electrical activity in 83 cells. Diabetes. 39:1457 1460. Holt, R. and Hanley, N. (2013). Endocrinology and Diabetes. West Sussex: John Wiley & Sons. Hraha, T. H., Westacott, M. J., Pozzoli, M., Notary, A. M., McClatchey, P. M., & Benni nger, R. K. P. (2014). Phase Transitions in the Multi cellular Regulatory Behavior of Pancreatic Islet Excitability. PLoS Computational Biology 10 (9), e1003819. Johnson JH, Newgard CB, Milburn JL, Lodish HF, Thorens B. The high Km glucose transporter of islets of Langerhans is functionally similar to the low affinity transporter of liver and has an identical primary sequence. J Biol Chem. 1990;265:6548 6551. Kanno T, Rorsman P, Gopel SO (2002) Glucose dependent regulation of rhythmic action potential fir channel modulation J Physiol, 545, 501 507.

PAGE 87

79 Kennedy R, Kauri L, Dahlgren G, Jung S. (2007) Metabolic Oscillations in Beta cells. Diabetes 51 S263 S161. Kim, Abraham et al. Islet Architecture: A Comparative Study. Islet s 1.2 (2009): 129 136. cell dysfunction and insulin resistance in type 2 diabetes: role of metabolic and genetic abnormalities. The American journal of medicine 113 (6), 3 11. MacDonald, P. E., Joseph, J. W., and Rorsman, P. (2005) Glucose Sensing Mechanisms Cells. Philosophical Transactions of the Royal Society B: Biological Sciences. 360.1464, 2211 2225. MacDonald PE, Rorsman P (2006) Oscillations, Intercellular Coupling, and Insulin s. PLoS Biol 4(2): e49. Maechler P, Kennedy ED, Pozzan T, Wollheim CB. (1997). Mitochondrial activation directly triggers the exocytosis of insulin in permeabilized pancreatic beta cells. Embo J 16(13), 3833 41. Matschinsky, F.M. (2002). Regulation of pa ncreatic beta cell glucokinase: from basics to therapeutics. Diabetes 51, S394 S404. Matschinsky, F.M. and Magnuson, M.A. (2004). Glucokinase and Glycemic Disease: From Basics to Novel Therapeutics. Front Diabetes. Basel, Karger, 2004, vol 16. Matschins ky, F.M., Ellerman, J.E. (1968). Metabolism of glucose in the islets of Langerhans. J Biol Chem 243, 2730 2736. Merrins, M. J., Poudel, C., McKenna, J. P., Ha, J., Sherman, A., Bertram, R., & Satin, L. S. (2016). Phase Analysis of Metabolic Oscillations and Membrane Potential in Pancreatic Islet Cells. Biophysical Journal 110 (3), 691 699. Moreno Snchez, R. (1985). Regulation of oxidative phosphorylation in mitochondria by external free Ca2+ concentrations J Biol Chem. 260(7), 4028 34. oupling Compensates for Overactive K ATP Channels to Restore Insulin Secretion and Prevent Hyperglycemia in a Diabetes 63.5 (2014): 1685 1697. PMC Web. 5 July 2017. Ni Q., Ganesan A., Aye Han N. N., Gao X., Allen M. D., Levchenko A., et al. (2011). Signaling diversity of PKA achieved via a Ca2 + cAMP PKA oscillatory circuit. Nat. Chem. Biol. 7 34 40.

PAGE 88

80 Niswender, K. D., Shiota, M., Postic, C., Cherrington, A. D., and Magnuson, M. A. (1997). Cell specific Expression and Re gulation of a Glucokinase Gene Locus Transgene. J. Biol. Chem. 272, 22604 22609 Niswender, K.D., Postic C., Jetton, T.L., Bennett, B.D., Piston, D.W. Efrat S., Magnuson, M.A. (1997). Cell specific expression and regulation of a glucokinase gene locus tra nsgene. J Biol Chem 272, 22570 22575. Njlstad PR Svik O Cuesta Muoz A Bjrkhaug L Massa O Barbetti F Undlien DE Shiota C Magnuson MA Molven A Matschinsky FM Bell GI (2001). Neonatal diabetes mellitus due to complete glucokinase deficiency. N Eng J Med 344, 1588 1592. Njlstad PR Sagen JV Bjrkhaug L Odili S Shehadeh N Bakry D Sarici SU Alpay F Molnes J Molven A Svik O Matschinsky FM (2003). Permanent neonatal diabetes caused by glucokinase d eficiency: inborn error of the glucose insulin signaling pathway. Diabetes 52(11), 2854 2860. Notary, A. M., Westacott, M. J., Hraha, T. H., Pozzoli, M., & Benninger, R. K. P. (2016). Decreases in Gap Junction Coupling Recovers Ca2+ and Insulin Secretion in Neonatal Diabetes Mellitus, Dependent on Beta Cell Heterogeneity and Noise. PLoS Computational Biology 12 (9), e1005116. Patel, N., & Gold, M. G. (2015). The genetically encoded tool set for investigating cAMP: more than the sum of its parts. Frontier s in Pharmacology 6 164. Pearson, E.R Velho, G Clark, P Stride, A Shepherd, M Frayling, T.M Bulman, M.P Ellard, S Froguel, P Hattersley, A.T .(2001). Beta cell genes and diabetes: quantitative and qualitative differences in the pathophysiology of hepatic nuclear factor 1alpha and glucokinase mutations. Diabetes 50, S101 S107. Pilkis, S.J., Hansen, R.J., Krahl, M.E. (1968). Hexose ATP phosphor transferases: Comparative aspects. Comp Biochem Phys 25, 681 684. Pilkis, S.J. (1968). Identification of human hepatic glucokinase and some properties of the enzyme. Proc Soc Exp Biol Med 129, 681 684. Pilkis, S.J., Hansen, R.J., Krahl, M.E. (1968). Hormonal control of hexokinase in animal tissues. Diabetes 17, 300 301. Prksen, N. (2002). The in vivo regulation of pulsatile insulin secretion. Diabetologia 45 (1), 3 20.

PAGE 89

81 Pueyo, M.E., Clement, K., Va xillaire, M., Passa P., Froguel P., Robert, J.J., Velho, G. (1994). Arginine induced insulin release in glucokinase deficient subjects. Diab Care 14, 1015 1021. Ravier MA, Guldenagel M, Charollais A, Gjinovci A, Caille D, Sohl G, Wollheim CB, Willecke K, Henquin J Cell Coupling, Islet Synchronization of Glucose Induced Ca2+ and Insulin Oscillations, and Basal Insulin Release. Diabetes. 2005; 54:1798 1807. Rocheleau, J. V., Remedi, M. S., Granada, B., Head, W. S., Koster, J. C., Nichols, C. G., & Piston, D. W. (2006). Critical Role of Gap Junction Coupled K ATP Channel Activity for Regulated Insulin Secretion. PLoS Biology 4 (2), e26. Sarici, S.U., Alpay, F., Dndaryz, M.R., Gyknay, E. 2001. Neonatal diabete s mellitus: patient report and review of the literature. J Pediatr Endocrinol Metab 14 451 454 Sharma, C, Manjeshwar, R., Weinhouse S. (1964). Hormonal and dietary regulation of hepatic glucokinase. Adv Enzyme Regul 2, 189 200. Shehadeh N 1 Gershoni Baruch R Mandel H Nutenko I Etzioni A (1996). Congenital permanent diabetes: a different type of diabetes? Acta Paediatr 85(12), 1415 1417. Sherman, A. (1996). Contributions of modeling to understanding stimulus secretion coupling in pancreatic beta cells. American Journal of Physiology Endocrinology and Metabolism 271 E362 E372. Sherman, A., and Rinzel, J. (1991). Model for synchronization of pancreatic beta cells by gap jun ction coupling. Biophys J 59 547 559. Sols A., Salas M., Vineula E. 1964. Induced Biosynthesis of liver glucokinase. Adv Enzyme Regul 2, 177 178. Sreenan, S.K Cockburn, B.N Baldwin, A.C Ostrega, D.M Levisetti, M Grupe, A Bell, G.I Stewart, T.A Roe, M.W Polonsky, K.S (1998). Adaptation to hyperglycemia enhances insulin secretion in glucokinase mutant mice. Diabetes 47(12), 1881 1888. Stamper, I.J., Jackson, E., Wang, X. (2014). Phase transitions in pancreatic islet cellular networks and implications for type 1 diabetes. Phys Rev E 89 012719 Stride, A Vaxillaire, M Tuomi, T Barbetti, F Njlstad, P.R Hansen, T Costa, A Conget, I Pedersen, O Svik, O Lorini, R Groop, L Froguel, P Hattersley, A.T (2002). The genetic ab normality in the beta cell determines the response to an oral glucose load. Diabetologia 45(3), 427 435.

PAGE 90

82 Suckale, J. and Solimena, M. (2008). Pancreas islets in metabolic signaling focus on the cell. Frontiers in Bioscience 13, 7156 717. Vetere, A. Choudhary, A., Burns, S.M., Wagner, B.K. (2014). Targeting the pancreatic cell to treat diabetes. Nature Reviews Drug Discovery 13, 278 289. Walker, D.G, Rao S. (1964). The role of glucokinase in the phosphorylation of glucose by rat liver. Biochem J 90, 360 368. World Health Organization. Global Report on Diabetes. 6 April 2016. Zhang M, Goforth P, Bertram R, Sherman A, Satin L. The Ca2+ dynamics of isolated mouse betacells and islets: implications for mathematical models. Biophysical Journal. 84, 2852 2870.