Quick viewing(Text Mode)

Evaluating the Glycogenic Activity and Therapeutic Capacity of PPP1R3D in a Mouse Model of Lafora Disease

Evaluating the Glycogenic Activity and Therapeutic Capacity of PPP1R3D in a Mouse Model of Lafora Disease

Evaluating the Glycogenic Activity and Therapeutic Capacity of PPP1R3D in a Mouse Model of Lafora Disease

by

Lori Israelian

A thesis submitted in conformity with the requirements for the degree of Master of Science Institute of Medical Science University of Toronto

© Copyright by Lori Israelian 2018

Evaluating the Glycogenic Activity and Therapeutic Capacity of PPP1R3D in a Mouse Model of Lafora Disease

Lori Israelian

Master of Science

Institute of Medical Science University of Toronto

2018

Abstract

Lafora disease (LD) is an intractable, neurodegenerative epilepsy caused by loss-of-function mutations in the EPM2A or EPM2B genes. Central to LD is the accumulation of malstructured (Lafora bodies; LB) in neurons, a consequence of dysregulated glycogen synthesis.

Glycogen synthase catalyzes glycogen formation and is activated by dephosphorylation. The latter is mediated by glycogen-targeting subunits of 1, including PTG (R5) and R6, known formally as PPP1R3D, both abundantly expressed in brain. PTG knockout in LD mice rescues LD, including near-complete disappearance of LB and neurodegeneration. I examined whether the same could be achieved with R6 knockout. Despite significant brain glycogen and LB reductions in R6-deficient-Epm2a-/- mice, substantial amounts of LB remained and neurodegeneration was not rescued. This partial effect remains unexplained. Future experiments to resolve the difference with PTG will shed important light on both LB formation and Lafora disease, and brain glycogen .

ii

Acknowledgments

As I approach the end of my master’s program, I cannot help but reflect on everyone that has helped me reach this stage of my academic development.

First and foremost, I would like to express my sincerest gratitude to my primary supervisor, Dr. Berge Minassian. Your faith in my abilities and continuous support is undoubtedly the reason for my success as a master’s student. Thank you providing me with countless opportunities to grow as a student and as an individual, and for being an exceptional role model throughout my academic journey. Lastly, thank you for introducing me to the dedicated Lafora community. The patients and affected families are an inspiration and have been my source of motivation throughout this process.

I would also like to extend my appreciation to the guidance provided by my co-supervisor, Dr. Steve Scherer, and my committee members, Dr. Lucy Osborne and Dr. Roman Melnyk. Thank you for insightful comments and encouraging words at each committee meeting.

Thank you to all past and present Minassian lab members that have assisted me throughout my graduate studies, or even as early as my summer student days. I am so lucky to have worked with such brilliant, patient, and kind-hearted individuals. Peter and Xiaochu – your expertises are unparalleled and I cannot thank you both enough for all the help you have given me throughout the years.

To my friends, new and old, that have ensured my time in the city has never been a dull moment - all of you mean the world to me.

And lastly, to my family, thank you for all that you do for me. Nyrie - thanks for synchronizing your coffee schedule with mine for the duration of my master’s.

iii

Table of Contents

Acknowledgments ...... iii

Table of Contents ...... iv

List of Tables ...... viii

List of Figures ...... ix

List of Abbreviations ...... xi

Chapter I: Glycogen and its role in Lafora disease ...... 1

1.1 Glycogen Biochemistry ...... 1

1.2 Glycogen metabolism ...... 2

1.2.A. Glycogen Synthesis ...... 2

1.2.A.i. ...... 3

1.2.A.ii. Glycogen Synthase ...... 4

1.2.A.iii. Glycogen Branching ...... 6

1.2.B. Glycogen Breakdown ...... 7

1.2.B.i. Glycogen ...... 7

1.2.B.ii. Glycogen Debranching Enzyme ...... 8

1.2.C. Glycogen in the Central Nervous System ...... 9

1.2.C.i. Localization of Glycogen in the Brain ...... 11

1.2.C.ii. Metabolism in Astrocyte and Neurons ...... 11

1.3 Regulation of Glycogen Metabolism ...... 12

1.3.A. Hormonal Regulation ...... 13

1.3.A.i. Norepinephrine, , and ...... 13

1.3.B. Exercise ...... 14

1.3.C. and Glycogen Targeting Subunits ...... 15

iv

1.3.C.i. Protein Phosphatase 1 ...... 16

1.3.C.ii. Glycogen-targeting Subunits...... 17

1.4 Lafora disease ...... 21

1.4.A. Clinical Presentation ...... 22

1.4.B. Laforin (EPM2A) ...... 22

1.4.C. Malin (EPM2B) ...... 24

1.4.D. Lafora bodies ...... 27

1.4.E. Models of Lafora Disease Pathogenesis ...... 29

1.4.E.i. Glycogen hyperphosphorylation and LB formation ...... 29

1.4.E.ii. Abnormal Glycogen Chain Length Underlies Lafora Disease...... 31

1.4.F. Animal Models of Lafora disease ...... 32

1.4.F.i. Mouse Models ...... 32

1.4.F.ii. Canine Models of Lafora disease ...... 33

1.4.G. Molecular Targets for Lafora Disease Therapy ...... 34

1.4.G.i. PTG ...... 34

1.4.G.ii. Glycogen Synthase ...... 35

Chapter 2 Evaluating the Glycogenic Activity of R6 and its Potential As A Therapeutic Target ...... 37

2.1 Rationale ...... 38

2.1.A. Aims ...... 38

2.1.B. Hypotheses ...... 39

2.2 Methods...... 41

2.2.A. Ethics Statement ...... 41

2.2.B. Generation of Ppp1r3d x Epm2a mouse line ...... 41

2.2.C. Generation of Ppp1r3c x Ppp1r3d x Gys1 mouse line ...... 43

v

2.2.D. Glycogen Quantification ...... 43

2.2.E. Histopathology and Immunohistochemistry ...... 44

2.2.F. Lafora body, anti-GFAP, and anti-Iba1 Quantification ...... 44

2.2.G. Tissue Homogenate Preparation...... 45

2.2.H. Glycogen Synthase Activation State Assay ...... 45

2.2.I. Kainic Acid Seizure Susceptibility Model ...... 46

2.2.J. Myoclonus Quantification ...... 47

2.2.K. Statistics ...... 47

2.3 Results ...... 49

2.3.A. Brain glycogen content is reduced in DKO and LKO/R6 het mice ...... 49

2.3.B. The knockout of one or both Ppp1r3d alleles leads to lower GS activation states in the brain ...... 51

2.3.C. R6KO and PTGKO mice have lower GS activation states than WT mice, but do not significantly differ from each other ...... 51

2.3.D. LB accumulation is reduced in the brain of LD mice lacking R6 expression ...... 54

2.3.E. DKO and LKO/R6 het mice are not rescued from gliosis ...... 58

2.3.F. R6 knockout does not reduce seizure susceptibility of LD mice ...... 64

2.3.G. Lack of R6 expression does not alter glycogen levels in skeletal muscle of male mice ...... 67

2.3.H. Skeletal muscle glycogen measurements are strikingly different between sexes of LKO, LKO/R6 het, and DKO mice ...... 67

2.3.I. GS activation state is unchanged in muscle of DKO and LKO/R6 het mice ...... 69

2.3.J. LB accumulation is comparable between males of DKO, LKO/R6 het, and LKO genotypes ...... 71

2.3.K. Male and female mice have different quantities of LB in skeletal muscle ...... 74

2.3.L. LB are reduced in the hearts of LKO/R6 Het and DKO male mice ...... 76

2.3.M. Male and Female LKO mice differ in LB quantities in the heart ...... 79

vi

2.4 Discussion ...... 81

2.4.1. The effects of R6 on glycogen metabolism...... 81

2.4.2. The effect of R6 deficiency on the histopathological and seizure susceptibility phenotypes of LD mice ...... 86

2.4.3. Limitations ...... 91

2.5 Conclusions ...... 93

2.6. Future Directions ...... 96

2.6.1 Towards a deeper understanding of glycogen-targeting subunits expressed in the brain ...... 97

2.6.2. Identifying the cause of the sex difference in muscle glycogen and LB quantities ...... 99

References ...... 101

Contributions...... 121

vii

List of Tables

Table 1. Modified Racine scale describing stages of seizure severity.

viii

List of Figures

Figure 1. Schematic showing structural differences between healthy glycogen and polyglucosans.

Figure 2. Schematic diagram of the disruption of Ppp1r3d through the insertion of a targeting vector.

Figure 3. Brain glycogen measurements in 14-month-old mice.

Figure 4. GS activity ratios from brain homogenates generated by the conduction of the GS activity assay in the absence (-) and presence (+) of G6P.

Figure 5. LB accumulation quantified by % of hippocampus occupied by PASD stain.

Figure 6. Representative images of PASD-stained hippocampi.

Figure 7. Quantification of gliosis in the hippocampus.

Figure 8. Representative images of the hippocampus following anti-GFAP immunolabelling.

Figure 9. Representative images of the hippocampus following anti-Iba1 immunolabelling.

Figure 10. Seizure susceptibility tests measuring myoclonus activity and Racine stage performance.

Figure 11. Skeletal muscle glycogen levels.

Figure 12. GS activity ratios from muscle homogenates generated by the conduction of the GS activity assay in the absence (-) and presence (+) of G6P.

Figure 13. PASD-stained skeletal muscle images.

Figure 14. Sex differences in the amount of LB accumulated in skeletal muscle.

Figure 15. PASD-stained images of the heart.

ix

Figure 16. Sex differences in amount of LB accumulated in heart.

x

List of Abbreviations

AMPK AMP-activated protein cAMP cyclic AMP G6P Glucose-6-phosphate GBE Glycogen branching enzyme GDE Glycogen debranching enzyme

GL Protein phosphatase 1, regulatory subunit 3B GLUT Glucose transporter

GM Protein phosphatase 1, regulatory subunit 3A GPh GS Glycogen synthase GS het GS heterozygous GSD (s) GSK3 Glycogen synthase kinase 3 GYG1 Glycogenin-1 GYG2 Glycogenin-2 GYS1 Glycogen synthase 1 LB Lafora Bodies LD Lafora disease LKO Laforin knockout LKO/R6 het Laforin knockout, R6 heterozygous MAPK Mitogen-activated PASD Periodic acid-Schiff with diastase treatment PI3K Phosphoinositide 3-kinase PIPs PP1 interacting proteins PKA PP1 Protein phosphatase 1 PP1c Protein phosphatase 1 catalytic subunit PTG Protein targeted to glycogen PTGKO PTG knockout xi

R3E Protein phosphatase 1, regulatory subunit 3E R3F Protein phosphatase 1, regulatory subunit 3F R6 Protein phosphatase 1 regulatory subunit 3D R6KO R6 knockout SD Standard Deviation WT Wild type

xii

Chapter I: Glycogen and its role in Lafora disease

1.1 Glycogen Biochemistry

Healthy glycogen metabolism requires that each glycogen macromolecule retain its water solubility and optimal structure as a spherical, highly organized polymer of glucose. Mathematical models of glycogen structure exist and have calculated that it can contain up to 12 tiers and consist of 55,000 glucose units for a diameter of 44 nm at its maximum (Meléndez, Meléndez-Hevia, & Cascante, 1997; Meléndez-Hevia, Waddell, & Shelton, 1993). Rather, glycogen particles fluctuate in size depending on the energy demands of the cell and average 25 nm in diameter. The inner, branched chains of glycogen are characterized as B-chains, and the outer, unbranched chains are called A-chains (Roach, Depaoli-Roach, Hurley, & Tagliabracci, 2012). Linear, inter-glucose chains of glycogen are polymerized as α-1,4-glycosidic linkages, while branch points formed on B-chains are introduced via α-1,6-glycosidic linkages (Meléndez- Hevia et al., 1993). Two key structural parameters comprise glycogen: chain length and degree of branching. Inter-glucose chains range from three to 30 glucosyl units, averaging a degree of polymerization of 13 glucose residues (Nitschke et al., 2013). Each B-chain consists of two branch points, imparting a uniform distribution of branching to the particle. The combination of relatively short chains and a high frequency of branching confers to glycogen its shape and solubility (Nitschke et al., 2017).

Glycogen particles also consist of other constituents that are present to a much lesser extent than glucose. For instance, glycogen contains small quantities of phosphate. There is approximately one phosphate incorporated for every 500-1000 glucosyl units (depending on which tissue glycogen was extracted from) incorporated as C2, C3 and C6 phosphomonoesters (Nitschke et al., 2013; De-Paoli Roach et al., 2015; Roach, 2015). Glycogen also contains small quantities of glucosamine that is possibly introduced by glycogen synthase during chain elongation. Lastly, glycogen is known to associate with proteins involved in its biosynthesis. Glycogenin, a self-

1

2

glucosylating enzyme that initiates glycogen synthesis, is covalently bound to glycogen granules and remains bound even after purification processes (Lomako, Lomako, & Whelan, 1988).

1.2 Glycogen metabolism

The mobilization and construction of glycogen in cells relies on the dynamic interplay of synthesizing and degrading . Decades of research and the use of in vitro and in vivo models have allowed for the elucidation of the molecular mechanisms underlying glycogen synthesis and breakdown. Each glycogen metabolizing enzyme is necessary for maintaining proper glycogen structure and for efficient storage and release of glucose. Deficiencies in any of these enzymes may cause glycogenoses, disorders of glycogen metabolism, which can vary in clinical severity, from manageable to fatal.

1.2.A. Glycogen Synthesis

Skeletal muscle and liver are the main glycogen-synthesizing organs of the body, but adipose tissue, kidney, heart, and brain are also capable of storing glycogen. The precursor for glycogen synthesis can be derived from direct and indirect pathways. In the direct pathway, blood glucose is taken up by an isoform-specific GLUT transporter, phosphorylated by gluco- or hexokinase into glucose-6-phosphate, and finally converted to glucose-1-phosphate. In the indirect pathway, glucose metabolism leads to non-carbohydrate precursors, such as lactate and amino acids, which through the gluconeogenic pathway, can also lead to glucose-6-phosphate and subsequently glucose-1-phosphate (Johnson & Bagby, 1988; Thorens & Mueckler, 2010 Kurland & Pilkis, 1989). In both cases, UDP-glucose pyrophosphorylase converts glucose-1-phosphate to UDP- glucose, the main glucose donor in (Gillet, Devine, Hansen, 1971).

Although isoforms of biosynthesizing enzymes can differ depending on the organ, the synthesis of glycogen follows three general steps: initiation, elongation, and branching. The self- glucosylating enzyme, glycogenin, is responsible for initiation; it generates a primer consisting

3

of 7-10 glucose units, which will serve as a substrate for the main chain elongating enzyme, glycogen synthase (Lomako, Lomako, & Whelan, 1988). Glycogen synthase then elongates the primer by catalyzing the formation of α-1,4-glycosidic linkages from its substrate, UDP-glucose. Next, glycogen branching enzyme will transfer 6-7 glucose units from the growing chain onto the C6 hydroxyl group of a glucose moiety to form an α-1,6-linked branch (Dashty, 2013). The concerted and repetitive actions of glycogen synthase and glycogen branching enzyme are responsible for synthesizing glycogen particles that are short chained, highly branched, and water soluble.

1.2.A.i. Glycogenin

Glycogenin is encoded by two genes, GYG1 and GYG2, the former encoding the muscle-specific that is also found in brain, lung, , and kidney, while the latter gene encodes liver glycogenin (Barbetti et al., 1996; Mu & Roach, 1998). Known as the glycogen synthesis initiator, glycogenin utilizes UDP-glucose as its substrate and will self-glucosylate its tyrosine- 194 residue on its acceptor site to form a C-1-O-tyrosyl linkage (Rodriguez & Whelan, 1985). This tyrosine residue is essential for facilitating the covalent bond between glycogenin and glycogen. After self-glucosylating, glycogenin will synthesize a short glucan primer of 7-10 units off which GS can continue to elongate (Lomako, Lomako, & Whelan, 1988).

It was once thought that glycogenin was a subunit of glycogen synthase (Pitcher et al., 1987). Shortly after, Lomako et al. (1988) showed that glycogenin was a separate, 38 kDa protein that has endogenous activity. Glycogenin binds glycogen synthase via 35 amino acids at its C-terminus, the purpose of which is to recruit glycogen synthase for elongation of the glycogen primer (Skurat, Dietrich, & Roach, 2006; Zeqiraj et al., 2014). This interaction between glycogenin and glycogen synthase was thought to be an essential process for glycogen synthesis, until this relationship was tested in mice lacking glycogenin expression. In rodents, it is encoded by only one gene, Gyg (Testoni et al., 2017). Testoni et al. (2017) determined that Gyg-/- mice can, in fact, synthesize glycogen without a glycogenin-formed glucan primer. Even more surprising was that the Gyg-/- mice accumulated 4 and 5 times more amylase-digestible glycogen

4

in skeletal muscle and cardiac muscle, respectively, as compared to wild type mice. Additionally, these mice presented with muscle weakness and impaired exercise capacity. These findings reinforced the biochemical (glycogen accumulation) and clinical (myopathy) phenotypes reported in humans with either reduced or completely absent glycogenin-1 (Malfatti et al., 2014; Hedberg-Oldfors et al., 2017). However, unlike the mouse model of Gyg deficiency, the glycogen that accumulated in the muscle of patients was amylase-resistant. This type of glycogen is referred to as polyglucosans (Hedberg-Oldfors et al., 2017; Malfatti et al., 2014; Colombo et al., 2016; Krag, Ruiz, & Vissing, 2017). Some patients with glycogenin-1 deficiencies also develop cardiomyopathy (Irgens et al., 2015). Reasons for the synthesis of glycogen in the absence of glycogenin-1 are still being formulated. Testoni et al. (2017) believe that glycogen synthase is capable of synthesizing glycogen without a glucan primer, while Krag et al. (2017) believe that glycogenin-1 deficiency in muscle leads to the expression of glycogenin-2, the liver isoform, in muscle as a form of compensation. Glycogenin deficiencies are now categorized as a glycogen storage disease (GSD), specifically GSD XV.

1.2.A.ii. Glycogen Synthase

Glycogen synthase (GS) is the rate determining enzyme in glycogen synthesis and utilizes UDP- glucose as its glucosyl donor. GS generates α-1,4-glycosidic linkages by transferring glucose to the non-reducing end of pre-existing glucan chains. GS is encoded by two genes: the first, GYS1, encodes muscle glycogen synthase, which is also found in the brain. The second gene, GYS2, is exclusively expressed in the liver. Despite their similar functions, muscle and liver GS are only 70% homologous in amino acid sequence, and are both multi-subunit proteins of roughly 85 kDa (Nuttall, Gannon, Bai, & Lee, 1994).

While it is greatly accepted that the primary function of GS is chain elongation during glycogen synthesis, recent studies have illuminated a possible side reaction of GS. It has been proposed that GS will rarely (once every 10,000 glucose additions from its UDP-glucose) incorporate glucose-phosphate to the growing glycogen chain, rather than solely glucose. This reaction is believed to account for the phosphorylation of glycogen at the C3 position of glucose. Further

5

evidence is needed to confirm this proposition and as a novel function of GS (Nitschke et al., 2013; Contreras et al., 2016; Chikwana et al., 2013a; Tagliabracci et al., 2011).

GS is a complicated enzyme, as it is regulated by both reversible, covalent phosphorylation and allosterically by glucose-6-phosphate (G6P) (Roach & Larner, 1977). Covalent phosphorylation has an inhibitory effect on GS activity. It can be phosphorylated by several enzymes, notably glycogen synthase kinase 3 (GSK3) and casein kinase II, where increasing amounts of phosphorylation leads to a lower activation status of GS (Roach, 1991). To date, at least nine different phosphorylation sites on GS have been identified, all of which are serine residues (Skurat, Wang, & Roach, 1994). While some sites of phosphorylation have very little effect on the overall activity of GS, other phosphorylation sites have a more prominent, inhibitory effect (Skurat et al., 1994). This inhibition can be reversed upon dephosphorylation by serine/threonine phosphatases, and entirely overridden by G6P to achieve maximal GS activity (Huang, Wilson, & Roach, 1997). The primary regulation of GS activity by G6P has been demonstrated using knock-in mice expressing G6P-insentive GS. These mice acquired half as much glycogen synthesized in skeletal muscle in comparison to their wild type counterparts (Bouskila et al., 2010).

The importance of GS activity on the quantity and chain length of glycogen has been exemplified using mouse models. When Pederson et al. (2003) overexpressed GS in mice, not only were skeletal muscle glycogen quantities increased roughly seven-fold in comparison to wild type muscle, but this glycogen was also less branched. Typically, the Gys1-/- genotype in mice results in perinatal lethality 90% of the time due to cardiac problems. Mice that survive, however, have been reported to live past 20 months of age and have normal exercise capacity despite glycogen depletion in muscle (Pederson et al., 2013; Pederson et al., 2005). In contrast, humans with null mutations in the GYS1 gene have been reported to have glycogen deficiencies leading to cardiomyopathies and poor exercise capacities (Kollberg et al., 2007). Although these patients from the case report of Kollberg et al. (2007) were described as having normal neurological functioning, Gys1 -/- deletions in mice have shown to lead to impairments in learning and memory processes (Duran, Saez, Gruart, Guinovart, & Delgado-García, 2013a).

6

1.2.A.iii. Glycogen Branching Enzyme

Glycogen branching enzyme (GBE) is the enzyme responsible for the introduction of branch points in glycogen. GBE works in concert with GS to generate the spherical and highly branched structure of glycogen. Following chain elongation by GS, GBE will first cleave approximately four to seven glucose units from the non-reducing end of a chain that has at least 11 glucosyl units. Next, GBE will introduce the branch point via an α-1,6 glycosidic linkage onto a neighbouring chain, or on the same chain, at a minimum of 4 units away from the pre-existing branch (Berg, Tymoczko, & Stryer, 2002; Froese et al., 2015). Exact numbers pertaining to chain length, number of glucose units transferred, and proximity of branch points varies between publications; however, what is uncontested is the fact that the consistency in branching pattern ensures glycogen hydrophilicity.

GBE is encoded by the gene GBE1. Mutations in GBE1 that lead to a complete absence of GBE or small percentages of residual GBE activity cause clinically distinguishable diseases, under the over-arching classification of glycogen storage disease type IV (GSD IV) (Akman et al., 2011). The hepatic form of GSD IV, Andersen’s disease, involves progressive liver cirrhosis and failure resulting in death by age five for patients without liver transplantation. Additionally, a neuromuscular version of GSD IV has been described, and is referred to as adult-onset polyglucosan body disease (APBD) (Moses & Parvari, 2002). While the reasons for the clinical variability is presumed to be associated with the type of GBE1 mutation, patients under the GSD IV umbrella accumulate malformed glycogen particles that are overall less branched, have longer chains, and lack solubility (Froese et al., 2015) These malconstructed glycogen molecules are termed polyglucosans, and can accumulate in several tissues including heart, liver, brain, and skeletal muscle, causing cell death and ultimately organ damage (Froese et al., 2015; Orhan Akman et al., 2015). Polyglucosan formation, either through GS overexpression or GBE knockdown in mice, underscores the need for balanced activity levels of biosynthesizing enzymes during glycogen synthesis (Pederson et al., 2003; Kakhlon et al., 2013).

7

1.2.B. Glycogen Breakdown

Glycogen mobilization serves different physiological purposes depending on the organ from which it is being degraded. In skeletal muscle, serves to provide energy for skeletal muscle contraction, while in liver, glycogen mobilization functions to regulate blood glucose homeostasis during periods of fasting (Jensen, Rustad, Kolnes, & Lai, 2011). In the brain, glycogen is believed to be used as an emergency glucose store during periods of hypoxia, but also during mental processes such as learning and memory formation (Mathieu et al., 2016; Jensen et al., 2011).

Upon extracellular and intracellular signaling, glycogen mobilization will take place in the cytosol and is catalyzed by two glycogenolytic enzymes: glycogen phosphorylase and glycogen debranching enzyme. Glycogen phosphorylase, the rate-limiting enzyme in this process, cleaves the α-1,4 glycosidic bonds synthesized by GS (Agius, 2015). Once glycogen phosphorylase is four glucose residues from a branch point, it ceases, and glycogen debranching enzyme will take over (Zhai, Feng, Xia, Yin, & Xiang, 2016a) . Debranching enzyme will transfer three of the four remaining glucose residues to an adjacent branch for phosphorylase to continue to trim, and then finally cleave the α-1,6 linkage to eliminate the branch point (Zhai et al., 2016). Functional glycogen phosphorylase and debranching enzyme are necessary to ensure proper glycogen degradation.

1.2.B.i. Glycogen Phosphorylase

The main enzyme responsible for catalyzing glycogen breakdown is glycogen phosphorylase (GPh). GPh utilizes inorganic phosphate to cleave inter-glucose linkages and release glucose in the form of glucose-1-phosphate (Mathieu et al., 2016). Three tissue-specific of GPh exist, each encoded by a different gene: PYGB (brain GPh), PYGM (muscle GPh), and PYGL (liver GPh). Each tissue-specific GPh has an inactive conformation, phosphorylase b, and an active conformation, phosphorylase a. They differ in the extent to which they are activated allosterically by AMP, following intracellular energy demands, and covalent phosphorylation at

8

serine 14 by the enzyme in response to extracellular signaling (Newgard, Hwang, & Fletterick, 1989; Mathieu, Dupret, & Rodrigues Lima, 2017). It is reported that brain GPh activity is enhanced more strongly through AMP binding than by phosphorylation, since Mathieu et al. (2016) found that phosphorylation of brain GPh resulted in activity that was just 60% of the activity level stimulated by AMP binding. This sensitivity to AMP suggests that brain GPh responds more so to intracellular energy needs. Mathieu et al. (2016) also determined that AMP binding was not cooperative, as the affinity of brain GPh for glycogen was not increased. This finding differs from the regulatory mechanisms of muscle GPh, which responds strongly to both allosteric activation and reversible phosphorylation, and does show following AMP binding (Mathieu et al., 2016; Morgan & Parmeggiani, 1964). Liver GPh is primarily regulated through reversible phosphorylation in response to hypo- and hyper-glycemic hormones (Mathieu et al., 2017). Phosphorylation of liver GPh results in near maximal activity, with AMP binding only increasing activity by 10-20% (Rath et al., 2000). Interestingly, the brain expresses both the brain isoform of GPh and the muscle isoform. Neurons express solely brain GPh, whereas astrocytes, which contain much of the glycogen in the brain, express both brain and muscle GPh isoforms (Pfeiffer‐Guglielmi, Fleckenstein, Jung, & Hamprecht, 2003).

GSD V (McArdle disease) and GSD VI (Hers disease) are inherited metabolic disorders resulting from deficiencies in muscle GPh and liver GPh, respectively. McArdle disease can have a juvenile or adult onset with patients presenting with exercise intolerance, myoglobinuria, myalgia and excessive serum creatine kinase. In contrast, Hers disease manifests in infants, which may present with enlarged livers, elevated liver enzymes, and short stature (Lucia et al., 2012; Burwinkel et al., 1998).

1.2.B.ii. Glycogen Debranching Enzyme

Glycogen debranching enzyme (GDE) is a bifunctional enzyme required for complete glycogen mobilization. GDE has inherent α-1,4 glucanotransferase and α-1,6 glucosidase activity that allows it to remove branches partially digested by GPh in two steps (Zhai, Feng, Xia, Yin, & Xiang, 2016b). First, GDE will act as a glucanotransferase and excise a maltotriosyl residue from

9

the branched chain, and place it on the non-reducing end of a linear chain for degradation by GPh. Next, it utilizes its glucosidase function to hydrolyze the α-1,6 linkage and release the single remaining glucosyl unit of the branch. The catalytic sites are located on the same polypeptide chain of GDE, with the glucanotransferase located at the N-terminal, and the glucosidase active site at the C-terminal. (Zhai et al., 2016; Nakayama, Yamamoto, & Tabata, 2001).

There is only one gene, AGL, encoding GDE in humans, with tissue-specific isoforms in existence (Bao, Dawson Jr, & Chen, 1996). Mutations in AGL can cause deficiencies in GDE, leading to another type of glycogen storage disease, GSD III, also referred to as Cori disease. GSD III is characterized by impairments in heart, liver, or skeletal muscle functioning, depending on the subtype. For instance, GSD IIIa initially presents with liver involvement, specifically hepatomegaly and hyperlipidemia in early childhood, until mid to late adulthood where cardio-and skeletal myopathies take place. GSD IIIb is a second subtype that has been described to only affect the liver (Dagli, Sentner, & Weinstein, 2016).

1.2.C. Glycogen in the Central Nervous System

Glycogen has been known to exist in the brain since the 1970’s, but investigation into its function was shorthanded by the fact that quantities in the brain are roughly 10 times less than that of muscle and 100 times less than liver glycogen (Nelson, Schulz, Passonneau, & Lowry, 1968). Once glycogen was identified in the brain, it was assumed that its purpose was to store and release glucose during times of need, similar to the function of glycogen in muscle and liver. As the breadth of knowledge surrounding brain glycogen metabolism increases, however, it is becoming evident that its role is to support neuronal functioning during pathological and physiological conditions.

10

Aglycemia and hypoxia are two instances during which brain glycogen is rapidly degraded to sustain neurons until the restoration of favorable conditions. In aglycemia, systemic delivery of glucose to the brain is insufficient; to delay neurological failure, astrocytic glycogenolysis will occur to generate lactate, a product of glycolysis, which is then shuttled to neurons and metabolized as an alternative source of energy (Brown et al., 2005; Brown, Tekkӧk, & Ransom, 2003). Swanson & Choi (1993) showed that astrocytic glycogen could sustain cultured neurons during glucose deprivation, but once the glycogen content was depleted, neurons degenerated. In hypoxic conditions, where there is a lack of oxygen to the brain, Saez et al. (2014) demonstrated that neuronal glycogen, although minimal, is degraded by brain GPh to protect against oxygen deprivation. This work was built upon the findings of Vilchez et al. (2007) who showed that neurons express muscle GS, which is kept inactive to prevent glycogen accumulation and neuronal apoptosis (Vilchez et al., 2007).

The physiological role of astrocytic glycogen has been linked to the uptake of extracellular K+, a product of neuronal activation. During times of high synaptic activity, astrocytes buffer + extracellular K through their membrane-bound NaHCO3 cotransporter. An increase in intracellular bicarbonate activates soluble adenylyl cyclase, which increases cAMP, triggering glycogenolysis. Glycogenolysis in astrocytes generates lactate, which is then transferred to neurons for energy (Waitt, Reed, Ransom, & Brown, 2017; Choi et al., 2012). In addition, astrocytic glycogen metabolism is believed to support neurons during times of high frequency stimulation, as Brown et al. (2005) found that inhibiting glycogenolysis using a compound known as isofagomine lead to compound activation potential failure in a mouse optic nerve, even in the presence of sufficient interstitial glucose. Lastly, brain glycogen has been determined, using rodent and chick models, to be important for learning and memory. One study generated mice lacking the expression GS only in the brain, and investigated their learning capacity in the form of operant conditioning and long-term potentiation following high-frequency stimulation (Duran, Saez, Gruart, Guinovart, & Delgado-García, 2013b). In comparison to control mice, these experimental mice underperformed in the Skinner box indicating a significant impairment in learning process, and showed no electrophysiological signs of long-term potentiation. These results suggest that with a lack of brain glycogen, the strengthening of synapses that occur in

11

learning and memory formation is compromised (Duran et al., 2013). The influence of glycogenolysis and glycolysis on learning and memory was also investigated using 1-day-old chicks. Authors found that when glycogenolysis, which provides precursors for glutamate, was inhibited, the learning process of chicks was attenuated. Also, they established that the neurotransmitter glutamate is necessary for memory consolidation, therefore making the process of glycogen breakdown necessary for learning and memory (Gibbs, Lloyd, Santa, & Hertz, 2007).

1.2.C.i. Localization of Glycogen in the Brain

Glycogen in the brain was believed to be localized solely in astrocytic processes to support nearby synapses, until Saez et al. (2014) discovered that neurons synthesize very small quantities of glycogen. This small quantity of glycogen in neurons was thought to only be degraded during hypoxic situations. Heightened interest in neuronal glycogen metabolism was also a result of experimental findings that determined large quantities of glycogen could induce neuronal apoptosis (Vilchez et al., 2007). The hippocampus, striatum, cortex, cerebellar molecular layer are all areas of high synaptic activity, and have been found to contain the greatest glycogen content (Rich & Brown, 2016).

1.2.C.ii. Glucose Metabolism in Astrocyte and Neurons

It is well known that glucose is the main source of energy for the brain. Glucose is supplied to the brain systemically, and will pass though glucose transporters to enter astrocytes (GLUT1) and neurons (GLUT3). Glucose is phosphorylated by hexokinase to G6P, with its metabolic fates (glycogenesis, glycolysis, and pentose phosphate pathway) determined mostly by cell type. Since glucose incorporation into glycogenesis has already been covered in previous sections, I will primarily focus on the glycolytic rates of neurons and astrocytes, and touch briefly upon the pentose phosphate pathway.

12

One metabolic difference between astrocytes and neurons is their rates of glycolysis. Astrocytes are highly glycolytic, the product of which is often lactate (Bittner et al., 2010). Lactate is then released by astrocytes into the extracellular space and taken up by neurons via membrane-bound monocarboxylate transporters. Lactate is a proven source of energy for neurons, undergoing oxidative metabolism to generate large quantities of ATP. It has even been found that neurons metabolize lactate even when glucose is sufficiently present (Bouzier‐Sore et al., 2006). In contrast to astrocytes, neurons have a low rate of glycolysis. Rather, they have high rates of aerobic metabolism, since they metabolize both the lactate received from astrocytes and of glucose absorbed via GLUT3 transporters (Turner & Adamson, 2011). The need for two oxidative fuels may be required for sustaining neuronal activation.

The differences in the rates of glycolysis and oxidative metabolism between neurons and astrocytes can be explained by examining expression levels of enzymes involved in the biochemical pathways. For example, the enzyme 6-phosphofructose-2-kinase/fructose-2,6- bisphosphatase-3 (pfkb3), which is a potent activator of the glycolytic enzyme phosphofructokinase-1, is expressed at very low levels in neurons, while highly expressed in astrocytes (Herrero-Mendez et al., 2009; Bélanger, Allaman, & Magistretti, 2011). In fact, when pfkb3 was overexpressed in cultured neurons, neurons underwent oxidative stress and apoptosis (Herrero-Mendez et al., 2009). This sensitivity to glycolysis could explain why neurons are less glycolytic than astrocytes.

Lastly, both astrocytes and neurons are believed to utilize glucose in the pentose phosphate pathway. Balaños et al. (2010) explain that glucose is used in this pathway to generate NADPH and reduce glutathione and maintain minimal levels of oxidative stress.

1.3 Regulation of Glycogen Metabolism

Regulating glycogen metabolism is essential, not only to avoid pathogenic circumstances, but also because of its implications in glucose homeostasis and energy production. Insulin and

13

glucagon, exercise, and important modulating proteins, initiate complex signaling cascades that modulate the activities of glycogen synthesizing and degrading enzymes to manage the fluctuation of glycogen content in cells in response to energy demands.

1.3.A. Hormonal Regulation

Glycogen stores are tightly regulated by hormonal stimulation. Norepinephrine is a neurotransmitter in the brain that can regulate the activity of neurons and glial cells. It plays a vital role in cellular metabolism by stimulating glycogenolysis in astrocytes for lactate transfer to neurons during periods of heightened synaptic activity. Systemic regulation of plasma glucose levels can be attributed to insulin and glucagon. Insulin and glucagon are produced from the beta and alpha cells of the pancreas, respectively. Insulin is an anabolic hormone which ultimately promotes glycogen synthesis, while glucagon has the opposite effect, stimulating the catabolism of glycogen and release of glucose into the bloodstream.

1.3.A.i. Norepinephrine, Insulin, and Glucagon

Norepinephrine is a neurotransmitter that elicits glycogenolysis in astrocytes via β-adrenergic receptor stimulation (Subbarao & Hertz, 1990). Activation of this G-coupled protein leads to increased cellular levels of cyclic AMP (cAMP), which stimulates protein kinase A (PKA) to phosphorylate and activate phosphorylase kinase. Phosphorylase kinase then phosphorylates glycogen phosphorylase b, changing its conformation to active phosphorylase a (O’Donnell, Zeppenfeld, McConnell, Pena, & Nedergaard, 2012). Norepinephrine stimulation of glycogenolysis is enhanced by the influx of extracellular calcium, since the activity of phosphorylase kinase has been shown to be dependent on calcium levels (Ververken, Veldhoven, Proost, Carton, & Wulf, 1982).

Insulin is a major anabolic hormone that facilitates the uptake of glucose by skeletal and cardiac muscle, liver, and adipose tissues for the synthesis of glycogen. Insulin stimulates an increase in the rate of glucose uptake by promoting the localization of glucose transporters to the cellular membrane. Hepatocytes express the GLUT2 transporter, while GLUT4 is expressed in

14

adipocytes and skeletal muscle. Intracellular glucose is phosphorylated by in liver, or hexokinase in muscle, to become G6P. Increased G6P allosterically activates GS, initiating glycogen synthesis (Bouskila et al., 2010). Insulin will also promote glycogen synthesis by initiating a cellular signaling pathway that ends in GS dephosphorylation and activation. Upon insulin binding to insulin receptors, tyrosine sites on the receptor are autophosphorylated, creating a motif for substrates to bind. The regulatory subunit of phosphoinositide 3-kinase (PI3K) binds to the receptor substrate, and is responsible for the formation of lipid products that activate downstream such as the AKT/PKB pathway, which phosphorylate GSK3 to downregulate its activity. By downregulating GSK3, GS will not be phosphorylated so glycogen synthesis can occur. To further promote the dephosphorylation of GS, downstream effects of insulin leads to the phosphorylation and activation of the regulatory subunit of protein phosphatase 1 (PP1). PP1, after to binding to tissue-specific, glycogen- targeting subunits, will be recruited to the glycogen particle where it can further dephosphorylate and activate GS (Dent et al., 1990).

Glucagon is another hormone that is crucial for the maintenance of blood glucose levels, and is responsible for the rapid mobilization of liver glycogen stores. Once bound to glucagon receptors, adenylate cyclase becomes activated and increases levels of cAMP, which in turn stimulates cAMP-dependent protein kinase A (PKA). PKA will go on to phosphorylate and activate phosphorylase kinase, which will subsequently activate GPh for glycogen breakdown (Johanns et al., 2016) .

1.3.B. Exercise

The energy for physical exercise is derived from glycogen stores in skeletal muscle; glycogen breakdown, primarily in type IIb muscle fibers, leads to glycolysis and ATP production to sustain muscular contraction. Rapid glycogenolysis is catalyzed by GPh upon activation by allosteric regulators and substrate availability. Upon neural stimulation during exercise, intracellular calcium concentrations increase and activate phosphorylase kinase, which will phosphorylate GPh to convert it from its inactive b conformation to active phosphorylase a

15

(Picton, Klee, & Cohen, 1981). In addition, excess AMP increases the affinity of phosphorylase a for its substrate, inorganic phosphate. Inorganic phosphate is primarily generated from phosphocreatine breakdown during ATP regeneration, and will increase the activity of phosphorylase a to promote glycogenolysis (Chasiotis, 1988; Chasiotis, Sahlin, & Hultman, 1982). While glycogen stores are depleted as a source of energy, it has been found that glycogen is simultaneously synthesized. One reason is because GS activity is regulated by the amount of glycogen in the myocyte; as glycogen is consumed, GS activity increases to replenish glycogen stores (Nielsen & Richter, 2003). It is also well known that the rate of glucose uptake increases following skeletal muscle contraction. Glucose is phosphorylated to G6P upon entering the cell, which at high physiological levels, will allosterically activate GS to initiate glycogen synthesis (Bloch et al., 1994). The pathway to GS activation during glycogen breakdown has also been found to involve PP1 and glycogen targeting subunits, specifically the GM subunit. Aschenbach et al. (2001) found that when GM was knocked out in mice that were subject to exercise, the activation state of GS was similar to the basal activation state, indicating no change in the degree of GS phosphorylation. In contrast, wild type mice experienced an increase in the GS activation state after exercise, indicating a lower phosphorylation status of GS (Aschenbach et al., 2001a).

These results suggested that GS dephosphorylation by the protein phosphatase-GM holoenzyme may play an important role in GS activation during exercise.

During physical activity, the brain’s demand for energy also increases. Astrocytic brain glycogen is degraded to generate lactate, which is shuttled to neurons through monocarboxylate transporters. Lactate is then oxidized to maintain sufficient levels of ATP in neurons. By this mechanism, glycogen in the brain is somewhat protective during exercise, in that it ensures neurons are constantly supplied with energy precursors (Matsui et al., 2017).

1.3.C. Protein Phosphatase 1 and Glycogen Targeting Subunits

Protein phosphatase 1 (PP1) plays a crucial role in regulating the activities of enzymes involved in glycogen metabolism. While PP1 is a redundant protein in that it executes only one function,

16

dephosphorylation, it is involved in multiple cellular processes due to its recruitment by PP1 interacting proteins (P1Ps). One category of PIPs is the glycogen-targeting subunits, which upon binding to PP1, relocate as a complex to the glycogen particle to facilitate its dephosphorylation of glycogenic enzymes. Such enzymes include GS and GPh. which are activated and inhibited, respectively, by PP1. Seven glycogen-targeting subunits exist, and each are expressed at different levels depending on tissue type. Together, the PP1-glycogen-targeting subunit holoenzyme is a regulatory component of glycogen metabolism.

1.3.C.i. Protein Phosphatase 1

PP1 is a serine/threonine phosphoprotein phosphatase that is highly conserved among . The PP1 enzyme dephosphorylates its targets as a holoenzyme composed of a regulatory subunit, which can belong to a variety of physiological processes, and a catalytic subunit (PP1c). In mammals, three genes encode the catalytic subunits: PPP1CA, PPP1CB, and

PPP1CC. Three PP1c isoforms (PP1α1, PP1α2, PP1α3) are generated from alternative splicing of the PPP1CA transcript, PPP1CB encodes one catalytic subunit PP1β, and PPP1CC generates the

PP1γ1 and PP1γ2 isoforms (Korrodi-Gregório, Esteves, & Fardilha, 2014). All isoforms are ubiquitously expressed, but can vary in the quantity at which they are found. For example,

PP1α1-3 and PP1γ1 isoforms are present at greatest quantities in the heart, while PP1β is more enriched in skeletal muscle. An exception to this is the PP1γ2 isoform, which is reported to only be expressed in the testes. Interestingly, the mammalian brain has been found to have the greatest number of kinases and protein phosphatases, with PP1 (all isoforms) being highly expressed in glial cells and neurons (Esteves et al., 2012).

PP1c requires a regulatory subunit for localization to its substrate. To date, up to 200 PIPs belonging to a variety of cellular processes, like muscle contraction and RNA splicing, have been identified and confer substrate specificity to PP1c (Ceulemans & Bollen, 2004) . In order for a PIP to bind to PP1, it requires a PP1c binding motif (Esteves et al., 2012). While there are several binding motifs for the large number of regulatory subunits involved in many physiological events, the RVXF is a common binding motif found on PIPs involved in glycogen

17

metabolism. This motif allows PIPs to bind to a hydrophobic region in the C terminal of PP1c (Egloff et al., 1997). PIPs that are involved in the regulation of glycogenic enzymes are referred to as glycogen-targeting subunits. Seven have been identified (PPP1R3A-G) and are differentially expressed throughout the body. Each holoenzyme formed between PP1c and a glycogen-targeting subunit may respond differently to intracellular or extracellular signals. The two most well-known substrates targeted for dephosphorylation are GS and GPh. It is possible that more exist, however more experiments are needed for their identification.

1.3.C.ii. Glycogen-targeting Subunits

Glycogen-targeting subunits are a subtype of PIPs that are involved in the regulation of glycogen metabolizing enzymes. They facilitate the dephosphorylation of GS and GPh by recruiting PP1 to the glycogen particle. Structurally, each subunit consists of a carbohydrate binding module (family 21) allowing it to bind to glycogen, an RVXF motif for binding PP1c, and a substrate- binding motif for GS and GPh interaction (Munro, Ceulemans, Bollen, Diplexcito, & Cohen, 2005). Currently, there are seven glycogen-targeting subunits (encoded by the genes PPP1R3A, PPP1R3B, PPP1R3C, PPP1R3D, PPP1R3E, PPP1R3F, PPP1R3G), each with a different pattern of tissue expression. Some subunits are referred to in multiple ways, and have a formal and common name.

PPP1R3A (protein phosphatase 1, regulatory subunit 3A) encodes the 124 kDa subunit RGL (also referred to as GM). GM is the most abundant glycogen-targeting subunit in striated muscle. In addition to its glycogen, PP1, and substrate binding motifs, unique to GM is its C terminus sarcoplasmic reticulum binding domain (Lerín et al., 2003). Its role in glycogen metabolism is mediated by its interactions with PP1c to form the PP1-GM complex. The association and dissociation of PP1-GM is reportedly regulated by multi-serine phosphorylation of GM, catalyzed by PKA and GSK3 (Dent, Campbell, Hubbard, & Cohen, 1989). Serine 67 is an important site as its phosphorylation, following a signaling cascade triggered by epinephrine binding, leads to the dissociation of PP1- GM to promote glycogenolysis (Kim, 2002; Walker, Watt, & Cohen, 2000) .

In contrast, the cell signaling cascade initiated by exercise appears to activate GM for promotion

18

of glycogen synthesis, however the exact pathway of GM regulation in this setting is still unknown (Aschenbach et al., 2001b). Two studies have analyzed skeletal muscle glycogen content in PPP1R3A knockout mouse models, and have shown that GM null mice have 10% of the wild type quantity of glycogen (Suzuki et al., 2001; Delibegovic et al., 2003).

PPP1R3B is the gene encoding the liver-dominant, glycogen-targeting subunit of PP1, GL. GL is a 33 kDa subunit that also localizes PP1 to the glycogen particle to regulate GS and GPh activity. A unique property of GL is that it contains a 16-amino acid at its C terminus for by liver glycogen phosphorylase a, promoting glycogenolysis. GPh is a potent allosteric inhibitor of the PP1-GL complex; once it binds to GL of PP1-GL, it prevents their activation of GS, without itself being dephosphorylated and inactivated (Kelsall, Munro, Hallyburton, Treadway, &

Cohen, 2007). Pautsch et al. (2008) found that during this event, the GL subunit interacts with the AMP binding site on liver glycogen phosphorylase a. If this interaction is inhibited, glycogen accumulation ensues as demonstrated in cultured rat hepatocytes (Zibrova, Grempler, Streicher, & Kauschke, 2008). Liver glycogen phosphorylase a is regulated more strongly by reversible phosphorylation than AMP binding. The activity of muscle glycogen phosphorylase a, in contrast, is stimulated by AMP binding. Since GL is also expressed in rabbit muscle but at lower levels,

Pautsh et al. (2008) tested whether the binding of GL to the AMP site on muscle GPh would increase its activity. Pautsch et al. (2008) found that GL peptides are able to activate muscle phosphorylase b to similar levels of activity as those found after 30 µm stimulation by AMP. While

GL is predominantly expressed in liver and at low levels in skeletal muscle, it is highly expressed in both liver and skeletal muscle in humans (Munro et al., 2002).

PPP1R3C encodes the ubiquitously expressed, PP1 regulatory subunit 3C, which is sometimes referred to as R5, but more commonly known as protein targeted to glycogen (PTG). PTG is most abundant in skeletal muscle, liver, and heart, and is also found in the brain. Cell culture experiments have revealed that PTG is ubiquitinated by the laforin-malin complex, two proteins heavily implicated in Lafora disease (Vilchez et al., 2007). In addition, in vitro experiments have determined that PTG is phosphorylated at two serine residues by the AMP-activated protein kinase (AMPK), which is believed to accelerate its targeting by the laforin-malin complex for

19

proteosomal degradation (Vernia et al., 2009). The interest surrounding PTG stems from its expression in the brain and its potent regulation of PP1-mediated GS activation. When PTG was overexpressed in mice, polyglucosans, which are insoluble and malformed glycogen particles, formed in the brain (Duran, Gruart, García-Rocha, Delgado-García, & Guinovart, 2014). Results from these two experiments reveal the strength of PTG’s regulation of glycogen synthesis in neurons, and have led to the consideration of PTG as a target for Lafora disease therapies. The effect of PTG knockout in Lafora disease mouse models will be discussed further in section 4.7.1.

PPP1R3D is the gene that encodes the fourth PP1 glycogen-targeting subunit, PPP1R3D, also known as R6. R6 is ubiquitously expressed, including in the brain. Like the other subunits, R6 has the three required domains for recruitment of PP1 to glycogen for dephosphorylation of glycogenic enzymes. Structural differences separate R6 from the other subunits. For instance, Armstrong et al. (1997) identified two extra consensus sequences at the N terminal of the protein. One sequences is recognized by mitogen-activated protein kinase (MAPK), and the other by MAPK-activated protein kinase 2 or calmodulin-dependent protein kinase 2 (Armstrong, Browne, Cohen, & Cohen, 1997). A separate structure-function analysis was conducted on R6, which uncovered an additional consensus sequence for 14-3-3 proteins which bind to phosphorylated serine/threonine residues. A serine residue within this binding site was found to be important for the binding of 14-3-3- proteins to R6, as its mutation to an alanine prevented protein binding. Also, the group discovered that preventing the interaction of R6 with 14-3-3 proteins led to a 9-fold increase in glycogen accumulation, indicating that this interaction may be a form of downregulating the glycogenic activity of R6 (Rubio-Villena, Sanz, & Garcia-Gimeno, 2015). In addition, R6 was the second subunit to be found to be regulated by laforin and malin, but unlike PTG, is degraded via the lysosomal pathway following ubiquitination (Rubio-Villena, Garcia-Gimeno, & Sanz, 2013). These latest findings have also led to the consideration of R6 as a therapeutic target for Lafora disease, since it may also play a role in the regulation of brain glycogen metabolism.

20

PPP1R3E is the gene encoding the fifth subunit, R3E. Munro et al. (2005) showed that R3E is expressed in large quantities in human skeletal and cardiac muscle, non-existent in human brain, and at low quantities in liver and kidney. In rodents, R3E was expressed greatly in the heart, had second greatest expression levels in the liver, and finally, was present in very low quantities in skeletal muscle and brain (Munro et al., 2005).

PPP1R3F encodes the protein R3F, which is expressed at highest levels in mouse brain, and at lower quantities in heart and liver. In mice skeletal muscle, R3F was barely detectable, however, biopsies of human skeletal muscle revealed that R3F is present in much larger quantities. Additionally, R3F was found to not only associate with the glycogen molecule, but also with the plasma membrane both in vivo and in vitro due to a hydrophobic region at its C terminus (Kelsall, Munro, Hallyburton, Treadway, & Cohen, 2007).

PPP1R3G encodes R3G, the seventh glycogen targeting subunit. Munro et al. investigated its mRNA expression solely in humans and found it to be present exclusively in the brain (Munro et al., 2005). In mice, Luo et al. (2011) found R3G to be highly expressed in liver and brain tissues. Results from recent studies have led to the consideration of R3G as a cyclic gene. It appears to be regulated by fasting, as mice that had fasted for 12 hours had R3G mRNA levels in the liver increased by 12-fold as compared to unfasted mice. Interestingly, when mice were refed after fasting, mRNA levels were significantly reduced. These results from Luo et al. (2011) suggest that PPP1R3G may fluctuate in expression depending on the fasting-fed state of the mouse. The effect of PPP1R3G knockout on liver glycogen measurements revealed that mice lacking R3G had 50% less liver glycogen than wild type mice and 33% less glycogen in white adipose tissue (Zhang et al., 2017).

It is interesting to consider why several glycogen-targeting subunits are expressed in a certain tissue. Each subunit is no more than 50% homologous to one another, and is well conserved between humans and murine models. This suggests that each subunit exists for a specific reason, and while appear to be functionally equivalent in their regulatory properties of glycogen

21

metabolism, may be modulated by different physiological stimulants and post-translational modifications.

1.4 Lafora disease

Lafora disease (LD) was first described in 1911 by Gonzalo Lafora, who distinguished the disease based on the presence of intracytoplasmic inclusion bodies in neurons of an adolescent patient (Delgado-Escueta, 2007). These inclusions are now known to be aggregates of malformed glycogen called Lafora bodies, and are the hallmark of LD. LD is an autosomal recessive disease, where 90% of LD patients have inherited loss of function mutations in one of two genes: EPM2A, which encodes the glycogen phosphatase, laforin, and EPM2B (also referred to as NHLRC1) which encodes malin, an E3 ubiquitin that is also involved in glycogen metabolism (Couarch et al., 2011). Two explanations have been generated to account for the 10% of LD patients with functional laforin and malin. The first explanation is that the mutation may lie within the non-coding region of EPM2A and EPM2B. The second is the possibility of a third locus of disease (Chan et al., 2004a). A third locus was discovered and mapped to the gene PRDM8. Turnbull et al. (2012) identified a gain-of-function mutation in this gene that led to early onset LD. The translocation of laforin and malin from the nucleus to the cytoplasm was disrupted, indicating that PRDM8 may interact with the two proteins (Turnbull et al., 2012).

LD is a very rare progressive myoclonus epilepsy, with just over 200 independent families identified (Singh & Ganesh, 2009). Most LD patients reside in in Mediterranean countries (France, Spain, Italy), the Middle East, Pakistan, India, Northern Africa and Central Asia, often in regions where consanguineous marriages are prevalent (Delgado-Escueta, 2007; Turnbull et al., 2016). LD is diagnosed by skin biopsies that show periodic acid-Schiff positive, diastase resistant (PASD) Lafora bodies, by the detection of homozygous mutations in either EPM2A or EPM2B, and by abnormal electroencephalogram recordings (Minassian, 2001). The several types of seizures endured by LD patients are extremely difficult to manage, and to date, no treatment or means of slowing disease progression exists.

22

1.4.A. Clinical Presentation

One of the most troubling aspects of LD is that affected children are seemingly healthy at birth. As they approach late childhood years, the preliminary symptoms of disease may encroach, and include frequent headaches, forgetfulness, depression, and a decline in academics. Early-mid adolescence is when the first epileptic episode can occur, placing LD in the category of adolescent-onset epilepsies. The initial seizure can be one of a variety, such as generalized myoclonic seizures, absence seizures, atonic or tonic-clonic seizures, occipital seizures with temporary blindness, and visual hallucinations. (Minassian, 2001; Turnbull et al., 2012; Turnbull et al., 2016). Shortly after seizure onset, epileptic episodes increase in frequency and severity, and cannot be attenuated with medications. Consequently, rapid deterioration of neurological functioning and complete loss of autonomy are inevitable. Near the end of the disease course, patients are left in a vegetative state, and are in a constant state of epilepsy known as status epilepticus. Despite the genetic heterogeneity, patients follow a similar clinical course, with the existence of few exceptions (Minassian et al., 1998). Death is inevitable and will commonly occur 10 years after disease onset from a large convulsive seizure or aspiration pneumonia from respiratory failure (Turnbull et al., 2016; Minassian, 2001).

1.4.B. Laforin (EPM2A)

EPM2A was the first causative gene identified for LD. Located on chromosome 6q24, its 4-exon transcript was found to encode a 38 kDa protein called laforin (Minassian et al., 1998). At its C terminus, laforin has a dual-specificity phosphatase domain which allows it to dephosphorylate substrates at tyrosine and serine/threonine residues (Ganesh et al., 2000). Laforin also has a carbohydrate binding module of family 20 at its N terminus that targets it to glycogen, allowing it to regulate enzymes in its vicinity and dephosphorylate glycogen (Wang, Stuckey, Wishart, & Dixon, 2002). Over 40 disease-causing mutations in EPM2A have been curated with nonsense, missense and insertion mutations all reported. Regardless of the type of mutation, a loss of function is conferred (Singh & Ganesh, 2009). Laforin is the only phosphatase in the entire

23

proteome that has a carbohydrate binding module, strongly implicating it in glycogen metabolism and the regulation of glycogenic proteins.

Since the discovery of laforin and its involvement in glycogen metabolism, its targets for dephosphorylation have been of longstanding interest for the understanding of LD pathogenesis. Laforin has been shown to localize not only to the rough endoplasmic reticulum, but also to the plasma membrane (Minassian et al., 2001). Its association with two subcellular locations may suggest the involvement of laforin in more than one cellular process and explain why laforin has been found to dephosphorylate several targets. Lohi et al. (2005) investigated a well-known inactivator of GS, GSK3β, and found, using overexpression experiments in Chinese hamster ovary cells, that laforin interacts with and activates GSK3. Activating GSK3 would lead to the phosphorylation and inactivation of GS to prevent glycogen synthesis (Lohi et al., 2005). Using yeast two-hybrid screens and co-immunoprecipitation experiments, the Hira-interacting protein 5 (HIRIP5), the EPM2A Interacting Protein 1 (EPM2AIP1), AMPK, PTG, and R6 have all been found to bind laforin (Fernández-Sánchez et al., 2003; Solaz-Fuster et al., 2007; Ganesh et al., 2003; Ianzano, Zhao, Minassian, & Scherer, 2003; Rubio-Villena et al., 2013). Based on the number of laforin-interacting proteins, it is evident that the physiological role of laforin is complex.

LD patients undergo a nearly identical course of disease progression regardless of a lack of laforin or malin, indicating that laforin and malin have interdependent physiological roles. To execute these roles, laforin and malin form an interactive complex. Evidence for this complex first came in 2005, when yeast two-hybrid analyses of human brain cDNA revealed that full- length laforin binds to the NHL (protein-interacting) domains of malin. Co-expression of tagged laforin and malin in HEK293 cells also revealed their interaction after co-immunoprecipitation experiments (Gentry, Worby, & Dixon, 2005a). Additional in vitro experiments were conducted showing that, when complexed to laforin, malin can ubiquitinate and downregulate both PTG and GS (Vilchez et al., 2007). Clinical evidence emphasizing the importance of a functional laforin-malin complex came when a LD patient was identified to have a mutation in one of the NHL domain coding regions (D146N) of malin, which prevented its interaction with laforin

24

(Sullivan, Nitschke, Steup, Minassian, & Nitschke, 2017; Vilchez et al., 2007). Although the interaction between laforin and malin is generally accepted, difficulties studying their interaction in vivo arise from a lack of for malin.

One undisputed binding partner of laforin is glycogen itself. It is well known that laforin can associate with glycogen particles with its carbohydrate binding module (Wang et al., 2002). Chan et al. (2004a) not only confirmed this binding, but found that laforin is drawn to polyglucosans composing Lafora bodies more strongly than to healthy glycogen in vivo. This binding pattern was similarly shown between solubilized starch and glycogen, where laforin was found at greater quantities at the starch granule, suggesting a preference for carbohydrates with longer glucan chains (Chan et al., 2004b). Considering that polyglucosans are structurally similar to starch, and that the absence of laforin causes glycogen particles to adopt starch-like structures, the overarching function of laforin may be to preserve chain length and glycogen structure. Upon binding to glycogen, laforin can hydrolyze phosphate esters at C2, C3, and C6 carbons that are normally found at low quantities within the glycogen particle. Until recently, it was thought that excess glycogen phosphate, as a result of deficiencies in laforin, was responsible for glycogen malformation, insolubility, and Lafora body formation (Tagliabracci et al., 2008). However, since the findings of Nitschke et al. (2017) and Gayarre et al. (2014) revealed that hyperphosphorylation is not detrimental to glycogen structure, it is possible that laforin’s carbohydrate binding module, rather than its phosphatase domain, is more important to its function, and that the recruitment of malin to glycogen is crucial for the preservation of proper glycogen structure (Gayarre et al., 2014; Nitschke et al., 2017).

1.4.C. Malin (EPM2B)

Shortly after the discovery of laforin, the second causative gene was identified as EPM2B, the locus on chromosome 6p22.3. This gene, also referred to as NHLRC1 in publications, consists of one exon and encodes an E3 ubiquitin ligase called malin. Characteristic to E3 , malin has a RING domain at its N terminus, and six NHL domains that facilitate protein-protein

25

interactions at its C terminus (Chan et al., 2003). Although loss of function mutations in EPM2B leads to histopathological outcomes that are indistinguishable from that of EPM2A mutations, it has been reported that patients with mutations in EPM2B have a slower rate of LD progression (Singh et al., 2006).

Malin has been shown, mostly via in vitro experiments, to have several substrates that it can ubiquitinate. Its binding partner, laforin, was the one of the first targets to be identified, and has been shown to be polyubiquitinated (Gentry, Worby, & Dixon, 2005b). Malin and laforin form a functional complex and together modulate enzymes involved in glycogen synthesis. In addition, malin has been found to occur in greater quantities in cases where laforin is also present, indicating that laforin may stabilize malin (Vilchez et al., 2007).

The physiological function of the laforin-malin complex has been a subject of great speculation, especially pertaining to the mechanisms by which laforin and malin prevent glycogen malformation and LB accumulation. Upon binding, the laforin-malin complex is recruited to the glycogen particle via the carbohydrate binding module of laforin. Enzymes in the vicinity of the glycogen particle can then be downregulated by malin in a laforin-dependent manner. Despite the difficulties assessing malin’s targets in vivo due to lack of antibody, in vitro, co-expression experiments have identified several substrates. These include PTG and R6, both of which promote the dephosphorylation and activation of GS. Worby et al. (2008) found that overexpression of PTG in cell culture models led to glycogen accumulation. This glycogen accumulation was prevented, however, when PTG was co-expressed with both malin and laforin. Similarly, expression of R6 led to cellular glycogen accumulation, which was prevented by co- expression with malin and laforin constructs (Worby, Gentry, & Dixon, 2008). In the same study, Worby et al. (2008) showed that in order for PTG to be ubiquitinated, laforin must be present, and that PTG, like laforin, is degraded by the 26 S proteosome. A separate study was not only able to identify R6 as an interacting partner to laforin, but also that it is a target of the laforin-malin complex as it was ubiquitinated by malin and subject to autophagic degradation (Rubio-Villena et al., 2013). These findings indicate that one function of the laforin-malin complex is to regulate glycogen synthesis by downregulating indirect activators of GS.

26

The laforin-malin complex has also been found to downregulate glycogen synthesis directly via ubiquitination of GS. Transfection of cultured neurons with malin, PTG, and laforin expressing plasmids revealed that muscle GS levels were only lowered when laforin and malin were co- expressed (Vilchez et al., 2007). In addition, Western blot analyses of cultured neurons showed that protein levels of muscle GS are lower due to ubiquitination by malin when the laforin-malin complex is present. They noted that muscle GS levels increased upon treatment of cells with proteosomal inhibitors (Vilchez et al., 2007). Despite this evidence, GS protein levels have been shown to be unchanged in Epm2a or Epm2b knockout mice, where one would expect GS levels to be increased in the absence of a functional laforin-malin complex. Nitschke et al. (2017) proposed that since laforin favors longer glucan chains, one function of the laforin-malin complex may be to downregulate GS activity at certain chains of the glycogen molecule that are being hyperextended by GS. By this mechanism, malin could ubiquitinate GS in a laforin- dependent manner and prevent further elongation of the chain (Nitschke et al., 2017). Although the functions of the laforin-malin complex remain to be elucidated, it is certain that proper glycogen metabolism relies on the presence of both proteins.

27

Figure 1. Schematic showing structural differences between healthy glycogen and polyglucosans. Normally, glycogen molecules have short glycosidic chains and a uniform branching pattern, whereas polyglucosans formed in LD average longer chains and are less branched due to absent laforin or malin.

1.4.D. Lafora bodies

LD can be distinguished from other progressive myoclonus epilepsies based on the presence PASD-positive inclusions in patient brain, muscle, skin, or liver biopsies. These inclusions are called Lafora bodies (LB), and are pathognomonic of LD (Andrade et al., 2003). LB only elicit symptoms in the brain, despite their accumulation in organs outside of the central nervous system.

There are two types of LB which are classified based on their locations in the neuron. Type I LB are found in neuronal dendritic processes, while type II LB are found in the soma of neurons,

28

occupying the cytoplasm and pushing organelles to the periphery of the neuron (Striano et al., 2008). LB can be very large ranging from 3-40 microns, making their dense core and spherical shape easily visible with a light microscope (Minassian, 2001). LB are predominantly found in neurons, despite the greatest concentration of glycogen in the brain occurring in astrocytes. One group, however, has identified LB in astrocytes of 11-month-old malin KO mice, suggesting that LB accumulation in glial cells is possible (Valles‐Ortega et al., 2011)

LB are aggregates of insoluble, malformed glycogen particles termed polyglucosans that form over the course of the disease. Generally, the severity of disease symptoms corresponds to the amount of LB present in the brain; during the seemingly-healthy years of childhood, insoluble glycogen molecules gradually form and accumulate at a speed concomitant with the slow progression of disease (Sullivan et al., 2017). Once LB quantities surpass a threshold, however, the severe epileptic episodes of LD will begin, becoming increasingly intractable and degenerative as LB accumulate over time.

Using mouse models, several studies have demonstrated that the prevention of LB formation in LD mice can normalize glycogen levels and neurobehavioral phenotypes. Therefore, it has become increasingly clear that the debilitating epilepsy behind LD is caused by LB accumulation in neurons. Exactly how LB elicit seizures is unknown. One possible explanation was reported by Ortolano et al. (2014) after having discovered that all LB formed in 3- month old LD mice were in GABAergic inhibitory neurons. They also found that the population of GABAergic cortical neurons were generally reduced in the LD mice, which led them to the hypothesis that that the inhibitory regulation of the cortex may be impaired in LD, resulting in seizures (Ortolano, Vieitez, Agis-Balboa, & Spuch, 2014). Another hypothesis involves impaired uptake of glutamate from the synapse. Muñoz-Bellaster et al. (2016) found that in Epm2a-/- and Epm2b-/- mice, the glutamate transporter GLT-1 that is expressed primarily in astrocytes was no longer localized to the plasma membrane, and had a reduced capacity to uptake glutamate. It was then inferred that laforin and malin assist in the homeostasis of this transporter, which may be compromised in LD. These results suggested that excessive glutamate in the synapse due to deficiencies in the transporter at the plasma membrane may contribute to LD etiology (Muñoz-

29

Ballester, Berthier, Viana, & Sanz, 2016). Although the neurological mechanisms underlying seizure susceptibility in LD patients are unknown, it is accepted that LB formation compromises neurons and contributes to the epilepsy associated with LD.

1.4.E. Models of Lafora Disease Pathogenesis

For decades, a handful of LD research groups have been dedicated to unraveling the pathogenesis of LD in hopes of identifying possible avenues of therapy. Over time, two models of disease progression have been established. The first model describes the hyperphosphorylation of glycogen as a potential cause of LB formation and LD progression. The second theory not only catalyzed the re-evaluation of the role glycogen phosphate, but also demonstrated the necessity of the laforin-malin complex in regulating glycogen chain length patterns for preservation of glycogen solubility.

1.4.E.i. Glycogen hyperphosphorylation and LB formation

Once the function of laforin as a glycogen phosphatase was established, it was believed that the hyperphosphorylation of glycogen, as measured in Epm2a-/- and Epm2b-/- LD mouse models, was causative of glycogen insolubility and subsequent LB formation. Since LD mouse models lacking expression of malin also had hyperphosphorylated insoluble glycogen, the general theory became that one of the functions of the laforin-malin complex was to maintain physiological levels of covalently-bound phosphate (Tagliabracci et al., 2008).

Phosphomonoesters of glycogen are present in minute quantities, at the approximate rate of 1 phosphate for every 500-100 glucose residues (depending on the source of glycogen). They were first identified at the C2 and C3 carbons of glucose (Roach, Depaoli-Roach, Hurley, & Tagliabracci, 2012; Roach et al., 2012) . In 2013, Nitschke et al. not only identified C6 as another location for glycogen phosphate, but also found that C6 phosphorylation was elevated in glycogen extracted from muscle of both Epm2a-/- and Epm2b -/- mouse models. These mice had a 7-fold increase in G6P content in comparison to wild type muscle glycogen (Nitschke et al.,

30

2013). Hyperphosphorylation at the C6 position was also found to correlate with glycogen chain length. Exhaustive debranching followed by chain length separation revealed that LD glycogen contained a greater proportion of longer chains than wild type glycogen. These results suggested that excessive phosphorylation at C6 carbons results in reduced availability of branch points and therefore molecules with longer chains and compromised hydrophilicity (Nitschke et al., 2013).

Two years later, a study determined that the proportion of C6 phosphate in wild type or LD glycogen remained at approximately 20% of total glycogen phosphate. Also, elevations in C2 and C3 phosphorylation contributed much greater to the increase in total phosphate than C6 phosphorylation (DePaoli-Roach et al., 2015). Based on these results, the authors opposed the presumption made by Nitschke et al. (2013) and stated that the increase in C6 phosphate is not profound enough to affect the branching frequency of the entire glycogen molecule. Instead, DePaoli-Roach proposed that glycogen hyperphosphorylation could disrupt the hydrogen bonds between chains as well as other stabilizing factors of glycogen structure, and that the decrease in glycogen branching could be attributed to a separate, unknown mechanism (DePaoli-Roach et al., 2015).

The pathogenicity of excessive glycogen phosphate was directly investigated by Gayerre et al. (2014) when they generated a transgenic LD mouse model that expressed a version of laforin that retained carbohydrate-binding ability, but lacked phosphatase activity (Epm2a-/- .LAFC265S). They compared this transgenic line with their Epm2a-/- LD mice, and their pathology analyses revealed that the Epm2a-/- mice phenotypically accumulated an overwhelming amount of LB in the hippocampus, while the LD mice expressing phosphatase- inactive laforin had very few LB in the hippocampus by 12 months of age (Gayarre et al., 2014). Nitschke et al. (2017) supplemented this result by quantifying total phosphate bound to brain glycogen from these mouse lines, and found that the Epm2a-/-.LAFC265S had similarly high levels of brain glycogen phosphate as the Epm2a-/- mice. Since brain glycogen phosphate from Epm2a-/-.LAFC265S mice was just as high as the Epm2a-/- LD mice, it became evident that glycogen hyperphosphorylation is not the cause of glycogen malformation, insolubility, and LB formation in LD.

31

These recent studies have established that LB formation is not primarily caused by excessive phosphorylation, and that the phosphatase activity of laforin is not necessary for preventing LD. While there is no doubt that glycogen phosphate is implicated in LD, its role in both healthy and dysregulated glycogen metabolism requires further investigation.

1.4.E.ii. Abnormal Glycogen Chain Length Underlies Lafora Disease

Once glycogen hyperphosphorylation was no longer considered to be a contributing factor to glycogen malformation in LD, a new hypothesis was formulated from data that revealed a major structural difference between the LB glycogen and healthy glycogen: Nitschke et al. (2013) reported different chain length distributions of skeletal muscle glycogen from wild type mice, Epm2a-/- mice, and Epm2b-/- mice, where the glycogen from the LD mice were found to have a greater proportion of longer glucan side chains (Nitschke et al., 2013). To further investigate this finding, Nitschke et al. (2017) studied glycogen from the transgenic line of Gayerre et al. (2014), namely the Epm2a-/-.LAFC265S line of LD mice that express a phosphatase-inactive laforin, and compared their glycogen phosphate and chain length distributions to that of Epm2a-/- and Epm2b- /- mice. Despite the fact that Epm2a-/-.LAFC265S mice were found to have hyperphosphorylated muscle and brain glycogen, their chain length distributions were normal, and nearly identical to that that of wild type muscle and brain glycogen. The Epm2a-/- and Epm2b-/- mice also had hyperphosphorylated brain and muscle glycogen, but had abnormal chain length distributions that showed a greater proportion of longer glucan chains. Most importantly, Epm2a-/- and Epm2b-/- mice had plenty of LB accumulated in the hippocampus, while Epm2a-/-.LAFC265S did not (Nitschke et al., 2017). These results indicated that correcting the brain glycogen chain length distribution in LD mice is sufficient for preventing the formation of LB, and that excess glycogen phosphate is not the major contributor to LD pathogenesis.

The mechanism by which an abnormal chain length distribution, specifically long glucan chains, may cause water insolubility can be explained when drawing comparisons to amylopectin, the branched constituent of plant starch. While amylopectin is chemically identical to glycogen in that it contains α-1,4 and α-1,6 inter-glucose linkages, it is insoluble due to the organization of

32

glucan chains. Amylopectin has long chains that are arranged into clusters, where vicinal glucan chains form double helices with one another. These double helices stack to generate the semi- crystalline structure of starch, rendering the particle water-insoluble (Cenci et al., 2014). In the absence of laforin or malin, regions of the glycogen molecule may be elongated to the extent where adjacent chains collapse on each other and form water-excluding double helices. As a result, the solubility of the glycogen particle becomes compromised and will eventually cause precipitation. The accumulation of aberrant, insoluble glycogen contributes to LB formation and LD progression.

More work is necessary for determining exactly how laforin and malin prevent the hyperextension of glycogen chains. Nitschke et al. (2017) believe that laforin binds to and recruits malin to certain glycogen molecules that have been synthesized with longer than normal chains (Nitschke et al., 2017). Malin could then ubiquitinate GS to halt further activity, allowing for glycogen degrading enzymes to shorten the glucan chain. Nevertheless, GS ubiquitination by malin has only been demonstration in vitro. Supplementation of this finding with evidence from in vivo experiments may help solidify GS as a target for malin.

1.4.F. Animal Models of Lafora disease

Genetic manipulation of mice has allowed for the development of LD mouse models. LD mice are crucial for studying disease pathogenesis, investigating laforin and malin function, characterizing enzymes involved in glycogen metabolism, and for testing the therapeutic efficacy of small molecule compounds and gene therapies. Naturally occurring LD animal models do exist, and have been reported in certain breeds of dogs.

1.4.F.i. Mouse Models

Most characterized LD mouse models have a loss of function mutation in either laforin or malin genes. In 2002, Ganesh et al. generated a LD mouse model by inserting a neomycin resistance

33

cassette in exon 4, which encodes the dual specificity phosphatase domain of the laforin gene. Their targeted disruption of Epm2a prevented the expression or maturation of the transcript, generating Epm2a-null mutants. These mice acquired the neuropathological phenotype associated with LD: polyglucosan bodies accumulated in the brain and in extra-neurological organs, which stained positive for PASD, indicative of LB formation (Ganesh et al., 2002). Chan et al. (2004) later generated a LD mouse line expressing human EPM2A with a dominant- negative mutation (C266S) in the dual-specificity phosphatase domain of laforin. This led to the 100-fold expression of the transgene over native laforin in the mice, which became useful for revealing substrates of laforin (Chan et al., 2004). Insight into LD pathogenesis arose when Gayarre et al. (2014) designed a transgenic mouse model expressing phosphatase-inactive laforin in a Epm2a-/- background. The pathology analyses from this study revealed that the phosphatase activity of laforin is not necessary for preventing LB formation, and that the CBM may hold the greatest importance in laforin function (Gayarre et al., 2014). A first malin knockout mouse model was generated by DePaoli-Roach et al. (2010) through the deletion of the single exon of Epm2b, and its replacement with a LacZ-neo cassette. Within the same year, Turnbull et al. (2010) reported on their Epm2b-/- mice, generated by the replacement of the lone exon with a loxp-flanked, NeoR cassette. Malin and laforin deficient mouse models of LD not only demonstrate identical histopathological changes and the glycogen accumulation phenotype, but both models accumulate excess glycogen phosphate and have abnormal glycogen chain length distributions regardless of the missing protein. One limitation of LD mouse models is that the epileptic phenotype is difficult to characterize, as it is not nearly as severe and debilitating as that seen in LD patients.

1.4.F.ii. Canine Models of Lafora disease

LD is known to naturally occur in canines. Dog breeds that experience the greatest prevalence of LD include the Miniature Wirehaired Dachshund, the Beagle, and Bassett Hounds. Clinically, LD dogs present with generalized seizures, hypnic jerks, and photosensitivity, all of which are reported in humans. The onset of clinical symptoms varies, ranging from five months to 12 years of age. They also accumulate the characteristic LB in the brain and liver, spleen, muscle and

34

lymph nodes (Swain et al., 2017; Gredal, Berendt, & Leifsson, 2003). Common to these breeds is a dodecamer expansion repeat in the EPM2B gene. Interestingly, this mutation is unique to dogs and has not been reported in humans (Swain et al., 2017). It is likely that the frequent inbreeding of these types of dogs has led to the genetic predisposition for LD.

1.4.G. Molecular Targets for Lafora Disease Therapy

Through the genetic manipulation of mouse models, two proteins involved in glycogen metabolism have emerged as potential targets for LD therapy. The first, PTG, promotes glycogen synthesis, specifically by facilitating the dephosphorylation and activation of GS via PP1. The second protein is the chain elongating enzyme in glycogen synthesis, GS. Although it is a complex enzyme, it is the only protein capable of synthesizing glycogen. Since LB consist of malstructured glycogen, targeting the glycogen synthesis pathway could prevent LB formation and, in turn, the neurodegenerative epilepsy of LD. Presently, only knockout models have been generated to investigate the therapeutic potential of these proteins; however, several alternative means of downregulating PTG and GS, such as with small molecules or gene therapy, are in progress to identify the most feasible target for LD patients.

1.4.G.i. PTG

The glycogen-targeting subunit PTG is recognized as a therapeutic target for LD therapy. It was first evaluated by Turnbull et al. (2011, 2014) when PTG was knocked out in both Epm2a-/- and Epm2b-/- mice. In both LD mouse lines, there was a near complete amelioration of LB in the brain. In addition, brain glycogen measurements from double-knockout mice were normalized to wild type levels of glycogen, indicating that the glycogen-accumulation phenotype had been corrected. Lastly, the PTG depletion in LD mice also led to a neurobehavioral rescue, as their myoclonic activity (number of myoclonic jerks per minute) were equivalent to that of wild type mice (Turnbull et al., 2011; Turnbull et al., 2014) These results have not only identified a molecular target for LD, but have also lead to the evaluation of PTG as a target for other

35

glycogen storage diseases. In addition, they have ignited an investigation into the therapeutic potential of other glycogen-targeting subunits involved in glycogen synthase activation and glycogen synthesis.

The appeal of PTG as a pharmacogenetic target was further revealed in a 2011 study that identified a LD patient with a mild phenotype and slow progression of disease (Guerrero et al., 2011). This patient had an uncharacterized mutation in the PPP1R3C gene that affected the ability of PTG to interact with GS and GPh. To determine whether this variant influenced glycogen synthesis in vitro, HEK 293T cells were transfected with plasmids expressing either wild type PPP1R3C or the version of the gene with the novel variant. Glycogen measurements revealed that expression of the PPP1R3C mutation lead to the accumulation of half as much glycogen as that of cells expressing wild type PPP1R3C (Guerrero et al., 2011) . The correlation of this PPP1R3C mutation with a milder LD phenotype further suggests that the design of molecular and genetic therapies against PTG should continue to be developed.

1.4.G.ii. Glycogen Synthase

Although Gys1-/- genotypes have been reported to be embryonic lethal, 10% of pups were found to be viable by Pederson et al. (2004). Lethality was suggested to stem from several defects of the heart, a consistent phenotype among mice lacking Gys1 expression. To explain why a small proportion of pups survive, Pederson et al. (2004) reasoned that slight physiological variations between mice may determine which surpass critical developmental phases. It was also proposed, based off a study by Meeson et al. (2001), that mice lacking glycogen in the heart may undergo reprogramming of certain cardiac genes that promote survival as part of a compensatory response.

To test the therapeutic effect of whole-body GS downregulation, double-knockout mice were bred (Epm2a-/- /Gys1-/-) and compared to Epm2a-/- and wild type mice (Pederson et al., 2013). Mice were analyzed between the ages of 20-26 months, and histopathological results revealed a total absence of LB in the brain of double-knockout mice, as well as complete prevention of

36

astrogliosis, a characteristic neurophenotype of LD mice. Unlike the Epm2a-/- mice, double- knockout mice did not show seizure susceptibility following injections of kainic acid, a neurotoxin and excitatory agent (Pederson et al., 2013). These data were interesting, as it demonstrated that preventing glycogen synthesis in LD was sufficient for halting LB formation and disease progression.

Duran et al. (2014) also investigated GS removal in their malin-knockout LD mouse model, and found similar results. Pathology analyses of Epm2b-/-/Gys1-/- mice revealed no indications of gliosis from either microglial or astrocytic infiltration, as their hippocampus resembled that of wild type mice. Most importantly, results showed that LD mice deficient in GS also had no LB accumulation in the brain. This finding correlated with the brain glycogen measurements of these mice, which were close to zero (Duran et al., 2014). These data cumulatively suggest that GS is a molecular target for LD therapy.

Currently, yeast and bacteria GS structures have been elucidated (Buschiazzo et al., 2004; Chikwana et al., 2013b). This information may advance the design of small molecules that can downregulate the glycogen synthesis pathway. Also, the structural basis underlying the interaction of glycogenin and GS during glycogen synthesis has also been determined. This interaction may serve as an alternative route for the inhibition of glycogen synthesis (Zeqiraj et al., 2014). Since there are several glycogen storage diseases, each caused by a mutated glycogen metabolizing enzyme, identifying ways of downregulating glycogen synthesis could benefit an entire genre of inherited metabolic diseases. Although patients with homozygous mutations in GYS1 present with clinical symptoms pertaining to skeletal and cardiac muscle dysfunction, heterozygous patients and siblings are unaffected and healthy (Kollberg et al., 2007). Therefore, reducing the expression of GYS1 to 50% or slightly less of its original would be a safe measure for therapies to strive for.

Chapter 2 Evaluating the Glycogenic Activity of R6 and its Potential As A Therapeutic Target

37

38

2.1 Rationale

The therapeutic rescue conferred by knocking out PTG in LD mice confirmed that downregulating glycogen synthesis can slow LD progression by minimizing the accumulation of insoluble glycogen in the brain, thereby preventing LB formation. Results from studies evaluating the in vivo regulatory capacity of PTG demonstrated how indirect activators of GS can have a strong effect on glycogen synthesis. In addition, these studies engendered an investigation into the remaining regulatory subunits of PP1, particularly those that contribute to the regulation of glycogen synthesis in the brain. Among the other subunits expressed in the brain (R6, R3F, and R3G), R6 has been the only subunit to reportedly interact with and be ubiquitinated by the laforin-malin complex in co-expression experiments using cultured neurons. For this reason, this thesis will investigate the potency of R6 in regulating glycogen synthesis in the brain of LD mice, and determine whether its influence warrants its consideration as an additional target for LD therapies.

2.1.A. Aims

Five genotypes were generated through the breedings of double heterozygous, Ppp1r3d- /+/Epm2a-/+ mice: wild type (WT) mice (Ppp1r3d+/+/Epm2a+/+), R6 knockout (R6KO) mice (Ppp1r3d-/-), laforin knockout (LKO) mice (Epm2a-/-), laforin knockout, R6 heterozygous (LKO/R6 het) mice (Ppp1r3d-/+/Epm2a-/-), and double-knockout (DKO) mice (Ppp1r3d-/- /Epm2a-/-). This thesis utilized these cohorts of mice to conduct experiments in determination of:

i) The effects of R6 removal on glycogen accumulation and GS activation state in muscle and brain

ii) Whether LB formation and gliosis are reduced in brains of mice lacking R6 expression

iii) Whether R6 downregulation in LD mice reduces seizure susceptibility

39

Results from this study will be discussed in terms of whether these three aims were achieved. A general aim of this study is to contribute to the breadth of literature surrounding two fields of research: glycogen metabolism and LD.

My final aim is for this project to conclude on whether the R6 subunit should be pursued as a potential therapeutic target for the treatment and management of LD in patients. Although there are two proteins that appear to be the best targets for downregulation (PTG and GS), the identification of novel targets is essential for providing alternative prospects in the event that difficulties arise when designing or optimizing knockdown therapies against PTG or GS.

2.1.B. Hypotheses

Since R6 is a glycogen-targeting subunit of PP1, which is responsible for dephosphorylating and activating GS, it is possible that downregulating R6 in LD mice will increase the phosphorylation status of GS and lower its activation state. R6 expression in the brain allows for the prediction that it is involved in brain glycogen metabolism, specifically the modulation of glycogen synthesis. The extent to which its absence will affect GS activity and inhibit glycogen synthesis is unknown and will be explored in this thesis. I hypothesize that knocking out R6 in LD mice will lower glycogen synthesis in the brain, but not as dramatically as was achieved with PTG depletion in LD mice. If R6 downregulation reduces glycogen quantities in the brain, then it can be expected that LB formation in the brain will also be reduced.

If LB accumulation in the brain is significantly reduced upon genetic depletion of R6, seizure susceptibility in LD mice should also be attenuated, as LD mice with dramatic reductions in LB formation have been shown to be less affected in behavioural tests designed to record and induce myoclonic or epileptic tendencies (Turnbull et al. 2011; Turnbull et al. 2014; Duran et al. 2014).

R6 is highly expressed in muscle, and therefore, could participate in the mechanisms surrounding glycogen synthesis in this tissue. My hypothesis is that the effect on glycogen synthesis in muscle will be minimal, as Turnbull et al. (2014) have already shown that LD mice lacking PTG have normalized muscle glycogen quantities. In addition, the GM glycogen-targeting subunit is

40

the most expressed subunit in muscle, and mice deficient in GM expression have 80% less glycogen than WT mice (Delibegovic et al., 2003). Therefore, I predict that a deficiency in the R6 subunit will provide modest reductions in glycogen accumulation and LB formation in muscle.

41

2.2 Methods

2.2.A. Ethics Statement

All animal procedures and behavioural tests were approved by The Toronto Centre for Phenogenomics Animal Care Committee.

2.2.B. Generation of Ppp1r3d x Epm2a mouse line

The Ppp1r3d knockout mouse line was generated using the embryonic stem cell clone 11161A- E11 (produced by Regeneron Pharmaceuticals Inc.) and acquired from KOMP repository. The Velocigene targeting vector described by Valenzuela et al. (2003), which consists of the ZEN- Ub1 cassette encoding a lacZ reporter and neomycin selection gene, was inserted within the only exon of Ppp1r3d on chromosome 2 to produce a deletion mutation (Figure 1). Embryonic stem cells electroporated with the targeted deletion were injected into C57BL/6N strain blastocysts, which were implanted into pseudopregnant CD1 mice. Heterozygous chimeras were bred to generate Ppp1r3d knockout mice. These mice were bred to Epm2a-/- mice maintained by the Minassian lab to generate Ppp1r3d-/+/Epm2a-/+ breeders for generation of the five genotypes used analyzed in this study: WT, LKO, LKO/R6 het, DKO, and R6KO mice.

42

Figure 2. Schematic diagram of the disruption of Ppp1r3d through the insertion of a targeting vector. An 835 base pair deletion of exon 1 was achieved through the insertion of a ZEN-Ub1 cassette that consists of a lacZ-p(A) reporter and a loxP-flanked hUbCpro-neo-p(A) selection marker.

43

2.2.C. Generation of Ppp1r3c x Ppp1r3d x Gys1 mouse line

An additional mouse line was designed for the generation of five genotypes bred within a wild type background. First, Ppp1r3c-/+ mice were crossed with Gys1-/+ mice to generate double heterozygous mice, Ppp1r3c-/+/Gys1-/+. These mice were bred to Ppp1r3d-/+ mice to generate triple heterozygous (Ppp1r3c-/+/Gys1-/+/Ppp1r3d-/+) breeders. The genotypes selected from each litter for experimentation were Ppp1r3c+/+/Gys1+/+/Ppp1r3d+/+ (wild type; WT), Ppp1r3c+/+/Gys1-/+/Ppp1r3d+/+ (GS heterozygous; GS het), Ppp1r3c-/-/Gys1+/+/Ppp1r3d+/+ (PTG knockout; PTGKO), Ppp1r3c-/+/Gys1+/+/Ppp1r3d+/+ (PTG heterozygous; PTG het), Ppp1r3c+/+/Gys1+/+/Ppp1r3d-/- (R6 knockout; R6KO), and finally Ppp1r3c+/+/Gys1+/+/Ppp1r3d- /+ (R6 heterozygous; R6 het). Brain tissue extracts were prepared from mice of each genotype for utilization in the GS activity assay.

2.2.D. Glycogen Quantification

Mice were sacrificed by cervical dislocation. Brain and muscle tissue were harvested and placed immediately into liquid nitrogen and transported to -80°C freezer for storage. Frozen tissue was pulverized by mortar and pestle in liquid nitrogen, and aliquots of 50-90 mg were transferred onto dry ice. Each sample was boiled in 30% potassium hydroxide for one hour with frequent agitation. Samples were then precipitated in 67% ethanol and placed at -20°C overnight. Samples were pelleted, and the supernatant aspirated prior to suspension in 300 µl H2O. To initiate a second precipitation, ethanol and lithium chloride (67% ethanol and 15 mM lithium chloride final concentrations) were added to each sample and placed at -20°C overnight. Samples were precipitated a third time in a manner identical to the second. After pelleting and aspirating samples, each was resuspended in water. The extracted glycogen from each sample was digested at 55°C for 1 hour by amyloglucosidase in 80 mM sodium acetate to release glucose. As described by Lowry & Passonneau (1972), glucose was then quantified indirectly through enzymatic treatment with hexokinase and glucose-6-phosphate dehydrogenase to generate NADPH, the amount of which is proportional to the quantity of glycogen-derived glucose in each sample. Glycogen is quantified as µmol glucose/g fresh weight (FW) tissue.

44

2.2.E. Histopathology and Immunohistochemistry

Mice were sacrificed by cervical dislocation, and brains and muscle were immediately fixed in 10% neutral buffered formalin. Tissues rotated for 48 hours at room temperature and were then embedded in paraffin. Five microns of tissue was coronally sectioned, placed on glass slides and stained with hematoxylin and eosin and PAS, followed by a short diastase treatment (PASD).

Mice hippocampi were stained with anti-GFAP antibody to measure astrogliosis at a concentration of 1:200 (BioLegend, San Diego, CA, USA) or anti-Iba1 antibody, to visualize reactive micgroglia, at a concentration of 1:2000 (Wako Chemicals, Richmond, VA, USA) and counterstained with Harris’s Hematoxylin and eosin.

2.2.F. Lafora body, anti-GFAP, and anti-Iba1 Quantification

To quantify LB, PASD-stained glass slides were scanned using the 3DHistech Pannoramic 250 Flash II Slide Scanner and analyzed using 3DHISTECH digital pathology software. Specifically, the HistoQuant module of 3DHISTECH was used to detect and quantify LB. LB, which stain with a bright magenta color following PASD staining of tissue, can be detected by HistoQuant based on the hue and saturation values of the magenta color that contrasts with the lighter background of the slide. Using Pannoramic Viewer, the entire hippocampal section on the slide was selected as the region of interest to determine the effect of genotype on LB accumulation. Using size exclusion parameters, HistoQuant was manually set to exclude from analysis any objects on the slide that stained with PASD and were not LB, such as blood vessels within the hippocampus. HistoQuant quantified the area (in µm2) of LB occupying the hippocampal section, while the area of the hippocampus (µm2) was quantified by Pannoramic viewer. LB accumulation was finally measured as the percent of hippocampus occupied by LB:

LB accumulation (% of hippocampus) = (area of PASD stain/area of hippocampus) x 100

Anti-GFAP and anti-Iba1 stains were quantified in a similar fashion to LB quantification. Stained slides were scanned and examined with Pannoramic Viewer, and the HistoQuant module

45

of 3DHISTECH was used to detect antibody-bound regions of the hippocampus. Again, the entire hippocampal section was quantified and the proportion of staining was determined by the following formulas:

Anti-GFAP stain (% of hippocampus) = (area of anti-GFAP/area of hippocampus) x 100

Anti-Iba1 stain (% of hippocampus) = (area of anti-Iba1/area of hippocampus) x 100

2.2.G. Tissue Homogenate Preparation

Mice frozen tissue was pulverized in liquid nitrogen and aliquoted into 1.5 ml Eppendorf tubes. For every 10 mg of tissue aliquoted, 100 µl of homogenization buffer was added to the sample. 1 Tissue lysis was achieved by two rounds of repetitive agitation using 18G x 1 /2 BD PrecisionGlideTM needles, during which all samples were kept on ice. Samples were then rotated at 4°C for 45 minutes prior to centrifugation at 4°C, 10,000 rpm for 20 minutes. The supernatants were transferred to clean 1.5 ml Eppendorf tubes, and pellets stored at -80°C. Aliquots of sample supernatants were diluted 5x for determination of protein concentration using the Bio-Rad Protein Assay kit that is based on the Bradford method of protein quantification.

2.2.H. Glycogen Synthase Activation State Assay

The in vitro GS activity assay is designed to measure the incorporation of 14C radiolabeled glucose into glycogen by the following reaction:

UDP-[14C]glucose + glycogen  UDP + glycogen-[14C]glucose

Conducted in a 96-well plate, 20 nM of GS from mice tissue extracts was added to each well containing reaction mixture that consists of 2 mg/ml glycogen, UDP-[14C]glucose, and either 0.17 mM G6P (absence) or 8 mM G6P (presence). The purpose of conducting the glycogen elongation reaction in the absence or presence of G6P is to generate an activity ratio (- G6P/+

46

G6P) that can provide an interpretation of the phosphorylation status of GS. A lower activity ratio indicates that GS in the protein extract is more phosphorylated, and therefore, less active. In contrast, a higher activity ratio indicates that endogenous GS is less phosphorylated and therefore more active, and is closer to maximal activity (as achieved by 8 mM G6P in vitro).

The reaction plate incubated at 37°C for 30 minutes. Ethanol, at a final concentration of 67% was added to the reaction mixture to precipitate the glycogen, which was then transferred to a 96-well filter plate. The glycogen was washed 6 more times with ethanol to remove any unbound UDP- [14C]glucose. The reaction is left to incubate for 30 minutes at 37°C prior to the addition of scintillation fluid. The amount of [14C]glucose incorporated into glycogen is read using the Microplate scintillation and luminescence counter.

2.2.I. Kainic Acid Seizure Susceptibility Model

Kainic acid (Sigma-Aldrich) was prepared by dissolving 10 mg of powder with 1-2 drops of 1 N NaOH and bring the volume to 10 ml with sterilized water for a concentration of 1 mg/ml. Mice 12-16 months of age were weighed to determine dosage with 8 mg/kg kainic acid. Mice were injected intraperitoneally and placed in a cage. Timing began immediately and the reaction to kainic acid was videotaped for a maximum of 1.5 hours. Mice were monitored following injection and noted for the induction of myoclonic and seizure activity. Seizure severity was compared between genotypes and assigned a number corresponding to stage 0-5 on the modified Racine Scale (Table 1) (Pederson et al. 2014; Racine, 1972). Mice were euthanized by cervical dislocation once recordings were complete.

47

Table 1. Modified Racine scale for scoring seizures.

Stage Seizure Activity

0 Unaffected

1 Immobility; head bobbing

2 Myoclonic jerk within 5 minutes

3 Forelimb clonus and rearing; tail shaking

4 Convulsive seizure; continuous rearing and falling

5 Death

2.2.J. Myoclonus Quantification

Mice 12 months of age were placed in an empty cage and videotaped for three hours. Mice were undisturbed and free to roam within the cage. Periods of low movement were closely monitored for myoclonic jerks. In addition, periods where mice were sleeping were also used to determine frequency of myoclonic movements. A total of 10 minutes was watched for each mouse. The observer was blind to genotype. Myoclonic activity was quantified as jerks/minute (Turnbull et al. 2011; Turnbull et al. 2014).

2.2.K. Statistics

Unless described otherwise, ordinary one-way analysis of variance (ANOVA) was used to compare the means of each genotype and identify whether any statistically significant relationships exist among independent groups. Following ANOVA, an unpaired student’s t-test with Welch’s correction was employed to quantify significance between two groups. Asterisks

48

are used to indicate the degree of statistical significance based on the P value: **** p <0.0001, *** p < 0.001, ** p < 0.01, and * p < 0.5. Error bars indicate standard deviation of the mean. All analyses were conducted using GraphPad Prism, version 7.04.

49

2.3 Results

2.3.A. Brain glycogen content is reduced in DKO and LKO/R6 het mice

Glycogen was quantified and compared between males and females of each genotype to determine the effect of R6 on glycogen synthesis in the brain (Figure 2A). As expected, LKO mice displayed the glycogen accumulation phenotype of LD mice, and had roughly 7 times more glycogen in the brain than WT mice (p < 0.0001). While the glycogen content in LKO/R6 het mice was increased five-fold in comparison to WT, this value was significantly lower than the amount of glycogen accumulated in LKO brain (p = 0.0001), at nearly half of its value. Interestingly, LKO/R6 het mice glycogen levels did not differ significantly from that of DKO mice which, notably, had more than a 50% reduction in glycogen in comparison to LKO mice (p < 0.0001). Despite the significant reduction in brain glycogen in DKO and LKO/R6 het mice in comparison to LKO mice, R6KO glycogen levels did not differ significantly from WT levels. No significant differences in glycogen between sexes was found (Figure 2B).

50

A *** ****

****

) W

F 15

g

/

e

s o

c 10

u

l

g

l

o m

 5

(

n

e

g

o c

y 0 l

G t T O O e O W K K h K 6 D L R 6 /R O K L

B )

W Male

F 15

g Female

/

e

s o

c 10

u

l

g

l

o m

 5

(

n

e

g

o c

y 0 l

G t T O e O O W K h K K L D 6 6 R /R O K L

Figure 3. Brain glycogen measurements in 14-month-old mice. (A) Glycogen quantities for all mice and (B) mice separated by sex. N=6-8 for (A) and n=2-5 for (B). ***, p < 0.001; ****, p < 0.0001. Error bars are standard deviation (SD) of the mean.

51

2.3.B. The knockout of one or both Ppp1r3d alleles leads to lower GS activation states in the brain

Considering that R6 is a member of the glycogen-targeting subunits of PP1, whose role it is to remove inhibitory phosphates from GS, the GS activation assay was conducted to determine the effect of R6 removal on the phosphorylation status of GS. Changes in phosphorylation status influence the activation state of GS, which is quantified using the activity ratio (-G6P/+G6P). Comparing the activity ratio among genotypes can be used to infer whether GS is, on average, more or less phosphorylated in a certain genotype.

In the brain, R6KO mice had a significantly lowered GS activity ratio than WT mice, indicating that GS from R6KO brain tissue is more phosphorylated, on average (p = 0.0320). While LKO/R6 het and DKO mice had comparable GS activity ratios, both were significantly lower than that of LKO mice (p = 0.0028 and 0.0013, respectively). These results indicate that PP1-R6 does affect the degree to which GS is phosphorylated, and consequently, its activation status in the brain (Figure 3A).

2.3.C. R6KO and PTGKO mice have lower GS activation states than WT mice, but do not significantly differ from each other

The activation status of GS was significantly lowered in PTGKO (p = 0.0005) and R6KO (p = 0.0009) mice brains in comparison to WT mice. However, the GS activity ratios of PTGKO and R6KO did not differ from each other, indicating that, on average, GS from these samples were phosphorylated and inhibited to similar degrees in the brain. GS het mice also had a significantly lowered activity ratio in comparison to WT mice (p = 0.0190). Interestingly, PTG het and R6 het genotypes showed no difference in GS activity ratio in comparison to WT mice, despite the highly significant reductions experienced by PTGKO and R6KO mice (Figure 3B).

52

A

**

* **

) P

6 0.15

G

+

/

P

6 G

- 0.10

(

o

i

t

a

r

y 0.05

t

i

v

i

t

c

a

S 0.00

G t T O O e O W K K h K 6 D L R 6 /R O K L

B

*** ***

) * P

6 0.15

G

+

/

P

6 G

- 0.10

(

o

i

t

a

r

y 0.05

t

i

v

i

t

c

a

S 0.00

G t t t T e O e O e W h K h K h S G G 6 6 G T T R R P P

Figure 4. GS activity ratios from brain homogenates generated by the conduction of the GS activity assay in the absence (-) and presence (+) of G6P. (A) Activity ratios generated using brain homogenates of mice bred from the crossing of the Ppp1r3d -/+/Epm2a -/+ mouse line, where mice were 14 months old and n=3-6 per group. (B) Activity ratios generated using brain

53

homogenates from the mouse line described in section 2.2.C. Mice were sacrificed at 1-3 months of age, and n=3-6. Error bars are SD of the mean. *, p < 0.05; **, p < 0.01; ***, p < 0.001.

54

2.3.D. LB accumulation is reduced in the brain of LD mice lacking R6 expression

The evaluation of R6 as a therapeutic target requires investigation into whether its absence in the brain can reduce the amount of accumulated LB, and therefore, slow down the progression of LD. LB that formed in the hippocampus of mice were quantified and expressed as the percent of the hippocampal area occupied by PASD-stained inclusions. As expected, WT and R6KO mice had low and comparable levels of PASD staining (Figure 4). In contrast, LKO mice averaged 7% of the hippocampus occupied by LB and was highly significant in comparison to WT mice (p < 0.0001). The LB accumulation in DKO mice was significantly lower than that in LKO mice (p = 0.03), and averaged 5% of the hippocampus. Despite the significant reduction in brain glycogen in LKO/R6 het mice in comparison to LKO mice (Figure 2A), the amount of LB accumulation did not differ significantly between these two genotypes. Figure 5 displays histopathology images of each genotype, which were captured using Pannoramic Viewer.

55

*

) s

u ****

p m

a 10

c

o p

p 8

i

h

f o

6

%

(

n 4

o

i

t a

l 2

u

m u

c 0 c

a T t O O e O W K K h K B 6 D 6 L

L R /R O K L

Figure 5. LB accumulation quantified by % of hippocampus occupied by PASD stain. Mice were sacrificed at 14 months for PASD staining. Mean LB accumulation is displayed for each genotype, and bars represent SD. N=5-7 mice. *, p < 0.05; ****, p < 0.0001.

56

57

Figure 6. Representative images of PASD-stained hippocampi. Images were acquired of 14- month-old (A) WT, (B) LKO, (C) LKO/R6 het, (D) DKO and (E) R6KO mice at 20x magnification. Scale bars in the top left corner of images are 100 µm

58

2.3.E. DKO and LKO/R6 het mice are not rescued from gliosis

Several groups have shown that gliosis, incited by the accumulation of neurotoxic LB, is a major component of LD mice neuropathology (Pederson et al. 2013; Turnbull et al. 2011; Duran et al. 2014). For this reason, mouse brains were immunostained with two different neuropathological markers, anti-Iba1 and anti-GFAP, to determine whether reductions in LB accumulation would result in lower levels of gliosis.

The amount of astrogliosis was quantified as the percent of the hippocampus occupied by anti- GFAP stain (Figure 6A). While LKO mice demonstrated a significant upregulation of glial scar formation in comparison to WT mice (p = 0.0074), the average percent of their hippocampi stained with anti-GFAP was similar to DKO and LKO/R6 het mice. Although the average quantity of anti-GFAP staining was lower in LKO/R6het mice, this value does not differ significantly from LKO or DKO mice. Lastly, R6KO mice had comparable levels of anti-GFAP stain to WT mice. Representative images of each genotype are shown in Figure 7.

The amount of anti-Iba1 was also quantified to determine whether mono- or biallelic knockout of R6 could limit microglial activation in the brain (Figure 6B). Microgliosis was significantly upregulated in LKO mice in comparison to WT mice (p = 0.0349). The quantity of anti-Iba1 stain was similar between DKO and LKO/R6 het mice. Again, standard deviation measurements revealed a high variability in staining between mice of each genotype. Representative images of each genotype are shown in Figure 8.

59

A

**

) s

u 30

p

m

a

c o

p 20

p

i

h

f

o

% (

10

P

A

F

G

- i

t 0 n

a t T O O e O W K K h K 6 D L R 6 /R O K L

B

*

) s

u 15

p

m

a

c o

p 10

p

i

h

f

o

% 5

(

1

a

b

I

-

i t

n 0 a t T O O e O W K K h K 6 D L R 6 /R O K L

Figure 7. Quantification of gliosis in the hippocampus. Gliosis was quantified as % of hippocampus stained with (A) anti-GFAP and (B) anti-Iba1 antibody. All mice were sacrificed at 14 months-old, where n= 5-11. Error bars are SD of the mean. *, p < 0.05; **, p < 0.01.

60

61

Figure 8. Representative images of the hippocampus following anti-GFAP immunolabelling. Images are of (A) WT, (B) LKO, (C) LKO/R6 het, (D) DKO, and (E) R6KO taken at 20x magnification. Scale bars in the top left corner of images are 100 µm.

62

63

Figure 9. Representative images of the hippocampus following anti-Iba1 immunolabelling. Images are of (A) WT, (B) LKO, (C) LKO/R6 het, (D) DKO, and (E) R6KO taken at 20x magnification. Scale bars in the top left corner of images are 100 µm

64

2.3.F. R6 knockout does not reduce seizure susceptibility of LD mice

Due to LB accumulation in the brain, LD mice have been reported to have spontaneous myoclonus and display epileptic activity in response to seizure-inducing agents like kainic acid. Although this phenotype is not as strong as what is seen in LD patients, the behavioural tendencies of LD mice can be sufficiently quantified. Spontaneous and induced seizure activities were measured to determine whether knocking out R6 could improve the behavioural phenotypes of LD mice in this study.

Analyses of unprovoked myoclonus activity revealed that LKO mice had significantly greater counts of myoclonic jerks per minute than WT mice (p = 0.0101). Both R6KO and WT mice demonstrated comparable levels of myoclonus. Not only did DKO and LKO/R6het mice have similar levels of myoclonus, but neither genotype differed significantly from LKO mice myoclonic activity (Figure 9A).

The second test of seizure susceptibility involved intraperitoneal injections of kainic acid. LB accumulation in the brain lowers the seizure threshold of LD mice, increasing their sensitivity to seizure-inducing compounds. Epileptic episodes of mice were scored on a modified Racine scale, as described in section 2.2.H. of Methods. As displayed in Figure 9B, WT mice remained in either stage 1 or 2, while all R6KO mice had category 2 seizures. Five of the 6 DKO stayed in stage 2, with one mouse reaching stage 3 on the modified Racine scale. Half of the LKO/R6 het mice had stage 2 seizures, while the remaining half had stage 3 seizures. Finally, one LKO mouse had a stage 2 seizure, two entered stage 3 seizures, and one LKO mouse had a category 4 seizure. No mice from any genotype reached stage 5 on the modified Racine scale. While statistical significance was only achieved between medians of WT and LKO mice (p = 0.0319), and upward trend in seizure severity is visible among genotypes.

65

A

*

) n

i 2.0

m

/

s

k r

e 1.5

j

(

y

t

i v

i 1.0

t

c

A

c

i 0.5

n

o

l

c o

y 0.0

M t T O O e O W K K h K 6 D L R 6 /R O K L

B

*

5

4

e

g a

t 3

S

e n

i 2

c

a R 1

0 t T O O e O W K K h K 6 D L R 6 /R O K L

Figure 10. Seizure susceptibility tests measuring myoclonus activity and Racine stage performance. (A) Myoclonic activity was recorded at 12 months of age, where n=6-8 per genotype. Bars represent SD of the mean, and *, p < 0.05. Kainic acid testing (B) was conducted on 12-16-month-old mice. Bars represent the median stage achieved by mice from each genotype, where n=4–7 mice. Statistical significance amongst medians of each group was

66

determined using the Kruskal-Wallis ANOVA, and the post hoc selected was the Mann-Whitney Test.

67

2.3.G. Lack of R6 expression does not alter glycogen levels in skeletal muscle of male mice

Armstrong et al. (1997) confirmed that R6 is not only ubiquitously expressed, but that its expression is greatest in skeletal muscle. The effect of R6 removal on muscle glycogen synthesis was determined (Figure 10A). When comparing male mice only, LKO mice differed significantly and displayed a 6-fold increase (p = 0.0039) in comparison to WT mice. Not only were DKO glycogen quantities comparable to LKO/R6 het mice, but both genotypes did not differ significantly from LKO mice. In addition, these three genotypes all showed great variability in glycogen quantities. Lastly, R6KO mice had glycogen quantities comparable to WT mice.

2.3.H. Skeletal muscle glycogen measurements are strikingly different between sexes of LKO, LKO/R6 het, and DKO mice

Originally, females from each genotype were included in muscle glycogen analyses. However, a sex difference was noticed, specifically among the genotypes that would be expected to display a glycogen accumulation phenotype. In Figure 10B, LKO, LKO/R6 het, and DKO males all have similar glycogen levels. The female mice from LKO/ LKO/R6 het, and DKO genotypes had quantities that were 5.4, 5.9, and 7-fold lower than their male counterparts, respectively. The reason for this large discrepancy in glycogen quantities between males and females of LKO, LKO/R6 het, and DKO genotypes is unknown.

68

A

**

) W

F 150

g

/

e

s o

c 100

u

l

g

l

o m

 50

(

n

e

g

o c

y 0 l

G t T O O e O W K K h K 6 D L R 6 /R O K L

B

) Male W

F 150

Female

g

/

e

s o

c 100

u

l

g

l

o m

 50

(

n

e

g

o c

y 0 l

G t T O e O O W K h K K L D 6 6 R /R O K L

Figure 11. Skeletal muscle glycogen levels. (A) Glycogen quantification for males and (B) between sexes of each genotype. All mice are between 12-14 months of age, where n=5-7 for (A) and n=3-5 for (B). Bars are SD of the mean. **, p < 0.01.

69

2.3.I. GS activation state is unchanged in muscle of DKO and LKO/R6 het mice

The GS activity ratio in R6KO mice was significantly reduced in comparison to WT mice (p = 0.0123), indicating that PP1-R6 does contribute to the phosphorylation status of GS in muscle. Nonetheless, the GS activity ratios of LKO, DKO, and LKO/R6het genotypes were all comparable (Figure 11).

70

*

) P

6 0.4

G

+

/ P

6 0.3

G

-

(

o

i 0.2

t

a

r

y t

i 0.1

v

i

t

c

a

S 0.0

G t T O O e O W K K h K 6 D L R 6 /R O K L

Figure 12. GS activity ratios from muscle homogenates generated by the conduction of the GS activity assay in the absence (-) and presence (+) of G6P. Muscle homogenates from 14- month-old mice were used for the assay, where n=3-6, and *, p < 0.05. Bars are SD of the mean.

71

2.3.J. LB accumulation is comparable between males of DKO, LKO/R6 het, and LKO genotypes

LD mice are known to accumulate LB in their skeletal muscle. LD mice lacking R6 partially or completely did not experience a reduction in LB, as indicated by the PASD-stained skeletal muscle images in Figure 12. The comparable levels of LB in muscle correlates with the lack of significant difference in muscle glycogen measurements of LKO, DKO, and LKO/R6 het mice. PASD images are of male mice, which was the sex chosen from which to conclude on the effects of R6 on LB formation in muscle. As expected, R6KO and WT mice had no LB form in muscle, male or female.

72

73

Figure 13. PASD-stained skeletal muscle images. Representative images are of (A) WT, (B) LKO, (C) LKO/R6 het, (D) DKO and (E) R6KO male mice. Images were acquired at 20x magnification. Scale bars in the top left corner of images are 100 µm.

74

2.3.K. Male and female mice have different quantities of LB in skeletal muscle

Male and female mice of LKO, DKO, and LKO/R6 het mice accumulated noticeably different quantities of LB in skeletal muscle, where male mice show large amounts of LB and females have much less (Figure 13). All mice were stained for PASD identically, and all mice are 14 months of age with correct genotypes.

75

Figure 14. Sex differences in the amount of LB accumulated in skeletal muscle. PASD- stained images were acquired at 20x magnification. Scale bars in the top left corner of the image are 100 µm.

76

2.3.L. LB are reduced in the hearts of LKO/R6 Het and DKO male mice

Comparing LB accumulation in the heart revealed that LKO/R6 het and DKO male mice had noticeably less LB form than LKO mice. As expected, WT and R6KO mice did not have any LB in the heart (Figure 14).

77

78

Figure 15. PASD-stained images of the heart. Representative images are of (A) WT, (B) LKO, (C) LKO/R6 het, (D) DKO and (E) R6KO male mice. Images were acquired at 20x magnification. Scale bars in the top left corner of images are 100 µm.

79

2.3.M. Male and Female LKO mice differ in LB quantities in the heart

Analysis of male and female LKO mice LB revealed differences in LB quantities in cardiac muscle as well. Males can be seen with overwhelming amounts of LB, whereas females appear to have less LB formed (Figure 15). Less obvious are the differences between males and females of LKO/R6 het and DKO mice, as a reduction in LB accumulation was achieved in males.

80

Figure 16. Sex differences in amount of LB accumulated in heart. PASD-stained images were acquired at 20x magnification. Scale bars in the top left corner of images are 100 µm.

81

2.4 Discussion

This study was the first to design a Ppp1r3d-/- mouse model and investigate the glycogenic activity of the R6 glycogen-targeting subunit in Epm2a-/- mice. The primary goal of this project was to determine the effects of mono- and biallelic Ppp1r3d knockout on glycogen accumulation, LB formation, and behavioural phenotypes of LD mice. Furthermore, the effectiveness of R6 as a therapeutic target for is also to be discussed, as well as several comparisons to PTG (R5), a closely related PP1 subunit.

2.4.1. The effects of R6 on glycogen metabolism

The intrigue surrounding R6 and its involvement in the molecular mechanisms of LD was sparked by the recent findings of Rubio-Villena et al. (2013). Using a neuronal cell model, the results of a co-immunoprecipitation experiment revealed an interaction between laforin and R6. Additionally, co-expression of laforin, R6, and malin in HEK 293 cells revealed that R6 is a ubiquitinated by malin in the presence of laforin. Following the overexpression of R6 in their neuronal cell model, Rubio-Villena et al. (2013) reported increased levels of glycogen in relation to their controls, and that this increase was due to endogenous GS activation in cells. Importantly, the accumulation of glycogen in neurons was reversed once laforin and malin were also overexpressed, further suggesting that R6 is an additional target of the laforin-malin complex. This experimental evidence solidified the likelihood that R6 contributes to mechanisms underlying glycogen synthesis in the brain, instigating our investigation into the glycogenic properties of R6 in the context of LD mice. Knocking out R6 was deemed an appropriate means of determining its influence on GS activity and glycogen deposition, and evaluating whether it could rescue the phenotypes associated with LD.

Biochemical analyses of glycogen accumulation in the brain revealed that LD mice lacking whole-body expression of R6 (DKO mice) accumulated 60% less glycogen than LKO mice, while LKO/R6 het mice achieved a near 50% reduction in brain glycogen. These data not only confirm the expression of R6 in the brain, but that R6 contributes to regulation of glycogen

82

synthesis, likely through modulation of GS activity. Primarily, these results indicate that knocking out R6 has the capacity to lower amounts of insoluble glycogen in the brain of aged LD mice. Interestingly, R6KO mice had similar levels of glycogen to WT mice, despite the prominent reduction of glycogen in DKO and LKO/R6 het mice. It is possible that soluble (non- pathogenic) glycogen degrades after cervical dislocation of mice and prior to snap-freezing of tissue, leaving behind solely insoluble glycogen for measurements. Consequently, glycogen quantities between R6KO and WT mice are comparable due to similar amounts of insoluble glycogen. Comparing this result with PTG KO mice from Turnbull et al. (2014) reveals that PTG is the stronger regulatory subunit of GS, as PTG KO brain glycogen quantities were reduced by nearly a third in comparison to WT mice. At present, PTG appears to be the main GS activator in the brain and the most promising therapeutic target for LD. Exploration of the two remaining PP1 glycogen-targeting subunits expressed in the brain, R3F and R3G, would provide further insight into which subunit has the greatest role in regulating glycogen synthesis in the brain.

In 2008, Worby et al. were the first to show that increased glycogen synthesis in CHO-IR cells, triggered by R6 or PTG expression, could be prevented through simultaneous expression with laforin and malin. Interestingly, this result was not replicated when GL, the major regulatory subunit in liver, was expressed. Here arose the possibility that laforin and malin function to regulate proteins a) ubiquitously expressed and b) found in the brain, where glycogen accumulation proves toxic to neurons. Potentially, the laforin-malin complex serves as a safeguard mechanism during glycogen metabolism, with only specific subunits, like PTG and R6, being substrates for ubiquitination.

The studies surrounding PTG in LD mice revealed that PTG, when genetically removed from either Epm2a-/- or Epm2b-/- mice, prevents the typical glycogen accumulation phenotype in brain and restores glycogen quantities to WT values. Knocking out R6 in Epm2a-/- mice, however, revealed that R6 has nearly half the regulatory strength of PTG: DKO mice from this study accumulated slightly less than half of the glycogen quantity of LKO mice, a value still 4-fold greater than WT quantities. Evidently, PTG and R6 differ in their ability to regulate GS and glycogen synthesis, which could be attributed to slight variations in their mechanisms of action. More research in this area is needed to unearth these differences and broaden our understanding

83

of glycogen metabolism. Perhaps, one reason for this difference is that PTG and R6 target PP1 to different phospho-serine residues on GS. Out of the nine phosphorylation sites of GS, certain sites have a stronger influence on GS activity. PTG could recruit PP1 to the critical serine residues that are known to increase GS activation the most when dephosphorylated, while PP1- R6 targets the more subordinate phosphor-serine sites on GS that have little effect on GS activity. An alternative explanation involves the possibility that the PTG subunit, rather than R6, is more responsive to extracellular signals in the brain that promote glycogen synthesis. For instance, Allaman et al. (2000) used mouse cortical astrocytes to determine that the neurotransmitters norepinephrine and vasoactive intestinal peptide induced the expression of PTG, following their short-term glycogenolytic effects in the brain. Mice lacking PTG may lose the ability to resynthesize glycogen in the long-term following stimulation by extracellular signals, explaining the normalization of glycogen levels in PTG-lacking LD mice. In addition, a study by Ruchti et al. (2016) prepared astrocytic cultures from PTG KO mice and WT mice to further investigate the role of PTG in brain glycogen metabolism upon stimulation with extracellular agents. When both astrocytic cultures were untreated, glycogen from PTG KO mice were roughly 100-fold lower than wild type cultures. Six hours following treatment with noradrenaline, however, the glycogen content from wild type astrocytic cultures increased 10- fold, while that of PTG KO culture was unchanged (Ruchti et al., 2016). These experiments suggest that extracellular stimulants that initiate glycogen synthesis, either in the short or long- term, may function through PP1-PTG. Considering that R6 has an additional consensus sequence for regulation by 14-3-3 proteins that is not found on PTG, one could reason that R6 promotion of glycogen synthesis is induced more so by intracellular, protein-protein interactions. These are possible explanations for the difference in regulatory capacities of PTG and R6 on GS activity and brain glycogen metabolism.

In this study, GS activity assays were conducted using brain homogenates to determine whether the reductions in brain glycogen accumulation in DKO and LKO/R6 het mice could be due to changes in the activation state of GS. Prior to executing this experiment, it was predicted that disrupting the PP1-R6 partnership would affect the phosphorylation status and activation of GS. The GS activity ratios of DKO and LKO/R6 het were significantly lower than that of LKO mice, indicating that GS from these genotypes had a greater inhibitory phosphorylation status than

84

LKO. Evidently, R6 contributes to the covalent regulatory mechanisms of GS by promoting its dephosphorylation through PP1 recruitment. The reduction in GS activation state achieved by R6 could be the main mechanism by which glycogen synthesis was reduced in the brains of DKO and LKO/R6 het mice. In cases where LB form, GS becomes trapped within the malformed branches of insoluble glycogen and accumulates. This finding has been confirmed though Western blot, where increased GS protein was found in the insoluble fraction of Epm2b-/- mouse brain homogenates in comparison to control mice. Interesting, similar levels of GS are found in the soluble fraction of both (Valles-Ortega et al., 2011). This extra GS does not affect GS activity or activity ratios, since it is inert. Therefore, the GS activity ratio generated for LKO mice in this study is certainly quantified from active GS molecules that still reside within the soluble component of brain homogenates.

With confirmation that R6 does regulate the phosphorylation status, and therefore activity, of GS, we sought to compare PTG and R6 in their ability to reduce the GS activity ratios in 1-3- month-old that will not develop LD. We reasoned that PTGKO mice would have the lower GS activity ratio, since PTG appears to have the greatest influence on brain glycogen. The in vitro GS activity assay revealed that, as expected, PTGKO and R6KO mice had significantly reduced GS activity ratios in comparison to WT mice. Interestingly, PTGKO and R6KO were comparable in their activity ratios. One possible reason for this result is that the GS from brains of each genotype is phosphorylated to a similar degree. Perhaps the use of a more sensitive assay may help in distinguishing slight differences in GS activity in these genotypes, and assist in determining whether GS activation is more dependent on PP1-PTG or PP1-R6 dephosphorylation. The GS activity ratios of WT and GS het mice also differed, with the ratio being lower in GS het brain homogenates. Since the dose of GS is lower in the GS het mice, the lower activity ratio could be due to the less GS molecules available for dephosphorylation, causing the average phosphorylation status of active GS molecules to be larger in the GS het mice.

When comparing the biochemical changes in the brain recorded by the present study and those of Turnbull et al (2011, 2014), it is possible to conclude that PP1-PTG is the greatest regulator of glycogen synthesis in the brain, and that R6 is secondary to PTG. Both subunits are responsible

85

for regulating the reversible phosphorylation of GS in the brain, and that knocking out R6 in LD mice is not sufficient for normalizing brain glycogen levels.

LD patients and mouse models accumulate LB in their muscle. Despite the greatest expression levels of R6 being in skeletal muscle, as determined by Armstrong et al. (1997), glycogen measurements revealed that LKO, DKO and LKO/R6 het male mice all had similar quantities of glycogen. Therefore, R6 deficiency did not prevent or reduce glycogen accumulation in these mice. While the GS activity ratio of R6KO mice was significantly lower than that of WT mice, this decrease in the GS activation state was possibly not sufficient for correcting the dysregulation of glycogen synthesis in muscle to prevent the glycogen accumulation phenotype in DKO and LKO/R6 het genotypes. The activity ratios of LKO, DKO, and LKO/R6 het mice were all similar, and may explain why glycogen quantities were unchanged. This contrasts with the muscle glycogenic effects of PTG deficiency in both Epm2a-/- and Epm2b-/- LD mouse models. Turnbull et al. (2014, 2011) showed that knocking out PTG in laforin deficient mice led to the normalization of glycogen levels, whereas PTG deficiency in malin knockout mice lowered the glycogen content only by one third, still significantly greater than WT levels. PTG KO mice, on the other hand, did not differ from WT glycogen levels, which is interesting, considering PTG-laforin DKO mice had normalized glycogen levels (Turnbull et al., 2014). It seems as though the only subunit to significantly lower muscle glycogen quantities passed wild type levels is the GM knockout mice. Disruption of the Ppp1r3a gene encoding GM led to a massive reduction in muscle glycogen, achieving an 80% decrease in comparison to wild type mice (Delibegovic et al., 2003). Therefore, the PP1-GM, and to a lesser extent PP1-PTG holoenzymes, appear to have the greatest influence on GS regulation and glycogen synthesis in muscle. Possibly, R6, despite its expression in skeletal muscle, plays a minor role in regulating glycogen synthesis and GS activity. Additionally, PP1-R6 may target enzymes aside from GS that are also regulated by reversible phosphorylation and are involved in glycogen metabolism. Future studies are required to identify other targets of PP1-R6.

The effects of R6 on muscle glycogen measurements have also been investigated using in vitro, cell culture experiments. One study overexpressed R6, PTG and GM in human skeletal myotubes and saw that PTG overexpression caused the greatest change in glycogen accumulation, followed

86

by R6 and finally by GM. This is an interesting result, considering that in vivo experiments have determined that GM is strongest regulatory of glycogen synthesis in muscle, and that regulatory capacity of R6 on glycogen accumulation, as demonstrated by this project, is weak. Evidently in vivo experiments appear to be the greatest tool for determining the role of a protein in glycogen metabolism, and that those data generated from cell culture experiments must be supplemented with additional in vivo analyses.

During the analysis of muscle glycogen measurements, it became obvious that a sex difference existed between our mice, possibly arising from several generations of inbreeding of the mouse line used for this project. To circumvent this unpredictable event, only male mice were used to evaluate the glycogenic activity of R6 in muscle. Although glycogen was not measured in the heart, its possible that this organ was also affected by this sex difference. Section 2.6.2 in Future Directions will further report on the biochemical and histopathological differences found between sexes.

Results from this study have revealed the difference in strength of GS regulation of PTG and R6, not only in the brain, but also in skeletal muscle. What is the purpose of multi-subunit expression in glycogen-storing organs? The conservation of R6 expression in muscle among species indicates a functional importance. Evidently, more research is needed to differentiate the glycogen-targeting subunits from one another.

2.4.2. The effect of R6 deficiency on the histopathological and seizure susceptibility phenotypes of LD mice

The gradual accumulation of LB in neurons is associated with the progression of disease in LD patients, specifically the worsening epilepsy and decline in cognitive functioning. Designing therapies that inhibit LB formation in the brain are of utmost importance in LD research, the development of which depend on the identification of molecular targets. Currently the two most attractive targets for knockdown in patients are PTG and GS, since knocking out GS or PTG has been shown to inhibit LB formation and correct behavioural abnormalities associated with LD

87

mice (Pederson et al., 2013; Turnbull et al., 2011; Turnbull et al., 2014). By knocking out R6 in a LD mouse model, this study aimed to investigate the resulting histopathological and behavioural changes in mice, and whether R6 should be considered as an additional molecular target in the treatment of LD.

LB quantification in the brain indicated that DKO mice accumulated less PASD-positive inclusions than LKO mice. Specifically, DKO mice had 5% of their hippocampus occupied by LB, which valued at 7% in LKO mice. This reduction in LB accumulation in DKO mice is due to the reduction in brain glycogen quantities, further confirming that R6 promotes glycogen synthesis in the brain. Moreover, it indicates that R6 knockout in LD mice is not sufficient for conferring a neuropathological rescue, and may be ineffective as a therapeutic target for patients. As previously stated, PTG removal led to a near complete amelioration of LB from both malin and laforin deficient LD models. Furthermore, LD mice lacking GS had no LB formation. Evidently, the proteins that lead to the greatest inhibition of insoluble glycogen accumulation when absent in LD mice are most likely to preventing polyglucosan aggregation into LB.

LB are undisputedly the hallmark of LD. However, conflicting theories have surfaced in terms of whether LB are responsible for the clinical manifestation of LD, or whether they are a consequence of the cellular impairments associated with the loss of laforin or malin. For instance, indications of oxidative stress and mitochondrial dysfunction have been identified in LD mouse models by other LD research groups (Roma-Mateo et al., 2015; Aguado et al., 2010). These studies have proposed that since LB have been found to be polyubiquitinated, their formation may be due to compromised autophagic processes engendered by laforin or malin deficiency. As a result, oxidative stress and reactive oxygen species could be the driving force of the epilepsy, rather than the occlusion of neurons with LB. Conclusions of these theories suggest that the formation of LB is the consequence of impaired clearance in the cells.

The seminal studies investigating PTG and Gys1 knockout in LD mice were the first to establish LB formation as the causative factor of seizure susceptibility. Behavioural testing results from PTG deficient LD mice clearly demonstrated a correlation between the near-complete reduction of LB in the brain and resolution of epileptic tendencies. Myoclonic activity was reduced in these mice, and they no longer succumbed to the seizure-inducing effects of kainic acid, behaving

88

comparably to WT mice (Turnbull et al., 2011; Turnbull et al., 2014). Pederson et al. (2013) also showed how the complete prevention of LB formation in Gys1 deficient LD mice translated to low scores on the modified Racine scale during kainic acid testing. Finally, Duran et al. (2014) conducted electrophysiological analyses to measure long-term potentiation at the CA3-CA1 synapse of the hippocampus of WT, LD mice (Epm2b-/-), and LD mice heterozygous for Gys1. They found that the neurophysiological patterns of the Gys1-/+, LD mice resembled wild type patterns, which they attributed to the reduction of LB in the hippocampus of these mice. Data from these independent studies have made it difficult to deny the role that LB formation plays in the myoclonus of LD.

In this present study, it was predicted that the performance of the mice on the seizure susceptibility tests would be reliant on the amount of LB formation in the brain. Our DKO mice did have a reduction in the amount of LB in the hippocampus, but was insufficient in normalizing their myoclonic activity or responses to kainic acid to WT levels. No differences in jerks/min were seen among the LKO, LKO/R6 het, and DKO mice. Additionally, no significant differences in median stage reached on the modified Racine scale were seen between DKO, LKO/R6 het, and LKO mice during kainic acid testing. An upward trend, however, is visible across the genotypes in Figure 9B, as the median stage increases from WT to LKO. More mice should be included in this behavioral test to increase the statistical power of the experiment, and to allow for the conclusion that greater amounts of LB cause more severe responses to seizure- inducing agents.

Previous studies have shown how astro- and microgliosis accompany LB formation in LD mice. To address this neuropathological phenotype in our own project, we sought to determine whether reductions in LB formation would correlate with less gliosis. The hippocampi of our mice were immunolabeled with anti-GFAP and anti-Iba1 to visualize and quantify the astrogliotic response and microglial activation, respectively. Comparison of all three genotypes that accumulated LB (DKO, LKO/R6 het, and LKO) revealed non-significant differences in levels of astrogliosis. However, this level of anti-GFAP labeling was much greater than that found in WT and R6KO hippocampi. Likewise, the amount of Iba1-immunoreactive microglia was also comparable between genotypes that accumulated LB, yet greater than quantities found

89

in WT and R6KO mice. Since WT and R6KO mice had lower inflammatory and astrogliotic responses in the hippocampus, we can conclude that the presence of LB in neurons contribute to these gliotic processes in the brain. Perhaps, the quantity of LB in DKO mice, although lowered, surpassed a certain threshold in the hippocampus that inevitably caused a neuropathological phenotype similar to LKO mice. A more dramatic reduction in LB formation may have prevented the upregulation of both GFAP and Iba1. Results from Duran et al. (2014), Turnbull et al. (2014), and Pederson et al. (2013), have all demonstrated that LB quantities parallel the amount of GFAP expression in the brain. Similarly, Duran et al. (2014) demonstrated that LD mice with lowered LB quantities have an attenuated inflammatory response by way of less microgliosis. As resident macrophages in the brain, its possible that their function is provoked by the presence of foreign inclusions in neurons. While they function to clear and regenerate damaged neurons, their recruitment of other immune cells through cytokine and chemokine release may exacerbate the inflammatory response in the brain and further neurodegenerative processes in LD mice. Possibly, microgliosis exacerbates the inflammatory response already occurring in the brain of LD mice, furthering the neurodegenerative process. López-González et al. (2017) also reported increased microgliosis in their LD mice through the investigation of changes in mRNA levels and cytokine expression in the brain. It was found that the expression and of certain pro-inflammatory cytokines are greater in LD mice in comparison to WT mice. They also compared the strength of the inflammatory response in young and aged LD mice, and found that older mice had greater indications of inflammation in comparison to younger mice, presumably due to greater LB quantities with age. While there may be several components of LD pathogenesis provoking gliosis in the brain, LB formation in neurons is a likely contributing factor.

While analyzing the histopathological results in muscle and heart, it was determined that the LB phenotype in female LKO mice was much weaker than in male mice. This pattern was also found between males and females of DKO and LKO/R6 genotypes. In order to conclude on the LB-reducing ability of R6 deficiency, only males of each genotype were compared, as males demonstrated the classic muscle glycogen phenotype. LB quantities in the muscle were unchanged between males of LKO/R6 het, DKO and LKO mice, indicating that a deficiency in R6 does not inhibit glycogen synthesis sufficiently to lower the insoluble glycogen content.

90

Currently, PTG is the best therapeutic target for reducing polyglucosan and LB formation in muscle. It would be interesting to determine whether the same histopathological result would appear in LD mice lacking GM, which has a much greater effect on muscle glycogen synthesis than PTG.

An interesting result from this study was that a deficiency in R6 had the capacity to confer a noticeable reduction in LB quantities in cardiac muscle. Although glycogen accumulation was not quantified from the heart, this result suggests that R6 regulates glycogen synthesis more strongly in the heart than in skeletal muscle. If future LD therapies targeting GS or PTG are successful, its possible that life expectancy of patients may increase because of a slower disease progression. With an extended lifespan, extra-neurological symptoms could develop, specifically in the heart. If cardiomyopathies became a life-threatening symptom in LD patients, R6 could be a useful target for lowering LB formation and preventing heart failure in patients.

The partial rescue achieved in the heart extends the applicability of R6 knockdown to several GSD where patients are described as experiencing complications in heart function as a result of glycogen accumulation. In these cases, targeting R6 would prove useful in preventing the cardiomyopathies associated with the accumulation of polyglucosans or large quantities of normal glycogen, an example of the former being ubiquitin ligase RBCK1 deficiency, and the latter, Pompe disease. In RBCK1 deficiency, described as a polyglucosan body myopathy, patients can present with dilated cardiomyopathy at adolescence. As the patient ages, they may eventually require heart transplantation (Nilsson et al. 2013). Pompe disease patients accumulate large quantities of lysosomal glycogen in several organs, including heart, which over time will lead to cardiorespiratory failure (Kishnani et al., 2009). Thankfully, research surrounding glycogen-targeting subunits will not only contribute to LD therapy development, but have applications to GSD that, although differ in their molecular pathogenesis, may benefit from the same treatment.

91

2.4.3. Limitations

The present study had limitations pertaining to biochemical analyses, histopathology and behavioural experimentation. While some hurdles were anticipated, others were unpredictable. By discussing the following limitations surrounding methodology and experimentation, my hope is that future research groups will benefit and remain vigilant when planning projects or designing experiments for studying LD or glycogen metabolism.

2.4.3.i. Glycogen measurements

Originally, this study planned to compare the glycogenic effects of R6 using a mouse model of LD, where male and female mice would be included in analyses. Unexpectedly, our analyses of muscle glycogen revealed stark differences in glycogen content between males and females of LB-accumulating genotypes. Male mice retained the glycogen accumulation phenotype, however females had glycogen quantities that resembled WT and R6KO values. This sex difference also became apparent when comparing LB quantities. While male LKO mice had typical LB quantities in muscle and heart, females had noticeable reductions in LB formation in comparison.

The ability to compare the effects of R6 on glycogen metabolism in mice was nearly hampered by an unforeseen sex difference. In such a study where one goal is to conclude on the therapeutic capacity of a molecular target, the exclusion of female mice in the analyses limits the applicability of the results to both genders of LD patients. However, in a disease that can be prevented by inhibiting one physiological process, glycogen synthesis, results from our analyses involving solely male mice can still provide insight into the effectiveness of R6 downregulation in muscle. Thankfully, a sex difference was not found in the brain; therefore, the conclusions drawn from brain glycogen data are reliable, and would be the most important for determining whether genetic knockdown mechanisms or small molecule inhibitors against R6 would be therapeutic in clinical settings.

92

2.4.3.ii. Pathology

Pathology is a large component of LD research. Its importance stems from the need to understand the extent of the damage in the brain caused by aberrant glycogen accumulation. In this study, the amount of GFAP and Iba1 immunolabelling, as well as PASD stain (representing LB) were quantified. In some LD projects, LB quantitation is not required because the reduction of LB in treated and untreated LD mice is highly noticeable. In this study, where differences in LB quantities are hard to discern, quantification is necessary. To closely analyze small changes in LB quantities, a 3D pathology software was used to detect LB in hippocampal sections based on size and color after PASD staining. While this tool was extremely helpful, there were two limitations. The first became apparent after looking at the correlation between brain glycogen quantities from each genotype and the respective LB quantities. LKO/R6 het mice had 50% less glycogen in the brain than LKO mice, yet our quantification methods indicated that they had similar levels of LB in the brain. The comparability between LKO and LKO/R6 het mice could be due to a number of reasons. Perhaps, the reduction in LKO/R6 het glycogen corresponded to less LB in regions of the brain other than the hippocampus, such as the cerebellum, which also accumulates LB. Alternatively, it could be that the quantitation methodology should be enhanced. For instance, the applied technique quantifies LB from 1-2 hippocampal sections per mouse. Instead, several coronal sections collected at set intervals throughout the hippocampus of each mouse could be quantified to achieve a more accurate representation of LB accumulation. LB quantities could differ in a section depending on its location within the hippocampus. For this reason, future studies that utilize this quantitation method should intend to collect multiple sections, throughout the hippocampus, to improve the reliability and accuracy of their results.

2.4.3.iii Behavioural experiments

LD patients have severe myoclonus and frequent epileptic episodes that worsen over the course of the disease. There is little variability in the clinical trajectory of LD patients; the type of seizure can differ at disease onset, but most LD patients will undergo the same rate of cognitive decline due to the frequency and severity of the seizures. While LD mouse models can

93

recapitulate the histopathological hallmarks associated with LD, the resulting clinical phenotype is not as prominent. Consequently, measuring the effectiveness of therapies in mice is dependent on reductions or elimination of LB, more so than the alleviation of symptoms.

The present study employed two behavioural tests to determine whether LD mice would portray the behavioural phenotype and whether knocking out R6 could provide some phenotypic relief. The test for myoclonic activity did confirm significant increases in jerks/min in LKO, DKO, LKO/R6 het mice in comparison to WT and R6KO mice. However, due to the a) variation in the behavioural phenotype of mice within each affected genotype and b) the weak phenotypic expression in LD mice, it proved challenging to determine whether the partial reduction in LB from knocking out R6 could alleviate symptoms. The kainic acid test, a measure of seizure susceptibility, often proves to be the more reliable of the two, and is used more in LD studies to determine relief of symptoms in mice. In the experiment of the present study, significance was not achieved based on differences in the median stage reached on the modified Racine scale. Perhaps if more mice had been available for kainic acid testing, there would have been a more evident discrepancy in seizure susceptibility between LKO, DKO, and LKO/R6 het mice. This test was conducted on aged mice (12-16 months) to allow for the mice to be most sensitive to seizure-inducing agents.

2.5 Conclusions

After investigating the glycogenic involvement of R6 in the background of a LD mouse model, it is apparent that the regulation of GS and glycogen synthesis by PP1-R6 is not as strong as PP1- PTG. Since knocking out PTG in an LD mouse model had the capacity to rescue the mice, it was suspected that R6 would have a weaker effect on glycogen metabolism. This proved true for two reasons. First, R6KO mice from this study did not have lower brain glycogen levels than WT mice, which PTGKO mice from the study of Turnbull et al. (2014) did. Secondly, DKO mice did not have normalized brain glycogen values, a feat that was achieved when knocking out PTG in LD mice. These results indicate that despite their expression in the brain, PTG and R6 must have

94

related, yet hierarchal roles in brain glycogen metabolism that regulate glycogen synthesis and GS activity at different strengths.

It was also predicted that disruption of the PP1-R6 complex would lead to greater phosphorylation of GS, lowering GS activity. Our results showed that R6KO mice have a lower GS activity ratio than WT mice, indicating that GS is, on average, more phosphorylated and has a lower activation status. The activity ratios in LKO/R6 het and DKO mice were also significantly lower than LKO mice in the brain, and is most likely the reason for the lower brain glycogen in these genotypes in comparison to LKO mice. In the attempt to clarify why PTG has a stronger effect on GS activation and glycogen synthesis, we generated a mouse line where PTGKO and R6KO mice were bred within the background of healthy mice. Since Turnbull et al. (2014) showed that PTGKO mice have less brain glycogen than WT mice, it was hypothesized that the GS activity ratio would be lower in PTGKO brain extracts in comparison to R6KO mice. This expectation was incorrect, as both the PTGKO and R6KO GS activity ratios, while significantly lower than WT, were comparable. Evidently, more experiments are needed to elucidate why the PTG effects brain glycogen metabolism more strongly than R6 removal in the brain.

Although knocking out R6 in LD mice led to a partial reduction in LB accumulation in the brain, the amount was not sufficient for lowering astroglial GFAP upregulation and microgliosis in the hippocampi of mice. LKO, DKO, and LKO/R6het had comparable levels of anti-GFAP and anti- Iba1, indicating that astrogliosis and reactive microglia were still rampant throughout the brains of genotypes that accumulated LB. Presumably, greater reductions of LB formation in the hippocampus would be necessary to attenuate gliosis in DKO and LKO/R6 het mice and rescue this neurological phenotype.

The performance of LD mice in seizure susceptibility tests seem to depend on the amount of LB in the brain. In the unprovoked myoclonus test, LKO, DKO, and LKO/R6 het mice had similar jerks/minute, which were greater than that of WT and R6KO mice. Seizure analyses from kainic acid injections also revealed that knocking out R6 fully or partially did not significantly improve seizure susceptibility, even though the median stage reached on the modified Racine scale of LKO/R6 het and DKO mice was lower than LKO mice. Adding more mice for testing for each

95

genotype may reveal clear improvements in seizure susceptibility, or solidify that a near- complete depletion of LB is necessary for there to be significant differences in seizure susceptibility. The latter has been witnessed in studies by Pederson et al. (2013), Duran et al. (2014), and Turnbull et al. (2011, 2014). Although it is difficult to study behavioural phenotypes in LD mice, it is understood that inhibiting LB formation can prevent LD progression.

An additional hypothesis of this study was that R6 would play an influential role in muscle glycogen metabolism since R6 was reported to be greatly expressed in skeletal muscle, and because PTGKO mice from the study of Turnbull et al. (2014) had similar levels of muscle glycogen as WT mice. Rather, eliminating the PP1-R6 holoenzyme had no effect on skeletal muscle glycogen in male mice: R6KO muscle glycogen levels were comparable to WT levels, and LKO, DKO and LKO/R6 het mice were not only comparable but all exemplified the glycogen accumulation phenotype. As a result, LB formation was not hindered in either LKO/R6 het or DKO muscle, indicating that downregulating R6 in muscle may not confer therapeutic benefits in GSD or LD patients.

Originally, R6 removal was not hypothesized to effect LB formation in the heart. Although it is expressed in cardiac muscle, the focus of the study was on the neurological implications of knocking out R6 in LD mice. Unexpectedly, eliminating R6 expression in cardiac muscle, both partially or fully, led to a noticeable reduction in LB formation. While LKO/R6 het and DKO mice had a clear diminishment of LB in comparison to LKO mice, negating the need for close quantification, a large difference was not visible between the hearts of LKO/R6 het and DKO mice. This finding is relevant for GSD that accumulate pathogenic polyglucosans in the heart. An example of one GSD that could benefit from this finding is APBD, where polyglucosans accumulate in multiple organs including the heart, and death of a patient is often caused by cardiomyopathies and heart failure. R6 downregulation would, therefore, be applicable for treatments that aim to downregulate glycogen synthesis in the heart.

A major goal of the present study was to identify whether the R6 subunit should be considered a novel therapeutic target for the treatment of LD. Interest surrounding the glycogen-targeting subunits heightened once LD studies demonstrated that knocking out PTG in LD mice leads to the near-abolishment of the neuropathological and behavioural phenotypes. Investigation of R6

96

from this present study, in contrast, suggest that R6 is not as strong of an indirect activator of GS, and plays a weaker regulatory role in glycogen synthesis of the brain. Therefore, downregulating R6 alone as a treatment may be insufficient in preventing LD in patients. Alternatively, targeting R6 in combination with PTG or other glycogen synthesis inhibitors could be an effective form of therapy. The discrepancy in rescuing capability between PTG and R6 could potentially be explained by slight differences in their mechanisms of action, the ways by which they are regulated, or their localization in the brain. Alternatively, PTG may be the more dominant subunit in neurons, where LB are known to form, whereas both PTG and R6 could be expressed in astrocytes, where the greatest proportion of glycogen resides in the brain. More experiments are needed to contribute to and clarify the knowledge surrounding the regulation of glycogen metabolism by glycogen-targeting subunits.

To conclude, knocking out R6 in laforin-deficient mice did not prevent the neuropathological or behavioural phenotypes associated with LD. At this stage of LD research, achieving GS or PTG knockdown in the brain of patients remains as the most effective options for treating LD. Reducing GS expression levels in the brain to zero is not only difficult but dangerous for the patient, emphasizing the need for therapies that can downregulate PTG and reduce GS activity to levels that prevent disease safely. To optimize therapies against PTG, a greater understanding of its role in glycogen metabolism and LD is necessary. Future investigations will hopefully uncover the reasons behind the discrepancy in rescuing capability between PTG and R6, and benefit not only our comprehension glycogen metabolism in the brain, but also forward our grasp of LD pathogenesis.

2.6. Future Directions

While the results of this study have contributed to the research surrounding LD, glycogen- targeting subunits, and glycogen metabolism, they have also identified gaps in knowledge of each field. Posing the relevant questions and executing well-intentioned experiments will be necessary for advancing our understanding in these areas and for resolving unanswered questions regarding the glycogenic roles of PTG and R6.

97

2.6.1 Towards a deeper understanding of glycogen-targeting subunits expressed in the brain

PTG was the first glycogen-targeting subunit of PP1 investigated in the context of LD through genetic knockout. Its ability to prevent LB formation by reducing GS activity established PTG as an ideal molecular target. This finding initiated the recruitment of gene therapy technologies for attempting PTG knockdown in mouse models, and the screening of small molecule inhibitors of PTG and GS activities in cellular models. The design and optimization of these technologies can prove challenging and time consuming, allowing for the search for other potential therapeutic targets in the interim. Such was the case for investigating R6, the less effective of the two closely-related subunits expressed in the brain. Two additional glycogen-targeting subunits function in the brain, R3F and R3G, indicating that glycogen metabolism requires tight regulation of biosynthesizing and degrading enzymes. Knocking out R3G or R3F in an LD mouse model would determine their effectiveness in preventing LD progression and uncover their regulatory strengths of glycogen synthesis in the brain, albeit a repetitive study. Potentially, results from this investigation would determine whether R3F or R3G are suitable targets for LD, establish a hierarchy of subunits in respect to their influence in glycogen metabolism, and whether the current focus should remain on optimizing PTG knockdown in LD mice.

The PP1-subunit holoenzymes also dephosphorylate and inactivate GPh, preventing glycogen breakdown and promoting a net synthesis of glycogen. It would be worth investigating the extent to which PTG and R6 promote GPh dephosphorylation. Possibly, one subunit may be more inclined to recruit PP1 to GPh rather than GS. If so, this would provide insight into how knocking out PTG or R6 in LD mice led to less brain glycogen: whether the result was due solely to inhibition of GS, increased activation of GPh, or a mixture of both. LB are typically found in neurons which express only the brain isoform of GPh. Astrocytes, on the other hand, express both brain and muscle GPh isoforms. Specifying which isoforms, if any, are targeted by the subunits would further inform of their mechanisms in the brain. Some subunits, such as the GL subunit in liver, are known to recruit PP1 to GS and inhibit phosphatase activity on GPh (Doherty et al., 1995). Also, Montori-Grau et al. (2011) have shown that when R6 is overexpressed in human cultured myotubes, the activity of muscle GPh is unchanged. If a similar

98

result was achieved with the brain isoform of GPh in a neuronal cell model, it could indicate that the recruitment of PP1 by R6 is solely for GS dephosphorylation and activation. A review on the glycogen-targeting subunits by Newgard et al. (2000) states that PTG has been reported to recruit PP1 to GPh, but there is no indication of whether this includes the brain isoform. Therefore, it is possible that PTG and R6 recruit PP1 to a preferred glycogenic substrate in the brain, which could help explain their mechanisms of action.

Another question surrounding the mechanisms of PTG and R6 is how exactly their absence led to less LB accumulation in LD mice. Was the reduction in LB due to lower glycogen quantities, i.e. less insoluble glycogen formed, and consequently, less aggregation into LB? Or rather, was knocking out PTG and R6 in LD mice capable of lowering GS activity levels to the point where the glycogen branching enzyme could incorporate branch points at a greater rate, improving glycogen structure and solubility? The latter proposition is possible, as Nitschke et al. (2017) determined that LB have a greater occurrence of abnormally long chains of glycogen that compromise branching pattern, promoting insolubility. It is also known, however, that in the absence of the laforin-malin complex, mice accumulate much more insoluble glycogen than WT mice, yet have similar quantities of soluble glycogen (Sullivan et al., unpublished). Measuring the degree of polymerization, or average chain length, of brain glycogen from DKO, LKO/R6 het, and LKO mice from this present study, as well as PTG-deficient LD mice from Turnbull et al. (2011, 2014), will assist in unraveling exactly how PTG and R6 exert their LB-reducing effects. Additionally, it will determine which is most important for preventing LB formation: reducing quantity, or the preservation of glycogen quality.

A major finding in this study was the partial rescue in the heart conferred by knocking out R6, as there was a noticeable reduction in the amount of LB in the DKO and LKO/R6 het genotypes in comparison to the LKO mice. Typically, LB quantities correlate with the amount of glycogen present in the organ. Therefore, measuring glycogen from heart samples would further reveal the regulatory strength of R6 on glycogen metabolism in the heart. This partial rescue in the heart is not only the first to be reported in LD research, but has also not been explored in PTG deficient, LD mice. It would be helpful to analyze LB accumulation and glycogen quantities of the heart from LD mice lacking PTG to compare the influences of R6 and PTG in glycogen synthesis in

99

the heart, and to also determine which subunit confers a clearer pathological rescue in the heart. Before R6 can be considered a potential target for downregulation in the treatment of GSD, it would first have to be determined if the reduction in LB in the heart is enough to preserve cardiac functioning in LD mice. Cardiac functioning has not been investigated in this project, since the major focus was to determine if LKO/R6 het and DKO mice had any neurophenotypic relief in comparison to LKO mice. Perhaps in future LD studies, the impact of LB accumulation in the heart could be explored, even though LD is neurodegenerative disease. The effects of LB in the brain are undeniably detrimental to neuronal functioning and cognitive processes in both mice and humans, but with the advent of glycogen synthesis-inhibiting therapies on the brink of introduction to the clinic, a rescue in the brain might then lead to novel cardiac complications that will, again, require a therapy.

2.6.2. Identifying the cause of the sex difference in muscle glycogen and LB quantities

For this study, males and females were bred for inclusion in both biochemical and histopathological experiments. Upon analyses of our muscle glycogen results, however, a sex difference became evident. It was very noticeable that males from LKO, LKO/R6 het, and DKO genotypes had much larger quantities of muscle glycogen than their female counterparts. LD mice, regardless of sex, are known to accumulate large quantities of glycogen in muscle due to the presence of LB. This sex difference was also noticed when comparing LB quantities in cardiac muscle of male and female LKO mice. Interestingly, brain glycogen was not affected by this sex difference, as males and females had comparable levels of glycogen.

The reason for the sex difference is unknown, but has been inherent to the mouse line used to generate mice for this study. During the early stages of breeding, it is possible that a generation of mice acquired a mutation in a gene that encodes a protein involved in glycogen metabolism specific to skeletal and cardiac cells. This mutation could involve a known isozyme directly involved in glycogen synthesis or degradation, or reside in an uncharacterized gene that modulates glycogenic enzymes. In this study, all females that had muscle glycogen quantified

100

showed the “low glycogen” phenotype, and all males showed the glycogen accumulation phenotype. One possible explanation for the lack of affected males is that this mutation is lethal for males in utero, and follows an X-linked pattern of inheritance, as it has only segregated with females. Any females from this project that have not had glycogen quantified from muscle tissue should be analyzed to confirm that this mutation segregates with every female. The importance of discovering where the mutation lies is underscored for several reasons. First, it has conferred protective effects in the muscle and heart. Identifying the affected protein will reveal a therapeutic target for certain GSD that are characterized as having patients with severe clinical myopathies caused by polyglucosan accumulation. Second, locating the mutation could become an important contribution to the literature surrounding the affected protein. For instance, the mutation could alter the amino acid sequence in a vital domain of the protein, compromising protein function. This would inform on how certain mutations affect the structure-function correlations of the protein. Finally, identifying the affected gene will resolve the mystery surrounding the sex difference that has proved to be a double-edged sword in this thesis project, and allow colleagues of mine to be mindful of this mutation when generating mouse lines for future studies.

Our laboratory has begun the attempt to identify the locus of the mutation and generate a novel mouse line devoid of this mutation. DNA has been extracted from males that have shown the LB and glycogen accumulation phenotype, and from female mice that have not. An array-based genotyping approach was conducted, using informative single nucleotide polymorphisms (SNP) markers throughout the mouse genome. As we await the genotyping results, we have attained new breeders from the Jackson Laboratory which we have been integrating with LD mice, separate from those used to generate the mouse line for this project. Currently, LD mice bred from this “clean” background are aging as we wait to determine whether the expected LD phenotypes have returned.

References

Agius, L. (2015). Role of glycogen phosphorylase in liver glycogen metabolism. Molecular Aspects of Medicine, 46, 34-45.

Akman, H. O., Sheiko, T., Tay, S. K., Finegold, M. J., DiMauro, S., & Craigen, W. J. (2011). Generation of a novel mouse model that recapitulates early and adult onset glycogenosis type IV. Human Molecular Genetics, 20(22), 4430-4439.

Allaman, I., Pellerin, L., & Magistretti, P. J. (2000). Protein targeting to glycogen mRNA expression is stimulated by noradrenaline in mouse cortical astrocytes. Glia, 30(4), 382-391.

Andrade, D. M., Ackerley, C. A., Minett, T., Teive, H., Bohlega, S., Scherer, S. W., & Minassian, B. A. (2003). Skin biopsy in lafora disease genotype–phenotype correlations and diagnostic pitfalls. Neurology, 61(11), 1611-1614.

Armstrong, C. G., Browne, G. J., Cohen, P., & Cohen, P. T. (1997). PPP1R6, a novel member of the family of glycogen‐targetting subunits of protein phosphatase 1. FEBS Letters, 418(1-2), 210-214.

Aschenbach, W. G., Suzuki, Y., Breeden, K., Prats, C., Hirshman, M. F., Dufresne, S. D., . . . Kim, J. (2001a). The muscle-specific protein phosphatase PP1G/RGL (GM) is essential for activation of glycogen synthase by exercise. Journal of Biological Chemistry, 276(43), 39959-39967.

Aschenbach, W. G., Suzuki, Y., Breeden, K., Prats, C., Hirshman, M. F., Dufresne, S. D., . . . Kim, J. (2001b). The muscle-specific protein phosphatase PP1G/RGL (GM) is essential for activation of glycogen synthase by exercise. Journal of Biological Chemistry, 276(43), 39959-39967.

Bélanger, M., Allaman, I., & Magistretti, P. J. (2011). Brain energy metabolism: Focus on astrocyte-neuron metabolic cooperation. Cell Metabolism, 14(6), 724-738.

Bao, Y., Dawson Jr, T. L., & Chen, Y. (1996). Human glycogen debranching enzyme gene (AGL): Complete structural organization and characterization of the 5′ flanking region. Genomics, 38(2), 155-165.

101 102

Barbetti, F., Rocchi, M., Bossolasco, M., Cordera, R., Sbraccia, P., Finelli, P., & Consalez, G. G. (1996). The human skeletal muscle glycogenin gene: cDNA, tissue expression, and chromosomal localization. Biochemical and Biophysical Research Communications, 220(1), 72-77.

Berg, J. M., Tymoczko, J. L., & Stryer, L. (2002). Biochemistry. 5th. New York: WH Freeman, 38(894), 76.

Bittner, C. X., Loaiza, A., Ruminot, I., Larenas, V., Sotelo-Hitschfeld, T., Gutiérrez, R., . . . Barros, L. F. (2010). High resolution measurement of the glycolytic rate. .Frontiers in Neuroenergetics, 2.

Bloch, G., Chase, J. R., Meyer, D. B., Avison, M. J., Shulman, G. I., & Shulman, R. G. (1994). In vivo regulation of rat muscle glycogen resynthesis after intense exercise. American Journal of Physiology-Endocrinology and Metabolism, 266(1), E91.

Bolaños, J. P., Almeida, A., & Moncada, S. (2010). Glycolysis: A bioenergetic or a survival pathway? Trends in Biochemical Sciences, 35(3), 145-149.

Bouskila, M., Hunter, R. W., Ibrahim, A. F., Delattre, L., Peggie, M., Van Diepen, J. A., . . . Sakamoto, K. (2010). Allosteric regulation of glycogen synthase controls glycogen synthesis in muscle. Cell Metabolism, 12(5), 456-466.

Bouzier‐Sore, A., Voisin, P., Bouchaud, V., Bezancon, E., Franconi, J., & Pellerin, L. (2006). Competition between glucose and lactate as oxidative energy substrates in both neurons and astrocytes: A comparative NMR study. European Journal of Neuroscience, 24(6), 1687- 1694.

Brown, A. M., & Ransom, B. R. (2007). Astrocyte glycogen and brain energy metabolism. Glia, 55(12), 1263-1271.

Brown, A. M., Sickmann, H. M., Fosgerau, K., Lund, T. M., Schousboe, A., Waagepetersen, H. S., & Ransom, B. R. (2005). Astrocyte glycogen metabolism is required for neural activity during aglycemia or intense stimulation in mouse white matter. Journal of Neuroscience Research, 79(1‐2), 74-80.

103

Brown, A. M., Tekkӧk, S. B., & Ransom, B. R. (2003). Glycogen regulation and functional role in mouse white matter. The Journal of Physiology, 549(2), 501-512.

Burwinkel, B., Bakker, H. D., Herschkovitz, E., Moses, S. W., Shin, Y. S., & Kilimann, M. W. (1998). Mutations in the liver glycogen phosphorylase gene (PYGL) underlying glycogenosis type VI (hers disease). The American Journal of Human Genetics, 62(4), 785- 791.

Buschiazzo, A., Ugalde, J. E., Guerin, M. E., Shepard, W., Ugalde, R. A., & Alzari, P. M. (2004). Crystal structure of glycogen synthase: Homologous enzymes catalyze glycogen synthesis and degradation. The EMBO Journal, 23(16), 3196-3205.

Cao, Y., Mahrenholz, A. M., DePaoli-Roach, A. A., & Roach, P. J. (1993). Characterization of rabbit skeletal muscle glycogenin. tyrosine 194 is essential for function. Journal of Biological Chemistry, 268(20), 14687-14693.

Cenci, U., Nitschke, F., Steup, M., Minassian, B. A., Colleoni, C., & Ball, S. G. (2014). Transition from glycogen to starch metabolism in archaeplastida. Trends in Plant Science, 19(1), 18-28.

Ceulemans, H., & Bollen, M. (2004). Functional diversity of protein phosphatase-1, a cellular economizer and reset button. Physiological Reviews, 84(1), 1-39.

Chan, E. M., Omer, S., Ahmed, M., Bridges, L. R., Bennett, C., Scherer, S. W., & Minassian, B. A. (2004a). Progressive myoclonus epilepsy with polyglucosans (lafora disease) evidence for a third locus. Neurology, 63(3), 565-567.

Chan, E. M., Ackerley, C. A., Lohi, H., Ianzano, L., Cortez, M. A., Shannon, P., . . . Minassian, B. A. (2004b). Laforin preferentially binds the neurotoxic starch-like polyglucosans, which form in its absence in progressive myoclonus epilepsy. Human Molecular Genetics, 13(11), 1117-1129.

Chan, E. M., Young, E. J., Ianzano, L., Munteanu, I., Zhao, X., Christopoulos, C. C., . . . Jovic, N. J. (2003). Mutations in NHLRC1 cause progressive myoclonus epilepsy. Nature Genetics, 35(2), 125.

104

Chasiotis, D. (1988). Role of cyclic AMP and inorganic phosphate in the regulation of muscle glycogenolysis during exercise. Medicine and Science in Sports and Exercise, 20(6), 545- 550.

Chasiotis, D., Sahlin, K., & Hultman, E. (1982). Regulation of glycogenolysis in human muscle at rest and during exercise. Journal of Applied Physiology, 53(3), 708-715.

Chikwana, V. M., Khanna, M., Baskaran, S., Tagliabracci, V. S., Contreras, C. J., DePaoli- Roach, A., . . . Hurley, T. D. (2013a). Structural basis for 2′-phosphate incorporation into glycogen by glycogen synthase. Proceedings of the National Academy of Sciences, 110(52), 20976-20981.

Chikwana, V. M., Khanna, M., Baskaran, S., Tagliabracci, V. S., Contreras, C. J., DePaoli- Roach, A., . . . Hurley, T. D. (2013b). Structural basis for 2′-phosphate incorporation into glycogen by glycogen synthase. Proceedings of the National Academy of Sciences, 110(52), 20976-20981.

Choi, H. B., Gordon, G. R., Zhou, N., Tai, C., Rungta, R. L., Martinez, J., . . . Tresguerres, M. (2012). Metabolic communication between astrocytes and neurons via bicarbonate- responsive soluble adenylyl cyclase. Neuron, 75(6), 1094-1104.

Colombo, I., Pagliarani, S., Testolin, S., Cinnante, C. M., Fagiolari, G., Ciscato, P., . . . Previtali, S. C. (2016). Longitudinal follow-up and muscle MRI pattern of two siblings with polyglucosan body myopathy due to glycogenin-1 mutation. J Neurol Neurosurg Psychiatry, 87(7), 797-800.

Contreras, C. J., Segvich, D. M., Mahalingan, K., Chikwana, V. M., Kirley, T. L., Hurley, T. D., . . . Roach, P. J. (2016). Incorporation of phosphate into glycogen by glycogen synthase. Archives of Biochemistry and Biophysics, 597, 21-29.

Couarch, P., Vernia, S., Gourfinkel-An, I., Lesca, G., Gataullina, S., Fedirko, E., . . . Steschenko, D. (2011). Lafora progressive myoclonus epilepsy: NHLRC1 mutations affect glycogen metabolism. Journal of Molecular Medicine, 89(9), 915.

Dagli, A., Sentner, C. P., & Weinstein, D. A. (2016). Glycogen storage disease type III.

105

Dashty, M. (2013). A quick look at biochemistry: . Clinical Biochemistry, 46(15), 1339-1352.

Delgado-Escueta, A. V. (2007). Advances in lafora progressive myoclonus epilepsy. Current Neurology and Neuroscience Reports, 7(5), 428-433.

Delibegovic, M., Armstrong, C. G., Dobbie, L., Watt, P. W., Smith, A. J., & Cohen, P. T. (2003). Disruption of the striated muscle glycogen targeting subunit PPP1R3A of protein phosphatase 1 leads to increased weight gain, fat deposition, and development of insulin resistance. Diabetes, 52(3), 596-604.

Dent, P., Campbell, D. G., Hubbard, M. J., & Cohen, P. (1989). Multisite phosphorylation of the glycogen-binding subunit of protein phosphatase-1G by cyclic AMP-dependent protein kinase and glycogen synthase kinase-3. FEBS Letters, 248(1-2), 67-72.

Dent, P., Lavoinne, A., Nakielny, S., Caudwell, F. B., Watt, P., & Cohen, P. (1990). The molecular mechanism by which insulin stimulates glycogen synthesis in mammalian skeletal muscle. Nature, 348(6299), 302.

DePaoli-Roach, A. A., Contreras, C. J., Segvich, D. M., Heiss, C., Ishihara, M., Azadi, P., & Roach, P. J. (2015a). Glycogen phosphomonoester distribution in mouse models of the progressive myoclonic epilepsy, lafora disease. Journal of Biological Chemistry, 290(2), 841-850.

DePaoli-Roach, A. A., Contreras, C. J., Segvich, D. M., Heiss, C., Ishihara, M., Azadi, P., & Roach, P. J. (2015b). Glycogen phosphomonoester distribution in mouse models of the progressive myoclonic epilepsy, lafora disease. Journal of Biological Chemistry, 290(2), 841-850.

DePaoli-Roach, A. A., Tagliabracci, V. S., Segvich, D. M., Meyer, C. M., Irimia, J. M., & Roach, P. J. (2010). Genetic depletion of the malin E3 ubiquitin ligase in mice leads to lafora bodies and the accumulation of insoluble laforin. Journal of Biological Chemistry, 285(33), 25372-25381.

106

Doherty, M. J., Moorhead, G., Morrice, N., Cohen, P., & Cohen, P. T. (1995). Amino acid sequence and expression of the hepatic glycogen-binding (GL-subunit of protein phosphatase-1. FEBS Letters, 375(3), 294-298.

Duran, J., Gruart, A., García-Rocha, M., Delgado-García, J. M., & Guinovart, J. J. (2014). Glycogen accumulation underlies neurodegeneration and autophagy impairment in lafora disease. Human Molecular Genetics, 23(12), 3147-3156.

Duran, J., Saez, I., Gruart, A., Guinovart, J. J., & Delgado-García, J. M. (2013). Impairment in long-term memory formation and learning-dependent synaptic plasticity in mice lacking glycogen synthase in the brain. Journal of Cerebral Blood Flow & Metabolism, 33(4), 550- 556.

Egloff, M., Johnson, D. F., Moorhead, G., Cohen, P. T., Cohen, P., & Barford, D. (1997). Structural basis for the recognition of regulatory subunits by the catalytic subunit of protein phosphatase 1. The EMBO Journal, 16(8), 1876-1887.

El Tahry, R., de Tourtchaninoff, M., Vrielynck, P., & Van Rijckevorsel, K. (2015). Lafora disease: Psychiatric manifestations, cognitive decline, and visual hallucinations. Acta Neurologica Belgica, 115(3), 471-474.

Esteves, S. L., Domingues, S. C., da Cruz e Silva, Odete AB, Fardilha, M., & da Cruz e Silva, Edgar F. (2012). Protein phosphatase 1α interacting proteins in the human brain. Omics: A Journal of Integrative Biology, 16(1-2), 3-17.

Fernández-Sánchez, M. E., Criado-García, O., Heath, K. E., García-Fojeda, B., Medrano- Fernandez, I., Gomez-Garre, P., . . . Rodríguez de Córdoba, S. (2003). Laforin, the dual- phosphatase responsible for lafora disease, interacts with R5 (PTG), a regulatory subunit of protein phosphatase-1 that enhances glycogen accumulation. Human Molecular Genetics, 12(23), 3161-3171.

Froese, D. S., Michaeli, A., McCorvie, T. J., Krojer, T., Sasi, M., Melaev, E., . . . Álvarez, R. (2015). Structural basis of glycogen branching enzyme deficiency and pharmacologic rescue by rational peptide design. Human Molecular Genetics, 24(20), 5667-5676.

107

Ganesh, S., Agarwala, K. L., Ueda, K., Akagi, T., Shoda, K., Usui, T., . . . Yamakawa, K. (2000). Laforin, defective in the progressive myoclonus epilepsy of lafora type, is a dual- specificity phosphatase associated with polyribosomes. Human Molecular Genetics, 9(15), 2251-2261.

Ganesh, S., Delgado-Escueta, A. V., Sakamoto, T., Avila, M. R., Machado-Salas, J., Hoshii, Y., . . . Amano, K. (2002). Targeted disruption of the Epm2a gene causes formation of lafora inclusion bodies, neurodegeneration, ataxia, myoclonus epilepsy and impaired behavioral response in mice. Human Molecular Genetics, 11(11), 1251-1262.

Ganesh, S., Tsurutani, N., Suzuki, T., Ueda, K., Agarwala, K. L., Osada, H., . . . Yamakawa, K. (2003). The lafora disease gene product laforin interacts with HIRIP5, a phylogenetically conserved protein containing a NifU-like domain. Human Molecular Genetics, 12(18), 2359-2368.

Gayarre, J., Duran-Trío, L., Criado Garcia, O., Aguado, C., Juana-López, L., Crespo, I., . . . Rodríguez de Córdoba, S. (2014). The phosphatase activity of laforin is dispensable to rescue Epm2a−/− mice from lafora disease. Brain, 137(3), 806-818.

Gentry, M. S., Worby, C. A., & Dixon, J. E. (2005). Insights into lafora disease: Malin is an E3 ubiquitin ligase that ubiquitinates and promotes the degradation of laforin. Proceedings of the National Academy of Sciences of the United States of America, 102(24), 8501-8506.

Gibbs, M. E., Lloyd, H. G., Santa, T., & Hertz, L. (2007). Glycogen is a preferred glutamate precursor during learning in 1‐day‐old chick: Biochemical and behavioral evidence. Journal of Neuroscience Research, 85(15), 3326-3333.

Gillett, T. A., Levine, S., & Hansen, R. G. (1971). glucose pyrophosphorylase III. catalytic mechanism. Journal of Biological Chemistry, 246(8), 2551- 2554.

Gredal, H., Berendt, M., & Leifsson, P. S. (2003). Progressive myoclonus epilepsy in a beagle. Journal of Small Animal Practice, 44(11), 511-514.

108

Greenberg, C. C., Jurczak, M. J., Danos, A. M., & Brady, M. J. (2006). Glycogen branches out: New perspectives on the role of glycogen metabolism in the integration of metabolic pathways. American Journal of Physiology-Endocrinology and Metabolism, 291(1), E8.

Guerrero, R., Vernia, S., Sanz, R., Abreu-Rodríguez, I., Almaraz, C., García-Hoyos, M., . . . Nobile, C. (2011). A PTG variant contributes to a milder phenotype in lafora disease. PLoS One, 6(6), e21294.

Hedberg-Oldfors, C., Glamuzina, E., Ruygrok, P., Anderson, L. J., Elliott, P., Watkinson, O., . . . Kingston, N. (2017). Cardiomyopathy as presenting sign of glycogenin-1 deficiency—report of three cases and review of the literature. Journal of Inherited Metabolic Disease, 40(1), 139-149.

Herrero-Mendez, A., Almeida, A., Fernández, E., Maestre, C., Moncada, S., & Bolaños, J. P. (2009). The bioenergetic and antioxidant status of neurons is controlled by continuous degradation of a key glycolytic enzyme by APC/C–Cdh1. Nature Cell Biology, 11(6), 747- 752.

Huang, D., Wilson, W. A., & Roach, P. J. (1997). Glucose-6-P control of glycogen synthase phosphorylation in yeast. Journal of Biological Chemistry, 272(36), 22495-22501.

Ianzano, L., Zhao, X. C., Minassian, B. A., & Scherer, S. W. (2003). Identification of a novel protein interacting with laforin, the EPM2a progressive myoclonus epilepsy gene product. Genomics, 81(6), 579-587.

Irgens, H. U., Fjeld, K., Johansson, B. B., Ringdal, M., Immervoll, H., Leh, S., . . . Njlstad, P. R. (2015). Glycogenin-2 is dispensable for liver glycogen synthesis and glucagon-stimulated glucose release. The Journal of Clinical Endocrinology & Metabolism, 100(5), E775.

Jensen, J., & Lai, Y. (2009). Regulation of muscle glycogen synthase phosphorylation and kinetic properties by insulin, exercise, adrenaline and role in insulin resistance. Archives of Physiology and Biochemistry, 115(1), 13-21.

Jensen, J., Rustad, P. I., Kolnes, A. J., & Lai, Y. (2011). The role of skeletal muscle glycogen breakdown for regulation of insulin sensitivity by exercise. Frontiers in Physiology, 2

109

Johanns, M., Lai, Y., Hsu, M., Jacobs, R., Vertommen, D., Van Sande, J., . . . Hue, L. (2016). AMPK antagonizes hepatic glucagon-stimulated cyclic AMP signalling via phosphorylation-induced activation of cyclic nucleotide phosphodiesterase 4B. Nature Communications, 7, 10856.

Johnson, J. L., & Bagby, G. J. (1988). Gluconeogenic pathway in liver and muscle glycogen synthesis after exercise. Journal of Applied Physiology, 64(4), 1591-1599.

Kakhlon, O., Glickstein, H., Feinstein, N., Liu, Y., Baba, O., Terashima, T., . . . Lossos, A. (2013). Polyglucosan neurotoxicity caused by glycogen branching enzyme deficiency can be reversed by inhibition of glycogen synthase. Journal of Neurochemistry, 127(1), 101- 113.

Kelsall, I. R., Munro, S., Hallyburton, I., Treadway, J. L., & Cohen, P. T. (2007). The hepatic PP1 glycogen‐targeting subunit interaction with phosphorylase a can be blocked by c‐ terminal tyrosine deletion or an indole drug. FEBS Letters, 581(24), 4749-4753.

Kelsall, I. R., Voss, M., Munro, S., Cuthbertson, D. J., & Cohen, P. T. (2011). R3F, a novel membrane‐associated glycogen targeting subunit of protein phosphatase 1 regulates glycogen synthase in astrocytoma cells in response to glucose and extracellular signals. Journal of Neurochemistry, 118(4), 596-610.

Kim, J. H. (2002). Epinephrine control of glycogen metabolism in glycogen-associated protein phosphatase PP1G/RGL knockout mice. BMB Reports, 35(3), 283-290.

Kirkman, B. R., & Whelan, W. J. (1986). Glucosamine is a normal component of liver glycogen. FEBS Letters, 194(1), 6-11.

Kishnani, P. S., Corzo, D., Leslie, N. D., Gruskin, D., Van der Ploeg, A., Clancy, J. P., . . . Bauer, M. S. (2009). Early treatment with alglucosidase alfa prolongs long-term survival of infants with pompe disease. Pediatric Research, 66(3), 329.

Kollberg, G., Tulinius, M., Gilljam, T., ӧstman-Smith, I., Forsander, G., Jotorp, P., . . . Holme, E. (2007). Cardiomyopathy and exercise intolerance in muscle glycogen storage disease 0. New England Journal of Medicine, 357(15), 1507-1514.

110

Korrodi-Gregório, L., Esteves, S. L., & Fardilha, M. (2014). Protein phosphatase 1 catalytic isoforms: Specificity toward interacting proteins. Translational Research, 164(5), 366-391.

Krag, T. O., Ruiz, C. R., & Vissing, J. (2017). Glycogen synthesis in glycogenin 1 deficient patients; a role for glycogenin 2 in muscle. The Journal of Clinical Endocrinology & Metabolism,

Kurland, I. J., & Pilkis, S. J. (1989). Indirect versus direct routes of hepatic glycogen synthesis. The FASEB Journal, 3(11), 2277-2281.

López-González, I., Viana, R., Sanz, P., & Ferrer, I. (2017). Inflammation in Lafora disease: Evolution with disease progression in laforin and malin knock-out mouse models. Molecular Neurobiology, 54(5), 3119-3130.

Lerín, C., Montell, E., Nolasco, T., Clark, C., Brady, M. J., Newgard, C. B., & Gómez-Foix, A. M. (2003). Regulation and function of the muscle glycogen-targeting subunit of protein phosphatase 1 (GM) in human muscle cells depends on the COOH-terminal region and glycogen content. Diabetes, 52(9), 2221-2226.

Lohi, H., Ianzano, L., Zhao, X., Chan, E. M., Turnbull, J., Scherer, S. W., . . . Minassian, B. A. (2005). Novel glycogen synthase kinase 3 and ubiquitination pathways in progressive myoclonus epilepsy. Human Molecular Genetics, 14(18), 2727-2736.

Lomako, J., Lomako, W. M., & Whelan, W. J. (1988). A self-glucosylating protein is the primer for rabbit muscle glycogen biosynthesis. The FASEB Journal, 2(15), 3097-3103.

Lowry, O. H., & Passonneau, J. V. (1972). A collection of metabolite assays. A Flexible System of Enzymatic Analysis, , 146-218.

Lucia, A., Ruiz, J. R., Santalla, A., Nogales-Gadea, G., Rubio, J. C., García-Consuegra, I., . . . Vieitez, I. (2012). Genotypic and phenotypic features of McArdle disease: Insights from the spanish national registry. J Neurol Neurosurg Psychiatry, 301593.

Luo, X., Zhang, Y., Ruan, X., Jiang, X., Zhu, L., Wang, X., . . . Wang, Z. (2011). Fasting- induced protein phosphatase 1 regulatory subunit contributes to postprandial blood glucose homeostasis via regulation of hepatic glycogenesis. Diabetes, 60(5), 1435-1445.

111

Malfatti, E., Nilsson, J., Hedberg‐Oldfors, C., Hernandez‐Lain, A., Michel, F., Dominguez‐ Gonzalez, C., . . . Van den Bergh, P. (2014). A new muscle glycogen storage disease associated with glycogenin‐1 deficiency. Annals of Neurology, 76(6), 891-898.

Mathieu, C., De La Sierra-Gallay, Ines Li, Duval, R., Xu, X., Cocaign, A., Léger, T., . . . Haouz, A. (2016). Insights into brain glycogen metabolism THE STRUCTURE OF HUMAN BRAIN GLYCOGEN PHOSPHORYLASE. Journal of Biological Chemistry, 291(35), 18072-18083.

Mathieu, C., Dupret, J. M., & Rodrigues Lima, F. (2017). The structure of brain glycogen phosphorylase—from allosteric regulation mechanisms to clinical perspectives. The FEBS journal, 284(4), 546-554.

Matsui, T., Omuro, H., Liu, Y., Soya, M., Shima, T., McEwen, B. S., & Soya, H. (2017). Astrocytic glycogen-derived lactate fuels the brain during exhaustive exercise to maintain endurance capacity. Proceedings of the National Academy of Sciences, 114(24), 6358-6363.

Meeson, A. P., Radford, N., Shelton, J. M., Mammen, P. P., DiMaio, J. M., Hutcheson, K., ... & Garry, D. J. (2001). Adaptive mechanisms that preserve cardiac function in mice without myoglobin. Circulation research, 88(7), 713-720.

Melendez-Hevia, E., Waddell, T. G., & Shelton, E. D. (1993). Optimization of molecular design in the evolution of metabolism: The glycogen molecule. Biochemical Journal, 295(2), 477- 483.

Melendez, R., Melendez-Hevia, E., & Cascante, M. (1997). How did glycogen structure evolve to satisfy the requirement for rapid mobilization of glucose? A problem of physical constraints in structure building. Journal of Molecular Evolution, 45(4), 446-455.

Minassian, B. A. (2001). Lafora’s disease: Towards a clinical, pathologic, and molecular synthesis. Pediatric Neurology, 25(1), 21-29.

Minassian, B. A., Andrade, D. M., Ianzano, L., Young, E. J., Chan, E., Ackerley, C. A., & Scherer, S. W. (2001). Laforin is a cell membrane and endoplasmic reticulum–associated protein tyrosine phosphatase. Annals of Neurology, 49(2), 271-275.

112

Minassian, B. A., Lee, J. R., Herbrick, J., Huizenga, J., Soder, S., Mungall, A. J., . . . Carpenter, S. (1998). Mutations in a gene encoding a novel protein tyrosine phosphatase cause progressive myoclonus epilepsy. Nature Genetics, 20(2), 171.

Morgan, H. E., & Parmeggiani, A. (1964). Regulation of glycogenolysis in muscle III. control of muscle glycogen phosphorylase activity. Journal of Biological Chemistry, 239(8), 2440- 2445.

Moses, S. W., & Parvari, R. (2002). The variable presentations of glycogen storage disease type IV: A review of clinical, enzymatic and molecular studies.Current Molecular Medicine, 2(2), 177-188.

Mu, J., & Roach, P. J. (1998). Characterization of human glycogenin-2, a self-glucosylating initiator of liver glycogen metabolism. Journal of Biological Chemistry, 273(52), 34850- 34856.

Muñoz-Ballester, C., Berthier, A., Viana, R., & Sanz, P. (2016). Homeostasis of the astrocytic glutamate transporter GLT-1 is altered in mouse models of lafora disease. Biochimica Et Biophysica Acta (BBA)-Molecular Basis of Disease, 1862(6), 1074-1083.

Munro, S., Ceulemans, H., Bollen, M., Diplexcito, J., & Cohen, P. T. (2005). A novel glycogen‐ targeting subunit of protein phosphatase 1 that is regulated by insulin and shows differential tissue distribution in humans and rodents. The FEBS Journal, 272(6), 1478-1489.

Munro, S., Cuthbertson, D. J., Cunningham, J., Sales, M., & Cohen, P. T. (2002). Human skeletal muscle expresses a glycogen-targeting subunit of PP1 that is identical to the insulin- sensitive glycogen-targeting subunit GL of liver. Diabetes, 51(3), 591-598.

Nakayama, A., Yamamoto, K., & Tabata, S. (2001). Identification of the catalytic residues of bifunctional glycogen debranching enzyme. Journal of Biological Chemistry, 276(31), 28824-28828.

Nelson, S. R., Schulz, D. W., Passonneau, J. V., & Lowry, O. H. (1968). Control of glycogen levels in brain. Journal of Neurochemistry, 15(11), 1271-1279.

113

Newgard, C. B., Hwang, P. K., & Fletterick, R. J. (1989). The family of glycogen : Structure and functio. Critical Reviews in Biochemistry and Molecular Biology, 24(1), 69-99.

Nielsen, J. N., & Richter, E. A. (2003). Regulation of glycogen synthase in skeletal muscle during exercise. Acta Physiologica, 178(4), 309-319.

Niittyl, T., Comparot-Moss, S., Lue, W., Messerli, G., Trevisan, M., Seymour, M. D., . . . Chen, J. (2006). Similar protein phosphatases control starch metabolism in plants and glycogen metabolism in mammals. Journal of Biological Chemistry, 281(17), 11815-11818.

Nilsson, J., Schoser, B., Laforet, P., Kalev, O., Lindberg, C., Romero, N. B., . . . Iglseder, S. (2013). Polyglucosan body myopathy caused by defective ubiquitin ligase RBCK1. Annals of Neurology, 74(6), 914-919.

Nitschke, F., Sullivan, M. A., Wang, P., Zhao, X., Chown, E. E., Perri, A. M., . . . de Córdoba, S. R. (2017). Abnormal glycogen chain length pattern, not hyperphosphorylation, is critical in lafora disease. EMBO Molecular Medicine, , e201707608.

Nitschke, F., Wang, P., Schmieder, P., Girard, J., Awrey, D. E., Wang, T., . . . Heydenreich, M. (2013). Hyperphosphorylation of glucosyl C6 carbons and altered structure of glycogen in the neurodegenerative epilepsy lafora disease. Cell Metabolism, 17(5), 756-767.

Nuttall, F. Q., Gannon, M. C., Bai, G., & Lee, E. Y. C. (1994). Primary structure of human liver glycogen synthase deduced by cDNA cloning//doi.org/10.1006/abbi.1994.1260

O’Donnell, J., Zeppenfeld, D., McConnell, E., Pena, S., & Nedergaard, M. (2012). Norepinephrine: A neuromodulator that boosts the function of multiple cell types to optimize CNS performance. Neurochemical Research, 37(11), 2496-2512.

Orhan Akman, H., Emmanuele, V., Kurt, Y. G., Kurt, B., Sheiko, T., DiMauro, S., & Craigen, W. J. (2015). A novel mouse model that recapitulates adult-onset glycogenosis type 4. Human Molecular Genetics, 24(23), 6801-6810.

Ortolano, S., Vieitez, I., Agis-Balboa, R. C., & Spuch, C. (2014). Loss of GABAergic cortical neurons underlies the neuropathology of lafora disease. Molecular Brain, 7(1), 7.

114

Pautsch, A., Stadler, N., Wissdorf, O., Langkopf, E., Moreth, W., & Streicher, R. (2008). Molecular recognition of the protein phosphatase 1 glycogen targeting subunit by glycogen phosphorylase. Journal of Biological Chemistry, 283(14), 8913-8918.

Pederson, B. A., Chen, H., Schroeder, J. M., Shou, W., DePaoli-Roach, A. A., & Roach, P. J. (2004). Abnormal cardiac development in the absence of heart glycogen. Molecular and Cellular Biology, 24(16), 7179-7187.

Pederson, B. A., Cope, C. R., Schroeder, J. M., Smith, M. W., Irimia, J. M., Thurberg, B. L., . . . Roach, P. J. (2005). Exercise capacity of mice genetically lacking muscle glycogen synthase in mice, muscle glycogen is not essential for exercise. Journal of Biological Chemistry, 280(17), 17260-17265.

Pederson, B. A., Csitkovits, A. G., Simon, R., Schroeder, J. M., Wang, W., Skurat, A. V., & Roach, P. J. (2003). Overexpression of glycogen synthase in mouse muscle results in less branched glycogen. Biochemical and Biophysical Research Communications, 305(4), 826- 830.

Pederson, B. A., Turnbull, J., Epp, J. R., Weaver, S. A., Zhao, X., Pencea, N., . . . Minassian, B. A. (2013). Inhibiting glycogen synthesis prevents lafora disease in a mouse model. Annals of Neurology, 74(2), 297-300.

Pfeiffer‐Guglielmi, B., Fleckenstein, B., Jung, G., & Hamprecht, B. (2003). Immunocytochemical localization of glycogen phosphorylase isozymes in rat nervous tissues by using isozyme‐specific antibodies. Journal of Neurochemistry, 85(1), 73-81.

Picton, C., Klee, C. B., & Cohen, P. (1981). The regulation of muscle phosphorylase kinase by calcium ions, calmodulin and troponin-C. Cell Calcium, 2(4), 281-294.

PITCHER, J., SMYTHE, C., CAMPBELL, D. G., & COHEN, P. (1987). Identification of the 38‐kDa subunit of rabbit skeletal muscle glycogen synthase as glycogenin. The FEBS Journal, 169(3), 497-502.

Racine, R. J. (1972). Modification of seizure activity by electrical stimulation: II. motor seizure. Electroencephalography and Clinical Neurophysiology, 32(3), 281-294.

115

Rath, V. L., Ammirati, M., LeMotte, P. K., Fennell, K. F., Mansour, M. N., Danley, D. E., . . . Pandit, J. (2000). Activation of human liver glycogen phosphorylase by alteration of the secondary structure and packing of the catalytic core. Molecular Cell, 6(1), 139-148.

Rich, L., & Brown, A. M. (2016). Glycogen: Multiple roles in the CNS. The Neuroscientist, , 1073858416672622.

Roach, P. J. (1991). Multisite and hierarchal protein phosphorylation. Journal of Biological Chemistry, 266(22), 14139-14142.

Roach, P. J. (2015). Glycogen phosphorylation and lafora disease. Molecular Aspects of Medicine, 46, 78-84.

Roach, P. J., Depaoli-Roach, A. A., Hurley, T. D., & Tagliabracci, V. S. (2012). Glycogen and its metabolism: Some new developments and old themes.Biochemical Journal, 441(3), 763- 787.

Roach, P. J., & Larner, J. (1977). Covalent phosphorylation in the regulation of glycogen synthase activity. Molecular and Cellular Biochemistry, 15(3), 179-200.

Rodriguez, I. R., & Whelan, W. J. (1985). A novel glycosyl-amino acid linkage: Rabbit-muscle glycogen is covalently linked to a protein via tyrosine. Biochemical and Biophysical Research Communications, 132(2), 829-836.

Romá-Mateo, C., Aguado, C., García-Giménez, J. L., Ibáñez-Cabellos, J. S., Seco-Cervera, M., Pallardó, F. V., . . . Sanz, P. (2015). Increased oxidative stress and impaired antioxidant response in lafora disease. Molecular Neurobiology, 51(3), 932-946.

Rubio-Villena, C., Garcia-Gimeno, M. A., & Sanz, P. (2013). Glycogenic activity of R6, a protein phosphatase 1 regulatory subunit, is modulated by the laforin–malin complex. The International Journal of Biochemistry & Cell Biology, 45(7), 1479-1488.

Rubio-Villena, C., Sanz, P., & Garcia-Gimeno, M. A. (2015). Structure-function analysis of PPP1R3D, a protein phosphatase 1 targeting subunit, reveals a binding motif for 14-3-3 proteins which regulates its glycogenic properties. PloS One, 10(6), e0131476.

116

Ruchti, E., Roach, P. J., DePaoli-Roach, A. A., Magistretti, P. J., & Allaman, I. (2016). Protein targeting to glycogen is a master regulator of glycogen synthesis in astrocytes. IBRO Reports, 1, 46-53.

Saez, I., Duran, J., Sinadinos, C., Beltran, A., Yanes, O., Tevy, M. F., . . . Guinovart, J. J. (2014). Neurons have an active glycogen metabolism that contributes to tolerance to hypoxia. Journal of Cerebral Blood Flow & Metabolism, 34(6), 945-955.

Singh, S., Sethi, I., Francheschetti, S., Riggio, C., Avanzini, G., Yamakawa, K., . . . Ganesh, S. (2006). Novel NHLRC1 mutations and genotype–phenotype correlations in patients with lafora’s progressive myoclonic epilepsy. Journal of Medical Genetics, 43(9), e48.

Singh, S., & Ganesh, S. (2009). Lafora progressive myoclonus epilepsy: A meta‐analysis of reported mutations in the first decade following the discovery of the EPM2A and NHLRC1 genes. Human Mutation, 30(5), 715-723.

Skurat, A. V., Dietrich, A. D., & Roach, P. J. (2006). Interaction between glycogenin and glycogen synthase. Archives of Biochemistry and Biophysics, 456(1), 93-97.

Skurat, A. V., Wang, Y., & Roach, P. J. (1994). Rabbit skeletal muscle glycogen synthase expressed in COS cells. identification of regulatory phosphorylation sites. Journal of Biological Chemistry, 269(41), 25534-25542.

Solaz-Fuster, M. C., Gimeno-Alcaniz, J. V., Ros, S., Fernandez-Sanchez, M. E., Garcia-Fojeda, B., Garcia, O. C., . . . Sanchez-Piris, M. (2007). Regulation of glycogen synthesis by the laforin–malin complex is modulated by the AMP-activated protein kinase pathway. Human Molecular Genetics, 17(5), 667-678.

Striano, P., Zara, F., Turnbull, J., Girard, J., Ackerley, C. A., Cervasio, M., . . . Minassian, B. A. (2008). Typical progression of myoclonic epilepsy of the lafora type: A case report. Nature Clinical Practice Neurology, 4(2), 106-111.

Subbarao, K. V., & Hertz, L. (1990). Effect of adrenergic agonists on glycogenolysis in primary cultures of astrocytes. Brain Research, 536(1-2), 220-226.

117

Sullivan, M. A., Nitschke, S., Steup, M., Minassian, B. A., & Nitschke, F. (2017). Pathogenesis of lafora disease: Transition of soluble glycogen to insoluble polyglucosan. International Journal of Molecular Sciences, 18(8), 1743.

Suzuki, Y., Lanner, C., Kim, J., Vilardo, P. G., Zhang, H., Yang, J., . . . Bock, C. B. (2001). Insulin control of glycogen metabolism in knockout mice lacking the muscle-specific protein phosphatase PP1G/RGL. Molecular and Cellular Biology, 21(8), 2683-2694.

Swain, L., Key, G., Tauro, A., Ahonen, S., Wang, P., Ackerley, C., . . . Rusbridge, C. (2017). Lafora disease in miniature wirehaired dachshunds. PloS One, 12(8), e0182024.

Swanson, R. A., & Choi, D. W. (1993). Glial glycogen stores affect neuronal survival during glucose deprivation in vitro. Journal of Cerebral Blood Flow & Metabolism, 13(1), 162- 169.

Tagliabracci, V. S., Girard, J. M., Segvich, D., Meyer, C., Turnbull, J., Zhao, X., . . . Roach, P. J. (2008). Abnormal metabolism of glycogen phosphate as a cause for lafora disease. Journal of Biological Chemistry, 283(49), 33816-33825.

Tagliabracci, V. S., Heiss, C., Karthik, C., Contreras, C. J., Glushka, J., Ishihara, M., . . . Roach, P. J. (2011). Phosphate incorporation during glycogen synthesis and lafora disease. Cell Metabolism, 13(3), 274-282.

Testoni, G., Duran, J., Garca-Rocha, M., Vilaplana, F., Serrano, A. L., Sebastin, D., . . . Vilaseca, M. (2017). Lack of glycogenin causes glycogen accumulation and muscle function impairment. Cell Metabolism, 26(1), 266. e4.

Thorens, B., & Mueckler, M. (2010a). Glucose transporters in the 21st century. American Journal of Physiology-Endocrinology and Metabolism, 298(2), E145.

Thorens, B., & Mueckler, M. (2010b). Glucose transporters in the 21st century. American Journal of Physiology-Endocrinology and Metabolism, 298(2), E145.

Turnbull, J., DePaoli-Roach, A. A., Zhao, X., Cortez, M. A., Pencea, N., Tiberia, E., . . . Ackerley, C. A. (2011). PTG depletion removes lafora bodies and rescues the fatal epilepsy of lafora disease. PLoS Genetics, 7(4), e1002037.

118

Turnbull, J., Epp, J. R., Goldsmith, D., Zhao, X., Pencea, N., Wang, P., . . . Minassian, B. A. (2014). PTG protein depletion rescues malin‐deficient lafora disease in mouse. Annals of Neurology, 75(3), 442-446.

Turnbull, J., Girard, J., Lohi, H., Chan, E. M., Wang, P., Tiberia, E., . . . Chakrabarty, A. (2012). Early-onset lafora body disease. Brain, 135(9), 2684-2698.

Turnbull, J., Tiberia, E., Striano, P., Genton, P., Carpenter, S., Ackerley, C. A., & Minassian, B. A. (2016). Lafora disease. Epileptic Disorders, 18(s2)

Turnbull, J., Wang, P., Girard, J., Ruggieri, A., Wang, T. J., Draginov, A. G., . . . Ackerley, C. A. (2010). Glycogen hyperphosphorylation underlies lafora body formation. Annals of Neurology, 68(6), 925-933.

Turner, D. A., & Adamson, D. C. (2011). Neuronal-astrocyte metabolic interactions: Understanding the transition into abnormal astrocytoma metabolism.Journal of Neuropathology & Experimental Neurology, 70(3), 167-176.

Valenzuela, D. M., Murphy, A. J., Frendewey, D., Gale, N. W., Economides, A. N., Auerbach, W., . . . Yasenchak, J. (2003). High-throughput engineering of the mouse genome coupled with high-resolution expression analysis. Nature Biotechnology, 21(6), 652.

Valles‐Ortega, J., Duran, J., Garcia‐Rocha, M., Bosch, C., Saez, I., Pujadas, L., . . . Delgado‐ García, J. M. (2011). Neurodegeneration and functional impairments associated with glycogen synthase accumulation in a mouse model of lafora disease. EMBO Molecular Medicine, 3(11), 667-681. van Maanen, M., Fournier, P. A., Palmer, T. N., & Abraham, L. J. (1999). Characterization of the human glycogenin-1 gene: Identification of a muscle-specific regulatory domain. Gene, 234(2), 217-226.

Vernia, S., Solaz-Fuster, M. C., Gimeno-Alcañiz, J. V., Rubio, T., García-Haro, L., Foretz, M., . . . Sanz, P. (2009). AMP-activated protein kinase phosphorylates R5/PTG, the glycogen targeting subunit of the R5/PTG-protein phosphatase 1 holoenzyme, and accelerates its

119

down-regulation by the laforin-malin complex. Journal of Biological Chemistry, 284(13), 8247-8255.

Ververken, D., Veldhoven, P., Proost, C., Carton, H., & Wulf, H. (1982). On the role of calcium ions in the regulation of glycogenolysis in mouse brain cortical slices. Journal of Neurochemistry, 38(5), 1286-1295.

Vilchez, D., Ros, S., Cifuentes, D., Pujadas, L., Vallès, J., Garcia-Fojeda, B., . . . Domínguez, J. (2007). Mechanism suppressing glycogen synthesis in neurons and its demise in progressive myoclonus epilepsy. Nature Neuroscience, 10(11), 1407-1413.

Waitt, A. E., Reed, L., Ransom, B. R., & Brown, A. M. (2017). Emerging roles for glycogen in the CNS. Frontiers in Molecular Neuroscience, 10

Walker, K. S., Watt, P. W., & Cohen, P. (2000). Phosphorylation of the skeletal muscle glycogen‐targeting subunit of protein phosphatase 1 in response to adrenaline in vivo. FEBS Letters, 466(1), 121-124.

Wang, J., Stuckey, J. A., Wishart, M. J., & Dixon, J. E. (2002). A unique carbohydrate binding domain targets the lafora disease phosphatase to glycogen. Journal of Biological Chemistry, 277(4), 2377-2380.

Worby, C. A., Gentry, M. S., & Dixon, J. E. (2008). Malin decreases glycogen accumulation by promoting the degradation of protein targeting to glycogen (PTG) Journal of Biological Chemistry, 283(7), 4069-4076.

Zeqiraj, E., Tang, X., Hunter, R. W., García-Rocha, M., Judd, A., Deak, M., . . . Tyers, M. (2014). Structural basis for the recruitment of glycogen synthase by glycogenin. Proceedings of the National Academy of Sciences, 111(28), E2840.

Zhai, L., Mu, J., Zong, H., DePaoli-Roach, A. A., & Roach, P. J. (2000). Structure and chromosomal localization of the human glycogenin-2 gene GYG2. Gene, 242(1), 229-235.

Zhai, L., Feng, L., Xia, L., Yin, H., & Xiang, S. (2016). Crystal structure of glycogen debranching enzyme and insights into its catalysis and disease-causing mutations. Nature Communications, 7

120

Zhang, Y., Gu, J., Wang, L., Zhao, Z., Pan, Y., & Chen, Y. (2017). Ablation of PPP1R3G reduces glycogen deposition and mitigates high-fat diet induced obesity.Molecular and Cellular Endocrinology, 439, 133-140.

Zibrova, D., Grempler, R., Streicher, R., & Kauschke, S. G. (2008). Inhibition of the interaction between protein phosphatase 1 glycogen-targeting subunit and glycogen phosphorylase increases glycogen synthesis in primary rat hepatocytes. Biochemical Journal, 412(2), 359- 366.

121

Contributions

The mouse line generated for this project was initiated by Dr. Julie Turnbull and bred by Xiaochu Zhao prior to my arrival to the lab. I would like to acknowledge both for their contributions to this project, as it likely would not have existed without them.

The author performed all analyses and experiments for this project with the exception of the following:

Ami Perri and Dr. Peter Wang conducted brain glycogen measurements.

The Pathology Core at the Centre for Phenogenomics conducted Periodic Acid-Schiff diastase staining of tissue, and immunohistochemistry.

Xiaochu Zhao assisted with tissue harvesting and kainic acid injections for mouse behavioural studies.

Dr. Peter Wang conducted glycogen synthase activity assays.