The Essential Role of O-Glcnacylation in Primary Sensory Neurons

The Essential Role of O-GlcNAcylation in Primary Sensory Neurons

The Harvard community has made this article openly available. Please share how this access benefits you. Your story matters

Citable link http://nrs.harvard.edu/urn-3:HUL.InstRepos:40046407

Terms of Use This article was downloaded from Harvard University’s DASH repository, and is made available under the terms and conditions applicable to Other Posted Material, as set forth at http:// nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of- use#LAA

The essential role of O-GlcNAcylation in primary sensory neurons

A dissertation presented

Cathy Su

The Division of Medical Sciences

in partial fulfillment of the requirements

for the degree of

Doctor of Philosophy

in the subject of

Biological and Biomedical Sciences

Harvard University

Cambridge, Massachusetts

February 2017

Dissertation Advisor: Dr. Thomas L. Schwarz Cathy Su

The essential role of O-GlcNAcylation in primary sensory neurons

Abstract

O-GlcNAcylation is the post-translational addition of β-N-acetylglucosamine to nuclear

and cytoplasmic proteins. This addition is mediated by a single enzyme, O-GlcNAc transferase

(OGT), which regulates a wide range of cellular processes through its thousands of protein substrates. The activity of OGT is affected by nutrient availability, and thus its role has been broadly studied in metabolic tissues. OGT is enriched in the nervous system, yet little is known about its importance in basic neuronal processes in vivo.

In this work, I utilized in vivo and neuronal culture systems to determine the effects of altered O-GlcNAc dynamics in primary sensory neurons. Sensory neurons lie outside of the blood brain barrier and therefore may have a particular need for mechanisms of metabolic sensing. I show that sensory neuron-specific knockout of OGT in mice results in behavioral hyposensitivity to thermal and mechanical stimuli accompanied by decreased epidermal innervation and cell body loss in the dorsal root ganglia. These effects are observed early in postnatal development and progress as the animals age. The deficits in neuronal health are not

solely due to disruption of developmental processes, because inducing OGT knockout in the

sensory neurons of adult mice results in a similar decrease in nerve fiber endings and cell bodies.

Significant nerve ending loss occurs prior to a decrease in cell bodies, indicative of axonal

dieback that progresses to neuronal death. Cultured sensory neurons lacking OGT also exhibit

decreased axonal outgrowth. These findings demonstrate that OGT is important in regulating

axonal maintenance in the periphery and the overall health and survival of sensory neurons.

iii

While previous studies have focused on the essential role of OGT in mediating survival of mitotic cells, this work is the first to find that loss of OGT results in cell death in neurons.

Moreover, it suggests that aberrant O-GlcNAc signaling can contribute to the development of neurodegeneration and neuropathy. Primary sensory neurons in particular are subject to degeneration in diabetes. My findings provide a foundation for understanding the role of neuronal OGT under normal physiological conditions, which will be important for understanding disease states such as diabetic neuropathy.

Table of Contents

Title page ...... i

Abstract ...... iii

Table of contents ...... v

List of figures and tables ...... viii

Acknowledgements ...... x

Chapter 1: Introduction

1.1 The O-GlcNAc modification ...... 2

Enzymes mediating O-GlcNAc cycling...... 2

Cellular functions of O-GlcNAcylation ...... 6

1.2 O-GlcNAc as a metabolic sensor ...... 8

Roles of OGT in peripheral metabolic processes ...... 10

1.3 O-GlcNAcylation in the nervous system ...... 11

O-GlcNAcylation in neurodegenerative disease ...... 14

1.4 Primary sensory neurons ...... 15

Sensory neuropathy ...... 19

1.5 Overview of dissertation ...... 20

Chapter 2: O-GlcNAc transferase is essential for sensory neuron survival and maintenance

2.1 Introduction ...... 24

2.2 Results

Nav1.8-Ogt knockout mice are overtly healthy but exhibit altered weight and glucose

tolerance ...... 26

Nav1.8-Ogt knockout mice show behavioral deficits in thermal and mechanical

sensitivity ...... 32

Nav1.8-Ogt knockout mice lose epidermal innervation and DRG cell bodies ...... 36

OGT knockout neurons exhibit axonal outgrowth deficits in culture...... 40

Loss of epidermal innervation and cell bodies occurs independently of developmental

processes ...... 43

2.3 Discussion ...... 49

2.4 Materials and methods

Mice ...... 54

OGT knockout ...... 55

Behavioral testing ...... 56

Glucose tolerance test ...... 58

Epidermal innervation ...... 59

DRG cell body count ...... 59

Axonal outgrowth assay ...... 60

Chapter 3: Additional observations: Effects of altered O-GlcNAcylation in sensory neurons

3.1 Introduction ...... 63

3.2 O-GlcNAc regulation of mitochondrial motility in DRG neurons ...... 64

Mitochondrial motility in DRG neurons under a high glucose condition ...... 71

3.3 Additional observations of OGT knockout neurons in culture ...... 73

Axonal outgrowth under varying glucose conditions ...... 73

Mitochondrial distribution ...... 76

Survival in culture ...... 78

3.4 Investigating candidate protein expression in Nav1.8-Ogt knockout DRGs ...... 79

3.5 Materials and methods

Lentivirus preparation ...... 83

Axonal transport assays ...... 84

Quantification of O-GlcNAc levels ...... 86

Axonal outgrowth and survival ...... 86

Immunofluorescence ...... 87

Chapter 4: Extended discussion

4.1 OGT is essential for neuronal survival ...... 89

4.2 Behavioral and global consequences of loss of sensory neurons ...... 92

4.3 Deciphering OGT-regulated pathways ...... 94

4.4 Implications in diabetic neuropathy ...... 95

References ...... 99

vii

List of Figures and Tables

Figure 1.1 O-GlcNAcylation is dynamically regulated to exert specific effects on target

proteins ...... 3

Table 1.1 Examples of O-GlcNAc regulation of cellular processes ...... 7

Figure 1.2 The hexosamine biosynthetic pathway...... 9

Figure 1.3 Sensory neuron structure ...... 17

Table 1.2 Classification of primary somatosensory neuron fiber types ...... 18

Figure 2.1 Nav1.8-Cre-mediated Ogt knockout occurs in the majority of small-diameter DRG

neurons ...... 28

Figure 2.2 Nav1.8-Ogt knockout mice exhibit normal global and motor phenotypes...... 29

Figure 2.3 Nav1.8-Ogt knockout mice exhibit altered weight ...... 30

Figure 2.4 Nav1.8-Ogt knockout mice exhibit altered glucose tolerance ...... 31

Figure 2.5 Nav1.8-Ogt knockout mice show clear behavioral deficits to thermal stimulation 33

Figure 2.6 Nav1.8-Ogt knockout mice show mild behavioral deficits to mechanical

stimulation ...... 35

Figure 2.7 Decreased epidermal innervation in Nav1.8-Ogt knockout mice...... 37

Figure 2.8 Loss of peripherin-expressing neuron cell bodies in Nav1.8-Ogt knockout DRGs 39

Figure 2.9 Decreased axonal outgrowth in cultured OGT knockout neurons ...... 41

Figure 2.10 Brn3a-CreERT2-mediated Ogt knockout occurs in the majority of small-diameter

DRG neurons ...... 45

Figure 2.11 OGT knockout in adult animals causes loss of epidermal innervation and DRG cell

bodies...... 46

viii

Figure 2.12 Brn3a-Ogt knockout mice show normal thermal sensitivity at 4 weeks after Cre

induction ...... 48

Figure 3.1 PUGNAc treatment results in increased O-GlcNAc levels in DRG neurons ...... 66

Figure 3.2 Increased O-GlcNAcylation decreases axonal mitochondrial motility in DRG

neurons ...... 67

Figure 3.3 Increased O-GlcNAcylation does not affect axonal transport of lysosomes in DRG

neurons ...... 69

Figure 3.4 A high extracellular glucose condition does not affect axonal mitochondrial

motility in DRG neurons ...... 72

Figure 3.5 Decreased axonal outgrowth under varying glucose conditions in cultured OGT

knockout neurons ...... 74

Figure 3.6 Loss of OGT has no apparent effects on mitochondrial distribution in cultured

DRG neurons ...... 77

Figure 3.7 Loss of OGT has no apparent effects on immunofluorescence patterns of candidate

proteins in DRGs ...... 82

Acknowledgements

My graduate school experience has been one of tremendous scientific and personal

growth, which would not have been possible without the mentorship of my advisor, Tom

Schwarz. Tom’s dedication to his trainees and genuine commitment towards scientific discovery

and learning were what first drew me to his lab, and I thank him for his continuous support,

feedback, and guidance in the years since. While my project has had its ups and downs, Tom has

been a constant source of positivity and reason, and for that I am very grateful.

I also acknowledge the past and present members of the Schwarz lab, who have made the

lab such a friendly and supportive environment in which to work: Amos Gutnick, Angelika

Harbauer, Asli Oztan Matos, Erica Gornstein, Evgeny Shlevkov, Ga Young Lee, Ghazaleh

Ashrafi, Gulcin Pekkurnaz, Guoli Zhao, Himanish Basu, Jarom Chung, Jill Falk, Kaan Apaydin,

Kayla Davis, Lala Mkhitaryan, Matt Boersma, Michelle Cronin, Rita Teodoro, Romain Cartoni,

Steve Raiker, and Tal Kramer. I have received advice from everyone at some point, and in

particular I thank Gulcin for teaching me about O-GlcNAcylation when I first joined the lab, a modification I had never even heard of till then. I am also especially thankful for Erica, Jarom, and Ga Young and the close friendships we’ve formed – grad school would not have been nearly as fun without them.

When I embarked on my project, I knew I would be seeking the expertise of many others beyond our lab. I am grateful for the advice, lessons, and reagents from many members of

Clifford Woolf’s lab: Ajay Yekkirala, Alban Latremoliere, Bhagat Singh, Catherine Ward, Eric

Huebner, Mike Costigan, Segun Babaniyi, and Takao Omura. I also acknowledge the core facility managers in the department – Anthony Hill, Gregorz Gorski, and Nick Andrews – for

their training in techniques.

I thank the members of my dissertation advisory committee – Marcia Haigis, Clifford

Woolf, and Gary Yellen – for their discussion, guidance, and kind encouragement at each and

every meeting.

I am grateful for all of the friends I have made in Boston, who have made living in a new

city feel like home over the past 5 years. I am especially thankful for Linda and Sherry, who

have become my closest friends and with whom I’ve shared countless adventures and meals.

I would like to thank Jigar, my best friend and partner, for always being “by my side,”

even when distance gets in the way. His understanding and companionship have been truly

invaluable through these years.

Lastly, I would like to thank my parents for their endless support. They are my greatest role models, and without their sacrifice, love, and encouragement, this would not have been

possible.

Chapter 1

Introduction

1.1 The O-GlcNAc modification

In the early 1980s, a novel form of protein glycosylation was discovered that defied the

characteristics of “classical” glycosylation in a cell. Termed O-GlcNAcylation, the modification is the addition of O-linked β-N-acetylglucosamine (O-GlcNAc) to serine and threonine residues

on proteins (Torres and Hart, 1984). Unlike the complex glycans that decorate extracellular

proteins and luminal proteins of the secretory system, O-GlcNAc exists as a monosaccharide on

proteins throughout the cytoplasmic and nuclear compartments of the cell (Holt and Hart, 1986).

Moreover, O-GlcNAcylation is highly dynamic and often substoichiometric at any site. In these

ways, the modification has been likened to phosphorylation (Hart et al., 2011).

Over the past 30 years, O-GlcNAcylation has been established as an important and

ubiquitous regulator of cellular processes. The modification is highly conserved and has been

documented in unicellular organisms, plants, and all metazoans (Hart et al., 2011). In mammals,

O-GlcNAc is present in essentially all cell types and within all intracellular compartments

(Kreppel et al., 1997; Lubas et al., 1997). Over 4000 proteins have been identified to be O-

GlcNAcylated, and these proteins cover a wide range of functional groups and pathways (Ma

and Hart, 2014).

Enzymes mediating O-GlcNAc cycling

O-GlcNAc cycling in mammals is mediated by only two enzymes: O-GlcNAc transferase

(OGT), which adds the modification to proteins, and O-GlcNAcase (OGA), which removes the modification. This is in contrast to other post-translational modifications, notably phosphorylation, which can be mediated by up to hundreds of different enzymes. OGT and OGA are therefore regulated on several levels to ensure proper O-GlcNAcylation of their many protein substrates under various conditions (Figure 1.1) (Nagel and Ball, 2014; Bond and Hanover,

Figure 1.1. O-GlcNAcylation is dynamically regulated to exert specific effects on target proteins

O-GlcNAcylation is mediated by two enzymes: O-GlcNAc transferase (OGT), which adds the modification, and O-GlcNAcase (OGA), which removes the modification. With over 4000 substrates,

OGT is regulated on many levels to ensure proper modification of different protein substrates. Alternative splicing of the gene, donor substrate availability, interactions with binding partners, and post-translational modifications (PTMs) can all have an effect on the localization and activity of OGT under different conditions. In turn, the activity of OGT on its target proteins can affect the localization, stability, activity, substrate specificity, and PTM profile of the proteins, and thus regulate pathways and processes within the cell.

2015).

OGT consists of an N-terminal tetratricopeptide repeat (TPR) domain and a catalytic C- terminal glycosyltransferase domain. The Ogt gene is located on the X chromosome and encodes for three major splice variants that vary in the number of TPR motifs and cellular localization (Shafi et al., 2000; Hanover et al., 2003; Lazarus et al., 2006). Nucleocytoplasmic

OGT (ncOGT) is the full-length variant, with 11-13 TPRs in mammals. Mitochondrial OGT

(mOGT) contains a mitochondrial targeting sequence and 9 TPRs, and localizes to the inner membrane of mitochondria. Soluble OGT (sOGT) is the shortest isoform, with 2.5 TPRs, and also localizes to the nucleus and cytoplasm. In vivo, OGT can oligomerize into a dimer or trimer.

The enzyme was originally purified from rat liver as a heterotrimer, consisting of two full-length

110-kDa variants and one 78-kDa unit, which may correspond to sOGT (Haltiwanger et al.,

1992). Since then, it has also been shown exist as ncOGT homodimers and homotrimers

(Kreppel and Hart, 1999; Jinek et al., 2004).

In vivo studies suggest that O-GlcNAc sites tend to be in disordered regions of proteins and contain a proline-valine-serine (PVS) motif, but there is no absolute consensus sequence for the modification (Lagerlof and Hart, 2014; Liu et al., 2014; Pathak et al., 2015). Therefore, while the primary structure of a given protein plays a role in whether or not it will be O-GlcNAcylated, protein interactions with the TPR motifs and other regions of OGT are critical in regulating the target specificity of the enzyme at any given time and place in a cell. The TPR region consists of anti-parallel α-helices that generate a superhelical structure, lending itself to protein-protein interactions (Jinek et al., 2004). In many cases, OGT is directed to its protein substrates through these interactions with binding partners, which can affect the localization and catalytic activity of the enzyme. For example, OGT forms a complex with the transcriptional co-activator PGC-1α,

4 which targets the enzyme to FoxO transcription factors and results in increased O-

GlcNAcylation of FoxOs and enhanced transcriptional activity (Housley et al., 2009). OGT also binds directly with several of its substrates, including Milton (TRAK1/2 or OIP106/98) (Iyer et al., 2003), host cell factor C1 (HCF-1) (Lazarus et al., 2013), myosin phosphatase target subunit

1 (MYPT1) (Cheung et al., 2008), and OGA (Whisenhunt et al., 2006). Moreover, OGT itself is post-translationally modified, and this too could alter its activity (Nagel and Ball, 2014). For instance, phosphorylation of OGT at Thr-444 increases its nuclear localization and affects its substrate selectivity in C2C12 myotubes and HEK293T cells (Bullen et al., 2014).

The donor substrate for OGT is uridine diphosphate N-acetylglucosamine (UDP-

GlcNAc), a nucleotide sugar produced by the hexosamine biosynthetic pathway (HBP).

Structural studies of OGT indicate that O-GlcNAc addition occurs through an ordered sequential bi-bi mechanism in which OGT first binds UDP-GlcNAc followed by the protein substrate

(Lazarus et al., 2011). The activity of OGT is in part regulated by the concentration of UDP-

GlcNAc. In vitro, OGT exhibits a wide range of affinities for peptide substrates depending on the levels of UDP-GlcNAc (Kreppel and Hart, 1999). Moreover, the kinetics of UDP-GlcNAc as a substrate are also different depending on the protein substrate (Shen et al., 2012). OGT activity is therefore highly dynamic and specific to different proteins: while some may be very responsive to the availability of UDP-GlcNAc, others may have more constant levels of O-GlcNAcylation.

OGA consists of a catalytic N-terminal glycosidase domain and a C-terminal histone acetyl transferase (HAT) domain (Schultz and Pils, 2002). The Oga gene was originally identified as meningioma-expressed antigen 5 (MGEA5) and encodes two major variants, a long isoform and short isoform (Comtesse et al., 2001). The full-length variant localizes to the nucleus and cytoplasm whereas the short isoform localizes to the nucleus and lipid droplets

(Keembiyehetty et al., 2011). While less is known about its regulation, OGA is likely to also

form important interactions with its binding partners and substrates to guide its recruitment and

activity (Lagerlof and Hart, 2014).

OGT and OGA are both essential for embryonic development in mice and for the survival of many somatic cell types. Genetic disruption of Ogt is lethal in mouse embryonic stem cells, so generation of an OGT knockout mouse is not possible (Shafi et al., 2000). In vivo studies have therefore utilized the Cre-loxP recombination system to examine the effects of OGT loss in specific populations of cells. Loss of OGT in several mitotic cell types, such as T cells, B cells, mouse embryonic fibroblasts (MEFs), and pancreatic β-cells, results in cell death (O'Donnell et al., 2004; Chiang et al., 2013; Alejandro et al., 2015). OGT knockout in post-mitotic cells, such as cardiomyocytes and neurons, also affects cellular and organ function, although cellular viability appears to be less affected (Watson et al., 2010; Watson et al., 2014; Lagerlof et al.,

2016). Similarly, genetic disruption of Oga in mice results in neonatal lethality, and OGA knockout MEFs exhibit defects in mitosis, genomic stability, and metabolic pathways (Yang et al., 2012; Keembiyehetty et al., 2015).

Cellular functions of O-GlcNAcylation

Through its thousands of substrates, OGT plays a role in regulating many fundamental cellular processes. O-GlcNAc is abundant on transcriptional and translational machinery, including RNA polymerase II, transcription factors, chromatin proteins, and ribosomal proteins

(Ozcan et al., 2010; Lagerlof and Hart, 2014). Cytoskeletal proteins are also heavily O-

GlcNAcylated, including tubulin, actin, and intermediate filament proteins (Hart et al., 2011).

Additional processes mediated by the modification include protein trafficking, protein degradation, stress response, cell cycle, mitochondrial dynamics, and nuclear pore function.

Table 1.1. Examples of O-GlcNAc regulation of cellular processes

Cellular process Role of O-GlcNAcylation Reference Sin3A recruits OGT to promoters to cooperatively repress Yang et al., 2002 basal and Sp1-activated transcription. Transcription OGT cleaves HCF-1 to mediate its proteolytic maturation. Capotosti et al., 2011 OGT is essential for RNA pol II promoter recruitment. Lewis et al., 2016 O-GlcNAcylation of p67 mediates eIF-2 and translation Datta et al., 1989 inhibition. Translation 20 core ribosomal proteins are O-GlcNAcylated, including Zeidan et al., 2010 RPS6. O-GlcNAcylation regulates the solubility, organization, Srikanth et al., 2010 and stability of keratins. Cytoskeleton O-GlcNAcylation of tubulin inhibits its polymerization. Ji et al., 2011

O-GlcNAcylation regulates E-cadherin cell surface Zhu et al., 2001 Protein trafficking. trafficking O-GlcNAcylation regulates the nuclear localization of β- Sayat et al., 2008 catenin. O-GlcNAcylation of Rpt2 ATPase in the 19S cap inhibits Zhang et al., 2003 Protein proteasome function. degradation Protein ubiquitination is modulated by O-GlcNAcylation. Guinez et al., 2008 Cells with increased O-GlcNAcylation are more tolerant Zachara et al., 2004 to stress. Stress response Stress-induced O-GlcNAcylation increases HSP Kazemi et al., 2010 expression through inhibition of GSK-3β. Disruption of O-GlcNAc dynamics results in cell cycle defects, including delayed M-phase progression, altered Slawson et al., 2005 mitotic protein phosphorylation, and abnormal Cell cycle cytokinesis. OGT and OGA form a mitotic complex with Aurora B and Slawson et al., 2008 PP1 to regulate the post-translational status of vimentin. Increased O-GlcNAcylation enhances recruitment of Gawlowski et al., Mitochondrial DRP1 to mitochondria and DRP1 GTP-binding ability. 2012 dynamics OGT regulates mitochondrial motility through Pekkurnaz et al., 2014 modification of Milton. O-GlcNAcylation of Nup98 increases pore permeability. Labokha et al., 2013 Nuclear pore Loss of O-GlcNAc results in decreased levels of Nups. Zhu et al., 2016

Table 1.1 highlights some specific examples of O-GlcNAc regulation in these different areas.

The addition of the O-GlcNAc moiety can produce a range of effects on its protein substrates (Figure 1.1). O-GlcNAcylation can alter the localization of a protein, activate or inhibit enzymatic activity, affect the stability of a protein, and even influence an enzyme’s substrate specificity (Bond and Hanover, 2015). A major element of O-GlcNAc regulation is its extensive interplay with phosphorylation. Cross talk between these two modifications can exist at multiple levels (Hart et al., 2011). O-GlcNAcylation and phosphorylation can occur at the

same serine or threonine residues, and thus have reciprocal occupancy, or at proximal sites,

producing possible steric competition. Both modifications can also exist simultaneously on the

same protein. In addition to having shared substrates, O-GlcNAc and phosphate cycling enzymes

are modified by each other. OGT and OGA can form complexes with and modify kinases and

phosphatases, and OGT and OGA themselves are phosphorylated as well. Together, these two

post-translational modifications can dynamically regulate virtually all cellular processes.

1.2 O-GlcNAc as a metabolic sensor

Research over the years has unveiled a key role of O-GlcNAcylation in nutrient sensing

and the regulation of cellular pathways under varying metabolic conditions. The HBP, which

synthesizes UDP-GlcNAc, receives input from multiple metabolic pathways, and changes in the

flux of these pathways can result in changes in the production of UDP-GlcNAc (Figure 1.2). In

particular, 2-5% of the glucose in a cell is funneled into the HBP, so the levels of UDP-GlcNAc

are sensitive to glucose availability (Marshall et al., 1991).

Shifts in glucose levels have been linked to changes in O-GlcNAc levels, although the

effects depend on the particular protein, pathway, or cell type. In certain cases, changes in the

levels of glucose are parallel to those in O-GlcNAcylation: increased glucose results in increased

Figure 1.2. The hexosamine biosynthetic pathway

The hexosamine biosynthetic pathway (HBP) incorporates multiple nutrient-derived precursors into the synthesis of uridine diphosphate N-acetylglucosamine (UDP-GlcNAc), including glucose, glutamine, acetyl coenzyme A (acetyl coA), and uridine triphosphate (UTP). These precursors are derived from the glucose, amino acid, fatty acid, and nucleotide metabolic pathways, respectively. The rate-limiting step in this pathway is the incorporation of glutamine by the enzyme glucose:fructose 6-phosphate amidotransferase (GFAT).

O-GlcNAc levels in cells such as cardiomyocytes and pancreatic β-cells, presumably in part due

to the increased availability of the donor substrate (Han et al., 2000; Liu et al., 2000; Akimoto et

al., 2001; Clark et al., 2003; Walgren et al., 2003). In other cases, changes in the levels of

glucose are inverse to those in O-GlcNAcylation. For instance, glucose deprivation results in

increased O-GlcNAcylation in HepG2 liver cells and Neuro-2a neuroblastoma cells, due to feedback mechanisms such as the upregulation of OGT or downregulation of OGA (Cheung and

Hart, 2008; Taylor et al., 2008; Taylor et al., 2009). Overall, because OGT is the only enzyme to add the O-GlcNAc modification, it can act as a global modulator of cellular pathways in response to changes in nutrient availability.

Roles of OGT in peripheral metabolic processes

Given its role as a metabolic sensor, OGT has been widely studied in peripheral

metabolic tissues such as the pancreas and liver. OGT is highly enriched in the pancreas and

plays an important role in the regulation of insulin production through the modification of two

key β-cell-specific transcription factors, PDX1 and NeuroD1 (Hanover et al., 1999; Gao et al.,

2003; Andrali et al., 2007). Moreover, O-GlcNAcylation mediates glucose toxicity and β-cell

survival (Ruan et al., 2013). Aside from insulin production, O-GlcNAc also modulates the

insulin signaling cascade through the modification of many of the signaling proteins, including

insulin receptor (IR), insulin receptor substrate 1 and 2 (IRS1/2), and protein kinase B (AKT)

(Yang et al., 2008; Whelan et al., 2010). In the liver, OGT modifies several proteins involved in

gluconeogenesis, including the transcription factor FOXO1 and the transcriptional coactivators

PGC-1α and CRTC2, which mediate the expression of gluconeogenic genes (Dentin et al., 2008;

Housley et al., 2008; Ruan et al., 2012). O-GlcNAc modification increases their activity,

stability, and nuclear localization, respectively, and in this way promotes glucose production.

Aside from regulating the normal metabolic functions of various organs, O-

GlcNAcylation also plays a role in metabolic disease. Diabetes mellitus is a common metabolic disease worldwide, and through in vitro and in vivo studies, O-GlcNAcylation has been implicated in insulin resistance and glucose toxicity, two hallmarks of the disease. The HBP has long been linked to diabetes; in 1991, Marshall et al. demonstrated that an influx of nutrients, including high glucose and glutamine, causes insulin resistance in primary adipocytes and that this effect is prevented by pharmacologically inhibiting glucose:fructose 6-phosphate amidotransferase (GFAT), the rate-limiting enzyme of the HBP. Studies using transgenic mice further demonstrated that overexpression of GFAT and OGT in skeletal muscle, fat, and liver leads to insulin resistance (McClain et al., 2002; Yang et al., 2008). Increased levels of O-

GlcNAc are present in multiple tissues and cell types of diabetic animals, including the pancreas

(Akimoto et al., 2000), liver (Ruan et al., 2012), heart (Fricovsky et al., 2012; Ramirez-Correa et al., 2015), erythrocytes (Park et al., 2010), and leukocytes (Springhorn et al., 2012). Over time, this hyper-glycosylated state can have a detrimental impact on cellular functions, although more research is needed to understand the specific mechanisms. Altogether, O-GlcNAcylation of insulin signaling proteins, transcription factors, and numerous other proteins in peripheral tissues such as muscle, fat, liver, and pancreas appears to contribute to the pathogenesis of diabetes and diabetic complications.

1.3 O-GlcNAcylation in the nervous system

Although a large body of work has focused on the roles of O-GlcNAc signaling in

peripheral organs, OGT and OGA are in fact highly enriched in the brain. The highest level of

OGA mRNA and one of the highest levels of OGT mRNA and enzymatic activity are found in

the brain as compared to tissues such as heart, kidney, liver, lung, muscle, and spleen (Kreppel et

al., 1997; Gao et al., 2001). This is perhaps unsurprising given the metabolic demand of the

nervous system and therefore its need for mechanisms of nutrient sensing: the brain consumes an

estimated 20% of all glucose-derived energy in the body (Mergenthaler et al., 2013).

Additionally, many metabolic processes and autonomic functions are mediated by regions in the

brain, in particular the hypothalamus (Roh et al., 2016).

To study the effects of loss of OGT in neurons in vivo, O'Donnell et al. (2004) used the

pan-neuronal Cre driver, Synapsin1-Cre (Syn1-Cre), and demonstrated that neuronal OGT is

essential for the viability of mice. Knockout animals were born at 50% of the expected

frequency, suffered from locomotor deficits, and were smaller than their littermate controls.

They did not survive past postnatal day 10. Additional analysis of protein extracts from brain and

spinal cord tissue revealed hyperphosphorylated Tau in the mice lacking neuronal OGT.

However, due to the lethal nature of the pan-neuronal OGT knockout, these mice have only been

broadly characterized, with limited investigation into the effects on the knockout neurons

themselves. Therefore, little is known about how loss of OGT affects global neuronal health and

fundamental neuronal processes such as development, axonal projection, and axonal

maintenance.

The growing body of literature on the role of the O-GlcNAcylation in the nervous system

indicates that the modification does regulate neuronal function in various ways. An estimated

20% of mouse synaptosome proteins are O-GlcNAcylated, including synapsin I/II and bassoon,

presynaptic proteins important for neurotransmitter release, as well as alpha Ca2+/calmodulin- dependent protein kinase II (αCaMKII), a kinase crucial in mediating signaling cascades involved in learning and memory (Trinidad et al., 2012). O-GlcNAc modification of a single residue on synapsin I regulates its localization to synapses and the size of the reserve pool of

synaptic vesicles, thus having direct implications in presynaptic plasticity (Skorobogatko et al.,

2014). Furthermore, pharmacological elevation of O-GlcNAc levels in vivo enhances long-term potentiation (LTP), a persistent strengthening of synaptic transmission based on activity, and reduces paired-pulse facilitation (PPF), a short-term form of synaptic plasticity, in mouse hippocampal slices (Tallent et al., 2009).

Studies using dissociated neurons have also implicated O-GlcNAc in mediating neurite outgrowth and axonal branching. Rexach et al. (2012) found that loss of OGT through the expression of Cre recombinase in cultured Ogt-floxed mouse cortical neurons resulted in increased axonal length. This was attributed to decreased O-GlcNAcylation of the transcription factor CREB and the consequent increase in CREB-mediated transcription. Francisco et al.

(2009) found that decreased O-GlcNAc levels, through overexpression OGA in cultured chick forebrain neurons, resulted in increased axonal branching. Conversely, pharmacological inhibition of OGA with either the compound 9d or PUGNAc led to a decrease in the number of axonal filopodia. Neuron morphology, therefore, can be altered based on the O-GlcNAcylated state of the cell in vitro.

Recent studies have highlighted the role of OGT as a metabolic sensor in the nervous

system as well. Pekkurnaz et al. (2014) demonstrated the importance of OGT in regulating

axonal mitochondrial transport in hippocampal neurons under different levels of extracellular

glucose. Under conditions of high glucose, there is increased modification of the mitochondrial

motor-adaptor protein Milton, which results in an arrest in mitochondrial motility. Additionally,

OGT has specific functions in hypothalamic neurons to mediate thermogenesis and feeding

behavior. Ruan et al. (2014) discovered that OGT regulates the neuronal excitability of AgRP

neurons through modification of the voltage-dependent potassium channel, Kcnq3, and that loss

of OGT results in impaired AgRP neuronal activity and, consequently, enhanced thermogenesis

of white adipose tissue. Lagerlof et al. (2016) showed that OGT regulates excitatory synaptic function in αCaMKII-positive neurons of the paraventricular nucleus (PVN) and that this mediates feeding behavior in mice. Altogether, these studies in dissociated neurons and in vivo provide increasing evidence of the importance of O-GlcNAcylation in neuronal function.

O-GlcNAcylation in neurodegenerative disease

The role of O-GlcNAcylation has been implicated not only in physiological neuronal

processes but also in a number of neurodegenerative diseases. The Ogt gene maps to a locus

linked to X-linked dystonia-parkinsonism (Haberhausen et al., 1995; Shafi et al., 2000), and the

Oga gene maps to a locus linked to late-onset Alzheimer’s disease (Heckel et al., 1998; Bertram

et al., 2000). Many proteins that are key players in neurodegenerative diseases are modified by

O-GlcNAc, including tau, β-amyloid precursor protein (APP), and α-synuclein. The brains of

Alzheimer’s disease patients exhibit decreased global O-GlcNAcylation as well as decreased O-

GlcNAc modification of tau (Liu et al., 2004; Liu et al., 2009). Additional studies in Alzheimer’s

disease mouse models have demonstrated that enhanced O-GlcNAcylation results in decreased

tau aggregation and β-amyloid accumulation as well as decreased neuron loss and memory

impairment (Yuzwa et al., 2012; Kim et al., 2013). Recently, O-GlcNAcylation of α-synuclein

has also been found to decrease aggregation and toxicity of the protein in cultured cells (Marotta

et al., 2015). These studies suggest a protective role of O-GlcNAc in models of Alzheimer’s and

Parkinson’s disease, but further research is needed to determine any causal roles of O-GlcNAc

signaling.

O-GlcNAcylation has been studied in the context of other neurodegenerative diseases as

well, although to a lesser extent. Decreased O-GlcNAc levels have been found in the spinal cord

tissue and motor neurons of mSOD mice, a transgenic mouse model of amyotrophic lateral

sclerosis (ALS) that overexpresses mutant superoxide dismutase (Shan et al., 2012). Decreased

O-GlcNAcylation was also shown to reduce the toxicity of mutant huntingtin exon 1-coded

protein by regulating its clearance through increased autophagosome-lysosome fusion; this

helped to suppress neurodegeneration in a Drosophila model of Huntington’s disease (Kumar et

al., 2014).

Overall, O-GlcNAc signaling has been demonstrated to be involved in mediating

neuronal processes under both physiological conditions and neurodegenerative pathologies.

However, a basic understanding of the role of the modification in neurons is still lacking,

especially given the limited characterization of OGT knockout neurons in vivo. Are neurons

lacking OGT still able to develop normally and innervate their targets? If so, are they able to

survive long-term? Are they able to maintain their morphology and function, or do they

experience functional deficits and axonal degeneration? Investigating these questions in a

relevant and distinct neuronal population could elucidate the fundamental roles of O-

GlcNAcylation in neurons and how altered O-GlcNAc signaling could contribute to

neurodegeneration.

1.4 Primary sensory neurons

The peripheral nervous system is a system in which the role of OGT is poorly understood

but in which metabolic sensing may be particularly important. The central nervous system is

protected by the blood brain barrier, a vascular network that serves as a selective filter between

the circulating blood and the extracellular fluid in the brain and spinal cord. While common

nutrients are not excluded by this barrier, molecules such as glucose need to pass through the

specialized brain endothelial cells via facilitated diffusion, adding an additional layer at which

transport could be regulated (Campos-Bedolla et al., 2014). Central neurons are thus maintained

in a highly controlled environment, and extracellular concentrations of glucose range from 2 to 6

mM in the mammalian brain (Silver and Erecinska, 1994).

The cell bodies of primary somatosensory neurons, however, reside in dorsal root ganglia

(DRGs) that extend outside the spinal cord. Depending on the genetic background, there are 30-

31 pairs of DRGs in mice that are positioned by each vertebra of the spinal column: 8 cervical

(C1-C8), 13 thoracic (T1-T13), 5-6 lumbar (L1-L5/6), and 4 sacral (S1-S4) (Sleigh et al., 2016).

These neurons lie outside of the blood brain barrier and can therefore be exposed to greater changes in metabolic flux from the environment. Under physiological conditions, blood glucose levels fluctuate throughout the day between the fasting and fed states, from 3.9 to 7.8 mM in humans. This is augmented in diabetic patients, who may experience blood glucose levels of greater than 11 mM (American Diabetes Association, 2000). In a rat model of diabetes, hyperglycemia was indeed associated with a significant increase in the glucose concentrations of sciatic nerve extracts (Stewart et al., 1966). Peripheral neurons are thus vulnerable to glucose toxicity under these conditions.

Primary sensory neurons have a pseudo-unipolar structure: their cell bodies project a single axon that bifurcates into a distal process that extends to the periphery and a proximal central process that enters into the dorsal column of the spinal cord (Figure 1.3). These neurons are responsible for relaying afferent sensory information from peripheral tissues such as the skin, visceral organs, and muscles to neurons in the central nervous system. Sensory neurons are highly heterogeneous and consist of subpopulations that respond specifically to different stimuli

(Table 1.2) (Dubin and Patapoutian, 2010; Abraira and Ginty, 2013; Catala and Kubis, 2013;

Chiu et al., 2014). The neuronal subtypes differ by various characteristics, including size, degree

Figure 1.3. Sensory neuron structure

The cell bodies of primary sensory neurons reside in dorsal root ganglia that are adjacent to the spinal cord. The neurons extend a single axon that splits into two processes: one that extends to the periphery and innervates the skin, muscles, and other organs, and one that extends into the spinal cord to relay afferent information to neurons of the central nervous system.

Table 1.2. Classification of primary somatosensory neuron fiber types

Nerve fiber type Characteristics Receptor type Sensory modality Myelinated, large Muscle spindle, Golgi tendon Aα (Group I) diameter, fastest Proprioception organ conduction Muscle spindle, Meissner’s Myelinated, large corpuscle, Merkel cell, Aβ (Group II) Touch, vibration diameter, fast conduction Pacinian corpuscle, Ruffini ending, lanceolate endings Myelinated, medium Pain, temperature, Aδ (Group III) diameter, intermediate Free nerve endings touch conduction Unmyelinated, small Pain, temperature, C (Group IV) Free nerve endings diameter, slow conduction itch

of myelination, conduction velocity, receptor expression, and molecular markers. Thermal,

mechanical, proprioceptive, nociceptive, and other sensations are thus transduced through these

specialized neurons and integrated into the central nervous system, where a response can be

evoked.

Sensory neuropathy

Peripheral neuropathy consists of a set of disorders in which neurons of the peripheral

nervous system are unable to function properly, leading to motor, sensory, or autonomic

abnormalities. One of the most common types of peripheral neuropathy is small fiber sensory

neuropathy. Patients with small fiber neuropathy exhibit positive sensory symptoms such as

tingling, burning, prickling, and pain, as well as negative sensory symptoms such as numbness

and loss of sensation (Lacomis, 2002). These symptoms occur most commonly in the distal

regions of the body, such as the hands and feet, suggesting a length-dependent nature of the pathology. Often, patients present with decreased intraepidermal nerve fiber density as evaluated from a skin biopsy.

Sensory neuropathy usually occurs secondary to a systemic medical condition, and several causes have been identified, including viral infections, autoimmune disorders, and the use of chemotherapeutic agents (Lacomis, 2002). The most common form of sensory neuropathy is diabetic neuropathy, affecting over 25% of diabetic patients (Ziegler, 2006). Clinically, diabetic neuropathy can manifest in many forms, and any of the sensory nerve fiber types could be affected to varying degrees or at different stages of the disease. Damage to the small unmyelinated C fibers is common and often occurs early, resulting in symptoms such as loss of pain and thermal sensation. While the underlying mechanisms contributing to diabetic neuropathy are not fully understood, hyperglycemia is believed to be the critical initiator of

neuropathy (Vincent et al., 2013). Given the susceptibility of peripheral sensory neurons to glucose toxicity, pathways involved in metabolic sensing and regulation may be critical in mediating the development of diabetic neuropathy.

It is conceivable, then, that altered O-GlcNAc signaling is one of the contributing factors towards neuropathy. In a recent study, Kim et al. (2016) demonstrated that OGT is important in the peripheral nervous system by deleting the gene specifically in Schwann cells, the glia that ensheath peripheral axons. Mice lacking OGT in Schwann cells developed tomaculous demyelinating neuropathy, characterized by progressive demyelination, impaired sensory function, and axonal degeneration. Studies in sensory neurons would further elucidate the potential roles of OGT in neuronal health and peripheral neuropathy.

1.5 Overview of dissertation

My studies have focused on investigating the role of OGT in primary sensory neurons by

genetically and pharmacologically manipulating the levels of O-GlcNAcylation. In doing so, I

aimed to address how altered O-GlcNAc signaling affects fundamental neuronal processes, such

as development, axonal outgrowth, axonal maintenance, and survival.

My primary approach was to characterize the effects of OGT loss specifically in sensory

neurons using tissue-specific Cre-loxP recombination to mutate the Ogt gene. I generated OGT

knockout mice by crossing mice that carry the floxed Ogt allele with mice that have the Cre

transgene under either a Nav1.8 promoter or a brn3a promoter. In the floxed Ogt allele, amino

acids 206-232 of the complete 1,037 amino acid Ogt sequence are encompassed by loxP sites,

such that excision of the exon results in a frameshift, leading to a highly truncated form of OGT that consists of only the first 206 amino acids (Shafi et al., 2000). This form of OGT thus lacks the C-terminal domain and any glycosyltransferase activity. Tissue-specific excision was driven

by either Nav1.8-Cre or brn3a-CreERT2. Nav1.8 (SNS) is a voltage-gated sodium channel expressed specifically in sensory neurons of the dorsal root, trigeminal, and nodose ganglia, with no expression in other peripheral or central neurons, glia, or non-neuronal tissues based on in situ hybridization studies (Akopian et al., 1996; Stirling et al., 2005). Early reports identified expression in up to 85% of neurons with C fibers and in around 10% of neurons with A fibers, particularly the low-threshold mechanoreceptors (Amaya et al., 2000; Djouhri et al., 2003). A more recent study suggests that Nav1.8 expression may be present in an even greater percentage of sensory neurons: 75% of all DRG neurons and up to 40% of A-fiber neurons (Shields et al.,

2012). The Nav1.8-Cre transgene exhibits a similar sensory neuron-specific pattern of expression

(Agarwal et al., 2004). Brn3a is a POU-domain transcription factor expressed in sensory neurons of the dorsal root and cranial ganglia, as well as in specific neurons in the caudal central nervous system, including the medial and lateral habenula, specific tectal lamina, red nucleus, interpeduncular nucleus, mesencephalic trigeminal nucleus, and inferior olive (Eng et al., 2001).

However, the brn3a-CreERT2 transgene construct was engineered using a promoter sequence that

included 11 kb of 5′-flanking sequence, which resulted in specific LacZ reporter expression in

the trigeminal (fifth) ganglia, vestibulocochlear (eighth) ganglia, superior (ninth) ganglia, and

DRGs, with no expression in the various central nuclei (Eng et al., 2001; O’Donovan et al.,

2014).

Using this in vivo knockout approach, I found that loss of OGT in the sensory neurons of

mice results in decreased weight and hyposensitivity to thermal and mechanical stimuli in

behavioral assays. The sensory deficits are caused by decreased epidermal innervation and loss

of DRG cell bodies. I determined that these effects are present in postnatal development as well

as in adulthood and conclude that OGT is essential for sensory neuron maintenance and survival.

I further explored specific neuronal processes in cultured sensory neurons and found that altered

O-GlcNAcylation affects axonal outgrowth and mitochondrial trafficking. Altogether, this work demonstrates that OGT is critical for neuronal survival and supports the hypothesis that changes in O-GlcNAc dynamics can contribute to neuropathy and neurodegeneration.

Chapter 2

O-GlcNAc transferase is essential for sensory neuron survival and maintenance

The body of work in this chapter forms the manuscript: Su C and Schwarz TL (2017) O-GlcNAc transferase is essential for sensory neuron survival and maintenance. J Neurosci 37:2125-2136.

Contributions: All experiments were designed by Cathy Su and Thomas Schwarz. Experiments and data analysis were performed by Cathy Su. 2.1 Introduction

O-GlcNAcylation is a dynamic post-translational modification that is present on

thousands of nuclear and cytoplasmic proteins (Ma and Hart, 2014). The addition of the O-linked

β-N-acetylglucosamine (O-GlcNAc) moiety is catalyzed by a single enzyme called O-GlcNAc

transferase (OGT) (Haltiwanger et al., 1992). Through its vast array of substrates, OGT regulates

a range of cellular processes, including transcription, translation, protein trafficking, protein

degradation, and stress response (Hart et al., 2007; Bond and Hanover, 2015). Moreover, the

levels of the donor substrate, UDP-GlcNAc, and therefore of the modification, are influenced by

the availability of glucose and other nutrients, allowing O-GlcNAcylation to play a key role in

metabolic sensing (Zachara and Hart, 2004; Ruan et al., 2013). Several studies have

demonstrated the significance of O-GlcNAc signaling in peripheral metabolic processes,

including insulin signaling (Yang et al., 2008; Whelan et al., 2010), β-cell function (Soesanto et

al., 2011; Durning et al., 2016), and gluconeogenesis (Dentin et al., 2008; Ruan et al., 2012).

Recent studies have also begun to focus on the roles of OGT in neuronal populations in the

context of metabolism and have identified OGT in hypothalamic neurons as a regulator of

thermogenesis (Ruan et al., 2014) and feeding (Lagerlof et al., 2016).

Despite its relative abundance in the nervous system (Kreppel et al., 1997; Gao et al.,

2001), little is still known about the global function of O-GlcNAcylation in neurons. An

estimated 20% of synaptosome proteins are O-GlcNAcylated (Trinidad et al., 2012), and the

modification has been implicated in processes such as neurite outgrowth (Rexach et al., 2012),

axonal branching (Francisco et al., 2009), synaptic plasticity (Tallent et al., 2009), and

mitochondrial trafficking (Pekkurnaz et al., 2014). Altered O-GlcNAc signaling has also been

shown to play a role in neurodegenerative diseases: multiple proteins involved in Alzheimer’s

disease and Parkinson’s disease have been shown to be O-GlcNAcylated, including tau, amyloid

precursor protein, and α-synuclein (Lazarus et al., 2009; Wani et al., 2016). However, because a pan-neuronal knockout of OGT in mice results in locomotor defects and postnatal mortality by

day 10, the effects of OGT loss could only be very broadly characterized (O'Donnell et al.,

2004). The in vivo significance of OGT in neurons remains little understood, particularly in

fundamental processes such as development, axonal projection, and survival.

Given its role as a metabolic sensor, one neuronal population in which OGT may be

important is the peripheral nervous system and, in particular, sensory neurons. These neurons

have cell bodies in ganglia along the spine and project sensory axons to multiple organs in the

periphery, including the skin, and central axons that enter the spinal cord (Catala and Kubis,

2013). Because the cell bodies and sensory axons lie outside of the blood brain barrier, they may

be subject to greater fluctuations in nutrient availability than central neurons (Tomlinson and

Gardiner, 2008). In particular, blood glucose levels normally change significantly between fasted

and fed states, and there can be even greater variations in pathological conditions such as

diabetes. Diabetic neuropathy is a common complication of diabetes that causes patients to lose

proper function of their peripheral nerves and experience symptoms such as numbness, pain, or

loss of sensation starting at the extremities (Zochodne, 2008). Additionally, a recent study

demonstrated that mice with OGT deleted in Schwann cells develop demyelinating peripheral

neuropathy (Kim et al., 2016). Therefore, OGT function likely has great significance in the

peripheral nervous system.

As a first step to understanding the role of OGT in peripheral sensory neurons, I

investigated the importance of OGT in the basic development and maintenance of sensory

neurons through the deletion of OGT. I show that sensory neuron-specific knockout of OGT in

mice results in functional and anatomical deficits and affects neuronal maintenance and survival over time. OGT is therefore essential for sensory neuron function, axonal projection, and long- term survival.

2.2 Results

Nav1.8-Ogt knockout mice are overtly healthy but exhibit altered weight and glucose

tolerance

Pan-neuronal loss of OGT has previously demonstrated an essential role of neuronal

OGT for the viability of the animals (O'Donnell et al., 2004), but due to their short lifespan, little

is understood about the characteristics of neurons lacking OGT. To investigate the role of OGT

in the ability of neurons to survive, project axons, and properly function in vivo, I used the

transgenic Nav1.8-Cre driver to selectively delete OGT in a major subset of DRG sensory

neurons. These neurons relay sensory input from the skin and other peripheral organs to the

central nervous system (Agarwal et al., 2004). Nav1.8 is a voltage-gated sodium channel expressed in the majority of DRG neurons with C fibers, and Nav1.8-expressing neurons comprise peptidergic and non-peptidergic nociceptors involved in sensing noxious thermal stimuli, as well as a percentage of A-fibers neurons, which are involved in sensing mechanical stimuli (Shields et al., 2012). Nav1.8-Cre expression is limited to the sensory ganglia as visualized through LacZ staining using ROSA26-LacZ reporter mice; staining is completely absent in sections of the spinal cord, brain, and brainstem, as well as in visceral and skeletal organs such as the liver, lung, heart, intestine, and skeletal muscles (Agarwal et al., 2004). I hypothesized that the knockout of OGT in only this subset of neurons would result in viable animals and allow for a more comprehensive characterization of the effects of OGT loss.

Nav1.8-Ogt knockout mice (Nav1.8-Cre-/+ OgtloxP/Y) were born at an expected rate and

developed normally alongside their littermate controls. Loss of OGT was assessed in 4-week-old mice by immunostaining for the O-GlcNAc modification and peripherin (Figure 2.1). Peripherin

is an intermediate filament expressed in small-diameter DRG neurons with C fibers and has

significant overlap with Nav1.8 expression: around 90% of peripherin-positive neurons contain

Nav1.8 and conversely around 60% of Nav1.8-positive cells contain peripherin (Shields et al.,

2012). Therefore, I used peripherin staining to roughly distinguish cells that would most likely be

Nav1.8-positive and therefore OGT-negative in this system. I observed the presence of O-

GlcNAc in only 23.7 ± 3.4% of peripherin-positive DRG neuron cell bodies in Nav1.8-Ogt

knockout tissue (Nav1.8-Cre-/+ OgtloxP/Y) as opposed to 94.4 ± 0.3% of peripherin-positive cell bodies in control tissue (Nav1.8-Cre-/- OgtloxP/Y) (n = 3 animals per genotype, 5 sections per

animal, p < 0.0001 by unpaired t-test) and thus achieved knockout in the large majority of small-

diameter DRG sensory neurons.

A global phenotype was assessed at 8 weeks of age using the modified SHIRPA protocol,

in which animals were scored for a range of morphological and behavioral phenotypes (see

Materials and Methods), and no significant differences were observed between Nav1.8-Ogt knockout and control mice (Figure 2.2, A). Nav1.8-Ogt knockout mice also displayed normal

locomotor activity and forelimb grip strength (Figure 2.2, B-C). Overall, the knockout mice were

viable up to at least 1 year of age, the longest length of time they were kept before being

euthanized, and no unexpected deaths were observed. Therefore, loss of OGT in sensory neurons

does not affect the global development and behavior of mice and can serve as an appropriate

system to further investigate its effects in neurons.

The one global morphological phenotype that was observed, however, was that Nav1.8-

Ogt knockout mice weighed less than their littermate controls, as measured at either 8 or 30

Figure 2.1. Nav1.8-Cre-mediated Ogt knockout occurs in the majority of small-diameter DRG neurons

Immunostaining of O-GlcNAc (magenta) and peripherin (green) in DRGs of control and Nav1.8-Ogt knockout (KO) mice to validate loss of O-GlcNAcylation. The remaining O-GlcNAc-positive cells are mostly peripherin-negative and presumably belong to the subset of sensory neurons that do not express

Nav1.8. Scale bar, 50 µm.

Figure 2.2. Nav1.8-Ogt knockout mice exhibit normal global and motor phenotypes

A 8-week-old mice were observed and scored in a phenotypic test battery following the modified

SHIRPA protocol. B Locomotor activity was measured by counting the total number of grid crossings of

8-week-old mice placed in an arena for 30 s. C Grip strength was measured as the average of three consecutive readings using a grip strength meter with 8-week-old mice. n = 16 control and 22 Nav1.8-Ogt knockout animals, unpaired t-test. All values shown as mean ± SEM.

Figure 2.3. Nav1.8-Ogt knockout mice exhibit altered weight

A Weights of control and Nav1.8-Ogt knockout mice at 8 and 30 weeks of age. n = 16-22 animals per

genotype, unpaired t-test, ****p < 0.0001. B Weights of wild-type (WT) and Nav1.8-Cre-positive mice at

8 weeks of age. n = 9 animals per genotype, unpaired t-test.

All values shown as mean ± SEM.

Figure 2.4. Nav1.8-Ogt knockout mice exhibit altered glucose tolerance

Blood glucose was measured after intraperitoneal injection of a dextrose solution (2 g glucose/kg body mass) in fasted mice at 8 and 32 weeks of age. n = 9 animals per genotype at each age, two-way repeated measures ANOVA with Bonferroni correction (p = 0.007 at 8 weeks, p = 0.031 at 32 weeks), *p < 0.05,

**p < 0.01. All values shown as mean ± SEM.

31 weeks of age (Figure 2.3, A). Animals expressing only Nav1.8-Cre without the floxed Ogt allele did not exhibit decreased weight, ruling out effects from Cre expression (Figure 2.3, B). To probe whether the difference in weight reflects differences in metabolism, I performed an intraperitoneal glucose tolerance test on mice at 8 weeks and 32 weeks of age. Nav1.8-Ogt knockout mice consistently exhibited increased glucose tolerance, as indicated by lower blood glucose levels, compared to control mice at 30, 60, 90, and 120 min after a glucose challenge

(Figure 2.4). Together, these data suggest that loss of OGT in sensory neurons could lead to altered metabolic processes in mice.

Nav1.8-Ogt knockout mice show behavioral deficits in thermal and mechanical sensitivity

To assess the behavioral consequences of OGT knockout in sensory neurons, I conducted assays aimed at testing sensitivity to thermal and mechanical stimuli. At 10 weeks of age,

Nav1.8-Ogt knockout mice exhibited a significant difference in response in the thermal place preference assay. This assay pairs a 30°C plate with one at 49°C, a temperature that normally evokes an aversive behavior. Whereas the control animals had a strong preference for the 30°C plate over the adjacent 49°C plate, knockout animals spent on average more time on the hotter plate than control animals. This hyposensitivity to the changes in plate temperature is visualized by a decrease in amplitude of the curve representing the average duration of time spent on each plate. The effect appeared to be slightly more severe at 30 weeks of age (Figure 2.5, A-B). In the hot plate assay, 10-week-old Nav1.8-Ogt knockout mice exhibited slower hindpaw responses when placed on a 52°C plate, but not on 49°C or 55°C plates. This increased response time was more pronounced as the animals aged, and by 30 weeks of age, knockout mice were taking over

50% longer to respond than control mice at all three temperatures (Figure 2.5, C).

The von Frey assay was conducted as a test of hindpaw mechanical sensitivity. At 11

Figure 2.5. Nav1.8-Ogt knockout mice show clear behavioral deficits to thermal stimulation

A Average duration of time in seconds that 10-week-old and 30-week-old mice spent on a plate in the thermal place preference assay as the plate oscillated between the temperatures of 30°C and 49°C, as indicated by the bottom graph. A random, temperature-insensitive distribution would result in an average of 15 s baseline on each plate throughout the assay. n = 17-21 animals per genotype. B The thermal place preference assay shown in (A) was quantified by measuring the total area between the curve and the 15 s baseline for one plate starting at 2 min. Unpaired t-test, ***p < 0.001, ****p < 0.0001. C Hot plate assay at 10 weeks and 30 weeks of age. The latency was measured for a hindpaw response following placement of the animal on a plate at the indicated temperatures. The 3 temperatures were tested on 3 consecutive days. n = 15-23 animals per genotype, unpaired t-test, *p < 0.05, **p < 0.01, ****p < 0.0001.

All values shown as mean ± SEM.

Figure 2.5 (Continued)

Figure 2.6. Nav1.8-Ogt knockout mice show mild behavioral deficits to mechanical stimulation

Number of hindpaw responses out of 10 applications of von Frey fiber at each corresponding pressure for

11-week-old and 31-week-old animals. Two trials were conducted over consecutive days and averaged

for each animal. n = 17-23 animals per genotype, two-way repeated measures ANOVA with Bonferroni

correction (p = 0.261 at 11 weeks, p = 0.092 at 31 weeks), ***p < 0.001. Because strong stimuli appeared

to overcome any deficits in the Nav1.8-Ogt knockout mice, individual unpaired t-tests at each pressure were also conducted to assess the difference between the genotypes. For 11-week-old animals: p = 0.016

at 0.04 g, p = 0.070 at 0.07 g, p = 0.009 at 0.16 g, p = 0.217 at 0.4 g, p = 0.417 at 0.6 g, and p = 0.987 at

1.0 g. For 31-week-old animals: p = 0.012 at 0.04 g, p = 0.065 at 0.07 g, p = 0.002 at 0.16 g, p = 0.459 at

0.4 g, p = 0.704 at 0.6 g, and p = 0.615 at 1.0 g. These comparisons suggest a loss of sensitivity to weaker stimuli at both ages. All values shown as mean ± SEM.

weeks of age, Nav1.8-Ogt knockout mice exhibited largely normal responses to punctate forces

as compared to control mice, though slight decreases in the numbers of responses were observed

at forces ranging from 0.16 to 0.6 g. At forces of ≥ 1.0 g, both knockout and control mice were responsive to the fibers, whereas at forces of < 0.07 g the mice were equally insensitive to the fibers. At 31 weeks of age, both mutant and control mice appeared to be more sensitive to the von Frey fibers. Knockout mice exhibited slightly fewer responses to the forces ranging from

0.04 to 0.16 g, and the effect was statistically significant at 0.16 g, with knockout mice responding less than half as often as control mice (Figure 2.6).

Altogether, the data indicate a clear hyposensitivity to thermal stimuli and a mild hyposensitivity to mechanical stimuli in Nav1.8-Ogt knockout mice. This demonstrates that loss of OGT results in functional deficits observed at the behavioral level in mice and that its effects can worsen with age.

Nav1.8-Ogt knockout mice lose epidermal innervation and DRG cell bodies

Sensory inputs relaying thermal and mechanical information originate in free nerve endings in the skin. To understand the cause of the behavioral deficits, I examined the small fiber innervation of glabrous hindpaw epidermis from control and Nav1.8-Ogt knockout mice.

Epidermal innervation was severely reduced already at 8 weeks of age in the knockout animals

(Figure 2.7, A), despite the relatively mild behavioral effects seen at that age. I therefore examined the innervation throughout postnatal development and as the animals aged. In control mice, the density of epidermal innervation increases for several days after birth, and is maximal between postnatal days 4 and 10 (P4-P10). The density of innervation then declines gradually to reach stable adult levels by 3 weeks of age (Figure 2.7, B). Nav1.8-Ogt knockout mice had consistently fewer nerve fiber endings in the skin already at P1 (Figure 2.7, A-B). Although this

Figure 2.7. Decreased epidermal innervation in Nav1.8-Ogt knockout mice

A PGP9.5 immunostaining (green) to visualize nerve fibers (white arrowheads) and Hoechst staining

(blue) to visualize the keratinocyte layer in glabrous hindpaw skin at 1 day (P1), 8 weeks, and 34 weeks of age. Scale bar, 50 µm. B Quantification of fiber number/250 μm of epidermal length. n = 3 animals per genotype at each age (5 sections per animal), unpaired t-test, *p < 0.05, **p < 0.01, ***p < 0.001, ****p

< 0.0001. All values shown as mean ± SEM. C Hairy back skin, immunostained as in (A), at 1 day and 8 weeks of age. Scale bar, 20 µm.

Figure 2.7 (Continued)

Figure 2.8. Loss of peripherin-expressing neuronal cell bodies in Nav1.8-Ogt knockout DRGs

A Immunostaining of L5 DRGs from 34-week-old mice with anti-peripherin (green) to visualize peripherin-positive (white arrowheads) and peripherin-negative (purple arrowheads) neuronal cell bodies.

Scale bar, 100 µm. B Cell bodies were counted every 42 μm through serially sectioned L5 DRGs from

10-day-old and 34-week-old mice. n = 3 animals per genotype (2 DRGs from each animal), unpaired t- test, **p < 0.01, ****p < 0.0001. All values shown as mean ± SEM.

innervation increased between P1 and P10, there remained 40-60% fewer nerve fibers/µm of

skin in the knockout animals than the controls. After 3 weeks of age, when the number of nerve

fibers in control animals had stabilized, innervation continued to decline in the knockout mice

until at 34 weeks of age almost no fibers remained in the skin. A similar phenomenon was

observed in hairy skin from the back (Figure 2.7, C). These results suggest that the decrease in

sensitivity is due to a progressive degeneration of epidermal nerve endings rather than intact but

dysfunctional nerves.

To determine whether this axonal degeneration is accompanied by loss of cell bodies, I

counted neuronal cell bodies in serial sections immunostained for peripherin from L5 DRGs of

control and Nav1.8-Ogt knockout mice. Brightly fluorescent peripherin-positive cells were selected as likely to represent small-diameter neurons and therefore much of the Nav1.8- expressing population targeted for loss of OGT. Cell bodies that displayed weak background fluorescence were analyzed separately as peripherin-negative cells. At P10, there were fewer peripherin-positive cell bodies in DRGs from Nav1.8-Ogt knockout animals than in controls and a less significant decrease in peripherin-negative cell bodies. At 34 weeks of age, many more cell bodies, and particularly the peripherin-positive population, appeared to have been lost (Figure

2.8). The smaller effect on peripherin-negative neurons can be attributed to the imperfect proxy of peripherin as a marker for Nav1.8-positive neurons; the expression pattern of Nav1.8-Cre

includes a fraction of large-diameter neurons as well (Agarwal et al., 2004). These findings

demonstrate that OGT is not only critical for epidermal innervation but also for neuronal

survival.

OGT knockout neurons exhibit axonal outgrowth deficits in culture

The loss of skin innervation and cell bodies in Nav1.8-Ogt knockout animals prompted

Figure 2.9. Decreased axonal outgrowth in cultured OGT knockout neurons

A Immunostaining of β-tubulin III (Tuj1) (green) and O-GlcNAc (red) to visualize neurite outgrowth of

dissociated control and OGT knockout neurons cultured for 12 h. Scale bar, 100 µm. B Quantification of

longest neurite length per neuron. n = 150 control and 162 Nav1.8-Ogt knockout cells from 3 animals per

genotype, unpaired t-test, *p < 0.05. C Quantification of neurite crossings at increasing distances from the

soma center using Sholl analysis. n = 152 control and 163 Nav1.8-Ogt knockout cells from 3 animals per

genotype, two-way repeated measures ANOVA with Bonferroni correction (p < 0.0001), *p < 0.05,

****p < 0.0001. D Quantification of neurite crossings, as in (C), to examine the effects of Cre

expression. n = 60 cells from 2 animals per genotype, two-way repeated measures ANOVA with

Bonferroni correction (p = 0.998). E Quantification of neurite crossings, as in (C), of large neurons only.

n = 38 control and 32 Nav1.8-Ogt knockout cells from 3 animals per genotype, two-way repeated

measures ANOVA with Bonferroni correction (p < 0.0001), *p < 0.05, ****p < 0.0001. F Quantification of neurite crossings, as in (C), of small neurons only. n = 114 control and 130 Nav1.8-Ogt knockout cells from 3 animals per genotype, two-way repeated measures ANOVA with Bonferroni correction (p =

0.022), ****p < 0.0001.

All values shown as mean ± SEM.

Figure 2.9 (Continued)

me to examine knockout neurons in vitro to assess their morphology and survival. DRG neurons

were dissected from 4-week-old control and Nav1.8-Ogt knockout mice, sparsely plated, and cultured for 12 h. Axons of individual neurons were visualized with β-tubulin III

immunostaining, and OGT knockout neurons were identified by the absence of O-GlcNAc immunostaining (Figure 2.9, A). Although the extent of axonal outgrowth varied considerably for each genotype, neurons lacking OGT were on average shorter and had fewer neurites (Figure

2.9, B-C). To determine whether only a subset of the neuronal population was contributing to this effect, the data from large (> 30 µm soma diameter) and small (≤ 30 µm soma diameter) neurons were analyzed separately. A significant difference in axonal outgrowth was observed for both cell types, though the decrease was more significant in the larger cells, perhaps because they had more neurites to lose (Figure 2.9, E-F). Neurons expressing only Nav1.8-Cre did not exhibit any differences in axonal outgrowth (Figure 2.9, D). These results suggest an effect of

OGT on axonal growth and branching, although the selective loss of a subset of fast-growing, highly branched sensory neurons is also a possibility.

Loss of epidermal innervation and cell bodies occurs independently of developmental processes

The Nav1.8-Ogt knockout mouse revealed that loss of OGT decreases epidermal innervation and DRG cell bodies. These effects were observed as early as P1, and while they appeared to progress, it was unclear whether developmental defects were the primary cause of the later phenotypes. I therefore turned to a tamoxifen-inducible Cre driver, brn3a-CreERT2, to

distinguish the effects of OGT loss on neuronal development and maintenance (O’Donovan et

al., 2014). Brn3a is a POU-domain transcription factor expressed in all DRG sensory neurons,

and through the inducible system, I deleted OGT in adult mice. Cre expression was induced at 7

weeks of age with tamoxifen injections. Two weeks later, loss of OGT was evaluated by O-

GlcNAc immunostaining (Figure 2.10). O-GlcNAc immunoreactivity was present in 46.7 ± 6.2%

of peripherin-positive DRG neuron cell bodies in brn3a-Ogt knockout ganglia (tamoxifen-treated

brn3a-CreERT2-/+ OgtloxP/Y) as opposed to 94.2 ± 1.5% of peripherin-positive cell bodies in

control ganglia (vehicle-treated brn3a-CreERT2-/+ OgtloxP/Y) (n = 5 animals per genotype, 5 sections per animal, p < 0.0001 by unpaired t-test). Thus, although the majority of peripherin- positive DRG neurons had lost OGT, the inducible Cre driver was not as penetrant as the

Nav1.8-Cre driver in the knockout mice.

As with the Nav1.8-Ogt knockout mice, I examined epidermal innervation, cell body count, and thermal sensitivity in brn3a-Ogt knockout mice (tamoxifen-treated brn3a-CreERT2-/+

OgtloxP/Y) to determine the consequences of OGT loss in adult sensory neurons. At 2 weeks post- treatment, there was a slight decrease in the number of nerve fibers in hindpaw epidermis from brn3a-Ogt knockout animals compared to that of vehicle control animals (vehicle-treated brn3a-

CreERT2-/+ OgtloxP/Y) and tamoxifen control animals (tamoxifen-treated brn3a-CreERT2-/-

OgtloxP/Y) (Figure 2.11, A-B) but no significant difference in L5 DRG cell body counts (Figure

2.11, C-D) or thermal place preference behavior (Figure 2.12). At 4 weeks post-treatment, there

was a much larger decrease in epidermal innervation, with 60% fewer fibers in brn3a-Ogt

knockout animals than in control animals (Figure 2.11, A-B). There were also fewer L5 DRG

cell bodies in the mice lacking OGT (Figure 2.11, C-D). These mice still exhibited normal to

mildly hyposensitive thermal place preference behavior (Figure 2.12). Because OGT was deleted

when the sensory neurons would have been fully developed and innervating their peripheral targets, these results demonstrate that OGT has an important role in the maintenance and survival of adult neurons. The time course of the phenotypes also suggests that loss of OGT first affects

Figure 2.10. Brn3a-CreERT2-mediated Ogt knockout occurs in the majority of small-diameter DRG neurons

Immunostaining of O-GlcNAc (magenta) and peripherin (green) in DRGs of control and brn3a-Ogt knockout mice to validate loss of O-GlcNAcylation. Scale bar, 50 µm.

Figure 2.11. OGT knockout in adult animals causes loss of epidermal innervation and DRG cell

bodies

A PGP9.5 immunostaining (green) to visualize nerve fibers (white arrowheads) and Hoechst staining

(blue) to visualize the keratinocyte layer in glabrous hindpaw skin at 2 and 4 weeks after either vehicle or

tamoxifen injection to induce Ogt knockout in adult brn3a-Ogt mice. Vehicle control and brn3a-Ogt knockout mice were of the genotype brn3a-CreERT2-/+ OgtloxP/Y and tamoxifen control mice were of the genotype brn3a-CreERT2-/- OgtloxP/Y. Scale bar, 50 µm. B Quantification of fiber number/250 μm of

epidermal length for genotypes and injections as in (A). n = 6-7 animals per condition at 2 weeks and 5 animals per condition at 4 weeks (5 sections per animal), one-way ANOVA with post hoc Fisher’s LSD test (p = 0.055 at 2 weeks, p = 0.001 at 4 weeks), *p < 0.05, ***p < 0.001. C Immunostaining of L5

DRGs from mice 4 weeks post-treatment with anti-peripherin (green) to visualize peripherin-positive

(white arrowheads) and peripherin-negative (purple arrowheads) neuronal cell bodies. Scale bar, 100 µm.

D Cell bodies were counted every 42 μm through serially sectioned L5 DRGs from mice 2 and 4 weeks post-treatment. n = 5 animals per condition (2 DRGs from each animal), unpaired t-test, *p < 0.05, **p <

0.01.

All values shown as mean ± SEM.

Figure 2.11 (Continued)

Figure 2.12. Brn3a-Ogt knockout mice show normal thermal sensitivity at 4 weeks after Cre

induction

A Average duration of time in seconds that mice 2 and 4 weeks post-treatment spent on a plate in the thermal place preference assay as the plate oscillated between the temperatures of 30°C and 49°C, as indicated by the bottom graph. A random, temperature-insensitive distribution would result in an average of 15 s baseline on each plate throughout the assay. n = 13-15 animals per genotype. B The thermal place preference assay shown in (A) was quantified by measuring the total area between the curve and the 15 s baseline for one plate starting at 2 min. Unpaired t-test.

All values shown as mean ± SEM.

the axons in the periphery and causes loss of the cell body subsequent to axonal degeneration.

2.3 Discussion

O-GlcNAcylation is a post-translational modification that broadly modulates cellular processes as metabolic conditions vary (Bond and Hanover, 2015). Previous studies examined the effects of complete neuronal loss of OGT, whereupon the early death of the mice precluded detailed analysis (O'Donnell et al., 2004). Other studies have deleted OGT in a specific neuronal subpopulation to investigate a particular metabolic function (Ruan et al., 2014; Lagerlof et al.,

2016). Here, I examined the effects of OGT knockout in primary sensory neurons to gain a better understanding of the importance of OGT in fundamental neuronal processes such as axonal projection and long-term maintenance and survival. I found that loss of OGT in a subset of sensory neurons decreased skin innervation and DRG cell bodies early in postnatal development.

These phenotypes worsened with age and manifested in functional deficits in thermal and mechanical sensation. These phenotypes were not solely due to disruption of developmental processes, as knockout of OGT in adult mice also decreased skin innervation and cell body counts.

Major consequences of OGT loss in sensory neurons are perhaps unsurprising given the number of protein substrates that are O-GlcNAcylated in a cell. For the same reason, it is also difficult to speculate on the particular causes of the axonal degeneration and cell death.

Proteomic analysis of mouse synaptosomes revealed an estimated 20% of synaptosome proteins to be O-GlcNAcylated, including several vesicular transport proteins, cytoskeletal proteins, and transcriptional regulators (Trinidad et al., 2012). Gene ontology of over 250 O-GlcNAcylated proteins found in mouse cortical tissue also supports the role of the modification in cytoskeletal organization, neurogenesis, and synaptic transmission (Alfaro et al., 2012). In adult rat sciatic

nerve, proteomic analysis identified over a hundred O-GlcNAcylated proteins, with an enrichment in focal adhesion and MAPK signaling pathways (Kim et al., 2016). Additionally,

OGT regulates axonal mitochondrial transport in response to changes in extracellular glucose

(Pekkurnaz et al., 2014). Globally, loss of OGT in sensory neurons could disrupt the metabolic homeostasis of the cells if pathways become less sensitive to changes in metabolic flux. In fact, metabolic perturbations associated with mitochondrial dysfunction have long been associated with mechanisms of axonal degeneration and decreased neuronal health (Court and Coleman,

2012). Altogether, it is likely that the observed phenotypes are the result of many altered processes occurring in the neurons. Additional proteomic studies with OGT knockout sensory neurons may yield further mechanistic insight into these changes, but no individual loss of O-

GlcNAcylation may prove to be essential to the degeneration. While studies have shown that

OGT is essential for the survival of mitotic cells in mice (Shafi et al., 2000; O'Donnell et al.,

2004; Levine and Walker, 2016), possibly due to the role of O-GlcNAcylation in the cell cycle and cytokinesis (Slawson et al., 2005), this is the first study to demonstrate the importance of the modification in the maintenance and ultimate survival of post-mitotic neurons. The severity of

the phenotype furthers our appreciation of O-GlcNAc signaling as crucial and ubiquitous in regulating cellular functions.

The effects of loss of OGT on thermal sensitivity were clearer than on mechanical sensitivity at both the 10-11 and 30-31 week ages. This is consistent with the fact that Nav1.8

expression is predominantly in small-diameter neurons responsible for nociception and thermal

sensation (Agarwal et al., 2004). However, Nav1.8-Cre expression is not strictly limited to C-

fiber neurons. Cre expression in a small population of A-fiber neurons, as well as the loss of

OGT in mechanosensitive C fibers, could account for the mild hyposensitivity of Nav1.8-Ogt

knockout mice in the von Frey assay.

The density of epidermal innervation in the footpad appeared to be dynamic across

different ages, not only in Nav1.8-Ogt knockout mice but in control mice as well. Nerve fiber

density sharply increases following birth and then decreases between P10 and 3 weeks of age. It

is unclear whether these changes in epidermal innervation during postnatal development are

accompanied by changes in DRG cell body number or axonal branching. It is possible that

postnatal pruning occurs at the sensory nerve endings, as is observed in other neuronal systems

such as the cortex, cerebellum, and at the neuromuscular junction (Low and Cheng, 2006).

Because the skin is rapidly expanding as the mice grow during this time, it is also possible that a

constant set of fibers becomes distributed over a greater area of skin, accounting for the decrease

in nerve fiber density.

In Nav1.8-Ogt knockout mice, both loss of epidermal innervation and loss of cell bodies were observed early in development but also seemed to progress as the animals aged. In particular, significant decreases in nerve fiber endings were measured in adult Nav1.8-Ogt knockout mice at a stage when control animal nerve fiber counts remained stable. Using the inducible brn3a-CreERT2 driver clarified that loss of OGT in the adult decreases the survival of adult neurons. In brn3a-Ogt knockout mice at 2 weeks post-tamoxifen treatment, there was already a decrease in nerve fiber density in the paw, although DRG cell body counts remained the same as controls. By 4 weeks post-treatment, epidermal innervation was significantly further decreased, whereas there was only a modest decrease in cell body counts. Together, this suggests that loss of OGT results in an axonal dieback phenotype first that then progresses to overall death of the neuron.

Interestingly, thermal sensitivity as measured by the thermal place preference assay was

only mildly affected in brn3a-Ogt knockout animals by 4 weeks post-treatment, despite

significant loss of epidermal innervation. This is similar to what was observed in the Nav1.8-Ogt

knockout mice, which did not exhibit very severe behavioral deficits at 10 weeks of age despite

severe loss of nerve fibers at that time point. It is likely that under normal circumstances, there are far more nerve fibers in the skin than are needed for proper sensation of the stimuli used in the tests, so that a partial reduction in nerve endings would not immediately translate to a functional deficit. Additionally, more nerve fibers may be present in the skin than can be visualized by immunostaining, which is dependent on the penetrance and specificity of the

PGP9.5 antibody in the tissue. Nonetheless, I observed in my study that a severe decrease in nerve fiber density in the skin is necessary before behavioral consequences can be detected.

In addition to hyposensitivity to thermal and mechanical stimuli, Nav1.8-Ogt knockout animals weighed less than controls. Because Nav1.8 is expressed in sensory neurons that innervate the viscera (Gautron et al., 2011), altered innervation of organs such as the pancreas

and gut could affect the overall metabolism of the animals and contribute to the weight and glucose tolerance phenotypes. Increased glucose tolerance signals a possible increase in insulin secretion and/or an increased sensitivity of target tissues to circulating insulin, such that glucose is cleared from the blood faster (Bowe et al., 2014). Previously, TRPV1-positive sensory neurons, which are likely to also express Nav1.8, have been linked to β-cell function and glucose

homeostasis in diabetic animal models, thus supporting this hypothesis (Razavi et al., 2006).

Impaired glucose tolerance is also associated with obesity and diabetes (Nathan et al., 2007), and therefore differences in glucose metabolism may be related to differences in weight here as well.

An additional possibility is that taste and olfactory sensation could be altered in these mice, which would affect feeding. Future experiments examining visceral nerve fibers, feeding intake,

and other metabolic parameters, such as plasma insulin levels, could elucidate the relationship of

sensory neuron OGT and weight gain.

Differences in axons were not only observed in vivo but also in cultured DRG neurons.

OGT knockout neurons exhibited less axonal outgrowth in vitro compared to control neurons.

This demonstrates that loss of OGT has cell autonomous effects that alter growth and morphology, which may affect the overall health of the neuron. DRG neurons that are dissected

and dissociated undergo an axotomy in the process, and so deficits in outgrowth may also

represent deficits in the regenerative capacities of these neurons. The decrease in axonal

outgrowth observed here is in contrast to previous studies in which decreased O-GlcNAcylation

increased axonal length in cultures of embryonic mouse cortical neurons (Rexach et al., 2012)

and increased axonal branching in embryonic chicken forebrain neurons (Francisco et al., 2009).

Differences in the neuronal subtypes and developmental stages, which can affect baseline

morphology and therefore the analytical methods, may explain the differences. The previous

studies also used in vitro methods of decreasing O-GlcNAcylation, such as transfected Cre

expression in dissociated Ogt-floxed neurons and overexpression of OGA. These methods may

not have achieved the same extent of loss of the modification as knockout of OGT in vivo.

I hypothesized that the peripheral nervous system might be particularly sensitive to

metabolic perturbations and that the absence of OGT would therefore have a great effect. A

recent study deleted OGT in Schwann cells in mice and found that the mice developed

neuropathy characterized by progressive myelin loss, axonal loss, and motor and sensory nerve

dysfunction (Kim et al., 2016). These findings demonstrate the importance of OGT in

maintaining normal myelin and healthy axons and provide a counterpart to my study in sensory

neurons. Together, the studies show that O-GlcNAcylation has crucial roles in the health and

proper functioning of the peripheral nervous system.

In summary, I find that OGT is essential for sensory neuron survival and target

innervation. It is therefore possible that altered O-GlcNAc signaling in neurons contributes to the development of neuropathy in different disease states. Given the role of OGT in metabolic sensing, this is particularly relevant to diabetic neuropathy. This common complication of diabetes often occurs in patients with uncontrolled high blood glucose levels and primarily manifests as an axonopathy, although a few studies have demonstrated the presence of neuronal cell death in rodent models of diabetic neuropathy as well (Vincent et al., 2002; Ziegler, 2006;

Callaghan et al., 2012). It is conceivable then that under diabetic conditions, O-GlcNAc signaling becomes misregulated, disrupts cellular processes, and eventually causes axonal degeneration. Additionally, the effects of OGT knockout on neuronal survival may apply to central nervous system neurons as well, and long-term studies that alter O-GlcNAc signaling in

the brain may encounter consequences such as synapse loss, axonal degeneration, and neuron

loss.

2.4 Materials and methods

Mice

Ogt-floxed mice on a C57BL/6 background (Shafi et al., 2000) were kindly provided by

Dr. Xiaoyong Yang (Yale University). Nav1.8-Cre (Agarwal et al., 2004) and brn3a-CreERT2

(O’Donovan et al., 2014) mice on a C57BL/6 background were kindly provided by Dr. Clifford

Woolf (Boston Children’s Hospital). Only male mice of the appropriate genotypes were used in the studies. All animals were housed on a 12 h light/dark cycle with access to standard chow and water ad libitum. All procedures were approved by the Boston Children’s Hospital Institutional

Animal Care and Use Committee.

OGT knockout

To selectively delete OGT in sensory neurons during early development or during adulthood, female Ogt-floxed mice (OgtloxP/loxP) were crossed to male Nav1.8-Cre mice (Nav1.8-

Cre-/+) or brn3a-CreERT2 mice (brn3a-CreERT2-/+). Nav1.8-Cre is expressed in most of the small-diameter sensory neurons in the DRGs and trigeminal ganglia, including all nociceptors

and thermoreceptive neurons, as well as a small proportion of large-diameter sensory neurons,

including proprioceptive and mechanoreceptive neurons (Agarwal et al., 2004). Brn3a-CreERT2 is expressed in all DRG neurons (O’Donovan et al., 2014). Because Ogt lies on the X chromosome, animals hemizygous for the Cre transgene and floxed Ogt allele were generated in one cross (Shafi et al., 2000). Male Nav1.8-Cre Ogt-floxed mice (Nav1.8-Cre-/+ OgtloxP/Y) were

identified as OGT knockout mice by PCR genotyping from tail samples using the following

primers: 5′-tgc acg ttc acc ggc atc aac g-3′; 5′-gat gca acg agt gat gag gtt c-3′; 5′-cat ctc tcc agc

ccc aca aac tg-3′; and 5′-gac gaa gca gga ggg gag agc ac-3’. OGT knockout was induced in 7-

week-old male brn3a-CreERT2 Ogt-floxed mice (brn3a-CreERT2-/+ OgtloxP/Y) by intraperitoneal

(i.p.) injection of 1 mg of tamoxifen (Sigma-Aldrich) for 5 consecutive days. Tamoxifen was

freshly prepared prior to each experiment by suspension in sunflower oil with 5% ethanol at a

concentration of 10 mg/ml.

Validation of the loss of OGT was conducted by staining DRG tissue with an antibody

against the O-GlcNAc moiety. Animals were killed, and L3-L5 lumbar DRGs were dissected,

fixed in 4% paraformaldehyde (PFA) for 1 h at room temperature, and cryoprotected in 30%

sucrose overnight at 4°C. DRGs were embedded in Tissue-Tek OCT compound (Sakura Finetek)

and frozen, and 20 µm sections were prepared. Antigen retrieval was performed on the sections

by submerging the sections in a 10 mM sodium citrate and 0.05% Tween 20 buffer adjusted to

pH 6.0 and then heating at 90°C for 10 min. After cooling, sections were blocked in 1% bovine

serum albumin (BSA) and 0.1% Triton X-100 in PBS for 1 h at room temperature and incubated

in rabbit anti-peripherin at 1:500 (Millipore) and mouse anti-O-GlcNAc at 1:200 (clone RL2,

Abcam) overnight at 4°C. Sections were then incubated in anti-rabbit Alexa Fluor 488 at 1:500

(Invitrogen) and anti-mouse Alexa Fluor 568 at 1:500 (Invitrogen) for 1 h at room temperature.

Images were acquired on a Zeiss LSM 700 confocal microscope using a 25X oil objective. The

numbers of O-GlcNAc-positive and peripherin-positive neurons were manually counted from 5 imaging fields per animal, and the overall proportion was calculated for each animal to represent knockout efficiency. Peripherin is present in small-diameter DRG neurons and has significant overlap with Nav1.8 expression; therefore, most of the peripherin-positive neurons are likely also to be positive for Nav1.8-Cre in this system. Significance was calculated by unpaired Student’s t- test.

Behavioral testing

All behavioral assays were conducted blinded as to genotype or tamoxifen treatment.

Global phenotyping was assessed through a phenotypic test battery following the modified

SHIRPA protocol (SmithKline Beecham Pharmaceuticals; Harwell, MRC Mouse Genome

Centre and Mammalian Genetics Unit; Imperial College School of Medicine at St. Mary’s;

Royal London Hospital, Bartholomew’s and the Royal London School of Medicine; Phenotype

Assessment). In this protocol, animals were placed in a viewing compartment and observed for the presence or absence of the following phenotypes: tremor, palpebral closure, whiskers, lacrimation, and defecation. General activity and coat appearance were also assessed. Animals were then transferred to an arena and assessed for the following phenotypes: transfer arousal, gait, tail elevation, startle response, touch escape, and visual placing. Animals were held by the

56 tail above the arena to assess the following phenotypes: positional passivity, paw skin color, trunk curl, and limb grasping. Additional handling of the animals was conducted to assess righting reflex, pinnal reflex, and corneal reflex. The presence or absence of biting or vocalization during the whole protocol was noted. The overall SHIRPA score was determined for each animal by scoring each observation using an arbitrary value (0-3) and summing the values.

Each animal was also weighed. Significance was calculated by unpaired Student’s t-test.

Grip strength was measured using a digital grip strength meter (TSE Systems), in which each animal was positioned with its forelimbs gripping a bar and subsequently pulled away from the bar to record the maximal strength exerted. Three consecutive trials were conducted in one session and averaged for each animal. Significance was calculated by unpaired Student’s t-test.

Locomotor activity was measured by placing the animal in a tiled arena and counting the total number of squares the animal crossed within 30 s. A square was counted if all 4 paws had entered the square while the animal was moving. Significance was calculated by unpaired

Student’s t-test.

Thermosensitivity was assessed using the thermal place preference and hot plate assays.

In the thermal place preference assay, animals were placed in an arena spanning two adjacent metallic plates (Bioseb) that reciprocally oscillated between 30°C and 49°C and video-recorded from above (Noldus EthoVision) while they were free to move over a span of 11 min. Thermal preference was assessed for each animal by calculating and graphing the duration of time spent on each oscillating plate over every 30 s interval of the trial. The resulting graphs were then quantified by measuring the total area “under” the curve for one of the plates starting at 2 min, in which the baseline was 15 s and peaks both above and below the baseline were included.

Significance was calculated by unpaired Student’s t-test.

In the hot plate assay, animals were acclimated to an arena on top of a metallic plate

(Bioseb) set at 30°C for 10 min the day before the experiment. Animals were placed within the

arena on the plate set at 52°C, 49°C, or 55°C, and the amount of time for a hindpaw response, such as shaking or licking of the paw, was measured. The assay was conducted over 3 consecutive days, with 1 temperature tested per day. Significance was calculated by unpaired

Student’s t-test at each temperature.

Mechanosensitivity was assessed using the von Frey assay. Animals were acclimated to individual arenas on top of a raised gridded wire surface for 2 h each day for 2 consecutive days prior to testing. Mechanical stimuli were applied to the plantar surface of the hindpaw using a graded series of von Frey fibers that produced bending forces of 0.04, 0.07, 0.16, 0.4, 0.6, 1.0,

1.4, and 2.0 g (Stoelting Touch Test). Each fiber was tested 10 consecutive times in increasing order from the lowest force, and the number of paw withdrawals was recorded. Two trials were conducted over 2 consecutive days and averaged for each animal. Significance was calculated by two-way repeated measures ANOVA with Bonferroni correction.

Glucose tolerance test

Animals were fasted for 6 h during the morning and midday. Initial blood glucose levels were determined by nicking the tail vein with a razor blade, collecting a droplet of blood, and measuring the blood using a glucose meter (ACCU-CHEK Aviva Plus). Animals were then injected with a 20% dextrose solution (Hospira) into the intraperitoneal cavity at 2 g glucose/kg body mass. Blood glucose was measured at 15, 30, 60, 90, and 120 min after glucose injection.

Animals that did not display a blood glucose level of over 200 mg/dl at 15 min after the glucose challenge were considered outliers, likely due to a technical error in the i.p. injection, and were not included in the data. There were 2 control outliers at 8 weeks and 1 Nav1.8-Ogt knockout

58 outlier at 30 weeks. Significance was calculated by two-way repeated measures ANOVA with

Bonferroni correction.

Epidermal innervation

Animals were killed, and footpad tissue from hindpaws or back skin were dissected, fixed in Zamboni’s fixative (Stefanini et al., 1967) overnight at 4°C, and then cryoprotected in 30% sucrose overnight at 4°C. Samples were embedded in Tissue-Tek OCT compound (Sakura

Finetek), frozen, and 30 µm sections were prepared. Sections were blocked in 1% BSA and 0.1%

Triton X-100 in PBS for 1 h at room temperature and incubated in rabbit anti-PGP9.5 at 1:1000

(UltraClone) or 1:500 (EnCor Biotechnology) overnight at 4°C. Sections were then incubated in anti-rabbit Alexa Fluor 488 at 1:500 (Invitrogen) for 1 h at room temperature and counterstained with Hoechst for 15 min at room temperature. Images were acquired on a Zeiss LSM 700 confocal microscope using a 25X oil objective. The number of PGP9.5-labeled fibers in the epidermis region devoid of dermal papillae was manually counted over a distance of 250 µm.

Fiber counts from 5 sections were averaged for each animal. Counting was performed in a blinded manner, and significance was calculated by unpaired Student’s t-test and one-way

ANOVA with post hoc Fisher’s least significant difference (LSD) test.

DRG cell body count

Animals were killed, and L5 DRGs were dissected and processed as described above.

Fourteen micrometer serial sections were prepared and incubated in rabbit anti-peripherin at

1:500 (Millipore) overnight at 4°C following antigen retrieval and block. Sections were then incubated in anti-rabbit Alexa Fluor 488 at 1:500 (Invitrogen) for 1 h at room temperature.

Images were acquired on a Zeiss LSM 700 confocal microscope using a 10X air objective. The

total numbers of peripherin-positive cell bodies, as determined through fluorescent labeling, and

peripherin-negative cell bodies, as determined through background staining, were manually

counted for every third section, or 42 µm. Cell body counts from the 2 L5 DRGs were averaged

for each animal. Counting was performed in a blinded manner, and significance was calculated

by unpaired Student’s t-test.

Axonal outgrowth assay

DRGs from all segments of the vertebral column were dissected from 4-week-old mice.

Ganglia were initially placed in HBSS with 1% penicillin/streptomycin (pen/strep), and then

incubated with 5 mg/ml collagenase (Sigma-Aldrich) and 1 mg/ml dispase (Roche) for 70 min at

37°C. Cells were then incubated with 0.25% trypsin for 5 min and washed with 0.25% trypsin

inhibitor and high glucose DMEM containing 10% fetal bovine serum (FBS) and 1% pen/strep.

Cells were triturated in the presence of DNase I inhibitor using a series of flame-polished Pasteur

pipettes and centrifuged through 10% BSA (Sigma-Aldrich) in PBS with 1% pen/strep. The

pellet was washed again in DMEM before being resuspended in Neurobasal-A medium (Life

Technologies) containing 2% B-27 supplement (Life Technologies), 1% pen/strep, 1 mM L-

glutamine, 50 ng/ml NGF (Life Technologies), and 2 ng/ml GDNF (Sigma-Aldrich). Cells were

plated onto 12 mm German glass coverslips (Bellco Glass) coated with 100 µg/ml poly-D-lysine

(Sigma-Aldrich) and 10 µg/ml laminin (Life Technologies) and grown in 24-well tissue culture

plates (Corning) at 37°C and 5% CO2.

Twelve hours after plating, cells were fixed in 4% PFA for 10 min and incubated in rabbit anti-β-tubulin III at 1:2000 (clone TUJ1 1-15-79, Covance) and mouse anti-O-GlcNAc at

1:200 (clone RL2, Abcam) in 1% BSA, 15% goat serum, and 0.1% Triton X-100 in PBS for 1 h at room temperature. Cells were then incubated in anti-rabbit Alexa Fluor 488 at 1:500

(Invitrogen) and anti-mouse Alexa Fluor 568 at 1:500 (Invitrogen) for 1 h at room temperature.

Images were acquired on a Zeiss LSM 700 confocal microscope using a 25X oil objective.

Axonal outgrowth of individual neurons was quantified using Sholl analysis in ImageJ, in which the total number of neurite crossings was counted at increasing intervals of 50 µm from the soma center as designated by concentric circles, starting at 25 µm from the center. To exclude dying cells, only neurons with visible axons were included. Cell body diameters and longest neurite lengths were measured in ImageJ. Significance was calculated by unpaired Student’s t-test and two-way repeated measures ANOVA with Bonferroni correction.

Chapter 3

Additional observations:

Effects of altered O-GlcNAcylation in sensory neurons

Contributions: All experiments were designed by Cathy Su and Thomas Schwarz. Experiments and data analysis were performed by Cathy Su. 3.1 Introduction

My findings in Chapter 2 demonstrate the essential role of O-GlcNAcylation in sensory neurons in vivo, particularly in maintaining innervation in the periphery and long-term neuronal

survival. In this Chapter, I discuss additional efforts toward understanding the role of O-

GlcNAcylation in these neurons, including investigations into specific pathways and further

examination of OGT knockout neurons in culture.

While my work has sought to address the global significance of OGT in sensory neurons,

I had two additional areas of interest at the outset of this project. First, our lab originally became

interested in OGT through our studies in mitochondrial motility. Mitochondrial transport is

mediated by motor/adaptor complexes that consist of molecular motors, kinesin heavy chain

(KHC) and dynein/dynactin, coupled to adaptor proteins, Miro (RhoT1/2) and Milton (TRAK1/2

or OIP106/98) (Schwarz, 2013). OGT was previously identified to be a binding partner of Milton

(Iyer et al., 2003), and work in our lab has shown that increased O-GlcNAcylation of Milton

under high extracellular glucose concentrations decreases axonal mitochondrial motility in

hippocampal neurons (Pekkurnaz et al., 2014). This mechanism could serve as a way for

neuronal mitochondria to respond to spatial differences or temporal changes in glucose

availability along an axon, allowing for enrichment of mitochondria in areas of high glucose to

facilitate efficient ATP production. Sensory neurons have especially long axons that extend from

the spinal cord to the distal extremities, so mitochondrial transport is crucial for proper

distribution of organelles throughout the cell. Given the potential for greater fluctuations in

glucose concentrations in the periphery, I aimed to determine whether O-GlcNAc regulation of

mitochondrial motility also occurs in the axons of sensory neurons.

Second, we recognized that the presence of a mechanism for glucose-mediated

63 mitochondrial motility has implications in the development of diabetic neuropathy, a common complication of diabetes that often manifests in a length-dependent manner. In our hypothesis, hyperglycemia could affect the transport and distribution of mitochondria through the activity of

OGT and thus contribute to the decline in sensory neuron health, particularly in the distal axons.

Additionally, my findings that altered O-GlcNAc dynamics could result in axonal degeneration and neuronal cell death further supports a possible role of OGT in diabetic neuropathy.

Therefore, I aimed to examine the effects of changes in glucose concentrations on sensory neuron processes, including mitochondrial motility and axonal outgrowth, as a first step to investigate this possibility.

Finally, after characterizing the effects of loss of OGT in vivo, I aimed to gain further mechanistic insight into the observed axonal degeneration and cell death phenotypes. To do so, I investigated mitochondrial distribution and survival of OGT knockout neurons in culture as well as the expression of candidate proteins in OGT knockout DRG tissue. While some of the experimental approaches presented in this chapter had limitations and were preliminary in nature, I discuss the logic of the strategies, alternative approaches, and future directions to address these aims.

3.2 O-GlcNAc regulation of mitochondrial motility in DRG neurons

Increased O-GlcNAcylation in cultured neurons can be induced by either overexpression of OGT, inhibition of OGA, or elevated extracellular glucose levels; all three manipulations result in decreased mitochondrial transport in hippocampal axons (Pekkurnaz et al., 2014).

To test whether this mechanism is present in sensory neurons as well, I examined the effects of pharmacological inhibition of OGA on mitochondrial movement in the axons of cultured DRG neurons. Adult DRG neurons were infected with lentivirus expressing Mito-

DsRed to label the mitochondria and treated with the OGA inhibitor, PUGNAc, to increase the levels of O-GlcNAc. Initially, treatment with 100 µM PUGNAc for 6 h prior to live cell imaging did not appear to have an effect on mitochondrial motility. This condition was previously shown to have an effect in hippocampal neurons, so I hypothesized that different neuronal subtypes may require different exposures to PUGNAc to yield an effect. Therefore, I tested varying PUGNAc treatments in the DRG neurons and observed that neurons treated with 100 µM PUGNAc for 24 h had an almost 20% increase in O-GlcNAc immunofluorescence intensity as compared to those only treated for 6 h.

Ultimately, DRG neurons were treated with 200 µM PUGNAc for 32 h prior to live cell imaging. This treatment resulted in nearly double the O-GlcNAc intensity in the neurons as compared to those treated with the vehicle, DMSO (Figure 3.1). Mitochondrial transport and density were analyzed from time-lapse movies taken from the mid-axon region of DIV5 DRG neurons. Treatment with PUGNAc resulted in a decrease in mitochondrial transport in both the anterograde (away from the cell body) and retrograde (towards the cell body) directions (Figure

3.2, A-B). Likewise, there was a slight decrease in the average travel length of the mitochondria per axon (Figure 3.2, C). However, there were no significant differences in mitochondrial speed or density after PUGNAc treatment (Figure 3.2, D-E).

To determine whether the decrease in axonal transport was specific to mitochondria, I assessed the motility of another cargo, late endosomes/lysosomes, by expressing LAMP1-RFP in the DRG neurons. Due to the large number of lysosomes, motility was quantified by analyzing flux rather than by tracing individual organelles. There was no difference in either anterograde or retrograde flux of LAMP1-positive vesicles after treatment with 200 µM PUGNAc for 32 h

(Figure 3.3). Because different cargoes are transported via different motor/adaptor protein

Figure 3.1. PUGNAc treatment results in increased O-GlcNAc levels in DRG neurons

A Immunostaining of O-GlcNAc (magenta) and β-tubulin III (green) in dissociated DRG neurons treated with DMSO as a vehicle control or 200 μM PUGNAc for 32 h. Scale bar, 25 µm. B Quantification of O-

GlcNAc levels as defined by integrated density normalized to the area of axons. n = 6 imaging fields per condition, unpaired t-test, ***p < 0.001. All values shown as mean ± SEM.

Figure 3.2. Increased O-GlcNAcylation decreases axonal mitochondrial motility in DRG neurons

A Representative kymographs of mitochondrial motility in DIV5 DRG axons treated with either DMSO or 200 μM PUGNAc for 32 h prior to imaging. Kymographs represent position (x axis) over time (y axis) such that stationary organelles appear as vertical lines and moving organelles appear as diagonal lines.

The first frame of each time-lapse movie appears above the generated kymograph. Scale bars, 10 µm and

100 s. B Percent of time that mitochondria moved in either the anterograde or retrograde directions per axon, as quantified from the generated kymographs. n = 27 axons from 3 experiments, unpaired t-test, *p

< 0.05, **p < 0.01. C Average total distance traveled by mitochondria in either direction per axon. n = 27 axons from 3 experiments, unpaired t-test. D Speed of each mitochondrion moving in either the anterograde or retrograde directions, calculated as the average of all non-zero instantaneous speeds in a trace. n = 179-230 anterograde and 112-196 retrograde mitochondria from 3 experiments, Mann-Whitney

U test. E Density of mitochondria, calculated as the number of mitochondria per 10 µm of axon in the first frame of each time-lapse movie. n = 27 axons from 3 experiments, unpaired t-test.

All values shown as mean ± SEM.

Figure 3.2 (Continued)

Figure 3.3. Increased O-GlcNAcylation does not affect axonal transport of lysosomes in DRG neurons

A Representative kymographs of lysosomal motility in DIV5 DRG axons treated with either DMSO or

200 μM PUGNAc for 32 h prior to imaging. The first frame of each time-lapse movie appears above the generated kymograph. Scale bars, 10 µm and 100 s. B Flux of LAMP1-positive vesicles in the anterograde and retrograde directions per axon, as quantified from the generated kymographs. n = 27 axons from 3 experiments, unpaired t-test. All values shown as mean ± SEM.

complexes, this finding suggests that mitochondrial arrest due to increased O-GlcNAcylation in sensory neurons is specific and likely mediated through the modification of the mitochondrial adaptor protein Milton as well.

Overall, these results demonstrate that O-GlcNAc regulation of mitochondrial motility

also occurs in primary sensory neurons and can therefore serve to mediate changes in motility in

response to changes in glucose availability. In an axon that extends up to a meter in length in

humans, regulation of mitochondrial transport is especially important. Different regions of a

neuron may be exposed to different nutrient levels, due to heterogenous distribution of glucose

transporters or proximity to nodes of Ranvier. Pekkurnaz et al. (2014) demonstrated that axonal

mitochondria do in fact accumulate in areas of higher glucose by using microfluidic chambers to

generate spatial differences in extracellular glucose concentrations. Therefore, O-GlcNAc

signaling in sensory neurons could play an important role in distributing mitochondria in areas of

high nutrient availability to maximize energy production within the cell. Regulation of

mitochondrial transport also affects other aspects of mitochondrial dynamics, such as fusion and

fission, which in turn can affect the balance of nutrient supply and energy demand within the

neuron (Liesa and Shirihai, 2013).

In addition to spatial differences in glucose availability over the length of a cell, sensory

neurons may also experience temporal changes in their exposure to glucose. This can occur

between fasting and feeding, during periods of high activity when both glucose uptake and ATP

consumption are increased (Ferreira et al., 2011), or under disease states such as diabetes. Given

the length-dependent nature of diabetic neuropathy, a defect in axonal transport, and in particular mitochondrial transport via altered O-GlcNAc signaling, is possible in the pathogenesis of the complication.

Mitochondrial motility in DRG neurons under a high glucose condition

Given that inhibition of OGA results in a decrease in mitochondrial motility in sensory

neurons, I next treated the neurons with high glucose to determine whether it would have a

similar effect. Previously, a 2 h high glucose treatment prior to live cell imaging was sufficient to

cause a decrease in both anterograde and retrograde mitochondrial transport in hippocampal

neurons (Pekkurnaz et al., 2014). Therefore, I followed a similar protocol with the sensory

neuron cultures.

DRG neurons were plated in the standard Neurobasal-A medium containing 25 mM glucose. This concentration of glucose, however, is much higher than physiological levels of glucose and likely saturates the glucose transporters of the neurons, such that a further increase in concentration would not be detected. Therefore at DIV3, the cultures were switched to a 5 mM glucose medium to precondition the neurons to a lower glucose concentration before a high glucose challenge. At DIV5, neurons were either maintained in 5 mM glucose or shifted to 25 mM glucose for 2 h prior to imaging. However, this elevated glucose condition did not result in any differences in the levels of mitochondrial motility in the DRG neurons (Figure 3.4).

The lack of an effect of high glucose in this study is in contrast with the findings in hippocampal neurons, but does not preclude the possibility of glucose-mediated mitochondrial transport in sensory neurons. For one, different neuronal subtypes may contain different concentrations or localization of glucose transporters and other metabolic enzymes. Therefore, the same glucose treatment protocol may elicit varied effects if the changes in glucose concentrations are sensed differently within the cells. In addition, the hippocampal neurons used in the previous studies were dissected from embryos whereas the DRG neurons here are from adult animals. Thus, differences in developmental stage could also explain the conflicting results.

Figure 3.4. A high extracellular glucose condition does not affect axonal mitochondrial motility in

DRG neurons

A Representative kymographs of mitochondrial motility in DIV5 DRG axons treated with either 5 mM glucose or 25 mM glucose for 2 h prior to imaging. The first frame of each time-lapse movie appears above the generated kymograph. Scale bars, 10 µm and 100 s. B Percent of time that mitochondria moved in either the anterograde or retrograde directions per axon, as quantified from the generated kymographs.

C Average total distance traveled by mitochondria in either direction per axon. n = 25-26 axons from 3 experiments, unpaired t-test. All values shown as mean ± SEM.

Future strategies to understand these differences could focus on evaluating the effects of

high glucose on pathways upstream of mitochondrial transport. For instance, does an increase in

extracellular glucose translate to an increase in intracellular glucose in the sensory neurons?

Does treatment with a given high glucose concentration result in a significant change in the

levels of O-GlcNAc? Using a fluorescent glucose sensor and immunostaining against O-

GlcNAc could serve to validate high glucose protocols in adult DRG neurons. With a different glucose concentration or treatment length, the effects of high glucose on mitochondrial motility could be evident in sensory neurons as well.

3.3 Additional observations of OGT knockout neurons in culture

As described in Chapter 2, in addition to characterizing the effects of loss of OGT in

sensory neurons in vivo, I also examined the morphology of dissociated OGT knockout neurons and found that neurons lacking OGT have decreased axonal outgrowth. Loss of OGT, therefore, has cell autonomous effects that could contribute to the decreased health of the neurons. In order to gain further insight into the consequences of OGT deletion on a cellular level, I investigated additional aspects of the cultured knockout neurons.

Axonal outgrowth under varying glucose conditions

Because I observed a clear phenotype using the axonal outgrowth assay, I decided to further assess the growth and morphology of the OGT knockout neurons under varying glucose conditions. I hypothesized that the neurons may exhibit differences in how they adapt to changes in glucose concentration due to the absence of O-GlcNAc signaling. For instance, a shift to elevated glucose levels may increase axonal outgrowth in control neurons but have no effect in

OGT knockout neurons. Such differences would support the role of O-GlcNAcylation in

Figure 3.5. Decreased axonal outgrowth under varying glucose conditions in cultured OGT knockout neurons

A-D Quantification of neurite crossings of DIV1 control and OGT knockout neurons at increasing distances from the soma center using Sholl analysis. Neurons were either cultured in A 25 mM glucose media for 12 h (p = 0.071), B 25 mM glucose media for 6 h followed by 5 mM glucose media for 6 h (p <

0.001), C 5 mM glucose media for 12 h (p = 0.117), or D 5mM glucose media for 6 h followed by 25 mM glucose media for 6 h (p < 0.001). E-F Same quantification as in (A-D), organized by genotype and analyzed across glucose conditions (p = 0.242 for control neurons, p = 0.27 for OGT knockout neurons). n = 35-36 cells from 1 animal per genotype, two-way repeated measures ANOVA with Bonferroni correction, *p < 0.05, ***p < 0.001, ****p < 0.0001. All values shown as mean ± SEM.

Figure 3.5 (Continued)

mediating neuronal responses to nutrient flux and establish axonal outgrowth as an in vitro

parameter for measuring the effects of glucose availability in the absence of OGT.

DRG neurons were dissected from 4-week-old control and Nav1.8-Ogt knockout mice and cultured for 12 h at 4 different glucose concentrations: 5 mM for 12 h, 5 mM for 6 h then 25 mM for 6 h, 25 mM for 12 h, or 25 mM for 6 h then 5 mM for 6 h. The OGT knockout neurons consistently exhibited decreased axonal outgrowth for each condition, and any additional differences between the two genotypes across the conditions were unable to be detected (Figure

3.5). Therefore, due to the robust baseline knockout phenotype, I did not pursue axonal outgrowth as a parameter affected by glucose availability.

Although shifts in glucose levels did not result in detectable differences in axonal outgrowth between OGT knockout and control neurons, there could be other processes affected by glucose flux. In particular, it would be interesting to discover parameters in which OGT knockout neurons exhibit abnormalities only when challenged with high or low glucose.

Identifying such processes would highlight the role of OGT in mediating cellular responses to glucose availability and lay the foundation for future work in studying O-GlcNAcylation in diabetic neuropathy.

Mitochondrial distribution

Because OGT regulates mitochondrial transport in axons, I hypothesized that mitochondrial distribution might be affected in OGT knockout neurons. For instance, if mitochondria are less motile and transported to the periphery at a lower rate, then neurons would have fewer mitochondria towards the axon tips. Therefore, I immunostained for TOM20, a mitochondrial outer membrane protein, to visualize the mitochondria in DIV1 DRG neurons.

Mitochondria were distributed evenly within the cell bodies and along the axons of neurons from

Figure 3.6. Loss of OGT has no apparent effects on mitochondrial distribution in cultured DRG

neurons

Immunostaining of β-tubulin III (green) to visualize neurons and TOM20 (red) to visualize mitochondria in DIV1 DRG neurons from 4-week-old mice. Scale bar, 50 µm.

both the control and Nav1.8-Ogt knockout DRG cultures (Figure 3.6). Due to this lack of

apparent differences, mitochondrial distribution was not a strong candidate for further

investigation. However, the possibility remains that distribution could be altered if mitochondrial

motility is misregulated under conditions of high glucose. To address this hypothesis, future

studies could focus on determining first whether loss of OGT alters mitochondrial motility, both

under baseline conditions and under a high glucose challenge.

Survival in culture

Loss of OGT in sensory neurons in vivo results in decreased neuronal survival over time.

To determine whether this occurs in dissociated OGT knockout neurons as well, DRG neurons

from 4-week-old control and Nav1.8-Ogt knockout mice were cultured for up to 10 days in vitro.

At DIV1, around 40-45% of the neurons from Nav1.8-Ogt knockout mice were positive for O-

GlcNAc immunoreactivity, as compared to 97% of neurons from control mice. If there is indeed a loss of OGT knockout neurons over time, then I would expect that the proportion of O-

GlcNAc-positive neurons would increase in older Nav1.8-Ogt knockout cultures. However at

DIV10, the percentage of O-GlcNAc-positive neurons from Nav1.8-Ogt knockout mice was still around 40-50%. This indicates that there was not a substantial specific loss of OGT knockout neurons in the culture between DIV1 and DIV10. Although these observations were preliminary, they suggest that neurons lacking OGT do not necessarily have immediate decreased survivability in culture.

The decrease in DRG cell bodies in vivo was evident within 10 days after birth but continued to worsen over the course of months. This suggests that the onset of neuronal cell death is heterogeneous and can be delayed for subsets of sensory neurons. The neurons observed in the DIV10 culture may represent a robust population of DRG neurons that have already

survived 4 weeks in vivo and through a dissociation process, and thus are less likely to die in

culture after a short period. Studies that extend beyond DIV10 may yield more significant

differences in the ability of OGT knockout neurons to survive in culture.

In general, neuronal survival and axonal growth in culture were observed to depend

heavily on the plating density of the neurons. Cultures that are plated at a higher density exhibit

healthier axonal morphology and a greater proportion of neurons that survive over time. Because

it was difficult to ensure identical plating densities between two sets of cultures, I was unable to

directly compare the axonal density of older control and Nav1.8-Ogt knockout DRG cultures.

However, it is possible that while OGT knockout neurons are capable of surviving in culture,

they may have smaller axonal arbors as compared to control neurons, consistent with their initial

decreased outgrowth seen at DIV1. Future work examining axonal density through an assay that

accurately normalizes for cell body density would help to elucidate whether loss of OGT in

neurons affects long-term axonal growth and morphology in vitro.

3.4 Investigating candidate protein expression in Nav1.8-Ogt knockout DRGs

The question remains as to why loss of OGT in neurons results in severe consequences in

vivo. While it is likely that multiple pathways are misregulated with the loss of O-GlcNAc signaling, it is possible that certain pathways important for neuronal health are more greatly affected. As a preliminary step to identifying these pathways, I tested the expression of candidate proteins in DRG sections from 4-week-old control and Nav1.8-Ogt knockout mice. O-

GlcNAcylation has been shown to regulate proteins in various ways, such as by affecting their localization or stability. Moreover, OGT can regulate protein expression through the modification of transcription factors and transcriptional and translational machinery (Ozcan et al., 2010; Lagerlof and Hart, 2014). I hypothesized then that in OGT-regulated pathways, there

could be observable differences in the expression levels or localization of proteins in the

knockout neurons as opposed to O-GlcNAc-positive neurons from either the same tissue or control tissue.

Proteins that are likely to be affected by loss of OGT are those that are direct substrates of the modification. However, given the wide network of O-GlcNAc-regulated processes, proteins beyond the O-GlcNAcylated proteome could also be indirectly affected. Due to a limited availability of antibodies, candidate proteins were thus selected based on their known roles in DRG neurons. DRG sections were immunostained for O-GlcNAc and one of 6 proteins: tropomyosin receptor kinase A (TrkA), activating transcription factor 3 (ATF3), extracellular signal-regulated kinases 1 and 2 (ERK1/2), phosphorylated ERK1/2 (phospho-ERK1/2), cyclic

AMP response element binding protein (CREB), and phosphorylated CREB (phospho-CREB).

Overall, the images revealed similar or equally heterogeneous levels of the proteins in wild-type

and OGT knockout neurons; no apparent differences in expression or localization were observed.

Below, I briefly describe the rationale behind examining each protein and the preliminary

findings.

TrkA is the receptor for nerve growth factor (NGF), and together, NGF/TrkA signaling is

important for sensory neuron survival and peripheral innervation (Patel et al., 2000), two

processes that are disrupted in OGT knockout neurons. Moreover, the expression of TrkA

determines sensory neuron differentiation during development, and the neurons that continue to

express the receptor become the peptidergic class of nociceptors (Woolf and Ma, 2007).

Therefore, in the DRG sections, the TrkA-positive small-diameter nociceptors are likely to also be O-GlcNAc-negative, due to Nav1.8 expression in the same neurons. I indeed observed that

OGT knockout neurons were more likely to have high TrkA immunostaining as compared to

wild-type neurons, but there were no additional unexpected differences (Figure 3.7, A).

ATF3 is upregulated in response to injury and stress and serves as a standard marker for neurons axotomized by peripheral nerve injury (Hunt et al., 2012). ATF3 is also upregulated in some models of peripheral neuropathy, including chemotherapy-induced peripheral neuropathy and diabetic neuropathy (Wright et al., 2004; Hunt et al., 2012). Since loss of OGT can potentially induce conditions of stress leading up to neuronal cell death, I examined ATF3 expression for upregulation in the knockout neurons. However, there was no clear expression of

ATF3 in either control or Nav1.8-Ogt knockout tissue, so ATF3 does not appear to be upregulated in neurons lacking OGT (Figure 3.7, B).

ERK1/2 are members of the mitogen-activated protein kinase (MAPK) family and have roles in mediating neuronal survival. Activation of ERK1/2 via phosphorylation has been observed in conditions of neuronal stress and injury and generally promotes cell survival

(Hetman and Gozdz, 2004); therefore, this could be a pathway involved in the death of OGT knockout neurons. However, immunostaining against ERK1/2 or phospho-ERK1/2 produced inconsistent results, and no definitive patterns of expression emerged (Figure 3.7, C-D).

CREB is a transcription factor that mediates several neuronal processes, including development and survival. Rexach et al. (2012) demonstrated that O-GlcNAcylation of CREB regulates neuronal growth and modulates long-term memory. Given the decreased axonal outgrowth in OGT knockout neurons, I investigated whether CREB expression is altered in neurons lacking OGT. CREB appeared to form more puncta in the nuclei of control DRG neurons, though this was not observed in the O-GlcNAc-positive DRG neurons from Nav1.8-Ogt knockout animals (Figure 3.7, E). Activated phospho-CREB had similar expression and nuclear localization in both control and OGT knockout neurons (Figure 3.7, F).

Figure 3.7. Loss of OGT has no apparent effects on immunofluorescence patterns of candidate proteins in DRGs

Immunostaining of lumbar DRGs from 4-week-old mice for O-GlcNAc (magenta) and A tropomyosin

receptor A (TrkA), B activating transcription factor 3 (ATF3), C extracellular signal-regulated kinases 1 and 2 (ERK1/2), D phosphorylated ERK1/2 (phospho-ERK1/2), E cyclic AMP response element binding

protein (CREB), and F phosphorylated CREB (phospho-CREB) (green). Scale bar, 100 µm.

Overall, this candidate-based approach was very limited in scope and served as a first- pass attempt to understand changes in OGT knockout sensory neurons. In addition to technical variability in immunofluorescence levels, the heterogeneity of sensory neurons also made it difficult to interpret the data, since variable fluorescence intensities could be due to different levels of the proteins in different sensory neuron subtypes. For these reasons, I did not continue with this approach or pursue any of the candidate proteins.

Ultimately, a future direction for this project is to investigate potential cellular changes on a more global scale using quantitative transcriptomic or proteomic approaches. Analyzing the changes across hundreds of transcripts or proteins between neurons expressing and lacking OGT could help to identify more specific pathways that are altered with the loss of OGT. These could serve as candidates for further investigation to elucidate the specific roles of OGT in neurons and how they mediate processes essential for growth, maintenance, and survival.

3.5 Materials and methods

Lentivirus preparation

The Mito-DsRed lentiviral construct was a gift from Dr. Orian Shirihai (Boston

University Medical Center). LAMP1-RFP-2A-GFP was generated by cloning LAMP1-RFP

(Addgene #1817) into pLenti-CMV-2A-GFP, which was a gift from Dr. Clifford Woolf (Boston

Children’s Hospital).

Lentiviral particles were produced by co-transfecting the expression vectors, envelope plasmid (encoding VSV-G), and packaging plasmids (encoding the gag, pol, rev, and tat genes) into HEK293T cells plated in 15 cm2 cell culture dishes (Corning). Media containing lentiviral particles were collected 48 and 72 h after transfection and filtered through a 0.45 μm filter. The particles were concentrated by spinning at 25,900 RPM at 4°C for 90 min (SW 32 Ti rotor,

Beckman Coulter). The resulting pellet was resuspended in 0.001% Pluronic F-68 (Sigma-

Aldrich) in PBS (concentrating 800-fold from the supernatant media), aliquoted, and stored at -

80°C.

For lentiviral transduction in DRG neurons, lentivirus was added to culture media 4 h

after plating, and media was changed 16-18 h later. Lentiviral expression was monitored by

examining expression of fluorescence in the neuronal cell bodies and axons.

Axonal transport assays

DRG neurons were dissected from 8-week-old C57BL/6J mice and cultured as described

in Chapter 2, with 10 µM cytosine arabinoside (Ara-C) (Sigma-Aldrich) added to the media beginning on DIV1. Neurons were maintained in culture by replacing half of the volume of old media with fresh media every 2 days. To assess the effects of increased O-GlcNAcylation on axonal transport, lentiviral-infected neurons were treated with 200 µM O-(2-acetamido-2-deoxy-

D-glucopyranosylidene)amino-N-phenylcarbamate (PUGNAc) (Tocris Bioscience) on DIV4 for

32 h prior to imaging. To assess the effects of high extracellular glucose levels, lentiviral- infected neurons were plated in standard media with 25 mM glucose and then transferred to 5 mM glucose media on DIV3, approximately 50 h before imaging. 5 mM culture media was made by combining 0 mM Neurobasal-A media (Life Technologies) with 25 mM Neurobasal-A media at the requisite proportion. Neurons were then shifted to either 25 mM media as a high glucose condition or maintained at 5 mM media for 2 h prior to imaging.

Live cell imaging of mitochondria or LAMP1-positive vesicles was conducted in DRG neurons at DIV5. Coverslips were transferred to cell culture dishes containing Hibernate A

(BrainBits) with 2% B-27 (Life Technologies) and 2 mM GlutaMAX (Life Technologies).

PUGNAc-treated neurons were transferred to imaging media that also contained 83 µM

PUGNAc. Glucose-treated neurons were transferred to imaging media that contained the same

glucose concentration as in the culture media. 5 mM glucose imaging media was made by

combining 0 mM Hibernate A media (BrainBits) with 25 mM Hibernate A media.

Time-lapse movies were acquired on a Zeiss LSM 700 confocal microscope using a 63X water objective. Cells were maintained at 37°C on a temperature-controlled stage. Portions of

axon 85-100 µm long that were at least 300 µm away from the cell body were selected for

imaging. Images were captured at 1 frame/5 s for mitochondria and 1 frame/2.5 s for LAMP1-

positive vesicles for 5 min.

The trajectories of cargoes in the axon were manually traced, graphed as kymographs,

and quantified in ImageJ. The mitochondrial movement parameters “percent time in motion,”

“speed,” “total distance traveled,” and “density” were measured using the Kymolyzer macro, as

previously described (Pekkurnaz et al., 2014). Percent time in motion per axon was calculated by

averaging the percent of frames (0-100%) for which each mitochondrion in the axon moved,

either in the anterograde or retrograde direction. The speed of each mitochondrion was calculated

as the average of all instantaneous speeds in a trace that were not zero. The total distance

travelled per axon was calculated as the average of total distances traveled in either direction for

each mitochondrion within the axon. Density was calculated using the first frame of each time-

lapse movie of the selected axons. Significance was calculated by unpaired Student’s t-test and

Mann-Whitney U test.

Lysosomal flux was determined by drawing 2 vertical lines in the same positions on each

kymograph and manually counting the number of tracks crossing the lines in the anterograde and

retrograde directions. The flux in each axon was determined by averaging the values obtained

from the 2 lines. Significance was calculated by unpaired Student’s t-test.

Quantification of O-GlcNAc levels

O-GlcNAc levels were assessed in PUGNAc-treated DRG neurons by immunostaining for O-GlcNAc and β-tubulin III, as described in Chapter 2. Maximum intensity projections were created for each set of Z-stack images, and imaging fields containing only axons were used for analysis. O-GlcNAc levels were quantified in ImageJ by measuring the average integrated density of the O-GlcNAc immunofluorescence signal within the area of the β-tubulin III signal.

The Huang threshold was first applied to the β-tubulin III signal, and the image calculator was used to capture the overlap between the O-GlcNAc and β-tubulin III channels (AND function).

The integrated density of the O-GlcNAc signal within the β-tubulin III region was then measured

and normalized to the total area of the thresholded β-tubulin III region. Significance was

calculated by unpaired Student’s t-test.

Axonal outgrowth and survival

DRG neurons were dissected from 4-week-old mice and cultured as described above and

in Chapter 2. To assess the effect of glucose levels on axonal outgrowth at DIV1, neurons were

plated in the following conditions: 5 mM glucose media for 12 h, 25 mM glucose media for 12 h,

5 mM media for 6 h then 25 mM media for 6 h, or 25 mM media for 6 h then 5 mM media for 6

h. β-tubulin III and O-GlcNAc immunostaining and image acquisition and analysis were

performed as described in Chapter 2. Significance was calculated by two-way repeated measures

ANOVA with Bonferroni correction.

To assess DRG neuron survival, neuronal cell bodies were manually counted and scored

for the presence of O-GlcNAc at DIV1 and DIV10. All cell bodies were included, regardless of

the extent of axonal outgrowth.

Immunofluorescence

Mitochondrial distribution was assessed in DIV1 DRG neurons by immunostaining with the primary antibody rabbit anti-Tom20 at 1:250 (Santa Cruz) and the secondary antibody anti-

rabbit Alexa Fluor 568 at 1:500 (Invitrogen) using methods described in Chapter 2.

Lumbar DRGs (L3-L5) from 4-week-old animals were dissected, processed, and

immunostained as described in Chapter 2. Briefly, antigen retrieval was performed on 14 µm

cryosections before incubation with primary antibody overnight at 4°C and secondary antibody

for 1 h at room temperature. Primary antibodies used include: rabbit anti-ATF3 at 1:400 (Aviva

Systems Biology), rabbit anti-CREB at 1:400 (clone 48H2, Cell Signaling), rabbit anti-phospho-

CREB (Ser133) at 1:400 (clone 87G3, Cell Signaling), rabbit anti-MAPK (ERK-1, ERK-2) at

1:400 (Sigma-Aldrich), rabbit anti-phospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204) at 1:400

(clone D13.14.4E, Cell Signaling), and rabbit anti-TrkA at 1:400 (Chemicon). Secondary

antibodies used include anti-rabbit Alexa Fluor 488 at 1:500 (Invitrogen) and anti-mouse Alexa

Fluor 568 at 1:500 (Invitrogen).

Chapter 4

Extended discussion

4.1 OGT is essential for neuronal survival

Research in the recent decades has established O-GlcNAcylation as a key form of regulation for basic cellular processes and metabolic pathways. Despite the abundance of OGT in the nervous system, its roles in neurons are still relatively unexplored. I aimed to investigate the importance of OGT in neurons by characterizing the effects of altered O-GlcNAc levels specifically in primary sensory neurons, a neuronal population susceptible to changes in metabolic conditions. My findings have revealed that OGT is essential for the long-term maintenance and survival of sensory neurons and plays roles in mediating axonal outgrowth and mitochondrial motility.

This work presents the first evidence that the survival of neurons can be substantially reduced due to the loss of OGT. In Nav1.8-Ogt knockout animals, there is already a significant reduction in DRG cell body counts at P10, suggesting that a subset of neurons are lost early in development or fail to develop at all. At 34 weeks of age, the number of cell bodies is further reduced, indicating that additional neurons are lost over time. It is likely, then, that loss of OGT affects both the development and maintenance of sensory neurons and that different neurons may have different susceptibilities to the deleterious effects of OGT deletion. DRG neurons are highly heterogeneous, and so it is possible that certain subtypes of neurons are more likely to die early in development, while others are able to last into adulthood.

Previous studies have suggested that while OGT deletion in mitotic cells is lethal, possibly due to the roles of O-GlcNAcylation in regulating the cell cycle, post-mitotic cells such as neurons can survive in the absence of OGT (Levine and Walker, 2016). Until recently, the in vivo effects of OGT loss in neurons have only been examined in a few studies. O'Donnell et al.

(2004) used the Syn1-Cre driver to create a pan-neuronal OGT knockout mouse; these mice were

smaller, exhibited locomotor defects, and died within 10 days of birth. However, the authors did

not further characterize the possible developmental or degenerative defects of the knockout

neurons on a cellular level. Ruan et al. (2014) used the AgRP-Cre driver to selectively delete

OGT in hypothalamic AgRP neurons and demonstrated effects on neuronal activity and the

regulation of thermogenesis. However, they did not report whether they observed any changes in

neuron number over time and only showed that some AgRP neurons lacking OGT were still

present in the knockout animals. Lagerlof et al. (2016) used the αCaMKII-CreERT2 driver to delete OGT in αCaMKII-expressing neurons and showed that this affected the excitatory synaptic function of neurons in the PVN, which in turn affected feeding behavior in mice. The authors briefly state that deletion of OGT in post-mitotic neurons did not affect cell number, based on a normal number of DAPI-stained knockout neurons either in primary cultures infected with Cre-expressing virus or in the CA1 region of hippocampal slices from the knockout animals. My findings demonstrate that loss of OGT can result in axonal degeneration and neuronal death, particularly in the long term. Thus, it is likely that pan-neuronal OGT knockout mice experience a physical loss or absence of neurons, resulting in the lethal phenotype.

A recently published study concurrent to my work has demonstrated that neuronal cell death due to the loss of OGT is a phenomenon not only in peripheral neurons but in neurons in the central nervous system as well. Wang et al. (2016) used the αCaMKII-Cre driver to delete

OGT in excitatory neurons of the forebrain, starting at P14-21, and discovered that loss of OGT resulted in progressive neurodegeneration that included neuronal cell death, increased expression of neuroinflammatory markers, increased levels of hyperphosphorylated tau and amyloidogenic

Aβ-peptides, and increased anxiety and memory impairment. Neuron loss was observed as a significant decrease in cortex size and cortical thickness by 4 months of age and a progressive

decrease in neuronal density at 2, 4, and 6 months of age in the dentate gyrus and CA1 region of the hippocampus. These results validate my findings that OGT is essential in maintaining neuronal health. The differing phenotypes in the Lagerlof and Wang studies, which both use

αCaMKII-mediated knockout models, could be explained by the timing of OGT knockout. In the former, loss of OGT was induced via tamoxifen injections at 6 weeks of age, and the effects were only observed for up to 4 weeks later (Lagerlof et al., 2016). In the latter, loss of OGT occurred at an earlier postnatal stage, and the effects were monitored for up to 6 months (Wang et al., 2016). Likewise, in my study, dramatic neuron loss in Nav1.8-Ogt knockout mice was not

observed until the older time point, at 34 weeks of age. In brn3a-Ogt knockout mice, cell body

loss was not detected until 4 weeks after Cre induction, and the effect at that point was modest.

Altogether, the data indicate that neurodegeneration due to the loss of OGT is progressive and

more easily detected in older knockout animals, and that in the short term, a rigorous

examination of neuronal cell body count is necessary to detect more subtle changes in neuron

number.

One aspect for further investigation is the nature of the neuronal death in this system and

whether loss of OGT eventually induces apoptotic cell death in sensory neurons. I show that

brn3a-Ogt knockout mice present with decreased epidermal innervation prior to cell body loss,

which suggests that the neurons undergo a period of axonal degeneration before cell death. Wang

et al. (2016) observed positive TUNEL and Fluoro-Jade C staining in their αCaMKII-Ogt

knockout model, demonstrating the presence of apoptotic and degenerating neurons. However,

preliminary experiments in which I immunostained for cleaved caspase-3, a marker for apoptosis, in Nav1.8-Ogt knockout DRGs from 4-week-old mice did not produce any positive results. I hypothesized that due to the progressive nature of neuron loss, it may be difficult to

capture apoptotic cells in any given DRG section from a single time point. Analysis of the

number of DRG cell bodies at additional ages could help to identify the period with the highest

rate of neuron loss, during which apoptotic markervs may be more readily detected in serially

sectioned DRGs. It is also possible that other methods for visualizing apoptotic cells, such as

TUNEL staining, would be more effective. Additional studies could focus on cultured OGT

knockout neurons as well, and whether they undergo apoptosis under various conditions of stress

or nutrient availability.

4.2 Behavioral and global consequences of loss of sensory neurons

Loss of sensory neurons due to the loss of OGT manifested in behavioral consequences

such as decreased thermal and mechanical sensitivity in the mice. Sensory neurons are highly

heterogeneous, and thus a diverse range of neuronal subtypes can be affected in the knockout

mice. Nav1.8 is expressed in up to 75% of sensory neurons, which encompass the majority of

neurons with C fibers as well as a subset of those with A fibers (Shields et al., 2012). In vivo

studies of afferent types in mouse glabrous skin in particular have identified several classes of neurons, including Aβ rapidly adapting (RA) and slowly adapting (SA) mechanoreceptors, Aδ fibers with high mechanical thresholds and those sensitive to temperature, and largely polymodal

C fibers sensitive to temperature and different levels of mechanical stimuli (Cain et al., 2001).

The behavioral phenotypes that I observed could therefore be attributed to the loss of specific subsets of these DRG neurons. For example, afferents positive for Mas-related G-protein- coupled receptor D (Mrgprd) constitute over 90% of non-peptidergic C fibers in the epidermis, and specific ablation of these neurons in mice results in deficits in mechanical sensitivity, as measured by the von Frey assay, but not in thermal or cold sensitivity. On the other hand, ablation of fibers positive for transient receptor potential vanilloid receptor 1 (TRPV1), a

capsaicin-sensitive receptor involved in nociception, causes decreased sensitivity to heat in the

hot plate and thermal place preference assays, but does not affect behavior in the von Frey assay

(Cavanaugh et al., 2009). The findings indicate that these neurons are important for specific

sensory modalities, and suggest that specific loss of these neuron types are contributing to the

behavioral phenotypes of Nav1.8-Ogt knockout mice. To better correlate the loss of specific

subtypes with the behavioral consequences, it would be interesting to differentiate the remaining

epidermal nerve fibers in the OGT knockout mice by immunostaining for specific markers or

receptors. An earlier or more severe loss of TRPV1-positive fibers, for example, could help to

explain the stronger thermal sensitivity phenotypes. Ultimately, I observed a significant

reduction of epidermal nerve fibers throughout adulthood in the mice, especially at 34 weeks of

age when almost no fibers could be visualized, suggesting that most sensory neuron subtypes

that innervate the epidermis are affected.

The loss of OGT in sensory neurons also had an effect on the weight and glucose

tolerance of the knockout mice. Because Nav1.8 is widely recognized to be expressed solely in

sensory neurons, it is unlikely that these phenotypes are due to the loss of OGT in other cell

types. Instead, it is possible that loss of innervation at organs such as the pancreas, stomach,

liver, and intestines affects their function, resulting in changes in the global metabolism of the

animals. Nav1.8 is expressed in the majority of neuronal cell bodies in the nodose ganglia, which gives rise to the vagus nerve and innervates the viscera, and both vagal and spinal afferents are

found throughout the alimentary canal, liver, and islets of Langherans in the pancreas (Stirling et

al., 2005; Gautron et al., 2011). A large proportion of these afferents are peptidergic and express

TRPV1. Although the roles of TRPV1 in metabolism have yet to be fully elucidated,

manipulating TRPV1 activity in mice has been shown to have an effect on body weight, fat

deposition, and pancreatic function (Ahern, 2013; Dunn and Adams, 2014; Riera and Dillin,

2016). Vagal and spinal afferents in the walls of the stomach and intestines also respond to

physiological and noxious chemical and mechanical stimuli, such as gut distension (Mayer,

2011). Vagal afferents in particular are sensitive to levels of nutrient uptake and can mediate functions such as gastric emptying, pancreatic secretions, and hepatic glucose production (Dunn and Adams, 2014). Therefore, a large reduction in visceral innervation, similar to what was observed in the skin of Nav1.8-Ogt knockout mice, could affect a range of metabolic processes.

Additional studies of weight and glucose tolerance over time in the brn3a-Ogt knockout mice would help to distinguish the effects of loss of vagal versus spinal innervation, since brn3a-

CreERT2 is expressed in DRGs but not in nodose ganglia.

4.3 Deciphering OGT-regulated pathways

While this dissertation work has focused largely on the global effects of OGT knockout

in sensory neurons, it can serve as the foundation for more mechanistic investigation into the

roles of O-GlcNAcylation in neurons. My initial findings show that there are deficits in neurite outgrowth with the loss of OGT and that axonal mitochondrial transport can also be affected with altered O-GlcNAc dynamics. These are likely to be two of many processes that OGT regulates in sensory neurons. Proteomic studies over the years have identified hundreds of O-

GlcNAcylated neuronal proteins from rat forebrain (Khidekel et al., 2004; Khidekel et al., 2007), mouse cortex (Alfaro et al., 2012), mouse synaptosomes (Trinidad et al., 2012), and rat sciatic nerve (Kim et al., 2016). The modified proteins cover a wide range of biological processes, with enrichment in signaling pathways, transcriptional regulation, and cytoskeletal structures (Rexach et al., 2008). As the technology for profiling O-GlcNAcylated proteins continues to improve, the extent to which we understand the role of O-GlcNAc regulation in neurons is likely to grow.

Given the number of processes that OGT has been shown to regulate, future directions

include investigating changes in these processes in the OGT knockout neurons using proteomic

or transcriptomic approaches. Wang et al. (2016) performed gene-expression microarray analysis

and observed an up-regulation of over 900 genes in the hippocampi of 2-month-old αCaMKII-

Ogt knockout mice. Gene ontology analysis identified 2 gene modules correlated with OGT

deletion: one was enriched with genes involved in immune response, including Bcl-2 homologous antagonist killer (Bak1) and members of the complement cascade, and the other was enriched with genes involved in cell-cycle arrest, including protein regulator of cytokinesis 1

(Prc1) and kinetochore protein Spc25 (Spc25). A similar study in sensory neurons to identify differentially regulated genes or proteins would further our understanding of the processes that are most affected in OGT knockout neurons and are thus likely to be contributing to the decreased survivability in these models. Additionally, it would be interesting to compare the O-

GlcNAcylated proteome of primary sensory neurons with that of central neurons to see if there are any uniquely modified proteins in the peripheral neurons, which could be indicative of differences in metabolic regulation and unique roles of OGT in sensory neurons.

4.4 Implications in diabetic neuropathy

Considering the essential role of OGT in sensory neuron maintenance, it is possible that

altered O-GlcNAcylation is a contributing factor to the development of peripheral neuropathies,

in particular diabetic neuropathy. In diabetes, peripheral tissues are exposed to chronic

hyperglycemia, and in neurons, this could lead to abnormal O-GlcNAc dynamics, misregulated

cellular processes, and eventual axonal degeneration.

While changes in O-GlcNAcylation have yet to be linked to diabetic neuropathy, the

modification has been demonstrated to play a role in diabetes and other diabetic complications.

Increased O-GlcNAcylation is observed in several tissue types in diabetic rodent models and is

believed to contribute to insulin resistance (Ma and Hart, 2013). Moreover, a variant of Oga

containing a single nucleotide polymorphism leading to possible decreased OGA activity, has

been linked to increased diabetes risk in Mexican Americans (Lehman et al., 2005). Increased O-

GlcNAcylation has been implicated in the pathogenesis of several diabetic complications.

Kidney biopsy samples from diabetic nephropathy patients exhibit increased O-GlcNAc levels in

the glomeruli and tubules (Degrell et al., 2009), and studies in rat mesangial cells show that O-

GlcNAcylation enhances profibrotic signaling in response to high glucose, suggesting a role in

the matrix expansion and fibrosis that occurs in diabetic nephropathy (Goldberg et al., 2011). In

rodent models of diabetic cardiomyopathy, O-GlcNAcylation appears to contribute to

myofilament Ca2+ insensitivity and cardiac mitochondrial dysfunction (Banerjee et al., 2015;

Ramirez-Correa et al., 2015). Studies on diabetic keratopathy and retinopathy also show

increased levels of O-GlcNAc in the corneas and retinas of diabetic rodents, possibly contributing to changes in cell-to-cell adhesion, tubulin dynamics, and cell migration (Semba et al., 2014). Finally, hyperglycemia increases O-GlcNAc levels and delays wound healing in human keratinocytes, whereas OGT knockdown accelerates wound healing, suggesting a role for the modification in the delayed wound healing observed in patients with diabetic skin ulcers

(Runager et al., 2014). Together, these findings support the possibility that altered O-

GlcNAcylation also occurs in the peripheral nerves of diabetic patients and contributes to the pathogenesis of neuropathy.

An original aim of this dissertation work was to investigate the effects of OGT loss in sensory neurons on the development of neuropathy in a diabetic mouse model. The hypothesis was that if loss of OGT did not have a severe phenotype on its own, then the absence of the

metabolic sensor may delay the development of diabetic neuropathy. Therefore, I initially

wanted to ask whether induction of diabetes in Nav1.8-Ogt knockout mice by streptozotocin

(STZ) treatment would exacerbate or alleviate the neuropathy that typically develops in STZ mice (O'Brien et al., 2014). STZ is a chemical that is toxic to pancreatic β-cells, and thus causes hyperglycemia (King, 2012). Upon discovering the severe reduction of epidermal nerve fibers in the Nav1.8-Ogt knockout mice, however, I decided to shift the focus to characterizing the effects of OGT loss, which in itself appeared to cause significant neuropathy.

As a future direction, it would be interesting to introduce metabolic changes in vivo and ask whether differences in feeding protocols or diet could affect the severity of neuropathy observed in OGT knockout mice. For example, would a high-fat diet, which over time can induce obesity and glucose tolerance, affect the progressive degeneration of the sensory neurons in brn3a-Ogt knockout mice? It is possible that more severe degenerative phenotypes would be observed if the neurons lacking OGT are unable to appropriately adjust to the abnormal metabolic conditions and are put under further stress.

In general, increased O-GlcNAcylation is a common feature in diabetic complications

and appears to mediate some of the detrimental effects of hyperglycemia. In this work, I

demonstrated that increased O-GlcNAcylation in DRG neurons results in decreased

mitochondrial motility; this suggests that mitochondrial transport could be one cellular process

that is misregulated under diabetic conditions. Proper distribution and trafficking of

mitochondria is especially important in neurons, and disrupted transport is a key feature of

several forms of neurodegeneration and neuropathy, including Charcot-Marie-Tooth disease and

ALS (Baloh et al., 2007; Magrane et al., 2014). OGT regulates mitochondrial transport through

the modification of Milton, and expression of a mutant form of Milton, in which 4 key O-

GlcNAcylated sites have been altered, rescues the motility arrest under high glucose conditions

(Pekkurnaz et al., 2014). As a future direction, this work can be moved into an in vivo model, wherein a mouse expresses a knock-in allele of Milton that is resistant to modification by OGT.

It would then be interesting to assess whether the model is more or less prone to developing neuropathy when induced with diabetes, such as using STZ treatment or a high-fat diet. If altered mitochondrial transport via hyper-glycosylation is indeed a key contributing factor in the pathogenesis of diabetic neuropathy, then one would expect that removal of this form of regulation would prevent or delay the development of neuropathy in vivo.

Ultimately, diabetic neuropathy remains a complex and heterogeneous disease state that likely has many causal factors. Further investigation of the roles of OGT in mediating essential neuronal processes would advance not only our knowledge of basic O-GlcNAc biology but also our understanding of the modification in mediating neurodegenerative states such as diabetic neuropathy.

References

Abraira VE, Ginty DD (2013) The sensory neurons of touch. Neuron 79:618-639.

Agarwal N, Offermanns S, Kuner R (2004) Conditional gene deletion in primary nociceptive neurons of trigeminal ganglia and dorsal root ganglia. Genesis 38:122-129.

Ahern GP (2013) Transient receptor potential channels and energy homeostasis. Trends Endocrinol Metab 24:554-560.

Akimoto Y, Kreppel LK, Hart GW (2000) Increased O-GlcNAc transferase in pancreas of rats with streptozotocin-induced diabetes. Diabetologia 43:1239-1247.

Akimoto Y, Kreppel LK, Hirano H, Hart GW (2001) Hyperglycemia and the O-GlcNAc transferase in rat aortic smooth muscle cells: Elevated expression and altered patterns of O-GlcNAcylation. Arch Biochem Biophys 389:166-175.

Akopian AN, Sivilotti L, Wood JN (1996) A tetrodotoxin-resistant voltage-gated sodium channel expressed by sensory neurons. Nature 379:257-262.

Alejandro EU, Bozadjieva N, Kumusoglu D, Abdulhamid S, Levine H, Haataja L, Vadrevu S, Satin LS, Arvan P, Bernal-Mizrachi E (2015) Disruption of O-linked N- acetylglucosamine signaling induces ER stress and β cell failure. Cell Rep 13:2527-2538.

Alfaro JF, Gong C-X, Monroe ME, Aldrich JT, Clauss TRW, Purvine SO, Wang Z, Camp DG 2nd, Shabanowitz J, Stanley P, Hart GW, Hunt DF, Yang F, Smith RD (2012) Tandem mass spectrometry identifies many mouse brain O-GlcNAcylated proteins including EGF domain-specific O-GlcNAc transferase targets. Proc Natl Acad Sci U S A 109:7280- 7285.

Andrali SS, Qian Q, Ozcan S (2007) Glucose mediates the translocation of NeuroD1 by O-linked glycosylation. J Biol Chem 282:15589-15596.

Amaya F, Decosterd I, Samad TA, Plumpton C, Tate S, Mannion RJ, Costigan M, Woolf CJ (2000) Diversity of expression of the sensory neuron-specific TTX-resistant voltage- gated sodium ion channels SNS and SNS2. Mol Cell Neurosci 15:331-342.

American Diabetes Association (2000) Screening for type 2 diabetes. Diabetes Care 23:S20-23.

Baloh RH, Schmidt RE, Pestronk A, Milbrandt J (2007) Altered axonal mitochondrial transport in the pathogenesis of Charcot-Marie-Tooth disease from mitofusin 2 mutations. J Neurosci 27:422-430.

Banerjee PS, Ma J, Hart GW (2015) Diabetes-associated dysregulation of O-GlcNAcylation in rat cardiac mitochondria. Proc Natl Acad Sci U S A 112:6050-6055.

Bertram L, Blacker D, Mullin K, Keeney D, Jones J, Basu S, Yhu S, McInnis MG, Go RCP, Vekrellis K, Selkoe DJ, Saunders AJ, Tanzi RE (2000) Evidence for genetic linkage of Alzheimer's disease to chromosome 10q. Science 290:2302-2303.

Bond MR, Hanover JA (2015) A little sugar goes a long way: The cell biology of O-GlcNAc. J Cell Biol 208:869-880.

Bowe JE, Franklin ZJ, Hauge-Evans AC, King AJ, Persaud SJ, Jones PM (2014) Metabolic phenotyping guidelines: Assessing glucose homeostasis in rodent models. J Endocrinol 222:G13-G25.

Bullen JW, Balsbaugh JL, Chanda D, Shabanowitz J, Hunt DF, Neumann D, Hart GW (2014) Cross-talk between two essential nutrient-sensitive enzymes: O-GlcNAc transferase (OGT) and AMP-activated protein kinase (AMPK). J Biol Chem 289:10592-10606.

Cain DM, Khasabov SG, Simone DA (2001) Response properties of mechanoreceptors and nociceptors in mouse glabrous skin: An in vivo study. J Neurophysiol 85:1561-1574.

Callaghan BC, Cheng HT, Stables CL, Smith AL, Feldman EL (2012) Diabetic neuropathy: Clinical manifestations and current treatments. Lancet Neurol 11:521-534.

Campos-Bedolla P, Walter FR, Veszelka S, Deli MA (2014) Role of the blood–brain barrier in the nutrition of the central nervous system. Arch Med Res 45:610-638.

Capotosti F, Guernier S, Lammers F, Waridel P, Cai Y, Jin J, Conaway JW, Conaway RC, Herr W (2011) O-GlcNAc transferase catalyzes site-specific proteolysis of HCF-1. Cell 144:376-388.

Catala M, Kubis N (2013) Gross anatomy and development of the peripheral nervous system. Handb Clin Neurol 115:29-41.

Cavanaugh DJ, Lee H, Lo L, Shields SD, Zylka MJ, Basbaum AI, Anderson DJ (2009) Distinct subsets of unmyelinated primary sensory fibers mediate behavioral responses to noxious thermal and mechanical stimuli. Proc Natl Acad Sci U S A 106:9075-9080.

Cheung WD, Hart GW (2008) AMP-activated protein kinase and p38 MAPK activate O- GlcNAcylation of neuronal proteins during glucose deprivation. J Biol Chem 283:13009- 13020.

Cheung WD, Sakabe K, Housley MP, Dias WB, Hart GW (2008) O-linked β-N- acetylglucosaminyltransferase substrate specificity is regulated by myosin phosphatase targeting and other interacting proteins. J Biol Chem 283:33935-33941.

Chiang M-F, Wu C-L, Lin K-I (2013) O-linked N-acetylglucosaminyl transferase (Ogt) promotes the survival of mature B cells. Front Immunol Conference Abstract.

100

Chiu IM, Barrett LB, Williams EK, Strochlic DE, Lee S, Weyer AD, Lou S, Bryman GS, Roberson DP, Ghasemlou N, Piccoli C, Ahat E, Wang V, Cobos EJ, Stucky CL, Ma Q, Liberles SD, Woolf CJ (2014) Transcriptional profiling at whole population and single cell levels reveals somatosensory neuron molecular diversity. eLife 3:e04660.

Clark RJ, McDonough PM, Swanson E, Trost SU, Suzuki M, Fukuda M, Dillmann WH (2003) Diabetes and the accompanying hyperglycemia impairs cardiomyocyte calcium cycling through increased nuclear O-GlcNAcylation. J Biol Chem 278:44230-44237.

Comtesse N, Maldener E, Meese E (2001) Identification of a nuclear variant of MGEA5, a cytoplasmic hyaluronidase and a β-N-acetylglucosaminidase. Biochem Biophys Res Commun 283:634-640.

Court FA, Coleman MP (2012) Mitochondria as a central sensor for axonal degenerative stimuli. Trends Neurosci 35:364-372.

Datta B, Ray MK, Chakrabarti D, Wylie DE, Gupta NK (1989) Glycosylation of eukaryotic peptide chain initiation factor 2 (eIF-2)-associated 67-kDa polypeptide (p67) and its possible role in the inhibition of eIF-2 kinase-catalyzed phosphorylation of the eIF-2 α- subunit. J Biol Chem 264:20620-20624.

Degrell P, Cseh J, Mohas M, Molnar GA, Pajor L, Chatham JC, Fulop N, Wittmann I (2009) Evidence of O-linked N-acetylglucosamine in diabetic nephropathy. Life Sci 84:389-393.

Dentin R, Hedrick S, Xie J, Yates J 3rd, Montminy M (2008) Hepatic glucose sensing via the CREB coactivator CRTC2. Science 319:1402-1405.

Djouhri L, Fang X, Okuse K, Wood JN, Berry CM, Lawson SN (2003) The TTX-resistant sodium channel Nav1.8 (SNS/PN3): Expression and correlation with membrane properties in rat nociceptive primary afferent neurons. J Physiol 550:739-752.

Dubin AE, Patapoutian A (2010) Nociceptors: the sensors of the pain pathway. J Clin Invest 120:3760-3772.

Dunn TN, Adams SH (2014) Relations between metabolic homeostasis, diet, and peripheral afferent neuron biology. Adv Nutr 5:386-393.

Durning SP, Flanagan-Steet H, Prasad N, Wells L (2016) O-linked β-N-acetylglucosamine (O- GlcNAc) acts as a glucose sensor to epigenetically regulate the insulin gene in pancreatic beta cells. J Biol Chem 291:2107-2118.

Eng SR, Gratwick K, Rhee JM, Fedtsova N, Gan L, Turner EE (2001) Defects in sensory axon growth precede neuronal death in Brn3a-deficient mice. J Neurosci 21:541-549.

Ferreira JM, Burnett AL, Rameau GA (2011) Activity-dependent regulation of surface glucose transporter-3. J Neurosci 31:1991-1999.

101

Francisco H, Kollins K, Varghis N, Vocadlo D, Vosseller K, Gallo G (2009) O-GlcNAc post- translational modifications regulate the entry of neurons into an axon branching program. Dev Neurobiol 69:162-173.

Fricovsky ES, Suarez J, Ihm S-H, Scott BT, Suarez-Ramirez JA, Banerjee I, Torres-Gonzalez M, Wang H, Ellrott I, Maya-Ramos L, Villarreal F, Dillmann WH (2012) Excess protein O- GlcNAcylation and the progression of diabetic cardiomyopathy. Am J Physiol Regul Integr Comp Physiol 303:R689-R699.

Gao Y, Miyazaki J-I, Hart GW (2003) The transcription factor PDX-1 is post-translationally modified by O-linked N-acetylglucosamine and this modification is correlated with its DNA binding activity and insulin secretion in min6 β-cells. Arch Biochem Biophys 415:155-163.

Gao Y, Wells L, Comer FI, Parker GJ, Hart GW (2001) Dynamic O-glycosylation of nuclear and cytosolic proteins. Cloning and characterization of a neutral, cytosolic β-N- acetylglucosaminidase from human brain. J Biol Chem 276:9838-9845.

Gautron L, Sakata I, Udit S, Zigman JM, Wood JN, Elmquist JK (2011) Genetic tracing of Nav1.8-expressing vagal afferents in the mouse. J Comp Neurol 519:3085-3101.

Gawlowski T, Suarez J, Scott B, Torres-Gonzalez M, Wang H, Schwappacher R, Han X, Yates JR 3rd, Hoshijima M, Dillmann W (2012) Modulation of dynamin-related protein 1 (DRP1) function by increased O-linked-β-N-acetylglucosamine modification (O- GlcNAc) in cardiac myocytes. J Biol Chem 287:30024-30034.

Goldberg H, Whiteside C, Fantus IG (2011) O-linked β-N-acetylglucosamine supports p38 MAPK activation by high glucose in glomerular mesangial cells. Am J Physiol Endocrinol Metab 301:E713-E726.

Guinez C, Mir A-M, Dehennaut V, Cacan R, Harduin-Lepers A, Michalski JC, Lefebvre T (2008) Protein ubiquitination is modulated by O-GlcNAc glycosylation. FASEB J 22:2901-2911.

Haberhausen G, Schmitt I, Kohler A, Peters U, Rider S, Chelly J, Terwilliger JD, Monaco AP, Muller U (1995) Assignment of the dystonia-parkinsonism syndrome locus, DYT3, to a small region within a 1.8-Mb YAC contig of Xq13.1. Am J Hum Genet 57:644-650.

Haltiwanger RS, Blomberg MA, Hart GW (1992) Glycosylation of nuclear and cytoplasmic proteins. Purification and characterization of a uridine diphospho-N-acetylglucosamine: polypeptide β-N-acetylglucosaminyltransferase. J Biol Chem 267:9005-9013.

Han I, Oh E-S, Kudlow JE (2000) Responsiveness of the state of O-linked N-acetylglucosamine modification of nuclear pore protein p62 to the extracellular glucose concentration. Biochem J 350:109-114.

102

Hanover JA, Lai Z, Lee G, Lubas WA, Sato SM (1999) Elevated O-linked N-acetylglucosamine metabolism in pancreatic β-cells. Arch Biochem Biophys 362:38-45.

Hanover JA, Yu S, Lubas WA, Shin S-H, Ragano-Caracciola M, Kochran J, Love DC (2003) Mitochondrial and nucleocytoplasmic isoforms of O-linked GlcNAc transferase encoded by a single mammalian gene. Arch Biochem Biophys 409:287-297.

Hart GW, Housley MP, Slawson C (2007) Cycling of O-linked β-N-acetylglucosamine on nucleocytoplasmic proteins. Nature 446:1017-1022.

Hart GW, Slawson C, Ramirez-Correa G, Lagerlof O (2011) Cross talk between O- GlcNAcylation and phosphorylation: Roles in signaling, transcription, and chronic disease. Annu Rev Biochem 80:825-858.

Heckel D, Comtesse N, Brass N, Blin N, Zang KD, Meese E (1998) Novel immunogenic antigen homologous to hyaluronidase in meningioma. Hum Mol Genet 7:1859-1872.

Hetman M, Gozdz A (2004) Role of extracellular signal regulated kinases 1 and 2 in neuronal survival. Eur J Biochem 271:2050-2055.

Holt GD, Hart GW (1986) The subcellular distribution of terminal N-acetylglucosamine moieties. J Biol Chem 261:8049-8057.

Housley MP, Udeshi ND, Rodgers JT, Shabanowitz J, Puigserver P, Hunt DF, Hart GW (2009) A PGC-1α-O-GlcNAc transferase complex regulates FoxO transcription factor activity in response to glucose. J Biol Chem 284:5148-5157.

Housley MP, Rodgers JT, Udeshi ND, Kelly TJ, Shabanowitz J, Hunt DF, Puigserver P, Hart GW (2008) O-GlcNAc regulates FoxO activation in response to glucose. J Biol Chem 283:16283-16292.

Hunt D, Raivich G, Anderson PN (2012) Activating transcription factor 3 and the nervous system. Front Mol Neurosci 5:7.

Iyer SPN, Akimoto Y, Hart GW (2003) Identification and cloning of a novel family of coiled- coil domain proteins that interact with O-GlcNAc transferase. J Biol Chem 278:5399- 5409.

Ji S, Kang JG, Park SY, Lee J, Oh YJ, Cho JW (2011) O-GlcNAcylation of tubulin inhibits its polymerization. Amino Acids 40:809-818.

Jinek M, Rehwinkel J, Lazarus BD, Izaurralde E, Hanover JA, Conti E (2004) The superhelical TPR-repeat domain of O-linked GlcNAc transferase exhibits structural similarities to importin α. Nat Struct Mol Biol 11:1001-1007.

Kazemi Z, Chang H, Haserodt S, McKen C, Zachara NE (2010) O-linked β-N-acetylglucosamine

103

(O-GlcNAc) regulates stress-induced heat shock protein expression in a GSK-3β- dependent manner. J Biol Chem 285:39096-39107.

Keembiyehetty C, Love DC, Harwood KR, Gavrilova O, Comly ME, Hanover JA (2015) Conditional knock-out reveals a requirement for O-linked N-acetylglucosaminase (O- GlcNAcase) in metabolic homeostasis. J Biol Chem 290:7097-7113.

Keembiyehetty CN, Krzeslak A, Love DC, Hanover JA (2011) A lipid-droplet-targeted O- GlcNAcase isoform is a key regulator of the proteasome. J Cell Sci 124:2851-2860.

Khidekel N, Ficarro SB, Peters EC, Hsieh-Wilson LC (2004) Exploring the O-GlcNAc proteome: Direct identification of O-GlcNAc-modified proteins from the brain. Proc Natl Acad Sci U S A 101:13132-13137.

Khidekel N, Ficarro SB, Clark PM, Bryan MC, Swaney DL, Rexach JE, Sun YE, Coon JJ, Peters EC, Hsieh-Wilson LC (2007) Probing the dynamics of O-GlcNAc glycosylation in the brain using quantitative proteomics. Nature Chem Biol 3:339-348.

Kim C, Nam DW, Park SY, Song H, Hong HS, Boo JH, Jung ES, Kim Y, Baek JY, Kim KS, Cho JW, Mook-Jung I (2013) O-linked β-N-acetylglucosaminidase inhibitor attenuates β- amyloid plaque and rescues memory impairment. Neurobiol Aging 34:275-285.

Kim S, Maynard JC, Sasaki Y, Strickland A, Sherman DL, Brophy PJ, Burlingame AL, Milbrandt J (2016) Schwann cell O-GlcNAc glycosylation is required for myelin maintenance and axon integrity. J Neurosci 36:9633-9646.

King AJF (2012) The use of animal models in diabetes research. Br J Pharmacol 166:877-894.

Kreppel LK, Hart GW (1999) Regulation of a cytosolic and nuclear O-GlcNAc transferase. Role of the tetratricopeptide repeats. J Biol Chem 274:32015-32022.

Kreppel LK, Blomberg MA, Hart GW (1997) Dynamic glycosylation of nuclear and cytosolic proteins. Cloning and characterization of a unique O-GlcNAc transferase with multiple tetratricopeptide repeats. J Biol Chem 272:9308-9315.

Kumar A, Singh PK, Parihar R, Dwivedi V, Lakhotia SC, Ganesh S (2014) Decreased O-linked GlcNAcylation protects from cytotoxicity mediated by huntingtin exon1 protein fragment. J Biol Chem 289:13543-13553.

Labokha AA, Gradmann S, Frey S, Hulsmann BB, Urlaub H, Baldus M, Gorlich D (2013) Systematic analysis of barrier-forming FG hydrogels from Xenopus nuclear pore complexes. EMBO J 32:204-218.

Lacomis D (2002) Small-fiber neuropathy. Muscle Nerve 26:173-188.

Lagerlof O, Hart GW (2014) O-GlcNAcylation of neuronal proteins: Roles in neuronal functions

104

and in neurodegeneration. Adv Neurobiol 9:343-366.

Lagerlof O, Slocomb JE, Hong I, Aponte Y, Blackshaw S, Hart GW, Huganir RL (2016) The nutrient sensor OGT in PVN neurons regulates feeding. Science 351:1293-1296.

Lazarus BD, Love DC, Hanover JA (2006) Recombinant O-GlcNAc transferase isoforms: identification of O-GlcNAcase, yes tyrosine kinase, and tau as isoform-specific substrates. Glycobiology 16:415-421.

Lazarus BD, Love DC, Hanover JA (2009) O-GlcNAc cycling: Implications for neurodegenerative disorders. Int J Biochem Cell Biol 41:2134-2146.

Lazarus MB, Nam Y, Jiang J, Sliz P, Walker S (2011) Structure of human O-GlcNAc transferase and its complex with a peptide substrate. Nature 469:564-567.

Lazarus MB, Jiang J, Kapuria V, Bhuiyan T, Janetzko J, Zandberg WF, Vocadlo DJ, Herr W, Walker S (2013) HCF-1 is cleaved in the active site of O-GlcNAc transferase. Science 342:1235-1239.

Lehman DM, Fu D-J, Freeman AB, Hunt KJ, Leach RJ, Johnson-Pais T, Hamlington J, Dyer TD, Arya R, Abboud H, Goring HHH, Duggirala R, Blangero J, Konrad RJ, Stern MP (2005) A single nucleotide polymorphism in MGEA5 encoding O-GlcNAc–selective N- acetyl-β-D glycosaminidase is associated with type 2 diabetes in Mexican Americans. Diabetes 54:1214-1221.

Levine ZG, Walker S (2016) The biochemistry of O-GlcNAc transferase: Which functions make it essential in mammalian cells? Annu Rev Biochem 85:631-657.

Lewis BA, Burlingame AL, Myers SA (2016) Human RNA polymerase II promoter recruitment in vitro is regulated by O-linked N-acetylglucosaminyltransferase (OGT). J Biol Chem 291:14056-14061.

Liesa M, Shirihai OS (2013) Mitochondrial dynamics in the regulation of nutrient utilization and energy expenditure. Cell Metab 17:491-506.

Liu F, Iqbal K, Grundke-Iqbal I, Hart GW, Gong C-X (2004) O-GlcNAcylation regulates phosphorylation of tau: A mechanism involved in Alzheimer's disease. Proc Natl Acad Sci U S A 101:10804-10809.

Liu F, Shi J, Tanimukai H, Gu J, Grundke-Iqbal I, Iqbal K, Gong C-X (2009) Reduced O- GlcNAcylation links lower brain glucose metabolism and tau pathology in Alzheimer's disease. Brain 132:1820-1832.

Liu K, Paterson AJ, Chin E, Kudlow JE (2000) Glucose stimulates protein modification by O- linked GlcNAc in pancreatic β cells: Linkage of O-linked GlcNAc to β cell death. Proc Natl Acad Sci U S A 97:2820-2825.

105

Liu X, Li L, Wang Y, Yan H, Ma X, Wang PG, Zhang L (2014) A peptide panel investigation reveals the acceptor specificity of O-GlcNAc transferase. FASEB J 28:3362-3372.

Low LK, Cheng H-J (2006) Axon pruning: an essential step underlying the developmental plasticity of neuronal connections. Philos Trans R Soc Lond B Biol Sci 361:1531-1544.

Lubas WA, Frank DW, Krause M, Hanover JA (1997) O-linked GlcNAc transferase is a conserved nucleocytoplasmic protein containing tetratricopeptide repeats. J Biol Chem 272:9316-9324.

Ma J, Hart GW (2013) Protein O-GlcNAcylation in diabetes and diabetic complications. Expert Rev Proteomics 10:365-380.

Ma J, Hart GW (2014) O-GlcNAc profiling: from proteins to proteomes. Clin Proteomics 11:8.

Magrane J, Cortez C, Gan WB, Manfredi G (2014) Abnormal mitochondrial transport and morphology are common pathological denominators in SOD1 and TDP43 ALS mouse models. Hum Mol Genet 23:1413-1424.

Marotta NP, Lin YH, Lewis YE, Ambroso MR, Zaro BW, Roth MT, Arnold DB, Langen R, Pratt MR (2015) O-GlcNAc modification blocks the aggregation and toxicity of the protein α-synuclein associated with Parkinson's disease. Nat Chem 7:913-920.

Marshall S, Bacote V, Traxinger RR (1991) Discovery of a metabolic pathway mediating glucose-induced desensitization of the glucose transport system. J Biol Chem 266:4706- 4712.

Mayer EA (2011) Gut feelings: the emerging biology of gut-brain communication. Nat Rev Neurosci 12:453-466.

McClain DA, Lubas WA, Cooksey RC, Hazel M, Parker GJ, Love DC, Hanover JA (2002) Altered glycan-dependent signaling induces insulin resistance and hyperleptinemia. Proc Natl Acad Sci U S A 99:10695-10699.

Mergenthaler P, Lindauer U, Dienel GA, Meisel A (2013) Sugar for the brain: the role of glucose in physiological and pathological brain function. Trends Neurosci 36:587-597.

Nagel AK, Ball LE (2014) O-GlcNAc transferase and O-GlcNAcase: achieving target substrate specificity. Amino Acids 46:2305-2316.

Nathan DM, Davidson MB, DeFronzo RA, Heine RJ, Henry RR, Pratley R, Zinman B (2007) Impaired fasting glucose and impaired glucose tolerance: Implications for care. Diabetes Care 30:753-759.

O'Brien PD, Sakowski SA, Feldman EL (2014) Mouse models of diabetic neuropathy. ILAR J 54:259-272.

106

O'Donnell N, Zachara NE, Hart GW, Marth JD (2004) Ogt-dependent X-chromosome-linked protein glycosylation is a requisite modification in somatic cell function and embryo viability. Mol Cell Biol 24:1680-1690.

O’Donovan KJ, Ma K, Guo H, Wang C, Sun F, Han SB, Kim H, Wong JK, Charron J, Zou H, Son Y-J, He Z, Zhong J (2014) B-RAF kinase drives developmental axon growth and promotes axon regeneration in the injured mature CNS. J Exp Med 211:801-814.

Ozcan S, Andrali SS, Cantrell JEL (2010) Modulation of transcription factor function by O- GlcNAc modification. Biochim Biophys Acta 1799:353-364.

Park K, Saudek CD, Hart GW (2010) Increased expression of β-N-acetylglucosaminidase in erythrocytes from individuals with pre-diabetes and diabetes. Diabetes 49:1845-1850.

Patel TD, Jackman A, Rice FL, Kucera J, Snider WD (2000) Development of sensory neurons in the absence of NGF/TrkA signaling in vivo. Neuron 25:345-357.

Pathak S, Alonso J, Schimpl M, Rafie K, Blair DE, Borodkin VS, Schuttelkopf AW, Albarbarawi O, van Aalten DMF (2015) The active site of O-GlcNAc transferase imposes constraints on substrate sequence. Nat Struct Mol Biol 22:744-750.

Pekkurnaz G, Trinidad JC, Wang X, Kong D, Schwarz TL (2014) Glucose regulates mitochondrial motility via Milton modification by O-GlcNAc transferase. Cell 158:54- 68.

Ramirez-Correa GA, Ma J, Slawson C, Zeidan Q, Lugo-Fagundo NS, Xu M, Shen X, Gao WD, Caceres V, Chakir K, DeVine L, Cole RN, Marchionni L, Paolocci N, Hart GW, Murphy AM (2015) Removal of abnormal myofilament O-GlcNAcylation restores Ca2+ sensitivity in diabetic cardiac muscle. Diabetes 64:3573-3587.

Razavi R, Chan Y, Afifiyan FN, Liu XJ, Wan X, Yantha J, Tsui H, Tang L, Tsai S, Santamaria P, Driver JP, Serreze D, Salter MW, Dosch H-M (2006) TRPV1+ sensory neurons control β cell stress and islet inflammation in autoimmune diabetes. Cell 127:1123-1135.

Rexach JE, Clark PM, Hsieh-Wilson LC (2008) Chemical approaches to understanding O- GlcNAc glycosylation in the brain. Nat Chem Biol 4:97-106.

Rexach JE, Clark PM, Mason DE, Neve RL, Peters EC, Hsieh-Wilson LC (2012) Dynamic O- GlcNAc modification regulates CREB-mediated gene expression and memory formation. Nature Chem Biol 8:253-261.

Riera CE, Dillin A (2016) Emerging role of sensory perception in aging and metabolism. Trends Endocrinol Metab 27:294-303.

Roh E, Song DK, Kim M-S (2016) Emerging role of the brain in the homeostatic regulation of energy and glucose metabolism. Exp Mol Med 48:e216.

107

Ruan H-B, Singh JP, Li M-D, Wu J, Yang X (2013) Cracking the O-GlcNAc code in metabolism. Trends Endocrinol Metab 24:301-309.

Ruan H-B, Dietrich MO, Liu Z-W, Zimmer MR, Li M-D, Singh JP, Zhang K, Yin R, Wu J, Horvath TL, Yang X (2014) O-GlcNAc transferase enables AgRP neurons to suppress browning of white fat. Cell 159:306-317.

Ruan H-B, Han X, Li M-D, Singh JP, Qian K, Azarhoush S, Zhao L, Bennett AM, Samuel VT, Wu J, Yates JR 3rd, Yang X (2012) O-GlcNAc transferase/host cell factor C1 complex regulates gluconeogenesis by modulating PGC-1α stability. Cell Metab 16:226-237.

Runager K, Bektas M, Berkowitz P, Rubenstein DS (2014) Targeting O-glycosyltransferase (OGT) to promote healing of diabetic skin wounds. J Biol Chem 289:5462-5466.

Sayat R, Leber B, Grubac V, Wiltshire L, Persad S (2008) O-GlcNAc-glycosylation of β-catenin regulates its nuclear localization and transcriptional activity. Exp Cell Res 314:2774- 2787.

Schultz J, Pils B (2002) Prediction of structure and functional residues for O-GlcNAcase, a divergent homologue of acetyltransferases. FEBS Lett 529:179-182.

Schwarz TL (2013) Mitochondrial trafficking in neurons. Cold Spring Harb Perspect Biol 5:a011304.

Semba RD, Huang H, Lutty GA, Van Eyk JE, Hart GW (2014) The role of O-GlcNAc signaling in the pathogenesis of diabetic retinopathy. Proteomics Clin Appl 8:218-231.

Shafi R, Iyer SPN, Ellies LG, O'Donnell N, Marek KW, Chui D, Hart GW, Marth JD (2000) The O-GlcNAc transferase gene resides on the X chromosome and is essential for embryonic stem cell viability and mouse ontogeny. Proc Natl Acad Sci U S A 97:5735-5739.

Shan X, Vocadlo DJ, Krieger C (2012) Reduced protein O-glycosylation in the nervous system of the mutant SOD1 transgenic mouse model of amyotrophic lateral sclerosis. Neurosci Lett 516:296-301.

Shen DL, Gloster TM, Yuzwa SA, Vocadlo DJ (2012) Insights into O-linked N- acetylglucosamine (O-GlcNAc) processing and dynamics through kinetic analysis of O- GlcNAc transferase and O-GlcNAcase activity on protein substrates. J Biol Chem 287:15395-15408.

Shields SD, Ahn H-S, Yang Y, Han C, Seal RP, Wood JN, Waxman SG, Dib-Hajj SD (2012) Nav1.8 expression is not restricted to nociceptors in mouse peripheral nervous system. Pain 153:2017-2030.

Silver IA, Erecinska M (1994) Extracellular glucose concentration in mammalian brain: Continuous monitoring of changes during increased neuronal activity and upon limitation

108

in oxygen supply in normo-, hypo-, and hyperglycemic animals. J Neurosci 14:5068- 5076.

Skorobogatko Y, Landicho A, Chalkley RJ, Kossenkov AV, Gallo G, Vosseller K (2014) O- linked β-N-acetylglucosamine (O-GlcNAc) site Thr-87 regulates synapsin I localization to synapses and size of the reserve pool of synaptic vesicles. J Biol Chem 289:3602-3612.

Slawson C, Lakshmanan T, Knapp S, Hart GW (2008) A mitotic GlcNAcylation/phosphorylation signaling complex alters the posttranslational state of the cytoskeletal protein vimentin. Mol Biol Cell 19:4130-4140.

Slawson C, Zachara NE, Vosseller K, Cheung WD, Lane MD, Hart GW (2005) Perturbations in O-linked β-N-acetylglucosamine protein modification cause severe defects in mitotic progression and cytokinesis. J Biol Chem 280:32944-32956.

Sleigh JN, Weir GA, Schiavo G (2016) A simple, step-by-step dissection protocol for the rapid isolation of mouse dorsal root ganglia. BMC Res Notes 9:82.

Soesanto Y, Luo B, Parker G, Jones D, Cooksey RC, McClain DA (2011) Pleiotropic and age- dependent effects of decreased protein modification by O-linked N-acetylglucosamine on pancreatic β-cell function and vascularization. J Biol Chem 286:26118-26126.

Springhorn C, Matsha TE, Erasmus RT, Essop MF (2012) Exploring leukocyte O- GlcNAcylation as a novel diagnostic tool for the earlier detection of type 2 diabetes mellitus. J Clin Endocrinol Metab 97:4640-4649.

Srikanth B, Vaidya MM, Kalraiya RD (2010) O-GlcNAcylation determines the solubility, filament organization, and stability of keratins 8 and 18. J Biol Chem 285:34062-34071.

Stefanini M, De Martino C, Zamboni L (1967) Fixation of ejaculated spermatozoa for electron microscopy. Nature 216:173-174.

Stewart MA, Sherman WR, Anthony S (1966) Free sugars in alloxan diabetic rat nerve. Biochem Biophys Res Commun 22:488-491.

Stirling LC, Forlani G, Baker MD, Wood JN, Matthews EA, Dickenson AH, Nassar MA (2005) Nociceptor-specific gene deletion using heterozygous NaV1.8-Cre recombinase mice. Pain 113:27-36.

Tallent MK, Varghis N, Skorobogatko Y, Hernandez-Cuebas L, Whelan K, Vocadlo DJ, Vosseller K (2009) In vivo modulation of O-GlcNAc levels regulates hippocampal synaptic plasticity through interplay with phosphorylation. J Biol Chem 284:174-181.

Taylor RP, Geisler TS, Chambers JH, McClain DA (2009) Up-regulation of O-GlcNAc transferase with glucose deprivation in HepG2 cells is mediated by decreased hexosamine pathway flux. J Biol Chem 284:3425-3432.

109

Taylor RP, Parker GJ, Hazel MW, Soesanto Y, Fuller W, Yazzie MJ, McClain DA (2008) Glucose deprivation stimulates O-GlcNAc modification of proteins through up-regulation of O-linked N-acetylglucosaminyltransferase. J Biol Chem 283:6050-6057.

Tomlinson DR, Gardiner NJ (2008) Glucose neurotoxicity. Nat Rev Neurosci 9:36-45.

Torres C-R, Hart GW (1984) Topography and polypeptide distribution of terminal N- acetylglucosamine residues on the surfaces of intact lymphocytes. J Biol Chem 259:3308- 3317.

Trinidad JC, Barkan DT, Gulledge BF, Thalhammer A, Sali A, Schoepfer R, Burlingame AL (2012) Global identification and characterization of both O-GlcNAcylation and phosphorylation at the murine synapse. Mol Cell Proteomics 11:215-229.

Vincent AM, Brownlee M, Russell JW (2002) Oxidative stress and programmed cell death in diabetic neuropathy. Ann N Y Acad Sci 959:368-383.

Vincent AM, Calabek B, Roberts L, Feldman EL (2013) Biology of diabetic neuropathy. Handb Clin Neurol 115:591-606.

Walgren JLE, Vincent TS, Schey KL, Buse MG (2003) High glucose and insulin promote O- GlcNAc modification of proteins, including α-tubulin. Am J Physiol Endocrinol Metab 284:E424-E434.

Wang AC, Jensen EH, Rexach JE, Vinters HV, Hsieh-Wilson LC (2016) Loss of O-GlcNAc glycosylation in forebrain excitatory neurons induces neurodegeneration. Proc Natl Acad Sci U S A 113:15120-15125.

Wani WY, Chatham JC, Darley-Usmar V, McMahon LL, Zhang J (2016) O-GlcNAcylation and neurodegeneration. Brain Res Bull.

Watson LJ, Facundo HT, Ngoh GA, Ameen M, Brainard RE, Lemma KM, Long BW, Prabhu SD, Xuan Y-T, Jones SP (2010) O-linked β-N-acetylglucosamine transferase is indispensable in the failing heart. Proc Natl Acad Sci U S A 107:17797-17802.

Watson LJ, Long BW, DeMartino AM, Brittian KR, Readnower RD, Brainard RE, Cummins TD, Lakshmanan A, Hill BG, Jones SP (2014) Cardiomyocyte Ogt is essential for postnatal viability. Am J Physiol Heart Circ Physiol 306:H142-H153.

Whelan SA, Dias WB, Thiruneelakantapillai L, Lane MD, Hart GW (2010) Regulation of insulin receptor substrate 1 (IRS-1)/AKT kinase-mediated insulin signaling by O-linked β-N- acetylglucosamine in 3T3-L1 adipocytes. J Biol Chem 285:5204-5211.

Whisenhunt TR, Yang X, Bowe DB, Paterson AJ, Van Tine BA, Kudlow JE (2006) Disrupting the enzyme complex regulating O-GlcNAcylation blocks signaling and development. Glycobiology 16:551-563.

110

Woolf CJ, Ma Q (2007) Nociceptors – noxious stimulus detectors. Neuron 55:353-364.

Wright DE, Ryals JM, McCarson KE, Christianson JA (2004) Diabetes-induced expression of activating transcription factor 3 in mouse primary sensory neurons. J Peripher Nerv Syst 9:242-254.

Yang X, Zhang F, Kudlow JE (2002) Recruitment of O-GlcNAc transferase to promoters by corepressor mSin3A: Coupling protein O-GlcNAcylation to transcriptional repression. Cell 110:69-80.

Yang X, Ongusaha PP, Miles PD, Havstad JC, Zhang F, So WV, Kudlow JE, Michell RH, Olefsky JM, Field SJ, Evans RM (2008) Phosphoinositide signalling links O-GlcNAc transferase to insulin resistance. Nature 451:964-969.

Yang YR, Song M, Lee H, Jeon Y, Choi E-J, Jang H-J, Moon HY, Byun H-Y, Kim E-K, Kim DH, Lee MN, Koh A, Ghim J, Choi JH, Lee-Kwon W, Kim KT, Ryu SH, Suh P-G (2012) O-GlcNAcase is essential for embryonic development and maintenance of genomic stability. Aging Cell 11:439-448.

Yuzwa SA, Shan X, Macauley MS, Clark T, Skorobogatko Y, Vosseller K, Vocadlo DJ (2012) Increasing O-GlcNAc slows neurodegeneration and stabilizes tau against aggregation. Nat Chem Biol 8:393-399.

Zachara NE, Hart GW (2004) O-GlcNAc a sensor of cellular state: The role of nucleocytoplasmic glycosylation in modulating cellular function in response to nutrition and stress. Biochim Biophys Acta 1673:13-28.

Zachara NE, O'Donnell N, Cheung WD, Mercer JJ, Marth JD, Hart GW (2004) Dynamic O- GlcNAc modification of nucleocytoplasmic proteins in response to stress: A survival response of mammalian cells. J Biol Chem 279:30133-30142.

Zeidan Q, Wang Z, De Maio A, Hart GW (2010) O-GlcNAc cycling enzymes associate with the translational machinery and modify core ribosomal proteins. Mol Biol Cell 21:1922- 1936.

Zhang F, Su K, Yang X, Bowe DB, Paterson AJ, Kudlow JE (2003) O-GlcNAc modification is an endogenous inhibitor of the proteasome. Cell 115:715-725.

Zhu W, Leber B, Andrews DW (2001) Cytoplasmic O-glycosylation prevents cell surface transport of E-cadherin during apoptosis. EMBO J 20:5999-6007.

Zhu Y, Liu T-W, Madden Z, Yuzwa SA, Murray K, Cecioni S, Zachara N, Vocadlo DJ (2016) Post-translational O-GlcNAcylation is essential for nuclear pore integrity and maintenance of the pore selectivity filter. J Mol Cell Biol 8:2-16.

Ziegler D (2006) Treatment of diabetic polyneuropathy: Update 2006. Ann N Y Acad Sci

111

1084:250-266.

Zochodne DW (2008) Diabetic polyneuropathy: An update. Curr Opin Neurol 21:527-533.

112