Open CR-Thesis-Final-Final.Pdf

The Pennsylvania State University

The Graduate School

IDENTIFICATION OF GENETIC FACTORS THAT AFFECT NEURONAL

PATTERNING, FUNCTION, AND DISEASE IN DROSOPHILA MELANOGASTER

A Dissertation in

Biochemistry, Microbiology, and Molecular Biology

Claire Elizabeth Reynolds

Submitted in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

December 2017

The dissertation of Claire Elizabeth Reynolds was reviewed and approved* by the following:

Scott B Selleck Professor of Biochemistry & Molecular Biology Dissertation Co-Advisor Chair of Committee

Santhosh Girirajan Assistant Professor of Biochemistry & Molecular Biology Assistant Professor of Anthropology Dissertation Co-Advisor

Joseph Reese Professor of Biochemistry & Molecular Biology

Marylyn Ritchie Director, Center for Systems Genomics Director, Biomedical & Translational Informatics Program of Geisinger Research Professor of Biochemistry & Molecular Biology

Richard Ordway Professor of Molecular Neuroscience and Genetics

Wendy Hanna-Rose Associate Professor of Biochemistry & Molecular Biology Interim Head of the Department of Biochemistry and Molecular Biology

*Signatures are on file in the Graduate School

ABSTRACT

Heparan Sulfate Proteoglycans (HSPGs) are required for normal synaptic development at the Drosophila melanogaster larval neuromuscular junction (NMJ).

When enzymes required for biosynthesis of HSPGs are inhibited through mutations of

RNA interference, a variety of morphological and electrophysiological defects are observed at the NMJ. These defects included changes in the post-synaptic specialization of the muscle (the SSR), loss of mitochondria from the sub-synaptic cytosol, and abnormal mitochondrial morphology. Identification of autophagic regulation as the mechanism by which HSPGs influenced synaptic properties was the foundation of this dissertation.

The present work more fully characterizes the influence of HSPG function on autophagic markers in muscle tissue. Confocal microscopic and Western blot analysis of both endogenous proteins and ectopically expressed reporter constructs was used to assess the level of autophagic degradation in muscle tissue. RNA inhibition of sfl and ttv, two HS biosynthetic enzymes, influenced these measures of autophagy in a direction that indicated an overall increase in autophagic flux.

HSPGs are ubiquitously expressed, and have a variety of critical functions throughout the organism. Similarly, autophagy is a globally important catabolic pathway.

To better understand the significance of this novel regulatory role for HSPGs, we extended our analysis into a second tissue. Drosophila fat body tissue is an energetically sensitive tissue in which autophagy is commonly studied. Using many of the same assays, we identified a similar increase in autophagy in response to HSPG biosynthetic inhibition that was non-cell autonomous.

iii

While autophagy is important for the maintenance of cellular homeostasis throughout the organism, certain cell types are more reliant on this pathway than others.

Neurons are one such autophagy sensitive cell type, and failure of neuronal autophagy is closely associated with neurodegenerative disease. When HSPGs were inhibited in the brain of adult flies, an increase in autophagic degradation was once again identified.

This increase in autophagy protected against the accumulation of oxidant-damaged cellular components after environmental exposure to hydrogen peroxide, and reduced neuronal death in a model of Alzheimer’s disease. These findings indicate that the regulation of autophagy by HSPGs is of high clinical relevance to human health.

This thesis also describes an attempt to develop a Drosophila behavioral assessment protocol intended to provide biological validation of Autism Spectrum

Disorder (ASD) candidate genes. Sequencing and microarray analysis of individuals with

Autism spectrum disorders has implicated an overwhelmingly large pool of candidate genes. Thorough assessment of the functional contributions of these genes to neurobehavioral development in model organisms is most drastically hindered by the generally low rate of recurrence in genetic changes affecting each candidate gene.

Behavioral screening in the inexpensive and genetically facile model D. melanogaster was attempted using RNA interference (RNAi) to model genetic changes associated with human ASD and identify high-throughput behavioral methods suitable for providing the lacking biological validation for these genetic models. Three behavioral methods were selected based on their compatibility and straightforwardness, with consideration for evidence of previous successful use in established fly models of

Autism. While the pilot run of this study was able to identify a few phenotypes of

potential interest, the majority of the data indicated that these methods were not well- suited to the use of RNAi.

TABLE OF CONTENTS

List of Figures ...... ix

List of Abbreviations ...... xi

Acknowledgements ...... xiii

Chapter 1 Heparan Sulfate Proteoglycans are required for Basal Regulation of Autophagy in Drosophila ...... 1

1.1 Abstract ...... 1 1.2 Introduction ...... 2 1.21 Heparan Sulfate Proteoglycans: An overview ...... 2 1.22 Autophagic degradation ...... 8 1.23 Studying HSPGs in Drosophila ...... 12 1.24 The requirement for HSPGs at the Drosophila Neuromuscular Junction ...... 16 1.3 Results ...... 24 1.31 Endogenously expressed autophagy substrates are upregulated in muscle with impaired HS function ...... 24 1.32 The degradative capacity of HS inhibited muscle tissue is not reduced...... 26 1.33 HS Biosynthetic inhibition increases levels of autophagy markers in fat body tissue ...... 30 1.34 HS-mediated autophagic regulation is non-cell autonomous ...... 33 1.4 Discussion ...... 35 1.6 Materials and Methods ...... 40 1.61 Real-time quantitative polymerase chain reaction ...... 40 1.62 Immunohistochemistry and LysoTracker Red staining ...... 41 1.63 Mosaic analysis ...... 43 1.64 Image acquisition and analysis ...... 43 1.65 Western blot ...... 44 1.66 Statistical analysis ...... 45

Chapter 2 Heparan Sulfate Proteoglycans Regulate Basal Autophagy and Stress Tolerance in the Central Nervous System...... 46

2.1 Abstract ...... 46 2.2 Introduction ...... 47 2.21 Autophagy-dependent synaptic sculpting in neurodevelopmental disorders ...... 48 2.22 Contributions of autophagy to neurodegenerative disease processes ...... 49 2.23 HSPGs in neuronal cell death ...... 51 2.3 Results ...... 54 2.31 HSPG biosynthetic inhibition enhances neuronal autophagy and improves clearance of oxidant damaged cellular components ...... 54

2.32 HSPG functional levels influence survival of chronic oxidant stress and age-associated proteostasis...... 58 2.33 Inhibition of HSPG function ameliorates neuronal death in a Drosophila model of familial Alzheimer’s disease ...... 64 2.4 Discussion ...... 68 2.6 Materials and Methods ...... 76 2.61 Head Collection ...... 77 2.62 Protein Purification ...... 78 2.63 Western Blot ...... 79 2.64 Oxidant Exposure ...... 79 2.65 Disease interaction studies ...... 80 2.66 Statistical Analysis ...... 81

Chapter 3 Inert Genetic Constructs Influence Behavior in Drosophila ...... 82

3.1 Abstract ...... 82 3.2 Introduction ...... 83 3.21 Autism Spectrum Disorder: the face of neurodiversity ...... 84 3.22 Neurobehavioral modeling in Drosophila melanogaster ...... 86 3.33 Assembling a Drosophila behavioral screen for ASD candidate gene validation ...... 88 3.3 Results ...... 94 3.31 The Drosophila Activity Multibeam Monitor: a high resolution method to track spontaneous locomotor activity and rhythmicity ...... 94 3.32 Startle-inducted negative geotaxis: an indicator of danger responsiveness and locomotor capability...... 97 3.33 The Social Spacing Index- a measure of preferred flocking distance within Drosophila social groups...... 99 3.4 Discussion ...... 102 3.6 Materials and methods ...... 111 3.61 Circadian Rhythm: ...... 111 3.62 Startle Induced Negative Geotaxis: ...... 112 3.63 Social Spacing ...... 113

Chapter 4 Future Research Directions ...... 116

4.1 Ongoing Research on the HSPG Project ...... 116 4.2 Future Aims for the HSPG project ...... 120 4.3 Problematic Design Features of the Drosophila Screen for Biological Validation of ASD Candidate Genes...... 123

References ...... 127

Appendix A PI3K Overexpression suppresses autophagic induction ...... 145 Appendix B qPCR Data ...... 146 Appendix C Supplementary western blot data ...... 148 Appendix D Enlarged images from Fig. 2.06 ...... 151 Appendix E Summary of the mucopolysaccharidoses and related disorders ... 153

vii

Appendix F Publication ...... 154

viii

LIST OF FIGURES

Figure 1.01: Cell surface and secreted heparan sulfate proteglycans ...... 3

Figure 1.02: Heparan sulfate glycosaminoglycan synthesis and modification ...... 7

Figure 1.03: Autophagosome biogenesis ...... 9

Figure 1.04: The organization and innervation of larval body wall muscles ...... 14

Figure 1.05: Changes in SSR structure, mitochondrial number and morphology in HS biosynthesis compromised larval muscles ...... 18

Figure 1.06: Accumulation of membrane-bound structures in muscle cells of larvae with compromised HS biosynthesis ...... 20

Figure 1.07: SSR phenotypes associated with impaired HS modification are suppressed by down-regulation of autophagy ...... 23

Figure 1.08: Compromised HS biosynthesis leads to increased levels of autophagy indicator proteins in larval muscle ...... 25

Figure 1.09: Accumulation of autophagic structures in muscle cells of larvae with compromised HS biosynthesis ...... 27

Figure 1.10: Reductions in HS biosynthesis and modification result in a net increase in autophagic flux ...... 29

Figure 1.11: Inhibition of HS biosynthesis in fat body cells induces autophagy in wandering 3rd instar larvae ...... 32

Figure 1.12: Homozygous mutant ttv00681 causes a cell non-autonomous increase in the number of acidic organelles ...... 34

Figure 2.01: Pan-neuronal HSPG biosynthetic inhibition increases autophagic turnover of an mCherry-Atg8a fusion protein ...... 55

Figure 2.02: Inhibition of HSPG function in neurons reduced the accumulation of autophagic substrates following environmental exposure to hydrogen peroxide ...... 56

Figure 2.03: Survival of heterozygous and homozygous mutants under continuous oxidative stress ...... 60

Figure 2.04: Loss of sulfatase activity dramatically increases age-associated autophagic senescence ...... 63

Figure 2.05: Loss of HSPG function partially rescues neurodegeneration in a model of Alzheimer’s disease ...... 66

Figure 2.06: Determining the balance between pro-growth and pro-survival pathways, and between the generation and clearance of ubiquitinated substrates ...... 70

Figure 2.07: Deletion of sumf mimics a wing developmental error found in animals lacking the HSPG negative regulatory sulfatase gene sulf1 ...... 72

Figure 3.01: Circadian Activity pattern of Drosophila melanogaster ...... 91

Figure 3.02: Multiple elements of the binary expression exert isolated effects on spontaneous activity levels ...... 95

Figure 3.03: Overexpression of ASD candidate gene Rheb reduced activity ...... 95

Figure 3.04: Circadian rhythmicity is disrupted by RNAi of an ASD candidate gene ...... 96

Figure 3.05: Off- and On-target genetic influences on startle induced negative geotaxis ...... 98

Figure 3.06: Normal flocking distances in the social spacing assay ...... 100

Figure 3.07: Performance in the social spacing assay is variable in stocks bearing inserted genetic elements ...... 101

Figure 4.01: sumf loss of function alters the quantity and sulfation level of HSPGs in Drosophila ...... 117

LIST OF ABBREVIATIONS

Aβ amyloid beta AD Alzheimer’s disease ADHD Attention deficit hyperactive disorder AKT Protein kinase B AMPK AMP-activated protein kinase AP Autophagosome ASD Autism spectrum disorder Atg3 Autophagy related 3, D.melanogaster gene Atg7 Autophagy related 7, D.melanogaster gene Atg8a Autophay related 8a, D.melanogaster gene, homolog of LC3 CNS Central nervous system CS Chondroitin sulfate DAMM Drosophila activity multibeam monitor DcrII Dicer-II, D.melanogaster gene DJ-1 Parkinsonism associated deglycase, human gene dlp dally-like, D.melanogaster gene ECM Extracellular matrix FGF Fibroblast growth factor FGFR Fibroblast growth factor receptor Fmr1 Fragile X mental retardation 1, D.melanogaster gene fz2 frizzled 2, D.melanogaster gene GAG Glycosaminoglycan gbb glass bottom boat, D.melanogaster gene GPI Glycosylphosphatidylinositol HGF Hepatocyte growth factor Hh Hedgehog HS Heparan sulfate Hs6st Heparan sulfate 6-O sulfotransferase, D.melanogaster gene HSPG Heparan sulfate proteoglycan IR Infrared IUP Insoluble ubiquitinated particle

JNK c-Jun N-terminal Kinase LRKK2 Leucine rich repeat kinase 2, human gene LSD Lysosomal storage disease Mad Mothers against decapentaplegic, D.mel protein MPS Mucopolysaccharidosis MPSPS Mucopolysaccharidosis plus syndrome MSD Multiple sulfatase deficiency mTORC1/mTOR mechanistic target of rapamycin containing complex 1 NMJ Neuromuscular junction OreR OreR, a control fly stock PD Parkinson’s disease PE Phosphatidlyethanolamine PG Proteoglycan pgant5 polypeptide GalNAc transferase 5, D.melanogaster gene Pten Phosphatase and tensin homolog, D.melanogaster gene Rheb Ras homolog enriched in brain, D.melanogaster gene sfl sulfateless, D.melanogaster gene SING Startle-induced negative geotaxis SSI Social spacing index SSR Subsynaptic reticulum Sulf1 Sulfated, D.melanogaster gene sumf (cg7049) D.melanogaster homolog of human SUMF1 SUMF1 Sulfatase modifying factor 1, human protein Tau Microtubule associated protein Tau TEM Transmission electron microscope TGFβ Transforming growth factor beta TSC1 Tuberous sclerosis complex 1, human gene ttv tout-velu, D.melanogaster gene VEGF Vascular endothelial growth factor w (w*, w+) white, D.melanogaster gene. *indicates mutation of unknown allele, +indicates a wildtype transcript wg wingless, D.melanogaster gene wit wishful thinking, D.melanogaster gene

xii

ACKNOWLEDGEMENTS

I would like to thank all of the individuals who assisted me in the completion of this work: my lab-mates, my committee members, and most especially my year-mates, the BMMBMMB (and honorary members). I have never been randomly assorted to such a wonderful group of humans. It has been a long journey full of difficulty, but thanks to your support, it has also been full of love, happiness, and belonging. Special thanks are due to my husband, who has forgiven me time and time again when I have put my love for science before my love for him, and who has cheered and challenged and soothed me through lo these many years of struggle.

I wouldn’t even be here writing an acknowledgment section for my thesis (let alone the rest of my thesis) if I hadn’t had such a wonderful, compassionate, and challenging mentor. Scott, I look up to you so much. I hope I can be just like you when I grow up, but maybe a little less silly. Thank you for letting me make my mistakes and for teaching me to believe in myself. Thank you for turning me from someone who just wanted to be told what to do into someone who asks her own questions.

Of course, I have to acknowledge my family. I am incredibly privileged to have been raised by such intelligent and forceful parents. You always told me to be exactly what I am and nothing less. I finally found the place where I belong, and I’m so glad I listened to you. Thanks are also due to my siblings, whom I will always love best. Neil, my brother in truth, and Cathryn, my sister in every way that matters, my very dearest friend: the two of you are my heart.

Finally, I would like to thank Rodrigo, Rob, Brian, Sam, Stephen, Adriana, and

Matthew for telling the most wonderful stories and for keeping me company through many long and lonely nights in the lab. Thank you for the beautiful work you do.

xiii

Chapter 1

Heparan Sulfate Proteoglycans are required for Basal Regulation of Autophagy in Drosophila

1.1 Abstract

Heparan Sulfate proteoglycans (HSPGs) are critically important regulators of developmental patterning and cell-signaling, found in the extracellular matrix and at the cell surface. Study of mutations and RNA interference (RNAi) of HS biosynthetic enzymes sfl, the Drosophila heparan sulfate N-deacetylate/N-Sulfotransferase, and ttv, a homolog of the mammalian Exostoses 1 heparan sulfate co-polymerase, at the neuromuscular junction (NMJ) of Drosophila melanogaster larva revealed a variety of structural and electrophysiological defects at and around the synapse. Transmission

Electron Microscopy (TEM) analysis of the subsynaptic reticulum (SSR), the postsynaptic specialization of the NMJ, and the underlying cytoplasm indicated that these phenotypes may arise from a disruption of autophagy in the muscle. Tracking of fluorophore-tagged Atg8a in HSPG inhibited muscle revealed changes consistent with increased autophagic degradation. Additional methods of examining autophagic flux in muscle supported the conclusion that HSPGs negatively regulate autophagy. HSPG function was also found to regulate autophagy in larval fat body tissue, the central hub of nutritional sensing and energy storage in Drosophila. In mosaic fat body mutants, HSPG regulation of autophagy was found to operate cell-non autonomously, suggesting a regulatory mechanism dependent on cell-surface signaling that can be activated in trans.

These findings indicate that the effects of HS biosynthetic inhibition at the NMJ function

at least partially through upregulation of autophagy and that this novel regulatory role for

HSPGs is conserved in the fat body tissue as well.

1.2 Introduction

1.21 Heparan Sulfate Proteoglycans: An overview

Proteoglycans (PGs) are a ubiquitously expressed family of proteins defined by the covalent attachment of one or more glycosaminoglycan (GAG) polymers to a core protein1 (Fig. 1.01). The GAGs are long unbranched polysaccharide chains formed from repeating disaccharide units composed of a hexosamine sugar and either uronic acid or galactose. GAG chains are categorized into subfamilies based on the composition of their constituent disaccharides. Heparan sulfate (HS) is one of these GAG subtypes, defined compositionally by the use of disaccharides made from N-acetylglucosamine

(GlcNAc) and glucuronic acid (GlcA)2. Heparan sulfate proteoglycans (HSPGs) are ubiquitously expressed and are found primarily at the cell surface and in the extracellular matrix.

There are three main core proteins that receive HS modifications; syndecan, glypican, and perlecan. Syndecan is a single pass transmembrane protein that can be modified by both HS (shown in blue, Fig. 1.01) and chondroitin sulfate, another GAG subtype (CS, shown in yellow, Fig 1.01)1. Glypicans are linked to the outer leaflet of the plasma membrane by a glycosylphosphatidylinositol (GPI) anchor, and receive only HS

GAGs3. Perlecan is a very large secreted HSPG, found primarily in basement membrane and pericellular spaces4, 5.

The cell surface HSPGs play a crucial role in cellular signaling, particularly during development. Wnt, Hedgehog (Hh), transforming growth factor-β (TGFβ1 and 2), fibroblast growth factors (FGFs) and their related receptors (FGFRs), growth factor binding proteins, hepatocyte growth factor (HGF), and vascular endothelial growth factor

(VEGF) are all known to require HS binding for their proper functioning. A wide variety of cytokines and chemokines are also functionally dependent on HS binding6.

Figure 1.01: Cell surface and secreted heparan sulfate proteoglycans

HSPGs consist of a core protein, (brown), which can be modified by one or more heparan sulfate glycosaminoglycan chains (blue). Some HSPGs can also receive chondroitin sulfate GAGs (yellow). The GAG domains of these proteins function in the extracellular space. Glypicans and syndecans are attached to the plasma membrane (light green), while perlecans are secreted into the extracellular space. Modified from Fig. 16.2 : Esko et. al. (2009) Essentials of Glycobiology 2nd Edition

Signaling through Heparan Sulfate Proteoglycans

HSPGs interact with signaling molecules through a variety of mechanisms, some better understood than others. One of the most developmentally critical functions of

HSPGs involves the establishment of morphogen gradients. Morphogens are soluble molecules secreted by cells at a distance from their targets that enact developmental programs such as cell fate determination7. Morphogen gradient establishment is thought to occur by the rapid sequestration of secreted morphogens into the GAG chains of cell surface and matrix HSPGs near the site of release to restrict diffusion8, 9. This process creates a steep gradient where morphogen concentrations are high near the site of release and quickly taper off distal to that source10.

Bound morphogens can be presented to receptors that also interact with HS or brought into the cell on PGs through endocytosis11-13. The specific contribution of HSPG binding to Wnt, Hh, and BMP gradients during tissue patterning and cell fate determination has been extensively characterized in Drosophila. Due to the increased complexity of HS core protein families and additional functional redundancy with other types of PGs, these processes have not been as thoroughly described in vertebrates.

Interactions between HSPGs and their ligands occur largely through electrostatic interactions based on the highly negative sulfated domains14, 15, however, these domains are relatively small (3-10 disaccharides) in comparison to the total length of the GAG chains (40-160 disaccharides)16. HS chains consist of long unsubstituted regions, theorized to provide flexibility, interspersed by short, highly modified regions and regions of mixed unmodified and lightly modified residues. While the pattern, length, and composition of these functional domains are highly conserved with the cell-type of origin, there is a great deal of variability in the modifications that occur between different

4 tissues17, 18. Cell type-specific sulfation patterns and disaccharide prevalence are likely due to a combination of cell-specific core protein expression levels and the concentration and location of the various biosynthetic enzymes that participate in chain elongation and modification19.

Tightly controlled cell type-specific patterning of sulfated domains on the GAG chain contribute heavily to the specific binding capabilities of individual HSPGs. 6-O sulfation in particular plays an important role in interaction between HS and a number of its ligands, including FGFs 20-22, HGF 17, 20, and VEGF 23, and PDGF 24. Importantly, 6-O sulfate groups on HSPGs are regulated not only during initial biosynthesis of the molecule in the golgi, but at the cell surface as well. The cell-surface Sulf enzymes, a unique class of sulfatases, remove select 6-O sulfate groups from the GAG without otherwise disrupting the structure of the polysaccharide. Like other sulfatases, the Sulfs possess a highly conserved cysteine bearing active site which receives a unique α- formylglycine modification25 required for their activity. However, outside of the catalytic domain, Sulfs are structurally dissimilar to other sulfatases and perform optimally at neutral, rather than acidic pH26, 27.

Sulf activity dynamically regulates HS-dependant signaling, promoting Wnt,

BDNF, BMP, and Hh signaling28-31 and inhibiting the activity of FGFs, VEGF, and

TGFβ32-35. In most higher eukaryotes, Sulfs and other sulfatases become active after receiving the post-translational α-formylglycine modification of a cysteine residue in the active site. This modification is only found in sulfatases, and is generated by a single enzyme, SUMF1 (sulfatase modifying factor 1)36, 37. SUMF1 is highly conserved throughout phyla within bilateria, and loss of enzyme function is incompatible with life in mammals. Partial loss of function is associated with the human disease multiple sulfatase deficiency (MSD), a lethal lysosomal storage disease37, 38.

As the master regulator of sulfatase activity, loss of SUMF1 activity results in cessation of function in all cell surface and lysosomal sulfatases. Inactivation of sulfatases causes failures of Sulf-orchestrated negative regulation of HSPGs at the cell surface and of their degradation in the lysosome, resulting in increased HSPG activity.

This can be seen in the form of hyperactive FGF signaling in SUMF1 null mice39.

Molecular hallmarks of MSD include global suppression of autophagy, accumulation of un-degraded, highly sulfated GAGs in lysosomes, and secondary accumulation of most other lysosomal degradation substrates40-42. This leads to skeletal and cartilaginous defects, organomegaly, and neurodegeneration. Secondary accumulation of substrates is thought to be the cause of cellular death in MSD and other lysosomal storage diseases; however, the cause of secondary accumulation is not understood. In the case of MSD specifically, one current theory proposes that HSPG over-activity may lead to an inhibition of autophagy that results in secondary accumulation and cell death in autophagy-sensitive tissues42.

Biosynthesis of Heparan Sulfate Proteoglycans

Synthesis and modification of GAG chains is performed by a series of enzymes in the Golgi. For HSPGs, this process is initiated by attachment of xylose to a serine of the core protein. Two galactose residues and a glucuronic acid are added to the xylose, after which the tetrasaccharide linker is complete and GAG synthesis may begin.

Transfer of N-acetylglucosamine to the tetrasaccharide linker commits that modification site to polymerization of a heparan sulfate GAG. GAG chain polymerization and modification proceeds from this point in a stepwise fashion, as shown in Fig. 1.02.

The HS GAG itself is composed of alternating glucosamine derived amino sugars and uronic acids that form disaccharide units. These units can then be modified individually by a variety of enzymes into at least 23 known distinct and naturally occurring disaccharides6. HS constituent disaccharides are made from combinations of

N-acetylglucosamine (GlcNAc), variably O-sulfated N-unsubstituted glucosamine (GlcN), or N-sulfated glucosamine (GlcNS) paired with either D-glucuronic acid (GlcA) with or without C-2 sulfation or L-iduronic acid (IdoA) with or without C-2 sulfation.

Figure 1.02: Heparan sulfate glycosaminoglycan synthesis and modification

N-GlcNAc transferase I attaches the first N-acetylglucosamine (GlcNAc, diamonds) to the tetrasaccharide linker, after which N-GlcNAc co-transferases, including ttv, and GlcA transferase extend the chain with additional GlcNAc and glucuronic acid residues (GlcA, hexagons). Chain modification begins with N-sulfation (orange) of GlcNAc performed by the N-deacetylase/N- sulfotransferase performed by sfl. Subsequent 2O- (green), 3O- (pink), and 6O-sulfations (blue) and epimerization of GlcA to iduronic acid (IdoA, circles) are dependent on completion of the previous step. Modified from Figure 1: Reynolds-Peterson et. at. 2017 7

1.22 Autophagic degradation

Macroautophagy, hereafter referred to simply as autophagy, is the process by which bulk cytoplasmic components, organelles, and proteinaceous aggregates are recycled as part of normal cellular metabolism and in response to cellular stresses such as starvation43 and oxidative stress44. Autophagy was originally discovered as a response to starvation, in which a wide variety of cellular components are catabolized for energy and components in what appeared to be an indiscriminate manner45. Many years of intense study have elaborated on this simple interpretation of autophagy, to what is known understood to be a multifaceted process regulated by several complex signaling arms.

Autophagic regulation

Nutritional sensing is the primary regulatory input in autophagic regulation. The mechanistic Target of Rapamycin Complex 1 (mTORC1), a serine/threonine protein kinase, acts as the central hub for this canonical autophagy regulation pathway. mTORC1 integrates inputs from growth factor signaling through protein kinase B (Akt)46, amino acid sufficiency and ATP/AMP sensing through AMP-activated Protein Kinase

(AMPK)47, 48, and inputs from cellular stress sensing pathways in order to allow an appropriate level of autophagy. The level of autophagy required to maintain homeostasis in the absence of pro-autophagy stressors such as starvation or hypoxia is referred to as

‘basal’ autophagy. The level of basal autophagy varies between cell types and decreases with increasing age49.

Autophagy is induced to protect cells for a wide variety of toxic insults. Hypoxia, oxidative stress, and DNA damage all induce autophagy through AMPK-mediated mTOR inhibition50-53. Proteasomal inhibition and ER stress induce a similar protective autophagic response through c-Jun N-terminal kinase (JNK) mediated inhibition of mTOR inhibition54-56. Direct inhibition of mTOR by pharmaceutical intervention is sufficient to induce autophagy in a wide variety of cultured cells and model organisms57-

60. Autophagy also responds to a variety of mTOR-independent inputs, including changes in intracellular levels of Ca2+, inositol, and cyclic AMP61.

Figure 1.03 Autophagosome biogenesis

In response to pro-autophagic signals, Atg8a becomes PE-conjugated to an extending double layered membrane projection, the phagophore. Bulk cytoplasmic and chaperone-targeted cellular components are encircled as the phagophore elongates, enclosing these substrates within a lumen. The fully enclosed autophagosome then fuses with a lysosome to become a mature autolysosome. The contents of the autophagosome are degraded and autophagosomal proteins, including LC3, are recycled from the outer membrane of the autolysosome. Inset: magnified view of two lipid bilayer membranes. Modified from Figure 1: Jing and Lim 2012

Mechanistic details of the autophagy pathway

The process of autophagic degradation exhibits a degree of mechanistic complexity that is at least as complicated as its regulatory inputs. Over 30 numbered autophagy-related (atg) proteins have been identified in yeast, many of which are conserved in eukaryotes62. These atg proteins work together with SNAREs (Soluble NSF

Attachment Protein Receptors), ESCRT (Endosomal Sorting Complexes Required for

Transport) machinery, and VPS (Vacuolar Protein Sorting) proteins as well as a number of adaptors and chaperones throughout the process of autophagosome formation, substrate loading, and maturation62.

The first step of autophagy involves nucleation of the isolation membrane or phagophore, a two-layered membranous extension (Fig. 1.03). Multiple possible membrane donors exist for phagophore initiation, including ER, golgi, and the mitochondrial outer membrane63. As the isolation membrane is extended, autophagy related proteins including Atg8a and Atg12 are attached to both faces of the phagophore by phosphatidylethanolamine (PE) linkage. Atg8a and Atg12 aid in elongation of the isolation membrane and substrate loading. Autophagy substrates are gathered into the lumen of the phagophore, and complete engulfment by the isolation membrane produces a mature autophagosome (AP). APs go on to fuse with acidic organelles, primarily lysosomes, to degrade their contents64.

Starvation induces formation of autophagosomes that engulf of bulk cytoplasmic components in a nonspecific manner, but the majority of autophagic degradation involves targeted loading of specific substrates. Selective autophagy is responsible for regulation of organelle number and quality through the processes of mitophagy, pexophagy, reticulophagy, and ribophagy (targeted degradation of mitochondria,

10 peroxisomes, endoplasmic reticulum, and ribosomes respectively)65. Clearance of intracellular aggregates (aggrephagy), pathogens (xenophagy) and endocytosed materials is also managed through selective autophagy63, 66, 67.

Selective loading of targeted autophagy substrates is accomplished through binding of autophagy receptors to members of the Atg8/LC3 protein family at LC3

Interacting Regions (LIRs)68. A few dozen such LIR-containing receptors are known to exist in mammalian cells69, 70, of which the most commonly studied is p62. By definition, autophagy receptors must bind both to phagophore-conjugated membrane proteins and autophagic substrates. p62 binds to Atg8a with a LIR domain, while binding polyubiquitinated substrates with a ubiquitin-associated (UBA) domain71-73. While other autophagy receptors can load substrates using this mechanism, p62 also has the unique ability to cluster misfolded proteins and defective mitochondria through self- polymerization in preparation for autophagosomal loading74.

A number of mechanisms exist by which autophagy receptors identify and interact with the substrates they recognize, but ubiquitin modification is the most common binding partner in these interactions. Ubiquitination of mitochondrial surface proteins, bacterial pathogens, peroxisomes, and some cytosolic proteins is both necessary and sufficient to commence their autophagic degradation75. Forced mono- ubiquitination of soluble red fluorescent protein (RFP) and peroxisome surface membrane proteins is sufficient to direct these substrates into autophagosomes75.

Polyubiquitin chains linked at K63 and K27 are required for p62-mediated mitophagy following membrane depolarization76. Beyond these findings, it is still unclear which forms of ubiquitination specifically route substrates to the autophagic degradation pathway.

1.23 Studying HSPGs in Drosophila

Drosophila melanogaster is an invaluable model organism that has been used extensively to study the genetic underpinnings of development and the function of

HSPGs. The accessibility of the genetic tools available for use in Drosophila is exceptional in comparison to other common multicellular model organisms. RNA interference (RNAi) methods are very efficient in the tissues of Drosophila, and libraries of RNAi constructs targeting every gene in the D. melanogaster genome are readily available77, 78. The expression of these constructs is controlled ectopically by the yeast transcription factor GAL4, which has been placed under the regulatory control of a wide variety of promoters to allow study of genetic inhibition in highly controlled patterns79, 80.

Efficiency of GAL4 function varies with temperature, allowing for the use of differing rearing temperatures to exert fine control over the expression level of constructs under control of the Upstream Activator Sequence (UAS, a cis-acting regulatory sequence) recognized by GAL479, 80. Furthermore, the GAL4-UAS binary expression system allows for severe inhibition of genes in a tissue specific manner that, if functionally inhibited to a similar degree ubiquitously by a traditional mutation, would result in organismal lethality. Total inhibition of any non-redundant HS biosynthetic enzyme is lethal due to the critical importance of these molecules in early developmental patterning, although maternal contribution of functional mRNA allows null mutant larvae to survive to early pupal phase at the latest. Maternal contribution of functional HS biosynthetic components lessens the effect of mutations, so tissue-restricted RNAi function is necessary both to allow organismal survival and to efficiently remove all transcripts of both maternal and zygotic origin during development.

The facile nature of Drosophila genetics is complemented by their rapid lifecycle, ease of care, and a degree of biological complexity that is sufficient to study the vast majority of cell types and functional systems found in vertebrates. All of the major signaling pathways are fully conserved in Drosophila, and many were initially discovered in this model81, 82. In larvae particularly, there are a wide variety of tissues and organs for which the structure and development are highly regular and well-described, and which can be readily whole-mounted for confocal imaging of both immunofluorescently-stained and ectopically expressed fluorescent markers. Larval tissues therefore present an ideal model in which to study the requirement of HSPGs in cell-cell contacts and signaling.

The larval neuromuscular junction (NMJ), where motor neurons extended from the ventral nerve chord interface with body wall muscles, is one such specialized model tissue commonly used to study signaling and synaptic development.

The larval neuromuscular junction as a model of cell communication and postsynaptic structural development

Larval neuromuscular junctions are formed by 32 fully characterized motor neurons that innervate the 30 body wall muscles repeated in every abdominal hemisegment (Fig. 1.04). Where the axon terminal boutons make contact with the muscle cell, an elaborate membranous structure called the subsynaptic reticulum (SSR) is formed. The SSR is the post-synaptic specialization of the NMJ, and throughout embryonic and larval development, it grows proportionally with the muscle until the axon terminal boutons are fully embedded within it83, 84. Both the morphological and electrophysiological properties of these synapses are highly stereotyped84-87.

Figure 1.04 The organization and innervation of larval body wall muscles

The body of Drosophila larvae is broken into distinctive sections called hemisegments. A Strap muscles of the larval body wall. The boundaries between most hemisegments are easily identified by examining the end points of the muscles (indicated by white bars to the right). Hemisegments are categorized as thoracic (T) or abdominal (A). B In hemisegments A2-A7, the same 30 muscles occur and are organized in an identical pattern. Each strap muscle in each hemisegment is innervated by a developmentally predetermined motor neuron. ISN, intersegmental nerve; SN, segmental nerve; B hemisegment orientation rotated 90° counter- clockwise from orientation in A. (A) Modified from: H. Bellen et al 2000 Drosophila Protocols (B) from Figure 2: Kohsaka et al 2012

Rather than using choline, the primary excitatory neurotransmitter of mammalian

NMJs, the Drosophila NMJ is glutamatergic, rendering this model synapse more similar to mammalian central nervous system (CNS) synapses88. However, unlike mammalian

CNS synapses, each Drosophila NMJ is formed by uniquely identifiable motoneurons that invariably locate and synapse upon a developmentally pre-defined muscle (Fig.

1.04)89, 90. These features of the NMJ allow for powerful genetic investigation of the requirements for glutamatergic synapse development and function in an easily accessible, identifiable, and stereotypical synaptic terminal.

Morphogen signaling controls NMJ developmental and activity-dependent plasticity

The NMJ is a highly plastic synapse, exhibiting activity-dependent structural and functional remodeling during development and throughout the lifetime of the animal.

Plasticity is seen in both pre- and post-synaptic NMJ structures in response to signals traveling bi-directionally across the synapse91, 92. Anterograde signaling (from neuron to muscle) is conducted by the Drosophila Wnt, wingless (wg). wg produced by the neuron is released into the synaptic cleft by axon terminal boutons and activates its receptor, frizzled 2 (fz-2), at the surface of the apposed muscle cell. wg-mediated activation of fz-

2 leads to cleavage and nuclear import of its C-terminal domain93. In the muscle cell, wg signaling initiates expansion of the post-synaptic specialization accompanied by clustering of scaffolding proteins and glutamate receptors94. Pre-synaptic wg is also required for assembly of the axonal active zone, the site of neurotransmitter release, and the growth and budding of terminal boutons as the animal matures.

Retrograde signaling (from muscle to neuron) by the BMP 4/5/6 homolog glass bottom boat (gbb) via its neuronal receptor wishful thinking (wit) is required alongside anterograde wg signaling for pre-synaptic expansion and neurotransmission at the

Drosophila NMJ95, 96. Assembly of active zones and maintenance of the close membrane apposition between the pre- and post-synaptic cells is dependent on retrograde transport of BMP through activation of its effector protein, the transcription factor Mothers against decapentaplegic (Mad)97, 98. Continuous release of gbb from the post-synaptic muscle cell is required to maintain normal neurotransmission99.

HSPGs play an enormous role in these trans-synaptic signaling pathways. Both wg and gbb rely on HS co-receptors present in the synaptic cleft to bind their receptors.

Alteration of HS biosynthesis through mutations in Hs6st (the Drosophila Heparan sulfate 6-O transferase) or loss of HS negative regulation through mutations in Sulfated

(Sulf1, the sole Drosophila extracellular 6O-endosulfatase) disrupts these signaling pathways, causing defects in both synaptic morphology and function100, 101. These functions are orchestrated by the different HSPGs present at and around the NMJ.

1.24 The requirement for HSPGs at the Drosophila Neuromuscular Junction

In Drosophila, there is a single syndecan (sdc, syndecan) and two glypicans

(dally, division abnormally delayed, and dlp, dally-like protein), as well as a single perlecan (troll, terribly reduced optic lobes). All three HS core proteins are found in large quantities around the synapse at the NMJ102, 103. sdc and dlp impact primarily pre- and postsynaptic development respectively104, while trol regulates bidirectional wg signaling to effect both pre- and postsynaptic structures103. When the functions of these PGs are concomitantly inhibited at the NMJ by null mutations in sfl (sulfateless, the Drosophila N-

16 deacetylase-N-sulfotransferase) or ttv (tout-velu, the Drosophila EXT1 homolog), affected animals exhibited changes in synaptic transmission and impaired locomotion102.

This was likely associated with the reduced number of axon terminal boutons at the NMJ and the increased activity-dependent endocytosis at the presynaptic terminal observed in these animals. Significant clearance of mitochondria from the subsynaptic muscle cytosol and a partially penetrant disorganization of the SSR were observed in these animals as well.

To better understand the mechanisms linking impaired HS biosynthesis and modification to changes in mitochondrial localization, membrane trafficking, and synaptic function, muscle directed RNAi was utilized to elicit more drastic inhibition of sfl and ttv expression. TEM images of the NMJ in these animals replicated the phenotypes seen in the mutants and revealed significant worsening of the partially penetrant SSR disruption.

In RNAi expressing animals, the SSR was drastically disorganized, exhibiting expansion of the normally narrow spaces between membrane layers (Fig. 1.05, compare A and A’ to B, B’, C, C’). Total SSR membrane content was significantly reduced in these animals even after accounting for the expanded post-synaptic area (Figure 1.05 D).

Figure 1.05: Changes in SSR structure, mitochondrial number and morphology in HS biosynthesis compromised larval muscles.

(A-C) TEM images of the SSR at a single synaptic bouton; SSR: subsynaptic reticulum; MN: motoneuron. In control animals (A and Ai: Mef2-GAL4/+), SSR comprised of multiple compactly assembled membrane stacks are tightly packed surrounding the presynaptic nerve terminal. In animals where sfl (B and Bi: Mef2-GAL4>UAS-sflRNAi) or ttv (C and Ci: Mef2-GAL4>UAS- ttvRNAi) function was compromised by muscle-directed RNAi, SSR organization was disrupted. (Ai-Ci) Higher magnification views of regions demarked by blue squares in (A-C). (D) Quantitative analysis of membrane content in the SSR showed significant decreases in sflRNAi and ttvRNAi expressing muscles compared to controls. (E-G) TEM micrographs of the cytoplasm below the NMJ synapse in controls (E: Mef2-GAL4/+) and animals expressing sflRNAi (F: Mef2- GAL4>UAS-sflRNAi) or ttvRNAi (G: Mef2-GAL4>UAS-ttvRNAi) in the muscle. Note the paucity of mitochondria in sflRNAi and ttvRNAi compared to controls. (H-I) Electron micrographs of mitochondrial morphology in controls (H), and animals expressing sflRNAi (I) or ttvRNAi (J) in the muscle. Red arrows indicate individual mitochondria of interest. In I, controls show mostly rod shaped mitochondria, while RNAi of sfl or ttv lead to a large number of mitochondria that were “bent” or shorter and rounder. (K) Quantification of mitochondrial density in the subsynaptic cytosol. Compared to controls, sflRNAi and ttvRNAi expressing animals have fewer mitochondria. Scale bars (A-C) 1 μm, (Ai-Ci) .5 μm, (E-G) 5 μm, (H-J) 0.5 μm. Error bars denote SEM. ***, p < 0.001. Numbers at the bottom of the bar indicate sample sizes. Figure 2, Reynolds-Peterson et. al 2017 (Experiments and analysis performed by N. Zhao, for additional information on methods see Appendix F)

Figure 1.06: Accumulation of membrane-bound structures in muscle cells of larvae with compromised HS biosynthesis.

(A, B and Bi) Electron micrographs of subsynaptic cytosol in controls (A: Mef2-GAL4/+) and animals expressing sflRNAi in the muscle (B and Bi: Mef2-GAL4>UAS-sflRNAi). (Bi) Higher magnification view of structures in the red box (whole panel) and the blue box (inset) from (B). In addition to classical autophagosomes, there are groups of membrane bound structures, some with single membrane layers (1), or nested within one another (2), and double membrane bound structures filled with electron dense material (3) and cytosol (inset). Mef2-GAL4>UAS- sflRNAi animals have many more vesicular structures than controls, quantified in (C). mito: mitochondria; N: nucleus; AP: autophagosome, PM: plasma membrane. (D) Mef2-GAL4>UAS- ttvRNAi animals harbor similar membranous structures. (Di) A higher magnification view of the red-boxed area in (D) (whole panel) and the blue box (inset). Vesicles appear adjacent to mitochondria (arrows). Crescent shaped thick membrane sheets containing regions of mitochondrial matrix (arrowhead) and closed vesicles with mitochondrial matrix-like material remaining between the layers (inset) are common. Scale Bars (A, B, D) 2 μm, (B’ and D’) 0.5 μm Error bars denote SEM. **, p < 0.01. Numbers at the bottom of the bars indicate sample sizes. Figure 3, Reynolds-Peterson et. al 2017 (N. Zhao)

Just below the muscle surface in the subsynaptic cytosol, similarly dramatic changes were seen in organellar composition. Both the muscle tissue itself and the SSR in particular have high energy requirements, and sub-synaptic cytosol in control animals is packed with mitochondria (Fig. 1.05 E). HS inhibition led to a highly significant loss of mitochondria in this cellular region (Fig. 1.05 F and G, K). Furthermore, the morphology of remaining mitochondria was abnormal (Fig. 1.05, compare H to I and J).

RNAi knockdown of sfl also resulted in the accumulation of a large number of unusual double membrane vesicular structures, seen to a lesser extent in ttv inhibited animals (Fig. 1.06 C, compare A to B). Some of these structures resembled traditional mature autophagosomes (AP, Figure 1.06 B’) while the majority lacked identifying features other than having two or more layers if enclosing membrane. These vesicular structures were commonly seen in close proximity to mitochondria (Fig. 1.06 D’, arrows).

A sizable number of these structures had remnants of electron dense material, similar in appearance to mitochondrial matrix, between their enclosing membrane layers (Fig. 1.06

D’, arrow head and blue inset).

Mitochondrial quality and numbers are regulated by autophagic degradation, and mitochondrial membrane is a known donor source for the initiation of autophagic phagophore formation105. Therefore, the appearance of these double layer vesicular structures and the concomitant disappearance of mitochondria raised the question whether loss of HS biosynthetic capability was altering levels of autophagy. This would explain the previously identified increase in pre-synaptic endocytosis, which is also symptomatic of increased autophagy. To probe the nature of the vesicular structures identified by TEM, the population of autophagosomes was examined using a fluorescent marker. The GFP-Atg8a marker is an autophagosome specific PE-conjugated membrane protein, which produces easily identified puncta when it is incorporated into

21 autophagosomes. Use of this marker revealed that the number of early autophagosomes was significantly increased in HS compromised animals106.

To test the applicability of this finding in terms of the HS dependent SSR phenotypes, SSR structure of sfl inhibited animals was investigated in the context of autophagic inhibition by co-inhibition of Atg8a. Blockage of autophagy through RNAi of

Atg8a in the muscle was sufficient to prevent both the sfl-induced loss of SSR membrane density previously observed (Fig. 1.07 A-C, D). Co-inhibition of sfl and Atg8a also rescued mitochondrial numbers in the subsynaptic cytosol (Fig. 1.07 E). This rescue directly tied the synaptic deficits caused by HS inhibition to a pathological increase in autophagy, and prompted a more thorough investigation of autophagic degradation in the context of HS biosynthetic perturbation.

Figure 1.07: SSR phenotypes associated with impaired HS modification are suppressed by downregulation of autophagy.

(A-C) Ultrastructure of SSR in control (A: Mef2-GAL4/+), sflRNAi (B: Mef2-GAL4>UAS-sflRNAi), and animals co-expressing sflRNAi and Atg8aRNAi (C: Mef2-GAL4>UAS-Atg8aRNAi, UAS- sflRNAi). Abnormally assembled SSR found in animals with loss of sfl function was partially rescued by expressing Atg8aRNAi. (D) Quantification of membrane content in the SSR region. Reduction of Atg8a function by expressing Atg8aRNAi in the muscle restored SSR membrane density in sflRNAi animals to a significant degree. SSR: subsynaptic reticulum; MN: motoneuron. (E) Quantitative analysis of mitochondrial densities in TEM micrographs. Co-expression of Atg8aRNAi significantly increased mitochondrial density in sflRNAi expressing animals. Figure 10, Reynolds-Peterson et. al. 2017 (N. Zhao)

1.3 Results

1.31 Endogenously expressed autophagy substrates are upregulated in muscle with impaired HS function

The level of autophagy substrate proteins is frequently examined to help determine the direction of a regulatory impact on autophagic regulation. Ubiquitin- modified proteins and ref(2)P (refractory to sigma P, the Drosophila p62 homolog) are two such commonly quantified autophagy substrates107. Ubiquitin and ref(2)P participate in substrate identification and autophagosomal loading, and thus become subject to degradation along with other substrates in the autolysosome.

Levels of ubiquitin and ref(2)P were measured in animals expressing RNAi or overexpression constructs under the control of the muscle specific GAL4 driver Mef2. In comparison to controls (Fig. 1.08 A, H), inhibition of autophagy through RNAi of Atg8a resulted in a massive accumulation of both ref(2)P and ubiquitin (Fig. 1.08 B, I).

Overexpression of Atg8a, a condition sufficient to increase autophagy108, 109, caused a much more modest increase in both proteins over basal levels (Fig. 1.08 C, J). A similar result was seen in animals expressing RNAi against the Drosophila Class I PI3K,

PI3K92E (Fig. 1.08 D, K), which is also known to enhance autophagy110, 111. RNAi of sfl and ttv (Fig. 1.08 E and M, F and N, respectively) resulted in a statistically significant increase in both markers much more similar to that seen in genetic conditions of enhanced autophagy than in situations of reduced autophagy (Fig. 1.08 G, O).

These observations clearly indicate a disruption in the levels of autophagy, but provide insufficient evidence to declare that shift an increase or decrease in overall autophagic flux. The levels of autophagy-associated proteins at any given time are determined by the balance of their production and degradation. For proteins that are

Figure 1.08: Compromised HS biosynthesis leads to increased levels of autophagy indicator proteins in larval muscle.

Confocal images of anti-ref(2)P (A-F) and anti-ubiquitin immunostaining (H-N) of muscles 6 and 7, segment A3, in 3rd instar larvae. A and H control: Mef2-GAL4>UAS-wRNAi, B and I inhibition of autophagy: Mef2-GAL4>UAS-Atg8aRNAi, C and J enhanced autophagy (overexpression of Atg8a): Mef2-GAL4>UAS-Atg8a, D and K enhanced autophagy: Mef2-GAL4>UAS-Pi3K92ERNAi, E and M: Mef2-GAL4>UAS-sflRNAi, F and N: Mef2-GAL4>UAS-ttvRNAi. G Quantification of ref(2)P immunofluorescent signal density. O Quantification of ubiquitin immunofluorescent signal density. Scale bar 50μm. Error bars denote SEM. *, p < 0.05; **, p < 0.01; ***, p < 0.001. Numbers at the bottom of the bars indicate sample sizes.

both required to advance autophagy and degraded as a result of participating in the process, cellular conditions that activate autophagy increase both production and turnover of these requisite proteins112. Upregulation of autophagy is associated with increased expression of ref(2)P112 and increased ubiquitination113. In aged tissues which experience a natural senescence of autophagic degredation and reduced transcription of autophagy-requisite proteins, increases and decreases in the level of autophagic flux can be clearly identified by decreased or increased levels of endogenous ref(2)P and ubiquitin, respectively107. In the tissues of young animals particularly, the increase in expression of ref(2)P can outstrip autophagic degradation, resulting in increased levels of the protein rather than the canonically expected decrease109, 114. For these reasons, additional evidence of autophagic progression in our larval muscle tissue model was sought.

1.32 The degradative capacity of HS inhibited muscle tissue is not reduced.

If an accumulation of autophagic substrates was occurring due to a failure of autophagosome maturation, this would be reflected in the number of autolysosomes and endosomes seen in affected cells. Increases in autophagic flux require an expansion in

the number of lysosomes to meet the need for autophagosome fusion and degradation,

while animals with normal or below average levels of autophagy have fewer acidic

organelles.

Figure 1.09: Accumulation of autophagic structures in muscle cells of larvae with compromised HS biosynthesis.

(A-C) Autophagy levels were measured with LysoTracker in controls (A: Mef2-GAL4/+) and animals with muscle-specific expression of sflRNAi (B: Mef2-GAL4>UAS-sflRNAi) or ttvRNAi (C: Mef2- GAL4>UAS-ttvRNAi). D Average number of eGFP-Atg8a-positive patches. H Average number of LysoTracker-positive structures. Scale bar 50μm. Error bars denote SEM. *, p < 0.05; **, p < 0.01; ***, p < 0.001. Numbers at the bottom of the bars indicate sample sizes.

In order to examine the degradative capacity of RNAi expressing muscles, the vital dye LysoTracker Red was used to visualize all acidic organelles in fresh muscle cells of control animals and animals expressing RNAi against sfl or ttv. In both cases,

(Fig. 1.09 B: sfl RNAi and C: ttv RNAi) the number of lysosomes exhibited a statistically significant increase beyond that seen in controls (Fig. 1.09 A). While not definitive proof of an increase in autophagic degradation, this evidence indicates that the increased number of early autophagosomes found in these animals are proceeding through the later stages of the autophagy vesicular cycle.

Although the results of LysoTracker staining agreed with an increase in autophagic flux, this observation does not directly address the observed increase in ubiquitinated proteins and ref(2)P. If autophagic flux is truly increased, a rise in the level of these proteins may be resulting out of the necessity to identify and direct substrates to autophagosomes. On the other hand, failure of lysosomal degradation might lead both to increased levels of lysosomes and increased quantities of substrate proteins if the blockade in degradation occurs after autophagosome-lysosome fusion.

To better understand the dynamics of ref(2)P degradation, an ectopically expressed GFP-tagged version of the protein was examined by western blot. Due to the somewhat high lysosomal pH of Drosophila, turnover of autophagy substrates can be readily measured by examining release of a relatively degradation-resistant fluorophore from a tagged protein following rapid degradation of the highly vulnerable peptide linker attaching it to the protein of interest107. Comparing the intensity of the parental GFP- tagged holoprotein to the intensity of the free GFP band can therefore give a rough indication of protein turnover.

Figure 1.10: Reductions in HS biosynthesis and modification result in a net increase in autophagic flux.

An anti-GFP Western blot to examine lysosomal degradation of ectopically expressed GFP- ref(2)P. Da-GAL4 was used to ubiquitously express UAS-GFP-ref(2)P alone or with an RNAi construct, and total protein was extracted from whole larvae for analysis. The intact fusion protein migrates at approximately 130 kDa (upper box) while the free GFP domain released by lysosomal degradation runs at approximately 27 kDa (lower box). The ratio of fusion protein to free GFP, calculated by division of background-corrected GFP-ref(2)P band signal density by free GFP background-corrected intensity within each sample, is displayed below the bottom box. da- GAL4>UAS-GFP-ref(2)P without an additional transgene served as the control.

Pancellular ectopic expression of a GFP-ref(2)P fusion protein was employed to examine the efficiency of ref(2)P degradation in HS biosynthetically impaired animals.

Anti-GFP staining was performed to measure proteolytic release of free GFP (~27 kDa) from GFP-ref(2)P (~130 kDa), as measured by the ratio between these two protein species (Fig. 1.10). This ratio was increased in animals overexpressing Atg8a to enhance autophagic degradation, while inhibition of autophagy through Atga8a RNAi resulted in a marked accumulation of the parental GFP-ref(2)P protein. RNAi of sfl caused a similar, though less dramatic, increase in the release of free GFP, while inhibition of ttv produced a conversion ratio very similar to that seen in controls.

Combined with the increased number of lysosomes and the slightly elevated levels of ref(2)P and ubiquitinated proteins, this evidence of adequate lysosomal degradation supports an increase in overall autophagic degradation in response to impaired HSPG function in muscle tissue. Muscle tissue acts as the primary reservoir of

29 protein in higher eukaryotes, making it a particular target of autophagy when the need for amino acids is unmet115. Muscles also rely heavily on autophagy for proteostasis of their many long-lived cytosolic proteins, which are mostly degraded by autophagy116. In order to determine whether this regulatory role of HSPG function was muscle specific, we moved on to examine levels of autophagy in another metabolically sensitive tissue.

1.33 HS Biosynthetic inhibition increases levels of autophagy markers in fat body tissue

Fat body is the tissue most commonly used to study autophagy in Drosophila.

The tissue consists of a single layer of exceptionally large, flat cells with large nuclei and prominent lipid droplets. Cells of the fat body perform the energetic storage roles of both mammalian adipose and liver, storing glycogen and lipids in a single location117.

Hormonal and growth factor signaling from this tissue controls organismal metabolism, body size specification, and many other critical processes. The thin, wide cellular shape also lends itself to the use of visual markers in the study of autophagy. For all of these reasons, we moved to the fat body to determine whether regulation of autophagy by

HSPGs could be generalized to other tissues.

The autophagy markers GFP-Atg8a and LysoTracker Red were examined in wandering 3rd instar larvae, which exhibit a slightly elevated level of autophagy as the animals have stopped feeding and are preparing to pupate (Fig. 1.11 A, E). Disruption of

HSPG biosynthesis through inhibition of sfl (Fig. 1.11 B, F) and ttv (Fig. 1.11 C, G) fat body cells produced a drastic increase in the density of both GFP-atg8a marked structures and acidic organelles (Fig. 1.11 D, H), as previously seen in muscle cells.

The significant increase in both early (GFP-Atg8a positive) and late (LysoTracker

Red positive) autophagic vesicles once again suggests an increase in autophagic flux.

Reproducing this effect in a second tissue that is not only energetically critical but also already using autophagy to mobilize resources during a fasting period indicates that

HSPGs may play a general role in regulating autophagy under a variety of physiological conditions.

As reviewed in section 1.21, HSPGs are required for a wide variety of cell- surface growth factor signaling pathways. Many such growth factor signaling events go on to mediate mTOR activity through the PI3K signaling cascade. The primary mechanism for canonical autophagy regulation operates through the manipulation of mTOR activity. It therefore seemed likely that the regulation of autophagy by HSPG activity would be exerted through cell surface signaling events upstream of class I PI3K activity. However, empirical support for this hypothesis and evidence of a potential mechanism remain to be identified.

Figure 1.11: Homozygous mutant ttv00681 causes a cell non-autonomous increase in the number of acidic organelles.

A and B Confocal images of unfixed fat body tissue from wandering 3rd instar larvae. Acidic organelles visualized using LysoTracker (red) and nuclei visualized using Hoechst (blue) in control (A: Oregon R) and homozygous mutant (B: ttv00681) larvae. C Density of LysoTracker positive puncta was significantly increased in homozygous mutants. D and E Homozygous ttv00681 clones were generated by heat shock-induced FRT mediated recombination in a heterozygous background. D The wild type chromosome arm is marked with UAS-mCD8 GFP, which appears the brightest in homozygous wild type cells, while homozygous mutant cells are identified by the absence of GFP. E Green lines indicate tissue boundaries. White lines demark homozygous ttv00681 mutant clones. The density of acidic organelles marked by LysoTracker (red) is similar in homozygous mutant clones and neighboring heterozygous cells or wild type clone cells. Scale bars, 50 µm. Error bars denote SEM. ***, p < 0.001. Numbers at the bottom of the bars indicate sample sizes.

1.34 HS-mediated autophagic regulation is non-cell autonomous

To further verify the effect of HS inhibition of autophagy, LysoTracker Red staining was performed in fat body tissue of the previously described ttv00681 P-element insertion mutant. In controls, (Fig. 1.12 A) there are few prominent lysosomes. The homozygous mutant (Fig. 1.12 B) displayed a significant increase in both the size and number of lysosomes, which translated to an increase in the tissue density of

LysoTracker positive structures (Fig. 1.12 C).

The identification of a positive phenotype in the ttv00681 mutant fat body cleared the way to begin probing the mechanism of HSPG mediated autophagic regulation using mosaic mutants. While muscle and fat body cells displayed similar responses to loss of sfl or ttv gene function, mosaics cannot be generated in muscle due to the multinucleate nature of the cells. Fat body tissue on the other hand is amenable to clonal generation.

Unfortunately, due to the lack of positional permanence during cellularization of the fat body, large clone patches are rare as daughter cells are free to drift away from one another. As a result, all of the homozygous mutant recombinant cells are bordered by homozygous or heterozygous wildtype cells on at least one side.

Homozygous mutant cells can be identified by the absence of GFP, which is readily apparent in heterozygous cells and doubly bright in homozygous wildtype cells

(Fig. 1.12 D). Homozygous mutant cells are outlined in white (Fig. 1.12 E). Converse to the bright LysoTracker Red staining in cells from entirely homozygous mutant fat body tissue (Fig. 1.12 B), homozygous mutant cells in a mosaic fat body show no change in their level of LysoTracker Red staining (Fig. 1.12 E). This non-cell autonomous

33 phenotype likely results from compensatory signaling provided by normal HS chains synthesized by neighboring cells, a phenomenon previously described in HSPGs.

Figure 1.12: Inhibition of HS biosynthesis in fat body cells induces autophagy in wandering 3rd instar larvae.

A-C Confocal images showing increased numbers of GFP-Atg8a-positive autophagosomes in fat body cells of animals expressing RNAi constructs targeting sfl (B, F: r4-GAL4>UAS-eGFP-Atg8a; UAS-sflRNAi) or ttv (C, G: r4-GAL4>UAS-eGFP-Atg8a; UAS-ttvRNAi) compared to controls (A, E: r4- GAL4>UAS-eGFP-Atg8a; UAS-wRNAi). Autophagosomes were visualized by eGFP-Atg8a (green), and nuclei were stained with Hoechst (blue). D The average density of eGFP-ATG8a puncta was significantly increased in animals expressing sflRNAi or ttvRNAi in comparison to animals expressing wRNAi (control) or no RNAi construct (not shown). (E-G) Within the same animals, autophagic degradation was measured by staining with LysoTracker (red) to visualize acidic organelles. F The average density of LysoTracker puncta was significantly increased in animals expressing sflRNAi or ttvRNAi in comparison to animals expressing wRNAi (control) or no RNAi construct (not shown). Scale bar 50 µm. Error bars denote SEM. *, p < 0.05; **, p < 0.01; ***, p < 0.001. Numbers at the bottom of the bars indicate sample sizes.

1.4 Discussion

Defects in HSPG biosynthesis elevate autophagy in muscle and fat body tissue

Loss of function mutations in two HS biosynthetic enzymes, sfl and ttv, were found to cause a variety of structural and functional changes at the neuromuscular junction of Drosophila larvae102. To determine the cause of these deficits, RNAi against these genes was used to induce more severe loss of function in muscle tissue. This not only reproduced the previously identified SSR structural defect with greater penetrance and severity, but also caused changes in the localization, morphology, and number of mitochondria in the subsynaptic muscle cytosol. Dysregulation of autophagy was identified as a likely effector in these phenotypic outcomes. Muscle- and fat body- specific expression of RNA interfering constructs was used to better characterize this relationship between inhibition of HS biosynthesis and changes in the level of autophagic degradation.

Multiple molecular markers of autophagy, including ubiquitin, GFP-Atg8a, and

LysoTracker Red were elevated in muscle and fat body cells experiencing loss of either sfl or ttv function due to RNAi. The presence of elevated numbers of both early (GFP-

Atg8a positive) and late (LysoTracker Red positive) structures suggested enhanced autophagy, by indicating that autophagosomes were proceeding normally through their maturation process. A western blot examining release of free GFP from GFP-ref(2)P, a commonly used method for examining autophagic flux in yeast and Drosophila107, 114 supported this notion. The increase in free GFP that accompanied ubiquitous inhibition of sfl gene function indicated an increase in autophagic degradation in these animals.

While the effect in ttv-inhibited animals was more equivocal, it is possible that both

35 release and subsequent degradation of GFP are increased, preventing the accumulation of free GFP in these animals.

Significantly, accumulation of endogenous ref(2)P and ubiquitin in larval muscle occurs under conditions in which autophagic flux is clearly elevated, supported both by previous characterization of mCherry-GFP-Atg8a106 and the GFP conversion assay.

Interpretation of ref(2)P levels in flies has primarily been performed in conjunction with treatments spanning days or weeks, and is generally examined in the tissue of adult animals118-120. In this context, an increase in ref(2)P indicates failure of autophagic flux.

However, in the context of larval tissues, increased levels of ref(2)P have been found in fat body tissue in response to starvation109 and inhibition of Tsc1114, both positive regulators of autophagy. In the tissues of these young animals, it may be that the transcriptional upregulation of ref(2)P, which occurs in response to various positive regulators of autophagy including starvation112, may outweigh increased autophagic clearance of the protein to a greater extent than in other experimental settings. The results described here provide another indication that caution is due when interpreting an increase in ref(2)P or ubiquitin outside the context of aged adult tissues in Drosophila.

HSPGs as general regulators of autophagy: a potential rheostat on input to mTOR

Confirming the increase in autophagy in a second tissue suggested a role for

HSPGs as general regulators of autophagy and afforded an opportunity to probe the mechanism of this regulation using mosaic mutants. In genetic mosaic fat body tissue, all of the homozygous mutant cells identified bordered on at least one cell producing normal HSPGs. In this setting, neighboring cells were able to rescue autophagy levels in the homozygous mutant cells, likely through non-cell autonomous signaling. This

36 phenomenon has been previously observed in several HS dependent signaling events and supports an ECM/cell surface signaling-mediated mechanism for the regulation of autophagy by these molecules121, 122.

HSPGs regulate signaling through a variety of cell surface receptors that use class one PI3Ks to propagate signaling cascades within the cell, which in turn activate mTOR, the canonical suppressor of autophagy. Due to the fundamental importance of

HSPGs in developmental pattering during embryogenesis, it can be difficult to tease apart the causality of phenotypes that may result from disruption of a specific signaling event or a broad change in cellular metabolism. This is a common complication in diseases associated with defects in HS regulation.

A great deal more evidence is required to confirm that HS-mediated signaling regulates autophagy through mTOR. Developing a model of increased HS function is a basic pre-requisite to progress in this direction. Manipulating PI3K in sfl inhibited tissues prevented the increased autophagy associated with HS loss (Appendix A, N. Zhao unpublished data). However, this only indicates that constitutive mTOR activation is sufficient to override whatever regulatory input is generated by HS inhibition.

Manipulating downstream signaling in the context of HS activation would provide much more mechanistic insight. Developing a Drosophila model of multiple sulfatase deficiency could provide one such system in which to investigate both the disease itself and the influence of increased HSPG activity on autophagy.

MSD is caused by partial loss of function of the gene SUMF1 in humans. Studies of the MSD mouse model indicate that loss of mouse Sumf1 results in increased HSPG activity, likely due to both loss of Sulf-mediated negative regulation at the cell surface and the failure of lysosomal HSPG degradation. Autophagy is also systemically

37 suppressed in these animals, supporting a model in which HSPGs function to negatively regulate basal autophagy42.

Work with Sumf1 null mice has begun to identify specific causes of certain disease associated phenotypes. Growth plate abnormalities were rescued in this model by correcting for HS hyperactivity-induced increases in FGF signaling, however, cartilage cellularity was not rescued39. In poorly vascularized tissue, autophagy plays a more critical role in meeting cellular energetic needs. The continuing blockade of autophagy in these animals was therefore theorized to ultimately result in the death of cartilage cells. Beyond the constellation of skin, musculoskeletal and truncal organ anomalies, MSD also causes severe and eventually fatal neurodegeneration123.

Neuronal cell death in this disease may be as much a result of general autophagic inhibition as it is a result of defective signaling.

HSPG-mediated autophagic regulation: clinical connections to chemoresistance and poor prognosis in post-chemotherapy cancer relapse.

This thesis discusses a novel, and potentially global role for HS-mediated autophagic regulation. However, the biological implications of such a regulatory relationship have been turning up in medical research even beyond the sphere of directly HS-related diseases. Tumor chemoresistance is very significantly associated with increased expression of heparanase125. Heparanase is the only endoglucuronidase capable of degrading heparan sulfate, making it an extremely important regulator of the

ECM126. Chemotherapy is a mostly effective first-line treatment for cancer. While most patients exhibit dramatic reductions in tumor mass following chemotherapy, those who experience later relapses will often present with chemoresistant tumors in subsequent

38 treatments. This is associated with poor prognosis and more aggressive tumor growth and metastasis124. In the context of cancer, increased heparanase activity is associated with the generation of a pro-survival tumor microenvironment and has been identified in nearly every type of cancer studied125, 127-130.

Recent research probing the role of heparanase in chemoresistance has begun to reveal intracellular changes caused by increased heparanase expression. Ectopic overexpression of heparanase was found to elevate levels of autophagy markers131, while cells from heparanase deficient mice show the opposite effect132. Autophagy is already known to protect cancer cells from starvation and hypoxia133, suggesting that the ability of this enzyme to promote tumor growth and chemotherapy resistance131, 134 functions in part through manipulation of autophagy130, 133. Increased autophagy in cancer cells prompted by heparanase-dependent depletion of HS mirrors our own findings in HS biosynthesis compromised animals. Together, evidence from studies in cancer, MSD, and our own work strongly suggest a role for HSPGs as global regulators of basal autophagy.

1.6 Materials and Methods

Fly strains were raised on standard cornmeal/sucrose/agar media at 25°C during embryogenesis and 30°C during larval development under a 12 h day/12 h night unless otherwise specified. Oregon-R, w1118, and VDRC60100 strains served as wild-type stocks. RNAi strains and a control strain with the same genetic background were obtained from the Vienna Drosophila RNAi Center (VDRC)77: UAS-Atg8a RNAi (43097),

UAS-Pi3K92E RNAi (38985), UAS-sfl RNAi (5070), UAS-ttv RNAi (4871), UAS-w RNAi

(30033) and empty vector control (60100). UAS-GFP-mcherry-Atg8a, UAS-mCD8-GFP,

UAS-GFP, UAS-mito-GFP, EP-Atg8a, and the Drosophila Transgenic RNAi Project

(TRiP)78 lines UAS-Atg5 RNAi (JF02703) and UAS-Atg7 RNAi (JF02787) were obtained from the Bloomington Drosophila Stock Center (BDSC). UAS-GFP-Atg8a 135 and UAS-

GFP-ref(2)P 136 are generous gifts from T. Neufeld, University of Minnesota,

Minneapolis, MN. UAS-Xbp1-eGFP 137 and UAS-ninaEG69D are generous gifts from H.

Steller, The Rockefeller University, New York, NY. Mef2-GAL4, r4-GAL4, and da-GAL4

(BDSC) transposon-containing stocks were used for muscle-specific, fat body-specific, and ubiquitous expression of RNAi constructs or transgenes. The c754-GAL4 hsFLP;

FRTG13 UAS-mCD8GFP and FRTG13 ttv00681/CyOGFP strains were generated using stocks obtained from the BDSC.

1.61 Real-time quantitative polymerase chain reaction

The results of RTqPCR analysis of RNAi efficacy are compiled in Appendix B.

Total RNA was extracted from whole second (n=~40) or third (n=~20) instar larvae using NeocleoSpin RNA extraction kit (Macherey-Nagel, 740955) following the

40 manufacturer’s manual. All genotypes were analyzed in biological triplicate. RNAi constructs were expressed ubiquitously under the control of Da-GAL4, and crosses were maintained at 25°C. For each sample, 0.5 mg of total RNA was used for reverse transcription of RNA in 10-μl reaction mix using qScript cDNA SuperMix (Quanta

Biosciences, 95048) to generate first strand cDNA. Real time quantitative PCR was performed using the PerfectaTM SYBR® Green Supermix, ROXTM (Quanta

Biosciences, 95055) on StepOnePlusTM (Applied Biosystems). Primers used: sfl: forward TCCCGCACCCATTATCAAC, Reverse TCAAACACAATCACGCCATAAC; ttv: forward ATCCCGCTCTTCCACAAAC, Reverse GTCGTATCTGTCGTATTCCCG; Atg5: forward GCACGCACGGCATTGATCTACA, reverse GCCCTGGGATTTGCTGGAAT;

Atg7: forward TTTTGCCTCACTCCATCCGTGG, reverse

ATCCTCGTCGCTATCGGACATG; Atg8a: forward CAACCAACCAACCAACTTTCC, reverse GCATTCGCACGGATCAATTAC; RpL32: Forward

GCAAGCCCAAGGGTATCGA, Reverse ACCGATGTTGGGCATCAGA. The relative

RNA levels of each sample were found by comparing the amplification of the gene(s) of interest first to the internal control gene RpL32, and then calculating the relative change in comparison to a likewise internally corrected value from the genotypic control.

1.62 Immunohistochemistry and LysoTracker Red staining

Wandering third instar larvae were dissected in ice-cold Ca2+ free HL-3 medium

(70 mM NaCl, 5 mM KCl, 20 mM MgCl2, 10 mM NaHCO3, 5 mM trehalose [Alfa Aesar,

A19434], 100 mM sucrose, 5 mM HEPES, pH 7.2). Samples were fixed in 4% paraformaldehyde for 30 min (except samples for anti-HS antibody staining), followed by intensive washing with phosphate-buffered saline (PBS: 137 mM NaCl, 2.7 mM KCl, 10

41 mM Na2HPO4, 1.8 mM KH2HPO4, pH 7.4). Muscle preparations with no trace of fixative solution left were blocked in 10% normal goat serum (MP Biomedicals, IC642921) for 1 h at room temperature and then incubated with primary antibodies for at least 12 h at

4°C. Samples were intensively washed with PBST (PBS plus .1% Triton X-100 [IBI

Scientific, IB07100]) and incubated with secondary antibody for 1 h at room temperature.

Muscle preparations were then washed as before and incubated overnight in mounting media before being mounted on glass slides. Primary antibodies were used at following concentrations: goat anti-HRP antibody (1:1000, conjugated with Alexa fluor 488;

Jackson ImmunoResearch Laboratories, 123-545-021), anti-ubiquitin antibody (1:1000,

FK2; Enzo, BML-PW8810-0100), anti-ref(2)P 114 (1:5000) and anti-mitochondria antibody

(1:500, 4C7; Developmental Studies Hybridoma Bank, 4C7). Alexa Fluor fluorescence- conjugated secondary antibodies (1:1000) were obtained from Life Technologies

(A21070, A21050, A11003, A11010, A11001, A11008).

For LysoTracker Red staining in muscle cells, wandering third instar larvae were dissected in Ca2+-free HL-3(70 mM NaCl, 5 mM KCl, 20 mM MgCl2, 10 mM NaHCO3, 5 mM trehalose, 115 mM sucrose, 5 mM HEPES, pH 7.2) and incubated with 100 nM

LysoTracker Red DND-99 (Life Technologies, L-7528) for 5 min in the absence of light, rinsed twice with PBS, and imaged immediately. All of these procedures were performed at room temperature.

For LysoTracker Red staining in fat body cells, fed or starved third instar larvae were dissected in PBS. Fat bodies were incubated in 100 nM LysoTracker Red DND-99,

1 μM Hoechst 33342 (Life Technologies, H3570) in PBS for 2 min. Fat bodies were then

42 rinsed twice with PBS, transferred to 40% glycerol on glass slides, covered, and imaged immediately. All of these procedures were performed at room temperature.

1.63 Mosaic analysis

c754-GAL4 hsFLP; FRTG13 UAS-mCD8-GFP virgin females were mated with

FRTG13 ttv00681/CyOGFP males. Eggs were collected in yeast paste on grape juice agar (2.2% agar, 1.2% sucrose, 25% grape juice concentrate) plates for 6 h, immediately followed by a 1-h heat shock at 37°C. The eggs were allowed to recover overnight at 25°C before being transferred to food medium. Larvae were reared at 25°C on instant food medium under a 12 h light/dark cycle. Feeding stage 3rd instar larvae were picked out of the food media and sorted by gender, discarding the females. Male larvae bearing the CyOGFP balancer were identified by strong GFP fluorescence in the testes and discarded. The remaining male larvae exhibiting GFP fluorescence in the pattern directed by c754-GAL4>UAS-mCD8GFP only were dissected and stained with

LysoTracker Red as described above. Homozygous mutant clones were identified by loss of the mCD8GFP marker.

1.64 Image acquisition and analysis

Images were acquired at room temperature using an Olympus Fluoview FV1000 laser-scanning confocal microscope (Olympus America, Waltham, Massachusetts,

USA). FV10-ASW 2.1 software (Olympus, Waltham, Massachusetts, USA) was used to capture images. When more than one fluorophore was detected, sequential line scanning was performed to avoid spectral bleed through artifacts. Images of samples

43 with different genotypes within a single experiment were captured, processed, and analyzed using the same settings. Images were presented as Z-stacks of maximum intensity projections using Imaris 7.3 software (Bitplane Inc.). Three-dimensional surface reconstructions were generated from serial images taken by confocal microscopy using

Imaris 7.3. The numbers of or area covered by GFP-Atg8a, GFP-mcherry-Atg8a, ref(2)P, ubiquitin, and LysoTracker Red-positive signals were measured using ImageJ1.42q. The absolute values from each animal were individually normalized by muscle size/tissue area. Adobe Photoshop CS5 (Adobe Systems Incorporated) was then used to crop images and adjust brightness and contrast. All adjustments to contrast and brightness made to ease interpretation of confocal images were applied identically to all genotypes within the experiment.

1.65 Western blot

Ubiquitous expression of both UAS-GFP-ref(2)P (GFP-ref[2]P) together with

RNAi or overexpression constructs was directed by da-GAL4. Larvae were reared at

25°C. Five wandering 3rd instar larvae were homogenized by plastic pestle in 200 µl ice cold protein extraction buffer (20 mM HEPES, 100 mM KCl, 10 mM EDTA, 1 mM DTT,

.1% Triton X-100, 5% glycerol, 1X protease inhibitor cocktail Complete [Roche,

10184600])138 for each genotype. Samples were centrifuged for 10 min, and the supernatant was removed to a clean tube. Protein concentration was measured by BCA

Protein Assay (Pierce, 23227), and samples were diluted to a standard concentration using 4x protein loading buffer139 and protein extraction buffer, then boiled at 95°C for 5 min. Approximately 25 μg total protein per sample was run on an 8% polyacrylamide gel.

After transfer, the nitrocellulose membrane was rinsed in TBS (50 mM Tris-Cl, 150 mM

NaCl, pH 7.6) and dried overnight. The membrane was rehydrated and blocked in 0.5%

TBST (TBS plus 0.5% Tween 20 [BDH, 4210])140 and incubated overnight at 4°C with rabbit polyclonal anti-GFP (1:1500, CAB3211; Thermo Scientific, CAB4211). Goat anti- rabbit-HRP secondary antibody was applied at 1:3000 (Invitrogen, 31430). ECL detection was performed using G:BOX Chemi XG4 (Syngene, Fredrick, Maryland, USA) which was also used to invert coloration of resulting images. Band signal density was assessed using ImageJ1.42q.

1.66 Statistical analysis

Statistical analyses for quantitative data were performed with Minitab Release 16

(Minitab). All data points were presented as mean ± SEM. Normality of data distribution was determined using probability plots. Comparisons between 2 groups were performed using the Student t test for normally distributed data or Mann-Whitney test for nonparametric data. Comparisons between more than 2 groups were performed using

ANOVA for normally distributed data or Kruskal-Wallis for nonparametric data, followed by the Tukey test or Mann-Whitney test.

1.67 Additional MethodsFigures in section 1.2 excerpted from the publication

Reynolds-Peterson et. al (2017) were generated using data and analysis contributed by the coauthors of that publication. Additional methodological information pertaining to these experiments can be found in the methods section of Appendix F.

Chapter 2

Heparan Sulfate Proteoglycans Regulate Basal Autophagy and Stress Tolerance in the Central Nervous System

2.1 Abstract

Heparan Sulfate Proteoglycans (HSPGs), critical components of cell surface signaling and the extracellular matrix (ECM), have been shown to negatively regulate autophagy in both muscle and fat body tissue of Drosophila melanogaster larvae.

Autophagic degradation is required for catabolism of organelles, long lived structural proteins, aggregates, and endocytosed materials. Autophagy in the central nervous system performs critical functions in development and neuronal survival. Dysregulation of neuronal autophagy is associated with neurodegenerative diseases and aging, as well as abnormal synaptic features observed in neurodevelopmental disorders.

RNAi inhibition of HSPG biosynthesis in the brain was found to enhance autophagy in the central nervous system (CNS) through assessment of mCherry-Atg8a, a reporter construct used to monitor autophagic flux. Reduction of HS function through

RNAi inhibition of the HS biosynthetic enzymes sfl and ttv, and heterozygosity for mutants in these genes, improved clearance of oxidant damaged substrates following exposure to hydrogen peroxide (H2O2). Heterozygosity for either sfl or ttv extended survival during prolonged exposure to H2O2. Inhibition of HSPG biosynthesis also protected against neurodegeneration in a Drosophila model of Alzheimer’s disease caused by overexpression of presenilin. Homozygous loss of function of sumf function was developed as a model of increased HSPG activity, and animals lacking this gene were sensitized to age- and oxidant exposure- induced accumulation of autophagic substrates.

2.2 Introduction

Autophagy is required in all cell types for basic homeostasis, and protects against cell death from stressors such as starvation, hypoxia, reactive oxidant exposure, and acute protein misfolding due to disruption of the endoplasmic reticulum141-143. While this cellular recycling pathway is broadly important for survival, certain tissues and cell types are more sensitive to its loss. The central nervous system is one such tissue in which blockage of autophagy carries an invariably heavy toll144, 145. Inhibition of autophagy results directly in neuronal death and the development of neurodegenerative disease in mice146, 147. This is likely due to the structural and energetic sensitivity of neurons.

Axons and dendrites require an array of specialized proteins to function, and the constant vesicular trafficking required for synaptic transmission consumes a great deal of energy. Furthermore, the need for specialized proteins and ATP is located at sites distil from the cell body. This need is met by localization of mitochondria and protein translation to the axons and dendrites148, 149. Appropriate mitochondrial turnover is particularly important in neurons, which, as post-mitotic cells, cannot dilute the accumulation of defective organelles through cell division.

The long, thin axonal structure is also highly sensitive to disruption when autophagic degradation is compromised. Failure of autophagy leads to disruption of microtubule-dependent transport in the axon, which causes accumulation of p62- and ubiquitin-positive aggregates in focal swellings along the length of the axon150. Axonal focal swelling is a pathological hallmark of both Alzheimer’s disease151 and Parkinson’s disease152. Changes in axonal width and membrane quality caused by axonal focal

47 swelling compromise the propagation of action potential along the membrane, slowing and even completely blocking their transmission153.

2.21 Autophagy-dependent synaptic sculpting in neurodevelopmental disorders

Autophagy is critical to the development and plasticity of synapses. Dendritic pruning to fine-tune synaptic connections during adolescence is driven by autophagy, with under-pruning resulting in behaviors associated with Autism Spectrum Disorder

(ASD)154. Hyperactivity of autophagy during this same period is associated with over- pruning of dendritic fields and the development of schizophrenia155. Autophagy also drives pre-synaptic activity dependent plasticity in mature neurons156.

During nervous system development, excessive synapses are formed first, followed by pruning of both axons and dendrites to achieve optimum connectivity157-159.

Axonal pruning proceeds either by caspase-dependent degeneration 160, 161 or by physical retraction of the axonal process and subsequent synaptic elimination162, 163.

Pruning of dendrites and dendritic spines on the other hand is dependent on ubiquitin modification and AKT/PI3K mediated suppression of mTORC1164, suggesting an autophagy-dependent mechanism.

Constitutive over activity of mTORC1 in a mouse model of tuberous sclerosis is associated with both excess dendritic spines and behavioral changes. The increased dendritic spine density identified in these mice recapitulated an increase in dendritic spines seen in the temporal lobe of human post-mortem brain tissue samples from patients with ASD165, 166. Autophagy-dependent under-pruning of dendritic fields has been repeatedly associated with ASD and behavioral disturbances.

Loss of Atg7, a gene required for autophagy, in mouse microglia cells was sufficient to prevent normal dendritic pruning. Affected mice had excess cortical neuron dendritic spines and exhibited ASD-like repetitive behaviors and impaired socialization154. Furthermore, both tuberous sclerosis and fragile-X syndrome, the most common syndromic causes of autism, are associated with hyperactive mTOR signaling167, 168. Treatments that genetically or pharmacologically suppress mTOR activity have been shown to improve cognitive and behavioral phenotypes in mouse models of ASD169-171.

2.22 Contributions of autophagy to neurodegenerative disease processes

Autophagy is a critical homeostatic process, without which cells succumb to oxidative stress generated by damaged mitochondria and the accumulation of autophagic substrates. While proteasomal degradation is thought to be upregulated in response to autophagic suppression172, 173, this redundancy can only assist in the clearance of substrates that the proteasome is physically capable of degrading which limits its ability to adjust for autophagic failure over long periods of time174. In no tissue is this effect more noticeable and severe than the central nervous system. Failure of autophagy is associated with neuronal death in both Parkinson’s disease (PD) and

Alzheimer’s disease (AD)144.

Failure of mitophagy and oxidative stress drive neurodegeneration in both familial and sporadic Parkinson’s disease

Parkinson’s disease is a neurodegenerative disease characterized by the death of dopaminergic neurons of the substantia nigra, one of the five nuclei that make up the

49 basal ganglia of the midbrain175, 176. While neurodegeneration in PD is not entirely restricted to this brain region, the characteristic motor deficits associated with this disease result from the specific loss of SN neurons177. Mutations in PINK, Parkin,

LRRK2, and DJ-1 that are associated with familial PD are required for mitophagy, the process by which old and damaged mitochondria are removed from cells178-181. Inherited mutations such as these are associated with approximately 10% of PD cases182.

When mitophagy is compromised, neurons experience accumulation of reactive oxygen species accompanied by insufficient energy production. Mitochondrial dysfunction is considered a highly probable source of the oxidative stress associated with cell death in PD183. Post-mortem analysis of PD brain tissue has identified significant increases in molecules known to be byproducts of oxidative stress, including markers of lipid peroxidation184, 185, carbonylated cytosolic proteins186, and oxidated DNA and RNA derivatives187, 188.

Normally, depolarization of mitochondria following damage leads to stabilization of PINK1 on the mitochondrial outer membrane, triggering ubiquitination of mitochondrial outer membrane proteins by the ubiquitin E3 ligase Parkin followed by and autophagic engulfment189-191. A recent assessment of mitochondrial motility and degradation in cell lines derived from patients with sporadic PD identified functionally similar impairments of mitophagy despite the absence of inherited mutations in any of these genes181, 192. The identification of failed mitophagy as a mechanistic link between the pathology of familial and sporadic PD is a crucial step towards identifying treatments that can effectively manage both forms of this disease. Mitochondrial dysfunction has also recently been identified as a contributing factor in the neurodegeneration associated with Alzheimer’s disease, identifying increased mitophagy as a potential therapeutic target treatment for both PD and AD144, 193.

Autophagic dysregulation in Alzheimer’s disease

Alzheimer’s disease (AD) is the most common cause of progressive senile dementia194. Neuronal death in the vertebral cortex and hippocampus is associated with intracellular Tau-containing neurofibrillary tangles and extracellular amyloid beta (Aβ) plaques195, 196. Deposition of Aβ and Tau are seen to precede symptomatic onset by as much as 20 years195. Symptoms of AD include memory deficits, dementia, and the decline of cognition, language, and visual-spatial perception197, 198.

The toxic Aβ fragments associated with Alzheimer’s disease are known to be degraded by autophagy, and in post-mortem brain tissue of AD patients, clear hallmarks of autophagic failure can be seen in the form of numerous enlarged autophagosomes199,

200. Progression of AD is correlated with hyper-activation of mTORC1 in the brain, a condition known to inhibit autophagy201. Furthermore, the production of amyloid beta is associated with increased mTORC1 activation in cell culture202. Inhibition of autophagy in AD is seen not only in upstream regulation, but also during lysosomal processing203 and final clearance of autophagic vesicles204. The suppression of autophagy at every step of the pathway appears to be closely associated with cell death in both familial and sporadic cases of AD205.

2.23 HSPGs in neuronal cell death

Autophagic insufficiency also plays an integral role in the fatal neurodegeneration that accompanies many lysosomal storage diseases (LSDs). LSDs are diseases caused by failures in lysosomal degradation, which present with accumulation of various types of un-degraded substrates. These diseases are categorized into subgroups based on the

51 type of defect present. Phenotypic severity in affected individuals is highly reliant upon the type of substrate accumulated, and within specific disorders by the degree of functional compromise caused by each patient’s particular genetic mutation(s).

Failure to degrade varying types of glycosaminoglycans (GAGs) is associated with one family of LSDs, called mucopolysaccaridoses (MPS) (see appendix E for a summary of MPS subtypes). MPS subtypes IV, VI, and IX cause moderate to severe somatic disease, predominantly affecting the musculoskeletal system and truncal organs without involvement of CNS features. These MPS subtypes present with accumulation of the GAGs hyaluronin, chondroitin sulfate, and dermatan sulfate206-208. The most severe

MPS subtypes, I, II, III, and VII, are associated with developmental regression and eventual death due to neurodegeneration209-214. These MPS subtypes are associated with accumulation of heparan sulfate, with or without accumulation of other GAGs.

Two additional LSDs have been identified that present with accumulation of

GAGs in addition to other substrates. One is the recently identified

Mucopolysaccharidosis-Plus Syndrome (MPSPS). This recessive disease is caused by loss of function in VPS33A, a gene required for autophagosome maturation. MPSPS causes clinically significant accumulation of glycosaminoglycans, including HS215, 216 and causes both somatic and CNS phenotypes in keeping with MPS types I, II, and VII.

Multiple sulfatase deficiency (MSD) is another symptomatically severe LSD that, like

MPSPS, combines features of the mucopolysaccharidoses and other storage diseases.

Features of MSD include accumulation of all sulfated compounds, global inhibition of autophagy, musculoskeletal defects, and severe neurodegeneration39, 217, 218.

Interestingly, CNS involvement in MPSs is limited to disease subtypes that present with accumulation of HSPGs, while severe somatic disease is more likely with accumulation of dermatan sulfate. In MSD and MPSPS, neurodegeneration is once

52 again concordant with accumulation of HS GAGs. This correlation suggests that an HS specific process is responsible for neurodegeneration in the MPS family of LSDs.

Furthermore, post-mortem analysis of brain tissue from patients with MPS types I, III, and VII revealed elevated levels of soluble amyloid beta, suggesting that this HS-specific accumulation is correlated with abnormal amyloid beta and amyloid beta precursor protein dynamics219.

Due to the implications of autophagy on neurodegenerative and neurodevelopmental disorders and the recently identified role HSPGs play in the regulation of basal autophagy in Drosophila muscle and fat body tissue, it was natural to inquire after the implications of HSPG function in neuronal tissue.

2.3 Results

2.31 HSPG biosynthetic inhibition enhances neuronal autophagy and improves clearance of oxidant damaged cellular components

To begin investigating the role of HSPGs in the regulation of neuronal autophagy, the reporter UAS-mCherry-Atg8a was expressed under control of the pan-neuronal elav

C155-Gal4 driver. Conversion of the parental mCherry-tagged protein to free mCherry through lysosomal degradation was measured to assess levels of autophagic degradation in whole head protein extracts. In comparison to control animals, ectopic overexpression of wildtype Atg8a to induce autophagy released more free mCherry protein, with a 43% decrease in the ratio of the parental protein to free mCherry (Fig.

2.01). Pan-neuronal RNA inhibition of both sfl and ttv similarly reduced the ratio of the parental fusion protein to free GFP, by 32% and 49% respectively, indicating an increase in autophagic degradation in these animals.

To determine the biological impact of this potential increase in autophagy, an oxidant stress challenge was applied. Environmental oxidant exposure causes systemic damage of cellular components, which are targeted for clearance by ubiquitin modification. These ubiquitinated molecules are then cleared through a combination of proteasomal and autophagic degradation, both of which are accelerated in response to oxidative stress. Failure to rapidly clear ubiquitinated substrates leads to the formation of insoluble aggregates that are subject to autophagic degradation. Enhanced activation of autophagy reduces the presence of such aggregates following oxidant exposure through some combination of substrate clearance prior to aggregation and rapid removal of the aggregates that do form108.

Figure 2.01: Pan-neuronal HSPG biosynthetic inhibition increases autophagic turnover of an mCherry-Atg8a fusion protein

Pan-neuronal expression was achieved using elavC155-Gal4 with UAS-DcrII to enhance RNAi efficacy and the UAS-mCherry-Atg8a reporter construct. UAS-w RNAi was used as a control. Overexpression of wildtype Atg8a was used as a positive control for enhanced autophagy. Density of the parental mCherry-Atg8a band was divided by density of the free mCherry band to determine the ratio of fluorophore conversion.

Figure 2.02: Inhibition of HSPG function in neurons reduced the accumulation of autophagic substrates following environmental exposure to hydrogen peroxide.

Accumulation of insoluble ubiquitinated particles (IUPs) was examined in the triton-X100 insoluble fraction of total head proteins. UAS-mCherry RNAi, an shRNAi with no predicted targets in the Drosophila genome, was used as a control. Inhibition of Atg3 was used as a control for inhibition of autophagy. A Selected lanes from anti-ubiquitin stained membranes are shown in control vs. oxidant exposed pairs. The ratio depicted over these images is the average density of anti-ubiqtuitin staining in (+) samples divided by (-) samples. (-) denotes the absence and (+) the presence of 2% H2O2 in the food medium. B Quantitative representation of the relative average IUP density in unexposed (set to 1) and exposed animals, error bars represent SEM. Analysis of loading was performed using ponceau total protein staining. Images in A were selected for display from within three biological replicates based on equality of loading. Baseline IUP concentrations for B were set to an arbitrary value of ‘1’ to better represent the relative change in IUP concentration with and without treatment within each genotype, which was the biological effect of interest. The baseline IUP concentration is not equal between the different genotypes.

Levels of insoluble ubiquitinated particles (IUPs) were compared in triton-X insoluble head protein extract of control and oxidant exposed animals to determine the effect of HS biosynthetic inhibition on the autophagy-dependent oxidant stress response.

Control animals exhibited a 1.20-fold increase in IUPs following a 24-hour exposure to

2% hydrogen peroxide (Fig. 2.02). Inhibition of autophagy through pan-neuronal RNA inhibition of Atg3, a protein required for the formation of autophagosomes, increased the

H2O2 induced accumulation of IUPs to 1.42-fold, an 18% increase over controls.

Animals with pan-neuronal RNAi of sfl or ttv, however, experienced a decrease in the concentration of IUPs following oxidant exposure. sfl inhibition produced a post- exposure IUP concentration of 0.90, a 25% decrease in comparison to controls. ttv inhibition produced a ratio of 0.72, a remarkable 40% decrease compared to controls.

The reduced accumulation of IUPs in HS biosynthetically compromised brain tissue points both to enhanced autophagy and a drastic change in the way these neurons respond to proteostatic stress.

2.32 HSPG functional levels influence survival of chronic oxidant stress and age- associated proteostasis.

Deletion of the sumf allele was used to model the human disease Multiple

Sulfatase Deficiency, the pathophysiology of which is caused in part by increased

HSPG-dependent signaling activity. Heterozygous ttv and sfl mutants and heterozygous and homozygous sumf mutants were aged one week and subjected to chronic hydrogen peroxide exposure in order to determine whether or not the apparent alterations in autophagic degradation in these animals influenced their survival during periods of prolonged stress. Continuous exposure to yeast-agar medium supplemented with 1.5%

H2O2 was fatal to control animals over a period of 108 (average 73.8) hours in females and 72 (average 44.0) hours in males (Fig. 2.03 A & B).

Heterozygosity for the null alleles ttv00681 and sfl03844 (yellow and blue respectively) provided substantial protection against the oxidant challenge, with average survival durations of 93.6 and 90.8 hours in females and 77.4 and 63.1 hours in males respectively. Heterozygosity for the hypomorphic sfl9B4 point mutation provided no apparent protection in females, with an average survival duration of 75.9 hours, and the suggestion of a protective effect in males, with an average survival duration of 49.6 hours. In female animals, homozygosity for the sumfΔ213 allele clearly decreased tolerance of the oxidant challenge (Fig. 2.03 A and B), while heterozygous sumf mutants exhibited an intermediate phenotype.

From the survival curves, it appears that male animals are much more sensitive to oxidant-induced lethality than females. Heterozygosity for the ttv mutant allele produced an even stronger effect in males than females, while male sfl9B4 heterozygotes exhibited a trend towards mild protection not seen in females. On the other side of the biological balance, sumf heterozygosity or homozygosity had very little additional effect on survival in the already sensitized males. For males, the average survival duration was

44.0 hours in controls as opposed to 43.0 hours in sumf heterozygotes and 39.3 hours in sumf homozygous mutants. This data suggests that autophagic capacity and sensitivity to the types of stresses that activate a protective autophagic response may have biological thresholds around which even a very modest increase in autophagy can provide clear benefits.

Figure 2.03: Survival of heterozygous and homozygous mutants under continuous oxidative stress.

One week old flies were exposed to 1.5% H2O2 in food medium and survival was assessed every 12 hours. OreR was the control, all heterozygous mutants were f1 progeny of the mutant stock crossed to OreR. The sumfΔ213 deletion mutants were isogenized to our OreR laboratory stock before use in the homozygous state. While all genotypes were exposed and scored concurrently, results were broken up by gender and by the direction of the effect on HSPG function for ease of interpretation. N=125(±10) animals per genotype, per gender. A Survival curves. B Graphical representation of average time to death and time of 50%lethality (median). Error bars represent SEM.

Δ213 The effect on survival under H2O2 stress in sumf homozygous mutants raised the question whether the decreased apparent level of autophagy in these animals would also influence the process of natural aging. With increasing age, expression of autophagy related genes decreases108 and the rate of autophagic degradation gradually senesces49. This natural decline in protein turnover results in the accumulation of autophagic substrates such as IUPs and may be associated with age-associated human diseases108.

To probe the effects of the sumfΔ213 mutation on the natural senescence of autophagy, both heterozygous and homozygous mutants were collected and maintained for up to eight weeks. While a detailed record of lifespan data was not collected, sumfΔ213 homozygous mutants began dying at an accelerated rate around five weeks of age, preventing the collection of protein samples at eight weeks. In comparison, heterozygous mutants survived to eight weeks with minimal lethality.

At two weeks of age, heterozygous and homozygous mutants exhibited a roughly similar degree of IUP accumulation (Fig. 2.04). However, by six weeks of age, homozygous mutants had experienced a dramatic accumulation of IUPs. IUP levels in homozygous mutants increased by 81% from week two to week six, in comparison to the

21% increase seen in heterozygous mutants during the same period. Even at eight weeks, heterozygous mutants failed to reach the catastrophic degree of IUP accumulation seen in six-week-old homozygous mutants, exhibiting a 64% increase in

IUP accumulation compared to levels seen at two weeks of age.

Figure 2.04: Loss of sulfatase activity dramatically increases age-associated autophagic senescence.

The Triton-X100 insoluble fraction of head total protein was assessed for accumulation of IUPs by anti-ubiquitin staining. Average ubiquitin density relative to loading was calculated for each age group and is displayed over the lanes excerpted from the membrane. Representative lanes were selected from 3-4 biological replicates based on equality of loading.

2.33 Inhibition of HSPG function ameliorates neuronal death in a Drosophila model of familial Alzheimer’s disease

We have collected a great deal of evidence supporting HSPG function as an important global regulator of autophagy in tissues throughout the body, including the brain. In an effort to better understand the implications of our findings in the context of health and human disease, we assessed inhibition of HSPG biosynthesis and modification as a potential suppressor of a Drosophila model of familial Alzheimer’s disease. Size and organization of the compound eye were used as biomarkers to assess the degree of neuronal death associated with these disease models.

Pan-neuronal overexpression of the Drosophila Presenilin homolog, psn, was used to induce neuronal cell death. This disease model produced a moderate reduction in eye area (Fig. 2.05 F, graph of eye sizes) and marked disruption of ommatidial organization (Fig. 2.05 G, graph of flynotyper result) consistent with the death of photoreceptor neurons and their support cells during larval and early pupal development.

Ommatidial size and shape varied dramatically throughout the eye. The anterior border of the eye suffered particularly severe disruption, resulting in a distinctive blistered appearance (Fig. 2.05 B, large picture with arrows to point this out). Patches of necrosis were also visible in this region of the eye.

Inhibition of w, a gene important for neurotransmitter synthesis but not required for neuronal survival, was used as a control for non-specific effects of RNAi (Fig. 2.05

C). The eyes of these animals were similar in size and organization to animals expressing only psn. In comparison to controls, RNA inhibition of either sfl or ttv significantly restored eye area (Figure 2.05 F). One of the two sfl RNAi constructs was significantly more effective at reducing mRNA levels than the other (Appendix B), and only this line (sfl RNAi 2) significantly improved ommatidial organization (Fig. 2.05 E, G).

Ommatidia in the compound eye are normally hexagonal in shape, allowing for the formation of straight rows. This orderliness breaks down only around the edge of the compound eye where ommatidial alignment is constrained by the tissue boundary. While the ttv RNAi (Fig. 2.05 F) and sfl RNAi 1(Fig. 2.05 D) lines were able to reduce blistering at the apical eye border and variation in ommatidial size across the eye, the ommatidia do not form lines through the central field of the eye. With greater inhibition of sfl function through use of the sfl RNAi 2 construct, ommatidia in the center of the eye field achieved a sufficiently regular size and shape to produce visible rows.

Interactions between inhibition of sfl and ttv function and severity of the neurodegenerative phenotype were very similar in male and female eyes. In females, sfl

RNAi 2 restored significantly more eye area than the other two RNAi constructs, further highlighting the dose-dependent nature of the genetic rescue. In males, all three RNAi constructs affected eye area to a similar degree. Male flies inherently have smaller eyes than females, which resulted in a narrower margin between the eye area of wildtype flies and flies overexpressing psn. Within this narrow range, it is more difficult to differentiate between effects of relatively similar size.

Figure 2.0 5: Loss of HSPG function partially rescues neurodegeneration in a model of Alzheimer’s disease.

Overexpression of wildtype Drosophila presenilin was exerted using elavC155-Gal4>UAS-DcrII and a UAS-psn transgene alone and in combination with RNAi constructs targeting control and HS biosynthetic genes. A Control animals (elavC155-Gal4>UAS-DcrII/+ displayed, elavC155- Gal4>UAS-DcrII/UAS-wRNAi used as control in boxplots) have straight rows of ommatidia with a uniform size and shape. Animals overexpressing psn alone (B) and in combination with a control RNAi transgene (C, elavC155-Gal4>UAS-DcrII/UAS-psn; UAS-wRNAi) exhibit reduced eye area and disruption of ommatidial structure and organization, particularly severe at the anterior eye margin (arrow in B). Rows are only present to a limited extent at the posterior margin of the eye to a limited degree. D (elavC155-Gal4>UAS-DcrII/UAS-sflRNAi GD, 1) Very weak inhibition of sfl partially restored eye area but not ommatidial organization. E (elavC155-Gal4>UAS-DcrII/UAS- sflRNAi HMS, 2) Strong inhibition of sfl partially restored both eye area and ommatidial organization. F (elavC155-Gal4>UAS-DcrII/UAS-ttvRNAi) Weak inhibition of ttv partially restored eye area but not ommatidial organization. G Quantification of eye area. All lines overexpressing psn experienced a statistically significant reduction in eye area. Inhibition of sfl or ttv but not w significantly rescued eye area. H FlynotyperJ was used to calculate P-score, a quantitative measure of ommatidial organization. All lines overexpressing psn experienced a statistically significant increase in P-score. Strong inhibition of sfl via the UAS-sflRNAi shRNA construct was able to significantly rescue ommatidial disorganization. * P < .05, **** P < .0001 in comparison to control. # P < .05, ### P < .001, #### P < .0001, genotypes being compared indicated by brackets.

See Appendix D for enlarged eye images.

2.4 Discussion

HSPGs have long been known to play critical roles in the development and homeostatic plasticity of synapses102. Our previous work has shown that HSPGs also serve as negative regulators of basal autophagy in the muscle and fat body tissue of

Drosophila106. The present study extends the regulation of autophagy by HSPGs into the central nervous system. Our work also further highlights the influence of HSPG function on the pathology of neurodegenerative diseases, confirming a previously identified interaction between inhibition of HSPG biosynthesis and the progression of Alzheimer’s disease.

HSPGs negatively regulate autophagy in the central nervous system

We have shown, directly and indirectly, that inhibition of HS function is sufficient to provoke an autophagic response in neurons. Assessment of proteolytic release of free mCherry from an ectopically expressed mCherry-Atg8a fusion protein showed enhanced lysosomal processing of the parental protein with loss of sfl or ttv function, indicating enhanced autophagic flux. This effect was similar to that of enhancing autophagy through overexpression of wildtype Atg8a. Detection of this fusion protein using anti- mCherry antibodies did not afford the ability to distinguish between Atg8a-I, the free cytoplasmic form of the protein, and Atg8a-II, the PE-lipid conjugated form of the protein.

Increased levels of Atg8a-II are associated with enhanced autophagy, and measuring the ratio of the two forms of the protein is a common method of assessing autophagy

68 levels. Failure to identify this protein isoform might result from the previously reported poor recovery of Atg8a-II from neuronal tissue107.

Indirect assessment of autophagy through monitoring degradative substrates also indicated that loss of HS function resulted in enhanced neuronal autophagy.

Specifically, the accumulation of IUPs in the brain in response to oxidant exposure was prevented in animals expressing RNAi constructs targeting sfl or ttv. When autophagy was suppressed through RNA inhibition of Atg3, a protein required for assembly of the autophagosome, accumulation of IUPs following peroxide exposure was enhanced over the level seen in control animals. These results are in agreement with previous reported uses of the IUP assessment method. Overexpression of wildtype Atg8a with a stronger pan-neuronal Gal4 driver was previously shown to result in a decrease in IUPs following oxidant exposure108.

A lessened accumulation of autophagic substrates in animals with enhanced autophagy in comparison to controls would logically be expected. The cellular constituents damaged by oxidative stress are ubiquitinated to mark them for clearance by both autophagy and the proteasome. Therefore, following oxidant exposure, an increase in the total quantity of ubiquitinated substrates over animals that had not been exposed at all would still be expected, even when the basal level of autophagy has been increased. This is not what we observed, nor was it shown to be the case in the studies that first established the method139. While this curious effect was not addressed by the originating research group, it is possible that chronically enhancing basal autophagy primes animals to respond more efficiently to oxidative stress.

Figure 2.06: Determining the balance between pro-growth and pro-survival pathways, and between the generation and clearance of ubiquitinated substrates.

We theorize that HSPGs regulate autophagy through their control over cell-surface growth factor signaling inputs on mTOR. A Activation of mTOR promotes cell division and inhibits autophagy to a low baseline level, also restricting the types and quantity of substrates tagged for degradation. Cytotoxic stresses such as exposure to reactive oxidant species damage cellular components and activate stress response pathways that inhibit mTOR. This results in an increase in ubiquitin modification, required for the targeted clearance of damaged cellular components, and an increase in autophagy, required to clear the surfeit of degradation-targeted molecules. B When HSPG activity is reduced, we believe that mTOR is inhibited through an as-of-yet unidentified intermediary, likely the class I PI3K signaling cascade. This would result in an increase in the targeting of cellular components to degradative pathways via ubiquitination, while also increasing the basal level of autophagy. However, as the degree of effect on each of these two processes remains to be determined, it is difficult to predict the ultimate impact on the balance between targeting and clearance of degradative substrates, which is represented here by the total abundance of ubiquitinated substrates at a given point in time.

Our previous work demonstrates that inhibition of sfl or ttv causes an increase in ubiquitin modification and expression of ref(2)P in larval muscle tissue (Fig. 1.08). In adults with neuron-specific RNAi of sfl or ttv, IUP analysis indicated that the baseline IUP density in adult brain tissue of these animals is also increased (Fig. 2.02), although additional experimentation is required to accurately measure this trend. We feel that

HSPG-mediated negative regulation of autophagy operates through mTOR activation, given that many HSPG-dependent signaling events influence mTOR activity. mTOR sets the proteostatic balance by promoting protein synthesis while inhibiting autophagy and the targeting of substrates for degradation through ubiquitination. The particular degree to which ubiquitination and autophagic turnover are increased by HSPG biosynthetic inhibition has yet to be determined, and adding an environmental stressor that also increases autophagy further complicates the issue of balance in these two interrelated pathways (Fig. 2.06). Only by measuring levels of autophagy, levels of ubiquitination, and levels of catabolism, can we hope to shed mechanistic light on the very interesting result afforded by the oxidant exposure assay.

Increasing HSPG function aggravates age-associated failure of autophagic clearance and sensitizes animals to chronic oxidant exposure

In order to better understand the mechanism by which HSPGs regulate autophagy, we first needed to develop a model in which HSPG function was increased, which would allow for dissection of downstream autophagy regulation pathways. To this end we assessed a variety of phenotypes in a Drosophila sumf null mutant, a potential model of Multiple Sulfatase Deficiency and a negative regulator of HSPGs. The sumfΔ213 null mutation is homozygous viable and recapitulates a wing bristle phenotype caused

Figure 2.07: Deletion of sumf mimics a wing developmental error found in animals lacking the HSPG negative regulatory sulfatase gene sulf1.

A Drosophila wings are covered by several different types of bristles that receive and convey sensory information important for the process of flight. The outermost edge of the wing margin is decorated by a single row of thick mechanosensory bristles (MSB, yellow arrows). The next prominent row of bristles just medial to the wing margin consists of chemosensory bristles (CSB green arrows), which are thinner than the MSBs. In sumfΔ213 homozygous mutants, ectopic MSBs can been found in the CSB row of the wing margin (orange arrows). In sulf1 mutants, this phenotype is caused by errors in HSPG-controlled wnt signaling during developmental patterning of the presumptive wing tissue220. B In addition to transformation of cell type such that some bristles in the CSB row are incorrectly specified as MSBs, sumf null mutants also show a statistically significant increase in the number of properly specified CSBs, another phenotype found in sulf1 mutants.N= ; error bars represent SEM; ***, p <.001.

72 by loss of sulf1, the only 6-O-endosulfatase present in Drosophila (Fig. 2.07)220. This finding confirms that Drosophila sumf, like its mammalian counterpart, is required for catalytic activation of sulfatases, including those required for degradation of HSPGs.

Using the sumfΔ213 mutation, we found that accumulation of IUPs in the brain was drastically worsened in aging animals. In comparison to heterozygotes, complete loss of sumf function shortened lifespan and accelerated the age-associated accumulation of these autophagic substrates. At six weeks of age, sumfΔ213 homozygous animals had a

28% higher burden of IUPs than eight week old sumfΔ213 heterozygotes, which was likely linked to the early mortality of these animals. Furthermore, loss of sumf sensitized animals to death by chronic exposure to low levels of hydrogen peroxide. While this sensitization was the mildest phenotype observed in sumf mutants, it was opposed by drastically improved viability in sfl and ttv under the same conditions.

By these indirect measures, autophagy appears to be reduced in sumf null animals, as would be expected from an increase in HSPG function based on our model.

The consistent observation in all of these assays of effects in the opposite direction of

HSPG functional inhibition further support this model of autophagy regulation. Having established the sumfΔ213 mutation as a viable model of sulfatase deficiency and autophagic inhibition, it will be possible to move forward on work that elucidates the mechanism of autophagy regulation by HSPGs.

This model will also allow for a more thorough investigation of the contributions of reduced autophagy to MSD disease progression. Suppression of autophagy has already been identified as an element of MSD42 and other lysosomal storage diseases221, 222, although in most cases it remains unclear whether autophagic depression is a cause or an effect of disease progression. These questions can be addressed using complex

73 genetic models in Drosophila much more readily than in mice, the predominant model in which MSD has been studied.

While some lysosomal storage diseases are caused by defects in vesicular trafficking216, MSD and the mucopolysaccharoidoses are caused not by a general failure of lysosomal degradation but by an inability to degrade specific and limited types of substrates. While accumulation of these substrates in the cell might be unavoidable, the ultimately lethal accumulation of secondary substrates should not be.

We aim to test whether chemical and genetic means of enhancing autophagy downstream of the global suppression caused by accumulation of HSPGs is sufficient to reduce secondary accumulation and slow disease progression using the phenotypes described here as markers of viability and autophagic competency. Drugs such as rapamycin (serolimus) that can be used to enhance autophagy are already in clinical use for the treatment of other diseases. Such treatments could be quickly brought to bear in human patients with MSD and similar lysosomal storage diseases if they were shown experimentally to improve quality of life and extend life expectancy.

HSPGs: a novel target for autophagy-sensitive neurodegenerative disease

Neurons are highly dependent on autophagy for activity dependent rejuvenation of presynaptic proteins181, negative regulation of synaptic transmission223, and survival during a variety of toxic and physiological insults224-226. Identification of autophagy as a suppressor of neuron death in certain neurodegenerative diseases has led to the pursuit of pharmalogical activation of autophagy as a potential treatment. Our work here demonstrates that inhibition of HS biosynthesis can provide similar protection. The ability

74 of HS functional reduction to moderate the toxicity of presenilin supports previous findings in another model of Alzheimer’s disease.

Cre conditional knockout of Ext1, the mouse homolog of ttv, in the adult forebrain delayed neurodegeneration and death in PS1/APP transheterozygous mutant mice.

Loss of Ext1 improved clearance of amyloid-β and reduced deposition of the amyloidogenic Aβ42 isoform227. Extracellular Aβ fibrils are known to co-localize with cell surface and ECM HS chains in vivo228. In vitro, Aβ40 has been shown to inhibit heparanase-mediated degradation of HS229, suggesting that accumulation of amyloid fibrils in vivo in the proximity of HSPGs may interfere with their turnover. Finally, overexpression of heparanase in a PS1 mutant mouse model of Alzheimer’s disease was found to reduce extracellular accumulation of Aβ230. When taken together with our data, the literature suggests a fatal positive feedback spiral involving the interaction of

Aβ and HS. Aβ forms amyloid deposits around HSPGs, which become resistant to degradation. As HSPGs accumulate, autophagy is suppressed and Aβ clearance is reduced. Failure to clear Aβ leads to increased extracellular deposition in the form of Aβ amyloid fibrils, perpetuating the cycle and eventually causing inflammation and cell death.

In addition to serving as a focal point for senile plaque formation, cell surface neuronal HSPGs are also required for propagation of certain types of misfolded proteins between cells. In addition to extracellular Aβ plaques, Alzheimer’s disease is associated with intracellular aggregates of misfolded tau. Tau misfolding due to problems in protein turnover or somatic mutations can lead to aggregation within affected neurons. These aggregates can then be propagated to neighboring neurons in a heparan sulfate- dependent proteopathic seeding event, allowing the toxic misfolding and aggregation of tau to spread throughout the brain231. α-Synuclein tangles can propagate between

75 neurons using the same mechanism, and it is likely that other forms of aggregates capable of being moved into the extracellular space may do so as well.

The potential to slow neurodegenerative disease progression through autophagy bolstering interventions is currently being investigated from many angles.

Pharmaceutical induction of autophagy through a variety of mTOR dependent and independent mechanisms is known to protect against neurodegeneration in a variety of animal and cell culture models of Alzheimer’s, Parkinson’s, and Huntington’s diseases59,

61, 232-237. Even activation of autophagy through caloric restriction is neuroprotective in some disease models238. Identification of HSPGs as regulators of autophagy and as a potential pathological link between MPS and Alzheimer’s disease provides a new target for therapeutic development. MPSs are relatively rare disorders, and current treatment options are limited to enzyme replacement therapy and bone marrow transplantation, neither of which is effective in ameliorating the associated neurodegeneration. Extending a connection to a potential early intervention for Alzheimer’s disease provides a greater incentive to develop treatments that might also benefit children with MPS, MPSP and

MSD.

2.6 Materials and Methods

Flies were raised on standard cornmeal/agar food media at 25°C during under a

12 h day/12 h night cycle. While aging animals for IUP analysis, flies were maintained on standard food in culture vials, with no more than 35 animals per vial. The animals were flipped onto fresh food every third day.

Oregon-R served as the wildtype control for analysis of the traditional mutant alleles, while VDRC60100 the control for RNAi experiments. RNAi strains and a control

76 strain with the same genetic background were obtained from the Vienna Drosophila

RNAi Center (VDRC)77: UAS-Atg3 RNAi (101364), UAS-ttv RNAi (4871), UAS-w RNAi

(30033), and empty vector control (60100). UAS-mcherry-Atg8a (37750), Elav(C155)-

Gal4>UAS-DcrII (25750), EP-Atg8a (10107), UAS-Psn (8309) and the Drosophila

Transgenic RNAi Project (TRiP)78 lines UAS-sfl RNAi GLC01656 (50538), UAS-sfl RNAi

HMS00543 (34601), and UAS-mCherry RNAi (35785) were obtained from the

Bloomington Drosophila Stock Center (BDSC).

The sfl03844 and ttv00681 P-element insertion alleles were generated by the Berkeley

Drosophila gene disruption project239 and have been previously described102, 240-242. The sfl9B4 ethylmethanesulfonate-induced point mutation was generated and described by

Norbert Perrimon241.

The sumfΔ213 P-element excision mutant was a generous gift from Hiroshi Nakato,

University of Minnesota. Prior to experimentation, this fly stock was backcrossed 8 generations into our OregonR control strain to remove potential second site mutations.

The mutant allele was maintained through non-lethal PCR genotyping of individual female progeny.

2.61 Head Collection

Flies were collected in groups of 30-60 under CO2 anesthesia and transferred to

1.7ml microcentrifuge tubes. These tubes were promptly frozen in liquid nitrogen and stored at -80°C until all samples in an experiment were ready for protein extraction.

Tubes were kept on liquid nitrogen for the duration of the head collecting process. Each tube was manually disrupted to break the heads from the thorax. Brass sieves were chilled in liquid nitrogen and stacked with a No. 25 sieve (710μm, Hogentogler & Co. Inc.

#1546) on top to remove thorax and abdomen and a No.40 (425μm, Hogentogler & Co.

Inc. #1549) on the bottom to collect heads while allowing legs and other small debris to pass through. Heads were passed through a chilled funnel into a clean chilled tube containing four ceramic beads (2.8 mm Ceramic Bead Media 19-646-3; Omni

International). Heads were then either immediately processed for either protein or RNA extraction or stored at -80°C.

2.62 Protein Purification

For the mCherry conversion assay, heads were homogenized by ceramic bead agitation using the using a Bead Ruptor 24 (Omni International, Kennesaw, Georgia,

USA) in 100μl of SDS extraction buffer (2% SDS, 50mM Tris pH7.4, 1X protease inhibitor cocktail Complete [Roche, 10184600])139. Samples were spun down at

10,000rpm for 10 minutes and the supernatant was removed to a clean tube.

A two-step protein extraction was used for IUP analysis, modified from the method described by Cumming, Simonsen and Finley, 2008. Heads were homogenized by ceramic bead agitation using the Bead Ruptor 24 in 150ul of Tritonx-100 extraction buffer (1% Triton-X100, 1x PBS, 1X protease inhibitor cocktail Complete [Roche,

10184600])139. Samples were spun down at 10,000 RPM for 10 minutes at 4°C. The supernatant was drawn off and saved as the soluble protein fraction. The pellets were washed with 100ul of TritonX-100 before being resuspended in 100ul of 2% SDS extraction buffer. Supernatant from this extraction was saved and is referred to as the

TritonX insoluble fraction. Protein sample concentration was determined by BCA Protein

Assay (Pierce, 23227).

2.63 Western Blot

IUP assessment utilized 9% acrylamide gels, while the analysis of mCherry-

Atg8a utilized 6% acrylamide gels. PAGE was performed using the BioRad mini-

PROTEAN electrophoretic and transfer system (BioRad, Hercules CA) with 1mm plates.

Protein samples were prepared for loading and membranes were processed as described in section 1.65. Primary antibody incubation was performed overnight at 4°C using 1:2000 mouse anti-Mono- and Polyubiquitinylated Conjugates (FK2) (Enzo, BML-

PW8810-0100) or 1:2000 rabbit anti-mCherry (abcam, ab183628), and 1:3000 mouse anti-tubulin (Developmental Studies Hybridoma Bank, 12G10). Secondary antibody incubation (1:3000 HRP conjugated goat anti-mouse or –rabbit, Invitrogen, 31430 and

31460) was performed for 45 minutes at room temperature. ECL detection and band densitometry were performed as described in section 1.65. For loading analysis, a combination of Ponceau staining and stripping and reprobing with anti-alpha tubulin was used. Membranes were stripped for re-probing in mild stripping buffer (0.1% SDS, 1.0%

Tween20, 0.2M glycine, pH 2.2). All blotting experiments were performed using three biological replicates for each genotype under each condition. Lanes with the most numerically similar sample loading, as determined by densitometry, were chosen from within the biological triplicates for display.

2.64 Oxidant Exposure

For IUP analysis in oxidant exposed flies, 7-10 day old flies were exposed to yeast agar food media with or without 2.0% H2O2 for 24 hours at 25˚C. The yeast agar food media was 1% yeast extract, 1% sucrose, 1.2% agar. For food containing H2O2, the

79 media was cooled to 45°C before the addition of peroxide. After exposure, samples were processed as described in sections 2.61-.62.

Survival curves were generated by exposing 7 day old flies to yeast agar food media containing 1.5% H2O2. The number of animals dead (no longer moving) was recorded every 12 hours. All groups of animals were transferred onto fresh 1.5% H2O2 food every 48 hours to maintain a consistent level of oxidant exposure over the course of the survival challenge. For each genotype and gender, 30 additional flies were maintained on yeast agar food without peroxide as a survival control group, and were similarly flipped into fresh food media every 24 hours. Survivorship was greater than or equal to 99% in all of the control groups for every gender/genotype combination over the duration of the experiment.

2.65 Disease interaction studies

UAS-Psn is a 541 amino acid wildtype Presenilin coding sequence. This construct was overexpressed pan-neuronally under the control of Elav(C155)-

Gal4>UAS-DcrII. This Gal4-expressing line also incorporates UAS-based overexpression of DcrII, which is needed to elicit effective RNAi in neurons. The UAS-

Psn construct was expressed on its own and in combination with several RNAi lines.

Eyes were imaged using an Olympus BX53 compound microscope and Olympus cellSens Dimension software (Olympus Optical, Waltham, MA) to obtain optical sections at a 9μm step. Optical sections for each animal were assembled into a single composite image using Zerene Stacker (Zerene Systems LLC, Richland, WA). Eye area was measured in ImageJ and ommatidial organization was assessed using the Flynotyper

ImageJ plugin (http://flynotyper.sourceforge.net/)

2.66 Statistical Analysis

See section 1.66

Chapter 3

Inert Genetic Constructs Influence Behavior in Drosophila

3.1 Abstract

Autism spectrum disorder (ASD) encompasses a range of neurodevelopmental defects that present with varying levels of behavioral, linguistic, and intellectual impairments. The specific impairments associated with ASD are heterogeneous, reflecting the diverse genetic basis of this family of disorders. As ofJune 2017, 910 candidate genes have been identified through the widespread effort to understand the genetic basis of ASD. Due to the incredible diversity of function in these genes and the low recurrence of mutations affecting individual candidates, little progress has been made in understanding the biological processes through which ASD associated symptoms manifest.

Genes of clear importance to neurological function have been prioritized for study in animal models, while a vast array of other candidate genes are dismissed and ignored due to a perceived low cost to reward ratio. A better understanding of the roles these diverse candidate genes play in the etiology of ASD would likely shed light on the complex phenotypic heterogeneity seen in affected individuals. Developing a high throughput and low cost screening method would help reduce this barrier to the development of more diverse animal models.

A panel of behavioral assays was assessed for its utility in approaching this question in the genetically accessible model organism, Drosophila melanogaster. Use of

RNAi libraries to inhibit the expression of candidate genes quickly and cost-effectively was combined with the wealth of Drosophila behavioral analysis to determine the

82 feasibility of this system for a medium- to high-throughput screen. While Drosophila behavior has been used successfully to model autism-related behaviors in traditional mutants, these assays were highly sensitive to spurious effects from genetic background and insertion of inert genetic elements. While some genetic combinations appeared to exert effects above and beyond those associated with each construct alone, it is difficult to interpret the significance of these findings in the context of their relevance to disease.

It seems that while Drosophila behavior is a rich resource for assessing ASD-related phenotypes, RNAi based screening is by and large not compatible with this experimental paradigm.

3.2 Introduction

The human central nervous system, seat of our self-identity and consciousness, is the feature of Homo sapiens that most drastically sets us apart from even our closest relatives among the great apes. This remarkable structure, formed from delicately interwoven neurons, glia, neurites, and distal sites of innervation performs tasks both as mundane as controlling the homeostasis of survival and as exceptional as crafting words and motions into art. When injury, non-congenital developmental errors, or mutations disrupt brain homeostasis, the results are dramatic and often incompatible with life.

However, small changes to developmental programs and alterations in the function of a remarkably wide variety of genes can be well-tolerated in terms of survival and yet can cause quite noticeable alterations to the complex behaviors that humans have evolved to form and maintain social structures.

The ability to connect and communicate with one another through language and behavior is one of the most fundamental components of humanity, helping to facilitate

83 learning and allowing individuals to integrate themselves into family units and larger social structures. It should therefore come as no surprise that behavior may be one of the most genetically sensitive outputs of the human brain. Any deficit in an individual’s ability to read and interpret the subtleties of body language and vocal intonation within their culture can lead to isolation, depression, and general malaise. While social structures in most of the developed world are moving to become more inclusive of individuals who function at or beyond the margins of what is considered to be neurotypical behavior, understanding the contributions of potentially avoidable neurologically unfavorable factors such as metabolic or environmental conditions could aid in protecting the social satisfaction and wellbeing of populations at-risk for neurodevelopmental disorders.

3.21 Autism Spectrum Disorder: the face of neurodiversity

Autism Spectrum Disorder (ASD) is a clinical diagnosis given to individuals who present with a variety of behavioral impairments. These include abnormal verbal and non-verbal communication, motor stereotypies, ritualistic adherence to patterns and schedules, and restricted and obsessive interests. ASD is a wide-ranging diagnosis that incorporates that incorporates the previously separate diagnoses of autistic disorder,

Asperger’s disorder, childhood disintegrative disorder, and pervasive developmental disorder-not otherwise specified243. Approximately 83% of ASD diagnoses are comorbid with another chromosomal or metabolic syndrome244, most commonly fragile-X syndrome, trisomy 21, or tuberous sclerosis245-248. A growing understanding of the incredible genetic and symptomatic heterogeneity of ASD is what led to the unification of

84 all four previous subcategories of the disorder into a spectrum, in recognition of the previous divisions as uninformative and overly simplistic.

Regardless of what these neurodevelopmental deficits are called by clinicians, a number of critical questions remain. First and most important, what can be done to help affected individuals enjoy the same quality of life as neurotypical individuals? Second, what, if anything, can be done to treat or prevent the abnormal neuronal development that causes these defects in behavior? The Autism and Developmental Disabilities

Monitoring Network report for 2016 indicates a 1 in 68 incidence of childhood autism during the study period249. Childhood ASD is associated with an extra annual cost per child of approximately $17,000 for special education, therapy and other non- pharmaceutical interventions, and medical costs, in comparison to neurotypical children250. This figure increases with symptomatic severity, with the most aggressive behavioral interventions incurring an annual cost of $40,000-$60,000 per child251.

While public schools provide services for children with ASD through age 21, the need for support does not end in adulthood. Very few adults with ASD live independently, with most individuals requiring either full residential care or some form of supportive living accommodation252, 253. This requirement for continuous lifelong care combined with the low employment rate in the ASD community accounts for most of the lifetime cost of support for a single affected individual, an estimated $1.14 million for

ASD alone or $2.4 million for ASD with intellectual disability254.

In order to better treat individuals with ASD, a better understanding of the underlying biology of the associated behavioral disturbances are needed. There has been over a decade of sequencing and microarray genetic analysis of patients and their families to understand the genetic underpinnings of this highly prevalent neurodevelopmental disorder. The fruit of these efforts is an ever growing collection of

over 900 candidate genes255. While some of these candidates fall into pathways known to be involved in neuronal function, many of the genes have functions that are either unidentified or bear no immediately apparent relationship to neuronal function.

Animal models have been developed for a number of strong candidates, genes in which unique mutations have been identified across multiple patients, or that are associated with syndromic causes of ASD. Unfortunately, this tactic abandons the vast majority of genes that have few identified mutagenic events or unclear functional significance. The high risk and high cost of developing models for these less promising candidates is a major roadblock to better understanding both the true genetic diversity of

ASD and by extension the full scope of biological systems that influence neurodevelopment.

3.22 Neurobehavioral modeling in Drosophila melanogaster

Rats and mice are currently the preferred models for autism associated behaviors. Rats are naturally gregarious and playful, engaging in a variety of behaviors ideal for observation in ASD research256, 257. Mouse models of autism are more common due to their greater genetic accessibility. These models commonly exhibit deficits in learning and memory, changes in vocalization and activity, and motor stereotypies257, 258.

However, rodents are resource and time intensive model organisms. Furthermore, experiments involving rodents require careful ethical consideration and oversight. There is very little to justify using either of these model organisms to perform a widespread assessment of high-risk candidate genes, but neither do these facts justify ignoring a large portion of the data provided through human genetics.

Another layer of biological validation is needed to bridge the gap between human genetics data and in-depth assessment with a mammalian model. Behavioral screening in Drosophila melanogaster could potentially provide the means to build such a bridge.

Like mice and rats, Drosophila engage in a variety of social behaviors ranging from intricate courtship displays to boxing and tussling. The song and dance of Drosophila courtship have been described and studied for over half a century259. These innate behaviors are highly plastic, changeable by training and individual experience to an amazing degree260. The patterns of aggressive behavior are similarly complex, with males and females exhibiting unique forms of intra-gender aggression261, 262. Male flies use a combination of chemosensory signals and combat to develop hierarchical relationships263, 264, while female aggression is determined by prior socialization and resource availability265.

Several Drosophila models of ASD candidate genes have successfully identified alterations in a variety of behaviors including aggression and courtship. Models of fragile-X syndrome in particular have been studied extensively. Loss of the Drosophila homolog of fragile-X mental retardation 1, dFmr1, is associated with defects in social interaction266, olfactory response267, circadian rhythm, and learning and memory268.

These behavioral and cognitive phenotypes recapitulate symptoms seen in affected human patients, and have even been used to identify effective treatments that improved the sleep and activity phenotypes in flies268.

Other monogenic Drosophila ASD models have seen similar success. Deletion of neuroligan was found to increase spacing between flies and reduce courting and aggression behaviors269. Mutations in rugose, homolog of the ASD candidate gene

Neurobeachin, identified synaptic structural and electrophysiological defects, along with hyperactivity, impaired habituation, and increased inter-fly distance in the social spacing

87 assay270. Neurofibramin loss of function has similarly been found to impair learning and memory271, induce hyperactivity and disrupt circadian rhythmicity272, and results in excessive grooming behavior273. All of these studies strongly support the use of

Drosophila to model neurobehavioral disorders.

3.33 Assembling a Drosophila behavioral screen for ASD candidate gene validation

Providing biological validation for the many mid- and high-risk candidate genes on a large scale would require a high-throughput screening paradigm with low operational costs. In order to assess the largest number of genes in the shortest time possible, the Gal4-UAS binary expression system was selected to model the genetic changes associated with autism. This system uses the yeast transcription factor Gal4 to express genetic elements placed under control of a UAS promoter, which is not recognized by any native Drosophila transcription factors79. The Gal4 open reading frame is placed under control of various endogenous promoters, termed Gal4-drivers, to exert spatio-temporal control over the expression of UAS-promoted genetic elements.

The -Gal4 and UAS- elements are kept in separate fly stocks to circumvent potential genetic effects associated with loss of gene function, including lethality, sterility, and natural selection for spontaneous reversion mutations.

Rather than generating and characterizing traditional mutants, UAS-RNAi constructs were obtained from Drosophila stock centers to inhibit the expression of candidate genes. Libraries of RNAi constructs targeting every known coding gene in the

Drosophila genome are readily available to the research community. Ready access to these interfering RNA stocks provides a vast genetic toolkit that has been widely

88 leveraged in forward-genetic screens. Researchers are currently working to generate similar libraries of overexpression constructs for all known open reading frames, which are being made available as they are completed, providing further opportunities for screening.

Behavioral assays were chosen with a similar consideration for speed and ease of use. Ideal methods would be reproducible and sensitive enough to detect potentially subtle alterations in behavior. In combination, the methods would ideally provide the ability to cross-check results to enhance the specificity of identified behavioral defects.

Output from the assays should not require much technical training to analyze and should be robust against unintentional failures of human objectivity.

The Drosophila Activity Multibeam Monitoring system

Hyperactivity and sleep disturbances are commonly identified in individuals with a primary ASD diagnosis. An estimated 30-50% of ASD cases also exhibit symptoms

Attention Deficit Hyperactive Disorder (ADHD), and co-occurance of these disorders is associated with significantly worse patient outcomes274, 275. Both ASD and ADHD are highly associated with sleep disturbances, predominantly sleep-onset insomnia, but also sleep movement and breathing disorders276-278.

Locomotor activity and circadian rhythmicity are among the best characterized and most readily assessed of all Drosophila behaviors. The ready availability of the

Drosophila Activity Multibeam Monitoring (DAMM) system (Fig 3.01 A) and the high occurrence of sleep and activity disturbances in ASD rendered circadian rhythm monitoring a clear candidate for inclusion in this methodological screen. Furthermore,

89 data collection via this method is fully automated. Once flies are inserted into tubes and placed into the monitor, no further user input is required for a period of several days.

The DA multibeam monitor uses 17 individual infrared (IR) beams spaced at

3mm intervals along a glass tube to track the activity of a single fly (Fig. 3.01 B). Each multibeam monitor can hold 16 such tubes. When a fly passes through one of the monitored locations, the IR beam at that location is blocked from reaching its receptor.

Each incidence of this event, termed ‘beam breaking’, is recorded as activity count.

Drosophila cannot visually perceive IR light, so monitor beams do not alter sleep or spontaneous locomotion. The data collection software individually tracks beam breaks for every beam on every tube, producing a rich dataset with information on total activity, entrainment to the light-dark cycle, and positioning of every fly monitored.

Drosophila orchestrate their daily activity in a pattern that responds to the natural photoperiod exerted upon them by their environment. In the laboratory setting

Drosophila exhibit a highly stereotyped bimodal diurnal activity cycle. Spontaneous locomotor activity increases in anticipation of ‘sunrise’ and ‘sunset’, with peaks of high activity occurring just after lights on and lights off, as seen in Figure 3.01C. Spontaneous activity levels began to rise between one and two hours prior to the beginning of the light cycle, reaching a sharp peak within 30 minutes of ‘sunrise’ and declining steadily thereafter for a period of three to four hours. An increase in activity of similar slope was seen in the three to four hours before the onset of the dark cycle, with a sharp peak in activity in the 15 minutes after ‘sunset’. The evening activity peak was followed by a steep decline in activity over approximately one hour. Between these activity peaks, wild type animals became largely inactive.

….

Figure 3.01: Circadian Activity pattern of the Fruit Fly.

A The TriKinetics Drosophila Activity Multibeam Monitor (DAMM) B Schematic representation of a single slot in the DAMM (for example, box in A, transverse section). A single fly in a glass tube with food (left, yellow), plugged with cotton (white, right). LED lights on the top of the monitor originate IR beams (red dashed lines) that are detected by receivers embedded in the bottom of the monitor. When a fly ‘breaks’ a beam by standing in or moving through the path of the light, the receiver records a single instance of instance movement, regardless of the duration of time for which the beam is interrupted. C The normal circadian rhythmicity of spontaneous locomotor activity in Drosophila melanogaster, with total activity binned at 15 minute intervals. The light period is indicated by the yellow segment of the line, while the dark period is indicated by the black line segment. Time is measured in Zeitgeber hours, wherein time zero is set to the onset of the light period.

Startle-induced negative geotaxis

Negative geotaxis, or movement in opposition of gravity, is a deeply ingrained behavior of Drosophila. When given the choice between moving with or against gravity in a simple maze, Drosophila will choose to move upwards nearly 75% of the time279. This preference for negative geotaxis is known to rely on genetic inputs from all three of the main gene-bearing chromosomes280, which indicates that genetic screening using this behavior stands a good chance of producing observable behavioral changes.

Threatening stimuli such as looming shadows, sudden changes in surrounding air currents or mechanical disruption induce a faster negative geotactic escape response that is more time-efficient to score, particular in flies that are capable of locomotion but display low levels of spontaneous activity. Adherence to this behavioral paradigm requires both sensorimotor integration and sufficient strength and coordination to successfully climb, jump, and fly.

Startle Induced Negative Geotaxis (SING) is a behavior that is both easy to assess and controlled by rich underlying biological processes. Manual assessment of

SING requires only materials that are already on hand in any Drosophila laboratory.

While manual performance of the assay is subject to unintended bias and technical differences between human experimenters, these issues can be easily circumvented by genotypic blinding and completion of all replicates within an experiment by the same individual. While this behavior is heavily dependent on the physical ability of flies to climb, performing SING in combination with circadian activity monitoring affords the ability to rule out locomotor deficiencies as a cause for poor performance.

The social spacing assay: an assessment of preferred flocking distance

While courtship and aggressive behaviors are known to have rich genetic underpinnings, scoring of these behaviors is time consuming and somewhat subjective.

However, the preference of groups of flies to ‘flock’ together when at rest281 provides the opportunity to measure a more easily interpreted underlying behavior282. Higher order interactions between individuals are dependent on the close proximity of conspecific individuals. Reproductive success and larval survival in Drosophila is influenced by population density and clustering of individuals. While overpopulation of a food source can lead to harmful competition for limited resources among larvae283, a certain degree of population concentration is required in order for larvae to generate conditions ideal for feeding and resistance to disease and parasites284. Social clustering of adults is requisite for mating and clustered egg laying that leads to a sufficiently large larval population.

While group clustering and food scavenging behaviors of wild flies respond to a variety of complex variables285-287, a preference for group closeness can be observed in controlled laboratory conditions as well. Fly social groups still exhibit a preferred flocking distance in behavioral observations arenas in the absence of food or other attractive stimuli such as differences in vertical height or lighting282, 288. Observation of fmr1 mutant flies revealed a decrease in attempts to approach other animals and in the amount of time spent near other animals. This finding indicates that this behavior is genetically- based and could potentially represent decreased interest in initiating social interactions seen in ASD. Furthermore, the chambers required to observe this behavior can be easily fabricated in a variety of shapes and designs from clear acrylic plastic pieces (as seen in Fig. 3.06), rendering it accessible for development in a screen.

3.3 Results

3.31 The Drosophila Activity Multibeam Monitor: a high resolution method to track spontaneous locomotor activity and rhythmicity

Average daily activity levels and adherence to the normal rhythmicity of spontaneous activity were measured in high resolution using the DAMM apparatus.

Initial investigation of daily activity levels revealed hyperactivity as a common effect of inert genetic constructs (Figure 3.02). Of the seven UAS- promoter driven expression constructs tested, five of them significantly increased average daily activity levels in the absence of the Gal4 transcription factor required to activate their expression. Of the two

Gal4 expression lines tested, the brain specific elavC155-Gal4>UAS-DcrII line also induced hyperactivity in the absence of additional genetic manipulations.

The dramatic disruption of spontaneous activity by the elements of the binary expression system when isolated from one another necessarily complicated interpretation of phenotypes in offspring with both transcription factor and expression construct present. Nevertheless, a few observations of interest were made.

Overexpression of ras homolog enriched in brain (Rheb) to emulate the effects of

Tuberous Sclerosis, a syndromic cause of ASD, was found to significantly reduce spontaneous activity in comparison both to wild type controls and animals heterozygous for elavC155-Gal4>UAS-DcrII (Figure 3.03).

While average daily activity exhibited a keen sensitivity to insertional effects, rhythmicity of behavior was only found to be altered in one circumstance. RNAi inhibition of polypeptide GalNAc transferase 5 (pgant5, homolog of the ASD candidate genes

GALNT13 and GALNT14) caused a mirroring of the evening activity period (Figure

3.04). Affected animals did not gradually increase activity in anticipation of ‘sunset’, and

Figure 3.02: Multiple elements of the binary expression exert isolated effects on spontaneous activity levels. Average daily activity was observed using the Drosophila multibeam monitor. CS is the behavioral standard line, Canton Special. Each of the binary expression elements displayed here were heterozygous offspring of a cross between the original stock and CS. *** P<.001; N=16 animals per genotype.

Figure 3.03: Overexpression of ASD candidate gene Rheb reduced activity.

As shown previously, the pan-neuronal Gal4 promoter fusion elavC155-Gal4>UAS-DcrII causes hyperactivity on its own, while the UAS-Rheb overexpression construct did not exhibit an independent effect on activity levels. Overexpression of Rheb through the combination of these lines however caused a reduction in spontaneous activity levels. *** P<.001, compared to control; ### P<.001 compared to +/Rheb; N=16 animals per genotype.

Figure 3.04: Circadian rhythmicity is disrupted by RNAi of an ASD candidate gene.

Animals heterozygous for only da-Gal4 or UAS-pgant5RNAi exhibit normal entrainment to a defined light period. Combining the two elements results in animals that fail to predict onset of the dark period and exhibit increased latency in their response to cycling from light to dark and dark to light. N=16 animals per genotype. 96 exhibited a gradual tapering off of activity following the evening peak that was similar in slope to the decline after the morning activity peak. These animals also experienced a delayed response to the onset of the dark period itself, with activity only beginning to increase after 15 minutes of darkness and reaching its maximum 30 minutes after

‘sunset’. The lack of spurious genetic effects on circadian rhythmicity suggests this particular behavioral output as a valuable measure of normal brain connectivity.

3.32 Startle-inducted negative geotaxis: an indicator of danger responsiveness and locomotor capability.

Among wild-type control animals, approximately 85% of individuals can climb at least 8cm within 10 seconds of mechanical disruption of the testing chamber (Figure

3.05 A). Expression of a neurodegenerative polyglutamine expanded peptide, UAS-

MJDtrQ78, confirmed the reliance of this activity on motor ability, exhibiting drastically reduced success in this climbing task.

However, startle-induced geotaxis was also somewhat sensitive to insertional effects of expression constructs, and activity of the elavC155-Gal4>UAS-DcrII combination line. Unlike the elav-Gal4 8760 construct (Fig 3.05A), the elavC155-

Gal4>UAS-DcrII combination line was associated with increased climbing ability (Fig.

3.05 B). Of the five UAS-expressed constructs tested rigorously, four of them significantly disrupted geotactic climbing in the absence Gal4 transcriptional activation

(Fig. 3.05 A & B). When placed under the control of elavC155-Gal4>UAS-DcrII, UAS-

Rheb and UAS-FMR1RNAi produced dramatic reductions in the rate of successful startle- induced geotaxis beyond the independent spurious effects of each individual genetic element. In the case of UAS-Rheb, this outcome was predicted by the overall reduction

Figure 3.05: Off- and On-target genetic influences on startle induced negative geotaxis.

Startle induced negative geotaxis (SING) relies on both proper locomotor function and proper sensorimotor integration, effects that can be separated by pre-screening against locomotor defects. A Neurodegeneration caused by overexpression of MJDtrQ78 results in poor negative geotaxis. B Isolated genetic expression elements influence success in SING, however strong On- target effects can still be identified with overexpression of Rheb and inhibition of Fmr1. C Locomotor activity pre-screening of Fmr1RNAi found normal mobility, suggesting a behavioral basis for their SING deficit. ** P<.01, *** P<.001 in comparison to CS. ### P<.001 in comparison to UAS-only heterozygous construct control, comparison indicated by brackets. N=10 groups per genotype (9-12 animals per group) 98 in locomotor activity previously identified using the DAMM system (Fig 3.03). The effect of UAS-FMR1RNAi was pinpointed as a specifically behavior-dependent phenotype, as these animals exhibited normal activity levels in the DAMM system (Fig. 3.05 C).

3.33 The Social Spacing Index- a measure of preferred flocking distance within Drosophila social groups.

Wild type control animals preferred to stay close to one another (Fig. 3.06 A), with the majority of animals located within 5mm (two body lengths) of at least one other animal. Very few animals were further than 10mm from another animal (Fig 3.06 B).

After sorting the distances between nearest neighbor pairs into 5mm bins, the social spacing index (SSI) was calculated by subtracting the constituency of the second bin

(flies between 5 and 10mm away from their nearest neighbor) from the constituency of the first bin (flies between 0 and 5mm from their nearest neighbor).

The behavior of control animals was consistent between independent experiments, with no significant difference in SSI between groups (P=.235, one-way

ANOVA). However, this behavior was quite variable in other genotypes (Figure 3.7 A, collection of CS [in grey], and ED SSI average across trials). The next most commonly tested animals bore only the elavC155-gal4>UAS-DcrII driver. These animals exhibited more behavioral variation between experiments, with significantly different SSI scores between groups (P= .001, one-way ANOVA)(in white, Figure 3.7 A). In two of the four trials using this Gal4 driver, the construct alone significantly reduced SSI in comparison to the control.

Most of the other constructs tested also negatively impacted flocking distances in the social spacing observational chambers. The other two other Gal4 driver lines tested

99 using this method also caused reduced SSI scores, as did all of the transcriptionally un- induced UAS-promotor driven expression constructs (Fig. 3.07 B). SSI scores were often quite variable within groups of the same genotype and experiment, which was not drastically improved upon by doubling or tripling the sample size. Only overexpression of

Rheb caused a change in SSI that was significant in comparison to the isolated effects of the genetic components used in the experiment (Fig. 3.07 B).

Figure 3.06: Normal flocking distances in the social spacing assay.

CS control flies exhibit a strong preference for a close flocking distance. A After exploring a horizontal chamber, flies come to rest and engage in stationary grooming behaviors. B Control flies generally settling within two body lengths (5mm) of at least one other fly. N=15 groups of 35- 40 animals

100

5 5 7 8 15 15 18 19 7 15 15 6 ` 10 16 15

Figure 3.07: Performance in the social spacing assay is variable in stocks bearing inserted genetic elements.

A CS behavioral controls perform quite reliably in the social spacing assay, but animals heterozygous for the inserted elavC155-Gal4>UAS-DcrII expression constructs exhibited highly variable behavior within and between trials. B All of the expression elements tested, when heterozygous and isolated from the other half of the binary expression system, negatively influenced behavior. This included da-Gal4 and elav-Gal4 8760, Gal4 promoter fusions that did not affect activity levels or SING success rate. While the combination of elavC155-Gal4>UAS- DcrII and UAS-Rheb reduced SSI beyond the level of either genetic component on its own, the combinatorial effect was in the same direction as the individual effects and most likely resulted from the additive off-target effects. ** P<.01, *** P<.001 compared to CS. # P<.05 compared to UAS-Rheb/+. Brackets clarify statistical comparisons. Numbers at the bottom of the bars indicate sample size (the number of groups tested). Each group contained 35-40 animals.

101

3.4 Discussion

The phenotypic heterogeneity of Autism Spectrum disorders has long complicated both diagnosis and the development of meaningful interventions. Diagnosis of ASDs is still dependent on behavioral interviews rather than physiological tests. As efforts to sequence both exomes and whole genomes of individuals on the spectrum advance, the enormous genetic heterogeneity identified shines a light on the exquisite sensitivity of complex human behaviors to genetic change. Development of animal models will be necessary to supplement human studies, as recurrent genetic hits are rare in the ASD individuals that have been sequenced so far. A suite of time and cost efficient behavioral analyses in Drosophila melanogaster with the Gal4-UAS binary expression system was appraised for its ability to perform medium to high throughput screening of ASD candidate genes. Such a screening method would provide an invaluable low-risk mechanism for identifying a variety of functionally diverse genetic targets for more in-depth biological study.

Assessment of technical utility for three behavioral analysis methods

Two of the three behavioral methods hold some potential for ASD candidate gene screening, although each suffers from its own shortcomings. Circadian rhythmicity and spontaneous activity monitoring via the DAMM system provided the greatest degree of experimental control due to its high level of automation. This experimental method requires little direct time from experimenters to set up per trial and produces raw data that can be used to analyze sleep patterns in addition to overall activity levels and circadian entrainment. Throughput is limited only by the number of monitors purchased

102 and the availability of incubator space that can be dedicated solely to the running of experiments. Overall, this method generated reliable data that was reproducible across trials and left experimenters free to pursue other work while the DAMM system collected data in the background, making it an ideal method for use as part of a behavioral screening paradigm.

The second method assessed, startle-induced negative geotaxis, was a fruitful addition to the information obtained through activity monitoring. For the purpose of this study, the tapping stimulus used to induce the startle response was applied by hand.

This necessitated genotypic blinding to prevent experimenter bias in the application of physical force to the testing chamber. Due to variations in technique between individuals, experiments needed to be performed entirely by a single person, which limited throughput.

A method for automated Rapid Iterative Negative Geotaxis (aRING) that eliminates the variation introduced by human error has been published289, however the equipment and software required for the method were produced in-house and could not be reproduced with the skillset available at the time of this methodological investigation.

Even in the absence of an automated method, behavioral variation within genotypes was quite low and behavior was reproducible between independent trials. Obtaining the setup for aRING would reduce the number of uncontrolled variables and dramatically increase throughput, further elevating the utility of this method.

The social spacing assay put forth the greatest number of pitfalls and provided the least useful information of the three methods. Loading of animals into the observation arena had to be performed by mouth applicator to circumvent the negative physiological and behavioral effects of cold and CO2 anesthesia. Use of compressed air was attempted as an alternate means of introducing flies to the arena, but was more

103 difficult to control and often proved injurious to the animals. A method for automated image capturing that produced sufficiently high quality images of multiple chambers without influencing behavior was similarly inaccessible.

Due to the lack of automation, this assay was cumbersome and time consuming to perform, rendering it largely inappropriate for large scale screening. Furthermore, the data generated using this method had the largest intra-genotype variance, requiring very large sample sizes to achieve any statistical power. Use of this method appeared to be largely incompatible with the genetic reagents being used and provided no added value to the behavioral screening. Review of pertinent literature suggests that this method is more reliable in background-isogenized mutants270.

Hyperactivity as a common effect of genetic background or insertion site and as a critical intersection between all three behavioral methods

In all three behavioral analysis methods, a notable proclivity towards hyperactivity was seen in animals bearing unexpressed UAS-promoted genetic elements. This confounding effect was most problematic in the interpretation of total daily activity, where it was difficult to ascertain whether an increase in activity in an experimental group was due to the targeted genetic effect or merely an additive effect of the hyperactivity induced by the -Gal4 and UAS- expression constructs. However, the

DAMM system was able to offer a more specific phenotype in the form of circadian rhythmicity, which was not altered by any off-target genetic effects.

Circadian rhythm entrainment relies on the ability of pacemaker neurons to integrate time cues such as light, temperature, and nutrient availability into the transcriptional cycles that control the endogenous biological rhythm of organisms.

104

Failure to entrain normally to the photoperiod of an environment could reflect any number of functional changes in synaptic plasticity, neurotransmission, or photoreception. Inhibition of pgant5 caused a dramatic change in the photoperiod entrainment of affected animals. Of particular note, a similar mirroring of the evening activity peak and delayed response to changes in chamber lighting were observed in the monogenic ASD neurofibromin null mutant model273. The commonality of this phenotype with an established ASD model provides a degree of biological validity to the candidate status of pgant5 and its human homologs GALNT13 and GALNT14.

The addition of DAMM system monitoring information to the results of the startle induced negative geotaxis assay was further critical in separating changes in behavioral patterns from basic motor competency. Unsurprisingly, genetic elements that produced overall increases in the level of spontaneous daily activity also increased the rate of successful geotaxis. Similarly, low spontaneous activity levels were associated with poor performance in the geotaxis assay. However, a unique behavioral defect was identified in fmr1 deficient animals through cross-comparison of its performance in these two assays.

Inhibition of fmr1, a well-established ASD model, did not reduce spontaneous activity in comparison to wild type controls, but resulted in a dramatic failure of startle- induced negative geotaxis. Failure of this basic reflexive response to a physical threat suggests a failure either in the ability to interpret environmental stimuli or in sensorimotor integration required to elicit a response to the stimulus. Correctly identifying this phenotype as uniquely behavioral rather than purely physiological was due entirely to the use of these two techniques in tandem, and exemplifies how a simple assay like

SING can be used to study complicated behavioral phenotypes.

105

The social spacing assay was also quite sensitive to hyperactivity due to its reliance on the establishment of a stable distribution of animals at rest for the ensuing analysis. As determined by the quantification of total daily activity in the DAMM system, the majority of the UAS-expression constructs being tested induced hyperactivity in animals even in the absence of the transcription factor required to express them.

Furthermore, the pan-neuronal Gal4 driver used to express the genetic regents also raised average activity levels. While wild type animals stopped exhibiting exploration and chasing behaviors and settled into a stable and stationary distribution after approximately 20 minutes, most of the -Gal4 and UAS- construct bearing animals took at least twice as long to become stationary.

Several of the genotypes tested never reached a stable distribution, with 5-10% of flies continuing to exhibit chasing behavior after an hour in the observation chamber.

While these flies represented the minority of individuals being observed, their persistent activity disturbed the other flies in the arena and often broke up groups that were previously clustered. While this behavior could itself be of interest, there was no clear mechanism by which to score it for comparison between genotypes. This significant complication rendered the social spacing assay unusable for the purposes of this foray into behavioral screening.

How to screen behavior in Drosophila: lessons from genetic personality

Drosophila behavioral analysis has been used for over 30 years to dissect the genetic basis of behavior. Behavioral analysis has largely focused on the study of single gene mutations in a consistent genetic background290. Nearly all behavioral assays use the wildtype stock Canton Special (CS) as the positive behavioral control, and compare

106 that control to mutants generated or back-crossed into the CS genetic background. Few behaviors have been examined for interfering effects of genetic background.

First, the learning and memory mutant mushroom body miniature (mbm) was found to be suppressed by back-crossing into CS291. The mbm1 allele was initially studied in the genetic background of Berlin, another wildtype stock. After outcrossing to replace the Berlin genetic background with that of CS, the neuroanatomical deficit associated with this mutation was suppressed to the point of being indistinguishable from controls. The accompanying olfactory learning phenotype was rescued to a lesser but still significant degree.

Positive phototaxis, the tendency of Drosophila to move toward a source of light, was used for a more intensive investigation of behavioral differences in CS and white1118, an eye color mutant background commonly used to generate fly stocks292. This study revealed that white1118 flies not only exhibit lower rates of positive phototaxis than CS flies, but also a much broader variation in the phototactic behavior of individual flies of the same genotype. The phototactic personality of individual flies was consistent throughout their lifetime and was not heritable- the offspring of individuals selected and mated on the basis of high, low, or intermediate phototactic personality exhibited the same range of overall positive phototaxis responses seen in the general population.

The cause of this anomalous behavior was linked to the normal function of the white (w) gene. w is an ATP-binding cassette transporter required for import of tryptophan and guanine293. Tryptophan and guanine serve not only as precursors for the pigment that produce eye color in Drosophila, but also as precursors for the synthesis of serotonin and dopamine294, 295. Drosophila with w* and other eye color mutations have significantly altered concentration and localization of neurotransmitters and their precursors in the brain296. Introducing a w+ allele into the w1118 genetic background or

107 feeding supplemental 5-hydroxytryptophan, a downstream serotonin precursor that bypasses the need for w, was able to rescue positive phototaxis and reduced the variance of individual phototactic personality to CS levels292.

These findings are of particular note in the context of behavioral screening using the Gal4-UAS binary expression system, as all of the lines used were generated in a w1118 or y1w1 genetic background and had not been backcrossed to remove the null w allele. w* is a common genetic background due to the inclusion of a truncated w+ open reading frame on introduced genetic elements that partially restores gene function. This allows for the use of eye color as an easy positive selection marker for incorporation of genetic elements into the genome. The mini-white fragment is expressed from an endogenous promoter, independent of whatever mechanism controls expression of the introduced genetic element. These mini-white marker genes do not fully restore w function; in addition to being only partially functional due to truncation of the coding sequence, transcription of the reporter construct is subject to local regulatory effects at the site of insertion. The degree of w function in the animals tested depended on the number of mini-white bearing genetic elements present and the unique degree of mini- white expression from the constructs being tested.

Beyond the vagaries of w gene function, other genetic variables were likely contributing to the behavioral variation observed. Each Gal4-promoter fusion and UAS- expression element has the potential to induce deleterious mutations at the site of its insertion. This issue could potentially be circumvented by only using stocks that place transgenes in the exact same insertion site through targeted integration. Unfortunately, using only genetic elements placed in a specific insertion site would limit the scope of genes available for study. Even after controlling for insertion site and genetic background, fly stocks are vulnerable to the accumulation of second site mutations.

108

Existing stocks must be perpetually cultivated in a living line as Drosophila cannot be reliably retrieved from frozen samples. Stocks therefore experience an unavoidable degree of natural genetic drift over time.

Additional investigation would be necessary to determine the specific cause of the background shifts in behavior before a screen such as the one described here could be pursued. If introducing a wildtype w locus were found sufficient to reduce behavioral variability, throughput in the circadian rhythm assay could be improved by potentially reducing the number of individuals required in a sample. This requires only a single generation of breeding to replace the X chromosome, which would not be an insurmountable barrier to a screening process.

To determine whether or not insertion site mutations caused the behavioral disturbances identified in this study, the constructs would need to be isogenized with a w+ control line. Isogenization is a time consuming process in which experimental strains are intensively inbred to eliminate the effects of genetic drift and bring animals back to a state of exceptionally high genetic similarity within a population. Once an isogenized control stock is established, the experimental stock must be backcrossed to the isogenized control for eight generations. The meiotic recombination that occurs over eight generations is sufficient to replace approximately 99% of the original genome.

Single fly selection and non-lethal PCR genotyping is used to identify animals that retain the genetic element of interest over this long process.

This course of action was not pursued due to the enormous time commitment it would require. The goal of this study was to determine whether behavioral analysis could be used in combination with the affordable and expansive genetic toolkit available through the popular Gal4-UAS binary expression system in Drosophila. If isogenization of stocks is required in order to obtain clear results in behavioral analysis, a screen with

109 any degree of useful throughput is not possible. Isogenization is commonly incorporated as part of the standard procedure in developing and assessing traditional mutants, in which behavioral analysis has been used to great effect.

Since the time of this study, the CRISPR method of mutagenesis has emerged and been optimized for use in a number of model organisms, including Drosophila297.

This method greatly reduces the time required to generate mutations in genes of interest, and the breeding and screening process required to obtain such animals is still much less time consuming in Drosophila than it is in mice. Using CRISPR to generate mutants for behavioral analysis would slow throughput and require more careful pre- screening and selection of genes for study. However, this method would circumvent issues that plague RNAi based methods, including variable efficiency of transcript suppression and off-target effects.

Gal4-UAS binary expression has been used extensively to study a wide variety of morphological phenotypes. It was therefore not unreasonable to consider using this accessible and time-saving genetic toolset to investigate behavior. Unfortunately, this body of work demonstrates that behavior is indeed exquisitely sensitive to relatively minor genetic variations to degree not seen in physiological defects, mirroring what we know from human genetics studies. Since 2011 when this project was initiated, the number of ASD candidate genes has exploded from just over 300 to 845 at the time of this writing255. These candidate genes encompass a truly monumental range of functionally diverse pathways and proteins, and an equally monumental effort will be required to understand how such a complex and varying array of genes work together to permit normal human behaviors and socialization.

110

3.6 Materials and methods

Flies were reared under standard culture conditions, as discussed in section 2.6.

Canton Special (CS) flies were used as the behavioral wildtype standard. In experiments testing the effect of overexpression or RNA inhibition through use of the Gal4-UAS binary expression system, CS flies were crossed to the Gal4 line and the UAS-line individually, and progeny heterozygous for each of the genetic elements in isolation served as controls for the ‘experimental genotype’ in which animals bore both elements of the expression system.

da-Gal4 (55851), elav-Gal4 (8760), elav(C155)-Gal4>UAS-DcrII (25750), UAS-

Rheb (9688), UAS-akt CA (8294) and UAS-MJDtrQ78 (8141) were obtained from the

Bloomington Drosophila Stock Center (BDSC). UAS-Fmr1 RNAi (110800), UAS-mGluR

RNAi (103736), UAS-pgant5 RNAi (2629), and UAS-sfl RNAi (5070) were obtained from the Vienna Drosophila RNAi Center (VDRC)77.

3.61 Circadian Rhythm:

Total locomotor activity and the circadian rhythmicity of that activity were measured using the Drosophila activity multibeam monitor (DAMM, TriKinetics, Waltham

MA). Flies were reared in the incubator in which the behavior was to be recorded under standard culture conditions and were collected under CO2 anesthesia promptly after eclosion. Virgin males were selected for analysis to eliminate the maximum number of behavioral variables possible. Single male flies were loaded into each tube by mouth applicator. The DAM system was set up as previously described298. In brief, individual

80x5mm glass tubes with approximately 1 cm of standard Drosophila food medium in

111 one end were loaded with a single male fly and inserted into the DAMM apparatus in a humidity-controlled incubator. The incubator was closed and undisturbed for a period of approximately six days while data was collected automatically by the associated

DAMSystemMB106 software. Activity data was automatically split into 24 hour chunks beginning at 2:01am and ending at 2:00am the following day. Data collected between loading flies into the monitor and the first instance of 2:00am was discarded, allowing time for the flies to recover from experimental setup and acclimate to the testing environment. Data was collected for five full 24 hour periods under a 12:12 light dark cycle. The testing incubator was located in a small room, of which the lights were kept off and the door closed during the dark period to prevent any external light from leaking into the incubator.

Activity data was binned by summing into 15 minute intervals. Matching time intervals were averaged across the five days of 24 hour monitoring to generate the circadian profile of each individual fly. Those individual profiles were then combined into an average circadian profile for the genotype. To calculate average daily activity, the number of recorded beam breaks from the 15 minute time bins were summed for each fly’s individual circadian profile and those sums were averaged by genotype.

3.62 Startle Induced Negative Geotaxis:

Animals were reared under standard culture conditions, virgin males were collected upon eclosion. Animals were kept in groups of ten in standard Drosophila culture vials for 2-5 days prior to testing. One day prior to testing, a person other than the experimenter assigned four digit random numbers to the testing groups to prevent

112 any unconscious bias during the procedure. A genotype identification key was then provided to the experimenter after data collection was complete.

Flies were removed from the incubator and allowed at least 30 minutes to acclimate to the laboratory environment prior to testing. Genotype blinded testing groups of ten flies were transferred into testing arenas made of two empty culture vials taped together. A red line was drawn in marker 8cm above the bottom of the chamber. Flies were allowed five minutes to acclimate to the testing chamber before beginning the assessment. The testing chamber was tapped firmly but gently on a padded surface to collect flies at the bottom of the chamber and to induce a startle response. The number of animals above the 8cm mark was recorded 10 seconds after tapping. After a one minute rest period, the same challenge was repeated. Each group of flies was tested a total of 10 times at one minute intervals to generate an average success rate for the group

3.63 Social Spacing

Arenas in which to observe fly behavior were fabricated by the Penn State machine shop from clear acrylic plastic. Flies were reared under standard laboratory conditions and collected promptly after eclosion into groups of 35-40 individuals, with an approximate 1:1 ratio of males and females. Each sample of flies was maintained together in a culture vial for 2-5 days to allow for final maturation adults and to ensure full behavioral recovery from CO2 anesthesia. Flies were flipped into fresh food vials every other day prior to experimentation. The procedure for Social Spacing Analysis was adapted from Simon et. al 2012282.

113

Experiments were performed between the hours of 1:00 and 4:00 PM on sunny to partly cloudy days when ambient natural light from the windows was sufficient.

Overhead fluorescent lights were turned off to reduce glare and avoid effects of phototactic preference of fly positioning in the arena. Noise and foot traffic near the testing area were strictly monitored and limited to prevent external stimuli from distressing the animals during experimental observation.

Flies were removed from their rearing incubators 30 minutes prior to testing to allow the animals to acclimate to the laboratory environment in their culture vials before being moved into the observation arenas. Animals were gently transferred into the testing chambers by suction and puffs of air applied through a mouth applicator with a two point cloth barrier system to prevent aspiration or injury to the animals. Once inside the chamber, animals were allowed 20 minutes to explore the chamber and come to a stationary distribution. After 20 minutes, if more than three flies were still moving in the chamber, animals were allowed up to 20 additional minutes before image capture to become settled.

Images of the chambers were captured by hand from above using a digital camera. Three to six chambers were filled and observed concurrently to maximize the amount of trials performed each day. The genotypes being tested were rotated through each position on the bench in sequential trials to mitigate the effects of any unforeseen environmental variables. Experiments took 2-4 days to complete. An equal number of sample groups from each genotype were tested each day to control for variations in testing conditions. Each sample group was used only once and all flies were naïve to the testing conditions. After each trial, chambers were disassembled, cleaned with 70% ethanol to remove waste and odorants, then rinsed in distilled water and dried before re- use.

114

Images were analyzed using the open source lispix image analysis program

(https://www.nist.gov/services-resources/software/lispix) as described in Simon 2012.

Nearest neighbor distances were binned at 5mm intervals using the histogram function in Minitab 16 (Minitab Inc., State College PA). Loss of some flies prior to experimentation and during chamber loading was inevitable. Groups with fewer than 30 individuals as ascertain through image analysis were excluded, as group size influences performance in the assay.

This assay was performed in a number of environments, including two incubators and an environmentally controlled room. The animals displayed acute sensitivity to environmental stimuli such as vibration and fans turning on and off in incubators and the environmental chamber, even after being raised and stored prior to testing in those environments. Due to the consistent quiet and stability of the bench top, it was chosen as the final testing environment for all of the social spacing data presented herein.

115

Chapter 4

Future Research Directions

4.1 Ongoing Research on the HSPG Project

The information presented in Chapter 1 has already been published in the

Journal of Autophagy (Appendix F), and subsequent progress on the project is presented in Chapter 2. A nearly infinite number of questions remain to be asked concerning how HSPGs regulate autophagy, and the implications of this relationship with respect to health, disease, and aging. We are continuing to investigate a select few of those questions in order to publish a second article chronicling this story. The current research goals are focused on further linking the biological phenotypes associated with the sfl, ttv, and sumf traditional mutants with changes in the level of autophagy in these animals.

First and foremost, the biochemical consequences of the sumfΔ213 mutant need to be more thoroughly characterized. This gene was chosen to advance our understanding of the relationship between HSPGs and autophagy because we needed a method by which to increase HS activity as a supplement to our work using mutants and RNAi to decrease HS activity. The sumf mutant should result in a global failure of sulfatase activity, which should cause an accumulation of HSPGs and increase the overall levels of HS sulfation. Fitting with a model in which HSPGs negatively regulate autophagy, sumf mutants should have increased levels of highly sulfated HSPGs and lower levels of autophagy.

Preliminary HPLC analysis of total HSPG content from sumfΔ213 heterozygous and homozygous mutant animals bears out our expectations for this model (Fig. 4.01).

116

Compared to control animals, sumfΔ213 homozygous mutants (Fig. 4.01 A and C respectively) have an enormous increase in the total concentration of HSPGs. This provides direct support for the effectiveness of sumf loss of function as a method by which to increase the concentration of HSPGs in vivo in Drosophila. Among the sulfated

HS disaccharides, the trend towards particularly large increases in the level of 6O- sulfation provides biochemical support for the loss of sulf1-dependant 6O-endosulfatase activity in these animals, which was assessed phenotypically in Fig. 2.07.

117

Figure 4.01: sumf loss of alters the quantity and sulfationΔ level of HSPGs in DrosophilaΔ. Δ

HPLC disaccharide profiling of HSPGs extracted from whole adult flies, courtesy of Dr. Hidenao Toyoda. -S, unsulfated disaccharide; NS, N-sulfated disaccharide; 6S, 6O-sulfated disaccharide; NS6S, N- and 6O-disulfated disaccharide; 2SNS, 2O- and N-disulfated disaccharide; 2SNS6S, 2O-, N- and 6O-trisulfated disaccharide. A The wildtype control, OreR, shows the basal level of HSPGs. NS and NS,6OS disaccharides are the predominant forms of sulfation. B In sumfΔ213 heterozygous mutants, there is no clear change in the overall quantity of HSPGs or in the proportional prevalence of sulfated disaccharides. C In sumfΔ213 homozygous mutants, there is an enormous increase in the overall quantity of HS disaccharides of all kinds. Unsulfated (-S) disaccharides remain the most prevalent form of HS, however, all forms of 6O-sulfated disaccharides (6S, NS6S, and 2SNS6S) appear to exhibit disproportionately large increases in prevalence. This is in keeping with the identification of wing patterning defects associated with loss of sulf1 function, the only Drosophila Heparan Sulfate 6O-endosufatase.

Quantitative and statistical analysis of this HPLC data is forthcoming from our collaborators. Additional HPLC experiments to replicate the preliminary findings and complement them with analysis of other sumfΔ213 alleles still need to be performed.

Ideally, an in-depth assessment of HS disaccharide profiles in the sfl and ttv heterozygous mutant adult should also be performed to clarify the biochemical underpinnings of the ROS tolerance exhibited by these animals.

118

The HSPG mutant panel (heterozygous mutations in sfl and ttv and heterozygous and homozygous mutations in sumf) shows that decreased HSPG biosynthetic capacity leads to improved survival during an ROS exposure challenge, while failure to degrade

HSPGs has the opposite effect (Fig 2.03). Preliminary data suggests that the HSPG mutant panel also shows some trends towards corresponding changes in the accumulation of IUPs in ROS exposed verses unexposed brain tissue (Appendix C).

Replication sets of this experiment are still required to clarify and build upon these results, which were not statistically significant, possibly due to the exceptionally dilute protein samples obtained from the preliminary experiment. Firm data indicating a change in the clearance of ubiquitinated substrates in whole head protein extracts from the mutant panel will provide insight to the biological basis for the changes in survival during the ROS challenge.

In addition to measuring changes in IUP clearance, a more direct method of assessing the level of neuronal autophagy in these animals is still required. To this end, the neuron-specific GAL4 driver appl-GAL4, the ubiquitious expression driver da-GAL4, and the UAS-mCherry-Atg8a constructs need to be bred into the sumf mutant background. For the sfl and ttv mutants, which can only be examined in the heterozygous condition, breeding one of the two expression constructs into the mutant background is sufficient. Breeding these genetic components into the mutant backgrounds will allow us to perform the fluorophore conversion assay to directly measure autophagic degradation both in the brain and globally in these mutants.

Finally, we intend to further confirm the importance of autophagy in the heightened susceptibility of sumf mutants to ROS-induced lethality by administering the mTOR inhibitor Rapamycin during a repetition of the ROS survival challenge. According to our model (Fig. 2.06), the ROS sensitivity in these mutants is due to constitutive

119 suppression of autophagy by an overabundance of HSPGs. Direct inhibition of mTOR downstream of the HSPG-mediated signaling defect should be sufficient to rescue the autophagy deficit and the sensitivity to ROS exposure.

4.2 Future Aims for the HSPG project

Section 4.1 outlines the work remaining before the next logical follow-up manuscript on the HSPG/autophagy project could be submitted for publication. However, the potential for this project extends well beyond that end point. At least three major branches are currently available for this project. The first possibility is a more thorough investigation of the interaction between HSPGs and neurodegenerative disease. The second possibility would pursue the original goal of developing sumf loss of function as a springboard to launch a mechanistic investigation of the signaling pathways linking HS activity to autophagy regulation. The third is an option we are already pursuing through collaboration, moving our theory into mouse models of MSD and HS biosynthetic inhibition.

Expanding the scope of HS/disease interactions

So far, only a single method of modeling a single disease has been investigated for interaction with HS biosynthetic inhibition (Fig. 2.05). As discussed in section 2.4

(subsection 3), the literature suggests a particularly close relationship between HS and the progression of Alzheimer’s disease. The presenilin overexpression model of AD was selected for the first interaction study because of this link and because of the ease with which the model could be used and the beautiful clarity of the phenotype it imparts.

120

However, there are many more models of AD and a variety of other neurodegenerative diseases available for study in Drosophila.

More of these options should be pursued to expand our understanding of how changes in HS activity contribute to disease pathogenesis. I personally feel it would be quite worthwhile to study interactions using the heterozygous sfl and ttv mutants. If even a 50% decrease in gene dosage of these biosynthetic enzymes is potent enough to palliate disease in these models, it is more likely that a pharmacological or gene-therapy intervention to adjust the level of HS activity could be developed and implemented safely in humans. Furthermore, identifying a broad range of more common neurodegenerative diseases that are responsive to reductions in the level of HS would create greater incentive to develop medical interventions that would accomplish this aim, treatments that could also be used to improve the lives of children with the much less common MPS family of lysosomal storage diseases.

Leveraging the sumf mutant for analysis of HSPG downstream signaling pathways required for the regulation of autophagy

Several years’ worth of effort has finally paid off in the identification of measurable phenotypes in sumf loss of function animals. These findings provide critical support for our model of HSPG-mediated autophagic regulation. However, the goal of assessing the sumf mutant and developing a Drosophila model for MSD was always to use this model as a platform from which to investigate the signaling events downstream of HSPG activity that are required for the regulation of autophagy. We now have ample evidence indicating that sumf loss of function is inhibiting sulfatase activity and HSPG degradation, as was predicted based on the close homology of the Drosophila sumf

121 gene to its human and mouse orthologs. With this evidence in hand, it has become logically reasonable to invest the effort required to undertake such a study.

As previously discussed, the numerous HS-dependent growth factor signaling pathways that operate through activation of class I PI3K kinase and AKT would suggest that HSPGs influence autophagy through the canonical mTOR-dependent pathway. As a first foray into testing this hypothesis, it would therefore be wise to combine sumf loss of function with loss of function in components of the Drosophila PI3K-Akt signaling pathway. This could be accomplished through inhibition of Pi3K92E (the p110 subunit of the Drosophila Class I PI3K) or Akt itself, or overexpressing Pten (the phosphatase that convers PIP3 to PIP2). The primary difficulty of this undertaking would be the identification of an appropriate phenotype to use as a marker of autophagy in these animals.

If the mCherry-Atg8a conversion assay provides a clear autophagy phenotype in sumf mutants, it would most likely be appropriate for this project as well. However, if the conversion assay is not sensitive enough for use in interaction studies, the peroxide survival challenge might be a viable alternative. Unfortunately, while the fluorophore conversion assay can be used in larvae, adapting the peroxide survival challenge for larvae would be incredibly difficult due to the physical fragility of larvae, their burrowing behaviors, and the introduction of developmental processes as a new variable. The ability to observe a phenotype in adult flies verses larvae is an important consideration for this study, as it will determine when, where, and how the inhibition of critical developmental genes such as Pi3K92E and Akt can be introduced.

Ubiquitous inhibition of Pi3K92E or Akt during embryonic and larval development is highly lethal, and even restriction of the effect to specific tissues can result in drastic changes in animal size and viability. If the animals need to reach adulthood before being

122 tested, the inhibition of PI3K/AKT signaling would need to be achieved either pharmacologically or through a temporally controlled method. If RNAi is being used as the mechanism of inhibition, temporal control can be exerted though use of a drug- inducible expression system or by introduction of the temperature sensitive GAL4 inhibitor, GAL80299. That being said, the need to introduce additional genetic elements into the sumf mutant background would add a great deal of time and difficulty to this process. Identifying the proper phenotype in which to carry out these experiments and the preparatory legwork for this project will be challenging require a substantial time commitment, which is why we aren’t already attempting to address these issues with the intention of including them in the next publication on this project.

4.3 Problematic Design Features of the Drosophila Screen for Biological Validation of ASD Candidate Genes.

As described in section 3.4, the attempt to use the GAL4-UAS binary expression system to model genetic lesions associated with ASD primarily served to highlight the incompatibility of this genetic method with the types of behavioral assays used to study

ASD-like phenotypes in fruit flies. This project, as it currently stands, does not merit continuation. Furthermore, a more thorough understanding of statistics and experimental design would have drastically shortened the duration of this course of experiments.

First and foremost, this research project should have been initiated using well- studied traditional mutants with known behavioral defects such as fmr1 to optimize the assays being tested. Initial verification of assay utility should have been followed by a thorough assessment of the GAL4-UAS expression system genetic components to confirm that conditions and sample sizes established using wildtype and positive control

123 genotypes were applicable to our study. The first experimental step beyond establishing these controls should have involved RNAi inhibition of the same genes affected by the positive control mutants used for assay optimization. Only after a reasonable effort at understanding the experimental system had been made should we have jumped to the selection of the panel of diverse and less stringently vetted candidate genes. While the goal of this project was to provide biological validation for these higher risk candidates, it was illogical to jump directly to working with them.

This project was spurred onward in large part by promising findings that turned out to be caused not by the effects of the genetic manipulations being implemented, but rather by independent insertional effects of the constructs being used. Having come directly to this project from working with NMJ morphology, I was used to assuming these constructs would have no effect on their own, and only including one of a number of potential control genotypes in any given experiment. In the lab’s study of NMJ morphology, GAL4 constructs or UAS-expression constructs have not been found to influence the phenotypes under investigation. However, the literature upon which I based this project set forth a clear precedent for the susceptibility of behavioral analysis to a wide variety of generally innocuous sources, so more aggressive attempts to control for genetic background effects was warranted from the outset.

For the social spacing assay in particular, optimization of the chamber design, procedural parameters, and sample size happened over the period of at least a year.

Had proper statistical methods been applied to the results of this assay earlier, the unsuitability of the social spacing assay would have become clear in a quarter of this time. Doubling and even tripling the initial sample size of the social spacing assay was not sufficient to garner consistent performance from non-wildtype control genotypes, such as animals expressing only the neuron specific GAL4 driver elav (as seen in figure

124

3.07). The only effect of dramatically increasing sample size was a marked drain on time and resources, as it necessitated the generation of huge crosses to produce enough progeny for the test.

Experimentation with the DAMM system began with a similarly insufficient degree of concern for inclusion of all possible genetic controls. As the DAMM system required some monetary investment up front, we started off with a single monitor capable of holding 16 animals. Due to the small number of animals that could be observed in a single trial, we tended to include only the experimental genotype (containing both of the binary expression elements being used) with only one of the appropriate genetic controls

(either the GAL4 driver alone or the UAS-element alone, but not both of these controls) and a wildtype control.

While I attempted to control for the limitations of time and equipment by assessing one of the genetic controls in the DAMM system while testing the other control in a concurrent social spacing experiment, this proved a highly insufficient effort as the insertional effects of a single construct did not necessarily produce an effect in both assays. Overall, the initial hit-and-miss approach to the inclusion of all the necessary genetic controls delayed the inevitable recognition that the genetic methods being used were inappropriate for the screen.

The goal of this project was to provide a method of quickly and cheaply vetting

ASD candidate genes to help diversify the development of vertebrate models away from the most popular and obvious candidate genes. However, the incompatibility of the

GAL4-UAS binary expression system with behavioral assessment complicates the interpretation of any data generated from such a screen and pares down the types and number of assays that can be relied upon.

125

It is uncertain whether genetic background or the insertion of the expression elements was to blame for the behavioral changes described in this thesis, however, isogenization of genetic backgrounds is a process that requires at least seven generations of backcrossing into the control genetic background, accompanied by careful genotyping to keep track of the inserted genetic element. Therefore, it is of little import which genetic variable is at fault, since isogenization of a large library of RNAi lines for a screen would pose an insurmountable barrier to progress. The only avenue to approach a large scale screen would have to rely upon traditional mutants, which would restrict throughput significantly. As such, we have no plans to continue this project at this time.

126

References

1. Esko JK, K.; Lindahl, U. Proteoglycans and sulfated glycosaminoglycans. In: Varki A, Cummings R, Esko J, Freeze H, Hart G, Marth J, eds. Essentials of Glycobiology. Cold Spring Harbor, NY: Cold Spring Harbor Press, 1999:469-77. 2. Roden L. In: Lennarz WJ, ed. The Biochemistry of Glycoproteins and Proteoglycans. New York: Plenum Press, 1980. 3. Filmus J, Selleck SB. Glypicans: proteoglycans with a surprise. J Clin Invest 2001; 108:497-501. 4. P. D. Yurchenco YC, G. C. Ruben. Self-assembly of a high molecular weight basement membrane heparan sulfate proteolycan into dimers and oligomers. The Journal of Biological Chemistry 1987; 262:17668-76. 5. Murdoch AD, Liu B, Schwarting R, Tuan RS, Iozzo RV. Widespread expression of perlecan proteoglycan in basement membranes and extracellular matrices of human tissues as detected by a novel monoclonal antibody against domain III and by in situ hybridization. The journal of histochemistry and cytochemistry : official journal of the Histochemistry Society 1994; 42:239-49. 6. Esko JD, Selleck SB. Order out of chaos: assembly of ligand binding sites in heparan sulfate. Annu Rev Biochem 2002; 71:435-71. 7. Wolpert L. Cellular basis of skeletal growth during development. Br Med Bull 1981; 37:215-9. 8. Lin X. Functions of heparan sulfate proteoglycans in cell signaling during development. Development 2004; 131:6009-21. 9. Gallet A, Staccini-Lavenant L, Therond PP. Cellular trafficking of the glypican Dally-like is required for full-strength Hedgehog signaling and wingless transcytosis. Dev Cell 2008; 14:712-25. 10. Yan D, Lin X. Shaping morphogen gradients by proteoglycans. Cold Spring Harb Perspect Biol 2009; 1:a002493. 11. Chen Y, Struhl G. Dual roles for patched in sequestering and transducing Hedgehog. Cell 1996; 87:553-63. 12. McLellan JS, Yao S, Zheng X, Geisbrecht BV, Ghirlando R, Beachy PA, et al. Structure of a heparin-dependent complex of Hedgehog and Ihog. Proc Natl Acad Sci U S A 2006; 103:17208-13. 13. Eugster C, Panakova D, Mahmoud A, Eaton S. Lipoprotein-heparan sulfate interactions in the Hh pathway. Dev Cell 2007; 13:57-71. 14. Jackson RL, Busch SJ, Cardin AD. Glycosaminoglycans: molecular properties, protein interactions, and role in physiological processes. Physiological reviews 1991; 71:481-539. 15. Kjellen L, Lindahl U. Proteoglycans: structures and interactions. Annu Rev Biochem 1991; 60:443-75. 16. Maccarana M, Sakura Y, Tawada A, Yoshida K, Lindahl U. Domain structure of heparan sulfates from bovine organs. Journal of Biological Chemistry 1996; 271:17804- 10. 17. Lyon M, Deakin JA, Mizuno K, Nakamura T, Gallagher JT. Interaction of hepatocyte growth factor with heparan sulfate. Elucidation of the major heparan sulfate structural determinants. J Biol Chem 1994; 269:11216-23.

127

18. Dietrich CP, Nader HB, Straus AH. Structural differences of heparan sulfates according to the tissue and species of origin. Biochem Biophys Res Commun 1983; 111:865-71. 19. Proteoglycans: structure, biology, and molecular interactions. New York: Marcel Dekker, 2000. 20. Ashikari-Hada S, Habuchi H, Kariya Y, Itoh N, Reddi AH, Kimata K. Characterization of growth factor-binding structures in heparin/heparan sulfate using an octasaccharide library. J Biol Chem 2004; 279:12346-54. 21. Kreuger J, Salmivirta M, Sturiale L, Gimenez-Gallego G, Lindahl U. Sequence analysis of heparan sulfate epitopes with graded affinities for fibroblast growth factors 1 and 2. J Biol Chem 2001; 276:30744-52. 22. Pye DA, Vives RR, Hyde P, Gallagher JT. Regulation of FGF-1 mitogenic activity by heparan sulfate oligosaccharides is dependent on specific structural features: differential requirements for the modulation of FGF-1 and FGF-2. Glycobiology 2000; 10:1183-92. 23. Ono K, Hattori H, Takeshita S, Kurita A, Ishihara M. Structural features in heparin that interact with VEGF165 and modulate its biological activity. Glycobiology 1999; 9:705-11. 24. Feyzi E, Trybala E, Bergstrom T, Lindahl U, Spillmann D. Structural requirement of heparan sulfate for interaction with herpes simplex virus type 1 virions and isolated glycoprotein C. Journal of Biological Chemistry 1997; 272:24850-7. 25. Dierks T, Schmidt B, von Figura K. Conversion of cysteine to formylglycine: a protein modification in the endoplasmic reticulum. Proc Natl Acad Sci U S A 1997; 94:11963-8. 26. Frese MA, Milz F, Dick M, Lamanna WC, Dierks T. Characterization of the human sulfatase Sulf1 and its high affinity heparin/heparan sulfate interaction domain. J Biol Chem 2009; 284:28033-44. 27. Tang R, Rosen SD. Functional consequences of the subdomain organization of the sulfs. J Biol Chem 2009; 284:21505-14. 28. Ai X, Do AT, Lozynska O, Kusche-Gullberg M, Lindahl U, Emerson CP, Jr. QSulf1 remodels the 6-O sulfation states of cell surface heparan sulfate proteoglycans to promote Wnt signaling. J Cell Biol 2003; 162:341-51. 29. Ai X, Kitazawa T, Do AT, Kusche-Gullberg M, Labosky PA, Emerson CP, Jr. SULF1 and SULF2 regulate heparan sulfate-mediated GDNF signaling for esophageal innervation. Development 2007; 134:3327-38. 30. Viviano BL, Paine-Saunders S, Gasiunas N, Gallagher J, Saunders S. Domain- specific modification of heparan sulfate by Qsulf1 modulates the binding of the bone morphogenetic protein antagonist Noggin. J Biol Chem 2004; 279:5604-11. 31. Danesin C, Agius E, Escalas N, Ai X, Emerson C, Cochard P, et al. Ventral neural progenitors switch toward an oligodendroglial fate in response to increased Sonic hedgehog (Shh) activity: involvement of Sulfatase 1 in modulating Shh signaling in the ventral spinal cord. J Neurosci 2006; 26:5037-48. 32. Seffouh A, Milz F, Przybylski C, Laguri C, Oosterhof A, Bourcier S, et al. HSulf sulfatases catalyze processive and oriented 6-O-desulfation of heparan sulfate that differentially regulates fibroblast growth factor activity. Faseb J 2013; 27:2431-9. 33. Yue X, Li X, Nguyen HT, Chin DR, Sullivan DE, Lasky JA. Transforming growth factor-beta1 induces heparan sulfate 6-O-endosulfatase 1 expression in vitro and in vivo. J Biol Chem 2008; 283:20397-407.

128

34. Li J, Kleeff J, Abiatari I, Kayed H, Giese NA, Felix K, et al. Enhanced levels of Hsulf-1 interfere with heparin-binding growth factor signaling in pancreatic cancer. Molecular cancer 2005; 4:14. 35. Lai J, Chien J, Staub J, Avula R, Greene EL, Matthews TA, et al. Loss of HSulf-1 up-regulates heparin-binding growth factor signaling in cancer. J Biol Chem 2003; 278:23107-17. 36. Cosma MP, Pepe S, Annunziata I, Newbold RF, Grompe M, Parenti G, et al. The multiple sulfatase deficiency gene encodes an essential and limiting factor for the activity of sulfatases. Cell 2003; 113:445-56. 37. Diez-Roux G, Ballabio A. Sulfatases and human disease. Annual review of genomics and human genetics 2005; 6:355-79. 38. Landgrebe J, Dierks T, Schmidt B, von Figura K. The human SUMF1 gene, required for posttranslational sulfatase modification, defines a new gene family which is conserved from pro- to eukaryotes. Gene 2003; 316:47-56. 39. Settembre C, Arteaga-Solis E, McKee MD, de Pablo R, Al Awqati Q, Ballabio A, et al. Proteoglycan desulfation determines the efficiency of chondrocyte autophagy and the extent of FGF signaling during endochondral ossification. Genes & Development 2008; 22:2645-50. 40. Guerra WF, Verity MA, Fluharty AL, Nguyen HT, Philippart M. Multiple sulfatase deficiency: clinical, neuropathological, ultrastructural and biochemical studies. Journal of neuropathology and experimental neurology 1990; 49:406-23. 41. Settembre C, Annunziata I, Spampanato C, Zarcone D, Cobellis G, Nusco E, et al. Systemic inflammation and neurodegeneration in a mouse model of multiple sulfatase deficiency. Proceedings of the National Academy of Sciences of the United States of America 2007; 104:4506-11. 42. Settembre C, Fraldi A, Jahreiss L, Spampanato C, Venturi C, Medina D, et al. A block of autophagy in lysosomal storage disorders. Hum Mol Genet 2008; 17:119-29. 43. Stipanuk MH. Macroautophagy and its role in nutrient homeostasis. Nutrition reviews 2009; 67:677-89. 44. Xie X, Le L, Fan Y, Lv L, Zhang J. Autophagy is induced through the ROS-TP53- DRAM1 pathway in response to mitochondrial protein synthesis inhibition. Autophagy 2012; 8:1071-84. 45. De Duve C, Wattiaux R. Functions of lysosomes. Annual review of physiology 1966; 28:435-92. 46. Inoki K, Li Y, Zhu T, Wu J, Guan KL. TSC2 is phosphorylated and inhibited by Akt and suppresses mTOR signalling. Nat Cell Biol 2002; 4:648-57. 47. Sancak Y, Bar-Peled L, Zoncu R, Markhard AL, Nada S, Sabatini DM. Ragulator- Rag complex targets mTORC1 to the lysosomal surface and is necessary for its activation by amino acids. Cell 2010; 141:290-303. 48. Russell RC, Yuan HX, Guan KL. Autophagy regulation by nutrient signaling. Cell research 2014; 24:42-57. 49. Chang JT, Kumsta C, Hellman AB, Adams LM, Hansen M. Spatiotemporal regulation of autophagy during Caenorhabditis elegans aging. eLife 2017; 6. 50. Marsin AS, Bouzin C, Bertrand L, Hue L. The stimulation of glycolysis by hypoxia in activated monocytes is mediated by AMP-activated protein kinase and inducible 6- phosphofructo-2-kinase. J Biol Chem 2002; 277:30778-83. 51. Qiang L, Wu C, Ming M, Viollet B, He YY. Autophagy controls p38 activation to promote cell survival under genotoxic stress. J Biol Chem 2013; 288:1603-11.

129

52. Sanli T, Steinberg GR, Singh G, Tsakiridis T. AMP-activated protein kinase (AMPK) beyond metabolism: a novel genomic stress sensor participating in the DNA damage response pathway. Cancer biology & therapy 2014; 15:156-69. 53. Zhang H, Bosch-Marce M, Shimoda LA, Tan YS, Baek JH, Wesley JB, et al. Mitochondrial autophagy is an HIF-1-dependent adaptive metabolic response to hypoxia. J Biol Chem 2008; 283:10892-903. 54. Feng D, Wang B, Wang L, Abraham N, Tao K, Huang L, et al. Pre-ischemia melatonin treatment alleviated acute neuronal injury after ischemic stroke by inhibiting endoplasmic reticulum stress-dependent autophagy via PERK and IRE1 signalings. Journal of pineal research 2017; 62. 55. Su KH, Cao J, Tang Z, Dai S, He Y, Sampson SB, et al. HSF1 critically attunes proteotoxic stress sensing by mTORC1 to combat stress and promote growth. Nat Cell Biol 2016; 18:527-39. 56. Su KH, Dai C. mTORC1 senses stresses: Coupling stress to proteostasis. Bioessays 2017; 39. 57. Spilman P, Podlutskaya N, Hart MJ, Debnath J, Gorostiza O, Bredesen D, et al. Inhibition of mTOR by rapamycin abolishes cognitive deficits and reduces amyloid-beta levels in a mouse model of Alzheimer's disease. PLoS One 2010; 5:e9979. 58. Zhang X, Yang H, Yue S, He G, Qu S, Zhang Z, et al. The mTOR inhibition in concurrence with ERK1/2 activation is involved in excessive autophagy induced by glycyrrhizin in hepatocellular carcinoma. Cancer medicine 2017. 59. Zhang ZH, Wu QY, Zheng R, Chen C, Chen Y, Liu Q, et al. Selenomethionine Mitigates Cognitive Decline by Targeting Both Tau Hyperphosphorylation and Autophagic Clearance in an Alzheimer's Disease Mouse Model. J Neurosci 2017; 37:2449-62. 60. Zhou X, Yue GG, Chan AM, Tsui SK, Fung KP, Sun H, et al. Eriocalyxin B, a novel autophagy inducer, exerts anti-tumor activity through the suppression of Akt/mTOR/p70S6K signaling pathway in breast cancer. Biochemical pharmacology 2017. 61. Sarkar S. Regulation of autophagy by mTOR-dependent and mTOR-independent pathways: autophagy dysfunction in neurodegenerative diseases and therapeutic application of autophagy enhancers. Biochem Soc Trans 2013; 41:1103-30. 62. Mizushima N, Yoshimori T, Ohsumi Y. The role of Atg proteins in autophagosome formation. Annu Rev Cell Dev Biol 2011; 27:107-32. 63. Stolz A, Ernst A, Dikic I. Cargo recognition and trafficking in selective autophagy. Nat Cell Biol 2014; 16:495-501. 64. Burman C, Ktistakis NT. Autophagosome formation in mammalian cells. Semin Immunopathol 2010; 32:397-413. 65. Anding AL, Baehrecke EH. Cleaning House: Selective Autophagy of Organelles. Dev Cell 2017; 41:10-22. 66. Levine B, Mizushima N, Virgin HW. Autophagy in immunity and inflammation. Nature 2011; 469:323-35. 67. Kraft C, Peter M, Hofmann K. Selective autophagy: ubiquitin-mediated recognition and beyond. Nat Cell Biol 2010; 12:836-41. 68. Slobodkin MR, Elazar Z. The Atg8 family: multifunctional ubiquitin-like key regulators of autophagy. Essays in biochemistry 2013; 55:51-64. 69. Kirkin V, Lamark T, Sou YS, Bjorkoy G, Nunn JL, Bruun JA, et al. A role for NBR1 in autophagosomal degradation of ubiquitinated substrates. Mol Cell 2009; 33:505-16.

130

70. Behrends C, Sowa ME, Gygi SP, Harper JW. Network organization of the human autophagy system. Nature 2010; 466:68-76. 71. Bjorkoy G, Lamark T, Brech A, Outzen H, Perander M, Overvatn A, et al. p62/SQSTM1 forms protein aggregates degraded by autophagy and has a protective effect on huntingtin-induced cell death. J Cell Biol 2005; 171:603-14. 72. Isogai S, Morimoto D, Arita K, Unzai S, Tenno T, Hasegawa J, et al. Crystal structure of the ubiquitin-associated (UBA) domain of p62 and its interaction with ubiquitin. J Biol Chem 2011; 286:31864-74. 73. Pankiv S, Clausen TH, Lamark T, Brech A, Bruun JA, Outzen H, et al. p62/SQSTM1 binds directly to Atg8/LC3 to facilitate degradation of ubiquitinated protein aggregates by autophagy. J Biol Chem 2007; 282:24131-45. 74. Ciuffa R, Lamark T, Tarafder AK, Guesdon A, Rybina S, Hagen WJ, et al. The selective autophagy receptor p62 forms a flexible filamentous helical scaffold. Cell reports 2015; 11:748-58. 75. Kim PK, Hailey DW, Mullen RT, Lippincott-Schwartz J. Ubiquitin signals autophagic degradation of cytosolic proteins and peroxisomes. Proc Natl Acad Sci U S A 2008; 105:20567-74. 76. Geisler S, Holmstrom KM, Skujat D, Fiesel FC, Rothfuss OC, Kahle PJ, et al. PINK1/Parkin-mediated mitophagy is dependent on VDAC1 and p62/SQSTM1. Nat Cell Biol 2010; 12:119-31. 77. Dietzl G, Chen D, Schnorrer F, Su KC, Barinova Y, Fellner M, et al. A genome- wide transgenic RNAi library for conditional gene inactivation in Drosophila. Nature 2007; 448:151-6. 78. Perkins LA, Holderbaum L, Tao R, Hu Y, Sopko R, McCall K, et al. The Transgenic RNAi Project at Harvard Medical School: Resources and Validation. Genetics 2015; 201:843-52. 79. Brand AH, Perrimon N. Targeted Gene-Expression as a Means of Altering Cell Fates and Generating Dominant Phenotypes. Development 1993; 118:401-15. 80. Ito K, Sass H, Urban J, Hofbauer A, Schneuwly S. GAL4-responsive UAS-tau as a tool for studying the anatomy and development of the Drosophila central nervous system. Cell Tissue Res 1997; 290:1-10. 81. Rijsewijk F, Schuermann M, Wagenaar E, Parren P, Weigel D, Nusse R. The Drosophila homolog of the mouse mammary oncogene int-1 is identical to the segment polarity gene wingless. Cell 1987; 50:649-57. 82. Nusslein-Volhard C, Wieschaus E. Mutations affecting segment number and polarity in Drosophila. Nature 1980; 287:795-801. 83. Atwood HL, Govind CK, Wu CF. Differential ultrastructure of synaptic terminals on ventral longitudinal abdominal muscles in Drosophila larvae. J Neurobiol 1993; 24:1008-24. 84. Johansen J, Halpern ME, Johansen KM, Keshishian H. Stereotypic morphology of glutamatergic synapses on identified muscle cells of Drosophila larvae. J Neurosci 1989; 9:710-25. 85. Broadie KS, Bate M. Development of the embryonic neuromuscular synapse of Drosophila melanogaster. J Neurosci 1993; 13:144-66. 86. Schmid A, Chiba A, Doe CQ. Clonal analysis of Drosophila embryonic neuroblasts: neural cell types, axon projections and muscle targets. Development 1999; 126:4653-89. 87. Jan LY, Jan YN. Properties of the larval neuromuscular junction in Drosophila melanogaster. The Journal of physiology 1976; 262:189-214.

131

88. Jan LY, Jan YN. L-glutamate as an excitatory transmitter at the Drosophila larval neuromuscular junction. The Journal of physiology 1976; 262:215-36. 89. Hoang B, Chiba A. Single-cell analysis of Drosophila larval neuromuscular synapses. Dev Biol 2001; 229:55-70. 90. Choi JC, Park D, Griffith LC. Electrophysiological and morphological characterization of identified motor neurons in the Drosophila third instar larva central nervous system. Journal of neurophysiology 2004; 91:2353-65. 91. Schuster CM, Davis GW, Fetter RD, Goodman CS. Genetic dissection of structural and functional components of synaptic plasticity. I. Fasciclin II controls synaptic stabilization and growth. Neuron 1996; 17:641-54. 92. Schuster CM, Davis GW, Fetter RD, Goodman CS. Genetic dissection of structural and functional components of synaptic plasticity. II. Fasciclin II controls presynaptic structural plasticity. Neuron 1996; 17:655-67. 93. Mathew D, Ataman B, Chen J, Zhang Y, Cumberledge S, Budnik V. Wingless signaling at synapses is through cleavage and nuclear import of receptor DFrizzled2. Science 2005; 310:1344-7. 94. Packard M, Koo ES, Gorczyca M, Sharpe J, Cumberledge S, Budnik V. The Drosophila Wnt, wingless, provides an essential signal for pre- and postsynaptic differentiation. Cell 2002; 111:319-30. 95. Marques G, Haerry TE, Crotty ML, Xue M, Zhang B, O'Connor MB. Retrograde Gbb signaling through the Bmp type 2 receptor wishful thinking regulates systemic FMRFa expression in Drosophila. Development 2003; 130:5457-70. 96. McCabe BD, Marques G, Haghighi AP, Fetter RD, Crotty ML, Haerry TE, et al. The BMP homolog Gbb provides a retrograde signal that regulates synaptic growth at the Drosophila neuromuscular junction. Neuron 2003; 39:241-54. 97. Aberle H, Haghighi AP, Fetter RD, McCabe BD, Magalhaes TR, Goodman CS. wishful thinking encodes a BMP type II receptor that regulates synaptic growth in Drosophila. Neuron 2002; 33:545-58. 98. Marques G, Bao H, Haerry TE, Shimell MJ, Duchek P, Zhang B, et al. The Drosophila BMP type II receptor Wishful Thinking regulates neuromuscular synapse morphology and function. Neuron 2002; 33:529-43. 99. Berke B, Wittnam J, McNeill E, Van Vactor DL, Keshishian H. Retrograde BMP signaling at the synapse: a permissive signal for synapse maturation and activity- dependent plasticity. J Neurosci 2013; 33:17937-50. 100. Dani N, Broadie K. Glycosylated synaptomatrix regulation of trans-synaptic signaling. Dev Neurobiol 2012; 72:2-21. 101. Dani N, Nahm M, Lee S, Broadie K. A targeted glycan-related gene screen reveals heparan sulfate proteoglycan sulfation regulates WNT and BMP trans-synaptic signaling. PLoS Genet 2012; 8:e1003031. 102. Ren Y, Kirkpatrick CA, Rawson JM, Sun M, Selleck SB. Cell Type-Specific Requirements for Heparan Sulfate Biosynthesis at the Drosophila Neuromuscular Junction: Effects on Synapse Function, Membrane Trafficking, and Mitochondrial Localization. Journal of Neuroscience 2009; 29:8539-50. 103. Kamimura K, Ueno K, Nakagawa J, Hamada R, Saitoe M, Maeda N. Perlecan regulates bidirectional Wnt signaling at the Drosophila neuromuscular junction. J Cell Biol 2013; 200:219-33. 104. Johnson KG, Tenney AP, Ghose A, Duckworth AM, Higashi ME, Parfitt K, et al. The HSPGs Syndecan and Dallylike Bind the Receptor Phosphatase LAR and Exert Distinct Effects on Synaptic Development. Neuron 2006; 49:517-31.

132

105. Hailey DW, Rambold AS, Satpute-Krishnan P, Mitra K, Sougrat R, Kim PK, et al. Mitochondria supply membranes for autophagosome biogenesis during starvation. Cell 2010; 141:656-67. 106. Reynolds-Peterson CE, Zhao N, Xu J, Serman TM, Xu J, Selleck SB. Heparan sulfate proteoglycans regulate autophagy in Drosophila. Autophagy 2017:1-18. 107. Klionsky DJ, Abdelmohsen K, Abe A, Abedin MJ, Abeliovich H, Acevedo Arozena A, et al. Guidelines for the use and interpretation of assays for monitoring autophagy (3rd edition). Autophagy 2016; 12:1-222. 108. Simonsen A, Cumming RC, Brech A, Isakson P, Schubert DR, Finley KD. Promoting basal levels of autophagy in the nervous system enhances longevity and oxidant resistance in adult Drosophila. Autophagy 2008; 4:176-84. 109. Bartlett BJ, Isakson P, Lewerenz J, Sanchez H, Kotzebue RW, Cumming RC, et al. p62, Ref(2)P and ubiquitinated proteins are conserved markers of neuronal aging, aggregate formation and progressive autophagic defects. Autophagy 2011; 7:572-83. 110. Petiot A, Ogier-Denis E, Blommaart EF, Meijer AJ, Codogno P. Distinct classes of phosphatidylinositol 3'-kinases are involved in signaling pathways that control macroautophagy in HT-29 cells. J Biol Chem 2000; 275:992-8. 111. Zhai C, Cheng J, Mujahid H, Wang H, Kong J, Yin Y, et al. Selective inhibition of PI3K/Akt/mTOR signaling pathway regulates autophagy of macrophage and vulnerability of atherosclerotic plaque. PLoS One 2014; 9:e90563. 112. Sahani MH, Itakura E, Mizushima N. Expression of the autophagy substrate SQSTM1/p62 is restored during prolonged starvation depending on transcriptional upregulation and autophagy-derived amino acids. Autophagy 2014; 10:431-41. 113. Zhao J, Zhai B, Gygi SP, Goldberg AL. mTOR inhibition activates overall protein degradation by the ubiquitin proteasome system as well as by autophagy. Proc Natl Acad Sci U S A 2015; 112:15790-7. 114. Pircs K, Nagy P, Varga A, Venkei Z, Erdi B, Hegedus K, et al. Advantages and limitations of different p62-based assays for estimating autophagic activity in Drosophila. PLoS One 2012; 7:e44214. 115. Felig P, Owen OE, Wahren J, Cahill GF, Jr. Amino acid metabolism during prolonged starvation. J Clin Invest 1969; 48:584-94. 116. Sandri M. Autophagy in skeletal muscle. FEBS Lett 2010; 584:1411-6. 117. Gutierrez E, Wiggins D, Fielding B, Gould AP. Specialized hepatocyte-like cells regulate Drosophila lipid metabolism. Nature 2007; 445:275-80. 118. DeVorkin L, Go NE, Hou YC, Moradian A, Morin GB, Gorski SM. The Drosophila effector caspase Dcp-1 regulates mitochondrial dynamics and autophagic flux via SesB. J Cell Biol 2014; 205:477-92. 119. Nagy P, Varga A, Kovacs AL, Takats S, Juhasz G. How and why to study autophagy in Drosophila: it's more than just a garbage chute. Methods 2015; 75:151-61. 120. Takats S, Nagy P, Varga A, Pircs K, Karpati M, Varga K, et al. Autophagosomal Syntaxin17-dependent lysosomal degradation maintains neuronal function in Drosophila. J Cell Biol 2013; 201:531-9. 121. Selleck SB. Signaling from across the way: transactivation of VEGF receptors by HSPGs. Mol Cell 2006; 22:431-2. 122. Jakobsson L, Kreuger J, Holmborn K, Lundin L, Eriksson I, Kjellen L, et al. Heparan sulfate in trans potentiates VEGFR-mediated angiogenesis. Dev Cell 2006; 10:625-34.

133

123. Annunziata I, Bouche V, Lombardi A, Settembre C, Ballabio A. Multiple sulfatase deficiency is due to hypomorphic mutations of the SUMF1 gene. Hum Mutat 2007; 28:928. 124. Ramani VC, Purushothaman A, Stewart MD, Thompson CA, Vlodavsky I, Au JL, et al. The heparanase/syndecan-1 axis in cancer: mechanisms and therapies. Febs J 2013; 280:2294-306. 125. Vlodavsky I, Beckhove P, Lerner I, Pisano C, Meirovitz A, Ilan N, et al. Significance of Heparanase in Cancer and Inflammation. Cancer Microenvironment 2012; 5:115-32. 126. Bernfield M, Gotte M, Park PW, Reizes O, Fitzgerald ML, Lincecum J, et al. Functions of cell surface heparan sulfate proteoglycans. Annu Rev Biochem 1999; 68:729-77. 127. Iozzo RV, Sanderson RD. Proteoglycans in cancer biology, tumour microenvironment and angiogenesis. J Cell Mol Med 2011; 15:1013-31. 128. Rivara S, Milazzo FM, Giannini G. Heparanase: a rainbow pharmacological target associated to multiple pathologies including rare diseases. Future medicinal chemistry 2016; 8:647-80. 129. Vlodavsky I, Singh P, Boyango I, Gutter-Kapon L, Elkin M, Sanderson RD, et al. Heparanase: From basic research to therapeutic applications in cancer and inflammation. Drug Resist Updat 2016; 29:54-75. 130. Vreys V, David G. Mammalian heparanase: what is the message? J Cell Mol Med 2007; 11:427-52. 131. Shteingauz A, Boyango I, Naroditsky I, Hammond E, Gruber M, Doweck I, et al. Heparanase Enhances Tumor Growth and Chemoresistance by Promoting Autophagy. Cancer Res 2015; 75:3946-57. 132. Ilan N, Shteingauz A, Vlodavsky I. Function from within: Autophagy induction by HPSE/heparanase—new possibilities for intervention. Autophagy 2015; 11:2387-9. 133. White E. Deconvoluting the context-dependent role for autophagy in cancer. Nature reviews Cancer 2012; 12:401-10. 134. Sanderson RD, Elkin M, Rapraeger AC, Ilan N, Vlodavsky I. Heparanase regulation of cancer, autophagy and inflammation: new mechanisms and targets for therapy. Febs J 2017; 284:42-55. 135. Juhasz G, Hill JH, Yan Y, Sass M, Baehrecke EH, Backer JM, et al. The class III PI(3)K Vps34 promotes autophagy and endocytosis but not TOR signaling in Drosophila. J Cell Biol 2008; 181:655-66. 136. Chang YY, Neufeld TP. An Atg1/Atg13 complex with multiple roles in TOR- mediated autophagy regulation. Mol Biol Cell 2009; 20:2004-14. 137. Ryoo HD, Domingos PM, Kang MJ, Steller H. Unfolded protein response in a Drosophila model for retinal degeneration. EMBO J 2007; 26:242-52. 138. Emery P. Protein extraction from Drosophila heads. Methods Mol Biol 2007; 362:375-7. 139. Cumming RC, Simonsen A, Finley KD. Quantitative Analysis of Autophagic Activity in Drosophila Neural Tissues by Measuring the Turnover Rates of Pathway Substrates. 2008; 451:639-51. 140. Batteiger B, Newhall WJt, Jones RB. The use of Tween 20 as a blocking agent in the immunological detection of proteins transferred to nitrocellulose membranes. Journal of immunological methods 1982; 55:297-307. 141. Benbrook DM, Long A. Integration of autophagy, proteasomal degradation, unfolded protein response and apoptosis. Exp Oncol 2012; 34:286-97.

134

142. Ding Z, Liu S, Sun C, Chen Z, Fan Y, Deng X, et al. Concentration polarization of ox-LDL activates autophagy and apoptosis via regulating LOX-1 expression. Sci Rep 2013; 3:2091. 143. He C, Klionsky DJ. Regulation mechanisms and signaling pathways of autophagy. Annu Rev Genet 2009; 43:67-93. 144. Cho DH, Nakamura T, Lipton SA. Mitochondrial dynamics in cell death and neurodegeneration. Cell Mol Life Sci 2010; 67:3435-47. 145. Chen H, Chan DC. Critical dependence of neurons on mitochondrial dynamics. Curr Opin Cell Biol 2006; 18:453-9. 146. Komatsu M, Waguri S, Chiba T, Murata S, Iwata J, Tanida I, et al. Loss of autophagy in the central nervous system causes neurodegeneration in mice. Nature 2006; 441:880-4. 147. Hara T, Nakamura K, Matsui M, Yamamoto A, Nakahara Y, Suzuki-Migishima R, et al. Suppression of basal autophagy in neural cells causes neurodegenerative disease in mice. Nature 2006; 441:885-9. 148. Brickley K, Stephenson FA. Trafficking kinesin protein (TRAK)-mediated transport of mitochondria in axons of hippocampal neurons. J Biol Chem 2011; 286:18079-92. 149. Glock C, Heumuller M, Schuman EM. mRNA transport & local translation in neurons. Curr Opin Neurobiol 2017; 45:169-77. 150. Zhao YG, Sun L, Miao G, Ji C, Zhao H, Sun H, et al. The autophagy gene Wdr45/Wipi4 regulates learning and memory function and axonal homeostasis. Autophagy 2015; 11:881-90. 151. Tsai J, Grutzendler J, Duff K, Gan WB. Fibrillar amyloid deposition leads to local synaptic abnormalities and breakage of neuronal branches. Nat Neurosci 2004; 7:1181- 3. 152. Galvin JE, Uryu K, Lee VM, Trojanowski JQ. Axon pathology in Parkinson's disease and Lewy body dementia hippocampus contains alpha-, beta-, and gamma- synuclein. Proc Natl Acad Sci U S A 1999; 96:13450-5. 153. Maia PD, Hemphill MA, Zehnder B, Zhang C, Parker KK, Kutz JN. Diagnostic tools for evaluating the impact of Focal Axonal Swellings arising in neurodegenerative diseases and/or traumatic brain injury. Journal of neuroscience methods 2015; 253:233- 43. 154. Kim HJ, Cho MH, Shim WH, Kim JK, Jeon EY, Kim DH, et al. Deficient autophagy in microglia impairs synaptic pruning and causes social behavioral defects. Mol Psychiatry 2016. 155. Moyer CE, Shelton MA, Sweet RA. Dendritic spine alterations in schizophrenia. Neurosci Lett 2015; 601:46-53. 156. Nikoletopoulou V, Sidiropoulou K, Kallergi E, Dalezios Y, Tavernarakis N. Modulation of Autophagy by BDNF Underlies Synaptic Plasticity. Cell Metab 2017; 26:230-42 e5. 157. Chechik G, Meilijson I, Ruppin E. Synaptic pruning in development: a computational account. Neural computation 1998; 10:1759-77. 158. Iglesias J, Eriksson J, Grize F, Tomassini M, Villa AE. Dynamics of pruning in simulated large-scale spiking neural networks. Bio Systems 2005; 79:11-20. 159. Riccomagno MM, Kolodkin AL. Sculpting neural circuits by axon and dendrite pruning. Annu Rev Cell Dev Biol 2015; 31:779-805.

135

160. Cusack CL, Swahari V, Hampton Henley W, Michael Ramsey J, Deshmukh M. Distinct pathways mediate axon degeneration during apoptosis and axon-specific pruning. Nature communications 2013; 4:1876. 161. Simon DJ, Weimer RM, McLaughlin T, Kallop D, Stanger K, Yang J, et al. A caspase cascade regulating developmental axon degeneration. J Neurosci 2012; 32:17540-53. 162. Bagri A, Cheng HJ, Yaron A, Pleasure SJ, Tessier-Lavigne M. Stereotyped pruning of long hippocampal axon branches triggered by retraction inducers of the semaphorin family. Cell 2003; 113:285-99. 163. Liu XB, Low LK, Jones EG, Cheng HJ. Stereotyped axon pruning via plexin signaling is associated with synaptic complex elimination in the hippocampus. J Neurosci 2005; 25:9124-34. 164. Wong JJ, Li S, Lim EK, Wang Y, Wang C, Zhang H, et al. A Cullin1-based SCF E3 ubiquitin ligase targets the InR/PI3K/TOR pathway to regulate neuronal pruning. PLoS biology 2013; 11:e1001657. 165. Tang G, Gudsnuk K, Kuo SH, Cotrina ML, Rosoklija G, Sosunov A, et al. Loss of mTOR-dependent macroautophagy causes autistic-like synaptic pruning deficits. Neuron 2014; 83:1131-43. 166. Gibson JR, Bartley AF, Hays SA, Huber KM. Imbalance of neocortical excitation and inhibition and altered UP states reflect network hyperexcitability in the mouse model of fragile X syndrome. Journal of neurophysiology 2008; 100:2615-26. 167. Sharma A, Hoeffer CA, Takayasu Y, Miyawaki T, McBride SM, Klann E, et al. Dysregulation of mTOR signaling in fragile X syndrome. J Neurosci 2010; 30:694-702. 168. de Vries PJ. Targeted treatments for cognitive and neurodevelopmental disorders in tuberous sclerosis complex. Neurotherapeutics : the journal of the American Society for Experimental NeuroTherapeutics 2010; 7:275-82. 169. Gross C, Nakamoto M, Yao X, Chan CB, Yim SY, Ye K, et al. Excess phosphoinositide 3-kinase subunit synthesis and activity as a novel therapeutic target in fragile X syndrome. J Neurosci 2010; 30:10624-38. 170. Bhattacharya A, Kaphzan H, Alvarez-Dieppa AC, Murphy JP, Pierre P, Klann E. Genetic removal of p70 S6 kinase 1 corrects molecular, synaptic, and behavioral phenotypes in fragile X syndrome mice. Neuron 2012; 76:325-37. 171. Busquets-Garcia A, Gomis-Gonzalez M, Guegan T, Agustin-Pavon C, Pastor A, Mato S, et al. Targeting the endocannabinoid system in the treatment of fragile X syndrome. Nat Med 2013; 19:603-7. 172. Riley BE, Kaiser SE, Shaler TA, Ng AC, Hara T, Hipp MS, et al. Ubiquitin accumulation in autophagy-deficient mice is dependent on the Nrf2-mediated stress response pathway: a potential role for protein aggregation in autophagic substrate selection. J Cell Biol 2010; 191:537-52. 173. Wang XJ, Yu J, Wong SH, Cheng AS, Chan FK, Ng SS, et al. A novel crosstalk between two major protein degradation systems: regulation of proteasomal activity by autophagy. Autophagy 2013; 9:1500-8. 174. Dikic I. Proteasomal and Autophagic Degradation Systems. Annu Rev Biochem 2017; 86:193-224. 175. Fahn S, Sulzer D. Neurodegeneration and neuroprotection in Parkinson disease. NeuroRx : the journal of the American Society for Experimental NeuroTherapeutics 2004; 1:139-54.

136

176. Sulzer D, Surmeier DJ. Neuronal vulnerability, pathogenesis, and Parkinson's disease. Movement disorders : official journal of the Movement Disorder Society 2013; 28:715-24. 177. Schapira AH, Jenner P. Etiology and pathogenesis of Parkinson's disease. Movement disorders : official journal of the Movement Disorder Society 2011; 26:1049- 55. 178. Dagda RK, Cherra SJ, 3rd, Kulich SM, Tandon A, Park D, Chu CT. Loss of PINK1 function promotes mitophagy through effects on oxidative stress and mitochondrial fission. J Biol Chem 2009; 284:13843-55. 179. Thomas KJ, McCoy MK, Blackinton J, Beilina A, van der Brug M, Sandebring A, et al. DJ-1 acts in parallel to the PINK1/parkin pathway to control mitochondrial function and autophagy. Hum Mol Genet 2011; 20:40-50. 180. Su YC, Guo X, Qi X. Threonine 56 phosphorylation of Bcl-2 is required for LRRK2 G2019S-induced mitochondrial depolarization and autophagy. Biochim Biophys Acta 2015; 1852:12-21. 181. Wang X. Destructive Cellular Paths Underlying Familial and Sporadic Parkinson Disease Converge on Mitophagy. Autophagy 2017:0. 182. Thomas B, Beal MF. Parkinson's disease. Hum Mol Genet 2007; 16 Spec No. 2:R183-94. 183. Zhu J, Chu CT. Mitochondrial dysfunction in Parkinson's disease. Journal of Alzheimer's disease : JAD 2010; 20 Suppl 2:S325-34. 184. Jenner P. Oxidative stress in Parkinson's disease. Annals of neurology 2003; 53 Suppl 3:S26-36; discussion S-8. 185. Yoritaka A, Hattori N, Uchida K, Tanaka M, Stadtman ER, Mizuno Y. Immunohistochemical detection of 4-hydroxynonenal protein adducts in Parkinson disease. Proc Natl Acad Sci U S A 1996; 93:2696-701. 186. Floor E, Wetzel MG. Increased protein oxidation in human substantia nigra pars compacta in comparison with basal ganglia and prefrontal cortex measured with an improved dinitrophenylhydrazine assay. J Neurochem 1998; 70:268-75. 187. Alam ZI, Jenner A, Daniel SE, Lees AJ, Cairns N, Marsden CD, et al. Oxidative DNA damage in the parkinsonian brain: an apparent selective increase in 8- hydroxyguanine levels in substantia nigra. J Neurochem 1997; 69:1196-203. 188. Zhang J, Perry G, Smith MA, Robertson D, Olson SJ, Graham DG, et al. Parkinson's disease is associated with oxidative damage to cytoplasmic DNA and RNA in substantia nigra neurons. The American journal of pathology 1999; 154:1423-9. 189. Lazarou M, Sliter DA, Kane LA, Sarraf SA, Wang C, Burman JL, et al. The ubiquitin kinase PINK1 recruits autophagy receptors to induce mitophagy. Nature 2015; 524:309-14. 190. Koyano F, Okatsu K, Kosako H, Tamura Y, Go E, Kimura M, et al. Ubiquitin is phosphorylated by PINK1 to activate parkin. Nature 2014; 510:162-6. 191. Sarraf SA, Raman M, Guarani-Pereira V, Sowa ME, Huttlin EL, Gygi SP, et al. Landscape of the PARKIN-dependent ubiquitylome in response to mitochondrial depolarization. Nature 2013; 496:372-6. 192. Hsieh CH, Shaltouki A, Gonzalez AE, Bettencourt da Cruz A, Burbulla LF, St Lawrence E, et al. Functional Impairment in Miro Degradation and Mitophagy Is a Shared Feature in Familial and Sporadic Parkinson's Disease. Cell stem cell 2016; 19:709-24.

137

193. Kerr JS, Adriaanse BA, Greig NH, Mattson MP, Cader MZ, Bohr VA, et al. Mitophagy and Alzheimer's Disease: Cellular and Molecular Mechanisms. Trends Neurosci 2017; 40:151-66. 194. Hardy J. Amyloid, the presenilins and Alzheimer's disease. Trends in Neurosciences 1997; 20:154-9. 195. Bateman RJ, Xiong C, Benzinger TL, Fagan AM, Goate A, Fox NC, et al. Clinical and biomarker changes in dominantly inherited Alzheimer's disease. N Engl J Med 2012; 367:795-804. 196. Wenk GL. Neuropathologic changes in Alzheimer's disease. The Journal of clinical psychiatry 2003; 64 Suppl 9:7-10. 197. Karlinsky H, Vaula G, Haines JL, Ridgley J, Bergeron C, Mortilla M, et al. Molecular and prospective phenotypic characterization of a pedigree with familial Alzheimer's disease and a missense mutation in codon 717 of the beta-amyloid precursor protein gene. Neurology 1992; 42:1445-53. 198. Farlow M, Murrell J, Ghetti B, Unverzagt F, Zeldenrust S, Benson M. Clinical characteristics in a kindred with early-onset Alzheimer's disease and their linkage to a G- ->T change at position 2149 of the amyloid precursor protein gene. Neurology 1994; 44:105-11. 199. Cataldo AM, Barnett JL, Pieroni C, Nixon RA. Increased neuronal endocytosis and protease delivery to early endosomes in sporadic Alzheimer's disease: neuropathologic evidence for a mechanism of increased beta-amyloidogenesis. J Neurosci 1997; 17:6142-51. 200. Nixon RA, Wegiel J, Kumar A, Yu WH, Peterhoff C, Cataldo A, et al. Extensive involvement of autophagy in Alzheimer disease: an immuno-electron microscopy study. Journal of neuropathology and experimental neurology 2005; 64:113-22. 201. Tramutola A, Triplett JC, Di Domenico F, Niedowicz DM, Murphy MP, Coccia R, et al. Alteration of mTOR signaling occurs early in the progression of Alzheimer disease (AD): analysis of brain from subjects with pre-clinical AD, amnestic mild cognitive impairment and late-stage AD. J Neurochem 2015; 133:739-49. 202. Caccamo A, Maldonado MA, Majumder S, Medina DX, Holbein W, Magri A, et al. Naturally secreted amyloid-beta increases mammalian target of rapamycin (mTOR) activity via a PRAS40-mediated mechanism. J Biol Chem 2011; 286:8924-32. 203. Cataldo AM, Barnett JL, Berman SA, Li J, Quarless S, Bursztajn S, et al. Gene expression and cellular content of cathepsin D in Alzheimer's disease brain: evidence for early up-regulation of the endosomal-lysosomal system. Neuron 1995; 14:671-80. 204. Small SA, Petsko GA. Retromer in Alzheimer disease, Parkinson disease and other neurological disorders. Nat Rev Neurosci 2015; 16:126-32. 205. Shafei MA, Harris M, Conway ME. Divergent Metabolic Regulation of Autophagy and mTORC1-Early Events in Alzheimer's Disease? Frontiers in aging neuroscience 2017; 9:173. 206. Holzgreve W, Grobe H, von Figura K, Kresse H, Beck H, Mattei JF. Morquio syndrome: clinical findings in 11 patients with MPS IVA and 2 patients with MPS IVB. Hum Genet 1981; 57:360-5. 207. Natowicz MR, Short MP, Wang Y, Dickersin GR, Gebhardt MC, Rosenthal DI, et al. Clinical and biochemical manifestations of hyaluronidase deficiency. N Engl J Med 1996; 335:1029-33. 208. Azevedo AC, Schwartz IV, Kalakun L, Brustolin S, Burin MG, Beheregaray AP, et al. Clinical and biochemical study of 28 patients with mucopolysaccharidosis type VI. Clin Genet 2004; 66:208-13.

138

209. Shipley JM, Klinkenberg M, Wu BM, Bachinsky DR, Grubb JH, Sly WS. Mutational analysis of a patient with mucopolysaccharidosis type VII, and identification of pseudogenes. Am J Hum Genet 1993; 52:517-26. 210. Lee C, Dineen TE, Brack M, Kirsch JE, Runge VM. The mucopolysaccharidoses: characterization by cranial MR imaging. AJNR American journal of neuroradiology 1993; 14:1285-92. 211. Cleary MA, Wraith JE. The presenting features of mucopolysaccharidosis type IH (Hurler syndrome). Acta paediatrica 1995; 84:337-9. 212. Whitley CB. The Mucopolysaccharidoses. In: Beighton P, ed. McKusick's Heritable Disorders of Connective Tissue. St. Luis: Mosby, 1993:377-98. 213. Wraith JE, Scarpa M, Beck M, Bodamer OA, De Meirleir L, Guffon N, et al. Mucopolysaccharidosis type II (Hunter syndrome): a clinical review and recommendations for treatment in the era of enzyme replacement therapy. European journal of pediatrics 2008; 167:267-77. 214. van de Kamp JJ, Niermeijer MF, von Figura K, Giesberts MA. Genetic heterogeneity and clinical variability in the Sanfilippo syndrome (types A, B, and C). Clin Genet 1981; 20:152-60. 215. Dursun A, Yalnizoglu D, Gerdan OF, Yucel-Yilmaz D, Sagiroglu MS, Yuksel B, et al. A probable new syndrome with the storage disease phenotype caused by the VPS33A gene mutation. Clinical dysmorphology 2017; 26:1-12. 216. Kondo H, Maksimova N, Otomo T, Kato H, Imai A, Asano Y, et al. Mutation in VPS33A affects metabolism of glycosaminoglycans: a new type of mucopolysaccharidosis with severe systemic symptoms. Hum Mol Genet 2017; 26:173- 83. 217. Burk RD, Valle D, Thomas GH, Miller C, Moser A, Moser H, et al. Early manifestations of multiple sulfatase deficiency. The Journal of pediatrics 1984; 104:574- 8. 218. Schlotawa L, Ennemann EC, Radhakrishnan K, Schmidt B, Chakrapani A, Christen HJ, et al. SUMF1 mutations affecting stability and activity of formylglycine generating enzyme predict clinical outcome in multiple sulfatase deficiency. Eur J Hum Genet 2011; 19:253-61. 219. Ginsberg SD, Galvin JE, Lee VM, Rorke LB, Dickson DW, Wolfe JH, et al. Accumulation of intracellular amyloid-beta peptide (A beta 1-40) in mucopolysaccharidosis brains. Journal of neuropathology and experimental neurology 1999; 58:815-24. 220. Kleinschmit A, Koyama T, Dejima K, Hayashi Y, Kamimura K, Nakato H. Drosophila heparan sulfate 6-O endosulfatase regulates Wingless morphogen gradient formation. Dev Biol 2010; 345:204-14. 221. Canonico B, Cesarini E, Salucci S, Luchetti F, Falcieri E, Di Sario G, et al. Defective Autophagy, Mitochondrial Clearance and Lipophagy in Niemann-Pick Type B Lymphocytes. PLoS One 2016; 11:e0165780. 222. Lieberman AP, Puertollano R, Raben N, Slaugenhaupt S, Walkley SU, Ballabio A. Autophagy in lysosomal storage disorders. Autophagy 2012; 8:719-30. 223. Hernandez D, Torres CA, Setlik W, Cebrian C, Mosharov EV, Tang G, et al. Regulation of presynaptic neurotransmission by macroautophagy. Neuron 2012; 74:277- 84. 224. Silva VDA, Cuevas C, Munoz P, Villa M, Ahumada-Castro U, Huenchuguala S, et al. Autophagy protects against neural cell death induced by piperidine alkaloids present

139 in Prosopis juliflora (Mesquite). Anais da Academia Brasileira de Ciencias 2017; 89:247- 61. 225. Otabe H, Nibuya M, Shimazaki K, Toda H, Suzuki G, Nomura S, et al. Electroconvulsive seizures enhance autophagy signaling in rat hippocampus. Prog Neuropsychopharmacol Biol Psychiatry 2014; 50:37-43. 226. Zhang X, Yan H, Yuan Y, Gao J, Shen Z, Cheng Y, et al. Cerebral ischemia- reperfusion-induced autophagy protects against neuronal injury by mitochondrial clearance. Autophagy 2013; 9:1321-33. 227. Liu CC, Zhao N, Yamaguchi Y, Cirrito JR, Kanekiyo T, Holtzman DM, et al. Neuronal heparan sulfates promote amyloid pathology by modulating brain amyloid-beta clearance and aggregation in Alzheimer's disease. Science translational medicine 2016; 8:332ra44. 228. Snow AD, Wight TN. Proteoglycans in the pathogenesis of Alzheimer's disease and other amyloidoses. Neurobiology of aging 1989; 10:481-97. 229. Bame KJ, Danda J, Hassall A, Tumova S. Abeta(1-40) prevents heparanase- catalyzed degradation of heparan sulfate glycosaminoglycans and proteoglycans in vitro. A role for heparan sulfate proteoglycan turnover in Alzheimer's disease. J Biol Chem 1997; 272:17005-11. 230. Jendresen CB, Cui H, Zhang X, Vlodavsky I, Nilsson LN, Li JP. Overexpression of heparanase lowers the amyloid burden in amyloid-beta precursor protein transgenic mice. J Biol Chem 2015; 290:5053-64. 231. Holmes BB, DeVos SL, Kfoury N, Li M, Jacks R, Yanamandra K, et al. Heparan sulfate proteoglycans mediate internalization and propagation of specific proteopathic seeds. Proc Natl Acad Sci U S A 2013; 110:E3138-47. 232. Sarkar S, Ravikumar B, Floto RA, Rubinsztein DC. Rapamycin and mTOR- independent autophagy inducers ameliorate toxicity of polyglutamine-expanded huntingtin and related proteinopathies. Cell Death Differ 2009; 16:46-56. 233. Ha JY, Kim JS, Kim SE, Son JH. Simultaneous activation of mitophagy and autophagy by staurosporine protects against dopaminergic neuronal cell death. Neurosci Lett 2014; 561:101-6. 234. Proenca CC, Stoehr N, Bernhard M, Seger S, Genoud C, Roscic A, et al. Atg4b- dependent autophagic flux alleviates Huntington's disease progression. PLoS One 2013; 8:e68357. 235. Steele JW, Ju S, Lachenmayer ML, Liken J, Stock A, Kim SH, et al. Latrepirdine stimulates autophagy and reduces accumulation of alpha-synuclein in cells and in mouse brain. Mol Psychiatry 2013; 18:882-8. 236. Singh AK, Kashyap MP, Tripathi VK, Singh S, Garg G, Rizvi SI. Neuroprotection Through Rapamycin-Induced Activation of Autophagy and PI3K/Akt1/mTOR/CREB Signaling Against Amyloid-beta-Induced Oxidative Stress, Synaptic/Neurotransmission Dysfunction, and Neurodegeneration in Adult Rats. Mol Neurobiol 2016. 237. Franco R, Martinez-Pinilla E, Navarro G, Zamarbide M. Potential of GPCRs to modulate MAPK and mTOR pathways in Alzheimer's disease. Prog Neurobiol 2017; 149-150:21-38. 238. Ruhlmann C, Wolk T, Blumel T, Stahn L, Vollmar B, Kuhla A. Long-term caloric restriction in ApoE-deficient mice results in neuroprotection via Fgf21-induced AMPK/mTOR pathway. Aging 2016; 8:2777-89. 239. Spradling AC, Stern D, Beaton A, Rhem EJ, Laverty T, Mozden N, et al. The Berkeley Drosophila Genome Project gene disruption project: Single P-element insertions mutating 25% of vital Drosophila genes. Genetics 1999; 153:135-77.

140

240. Lin X, Buff EM, Perrimon N, Michelson AM. Heparan sulfate proteoglycans are essential for FGF receptor signaling during Drosophila embryonic development. Development 1999; 126:3715-23. 241. Lin X, Perrimon N. Dally cooperates with Drosophila Frizzled 2 to transduce Wingless signalling. Nature 1999; 400:281-4. 242. The I, Bellaiche Y, Perrimon N. Hedgehog movement is regulated through tout velu-dependent synthesis of a heparan sulfate proteoglycan. Mol Cell 1999; 4:633-9. 243. Association AP. Diagnostic and statistical manula of mental disorders (5th ed.). Washington, DC, 2013. 244. Levy SE, Giarelli E, Lee LC, Schieve LA, Kirby RS, Cunniff C, et al. Autism spectrum disorder and co-occurring developmental, psychiatric, and medical conditions among children in multiple populations of the United States. J Dev Behav Pediatr 2010; 31:267-75. 245. DiGuiseppi C, Hepburn S, Davis JM, Fidler DJ, Hartway S, Lee NR, et al. Screening for autism spectrum disorders in children with Down syndrome: population prevalence and screening test characteristics. J Dev Behav Pediatr 2010; 31:181-91. 246. Cohen D, Pichard N, Tordjman S, Baumann C, Burglen L, Excoffier E, et al. Specific genetic disorders and autism: clinical contribution towards their identification. J Autism Dev Disord 2005; 35:103-16. 247. Hall SS, Lightbody AA, Reiss AL. Compulsive, self-injurious, and autistic behavior in children and adolescents with fragile X syndrome. American journal of mental retardation : AJMR 2008; 113:44-53. 248. Zecavati N, Spence SJ. Neurometabolic disorders and dysfunction in autism spectrum disorders. Current neurology and neuroscience reports 2009; 9:129-36. 249. Christensen DL, Baio J, Van Naarden Braun K, Bilder D, Charles J, Constantino JN, et al. Prevalence and Characteristics of Autism Spectrum Disorder Among Children Aged 8 Years--Autism and Developmental Disabilities Monitoring Network, 11 Sites, United States, 2012. MMWR Surveill Summ 2016; 65:1-23. 250. Lavelle TA, Weinstein MC, Newhouse JP, Munir K, Kuhlthau KA, Prosser LA. Economic burden of childhood autism spectrum disorders. Pediatrics 2014; 133:e520-9. 251. Amendah D, Grosse SD, Peacock G, Mandell DS. The economic costs of autism: A review. In: D. Amaral DG, & G. Dawson, ed. Autism spectrum disorders. Oxford: Oxford University Press, 2011:1347-60. 252. Anderson KA, Shattuck PT, Cooper BP, Roux AM, Wagner M. Prevalence and correlates of postsecondary residential status among young adults with an autism spectrum disorder. Autism : the international journal of research and practice 2014; 18:562-70. 253. Howlin P, Goode S, Hutton J, Rutter M. Adult outcome for children with autism. J Child Psychol Psychiatry 2004; 45:212-29. 254. Buescher AV, Cidav Z, Knapp M, Mandell DS. Costs of autism spectrum disorders in the United Kingdom and the United States. JAMA pediatrics 2014; 168:721- 8. 255. Basu SN, Kollu R, Banerjee-Basu S. AutDB: a gene reference resource for autism research. Nucleic Acids Res 2009; 37:D832-6. 256. Hamilton SM, Green JR, Veeraragavan S, Yuva L, McCoy A, Wu Y, et al. Fmr1 and Nlgn3 knockout rats: novel tools for investigating autism spectrum disorders. Behav Neurosci 2014; 128:103-9.

141

257. Servadio M, Vanderschuren LJ, Trezza V. Modeling autism-relevant behavioral phenotypes in rats and mice: Do 'autistic' rodents exist? Behavioural pharmacology 2015; 26:522-40. 258. Pasciuto E, Borrie SC, Kanellopoulos AK, Santos AR, Cappuyns E, D'Andrea L, et al. Autism Spectrum Disorders: Translating human deficits into mouse behavior. Neurobiology of learning and memory 2015; 124:71-87. 259. Bastock M, Manning A. The courtship of Drosophila melanogaster. Behavior 1955; 8:85–111. 260. Griffith LC, Ejima A. Courtship learning in Drosophila melanogaster: diverse plasticity of a reproductive behavior. Learn Mem 2009; 16:743-50. 261. Dow MA, von Schilcher F. Aggression and mating success in Drosophila melanogaster. Nature 1975; 254:511-2. 262. Nilsen SP, Chan YB, Huber R, Kravitz EA. Gender-selective patterns of aggressive behavior in Drosophila melanogaster. Proc Natl Acad Sci U S A 2004; 101:12342-7. 263. Wang L, Han X, Mehren J, Hiroi M, Billeter JC, Miyamoto T, et al. Hierarchical chemosensory regulation of male-male social interactions in Drosophila. Nat Neurosci 2011; 14:757-62. 264. Yurkovic A, Wang O, Basu AC, Kravitz EA. Learning and memory associated with aggression in Drosophila melanogaster. Proc Natl Acad Sci U S A 2006; 103:17519-24. 265. Ueda A, Kidokoro Y. Aggressive behaviors of female Drosophila melanogaster are influenced by their social experience and food resources. Physiological Entomology 2002; 27:21-8. 266. Bolduc FV, Valente D, Nguyen AT, Mitra PP, Tully T. An assay for social interaction in Drosophila fragile X mutants. Fly 2010; 4:216-25. 267. Franco LM, Okray Z, Linneweber GA, Hassan BA, Yaksi E. Reduced Lateral Inhibition Impairs Olfactory Computations and Behaviors in a Drosophila Model of Fragile X Syndrome. Curr Biol 2017; 27:1111-23. 268. Monyak RE, Emerson D, Schoenfeld BP, Zheng X, Chambers DB, Rosenfelt C, et al. Insulin signaling misregulation underlies circadian and cognitive deficits in a Drosophila fragile X model. Mol Psychiatry 2016. 269. Hahn N, Geurten B, Gurvich A, Piepenbrock D, Kastner A, Zanini D, et al. Monogenic heritable autism gene neuroligin impacts Drosophila social behaviour. Behav Brain Res 2013; 252:450-7. 270. Wise A, Tenezaca L, Fernandez RW, Schatoff E, Flores J, Ueda A, et al. Drosophila mutants of the autism candidate gene neurobeachin (rugose) exhibit neurodevelopmental disorders, aberrant synaptic properties, altered locomotion, and impaired adult social behavior and activity patterns. J Neurogenet 2015; 29:135-43. 271. Guo HF, Tong J, Hannan F, Luo L, Zhong Y. A neurofibromatosis-1-regulated pathway is required for learning in Drosophila. Nature 2000; 403:895-8. 272. Williams JA, Su HS, Bernards A, Field J, Sehgal A. A circadian output in Drosophila mediated by neurofibromatosis-1 and Ras/MAPK. Science 2001; 293:2251-6. 273. King LB, Koch M, Murphy KR, Velazquez Y, Ja WW, Tomchik SM. Neurofibromin Loss of Function Drives Excessive Grooming in Drosophila. G3 2016; 6:1083-93. 274. Leitner Y. The co-occurrence of autism and attention deficit hyperactivity disorder in children - what do we know? Front Hum Neurosci 2014; 8:268.

142

275. Davis NO, Kollins SH. Treatment for co-occurring attention deficit/hyperactivity disorder and autism spectrum disorder. Neurotherapeutics : the journal of the American Society for Experimental NeuroTherapeutics 2012; 9:518-30. 276. Allik H, Larsson JO, Smedje H. Insomnia in school-age children with Asperger syndrome or high-functioning autism. BMC psychiatry 2006; 6:18. 277. Liu X, Hubbard JA, Fabes RA, Adam JB. Sleep disturbances and correlates of children with autism spectrum disorders. Child psychiatry and human development 2006; 37:179-91. 278. Singh K, Zimmerman AW. Sleep in Autism Spectrum Disorder and Attention Deficit Hyperactivity Disorder. Semin Pediatr Neurol 2015; 22:113-25. 279. Hirsch J. Studies in experimental behavior genetics. II. Individual differences in geotaxis as a function of chromosome variations in synthesized Drosophila populations. Journal of comparative and physiological psychology 1959; 52:304-8. 280. Hirsch J, Ksander G. Studies in experimental behavior genetics. V. Negative geotaxis and further chromosome analyses in Drosophila melanogaster. Journal of comparative and physiological psychology 1969; 67:118-22. 281. Navarro J, del Solar E. Pattern of spatial distribution in Drosophila melanogaster. Behavior genetics 1975; 5:9-16. 282. Simon AF, Chou MT, Salazar ED, Nicholson T, Saini N, Metchev S, et al. A simple assay to study social behavior in Drosophila: measurement of social space within a group. Genes Brain Behav 2012; 11:243-52. 283. Moghadam NN, Holmstrup M, Manenti T, Loeschcke V. Phospholipid fatty acid composition linking larval-density to lifespan of adult Drosophila melanogaster. Experimental gerontology 2015; 72:177-83. 284. Luong LT, Polak M. Environment-dependent trade-offs between ectoparasite resistance and larval competitive ability in the Drosophila-Macrocheles system. Heredity 2007; 99:632-40. 285. de Gee M, Lof ME, Hemerik L. The effect of chemical information on the spatial distribution of fruit flies: II Parameterization, calibration, and sensitivity. Bulletin of mathematical biology 2008; 70:1850-68. 286. Lof ME, de Gee M, Hemerik L. Odor-mediated aggregation enhances the colonization ability of Drosophila melanogaster. Journal of theoretical biology 2009; 258:363-70. 287. Lof ME, Etienne RS, Powell J, de Gee M, Hemerik L. The effect of chemical information on the spatial distribution of fruit flies: I Model results. Bulletin of mathematical biology 2008; 70:1827-49. 288. McNeil AR, Jolley SN, Akinleye AA, Nurilov M, Rouzyi Z, Milunovich AJ, et al. Conditions Affecting Social Space in Drosophila melanogaster. J Vis Exp 2015:e53242. 289. Liu H, Han M, Li Q, Zhang X, Wang WA, Huang FD. Automated rapid iterative negative geotaxis assay and its use in a genetic screen for modifiers of Abeta(42)- induced locomotor decline in Drosophila. Neuroscience bulletin 2015; 31:541-9. 290. Benzer S. Genetic dissection of behavior. Scientific American 1973; 229:24-37. 291. de Belle JS, Heisenberg M. Expression of Drosophila mushroom body mutations in alternative genetic backgrounds: a case study of the mushroom body miniature gene (mbm). Proc Natl Acad Sci U S A 1996; 93:9875-80. 292. Kain JS, Stokes C, de Bivort BL. Phototactic personality in fruit flies and its suppression by serotonin and white. Proc Natl Acad Sci U S A 2012; 109:19834-9. 293. Nolte DJ. The eye-pigmentary system of Drosophila. III. The action of eye-colour genes. Journal of Genetics 1952; 51:142-85.

143

294. Joh TH. Tryptophan hydroxylase: Molecular biology and regulation. Handb Exp Pharmacol 1997; 129:117-29. 295. Goodwill KE, Sabatier C, Stevens RC. Crystal structure of tyrosine hydroxylase with bound cofactor analogue and iron at 2.3 A resolution: self-hydroxylation of Phe300 and the pterin-binding site. Biochemistry 1998; 37:13437-45. 296. Borycz J, Borycz JA, Kubow A, Lloyd V, Meinertzhagen IA. Drosophila ABC transporter mutants white, brown and scarlet have altered contents and distribution of biogenic amines in the brain. J Exp Biol 2008; 211:3454-66. 297. Bassett AR, Tibbit C, Ponting CP, Liu JL. Highly efficient targeted mutagenesis of Drosophila with the CRISPR/Cas9 system. Cell reports 2013; 4:220-8. 298. Pfeiffenberger C, Lear BC, Keegan KP, Allada R. Locomotor activity level monitoring using the Drosophila Activity Monitoring (DAM) System. Cold Spring Harbor protocols 2010; 2010:pdb prot5518. 299. McGuire SE, Le PT, Osborn AJ, Matsumoto K, Davis RL. Spatiotemporal rescue of memory dysfunction in Drosophila. Science 2003; 302:1765-8 300. Bellen HJ, Budnik V. The neuromuscular junction. Drosophila Protocols. Cold Spring Harbor Laboratory Press, 2000; 179-199 301. Kohsaka H, Okusawa S, Itakura Y, Fukushiki A, Nose A. Develeopment of larval motor circuits in Drosophila. Development, Growth, & Differentiation 2012; 54:408-419

144

Appendix A

PI3K Overexpression suppresses autophagic induction

Appendix A: Overexpression of the Drosophila p110 subunit of the class I PI3K suppresses the autophagic response induced by inhibition of sfl

Muscles 6 and 7 of the larval A3 hemisegment. I In control animals, pronounced GFP-atg8a puncta occur in small numbers. J sfl inhibition induces the formation of numerous GFP-atg8a positive structures (early autophagosomes). K Overexpression of wild-type Drosophila p110 (Dp110WT) suppresses the formation of both large, bright GFP-atg8a puncta and the small, dim puncta seen throughout muscles 6 and 7 in panels J and I. L Co-expression of Dp110WT with sfl RNAi suppresses the ability of sfl loss of function to induce autophagosome formation. M Quantification of the prevalence of bright GFP-Atg8a puncta. ** p<.01, *** p<.001 (compared to control), ### p<.001 (comparison indicated by brackets). 145

Appendix B

qPCR Data

146

Appendix B: RTqPCR analysis of gene expression

All transcription levels are displayed as a percentage relative the level of transcript found in the control, and the gene name over each graph indicates the transcript being measured. A RTqPCR performed on RNA collected from whole larval tissue. In all cases other than the mutant ttv00681, the change in expression was exacted using the ubiquitous da-GAL4 driver and an RNAi construct. B RTqPCR performed on RNA collected from adult heads. RNAi and UAS- overexpression was induced by the elav C155-GAL4 neuron specific driver, with UAS-DcrII to enhance RNAi efficacy. Error bars represent SEM.

147

Appendix C

Supplementary western blot data

Figure 2.02 Appendix C: Loading control

Ponceau S staining of the same representative lanes selected for Fig.2.02. Each combination of genotype and treatment was performed in biological triplicate. The lanes from each set with the closest background-corrected Ponceau staining intensities were selected to be displayed in the primary portion of Fig. 2.02.

148

2% H2O2: - + - + - + - + - + - +

Appendix C: Preliminary results from exposure of the mutant panel to 2% H2O2

A α-ubiquitin staining of insoluble total head extract. B Ponceau S loading controls for samples in A. C Quantitation of IUP density (relative to loading) in control and peroxide exposed animals. All pairwise differences are not statistically significant.

149

150

Appendix D

Enlarged images from Fig. 2.06

ElavC155-Gal4>UAS-dcsrII /+ (Control) /UAS-wRNAi

/UAS-psn /UAS-psn; UAS-wRNAi

151

/UAS-psn; UAS-sflRNAi 1 /UAS-psn; UAS-sflRNAi 2

/UAS-psn; UAS-ttvRNAi

152

Appendix E

Summary of the mucopolysaccharidoses and related disorders

Most MPSs have multiple sub-categories, generally distinguished by severity, which have been collapsed here for the sake of brevity. All MPSs exhibit a range of symptomatic severity depending on the degree of functional compromise in the affected gene. MPSs that do not present with neurodegeneration may still cause reduced lifespan due to complications of respiratory and cardiovascular symptoms. CNS complications may also occur in these MPSs due to skeletal abnormalities of the cervical spine and/or skull. HS, Heparan Sulfate; DS, Dermatan Sulfate; CD, Chondroitin Sulfate; KS, Keratan Sulfate; HS, Hyaluronin; MPS, Mucopolysaccharidosis; MPSPS, Mucopolysaccharidosis Plus Syndrome; MSD, Multiple Sulfatase Deficiency.

153

Appendix F

Publication

154 AUTOPHAGY 2017, VOL. 13, NO. 8, 1262–1279 https://doi.org/10.1080/15548627.2017.1304867

BASIC RESEARCH PAPER Heparan sulfate proteoglycans regulate autophagy in Drosophila

Claire E. Reynolds-Peterson†, Na Zhao†, Jie Xu, Taryn M. Serman, Jielin Xu, and Scott B. Selleck Department of Biochemistry and Molecular Biology, The Pennsylvania State University, University Park, PA, USA

ABSTRACT ARTICLE HISTORY Heparan sulfate-modified proteoglycans (HSPGs) are important regulators of signaling and molecular Received 22 June 2015 recognition at the cell surface and in the extracellular space. Disruption of HSPG core proteins, Revised 22 February 2017 HS-synthesis, or HS-degradation can have profound effects on growth, patterning, and cell survival. The Accepted 6 March 2017 Drosophila neuromuscular junction provides a tractable model for understanding the activities of HSPGs at KEYWORDS a synapse that displays developmental and activity-dependent plasticity. Muscle cell-specific knockdown autophagy; Drosophila; of HS biosynthesis disrupted the organization of a specialized postsynaptic membrane, the subsynaptic heparan sulfate reticulum (SSR), and affected the number and morphology of mitochondria. We provide evidence that proteoglycan; mitochondria; these changes result from a dysregulation of macroautophagy (hereafter referred to as autophagy). subsynaptic reticulum Cellular and molecular markers of autophagy are all consistent with an increase in the levels of autophagy in the absence of normal HS-chain biosynthesis and modification. HS production is also required for normal levels of autophagy in the fat body, the central energy storage and nutritional sensing organ in Drosophila. Genetic mosaic analysis indicates that HS-dependent regulation of autophagy occurs non-cell autonomously, consistent with HSPGs influencing this cellular process via signaling in the extracellular space. These findings demonstrate that HS biosynthesis has important regulatory effects on autophagy and that autophagy is critical for normal assembly of postsynaptic membrane specializations.

Introduction and active zone assembly.8,9 The protein trol (terribly Heparan sulfate-modified proteins are ubiquitous components reduced optic lobes), a secreted HSPG, affects both moto- of the cell surface and extracellular matrix. These highly sul- neuron terminal outgrowth and the elaboration of a post- fated macromolecules play crucial roles in regulating develop- synaptic specialization, the subsynaptic reticulum (SSR), mental signaling pathways, including WNT/Wingless, apparently by governing wg/Wnt signaling at the synapse.10 Hedgehog, BMP/Bone Morphogenetic Protein families, TGFB/ Our earlier work examining the effects of HS biosynthesis Transforming Growth Factor-b, and FGF/Fibroblast Growth on NMJ structure reveals the critical requirement for – Factors in a myriad of systems.1 5 HSPGs have diverse biologi- HSPGs in both the pre- and postsynaptic cells, consistent cal functions, affecting assembly of growth factor-receptor with the aforementioned studies examining the functions of complexes, the stability of growth factors in the matrix, produc- individual HSPGs and their molecular partners. In particu- tion of morphogen gradients, and endocytic clearance of cell lar, we note several striking effects on the cellular organiza- surface-bound ligands. Disruption of individual HSPG core tion of the postsynaptic cell, namely disorganization of the proteins, HS-biosynthesis or modification, HS-desulfation or SSR and loss of mitochondria.11 The work described here degradation can have profound effects on morphogenesis, lipid revealsthecellularbasisforthesephenotypes,showingthat catabolism, and cell survival.6,7 loss of HSPGs produces activation of autophagy in the mus- We have been using the Drosophila neuromuscular junc- cle cell. The function of HSPGs in suppressing autophagy is tion to investigate the molecular and cellular functions of not limited to muscle. A mutation affecting HS polymer HSPGs during synapse development. HSPGs are concen- synthesis or RNA interference of genes required for either trated at the neuromuscular junction (NMJ) and play a crit- HS synthesis or sulfation produced elevated levels of auto- ical role in signaling events that coordinate motoneuron phagy in the fat body, an organ critical for energy sensing and muscle cell interactions. The HSPGs Sdc (syndecan) and storage. These findings suggest that HSPG-mediated and dlp (dally-like) influence the signaling of the protein signaling can affect autophagy in multiple cell types and tyrosine phosphatase Lar (leukocyte-antigen-related-like) at provide a general mechanism for affecting the levels of this the presynaptic-terminal, affecting both terminal growth critical cellular process.

CONTACT Scott B. Selleck [email protected] Department of Biochemistry and Molecular Biology, The Pennsylvania State University, University Park, PA 16802, USA. y Contributed equally to this work. Supplemental data for this article can be accessed on the publisher’s website. © 2017 Claire E. Reynolds-Peterson, Na Zhao, Jie Xu, Taryn M. Serman, Jielin Xu, and Scott B. Selleck. Published with license by Taylor & Francis. This is an Open Access article distributed under the terms of the Creative Commons Attribution-Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0/), which permits unre- stricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. The moral rights of the named author(s) have been asserted. AUTOPHAGY 1263

Results mRNA in animals with pancellular expression of the RNAi construct (see Methods section and Fig. S1). These measures showed HS biosynthesis is required for the organization of that the RNAi constructs produced significant reduction of their postsynaptic specializations at the neuromuscular junction respective target mRNAs, and for ttv, where function would Earlier analysis of genes affecting HS-biosynthesis had shown affect the levels of HS, the 3G10 epitope was dramatically low- effects on the morphology and physiology of the neuromuscular ered at the NMJ (Fig. S1, A to F and G). The cellular architec- junction,11 a glutaminergic synapse that demonstrates both ture of the synapse was examined in detail by transmission developmental and activity-dependent plasticity. Animals defec- electron and confocal microscopy analysis of animals bearing tive in HS biosynthesis or modification show a diverse set of cel- these RNA interfering constructs expressed selectively in muscle. lular phenotypes in the muscle, including changes in The neuromuscular junction is a well-studied and well-charac- mitochondrial density, altered levels of ER and golgi markers, terized synapse ideally suited for examining the molecular and cel- and disruptions of a specialized postsynaptic membrane struc- lular processes affected by compromising HS biosynthesis. The ture, the subsynaptic reticulum (SSR).11 To evaluate the function larval body wall muscles are innervated by identified and segmen- of HSPGs in the postsynaptic cell and its membrane specializa- tally repeated motoneurons. The arrangement of tions, RNAi constructs directed against tout velu (ttv)orsulfate- Drosophila larval body wall muscles is readily revealed by the mus- less (sfl) mRNA were expressed with muscle-specificor cle specific expression of a GFP-tagged MusmusculusCD8trans- – pancellular GAL4 lines.12 15 Each of these genes encodes a gene encoding a transmembrane protein (Mef2-Gal4> UAS- unique enzyme affecting different steps of HS biosynthesis, sum- mCD8-GFP). The somatic muscles of Drosophila larvae are large, marized in Fig. 1 A. Knockdown of these components compro- multinucleated cells arranged in a stereotyped pattern that is mises either the generation of the HS polymer, or its sulfation, repeated in each abdominal segment (Fig. 1 B). Within each hemi- both critical for the function of HS-modified proteins.16,17 The segment, 30 muscle cells are innervated by 32 motoneurons.18 efficacy and specificity of the RNA-interference constructs were Muscles 6 and 7 of segment A3 are the most commonly used evaluated by 2 methods: 1) detection of an HS-derived epitope muscles for microscopy and physiology recordings, and are shown at the NMJ with a monoclonal antibody (3G10) that reflects the with mCD8-GFP (green) labeling the muscle membranes and anti- level of HS polymer and 2) quantitative PCR of the target HRP antibody staining (red) labeling motoneuron axonal

Figure 1. HS biosynthesis and morphology of the larval neuromuscular junction. (A) Diagram of HS biosynthesis and modification. HS polymers are attached to serine residues of the core protein via a tetrasaccharide linker. Chain initiation and polymerization are catalyzed by N-GlcNAc transferase I, N-GlcNAc transferase II and GlcA transferase II (encoded in Drosophila by botv, ttv and Ext2), respectively. Further modification requires the N-deacetylase/N-sulfotransferase (encoded by sfl) and specific O- sulfotransferases. (B to D) Images of Drosophila third instar larval ventral body wall muscle and NMJ. (B) UAS-mCD8-GFP expression was directed to muscle membranes using Mef2-GAL4. The body wall muscles of Drosophila larvae are arranged in a stereotypical pattern repeated in each abdominal segment. Muscles 6 and 7 of abdominal segment 3 are highlighted in the blue rectangle. A, anterior; P, posterior; L, left; R, right. (C) Motoneuron projections were visualized by anti-HRP immunostaining (red), muscle membrane is tagged with the integral membrane protein mCD8-GFP (green). The subsynaptic reticulum (SSR), with its concentration of membrane, is readily visualized with mCD8-GFP. Several synaptic boutons of the NMJ are highlighted in the blue rectangle. (D) High magnification view of the region enclosed by the blue rectangle in (C). The white bar shows the extent of one large synaptic bouton, which is represented in TEM micrographs that follow. 1264 C. E. REYNOLDS-PETERSON ET AL. projections (Fig. 1 C). Each synaptic connection consists of numer- function of synaptic activity.21 Genes required for SSR elabora- ous boutons, defined as a single bud of the axonal terminal (red) tion such as Syx18 and Akt1 are also critical for the physiological and the surrounding subsynaptic membrane compartments properties of this synapse, independent of the changes in neuro- (green) (Fig. 1 D). The intense GFP signal seen at the NMJ sur- transmitter receptors.22,23 Therefore, assembly and structure of rounding the motoneuron bouton is a function of the complex and the SSR is a critical determinant of NMJ function. Ultrastruc- dense membrane assembly of the SSR, visualized with mCD8-GFP. tural changes of the NMJ in response to reduction of sfl or ttv The SSR is a highly ordered postsynaptic membrane structure function were assessed by transmission electron microscopy where cytoskeletal, neurotransmitter receptor, and signaling mol- (TEM) of motoneuron terminals and the underlying postsynap- ecules are spatially localized.19,20 During development, the SSR tic specializations (Fig.2,AtoC,AitoCi). Reduction of either expands proportionally as the muscle cell grows. The SSR also sfl or ttv function with muscle-directed RNAi expanded the shows use-dependent plasticity, with expansion occurring as a spaces between membrane layers of the SSR. This SSR structural

Figure 2. Changes in SSR structure, mitochondrial number and morphology in HS biosynthesis compromised larval muscles. (A to C) TEM images of the SSR at a single synaptic bouton; SSR: subsynaptic reticulum; MN: motoneuron. In control animals (A and Ai: Mef2-GAL4/C), SSR comprised of multiple compactly assembled membrane stacks are tightly packed surrounding the presynaptic nerve terminal. In animals where sfl (B and Bi: Mef2-GAL4>UAS-sfl RNAi) or ttv (C and Ci: Mef2-GAL4>UAS-ttv RNAi) function was compromised by muscle-directed RNAi, SSR organization was disrupted. (Ai to Ci) Higher magnification views of regions demarked by blue squares in (A to C). (D) Quantitative analysis of membrane content in the SSR showed significant decreases in sfl RNAiand ttv RNAi-expressing muscles compared with controls. (E to G) TEM micrographs of the cytoplasm below the NMJ synapse in controls (E: Mef2-GAL4/C) and animals expressing sfl RNAi (F: Mef2-GAL4>UAS-sfl RNAi) or ttv RNAi (G: Mef2-GAL4>UAS-ttv RNAi) in the muscle. Note the paucity of mitochondria in sfl RNAi and ttv RNAi compared with controls. (H and I) Electron micrographs of mitochondrial morphology in controls (H), and animals expressing sfl RNAi (I) or ttv RNAi (J) in the muscle. Red arrows indicate individual mitochondria of interest. In I, controls show mostly rod shaped mitochondria, while RNAi against sfl or ttv lead to a large number of mitochondria that were “bent” or shorter and rounder. (K) Quantification of mitochondrial density in the subsynaptic cytosol. Compared to controls, sfl RNAi and ttv RNAi-expressing animals have fewer mitochondria. Scale bars (A to C) 1 mm, (Ai to Ci) .5 mm, (E to G) 5 mm, (H to J) 0.5 mm. Error bars denote SEM. , P < 0.001. Numbers at the bottom of the bar indicate sample sizes. AUTOPHAGY 1265 change was measured by determining the membrane area (the between the 2 membrane layers (Fig. 3 B, Bi, D, and Di show electron dense membrane bilayer) divided by the total area these structures in sfl RNAi and ttv RNAi-bearing animals), an encompassed by the SSR (with the motoneuron terminal area organization not associated with mature autophagosomes. subtracted out). Membrane elaboration was significantly reduced These double-membrane structures were often found in close in animals with either sfl or ttv RNA interference in the muscle proximity to mitochondria (Fig. 3 Di). The outer membrane of cell (Fig. 2 D). Per unit area of the postsynaptic specialization, mitochondria has been shown to contribute to autophagosome less membrane complexity is found within the SSR of animals biogenesis in mammalian cells,24 suggesting the possibility that with compromised HS biosynthesis. these vesicular structures are an intermediate in autophago- TEM analysis of animals with RNAi against sfl or ttv some formation, and that autophagy is altered in muscle cells revealed other changes in the muscle cell. Mitochondria were with compromised HS production. reduced in number (Fig. 2 E to G and K) and altered in their Autophagy is a process where cytoplasmic components are morphology (Fig. 2 H to J). In control animals mitochondria enveloped and delivered to lysosomes for degradation.25 Anum- were elongated and cylindrical (Fig. 2 H) whereas in both sfl ber of proteins are critical for the formation of autophagosomes and ttv RNAi-bearing animals mitochondria were more ellipti- and delivery of cargo destined for degradation. ref(2)P is a recep- cal (Fig. 2 I and J), in some cases horseshoe-shaped in cross- tor protein that binds to both ubiquitin-modified proteins and section (see sfl RNAi in Fig. 2 I). Atg8a, a ubiquitin-like protein required for autophagy and cova- Another notable feature observed in the TEM images was an lently attached to the outer membrane of the assembling auto- abundance of double-membrane vesicular structures (Fig. 3 A, phagosome. The levels and cellular distribution of ref(2)P B, and D), particularly in sfl RNAi-bearing animals, where their provide a measure of changes in autophagy, and anti-ref(2)P density was significantly increased compared with controls antibodies were used to examine cells with RNA interference of (Fig. 3 C). Many of these intracellular structures had material HS-biosynthetic components. Consistent with studies of

Figure 3. Accumulation of membrane-bound structures in muscle cells of larvae with compromised HS biosynthesis. (A, B and Bi) Electron micrographs of subsynaptic cytosol in controls (A: Mef2-GAL4/C) and animals expressing sfl RNAi in the muscle (B and Bi: Mef2-GAL4>UAS-sfl RNAi). (Bi) Higher magnification view of structures in the red box (whole panel) and the blue box (inset) from (B). In addition to classical autophagosomes, there are groups of membrane bound structures, some with single membrane layers (1), or nested within one another (2), and double-membrane-bound structures filled with electron dense material (3) and cytosol (inset). Mef2- GAL4>UAS-sfl RNAi animals have many more vesicular structures than controls, quantified in (C). mito: mitochondria; N: nucleus; AP: autophagosome, PM: plasma membrane. (D) Mef2-GAL4>UAS-ttv RNAi animals harbor similar membranous structures. (Di) A higher magnification view of the red-boxed area in (D) (whole panel) and the blue box (inset). Vesicles appear adjacent to mitochondria (arrows). Crescent shaped thick membrane sheets containing regions of mitochondrial matrix (arrowhead) and closed vesicles with mitochondrial matrix-like material remaining between the layers (inset) are common. Scale bars: 2 mm (A and B as well as D), 0.5 mm (Bi and Di). Error bars denote SEM. , P < 0.01. Numbers at the bottom of the bars indicate sample sizes. 1266 C. E. REYNOLDS-PETERSON ET AL. autophagy in other Drosophila cell types, blockade of autophagy largely colocalize with the ref(2)P-positive structures (Fig. 4, by RNAi against Atg8a in larval muscle produced an elevated compare A to H, and B to I). Overexpression of Atg8a28,29 or level of ref(2)P-positive puncta in the cytoplasm.26,27 This is inhibition of the Akt-MTOR signaling pathway30,31 through accompanied by an increase in ubiquitin-modified proteins that RNAi inhibition of the Drosophila class I phosphotidylinositol 3-

Figure 4. Compromised HS biosynthesis leads to increased levels of autophagy indicator proteins in larval muscle. (A to F) Confocal images of anti-ref(2)P and (H to N) anti-ubiquitin immunostaining of muscles 6 and 7, segment A3, in 3rd instar larvae. (A and H) control: Mef2-GAL4>UAS-w RNAi, (B and I) inhibition of autophagy: Mef2- GAL4>UAS-Atg8a RNAi, (C and J) enhanced autophagy (overexpression of Atg8a): Mef2-GAL4>UAS-Atg8a, (D and K, enhanced autophagy): Mef2-GAL4>UAS-Pi3K92E RNAi, (E and M): Mef2-GAL4>UAS-sfl RNAi, (F and N): Mef2-GAL4>UAS-ttv RNAi. (G) Quantification of ref(2)P immunofluorescent signal density. (O) Quantification of ubiquitin immunofluorescent signal density. Scale bar: 50 mm. Error bars denote SEM. , P < 0.05; , P < 0.01; , P < 0.001. Numbers at the bottom of the bars indicate sample sizes. AUTOPHAGY 1267 kinase Pi3K92E, conditions shown to increase autophagy in (generated by autophagosome-lysosome vesicle fusion), provid- other cell types, also increased the levels of ref(2)P and ubiquitin ing a measure of delivery of material to the degradative machin- intracellular puncta, although to a much lesser degree (Fig. 4 C ery.32 RNAi against either sfl or ttv significantly increased the andJ,DandK), presumably reflecting the increase in molecules numbers of GFP-Atg8a-labeled structures (Fig. 5 A to C and G) destined for destruction. RNA interference of either sfl or ttv as well as LysoTracker Red-positive vesicles (Fig. D to F and H). also produced an increase in both ref(2)P and ubiquitin-modi- These findings show that disruptions of HS-synthesis increased fied proteins in muscle cells (Fig.4EandM,FandN). The intracellular structures associated with autophagic degradation increase in both markers in response to sfl and ttv RNAis was from early to late steps and provide evidence that changes in ref on a scale similar to that seen in Atg8a overexpression and (2)P-positive or Atg8a-tagged components were not the conse- Pi3K92E RNAi, rather than the drastic accumulation achieved quence of a blockade in progression to lysosomes. with knockdown of Atg8a (Fig. 4 G and O). These results indi- Together, the increase in both GFP-Atg8a and LysoTracker cate that autophagy is disrupted in muscle cells with compro- Red-positive vesicles in animals with decreased HS biosynthesis mised HS biosynthesis but do not address whether there is a is consistent with an increased flux of material through the blockade or conversely an increase in autophagic flux. autophagy process. To monitor this process directly we used a Changes in autophagy mediated by compromising HS-bio- form of Atg8a that is tagged with both GFP and mCherry fluo- synthesis were further examined by monitoring both early and rescent epitopes.33 This construct produces both GFP and late elements of the degradative process (Fig. 5). Ectopically mCherry fluorescence in nonacidic compartments, but GFP expressed GFP-tagged Atg8a has been widely used to monitor emission is lost when the molecule arrives in the lysosome, the levels and cellular distribution of autophagosomes.32 Avital where protonation destroys the structure necessary for GFP but dye, LysoTracker Red selectively accumulates within acidic intra- not mCherry fluorescence.34,35 Thus, GFP and mCherry dou- cellular compartments, mainly lysosomes and autolysosomes ble-fluorescing structures (yellow) provide a measure of the

Figure 5. Accumulation of autophagic structures in muscle cells of larvae with compromised HS biosynthesis. (A to C) Confocal images of muscles 7 and 6 in controls (A: Mef2-GAL4>UAS-GFP-Atg8a) and animals also expressing RNAi constructs targeting sfl (B: Mef2-GAL4>UAS- GFP-Atg8a; UAS-sfl RNAi) or ttv (C: Mef2-GAL4>UAS- GFP- Atg8a; UAS-ttv RNAi). Autophagosomes were visualized with GFP-Atg8a. (D to F) Autophagy levels were measured with LysoTracker Red in controls (D: Mef2-GAL4/C) and animals with muscle-specific expression of sfl RNAi (E: Mef2-GAL4>UAS-sfl RNAi) or ttv RNAi (F: Mef2-GAL4>UAS-ttv RNAi). (G) Average number of GFP-Atg8a-positive patches. (H) Average number of LysoTracker Red-positive structures. Scale bar: 50 mm. Error bars denote SEM. , P < 0.05; , P < 0.01; , P < 0.001. Numbers at the bottom of the bars indicate sample sizes. 1268 C. E. REYNOLDS-PETERSON ET AL.

Figure 6. Reductions in HS biosynthesis and modification, result in a net increase in autophagic flux. (A) Changes in autophagic flux were monitored using a GFP- mCherry-Atg8a marker in muscles 6 and 7 of third instar larvae. (B to J) Early autophagosomes show both GFP and mCherry fluorescence. Autolysosomes are labeled solely with mCherry fluorescence as the acidic environment quenches GFP signal. Controls (B to D: Mef2-GAL4>UAS-GFP-mcherry-Atg8a), larvae with muscle-directed RNAi against sfl (E to G: Mef2-GAL4>UAS-GFP-mcherry-Atg8a; UAS-sfl RNAi) or ttv (H to J: Mef2-GAL4>UAS-GFP-mcherry-Atg8a; UAS-ttv RNAi). Merged images showed several mCherry-labeled patches were not colabeled with GFP in (G) sfl RNAi and (J) ttv RNAi-expressing muscle cells. (K) Compared with controls, muscles expressing RNAi constructs of sfl or ttv showed significantly more mCherry-exclusive puncta. (L) Fluorescent intensity of mCherry was similar to that of controls in sfl RNAi and significantly increased in ttv RNAi-expressing animals. (M) An anti-GFP western blot to examine lysosomal degradation of exogenous GFP-ref(2)P expressed ubiquitously under the control of da-GAL4 in total protein extracts of whole third instar larvae. The intact fusion protein migrates at approximately 130 kDa (upper box) while the free GFP domain released by lysosomal degradation runs at approximately 27 kDa (lower box). The ratio of fusion protein to free GFP, calculated by division of background-corrected GFP-ref(2)P band signal density by free GFP background-corrected intensity within each sample, is displayed below the bottom box. da-GAL4>UAS-GFP-ref(2)P without an additional transgene served as the control. Scale bar: 50 mm (B to J). Error bars denote SEM. , P < 0.05. Numbers at the bottom of the bars indicate sample sizes. number of early autophagosomes, whereas structures with only structures was significantly increased in both sfl and ttv RNAi- mCherry signal (red) indicate the relative number of autolyso- bearing animals compared with controls, with mCherry fluo- somes (Fig. 6 A). In control animals, a proportion of the rescent intensity also significantly increased by ttv inhibition mCherry fluorescence overlapped with the GFP fluorescence, (Fig. 6 K and L). These findings are consistent with published consistent with the shared labeling of early autophagosomes reports documenting the behavior of this molecular reporter (Fig. 6 B to D). In animals bearing either sfl or ttv RNAi con- under conditions of a net increase in autophagic flux.32 structs, the levels of GFP and mCherry double-positive vesicles We also investigated the previously identified increase in were elevated compared with control animals (Fig. 6 E to J). In endogenous ref(2)P levels pursuant to its implication in auto- addition, the number of mCherry-exclusive fluorescing phagic flux. Pancellular ectopic expression of a GFP-ref(2)P AUTOPHAGY 1269 fusion protein was used to examine the efficacy of ref(2)P deg- not show a nutritional stress response as sensitive to starvation radation in HS biosynthetically impaired animals by comparing as that documented for the fat body. levels of free GFP to levels of the parental GFP-ref(2)P pro- We also evaluated whether disruption of HS-biosynthesis – tein.32,36 38 An anti-GFP western blot revealed that proteolytic could alter the secretory apparatus in some manner and activate release of free GFP (»27 kDa) from GFP-ref(2)P (»130 kDa), ER stress, a known trigger of autophagy.40,41 Activation of ER as measured by the ratio between these 2 protein species, was stress mediated by Ire142-44 produces alternative splicing of an increased in animals overexpressing Atg8a to enhance autopha- Xbp1 transcript engineered to create an open reading frame for gic degradation. Treatment with RNAi against Atg8a to inhibit eGFP translation only after splicing. This construct has been used autophagy resulted in a pronounced accumulation of the GFP- successfully as a sensitive indicator of ER stress activation.45 ref(2)P fusion protein (Fig. 6 M). Conversion of GFP-ref(2)P to RNAi of sfl or ttv in the muscle, conditions that produce elevated free GFP was increased in sfl RNAi similar to the effect seen in autophagy, did not result in activation of the ER-stress reporter, Atg8a overexpression, indicating that while endogenous ref(2)P while expression of a misfolded version of the ninaE protein levels may be increased in these animals, this does not reflect (ninaEG69D)45,46 produced a robust response (Fig. S3). By this reduced degradative capacity. Inhibition of ttv caused a modest measure, elevated autophagy in animals with RNAi of HS biosyn- increase in the ratio of fusion protein to free GFP, however, thesis was not produced by activation of an ER stress response. this effect was mild in comparison to inhibition of Atg8a.In combination with the analysis of the GFP-mCherry-Atg8a, Autophagosome and mitochondrial association in sfl RNAi these results suggest that autophagosome maturation proceeds animals normally, and the increase in autophagy components indicates an overall increase in autophagic degradation. Selective autophagic elimination of mitochondria, referred to as A number of experiments were conducted to determine if mitophagy, provides a mechanism to regulate mitochondrial the effect of HS biosynthesis on autophagy in muscle was a numbers and remove damaged organelles.47,48 In sfl and ttv direct or indirect effect. Disrupting HS synthesis in muscle RNAi-expressing muscle, autophagosomes were found in the could potentially affect feeding behavior or otherwise produce vicinity of mitochondria (Fig. 3 Di). The reductions in mito- systemic activation of autophagy. Systemic autophagy can be chondrial densities were also accompanied by changes in their monitored by examining levels of autophagy markers in the fat morphology. Instead of the typical rod-like, elongated mito- body, the principal energy storage and nutritional sensing chondria seen in control animals, sfl RNAi produced curved or organ in Drosophila. Muscle-specific RNAi of sfl and ttv did cup-like mitochondria (Fig. 2 H and I). Reduction of ttv func- not induce an increase in LysoTracker Red staining in the fat tion resulted in more rounded mitochondria (Fig. 2 J). The body, a response readily observed with starvation (Fig. S2 A to proximity of vesicular structures that bear molecular tags of C, as well as H, L and M). We also monitored food ingestion in autophagosomes near mitochondria was of interest for 2 rea- animals with RNAi of HS biosynthesis in muscle using fluores- sons. First, it would inform the possibility that autophagosomal cein, a fluorescent dye incorporated in the food.39 Animals elements could derive at least in part from mitochondria, a phe- expressing muscle-directed RNAi of sfl had comparable levels nomenon documented in mammalian cells.24 Second, the loss of food ingestion compared with wild-type larvae (data not of mitochondria in the muscle cells of animals with defective shown). We also determined that a starvation protocol that HS biosynthesis could result from either a “conversion” of produces robust activation of autophagy in the fat body (Fig. S2 mitochondria to autophagosomal structures or mitophagy D, H, and M), produced no change in either LysoTracker Red directly. Mitochondria and autophagosomes were examined or Atg8a-GFP markers in the muscle (Fig. S2 D to G, H to K, simultaneously using a monoclonal mitochondria-specific anti- and M to O). Collectively these experiments show that RNAi of body 4C7 and GFP-Atg8a. Visualization of mitochondria (red) genes required for HS biosynthesis in the muscle produced ele- and autophagosomes (green) showed some mitochondria in sfl vated levels of autophagy independent of nutritional or sys- RNAi muscles in close association with autophagosomes (Fig. 7 temic activation of autophagy, and that body wall muscles do AtoC), and mitochondria enveloped by GFP-Atg8a-positive

Figure 7. Proximity of mitochondria and autophagosomes in sfl RNAi-expressing animals. (A to D) Autophagosomes were visualized with GFP-Atg8a (green) and mitochondria were detected with anti-mitochondria antibody (red) in sfl RNAi expressing muscle (Mef2-GAL4>UAS-GFP-Atg8a; UAS-sfl RNAi). Autophagosomal vesicles containing mitochondrial spheroids are indicated by arrows. (D) Three-dimensional reconstruction from serial confocal optic sections depicting autophagosomal and mitochondrial localizations in an animal expressing the sfl RNAiconstruct. Some optical sections not included to reveal interior of structure. n, nuclei. Scale bars: 10 mm(A to C), 2 mm (D). 1270 C. E. REYNOLDS-PETERSON ET AL. structures (Fig. 7 C and D; the high magnification view shown in D provides a cross-section of a mitochondrion surrounded by a GFP-Atg8a-tagged autophagosome, the 3D-reconstruction includes several confocal optical sections, but not the entire set which showed envelopment of the mitochondrion). These findings suggest a role of autophagy in the reduction of mitochondrial number in animals with compromised HS biosynthesis (Fig. 2 K).

HS biosynthesis is required for suppression of autophagy in fat body and is noncell autonomous Given the profound effects of abrogating HS-polymer biosynthesis or sulfation on autophagy in developing muscle, it was of interest to determine if this regulatory relationship exists in other cell types. The requirement for HS biosynthesis in muscle for the normal regulation of autophagy suggested the possibility that HSPGs serve this function in multiple tissues. To assess this hypothesis, the level of autophagy in fat body cells was evaluated in animals with compromised HS biosynthesis. The fat body is a tissue central to energy storage and utilization, and mounts a robust autophagic response during periods of nutritional stress, characterized by increased levels of both autophagosomes and lysosomes.49 RNA interference of sfl or ttv directed to the fat body with r4-GAL4 produced significant increases in GFP-Atg8a labeled intracellular structures (Fig. 8 A to D) as well as LysoTracker Red-positive organelles (Fig. 8 E to H) indicating that loss of HS biosynthesis permits activation of autophagy in the absence of starvation in this tissue. The importance of HS biosynthesis in regulation of autophagy in the fat body was further examined using a ttv mutant that survives to the third instar larval stage. Larvae homozygous for ttv00681 showed dramatic increases in LysoTracker Red-positive structures in the fed state (Fig. 9 A to C), consistent with the RNAi experiments. The robust activation of autophagy in fat body of ttv mutant larvae provided the tool for addressing an important mechanistic question: Is HS biosynthesis required cell autonomously, or can neighboring cells with normal HS production provide the function to achieve normal autophagy regulation? A non-cell autonomous function would support a mechanism that requires signaling in the extracel- Figure 8. Inhibition of HS biosynthesis in fat body cells induces autophagy in fed lular space, where HSPGs are known to regulate the activity larvae. (A to C) Confocal images showing increased numbers of GFP-Atg8a-positive fl autophagosomes in fat body cells of animals expressing RNAi constructs targeting of several growth factor pathways that in uence regulators sfl (B, F: r4-GAL4>UAS-GFP-Atg8a; UAS-sfl RNAi) or ttv (C, G: r4-GAL4>UAS-GFP- of autophagy, most notably PI3 kinase. Using a system for Atg8a; UAS-ttv RNAi) compared with controls (A, E: r4-GAL4>UAS-GFP-Atg8a; UAS- inducing mitotic recombination in the fat body to produce w RNAi). Autophagosomes were visualized by GFP-Atg8a (green), and nuclei were 00681 00681 stained with Hoechst (blue). (D) The average density of GFP-Atg8a puncta was sig- ttv /ttv (Fig. 9 D, cells without expression of a GFP nificantly increased in animals expressing sfl RNAi or ttv RNAi in comparison to ani- marker) and C/C cell clones (Fig. 9, bright green cells, mals expressing w RNAi (control) or no RNAi construct (not shown). (E to G) Within bearing 2 copies of GFP-expressing marker gene construct) the same animals, autophagic degradation was measured by staining with Lyso- 00681 C Tracker Red to visualize acidic organelles. (F) The average density of LysoTracker from ttv / heterozygous parental cells, genetic mosaics Red puncta was significantly increased in animals expressing sfl RNAi or ttv RNAi in comparison to animals expressing w RNAi (control) or no RNAi construct (not were created in third instar larval fat body. LysoTracker Red-positive structures were not elevated in ttv00681/ttv00681 shown). Scale bar: 50 mm. Error bars denote SEM. , P < 0.05; , P < 0.01; , P < 0.001. Numbers at the bottom of the bars indicate sample sizes. cell clones (Fig. 9 E). In all 15 cases examined, the mutant cells were in contact with either ttv00681/C or C/C cells, a common occurrence in fat body, where large isolated contact with other HS-producing cells. This result shows mutant clones are rare. These experiments demonstrated that some signaling process in the extracellular space is that ttv function was not required in a cell to maintain nor- required for the normal suppression of autophagy by mal levels of autophagy under conditions where it was in HSPGs. AUTOPHAGY 1271

Figure 9. Homozygous mutant ttv00681 causes a cell non-autonomous increase in the number of acidic organelles. (A and B) Confocal images of unfixed fat body tissue from wandering 3rd instar larvae. Acidic organelles visualized using LysoTracker Red and nuclei visualized using Hoechst (blue) in control (A: Oregon R) and homozygous mutant (B: ttv00681) larvae. (C) Density of LysoTracker Red-positive puncta was significantly increased in homozygous mutants. (D and E) Homozygous ttv00681 clones were generated by heat shock-induced FRT mediated recombination in a heterozygous background. (D) The wild-type chromosome arm is marked with UAS-mCD8 GFP, which appears the brightest in homozygous wild-type cells, while homozygous mutant cells are identified by the absence of GFP. (E) Green lines indicate tissue boundaries. White lines demark homozygous ttv00681 mutant clones. The density of acidic organelles marked by LysoTracker Red is similar in homozygous mutant clones and neighboring heterozygous cells or wild-type clone cells. Scale bars: 50 mm. Error bars denote SEM. , P < 0.001. Numbers at the bottom of the bars indicate sample sizes.

Compromising Atg gene function rescues SSR Reducing the levels of autophagy in animals with and mitochondrial phenotypes in animals with reduced sfl compromised HS biosynthesis rescues lethality function HS biosynthesis is required for viability; animals bearing The findings described above establish that HS biosynthesis mutations in key enzyme-encoding genes show developmen- is an important determinant of autophagy levels in 2 cell tal arrest and do not survive to the adult stage.11 The stage types, muscle and fat body. Compromising HS assembly of arrest is affected by the point during development at and hence HSPG structure also has morphological and which gene function is compromised, and the degree of functional consequences in muscle tissue. To determine if functional loss induced by any particular allele.11,50-53 Con- the change in autophagy mediated by altered HS biosynthe- sistent with published work describing mutant alleles, RNAi sis was responsible for these phenotypes, we investigated against sfl and ttv using a ubiquitously-expressed GAL4 whether inhibiting expression of known autophagy genes transcriptional activator, Da-GAL4, produced lethality at could ameliorate phenotypes associated with compromised second instar larval and early-pupal stages, respectively HS assembly. TEM analysis revealed that changes in SSR (Table 1). To determine if the lethality produced by reduc- morphology and membrane content of the postsynaptic tion of sfl or ttv function is due in some measure to ele- muscle cell that accompanied interference of sfl were sup- vated levels of autophagy, we examined the development of pressed with expression of an RNAi construct directed animals with both compromised HS biosynthesis and against Atg8a (Fig. 10 A to D). This reduction in Atg8a reduced Atg gene function. Ubiquitous expression of RNAi function also served to rescue the changes in mitochondrial constructs directed against Atg8a, Atg5,andAtg7 all par- density found in sfl RNAi muscle cells (Fig. 10 E). tially rescued lethality in sfl RNAi and ttv RNAi-expressing The interplay between HS-biosynthesis and autophagy animals, improving the survival stage from the second levels in affecting muscle phenotypes was further explored instar larval stage to mid-pupal stage and early-pupal to using RNAi constructs directed against other autophagy mid-pupal stage, respectively (Table 1). QPCR of animals components. Scoring animals for mitochondrial morphol- expressing these constructs was done to assess the levels of ogy, RNAi against Atg5, Atg7,orAtg8a reduced the degree sfl, ttv,andAtg gene mRNAs (Fig. S1 G). These experi- of mitochondrial morphological abnormality compared with ments showed that sfl and ttv mRNA levels were not animals expressing sfl RNAi alone (Fig. 10 F). These find- affected by Atg RNAi constructs and vice versa, indicating ings support the conclusion that HS-mediated regulation of the interaction was at the level of autophagy function. autophagy is an important function of these molecules, Expression of UAS-GFP did not produce any improvement affecting both the assembly of an important postsynaptic in survival, showing that the rescue was due to the reduced membrane specialization and the density of mitochondria at activities of the Atg genes, and not the presence of an addi- this energy-demanding cell-cell contact. tional UAS transgene (Table 1). 1272 C. E. REYNOLDS-PETERSON ET AL.

Figure 10. Mitochondrial and SSR phenotypes associated with impaired HS modification are suppressed by downregulation of autophagy. (A to C) Ultrastructure of SSR in control (A: Mef2-GAL4/C), sfl RNAi (B: Mef2-GAL4>UAS-sfl RNAi), and animals coexpressing sfl RNAi and Atg8a RNAi (C: Mef2-GAL4>UAS-Atg8a RNAi, UAS-sfl RNAi). Abnor- mally assembled SSR found in animals with loss of sfl function was partially rescued by expressing Atg8a RNAi. (D) Quantification of membrane content in the SSR region. Reduction of Atg8a function by expressing Atg8a RNAi in the muscle restored SSR membrane density in sfl RNAi animals to a significant degree. SSR, subsynaptic reticulum; MN, motoneuron. (E) Quantitative analysis of mitochondrial densities in TEM micrographs. Coexpression of Atg8a RNAi significantly increased mitochondrial density in sfl RNAi expressing animals. (F) Mito-GFP was used as a marker of mitochondrial morphology for confocal imaging in larval muscle. Mitochondrial phenotypes were assessed in animals expressing a control RNAi (UAS-w RNAi), RNAi targeting sfl, Atg5, Atg7, and Atg8a alone, and concomitant RNAi inhibition of sfl together with each of the Atg genes driven by Mef2-GAL4. Mitochondrial morphology in genotype blinded confocal images was scored as either ‘wild type’ or ‘abnormal’ using images from Oregon R larvae as baseline. Inhibition of sfl significantly increased the incidence of abnormal mitochondrial phenotypes, while the Atg genes and the RNAi control were not associated with a change in relative score frequencies. Coinhibition of Atg5 or Atg8a with sfl RNAi significantly rescued the score frequencies associated with sfl inhibition. Coinhibition of sfl and Atg7 produced an intermediate phenotype. Scale bar: 1 mm (A to C). Error bars denote SEM. , P < 0.01; , P < 0.001 for comparisons with controls. ##, P < 0.01; ###, P < 0.001 for comparisons with Mef2-GAL4>UAS-sfl RNAi. Numbers at the bottoms of the bars ((D)and E) and above the bars (F) indicate sample sizes. AUTOPHAGY 1273

Table 1. The survival stages of Drosophila with HS biosynthetic inhibition and inhi- indicates an increase in autophagic degradation in these ani- bition of Atg genes. mals. While the effect in ttv-inhibited animals is more equivo- Genotype Viability cal, it is possible that both release and degradation of GFP are increased, preventing the accumulation of free GFP in these da-GAL4/C adults da-GAL4 > UAS-sflRNAi 2nd instar larvae animals. da-GAL4 > UAS-GFP; UAS-sflRNAi 2nd instar larvae Significantly, accumulation of endogenous ref(2)P and ubiq- da-GAL4 > UAS-Atg5RNAi; UAS-sflRNAi pupae RNAi RNAi uitin in larval muscle occurs under conditions in which auto- da-GAL4 > UAS-Atg7 ; UAS-sfl pupae fl da-GAL4 > UAS-Atg8aRNAi; UAS-sflRNAi pupae phagic ux is clearly elevated, supported both by previous da-GAL4 > UASttvRNAi early pupae (white pupae) characterization of the Atg8a overexpression construct and the da-GAL4 > UAS-GFP; UAS-ttvRNAi early pupae (white pupae) RNAi RNAi results of both the mCherry-GFP-Atg8a analysis and the GFP da-GAL4 > UAS-Atg5 ; UAS-ttv pupae fl da-GAL4 > UAS-Atg7RNAi; UAS-ttvRNAi pupae conversion assay. Interpretation of ref(2)P levels in ies has pri- da-GAL4 > UAS-Atg8aRNAi; UAS-ttvRNAi pupae marily been performed in conjunction with treatments spanning days or weeks, and is generally examined in the tissue of da-GAL4 was used to ubiquitously express either no construct (the control, animals 27,32,54 experience a normal life span) or an RNAi construct targeting either sfl or ttv, adult animals. In this context, an increase in ref(2)P indi- alone or in combination with an RNAi construct targeting Atg5, Atg7,orAtg8a,or cates failure of autophagic flux. However, in the context of lar- with UAS-GFP which is not predicted to effect the lethality of HS biosynthetic val tissues, increased levels of ref(2)P have been found in fat inhibition. Genotypes are displayed on the left and the stage to which progeny 29 26 survive under standard culture conditions is displayed on the right. body tissue in response to starvation and inhibition of Tsc1, both positive regulators of autophagy. In the tissues of these young animals, it may be that the transcriptional upregulation Discussion of ref(2)P, which is activated in response to various positive regulators of autophagy including starvation,55 may outweigh HS biosynthesis deficits produce elevated levels the increased degradation of the protein through autophagy to of autophagy in muscle and fat body a greater extent than in other experimental settings. Loss of heparan sulfate biosynthesis at the NMJ produces sev- The association of double-membrane organelles with mito- eral diverse phenotypes, affecting both motoneurons and post- chondria observed in TEM images and the close proximity synaptic muscle cells. The work described here was undertaken between GFP-Atg8a-positive structures and anti-mito-stained to understand the molecular basis of 2 remarkable postsynaptic structures is intriguing for 2 reasons. First, in mammalian cells cellular phenotypes; disruption of the SSR, a postsynaptic it has been demonstrated that the outer membrane of mito- membrane specialization, and loss of mitochondria at the syn- chondria can participate in autophagosome biogenesis.24 Sec- apse. The interfering RNA-generating transgenes described ond, it raises the possibility that the reduced number of here provided cell-type specific blockade of function and postsynaptic mitochondria in muscle that is observed with allowed the characterization of phenotypes that were incom- inhibition of HS biosynthesis is a consequence of either pletely penetrant and of lesser severity in animals bearing mitophagy or some “conversion” of mitochondria to autopha- zygotic mutations. gosome precursors, and eventually autophagosomes and Expression of interfering RNAs against 2 distinct compo- autolysosomes. nents of the HS synthesis and modification apparatus altered RNA interference of 2 HS biosynthetic genes in the fat body the organization of the SSR, an elaborate membrane specializa- also produced elevated levels of autophagy markers, indicating tion that expands during development to support the synaptic that the role of HSPGs in regulating autophagy is not limited to function of the growing larval muscle. In animals with compro- muscle. Genetic mosaic analysis showed the requirement of HS mised HS biosynthesis, the SSR showed a reduction of mem- production for the control of autophagy to be non-cell autono- brane complexity indicating a deficit in either elaboration or an mous. Cells that are wild type for HS synthesis could rescue the excess of membrane retrieval from this organelle. TEM analysis level of autophagy in adjacent HS-deficient cells. This phenom- of the NMJ in these RNAi-treated animals revealed an abun- enon has been observed for HS-dependent signaling, and sup- dance of double-membrane intracellular organelles, many of ports a mechanism that involves function of HSPGs in the them in close proximity to mitochondria. A variety of molecu- extracellular environment.56 lar markers, including ref(2)P, Atg8a, and LysoTracker Red, The involvement of HS biosynthesis in controlling auto- were all elevated in animals with compromised HS biosynthe- phagy levels in Drosophila begs the question as to the generality sis. A doubly-tagged GFP-mCherry-Atg8a transgene that per- of this regulation. In human epithelial cells, degradation of mits tracking of both autophagosomes and autolysosomes HSPGs by heparinase III increased the levels of autophagic demonstrated that these vesicle types were increased upon markers, consistent with our findings in Drosophila muscle and compromising HS-biosynthesis, showing that loss of HS bio- fat body.57 There is also recent evidence that heparanase, an synthetic capacity produced a net increase in autophagy and endogenous HS-cleaving enzyme, enhances autophagy. Fur- not simply an accumulation of one vesicular intermediate. thermore, the proautophagic function of heparanase promotes A western blot examining release of free GFP from GFP-ref tumor growth and chemoresistance.58 (2)P, a commonly used method for examining autophagic flux Additional clues that HSPGs can broadly suppress auto- in yeast and Drosophila,26,38 provided further support for an phagy come from studies of SUMF1 (sulfatase modifying factor increase in the level of autophagic degradation in HS-biosyn- 1). SUMF1 catalyzes a post-translational modification that is thetically compromised animals. The increase in free GFP that required for the activity of sulfatases, including those that accompanies ubiquitous inhibition of sfl gene function degrade HS.59,60 Loss of SUMF1 activity is therefore predicted 1274 C. E. REYNOLDS-PETERSON ET AL. to increase HS levels and hence HSPG-dependent signaling. Autophagy has been previously implicated in synaptic Consistent with this model, Sumf1 mutant mice show hyperac- growth in Drosophila, negatively regulating the level of the E3 tivation of FGF18 signaling, a known HS-dependent signaling ubiquitin ligase hiw (highwire) in the motoneuron.71 However, process. Interestingly, autophagy is suppressed in Sumf1 its function in postsynaptic development has not been as well mutants, a finding consistent with a role for HSPGs in limiting studied. In C. elegans, loss of presynaptic inputs at the NMJ 61–64 fi autophagy. In humans, loss of SUMF1 is responsible for results in postsynaptic GABAA receptor traf cking to autopha- multiple sulfatase deficiency (MSD),65 a syndrome character- gosomes, a mechanism for selective removal of this neurotrans- ized by neurological deficits and bone patterning abnormalities. mitter receptor.72 Autophagy could affect postsynaptic function These findings suggest a role for HSPGs in the hypersuppres- in several ways, controlling degradation of key receptors or sion of autophagy in Sumf1 mutant animals and MSD patients. cytoskeletal elements, altering membrane trafficking, and affecting the number of mitochondria localized at synapses to provide for the high-energy demands of synaptic transmission. Mechanism of HS-dependent inhibition of autophagy In short, the precise regulation of autophagy in the nervous sys- HSPGs are ubiquitous cell surface-associated proteins that influ- tem is critical: modest decreases achieved by activation of Tor ence a large number of growth factor signaling systems, many of can result in a failure of dendritic pruning leading to behavioral which affect class I phosphoinositide 3-kinase (PI3K),6,66,67 a deficits,21 and complete abrogation of autophagy produces neu- key activator of TOR (MTOR in mammals). TOR activity in ronal death. The work reported here demonstrated that ele- turn serves to limit autophagy. Downregulation of PI3K activity vated autophagy also has consequences, decreasing the is necessary for induction of autophagy in response to nutrient membrane complexity of postsynaptic specializations and deprivation or developmentally regulated autophagy during reducing mitochondrial density at synapses. metamorphosis.49,68,69 The simplest hypothesis for how HSPGs serve to suppress autophagy is that their loss compromises several signaling pathways that activate PI3K and thus reduces the Materials and methods inhibitory input to autophagy provided by TOR. We have made Fly husbandry several observations that support this model (data not shown): Fly strains were raised on standard cornmeal/sucrose/agar 1) overexpression of Pi3K92E in the muscle reduced autophagy and suppressed the elevated autophagy resulting from loss of HS media at 25 C during embryogenesis and 30 C during larval biogenesis, 2) reduction of Pi3K92E activity produced a mito- development under a 12 h day/12 h night unless otherwise fi 1118 chondrial morphological phenotype like that found in HS-com- speci ed. Oregon-R, w , and VDRC60100 strains served as promised larvae, and 3) the mitochondrial deficit associated wild-type stocks. RNAi strains and a control strain with the with loss of HS function could be rescued by increasing Pi3K92E same genetic background were obtained from the Vienna Dro- 73 activity. These findings are suggestive, but not a direct demon- sophila RNAi Center (VDRC) : UAS-Atg8a RNAi (43097), fl stration, that HS biosynthesis affects autophagy levels via regula- UAS-Pi3K92E RNAi(38985), UAS-s RNAi(5070), UAS-ttv tion of class I PI3K activity. This mechanism is best evaluated in RNAi(4871), UAS-w RNAi (30033) and empty vector control a system where HS-directed events are elevated and their depen- (60100). UAS-GFP-mcherry-Atg8a, UAS-mCD8-GFP, UAS- dence on PI3K activity can be directly tested. GFP, UAS-mito-GFP, EP-Atg8a, and the Drosophila Trans- genic RNAi Project (TRiP)74 lines UAS-Atg5 RNAi (JF02703) and UAS-Atg7 RNAi (JF02787) were obtained from the Bloo- Autophagic regulation of postsynaptic specializations mington Drosophila Stock Center (BDSC). UAS-GFP-Atg8a75 76 Compromising HS biosynthesis produced loss of mitochondria and UAS-GFP-ref(2)P are generous gifts from T. Neufeld, 45 in the postsynaptic cell and disruption of the SSR, a postsynap- University of Minnesota, Minneapolis, MN. UAS-Xbp1-eGFP G69D tic specialization that expands during synaptic growth and is and UAS-ninaE are generous gifts from H. Steller, The responsive to synaptic activity.21 The SSR and mitochondrial Rockefeller University, New York, NY. Mef2-GAL4, r4-GAL4, phenotypes in the muscle cell were rescued by reducing the and da-GAL4 (BDSC) transposon-containing stocks were used fi fi function of genes critical for autophagy, demonstrating that for muscle-speci c, fat body-speci c, and ubiquitous expres- these cellular changes were a consequence of increased auto- sion of RNAi constructs or transgenes. The c754-GAL4 hsFLP; G13 G13 00681 phagy. These results provide evidence that the precise control FRT UAS-mCD8GFP and FRT ttv /CyOGFP strains of autophagy is critical for the assembly of postsynaptic mem- were generated using stocks obtained from the BDSC. brane specializations; enhancing autophagy decreased the com- When starvation was involved in an experimental procedure, plexity of the SSR membrane structure. These findings are in third instar larvae were removed from food media and placed on fi concert with recent work describing the role of autophagy in lter paper soaked with 20% sucrose (Mallinckrodt Chemicals, – dendritic spine elaboration and pruning.70 Hyperactive MTOR 8360 06) solution in Petri dishes for 3 h before dissection. Fed produced both reductions in autophagy and a deficiency in control larvae were maintained in food media until dissection. dendritic pruning in mouse cortical projection neurons. This Tor-mediated increase in dendritic spine density could be res- Real-time quantitative polymerase chain reaction cued by inhibition of Tor by rapamycin, and the rescue required autophagy gene function. Our data support this link Total RNA was extracted from whole second (n D»40) or between autophagy and the control of postsynaptic third (n D»20) instar larvae using NeocleoSpin RNA extrac- specializations. tion kit (Macherey-Nagel, 740955) following the AUTOPHAGY 1275 manufacturer’s manual. All genotypes were analyzed in biologi- Samples were then incubated overnight with anti-heparan sul- cal triplicate. RNAi constructs were expressed ubiquitously fate antibody (1:100, 3G10; Seikagaku, 270426) at 4C. Signals under the control of Da-GAL4, and crosses were maintained at were amplified using Tyramide Signal Amplification Fluores- 25C. For each sample, 0.5 mg of total RNA was used for cence Systems (PerkinElmer, NEL744001KT). All of these pro- reverse transcription of RNA in 10-ml reaction mix using cedures were performed at room temperature. qScript cDNA SuperMix (Quanta Biosciences, 95048) to gener- For LysoTracker Red staining in muscle cells, wandering C ate first strand cDNA. Real-time quantitative PCR was per- third instar larvae were dissected in Ca2 -free HL-3 (70 mM TM Ò TM formed using the Perfecta SYBR Green Supermix, ROX NaCl, 5 mM KCl, 20 mM MgCl2, 10 mM NaHCO3, 5 mM tre- (Quanta Biosciences, 95055) on StepOnePlusTM (Applied halose, 115 mM sucrose, 5 mM HEPES, pH 7.2) and incubated Biosystems). Primers used: sfl: forward TCCCGCACCCAT- with 100 nM LysoTracker Red DND-99 (Life Technologies, L- TATCAAC, reverse TCAAACACAATCACGCCATAAC; ttv: 7528) for 5 min in the absence of light, rinsed twice with PBS, forward ATCCCGCTCTTCCACAAAC, reverse GTCGTA- and imaged immediately. All of these procedures were per- TCTGTCGTATTCCCG; Atg5: forward GCACGCACGG- formed at room temperature. CATTGATCTACA, reverse GCCCTGGGATTTGCTGGAAT; For LysoTracker Red staining in fat body cells, fed or starved Atg7: forward TTTTGCCTCACTCCATCCGTGG, reverse third instar larvae were dissected in PBS. Fat bodies were incu- ATCCTCGTCGCTATCGGACATG; Atg8a: forward CAAC- bated in 100 nM LysoTracker Red DND-99, 1 mM Hoechst CAACCAACCAACTTTCC, reverse GCATTCGCACGGAT- 33342 (Life Technologies, H3570) in PBS for 2 min. Fat bodies CAATTAC; RpL32: forward GCAAGCCCAAGGGTATCGA, were then rinsed twice with PBS, transferred to 40% glycerol on reverse ACCGATGTTGGGCATCAGA. The relative RNA lev- glass slides, covered, and imaged immediately. All of these pro- els of each sample were found by comparing the amplification cedures were performed at room temperature. of the gene(s) of interest first to the internal control gene RpL32, and then calculating the relative change in comparison Mosaic analysis to a likewise internally corrected value from the genotypic control. c754-GAL4 hsFLP; FRTG13 UAS-mCD8-GFP virgin females were mated with FRTG13 ttv00681/CyOGFP males. Eggs were collected in yeast paste on grape juice agar (2.2% agar, 1.2% Immunohistochemistry and lysotracker red staining sucrose, 25% grape juice concentrate) plates for 6 h, immedi- Wandering third instar larvae were dissected in ice-cold ately followed by a 1-h heat shock at 37C. The eggs were C Ca2 freeHL-3medium(70mMNaCl,5mMKCl,20mM allowed to recover overnight at 25C before being transferred MgCl2,10mMNaHCO3, 5 mM trehalose [Alfa Aesar, to food medium. Larvae were reared at 25 C on instant food A19434], 100 mM sucrose, 5 mM HEPES, pH 7.2). Samples medium under a 12 h light/dark cycle. Feeding stage 3rd instar were fixed in 4% paraformaldehyde for 30 min (except sam- larvae were picked out of the food media and sorted by gender, ples for anti-HS antibody staining), followed by intensive discarding the females. Male larvae bearing the CyOGFP bal- washing with phosphate-buffered saline (PBS: 137 mM ancer were identified by strong GFP fluorescence in the testes fl NaCl, 2.7 mM KCl, 10 mM Na2HPO4,1.8mMKH2HPO4, and discarded. The remaining male larvae exhibiting GFP uo- pH 7.4). Muscle preparations with no trace of fixative solu- rescence in the pattern directed by c754-GAL4>UAS- tion left were blocked in 10% normal goat serum (MP Bio- mCD8GFP only were dissected and stained with LysoTracker medicals, IC642921) for 1 h at room temperature and then Red as described above. Homozygous mutant clones were iden- incubated with primary antibodies for at least 12 h at 4C. tified by loss of the mCD8GFP marker. Samples were intensively washed with PBST (PBS plus .1% Triton X-100 [IBI Scientific, IB07100]) and incubated with Image acquisition and analysis secondary antibody for 1 h at room temperature. Muscle preparations were then washed as before and incubated Images were acquired at room temperature using an Olympus overnight in mounting media before being mounted on Fluoview FV1000 laser-scanning confocal microscope (Olym- glass slides. Primary antibodies were used at following con- pus America, Waltham, MA, USA). FV10-ASW 2.1 software centrations: goat anti-HRP antibody (1:1000, conjugated (Olympus, Waltham, MA, USA) was used to capture images. with Alexa fluor 488; Jackson ImmunoResearch Laborato- When more than one fluorophore was detected, sequential line ries, 123–545–021), anti-ubiquitin antibody (1:1000, FK2; scanning was performed to avoid spectral bleed through arti- Enzo, BML-PW8810–0100), anti-ref(2)P26 (1:5000) and facts. Images of samples with different genotypes within a sin- anti-mitochondria antibody (1:500, 4C7; Developmental gle experiment were captured, processed, and analyzed using Studies Hybridoma Bank, 4C7). Alexa Fluor fluorescence- the same settings. Images were presented as Z-stacks of maxi- conjugated secondary antibodies (1:1000) were obtained mum intensity projections using Imaris 7.3 software (Bitplane from Life Technologies (A21070, A21050, A11003, A11010, Inc.). Three-dimensional surface reconstructions were gener- A11001, A11008). ated from serial images taken by confocal microscopy using For heparan sulfate staining, body-wall muscle preparations Imaris 7.3. The numbers of or area covered by GFP-Atg8a, were fixed in Bouin fixative (70% picric acid, 25% formalde- GFP-mcherry-Atg8a, ref(2)P, ubiquitin, and LysoTracker Red- hyde [37%], 5% glacial acetic acid) for 80 min, quenched with positive signals were measured using ImageJ1.42q. The abso- 0.5% H2O2 for 1 h, and treated with 20 mU heparitinase I (Sei- lute values from each animal were individually normalized kagaku, 100704) in heparitinase I buffer for 2 h at 37C. by muscle size/tissue area. Adobe Photoshop CS5 (Adobe 1276 C. E. REYNOLDS-PETERSON ET AL.

Systems Inc.) was then used to crop images and adjust bright- buffer, pH 7.0) and en bloc staining (2% uranyl acetate) were ness and contrast. All adjustments to contrast and brightness performed in a dark container for 2 h each. Samples were made to ease interpretation of confocal images were applied washed with 0.1 M sodium cacodylate buffer (pH 7.4) between identically to all genotypes within the experiment. fixations and before staining. Muscle preparations were then dehydrated (gradient ethanol range from 50% to 100%, 100% acetone, 70 min in total) and infiltrated (1:1 acetone:resin, 1:3 Western blot acetone:resin, 100% resin, 2 d in total). Samples were embed- Ubiquitous expression of both UAS-GFP-ref(2)P (GFP-ref[2]P) ded in Spurr’s low viscosity resin (Electron Microscopy Scien- together with RNAi or overexpression constructs was directed ces, 14300) at 60C for »60 h, allowing for polymerization. by da-GAL4. Larvae were reared at 25C. Five wandering 3rd The sample cubes were sectioned to thin (»70 nm) slices instar larvae were homogenized by plastic pestle in 200 ml ice and images were obtained with a JEOL1200 transmission elec- cold protein extraction buffer (20 mM HEPES, 100 mM KCl, tron microscope (Peabody, MA, USA) and analyzed by 10 mM EDTA, 1 mM DTT, .1% Triton X-100, 5% glycerol, 1X ImageJ1.42q. protease inhibitor cocktail Complete [Roche, 10184600])77 for The measurement of membrane content provided a way to each genotype. Samples were centrifuged for 10 min, and the assess the density of SSR. It was calculated as the area of the supernatant was removed to a clean tube. Protein concentra- electron dense membrane bilayers divided by the total area tion was measured by BCA Protein Assay (Pierce, 23227), and encompassed by the SSR (with the motoneuron terminal area samples were diluted to a standard concentration using 4x pro- subtracted out). The number of autophagosomes was counted tein loading buffer78 and protein extraction buffer, then boiled manually and the values were normalized by muscle area for at 95C for 5 min. Approximately 25 mg total protein per sam- each graph. Mitochondrial density was determined as the area ple was run on an 8% polyacrylamide gel. After transfer, the of electron dense mitochondria divided by the total area of nitrocellulose membrane was rinsed in TBS (50 mM Tris-Cl, muscle surface. 150 mM NaCl, pH 7.6) and dried overnight. The membrane was rehydrated and blocked in 0.5% TBST (TBS plus 0.5% Statistical analysis Tween 20 [BDH, 4210])79 and incubated overnight at 4C with rabbit polyclonal anti-GFP (1:1500, CAB3211; Thermo Scien- Statistical analyses for quantitative data were performed with tific, CAB4211). Goat anti-rabbit-HRP secondary antibody was Minitab Release 16 (Minitab). All data points were presented as applied at 1:3000 (Invitrogen, 31430). ECL detection was per- mean § SEM. Normality of data distribution was determined formed using G:BOX Chemi XG4 (Syngene, Frederick, MD, using probability plots. Comparisons between 2 groups were USA) which was also used to invert coloration of resulting performed using the Student t test for normally distributed images. Band signal density was assessed using ImageJ1.42q. data or Mann-Whitney test for nonparametric data. Compari- sons between more than 2 groups were performed using ANOVA for normally distributed data or Kruskal-Wallis for Assessment of UAS-mito-GFP phenotypes nonparametric data, followed by the Tukey test or Mann-Whit- Confocal images taken from animals bearing no RNAi expres- ney test. When analyzing the contingency table of mitochon- sion construct were carefully examined and analyzed based on drial morphology, expected values for occurrence of an visible features. A summary of common features and the scope abnormal score were often less than 5 so pairwise comparisons of variation seen in those features between images was gener- were performed using the Fisher exact test. ated by assessing the fluorescent intensity of the mito-GFP reporter, the mitochondrial shapes, sizes, and density within Abbreviations each animal. The summaries of these 4 features were then used as a qualitative standard for what constitutes a normal mito- Atg5 autophagy-related 5 chondrial phenotype as observed by mito-GFP. The confocal Atg7 autophagy-related 7 images from all animals, including these wild-type controls, Atg8a autophagy-related 8a were then blinded by replacing the image identifier with a ran- botv brother of tout-velu (Drosophila EXT3 homolog) domly generated number (nonrepeating integer sequence gen- ER endoplasmic reticulum erated by random.org) and scored as ‘normal’ or ‘substantially Ext2 exostosin glycosyltransferase 2 abnormal’ based on comparison to the qualitative standard of HS heparan sulfate normality. The scored images were then unblinded and the HSPG heparan sulfate proteoglycan results were assembled into a contingency table. mCD8-GFP Mus musculus CD8 protein fused with green fluorescent protein MSD multiple sulfatase deficiency Transmission electron microscopy NMJ neuromuscular junction C Wandering third instar larvae were dissected in ice-cold Ca2 - Pi3K92E phosphatidylinositol 3-kinase class I (Drosophila free HL-3 medium and fixed overnight with TEM fixative solu- gene) tion (1.5% glutaraldehyde, 2.5% paraformaldehyde, 1.8 mM PI3K class I phosphatidylinositol 3-kinase, species C Ca2 in 0.1 M sodium-cacodylate buffer, pH 7.0) at 4C. Post- nonspecific fixation (1% osmium tetroxide in 0.1 M sodium-cacodylate QPCR quantitative polymerase chain reaction AUTOPHAGY 1277 ref(2)P refractory to sigma P (Drosophila SQSTM1 neuromuscular junction. J Cell Biol 2013; 200:219-33; homolog) PMID:23319599; https://doi.org/10.1083/jcb.201207036 SUMF1 sulfatase modifying factor 1 (Homo sapiens) [11] Ren Y, Kirkpatrick CA, Rawson JM, Sun M, Selleck SB. Cell Sumf1 sulfataste modifying factor 1 (Mus musculus) type-specific requirements for heparan sulfate biosynthesis at the drosophila neuromuscular junction: Effects on synapse function, SSR subsynaptic reticulum fi fl membrane traf cking, and mitochondrial localization. J Neurosci s sulfateless (Drosophila N-deacetylase N- 2009; 29:8539-50; PMID:19571145; https://doi.org/10.1523/ sulfotransferase) JNEUROSCI.5587-08.2009 TEM transmission electron microscopy [12] Bellaiche Y, The I, Perrimon N. Tout-velu is a Drosophila homo- TOR target of rapamycin (MTOR in mammalian logue of the putative tumour suppressor EXT-1 and is needed for Hh species) diffusion. Nature 1998; 394:85-8; PMID:9665133; https://doi.org/ 10.1038/27932 ttv tout-velu (Drosophila EXT1 homolog) [13] Duffy JB. GAL4 system in Drosophila: a fly geneticist’s Swiss army w white knife. Genesis 2002; 34:1-15; PMID:12324939; https://doi.org/ 10.1002/gene.10150 Disclosure of potential conflicts of interest [14] Han C, Belenkaya TY, Khodoun M, Tauchi M, Lin X, Lin X. Distinct and collaborative roles of Drosophila EXT family proteins in mor- No potential conflicts of interest were disclosed. phogen signalling and gradient formation. Development 2004; 131:1563-75; PMID:14998928; https://doi.org/10.1242/dev.01051 [15] Lin H, Huber R, Schlessinger D, Morin PJ. Frequent silencing of the Acknowledgments GPC3 gene in ovarian cancer cell lines. Cancer Res 1999; 59:807-10; PMID:10029067 We are grateful to Thomas Neufeld, Hermann Steller, the Bloomington [16] Toyoda H, Kinoshita-Toyoda A, Fox B, Selleck SB. Structural analy- fl Drosophila stock center and the Vienna Drosophila RNAi Center for y sis of glycosaminoglycans in animals bearing mutations in sugarless, stocks. We are grateful to Richard Ordway for sharing the ref(2)P anti- sulfateless, and tout-velu. Drosophila homologues of vertebrate genes body. We also acknowledge the Microscopy and Cytometry Facility at The encoding glycosaminoglycan biosynthetic enzymes. J Biol Chem Pennsylvania State University for technical assistance with electron 2000; 275:21856-61; PMID:10806213; https://doi.org/10.1074/jbc. microscopy. M003540200 [17] Toyoda H, Kinoshita-Toyoda A, Selleck SB. Structural analysis of glycosaminoglycans in Drosophila and Caenorhabditis elegans and Funding demonstration that tout-velu, a Drosophila gene related to EXT This work was supported by National Institutes of Health grant 5 R01 tumor suppressors, affects heparan sulfate in vivo. J Biol Chem 2000; GM054832–15 and funds to SBS from The Pennsylvania State University. 275:2269-75; PMID:10644674; https://doi.org/10.1074/jbc.275.4. 2269 [18] Schmid A, Chiba A, Doe CQ. Clonal analysis of Drosophila embry- References onic neuroblasts: neural cell types, axon projections and muscle targets. Development 1999; 126:4653-89; PMID:10518486 [1] Lin X. Functions of heparan sulfate proteoglycans in cell signaling [19] Atwood HL, Govind CK, Wu CF. Differential ultrastructure of syn- during development. Development 2004; 131:6009-21; aptic terminals on ventral longitudinal abdominal muscles in Dro- PMID:15563523; https://doi.org/10.1242/dev.01522 sophila larvae. J Neurobiol 1993; 24:1008-24; PMID:8409966; https:// [2] Lander AD, Selleck SB. The elusive functions of proteoglycans: in doi.org/10.1002/neu.480240803 vivo veritas. J Cell Biol 2000; 148:227-32; PMID:10648554; https:// [20] Gorczyca M, Popova E, Jia XX, Budnik V. The gene mod(mdg4) doi.org/10.1083/jcb.148.2.227 affects synapse specificity and structure in Drosophila. J Neurobiol [3] Bishop JR, Schuksz M, Esko JD. Heparan sulphate proteoglycans 1999; 39:447-60; PMID:10363916; https://doi.org/10.1002/(SICI) fine-tune mammalian physiology. Nature 2007; 446:1030-7; 1097-4695(19990605)39:3%3c447::AID-NEU10%3e3.0.CO;2-Q PMID:17460664; https://doi.org/10.1038/nature05817 [21] Teodoro RO, Pekkurnaz G, Nasser A, Higashi-Kovtun ME, Balakir- [4] Hacker U, Nybakken K, Perrimon N. Heparan sulphate proteogly- eva M, McLachlan IG, Camonis J, Schwarz TL. Ral mediates activity- cans: the sweet side of development. Nat Rev Mol Cell Biol 2005; dependent growth of postsynaptic membranes via recruitment of the 6:530-41; PMID:16072037; https://doi.org/10.1038/nrm1681 exocyst. EMBO J 2013; 32:2039-55; PMID:23812009; https://doi.org/ [5] Kirkpatrick CA, Selleck SB. Heparan sulfate proteoglycans at a 10.1038/emboj.2013.147 glance. J Cell Sci 2007; 120:1829-32; PMID:17515480; https://doi. [22] Gorczyca D, Ashley J, Speese S, Gherbesi N, Thomas U, Gundel- org/10.1242/jcs.03432 finger E, Gramates LS, Budnik V. Postsynaptic membrane addi- [6] Sarrazin S, Lamanna WC, Esko JD. Heparan sulfate proteoglycans. tion depends on the Discs-Large-interacting t-SNARE Gtaxin. J Cold Spring Harb Perspect Biol 2011; 3:a004952; PMID:21690215; Neurosci 2007; 27:1033-44; PMID:17267557; https://doi.org/ https://doi.org/10.1101/cshperspect.a004952 10.1523/JNEUROSCI.3160-06.2007 [7] Selleck SB. Proteoglycans and pattern formation: sugar biochemistry [23] Lee HG, Zhao N, Campion BK, Nguyen MM, Selleck SB. Akt meets developmental genetics. Trends Genet 2000; 16:206-12; regulates glutamate receptor trafficking and postsynaptic mem- PMID:10782114; https://doi.org/10.1016/S0168-9525(00)01997-1 brane elaboration at the Drosophila neuromuscular junction. [8] Johnson KG, Tenney AP, Ghose A, Duckworth AM, Higashi ME, Dev Neurobiol 2013; 73:723-43; PMID:23592328; https://doi.org/ Parfitt K, Marcu O, Heslip TR, Marsh JL, Schwarz TL, et al. The 10.1002/dneu.22086 HSPGs syndecan and dallylike bind the receptor phosphatase [24] Hailey DW, Rambold AS, Satpute-Krishnan P, Mitra K, Sougrat R, LAR and exert distinct effects on synaptic development. Neuron Kim PK, Lippincott-Schwartz J. Mitochondria supply membranes 2006; 49:517-31; PMID:16476662; https://doi.org/10.1016/j. for autophagosome biogenesis during starvation. Cell 2010; 141:656- neuron.2006.01.026 67; PMID:20478256; https://doi.org/10.1016/j.cell.2010.04.009 [9] Stryker E, Johnson KG. LAR, liprin alpha and the regulation of active [25] Klionsky DJ. Autophagy: from phenomenology to molecular under- zone morphogenesis. J Cell Sci 2007; 120:3723-8; PMID:17959628; standing in less than a decade. Nat Rev Mol Cell Biol 2007; 8:931-7; https://doi.org/10.1242/jcs.03491 PMID:17712358; https://doi.org/10.1038/nrm2245 [10] Kamimura K, Ueno K, Nakagawa J, Hamada R, Saitoe M, Maeda N. [26] Pircs K, Nagy P, Varga A, Venkei Z, Erdi B, Hegedus K, Juhasz G. Perlecan regulates bidirectional Wnt signaling at the Drosophila Advantages and limitations of different p62-based assays for 1278 C. E. REYNOLDS-PETERSON ET AL.

estimating autophagic activity in Drosophila. PLoS One 2012; 7: [42] Ruegsegger U, Leber JH, Walter P. Block of HAC1 mRNA translation e44214; PMID:22952930; https://doi.org/10.1371/journal. by long-range base pairing is released by cytoplasmic splicing upon pone.0044214 induction of the unfolded protein response. Cell 2001; 107:103-14; [27] DeVorkin L, Go NE, Hou YC, Moradian A, Morin GB, Gorski SM. PMID:11595189; https://doi.org/10.1016/S0092-8674(01)00505-0 The Drosophila effector caspase Dcp-1 regulates mitochondrial [43] Shen X, Ellis RE, Lee K, Liu CY, Yang K, Solomon A, Yoshida H, dynamics and autophagic flux via SesB. J Cell Biol 2014; 205:477-92; Morimoto R, Kurnit DM, Mori K, et al. Complementary signaling PMID:24862573; https://doi.org/10.1083/jcb.201303144 pathways regulate the unfolded protein response and are required for [28] Simonsen A, Cumming RC, Brech A, Isakson P, Schubert DR, Finley C. elegans development. Cell 2001; 107:893-903; PMID:11779465; KD. Promoting basal levels of autophagy in the nervous system https://doi.org/10.1016/S0092-8674(01)00612-2 enhances longevity and oxidant resistance in adult Drosophila. Auto- [44] Yoshida H, Oku M, Suzuki M, Mori K. pXBP1(U) encoded in phagy 2008; 4:176-84; PMID:18059160; https://doi.org/10.4161/ XBP1 pre-mRNA negatively regulates unfolded protein response auto.5269 activator pXBP1(S) in mammalian ER stress response. J Cell Biol [29] Bartlett BJ, Isakson P, Lewerenz J, Sanchez H, Kotzebue RW, 2006; 172:565-75; PMID:16461360; https://doi.org/10.1083/ Cumming RC, Harris GL, Nezis IP, Schubert DR, Simonsen A, et al. jcb.200508145 p62, Ref(2)P and ubiquitinated proteins are conserved markers of [45] Ryoo HD, Domingos PM, Kang MJ, Steller H. Unfolded protein neuronal aging, aggregate formation and progressive autophagic response in a Drosophila model for retinal degeneration. EMBO J defects. Autophagy 2011; 7:572-83; PMID:21325881; https://doi.org/ 2007; 26:242-52; PMID:17170705; https://doi.org/10.1038/sj. 10.4161/auto.7.6.14943 emboj.7601477 [30] Petiot A, Ogier-Denis E, Blommaart EF, Meijer AJ, Codogno P. [46] O’Tousa JE, Baehr W, Martin RL, Hirsh J, Pak WL, Applebury ML. Distinct classes of phosphatidylinositol 30-kinases are involved in sig- The Drosophila ninaE gene encodes an opsin. Cell 1985; 40:839-50; naling pathways that control macroautophagy in HT-29 cells. J Biol PMID:2985266; https://doi.org/10.1016/0092-8674(85)90343-5 Chem 2000; 275:992-8; PMID:10625637; https://doi.org/10.1074/ [47] Johansen T, Lamark T. Selective autophagy mediated by autophagic jbc.275.2.992 adapter proteins. Autophagy 2011; 7:279-96; PMID:21189453; [31] Zhai C, Cheng J, Mujahid H, Wang H, Kong J, Yin Y, Li J, Zhang Y, https://doi.org/10.4161/auto.7.3.14487 Ji X, Chen W. Selective inhibition of PI3K/Akt/mTOR signaling [48] Youle RJ, Narendra DP. Mechanisms of mitophagy. Nat Rev Mol pathway regulates autophagy of macrophage and vulnerability of Cell Biol 2011; 12:9-14; PMID:21179058; https://doi.org/10.1038/ atherosclerotic plaque. PLoS One 2014; 9:e90563; PMID:24599185; nrm3028 https://doi.org/10.1371/journal.pone.0090563 [49] Scott RC, Schuldiner O, Neufeld TP. Role and regulation of starva- [32] Nagy P, Varga A, Kovacs AL, Takats S, Juhasz G. How and why to tion-induced autophagy in the Drosophila fat body. Dev Cell 2004; study autophagy in Drosophila: it’s more than just a garbage chute. 7:167-78; PMID:15296714; https://doi.org/10.1016/j.devcel.2004.07.009 Methods 2015; 75:151-61; PMID:25481477; https://doi.org/10.1016/j. [50] Baeg GH, Lin X, Khare N, Baumgartner S, Perrimon N. Heparan sul- ymeth.2014.11.016 fate proteoglycans are critical for the organization of the extracellular [33] Nezis IP, Shravage BV, Sagona AP, Lamark T, Bjorkoy G, Johansen distribution of Wingless. Development 2001; 128:87-94; T, Rusten TE, Brech A, Baehrecke EH, Stenmark H. Autophagic deg- PMID:11092814 radation of dBruce controls DNA fragmentation in nurse cells during [51] Lin X, Buff EM, Perrimon N, Michelson AM. Heparan sulfate pro- late Drosophila melanogaster oogenesis. J Cell Biol 2010; 190:523-31; teoglycans are essential for FGF receptor signaling during Drosophila PMID:20713604; https://doi.org/10.1083/jcb.201002035 embryonic development. Development 1999; 126:3715-23; [34] Kimura S, Noda T, Yoshimori T. Dissection of the autophagosome PMID:10433902 maturation process by a novel reporter protein, tandem fluorescent- [52] Lin X, Perrimon N. Dally cooperates with Drosophila Frizzled 2 to tagged LC3. Autophagy 2007; 3:452-60; PMID:17534139; https://doi. transduce Wingless signalling. Nature 1999; 400:281-4; org/10.4161/auto.4451 PMID:10421372; https://doi.org/10.1038/22343 [35] Kneen M, Farinas J, Li Y, Verkman AS. Green fluorescent protein as [53] Perrimon N, Lanjuin A, Arnold C, Noll E. Zygotic lethal mutations a noninvasive intracellular pH indicator. Biophys J 1998; 74:1591-9; with maternal effect phenotypes in Drosophila melanogaster. II. Loci PMID:9512054; https://doi.org/10.1016/S0006-3495(98)77870-1 on the second and third chromosomes identified by P-element- [36] Nagy P, Varga A, Pircs K, Hegedus K, Juhasz G. Myc-driven over- induced mutations. Genetics 1996; 144:1681-92; PMID:8978055 growth requires unfolded protein response-mediated induction of [54] Takats S, Nagy P, Varga A, Pircs K, Karpati M, Varga K, Kovacs AL, autophagy and antioxidant responses in Drosophila melanogaster. Hegedus K, Juhasz G. Autophagosomal Syntaxin17-dependent lyso- PLoS Genet 2013; 9:e1003664; PMID:23950728; https://doi.org/ somal degradation maintains neuronal function in Drosophila. J Cell 10.1371/journal.pgen.1003664 Biol 2013; 201:531-9; PMID:23671310; https://doi.org/10.1083/ [37] Mauvezin C, Ayala C, Braden CR, Kim J, Neufeld TP. Assays to jcb.201211160 monitor autophagy in Drosophila. Methods 2014; 68:134-9; [55] Sahani MH, Itakura E, Mizushima N. Expression of the autophagy PMID:24667416; https://doi.org/10.1016/j.ymeth.2014.03.014 substrate SQSTM1/p62 is restored during prolonged starvation [38] Klionsky DJ, Abdelmohsen K, Abe A, Abedin MJ, Abeliovich H, Ace- depending on transcriptional upregulation and autophagy-derived vedo Arozena A, Adachi H, Adams CM, Adams PD, Adeli K, et al. amino acids. Autophagy 2014; 10:431-41; PMID:24394643; https:// Guidelines for the use and interpretation of assays for monitoring doi.org/10.4161/auto.27344 autophagy (3rd edition). Autophagy 2016; 12:1-222; [56] Jakobsson L, Kreuger J, Holmborn K, Lundin L, Eriksson I, Kjellen L, PMID:26799652; https://doi.org/10.1080/15548627.2015.1100356 Claesson-Welsh L. Heparan sulfate in trans potentiates VEGFR- [39] Gasque G, Conway S, Huang J, Rao Y, Vosshall LB. Small molecule mediated angiogenesis. Dev Cell 2006; 10:625-34; PMID:16678777; drug screening in Drosophila identifies the 5HT2A receptor as a https://doi.org/10.1016/j.devcel.2006.03.009 feeding modulation target. Sci Rep 2013; 3:srep02120; [57] Ding Z, Wang X, Khaidakov M, Liu S, Dai Y, Mehta JL. Degradation PMID:23817146; https://doi.org/10.1038/srep02120 of heparan sulfate proteoglycans enhances oxidized-LDL-mediated [40] Ogata M, Hino S, Saito A, Morikawa K, Kondo S, Kanemoto S, autophagy and apoptosis in human endothelial cells. Biochem Bio- Murakami T, Taniguchi M, Tanii I, Yoshinaga K, et al. Autophagy is phys Res Commun 2012; 426:106-11; PMID:22910414; https://doi. activated for cell survival after endoplasmic reticulum stress. Mol org/10.1016/j.bbrc.2012.08.044 Cell Biol 2006; 26:9220-31; PMID:17030611; https://doi.org/10.1128/ [58] Shteingauz A, Boyango I, Naroditsky I, Hammond E, Gruber M, MCB.01453-06 Doweck I, Ilan N, Vlodavsky I. Heparanase enhances Tumor growth [41] Yorimitsu T, Nair U, Yang Z, Klionsky DJ. Endoplasmic reticulum and Chemoresistance by promoting autophagy. Cancer Res 2015; stress triggers autophagy. J Biol Chem 2006; 281:30299-304; 75:3946-57; PMID:26249176; https://doi.org/10.1158/0008-5472. PMID:16901900; https://doi.org/10.1074/jbc.M607007200 CAN-15-0037 AUTOPHAGY 1279

[59] Cosma MP, Pepe S, Annunziata I, Newbold RF, Grompe M, Parenti induced by ecdysone through regulation of the PI3K pathway. Dev G, Ballabio A. The multiple sulfatase deficiency gene encodes an Cell 2004; 7:179-92; PMID:15296715; https://doi.org/10.1016/j. essential and limiting factor for the activity of sulfatases. Cell 2003; devcel.2004.07.005 113:445-56; PMID:12757706; https://doi.org/10.1016/S0092-8674 [70] Tang G, Gudsnuk K, Kuo SH, Cotrina ML, Rosoklija G, Sosunov A, (03)00348-9 Sonders MS, Kanter E, Castagna C, Yamamoto A, et al. Loss of [60] Dierks T, Schmidt B, Borissenko LV, Peng J, Preusser A, Mariappan mTOR-dependent macroautophagy causes autistic-like synaptic M, von Figura K. Multiple sulfatase deficiency is caused by mutations pruning deficits. Neuron 2014; 83:1131-43; PMID:25155956; https:// in the gene encoding the human C(alpha)-formylglycine generating doi.org/10.1016/j.neuron.2014.07.040 enzyme. Cell 2003; 113:435-44; PMID:12757705; https://doi.org/ [71] Shen W, Ganetzky B. Autophagy promotes synapse development in 10.1016/S0092-8674(03)00347-7 Drosophila. J Cell Biol 2009; 187:71-9; PMID:19786572; https://doi. [61] Settembre C, Arteaga-Solis E, McKee MD, de Pablo R, Al Awqati Q, org/10.1083/jcb.200907109 Ballabio A, Karsenty G. Proteoglycan desulfation determines the effi- [72] Rowland AM, Richmond JE, Olsen JG, Hall DH, Bamber BA. Pre- ciency of chondrocyte autophagy and the extent of FGF signaling synaptic terminals independently regulate synaptic clustering and during endochondral ossification. Genes Dev 2008; 22:2645-50; autophagy of GABAA receptors in Caenorhabditis elegans. J Neuro- https://doi.org/10.1101/gad.1711308 sci 2006; 26:1711-20; https://doi.org/10.1523/JNEUROSCI.2279- [62] Settembre C, Fraldi A, Jahreiss L, Spampanato C, Venturi C, Medina 05.2006 D, de Pablo R, Tacchetti C, Rubinsztein DC, Ballabio A. A block of [73] Dietzl G, Chen D, Schnorrer F, Su KC, Barinova Y, Fellner M, Gasser autophagy in lysosomal storage disorders. Hum Mol Genet 2008; B, Kinsey K, Oppel S, Scheiblauer S, et al. A genome-wide transgenic 17:119-29; PMID:17913701; https://doi.org/10.1093/hmg/ddm289 RNAi library for conditional gene inactivation in Drosophila. Nature [63] Settembre C, Fraldi A, Rubinsztein DC, Ballabio A. Lysosomal stor- 2007; 448:151-6; PMID:17625558; https://doi.org/10.1038/ age diseases as disorders of autophagy. Autophagy 2008; 4:113-4; nature05954 PMID:18000397; https://doi.org/10.4161/auto.5227 [74] Perkins LA, Holderbaum L, Tao R, Hu Y, Sopko R, McCall K, [64] Settembre C, Arteaga-Solis E, Ballabio A, Karsenty G. Self-eating in Yang-Zhou D, Flockhart I, Binari R, Shim HS, et al. The Transgenic skeletal development: implications for lysosomal storage disorders. RNAi Project at Harvard Medical School: Resources and Validation. Autophagy 2009; 5:228-9; PMID:19029806; https://doi.org/10.4161/ Genetics 2015; 201:843-52; PMID:26320097; https://doi.org/10.1534/ auto.5.2.7390 genetics.115.180208 [65] Annunziata I, Bouche V, Lombardi A, Settembre C, Ballabio A. Mul- [75] Juhasz G, Hill JH, Yan Y, Sass M, Baehrecke EH, Backer JM, Neufeld tiple sulfatase deficiency is due to hypomorphic mutations of the TP. The class III PI(3)K Vps34 promotes autophagy and endocytosis SUMF1 gene. Hum Mutat 2007; 28:928; PMID:17657823; https:// but not TOR signaling in Drosophila. J Cell Biol 2008; 181:655-66; doi.org/10.1002/humu.9504 PMID:18474623; https://doi.org/10.1083/jcb.200712051 [66] Abid MR, Guo S, Minami T, Spokes KC, Ueki K, Skurk C, Walsh K, [76] Chang YY, Neufeld TP. An Atg1/Atg13 complex with multiple roles Aird WC. Vascular endothelial growth factor activates PI3K/Akt/ in TOR-mediated autophagy regulation. Mol Biol Cell 2009; forkhead signaling in endothelial cells. Arterioscler Thromb Vasc 20:2004-14; PMID:19225150; https://doi.org/10.1091/mbc.E08-12- Biol 2004; 24:294-300; PMID:14656735; https://doi.org/10.1161/01. 1250 ATV.0000110502.10593.06 [77] Emery P. Protein extraction from Drosophila heads. Methods Mol [67] Dey JH, Bianchi F, Voshol J, Bonenfant D, Oakeley EJ, Hynes NE. Biol 2007; 362:375-7; PMID:17417024 Targeting fibroblast growth factor receptors blocks PI3K/AKT sig- [78] Cumming RC, Simonsen A, Finley KD. Quantitative analysis of auto- naling, induces apoptosis, and impairs mammary tumor outgrowth phagic activity in Drosophila neural tissues by measuring the turn- and metastasis. Cancer Res 2010; 70:4151-62; PMID:20460524; over rates of pathway substrates. Methods Enzymol 2008; 451:639- https://doi.org/10.1158/0008-5472.CAN-09-4479 51; PMID:19185743 [68] Berry DL, Baehrecke EH. Growth arrest and autophagy are required [79] Batteiger B, Newhall WJt, Jones RB. The use of Tween 20 as a for salivary gland cell degradation in Drosophila. Cell 2007; 131:1137- blocking agent in the immunological detection of proteins trans- 48; PMID:18083103; https://doi.org/10.1016/j.cell.2007.10.048 ferred to nitrocellulose membranes. J Immunol Methods 1982; [69] Rusten TE, Lindmo K, Juhasz G, Sass M, Seglen PO, Brech A, 55:297-307; PMID:6820029; https://doi.org/10.1016/0022-1759 Stenmark H. Programmed autophagy in the Drosophila fat body is (82)90089-8

ACADEMIC VITA ‒ CLAIRE ELIZABETH REYNOLDS

Education: 2008-2011: B.S. Agricultural Biotechnology, State University of New York at Cobleskill, GPA 3.98 2011-2017: Doctoral Candidate in the department of Biochemistry, Microbiology, and Molecular Biology, The Pennsylvania State University, GPA 3.84 Awards: 2016: Outstanding Poster Award, Proteoglycans Gordon Research Conference 2012: Paul M Althouse Outstanding Teaching Assistant Award, PSU 2011: Roberts Incentive Fellowship for outstanding applicant, PSU Chancellor’s Award for Student Excellence, State University of New York

Presentations:

2017: 36th Annual summer symposium in Molecular Biology: Metabolism and Metabolomics: Disease Models and Model Organisms, University Park, PA. (poster)

2016: Proteoglycans Gordon Research Conference, Andover NH (poster and talk)

2013: Epistasis Discovery in Genetic Epidemiology Conference, Key West FL (talk)

Publications:

C. E. Reynolds-Peterson, N. Zhao, J. Xu, T. M. Serman, J. Xu, and S. B. Selleck. (2017) Heparan Sulfate Proteoglycans Regulate Autophagy in Drosophila. Autophagy 13: 1262-1279

J.M. O’Hara, J.C. Kasten-Jolly, C.E. Reynolds, and N.J. Mantis. (2014) Localization of non-linear neutralizing B cell epitopes on ricin toxin’s enzymatic subunit (RTA). Immunology Letters 158: 7-13