Mapping Genetic Interactions for Rare Disease- Associated

by

Kristin Kantautas

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Molecular Genetics University of Toronto

© Copyright by Kristin Kantautas 2020

Mapping Genetic Interactions for Rare Disease-Associated Genes

Kristin Kantautas

Doctor of Philosophy

Molecular Genetics University of Toronto

2020 Abstract

Over 7,000 rare diseases have been identified in the human population that, in aggregate, affect over 400 million people worldwide. The majority of rare diseases that have been identified are genetic in origin. However, a key challenge in developing therapies is that the functions of many rare disease-associated genes are poorly understood and consequently, their role in disease pathogenesis. There is strong appreciation that genes do not act in isolation and many disease phenotypes may arise from complex genetic interactions (GIs). Systematic mapping of GIs offers a wealth of information about function and advancements in gene-editing technology, such as clustered regularly interspaced short palindromic repeats (CRISPR), allows for GI mapping in human cells.

In this thesis, I describe how lentiviral-based genome-wide pooled CRISPR/Cas9 screens offer insight into the function of rare disease-associated genes including NGLY1 and the genes encoding the complement regulatory proteins (CRPs): CD46, CD55, and CD59. Autosomal recessive mutations in NGLY1 give rise to NGLY1-deficiency, a congenital disorder of deglycosylation. How the loss of NGLY1 function contributes to disease pathogenesis is poorly understood. The CRPs are key inhibitors of the complement cascade and their impaired function contributes to a wide range of common and rare diseases. Similar to NGLY1, the functions of the

CRPs independent of complement-regulation and roles in disease have not been explored using

ii functional genomics approaches. To further our functional understanding of NGLY1 and the genes encoding the CRPs, I performed pooled genome-wide CRISPR/Cas9 screens to identify

GIs, and uncovered a novel genetic relationship between the CRPs and the NGLY1 pathway involving genes in the secretory pathway that may coordinate different aspects of intracellular cholesterol trafficking.

My work demonstrates the utility in mapping GIs in order to reveal functional relationships with rare disease genes, offering critical insight into the genetic underpinnings of NGLY1-deficiency and complement-independent functions of the CRPs. As the functions of many rare disease- associated genes are poorly characterized, mapping genetic interactions in co-isogenic human cell lines using genome-wide CRISPR screening approaches offers a powerful framework that may be exploited to gain insight into gene function and direct future therapeutic efforts.

iii Acknowledgments

I would like to thank my supervisor Jason Moffat for the opportunity to pursue a project that I am truly grateful to have been a part of. Thank you for your guidance and for giving me the opportunity to share my research at conferences where I have both learned a lot and have been inspired by incredible scientists.

I am very grateful to my committee members, James Dennis and James Rini, for their support and guidance throughout the past six years, especially while writing my thesis. Thank you for always challenging me to consider alternative hypotheses, your insight and advice have truly helped shape my critical thinking skills.

Thank you to all members of the Moffat and Sidhu labs, past and present, for their support throughout my time in the lab. I would especially like to thank Ashwin Seetharaman. You have been an incredible mentor, collaborator and friend. I truly could not imagine this journey without your support and friendship. Thank you for always making science fun, for morning gratitudes in our office, and for encouraging me to believe in myself.

Thank you to my incredible friends who have been there for me since the beginning, especially Geeta Chopra, Zoë Abreder and Natasha Pascoe for their friendship, perspective and support. I am so lucky to share the ups and downs in life with all of you.

To my incredibly loving and supportive parents, words cannot express my gratitude. You have sacrificed so much so that I could pursue my passions and you both have been there to encourage me every step of the way. Thank you for always believing in me and for reminding me of the light at the end of this challenging journey.

Finally, to my husband Isaac, thank you. This accomplishment would not have been possible without your love, support and encouragement. Thank you for all the laughter, surprise trips, cool outfits and for always ensuring there was a nice bottle of wine waiting for me at home.

iv Table of Contents

ACKNOWLEDGMENTS ...... IV

TABLE OF CONTENTS ...... V

LIST OF TABLES ...... VIII

LIST OF FIGURES ...... IX

LIST OF ABBREVIATIONS...... XI

CHAPTER 1 ...... 1

GENERAL INTRODUCTION ...... 1

1.1 RARE DISEASES ...... 1 1.1.1 RARE DISEASE CATEGORIES ...... 2 1.2 GLYCOSYLATION ...... 4 1.2.1 MAIN TYPES OF PROTEIN GLYCOSYLATION IN HUMANS ...... 5 1.2.2 SYNTHESIS OF N- ...... 8 1.2.3 CATABOLISM OF N-GLYCOPROTEINS ...... 8 1.2.4 DISORDERS OF N- METABOLISM ...... 11 1.3 PEPTIDE:N-GLYCANASE I (PNGASE; NGLY1) ...... 14 1.3.1 NGLY1-DEFICIENCY...... 14 1.3.2 PNGASE ENZYMATIC PROPERTIES ...... 17 1.3.3 NGLY1 GENOMIC AND PROTEIN INFORMATION ...... 18 1.3.4 BIOLOGICAL FUNCTIONS OF NGLY1 ...... 20 1.4 THE COMPLEMENT SYSTEM ...... 33 1.4.1 THE COMPLEMENT SYSTEM ...... 33 1.4.2 THE COMPLEMENT CASCADE ...... 35 1.4.3 FUNCTIONS OF COMPLEMENT ...... 38 1.5 COMPLEMENT REGULATORY PROTEINS ...... 40 1.5.1 STRUCTURE, EXPRESSION AND INHIBITORY FUNCTIONS ...... 41 1.5.2 FUNCTIONS BEYOND COMPLEMENT REGULATION...... 44 1.5.3 COMPLEMENT REGULATORY PROTEINS IN DISEASE ...... 48

v 1.6 GENETIC INTERACTIONS ...... 53 1.6.1 DEFINING GENETIC INTERACTIONS ...... 53 1.6.2 MAPPING GENETIC INTERACTIONS ...... 54 1.6.3 GENETIC INTERACTIONS AND IMPLICATIONS IN DISEASE PHENOTYPES ...... 57 1.6.4 CONCLUDING STATEMENT ...... 58 1.7 OVERVIEW OF DOCTORAL RESEARCH PROJECT ...... 59

CHAPTER 2 ...... 60

GENOME-WIDE GENETIC INTERACTION CRISPR SCREENS OFFER NEW INSIGHTS INTO THE GENETIC DEPENDENCIES UNDERLYING NGLY1-DEFICIENCY ...... 60

2.1 ABSTRACT ...... 61 2.2 INTRODUCTION ...... 62 2.3 RESULTS ...... 64 2.3.1 AN UNBIASED GENOME-WIDE GENETIC SCREEN FOR SENSITIZERS OF BORTEZOMIB HELPS DEFINE AN NGLY1 GENETIC PATHWAY IN HUMAN CELLS ...... 64 2.3.2 MAPPING GLOBAL NGLY1 GIS IN HUMAN CELLS ...... 67 2.3.3 NGLY1, NFE2L1, AND DDI2 SHARE NEGATIVE GIS WITH THE GARP COMPLEX ...... 73 2.3.4 NUCLEAR IMPORT OF CLEAVED-NFE2L1 REQUIRES THE ACTIVITY OF NGLY1 ...... 76 2.3.5 IDENTIFYING A SHARED TRANSCRIPTIONAL SIGNATURE IN NGLY1Δ, NFE2L1Δ AND DDI2Δ CELLS 77 2.3.6 LOSS OF NGLY1, NFE2L1, OR DDI2 RESULTS IN DYSREGULATED EARLY ENDOSOME AND CHOLESTEROL TRAFFICKING ...... 84 2.4 DISCUSSION ...... 87

CHAPTER 3 ...... 93

GENOME-WIDE CRISPR SCREEN IDENTIFIES COMPLEMENT-INDEPENDENT ROLES FOR THE COMPLEMENT REGULATORY PROTEINS ...... 93

3.1 ABSTRACT ...... 94 3.2 INTRODUCTION ...... 95 3.3 RESULTS ...... 97 3.3.1 GENERATION OF CLONAL HAP1 CELL LINES LACKING SURFACE CRPS...... 97 3.3.2 GENOME-WIDE CRISPR SCREEN OF CRP-KO CELLS IDENTIFIES SENSITIZERS AND SUPPRESSORS OF COMPLEMENT IN THE ABSENCE OF CRPS ...... 97 3.3.4 MEMBERS OF THE NGLY1 GENETIC PATHWAY SHARE STRONG NEGATIVE GIS WITH THE CRPS 103

vi 3.3.5 LOSS OF THE CRPS PERTURBS NFE2L1 PROCESSING AND TURNOVER DURING CHOLESTEROL- INDUCED STRESS ...... 105 3.3.6 HUMAN CELLS LACKING THE CRPS DISPLAY INCREASED NUMBERS OF EARLY ENDOSOMES AND CHOLESTEROL ACCUMULATION ...... 105 3.4 DISCUSSION ...... 107

CHAPTER 4 ...... 114

METHODS ...... 114

4.1 CELL CULTURE ...... 114 4.2 CLONAL KNOCKOUT CELL LINES ...... 114 4.2.1 NGLY1, NFE2L1 AND DDI2 ...... 114 4.2.2 GENERATION OF CRP KNOCKOUT CELL LINES ...... 114 4.3 FLOW CYTOMETRY ...... 115 4.4 COMPLEMENT-MEDIATED LYSIS ...... 115 4.5 GENOME-WIDE GENETIC INTERACTION CRISPR SCREENS ...... 115 4.6 CRISPR/CAS9-MEDIATED GENE TARGETING ...... 116 4.7 GENE SET ENRICHMENT ANALYSIS ...... 116 4.8 RNA SEQUENCING ...... 116 4.9 IMMUNOBLOTS ...... 117 4.10 CHOLESTEROL SOLUBILIZATION ...... 117 4.11 SUBCELLULAR FRACTIONATION ...... 118 4.12 CONFOCAL MICROSCOPY ...... 118 4.12.1 CHOLESTEROL STAINING FOR CONFOCAL MICROSCOPY ...... 118

CHAPTER 5 ...... 120

DISCUSSION ...... 120

REFERENCES ...... 132

vii List of Tables

Table 1.1: Summary of NGLY1-deficiency models...... 21

Table 2.1 Top 25 predicted negative genetic interactions of bortezomib ...... 70

Table 2.2 Differentially expressed genes across NGLY1, NFE2L1 and DDI2 knockout cells (FDR < 0.5) identified by GSEA. Enriched gene sets for NGLY1 analysis...... 81

viii List of Figures

Figure 1.1: Distribution of FDA approved drugs by rare disease therapeutic area...... 3

Figure 1.2: Classification of N-...... 7

Figure 1.3: Protein N-glycosylation and quality control of protein folding in the endoplasmic reticulum...... 9

Figure 1.4: Degradation pathway for high-mannose type N-glycans on N-glycoproteins in mammals...... 12

Figure 1.5: Schematic representation of the domain structure of PNGase (NGLY1) from various organisms and mutations identified in NGLY1-deficiency...... 16

Figure 1.6: Proposed NGLY1 genetic pathway and proteasome bounce-back response...... 27

Figure 1.7: Systemic complement activation...... 36

Figure 1.8: Schematic of the structure of complement regulatory proteins...... 43

Figure 1.9: CD46 in T cell homeostasis and effector function...... 46

Figure 1.10: Graphical representation of quantitative genetic interactions...... 55

Figure 2.1: Genome-wide CRISPR screen schematic to identify genetic pathways that mediate sensitivity and resistance to bortezomib...... 66

Figure 2.2: Genome-wide CRISPR screen identifies members of the NGLY1 genetic pathway as sensitizers to bortezomib...... 69

Figure 2.3: Analysis of genome-wide CRISPR/Cas9 screening for genetic interactions of NGLY1...... 72

Figure 2.4: Mapping genetic interaction partners of NGLY1, NFE2L1 and DDI2 in human HAP1 cells...... 75

Figure 2.5: NFE2L1 nuclear import is impaired in the absence of NGLY1 and DDI2...... 78

ix Figure 2.6: Gene Set Enrichment Analysis of RNA sequencing of NGLY1∆, NFE2L1∆, and DDI2∆ cells...... 80

Figure 2.7: The NGLY1 genetic pathway regulates cholesterol homeostasis and immune processes...... 83

Figure 2.8: Loss of NGLY1, NFE2L1 and DDI2 result in dysregulated intracellular vesicle trafficking...... 85

Figure 2.9: Loss of NGLY1, NFE2L1 and DDI2 results in cholesterol accumulation...... 86

Figure 2.10: Proposed model of how the NGLY1 pathway and the GARP complex may coordinate intracellular vesicle trafficking in human cells...... 90

Figure 3.1: Generation of human HAP1 cell lines lacking all surface CRPs...... 98

Figure 3.2: Genome-wide CRISPR/Cas9 screen predicts sensitizers and suppressors of complement in the absence of the complement regulatory proteins...... 100

Figure 3.3: Complement regulatory proteins have negative genetic interactions with NGLY1, NFE2L1 and DDI2...... 104

Figure 3.4: Loss of the complement regulatory proteins impairs NFE2L1 processing during a high-cholesterol challenge...... 106

Figure 3.5: Loss of the complement regulatory proteins results in intracellular vesicle trafficking defects...... 108

Figure 3.6: Loss of the complement regulatory proteins results in cholesterol accumulation.... 109

x List of Abbreviations

AAS: Aggregated agreement score aHUS: Atypical hemolytic uremic syndrome

AMD: Age-related macular degeneration

APC: Antigen presenting cell

ARE: Antioxidant response element

Asn: Asparagine

ATG: Autophagy-related protein

BMP: Bone morphogenic protein

C3aR: C3a receptor

C5aR: C5a receptor

CCP: Complement control protein

CDDG: Congenital disorder of deglycosylation

CDG: Congenital disorders of glycosylation

CEG: Core essential gene

COG: Conserved oligomeric Golgi

CR1: Complement receptor 1; CD35

CRISPR: Clustered regularly interspaced short palindromic repeats

CRP: Complement regulatory protein

DAF: Decay accelerating factor; CD55

DAMPs: Damage-associated molecular patterns

xi DDI2: DNA-damage inducible 1 homolog 2

ENGase: Endo-beta-N-acetylglucosaminidase

ER: Endoplasmic reticulum

ERAD: Endoplasmic reticulum associated degradation

Fuc: Fucose

GAG: Glycosaminoglycan

GalNAc: N-acetylglucosamine

GI: Genetic interaction

Glc: Glucose

GlcNAc: N-acetylglucosamine

GPI: Glycosylphosphatidylinositol gRNA: Guide ribonucleic acid

GO:

GWAS: Genome-wide association studies

HFrD: High fructose diet

IEMs: Inborn errors of metabolism

IFN: Interferon

KEGG: Kyoto Encyclopedia of Genes and Genomes

LFC: Log-fold change

LLO: Lipid-linked oligosaccharide

xii LOF: Loss-of-function

LSD: Lysosomal storage diseases

MAC: Membrane attack complex

Man: Mannose

MCP: Membrane cofactor protein; CD46

MBL: Mannose-binding lectin mtDNA: Mitochondrial deoxyribonucleic acid mtRNA: Mitochondrial ribonucleic acid

MS: Multiple sclerosis

MSigDB: Molecular Signatures Database

NES: Non-essential gene

NGS: Next-generation sequencing

NFE2L1: Nuclear factor erythroid derived 2-related factor 1

NFE2L2: Nuclear factor erythroid derived 2-related factor 2

NHEJ: Non-homologous end joining

OMIM: Online Mendelian Inheritance in Man

PAMPs: Pathogen-associated molecular patterns

PIG: phosphatidylinositol- biosynthesis class

PNGase: Peptide:N-glycanase; NGLY1

PNH: Paroxysmal nocturnal hemoglobinuria

xiii PRM: Pattern recognition molecule qGI: Quantitative genetic interaction

RA: Rheumatoid arthritis

RC: Respiratory chain

RCA: Regulators of complement activation

RNAi: RNA interference

RNP: Ribonucleic protein

SCR: short consensus repeat

Ser: Serine

SGA: Synthetic gene array sgRNA: Single-guide ribonucleic acid

SLE: Systemic lupus erythematosus

SNARE: Soluble NSF attachment protein receptor

StAR: Steroidogenic acute regulatory protein

START: StAR-related lipid transfer

TCR: T cell receptor

TKO: Toronto Knockout

Thr: Threonine

WES: Whole exome sequencing

Xyl: Xylose

xiv Chapter 1 General Introduction General Introduction 1.1 Rare Diseases

Rare diseases are classified as any disease that affects a small percentage of the human population, many of which result in life-threatening conditions. When accounting for the over 7,000 rare diseases that have been identified to date (Hmeljak and Justice 2019), roughly 350 million people worldwide are currently living with a rare disease (Liu et al. 2019). Approximately 50% of rare diseases affect children, 30% of whom will die before the age of 5 years (Liu et al. 2019). Despite the significant global impact of rare diseases on human health, research advances on rare diseases are lagging compared to common diseases such as cancer or heart disease. Consequently, many rare diseases are poorly understood and therapies are only available for less than 5% of them (Kaufmann, Pariser, and Austin 2018).

80% of diseases are thought to be genetic in origin, either monogenic or polygenic, while the remaining rare diseases may be rare infectious diseases, rare cancers or rare auto-immune diseases (Liu et al. 2019). The application of next-generation sequencing (NGS) technology has dramatically increased rare disease gene discovery (Boycott et al. 2013).A substantive number of rare diseases are a result of a mutation in a single gene (Boycott et al. 2017). Indeed, a recent analysis of the Online Mendelian Inheritance in Man (OMIM) data base reported that approximately 3,200 unique genes are associated with a 4,500 monogenic rare diseases (Boycott et al. 2017). Monogenic or “Mendelian” diseases result from alterations in a single gene and follow Mendelian patterns of inheritance; either autosomal dominant, autosomal recessive or X- linked (Tripathi 2017). Conversely, polygenic or complex diseases are multifactorial in nature; they arise due to the complex interactions between many genes and other factors and therefore do not obey the rule of Mendelian inheritance (Tripathi 2017). One factor that has hindered therapeutic development for rare diseases is an incomplete understanding of the function of most genes and the biological pathways that they are involved in. A deeper understanding of rare- disease gene function will offer valuable insight into disease pathogenesis and may enable the identification of potential targets for therapeutic intervention.

1 1.1.1 Rare Disease Categories

Rare genetic diseases may be classified into various categories, with some of the major classes including neurological diseases, developmental anomalies during embryogenesis, inborn errors of metabolism, eye diseases, bone diseases and immunological diseases based on Orphanet rare disease classification. However, this distribution is not reflected in the number of FDA approved therapies for each rare disease category (Figure 1.1a) (Tambuyzer et al. 2020). This observation may in part reflect challenges in targeting particular disease pathways or affected organs and/or lack of resources and interest to research these diseases.

One of the largest classes of rare diseases includes inborn errors of metabolism (IEMs) which are a heterogenous group of diseases resulting from a defective or transport protein in a metabolic pathway involving the break-down and/or storage of carbohydrates, fatty acids and proteins (Jeanmonod and Jeanmonod 2019). Within IEMs includes disorders that arise due to defective activity of that participate in glycosylation (congenital disorders of glycosylation; CDGs) and the breakdown of lipids or glycoproteins (lysosomal storage diseases; LSDs) (Agana et al. 2018). There are currently approved therapies for the majority of CDGs (Chang, He, and Lam 2018), while some LSDs can be treated with enzyme replacement therapy (Platt et al. 2017). Recently, the first congenital disorder of deglycosylation (CDDG) known as NGLY1-deficiency, was identified (Need et al. 2012). NGLY1-deficiency arises due to inherited mutations in the NGLY1 gene which encodes a de-N-glycosylating enzyme(Enns et al. 2014). While many research groups have made significant advances in unravelling the functions of NGLY1 since it was first identified as a rare-disease causing gene, how its loss-of-function (LOF) contributes to disease pathogenesis is poorly understood.

Rare diseases due to dysregulation of the complement system have garnered increasing interest over the last decade since the first complement-specific drug was approved for the treatment of the rare diseases paroxysmal nocturnal hemoglobinuria (PNH) and atypical hemolytic uremic syndrome (aHUS) (Harris 2018). The complement system is a key component of innate immunity, providing a first line of defense against invading pathogens, maintaining tissue homeostasis and bridging innate and adaptive immunity (Merle, Noe, et al. 2015). Rare diseases involving the complement system may arise due to mutations in a complement protein or regulators which may manifest as uncontrolled activation of complement resulting in tissue

2 a

Other (7%) Ophthalmic (2%) Pulmonary (2%) Gastroenterology (3%) Cardiovascular (3%) Antidotes and medical countermeasures (4%) Rheumatology and immunology (5%) Neurology (7%) Infectious disease (7%) Hematology (%) Metabolism and endocrinology (15%) Oncology (34%)

Figure 1.1: Distribution of FDA approved drugs by rare disease therapeutic area. a. FDA orphan drug approvals by rare disease therapeutic area at the end of 2018. n = 770 (Recreated from Tambuyzer et al. 2020)

3 damage, under-activation resulting in recurrent infections, or autoimmunity (Degn, Jensenius, and Thiel 2011; Vignesh et al. 2017; Daniel Ricklin, Reis, and Lambris 2016). Lastly, strong genetic associations of complement with common diseases such as age-related macular degeneration (Geerlings, de Jong, and den Hollander 2017), have illuminated the involvement of the complement system in various human diseases (Daniel Ricklin, Reis, and Lambris 2016; Liszewski et al. 2016).

The objective of my doctoral research has been to investigate the function of NGLY1, a rare disease gene that when mutated gives rise to NGLY1-deficiency, and the function of complement regulatory proteins (CRPs) encoded by CD46, CD55, and CD59, which are critical regulators of the complement cascade.

1.2 Glycosylation

Glycans (carbohydrates), together with nucleic acids, proteins, and lipids, are the basic components of cells (Ohtsubo and Marth 2006). Monosaccharides, also called simple sugars, are the basic structural units of glycans and are used in the enzymatic process of glycosylation which is the process by which glycans are covalently attached to proteins (Ajit and Sharon 2009). Glycosylation is one of the most common post-translational modifications of proteins and it can affect their structure, stability, trafficking, function and receptor recognition (Ohtsubo and Marth 2006).

Glycosylation produces a diverse repertoire of cellular glycans that contribute to many crucial biological processes in animal systems. Often glycans, or oligosaccharide chains, are attached to lipids or proteins (glycoconjugates) where they participate in diverse cellular processes (Moremen, Tiemeyer, and Nairn 2012; Reily et al. 2019). Glycans associated with cell surface receptors and proteins influence cell adhesion, modulate protein function, cell signaling and alter the dynamics of glycoprotein endocytosis (Moremen, Tiemeyer, and Nairn 2012; Reily et al. 2019). Glycan structures on newly synthesized glycoproteins are critical for numerous aspects of protein homeostasis including protein folding, providing ligands for chaperones, quality control surveillance in the endoplasmic reticulum (ER) and facilitating transit and targeting through the secretory pathway functions (Ohtsubo and Marth 2006). It is therefore not surprising that defects in glycan synthesis or glycoconjugate degradation underlie the pathogenesis of numerous human diseases such as CDGs and LSDs (Freeze 2006; Platt et al. 2017).

4 1.2.1 Main Types of Protein Glycosylation in Humans

Glycans can be covalently attached to amino acid side-chains of proteins to form glycoproteins. Glycosylation of proteins begins in the ER with much of the terminal processing occurring in the cis-, medial- and trans-Golgi compartments (Colley, Varki, and Kinoshita 2017). Protein glycosylation includes the addition of N-linked or O-linked glycans, phosphorylated glycans, glycosaminoglycans (GAGs), and glycosylphosphatidylinositol (GPI) anchors to peptide backbones, as well as C-mannosylation (Reily et al. 2019). Most glycans found on membrane and secreted proteins are attached by N- and O-linkages (Moremen, Tiemeyer, and Nairn 2012).

1.2.1.1 O-Glycosylation

O-glycosylation refers to the attachment of a glycan to the oxygen atom of amino acids with functional hydroxyl groups, which are most often serine (Ser) and threonine (Thr), but may also be hydroxylysine or tyrosine (Reily et al. 2019; Moremen, Tiemeyer, and Nairn 2012). This includes the attachment of mannose (Man), N-acetylglucosamine (GlcNAc), xylose (Xyl), fucose (Fuc), glucose (Glc) and N-acetylgalactosamine (GalNAc) (Moremen, Tiemeyer, and Nairn 2012). O-glycans can therefore be further subclassified based on the initial sugar attachment and the additional sugar structures that are added to the glycan (Reily et al. 2019). O-glycosylation of membrane bound and secreted proteins takes place in the Golgi after N-glycosylation and protein folding has occurred, with the exception of the addition of Man and GlcNAc residues which take place in the endoplasmic reticulum (ER) and cytosol/nucleus, respectively (Reily et al. 2019). O- glycosylation modifications to proteins play important roles in protein trafficking, localization, stability, cell-cell interactions and immunity (Wopereis et al. 2006).

The most common type of O-glycosylation includes the addition of GalcNAc to Ser/Thr, which is also referred to as a mucin-type O-glycan. The GalNAc residue can be further extended with galactose (Gal), GlcNAc or GalNAc to produce eight different core structures (Moremen, Tiemeyer, and Nairn 2012). These glycans are abundant on extracellular and secreted proteins, including mucins, where they serve as lubricants and protect against pathogens and self- recognition by the immune system (Reily et al. 2019).

O-GlcNAcylation, the attachment of a GlcNAc residue to Ser/Thr, is a unique type of O- glycosylation that is carried out by O-GlcNAc transferases in the cytosol and nucleus and is not extended by the attachment of additional residues (Reily et al. 2019). O-GlcNAcylation plays an

5 important role in the modification of the biological activity of intracellular proteins and is thought to regulate many cellular functions, including metabolism (Reily et al. 2019).

Glycans that are attached through Xyl to Ser are called GAGs which are long unbranched polysaccharides and include heparin, heparin sulphate, chondroitin sulphate and dermatan sulphate (Freeze 2006). Proteins that contain GAGs chains are called proteoglycans. GAGs are initiated by a conserved tetrasaccharide, followed by disaccharide repeats, which is the basis for their classification (Wopereis et al. 2006). GAGs may be further modified by sulphation. Keratin sulfate is another type of GAG which can be O-linked via GalNAc Ser/Thr and is extended by disaccharide repeats (Wopereis et al. 2006). GAGs perform a variety of biological functions, some of which include providing physical integrity for the cell membrane, facilitating cell adhesion, and binding growth factors (Freeze 2006).

1.2.1.2 N-glycosylation

As reviewed by Stanley et al. 2017, N-glycosylation refers to the attachment of an N-glycan to the nitrogen atom of an asparagine (Asn) side-chain at the consensus glycosylation motif of Asn- X-Ser/Thr, where X is any amino acid except for Pro. N-glycans are highly branched and heterogeneous structures. All eukaryotic N-glycans share a common core structure consisting of two GlcNac and 3 Man residues. Monosaccharides such as Man, GlcNAc, Gal, N- acetylneuraminic acid and Fuc, may be added to the core structure creating extended and sometimes more branched structures. N-glycans are therefore classified into three types: high- mannose glycans, complex glycans, or hybrid N-glycans based on the type of monosaccharides that extend the core (Figure 1.2). High-mannose N-glycans contain only mannose residues attached to the core. Hybrid N-glycans contain mannose residues on one branch while N- acetylglucosamine initiates the other branch. Complex N-glycans are those in which extending branches are initiated by GlcNAc.

Many proteins are modified by N-glycosylation and N-glycans have a crucial role in regulating many intracellular and extracellular functions (Moremen, Tiemeyer, and Nairn 2012). Furthermore, defects in N-glycosylation represent the majority of subtypes of CDGs (Chang, He, and Lam 2018).

6

Figure 1.2: Classification of N-Glycans. N-glycans added to a protein at Asn-X-Ser/Thr sequons may be classified as high-mannose, complex or hybrid. N- glycans contain a core of 3 mannose, 2 N-Acetylglucosamine and asparagine (Man3GlcNAc2Asn). Shown is a bi- antennary complex N-glycan. Created with BioRender.

7 1.2.2 Synthesis of N-Glycoproteins

N-glycosylation of proteins involves the use of a lipid-linked oligosaccharide (LLO) as a donor substrate (Figure 1.3) (Stanley, Taniguchi, and Aebi 2017). The LLO contains 2 GlcNAc, 9 Man, and 3 Glc units and is constructed on a lipid carrier such as dolichol, by glycosyltransferases (Stanley, Taniguchi, and Aebi 2017). Once it has been synthesized, the glycan is transferred to an asparagine on a nascent polypeptide chain in the ER lumen during translation (Stanley, Taniguchi, and Aebi 2017). Following the transfer of the glycan, two glucose residues are removed which allows for binding to the chaperones calnexin and calreticulin to help fold the protein (Helenius and Aebi 2004). Iterative cycles of removing and adding the glucose moiety and subsequent chaperone re-binding, help facilitate folding of newly synthesized glycoproteins in the ER (Helenius and Aebi 2004).

After leaving the ER, glycoproteins travel through the Golgi apparatus where the glycans are susceptible to further processing (Moremen, Tiemeyer, and Nairn 2012). Almost all glycoprotein glycans undergo trimming and extension as they traverse the Golgi (Moremen, Tiemeyer, and Nairn 2012). Glycan processing enzymes are distributed across the Golgi in a cis-to-trans topography and successive glycan maturation steps occur as glycoproteins traffic through the organelle (Moremen, Tiemeyer, and Nairn 2012). The transitional ER and cis-Golgi compartments contain mannosidases that trim high-mannose N-glycans, and as the glycoproteins move through the medial and trans-Golgi compartments, hybrid and complex N-glycans are produced through the addition of GlcNAc, galactose, sialic acid and fucose sugars (Moremen, Tiemeyer, and Nairn 2012).

1.2.3 Catabolism of N-Glycoproteins

Catabolism of N-glycoproteins is part of the normal turnover of cellular components (Winchester 2005). This process occurs in the lysosome where proteases break down the glycoprotein and glycosidases degrade N-glycans into monosaccharides that can be recycled in the cell (Winchester 2005). In addition to lysosomal degradation, cytosolic enzymes may degrade N- glycoproteins or their free N-glycan derivatives (Suzuki 2007). Both lysosomal and non- lysosomal degradation systems of N-glycoproteins are critical for normal development in humans (Suzuki 2016).

8

Figure 1.3: Protein N-glycosylation and quality control of protein folding in the endoplasmic reticulum. During glycoprotein synthesis, the nascent polypeptide enters the endoplasmic reticulum (ER) through the Sec61 translocon. Simultaneously, a 14-residue oligosaccharide is transferred from dolichol di-phosphate to the nascent chain at Asn-X-Ser/Thr by oligosaccharyltransferase (OST). The terminal two glucoses are removed by glucosidase I and II (Gls-I and Gls-II) sequentially. The monoglucosylated glycoprotein can then bind to lectins calnexin (CNX) or calreticulin (CRT) to facilitate its folding. Removal of the last glucose reside by Gls-II causes the glycoprotein to be released from CNX/CRT. Correctly folded glycoproteins proteins are packaged for transport to the Golgi and exit the ER. Incompletely folded glycoproteins are recognized by UDP-glucose:glycoprotein glycosyltransferase (UGGT). UGGT adds the glucose back onto the glycoprotein from a UDP-glucose donor. The-reglucosylated glycoprotein then re-enters the CNX/CRT cycle. A protein may cycle through the CNX/CRT cycle several times before it reaches its native conformation. Terminally misfolded proteins are subjected to ER-associated degradation (ERAD) by mannose trimming, assisted by EDEM an ER resident mannosidase. The misfolded protein is then retrotranslocated to the cytosol, deglycosylated by PNGase and degraded by the proteasome. LLO: lipid-linked oligosaccharide. Created with BioRender.

9 1.2.3.1 Autophagy and Lysosomal Degradation

The majority of both long and short lived proteins undergo lysosomal degradation (Suzuki 2016). N-glycoproteins that are destined for degradation are delivered to lysosomes either by endocytosis from the cell surface or extracellular spaces, or by autophagic processes for intracellular protein degradation (Winchester 2005).

As reviewed by Dikic and Elazar, 2018, autophagy occurs in response to different forms of stress such as nutrient deprivation, ER stress, oxidative stress and hypoxia in order to provide nutrients for essential cellular functions. However, it also eliminates unwanted and potentially harmful cytosolic material such as protein aggregates or damaged mitochondria (mitophagy). Autophagy is carried out by autophagy-related proteins (ATGs) which form complexes with other autophagic proteins which facilitate different steps of the autophagy process. Induction of autophagy results in the recruitment of ATGs to a specific cellular location and the formation of an isolation membrane that forms a phagophore. Elongation of the membrane leads to the expansion of the phagophore into a double-membraned vesicle, termed the autophagosome, which contains the sequestered cargo. As the autophagosome matures, it eventually fuses with the lysosome where it will be degraded (Dikic and Elazar 2018).

In the endocytic pathway, Rab GTPases which are expressed on the cytoplasmic side of various organelles and vesicles, regulate their targeting and fusion with the appropriate membranes (Elkin, Lakoduk, and Schmid 2016). Glycoproteins that are internalized by endocytosis are transported to the endosomes by endocytic vesicles (Elkin, Lakoduk, and Schmid 2016). Early endosomes mature into late endosomes which eventually fuse with the lysosome to degrade the contained protein (Elkin, Lakoduk, and Schmid 2016).

Regardless of how the glycoprotein is delivered to the lysosome, the catabolism of the glycoprotein is catalyzed by hydrolases in the low pH environment (Winchester 2005). Various proteases degrade the glycoprotein and the N-glycans are broken down by glycosidases (Winchester 2005). Proteases within the lysosome include a mixture of endopeptidases and exopeptidases, which degrade the protein into amino acids and in some cases, dipeptides (Winchester 2005). The amino acids/dipeptides are transported across the lysosomal membrane into the cytosol by a combination of diffusion and carrier-mediated transport (Winchester 2005). The catabolic pathways for all types of N-glycans have been established where the sequential

10 removal of monosaccharides by various glycosidases follows a defined process (Suzuki 2016). After glycans are degraded into monomeric sugars, they are transported across the lysosomal membrane into the cytosol where they can be reused by the cell (Winchester 2005).

1.2.3.2 ERAD and Non-Lysosomal Degradation

N-glycoproteins synthesized in the lumen of the ER are monitored by a surveillance system called “ER quality control” (Lamriben et al. 2016). Glycoproteins that fold too slowly due to mutations or incomplete assembly of subunits are retro-translocated into the cytosol for proteasomal degradation by ER-associated degradation (ERAD)(Helenius and Aebi 2004). Once a glycoprotein reaches the cytosol, its N-glycans are removed by cytosolic peptide:N-glycanase (PNGase, which is NGLY1 in humans), which hydrolyzes the amide bond between the innermost GlcNAc on the N-glycan and the Asn residue on the glycoprotein (Figure 1.4) (Hirayama, Hosomi, and Suzuki 2015; Suzuki, Huang, and Fujihira 2016). The free N-glycans are subsequently processed by two cytosolic enzymes, endo- β-N -acetylglucosaminidase (ENGASE) and -mannosidase (MAN2C1) and the final glycan structure is transported to the lysosome for further degradation (Suzuki 2007).

1.2.3.3 ERAD and Lysosomal Degradation

Some misfolded proteins in the ER cannot be degraded by ERAD and are instead delivered to the endo-lysosome by autophagic processes for clearance. This includes proteins that form aggregates or polymers in the ER, do not engage ERAD factors, are too large or display properties that prevent dislocation across the ER membrane (Fregno and Molinari 2018).These ERAD-resistant proteins recruit autophagic proteins to the ER, by mechanisms that have yet to be defined, and are delivered to the lysosome as cargo in an autophagosome where they are then degraded (Fregno and Molinari 2018).

1.2.4 Disorders of N-Glycoprotein Metabolism

Mutations in genes involved in N-glycoprotein metabolism give rise to numerous rare genetic disorders including congenital disorders of glycosylation (CDGs), lysosomal storage diseases (LSDs), and NGLY1-deficiency.

11

Figure 1.4: Degradation pathway for high-mannose type N-glycans on N-glycoproteins in mammals. N-glycans are removed from misfolded glycoproteins in the cytosol by NGLY1 and are sequentially processed by ENGASE and MAN2C1. The final glycan structure is transported to the lysosome for further degradation.

12 1.2.4.1 Congenital Disorders of Glycosylation

CDGs are a group of rare genetic and metabolic disorders that arise from defects in glycosylation (Grünewald, Matthijs, and Jaeken 2002). To date, over 130 different genes have been determined to cause CDGs (Chang, He, and Lam 2018). CDGs are classified into different groups based on the synthetic pathway affected and include disorders in N-glycosylation, O-glycosylation, combined N- and O-linked glycosylation, and lipid and (GPI) anchor biosynthesis defects (Chang, He, and Lam 2018). CDGs are also often divided into either type I (early) or type II (late) glycosylation deficiencies (Reily et al. 2019). Type I CDGs (CDG-I) are caused by abnormalities in the formation of the oligosaccharide structure on the LLO prior to the attachment to the Asn residue of a protein and include disorders of N-glycosylation. Type II CDGs (CDG-II) involve defects in the N-linked branching structure on the nascent glycoprotein (Reily et al. 2019).

Forty-two different enzymes in the N-linked oligosaccharide synthetic pathway are currently recognized to give rise to CDGs (Sparks and Krasnewich 2017). The vast majority of CDGs are autosomal recessive and are often embryonic lethal, however when they do manifest, they are typically severe and multisystemic as they affect many muscular, developmental and neurological functions (Chang, He, and Lam 2018). The severity and phenotypic diversity of CDGs clearly highlight the vital role of glycans and glycosylation in normal development and cellular processes (Reily et al. 2019).

The role of N-glycans in normal development, the severity of N-glycosylation CDGs as well as disease phenotypes, is also highlighted in mouse knockout studies (Thiel and Körner 2011). The International Mouse Phenotyping Consortium is a continuously growing catalog of documented phenotypes in mice associated with a given gene knockout and includes many genes known to give rise to CDGs (International Mouse Phenotyping Consortium). Of the knockout mice that have been documented by the Consortium, the majority with homozygous knockouts in genes known to give rise to N-glycosylation CDGs are embryonic lethal. While this observation highlights the role of N-glycans in normal development in mice, it poses a significant challenge in using mouse models to gain insight into the pathogenesis of N-glycosylation CDGs (Thiel and Körner 2011). Therefore, mouse models with hypomorphic alleles and conditional knockouts of genes known to give rise to N-glycosylation CDGs will provide further insight into the

13 consequence of impaired N-glycosylation in both embryonic and post-natal development (Thiel and Körner 2011).

1.2.4.2 Disorders of N-Glycan/N-Glycoprotein Degradation

Defects in both lysosomal and non-lysosomal degradation of N-glycoproteins give rise to rare human diseases, highlighting the consequences of disrupting these biological processes. LSDs are a group of over rare 70 diseases that are characterized by lysosomal dysfunction (Platt et al. 2017). LSDs are genetically and clinically heterogeneous disorders but frequently present as pediatric neurodegenerative diseases (Platt et al. 2017). These disorders are caused by mutations in genes encoding various lysosomal proteins including those involved in the degradation of N- glycoproteins (Platt et al. 2017). Mutations in lysosomal genes result in the gradual accumulation of substrates inside the lysosome which leads to cell dysfunction and cell death (Platt et al. 2017). Recently, the first congenital disorder of deglycosylation, NGLY1-deficiency, has been identified (Need et al. 2012) and will be discussed in greater detail below. NGLY1-deficiency is a rare multi-systemic genetic disorder where mutations in the NGLY1 gene are thought to lead to the accumulation of misfolded proteins in certain tissues of the body (Suzuki, Huang, and Fujihira 2016). Taken together, the controlled degradation of N-glycans and N-glycoproteins is critical for cellular homeostasis. Disrupting these pathways results in serious biological consequences that give rise to clinically and phenotypically diverse rare diseases.

1.3 Peptide:N-glycanase I (PNGase; NGLY1) 1.3.1 NGLY1-deficiency

NGLY1 encodes the cytoplasmic peptide: N-glycanase I (PNGase, NGLY1 in humans, Ngly1 in mice) and functions as an enzyme responsible for the deglycosylation of misfolded N- glycosylated proteins in the cytoplasm prior to their proteasome-mediated degradation (Suzuki et al. 1993). NGLY1 catalyzes the cleavage of the amide bond between the proximal GlcNAc residue of glycans and the asparagine residue of a glycoprotein (Suzuki et al. 1993). The process by which misfolded N-glycoproteins undergo proteasomal degradation is termed ER-associated degradation (ERAD) and is an essential quality control mechanism for glycoproteins that are produced in the ER (Chakrabarti, Chen, and Varner 2011). Disruption of ERAD has been associated with various neurological diseases including amyotrophic lateral sclerosis and Parkinson’s disease (Chakrabarti, Chen, and Varner 2011) and may result in an abnormal

14 accumulation of misfolded glycoproteins (Rao and Bredesen 2004). In 2012, a rare autosomal recessive human genetic disorder involving mutations in the NGLY1 gene was identified through an exome analysis (Need et al. 2012). NGLY1-deficiency is the first known congenital disorder of deglycosylation (CDDG) and almost 70 patients have been identified to date worldwide (NGLY1.Org). NGLY1-deficiency is a multi-systemic neurodevelopment disorder and patients most commonly exhibit symptoms which include global developmental delay, alacrima (the absence of tears), complex movement disorder, hypotonia (poor muscle tone), elevated liver transaminases and approximately half of them develop clinical seizures (Lam et al. 2018). Most patients with NGLY1-deficiency survive into early adulthood, however, death at younger ages has been reported due to severe liver diseases, infections and/or seizure complications (Lam et al. 2018). There is currently no standardized treatment or cure for NGLY1-deficiency.

The diverse and debilitating symptoms of NGLY1-deficiency highlight the functional importance of NGLY1 as the only de-N-glycosylating enzyme in the cell. Although significant advances have been made towards understanding NGLY1 function since it was first identified as a rare disease-causing gene, the role of NGLY1 in disease pathogenesis has not been fully resolved.

1.3.1.1 Mutations in NGLY1-deficiency

A number of heterozygous and homozygous predicted LOF variants in NGLY1 have been reported including nonsense, missense, and frameshift mutations (Figure 1.5) (Enns et al. 2014). These mutations occur throughout the gene with no obvious hotspots; the nonsense mutation R401X is the most commonly detected allele and observed in approximately one third of patients (Enns et al. 2014; Lam et al. 2017). Affected individuals harboring at least one copy of this pathogenic variant tend to have a more severe clinical phenotypes compared to patients bearing other mutations (Enns et al. 2014; Lam et al. 2017). Additionally, phenotypic heterogeneity exists between patients with different mutations, as well as with patients who harbor identical mutations, further highlighting the complexity of NGLY1-deficiency (Caglayan et al. 2015).

1.3.1.2 Biomarkers in NGLY1-deficiency

The majority of NGLY1-deficiency patients have been diagnosed through whole-exome sequencing (WES) or whole-genome sequencing (WGS) as identification of the disease has been

15 Figure 1.5: Schematic representation of the domain structure of PNGase (NGLY1) from various organisms and mutations identified in NGLY1-deficiency. PUB, peptide:N-glycanase ubiquitin association domain (protein–protein interaction domain); TG, domain (catalytic domain); PAW, present in PNGase and other worm proteins domain (mannose-binding domain);TRX, thioredoxin domain. Pink line is a putative zinc-binding motif. Mutations found in NGLY1-deficiency patients are indicated on human NGLY1 (top) (as identified in Lam et al. 2017). R401X (red) is the most common mutation. Created with BioRender. (Figure adapted from Hirayama, Hosomi & Suzuki, 2015).

16 challenging owing to the clinical heterogeneity of phenotypes, broad differential diagnosis and a lack of diagnostic biomarkers (Haijes et al. 2019). Diagnostic biomarkers for NGLY1-deficiency would be valuable given some of the challenges associated with diagnosis through NGS which include both costs and diagnostic delays (Haijes et al. 2019).

A potential urinary oligosaccharide biomarker Neu5Ac1Hex1GlcNAc1-Asn (Asn-N) was previously identified in NGLY1-deficiency patient urine samples using MALDI-TOF mass spectrometry (Hall et al. 2018). However, the same species has been observed in profiles of individuals with aspartylglucosaminuria, an inherited lysosomal storage disorder, and the relative amounts of Asn-N can vary over time suggesting that this metabolite is intermittently excreted (Hall et al. 2018).

Recently, aspartylglycosamine was identified as the first potential biomarker that can be detected in dried blood spots for NGLY1-deficiency. The detection of elevated levels of this species can be explained by the lack of NGLY1 function as NGLY1-deficiency results in a single GlcNAc residue left attached to the asparagine residue of a protein, which results in a free aspartylglycosamine when the glycoprotein is degraded. This potential biomarker may enable biochemical diagnosis for NGLY1-deficiency rather than diagnosis from WES (Haijes et al. 2019).

1.3.2 PNGase Enzymatic Properties

The gene encoding the cytoplasmic PNGase was first identified in the budding yeast S. cerevisiae (referred to as PNG1) (Suzuki et al. 2000). Prior to the discovery of the gene, the activity of cytoplasmic PNGase was reported in several mammalian-derived cultured cells (Suzuki et al. 1993). The discovery of PNG1 occurred upon screening a collection of mutagenized strains for the absence of PNGase activity in cell extracts. However, since PNG1 is not essential in yeast, the png1-mutant did not display any significant phenotypes (Suzuki et al. 2000). Subsequently, gene orthologues of PNG1 were found in a wide variety of eukaryotes including mammals (Suzuki et al. 2000; Suzuki, Park, and Lennarz 2002). Biochemical and structural studies have provided insight into the enzymatic properties of PNGase and its domains, as well as how it varies between species.

17 NGLY1 removes intact N-glycans by a two-step reaction. First, N-linked glycans on the consensus sequence of a N-glycoprotein (Asn-X-Ser/Thr where X is any amino acid except proline) are hydrolyzed by NGLY1 at the amide bond between the innermost GlcNAc and the Asn residue (Huang et al. 2015). This reaction generates a de-N-glycosylated protein in which the Asn residue is converted to Asp, and a 1-amino-GlcNAc containing free oligosaccharide (Huang et al. 2015). In the subsequent step, ammonia is spontaneous released from the reducing end, resulting in the production of a free oligosaccharide that is recycled in the cell (Huang et al. 2015).

PNGase preferentially deglycosylates proteins that contain high-mannose oligosaccharides over those bearing complex-type oligosaccharides (Hirsch et al. 2003) and proteins in their non-native state (Hirsch et al. 2004). NGLY1 enzymatic activity differs from other PNGases, such as PNGase A in plants or PNGase F in bacteria, in its requirements for reducing reagent activity, a neutral PH for optimal activity and an intrinsic carbohydrate-binding property ( Kitajima et al. 1995; Suzuki et al. 1993; Suzuki et al. 1994).

1.3.3 NGLY1 Genomic and Protein Information

The human NGLY1 gene locus is located on 3, has 12 exons and encodes a 654 amino acid protein (Suzuki, Huang, and Fujihira 2016). NGLY1 is comprised of three domains that are present in higher eukaryotes: the central transglutaminase-like catalytic domain (TG), an N-terminal PUB domain (peptide:N-glycanase ubiquitin association) and a C-terminal PAW domain (present in PNGases and other worm proteins) (Figure 1.5) (Suzuki, Huang, and Fujihira 2016).

Two regions that are largely conserved among PNGases include the TG domain and a pair of CXXC motifs which are thought to be involved in zinc binding (Lee et al. 2005). Structural and biochemical studies have elucidated the function of the three domains in PNGase, offering insight into its enzymatic activity, as well as its role in recognizing N-glycans and facilitating the degradation of proteins by ERAD. To date, the crystal structure of yeast, mouse (all domains) and human (PUB domain) PNGase have been determined (Allen, Buchberger, and Bycroft 2006; Lee et al. 2005; Zhao et al. 2006b, 2007, 2009; Zhou et al. 2006).

18 1.3.3.1 Transglutaminase Domain

The TG domain contains a conserved catalytic triad of residues composed of Cys, His, and Asp which are essential for its deglycosylation activity (Katiyar et al. 2002; Suzuki et al. 2006; Zhao et al. 2009). Yeast, mouse, and human PNGase associates with the proteasome through binding of the TG domain to HR23 (the mammalian homologue of Rad23 in yeast) (Katiyar, Li, and Lennarz 2004; Lee et al. 2005; Kim et al. 2006). HR23 is an essential DNA repair protein and participates in both nucleotide excision repair and ubiquitin-mediated protein degradation (Kim et al. 2006). In the ubiquitin-proteasome pathway, HR23 acts as a proteasome shuttling factor by interacting with components of ubiquitin and proteasome (Hirayama, Hosomi, and Suzuki 2015). Therefore, the interaction between HR23 and PNGase serves to couple PNGase-mediated deglycosylation and degradation of misfolded glycoproteins by ERAD (Suzuki, Park, Kwofie, et al. 2001; Kim et al. 2006). Interestingly, a mutation in the Cys residues in the putative Zn- binding CXXC motifs found near the TG domain results in the loss of PNGase enzymatic activity in yeast (Katiyar et al. 2002).

1.3.3.2 PUB Domain

In higher eukaryotes, the PUB domain was identified by bioinformatic analysis (Suzuki, Park, Till, et al. 2001) and was found to interact with the cytosolic p97 (Allen, Buchberger, and Bycroft 2006). p97, also known as vasolin-containing protein (VCP), is a barrel-like AAA ATPase enzyme that functions in extracting membrane-bound proteins from the ER to the cytoplasm that are destined for ERAD (Woodman 2003). During ERAD, p97 is part of a protein degradation complex and associates with various cofactors to recruit ubiquitinated substrates (Zhao et al. 2007). In addition to its role in ERAD, p97 participates in other cellular processes such as DNA repair and fulfills its various functions by interacting with cofactors (Zhao et al. 2007). The interaction between p97 and PNGase occurs at the C terminus of p97 whereas most of its other cofactors interaction with its N-terminal domain (Zhao et al. 2007). Thus through its interaction between p97, the PUB domain of PNGase is predicted to be involved in the extraction of ubiquitinated-misfolded proteins from the ER to the cytosol for degradation during ERAD (Allen, Buchberger, and Bycroft 2006; Zhao et al. 2007).

19 1.3.3.3 Paw Domain

Structural and biochemical studies have demonstrated that the C-terminal PAW domain serves as a carbohydrate domain for high-mannose type glycans (Zhou et al. 2006). Through this interaction, the binding affinity between PNGase and its substrates may be increased (Zhou et al. 2006).

PNGases in some eukaryotes appear to have additional or alternative functions. Caenorhabditis elegans (C. elegans) has a thioredoxin domain (TRX domain) in its N-terminus, which exhibits oxidoreductase activity (Kato et al. 2007). Orthologues of PNGase in Neurospora crassa (N. crassa) show no PNGase activity due to the intrinsic mutations in two residues of the catalytic triad (Maerz et al. 2010). However, deletion of the N. crassa PNGase results in a temperature sensitive phenotype as well as an abnormal cell polarity, suggesting that its function is likely distinct from deglycosylation activity (Maerz et al. 2010). The Drosophila melanogaster (D. melanogaster) PNGase ortholog, PNGase-like (Pngl) retains carbohydrate-binding activity, but lacks conventional PNGase activity, suggesting that the deglycosylation activity may have been lost during the evolutionary process. Pngl has been shown to be critical for normal development in D. melanogaster, and thus may not lie within its deglycosylation activity (Funakoshi et al. 2010).

1.3.4 Biological Functions of NGLY1

Since its discovery in yeast, the phenotypes associated with PNGase activity have been investigated in various eukaryotes using loss-of-function (LOF) studies (Table 1.1). Although much progress has been made in clarifying the role of NGLY1 in ERAD, there is now increasing evidence that by deglycosylating substrates such as NFE2L1, and perhaps others that have not yet been identified, NGLY1 might orchestrate important roles in various biological processes. These include proteasome and mitochondrial homeostasis, inflammation, lipid metabolism, aquaporin regulation, antigen presentation, and developmental signaling pathways (Suzuki 2007; Suzuki, Huang, and Fujihira 2016; Altrich-VanLith et al. 2006; Galeone et al. 2017; Tomlin et al. 2017; Yang et al. 2018; Kong et al. 2018; Fujihira et al. 2019; Tambe, Ng, and Freeze 2019). Until recently, little was known about the function of NGLY1 in normal animal development, and several disease models of NGLY1-deficiency have been developed using model organisms (D. melanogaster, C. elegans, M. musculus) and cultured human cells, which have provided valuable

20 Table 1.1: Summary of NGLY1-deficiency models.

Model Gene Name Phenotype Molecular Mechanisms

S. cerevisiae PNG1 No observable phenotype1 N/A

C. elegans png-1 Reduced lifespan2 Impaired SKN1-A (NFE2L1) processing4,5 Neuronal branching defect3 Sequencing editing of SKN-1A via Abnormal mitochondrial PNG-1 mediated deglycosylation is physiology1 necessary4,5

Hypersensitive to proteasome Impaired proteasome bounce-back inhibition4 response4

D. melanogaster Pngl Global developmental Impaired BMP signaling via decreased Dpp homodimer delay6,7,8 formation9

Larval and adult lethality6,7,8 Defects in neuroendocrine axis6

7 Small body size6 No ER stress or abnormal ERAD

Impaired Cnc (NFE2L1) function7 Abnormal gut development9 Dysregulated expression of Hypersensitive to proteasome proteasome subunit genes inhibition (proteasome bounce-back) and proteins with oxidation-reduction functions7

M. musculus Ngly1 Abnormal mitochondrial Increased expression of interferon- physiology2 stimulated genes12 (and cultured cells) Sensitive to mitochondrial Delayed ERAD10 inhibition2 Impaired NFE2L1 processing12 Abnormal hepatocyte nuclear size/morphology14 Impaired proteasome bounce-back response12 Accumulation of N-GlcNAc aggregate proteins10 NFE2L1 regulates mitochondrial homeostasis (mitophagy-related Ventricular septal defect genes)12 (embryo)11

Embryonic lethal (C57BL/6 background)11

21 Partial rescue by Ngly1-/-; Engase-/- knockout11

Bent spine, hind-limb clasping, front-limb shaking reduced body weight (Ngly1-/- ; Engase-/- knockout)11

H. sapiens NGLY1 Abnormal mitochondrial Increased expression of interferon- physiology (cell lines and stimulated genes (patient cells and (cell lines) patient liver and muscle cell lines)12 cells)2 Impaired NFE2L1 processing13 Hypersensitive to proteasome inhibition13 Impaired proteasome bounce-back response13 Sensitive to mitochondrial inhibition2 NFE2L1 regulates mitochondrial homeostasis (mitophagy-related genes)12

1 Suzuki et al. 2000 2 Kong et al. 2018 3 Habibi-Babadi et al. 2010 4 Lehrbach & Ruvkun 2016 5 Lehrbach, Breen & Ruvkun 2019 6 Portillo et al. 2018 7 Owings et al. 2018 8 Funakoshi et al. 2010 9 Galeone et al. 2017 10 Huang et al. 2015 11 Fujihira et al. 2017 12 Yang et al. 2018 13 Tomlin et al. 2017 14 Fujihira et al. 2020

22 insights into the function of NGLY1 and disease pathogenesis. From these studies, two models of the pathogenesis of NGLY1-deficiency have been proposed; defects in cellular glycoprotein homeostasis and defects in glycosylation of specific glycoproteins.

1.3.4.1 Endoplasmic Reticulum-Associated Degradation

ER stress occurs when misfolded proteins accumulate in the ER lumen (Bernales, Papa, and Walter 2006). Through the ERAD pathway, terminally misfolded proteins are translocated from the ER lumen to the cytosol and are degraded by the proteasome (Bernales, Papa, and Walter 2006). Current evidence suggests that NGLY1 participates as a key component of ERAD by de- glycosylating misfolded proteins prior to their degradation in the cytosol (Park, Suzuki, and Lennarz 2001). Further, the PUB domain of NGLY1 is thought to be involved in the extraction of ubiquitinated and misfolded proteins that are retro-translocated from the ER to the cytosol for degradation during ERAD through its interactions with p97 and HR23 (Li et al. 2005; Kamiya et al. 2012). Together, NGLY1, HR23 and the 26S proteasome comprise a ternary complex (Li et al. 2005). It has been hypothesized that interaction between the ternary complex and retro- translocon, a protein complex in the ER membrane that facilitates retrotranslocation of misfolded proteins, via p97, may facilitate more efficient de-glycosylation and degradation of misfolded proteins (Hirayama, Hosomi, and Suzuki 2015).

Although it is clear that NGLY1 is involved in the ERAD process, the critical role of NGLY1- mediated deglycosylation in the degradation of ERAD substrates continues to be elucidated (Suzuki, Huang, and Fujihira 2016). It is unknown if NGLY1 is required to de-glycosylate all misfolded proteins or just a subset (Suzuki, Huang, and Fujihira 2016). Several ERAD substrates have been reported to be de-glycosylated by cytoplasmic PNGase during their degradation including the MHC class I heavy chain, the TCR α subunit, and nuclear factor erythroid derived 2-related factor (NRF1/NFE2L1) (Hughes, Hammond, and Cresswell 1997; Huppa and Ploegh 1997; Lehrbach and Ruvkun 2016). However, N-glycan removal from proteins is not a prerequisite for proteasomal degradation, and proteasomes can degrade glycoproteins in vitro (Kario et al. 2008). Furthermore, the loss of NGLY1 does not result in a general ERAD defect (Hirsch, Blom, and Ploegh 2003). Recently, transcriptome analysis of NGLY1 (Pngl) knockdown flies by RNA-sequencing found no evidence of ER stress or ERAD dysfunction (Owings et al. 2018).

23 1.3.4.2 Glycoprotein Homeostasis

The loss of NGLY1 function may have important biological consequences in glycoprotein homeostasis and offers insight into the pathophysiology of NGLY1-deficiency (Huang et al. 2015). In absence of NGLY1 activity, it has been proposed that N-glycans are inappropriately cleaved from glycoproteins by the downstream cytoplasmic enzyme ENGASE (Huang et al. 2015). When NGLY1 is present, the normal substrate of ENGASE is the soluble oligosaccharide liberated by NGLY1 (Huang et al. 2015). However, glycoproteins misprocessed by ENGASE in the absence of NGLY1 would retain a single GlcNAc residue that may promote their aggregation (Huang et al. 2015). For example, a model ERAD substrate was shown to be stabilized in Ngly1 knockout mouse embryonic fibroblasts and de-glycosylated by ENGase (Huang et al. 2015). Upon knockout of both Ngly1 and Engase, the ERAD substrate was quickly degraded by the proteasome (Huang et al. 2015). This degradation defect may have detrimental effects on cells by creating toxic aggregates and or through impairment of O-GlcNAc signaling, and has been proposed as a model of pathogenesis for NGLY1-deficiency (Huang et al. 2015).

In support of this hypothesis, Suzuki and colleagues found that Ngly1-deficient mice display varying degrees of lethality depending on the genetic background or genotype of the mouse strain (Fujihira et al. 2017). Deletion of Engase in a Ngly1-deficient background partially rescues this lethality (Fujihira et al. 2017). However, viable double mutants (Ngly1-/-Engase-/-) were reported not to be healthy and showed a severe phenotype reminiscent of the symptoms in human NGLY1-deficient patients (Fujihira et al. 2017). Therefore, the Ngly1-/-Engase-/- - deficient mice may serve as a valuable model for investigating the pathogenesis of NGLY1- deficiency, and ENGase may represent a therapeutic target for treating this disorder ( Bi et al. 2017; Fujihira et al. 2017).

1.3.4.3 Proteasome Homeostasis and Transcriptional Regulation via NFE2L1

Proteasome dysfunction causes adult-onset neurodegenerative diseases and may also be a general feature of decline during aging (Saez and Vilchez 2014). A second model proposes that NGLY1- deficiency pathology may be a result of defects in glycosylation of specific glycoproteins, including the regulator of the proteasome bounce-back response, transcription factor NRF1/NFE2L1(Lehrbach and Ruvkun 2016; van Keulen, Rotteveel, and Finken 2019). Indeed, recent studies have revealed that NGLY1 plays a critical role in mitigating proteotoxic stress

24 through regulation of NFE2L1 (Lehrbach and Ruvkun 2016; Tomlin et al. 2017; Lehrbach, Breen, and Ruvkun 2019).

A conserved response to proteasome dysfunction caused by pathogen attacks that target the proteasome, or proteasome inhibitor drugs, is the transcriptional upregulation of proteasome subunit genes (Meiners et al. 2003). In mammalian cells, this “bounce-back response” is regulated by NFE2L1, a member of the stress-responsive NRF/NFE2 family of Cap “n” Collar basic leucine zipper (CnC-bZip) transcription factor family (Radhakrishnan et al. 2010). NFE2L1 is a glycosylated ER-resident transcription factor that is maintained at low basal levels by ERAD. NFE2L1 is continuously retro-translocated from the lumen to the cytosol by the p97/VCP ATPase complex, deglycosylated by NGLY1(Radhakrishnan, den Besten, and Deshaies 2014; Tomlin et al. 2017), ubiquitinated (Steffen et al. 2010) and degraded by the proteasome (Radhakrishnan et al. 2010). However, when proteasome activity is compromised, some NFE2L1 escapes degradation and undergoes deglycosylation and subsequent proteolytic cleavage (Radhakrishnan et al. 2010; Radhakrishnan, den Besten, and Deshaies 2014). This cleavage results in a transcriptionally active form of NFE2L1, which enters the nucleus and activates the expression of target genes including proteasome subunit genes by binding to antioxidant response elements (ARE) (Radhakrishnan, den Besten, and Deshaies 2014).

NRF2/NFE2L2 is a closely related family member to NFE2L1, and binds to the same DNA sequence motif; however, it is not glycosylated and does not require NGLY1 enzymatic activity to function as a transcription regulator (Digaleh, Kiaei, and Khodagholi 2013). Furthermore, it regulates oxidative stress responses and plays a role in autophagy and mitophagy ( Digaleh, Kiaei, and Khodagholi 2013; Radhakrishnan et al. 2010). In flies, there is a single NFE2L1 homolog, cap-n-collar (Cnc), which increases the expression of proteasome subunit genes, as well as oxidative and redox stress response pathways (Grimberg et al. 2011). Similarly, in C. elegans the functions of both NRF1/2 depend on a single NRFF ortholog encoded by skn-1, which generates three isoforms (SKN-1A, SKN-1B, SKN-1C) (Blackwell et al. 2015). Recently, Lehrbach et al. 2019 demonstrated that SKN-1A and SKN-1C are two functionally distinct transcription factors that are activated in different contexts and regulate different target genes, mirroring the relationship between NFE2L1 and NFE2L2 (Lehrbach, Breen, and Ruvkun 2019).

25 A role for NGLY1 in regulating NFE2L1 processing and proteasome homeostasis was first identified in C. elegans (Lehrbach and Ruvkun 2016). The ER-associated SKN-1A isoform regulates proteasomal , and through genetic analysis a pathway for activation of SKN-1A was recently identified (Lehrbach and Ruvkun 2016). An unbiased forward genetic screen identified PNG-1 (NGLY1 ortholog) and aspartic protease DNA-damage inducible 1 homologue 1 (DDI-1; DDI2 in humans) as essential activators of SKN1-A, and like skn1-a mutants, both png-1 and ddi-1 C. elegans mutants display increased sensitivity to bortezomib (Lehrbach and Ruvkun 2016). A model was proposed where, in response to proteotoxic stress, SKN1-A/NFE2L1 is retro-translocated to the cytosol where it is de-glycosylated by PNG- 1/NGLY1 and cleaved by DDI-1/DDI2 to enter the nucleus and activate transcription (Figure 1.6)(Lehrbach and Ruvkun 2016). This seminal study was the first to propose that SKN- 1A/NFE2L1, PNG-1/NGLY1 and DDI-1/DDI2 function in a common genetic pathway.

A gene essentiality profile in 14 human leukemia cell lines subsequently identified a correlation in gene essentiality profiles with NFE2L1, NGLY1 and DDI2 (Wang et al. 2017). Concurrently, the mechanism for NFE2L1 processing and activation was demonstrated in human cells (Koizumi et al. 2016; Tomlin et al. 2017). The loss of DDI2 prevents NFE2L1 from being cleaved and entering the nucleus resulting in low levels of proteasome production and an accumulation of cytoplasmic full length NFE2L1 (Koizumi et al. 2016). Similarly, genetic or chemical-genetic disruption of NGLY1 impairs the nuclear import and transcriptional activity of NFE2L1 (Tomlin et al. 2017).

Recently, the molecular basis for SKN1-A activation by PNG-1 and DDI-I was determined and revealed that its activation requires the conversion of specific N-glycosylated asparagine residues with SKN-1A to aspartate via PNG-1 enzymatic activity. (Lehrbach, Breen, and Ruvkun 2019). Lehrbach et al. 2019 demonstrated that a non-glycosylated (bearing asparagine residues at N- glycosylation motifs) N-terminally truncated form of SKN-1A (mimics the product of DDI-I- dependent cleavage) does not rescue bortezomib hypersensitivity of skn1-a mutants. The authors hypothesized that sequence editing of SKN-1A may be required for it to interact with cofactors that specifically recognize the sequence-edited domain and facilitate activation of proteasome subunit genes (Lehrbach, Breen, and Ruvkun 2019). This mechanism of NFE2L1 activation has yet to be demonstrated in human cells and also poses the question as to whether it is a

26 Figure 1.6: Proposed NGLY1 genetic pathway and proteasome bounce-back response. (1) Full length NFE2L1 is glycosylated in the ER lumen and is retrotranslocated to the cytosol by p97/VCP. (2) NFE2L1 is de-N-glycosylated by NGLY1, ubiquitinated and (3) rapidly degraded by the proteasome rapidly degraded by the proteasome and thus is maintained at low levels in the cell. In cells with impaired proteasome capacity due to proteasome inhibition or an overload of misfolded proteins, NFE2L1 is deglycosylated by NGLY1 then (4) cleaved by DDI2. (5) Processed NFE2L1 accumulates and enters the nucleus where it heterodimerizes with Maf protein cofactors and activates the transcription of proteasome subunit genes by binding to antioxidant response elements (ARE). Created with BioRender.

27 fundamental requirement for its activity as a transcription factor or if it is specific to the proteasome bounce-back response.

Lastly, SKN-1A and PNG-1 were shown to control basal proteasome subunit gene expression and proteasome homeostasis in C. elegans (Lehrbach, Breen, and Ruvkun 2019). Impaired proteasome function is associated with diverse human diseases including neurodegeneration (Rousseau and Bertolotti 2018) which is also a feature of NGLY1-deficiency (Lam et al. 2017). Therefore, defective regulation of the proteasome by NFE2L1 may be causal for neurodegeneration in NGLY1-deficiency.

1.3.4.4 Mitochondrial Function

Recent findings have demonstrated a physiological connection between Ngly1 deglycosylation capacity and mitochondrial function (Kong et al. 2018; Yang et al. 2018). Mitochondrial respiratory chain diseases (RC) and CDGs display extensive clinical and genetic heterogeneity but also share overlapping clinical manifestations (Kong et al. 2018). A recent transcriptome meta-analysis of human RC diseases of diverse etiologies found that NGLY1 was one of the most dysregulated genes and was often upregulated, thus providing the first link between NGLY1 and mitochondrial function (Zhang and Falk 2014). In support of this finding, muscle and liver mitochondrial abnormalities have been identified in several patients with NGLY1-deficiency and mitochondrial physiology is impaired in several NGLY1-deficiency models (Kong et al. 2018). NGLY1 disease human fibroblasts, Ngly1-knockout mouse embryonic fibroblasts (MEFs) and C. elegans (png-1) whole animals displayed significantly impaired mitochondrial membrane potential, increased mitochondrial matrix oxidative burden and reduced cellular respiratory capacity (Kong et al. 2018). Taken together, these findings highlight that NGLY1 function is necessary for normal mitochondrial function and that cellular adaptation to RC dysfunction requires upregulation of NGLY1 expression (Zhang and Falk 2014; Kong et al. 2018).

NGLY1 has also been shown to regulate mitochondrial homeostasis and inflammation (Yang et al. 2018). NGLY1-deficient human and mouse cells display impaired mitophagy resulting in severely fragmented mitochondria and reduced mitochondrial function (Yang et al. 2018). The damaged mitochondria leak mitochondrial DNA (mtDNA) and RNA (mtRNA) into the cytosol which activates cytosolic innate immune DNA and RNA-sensing pathways, leading to chronic activation of type-I interferon (IFN) signaling (Yang et al. 2018). Interestingly, children with

28 NGLY1-deficiency appear to be resistant to common viral infections (Mnookin 2014), a phenotype that has been reported in patients with chronically activated type-I interferon responses that are caused by mutations in nucleic acid metabolizing enzymes (Hasan et al. 2013). Further, NGLY1 was found to regulate transcriptional programs of mitophagy through NFE2L1 activity. Expression of a transcriptionally active version of Nfe2l1 (cleaved and does not require deglycosylation by Ngly1), restored expression of mitophagy-related genes in Ngly1-/- mouse cells (Yang et al. 2018). This observation suggests that sequence editing of Nfe2l1 may not be a requirement for its transcriptional activity in regulating mitophagy genes or that it may be specific to C. elegans and warrants further investigation.

1.3.4.5 Lipid Metabolism and Transcriptional Regulation via NFE2L1

While a role for NFE2L1 in cholesterol homeostasis has been previously been demonstrated (Widenmaier et al. 2017), NGLY1 was recently shown to play a role in lipid metabolism in the liver (Fujihira et al. 2019). These findings may offer valuable insight into the pathogenesis of NGLY1-deficiency as one of the symptoms of the disease is abnormal liver function including elevated liver transaminases and steatosis (Enns et al. 2014; Lam et al. 2017)

Nrf1 is a critical sensor and mediator of cholesterol homeostasis and inflammation through a mechanism in which it senses cholesterol levels in the ER and also regulates a transcriptional program for cholesterol efflux (Widenmaier et al. 2017). When cholesterol levels are low, Nrf1 and the ER-resident transcription factor sterol regulatory element binding transcription factor 2 (SREBP2) coordinate cholesterol synthesis. SREBP2 is activated and enters the nucleus where it drives cholesterol synthesis by stimulating transcription of sterol-regulated genes (Madison 2016). Nrf1 also enters the nucleus and represses a cholesterol removal program by inhibiting liver X receptor (LXR) and CD36-inflammatory signaling (Widenmaier et al. 2017). LXRs are transcription factors that drive cholesterol export and CD36 is a scavenger receptor that facilitates lipid uptake, promotes inflammatory responses and regulates stress kinases such as JNK (Silverstein and Febbraio 2009). In the presence of excess cholesterol, SREBP2 is inhibited, and Nrf1 is retained in the ER where it directly binds and senses excess cholesterol in the ER membrane through its CRAC domain (Widenmaier et al. 2017). Retention of Nrf1 results in the depression of the LXR-mediated transcriptional cholesterol removal program and primes inflammatory signaling (Widenmaier et al. 2017). Taken together, the authors proposed a model where in the absence of Nrf1, CD36 is de-repressed resulting in chronic inflammatory signaling,

29 where in the liver, this causes hyperactivation of JNK leading to diminished cholesterol excretion and cholesterol accumulation. Indeed, the critical role of Nrf1 in defending against excess cholesterol is highlighted by the fact that in vivo cholesterol challenges in Nrf1-deficient livers induce massive hepatic cholesterol accumulation and damage (Widenmaier et al. 2017). However, the study by Widenmaier et al did not investigate NFE2L1’s function in cholesterol homeostasis in the absence of NGLY1 or DDI2.

Recently, Fujihira et al. 2020 found that hepatocyte-specific Ngly1-deficient mice had elevated liver transaminases, formed fatty liver and had increased lipid droplet accumulation under a high fructose diet, suggesting that lipid metabolism is dysregulated in the absence of Ngly1. A lipidome analysis revealed that lipids such as triacylglycerol and diacylglycerol were increased in the liver and decreased in the serum in hepatocyte Ngly1-deficient mice, suggesting that lipid release from the liver, as lipoproteins may be compromised in the absence of Ngly1 (Fujihira et al. 2019). While mice did not exhibit these phenotypes on a regular diet, transcriptome analysis supported the possibility that Ngly1-deficient livers have lipid metabolism defects as several cytochrome p450 enzymes, which are important in the synthesis and metabolism of fatty acids and sterols, were upregulated (Fujihira et al. 2019). Additionally, Cyp2j9, a cytochrome p450 enzyme that metabolizes arachidonic acid, was one of the top down-regulated genes (Fujihira et al. 2019). Arachidonic acid stimulation has been shown to induce the formation of lipid droplets in a JNK-specific manner (Guijas et al. 2012). Also, impaired Nfe2l1 function may in part contribute to dysregulated lipid metabolism observed in hepatocyte-specific Ngly1-deficient mice as its processing and localization was found to be impaired in this model. Lastly, while abnormal liver mitochondrial morphology has been reported in NGLY1-deficiency patient biopsies (Kong et al. 2018) and Ngly1-mouse cells (Yang et al. 2018), the authors did not observe any abnormalities in mitochondrial morphology, suggesting that Ngly1 LOF does not always impair mitochondrial physiology.

While dysregulated lipid metabolism and cholesterol homeostasis due to impaired NFE2L1 activity offers insight into the liver pathology observed in NGLY1-deficiency, it may also contribute to some of the neurological symptoms. A small molecule screen carried out in a fly model of NGLY1-deficiency found that -hydroxyecdysone (20E), cholesterol-derived ecdysteroid molting hormone 20, partially rescued the global developmental delay in mutant homozygotes (Rodriguez et al. 2018). Given the fact that impairments in cholesterol homeostasis

30 cause neurodegeneration (Anchisi et al. 2013), and that NGLY1-deficiency is a neurodevelopmental disorder, it is of interest to gain further insight into if and how NGLY1 may be involved in this critical biological process.

1.3.4.6 BMP Signaling and Digestive Tract Development

NGLY1 is thought to be important during embryonic development, which is highlighted by the fact the Ngly1 knockout mice are embryonic lethal in certain strain backgrounds (Fujihira et al. 2017). Loss of function of the D. melanogaster Pngl gene results in developmental delay and both larval and adult lethality (Funakoshi et al. 2010; Owings et al. 2018). It was recently shown that the loss of Pngl leads to defects in developmental bone morphogenic protein (BMP) signaling, a signaling pathway that regulates many aspects of embryonic development in both fruit flies and humans (Galeone et al. 2017). Disruptions of BMP signaling due to the loss of Pngl may also provide insight into the pathogenesis of NGLY1-deficiency and some of the symptoms observed in patients.

Cells secrete BMP ligands that exist as either hetero- or homo-dimers that bind to specific receptors on the surface of cells to stimulate BMP signaling cascades (O’Connor et al. 2006). Mutant png1 flies form fewer homodimers of a BMP ligand called Dpp, which results in defects in BMP signaling in two layers of tissue that are required for normal digestive system development, thus causing abnormal gut development (Galeone et al. 2017). Various human pathologies are associated with dysregulated BMP signaling in ophthalmic, gastrointestinal and musculoskeletal systems (Wang et al. 2014). Therefore, tissue-specific alterations in BMP signaling may contribute to some of the NGLY1-deficiency phenotypes including retinal abnormalities, delayed bone age and osteopenia, small feet and hands, and chronic constipation (Enns et al. 2014; Lam et al. 2017).

It is not known whether Pngl acts directly on Dpp to impair its ability to form homodimers, and not all of the observed Pngl phenotypes are due to defects in BMP signaling (Galeone et al. 2017). A recent transcriptome analysis in Pngl knockout flies determined that the loss of Pngl results in dysregulation of Cnc/Nfe2l1. Cnc target genes were downregulated, including genes encoding proteasome subunits and proteins with oxidation-reduction functions (Owings et al. 2018). Interestingly, a number of other genes were dysregulated in the absence of Pngl including downregulation of genes involved GlcNAc synthesis and upregulation of heat shock response

31 genes, and increasing the function of both pathways rescued lethality (Owings et al. 2018). Taken together, these findings open up the possibility that NGLY1 is required for the function of multiple glycoproteins and their dysregulation may contribute to various aspects of NGLY1- deficiency pathogenesis.

1.3.4.7 MHC Class I-Mediated Antigen Presentation

NGLY1 plays a critical role in MHC class I-mediated antigen presentation by the necessary de- glycosylation of tyrosinase (Ostankovitch et al. 2009), a key enzyme involved in melanin production. MHC class I molecules bind to peptides derived from intracellular pathogens or from proteins expressed in tumor cells and present them on the cell surface to the host immune system (Skipper et al. 1996). Post translational modifications such as de-amidation of Asn to Asp alter the repertoire of MHC class I peptides displayed on the cell (Skipper et al. 1996). A peptide derived from tyrosinase, the natural peptide target for melanoma-reactive cytotoxic T-cells, is de- amidated; which was determined to be dependent on glycosylation in the ER and subsequent deglycosylation by NGLY1 in the cytosol (Ostankovitch et al. 2009). Non-glycosylated tyrosinase is known to be rapidly degraded and fails to be efficiently presented on the MHC class I molecule HLA-A (Ostankovitch et al. 2009). Additionally, NGLY1 activity has been shown to affect the efficiency of antigen presentation, thus highlighting the functional importance of NGLY1 activity in this immunological process (Altrich-VanLith et al. 2006).

1.3.4.8 Aquaporin Expression

While NGLY1’s role in various biological processes have been largely attributed to its deglycosylating activity, NGLY1 regulates the expression of aquaporins by a mechanism that is independent of its catalytic activity (Tambe, Ng, and Freeze 2019). Aquaporins are membrane proteins that act as water channels and dysregulation expression of these proteins may explain multiple symptoms observed in NGLY1-deficiency.

Tambe et al. 2019 demonstrated that NGLY1-deficient human and mouse cell lines are resistant to hypotonic cell lysis as a result of reduced expression of multiple aquaporins which occurs at both the transcript and protein level. Interestingly, this regulation was found to be independent of NGLY1’s catalytic activity as a catalytically inactive Ngly1 restored normal hypotonic lysis in NGLY1 knockout human HAP1 cells and Ngly1 knockout MEFs, as well as aquaporin1 expression in Ngly1 knockout MEFs (Tambe, Ng, and Freeze 2019). Furthermore, Ngly1-

32 mediated regulation of aquaporins occurs through regulation of the Aft1/Creb1 signaling pathway; however, the authors did not investigate whether it is achieved by direct interaction between Ngly1 and these transcription factors (Tambe, Ng, and Freeze 2019). These findings suggest that NGLY1 may have a function independent of its catalytic activity in some organisms.

Lastly, ENGASE also appears to regulate aquaporins as aquaporin1 protein levels were reduced by ~50% in Engase knockout MEFs and by ~75-80% in Ngly1/Engase double knockouts which were even more resistant hypotonic cell lysis than Ngly1 knockout MEFs (Tambe, Ng, and Freeze 2019). However, this observation is inconsistent with previous studies have found that additional knockout of Engase in Ngly1 knockout cell and animal models has a protective effect on ERAD.

The role of Ngly1 in regulating aquaporin expression may explain several symptoms observed in NGLY1-deficiency including alacrima and seizures. Dysregulated aquaporin expression may result in reduced or absent tear secretion in patients with NGLY1-deficiency. Patients with Sjogren’s syndrome, an autoimmune disease characterized by extensive dryness, exhibit decreased salivary and lacrimal gland secretion as a result of defective aquaporin trafficking (Soyfoo et al. 2018; Tsubota et al. 2001). Additionally, aquaporin levels may contribute to neurological symptoms of NGLY1-deficiency such as seizures. Aquaporin4 has been implicated in neuroinflammation and AQP4-/- mice are more vulnerable to seizures after brain injury (Badaut et al. 2014). Lastly, the discovery of Ngly1-mediated regulation of aquaporins has resulted in the identification of a new cellular phenotype of NGLY1-deficient cells (i.e. resistance to hypotonic lysis), which may facilitate drug screening for future therapies.

1.4 The Complement System 1.4.1 The Complement System

The complement system is an evolutionarily ancient component of the innate immune system, playing a key role in defense against pathogens, immune surveillance and in host homeostasis (Walport 2001a, 2001b; Zipfel and Skerka 2009; Daniel Ricklin et al. 2010). As one of the first lines of defense, complement rapidly recognizes, tags and eliminates microbial intruders by phagocytosis, direct killing, and generation of an inflammatory response (Merle, Church, et al. 2015). Complement plays an important role in immune surveillance and host homeostasis by

33 recognizing and assisting in the clearance of immune complexes, cellular debris and modified or apoptotic host cells (Ricklin et al. 2010). This complement network is comprised of over fifty blood- and lymph-circulating, as well membrane-bound proteins that work in concert to discriminate among microorganisms, healthy host tissue, and modified cells, to stimulate the appropriate downstream immune response (Kolev and Kemper 2017). Once activated, complement progresses in a sequential manner, allowing for regulation at several critical stages of the complement cascade. Complement regulation is self-controlled, and thus both directs immune effector functions and modulates the intensity of the response accordingly (Zipfel and Skerka 2009). This allows for a controlled outcome of complement activation, which is needed for the recognition and appropriate removal of infectious or modified host cells (Merle, Noe, et al. 2015).

In the past decade, there has been a shift in understanding surrounding the complement system. While protection against infection is a defining feature of this pathway, it is now appreciated that the complement system exerts much broader roles in shaping adaptive immune responses, is important for cell and tissue integrity and homeostasis, and is at the nexus of extensive crosstalk between many biological pathways (Ricklin et al. 2010; Hajishengallis et al. 2017; Kolev, Friec, and Kemper 2014; Reis et al. 2019). Furthermore, the recent discovery that complement activation and function is not restricted to the extracellular space as previously thought, but also occurs intracellularly, has revealed that complement regulates basic cellular processes, particularly those of a metabolic nature (Arbore, Kemper, and Kolev 2017; Hess and Kemper 2016; Liszewski et al. 2013; Kolev and Kemper 2017; Kolev et al. 2015). Complement engages receptors both intracellularly and on the surface to induce cellular responses (Arbore, Kemper, and Kolev 2017; Liszewski et al. 2013). Although most studies on intracellular complement, or the “complosome”, have been in CD4+ T cells, intracellular complement has been identified in other cell types, highlighting that it likely regulates homeostatic functions for both immune and non-immune cells (Liszewski et al. 2013).

The involvement of complement in basic cellular processes may be due to the fact that all complement receptors, and some regulators like the complement regulatory proteins (CRPs), are membrane-bound proteins with signaling capacity, so ligand binding is likely to impact the functional behavior of the cell (Kemper and Atkinson 2007). As our understanding of the functions of complement now encompasses metabolism and basic cellular processes, it is not

34 surprising that dysregulation of normal complement function, or its regulators, is involved in a broader range of human diseases than what has previously been thought (Ricklin and Lambris 2013; Ricklin, Reis, and Lambris 2016). For example, the widely expressed CRPs are involved in the pathogenesis of numerous common and rare diseases including cancer, multiple sclerosis (MS), age-related macular degeneration (AMD), systemic lupus erythematosus (SLE), paroxysmal nocturnal hemoglobinuria (PNH) and atypical hemolytic uremic syndrome (aHUS) (Astier 2008; Das et al. 2019; Dho, Lim, and Kim 2018; Ellinghaus et al. 2017; Fremeaux- Bacchi et al. 2006 Hill et al. 2017; Liszewski and Atkinson 2015).

The critical functions of the CRPs as regulators of complement has been well understood for many decades. However, there is now accumulating evidence that the CRPs are involved in cellular processes that are independent of serum-complement inhibition, which has offered new insight into their roles in the pathogenesis of various diseases. Despite these advances, an unbiased genetic approach to investigating the complement-independent functions of the CRPs has not been carried out. Given the utility of GIs for identifying functional relationships between genes that are indicative of their roles within the cell, employing genome-wide CRISPR screens may offer valuable insight into CRP function.

1.4.2 The Complement Cascade

The complement system becomes activated in a cascade-like fashion when triggered by one or more of three activation pathways: the classical, lectin and alternative or alternative pathways (Figure 1.7). While distinct and initiated by different molecules, these pathways converge upon the generation of the enzymatic complex known as the C3 convertase (Merle, Church, et al. 2015). As the cascade progresses, effector molecules are generated and facilitate diverse biological functions through binding to receptors expressed on various immune cells.

C1q is the pattern recognition molecule (PRM) of the classical pathway and recognizes distinct structures on microbial cells (pathogen-associated molecular patterns; PAMPs) or apoptotic cells (damage associated molecular patterns, DAMPs) as well as endogenous immunoglobulins such as IgM or IgG ((Merle et al. 2015); Daniel Ricklin, Reis, and Lambris 2016). Surface binding of C1q activates the C1 complex to which it belongs (Merle, Church, et al. 2015). Another component of the C1 complex, serine protease C1s, cleaves plasma proteins C4 and C2 into C4b

35 Figure 1.7: Systemic complement activation. Serum circulating complement can be activated via three pathways: the classical, lectin and alternative pathways, which all lead to the formation of C3 convertases. When pattern recognition molecules of the classical (C1q) and lectin (MBL, Ficolins, Collectins) pathways encounter antibody clusters or PAMPs on a foreign cell, activated serine proteases (C1r, C1s, or MASPs) cleave C4 and C2 to form the classical/lectin pathway C3 convertase (C4b2b). In the alternative pathway, tick-over or cell surface interactions of C3 leads to its hydrolysis, and Factor D (FD) cleaves Factor B (FB), together generating the initial fluid-phase alternative C3 convertase (C3H20Bb). All C3 convertases cleave C3 into anaphylatoxin, C3a, which binds to C3aR on immune cells to induce chemotaxis and inflammatory signaling, and opsonin, C3b, which is deposited on the cell surface and its phagocytic uptake. Factor B binds to deposited C3b (via Factor D activation) to form the final alternative C3 convertase (C3bBb), which produces more C3b (and C3 convertases) through an amplification loop. Factor P (FP; Properdin) stabilizes the alternative C3 and C5 convertases. C3b is also degraded to iC3b and C3dg which interact with complement receptors on immune cells to mediate adhesion or phagocytic uptake. Increased deposition of C3b leads to the generation of C5 convertase which cleave C5 to generate anaphylatoxin, C5a which binds to C5aR on immune cells, and C5b, which combines with C6-C9 to form membrane attack complexes (MACs), resulting in lysis of the target cell. Healthy cells express membrane-bound complement regulatory proteins (CRPs) on their surface to control complement activation. CD35 and CD46 act as cofactors for Factor I-mediated mediated cleavage of C3b and C4b and CD55 binds to C3b/C4b to accelerate the decay of convertases. CD59 prevents the formation of the MAC. Soluble complement inhibitors are not shown. Silent removal of damaged and apoptotic cells, debris, and immune complexes is achieved through the sensing of DAMPs, and limited deposition of opsonins that facilitate phagocytic clearance (not shown). Created with BioRender.

36

and C2b, respectively, thus forming the classical pathway C3 convertase enzyme (C4bC2b) (Merle, Church, et al. 2015). Similarly, in the lectin pathway, PRMs such as mannose-binding lectin (MBL), ficolins1-3, and collectin-11, recognize carbohydrates on microbial and apoptotic cells (Merle, Church, et al. 2015). After recognition, they form complexes with MBL-associated serine proteases (MASPs), which then cleave proteins C4 and C2, resulting in the formation of the C3 convertase (C4bC2b) (Merle, Church, et al. 2015). The alternative complement pathway does not require specific activation and is constantly active at low-levels on biological surfaces in plasma and other body fluids (Merle, Church, et al. 2015). This allows for constant probing of surfaces and initiates rapid amplification of the cascade (Merle, Church, et al. 2015). Through a process termed “tick-over”, C3 undergoes spontaneous hydrolysis leading to conformational rearrangements that allow C3H20 to bind factor B and form the alternative pathway C3 convertase enzyme (C3H20Bb) (Merle, Church, et al. 2015).

Activation of each of these pathways converges on the assembly of the first enzyme in the cascade, the C3 convertase (Merle, Church, et al. 2015). Both pathway C3 convertases cleave C3 to liberate the anaphylatoxin C3a and generate C3b (Merle, Church, et al. 2015). Newly released C3a recruits immune cells to induce a proinflammatory response and the C3b fragment is rapidly deposited on the particle surface, thereby marking the cell for phagocytosis (Merle, Church, et al. 2015). Additionally, deposited C3b in the presence of factor B and factor D, results in the generation of the alternative pathway C3 Convertase (C3bBb) (Merle, Church, et al. 2015) . This convertase is able to cleave additional C3 proteins and produce more convertases, thereby creating a rapid self-amplification loop that leads to efficient opsonization of cells (Merle, Church, et al. 2015).

If activation proceeds beyond the cleaving of C3, additional C3b molecules bind to the C3 convertases to generate the C5 convertases for the alternative (C3bBbC3b) or classical/lectin (C4bC2bC3b) pathways (Merle, Church, et al. 2015). Both enzymes cleave C5 into the anaphylatoxin and chemoattractant C5a, and C5b (Merle, Church, et al. 2015). Liberation of the C5b fragment results in the non-enzymatic assembly of terminal pathway components C5b, C6, C7, C8 and multiple copies of C9, which together form the membrane attack complex (MAC) (Merle, Church, et al. 2015). The MAC becomes inserted into the cell membrane, thereby forming a large pore that results in cell lysis (Merle, Church, et al. 2015).

37

1.4.3 Functions of Complement

Activation of the complement system leads to a number of biological outcomes. The complement system works to target, kill and clear pathogens such as bacteria, fungi and viruses via both direct and indirect mechanisms (Merle, Noe, et al. 2015). Furthermore, the complement system enables the clearance of host machinery, such as immune complexes that are present after the completion of an immune response (Merle, Noe, et al. 2015). In addition to pathogen killing and clearance, the complement system also acts on apoptotic host cells, facilitating their immunologically silent clearance from healthy tissues (Merle, Noe, et al. 2015). Many of these functions are achieved through the activity of complement activation products and effector compounds, such as the anaphylatoxins, opsonins and C1q (Merle, Noe, et al. 2015). By binding to various complement receptors on different cell types, these cascade components induce and control biological responses.

1.4.3.1 Defense Against Pathogens

The complement system acts as a first line of defense against pathogens and functions to eliminate foreign invaders in a pre-emptive attempt to prevent infections (Meri 2016). Complement operates based on discrimination of “self” from “non-self”; host cells either express inhibitory molecules on their surface, or recruit them to the cell membrane from the plasma (Zipfel and Skerka 2009). Cells, debris or microorganisms that do not bear these protective inhibitors will serve as a surface for complement activation. Pathogens are targeted by all three complement pathways, depending on the exact membrane composition of the invading organism (Merle et al. 2015). The complement system eliminates pathogens by inducing inflammation, opsonization and phagocytosis, and direct lysis via the membrane attack complex (Merle, Noe, et al. 2015).

1.4.3.1.1 Opsonization and Phagocytosis

Although complement employs various strategies to destroy pathogens, its primary function is to eliminate the pathogen indirectly, by depositing opsonins on the surface of pathogen targets allowing for their recognition, ingestion and destruction by phagocytic cells, neutrophils, monocytes and macrophages (Merle, Noe, et al. 2015). C3 fragments (C3b and iC3b) and IgG antibodies function as opsonins upon activation of complement. Direct activation of the alternative pathway (which does not involve antibodies) and the subsequent deposition of C3 38

fragments on the pathogen, allows for their quick removal before an adaptive immune response has been mounted (Merle, Noe, et al. 2015). Phagocytes express specific receptors for C3 fragments that mediate or enhance phagocytosis of coated targets (Merle, Church, et al. 2015). Phagocytes also express Fc gamma receptors (FcyR) that bind to IgG molecules for phagocytosis of IgG-coated pathogens (Merle, Noe, et al. 2015). Once internalized, pathogens are killed by toxic reactive oxygen species and by microbicidal components such and proteases (e.g. lysozyme) present in phagocytic granules (Merle, Noe, et al. 2015).

1.4.3.1.2 Inflammation

Complement plays a pro-inflammatory role in the immune response to pathogens (Merle, Noe, et al. 2015). Complement-mediated inflammation is mediated by the anaphylatoxins generated during the activation of C3 and C5 (C3a and C5a), and to a lesser extent the small fragment of C4 (C4a) (Dunkelberger and Song 2010). The anaphylatoxins recruit immune cells by binding to specific receptors including the C3a receptor (C3aR) and C5a receptor (C5aR) expressed on various immune cells such as macrophages, neutrophils, eosinophils, mast cells, basophils, endothelial cells, epithelial cells and Kupffer cells (Klos et al. 2009; Holers 2014). Through receptor interactions, C3a and C5a upregulate inflammatory pathways by stimulating oxidative bursts in several immune cell types, histamine production by basophils and mast cells, and degranulation of epithelial and mast cells (Holers 2014; Merle, Noe, et al. 2015). Macrophages also express receptors that bind C1q (C1qRs), which modulates macrophage inflammatory responses such as enhanced cytokine production and phagocytosis (Bohlson et al. 2014).

1.4.3.1.3 Direct lysis via the MAC

Complement directly eliminates pathogens upon activation of the terminal complement pathway, in which formation of the MAC complex creates large pores in the target membrane. These large pores perturb the osmotic balance of a pathogen and enables access of additional enzymes for degradation of any remaining membrane (Bayly-Jones, Bubeck, and Dunstone 2017). These pores result in pathogen lysis and subsequently clearance.

1.4.3.2 Immune Complex Clearance

After the clearance of an infection and/or inflammatory event, the complement system will clear the host machinery assembled within the immune complex (Merle, Noe, et al. 2015). CR1 or

39

CD35 is a complement receptor and inhibitor protein involved in the regulation of C3 convertases. It is expressed primarily on erythrocytes and phagocytes and binds opsonin C3b (Zipfel and Skerka 2009). CR1 present on the surface of erythrocytes facilitates clearance of soluble immune complexes by transporting them to the liver and spleen to be cleared by macrophages (Merle, Noe, et al. 2015).

1.4.3.3 Apoptotic Cell Clearance

Between the surveillance of healthy host cells and pathogens, the complement system must also recognize altered and apoptotic host cells and facilitate their safe or immunologically-silent clearance by phagocytosis (Merle, Noe, et al. 2015). Clearance of apoptotic cells is critical for many physiological processes, including development, tissue remodeling, and maintenance of homeostasis (Merle, Noe, et al. 2015; Daniel Ricklin et al. 2010). Apoptotic cells undergo cellular modifications, such as the release of DAMPs and the decreased expression of CRPs, that activate complement. Apoptotic cells are then opsonized by C1q and C3b which allows them to be recognized and degraded by phagocytic cells without mounting an immune response. (Merle, Church, et al. 2015)

1.5 Complement Regulatory Proteins

Regulators of complement activity are essential in protecting healthy host cells from complement attack and in tuning the response of the complement system to microbial invaders and homeostatic insults. There are a number of soluble and membrane-expressed proteins that regulate the activity of the complement cascade (Zipfel and Skerka 2009). Due to the proinflammatory and destructive capabilities of complement, many of the complement regulators function as inhibitors of the cascade (Zipfel and Skerka 2009). Negative regulation of complement occurs at all four stages of the cascade by numerous fluid-phase and membrane- bound regulatory proteins, and through various mechanisms of action; for example, inactivation of activating proteases (C1s and MASPs), competitive inhibition/dissociation convertase complexes, cofactor activity for degradation of cleavage products and blocking terminal MAC formation (Holers 2014).

The CRPs are key regulators of complement activation on the cell surface and are comprised of complement receptor 1 (CR1/CD35), membrane cofactor protein (MCP/CD46), decay-

40

accelerating factor (DAF/CD55) and protectin (CD59) (Fishelson 2003). They aid in “non-self” versus “self” discrimination in that foreign cells are usually lacking them, and thus are recognized and attacked (Gorter and Meri 1999). Although their individual expression is variable, CD46, CD55, and CD59 are widely expressed by most cell types (Zipfel and Skerka 2009), and thus are considered to be the main membrane-bound regulators of complement. Although complement control by CRPs is vital in the prevention of tissue damage, it is now understood that they have a role beyond sole regulation on the cell surface (Kolev and Kemper 2017; Melania Capasso et al. 2006; Sutavani et al. 2013; Sivasankar et al. 2009; Frolíková et al. 2012). Due to their critical roles in complement regulation as well as other biological processes, it is not surprising that they are involved in the pathogenesis of numerous common and rare diseases.

1.5.1 Structure, Expression and Inhibitory Functions

1.5.1.1 CD46; Membrane Cofactor Protein

CD46/MCP belongs to a family of genetically-, structurally-, and functionally-related proteins called the regulators of complement activation (RCA), or complement control proteins (CCP) (Liszewski, Post, and Atkinson 1991; Zipfel and Skerka 2009). The RCA proteins control the assembly of the C3 and C5 convertases; they are either co-factors in the factor I-mediated cleavage of C3b and/or C4b, or can accelerate the decay of already formed complexes (Holers 2014). As its name implies, CD46 functions as a cofactor for factor-I mediated cleavage and inactivation of opsonins C3b and C4b (Liszewski, Post, and Atkinson 1991; Barilla-labarca et al. 2002). In turn, CD46 prevents amplification of the alternative pathway feedback loop on the same cell on which it is expressed (Oglesby et al. 1992).

CD46 is the only ubiquitously expressed (except on erythrocytes) co-factor for the factor-I mediated cleavage of C3b and C4b, and thus plays a key role in protecting host cells from unwanted complement attack (Cardone, Le Friec, and Kemper 2011). However, in mice CD46 is largely restricted to the testis while another related protein, Crry, is widely expressed and performs the complement-regulatory activity of CD46 (Kemper et al. 2001). CD46 is a type I transmembrane glycoprotein that is commonly co-expressed on cells as four distinct isoforms that arise from alternative splicing of a single gene (Yamamoto et al. 2013). While most cells express all four isoforms, tissue specific predominance is observed (Tang et al. 2016).

41

All CD46 isoforms have four repeating units beginning at the N-terminus, known as complement control protein (CCP) domains, short consensus repeats (SCR) or sushi domains which are conserved across the RCA family (Liszewski, Post, and Atkinson 1991) (Figure 1.8). The SCR modules of CD46 contain the sites for its regulatory activities (and structural studies have mapped binding sites for complement ligands C3b and C4b (Persson et al. 2010). The SCRs are followed by a heavily O-glycosylated serine, threonine and proline-rich (STP) region that arises from alternative splicing of exons and plays a key role in controlling CD46 cleavage and its regulatory functions (Choileain et al. 2011). Following the STP region is a region of unknown function followed by a transmembrane anchor and one of two distinct cytoplasmic domains, CYT-1 and CYT-2. Both cytoplasmic domains contain signaling motifs and mediate distinct cellular functions, highlighted by their opposing roles in T cell activation (Choileain et al. 2017; Choileain et al. 2011).

1.5.1.2 CD55; Decay Accelerating Factor

Like CD46, CD55/DAF belongs to the RCA family of proteins and its primary function is to accelerate the spontaneous decay of the classical (and lectin) and alternative pathway C3 convertases by dissociating them into their constituent proteins (Lublin and Atkinson 1989). By binding to C3b, CD55 prevents the formation and accelerates the dissociation of the alternative pathway C3 convertase (C3bBb), whereas binding to C4b accelerates the dissociation of the classical/lectin pathway C3 convertases (C4bC2b) (Noris and Remuzzi 2013). CD55 is expressed on all cells including erythrocytes, epithelial cells and endothelial cells (Kim and Song 2006).

CD55 is a glycoprotein that is expressed as two different isoforms; a glycosylphosphatidylinositol (GPI)-anchored form and a soluble form generated by alternative splicing (Vainer et al. 2013). The GPI-anchored form of CD55 is broadly expressed on blood, stroma, epithelial, and endothelial cells whereas the soluble form present in body fluids including plasma, urine, saliva, tears, and synovial fluids (Dho, Lim, and Kim 2018). However, GPI-linked CD55 is regarded as the major form of CD55 rather than soluble CD55 which is expressed at lower levels (Vainer et al. 2013).

Like other RCA proteins, CD55 has four repeating SCR domains (Figure 1.8) (Lukacik et al. 2004) which contain binding sites for C3b and C4b. The SCRs are followed by a heavily O- glycosylated STP region and a C-terminal GPI-anchor. 42

Figure 1.8: Schematic of the structure of complement regulatory proteins. CD46, CD55 and CD59. CD46 (MCP; membrane cofactor protein) is comprised of four short-consensus repeat (SCR) domains (also called sushi or complement control protein domains) which bind C3b, C4b and Factor I, a heavily O-glycosylated region rich in serine, threonine and proline (STP) and a region of unknown function. Following a transmembrane (TM) segment is a short cytoplasmic tail (CYT). Four isoforms of CD46 (BC1, C1, BC2, C2) arise through alternative splicing to give rise to different STP domains (BC or C isoform) and cytoplasmic tails (CYT1 or CYT2). CD55 (DAF; decay accelerating factor) is also comprised of four SCR domains which bind C3b and C4b to accelerate the decay of pathway convertases, a STP domain, followed by a GPI-anchor. CD59 is structurally different than the other CRPs and contains a cysteine-rich LY-6 domain followed by a GPI-anchor. CD59 inhibits the formation of the membrane attack complex by binding to C8 and C9. Created with BioRender.

43

In mice, there are two CD55/DAF genes (Daf1 and Daf2), with Daf1 encoding a GPI-anchored Daf that is widely expressed while Daf2 encodes a transmembrane Daf that is restricted to the testes (Turnberg and Botto 2003).

1.5.1.3 CD59; Protectin

Unlike the other CRPs, CD59 is not a member of the RCA family and serves as the primary inhibitor of the MAC (Zipfel and Skerka 2009). CD59 plays a critical role in the protection of self-tissues and is widely expressed on cells including erythrocytes (Zipfel and Skerka 2009). CD59 is a small glycoprotein protein that is attached to cell membranes via a GPI anchor (Figure 1.8). Mice harbor two genes encoding Cd59 with the two forms of Cd59 protein (Cd59a and CD59b) exhibiting 60% identity with one-another, where CD59b displays the greatest degree of homology to CD59 (Blom 2017). Structurally, CD59 is quite different than the other CRPs and belongs to the leukocyte antigen-6 (Ly6) family of proteins characterized by a LU domain (Loughner et al. 2016).

The MAC is formed by the self-assembly of complement proteins C5b, C6, C7, C8, and several molecules of C9 (Huang et al. 2006). CD59 inhibits the formation of the MAC by binding C8 and in the C5b-8 complex, preventing subsequent C9 binding in the C5b-9 complex, and preventing the requirement of additional C9 (Huang et al. 2006).

1.5.2 Functions Beyond Complement Regulation

Although the CRPs are key regulators of the complement cascade and play an important role in defence against pathogens, roles for the CRPs beyond complement cascade regulation have been identified (Hess and Kemper 2016; Capasso et al. 2014; Kimberley, Sivasankar, and Paul Morgan 2007; Blom 2017; Golec et al. 2019). The CRPs, and in particular CD46, are important regulators of T- cell function and homeostasis which is highlighted by the fact that dysregulation of CRP-mediated T cell signaling is observed in several diseases (Astier et al. 2006; Ellinghaus et al. 2017; Visser et al. 2002). Furthermore, the CRPs appear to be involved in different aspects of fertilization through their expression on sperm membranes (Riley-Vargas et al. 2004; Frolíková et al. 2012). It is therefore possible that the CRPs may have other important cellular functions that extend beyond complement regulation that have yet to be identified.

44

1.5.2.1 T Cell Regulation

Although CD46 was initially discovered as a complement regulator, it has become- increasingly apparent that it also functions as a complement receptor and that is central in T cell regulation (Hess and Kemper 2016; Kolev and Kemper 2017; Cardone et al. 2010; Liszewski et al. 2013). Efficient CD4+ T cell responses require three main types of signals : T cell receptor (TCR) engagement by antigens expressed on MHC class II molecules of antigen presenting cells (APCs) (signal 1), the stimulation of a costimulatory molecule such as CD28 on T cells (signal 2) and a T-cell polarizing signal from the cytokine milieu (Reis et al. 2019) Studies over the past decade have revealed that locally and intracellularly produced complement acts in an autocrine fashion to modulate T cell responses, independent of APCs (Liszewski et al. 2013; Arbore et al. 2016). Furthermore, CD46 has been shown to be integral to both CD4+ T cell homeostasis and metabolic reprogramming of T cells during activation (Figure 1.9) (Hess and Kemper 2016) .

In resting human CD4+ T cells, C3 is cleaved intracellularly via cathepsin-L-mediated cleavage (CTSL) into C3a and C3b in the lysosome (Liszewski et al. 2013). C3a binds to the C3aR on lysosomes and stimulates the production of low-levels of mechanistic target of rapamycin (mTOR), thereby sustaining homeostatic T cell survival (Liszewski et al. 2013). This homeostatic mechanism is sustained by the inhibition of Notch signaling via the interaction between CD46 and Notch ligand Jagged1 (Le Friec et al. 2012).

Engagement of CD46 on activated T cells results in either TH1 cell or Treg (Tr1) cell polarization depending on local levels of IL-2 (Cardone et al. 2010). TH1 cells are effector T cells that promote cell mediated immunity, whereas Tr1 cells are a class of regulatory T cells that regulate tolerance towards antigens and suppress tissue inflammation in autoimmunity (Cardone et al. 2010). Upon TCR and CD28 engagement on CD4+ T cells, intracellular C3a and C3b translocate to the surface where they engage cell-surface C3aR and activate CD46 (via C3b)(Kolev, Friec, and Kemper 2014). Activation of CD46 leads to the assembly of the IL-2 receptor, increased expression of the late endosomal/lysosomal adaptor and MAPK and mTOR activator 5 (LAMTOR5) and the assembly of the mTOR complex 1, leading to nutrient influx (amino acids and glucose) and OXPHOS (Kolev et al. 2015). These metabolic changes lead to the induction of

TH1 cells and secretion of IF- (Kolev et al. 2015). CD4+ T cells also continuously generate intracellular stores of C5a through a mechanism that has not yet been defined

45

Figure 1.9: CD46 in T cell homeostasis and effector function. In resting human CD4+ T cells, C3 is cleaved intracellularly via cathepsin-L-mediated cleavage (CTSL) into C3a and C3b in the lysosome. C3a binding to C3aR on lysosomes stimulates the production of low-levels of mechanistic target of rapamycin (mTOR), sustaining homeostatic T cell survival. Homeostasis is sustained by the inhibition of Notch signaling via the interaction between CD46 and Jagged1. Upon TCR and CD28 engagement, intracellular C3a and C3b translocate to the surface where they engage cell-surface C3aR and activate CD46. CD46 activation leads to the assembly of the IL-2 receptor and the mTOR complex 1 (mTORC1), leading to the influx of amino acids (AA) and glucose (Glu) and OXPHOS, resulting in Th1 induction secretion of IF-. Additionally, intracellular C5a binding to C5aR induces mitochondrial ROS production and inflammasome activation which sustains TH1 induction and IF- production. CD46 engagement (via C3b) and the accumulation of IL-2 induces a switch to IL- 10 production, resulting in decreased IF- production, glycolysis and OXPHOS and contraction to Treg cell phenotype. LAMTOR5: late endosomal/lysosomal adaptor and MAPK and mTOR activator 5, GLUT1: glucose transporter 1, LAT1: L type amino acid transporter 1.

46

(Arbore et al. 2016). Upon T cell activation, intracellular C5a binding to C5aR induces mitochondrial ROS production and inflammasome activation which sustains TH1 cell induction and IF- production (Arbore et al. 2016). CD46 engagement and the accumulation of IL-2 induces a switch to IL-10 production, resulting in decreased IF- production, glycolysis and

OXPHOS and contraction to Treg cell phenotype (Kolev, Friec, and Kemper 2014).

Additionally, a key component of primary T cell activation and function is the enzymatic processing of CD46 culminating in the cleavage of its cytoplasmic tails and their translocation to the nucleus where they control target genes (Choileain et al. 2011). Processing of CYT-1 is mediates the upregulation of the glucose transport GLUT1and the amino acid (AA) channel for nutrient influx into the cell and the assembly of the nutrient-sensing mTOR machinery, which facilitates the high levels of glycolysis and oxidative phosphorylation that are needed for IFN-γ production and subsequent TH1 activation (Kolev et al. 2015; Hess and Kemper 2016). Conversely, cleavage of CYT-2 results in inhibition of T cell activation and thus is critical to ensuring T cell homeostasis (Choileain et al. 2011). Taken together, these findings highlight the critical role of CD46 as a complement receptor that is integral to T cell function and suggests compartmentalization of its functions during serum and immune cell stimulation.

CD55 has been shown to play both an activating and suppressing role in T cell function, acting either directly on the T cells or indirectly via interaction with CD97 on antigen presenting cells (APCs) (Clarke and Tenner 2014). CD97 engagement of CD55, combined with CD3 stimulation in the presence of IL-2, activates human CD4+ T cells, resulting in enhanced T cell proliferation and secretion of IL-10 (Capasso et al. 2014). Further, costimulation of human naïve CD4+ T cells through the CD55-CD97 interaction drives Tr1 activation, expansion and function (Sutavani et al. 2013). Taken together, these findings provide evidence of a direct effect of CD55 on human T cell function that is independent of complement.

CD59 plays a suppressive role in T cell function. CD59 is upregulated on activated human CD4+ T cells and leads to down regulation of their activity (Sivasankar et al. 2009). Blockade of CD59 enhances CD4+ T cell responses including increased proliferation, decreased IL-10 production and increased IF-γ production (Sivasankar et al. 2009). It has been proposed that CD59– mediated signaling in T cells may be due to binding of a non-complement ligand expressed on the surface of APCs (Sivasankar et al. 2009). However, the effects of CD59 as well as CD55

47

may be mediated by modulating lipid rafts on T cells (Blom 2017) which is important for intracellular TCR signaling (Kabouridis 2006).

1.5.2.2 Fertilization

There is evidence to suggest that some or all of the CRPs play a role during fertilization (Frolíková et al. 2012). CD46 is expressed on the inner acrosomal membrane of sperm where its mediates fusion of the sperm and oocyte (Rooney, Oglesby, and Atkinson 1993; Riley et al. 2002).The sperm-egg interaction driven by CD46 is mediated via its SCR1 domain, which is not required for binding to complement ligands (Riley et al. 2002). In humans, the lack or severely reduced expression of CD46 on spermatozoa is connected with idiopathic male infertility (Nomura et al. 2001).

CD55 and CD59 are also expressed on the acrosomal membrane in human sperm however their roles in fertilization remain to be clarified (Frolíková et al. 2012). In mice, CD59b is highly expressed in the testes (Blom 2017) and reproductive abnormalities such as decreased sperm count and morphological abnormalities have been reported in mice lacking CD59b (Frolíková et al. 2012).

1.5.3 Complement Regulatory Proteins in Disease

Owing to their functions as complement inhibitors as well as their roles in T cell regulation, the CRPs contribute to the pathogenesis of numerous common and rare diseases. Overexpression of the CRPs on tumors play a protective role against immune detection and destruction regulators that may be augmented by their roles in T cell regulation (Fishelson 2003; Gorter and Meri 1999; Jurianz et al. 1999). Conversely, mutations in the CRPs or in genes involved in their synthesis resulting in decreased CRP expression, renders cells susceptible to excessive complement activation and destruction (Hill et al. 2017). Furthermore, abnormalities in CRP expression and function that result in chronic proinflammatory states contribute to the pathogenesis of autoimmune diseases such as MS and SLE (Ellinghaus et al. 2017; Astier 2008; Liszewski and Atkinson 2015; Toomey, Cauvi, and Pollard 2014). Taken together, the CRPs are involved in the pathogenesis of many human diseases, so gaining insight into their complement-dependent and independent functions may provide new opportunities for therapeutic intervention.

48

1.5.3.1 Cancer

Although the CRPs are expressed throughout the body to prevent over-activation of the complement system, there is increasing evidence that they act as double-edged swords. The CRPs can restrict complement activation such that it is no longer effective at eliminating cancer cells (Geller and Yan 2019). Their involvement in cancer pathogenesis is highlighted by studies over the last decade that indicate they may function as biomarkers of malignant transformation, and CRP expression is often associated with poor prognosis (Guc et al. 2000; Surowiak et al. 2006; Reis et al. 2018; Durrant et al. 2003). Furthermore, the expression of CRPs on tumors has been shown to interfere with cancer immunotherapies by restricting them in harnessing complement-dependent cytotoxicity (CDC) as a mechanism of action (Zhao et al. 2009). Thus, inhibition of CRPs has been explored to improve responses to various chemotherapies. Indeed, blocking of CRPs with neutralizing antibodies alongside existing therapeutic antibodies, such as rituximab or trastuzumab, leads to improved killing of tumor cells in vitro and in mouse models (Golay et al. 2000; Macor et al. 2007; P Macor et al. 2014).

However, given the role of the CRPs in regulating T cell responses, the CRPs may be involved in cancer pathogenesis as both inhibitors of complement and also modulators of adaptive T cell responses that promote immune evasion and tumor growth. Dysregulation of CD46 has been implicated in chronic inflammatory diseases by impairing TH1 – Tr1 switching (Astier et al. 2006; John Cardone et al. 2010) and chronic inflammation lends to proliferation and survival of malignant cells (Colotta et al. 2009). CD55 expression on tumors may also be functionally related to its role as a ligand for CD97 in T cell activation (Geller and Yan 2019). T cells stimulated in response to the CD55-CD97 interaction produce cells that behave like Tregs, which would promote tumor progression if expressed in the tumor microenvironment (TME) (Geller and Yan 2019).

Although studies have shown that enhancing complement activation improves tumor cell killing, it may in fact be beneficial for the tumor, as prolonged inflammation, which could be caused by activation of the complement cascade (and generation of the anaphylatoxins), promotes tumorigenesis (Colotta et al. 2009).

A new paradigm for understanding the role of CRPs in cancer has been proposed, where with increasing inflammation in the TME, CRP expression continually increases as a way to limit 49

complement activation (Geller and Yan 2019). CRP expression may therefore represent an important biomarker for highly inflammatory TMEs that lend to aggressive tumor growth and metastasis (Geller and Yan 2019). To gain insight into the complexity of the CRPs in tumor development and how they may be harnessed for developing future cancer therapies, a deeper understanding of their role in innate and adaptive immune responses are needed.

1.5.3.2 Diabetes

Recent evidence suggests CD59 is involved in the development of diabetes and the complication associated with the disease (Golec et al. 2019). One of the primary reasons for organ damage that is commonly observed in diabetes is sustained hyperglycemia leading to the formation of advanced glycation products which affects the function of many proteins, including CD59 (Acosta et al. 2000; Blom 2017). Thus, enhanced complement activation, due to inactivation of CD59 by hyperglycemia, leads to the production of proinflammatory cytokines (Blom 2017). Additionally, CD59 has been implicated in insulin secretion (Krus et al. 2014; Golec et al. 2019). Healthy pancreatic islets, including B cells, express high levels of CD59 and downregulation of CD59 was shown to decrease the ability of rat B-cells to secrete insulin upon stimulation with glucose (Krus et al. 2014). However, it is thought that an intracellular, soluble form of CD59 is involved in insulin release by facilitating insulin granule interactions with the cell membrane through a mechanism involving soluble NSF attachment protein receptor (SNARE proteins) VAMP2 and syntaxin-1 (Krus et al. 2014). VAMP2 localizes to insulin granules, whereas syntaxin-1 is present in the cell surface membrane. The VAMP-2 syntaxin-1 interaction under high-glucose conditions allows the release of insulin to the extracellular environment (Krus et al. 2014). Under high glucose conditions, CD59 may interact with SNARE proteins (Krus et al. 2014). Taken together, CD59 may play a role in diabetes through its role as a complement regulator and involvement in insulin secretion. Further investigation is required to elucidate the role of CD59 in trafficking of insulin granules which perhaps suggests it may be more broadly involved in secretory processes within cells.

1.5.3.3 Multiple Sclerosis

The presence of autoreactive T cells, as well as a lack in their regulation, contribute to the pathogenesis of autoimmune diseases (Reis et al. 2019). The switching of TH 1 to IL-10 secreting Tr1 cells, a process driven by CD46 signaling on T cells, is defective in chronic inflammatory

50

diseases such as multiple sclerosis (MS) (Astier and Hafler 2007; Astier 2008), asthma (Xu et al. 2010) and rheumatoid arthritis (RA) (Cardone et al. 2010). CD55 may also be involved in the pathogenesis of MS through its interaction with CD97 (Visser et al. 2002). One of the defining features of MS is the presence of inflammatory demyelinating lesions axonal loss in the central nervous system (Visser et al. 2002). CD55 is highly expressed on endothelial cells in MS lesions while there is robust expression of CD97 on infiltrating T cells, macrophages and microglia (Visser et al. 2002). Thus, binding of CD55 to CD97 on infiltrating leukocytes may facilitate immune cell activation and migration through the blood brain barrier (Visser et al. 2002). Taken together, CD46 and CD55 appear to play roles in autoimmune disorders that are independent of their traditional functions as complement cascade regulators.

1.5.3.4 Systemic Lupus Erythematosus

The CRPs are thought to be involved in the pathogenesis of systemic lupus erythematosus (SLE) through their regulation of complement activity as well as T cell responses (Ellinghaus et al. 2017; Le Buanec et al. 2011; Alegretti et al. 2012). SLE is a complex rare autoimmune disease characterized by dysregulated cytokine production and impaired self- tolerance which contributes to tissue inflammation (Zharkova et al. 2017). The complement system is involved in the pathogenesis of tissue injury in SLE as tissue deposition of immunoglobulin in SLE causes continued complement activation (Walport and Walport 2002). Reduced expression of CD55 and CD59 on blood cells appears to be associated with disease activity in SLE (Alegretti et al. 2012). However, it has been proposed that both CD55 and CD46 contribute to SLE pathogenesis through T cell regulation.

IFN-γ plays an essential role in the development and severity of systemic autoimmunity (Pollard et al. 2013), and as discussed above, dysregulation of CD46 signaling leading to impaired TH1 contraction (and continued production of IFN-γ) contributes to the disease pathology in RA (Cardone et al. 2010) and MS (Astier and Hafler 2007; Astier 2008). As CD46 mediated signals are required for both TH1 induction and contraction (IFN-γ to IL-10 switching), cell surface expression of CD46 on activated T cells is tightly controlled (Choileain et al. 2011;Choileain and Astier 2012). As observed in several chronic inflammatory diseases, defects in CD46-regulated

TH1 contraction has been identified in patients with SLE (Ellinghaus et al. 2017).

51

Another important factor in T cell activation is the formation of lipid rafts on T cells at the immunological synapse (interface between T cell and an APC) (Kabouridis 2006). T cell costimulatory molecule CD28, and CD55 are both GPI anchored proteins that exist in lipid rafts and their engagement leads to redistribution and clustering at the site of the TCR (Legembre et al. 2006). Conventional CD28 cross-linking leads to the formation of lipid raft clusters, which exclude CD55 and vice versa (Toomey, Cauvi, and Pollard 2014). Additionally, the cytokine profiles of murine CD4+ T cells differ between CD3/CD97 (i.e. CD55 interaction partner) activation compared to costimulation by CD3/CD28, with the latter eliciting a more proinflammatory profile (Toomey, Cauvi, and Pollard 2014). Therefore, given that CD55 levels are decreased on lymphocytes of patients with SLE (Alegretti et al. 2012), it has been proposed that T cell activation by CD28 cross-linking leads to the formation of lipid rafts, which exclude CD55 resulting in proinflammatory cytokine production (IFN- γ) (Toomey, Cauvi, and Pollard 2014).

1.5.3.5 Paroxysmal Nocturnal Hemoglobinuria

Paroxysmal nocturnal hemoglobinuria (PNH) is an ultra-rare acquired blood disorder that is life threatening (Hill et al. 2017). The disease is characterized by the destruction of red blood cells, blood clots and impaired bone marrow function (Hill et al. 2017). PNH arises from a hematopoietic stem cell that acquires a mutation in the PIGA gene whose gene product is essential for the synthesis of GPI (Hill et al. 2017). Consequently, blood cells lacking GPI- anchored proteins CD55 and CD59 are overly susceptible to complement-mediated attack (Hill et al. 2017; Hill, Kelly, and Hillmen 2013). Platelets lacking these CRPs become activated leading to an increased risk of thrombosis (Hill et al. 2017; Hill, Kelly, and Hillmen 2013).

1.5.3.6 Atypical Hemolytic Uremic Syndrome

GWAS studies and advancements in next-generation sequencing have revealed that there are now over 60 disease-associated CD46 mutations (Liszewski and Kemper 2019). Many of the mutations identified have been linked to a rare thrombotic microangiopathic-based disease known as atypical hemolytic uremic syndrome (aHUS), similar to other diseases such as pregnancy-related disorders (preeclampsia, HELLP syndrome) and age-related macular degeneration (AMD)(Liszewski and Atkinson 2015). aHUS is characterized by low levels of circulating blood cells due to their destruction and acute renal failure due to the inability of the

52

kidneys to process and excrete waste products from the blood into the urine (uremia) (Zhang et al. 2017). Most cases of aHUS are genetic and are associated with excess activation or dysregulation of the alternate pathway of complement resulting in harm to “self” cells. Mutations in CD46 are found in 10-20% of patients with aHUS (Liszewski and Atkinson 2015). In the remainder of aHUS cases, mutations occur in complement activators (C3 and factor B) or regulators (factor H and factor I) of the alternative pathway (Liszewski and Atkinson 2015).

1.6 Genetic Interactions

While the increasing discovery of rare disease-causing genes/mutations/variants is a significant advance (Hieter and Boycott 2014), there is a knowledge gap in our understanding of the function of may rare disease-causing genes and consequently, how a mutation in given gene contributes to disease pathogenesis. However, genes don’t function in isolation, a mutation in a single gene can impact the function of many other genes. Indeed, pioneering studies in model systems have shown the effect of one genetic change on the phenotype of an organism greatly depends upon its interaction(s) with other genes in the genome (Costanzo et al. 2010, 2016; Davierwala et al. 2005). Therefore, uncovering the functional relationships, or genetic interactions (GIs) of rare disease genes may be a valuable approach for unraveling their functions.

1.6.1 Defining Genetic Interactions

The simplest GI is between two genes (i.e. digenic interaction) and occurs when a combination of mutations in different genes results in a unexpected phenotype (Costanzo et al. 2011). For simplicity, I will use GI synonymously throughout my thesis with digenic interaction. GIs may be quantified by measuring how a double-mutant phenotype compares in severity to the predicted phenotype that is determined by the combined effects of the single mutants (Costanzo et al. 2013). Any phenotype that can be measured can be used to detect GIs; however, in model systems such as yeast and mammalian cells, cellular fitness is often the phenotype of choice since it can be easily measured and quantified and is reflective of the general state of the cell (Costanzo et al. 2019).

GIs have been studied in various model organisms as a means of identifying functional relationships among genes, with the nature of these relationships depending on the type of GI

53

(Error! Reference source not found.) (Hartman, Garvik, and Hartwell 2001; Byrne et al. 2007; St. Onge et al. 2007; Costanzo et al. 2016). GIs may be broadly classified as either positive or negative (Mani et al. 2008). Positive GIs correspond to interactions between genes in which a double mutant displays a less severe phenotype than the expected effects of combined single mutants and often identifies genes that act in the same pathway (Mani et al. 2008; Costanzo et al. 2016). Positive GIs can be further sub-classified into symmetric, masking and suppressive genetic interactions (Mani et al. 2008; Costanzo et al. 2013). Conversely, negative GIs, also called aggravating, synergistic, or synthetic lethal interactions, correspond to interactions in which a double mutant shows a greater fitness defect than the expected effect of the combined single mutants (Mani et al. 2008). Negative GIs are very useful in identifying genes that are in parallel pathways and converge upon a common cellular function (Costanzo et al. 2010, 2016; Tong et al. 2001). Synthetic lethality is an example of an extreme negative GI whereby the cell can survive a loss of function of either gene individually, but not both. Although rare, synthetic lethal interactions often connect genes with related biological functions and uncover novel functional relationships (Costanzo et al. 2016).

1.6.2 Mapping Genetic Interactions

Given the importance of GIs in understanding diseases phenotypes and gene function, functional genomic efforts in model eukaryotic organisms have shifted from investigating individual gene functions in isolation towards systematic mapping GI networks (Baliga et al. 2017). This approach requires systematic mapping of all pairwise gene mutation combinations in a given organism to a measurable phenotypic outcome such as colony size in yeast or proliferation in cell culture systems (Costanzo et al. 2016; Vizeacoumar et al. 2013). Each GI is categorized based on the observed phenotype of the double mutant and the strength of the GI is quantified. Similarly, a systematic approach may also be used to map chemical-genetic GI where a specific mutant is either hypersensitive or resistant to a compound compared to a wild-type control (Costanzo et al. 2010; Wang et al. 2014). Model systems such as yeast and worms, as well as mammalian cell culture systems, provide a format for mapping GIs systematically through the use of genome- wide collections of defined mutants or genome-wide gene-perturbation systems such as RNA interference (RNAi) or CRISPR/Cas technology (Byrne et al. 2007; Costanzo et al. 2016; Fischer et al. 2015; Hart et al. 2015).

54

Figure 1.10: Graphical representation of quantitative genetic interactions. Wild-type fitness is defined as 1 and each single mutant (a and b) exhibits a fitness defect relative to wild-type. In this example, the fitness of a single mutant in gene a is 0.8. The x-axis is the fitness of the single mutant in gene b and the y-axis is the fitness of the double mutant (ab). The expected fitness of the double mutant based on a multiplicative model (a x b) is 0.8 x b and is plotted as a black line. Negative genetic interactions: A negative GI occurs if the observed fitness of the double mutant is less than expected and can be categorized and synthetic sick or synthetic lethal. Negative GIs often occur between genes that impinge upon a common essential biological function. Positive genetic interactions: A positive GI occurs if the observed fitness of the double mutant is greater than expected and can be classified as symmetric or asymmetric. In symmetric positive GIs, the two single mutants and the double mutant exhibit the same fitness defects (in this example, ab=a=b=0.8). These GIs are frequently observed between genes encoding members of the same non-essential protein complex. In asymmetric positive GIs the single and double mutants differ in fitness and can be further sub classified into masking GIs and suppression. In masking GIs, the fitness of the double mutant is better than expected and is less than or equal to the fitness of the sickest single mutant. Suppression occurs when the fitness of the double mutant is greater than that of the sickest single mutant. In yeast, suppression between loss-of-function (LOF) deletion alleles is often indicative of a negative regulatory relationship between genes a and b. Created with BioRender (Figure adapted from Costanzo et al. 2013).

55

To date, the most comprehensive analysis of mapping GIs has been carried out using the budding yeast (S. cerevisiae) via Synthetic Genetic Array (SGA) analysis and has offered important insights into our understanding of GIs (Costanzo et al. 2010, 2016; Tong et al. 2001; Davierwala et al. 2005). Recently, the first complete global GI network of a yeast cell was completed by screening 23 million double mutants for growth defects, identifying nearly 1 million GIs (Costanzo et al. 2016). This hallmark study revealed the complex functional relationship between genes within a cell and the wealth of information that may be captured by positive and negative GIs. Importantly, findings from this study further demonstrates that grouping genes according to their GI profiles is a powerful and effective way to predict gene function and that genes belonging to similar processes share overlapping sets of both negative and positive GIs.

While yeast studies have established the value of comprehensively mapping GIs for inferring gene function, large-scale GI mapping in human cells remains challenging. The size and complexity of the , as well as the limited availability of efficient and scalable genetic engineering tools, has limited the comprehensive mapping of GIs on a genome-wide scale. The utility of conventional RNAi based screening in cultured cells has been limited by incomplete gene knockdown and confounding off-target effects (Moffat, Reiling, and Sabatini 2007). However, the development of CRISPR-based gene editing has enabled high-throughput genome-wide interrogation of gene function, and overcomes many of the limitations of conventional screening approaches. In CRISPR/Cas9 genome editing, the Cas9 nuclease is directed by a single guide RNA (sgRNA) that is complementary with the target DNA strand in the form of a ribonucleoprotein (RNP) (Cong et al. 2013; Mali et al. 2013). Upon binding to its target sequence, the Cas9-RNP generates a site-specific double-strand break on the target DNA sequence, which is dominantly repaired by error-prone non-homologous end joining (NHEJ) and often results in frameshift indel mutations that abolish target gene function (Cong et al. 2013; Mali et al. 2013). By harnessing the power of CRISPR technology, gene function may now be interrogated on a genome-wide scale in a high-throughput manner by pooled genome-scale libraries of CRISPR gRNAs (Hart et al. 2015; Shalem et al. 2014; Wang et al. 2014). Previous work by the Moffat lab and others have demonstrated that genome-wide CRISPR screens offer significant advantages over pooled-library shRNA screens in that they are more sensitive and specific (Hart et al. 2015, 2014).

56

The adaptation of CRISPR/Cas9 technology to mammalian cell lines has enabled genome-wide fitness screens which allow for the systematic identification of ‘‘core’’ and ‘‘context-specific’’ fitness genes (Blomen et al. 2015; Hart et al. 2015; Wang et al. 2014, 2015). Core fitness genes are genes that affect cell fitness across multiple cell lines whereas context-specific fitness genes are genes that are essential in a given genetic or environmental context. Identifying “context- specific” fitness genes may yield therapeutic targets for specific human diseases. For example, synthetic-lethal GIs enable rational design strategies for certain cancer types, whereas identifying positive genetic interactions provides molecular targets whose inhibition could ameliorate the consequences of a particular genetic mutation. Previous work in the Moffat lab led to the identification of a set of core fitness genes across multiple human cell lines, as well as novel context-specific vulnerabilities in different cancer lines (Hart et al. 2015). Furthermore, loss-of- function (LOF) CRISPR screens carried out in cells harboring a disease-relevant mutation (i.e. not a core fitness gene) offers an entry point to gain biological insight into a gene’s function, as well as the global consequences of specific mutations.

1.6.3 Genetic Interactions and Implications in Disease Phenotypes

Understanding how genetic variation contributes to inherited phenotypes and human disease is one of the greatest puzzles of our time (Costanzo et al. 2019; Ritchie and Van Steen 2018). Genome-wide association studies (GWAS) have discovered many genetic variants associated with complex diseases and traits, however, these variants typically represent a only small fraction of disease heritability (Manolio et al. 2009; Zuk et al. 2012, 2014). While several reasons have been proposed to explain the is discrepancy, including the presence of rare genetic variants, genetic interactions may account for a significant component of the “missing heritability” (Zuk et al. 2012). Indeed, GIs are known to affect the susceptibility and/or onset of several complex diseases (Eichler et al. 2010; Zuk et al. 2012). While detection of significant GIs between individual gene-gene pairs in GWAS datasets remains challenging due to a lack of statistical power, properties of the global yeast GI network offers important insights (Costanzo et al. 2016). The yeast GI network is highly organized with GIs clustering into discrete functional modules where coherent sets of negative GIs often occur between genes in two pathways or complexes (between pathway module) or within an essential pathway and complex (within pathway module) (Costanzo et al. 2016). These insights were recently exploited to identify

57

pathway-level GIs from human GWAS studies in breast cancer, highlighting various pathways as modifiers of breast cancer risk (Wang et al. 2017).

While GWAS studies have uncovered new associations for common diseases and complex phenotypes, this approach has not been amenable to rare diseases due to the rarity of individual diseases and the lack of large cohorts. Furthermore, while “missing heritability” has largely been discussed in the context of common diseases, the majority of rare diseases also face this problem (Maroilley and Tarailo-Graovac 2019). For example, phenotypic heterogeneity, the diverse phenotypes associated with variants of the same gene, is observed in various rare diseases (Maroilley and Tarailo-Graovac 2019), even in individuals with the same causal variant (Tahsin Hassan Rahit and Tarailo-Graovac 2020). Indeed, in NGLY1-deficiency, specific mutations are associated with more severe phenotypes (Enns et al. 2014; Lam et al. 2017) and diverse phenotypes have been reported in patients with the same mutation (Enns et al. 2014). Conversely, there is evidence of individuals carrying mutations that have been associated with severe Mendelian childhood diseases that do not present with the disease (Chen et al. 2016). Diverse phenotypes observed with the same causal variant and the discovery of “disease resilient” individuals may be due to the presence of additional mutations in genetic modifiers. Genetic modifiers are genetic variants that can modify the phenotypic outcome of the primary- disease causing variant, by enhancement or suppression, but may not be disease causing themselves (Tahsin Hassan Rahit and Tarailo-Graovac 2020). Taken together, these findings highlight the complexity of the genotype-to-phenotype relationship and the importance of GIs underlying both common and rare disease phenotypes.

1.6.4 Concluding Statement

To date, the function of many rare disease-associated genes remains poorly understood (Boycott et al. 2017). Consequently, FDA approved therapeutic interventions are available for less than 5% of rare diseases (Kaufmann, Pariser, and Austin 2018). The utility of using GIs to interrogate gene function, coupled with CRISPR/Cas9 screening platforms, may provide an opportunity to begin unravelling the function of rare disease-associated genes on a genome-wide scale. Furthermore, lessons from yeast studies have revealed that genes belonging to similar processes share overlapping sets of GIs (Costanzo et al. 2016). Therefore, mapping GIs of genes known to function in the same pathway as rare disease-associated gene may provide greater insight into is

58

function. Lastly, uncovering rare disease gene function through GI may offer valuable insight into disease pathogenesis and help direct future therapeutic efforts.

1.7 Overview of Doctoral Research Project

The overarching goal of my doctoral research has been to gain functional insight into NGLY1 and the CRP-encoding genes CD46, CD55, and CD59 by mapping their GIs in human HAP1 cell lines. HAP1 cells were used as a model to explore GIs in human cells because i) HAP1 cells are nearly haploid, thereby facilitating successful CRISPR-mediated gene knockout ii) the genome of HAP1 cells has been sequenced (Essletzbichler et al. 2014) iii) HAP1 cells can sustain growth at single cell dilutions, enabling the generation of clonal knockout lines iv) HAP1 cells are amenable to multiple gene knockouts in the same cell and lastly v) the Moffat lab has access to large collection of isogenic HAP1 cell lines which harbor a single mutation in a gene of interest.

I employed pooled genome-wide CRISPR/Cas9 screens to map the negative genetic interactions of NGLY1 pathway members (NGLY1, NFE2L1, DDI2) and the genes encoding the complement regulatory proteins (CRPs; CD46, CD55, CD59). I identified a novel functional link between the CRPs and members of the NGLY1 genetic pathway, that together may regulate opposing aspects of intracellular cholesterol trafficking. In chapter two, I describe a novel network of shared genetic interactions between NGLY1, NFE2L1, DDI2 and several genes involved in the secretory pathway, and in particular, members of the Golgi-associated retrograde protein (GARP) complex. Through an NFE2L1-driven transcriptional program, the NGLY1 genetic pathway likely regulates intracellular vesicle and cholesterol transport as well as immunological pathways. In chapter three, I describe a genetic relationship between the CRPs and the NGLY1 genetic pathway which revealed a possible complement-independent role for the CRPs in regulating intracellular cholesterol transport. Furthermore, the CRPs may regulate NFE2L1 turnover and processing during cholesterol-induced stress.

Taken together, these studies offer critical insight into the genetic underpinnings of NGLY1- deficiency and complement-independent functions of the CRPs. Furthermore, as the function of many rare disease-associated genes have not yet been unraveled, this work offers powerful framework that may be exploited to gain functional insight into unknown and poorly characterized genes which may help direct future therapeutic efforts for rare diseases. 59

Chapter 2 Genome-Wide Genetic Interaction CRISPR Screens Offer New Insights into the Genetic Dependencies Underlying NGLY1-deficiency

The work discussed in this chapter was a collaborative effort between myself, Dr. Ashwin Seetharaman, and other members of the Moffat and Myers groups.

I carried out the experiments described in Figures 2.2, 2.4a, 2.5, 2.6, 2.7, 2.8, and 2.9 in equal contribution with Dr. Ashwin Seetharaman (former postdoctoral fellow in the Moffat and Boone labs). Dr. Ashwin Seetharaman carried out several of the CRISPR screens and the bortezomib sensitivity assays in Figure 2.1b and 2.2d. Dr. Traver Hart (former research associate in the Moffat lab) analyzed the bortezomib screen in Figure 2.2a. Dr. Max Billmann (postdoctoral fellow in the Myers lab) analyzed all other screens in this study and generated the corresponding plots shown in Figures 2.2a,b, 2.3, and 2.4. Dr. Kevin Brown (a research associate in the Moffat lab) analyzed the RNA sequencing data in Figure 2.6 and 2.7. Please refer to the figure legends for further details on the specific contribution made by each individual towards the experiment in question.

60

Genome-Wide Genetic Interaction CRISPR Screens Offer New Insights into the Genetic Dependencies Underlying NGLY1-deficiency 2.1 Abstract

NGLY1 encodes a deglycosylating enzyme that is critical for human development. Its loss-of-function results in a rare genetic disorder called NGLY1-deficiency. How the loss of NGLY1 function leads to the diverse clinical phenotypes observed in NGLY1-deficient patients is poorly understood. In this chapter I describe how I employed pooled genome-scale CRISPR- Cas9 screens to map the negative genetic interactions of NGLY1 and its related pathway members, NFE2L1and DDI2, in human cells. I have uncovered a novel network of shared genetic interactions between NGLY1, NFE2L1, DDI2 and several genes involved in the secretory pathway, particularly members of the GARP complex VPS52 and VPS54. I demonstrate that an NGLY1 genetic pathway likely regulates intracellular vesicle and cholesterol transport. Additionally, I uncovered a shared set of genes involved in immunological and secretory pathway functions that are likely transcriptionally regulated by the NGLY1 pathway. Overall, this study highlights the utility of genetic interactions to shed light on the genetic dependencies underlying rare disorders.

61

2.2 Introduction

GIs provide a window into the underlying mechanistic connections between genes and their respective genetic pathways. Synthetic lethality is an extreme form of a negative GI whereby a simultaneous LOF mutation in two different genes results in an inviable phenotype, while both mutations on their own are viable (Baryshnikova et al. 2010; Costanzo et al. 2016). Synthetic lethal GIs often occur between genes that act within parallel pathways impinging upon a common essential function and therefore can provide a powerful window into elucidating gene function. Conversely, a positive GI is identified when perturbation of two genes results in improved cellular fitness that is significantly greater than the effect of a single gene perturbation (Baryshnikova et al. 2010). GI profiles of genes have helped define genes within specific cellular pathways, illuminate hidden inter-dependencies between different cellular processes, and provide a powerful way to predict gene function (Costanzo et al. 2010, 2016). Here, I along with my collaborators, employed genome-wide CRISPR-based GI screens to interrogate the biological role of a rare disease-associated gene called NGLY1 and to help delineate genetic dependencies underlying NGLY1-deficiency.

NGLY1/N-glycanase I is a well-conserved cytosolic deglycosylating enzyme that mediates non- lysosomal degradation of N-glycoproteins via the ERAD pathway (Heeley and Shinawi 2015; Enns et al. 2014). NGLY1 function is critical for human development and deleterious mutations in NGLY1 result in an ultra-rare inherited genetic disorder characterized by a wide spectrum of clinical manifestations, including global developmental delay, seizures, liver dysfunction, microcephaly, alacrima, hypotonia, diminished reflexes and peripheral neuropathy (Heeley and Shinawi 2015; Enns et al. 2014; Tickotsky-Moskovitz 2015; Caglayan et al. 2015). Previous studies also provided evidence that NGLY1 function is integral to a variety of cellular processes including ERAD, proteasome and mitochondrial homeostasis, inflammation, antigen presentation, developmental signaling, aquaporin expression and lipid metabolism (Suzuki 2007; Suzuki, Huang, and Fujihira 2016; Altrich-VanLith et al. 2006; Galeone et al. 2017; Tomlin et al. 2017; Yang et al. 2018; Kong et al. 2018; Tambe, Ng, and Freeze 2019; Fujihira et al. 2019). However, to date an unbiased global investigation of NGLY1 gene function in any organism has not been carried out.

62

A recent study using the nematode C. elegans model revealed that PNG-1, the worm ortholog of NGLY1, functions with two well-known endoplasmic reticulum (ER) stress regulators; the aspartic acid protease DDI-1 (human DDI1/DDI2), and the transcription regulator SKN-1 (human NFE2L1 and/or NFE2L2). NFE2L1 (often referred to as NRF1) is an ER-localized transcription factor constitutively targeted for proteasomal degradation through ERAD (Lehrbach and Ruvkun 2016). The prevailing perspective is that upon induction of proteotoxic stress, NFE2L1 is stabilized, then cleaved and subsequently de-glycosylated. Following de- glycosylation, the cleaved cytoplasmic portion of NFE2L1 localizes to the nucleus and initiates transcription of ARE-dependent target genes to mitigate proteotoxic stress by up-regulating proteasome-encoding genes (Radhakrishnan et al. 2010). Studies have identified DDI2 as the enzyme primarily responsible for cleaving NFE2L1 in response to proteotoxic stress (Koizumi et al. 2016). Importantly, it has been shown that both cleavage by DDI2 and its subsequent de- glycosylation by NGLY1 are critical for NFE2L1-dependent transcriptional changes (Koizumi et al. 2016; Tomlin et al. 2017; Radhakrishnan, den Besten, and Deshaies 2014; Lehrbach and Ruvkun 2016).

NFE2L1 has also recently been identified as a cholesterol sensor in the ER (Widenmaier et al. 2017). Changes in NFE2L1 turnover, localization and processing were found to be integral in orchestrating the cellular response to proteotoxic and cholesterol-induced stress (Widenmaier et al. 2017). These observations are significant as the mechanism for how mammalian cells establish a cholesterol gradient from ‘high’ concentrations in the plasma membrane to ‘low’ concentrations in the ER is largely unknown.

Although NFE2L1 has been linked to cholesterol sensing (Widenmaier et al. 2017), regulatory proteins of NFE2L1 such as NGLY1 and DDI2, have thus far not been connected to cholesterol sensing in cells. In this study, I describe the development and utility of a computational scoring pipeline for defining the global negative GIs of the NGLY1 gene in the human HAP1 cell line and the GIs that are shared between NGLY1, NFE2L1 and DDI2. The results from this study suggest that NGLY1, NFE2L1 and DDI2 likely function in the same genetic pathway to regulate multiple aspects of the secretory pathway, including intracellular vesicle and cholesterol transport. Finally, perturbations in the NGLY1 genetic pathway result in the downregulation of a set of genes involved in immunological pathways and cholesterol homeostasis.

63

Overall, this study not only extends our understanding of the genetic dependencies underlying NGLY1-deficiency, but also offers a powerful framework that may be employed to investigate the function of other rare disease-associated genes in human cells and direct future therapeutic efforts.

2.3 Results

2.3.1 An unbiased genome-wide genetic screen for sensitizers of bortezomib helps define an NGLY1 genetic pathway in human cells

Previous studies employing the budding yeast S. cerevisiae have revealed that genes that regulate similar biological processes tend to display similar GI profiles (Baryshnikova et al. 2010)(Costanzo et al. 2016). Therefore, we reasoned that mapping GIs with NGLY1 to gain a broader understanding of NGLY1 molecular function in human cells may offer new insights into its biological role and how it contributes to NGLY1-deficiency. Lehrbach et al. 2016, reported the first description of a genetic pathway in C. elegans comprising png-1 (ortholog of mammalian NGLY1), skn-1a (ortholog of mammalian NFE2L1/NFE2L2) and ddi-1 (ortholog of mammalian DDI2), to orchestrate the cellular response to proteotoxic stress (Lehrbach and Ruvkun 2016). Subsequently, a gene essentiality profile study conducted across 14 distinct human leukemia cell lines identified a correlated essentiality between NFE2L1, NGLY1 and DDI2, suggesting the existence of functional genetic relationships between these three genes (T. Wang et al. 2017). In addition, other studies have independently corroborated that NGLY1, NFE2L1 and DDI2 are integral to orchestrating a “proteasome bounce-back response” in mammalian cells when challenged with a proteasome inhibitor (Koizumi et al. 2016; Tomlin et al. 2017), such that a LOF in any of these three genes results in heightened sensitivity to proteasome inhibition. Cleavage of NFE2L1 by DDI2 and de-glycosylation by NGLY1 are critical for the NFE2L1-dependent transcription of genes encoding proteasome subunits (Radhakrishnan et al. 2010). Taken together, it is conceivable that a genetic pathway comprised of NGLY1, NFE2L1 and DDI2 may operate in human cells to mitigate proteotoxic stress.

Towards conclusively determining whether a genetic pathway comprising NGLY1, NFE2L1 and DDI2 exists in human cells, Dr. Ashwin Seetharaman (former postdoctoral fellow in the Moffat and Boone labs) performed an unbiased genome-wide CRISPR screen in human HAP1 cells that were treated with bortezomib (Velcade), a chemical inhibitor of the 20S catalytic subunit of the

64

proteasome that is used to treat multiple myeloma (Kubiczkova et al. 2014). If NGLY1, NFE2L1 and DDI2 indeed function within the same genetic pathway in human cells to mitigate proteotoxic stress, then all three genes should be found among the top bortezomib sensitizing hits.

To carry out the screen, a HAP1 cell line stably expressing Cas9 was used (HAP-Cas9C1) and infected with the first-generation Toronto Gene Knockout (TKO) 90k guide genome-wide CRISPR library (TKOv1), in order to perform a systematic gene knockout screen similar to what has previously been reported (Hart et al. 2015). Briefly, following puromycin selection of TKOv1 library-infected cells at ~200-fold library coverage, cells were split into either non-drug treated control or bortezomib-treated groups. To identify sensitizers of bortezomib treatment, cells were cultured at an IC50 dose of bortezomib (Figure 2.1b). Cells were passaged (and treated with bortezomib in treatment group) every three days for approximately 20 doublings. Genomic DNA from the initial (T0) and final (T18) time points was collected and subjected to deep sequencing (Figure 2.1a). Single guide RNAs (sgRNAs) targeting genes that sensitize cells to bortezomib are expected to drop out and deplete from the collective pool of library-infected cells. Conversely, genes that mediate resistance to bortezomib when knocked out, are expected to be enriched in the final pool of cells. Fold-changes in sgRNA abundance between the bortezomib treated and untreated groups were identified computationally using DrugZ-scoring (Colic et al. 2019).

Through this screen, we discovered that LOF in genes encoding 19S proteasome subunits confer resistance, while LOF of the 20S subunits sensitized cells to bortezomib (Figure 2.2a). These findings agree with previous reports from other groups (Acosta-Alvear et al. 2015; Tsvetkov et al. 2015). Genetic pathways such as glycan biosynthesis and mTOR signaling, which have been linked to synergy with bortezomib by other groups were also identified (Figure 2.2b)(Wlodkowic et al. 2009; Hill et al. 2018). For example, ALG3 and ALG12, found among the top sensitizing hits (Table 2.1) encode glycosyltransferases that regulate lipid-linked oligosaccharide biosynthesis (Freeze 2006). Previous studies have shown that perturbation of N-glycosylation by tunicamycin, a global inhibitor of N-linked glycosylation, synergizes with bortezomib in killing lymphoma and pancreatic cancer cells in vitro (Wlodkowic et al. 2009). Lastly, two of the top predicted sensitizers of bortezomib included NCOR1 and HDAC3 which encode a nuclear receptor co-repressor and a histone deacetylase, that together form a corepressor complex that 65

Figure 2.1: Genome-wide CRISPR screen schematic to identify genetic pathways that mediate sensitivity and resistance to bortezomib. a. Schematic of the genome-wide CRISPR-Cas9 screening pipeline to identify sensitizing and resistance genes of bortezomib in human HAP1 cells. b. Estimation of IC50 concentrations of HAP1-Cas9C1 cells when challenged with increasing concentrations of bortezomib. Both experiments were performed by Dr. Ashwin Seetharaman.

66

regulates a wide range of cellular processes through transcriptional repression (Karagianni and Wong 2007). HDAC3 has also previously been reported as sensitizer to bortezomib (Minami et al. 2014) with HDAC inhibitors currently being used to treat relapsed, refractory multiple myeloma (Raedler 2016). Interestingly, the activity of the NCoR1-HDAC3 complex likely plays a role in the disease pathogenesis of the neurological disease Rett syndrome, which arises due to mutations in the MECP2 gene (Kyle et al. 2016). MeCP2 coordinates lipid metabolism by anchoring the NCoR1-HDAC3 complex to lipogenesis target genes, therefore mutations in MECP2 may contribute to disease phenotypes as a result of impaired lipid homeostasis (Kyle et al. 2016). Taken together, these findings highlight that the chemical-genetic relationships uncovered in our unbiased bortezomib screen, recapitulate findings that have previously been reported using a variety of different experimental methods, and provide a strong benchmark for the quality of the screen.

As predicted, NGLY1, NFE2L1 and DDI2 were among the top bortezomib-sensitizing hits (Figure 2.2a). To validate these hits, CRISPR sgRNAs were designed by Dr. Seetharaman targeting each of these three genes. gRNAs were infected into HAP1 cells and other cell lines including the osteosarcoma cell line U2OS and the colorectal carcinoma cell line HCT116. In each case, he found that perturbing NGLY1, NFE2L1 or DDI2 function confers a significant increase in sensitivity to bortezomib (Figure 2.2d). These observations suggest that the increase in sensitivity to bortezomib following perturbation of NGLY1, NFE2L1 or DDI2 activity is not restricted to the HAP1 background and may generally be true of most human cell lines (Figure 2.2d).These findings are consistent with previous reports in other model systems and suggest that a genetic pathway comprising NGLY1, NFE2L1 and DDI2 may function within human cells.

2.3.2 Mapping global NGLY1 GIs in human cells

Given the power of GIs for revealing functional relationships between genes and genetic pathways, I sought to map negative GI profiles for NGLY1 using human HAP1 cells. In collaboration with Drs. Ashwin Seetharaman and Max Billmann, we performed lentiviral-based pooled genome-wide loss-of-function screens in co-isogenic clonal HAP1 lines either wild-type (parental) or deficient (NGLY1Δ) in NGLY1 (see Methods for details). Briefly, we performed pooled genome-wide CRISPR-Cas9 gRNA screens using the sequence optimized 71k guide TKOv3 library in both the NGLY1Δ query cell line and the wild-type (WT) HAP1 cell lines.

67

Similar to what was previously described, we compared the relative gRNA abundance of individual gRNAs between T0 and T18 time points for each screen. The relative abundance of gRNAs target each of the genes in WT cells provides an estimate of single mutant fitness whereas the abundance of gRNAs in the NGLY1Δ line provides an estimate of double mutant fitness. To robustly identify genetic dependencies of NGLY1, three independent genome-wide CRISPR screens in NGLY1Δ cells were performed.

68

Figure 2.2: Genome-wide CRISPR screen identifies members of the NGLY1 genetic pathway as sensitizers to bortezomib. a. Sensitizing and resistance genes after bortezomib-mediated inhibition of the 19S proteasome. Shown are 16318 genes screened using the TKOv1 library in HAP1 cells with and without bortezomib treatment. Z-scores were computed using DrugZ. Members of the 19S (blue) and 20S (green) proteasome are labeled as well as the top sensitizer DDI2, NGLY1 and NFE2L1 (shown in orange). b. Pathways that sensitize or convey resistance to bortezomib treatment. 1237 pathways as annotated in KEGG, REACTOME or BIOCARTA were tested (two-sided Wilcoxon rank-sum test, BH-corrected). Dot size corresponds to number of pathway members. c. Illustration of all gene z-scores for the mTOR signaling pathway as annotated in REACTOME. d. Validation of NGLY1, NFE2L1 and DDI2-mediated sensitivity in three distinct human cell lines; HAP1, HCT116 and U2OS. Cells were infected with lentivirus containing sgRNAs targeting lacZ, NGLY1, NFE2L1 or DDI2 and were treated with different doses of bortezomib. Relative cell viability for each of the knockout cell populations was estimated by Alamar Blue. Data was normalized to lacZ controls in HAP1-WT cells and represented as mean + SD. * p < 0.05, ** p < 0.01,*** p<0.001 unpaired two-tailed Student’s t-test. Dr. Ashwin Seetharaman carried out the bortezomib screen in 2.1a which was analyzed by Dr. Traver Hart. I carried out screen analyses to determine enriched biologically relevant pathways that were used to generate 2.1a,b. Dr. Max Billmann generated the plots in 2a,b and performed the enrichment analyses in 2b. Dr. Ashwin Seetharaman generated the knockout constructs and carried out the bortezomib assays in 2d.

69

Table 2.1 Top 25 predicted negative genetic interactions of bortezomib

Gene Z Score FDR DDI2 -11.16 5.44E-25 NGLY1 -11.03 1.10E-24 NCOR1 -10.38 8.60E-22 UBE2A -9.45 6.82E-18 HDAC3 -9.30 2.21E-17 ALG3 -9.23 3.53E-17 MTF1 -9.10 1.02E-16 ALG12 -8.83 1.09E-15 CTR9 -8.40 4.14E-14 CXXC1 -8.17 2.51E-13 DOT1L -8.14 3.05E-13 DERL2 -8.10 3.80E-13 BIRC6 -7.85 2.54E-12 G6PD -7.74 5.80E-12 MNT -7.56 2.17E-11 KCTD5 -7.23 2.38E-10 GPS2 -7.18 3.27E-10 SLC25A1 -7.09 6.12E-10 TSC2 -7.05 7.63E-10 PSMG1 -6.99 1.11E-09 NFE2L1 -6.95 1.41E-09 FLCN -6.90 1.98E-09 FBXO42 -6.86 2.42E-09 SZT2 -6.74 5.44E-09 HECTD1 -6.72 5.76E-09

70

To identify negative GIs of NGLY1, Dr. Billmann from the Myers group at the University of Minnesota developed a quantitative GI (qGI) score. The qGI measures the strength and significance of GIs by comparing the relative abundance of gRNAs in a given query mutant cell line (double mutant) to the abundance of gRNAs targeting the corresponding genes to the control WT cell line. Negative GIs are identified as genes whose corresponding gRNAs display significantly decreased abundance in the NGYL Δ background compared to the control WT cell line.

We observed that CRISPR sgRNAs targeting core essential genes (CEGs), genes that are essential across all cell lines dropped out of the pool reliably, allowing to us separate those genes from a reference set of non-essential genes (NEGs) (Figure 2.3a). Moreover, genome-wide fitness defects, as measured by a log2 fold-change (LFC) between initial and endpoint sgRNA abundance, was reproducible between the three NGLY1Δ screens (Figure 2.3b-d).

To evaluate the quality of the screens and identify the top negative GI partners of NGLY1, Dr. Billmann developed an algorithm to quantify the agreement of each GI between a pair of screens. First, GI scores for each gene are scaled using the global, per-screen GI score mean and standard deviation. Scaled values are then multiplied between a pair of screens providing a gene-level agreement score. For each gene, this score is aggregated across the three NGLY1 knockout screen pairs to form the aggregated agreement score (AAS). Genetic pathways and protein complex identities for the GI hits were derived through CORUM 3.0, KEGG and REACTOME databases respectively and the AAS ranking was tested using a one-sided Wilcoxon rank-sum test with multiple hypothesis correction. The pathways and complexes that register higher than expected AAS ranks at an FDR of 10% and a negative protein complex or pathway GI score are shown in Figure 2.4a.

The above analysis revealed that a significant fraction of NGLY1’s negative GIs are with several genes that comprise the pyruvate dehydrogenase complex (PDC), which is integral to regulating mitochondrial homeostasis and cellular metabolism via the conversion of pyruvate to acetyl-CoA in the TCA cycle (Ng and Tang 2014; Birsoy et al. 2015)(Figure 2.4a, e). Previous studies have highlighted a key role for NGLY1 in coordinating various aspects of mitochondrial function across multiple model systems (Yang et al. 2018; Kong et al. 2018) and suggest that NGLY1

71

Figure 2.3: Analysis of genome-wide CRISPR/Cas9 screening for genetic interactions of NGLY1. a. Precision-TP (true positive) curves showing separation of 684 gold-standard core essential (defined in Hart et al., 2017) and non-essential genes using gene-level guide drop-out (log-fold change) measurements across 5 genome- wide screens. Three replicated NGLY1∆ screens (blue), DDI2∆ (violet) and NFE2L1∆ (orange) screens. b. Replication of gene-level drop-out as a measure of fitness effects of the three NGLY1 screens. All three pairwise comparisons and the corresponding Pearson correlation coefficient are shown. c. Comparison of fitness scores of 17,804 genes in NGLY1∆ and HAP1-WT cells upon pooled CRISPR/Cas9 screening using the TKOv3 guide RNA (gRNA) library. Fitness scores for each gene are derived from the log2 FC values for four sequence-independent gRNAs. Fitness scores in HAP1-WT cells represent mean-summarized scores from four genome-wide screens. Less severe fitness defects in NGLY1∆ cells identify positive (yellow) and more severe fitness defects negative (blue) genetic interactions (GI) with NGLY1. Dot size corresponds to amplitude and FDR of a quantitative GI score. The Null-model defining the expected, non-interacting phenotype of a gene was inferred from weighted contrasts between NGLY1∆ and each of the four HAP1-WT screens using a stabilized loess regression. d. Replication of genetic interactions across the three NGLY1∆ screens. All three pairwise comparisons and the corresponding Pearson correlation coefficient are shown. I carried out two screens (NGLY1, NFE2L1) and Dr. Ashwin Seetharaman carried out the others. Dr. Max Billmann carried out the analyses shown in the figure and generated the plots.

72

may play a role in oxidative phosphorylation (OXPHOS) (Kong et al. 2018). Mutations in the PDC, such as in the rare disease PDC deficiency, impairs the mitochondrial oxidation of pyruvate and promotes its reduction to lactate in the cytoplasm causing lactic acidosis (Patel et al. 2012). Therefore, the combined loss of NGLY1 and members of the PDC may result in a negative GI by impairing cellular metabolism.

We also observed that many of NGLY1’s strongest negative GIs occur with genes that function in genetic pathways/complexes that are known to regulate various aspects of the secretory pathway (Figure 2.4b-d). These include members of the Conserved Oligomeric Golgi (COG) complex, which regulates vesicular routes of protein substrate trafficking in the Golgi (Smith and Lupashin 2008), several genes that are predicted to function within the GPI anchor biosynthesis pathway (phosphatidylinositol glycan classes; PIGs) and members of the Golgi-Associated Retrograde Protein (GARP) complex that regulate substrate trafficking between the ER and Golgi (Wei et al. 2017).

2.3.3 NGLY1, NFE2L1, and DDI2 share negative GIs with the GARP complex

Given that NGLY1, NFE2L1 and DDI2 likely function within a common pathway to defend against proteotoxic stress (Figure 2.1a,d), we also sought to identify the shared negative GI partners of NFE2L1 and DDI2 to broaden our understanding of the genetic dependencies underlying NGLY1-deficiency.

Towards this goal and in collaboration with Dr. Seetharaman, I screened clonal knockout HAP1 cell line of NFE2L1 (NFE2L1Δ) and DDI2 (DDI2Δ) and mapped their negative GIs with the intent of identifying genetic pathways or complexes that are shared with NGLY1. Mapping these genetic pathways should reveal biological functions that are likely coordinately regulated by the NGLY1 genetic pathway (Figure 2.4b-e). Consistent with the idea that NGLY1, NFE2L1 and DDI2 act within the same genetic pathway, we did not find NFE2L1 or DDI2 among NGLY1’s negative genetic interactions or vice-versa. Interestingly, some of the strongest negative GIs that were shared across the NFE2L1, DDI2 and the three NGLY1 screens occurred with members of the GARP complex (Figure 2.4b). The GARP complex is comprised of four subunits encoded by the VPS51, VPS52, VPS53 and VPS54 genes. Together, the four subunits of the GARP contribute to the functioning of the secretory pathway by capturing endosome-derived transport vesicles and promoting their fusion to the trans-Golgi network (TGN) (Bonifacino and Hierro 2011). 73

74

Figure 2.4: Mapping genetic interaction partners of NGLY1, NFE2L1 and DDI2 in human HAP1 cells. a. Protein complex and pathway genetic interaction (GI) score and reproducibility highlights the GARP complex as a major negative GI partner of NGLY1. The protein complex GI score represents the mean GI score across the three NGLY1-knockout screens and respective complex members. Per screen pair, an agreement score was computed from the product of two GI scores for each gene pair after adjusting for mean and standard deviation. The aggregated agreement score (AAS) represents the sum of all three possible pairwise agreement scores. For each of the 2916 CORUM 3.0 complexes (Giurgiu et al. 2019) and KEGG and REACTOME pathway standards, higher than background AAS ranking was tested using a one-sided Wilcoxon rank-sum test with BH-correction. Complexes are labeled if they have higher than expected AAS ranks at an FDR of 10% and a negative protein complex or pathway GI score. (b-e) Cluster plots depicting the GI score for GARP, COG, GPI biosynthesis genes and PDH complex members across the NGLY1, NFE2L1 and DDI2 CRISPR screens. f. Shown is a quantification of the relative cell viability in HAP1-WT, NGLY1∆, NFE2L1∆ and DDI2∆ clonal knockout HAP1-lines targeted by CRISPR/Cas9 gRNAs against either AAVS1 or VPS52 and VPS54 GARP complex components. g. Quantification of the relative cell viability in VPS52∆ and VPS54∆ clonal knockout HAP1 cell lines targeted by CRISPR/Cas9 gRNAs against either AAVS1 or NGLY1, NFE2L1 or DDI2. Data was normalized to AAVS1 controls in WT HAP1 cells and represented as mean + SD. * p < 0.05, ** p < 0.01, *** p<0.001 unpaired two-tailed Student’s t-test. Cell viability was measured by Alamar Blue. I carried out two screens (NGLY1, NFE2L1) and Dr. Ashwin Seetharaman carried out the others. Dr. Max Billmann analyzed the screens and generated plots a-e. I analyzed the screens in equal contribution with Dr. Ashwin Seetharaman, I analyzed the screens to determine biologically relevant pathways in b-e and carried out the validations in f,g.

75

Recently, the GARP complex has been shown to regulate targeting of the cholesterol-transport protein NPC2 to the lysosomes and facilitating intracellular cholesterol transport (Wei et al. 2017). Impairment of NPC2 function has been shown to result in Niemann-Pick disease type C2 in humans (Vanier and Millat 2003). This finding is interesting given the recent report identifying NFE2L1 as a sensor of cholesterol-induced stress in mammalian cells (Widenmaier et al. 2017). I was therefore curious to further validate the GIs between members of the GARP complex and NGLY1, NFE2L1 and DDI2. Given that loss of VPS52 and VPS54 are the two strongest predicted negative GIs that are shared across all of the NGLY1 pathway knockout backgrounds, we investigated whether the loss of these GARP components in NGLY1Δ, NFE2L1Δ and DDI2Δ cells result in decreased cellular fitness compared to HAP1-WT cells.

To confirm this idea, NGLY1Δ, NFE2L1Δ, DDI2Δ and HAP1-WT cell lines were transduced with CRISPR/Cas9-sgRNAs targeting VPS52, VPS54 or AAVS1 as a control. Conversely, clonal knockout HAP1 VPS52 (VPS52Δ) and VPS54 (VPS54Δ) knockout cells were transduced with CRISPR/Cas9-sgRNAs targeting either NGLY1, NFE2L1 or DDI2. Consistent with the screen predictions, we found that in both scenarios, a double knockout comprising any NGLY1 pathway member, with either VPS52 or VPS54, resulted in significant cell proliferation defects (Figure 2.4f,g). In conclusion, these results validate the efficacy of the computational pipeline developed by Dr. Billmann for scoring biologically relevant GIs. Furthermore, these findings predict a novel role for the NGLY1 genetic pathway in regulating the secretory processes with respect to coordinating intracellular vesicle and substrate trafficking between the ER and Golgi in human cells.

2.3.4 Nuclear import of cleaved-NFE2L1 requires the activity of NGLY1

A recent study highlighted a novel dual role for NFE2L1 in regulating the cellular response to proteotoxic and cholesterol-induced stress (Widenmaier et al. 2017). In this study, it was shown that exposure to proteotoxic or excess cholesterol-induced stress caused changes in NFE2L1 processing and turnover, which were found to be critical for orchestrating the cellular response to proteotoxic and cholesterol-induced stress (Widenmaier et al. 2017). How excess cholesterol and proteotoxic stress affect NFE2L1 processing and turnover in human cells lacking NGLY1 or DDI2 is not well understood. Our HAP1 model system provides an opportunity to explore this avenue.

76

HAP1-WT, NGLY1Δ, NFE2L1Δ and DDI2Δ cells were treated with a proteotoxic stress-inducing agent, epoxomycin, or excess cholesterol over a 12h period as previously reported (Widenmaier et al. 2017). Following treatment, protein lysates were prepared from each treatment background and NFE2L1 protein bands were visualized by western blot. From these experiments, it was clear that treating cells with epoxomycin resulted in an increase in the cleaved form of NFE2L1 in HAP1-WT and NGLY1Δ backgrounds (Figure 2.5a). Furthermore, given that DDI2 is the key enzyme responsible for NFE2L1 cleavage, it is not surprising that we observed incomplete processing of NFE2L1 in DDI2Δ cells (Figure 2.5a). These findings are consistent with the predicted roles for these gene products (Lehrbach and Ruvkun 2016; Radhakrishnan et al. 2010; Koizumi et al. 2016; Tomlin et al. 2017). Treatment with cholesterol, on the other hand, resulted in the accumulation of the uncleaved full-length form of NFE2L1 in parental HAP1 cells, as well as the NGLY1 and DDI2 cell lines (Figure 2.5b).

We next investigated the nuclear localization of cleaved-NFE2L1 in response to proteotoxic stress via subcellular protein fractionation analysis. Consistent with the predicted role for DDI2 in cleaving NFE2L1, little to no nuclear localization of cleaved-NFE2L1 was observed in DDI2Δ cells or NFE2L1Δ control cells (Figure 2.5c). Interestingly, NGLY1Δ cells displayed diminished nuclear import of cleaved-NFE2L1 (Figure 2.5c). These findings suggest that de-glycosylation by NGLY1 is necessary for facilitating the import of cleaved-NFE2L1 into the nucleus. These observations are consistent with other findings, whereby a decrease in the nuclear import of fluorescently-tagged NFE2L1 in an NGLY1 LOF background was observed by immunofluorescence imaging (Tomlin et al. 2017). These results complement findings from previous studies on NFE2L1 processing and independently validate a role for NGLY1 in facilitating the nuclear import of cleaved-NFE2L1 in human cells.

2.3.5 Identifying a shared transcriptional signature in NGLY1Δ, NFE2L1Δ and DDI2Δ cells

Our current understanding indicates that both DDI2 and NGLY1 are necessary for NFE2L1’s processing and subsequent nuclear import (Tomlin et al. 2017). Of note is that NFE2L1’s ability to activate genes encoding proteasome subunits in response to proteotoxic stress is dependent

77

Figure 2.5: NFE2L1 nuclear import is impaired in the absence of NGLY1 and DDI2. a-b NFE2L1 full-length and cleaved forms are visualized using an anti- NFE2L1 antibody in different genetic backgrounds. Samples in a. were treated with 50nM of Epoxomycin and b. with 100μM of cholesterol for different lengths of time as indicated. Experiments a and b are a representation of n=2 c. Immunoblots of protein lysates from different genetic backgrounds separated into cytoplasmic and nuclear fractions through subcellular protein fractionation. Different treatments used are indicated. Fibrillarin was used as a loading control for the nuclear fraction while tubulin was used as a control for the cytoplasmic fraction. Cleaved-NFE2L1 was visualized using an anti- NFE2L1 antibody. I performed all experiments in 2.5 in equal contribution with Dr. Ashwin Seetharaman.

78

upon the activity of both NGLY1 and DDI2 (Lehrbach and Ruvkun 2016; Koizumi et al. 2016; Tomlin et al. 2017). Based on this information, we hypothesized that the regulation of a transcriptional cascade by NFE2L1 may account for some of the functional links between NGLY1, NFE2L1 and DDI2 and the secretory pathway that were predicted in the CRISPR screen data.

To test this hypothesis, RNA sequencing (RNA-Seq) was employed to identify genes that were differentially expressed across NGLY1Δ, NFE2L1Δ and DDI2Δ cells compared to HAP1-WT cells. Given that exposure to proteotoxic stress is known to trigger NFE2L1 nuclear translocation and transcriptional activity (Lehrbach and Ruvkun 2016; Tomlin et al. 2017), NGLY1Δ,

NFE2L1Δ, DDI2Δ and HAP1-WT cell lines were treated with a low dose (~IC40) of bortezomib to stimulate an NFE2L1-dependent transcriptional response and performed RNA-Seq. Genes that were differentially expressed in each background were identified using the Bioconductor package ‘limma’. I then performed a Gene Set Enrichment Analysis (GSEA) for each genetic background using the ‘Hallmark’ gene set within the Molecular Signatures Database (MsigDB). Through this analysis, I identified several negatively enriched pathways (FDR <0.4), some of which were shared across all three genetic backgrounds (Figure 2.6).

Within the predicted negatively enriched processes that were shared across all three genetic backgrounds, I identified 56 genes within these processes that were differentially expressed based on the limma analysis compared HAP1-WT cells (FDR of <0.05) (Table 2.2). Of this differentially expressed gene set, I observed several genes mapped to immunological pathways such as the interferon gamma response and the complement cascade (Figure 2.7). This finding was noteworthy since NGLY1 has been shown to regulate inflammation in an NFE2L1- dependent manner (Yang et al. 2018).

Consistent with the predictions from our CRISPR screen, several genes were identified that mapped to secretory pathway functions, such as cholesterol homeostasis, across NGLY1Δ, NFE2L1Δ and DDI2Δ backgrounds (Figure 2.7). Taken together, these findings suggest that the NGLY1 genetic pathway likely regulates the transcription of a variety of genes connected to various aspects of the secretory pathway and immune functions through NFE2L1.

79

Figure 2.6: Gene Set Enrichment Analysis of RNA sequencing of NGLY1∆, NFE2L1∆, and DDI2∆ cells. a-c. Gene set enrichment analysis (GSEA) was performed and genes were mapped to the Hallmark gene sets for NGLY1-, NFE2L1- and DDI2-knockout cells. Negatively enriched pathways (FDR < 0.4) were plotted for each genetic background. I performed the GSEA analysis of the screens.

80

Table 2.2 Differentially expressed genes across NGLY1, NFE2L1 and DDI2 knockout cells (FDR < 0.5) identified by GSEA. Enriched gene sets for NGLY1 analysis. I carried out the analysis to identify the shared enriched gene set.

Gene Log2 Fold-change Enriched MSigDB Gene Sets (in NGLY1) RPS3A -0.628172 Allograft Rejection HLA-E -0.5322723 Allograft Rejection RPL9 -0.3209875 Allograft Rejection HDAC9 0.22816319 Allograft Rejection FLNA 1.23098474 Allograft Rejection, Epithelial Mesenchymal Transition MERTK -0.4644932 Androgen Response RAB4A -0.434801 Androgen Response PGM3 -0.4079583 Androgen Response UAP1 -0.3876064 Androgen Response TPD52 -0.6454693 Androgen Response, Allograft Rejection, Hypoxia INSIG1 -0.8917374 Androgen Response, UV Response STARD4 -0.6420448 Cholesterol Homeostasis GUSB -0.5843833 Cholesterol Homeostasis HSD17B7 -0.5843047 Cholesterol Homeostasis ACSS2 -0.349104 Cholesterol Homeostasis FDPS -0.3268296 Cholesterol Homeostasis HMGCS1 -0.6656724 Cholesterol Homeostasis, Androgen Response SCD -0.3813036 Cholesterol Homeostasis, Androgen Response HMGCR -0.3705107 Cholesterol Homeostasis, Androgen Response FDFT1 -0.4872078 Cholesterol Homeostasis, Estrogen Response ALDOC -0.9553562 Cholesterol Homeostasis, Hypoxia CTSC -0.7496239 Complement CD46 -0.6485956 Complement LTA4H -0.4401106 Complement SH2B3 0.34557132 Complement CD55 -1.0799798 Complement, Inflammatory Response PLA2G4A -0.3896354 Complement, Interferon Gamma Response DPP4 -2.0875132 Complement, Pancreas Beta Cells ELOVL5 -0.4084509 Estrogen Response, Androgen Response COCH -0.8232901 IL2/STAT5 Signaling HUWE1 0.31792818 IL2/STAT5 Signaling PUS1 0.51173583 IL2/STAT5 Signaling NOP2 0.51350527 IL2/STAT5 Signaling IGF2R 0.8064216 IL2/STAT5 Signaling SOCS2 0.8744805 IL2/STAT5 Signaling GABARAPL1 1.29917372 IL2/STAT5 Signaling CDKN1C -1.300939 IL2/STAT5 Signaling, Hypoxia GPC3 -0.6798785 Inflammatory Response CDKN1A Inflammatory Response, Interferon Gamma Response, P53 -1.6976763 Pathway, Hypoxia, TNF Alpha Signaling via NFKB NFKB1 Inflammatory Response, Interferon Gamma Response, TNF 0.40724451 Alpha Signaling via NFKB, UV Response IFNGR2 Inflammatory Response, TNF Alpha Signaling via NFKB, -0.3703878 Allograft Rejection HLA-C -0.3734063 Interferon Alpha Response HLA-A -0.7285595 Interferon Gamma Response, Allograft Rejection RNF31 0.54276312 Interferon Gamma Response, Interferon Alpha Response Interferon Gamma Response, Interferon Alpha Response, B2M -0.8258729 Hypoxia, Allograft Rejection, Androgen Response Interferon Gamma Response, Interferon Alpha Response, TXNIP 0.96947608 P53 Pathway

81

CCND3 P53 Pathway, IL2/STAT5 Signaling, Allograft Rejection, -0.4234585 Androgen Response AKT3 0.8424916 Pancreatic Beta Cells, UV Response NEK2 0.27790278 Spermatogenesis PHF7 -0.5608131 Spermatogenesis KIF2C 0.31784163 Spermatogenesis CDK1 -0.3734063 Spermatogenesis PARP2 -0.8258729 Spermatogenesis CCNB2 -1.6976763 Spermatogenesis

82

Figure 2.7: The NGLY1 genetic pathway regulates cholesterol homeostasis and immune processes. GSEA enrichment of differentially expressed genes by RNA-sequencing across all genetic backgrounds compared to WT , annotated according to the Hallmark gene sets. Global enrichment of all genes comprising the a. cholesterol homeostasis c. complement e. inflammatory response and g. interferon alpha response gene sets. Insets b,d,f,h are genes that are enriched at an FDR <0.05 for the given gene set. I performed the RNA sequencing experiment in equal contribution with Dr. Ashwin Seetharaman. Dr. Kevin Brown analyzed the data. I carried out the subsequent analysis leading to the identification of the gene signature and generated the heat maps.

83

2.3.6 Loss of NGLY1, NFE2L1, or DDI2 results in dysregulated early endosome and cholesterol trafficking

Our findings that NGLY1Δ, NFE2L1Δ and DDI2Δ cells display negative GIs with several genes that function within the secretory pathway, including one or more members of the GARP complex, suggests a novel role for these gene products in regulating intracellular vesicle and substrate trafficking. The GARP complex is a regulator of retrograde trafficking of substrates between endosomes, lysosomes and Golgi and was also identified as a critical regulator of intracellular cholesterol transport (Wei et al. 2017). Based on these reports and the predictions from our CRISPR screen, we hypothesized that NGLY1Δ, NFE2L1Δ, and DDI2Δ cells may display phenotypes that are consistent with impairments in subcellular vesicle and cholesterol trafficking.

To test this hypothesis, we examined the subcellular abundance of early endosomes as well as intracellular cholesterol by immunofluorescence confocal imaging in NGLY1Δ, NFE2L1Δ and DDI2Δ cells. Early endosomes were visualized using an antibody against early endosome antigen 1 (EEA1). Consistent with my hypothesis, we found that cells lacking NGLY1, NFE2L1 or DDI2 have significantly more early endosomes compared to HAP1-WT cells (Figure 2.8), suggesting a shared dysregulation in intracellular vesicle trafficking.

To visualize intracellular cholesterol, we employed filipin staining, based on previous reports (Wei et al. 2017) and detected a dramatic increase in the abundance of intracellular cholesterol particles across NGLY1Δ, NFE2L1Δ and DDI2Δ cells (Figure 2.9a,b). As positive control for detecting intracellular cholesterol, we also treated HAP1-WT and knockout cells with U18666a. Previous studies have shown that treating cells with U18666a, a chemical inhibitor of NPC1 forces cholesterol to accumulate within the lysosomes (Wei et al. 2017). Consistent with previous reports, we observed a significant increase in filipin staining in HAP1-WT cells (Figure 2.9c,d), suggesting that an active intracellular cholesterol transport mechanism is normally in place in HAP1-WT cells and cholesterol accumulation would not normally be observed in these cells when untreated.

Taken together, these findings are consistent with a predicted role for NGLY1, NFE2L1 and DDI2 within the secretory pathway in coordinating intracellular vesicle and substrate trafficking.

84

Figure 2.8: Loss of NGLY1, NFE2L1 and DDI2 result in dysregulated intracellular vesicle trafficking. a. Immunofluorescence confocal microscopy images of early endosomes in HAP1-WT and knockout cells. b. Quantification of the number of early endosomes are shown on the right. The number of early endosomes per well was normalized to the cell surface area per well. Final normalization for knockout cell lines was compared to WT. *** denotes p< 0.001, unpaired students t-test. Scale bar represents 20μm. Anti-phalloidin = actin, EEA1 = early endosomes. I carried out the immunofluorescence experiments in equal contribution with Dr. Ashwin Seetharaman and I carried out the analyses.

85

Figure 2.9: Loss of NGLY1, NFE2L1 and DDI2 results in cholesterol accumulation. Immunofluorescence confocal microscopy images of cholesterol particles in HAP1-WT and knockout cells in untreated or NPC1 inhibitor U18666 treated. a. Cholesterol abundance in untreated cell lines. b. Quantification of cholesterol abundance in untreated cell lines. The number of cholesterol particles per each well was normalized to the cell surface area per each well. c. cholesterol abundance in cell lines treated with NPC1 inhibitor to induce cholesterol accumulation as a positive control. d. Quantification of cholesterol abundance in treated compared to untreated cell lines. Cholesterol abundance was quantified as in b. anti-phalloidin = actin, Filipin III = cholesterol. *** denotes p< 0.001, unpaired students t-test. Scale bar represents 20μm. I carried out the above experiment in equal contribution with Dr. Ashwin Seetharaman and I carried out the analyses.

86

2.4 Discussion

Although recent advances in next generation sequencing technology have facilitated identification of numerous genes associated with rare diseases, functions for many of these genes remain poorly characterized (Boycott et al. 2013). In collaboration with Dr. Seetharaman, I sought to delineate the functions of a rare disease-associated gene NGLY1 from the context of its GIs. It has been shown in yeast that genes that are involved in similar biological processes share overlapping sets of GIs (Baryshnikova et al. 2010; Costanzo et al. 2016), Therefore, to gain a broader understanding of the genetic dependencies underlying NGLY1-deficiency, genome-wide CRISPR-based genetic screens were employed to elucidate negative GIs of NGLY1 and that of its related pathway members NFE2L1 and DDI2 in human HAP1 cells.

Previous studies have provided putative evidence for the existence of an NGLY1 genetic pathway in human cells in defending against proteotoxic stress (Lehrbach and Ruvkun 2016; Tomlin et al. 2017). We hypothesized that if an NGLY1 genetic pathway exists in human cells, an unbiased genome-wide CRISPR screen for genes that mediate sensitivity to the proteasome inhibitor bortezomib, would reveal NGLY1, NFE2L1 and DDI2 as top hits. These genes were indeed identified among the top sensitizing hits from the genome-wide bortezomib screen and were subsequently validated in several distinct human cancer cell lines as sensitizers of bortezomib. These findings are consistent with previous reports from other groups and strongly suggest that a conserved genetic pathway comprising NGLY1, NFE2L1 and DDI2 operates in human cells in response to proteotoxic stress and may also collectively coordinate other cellular processes.

We then carried out three independent genome-wide CRISPR screens in NGLY1Δ cells to identify the top GIs of NGLY1. We observed that the screens strongly correlated with each other and members of the GARP complex were revealed as one of the top predicted negative GIs of NGLY1 in addition to other genes involved in secretory processes. Upon screening NFE2L1Δ and DDI2Δ cells, we observed that the top negative GIs that were shared with NGLY1 occurred with the GARP complex. Some of the other shared negative GIs included genes involved in GPI- anchor biosynthesis, Golgi trafficking via the COG complex, and the PDC complex but were nevertheless more prominent hits in the NGLY1Δ screens compared to NFE2L1Δ and DDI2Δ screens. These findings suggest that while an NGLY1 genetic pathway comprising NGLY1,

87

NFE2L1 and DDI2 may regulate some aspects of secretory pathway functions, NGLY1 likely de-glycosylates other substrates in an addition to NFE2L1.

The GARP complex regulates intracellular vesicle and substrate trafficking including intracellular cholesterol transport (Wei et al. 2017), and the observed GIs between members of the GARP complex and the NGLY1 pathway suggests a role for the NGLY1 pathway in regulating some aspects of intracellular substrate trafficking. Consistent with this hypothesis, it was observed that NGLY1Δ, NFE2L1Δ and DDI2Δ cells display cellular phenotypes that are consistent with defects in impaired intracellular vesicle and substrate trafficking. These phenotypic findings highlight the biological implications predicted through the CRISPR screens and lend additional support to the computational framework developed for analyzing genome- wide GI screen data in human cells.

The NFE2L1 transcription factor requires processing by NGLY1 and DDI2 in order to enter the nucleus, it remained to be seen whether they drive a transcriptional program of secretory pathway functions in HAP1 cells. Consistent with this hypothesis, several genes were identified through RNA-Seq that mapped to secretory pathway functions including cholesterol homeostasis across NGLY1Δ, NFE2L1Δ and DDI2Δ backgrounds. STARD4 is particularly noteworthy as one of the significantly downregulated genes across the NGLY1Δ, NFE2L1Δ and DDI2Δ backgrounds. Intracellular cholesterol trafficking occurs through vesicular and non-vesicular transport mechanisms, with non-vesicular transport playing an essential role cholesterol distribution and regulation (Maxfield and Wüstner 2002). Sterol transfer proteins such as the steroidogenic acute regulatory protein (StAR)-related lipid transfer (START) domain proteins have been implicated in non-vesicular transport of cholesterol between membranes of various organelles as well as to and from the plasma membrane (Luo et al. 2017). STARD4 in particular has been implicated as a critical lipid transfer protein in maintaining cholesterol homeostasis (Calderon-Dominguez et al. 2014). STARD4 traffics sterols to and from various organelles, including the ER and mitochondria (Calderon-Dominguez et al. 2014), and is transcriptionally regulated by SREBP-2 where its transcription is repressed when sterols are in excess (Soccio et al. 2005). It has been hypothesized that STARD4 regulates cholesterol homeostasis by transporting cholesterol to the ER for esterification and the mitochondria for the generation of oxysterols which in turn would regulate SREB2 levels (Rodriguez-Agudo et al. 2008). It was recently shown that reduced expression of STARD4 results in intracellular cholesterol 88

accumulation, which is part due to excess cholesterol storage in late endosomes or lysosomes, as well as slower trafficking of cholesterol (Iaea et al. 2020). Conversely, overexpression of STARD4 increases cholesterol ester formation and bile acid synthesis, indicating an increase in cholesterol transport to the ER and mitochondria (Rodriguez-Agudo et al. 2008). Taken together, decreased expression of STARD4 in NGLY1 pathway knockouts may result in intracellular cholesterol accumulation and impaired cholesterol trafficking the ER and/or mitochondria, thereby disrupting cholesterol homeostatic mechanisms.

Based on these findings, I propose a speculative model whereby the NGLY1 genetic pathway in human cells may regulate a non-vesicular route of intracellular cholesterol transport that acts in parallel to vesicular substrate trafficking processes that would involve the GARP complex. Therefore, the loss of any member of the NGLY1 genetic pathway with a member of the GARP complex, would result in synthetic lethality due to abrogation of both vesicular and non-vesicular modes of cholesterol trafficking within the cells (Figure 2.10). Further characterization of cholesterol trafficking in NGLY1 knockout cells, including whether intracellular cholesterol accumulation occurs as free cholesterol or within specific organelles, would offer valuable insight into the role of the NGLY1 pathway in cholesterol homeostasis.

Importantly, impaired cholesterol trafficking resulting in intracellular cholesterol accumulatio, may explain some of neurological symptoms and liver pathology in NGLY1-deficiency. The endocytic process is known to play significant roles in pathology, as it is responsible for the uptake of nutrients such as cholesterol, as well as regulating the surface expression levels of membrane proteins. A disruption of these processes is a hallmark to multiple human diseases such as Alzheimer’s, Parkinson’s, amyotrophic lateral sclerosis and lysosomal storage diseases (Schreij, Fon, and McPherson 2016)and recent studies suggest that impaired cholesterol trafficking may also contribute towards the onset of other neurodegenerative diseases (Wei et al. 2017). Therefore cholesterol accumulation in NGLY1-deficient neurons, that may be in part a result of impaired cholesterol trafficking, could explain clinical seizures observed in some patients (Lam et al. 2017). Investigation of possible secretory defects secretory and cholesterol abundance in NGLY1-deficient neuronal cell lines would offer insight into the potential involvement of these processes in disease pathogenesis.

89

Figure 2.10: Proposed model of how the NGLY1 pathway and the GARP complex may coordinate intracellular vesicle trafficking in human cells. (top left) In wild type cells, the GARP complex and NGLY1 pathway regulate cholesterol trafficking through vesicular and non- vesicular routes. NFE2L1 enters the nucleus where it upregulates the expression of STARD4 for non-vesicular cholesterol transport. (top right) The loss of any members of the NGLY1 pathway impairs the nuclear import of NFE2L1 and its transcriptional activity resulting in downregulation of STARD4. Non-vesicular cholesterol transport is impaired. (bottom left) The loss of the GARP complex impairs vesicular cholesterol transport. (bottom right) The loss of GARP complex and the NGLY1 pathway results in a synthetic lethal genetic interaction due to impaired vesicular and non-vesicular transport of cholesterol. Created with BioRender.

Figure 2.12. Proposed model of how the NGLY1 pathway and the GARP complex may coordinate intracellular vesicle trafficking in human cells. (top left) In wild type cells, the functioning GARP and NGLY1 pathway regulate cholesterol trafficking through vesicular and non- vesicular routes. NFE2L1 enters the nucleus where it upregulates the expression of STARD4 for non-vesicular cholesterol transport. (top right) The loss of any members of the NGLY1 pathway impairs the nuclear import of NFE2L1 and its transcriptional activity resulting in downregulation of STARD4. Non-vesicular cholesterol transport is impaired. (bottom left) The loss of the GARP complex impairs vesicular cholesterol transport. (bottom right) The loss of GARP complex and the NGLY1 pathway results in a synthetic lethal genetic interaction due to impaired vesicular and non-vesicular transport of cholesterol. (Created with BioRender)

90

Hepatocyte-specific Ngly1-deficient mice form fatty liver and display increased droplet accumulation under a high fructose diet (HFrD) induced stress, which may in part be a result of impaired lipid release from liver (Fujihira et al. 2019). As Nfe2l1 processing was impaired in HFrD fed hepatocyte-specific Ngly1-deficienct mice, impaired Nfe2l1 activity likely contributes to this phenotype (Fujihira et al. 2019). Our observation of cholesterol accumulation in NGLY1 pathway knockout cells, which may be due to impaired intracellular trafficking, could additionally contribute to liver pathology in NGLY1-deficiency. Intriguingly, hypocholesterolemia is observed in patients with NGLY1-deficiency (Lam et al. 2017). It is therefore possible that liver pathology observed in patients with NGLY1-deficiency may be in part be a result of impaired regulation of lipid metabolism as well intracellular cholesterol transport. Taken together, findings from this study suggest that the loss of NGLY1 function may contribute to some of the diverse clinical symptoms through impaired vesicle trafficking and cholesterol accumulation and offers new avenues for further investigation and therapeutic intervention. Possible secretory defects and cholesterol abundance should therefore be investigated in NGLY1-deficient hepatocytes.

Previous studies have uncovered a role for NGLY1 in regulating various immune processes including antigen presentation (Altrich-VanLith et al. 2006) and inflammation (Yang et al. 2018). GSEA analysis of the RNA-Seq data revealed that several shared immune processes were negatively enriched across NGLY1Δ, NFE2L1Δ and DDI2Δ cells. Interestingly, while NGLY1 has been shown to regulate antigen presentation through the necessary deamidation of antigen peptides (Altrich-VanLith et al. 2006), downregulation of MHC class I genes B2M and HLA-A/B was also observed, suggesting that the NGLY1 genetic pathway may also be involved in antigen presentation via NFE2L1-mediated transcriptional regulation and is worthy of further investigation. Additionally, genes involved in the complement response were also observed. CD46 and CD55 are ubiquitously expressed complement regulatory proteins that protect autologous cells from complement-mediated attack. Reduced expression or impaired function of these regulatory proteins results in a chronic proinflammatory state and thus contributes to the pathogenesis of autoimmune, inflammatory and neurodegenerative diseases (Cardone, Le Friec, and Kemper 2011; Dho, Lim, and Kim 2018; Orsini et al. 2014) Although NGLY1-deficiency has not been formally considered as an autoinflammatory disease, abnormal brain imaging,

91

progressive loss of neurons, and evidence of neuropathy have been reported for many NGLY1- deficiency patients (Lam et al. 2017).

In conclusion, this work highlights not only the importance of defining global GI profiles for rare disease-associated genes such as NGLY1, but also that of genes known to function in a shared pathway. This approach has led to the discovery of novel genetic relationships for members of the NGLY1 pathway and offers insight into additional biological processes in which NGLY1’s involvement is critical. Dysregulation of secretory processes within the cell as a result of loss of NGLY1, and subsequently NFE2L1 function, may contribute to the underlying pathology of NGLY1-deficiency. Taken together, this work has revealed new aspects of NGLY1-deficiency that may offer insight into the development of successful therapeutic strategies for affected individuals.

92

Chapter 3 Genome-Wide CRISPR Screen Identifies Complement-Independent Roles for the Complement Regulatory Proteins

Components of the work discussed in this chapter were a collaborative effort between myself and Dr. Ashwin Seetharaman.

The experiments and analyses described in Figures 3.1, 3.2a-d and 3.3a.b were performed entirely by me. I performed the experiments in Figures 3.3c, 3.4, 3.5, 3.6 and in equal contribution with Dr. Ashwin Seetharaman. Ellis Kelly (a summer student whom I mentored and supervised) generated the constructs in Figure 3.2e and carried out the complement sensitivity assay under my supervision. Dr. Max Billmann analyzed the CRISPR screens. Please refer to the figure legends for further details on the specific contribution made by each individual towards the experiment in question.

93

Genome-Wide CRISPR Screen Identifies Complement- Independent Roles for the Complement Regulatory Proteins 3.1 Abstract

The complement system has traditionally been viewed as a critical arm of innate immunity that aids in the removal of pathogens, immune complexes, cellular debris and dying cells. However, there is now a growing appreciation for the complement system as a regulator of diverse biological processes that are critical for cell and tissue integrity and homeostasis, as well as basic cellular processes such as metabolism. Dysregulation of complement activity, or its regulators, may therefore have broader implications in human diseases than previously anticipated. The complement regulatory proteins (CRPs) include CD46, CD55 and CD59. The CRPs are key membrane-bound inhibitors of complement and are involved in the pathogenesis of a wide range of human diseases. However, their potential roles that are independent of complement inhibition are poorly understood. By harnessing the power of genetic interactions to provide insight into gene function, I carried out a genome-wide CRISPR in a CRP triple knockout cell line (CD46-CD55-CD59-), to delineate the complement-independent biological pathways in which the CRPs may function in. I identified a novel functional link between the CRPs and members of the NGLY1 genetic pathway, that together may regulate opposing aspects of intracellular cholesterol trafficking. Furthermore, the CRPs likely regulate NFE2L1 turnover and processing during cholesterol-induced stress. These findings highlight new avenues for interrogating the functions of the CRPs beyond systemic complement regulators and their roles in human disease.

94

3.2 Introduction

The complement system is a critical arm of the innate immune system, which functions to clear pathogens, dying cells and immune complexes (Merle, Noe, et al. 2015). Because of its potential for generating potent pro-inflammatory effects including tissue destruction, complement activation must be tightly controlled on healthy host cells (Merle, Noe, et al. 2015). It is no surprise that dysregulation of the complement system due to ineffective regulation and/or over- activation, can result in significant damage to host cells and contributes to the pathology of a wide range of human diseases (Baines and Brodsky 2017; Ricklin and Lambris 2013; Daniel Ricklin, Reis, and Lambris 2016; Carroll and Sim 2011).

The complement regulatory proteins (CRPs) are encoded by the CD46, CD55 and CD59 genes and are key membrane-bound inhibitors of complement and are widely expressed on host cells (Zipfel and Skerka 2009). CD46 and CD55 function in deactivating key multi-protein enzymatic complexes of the cascade, the C3 and C5 convertases, by acting as either a cofactor for the cleavage of convertase components, or accelerating their dissociation (Zipfel and Skerka 2009). CD59 prevents the formation of the membrane attack complex in the cell membrane, sparing host cells from osmolytic lysis (Zipfel and Skerka 2009). Genes encoding the CRPs are involved in the pathogenesis of numerous common and rare diseases (Meri and Jarva 2013; Liszewski and Atkinson 2015; Liszewski and Atkinson 2015; Baines and Brodsky 2017; Ellinghaus et al. 2017).

Roles for the CRPs beyond regulators of systemic complement have been identified over the last decade (Kolev et al. 2015; Capasso et al. 2006; Sutavani et al. 2013; Blom 2017; Golec et al. 2019; Liszewski et al. 2013; Arbore et al. 2016). For example, Autocrine CD46 stimulation via intracellularly generated complement fragments is critical in driving key T cell metabolic pathways during Th1 response (Liszewski et al. 2013; Arbore et al. 2016). Consequently, dysregulation of CD46 contributes to the pathology of several autoimmune diseases as a result of impaired T cell responses (Astier et al. 2006; Astier 2008; Xu et al. 2010). CD55 and CD59 have also been shown to have a direct impact on T cell responses ( Sutavani et al. 2013; Melania Capasso et al. 2006) which may in part be mediated through modulating lipid raft composition (Kimberley, Sivasankar, and Paul Morgan 2007; Wang et al. 2018; Toomey, Cauvi, and Pollard 2014) on T cells which is critical for intracellular T-cell receptor signaling (Kabouridis 2006).

95

Lastly, CD59 may be involved in the trafficking of insulin through interactions with SNARE proteins and consequently, the pathogenesis of diabetes (Krus et al. 2014; Golec et al. 2019). Taken together, it is likely that the cellular functions of the CRPs are much broader than originally anticipated. Gaining a deeper understanding of the function of the CRPs and the biological processes in which they are involved will be helpful for determine their precise role in human disease.

Advancements in functional genomic technologies such as CRISPR offer new avenues for investigating the biological functions of the CRPs in human cells. GIs offer a powerful approach for predicting gene function and illuminating the biological processes in which a gene may be involved (Costanzo et al. 2016; A. Tong et al. 2001). Systematic mapping of GIs through the use of genome-wide CRISPR/Cas9 libraries in human cells has produced valuable insight into the genetic underpinnings of human diseases as well avenues for investigating therapeutic interventions (Wang et al. 2019; Macleod et al. 2019; Steinhart et al. 2017). To date, the investigation of CRP function via GIs has not been carried out, and may offer insight into the biological roles beyond complement.

However, a major challenge in gaining insight into the biological roles of the CRPs through this genetic approach lies in the fact that CRPs share structural and functional similarities that may confound potential genetic interactions. For example, CD46 and CD55 are comprised of a similar extracellular domain structure and their inhibitory activity is achieved through their interactions with both C3b and C4b (Persson et al. 2010; Lukacik et al. 2004) . Although the extracellular domain of CD59 is unlike the other CRPs, CD59 and CD55 are both GPI-anchored proteins which has been shown to be important in their signaling capacity (Bohana-Kashtan et al. 2004; Wang et al. 2018; Shenoy-Scaria et al. 1992).

In this chapter, I describe how I designed and developed a human HAP1 cell line lacking all three CRPs and carried out pooled genome-wide CRISPR/Cas9 screens to investigate their GIs. I identified biological pathways that may be impaired in the absence of the CRPs as well as a novel functional link to members of the NGLY1 genetic pathway; NGLY1, NFE2L1 and DDI2. Furthermore, findings from this study suggest that the CRPs likely contribute to the processing of transcription factor NFE2L1 under cholesterol-induced stress and that together, the CRPs and the NGLY1 pathway may regulate different aspects of cholesterol trafficking. These findings

96

offer novel insights into the biological functions of the CRPs beyond systemic complement regulators and serves as a foundation for further investigation into their roles in human disease.

3.3 Results 3.3.1 Generation of clonal HAP1 cell lines lacking surface CRPs

CRPs are ubiquitously expressed across all human cells, with most cell types expressing at least two CRPs on their surface (Zipfel and Skerka 2009), including human HAP1 cells which expresses all three CRPs (Thielen et al. 2018). Therefore, to investigate the biological role of the CRPs by mapping their GIs, a cell line that lacks all three CRPs is necessary. To generate two unique clonal HAP1 cell lines lacking all CRPs, I obtained a CD46 knockout cell line (CD46) through Horizon Discovery Inc. and targeted CD55 and CD59 with CRISPR/Cas9 sgRNAs (Figure 3.1a). CD46 cells harboring sgRNAs targeted against CD55 and CD59 were passaged for two weeks and using antibodies against all CRPs, single cells lacking all surface CRPs were obtained by fluorescence-activated cell sorting (FACS). Clonal cell lines were confirmed to contain indels by conventional Sanger sequencing. Next, I analyzed the cell lines by flow cytometry and found that in contrast to the parental CD46 cells, the CRPΔ cell lines also lacked CD55 or CD59 expression (Figure 3.1b).

Finally, to confirm that the CRP cell lines were indeed lacking all CRPs on the cell surface, I performed a complement sensitivity assay. HAP1 wild-type (WT), CD46 and CRP cell lines were treated with increasing doses of complement and cell viability was measured. As expected, cells lacking CD46 or all three CRPs are significantly more sensitive to complement compared to wild-type cells (Figure 3.1c). Together, these findings confirm that single cell clonal cell lines CRPΔc1.3 and CRPΔc2.3 are indeed a triple KO of all three CRPs.

3.3.2 Genome-wide CRISPR screen of CRP-KO cells identifies sensitizers and suppressors of complement in the absence of CRPs

Although novel roles for the CRPs beyond regulators of the complement cascade have been identified, their biological functions in the absence of complement have not been investigated using a global genome-wide approach. As GIs reveal functional relationships between genes,

97

Figure 3.1: Generation of human HAP1 cell lines lacking all surface CRPs. a. CRISPR/Cas9 gRNAs were used to target CD55 and CD59 in a human HAP1 CD46 knockout (CD46) cell line to generate clonal CD46/CD55/CD59-knockout (CRP) cell lines c1.3 and c2.3. Gene location of CRISPR/Cas9 gRNAs targeting CD55 (left) and CD59 (right) to generate cell lines c1.3 and c2.3 (Created with BioRender). b. Flow cytometry was performed to confirm the loss of CD55 and CD59 surface expression on both clonal cell lines (middle, right) compared to the parental CD46 cell line. c. The loss of all CRPs sensitizes cells to complement-mediated cell death. Wild-type (WT), CD46 and clonal CRP cell lines were incubated with complement and cell viability was measured. SCR; short consensus repeat domain, LU; Ly-6 antigen/uPAR domain. I performed the above experiments.

98

I sought to map the negative GIs of the CRPs by carrying out a genome-wide CRISPR/Cas9 gRNA screen in a human HAP1 CRP cell line to gain insight into their functions (Figure 3.2a). Because CRP-KO cells are hypersensitive to complement, I also carried out a genome-wide

CRISPR/Cas9 gRNA screen in CRP cells treated with an IC40 dose of complement (Figure 3.2a). This approach would allow for the identification of genes or biological processes that mediate sensitivity or resistance to complement, some of which may be predicted due to the mechanism of action of key soluble complement inhibitors, such as complement factor H, whose function would be particularly important in protecting CRP cells from complement-mediated destruction (Ferreira and Pangburn 2007).

Towards this goal, CRPΔ-c2.3 (referred to as CRPΔ from here on) HAP1 cells were infected with a 71k guide TKOv3 library to perform a systematic gene knockout screen similar to what the Moffat lab has reported previously (Figure 3.2a) (Hart et al. 2015, 2017). Briefly, following puromycin selection of TKOv3 library-infected cells at ~200-fold library coverage, CRPΔ cells were split into either untreated or complement-treated groups. To identify sensitizers and suppressors of complement, cells were cultured at an IC40 dose of complement and both arms of the screen were passaged every three days for approximately 20 doublings. Genomic DNA from the initial (T0) and final (T18) time-points was collected and subjected to deep sequencing. Genes that sensitize cells to complement are expected to drop out and deplete from the collective pool of library-infected cells. Conversely, genes that mediate resistance to complement when knocked out, are expected to be enriched in the final pool of cells. Genes that represent negative GIs with CRPs (untreated group) are expected to drop out from the pool of library infected cells. Fold-changes in sgRNA abundance were determined and quantified using the scoring pipeline for CRISPR/Cas9 screens performed with TKOv3 and described in Chapter 2. The resulting list of predicted GIs for each screen were subjected to further analyses by GSEA or DrugZ.

To begin investigating sensitizers and suppressors of complement as a measure of confidence for the screens, I performed GSEA on the ranked gene output list from the complement-treated screen using the “KEGG” gene set within the MSigDB. Through this analysis, I identified negatively (FDR<0.25) and positively (FDR <0.025) enriched pathways (Figure 3.2b,c). GPI anchor biosynthesis was one of the most significantly positively enriched pathways with core enrichment genes comprised of almost all of the PIG anchor biosynthesis genes (Figure 3.2b).

99

Figure 3.2: Genome-wide CRISPR/Cas9 screen predicts sensitizers and suppressors of complement in the absence of the complement regulatory proteins. a. Schematic of the screens carried out in CRP∆ cells. CRP∆ cell line c2.3 was infected with CRISPR/Cas9 gRNA library TKOv3. Cells harboring gRNAs were puromycin selected and passaged. Cells were split into untreated and complement-treated conditions (T3) with treatment beginning on T6 ( IC40 dose) Cells were passaged (retreated in treatment group) every 3 days and gDNA was collected. T18 and T0 gDNA was subjected to deep sequencing and analyzed by GSEA or DrugZ. Positive (b) and negative (c) enrichment pathway plots from the complement-treated screen. The KEGG gene set list within the MSigDB was used. Gene ratio = ratio of core enrichment genes within the gene set to the size of the total gene set, FDR q-val = false discovery rate of GSEA enriched pathway. d. DrugZ plot highlighting top sensitizers and suppressors from pathways in enrichment plots b,c; glycosyltransferases (green), PIG anchor biosynthesis genes, and mTOR signaling genes. e. Top 10 predicted sensitizers of complement ranked by fitness score. f. A complement cell viability assay was performed on CRP∆ cells (~IC40 dose) uninfected or infected with CRISPR/Cas9 gRNAs targeting AAVS1 (non-target control), FXYD5 or MGAT2 and cell viability was measured. 100

The analysis includes two technical replicates, each of three wells. Data are normalized to the untreated CRP∆ cell line infected with the gRNA indicated and depicted as mean + SD. Unpaired two-tailed Student’s t-test between treated AAVS1 and target genes, * p < 0.05, ** p < 0.01, *** p < 0.001. GSL; glycosphingolipid. I carried out the experiments and analyses in a-e. Dr. Max Billmann analyzed the screen. Ellis Kelly generated the constructs and carried out the complement assay in f under my supervision.

101

Additionally, several of the positively enriched pathways are involved in the pathogenesis of various cancers and within these gene sets, many of the core enrichment genes are also involved in mTOR signaling. Taken together, these findings suggest that a LOF in genes involved in GPI- anchor biosynthesis and mTOR signaling would render cells more resistant to complement in the absence of the CRPs.

A significant number of the negatively enriched pathways were involved in the biosynthesis and metabolism of glycans and glycoconjugates (Figure 3.2c). Within these sets, many of the core enrichment genes encoded for various classes of glycosyltransferases. It is perhaps not unexpected that a LOF in particular glycosyltransferase genes render cells more sensitive to complement given the mechanism of action of complement factor H (CFH), the main soluble inhibitor of the alternative pathway (AP) of complement activation. In the absence of the CRPs, HAP1 cells do not possess any other surface complement inhibitors and thus the activity of CFH is critical for protecting cells against complement. In binding C3b, CFH inhibits the formation of the AP C3 convertase, accelerates its dissociation and acts as a cofactor for its cleavage. CFH interacts with surface-bound C3b by binding sialic acid-capped glycans and the glycosaminoglycan (GAG) chains of proteoglycans on the cell surface (Parente, Clark, Inforzato, and Day, 2017). Therefore, a LOF in genes involved in the synthesis of sialic acid and GAGs could impair CFH’s inhibitory activity and render cells more sensitive to complement.

In addition to subjecting the list of genes from the complement treated screen (assigned and ranked by fitness defect score) to GSEA, I used DrugZ to identify sensitizers and suppressors of complement. Fold-changes in sgRNA abundance between the complement-treated and untreated groups were compared computationally using DrugZ-scoring to identify suppressors/sensitizers, or chemical-GIs of complement. Core enrichment genes identified the GSEA analysis from Figure 3.2b and c including glycosyltransferases, PIGs and genes involved in mTOR signaling were again identified as the top sensitizers and suppressors of complement and highlighted along the DrugZ plot (Figure 3.2d).

To validate predictions from the complement-treated screen, I focused on two of the top predicted sensitizers of complement, FXYD5 and MGAT2. FXDY5, which encodes a cancer- associated cell membrane glycoprotein (Sato et al. 2003) and was ranked the top predicted negative GI, whereas MGAT2 encodes a N-acetylglucosaminyltransferase (GlcNacT) whose

102

activity represents the committed step in the synthesis of complex-type N-glycan structures (Wang et al. 2001). Validation experiments proceeded with transduction of CRPΔ cells with CRISPR-sgRNAs targeting FXYD5, MGAT2, or AAVS1 as a control. A complement assay was then performed and cell viability was compared between targeted knockout cells lines and the AAVS1 control. I found that targeting FXYD5 or MGAT2 in CRPΔ cells sensitized cells to complement (Figure 3.2e).However, it is worth mentioning that although the observed effect of targeting these genes was statistically significant, it nevertheless was modest. This raises the question of whether or not these genes were fully knocked out in the CRPΔ cells or if in fact losing these genes renders cells only slightly more sensitive to complement. Further investigation to confirm gene knockouts, as well as orthogonal complement assays may help establish these predicted GIs of the CRPs.

3.3.4 Members of the NGLY1 genetic pathway share strong negative GIs with the CRPs

To gain insight into the biological roles of the CRPs in the absence of complement through their negative GIs, GSEA analysis was performed on the gene output list from the untreated screen using the GO gene set within the MSigDB. Through this analysis, negatively enriched gene pathways were identified (FDR<0.15) including several immune pathways (inflammatory cell apoptosis and innate immune response activation via cell surface receptor signaling), spinal cord development, ubiquitination, and responses to sterol (Figure 3.3a). Strikingly, NGLY1, NFE2L1 and DDI2 were all identified as negative GIs (Figure 3.3b). This finding was interesting given that the unbiased RNA sequencing analysis I previously carried out (Chapter 2; Figure 2.8) identified a set of differentially expressed genes in NGLY1Δ, NFE2L1Δ and DDI2Δ cells that mapped to the complement cascade.

We next sought to investigate the connection between the CRPs and the NGLY1-NFE2L1-DDI2 genetic pathway. HAP1 wild-type and CRPΔ cells were targeted with sgRNAs targeting NGLY1, NFE2L1, DDI2, or AAVS1 as a control and we observed that targeting any member of the NGLY1 pathway in CRPΔ cells resulted in a significantly decreased cellular fitness compared to the AAVS1 control (Figure 3.3c). This result not only validates the screen predictions and also predicts a novel genetic relationship between the CRPs and the NGLY1 genetic pathway.

103

Figure 3.3: Complement regulatory proteins have negative genetic interactions with NGLY1, NFE2L1 and DDI2. a. Negative enrichment GSEA of the gene output list ranked by fitness score from the CRP∆ untreated CRISPR/Cas9 screen. The GO gene set list within the MSigDB was used with a gene set cut off 15-500. Gene ratio represents the ratio of core enrichment genes within the gene set to the size of the total gene set. b. Genetic interaction map of the CRP∆ untreated CRISPR/Cas9 screen of predicted positive (yellow) and negative (blue) genetic interactions at a false- discovery rate (FDR) of 20%. Genetic interactions are determined by the fold-change in the abundance of a gRNA targeting a specific gene in CRP∆ cells compared to WT-HAP1 cells at T18. c. Cell viability assays of CRP∆ cells infected with lentiviral gRNAs targeting AAVS1 (non-target control), NGLY1, NFE2L1 and DDI2 confirming the predicted genetic interactions from the CRP∆ CRISPR/Cas9 untreated screen. Data are normalized to AAVS1 controls in WT or CRP∆ cells and depicted as mean + SD, * p < 0.05, ** p < 0.01, *** p < 0.001, unpaired two-tailed Student’s t-test. IIR; innate immune response, RNAP; RNA Polymerase. I carried out the analyses and generated the plots in a-b. I carried out the validation experiment of the NGLY1 pathway in c in equal contribution with Dr. Ashwin Seetharaman.

104

3.3.5 Loss of the CRPs perturbs NFE2L1 processing and turnover during cholesterol-induced stress

Previous findings have delineated a dual role NFE2L1 as an orchestrator of the cellular responses to proteotoxic and cholesterol-induced stress and that its differential processing is integral to these responses (Lehrbach and Ruvkun 2016; Koizumi et al. 2016; Tomlin et al. 2017; Widenmaier et al. 2017). Furthermore, the response to sterol was one of the top negatively enriched biological pathways from the untreated CRPΔ screen which included NFE2L1 as a core enrichment gene (Figure 3.3a-b). This raised the possibility that the CRPs may contribute to the differential processing of NFE2L1 under these stress conditions.

HAP1-WT and CRPΔ cells were treated with a proteotoxic stress-inducing agent, epoxomycin, or excess cholesterol over a 12h period as previously reported by Widenmaier et.al, 2017. In our hands, CRPΔ cells treated with excess cholesterol showed that processing and turnover of NFE2L1 was dysregulated compared to WT cells (Figure 3.4a). In CRPΔ cells, a greater accumulation of the full-length form of NFE2L1 was observed compared to WT cells. However, there were no differences observed in NFE2L1 processing between WT and CRPΔ cells when treated with epoxomicin (Figure 3.4a). Intriguingly, the presence of cleaved NFE2L1 was observed under both proteotoxic and cholesterol induced stress conditions; the cleaved fragment enters the nucleus as expected under epoxomycin treatment but not under high cholesterol challenge (Figure 3.4b). The reason for this disparity remains a mystery and awaits further investigation. Together, these findings suggest that under cholesterol-induced stress, the loss of the CRPs disrupts NFE2L1 processing and turnover.

3.3.6 Human cells lacking the CRPs display increased numbers of early endosomes and cholesterol accumulation

In the previous chapter, a link between the NGLY1 pathway and cholesterol homeostasis was observed; all single KO cell lines displayed dysregulated early endosome and cholesterol trafficking defects. Given the observation that the loss of the CRPs disrupts NFE2L1 turnover, and the link between the NGLY1 pathway and cholesterol metabolism, we were curious to see if CRPΔ cells shared any of the trafficking defects observed in the NGLY1 pathway KOs.

105

Figure 3.4: Loss of the complement regulatory proteins impairs NFE2L1 processing during a high- cholesterol challenge. . a. NFE2L1 processing was detected in Wild-type and CRP∆ cells upon Epoxomycin- (top) and cholesterol-induced (bottom) stress conditions. Wild-type and CRP∆ cells were treated with 50nM of Epoxomycin or 100µM of cholesterol for up to 12 hours. Experiment is representative of n=2. b. Nuclear localization of cleaved NFE2L1 in CRP∆ cells treated with Epoxomycin or Cholesterol compared to WT, NGLY1∆, NFE2L1∆ and DDI2 ∆ cells. Cells were treated for 4 hours and nuclear fraction was extracted from lysates. I carried out the above experiments in equal contribution with Dr. Ashwin Seetharaman.

106

To test this hypothesis, we examined the subcellular abundance of early endosomes as well as intracellular cholesterol by immunofluorescence confocal imaging in CRPΔ and WT cells as described in the previous chapter. We observed that CRPΔ cells have significantly more early endosomes (Figure 3.5a,b) and an accumulation of cholesterol compared to HAP1-WT cells (Figure 3.6a,b).

As a positive control for detecting intracellular cholesterol, we also treated HAP1-WT and CRPΔ cells with U18666a as previously described (Chapter 2.3.6). Consistent with previous reports, significant increase in filipin staining in HAP1-WT cells was observed (Figure 3.6a) suggesting that an active intracellular cholesterol transport mechanism is normally in place in HAP1-WT cells and thus cholesterol accumulation would not normally be observed in these cells when untreated. Taken together, these findings demonstrate that CRPΔ cells also display trafficking defects that are consistent with the NGLY1 pathway mutants and suggest that the CRPs may also be involved in regulating cholesterol homeostasis and vesicle trafficking.

3.4 Discussion

CRPs CD46, CD55 and CD59 are membrane-bound inhibitors of systemic complement whose deficiency contributes to the pathogenesis of several human diseases, many of which are rare (Astier 2008; Dho, Lim, and Kim 2018; Ellinghaus et al. 2017; Liszewski and Atkinson 2015; Das et al. 2019; Hill et al. 2017; Fremeaux-Bacchi et al. 2006). However, there is now a growing appreciation that the CRPs not only regulators of systemic complement but proteins that may also play important roles in other biological processes (Kolev et al. 2015; Capasso et al. 2006; Sutavani et al. 2013; Blom 2017; Golec et al. 2019; Liszewski et al. 2013; Arbore et al. 2016), therefore their involvement in disease may be much broader than initially thought. Identifying GIs offers a powerful approach for interrogating gene function and technological developments such as the emergence of genome-scale CRISPR libraries, now provide a method for mapping GIs of disease-associated genes in human cells, such as the CRPs. In this study, I sought to delineate the GIs of the CRPs that are independent of their functions as complement inhibitors.

The complement-treated screen was carried out in order to gain confidence in the ability of genome-wide CRISPR screens in CRP knockout cells to predict biologically relevant GIs.

107

Figure 3.5: Loss of the complement regulatory proteins results in intracellular vesicle trafficking defects. a. Immunofluorescence confocal microscopy images of early endosomes in HAP1-WT and knockout cells. b. Quantification of the number of early endosomes are shown on the right. The number of early endosomes per well was normalized to the cell surface area per well. Final normalization for knockout cell lines was compared to WT. *** denotes p< 0.001, unpaired students t-test. Scale bar represents 20μm. Anti-phalloidin = actin, EEA1 = early endosomes. I carried out the immunofluorescence experiments in equal contribution with Dr. Ashwin Seetharaman and I carried out the analyses.

108

Figure 3.6: Loss of the complement regulatory proteins results in cholesterol accumulation. Immunofluorescence confocal microscopy images of cholesterol particles in HAP1-WT and knockout cells in untreated or NPC1 inhibitor U18666a treated. a. Cholesterol abundance in untreated cell lines. b. Quantification of cholesterol abundance in untreated cell lines. The number of cholesterol particles per each well was normalized to the cell surface area per each well. c. cholesterol abundance in cell lines treated with NPC1 inhibitor to induce cholesterol accumulation as a positive control. d. Quantification of cholesterol abundance in treated compared to untreated cell lines. Cholesterol abundance was quantified as in b. anti-phalloidin = actin, Filipin III = cholesterol. *** denotes p< 0.001, unpaired students t-test. Scale bar represents 20μm. I carried out the above experiment in equal contribution with Dr. Ashwin Seetharaman and I carried out the analyses.

109

In the absence of the CRPs, the function of serum complement inhibitors would be critical in protecting cells from complement. Therefore, biologically relevant sensitizers of complement would likely include genes whose functions are integral to serum complement inhibitor activity.

GSEA and DrugZ analyses of the predicted GIs from the screen, revealed positively and negatively enriched pathways that are comprised of genes acting as suppressors or sensitizers of complement, respectively. Interestingly, I observed that many of the positively enriched pathways included those contributing to various cancers. The dynamic relationship between complement and tumor development has been of great interest, with accumulating evidence over the last decade that imbalanced complement activation plays a key role in cancer progression (Reis et al. 2018). Therefore, the observation that genes involved in several cancer pathways, and in particular those within the mTOR signaling cascade, are predicted suppressors of complement lend confidence to the screen predictions.

The negatively enriched pathways were largely comprised of those involved in the synthesis of glycans and glycoconjugates. Given the fact that the in the absence of the CRPs, HAP1 cells do not possess any other surface complement inhibitors, it is likely that the cells with largely depend on the activity of the soluble and surface-acquired complement inhibitor, FH, for protection against complement-mediated lysis. The importance of FH as an inhibitor of complement is highlighted by the fact that mutations in FH are associated with aHUS, a life-threatening rare- disease in which dysregulation of complement greatly contributes to disease symptoms

(Jokiranta 2018). FH functions as an inhibitor of both the fluid-phase (C3H20Bb) and surface (C3bBb) C3 Convertases of the alternative pathway (Bexborn et al. 2008) by binding sialic acid- capped glycans and the GAG chains of proteoglycans on the cell surface and acting as a co- factor in the cleavage of C3b (Parente et al. 2017). Many of the patients with aHUS possess mutations in the regions of FH that bind heparin and sialic acid on the surface of cells, thus highlighting critical interactions that are required for FH activity (Caprioli et al. 2001; Richards et al. 2001). Therefore, a LOF in genes encoding glycosyltransferases that contribute to the biosynthesis of sialic acid-containing glycans and GAGs should render cells highly sensitive to complement.

To validate the screen predictions, complement sensitivity assays were carried out to determine cell viability of CRPΔ cells with a LOF in top-predicted sensitizers FXYD5 or MGAT2 and

110

observed that in agreement with the screen prediction, targeting these genes renders CRPΔ cells more sensitive to complement. MGAT2 is glycosyltransferase involved in the synthesis of complex-type N-glycan structures. Overexpressing MGAT genes has been shown to increase sialic acid content on the cell by enhancing glycan antennary branching (Cha et al. 2017) and a reduction in terminal sialic acids is observed with the loss of MGAT2 (Wang et al. 2001). Therefore, it is tempting to speculate that the loss of MGAT2 could impair the inhibitory activity of FH by altering sialic acid content on the cell surface and thus by preventing its binding and cells are unprotected from complement.

Strikingly, analysis of the untreated screen identified NGLY1, NFE2L1 and DDI2 as the top predicted negative GIs of the CRPs. Together, these three genes comprise what is now referred to as the NGLY1 genetic pathway, a key regulator of the cellular response to proteotoxic stress (Lehrbach and Ruvkun 2016; Koizumi et al. 2016). While no definitive link between the NGLY1 pathway and the CRPs has previously been reported, I previously discovered that the expression levels of genes involved in the complement response, including two of the CRPs, CD46 and CD55, were downregulated across NGLY1Δ, NFE2L1Δ and DDI2Δ cells.

Findings from this study suggest that the CRPs may be involved in aspects of cholesterol homeostasis at the level of both NFE2L1 processing as well as independently coordinating aspects of intracellular cholesterol transport. Under normal cellular conditions, the CRPs are negative GI partners with NGLY1, NFE2L1 and DDI2. We discovered that the NGLY1 pathway, through NFE2L1 processing and its subsequent activity as a transcription factor, regulates a transcriptional program coordinating the non-vesicular transport of intracellular cholesterol. Consistent with this proposed model, NGLY1Δ, NFE2L1Δ and DDI2Δ cells display cellular phenotypes that are consistent with defects in impaired intracellular vesicle and substrate trafficking which include the accumulation of intracellular cholesterol and early endosomes. In this study, we also observed that CRPΔ cells displayed an accumulation of intracellular cholesterol and early endosomes that was significantly greater than what is observed in HAP1 wild-type cells.

Furthermore, one of the top negatively enriched pathways from the GSEA analysis of the screen was the cellular response to sterol suggesting that the losing the CRPs disrupts aspects of cholesterol homeostasis. Lastly, there is evidence that CD59 interacts with SNARE proteins

111

exocytic SNARE proteins VAMP2 and Syntaxin-1 thus highlighting its involvement in vesicle mediated transport (Krus et al. 2014). Taken together, it is possible that the CRPs may be coordinating aspects of vesicular-mediated intracellular cholesterol transport, and therefore the loss of any member of the NGLY1 pathway with the CRPs would result in synthetic lethality due to abrogation of both vesicular and non-vesicular modes of cholesterol trafficking within the cells. Further investigation into the potential role of the CRPs in secretory processes is warranted.

Changes in NFE2L1 processing, turnover, and localization are integral in orchestrating the cellular response to proteotoxic and cholesterol-induced stress (Widenmaier et al. 2017). Widenmaier et al. 2017 proposed a model where under a high cholesterol challenge, NFE2L1 detects and binds cholesterol that is escaping into the ER lumen through via its CRAC domain that is buried in the lumen. In binding to cholesterol, full-length NFE2L1 gets retained at the ER membrane resulting in the depression of a transcriptional program of coordinating cholesterol removal. I observed that under a high-cholesterol challenge, the loss of the CRPs resulted in aberrant processing of NFE2L1. Strikingly, there was a greater accumulation of full-length NFE2L1 over time compared to WT, however NFE2L1 also continued to be cleaved as shown by the steady presence of cleaved-NFE2L1 which unexpectedly, was not detected in the nucleus.

The presence of both cleaved and full-length NFE2L1 in CRPΔ cells was reminiscent of an observation by Widenmaier et al. 2017 in which a CRAC domain mutant of NFE2L1 was observed as both the full-length and cleaved forms of NFE2L1 in equal abundance. However, CRAC mutant protein levels did not increase upon exposure to cholesterol, as NFE2L1 could no longer bind escaping cholesterol, and furthermore, the mutant was transcriptionally active as observed by a diminished expression of genes coordinating cholesterol removal. These findings suggest that in CRP knockout cells, NFE2L1’s ability to detect cholesterol escaping from the ER membrane remains intact. However, NFE2L1 accumulation likely remains unchecked and is therefore in excess, and in turn, may be retrotranslocated and subsequently cleaved by DDI2. This dysregulation however, might not result in impaired depression of a cholesterol removal program as the cleaved NFE2L1 does not appear to enter the nucleus. Taken together, these findings predict a possible role for the CRPs in regulating NFE2L1 turnover in high cholesterol conditions.

112

Finally, the previous genome-wide CRISPR screens NGLY1Δ, NFE2L1Δ or DDI2Δ cells did not identify any CRPs as predicted GIs, which may be due to masking effects resulting from shared functional and structural similarities between the CRPs. However, this finding does not preclude the possibility that two and not all of the CRPs are involved in coordinating aspects of cholesterol homeostasis in connection to the NGLY1 pathway or in regulating NFE2L1 turnover. To address this question, NGLY1 pathway knockouts should be tested in both single and double CRP knockout lines.

Additionally, it is possible that the synthetic lethal phenotype observed in CRP-NGLY1 pathway knockouts is a consequence of editing multiple genes in a cell. To address this question, future studies should be carried out to rescue the synthetic lethal phenotype (i.e. expressing the CRPs) as well as investigate levels of ER stress in CRPΔ cells. Nevertheless, this study provides a foundation for interrogating CRP function in the absence of complement and future studies investigating the GIs of two or a single CRP would further our understanding of their unique and shared roles in cellular processes.

This study offers for the first time, insight into the biological functions of the CRPs that are independent of their roles as complement regulators from the perspective of GIs. This approach has allowed for the identification of GIs between the CRPs and members of the NGLY1 pathway and offers insight into possible complementary roles in coordinating intracellular cholesterol transport. Insights into the pathogenicity associated with rare diseases resulting from loss of CRP function have been largely ascribed to their complement-inhibitory functions or role in T-cell signaling via binding complement fragments (Ellinghaus et al. 2017; Astier et al. 2006; Astier and Hafler 2007; Ozen et al. 2017; Hill et al. 2017). Therefore, further investigation of the complement-independent roles for the CRPs would not only broaden our understanding of these proteins, but importantly illuminate their impact on human disease.

113

Chapter 4 Methods Methods 4.1 Cell Culture

All HAP1 cell lines used in this study were maintained in DMEM media (Wisent Inc.), with sodium bicarbonate, 1.982g/L glucose, 0.161g/L L-glutamine (Wisent Inc.), without sodium pyruvate and supplemented with 10% Heat inactivated FBS (Thermo Fischer Scientific) and 1% Penicillin/Streptomycin (Thermo Fischer Scientific).

4.2 Clonal Knockout Cell Lines 4.2.1 NGLY1, NFE2L1 and DDI2

HAP1 NGLY1Δ (HZGHC004929c006), DDI2Δ (HZGHC000396c007), NFE2L1Δ (HZGHC005847c001), VPS52Δ (HZGHC004879c006), VPS54Δ (HZGHC000854c012) and HAP1-WT cells lines were obtained from Horizon Genomics and their genotypes were confirmed via Sanger sequencing.

4.2.2 Generation of CRP Knockout Cell Lines

Guide RNA targeting CD55 (clone 1.3; forward: CTTGTACGGCCTTCCAAAGC)(clone 2.3; forward sequence: GAAGGTTCTCTTCTGTAACC) and CD59 (forward sequence: ATCACAATGGGAATCCAAGG) were each cloned into pSpCas9n(BB)-2A-Puro (PX459) V2.0) (Addgene). HAP1 CD46 knockout cells were obtained from Horizon (HZGHC003436c005) and 10µg of each of the above plasmids were transfected into the cells by electroporation (Neon Transfection System; Life Technologies) and seeded into a plate. After 24 hours the medium was replaced with medium containing puromycin (2µg/ml) and incubated for 48 hours before expansion. Medium was removed, cells were washed with PBS and then detached with 1mM filtered sterilized EDTA (Life Technologies) and resuspended in FACS buffer (PBS + 2% Heat Inactivated FBS). Keeping the cells on ice, 1 million cells were filtered in FACS tubes, centrifuged at 1250rpm for 5 min and the medium was removed. A primary antibody mix containing 10 µg/ml of CD46 antibody (lab reagent), 10 µg/mL CD55 antibody

114

(lab reagent) and 5 µg/mL CD59 antibody (Abcam, Cat.no. ab9182) was added to the cells for 30 min, then washed with FACS buffer and centrifuged for 5 min at 1250rpm twice. Cells were incubated for 30min with Alexa-488 conjugated (Thermo Fischer Scientific) and Alexa-633 conjugated (Thermo Fischer Scientific) secondary antibodies used at 5µg/ml then washed and centrifuged twice. Cells were filtered into a new tube and sorted for CD46-/CD55-/CD59- single cell clones by flow cytometry (BD FACSAria III) and seeded into a 96 well plate containing IMDM media with 20% Heat Inactive FBS and Penicillin/Streptomycin. Triple CRP negative clones were expanded and 250K cells were harvested for gDNA extraction.

4.3 Flow Cytometry

HAP1 wild-type, CD46 knockout or CRP knockout cells were incubated with primary antibodies to CD46, CD55, and CD59 (as previously described) in PBS containing 2% heat inactivated FBS (wash buffer) for 1 hour. After washing, the cells were incubated with the appropriate Alexa-488 conjugated secondary antibodies (Thermo Fischer Scientific) for 30 min and washed. Cell fluorescence was evaluated on a CytoFLEX instrument (Beckman Coulter) using CytExpert software.

4.4 Complement-Mediated Lysis

HAP1 cells/well were seeded in a 48 well plate for 24 hours. Lyophilized rabbit complement (Cedar Lane) was resuspended in supplemented DMEM media (previously listed) and filter- sterilized. Medium was removed, complement was diluted in media and adjusted to the desired concentration for a final volume of 500µl/well, added to the cells and incubated at 37°C for 6 hours. After incubation, medium was removed and 500µl of Alamar Blue (Thermo Fisher Scientific) was added to each well and cell viability was measured.

4.5 Genome-Wide Genetic Interaction CRISPR Screens

All the experimental steps for setting up pooled CRISPR-Cas9 genetic screens employed in this study were carried out as described in Hart et al, 2015 however, the CRISPR screen carried out in figure 2.2 was using the TKOv1 CRISPR library while the rest of the screens carried out were using TKOv3. Cells were infected with either TKOv1 or TKOv3 lentiviral libraries at a MOI~0.3. 24 hours after selection, cells were split into three replicates, passaged every 3 days, and maintained at 200-fold coverage. In the bortezomib screen, cells were split into treated and 115

untreated groups at T6, and the treatment group received an IC50 dose of bortezomib upon each passaging. Cells were collected at a 200-fold coverage for genomic extraction at day 0 and every 3 days from day 6 to day 18 post-selection. Genomic DNA was extracted from cell pellets. gRNA inserts were amplified via PCR using primers harboring Illumina TruSeq adapters with i5 and i7 barcodes, and the resulting libraries were sequenced on an Illumina HiSeq2500.

4.6 CRISPR/Cas9-Mediated Gene Targeting

Lentiviruses containing sgRNA targeting different genes mentioned in this study; LacZ (forward: CCCGAATCTCTATCGTGCGG), NFE2L1 guide 1 (Figures 2.2d and 2.4g; forward: AGGGCGCGGAAGCTCTGGCA ) , NFE2L1 guide 2 (Figure 2.4g; forward: ACAATTACTTCACTGCCCGG); NGLY1 guide 1 (Figures 2.2d and 2.4g; forward: GTTCTGGCAGAGCTCAGCCA), NGLY1 guide 2 (Figure 2.4g; forward: AGTTCAGCAATCGATTCCCA); DDI2 guide 1 (Figures 2.2d and 2.4g; forward: ACTGAGCAGAGCTTCTGCCA), DDI2 guide 2 (Figure 2.4g; forward: GCGTACCTGGCTCTCGGCTG) were generated in 293T cells along with the viral envelope plasmid pMD2.G and the packaging plasmids psPAX2 (Addgene). Successfully transduced cells were selected using puromycin (1 μg/ml) for 48h.

4.7 Gene Set Enrichment Analysis

A ranked list of genes was created as a .txt file and used as the input for the analysis. The GSEA was carried out using the default parameters as outlined in http://www.broadinstitute.org/gsea.

4.8 RNA Sequencing

NGLY1Δ, NFE2L1Δ and DDI2Δ cells were treated with Bortezomib at an ~IC40 dose for 72 hours, in duplicate. Following this, total RNA was isolated from 5 million cells per sample using the RNeasy Mini Kit (Qiagen, Cat.no. 74104). Total RNA was quantified using Qubit RNA BR (Thermo Fisher Scientific). Stranded mRNA sequencing libraries were prepared using the TruSeq Stranded mRNA Library Prep kit (Illumina). Sequencing was carried out on the Illumina NextSeq500 to a target depth of 20 million reads per sample. Reads were trimmed of adaptor contamination using the bcl2fastq software, and reads shorter than 36bp were removed using a bespoke Perl script. Reads were mapped to the human genome version hg38 and Gencode v25

116

gene definitions using the STAR short-read aligner (v2.4.2a). Reads counts per gene for each sample were then merged into a single matrix for differential expression analysis.

Differential expression was performed in R using the Bioconductor package ‘limma’. Only genes with expression more than 0.5 CPM in at least two samples were retained for analyses. Reads were normalized using the TMM method and observation-level weights estimated using voom. To identify genes specific to the mutant backgrounds, simple linear models were used. Where treatments were compared across mutant backgrounds, an interaction term was added to the linear model. Significant genes were selected with an FDR < 0.05.

4.9 Immunoblots

The cell-lines used for the immunoblot analysis were harvested and washed with 1xPBS (Life Technologies) and then lysed in a buffer comprised of 50mM TRIS-HCL (pH 8.0), 1mM EDTA, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 150mM NaCl together with HALT protease inhibitor cocktail (Thermo Fisher Scientific, Cat.no. 1861280). Lysates were then centrifuged at 4°C for 20min and protein concentration was estimated via a Bradford assay (Thermo Fisher Scientific). Protein samples were run in 4%-12% NuPAGE Bolt Bis-Tris Plus protein gels in a NuPAGE Bolt MES SDS running buffer. The protein samples were then transferred onto nitrocellulose membranes (Thermo Fisher Scientific) and later blocked in TBST with 5% milk. The membranes were incubated overnight in TBST with 1% milk with primary antibodies with gentle shaking. Next, after 3 x 5min washes with TBST, anti-mouse or anti-rabbit HRP- conjugated secondary antibodies (Santa Cruz Biotechnology), were applied to the membranes for 1 hour. After 6x5min washes with TBST, membranes were incubated for ~ 3min with the Super Signal West Femto Maximum Sensitivity chemiluminescent Substrate (Thermo Fisher Scientific) and analyzed using the MicroChemi digital imaging system (from FroggaBio).

4.10 Cholesterol Solubilization

Cholesterol solubilization and the subsequent treatment of cells with cholesterol was performed as described in Widenmaier et. al 2017. This method causes cells to accumulate cholesterol and circulate through the normal cellular cholesterol pools. Dry cholesterol powder (Sigma Aldrich) was solubilized in 100% ethanol at 65°C and then added to methyl-b-cyclodextrin (MbCD) (Sigma Aldrich) in water for a final concentration of 5 mM cholesterol to 42 mg/ml MbCD,

117

which was then cooled and sterile filtered. Cells were with cholesterol for a final concentration of 100mM.

4.11 Subcellular Fractionation

Cells were seeded and incubated for 37°C overnight. The following day, media was removed and replaced with media containing a final concentration of 50nM epoxomycin or 100 µM cholesterol. Cells were treated for 6 hours and the cytoplasmic and nuclear protein extracts were obtained using a kit for Subcellular Protein Fractionation Kit for Cultured Cells (Thermo Fischer). Fractionation was carried out following the protocol provided with the kit.

4.12 Confocal Microscopy

HAP1 wild-type or CRP knockout cells were seeded on tissue-cultured treated polystyrene plates from Perkin Elmer and incubated at 37°C overnight. Cells were washed with PBS then fixed with 4% paraformaldehyde (PFA) in PBS for 30 min, washed again then permeabilized in 0.1% Triton X-100 for 10 min. Cells were then blocked with 1% BSA in PBS for 1hour with gentle shaking then washed twice. Cells were incubated with an anti-EEA1 primary antibody (Cell Signaling Technologies) and Alexa-647 conjugated Phalloidin (Thermo Fischer Scientific), overnight at 4°C in the dark, washed with PBS three times and incubated with a secondary antibody (Thermo Fischer Scientific) for 1hour. Cells were washed three times and Hoechst dye was added before visualization. Wells were approximately 40% confluent. Fluorescent images were taken in an Opera Phenix High Content Screening Platform with a confocal microscope at a 40x water immersion objective. Images were analyzed with Harmony Imaging Software with the following parameters: cell surface area; image region with common threshold = 0.3, cholesterol; find spots method D, endosomes: find spots method D with detection sensitivity = 0.2 and splitting sensitivity = 0.378. 140 fields were analyzed per well.

4.12.1 Cholesterol Staining for Confocal Microscopy

Cells were left untreated or incubated with 1.25 µM of the cholesterol tracking inhibitor U18666a (Abcam) 1.25 µM for 24 hours. Cells were washed with PBS with 4% PFA in PBS for 30 min, then permeabilized in 0.1% Triton X-100 for 10 min. Cells were then blocked with 1% BSA in PBS for 1hour with gentle shaking then washed twice. A stock solution of Filipin III (Abcam) was prepared in 100% ethanol and diluted 1:100 in PBS for 1 hour at 4°C in the dark. 118

119

Chapter 5 Discussion Discussion

While each of the estimated 7000 rare diseases are individually rare (Hmeljak and Justice 2019), they collectively affect up to 400 million people worldwide (de Vrueh, Baekelandt, and de Haan 2013). The scientific landscape for rare diseases has been rapidly changing and accelerating, largely owing to significant advancements in technology. The application of next generation sequencing has ushered in an unprecedented era of rare disease gene discovery such that up to 80% of rare diseases have been linked to genetic abnormalities (Boycott et al. 2013). Furthermore, scientific advancements in treatments such as gene-based therapies, continue to offer hope for many patients living with monogenic rare diseases. However, despite these significant advancements, less than 5% of rare disease patients currently have an effective treatment (Kaufmann, Pariser, and Austin 2018). A key challenge remains in how the discovery of rare disease-causing genes is translated into a mechanistic understanding of the gene’s function, the consequence of disrupting its function, and ultimately the identification of therapeutic targets.

Studies in yeast have proven that GIs and in particular GIs such as synthetic lethality, offer a wealth of information about a gene’s function (Costanzo et al. 2010, 2016;Tong et al. 2001;Tong et al. 2004). Many non-essential genes, impinge upon essential biological functions, however, the cell is wired with complex back up pathways that buffer the loss of a gene’s function (Costanzo et al. 2016). The essential functions of non-essential gene pairs can be examined in GI screens where the loss of two genes impinging upon the same function, results in synthetic lethality in which the cell is no longer viable (Costanzo et al. 2010, 2016;Tong et al. 2001; 2004). As synthetic lethal GIs often occur between functionally related genes, examining digenic synthetic lethal combinations provides insight into gene function (Costanzo et al. 2016). Importantly, genes that belong to similar biological processes tend to have similar genetic interaction profiles, and grouping genes according to their GI profiles is powerful way to predict gene function (Tong et al. 2004; Costanzo et al. 2010).

120

Exploring global GI profiles in human cells had remained a distant dream for a long time mainly due to lack of sufficient technological advancements. However, owing to revolutionary gene- editing technologies such as CRISPR (Cong et al. 2013) over the past five years, GI mapping on a global scale in human cells has now become feasible and early work has revealed that several properties of genetic networks mapped in model organisms are conserved in human cells (Blomen et al. 2015). Therefore, mapping GIs of monogenic rare-disease associated genes in human cells will undoubtedly help bridge the gap in understanding gene function, help characterize genes of unknown function and provide insight into potentially targetable pathways. Using a CRISPR-based genome wide mapping of GIs in the human HAP1 cell line, I along with my collaborators, have uncovered novel functional insights into the rare-disease associated genes namely NGLY1 and the complement regulatory proteins.

NGLY1-deficiency is a complex, multi-system rare disorder and is the first congenital disorder of deglycosylation described in the biomedical literature (Need et al. 2012). Clinically, affected individuals present with global developmental delay, a complex hyperkinetic movement disorder, small body size, seizures, and an absence of tears, however, the specific symptoms and severity vary dramatically among patients (Chang, He, and Lam 2018). The diverse and debilitating symptoms of NGLY1-deficiency clearly highlight the complexity of the disease and functional importance of NGLY1 enzymatic activity. In the five years since the clinical presentation of a cohort of NGLY1-deficiency was first described (Enns et al. 2014), multiple research groups have contributed to our understanding of the consequences of losing NGLY1 function using animal, cell-culture, and patient-derived cell models (Lehrbach and Ruvkun 2016; Huang et al. 2015; Yang et al. 2018; Kong et al. 2018; Owings et al. 2018; Rodriguez et al. 2018; Galeone et al. 2017).

NGLY1 is thought to function is the protein disposal ERAD pathway and has since been linked to mitochondrial homeostasis, developmental and inflammatory pathways through regulating the activity of specific glycoproteins, such as NFE2L1 (Suzuki 2015; Huang et al. 2015; Yang et al. 2018; Kong et al. 2018; Lehrbach and Ruvkun 2016; Galeone et al. 2017). Disruption of these biological processes offers mechanistic insight into NGLY1-deficiency symptoms as proteasome dysfunction is known to cause adult-neurodegenerative diseases (Pilla, Schneider, and Bertolotti 2017), impaired mitophagy (Yang et al. 2018) may explain some of the known abnormal mitochondrial physiology that is observed in patient cells (Kong et al. 2018), and lastly 121

inflammation (Yang et al. 2018) may contribute to brain abnormalities and neuropathy in patients (Lam et al. 2017). Furthermore, the fact the NGLY1 functions as the only de-N- glycosylating enzyme in the cell, whose function is also critical to NFE2L1’s transcriptional activity, clearly highlights how genes do not function in isolation and thus the loss of a single gene’s function has widespread consequences.

Although there have been significant advancements in our understanding of NGLY1’s function since it was first identified as a disease-causing gene, how the loss of NGLY1 function contributes to the pathogenesis of NGLY1-deficiency has not been resolved. However, what may now be appreciated is that NGLY1 regulates diverse biological processes that are critical for cellular homeostasis and therefore the loss of NGLY1 activity has a “ripple effect” across the genome. With that in mind, systematic mapping of NGLY1’s GIs is a valuable approach for capturing its involvement these processes and provides a platform for gaining a deeper mechanistic understanding of NGLY1 function.

Indeed, genome-wide CRISPR screening of NGLY1 knockout cells highlighted the power of GIs in uncovering meaningful biological relationships that offer insight into NGLY1 function. Importantly, the predicted negative GIs from the screen comprised pathways in which NGLY1’s function is already known to be critical, as well as novel processes in which NGLY1 had not yet been implicated in. For example, predicted GIs included genes involved in mitochondrial function and OXPHOS, further corroborating previous studies that demonstrated across multiple model systems a role for NGLY1 in these processes. Interestingly, it also predicted that NGLY1 may contribute to mitochondrial and cellular homeostasis via the TCA cycle. These findings served to corroborate a role for NGLY1 in mitochondrial function (Kong et al. 2018; Yang et al. 2018) and offers further avenues for investigation.

Many of the strongest negative GIs of NGLY1 included occurred with genes/complexes that are known to function in the secretory pathway. These included genes that function in GPI anchor biosynthesis (phosphatidylinositol glycan classes; PIGs) as well as genes that comprised complexes that regulate aspects of vesicular substrate trafficking; the Conserved Oligomeric Golgi (COG) and Golgi-Associated Retrograde Protein (GARP) complexes. Disrupting secretory pathway functions contributes to the pathogenesis of several diseases, highlight the importance of secretory functions in normal cellular homeostasis. For example, mutations in seven of the

122

eight COG subunit genes are linked to Congenital Disorders of Glycosylation type II (CDG-II) (Climer, Dobretsov, and Lupashin 2015), whereas mutations in the GARP complex cause a neurodegenerative disease known as progressive cerebello-cerebral atrophy type 2 (PCCA2) (Feinstein et al. 2014). Furthermore, patients with COG deficiencies and PCCA2 share some of the neurological symptoms observed in individuals with NGLY1-deficiency (Feinstein et al. 2014; Climer, Dobretsov, and Lupashin 2015). Taken together, the negative GI profile of NGLY1 predicted a novel function for NGLY1 in regulating the secretory pathway and offered insight into how disrupting this process could contribute to disease pathogenesis.

In taking advantage of the fact that genes that function in similar biological processes share overlapping GIs, I along with my collaborator, screened all members of the NGLY1 genetic pathway to gain broader insight into NGLY1’s predicted secretory pathway function. Indeed, we uncovered shared genetic relationships with several genes that function in the secretory pathway including members of the GARP complex. Additional studies further suggested that the synthetic lethal relationship between the NGLY1 pathway and GARP complex may be due their complimentary roles in vesicular and non-vesicular transport, including that of key substrates such as cholesterol. RNA-Seq analysis revealed that this putative link between NGLY1 and non- vesicular transport is likely through enzymatic processing of NFE2L1 and its subsequent regulation of a transcriptional program.

NFE2L1 is now known to function as an ER sensor of cholesterol levels and a transcriptional regulator of cholesterol homeostasis (Widenmaier et al. 2017). My findings further support a role for NFE2L1 is cholesterol homeostasis and extend its involvement to possible regulation of non- vesicular transport. Importantly, it also for the first time, links NGLY1 to this process which may offer additional insights into NGLY1-deficiency clinical manifestations as impaired cholesterol trafficking is involved in the pathogenesis of neurodegenerative diseases.

RNA-Seq analysis of NGLY1-, NFE2LI- and DDI2 -knockout cells also revealed that the NGLY1 pathway likely regulates inflammatory response pathways such as the interferon alpha response and the complement system. I observed that two of the three CRPs expressed in the surface of HAP1 cells, CD46 and CD55, were down regulated across all NGLY1 pathway knockout cells. This finding was interesting as NGLY1 has been shown to regulate inflammation in an NFE2L1- dependent manner however (Yang et al. 2018). Reduced expression or impaired function of

123

CRPs results in a chronic proinflammatory state and has been shown to contribute to the pathogenesis of autoimmune, inflammatory and neurodegenerative diseases (Cardone, Le Friec, and Kemper 2011; Dho, Lim, and Kim 2018; Orsini et al. 2014). Therefore, impaired NFE2L1 transcriptional activity, and consequently the downregulation of CRP expression, may contribute to the observed pro-inflammatory state in NGLY1-deficiency.

Interestingly, Widenmaier et al. 2017 also demonstrated that NFE2L1 regulates CD36, a scavenger receptor that facilitates lipid uptake and impacts inflammation through forming signaling complex, and that the loss of NFE2L1 (resulting in de-repression of CD36) contributes to the liver damage observed in hepatocyte-NFE2L1 deficient mice. NGLY1 has not been shown to be involved in the cholesterol removal program delineated by Widenmaier et al. 2017, however, upon examination of the RNA sequencing data, I observed that NGLY1 knockout cells displayed increased expression of CD36 compared to wild-type cells. Therefore, it’s possible that through NFE2L1’s regulation of various immune-related genes, NGLY1 is further contributing to liver pathology and inflammation observed in NGLY1-deficiency patients.

Findings from my research, along with other groups, supports that a loss of NGLY1 contributes to impaired mitochondrial function, impaired lipid metabolism and cholesterol homeostasis, and results in inflammation. It is likely that dysregulation of these processes does not occur independent of each other and together may contribute to disease pathogenesis in NGLY1- deficiency. Impaired mitochondrial function in NGLY1-deficient cells (Yang et al. 2018; Kong et al. 2018), as a result of impaired NFE2L1-mediated regulation of mitophagy, and activates inflammatory responses (Yang et al. 2018). This “anti-viral state” may be linked to the transcriptional downregulation of CRPs as they have been observed to be downregulated in several different cell types in response to infection (Thurman and Renner 2011). This may occur in order activate complement and limit the spread of a perceived infection, but chronic activation as a result of impaired mitophagy, may cause inflammatory damage to the tissue. It is also a possibility that impaired mitochondrial function in NGLY1-deficient macrophages could also play a role in inflammation as mitochondrial metabolism governs macrophage activation. While pro-inflammatory M1 macrophage activation is driven by glycolysis, OXPHOS supports activation of anti-inflammatory M2 macrophages, which are also characterized by increased fatty acid oxidation (Tur et al. 2017). In a mouse model of mitochondrial complex I deficiency, the mice show increased systemic inflammation and their macrophages switch to an inflammatory 124

phenotype, highlighting the critical role of mitochondrial metabolism in macrophage function and inflammation (Jin et al. 2014).

Mitochondria also play a central role in lipid metabolism, as they are the site of fatty acid catabolism to generate acetyl-CoA for the TCA cycle while acetyl-CoA generated in the mitochondria from pyruvate oxidation is used for fatty acid synthesis (Houten and Wanders 2010; Kastaniotis et al. 2017). Therefore, impaired mitochondrial function in NGLY1-deficiency could impact lipid metabolism and contribute to liver pathology. One of the symptoms of NGLY1-deficient patients is abnormal liver function and steatosis (Enns et al. 2014; Lam et al. 2017) and abnormal liver mitochondrial morphology has also been observed in patient biopsies (Kong et al. 2018). While Fujihira et al. 2019 demonstrated that lipid metabolism was impaired in Ngly1-deficient mouse livers under food stress, and transcriptome analysis of the livers on a regular diet did suggest that they were undergoing an oxidative stress, they did not observe any abnormal mitochondrial morphology (Fujihira et al. 2019). Nevertheless, this possible connection warrants further investigation.

Lastly, impaired cholesterol homeostasis in NGLY1-deficient cells may also impact mitochondrial function. Mitochondria require cholesterol for proper functioning and are highly sensitive to changes in cholesterol content (Martin, Kennedy, and Karten 2016). Indeed, increased mitochondrial cholesterol levels are associated with impaired mitochondrial function which has been observed in several diseases including steatohepatitis and neurodegenerative diseases (Martin, Kennedy, and Karten 2016). For example, in a mouse model of the lysosomal storage disorder NPC-1 where cholesterol accumulation in late endosomes, elevated mitochondrial cholesterol is also observed and affects mitochondrial function (Yu et al. 2005) which may be due to increased cholesterol trafficking from the late endosomes to the mitochondria via lipid transfer protein STARD3 (Charman et al. 2010). Therefore, cholesterol accumulation in NGLY1-deficient cells may disrupt mitochondrial function and further characterization of cholesterol abundance (i.e. within specific organelles) and transport would offer valuable insight.

It’s unclear if dysregulation of one of the above processes, such as impaired mitochondrial function, is an initiating factor in disrupting additional biological processes in NGLY1- deficiency, or if they occur in parallel but nevertheless impact each other. Nevertheless, NFE2L1

125

appears to be a key player in several of these processes and further investigation of its role in NGLY1-deficiency may shed light on its involvement in disease pathogenesis.

While my work offers further insight into how the relationship between NGLY1 and NFE2L1 may contribute to NGLY1-deficiency, it also poses the question of whether or not NGLY1 has additional substrates. The GI profiles as well as the GSEA enrichment pathways of individual NGLY1 pathway members highlighted that NGLY1 has both shared and independent functions that may contribute to cellular homeostasis. Employing protein interaction techniques such as BioID may reveal substrates of NGLY1 and offer further insight into NGLY1-deficiency. Given that NGLY1 was recently shown to regulate aquaporin expression independent of its catalytic activity (Tambe, Ng, and Freeze 2019), identifying NGLY1 substrates may reveal distinct sets of proteins that either require NGLY1 for de-glycosylation or for another aspect of processing.

Predictions from the CRISPR screens, as well as work from other groups, highlights a role for NGLY1 in mitochondrial physiology (Kong et al. 2018; Yang et al. 2018). Interestingly, a recent glycoproteomic study using mouse brain tissue identified N-linked glycosylated forms of several subunits of the mitochondrial protein import complex (Fang et al. 2016), which may implicate N- glycosylation or de-N-glycosylation in the regulation of mitochondrial proteins.

Identifying possible NGLY1 substrates may also provide valuable insight into potential treatment strategies for NGLY1-deficiency. For example, the absence of NGLY1’s de-N- glycosylating activity may lead to the accumulation of specific glycoproteins which may disrupt downstream signaling pathways or result in cellular toxicity. Therefore, therapeutic interventions directed towards reducing accumulation of the substrate may alleviate cell stress and help restore homeostasis.

Alternatively, identifying NGLY1 substrates may reveal targets whose increase in activity may compensate for impaired NGLY1 substrate function. Recently, Yang et al. 2018 demonstrated how targeting Keap1 may compensate for the loss of NFE2L1 activity in NGLY1- deficient cells and animal model systems. As discussed in Chapter 1, both NFE2L1 and NFE2L2 bind to ARE and transcriptionally activate the expression of proteasome subunit genes, but differ in their response cues and processing. NFE2L2 responds to oxidative stress, resides in the cytosol and is not glycosylated (Kensler, Wakabayashi, and Biswal 2006). Under normal conditions, NFE2L2 binds to KEAP1 which mediates its ubiquitination and degradation (Kensler, Wakabayashi, and 126

Biswal 2006). Yang et al. 2018 found that treating Ngly1-/- mouse embryonic fibroblast cells with Keap1 inhibitor sulforaphane, increased Nrf2 expression which resulted in increased expression of mitophagy-related genes, restored mitochondrial integrity and suppressed immune activation. Additionally, drug repurposing screens of NGLY1-deficiency in worm and fly models also identified an NRF2 activator as a potential therapeutic approach (Iyer et al. 2019). However, Lehrbach et al. 2019 demonstrated that in C. elegans, SKN-1A and SKN-1C, which functions analogously to NFE2L2, regulate distinct target genes. Therefore, NFE2L2 activators may alleviate some symptoms that arise due to impaired mitochondrial function and inflammation but may not restore basal proteasome function.

Developing effective therapies for multi-systemic disorders such as NGLY1-deficiency is undoubtedly a great challenge. Presently, small molecule activators of NFE2L1, enzyme- replacement therapy and gene therapy are being pursued as therapeutic strategies for treating NGLY1-deficiency (Grace Science LLC). While gene therapy holds great promise in treating monogenic diseases such as NGLY1-deficiency, many hurdles must first be overcome. Key challenges include how to target the correct tissues and ensure that the gene is (and remains) activated and that it does not cause an immune response in the patient (Sun, Zheng, and Simeonov 2017).

As NGLY1’s function continues to be unraveled, one can envision how its general activity as a de-N-glycosylating enzyme, as well as regulator of specific proteins, such as NFE2L1, disrupts numerous biological processes that may account for the observed clinical manifestations of NGLY1-deficiency. Our findings linking NGLY1 to secretory processes via NFE2L1 processing, and possibly other aspects of immune function such as the complement cascade, offers an additional piece of the puzzle as researchers continue to determine the biological consequences of losing NGLY1 function. Lastly, identifying other potential substrates that are regulated by NGLY1 activity, which may be also be independent of NGLY1 catalytic activity, will offer critical insight into the disrupted pathways in NGLY1-deficiency and direct future therapeutic efforts.

Determining the biological processes in which a rare-disease gene is involved in, not only offers insight into gene function and disease manifestations, but may also help direct drug screening efforts. Drug repurposing continues to be an important route for orphan drug development (Xu

127

and Coté 2011) and thus screening for drugs that are known to function in the pathways affected by the faulty gene may help discover effective therapies sooner. For example, investigating drugs that restore cholesterol homeostasis may offer a new avenue for alleviating some of the symptoms associated with NGLY1-deficiency. Cyclodextrin, a compound that solubilizes cholesterol, is currently being explored in clinical trials for the treatment of NPC, a neurological disease characterized by the accumulation of cholesterol in lysosomes (Ory et al. 2017). However, identifying quantitative phenotypes is critical for drug screening to identify candidate therapeutics that may alleviate symptoms associated with disruption of given pathways/processes due to LOF (Sun, Zheng, and Simeonov 2017). For example, in a fly model of NGLY1- deficiency, the mutants display global development delay, pupal lethality and small body size as adults and therefore (Rodriguez et al. 2018; Iyer et al. 2019) carried out image-based quantitative screens of FDA-approved drugs using larval size as a phenotypic read-out. In the context of cell- based screening approaches, GIs offer insight into the disrupted pathways due to a LOF in a disease-associated gene and may help delineate quantifiable cellular phenotypes for image-based drug screens.

Predicted GIs from the CRISPR screen helped identify biological processes in which NGLY1’s function may be critical such as mitochondrial homeostasis and cholesterol trafficking. The phenotypic consequences of impaired NGLY1 function in secretory processes was observed in the accumulation of cholesterol particles and early endosomes by immunofluorescence, which offers new opportunities for drug screening in NGLY1 deficient cells. As additional phenotypes of NGLY1 deficient cells continue to be determined, machine learning approaches secretory defects or other quantifiable phenotypes, may be coupled with drug screening to provide a phenotypic read out for potential drug candidates.

Employing unbiased genome-wide CRISPR screens to uncover GIs of NGLY1, coupled with RNA-Seq analysis, helped delineate a role for the NGLY1 pathway in regulating secretory processes via NFE2L1. Furthermore, it also highlighted a possible role in regulating immune pathways including the complement system and some of the CRPs were coordinately downregulated across NGLY1 pathway knockout cells. The protective role of the CRPs as membrane-bound inhibitors of systemic complement activation has been well understood for many decades. Consequently, changes in their expression or function resulting in excessive or restricted complement has been linked to the pathogenesis of many diseases (Astier 2008; Dho, 128

Lim, and Kim 2018; Ellinghaus et al. 2017; Liszewski and Atkinson 2015; Das et al. 2019; Hill et al. 2017; Fremeaux-Bacchi et al. 2006). However, there is now a shift in our understanding of the role of complement in disease pathogenesis. For example, the contribution of complement to cancer pathophysiology is now viewed as largely contextual; complement activation can act as both a positive and negative regulator of tumorigenesis (Reis et al. 2018), and the discovery of intracellular complement activation has highlighted its important role in T cell responses (Kolev and Kemper 2017) and thus autoimmune diseases (Astier et al. 2006; Xu et al. 2010). Furthermore, the CRPs have now been linked to systemic complement-independent roles in T cell responses (John Cardone et al. 2010; Kolev et al. 2015) , metabolism (Hess and Kemper 2016), and secretory granule trafficking (Krus et al. 2014; Golec et al. 2019). Taken together, as the complexity of how complement contributes to disease continues to be unraveled, this too warrants a re-evaluation of how the CRPs contribute to disease pathogenesis in which unbiased GI studies may be of clear value. Therefore, I sought to gain insight into the complement- independent functions of the CRPs. In doing so, I uncovered a novel genetic relationship between the CRPs and the NGLY1 pathway that together may regulate opposing aspects of intracellular cholesterol trafficking.

The identification of predicted suppressors and sensitizers of complement in the absence of the CRPs was used to benchmark the screens for predicting biologically relevant GIs. However, further investigation into these GIs would provide insight into additional mechanisms that cells employ to protect themselves from complement activation. Many of the negatively enriched pathways encompassed genes involved in glycan synthesis. The enzymatic activity of glycosyltransferases as well as their gene expression is known to be involved in cancer pathogenesis and resistance to therapies (Wu et al. 2019). Given the intricate relationship between complement and cancer, as well as the role of glycans in complement activation, further investigation of how changes in surface glycans affect complement activation in cancer would perhaps offer additional insight into disease pathogenesis.

The GIs observed between the CRPs and all members of the NGLY1 pathway suggested that these genes may be functioning in parallel pathways that each impinge upon an essential biological function. However, it was interesting that none of the CRPs were predicted negative GIs of any of the NGLY1 pathway members. This observation suggested that structural or functional similarities between the CRPs in their activities as complement regulators, may limit 129

the detection of GIs and ultimately, biological pathways in which they are involved. This is an important observation as it suggests that some GIs between genes may not be detected by genome-wide screens if there are other genes within the same pathway that share highly similar functions. However, as discussed in chapter 3, investigating synthetic lethality in single and double CRPΔ backgrounds may offer further insight into this observation.

The GIs between the CRPs and all members of the NGLY1 pathway, as well as the secretory defects in CRP knockout cells observed by immunofluorescence, suggests that the CRPs may also be involved in secretory processes. If the NGLY1 pathway is in fact coordinating aspects of non-vesicular transport, the CRPs may function in vesicular transport and thus disruption of both pathways would be synthetic lethal. Additionally, one of the top negatively enriched pathways from the untreated CRP knockout screen included peroxisomal membrane transport. Trafficking of intracellular cholesterol from late endosomes/lysosomes to the peroxisome is critical and disruption of peroxisome genes has been shown to result in cholesterol accumulation in the lysosome (Chu et al. 2015).Therefore it’s also a possibility that the CRPs may be involved in cholesterol homeostasis through supporting peroxisome function.

Additionally, the tethering of peroxisomes to the ER via VAMP-associated proteins, is important for the maintenance of cellular cholesterol levels (Hua et al. 2017). To date, only CD59 has been linked to secretory functions; through interactions with SNARE proteins VAMP2 and Syntaxin- 1, a soluble form of CD59 coordinates the secretion of insulin from Islet cells (Krus et al. 2014). However, GPI-anchored proteins, such as CD55 and CD59 and cholesterol, are key components of lipid rafts which function as coordinated signaling platforms in cells (Simons and Ehehalt 2002). It is also possible that the loss of CD55 and CD59 may disrupt lipid raft organization and consequently, downstream signaling cascades. Therefore, the loss of genes involved in cholesterol trafficking, such as members of the NGLY1 pathway, as well as the loss of key GPI- anchored proteins, may impair signaling pathways critical for cellular homeostasis. Taken together, further characterization of secretory defects in CRP knockout cells, as well as determining if all or some of the CRPs are critical to this process, would lend further insight into their possible roles within the secretory pathway.

Genome wide CRISPR screens uncovered GIs that offered new insight into the biological functions of rare disease-associated genes NGLY1 and the CRPs. However, in order to gain

130

further insight into how these functions contribute to disease, a necessary consideration is if these genetic relationships are consistent in other cell lines of various tissue types. Furthermore, while GIs do offer a powerful approach for gaining insight into rare disease gene function, the consequence of specific mutations in the gene and how it disrupts biological processes are not captured. Indeed, although various mutations in the NGLY1 gene give rise to NGLY1-deficiency, patients with a particular homozygous mutation show more severe phenotypes (Enns et al. 2014). Additionally, NGLY1-deficiency patients with the same mutation display diverse phenotypes (Enns et al. 2014) thus highlighting the complexity of the genotype-phenotype relationship.

The genotype-phenotype problem remains a key challenge in understanding human diseases as phenotypes may arise from complex GIs involving three or more genes (Costanzo et al. 2019). Systematic quantification of complex GIs is an immense challenge, however, studies in yeast have illustrated the profound effect that complex GIs can have on phenotypes (Kuzmin et al. 2018; Mullis et al. 2018). Several RNAi and CRISPR technologies based on multiplexing shRNAs and gRNAs have been developed towards combinatorial mapping of GIs and will undoubtedly offer insight into the complexity of human diseases (Horlbeck et al. 2018).

In summary, a key challenge in developing effective therapies for rare diseases lies in understanding the functions of a rare disease-causing gene and how its loss contributes to disease pathogenesis. Systematic mapping of genetic interactions offers a window into gene function and may provide a wealth of information about the biological processes in which a gene is involved. By employing genome-wide GI CRISPR screens, my doctoral research uncovered novel roles for known rare-disease causing genes; NGLY1 and the complement regulatory proteins. My research offers critical insight into NGLY1’s function in the secretory pathway and how disrupting this biological process may contribute to NGLY1-deficiency pathogenesis. Additionally, I have carried out the first global genetic investigation of CRP function that is independent of their roles as complement regulators and uncovered a GI between the NGLY1 pathway that may be indicative of functions within the secretory pathway. Taken together, my research demonstrates the utility of employing genome-wide GI CRISPR screens in human cells to gain functional insight into rare disease-associated genes and offers a useful framework that may help direct future therapeutic efforts for rare diseases.

131

References

Acosta-Alvear, Diego, Min Y. Cho, Thomas Wild, Tonia J. Buchholz, Alana G. Lerner, Olga Simakova, Jamie Hahn, et al. 2015. “Paradoxical Resistance of Multiple Myeloma to Proteasome Inhibitors by Decreased Levels of 19S Proteasomal Subunits.” ELife 4 (September). https://doi.org/10.7554/eLife.08153.

Acosta, Juan, Judith Hettinga, Rudolf Flückiger, Nicole Krumrei, Allison Goldfine, Luis Angarita, and Jose Halperin. 2000. “Molecular Basis for a Link between Complement and the Vascular Complications of Diabetes.” Proceedings of the National Academy of Sciences of the United States of America 97 (10): 5450–55. https://doi.org/10.1073/pnas.97.10.5450.

Agana, Marisha, Julia Frueh, Manmohan Kamboj, Dilip R. Patel, and Shibani Kanungo. 2018. “Common Metabolic Disorder (Inborn Errors of Metabolism) Concerns in Primary Care Practice.” Annals of Translational Medicine 6 (24): 469–469. https://doi.org/10.21037/atm.2018.12.34.

Ajit, V, and N Sharon. 2009. “Chapter 1 Historical Background and Overview.” Essentials of Glycobiology, 1–21. https://doi.org/10.1101/glycobiology.3e.001.

Alegretti, Ana Paula, Laiana Schneider, Amanda Kirchner Piccoli, Odirlei Andre Monticielo, Priscila Schmidt Lora, João Carlos Tavares Brenol, and Ricardo Mac Hado Xavier. 2012. “Diminished Expression of Complement Regulatory Proteins on Peripheral Blood Cells from Systemic Lupus Erythematosus Patients.” Clinical and Developmental Immunology 2012. https://doi.org/10.1155/2012/725684.

Allen, Mark D., Alexander Buchberger, and Mark Bycroft. 2006. “The PUB Domain Functions as a P97 Binding Module in Human Peptide N-Glycanase.” Journal of Biological Chemistry. https://doi.org/10.1074/jbc.M601173200.

Altrich-VanLith, Michelle L., Marina Ostankovitch, Joy M. Polefrone, Claudio A. Mosse, Jeffrey Shabanowitz, Donald F. Hunt, and Victor H. Engelhard. 2006. “ Processing of a Class I-Restricted Epitope from Tyrosinase Requires Peptide N -Glycanase and the Cooperative Action of Endoplasmic Reticulum Aminopeptidase 1 and Cytosolic Proteases .” The Journal of Immunology 177 (8): 5440–50. 132

https://doi.org/10.4049/jimmunol.177.8.5440.

Anchisi, Laura, Sandra Dessì, Alessandra Pani, and Antonella Mandas. 2013. “Cholesterol Homeostasis: A Key to Prevent or Slow down Neurodegeneration.” Frontiers in Physiology 3 JAN (January): 1–12. https://doi.org/10.3389/fphys.2012.00486.

Arbore, Giuseppina, Claudia Kemper, and Martin Kolev. 2017. “Intracellular Complement − the Complosome − in Immune Cell Regulation.” Molecular Immunology 89 (May): 2–9. https://doi.org/10.1016/j.molimm.2017.05.012.

Arbore, Giuseppina, Erin E. West, Rosanne Spolski, Avril A.B. Robertson, Andreas Klos, Claudia Rheinheimer, Pavel Dutow, et al. 2016. “T Helper 1 Immunity Requires Complement-Driven NLRP3 Inflammasome Activity in CD4+ T Cells.” Science 352 (6292). https://doi.org/10.1126/science.aad1210.

Astier, Anne L. 2008. “T-Cell Regulation by CD46 and Its Relevance in Multiple Sclerosis.” Immunology. https://doi.org/10.1111/j.1365-2567.2008.02821.x.

Astier, Anne L., and David A. Hafler. 2007. “Abnormal Tr1 Differentiation in Multiple Sclerosis.” Journal of Neuroimmunology. https://doi.org/10.1016/j.jneuroim.2007.09.018.

Astier, Anne L., Gregory Meiffren, Samuel Freeman, and David A. Hafler. 2006. “Alterations in CD46-Mediated Tr1 Regulatory T Cells in Patients with Multiple Sclerosis.” Journal of Clinical Investigation. https://doi.org/10.1172/JCI29251.

Badaut, Jérôme, Andrew M. Fukuda, Amandine Jullienne, and Klaus G. Petry. 2014. “Aquaporin and Brain Diseases.” Biochimica et Biophysica Acta - General Subjects. Elsevier. https://doi.org/10.1016/j.bbagen.2013.10.032.

Baines, Andrea C., and Robert A. Brodsky. 2017. “Complementopathies.” Blood Reviews. Churchill Livingstone. https://doi.org/10.1016/j.blre.2017.02.003.

Baliga, Nitin S, Johan LM Bjö rkegren, Jef D Boeke, Michael Boutros, Nigel PS Crawford, Aimé M Dudley, Charles R Farber, et al. 2017. “The State of Systems Genetics in 2017.” Cell Systems Commentary 8: 18. https://doi.org/10.1016/j.cels.2017.01.005.

133

Barilla-labarca, Maria L, M Kathryn Liszewski, John D Lambris, Dennis Hourcade, and John P Atkinson. 2002. “Role of Membrane Cofactor Protein (CD46) in Regulation of C4b and C3b Deposited on Cells 1.”

Baryshnikova, Anastasia, Michael Costanzo, Yungil Kim, Huiming Ding, Judice Koh, Kiana Toufighi, Ji Young Youn, et al. 2010. “Quantitative Analysis of Fitness and Genetic Interactions in Yeast on a Genome Scale.” Nature Methods 7 (12): 1017–24. https://doi.org/10.1038/nmeth.1534.

Bayly-Jones, Charles, Doryen Bubeck, and Michelle A. Dunstone. 2017. “The Mystery behind Membrane Insertion: A Review of the Complement Membrane Attack Complex.” Philosophical Transactions of the Royal Society B: Biological Sciences. Royal Society. https://doi.org/10.1098/rstb.2016.0221.

Bernales, Sebastián, Feroz R. Papa, and Peter Walter. 2006. “Intracellular Signaling by the Unfolded Protein Response.” Annual Review of Cell and Developmental Biology 22 (1): 487–508. https://doi.org/10.1146/annurev.cellbio.21.122303.120200.

Bexborn, Fredrik, Per Ola Andersson, Hui Chen, Bo Nilsson, and Kristina N. Ekdahl. 2008. “The Tick-over Theory Revisited: Formation and Regulation of the Soluble Alternative Complement C3 Convertase (C3(H 2 O)Bb).” Molecular Immunology 45 (8): 2370–79. https://doi.org/10.1016/j.molimm.2007.11.003.

Bi, Yiling, Matthew Might, Hariprasad Vankayalapati, and Balagurunathan Kuberan. 2017. “Repurposing of Proton Pump Inhibitors as First Identified Small Molecule Inhibitors of Endo-β-N-Acetylglucosaminidase (ENGase) for the Treatment of NGLY1 Deficiency, a Rare Genetic Disease.” Bioorganic and Medicinal Chemistry Letters 27 (13): 2962–66. https://doi.org/10.1016/j.bmcl.2017.05.010.

Birsoy, Kivanç, Tim Wang, Walter W. Chen, Elizaveta Freinkman, Monther Abu-Remaileh, and David M. Sabatini. 2015. “An Essential Role of the Mitochondrial Electron Transport Chain in Cell Proliferation Is to Enable Aspartate Synthesis.” Cell 162 (3): 540–51. https://doi.org/10.1016/j.cell.2015.07.016.

Blackwell, T. Keith, Michael J. Steinbaugh, John M. Hourihan, Collin Y. Ewald, and Meltem 134

Isik. 2015. “SKN-1/Nrf, Stress Responses, and Aging in Caenorhabditis Elegans.” Free Radical Biology and Medicine. Elsevier Inc. https://doi.org/10.1016/j.freeradbiomed.2015.06.008.

Blom, A. M. 2017. “The Role of Complement Inhibitors beyond Controlling Inflammation.” Journal of Internal Medicine 282 (2): 116–28. https://doi.org/10.1111/joim.12606.

Blomen, V, P Majek, L Jae, J Bigenzahn, and Joppe Nieuwenhuis. 2015. “Gene Essentiality and Synthetic Lethality in Haploid Human Cells.” Science 350 (6264): 1092–96.

Bohana-Kashtan, Osnat, Lea Ziporen, Natalie Donin, Sarah Kraus, and Zvi Fishelson. 2004. “Cell Signals Transduced by Complement.” Molecular Immunology 41 (6–7): 583–97. https://doi.org/10.1016/j.molimm.2004.04.007.

Bohlson, Suzanne S., Sean D. O’Conner, Holly Jo Hulsebus, Minh Minh Ho, and Deborah A. Fraser. 2014. “Complement, C1Q, and C1q-Related Molecules Regulate Macrophage Polarization.” Frontiers in Immunology. https://doi.org/10.3389/fimmu.2014.00402.

Bonifacino, Juan S., and Aitor Hierro. 2011. “Transport According to GARP: Receiving Retrograde Cargo at the Trans-Golgi Network.” Trends in Cell Biology 21 (3): 159–67. https://doi.org/10.1016/j.tcb.2010.11.003.

Boycott, Kym M., Ana Rath, Jessica X. Chong, Taila Hartley, Fowzan S. Alkuraya, Gareth Baynam, Anthony J. Brookes, et al. 2017. “International Cooperation to Enable the Diagnosis of All Rare Genetic Diseases.” American Journal of Human Genetics 100 (5): 695–705. https://doi.org/10.1016/j.ajhg.2017.04.003.

Boycott, Kym M., Megan R. Vanstone, Dennis E. Bulman, and Alex E. MacKenzie. 2013. “Rare-Disease Genetics in the Era of next-Generation Sequencing: Discovery to Translation.” Nature Reviews Genetics. https://doi.org/10.1038/nrg3555.

Buanec, H. Le, M.-L. Gougeon, A. Mathian, P. Lebon, J.-M. Dupont, G. Peltre, P. Hemon, et al. 2011. “IFN- and CD46 Stimulation Are Associated with Active Lupus and Skew Natural T Regulatory Cell Differentiation to Type 1 Regulatory T (Tr1) Cells.” Proceedings of the National Academy of Sciences 108 (47): 18995–0.

135

https://doi.org/10.1073/pnas.1113301108.

Byrne, Alexandra B, Matthew T Weirauch, Victoria Wong, Martina Koeva, Scott J Dixon, Joshua M Stuart, and Peter J Roy. 2007. “A Global Analysis of Genetic Interactions in Caenorhabditis Elegans.” Journal of Biology 6 (3): 8. https://doi.org/10.1186/jbiol58.

Caglayan, Ahmet Okay, Sinan Comu, Jacob F. Baranoski, Yesim Parman, Hande Kaymakçalan, Gozde Tugce Akgumus, Caner Caglar, et al. 2015. “NGLY1 Mutation Causes Neuromotor Impairment, Intellectual Disability, and Neuropathy.” European Journal of Medical Genetics 58 (1): 39–43. https://doi.org/10.1016/j.ejmg.2014.08.008.

Calderon-Dominguez, Maria, Gregorio Gil, Miguel Angel Medina, William M. Pandak, and Daniel Rodríguez-Agudo. 2014. “The StarD4 Subfamily of Steroidogenic Acute Regulatory-Related Lipid Transfer (START) Domain Proteins: New Players in Cholesterol Metabolism.” International Journal of Biochemistry and Cell Biology 49 (1): 64–68. https://doi.org/10.1016/j.biocel.2014.01.002.

Capasso, M., L. G. Durrant, M. Stacey, S. Gordon, J. Ramage, and I. Spendlove. 2014. “Costimulation via CD55 on Human CD4+ T Cells Mediated by CD97.” The Journal of Immunology 177 (2): 1070–77. https://doi.org/10.4049/jimmunol.177.2.1070.

Caprioli, J., P. Bettinaglio, P. F. Zipfel, B. Amadei, E. Daina, S. Gamba, C. Skerka, N. Marziliano, G. Remuzzi, and M. Noris. 2001. “The Molecular Basis of Familial Hemolytic Uremic Syndrome: Mutation Analysis of Factor H Gene Reveals a Hot Spot in Short Consensus Repeat 20.” Journal of the American Society of Nephrology 12 (2): 297–307.

Cardone, J, G Le Friec, and C Kemper. 2011. “CD46 in Innate and Adaptive Immunity: An Update.” Clinical and Experimental Immunology 164 (3): 301–11. https://doi.org/10.1111/j.1365-2249.2011.04400.x.

Cardone, John, Gaelle Le Friec, Pierre Vantourout, Andrew Roberts, Anja Fuchs, Ian Jackson, Tesha Suddason, et al. 2010. “Complement Regulator CD46 Temporally Regulates Cytokine Production by Conventional and Unconventional T Cells.” Nature Immunology 11 (9): 862–71. https://doi.org/10.1038/ni.1917.

136

Carroll, Maria V., and Robert B. Sim. 2011. “Complement in Health and Disease.” Advanced Drug Delivery Reviews 63 (12): 965–75. https://doi.org/10.1016/j.addr.2011.06.005.

Cha, Hyun Myoung, Jin Hyuk Lim, Jung Heum Yeon, Jeong Min Hwang, and Dong Il Kim. 2017. “Co-Overexpression of Mgat1 and Mgat4 in CHO Cells for Production of Highly Sialylated Albumin-Erythropoietin.” Enzyme and Microbial Technology 103 (March): 53– 58. https://doi.org/10.1016/j.enzmictec.2017.04.010.

Chakrabarti, Anirikh, Aaron W. Chen, and Jeffrey D. Varner. 2011. “A Review of the Mammalian Unfolded Protein Response.” Biotechnology and Bioengineering 108 (12): 2777–93. https://doi.org/10.1002/bit.23282.

Chang, Irene J, Miao He, and Christina T Lam. 2018. “Congenital Disorders of Glycosylation.” Annals of Translational Medicine 6 (24): 1–13. https://doi.org/10.21037/atm.2018.10.45.

Charman, Mark, Barry E. Kennedy, Nolan Osborne, and Barbara Karten. 2010. “MLN64 Mediates Egress of Cholesterol from Endosomes to Mitochondria in the Absence of Functional Niemann-Pick Type C1 Protein.” Journal of Lipid Research 51 (5): 1023–34. https://doi.org/10.1194/jlr.M002345.

Chen, Rong, Lisong Shi, Jörg Hakenberg, Brian Naughton, Pamela Sklar, Jianguo Zhang, Hanlin Zhou, et al. 2016. “Analysis of 589,306 Genomes Identifies Individuals Resilient to Severe Mendelian Childhood Diseases.” Nature Biotechnology 34 (5): 531–38. https://doi.org/10.1038/nbt.3514.

Choileain, Ni S., and Anne L. Astier. 2012. “CD46 Processing: A Means of Expression.” Immunobiology, 169–75. https://doi.org/10.1016/j.imbio.2011.06.003.

Choileain, S, N Weyand, C Neumann, J Thomas, M So, and A Astier. 2011. “The Dynamic Processing of CD46 Intracellular Domains Provides a Molecular Rheostat for T Cell Activation.” PLoS ONE 6 (1). https://doi.org/10.1371/journal.pone.0016287.

Choileain, Siobhan Ni, Joanne Hay, Joelle Thomas, Anna Williams, Matthieu M. Vermeren, Cecile Benezech, Mario Gomez-Salazar, et al. 2017. “TCR-Stimulated Changes in Cell Surface CD46 Expression Generate Type 1 Regulatory T Cells.” Science Signaling 10

137

(502): 1–13. https://doi.org/10.1126/scisignal.aah6163.

Chu, Bei Bei, Ya Cheng Liao, Wei Qi, Chang Xie, Ximing Du, Jiang Wang, Hongyuan Yang, Hong Hua Miao, Bo Liang Li, and Bao Liang Song. 2015. “Cholesterol Transport through Lysosome-Peroxisome Membrane Contacts.” Cell 161 (2): 291–306. https://doi.org/10.1016/j.cell.2015.02.019.

Clarke, Elizabeth V., and Andrea J. Tenner. 2014. “Complement Modulation of T Cell Immune Responses during Homeostasis and Disease.” Journal of Leukocyte Biology 96 (5): 745–56. https://doi.org/10.1189/jlb.3mr0214-109r.

Climer, Leslie K., Maxim Dobretsov, and Vladimir Lupashin. 2015. “Defects in the COG Complex and COG-Related Trafficking Regulators Affect Neuronal Golgi Function.” Frontiers in Neuroscience. Frontiers Media S.A. https://doi.org/10.3389/fnins.2015.00405.

Colic, Medina, Gang Wang, Michal Zimmermann, Keith Mascall, Megan McLaughlin, Lori Bertolet, W. Frank Lenoir, et al. 2019. “Identifying Chemogenetic Interactions from CRISPR Screens with DrugZ.” Genome Medicine 11 (1): 1–12. https://doi.org/10.1186/s13073-019-0665-3.

Colley, KJ, A. Varki, and T Kinoshita. 2017. “Cellular Organization of Glycosylation In: Essentials of Glycobiology.” Essentials of Glycobiology, 4–11. https://doi.org/10.1101/glycobiology.3e.004.

Colotta, Francesco, Paola Allavena, Antonio Sica, Cecilia Garlanda, and Alberto Mantovani. 2009. “Cancer-Related Inflammation, the Seventh Hallmark of Cancer: Links to Genetic Instability.” Carcinogenesis. https://doi.org/10.1093/carcin/bgp127.

Cong, L, FA Ran, D Cox, S Lin, R Barretto, N Habib, PD Hsu, et al. 2013. “Multiplex Genome Engineering Using CRISPR/Cas Systems.” Science 339 (February): 819–24.

Costanzo, Michael, Anastasia Baryshnikova, Jeremy Bellay, Yungil Kim, Eric D. Spear, Carolyn S. Sevier, Huiming Ding, et al. 2010. “The Genetic Landscape of a Cell.” Science 327 (5964): 425–31. https://doi.org/10.1126/science.1180823.

Costanzo, Michael, Anastasia Baryshnikova, Chad L Myers, Brenda Andrews, Charles Boone, 138

Matthias Heinemann, and Uwe Sauer. 2011. “Charting the Genetic Interaction Map of a Cell.” Current Opinion in Biotechnology 22: 66–74. https://doi.org/10.1016/j.copbio.2010.11.001.

Costanzo, Michael, Anastasia Baryshnikova, Benjamin VanderSluis, Brenda Andrews, Chad L. Myers, and Charles Boone. 2013. “Genetic Networks.” Handbook of Systems Biology, 115– 35. https://doi.org/10.1016/B978-0-12-385944-0.00006-X.

Costanzo, Michael, Elena Kuzmin, Jolanda van Leeuwen, Barbara Mair, Jason Moffat, Charles Boone, and Brenda Andrews. 2019. “Global Genetic Networks and the Genotype-to- Phenotype Relationship.” Cell 177 (1): 85–100. https://doi.org/10.1016/j.cell.2019.01.033.

Costanzo, Michael, Benjamin VanderSluis, Elizabeth N. Koch, Anastasia Baryshnikova, Carles Pons, Guihong Tan, Wen Wang, et al. 2016. “A Global Genetic Interaction Network Maps a Wiring Diagram of Cellular Function.” Science 353 (6306). https://doi.org/10.1126/science.aaf1420.

Das, Nibhriti, Devyani Anand, Bintili Biswas, Deepa Kumari, and Monika Gandhi. 2019. “The Membrane Complement Regulatory Protein CD59 and Its Association with Rheumatoid Arthritis and Systemic Lupus Erythematosus.” Current Medicine Research and Practice, August. https://doi.org/10.1016/j.cmrp.2019.07.013.

Davierwala, Armaity P., Jennifer Haynes, Zhijian Li, Renée L. Brost, Mark D. Robinson, Lisa Yu, Sanie Mnaimneh, et al. 2005. “The Synthetic Genetic Interaction Spectrum of Essential Genes.” Nature Genetics 37 (10): 1147–52. https://doi.org/10.1038/ng1640.

Degn, Søren E., Jens C. Jensenius, and Steffen Thiel. 2011. “Disease-Causing Mutations in Genes of the Complement System.” American Journal of Human Genetics. Elsevier. https://doi.org/10.1016/j.ajhg.2011.05.011.

Dho, So Hee, Jae Cheong Lim, and Lark Kyun Kim. 2018. “Beyond the Role of CD55 as a Complement Component.” Immune Network 18 (1): 1–13. https://doi.org/10.4110/in.2018.18.e11.

Digaleh, Hadi, Mahmoud Kiaei, and Fariba Khodagholi. 2013. “Nrf2 and Nrf1 Signaling and ER

139

Stress Crosstalk: Implication for Proteasomal Degradation and Autophagy.” Cellular and Molecular Life Sciences 70 (24): 4681–94. https://doi.org/10.1007/s00018-013-1409-y.

Dikic, Ivan, and Zvulun Elazar. 2018. “Mechanism and Medical Implications of Mammalian Autophagy.” Nature Reviews Molecular Cell Biology. https://doi.org/10.1038/s41580-018- 0003-4.

Dunkelberger, Jason R, and Wen-Chao Song. 2010. “Complement and Its Role in Innate and Adaptive Immune Responses.” Cell Research 20 (1): 34–50. https://doi.org/10.1038/cr.2009.139.

Durrant, L. G., M. A. Chapman, D. J. Buckley, I. Spendlove, R. A. Robins, and N. C. Armitage. 2003. “Enhanced Expression of the Complement Regulatory Protein CD55 Predicts a Poor Prognosis in Colorectal Cancer Patients.” Cancer Immunology, Immunotherapy 52 (10): 638–42. https://doi.org/10.1007/s00262-003-0402-y.

Eichler, Evan E., Jonathan Flint, Greg Gibson, Augustine Kong, Suzanne M. Leal, Jason H. Moore, and Joseph H. Nadeau. 2010. “Missing Heritability and Strategies for Finding the Underlying Causes of Complex Disease.” Nature Reviews Genetics. https://doi.org/10.1038/nrg2809.

Elkin, Sarah R, Ashley M Lakoduk, and Sandra L Schmid. 2016. “Endocytic Pathways and Endosomal Trafficking: A Primer.” Wien Med Wochenschr 166: 196–204. https://doi.org/10.1007/s10354-016-0432-7.

Ellinghaus, Ursula, Andrea Cortini, Christopher L. Pinder, Gaelle Le Friec, Claudia Kemper, and Timothy J. Vyse. 2017. “Dysregulated CD46 Shedding Interferes with Th1-Contraction in Systemic Lupus Erythematosus.” European Journal of Immunology 47 (7): 1200–1210. https://doi.org/10.1002/eji.201646822.

Enns, Gregory M., Vandana Shashi, Matthew Bainbridge, Michael J. Gambello, Farah R. Zahir, Thomas Bast, Rebecca Crimian, et al. 2014. “Mutations in NGLY1 Cause an Inherited Disorder of the Endoplasmic Reticulum-Associated Degradation Pathway.” Genetics in Medicine : Official Journal of the American College of Medical Genetics 16 (10): 751–58. https://doi.org/10.1038/gim.2014.22. 140

Essletzbichler, Patrick, Tomasz Konopka, Federica Santoro, Doris Chen, Bianca V Gapp, Robert Kralovics, Thijn R Brummelkamp, Sebastian M B Nijman, and Tilmann Bürckstümmer. 2014. “Megabase-Scale Deletion Using CRISPR/Cas9 to Generate a Fully Haploid Human Cell Line.” Genome Research 24 (12): 2059–65. https://doi.org/10.1101/gr.177220.114.

Fang, Pan, Xin Jian Wang, Yu Xue, Ming Qi Liu, Wen Feng Zeng, Yang Zhang, Lei Zhang, et al. 2016. “In-Depth Mapping of the Mouse Brain N-Glycoproteome Reveals Widespread N- Glycosylation of Diverse Brain Proteins.” Oncotarget 7 (25): 38796–809. https://doi.org/10.18632/oncotarget.9737.

Feinstein, Miora, Hagit Flusser, Lerman Sagie Tally, Bruria Ben-Zeev, Dorit Lev, Orly Agamy, Idan Cohen, et al. 2014. “VPS53 Mutations Cause Progressive Cerebello-Cerebral Atrophy Type 2 (PCCA2).” Journal of Medical Genetics 51 (5): 303–8. https://doi.org/10.1136/jmedgenet-2013-101823.

Ferreira, Viviana P., and Michael K. Pangburn. 2007. “Factor H-Mediated Cell Surface Protection from Complement Is Critical for the Survival of PNH Erythrocytes.” Blood 110 (6): 2190–92. https://doi.org/10.1182/blood-2007-04-083170.

“Fighting a One-of-a-Kind Disease | The New Yorker.” n.d. Accessed November 19, 2019. https://www.newyorker.com/magazine/2014/07/21/one-of-a-kind-2.

Fischer, Berad, Thomas Sandmann, Thomas Horn, Maximilian Billmann, Vanin Chaudhary, Wolfgang Huber, and Michael Boutros. 2015. “A Map of Directional Genetic Interactions in a Metazoan Cell.” ELife 2015 (4): 1–21. https://doi.org/10.7554/eLife.05464.

Fishelson, Z. 2003. “Obstacles to Cancer Immunotherapy: Expression of Membrane Complement Regulatory Proteins (MCRPs) in Tumors.” Molecular Immunology 40 (2–4): 109–23. https://doi.org/10.1016/S0161-5890(03)00112-3.

Freeze, Hudson H. 2006. “Genetic Defects in the Human Glycome.” Nature Reviews Genetics. Nature Publishing Group. https://doi.org/10.1038/nrg1894.

Fregno, Ilaria, and Maurizio Molinari. 2018. “ER-Associated Degradation (ERAD) and ER-to- Lysosome-Associated Degradation (ERLAD) Pathways.” Critical Reviews in Biochemistry

141

and Molecular Biology 54 (2): 153–63. https://doi.org/10.1080/10409238.2019.1610351.

Fremeaux-Bacchi, Véronique, Elizabeth A. Moulton, David Kavanagh, Marie Agnès Dragon- Durey, Jacques Blouin, Amy Caudy, Nadia Arzouk, et al. 2006. “Genetic and Functional Analyses of Membrane Cofactor Protein (CD46) Mutations in Atypical Hemolytic Uremic Syndrome.” Journal of the American Society of Nephrology 17 (7): 2017–25. https://doi.org/10.1681/ASN.2005101051.

Friec, Gaëlle Le, Devon Sheppard, Pat Whiteman, Christian M. Karsten, Salley Al Tilib Shamoun, Adam Laing, Laurence Bugeon, et al. 2012. “The CD46-Jagged1 Interaction Is Critical for Human T H 1 Immunity.” Nature Immunology 13 (12): 1213–21. https://doi.org/10.1038/ni.2454.

Frolíková, Michaela, Romana Stopková, Jana Antalíková, Peter M. Johnson, Pavel Stopka, and Kateřina Dvořákova-Hortova. 2012. “Role of Complement Regulatory Proteins CD46, CD55 and CD59 in Reproduction.” Folia Zoologica 61 (1): 84–94. https://doi.org/10.25225/fozo.v61.i1.a12.2012.

Fujihira, Haruhiko, Yuki Masahara-Negishi, Yoshihiro Akimoto, Hiroto Hirayama, Hyeon- Cheol Lee, Benjamin A Story, William F Mueller, et al. 2019. “Liver-Specific Deletion of Ngly1 Causes Abnormal Nuclear Morphology and Lipid Metabolism under Food Stress.” https://doi.org/10.1016/j.bbadis.2019.165588.

Fujihira, Haruhiko, Yuki Masahara-Negishi, Masaru Tamura, Chengcheng Huang, Yoichiro Harada, Shigeharu Wakana, Daisuke Takakura, et al. 2017. “Lethality of Mice Bearing a Knockout of the Ngly1-Gene Is Partially Rescued by the Additional Deletion of the Engase Gene.” PLoS Genetics, 1–23. https://doi.org/10.1371/journal.pgen.1006696.

Funakoshi, Yoko, Yuki Negishi, J. Peter Gergen, Junichi Seino, Kumiko Ishii, William J. Lennarz, Ichiro Matsuo, Yukishige Ito, Naoyuki Taniguchi, and Tadashi Suzuki. 2010. “Evidence for an Essential Deglycosylation-Independent Activity of PNGase in Drosophila Melanogaster.” PLoS ONE. https://doi.org/10.1371/journal.pone.0010545.

Galeone, Antonio, Seung Yeop Han, Chengcheng Huang, Akira Hosomi, Tadashi Suzuki, and Hamed Jafar-Nejad. 2017. “Tissue-Specific Regulation of BMP Signaling by Drosophila N- 142

Glycanase 1.” ELife, 1–24. https://doi.org/10.7554/elife.27612.

Geerlings, Maartje J., Eiko K. de Jong, and Anneke I. den Hollander. 2017. “The Complement System in Age-Related Macular Degeneration: A Review of Rare Genetic Variants and Implications for Personalized Treatment.” Molecular Immunology 84 (April): 65–76. https://doi.org/10.1016/j.molimm.2016.11.016.

Geller, Anne, and Jun Yan. 2019. “The Role of Membrane Bound Complement Regulatory Proteins in Tumor Development and Cancer Immunotherapy.” Frontiers in Immunology 10 (May): 1–13. https://doi.org/10.3389/fimmu.2019.01074.

Giurgiu, Madalina, Julian Reinhard, Barbara Brauner, Irmtraud Dunger-Kaltenbach, Gisela Fobo, Goar Frishman, Corinna Montrone, and Andreas Ruepp. 2019. “CORUM: The Comprehensive Resource of Mammalian Protein Complexes-2019.” Nucleic Acids Research 47 (D1): D559–63. https://doi.org/10.1093/nar/gky973.

Golay, J, L Zaffaroni, T Vaccari, M Lazzari, G M Borleri, S Bernasconi, F Tedesco, a Rambaldi, and M Introna. 2000. “Biologic Response of B Lymphoma Cells to Anti-CD20 Monoclonal Antibody Rituximab in Vitro: CD55 and CD59 Regulate Complement-Mediated Cell Lysis.” Blood 95 (12): 3900–3908. http://www.ncbi.nlm.nih.gov/pubmed/10845926.

Golec, Ewelina, Rebecca Rosberg, Enming Zhang, Erik Renström, Anna M Blom, and Ben C King. 2019. “A Cryptic Non-GPI-Anchored Cytosolic Isoform of CD59 Controls Insulin Exocytosis in Pancreatic β-Cells by Interaction with SNARE Proteins.” FASEB Journal : Official Publication of the Federation of American Societies for Experimental Biology 33 (11): 12425–34. https://doi.org/10.1096/fj.201901007R.

Gorter, Arko, and Seppo Meri. 1999. “Immune Evasion of Tumor Cells Using Membrane-Bound Complement Regulatory Proteins.” Review Immunology Today, no. 12: 576–82.

Grimberg, K. B., A. Beskow, D. Lundin, M. M. Davis, and P. Young. 2011. “Basic Leucine Zipper Protein Cnc-C Is a Substrate and Transcriptional Regulator of the Drosophila 26S Proteasome.” Molecular and Cellular Biology 31 (4): 897–909. https://doi.org/10.1128/mcb.00799-10.

143

Grünewald, Stephanie, Gert Matthijs, and Jaak Jaeken. 2002. “Congenital Disorders of Glycosylation: A Review.” Pediatric Research 52 (5): 618–24. https://doi.org/10.1203/00006450-200211000-00003.

Guc, Dicle, Hande Canpinar, Can Kucukaksu, and Emin Kansu. 2000. “Expression of Complement Regulatory Proteins CR1, DAF, MCP and CD59 in Haematological Malignancies.” European Journal of Haematology 64 (1): 3–9. https://doi.org/10.1034/j.1600-0609.2000.80097.x.

Guijas, Carlos, Gema Pérez-Chacón, Alma M. Astudillo, Julio M. Rubio, Luis Gil-de-Gómez, María A. Balboa, and Jesús Balsinde. 2012. “Simultaneous Activation of P38 and JNK by Arachidonic Acid Stimulates the Cytosolic Phospholipase A2-Dependent Synthesis of Lipid Droplets in Human Monocytes.” Journal of Lipid Research 53 (11): 2343–54. https://doi.org/10.1194/jlr.M028423.

Haijes, Hanneke A., Monique G.M. de Sain-van der Velden, Hubertus C.M.T. Prinsen, Anke P. Willems, Maria van der Ham, Johan Gerrits, Madeline H. Couse, et al. 2019. “Aspartylglycosamine Is a Biomarker for NGLY1-CDDG, a Congenital Disorder of Deglycosylation.” Molecular Genetics and Metabolism, no. May: 0–1. https://doi.org/10.1016/j.ymgme.2019.07.001.

Hajishengallis, George, Edimara S. Reis, Dimitrios C. Mastellos, Daniel Ricklin, and John D. Lambris. 2017. “Novel Mechanisms and Functions of Complement.” Nature Immunology. https://doi.org/10.1038/ni.3858.

Hall, Patricia L., Christina Lam, John J. Alexander, Ghazia Asif, Gerard T. Berry, Carlos Ferreira, Hudson H. Freeze, et al. 2018. “Urine Oligosaccharide Screening by MALDI-TOF for the Identification of NGLY1 Deficiency.” Molecular Genetics and Metabolism 124 (1): 82–86. https://doi.org/10.1016/j.ymgme.2018.03.002.

Harris, Claire L. 2018. “Expanding Horizons in Complement Drug Discovery: Challenges and Emerging Strategies.” Seminars in Immunopathology. Springer Verlag. https://doi.org/10.1007/s00281-017-0655-8.

Hart, Traver, Kevin R Brown, Fabrice Sircoulomb, Robert Rottapel, and Jason Moffat. 2014. 144

“Measuring Error Rates in Genomic Perturbation Screens: Gold Standards for Human Functional Genomics.” Molecular Systems Biology 10 (733). https://doi.org/10.15252/msb.20145216.

Hart, Traver, Megha Chandrashekhar, Michael Aregger, Zachary Steinhart, Kevin R. Brown, Graham MacLeod, Monika Mis, et al. 2015. “High-Resolution CRISPR Screens Reveal Fitness Genes and Genotype-Specific Cancer Liabilities.” Cell 163 (6): 1515–26. https://doi.org/10.1016/j.cell.2015.11.015.

Hart, Traver, Amy Hin Yan Tong, Katie Chan, Jolanda Van Leeuwen, Ashwin Seetharaman, Michael Aregger, Megha Chandrashekhar, et al. 2017. “Evaluation and Design of Genome- Wide CRISPR/SpCas9 Knockout Screens.” G3: Genes, Genomes, Genetics 7 (8): 2719–27. https://doi.org/10.1534/g3.117.041277.

Hartman, John L., Barbara Garvik, and Lee Hartwell. 2001. “Principles for the Buffering of Genetic Variation.” Science 291 (5506): 1001–4. https://doi.org/10.1126/science.1056072.

Hasan, Maroof, James Koch, Dinesh Rakheja, Asit K. Pattnaik, James Brugarolas, Igor Dozmorov, Beth Levine, Edward K. Wakeland, Min Ae Lee-Kirsch, and Nan Yan. 2013. “Trex1 Regulates Lysosomal Biogenesis and Interferon-Independent Activation of Antiviral Genes.” Nature Immunology 14 (1): 61–71. https://doi.org/10.1038/ni.2475.

Heeley, Jennifer, and Marwan Shinawi. 2015. “Multi-Systemic Involvement in NGLY1-Related Disorder Caused by Two Novel Mutations.” American Journal of Medical Genetics, Part A 167 (4): 816–20. https://doi.org/10.1002/ajmg.a.36889.

Helenius, Ari, and Markus Aebi. 2004. “Roles of N-Linked Glycans in the Endoplasmic Reticulusm.” Annu. Rev. Biochem., 1019–49. https://doi.org/10.1146/annurev.biochem.73.011303.073752.

Hess, Christoph, and Claudia Kemper. 2016. “Complement-Mediated Regulation of Metabolism and Basic Cellular Processes.” Immunity. https://doi.org/10.1016/j.immuni.2016.08.003.

Hieter, Philip, and Kym M. Boycott. 2014. “Understanding Rare Disease Pathogenesis: A Grand Challenge for Model Organisms.” Genetics 198 (2): 443–45.

145

https://doi.org/10.1534/genetics.114.170217.

Hill, Anita, Amy E. DeZern, Taroh Kinoshita, and Robert A. Brodsky. 2017. “Paroxysmal Nocturnal Haemoglobinuria.” Nature Reviews Disease Primers 3 (1): 17028. https://doi.org/10.1038/nrdp.2017.28.

Hill, Anita, Richard J. Kelly, and Peter Hillmen. 2013. “Thrombosis in Paroxysmal Nocturnal Hemoglobinuria.” Blood. American Society of Hematology. https://doi.org/10.1182/blood- 2012-09-311381.

Hill, Brian T., Mitchell R. Smith, Meredeth Shelley, Deepa Jagadeesh, Robert M. Dean, Brad Pohlman, John W. Sweetenham, Brian J. Bolwell, and Stephen D. Smith. 2018. “A Phase I Trial of Bortezomib in Combination with Everolimus for Treatment of Relapsed/Refractory Non-Hodgkin Lymphoma.” Leukemia and Lymphoma 59 (3): 690–94. https://doi.org/10.1080/10428194.2017.1347932.

Hirayama, Hiroto, Akira Hosomi, and Tadashi Suzuki. 2015. “Physiological and Molecular Functions of the Cytosolic Peptide: N-Glycanase.” Seminars in Cell and Developmental Biology 41: 110–20. https://doi.org/10.1016/j.semcdb.2014.11.009.

Hirsch, Christian, Daniël Blom, and Hidde L. Ploegh. 2003. “A Role for N-Glycanase in the Cytosolic Turnover of Glycoproteins.” EMBO Journal 22 (5): 1036–46. https://doi.org/10.1093/emboj/cdg107.

Hmeljak, Julija, and Monica J. Justice. 2019. “From Gene to Treatment: Supporting Rare Disease Translational Research through Model Systems.” Disease Models & Mechanisms 12 (2): dmm039271. https://doi.org/10.1242/dmm.039271.

Holers, V. Michael. 2014. “Complement and Its Receptors: New Insights into Human Disease.” Annual Review of Immunology 32 (1): 433–59. https://doi.org/10.1146/annurev-immunol- 032713-120154.

“Home - IMPC | International Mouse Phenotyping Consortium.” n.d. Accessed December 8, 2019. https://www.mousephenotype.org/.

Horlbeck, Max A., Albert Xu, Min Wang, Neal K. Bennett, Chong Y. Park, Derek Bogdanoff, 146

Britt Adamson, et al. 2018. “Mapping the Genetic Landscape of Human Cells.” Cell 174 (4): 953-967.e22. https://doi.org/10.1016/j.cell.2018.06.010.

Houten, Sander Michel, and Ronald J.A. Wanders. 2010. “A General Introduction to the Biochemistry of Mitochondrial Fatty Acid β-Oxidation.” Journal of Inherited Metabolic Disease. Springer. https://doi.org/10.1007/s10545-010-9061-2.

Hua, Rong, Derrick Cheng, Étienne Coyaud, Spencer Freeman, Erminia Di Pietro, Yuqing Wang, Adriano Vissa, et al. 2017. “VAPs and ACBD5 Tether Peroxisomes to the ER for Peroxisome Maintenance and Lipid Homeostasis.” Journal of Cell Biology 216 (2): 367–77. https://doi.org/10.1083/jcb.201608128.

Huang, Chengcheng, Yoichiro Harada, Akira Hosomi, Yuki Masahara-Negishi, Junichi Seino, Haruhiko Fujihira, Yoko Funakoshi, Takehiro Tadashi Suzuki, Naoshi Dohmae, and Takehiro Tadashi Suzuki. 2015. “Endo-β-N-Acetylglucosaminidase Forms N-GlcNAc Protein Aggregates during ER-Associated Degradation in Ngly1-Defective Cells.” PNAS 112 (5): 1398–1403. https://doi.org/10.1073/pnas.1414593112.

Huang, Yuxiang, Fei Qiao, Ruben Abagyan, Starr Hazard, and Stephen Tomlinson. 2006. “Defining the CD59-C9 Binding Interaction.” Journal of Biological Chemistry 281 (37): 27398–404. https://doi.org/10.1074/jbc.M603690200.

Hughes, Eric A., Craig Hammond, and Peter Cresswell. 1997. “Misfolded Major Histocompatibility Complex Class I Heavy Chains Are Translocated into the Cytoplasm and Degraded by the Proteasome.” Proceedings of the National Academy of Sciences of the United States of America 94 (5): 1896–1901. https://doi.org/10.1073/pnas.94.5.1896.

Huppa, Johannes B., and Hidde L. Ploegh. 1997. “The α Chain of the T Cell Antigen Receptor Is Degraded in the Cytosol.” Immunity 7 (1): 113–22. https://doi.org/10.1016/S1074- 7613(00)80514-2.

Iyer, Sangeetha, Joshua D. Mast, Hillary Tsang, Tamy P. Rodriguez, Nina DiPrimio, Madeleine Prangley, Feba S. Sam, Zachary Parton, and Ethan O. Perlstein. 2019. “Drug Screens of NGLY1 Deficiency Worm and Fly Models Reveal Catecholamine, NRF2 and Anti- Inflammatory Pathway Activation as Potential Clinical Approaches.” Disease Models & 147

Mechanisms, September, dmm.040576. https://doi.org/10.1242/dmm.040576.

Jeanmonod, R., and D. Jeanmonod. 2019. “Inborn Errors of Metabolism.” In Handbook of Clinical Neurology, 162:449–81. Elsevier B.V. https://doi.org/10.1016/B978-0-444-64029- 1.00022-9.

Jin, Zixue, Wei Wei, Marie Yang, Yang Du, and Yihong Wan. 2014. “Mitochondrial Complex i Activity Suppresses Inflammation and Enhances Bone Resorption by Shifting Macrophage- Osteoclast Polarization.” Cell Metabolism 20 (3): 483–98. https://doi.org/10.1016/j.cmet.2014.07.011.

Jokiranta, T Sakari. 2018. “HUS and Atypical HUS.” Blood 129 (21): 2847–56. https://doi.org/10.1182/blood-2016-11-709865.

Jurianz, K, S Ziegler, H Garcia-Schüler, S Kraus, O Bohana-Kashtan, Z Fishelson, and M Kirschfink. 1999. “Complement Resistance of Tumor Cells: Basal and Induced Mechanisms.” Molecular Immunology 36 (13–14): 929–39. http://www.ncbi.nlm.nih.gov/pubmed/10698347.

Kabouridis, Panagiotis S. 2006. “Lipid Rafts in T Cell Receptor Signalling (Review).” Molecular Membrane Biology. https://doi.org/10.1080/09687860500453673.

Kamiya, Yukiko, Yoshinori Uekusa, Akira Sumiyoshi, Hiroaki Sasakawa, Takeshi Hirao, Tadashi Suzuki, and Koichi Kato. 2012. “NMR Characterization of the Interaction between the PUB Domain of Peptide: N -Glycanase and Ubiquitin-like Domain of HR23.” FEBS Letters 586 (8): 1141–46. https://doi.org/10.1016/j.febslet.2012.03.027.

Karagianni, P., and J. Wong. 2007. “HDAC3: Taking the SMRT-N-CoRrect Road to Repression.” Oncogene. https://doi.org/10.1038/sj.onc.1210612.

Kastaniotis, Alexander J., Kaija J. Autio, Juha M. Kerätär, Geoffray Monteuuis, Anne M. Mäkelä, Remya R. Nair, Laura P. Pietikäinen, Antonina Shvetsova, Zhijun Chen, and J. Kalervo Hiltunen. 2017. “Mitochondrial Fatty Acid Synthesis, Fatty Acids and Mitochondrial Physiology.” Biochimica et Biophysica Acta - Molecular and Cell Biology of Lipids. Elsevier B.V. https://doi.org/10.1016/j.bbalip.2016.08.011.

148

Katiyar, S., G. Li, and W. J. Lennarz. 2004. “A Complex between Peptide:N-Glycanase and Two Proteasome-Linked Proteins Suggests a Mechanism for the Degradation of Misfolded Glycoproteins.” Proceedings of the National Academy of Sciences 101 (38): 13774–79. https://doi.org/10.1073/pnas.0405663101.

Katiyar, Samiksha, Tadashi Suzuki, Bhumika J. Balgobin, and William J. Lennarz. 2002. “Site- Directed Mutagenesis Study of Yeast Peptide: N-Glycanase. Insight into the Reaction Mechanism of Deglycosylation.” Journal of Biological Chemistry 277 (15): 12953–59. https://doi.org/10.1074/jbc.M111383200.

Kato, T., A. Kawahara, H. Ashida, and K. Yamamoto. 2007. “Unique Peptide:N-Glycanase of Caenorhabditis Elegans Has Activity of Protein Disulphide Reductase as Well as of Deglycosylation.” Journal of Biochemistry 142 (2): 175–81. https://doi.org/10.1093/jb/mvm117.

Kaufmann, Petra, Anne R. Pariser, and Christopher Austin. 2018. “From Scientific Discovery to Treatments for Rare Diseases - The View from the National Center for Advancing Translational Sciences - Office of Rare Diseases Research.” Orphanet Journal of Rare Diseases 13 (1): 1–8. https://doi.org/10.1186/s13023-018-0936-x.

Kemper, C., M. Leung, C. B. Stephensen, C. A. Pinkert, M. K. Liszewski, R. Cattaneo, and J. P. Atkinson. 2001. “Membrane Cofactor Protein (MCP; CD46) Expression in Transgenic Mice.” Clinical and Experimental Immunology 124 (2): 180–89. https://doi.org/10.1046/j.1365-2249.2001.01458.x.

Kemper, Claudia, and John P. Atkinson. 2007. “T-Cell Regulation: With Complements from Innate Immunity.” Nature Reviews Immunology. https://doi.org/10.1038/nri1994.

Kensler, Thomas W, Nobunao Wakabayashi, and Shyam Biswal. 2006. “Cell Survival Responses to Environmental Stresses Via the Keap1-Nrf2-ARE Pathway.” https://doi.org/10.1146/annurev.pharmtox.46.120604.141046.

Keulen, Britt J. van, Joost Rotteveel, and Martijn J.J. Finken. 2019. “Unexplained Death in Patients with NGLY1 Mutations May Be Explained by Adrenal Insufficiency.” Physiological Reports 7 (3): 1–4. https://doi.org/10.14814/phy2.13979. 149

Kim, David D, and Wen-Chao Song. 2006. “Membrane Complement Regulatory Proteins.” Clinical Immunology (Orlando, Fla.) 118 (2–3): 127–36. https://doi.org/10.1016/j.clim.2005.10.014.

Kim, Ikjin, Jungmi Ahn, Chang Liu, Kaori Tanabe, Jennifer Apodaca, Tadashi Suzuki, and Hai Rao. 2006. “The Png1-Rad23 Complex Regulates Glycoprotein Turnover.” Journal of Cell Biology 172 (2): 211–19. https://doi.org/10.1083/jcb.200507149.

Kimberley, Fiona C, Baalasubramanian Sivasankar, and B Paul Morgan. 2007. “Alternative Roles for CD59.” Molecular Immunology 44 (1–3): 73–81. https://doi.org/10.1016/j.molimm.2006.06.019.

Klos, Andreas, Andrea J. Tenner, Kay Ole Johswich, Rahasson R. Ager, Edimara S. Reis, and Jörg Köhl. 2009. “The Role of the Anaphylatoxins in Health and Disease.” Molecular Immunology. https://doi.org/10.1016/j.molimm.2009.04.027.

Koizumi, Shun, Taro Irie, Shoshiro Hirayama, Yasuyuki Sakurai, Hideki Yashiroda, Isao Naguro, Hidenori Ichijo, Jun Hamazaki, and Shigeo Murata. 2016. “The Aspartyl Protease DDI2 Activates Nrf1 to Compensate for Proteasome Dysfunction.” ELife. https://doi.org/10.7554/eLife.18357.

Kolev, Martin, Sarah Dimeloe, Gaelle Le Friec, Alexander Navarini, Giuseppina Arbore, Giovanni A Povoleri, Marco Fischer, et al. 2015. “Complement Regulates Nutrient Influx and Metabolic Reprogramming during Th1 Cell Responses.” Immunity 42 (6): 1033–47. https://doi.org/10.1016/j.immuni.2015.05.024.

Kolev, Martin, Gaelle Le Friec, and Claudia Kemper. 2014. “Complement-Tapping into New Sites and Effector Systems.” Nature Reviews Immunology. https://doi.org/10.1038/nri3761.

Kolev, Martin, and Claudia Kemper. 2017. “Keeping It All Going-Complement Meets Metabolism.” Frontiers in Immunology 8 (JAN). https://doi.org/10.3389/fimmu.2017.00001.

Kong, Jianping, Min Peng, Julian Ostrovsky, Young Joon Kwon, Olga Oretsky, Elizabeth M. McCormick, Miao He, Yair Argon, and Marni J. Falk. 2018. “Mitochondrial Function

150

Requires NGLY1.” Mitochondrion 38 (July 2017): 6–16. https://doi.org/10.1016/j.mito.2017.07.008.

Krus, Ulrika, Ben C. King, Vini Nagaraj, Nikhil R. Gandasi, Jonatan Sjölander, Pawel Buda, Eliana Garcia-Vaz, et al. 2014. “The Complement Inhibitor CD59 Regulates Insulin Secretion by Modulating Exocytotic Events.” Cell Metabolism 19 (5): 883–90. https://doi.org/10.1016/j.cmet.2014.03.001.

Kubiczkova, Lenka, Ludek Pour, Lenka Sedlarikova, Roman Hajek, and Sabina Sevcikova. 2014. “Proteasome Inhibitors - Molecular Basis and Current Perspectives in Multiple Myeloma.” Journal of Cellular and Molecular Medicine 18 (6): 947–61. https://doi.org/10.1111/jcmm.12279.

Kuzmin, Elena, Benjamin VanderSluis, Wen Wang, Guihong Tan, Raamesh Deshpande, Yiqun Chen, Matej Usaj, et al. 2018. “Systematic Analysis of Complex Genetic Interactions.” Science 360 (6386). https://doi.org/10.1126/science.aao1729.

Kyle, Stephanie M, Pradip K Saha, Hannah M Brown, Lawrence C Chan, and Monica J Justice. 2016. “MeCP2 Co-Ordinates Liver Lipid Metabolism with the NCoR1/HDAC3 Corepressor Complex.” Human Molecular Genetics 25 (14): 3029–2041. https://doi.org/10.1093/hmg/ddw156.

Lam, Christina, Carlos Ferreira, Donna Krasnewich, Camilo Toro, Lea Latham, Wadih M. Zein, Tanya Lehky, et al. 2017. “Prospective Phenotyping of NGLY1-CDDG, the First Congenital Disorder of Deglycosylation.” Genetics in Medicine 19 (2): 160–68. https://doi.org/10.1038/gim.2016.75.

Lamriben, Lydia, Jill B. Graham, Benjamin M. Adams, and Daniel N. Hebert. 2016. “N -Glycan- Based ER Molecular Chaperone and Protein Quality Control System: The Calnexin Binding Cycle.” Traffic 17 (4): 308–26. https://doi.org/10.1111/tra.12358.

Lee, J.-H., J. M. Choi, C. Lee, K. J. Yi, and Y. Cho. 2005. “Structure of a Peptide:N-Glycanase- Rad23 Complex: Insight into the Deglycosylation for Denatured Glycoproteins.” Proceedings of the National Academy of Sciences 102 (26): 9144–49. https://doi.org/10.1073/pnas.0502082102. 151

Legembre, Patrick, Sophie Daburon, Patrick Moreau, Jean-François Moreau, and Jean-Luc Taupin. 2006. “Cutting Edge: Modulation of Fas-Mediated Apoptosis by Lipid Rafts in T Lymphocytes.” The Journal of Immunology 176 (2): 716–20. https://doi.org/10.4049/jimmunol.176.2.716.

Lehrbach, Nicolas J., and Gary Ruvkun. 2016. “Proteasome Dysfunction Triggers Activation of SKN-1A/Nrf1 by the Aspartic Protease DDI-1.” ELife. https://doi.org/10.7554/eLife.17721.001.

Lehrbach, Nicolas J, Peter C Breen, and Gary Ruvkun. 2019. “Protein Sequence Editing of SKN- 1A / Nrf1 by Article Protein Sequence Editing of SKN-1A / Nrf1 by Peptide : N-Glycanase Controls.” Cell 177 (3): 737-750.e15. https://doi.org/10.1016/j.cell.2019.03.035.

Li, G., X. Zhou, G. Zhao, H. Schindelin, and W. J. Lennarz. 2005. “Multiple Modes of Interaction of the Deglycosylation Enzyme, Mouse Peptide N-Glycanase, with the Proteasome.” Proceedings of the National Academy of Sciences 102 (44): 15809–14. https://doi.org/10.1073/pnas.0507155102.

Liszewski, K. M., and J. P. Atkinson. 2015. “Complement Regulators in Human Disease: Lessons from Modern Genetics.” Journal of Internal Medicine 277 (3): 294–305. https://doi.org/10.1111/joim.12338.

Liszewski, M., A. Java, E. Schramm, and J. Atkinson. 2016. “Complement Dysregulation and Disease: Insights from Contemporary Genetics.” Annu. Rev. Pathol. Mech. Dis. 2017 12: 25–52. https://doi.org/10.1146/annurev-pathol-012615-044145.

Liszewski, M. Kathryn, and John P. Atkinson. 2015. “Complement Regulator CD46: Genetic Variants and Disease Associations.” Human Genomics. https://doi.org/10.1186/s40246-015- 0029-z.

Liszewski, M. Kathryn, and Claudia Kemper. 2019. “Complement in Motion: The Evolution of CD46 from a Complement Regulator to an Orchestrator of Normal Cell Physiology.” The Journal of Immunology 203 (1): 3–5. https://doi.org/10.4049/jimmunol.1900527.

Liszewski, M., Martin Kolev, Gaelle Le Friec, Marilyn Leung, Paula G. Bertram, Antonella F.

152

Fara, Marta Subias, et al. 2013. “Intracellular Complement Activation Sustains T Cell Homeostasis and Mediates Effector Differentiation.” Immunity 39 (6): 1143–57. https://doi.org/10.1016/j.immuni.2013.10.018.

Liszewski, M K, T W Post, and J P Atkinson. 1991. “Membrane Cofactor Protein (MCP or CD46): Newest Member of the Regulators of Complement Activation Gene Cluster.” Annual Review of Immunology 9 (January): 431–55. https://doi.org/10.1146/annurev.iy.09.040191.002243.

Liu, Zhichao, Liyuan Zhu, Ruth Roberts, and Weida Tong. 2019. “Toward Clinical Implementation of Next-Generation Sequencing-Based Genetic Testing in Rare Diseases: Where Are We?” Trends in Genetics. Elsevier Ltd. https://doi.org/10.1016/j.tig.2019.08.006.

Loughner, Chelsea L., Elspeth A. Bruford, Monica S. McAndrews, Emili E. Delp, Sudha Swamynathan, and Shivalingappa K. Swamynathan. 2016. “Organization, Evolution and Functions of the Human and Mouse Ly6/UPAR Family Genes.” Human Genomics. https://doi.org/10.1186/s40246-016-0074-2.

Lublin, D. M., and J. P. Atkinson. 1989. “Decay-Accelerating Factor: Biochemistry, Molecular Biology, and Function.” Annual Review of Immunology. https://doi.org/10.1146/annurev.iy.07.040189.000343.

Lukacik, P, P Roversi, J White, D Esser, G P Smith, J Billington, P a Williams, et al. 2004. “Complement Regulation at the Molecular Level: The Structure of Decay-Accelerating Factor.” Proceedings of the National Academy of Sciences of the United States of America 101 (5): 1279–84. https://doi.org/10.1073/pnas.0307200101.

Luo, Jie, Luyi Jiang, Hongyuan Yang, and Bao-Liang Song. 2017. “Routes and Mechanisms of Post-Endosomal Cholesterol Trafficking: A Story That Never Ends.” Traffic 18 (4): 209– 17. https://doi.org/10.1111/tra.12471.

Macleod, Graham, Danielle A Bozek, Nishani Rajakulendran, Samuel Weiss, Peter B Dirks, Stephane Angers, Vernon Monteiro, et al. 2019. “Genome-Wide CRISPR-Cas9 Screens Expose Genetic Vulnerabilities and Mechanisms of Temozolomide Sensitivity in 153

Glioblastoma Stem Cells Resource Genome-Wide CRISPR-Cas9 Screens Expose Genetic Vulnerabilities and Mechanisms of Temozolomide Sensitivity in Glioblastoma Stem Cells.” CellReports 27: 971-986.e9. https://doi.org/10.1016/j.celrep.2019.03.047.

Macor, P, E Secco, N Mezzaroba, S Zorzet, P Durigutto, T Gaiotto, L De Maso, et al. 2014. “Bispecific Antibodies Targeting Tumor-Associated Antigens and Neutralizing Complement Regulators Increase the Efficacy of Antibody-Based Immunotherapy in Mice.” Leukemia 29 (2): 406–14. https://doi.org/10.1038/leu.2014.185.

Macor, Paolo, Claudio Tripodo, Sonia Zorzet, Erich Piovan, Fleur Bossi, Roberto Marzari, Alberto Amadori, and Francesco Tedesco. 2007. “In Vivo Targeting of Human Neutralizing Antibodies against CD55 and CD59 to Lymphoma Cells Increases the Antitumor Activity of Rituximab.” Cancer Research 67 (21): 10556–63. https://doi.org/10.1158/0008- 5472.CAN-07-1811.

Madison, Blair B. 2016. “Srebp2: A Master Regulator of Sterol and Fatty Acid Synthesis.” Journal of Lipid Research 57 (3): 333–35. https://doi.org/10.1194/jlr.C066712.

Maerz, Sabine, Yoko Funakoshi, Yuki Negishi, Tadashi Suzuki, and Stephan Seiler. 2010. “The Neurospora Peptide:N-Glycanase Ortholog PNG1 Is Essential for Cell Polarity despite Its Lack of Enzymatic Activity.” Journal of Biological Chemistry 285 (4): 2326–32. https://doi.org/10.1074/jbc.M109.045302.

Mali, Prashant, Luhan Yang, Kevin M. Esvelt, John Aach, Marc Guell, James E. DiCarlo, Julie E. Norville, and George M. Church. 2013. “RNA-Guided Human Genome Engineering via Cas9.” Science 339 (6121): 823–26. https://doi.org/10.1126/science.1232033.

Mani, Ramamurthy, Robert P. St. Onge, John L. Hartman IV, Guri Giaever, and Frederick P. Roth. 2008. “Defining Genetic Interaction.” Proceedings of the National Academy of Sciences of the United States of America 105 (9): 3461–66. https://doi.org/10.1073/pnas.0712255105.

Manolio, Teri A, Francis S Collins, Nancy J Cox, David B Goldstein, Lucia A Hindorff, David J Hunter, Mark I McCarthy, et al. 2009. “Finding the Missing Heritability of Complex Diseases.” Nature 461 (7265): 747–53. https://doi.org/10.1038/nature08494. 154

Maroilley, Tatiana, and Maja Tarailo-Graovac. 2019. “Uncovering Missing Heritability in Rare Diseases.” Genes. MDPI AG. https://doi.org/10.3390/genes10040275.

Martin, Laura A., Barry E. Kennedy, and Barbara Karten. 2016. “Mitochondrial Cholesterol: Mechanisms of Import and Effects on Mitochondrial Function.” Journal of Bioenergetics and Biomembranes. Springer New York LLC. https://doi.org/10.1007/s10863-014-9592-6.

Maxfield, Frederick R, and Daniel Wüstner. 2002. “Intracellular Cholesterol Transport” 110 (7): 891–98. https://doi.org/10.1172/JCI200216500.The.

Meiners, Silke, Dirk Heyken, Andrea Weller, Antje Ludwig, Karl Stangl, Peter M. Kloetzel, and Elke Krüger. 2003. “Inhibition of Proteasome Activity Induces Concerted Expression of Proteasome Genes and de Novo Formation of Mammalian Proteasomes.” Journal of Biological Chemistry 278 (24): 21517–25. https://doi.org/10.1074/jbc.M301032200.

Meri, Seppo. 2016. “Self-Nonself Discrimination by the Complement System.” FEBS Letters. https://doi.org/10.1002/1873-3468.12284.

Meri, Seppo, and Hanna Jarva. 2013. “Complement Regulatory Proteins and Related Diseases.” ELS 1. https://doi.org/10.1002/9780470015902.a0001434.pub3.

Merle, Nicolas S., Sarah E. Church, Veronique. Fremeaux-Bacchi, and Lubka T. Roumenina. 2015. “Complement System Part I - Molecular Mechanisms of Activation and Regulation.” Frontiers in Immunology. https://doi.org/10.3389/fimmu.2015.00262.

Merle, Nicolas S., Remi Noe, Lise Halbwachs-Mecarelli, Veronique Fremeaux-Bacchi, and Lubka T. Roumenina. 2015. “Complement System Part II: Role in Immunity.” Frontiers in Immunology. https://doi.org/10.3389/fimmu.2015.00257.

Minami, J., R. Suzuki, R. Mazitschek, G. Gorgun, B. Ghosh, D. Cirstea, Y. Hu, et al. 2014. “Histone Deacetylase 3 as a Novel Therapeutic Target in Multiple Myeloma.” Leukemia 28 (3): 680–89. https://doi.org/10.1038/leu.2013.231.

Moffat, Jason, Jan H. Reiling, and David M. Sabatini. 2007. “Off-Target Effects Associated with Long DsRNAs in Drosophila RNAi Screens.” Trends in Pharmacological Sciences 28 (4): 149–51. https://doi.org/10.1016/j.tips.2007.02.009. 155

Moremen, Kelley W., Michael Tiemeyer, and Alison V. Nairn. 2012. “Vertebrate Protein Glycosylation: Diversity, Synthesis and Function.” Nature Reviews Molecular Cell Biology. https://doi.org/10.1038/nrm3383.

Mullis, Martin N., Takeshi Matsui, Rachel Schell, Ryan Foree, and Ian M. Ehrenreich. 2018. “The Complex Underpinnings of Genetic Background Effects.” Nature Communications 9 (1): 3548. https://doi.org/10.1038/s41467-018-06023-5.

Need, Anna C., Vandana Shashi, Yuki Hitomi, Kelly Schoch, Kevin V. Shianna, Marie T. McDonald, Miriam H. Meisler, and David B. Goldstein. 2012. “Clinical Application of Exome Sequencing in Undiagnosed Genetic Conditions.” Journal of Medical Genetics. https://doi.org/10.1136/jmedgenet-2012-100819.

Ng, Fanny, and Bor Luen Tang. 2014. “Pyruvate Dehydrogenase Complex (PDC) Export from the Mitochondrial Matrix.” Molecular Membrane Biology 31 (7–8): 207–10. https://doi.org/10.3109/09687688.2014.987183.

“NGLY1-Related Congenital Disorder of Deglycosylation - GeneReviews® - NCBI Bookshelf.” n.d. Accessed November 18, 2019. https://www.ncbi.nlm.nih.gov/books/NBK481554/.

“NGLY1.Org - N-Glycanase Deficiency - NGLY1.” n.d. Accessed November 18, 2019. https://www.ngly1.org/.

Nomura, Midori, Masaya Kitamura, Kiyomi Matsumiya, Akira Tsujimura, Akihiko Okuyama, Misako Matsumoto, Kumao Toyoshima, and Tsukasa Seya. 2001. “Genomic Analysis of Idiopathic Infertile Patients with Sperm-Specific Depletion of CD46.” Experimental and Clinical Immunogenetics 18 (1): 42–50. https://doi.org/10.1159/000049086.

Noris, Marina, and Giuseppe Remuzzi. 2013. “Overview of Complement Activation and Regulation.” Seminars in Nephrology 33 (6): 479–92. https://doi.org/10.1016/j.semnephrol.2013.08.001.

O’Connor, Michael B., David Umulis, Hans G. Othmer, and Seth S. Blair. 2006. “Shaping BMP Morphogen Gradients in the Drosophila Embryo and Pupal Wing.” Development. https://doi.org/10.1242/dev.02214.

156

Oglesby, Teresa J., Christopher J. Allen, M. Kathryn Liszewski, David J.G. White, and John P. Atkinson. 1992. “Membrane Cofactor Protein (CD46) Protects Cells from Complement- Mediated Attack by an Intrinsic Mechanism.” Journal of Experimental Medicine 175 (6): 1547–51. https://doi.org/10.1084/jem.175.6.1547.

Ohtsubo, Kazuaki, and Jamey D. Marth. 2006. “Glycosylation in Cellular Mechanisms of Health and Disease.” Cell 126 (5): 855–67. https://doi.org/10.1016/j.cell.2006.08.019.

Onge, Robert P. St., Ramamurthy Mani, Julia Oh, Michael Proctor, Eula Fung, Ronald W. Davis, Corey Nislow, Frederick P. Roth, and Guri Giaever. 2007. “Systematic Pathway Analysis Using High-Resolution Fitness Profiling of Combinatorial Gene Deletions.” Nature Genetics 39 (2): 199–206. https://doi.org/10.1038/ng1948.

Orsini, Franca, Daiana De Blasio, Rosalia Zangari, Elisa R. Zanier, and Maria Grazia De Simoni. 2014. “Versatility of the Complement System in Neuroinflammation, Neurodegeneration and Brain Homeostasis.” Frontiers in Cellular Neuroscience. Frontiers Media S.A. https://doi.org/10.3389/fncel.2014.00380.

Ory, Daniel S., Elizabeth A. Ottinger, Nicole Yanjanin Farhat, Kelly A. King, Xuntian Jiang, Lisa Weissfeld, Elizabeth Berry-Kravis, et al. 2017. “Intrathecal 2-Hydroxypropyl-β- Cyclodextrin Decreases Neurological Disease Progression in Niemann-Pick Disease, Type C1: A Non-Randomised, Open-Label, Phase 1–2 Trial.” The Lancet 390 (10104): 1758–68. https://doi.org/10.1016/S0140-6736(17)31465-4.

Ostankovitch, Marina, Michelle Altrich-VanLith, Valentina Robila, and Victor H. Engelhard. 2009. “ N -Glycosylation Enhances Presentation of a MHC Class I-Restricted Epitope from Tyrosinase .” The Journal of Immunology 182 (8): 4830–35. https://doi.org/10.4049/jimmunol.0802902.

Owings, Katie G., Joshua B. Lowry, Yiling Bi, Matthew Might, and Clement Y. Chow. 2018. “Transcriptome and Functional Analysis in a Drosophila Model of NGLY1 Deficiency Provides Insight into Therapeutic Approaches.” Human Molecular Genetics. https://doi.org/10.1093/hmg/ddy026.

Ozen, Ahmet, William A. Comrie, Rico C. Ardy, Cecilia Domínguez Conde, Buket Dalgic, 157

Ömer F. Beser, Aaron R. Morawski, et al. 2017. “CD55 Deficiency, Early-Onset Protein- Losing Enteropathy, and Thrombosis.” New England Journal of Medicine 377 (1): 52–61. https://doi.org/10.1056/nejmoa1615887.

Parente, Raffaella, Simon J. Clark, Antonio Inforzato, and Anthony J. Day. 2017. “Complement Factor H in Host Defense and Immune Evasion.” Cellular and Molecular Life Sciences 74 (9): 1605–24. https://doi.org/10.1007/s00018-016-2418-4.

Park, H., T. Suzuki, and W. J. Lennarz. 2001. “Identification of Proteins That Interact with Mammalian Peptide:N-Glycanase and Implicate This Hydrolase in the Proteasome- Dependent Pathway for Protein Degradation.” Proceedings of the National Academy of Sciences. https://doi.org/10.1073/pnas.201393498.

Patel, Kavi P, Thomas W O’brien, Sankarasubramon H Subramony, Jonathan Shuster, and Peter W Stacpoole. 2012. “The Spectrum of Pyruvate Dehydrogenase Complex Deficiency: Clinical, Biochemical and Genetic Features in 371 Patients.” Mol Genet Metab 105 (1): 34– 43. https://doi.org/10.1016/j.ymgme.2011.09.032.

Persson, B David, Nikolaus B Schmitz, César Santiago, Georg Zocher, Mykol Larvie, Ulrike Scheu, José M Casasnovas, and Thilo Stehle. 2010. “Structure of the Extracellular Portion of CD46 Provides Insights into Its Interactions with Complement Proteins and Pathogens.” PLoS Pathogens 6 (9): e1001122. https://doi.org/10.1371/journal.ppat.1001122.

Pilla, Esther, Kim Schneider, and Anne Bertolotti. 2017. “Coping with Protein Quality Control Failure.” Annual Review of Cell and Developmental Biology 33 (1): 439–65. https://doi.org/10.1146/annurev-cellbio-111315-125334.

Platt, F, A D’Azzo, B Davidson, E Neufeld, and C Tifft. 2017. “Lysosomal Storage Diseases.” Nature Reviews Disease Primers, January, 367–440. https://doi.org/10.3233/978-1-61499- 718-4-367.

Pollard, Kenneth M, David M Cauvi, Christopher B Toomey, Kevin V Morris, and Dwight H Kono. 2013. “Interferon-γ and Systemic Autoimmunity.” Discovery Medicine 16 (87): 123– 31. http://www.ncbi.nlm.nih.gov/pubmed/23998448.

158

Radhakrishnan, Senthil K., Willem den Besten, and Raymond J. Deshaies. 2014. “P97- Dependent Retrotranslocation and Proteolytic Processing Govern Formation of Active Nrf1 upon Proteasome Inhibition.” ELife 2014 (3): 1–15. https://doi.org/10.7554/eLife.01856.001.

Radhakrishnan, Senthil K., Candy S. Lee, Patrick Young, Anne Beskow, Jefferson Y. Chan, and Raymond J. Deshaies. 2010. “Transcription Factor Nrf1 Mediates the Proteasome Recovery Pathway after Proteasome Inhibition in Mammalian Cells.” Molecular Cell 38 (1): 17–28. https://doi.org/10.1016/j.molcel.2010.02.029.

Raedler, Lisa A. 2016. “Farydak (Panobinostat): First HDAC Inhibitor Approved for Patients with Relapsed Multiple Myeloma.” American Health & Drug Benefits 9 (Spec Feature): 84–87. http://www.ncbi.nlm.nih.gov/pubmed/27668050.

Rao, Rammohan V., and Dale E. Bredesen. 2004. “Misfolded Proteins, Endoplasmic Reticulum Stress and Neurodegeneration.” Current Opinion in Cell Biology. https://doi.org/10.1016/j.ceb.2004.09.012.

Reily, Colin, Tyler J. Stewart, Matthew B. Renfrow, and Jan Novak. 2019. “Glycosylation in Health and Disease.” Nature Reviews Nephrology 15 (6): 346–66. https://doi.org/10.1038/s41581-019-0129-4.

Reis, Edimara S., Dimitrios C. Mastellos, George Hajishengallis, and John D. Lambris. 2019. “New Insights into the Immune Functions of Complement.” Nature Reviews Immunology. https://doi.org/10.1038/s41577-019-0168-x.

Reis, Edimara S., Dimitrios C. Mastellos, Daniel Ricklin, Alberto Mantovani, and John D. Lambris. 2018. “Complement in Cancer: Untangling an Intricate Relationship.” Nature Reviews Immunology 18 (1): 5–18. https://doi.org/10.1038/nri.2017.97.

Richards, Anna, Mark R. Buddles, Rosemary L. Donne, Bernard S. Kaplan, Edwin Kirk, Michael C. Venning, Christian L. Tielemans, Judith A. Goodship, and Timothy H.J. Goodship. 2001. “Factor H Mutations in Hemolytic Uremic Syndrome Cluster in Exons 18- 20, a Domain Important for Host Cell Recognition.” American Journal of Human Genetics 68 (2): 485–90. https://doi.org/10.1086/318203. 159

Ricklin, D., and J. D. Lambris. 2013. “Complement in Immune and Inflammatory Disorders: Pathophysiological Mechanisms.” The Journal of Immunology. https://doi.org/10.4049/jimmunol.1203487.

Ricklin, Daniel, George Hajishengallis, Kun Yang, and John D Lambris. 2010. “Complement: A Key System for Immune Surveillance and Homeostasis.” Nature Immunology 11 (9): 785– 97. https://doi.org/10.1038/ni.1923.

Ricklin, Daniel, Edimara S. Reis, and John D. Lambris. 2016. “Complement in Disease: A Defence System Turning Offensive.” Nature Reviews Nephrology. https://doi.org/10.1038/nrneph.2016.70.

Riley-Vargas, Rebecca C, Darcy B Gill, Claudia Kemper, M Kathryn Liszewski, and John P Atkinson. 2004. “CD46: Expanding beyond Complement Regulation.” Trends in Immunology 25 (9): 496–503. https://doi.org/10.1016/j.it.2004.07.004.

Riley, Rebecca C., Claudia Kemper, Marilyn Leung, and John P. Atkinson. 2002. “Characterization of Human Membrane Cofactor Protein (MCP; CD46) on Spermatozoa.” Molecular Reproduction and Development 62 (4): 534–46. https://doi.org/10.1002/mrd.10144.

Ritchie, Marylyn D., and Kristel Van Steen. 2018. “The Search for Gene-Gene Interactions in Genome-Wide Association Studies: Challenges in Abundance of Methods, Practical Considerations, and Biological Interpretation.” Annals of Translational Medicine 6 (8): 157–157. https://doi.org/10.21037/atm.2018.04.05.

Rodriguez-Agudo, Daniel, Shunlin Ren, Eric Wong, Dalila Marques, Kaye Redford, Gregorio Gil, Phillip Hylemon, and William M. Pandak. 2008. “Intracellular Cholesterol Trnasporter StarD4 Binds Free Cholesterol and Increases Cholesteryl Ester Formation.” Journal of Lipid Research 49 (7): 1409–19. https://doi.org/10.1194/jlr.M700537-JLR200.

Rodriguez, Tamy Portillo, Joshua D. Mast, Tom Hartl, Tom Lee, Peter Sand, and Ethan O. Perlstein. 2018. “ Defects in the Neuroendocrine Axis Contribute to Global Development Delay in a Drosophila Model of NGLY1 Deficiency .” G3&#58; Genes|Genomes|Genetics 8 (7): 2193–2204. https://doi.org/10.1534/g3.118.300578. 160

Rooney, I A, T J Oglesby, and J P Atkinson. 1993. “Complement in Human Reproduction: Activation and Control.” Immunologic Research 12 (3): 276–94. https://doi.org/10.1007/bf02918258.

Rousseau, Adrien, and Anne Bertolotti. 2018. “Regulation of Proteasome Assembly and Activity in Health and Disease.” Nature Reviews Molecular Cell Biology. Nature Publishing Group. https://doi.org/10.1038/s41580-018-0040-z.

Saez, Isabel, and David Vilchez. 2014. “The Mechanistic Links Between Proteasome Activity, Aging and Agerelated Diseases.” Current Genomics 15 (1): 38–51. https://doi.org/10.2174/138920291501140306113344.

Sato, Haruhiro, Yoshinori Ino, Ayaka Miura, Yoshifumi Abe, Hideto Sakai, Koichi Ito, and Setsuo Hirohashi. 2003. “Dysadherin: Expression and Clinical Significance in Thyroid Carcinoma.” The Journal of Clinical Endocrinology & Metabolism 88 (9): 4407–12. https://doi.org/10.1210/jc.2002-021757.

Schreij, Andrea M.A., Edward A. Fon, and Peter S. McPherson. 2016. “Endocytic Membrane Trafficking and Neurodegenerative Disease.” Cellular and Molecular Life Sciences. Birkhauser Verlag AG. https://doi.org/10.1007/s00018-015-2105-x.

Shalem, Ophir, Neville E. Sanjana, Ella Hartenian, Xi Shi, David A. Scott, Tarjei S. Mikkelsen, Dirk Heckl, et al. 2014. “Genome-Scale CRISPR-Cas9 Knockout Screening in Human Cells.” Science 343 (6166): 84–87. https://doi.org/10.1126/science.1247005.

Shenoy-Scaria, A M, J Kwong, T Fujita, M W Olszowy, A S Shaw, and D M Lublin. 1992. “Signal Transduction through Decay-Accelerating Factor. Interaction of Glycosyl- Phosphatidylinositol Anchor and Protein Tyrosine Kinases P56lck and P59fyn 1.” Journal of Immunology (Baltimore, Md. : 1950) 149 (11): 3535–41. http://www.ncbi.nlm.nih.gov/pubmed/1385527.

Silverstein, Roy L., and Maria Febbraio. 2009. “CD36, a Scavenger Receptor Involved in Immunity, Metabolism, Angiogenesis, and Behavior.” Science Signaling. American Association for the Advancement of Science. https://doi.org/10.1126/scisignal.272re3.

161

Simons, Kai, and Robert Ehehalt. 2002. “Cholesterol, Lipid Rafts, and Disease.” Journal of Clinical Investigation 110 (5): 597–603. https://doi.org/10.1172/JCI16390.

Sivasankar, B., M. P. Longhi, K. M. E. Gallagher, G. J. Betts, B. P. Morgan, A. J. Godkin, and A. M. Gallimore. 2009. “CD59 Blockade Enhances Antigen-Specific CD4+ T Cell Responses in Humans: A New Target for Cancer Immunotherapy?” The Journal of Immunology 182 (9): 5203–7. https://doi.org/10.4049/jimmunol.0804243.

Skipper, Jonathan C.A., Ronald C. Hendrickson, Pamela H. Gulden, Vincent Brichard, Ailene Van Pel, Ye Chen, Jeffrey Shabanowitz, et al. 1996. “An HLA-A2-Restricted Tyrosinase Antigen on Melanoma Cells Results from Posttranslational Modification and Suggests a Novel Pathway for Processing of Membrane Proteins.” Journal of Experimental Medicine 183 (2): 527–34. https://doi.org/10.1084/jem.183.2.527.

Smith, Richard D., and Vladimir V. Lupashin. 2008. “Role of the Conserved Oligomeric Golgi (COG) Complex in Protein Glycosylation.” Carbohydrate Research. https://doi.org/10.1016/j.carres.2008.01.034.

Soccio, Raymond E., Rachel M. Adams, Kara N. Maxwell, and Jan L. Breslow. 2005. “Differential Gene Regulation of StarD4 and StarD5 Cholesterol Transfer Proteins: Activation of StarD4 by Sterol Regulatory Element-Binding Protein-2 and StarD5 by Endoplasmic Reticulum Stress.” Journal of Biological Chemistry 280 (19): 19410–18. https://doi.org/10.1074/jbc.M501778200.

Soyfoo, Muhammad Shahnawaz, Clara Chivasso, Jason Perret, and Christine Delporte. 2018. “Involvement of Aquaporins in the Pathogenesis, Diagnosis and Treatment of Sjögren’s Syndrome.” International Journal of Molecular Sciences. MDPI AG. https://doi.org/10.3390/ijms19113392.

Sparks, Susan E, and Donna M Krasnewich. 2017. Congenital Disorders of N-Linked Glycosylation and Multiple Pathway Overview. GeneReviews®. http://www.ncbi.nlm.nih.gov/pubmed/20301507.

Stanley, P., N. Taniguchi, and M. Aebi. 2017. “N-Glycans.” In Essentials of Glycobiology, edited by et al. Varki A, Cummings RD, Esko JD. Cold Spring Harbor Laboratory Press. 162

https://doi.org/10.1101/glycobiology.3e.009.

Steffen, Janos, Michael Seeger, Annett Koch, and Elke Krüger. 2010. “Proteasomal Degradation Is Transcriptionally Controlled by TCF11 via an ERAD-Dependent Feedback Loop.” Molecular Cell 40 (1): 147–58. https://doi.org/10.1016/j.molcel.2010.09.012.

Steinhart, Zachary, Zvezdan Pavlovic, Megha Chandrashekhar, Traver Hart, Xiaowei Wang, Xiaoyu Zhang, Mélanie Robitaille, et al. 2017. “Genome-Wide CRISPR Screens Reveal a Wnt-FZD5 Signaling Circuit as a Druggable Vulnerability of RNF43-Mutant Pancreatic Tumors.” Nature Medicine 23 (1): 60–68. https://doi.org/10.1038/nm.4219.

Sun, Wei, Wei Zheng, and Anton Simeonov. 2017. “Drug Discovery and Development for Rare Genetic Disorders.” American Journal of Medical Genetics, Part A 173 (9): 2307–22. https://doi.org/10.1002/ajmg.a.38326.

Surowiak, Pawel, Verena Materna, Adam Maciejczyk, Irina Kaplenko, Marek Spaczynski, Manfred Dietel, Hermann Lage, and Maciej Zabel. 2006. “CD46 Expression Is Indicative of Shorter Revival-Free Survival for Ovarian Cancer Patients.” Anticancer Research 26: 4943–48.

Sutavani, R. V., R. G. Bradley, J. M. Ramage, A. M. Jackson, L. G. Durrant, and I. Spendlove. 2013. “CD55 Costimulation Induces Differentiation of a Discrete T Regulatory Type 1 Cell Population with a Stable Phenotype.” The Journal of Immunology 191 (12): 5895–5903. https://doi.org/10.4049/jimmunol.1301458.

Suzuki, T. 2007. “Cytoplasmic Peptide:N-Glycanase and Catabolic Pathway for Free N-Glycans in the Cytosol.” Seminars in Cell and Developmental Biology 18 (6): 762–69. https://doi.org/10.1016/j.semcdb.2007.09.010.

———. 2016. “Catabolism of N-Glycoproteins in Mammalian Cells: Molecular Mechanisms and Genetic Disorders Related to the Processes.” Molecular Aspects of Medicine 51: 89– 103. https://doi.org/10.1016/j.mam.2016.05.004.

Suzuki, T., C. Huang, and H. Fujihira. 2016. “The Cytoplasmic Peptide: N-Glycanase (NGLY1) - Structure, Expression and Cellular Functions.” Gene 577 (1): 1–7.

163

https://doi.org/10.1016/j.gene.2015.11.021.

Suzuki, T., K. Kitajima, Y. Inoue, and S. Inoue. 1995. “Carbohydrate-Binding Property of Peptide:N-Glycanase from Mouse Fibroblast L-929 Cells as Evaluated by Inhibition and Binding Experiments Using Various Oligosaccharides.” Journal of Biological Chemistry. https://doi.org/10.1074/jbc.270.25.15181.

Suzuki, T., A. Seko, K. Katajima, Y. Inoue, and S. Inoue. 1993. “Identification of Peptide:N- Glycanase Activity in Mammalian-Derived Culture Cells.” Biochemical and Biophysical Research Communications 194 (3): 1124–30. https://doi.org/10.1006/bbrc.1993.1938.

Suzuki, Tadashi. 2015. “The Cytoplasmic Peptide:N-Glycanase (Ngly1) - Basic Science Encounters a Human Genetic Disorder.” Journal of Biochemistry 157 (1): 23–34. https://doi.org/10.1093/jb/mvu068.

Suzuki, Tadashi, Izumi Hara, Miyako Nakano, Gang Zhao, William J. Lennarz, Hermann Schindelin, Naoyuki Taniguchi, Kiichiro Totani, Ichiro Matsuo, and Yukishige Ito. 2006. “Site-Specific Labeling of Cytoplasmic Peptide:N-Glycanase by N,N′-Diacetylchitobiose- Related Compounds.” Journal of Biological Chemistry 281 (31): 22152–60. https://doi.org/10.1074/jbc.M603236200.

Suzuki, Tadashi, Ken Kitajima, Sadako Inoue, and Yasuo Inoue. 1994. “Does an Animal Peptide:N-Glycanase Have the Dual Role as an Enzyme and a Carbohydrate-Binding Protein?” Glycoconjugate Journal 11 (5): 469–76. https://doi.org/10.1007/BF00731283.

Suzuki, Tadashi, Hangil Park, Elizabeth Anderson Till, and William J. Lennarz. 2001. “The PUB Domain: A Putative Protein-Protein Interaction Domain Implicated in the Ubiquitin- Proteasome Pathway.” Biochemical and Biophysical Research Communications. https://doi.org/10.1006/bbrc.2001.5688.

Suzuki, Tadashi, Hangil Park, Nancy M. Hollingsworth, Rolf Sternglanz, and William J. Lennarz. 2000. “PNG1, a Yeast Gene Encoding a Highly Conserved Peptide: N- Glycanase.” Journal of Cell Biology. https://doi.org/10.1083/jcb.149.5.1039.

Suzuki, Tadashi, Hangil Park, Michael A. Kwofie, and William J. Lennarz. 2001. “Rad23

164

Provides a Link between the Png1 Deglycosylating Enzyme and the 26 S Proteasome in Yeast.” Journal of Biological Chemistry 276 (24): 21601–7. https://doi.org/10.1074/jbc.M100826200.

Suzuki, Tadashi, Hangil Park, and William J Lennarz. 2002. “Cytoplasmic Peptide:N-Glycanase (PNGase) in Eukaryotic Cells: Occurrence, Primary Structure, and Potential Functions.” FASEB Journal 16 (7): 635–41. www.fasebj.org.

Tahsin Hassan Rahit, K. M., and Maja Tarailo-Graovac. 2020. “Genetic Modifiers and Rare Mendelian Disease.” Genes. MDPI AG. https://doi.org/10.3390/genes11030239.

Tambe, M., B. Ng, and H. Freeze. 2019. “N-Glycanase 1 Transcriptionally Regulates Aquaporins Independent of Its Enzymatic Activity.” Cell Reports 29: 4620–31. https://doi.org/10.1016/j.celrep.2019.11.097.

Tambuyzer, Erik, Benjamin Vandendriessche, Christopher P Austin, Philip J Brooks, Kristina Larsson, Katherine I Miller Needleman, James Valentine, et al. 2020. “Therapies for Rare Diseases: Therapeutic Modalities, Progress and Challenges Ahead.” Nature Reviews Drug Discovery 19: 93–111. https://doi.org/10.1038/s41573-019-0049-9.

Tang, Sze Jing, Shufang Luo, Jia Xin Jessie Ho, Phuong Thao Ly, Eling Goh, and Xavier Roca. 2016. “Characterization of the Regulation of CD46 RNA Alternative Splicing.” Journal of Biological Chemistry 291 (27): 14311–23. https://doi.org/10.1074/jbc.M115.710350.

Thiel, Christian, and Christian Körner. 2011. “Mouse Models for Congenital Disorders of Glycosylation.” Journal of Inherited Metabolic Disease 34 (4): 879–89. https://doi.org/10.1007/s10545-011-9295-7.

Thielen, Astrid J.F., Iris M. van Baarsen, Marlieke L. Jongsma, Sacha Zeerleder, Robbert M. Spaapen, and Diana Wouters. 2018. “CRISPR/Cas9 Generated Human CD46, CD55 and CD59 Knockout Cell Lines as a Tool for Complement Research.” Journal of Immunological Methods 456 (November 2017): 15–22. https://doi.org/10.1016/j.jim.2018.02.004.

Thurman, Joshua M., and Brandon Renner. 2011. “Dynamic Control of the Complement System

165

by Modulated Expression of Regulatory Proteins.” Laboratory Investigation. Nature Publishing Group. https://doi.org/10.1038/labinvest.2010.173.

Tickotsky-Moskovitz, Nili. 2015. “New Perspectives on the Mutated NGLY1 Enigma.” Medical Hypotheses 85 (5): 584–85. https://doi.org/10.1016/j.mehy.2015.07.019.

Tomlin, Frederick M., Ulla I.M. Gerling-Driessen, Yi Chang Liu, Ryan A. Flynn, Janakiram R. Vangala, Christian S. Lentz, Sandra Clauder-Muenster, et al. 2017. “Inhibition of NGLY1 Inactivates the Transcription Factor Nrf1 and Potentiates Proteasome Inhibitor Cytotoxicity.” ACS Central Science. https://doi.org/10.1021/acscentsci.7b00224.

Tong, A., M. Evangelista, A. B. Parsons, H. Xu, G. D. Bader, N. Pagé, M. Robinson, et al. 2001. “Systematic Genetic Analysis with Ordered Arrays of Yeast Deletion Mutants.” Science 294 (5550): 2364–68. https://doi.org/10.1126/science.1065810.

Tong, Amy Hin Yan, Guillaume Lesage, Gary D. Bader, Huiming Ding, Hong Xu, Xiaofeng Xin, James Young, et al. 2004. “Global Mapping of the Yeast Genetic Interaction Network.” Science 303 (5659): 808–13. https://doi.org/10.1126/science.1091317.

Toomey, Christopher B., David M. Cauvi, and Kenneth M. Pollard. 2014. “The Role of Decay Accelerating Factor in Environmentally Induced and Idiopathic Systemic Autoimmune Disease.” Autoimmune Diseases 2014: 1–12. https://doi.org/10.1155/2014/452853.

Tripathi, Priya. 2017. “Steps in Genetic Research of Complex Disease.” Global Journal of Research and Review 04 (01): 1–3. https://doi.org/10.21767/2393-8854.10010.

Tsubota, Kazuo, Shinichiro Hirai, Landon S. King, Peter Agre, and Naruhiro Ishida. 2001. “Defective Cellular Trafficking of Lacrimal Gland Aquaporin-5 in Sjögren’s Syndrome.” Lancet 357 (9257): 688–89. https://doi.org/10.1016/S0140-6736(00)04140-4.

Tsvetkov, Peter, Marc L. Mendillo, Jinghui Zhao, Jan E. Carette, Parker H. Merrill, Domagoj Cikes, Malini Varadarajan, et al. 2015. “Compromising the 19S Proteasome Complex Protects Cells from Reduced Flux through the Proteasome.” ELife 4 (September2015). https://doi.org/10.7554/eLife.08467.

Tur, J., T. Vico, J. Lloberas, A. Zorzano, and A. Celada. 2017. “Macrophages and Mitochondria: 166

A Critical Interplay Between Metabolism, Signaling, and the Functional Activity.” In Advances in Immunology, 133:1–36. Academic Press Inc. https://doi.org/10.1016/bs.ai.2016.12.001.

Turnberg, Daniel, and Marina Botto. 2003. “The Regulation of the Complement System: Insights from Genetically-Engineered Mice.” Molecular Immunology 40: 145–53. https://doi.org/10.1016/S0161-5890(03)00110-X.

Vainer, E D, K Meir, M Furman, I Semenenko, F Konikoff, and G W Vainer. 2013. “Characterization of Novel CD55 Isoforms Expression in Normal and Neoplastic Tissues.” Tissue Antigens 82 (1): 26–34. https://doi.org/10.1111/tan.12138.

Vanier, MT, and G Millat. 2003. “Niemann-Pick Disease Type C.” Clinical Genetics 64 (4): 269–81. https://doi.org/10.1034/j.1399-0004.2003.00147.x.

Vignesh, Pandiarajan, Amit Rawat, Madhubala Sharma, and Surjit Singh. 2017. “Complement in Autoimmune Diseases.” Clinica Chimica Acta. Elsevier B.V. https://doi.org/10.1016/j.cca.2016.12.017.

Visser, Lizette, Alex F de Vos, Jörg Hamann, Marie-José Melief, Marjan van Meurs, René A W van Lier, Jon D Laman, and Rogier Q Hintzen. 2002. “Expression of the EGF-TM7 Receptor CD97 and Its Ligand CD55 (DAF) in Multiple Sclerosis.” Journal of Neuroimmunology 132 (1–2): 156–63. https://doi.org/10.1016/s0165-5728(02)00306-5.

Vizeacoumar, Franco J., Roland Arnold, Frederick S. Vizeacoumar, Megha Chandrashekhar, Alla Buzina, Jordan T.F. Young, Julian H.M. Kwan, et al. 2013. “A Negative Genetic Interaction Map in Isogenic Cancer Cell Lines Reveals Cancer Cell Vulnerabilities.” Molecular Systems Biology 9 (1). https://doi.org/10.1038/msb.2013.54.

Vrueh, R de, E R F Baekelandt, and J M H de Haan. 2013. “Background Paper 6.19 Rare Diseases.” Background Paper 6.19 Rare Diseases, no. March: 1–46.

Walport, Mark J. 2001a. “Complement (First of Two Parts).” New England Journal of Medicine 344 (14): 1058–66. https://doi.org/10.1056/NEJM200104053441406.

———. 2001b. “Complement (Second of Two Parts).” New England Journal of Medicine 344 167

(15): 1140–44. https://doi.org/10.1056/NEJM200104123441506.

Walport, Mark J, and Correspondence Mark J Walport. 2002. “Complement and Systemic Lupus Erythematosus Complement and SLE : The Clinical Observations,” 1–28. https://doi.org/10.1186/ar586.

Wang, Chao, Gang Wang, Xu Feng, Peter Shepherd, Jie Zhang, Mengfan Tang, Zhen Chen, et al. 2019. “Genome-Wide CRISPR Screens Reveal Synthetic Lethality of RNASEH2 Deficiency and ATR Inhibition.” Oncogene 38 (14): 2451–63. https://doi.org/10.1038/s41388-018-0606-4.

Wang, Li Na, Mei Hua Gao, Bing Wang, Bei Bei Cong, and Shu Chao Zhang. 2018. “A Role for GPI-CD59 in Promoting T-Cell Signal Transduction via LAT.” Oncology Letters 15 (4): 4873–81. https://doi.org/10.3892/ol.2018.7908.

Wang, Richard N., Jordan Green, Zhongliang Wang, Youlin Deng, Min Qiao, Michael Peabody, Qian Zhang, et al. 2014. “Bone Morphogenetic Protein (BMP) Signaling in Development and Human Diseases.” Genes and Diseases. Chongqing Medical University. https://doi.org/10.1016/j.gendis.2014.07.005.

Wang, Tim., Jenny J. Wei, David M. Sabatini, and Eric S. Lander. 2014. “Genetic Screens in Human Cells Using the CRISPR-Cas9 System.” Science 2 (3): 256–63. https://doi.org/10.1136/bmjspcare-2011-000063.

Wang, Tim, Kivanç Birsoy, Nicholas W. Hughes, Kevin M. Krupczak, Yorick Post, Jenny J. Wei, Eric S. Lander, and David M. Sabatini. 2015. “Identification and Characterization of Essential Genes in the Human Genome.” Science 350 (6264): 1096–1101. https://doi.org/10.1126/science.aac7041.

Wang, Tim, Haiyan Yu, Nicholas W. Hughes, Bingxu Liu, Arek Kendirli, Klara Klein, Walter W. Chen, Eric S. Lander, and David M. Sabatini. 2017. “Gene Essentiality Profiling Reveals Gene Networks and Synthetic Lethal Interactions with Oncogenic Ras.” Cell 168 (5): 890-903.e15. https://doi.org/10.1016/j.cell.2017.01.013.

Wang, Wen, Zack Z. Xu, Michael Costanzo, Charles Boone, Carol A. Lange, and Chad L.

168

Myers. 2017. “Pathway-Based Discovery of Genetic Interactions in Breast Cancer.” PLoS Genetics 13 (9): e1006973. https://doi.org/10.1371/journal.pgen.1006973.

Wang, Yan, Jenny Tan, Mark Sutton-Smith, David Ditto, Maria Panico, Robert M. Campbell, Nissi M. Varki, et al. 2001. “Modeling Human Congenital Disorder of Glycosylation Type IIa in the Mouse: Conservation of Asparagine-Linked Glycan-Dependent Functions in Mammalian Physiology and Insights into Disease Pathogenesis.” Glycobiology 11 (12): 1051–70. https://doi.org/10.1093/glycob/11.12.1051.

Wei, Jian, Ying-Yu Zhang, Jie Luo, Ju-Qiong Wang, Yu-Xia Zhou, Hong-Hua Miao, Xiong-Jie Shi, et al. 2017. “The GARP Complex Is Involved in Intracellular Cholesterol Transport via Targeting NPC2 to Lysosomes.” Cell Reports 19 (13): 2823–35. https://doi.org/10.1016/j.celrep.2017.06.012.

Widenmaier, Scott B., Nicole A. Snyder, Truc B. Nguyen, Alessandro Arduini, Grace Y. Lee, Ana Paula Arruda, Jani Saksi, Alexander Bartelt, and Gökhan S. Hotamisligil. 2017. “NRF1 Is an ER Membrane Sensor That Is Central to Cholesterol Homeostasis.” Cell. https://doi.org/10.1016/j.cell.2017.10.003.

Winchester, Bryan. 2005. “Lysosomal Metabolism of Glycoproteins.” Glycobiology 15 (6): 1– 15. https://doi.org/10.1093/glycob/cwi041.

Wlodkowic, Donald, Joanna Skommer, Dagmara McGuinness, Chris Hillier, and Zbigniew Darzynkiewicz. 2009. “ER-Golgi Network-A Future Target for Anti-Cancer Therapy.” Leukemia Research. https://doi.org/10.1016/j.leukres.2009.05.025.

Woodman, Philip G. 2003. “P97, a Protein Coping with Multiple Identities.” Journal of Cell Science 116 (21): 4283–90. https://doi.org/10.1242/jcs.00817.

Wopereis, Suzan, Dirk J Lefeber, Éva Morava, and Ron A Wevers. 2006. “Mechanisms in Protein O-Glycan Biosynthesis and Clinical and Molecular Aspects of Protein O-Glycan Biosynthesis Defects: A Review.” Clinical Chemistry 52 (4): 574–600. https://doi.org/10.1373/clinchem.2005.063040.

Wu, Yinshuang, Xixi Chen, Shidan Wang, and Shujing Wang. 2019. “Advances in the

169

Relationship between Glycosyltransferases and Multidrug Resistance in Cancer.” Clinica Chimica Acta. Elsevier B.V. https://doi.org/10.1016/j.cca.2019.05.015.

Xu, Kui., and Timothy R. Coté. 2011. “Database Identifies FDA-Approved Drugs with Potential to Be Repurposed for Treatment of Orphan Diseases.” Briefings in Bioinformatics 12 (4): 341–45. https://doi.org/10.1093/bib/bbr006.

Xu, Ya Qing, Ya Dong Gao, Jiong Yang, and Wei Guo. 2010. “A Defect of CD4 + CD25 + Regulatory T Cells in Inducing Interleukin-10 Production from CD4 + T Cells under CD46 Costimulation in Asthma Patients.” Journal of Asthma 47 (4): 367–73. https://doi.org/10.3109/02770903.2010.481340.

Yamamoto, Hidekazu, Antonella Francesca Fara, Prokar Dasgupta, and Claudia Kemper. 2013. “CD46: The ‘multitasker’ of Complement Proteins.” The International Journal of Biochemistry & Cell Biology 45 (12): 2808–20. https://doi.org/10.1016/j.biocel.2013.09.016.

Yang, Kun, Ryan Huang, Haruhiko Fujihira, Tadashi Suzuki, and Nan Yan. 2018. “N-Glycanase NGLY1 Regulates Mitochondrial Homeostasis and Inflammation through NRF1.” J. Exp. Med 215: 2600–2616. https://doi.org/10.1084/jem.20180783.

Yu, Wenxin, Jian Sheng Gong, Mihee Ko, William S. Garver, Katsuhiko Yanagisawa, and Makoto Michikawa. 2005. “Altered Cholesterol Metabolism in Niemann-Pick Type C1 Mouse Brains Affects Mitochondrial Function.” Journal of Biological Chemistry 280 (12): 11731–39. https://doi.org/10.1074/jbc.M412898200.

Zhang, Kuixing, Yuxin Lu, Kevin T. Harley, and Minh Ha Tran. 2017. “Atypical Hemolytic Uremic Syndrome: A Brief Review.” Hematology Reports. Page Press Publications. https://doi.org/10.4081/hr.2017.7053.

Zhang, Zhe, and Marni J Falk. 2014. “Integrated Transcriptome Analysis across Mitochondrial Disease Etiologies and Tissues Improves Understanding of Common Cellular Adaptations to Respiratory Chain Dysfunction.” The International Journal of Biochemistry & Cell Biology 50: 106–11. https://doi.org/10.1016/j.biocel.2014.02.012.

170

Zhao, G., X. Zhou, L. Wang, G. Li, C. Kisker, W.J. Lennarz, and H. Schindelin. 2006. “Structure of the Mouse Peptide N-Glycanase-HR23 Complex Suggests Co-Evolution of the Endoplasmic Reticulum-Associated Degradation and DNA Repair Pathways.” Journal of Biological Chemistry. https://doi.org/10.1074/jbc.M600137200.

Zhao, Gang, Guangtao Li, Xiaoke Zhou, Ichiro Matsuo, Yukishige Ito, Tadashi Suzuki, William J. Lennarz, and Hermann Schindelin. 2009. “Structural and Mutational Studies on the Importance of Oligosaccharide Binding for the Activity of Yeast PNGase.” Glycobiology 19 (2): 118–25. https://doi.org/10.1093/glycob/cwn108.

Zhao, Gang, Xiaoke Zhou, Liqun Wang, Guangtao Li, Hermann Schindelin, and William J. Lennarz. 2007. “Studies on Peptide:N-Glycanase-P97 Interaction Suggest That P97 Phosphorylation Modulates Endoplasmic Reticulum-Associated Degradation.” Proceedings of the National Academy of Sciences of the United States of America 104 (21): 8785–90. https://doi.org/10.1073/pnas.0702966104.

Zhao, Wei-peng, B O Zhu, Yu-zhong Duan, and Zheng-tang Chen. 2009. “Neutralization of Complement Regulatory Proteins CD55 and CD59 Augments Therapeutic Effect of Herceptin against Lung Carcinoma Cells,” 1405–11. https://doi.org/10.3892/or.

Zharkova, Olga, Teja Celhar, Petra D. Cravens, Anne B. Satterthwaite, Anna Marie Fairhurst, and Laurie S. Davis. 2017. “Pathways Leading to an Immunological Disease: Systemic Lupus Erythematosus.” Rheumatology (Oxford, England). https://doi.org/10.1093/rheumatology/kew427.

Zhou, X., G. Zhao, J.J. Truglio, L. Wang, G. Li, W.J. Lennarz, and H. Schindelin. 2006. “Structural and Biochemical Studies of the C-Terminal Domain of Mouse Peptide-N- Glycanase Identify It as a Mannose-Binding Module.” Proc.Natl.Acad.Sci.Usa 103: 17214– 19. https://doi.org/10.2210/PDB2G9G/PDB.

Zhou, Xiaoke, Gang Zhao, James J. Truglio, Liqun Wang, Guangtao Li, William J. Lennarz, and Hermann Schindelin. 2006. “Structural and Biochemical Studies of the C-Terminal Domain of Mouse Peptide-N-Glycanase Identify It as a Mannose-Binding Module.” Proceedings of the National Academy of Sciences of the United States of America 103 (46): 17214–19.

171

https://doi.org/10.1073/pnas.0602954103.

Zipfel, Peter F, and Christine Skerka. 2009. “Complement Regulators and Inhibitory Proteins.” Nature Reviews. Immunology 9 (10): 729–40. https://doi.org/10.1038/nri2620.

Zuk, Or, Eliana Hechter, Shamil R. Sunyaev, and Eric S. Lander. 2012. “The Mystery of Missing Heritability: Genetic Interactions Create Phantom Heritability.” Proceedings of the National Academy of Sciences of the United States of America 109 (4): 1193–98. https://doi.org/10.1073/pnas.1119675109.

Zuk, Or, Stephen F Schaffner, Kaitlin Samocha, Ron Do, Eliana Hechter, Sekar Kathiresan, Mark J Daly, Benjamin M Neale, Shamil R Sunyaev, and Eric S Lander. 2014. “Searching for Missing Heritability: Designing Rare Variant Association Studies.” PNAS, 455–564. https://doi.org/10.1073/pnas.1322563111.

172

173