Developing Drug and Gene Therapies for Peroxisome Biogenesis Disorders of the Zellweger Spectrum

Home , Ayanin, PEX1, Retusin (flavonol), Retusin (isoflavone)

Developing drug and gene therapies for peroxisome biogenesis disorders of the Zellweger Spectrum

Catherine Argyriou

Department of Human Genetics McGill University, Montréal, Canada June 2018

A thesis submitted to McGill University in partial fulfillment of the requirements of the degree of Doctor of Philosophy

Zellweger spectrum disorder (ZSD) usually results from biallelic mutations in PEX genes required for peroxisome biogenesis. PEX1-G843D is a common hypomorphic allele associated with milder disease. We previously showed that fibroblasts from patients with a PEX1-G843D allele recovered peroxisome functions when cultured with the nonspecific chaperone betaine and flavonoid acacetin diacetate. To identify more effective flavonoids for preclinical trials, we compared 54 flavonoids using our cell-based peroxisomal assays. Diosmetin showed the most promising combination of potency and efficacy; co-treatments of diosmetin and betaine showed the most robust additive effects. This was confirmed by 5 independent assays in primary PEX1-G843D patient cells. Neither agent was active in PEX1 null cells. I propose that diosmetin acts as a pharmacological chaperone to improve stability, conformation, and function of PEX1/PEX6 exportomer complexes. All individuals with a PEX1-G843D allele develop a retinopathy that progresses to blindness. To investigate pathophysiology and identify endpoints for experimental trials, I used the knock-in mouse model for the equivalent human mutation, PEX1-G844D. I characterized the progression of retinopathy and found reduced cone cell function and number early in life with more gradual deterioration of rod cell function. Electron microscopy at later stage retinopathy showed disorganization of photoreceptor inner segments and enlarged mitochondria. As retino-cortical function was relatively well-preserved, I propose that the vision defect in the Pex1-G844D mouse is primarily at the retinal level. In contrast to the human PEX1-G843D which behaves as a misfolded, degraded protein in fibroblasts, I found equivalent amounts of murine Pex1-G844D and wild-type Pex1 proteins in retina and other tissues. I propose that murine PEX1-G844D is stable but subfunctional. To determine if we could slow visual loss in PEX1-mediated ZSD, I performed a proof-of-concept trial for PEX1 retinal gene augmentation therapy using the PEX1-G844D mouse. Subretinal injections of AAV-PEX1 at both early and later disease stages resulted in a non-statistically significant trend of improved visual acuity after 2 months, and a twofold improvement in both rod- and cone-mediated retinal electrophysiological response after 5-6 months. Neither injection, viral capsid exposure nor the transgenic protein negatively altered retinal histology or visual response. These results support the potential of retinal gene augmentation to improve vision in patients with ZSD at both earlier and later stages of disease.

RÉSUMÉ

Le Trouble de la biogénèse des peroxysomes du spectre de Zellweger (ZSD) est généralement causé par des mutations bialléliques dans les gènes PEX, qui sont requis pour la biogénèse des peroxysomes. L’allèle hypomorphe PEX1-G843D est le plus fréquent et est associé à une présentation plus légère de la maladie. Nous avons démontré précédemment que les fibroblastes provenant de patients avec un allèle PEX1-G843D récupèrent leurs fonctions peroxysomales lorsque le chaperon non-spécifique bétaïne et le flavonoïde diacétate d’acacétine sont ajoutés à la culture. Dans le but d’identifier les flavoinoïdes les plus efficaces pour des essais pré-cliniques, nous avons comparé 54 flavonoïdes en utilisant des tests cellulaires de la fonction peroxysomale. La combinaison de puissance et d’efficacité la plus prometteuse a été attribuée à la diosmétine; le co-traitement avec la diosmétine et le bétaïne a eu l’effet additif le plus robuste. Cela a été confirmé par 5 tests indépendants dans les cellules primaires PEX1-G843D provenant de patients. Aucun de ces agents n’a été actif dans les cellules nulles pour PEX1. Je propose que la diosmétine agisse comme un chaperon pharmacologique pour améliorer la stabilité, la conformation et la fonction des complexes d’exportation PEX1/PEX6.

Toutes les personnes atteintes de ZSD qui ont un allèle PEX1-G843D développent une rétinopathie qui progresse jusqu’à la cécité. J’ai utilisé un modèle de souris knock-in avec la mutation PEX1- G844D, équivalente à la mutation chez les humains, dans le but d’investiguer sur la physiopathologie et d’identifier des paramètres cliniques à mesurer pour les essais expérimentaux. J’ai caractérisé la progression de la rétinopathie et démontré une diminution de la fonction et du nombre de cônes tôt dans la vie des souris, avec une détérioration plus graduelle de la fonction des bâtonnets. L’imagerie à microscope électronique à un stade plus avancé de la rétinopathie a démontré une désorganisation des segments intérieurs des photorécepteurs et des mitochondries élargies. Puisque la fonction rétino- corticale était relativement bien préservée, je propose que la défectuosité de la vision chez les souris PEX1-G844D se situe d’abord au niveau de la rétine. Alors que la protéine PEX1-G843D humaine se comporte comme une protéine mal repliée qui est dégradée dans les fibroblastes, j’ai découvert qu’il y a des quantités similaires de protéines PEX1-G844D et de Pex1de type sauvage dans la rétine et d’autres tissus chez la souris. Je propose donc que le Pex1-G844D chez la souris est stable, mais moins fonctionnelle. Pour déterminer si on peut ralentir la perte de vision chez les personnes atteintes de ZSD dû à un défaut du gène PEX1, j’ai réalisé un essai de preuve de concept pour une thérapie d’augmentation du gène PEX1 dans la rétine des souris PEX1-G844D. Des injections sous-rétiniennes de l’AAV-PEX1 aux stades précoces et avancés de la maladie ont entraîné tendance vers une amélioration de l’acuité visuelle, sans être statistiquement significatives. De plus, cela a permis de doubler les réponses iii

électrophysiologiques de la rétine après 5-6 mois. Ni l’injection, ni l’exposition à la capside virale ou la protéine transgénique n’a négativement altéré l’histologie de la rétine ou la réponse visuelle. Ces résultats démontrent le potentiel d’une augmentation génique rétinienne pour améliorer la vision chez les patients atteints de ZSD, aux stades précoce et avancé de la maladie.

TABLE OF CONTENTS

ABSTRACT ...... ii RÉSUMÉ ...... iii LIST OF ABBREVIATIONS ...... 6 LIST OF FIGURES ...... 8 LIST OF TABLES ...... 9 ACKNOWLEDGEMENTS ...... 10 PREFACE TO THE THESIS ...... 11 CONTRIBUTION OF AUTHORS ...... 12 CHAPTER I: INTRODUCTION AND LITERATURE REVIEW ...... 13 1.1 OVERVIEW AND STUDY RATIONALE ...... 14 1.2 PEROXISOME BIOLOGY AND CLINICAL DISORDERS ...... 15 1.2.1 Peroxisome biochemistry ...... 15 1.2.2 Peroxisome biogenesis ...... 18 1.2.3 Clinical disorders ...... 22 1.2.3.1 Zellweger Spectrum Disorder (ZSD) ...... 24 1.2.3.2 Atypical ZSD presentations ...... 26 1.2.3.3 Other disorders that affect peroxisome biogenesis ...... 28 1.2.4 Clinical diagnosis ...... 28 1.2.5 Disease management ...... 29 1.3 MOUSE MODELS OF PBD-ZSD ...... 31 1.4 THE MOUSE VISUAL SYSTEM AND ITS ASSESSMENT ...... 32 1.4.1 The mouse visual pathway ...... 33 1.4.2 Assessing visual function in the laboratory mouse ...... 34 1.4.2.1 Electrophysiology ...... 35 1.4.2.2 Visual acuity ...... 37 1.5 HYPOTHESES AND OBJECTIVES ...... 38 CHAPTER II: Zellweger spectrum disorder patient-derived fibroblasts with the PEX1- Gly843Asp allele recover peroxisome functions in response to flavonoids ...... 41 2.1 ABSTRACT ...... 42 2.2 INTRODUCTION ...... 43

2.3 MATERIALS AND METHODS ...... 45 2.3.1 Cell lines and culture conditions ...... 45 2.3.2 Chemicals ...... 45 2.3.3 Flavonoid screen ...... 45 2.3.4 Immunoblotting...... 46 2.3.5 PEX5 localization ...... 46 2.3.6 Lipid analyses ...... 47 2.3.7 Protein bioinformatics ...... 48 2.4 RESULTS...... 48 2.4.1 Recovery of peroxisome matrix protein import by flavonoids ...... 48 2.4.2 Dose response studies ...... 49 2.4.3 SAR of flavonoids...... 49 2.4.4 Combination treatments ...... 52 2.4.5 Independent validation of peroxisome import recovery ...... 52 2.4.6 PEX1 ATP-binding domain 2 and its role in ligand binding ...... 55 2.5 DISCUSSION ...... 55 2.5.1 SAR of flavonoids tested ...... 55 2.5.2 Proposed mechanism of action of flavonoids ...... 57 2.5.3 Potential therapeutic implications ...... 58 2.6 ACKNOWLEDGMENTS ...... 59 2.7 CONFLICTS OF INTEREST ...... 60 2.8 CONNECTING TEXT BETWEEN CHAPTERS II AND III ...... 73 Chapter III: Characterization of Retinopathy in the PEX1-Gly844Asp Mouse Model for Mild Zellweger Spectrum Disorder ...... 74 3.1 ABSTRACT ...... 75 3.2 INTRODUCTION ...... 76 3.3 MATERIALS AND METHODS ...... 78 3.3.1 Animal husbandry ...... 78 3.3.2 Electrophysiology ...... 78 3.3.3 Assessing visual acuity with the Virtual Optomotor System...... 80 3.3.4 Lipid analysis ...... 80

3.3.5 Immunoblotting...... 82 3.3.6 Retinal histology and immunohistochemistry ...... 82 3.3.7 Transmission electron microscopy ...... 83 3.4 RESULTS...... 83 3.4.1 Assessing retinal function ...... 83 3.4.2 Assessing the primary visual pathway (from the retina to the visual cortex) ...... 84 3.4.3 Retinal structure ...... 85 3.4.4 Peroxisome protein and biochemical metabolite levels ...... 86 3.5 DISCUSSION ...... 87 3.5.1 Visual response and retinal architecture ...... 87 3.5.2 Peroxisome dysfunction and the retina ...... 90 3.5.3 Clinical implications: The translatability of the model...... 91 3.6 ACKNOWLEDGMENTS ...... 93 3.7 CONFLICTS OF INTEREST ...... 93 3.8 CONNECTING TEXT BETWEEN CHAPTERS III AND IIV ...... 102 CHAPTER IV: rAAV-mediated PEX1 Gene Augmentation Improves Visual Function in the PEX1-Gly844Asp Mouse Model for Mild Zellweger Spectrum Disorder ...... 103 4.1 ABSTRACT ...... 104 4.2 INTRODUCTION ...... 106 4.3 MATERIALS AND METHODS ...... 108 4.3.1 Proviral plasmids and AAV production ...... 108 4.3.2 In vitro titer assay for individual AAV capsid variants ...... 109 4.3.3 Cell transduction and immunoblotting...... 109 4.3.4 Peroxisome import after viral transduction...... 109 4.3.5 Animal husbandry ...... 110 4.3.6 Design of in vivo experiments...... 111 4.3.7 Vector delivery...... 111 4.3.8 Electrophysiology ...... 112 4.3.9 Visual acuity using the Virtual Optomotor System ...... 112 4.3.10 Retinal immunohistochemistry ...... 113 4.4 RESULTS...... 113

4.4.1 Recovery of peroxisome import by rAAV-delivered hPEX1 in human and mouse cells 113 4.4.2 rAAV8-mediated hPEX1.HA protein is expressed in mouse retina ...... 114 4.4.3 Visual function improves by 8 weeks post subretinal gene delivery ...... 115 4.4.4 Visual function but not functional vision improves 5 or 6 months post subretinal gene delivery ...... 116 4.5 DISCUSSION ...... 117 4.5.1 PEX1 gene augmentation improves retinal function ...... 117 4.5.2 Functional relevance ...... 117 4.5.3 Clinical implications ...... 119 4.6 ACKNOWLEDGEMENTS ...... 120 4.7 CONFLICTS OF INTEREST ...... 121 CHAPTER V: DISCUSSION ...... 131 5.1 OVERVIEW AND MAJOR FINDINGS ...... 132 5.1.1 Small molecule therapy is a viable treatment option for ZSD ...... 132 5.1.2 The study of PEX1-G844D retinopathy identifies therapeutic outcome measures ...... 133 5.1.3 Gene augmentation therapy is a viable treatment option for ZSD ...... 134 5.2 GENERAL DISCUSSION ...... 134 5.2.1 Additional pharmacological therapies under development for ZSD ...... 134 5.2.1.1 Pharmacological enhancement of activity of a defective PEX protein ...... 134 5.2.1.2 Regulation of peroxisome numbers ...... 135 5.2.1.3 Nonsense suppression ...... 136 5.2.2 Alternative therapeutic approaches for ZSD ...... 136 5.2.2.1 Gene therapy ...... 136 5.2.2.1 Gene editing ...... 136 5.2.3 Interventional clinical trials ...... 137 5.2.3.1 Natural history studies ...... 137 5.2.3.2 Newborn screening ...... 138 5.2.4 An expanding role for peroxisomes in disease ...... 138 5.3 FUTURE DIRECTIONS...... 139 5.3.1 Further development of flavonoid and other pharmacological therapies ...... 139

5.3.2 Deeper phenotyping of the PEX1-G844D mouse model ...... 140 5.3.3 Retinal gene therapy continued ...... 141 5.4 CONCLUDING SUMMARY ...... 142 5.5 ORIGINAL CONTRIBUTIONS ...... 142 REFERENCES ...... 144

LIST OF ABBREVIATIONS µg – microgram µL – microlitre µM – micromolar µm – micrometre μs - microsecond μV – microvolt AAA-ATPase – ATPase associated with various cellular activities AAV – adeno-associated virus ABC – ATP binding cassette ABCD1 – ATP-binding cassette sub-family D member 1 ABCD3 – ATP-binding cassette sub-family D member 3 ATP – adenosine triphosphate bGH – bovine growth hormone BSA – bovine serum albumin c/d – cycles per degree CMV – cytomegalovirus db – decibel DHA – docosahexanoic acid, C22:6 n-3 DNA – deoxyribonucleic acid ffERG – full-field flash electroretinogram g – gram gc – genome copies GFAP – glial fibrillary acidic protein GFP – green fluorescent protein HA – Human influenza hemagglutinin HEK – human embryonic kidney HepG2 – human hepatocellular carcinoma hz – hertz IF – immunofluorescence IHC – immunohistochemistry INL – inner nuclear layer IRD – Infantile Refsum Disease IS – inner segment ITR – inverted terminal repeat L – litre LC-MS/MS – Liquid chromatography-tandem mass spectrometry LPM – litres per minute lyso-PC – lysophosphatidylcholine mg – milligram mL – millilitre mM – millimolar MOI – multiplicity of infection ms – millisecond NALD – Neonatal adrenoleukodystrophy NGS – normal goat serum

NMR – nuclear magnetic resonance spectroscopy OCT – optical coherence tomography OKN – optokinetic nystagmus ONL – outer nuclear layer OP – oscillatory potential OPL – outer plexiform layer OS – outer segment PBD – peroxisome biogenesis disorder PBS – phosphate buffered saline PC – phophatidylcholine PCR – polymerase chain reaction PE – phosphatidylethanolamine PEX - Peroxin PL – plasmalogen pmol – picomole PPAR – peroxisome proliferator-activated receptor PTS – peroxisome targeting signal qPCR – quantitative polymerase chain reaction rAAV – recombinant adeno-associated virus RNA – ribonucleic acid RT-PCR – reverse transcriptase polymerase chain reaction rpm – revolutions per minute SAR – structure activity relationships sec - second SPR – surface plasmon resonance TEM – transmission electron microscopy UTR – untranslated region VCP – Valosin containing protein VEP – visual evoked potential vg – vector genomes VLCFA – very long chain fatty-acid X-ALD – X-linked adrenoleukodystrophy ZS – Zellweger Syndrome ZSD – Zellweger Spectrum Disorder

LIST OF FIGURES

Figure 1.1: Peroxisome matrix protein import. (1) PTS receptor binding...... 22 Figure 1.2. The mouse eye and retina...... 34 Figure 1.3. Representation of a scotopic ffERG waveform...... 36 Figure 1. 4. The virtual optomotor system (OptoMotry, CerebralMechanics Inc.) for determining rodent visual acuity...... 38 Figure 2.1. Flavonoid subclasses...... 61 Figure 2.2 Recovery of peroxisome import in PEX1-G843D-GFP-PTS1 cells...... 61 Figure 2.3. Chemical structures of the most active flavonoids ...... 62 Figure 2.4. Effects of combination treatment on PEX1-G843D cells...... 63 Figure 2.5. Effect of betaine, diosmetin, and combination treatments on PEX5 localization and peroxisome metabolite levels in PEX1-deficient primary fibroblasts...... 64 Figure 2.6. Effect of betaine, diosmetin, and combination treatments on PEX protein levels in primary fibroblasts...... 65 Supplemental Figure 2.1 Dose responses for GFP-PTS1 import in PEX1-G843D-GFP-PTS1 cells treated with flavonoids...... 68 Supplemental Figure 2.2. Conservation of the ATP binding domain in VCP and PEX1...... 69 Figure 3.1. ffERG maturation in PEX1-G844D mice...... 94 Figure 3.2. Prominence of oscillatory potentials (OPs) in ffERGs of PEX1-G844D mice...... 95 ...... 96 Figure 3.3 Visual evoked potential (VEP) and visual acuity in PEX1-G844D mice...... 96 Figure 3.4. Immunohistochemistry shows preservation of various cell structures in PEX1- G844D retinas...... 97 Figure 3.5. Immunohistochemistry shows decreased staining of cone and bipolar cells in Pex1-G844D retinas...... 98 Figure 3.6. Prominence of glial fibrillary acidic protein with age in PEX1-G844D retinas...... 98 Figure 3.7. Transmission electron microscopy of PEX1-G844D retinas...... 99 Figure 3.8. Peroxisome protein and biochemical metabolite levels in Pex1-G844D retinas...... 100 Supplemental figure 3.1. Retinal histology...... 101 Figure 4.1. Proviral plasmid design and rAAV8 vector expression...... 122 Figure 4.2. Recovery of peroxisome import by rAAV-delivered hPEX1 in human and mouse cells...... 123 Figure 4.3. Schematic representation of experimental design...... 124 Figure 4.4. Preliminary ffERG, visual acuity, and confirmation of rAAV-delivered protein expression in the ‘sacrificial’ cohort...... 125 Figure 4.5. Preliminary ffERG on a subset of animals from ‘prevention’ and ‘recovery’ cohorts...... 126 Figure 4.6. Endpoint ffERG and visual acuity measures in ‘prevention’ and ‘recovery’ cohorts...... 127

Supplemental figure 4.1. Baseline ffERG measures...... 129 Supplemental figure 4.2. Endpoint ffERG waveforms...... 130

LIST OF TABLES

Table 2.1. All informative comparisons at the indicated A-, B-, and C-ring positions...... 66 Table 2.2 All informative comparisons at the indicated A-, B-, and C-ring positions...... 67 Supplemental Table2.1. All flavonoids evaluated in this study...... 70 Supplemental Table 2.2 All informative comparisons of an active flavonoid and another flavonoid that differ by substitutions at a single position or by subclass...... 71 Table 4.1. Comparison of baseline to endpoint ffERG...... 128

ACKNOWLEDGEMENTS

First and foremost, I must thank my supervisor Dr. Nancy Braverman, who allows her students to be the star of their work. Owing to her, I have learned so much and become a stronger scientist through these past five years. I thank my supervisory committee members, Dr. Shoubridge and Dr. Jardim, for reading those yearly reports, attending the meetings, and enthusiastically helping me design the best experiments possible. Next, I must thank our collaborators across the globe for their invaluable mentoring and contributions to our projects, primarily Dr. Joseph Hacia, who is always there to bounce around ideas and find a way to get things done. My lab-mates and colleagues at the MUHC-RI, too many to name (but you know who you are), have helped in every way from sharing protocols, critiquing results, to just making work a fun place to be. I consider you not only colleagues, but friends.

To all those affected by peroxisomal disorders: Soon after starting at the lab, I attended the GFPD Family and Scientific Meeting, where I met many people touched by PBD-ZSD and their families. It has been a privilege to know you and witness your courage. Thank-you to the foundations for motivating us and believing in our research. I look forward to more collaborations in years to come!

To my friends old and new: No matter how far apart we are, it is as though we have not skipped a beat. Thank-you for being there through it all, some for over 20 years. You can finally stop asking me when I’ll be done!

My family: I would not be here (or a useful person) if not for them. They fostered an environment of curiosity and original thought, and provided every opportunity for my development. They let me take my own path, but cleared it as best they could.

Last but certainly not least, I thank my partner in life, Yioti, who is my biggest fan and believes in me to an absurd extent. Thank-you for keeping me well-fed and well-supported through these last five years; this degree is not mine, it’s ‘ours’.

PREFACE TO THE THESIS

This thesis is written in accordance with the guidelines of the McGill University Graduate and Postdoctoral Studies Office. This thesis is comprised of five chapters. Three of these chapters are scientific data chapters written in manuscript format. Chapter II contains a manuscript accepted for publication by the Journal of Cellular Biochemistry. Chapter III and IV are manuscripts that are currently in preparation. Chapter I is a review of the current background literature relevant to this thesis. Sections 1.2.1 to 1.3 are mainly excerpted from a published literature review of which I am first author, reproduced in this thesis with publisher permissions. This chapter reviews peroxisome biology, disorders of peroxisome biogenesis of the Zellweger Spectrum (PBD-ZSD), mouse models of peroxisome disease, mouse ocular biology, and assessment of visual function in mice. Also included are the objectives of the thesis. Chapter II, III, and IV contain experimental data in manuscript format. Chapter II describes candidate drug studies in PBD-ZSD cells, including structure-activity relationship studies and validation of candidate compounds. Chapter III describes the characterization of the retinal phenotype in the PEX1-G844D mouse model for mild PBD-ZSD. Chapter IV describes the results of a proof-of-concept preclinical retinal gene therapy trial to recover retinal function in the PEX1-G844D mouse model. Connecting texts between Chapters II, III, and IV link the manuscripts. Chapter V is a discussion of the thesis findings and suggested future studies.

CONTRIBUTION OF AUTHORS

The experimental research presented herein was carried out by Catherine Argyriou under the supervision of Dr. Nancy Braverman. Catherine Argyriou carried out the majority of experiments, experimental design, data analysis and interpretation, and manuscript writing. Dr. Nancy Braverman supervised the work, and participated in experimental design, data interpretation, and manuscript editing. The contribution of the remaining authors to each chapter is described below.

For Chapter II, Gillian MacLean performed SAR and dose response experiments. Xuting Sun assisted with PEX5 localization experiments. Dr. Erminia Di Pietro performed LC-MSMS lipid analyses. Sara Birjandian and Panteha Saberian performed preliminary protein quantification studies. Dr. Joseph Hacia participated in statistical analysis and manuscript preparation.

For Chapter III, Dr. Anna Polosa (supervised by Dr. Pierre Lachapelle) assisted with electrophysiology and histology experiments and data analysis. Bruno Ceçyre (supervised by Dr. Jean-François Bouchard) performed visual acuity measures. Dr. Erminia Di Pietro performed LC-MSMS lipid analyses. Monica Hsieh assisted with histology imaging and electrophysiology quantification. Dr. Wei Cui established the genotyping protocol for our PEX1-G844D mouse colony.

For Chapter IV, Dr. Ji Yun Song (supervised by Dr. Jean Bennett) performed subretinal injections. Dr. Anna Polosa (supervised by Dr. Pierre Lachapelle) assisted with electrophysiology and histology data analysis. Devin McDougald (supervised by Dr. Jean Bennett) cloned the expression plasmids. Bruno Ceçyre (supervised by Dr. Jean-François Bouchard) performed visual acuity measures. Dr. Erminia Di Pietro performed LC-MSMS lipid analyses. Ning Huang (supervised by Dr. Joseph Hacia) performed PEX1-G844D mouse fibroblast studies.

CHAPTER I: INTRODUCTION AND LITERATURE REVIEW

1.1 OVERVIEW AND STUDY RATIONALE

Peroxisomes biogenesis disorders (PBD) are a heterogeneous group of diseases caused by a defect in peroxisome biogenesis. They encompass two phenotypic groups: 1. Zellweger spectrum disorder, (ZSD) and 2. Rhizomelic Chondrodysplasia Punctata type 1 (RCDP1). As the name suggests, ZSD presents in a spectrum of manifestations, with severe, intermediate and milder forms. These were previously named as Zellweger syndrome (ZS), Neonatal Adrenoleukodystrophy (NALD) and Infantile Refsum Disease (IRD), respectively. The phenotypic description of ZSD continues to expand as ‘atypical’ manifestations are described [1].

Peroxisomes are spherical, membrane bound organelles up to 1µm in diameter, numbering up to several hundred per mammalian cell and derived from endoplasmic reticulum and mitochondrial membranes. Highly conserved throughout evolution amongst most eukaryotes, peroxisomes are critical for normal life. Peroxisomes were first identified by biochemist Christian DeDuve in the 1960s, as cytoplasmic particles containing hydrogen peroxide-generating oxidases [2]. Later, the neurologist Hans Zellweger reported a group of children with craniofacial dysmorphism and malformations in the brain, eye, liver, and kidney [3]. These cases, initially described under the eponym of “cerebro-hepato-renal-syndrome”, were subsequently termed Zellweger syndrome (ZS) [4], and represent the most severe manifestation of ZSD as we know it today. The reported absence of morphologically identifiable peroxisomes in hepatocytes and renal tubular cells of affected individuals using the peroxisomal matrix enzyme catalase as a marker, [5] demonstrated a connection between ZS and peroxisome dysfunction. The etiological connection, however, was not recognized until the identification of multiple peroxisomal enzymes and their deficiencies in affected patients. Similar findings were later demonstrated in the intermediate and milder ZSD phenotypes and RCDP1 [6-8].

Peroxisome biogenesis disorders of the Zellweger spectrum (PBD-ZSD) are caused by defects in any one of 13 PEX proteins, encoded by PEX genes, which are required for peroxisome biogenesis [1, 9]. One common hypomorphic allele, PEX1 p.[Gly843Asp] or G843D (c.2528G>A), has high residual activity and confers a milder ZSD phenotype [10, 11]. As this allele represents up to 30% of all ZSD alleles [9, 12, 13], identifying treatments for it would

benefit a significant number of ZSD patients, and potentially prevent or ameliorate disease progression. In this thesis I report the progress of therapies in development to target this allele, namely small molecule therapy in patient cell-based studies, and rAAV-mediate gene augmentation (gene therapy) in the Pex1-G844D mouse model for mild ZSD.

1.2 PEROXISOME BIOLOGY AND CLINICAL DISORDERS

Sections of this introduction are reprinted from Translational Science of Rare Diseases, Metabolic Diseases, Vol 2, Agyriou C, D’Agostino M D, Braverman N, Peroxisome Biogenesis Disorders, Pages 847-865., Copyright (2017), with permission from IOS Press. The publication is available at IOS Press through http://dx.doi.org/10.3233/TRD-160003. The reprinted sections are italicized and demarcated by quotations.

1.2.1 Peroxisome biochemistry

“There are several hundred peroxisomes in all mammalian cells, each containing more than 50 matrix enzymes required for multiple vital metabolic pathways, including the catabolism and synthesis of important cellular lipids [reviewed in [14]]. The best characterized peroxisomal pathways include β-oxidation of very long chain fatty acids (VLCFA), -oxidation of methyl-

branched phytanic acid and the biosynthesis of ether phospholipid (plasmalogen)𝛼𝛼 . The metabolites of these pathways are routinely screened when a peroxisome biogenesis defect is suspected.

Peroxisomal β-oxidation utilizes a variety of different fatty acid substrates in catabolic and synthetic processes: very long (≥ C22) straight chain saturated and unsaturated fatty acids (VLCFA), (2S) methyl-branched chain pristanic acid and (25S) di- and tri-hydroxycholestanoic acids (D/THCA). Docosahexanoic acid (DHA), a critical polyunsaturated omega-3 fatty acid is synthesized in the peroxisome after β-oxidation of its precursor [15-17]. D/THCA are C27 intermediary bile acids that undergo one step of β-oxidation to form the primary C24 bile acids, cholic and chenodeoxycholic acid. Peroxisomal β-oxidation also participates in the regulation of other complex lipids, including pro-inflammatory molecules such as eicosanoids, leukotrienes, and prostaglandins [18-20] [reviewed in [14]]. Furthermore, peroxisomes are the site of β- oxidation of dicarboxylic acids [21]. The initial oxidation step is performed in peroxisomes by

acyl- CoA oxidase 1 (ACOX1 for VLCFA, ACOX2 for pristanic and D/THCA, and ACOX3 for pristanic acid) generating hydrogen peroxide (hence the name, peroxisome) as a byproduct that is detoxified by peroxisomal catalase [22-26] [reviewed in [27]]. D-bifunctional protein (DBP or HSD17B4) catalyzes the second (hydration) and third (dehydration) step of peroxisomal β- oxidation [28-30]. The final step is catalyzed by thiolase (either 3-oxoacyl-CoA thiolase (ACAA1) or SCPx in humans) [reviewed in [14]]. The peroxisomal fatty acid oxidation system is generally not able to degrade fatty acids to completion, so C16 or shorter fatty acids are transported as an acylcarnitine derivative to mitochondria for complete oxidation [17, 31-34].

3-methyl branched fatty acids (predominantly phytanic acid), because of the position of the methyl group, require a preliminary -oxidation step in order to undergo β-oxidation [35, 36].

This requires the peroxisomal enzymes,𝛼𝛼 phytanyl-CoA hydroxylase (PhyH), phytanyl-CoA lyase, and pristanal dehydrogenase, to generate pristanic acid [37-40] [reviewed in [14]]. Both pristanic acid and D/THCA are generated as (R) and (S) enantiomers, which require conversion to the (S) form by peroxisomal alpha-methyl-CoA racemase (AMACR) before β-oxidation [41]. Phytanic acid is exclusively dietary in origin and obtained from meats and dairy products of ruminant animals and certain fatty fishes [42, 43]. It accumulates in patients with ZSD and RCDP1, as well as in patients with single enzyme defects in PhyH [reviewed in [44]].

The committing steps for ether phospholipid synthesis occur in the peroxisome to generate the alkylglycerol backbone and require the peroxisomal enzymes fatty acyl-CoA reductase (FAR1), dihydroxyacetonephosphate acyltransferase (GNPAT), and alkyl-dihydroxy-acetonephosphate synthase (AGPS). Further metabolism in the ER generates the mature ether phospholipids, the majority of which are plasmalogens and contain a fatty alcohol linked by a vinyl ether bond to the sn-1 position of the glycerol backbone and are enriched in DHA and arachidonic acid (AA) at the sn-2 position [reviewed in [45, 46]]. Plasmalogens are a unique class of membrane glycerophospholipids by virtue of their vinyl ether bond, which provides specialized and critical functions to cellular membranes [47-49]. These properties include unique physical membrane attributes [50], signal transduction through released DHA and AA, as well as the remaining lysophospholipid [51], and the ability to provide oxidant protection to other membrane lipids [reviewed in [45, 46]]. Plasmalogens are enriched in heart and skeletal muscles, kidney, brain, and erthrocytes, whereas liver has very low concentration of plasmalogens [52]. Plasmalogen

deficiency causes the RCDP phenotype [53-55]. In ZSD, the phenotype is more complex due to multiple enzyme deficiencies.

In addition, peroxisomes contain enzymes involved in the catabolism of D-amino acids (D- amino acid oxidase) and L-lysine (pipecolic acid oxidase), polyamine oxidation, detoxification of glyoxylate (alanine-glyoxylate aminotransferase, AGXT), reactive oxygen species, and xenobiotics. [For a full catalog of peroxisomal enzymes see [14]]. Cholesterol levels tend to be low in PBD patients [56-58], although the underlying mechanism remains to be resolved [59]. Recent studies suggest that peroxisome dysfunction leads to chronic ER stress and consequent dysregulation of the sterol response pathway [60].

The catalog of peroxisomal enzymes continues to increase through the use of bioinformatic searches for proteins containing the conserved targeting signals [61-64], as well as proteomics and mass spectrometry techniques [65, 66]. Most peroxisomal enzymes are unique to peroxisomes, but some are shared with other cellular compartments, including mitochondria and cytosol. These multiple subcellular localizations may be due to the presence of multiple targeting signals within the same protein, as shown for 3- hydroxymethylglutary-CoA lyase [67], alpha methyl-CoA racemase [68], malonyl-CoA decarboxylase [69] and others, or relate to properties of the oligomeric protein complex as in epoxide hydrolase [70], copper-zinc superoxide dismutase [71], and alanine glyoxylate aminotransferase [72].

Shared enzymes and the movement of substrates and products through various cellular compartments position peroxisomes to be versatile organelles that dynamically adapt their number and metabolic functions in response to cellular needs. This is observed in yeast when grown on fatty acid media, by concomitant transcriptional up-regulation of peroxisomal β- oxidation enzymes and peroxisome proliferation. Rodent peroxisomes also proliferate secondary to increased β-oxidation activity and this is controlled by PPAR [73]. PPAR belongs to a

transcription factor superfamily that binds to, and is activated by,𝛼𝛼 a broad range𝛼𝛼 of fatty acids and is a key regulator of peroxisomal and mitochondrial β-oxidation of fatty acids. Nevertheless, PPAR agonists have thus far not been associated with peroxisome proliferation in humans

[74]. However,𝛼𝛼 a metabolic control of peroxisome abundance in humans is expected considering the reduced number and enlargement of peroxisomes observed in fibroblast cultures from both ZSD patients and patients with single enzyme defects in β-oxidation pathways (ACOX1 and

DBP) [75]. Human peroxisome numbers can also be increased through a PPARα independent mechanism that may rely on PEX11 activation [reviewed in [76]]. Significantly, the PPARγ coactivator (PGC)-1 , a regulator of cellular energy metabolism and mitochondrial biogenesis,

was shown also to stimulate𝛼𝛼 peroxisome biogenesis by upregulating PEX11, as well as other PEX genes [77].”

1.2.2 Peroxisome biogenesis

“Peroxisome biogenesis encompasses the different processes required to assemble and maintain functional peroxisomes including matrix protein import, synthesis of new organelles, and fission of existing organelles. The coordinated activity of 16 PEX proteins, or peroxins, encoded by their corresponding PEX genes, is required for this process in mammals. There is evidence that that peroxisome movement and assembly utilizes the cytoskeleton [78][reviewed in [79]]. Current models of eukaryotic peroxisome biogenesis have been recently reviewed [76, 80-82] and briefly are summarized below with emphasis on mammalian systems (see Figure 1.1 for diagram of matrix protein import). Peroxisomal enzymes are synthesized on cytosolic ribosomes and contain one of two peroxisomal targeting signals (PTS) required for their import into the peroxisome [reviewed in [83]]. PTS1, the most common, is a C-terminal tripeptide, -Ser-Lys-Leu or consensus variant thereof [84-86]. PTS2 is a N-terminal nonapeptide found in at least 3 mammalian enzymes: 3-oxoacyl-CoA thiolase (ACAA1), PhyH, and AGPS [61, 87-91]. PTS1 proteins are recognized by the cytosolic receptor, PEX5, which also acts as a chaperone for PTS1 proteins [92-94]. The crystal structure of PEX5 demonstrating the PTS1 binding site has been solved [95]. There are two isomers of PEX5: PEX5S and PEX5L, resulting from differential splicing of the primary transcript of the PEX5 gene [96]. The PTS2 sequence forms a structural motif that is recognized by the cytosolic receptor PEX7 [97]. Comparative homology modeling has identified the PEX7-PTS2- binding groove [61]. In mammals, PEX7 binds to PEX5L for targeting to the peroxisomal membrane [96, 98]. PEX5L and PEX7 receptors form cytosolic complexes with their ligands and efficiently ferry them to the peroxisome membrane [98]. The receptor/cargo complex binds at the peroxisome membrane to the PEX13 and PEX14 docking complex [99-103]. The cargo proteins are subsequently translocated into the matrix, possibly through a transient pore, which assembles at the peroxisome membrane to enable matrix protein translocation, before it is disassembled afterwards and its components recycled

for further rounds of protein import [104, 105]. Direct evidence for the existence of a transient import pore was shown by electrophysiological studies, which demonstrated that a Pex5p- Pex14p sub-complex of S. cerevisiae harbors pore forming activity [106]. Through this transient pore, which can widen to 9 nm, peroxisomes can import folded, oligomeric and cofactor bound proteins [106-108]. An alternate emerging model postulates that the peroxisomal import machinery displays a preference for monomeric proteins. In this model: (i) newly synthesized peroxisomal proteins are folded by cytosolic chaperones and released as soluble monomers; (ii) PEX5 binding to these monomers (PTS1) blocks their oligomerization; and finally, (iii) these monomeric cargoes are translocated by a single PEX5 molecule into the matrix of the organelle where oligomerization occurs [109].

After translocation, the cargo dissociates from the PTS-receptors and is released into the peroxisomal lumen [110]. The PTS receptors are released from the peroxisomal membrane for another import cycle or directed to the proteasome for their degradation [111-113]. Translocation of the ligands into the peroxisome matrix is also associated with the integral membrane complex: PEX2, PEX10, and PEX12. PEX2/10/12 are ubiquitin ligase (E3)-like proteins containing RING (really interesting new gene) domains and are involved in facilitating the mono- or polyubuiquitination based PTS receptor recycling and degradation [114]. In humans, PEX2 mono-ubiquitination of PEX5 occurs on a conserved cysteine residue in the N- terminal region, and is mediated by redundant ubiquitin-conjugating (E2)-enzymes (UbcH5a, UbcH5b and UbcH5c) [115, 116]. This mono-ubiquitination could provide a handle for PEX5 recycling back to the cytosol [117, 118]. If the monoubiquitination dependent recycling pathway is impaired, the PTS-receptors enter a polyubiquitination-dependent alternative pathway where PEX10/12 poly-ubiquitinate PEX5 on lysine residues, targeting it for degradation by the 26S proteasome [118, 119]. In wild type cells there is very little receptor turnover by degradation [119, 120].” Furthermore, the ubiquitination of PEX5 is redox sensitive, modulating its activity in times of extra-peroxisomal oxidative stress [121]. For a complete review of PEX5

ubiquitination and dislocation, see [122, 123].

“The ‘exportomer’ complex of PEX1, PEX6, and PEX26 is involved in the recycling of the PTS1- receptor, PEX5, and presumably PEX7, back to the cytosol for additional rounds of import [124]. PEX1 and PEX6 are members of the AAA protein family (ATPases Associated with

various cellular Activities) that contain one or two AAA cassettes responsible for ATP-binding and ATP hydrolysis. PEX1 and PEX6 each contain two ATPase domains (D1 and D2), and are therefore members of the double-ring ATPase family [reviewed in [125]]. These proteins typically form hexameric rings that change conformation during the ATPase cycle, providing the motion and force required for the variety of cellular functions they perform [126-128]. ATP binding is required for the interaction of PEX1 and PEX6 [129], and association with the peroxisome membrane occurs by interaction of the N-terminal region of PEX6 with PEX26 [130]. PEX5 export, but not import, is an ATP driven process, with the energy of ATP hydrolysis providing the conformational change required in the PEX1/6 complex to pull monoubiquitinated PEX5 out of the peroxisome membrane [131]. Recently, cryo-electron microscopy studies have shown that yeast Pex1/Pex6 ATPases form a unique double-ring structure in which the two proteins alternate around the ring [132-134]. These data shed light on the mechanism and function of this ATPase complex and suggest a role in peroxisomal protein import analogous to that of p97 in ER-associated protein degradation, in which the ATPase extracts polyubiquitinated proteins from the ER membrane [discussed in [125, 135]].

Peroxisomes can form de novo or by growth and division of pre-existing peroxisomes. PEX3, PEX16, and PEX19 are required for these processes and for correct localization of peroxisome membrane proteins (PMPs) [136-144]. In the de novo pathway, PEX16 is first incorporated into the ER, followed by PEX3 and other PMPs [145]. This leads to the formation of specialized “peroxisome-like” domains, or pre-peroxisomal vesicles, which can detach from the ER to form peroxisomes [145, 146] [reviewed in [147]]. During peroxisome growth, PEX19, a cytosolic protein, recognizes newly synthesized PMPs via their internal targeting signal motifs and binds them [148]. The complex then docks to PEX3 and PEX16, located in the peroxisomal membrane [149, 150]. In this pathway, PEX19-PEX3-PEX16 mediates import for most PMPs, including metabolite transporters and PEX proteins [151][ reviewed in [152]]. Peroxisome fission requires the integral membrane protein PEX11β and shares several components of its division machinery with mitochondria. These include dynamin-like proteins (DLPs), Fis1, Mff and ganglioside-induced differentiation-associated protein 1 (GDAP1) [153-156]. PEX11β is required for membrane elongation, and binds to Fis1, which recruits DLPs [reviewed in [76]]. PEX11β is constitutively expressed in all tissues, whereas its isoform PEX11α is inducible [157, 158], and PEX11γ is constitutively expressed in liver [159]. PEX11γ is involved in peroxisome

membrane protrusion [160] and may function in the de novo pathway to mediate the import of peroxisomal matrix proteins into the pre-peroxisomal vesicles that are derived from the ER [161]. The relative contribution of the ER-derived versus fission pathway to the total peroxisome population in vivo, and the detailed mechanisms of ER entry and exit of PMPs are controversially discussed. For a detailed review, see [162-164]. Recently, mitochondrial membranes were shown also to contribute to new peroxisome membrane formation. Upon overexpression PEX3 in a PEX3 null cell line, PEX3 and PEX14-containing vesicles were shown to bud from mitochondria and fuse with ER vesicles containing PEX16 to then form a pre-peroxisomal vesicle, followed by PEX19 mediated PMP recruitment [165].

Peroxisome turnover, termed pexophagy, is by general mechanisms related to autophagy of cellular contents, as well as mechanisms selective for peroxisomes [reviewed in [166, 167]]. Deficiencies in various PEX proteins increase pexophagy, including PEX3 [168] and PEX14 [169]. Studies in yeast show that deficiencies in exportomer components Pex1, Pex6, or Pex15 (the yeast ortholog of PEX26) increase pexophagy more than other peroxin deficiencies [170]” This work is corroborated in human cells, highlighting the importance of the ‘exportomer’ complex in pexophagy regulation [171]. The constitutive turnover of peroxisomes is estimated to be around 30% per day [172].

Argyriou, D’Agostino, Braverman, 2016

Figure 1.1: Peroxisome matrix protein import. (1) PTS receptor binding. PEX5 and PEX7 cytosolic receptors bind their cognate ligands (PTS1 and PTS2 enzymes, respectively) in the cytosol. PEX5 has two isoforms that differ by alternative splicing. The longer isoform, PEX5L, binds both PTS1 enzymes and PEX7 and delivers them to the peroxisome membrane. (2) Docking. The receptor-ligand complex docks at the peroxisome membrane by binding PEX13 and PEX14. (3) Matrix enzyme translocation. PEX5, together with PEX14, forms a dynamic membrane pore through which the ligands are transported into the peroxisome matrix. (4) Receptor recycling. PEX2, PEX10, and PEX12 mono-ubiquitinate PEX5, allowing its removal from the membrane. The PEX1-PEX6 AAA-ATPase heterohexamer (anchored to the peroxisome membrane by PEX26) uses the energy from ATP hydrolysis to remove PEX5-Ub from the peroxisome membrane for another round of import. PEX7 is recycled to the cytosol after PEX5 in an ATP independent manner. Note that defects in PEX7 prevent import of PTS2 enzymes, but do not disrupt the PEX5/PTS1 import pathway. This figure is from Argyriou et al. 2016 [1].

1.2.3 Clinical disorders

Pathogenic mutations in any one of 14 PEX genes cause disorders of peroxisome assembly. These are collectively and classically known as peroxisome biogenesis disorders (PBD). Biallelic mutations are typically required for disease, although recently a missense mutation in PEX6 (c.2578C>T [p.Arg860Trp]) was reported to exert a dominant effect when in cis with a frequent 3’ untranslated region (UTR) variant via allelic expression imbalance [173].

“No human disease has yet been associated with defects in PEX11α and PEX11ɣ. Defects in PEX1, PEX2, PEX3, PEX5, PEX6, PEX10, PEX11β, PEX12, PEX13, PEX14, PEX16, PEX19 and PEX26 genes cause Zellweger spectrum disorder (ZSD) including Zellweger syndrome (ZS), Neonatal Adrenoleukodystrophy (NALD), and Infantile Refsum Disease (IRD). These historically distinct phenotypes are now considered different presentations within the same clinical and biochemical spectrum with ZS being the most severe presentation, NALD intermediate and IRD milder [5, 174, 175]. The term ZSD is preferred, given the overlap in individual clinical presentations, and the atypical phenotypes currently being identified by new sequencing technologies. In general, disease severity correlates with patient age at onset of symptoms and the predicted consequence of the mutation on peroxin function, capacity for matrix protein import, residual enzyme functions, and peroxisome numbers. At the cellular level, defects causing ZSD lead to decreased numbers of enlarged peroxisomes that show reduced import of matrix enzymes [176]. These structures are termed peroxisome ‘ghosts’ [176, 177]. Defects in PEX7 cause the distinct phenotype of RCDP1 [178]. Since only PTS2 protein import is disrupted in RCDP1, peroxisome number and morphology are normal and there is deficiency of only a subset of matrix enzymes. A phenotypic spectrum of severity is also present in RCDP1 and correlates to residual plasmalogen levels [53-55]. The overall estimated birth prevalence for ZSD is 1/50, 000 [179] and for RCDP 1/100, 000 [180]. The prevalence of both ZSD and RCDP1 varies among different populations [181-184]. These disorders are autosomal recessive and panethnic, however founder mutations account for the high frequency of the PEX1- c.[2528G>A] allele (p.[Gly843Asp], or p.G843D) [9, 185] and the PEX7-c.[2097_2098insT] allele [186, 187] in individuals of Northern European heritage. Other founder alleles include PEX10-c.[815-815delCT] in the Japanese population [182], PEX6-c.[802_815del] in French Canadians [181], and PEX12- c.[-26G>A; 102A>T] in individuals of Persian Jewish descent [188] (The PEX12-c.[-26G>A] cis mutation is provided by N. Braverman, personal communication).”

1.2.3.1 Zellweger Spectrum Disorder (ZSD)

“Severe ZSD is a classic malformation syndrome described by its characteristic features as cerebro-hepato-renal syndrome [reviewed in [152, 189, 190]]. Patients present as newborns with severe hypotonia, seizures, and characteristic craniofacial dysmorphisms. Brain MRI shows microgyria, pachygyria and heterotopia consequent to neuronal migration defects [reviewed in [191]]. The pattern of medial pachygyria and lateral (perisylvian) polymicrogyria is a distinguishing feature [192]. Eye disease includes cataracts, congenital glaucoma, retinal dystrophy and optic atrophy. Sensorineural deafness is present. The liver is enlarged and hepatic functions show elevated transaminases, cholestasis and coagulopathy. Cortical renal cysts are observed on ultrasound, but are not usually clinically significant. Chondrodysplasia punctata especially in the knees and hips, can be seen on skeletal x-rays. Cardiovascular malformations and pulmonary hypoplasia have also been reported [193]. These infants generally do not survive beyond the first year of life [190].

The intermediate and milder (previously termed NALD and IRD) group of patients present after the newborn period with more varied symptomatology. Hypotonia, failure to thrive, developmental delays, sensory deficits, amelogenesis imperfecta and history of neonatal cholestasis are usually present. The characteristic Zellweger facies is attenuated or absent. Subtle neuronal migration defects have been described in some intermediate patients, but are absent in the milder forms of ZSD [194, 195] [reviewed in [191]]. Overall, this group is clinically distinguished from severe ZSD by the absence of congenital malformations, and is instead a disorder marked by progressive peroxisome dysfunction over time. Most individuals have progressive sensorineural hearing loss and visual loss due to retinal degeneration, leading to deafness and blindness. Other complications include seizures, adrenal insufficiency, leukodystrophy, and osteopenia, the latter leading to pathological fractures [196]. Nephrolithiasis from hyperoxaluria due to deficiency of AGXT has been reported [197]. Liver dysfunction can quiesce, or less frequently, deteriorate [198]. The prediction of which children will develop these complications and when, as well as the timeline of their progression remains unknown. However, children with the intermediate phenotype generally develop more complications at earlier times, almost all develop adrenal insufficiency, and some will develop leukodystrophy. Lifespan is usually shortened, and many of these individuals do not survive past

24 late childhood. Most individuals with milder ZSD have better cognitive functions, less symptoms of disease and can survive through adulthood [198].

Leukodystrophy was classically described in intermediate ZSD patients in late infancy and early childhood. It manifests with active demyelination in the cerebrum, midbrain and cerebellum and consequent psychomotor regression [192, 199-201]. However, leukodystrophy can occur at any age, remain stable, or progress [198, 202]. Limited pathological studies in brain from intermediate patients have shown reduced neuron number and cerebellar atrophy without active demyelination. There can be dysplastic olives and irregular, heterotopic, clumps of Purkinje cells [194, 203]. In the milder forms of ZSD, where patients can survive into adulthood, brain MRI has also revealed white matter abnormalities, although it can be normal. A study including neuroimaging findings in 19 such patients demonstrated high proportion of white matter abnormalities restricted to the cerebellar hilus of the dentate nucleus and/or the peridentate region. Interestingly, at the time of diagnosis, 17 of these patients had a blood metabolite profile typical of a peroxisomal disorder; but at later time points the concentration of many originally accumulating metabolites had declined and, in some patients, even a complete normalization was observed [198]. Histopathological analysis of the brain from an IRD patient showed hypoplasia of the cerebellar granular layer and ectopic location of purkinje cells in the molecular layer without active demyelination [204].

Retinal degeneration is a prominent feature of all ZSD patients and can manifest as early as the first year of life in the intermediate and milder spectrum [205]. Cataracts can also present [190, 206]. Clinical case descriptions have included Leopard spot retinal pigmentation [207], congenital amaurosis [208, 209], Usher syndrome [210, 211], as sensorineural hearing loss is typically also present, and retinitis pigmentosa with a predominant macular dystrophic component [208, 209, 212, 213]. ERG recordings typically show severely reduced or extinguished cone and rod responses [205, 213, 214]. The course and progression of the retinal degeneration is unknown, but in general the progression is slower in the milder phenotypes. Histological reports showed degeneration of photoreceptor outer segments and reduced inner segments, and extensive loss of nerve fiber and ganglion cell layers [215]. Areas of photoreceptor degeneration were associated with pigmentary changes. The retinal pigment epithelium can show patchy hypertrophy, nodular hyperplasia, and atrophy. Pigmented cells can

be found in the subretinal space and in other layers in the retina. Macrophages may contain pigment and degenerated photoreceptors. Lamellar lipid inclusions were observed in most layers of the retina and within the optic nerve by electron microscopy [215]. These inclusions have also been observed in brain, adrenal glands, and liver [204, 207]. Recently, peroxisomes were shown to cluster at the base of photoreceptor outer segments, suggesting that peroxisomes are intimately involved in maintaining outer segments [210]. Cochlear histology has shown degeneration of the sensory epithelium, with preservation of ganglion cells and nerve fibers [204].”

1.2.3.2 Atypical ZSD presentations

“Recently, atypical and milder phenotypes have been reported that may relate to specific mutations in specific PEX genes. These children and adults with normal or near normal intellect, slow disease course and prolonged survival, illustrate that the phenotypic spectrum of these disorders is much broader than initially thought [161, 188, 195, 206, 211, 212, 216-226]. These patients are arbitrarily classified according to major presenting symptomatology: those with prominent sensory deficits (visual and hearing loss), and those with cerebellar dysfunction, peripheral neuropathy or relapsing encephalopathy. These individuals inspire fundamental reflections on PBDs and highlight some of the molecular, biochemical and cellular mechanisms by which these variant phenotypes arise.

The underlying molecular mechanisms of these atypical presentations often involve hypomorphic PEX alleles encoding peroxins with residual protein function. This can include bi-allelic hypomorphic alleles, as well as a mono-allelic, hypomorphic mutation in trans with a severe, usually loss of function mutation [227]. Mutations in specific domains (ring finger) of PEX2, PEX10 and PEX12 can lead to specific phenotypes of cerebellar dysfunction and peripheral neuropathy [discussed in [218]]. In addition, an adult with milder disease was identified with two null mutations in PEX11β. It was suggested that functional redundancy of PEX11α and PEX11ɣ supported the absence of PEX11β [161]. In most of these cases, routine peroxisome metabolite testing in blood (VLCFA, plasmalogen and phytanic acid levels) showed only subtle dysfunction or none at all [161, 195, 216, 217, 227, 228]. In this context, if a PBD is strongly suspected, additional peroxisome functions should be investigated. Pipecolic acid, D/THCA and

pristanic acid levels, as well as fibroblast enzymatic testing, morphological analysis of cellular peroxisomes and catalase distribution has targeted PBD in difficult cases [229, 230].

Morphological analyses of peroxisomes in fibroblast cultures have shown more peroxisomes with residual function, mosaicism, and temperature sensitivity in milder cases, phenomena that likely reflect a partially functional peroxin [227]. Mosaicism refers to the observation of a heterogeneous cell population (by both immunohistochemistry and electron microscopy techniques) in fibroblast cultures, and directly, in hepatocytes from liver cell biopsies, in which some cells contain functional peroxisomes that import matrix proteins, while adjacent cells contain peroxisomes (ghosts) that show minimal or no import of matrix proteins [231-233]. Temperature sensitivity refers to improvement in peroxisome formation and biochemical function in ZSD fibroblasts cultured at 30oC rather than 37oC [234-236]. It has been suggested that temperature sensitivity is a consequence of an imbalance of folding and unfolding kinetics of mutant proteins, resulting in an increase in the correctly folded protein as temperature decreases [237]. The etiology for peroxisome mosaicism is not known, but could reflect microenvironment or cell cycle differences that enable recovery of a partially functional peroxin, or a partially functional import system. Taken together, this data implies residual peroxisome functions in body tissues that are dependent on a threshold of temperature, microenvironment and other unknown factors which can induce further expression. In fibroblast cultures from some patients with atypical phenotypes and normal peroxisomal morphology, increasing the culture temperature to 40oC unmasks the underlying peroxisome dysfunction [188]. Finally, peroxisome morphology can provide a clue to the underlying gene defect. For example, fibroblast cultures from patients with defects in DLP1 or Mff exhibit elongated, constricted peroxisomes that are unable to divide (i.e. “pearls on a string” phenotype) [238, 239]. Fibroblasts from a PEX11β patient have enlarged, elongated peroxisomes, indicating a defect in peroxisome constriction and/or division [161]. Cells from patients with milder variants in PEX16 show a reduced number of enlarged, but import competent peroxisomes [206]. Thus far, peroxisomes in patient cell lines with PEX7 defects (RCDP1) have normal morphology and do not show a temperature response.”

1.2.3.3 Other disorders that affect peroxisome biogenesis

“A few patients were reported with mutations in the shared mitochondrial and peroxisome fission proteins, DLP1 [238], Mff [239], and GDAP1 [156]. These individuals displayed clinical evidence of combined mitochondrial and peroxisome dysfunction. Mitochondrial and peroxisomal morphology indicated fission dysfunction. The patient with DLP1 deficiency presented with affected brain development including an abnormal gyral pattern in both frontal lobes associated with dysmyelination, microcephaly, optic atrophy, and hypoplasia. She died a few weeks after birth. The patients with Mff deficiency showed delayed psychomotor development, microcephaly, pale optic discs, and mild hypertonia. Brain MRI revealed an abnormality in the globus pallidus. The eldest of the 2 patients (brothers) was 4.5 years old at the time of publication. GDAP1 mutations are known to affect peroxisome and mitochondrial dynamics and cause the peripheral neuropathy Charcot-Marie-Tooth disease (CMT) [156, 240]. Interestingly, peroxisome metabolic functions (as measured by biomarkers indicative of classic peroxisomal disorders) were not altered in the Mff patient. The DLP1 patient showed only a slight elevation of VLCFA. This minimal or nonexistent biochemical evidence of peroxisomal dysfunction highlights the importance of alternate methods in patient diagnosis (discussed previously). It is likely that defects in other proteins involved in peroxisome assembly and maintenance remain to be identified that can result in a clinical disease.

The increased use of next-generation sequencing technologies has led to the identification of many of these atypical patients [161, 188, 195, 206, 212, 216, 218, 219, 223-226] and will continue to do so. However, analysis of peroxisome biochemical and cellular functions remains necessary to prove a peroxisome etiology, especially in cases where there are novel PEX gene variations, novel phenotypes, and defects identified in other pathways involved in peroxisome assembly and maintenance. All this will lead an improved understanding of peroxisome biology in various tissues over time and thus enable more precise genotype and phenotype correlations.”

1.2.4 Clinical diagnosis

“Clinical suspicion of PBD requires confirmation by determining peroxisome functions. The fundamental problem in ZSD is the inability to generate and maintain functional peroxisomes [5, 6]. Consequently, the typical biochemical signature of ZSD is concomitant dysfunction of

multiple peroxisomal pathways. This is detected in blood, urine and cells by accumulation of VLCFA, phytanic, pristanic and pipecolic acids, D/THCA and deficiency of plasmlaogens, docosahexanoic and primary bile acids (cholic and chenodeoxycholic acids) [diagnosis reviewed in [241]]. Metabolite testing is usually followed by enzymatic confirmation of PBD in fibroblast cultures and/or molecular confirmation of the PEX gene defect.

With the advent of next generation sequencing, molecular diagnosis has become cost efficient, comprehensive, and has enabled additional genotype-phenotype correlations in atypical patients. Over the past decade, studies on more than 800 patients show that defects in PEX1 represent around 60% of all ZSD alleles [9, 12, 13]. In these reports, 12–30% of ZSD alleles are PEX1-p.G843D, and 5–20% are PEX1-p.[Ile700Tyrfs42*]. About 50% of all RCDP alleles are PEX7-p.[Leu292*]. The high frequency of these alleles is due to a founder effect in persons of northern European extraction [55, 185]. PEX1-p.G843D is a missense protein with residual functions; the presence of at least one allele predicts an intermediate phenotype, and homozygosity generally results in milder phenotypes [10, 11, 234, 242]. The large variations in phenotype severity observed amongst PEX1-p.G843D homozygotes (unrelated, as well as intrafamilial) suggest the influence of modifier genes [198, 212].

PEX1-p.[Ile700Tyrfs42*] and PEX7-p.[Leu292*], are null alleles and predict a severe phenotype when homozygous [55, 187, 243]. Overall, bi-allelic mutations in PEX1, 6, 26, 10, 12, and 2 contribute to the majority of ZSD alleles [9]. When previously unknown or uncharacterized variants are identified, functional biochemical studies remain a necessary tool to determine causality and pathogenicity [reviewed in [152, 230]].”

1.2.5 Disease management

“There are no curative therapies for PBD. Management is multidisciplinary, symptomatic, and based on surveillance of the multiple systems involved. This includes regular developmental, neurological, auditory, visual, nutritional, and orthopedic assessment, and monitoring of liver and adrenal functions [241, 244]. Vitamin K and fat soluble vitamin supplements are usually prescribed. Care is often palliative for those with severe disease.

Dietary interventions have included restriction of metabolites that accumulate and replacement of those that are deficient. Reports of these approaches in patients are mostly anecdotal and have not been systematically studied. Furthermore, the extensive inter- and intrafamilial variations in the intermediate and milder ZSD patients requires better understanding of the natural history of these disorders and the establishment of reliable clinical endpoints for future interventional trials.”

Patients have also received DHA supplementation, with controversial clinical benefit. In a small case series, improvements in electroretinograms (ERG) and growth were shown [214, 245]. In a double-blind randomized placebo controlled trial that included 48 patients over one year of study, there were no overall improvements in ERG and growth in the treated group compared to the placebo group [214]. Due to the benefit of dietary phytanic acid restriction in adult Refsum disease [246], and the fact that it is not a difficult diet regimen, this is often prescribed for surviving PBD patients but formal studies have not been done to determine if this improves the disease course.”

Administering the oral plasmalogen precursor batyl alcohol recovered erythrocyte plasmalogens in a small case series of ZSD patients, though clinical outcome did not change [247]. Oral bile acid supplementation with chenodeoxycholic and ursodeoxycholic acids in two infants with severe ZSD reduced the C27 bile acid intermediates DHCA and THCA and improved hepatobiliary function, but did not change the clinical course [248, 249]. Cholic acid supplementation in a third infant with severe ZSD improved histology with a decrease in bile duct proliferation and inflammation [248]. A larger study of cholic acid supplementation in 17 ZSD patients without severe liver fibrosis or cirrhosis found reduced C27 bile acid intermediates after 21 months of treatment, with no other clinical improvement (i.e. liver tests, elasticity, coagulation parameters, vitamin levels, or body weight). The levels of C27 bile acids intermediates increased upon discontinuing treatment [250].

In two siblings with an intermediate-milder ZSD phenotype, the older child received an hepatocyte transplantation [251] and the younger received a liver transplant [252]. Peroxisome biochemical functions improved, but clinical outcome of the younger sibling was relatively better, with sustained biochemical improvement, stable hearing and vision, and better neurodevelopmental and psychomotor status over 17 years [253]. Another patient received a

living donor liver transplant, and biochemical functions improved, but long-term outcome remains to be seen [254].

1.3 MOUSE MODELS OF PBD-ZSD

“In order to gain further insight into the pathogenesis of ZSD, mouse models with complete deficiency of Pex2, Pex5, Pex13, and Pex11β have been engineered [255]. These mice recapitulate many aspects of severe ZSD, including the metabolic consequences of peroxisome elimination and abnormalities in brain formation, hypotonia, and early demise. Detailed investigations of these models revealed neuronal migration defects commencing early in the cytogenetic epoch and demonstrated by birth dating experiments using 5-BrdU. Other abnormalities include delayed neuron differentiation and extensive neuronal apoptosis (Pex5, Pex11β). In the cerebellum, there was delayed neuronal layering, increased apoptosis in granular neurons and abnormal morphology of Purkinje cells (Pex2) [256]. In Pex11β null mice, there was reduced numbers of importing peroxisomes and only mild deficiencies of β- oxidation and plasmalogen biosynthesis, challenging the idea that PBD clinical features are directly related to the severity of the metabolic abnormalities [257, 258]. Overall, the null models survive gestation, but die soon thereafter, indicating that severe peroxisome deficiency is compatible with embryonic, but not postnatal life in both mice and humans.

To gain further information, conditional PEX5 null mice were engineered to evaluate peroxisome dysfunction in key organs. Targeted elimination of peroxisomes in neural precursor cells results in moderate cortical migration defects that correct postnatally and might reflect migration delay [259]. Peroxisome elimination in hepatocytes results in postnatal arrest of neuron migration, particularly in the cerebellum, which still develops in the postnatal period [259]. Thus, peroxisome metabolism in the liver influences the development of the murine brain. It was suggested that elevated C27 bile acids, which are more hydrophobic than the terminal conjugated forms, may enter the brain and impair cerebellar development [259]. Selective inactivation of Pex5 in neural cells, although causing mild neurodevelopmental defects, resulted in a progressive motor and cognitive impairment with death before 6 months of age [260]. Myelinated axons and glial cells were severely affected in the brain and spinal cord, showing

decreased compaction of myelin, loss of myelin and axonal damage. Mice with Pex5 inactivation in oligodendrocytes also developed similar axonal damage and demyelination [261]. Thus, peroxisomes may not be required for myelination, but for the preservation of axonal integrity and maintainence of myelin, and this role may be specific to oligodendrocyte peroxisomes [261].

Considering the interrelationship in mitochondrial and peroxisome functions, and the mitochondrial abnormalities noted in tissues from severe ZSD patients [5], mitochondrial abnormalities were investigated in the liver selective PEX5 null mouse model [262]. In this model, morphologically abnormal mitochondria with decreased respiratory chain complex activity were observed in liver. However, oxidative damage was not observed, possibly due to a compensatory increase in glycolysis and mitochondrial proliferation [262]. The decreased gluconeogenesis and glycogen synthesis resulted in the requirement for additional carbohydrates, increased food intake and lower body weights [263].

A PEX1-p.G843D equivalent knock-in mouse model (PEX1-p.G844D) was recently engineered to mimic the common human mutation, and to generate a model for the intermediate and milder phenoytpes that feature progressive peroxisome dysfunction over time. PEX1-p.G844D homozygous mice exhibit many of the typical ZSD manifestations, including growth retardation, progressive retinopathy, bile acid defects resulting in fatty liver with cholestasis, elevated VLCFA, and decreased plasmalogens [264].” Further characterization of the course and mechanisms for disease pathophysiology are currently ongoing in our laboratory and by other colleagues.

“The PBD mouse models have also been used to test potential therapeutic interventions. In the Pex5 null models, supplementation with DHA, beginning in utero, normalizes DHA levels in brain but does not improve the neuronal migration disorder [265]. Bile acid supplementation in Pex2 null mice improves cerebellar abnormalities, implicating the accumulation of C27 bile acid intermediates in the cerebellar toxicity observed postnatally in these mice [266].”

1.4 THE MOUSE VISUAL SYSTEM AND ITS ASSESSMENT

Chapters III and IV of this thesis focus on i) the characterization of retinopathy in the Pex1- G844D mouse model, and ii) using retinal gene therapy to improve visual function in this model. An introduction to the visual system and its study in the mouse is thus warranted, with focus on the retina.

1.4.1 The mouse visual pathway

The mouse eye is anatomically similar to that of most other vertebrates, comprised of a cornea and lens that refract light to form an image on the retina, and an iris that controls pupillary diameter, and consequently the amount of light reaching the posterior eye segment [267, 268] (Figure 1.2A). In mammals, vision depends on the detection of light, the first step of which occurs at the outer retina. The visual cascade is initiated when photons trigger chemical changes in the melanopsin pigments of the photoreceptor outer segment [268]. This signal is transmitted via the outer plexiform layer (OPL), which is comprised of photoreceptor and bipolar cell synapses, to the inner nuclear layer (INL), comprised of bipolar, amacrine, horizontal, and Müller cell nuclei. These cells extend to the inner plexiform layer (IPL, a second synaptic layer), where they connect to the retinal ganglion cells. The ganglion cell axons will form the optic nerve, which transmits retinal output to the brain. Above the photoreceptors is the retinal pigment epithelium (RPE), which contains melanin pigment granules that absorb residual photons [269]. In addition, the RPE is the site of photopigment regeneration, performed via the visual cycle between the RPE and photoreceptors [270, 271]. For a schematic representation of retinal architecture, see Figure 1.2B.

Two types of photoreceptors, rods, and cones, mediate scotopic (dim light) and photopic (bright light) vision, respectively. Like that of most mammals, the mouse retina is rod-dominated with rods comprising approximately 97% of mouse photoreceptors, and cones the remaining 3% [268]. Photoreceptor morphology is comprised of three components: the outer segment (OS), inner segment (IS), and nucleus. The OS contains the light-sensing rhodopsin in rods, and L-, M- , and S-opsin pigments in cones, that absorb and convert light signal [272]. The diversity of cone opsins, each absorbing light of distinct wavelengths, allows for colour perception. Connecting the OS and nucleus is the IS, a mitochondria-rich, structure that provides the metabolic support needed for constant OS regeneration. The IS is also involved in regulating membrane potential [273]. The photoreceptor nuclei comprise the retinal outer nuclear layer (ONL). Bipolar cells

transform the input received from photoreceptors to an output conveying detailed properties of the visual stimulus, and transmit this signal to retinal ganglion cells [274]. The horizontal and amacrine cells are retinal interneurons that modulate signal from the photoreceptors to bipolar cells, and from the bipolar to retinal ganglion cells, respectively. The Müller cells span the entire neural retina as glial support [275].

Peirson et al., 2017

Figure 1.2. The mouse eye and retina. A) The mouse eye is similar in structure to that of most other vertebrates B) The retina is a layered structure and light must pass through the inner retinal layers to reach the light sensitive photoreceptors in the outer retina. The retina contains two classes of visual photoreceptors: rods which mediate low light (scotopic) vision, and cones which mediate bright light (photopic) vision and provide colour vision. In addition to the rods and cones, a subset of melanopsin-expressing photosensitive retinal ganglion cells (pRGCs) have recently been identified, mediating many non-visual responses to light. This figure and legend are adapted from Peirson et al. 2017 [267].

1.4.2 Assessing visual function in the laboratory mouse

Although one may consider nocturnal rodents poor subjects for vision research, the study of vision in mice has been invaluable for understanding both basic neuroscience and the pathophysiology of human disease. Indeed, several methods similar to those used in humans have been established to reliably measure visual function in rodents, including fundus examination, electrophysiology, optokinetic reflex, and pupillary reflex [276]. Behavioral visual assessments include maze-based tests, visually cued conditioning, visual placing, and light- induced shifts in circadian phase. Below are summarized the electrophysiological and behavioural tests included in this thesis.

1.4.2.1 Electrophysiology

Electrophysiology provides an objective way to measure neurological function in the form of an evoked potential. This technology is widely used for the characterization and diagnosis of both human pathology in the clinic, and of animals in an experimental setting.

The flash electroretinogram (ffERG)

The full-field flash electroretinogram, or ffERG, is an evoked potential that represents the pan- retinal response to a light stimulus. A low-mass, silver-coated, conductive ﬁber electrode (Dawson, Trick, and Litzkow, DTL electrode) placed across the eye can be used to detect the electrical response [277]. A schematic representation of the resulting waveform and the cells contributing to each component are shown in Figure 1.3. Upon applying a light stimulus, the first component that appears is a negative deflection called the ‘a-wave’, which results from the hyperpolarization of photoreceptors [278, 279]. Although photoreceptors are the major contributing cells, there is evidence that horizontal and/or bipolar cell response (i.e. the inner retina) has a small contribution to a-wave generation [280]. Following the ‘a-wave’ is a positive deflection called the ‘b-wave’. This is the most prominent ffERG component, and results from the transfer of signal through the photoreceptor axon to the inner retina, inducing depolarization of inner retinal cells [281]. Bipolar cell, and to a much smaller extent Müller cell, responses generate the b-wave [282, 283]. Oscillatory potentials (OPs) are low amplitude, high frequency waves occurring within the electroretinogram, typically on the ascending limb of the b-wave.

Although their exact source is still debated, OPs originate in the inner retina and are considered the result of negative feedback reactions involving amacrine and bipolar cells [284].

When performed under dim light (scotopic) or high light (photopic) conditions, the ffERG can distinguish between rod and cone mediated retinal response, respectively. There is no a-wave present in the murine photopic ffERG waveform, which is composed of only a b-wave.

Figure 1.3. Representation of a scotopic ffERG waveform. Upon flash stimulus, a negative deflection (a-wave) occurs, caused by photoreceptor hyperpolarization. As the light-induced signal is transduced through the retina, a positive deflection (b-wave) occurs, caused by depolarization of the inner retinal cells. Within the b-wave are oscillatory potentials, the result of negative feedback of bipolar to amacrine cells. The amplitude of the a- or b-waves corresponds to the intensity of retinal response. This figure is adapted from Rossmiller et al. 2015 [285].

Visually evoked potential (VEP)

Although the ffERG is considered the ‘gold standard’ for measuring mouse retinal function, it does not provide information about signal transduction from the eye to the visual cortex. The visually evoked potential (VEP), is determined from an electroencephalogram (EEG) signal in response to light stimulus. As such, the cells generating this electrical potential are in the brain, specifically the superior colliculus [286]. In mice, the VEP can be recorded using a subdermal

electrode placed at the base of the skull, terminating above the first vertebrae. The amplitude of the resulting wave corresponds to visual cortex activation in response to light stimulus. A diminished VEP amplitude or delayed peak time in the presence of a normal ffERG indicates a post-retinal defect in signal transduction.

1.4.2.2 Visual acuity

Although visual acuity in mice is approximately ten-fold lower than that of humans [287], it is nonetheless possible to reliably measure this parameter in the rodent. One method to measure visual acuity is an optomotor response test. This system exploits the reflexive head movements that occur when a visual stimulus moves in an animal’s visual field [288]. Freely moving mice are placed on an elevated platform inside a chamber and exposed to rotating vertical sine wave gratings on surrounding digital monitors. When able to perceive the stimulus, the mouse will normally stop moving its body and begin to track the grating with reflexive head movements in concert with the rotation. The head movements are detected by a video camera overlying the platform. The spatial frequency (width) of the grating is gradually decreased until the mouse no longer exhibits a tracking behavior. The highest spatial frequency that can be followed (i.e. spatial frequency threshold) determines the visual acuity. This technique yields independent measures of right- and left-eye acuity, as only motion in the temporal-to-nasal direction evokes a tracking response [289]. Images of the system are shown in Figure 1.4.

Figure 1. 4. The virtual optomotor system (OptoMotry, CerebralMechanics Inc.) for determining rodent visual acuity. (A) Schematic representation of mouse virtual chamber from manufacturer (Cerebral Mechanics). The mouse is positioned on a platform as vertical gratings rotate on surrounding digital monitors. Perception of the visual stimulus results in a reflexive tracking motion in concert with the rotation. The thickness of perceived gratings determines ‘spatial frequency threshold’, or visual acuity. (B) Photos of our OptoMotry chamber and output monitor.

1.5 HYPOTHESES AND OBJECTIVES

Our research group and others previously discovered that PEX1-G843D activity can be recovered by small molecule therapy [290, 291], including two related flavonoid compounds. Flavonoid compounds were shown to bind at or near the ATP-binding pocket in ABC transporters and protein kinases [292-294]. Thus, we hypothesized that they may function as pharmacological chaperones for PEX1, a AAA-ATPase, by binding at or near its ATP binding site. We expanded our investigation of flavonoid compounds through structure-activity relationship and dose response analysis, and identified diosmetin as best recovering peroxisome import. I report here on the optimization of diosmetin and other small molecule compounds to improve peroxisome functions in PEX1-G843D patient cells. I hypothesized that recovering peroxisome import would translate to recovery of PEX1-G843D and downstream peroxisome functions, and confirmed this by studying (i) PEX5 localization by immunofluorescent microscopy, and (ii) peroxisome biochemical metabolite levels by LC-MS/MS after treatment. As several studies suggest that PEX1-G843D is a misfolded, degraded protein [11, 290], I then hypothesized that if diosmetin is indeed chaperoning PEX1-G843D, this should result in increased ‘exportomer’ complex stability, and thus increased levels of PEX1-G843D and PEX6, and tested this using immunoblotting. The results of these experiments are described in Chapter II.

To test potential therapies in vivo, I used the Pex1-G844D mouse model for mild ZSD, which bears the murine equivalent to the human mutation. To establish endpoints for preclinical studies, it was necessary to first characterize the mouse phenotype. In this thesis I report on the characterization of retinopathy in this model, in preparation for proof-of-concept studies for vision improvement in ZSD. I hypothesized that the visual defect in Pex1-G844D mice is at the retinal level, and like many retinopathy models, the mice would exhibit a progressive deterioration of visual function due to retinal cell loss and thinning, and tested this using electroretinograms, retinal histology, and immunohistochemistry in mice from 2-32 weeks of age. I also hypothesized that signal transduction from the retina to visual cortex would be preserved, and tested this using visual evoked potential electrophysiological measures at young and old ages. Finally, I hypothesized that, like in human PEX1-G843D fibroblasts, Pex1-G844D protein levels in the mouse retina would be reduced, as would plasmalogen levels, and that

VLCFA levels would be elevated. I tested this hypothesis using immunoblotting, immunohistochemistry, and LC-MS/MS lipid analysis. The results of these experiments are described in Chapter III.

As ZSD results from monogenic mutations in a single PEX gene, it is an ideal candidate for gene replacement therapy, which uses a viral vector to deliver a functional gene to affected cells. To test whether retinal gene therapy can improve vision in ZSD, I used the Pex1-G844D mouse model in a proof-of-concept study. I hypothesized that rAAV-mediated subretinal delivery of a functional human PEX1 gene would improve retinal response by preventing or slowing functional deterioration. I tested this by first confirming that human PEX1 rescues peroxisome import in mouse cells, and that the therapeutic vector rescues peroxisome import in PEX1 null HepG2 cells. I then followed the visual function of two Pex1-G844D mouse cohorts for 5-6 months after receiving retinal gene augmentation at 5 or 9 weeks of age. The results of these experiments are described in Chapter IV.

CHAPTER II: Zellweger spectrum disorder patient-derived fibroblasts with the PEX1-Gly843Asp allele recover peroxisome functions in response to flavonoids

+Gillian MacLean 1, +Catherine Arygriou 1, Erminia Di Pietro2, Xuting Sun3, Sara Birjandian3, Panteha Saberian3, Joseph G. Hacia4, *Nancy Braverman1,2

1) Department of Human Genetics, McGill University, Montreal, Quebec, Canada 2) Department of Pediatrics and Research Institute of the McGill University Health Center, Montreal, Quebec, Canada 3) Department of Biotechnology, McGill University, Montreal, Quebec, Canada 4) Department of Biochemistry and Molecular Medicine, Keck School of Medicine, Los Angeles, California, USA

+These authors contributed equally

Accepted for publication by the Journal of Cellular Biochemistry, DOI 10.1002/jcb.27591

2.1 ABSTRACT

Zellweger spectrum disorder (ZSD) results from biallelic mutations in PEX genes required for peroxisome biogenesis. PEX1-G843D is a common hypomorphic allele in the patient population that is associated with milder disease. In prior work using a PEX1-G843D/null patient fibroblast line expressing a GFP reporter with a peroxisome-targeting signal (GFP-PTS1), we demonstrated that treatments with the chemical chaperone betaine and flavonoid acacetin diacetate recovered peroxisome functions. To identify more effective compounds for preclinical investigation, we evaluated 54 flavonoids using this cell-based phenotype assay. Diosmetin showed the most promising combination of potency and efficacy (EC50 2.5 µM). All active 5,7- dihydroxyflavones showed greater average efficacy than their corresponding flavonols, while the corresponding flavanones, isoflavones and chalcones tested were inactive. Additional treatment with the proteostasis regulator bortezomib increased the percentage of import-rescued cells over treatment with flavonoids alone. Co-treatments of diosmetin and betaine showed the most robust additive effects, as confirmed by three independent functional assays in primary PEX1-G843D patient cells, but neither agent was active alone or in combination in patient cells homozygous for the PEX1 c.2097_2098insT null allele. Moreover, diosmetin treatment increased PEX1, PEX6 and PEX5 protein levels in PEX1-G843D patient cells, but none of these proteins increased in PEX1 null cells. We propose that diosmetin acts as a pharmacological chaperone that improves the stability, conformation, and functions of PEX1/PEX6 exportomer complexes required for peroxisome assembly. We suggest that diosmetin, in clinical use for chronic venous disease, and related flavonoids warrant further preclinical investigation for the treatment of PEX1-G843D-associated ZSD.

2.2 INTRODUCTION

Peroxisome biogenesis disorders of the Zellweger spectrum (PBD-ZSD) are a heterogeneous group of autosomal recessive disorders caused by defects in any one of 13 PEX proteins, encoded by PEX genes, which are required for peroxisome biogenesis [1, 9]. Mutations in PEX genes lead to the disruption of peroxisome assembly and, consequently, peroxisomal biochemical pathways (reviewed in [14]). Patients with ZSD show a spectrum of clinical severity from severe to milder (previously called Zellweger Syndrome (ZS), Neonatal Adrenoleukodystrophy (NALD), and Infantile Refsum Disease (IRD), respectively) [241]. The level of residual PEX protein activity and consequent peroxisome function is associated with phenotypic severity. One common hypomorphic allele, PEX1 p.[Gly843Asp] or G843D (c.2528G>A), has high residual activity and confers a milder ZSD phenotype [10, 11]. This is a founder allele in persons of Northern European ancestry, and represents up to 30% of all ZSD alleles [9, 12, 13]. Thus, identifying treatments for this hypomorphic allele would benefit a significant number of ZSD patients, and potentially prevent or ameliorate disease progression.

The PEX1 and PEX6 proteins are members of the AAA ATPase family (ATPases associated with various cellular activities) that form part of the PEX1-PEX6-PEX26 exportomer complex required for peroxisome assembly in mammalian cells. Cryo-electron microscopy studies in yeast have shown that Pex1 and Pex6 can form a unique heterohexameric double-ring structure with alternating Pex1 and Pex6 protein subunits [132-134]. In mammalian cells, PEX1-PEX6 complexes are anchored at the cytosolic surface of the peroxisome membrane via interactions between the N-terminal domain of PEX6 and PEX26, an integral peroxisome membrane protein [295]. In an ATP-driven process, the PEX1-PEX6 complex ‘pulls’ the PEX5 receptor out of the peroxisome so that it can be recycled for additional rounds of import [reviewed in [296]]. If not recycled, PEX5 is targeted for proteasomal degradation [113]. There is evidence that the PEX1- PEX6 complex also regulates pexophagy [171]. Overall, defects in PEX1 result in abnormal peroxisome assembly, structure, abundance, and downstream functions.

Although PEX1 transcript levels in PEX1-G843D homozygous patient fibroblast lines are reported to be 50-100% of normal [297] , PEX1-G843D protein amounts are reduced to 5-15% of normal PEX1 levels [11]. PEX6, its partner protein, and PEX5 are also reduced [11, 290]. Several studies on PEX1-G843D patient fibroblasts show that peroxisomal matrix protein import improves when cells are grown at 30°C [11], in cells where PEX6 is overexpressed [298], or when cells are treated with chemical chaperones [290, 291]. Taken together with the clinical phenotype, these observations suggest that PEX1-G843D is a misfolded protein with residual activity and a reduced half-life, and amenable to chaperone therapy. Compounds with chaperone activity include chemical chaperones (e.g. glycerol, DMSO, trimethylamines) that non- selectively stabilize misfolded proteins through weak thermodynamic interactions, thus requiring high concentrations to be effective [reviewed in [299]]. In contrast, pharmacological chaperones selectively bind target proteins and are thus effective at much lower concentrations. These include enzyme substrates, inhibitors, or mimics that bind and facilitate folding of non-native protein intermediates to their native state [reviewed in[300]].

To identify small molecule compounds that recover peroxisome assembly, we previously designed a high content cell-based phenotype assay using a transformed and immortalized PEX1-G843D/null compound heterozygous ZSD patient fibroblast line expressing a GFP-PTS1 (peroxisome targeting signal 1) reporter protein. The reporter, cytosolic at baseline, redistributes to punctate structures that co-localize with peroxisomes following effective treatment, correlating to functional recovery of peroxisomal matrix protein import. Using this assay, we identified two related flavonoids and protein kinase C inhibitors that improved peroxisome matrix protein import [290]. Compounds from these classes were shown to bind at or near the ATP-binding pocket in ABC transporters and protein kinases [292-294]. Thus, we hypothesized that they may function as pharmacological chaperones for PEX1. In this report, we expanded our investigation of flavonoids, naturally occurring compounds found in edible plants, to identify lead compounds for further optimization. We also tested combination therapies of flavonoids with non-specific chemical chaperones or a proteostasis regulator to rescue peroxisome assembly in PEX1-G843D ZSD patient cells. Finally, we validated our primary results using independent assays of peroxisome activities in primary ZSD patient cells: (i) examination of endogenous enzyme import, (ii) PEX5 localization, (iii) biochemical functions, (iv) and protein amounts.

2.3 MATERIALS AND METHODS

2.3.1 Cell lines and culture conditions

Fibroblasts under 20 passages (Kennedy Krieger Institute, Baltimore, MD, USA), consented for

research through the Institutional Review Board, were cultured at 37ºC, 5% or 7% CO2 in DMEM (Gibco) with 10% (vol/vol) FBS (Wisent Bioproducts). Primary fibroblasts used were: 3 cell lines (PEX1-643, PBD214, PBD664) with alleles c.[2528G>A];[2528G>A], PEX1 p.[G843D]; 2 cell lines (PEX1-900, CA1) with alleles c.[2528G>A];[2097_2098insT], PEX1 p.[G843D];[0]; 1 cell line (CA3) with alleles c.[2528G>A];[382C>T], PEX1 p.[G843D];[0]; 2 cell lines (PBD602, PBD651) with alleles c.[2097_2098insT]; [2097_2098insT], PEX1 p.[0];[0]; and 2 in-house controls. Transformed immortalized cells used were: PEX1-G843D-GFP-PTS1 [290] with alleles c.[2528G>A]; [2097_2098insT], PEX1 p.[G843D];[0] expressing pEGFP- PTS1 and control GM09503 (healthy 10 year old male; Coriell Biorepository, Camden, New Jersey).

2.3.2 Chemicals

Flavonoids were purchased from Apin Chemicals Limited (Oxon, United Kingdom), Extrasynthese (Genay, France), Indofine Chemical Company, Inc. (Hillsborough, NJ), Sigma- Aldrich (St. Louis, MO), and Zamboni Chem Solutions Inc. (Montreal, Quebec, Canada) (Supplementary Table 1), and dissolved in 100% DMSO as 10 mM stock solutions. Bortezomib (Selleck Chemicals (Houston, TX), S1013) was dissolved in 60% DMSO at 13 mM, diluted in water to 2 μM stock solution (DMSO<0.1%) and filter (0.2µm) sterilized. Anhydrous betaine (Orphan Europe, Puteaux - France) was dissolved in water to 5 mM stock solution and filter (0.2 µm) sterilized. Stock solutions were stored at -20ºC.

2.3.3 Flavonoid screen

PEX1-G843D-GFP-PTS1 cells were seeded at 2200 cells/well in 100 µl in 96-well plates. Flavonoid stocks were diluted to 20 µM in culture medium. 5 hours after seeding, 100 µL of each flavonoid preparation was added per well, to 10 µM final. Each treatment was done in triplicate. Negative controls were 0.002% (vol/vol) DMSO or culture medium only; positive control was 100 mM betaine. The plates were cultured for 48 hours, to 60-80% confluence. A second experiment was done at 20 µM flavonoid, except 5 µM acacetin, and cells cultured 4

days. Following treatment, cells were washed in PBS, fixed in 3% formaldehyde. Plates were examined using a Leica DM IRB inverted microscope at 40X magnification. At least 100 importing and non-importing cells in each well were counted by scanning fields of view from the top left to bottom right. Cells were classified in a binary manner as either “importing” or “non- importing”, which is not indicative of the proportion of import competent peroxisomes per cell. Positive cells were defined as having >15 punctate structures (importing peroxisomes) to exclude any cells with chance aggregation of GFP [290]. Fluorescent images were taken using a QImaging (Surrey, BC, Canada) Retiga 2000R camera and Image-Pro Plus 6.1 software (Media Cybernetics, Rockville, MD). In blinded dose response and combination drug experiments, we seeded 500 cells/well and treated for 2-5 days.

2.3.4 Immunoblotting

Fibroblasts in T25 or T75 flasks at approximately 40% confluence were treated with 10 μM diosmetin, 50 or 100 mM betaine, 1.5 nM bortezomib, alone or in combination for 2 or 6 days (medium changed every 2 days). Cells were trypsinized and lysates were prepared in IGEPAL lysis buffer. Lysates (20 μg) were separated on 12% or 7.5% polyacrylamide gels and transferred to nitrocellulose membranes. Membranes were blocked and hybridized in 5% milk with primary antibodies: mouse anti-human PHYH, rabbit anti-human ACAA1 (thiolase) [290], 1:1000 rabbit anti-human PEX1 (Proteintech (Rosemont, IL) 13669-1-AP), 1:2000 rabbit anti-human PEX5, PEX6 (gifts from Gabriele Dodt, University of Tübingen), 1:2000 rabbit anti-human ABCD3 (gift from Steven Gould, Johns Hopkins University), 1:17000 rabbit anti-human β-tubulin (Abcam (Cambridge, MA) ab6046), followed by appropriate HRP-conjugated secondary antibodies, and visualized by ECL. Band quantification was done using BioRad (Hercules, California) Image Lab 4.0.1 or ImageJ (NIH).

2.3.5 PEX5 localization

Primary fibroblasts were seeded onto coverslips in 12-well plates and received either: culture medium, 100 mM betaine, 10 µM diosmetin, or 50 mM betaine + 10 µM diosmetin for 48 hours. Cells were prepared for indirect immunofluorescence as described [301]. Primary antibodies: 1:400 rabbit anti-human PEX5 (gift from Gabriele Dodt, University of Tübingen); 1:400 sheep anti-human ABCD3 (gift from Steven Gould, Johns Hopkins University); secondary antibodies: 1:400 anti-rabbit 488 (Invitrogen), 1:300 anti-sheep Texas Red (Sigma-Aldrich). Slides were

visualized using an Olympus (Center Valley, PA) BX51 microscope at 60X magnification; images were captured using an Olympus CCD camera and MagnaFire software. Cells with ‘peroxisomal’ localization of PEX5 were defined as having PEX5 co-localized with ≥ 10 peroxisomes, and ‘cytosolic’ localization as having PEX5 co-localized with < 10 peroxisomes.

2.3.6 Lipid analyses

Primary fibroblasts in T25 flasks were exposed to culture medium, 100 mM betaine, 10 µM diosmetin, or 50 mM betaine + 10 µM diosmetin; medium was replaced every 2 days. After 10 days, cells were trypsinized, washed and pelleted twice; cell pellets were flash frozen and stored at -80°C. For liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis, cell pellets were homogenized in PBS using a mini pestle. An extraction solution of methanol containing 10 ng of each internal standard, 16:0-D4 lyso-PAF (20.6 pmol) and D4-26:0-lyso-PC (15.6 pmol) was added to 50 µg protein cell extract in a glass tube and incubated shaking at room temperature for 1 hour. Samples were filtered by centrifugation (Corning (Corning, NY) Costar spin-X centrifuge tube filters) for 5 minutes. Filtrates were analyzed in Verex auto-sampler vials (Phenomenex, Torrance, CA). A 2.1 X 50 mm, 1.7 µm chromatography column and a Waters (Milford, MA) TQD (Triple Quadrupole Mass Spectrometer) interfaced with an Acquity UPLC (ultra-performance liquid chromatography) was used in positive ion electrospray (ESI)-MS/MS ionization. The solvent systems were: mobile phase A = 54.5% water/45% acetonitrile/0.5% formic acid, mobile phase B = 99.5% acetonitrile/0.5% formic acid with both solutions containing 2 mM ammonium formate. Injections of extracts dissolved in methanol were made with initial solvent conditions of 85% mobile phase A/15% mobile phase B. The gradient employed was from 15% to 100% mobile phase B over a period of 2.5 min, held at 100% mobile phase B for 1.5 min before reconditioning the column back to 85% mobile phase A/ 15% mobile phase B for 1 min at a solvent rate of 0.7 ml/min. A column temperature of 35oC and an injection volume of 5µl for plasmalogen and 10 µl for lysoPC were used for analysis. Ethanolamine plasmalogens were detected by multiple reaction monitoring (MRM) transitions representing fragmentation of [M+H]+ species to m/z 311, 339, 361, 385, 389, 390 for compounds with 16:1, 18:1, 20:4. 22:6, 22:4 and 18:0, at the sn-2 position, respectively. Lysophosphatidylcholine (LysoPC) species were detected by multiple reaction monitoring (MRM) transitions representing fragmentation of [M+H]+ species to m/z 104. Reagents used

were authentic plasmalogen standards, tetradeuterated internal standards 26:0-D4 lysoPC (Avanti Polar Lipids, Alabaster, Alabama), 16:0-D4 lyso PAF (Cayman Chemical Company, Ann Arbor, Michigan) and HPLC grade solvents (methanol, acetonitrile, chloroform, water) (Fisher Scientific, Waltham, MA), formic acid (Sigma-Aldrich), and PBS (Thermo Fisher Scientific, Waltham, MA).

2.3.7 Protein bioinformatics

Human PEX1 and VCP protein sequences were aligned using Accelrys Discovery Studio 2.5 multiple sequence alignment. A ligand interaction diagram was produced (Accelrys (San Diego, CA) Discovery Studio 2.5) using the crystal structure of the VCP D2 subdomain with ADP- bound (PDB ID: 3CF0) [302].

2.4 RESULTS

2.4.1 Recovery of peroxisome matrix protein import by flavonoids

Flavonoids are a large family of naturally occurring polyphenolic compounds that can be divided into subclasses that include flavones, flavonols, flavanones, isoflavones, and chalcones (Figure 2.1) [293, 303]. Previously, we reported functional recovery of peroxisomal import in PEX1- G843D/null patient fibroblasts by the flavones acacetin and acacetin diacetate [290]. To investigate other flavonoids for higher potency, we tested an initial panel of 34 commercially available flavonoids (Supplemental Table 2.1). Using our engineered PEX1-G843D/null patient cell line expressing a GFP-PTS1 reporter, referred to as PEX1-G843D-GFP-PTS1 (a telomerase immortalized ZSD patient-derived fibroblast line that is compound heterozygous for PEX1 c.[2528G>A]; [2097_2098insT], PEX1 p.[G843D];[0] ) [290], we quantified recovery of GFP- PTS1 import by scoring the proportion of cells with peroxisomal (punctate) or cytosolic GFP fluorescence, indicative of importing or non-importing cells, respectively. Representative cell images are shown in Figure 2.2A. In our initial survey, 8 / 34 (23.5%) of the tested flavonoids showed activity above the upper baseline limit of 20% importing cells (P ≤ 0.05; two-tailed Student’s t-test): diosmetin, acacetin, acacetin diacetate, apigenin, kaempferol, galangin, chrysin, and tamarixetin (Supplemental Table 2.1).

2.4.2 Dose response studies

To confirm activity, dose response experiments were performed using the 8 best candidates identified above. Since import recovery and toxicity were similar for 2 and 5 day treatments, and blinded dose response (Supplemental Figure 2.1), we show the combined results in Figure 2.2B. Structure comparisons for these compounds are provided in Figure 2.3. The 8 candidates included 5 flavones and 3 flavonols. Herein, we will discuss the results from testing these flavones followed by those of the flavonols.

Our initial group of five active flavones included four 5,7-dihydroxyflavones (diosmetin: 5,7,3'- trihydroxy-4'-methoxyflavone, acacetin: 5,7-dihydroxy-4'-methoxyflavone, apigenin: 5,7,4- trihydroxyflavone, and chrysin: 5,7-dihydroxyflavone) and acacetin diacetate (5,7-diacetoxy-4'- methoxyflavone). Diosmetin showed the best combination of potency and efficacy of all 8 candidate active flavonoids with an EC50 of approximately 2.5 µM (0.4 log10[µM]) while inducing complete import of the GFP-PTS1 reporter in nearly all cells at a 10 or 20 µM dosage (1 and 1.3 log10 [µM], respectively). Acacetin showed similar potency and activity relative to diosmetin at 5 µM (about 80% cells with import, 0.7 log10[µM]); however, cell toxicity precluded testing it at higher doses. Acacetin diacetate was less potent than acacetin, the corresponding 5,7-dihydroxy-substituted flavone, and required a 20 µM dose (1.3 log10[µM]) to achieve import in about 80% of cells. The other 5,7-dihydroxyflavones (apigenin and chrysin) were less active. At 20 µM doses, apigenin- and chrysin-treated cultures showed maximum efficacies of approximately 75% and 50% importing cells, respectively (Figure 2.2B, Supplemental table 2.2). The maximal efficacies of kaempferol and galangin were approximately 55% and 35% importing cells, respectively, at 20-30 µM dosage (1.3-1.5 log10 [µM]). Tamarixetin was the least active of these flavonols, with near baseline levels of importing cells for all concentrations tested.

2.4.3 SAR of flavonoids

In order to better define structure-activity relationships (SAR) in our system, we tested an additional 20 flavonoids to augment our initial survey of 34 compounds (Supplemental Table 2.1, Supplemental figure 2.1B, Table 2.1). In part, these compounds were chosen to survey the effects of combinations of hydroxyl- and methoxyl-substitutions at ring positions that influenced flavonoid activity in our initial survey. Although it is possible to draw numerous inferences

based on the activities of this expanded group of flavonoids, we will focus on the most robust observations that were reproducible in multiple comparisons.

Our expanded group of 54 compounds allowed us to compare the activities of all the major flavonoid groups surveyed. In all 5 cases where an informative comparison of a flavone and the corresponding flavonol (i.e. at least one compound showed activity), the flavone-treated cultures had a greater average number of importing cells (diosmetin> tamarixitin, acacetin>kaempferide, apigenin>kaempferol, chrysoeriol>isorhamnetin, and chrysin>galangin) (Table 2.1). This was especially striking when comparing diosmetin and tamarixitin (80% more importing cells in diosmetin-treated cultures at 20 µM dosage, P < 0.0001). This highlights the deleterious effects the flavonol B-ring 3-OH group can have on activity. In one case, we could compare an active flavone (apigenin, 75% importing cells) to the corresponding less active flavonol (kaempferol, 55% importing cells), isoflavone (genistein, inactive), and flavanone (naringenin, inactive) (Table 2.1, citing data for 20µM dosage). In agreement, the highly active flavone diosmetin had an inactive corresponding flavanone (hesperetin). Finally, the highly active flavone acacetin has a corresponding inactive chalcone (2',4'-dihydroxy-4-methoxychalcone) (Table 2.1).

In our expanded group, there were 33 possible informative comparisons involving an active flavonoid and another flavonoid (active or inactive) that differed by substitutions at a single position or by subclass, as discussed above (Supplemental Table 2.2). Below, we will discuss informative comparisons involving flavones and flavonols. The other flavonoids investigated herein (flavanones, isoflavones, and chalcones) did not provide additional informative comparisons of this nature due to their poor activities.

All informative comparisons involving A-ring substituent groups related to the 5-, 6-, and 7- positions (Table 2.2 and Supplemental Table 2.2): In two cases, addition of a 5-OH group to an inactive flavone (7-hydroxyflavone and 7-hydroxy-4'-methoxyflavone) resulted in a compound of increased activity (chrysin and acacetin). In another comparison, the modestly active 4',7- dimethoxy-5-hydroxyflavone and inactive 5,7, 4'-trimethoxyflavone only differ by the 5-OH group in the former and 5-OMe group in the latter. In contrast, addition of a 6-OH group to a modestly active chrysin results in an inactive compound (baicalein, see Supplemental Table 2.2). In two cases involving flavones only differing at the 7-position, flavones with a 7-OH group were more active than those with 7-OMe groups (acacetin and apigenin vs 4',7-dimethoxy-5-

hydroxyflavone and genkwanin, respectively). Finally, we note that diosmin, the 7-O-glycoside of diosmetin, was inactive in our assay.

All informative comparisons involving B-ring substituent groups related to the 2’-, 3’-, and 4’- positions (Table 2.2 and Supplemental Table 2.2): Addition of a 2’-OH to a moderately active flavonol (kaempferol) resulted in an inactive compound (morin hydrate, see Supplemental Table 2.2). The results from 3’-substitutions were varied. Addition of a 3’- OMe or -OH group to apigenin yields a similar moderately active flavone (3’-OMe, chrysoeriol) or an inactive flavone (3’-OH, luteolin). Likewise, addition of a 3’-OH group to the moderately active kaempferide results in the inactive tamarixitin. In contrast, the addition of 3’-OH group to acacetin yields diosmetin, which is also highly active, but has reduced cell toxicity. For the moderately active kaempferol, addition of either a 3’-OMe or 3’-OH group results in an inactive compound (isorhamnetin and quercetin, respectively). Further highlighting the context dependency of 3’-substituents on compound activity, the moderately active 3’-OMe flavone chrysoeriol has a corresponding 3’-OH flavone luteolin that is inactive.

All eight informative comparisons relevant to B-ring 4’-groups and flavone activity (Table 2.2 and Supplemental Table 2.2): For the modestly active chrysin and galangin, adding a 4’-OH group results in a compound with enhanced activity (apigenin and kaempferol, respectively) while adding a 4’-OMe group results in an even greater enhancement in activity (acacetin and kaempferide, respectively). In agreement, there were two other active 4’-OMe flavones (diosmetin and 4',7-dimethoxy-5-hydroxyflavone) whose corresponding 4’-OH flavones (luteolin and genkwanin, respectively) were inactive (Table 2.2).

All the informative comparisons involving the C-ring involved the 3-position (Supplemental Table 2.2): Earlier, we discussed the five cases in which the addition of a 3-OH group to an active flavone resulted in a less active or inactive flavonol (Table 2.1). We also note that the addition of a 3-OMe to diosmetin (5,7,3'-trihydroxy-4'-methoxyflavone) results in an inactive compound (3,4'-dimethoxy-3',5,7-trihydroxyflavone).

While numerous activity comparisons can be made among compounds that differ by two substitutions, we briefly highlight cases involving the 5- and 7-positions of the A-ring and 3’- and 4’-positions of the B-ring. Although both are highly active, acacetin diacetate differs from

diosmetin at both the 5- and 7-positions (5,7-acetoxy vs 5,7-(OH) groups, respectively) (Figure 2.2B, 2.3). In a related example, highly active acacetin differs from the inactive 5,7,4’- trimethoxyflavone at both the 5- and 7-positions (5,7-(OH) vs 5,7-(OMe) groups, respectively). Finally, we noted two highly active flavones that only differ by their 3’- and 4’-substituents (3’- OMe, 4’-OH: chrysoeriol vs 3’-OH, 4’-OMe: diosmetin). The corresponding flavone chrysin (5,7-dihydroxyflavone) that lacks 3’- and 4’-substituent groups shows modest activity in our assays.

2.4.4 Combination treatments

The five most effective candidate flavonoids, diosmetin, acacetin, acacetin diacetate, apigenin, and chrysoeriol (all flavones), were also tested in combination with other potential therapeutic modalities using our PEX1-G843D-GFP-PTS1 assay [290]. Betaine (trimethyl glycine) is a trimethylamine, a class of compounds with chemical chaperone activity [304], and bortezomib is a proteasome inhibitor that has also been described as a proteostasis regulator [305]. Results using flavones, betaine, and bortezomib independently or in combination are shown in Figure 2.4A. Note that lower doses for betaine (25 mM) and flavonoids (5 µM) were used than in previous experiments. Bortezomib treatment alone did not significantly increase peroxisome import. Betaine treatment alone caused moderate improvement (from 10% to 30% importing cells, P < 0.01; two-tailed Student’s t-test), while co-treatment of betaine with each flavonoid resulted in recovery 30-75% above that of the corresponding flavonoid alone (P < 0.015 in all five comparisons; two-tailed Student’s t-test). The effect of combining low doses of betaine and a flavonoid was equivalent to the sum of the individual treatments and resulted in 7-55% higher recovery than even the 10 µM flavonoid treatment alone (Figure 2.4A). Adjusting for the effect of the different drug treatments, the average outcome from additional treatment with bortezomib was approximately 7% importing cells above treatment with flavonoid alone (P = 0.015 by multiple linear regression analysis) suggesting possible modest additive or synergistic effects for these combination therapies.

2.4.5 Independent validation of peroxisome import recovery

In the GFP-PTS1 reporter assay, cells were classified in a binary manner as either “importing” or “non-importing”, which does not consider the proportion of import competent peroxisomes per cell, or the degree of functional rescue. To corroborate the results of the GFP-PTS1 reporter

assay, we performed four independent measures of peroxisome function using primary patient fibroblast lines.

Recovery of PTS2 protein import into peroxisomes. To corroborate the results of the GFP- PTS1 reporter assay, we used homozygous PEX1-G843D ZSD patient-derived primary fibroblasts to evaluate import of two endogenous PTS2 (peroxisome targeting signal 2, a nonapeptide sequence located near the N-terminus)-peroxisomal matrix enzymes, ACAA1 (thiolase) and PHYH (phytanoyl-CoA hydroxylase). After matrix import, these proteins are processed by the peroxisomal TYSND1 (Trypsin domain containing 1) protease that removes the N-terminal part of these proteins including the PTS2 sequence (30) and a smaller, mature protein is produced that can be distinguished by immunoblotting (Figure 2.4B). Since TYSND1 is a PTS1-containing protein, this assay evaluates the recovery of both PTS1- and PTS2-mediated peroxisomal matrix protein import. In control cells, the majority of ACAA1 (77%) and PHYH (83%) protein is in its mature processed form, consistent with peroxisomal processing and localization. In contrast, PEX1-G843D patient cells have no detectable mature processed ACAA1, and only 17% mature PHYH, consistent with severely impaired peroxisome assembly. Cells treated with bortezomib alone demonstrated no recovery of processing, while those treated with 50 mM betaine showed 17% mature ACAA1, and 31% mature PHYH, both observations consistent with the results of the PEX1-G843D-GFP-PTS1 assays described above. Treatment with 10 µM diosmetin was most effective and resulted in 31% mature ACAA1, and 52% mature PHYH, again consistent with PEX1-G843D-GFP-PTS1 assay results. The combination of diosmetin and bortezomib did not result in additive or synergistic effects (possibly due to difference in the sensitivities of the PEX1-G843D-GFP-PTS1 and immunoblotting assays); however, diosmetin and betaine co-treatments increased the relative levels of mature ACAA1 and PHYH to 57% and 68%, respectively.

Recovery of PEX5 localization. As an additional measure of peroxisome function recovery in PEX1-G843D patient cells, we assessed whether our most promising therapeutic candidates (betaine and diosmetin) also improve PEX5 recycling. For these and subsequent experiments, we tested primary fibroblasts homozygous for PEX1-G843D, compound heterozygous for PEX1- G843D and a null allele, and homozygous for PEX1-null alleles. PEX5 was primarily cytosolic in control cells, but localized to peroxisomes in PEX1-mutant cells (Figure 2.5A). A 2-day

exposure to betaine, diosmetin, or the combination, corrected PEX5 localization to the cytosol in 65-75% of PEX1-G843D homozygous cells, and 55, 65, and 75% of PEX1-G843D/null compound heterozygous cells, respectively (Figure 2.5B) (P < 0.001 in all comparisons; two- tailed Student’s t-test). Treatment had no effect in PEX1-null cells. In control cells, PEX5 location was invariably cytosolic, regardless of intervention (Figure 2.5A).

Recovery of peroxisome metabolite levels. Peroxisomal enzymatic activities are essential for the catabolism of very long chain fatty acids (VLCFA) and the synthesis of plasmalogens, membrane vinyl ether phospholipids [reviewed in [152]]. Accordingly, in ZSD cells, C26:0 VLCFA levels are elevated and plasmalogen levels are decreased relative to those derived from healthy controls. To determine if betaine and diosmetin can improve these metabolite levels, we measured C26:0 lyso-PC (C26:0 incorporated into phosphatidylcholine) and ethanolamine plasmalogen lipids in primary fibroblasts after 10 days of treatment (Figure 2.5C, D). Diosmetin decreased C26:0 lyso-PC levels by 60-70% in the PEX1-G843D patient cells, but betaine had no effect, and the combination of betaine and diosmetin was not synergistic. Treatment had a more robust effect on plasmalogens. Betaine increased plasmalogen levels by 25% and 40%, and diosmetin increased plasmalogen levels by 100% and 150% in the PEX1-G843D homozygous and PEX1-G843D/null compound heterozygous cells, respectively. With combination treatment, there was 200% improvement in plasmalogens in the compound heterozygous cells (P < 0.05- 0.001 in all comparisons; Two-tailed Student’s t-test). In PEX1-null cells, betaine and diosmetin alone or in combination did not improve peroxisome import, PEX5 recycling, or metabolite levels, consistent with their proposed mechanism of action as chemical and pharmacological chaperones that require residual PEX1 protein (PEX1-G843D and possibly other misfolded PEX proteins) for their activity. Treatment neither reduced C26:0 lyso-PC nor increased plasmalogens in control cells, supporting compound action on peroxisome-mediated pathways.

PEX1, PEX6, and PEX5 protein levels. To determine if functional recovery corresponds to an increase in PEX1, PEX6, and PEX5 protein amounts, we performed immunoblotting of whole cell lysates after drug treatment in PEX1-G843D homozygous, PEX1-G843D/null compound heterozygous, and PEX1-null cells (Figure 2.6). In PEX1-G843D homozygous cells, both diosmetin and combination therapy with betaine and diosmetin improved PEX1-G843D, PEX6, and PEX5 levels. In the PEX1-G843D/null compound heterozygous cells a small, but significant,

improvement in PEX5 was achieved with combination therapy, as was a non-significant trend toward higher PEX1-G843D levels with diosmetin and combination treatment (P < 0.05-0.01 in all comparisons; Two-tailed Student’s t-test). There was no improvement in PEX6 or PEX5 protein levels in PEX1-null cells, consistent with the results of the biochemical analyses. Levels of the peroxisome membrane protein PEX14 were unchanged by drug treatment.

2.4.6 PEX1 ATP-binding domain 2 and its role in ligand binding

We aligned the PEX1 protein sequence with that of its most closely related structural neighbor valosin-containing protein (VCP or p97), determined using I-TASSER (Iterative Threading ASSEmbly Refinement) and RaptorX programs. This alignment demonstrated conservation of PEX1 glycine 843, immediately upstream of the second AAA cassette (D2) (Supplemental Figure 2.2A). A ligand interaction diagram was generated for VCP with ADP bound (using PDB ID: 3CF0 [302]), and showed that glycine 480, the residue corresponding to PEX1-G843, directly interacts with the bound ADP molecule (Supplemental Figure 2.2B). This model suggests that the aspartate substitution could influence the ability of PEX1-G843D to effectively bind and/or hydrolyze ATP.

2.5 DISCUSSION

In this study, we compared a targeted panel of flavonoids for their ability to recover peroxisomal enzyme import and to elucidate structure-activity relationships in PEX1-G843D patient cells. We used 5 independent measures to confirm recovery: (i) direct counting of importing cells in our GFP-PTS1 reporter assay, (ii) assessing proportions of processed (imported) endogenous peroxisome matrix enzymes, (iii) determining PEX5 sub-cellular localization, (iv) measuring VLCFA and plasmalogen levels, and (v) quantifying PEX1, PEX6 and PEX5 protein levels. In these experiments, diosmetin was the most active flavonoid and combination treatments of diosmetin with betaine generally produced an additive effect at lower compound doses.

2.5.1 SAR of flavonoids tested

Structure-activity relationships were determined based on structure comparisons among flavonoids of varying activity (see Tables 2.1-2 and Supplemental Tables 2.1-2). All the highly active flavonoids in our assay were flavones, and showed higher activity than their corresponding flavonol, flavanone, isoflavone, or chalcone in all comparable cases.

The reduced activity of the flavonols highlights the deleterious effect of the C-ring 3-OH group on activity in our assay, which is further supported by the deleterious effects of 3-OMe and 3-O- glycoside substitutions. Based on structural aspects alone, the inactivity in our GFP-PTS1 system of (i) flavanones relates to the absence of the C2-C3 double bond of the C-ring, (ii) isoflavones relates to the position of the linkage of the B-ring to the C-ring, and (iii) chalcones relates to the 3-carbon open chain in place of the C-ring. The mechanism(s) by which these structural changes directly relate to the inactivity of these compounds in our assays is unclear and warrants further investigation.

SAR also highlighted the importance of specific groups on all three rings in multiple contexts, highlighting the influence of the functional groups located at the 5- and 7-positions of the A-ring as well as 3’- and 4’-substituents of the B-ring. In particular, all highly active compounds were 5,7-dihydroxyflavones, with the exception of acacetin diacetate, a 5,7-diacetoxyflavone. In informative comparisons involving chrysin and galangin, addition of a B-ring 4’-OMe or 4’-OH group resulted in a more active compound. In all four comparisons in which compounds differed only by having a 4’-OMe or 4’-OH group, the compound with a 4’-OMe group was more active, but the magnitude of the changes were context dependent. As discussed in the Results sections, the effects of B-ring 3’-substituents on compound activity were context-dependent.

In general, the SAR we observed were consistent with findings from others who evaluated the ability of flavonoids to bind ATP-binding sites and inhibit ATPases (protein kinases and ABC transporters). Similar findings include: (i) greater activity of flavones relative to flavonols [306- 308], though some studies suggest the reverse relationship [293], (ii) 5, 7-hydroxyl groups enhance activity [293, 308]; (iii) 4’-substitutions enhance activity, with methoxyl substitutions conferring greater potency than hydroxyl substitutions [293, 306, 307]; (iv) flavanones, isoflavones and chalcones being least potent [293, 306, 308].

The interactions of flavonoids with ATP-binding proteins have been extensively studied. For example, the 5,7 –hydroxyl groups of kaempferol were modeled to interact directly with the ATP-binding site of the ribosomal S6 kinase 2 (RSK2) [294]. Acacetin and other flavonoids were modeled to directly bind the ATP-binding site in phosphoinositol-3-kinase (PI3K), and also found to compete with ATP-binding [309, 310]. Several flavonoids, including acacetin, directly compete with ATP to bind to ABCG2 (aka BCRP1), a xenobiotic transporter that is a member of

the superfamily of ATP-binding cassette (ABC) transporters, to prevent drug efflux [306, 311, 312].

2.5.2 Proposed mechanism of action of flavonoids

The effect of flavonoids on the stability and activity of AAA ATPases has not been previously reported. Although we appreciate alternative hypotheses discussed later, we propose that flavonoids could act as pharmacological chaperones for the PEX1-G843D protein and/or the corresponding PEX1-PEX6-PEX26 exportomer complex. Given the predicted position of G843 in the ATP pocket (Supplemental Figure 2.2B), and the literature supporting the binding of flavonoids at or near ATP-binding sites [reviewed in [313, 314]], we speculate that our candidate compounds interact dynamically with the ATP-binding sites of PEX1 and/or the PEX1-PEX6 complex, resulting downstream in increased peroxisome import.

The increase in PEX1-G843D protein levels after diosmetin treatment, and the corresponding increase in its binding partner PEX6 (Figure 2.6), is consistent with our hypothesis. This stabilization effect could occur either in the cytosol to increase PEX1-G843D-PEX6 complex formation and/or stability, or on the sub-population of PEX1-G843D-PEX6 complexes on the peroxisome membrane that interact with the PEX26 peroxisomal membrane protein. In patient cells with at least one PEX1-G843D allele, the most marked increase was in PEX5 levels, likely due to improved ‘exportomer’ function and decreased PEX5 degradation. The recovery of PEX5 localization after diosmetin treatment corroborates this observation (Figure 2.5B). The failure of diosmetin to improve PEX6 or PEX5 levels, PEX5 localization, or peroxisome metabolites in PEX1-null cells (Figure 2.6, 2.5B, 2.5C) suggests an effect either directly on PEX1 or on the PEX1-PEX6 complex. Alternatively, diosmetin could, directly or indirectly, upregulate PEX1 or PEX6 gene expression.

Unexpectedly, in PEX1 null cells there were more cells classified as having cytosolic versus peroxisomal PEX5 localization compared to the PEX1-G843D cell lines at baseline (Figure 2.5B). The absence of the ‘exportomer’ complex in PEX1-null cells could exacerbate the degradation of PEX5 trapped at the peroxisome membrane, as well as pexophagy. PEX1 null cells have markedly fewer peroxisomes than PEX1-G843D cells, and thus identifying cells with PEX5 protein localized to >10 peroxisomes becomes less likely. Furthermore, since PEX1-null

57 cells contain less PEX5 protein than PEX1-G843D cells (Figure 2.6A), the data suggest that the greater proportion of null cells with cytosolic PEX5 does not imply greater PEX5 stability.

Although our discussion above primarily focuses on the hypothesis that flavonoids differ in activity in our assay systems based on their ability to interact directly with PEX1 and/or PEX1- PEX6 exportomer complexes, we also note that our observations could also be influenced by differences in cell permeability, stability, and/or intracellular modification. Furthermore, we recognize the possibility that the flavonoids could have an indirect mechanism of action by modulating the activity of other proteins involved in mediating the turnover, folding, and functions of PEX proteins, including, but not limited to, PEX1 and PEX6.

2.5.3 Potential therapeutic implications

We have shown that by combining a chemical chaperone (betaine) with a pharmacological chaperone (flavonoids) we can achieve additive effects at lower doses, which may reduce potential side effects and achieve better results than the individual high dose alone. Diosmin (Daflon®), betaine (Cystadane®) and bortezomib (Velcade®) are approved drugs for other conditions and thus have established pharmacological and toxicity profiles. Diosmin, the glycosylated form of diosmetin, is used in the treatment of chronic venous disease. Human intestinal flora remove the glycoside group of diosmin, leaving diosmetin to be absorbed and rapidly distributed widely throughout the body [315]. After a single oral 500mg dose of a diosmin preparation, plasma diosmetin levels in healthy volunteers were above 1 mM [315], which far surpasses the effective concentration of diosmetin used in our cell based assay. Flos Chrysanthemi extract, another available supplement, leads to increased circulating plasma levels of luteolin, apigenin, diosmetin, and chrysoeriol when administered orally to rats [316], including 3 of the most potent flavonoids identified in our assay. Importantly, flavonoids, such as diosmetin and other flavones, reportedly cross the blood-brain barrier to some extent [317] which, in principle, would be required for treating the neurological aspects of ZSD. In addition, flavones have neuroprotective effects [318, 319], and also modulate neurologic behavior in mice [320, 321] indicating that they can have a positive influence on CNS functions.

In principle, small molecule compounds that improve PEX1-PEX6 functions in ZSD cells by different mechanisms could be used in combination and have increased potency and/or efficacy

relative to the corresponding monotherapies. One example, as we have demonstrated, is the combination of potential chemical and pharmacologic chaperones, such as betaine and diosmetin. Furthermore, the addition of a proteostasis regulator may increase the half-life of a misfolded protein, such as PEX1-G843D, and thus potentially enhance the effectiveness of pharmacological chaperone therapy [322, 323]. Our observations in PEX1-G843D-GFP-PTS1 patient cells co-treated with diosmetin and bortezomib provide evidence that this approach could enhance peroxisome assembly in PEX1-G843D ZSD patient-derived cells. However, the chemotherapeutic agent bortezomib is likely ill-suited for the long-term treatment of chronic genetic disorders, such as ZSD and thus other proteostasis regulators should be evaluated.

In summary, we have identified PEX1-G843D as a potential molecular target for chaperone therapy, which could benefit a large proportion of people with ZSD. We have also identified a group of potential pharmacological chaperones that could be optimized once their mechanism of action is more fully explored and validated. Given that certain flavonoids are currently used for the treatment of other indications in adults without noteworthy serious adverse effects, we suggest they warrant preclinical testing in induced pluripotent stem cell (IPSC) models of ZSD and in the Pex1-G844D mouse model that is homozygous for the murine equivalent of the human PEX1-G843D allele and reflects numerous aspects of mild ZSD [264, 324]. Lastly, bioactive flavonoids also warrant testing in ZSD patient-derived cells with other hypomorphic PEX gene missense alleles that could be responsive to pharmacological chaperone therapy.

2.6 ACKNOWLEDGMENTS

This work was supported by the Rare Disease Foundation (Children’s & Women’s Health Centre of British Columbia), Research in RCDP & Zellweger Fund (217703), and Canadian Institutes of Health Research to NB (CIHR 126108 and 34575) and The Global Foundation for Peroxisomal Disorders (GFPD) to JGH. GM and CA are graduate students in the Department of Human Genetics of McGill University. GM was supported by an NSERC training award. CA is supported by the Fonds de recherche du Québec - Santé (FRQS Doctoral training award 32105). Ning Huang (Hacia Laboratory, USC) provided initial observations for bortezomib treatment, with which Kim Siegmund (USC) assisted in statistical analysis. This work was spearheaded by GM, who passed away prior to the submission of this manuscript, and is dedicated to her family,

59 friends and colleagues and all the families affected by ZSD whose lives she has touched through her compassion and devotion to finding more effective therapies for this disease.

2.7 CONFLICTS OF INTEREST

There are no conflicts of interest.

Figure 2.1. Flavonoid subclasses. The basic chemical structures and numbering systems of the major classes of flavonoids investigated are provided.

Figure 2.2 Recovery of peroxisome import in PEX1-G843D-GFP-PTS1 cells. (A) Representative image showing redistribution of the GFP-PTS1 reporter from the cytosol to peroxisomes (puncta) after treatment with diosmetin. (B) Dose response for the eight most effective flavonoids. The number of importing and non-importing cells were scored visually, and expressed as a percentage of importing cells. Flavones are shown by a solid symbol; corresponding flavonols are shown by an open symbol of the same color. Untreated or vehicle treated cells had an average baseline import of 8% (lower dotted line, range = 0 - 20%). The chemical chaperone betaine at 100 mM recovered import in 87% of cells on average (upper dotted line). Results are pooled from 3 independent experiments, each performed in triplicate (≥

100 cells scored per replicate). Error bars indicate standard error; Student’s two-tailed t-test was used to determine recovery over baseline.

Figure 2.3. Chemical structures of the most active flavonoids Flavones are on the left and their corresponding less active or inactive flavonols on the right. Functional groups discussed in the text are highlighted.

Figure 2.4. Effects of combination treatment on PEX1-G843D cells. (A) PEX1-G843D-GFP- PTS1 cells were treated with single compounds (bortezomib 1.5 nM, betaine 25 mM, flavone 5 µM), independently or in combination for 2 days. The last column is flavone at 10 µM. Numbers of importing and non-importing cells were scored visually and expressed as a percentage of importing cells; error bars indicate standard error. Low dose betaine moderately recovered import (Student’s two-tailed t-test; P < 0.01), bortezomib plus flavone was slightly more effective than 5 µM flavone alone (multiple linear regression, P = 0.015) and betaine plus flavone was most effective (Student’s two-tailed t-test; P <0.0001), more so than high dose flavone alone. N=3; Error bard indicate standard error. (B) Primary PEX1-G843D homozygous fibroblasts were treated with 1.5 nM bortezomib (Bort, Bo), 50mM betaine (Bet, Be), or 10µM diosmetin (Dios, D), independently or in combination. “Ctrl” denotes non-PBD-ZSD control cells. Immunoblots show the larger precursor which is cleaved inside peroxisomes to mature ACAA1 and PHYH. Numbers under bands represent percentage of mature enzyme to total enzyme, plotted below (quantified by densitometry). In control cells the majority of ACAA1 and

PHYH show peroxisomal in localization, while they are primarily in the cytoplasm in PEX1- G843D cells. Enzyme localization was improved by betaine and diosmetin, and their combination was synergistic. This experiment was performed in duplicate. Error bars indicate standard error.

Figure 2.5. Effect of betaine, diosmetin, and combination treatments on PEX5 localization and peroxisome metabolite levels in PEX1-deficient primary fibroblasts. Cells were treated with 100 mM betaine, 10 µM diosmetin, or 50 mM betaine and 10 µM diosmetin, as indicated. (A) In control cells PEX5 is distributed throughout the cytosol; in PEX1 mutant cells PEX5 is primarily localized at peroxisomes; PEX5 relocalizes to the cytosol after 2-day treatment in cells with at least 1 PEX1-G843D allele. Images were visualized by indirect immunofluorescent microscopy with PEX5 (green) and peroxisome membrane protein ABCD3 (red/magenta); co- localization (yellow). (B) Candidate compounds recovered cytosolic PEX5 in 65-75% of PEX1- G843D homozygous cells, and 55-75% of PEX1-G843D/null compound heterozygous cells, but have no effect on PEX1 null cells (Student’s two-tailed t-test; P < 0.001). Error bars indicate standard error. Over 400 cells were scored visually per treatment in 2 separate experiments. (C, D) Levels of VLCFA (C26:0 lyso-PC) and ethanolamine plasmalogens (PE) after 10 day treatments were measured by LC/MSMS. C26:0 lyso-PC levels decreased with 10 µM diosmetin

in PEX1-G843D cell lines; betaine and diosmetin combination was not synergistic. PE plasmalogens increased with all compounds. Compounds did not recover PEX1 null cells; Error bars indicate standard error; Student’s two-tailed t-test, * P < 0.05, ** P < 0.01, *** P < 0.001; n=4 (2 separate experiments performed in duplicate).

Figure 2.6. Effect of betaine, diosmetin, and combination treatments on PEX protein levels in primary fibroblasts. Cells were treated with 100 mM betaine, 10 µM diosmetin, or 50 mM betaine and 10 µM diosmetin, as indicated. (A) Immunoblots show the amount of PEX1, PEX6, PEX5, PEX14, and β-tubulin (loading control) in control, PEX1-G843D homozygous, PEX1- G843D/null compound heterozygous and PEX1 null cells after 6-day treatment. (B) Diosmetin treatments increased PEX1, PEX6, and PEX5 levels in PEX1-G843D homozygous cells, and PEX5 levels in PEX1-G843D/null cells (quantified using densitometry). Dotted line represents baseline protein amount for each cell line, set to “1”. Two different patient cell lines were used for each genotype and controls, and lysates were tested in duplicate in independent experiments (n=4); Error bars indicate standard error; Student’s two-tailed t-test, * P =< 0.05, ** P < 0.01, *** P < 0.001.

Table 1. All informative comparisons of at least one active flavonoid and another flavonoid that differs by subclass % Import % Import Comparison Compound 3 5 7 3' 4' P P (10 µM) (20 µM)

Diosmetin OH OH OH OCH3 92 95 <0.0001 <0.0001 Tamarixitin OH OH OH OH OCH3 < 20 < 20

Acacetin OH OH OCH3 72* 72* 0.331 - Kaempferide OH OH OH OCH 62 ND flavone 3 Apigenin OH OH OH 51 75 vs 0.361 0.024 flavonol Kaempferol OH OH OH OH 42 55 Chrysoeriol OH OH OCH3 OH 52 78 0.051 0.081 Isorhamnetin OH OH OH OCH3 OH < 20 43 Chrysin OH OH 33 48 0.128 0.027 Galangin OH OH OH 20 37

Diosmetin OH OH OH OCH3 92 95 flavone - <0.0001 Hesperetin OH OH OH OCH ND < 20 vs 3 Apigenin OH OH OH 51 75 flavanone - <0.0001 Naringenin OH OH OH - < 20 flavone vs Apigenin OH OH OH 51 75 - isoflavone Genistein OH OH OH ND < 20 <0.0001

flavone Acacetin OH OH OCH3 72* 72* vs 2',4'-dihydroxy-4- <0.0001 - OH OH OCH < 20 ND chalcone methoxychalcone** 3 *data from 5uM dosage; **chemical groups assigned to corresponding flavone numbering ND - Not done due to inactivity of compound at higher dose or toxicity

Table 2.1. All informative comparisons at the indicated A-, B-, and C-ring positions. Percent importing cells at 10 and 20 µM is shown for compounds that rescued import in over 20% of cells. P-values represent comparison of percent importing cells between 2 compounds by Student’s two-tailed t-test.

Table 2. All informative comparisons at the indicated A-, B-, and C-ring positions % Import % Import Position Compound 3 5 7 3' 4' P P (10 µM) (20 µM) 7-hydroxyflavone OH ND < 20 - <0.0001 Chrysin OH OH 33 48 7-hydroxy-4'- OH OCH3 < 20 < 20 methoxyflavone <0.0001 0.002

A-5 Acacetin OH OH OCH3 72* 72* 4',7-dimethoxy-5- OH OCH OCH 30 34 hydroxyflavone 3 3 0.082 0.0004 5,7,4'- OCH OCH OCH < 20 < 20 trimethoxyflavone 3 3 3

Acacetin OH OH OCH3 72* 72* 4',7-dimethoxy-5- <0.0001 0.017 OH OCH OCH 30 34 A-7 hydroxyflavone 3 3 Apigenin OH OH OH 51 75 - 0.0002 Genkwanin OH OCH3 OH ND < 20 Diosmetin OH OH OH OCH 92 95 3 <0.0001 <0.0001 Diosmin OH OH OH O- Glu** < 20 < 20

Acacetin OH OH OCH3 72* 72* 0.002 0.0005 Diosmetin OH OH OH OCH3 92 95 Apigenin OH OH OH 51 75 - <0.0001 Luteolin OH OH OH OH ND 3

Kaempferide OH OH OH OCH3 62 ND 0.002 - Tamarixitin OH OH OH OH OCH3 < 20 15 Apigenin OH OH OH 51 75 B-3' 0.955 0.787 Chrysoeriol OH OH OCH3 OH 52 78 Chrysoeriol OH OH OCH OH 52 78 3 - <0.0001 Luteolin OH OH OH OH ND < 20 Kaempferol OH OH OH OH 42 55 0.003 0.358 Isorhamnetin OH OH OH OCH3 OH < 20 43 Kaempferol OH OH OH OH 42 55 - 0.002 Quercetin OH OH OH OH OH - < 20 Diosmetin OH OH OH OCH 92 95 3 - <0.0001 Luteolin OH OH OH OH ND < 20 Chrysin OH OH 33 48 0.162 0.0008 Apigenin OH OH OH 51 75 Chrysin OH OH 33 48 0.0003 0.024 Acacetin OH OH OCH3 72* 72* Acacetin OH OH OCH 72* 72* 3 0.049 0.409 Apigenin OH OH OH 51 75 B-4' Galangin OH OH OH 20 37 0.016 0.021 Kaempferol OH OH OH OH 42 55 Galangin OH OH OH 20 37 0.013 - Kaempferide OH OH OH OCH3 62 ND Kaempferide OH OH OH OCH 62 ND 3 0.094 - Kaempferol OH OH OH OH 42 55 4',7-dimethoxy-5- OH OCH3 OCH3 30 34 hydroxyflavone - 0.005

Genkwanin OH OCH3 OH ND < 20 *experiment conducted at 5uM dosage;**O-glucoside ND - Not done due to inactivity of compound at higher dose or toxicity Table 2.2 All informative comparisons at the indicated A-, B-, and C-ring positions. Percent importing cells at 10 and 20 µM is shown for compounds that rescued import in over 20% of

cells. P-values represent comparison of percent importing cells between 2 compounds by AStudent’s two-tailed t-test.

Supplemental Figure 2.1 Dose responses for GFP-PTS1 import in PEX1-G843D-GFP-PTS1 cells treated with flavonoids. (A) Compounds had a similar effect on import recovery after 2- day or 5-day treatment, or after blinded evaluation following 4-day treatment. (B) Follow-up targeted analysis for SAR. Untreated cells had an average baseline import of 7% (lower dotted line, range = 2-20%). The chemical chaperone betaine at 100mM recovered import in 87% of cells on average (upper dotted line). The number of importing and non-importing cells were scored visually, and expressed as a percentage of importing cells; error bars indicate standard error. Flavones are represented by a solid symbol; corresponding flavonols are represented by an open symbol of the same color. N=3 for each experiment.

Supplemental Figure 2.2. Conservation of the ATP binding domain in VCP and PEX1. (A) An alignment of PEX1 and VCP, another AAA ATPase, illustrates the high conservation of the second AAA cassette (D2), including conservation of the Walker A and Walker B motifs (responsible for ATP binding and hydrolysis, respectively) and the glycine residue which is mutated in PEX1-G843D. (B) A ligand interaction diagram for the crystallized D2 subdomain of VCP with ADP bound shows that the VCP glycine 480 residue, which corresponds to the PEX1- G843 residue, directly interacts with the bound ADP molecule. Glycine 843 (*) and corresponding residues predicted to interact with ADP are indicated.

Supplemental Table 1. All flavonoids evaluated in this study % Importing EC50 Class Chemical Name Supplier IUPAC Name Screening Cells (10 µM) (µM) 3',4'- dihydroxyflavone < 20% - Extrasynthese 1204 2-(3,4-dihydroxyphenyl)chromen-4-one + 3,4'-dimethoxy-3',5,7-trihydroxyflavone < 20% - Apin Chemicals N04770 d 5,7-dihydroxy-2-(3-hydroxy-4-methoxyphenyl)-3-methoxy-4H-chromen-4-one * 4',7-dimethoxy-5-hydroxyflavone 30% 5 Apin Chemicals N06144 a 5-hydroxy-7-methoxy-2-(4-methoxyphenyl)-4H-chromen-4-one * 4'-hydroxyflavone < 20% - Zamboni Chem Solutions 2-(4-hydroxyphenyl)chromen-4-one * 4'-methoxyflavone < 20% - Zamboni Chem Solutions 2-(4-methoxyphenyl)chromen-4-one * 5,7,3',4'-tetramethoxyflavone < 20% - Indofine L-102 2-(3,4-dimethoxyphenyl)-5,7-dimethoxy-4H-chromen-4-one * 5,7,3'-trimethoxyflavone < 20% - Indofine 22-358 5,7-dimethoxy-2-(3-methoxyphenyl)-4H-chromen-4-one * 5,7,4'-trimethoxyflavone < 20% - Indofine A-106 5,7-dimethoxy-2-(4-methoxyphenyl)-4H-chromen-4-one * 6,7- dihydroxyflavone < 20% - Extrasynthese 1012 6,7-dihydroxy-2-phenyl-4H-chromen-4-one + 7,3',4'-trimethoxyflavone < 20% - Indofine T-204 2-(3,4-dimethoxyphenyl)-7-methoxy-4H-chromen-4-one * 7-hydroxy-4'-methoxyflavone < 20% - Indofine H-427 7-hydroxy-2-(4-methoxyphenyl)-4H-chromen-4-one * 7-hydroxyflavone < 20% - Extrasynthese 1060 7-hydroxy-2-phenylchromen-4-one + Acacetin (5,7-dihydroxy-4'-methoxyflavone) 72% (5 µM) 0.5 Sigma-Aldrich 00017 5,7-dihydroxy-2-(4-methoxyphenyl)chromen-4-one +

Microsource Discovery Systems Acacetin diacetate (5,7-diacetoxy-4'-methoxyflavone) 60% 5 [5-acetyloxy-2-(4-methoxyphenyl)-4-oxochromen-7-yl] acetate + 200833 FLAVONES Apigenin (5,7,4-trihydroxyflavone) 60% 2.5 Sigma-Aldrich A3145 5,7-dihydroxy-2-(4-hydroxyphenyl)chromen-4-one +

Ayanin (3',5-Dihydroxy-3,4',7-trimethoxyflavone) 20% 2.5 Apin Chemicals N34055 a 5-hydroxy-2-(3-hydroxy-4-methoxyphenyl)-3,7-dimethoxy-4H-chromen-4-one *

Baicalein (5,6,7-trihydroxyflavone) < 20% - Sigma-Aldrich 465119 5,6,7-trihydroxy-2-phenylchromen-4-one + Baicalein (5,6,7-trimethylether) < 20% - 5,6,7-trimethoxy-2-phenylchromen-4-one + Chrysin (5,7-dihydroxyflavone) 33% 2.5 Sigma-Aldrich C80105 5,7-dihydroxy-2-phenylchromen-4-one + Chrysoeriol (4',5,7-trihydroxy-3'-methoxyflavone) 52% 2.5 Apin Chemicals N07188 c 5,7-dihydroxy-2-(4-hydroxy-3-methoxyphenyl)-4H-chromen-4-one * Diosmetin (5,7,3'-trihydroxy-4'-methoxyflavone) 92% 2.5 Extrasynthese 1108 S 5,7-dihydroxy-2-(3-hydroxy-4-methoxyphenyl)chromen-4-one + Flavone < 20% - Zamboni Chem Solutions 2-phenylchromen-4-one * Genkwanin (4',5-dihydroxy-7-methoxyflavone) < 20% - Extrasynthese 1147 5-hydroxy-2-(4-hydroxyphenyl)-7-methoxychromen-4-one + Luteolin (3',4',5,7-tetrahydroxyflavone) < 20% - Sigma-Aldrich L9283 2-(3,4-dihydroxyphenyl)-5,7-dihydroxychromen-4-one + Retusin (5-hydroxy-3,7,3',4'-tetramethoxyflavone) < 20% - Indofine 21074 2-(3,4-dimethoxyphenyl)-5-hydroxy-3,7-dimethoxy-4H-chromen-4-one * Tangeretin (4',5,6,7,8-pentamethoxyflavone) < 20% - Extrasynthese 1033 5,6,7,8-tetramethoxy-2-(4-methoxyphenyl)chromen-4-one + 5-hydroxy-2-(3-hydroxy-4-methoxyphenyl)-4-oxo-4H-chromen-7-yl 6-O-(6- Diosmin (3',5,7-trihydroxy-4'-methoxyflavone-7-rutinoside) < 20% - Sigma-Aldrich D3525 + deoxy-α-D-mannopyranosyl)-β-D-glucopyranoside Fisetin (3,3',4',7-tetrahydroxyflavone) <20% - Sigma-Aldrich F4043 2-(3,4-dihydroxyphenyl)-3,7-dihydroxychromen-4-one + Galangin (3,5,7-trihydroxyflavone) 20% 5 Sigma-Aldrich 282200 3,5,7-trihydroxy-2-phenylchromen-4-one + Isorhamnetin (3,4',5,7-tetrahydroxy-3'-methoxyflavone) < 20% 15 Indofine 021120s 3,5,7-trihydroxy-2-(4-hydroxy-3-methoxyphenyl)-4H-chromen-4-one * Kaempferide (4'-methoxy-3,5,7-trihydroxyflavone) 62% 2.5 Indofine K-101 3,5,7-trihydroxy-2-(4-methoxyphenyl)-4H-chromen-4-one * Kaempferol (3,4',5,7-tetrahydroxyflavone) 42% 2.5 Sigma-Aldrich 60010 3,5,7-trihydroxy-2-(4-hydroxyphenyl)chromen-4-one + Morin hydrate (2',3,4',5,7-pentahydroxyflavone) <20% - Sigma-Aldrich M4008 2-(2,4-dihydroxyphenyl)-3,5,7-trihydroxychromen-4-one hydrate + Myricetin (3,5,7,3',4',5'-hexahydroxyflavone) <20% - Sigma-Aldrich M6760 3,5,7-trihydroxy-2-(3,4,5-trihydroxyphenyl)chromen-4-one + FLAVONOLS Quercetin (3,3',4',5,7-pentahydroxyflavone) < 20% - Sigma-Aldrich G4951 2-(3,4-dihydroxyphenyl)-3,5,7-trihydroxychromen-4-one + Rhamnetin (3,5,3',4'-tetrahydroxy-7-methoxyflavone) < 20% - Extrasynthese 1136 S 2-(3,4-dihydroxyphenyl)-3,5-dihydroxy-7-methoxychromen-4-one + Robinetin (3,7,3',4',5'-pentahydroxyflavone) < 20% - Extrasynthese 1138 S 3,7-dihydroxy-2-(3,4,5-trihydroxyphenyl)chromen-4-one + Tamarixetin (3,3',5,7-tetrahydroxy-4'-methoxyflavone) < 20% - Extrasynthese 1140 S 3,5,7-trihydroxy-2-(3-hydroxy-4-methoxyphenyl)chromen-4-one + 2-(3,4-dihydroxyphenyl)-5,7-dihydroxy-4-oxo-4H-chromen-3-yl 6-deoxy-α-L- Quercitrin (3,3',4',5,7-pentahydroxyflavone-3-L-rhamnoside) < 20% - Extrasynthese 1236 S + mannopyranoside

2-(3,4-dihydroxyphenyl)-5,7-dihydroxy-4-oxo-4H-chromen-3-yl 6-O-(6-deoxy-α- Rutin (3,3',4',5,7-pentahydroxyflavone-3-rutinoside) < 20% - Extrasynthese 1139 S + L-mannopyranosyl)-β-D-glucopyranoside

(2S)-5,7-dihydroxy-2-(3-hydroxy-4-methoxyphenyl)-2, Hesperetin (3',5,7-trihydroxy-4'-methoxyflavanone) < 20% - Sigma-Aldrich H4125 + 3-dihydrochromen-4-one

Naringenin (4',5,7-trihydroxyflavanone) < 20% - Sigma-Aldrich N5893 (2S)-5,7-dihydroxy-2-(4-hydroxyphenyl)-2,3-dihydrochromen-4-one +

(2S)-5-hydroxy-2-(3-hydroxy-4-methoxyphenyl)-4-oxo-3,4-dihydro-2H-chromen- Hesperidin (3',5,7-trihydroxy-4'-methoxyflavanone-7-rutinoside) < 20% - Sigma-Aldrich H5254 + FLAVONONES 7-yl 6-O-(6-deoxy-α-L-mannopyranosyl)-β-D-glucopyranoside

(S )-4′-Methoxy-3′,5,7-trihydroxyflavanone-7-[2-O-(α-L-rhamnopyranosyl)-β-D- Neohesperidin < 20% - Sigma-Aldrich N1887 + glucopyranoside]

(2S)-5-hydroxy-2-(4-hydroxyphenyl)-4-oxo-3,4-dihydro-2H-chromen-7-yl 2-O- Naringin(4',5,7-trihydroxyflavanone-7-rhamnoglucoside) < 20% - Sigma-Aldrich N1376 + (6-deoxy-α-L-mannopyranosyl)-β-D-glucopyranoside

Daidzein (4',7-dihydroxyisoflavone) < 20% - Sigma-Aldrich D7802 7-hydroxy-3-(4-hydroxyphenyl)chromen-4-one + ISOFLAVONES Formononetin (7-hydroxy-4'-methoxyisoflavone) < 20% - Extrasynthase 1248 S 7-hydroxy-3-(4-methoxyphenyl)chromen-4-one + Genistein (4',5,7-trihydroxyisoflavone) < 20% - Sigma-Aldrich G6649 5,7-dihydroxy-3-(4-hydroxyphenyl)chromen-4-one +

Class Chemical Name Recovery (10 µM) Supplier IUPAC Name Screening

2',4'-dihydroxy-4-methoxychalcone < 20% - Indofine D-507 (2E)-1-(2,4-dihydroxyphenyl)-3-(4-methoxyphenyl)-2-propen-1-one * 2',4'-dihydroxy-3,4-dimethoxychalcone < 20% - Indofine 22-208 (2E)-1-(2,4-dihydroxyphenyl)-3-(3,4-dimethoxyphenyl)-2-propen-1-one * 3,4,2',4',6'-pentamethoxychalcone < 20% - Indofine 19-439 3-(3,4-dimethoxyphenyl)-1-(2,4,6-trimethoxyphenyl)-2-propen-1-one * 3,2'-dihydroxy-4,4',6'-trimethoxychalcone < 20% - Indofine 19-536 1-(2-hydroxy-4,6-dimethoxyphenyl)-3-(3-hydroxy-4-methoxyphenyl)-2-propen-1-o * CHALCONES (2E)-3-(3,4-dimethoxyphenyl)-1-(2-hydroxy-4,6-dimethoxyphenyl)-2-propen-1- 2'-hydroxy-3,4,4',6'-tetramethoxychalcone < 20% - Indofine 19-438 * one

1-(4-((2-O -[6-deoxy-α-L-mannopyranosyl]-β-D-glucopyranosyl)oxy)-2,6- neohesperidin dihydrochalcone < 20% - Sigma-Aldrich 8757 + dihydroxyphenyl)-3-[3-hydroxy-4-methoxyphenyl]-1-propanone

+ 34 initially screened flavonoids; * additional flavonoids for SAR refinement

Supplemental Table2.1. All flavonoids evaluated in this study. Percent importing cells and EC50 at 10 µM is shown for compounds that rescued import in over 20% of cells.

Supplemental Table 2.2. All informative comparisons of an active flavonoid and another flavonoid that differ by substitutions at a single position or by subclass % Import % Import Position Compound 3 5 6 7 2' 3' 4' P P (10 µM) (20 µM)

Diosmetin OH OH OH OCH3 92 95 <0.0001 <0.0001 Tamarixitin OH OH OH OH OCH3 < 20 < 20

Acacetin OH OH OCH3 72* 72* 0.331 - Kaempferide OH OH OH OCH3 62 ND flavone vs flavonol Apigenin OH OH OH 51 75 0.361 0.024 (C-ring 3-OH) Kaempferol OH OH OH OH 42 55

Chrysoeriol OH OH OCH3 OH 52 78 0.051 0.081 Isorhamnetin OH OH OH OCH3 OH < 20 43 Chrysin OH OH 33 48 0.128 0.027 Galangin OH OH OH 20 37

Diosmetin OH OH OH OCH3 92 95 - <0.0001 Hesperetin OH OH OH OCH ND < 20 flavone vs flavanone 3 Apigenin OH OH OH 51 75 - <0.0001 Naringenin OH OH OH ND < 20 Apigenin OH OH OH 51 75 flavone vs isoflavone - <0.0001 Genistein OH OH OH ND < 20

Acacetin OH OH OCH3 72* 72* flavone vs. chalcone <0.0001 - 2',4'-dihydroxy-4-methoxychalcone OH OH OCH3 < 20 ND 7-hydroxyflavone OH ND < 20 - <0.0001 Chrysin OH OH 33 48

7-hydroxy-4'-methoxyflavone OH OCH3 < 20 < 20 A-ring 5-position <0.0001 0.002 Acacetin OH OH OCH3 72* 72*

4',7-dimethoxy-5-hydroxyflavone OH OCH3 OCH3 30 34 0.082 0.0004 5,7,4'-trimethoxyflavone OCH3 OCH3 OCH3 < 20 < 20 Chrysin OH OH 33 48 A-ring 6-position - 0.0009 Baicalein OH OH OH ND < 20

Acacetin OH OH OCH3 72* 72* <0.0001 0.017 4',7-dimethoxy-5-hydroxyflavone OH OCH OCH 30 34 A-ring 7-position 3 3 Apigenin OH OH OH 51 75 - 0.0002 Genkwanin OH OCH3 OH ND < 20

Diosmetin OH OH OH OCH3 92 95 <0.0001 <0.0001 Diosmin OH O-glucoside OH OCH3 < 20 < 20 Kaempferol OH OH OH OH 42 55 B-ring 2'-position - 0.002 Morin hydrate OH OH OH OH OH ND < 20

Acacetin OH OH OCH3 72* 72* 0.002 0.0005 Diosmetin OH OH OH OCH3 92 95 Apigenin OH OH OH 51 75 - <0.0001 Luteolin OH OH OH OH ND < 20

Kaempferide OH OH OH OCH3 62 ND 0.002 - Tamarixitin OH OH OH OH OCH3 < 20 < 20 B-ring 3'-position Apigenin OH OH OH 51 75 0.955 0.787 Chrysoeriol OH OH OCH3 OH 52 78 Chrysoeriol OH OH OCH OH 52 78 3 - <0.0001 Luteolin OH OH OH OH ND < 20 Kaempferol OH OH OH OH 42 55 0.003 0.358 Isorhamnetin OH OH OH OCH3 OH < 20 43 Kaempferol OH OH OH OH 42 55 - 0.002 Quercetin OH OH OH OH OH ND < 20 Diosmetin OH OH OH OCH 92 95 3 - <0.0001 Luteolin OH OH OH OH ND < 20 Chrysin OH OH 33 48 0.162 0.0008 Apigenin OH OH OH 51 75 Chrysin OH OH 33 48 0.0003 0.024 Acacetin OH OH OCH3 72* 72* Acacetin OH OH OCH 72* 72* 3 0.049 0.409 Apigenin OH OH OH 51 75 B-ring 4'-position Galangin OH OH OH 20 37 0.016 0.021 Kaempferol OH OH OH OH 42 55 Galangin OH OH OH 20 37 0.013 - Kaempferide OH OH OH OCH3 62 ND Kaempferide OH OH OH OCH 62 ND 3 0.094 - Kaempferol OH OH OH OH 42 55

4',7-dimethoxy-5-hydroxyflavone OH OCH3 OCH3 30 34 - 0.005 Genkwanin OH OCH3 OH ND < 20

Diosmetin OH OH OH OCH3 92 95 C-ring 3-position <0.0001 <0.0001 3,4'-dimethoxy-3',5,7-trihydroxyflavone OCH3 OH OH OH OCH3 < 20 < 20 *data from 5uM dosage ND - Not done due to inactivity of compound at higher dose or toxicity Supplemental Table 2.2 All informative comparisons of an active flavonoid and another flavonoid that differ by substitutions at a single position or by subclass. Percent importing cells at 10 and 20 µM is shown for compounds that rescued import in over 20% of cells. P-values represent comparison of percent importing cells between 2 compounds by Student’s two-tailed t- test.

2.8 CONNECTING TEXT BETWEEN CHAPTERS II AND III

To test potential ZSD therapies in vivo and to inform disease pathophysiology, we established a PEX1-G844D mouse model, which has the murine equivalent to the common human PEX1- G843D mutation. Prior to intervention, phenotypic characterization of this model is necessary for understanding the natural progression of disease in the mouse and establishing therapeutic endpoints. Presented in Chapter III are the results of an in-depth characterization of the retinal phenotype of the PEX1-G844D homozygous mouse.

Chapter III: Characterization of Retinopathy in the PEX1-Gly844Asp Mouse Model for Mild Zellweger Spectrum Disorder

Catherine Argyriou1, Anna Polosa2, Bruno Ceçyre3, Monica Hsieh4, Erminia Di Pietro4, Wei Cui4, Jean-François Bouchard3, Pierre Lachapelle2,4, Nancy Braverman1,4

1Department of Human Genetics, McGill University, Montreal, Canada 2Department of Ophthalmology, McGill University, Montreal, Canada 3School of Optometry, Université de Montréal, Montreal, Canada 4Montreal Children’s Hospital Research Institute, Montreal, Canada

Manuscript in preparation

3.1 ABSTRACT

Zellweger Spectrum Disorder (ZSD) is an autosomal recessive disease caused by mutations in any one of 13 PEX genes whose protein products are required for peroxisome assembly. Retinopathy leading to blindness is one of the major untreatable handicaps faced by patients with ZSD but is not well characterized, and the requirement for peroxisomes in retinal health is unknown. To address this, we examined the progression of retinopathy from 2- 32 weeks of age in our murine model for the common PEX1-G843D allele (PEX1-G844D) using electrophysiology, histology, electron microscopy, biochemistry, and behavioural tests. We found that retinopathy in the PEX1-G844D mouse model is marked by low cone cell function and number early in life, with gradually decreasing rod cell function. Structural defects at the inner retina occur later in the form of bipolar cell degradation (between 13 and 32 weeks), inner segment disorganization and enlarged mitochondria (32 weeks), while other inner retinal cells appear preserved. Visual acuity is diminished by 11 weeks of age, while signal transmission from the eye to the brain is relatively intact from 7 to 32 weeks of age. PEX1-G844D, PEX6, and PEX5 protein levels are normal in the PEX1-G844D mouse model, contrary to human PEX1- G843D cells. Finally, C26:0 lysophosphatidylcholine (lyso-PC, a sensitive measurement of very long chain fatty-acids) are elevated in the Pex1-G844D retina, while phopshoethanolamine plasmalogen lipids are present at normal levels. These characterization studies identify therapeutic endpoints for future preclinical trials, including improving or preserving ffERG response, improving visual acuity, and/or preventing loss of bipolar cells.

3.2 INTRODUCTION

Peroxisomes are ubiquitous, mostly spherical, single membrane-bound organelles numbering up to several hundred per mammalian cell. They are highly conserved amongst eukaryotes and indispensable for normal life, each containing over 50 matrix enzymes required for multiple metabolic pathways [reviewed in [1],[14]]. In mammals, 16 PEX proteins, encoded by PEX genes, regulate peroxisome biogenesis, which includes synthesis, assembly, matrix protein import, and division. Primarily biallelic mutations in any of 13 PEX genes result Zellweger Spectrum Disorder (ZSD), a heterogeneous group of disorders with a range of multi-system manifestations [9, 12, 173, 189]. Phenotypic severity in ZSD is associated with the level of residual PEX protein activity and consequent peroxisome function.

Mutations in PEX1 are the most common cause of ZSD, accounting for roughly 70% of known North American patients [9]. One mutation, PEX1-c.[2528G>A] encodes the missense protein PEX1-p.[Gly843Asp] (or G843D), and results in a milder ZSD phenotype without developmental defects, rather a progressive multi-system disorder due to ongoing peroxisome dysfunction over time [10, 185, 242]. The PEX1-G843D allele has a high frequency due to a founder effect in persons of North European ancestry, and represents roughly 30% of all ZSD alleles [9, 12, 13, 325].

PEX1 and its partner protein PEX6 are AAA ATPases (ATPases associated with various cellular activities) that form part of the peroxisome ‘exportomer’ complex. The PEX1-PEX6 complex is anchored to the peroxisome membrane by PEX26 [295] and uses the energy from ATP hydrolysis to ‘pull’ the PEX5 enzyme receptor from the peroxisome membrane so that it can be recycled for additional rounds of import [129, 131][reviewed in [296]]. If not recycled, PEX5 is targeted for proteasomal degradation [113], and enzyme import into the peroxisome is impaired.

Although the clinical presentation of ZSD is highly variable and can include brain, liver, adrenal, and bone involvements [reviewed in [1, 190]], nearly all ZSD patients develop sensorineural hearing loss and a progressive retinopathy leading to blindness. The natural history of ZSD retinopathy has not been systematically studied in these patients, and observations have been reported mainly through case reviews. Although these reports are limited by inadequate details of disease severity, PEX genotypes, and evolution of the retinopathy over time, they provide a

description of retinal findings in various patients. In severely affected patients, ophthalmic manifestations can include, in addition to retinopathy, also cataracts, glaucoma, and corneal clouding [reviewed in [326]]. Cataracts can develop with age in some patients [206]. Across the disease spectrum, however, posterior segment abnormalities, particularly photoreceptor loss, are the most severe of all ophthalmic manifestations [reviewed in [326]]. On the milder end of the spectrum, patients have been diagnosed with Usher Syndrome [210, 327, 328], and presentations are most typically characterized by posterior segment anomalies including retinitis pigmentosa, macular atrophy, reduced visual acuity, and reduced or extinguished electroretinograms [198, 329]. Additional manifestations include retinal arteriolar attenuation, optic nerve atrophy, nystagmus, and foveal thinning [212, 329][reviewed in [326]]. One ophthalmic review of a young adult homozygous for PEX1-G843D reported a progressive course of peripheral visual field loss, nyctalopia, and dyschromatopsia from childhood. At 25 years she had a visual acuity of 20/100 and 20/80 in her right and left eyes respectively, and generalized constriction of visual fields with a few preserved regions of vision in the mid-periphery [330]. Pigmentary retinopathy, degeneration of both fundi, optic nerve head drusen, and attenuated retinal vasculature were also present, as was unilateral macular edema that improved with a topical carbonic anhydrase II inhibitor (dorzolamide) [330]. Another PEX1-G843D homozygote presented with pigmentary retinopathy and at 19 months had severely reduced rod and cone function, as measured by full- field flash electroretinogram (ffERG). Optical coherence tomography at 6 years showed cystoid macular edema and absence of the photoreceptor layer in the macula in both eyes [328].

To begin understanding the course and mechanism of peroxisome-mediated visual deterioration, and to guide future therapeutic studies, we engineered a mouse model for mild ZSD, which has the PEX1-G843D equivalent knock-in mutation (PEX1-G844D). We previously showed that this model has diminished ffERG at 2-4 months, particularly in the cone-mediated system, and reduced cone photoreceptor staining at 3 weeks and 22 weeks [264]. There was decreased expression of the cone-specific genes Arr3, Pde6h, Gnat2, and Opn1mw, and increased expression of Cyp4a14, involved in the omega oxidation of fatty acids and Ufd1l, which plays a critical role in endoplasmic reticulum-associated degradation and cholesterol metabolism. No further ophthalmic characterization was performed, nor was the retinopathy followed over time. In this report, we have characterized the progression of PEX1-G844D mouse retinopathy with electrophysiology (ffERG and visual evoked potential, VEP), visual acuity (optokinetic

nystagmus, OKN), peroxisome protein levels, and biochemical metabolites. We also examined retinal structure and cell types using histology, immunohistochemistry, and electron microscopy. These studies established which functional and structural visual components were affected first and most severely, and which remained unaffected in the PEX1-G844D model.

3.3 MATERIALS AND METHODS

3.3.1 Animal husbandry

PEX1-G844D mice were maintained on a mixed 129/SvEv and C57BL/6N Taconic background. Strain background was evaluated yearly by SNP genotyping (MaxBax, Charles River) and showed a stable 70% 129/SvEv and 30% C57BL/6N Taconic. Mice were housed at the RI- MUHC Glen site animal care facility with ad libitum access to food and water. All experiments were performed at the RI-MUHC Glen site, except visual acuity measures were performed at the Pavillon Liliane-de-Stewart de l’Université de Montréal animal care facility. All experiments were approved by the Research Institute of the McGill University Health Centre Animal Care

Committee or the Université de Montréal Ethics Committee. Euthanasia was performed by CO2 asphyxiation under isoflurane anaesthesia (5% isoflurane in oxygen until loss of consciousness,

immediately followed by CO2 at maximum flow rate, 4 LPM). Both males and females were used for all experiments.

Genotyping: For routine genotyping, genomic DNA was isolated from ear punches of 21 day old mice by incubating in 75µl alkaline lysis buffer (25mM NaOH, 0.2mM Na2EDTA) at 95°C for 20 min, then neutralized with 250µl 40mM Tris-HCl. Genotypes were determined by PCR amplification (forward primer 5’-TCAATGTGTCCAGCACCTTC-3’; reverse primer 5’- TATGGAACGGAATGAGGC-3’) that produces amplicons of different sizes depending on the presence of a 177 bp residual Neo cassette fragment in intron 13 near the Pex1 c.2531G>A mutation in exon 15. The Pex1 c.2531G>A (Pex1 p.Gly844Asp) allele yields a 852 base pair product and the wild-type allele yields a 672 base pair product.

3.3.2 Electrophysiology

Full-field flash electroretinogram (ffERG): Retinal function was assessed using ffERG as previously described [331]. Briefly, following a 12 hours dark adaptation, mice were anesthetized [intraperitoneal injection of 10% ketamine, 5% xylazine solution in PBS

(130µL/10g body weight)] and their pupils were dilated [1% Mydriacyl Tropicamide (Alcon, Mississauga, CA)]. All experimental procedures were performed in a dark room under red light illumination. Two types of recordings were performed to assess rod and cone function: scotopic or dark-adapted ERG and photopic or light adapted ERG, respectively. A set of three electrodes was used to record both ERG types: a DTL fibre electrode (27/7 X-Static® silver coated conductive nylon yarn, Sauquoit Industries, Scranton, PA, USA) placed on the cornea and held in place with a moisturizing solution (Tear-Gel, Novartis Ophthalmic, Novartis Pharmaceuticals Inc, Canada), a reference electrode (Grass E5 disc electrode; Grass Instruments, Quincy, MA, USA) positioned in the mouth and a ground electrode (Grass E2 subdermal electrode; Grass Instruments, Quincy, MA, USA) inserted in the tail. The animal body temperature was maintained at 37°C (homeothermic heating blanket, Harvard Apparatus, Holliston, MA). Recordings of full-field ERGs (bandwidth: 1-1000 Hz; 10 000 X; 6 db attenuation; Grass P-511 amplifiers; Grass Instruments, Quincy, MA, USA) and oscillatory potentials (OPs) (bandwidth: 100-1000Hz; 50 000X; 6db attenuation; Grass P511 amplifiers; Grass Instruments, Quincy, MA, USA) were performed simultaneously with the Biopac data acquisition system (Biopac MP 100 WS, Biopac System Inc., Goleta, CA, USA). Scotopic ERGs and OPs were obtained in response to progressively brighter flashes of white light ranging in intensity from -6.3 log cd.s.m-2 to -1.5 log cd.s.m-2 in 0.9 log-unit increments, and from -1.5 log cd.s.m-2 to 0.9 log cd.s.m-2 in 0.3 log- unit increments [Grass PS-22 photostimulator (Grass Instruments; Quincy, MA, USA), interstimulus interval: 10 sec, flash duration 20 µs, average of 2-5 flashes depending on intensity]. Photopic ERGs and OPs were evoked to flashes of 0.9 log cd.s.m-2 (photopic background, 30 cd/m2, interstimulus interval: 1 sec, flash duration 20 µs, average of 20 flashes). Furthermore, in order to avoid the previously reported light adaptation effect, the photopic recordings were obtained 20 minutes following the opening of the background light [332, 333]. Mice were always tested within three hours of their daylight hour onset.

Quantification of the amplitude of the a-wave, b-wave and oscillatory potentials (OPs) was performed as previously described [334]. Briefly, the amplitude of the a-wave was measured from baseline to the most negative trough, while the amplitude of the b-wave was measured from the trough of the a-wave to the most positive peak of the ERG (scotopic ERG) and from the baseline to the highest peak of the b-wave (photopic ERG), while that of OPs by measuring each OP individually from trough to the peak and summing all amplitudes together.

Visual evoked potential (VEP): Retinocortical function was assessed by measuring the VEP as previously described [335]. An active electrode (Grass E2 subdermal electrode) was inserted subcutaneously over the occipital cortex (lambda stereotaxic coordinate as per [336]), while the ground and the reference electrodes remained in the same location as for the ffERG recordings. VEPs (bandwidth: 1–100Hz, 10 000X; 6db attenuation; Grass P-511 amplifiers) were evoked to flashes of 0.9 log cd.s.m-2 (interstimulus interval: 1 sec, flash duration 20µs, average of 100 flashes; background light at 30 cd.m-2). Only the left eye was exposed to light stimulus. Two components could be identified on the VEP tracings: a negative (N1) and a positive (P1). The N1 value was measured from baseline to the most negative trough, and the P1 value from N1 to the most positive peak. Latency was measured from flash onset to trough (N1) or peak (P1) and represents the time required for signal transmission from retina to occipital cortex. Concurrent photopic ffERG was recorded on all animals that underwent VEP recording.

3.3.3 Assessing visual acuity with the Virtual Optomotor System

The visual acuity of mice was determined using a virtual-reality optomotor system (CerebralMechanics, Lethbridge, AB) as previously described [289]. Briefly, freely moving mice were placed on an elevated platform and exposed to vertical sine wave gratings rotating at 12 degrees/second. The grating spatial frequencies increased from 0.01 to 0.5 cycles/degree in varying increments. When able to perceive the stimulus, the mouse normally stopped moving its body and began to track the grating with reflexive head movements in concert with the rotation. Spatial frequency of the grating at full contrast (100%) was gradually increased until the mice no longer exhibited a tracking behavior. The highest spatial frequency that could be followed (i.e. spatial frequency threshold) determined the visual acuity (in cycle per degree) for each eye. Mice were always tested within three hours of their daylight hour onset.

3.3.4 Lipid analysis

For liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis, retinas were homogenized in PBS using a mini pestle. 2:1 chloroform/methanol containing 0.05% butylhydroxytoluene (BHT) was added to 50 µg protein extract in a glass tube and incubated on an orbital shaker at room temperature for 2 hours. Samples were centrifuged at 2500 rpm for 10 min and the supernatant was transferred to a clean glass tube. The supernatant was washed with 0.2 volumes of purified water, mixed and centrifuged at 2000 rpm room temperature for 5 min to

80 separate the two phases. The upper phase was removed and the lower phase was washed with Folch theoretical upper phase (3:48:47 chloroform:methanol:water). Samples were mixed and centrifuged at 2000 rpm for 5 min and the upper phase was removed. The lower phase was dried under nitrogen and then under vacuum in a desiccator for 30 min. The dried lipid was dissolved in 3:2 hexane:isopropanol containing 10 ng each of the internal standards, 16:0-D4 lyso-PAF (20.6 pmol) and D4-26:0-lyso-PC (15.6 pmol). Samples were filtered by centrifugation (Corning (Corning, NY) Costar spin-X centrifuge tube filters) for 5 minutes. Filtrates were analyzed in Verex auto-sampler vials (Phenomenex, Torrance, CA). A 2.1 X 50 mm, 1.7 µm chromatography column and a Waters (Milford, MA) TQD (Triple Quadrupole Mass Spectrometer) interfaced with an Acquity UPLC (ultra-performance liquid chromatography) was used in positive ion electrospray (ESI)-MS/MS ionization. The solvent systems were: mobile phase A = 54.5% water/45% acetonitrile/0.5% formic acid, mobile phase B = 99.5% acetonitrile/0.5% formic acid with both solutions containing 2 mM ammonium formate. Injections of extracts dissolved in 3:2 isopropanol/hexane were made with initial solvent conditions of 85% mobile phase A/15% mobile phase B. The gradient employed was from 15% to 100% mobile phase B over a period of 2.5 min, held at 100% mobile phase B for 1.5 min before reconditioning the column back to 85% mobile phase A/ 15% mobile phase B for 1 min at a solvent rate of 0.7 ml/min. A column temperature of 35 °C and an injection volume of 5 µl for plasmalogen and 10 µl for lysoPC were used for analysis. Ethanolamine plasmalogens were detected by multiple reaction monitoring (MRM) transitions representing fragmentation of [M+H]+ species to m/z 311, 339, 361, 385, 389, 390 for compounds with 16:1, 18:1, 20:4. 22:6, 22:4 and 18:0, at the sn-2 position, respectively. Lysophosphatidylcholine (LysoPC) species were detected by multiple reaction monitoring (MRM) transitions representing fragmentation of [M+H]+ species to m/z 104. Reagents used were authentic plasmalogen standards, tetradeuterated internal standards 26:0-D4 lysoPC (Avanti Polar Lipids, Alabaster, Alabama), 16:0-D4 lyso PAF (Cayman Chemical Company, Ann Arbor, Michigan) and HPLC grade solvents (methanol, acetonitrile, chloroform, water) (Fisher Scientific, Waltham, MA), formic acid (Honeywell Fluka), ammonium formate (Sigma-Aldrich), and PBS (Thermo Fisher Scientific, Waltham, MA).

3.3.5 Immunoblotting

Mouse retinas were homogenized in IGEPAL lysis buffer. Lysates (20 μg) were separated on 7.5% polyacrylamide gel and transferred to nitrocellulose membrane. Membranes were blocked and hybridized in 5% milk with primary antibodies: 1:1000 rabbit anti-human PEX1 (Proteintech 13669-1-AP), 1:2000 rabbit anti-human PEX5, PEX6 (gifts from Dr. Gabriele Dodt, University of Tübingen), 1:17000 rabbit anti-human β-tubulin (Abcam, ab6046), followed by appropriate HRP-conjugated secondary antibodies, and visualized by ECL using an Amersham 600 platform.

3.3.6 Retinal histology and immunohistochemistry

Retinal histology: was performed as previously described [331]. Briefly, following overnight fixation in 4% paraformaldehyde, eyes were embedded in Epon resin. 1µm sections cut at the level of the optic nerve head were collected on glass slides and stained with 0.1% Toluidine Blue. Samples were visualized using a Zeiss CZC0M0042 microscope and images captured using a Zeiss AxioCam MRc camera. Thickness of retinal layers was measured using AxioVision software (version 4.8.2.0; Carl Zeiss Microscopy GmbH).

Immunohistochemistry: Eye cups were fixed 3 hr in 4% paraformaldehyde, incubated in 10% (30 min on ice), 20% (1 hr on ice), and 30% (4°C overnight) sucrose in 0.1M PB, then embedded and frozen in frozen section compound (VWR). 5 µm retinal cryo-sections were blocked (1% NGS, 0.1% triton, 10% BSA in PBS) for 1 hr, washed, incubated at 4°C overnight with primary antibody in incubation buffer (0.1% triton, 10% BSA in PBS), washed, incubated 90 min with secondary antibody and washed. Coverslips were mounted using ProLong Gold antifade reagent with DAPI (Invitrogen) and retinas were visualized using a Leica DMI600 microscope with DFC345FX camera and LASX software. Primary antibodies used were 1:450 rabbit anti-human PEX1 (Proteintech 13669-1-AP), 1:50 goat anti-human PEX6 (Abcam ab175064), 1:300 rabbit anti-human PEX14 (gift from Paul Watkins, Johns Hopkins University), 1:200 rabbit anti-human glial fibrillary acidic protein (Sigma G9269), 1:200 rabbit anti-human calbindin D-28K (Sigma C9848), 1:300 mouse anti-human rhodopsin (Phosphosolutions 1840-RHO), 1:200 rabbit anti- human synaptophysin (Abcam ab14692), 1:300 rabbit anti-human cone arrestin (Millipore AB15282), 1:150 mouse anti-human glutamine synthetase (Millipore MAB302), 1:300 rabbit anti-human PKC-α (Abcam ab32376), 1:200 rabbit anti-human parvalbumin (Abcam ab11427). Secondary antibodies used: 1:400 anti-rabbit 488 (Jackson ImmunoResearch Laboratories 711-

095-152),1:300 anti-goat Texas Red (Invitrogen PA1-28662), 1:300 anti-mouse Texas Red (Invitrogen T-862), 1:100 anti-mouse 488 (Invitrogen A32723).

3.3.7 Transmission electron microscopy

Mice were perfused with 0.1M sodium cacodylate buffer (Electron Microscopy Sciences, EMS) containing 0.1% calcium chloride (Bioshop), 2.5% glutaraldehyde (EMS), 2% paraformaldehyde (Fisher). Eyes were fixed in 2.5% glutaraldehyde in 0.1M sodium cacodylate buffer for 1 hr; lens were removed and eye cups were quartered, washed, then post stained with 1% osmium+1.5% potassium ferrocyanide (both EMS), dehydrated with increasing concentrations of acetone (Fisher), infiltrated with increasing concentrations of Epon (EMS), then polymerized with pure Epon at 60°C for 48 hrs. 100 nm sections were cut with a Leica UCT ultramicrotome, placed onto a 200 mesh copper grid, post stained with uranyl acetate and Reynold's lead, and imaged with the FEI Tecnai 12 BioTwin TEM equipped with the AMT XR80C CCD camera at an accelerating voltage of 120 kV.

3.4 RESULTS

3.4.1 Assessing retinal function

To assess retinal electrophysiological response over time, full field flash electroretinograms (ffERG) were performed in PEX1-G844D and non-mutant littermate control mice from eye opening (2 weeks) to 32 weeks of age. ffERG responses at maximum stimulus intensity are shown in Figures 3.1A and 3.1B. At 2 weeks of age, the scotopic retinal response in Pex1- G844D mice was low but recordable, averaging 16 ± 9 µV and 20 ±7 µV for a- and b-waves, respectively, 24% and 17% of average control values. The maximal scotopic b-wave response (on average 45-60% of control) was reached at 4 to 6 weeks, and then progressively decreased with age, falling to 10% of control at 32 weeks. Although the control scotopic response also peaked around 6 weeks, it did not diminish with time, remaining consistent to 32 weeks (166 ± 69 µV versus 176 ± 27 µV for a-wave, and 447 ± 177 µV to 464 ± 63 µV for b-wave, respectively). The scotopic a-wave in PEX1-G844D mice was less affected than the b-wave. Although low at 2-3 weeks, the a-wave value peaked at 60-80% of control at 4-6 weeks, with some animals overlapping controls, then gradually declined to 15% of normal at 32 weeks. Although the scotopic a-wave was relatively better preserved than the b-wave in mutants

compared to controls, the degree of decline of the two waves in the mutants from 4 to 32 weeks was not significantly different (t-test comparing the ratio of a-to b-wave decline from 4 to 32 weeks). The photopic response was not recordable in PEX1-G844D mice until 4 weeks of age. The average photopic response at 4 weeks was less than 5% that of controls, and remained low with age, never surpassing 25% of normal. Peak latency measurements revealed normal a- and b- wave peak times in mutants in both photopic and scotopic conditions.

Despite a marked decline in b-wave amplitude, oscillatory potentials (OPs) remained prominent in PEX1-G844D mice. A representative waveform is shown in Figure 3.2A. To quantify this effect, the average ratio of the sum of OP amplitudes to the b-wave amplitude was evaluated (Figure 3.2B). This ratio was significantly higher in PEX1-G844D mice compared to controls at 4, 10-11, and 21-32 weeks. The number of OPs was unchanged between mutants and controls at all ages.

3.4.2 Assessing the primary visual pathway (from the retina to the visual cortex)

To assess signal transmission from the retina to the visual cortex, photopic visual evoked potentials (VEPs) were recorded at 7 and 32 weeks. Different animals were used in each age cohort. Representative VEP recorded from control and mutant mice at 7 and 32 weeks of age are shown in Figure 3.3A. Photopic VEP amplitude in PEX1-G844D mice was on average 34% that of age-matched controls at 7 weeks (110 µV versus 327 µV), and 53% that of age-matched controls at 32 weeks (121 µV versus 229 µV) (Figure 3.3B). Although the average VEP amplitude in control mice was lower in the older compared to younger age group (229 µV versus 327 µV, respectively), there was no difference between age groups in PEX1-G844D mice (110 µV versus 121 µV) (Figure 3.3B). In PEX1-G844D mice, the average peak time at 7 weeks was 148 ms, delayed by 32 ms from the average control peak time of 116 ms (Figure 3.3B). This delay increased to 50 ms in the 32-week old mice, in which mutants had a peak time of 177 ms, compared to the control average of 127 ms. Although delayed, signal incorporated at the retinal level was proportionally transduced to the visual cortex in PEX1-G844D mice, as VEP amplitude correlated to photopic ffERG b-wave amplitude (Figure 3.3C). That is, the higher the photopic ffERG response, the greater the VEP, and vice versa.

Visual acuity was determined with the optomotor reflex evoked at gradually increasing spatial frequencies at 100% contrast. PEX1-G844D mice at 11-13 weeks age had diminished visual acuity, on average 10% that of littermate controls, and ranging from 0 to 33% of normal (Figure 3.3D). This technique yields independent measures of right- and left-eye acuity, as only motion in the temporal-to-nasal direction evokes a tracking response. For each animal, there was little variation between eyes, and so visual acuity was reported using the average of the two values. A longitudinal study of visual acuity has not yet been done.

3.4.3 Retinal structure

To inform the pathophysiology of retinal dysfunction in PEX1-G844D mice, histology was performed to identify any structural defects at 3, 6, 10, 28, and 32 weeks. Retinal sections stained with Toluidine blue revealed apparently normal presence and thickness of all retinal layers across this age range (Supplemental figure 3.1). As no significant histological changes were observed using light microscopy, immunohistochemistry was then used to visualize various cell types at 6, 13, and 32 weeks of age. As seen in Figure 3.4, results obtained in our mutants revealed a normal immunoreactivity for rod cell outer segments (rhodopsin), horizontal cells (calbindin), amacrine cells (parvalbumin), Müller cells (glutamine synthetase), and all synaptic layers (synaptophysin).

In contrast, antibodies against cone arrestin, which specifically detects cone cells from the outer segment (OS) to the synaptic terminal, revealed an altered cone structure in PEX1-G844D mice (Figure 3.5A). In addition to significant cell dropout, most of the outer (OS) and inner (IS) segments were absent, disorganized, or disrupted. There was a drastic synaptic retraction of cone axons and pedicles, resulting in an almost complete loss of contact with the inner nuclear layer (INL) cells, with some cone photoreceptors only exhibited staining of their nucleus. This pattern of morphological alterations was observed as early as 6 weeks of age and remained similar at 32 weeks of age. Similar changes were observed in the rod ON bipolar cells, which were labeled with antibodies against protein kinase C alpha (PKC-α). Although in the early phase of the retinopathy (6 and 13 weeks of age), PKC-α immunoreactivity was relatively normal, by 32 weeks, fewer ON rod bipolar cells were stained in mutants compared to controls (Figure 3.5B). Some of these remaining cells migrated towards the center of the inner plexiform layer (IPL) (Figure 3.5B; white arrows). In addition, both bipolar cell dendritic terminals (which establish

85 connections with rod spherules in the outer plexiform layer, OPL) and bipolar end-bulb axon terminals (which extend into the IPL connecting with retinal ganglion cell layer, RGC) were significantly reduced. Finally, immunolabelling for the Glial Fibrillary Acidic Protein (GFAP), a gliosis marker expressed by activated Müller cells, was present in Pex1-G844D retinas only at 32 weeks of age (no GFAP-positive staining was evidenced at 6 or 13 weeks of age) (Figure 3.6). GFAP staining was mainly confined to the RGC and the retinal fiber layer, but lighter staining was also present in some radial processes of Müller cells.

Transmission electron microscopy was used to visualise retinal ultrastructure from 32-week-old Pex1-G844D and control mice. Photoreceptor outer segments appeared to have normal density and opsin organization (Figure 3.7B). However, photoreceptor connections from outer to inner segments were not distinguishable in the mutants, in contrast to the linearly organized inner segments seen in controls (Figure 3.7A). This could imply that the mutant inner segments were either disorganized or degenerated to the point of non-detection. In addition, mitochondria were dramatically enlarged throughout Pex1-G844D retinas, ranging from one to five times the size of those in controls. This was particularly remarkable in the mitochondria-rich inner segment (Figure 3.7C). No other defects were visible in the remaining cellular layers.

3.4.4 Peroxisome protein and biochemical metabolite levels

Immunoblotting was performed to compare the amounts of Pex1-G844D to wild-type Pex1 in the retina. There were normal amounts of Pex1-G844D, its binding partner Pex6, and its putative ligand Pex5 (the peroxisome enzyme receptor) protein in mutant compared to control retinas, suggesting that the mutated protein is not degraded (Figure 3.8A). This was consistent from 7 to 32 weeks of age. Immunohistochemistry showed localization of Pex1-G844D and Pex6 at the photoreceptor inner segment and outer plexiform layer, similar to their localization in the control mice (Figure 3.8B), with slightly lower Pex1-G844D at the OPL. The peroxisome membrane protein PEX14, enriched at the inner segment, was also normally localized in the Pex1-G844D retina through 32 weeks.

PEX mutations disrupt peroxisome metabolism, resulting in an accumulation or deficiency in the biochemical metabolites of these pathways. The affected pathways and extent of these effects vary by tissue. Two well-known peroxisome-mediated processes are very long-chain fatty acid

(VLCFA) β-oxidation and plasmalogen synthesis [reviewed in [14]]. Disruption in these pathways results in increased C26:0 VLCFA and decreased plasmalogens, and these metabolites are commonly measured for ZSD diagnosis [241]. In the Pex1-G844D retina, C26:0 VLCFAs (measured as C26:0 lyso-PCs) were elevated on average four-fold compared to average wild- type control levels, and ranged from normal to seven-fold wild-type levels (Figure 3.8C). Total phosphoethanolamine (PE) plasmalogen levels in Pex1-G844D retinas matched wild-type controls (Figure 3.8C). Levels of individual PE-plasmalogen sub-species (sn1- C16:0, C18:0, C18:1, and sn2- 16:1, 18:0, 18:1, 20:4, 22:4 22:6) measured were also no different from wild- type (not shown).

3.5 DISCUSSION

3.5.1 Visual response and retinal architecture

The electroretinogram in PEX1-G844D mice is markedly diminished, with b-waves affected earlier and more severely than a-waves (Figure 3.1A, 3.1B). As a-waves result from the hyperpolarization of photoreceptors [278, 279], the relative a-wave preservation in our model is consistent with its preserved OS organization (Figure 3.4, 3.7A, 3.7B). B-waves result from the transfer of signal through the photoreceptor axon to the inner retina, inducing depolarization of inner retinal cells [281]. As bipolar cells, and to a much smaller extent Müller cells, generate the b-wave [282, 283], it is also consistent that b-wave response declines with age in our model, correlating with the decline in bipolar cell numbers (Figure 3.5B). Synaptophysin staining appears normal in Pex1-G844D; however as most retinal cell types are not decreased in our model, this staining may not distinguish a subtle change in only one connection type. Increased GFAP staining occurs commonly in retinopathies and is associated with photoreceptor degeneration, increased oxidative stress, and/or Müller cell de-differentiation [reviewed in [337, 338]]. Its appearance only at an older age in our model (Figure 3.6) may be the result of ROS accumulation over time. Considering that Müller cell staining in not reduced in PEX1-G844D mice, GFAP staining is likely not indicating Muller cell de-differentiation (Figure 3.4).

Scotopic response is better preserved than photopic response in PEX1-G844D mice, at least earlier in life, suggesting a greater impairment of the cone versus rod mediated visual pathway. This is corroborated by retinal immunohistochemistry, in which the rod outer segment (rhodopsin staining) remains intact in the mutants, while cone numbers are diminished, even at

early age (arrestin staining) (Figure 3.4, 3.5A). However, rhodopsin antibodies only mark the outer segment portion of the photoreceptor cell. Ultrastructural analysis revealed that although the rod OS and nuclear layers appear intact, the photoreceptor inner segment portion is disorganized or degenerated at 32 weeks (Figure 3.7A).

High frequency ERG components (i.e. OPs) are better preserved than low frequency ERG components (i.e. a- and b-waves) in PEX1-G844D mice (Figure 3.2B). Although their exact source is still debated, OPs originate in the inner retina and are considered the result of a negative feedback of bipolar on amacrine cells [284]. Taken together, the OP prominence and preserved parvalbumin staining in Pex1-G844D mice suggests that amacrine cell function, and thus synapses, are preserved. This relative OP preservation could be due to decreased inhibition of OP-generating amacrine cells (by bipolar cells) with age and retinopathy progression. Alternatively, the OPs in our model could become more prominent with age due to the loss of b- wave generating bipolar cells. As some of the OPs are typically camouflaged by a large b-wave, they may become more distinguishable with b-wave diminishment. In interpreting our findings it is important to note that electrophysiology measures a pan-retinal average of function, while IHC offers a localized structural view of one retinal section, and that the two may not always directly correlate.

Visual acuity measured by optokinetic nystagmus was diminished by 11-13 weeks in the Pex1- G844D mice. The OptoMotry system operates in an ambient light environment, thus depending on the photopic pathway to transduce the visual stimulus at the retinal level. Given the cone deficit in our model, greatly reduced vision at this age and these conditions is unsurprising.

The positive correlation of VEP to photopic ffERG amplitude indicates complete, or near complete retinocortical function (Figure 3.3C). The delayed VEP response time in PEX1- G844D mice complements the delayed VEP peak time reported in two ZSD patients, one of which was homozygous for PEX1-G843D, and the other compound heterozygous for two point mutations in PEX6 [198]. In this same report, VEPs from six additional PEX1-G843D patients and two patients with different point mutations in PEX1 or PEX2 were either normal or had latent peaks. We hypothesize that the delayed VEP response in our mice could be due to optic nerve pathology. Imaging mass spectrometry of the human eye shows enrichment of peroxisome-dependent PE plasmalogens, a class of ether-phospholipids, at the optic nerve [339],

as well as enrichment in PE plasmalogen in the nerve fibers passing through the lamina cribrosa, an unmyelinated region of the optic nerve head [340]. The delayed VEP in our mice could be consistent with reduced or otherwise abnormal optic nerve myelination, as plasmalogen lipids constitute around 80% of the glycerophosphlipid portion of myelin and the committing steps of their synthesis occur in peroxisomes [reviewed in [45]]. Potential optic nerve pathophysiology could be addressed by lipidomic analysis and looking at the thickness of myelin along the optic nerve tracts using histology.

Alternatively, the deficit resulting in delayed VEP peak time could be at the visual cortex level. In a different peroxisome disease model, selective fetal Pex5 inactivation in mouse neural cells caused mild neurodevelopmental defects with progressive motor and cognitive impairment, progressive compaction and loss of myelin in axons and glial cells, and axonal damage [260]. Similarly, although myelin with reduced plasmalogen and elevated VLCFA levels was initially formed in mice with fetal Pex5 inactivation in oligodendrocytes, they also developed axonal damage and demyelination [261]. Thus, peroxisomes may not be required for initial myelination, but the preservation of axonal integrity and maintenance of myelin. Although brain architecture in the PEX1-G844D model looks normal, as observed by hematoxylin and eosin stating at 15 weeks (personal correspondence with Dr. Steven Steinberg, Kennedy Krieger Institute, Baltimore, MD) this should be re-examined with specific myelin staining over time.

Electron microscopy revealed enlarged mitochondria in the PEX1-G844D retina (Figure 3.7C). Besides sharing fission proteins and membrane material [341], peroxisomes and mitochondria share extensive signalling interplay, and the concerted effort of these organelles is required for several metabolic processes [reviewed in [342]]. In the context of peroxisome dysfunction, mitochondria could secondarily be affected by altered lipid composition of their membranes and increased ROS [342]. Thus, the mitochondrial enlargement observed could be a compensatory response to peroxisome deficiency. Mitochondrial defects have been reported in severe cases of ZSD, in which myofibrillar mitochondria exhibit ultrastructural and functional defects, and liver mitochondria were reduced and dense, with irregular crystae and dilated intercristae space [5, 343]. Enlarged mitochondria with increased oxidative stress were observed in Pex13 null mouse brains [344]. In a liver selective Pex5 null mouse model, morphologically abnormal mitochondria with decreased respiratory chain complex activity were observed. However,

oxidative damage was absent, likely compensated by increased glycogen breakdown and glycolysis, also observed in Pex5 null liver tissue [262]. However, it is of note that these null models represent a more severe ZSD manifestation than that of patients homozygous for PEX1- G843D or the PEX1-G844D mouse model. Nonetheless, secondary mitochondrial dysfunction has been observed in mice and humans with various non-etiologically related retinal degenerative diseases [reviewed in [345]], and a possible mitochondrial contribution to PEX1- G844D mouse pathology warrants further study.

3.5.2 Peroxisome dysfunction and the retina

Protein levels of PEX1-G844D, its partner protein PEX6, and its ligand PEX5, are normal in the PEX1-G844D retina (Figure 3.8A), indicating the missense protein is present, but subfunctional. This is contrary to patient fibroblasts, in which PEX1-G843D protein amounts are degraded to 5- 15% of normal PEX1 levels [11], with equivalent reductions in PEX6, with which PEX1 forms a hetero-oligomeric complex. PEX5, which becomes polyubiquitinated and targeted for degradation, is also reduced [11, 290]. Whether the PEX1-G843D-PEX6 complex is degraded while in the cytosol, or when at the peroxisomal membrane is unknown. It may be that the structure of mouse PEX1-G844D and the murine PEX1-PEX6 complex is more stable and thus less degraded than its human counterpart, or that the murine protein degradation process is more tolerant of certain structural changes. Normal PEX5 levels in the presence of reduced function of the PEX1-PEX6 complex would support reduced PEX5 polyubiquitinaiton and subsequent degradation. It could be that pexophagy in mice is mediated differently than it is in humans, in which disruptions in the PEX1-PEX6-PEX26 complex trigger autophagic degradation of the organelle [171].

PEX1, PEX6, and PEX14 were found throughout control and PEX1-G844D retinas, as expected for peroxisomal biogenesis proteins, with PEX14 more concentrated at the inner segment (Figure 3.8B), and PEX1/PEX6 enriched at the inner segment and inner plexiform layer. Our results corroborate the localization of PEX1, PEX6, and PEX14 in normal mouse retina observed by Smith et al. [346]. In both mouse and human retina, PEX6 was observed by immunofluorescent and electron microscopy to concentrate between the outer and inner segment, localized predominantly distal to the connecting cilia [210]. As the OS requires continuous regeneration due to light exposure [347], it could depend more heavily than the other retinal

layers on peroxisome-dependent processes enabling high lipid and free radical turnover. Moreover, docosahexanoic acid (DHA) is synthesized in the peroxisome after β-oxidation of its precursor [15, 16, 31], and is integral for OS membrane formation, phototransduction, and oxidant damage protection [33, 348]. Even so, peroxisomes are ubiquitous and found across neuronal cell types [reviewed in [349]], and given the high metabolic demand of not only the photoreceptors, but also the outer and inner plexiform layers [350-353], the importance of peroxisome function in the retina is likely not limited to a single layer.

Imaging mass spectrometry allows for differentiating the lipid content of the various retinal layers, each found enriched in different peroxisome-dependent phosphatidylcholine (PC) and phosphatidylethanolamine (PE) plasmalogen species in mice [354, 355]. In human tissue, the DHA-containing PE plasmalogen (PE18:0/22:6) was found only in the inner retina, while the DHA-containing PC and PE lipids were exclusive to photoreceptors [339]. In whole retina from our PEX1-G844D model, total PE plasmalogens were not decreased (Figure 3.8C). PC plasmalogen levels were relatively low in mouse retina, but no different between controls and mutants (not shown). This suggests that at least at a whole-retina level, the residual function of the PEX1-G844D protein is sufficient for the committing peroxisomal enzymatic steps in plasmalogen synthesis. The initial β-oxidation steps of very long-chain fatty-acid (VLCFA ≥C22) catabolism occur in the peroxisome, resulting in an accumulation of these molecules in the presence of peroxisome deficiency. VLCFA-containing PC lipids, though scarce in retina, were detectable in normal mouse photoreceptors [339]. In our PEX1-G844D model, we detected elevated C26:0 lyso-PCs in the retina (Figure 3.8C), though the functional effect of this is unknown. It is important to note that there are many other peroxisome-mediated lipid species not measured by our methods that could contribute to PEX1-G844D pathology.

3.5.3 Clinical implications: The translatability of the model

The Pex1-G844D mouse model for mild ZSD exhibits an early onset cone deficit and progressive rod degeneration, with functional impairment precluding some cellular events. Taken together, the retinal electrophysiological and structural observations in this model suggest that the most important structural defect is at the outer retina, specifically at the photoreceptor level, with degeneration at the inner retina occurring later. Our characterization complements the progression seen in the equivalent human disease, in which patients can exhibit photoreceptor

loss, primarily in the cone-rich macular region, as reported in the case of a six-year-old homozygous for PEX1-G843D [328]. In the mouse model, cone cell response is decreased earlier and relatively more than rod response. This contrasts with one reported PEX1-G843D homozygous patient, in which ffERG at age 14 demonstrated “markedly subnormal scotopic responses, with minimally subnormal photopic responses” bilaterally, suggesting greater rod than cone involvement in the retinal degeneration [330]. Although this discrepancy warrants further investigation, it is important to consider the anatomical differences between rodents and humans, such as the absence of a macular region in the mouse. In Stargart macular dystrophy 3, for instance, cones are predominantly affected in humans, while mice with equivalent mutations manifest a rod pathology, possibly due to different biochemical requirements in rodent versus human photoreceptors [356]. Further patient characterization will enable a more accurate picture of natural history and variation within the human ZSD disease.

Our in-depth retinal phenotyping in the Pex1-G844D mouse model provides us accurate therapeutic endpoints for preclinical trials. These endpoints include improving or preserving ffERG response, and preventing bipolar cell diminishment. The progressive scotopic ffERG decline suggests a therapeutic window at or before the peak age of 4-6 weeks, during which this decline could be prevented. Although cone cell numbers are low at early age, their function could be improved by rescuing peroxisome deficiencies, and improvement measured by photopic ffERG and visual acuity.

Visual impairment is a major, currently untreatable handicap for ZSD patients, and alleviating this burden would enable improved communication, learning, and mobility. In more mildly affected patients, as in the case of PEX1-G843D, this effect could extend to autonomy and employment. Improvement in peroxisome function, particularly at an early disease stage, could slow, or even prevent, retinopathy progression and cellular deterioration. Small molecule compounds such as the chemical chaperones betaine and arginine have been shown to improve peroxisome functions in PEX1-G844D patient fibroblast lines [290, 291], and could be tested in vivo using the Pex1-G844D mouse model. In addition, the Pex1-G844D retina is an ideal candidate for gene augmentation therapy, which could also be tested for functional efficacy in this model.

3.6 ACKNOWLEDGMENTS

We would like to acknowledge Dr. Heidi McBride and Dr. Eric Shoubridge for their helpful discussions of the mitochondrial phenotype in the retinal EM images. We also acknowledge the patients and families affected by ZSD and The Global Foundation for Peroxisomal disorders for inspiring and encouraging us to continue researching these conditions. This work was funded by the Canadian Institutes of Health Research to NB (CIHR 12610) and NB and PL (CIHR 34575) and Fonds de recherche du Québec – Santé to CA (FRQS doctoral training award 32105).

3.7 CONFLICTS OF INTEREST

There are no conflicts of interest.

Figure 3.1. ffERG maturation in PEX1-G844D mice. (A) Representative waveforms show diminished scotopic and photopic retinal electrophysiological response in PEX1-G844D homozygous mice compared to controls from 2 to 32 weeks of age. (B) Quantification of maximum ffERG response shows that scotopic a- and b-wave amplitudes peak at 4-6 weeks then gradually diminish, while photopic b-waves remain consistently low. B-waves are affected earlier and more severely than a-waves. Error bars indicate standard deviation; Student’s t-test *=P<0.05, **=P<0.01, ***=P<0.001; n=4-7 per group.

Figure 3.2. Prominence of oscillatory potentials (OPs) in ffERGs of PEX1-G844D mice. (A) Representative waveforms show that OPs are preserved in PEX1-G844D mice, despite diminished a- and b-wave amplitude. (B) The sum of ascending and descending OP amplitudes divided by the b-wave amplitude was calculated for each ffERG. Starting at 4 weeks, this ratio is greater in PEX1-G844D mice than controls, and highest at the oldest ages tested (21-32 weeks), indicating that OPs are better preserved over time than the b-wave. Error bars indicate standard deviation; Student’s t-test *=P<0.05, ***=P<0.001; n=4-7 per group.

Figure 3.3 Visual evoked potential (VEP) and visual acuity in PEX1-G844D mice. (A) Representative photopic VEP waveforms show decreased but variable VEP at 7 and 32 weeks. (B) Quantification of photopic VEP amplitudes shows that although diminished, the VEP of PEX1-G844D mice remained consistent from 7 to 32 weeks, while control VEP decreased with age. PEX1-G844D VEP peak time was delayed compared to controls at 7 weeks and more so at 32 weeks. (C) VEP amplitude is positively correlated to photopic ffERG b-wave amplitude in PEX1-G844D mice (linear regression analysis, P=0.02). (D) Visual acuity was measured by determining the spatial frequency threshold of the optomotor reflex. Visual acuity was severely diminished in PEX1-G844D mice. Each mouse was tested twice and the average visual acuity from both eyes per test was taken. Each point represents 1 animal; Student’s t-test *=P<0.05, **=P<0.01, ***=P<0.001.

Figure 3.4. Immunohistochemistry shows preservation of various cell structures in PEX1- G844D retinas. Retinal immunohistochemistry reveals normal immunoreactivity of the rod cell outer segment (rhodopsin), amacrine cells (parvalbumin), horizontal cells (calbindin), Müller cells (glutamine synthetase), and synaptic layers (synaptophysin) in Pex1-G844D mice compared to littermate controls. Retinas were assessed at 6, 13, and 32 weeks of age and representative images shown (n=3).

Figure 3.5. Immunohistochemistry shows decreased staining of cone and bipolar cells in Pex1-G844D retinas. (A) Cone cell numbers (arrestin) are decreased from 6 to 32 weeks in Pex1-G844D mice compared to littermate controls (white arrows) (B) ON bipolar cells (PKC-α) were present at normal numbers at 6 and 13 weeks, but diminished and displaced into the IPL by 32 weeks in Pex1-G844D mice compared to littermate controls (white arrow). Representative images shown at each age point (n=3).

Figure 3.6. Prominence of glial fibrillary acidic protein with age in PEX1-G844D retinas. Retinal immunohistochemistry against GFAP is absent at younger ages and appears by 32 weeks in Pex1-G844D retinas, but not in littermate controls. Representative images shown (n=3).

Figure 3.7. Transmission electron microscopy of PEX1-G844D retinas. (A) Photoreceptor connections from outer (OS) to inner segments (IS) were not distinguishable in the mutants (1900x magnification). (B) Photoreceptor outer segments had normal density and opsin organization (11,000x magnification). (C) Mitochondria were dramatically enlarged throughout Pex1-G844D retinas, particularly evident in the mitochondria-rich inner segment (11,000x magnification). Findings were consistent in 3 mutant and 3 control retinas analyzed from 32- week-old animals.

Figure 3.8. Peroxisome protein and biochemical metabolite levels in PEX1-G844D retinas. (A) Immunoblotting of retinal lysates shows that PEX1-G844D protein is not degraded, and present at levels equal to that of wild-type PEX1 in controls. Pex6 and Pex5 levels are also normal. Both retinas from each mouse were pooled (n=3). Representative image from 32 week old mice is shown, though the observation was consistent from 7 to 32 weeks. (B) Immunohistochemistry showed normal localization of PEX1-G843D (primarily at the inner segment and outer plexiform layer), and Pex6 in the mutant compared to control retinas. (C) Levels of VLCFA (C26:0 lyso-PC) and ethanolamine plasmalogens (PE) were measured by LC/MSMS in mouse retinas. Average C26:0 lyso-PC levels were increase nearly four-fold compared to controls. Total PE plasmalogens levels were unchanged. Results from two separate experiments were pooled by normalizing all values against the control average for each experiment. Each point represents one whole retina; Student’s two-tailed t-test, ** P < 0.01.

100

Supplemental figure 3.1. Retinal histology. Retinal sections stained with Toluidine blue show preserved retinal layers and thickness in PEX1-G844D retinas compared to controls. Samples were assessed at 3, 6, 10, 28, and 32 weeks. Representative images are shown at 6 and 32 weeks.

101

3.8 CONNECTING TEXT BETWEEN CHAPTERS III AND IIV

As a monogenic disease, PEX1-mediated ZSD is a fitting candidate for gene augmentation therapy. This approach is based on expressing a functional gene in a tissue deficient for this gene, thus curing the disease. The compartmentalized structure of the eye makes it an ideal organ for gene therapy. It is accessible, small, and immune-privileged via the blood-retina barrier. These characteristics allow for the use low dose vector administration, and limit systemic spread and immune reactivity. For these reasons, I chose to target the PEX1-G844D retina using an adeno-associated vector to deliver PEX1. The results of these experiments are presented in Chapter IV.

102

CHAPTER IV: rAAV-mediated PEX1 Gene Augmentation Improves Visual Function in the PEX1-Gly844Asp Mouse Model for Mild Zellweger Spectrum Disorder

Catherine Argyriou1, Ji Yun Song2, Anna Polosa3, Devin McDougald2, Bruno Ceçyre4, Erminia Di Pietro5, Jean-François Bouchard4, Joseph Hacia6, Pierre Lachapelle3, Jean Bennett2, Nancy Braverman1,5

1Department of Human Genetics, McGill University, Montreal, Canada 2Center for Advanced Retinal and Ophthalmic Therapeutics, F.M. Kirby Center for Molecular Ophthalmology, Perelman School of Medicine, University of Pennsylvania 3Department of Ophthalmology, McGill University, Montreal, Canada 4School of Optometry, Université de Montréal, Montreal, Canada 5Montreal Children’s Hospital Research Institute, Montreal, Canada 6Department of Biochemistry and Molecular Medicine, Keck School of Medicine, Los Angeles, California, USA

Manuscript in preparation

103

4.1 ABSTRACT

Purpose. Zellweger spectrum disorders (ZSDs) are a group of autosomal recessive disorders caused by mutations in any one of 13 PEX genes whose protein products are required for peroxisome assembly and function. Individuals with the common PEX1-G843D mutation have an intermediate or milder disease phenotype and consistently develop a retinopathy that progresses to blindness. To test whether we could slow visual loss in these patients, we performed a proof-of-concept trial for PEX1 retinal gene augmentation therapy using our mouse model homozygous for the equivalent PEX1-G844D mutation. Methods. AAV8.CMV.hPEX1.HA vector was administered by subretinal injection to 2 mouse cohorts of 5 or 9 weeks of age (early or later stage retinopathy respectively). AAV8.CMV.GFP vector was used as a control in the contralateral eye. Efficient expression of the virus was confirmed by retinal histology/immunohistochemistry against the HA-tag. Functionality of the vector in recovering peroxisome import was confirmed in vitro using PEX1-deficient cell lines. ffERG and optokinetic (OKN) analyses were performed at regular intervals for 5 to 6 months post gene delivery. Results. Successful expression of hPEX1.HA protein in the photoreceptor layer with no gross histologic side effect was observed. Preliminary ffERG and OKN analyses at 8 weeks post injection showed twofold better cone-mediated retinal response and a non-statistically significant improvement in visual acuity, respectively, in the PEX1-injected eyes. ffERG analysis was improved by 1.6 to 2.5-fold when each cohort reached 25-weeks of age (16 or 20 weeks after gene delivery). At 32 weeks of age, the average ffERG response in the PEX1-injected eyes was double that of GFP-injected eyes in both cohorts. Furthermore, in PEX1-injected eyes the photopic ffERG response improved over time, and the decline in scotopic b-wave amplitude was ameliorated compared to GFP-injected eyes. OKN results showed a non-significant trend towards improvement. Neither the injection, exposure to AAV8 capsid or transgenic protein negatively altered the ffERG or OKN response. Conclusions: At 20-24 weeks after gene delivery, therapeutic vector-treated eyes showed improved ffERG compared to control eyes, on average, in both the younger and older cohorts.

104

This supports the clinical potential of retinal gene augmentation to improve vision in patients with ZSD at both earlier and later stages of disease.

105

4.2 INTRODUCTION

Zellweger Spectrum Disorder (ZSD) is a heterogeneous disorder typically caused by biallelic mutations in any of 13 PEX genes, resulting in a range of multi-system manifestations [9, 12, 173, 189] . PEX genes encode PEX proteins, which regulate peroxisome biogenesis, including synthesis, assembly, matrix protein import, and division. Peroxisomes are ubiquitous, single membrane-bound organelles, each containing over 50 enzymes required for multiple vital metabolic processes [reviewed in [14]]. The best characterized of these pathways include β- oxidation of very long chain fatty-acids, α-oxidation of methyl-branched phytanic acid, the biosynthesis of ether phospholipids (plasmalogens), and free radical metabolism.

Phenotypic severity in ZSD is associated with the level of residual PEX protein activity and consequent peroxisome function. Mutations in PEX1 are the most common cause of ZSD, accounting for roughly 70% of known North American cases [9]. One mutation, PEX1- c.[2528G>A] encodes the missense protein PEX1-p.[Gly843Asp] (or G843D), and results in a milder ZSD phenotype without major developmental defects, but instead a progressive multi- system disorder due to ongoing peroxisome dysfunction over time [10, 185, 242]. The PEX1- G843D allele has a high frequency due to a founder effect in persons of North European ancestry, and represents roughly 30% of all ZSD alleles [9, 12, 13, 325].

PEX1 and its binding partner PEX6 are AAA ATPases (ATPases associated with various cellular activities) that form part of the peroxisome ‘exportomer’ complex. The PEX1-PEX6 complex is anchored to the peroxisome membrane by PEX26 [295] and uses the energy from ATP hydrolysis to ‘pull’ the PEX5 enzyme receptor from the peroxisome membrane so that it can be recycled for additional rounds of import [129, 131][reviewed in [296]]. Peroxisomal enzymes most commonly contain a Peroxisome targeting signal 1 motif, typically the C-terminal tripeptide serine-lysine-leucine [84-86]. PEX5 uses this motif to bind and deliver these enzymes to the peroxisome [92]. If not recycled, PEX5 is targeted for proteasomal degradation [113], and enzyme import into the peroxisome is impaired.

Although ZSD clinical presentation is highly variable and can include brain, liver, adrenal, and bone involvement [reviewed in [1, 190]], nearly all ZSD patients develop sensorineural hearing loss and a progressive retinopathy leading to blindness. The natural history of ZSD retinopathy

106

has not been systematically studied in these patients, and observations have been reported mainly through case reviews. Although these reports are limited by inadequate details of disease severity, PEX genotypes, and evolution of the retinopathy, they provide a description of retinal findings in various patients. In severely affected patients, ophthalmic manifestations can, in addition to retinopathy, also include cataracts, glaucoma, and corneal clouding [reviewed in [326]]. Across the disease spectrum, however, posterior segment abnormalities, particularly photoreceptor loss, are the most severe of all ophthalmic manifestations [reviewed in [326]]. On the milder end of the spectrum, patients have been diagnosed with Usher Syndrome [210, 327, 328], and presentations are most typically characterized by posterior segment anomalies including retinitis pigmentosa, macular atrophy, reduced visual acuity, and reduced or extinguished electroretinograms [198, 329]. Additional manifestations include retinal arteriolar attenuation, optic nerve atrophy, nystagmus, and foveal thinning [212, 329][reviewed in [326]].

Visual impairment is a major, currently untreatable handicap for ZSD patients, and alleviating this burden would enable improved communication, learning, mobility, and autonomy. Improvement in peroxisome function, particularly at an early disease stage, could slow, or even prevent, retinopathy progression and cellular deterioration. In recent years, recombinant adeno- associated virus (rAAV) delivered gene augmentation therapy for monogenic ocular diseases has gained traction, with 30 clinical trials currently listed (clinicaltrials.gov). AAVs are naturally- occurring, non-pathogenic viruses amenable to in vitro manipulation, and able to transduce differentiated, non-dividing cells [357]. As they do not insert in the genome, AAVs can be used as relatively safe vectors to deliver an episomally-expressed gene sequence to a target cell, in order to recover expression of a functional protein, thus alleviating the disease [358]. Studies optimizing the viral capsid (serotype), administration route, species, and target tissue have allowed the translation of these vectors from the bench to the clinic [reviewed in [359]]. The first regulatory approval of gene therapy to treat an inherited disease occurred in December 2017 for voretigene neparvovec (LuxturnaTM , Spark Therapeutics), which uses an AAV2 vector to deliver a functional RPE65 gene by subretinal injection to patients with RPE65-associated Leber’s Congetinal Amaurosis [360].

To test whether rAAV-mediated gene augmentation could improve retinal function in ZSD, we used our mouse model for mild ZSD, which has the PEX1-G843D equivalent knock-in mutation

107

(Pex1-G844D) as a proof-of-concept system [264]. We previously showed that this model exhibits consistently diminished cone photoreceptor function (photopic full-field flash electroretinogram, (ffERG) to 32 weeks, with rod photoreceptor function (scotopic ffERG) diminishing over time, and reduced visual acuity by 11-13 weeks (Figure 3.1A, B, 3.3D). This model also exhibits decreased cone cell numbers through life, decreased bipolar cell numbers with age, and photoreceptor inner segment disorganization (Figure 3.5A, 3.5B, 3.7A). As photoreceptor structure and function is primarily affected in this model, and the retinal inner segment is enriched in peroxisomes and a primary site of PEX1 localization [210, 346](Figure 3.8B), the vector capsid and delivery route selected would have to efficiently infect the outer and inner segment of the retina. We selected to use the AAV8 serotype delivered by subretinal injection, as this combination demonstrates superior targeting of the photoreceptor cells in both mice [361, 362] and non-human primates [363, 364]. As all cell types in our model lack functional PEX1, and overexpression is not expected to be problematic, we used the cytomegalovirus (CMV) promoter and enhancer to drive the expression of human PEX1 in the mouse retina, and included an N-terminal human influenza hemagglutinin (HA) tag to distinguish vector-delivered protein from endogenous PEX1-G844D. The same vector containing eGFP was used as a control.

After confirming that AAV8.hPEX1 recovered peroxisome function in mouse and human PEX1- deficient cell lines, we report here on the effects of AAV8.CMV.hPEX1.HA delivered by subretinal injection into the left eye of 5 and 9-week-old homozygous PEX1-G844D mice. Outcome measures included PEX1 protein expression, effect on retinal function (ffERG), and visual acuity (OKN) over the course of 5-6 months.

4.3 MATERIALS AND METHODS

4.3.1 Proviral plasmids and AAV production

Human codon optimized PEX1 sequence (synthesized by ATUM/DNA2.0) was amplified with Q5 DNA polymerase (NEB) to include a Kozak consensus sequence preceding the translational start site and a C-terminal HA epitope tag. This PCR product was digested with NotI-HF and ScaI-HF restriction enzymes (NEB) and cloned into an AAV proviral plasmid using T4 DNA ligase (NEB). The completed proviral vector consisted of the cytomegalovirus (CMV) enhancer/promoter driving transgene expression and terminating with bovine growth hormone

108

(bGH) polyadenylation signal. The expression cassette was flanked by the canonical AAV2 inverted terminal repeats (ITRs). AAV8 vectors (Center for Advanced Retinal and Ocular Therapeutics (CAROT) Research Vector Core, University of Pennsylvania) were generated using previously described methods [365] by branched polyethylenimine (PEI) (Polysciences, no. 23966)-mediated triple transfection of HEK293 cells with a plasmid containing the transgene between the ITRs of AAV2, the AAV-helper plasmid encoding Rep2 and Cap for serotype variants, and the pHGTI-Adeno1 plasmid harboring helper adenoviral genes. The HEK293 cells express the helper E1A/E1b gene (American Type Culture Collection, catalogue number CRL- 157). Vectors were purified using a discontinuous iodixanol gradient (Sigma, Optiprep). Encapsidated DNA was quantified by TaqMan RT–PCR, following denaturation of the AAV particles by proteinase-K; titers were calculated as genome copies (gc) per ml.

4.3.2 In vitro titer assay for individual AAV capsid variants

Capsid genes were cloned in an AAV packaging plasmid for vector production, and used for small-scale vector preparations encoding firefly luciferase to obtain the titer. Physical particle titers were established by TaqMan qPCR. Subsequently, AAV2/8bp variants were assayed for transduction at equal multiplicity of infection onto HEK293 cells. For large-scale viral titer, the encapsidated DNA was quantified by TaqMan RT–PCR, following denaturation of the AAV particles by proteinase-K, and titers were calculated as genome copies (gc) per ml.

4.3.3 Cell transduction and immunoblotting

A AV 8 . C M V. hPEX1.HA was added to 84-31 cells (a modified KEK293 cell line) at a multiplicity of infection (MOI) of 105 or 2x105 (10μl or 20μl of 1010 vg/μl per 2x106 cells at time of viral expression). AAV8.CMV.eGFP was applied at an MOI of 105. Cells were harvested after 48 hrs, lysed, and separated on a 4-12% polyacrylamide gradient gel and transferred to nitrocellulose membrane. Membranes were blocked and hybridized in 5% milk with 1:1000 rabbit anti-HA- Tag (Cell Signaling Technology 3724) followed by rabbit anti-HRP conjugated secondary antibody, and visualized by ECL using an Amersham 600 platform and 20 sec exposure.

4.3.4 Peroxisome import after viral transduction

HepG2 control and PEX1 null cells (generated using CRSPR-Cas9 mediated gene editing to disrupt the gene) were seeded onto coverslips in 12-well plates and transduced with

109

AAV8.CMV.hPEX1.HA or AAV8.CMV.eGFP at an MOI of 105. Cells were prepared for indirect immunofluorescence as previously described [301]. Primary antibodies used were 1:200 anti-PTS1 and1:400 sheep anti-human ABCD3 (gifts from Steven Gould, Johns Hopkins University). Secondary antibodies used were 1:400 anti-rabbit 488 (Invitrogen), 1:300 anti-sheep Texas Red (Sigma-Aldrich). PTS1 (Peroxisome targeting signal 1) is a marker for peroxisome import that co-localizes with peroxisome punctate structures when import is functional, and is localized to the cytosol and/or degraded when import is impaired. Slides were visualized using an Olympus BX51 microscope (Center Valley, PA) at 60X magnification and images captured using an Olympus CCD camera and MagnaFire software.

Primary mouse wild-type and PEX1-G844D homozygous fibroblasts expressing a GFP-tagged PTS1 reporter were transduced with MOI 105 AAV9.CMV.hPEX1. Cells were fixed 48hrs later and visualized by indirect immunofluorescence.

4.3.5 Animal husbandry

Pex1-G844D mice were maintained on a mixed 129/SvEv and C57BL/6N Taconic background. Strain background was evaluated yearly by SNP genotyping (MaxBax, Charles River) and showed a stable 70% 129/SvEv and 30% C57BL/6N Taconic. Mice were housed at the RI- MUHC Glen site animal care facility with ad libitum access to food and water. All experiments were performed at the RI-MUHC Glen site, except for visual acuity measures, which were performed at the Pavillon Liliane-de-Stewart de l’Université de Montréal animal care facility. All experiments were approved by the Research Institute of the McGill University Health Centre Animal Care Committee or the Université de Montréal Ethics Committee. Euthanasia was

performed by CO2 asphyxiation under isoflurane anaesthesia (5% isoflurane in oxygen until loss

of consciousness, immediately followed by CO2 at maximum flow rate, 4 LMP). Both males and females were used for all experiments.

110

presence of a 177 bp residual Neo cassette fragment in intron 13 near the Pex1 c.2531G>A mutation in exon 15. The Pex1 c.2531G>A (Pex1 p.Gly844Asp) allele yields a 852 base pair product and the wild-type allele yields a 672 base pair product.

4.3.6 Design of in vivo experiments

Experiments were done using 2 different age cohorts to test the effect of intervention at early and later disease stages. A ‘validation’ cohort was included for the purpose of early confirmation of vector-mediated protein expression and to determine extent and localization of expression. To account for variability among individual mice, each injected mouse received therapeutic vector in the left eye, and GFP vector in the right eye.

‘Prevention’ cohort: Baseline ffERGs at 4 weeks of age, injection at 5 weeks of age. Preliminary ffERGs were done on a subset of animals 5 months post injection (25 weeks of age). Final ffERGs were done 6 months post injection and visual acuity (OKN) 6.5 months post injection (31-33 weeks of age), after which mice were sacrificed. Numbers at end: 15 injected mutants, 4 non-injected mutants, 2 injected wild-type littermates, 4 non-injected wild-type littermates.

‘Recovery’ cohort: Baseline ffERGs at 8 weeks of age, injection at 9 weeks of age. Preliminary ffERGs were done on a subset of animals 4 months post injection (25 weeks of age). Final ffERGs were done 5 months post injection and OKN 5.5 months post injection (31-33 weeks of age), after which mice were sacrificed. Numbers at end: 18 injected mutants, 6 non-injected mutants, 4 injected wild-type littermates, 4 non-injected wild-type littermates.

‘Validation’ cohort: ffERG done 8 weeks post injection (14 or 17 weeks of age), and no baseline ffERG. OKN was done 11 weeks post injection (17 or 20 weeks of age). Mice were sacrificed 12 weeks post injection and eyes processed for IHC. Numbers: 3 injected at 6 weeks of age, 6 injected at 9 weeks of age.

4.3.7 Vector delivery

Virus was diluted to 1.03 x 1010 vg/µl (AAV8.CMV.hPEX1.HA) or 1.40 x 1010 vg/µl (AAV8.CMV.eGFP) in Pluronic F-127 buffer (Sigma-Aldrich). Mice received Meloxicam oral analgesic in suspension and were anaesthetized by intraperitoneal injection of 10% ketamine, 5% xylazine solution (130µL/10g body weight). Proparacaine hydrochloride (Alcaine®) was applied

111

to the eye and pupils dilated using tropicamide (Mydriacyl®). Bilateral subretinal injections to deliver 1μL virus dilution per eye were performed at previously described [366]. AAV8.CMV.hPEX1.HA was delivered to the left eye, and AAV8.CMV.eGFP to the right. Tobramycin/dexamethasone ointment (Tobradex®) was applied to eyes for 2 days following surgery. Injections were performed in the surgical suite of the Glen-MUHC under a dissecting microscope. Mice were observed for signs of discomfort or corneal injury.

4.3.8 Electrophysiology

Retinal function was assessed using full-field flash electroretinogram (ffERG) as previously described [331][Chapter 3.3.2]. Following a 12 hour dark adaptation, mice were anesthetized (intraperitoneal injection of 10% ketamine, 5% xylazine solution in PBS (130µL/10g body weight)) and their pupils were dilated (1% Mydriacyl Tropicamide (Alcon, Mississauga, CA)). All experimental procedures were performed in a dark room under red light illumination. Two types of recordings were performed to assess rod and cone function: scotopic or dark-adapted ffERG and photopic or light adapted ffERG, respectively. Quantification of the amplitude of the a-wave and b-wave was performed as previously described [334]. Briefly, the amplitude of the a- wave was measured from baseline to the most negative trough, while the amplitude of the b- wave was measured form the trough of the a-wave to the most positive peak of the ffERG (scotopic ffERG) and from the baseline to the highest peak of the b-wave (photopic ffERG). ffERG was performed at baseline and at 2, 4, 5, and 6 months post gene delivery.

4.3.9 Visual acuity using the Virtual Optomotor System

The spatial frequency threshold (“visual acuity”) of the optomotor reflex of mice was determined using a virtual-reality optomotor system (Cerebral Mechanics, Lethbridge, Alberta) as previously described [289]. Briefly, freely moving mice were placed on an elevated platform and exposed to vertical sine wave gratings rotating at 12 degrees/sec. The grating spatial frequencies increased from 0.01 to 0.5 cycles/degree in varying increments. When able to perceive the stimulus, the mouse normally stopped moving its body and began to track the grating with reflexive head movements in concert with the rotation. Spatial frequency of the grating at full contrast (100%) was gradually increased until the mice no longer exhibited a tracking behavior. The highest spatial frequency that could be followed (i.e. spatial frequency threshold) determined the visual acuity (in cycle per degree) for each eye. This technique yields independent measures of right-

112

and left-eye acuity, as only motion in the temporal-to-nasal direction evokes a tracking response. Mice were always tested within three hours of their daylight hour onset.

4.3.10 Retinal immunohistochemistry

Eye cups from PBS-perfused mice were fixed 3 hr. in 10% formaldehyde, incubated in 10% (30 min. on ice), 20% (1 hr. on ice), and 30% (4°C overnight) sucrose in 0.1M PB, then embedded and frozen in frozen section compound (VWR). 5µm retinal cryo-sections were blocked (1% NGS, 0.1% triton, 10% BSA in PBS) for 1 hr., washed, incubated at 4°C overnight with primary antibody in incubation buffer (0.1% triton, 10% BSA in PBS), washed, incubated 90 min. with secondary antibody and washed. Coverslips were mounted using ProLong Gold antifade reagent with DAPI (Invitrogen) and slides visualized using a Leica DMI600 microscope, DFC345FX camera and LASX software. Primary antibodies used were 1:300 rabbit anti-HA-Tag (Cell Signaling Technology 3724).

Whole retinas were also removed, flash frozen, and stored at -80°C for future biochemical analysis of peroxisome metabolites by LC-MS/MS. Tail snips were collected and stored for future SNP genotyping to confirm strain background, and blood (plasma and red cells) to correlate treatment effect with blood biomarkers.

4.4 RESULTS

4.4.1 Recovery of peroxisome import by rAAV-delivered hPEX1 in human and mouse cells

Our vectors consisted of the transgenes codon-optimized human PEX1 with a C-terminal HA epitope tag (Figure 4.1A), or eGFP (Figure 4.1B), driven by a CMV enhancer/promoter, and terminating into a bovine growth hormone (bGH) polyadenylation signal. The expression cassette was flanked by the canonical AAV2 inverted terminal repeats (ITRs) and packaged into an AAV8 vector.

The AAV8.CMV.hPEX1.HA vector was tested for expression of the appropriate sized protein using 84-31 cells transduced with AAV8.CMV.hPEX1.HA at a multiplicity of infection (MOI) of 105 or 2x105 vector genomes (vg) per cell. Immunoblotting against the HA-tag revealed a single band of approximately 150kDA, corresponding to the size of PEX1, with increased

113

intensity at the higher compared to lower MOI (Figure 4.1C). No HA-tag was detected in cells transduced with the control vector, AAV8.CMV.eGFP.

To ensure that the HA-tag does not interfere with PEX1 function, PEX1 null HepG2 cells were transfected with MOI 105 AAV8.CMV.hPEX1.HA and visualized after 48 hours by immunofluorescent microscopy. In control cells, enzymes with the Peroxisome targeting signal 1 (PTS1) motif are localized to peroxisomes, co-localizing with the punctate peroxisome membrane protein ABCD3. In cells with a peroxisome defect, such as PEX1-deficient HepG2 cells, PTS1-enzymes are not imported by peroxisomes, and thus localized to the cytosol and degraded. AAV8-mediated expression of HA-tagged PEX1 (MOI 105) recovered PTS1 localization in PEX1 null HepG2 cells (Figure 4.2A), implying that the tagged transgene is functional within the PEX1-6-26 ‘exportomer’ complex. As the virus is largely episomal and thus diluted with each cell division, only a few rescued cells are seen per field of view. Transduction of PEX1 null HepG2 cells with AAV8.CMV.eGFP, or wild-type HepG2 cells with AAV8.CMV.hPEX1.HA had no visible effect on PTS1 localization (not shown).

To confirm that human PEX1 is functional within the mouse ‘exportomer’ complex, primary PEX1-G844D mouse fibroblasts over-expressing a GFP-tagged PTS1 reporter (GFP-PTS1) were used to assess peroxisome import. As the AAV8 does not efficiently transduce fibroblasts, a serotype 9 vector (AAV9.CMV.hPEX1) was used to test recovery in the mouse fibroblasts after hPEX1 delivery. At baseline, the GFP-PTS1 reporter was localized throughout the cytosol, but after transduction with AAV9.CMV.hPEX1 at MOI 106 redistributed to a more punctate localization, indicating recovered peroxisome import (Figure 4.2B). This supports functionality of the human PEX1 protein in PEX1-G844D murine cells.

4.4.2 rAAV8-mediated hPEX1.HA protein is expressed in mouse retina

The effect of AAV8-mediated gene delivery was tested at two different ages: mice injected at i) 5 weeks of age, representing the ‘prevention’ cohort, or ii) 9 weeks of age, representing the ‘recovery’ cohort. Full-field flash electroretinograms (ffERG) were recorded for all mice in the week prior to injection (waveforms and quantification are shown in Supplemental figure 4.1). To account for phenotypic variation among individuals, we used the contralateral eye of each treated animal as a control: PEX1-G844D and wild-type littermate mice received

114

AAV8.CMV.hPEX1.HA in the left eye and AAV8.CMV.eGFP in the right eye by subretinal injection. ‘Non-injected’ mutants and wild-type littermates received no intervention and served as controls for the surgical procedure. A schematic representation of the experimental design is shown in Figure 4.3.

Eight PEX1-G844D mice were sacrificed 11 weeks after subretinal injection to determine the location and intensity of hPEX1-HA or eGFP expression in the retina. The HA-tag was localized to the photoreceptor inner segment and the outer plexiform layer, matching the reported sites of endogenous Pex1 enrichment [210, 346] (Figure 4.4C), suggesting that the vector-delivered gene is translated and appropriately localized. GFP expression was more general, extending from the outer segment through to the outer plexiform layer, and at times reaching the inner nuclear layer. Immunohistochemistry was performed across all retinal sections from each eye, and all sections from each retina were visually scored for intensity of protein expression and degree of coverage. Typically, expression was strongest towards the nasal end of the retina, and gradually diminished towards the temporal end. Although all eight retinas were positive for HA-tag or GFP expression, the amount, location, and scope of expression were variable, with representative images shown in Figure 4.3C. Approximately 30% of the retina in each mouse was positive for some degree of expression. No gross histologic side-effects or infiltration of the HA-tag to the nuclear layers was observed.

4.4.3 Visual function improves by 8 weeks post subretinal gene delivery ffERGs were performed eight weeks following gene delivery on eight PEX1-G844D mice and eight non-injected mutant controls prior to sacrifice for transgene expression and localization studies. The average cone mediated (photopic) ffERG response of the hPEX1-injected (left) eyes was two-fold that of the GFP-injected (right) eyes and more than two-fold that of the non- injected left or right eyes (Figure 4.4A). The average rod-mediated (scotopic) b-wave trended towards improvement though this was not significant. Visual acuity of the same eight injected mice was determined using the OptoMotry system to measure the optokinetic nystagmus reflex (OKN) 11 weeks after subretinal gene delivery, when mice were 14 or 17 weeks old. Although the average visual acuity was nearly four-fold higher in the left versus right eyes, this difference was just shy of statistical significance (t-test, P=0.054) (Figure 4.4B).

115

ffERGs were recorded on a subset of animals (6) when each cohort reached 25 weeks of age (16 or 20 weeks after gene delivery). In the ‘prevention’ cohort, scotopic a-wave, b-wave, and photopic b-wave were on average 1.6, 1.7, and 2.5 -fold higher, respectively, in the left versus right injected eyes (Figure 4.5A). In the ‘recovery’ cohort, the average scotopic b-wave response was 2.7-fold higher in the left versus right eye (Figure 4.5B). Although the average scotopic a- wave and photopic b-wave responses were 1.6 and 2.5 -fold higher in the left eye of the ‘recovery’ cohort, this effect was not statistically significant.

4.4.4 Visual function but not functional vision improves 5 or 6 months post subretinal gene delivery

Endpoint ffERGs were performed when each cohort reached 31 weeks of age, 5 or 6 months post gene delivery for the ‘recovery’ or ‘prevention’ cohorts, respectively. In both cohorts, the average scotopic a-wave, scotopic b-wave, and photopic b-wave amplitude of the therapeutic vector-treated (left) eyes, was two-fold that of the control injected (right) eyes (Figure 4.6A, 4.6B). In PEX1-G844D mice, the retinal response of the control injected (right) eyes did not differ from that of either eye in non-injected mutant concurrent controls (Figure 4.6A, 4.6B). Furthermore, in PEX1-injected eyes the endpoint photopic ffERG response improved one- to ten- fold over baseline in 72% (24/33) of mice, compared to 20% (2/10) of non-injected mutants, which showed a more modest improvement. The average decline in scotopic a-wave and b-wave response was ameliorated by 41-58% compared to non-injected eyes in both ‘prevention’ and ‘recovery’ groups, with 12% (4/33) of treated animals improving over time, compared to 0/10 controls. Table 4.1 shows the average ffERG amplitude and standard deviation at baseline and endpoint for each treatment group, as well as the average amplitude change over this time. The proportion of mice in which retinal response improved or declined is also shown for each group. A limitation of this comparison is that baseline ffERGs were not acquired in contralateral control (right) eyes prior to injection with AAV8.CMV.eGFP.

OKN measures could only be done at the experimental endpoint, since they were done at a different animal care facility, and the mice could not be re-imported for subsequent ffERGs. Average visual acuity trended higher in the PEX1-injected (left) versus GFP-injected (right) eyes, but this was not statistically significant (Figure 4.6A, 4.6B). For both ffERG and visual acuity measures, there was no difference between wild-type animals with or without subretinal

116

injection. These values are thus grouped together as “wild-type” for representation in Figure 4.6A, 4.6B.

4.5 DISCUSSION

4.5.1 PEX1 gene augmentation improves retinal function

In PEX1-G844D and wild-type littermates, neither the injection, nor exposure to the AAV8 capsid or the transgenic protein negatively altered gross retinal histology, ffERG, or OKN response. Preliminary ffERG and optokinetic (OKN) analyses at 8 weeks post injection showed mildly better retinal response and visual acuity (Figure 4.4A, 4.4B), respectively, in the PEX1- injected eyes, as did ffERG analysis when each cohort reached 25-weeks of age (16 or 20 weeks after gene delivery) (Figure 4.5A, 4.5B). This effect was more pronounced in the cohort with intervention at 5 weeks of age, when ffERG response is highest in homozygous PEX1-G844D mice. At 5-6 months after gene delivery, therapeutic vector-treated eyes showed improved ffERG compared to control eyes, by twofold on average, in both the younger ‘prevention’ and older ‘recovery’ cohorts.

PEX1-G844D mice typically exhibit a gradual decline in scotopic ffERG, and a photopic ffERG that remains consistently low from 4-32 weeks (Figure 3.1). In these experiments, hPEX1 gene augmentation resulted in 41-58% lower scotopic ffERG decline compared to non-injected mutants, and improved the photopic response one- to ten-fold over time in the majority of treated animals (Table 4.1). This highlights the possibility to slow or prevent the deterioration of retinal response, or in the case of photopic ffERG, even improve response. The effect of retinal gene augmentation was relatively greater in the cone versus rod-mediated response, despite cone function being affected earlier in this model. It may be that peroxisome dysfunction affects rods and cones differently, and that a small improvement in peroxisome function, or an improvement in fewer peroxisome-mediated pathways, is sufficient to greatly recover cone function. Alternatively, the relative preservation of rod function in this model, and variability therein, could make the effect of therapy less apparent.

4.5.2 Functional relevance

Based on these studies, the effect of AAV8.CMV.hPEX1.HA delivery is durable to at least six months. In the PEX1-G844D mouse model, rod-mediated ffERG (scotopic, b-wave) diminishes

117

to below 15% of wild-type, on average, by 32 weeks (Figure 3.1). As variation among mutants also diminishes by this age, we selected 32 weeks as the endpoint of our experiments, and thus do not know the therapeutic efficacy beyond this point. There is community concern about long- term benefits of gene therapy, and whether therapeutic effect could be limited by progressive retinopathy [reviewed in [358]]. The effect of LuxturnaTM, which uses subretinally injected AAV2-delivered RPE65 to improve vision in Leber’s Congenital Amaurosis, is confirmed durable to at least three years in patients, with follow-on studies continuing [367, 368]. The numerous gene therapy trials currently in the clinic will inform durability concerns in the coming years. Regardless, any degree or length of effect is valuable to the individual with progressive retinopathy, where no other curative interventions are available.

Despite the twofold improvement in scotopic ffERG response over controls 5-6 months post hPEX1 gene augmentation, the scotopic and photopic retinal function of treated Pex1-G844D mice remained 22-33% the wild-type average (Figure 4.6A, 4.6B). As with many therapies in development, a challenge of retinal gene augmentation is determining the sufficient expression level and retinal coverage to meaningfully impact functional vision and thus quality of life. In the subset of animals tested 11 weeks following gene delivery, the average visual acuity of therapeutic vector treated eyes was four-fold that of the contralateral eyes, and this difference neared statistical significance (Figure 4.4B). By 5-6 months following gene delivery, only a subtle trend of improved visual acuity compared to the AAV8.CMV.eGFP injected contralateral eye remained (Figure 4.6A, 4.6B). To address this, we could expand our functional vision testing to include measures of contrast sensitivity [369] and colour perception [267], as PEX1- G844D mice could be more sensitive to these parameters than to full-contrast optomotor testing. The addition of more endpoint measures could also add power to our visual acuity testing, in that a subtle improvement across several tests may prove statistically meaningful.

Some limitations in interpreting our results include the large range of phenotypic severity at early age in PEX1-G844D mice in both ffERG and visual acuity (Supplemental figure 4.1, Figure 4.4A), and that baseline disease severity may influence therapeutic outcome in each animal. It is also important to note that approximately 30% of the retina in each animal was exposed to vector, and that transgene expression varied among mice (Figure 4.4C, Supplemental table 4.1). This range in efficiency may also lead to outcome variation. In future experiments, a

118

method to correlate location and amount of vector expression with peroxisome recovery should be considered. For example, imaging mass spectrometry allows detection of various lipids in the different retinal layers and across the span of the retina [339, 354, 355], including peroxisome dependent lipids such as very long chain fatty-acids, which are elevated in the PEX1-G844D retina (Figure 3.8C). This method could also be used to measure the relative amount of human PEX1 to mouse PEX1-G844D protein, thus allowing for correlation of vector delivered protein with lipidomics in consecutive sections. Moreover, the amount of hPEX1 across retinal sections could be quantified with this technology and correlated to functional outcome.

Retinopathy in PEX1-G844D mice occurs predominantly at the photoreceptor level, with bipolar cells diminishing later in life. In these preclinical experiments, we have thus focused on targeting the photoreceptor cells, and site of endogenous PEX1enrichment, but do not know if outer retina preservation is sufficient to prevent degeneration at the inner retina. Following subretinal delivery, our vector-delivered proteins are most strongly localized in the photoreceptor outer and inner segment, but also consistently expressed up to the outer plexiform layer (Figure 4.4C), the site of photoreceptor and bipolar cell synapses. Although subretinal injection in mice is the best way to target outer retinal cells [370], intravitreal injection allows efficient targeting of vector expression to inner retinal cells, and could be considered as an additional experimental approach [371].

4.5.3 Clinical implications

In the PEX1-G844D mouse retina, protein levels of PEX1-G844D, its partner protein PEX6, and its ligand PEX5, are normal (Figure 3.9A), indicating the missense protein is present, but subfunctional. This implies that the rAAV-delivered human PEX1 is competing against endogenous PEX1-G844D for a place on the PEX1-PEX6-PEX26 complex, which may be deleterious to efficient complex recovery. In contrast, PEX1-G843D protein amounts are degraded to 5-15% of normal PEX1 levels in patient fibroblasts [11], with PEX6 and PEX5 also reduced [11, 290]. This may have positive implications for gene augmentation therapy, as competition against endogenous subfunctional protein is reduced.

In Pex1-G844D mice, AAV8-mediated gene augmentation therapy had a relatively more robust effect on cone versus rod function. Although one cannot predict whether this effect would

119

translate to humans, robust cone rescue could ameliorate ambient light vision and thus facilitate the majority of daily tasks for ZSD patients. Moreover, humans have a cone-enriched macular region, which could be targeted by guided subretinal injection. In several PEX1-G843D homozygous patients, macular edema with photoreceptor degeneration, as well as peripheral visual field constriction have been reported [212, 328, 330], suggesting that the central macula, and thus cone photoreceptors, may be a logical therapeutic target.

Currently, there are 3 clinical trials using AAV8-mediated retinal gene delivery to treat the inherited ocular diseases achromatopsia (CNGB3), retinoschisis (RS1), and retinitis pigmentosa (RLBP1) (clinicaltrials.gov identifiers NCT03001310, NCT02317887, and NCT03374657, respectively), all of which progressed from murine-based proof-of-concept and optimization studies [117, 372, 373]. Our proof-of-concept studies using the PEX1-G844D mouse model for mild ZSD highlight the ability to improve retinal function even after the onset of retinal degeneration. There are currently no targeted treatments available for ZSD, and visual care is supportive only. Therapy for PEX1-mediated retinal degeneration would benefit a large proportion of ZSD patients, improving communication, learning, autonomy, and overall quality of life.

4.6 ACKNOWLEDGEMENTS

We would like to thank Gregory Robinson and John Tobin from Nightstar Therapeutics, and Deborah Slipetz, Eunkyung Kauh, Leonard Dragone, Paul Carrington, Xiaoyan Li, and Jesse Nussbaum from Merck for kindly donating their time and expertise to advise us on the next strategic steps for this project. We also acknowledge the ZSD patients and families, whose courage inspires us. We would especially like to thank all those at The Global Foundation for Peroxisome Disorders and the Wynne Mattefy Foundation for believing in this project; your support made this work possible and has opened the door to developing treatments for this condition. This work was funded by the Global Foundation for Peroxisome Disorders and the Wynne Mattefy Foundation to NB, JH, and JB, The Canadian Institutes of Health Research to NB (CIHR 12610) and NB and PL ( CIHR 34575) and the Fonds de recherche du Québec – Santé to CA (FRQS doctoral training award 32105).

120

4.7 CONFLICTS OF INTEREST

There are no conflicts of interest.

121

Figure 4.1. Proviral plasmid design and rAAV8 vector expression. (A) Proviral plasmid maps of the transgenes codon-optimized human PEX1 with a C-terminal HA epitope tag, or (B) eGFP or driven by a CMV enhancer/promoter, and terminating into a bovine growth hormone (bGH) polyadenylation signal. The expression cassette is flanked by the canonical AAV2 inverted terminal repeats (ITRs) and was packaged into an AAV8 vector. (C) Immunoblotting of 84-31 cells transduced with the AAV8.CMV.hPEX1.HA vector shows a single 150 kDA band corresponding to the size of PEX1 when membrane is probed with anti-HA. Band intensity correlates to MOI used for trandsduction. No HA-tag was detected in cells transduced with the control vector, AAV8.CMV.eGFP

122

Figure 4.2. Recovery of peroxisome import by rAAV-delivered hPEX1 in human and mouse cells. (A) At baseline, control HepG2 cells exhibit punctate Peroxisome targeting signal 1 (PTS1) distribution (white arrows) that co-localizes with the peroxisome membrane protein ABCD3.In contrast, PTS1 is diffuse in PEX1 null HepG2 cells. Following transduction with AAV8.CMV.hPEX1.HA (MOI 105), PTS1 redistributes from the cytosol to peroxisomes (puncta, white arrows), indicating recovered peroxisome import. (B) Representative image of Pex1- G844D mouse primary fibroblasts expressing a GFP-PTS1 reporter. Transduction with AAV9.CMV.hPEX1 (MOI 106) causes redistribution of the reporter from the cytosol to peroxisomes (puncta), similar to wild-type cells. Middle ‘magnification’ panel shows 10x magnification of image.

123

Figure 4.3. Schematic representation of experimental design. The effect of AAV8-mediated gene delivery was tested by full-field electroretinogram (ffERG) and visual acuity (optokinetic reflex, OKN) at two different ages, representing the ‘prevention’ and ‘recovery’ cohorts, exposed to vector for 6 or 5 months, respectively. A ‘validation’ cohort was used to obtain preliminary functional measures and validate vector expression in the retina. PEX1-G844D and wild-type littermate mice received AAV8.CMV.hPEX1.HA in the left eye and AAV8.CMV.eGFP in the right eye by subretinal injection. Non-injected PEX1-G844D and wild-type mice were also included in each cohort. The flow chart shows the ages of mice at intervention and assessment, and time between each event.

124

Figure 4.4. Preliminary ffERG, visual acuity, and confirmation of rAAV-delivered protein expression in the ‘sacrificial’ cohort. (A) Quantification of maximum ffERG response 8 weeks post subretinal injection showed improved average photopic b-wave response in the PEX1- injected left eye compared to the GFP-injected right eye, or either eye in non-injected controls. The scotopic ffERG response was not significantly altered. Each point represents 1 animal; Student’s t-test *=P<0.05, **=P<0.01, ***=P<0.001 (B) Optokinetic analyses 11 weeks post injection showed a trend toward improved average visual acuity in the PEX1-injected left eyes compared to right eyes. Animals were tested twice and an average value plotted for each individual eye. (C) Representative immunohistochemistry of retinal sections used to determine the location and scope of hsPEX1-HA expression in the retina. HA-tag expression was localized to the base of the outer segment/inner segment interface (OS/IS) and the outer plexiform layer (OPL), the sites of endogenous Pex1 expression, while GFP expression was more general, extending to the inner nuclear layer. No gross histologic side-effects or infiltration were observed.

125

Figure 4.5. Preliminary ffERG on a subset of animals from ‘prevention’ and ‘recovery’ cohorts. Analyses were performed when each cohort reached 25 weeks of age (16 or 20 weeks after gene delivery). (A) In the ‘prevention’ cohort, maximal scotopic a-wave, b-wave, and photopic b-wave were improved on average in the left versus right injected eyes. (B) In the ‘recovery’ cohort, the average scotopic b-wave response was significantly higher in the left versus right eyes. Scotopic a-wave and photopic b-wave responses trended towards improvement though not statistically significant. Upper dotted line represents wild-type average, and lower dotted line non-injected mutant average (n=5). Each point represents 1 animal; Student’s t-test *=P<0.05, **=P<0.01, ***=P<0.001.

126

Figure 4.6. Endpoint ffERG and visual acuity measures in ‘prevention’ and ‘recovery’ cohorts. Endpoint ffERGs were performed when each cohort reached 31 weeks of age, 5 or 6 months post gene delivery for the ‘recovery’ or ‘prevention’ cohorts, respectively. (A) In the ‘prevention’ cohort, the average maximal scotopic a-wave, scotopic b-wave, and photopic b- wave amplitude of the PEX1-injected left eyes was two-fold that of the GFP-injected right eyes. Average visual acuity trended higher in the left versus right injected eyes, but this was not statistically significant. (B) Similarly, in the ‘recovery’ cohort, the average maximal scotopic a- wave, scotopic b-wave, and photopic b-wave amplitude of the PEX1-injected left eyes was twofold that of the GFP-injected right eyes. Average visual acuity trended higher in the left versus right injected eyes, but this was not statistically significant. In both cohorts, the retinal response of the GFP-injected eyes did not differ from that of either eye in non-injected mutant concurrent controls. As there was no difference between wild-type animals with or without subretinal injection, these values are thus grouped together. Upper dotted line represents wild-type average (n=4-8). Each point represents 1 animal; Student’s t-test *=P<0.05, **=P<0.01, ***=P<0.001.

127

Table 4.1 Comparison of baseline to endpoint ffERG 'Prevention' group ffERG Scotopic, a-wave Scotopic, b-wave Photopic, b-wave Average amplitude at baseline (µV) 124 ± 58 152 ± 130 9± 8 Average amplitude at endpoint (µV) 47 ± 36 100 ± 86 25 ± 18 Pex1-G844D + hPEX1 Average % change from baseline to endpoint -49 ± 59 -23 ± 116 +242 ± 380 (n=15) Percentage of mice that improved 13 (n=2) 20 (n=3) 86 (n=13) Percentage of mice that declined 87 (n=13) 80 (n=12) 14 (n=2) Average amplitude at baseline (µV) 86 ± 51 188 ± 82 8 ± 7 Average amplitude at endpoint (µV) 37 ± 13 61 ± 39 14 ± 2 Pex1-G844D non-injected Average % change from baseline to endpoint -33 ± 60 -56 ± 44 +40 ± 115 (n=5) Percentage of mice that improved 20 (n=1) 0 20 (n=1) Percentage of mice that declined 80 (n=4) 100 (n=5) 80 (n=4) Average amplitude at baseline 153 ± 57 446 ± 167 114 ± 53 Average amplitude at endpoint 179 ± 36 487 ± 120 121 ± 27 Wild- type Average % change from baseline to endpoint +54 ± 75 +45 ± 79 +32 ± 82 (n=4) Percentage of mice that improved 75 (n=3) 50 (n=2) 50 (n=2) Percentage of mice that declined 25 (n=1) 50 (n=2) 25 (n=1) 'Recovery' group ffERG Scotopic, a-wave Scotopic, b-wave Photopic, b-wave Average amplitude at baseline 108 ± 42 234 ± 88 15 ± 8 Average amplitude at endpoint 54 ± 31 97 ± 58 20 ± 18 Pex1-G844D + hPEX1 Average % change from baseline to endpoint -30 ± 102 - 49 ± 57 +82 ± 235 (n=18) Percentage of mice that improved 11 (n=2) 6 (n=1) 56 (n=10) Percentage of mice that declined 78 (n=14) 17 (n=17) 39 (n=7) Average amplitude at baseline 115 ± 30 263 ± 76 35 ± 7 Average amplitude at endpoint 36 ± 23 34 ± 19 17 ± 5 Pex1-G844D non-injected Average % change -64 ± 27 -84 ± 14 -34 ± 67 (n=5) Percentage of mice that improved 0 0 20 (n=1) Percentage of mice that declined 100 (n=5) 100 (n=5) 80 (n=4) Average amplitude at baseline 182 ± 54 545 ± 144 123 ± 29 Average amplitude at endpoint 203 ± 16 460 ± 165 113 ± 21 Wild- type Average % change from baseline to endpoint +17 ± 31 -7 ± 45 +115 ± 226 (n=5) Percentage of mice that improved 20 (n=1) 20 (n=1) 20 (n=1) Percentage of mice that declined 0 40 (n=2) 40 (n=2) % change per baseline compares each animal to itself Improvement or decline was counted as over 10% change from baseline ± indicates standard deviation

Table 4.1. Comparison of baseline to endpoint ffERG.

128

Supplemental figure 4.1. Baseline ffERG measures. Maximal (A) scotopic and (B) photopic ffERG waveforms are shown for all animals. (C) Quantification of maximal ffERG response highlights variation among individual mice at early age.

129

Supplemental figure 4.2. Endpoint ffERG waveforms. Maximal (A) scotopic and (B) photopic ffERG waveforms are shown for all animals.

130

CHAPTER V: DISCUSSION

131

5.1 OVERVIEW AND MAJOR FINDINGS

This thesis presents findings that advance treatment development for PBD-ZSD. These include identifying strong pharmacological candidates, establishing in vivo outcome measures, and supporting the use of gene replacement therapy for this disease.

5.1.1 Small molecule therapy is a viable treatment option for ZSD

In Chapter II, we compared a targeted panel of flavonoids for their ability to recover peroxisomal enzyme import and to elucidate structure-activity relationships in PEX1-G843D patient cells. We used 5 independent measures to confirm recovery: (i) direct counting of importing cells in our GFP-PTS1 reporter assay, (ii) assessing proportions of processed (imported) endogenous peroxisome matrix enzymes, (iii) determining PEX5 sub-cellular localization, (iv) measuring VLCFA and plasmalogen levels, and (v) quantifying PEX1, PEX6 and PEX5 protein levels. In these experiments, diosmetin was the most active flavonoid and combination treatments of diosmetin with betaine generally produced an additive effect at lower compound doses.

The use of multiple measures of peroxisome recovery is essential to confirm the global effect of a candidate compound. As in the case of improved PEX5 localization (IF) concurrent with improved protein amounts (immunoblotting), multiple measures can corroborate findings. In this example, redistribution of PEX5 from the peroxisome to the cytosol also supports a drug mechanism of action that involves functional improvement of the PEX1-6-26 complex. Adding more outcome measures to our repertoire, and thus additional validation tests for compound efficacy, would further strengthen our drug development platform and potentially assist in informing mechanism of action of a therapeutic compound. Our laboratory recently optimized a catalase distribution assay, which quantifies the proportion cytosolic versus peroxisomal catalase [adapted from [374]], and developed a method for quantifying peroxisome size and number [adapted from [375]]. Catalase, one of the most sensitive markers of peroxisome dysfunction due to its non-canonical PTS1 signal (-NAKL), is mainly a peroxisomal enzyme, but in ZSD cells it is largely cytosolic. In ZSD, there is typically reduced numbers of enlarged peroxisomes. Effective treatment should thus also increase the amount of peroxisomal catalase, increase peroxisome numbers, and decrease size. In addition, subcellular redox potential or reactive oxygen species amount could be used as a post-therapeutic readout [376].

132

As a mechanism of drug action, we proposed that flavonoids could act as pharmacological chaperones for the PEX1-G843D protein and/or the corresponding PEX1-6-26 exportomer complex. Given the predicted position of G843 in the ATP pocket (Supplemental Figure 2.2B), and the literature supporting the binding of flavonoids at or near ATP-binding sites [reviewed in [313, 314]], we speculate that diosmetin interacts dynamically with the ATP-binding sites of PEX1 and/or the PEX1-PEX6 complex, resulting downstream in increased peroxisome import. In addition to our lab, other research groups have successfully recovered peroxisome functions in vitro using chaperone compounds. For example, the chemical chaperone arginine, in addition to improving peroxisome function in PEX1-G843D-mediated ZSD, improved peroxisome biogenesis and functions in patient cell lines with missense mutations in PEX6 and PEX12, as well as another missense mutation in PEX1 (P274L) [290, 291]. These results highlight that candidate compounds may be effective for multiple PEX mutations, which is discussed later.

5.1.2 The study of Pex1-G844D retinopathy identifies therapeutic outcome measures

The PEX1-G844D mouse model for mild ZSD can be used to better understand disease pathophysiology and identify outcome measures for in vivo preclinical trials. Vision loss is experienced by almost all people with ZSD, and is prioritized by the Global Foundation for Peroxisome Disorders as a major barrier to quality of life in these patients. We thus prioritized vision as a target system in our experiments.

I have characterized the progression of PEX1-G844D mouse retinopathy by measuring electrophysiology (ffERG, VEP), visual acuity (OKN), peroxisome protein levels, and biochemical metabolites. I also examined retinal structure and cell types using histology, immunohistochemistry, and electron microscopy. These studies enabled identification of which functional and structural visual components were affected first and most severely, and which remained unaffected in the PEX1-G844D model, thus paving the way for subsequent preclinical trials aimed at ameliorating vision.

Still unclear is how well the human versus murine ophthalmic manifestations correlate. Retrospective and prospective natural history studies in ZSD patients will help address this and are discussed later.

133

5.1.3 Gene augmentation therapy is a viable treatment option for ZSD

By delivering a functional gene to a tissue of interest, gene therapy targets the root cause of monogenic hereditary diseases. As a small, accessible, immuno-privileged compartment, the eye has been at the forefront of gene replacement therapies. In the PEX1-G844D mouse model, AAV8-mediated PEX1 augmentation has proved a safe approach for recovering retinal function, with the possibility of also improving functional vision. Efficacy in young adult animals with established retinopathy holds promise for the translatability of this therapy to human patients, which may have experienced visual decline prior to diagnosis. Our studies show that in principle, gene augmentation could be a viable therapeutic approach for ZSD, or more broadly peroxisome disorders, regardless of affected gene or tissue type.

5.2 GENERAL DISCUSSION

Over the past 50 years, an intense, multidisciplinary effort has led to elucidating the fundamental biochemical, pathophysiological and molecular bases for ZSD. This knowledge has paved the way for therapeutic interventions and trials. Below are summarized therapeutic approaches either in development or of potential significance for ZSD.

5.2.1 Additional pharmacological therapies under development for ZSD

5.2.1.1 Pharmacological enhancement of activity of a defective PEX protein

Given its milder phenotype and higher prevalence relative to other PEX mutations, much drug development for ZSD has focused on improving PEX1-G843D function. Our lab designed a high throughput cell-based phenotype assay using a patient fibroblast cell line with this allele and expressing a GFP-tagged PTS1 reporter that was cytosolic at baseline [290]. After compound treatment the cells were evaluated for recovery of GFP-PTS1 peroxisome importation. The advantages of using a cell-based assay include demonstrating that the molecules can cross at least one cell membrane and are functional in the cellular milieu. By measuring the downstream effect of peroxisome protein import recovery, this assay should potentially enable the identification of any drug recovery mechanism. One more recent hit compound is naltriben, a delta opioid receptor antagonist. Naltriben recovered peroxisome import in PEX1-G843D primary fibroblasts nearly as well as diosmetin, but improvement in downstream biochemical

134

metabolites was relatively modest. In addition, some recovery was seen in PEX1 null cells, suggesting Naltriben may not be acting directly on PEX1 [unpublished data from Dr. Joseph Hacia’s lab (USC) and our lab]. Compound library screening using the above-mentioned assay, and the development of alternate screening assays for peroxisome recovery are ongoing at the National Centre for Advancing Translational Sciences (NCATS) in Rockville, Maryland, and at Dr. Hans Waterham’s lab at the Academic Medical Centre of the University of Amsterdam. Hit compounds identified from these screens can be further developed into lead compounds if efficacious in animal models.

5.2.1.2 Regulation of peroxisome numbers

ZSD cells typically exhibit fewer, partially functional peroxisomes. Hypothetically, increasing the number of these organelles should improve overall cell functions. Peroxisome proliferation in ZSD fibroblasts has been achieved using 4-phenylbutyrate and related compounds that act independently of PPARα via PEX11 stimulation [377, 378]. In cell lines from patients with intermediate and milder phenotypes, peroxisome proliferation results in improvement in peroxisome enzyme functions [377]. High throughput cell-based phenotype assays to identify additional compounds that can induce human peroxisome proliferation have been reported [74].

An alternative approach to improving peroxisome numbers is preventing the autophagic degradation of peroxisomes, or ‘pexophagy’. Recent evidence highlights the importance of the peroxisome ‘exportomer’ complex in preventing pexophagy in yeast [170]. In human cells, this process depends on the accumulation of ubiquitinated PEX5, which occurs in the case of PEX1, PEX6, or PEX26 deficiency [379], suggesting pexophagy inhibition as a possible approach to treat mutations in these genes. In fact, treatment with three small molecule autophagy inhibitors led to improved peroxisome numbers in HELA cells [171]. The most promising of these compounds, the antimalarial chloroquine, improved peroxisome numbers in PEX1 null and PEX1-G843D patient fibroblasts, but had no effect in PEX16 null cells, which cannot assemble peroxisome membranes. Although VLCFA levels improved upon chloroquine treatment, plasmalogen levels did not. Measurement of other peroxisome functions and testing in the Pex1- G844D mouse is ongoing [personal correspondence with Dr. Kim, University of Toronto]. However, caution should be exercised when considering a known pro-oxidant such as

135

chloroquine for the treatment of peroxisome dysfunction, in which free radical detoxification is impaired [380, 381]. Retinopathy is also a known side-effect of prolonged chloroquine treatment [382].

5.2.1.3 Nonsense suppression

The development of safe nonsense suppressor therapies might benefit ZSD patients with the appropriate mutations. For example, in fibroblast lines from patients with PEX2 and PEX12 mutations that produce stable nonsense transcripts, the nonsense suppressor aminoglycoside drug, G418 (geneticin), was shown to promote peroxisome recovery [383]. However, the nonsense suppressor PTC124 had no effect.

5.2.2 Alternative therapeutic approaches for ZSD

5.2.2.1 Gene therapy

In addition to its use in ocular disorders, gene replacement therapy is in development for inner ear, neurologic, and liver disorders [384-387]. For example, AAV2-mediated delivery of the wild-type Ush1c gene to the cochlea restored sensory cell function, complex auditory function, and recovered hearing and balance in a mouse model for Usher Syndrome [388]. In the clinic, several hepatocyte-directed gene therapy trials using AAV-delivered human factor IX have been successful in treating haemophilia B [384]. Intracerebroventricular delivery of ASPA cDNA using an AAV9 vector was well-tolerated over nine months and had positive white matter and motor/cognitive outcomes in a patient with Canavan disease, a severe childhood leukodystrophy [389]. As all of these organ systems are affected in ZSD, they could each be targeted by gene therapy.

5.2.2.1 Gene editing

With the advent of CRISPR-Cas9 gene editing technology came the promise of genetic correction for inherited disorders. The ongoing improvements in these technologies to increase fidelity and minimize off-target effects will eventually enable their application for therapeutic use [390]. Small accessible compartments such as the eye or cochlea will likely be the initial target systems until systemic administration becomes possible. For example, Hung et al. (2016) used AAV2 to deliver CRISPR/Cas for the genetic modification of murine retinal cells, and achieved 84% efficiency in successfully targeted cells with no identified negative effects five

136

weeks after administration [391]. Moreover, new editing techniques allow for the conversion of single DNA base pairs (A-T to C-G or vice versa) without double stranded DNA cleavage [392, 393]. This approach could be applicable for correcting genetic disorders caused by single base pair substitutions, such as the PEX1 c.[2528G>A] mutation that translates to PEX1-G843D.

5.2.3 Interventional clinical trials

A six-month pilot, open-label trial assessing the safety and efficacy of betaine in ten ZSD patients with PEX1-G843D genotypes was completed in 2015 (clinical trials.gov NCT01838941). Betaine was shown to be safe in this patient population. The clinical endpoint was the effect of betaine on multiple peroxisome metabolites from blood. Although the results of this trial were inconclusive, they revealed large variations in peroxisome metabolite levels at the pre-trial baseline both within a single patient and among patients with the same genotypes. This effect obscured any demonstration of a subtle or even moderate drug effect. This study highlighted the need to identify biomarkers with less variability, as well as the need to incorporate functional clinical endpoints into trials. The need for accurate clinical endpoints underscores the importance of establishing a natural history of disease prior to any future trials.

5.2.3.1 Natural history studies

Establishing a natural history of disease progression is especially challenging when studying a rare genetic disorder with a spectrum of manifestations, but is nonetheless critical for selecting meaningful clinical endpoints. Our lab is currently conducting a retrospective natural history study of ZSD in which participant medical record data is systematically entered into a custom- designed database (clinicaltrial.gov NCT01668186). This database can be mined for answers to specific questions, and has been used to inform liver, brain, and growth pathologies, allowing for genotype-phenotype correlations. We are also beginning a prospective natural history study of the ophthalmic disease, in which participants will undergo yearly examinations including ffERG, VEP, OCT, pupillometry, visual fields, and visual acuity. This will, for the first time, allow the systematic study of several ophthalmic observations in the same person over time. Finally, a proxy-reported quality of life questionnaire is currently being administered to caregivers of ZSD patients to assess communication, medical care, emotional distress and well-being, role function, family interaction, parenting, and disability-related support (clinicaltrials.org NCT03440905).

137

This questionnaire could potentially be used to assess quality of life in a therapeutic clinical trial before and after intervention.

5.2.3.2 Newborn screening

The detection of elevated VLCFAs in newborn blood spots, measured by C26:0 lyso-PC by LC– MS/MS was validated as a diagnostic approach for X-linked adrenoleukodystrophy (X-ALD) [394, 395], a single enzyme/protein defect that impairs VLCFA entry to peroxisomes [396]. X- ALD screening was recently included in the United States Recommended Uniform Screening Panel [397], and is now being adopted by several states. Newborn screening for X-ALD will also detect the majority of PBD-ZSD cases that feature elevated blood VLCFA levels, thereby facilitating early diagnosis of relatively asymptomatic infants and more timely institution of supportive management. It will also enable determination of accurate incidence figures and continued expansion of the clinical phenotype as variant patients are identified, all of which inform the development of new therapies. Finally, the early detection and wider capture of affected individuals will maximize the patient population available to participate in clinical trials.

5.2.4 An expanding role for peroxisomes in disease

In recent years, the role of peroxisomes in health and disease has broadened beyond the traditional biochemical pathways. Emerging evidence highlights the importance of peroxisomes in more prevalent diseases that afflict a much greater population than ZSD. Several recent studies have shown a link between the upregulation of peroxisomal metabolic processes and cancer progression [398], and peroxisomes have been shown to mediate drug resistance in lymphoma cells [399]. Peroxisome dysregulation is present in Alzheimer’s disease, in that postmortem brains exhibit reduced plasmalogens and elevated VLCFAs, with exacerbated changes correlating to pathological severity [400]. Finally, peroxisome processes such as fatty- acid oxidation, free radical metabolism, and pexophagy are reportedly disturbed in ischemic acute kidney injury, diabetic nephropathy, and septic acute kidney injury [401]. The growing awareness of peroxisome importance in non-Mendelian diseases may break open the field of peroxisome modulation as a therapeutic approach for a broad range of conditions.

138

5.3 FUTURE DIRECTIONS

5.3.1 Further development of flavonoid and other pharmacological therapies

Diosmetin has proved effective in recovering peroxisome functions to near normal levels in PEX1-G843D homozygous patient cells, with the addition of betaine producing a synergistic effect for some measures. The mechanism of action of diosmetin should be investigated, and our hypothesis tested of the drug binding at or near the ATP binding sites of the PEX1-G843D-PEX6 complex. Techniques such as surface plasmon resonance (SPR) or nuclear magnetic resonance spectroscopy (NMR) could be used to test whether diosmetin binds the purified protein complex. Together with the structure activity relationships already delineated in our studies, further understanding of mechanism will ultimately inform medicinal chemistry to optimize structure for efficacy. This would allow for designing a more potent and selective molecule that could be used at lower dose.

Diosmetin and betaine should be tested in vivo using the PEX1-G844D mouse model. This experiment would allow proof-of-concept for improvement in clinical phenotype after drug treatment. Outcome measures could include improvement in visual functions and liver pathology, and could be paired with biomarkers for peroxisome metabolites and liver functions, as well as histology. Contrary to human cell observations, in PEX1-G844D mouse fibroblasts, I found that betaine was more effective than diosmetin in recovering PTS1 import, PEX5 localization, and biochemical metabolites in primary mouse fibroblasts, though both compounds were effective [unpublished results]. Since PEX1-G844D (unlike PEX1-G843D) is apparently not degraded, drug action on the peroxisome ‘exportomer’ complex may vary between the mouse and human and thus limit translatability of this preclinical model.

Although PEX1-G843D is the most common, ZSD is caused by diverse mutations in any of 13 PEX genes. It is thus important to test the efficacy of our candidate drugs on other alleles to hopefully benefit other patients. To address this, I have been using a lentivirus to deliver GFP- PTS1 to primary fibroblasts lines from patients with various PEX mutations, and scoring recovery of PTS1 import after treatment with candidate compounds. Positive lines are then assessed for PEX5 localization, catalase solubility, and peroxisome metabolites. With the help of summer students, I have tested the effect of betaine and diosmetin in various PEX1, PEX6, and PEX26 lines, using a PEX16 null line that does not assemble peroxisome membranes as one of

139

our controls. To complete this analysis, the effect of betaine and diosmetin must be assessed beyond the ‘exportomer’ complex, on cell lines with missense alleles in the remaining PEX genes (i.e. PEX2/10/12, PEX13/14, PEX5/7, PEX3/16/19). We will also test cell lines from X- ALD patients with missense alleles in ABCD1, a peroxisomal ATPase, which mediates the entry of VLCFA into the peroxisome. The results of these analyses will inform the mechanism by which these compounds rescue peroxisome function, and partly test our hypothesis that diosmetin acts as a pharmacological chaperone that specifically interacts with the PEX1/PEX6 AAA ATPase complex.

5.3.2 Deeper phenotyping of the PEX1-G844D mouse model

The visual system

In the PEX1-G844D mouse model, the largest phenotyping effort thus far involves visual function and the retina. These efforts have allowed us to identify additional parameters to assess. Optical coherence tomography (OCT) examinations would allow a pan-retinal view of retinal layers and thickness in the same animal over time, possibly revealing more than histology, which focuses on a single retinal section at a single time point. To continue structural analysis of the retina, immunohistochemistry could be used to label additional cell components, such as synaptic and axonal terminals of specific cell types. TEM should be used to identify whether inner segment disorganization and mitochondrial enlargement occur in early life, and immunogold labelling of peroxisomes can identify alterations in peroxisome size, shape, and import capacity. Extending away from the neural retina, structures such as the eye anterior chamber, retinal pigment epithelium and optic nerve should also be analyzed for defects, as these structures may contribute to pathology.

To expand our understanding of visual function in the mouse, several additional characterizations should be performed. This includes a natural history of visual acuity (OKN), with measures expanded to include contrast sensitivity and colour perception. Multifocal ERG (mfERG) would allow monitoring the response at a specific retinal location over time, in contrast to the pan-retinal average measured by the ffERG. It may be that response is maintained differently at different locations, which would inform injection targeting for gene therapy. To better assess retinocortical function, optic nerve response could be tested directly.

140

The mechanism of retinopathy could be tested by comparing the functional and structural effects of bright light exposure or protection from light in the mutant versus control animals. Due to their peroxisomal dysfunction, it may be that light-induced retinopathy is exacerbated or protection from light is protective in PEX1-G844D animals. More extensive lipidomics analyses could identify aberrant biochemical (including mitochondrial) pathways that may contribute to the retinal or optic nerve phenotype. Given the enlarged mitochondria in our model, and that mitochondrial dysfunction has been associated with various retinal degenerative diseases [reviewed in [345]], it is possible that mitochondrial dysregulation is contributing to PEX1- G844D mouse pathology. Respiratory chain function and markers of mitochondrial fission and/or fusion could be measured to elucidate this link. Taken together, these more mechanistic studies could suggest alternative approaches for targeting vision loss in ZSD.

Finally, the planned prospective natural history study of vision in ZSD will allow a comparison of human to mouse visual function. To further complement mouse studies, a comparison of retinal structure or biochemistry could be performed using ZSD autopsy samples.

Other organ systems

Characterization of the natural history of the liver disease in the PEX1-G844D mouse model is already underway, and includes analysis of gross liver pathology, serum biomarkers, liver biochemistry, histology, and immunohistochemistry over time. Bone and auditory characterizations have been started. Although no neurological deficits have been identified thus far, a comprehensive study of central and peripheral nervous system has not been performed and warrants investigation. Extensive phenotyping of our model will identify additional endpoints for future preclinical trials using this model. The utility of developing other models for preclinical trials is being discussed (personal communication with Dr. Joseph Hacia, University of Southern California, and Dr. Cat Lutz, Jackson Labs)

5.3.3 Retinal gene therapy continued

Given the success of the first proof-of-concept retinal gene therapy trial in the PEX1-G844D mouse model, this therapeutic approach warrants further development. Dose finding experiments to determine the minimum viral dose that produces maximal retina coverage must be performed prior to the next preclinical trials. The subsequent mouse experiments should utilize a clinical

141

vector without an epitope tag. This necessitates developing a method to confirm the presence of vector-delivered PEX1 protein in the mouse retina. To address this, one could use imaging mass spectrometry to detect human versus mouse PEX1, which would also allow the correlation of vector delivered protein with lipidomics in consecutive sections.

Although it has not been required for all murine-based therapies that have progressed to clinic, statistically significant improvement in functional vision would strengthen the case for retinal gene therapy in ZSD. This would involve using expanded visual acuity measures and regularly following visual acuity in each mouse over the course of treatment.

5.4 CONCLUDING SUMMARY

I have shown in this thesis that diosmetin recovers five independent measures of peroxisome function in several PEX1-G843D primary patient fibroblast lines, and hypothesize that it is acting as a pharmacological chaperone for PEX1-G843D. I have also shown that the visual deficit in the PEX1-G844D model for mild ZSD is marked by low cone cell function early in life, with rod function gradually diminishing. Structural defects occur primarily at the photoreceptor inner segment. Finally, I have shown that rAAV-mediated PEX1 gene augmentation improves retinal electrophysiological function in the Pex1-G844D mouse model.

5.5 ORIGINAL CONTRIBUTIONS

The following are the original contributions that I have made to developing therapies for Peroxisome Biogenesis Disorders of the Zellweger Spectrum.

Chapter II: In this chapter, I demonstrate that diosmetin recovers five independent measures of peroxisome function in several primary patient fibroblast lines with at least one PEX1-G843D allele. This work shows for the first time that cell treatment with betaine and/or diosmetin recovers PTS1 enzyme import to the peroxisome, PTS2 enzyme processing, PEX5 localization (a direct function of PEX1), improvement in downstream peroxisome metabolites, and increased PEX1, PEX6, and PEX5 levels.

142

Chapter III: In this chapter, I showed that retinopathy in the PEX1-G844D mouse model is marked by low cone cell function and number early in life, with gradually decreasing rod cell function concurrent with inner segment disorganization and enlarged mitochondria. Structural defects at the inner retina occur later in the form of bipolar cell degradation, while other inner retinal cells are apparently preserved. I also demonstrated low functional vision in the PEX1- G844D mouse and relatively well preserved retino-cortical function. This work shows for the first time a thorough analysis of retinal structure and function in this model over time, and is the first report of visual acuity and visual evoked potentials in a ZSD mouse model. It is also the first report of PEX1-G844D, PEX6, and PEX5 protein levels in the PEX1-G843D mouse model, showing that contrary to human PEX1-G843D cells, the amounts of these proteins are normal compared to non-ZSD controls.

Chapter IV: In this chapter, I showed that AAV8.CMV.hPEX1.HA is efficiently expressed and localized similarly to endogenous PEX1 in the PEX1-G844D mouse retina following subretinal injection. I also demonstrated that rAAV-mediated PEX1 gene augmentation at early and later disease stages is sufficient to improve retinal electrophysiological response in the PEX1-G844D mouse model. This is the first in vivo report of a gene therapy intervention for ZSD.

143

REFERENCES

1. Argyriou, C., M.D. D'Agostino, and N. Braverman, Peroxisome biogenesis disorders. Transl Sci Rare Dis, 2016. 1(2): p. 111-144. 2. Bowers, W.E., Christian de Duve and the discovery of lysosomes and peroxisomes. Trends Cell Biol, 1998. 8(8): p. 330-3. 3. Zellweger, H., et al., History of the cerebrohepatorenal syndrome of Zellweger and other peroxisomal disorders. South Med J, 1988. 81(3): p. 357-64. 4. Opitz J. M. ZuRhein, G.M., et al., The Zellweger syndrome (cerebro-hepato-renal syndrome). Birth Defects Orig. Art. Ser. , 1969. 2(2): p. 144-160. 5. Goldfischer, S., et al., Peroxisomal and mitochondrial defects in the cerebro-hepato-renal syndrome. Science, 1973. 182(4107): p. 62-4. 6. Heymans, H.S., et al., Severe plasmalogen deficiency in tissues of infants without peroxisomes (Zellweger syndrome). Nature, 1983. 306(5938): p. 69-70. 7. Heymans, H.S., et al., Rhizomelic chondrodysplasia punctata: another peroxisomal disorder. N Engl J Med, 1985. 313(3): p. 187-8. 8. Moser, A.E., et al., The cerebrohepatorenal (Zellweger) syndrome. Increased levels and impaired degradation of very-long-chain fatty acids and their use in prenatal diagnosis. N Engl J Med, 1984. 310(18): p. 1141-6. 9. Steinberg, S., et al., The PEX Gene Screen: molecular diagnosis of peroxisome biogenesis disorders in the Zellweger syndrome spectrum. Mol Genet Metab, 2004. 83(3): p. 252-63. 10. Preuss, N., et al., PEX1 mutations in complementation group 1 of Zellweger spectrum patients correlate with severity of disease. Pediatr Res, 2002. 51(6): p. 706-14. 11. Walter, C., et al., Disorders of peroxisome biogenesis due to mutations in PEX1: phenotypes and PEX1 protein levels. Am J Hum Genet, 2001. 69(1): p. 35-48. 12. Ebberink, M.S., et al., Genetic classification and mutational spectrum of more than 600 patients with a Zellweger syndrome spectrum disorder. Hum Mutat, 2011. 32(1): p. 59-69. 13. Yik, W.Y., et al., Identification of novel mutations and sequence variation in the Zellweger syndrome spectrum of peroxisome biogenesis disorders. Hum Mutat, 2009. 30(3): p. E467-80. 14. Wanders, R.J. and H.R. Waterham, Biochemistry of mammalian peroxisomes revisited. Annu Rev Biochem, 2006. 75: p. 295-332. 15. Ferdinandusse, S., et al., Studies on the metabolic fate of n-3 polyunsaturated fatty acids. J Lipid Res, 2003. 44(10): p. 1992-7. 16. Moore, S.A., et al., Docosahexaenoic acid synthesis in human skin fibroblasts involves peroxisomal retroconversion of tetracosahexaenoic acid. J Lipid Res, 1995. 36(11): p. 2433-43. 17. Wanders, R.J., et al., Peroxisomal very long-chain fatty acid beta-oxidation in human skin fibroblasts: activity in Zellweger syndrome and other peroxisomal disorders. Clin Chim Acta, 1987. 166(2-3): p. 255-63. 18. Pahan, K., B.T. Smith, and I. Singh, Epoxide hydrolase in human and rat peroxisomes: implication for disorders of peroxisomal biogenesis. J Lipid Res, 1996. 37(1): p. 159-67. 19. Arand, M., et al., An impaired peroxisomal targeting sequence leading to an unusual bicompartmental distribution of cytosolic epoxide hydrolase. FEBS Lett, 1991. 294(1-2): p. 19-22. 20. Waechter, F., et al., Epoxide hydrolase activity in isolated peroxisomes of mouse liver. FEBS Lett, 1983. 158(2): p. 225-8. 21. Nguyen, S.D., M. Baes, and P.P. Van Veldhoven, Degradation of very long chain dicarboxylic polyunsaturated fatty acids in mouse hepatocytes, a peroxisomal process. Biochim Biophys Acta, 2008. 1781(8): p. 400-5.

144

22. Singh, H. and A. Poulos, Distinct long chain and very long chain fatty acyl CoA synthetases in rat liver peroxisomes and microsomes. Arch Biochem Biophys, 1988. 266(2): p. 486-95. 23. Street, J.M., et al., Metabolism of saturated and polyunsaturated fatty acids by normal and Zellweger syndrome skin fibroblasts. Biochem J, 1989. 260(3): p. 647-55. 24. Uchiyama, A., et al., Molecular cloning of cDNA encoding rat very long-chain acyl-CoA synthetase. J Biol Chem, 1996. 271(48): p. 30360-5. 25. Schepers, L., et al., Presence of three acyl-CoA oxidases in rat liver peroxisomes. An inducible fatty acyl-CoA oxidase, a noninducible fatty acyl-CoA oxidase, and a noninducible trihydroxycoprostanoyl-CoA oxidase. J Biol Chem, 1990. 265(9): p. 5242-6. 26. Ferdinandusse, S., et al., A novel case of ACOX2 deficiency leads to recognition of a third human peroxisomal acyl-CoA oxidase. Biochim Biophys Acta, 2018. 1864(3): p. 952-958. 27. Ferdinandusse, S., et al., Bile acids: the role of peroxisomes. J Lipid Res, 2009. 50(11): p. 2139-47. 28. Baes, M., et al., Inactivation of the peroxisomal multifunctional protein-2 in mice impedes the degradation of not only 2-methyl-branched fatty acids and bile acid intermediates but also of very long chain fatty acids. J Biol Chem, 2000. 275(21): p. 16329-36. 29. Ferdinandusse, S., et al., Identification of the peroxisomal beta-oxidation enzymes involved in the degradation of long-chain dicarboxylic acids. J Lipid Res, 2004. 45(6): p. 1104-11. 30. Ferdinandusse, S., et al., Identification of the peroxisomal beta-oxidation enzymes involved in the degradation of leukotrienes. Biochem Biophys Res Commun, 2002. 293(1): p. 269-73. 31. Wanders, R.J., et al., Peroxisomal fatty acid beta-oxidation in relation to the accumulation of very long chain fatty acids in cultured skin fibroblasts from patients with Zellweger syndrome and other peroxisomal disorders. J Clin Invest, 1987. 80(6): p. 1778-83. 32. Wanders, R.J., et al., Direct demonstration that the deficient oxidation of very long chain fatty acids in X-linked adrenoleukodystrophy is due to an impaired ability of peroxisomes to activate very long chain fatty acids. Biochem Biophys Res Commun, 1988. 153(2): p. 618-24. 33. Su, H.M., et al., Peroxisomal straight-chain Acyl-CoA oxidase and D-bifunctional protein are essential for the retroconversion step in docosahexaenoic acid synthesis. J Biol Chem, 2001. 276(41): p. 38115-20. 34. Jakobs, B.S. and R.J. Wanders, Conclusive evidence that very-long-chain fatty acids are oxidized exclusively in peroxisomes in human skin fibroblasts. Biochem Biophys Res Commun, 1991. 178(3): p. 842-7. 35. Mize, C.E., et al., A pathway for oxidative degradation of phytanic acid in mammals. Biochem Biophys Res Commun, 1966. 25(3): p. 359-65. 36. ELDJARN, L., HEREDOPATHIA ATACTICA POLYNEURITIFORMIS (REFSUM'S DISEASE)--A DEFECT IN THE OMEGA-OXIDATION MECHANISM OF FATTY ACIDS. Scand J Clin Lab Invest, 1965. 17: p. 178- 81. 37. van den Brink, D.M., et al., Characterization of the final step in the conversion of phytol into phytanic acid. J Biol Chem, 2005. 280(29): p. 26838-44. 38. Jansen, G.A., et al., Identification of pristanal dehydrogenase activity in peroxisomes: conclusive evidence that the complete phytanic acid alpha-oxidation pathway is localized in peroxisomes. Biochem Biophys Res Commun, 2001. 283(3): p. 674-9. 39. Mukherji, M., et al., Structure-function analysis of phytanoyl-CoA 2-hydroxylase mutations causing Refsum's disease. Hum Mol Genet, 2001. 10(18): p. 1971-82. 40. Foulon, V., et al., Purification, molecular cloning, and expression of 2-hydroxyphytanoyl-CoA lyase, a peroxisomal thiamine pyrophosphate-dependent enzyme that catalyzes the carbon- carbon bond cleavage during alpha-oxidation of 3-methyl-branched fatty acids. Proc Natl Acad Sci U S A, 1999. 96(18): p. 10039-44.

145

41. Schmitz, W., R. Fingerhut, and E. Conzelmann, Purification and properties of an alpha- methylacyl-CoA racemase from rat liver. Eur J Biochem, 1994. 222(2): p. 313-23. 42. Steinberg, D., et al., Conversion of U-C14-phytol to phytanic acid and its oxidation in heredopathia atactica polyneuritiformis. Biochem Biophys Res Commun, 1965. 19(6): p. 783-9. 43. Brown, J., et al., Diet and Refsum’s disease. The determination of phytanic acid and phytol in certain foods and the application of this knowledge to the choice of suitable convenience foods for patients with Refsum’s disease Journal of Human Nutrition and Dietetics, 1993. 6: p. 295-305. 44. Wanders, R.J., J. Komen, and S. Ferdinandusse, Phytanic acid metabolism in health and disease. Biochim Biophys Acta, 2011. 1811(9): p. 498-507. 45. Braverman, N.E. and A.B. Moser, Functions of plasmalogen lipids in health and disease. Biochim Biophys Acta, 2012. 1822(9): p. 1442-52. 46. da Silva, T.F., et al., The importance of ether-phospholipids: a view from the perspective of mouse models. Biochim Biophys Acta, 2012. 1822(9): p. 1501-8. 47. MARINETTI, G.V. and J. ERBLAND, The structure of pig heart plasmalogen. Biochim Biophys Acta, 1957. 26(2): p. 429-30. 48. Han, X.L. and R.W. Gross, Plasmenylcholine and phosphatidylcholine membrane bilayers possess distinct conformational motifs. Biochemistry, 1990. 29(20): p. 4992-6. 49. Rog, T. and A. Koivuniemi, The biophysical properties of ethanolamine plasmalogens revealed by atomistic molecular dynamics simulations. Biochim Biophys Acta, 2016. 1858(1): p. 97-103. 50. Hermetter, A., et al., Influence of plasmalogen deficiency on membrane fluidity of human skin fibroblasts: a fluorescence anisotropy study. Biochim Biophys Acta, 1989. 978(1): p. 151-7. 51. Farooqui, A.A. and L.A. Horrocks, Plasmalogens, phospholipase A2, and docosahexaenoic acid turnover in brain tissue. J Mol Neurosci, 2001. 16(2-3): p. 263-72; discussion 279-84. 52. Brites, P., et al., Alkyl-glycerol rescues plasmalogen levels and pathology of ether-phospholipid deficient mice. PLoS One, 2011. 6(12): p. e28539. 53. Bams-Mengerink, A.M., et al., MRI of the brain and cervical spinal cord in rhizomelic chondrodysplasia punctata. Neurology, 2006. 66(6): p. 798-803; discussion 789. 54. Bams-Mengerink, A.M., et al., The neurology of rhizomelic chondrodysplasia punctata. Orphanet J Rare Dis, 2013. 8: p. 174. 55. Braverman, N., et al., Mutation analysis of PEX7 in 60 probands with rhizomelic chondrodysplasia punctata and functional correlations of genotype with phenotype. Hum Mutat, 2002. 20(4): p. 284-97. 56. Leroy, J.G., et al., Diagnostic work-up of a peroxisomal patient. J Inherit Metab Dis, 1995. 18 Suppl 1: p. 214-22. 57. Mandel, H., et al., Plasma lipoproteins and monocyte-macrophages in a peroxisome-deficient system: study of a patient with infantile refsum disease. J Inherit Metab Dis, 1992. 15(5): p. 774- 84. 58. Poll-The, B.T., et al., Infantile Refsum disease: an inherited peroxisomal disorder. Comparison with Zellweger syndrome and neonatal adrenoleukodystrophy. Eur J Pediatr, 1987. 146(5): p. 477-83. 59. Kovacs, W.J., et al., Disturbed cholesterol homeostasis in a peroxisome-deficient PEX2 knockout mouse model. Mol Cell Biol, 2004. 24(1): p. 1-13. 60. Faust, P.L. and W.J. Kovacs, Cholesterol biosynthesis and ER stress in peroxisome deficiency. Biochimie, 2014. 98: p. 75-85. 61. Kunze, M., et al., Structural requirements for interaction of peroxisomal targeting signal 2 and its receptor PEX7. J Biol Chem, 2011. 286(52): p. 45048-62.

146

62. Chowdhary, G., et al., Non-canonical peroxisome targeting signals: identification of novel PTS1 tripeptides and characterization of enhancer elements by computational permutation analysis. BMC Plant Biol, 2012. 12: p. 142. 63. Lingner, T., A.R. Kataya, and S. Reumann, Experimental and statistical post-validation of positive example EST sequences carrying peroxisome targeting signals type 1 (PTS1). Plant Signal Behav, 2012. 7(2): p. 263-8. 64. Mizuno, Y., et al., Predicted mouse peroxisome-targeted proteins and their actual subcellular locations. BMC Bioinformatics, 2008. 9 Suppl 12: p. S16. 65. Wiese, S., et al., Proteomics characterization of mouse kidney peroxisomes by tandem mass spectrometry and protein correlation profiling. Mol Cell Proteomics, 2007. 6(12): p. 2045-57. 66. Wiese, S., et al., Comparative profiling of the peroxisomal proteome of wildtype and Pex7 knockout mice by quantitative mass spectrometry. International Journal of Mass Spectrometry, 2012. 312: p. 30-40. 67. Ashmarina, L.I., et al., 3-Hydroxy-3-methylglutaryl coenzyme A lyase: targeting and processing in peroxisomes and mitochondria. J Lipid Res, 1999. 40(1): p. 70-5. 68. Ferdinandusse, S., et al., Subcellular localization and physiological role of alpha-methylacyl-CoA racemase. J Lipid Res, 2000. 41(11): p. 1890-6. 69. Wightman, P.J., et al., MLYCD mutation analysis: evidence for protein mistargeting as a cause of MLYCD deficiency. Hum Mutat, 2003. 22(4): p. 288-300. 70. Luo, B., et al., Protein quaternary structure and expression levels contribute to peroxisomal- targeting-sequence-1-mediated peroxisomal import of human soluble epoxide hydrolase. J Mol Biol, 2008. 380(1): p. 31-41. 71. Islinger, M., et al., Hitchhiking of Cu/Zn superoxide dismutase to peroxisomes--evidence for a natural piggyback import mechanism in mammals. Traffic, 2009. 10(11): p. 1711-21. 72. Leiper, J.M. and C.J. Danpure, The role of dimerization of alanine:glyoxylate aminotransferase 1 in its peroxisomal and mitochondrial import. Ann N Y Acad Sci, 1996. 804: p. 765-7. 73. Gonzalez, F.J. and Y.M. Shah, PPARalpha: mechanism of species differences and hepatocarcinogenesis of peroxisome proliferators. Toxicology, 2008. 246(1): p. 2-8. 74. Sexton, J.Z., et al., High content screening for non-classical peroxisome proliferators. Int J High Throughput Screen, 2010. 2010(1): p. 127-140. 75. Chang, C.C., et al., Metabolic control of peroxisome abundance. J Cell Sci, 1999. 112 ( Pt 10): p. 1579-90. 76. Schrader, M., et al., Proliferation and fission of peroxisomes - An update. Biochim Biophys Acta, 2016. 1863(5): p. 971-83. 77. Bagattin, A., L. Hugendubler, and E. Mueller, Transcriptional coactivator PGC-1alpha promotes peroxisomal remodeling and biogenesis. Proc Natl Acad Sci U S A, 2010. 107(47): p. 20376-81. 78. Schrader, M., M. Thiemann, and H.D. Fahimi, Peroxisomal motility and interaction with microtubules. Microsc Res Tech, 2003. 61(2): p. 171-8. 79. Neuhaus, A., et al., Why do peroxisomes associate with the cytoskeleton? Biochim Biophys Acta, 2016. 1863(5): p. 1019-26. 80. Agrawal, G. and S. Subramani, De novo peroxisome biogenesis: Evolving concepts and conundrums. Biochim Biophys Acta, 2016. 1863(5): p. 892-901. 81. Farr, R.L., et al., Peroxisome biogenesis in mammalian cells: The impact of genes and environment. Biochim Biophys Acta, 2016. 1863(5): p. 1049-60. 82. Dias, A.F., et al., The first minutes in the life of a peroxisomal matrix protein. Biochim Biophys Acta, 2016. 1863(5): p. 814-20. 83. Emmanouilidis, L., et al., Structural biology of the import pathways of peroxisomal matrix proteins. Biochim Biophys Acta, 2016. 1863(5): p. 804-13.

147

84. Gould, S.J., et al., A conserved tripeptide sorts proteins to peroxisomes. J Cell Biol, 1989. 108(5): p. 1657-64. 85. Gould, S.J., G.A. Keller, and S. Subramani, Identification of peroxisomal targeting signals located at the carboxy terminus of four peroxisomal proteins. J Cell Biol, 1988. 107(3): p. 897-905. 86. Gould, S.G., G.A. Keller, and S. Subramani, Identification of a peroxisomal targeting signal at the carboxy terminus of firefly luciferase. J Cell Biol, 1987. 105(6 Pt 2): p. 2923-31. 87. Petriv, O.I., et al., A new definition for the consensus sequence of the peroxisome targeting signal type 2. J Mol Biol, 2004. 341(1): p. 119-34. 88. Swinkels, B.W., et al., A novel, cleavable peroxisomal targeting signal at the amino-terminus of the rat 3-ketoacyl-CoA thiolase. EMBO J, 1991. 10(11): p. 3255-62. 89. Osumi, T., et al., Amino-terminal presequence of the precursor of peroxisomal 3-ketoacyl-CoA thiolase is a cleavable signal peptide for peroxisomal targeting. Biochem Biophys Res Commun, 1991. 181(3): p. 947-54. 90. Mihalik, S.J., et al., Identification of PAHX, a Refsum disease gene. Nat Genet, 1997. 17(2): p. 185-9. 91. de Vet, E.C., B.T. van den Broek, and H. van den Bosch, Nucleotide sequence of human alkyl- dihydroxyacetonephosphate synthase cDNA reveals the presence of a peroxisomal targeting signal 2. Biochim Biophys Acta, 1997. 1346(1): p. 25-9. 92. Freitas, M.O., et al., PEX5 protein binds monomeric catalase blocking its tetramerization and releases it upon binding the N-terminal domain of PEX14. J Biol Chem, 2011. 286(47): p. 40509- 19. 93. Brocard, C., et al., The tetratricopeptide repeat-domain of the PAS10 protein of Saccharomyces cerevisiae is essential for binding the peroxisomal targeting signal-SKL. Biochem Biophys Res Commun, 1994. 204(3): p. 1016-22. 94. Terlecky, S.R., et al., The Pichia pastoris peroxisomal protein PAS8p is the receptor for the C- terminal tripeptide peroxisomal targeting signal. EMBO J, 1995. 14(15): p. 3627-34. 95. Gatto, G.J., et al., Peroxisomal targeting signal-1 recognition by the TPR domains of human PEX5. Nat Struct Biol, 2000. 7(12): p. 1091-5. 96. Braverman, N., et al., An isoform of pex5p, the human PTS1 receptor, is required for the import of PTS2 proteins into peroxisomes. Hum Mol Genet, 1998. 7(8): p. 1195-205. 97. Rehling, P., et al., The import receptor for the peroxisomal targeting signal 2 (PTS2) in Saccharomyces cerevisiae is encoded by the PAS7 gene. EMBO J, 1996. 15(12): p. 2901-13. 98. Dodt, G., et al., Domain mapping of human PEX5 reveals functional and structural similarities to Saccharomyces cerevisiae Pex18p and Pex21p. J Biol Chem, 2001. 276(45): p. 41769-81. 99. Gould, S.J., et al., Pex13p is an SH3 protein of the peroxisome membrane and a docking factor for the predominantly cytoplasmic PTs1 receptor. J Cell Biol, 1996. 135(1): p. 85-95. 100. Elgersma, Y., et al., The SH3 domain of the Saccharomyces cerevisiae peroxisomal membrane protein Pex13p functions as a docking site for Pex5p, a mobile receptor for the import PTS1- containing proteins. J Cell Biol, 1996. 135(1): p. 97-109. 101. Erdmann, R. and G. Blobel, Identification of Pex13p a peroxisomal membrane receptor for the PTS1 recognition factor. J Cell Biol, 1996. 135(1): p. 111-21. 102. Albertini, M., et al., Pex14p, a peroxisomal membrane protein binding both receptors of the two PTS-dependent import pathways. Cell, 1997. 89(1): p. 83-92. 103. Otera, H., et al., The mammalian peroxin Pex5pL, the longer isoform of the mobile peroxisome targeting signal (PTS) type 1 transporter, translocates the Pex7p.PTS2 protein complex into peroxisomes via its initial docking site, Pex14p. J Biol Chem, 2000. 275(28): p. 21703-14. 104. Rucktäschel, R., W. Girzalsky, and R. Erdmann, Protein import machineries of peroxisomes. Biochim Biophys Acta, 2011. 1808(3): p. 892-900.

148

105. Erdmann, R. and W. Schliebs, Peroxisomal matrix protein import: the transient pore model. Nat Rev Mol Cell Biol, 2005. 6(9): p. 738-42. 106. Meinecke, M., et al., The peroxisomal importomer constitutes a large and highly dynamic pore. Nat Cell Biol, 2010. 12(3): p. 273-7. 107. McNew, J.A. and J.M. Goodman, An oligomeric protein is imported into peroxisomes in vivo. J Cell Biol, 1994. 127(5): p. 1245-57. 108. Glover, J.R., D.W. Andrews, and R.A. Rachubinski, Saccharomyces cerevisiae peroxisomal thiolase is imported as a dimer. Proc Natl Acad Sci U S A, 1994. 91(22): p. 10541-5. 109. Freitas, M.O., et al., The peroxisomal protein import machinery displays a preference for monomeric substrates. Open Biol, 2015. 5(4): p. 140236. 110. Alencastre, I.S., et al., Mapping the cargo protein membrane translocation step into the PEX5 cycling pathway. J Biol Chem, 2009. 284(40): p. 27243-51. 111. Dodt, G. and S.J. Gould, Multiple PEX genes are required for proper subcellular distribution and stability of Pex5p, the PTS1 receptor: evidence that PTS1 protein import is mediated by a cycling receptor. J Cell Biol, 1996. 135(6 Pt 2): p. 1763-74. 112. Dammai, V. and S. Subramani, The human peroxisomal targeting signal receptor, Pex5p, is translocated into the peroxisomal matrix and recycled to the cytosol. Cell, 2001. 105(2): p. 187- 96. 113. Kiel, J.A., et al., Ubiquitination of the peroxisomal targeting signal type 1 receptor, Pex5p, suggests the presence of a quality control mechanism during peroxisomal matrix protein import. J Biol Chem, 2005. 280(3): p. 1921-30. 114. Prestele, J., et al., Different functions of the C3HC4 zinc RING finger peroxins PEX10, PEX2, and PEX12 in peroxisome formation and matrix protein import. Proc Natl Acad Sci U S A, 2010. 107(33): p. 14915-20. 115. Williams, C., et al., A conserved cysteine is essential for Pex4p-dependent ubiquitination of the peroxisomal import receptor Pex5p. J Biol Chem, 2007. 282(31): p. 22534-43. 116. Costa-Rodrigues, J., et al., The N terminus of the peroxisomal cycling receptor, Pex5p, is required for redirecting the peroxisome-associated peroxin back to the cytosol. J Biol Chem, 2004. 279(45): p. 46573-9. 117. Carvalho, A.F., et al., Ubiquitination of mammalian Pex5p, the peroxisomal import receptor. J Biol Chem, 2007. 282(43): p. 31267-72. 118. Platta, H.W., et al., Pex2 and pex12 function as protein-ubiquitin ligases in peroxisomal protein import. Mol Cell Biol, 2009. 29(20): p. 5505-16. 119. Platta, H.W., W. Girzalsky, and R. Erdmann, Ubiquitination of the peroxisomal import receptor Pex5p. Biochem J, 2004. 384(Pt 1): p. 37-45. 120. Kragt, A., et al., The Saccharomyces cerevisiae peroxisomal import receptor Pex5p is monoubiquitinated in wild type cells. J Biol Chem, 2005. 280(9): p. 7867-74. 121. Walton, P.A., et al., The peroxisomal import receptor PEX5 functions as a stress sensor, retaining catalase in the cytosol in times of oxidative stress. Biochim Biophys Acta, 2017. 1864(10): p. 1833-1843. 122. Platta, H.W., et al., Regulation of peroxisomal matrix protein import by ubiquitination. Biochim Biophys Acta, 2016. 1863(5): p. 838-49. 123. Platta, H.W., et al., The peroxisomal receptor dislocation pathway: to the exportomer and beyond. Biochimie, 2014. 98: p. 16-28. 124. Platta, H.W., et al., Functional role of the AAA peroxins in dislocation of the cycling PTS1 receptor back to the cytosol. Nat Cell Biol, 2005. 7(8): p. 817-22. 125. Tan, D., et al., Structures of the double-ring AAA ATPase Pex1-Pex6 involved in peroxisome biogenesis. FEBS J, 2016. 283(6): p. 986-92.

149

126. Carroni, M., et al., Head-to-tail interactions of the coiled-coil domains regulate ClpB activity and cooperation with Hsp70 in protein disaggregation. Elife, 2014. 3: p. e02481. 127. Wendler, P., et al., Motor mechanism for protein threading through Hsp104. Mol Cell, 2009. 34(1): p. 81-92. 128. Martin, A., T.A. Baker, and R.T. Sauer, Rebuilt AAA + motors reveal operating principles for ATP- fuelled machines. Nature, 2005. 437(7062): p. 1115-20. 129. Birschmann, I., et al., Structural and functional analysis of the interaction of the AAA-peroxins Pex1p and Pex6p. FEBS J, 2005. 272(1): p. 47-58. 130. Weller, S., et al., Alternative splicing suggests extended function of PEX26 in peroxisome biogenesis. Am J Hum Genet, 2005. 76(6): p. 987-1007. 131. Miyata, N. and Y. Fujiki, Shuttling mechanism of peroxisome targeting signal type 1 receptor Pex5: ATP-independent import and ATP-dependent export. Mol Cell Biol, 2005. 25(24): p. 10822- 32. 132. Gardner, B.M., et al., The Pex1/Pex6 complex is a heterohexameric AAA+ motor with alternating and highly coordinated subunits. J Mol Biol, 2015. 427(6 Pt B): p. 1375-88. 133. Blok, N.B., et al., Unique double-ring structure of the peroxisomal Pex1/Pex6 ATPase complex revealed by cryo-electron microscopy. Proc Natl Acad Sci U S A, 2015. 112(30): p. E4017-25. 134. Ciniawsky, S., et al., Molecular snapshots of the Pex1/6 AAA+ complex in action. Nat Commun, 2015. 6: p. 7331. 135. Schliebs, W., W. Girzalsky, and R. Erdmann, Peroxisomal protein import and ERAD: variations on a common theme. Nat Rev Mol Cell Biol, 2010. 11(12): p. 885-90. 136. Baerends, R.J., et al., The Hansenula polymorpha PER9 gene encodes a peroxisomal membrane protein essential for peroxisome assembly and integrity. J Biol Chem, 1996. 271(15): p. 8887-94. 137. Honsho, M., et al., Mutation in PEX16 is causal in the peroxisome-deficient Zellweger syndrome of complementation group D. Am J Hum Genet, 1998. 63(6): p. 1622-30. 138. Götte, K., et al., Pex19p, a farnesylated protein essential for peroxisome biogenesis. Mol Cell Biol, 1998. 18(1): p. 616-28. 139. Matsuzono, Y., et al., Human PEX19: cDNA cloning by functional complementation, mutation analysis in a patient with Zellweger syndrome, and potential role in peroxisomal membrane assembly. Proc Natl Acad Sci U S A, 1999. 96(5): p. 2116-21. 140. Ghaedi, K., et al., PEX3 is the causal gene responsible for peroxisome membrane assembly- defective Zellweger syndrome of complementation group G. Am J Hum Genet, 2000. 67(4): p. 976-81. 141. Sacksteder, K.A., et al., PEX19 binds multiple peroxisomal membrane proteins, is predominantly cytoplasmic, and is required for peroxisome membrane synthesis. J Cell Biol, 2000. 148(5): p. 931-44. 142. Hettema, E.H., et al., Saccharomyces cerevisiae pex3p and pex19p are required for proper localization and stability of peroxisomal membrane proteins. EMBO J, 2000. 19(2): p. 223-33. 143. South, S.T. and S.J. Gould, Peroxisome synthesis in the absence of preexisting peroxisomes. J Cell Biol, 1999. 144(2): p. 255-66. 144. Otzen, M., et al., Hansenula polymorpha Pex19p is essential for the formation of functional peroxisomal membranes. J Biol Chem, 2004. 279(18): p. 19181-90. 145. Kim, P.K., et al., The origin and maintenance of mammalian peroxisomes involves a de novo PEX16-dependent pathway from the ER. J Cell Biol, 2006. 173(4): p. 521-32. 146. Geuze, H.J., et al., Involvement of the endoplasmic reticulum in peroxisome formation. Mol Biol Cell, 2003. 14(7): p. 2900-7. 147. Tabak, H.F., I. Braakman, and A. van der Zand, Peroxisome formation and maintenance are dependent on the endoplasmic reticulum. Annu Rev Biochem, 2013. 82: p. 723-44.

150

148. Jones, J.M., J.C. Morrell, and S.J. Gould, PEX19 is a predominantly cytosolic chaperone and import receptor for class 1 peroxisomal membrane proteins. J Cell Biol, 2004. 164(1): p. 57-67. 149. Fang, Y., et al., PEX3 functions as a PEX19 docking factor in the import of class I peroxisomal membrane proteins. J Cell Biol, 2004. 164(6): p. 863-75. 150. Honsho, M., T. Hiroshige, and Y. Fujiki, The membrane biogenesis peroxin Pex16p. Topogenesis and functional roles in peroxisomal membrane assembly. J Biol Chem, 2002. 277(46): p. 44513- 24. 151. Matsuzaki, T. and Y. Fujiki, The peroxisomal membrane protein import receptor Pex3p is directly transported to peroxisomes by a novel Pex19p- and Pex16p-dependent pathway. J Cell Biol, 2008. 183(7): p. 1275-86. 152. Waterham, H.R., S. Ferdinandusse, and R.J. Wanders, Human disorders of peroxisome metabolism and biogenesis. Biochim Biophys Acta, 2016. 1863(5): p. 922-33. 153. Koch, A., et al., A role for Fis1 in both mitochondrial and peroxisomal fission in mammalian cells. Mol Biol Cell, 2005. 16(11): p. 5077-86. 154. Li, X. and S.J. Gould, The dynamin-like GTPase DLP1 is essential for peroxisome division and is recruited to peroxisomes in part by PEX11. J Biol Chem, 2003. 278(19): p. 17012-20. 155. Koch, A., et al., Dynamin-like protein 1 is involved in peroxisomal fission. J Biol Chem, 2003. 278(10): p. 8597-605. 156. Huber, N., et al., Charcot-Marie-Tooth disease-associated mutants of GDAP1 dissociate its roles in peroxisomal and mitochondrial fission. EMBO Rep, 2013. 14(6): p. 545-52. 157. Schrader, M., et al., Expression of PEX11beta mediates peroxisome proliferation in the absence of extracellular stimuli. J Biol Chem, 1998. 273(45): p. 29607-14. 158. Abe, I. and Y. Fujiki, cDNA cloning and characterization of a constitutively expressed isoform of the human peroxin Pex11p. Biochem Biophys Res Commun, 1998. 252(2): p. 529-33. 159. Tanaka, A., K. Okumoto, and Y. Fujiki, cDNA cloning and characterization of the third isoform of human peroxin Pex11p. Biochem Biophys Res Commun, 2003. 300(4): p. 819-23. 160. Koch, J. and C. Brocard, PEX11 proteins attract Mff and human Fis1 to coordinate peroxisomal fission. J Cell Sci, 2012. 125(Pt 16): p. 3813-26. 161. Ebberink, M.S., et al., A novel defect of peroxisome division due to a homozygous non-sense mutation in the PEX11β gene. J Med Genet, 2012. 49(5): p. 307-13. 162. Mayerhofer, P.U., Targeting and insertion of peroxisomal membrane proteins: ER trafficking versus direct delivery to peroxisomes. Biochim Biophys Acta, 2016. 1863(5): p. 870-80. 163. Agrawal, G. and S. Subramani, Emerging role of the endoplasmic reticulum in peroxisome biogenesis. Front Physiol, 2013. 4: p. 286. 164. Dimitrov, L., S.K. Lam, and R. Schekman, The role of the endoplasmic reticulum in peroxisome biogenesis. Cold Spring Harb Perspect Biol, 2013. 5(5): p. a013243. 165. Sugiura, A., et al., Newly born peroxisomes are a hybrid of mitochondrial and ER-derived pre- peroxisomes. Nature, 2017. 542(7640): p. 251-254. 166. Katarzyna, Z.R. and S. Suresh, Autophagic degradation of peroxisomes in mammals. Biochem Soc Trans, 2016. 44(2): p. 431-40. 167. Honsho, M., S. Yamashita, and Y. Fujiki, Peroxisome homeostasis: Mechanisms of division and selective degradation of peroxisomes in mammals. Biochim Biophys Acta, 2016. 1863(5): p. 984- 91. 168. van Zutphen, T., M. Veenhuis, and I.J. van der Klei, Damaged peroxisomes are subject to rapid autophagic degradation in the yeast Hansenula polymorpha. Autophagy, 2011. 7(8): p. 863-72. 169. Hara-Kuge, S. and Y. Fujiki, The peroxin Pex14p is involved in LC3-dependent degradation of mammalian peroxisomes. Exp Cell Res, 2008. 314(19): p. 3531-41.

151

170. Nuttall, J.M., A.M. Motley, and E.H. Hettema, Deficiency of the exportomer components Pex1, Pex6, and Pex15 causes enhanced pexophagy in Saccharomyces cerevisiae. Autophagy, 2014. 10(5): p. 835-45. 171. Law, K.B., et al., The peroxisomal AAA ATPase complex prevents pexophagy and development of peroxisome biogenesis disorders. Autophagy, 2017. 13(5): p. 868-884. 172. Huybrechts, S.J., et al., Peroxisome dynamics in cultured mammalian cells. Traffic, 2009. 10(11): p. 1722-33. 173. Falkenberg, K.D., et al., Allelic Expression Imbalance Promoting a Mutant PEX6 Allele Causes Zellweger Spectrum Disorder. Am J Hum Genet, 2017. 101(6): p. 965-976. 174. Kelley, R.I., et al., Neonatal adrenoleukodystrophy: new cases, biochemical studies, and differentiation from Zellweger and related peroxisomal polydystrophy syndromes. Am J Med Genet, 1986. 23(4): p. 869-901. 175. Poll-The, B.T., et al., Infantile Refsum's disease: biochemical findings suggesting multiple peroxisomal dysfunction. J Inherit Metab Dis, 1986. 9(2): p. 169-74. 176. Santos, M.J., et al., Peroxisomal integral membrane proteins in control and Zellweger fibroblasts. J Biol Chem, 1988. 263(21): p. 10502-9. 177. Santos, M.J., et al., Peroxisomal membrane ghosts in Zellweger syndrome--aberrant organelle assembly. Science, 1988. 239(4847): p. 1536-8. 178. Braverman, N., et al., Human PEX7 encodes the peroxisomal PTS2 receptor and is responsible for rhizomelic chondrodysplasia punctata. Nat Genet, 1997. 15(4): p. 369-76. 179. Gould, S.J., G.V. Raymond, and D. Valle, The peroxisome biogenesis disorders, in The Metabolic and Molecular Bases of Inherited Disease, C.R. Scriver, Beaudet, A.L., Sly, W.S., Valle, D, Editor. 2001, McGraw Hill International Book Company: New York, NY. p. 3181-3217. 180. Stoll, C., et al., Birth prevalence rates of skeletal dysplasias. Clin Genet, 1989. 35(2): p. 88-92. 181. Levesque, S., et al., A founder mutation in the PEX6 gene is responsible for increased incidence of Zellweger syndrome in a French Canadian population. BMC Med Genet, 2012. 13: p. 72. 182. Shimozawa, N., et al., Genetic heterogeneity in Japanese patients with peroxisome biogenesis disorders and evidence for a founder haplotype for the most common mutation in PEX10 gene. Adv Exp Med Biol, 2003. 544: p. 71. 183. Mathijssen, I.B., et al., Targeted carrier screening for four recessive disorders: high detection rate within a founder population. Eur J Med Genet, 2015. 58(3): p. 123-8. 184. Shaheen, R., et al., Clinical, biochemical and molecular characterization of peroxisomal diseases in Arabs. Clin Genet, 2011. 79(1): p. 60-70. 185. Collins, C.S. and S.J. Gould, Identification of a common PEX1 mutation in Zellweger syndrome. Hum Mutat, 1999. 14(1): p. 45-53. 186. Braverman, N., et al., PEX7 gene structure, alternative transcripts, and evidence for a founder haplotype for the frequent RCDP allele, L292ter. Genomics, 2000. 63(2): p. 181-92. 187. Motley, A.M., et al., Mutational spectrum in the PEX7 gene and functional analysis of mutant alleles in 78 patients with rhizomelic chondrodysplasia punctata type 1. Am J Hum Genet, 2002. 70(3): p. 612-24. 188. Zeharia, A., et al., A novel PEX12 mutation identified as the cause of a peroxisomal biogenesis disorder with mild clinical phenotype, mild biochemical abnormalities in fibroblasts and a mosaic catalase immunofluorescence pattern, even at 40 degrees C. J Hum Genet, 2007. 52(7): p. 599- 606. 189. Braverman, N.E., M.D. D'Agostino, and G.E. Maclean, Peroxisome biogenesis disorders: Biological, clinical and pathophysiological perspectives. Dev Disabil Res Rev, 2013. 17(3): p. 187- 96.

152

190. Govaerts, L., et al., Cerebro-hepato-renal syndrome of Zellweger: clinical symptoms and relevant laboratory findings in 16 patients. Eur J Pediatr, 1982. 139(2): p. 125-8. 191. Berger, J., et al., Peroxisomes in brain development and function. Biochim Biophys Acta, 2016. 1863(5): p. 934-55. 192. Barkovich, A.J. and W.W. Peck, MR of Zellweger syndrome. AJNR Am J Neuroradiol, 1997. 18(6): p. 1163-70. 193. Gilchrist, K.W., et al., Studies of malformation syndromes of man XIB: the cerebro-hepato-renal syndrome of Zellweger: comparative pathology. Eur J Pediatr, 1976. 121(2): p. 99-118. 194. Aubourg, P., et al., Neonatal adrenoleukodystrophy. J Neurol Neurosurg Psychiatry, 1986. 49(1): p. 77-86. 195. Sevin, C., et al., Autosomal recessive cerebellar ataxia caused by mutations in the PEX2 gene. Orphanet J Rare Dis, 2011. 6: p. 8. 196. Rush, E.T., et al., Low bone mineral density is a common feature of Zellweger spectrum disorders. Mol Genet Metab, 2016. 117(1): p. 33-7. 197. Poll-The, B.T., et al., Peroxisome biogenesis disorders with prolonged survival: phenotypic expression in a cohort of 31 patients. Am J Med Genet A, 2004. 126A(4): p. 333-8. 198. Berendse, K., et al., Zellweger spectrum disorders: clinical manifestations in patients surviving into adulthood. J Inherit Metab Dis, 2016. 39(1): p. 93-106. 199. Baes, M. and P. Aubourg, Peroxisomes, myelination, and axonal integrity in the CNS. Neuroscientist, 2009. 15(4): p. 367-79. 200. Kulkarni, K.S., et al., Contrast enhancement of brainstem tracts in Zellweger spectrum disorder: evidence of inflammatory demyelination? Neuropediatrics, 2011. 42(1): p. 32-4. 201. Poll-The, B.T. and M. Engelen, Peroxisomal leukoencephalopathy. Semin Neurol, 2012. 32(1): p. 42-50. 202. Barth, P.G., et al., Neuroimaging of peroxisome biogenesis disorders (Zellweger spectrum) with prolonged survival. Neurology, 2004. 62(3): p. 439-44. 203. Powers, J.M. and H.W. Moser, Peroxisomal disorders: genotype, phenotype, major neuropathologic lesions, and pathogenesis. Brain Pathol, 1998. 8(1): p. 101-20. 204. Torvik, A., et al., Infantile Refsum's disease: a generalized peroxisomal disorder. Case report with postmortem examination. J Neurol Sci, 1988. 85(1): p. 39-53. 205. Verma, N.P., Z.H. Hart, and M. Nigro, Electrophysiologic studies in neonatal adrenoleukodystrophy. Electroencephalogr Clin Neurophysiol, 1985. 60(1): p. 7-15. 206. Ebberink, M.S., et al., Identification of an unusual variant peroxisome biogenesis disorder caused by mutations in the PEX16 gene. J Med Genet, 2010. 47(9): p. 608-15. 207. Lyons, C.J., et al., Leopard spot retinal pigmentation in infancy indicating a peroxisomal disorder. Br J Ophthalmol, 2004. 88(2): p. 191-2. 208. Michelakakis, H.M., et al., PEX1 deficiency presenting as Leber congenital amaurosis. Pediatr Neurol, 2004. 31(2): p. 146-9. 209. Ek, J., et al., Peroxisomal dysfunction in a boy with neurologic symptoms and amaurosis (Leber disease): clinical and biochemical findings similar to those observed in Zellweger syndrome. J Pediatr, 1986. 108(1): p. 19-24. 210. Zaki, M.S., et al., PEX6 is Expressed in Photoreceptor Cilia and Mutated in Deafblindness with Enamel Dysplasia and Microcephaly. Hum Mutat, 2016. 37(2): p. 170-4. 211. Raas-Rothschild, A., et al., A PEX6-defective peroxisomal biogenesis disorder with severe phenotype in an infant, versus mild phenotype resembling Usher syndrome in the affected parents. Am J Hum Genet, 2002. 70(4): p. 1062-8.

153

212. Majewski, J., et al., A new ocular phenotype associated with an unexpected but known systemic disorder and mutation: novel use of genomic diagnostics and exome sequencing. J Med Genet, 2011. 48(9): p. 593-6. 213. Weleber, R.G., et al., Ophthalmic manifestations of infantile phytanic acid storage disease. Arch Ophthalmol, 1984. 102(9): p. 1317-21. 214. Paker, A.M., et al., Docosahexaenoic acid therapy in peroxisomal diseases: results of a double- blind, randomized trial. Neurology, 2010. 75(9): p. 826-30. 215. Cohen, S.M., et al., Ocular histopathologic studies of neonatal and childhood adrenoleukodystrophy. Am J Ophthalmol, 1983. 95(1): p. 82-96. 216. Mignarri, A., et al., Zellweger Spectrum Disorder with Mild Phenotype Caused by PEX2 Gene Mutations. JIMD Rep, 2012. 6: p. 43-6. 217. Régal, L., et al., Mutations in PEX10 are a cause of autosomal recessive ataxia. Ann Neurol, 2010. 68(2): p. 259-63. 218. Steinberg, S.J., et al., A PEX10 defect in a patient with no detectable defect in peroxisome assembly or metabolism in cultured fibroblasts. J Inherit Metab Dis, 2009. 32(1): p. 109-19. 219. Schabhüttl, M., et al., Whole-exome sequencing in patients with inherited neuropathies: outcome and challenges. J Neurol, 2014. 261(5): p. 970-82. 220. Ferdinandusse, S., et al., Mutations in the gene encoding peroxisomal alpha-methylacyl-CoA racemase cause adult-onset sensory motor neuropathy. Nat Genet, 2000. 24(2): p. 188-91. 221. Clarke, C.E., et al., Tremor and deep white matter changes in alpha-methylacyl-CoA racemase deficiency. Neurology, 2004. 63(1): p. 188-9. 222. Thompson, S.A., et al., Relapsing encephalopathy in a patient with α-methylacyl-CoA racemase deficiency. BMJ Case Rep, 2009. 2009. 223. Lieber, D.S., et al., Next generation sequencing with copy number variant detection expands the phenotypic spectrum of HSD17B4-deficiency. BMC Med Genet, 2014. 15: p. 30. 224. Lines, M.A., et al., Peroxisomal D-bifunctional protein deficiency: three adults diagnosed by whole-exome sequencing. Neurology, 2014. 82(11): p. 963-8. 225. Pierce, S.B., et al., Mutations in the DBP-deficiency protein HSD17B4 cause ovarian dysgenesis, hearing loss, and ataxia of Perrault Syndrome. Am J Hum Genet, 2010. 87(2): p. 282-8. 226. McMillan, H.J., et al., Specific combination of compound heterozygous mutations in 17β- hydroxysteroid dehydrogenase type 4 (HSD17B4) defines a new subtype of D-bifunctional protein deficiency. Orphanet J Rare Dis, 2012. 7: p. 90. 227. Ratbi, I., et al., Heimler Syndrome Is Caused by Hypomorphic Mutations in the Peroxisome- Biogenesis Genes PEX1 and PEX6. Am J Hum Genet, 2015. 97(4): p. 535-45. 228. Moser, A.B., et al., Phenotype of patients with peroxisomal disorders subdivided into sixteen complementation groups. J Pediatr, 1995. 127(1): p. 13-22. 229. Waterham, H.R. and M.S. Ebberink, Genetics and molecular basis of human peroxisome biogenesis disorders. Biochim Biophys Acta, 2012. 1822(9): p. 1430-41. 230. Ferdinandusse, S., et al., The important role of biochemical and functional studies in the diagnostics of peroxisomal disorders. J Inherit Metab Dis, 2016. 231. Mandel, H., et al., A new type of peroxisomal disorder with variable expression in liver and fibroblasts. J Pediatr, 1994. 125(4): p. 549-55. 232. Roels, F., et al., Peroxisome mosaics. Adv Exp Med Biol, 2003. 544: p. 97-106. 233. Steinberg, S.J., et al., Peroxisome biogenesis disorders. Biochim Biophys Acta, 2006. 1763(12): p. 1733-48. 234. Imamura, A., et al., Temperature-sensitive mutation in PEX1 moderates the phenotypes of peroxisome deficiency disorders. Hum Mol Genet, 1998. 7(13): p. 2089-94.

154

235. Imamura, A., et al., Temperature-sensitive phenotypes of peroxisome-assembly processes represent the milder forms of human peroxisome-biogenesis disorders. Am J Hum Genet, 1998. 62(6): p. 1539-43. 236. Osumi, T., et al., Temperature sensitivity in peroxisome assembly processes characterizes milder forms of peroxisome biogenesis disorders. Cell Biochem Biophys, 2000. 32 Spring: p. 165-70. 237. Hashimoto, K., et al., Molecular mechanism of a temperature-sensitive phenotype in peroxisomal biogenesis disorder. Pediatr Res, 2005. 58(2): p. 263-9. 238. Waterham, H.R., et al., A lethal defect of mitochondrial and peroxisomal fission. N Engl J Med, 2007. 356(17): p. 1736-41. 239. Shamseldin, H.E., et al., Genomic analysis of mitochondrial diseases in a consanguineous population reveals novel candidate disease genes. J Med Genet, 2012. 49(4): p. 234-41. 240. Niemann, A., P. Berger, and U. Suter, Pathomechanisms of mutant proteins in Charcot-Marie- Tooth disease. Neuromolecular Med, 2006. 8(1-2): p. 217-42. 241. Braverman, N.E., et al., Peroxisome biogenesis disorders in the Zellweger spectrum: An overview of current diagnosis, clinical manifestations, and treatment guidelines. Mol Genet Metab, 2016. 117(3): p. 313-21. 242. Gärtner, J., et al., Mutations in PEX1 in peroxisome biogenesis disorders: G843D and a mild clinical phenotype. J Inherit Metab Dis, 1999. 22(3): p. 311-3. 243. Maxwell, M.A., et al., A common PEX1 frameshift mutation in patients with disorders of peroxisome biogenesis correlates with the severe Zellweger syndrome phenotype. Hum Genet, 1999. 105(1-2): p. 38-44. 244. Klouwer, F.C., et al., Zellweger spectrum disorders: clinical overview and management approach. Orphanet J Rare Dis, 2015. 10: p. 151. 245. Noguer, M.T. and M. Martinez, Visual follow-up in peroxisomal-disorder patients treated with docosahexaenoic Acid ethyl ester. Invest Ophthalmol Vis Sci, 2010. 51(4): p. 2277-85. 246. Baldwin, E.J., et al., The effectiveness of long-term dietary therapy in the treatment of adult Refsum disease. J Neurol Neurosurg Psychiatry, 2010. 81(9): p. 954-7. 247. Das, A.K., et al., Dietary ether lipid incorporation into tissue plasmalogens of humans and rodents. Lipids, 1992. 27(6): p. 401-5. 248. Setchell, K.D., et al., Oral bile acid treatment and the patient with Zellweger syndrome. Hepatology, 1992. 15(2): p. 198-207. 249. Maeda, K., et al., Oral bile Acid treatment in two Japanese patients with Zellweger syndrome. J Pediatr Gastroenterol Nutr, 2002. 35(2): p. 227-30. 250. Klouwer, F.C.C., et al., The cholic acid extension study in Zellweger spectrum disorders: results and implications for therapy. J Inherit Metab Dis, 2018. 251. Sokal, E.M., et al., Hepatocyte transplantation in a 4-year-old girl with peroxisomal biogenesis disease: technique, safety, and metabolic follow-up. Transplantation, 2003. 76(4): p. 735-8. 252. Van Maldergem, L., et al., Orthotopic liver transplantation from a living-related donor in an infant with a peroxisome biogenesis defect of the infantile Refsum disease type. J Inherit Metab Dis, 2005. 28(4): p. 593-600. 253. Demaret, T., et al., Living-donor liver transplantation for mild Zellweger spectrum disorder: Up to 17 years follow-up. Pediatr Transplant, 2018. 22(3): p. e13112. 254. Matsunami, M., et al., Living-Donor Liver Transplantation From a Heterozygous Parent for Infantile Refsum Disease. Pediatrics, 2016. 255. Baes, M. and P.P. Van Veldhoven, Generalised and conditional inactivation of Pex genes in mice. Biochim Biophys Acta, 2006. 1763(12): p. 1785-93. 256. Faust, P.L., Abnormal cerebellar histogenesis in PEX2 Zellweger mice reflects multiple neuronal defects induced by peroxisome deficiency. J Comp Neurol, 2003. 461(3): p. 394-413.

155

257. Li, X., et al., PEX11 beta deficiency is lethal and impairs neuronal migration but does not abrogate peroxisome function. Mol Cell Biol, 2002. 22(12): p. 4358-65. 258. Li, X. and S.J. Gould, PEX11 promotes peroxisome division independently of peroxisome metabolism. J Cell Biol, 2002. 156(4): p. 643-51. 259. Krysko, O., et al., Neocortical and cerebellar developmental abnormalities in conditions of selective elimination of peroxisomes from brain or from liver. J Neurosci Res, 2007. 85(1): p. 58- 72. 260. Hulshagen, L., et al., Absence of functional peroxisomes from mouse CNS causes dysmyelination and axon degeneration. J Neurosci, 2008. 28(15): p. 4015-27. 261. Kassmann, C.M., et al., Axonal loss and neuroinflammation caused by peroxisome-deficient oligodendrocytes. Nat Genet, 2007. 39(8): p. 969-76. 262. Dirkx, R., et al., Absence of peroxisomes in mouse hepatocytes causes mitochondrial and ER abnormalities. Hepatology, 2005. 41(4): p. 868-78. 263. Peeters, A., et al., Carbohydrate metabolism is perturbed in peroxisome-deficient hepatocytes due to mitochondrial dysfunction, AMP-activated protein kinase (AMPK) activation, and peroxisome proliferator-activated receptor γ coactivator 1α (PGC-1α) suppression. J Biol Chem, 2011. 286(49): p. 42162-79. 264. Hiebler, S., et al., The Pex1-G844D mouse: a model for mild human Zellweger spectrum disorder. Mol Genet Metab, 2014. 111(4): p. 522-32. 265. Janssen, A., et al., Docosahexaenoic acid deficit is not a major pathogenic factor in peroxisome- deficient mice. Lab Invest, 2000. 80(1): p. 31-5. 266. Keane, M.H., et al., Bile acid treatment alters hepatic disease and bile acid transport in peroxisome-deficient PEX2 Zellweger mice. Hepatology, 2007. 45(4): p. 982-97. 267. Peirson, S.N., et al., Light and the laboratory mouse. J Neurosci Methods, 2018. 300: p. 26-36. 268. Smith, R.S., et al., Mouse Eye: Anatomy, Pathology, and Biomethods. Research Methods for Mutant Mice, ed. J.P. Sundberg. 2002, Boca Raton, USA: CRC Press. 269. Peyman, G.A., M. Spitznas, and B.R. Straatsma, Peroxidase diffusion in the normal and photocoagulated retina. Invest Ophthalmol, 1971. 10(3): p. 181-9. 270. Thompson, D.A. and A. Gal, Vitamin A metabolism in the retinal pigment epithelium: genes, mutations, and diseases. Prog Retin Eye Res, 2003. 22(5): p. 683-703. 271. Wright, C.B., T.M. Redmond, and J.M. Nickerson, A History of the Classical Visual Cycle. Prog Mol Biol Transl Sci, 2015. 134: p. 433-48. 272. Rister, J. and C. Desplan, The retinal mosaics of opsin expression in invertebrates and vertebrates. Dev Neurobiol, 2011. 71(12): p. 1212-26. 273. Baker, S.A. and V. Kerov, Photoreceptor inner and outer segments. Curr Top Membr, 2013. 72: p. 231-65. 274. T, E., et al., Retinal bipolar cells: Elementary building blocks of vision. 2014. 15(8): p. 507-19. 275. Bringmann, A., et al., Müller cells in the healthy and diseased retina. Prog Retin Eye Res, 2006. 25(4): p. 397-424. 276. Pinto, L.H. and C. Enroth-Cugell, Tests of the mouse visual system. Mamm Genome, 2000. 11(7): p. 531-6. 277. Dawson, W.W., G.L. Trick, and C.A. Litzkow, Improved electrode for electroretinography. Invest Ophthalmol Vis Sci, 1979. 18(9): p. 988-91. 278. Brown, K.T., The eclectroretinogram: its components and their origins. Vision Res, 1968. 8(6): p. 633-77. 279. Penn, R.D. and W.A. Hagins, Signal transmission along retinal rods and the origin of the electroretinographic a-wave. Nature, 1969. 223(5202): p. 201-4.

156

280. Bush, R.A. and P.A. Sieving, A proximal retinal component in the primate photopic ERG a-wave. Invest Ophthalmol Vis Sci, 1994. 35(2): p. 635-45. 281. BROWN, K.T. and T.N. WIESEL, Localization of origins of electroretinogram components by intraretinal recording in the intact cat eye. J Physiol, 1961. 158: p. 257-80. 282. Miller, R.F. and J.E. Dowling, Intracellular responses of the Müller (glial) cells of mudpuppy retina: their relation to b-wave of the electroretinogram. J Neurophysiol, 1970. 33(3): p. 323-41. 283. Dick, E. and R.F. Miller, Light-evoked potassium activity in mudpuppy retina: its relationship to the b-wave of the electroretinogram. Brain Res, 1978. 154(2): p. 388-94. 284. Heynen, H., L. Wachtmeister, and D. van Norren, Origin of the oscillatory potentials in the primate retina. Vision Res, 1985. 25(10): p. 1365-73. 285. Rossmiller, B.P., R.C. Ryals, and A.S. Lewin, Gene therapy to rescue retinal degeneration caused by mutations in rhodopsin. Methods Mol Biol, 2015. 1271: p. 391-410. 286. Leinonen, H. and H. Tanila, Vision in laboratory rodents-Tools to measure it and implications for behavioral research. Behav Brain Res, 2017. 287. Baker, M., Neuroscience: Through the eyes of a mouse. Nature, 2013. 502(7470): p. 156-8. 288. Kretschmer, F., et al., A system to measure the Optokinetic and Optomotor response in mice. J Neurosci Methods, 2015. 256: p. 91-105. 289. Prusky, G.T., et al., Rapid quantification of adult and developing mouse spatial vision using a virtual optomotor system. Invest Ophthalmol Vis Sci, 2004. 45(12): p. 4611-6. 290. Zhang, R., et al., Recovery of PEX1-Gly843Asp peroxisome dysfunction by small-molecule compounds. Proc Natl Acad Sci U S A, 2010. 107(12): p. 5569-74. 291. Berendse, K., et al., Arginine improves peroxisome functioning in cells from patients with a mild peroxisome biogenesis disorder. Orphanet J Rare Dis, 2013. 8: p. 138. 292. Alvarez, A.I., et al., Modulation of the activity of ABC transporters (P-glycoprotein, MRP2, BCRP) by flavonoids and drug response. J Pharm Sci, 2010. 99(2): p. 598-617. 293. Di Pietro, A., et al., Modulation by flavonoids of cell multidrug resistance mediated by P- glycoprotein and related ABC transporters. Cell Mol Life Sci, 2002. 59(2): p. 307-22. 294. Nguyen, T.L., et al., Homology model of RSK2 N-terminal kinase domain, structure-based identification of novel RSK2 inhibitors, and preliminary common pharmacophore. Bioorg Med Chem, 2006. 14(17): p. 6097-105. 295. Nashiro, C., et al., Recruiting mechanism of the AAA peroxins, Pex1p and Pex6p, to Pex26p on the peroxisomal membrane. Traffic, 2011. 12(6): p. 774-88. 296. Schwerter, D.P., et al., ATP-driven processes of peroxisomal matrix protein import. Biol Chem, 2017. 398(5-6): p. 607-624. 297. Maxwell, M.A., et al., Novel PEX1 mutations and genotype-phenotype correlations in Australasian peroxisome biogenesis disorder patients. Hum Mutat, 2002. 20(5): p. 342-51. 298. Geisbrecht, B.V., et al., Disruption of a PEX1-PEX6 interaction is the most common cause of the neurologic disorders Zellweger syndrome, neonatal adrenoleukodystrophy, and infantile Refsum disease. Proc Natl Acad Sci U S A, 1998. 95(15): p. 8630-5. 299. Hamdane, D., et al., A chemical chaperone induces inhomogeneous conformational changes in flexible proteins. Phys Chem Chem Phys, 2016. 18(30): p. 20410-21. 300. Shin, M.H. and H.S. Lim, Screening methods for identifying pharmacological chaperones. Mol Biosyst, 2017. 13(4): p. 638-647. 301. Slawecki, M.L., et al., Identification of three distinct peroxisomal protein import defects in patients with peroxisome biogenesis disorders. J Cell Sci, 1995. 108 ( Pt 5): p. 1817-29. 302. Davies, J.M., A.T. Brunger, and W.I. Weis, Improved structures of full-length p97, an AAA ATPase: implications for mechanisms of nucleotide-dependent conformational change. Structure, 2008. 16(5): p. 715-26.

157

303. Pandey, K.B. and S.I. Rizvi, Plant polyphenols as dietary antioxidants in human health and disease. Oxid Med Cell Longev, 2009. 2(5): p. 270-8. 304. Roth, S.D., et al., Chemical chaperones improve protein secretion and rescue mutant factor VIII in mice with hemophilia A. PLoS One, 2012. 7(9): p. e44505. 305. Mohamed, F.E., et al., Pharmaceutical Chaperones and Proteostasis Regulators in the Therapy of Lysosomal Storage Disorders: Current Perspective and Future Promises. Front Pharmacol, 2017. 8: p. 448. 306. Katayama, K., et al., Flavonoids inhibit breast cancer resistance protein-mediated drug resistance: transporter specificity and structure-activity relationship. Cancer Chemother Pharmacol, 2007. 60(6): p. 789-97. 307. Nicolle, E., et al., QSAR analysis and molecular modeling of ABCG2-specific inhibitors. Adv Drug Deliv Rev, 2009. 61(1): p. 34-46. 308. Morris, M.E. and S. Zhang, Flavonoid-drug interactions: effects of flavonoids on ABC transporters. Life Sci, 2006. 78(18): p. 2116-30. 309. Jung, S.K., et al., The P110 subunit of PI3-K is a therapeutic target of acacetin in skin cancer. Carcinogenesis, 2014. 35(1): p. 123-30. 310. Walker, E.H., et al., Structural determinants of phosphoinositide 3-kinase inhibition by wortmannin, LY294002, quercetin, myricetin, and staurosporine. Mol Cell, 2000. 6(4): p. 909-19. 311. Imai, Y., et al., Phytoestrogens/flavonoids reverse breast cancer resistance protein/ABCG2- mediated multidrug resistance. Cancer Res, 2004. 64(12): p. 4346-52. 312. Jiang, H., et al., Breast Cancer Resistance Protein and Multidrug Resistance Protein 2 Regulate the Disposition of Acacetin Glucuronides. Pharm Res, 2017. 34(7): p. 1402-1415. 313. Palmeira, A., et al., Dual inhibitors of P-glycoprotein and tumor cell growth: (re)discovering thioxanthones. Biochem Pharmacol, 2012. 83(1): p. 57-68. 314. Miron, A., et al., Flavonoids as modulators of metabolic enzymes and drug transporters. Ann N Y Acad Sci, 2017. 1398(1): p. 152-167. 315. Campanero, M.A., et al., Simultaneous determination of diosmin and diosmetin in human plasma by ion trap liquid chromatography-atmospheric pressure chemical ionization tandem mass spectrometry: Application to a clinical pharmacokinetic study. J Pharm Biomed Anal, 2010. 51(4): p. 875-81. 316. Chen, Z., et al., Pharmacokinetic study of luteolin, apigenin, chrysoeriol and diosmetin after oral administration of Flos Chrysanthemi extract in rats. Fitoterapia, 2012. 83(8): p. 1616-22. 317. Rendeiro, C., J.S. Rhodes, and J.P. Spencer, The mechanisms of action of flavonoids in the brain: Direct versus indirect effects. Neurochem Int, 2015. 89: p. 126-39. 318. Han, Y., et al., Apigenin attenuates oxidative stress and neuronal apoptosis in early brain injury following subarachnoid hemorrhage. J Clin Neurosci, 2017. 40: p. 157-162. 319. Min, J.W., et al., Vitexin reduces hypoxia-ischemia neonatal brain injury by the inhibition of HIF- 1alpha in a rat pup model. Neuropharmacology, 2015. 99: p. 38-50. 320. Okuyama, S., et al., Permeation of Polymethoxyflavones into the Mouse Brain and Their Effect on MK-801-Induced Locomotive Hyperactivity. Int J Mol Sci, 2017. 18(3). 321. Zhang, M.W., et al., 7,8-Dihydroxyflavone reverses the depressive symptoms in mouse chronic mild stress. Neurosci Lett, 2016. 635: p. 33-38. 322. Balch, W.E., et al., Adapting proteostasis for disease intervention. Science, 2008. 319(5865): p. 916-9. 323. Mu, T.W., et al., Chemical and biological approaches synergize to ameliorate protein-folding diseases. Cell, 2008. 134(5): p. 769-81.

158

324. Wang, X.M., et al., Induced pluripotent stem cell models of Zellweger spectrum disorder show impaired peroxisome assembly and cell type-specific lipid abnormalities. Stem Cell Res Ther, 2015. 6: p. 158. 325. Reuber, B.E., et al., Mutations in PEX1 are the most common cause of peroxisome biogenesis disorders. Nat Genet, 1997. 17(4): p. 445-8. 326. Folz, S. and J. Trobe, The Peroxisome and the Eye. Survey of Ophthalmology, 1991. 35(5): p. 353- 368. 327. Neuhaus, C., et al., Next-generation sequencing reveals the mutational landscape of clinically diagnosed Usher syndrome: copy number variations, phenocopies, a predominant target for translational read-through, and. Mol Genet Genomic Med, 2017. 5(5): p. 531-552. 328. Ventura, M.J., et al., Diagnosis of a mild peroxisomal phenotype with next-generation sequencing. Mol Genet Metab Rep, 2016. 9: p. 75-78. 329. Ratbi, I., et al., Severe early onset retinitis pigmentosa in a Moroccan patient with Heimler syndrome due to novel homozygous mutation of PEX1 gene. Eur J Med Genet, 2016. 59(10): p. 507-11. 330. Pakzad-Vaezi, K.L. and D.A. Maberley, Infantile Refsum disease in a young adult: case presentation and brief review. Retin Cases Brief Rep, 2014. 8(1): p. 56-9. 331. Polosa, A., H. Bessaklia, and P. Lachapelle, Light-Induced Retinopathy: Young Age Protects more than Ocular Pigmentation. Curr Eye Res, 2017. 42(6): p. 924-935. 332. Molotchnikoff, S., E. Sicard, and P. Lachapelle, Modulations of collicular visual responses by acoustic stimuli in rabbits. Neurosci Res, 1987. 4(5): p. 385-95. 333. Peachey, N.S., et al., Properties of the mouse cone-mediated electroretinogram during light adaptation. Neurosci Lett, 1993. 162(1-2): p. 9-11. 334. Joly, S., et al., Light-induced retinopathy: comparing adult and juvenile rats. Invest Ophthalmol Vis Sci, 2006. 47(7): p. 3202-12. 335. Jung, S., et al., Visual Impairments Following Term Neonatal Encephalopathy: Do Retinal Impairments Also Play a Role? Invest Ophthalmol Vis Sci, 2015. 56(9): p. 5182-93. 336. Swanson, L.W., A Laboratory Guide with Printed and Electronic Templates for Data, Models and Schematic., in Brain Maps: Structure of the Rat Brain. 2004, Elsevier: Amsterdam. 337. Bramall, A.N., et al., The genomic, biochemical, and cellular responses of the retina in inherited photoreceptor degenerations and prospects for the treatment of these disorders. Annu Rev Neurosci, 2010. 33: p. 441-72. 338. Bringmann, A., et al., Cellular signaling and factors involved in Müller cell gliosis: neuroprotective and detrimental effects. Prog Retin Eye Res, 2009. 28(6): p. 423-51. 339. Zemski Berry, K.A., et al., Spatial organization of lipids in the human retina and optic nerve by MALDI imaging mass spectrometry. J Lipid Res, 2014. 55(3): p. 504-15. 340. Anderson, D.M., et al., High spatial resolution imaging mass spectrometry of human optic nerve lipids and proteins. J Am Soc Mass Spectrom, 2015. 26(6): p. 940-7. 341. Mohanty, A. and H.M. McBride, Emerging roles of mitochondria in the evolution, biogenesis, and function of peroxisomes. Front Physiol, 2013. 4: p. 268. 342. Fransen, M., C. Lismont, and P. Walton, The Peroxisome-Mitochondria Connection: How and Why? Int J Mol Sci, 2017. 18(6). 343. Salpietro, V., et al., Zellweger syndrome and secondary mitochondrial myopathy. Eur J Pediatr, 2015. 174(4): p. 557-63. 344. Rahim, R.S., et al., Mitochondrial changes and oxidative stress in a mouse model of Zellweger syndrome neuropathogenesis. Neuroscience, 2016. 334: p. 201-213. 345. Barot, M., M.R. Gokulgandhi, and A.K. Mitra, Mitochondrial dysfunction in retinal diseases. Curr Eye Res, 2011. 36(12): p. 1069-77.

159

346. Smith, C.E., et al., Spectrum of PEX1 and PEX6 variants in Heimler syndrome. Eur J Hum Genet, 2016. 24(11): p. 1565-1571. 347. Organisciak, D.T. and D.K. Vaughan, Retinal light damage: mechanisms and protection. Prog Retin Eye Res, 2010. 29(2): p. 113-34. 348. Nordgren, M. and M. Fransen, Peroxisomal metabolism and oxidative stress. Biochimie, 2014. 98: p. 56-62. 349. Kassmann, C.M., Myelin peroxisomes - essential organelles for the maintenance of white matter in the nervous system. Biochimie, 2014. 98: p. 111-8. 350. Schmidt, M., et al., How does the eye breathe? Evidence for neuroglobin-mediated oxygen supply in the mammalian retina. J Biol Chem, 2003. 278(3): p. 1932-5. 351. Cringle, S.J. and D.Y. Yu, A multi-layer model of retinal oxygen supply and consumption helps explain the muted rise in inner retinal PO(2) during systemic hyperoxia. Comp Biochem Physiol A Mol Integr Physiol, 2002. 132(1): p. 61-6. 352. Cringle, S.J., et al., Intraretinal oxygen consumption in the rat in vivo. Invest Ophthalmol Vis Sci, 2002. 43(6): p. 1922-7. 353. Yu, D.Y. and S.J. Cringle, Oxygen distribution and consumption within the retina in vascularised and avascular retinas and in animal models of retinal disease. Prog Retin Eye Res, 2001. 20(2): p. 175-208. 354. Hayasaka, T., et al., Matrix-assisted laser desorption/ionization quadrupole ion trap time-of- flight (MALDI-QIT-TOF)-based imaging mass spectrometry reveals a layered distribution of phospholipid molecular species in the mouse retina. Rapid Commun Mass Spectrom, 2008. 22(21): p. 3415-26. 355. Anderson, D.M., et al., High resolution MALDI imaging mass spectrometry of retinal tissue lipids. J Am Soc Mass Spectrom, 2014. 25(8): p. 1394-403. 356. Schori, C., et al., Elovl4 5-bp deletion does not accelerate cone photoreceptor degeneration in an all-cone mouse. PLoS One, 2018. 13(1): p. e0190514. 357. Fields, B., D. Knipe, and P. Howley, Fields Virology. Vol. 5. 2007, Philadelphia: Wolters Kluwer Health/Lippincott Williams & Wilkins. 358. Kumaran, N., et al., Retinal gene therapy. Br Med Bull, 2018. 359. Vandenberghe, L.H. and A. Auricchio, Novel adeno-associated viral vectors for retinal gene therapy. Gene Ther, 2012. 19(2): p. 162-8. 360. Administration, U.S.F.a.D. FDA approves novel gene therapy to treat patients with a rare form of inherited vision loss. FDA News Release 2017 [cited 2017 December 19]; Available from: https://www.fda.gov/NewsEvents/Newsroom/PressAnnouncements/ucm589467.htm. 361. Byrne, L.C., et al., Retinoschisin gene therapy in photoreceptors, Müller glia or all retinal cells in the Rs1h-/- mouse. Gene Ther, 2014. 21(6): p. 585-92. 362. Dai, X., et al., Long-term retinal cone rescue using a capsid mutant AAV8 vector in a mouse model of CNGA3-achromatopsia. PLoS One, 2017. 12(11): p. e0188032. 363. Vandenberghe, L.H., et al., Dosage thresholds for AAV2 and AAV8 photoreceptor gene therapy in monkey. Sci Transl Med, 2011. 3(88): p. 88ra54. 364. Vandenberghe, L.H., et al., AAV9 targets cone photoreceptors in the nonhuman primate retina. PLoS One, 2013. 8(1): p. e53463. 365. Bennicelli, J., et al., Reversal of blindness in animal models of leber congenital amaurosis using optimized AAV2-mediated gene transfer. Mol Ther, 2008. 16(3): p. 458-65. 366. Liang, F.Q., et al., Intraocular delivery of recombinant virus. Methods Mol Med, 2001. 47: p. 125- 39.

160

367. Ashtari, M., et al., The Role of the Human Visual Cortex in Assessment of the Long-Term Durability of Retinal Gene Therapy in Follow-on RPE65 Clinical Trial Patients. Ophthalmology, 2017. 124(6): p. 873-883. 368. Bennett, J., et al., Safety and durability of effect of contralateral-eye administration of AAV2 gene therapy in patients with childhood-onset blindness caused by RPE65 mutations: a follow-on phase 1 trial. Lancet, 2016. 388(10045): p. 661-72. 369. Umino, Y. and E. Solessio, Loss of scotopic contrast sensitivity in the optomotor response of diabetic mice. Invest Ophthalmol Vis Sci, 2013. 54(2): p. 1536-43. 370. Xiong, W. and C. Cepko, Distinct Expression Patterns of AAV8 Vectors with Broadly Active Promoters from Subretinal Injections of Neonatal Mouse Eyes at Two Different Ages. Adv Exp Med Biol, 2016. 854: p. 501-7. 371. Pang, J.J., et al., Comparative analysis of in vivo and in vitro AAV vector transduction in the neonatal mouse retina: effects of serotype and site of administration. Vision Res, 2008. 48(3): p. 377-85. 372. Carvalho, L.S., et al., Long-term and age-dependent restoration of visual function in a mouse model of CNGB3-associated achromatopsia following gene therapy. Hum Mol Genet, 2011. 20(16): p. 3161-75. 373. Choi, V.W., et al., AAV-mediated RLBP1 gene therapy improves the rate of dark adaptation in Rlbp1 knockout mice. Mol Ther Methods Clin Dev, 2015. 2: p. 15022. 374. Watkins, P.A., et al., Altered expression of ALDP in X-linked adrenoleukodystrophy. Am J Hum Genet, 1995. 57(2): p. 292-301. 375. Jauregui, M. and P.K. Kim, Probing peroxisome dynamics and biogenesis by fluorescence imaging. Curr Protoc Cell Biol, 2014. 62: p. Unit 21.9.1-20. 376. Lismont, C., et al., Peroxisomes as Modulators of Cellular Protein Thiol Oxidation: A New Model System. Antioxid Redox Signal, 2017. 377. Wei, H., et al., Pharmacological induction of peroxisomes in peroxisome biogenesis disorders. Ann Neurol, 2000. 47(3): p. 286-96. 378. Li, X., et al., PEX11alpha is required for peroxisome proliferation in response to 4-phenylbutyrate but is dispensable for peroxisome proliferator-activated receptor alpha-mediated peroxisome proliferation. Mol Cell Biol, 2002. 22(23): p. 8226-40. 379. Kim, P.K., et al., Ubiquitin signals autophagic degradation of cytosolic proteins and peroxisomes. Proc Natl Acad Sci U S A, 2008. 105(52): p. 20567-74. 380. Yen, N.T., et al., Pro-oxidant properties of indolone-N-oxides in relation to their antimalarial properties. J Inorg Biochem, 2013. 126: p. 7-16. 381. Ganguli, A., et al., Inhibition of autophagy by chloroquine potentiates synergistically anti-cancer property of artemisinin by promoting ROS dependent apoptosis. Biochimie, 2014. 107 Pt B: p. 338-49. 382. Abdulaziz, N., A.R. Shah, and W.J. McCune, Hydroxychloroquine: balancing the need to maintain therapeutic levels with ocular safety: an update. Curr Opin Rheumatol, 2018. 30(3): p. 249-255. 383. Dranchak, P.K., et al., Nonsense suppressor therapies rescue peroxisome lipid metabolism and assembly in cells from patients with specific PEX gene mutations. J Cell Biochem, 2011. 112(5): p. 1250-8. 384. Baruteau, J., et al., Gene therapy for monogenic liver diseases: clinical successes, current challenges and future prospects. J Inherit Metab Dis, 2017. 40(4): p. 497-517. 385. Choudhury, S.R., et al., Viral vectors for therapy of neurologic diseases. Neuropharmacology, 2017. 120: p. 63-80. 386. Gessler, D.J. and G. Gao, Gene Therapy for the Treatment of Neurological Disorders: Metabolic Disorders. Methods Mol Biol, 2016. 1382: p. 429-65.

161

387. Sun, H., A. Huang, and S. Cao, Current status and prospects of gene therapy for the inner ear. Hum Gene Ther, 2011. 22(11): p. 1311-22. 388. Pan, B., et al., Gene therapy restores auditory and vestibular function in a mouse model of Usher syndrome type 1c. Nat Biotechnol, 2017. 35(3): p. 264-272. 389. Corti, M., et al. Immune Blockade in CNS Gene Therapy Improves Safety and Clinical Outcome. in American Society of Gene and Cell Therapy 21st Annual Meeting. 2018. Chicago, IL. 390. DiCarlo, J.E., et al., CRISPR-Cas Genome Surgery in Ophthalmology. Transl Vis Sci Technol, 2017. 6(3): p. 13. 391. Hung, S.S., et al., AAV-Mediated CRISPR/Cas Gene Editing of Retinal Cells In Vivo. Invest Ophthalmol Vis Sci, 2016. 57(7): p. 3470-6. 392. Gaudelli, N.M., et al., Programmable base editing of A•T to G•C in genomic DNA without DNA cleavage. Nature, 2017. 551(7681): p. 464-471. 393. Komor, A.C., et al., Programmable editing of a target base in genomic DNA without double- stranded DNA cleavage. Nature, 2016. 533(7603): p. 420-4. 394. Theda, C., et al., Newborn screening for X-linked adrenoleukodystrophy: further evidence high throughput screening is feasible. Mol Genet Metab, 2014. 111(1): p. 55-7. 395. Hubbard, W.C., et al., Newborn screening for X-linked adrenoleukodystrophy (X-ALD): validation of a combined liquid chromatography-tandem mass spectrometric (LC-MS/MS) method. Mol Genet Metab, 2009. 97(3): p. 212-20. 396. Moser, H.W., A. Mahmood, and G.V. Raymond, X-linked adrenoleukodystrophy. Nat Clin Pract Neurol, 2007. 3(3): p. 140-51. 397. Children, A.C.o.H.D.i.N.a. Recommended Uniform Screening Panel. 2016; Available from: http://www.hrsa.gov/advisorycommittees/mchbadvisory/heritabledisorders/recommend edpanel/. 398. Cipolla, C.M. and I.J. Lodhi, Peroxisomal Dysfunction in Age-Related Diseases. Trends Endocrinol Metab, 2017. 28(4): p. 297-308. 399. Dahabieh, M.S., et al., Peroxisomes protect lymphoma cells from HDAC inhibitor-mediated apoptosis. Cell Death Differ, 2017. 24(11): p. 1912-1924. 400. Kou, J., et al., Peroxisomal alterations in Alzheimer's disease. Acta Neuropathol, 2011. 122(3): p. 271-83. 401. Vasko, R., Peroxisomes and Kidney Injury. Antioxid Redox Signal, 2016. 25(4): p. 217-31.

162