Improving Diagnosis, Understanding, and Treatment of Farber Disease

Home , ACER1, ACER3, Christianson syndrome

Improving diagnosis, understanding, and treatment of Farber disease

Shaalee Dworski

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

Improving diagnosis, understanding, and treatment of Farber disease

Shaalee Dworski

Doctor of Philosophy

Institute of Medical Science University of Toronto

2017 Abstract

Farber disease (FD) is an ultra-rare Lysosomal Storage Disorder. It is caused by mutations in

ASAH1, resulting in reduced activity of acid ceramidase and ceramide accumulation. The disease is poorly understood due to its rarity and the often short lifespan of patients. FD is systemic, with prominent hematological and sometimes neurological components. To better understand the disease, I characterized the hematopoietic and neurological effects of FD in the first mouse model, where the ASAH1 patient mutation P362R was knocked-in. Mice with FD have enlarged organs due to the accumulation of Mac-2+, foamy macrophages. This accumulation disrupts the organ architecture of hematopoietic-associated organs, including the BM and thymus, resulting in an almost complete loss of developing B and T cells in these organs, respectively, and an excess of hematopoietic stem and progenitor cells in the bone marrow. In the brain, mice with

FD also have excess macrophages/microglia, astrocytosis, and hydrocephaly. To improve diagnosis of FD, I identified a plasma cytokine profile that distinguishes patients with FD from those with a disease that it is commonly misdiagnosed as, Juvenile Idiopathic Arthritis. The most elevated of these cytokines was monocyte chemotactic protein 1, and it alone or with the other elevated cytokines was associated with the presence of FD with 80% accuracy. These cytokines were normalized in FD patients who had received hematopoietic stem cell transplantation ii

(HSCT). Finally, to reduce these signs of FD, I tested the efficacy of HSCT. HSCT from WT mice to FD mice more than doubled their lifespan from 7-13 weeks to a median of 27 weeks and a maximum of 40 weeks. Ceramide levels were normalized, and some peripheral signs of the disease were reduced. While beneficial, HSCT did not improve all symptoms. Through better diagnosis and understanding of FD, more effective treatments can be developed.

iii

Acknowledgments

Thank you to my supervisor, Dr. Jeffrey Medin, for the opportunity that he has given me and for his ongoing support. Thank you to my committee members Dr. Armand Keating and Dr. Norman

Iscove for their guidance and intellectual input.

Thank you to the patients and their families who participated in these studies.

I am thankful for the generous funding to study and travel to conferences provided during my

PhD from the Canadian Institutes of Health Research (CIHR) Biological Therapeutics Program, the CIHR Institute Community Support Program, the Queen Elizabeth II/Dr. Dina Gordon

Malkin Graduate Scholarship in Science and Technology, the Institute of Medical Science, the

University of Toronto School of Graduate Studies, the University Health Network Office of

Research Trainee, the WORLD Symposium, the American Society of Hematology, and the

Garrod Society.

For my father.

Contributions

Introduction and discussion

The predicted mouse and human model of acid ceramidase was developed by Zi Jian

Xiong. I annotated it with colour-coded mutations.

Chapter 3

Figures and text adapted from Dworski et al., 2015 with permission from Haematologica.

Jessa Trentadue and I photographed the mouse organs (Figure 10). Alexandra Berger, Joshua M.

Moreau, and I planned and collected data for the flow cytometry (Figure 16-Figure 18; Figure

20). Caren Furlonger performed the IL-7 stimulation assays (Figure 19). The Centre for

Modeling Human Disease (CMHD) performed the immunohistochemistry staining (Figure 11-

Figure 14).

Chapter 4

Figures and text adapted from Sikora & Dworski et al., 2017 with permission from the

American Journal of Pathology. Jakub Sikora, Matthew C. Micsenyi, and Tomo Sawada performed the histological analysis of the brains of the mice (Figure 29A; Figure 30A,C; Figure

31-Figure 38). E. Ellen Jones performed the imaging mass spectrometry analysis of the brains of the mice (Figure 27; Figure 28). Pauline Le Faouder, Justine Bertrand-Michel, Aude Dupuy,

Thierry Levade, Josefina Casas, and Gemma Fabrias identified the ceramide species present and

I analyzed the data (Figure 25; Figure 26). Christopher K. Dunn performed the grip test and activity analysis under my supervision and I analyzed the data (Figure 24G,H). Ingrid Xuan

v performed the open field test, rotarod test, and marble burying assay under my supervision

(Figure 24A-F,I).

Chapter 5

Joshua M. Moreau and Alexandra Berger assisted with planning the staining for flow cytometry and data collection (Figure 66). Jakub Sikora performed the histological analysis of the brains of the mice (Figure 67). Thierry Levade identified the ceramides in the organs and I analyzed the data (Figure 63). Maria (Mafe) Monroy and Sadiya Yousef performed the mouse tail vein injections for the hematopoietic stem cell transplantations. The Centre for Modeling

Human Disease (CMHD) performed the immunohistochemistry staining (Figure 65).

Chapter 6

Figures and text adapted from Dworski et al., 2017 with permission from Biochimica et

Biophysica Acta (BBA) – Molecular Basis of Disease. Ping Lu quantified the ceramide species in the mouse and human plasma samples (Figure 47-Figure 53). Xingxuan He and Edward H.

Schuchman performed the acid ceramidase enzyme activity test on human plasma samples

(Figure 46; Figure 56).

Table of Contents

Acknowledgments ...... iv

Contributions ...... v

Table of Contents ...... vii

List of Abbreviations ...... xvii

List of Tables ...... xxi

List of Figures ...... xxii

Chapter 1 Literature Review ...... 1

1.1 Sphingolipids ...... 1

1.1.1 Structure ...... 2

1.1.2 Source ...... 2

1.1.3 Ceramide ...... 4

1.1.4 Ceramide metabolism ...... 5

1.1.5 Difficulty in studying ceramides and sphingolipids ...... 6

1.2 Acid ceramidase ...... 8

1.2.1 Introduction ...... 8

1.2.2 Family ...... 8

1.2.3 Gene ...... 8

1.2.4 Protein ...... 9

1.2.4.1 Enzyme activity ...... 11

1.2.4.2 Reverse enzyme activity ...... 11

1.2.4.3 Extra-lysosomal function ...... 12

1.2.4.4 Apoptosis ...... 12

1.2.5 Diseases relating to perturbed ACDase activity ...... 14 vii

1.3 Lysosomal Storage Disorders ...... 15

1.3.1 Prevalence ...... 17

1.3.1.1 Difficulty in determining the prevalence of individual LSDs ...... 18

1.3.2 Diagnosis ...... 19

1.3.2.1 Methods ...... 19

1.3.2.2 Different ages of diagnosis ...... 19

1.3.2.3 Carrier screening...... 20

1.3.3 Treatments ...... 21

1.3.3.1 Enzyme replacement therapy...... 21

1.3.3.1.1 Limitations of ERT ...... 22

1.3.3.2 Hematopoietic stem cell transplantation...... 24

1.3.3.3 Gene therapy ...... 26

1.3.3.3.1 Gene therapy and HSCT ...... 27

1.3.3.4 Other therapies ...... 27

1.4 Farber disease ...... 28

1.4.1 Discovery and history ...... 28

1.4.2 Prevalence ...... 29

1.4.3 Signs and symptoms ...... 30

1.4.3.1 Classical triad of symptoms ...... 30

1.4.3.2 Hematopoietic symptoms ...... 31

1.4.3.2.1 Peripheral osteolysis ...... 32

1.4.3.3 Neurological symptoms ...... 33

1.4.3.3.1 SMA-PME ...... 34

1.4.3.4 Ultrastructural changes ...... 35

viii

1.4.4 Disease spectrum ...... 36

1.4.5 Cause ...... 38

1.4.6 Diagnosis ...... 39

1.4.6.1 Problems with FD diagnosis ...... 40

1.4.6.2 Misdiagnosis as Juvenile Idiopathic Arthritis ...... 41

1.4.7 Cause of death ...... 41

1.4.8 Current treatment options ...... 42

1.4.9 Correlation of phenotype to prognosis ...... 43

1.4.10 Source of ceramide accumulation in FD patients ...... 44

1.4.11 Contribution of accumulated ceramide on cell signaling ...... 45

1.5 Mouse models of FD ...... 46

1.5.1 ASAH1 knock-out mouse ...... 46

1.5.2 ASAH1 conditional knock-out mouse ...... 47

1.5.3 ASAH1 knock-in mouse ...... 48

1.5.3.1 ASAH1 Mutation ...... 48

1.5.3.2 Survival ...... 50

1.5.3.3 Ceramide accumulation ...... 51

1.5.3.4 Hematopoietic phenotype ...... 52

1.5.3.5 Neurological phenotype ...... 52

1.5.3.6 Cytokines and ceramides in the mouse model of FD ...... 53

1.5.3.7 Gene therapy of the mouse model ...... 53

1.5.3.7.1 Improvements following gene therapy ...... 54

1.5.3.7.2 Limitations of gene therapy in FD mice ...... 55

Chapter 2 Aims/Hypotheses ...... 56

2.1 Aim 1: To elucidate the effects of FD on different organ systems in a mouse model...... 56

2.2 Aim 2: To improve diagnosis of FD...... 57

2.3 Aim 3: To identify the benefits of HSCT treatment in a mouse model...... 57

Chapter 3 Markedly perturbed hematopoiesis in ACDase deficient mice ...... 59

3.1 Introduction ...... 59

3.2 Methods ...... 61

3.2.1 Mice ...... 61

3.2.2 Organ analyses ...... 62

3.2.3 Histology ...... 62

3.2.4 Flow cytometry ...... 62

3.2.5 Colony-forming cell assays ...... 63

3.2.6 Lymphocyte proliferation assay ...... 64

3.2.7 Bone marrow transplantation ...... 64

3.2.8 Statistical analysis ...... 65

3.3 Results ...... 65

3.3.1 Hematopoietic organs from Hom mice are enlarged and have reduced

cellularity ...... 65

3.3.2 Organ enlargement correlates with macrophage infiltration, which destroys

organ architecture ...... 68

3.3.3 B cell progenitors are dramatically reduced as Hom mice age ...... 78

3.3.4 Lack of B and T cell populations is not due to an inability to respond to

stimulation in Hom mice ...... 80

3.3.5 CD4+ CD8+ T cell progenitors are dramatically reduced as the Hom mice age ... 81

3.3.6 Differentiating myeloid progenitor cells from Hom mice were not biased

towards granulocyte or monocyte lineages ...... 84

3.3.7 HSPCs from Hom mice are able to reconstitute the hematopoietic system of

WT and Het mice and are not sufficient in themselves to induce a FD

phenotype ...... 87

3.4 Discussion/Conclusion ...... 92

3.4.1 Hematopoietic organ architecture ...... 92

3.4.2 Source of monocytes ...... 93

3.4.3 Lymphocyte progenitor loss ...... 94

Chapter 4 Acid Ceramidase Deficiency in Mice Results in a Broad Range of Central Nervous

System Abnormalities ...... 97

4.1 Introduction ...... 97

4.2 Materials and Methods ...... 99

4.2.1 Animals ...... 99

4.2.2 Behavioural testing ...... 99

4.2.3 Collection and processing of the brains ...... 101

4.2.4 Lipid analyses by liquid chromatography-mass spectrometry ...... 101

4.2.5 Mass spectrometry imaging ...... 102

4.2.6 Magnetic resonance imaging ...... 103

4.2.7 Histopathology, antibodies, immunohistochemistry and immunofluorescence . 103

4.2.8 Light microscopy ...... 104

4.2.9 Electron microscopy ...... 105

4.2.10 Statistical analyses and figure preparation ...... 106

4.3 Results ...... 106

4.3.1 Asah1P361R/P361R (Hom Farber) mice express a complex behavioural phenotype 106

4.3.2 The abnormal sphingolipid profiles in the brains of Hom Farber mice are

dominated by the accumulation of ceramide, hydroxyceramide, and

dihydroceramide ...... 109

4.3.3 MALDI-MSI visualization of ceramide accumulation shows regional

accumulation of selected lipid species ...... 114

4.3.4 Hom Farber mice develop hydrocephalus, extensive tissue pathology, and

subcellular storage changes in a broad range of CNS cell types ...... 119

4.3.4.1 Gross brain pathology ...... 119

4.3.4.2 Cerebral and cerebellar (immuno)histopathology ...... 122

4.3.4.3 Ultrastructural pathology ...... 131

4.4 Discussion/conclusions ...... 138

4.4.1 Contribution of brain pathology to behavioural phenotype ...... 139

4.4.2 The effects of storage on various cell types ...... 139

4.4.3 Ceramides linked to disease ...... 143

Chapter 5 Acid ceramidase deficiency is characterized by a unique plasma cytokine and

ceramide profile that is altered by therapy ...... 148

5.1 Introduction ...... 148

5.2 Materials and Methods ...... 151

5.2.1 Sample collection ...... 151

5.2.2 Determination of ACDase activity ...... 152

5.2.3 Determination of chitotriosidase activity ...... 152

5.2.4 Quantitation of lipids by mass spectrometry ...... 152

5.2.5 Cytokine analysis ...... 153

xii

5.2.6 Statistics ...... 153

5.3 Results ...... 154

5.3.1 Circulating cytokine levels ...... 154

5.3.2 Plasma ceramide levels ...... 166

5.3.3 Cytokines and ceramides ...... 179

5.3.4 Chitotriosidase activity ...... 184

5.4 Discussion/conclusions ...... 185

5.4.1 Understanding the biology of Farber disease ...... 186

5.4.1.1 The role of cytokines in the initiation and progression of Farber

disease ...... 186

5.4.1.2 Ceramides in Farber disease ...... 188

5.4.2 Improving diagnosis of Farber disease: a new method of verification that

distinguishes between Farber and JIA patients ...... 189

Chapter 6 Bone Marrow Transplantation Doubles the Lifespan of Acid Ceramidase-Deficient

Mice ...... 191

6.1 Introduction ...... 191

6.2 Materials and methods ...... 192

6.2.1 Animals ...... 192

6.2.2 Bone marrow transplantation (BMT) ...... 193

6.2.3 Skin stretch ...... 193

6.2.4 Peripheral blood counts ...... 194

6.2.5 Quantitation of ceramides ...... 194

6.2.6 Cytokine analysis ...... 194

6.2.7 Immunohistochemistry ...... 194

xiii

6.2.8 Flow cytometry ...... 195

6.2.9 Brain tissue collection and processing ...... 195

6.2.10 Brain immunofluorescence ...... 196

6.2.11 Brain microscopy ...... 196

6.2.12 Statistics ...... 197

6.3 Results ...... 197

6.3.1 BMT with wild-type cells doubles the lifespan of FD mice ...... 197

6.3.2 BMT with WT cells corrects physical signs of FD in homozygous ACDase-

deficient mice ...... 199

6.3.3 BMT reduces ceramide accumulation in peripheral organs and the central

nervous system ...... 201

6.3.4 BMT is able to partially correct the hematopoietic defects in Farber mice ...... 202

6.3.5 BMT reduces brain alterations in Hom Farber mice ...... 207

6.3.6 Circulating cytokines are normalized in BMT-treated Hom mice ...... 210

6.4 Discussion/conclusions ...... 213

6.4.1 Comparison to direct lentiviral injection ...... 213

6.4.2 Treatment limitations ...... 214

6.4.3 Cause of death ...... 215

Chapter 7 General Discussion ...... 217

7.1 Evaluation of hypotheses and success of aims ...... 217

7.1.1 Aim 1 ...... 217

7.1.2 Aim 2 ...... 218

7.1.3 Aim 3 ...... 219

7.2 Insights into the biology of Farber disease ...... 220

xiv

7.2.1 Onset of observable phenotype after weaning ...... 220

7.2.2 Variability in affected organs ...... 221

7.2.3 Inflammation ...... 223

7.2.4 Apoptosis ...... 224

7.2.5 Limitations of the mouse model in recapitulating FD ...... 224

7.3 Sphingolipids ...... 225

7.3.1 Sphingolipid backlog ...... 225

7.3.2 Ceramides in hematopoiesis ...... 226

7.3.2.1 Maturity of excess circulating cells ...... 226

7.3.2.2 Macrophages ...... 226

7.3.2.2.1 Source of macrophages ...... 226

7.3.2.2.2 Macrophages change due to the environment ...... 227

7.3.2.2.3 Macrophages change intrinsically with ceramide

accumulation ...... 228

7.3.3 Ceramides in the nervous system ...... 229

7.3.3.1 Effects on development ...... 229

7.3.3.2 Effects in adulthood ...... 229

7.3.3.3 Comparison of nervous system dysfunction to another rodent model . 230

7.4 Toward improved diagnosis ...... 231

7.4.1 Position of FD among other LSDs ...... 232

7.4.2 Misdiagnosed patients ...... 232

7.4.3 Improved prognostication of disease ...... 233

7.4.3.1 Mutation analysis ...... 234

7.5 Toward improved treatment ...... 238

7.5.1 Importance of early intervention ...... 238

7.5.1.1 Prevention of irreversible damage ...... 238

7.5.1.2 Prevention of neurological phenotype ...... 239

7.5.1.3 Perinatal treatment ...... 240

Chapter 8 Conclusions ...... 241

Chapter 9 Future Directions ...... 243

9.1 The subcellular localization of ceramide accumulation and ACDase ...... 243

9.2 Functionality of the immune system ...... 244

9.3 Investigation of the PNS and muscles ...... 245

9.4 Expanding the cytokine panel to other patients ...... 246

9.5 Developing a gene therapy treatment combined with HSCT ...... 247

References ...... 249

xvi

List of Abbreviations a-OH-Cer, alpha-hydroxylated ceramide

ACDase, acid ceramidase

AF, Alexa Fluor bFGF, basic fibroblast growth factor

BM, bone marrow

BMT, bone marrow transplant

CathD, cathepsin D

Cer, ceramide

Cer-1P, ceramide-1-phosphate

Cer-OH, hydroxy-ceramide

CerS, ceramide synthase

CFC, colony-forming cell

CNS, central nervous system conA, concanavalin A

CSF, cerebrospinal fluid

CTB, curvilinear tubular profiles

DAG, diacylglycerol

DCN, deep cerebellar nuclei

DHC, dihexosylceramide dhCer, dihydroceramide dhSph, dihydrosphingosine dhSph-1P, dihydrosphingosine-1-phosphate

DTIC, dacarbazine xvii

EC, external capsule

EGF, epidermal growth factor enGFP, enhanced green fluorescent protein

EM, electron microscopy

ER, endoplasmic reticulum

ERT, enzyme replacement therapy

FCS, fetal calf serum

FD, Farber disease

G-CSF, granulocyte colony-stimulating factor

GalCer, galactosylceramide

GFAP, glial fibrillary acidic protein

GlcCer, glucosylceramide

GM, grey matter

GM-CSF, granulocyte macrophage colony-stimulating factor

H&E, hematoxylin & eosin

Het, heterozygote

HGF, hepatocyte growth factor

Hom, homozygote

HSC, hematopoietic stem cell

HSCT, hematopoietic stem cell transplantation

HSPC, hematopoietic stem and progenitor cells

IF, immunofluorescence

IFNa, interferon alpha

IFNg, interferon gamma

xviii

IHC, immunohistochemistry

IL, interleukin

IP-10, interferon gamma-induced protein 10

JIA, Juvenile Idiopathic Arthritis

KC, keratinocyte chemoattractant

LC-MS, liquid chromatography-mass spectrometry

LC-MS/MS, liquid chromatography-tandem mass spectrometry

LFB, luxol fast blue

LN, lymph nodes

LPS, lipopolysaccharide

LSD, lysosomal storage disorder

LSK, lineage- Sca1+ cKit+

LV, lentiviral vector

M6P, mannose-6-phosphate

MALDI, matrix-assisted laser desorption ionization

MCP-1, monocyte chemotactic protein-1

MHC, monohexosylceramide

MHC-OH, hydroxy-monohexosylceramide

MIG, monokine induced by gamma interferon

MIP-1a, Macrophage inflammatory protein 1a

MLD, metachromatic leukodystrophy

MPS, mucopolysaccharidosis

MRI, magnetic resonance imaging

MSI, mass spectrometry imaging

xix

ND, not determined

NeuN, neuronal nuclei

NS, nonsignificant

PAS, periodic acid-Schiff

PB, peripheral blood

PBMC, peripheral blood mononuclear cell

PBS, phosphate-buffered saline

PC, Purkinje cell

PFA, paraformaldehyde

PKU, phenylketonuria

PNS, peripheral nervous system

RANTES, regulated on activation, normal T cell expressed and secreted

RBC, red blood cell rm, recombinant mouse rh, recombinant human

Sph-1P, sphingosine-1-phosphate

Sph, sphingosine

SM, sphingomyelin

SMA-PME, spinal muscular atrophy with progressive myoclonic epilepsy

TNFa, tumor necrosis factor alpha

VEGF, vascular endothelial growth factor

WM, white matter

WT, wild-type

List of Tables

Table 1. Specificity of ceramide synthases to different ceramide species and organ distribution. 5

Table 2. Select LSDs classified by stored material...... 17

Table 3. Frequency of ERT-related immune reactions...... 24

Table 4. Characteristics of the 9 types of FD...... 37

Table 5. Weights and total cells isolated from WT, Het, Hom, and BMT mouse organs...... 66

Table 6. The level of each cytokine was compared between male and female patients...... 166

Table 7. Known substitution and deletion mutations in ASAH1...... 235

xxi

List of Figures

Figure 1. De novo sphingolipid synthesis and some ceramide metabolism...... 3

Figure 2. Features that create variations in ceramide species...... 4

Figure 3. Predicted structure of ACDase...... 10

Figure 4. Acid ceramidase converts ceramide into sphingosine and a free fatty acid...... 11

Figure 5. Ceramide and sphingosine are pro-death, while their phosphorylated forms are pro- growth...... 13

Figure 6. Schwann cell with lipid inclusions that compress an axon...... 35

Figure 7. Spectrum of ACDase deficiency...... 38

Figure 8. The P361 R mutation used in the knock-in mouse is not in the active site on the predicted structure of ACDase...... 49

Figure 9. Farber mice weigh less than WT and Het littermates and their weight decreases over time...... 50

Figure 10. Hom mice are smaller but their hematopoietic organs are enlarged and do not have more cells...... 68

Figure 11. The BM of Hom mice fills with Mac-2+ macrophages as the mice ages, and these become foamy...... 70

Figure 12. The thymus of Hom mice fills with Mac-2+ macrophages and the cortex/medulla architecture is lost...... 71

Figure 13. The spleen of Hom mice fills with Mac-2+ macrophages and the germinal center/marginal zone architecture is lost...... 73

Figure 14. The LN of Hom mice are filled with Mac-2+ macrophages by 9 weeks-of-age...... 73

Figure 15. The liver of Hom mice has a modest increase in Mac-2+ macrophages...... 74

xxii

Figure 16. Phenotyping of foamy macrophages...... 76

Figure 17. Granulocytes accumulate in the thymus, LN and spleen, but not in the bone marrow...... 77

Figure 18. B cell progenitors are depleted in the BM of ACDase deficient mice...... 80

Figure 19. Lymphoid progenitors are not intrinsically altered in ACDase deficient mice and mature B and T cells are able to respond to stimulation in vitro...... 81

Figure 20. CD4+ CD8+ T cell progenitors are depleted in the thymus of ACDase deficient mice...... 84

Figure 21. There is an excess of myeloid progenitors in the BM but they are not biased towards the monocyte lineage...... 86

Figure 22. Hom BM rescues lethally-irradiated control mice and is not sufficient to induce Farber disease...... 90

Figure 23. Transplantation of Hom cells into WT and Het mice does not recapitulate the peripheral blood symptoms of Farber disease. Hom mice develop leukocytosis by 9 weeks...... 92

Figure 24. Hom Farber (Hom) mice have decreased locomotor/exploration activity, increased thigmotaxis and perform poorly in grip strength and rotating rod tests...... 108

Figure 25. Absolute levels of sphingolipids species are altered in Asah1P361R/P361R (Hom) mice...... 112

Figure 26. Specific sphingolipid species are altered in Hom Farber (Hom) mice...... 114

Figure 27. MALDI-MSI analysis reveals selective distribution of ceramides by specific brain region in 9-week-old Hom Farber (Hom) mice...... 116

Figure 28. MALDI-MSI analysis reveals selective distribution of sulfatide and gangliosides by specific brain region in Hom Farber (Hom) mice...... 118

Figure 29. Hydrocephalus affects all ventricles in the brains of Asah1P361R/P361R (Hom) mice. . 120

xxiii

Figure 30. The brain weight in Asah1P361R/P361R (Hom) mice is not different from Het and WT mice...... 121

Figure 31. Neurons in Asah1P361R/P361R (Hom) mice contain distended and coarsely granular lysosomes but do not undergo ballooning transformation...... 124

Figure 32. Temporal and spatial progress of the cell type-specific pathologies in the cerebrum of Asah1P361R/P361R (Hom) mice...... 126

Figure 33. Microglia/macrophage pathology in the cerebral white and grey matter of Asah1P361R/P361R (Hom) mice...... 127

Figure 34. Cerebellar pathology in Asah1P361R/P361R (Hom) mice...... 130

Figure 35. White matter and vascular ultrastructural storage pathology in Asah1P361R/P361R (Hom) mice...... 133

Figure 36. CTB-like storage material in Asah1P361R/P361R (Hom) mice...... 134

Figure 37. Neuronal ultrastructural storage pathology in Asah1P361R/P361R (Hom) mice...... 135

Figure 38. Choroid plexus and ependymal storage pathology in Asah1P361R/P361R (Hom) mice. 137

Figure 39. Cytokines elevated in the plasma of Farber mice...... 155

Figure 40. Cytokines from Farber mouse plasma and controls that were not significantly different...... 156

Figure 41. Farber patients have a unique plasma cytokine profile that is normalized following HSCT and different from JIA...... 158

Figure 42. Cytokines from Farber patient plasma and controls that were not significantly different...... 159

Figure 43. The elevation of cytokines in Farber disease is different than in Gaucher disease. .. 161

Figure 44. Cytokine groupings that do not result in productive clustering of Farber patients. .. 162

xxiv

Figure 45. Cytokines with weak or no correlations to each other...... 164

Figure 46. Cytokines with weak or no correlations to the age at sample collection or ACDase activity...... 165

Figure 47. Ceramides with a chain length of 16 carbons are elevated in the plasma of Farber mice...... 167

Figure 48. Ceramide levels vary in plasma from patients with different diseases...... 168

Figure 49. Ceramides that were not significantly different in Farber mouse plasma versus controls...... 170

Figure 50. Sphingosine levels were not significantly different in Farber mouse plasma versus controls...... 171

Figure 51. Ceramides and phosphorylated ceramides that were not significantly different in Farber patient plasma versus controls...... 172

Figure 52. Modified ceramides that were not significantly different in Farber patient plasma versus controls...... 173

Figure 53. Sphingosines were not significantly different in Farber patient plasma versus controls...... 174

Figure 54. Ceramides that do not result in productive clustering of Farber patients...... 175

Figure 55. Ceramides with weak or no correlation to each other...... 177

Figure 56. Ceramides with weak or no correlation to the age at sample collection or ACDase activity...... 178

Figure 57. Cytokine changes correlate with ceramide changes in plasma...... 180

Figure 58. Ceramides with weak or no correlation to MCP-1 or IP-10 levels...... 182

Figure 59. Ceramides with weak or no correlation to IL-6 or IL-12 levels...... 183

xxv

Figure 60. Ceramides with weak or no correlations to VEGF levels...... 184

Figure 61. Chitotriosidase activity in Farber patient plasma...... 185

Figure 62. BMT with WT cells prolongs the lifespan of Hom ACDase-deficient mice and also delays weight loss...... 199

Figure 63. BMT with WT cells prevents ceramide accumulation in Farber mice...... 201

Figure 64. BMT with WT cells prevents leukocytosis in Farber mice...... 203

Figure 65. Fewer macrophages accumulate in organs from transplanted Hom(WT) mice and they do not have a 'foamy' morphology...... 206

Figure 66. B cell progenitors are still depleted in transplanted Hom(WT) bone marrow...... 207

Figure 67. Transplanted Hom(WT) mouse brains retain hydrocephaly, macrophage/microglial infiltration, and astrocytosis...... 209

Figure 68. Plasma cytokines are normalized in transplanted Hom(WT) mice...... 211

Figure 69. Plasma cytokines that are not significantly different between untreated and transplanted mice...... 212

Figure 70. Mutation reported in ASAH1 visualized on the predicted structure of ACDase...... 236

xxvi 1

Chapter 1 Literature Review

1.1 Sphingolipids

Lipids containing a sphingoid base belong to the sphingolipid family. Named for the

Greek myth of the sphinx, whose difficult riddle one had to solve to enter the city of Thebes or else be devoured, they continue to be a riddle to scientists for decades. This name was given by Johann Ludwig Wilhelm Thudichum in 1884, the first to isolate sphingolipids

(Futerman & Hannun, 2004).

Sphingolipids are ubiquitously expressed, playing a wide range of roles depending on the cell type. Their most well-known role is to act as a barrier as components of the plasma membrane of eukaryotic cells: they may act as hydrophobic walls to decrease water loss at the epidermis, insulate axons, or perform other segmenting functions (Gault et al., 2010). In membranes some sphingolipids congregate into lipid rafts, clusters of lipids and proteins that function together, to form cell-type specific groupings with unique roles. The abundance of sphingolipids in the membranes of every cell is so large that they constitute 20% of the lipids in blood plasma (Shui et al., 2011). Sphingolipids may also act as intracellular signaling molecules, a role that is still being explored (Futerman & Hannun, 2004).

Due to their prevalence and importance, severe disruptions in sphingolipid metabolism result in one of many Lysosomal Storage Disorders (LSDs), discussed in detail in Section 1.3. Milder disruptions play a role in many neurological disorders, including dementia, neuroinflammation, and stroke (Lai et al., 2016). Early disruption of sphingolipid

2 metabolism can impact development, such as resulting in neural tube defects (Lai et al.,

2016).

1.1.1 Structure

Sphingolipids have the commonality of an 18-carbon chain called a sphingoid base

(Gault et al., 2010) (Figure 1). This hydrophobic base may be in the form of sphingosine, dihydrosphingosine, or phytosphingosine, and is attached to a variable length fatty acid chain. They also have a hydrophilic component that may be a phosphate, phosphorylcholine, sugar, or simple hydroxyl groups (Futerman & Hannun, 2004).

There are theoretically tens of thousands of sphingolipids that can be created by combining different hydrophobic and hydrophilic parts. These are differentiated by their hydrophilic headgroup, hydroxylation and length of the sphingoid base and fatty acid chain, and degree of saturation of the sphingoid base. To date, 620 unique species have been identified (Merrill et al., 2005; Fahy et al., 2009; Quehenberger et al., 2010).

1.1.2 Source

There are 3 sources of sphingolipids. In their order of prevalence, they can be formed de novo, by conversion from other sphingolipids, or obtained through the diet (Lai et al.,

2016). Conversely, there is only one way to remove sphingolipids: through metabolism of the sphingolipid into ceramide, which is then deacylated into sphingosine (Sph), phosphorylated to sphingosine-1-phosphate (Sph-1P) and irreversibly degraded by Sph-1P lyase into a fatty aldehyde and phosphoethanolamine (Saba et al., 1997; Lai et al., 2016; Figure 1).

Figure 1. De novo sphingolipid synthesis and some ceramide metabolism. Ceramide and sphingosine are formed on the cytosolic side of the ER de novo or formed in the salvage pathway in the lysosome. Ceramide can be transported to the Golgi for further metabolism to be converted into other sphingolipids. The only way to degrade sphingolipids is through conversion to sphingosine-1-phosphate and using its lyase. Adapted from Ogretmen & Hannun, 2004, Wymann & Schneiter, 2006, and Lai et al., 2016.

De novo synthesis occurs on the cytosolic side of the endoplasmic reticulum (ER)

(Figure 1). Palmitoyl-CoA and serine are condensed to form dihydrosphingosine (dhSph, sphinganine) (Gault et al., 2010). dhSph is acetylated by ceramide synthase (CerS) to form dihydroceramide, and then a double bond is introduced to create ceramide. A ceramidase then converts ceramide into sphingosine.

Modifications to the basic sphingosine base may occur. Addition of a phosphate group by sphingosine kinase results in Sph-1P, and acylation of sphingosine by CerS produces ceramide (Gault et al., 2010). Further modifications can be made to the sphingolipid by transferring it to the Golgi apparatus (Figure 1). The salvage pathway in the lysosome can be used to generate more ceramide or sphingosine as needed (Figure 1)

(Ogretmen & Hannun, 2004).

1.1.3 Ceramide

Ceramide is the central lipid in sphingolipid metabolism (Figure 1). It is formed during de novo synthesis of most sphingolipids, and often as an intermediate when one sphingolipid is metabolized into another. It is a fundamental molecule that can be converted to dozens of metabolites by various enzymes (Futerman & Hannun, 2004).

Ceramide is a misleading name; in reality there are dozens of ceramide species. They differ based on the length of their fatty acid chain, degree of saturation, and the presence of a head group, such as a phosphate group (Figure 2).

Figure 2. Features that create variations in ceramide species.

1.1.4 Ceramide metabolism

In de novo sphingolipid synthesis, CerS adds an acyl chain to dhSph. In sphingolipid metabolism it adds an acyl chain to sphingosine. The length and degree of saturation of the acyl chains determines the type of ceramide species formed. C16 and C18 are the most abundant ceramide species in mammals (Sassa & Kihara, 2014). Each organ has different prevalent ceramide species, in part due to a difference in the expression of each CerS (Table

1) (Grosch et al., 2012).

C14 C16 C18 C20 C22 C24 C26 C28 C30 C32 Organ Brain, skeletal muscle CerS1 P Ubiquitous, high in liver, CerS2 P P kidney, lung Skin, testis CerS3 P P P P Lung, heart CerS4 P P Ubiquitous, high in brain, CerS5 P kidney, testis Ubiquitous, high in brain, CerS6 P liver, thymus

Table 1. Specificity of ceramide synthases to different ceramide species and organ distribution. Ceramide synthases preferentially catalyze the addition of fatty acid chains of different lengths to produce different ceramides. Coloured squares indicate the ceramide species each synthase can make, with the preferred species labelled P. Darker purple squares indicate a longer acyl chain length. CerS, ceramide synthase. Adapted from Grosch et al., 2012 and Sassa & Kihara, 2014.

Ceramide species have different functions, many of which are not yet elucidated. To accommodate their different needs, organs have different prevalent CerS. For example, very long chain ceramides are needed to form a water permeability barrier in the skin (Feingold,

2007) and to maintain a hydrophobic environment in the stratum corneum of the eye (Choi &

Maibach, 2005), so CerS3 is prevalent there. However, the levels of CerS mRNA in an organ do not always match the predominant ceramide chain length present, suggesting that post- transcriptional or post-translational control may be important in ceramide production (Grosch et al., 2012).

Organs will also have different ratios of fatty acids available to match with the ceramide needed. Fatty acids with chain-lengths of C22 and C24 are ubiquitous, but the ultra- long-chain fatty acids C26 and longer are only found in areas where protective layers are needed (the skin, retina, meibomian gland of the eye, testes, and brain) (Sassa & Kihara,

2014).

Following its de novo generation, ceramide can be converted into other complex sphingolipids. Ceramide is transferred from the ER to the Golgi, where it can gain a headgroup, carbohydrate group, or sialic acid (Lai et al., 2016). For example, sphingomyelin

(SM) is made by adding a phosphorylcholine or phosphorylethanolamine headgroup to ceramide. The reverse metabolic reactions can also be used to create ceramide by the salvage pathway, or degrade ceramide back to sphingosine. (Lai et al., 2016).

1.1.5 Difficulty in studying ceramides and sphingolipids

Sphingolipids are difficult to study for historical and technical reasons. Initially, sphingolipids were seen mostly as components of membranes, whose function was limited to providing a physical barrier for cells and organelles (Futerman & Hannun, 2004). More recently their biological significance in signaling has been recognized and interest in their study has increased.

From a technical perspective, the methods with which to extract and manipulate lipids are in their infancy compared to those available to study DNA and protein. There is no equivalent to PCR for duplicating lipids, and the antibodies that recognize lipids are rare and unreliable. There are stains for categories of lipids, such as Sudan Black for neutral triglycerides, but no such specific stain for sphingolipids or ceramides. The closest commercially-available alternative are sphingolipid analogues with fluorescently-labelled tags that can be used in live cells to study trafficking (Mason, 1999).

Mass spectrometry is currently the most robust method to identify individual species of sphingolipids. This method is still being developed for the hundreds of unique species; in fact, we used the mouse model described in this thesis to develop a technique similar to immunohistochemistry for lipids using mass spectrometry (Jones et al., 2014; Jones et al.,

2015).

Another obstacle is the transformability of sphingolipids: an excess of one sphingolipid is compensated for by converting it into another sphingolipid. This is especially true for ceramide, a central sphingolipid connected to many metabolic pathways. The exception to this adaptability is seen in LSDs, where there is a continuous excess that eventually overcomes the cell and results in lipid storage.

Due to these difficulties, sphingolipids, including ceramide, have not been as well- studied as certain proteins. In this thesis, I describe the varied effects of a chronic excess of ceramide using a mouse model, which highlights the different roles ceramide plays in many tissues.

1.2 Acid ceramidase

1.2.1 Introduction

Acid ceramidase (ACDase, EC 3.5.1.23, also known as N-acylsphingosine amidohydrolase or N-acylsphingosine deacylase) is one of the enzymes that acts on ceramide to convert it into other sphingolipids. In the lysosome it degrades ceramide into sphingosine and a fatty acid.

ACDase was first identified in rat brain homogenates in 1963 (Gatt, 1963), and more than 30 years later isolated in significant amounts from human urine (Bernardo et al., 1995).

1.2.2 Family

Ceramidases remove the fatty acid from ceramide to convert it into sphingosine.

There are 3 types of ceramidases in humans – acid, neutral, and alkaline – that are named for the optimal pH of their activity. There is one acidic ceramidase (encoded by ASAH1) one neutral ceramidase (encoded by ASAH2), and 3 alkaline ceramidases (encoded by ACER1,

ACER2, ACER3) (Mao & Obeid, 2008).

1.2.3 Gene

ACDase is encoded by the ASAH1 gene on chromosome 8(p21.3-22). The cDNA is

29 kb and produces a 395 amino acid protein (Koch et al., 1996).

The homology of mouse to human ACDase is high: mouse cDNA is ~80% similar to human cDNA, and the protein is 90% similar (Park et al., 2006). Both the mouse and human genes contain 14 exons (Li et al., 1999).

1.2.4 Protein

ACDase functions as a heterodimeric protein consisting of an alpha and a beta chain

(Figure 3). ASAH1 mRNA is transcribed into a single precursor protein that, when transferred to the lysosome, self-cleaves into the two subunits that form the active enzyme (Shtraizent et al., 2008). This autoproteolytic cleavage occurs more quickly at pH 4.5 than at a neutral pH, which prevents premature activation of ACDase prior to reaching the lysosome (Shtraizent et al., 2008). A portion of the newly synthesized ACDase is secreted from cells as a 47 kDa monomeric protein (Bernardo et al., 1995). The alpha chain is 13 kDa and is not glycosylated, while the larger 40 kDa beta chain is glycosylated (Bernardo et al., 1995;

Ferlinz et al., 2001).

The definitive structure of ACDase is unknown as its crystal structure has not been published. However, we created a predicted structure by modeling ACDase after the bacterial acyl-coenzyme A: isopenicillin N-acyltransferase, the protein with the most similar amino acid sequence (Alayoubi et al., 2013). The predicted active site is in the beta chain (Figure

3).

Figure 3. Predicted structure of ACDase. The human structure is on top and the mouse structure on the bottom. The alpha chain is in cyan on the left and the beta chain in green on the right. Active site residues are in yellow with pink dots in the predicted active site.

1.2.4.1 Enzyme activity

The role of ACDase is to hydrolyse ceramide into sphingosine and a free fatty acid when in an acidic environment (Figure 4). ACDase is active in the lysosome, having an optimal pH of 4.5. It acts on ceramides with various chain length fatty acids, but requires a chain length of at least 12 carbons (Linke et al., 2001; Bhabak et al., 2013).

Figure 4. Acid ceramidase converts ceramide into sphingosine and a free fatty acid.

Deacylation by a ceramidase is the only known method to naturally form sphingosine

(Kitatani et al., 2008). In the absence of ACDase, only neutral and alkaline ceramidases can compensate, though these are not active in the lysosome.

1.2.4.2 Reverse enzyme activity

ACDase can also form ceramide from sphingosine and a free fatty acid (Gatt, 1963).

This reverse activity is optimal at a pH of 6.0 and preferentially uses lauric acid as the fatty acid (Okino et al., 2003). Phosphatidic acid and phosphatidylserine promote this reverse activity, while SM inhibits it (Okino et al., 2003). In patients with Farber disease (see section

1.4), whose fibroblasts accumulate ceramide due to reduced ACDase activity in the forward direction, also show a reduction in the reverse activity of ACDase (Okino et al., 2003).

1.2.4.3 Extra-lysosomal function

Localization studies of ACDase in the organelles of cells from normal tissue have not been performed. The localization of ACDase is critical, as its activity may be reversed in a more neutral pH, causing it to create ceramide instead of degrade it (Okino et al., 2003).

There is evidence that ACDase is present in extra-lysosomal compartments.

Immunohistochemical staining of an unfertilized oocyte showed ACDase is present in the nucleus, although this was not discussed by the authors (Eliyahu et al., 2007). Definitive localization to the nucleus was reported in a human adrenocortical cell line following immunocytochemistry and Western blotting of purified nuclei (Lucki et al., 2012). In these cells, ACDase binds to a nuclear receptor and regulates nuclear lipid metabolism (Lucki et al., 2012). It appears that the subcellular localization of ACDase can change: in melanocytes it was found in the cytosol and nucleus, but in mature melanoma cells it was primarily found in the cytosol (Realini et al., 2016).

1.2.4.4 Apoptosis

The most studied function of ACDase is its role in apoptosis. Generally, ceramide is a pro-apoptotic molecule and Sph-1P, the phosphorylated form of sphingosine, exerts a pro- survival function (Matthias et al., 2008) (Figure 5). ACDase activity is reduced, ceramides accumulate and Sph-1P decreases in apoptosing cells. In cancer cells, the opposite occurs to

13 prevent cell death: ACDase activity is increased, ceramide is reduced, and Sph-1P accumulates.

Figure 5. Ceramide and sphingosine are pro-death, while their phosphorylated forms are pro-growth. Adapted from Matthias et al., 2008.

This has been demonstrated using drugs that block ACDase activity. Dacarbazine

(DTIC) is one such drug that causes ACDase to degrade in the lysosomes (Bedia et al.,

2011). In human A375 melanoma cells, the loss of ACDase causes an increase in ceramides

(mainly C16 and C18 ceramides) and a decrease in sphingosine and Sph-1P, leading to cell death through autophagy (Bedia et al., 2011).

DTIC also effectively reduces ACDase in vivo, and is given to patients with metastatic melanoma to stimulate cell death. Similarly, temozolomide, an oral DTIC analog, can penetrate the central nervous system (CNS) to induce cell death (Middleton et al., 2000).

1.2.5 Diseases relating to perturbed ACDase activity

ACDase plays many roles in mammalian development due to its critical position in the balance between cell survival and cell death. It is so essential that when ACDase was knocked out genetically, embryos could not develop past the 2-cell stage (Eliyahu et al.,

2007). However, mice heterozygous for the deletion were viable, indicating that 50% of

ACDase activity is sufficient for normal embryonic development. The earliest ACDase mRNA expression seen by Northern blot is at E7.0 in mice (Li et al., 2002).

While some ACDase activity is needed for survival, varying levels of reduced or elevated activity cause different diseases at different life stages. Numerous diseases have reported changes in ACDase expression in adults, including cancer, obesity, diabetes, inflammation, and neurodegenerative diseases (Morad et al., 2013; Bikman 2012; Gomez-

Muñoz et al., 2013; Mencarelli & Martinez-Martinez, 2013; Mielke et al. 2012). Elevated

ACDase was reported in postmortem brains from patients with Alzheimer’s disease (Huang et al., 2004), though it is unclear when this elevation began. Conversely, a lifelong, systemic reduction in ACDase activity results in Farber disease, an ultra-rare childhood disorder with a spectrum of severity and signs (see 1.4 Farber disease) (Levade et al., 2009).

Due to its role in reducing pro-apoptotic ceramides, ACDase overexpression is a common finding in cancers. In prostate cancer ACDase is upregulated (Seelan et al., 2000;

Saad et al., 2007), which gives cancer cells resistance to radiation and cytotoxic drugs

(Mahdy et al., 2009; Saad et al., 2007). To counteract this, pharmacological inhibition of

ACDase completely prevents tumor growth of prostate and colon cancer (Selzner et al.,

2001; Samsel et al., 2004).

1.3 Lysosomal Storage Disorders

Lysosomal Storage Disorders (LSDs) are a spectrum of more than 50 inherited metabolic disorders (Filocamo & Morrone, 2011). Each disorder is rare but as a group LSDs are more common. The rarity and broad range of diseases leave many individual LSDs understudied and patients with these disorders with few treatment options.

Most LSDs are caused by a deficiency in a single enzyme required for the metabolism of macromolecules in the lysosome, including sphingolipids (Table 2). The substances that should be degraded instead accumulate in the cell, usually in the lysosome, and may cause additional lipids to accumulate as well. Some LSDs result from a deficiency in non- enzymatic lysosomal proteins, such as transmembrane protein transporters or structural support proteins, or lysosomal biogenesis proteins (Filocamo & Morrone, 2011).

The deficiency in the protein is caused by a mutation in the gene encoding it, making

LSDs heritable. Most are inherited in an autosomal recessive mode, and a few are X-linked recessive (such as Fabry disease and mucopolysaccharidosis (MPS) II).

Tay Sachs and Gaucher disease were the first two LSDs to be described, reported in

1881 (with additions in 1887) and 1882, respectively (Tay, 1881; Sachs 1887; Gaucher

1882). Pompe disease was the first disease to be labelled as a LSD in 1963 (Hers, 1963), and it was quickly predicted that other LSDs could exist due to different dysfunctional enzymes

(Hers, 1965).

The phenotype of LSDs varies by disease, but some features are common among the group, such as organomegaly, skeletal defects, and coarse hair and facial features (Ortolano et al., 2014). Accumulation of substrates in the CNS is common, and two-thirds of untreated

LSD patients will develop some sort of neurological symptoms (Wraith, 2004). Most LSDs are multi-systemic, and require a team of healthcare professionals to manage the symptoms, including neurologists, nephrologists, and cardiologists (Ortolano et al., 2014). The negative effects of LSDs worsen and increase in number as more storage material builds up in cells, leading to premature death, ranging in infancy to mid-adulthood depending on the disorder

(Ortolano et al., 2014).

Classification Disease Stored Material Sphingolipidoses Fabry Globotriaosylceramide Farber Ceramide Gangliosidosis GM1 GM1 ganglioside, keratan sulphate, oligosaccharides, glycolipids Gangliosidosis GM2 (Tay-Sachs) GM2 ganglioside, oligosaccharides, glycolipids Gangliosidosis GM2 (Sandhoff) GM2 ganglioside, oligosaccharides Gaucher Glucosylceramide Krabbe Galactosylceramide Metachromatic leukodystrophy Sulphatides Niemann-Pick Sphingomyelin Mucopolysaccha- MPS I (Hurler, Scheie, Dermatan sulphate, heparan sulphate ridoses (MPS) Hurler/Scheie) MPS II (Hunter) Dermatan sulphate, heparan sulphate MPS III A, B, C, D (Sanfilippo A, Heparan sulphate B, C, D) MPS IVA (Morquio A) Keratan sulphate, chondroitin 6-sulphate MPS IV B (Morquio B) Keratan sulphate MPS VI (Maroteaux-Lamy) Dermatan sulphate MPS VII (Sly) Dermatan sulphate, heparan sulphate, chondroitin 6-sulphate MPS IX (Natowicz) Hyaluronan Glycoproteinoses Aspartylglicosaminuria Aspartylglucosamine Fucosidosis Glycoproteins, glycolipids, fucoside-rich oligosaccharides α-Mannosidosis Mannose-rich oligosaccharides β-Mannosidosis Man(β1-4)GlnNAc Schindler Sialylated/asialo-glycopeptides, glycolipids Sialidosis Oligosaccharides, glycopeptides Glycogenoses Glycogenosis II/Pompe Glycogen Lipidoses Wolman/CESD Cholesterol esters

Table 2. Select LSDs classified by stored material. Adapted from Filocamo & Morrone, 2011.

1.3.1 Prevalence

The exact prevalence of each LSD is difficult to determine since most are not regularly screened for, leaving many patients undiagnosed or misdiagnosed. Because they are inherited diseases, the prevalence also depends on the population (Kingma et al., 2015;

Matern et al., 2015). For example, through newborn screening for Pompe disease it was found that Pompe disease had a frequency of 1 in 8 684 in Austria, but was twice as common in neighbouring Hungary at 1 in 4 447 (Mechtler et al., 2012; Wittmann et al., 2012). Even

18 within a country different populations have different frequencies: Pompe disease was found in 1 in 27 886 newborns screened in Washington state, USA, and 1 in 5 463 newborns screened in Missouri, USA (Hopkins et al., 2015; Scott et al., 2013).

One estimate made in 1999 of the frequency of LSDs worldwide was 1 in 7 700 live births, and shortly after in 2006 the estimate was 1 in 5 000 (Meikle et al., 1999; Fuller et al.,

2006). The trend seems to be that the more thoroughly screening is performed the greater the prevalence of the disease (Platt et al., 2012; Kingma et al., 2015). Therefore, these numbers are likely still underestimations, as most countries screen newborns for very few LSDs, if at all.

1.3.1.1 Difficulty in determining the prevalence of individual LSDs

The prevalence of an individual LSD is equally difficult to estimate. Even within a given population where genetic testing is performed, it is difficult to quantify the frequency due to the unknown mutation-phenotype correlations. For example, Fabry disease, one of the more common LSDs, was previously estimated to be around 1 in 40 000 (Desnick et al.,

2001). Newborn screening programs suggest it is higher, at 1 in 3 000 (Spada et al., 2006). It is difficult to gauge the population size through genetic screening because there are hundreds of mutations in the gene defective in Fabry disease, GLA, and it is not clear which are disease-causing and which are benign genetic variants. Therefore, newborn screening may over-estimate the prevalence of individual LSDs (van der Tol et al., 2014).

1.3.2 Diagnosis

1.3.2.1 Methods

Diagnosis of an LSD may occur (from youngest to oldest age) prenatally when parents are identified as carriers, after birth from newborn screening, after symptoms present in the patient, or in familial screening when a family member is diagnosed (Filocamo &

Morrone, 2011). The methods used to diagnose LSDs include gene sequencing to identify mutations, metabolic genetic testing (analysis of unmetabolized substrate in the urine or blood), and biochemical genetic testing (quantification of enzyme activity in blood). One or more of these techniques may be used to confirm a diagnosis (Filocamo & Morrone, 2011).

Accumulation of the problematic substrate in LSDs may begin during embryonic development, resulting in a disease that presents early and severely, or it may only accumulate to a significant level later in life, resulting in a late-onset, milder disease

(Filocamo & Morrone, 2011).

1.3.2.2 Different ages of diagnosis

Most newborns with LSDs appear normal at birth if the stored material does not affect bone and cartilage growth (Platt et al., 2012). Broad metabolic genetic screening of all newborns can identify those with a LSD. This is especially important for diseases that progress quickly and can be relatively easily treated or managed.

One such disease is phenylketonuria (PKU) where patients are unable to digest phenylalanine (Singh et al., 2016). The amount of this amino acid found in regular baby formula is too high for PKU patients, and ingestion of phenylalanine-containing formula will

20 cause irreversible brain damage and developmental delay. These life-altering affects can be avoided by following a low-phenylalanine diet (Singh et al., 2016). The most important aspect is identifying these patients as early as possible so that they may prevent exposure to excess phenylalanine.

LSDs are also diagnosed in adults. A recent survey of the Adult Metabolic Physicians working group of the Society for the Study of Inborn Errors reported that many patients with

LSDs were diagnosed as adults (Sirrs et al., 2015). Almost one quarter of the adult patients reported had a LSD, the most prevalent of which being Fabry, Gaucher, and Pompe disease.

Almost half of these patients were diagnosed as adults, underscoring the importance of LSD awareness for physicians (Sirrs et al., 2015).

1.3.2.3 Carrier screening

In populations where a certain LSD is known to be common, carrier screening can be performed. Parents are genetically screened to see if they are carriers for the disease. If both parents are carriers, they could screen their embryos for the disease and choose not to implant an embryo or to terminate the pregnancy. Alternatively, potential parents could decide not to have biological children, or, in the case of arranged marriages, not to match two carriers together (Burke et al., 2011).

Carrier screening was first used successfully to screen for carriers of Tay Sachs disease in the Ashkenazi Jewish population (Burke et al., 2011). There is an established set of mutations in HEXA that are known to cause Tay Sachs disease, a lethal childhood disorder. The Ashkenazi Jewish population has a relatively high frequency of carriers, 1 in

30, compared to 1 in 167 in the general population (Blitzer & McDowell, 1992).

The first screening program for carriers began in the US in the 1970s (Kaback et al.,

1977). This screening program is so successful that the prevalence of babies with Tay Sachs in the Ashkenazi Jewish community has plummeted (Burke et al., 2011). Anecdotally, most cases of Tay Sachs presented to physicians are of non-Ashkenazi Jewish decent, as parents from any other background are typically not screened (Dr. Joe Clarke, personal communication).

The success of Tay Sachs carrier screening is a model for LSD carrier screening in at- risk carrier populations (Blitzer & McDowell, 1992). This model has been extended to cystic fibrosis and spinal muscular atrophy (Burke et al., 2011).

1.3.3 Treatments

Few individual LSDs have treatments and none have been cured. Many severe LSDs have no treatment options other than palliative care.

1.3.3.1 Enzyme replacement therapy

Enzyme replacement therapy (ERT) was the first treatment theoretically proposed to treat LSDs. In 1964, when Henri Hers and colleagues discovered that Pompe disease is caused by an enzymatic deficiency, his mentor and the discoverer of the lysosome, Christian de Duve, speculated that replacing the deficient enzyme would be the ideal treatment (Hers,

1963; de Duve, 1964; Coutinho et al., 2016).

A mere five years later, Elizabeth Neufeld’s group demonstrated the proof-of-concept of ERT by correcting fibroblasts from patients with Hurler syndrome with cultured media

22 from the fibroblasts of patients with another LSD with a different enzyme deficiency, Hunter syndrome, and vice versa (Fratantoni et al., 1969).

The critical component of this enzyme transfer, mannose-6-phosphate (M6P), was identified in 1977 (Kaplan et al., 1977). M6P on the enzyme docks onto a M6P receptor on the target cell, leading to the endocytosis of the enzyme (Dahms et al., 1989).

Fifty years later, ERT has been developed for a handful of LSDs: Gaucher, Fabry,

Pompe disease, and MPS I, II, IVA and VI (Coutinho et al., 2016). The deficient enzyme is now generated in vitro and administered intravenously weekly or biweekly. In this treatment schema, the substrate accumulates in cells in between infusions and is mostly degraded when the enzyme is delivered as a bolus over a few minutes or hours in a single day.

Despite the theoretical curative potential of ERT, practically, its efficacy is highly variable. ERT is the most successful in Gaucher disease, the first LSD for which ERT was developed (Brady et al., 1974). Regular infusions can stop the progression of the disease and even repair damage (Barton et al., 1990; Barton et al., 1991). On the opposite side of the spectrum, in Pompe disease the enzyme has difficulty penetrating neurons, and antibodies against the infused enzyme quickly accumulate, drastically reducing the efficacy of ERT treatment (Banugaria et al., 2011; Banugaria et al., 2013).

1.3.3.1.1 Limitations of ERT

Even in diseases where ERT is beneficial, it is not a perfect solution. Generally, ERT treatment cannot reverse the effects of the disease, only stabilize or slow its progression. The infused enzyme does not penetrate all tissues equally, leaving some organs untreated. CNS

23 treatment is especially lacking in ERT, as the enzymes cannot cross the blood-brain barrier, yet CNS manifestations are common in LSDs (Brooks et al., 2006; Coutinho et al., 2016).

The infused enzyme is short-lived within the blood, meaning that infusions must be repeated at regular intervals for the duration of the patient’s life. This treatment is also expensive, costing several hundred thousand dollars yearly (Coutinho et al., 2016).

For Fabry disease, a dependence on access to the enzyme has been problematic. In

2009 a manufacturing problem with one brand of ERT, Fabrazyme (agalsidase beta), resulted in a universal shortage of enzyme until 2012. This left patients either without enzyme, on a reduced dose, or waiting to switch to the competitor’s brand, Replagal (agalsidase alfa). The effects of this switch have been studied and used to gain insight on dosing needs

(Weidemann et al., 2014).

Finally, the patient’s body may ultimately reject ERT treatment. This rejection is disease- and patient-specific. Pompe patients consistently develop an immune reaction to the enzyme, and immunomodulation is provided with the enzyme in some patients (Banugaria et al., 2011; Banugaria et al., 2013). Conversely, few Gaucher patients develop antibodies to

ERT, and no immunomodulation is necessary (Kishnani et al., 2016) (Table 3). The features that make one ERT enzyme more immunogenic than another are not understood.

% with IgG LSD Recombinant enzyme Generic name (trade name, company) antibody formation Pompe Acid α-glucosidase Alglucosidase alfa (Myozyme, Sanofi Genzyme) 95% Alglucosidase alfa (Lumizyme, Sanofi Genzyme) 100% MPS I α-L-iduronidase Laronidase (Aldurazyme, BioMarin 97% Pharmaceutical / Sanofi Genzyme) MPS II Iduronate-2-sulfatase Idursulfase (Elaprase, Shire HGT) 47% Fabry Α-galactosidase A Agalsidase beta (Fabrazyme, Sanofi Genzyme) 68% Agalsidase alpha (Replagal, Shire HGT) 64% Gaucher β-glucocerebrosidase Imiglucerase (Cerezyme, Sanofi Genzyme) 15% Velaglucerase alpha (VPRIV, Shire HGT) 1.9% Taliglucerase alfa (Elelyso, Pfizer) 53

Table 3. Frequency of ERT-related immune reactions. MPS, mucopolysaccharidosis. Modified from Kishnani et al., 2016.

An ideal treatment for LSDs would deliver the benefits of ERT while avoiding the risks. This treatment would provide a source of functional enzyme that is consistently available to all tissues in the body, yet does not elicit an immune response.

1.3.3.2 Hematopoietic stem cell transplantation

Hematopoietic stem cell transplantation (HSCT) is, in a sense, endogenous ERT.

Hematopoietic stem cells (HSC) are injected into the patient, usually following myeloablative conditioning (Boelens et al., 2014). After successful engraftment into the BM,

HSCs give rise to all cells of the blood system. The source of the cells may be the bone marrow (bone marrow transplant (BMT)), growth factor-mobilized peripheral blood (PB), or umbilical cord blood (Boelens et al., 2016).

The benefits of HSCT over ERT are many. Practically, not needing to repeatedly infuse an enzyme avoids infections at the injection site or missing doses due to drug

25 availability shortages. Biologically, having constant availability of enzyme in the body simplifies the processing of lipids in cells and prevents storage.

An important advantage of HSCT over ERT is that in some diseases it is able to reach the brain. However, the success is variable between diseases and even between patients with the same disease. BMT can delay brain pathology in MPS I but not MPS II (Hopwood et al.,

1993; Vellodi et al., 1999; Peters et al., 1998) (although earlier treatment in MPS II may render it more effective (Boelens et al., 2014). In Krabbe disease, an umbilical cord blood transplant could only prevent neurological manifestations in asymptomatic neonates, but not resolve them in symptomatic infants (Escolar et al., 2005).

From 1980 to 2013, 2 000 patients with LSDs have been transplanted, and HSCT is the standard practice for MPS IH, adult onset metachromatic leukodystrophy (MLD), and early onset Krabbe disease (Boelens et al., 2014).

As with any LSD treatment, early intervention is critical. A Niemann Pick C type 2 patient transplanted at age 16 months did not develop the respiratory problems by age 5 that untreated patients develop (Breen et al., 2013). This treatment was able to prevent motor difficulties. It has shown promise in treating or stabilizing peripheral symptoms in MPS II but not in neurological symptoms (Guffon et al., 2009).

There are some risks associated with HSCT, including infection and anemia.

Engraftment failure may occur, leaving the patient with an insufficient amount of blood cells following the myeloablative conditioning; however, with progress in the understanding of immunology and a greater donor pool, this has reduced significantly (Boelens et al., 2016).

The survival rate is over 90% for LSD patients receiving HSCT (Boelens et al., 2013;

Aldenhoven et al., 2008, 2009).

The success of HSCT has been associated with the amount of enzyme that the donor cells produce: children with MPS I who received cells with relatively high enzyme expression had a more healthy growth phase and required less orthopedic surgery than patients who received cells with relatively low enzyme expression (Aldenhoven et al., 2008,

2009). This suggests that increasing enzyme expression, perhaps above endogenous levels, could increase the efficacy of HSCT.

1.3.3.3 Gene therapy

In gene therapy, a virus is used to express a large amount functional protein. Using a lentivirus, herpes simplex virus, adenovirus, or adeno-associated virus, a vector is introduced into cells that expresses the functional version of the affected protein (Ortolano et al., 2014).

The vector can be introduced into the patient’s own enzyme-deficient cells to allow the patient’s cells to produce physiological levels of the enzyme, or introduced into a wild-type donor’s cells to allow the donor’s cells to produce supraphysiological levels of enzyme.

This virus can be delivered directly by intravenous or intrathecal injection, where the virus infects any cells that it comec into contact with in the blood or brain, respectively. A limitation of this is that a large amount of virus must be injected to ensure that a sufficient number of cells are infected. However, unless a stem cell is infected, cells that apoptose leave behind few or no infected cells with which to replace them, making the treatment short- lasting in cells with high turnover (such as blood cells) but not in cells with low turnover

(such as neurons) (Ortolano et al., 2014).

1.3.3.3.1 Gene therapy and HSCT

A longer-lasting modality for gene therapy is to infect stem cells, the most useful of which is the HSC. When a HSC is infected and the gene of interest integrates, all daughter cells will also contain and express the gene of interest (Aubourg, 2016). This allows for the long-term presence of enzyme-expressing cells in a combination of gene therapy and HSCT

(Boelens et al., 2014). This treatment would overcome the problem of an inconsistent presence of enzyme seen in ERT, and instead would provide an endogenous, consistent source of enzyme.

It is also an improvement on traditional HSCT. By infecting donor cells with the virus, the cells express enzyme from the vector in addition to their endogenous enzyme expression. These supraphysiological enzyme levels may better treat LSD patients. This is the case for MLD, where functional enzyme was able to reach the CNS following HSCT gene therapy (Biffi et al., 2013).

In the case where a matched donor cannot be found, the virus can be used to infect the patient’s own cells, granting them the ability to express functional enzyme (Ortolano et al.,

2014). An autologous transplant may even be preferential as it eliminates the risk of graft- versus-host disease and reduces other transplant-related complications since the cells engraft faster (Boelens et al., 2014).

1.3.3.4 Other therapies

Depending on the type of mutation in the enzyme causing the LSD, additional therapies may be appropriate. If the disease-causing mutation introduces a premature stop codon into the gene, the mRNA transcript is degraded by the cell. Unique small molecules or

28 aminoglycoside antibiotics (such as gentamycin) can force the ribosome to overlook the stop codon and read through the whole transcript (Brooks et al., 2006). This therapy may be useful for MPS I and ceroid lipofuscinosis neuronal (CLN) 1, where premature stop codons are present in 77% and 53% of patients in some populations, respectively (Brooks et al.,

2006).

Other disease-causing mutations may destabilize protein folding or cause the protein to target the wrong organelle. Chaperone therapy utilizes small molecules to help the peptide overcome these difficulties and form a properly folded protein in the correct location (Brooks et al., 2006).

Finally, an alternate approach for macromolecule accumulation in LSDs is substrate reduction therapy. In this modality, instead of trying to consume the accumulating macromolecule by enhancing the function of the mutated protein, the treatment tries to prevent the accumulation of the macromolecule. In order to slow down the generation of the storage material, a drug is used to inhibit one of the enzymes needed for its formation

(Ortolano et al., 2014). Substrate reduction therapy has been approved for type I Gaucher disease, Niemann Pick type C disease, and infantile nephropathic cystinosis (Platt et al.,

2012; Ortolano et al., 2014).

1.4 Farber disease

1.4.1 Discovery and history

Farber disease (FD, OMIM #228000) is a very rare and severe LSD. The first report of FD (then called disseminated lipogranulomatosis) was in 1947 by Dr. Sidney Farber, the namesake of the Dana-Farber Cancer Institute (Farber 1952). The first patient described had

29 a presentation similar to Hand-Schüller-Christian disease and histology similar to Niemann-

Pick disease at the age of 14 months. He called this disease lipogranulomatosis, as the patient presented with granulomas and lipid storage in various organs.

The identification of ceramide as one of the central molecules in the lipid accumulation of FD was by Dr. Arthur Prensky in 1967 (Prensky et al., 1967). Twenty-five years after Dr. Farber first described the disease its cause was found: lysosomal ACDase deficiency (Sugita et al., 1972).

With advances in molecular biology, the 1990’s introduced the ability to manipulate

ACDase. ACDase was purified and cloned in 1996 (Koch et al., 1996), and in 1999 the same group characterized the gene that encodes ACDase, ASAH1, and demonstrated pathogenic mutations in patients with FD (Li et al., 1999).

To better understand the disease and test novel treatments, the first mouse model of

FD was attempted in 2002, but this model failed to survive (Li et al., 2002). The first viable mouse model of FD was reported by our group in 2013 (Alayoubi et al., 2013) (see section

1.5 Mouse models of FD).

1.4.2 Prevalence

There have been approximately 100 cases of FD reported in the literature over the past 70 years since it was first reported (Levade et al., 2009; Bashyam et al., 2014). With such a small number of patients, the disease is classified as ultra-rare.

There are likely many undiagnosed or misdiagnosed cases (see section 1.4.6.2

Misdiagnosis as Juvenile Idiopathic Arthritis). To support this, a recent study published 11

30 new cases (10 patients from India and 1 from Afghanistan) (Bashyam et al., 2014). This report represents the largest group of patients from one country at one temporal point.

1.4.3 Signs and symptoms

FD is primarily a childhood disorder. Classical FD is severe and signs of the disease are evident in the first few months of life. The signs are so prominent that it has been said it can be diagnosed at a glance (Levade et al., 2009). There are also milder forms of the disease, including newly-described patients diagnosed in adulthood (see Table 4 for summary of types).

1.4.3.1 Classical triad of symptoms

The classical triad of symptoms that characterize FD are subcutaneous nodules; painful, swollen joints; and a hoarse voice (Levade et al., 2009). The subcutaneous nodules are palpable and cause hyperesthesia, an extreme sensitivity to touch or movement. This is often the first sign of the disease, usually evident in the first few weeks of life.

The nodules initially appear on joints and over pressure points with swelling. Over time the nodules thicken and increase in number and size, and the swelling contracts. The joints affected are numerous, and often include the interphalanges, metacarpals, ankles, wrists, knees, elbows, and spine (Burck et al., 1985; Fujiwaki et al., 1992; Mondal et al.,

2009; Ekici et al., 2012). Ultimately, pain in the joints is so severe that young patients choose to minimize their movements and develop joint contractures, where they hold a constant, flexed position (Fujiwaki et al., 1992; Kim et al., 1998).

When nodules appear on the larynx, a hoarse voice occurs, which may progress to aphonia, an inability to speak (Burck et al., 1985). Laryngeal nodules, as well as those on the epiglottis and swelling in the throat area, lead to difficulty in feeding and respiration. These nodules are the most detrimental and lead to the death of some Farber patients. A tracheostomy may be required (Zetterstorm et al., 1958; Samuelsson et al., 1972).

The nodules and swelling in the throat and tongue also contribute to poor weight gain, which is especially detrimental to infants (Cvitanovic-Sojat et al., 1997; Levade et al., 2009).

Repeated vomiting may lead to liquid filling the lungs, resulting in pulmonary disease that is ultimately lethal (Levade et al., 2009).

1.4.3.2 Hematopoietic symptoms

Two of the three classical FD symptoms are directly due to nodules (subcutaneous nodules and a hoarse voice). This feature highlights the importance of the hematopoietic system in the manifestations of FD: the characteristic nodules found in Farber disease are composed of macrophages called foamy histiocytes. The macrophages have acquired a foamy appearance, due to stored material, and may also be multinucleated (Burck et al.,

1985; Ekici et al., 2012; Devi et al., 2006; Fujiwaki et al., 1992; Kattner et al., 1997; Mondal et al., 2009).

In addition to their presence in nodules, histiocytic infiltrations have been reported in the liver, spleen, bone marrow (BM), lungs, thymus, LN, heart, lumbar spine, and peritoneal fluid of FD patients (Antonarakis et al., 1984; Kattner et al., 1997; Van Lijnschoten et al.,

2000). The histiocytes in liver were further studied and found to have a cytoplasmic

32 accumulation of soluble glycoprotein or neutral mucopolysaccharide in a single membrane- bound vesicle (Antonarakis et al., 1984).

Other hematopoietic symptoms of FD vary. Case reports have reported a variety of hematological characteristics, all of which have been present in some reports and absent in others. These include leukocytosis, anemia, thrombocytopenia, occasional nucleated red blood cells (RBCs), hepatosplenomegaly, lymphadenopathy, and calcification of axillary LN

(Antonarakis et al., 1984; Fujiwaki et al., 1992; Mondal et al., 2009). Circulating leukocytes are normal, not foamy, and do not appear to contain inclusions (Burck et al., 1985).

1.4.3.2.1 Peripheral osteolysis

Very recently, a new type of FD was discovered with hematopoietic-associated symptoms in the oldest set of patients. Three siblings, aged 40-60, with peripheral osteolysis did not have mutations in MMP2 or MMP14, as is typical for the recessive form of this disease, and instead had ACDase mutations (Bonafe et al., 2016). The ACDase activity quantified in 2 of the siblings’ fibroblasts was 7-8% of controls.

The clinical picture of these siblings is reminiscent of FD, at least in its initial years.

In childhood, they presented with subcutaneous nodules over the joints and ears, a hoarse voice, and episodic fever and pain. Over time their hands and feet have been affected by osteolysis and arthritis, and the fingers and toes had shortened, leaving redundant skin. As a

40-year-old at the time of the report, one sibling has limited movement in his knees and toes, but his fingers can be hyperextended. No neurological deficits are apparent, nor spine abnormalities (Bonafe et al., 2016).

There are some similarities between this new case and a report of one of the oldest patients with FD. At age 29.5, the Farber patient had deformed hands with thick, short fingers and redundant skin (Fiumara et al., 1993).

1.4.3.3 Neurological symptoms

For some FD patients, the nervous system is affected. This includes both the central and peripheral nervous systems (CNS and PNS, respectively).

Regarding the brain, patients present with hydrocephaly, cortical brain atrophy, and swollen neurons filled with stored lipid. The neuronal storage is broad, and has been seen in the cerebral cortex, cerebellum, brain stem, and anterior horns of the spinal cord (Bierman et al., 1966; Molz et al., 1968; Zappatini-Tommasi et al., 1992; Chedwari et al., 2012; Levade et al., 2009).

Along with lipid storage in CNS neurons, storage is also seen in the Schwann cells of the PNS. Both myelinating and non-myelinating Schwann cells have large membrane-bound inclusions that visibly compress the axonal body and reduce nerve conduction velocity

(Burck et al., 1985; Pellissier et al., 1986; Zappatini-Tommasi et al., 1992).

These changes to the nervous system can lead to seizures, developmental delays, and intellectual disability. Because of the storage in anterior horn cells and peripheral neuropathy, patients have poor deep tendon reflexes, hypotonia, and muscular atrophy. The poor muscle control may lead to older patients requiring a wheelchair (Levade et al., 2009; Bierman et al.,

1966; Molz et al., 1968; Eviatar et al., 1986; Zappatini-Tommasi et al., 1992; Chedwari et al., 2012).

As seen in other LSDs, a cherry-red spot may be seen in the macula of the eye (Chen et al., 2014). This spot is surrounded by a diffuse, gray retinal opacity, but no disturbance in visual function has been reported (Cogan et al., 1966; Pellissier et al., 1986; Levade et al.,

2009).

1.4.3.3.1 SMA-PME

Most Farber patients with neurological signs also have other classic FD elements, such as nodules. However, recently a new type of FD patient has been described that has no classical FD signs. These patients have spinal muscular atrophy with progressive myoclonic epilepsy (SMA-PME; MIM ID #159950).

Most patients with SMA have mutations in the SMN1 or SMN2 genes (Zhou et al.,

2012). However, there are some SMA patients who do not have mutations in these genes.

When sequencing the exomes of these rare patients, it was discovered that a subset of them instead have mutations in ASAH1, the same gene that causes FD (Zhou et al., 2012; Dyment et al., 2014; Gan et al., 2015; Giraldez et al., 2015; Rubboli et al., 2015).

Surprisingly, despite being caused by the same enzyme deficiency, SMA-PME has little similarities in its presentation with FD. Presentation of the disease is around age 5, an age which many Farber patients never reach, with difficulty walking and sporadic falls, tremors, and tongue fasciculations. As the disease progresses, muscle atrophy and nerve loss occur, and frequent seizures begin. Patients eventually lose the ability to walk and swallow.

Death occurs in the teenage years, around age 15 (Zhou et al., 2012; Dyment et al., 2014;

Gan et al., 2015; Giraldez et al., 2015; Rubboli et al., 2015).

1.4.3.4 Ultrastructural changes

FD causes changes inside individual cells. Electron microscopy (EM) images of skin biopsies reveal that fibroblasts, epidermal cells, phagocytes, and Schwann cells have numerous, large pale inclusions. In the sciatic nerve of a mouse with FD (see section 1.5.3

ASAH1 knock-in mouse), I identified inclusions in Schwann cells as well (Figure 6). There may also be curvilinear black lines visible inside these inclusions, called “Farber bodies”

(Burck et al., 1985; Eviatar et al., 1986; Fusch et al., 1989; Zappatini-Tommasi et al., 1992).

The inclusions are likely stored lipid, and the Farber bodies may be accumulated ceramide, as they can be induced to form in vitro by incubating fibroblasts from patients with FD in ceramide-rich media (Rutseart et al., 1977).

Figure 6. Schwann cell with lipid inclusions that compress an axon. The sciatic nerve of a mouse with FD was sectioned and visualized with an electron microscope. The large white areas indicated by * are seen changing the shape of the Schwann cell, compressing the ensheathed nerve. A smaller, non-compressed axon is seen above.

1.4.4 Disease spectrum

Dr. Farber first reported the disease in 1947 at a Mayo Foundation lecture (Farber

1952). When discussing this same case at the American Pediatric Society’s 62nd Annual

Meeting in 1952, Dr. Farber remarked, “The clinical picture I described may be found to be typical for these 3 cases and may not be encountered in the next 20 or 30. We should, with a disease of this kind, expect to see a number of unrelated clinical pictures in the future”

(Farber 1952). His foresight for the future of the disease could not have been more apt. Sixty years later researchers encountered three dissimilar FD types: an only neuronal form (SMA-

PME) (Zhou et al., 2012), a mild, peripheral form with no neuronal contribution (Dr. Edward

Schuchman, personal communication), and, most recently, a peripheral form affecting the bones and cartilage (Bonafe et al., 2016).

Until recently when these diverse patients were identified, FD was classified into 7 categories using 78 patients (Levade et al., 2009). The characteristics of each type, and the 2 new types, are in Table 4.

Mean age of Peripheral signs Neurological signs … (years)

red spot

Type

Onset

Death

Nodules

Epilepsy

Hoarseness

involvement

Splenomegaly

CNS impairedCNS

Hepatomegaly

Lung infiltrates

Joint involvement

Lower motor neuron

Macular cherry 1: Classic 0.3 1.45 x x x x x x x x x 2: Intermediate 0.66 6.25 x x x x x x - x x 3:Mild 1.6 15.8 x x x x x x - x x 4: Neonatal 0 0.16 x x x x x 5: Neurologic with Progressive 1.66 3.6 x x x x x - x x x 6: Combined with Sandhoff 0.13 1.1 x x x x x - - x 7: Prosaposin Deficiency 0 0.3 - - - x x - x SMA-PME 5 >15 ------x x x Peripheral Osteolysis 3 >60 x x x - - -

Table 4. Characteristics of the 9 types of FD. An ‘x’ indicates the presence of the phenotype; ‘-’ indicates absence; and blank indicates not assessed. Adapted from Levade et al., 2009, with additions from SMA-PME case reports from Zhou et al., 2012 and peripheral osteolysis case reports from Bonafe et al., 2016.

As additional FD patients are described, it has become clear that the old classification system for FD by type can be replaced by a spectrum of disease (Beck et al., 2015). Five of the seven classical types, plus SMA-PME and peripheral osteolysis, can be placed on a gradient based on the presence of peripheral and neurological signs (Figure 7). Type 6 and 7 have other genetic mutations causing the disease (in addition to ACDase) and are excluded from the spectrum.

Figure 7. Spectrum of ACDase deficiency. The types of FD disease due to ACDase deficiency can be placed on a spectrum based on the presence of peripheral and neurological signs.

1.4.5 Cause

FD is caused by a mutation in the ASAH1 gene that encodes ACDase (Levade et al.,

2009). One functional copy of the gene is sufficient to prevent FD: there are no recorded cases of a disease related to heterozygosity for ACDase deficiency.

FD patients have mutations on both ASAH1 alleles. There is usually a different mutation on each allele, however in cases of consanguinity the same mutation may be present on each allele (Levade et al., 2009). There has not been a study that correlates specific mutations with disease severity, potentially due to the small number of case reports available.

For FD patients, the mutation in ASAH1 severely reduced the functionality of the enzyme, yet some activity remains. The level of ACDase activity is much lower in patients with FD than in unaffected individuals, and may be at 6% of control values or lower (Ehlert

39 et al., 2007; Levade et al., 2009). In the few SMA-PME cases reports, activity varied from

3% to 32% (Zhou et al., 2012; Dyment et al., 2013), and in the only peripheral osteolysis case report the enzyme activity was 7-8% of control values (Bonafe et al., 2016).

No human cases have been reported with homozygous null deletions. This is because some ACDase activity is needed for embryo formation, as seen in the failure of generating an

ACDase knock-out mouse (Li et al., 2002; Eliyahu et al., 2007).

1.4.6 Diagnosis

FD is inherited in an autosomal recessive manner. Because it is so rare it is not included in prenatal screening or neonatal testing, unless an older sibling had been diagnosed with FD. Suspicion of FD begins with characteristic signs and symptoms, often in infancy, and may follow misdiagnosis as other diseases, such as Juvenile Idiopathic Arthritis (JIA).

Definite diagnosis of FD can be established by measuring ACDase activity and/or sequencing the ASAH1 gene. ACDase activity is determined in leukocytes collected from PB, from skin fibroblasts obtained by a biopsy, or from amniocytes obtained in utero. ACDase activity is usually below 6% of control values in affected patients, while other lysosomal enzymes have normal activity (Ehlert et al., 2007).

FD can also be diagnosed by gene sequencing. This is particularly informative in patients with non-classical FD, as seen in the newly described SMA-PME and peripheral osteolysis patients (Zhou et al., 2012; Bonafe et al., 2016).

Other common signs of FD that may help with diagnosis include observation of

“Farber bodies” by EM analysis of a biopsy, and quantification of high ceramide levels in a biopsy sample, urine, or cultured cells (Chatelut et al., 1996; Levade et al., 2009).

1.4.6.1 Problems with FD diagnosis

Analysis of ACDase enzyme activity is a specialized skill. Currently, many patients in Europe and the Middle East are diagnosed by a single lab in Toulouse, France whose expertise is rare disease diagnosis (Dr. Thierry Levade, personal communication).

Sequencing is a more widely available technique but can be expensive and take weeks for results (Sawyer et al., 2015). Whole-exome sequencing is becoming an increasingly attractive method of rare disease diagnosis (Sawyer et al., 2015). However, any type of sequencing requires the mutation results to be interpreted against previously reported mutations; the clinician must determine if the ASAH1 polymorphism is disease-causing or benign. This can be achieved by comparison with mutations available in the literature, yet there are a relatively small number of case reports with patient mutations. Should a patient have a unique mutation, software can be used to predict the detrimental impact of mutations, but this is unreliable.

These diagnostic methods are also unable to measure treatment success. Following an intervention, measuring ACDase activity is informative as to whether there is functional enzyme following the treatment, but is not informative as to whether the effects of this enzyme increase are positively influencing the patient.

An ideal FD diagnostic measure would be one that is as specific for FD as ACDase enzyme analysis and ASAH1 sequencing, but that is inexpensive and accessible, and that can also be re-assessed at intervals after a treatment to measure success.

1.4.6.2 Misdiagnosis as Juvenile Idiopathic Arthritis

In addition to being poorly understood, FD is likely underdiagnosed or misdiagnosed.

The painful and contracted joints seen in patients with FD are similar to those that occur in patients with Juvenile Idiopathic Arthritis (JIA). This can result in the misdiagnosis of patients with moderate FD as having JIA, a far more common disease (Kostik et al., 2013;

Torcoletti et al., 2014; Erfan et al., 2015). A literature review of case reports revealed that

36% of case reports of patients with moderate FD were initially misdiagnosed as JIA

(Schuchman, 2014).

Interestingly, Dr. Farber warned of the misdiagnosis of Farber patients for JIA. He described patients as having, “such extensive involvement of the joints and periarticular tissues that a diagnosis of rheumatoid arthritis has been made by workers in that field on clinical examination alone” (Farber, 1952).

Contributing to this confusion, the TNF-α inhibitors and Interleukin (IL)-6 receptor blockers that JIA patients receive to reduce joint inflammation are also somewhat effective in

Farber patients (Dr. Bruno Maranda, Dr. John J. Mitchell, Dr. Boris Hugle, Dr. Bo

Magnusson, personal communication).

A test that can distinguish between FD and JIA is needed to allow physicians to more quickly diagnose and treat patients. Additionally, the JIA patient populations could be screened to identify misdiagnosed Farber patients.

1.4.7 Cause of death

The cause of death depends on the symptoms and severity of the disease. The presence of neurological symptoms significantly affects survival – those with severe

42 neurological symptoms die in infancy, while those without them can survive into their 30’s

(Ehlert et al., 2007).

Hepatosplenomegaly is common, and can lead to death as early as 1 year of age

(Antonarakis et al., 1984). Most patients with severe hepatosplenomegaly succumb to the disease in the first few years, but there have been reports of FD where death was in utero

(hydrops fetalis) due to disseminated intravascular coagulation (Kattner et al., 1997, van

Lijnschoten et al., 2000).

Patients that survive into childhood may suffer from disturbances in swallowing, due to nodules and swelling in the throat. This may lead to aspiration pneumonia that can be fatal

(Levade et al. 2009; Ehlert et al. 2007). Similarly, SMA-PME patients also die from lung infections and pneumonia (Zhou et al., 2012).

There have been very few reports of older-surviving patients. Two sisters died at the ages 11 and 18 (Pavone et al., 1980), and their cousin, who was diagnosed at age 17 died at age 30 (Fiumara et al., 1993). One of these older patients died from respiratory infection

(Fiumara et al., 1993).

1.4.8 Current treatment options

There are currently 2 treatments available for FD: palliative care and HSCT. ERT is also being explored as a future treatment for FD (Schuchman, 2016).

All Farber patients can receive palliative care, which includes surgical and medicinal management of the disease (Levade et al., 2009). This includes surgical correction of contractures, steroid administration to manage the pain of the nodules, and anti-inflammatory

43 medications to reduce hematopoietic symptoms. Physical therapy can also help with mobility. (Ehlert et al., 2007)

HSCT reduces the size and number of subcutaneous nodules and helps increase joint mobility. However, it is not consistently successful and is associated with severe, even life- threatening, side-effects. HSCT is only effective for a subset of patients: those without neurological symptoms (Yeager et al., 2000; Vormoor et al., 2004; Ehlert et al., 2006).

For patients who have neurological symptoms before the transplant, HSCT is not effective. Even patients who did not have neurological symptoms before HSCT may still develop them afterwards (Yeager et al., 2000; Cappellari et al., 2016). Most recently, a child with FD displayed difficulty standing pre-transplantation, which was attributed to joint pain and suggested peripheral-only disease. Yet twenty-two months following HSCT, he displayed nystagmus, titubation, and gait ataxia, followed by a progression to dysphagia, dyspnea, and arm weakness two months later (Cappellari et al., 2016), demonstrating how new neurological symptoms can appear post-HSCT.

HSCT does not specifically target the brain, and the specific changes in the brain following HSCT in FD have not been assessed. An understanding of this would be critical to understanding how neurological symptoms develop post-HSCT in FD, and can be used to develop future therapies that target these neurological manifestations.

1.4.9 Correlation of phenotype to prognosis

The ability to predict the features of a patient’s disease following a diagnosis of FD would be beneficial. Correlations have not yet been reported between ASAH1 mutations and disease severity or particular phenotypes. This may be because there are so few Farber

44 patients that the database of mutations and case reports is relatively small (Levade et al.,

2009).

Attempts have also been made to correlate biochemical measures with a phenotype.

One study using cultured patient fibroblasts found no correlation between ceramide accumulation and residual AC activity in vitro (van Echten-Deckert et al., 1997). Another found that the amount of sphingomyelin-derived ceramide accumulated in situ correlated negatively in a logarithmic fashion with the age of patient death; that is the more ceramide that accumulated the younger the patient died (Levade et al., 1995).

These correlations are sparse and could be better developed with additional patients.

1.4.10 Source of ceramide accumulation in FD patients

Ceramide accumulates in FD due to ACDase deficiency, but there are several sources of ceramide: de novo synthesis, conversion from other sphingolipids, or ingested in the diet.

Identifying the main source(s) of the lipid is critical for delivering effective treatments, especially considering the volume of accumulation. In a patient with FD, ceramides constituted 16% of the total lipid extract of the liver, compared to 0.3% in controls

(Antonarakis et al., 1984).

One study indicates that the excess ceramide is not due to an increase in de novo ceramide formation. In vitro, cultured Farber patient fibroblasts did not generate an excessive amount of radio-labelled ceramide (van Echten-Deckert et al., 1997). This suggests that treatments that reduce de novo ceramide formation would not benefit patients with FD.

Instead, the same study suggests that the excess ceramide is due to delayed metabolism; both conversion of ceramide into other sphingolipids and degradation of ceramides were delayed in Farber cells (van Echten-Deckert et al., 1997).

1.4.11 Contribution of accumulated ceramide on cell signaling

The ceramide that accumulates in FD does not appear to participate in known cell signaling pathways. Ceramide signaling that affects the cells has been seen from ceramides in the mitochondria, the inner leaflet of the plasma membrane, and the surface of caveolae lipid rafts (Liu & Anderson, 1995; Zhang et al., 1997; Birbes et al., 2001). The ceramide accumulation in FD appears to remain in the lysosome (Chatelut et al., 1996) and therefore would not participate in that signaling.

The most well-known signaling role of ceramide is to accumulate and stimulate apoptosis (Matthias et al., 2008). In FD, where cells already have an accumulation of ceramide, one would expect there to be increased apoptosis. However, this is not seen in in vitro. Fibroblasts from patients with FD do not apoptose more frequently in vitro despite having more ceramides (Tohyama et al., 1999).

This may be because the organelle in which ACDase is most active is the lysosome, an acidic organelle, and ceramide that accumulates in acidic compartments does not trigger apoptosis (Ségui et al., 2000). In fact, lymphocytes and fibroblasts from patients with FD are equally sensitive, not more sensitive, to externally-introduced apoptotic stimuli as normal cells (Ségui et al., 2000; Burek et al., 2001).

One instance when Farber cells are more susceptible to apoptosis in vitro is when cell-permeable ceramides are introduced into the medium. Farber cells are more sensitive to apoptosis in this situation because their AC deficiency slows down the ability to metabolize this excess ceramide (Burek et al., 2001).

1.5 Mouse models of FD

1.5.1 ASAH1 knock-out mouse

The first attempt at making a mouse model of FD was using a knock-out strategy. No homozygous (Hom) knockout mice, Asah1-/-, were born, nor were any Hom embryos identified at E8.5 or later (Li et al., 2002). This is because without ACDase, embryos cannot progress from the 2-cell stage to the 4-cell stage (Eliyahu et al., 2007). This identifies

ACDase as a fundamental molecule for early embryonic development.

Mice heterozygous (Het) for the knock-out, Asah1+/-, were viable and appeared normal (Li et al., 2002). This is similar to heterozygous patients: the parents of FD patients appear normal. Het mice had a normal lifespan of at least 1.5 years and no observable deficits

(Li et al., 2002).

Only after several months could some changes be seen in aging Het mice. Lipid inclusions in the liver, lung, skin, and bone could be seen after 6 months. These were most remarkable in the liver, which became pale and fibrous. Most liver cell types had lipid-filled inclusions, and these were most significant in Kupffer cells. By 9 months, some ceramide accumulation could be seen. The greatest accumulation was in the liver at 1.5-2 fold over wild-type (WT) mice. (Li et al., 2002).

The most dramatic effect seen in these Het knock-out mice was a change in the expected offspring after long-term breeding (Park et al., 2006). Initially, offspring were born in a 1:2 WT:Het ratio, as expected. As the breeding pairs aged, they produced offspring at a ratio of 1:10 WT:Het. This suggests that as Het mice accumulate ceramide in half of their cells, their gametes with the accumulated ceramide are more likely to be fertilized (Park et al., 2006). Indeed increased ceramide was reported in ovarian tissue (Li et al., 2002). This ceramide could only be accumulating in the Asah1- gametes, not the Asah1+ gametes, suggesting that lipid accumulation aids in fertility.

1.5.2 ASAH1 conditional knock-out mouse

Because full knock-out of ASAH1 proved lethal (Li et al., 2002) a conditional knock- out was developed (Eliyahu et al., 2012). Gene excision was tamoxifen-inducible.

Intraperitoneal delivery of tamoxifen in 5-week-old female mice rendered them infertile: the few mice that were able to become pregnant did not deliver viable pups. The infertility was due to a lack of mature follicles in the ovaries. Follicles could not develop fully and apoptosed during their transition from the secondary to antral stages, indicating that ACDase is essential for follicle maturation and fertility (Eliyahu et al., 2012).

Although the tamoxifen injection induced ASAH1 excision throughout the body, albeit to various penetration (100% excision was reported in the skin, and 70% in the ovaries

(Eliyahu et al., 2012)), the authors limited their study to the ovaries. This is informative for fertility research, but not for FD, where most patents do not even survive to sexual maturity.

1.5.3 ASAH1 knock-in mouse

1.5.3.1 ASAH1 Mutation

To develop a more thorough understanding of the systemic effects of ACDase deficiency, we recently generated the first viable mouse model of FD using a knock-in approach (Alayoubi et al., 2013). An ASAH1 patient mutation (P362R) was introduced into the analogous murine locus (P361R), resulting in a mouse model that recapitulates the phenotype of FD.

The P362R mutation selected was identified in two patients with FD. One patient was homoallelic for this mutation, had severe FD, and died at age 1.5 years. The other was heteroallelic for P362R and E138V, had a milder disease, and died at age 8 (Li et al., 1999).

In vitro expression of ACDase with either of these mutations had drastically lower activity than WT ACDase (Li et al., 1999).

This mutation site was chosen to be introduced into the mouse model because it is the most conserved region of the gene (Alayoubi et al., 2013). It is also not in the predicted active site when mapped onto the predicted structure of ACDase (Figure 8).

Figure 8. The P361 R mutation used in the knock-in mouse is not in the active site on the predicted structure of ACDase. The predicted P362R mutation is shown in red on the predicted human ACDase structure (top) and the corresponding P361R mutation is shown in magenta on the predicted mouse ACDase structure. The alpha chain is in cyan on the left and the beta chain in green on the right. Active site residues are in yellow with pink dots in the predicted active site.

1.5.3.2 Survival

Mice homozygous for Asah1P361R/P361R (Hom) are viable but do not breed. Similar to the conditional knock-in mouse, the ovaries have a trend towards having fewer follicles, especially those at the mature antral stage. They are also smaller and covered with less fat than WT and Het littermates (Alayoubi et al., 2013). When Het mice are mated to generate

Hom mice, the offspring are in the expected Mendelian ratios (1:2:1 for WT:Het:Hom)

(Alayoubi et al., 2013).

Hom mice have many characteristics of FD. Their reduced lifespan of 7-13 weeks mimics the childhood lethality seen in patients. They are smaller than their littermates, a difference that is quantifiable from 4 weeks onwards (Alayoubi et al., 2013, independent repeat in Figure 9).

Figure 9. Farber mice weigh less than WT and Het littermates and their weight decreases over time. WT, Het, and Hom littermates were weighed at different ages, and results were similar to those published in Alayoubi et al., 2013. Hom mouse weight was significantly different between 5 and 9 weeks.

There are many features in these mice that are similar to one report of an aged Farber patient (29.5-years-old). The patient was, “extremely thin and the subcutaneous fat was almost absent” (Fiumara et al., 1993). As Farber mice age, their weight declines (Alayoubi et al., 2013) and their subcutaneous fat appears to reduce (unpublished observation, reduction in fat is visualized in the ovaries (Alayoubi et al., 2013)).

The cause of death of Farber mice is unknown. At some point between 7 and 13 weeks of age, Hom mice become moribund: their movements are minimal and with less vigour, and it is easy to pick them up without resistance. Their skin is extremely tight and it is difficult to scruff them. Pelvic prolapse occurs in all aged male Hom mice. As the mice progress towards a moribund state, they are unable to urinate and are sacrificed.

1.5.3.3 Ceramide accumulation

By 7-10 weeks of age, the ASAH1 mutation reduces ACDase activity in Hom mice in all organs tested: the spleen, liver, brain, and heart. This results in robust ceramide accumulation in all of these organs, and the lung and kidney (Alayoubi et al., 2013).

The degree of ACDase reduction and ceramide accumulation varies by organ. The greatest reduction in ACDase activity is in the spleen, thymus, and liver. The greatest ceramide accumulation is in the spleen, and the least accumulation was in the brain, suggesting that the brain may be able to better manage excess lipids (Alayoubi et al., 2013).

Importantly, the amount of immature ACDase protein detected by Western blot analysis is not reduced, only the activity of the enzyme (Alayoubi et al., 2013). This indicates that the P361R mutation reduces the activity of ACDase but does not promote its

52 degradation. The subcellular localization of ACDase has not been examined, and it is unknown if the mutation affects organelle distribution.

1.5.3.4 Hematopoietic phenotype

As in Farber patients, Hom Farber mice exhibit hepatosplenomegaly, as well as an enlarged thymus and lymph nodes. The spleen is paler in Hom mice and has a firm texture

(Alayoubi et al., 2013).

Leukocytosis occurs in the mouse by 7-10 weeks of age. Specifically neutrophils, monocytes, and eosinophils accumulate in the PB. Basophils and lymphocytes trend towards elevation, but are not significantly different in Hom mice. RBC and hemoglobin are moderately elevated in Hom mice over Het but not WT mice (Alayoubi et al., 2013).

As in patients, the macrophage infiltration is broad; foamy macrophages are present in almost all organs investigated. In 7-10-week-old mice, histiocytes accumulate and disrupt the parenchyma in the spleen, liver, thymus, BM, LN, skin, and nerves. There is also an increase in non-foamy macrophages and neutrophils. The only organs where no histological changes are found are the kidneys, testes, heart, and skeletal muscle. (Alayoubi et al., 2013).

The cause and characteristics of these histiocytes are unclear. There appears to be a defect in hematopoietic cell development, and investigation of the BM and HSC may reveal the defect. As well, the reason for hepatosplenomegaly was not identified.

1.5.3.5 Neurological phenotype

MRI scans of the brains of 10-week-old mice reveal hydrocephaly in 5 out of 7 Hom mice, but in no Het or WT mice. Excess histiocytes interrupt axons in both the CNS and

PNS: in the spinal cord and the sciatic nerve macrophages are present in between axons and disrupt myelin (Alayoubi et al., 2013). Lipid inclusions in Schwann cells are so prominent that they cause the Schwann cell to compress the axons that it ensheaths (Figure 6). These changes likely negatively affect Farber mice, yet their effects have not been investigated.

1.5.3.6 Cytokines and ceramides in the mouse model of FD

Given the broad distribution of the macrophage infiltration, signaling molecules in the blood may play a role in monocyte recruitment. To investigate this, the plasma of Farber mice was analyzed for a small set of inflammatory cytokines (Alayoubi et al., 2013).

Monocyte chemotactic protein-1 (MCP-1) and interleukin (IL)-12(p40) are elevated in the plasma of 7-9 week old Hom mice. MCP-1 promotes monocyte migration from the blood into organs down a concentration gradient. Correspondingly, the authors identified that

MCP-1 is also elevated in organs that contain macrophages, specifically, the liver, thymus, spleen, and brain of 9-week-old Hom mice (Alayoubi et al., 2013).

Ceramides are another class of molecules that contribute to cell signaling. As the primary feature of FD, they likely influence the Farber pathology. Only bulk ceramide was measured in Farber mice, and this was found to be elevated in all organs tested (Alayoubi et al., 2013). A detailed exploration of the many ceramide and sphingolipid species is needed.

1.5.3.7 Gene therapy of the mouse model

Many features of the Farber mouse are similar to those in FD patients, indicating that this knock-in mouse recapitulates FD and is a good model that can be used to better understand the disease and develop therapies.

1.5.3.7.1 Improvements following gene therapy

Gene therapy was tested in the Farber mouse model. A lentiviral vector (LV) expressing WT ACDase (LV/ACDase) was generated to test its therapeutic ability in Farber mice (Alayoubi et al., 2013). When compared to a control LV expressing enhanced green fluorescent protein (enGFP) (LV/enGFP), treatment with LV/ACDase improves the phenotype of Hom mice. A single, systemic bolus injection of LV/ACDase in neonatal Hom mice resulted in a median increased lifespan of 13 weeks with a maximum of 16 weeks. In this time, the mice prevent weight loss until 5 weeks of age, maintaining a weight similar to their WT and Het littermates. After 5 weeks they lose weight at a similar rate as Hom mice treated with the control LV/enGFP, ending their lives at a weight that is closer to these control mice than to their healthy WT and Het littermates (Alayoubi et al., 2013).

LV/ACDase treatment also successfully reduces the levels of total liver ceramides in treated mice to WT levels. In the spleen, ceramides are reduced in treated mice but not completely, and in the brain ceramides are unchanged and remain elevated. All 3 organs have fewer macrophage accumulations (though the brain is only mildly improved). Macrophage accumulation is still present in the spinal cord, sciatic nerve, thymus, LN, BM, lung, and skin

(Alayoubi et al., 2013).

Some positive effects of LV treatment are seen on PB cells. Leukocytes and eosinophils are reduced in 10-12-week-old LV/ACDase-treated Hom mice, but neutrophils, monocytes, basophils, and lymphocytes only have a trend towards reduction. (Alayoubi et al., 2013).

1.5.3.7.2 Limitations of gene therapy in FD mice

While very impressive for a single, systemic dose of lentivirus, LV/ACDase treatment does not correct many important phenotypes of Hom mice. The lifespan of Hom mice is increased by a few weeks, ceramides are not reduced in the brain, and macrophages still infiltrate all of the same organs, though to a lesser extent in some. Moribund male mice still develop pelvic prolapse.

A greater improvement in phenotype may be seen with long-term exposure to

ACDase. This may be from repeated systemic doses of LV/ACDase, or from coupling gene therapy with HSCT to create circulating cells that continuously express functional ACDase.

Asah1P361R/P361R mice effectively model of FD and can be used to better understand the phenotypes of the disease and to test novel treatment modalities.

Chapter 2 Aims/Hypotheses

2.1 Aim 1: To elucidate the effects of FD on different organ

systems in a mouse model.

Hypothesis: Hematopoietic and neurological tissues are negatively affected by FD.

ACDase is ubiquitous and its reduced functionality is expected to affect a wide-range of cell types. Due to the severity of classical FD and the young age of death, there has been limited opportunity to investigate the effects of ACDase deficiency on individual organ systems. This aim will be investigated using the mouse model of FD, as it facilitates organ collection, manipulation, and allows for sufficient numbers of subjects to study.

Macrophage accumulation and hepatosplenomegaly are key features of FD.

Hematopoietic organs including the liver, spleen, and BM (the source of macrophages), will be explored. Abnormalities in hematopoiesis will be investigated.

Some patients with FD have neurological manifestations, and there are patients with an only neurological phenotype due to ACDase deficiency, implicating that the brain may be affected. The neurological phenotype of Hom mice will be described, followed by in-depth investigation of the cells in the brain and the ceramide species that accumulate in this organ.

This aim is explored in Chapter 3 and Chapter 4.

2.2 Aim 2: To improve diagnosis of FD.

Hypothesis: FD can be diagnosed by circulating cytokines and ceramides.

FD is underdiagnosed due to its rarity and similarity to JIA. However, JIA is not caused by ACDase deficiency. The only treatment available for patients with FD is HSCT, and misdiagnosed patients would not receive this treatment.

Unique markers for FD may be in the PB. ACDase deficiency leads to ubiquitous ceramide accumulation, including in the PB. An imbalance in ceramides may result in abnormal cell signaling, including cytokine expression. Ceramides and cytokines in the PB should be different in patients and mice with FD compared to controls and JIA patients.

These differences will be identified and compared to cytokines in other LSDs.

The blood markers in FD will also be compared before and after treatment by HSCT.

HSCT has been shown to reduce nodules, suggesting that PB cytokines may be effected by

HSCT.

This aim is explored in Chapter 5.

2.3 Aim 3: To identify the benefits of HSCT treatment in a mouse model.

Hypothesis: Consistent, systemic availability of ACDase will improve FD signs in mice.

HSCT should theoretically cure FD. HSCT in patients with FD delivers donor cells that endogenously express WT ACDase. The blood cells generated circulate and deliver

ACDase systemically.

HSCT has only been effective in a subset of patients with FD, and is not curative in those patients. The patients in which it is ineffective are those who have a neurological phenotype before transplantation or develop it after transplantation.

The Farber mouse is the ideal model in which to identify the benefits and drawbacks of HSCT in patients with a neurological phenotype. Physiological changes following HSCT will be investigated. These include changes to the hematopoietic and neurological phenotypes identified in Aim 1, and changes in the PB markers identified in Aim 2.

This aim is explored in Chapter 6.

Chapter 3

Markedly perturbed hematopoiesis in ACDase deficient mice

This work is adapted from Dworski et al., 2015 with permission from Haematologica.

3.1 Introduction

Acid ceramidase (ACDase) is ubiquitous and catalyzes the degradation of ceramide.

ACDase and ceramides have been implicated in many disorders, including cancer, obesity, diabetes, inflammation, and neurodegenerative diseases (Morad et al., 2013; Bikman et al.,

2012; Gomez-Muñoz et al., 2013; Mencarelli et al., 2013; Mielke et al., 2012). The range of disorders that ACDase is involved in is not surprising given its pivotal biochemical position: it controls the balance between ceramide, a classically pro-apoptotic molecule, and sphingosine, which can be phosphorylated to Sph-1P, a classically pro-proliferative molecule

(Bieberich et al., 2008).

Given its central position in cell fate determination, ACDase is an essential enzyme.

Homozygous deletion of the ACDase gene, ASAH1, resulted in embryonic lethality at the 4- cell embryo stage in mice (Eliyahu et al., 2007), and no human cases have been reported with homozygous null deletions. Deficiencies in ACDase activity lead to FD, but the specific role of this reaction in hematopoiesis has not been elucidated.

Due to the rarity of patients diagnosed with FD and their shortened lifespan, hematopoietic defects of the disorder have only been described superficially. Case reports have mentioned some hematological manifestations, but they have presented inconsistently.

These manifestations include hepatosplenomegaly, enlargement and calcification of axillary lymph nodes (LN), PB leukocytosis, anemia, thrombocytopenia, and occasional nucleated

RBCs (Levade et al., 2009; Antonarakis et al., 1984; Mondal et al., 2009; Fujiwaki et al.,

2002).

Most Farber patients present with subcutaneous nodules that may contain foamy histiocytes (Mondal et al., 2009; Fujiwaki et al., 2002; Burck et al., 1985; Devi et al., 2006;

Kattner et al., 1997; Cvitanovic-Sojat et al., 2011; Pellissier et al., 1986; Kim et al., 1998).

Histiocytic infiltrations have also been found in the liver, spleen, BM, lungs, thymus, LN, heart, spine, and peritoneal fluid of patients (Antonarakis et al., 1984; Kattner et al., 1997;

Van Lijnschoten et al., 2003; Jarisch et al., 2013). These histiocytes have been described as both normal in appearance and also foamy (Antonarakis et al., 1984). Histiocytes in the liver were found to contain cytoplasmic accumulation of soluble glycoproteins or neutral mucopolysaccharides in a single membrane-bound vesicle (Antonarakis et al., 1984). In contrast, circulating leukocytes appeared normal (not foamy, not containing inclusions) in another study (Burck et al., 1985).

To develop a more thorough understanding of the systemic effects of ACDase deficiency, we generated the first viable mouse model of FD (Alayoubi et al., 2013). A patient mutation observed in ASAH1 that gave some minimal residual enzyme activity was introduced into the analogous murine locus resulting in a model that recapitulates the manifestations of FD. Homozygous Asah1P361R/P361R mice (Hom) are smaller than their Het

61 and WT littermates. Hom mice have a reduced lifespan of 7-13 weeks. Robust ceramide accumulation was seen in all organs tested. Leukocytosis, specifically neutrophilia and monocytosis, was present in the mouse (Alayoubi et al., 2013).

Given the varied and drastic effects ACDase deficiency causes in Farber patients, we hypothesized that ACDase deficiency affects hematopoiesis directly. We establish that the organ enlargement is due to a gradual accumulation of foamy macrophages, which destroys the tissue architecture. Lymphoid progenitors in the BM and thymus of Hom mice were severely reduced over time and myeloid progenitors were increased. Both progenitor populations were not intrinsically altered by ceramide accumulation, and, importantly, were able to reconstitute a WT mouse without inducing FD.

3.2 Methods

3.2.1 Mice

Mice heterozygous (Het) or homozygous (Hom) for the P361R mutation

(Asah1P361R/P361R) were used, as well as WT littermates (Alayoubi et al., 2013). All mice used were bred from Hets and were born in normal Mendelian proportions. Mice were on a mixed

CD1/129 background. Mice were housed at the Animal Resource Centre at the University

Health Network under specific pathogen-free conditions. Experiments were performed with male and female mice at ages stated in the text. All animal procedures were approved by the

University Health Network Animal Care Committee.

3.2.2 Organ analyses

Spleens, LN, thymuses, livers, and kidneys were collected and weighed from 3, 5, 7, and 9 week-old mice upon killing. Macroscopic images were taken with a Canon EOS 20D camera with a 100.0 mm lens and adjusted using Adobe Photoshop CS2. Spleens and thymuses were forced through a 40 µM nylon cell strainer (BD, Falcon) and the isolated cells were counted. BM was harvested by flushing the bones with media and the cells were counted.

3.2.3 Histology

Spleens, LN, thymuses, livers, and kidneys were collected and weighed from 5-, 7-, and 9-week-old mice. Tibias, thymuses, spleens, and LN of WT and Hom mice were fixed for 24 hours in 10% buffered formalin, embedded in paraffin, and sectioned. Bones were decalcified prior to sectioning. Sections were stained using antibodies against B220 (BD

Bioscience; #550286, 1:100), CD3 (Sigma; #C7930, 1:1000), and mac-2 myeloid cells

(Cedarlane; #CL8942AP, 1:2000). Biotinylated Anti-Rat IgG (H+L) and DAB were used to detect the antibodies in the Vectorstain ABC kit Elite Standard (Vector Laboratories).

Macroscopic images were taken with a Canon EOS 20D camera with a 100.0 mm lens and adjusted using Adobe Photoshop CS2.

3.2.4 Flow cytometry

Flow cytometry was performed on cells from the BM, thymus, spleen, LN, and PB.

The following techniques were used to prepare cell suspensions: LN were teased apart in

63 phosphate-buffered saline (PBS) containing 2% fetal calf serum (FCS) using syringe needles.

Cells from thymuses and spleens were forced through a 40 µM nylon cell strainer in PBS with 2% FCS. BM was harvested by flushing femurs and tibias with PBS containing 2%

FCS. After RBC lysis, cells were washed and resuspended in PBS with 3% FCS and counted.

PB was harvested by cardiac perfusion; cells were collected into an EDTA-coated tube. For

FACS analysis, cells were stained for 30 min at 4ºC with a combination of antibodies.

Antibodies were directly conjugated to FITC, PE, biotin, or APC and were purchased from

BD Biosciences: CD19 FITC (clone MB19-1), IgD FITC (clone 11-26c.2a), CD117 PE

(clone ACK45), CD43 PE (clone S7), CD2 PE (clone RM2-5), IL7R PE (clone A7R34),

B220 biotin (clone RA3-6B2), CD2 biotin (clone RM2-5), IgM biotin (clone 33.60) (Kincade et al., 1980), B220 APC (clone RA3-6B2), CD117 APC (clone ACK45), and CD19 APC

(clone MB19-1)). PerCP Streptavidin (Streptavidin-Peridinin Chlorophyll-a Protein,

BioLegend) was used as a second-step reagent for the indirect immunofluorescent staining of cells in combination with biotinylated primary antibodies. FACS analysis was performed on a FACSCalibur or LSRFortessa flow cytometer (BD Biosciences) and CellQuest or

FACSDiva software (BD Biosciences), and data analysis was conducted using FlowJo (Tree

Star, Inc.).

3.2.5 Colony-forming cell assays

BM and spleen were harvested from WT, Het, and Hom mice at 5, 7, and 9 weeks of age. For each time point and genotype, at least 3 males and 3 females were used as donors.

Cells from each mouse were plated in triplicate and the results of the three plates were averaged. BM was aspirated from the femurs and tibias of mice with Iscove’s Modified

Dulbecco’s Medium (Invitrogen) with 2% FBS. Cells were collected and plated in triplicate at 2x105 cells/ml in 35 mm culture dishes in MethoCult M03434 (StemCell Technologies), containing recombinant mouse (rm) SCF, rm IL-3, rh IL-6, and recombinant human (rh) Epo.

Spleens and livers were diced and forced through a 40 um cell strainer, centrifuged, counted, and plated at 1x106 cells/ml. After 11-13 days of incubation, colonies were scored. Colonies were picked and stained with May-Grunwald Stain (Sigma) and Modified GIEMSA Stain

(Sigma) to identify colony types.

3.2.6 Lymphocyte proliferation assay

Spleens were harvested as above. B cells were isolated using the EasySep Mouse

CD19 Positive Selection Kit (Stemcell Technologies), and T cells were isolated using the

EasySep Mouse T cell Isolation Kit (Stemcell Technologies). Responses to IL-7, lipopolysaccharide (LPS), anti-mu, and concanavalin A (conA) were measured by quantifying the extent of tritiated thymidine incorporation during proliferation by a Betaplate

Counter (Packard Instrument Co.; Velazquez et al., 2002). Data were pooled from 2 experiments, each comparing WT and Hom samples.

3.2.7 Bone marrow transplantation

Hom donor mice (4-5 weeks-old) received an intraperitoneal injection of 150 mg/kg

5-fluorouracil (Sigma) to enrich for HSCs. Four days later, BM was harvested from the tibia and femurs of donors and peripheral blood mononuclear cells (PBMCs) were isolated over a

NycoPrep 1.077 gradient (Axis-Shield; Takenaka et al., 2000; Yoshimitsu et al., 2007; Liang

65 et al., 2007). Recipient WT and Het mice (5 weeks-old) received a lethal dose of radiation

(11 Gy) followed by tail vein injection of the donor PBMCs. Control mice received radiation alone. Experiments were performed on 2 independent occasions and data were pooled.

3.2.8 Statistical analysis

Data and graphs represent mean ± standard error of the mean. GraphPad Prism v5.0 was used to calculate statistical significance with 2-tailed unpaired t test or with a one-way analysis of variance with a Bonferroni post-hoc test comparing WT and Hom mice at each week. Survival significance was assessed by the log-rank Mantel-Cox test in Graph Pad

Prism 5.0.

3.3 Results

3.3.1 Hematopoietic organs from Hom mice are enlarged and have

reduced cellularity

We previously reported that Hom mice are smaller than their littermates and have visually enlarged spleens, thymuses, and LN (Alayoubi et al., 2013). Here we have quantified the size increase (Table 5) and have found that the liver is also enlarged, and non- hematopoietic organs such as the kidney (Figure 10E-F; Table 5) and heart (data not shown) are not enlarged. The enlarged organs are also visibly different upon inspection: the spleen is pale pink instead of deep red (Figure 10B), the thymus occasionally has bloody red or necrotic black patches (Figure 10C), the LN are occasionally brown (data not shown), and the BM is pale pink instead of red (data not shown).

Interestingly, despite the enlargement of the hematopoietic organs, the total number of cells recovered was not increased. In fact, it was either similar (as in the LN) or reduced by 41-84% (as in the BM, spleen, thymus; Table 5). This cell number reduction is gradual and occurs near the end of the lifespan of the animal. For example, at age 3 weeks and 5 weeks the cell counts in these organs are similar to that from the WT mouse, but by 7 and 9 weeks the cell numbers decrease (Figure 10G-J).

Measurement WT Het Hom BMT, Fold- Fold Fold (unit ± (unit ± (unit ± (unit ± difference, difference, difference, SEM) SEM) SEM) SEM) Hom/WT BMT/Hom BMT/WT Body weight 28.36 ± 26.43 ± 11.87 ± 30.82 ± 0.419x 2.60x 1.09x (g) 1.21 1.05 0.29 2.97 *** *** N.S.

Spleen % 0.29% ± 0.29% ± 1.2% ± 0.92% ± 4.10x .0621x 3.15x weight 0.02% 0.03% 0.1% 0.68% *** NS NS Thymus % 0.27% ± 0.29% ± 0.78% ± 0.17% 3.00x 0.0101x 0.64x weight 0.02% 0.02% 0.12% *** ND ND Lymph node 0.0089% ± 0.013% ± 0.11% ± 0.0049% ± 12.2x 0.0003x 0.56x % weight 0.0020% 0.001% 0.01% 0.0008% *** *** NS Liver % 5.4% ± 5.3% ± 7.4% ± 4.9% ± 1.37x 0.3774x 0.91x weight 0.3% 0.4% 0.5% 0.7% ** * NS Kidney % 0.81% ± 0.90% ± 1.0% ± 0.81% ± 1.27x 0.0582x 0.99x weight 0.08% 0.11% 0.1% 0.08% NS NS NS

Total bone 4.17E7 ± ND 2.44E7 ± 1.22E8 ± 0.585x 5.0x 3x marrow cells 0.55E7 0.28E7 2.15E7 ** *** *** per leg Total spleen 1.17E8 ± ND 3.75E7 ± 2.06E9 ± 0.321x 55x 17x cells 0.26E8 0.99E7 1.64E9 ** * * Total thymus 8.96E7 ± ND 1.42E7 ± 4.2E8 0.158x 30x 5x cells 0.53E7 0.63E7 *** ND ND Total lymph 2.56E6 ± ND 4.01E6 ± ND 1.57x ND ND node cells 0.41E6 1.93E6 NS

Table 5. Weights and total cells isolated from WT, Het, Hom, and BMT mouse organs. WT, Het, and Hom mice were weighed at age nine weeks. BMT mice were weighed at six months following control mice transplanted with age-matched Hom donor cells. Body weight is expressed in grams (n=11-13 for WT, Het, Hom; n=4 for BMT).Organ values are expressed as percent of body weight (n=4-9 per group). Lymph node weight was calculated by averaging 2-4 proper axillary and accessory axillary lymph nodes. Total cells were counted from bone marrow and spleen (n=15-16 for WT, Het, Hom; n=4 for BMT) and from lymph nodes and thymus (n=4 per group, except for thymus BMT where n=1 (thymuses could not be seen in the other 3 mice)). Significance for an unpaired, two-tailed t-test is listed under each fold difference. NS, nonsignificant; ND, not determined.

Figure 10. Hom mice are smaller but their hematopoietic organs are enlarged and do not have more cells. Similar to what we previously demonstrated at 8 weeks (Alayoubi et al., 2013), closer examination of an even larger cohort of animals at 9 weeks demonstrates that (A) Hom mice are smaller than WT and Het littermates, but their organs involved in hematopoiesis are enlarged: (B) the spleen is enlarged and pale, (C) the thymus is enlarged with occasional necrotic patches, (D) and the axillary LN (and others not shown) are severely enlarged. (E) The liver is not visibly enlarged but is a greater percent of the mouse body weight (see Table 5). (F) The kidneys are not enlarged. Bars represent 1 cm. Cells were isolated from the BM by flushing, and from spleen, thymus, and LNs from mice of 3, 5, 7, and 9 weeks of age by pushing tissue through a cell strainer. (G) BM and (H) spleen cells are similar at age 3, 5, and 7 weeks in WT and Hom mice but there is a significant reduction in cells in Asah1P361R/P361R mice at 9 weeks, near the end of their lifespan. (I) Thymus cells begin to be reduced at 7 weeks. (J) LN cells are not significantly different at any age. Data points represent individual mice with a line representing the mean.

3.3.2 Organ enlargement correlates with macrophage infiltration, which

destroys organ architecture

Farber patients have histiocytosis in several organs and Hom mice do as well

(Alayoubi et al., 2013). In order to determine why the hematopoietic organs progressively enlarge without an increase in total cell number, histology was performed on bone, spleen,

LN, and thymus sections of 5, 7, and 9 week-old Farber mice. Over this short time course, an increase of unstained (pale) areas could be detected in Hom animals in hematoxylin & eosin

(H&E) stained sections (Figure 11-Figure 14). Immunohistochemistry revealed that these unstained areas are clusters of histiocytes, as they stained positive for Mac-2, but not for

B220 or CD3, suggesting the infiltrating cells were of the myeloid lineage. Macrophage infiltration was also seen in the liver, albeit to a lesser extent (Figure 15), but not in the kidney and heart (Alayoubi et al., 2013). Interestingly, these Mac-2+ cells were much larger in size than lymphocytes or Mac-2+ macrophages in WT BM, and were consistent in size and shape to foamy macrophages described in other LSDs (Pacheco et al., 2008; Kanzaki et al.,

2010). Foamy macrophages increase in size and numbers and towards the end of the short

69 life span of these mice, foamy macrophages dominated the organs, and the B220+ and CD3+ cells appeared in reduced numbers in random clusters throughout the organs.

The macrophage infiltration into tissues was so severe that the architecture of the organs was disturbed. While WT thymuses showed the distinctive architecture of cortex and medulla, no such architecture could be detected in sections from Hom mice after 5 weeks of age (Figure 12). In spleens of WT mice the red pulp and white pulp are two morphologically distinct compartments, with the white pulp containing germinal centers (T cell areas) and being surrounded by marginal zones (B cell and macrophages) (Figure 13). In Hom mice the splenic architecture was severely disrupted by the foamy macrophages and B220+ and CD3+ cells appeared in random clusters throughout the organ. The macrophage infiltration in the spleen was localized to the red pulp at 5 and 7 weeks. In one mouse at 9 weeks they clustered into focal areas reminiscent of white pulp, and in another they localized to red pulp as before.

Figure 11. The BM of Hom mice fills with Mac-2+ macrophages as the mice ages, and these become foamy. Staining of the BM with hematoxylin and eosin (H/E), B220, CD3, and Mac-2 revealed fewer B220+ and CD3+ cells, and instead a massive increase of large, Mac-2+ cells reminiscent of foamy macrophages. BM, bone marrow.

Figure 12. The thymus of Hom mice fills with Mac-2+ macrophages and the cortex/medulla architecture is lost. Staining of the thymus with hematoxylin and eosin (H/E), B220, CD3, and Mac-2 revealed fewer B220+ and CD3+ cells, and instead a massive increase of Mac-2+ macrophages as Hom mice increase in age. The characteristic cortex/medulla architecture of the thymus is destroyed by 7 weeks.

Figure 13. The spleen of Hom mice fills with Mac-2+ macrophages and the germinal center/marginal zone architecture is lost. Staining of the spleen, thymus, and LN with hematoxylin and eosin (H/E), B220, CD3, and Mac-2 revealed fewer B220+ and CD3+ cells, and instead a massive increase of Mac-2+ macrophages. Destroyed germinal center/marginal zone architecture of the spleen is seen. The bottom 2 rows are from a different Hom mouse than the 2 rows above them.

Figure 14. The LN of Hom mice are filled with Mac-2+ macrophages by 9 weeks-of-age. Staining of the LN with hematoxylin and eosin (H/E), B220, CD3, and Mac-2 revealed fewer B220+ and CD3+ cells, and instead a massive increase of large, Mac-2+ cells reminiscent of foamy macrophages. LN, lymph nodes.

Figure 15. The liver of Hom mice has a modest increase in Mac-2+ macrophages. Staining of the liver with hematoxylin and eosin (H/E), B220, CD3, and Mac-2 revealed a slight increase in Mac-2+ macrophages as Hom mice aged.

These Mac-2+ cells were much larger in size than lymphocytes or Mac-2+ macrophages in WT BM (Figure 16A) and were consistent with foamy macrophages

75 described in other LSDs (Pacheco et al., 2008; Kanzaki et al., 2010). These macrophages were also F4/80- CD23-, and could bind IgM (Figure 16B). However, they were not “sea blue” after May-Grunwald and Giemsa staining (Figure 16C), as described in Niemann-Pick disease (Sharma et al., 2009). These foamy macrophages increased in size and number from

5 to 9 weeks (Figure 11-Figure 13; Figure 15). Histiocytic infiltrations have been found in biopsies of the liver, spleen, BM, lungs, thymus, LN, heart, spine, and peritoneal fluid of

Farber patients (Antonarakis et al., 1984; Kattner et al., 1997; Van Lijnschoten et al., 2003;

Jarisch et al., 2014). The age of onset of dermal nodules (filled with histiocytes) may correlate with life expectancy (Burck et al., 1985). Antonarakis et al. describe a patient with normal PB counts initially that rapidly succumbed to the disorder as the PB counts increased; post-mortem analysis revealed massive histiocytic infiltration in organs (Antonarakis et al.,

1984).

Figure 16. Phenotyping of foamy macrophages. (A) Foamy macrophages are severely enlarged and express Mac-2 (consecutive panels are a close-up of boxed area) (B) The spleen of a 9-week-old Hom mouse was stained with hematoxylin and eosin (H/E), F4/80, IgM, and CD23. (C) Bone marrow from a 9-week-old WT and Hom mouse was stained with May-Grunwald-Giemsa (MGG) stain. Scale bar represents 50 μm.

Foamy macrophages are thought to develop from regular monocytic cells during pathological processes. Accumulation of non-foamy granulocytic myeloid cells was seen in the thymus, LNs, and spleen as determined by flow cytometry analyses of Mac-1 and Gr-1 expression (Figure 17A,B). In contrast, Mac-1+ Gr-1+ cells were present at normal numbers in the BM, where Mac-1 and Gr-1 mark monocytes (Figure 17B). This suggests that the developing monocytes are present at normal levels in Hom mice, a finding supported by results shown in Figure 21.

Figure 17. Granulocytes accumulate in the thymus, LN and spleen, but not in the bone marrow. Thymus, spleen, BM, and LN tissue from 9 week-old mice were forced through a cell strainer and stained for Mac-1 and Gr-1. (A) Mac-1+ Gr-1+ cells constituted a greater percentage of thymus cells in Hom mice. (B) The absolute number of Mac-1+ Gr-1+ cells was increased in the thymus and LN but are unchanged in the spleen and BM. LN, lymph nodes; BM, bone marrow.

3.3.3 B cell progenitors are dramatically reduced as Hom mice age

Flow cytometry was performed on BM, thymus, spleen, LN, and PB cells from

Farber and control mice aged 3, 5, 7, and 9 weeks. Flow cytometry analyses revealed a specific decline of the B cell progenitor populations in the BM of Hom mice, both in population percentages (Figure 18A-F) and absolute cell numbers (Figure 18H). This includes pro-B, pre-B, immature-B, and transitional-B cells. The time-dependent decline of B cell progenitors seen in flow cytometry is consistent with data obtained from histology

(Figure 11). Despite the reduction in B cell progenitors by 9 weeks, the BM, thymus, LN, and PB had a normal amount of mature IgD+ IgM+ B cells (Figure 18G-I), but there was a decline in IgD+ IgM+ B cells in the spleen (Figure 18G).

Due to the poor fidelity of the BM environment it is possible that the progenitors have left the BM or have died due to the destruction of the niche. However, an abnormal amount of B cell progenitors were not found in the PB (Figure 18I).

Figure 18. B cell progenitors are depleted in the BM of ACDase deficient mice. Flow cytometry analysis was performed on BM from WT and Hom mice following staining for B lineage markers. (A) B cell progenitors constituted a smaller percent of total BM in Hom mice. This includes pro-B cell, pre-B cell, immature B cell, and transitional B cells. In contrast, mature B cells were present at a normal percentage in the BM. (B-F) B cell progenitors are present in the appropriate proportion at 3 weeks and begin to decline at 5 weeks. Mature B cells gradually constitute a greater percentage of the BM environment due to the decline in B cell progenitors. (G-H) The absolute number of mature B cells was unaffected, except in the spleen where it was significantly reduced. (I) Peripheral blood was collected from 9 week-old WT and Asah1P361R/P361R mice and stained. The B cell progenitors missing from the BM were not found in excess in the peripheral blood. Data represent individual mice with a line representing the mean. Pro, Pro-B cell; Pre, Pre-B cell; Imm, immature B cell; Trans, transitional B cell; BM, bone marrow.

3.3.4 Lack of B and T cell populations is not due to an inability to

respond to stimulation in Hom mice

The decrease in B cell progenitors may be due to the destruction of their niche and/or their inability to respond to IL-7, a cytokine critical for lymphocyte development. Indeed, the decline of CD2- IL7Rα+ cells, identified by flow cytometry analyses, followed the same pattern as the decline of pro-B and pre-B cells as expected (Figure 19A-B), culminating in a drastic loss of these cells by 9 weeks (Figure 19C). Yet these cells maintain the capacity to respond to IL-7 stimulation in vitro: the absolute numbers of IL-7 responsive cells is not significantly different between WT and mutant mice at 3 and 5 weeks, indicating that there are IL-7 responsive cells present in the BM. However, after 7 weeks the number and frequency of these cells declines in the niche while their ability to respond remains unchanged (Figure 19D).

Figure 19. Lymphoid progenitors are not intrinsically altered in ACDase deficient mice and mature B and T cells are able to respond to stimulation in vitro. Flow cytometry analyses were performed on 3-, 5,-, 7-, and 9-week-old BM from WT and Hom mice after staining for CD2 and IL7Rα expression. (A, C) The population of BM cells expressing IL7Rα is decreased in Hom mice at 9 weeks. (B, D) The percent and number of cells able to respond to IL-7 in vitro is normal at 3 weeks, and declines at 5 weeks and onwards.

3.3.5 CD4+ CD8+ T cell progenitors are dramatically reduced as the

Hom mice age

Consistent with our histology results, flow cytometry analyses also revealed a specific decline of the CD4+ CD8+ T cell populations in the thymuses of Hom mice (Figure 20A).

Whereas in WT mice more than 77% of thymus T cells are double positive, in Hom mice less than 9% are. Flow cytometry results showed a time-dependent and precipitous decline of

CD4+ CD8+ T cells in percentages as well as total cell numbers with disease progression in

Farber mice (Figure 20B,C). Fewer CD4+ CD8- and CD8+ CD4- single positive cells were also found in the thymus over time and by 9 weeks there was a severe decrease in these cells in the thymus and spleen, but not in the BM or LN (Figure 20D,E). Based on our histology results, T cell development in this model may thus be arrested due to massive destruction of the thymic architecture.

To measure the function of the mature T cells produced, an in vitro lymphocyte proliferation assays. The responsiveness of CD19-selected splenic B cells to LPS and anti- mu was measured by quantifying proliferation as gauged by tritiated thymidine incorporation. At 9 weeks, Hom B cells were able to respond to LPS and anti-mu. T cells were also able to respond to concanavalin A (Figure 20F). Responsiveness at 3, 5, and 7 weeks was similar between cells originating from WT and Hom animals (data not shown).

Figure 20. CD4+ CD8+ T cell progenitors are depleted in the thymus of ACDase deficient mice. (A) Flow cytometry analysis was performed on cells from 9 week-old thymuses from WT and Hom mice following staining for CD4 and CD8. CD4+ CD8+ T cells comprise ~80% of the thymus in WT mice, yet are almost absent in Hom mice. (B) CD4+ CD8+ cells are present in normal numbers in Hom mice at 3 and 5 weeks; a drastic decline is seen at 7 weeks. (C) The reduction in cell population percentage is also seen when presented as absolute cell numbers. (D-E) Mature CD4+ CD8- and CD8+ CD4- T cells are reduced in the thymus and spleen, but unchanged in the BM and LN. Data represent individual mice with a line representing the mean. BM, bone marrow; LN, lymph nodes. (F) Spleen cells from 9 week-old WT and Hom were stained for CD19 to select for B cells. Their response to IL-7, LPS, and anti-mu in vitro was similar. Spleen cells selected for CD3 (T cells) were tested for their response to conA and their response was also similar. n=4 per genotype for IL-7, LPS, and anti-mu. n=2 for conA. Experiment performed twice. US, unstimulated; LPS, lipopolysaccharide; conA, concanavalin A.

3.3.6 Differentiating myeloid progenitor cells from Hom mice were not

biased towards granulocyte or monocyte lineages

We next asked whether the monocytosis in the PB (Alayoubi et al., 2013) and the foamy macrophages found in the organs of Hom mice were due to an excess of monocyte progenitor cells in the BM. In vitro colony-forming cell (CFC) assays were used. In this controlled environment the intrinsic effects of ceramide accumulation on the myeloid progenitors could be assessed without the confounding changes in the stromal environment.

There was an increase in total myeloid progenitor cell colonies seen in BM from Hom mice at 5, 7, and 9 weeks (Figure 21A). The difference in colony numbers in the CFC assays in samples originating from Hom mice was smaller at 9 weeks compared to 5 and 7 weeks.

However, the differentiation potential of these progenitors was not biased towards a monocyte lineage; all lineages were represented at similar levels as in the WT analyses (9 weeks: Figure 21B; 5 and 7 weeks: data not shown). This is supported by the unchanged

Mac-1+ Gr-1+ cell numbers in the BM (Figure 17). The CFC results were recapitulated by

85 flow analyses of lineage- Sca1+ cKit+ (LSK) cells; Hom BM at 9 weeks had an increase in

LSK cells (Figure 21C,D).

Spleen myeloid progenitors were decreased in Hom mice only at 7 weeks but, like the

BM, the differentiation potential was similar to cells from WT controls at all weeks (9 weeks:

Figure 21A,B; 5 and 7 weeks: data not shown).

Surprisingly, Hom livers contained many more cells that formed myeloid colonies than control livers at 5 weeks, with the difference gradually decreasing at 7 and 9 weeks

(Figure 21A). The differentiation potential of the myeloid progenitors was not statistically different between genotypes (Figure 21B). This suggests that the liver may be contributing to the monocytosis seen in the PB. Overall, these data suggest that the number and location of myeloid progenitors differs in Hom mice but the intrinsic differentiation potential in a controlled, in vitro environment, is unchanged.

Figure 21. There is an excess of myeloid progenitors in the BM but they are not biased towards the monocyte lineage. (A) Myeloid cells were assessed by the CFC assay. Colony types were scored by visual examination. The differentiation ability of the progenitors in the BM, spleen, and liver of Asah1P361R/P361R mice were not significantly different from WT and Het mice. Significance was assessed by a two-way ANOVA with Bonferroni post-test. (B) More myeloid progenitors are present in the BM of Asah1P361R/P361R mice compared to WT and Het mice at all ages. In the spleen, fewer colonies were in the spleen of Asah1P361R/P361R mice at 7 weeks compared to WT mice. In the liver, an increase in myeloid progenitors was seen at 5 weeks, which

87 declined over time. CFU-Meg were rare in the CFC of both WT and Asah1P361R/P361R assays and were not included in the counts. Significance was assessed by a two-way ANOVA with Bonferroni post-test. n=6-7 per genotype per week for CFC assays. CFU, colony forming unit; G, granulocyte; M, monocyte; GM, granulocyte/ monocyte; GEMM, granulocyte/ monocyte/ megakaryocyte/ erythrocyte; BFU-E, burst forming unit erythrocyte. (C) BM from 9-week-old mice was stained for lineage markers, Sca1 and cKit (LSK) to identify HSPC. An increase in the percent of LSK cells was seen in BM from Asah1P361R/P361R mice. (D) A similar increase was seen in the absolute number of LSK cells. Significance was assessed by a one-way ANOVA with Bonferroni post-test. N=8 for WT and Asah1P361R/P361R; n=3 for Het.

3.3.7 HSPCs from Hom mice are able to reconstitute the hematopoietic

system of WT and Het mice and are not sufficient in themselves to

induce a FD phenotype

The presence of mature B and T cells that appear functional, and the assessment of myeloid progenitors by the CFC assay data, together suggest that lymphoid and myeloid progenitor cells themselves are not negatively affected by ACDase deficiency. Instead we hypothesize that there is the degeneration of the niche that can no longer support these lineages. To support this hypothesis, we tested the ability of the myeloid and lymphoid cells to differentiate in a normal environment in vivo by transplanting Hom donor cells into WT and Het mice. The hematopoietic stem and progenitor cells (HSPC) present in the Hom donor cells (BMT) were able to reconstitute hematopoiesis in this normal environment at 60-

100% donor chimerism at 8 months post-transplant as measured by CFC assay with and without G418, and PCR of these colonies (Figure 22B,C). Infusion of the Hom donor cells did not result in the WT recipient mice developing signs of Farber disease. The mice survived past the lifespan of Farber mice (Figure 22A), maintained a normal weight (Table 5;

Figure 22B), and did not develop leukocytosis (Figure 23). The spleen, LN, liver, and thymus

88 sizes were reduced to normal WT sizes (Table 5). The total amount of cells recovered from the BM, spleen, and thymus of BMT mice was greater than in control WT or Hom animals.

Additionally, these organs were not filled with foamy Mac-2+ macrophages (Figure 22E).

The single thymus that was found had a normal proportion of CD4+ CD8+ T cells (Figure

22F). Similarly, B cell progenitors were present at normal levels (Figure 22G). Together, these data suggest that Hom hematopoietic cells alone are not sufficient to induce Farber disease in a WT environment. Alternatively, WT tissues transferred low levels of functional

ACDase to the donor Hom cells through M6P receptor-mediated uptake.

Figure 22. Hom BM rescues lethally-irradiated control mice and is not sufficient to induce Farber disease. (A) BM enriched for HSPC by 5-fluorouracil pre-treatment of donors was transplanted from Hom mice into lethally-irradiated WT mice (BMT). Hom cells were able to reconstitute the hematopoietic system of the recipients. BMT mice had a lifespan longer than untreated Hom mice (range indicated by arrows). Control mice not receiving a BMT after radiation had a lethality rate of 85%. Significance was assessed by a log-rank Mantel- Cox test. (B) Transplanted mice maintained a normal weight, greater than that of Hom mice. Data represent individual mice with a line representing the mean. (C) WT mice were lethally irradiated and received a bone marrow transplant from Hom mice. Hom mice contain a neomycin cassette (as a remnant from the generation of the mouse line). This confers G418 resistance to Asah1P361R/P361R cells. To assess chimerism, bone marrow from transplanted mice was grown in cytokine-enriched methylcellulose with or without 1.5 mg/ml G418. WT bone marrow could only grow in the absence of G418, while 6 BMT samples grew similarly well in the presence or absence of G418. (D) Colonies were picked from the CFC in (B). The colonies were genotyped by PCR to determine if they came from WT or Hom mice using the same primers as when determining the genotype of mice. 60-100% of colonies were from Hom donor mice. (E) Six months post-transplant hematopoietic organs were analyzed. Macrophage infiltration was not present in the spleen in one mouse. (F) B cell progenitors were present at normal levels in the bone marrow, (G) and T cell progenitors were present at normal levels in the thymus. Genotype of mice is first followed by their treatment in brackets. Ctrl, control mice (WT or Het); rad, radiation only; Hom, BMT of Hom donor cells.

Figure 23. Transplantation of Hom cells into WT and Het mice does not recapitulate the peripheral blood symptoms of Farber disease. Hom mice develop leukocytosis by 9 weeks. When BM was transplanted from Hom cells into irradiated WT and Het mice leukocytosis was not seen. Dashed lines indicate normal range of blood cell count in mice. All comparisons are not statistically significant.

3.4 Discussion/Conclusion

3.4.1 Hematopoietic organ architecture

We previously reported that Hom mice have enlarged hematopoietic organs with leukocytosis, specifically neutrophilia and monocytosis, along with an excess of macrophages (Alayoubi et al., 2013). Here we establish that ACDase deficiency resulted in hematopoietic organs that were swollen with foamy macrophages and devoid of architecture, reminiscent of the foamy histiocytes found in patients. ACDase deficiency also resulted in the wholesale depletion of B and T cell progenitors, but did not affect the ability of the remaining mature lymphocytes to respond to stimuli in vitro, the intrinsic differentiation ability of myeloid stem and progenitor cells, or their ability to reconstitute the BM of WT mice. The coincidental timing of the increase in macrophage infiltration into the thymus and

BM with the reduction in T and B progenitor cells, respectively, suggests that destruction of the architecture by macrophages may be disrupting the niche for developing cells.

In our Farber mouse model, the age of onset of catastrophic hematological indications is 5-7 weeks. The timing of the appearance of these abnormalities seems to correlate with survival. In Farber patients, the age of onset of dermal nodules (which are filled with histiocytes) has been likewise suggested to be correlated to the life expectancy (Burck et al.,

1985). Antonarakis and colleagues describe a patient with normal PB counts initially that rapidly succumbed to the disorder as the PB counts increased; post-mortem analysis revealed

93 massive histiocytic infiltration in organs (Antonarakis et al., 1984). This suggests that the histiocyte/macrophage infiltration propels the severity of the disorder. Based on our histology results, which show a massive invasion of foamy macrophages into many tissues, we conclude that the maintenance of these populations may be affected due to the destruction of the developmental niches for all hematopoietic cell types, and that threshold ceramide accumulation is the initiator of the severe hematopoietic abnormalities.

3.4.2 Source of monocytes

We previously reported that Hom mice demonstrate monocytosis (Alayoubi et al.,

2013). The CFC results indicate that the monocytosis and foamy macrophages found in their organs is not due to a disproportionate increase of monocyte progenitor cells (CFU-M or

CFU-GM) in the BM. Similarly, we previously reported that Hom mice have an excess of

RBCs (Alayoubi et al., 2013), but it is not from a disproportionate increase of RBC progenitors (BFU-E). Our current results suggest two possible sources of the monocytes: The cells may be from the BM, as there were a greater number of myeloid progenitors observed in the CFC assay. Instead these cell may be from the liver: Hom livers contained many more myeloid CFCs than control livers at 5 weeks, with the difference gradually decreasing at 7 and 9 weeks. Extramedullary hematopoiesis has been noted in one very severe FD patient that died 3 days after birth and had hydrops fetalis (Kattner et al., 1997), and would be consistent with the emergent granulopoiesis observed during inflammation. Indeed, an accumulation of granulocytes was seen in the thymus, LN, and spleen of our Hom mice as determined by Mac-1 and Gr-1 staining.

The difference in colony numbers in the CFC assays in samples originating from

Hom mice was smaller at 9 weeks compared to 5 and 7 weeks. This may be due to the destruction of the BM niche by 9 weeks: the myeloid progenitors may not have appropriate contact with the stromal cells in order to maintain survival or to self-renew. It is also possible that the committed progenitors have begun to exhaust their self-renewal potential as they were proliferating more than normal early on to create an increase in progenitor numbers at 5 and 7 weeks.

The complete cascade of events leading from ceramide accumulation to the abovementioned phenotypic alterations remains to be elucidated. One consequence we have described previously is an increase in monocyte chemotactic protein-1 (MCP-1) levels

(Alayoubi et al., 2013). MCP-1 is a chemokine that attracts monocytes from the blood into organs by a concentration gradient. MCP-1 was found to be elevated in the plasma and organs of Farber mice (Alayoubi et al., 2013), suggesting that it may be recruiting monocytes into organs and resulting in the histiocytic infiltration. The sequence of events leading from ceramide accumulation to MCP-1 over-expression may involve the phosphorylation of ceramide to ceramide-1-phosphate (Cer-1P) (Arana et al., 2013).

3.4.3 Lymphocyte progenitor loss

Despite the absence of B and T cell progenitors in the BM and thymus of Hom mice, respectively, mature B and T cells were found in the circulation and organs. These mature cells were able to respond to stimulation in vitro in lymphocyte proliferation assays. This demonstrates that the mature B and T cells from this model maintain inherent functionality.

The presence of mature B and T cells at an age where B and T progenitors are almost completely depleted suggests that these cells were produced earlier in the life of the mouse, when B and T cells progenitors were present and the environment was able to support differentiation. It is only by 9 weeks, when the niche has been uniformly destroyed, that the progenitors are absent. Either the B and T progenitors lost their ability to differentiate or the environment lost its ability to support them. The CFC and stimulation data argues against the former.

We conclude that myeloid and lymphoid progenitor cells themselves are not negatively affected by ACDase deficiency to the extent that their differentiation ability is altered. We instead hypothesize that it is the degeneration of the niche that can no longer support these lineages. This is supported by the BM transplantation data, where Hom donor cells were able to engraft and reconstitute hematopoiesis in both a WT and Het background.

An alternative explanation for the success of the transplant that cannot be excluded is that the

WT stroma and other tissues transferred low levels of functional ACDase to the donor Hom cells through M6P receptor-mediated uptake, thereby giving them sufficient enzyme to limit substrate accumulation.

Together, these data show for the first time that ubiquitous ACDase deficiency leads to the generation of an abnormal hematopoietic environment that initiates histiocytosis, which in return leads to the complete destruction of organ architecture. ACDase deficiency does not appear to intrinsically affect the differentiation of hematopoietic progenitor cells, but our data suggest it has a detrimental effect on the environment, which provides the essential developmental niche for developing hematopoietic cells. This previously unidentified role of ACDase suggests that it could be playing a broader role in hematopoietic

96 development, as well as being a critical modulator of the disease state. Identifying the roles of ACDase and ceramide at important junctures in hematopoiesis is critical for understanding and developing therapies for FD and other disorders in which ceramide accumulates, including inflammation, cancer, obesity, and diabetes.

Chapter 4

Acid Ceramidase Deficiency in Mice Results in a Broad Range

of Central Nervous System Abnormalities

This work is adapted from Sikora & Dworski et al., 2017 with permission from the

American Journal of Pathology.

4.1 Introduction

FD is a rare autosomal recessive lysosomal disorder caused by mutations in the

ASAH1 gene (8p22) that encodes ACDase. This enzyme is a key lysosomal enzyme of sphingolipid catabolism that converts ceramide (Cer) into sphingosine (Sph) and a free fatty acid. The FD clinical phenotype is variable, with the most common variant manifesting with progressive neurovisceral disease that associates with early-onset subcutaneous nodules, progressive arthritis with joint deformities, and laryngeal hoarseness (Levade et al., 2009). At the tissue level, FD is characterized by accumulations of storage-laden macrophages. Other non-neuronal cells such as hepatocytes, endothelial cells, sweat gland epithelial cells, fibroblasts and/or Schwann cells (Abenoza & Sibley, 1987; Burck et al., 1985; Zappatini-

Tommasi et al., 1992) also demonstrate sub-cellular storage pathology. While variable between cell types, ultrastructural storage profiles are characteristic, including comma- shaped curvilinear tubular profiles (CTBs or Farber bodies), “empty” banana/spindle/needle-

98 like bodies, and zebra-like bodies (Zappatini-Tommasi et al., 1992). Even though the most severely affected FD patients present with substantial neurological deficits, an understanding of CNS pathology is limited (Levade et al., 2009; Molz 1968; Moser et al., 1969).

Importantly, an additional clinical variant in which a neuronal phenotype is not associated with classical FD symptoms has been recently described (Zhou et al., 2012). That study identified six patients with SMA-PME, with each individual carrying ASAH1 mutations and demonstrating reduced ACDase activity. Since that initial description, an additional five

SMA-PME FD patients have been reported (Dyment et al., 2014; Gan et al., 2015; Giraldez et al., 2015; Rubboli et al., 2015). These studies suggest that FD may be more common than previously believed, and that rather than occurring as discrete variants or subtypes, actually occurs as a broad, continuous spectrum of clinical phenotypes.

To study the broad impact of ACDase deficiency in the mammalian model system and the mechanisms of FD progression, we recently generated an Asah1 knock-in mouse model that carries a P361R ACDase mutation corresponding to the P362R patient variant previously identified (Li et al., 1999). While Het mice appear unaffected and similar in all measures to WT mice, mice homozygous for the P361R mutation (Asah1P361R/P361R, Hom

Farber) develop a progressive and lethal phenotype (survival is 7-13 weeks) with recapitulation of major aspects (e.g. reduced ACDase activity and bulk Cer accumulation in tissues) of human FD (Alayoubi et al., 2013). Hom Farber mice replicate the macrophage aspect of tissue pathology observed in FD patients, suggesting that the mouse model is highly relevant. Furthermore, a follow-up study identified substantial dysregulation of the hematopoietic microenvironment in the Hom Farber mice (Dworski et al., 2015). Given the critical need to explore and understand the mechanisms underlying the neurological defects

99 observed among FD patients, we describe the hallmark behavioural, biochemical, and tissue/cell pathological abnormalities in the CNS of Asah1P361R/P361R mice in this current study.

4.2 Materials and Methods

4.2.1 Animals

Mice deficient in ACDase activity (Asah1P361R/P361R) were generated by our group

(Alayoubi et al., 2013). Mutant (Asah1P361R/P361R, Hom Farber) and wild-type (Asah1+/+, WT) animals were obtained by mating heterozygous (Asah1P361R/+, Het) mice. Genotyping was performed as described previously (Alayoubi et al., 2013). Mice were housed at the Animal

Resource Centre at the University Health Network (Toronto, Canada) under specific pathogen-free conditions. All animal procedures were approved by the University Health

Network Animal Care Committee.

4.2.2 Behavioural testing

Voluntary locomotor and exploratory activities along with assessments of anxiety- like behaviour were tested in an open field arena. Mice were further profiled for a spectrum of behavioural activities by observation. Muscle strength and motor coordination abilities were assessed by a grip test and an accelerating rotating rod (rotarod) test, respectively. Open field, rotarod, and marble burying tests were performed in animals aged 7-8 weeks, and activity observations were collated from 9-week-old mice. Mice were evaluated repetitively

100 at 5, 7, and 9 weeks of age on the grip test. At 7 weeks, the sequence of tests was as follows: open field on day 1, rotating rod on day 3 and 4, and marble burying test on day 4.

During testing, circadian rhythms were observed, with all test trials performed between 8 am and 1 pm. Non-test-related variability was reduced by using the same biosafety cabinet in a closed room. Experimenter variability was reduced by having the same person perform each trial of a given test on all animals. Data from any animal disturbed by an outside interference during testing were excluded from the analyses. An automatic tracking system (TrueScan, Coulbourn Instruments, Whitehall, PA, USA) was used to record the total ambulatory movement time and distance traveled, the number of rears and the number of entries into the center zone (18.4x18.4 cm) of the arena. The last parameter was used as a measure of thigmotaxis, which is a tendency for mice to stay close to the walls of the testing apparatus because of anxiety induced by the novel environment.

For activity profiling, each mouse was observed in its home cage for 5 minutes and an experimenter (blinded to the genotype) noted what activity (sniffing, rearing, locomotion, grooming, freezing, chewing or burrowing) was being performed every 30 seconds for 5 minutes (Martin et al., 1993). In the marble burying test that we employed (Thomas et al.,

2009), mice were also first habituated in their home cage and then transferred to a clean cage

(29.5x17.5x12.5 cm) with 5 cm of bedding and 20 clean, identical navy blue marbles (1.2 cm diameter) in a 4x5 matrix in two-thirds of the cage. Every mouse was placed in the one-third of the cage with no marbles and the lid was closed. After 30 minutes, the mouse was removed and the number of marbles buried was recorded. Marbles were considered buried when covered by more than 50% with bedding. For grip testing, mice were placed on metal mesh that was boxed in by smooth metal walls (Carlson et al., 2010). This box was slowly

101 turned upside-down, allowing mice to hold on. The time to fall was recorded to a maximum of 6 minutes. Mice were tested three times with a 10-minute rest in between the trials at each age. The average of the 3 trials was used for statistical analyses. For rotarod testing, mice were placed on the device which was then accelerated from 4 to 40 rpm over 5 minutes. The time when mice fell off the rotating rod was automatically recorded. Animals were tested three times on two consecutive days.

4.2.3 Collection and processing of the brains

For magnetic resonance imaging (MRI), light, and electron microscopy (EM) studies, mice were perfused transcardially with PBS followed by 4% paraformaldehyde (PFA). The extracted brains were further immersion-fixed overnight in 4% PFA at 4°C and then transferred to and stored in PBS at 4°C.

For liquid chromatography-mass spectrometry (LC-MS) and mass spectrometry imaging (MSI) brains were isolated from 9-week-old euthanized male mice and divided along the midline sagittal plane. One half of the brain was frozen on dry ice for bulk LC-MS analyses, and the other in liquid nitrogen in the vapor phase for MSI spectrometry.

Details of the consecutive tissue processing for individual types of analyses are provided in the following methodological sections.

4.2.4 Lipid analyses by liquid chromatography-mass spectrometry

For LC-MS, the complete half-brains were homogenized in 0.25 M sucrose (to get a concentration of 20% (w/v)) using the Fast-Prep apparatus (MP Biomedicals) and green-

102 capped tubes (Lysing Matrix D). Protein concentrations were determined with bovine serum albumin (BSA) as a standard using a bicinchoninic acid protein determination kit (Thermo

Scientific) according to the manufacturer’s instructions. Sphingoid bases were measured by

LC-triple quadrupole mass spectrometer as reported (Bedia et al., 2011). Other lipids were analyzed by UPLC-TOF mass spectrometry and Exactive (Thermo Electron) as described

(Garanto et al., 2013). The number of brains analyzed was 3 for each genotype.

4.2.5 Mass spectrometry imaging

For mass spectrometry imaging (MSI), frozen brain halves were sectioned sagittally

(10 m) in the 0.24<>0.48 lateral ranges (Franklin & Paxinos, 2008) and analyzed by matrix- assisted laser desorption ionization (MALDI) imaging mass spectrometry as previously described (Jones et al., 2014). Briefly, tissue samples were mounted on ITO coated slides and sprayed with 2,5-dihydroxybenzoic acid (DHB) matrix. MALDI spectra were acquired across the entire tissue section on a Solarix dual source 7T FT-ICR mass spectrometer (Bruker

Daltonics Inc., Billerica, MA, USA) to detect the lipid species of interest in positive ion mode (m/z = 200-1200) or negative ion mode (m/z = 200 – 2000) with a SmartBeam II laser operating at 1000 Hz, a laser spot size of 25 μm, and a raster width of 200 μm for general profiling or 75 μm for high resolution images. Images were generated using FlexImaging 4.0 software (Bruker Daltonics), and structures confirmed using an in-house database of FD ceramides as previously described (Jones et al., 2014). H&E staining of the brain sections was performed after the MALDI-MSI imaging. The number of brains analyzed was 3 for each genotype.

103

4.2.6 Magnetic resonance imaging

Fixed mouse brains (5, 7, and 9-week-old) were immersed in Fomblin and scanned using a 9.4T Varian Direct Drive animal MRI/MRS system (Agilent Technologies Inc.) with a 14-mm ID single loop receiver coil (Doty Scientific Inc.) and a 7-cm ID 1H transmission coil (m2m imaging Co.). T2-weighted images (78x78x500 μm3, no gap) were acquired using an fse sequence: TR/TE = 5000/44 ms, ETL = 4, NEX = 8. Image sequences were evaluated and individual images were extracted for Figure preparation in MIPAV software (Medical

Imaging Processing and Visualization, NIH). The number of brains analyzed was 1 for each genotype and age group.

4.2.7 Histopathology, antibodies, immunohistochemistry and

immunofluorescence

Cerebra and cerebella from Hom Farber and WT mice aged 3, 5, 7, and 9 weeks were embedded in paraffin, sectioned (coronally in cerebrum and sagittally in cerebellum) 5-8 m, and stained according to the standard H&E, periodic acid-Schiff (PAS), and luxol fast blue

(LFB) protocols. Immunohistochemical (IHC) detection of cathepsin D (CathD) was performed with the monoclonal rabbit Ab (1:500, LifeSpan Biosciences) on 2-5 m thick cerebral and cerebellar sections. Anti-mouse Dako Envision kit (DAKO) was used for secondary detection according to the manufacturer’s instructions. For immunofluorescence

(IF) labeling, PFA-fixed brains were cut sagittally in the midline and then divided into cerebral and cerebellar/brain stem parts by a single coronal cut at the mesencephalic level.

The separated fragments were embedded into 8.0% sucrose/3.5% agarose. Thirty-five m

104 thick serial sections (coronally in cerebrum and sagittally in cerebellum) were cut on Leica

VT-1000S vibratome (Leica Microsystems). Sections corresponding to the rostral/caudal bregma -1.94<>-2.30 and lateral 0.40<>0.84 ranges (Franklin & Paxinos, 2008) were selected from cerebra and cerebella, respectively. Matched sections were stained by multi- labeling IF protocols (McGlynn et al., 2004) using the following primary antibodies and dilutions: mouse anti-NeuN mAb (1:1000, Chemicon), mouse anti-calbindin mAb D-28K

(1:3000, Sigma-Aldrich), rabbit anti-calbindin pAb (1:800, Chemicon), rat anti-CD68 mAb

(1:1000, AbD Serotec), mouse anti-GFAP mAb G-A-5 (1:3000, Sigma-Aldrich), mouse anti-

GM2 ganglioside IgM mAb (1:15, cell culture supernatant produced in-house from the 10–11 hybridoma line by Progenics Pharmaceuticals), mouse anti-neurofilament medium chain

(NF-M) mAb (1:500, Novus Biologicals), and guinea pig anti-p62 (1:500, American

Research Products). Anti-LAMP2 rat mAb developed by J. T. August was purchased from the Developmental Studies Hybridoma Bank developed under the auspices of the National

Institute of Child Health and Human Development and maintained by the University of Iowa,

Department of Biological Sciences. The CathD rabbit pAb was a generous gift from R. A.

Nixon (Nathan Kline Institute, Orangeburg, NY). Species-specific detection of the primary antibodies used secondary antibodies conjugated to Alexa Fluor (AF) 488 and 546 dyes

(Invitrogen). A minimum of 2-3 male brains per genotype/age group were analyzed.

4.2.8 Light microscopy

H&E, PAS, LFB, and IHC stained tissue sections were imaged using Olympus AX70

(Olympus) and Nikon E800 (Nikon, Tokyo, Japan) microscopes with MagnaFire or Olympus

DP70 cameras, respectively. Olympus IX70 microscope with HQ2 camera (Photometrics),

105

Proscan II xyz stage (Prior Scientific), and SmartSutter 10 position filter wheel shutters

(Sutter) were used to serially xy image multi-fluorescently labeled sections as previously described (Sikora et al., 2016). Exposure time for all antibody combinations and channels was 750 ms (gain 2. Excitation/emission conditions for the AF488 were exc.490/20/b.s.

480-513/em.535/40 and for AF546 the conditions were exc.572/23/b.s.555-588/em.630/60 nm. Spatially overlapping double-channeled 14-bit images were acquired with the Plan 10x

(NA 0.25) objective, down-sampled to 8-bit, and digitally stitched in Metamorph/MetaFluor

(Molecular Devices) and Photoshop CS6 (Adobe) software. To gain higher optical resolution, IF labeled sections were imaged by laser scanning confocal microscope Zeiss

Meta Duo V2 (Zeis) using either Plan AF 63x (NA 1.4) or Plan AF 20x (NA 0.40) objectives. Z-stack image datasets are presented as maximum intensity projections (MIP).

4.2.9 Electron microscopy

After collecting serial vibratome sections for light microscopy analyses, a 250 mm thick single section was cut from the cerebrum and cerebellum of 3- and 9-week-old FD mice and controls. The following anatomical regions (nomenclature and abbreviations correspond to Franklin and Paxinos (Franklin & Paxinos, 2008)) were dissected into separate tissue blocks and were processed for EM as reported before (Sikora et al., 2016; Micsenyi et al.,

2013): neocortex+hippocampus (V2ML/V2MM/CA1and V1/V2L/S1/CA1/CA2/CA3/dentate gyrus), internal capsule/hippocampal fimbria/thalamus/lateral ventricle

(DLG/VPM/VPL/fi/ic/LV/CPu), and cerebellar lobules 1-5/medial DCN. Thin sections were imaged using Philips CM10 electron microscope (Philips Electron Optics). Individual CNS cell types were identified based on their morphological characteristics - neurons, astrocytes,

106 and microglia as by Peters and colleagues (Peters et al., 1991), and oligodendrocytes as by

Fletcher and colleagues (Fletcher et al., 2014). Categorization of the vascular organization into endothelial cells, pericytes and perivascular macrophages follow Guillemin and colleagues (Guillemin et al., 2004).

4.2.10 Statistical analyses and figure preparation

Statistical analyses of the open field, marble burying, and activity observation assays were assessed by an unpaired t-test in GraphPad Prism 5 (GraphPad Software, Inc.). For the grip test and rotarod assays, significance was assessed by repeated measures ANOVA followed by t-tests performed in JMP statistical software (v.11, SAS). Significance in biochemical and weight studies was assessed by one-way ANOVA or two-way ANOVA, respectively, with

Bonferroni post-tests in GraphPad Prism 5. Graphs were plotted in GraphPad Prism 5.

Figures were prepared in Photoshop CS6 (Adobe). For better contrast and visibility, and without altering the biological information, images were stretched to fill the full dynamic 8 bit-ranges. Neuroanatomical nomenclature and abbreviations used in the manuscript correspond to Franklin and Paxinos (Franklin & Paxinos, 2008).

4.3 Results

4.3.1 Asah1P361R/P361R (Hom Farber) mice express a complex

behavioural phenotype

Hom Farber mice replicate many of the phenotypic features of FD patients (Alayoubi et al., 2013). To delineate gross behavioural and neurological consequences of ACDase deficiency

107 in this model, we first compared Hom Farber mice to WT mice in their ambulation and exploration abilities, their compulsive/stereotypic and anxiety behaviours, and their muscle strength and motor coordination capabilities. In the open field test, Hom Farber mice spent less time ambulating (Figure 24A) and covered half as much distance (Figure 24B). They also had one-fifth of the rearing movements of controls (Figure 24C), which is also considered a measure of exploratory activity. As well, Hom Farber mice displayed an increased thigmotaxis behaviour by entering the center zone one-third as frequently (Figure 24D) and increased stereotypic behaviour in the open field analysis (back and forth movements in the corners of the arena) (Figure 24E). While the marble burying test results appear to oppose the open field test results and actually suggest a reduced stereotypic behaviour (Hom mice buried fewer marbles; Figure 24F), it is more likely that Hom Farber mice buried fewer marbles because they ambulate less (Figure 24B).

108

Figure 24. Hom Farber (Hom) mice have decreased locomotor/exploration activity, increased thigmotaxis and perform poorly in grip strength and rotating rod tests. (A-E) The activity of 7-8-week-old WT and Hom mice in an open field arena was recorded for 10 minutes. (A- C) Hom mice displayed decreased voluntary locomotion and rearing. (D) Hom mice show increased thigmotaxis (tendency to stay close to the walls of the open field apparatus) behaviour by entering the center zone of the arena less frequently. (E-F) 7-8-week-old Hom mice demonstrated increased stereotype behaviour in open field test but decreased stereotype behaviour in marble burying test. G: 9-week-old Hom mice spent a greater percentage of their time chewing compared to WT animals, and less time on locomotion and rearing. (H) Hom mice underperformed in the grip test: when the same mice were subjected to the grip test at age 5, 7, and 9 weeks, Hom mice performed significantly poorer at age 9 weeks. Neither WT nor Hom mice significantly increased the length of time for which they could grip over time. (I) Hom mice fell off more quickly on an accelerating rotating rod. Both WT and Hom mice were able to learn over trials and maintained that ability to the second day of testing. *p<0.05 in open field, marble burying, and activity observation was assessed by an unpaired t-test. For the grip test and rotarod, *p<0.05 was assessed by repeated measures multivariate ANOVA, followed by t-tests. Number of animals tested: open field - WT n=8, Hom n=5; marble burying test - WT n=10, Hom n=9; activity observation - WT n=13, Hom n=12; grip test - WT n=6, Hom n=5; rotarod - WT n=8, Hom n=4. All data are presented as average ± S.E.M. Male and female animals were combined for all tests and analyses.

109

Observation-based activity assessments also demonstrated significant differences between Hom Farber mice and WT mice. Hom Farber mice spent less time rearing and locomoting and spent more time chewing (Figure 24G). This is consistent with the results from the open field test. Next, the strength of the mice was assessed biweekly using an inverted grip test. Testing began at age 5 weeks, and the same mice were tested every 2 weeks. WT and Hom Farber mice performed similarly when first tested at age 5 weeks, but by 9 weeks of age Hom Farber mice performed significantly worse than WT mice (Figure

24H). Neither genotype showed significant improvements over time. Finally, Hom Farber mice presented with impaired motor coordination, falling more quickly off the rotating rod

(Figure 24I). Both genotypes significantly improved on each trial, yet Hom Farber mice still fell off more quickly than WT mice.

4.3.2 The abnormal sphingolipid profiles in the brains of Hom Farber

mice are dominated by the accumulation of ceramide,

hydroxyceramide, and dihydroceramide

The total sphingolipid profile of WT, Het, and Hom Farber mouse brains was analyzed by LC-MS. As a family, ceramides (Cer), hydroxy-ceramides (Cer-OH), dihydroceramides (dhCer), sphingosine (Sph), dihexosylceramides (DHC), and GM3 ganglioside were all elevated in Hom Farber mice compared to WT by absolute values and by fold-increases over WT levels (Figure 25A; Figure 26A). No significant changes in sphingolipid profiles were seen in Het animals compared to WT.

110

In Hom Farber mice, the elevations resulted in altered brain sphingolipid composition: Cer, Cer-OH, dhCer, DHC, and GM3 ganglioside constituted a significantly greater percent of total brain lipids, with a concomitant reduction in the proportion of monohexosylceramides (MHC) and sphingomyelin (SM) (Figure 26B). The change in distribution due to increased Cer and Cer-OH levels was a consequence of every chain-length isoform of these classes being elevated (Figure 26C,E; Figure 25B,C) but to different degrees

(Figure 26D,F).

Cer-OH was the primary class of sphingolipids elevated in brains, increasing by more than 100-fold in Hom Farber mice (Figure 26A). All Cer-OH chain-length isoforms contributed to this elevation, each being increased more than 40-fold (Figure 26E).

Similarly, all isoform species of dhCer were elevated (Figure 25D, Figure 26G,H), resulting in the dhCer contribution to brain lipid composition tripling from 0.05% to 0.18%

(Figure 26B).

Sph content greatly exceeded dihydrosphingosine content (dhSph) in WT and Het mice; this difference was further pronounced in Hom Farber mice (Figure 25E; Figure 26I,J).

Interestingly, whereas nearly all DHC species (Figure 26O) as well as the GM3 ganglioside (Figure 25I; Figure 28) were elevated in Hom Farber animals, most MHC chain lengths were unchanged (only 16:0 was elevated (Figure 26K; Figure 25F)) resulting in a reduction in MHC as a brain lipid class (Figure 26B). Hydroxy-monohexosylceramide

(MHC-OH) was the only lipid class measured that did not have any species lengths changed

(Figure 26M; Figure 25G), nor was its proportion in brain altered (Figure 26B). Finally, most sphingomyelin (SM) species were unchanged, and the ones changed were altered to a lesser

111 fold than other sphingolipid classes (Figure 26R; Figure 25J), resulting in the overall SM proportion being decreased in global sphingolipid composition (Figure 26B).

112

Figure 25. Absolute levels of sphingolipids species are altered in Asah1P361R/P361R (Hom) mice. Sphingolipids in the brain of 9-week-old WT, Het, and Hom mice were measured by LC-MS. (A) The absolute levels of each sphingolipid group are plotted. (B-J) Each sphingolipid group’s individual species. *p<0.05 in a one-way ANOVA with Bonferroni post-tests between WT vs Hom mice. Ceramide (Cer); hydroxy-ceramide (Cer-OH); dihydroceramide (dhCer); dihexosylceramide (DHC); monohexosylceramide (MHC); hydroxylated monohexosylceramide (MHC-OH); sphingosine (Sph); dihydrosphingosine (dhSph); sphingomyelin (SM); monosialodihexosylganglioside (GM3).

113

114

Figure 26. Specific sphingolipid species are altered in Hom Farber (Hom) mice. Sphingolipids in the brain of 9-week-old WT, Het, and Hom mice were measured by LC-MS. (A) Fold elevation over WT for each sphingolipid class measured. (B) The percent composition of the various sphingolipid classes measured. (D, F, H, J, L, N, P, R) Fold elevation over WT of individual isoforms within each sphingolipid class. (C, E, G, I, K, M, O, Q, S) Each isoform chain length is shown in % relation to other chain lengths in its respective sphingolipid class. *p<0.05 in a one-way ANOVA with Bonferroni post-tests between WT vs Hom mice. All data are presented as average ± S.E.M. n=3 animals for each genotype. Ceramide (Cer); hydroxy-ceramide (Cer-OH); dihydroceramide (dhCer); sphingosine (Sph); dihydrosphingosine (dhSph); monohexosylceramide (MHC); hydroxylated monohexosylceramide (MHC-OH); dihexosylceramide (DHC,); sphingomyelin (SM); monosialodihexosylganglioside (GM3).

4.3.3 MALDI-MSI visualization of ceramide accumulation shows regional

accumulation of selected lipid species

MALDI-MSI allows the identification and mapping of biomolecules directly in tissues, and a previous study using this method and kidneys from the same Hom Farber mouse model to characterize ceramide accumulation has been published (Jones et al., 2014).

To complement the quantitative LC-MS data obtained for the different ceramide species in

Hom Farber brain tissues, MALDI-MSI was used to determine the anatomical locations of selected ceramide and ganglioside species in the same brains used for LC-MS. Two of the most abundant ceramides detected by LC-MS in Hom Farber brain were Cer 18:1/16:0 and

Cer 18:1/18:0, and images of their distribution in sagittal sections of Hom Farber and WT brains are shown in Figure 27. In the Hom Farber brains, Cer 18:1/16:0 was primarily localized in the cerebellar granule cells layer, cerebral cortex, and paraventricular/medial thalamic nuclei (Figure 27), and strikingly absent in cerebellar white matter (WM). In contrast, localization of Cer 18:1/18:0 was in FD brain was increased in septal nuclei and corpus callosum. An overlay image of the two ceramides highlights these different distributions (Figure 27), with their co-localization cerebellar granule cells layer, and the

115 predominant Cer 18:0/16:0 midbrain and Cer 18:1/18:0 cerebral cortex accumulations. This representative data for these three ceramides indicates that their distributions in the Hom

Faber brains are localized to specific brain regions, and can be differential specie to specie.

116

Figure 27. MALDI-MSI analysis reveals selective distribution of ceramides by specific brain region in 9- week-old Hom Farber (Hom) mice. (A) Sagittal tissue sections of Hom Farber and WT brains were analyzed by MALDI-MSI in positive ion mode (B-E) as previously described (Jones et al., 2014). Shown are the localizations for Cer(d18:1/16:0) +K (m/z = 576.561, B), Cer(d18:1/18:0) +K (m/z = 604.496, (C) and Cer(d18:1/20:0) +K (m/z = 632.526, E). An overlay image (D) in the fourth panel is included for Cer(d18:1/16:0) (red) and Cer(d18:1/18:0) green, with overlapping regions appearing as yellow. The relative intensity levels of each specie are represented as colour pixels as indicated by the heatmap bars. MALDI-MSI data for specific ganglioside species were acquired in negative ion

117 mode and are given in Supplemental Figure S2. Sections were stained with H&E stain (A) after the MALDI- MSI imaging sequence for spatial determination, and anatomic regions of each brain are indicated as follows 1- cerebral cortex (ctx); 2- corpus callosum (cc); 3 -septal nuclei (sn); 4 - dorsal fornix (df) + ventral hippocampal commisure (vhc); 5 - hippocampus (hipp); 6- anterior commissure (ac); 7- stria medullaris thalami (sm); 8 - paraventricular/medial thalamic nuclei (p/mtn); 9 - fascicullus retroflexus (fr); 10- superior cerebellar peduncle (scp); 11 -cerebellar WM; I-X – cerebellar lobules. H&E images are identical to H&E images in Supplemental Figure S2. Cer 18:1/16:0 corresponds to Cer(d18:1/16:0), Cer(d18:1/18:0) to Cer 18:1/18:0, and Cer(d18:1/20:0) to Cer 18:1/20:0. The width of the heatmap bars corresponds to 5 mm. Representative images are shown for a group of brains tested (3 per genotype).

The most abundant ganglioside and sulfatide species were also measured in negative ion mode in the same sagittal sections (Figure 28). Shown are the GM1, GM2, GM3 and sulfatide versions with Cer 18:1/18:0. In Hom Farber brain, the GM1 was elevated in the cerebral cortex compared to WT, and GM2 and GM3 were elevated in the cerebellar granule cells layer, septal nucleus, and cerebral cortex. Sulfatide was elevated in the cerebellar and cerebral (corpus callosum) white matter and the midbrain (except the septal nuclei) (Figure

28). Interestingly, the GM2 and GM3 Cer 18:1/18:0 co-localized with the accumulated Cer

18:1/18:0 in the Hom Farber brains (Figure 27).

118

Figure 28. MALDI-MSI analysis reveals selective distribution of sulfatide and gangliosides by specific brain region in Hom Farber (Hom) mice. (A) Sagittal tissue sections of Hom Farber and WT brains were analyzed by MALDI-MSI in negative ion mode (B-E) as previously described (Jones et al., 2014). Shown are the localizations for Sulfatide(d18:1/18:0) (m/z =

119

806.786, B), GM1(d18:1/18:0) (m/z = 1545.775, (C), GM2(d18:1/18:0) (m/z = 1383.502, D), and GM3(d18:1/18:0) (m/z = 1181.259, E). The relative intensity levels of each specie are represented as colour pixels as indicated by the heatmap bars. Sections were stained with H&E stain (top panel) after the MALDI- MSI imaging sequence. Anatomic regions of each brain are indicated as follows: 1- cerebral cortex (ctx); 2- corpus callosum (cc); 3 -septal nuclei (sn); 4 - dorsal fornix (df) + ventral hippocampal commisure (vhc); 5 - hippocampus (hipp); 6- anterior commissure (ac); 7- stria medullaris thalami (sm); 8 -paraventricular/medial thalamic nuclei (p/mtn); 9 - fascicullus retroflexus (fr); 10- superior cerebellar peduncle (scp); 11 -cerebellar WM; I-X – cerebellar lobules. H&E images are identical to H&E images in Figure 3. Sulfatide 18:1/18:0 corresponds to Sulfatide(d18:1/18:0), GM1 18:1/18:0 corresponds to GM1(d18:1/18:0), GM2 18:1/18:0 corresponds to GM2(d18:1/18:0), and GM3 18:1/18:0 corresponds to GM3(d18:1/18:0).

4.3.4 Hom Farber mice develop hydrocephalus, extensive tissue

pathology, and subcellular storage changes in a broad range of

CNS cell types

4.3.4.1 Gross brain pathology

Hydrocephalus was previously detected in approximately 70% of 10-week-old Hom

Farber mice by MRI (Alayoubi et al., 2013). In agreement with the latter study and to further explore its initiation and temporal progress of this abnormality, in the current study we performed MRI on fixed brains of Hom Farber mice and WT mice aged 5, 7, and 9 weeks.

Hydrocephalus that affected all ventricles (lateral, 3rd and 4th) was evident in a fraction of

Hom Farber mice already at 7 weeks (Figure 29A; Figure 30A).

The additional fluid in the brains of Hom Farber mice contributed to an increase in brain weight relative to total body weight (Figure 29B). The absolute weight of Hom Farber brains was similar to that of WT and Het mice at ages 5, 7, and 9 weeks (Figure 30B), yet

Farber mice weight less than WT and Het mice (Alayoubi et al., 2013; Dworski et al., 2015).

120

Therefore, the similarly-sized brain constitutes double the percent of total body weight of

Hom Farber mice.

Because of the fragility of the hydrocephalic brains, we primarily focused further neurohistological and ultrastructural studies on non-hydrocephalic Hom Farber brains.

Figure 29. Hydrocephalus affects all ventricles in the brains of Asah1P361R/P361R (Hom) mice. (A) Hydrocephalus of all ventricles (lateral – white arrows, 3rd – grey arrows and 4th – black arrows) was first detected in Hom mice by MRI (axial projection) at 7 weeks of age. (B) The whole body and brains of 5-,7-, and 9-week-old mice were weighed. Hom brains represent a greater percent of the body, as assessed by a two-way ANOVA with Bonferroni post-tests. WT n=4-7, Het n=2-5, Hom n=5-6. Means of WT, Het and Hom brain/body weights are indicated by solid, dotted and dashed lines, respectively. Comparison to coronal brain MRI projections is provided by Figure 30A and C. Scale bar: 5 mm.

121

Figure 30. The brain weight in Asah1P361R/P361R (Hom) mice is not different from Het and WT mice. (A) Hydrocephaly of the 3rd (grey arrows) and lateral (white arrows) ventricles was detectable in Hom mice also by coronal MRI already at 7 weeks of age. (B) The weight of Hom brains is similar to WT and Het brains. The brains of 5-, 7-, and 9-week-old mice were measured. Hom brains were not significantly different, as assessed by a two-way ANOVA with Bonferroni post-tests. WT n=4-7, Het n=2-5, Hom n=5-6. Means of WT, Het and Hom brain weights are indicated by solid, dotted and dashed lines, respectively. (C) IF detection of NeuN+ neurons, CD68+ microglia/macrophages and GFAP+ astrocytes in a brain hemisphere of a 9 weeks-old Hom animal that is affected by hydrocephaly. Note that the CD68+ pathology affects the same cerebral regions as in the animals not affected by the hydrocephaly (Figure 32). Enlarged lateral ventricle is highlighted by the white arrow. Also note the substantial thinning of the overlying cerebral cortex. Compare to findings in animals not affected by hydrocephaly (Figure 32). Scale bar: 5 mm in (A), 1 mm in (C).

122

4.3.4.2 Cerebral and cerebellar (immuno)histopathology

We analyzed cerebral and cerebellar histopathology in HE, PAS, LFB, and CathD

IHC stained sections from 3-, 5-, 7- and 9-week-old Hom Farber mice and WT mice. Given the limited comparative data on the spatial distribution and cellular and tissue nature of the human FD CNS pathology (Molz 1968; Mosher et al., 1969), we decided to focus our studies at cerebral and cerebellar areas previously reported as preferentially affected in murine models of other LSDs (Macauley et al., 2008; Praggastis et al., 2015; Sarna et al., 2003;

Partanen et al., 2008; Pressey et al., 2012). The cerebral volume that we analyzed on coronal sections is highlighted by red arrows in Figure 27 and contains the primary somatosensory cortex, hippocampus, VPM, VPL and posterior group of thalamic nuclei, and subcortical white matter (internal capsule included). In cerebellum, we tested the tissue volume that was laterally limited by the extent of the lobule X.

We found no profound neuronal migration abnormalities in sections from Hom

Farber mice. We could identify evidence of lysosomal abnormalities by CathD IHC in specific cerebral and cerebellar neuronal populations in Hom Farber mouse brains at 3 weeks despite major neuronal cytoplasmic vacuolization not being discernible globally even at 9 weeks (Figure 31). As a dominant feature, we found numerous abnormal and variably PAS+ cells with expanded foamy cytoplasm both in cerebra and cerebella of Hom Farber mice. At

3 weeks of age these cells already appeared in the subcortical WM. The regions specifically affected were the external capsule (EC) and internal capsule/hippocampal fimbria regions. At later time points (>5 weeks), these cells started forming granuloma-like accumulations and frequently organized into perivascular cuffs.

123

Other areas populated by these cells were the meninges and to a lesser extent the stroma of choroid plexi of all cerebral ventricles. In contrast to what was observed in WM, in specific grey matter (GM) areas these cells accumulated into discrete and sparsely distributed perineuronal clusters.

124

Figure 31. Neurons in Asah1P361R/P361R (Hom) mice contain distended and coarsely granular lysosomes but do not undergo ballooning transformation. Brown granular IHC signal specifically marks Cathepsin D (lysosomal luminal marker). (A) Neurons of the hippocampal CA1 region. White arrows highlight distended lysosomes or accumulations of lysosomes (compare to Figure 34). (B) Abnormal lysosomal patterns are also present in Purkinje cells (PC) of Hom mice. Black arrows highlight individual PCs in Hom and WT mice. Images were taken from the anterior cerebellar zone (lobules I-V) (compare to Figure 35 and Figure 36). Scale bars: 50 m in (A) and (B). .

125

To delineate the cell type-specific pathologies, we co-stained sections from Hom

Farber and WT mouse brains with anti-neuronal nuclei (NeuN), anti-CD68, and anti-glial fibrillary acidic protein (GFAP) primary Abs to highlight neurons, microglia/macrophages, and astrocytes, respectively (Figure 32). Similar to the observations in the histological stains,

NeuN staining suggested that the gross overall neuronal density and organization of the cortical and subcortical GM was not substantially affected at any age tested.

Abnormal CD68+ microglia/macrophages, which corresponded to the abnormal and variably PAS+ cells both in the WM and GM, represented the major pathology. Contrary to the coalescing granuloma-like accumulations in WM (Figure 32; Figure 33A,B), CD68+ cells formed well-discernible neuronophagic clusters in specific GM regions. Within the analyzed cerebral volume, the GM areas affected by the latter pathology were the CA1 region of hippocampus (Figure 32; Figure 33C), layer II of the primary somatosensory (S1) cortex

(posterior group of thalamic nuclei as well as the thalamic VPM nucleus (data not shown).

Similar to the CD68+ microglial/macrophage pathology, GFAP+ gliosis was initiated in the

WM and only later appeared in the GM (Figure 32). A comparison to abnormalities in a brain of a 9-week-old animal affected by hydrocephaly (GFAP+ abnormality is particularly accentuated), is provided in Figure 30C.

126

Figure 32. Temporal and spatial progress of the cell type-specific pathologies in the cerebrum of Asah1P361R/P361R (Hom) mice. Cerebral pathology in 3-, 5- and 9-week-old Hom mice is compared to 9-week-old WT animals. Overall organization of the GM (highlighted by NeuN+) is not substantially impacted in Hom mice. CD68+ microglia/macrophage (white arrows) and GFAP+ astrocytic abnormalities are detectable already at 3 weeks of age in the external capsule (subcortical WM), internal capsule and hippocampal fimbria, and are progressive with age in Hom animals. Likely discrete neurodegeneration in specific GM areas (thalamus (grey arrows), CA1 region of the hippocampus (grey hollow arrowheads) and layer II of the primary somatosensory (S1) cortex (grey hollow arrows)) is reflected by neuronophagic clustering of CD68+ microglia/macrophages. This pathology first appears at 5 weeks and progresses until 9 weeks of age. Pathology of the external capsule (EC – white arrows) is shown in detail in Figure 33A and B. Hippocampal CA1 (grey hollow arrowheads) and layer II of S1 cortex (grey hollow arrows) changes are shown in detail in the C and D, respectively. White arrowhead highlights the approximate position of the sagittal section planes shown as MSI data in Figure 27. Scale bar: 1 mm.

127

Figure 33. Microglia/macrophage pathology in the cerebral white and grey matter of Asah1P361R/P361R (Hom) mice. (A) Luxol Fast Blue (LFB) staining outlines the external capsule (EC, subcortical WM) which contains, in 9- week-old Hom animals, abnormally clustered cells (black arrows) that correspond to CD68+ microglia/macrophages (white arrows) in (B). (C) In CA1 hippocampal region neurodegeneration can be implied by localized microglial (CD68+ cells in red by co-IF) neuronophagic clusters (white arrows) in 9-week- old animals. Ultrastructural findings highlight individual degenerated neurons with karyorrhectic (blebbed) nuclei (white arrows) and cytoplasmic storage changes (black arrows) in CA1 hippocampal neurons at 9 weeks. (D) Pathology similar to changes observed in CA1 can be identified also in the cortical layer II (here shown in the cortical somatosensory (S1) area). Inset (corresponds to the area highlighted by the white arrow) in the 9-

128 week-old Hom animal shows a cluster of CD68+ (red) cells phagocytosing a NeuN+ neuron (green). For neuroanatomic localization of areas shown in (A-D) and also findings in 9-week-old WT animals compare to Figure 32. Dotted lines in (A) and (B) highlight the external capsule (EC, subcortical WM). ctx. VI - cortical layer VI; EC - external capsule; CA1 - CA1 region of hippocampus; ctx. II - cortical layer II. IF confocal image stacks (13 m thick in z) shown in (B-D) are presented as maximum intensity projections. Scale bars: 100 m in (A), 50 m in (B), 50 m (confocal) and 5 m (EM) in (C), 50 m in (D).

We also performed LFB staining to highlight any major abnormality in WM formation and/or myelination. At 3 weeks of age, the cerebral subcortical WM appeared thinner in Hom Farber mice when compared to WT littermates. In 5- to 9-week-old Hom

Farber animals, the abnormal build-up of CD68+ microglia/macrophage granuloma-like accumulations was the dominant WM pathology (Figure 32; Figure 33A,B). Anti-NF-M antibody and IF staining was used to examine axonal bundles in the WM but this did not reveal significant neuroaxonal dystrophy in Hom Farber mice (data not shown).

Cerebellar pathology was, in general, similar (Figure 34A) to the abnormalities observed in the cerebra of Hom Farber mice. However, even in 9-week-old animals, the density of calbindin+ Purkinje cells (PCs) was comparable in all the cerebellar zones/lobules between Hom Farber mice and WT mice. Correspondingly the thickness of the molecular layer was also not substantially reduced in any of the cerebellar zones/lobules. Furthermore, we did not detect either any notable abnormalities of calbindin+ or NF-M+/calbindin- axons

(data not shown) or substantial morphological alterations of PC dendritic arbors by IF

(Figure 31B; Figure 34B).

As in the cerebrum, the abnormal CD68+ microglial/macrophages were the hallmark of the cerebellar pathology in Hom Farber mice. At 3 weeks of age, we found foamy CD68+ cells in the cerebellar WM (data not shown). Their distribution, again, was closely spatially

129 linked to brain vessels. Importantly, these CD68+ cells did not preferentially populate the molecular layer of the anterior and/or posterior cerebellar cortical zones or the deep cerebellar nuclei (DCN). However, at later ages, we found granuloma-like clustering of

CD68+ cells within the cerebellar GM (PC layer included) and WM, subependymal zone of the 4th ventricle, and in cerebellar meninges. Interestingly, the GFAP signal originating from astrocytes of the PC layer (Bergmann glia) was stronger in Hom Farber mice than in WT mice even in the absence of extensive and patterned PC degeneration in Hom Farber animals.

130

Figure 34. Cerebellar pathology in Asah1P361R/P361R (Hom) mice. (A) Comparison of 9-week-old Hom and WT cerebella does not suggest substantial and/or patterned degeneration of Purkinje cells (PCs). In Hom mice lobules I-V and VIII, that are first affected by PC degeneration in murine models of several other lysosomal storage conditions (Macauley et al., 2008; Praggastis et al., 2015; Sarna et al., 2003), do not systematically loose PCs even at 9 weeks of age. The molecular layer (usually densely populated by reactive microglia when patterned PC degeneration develops) is devoid of changes indicative of clearance of PC dendritic arbors. In contrast, CD68+ microglia/macrophages are distributed both in GM as well as in WM cerebellar areas. Local clustering of CD68+ cells can be observed within the PC layer, WM and meninges (white arrows). Abnormal GFAP expression in the PC layer of Hom mice is compared with the normal patterns in WT mice. (B) Details (correspond to areas outlined by white rectangles in (A)) of calbindin (PCs), CD68 (microglia/macrophages) and GFAP (astrocytes) immunofluorescent stains in the cerebellar lobule I. White arrows highlight CD68+ microglia/macrophages. PC – Purkinje cells layer; ml – molecular layer. Scale bar: 1 mm in (A).

131

4.3.4.3 Ultrastructural pathology

We performed ultrastructural analyses on the brains of 3- and 9-week-old Hom Farber mice and WT mice to evaluate the character and progression of the subcellular storage pathology. Storage changes could be detected in a broad range of neuronal and non-neuronal cell types in brains from Hom Farber mice as early as 3 weeks of age; these were not found in the brains of their WT littermates. All storage patterns (CTBs, banana/spindle/needle-like bodies, and zebra-like bodies) previously linked to FD were identified in Hom Farber mice.

That said, the appearance of the storage bodies was variable between specific cell types.

Likewise, individual cells often expressed a combination of several morphological storage patterns.

The abnormal microglia/macrophages in the brain presented with a dominant storage pathology. This was comparable to observations made previously concerning liver and peripheral nerves of Hom Farber mice (Alayoubi et al., 2013). The confluent storage bodies in these cells had a finely granular matrix (Figure 35A,C,E) with occasionally embedded

CTBs (Figure 36A). Similar to findings under light microscopy, granuloma-like accumulations of storage-laden microglia/macrophages were frequently (but not exclusively) observed in WM. These developed in close vicinity to brain vessels (Figure 35C,E).

Moreover, a fraction of the cells contained within these abnormal clusters likely degenerated.

As a result, we identified foci of accumulated cellular debris with embedded storage material

(Figure 36B) in older animals.

Unlike in 3-week-old Hom Farber mice (Figure 35B), in WM of 9-week-old Hom

Farber animals, we found zebra-like and/or cleared banana-like bodies in the cytoplasm of cells with oligodendroglial morphology (Figure 35D). Zebra-like bodies could also be found

132 in WM within the processes of cells with electron-dense cytoplasm, yet again, suggesting their oligodendroglial nature. However, in some instances it could not be clearly discerned whether or not these processes were astrocytic. Storage changes combining CTB-like and banana-like profiles were further identified in endothelial cells (Figure 35E) of brain vessels.

The endothelial storage, on a number of occasions, resulted in the formation of major intraluminal endothelial protrusions. To a lesser extent, storage bodies could also be detected in smooth muscle cells of the muscularis layer of larger-sized vessels (data not shown).

133

Figure 35. White matter and vascular ultrastructural storage pathology in Asah1P361R/P361R (Hom) mice. (A) EM demonstrates storage changes in microglia/macrophages in subcortical WM (EC) already at 3 weeks of age (black arrows). (C) At 9 weeks of age the storage bodies (black arrows) in microglial/macrophage-like cells are confluent. These abnormal cells (nuclei highlighted by white asterisks) form granulomatous clusters (compare to Figure 32 and Figure 33A and B), many of which are located close to vessels (the vascular lumen is highlighted by the black asterisk). (B) Cells with oligodendroglial morphology in Hom animals are largely devoid of storage vacuoles at 3 weeks. (D) Cytoplasmic zebra-like storage changes (black arrow) can be identified in WM cells with oligodendroglial ultrastructural characteristics in 9-week-old Hom animals. (E) Storage in endothelial (black arrows) and in perivascular macrophages (small white arrow) is present in Hom brain vessels (cerebellar WM vessels are shown) already at 3 weeks of age. At 9 weeks, endothelial (black arrows), pericytic (large white arrows) and perivascular macrophage (small white arrow) storage pathology is extensively developed. Scale bars: 1 m (A-E).

134

Figure 36. CTB-like storage material in Asah1P361R/P361R (Hom) mice. (A) Detail of the CTB-like morphology of the confluent storage bodies in perivascular macrophages. Farber bodies were found relatively occasionally (one such is highlighted by the black arrow in the inset). (B) Large area of the cerebellar molecular layer is occupied by CTB-like storage material (black arrows) and cellular debris. Asterisk highlights a nucleus of a likely disintegrating microglial/macrophage cell. Similar changes were also found in cerebral internal capsule and/or hippocampal fimbria. Scale bars: 1m in (A) and (B).

Contrary to the abnormal microglia/macrophages described earlier, the predominant finding in neurons were zebra-body-like profiles (Figure 37A) that varied from finely striated

(e.g. PCs; Figure 37B) to nearly “cleared” vesicles (e.g. see granule cell layer neurons;

Figure 37C) that did not form large-sized confluent structures. Similar to histopathology and

CathD IHC (Figure 31), EM did not reveal any major and/or widely distributed changes in overall neuronal cytoarchitecture (e.g. cytoplasmic ballooning). In some brain areas where IF analyses suggested neuronophagy (Figure 32), apoptotic-like changes (karyorrhexis) of neuronal nuclei could be identified (Figure 34B) in 9-week-old mice. We further searched for cytoplasmic non-membrane bound protein-aggregate-like structures, however, we identified only a minimum of such changes in neurons of Hom Farber mice.

135

Figure 37. Neuronal ultrastructural storage pathology in Asah1P361R/P361R (Hom) mice. (A) Neuronal storage changes (shown in neurons of various regions) are detectable already at 3 weeks of age. Storage bodies (highlighted by black arrows and shown enlarged in the inserts) often present as zebra-like structures. (B) Progressive storage changes (white arrows) are shown in PCs from anterior cerebellar zones (lobules I-V). Zebra-like storage bodies have variable appearance and only occasionally form larger confluent structures. (C) Storage bodies (white arrows) in granule cell layer (GCL) have “cleared” appearance. ctx. II – cortical layer II; CA1 - CA1 region of hippocampus. Scale bars: 1 m in (A-C).

136

Lastly, excess of CathD by IHC was found in choroid plexus cells already at 3 weeks of age (Figure 38A). Well-discernible and numerous CTBs and/or zebra-like storage profiles could be identified in choroid plexus cells in 9-week-old Hom Farber mice (Figure 38B). In contrast, ependymal cells seemed largely devoid of storage bodies or were affected only minimally. Similar to the histopathology and IF observations, we found local accumulation of storage laden microglial/macrophages in the subependymal zones (Figure 38C).

137

Figure 38. Choroid plexus and ependymal storage pathology in Asah1P361R/P361R (Hom) mice. (A) Accumulation of CathD+ (lysosomal luminal marker) vacuoles in choroid plexus (lateral ventricles are shown) cells is initiated already in 3-week-old Hom mice and progresses with age. (B) Choroid plexus cells (4th ventricle) in Hom animals contain CTB/zebra-like storage bodies (white arrows) by EM. (C) Ependymal cells (4th ventricle, nuclei marked by black asterisks) seem to be relatively spared of storage changes, nonetheless, subependymal zone contains microglia/macrophages (nuclei marked by white asterisks) filled with confluent storage bodies (white arrows). Scale bars: 20 m in (A), 1m in (B) and (C).

138

4.4 Discussion/conclusions

Farber disease is an ultra-rare and severe multi-systemic disorder. Given the small number of patients documented to date, the opportunity to study the clinical, biochemical, and cellular/tissue CNS-associated pathologies in FD has been limited. However, several reports suggest neuronal lysosomal storage changes in individuals affected by FD (Levade et al., 2009; Molz 1968; Moser et al., 1986). In contrast, data on lysosomal pathology involving non-neuronal CNS cell types is scant in FD. Individual brain MRI studies recently suggested loss of deep WM, hydrocephalus, and brain atrophy (Chedwari et al., 2012). Likewise,

Cappellari and colleagues documented increased size of peri-cerebellar sulci and the 4th ventricle in an HSCT-treated FD patient (Cappellari et al., 2016).

We recently generated the first viable animal model of FD (Alayoubi et al., 2013) and in this study we characterize the hallmark CNS deficits associated with ACDase deficiency.

In addition to describing the key features of the brain disease, we also sought to identify read-outs that could be used to help gauge the impact of novel treatment schemes for FD.

As noted earlier, Asah1P361R/P361R (Hom Farber) mice present with a progressive and lethal phenotype. Macrophages are affected by ACDase deficiency and are thought to be key contributors to the progression of a wide range of tissue pathologies observed in Hom Farber mice. Indeed, there is an increase in circulating monocytes and abnormal macrophages are numerous in the skin and liver. Disease-specific macrophages were previously also found in the spinal cord disrupting neuronal tracts and in sciatic nerves pushing on axons and impacting myelin sheaths (Alayoubi et al., 2013). These cells also infiltrate BM, thymus, spleen, and LN

(Alayoubi et al., 2013) and contribute to the overall perturbed hematopoiesis we previously described in Hom Farber mice (Dworski et al., 2015).

139

4.4.1 Contribution of brain pathology to behavioural phenotype

Given their severely impaired activity and motor coordination, a limited number of behavioural tests were chosen for Hom Farber mice. Similar limitations apply to testing cognitive abilities of these mice. Furthermore, the behavioural phenotype that we describe in this study in

Hom Farber mice should be interpreted within the context of their short lifespan and the progressive nature of their peripheral organ pathology. As a result, it is relatively difficult to establish clear neuroanatomical/neuropathological correlates for individual behavioural abnormalities observed in these mice. As a dominant feature, Hom Farber mice become increasingly weaker over time compared to WT mice. It is likely that the weak grip strength and compromised locomotion and motor coordination observed in Hom Farber mice cannot be attributed solely to the pathologies in the cerebral and cerebellar motor control centers. Aside from the systemic toxic effects of the visceral disease, additional contributors to the muscle weakness could be (still unidentified) degeneration or dysfunction of spinal motor neurons

(noticed by Moser et al. in the human FD (Moser et al., 1969)), pathology of peripheral motor nerves and theoretically also improper communication at the neuromuscular junction and/or myopathy.

4.4.2 The effects of storage on various cell types

While behavioural studies do not pinpoint specific regions of the brain contributing to the clinical phenotype, our biochemical and morphological studies identify a broad range of abnormalities at the subcellular, cellular, and organ levels. At the subcellular level, a plethora of neuronal and non-neuronal cell types develop storage changes compatible with those

140 previously identified in non-neuronal tissues of FD patients (Abenoza & Sibley, 1987; Burck et al., 1985; Zappatini-Tommasi et al., 1992) as well as Hom Farber mice.

Regardless of the complexity, the cerebral pathology in Hom Farber mice can be categorized into two major areas based on the cell type affected – CD68+ microglia/macrophages and neurons. Similar to that observed in non-neuronal tissues, the

CNS in the Farber mice is extensively affected by the granuloma-like accumulation of the abnormal CD68+ cells. In cerebrum, this abnormality is initiated and progresses in white matter areas, often associating with cerebral vessels as perivascular cuffs. At later stages of disease, it is likely that a fraction of these clustered abnormal cells degenerate/disintegrate and release their content into the extracellular space. Given this vessel-linked distribution pattern in the white matter, it is possible that these abnormal cells do not represent brain endogenous pool of microglia/macrophages, but rather originate from the periphery in a manner similar to other tissues in Hom Farber mice.

Various cerebral neuronal populations are also affected by the subcellular storage changes. Interestingly, the overall morphological alteration of neurons is not as pronounced as in the CD68+ macrophages/microglia. However, specific neuronal populations seem to be more sensitive to the primary enzymatic defect than others and degenerate as a result. This localized degeneration is associated with clearance by microglia that differ morphologically from the granuloma-like accumulations of foamy CD68+ cells in WM areas. Interestingly, some of the neuroanatomic areas likely affected by neuronal degeneration (e.g. specific thalamic nuclei) were previously reported as similarly compromised in mouse models of other LSDs (Pressey et al., 2012; Chedrawi et al., 2012). In future studies it will be important to identify whether particular functionally specialized neuronal populations (e.g. GABAergic

141 interneurons (Walkley et al., 1991)), previously shown to be preferentially compromised in

LSDs, are also affected in Hom Farber mice.

We also found ultrastructural storage changes also in the WM cells with oligodendroglial (and potentially also astrocytic) morphology, however, the extent of the impact these changes have on myelin build-up and overall WM integrity requires further detailed studies.

The dominant neuronal findings in the cerebellum of Hom Farber mice are the accumulation of zebra bodies in PCs and excess of the “cleared” storage profiles in granule cell layer neurons. Similar to the cerebrum, the local accumulation of abnormal macrophages could be identified in cerebellar WM and GM. Cerebellar WM oligodendroglia-like cells also presented with ultrastructural storage profiles similar to those found in cerebrum. Critically, however, in Hom Farber mice we did not observe either the patterned degeneration of

Purkinje cells and/or the macrophage/microglial and astrocytic pathologies usually associating with this degeneration (Macauley et al., 2008; Praggastis et al., 2015; Sarna et al., 2003).

Findings in the cerebral vasculature are also an important part of the overall CNS pathology in Hom Farber mice. Endothelial cells contained large amounts of storage material that organized into large-sized confluent storage bodies. By forming luminal protrusions, these bodies on many occasions substantially disfigured the overall morphology of the endothelial cells. Given the size of these protrusions and the global extent of this pathology, it is plausible that the subcellular storage pathology might impact global blood perfusion of the brain and negatively contribute to the neurological phenotype in this disease. As outlined

142 above, the extensive perivascular macrophage/microglial pathology associated with vessels in both the cerebrum and cerebellum could also potentially contribute to impaired perfusion.

As reported before (Alayoubi et al., 2013), hydrocephalus affecting all ventricles was identified in a substantial fraction of Hom Farber mouse brains. The general causes of hydrocephalus include increased production, compromised resorption, or obstructed flow of the cerebrospinal fluid (CSF). Alternatively, the volume of CSF increases with the atrophy of the brain tissue, which is not likely in Hom Farber mice given their lifespan and lack of evidence of numerically large-scale neuronal degeneration. Reviewing abnormalities in cell types participating in CSF homeostasis and circulation in Hom Farber mice, we identified subcellular storage pathology in choroid plexus cells. Furthermore, abnormal CD68+ cells were present in the stroma of choroid plexi. Importantly, the storage abnormalities seemed minimal in ependymal cells. However, local granulomatous accumulations of the abnormal microglia/macrophages could be found in the subependymal periventricular areas and also in the meninges. The latter localization could possibly also include the presence and local negative impacts of these abnormal cells in the arachnoid granulations, which are sites critical for CSF resorption. Alternatively, we could speculate that the subcellular pathology in choroid plexus cells may trigger CSF overproduction. Overall, however, the exact mechanisms causing the communicating hydrocephalus in Hom Farber mice still remain unresolved.

Although comparisons are difficult to establish as the human FD neuropathological studies are sporadic, Moser et al. reported neuronal storage changes, spatially restricted cortical and subcortical neurodegeneration associated with neuronophagy, fibrous gliosis, microglial proliferation and some myelin loss in a 9.5-month-old patient with FD (Moser et

143 al., 1969). Interestingly, cortical layers 2-3 were preferentially affected by neurodegeneration, and neurodegeneration could also be found in the hippocampus of this

FD patient. Further, similar to our findings in Hom Farber mice, a normal complement of

Purkinje cells some of which appeared distended was reported by Moser et al (Moser et al.,

1969). Interestingly, the latter FD case report mentions microglial/macrophage pathology only briefly.

Similarly, little information on the lipid storage in the brain of human FD patients is available. Accumulation of ceramide in the visceral organs as well as cerebral white matter of severely affected FD patients was first reported in 1967 and 1969 (Prensky et al., 1967;

Moser et al., 1969). In the cerebellum, storage of hydroxy-C24:0 and C22:0 ceramides was documented by Sugita and coworkers (Sugita et al., 1974; Sugita et al., 1973). As compared to control individuals, the non-hydroxy fatty acid-containing ceramide content was doubled in the cerebellum of the FD patient, whereas that of the 2-hydroxy fatty acid-containing ceramide was increased by almost 50-fold, representing more than one third of total ceramides, a finding very similar to that reported in Figure 26. Accumulation of a glycolipid in brain was also observed (Moser et al., 1969; Clausen et al., 1970) in FD patients.

4.4.3 Ceramides linked to disease

As a class of bioactive signaling molecules, it is important to know which specific ceramide species are accumulating in FD, as they are likely the initiating agents of the observed downstream histopathological and behavioural changes reported here. Yet current knowledge of the functions of individual sphingolipid species is very limited. Most studies

144 have been centered on the role of ceramides in apoptosis and autophagy, including in neurons

(Tong & de la Monte, 2009). C16:0-Cer is involved at multiple points of the apoptotic signaling pathway (reviewed by Grösch and colleagues (Grosch et al., 2012)), while dhC16:0-Cer is elevated in response to molecules that induce autophagy (Cruickshanks et al.,

2015; Signorelli et al., 2009). Both C16:0-Cer and dhC16:0-Cer had large fold elevations in levels in brains from Hom Farber mice, and some changes indicative of apoptosis were seen in neurons. Increases in C16 ceramides and decreases in C18 ceramides are implicated in cerebellar PC neurodegeneration in the flincher mouse (Zhao et al., 2011), yet in our mouse we did not see significant PC loss, perhaps due to an elevation instead of a reduction of C18.

All such attempts to link particular ceramide species with biological function not only require knowledge as to the cell types in which they accumulate but also at what subcellular location(s). It could be argued that the accumulation of ceramides in FD, given the absence of function of a lysosomal ACDase, would be limited to the lumen of late endosomes and lysosomes. In such a case little impact on cell death pathways might be expected; rather, deficiency of ceramide precursors for various metabolic pathways might be more likely.

Unfortunately, detailed information on subcellular location and dynamics of turnover remains largely unavailable at this time. Thus in the FD mouse model cells may be compensating for the entrapment of ceramide precursors secondary to ACDase deficiency through changes in the synthesis of ceramides (such as by altered CerS) and/or by increased degradation by other ceramidases (e.g. neutral ceramidase).

The results of the sphingoid base analyses support this hypothesis. Sph is the direct product of ACDase activity, so with dysfunctional ACDase there should be less Sph. Instead,

Sph is increased, while sphingoid bases remain unchanged. This suggests that neutral and/or

145 alkaline ceramidases may compensate somewhat for the ACDase deficiency. This compensation appears to be partially effective, as the brain has a relatively low level of ceramide accumulation compared to other organs (the spleen, liver, lung, heart and kidney)

(Alayoubi et al., 2013). Yet there are still ceramide species that are altered, suggesting that neutral and alkaline ceramidases have specificity for particular acyl chains. Interestingly, the ceramide species exhibiting the lowest fold elevation in Hom mice is C18:1 ceramide, a species that is known to be preferentially catabolized by the alkaline ceramidase ACER3 (Hu et al., 2010).

Ceramide changes have been seen in several other neurological diseases. Elevated levels of C16, C18, C20, and C24 ceramides have been reported in the brains of patients with

Alzheimer’s disease and other neurodegenerative diseases (Filippov et al., 2012; Han et al.,

2001). Astrocytes surrounding Alzheimer’s disease plaques mostly have C18 and C24:1 ceramides (Wang et al., 2012). In fact, ceramides have been reported to induce generation of the pathological molecule, Aβ (Puglielli et al., 2003) . In Parkinson’s disease, there is a general ceramide decrease in the anterior cingulate cortex (Abbott et al., 2014) . Decreased

C18 ceramides also lead to progressive myoclonus epilepsy (Vanni et al., 2014). These latter changes are accompanied by alterations in CERS gene expression, which encodes for ceramide synthase. Injection of ceramides into the brains of rat pups alters their behavioural phenotype and lipid composition similar to findings in the Hom Farber mice (de la Monte &

Tong, 2014). Yet as described above, lack of detailed information on where the ceramides are located subcellularly limit insights into a role in pathogenesis.

In addition to the absolute value of the accumulation of ceramides, the relative abundance of ceramide species is also important for normal cell/tissue/organ function. In

146 other neurological disorders such as Alzheimer’s disease the fractional composition of each species is unchanged between normal and diseased brains (Filippov et al., 2012). Here, we saw changes in the fractional composition. In brains from Hom mice, Cer-OH species had the largest fold-increase over WT. This may be attributed to the high abundance of hydroxylated galactosphingolipids in myelin; thus, turnover of these sphingolipids leads to an accumulation of hydroxylated ceramides. The increased levels of MHC-OH relative to MHC would then likely represent hydroxylated galactosylceramide (GalCer). This, however, suggests that GalCer turnover is much greater than that for other sphingolipids, or that sphingolipids in myelin are more rapidly turned over. Elevation of DHC levels has also been seen in other sphingolipidoses.

This increase (with little change in glucosylceramide (GlcCer) and GalCer), as well as that of GM3 ganglioside could be in line with the “traffic jam” described to accompany many lysosomal lipid storage disorders (Chen et al., 1999). We also used MALDI-MSI to identify the specific locations of a subset of ceramides and gangliosides. Further analysis of this sort examining the entire brain may help in connecting the sphingolipid accumulation to the brain regions affected, and link this to behavioural outcomes. For example, the cerebellum plays a role in controlling motor coordination, and C16:0 is elevated in that region. This may be contributing to the motor deficits seen behavioural tests. As the FD model will be used for testing of new genetic therapeutic approaches, the ability to monitor the sites of accumulation following therapeutic intervention will provide crucial tissue specific outcome measures.

Neurological disease represents a critical part of the clinical phenotype of FD.

However, given the lack of patient tissues and previous long-term unavailability of a suitable

147 animal model, the CNS pathology remained largely unexplored in FD. Utilizing our recently developed mouse model, we demonstrate the rapidly progressing and profound biochemical and structural impacts of the ACDase deficiency on a broad array of neuronal and non- neuronal cell types and the brain as a whole. By characterizing these hallmark abnormalities, we have not only demonstrated the crucial physiological roles of the lysosomal ACDase- mediated sphingolipid degradation for CNS integrity, but also outlined the range of target cell types and pathologies for any future therapies of this devastating and currently untreatable neurovisceral disease.

148

Chapter 5

Acid ceramidase deficiency is characterized by a unique

plasma cytokine and ceramide profile that is altered by therapy

This work is adapted from Dworski et al., 2017 with permission from Biochimica et

Biophysica Acta (BBA) – Molecular Basis of Disease.

5.1 Introduction

Acid Ceramidase Deficiency (Farber Disease, FD; Farber Lipogranulomatosis,

OMIM #228000) is an ultra-rare, autosomally recessive inherited LSD. It is caused by mutations in ASAH1, which encodes for the lipid-processing enzyme, acid ceramidase

(ACDase, EC 3.5.1.23). This deficiency results in the accumulation of ceramides and many downstream effects, leading to a multisystemic disorder that is often lethal in childhood.

HSCT is currently the only treatment available for FD, but it is not consistently successful and is associated with severe, even life-threatening, side effects. A better understanding of the biology of FD is needed to develop more effective treatments. As well, ceramides are fundamental sphingolipids involved in many cellular processes. Knowledge of their biology will have far-reaching impact.

One of the most pronounced results of FD is perturbation of the hematopoietic system. Patients and mice with FD have splenomegaly, lymphocytosis, and excess macrophage infiltration into many organs (Levade et al., 2009; Alayoubi et al., 2013).

149

Patients also develop granuloma-like lesions, principally at cartilage sites, that are composed of immune cells including lipid-filled macrophages and neutrophils. We previously described the hematopoietic abnormalities in our mouse model of FD and found that by the end of life

(9 weeks of age) the BM, thymus, spleen, and LN of Farber mice were packed with foamy macrophages; B and T progenitor cells were almost completely absent; and HSPCs were significantly increased in the BM (Dworski et al., 2015). In other LSDs that present with macrophage involvement, such as Gaucher disease and Niemann-Pick disease, chitotriosidase levels in plasma can be elevated (Boot et al., 1995; Ries et al., 2006).

However, chitotriosidase activity has not been evaluated in the plasma of patients with FD.

Sphingolipids also play a role in cell signaling. Normally, ACDase converts ceramide into sphingosine and a free fatty acid. In FD, this enzyme has minimal functionality, resulting in bulk ceramide accumulation in patients and mice (Levade et al., 2009; Alayoubi et al.,

2013). There are dozens of ceramide molecular species that vary by their carbon chain length, degree of unsaturation and hydroxylation, and the presence or absence of phosphates.

The specific ceramide species that accumulate in FD have not been investigated in detail.

Identification of these ceramide species is critical to understand which lipids may be causing the detrimental effects of FD. Additionally, it may be expected that sphingosine levels are reduced in ACDase deficiency, but this has not been confirmed. Along these lines, we previously found that some cytokines were elevated in aged Farber mice (Alayoubi et al.,

2013); here we supplement that study by adding younger mice to visualize the timeline of cytokine changes and extend such analyses to include samples obtained from FD patients. To examine whether this elevated cytokine pattern is unique to FD, we also compare the profile with that obtained from another macrophage-prominent LSD, Gaucher disease. In addition to

150 analyzing cytokines and ceramides, we also measure the activity of chitotriosidase in plasma from patients with FD, which can be elevated up to 1000-fold in plasma from patients with

Gaucher disease (Hollak et al., 1994).

In addition to being poorly understood, FD is likely underdiagnosed or misdiagnosed.

Definite diagnosis of FD in patients can be established by measuring ACDase enzymatic activity and/or sequencing the ASAH1 gene. ACDase activity is determined in leukocytes collected from PB or from skin fibroblasts obtained by a biopsy. Analysis of variations in the

ASAH1 sequence is done by comparisons with known Farber mutations, which are limited by the small number of case reports, and by software that predict the detrimental impact of mutations. Should a patient have a unique mutation, the geneticist may not be able to differentiate between a disease-causing or non-disease-causing ASAH1 polymorphism due to limited historical information.

Another obstacle is misdiagnosis. Joint abnormalities (e.g., contractures, pain) may also occur in JIA, resulting in the misdiagnosis of Farber patients (Kostik et al., 2013;

Torcoletti et al., 2014; Erfan et al., 2015). Several physicians submitting samples to this study (Bruno Maranda, John J. Mitchell, Boris Hugle, Bo Magnusson) have reported moderate responses of certain Farber symptoms (joint disease and inflammation) to treatment with biologic therapies used in treating JIA (TNF-α inhibitors, IL-6 receptor blockers), which indicates an additional facet that may perpetuate a misdiagnosis of JIA in the clinic. Indeed,

36% of case reports of patients with moderate FD were initially misdiagnosed as JIA

(Schuchman, 2014). Here, we identify the cytokines and ceramides that are changed in the plasma of patients with FD. These results can help investigators understand the pathobiology

151 of FD and allow for better differentiation between FD and inflammatory arthropathies such as JIA.

5.2 Materials and Methods

5.2.1 Sample collection

Blood samples were collected in ethylenediaminetetraacetic acid (EDTA)-coated tubes from 5-, 7-, 9- and 11-week-old mice homozygous for the FD mutation Asah1P361R/P361R

(Hom), heterozygous for the mutation (Het), or wild-type (WT) (Alayoubi et al. 2013). Three to four samples were collected for each genotype for 5-, 7-, and 9-week-old mice, and two samples per genotype for 11-week-old mice. Samples were centrifuged at 1377xg for 5 minutes at room temperature, and plasma samples were stored at -80°C until use.

Human samples were collected from 15 patients with FD, 5 patients with FD who underwent HSCT, 5 patients with JIA, and 11 patients with Gaucher disease. The unaffected parents and siblings of these patients were used as controls, for a total of 39 samples. Blood was collected in EDTA-coated tubes, plasma was separated from cells by centrifugation as above, and white blood cells were isolated. Treating physicians also filled in a questionnaire about the patient’s history with FD. Samples and patient information were collected in accordance with a protocol approved by the University Health Network Research Ethics

Board. The data were kept anonymous.

152

5.2.2 Determination of ACDase activity

The in vitro activity of ACDase was measured from patient leukocytes using a modification of the method previously published (He et al., 1999). Briefly, for each assay 3

µl of substrate buffer (0.2 M citrate-phosphate, pH 5 containing 0.2% Igepal CA-630, 0.3 M

NaCl, and 200 µM C12-NBD ceramide (Cayman Chemical)) was mixed with 3 µl of leukocyte cell lysate, vortexed and then incubated at 37ºC for 1 hour. The reaction was stopped by adding 100% ethanol followed by centrifugation at 13 000 x g for 5 min. Thirty- five µl of supernatant was removed and applied onto a UPLC system (Waters Acquity H-

Class) with fluorescent detector for analysis.

5.2.3 Determination of chitotriosidase activity

Chitotriosidase activity was measured in the plasma of Farber patients and parental/sibling controls. Gaucher patient plasma was used as a reference. The chitotriosidase activity of the Gaucher patient plasma was independently determined by Dr. Thierry Levade, and the samples were classified as ‘Gaucher High’ and ‘Gaucher Low.’ Chitotriosidase activity was determined by incubation of 5 µl of plasma with 4-methylumbelliferyl β-D-

N,N′,N′′-triacetylchitotrioside (Sigma) for 1 hour. The reaction was stopped by adding 0.5 M glycine-NaOH at pH 10.5. Fluorescence intensity was measured at 358/448 nm.

5.2.4 Quantitation of lipids by mass spectrometry

Sphingolipid analysis was performed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) by the Lipidomics Shared Resource at the Medical University of

153

South Carolina as described previously (Pettus et al., 2004). The specific sphingolipids measured were ceramides (Cer), dihydroceramides (dhCer), alpha-hydroxylated ceramides

(a-OH-Cer), and phosphorylated ceramides (Cer-1P) of chain lengths C14 to C26.

Sphingosine (Sph), phosphorylated sphingosine (Sph-1P), dihydrosphingosine (dhSph), and phosphorylated dihydrosphingosine (dhSph-1P) were also measured.

5.2.5 Cytokine analysis

Cytokine levels were analyzed from mouse plasma using the Bio-Plex Pro Mouse

Cytokine 23-plex Assay (Bio-Rad) as per the manufacturer’s instructions. Human plasma was analyzed for cytokine levels using the Cytokine Human Magnetic 30-Plex Panel

(Novex). Luminescence was quantified on the Luminex 100 instrument (Luminex). Data where low bead count was observed (<45 beads) was omitted. Where individual data points were below the detection limit of the kit, values were set to half of the detection limit. Where individual data points were above the detection limit, values were set to the top of the detection level. Heatmaps were generated in the statistical software program R using the gplots package (Warnes et al., 2011).

5.2.6 Statistics

Comparisons of cytokine and ceramide levels between Farber mice of different ages and humans with different disease status were performed using Prism 5.0c. Each cytokine was analyzed by a one-way-ANOVA followed by Bonferroni’s Multiple Comparison Test. If

Bartlett’s test for equal variances indicated that the variance was significantly different, then the Kruskal-Wallis test was performed followed by Dunn’s Multiple Comparison test. The

154 significance between cytokines and gender of the patient was assessed using a two-tailed t- test in Microsoft Excel. The R2 value when correlating cytokines with the age at sample collection or ACDase activity, or when correlating one cytokine with another, was calculated using Microsoft Excel.

5.3 Results

5.3.1 Circulating cytokine levels

Plasma was isolated from WT, Het, and Hom Asah1P361R mice aged 5, 7, and 9 weeks, and WT and Hom mice aged 11 weeks, and cytokines were analyzed by a multiplex assay. The selected data points span the start of the observable disease in this mouse model, as the immune system is compromised between 5 and 7 weeks (Dworski et al., 2015), and the end of their life (9-11 weeks). Circulating levels of monocyte chemotactic protein 1

(MCP-1, CCL2), macrophage inflammatory protein1a (MIP-1a, CCL3), and keratinocyte chemoattractant (KC, CXCL1) were elevated in Hom mice compared to WT and Het mice during this time-course (Figure 39). These cytokines were elevated early, but each peaked at a different age: KC at 7 weeks, MIP-1a at 9 weeks, and MCP-1 was highest at the end of life at 11 weeks. Cytokine levels that were not significantly different (though some showed trends towards elevation) were interleukin (IL)-1a, IL-1b, IL-2, IL-4, IL-5, IL-6, IL-10, IL-

12, IL-13, IL-17, basic fibroblast growth factor (bFGF), granulocyte macrophage colony- stimulating factor (GM-CSF), interferon gamma (IFNg), interferon gamma-induced protein

10 (IP-10), monokine induced by gamma interferon (MIG), tumor necrosis factor alpha

(TNFa), and vascular endothelial growth factor (VEGF) (Figure 40).

155

Figure 39. Cytokines elevated in the plasma of Farber mice. Cytokines were measured in plasma samples from WT, Het, and Hom Farber mice aged 5, 7, 9, and 11 weeks using a multiplex assay. MCP-1 (CCL2) (A), MIP-1a (CCL3) (B) and KC (CXCL1) (C) levels are illustrated. n=2-4 for each genotype at each time point. *p<0.05. MCP-1, monocyte chemotactic protein 1, MIP-1a, macrophage inflammatory 1a; KC, keratinocyte chemoattractant.

156

Figure 40. Cytokines from Farber mouse plasma and controls that were not significantly different. Cytokines were measured in plasma samples from WT, Het, and Hom Farber mice aged 5, 7, 9, and 11 weeks using a multiplex assay. *p<0.05. n=2-4 for each genotype at each time point. IL, interleukin; bFGF, basic fibroblast growth factor; GM-CSF, granulocyte macrophage colony-stimulating factor; IFNg, interferon gamma; IP-10, interferon gamma-induced protein 10; MIG, monokine induced by gamma interferon; TNFa, tumor necrosis factor alpha; VEGF, vascular endothelial growth factor.

157

Plasma was also obtained from the following human patient populations: patients with confirmed FD, patients with FD treated by HSCT, and patients with JIA confirmed not to have FD. Cytokine levels in these human samples were analyzed by multiplex assays and compared to parental and sibling controls (when available). MCP-1, IP-10, and IL-6 were all dramatically elevated in samples from Farber patients compared to those obtained from control and JIA patients. Strikingly, levels of these cytokines were normalized in Farber patients that had received HSCT (Figure 41A-C). IL-12 was also elevated significantly in

Farber patients compared to controls, and VEGF was significantly elevated in Farber patients compared to patients with JIA (Figure 41D-E). As observed in mice, the levels of several cytokines were not significantly different between FD and controls. These were IL-1b, IL-

1RA, IL-2, IL-2R, IL-4, IL-5, IL-7, IL-8, IL-10, IL-13, IL-15, IL-17, bFGF, epidermal growth factor (EGF), eotaxin, granulocyte colony-stimulating factor (G-CSF), GM-CSF, hepatocyte growth factor (HGF), interferon alpha (IFNa), IFNg, MIG, MIP-1a, MIP-1b, regulated on activation, normal T cell expressed and secreted (RANTES), and TNFa (Figure

42).

158

Figure 41. Farber patients have a unique plasma cytokine profile that is normalized following HSCT and different from JIA. Cytokines were measured in plasma from controls (n=39), untreated Farber patients (n=13), HSCT-treated Farber patients (n=5), and JIA patients (n=5) using a multiplex assay. MCP-1 (A), IP-10 (B), IL-6 (C), IL-12 (D) and VEGF (E) levels are shown. Unbiased hierarchical clustering was performed on all patients using these five cytokines (F) or MCP-1 (G). A cluster dendrogram is seen at the top (branches), followed by the patient colour key (single row of coloured rectangles) and a heatmap of the relative cytokine levels (green, black, and red rectangles). The relative amount of each cytokine is seen horizontally corresponding to its row, where green represents relatively low expression, black medium expression, and red relatively high expression. *p<0.05. HSCT, hematopoietic stem cell transplanted Farber patients; JIA, Juvenile Idiopathic Arthritis; ECD, Erdheim- Chester disease; MCP-1, monocyte chemotactic protein 1; IP-10, interferon gamma-induced protein 10; IL-6, interleukin 6; IL-12, interleukin 12; VEGF, vascular endothelial growth factor; ACDase, acid ceramidase.

159

Figure 42. Cytokines from Farber patient plasma and controls that were not significantly different. Cytokines were measured in plasma from controls (n=39), untreated Farber patients (n=13), HSCT-treated Farber patients (n=5), and JIA patients (n=5) using a multiplex assay. *p<0.05. IL, interleukin; bFGF, basic fibroblast growth factor; EGF, epidermal growth factor; G-CSF, granulocyte colony-stimulating factor; GM-

160

CSF, granulocyte macrophage colony-stimulating factor; HGF, hepatocyte growth factor; IFNA, interferon alpha; IFNg, interferon gamma; MIG, monokine induced by gamma; MIP-1a, macrophage inflammatory protein 1a; MIP-1b, macrophage inflammatory protein 1b; RANTES, regulated on activation, normal T cell expressed and secreted; TNFa, tumor necrosis factor alpha.

To examine whether the elevated cytokine profile we saw in plasma from patients with FD was specific to this disorder and not a general pattern seen in LSDs with inflammatory manifestations, we compared our results to those from samples obtained from patients with Gaucher disease. Gaucher disease also has macrophage involvement, and has been reported to manifest with elevated inflammatory markers (Pandey et al., 2013; Tantawy et al., 2015). The cytokine elevation was much more pronounced in FD than in Gaucher disease for MCP-1, IP-10, and IL-6. IL-12 was much more elevated in Gaucher disease than in FD (Figure 43).

161

Figure 43. The elevation of cytokines in Farber disease is different than in Gaucher disease. Select control (n=10) and Farber patient (n=8) plasma samples were compared to Gaucher patient plasma (n=11) using a multiplex assay. *p<0.05 by one-way ANOVA. A two-tailed t-test between control and Farber patient plasma was also significant. MCP-1, monocyte chemotactic protein 1; IP-10, interferon gamma-induced protein 10; IL, interleukin.

The significantly altered cytokines were informative in classifying the samples. When unbiased hierarchical clustering was performed utilizing the top five cytokines in terms of expression increases, 10 of 13 Farber patients clustered together, away from controls, HSCT, and JIA samples (Figure 41F). Two of the three patients who did not cluster with the Farber patients had an attenuated Farber phenotype: relatively mild symptoms and little impairment of daily functioning at over 20 years old, a rare age for patients with FD to reach based on reports published in the literature so far. Importantly, MCP-1 levels generated the same clustering pattern alone (Figure 41G). Using all of the cytokines together did not result in better clustering, nor did using any of the other top five cytokines individually (Figure 44).

Interestingly, plasma from the single Erdheim-Chester disease patient analyzed, a disease

162 with a different clinical phenotype but similar histopathologic changes (Diamond et al.,

2014), clustered near the Farber patients.

Figure 44. Cytokine groupings that do not result in productive clustering of Farber patients. Unbiased hierarchical clustering was performed on all patients using all of the cytokines analyzed (A), or IP-10 (B), IL-6 (C), IL-12 (D) and VEGF (E) alone. A cluster dendrogram is seen at the top, followed by the patient key and a heatmap of the relative cytokine levels.

163

To better understand how these cytokines could play a role in FD, we identified patterns through correlation analysis. Cytokine pairs did not correlate with each other (Figure

45). We also investigated the relationship between the ACDase activity measured in Farber patient leukocytes or the age of patients at sample collection with these cytokines. Only

MCP-1 strongly negatively correlated with ACDase activity (Figure 46F). No cytokine levels correlated with the age at sample collection or the sex of the patient, nor did any additional cytokines correlate with ACDase activity (Figure 46, Table 6).

164

Figure 45. Cytokines with weak or no correlations to each other.

165

Figure 46. Cytokines with weak or no correlations to the age at sample collection or ACDase activity. ACDase, acid ceramidase.

166

Cytokine Fold male/female p-value MCP-1 0.83 0.38 IP-10 8.12 0.19 IL-6 0.90 0.41 IL-12 0.67 0.07 VEGF 0.56 0.14

*The p-value was calculated using a two-tailed t-test.

Table 6. The level of each cytokine was compared between male and female patients.

5.3.2 Plasma ceramide levels

ACDase degrades ceramides into sphingosine and a free fatty acid. In FD, bulk ceramides accumulate due to deficient ACDase activity. To identify which of the dozens of ceramide species are altered in plasma from FD patients and mice, mass spectrometry was used. Ceramides were analyzed in the plasma of 5- and 9-week-old WT, Het, and Hom

Farber mice, and human Farber and JIA patients, controls, and Farber patients treated with

HSCT. Only C16-Ceramide (C16-Cer) and dhC16-Cer were significantly elevated in Hom

Farber mouse plasma (Figure 47), and in Farber patient plasmas only alpha-hydroxy-C18-

Cer, dhC12-Cer, dhC24:1-Cer, and C22:1-Cer-1-phosphate (C22:1-Cer-1P) were significantly elevated compared to controls (Figure 48A-D). Of those ceramides elevated in

Farber patient plasma, only alpha-hydroxy-C18-Cer was normalized following HSCT. In samples from Farber patients that had received HSCT, C20:1-Cer was slightly elevated compared to samples from untreated Farber patients (Figure 48E), and dhC18-Cer was higher than seen in control samples or those from JIA patients (Figure 48F). None of these ceramides could be used to group Farber patients using unbiased hierarchical clustering

(Figure 48G; Figure 54). Most were not significantly different (Figure 49; Figure 51; Figure

167

52). Quite surprisingly, the levels of sphingosine and its derivatives were also unchanged

(Figure 50; Figure 53).

Figure 47. Ceramides with a chain length of 16 carbons are elevated in the plasma of Farber mice. Plasma was collected from 5- and 9-week-old WT, Het, and Hom Farber mice. Ceramides were analyzed by mass spectrometry. C16-Cer (A) and dhC16-Cer (B) levels are shown. *p<0.05. n=3-4 for each genotype at each time point. Cer, ceramide.

168

Figure 48. Ceramide levels vary in plasma from patients with different diseases. Plasma was collected from controls (n=6-27), untreated Farber patients (n=4-8), HSCT-treated Farber patients (n=1-5), and JIA patients (n=5). Ceramides were analyzed by mass spectrometry. a-OH-C18-Cer (A), dhC12- Cer (B), dhC24:1-Cer (C), C22:1-Cer-1P (D), C20:1-Cer (E) and dhC18-Cer (F) levels are illustrated. *p<0.05. Unbiased hierarchical clustering was performed using four significantly altered ceramides (G). A cluster dendrogram is seen at the top (branches), followed by the patient colour key (single row of coloured rectangles) and a heatmap of the relative ceramide levels (green, black, and red rectangles). The relative amount of each

169 ceramide is seen horizontally corresponding to its row, where green represents relatively low expression, black medium expression, and red relatively high expression. There was a very strong negative correlation between the expression level of C20:1-Cer and dhC18-Cer (H). dhC12-Cer had a strong and very strong positive correlation, respectively, with dhC24:1-Cer (I) and aOH-C18-Cer (J). The age at sample collection correlated very strongly negatively with dhC24:1-Cer (K) and aOH-C18-Cer (L). Cer, ceramide; HSCT, hematopoietic stem cell transplanted Farber patients; JIA, Juvenile Idiopathic Arthritis.

170

Figure 49. Ceramides that were not significantly different in Farber mouse plasma versus controls. Plasma was collected from 5- and 9-week-old WT, Het, and Hom Farber mice.

171

Figure 50. Sphingosine levels were not significantly different in Farber mouse plasma versus controls. Plasma was collected from 5- and 9-week-old WT, Het, and Hom Farber mice.

172

Figure 51. Ceramides and phosphorylated ceramides that were not significantly different in Farber patient plasma versus controls. Plasma was collected from controls (A-L n=27; M-W n=6), untreated Farber patients (A-L n=8; M-W n=4), HSCT-treated Farber patients (A-L n=5; M-W n=1), and JIA patients (A-L n=5).

173

Figure 52. Modified ceramides that were not significantly different in Farber patient plasma versus controls. Plasma was collected from controls (n=21, except in B where n=27), untreated Farber patients (n=4, except in B where n=8), and HSCT-treated Farber patients (n=4, except in B where n=5), and JIA patients (n=5).

174

Figure 53. Sphingosines were not significantly different in Farber patient plasma versus controls. Plasma was collected from controls (n=27), untreated Farber patients (n=8), and HSCT-treated Farber patients (n=5), and JIA patients (n=5).

175

Figure 54. Ceramides that do not result in productive clustering of Farber patients. Unbiased hierarchical clustering was performed on samples using individual significantly different ceramides. Cluster dendrogram is seen at the top, followed by the patient key and a heatmap of the relative cytokine levels.

There were strong correlations between some of the altered ceramide species: C20:1-

Cer and dhC18-Cer had a very strong negative correlation, meaning that one may be metabolically related to the other. dhC12-Cer had a strong and very strong positive correlation, respectively, with dhC24:1-Cer and aOH-C18-Cer (Figure 48H-J). Other correlations were not found (Figure 55). There was also negative correlation of ceramide

176 species with the age at sample collection: dhC24:1-Cer, and aOH-C18-Cer correlated very strongly with the age at sample collection (Figure 48K-L). Other ceramides did not correlate with the age at sample collection, and none correlated with leukocyte ACDase activity

(Figure 56).

177

Figure 55. Ceramides with weak or no correlation to each other. Plasma was collected from untreated Farber patients (n=4). The correlation between ceramide species in the patient plasma was measured. Ceramides were analyzed by mass spectrometry. Cer, ceramide.

178

Figure 56. Ceramides with weak or no correlation to the age at sample collection or ACDase activity. Plasma was collected from untreated Farber patients (n=4). The correlation between ceramide species in the patient plasma was measured. Ceramides were analyzed by mass spectrometry. Cer, ceramide. ACDase, acid ceramidase.

179

5.3.3 Cytokines and ceramides

Inter-relationships between cytokine and ceramide levels were also probed. MCP-1,

IP-10, and VEGF correlated positively with dhC22:1-Cer-1P, C20:1-Cer, and dhC12-Cer, respectively (Figure 57). dhC24:1 and aOH-C18-Cer correlated positively with IL-12 and

VEGF levels. dhC18-Cer correlated negatively with levels of IP-10 but positively with that of IL-6. There was no correlation between other cytokines and ceramides (Figure 58-Figure

60).

180

Figure 57. Cytokine changes correlate with ceramide changes in plasma. The change in significantly different cytokines was compared with the change in significantly different ceramides. There was a strong positive correlation between MCP-1 and dhC22:1-Cer-1P (A). IP-10 correlated very strongly positively with C20:1-Cer (B), and strongly negatively with dhC18-Cer (C). IL-6 correlated very strongly positively with dhC18-Cer (D). IL-12 correlated moderately with dhC24:1-Cer (E) and strongly with

181 aOH-C18-Cer (F). VEGF correlated strongly positively with dhC12-Cer (G), and very strongly positively with dhC24:1-Cer (H), and aOH-C18-Cer (I). MCP-1, monocyte chemotactic protein 1; IP-10, interferon gamma- induced protein 10; IL-6, interleukin 6; IL-12, interleukin 12; VEGF, vascular endothelial growth factor; Cer, ceramide.

182

Figure 58. Ceramides with weak or no correlation to MCP-1 or IP-10 levels. The change in significantly different cytokines was compared with the change in significantly different ceramides. MCP-1, monocyte chemotactic protein 1; IP-10, interferon gamma-induced protein 10; Cer, ceramide.

183

Figure 59. Ceramides with weak or no correlation to IL-6 or IL-12 levels. The change in significantly different cytokines was compared with the change in significantly different ceramides. IL-6, interleukin 6; IL, interleukin; Cer, ceramide.

184

Figure 60. Ceramides with weak or no correlations to VEGF levels. The change in significantly different cytokines was compared with the change in significantly different ceramides. VEGF, vascular endothelial growth factor; Cer, ceramide.

5.3.4 Chitotriosidase activity

Given the dominant role of macrophages in FD, we investigated the activity of chitotriosidase in plasma. Chitotriosidase is a chitinase that is secreted by activated macrophages (Boot et al., 1995). Plasma chitotriosidase activity is high in patients with macrophage-involved disorders, such as Gaucher disease and Niemann-Pick disease (Ries et al., 2006). To determine if this biomarker was altered in FD, we evaluated its activity in plasma.

Chitotriosidase activity in plasma from patients with FD was higher than parental/sibling controls but lower than Gaucher patients with characteristically 'high' activity

(Figure 61). The activity in FD was in the same range as Gaucher patients with characteristically 'low' activity.

185

Figure 61. Chitotriosidase activity in Farber patient plasma. Plasma was collected from controls (n=18), Gaucher patients with independently identified 'high' chitotriosidase activity (n=8), Gaucher patients with independently identified 'low' chitotriosidase activity (n=3), untreated Farber patients (n=11), HSCT-treated Farber patients (n=2), and an ECD patient (n=1). Chitotriosidase activity was measured in vitro using a fluorescent substrate. *p<0.0001.

5.4 Discussion/conclusions

We identified patterns in plasma cytokines and ceramides due to ACDase deficiency that are unique to mice and humans with FD and not present in controls of either species,

Farber patients treated with HSCT, or JIA patients. These changes are critical to understanding the pathophysiology of FD. The cytokines identified can also be useful for the differential diagnosis of FD and JIA.

186

5.4.1 Understanding the biology of Farber disease

5.4.1.1 The role of cytokines in the initiation and progression of Farber

disease

In general, neutrophils are the first to arrive to a region of inflammation, followed by monocytes/macrophages. We observed the same sequence in the organs of Farber mice. The timing of the elevated cytokines in Farber mice (Figure 68) matches this process. KC, an inflammatory cytokine that recruits neutrophils, peaks first at 7 weeks. Next, MIP-1a and

MCP-1, inflammatory cytokines that recruit monocytes/macrophages, peak at 9 and 11 weeks, respectively. These cytokines are elevated in plasma and may be recruiting neutrophils and monocytes out of the BM and into the circulation where they can travel to and infiltrate organs. MCP-1 levels were elevated in the plasma of 11 week old Farber mice compared to WT mice of the same age (Figure 68). In addition, higher levels of MCP-1 were observed in organs compared to plasma, suggesting that circulating monocytes are the source cells that infiltrate organs in this disorder (Alayoubi et al., 2013).

We previously demonstrated that the macrophage infiltration is so severe in the

Farber mouse BM, thymus, spleen, and LN that the endogenous architecture is disrupted

(Dworski et al., 2015). Subcutaneous nodules, common to all reported FD patients, are composed of lipid-filled macrophages (Levade et al., 2009). Activated macrophages are present, as seen by elevations in chitotriosidase plasma activity, but not to the levels of 'high'

Gaucher disease.

Based on our results, future treatment studies in Farber disease may investigate the correlation between therapeutic efficacy and decrease in MCP-1 and MIP-1a as markers of

187 lower levels of macrophage infiltration and tissue damage. The anecdotal evidence of the moderate effect of powerful anti-inflammatory medications targeting the innate immune system (IL-6 receptor blockers) as symptomatic treatments in Farber disease, first related by

Dr. Bo Magnusson and confirmed by other co-authors, also can be seen to support this proposition. This is further supported by the suggestion that interruption of MCP-1 signaling may be a therapeutic intervention for rheumatoid arthritis and atherosclerosis (Hayashida et al., 2001; Kusano et al., 2004). Modulation of these specific cytokines is critical, based on our findings. Highlighting the importance of MCP-1 in the etiology of FD, the absolute levels of the cytokine in plasma were 10-fold higher in patients with FD than in controls (Figure

43). In another LSD where macrophages are a key feature, i.e. Gaucher disease, this elevation was less than 2-fold (Pavlova et al., 2011).

As shown above, MCP-1 and MIP-1a are uniquely elevated in mice and humans with

FD. Together they may be recruiting monocytes into organs. Furthermore, this may be a feed-forward mechanism, as IP-10, IL-6, and IL-12 are themselves secreted by macrophages

(Keeley et al., 2008; Hurst et al., 2001). Increasing ceramide levels by stimulating its de novo formation resulted in MCP-1 and IL-6 expression in a macrophage cell line (Hamada et al.,

2014). MIP-1a, IP-10, IL-6, and IL-12 are implicated in arthritis, where they may exacerbate inflammation (Patel et al., 2001; Nishimoto et al., 2004; Petrovic-Rackov et al., 2006).

C22:1-Cer-1P was also elevated in Farber patient plasma. Cer-1P has been shown in a macrophage cell line to stimulate MCP-1 release and cell migration through the phosphatidylinositol 3-kinase (PI3K)/Akt, mitogen-activated protein kinase kinase

(MEK)/extracellularly regulated kinases (ERK), and p38 pathways (Arana et al., 2013).

188

Other specific cytokines were also elevated in human Farber patients. IP-10, IL-6, and IL-12 were all increased in Farber patients compared to controls.

It is important to note that three out of four of the elevated cytokines returned to baseline levels following HSCT. This may be a sign of the positive effects of HSCT on FD, or HSCT may reduce inflammatory cytokine secretion. Plasma cytokines can be added to the toolbox of read-outs when assessing the efficacy of experimental therapeutics for FD, such as enzyme replacement therapy and gene therapy (Levade et al., 2009; Ramsubir et al., 2008;

Walia et al., 2011).

5.4.1.2 Ceramides in Farber disease

The ceramide trends in non-Farber controls in this report corroborate previous mass spectrometry analyses of plasma ceramides in healthy humans (Drobnik et al., 2003;

Hammad et al., 2010; Ichi et al., 2006). C24-Cer was the main product, and C18-Cer was a relatively minor product in plasma samples from both controls and Farber patients. This study analyzed more ceramides than the others, and we found several additional ceramide species in non-Farber human controls at lower levels than C18-Cer, including C18:1-Cer,

C20:1-Cer, C20:4-Cer, C22:1-Cer, C26-Cer, and most of the Cer-1P. In our analysis, C22:1-

Cer-1P was a minor product in controls.

Interestingly, not all ceramide species were elevated in the plasma of Farber patients.

This is not surprising as there are other ceramidases (one neutral ceramidase and three alkaline ceramidases) that also degrade ceramides. The only ceramides significantly elevated in the plasma from FD mice were C16-Cer and dhC16-cer (Figure 47); 24 other ceramides analyzed were not changed (Figure 49). This suggests that there is special significance of

189 ceramides with a fatty acid chain length of 16 carbons. We did not find a unique pattern of ceramide accumulation in the Farber patients in this cohort, and were not able to demonstrate clustering by the unique ceramides identified in the Farber mouse model. However, a few ceramide species correlated with each other, suggesting a possible common defect.

Specifically, dhC12-Cer, dhC24:1-Cer, and aOH-C18-Cer are positively correlated with each other, and negatively correlated with age at sample collection.

The product of ceramide degradation is sphingosine and a free fatty acid. The assumption is that reduced degradation of ceramide would lead to lower levels of sphingosine. This was not the case for plasma: sphingosine and its derivatives were unchanged (Figure 50; Figure 53). The levels of sphingosine are critical to a cell: in the treatment of another LSD, Niemann-Pick Type C, too much sphingosine resulted in neurodegeneration (Lloyd-Evans & Platt, 2010). These data suggest that either there is an alternate pathway through which sphingosine is formed (such as by other ceramidases), or that its degradation is reduced so that the small amount of sphingosine produced by ACDase is maintained at sufficient levels for its function.

5.4.2 Improving diagnosis of Farber disease: a new method of

verification that distinguishes between Farber and JIA patients

We identified an alternate method of verification of FD: by analysis of plasma cytokines. We demonstrated that MCP-1, IP-10, IL-6, and VEGF are elevated in Farber patients. These four markers may be useful for physicians to differentiate between FD and

JIA. Expression of MCP-1, IP-10, IL-6, IL-12 and VEGF results in discrimination of 10 of

13 Farber patients from JIA patients in cluster analysis (Figure 41F). An identical success

190 rate was achieved when clustering the patients based on MCP-1 expression alone (Figure

41G). This result, coupled with the observation that MCP-1 was the only cytokine elevated in both mouse and human Farber plasma, suggests that MCP-1 is a key player or biomarker in the etiology of FD. It is not the only cytokine that attracts monocytes that was elevated, raising the possibility that the role of MCP-1 could be shared/replaced by other cytokines. A further benefit of MCP-1 as a differential marker between Farber and JIA patients is the possibility of large-scale screening for FD in JIA populations of unknown etiology to identify misdiagnosed patients.

Using MCP-1 and other cytokines may also be helpful in identifying misdiagnosed

Farber patients in other populations. Spinal muscular atrophy with progressive myoclonic epilepsy (SMA-PME) is caused by ACDase deficiency but does not present the same way as

FD (Zhou et al., 2012; Dyment et al., 2014; Rubboli et al., 2015; Gan et al., 2015).

Comparison of plasma from SMA patients with SMN1 or SMN2 mutations compared to

SMA-PME patients with ASAH1 mutations may reveal cytokine differences that can be used to screen larger SMA populations. Samples from SMA-PME patients were not available at the time of analysis.

Collectively, our data support that ACDase deficiency results in changes in the cytokine and ceramide profiles found in the plasma of mice and patients. These changes are unique to untreated FD and are not seen in controls, HSCT-treated FD patients, or JIA patients. These changes can be used to better understand the biology of FD and to develop novel treatments. MCP-1 plasma levels may also be useful as a tool to differentiate between a diagnosis of FD and JIA.

191

Chapter 6 Bone Marrow Transplantation Doubles the Lifespan of Acid Ceramidase-Deficient Mice

6.1 Introduction

Due to the rarity of FD and the often rapid disease progression, limited hematopoietic and neurological studies have been performed on untreated Farber patients and those who have received HSCT. To better understand this disorder, we previously developed a mouse model with high fidelity to human FD (Alayoubi et al., 2013). Using this model, we recently described the specific hematopoietic and neurological defects that occur in FD (Alayoubi et al., 2013; Dworski et al., 2015; Dworski et al., 2017; Sikora et al., 2017; Chapter 3; Chapter

4; Chapter 5). We identified that the hepatosplenomegaly is primarily due to organ infiltration by foamy, Mac-2+ macrophages (Dworski et al., 2015). This macrophage infiltration is also present in other hematopoietic-associated organs, e.g., BM, thymus, and

LN, causing the latter two to be severely enlarged. The cellular architecture in these organs is disrupted to the point that, by the end of their lives, the thymus is nearly devoid of CD4+

CD8+ T cells, and the BM is nearly devoid of B cells at all developmental stages (Dworski et al., 2015). In the brain, macrophage/microglial infiltration is also seen, along with astrocytosis and hydrocephaly (Sikora et al., 2017). We also identified a pattern of elevated plasma cytokines that are unique to FD and may be playing a role in these outcomes

(Dworski et al., 2017). Note that previous case reports have not included analyses of these signs.

192

Knowing specific changes that occur in FD, we have now compared untreated mice with FD with those receiving HSCT. This analysis is critical to understanding the effects of

HSCT on FD, as it is a limited treatment. HSCT has been most helpful in patients with a peripheral-only disease: subcutaneous nodules disappear and joint mobility increases (Yeager et al., 2000; Vormoor et al., 2004; Ehlert et al., 2006). However, for FD patients who have neurological symptoms before the transplant, HSCT is not as effective. Even patients who did not have neurological symptoms before HSCT may still develop them afterwards (Yeager et al., 2000; Cappellari et al., 2016). Recently, a child with FD displayed difficulty standing pre-transplantation, likely due to joint pain and suggesting peripheral-only disease. Yet twenty-two months following HSCT, he displayed nystagmus, titubation, and ataxia, followed by a progression to dysphagia, dyspnea, and arm weakness two months later

(Cappellari et al., 2016). HSCT does not specifically target the brain, and the specific changes in the brain following HSCT in FD have not been assessed. An understanding of this would be critical to developing future therapies that better target neurological manifestations.

In this study we performed HSCT on mice with FD for the first time. HSCT proved to be the most effective treatment of FD in a mouse model to date, resulting in more than doubling of the lifespan of mice with FD. We identify the many positive effects of HSCT, both in the periphery and brain, and assess the model for impact on newly characterized signs of disease.

6.2 Materials and methods

6.2.1 Animals

Mice with mutations in the ACDase gene Asah1 (Asah1P361R/P361R) were previously generated by our group (Alayoubi et al., 2013). Mutant (Asah1P361R/P361R - Hom) and wild-

193 type (Asah1+/+ - WT) animals were obtained by mating heterozygous (Asah1P361R/+ - Het) mice. Genotyping was performed as described previously (Alayoubi et al., 2013). Mice were housed at the Animal Resource Centre at the University Health Network (Toronto, Canada) under specific pathogen-free conditions. All animal studies were approved by the University

Health Network Animal Care Committee.

6.2.2 Bone marrow transplantation (BMT)

BMT was performed as done previously by our group (Liang et al., 2007; Takenaka et al., 2000; Yoshimitsu et al., 2007). Donor WT mice received 150 mg/kg 5-fluorouracil

(Sigma) by intraperitoneal injection to enrich for hematopoietic stem cells. Four days later, donor animals were killed and BM was flushed from tibia and femurs. BM mononuclear cells were isolated by centrifugation over a Nycoprep 1.077 density gradient (AXIS-SHIELD).

One million donor cells were injected via tail vein into lethally-irradiated (11 Gy) recipient

Hom Farber or Ctrl (WT and Het) mice. We previously reported that our BMT techniques result in >80% engraftment (Dworski et al., 2015). All data presented here is an amalgamation of 4 independent BMT cohorts.

6.2.3 Skin stretch

Mice were scruffed and the height of the tent made by the skin at the back of the neck was measured with vernier calipers, as done previously (Lopez-Vazquez et al., 2016).

194

6.2.4 Peripheral blood counts

Blood was collected from mice by saphenous vein bleeding into EDTA-coated tubes.

Complete blood counts were performed on a Hemavet 950 FS (Drew Scientific Group).

6.2.5 Quantitation of ceramides

Lipids were extracted from the liver and brain of untreated and BMT-treated mice.

Ceramide and diacylglycerol (DAG) levels were determined using E. coli DAG kinase and

[γ32P]-ATP as previously described (Bielawska et al., 2001).

6.2.6 Cytokine analysis

Cytokine levels were analyzed as described previously from mouse plasma using the

Bio-Plex Pro Mouse Cytokine 23-plex Assay (Bio-Rad) as per the manufacturer’s instructions (Dworski et al., 2017). Luminescence was quantified on the Luminex 100 instrument (Luminex). Data where low bead count was observed (<45 beads) was omitted.

Where individual data points were below the detection limit of the kit, values were set to half of the detection limit. Where individual data points were above the detection limit, values were set to the top of the detection level. Heatmaps were generated in the statistical software program R using the gplots package (Warnes et al., 2011).

6.2.7 Immunohistochemistry

Tibias, thymuses, spleens, and LN of WT, Hom Farber mice, transplanted Ctrl(WT), and transplanted Hom(WT) mice were fixed for 24 hours in 10% buffered formalin,

195 embedded in paraffin, and sectioned. Bones were decalcified prior to sectioning. Sections were stained using antibodies against B220 (BD Bioscience; #550286, 1:100), CD3 (Sigma;

#C7930, 1:1000), and Mac-2 myeloid cells (Cedarlane; #CL8942AP, 1:2000). Biotinylated

Anti-Rat IgG (H+L) and DAB were used to detect the antibodies in the Vectorstain ABC kit

Elite Standard (Vector Laboratories).

6.2.8 Flow cytometry

BM was harvested by flushing femurs and tibias with PBS containing 2% FCS. After

RBC lysis, cells were washed and resuspended in PBS with 3% FCS and counted. For flow cytometry analysis, cells were stained for 30 min at 4ºC with a combination of antibodies.

Antibodies were directly conjugated to FITC, PE, biotin, or APC, and were purchased from

BD Biosciences: IgD FITC (clone 11-26c.2a), CD43 PE (clone S7), IgM biotin (clone 33.60;

Kincade et al., 1980), B220 APC (clone RA3-6B2). PerCP Streptavidin (Streptavidin-

Peridinin Chlorophyll-a Protein, BioLegend) was used as a second-step reagent for the indirect immunofluorescent staining of cells in combination with biotinylated primary antibodies. Flow cytometry analysis was performed on a FACSCalibur or LSRFortessa flow cytometer (BD Biosciences) using CellQuest or FACSDiva software (BD Biosciences). Data analysis was conducted using FlowJo (Tree Star, Inc.).

6.2.9 Brain tissue collection and processing

For light microscopy analyses, a 10-week-old Hom Farber mouse was perfused transcardially with PBS followed by 4% PFA. The entire brain was further immersion-fixed overnight in 4% PFA at 4°C and then transferred to and stored in PBS at 4°C. Half of the

196 brains (cerebrum and cerebellum) of one transplanted Hom(WT) and one transplanted

Het(WT) 30-week-old mice were collected and immersion-fixed in 4% PFA at 4°C and then transferred to and stored in PBS at 4°C.

6.2.10 Brain immunofluorescence

For IF labeling, half brains were split into cerebral and cerebellar/brain stem parts at the mesencephalic level. The separated fragments were embedded into 8.0% sucrose/3.5% agarose and 35 m thick serial sections (coronal in cerebrum and sagittal in cerebellum) were cut using Leica VT-1000S Vibratome (Leica Microsystems, Wetzlar, Germany). Sections representing rostral<>caudal bregma (-1.94<>-2.30) and lateral (0.40<>0.84) ranges

(Franklin & Paxinos, 2008) were selected from cerebra and cerebella, respectively. Matched sections were stained by multi-IF protocols as described before (McGlynn et al., 2004) and mounted using ProLong Gold medium with DAPI (LifeSciences). The following primary antibodies were used for staining of specific epitopes: rat anti-CD68 mAb (AbD Serotec,

1:1000), and mouse anti-glial acidic fibrillary protein (GFAP) mAb G-A-5 (Sigma-Aldrich,

1:3000). Species-specific secondary antibodies conjugated to AF488 and AF546 (Invitrogen,

Carlsbad) dyes were used for detection of primary antibodies.

6.2.11 Brain microscopy

Overview images of the cerebral and cerebellar IF-labeled sections were acquired using IX70 microscope (Olympus) equipped with an HQ2 camera (Photometrics), Proscan II encoded xyz stage (Prior Scientific) equipped with 10 position excitation/emission

197

SmartShutters filter wheels (Sutter). Exposure time for all antibody combinations was 750 ms (gain 2 per channel. Excitation/emission conditions for the AF488 and AF546 dyes were exc.490/20/b.s. 480-513/em.535/40 and exc.572/23/b.s.555-588/em.630/60 nm, respectively. Individual but partially overlapping double-channeled 14-bit images (down- sampled to 8-bit) were acquired with Plan 10x (NA 0.25) objective and were digitally stitched in Metamorph/MetaFluor (Molecular Devices) and consequently in Photoshop CS6

(Adobe) software. Figures were prepared in Photoshop CS6 (Adobe). For better contrast and visibility and not to alter the biological information, images were stretched to fill the full dynamic 8 bit-ranges.

6.2.12 Statistics

Survival significance was tested with the log-rank (Mantel-Cox) test comparing pairs of curves. Body weight was analyzed by two-way ANOVA and skin stretch by a 1-way

ANOVA, and both followed by a Bonferroni post-test. Organ weights, ceramide levels, and

Cer/DAG ratios were compared by one-way ANOVA with Bonferroni post-test. All statistics were analyzed using GraphPad Prism 5.0.

6.3 Results

6.3.1 BMT with wild-type cells doubles the lifespan of FD mice

Homozygous Farber mice (Hom) were lethally-irradiated and received a bone marrow transplant (BMT) from wild-type (WT) littermates; these are designated as transplanted

Hom(WT) mice. Some Hom mice were irradiated and not given a BMT; these controls are

198 designated as Hom(rad) mice. As well, WT and heterozygous (Het) mice were irradiated and received BMT with cells from WT mice; these are designated as Ctrl(WT) mice.

Untreated Hom ACDase-deficient mice have a lifespan of 49-91 days (7-13 weeks)

(Alayoubi et al., 2013). Hom(rad) mouse survival was found to be within the untreated Hom mouse timeframe, reaching a maximum of 77 days (Figure 62A). The median transplanted

Hom(WT) mouse survival was 60 days; however, if mice that died at day 77 or earlier

(presumably from failure to receive adequate BM cells for technical reasons or from failure to engraft) are excluded, the median survival was 188 days (27 weeks). This is more than double the lifespan of untreated Hom Farber mice. This extension of lifespan was unique to

Hom mice receiving WT cells: Hom Farber mice receiving Hom cells did not demonstrate an extended lifespan, living to a maximum of 81 days (data not shown).

199

Figure 62. BMT with WT cells prolongs the lifespan of Hom ACDase-deficient mice and also delays weight loss. (A) Farber Hom and control (WT and Het) mice received radiation alone (rad) or radiation and BMT (WT). Transplanted Hom(WT) mice had an extended lifespan beyond the lifespan of untreated Hom Farber mice. (B) Transplanted Hom(WT) mice maintained a similar weight to transplanted Ctrl(WT) animals for the first 2 months post-BMT, and did not fall below the average weight of untreated Hom Farber mice at the end of their lives (age 9 weeks; average from Dworski et al., 2017). The following pairs were significantly different: Ctrl(WT) vs Ctrl(rad), Hom(WT) vs Hom(rad), Hom(WT) vs Ctrl(WT), and Hom(WT) vs Ctrl(rad). (C) Skin stretch was quantified by scruffing mice and measuring the height of the tent produced using vernier calipers. Ctrl and Hom mice were measured at 9 weeks of age, and transplanted Ctrl(WT) and Hom(WT) animals were measured at 27-29 weeks of age. (D-H) Organs were isolated from 9-week-old Ctrl (WT or Het) mice, Hom Farber mice, and from 29-week-old transplanted Hom(WT) mice. Organ weights are expressed as a percent of total body weight of the animal. Error bars represent mean ± SEM. Hom(WT), Hom mice receiving radiation and BMT with WT cells; Ctrl(WT), WT and Het mice receiving radiation and BMT with WT cells; Hom(rad), Hom mice receiving radiation alone; Ctrl(rad), WT and Het mice receiving radiation alone.

6.3.2 BMT with WT cells corrects physical signs of FD in homozygous

ACDase-deficient mice

Untreated Hom Farber mice are smaller than their WT and Het littermates. This difference begins shortly after weaning. They also continue to decline in weight as they near their end-point (Alayoubi et al., 2013; Dworski et al., 2015). Near that end-point, the average

200 weight of an untreated Hom Farber mouse is 11.9±1.0 g (Dworski et al., 2015). Transplanted

Hom(WT) mice in this study maintained a weight above their untreated counterparts for the duration of their lives. They also kept up with Ctrl(rad) and Ctrl(WT) mice in the slope of their growth increases (Figure 62B) for a full 2 months following BMT (until ~12 weeks of age, i.e. beyond the end of the untreated Hom Farber mouse lifespan).

Our Farber mice have many visceral physical manifestations. Pelvic prolapse occurs in Hom Farber males near the end of their lifespan. In transplanted Hom(WT) mice pelvic prolapse was delayed by many weeks until the end of their lifespan (data not shown). The skin of untreated Hom Farber mice was also found to be less elastic than that of Ctrl mice at

9 weeks of age (Lopez-Vazquez et al., 2016). Transplanted Hom(WT) mice did not manifest this: their skin was as elastic as Ctrl and Ctrl(WT) animals, even at their end-point at 27-29 weeks of age (Figure 62C). Organomegaly of hematopoietic-associated organs was prevented by BMT: the spleen and liver of transplanted test animals were of normal size (Figure

62D,E). Further, the LN and thymuses were drastically reduced in size: the thymus in one mouse was so small it could not be found, in another it was at Ctrl mouse size, but in a third it was enlarged (Figure 62F). Non-hematopoietic-associated organs, such as the kidney, did not demonstrate organomegaly, as with untreated mice (Figure 62G). Brains were not specifically targeted for correction in this treatment schema and still weighed more than those from Ctrl mice (Figure 62H).

201

6.3.3 BMT reduces ceramide accumulation in peripheral organs and the

central nervous system

The goal of the BMT is to introduce WT hematopoietic cells that can produce functional ACDase to metabolize the excess ceramides. In our mouse model of FD, ceramide accumulation is seen in all tissues tested, including the liver and brain, but to varying degrees

(Alayoubi et al., 2013). Transplanted Hom(WT) mice had near-normal levels of ceramides in the liver and, very interestingly, in brain, both in absolute values and as a ratio to DAG content (Figure 63). This indicates that the impact of BMT-induced ceramide degradation was broad, from the liver, a peripheral organ, to the brain, a central nervous system organ.

Figure 63. BMT with WT cells prevents ceramide accumulation in Farber mice. Ceramide was measured in the liver (A) and brain (B) of untreated 9-week-old Ctrl and Hom mice, and transplanted 28-week-old Hom(WT) mice. The ratio of ceramide to DAG in the liver (C) and brain (D) of these mice was also compared. Error bars represent mean + SEM. Ctrl, n= 4-6; Hom, n=2; Hom(WT), n=2. Hom(WT), Hom Farber mice receiving radiation and a BMT with WT cells; Ctrl(WT), WT and Het mice receiving radiation and BMT with WT cells; Hom(rad), Hom Farber mice receiving radiation alone. DAG, diacylglycerol.

202

6.3.4 BMT is able to partially correct the hematopoietic defects in Farber

mice

We next investigated the effects of BMT on individual hematopoietic cell populations. Hom Farber mice have leukocytosis, with the bulk of the cell increase being monocytes and neutrophils (Alayoubi et al., 2013). Transplanted Hom(WT) mice did not exhibit this abnormality (Figure 64A,B,D). Importantly, at all ages post-BMT, lymphocyte, neutrophil, and monocyte counts were similar to that seen in Ctrl(WT) and Ctrl(rad) animals.

Transplanted Hom(WT) mice maintained a PB cell count similar to Ctrl(WT) animals throughout the duration of their lives post-BMT; some fluctuations were seen in certain months but these were normalized by the next month. Two exceptions were hemoglobin

(Figure 64H) and hematocrit (Figure 64I), which showed a trend to decline as transplanted

Hom(WT) mice aged, and were significantly lower by 5 months post-BMT. By this age transplanted Hom(WT) mice had become anemic.

203

A B C

D E F

G H I

J K L

M N

Figure 64. BMT with WT cells prevents leukocytosis in Farber mice. Peripheral blood was collected from mice receiving BMT and from control mice. Complete blood counts were performed to quantify cell types. The range of number of animals per group includes high starting numbers (Ctrl(rad), n=16; Hom(rad), n=7; Ctrl(WT), n=6; Hom(WT), n=15) and lower ending numbers (Ctrl(rad), n=4; Hom(rad), n=2; Ctrl(WT), n=4; Hom(WT), n=4) as mice die over time.

204

In Hom Farber mice, the monocytosis extended from the PB into the organs.

Hematopoietic-associated organs in Farber mice are filled with Mac-2+ macrophages, which affect organ architecture (Dworski et al., 2015). In this present study, transplanted Hom(WT) mice had reduced macrophage infiltration (Figure 65). The BM of Hom(WT) mice had a similar amount of Mac-2+ macrophages as Ctrl(WT) animals, but the architecture appeared more sparse (Figure 65). The LN from transplanted Hom(WT) recipients had more macrophages than Ctrl(WT) animals but the individual macrophages were not as large in morphology and foamy as in untreated Farber Hom mice (Figure 65). In the thymus the architecture was still somewhat disturbed: macrophages were in their appropriate location

(more in the medulla than the cortex) but there was an excess of macrophages in the medulla

(Figure 65). Similarly, in the spleen there were fewer macrophages but the B220+ cell and

CD3+ cell architecture was not entirely intact (data not shown).

205

206

Figure 65. Fewer macrophages accumulate in organs from transplanted Hom(WT) mice and they do not have a 'foamy' morphology. Bone marrow, lymph nodes, and thymuses from 9-week-old WT and Hom Farber mice and 29-week-old transplanted Hom(WT) and Ctrl(WT) mice were stained with H&E to demarcate general organ architecture, anti-B220 to mark B cells, anti-CD3 to mark T cells, and anti-Mac-2 to mark macrophages.

The excess macrophages in Hom Farber mice interrupt the cellular architecture of the

BM and thymus, thus disrupting hematopoiesis there. B and T cell progenitors are almost completely absent by 9 weeks of age in the BM and thymus of Farber mice, respectively

(Dworski et al., 2015). Here, BMT partially rescued the macrophage infiltration, but it did not appear to rescue the loss of the B cell progenitor phenotype. Pre-, pro-, immature, and transitional B cells were lower in the BM of transplanted Hom(WT) than Ctrl(WT) animals

(Figure 66). Only mature B cells were similar in transplanted Hom(WT) and Ctrl(WT) animals at 29 weeks, as they were similar in Hom Farber and WT mice at 9 weeks (Dworski et al., 2015). The prevalence of T cell progenitor phenotypes following transplant could not be assessed as most thymuses were too small to collect.

207

Figure 66. B cell progenitors are still depleted in transplanted Hom(WT) bone marrow. Bone marrow was isolated from the tibia and femurs of mice and stained with antibodies that mark different stages of B cell development (Pre = pre-B cell, Pro = pro-B cell, Imm = immature B cell, T = transitional B cell, M = mature B cell). Representative flow cytometry plots show that developmental B cells are reduced in transplanted Hom(WT) mice compared to Ctrl(WT) mice at a similar age. Developmental stages are in gates with percent cells corresponding to that stage listed.

6.3.5 BMT reduces brain alterations in Hom Farber mice

In Farber patients, neurological manifestations of the disorder are not corrected by

HSCT. Therefore, we investigated the changes post-BMT in the Farber mouse brain.

Macrophage/microglial and astrocyte pathology in 30-week-old transplanted Hom(WT) animals were compared to that in a Ctrl(WT) mice and also a 10-week-old (disease end- stage) untreated Hom Farber mouse. The untreated Hom Farber mouse expressed the full end-stage pathology characterized by hydrocephaly in all ventricles. As a previously identified hallmark (Sikora et al., 2017), massive and mostly perivascular granuloma-like

208 macrophage/microglia accumulations were present, that particularly affect the cerebral white matter, periventricular zones, choroid plexus, and meninges (Figure 67, top panels).

Additionally, extensive astrocytic gliosis could be identified in the cerebrum of the untreated

Hom mouse (Figure 67, bottom panels). The cerebellum was, similar to our previous observations in up to 10-week-old, untreated Hom Farber mice, less affected both by macrophage/microglial and astrocytic pathologies than the cerebrum (data not shown).

209

Figure 67. Transplanted Hom(WT) mouse brains retain hydrocephaly, macrophage/microglial infiltration, and astrocytosis. Brains from 10-week-old Hom mice and 30-week-old transplanted Hom(WT) and Ctrl(WT) mice were stained for CD68 to mark macrophage/microglial (CD68+) and glial fibrillary acidic protein (GFAP) to mark astrocytes. Hydrocephaly in Hom Farber mice is marked by white asterisks and in Hom(WT) by white and black arrows. Subcortical white matter areas and corpus callosum are marked by arrowheads.

Compared to the untreated Hom Farber mice, transplanted Hom(WT) animals had substantially less pronounced hydrocephaly (Figure 67), although it was still evident in the lateral ventricle, and this was not enough to normalize brain weight (Figure 62H). The microglial/macrophage and astrocytic pathologies were present in identical neuro-anatomic locations of transplanted Hom(WT) animals as in the cerebrum of Hom Farber mice, yet to a

210 lower degree (Figure 67, top panels). In Hom Farber mouse brains, accumulations of CD68+ macrophages /microglia were present in grey matter but predominated in white matter, periventricular zones, and the choroid plexus. In contrast, in transplanted Hom(WT) mouse brains the CD68+ macrophage/microglial conglomerates were minimal and limited to subcortical white matter areas and the corpus callosum. As in the cerebrum, cerebellar pathology was also mitigated in transplanted Hom(WT) mice (data not shown). Findings in cerebra and cerebella of Ctrl(WT) animals were comparable to 10-week-old WT mice as identified previously (Sikora et al., 2017).

6.3.6 Circulating cytokines are normalized in BMT-treated Hom mice

We previously reported that monocyte chemotactic protein 1 (MCP-1) was elevated in the plasma of end-point (11-week-old) Hom Farber mice compared to Ctrl mice (Dworski et al., 2017). MIP-1a was also elevated at 9 weeks of age. We compared the plasma levels of these and other cytokines in the BMT-treated mice. MCP-1 and MIP-1a were both at normal levels in transplanted, 29-week-old Hom(WT) mice (Figure 68A,B). As well, IL-1a, which was significantly elevated in untreated 9-week-old Farber mice, was normalized in transplanted Hom(WT) recipients (Figure 68C). No additional cytokines that we quantified were significantly different in transplanted Hom(WT) compared to Ctrl(WT) animals (Figure

69).

211

) A B ) C )

L a MCP-1 b MIP-1a c L IL-1a *

m m *

/ * 2800 1000 / 6000

g g g 4000

p 2400 p

(

( 800 (

2000 2000

n n 300

i 1600 600 i

t t 250

a a

r 1200 r 200

t 400 t

n 800 n 150

e 200 e 100

c 400 c

n n 50

l l 0 0 l l 0 l l

) )

r r

t m t m

T T

T T T T

o o

o o o o

C C

W W

H H

H H H H

( (

( ( ( (

r r

t m

C 9 wk 11 wk 9 wk 11 wk 9 wk 11 wk C

H H Untreated BMT Untreated BMT Untreated BMT

Figure 68. Plasma cytokines are normalized in transplanted Hom(WT) mice. Plasma was isolated from 9- and 11-week-old Ctrl (WT and Het) and Hom Farber mice, and 29-week-old transplanted Ctrl(WT) and Hom(WT) animals. Cytokines were analyzed in a multiplex panel. 9-week-old Ctrl, n=8; 9-week-old Hom, n=4; 11-week-old Ctrl and Hom, n=2; Ctrl(WT) and Hom(WT), n=3. Ctrl and Hom data are adapted from our previous publication (Dworski et al., 2017).

212

) )

L L

L Aa IL-1b Bb L IL-2 cC IL-4 Dd IL-5

m m

/ / / 1500 / 500 900 500

g g g g 800

p p

p 1250 p 400

( ( ( ( 400

150

n n n 1000 n 60

o o

i i i i 120 300

t t t 750 t

a a a a 40

r r r r 90

t t t 500 t 200

n n n n 60

e e e e 20 100

c c c 250 c 30

n n n

o o o

0 l l 0 l l 0 l l 0 l l

) ) ) ) ) ) ) )

r r r r r r r r

C C C

t m t m t m t m

T T T T

o o o o

C C C C

W W

W W W W W W

H H H H H H

H H

( ( ( ( ( ( ( (

l l l

r r r

t m t m

o o o

C 9 wk 11 wk 9 wk 11 wk C 9 wk 11 wk C 9 wk 11 wk C

H H H

H Untreated BMT Untreated BMT Untreated BMT Untreated BMT

) )

)

L L

IL-6 L IL-12 Ee Ff IL-10 gG Hh L IL-13

m m

/ /

/ 500 1000 700 / 2200

g g

p p

p 600 p 2100

( (

( 400 800 (

500 500

n n

o o

i 300 i

i 600 400 i 400

t t

a a

r r r 300 r 300

t t 200 t 400 t

n n

n 200 n 200

e e

e 100 200 e

c c c 100 c 100

n n

o o

l l

0 l l 0 0 l l 0 l l

) )

) ) ) ) ) )

r r

r r r r r r

C C C

t m t m

t m t m t m t m

t m t m

T T

T T T T

T T

o o

o o o o o o

C C

C C C C C C

W W

W W W W W W

H H

H H H H H H

( (

( ( ( ( ( (

l l l

r r r

t m

t m t m

o o o

C 9 wk 11 wk C 9 wk 11 wk 9 wk 11 wk C 9 wk 11 wk C

H H H Untreated BMT Untreated BMT Untreated BMT Untreated BMT

)

) )

)

L L IL-17 GM-CSF IFNg

i j L bFGF k l

m I m J K m L

/ / 900 m 500 500

/ 8000

800 p

p p 400

( ( ( 400

(

8 6000 80

i 300

i i

t t 6 t 60

a a 4000 a

t t 200 4 t 40

n n

e e e 2000 100

c c 2 20

n n

o o

l l 0 l l 0 0 l l 0 l l

) )

) ) ) ) ) )

r r

r r r r r r

t m t m

t m t m t m t m

t m t m

T T

T T T T

T T

o o

o o o o o o

C C

C C C C C C

W W

W W W W W W

H H

H H H H H H

( (

( ( ( ( ( (

l l l

r r

t m

t m t m t m

o o o 9 wk 11 wk C 9 wk 11 wk C 9 wk 11 wk C 9 wk 11 wk C

H H H Untreated BMT Untreated BMT Untreated BMT Untreated BMT

)

L MIG L IP-10

L Mm Nn L KC oO Pp TNFa

/ 1200 / 8000 1400 600

p 1200

p 1000 p 7000

(

1500 1000

n 800 n 400

i i 1200 800

t 600 t

a a

r r 900 600 r

t t

400 n 200

n n 600 400 n

e e

c c 200 300 200 c

n n

o o

l l

l l 0 0 0 0 l l

) )

r r

C C

t m t m

T T

o o

o o o o

C C

C C C C

W W W W

H H H H

H H

( ( ( (

l l

r r

t m t m

o o

C C 9 wk 11 wk C 9 wk 11 wk 9 wk 11 wk 9 wk 11 wk C

H Untreated BMT Untreated BMT Untreated BMT Untreated BMT

)

q L VEGF

Q / 500

( 400

i 300

t 200

e 100

0 l l

) )

r r

t m t m

T T

o o

C C

W W

H H

( (

t m

9 wk 11 wk C

H Untreated BMT

Figure 69. Plasma cytokines that are not significantly different between untreated and transplanted mice. Plasma was isolated from 9- and 11-week old Ctrl (WT and Het) and Hom Farber mice, and 29-week-old transplanted Ctrl(WT) and Hom(WT). Cytokines were analyzed in a multiplex panel. 9-week-old Ctrl, n=8; 9- week-old Hom, n=4; 11-week-old Ctrl and Hom, n=2; Ctrl(WT) and Hom(WT), n=3. Ctrl and Hom data are adapted from our previous publication (Dworski et al., 2017).

213

6.4 Discussion/conclusions

6.4.1 Comparison to direct lentiviral injection

The current study demonstrates the efficacy of BMT in a mouse model of FD. Our

BMT protocol is the most efficacious treatment developed to date for the mouse model of

FD, doubling its lifespan and significantly reducing signs of the disease.

BMT dramatically improved Hom Farber mice. Compared to our initial characterization of the FD mouse (Alayoubi et al., 2013), the lifespan was more than doubled from 7-13 weeks in untreated Hom mice to 11-40 weeks in transplanted Hom(WT) mice.

Hematopoietic-associated organs were no longer enlarged and leukocytosis was corrected.

Ceramide levels virtually returned to those seen in control mice in both peripheral organs and brain. Leukocytosis was fully prevented: in untreated Hom Farber mice near their end-point, neutrophils were the cell type with the greatest fold elevation, reaching 8,000 cells/µl

(Alayoubi et al., 2013), but in transplanted Hom(WT) mice near their endpoint, neutrophils remained below 5,000 cells/µl, the range seen in WT mice of the same age.

Comparison of the results presented here with those of our previously published report involving the treatment of Farber mice with a postnatal single-dose injection of an

ACDase-expressing lentiviral vector (LV) (Alayoubi et al., 2013) indicates that BMT is a more effective treatment. With LV injection, the mice lived up to 16 weeks of age (median survival 13 weeks; Alayoubi et al., 2013), while in this study transplanted Hom(WT) mice lived up to 40 weeks (median survival 27 weeks). With direct neonatal LV injection, weight loss was prevented until age 5 weeks, followed by a steep decline in weight at the same rate as control-treated Hom Farber mice (Alayoubi et al., 2013). With BMT, weight loss was prevented for a full 2 months post-treatment. LV injection was able to reduce liver ceramides

214 but not brain ceramides (Alayoubi et al., 2013) and BMT was able to reduce both to levels found in WT mice. Finally, leukocytosis was reduced with LV injection (Alayoubi et al.,

2013), but not normalized to WT mouse levels as it was with BMT. The increased efficacy of

BMT over single-dose, neonatal LV injection may be due to the ongoing and sustained presence of normal ACDase-expressing cells, as the engrafted cells circulate and multiply.

6.4.2 Treatment limitations

Ceramide accumulation is a characteristic feature of FD, yet normalization of ceramide levels by BMT (in the organs tested here) is seemingly not sufficient to cure FD.

This suggests that other organs may still be ceramide sinks or that the effects of ACDase deficiency and/or ceramide accumulation from early in life (before the transplant) create long-term pernicious effects that are not overcome by BMT or that non-ceramide-related aspects of ACDase deficiency perpetuate FD. Indeed, treatment of FD patients in the clinic is often delayed. Diagnosis is difficult, as it is an ultra-rare disease and not thought of immediately, or it is misdiagnosed as JIA (Schuchman, 2014). Following diagnosis, treatment of Farber patients by HSCT may be further complicated by difficulty in finding appropriate donors.

BMT was not successful in all of the recipient animals. This may be due to technical issues. It may also be due in part to the Farber mice being in a mixed background; i.e. not all offspring are syngeneic. Indeed there was a similar decline in survival in Hom(WT) and

Ctrl(WT) mice around 6 months. There were, however, slightly more Hom(WT) than

Ctrl(WT) mice that had a failed BMT. This may implicate other reasons beyond non- syngeneity in the failure to engraft, such as the long-lasting effects of ACDase deficiency.

215

One problem may be that the Hom BM has difficulty supporting the WT donor cells. Along these lines we previously showed that the BM architecture is disrupted in FD (Dworski et al.,

2015).

6.4.3 Cause of death

Considering that the primary cause of death of untreated Hom Farber mice is unknown, we cannot ascribe a cause of death to the transplanted Hom(WT) mice. We can, however, eliminate certain manifestations of FD that were corrected by BMT as causes of death. Death was not due to peripheral symptoms, such as leukocytosis, as BMT-treated

Farber mice had this corrected. Another major challenge to overcome in the treatment of FD is macrophage/microglia reduction. In our transplanted Hom(WT) Farber mice, there was a reduction in macrophages/microglia in the brain and peripheral organs yet it was not normalized. It is unknown whether the macrophage accumulation seen in Farber mice is from tissue-resident macrophages, circulating monocytes, or a combination of both. In the reverse

BMT experiment, where Hom Farber mouse donor cells were transplanted into WT murine recipients, foamy macrophages were no longer seen in the organs (Dworski et al., 2015).

Here, where WT donor cells were transplanted into Hom Farber mouse recipients, we observed a similar result. Together, these data suggest that the presence of WT ACDase is sufficient to prevent foamy cell formation, yet it is not sufficient to normalize macrophage quantity and/or localization.

We also identified a potential mechanism through which monocyte accumulation is occurring. In untreated Hom mice, the circulating levels of MCP-1 and MIP-1a, known monocyte chemoattractants, are higher than in WT mice, and monocyte accumulation is high

216 in organs. With BMT treatment, the levels of these cytokines are normalized, monocyte accumulation is greatly reduced, and foamy cell formation is abolished. An additional cytokine that was elevated in untreated Farber mice is IL-1a. One role of IL-1a is to increase blood neutrophils. Following BMT, IL-1a levels and neutrophil levels were both normalized.

This study demonstrated the many positive effects of HSCT treatment of FD for the first time in a mouse model. Farber mice that received HSCT had longer and healthier lives than those that were untreated, or than those previously treated with single-dose LV injection. Should the results of this study be translated to the human disorder, transplant success could be measured using the same parameters in HSCT-treated Farber patients.

Furthermore, the increased understanding of the disease is informative for the development of advanced therapies for FD.

217

Chapter 7 General Discussion

7.1 Evaluation of hypotheses and success of aims

7.1.1 Aim 1

Hypothesis: Hematopoietic and neurological tissues are negatively affected by FD.

We identified robust differences in the hematopoietic and nervous systems of mice with FD compared to WT and Het mice. There were many diverse effects of ceramide accumulation and/or ACDase deficiency, but also some common themes.

Accumulated, foamy macrophages are prominent in the brain and hematopoietic- associated organs, and may physically disrupt the parenchyma. The amount of macrophages present in organs increases as the mice age.

Along with the macrophages, other signs of FD worsen with age. Loss of B and T progenitor cells begins at 5 weeks and steeply declines as the mice age, and the physical abilities of Hom mice (such as grip strength) decline over time. The phenotype of FD is progressive in both the hematopoietic and nervous systems.

It was surprising the macrophage progenitor cells in the BM (CFU-GM and CFU-M) appear in normal numbers in Hom mice at all ages. This diminishes the possibility that the excess macrophages originate from dysfunctional BM.

A limitation of these studies is the variability seen in mice. In addition to inherent biological variability, the age of onset of the symptoms in Hom mice is between 5 and 7

218 weeks, and mice at 7 weeks may have severe disease and are nearly moribund, or may just be starting their decline.

A further limitation is the inability to differentiate between macrophages and microglia in the brain. Both are CD68+ and it is unclear whether one or both cell types are contributing to the Farber phenotype.

7.1.2 Aim 2

Hypothesis: FD can be diagnosed by circulating cytokines and ceramides.

We identified a set of cytokines that were characteristic of FD, but not a set of ceramides. The cytokine set could also distinguish patients with FD from those with JIA and

Gaucher disease, and were no longer characteristic of FD once the patients received HSCT.

It was surprising that there was not a specific ceramide accumulation pattern in the

PB of all Farber patients. In Alzheimer’s disease, where ceramide metabolism is not the defining feature, elevated serum ceramides may be used as a biomarker (Mielke et al., 2012).

It is possible that blood is not the appropriate tissue to target for ceramide pattern analysis. When collecting blood samples, we did not control for dietary intake (a high-fat diet leads to increased ceramide in plasma and adipose tissue), time of day of sample collection, and whether the donor was fasting or not, all of which may affect the fats present in the blood

(Shah et al., 2008; Nigam, 2011). In mice, all mice eat the same chow so the effect of a high- fat diet can be excluded. The time of day and the quantity of food eaten was not controlled in mice in this study.

219

Another possibility is that plasma may not have enough ceramide accumulation to be indicative of FD. The fold elevation of each ceramide species in the Farber patients compared to controls were not as drastic as measures of ceramide in other organs (brain

(Chapter 4), spleen, and liver (Alayoubi et al., 2013).

A major limitation of this study are the number of Farber patients included. For such a rare disease, the number of samples included is impressive and rivals the number of patients collected in the largest report to date on Farber patients (Bashyam et al., 2014).

Inclusion of additional patients would give more power to the study and allow for additional analysis, such as correlating a specific ASAH1 mutation with phenotype.

7.1.3 Aim 3

Hypothesis: Consistent, systemic availability of ACDase will improve FD signs in mice.

The presence of functional ACDase expressed by WT cells in Hom(WT) mice increased the lifespan of the treated mice to double that of untreated Hom mice. Many features of the disease were also improved. However, FD was not cured and the mice still died at a premature age. Because the cause of death of untreated Hom mice is unknown, we cannot know if Hom(WT) mice are dying for the same reason.

Similar to the results of the previous aim, we found that plasma cytokines are elevated in humans and mice with FD, and are normalized following HSCT. This provides evidence that the mechanism of action for the onset of phenotypes of FD may be shared between humans and mice, and that the two species can be compared.

Because HSCT is already performed on patients with FD, a secondary aim of performing HSCT on mice was to identify the specific benefits and short-falls of

220 transplantation. This analysis revealed that transplantation is able to significantly reduce macrophage accumulation in many organs, which is likely responsible for the prevention of hepatosplenomegaly. However, it is not able to correct the BM and thymus enough to fully restore B and T progenitor cells in all mice, nor completely prevent hydrocephaly in the brains of transplanted mice.

One limitation of the BMT experiment is that we did not measure the levels of

ACDase circulating in the plasma or present in organs. As a surrogate, we identified whether any functional ACDase was present by quantifying the amount of ceramide present in the liver and brain.

Another limitation of the study was that many Hom mice transplanted with WT BM died shortly after receiving the transplant. Their deaths were within the expected timeframe for mice whose transplants do not engraft. These deaths reduced the power of the study. It is unknown whether the engraftment failure is due to the mixed background of the mice, or due to a perturbation in the BM environment of Hom mice that makes it difficult for WT BM to engraft.

7.2 Insights into the biology of Farber disease

7.2.1 Onset of observable phenotype after weaning

An engaging feature of the mouse model of FD is that, of the parameters tested to date, Farber mice appear normal until they are weaned. Currently, the only physical feature that can be used to distinguish young Farber mice from the WT and Het littermates is the stiffness of their neck skin when scruffed - Farber mice have tighter skin, even at 3 weeks of age. This tightness continues and worsens as the mice age (Lopez-Vasquez et al., 2016).

221

There are at least two hypotheses for the lack of general phenotype until weaning.

The first is that brief ceramide accumulation (up to 3 weeks) is not drastic enough to affect most organs, and it is only after 5-7 weeks that it becomes problematic (a timeline seen in hematopoietic cells, Chapter 3).

Another explanation is that before being weaned at 3 weeks of age, the mice are being nursed by their mother. The dam may be transferring functional ACDase that is being absorbed and used. Support for this hypothesis is that acid sphingomyelinase has been reported to be transferred from mother to offspring in tsetse flies (Benoit et al., 2012). Mice can also incorporate SM into their skin from cow’s milk that is received orally (Haruta-Ono et al., 2012). This hypothesis could be explored by examining the proteins in mouse breast milk.

7.2.2 Variability in affected organs

Not all organs in the Farber mouse appear to be affected. Hematopoietic-associated organs are particularly impacted by FD, such as the spleen, liver, and thymus (Chapter 3), and, to a lesser extent, the brain (Chapter 4) and lungs (data not shown). In contrast, the kidneys and heart appear unaffected, as they are not infiltrated by macrophages (Alayoubi et al., 2013) and are not enlarged, though functional studies have not yet been performed.

The reason why some organs are affected while others are not may be explained by variable ACDase demand. One hypothesis is that the organs with the greatest ACDase demand are those most affected by ACDase deficiency. In humans, the tissues with the highest ACDase mRNA expression are the heart and kidney (Li et al., 1999), which directly contradicts this hypothesis (the other organs tested were placenta, lung, liver, skeletal muscle,

222 and pancreas). In mice, the organs with the highest ACDase mRNA expression are the kidney and brain (Li et al., 1998), which partially rejects this hypothesis (the other organs tested were heart, spleen, lung, liver, skeletal muscles, and testes). This difference in ACDase mRNA expression between mice and humans may be because the regulatory region upstream of the gene varies between the two species, despite the ASAH1 coding sequence being very similar (Li et al., 1999). Despite this variability, the kidney, heart, and brain have the greatest

ACDase mRNA expression yet are not the most affected organs in FD.

The opposite hypothesis can also be proposed: organs with the greatest ACDase demand are those least affected by ACDase deficiency. Organs with the greatest ACDase demand naturally produce more ACDase, and in FD, even though this ACDase has reduced functionality, these organs still produce more ACDase overall, resulting in a greater total enzyme output.

To support this hypothesis, in Farber mice the organs with the greatest ACDase demand had the smallest reduction in ACDase activity and the lowest amount of ceramide accumulation. ACDase activity in Hom brain and heart was half of that in WT brain and heart, while in the liver, spleen, and thymus the reduction in activity was more than half

(Alayoubi et al., 2013) (ACDase activity in the kidney was not performed). Furthermore, in

Hom mice the organs with the least ceramide accumulated are the brain (~2x), heart (~4x), and kidney (~7x), compared to the lung (~35x), liver (~35x), and spleen (~100x). (Alayoubi et al., 2013).

The heart, kidney, and brain may be the least affected because they more easily pass the threshold of ACDase activity needed for daily function. The exact threshold of activity is unclear, except that it is below 50% of wild-type levels. In all of our analyses Het mice are

223 similar to WT mice and do not exhibit ceramide accumulation (Alayoubi et al., 2013). Het

ACDase knock-out mice were also mostly normal (Li et al., 2002). Similarly, in humans no adverse events have been reported in heterozygous patients (i.e. parents of Farber patients).

7.2.3 Inflammation

As ceramide is ubiquitously expressed in all cells, understanding its homeostasis is critical to understanding the etiology of diseases linked to ceramide metabolism deficits, including FD, but extending to normal development, cancer, obesity, diabetes, inflammation, and neurodegenerative disorders.

The inflammation seen in the joints of patients with FD may be initiated by ceramide signaling (Ehlert et al., 2007). In arthritis, Fas (CD95) signaling plays a role in inflammation

(Tu-Rapp et al., 2004; Ma et al., 2004). Ceramide may be increasing this signaling in FD by helping lipid rafts form that contain and activate Fas (Cremesti et al., 2001; Grassme et al.,

2001). The ceramide in the cell membrane is normally formed by sphingomyelinase activity

(Cremesti et al., 2001), but if excess ceramides accumulate in the membranes of patients and mice with FD, aberrant lipid raft formation may occur.

Similarly, transcriptional activation of Fas ligand (FasL, CD95L) can be stimulated by IL-6, which is elevated in Farber patients (Chapter 5). This emphasizes the need for therapies that lower ceramides in joint cells as early as possible to prevent inflammation.

Additionally, it suggests that Fas signaling may be an effective target in managing inflammation in FD.

224

7.2.4 Apoptosis

Fas is also a death receptor whose activation leads to apoptosis, including in lymphocytes (Burek et al., 2001). In FD, where ceramides increase Fas signaling by helping lipid rafts form, an increased rate of apoptosis may be expected. This may be the reason for reduced T progenitor cells seen in Hom mice: normally, Fas activation removes autoreactive thymocytes during negative selection in the thymus (Castro et al., 1996). Overstimulation of

Fas signaling due to ceramide-induced lipid rafts may lead to the loss of immature T cells in the thymus of aged Farber mice.

7.2.5 Limitations of the mouse model in recapitulating FD

While our knock-in mouse model does have high fidelity to human FD, there are some features that are not recapitulated in the mouse model. This includes the cherry-red spot in the eye and subcutaneous nodules on joints.

Farber patients have a cherry-red spot in the macula of their eye. This phenotype is also seen in other LSDs (Chen et al., 2014). The explanation for Farber mice not mimicking this phenotype is simple: mice eyes do not contain a macula (Volland et al., 2015).

The reason for the absence of subcutaneous nodules in Farber mice is not clear. The lack of nodules is a limitation of the mouse model, both for understanding the disease and to assess the efficacy of treatments. Nodules are a key feature of FD that occur on a broad range of patients – from those with severe FD to those with mild FD where nodules are the only feature. Nodule reduction is also used as an indicator of HSCT treatment success in patients.

Elucidating the reason for the lack of nodules may be informative in understanding the criteria required for their formation in FD (that are missing in mice), or in understanding

225 mouse biology (where a joint structure needed for nodule formation is missing in mice, similar to the lack of macula resulting in the absence of a cherry-red spot).

7.3 Sphingolipids

7.3.1 Sphingolipid backlog

The only known pathway to eliminate sphingolipids is by irreversible degradation of

Sph-1P with Sph-1P lyase (Saba et al., 1997; Lai et al., 2016). To convert any sphingolipid into Sph-1P the sphingolipid must first be converted into ceramide (by a number of possible enzymes), then Sph (by a ceramidase), then phosphorylated to Sph-1P (by sphingosine kinase) (Ogretmen & Hannun, 2004). This places ceramidases in a central role to sphingolipid removal. Considering that in FD ACDase activity is low, this could create a backlog of all sphingolipids. Indeed, we observed an elevation of non-ceramide sphingolipids in the brain (Chapter 4).

Our data suggest that the backlog is specifically up to the ceramidase step. We saw normal levels of Sph and Sph-1P in plasma (Chapter 5), the downstream products in the pathway of sphingolipid degradation.

To help with this backlog, neutral and alkaline ceramidases may be contributing a larger role than normal to ceramide’s conversion to sphingosine. This would require intracellular transport of ceramides out of the lysosome to the plasma membrane or other organelles where these ceramidases would have optimal activity.

226

7.3.2 Ceramides in hematopoiesis

7.3.2.1 Maturity of excess circulating cells

The PB of Farber mice contains an abnormal amount of cells. Our investigations suggest that these cells are mostly mature (not immature) cells. Evidence for this is the identification of mature leukocytes, including monocytes and neutrophils, circulating in the

PB (Alayoubi et al., 2013). We did not identify B cell progenitor cells in the PB (Chapter 3).

Sphingolipids play a role in the egress of cells from the BM to the PB. Specifically,

Sph-1P and Cer-1P in the PB promote HSC recruitment from the BM (Ratajczak et al., 2010;

Juarez et al., 2011; Ratajczak et al., 2012; Kim et al., 2012), leading to the possibility that excess HSC are circulating as well in the PB. In the BM there was an increase in Lin- Sca1+ cKit+ HSC (Chapter 4) but the presence of these cells was not measured in the PB. Two pieces of data do not support the hypothesis that HSC or immature cells are circulating in the blood. First, in the PB of Farber mice, no species of Cer-1P or Sph-1P that were investigated were elevated (Chapter 5). Second, initial attempts to perform the CFC assay on mouse PB did not yield an abnormal amount of colonies (data not shown). In humans, C22:1-Cer-1P was elevated in the PB, but the presence of HSC in the PB could not be assessed.

7.3.2.2 Macrophages

7.3.2.2.1 Source of macrophages

Macrophages are a key feature of FD. We saw their massive accumulation in hematopoietic organs and the brain, resulting in a disruption in organ architecture that may

227 have led to impaired lymphocyte development, astrocytosis, and hydrocephaly (Chapter 3,

Chapter 4).

The source of the excess macrophages in the PB and organs is unclear. In the brain macrophages/microglia are perivascular, suggesting that they may have entered the brain from the circulation. Our data suggest that the macrophages are in excess due to the environment and not intrinsic factors within macrophages, though this has not been demonstrated definitively. Two opposing hypotheses regarding macrophages are presented below.

7.3.2.2.2 Macrophages change due to the environment

Our data indicate that intrinsically, macrophages appear normal. Macrophage development in the BM is not overstimulated, as the proportion of CFU-M was similar, though there was a greater overall number of myeloid progenitor cells in the BM of WT and

Hom mice (Chapter 3). When Hom BM was transplanted into a WT mouse (WT(Hom)) monocytosis was not seen in the PB, nor were excess macrophages found in organs. This suggests that the Hom monocytes are intrinsically normal, and that they accumulate in Hom mice due to its environment, but do not accumulate in a WT environment.

Environmental factors that could stimulate macrophage proliferation and migration include cytokines and ceramides. We identified cytokines that promote monocyte/macrophage recruitment that are elevated up to 10-fold in the PB of mice and humans (Chapter 5). In a related LSD, Gaucher disease, the level of one of these cytokines,

MCP-1, was only elevated 2-fold (Pavlova et al., 2011), and macrophages are still a key feature of the disease. The other cytokines that were elevated stimulate each other and MCP-

228

1, suggesting a feed-forward mechanism for cytokine production and macrophage stimulation (Chapter 5).

In addition to PB elevation, MCP-1 is also increased in tissues (Alayoubi et al.,

2013), providing an environmental signal for monocytes to enter organs. In Hom(WT) mice, the levels of these cytokines in the PB were lower than in untreated Hom mice. This may have contributed to the macrophage reduction seen in organs. Similarly, in Farber patients, following HSCT the levels of these cytokines in the PB were reduced. While macrophage presence in the PB and organs was not assessed specifically before and after HSCT in these patients, in general, HSCT is associated with a reduction in nodules, which are composed of groups of macrophages.

7.3.2.2.3 Macrophages change intrinsically with ceramide

accumulation

An alternate hypothesis is that the lipid accumulation in macrophages causes the macrophages to change. This may be due to a greater need for ACDase in macrophages, or because they take in excess lipids when they phagocytose other cells/debris. This hypothesis does not have as much support.

In a macrophage cell line, increased Cer-1P and de novo ceramide production resulted in an increase in the cytokines seen elevated in the Farber mouse (Hamada et al., 2014;

Arana et al., 2013). However, de novo ceramide is likely not a large source of the ceramide accumulation (van Echten-Deckert et al., 1997).

The knock-out mouse model of ACDase deficiency suggests that macrophages are particularly susceptible to lipid accumulation. The only notable phenotype of mice that were

229 heterozygous for ACDase deficiency, Asah1+/-, was lipid accumulation in liver cells, most significantly in Kupffer cells, the tissue-resident macrophages of the liver (Li et al., 2002).

This may suggest that macrophage-like cells are the most susceptible to lipid accumulation.

7.3.3 Ceramides in the nervous system

We identified lipid accumulation and its effects at various levels of magnification of the CNS and PNS, and linked a physical cell characteristic with its resulting phenotype.

7.3.3.1 Effects on development

Our initial analyses do not suggest that ACDase deficiency affects brain development.

The general structures in the brain were present and appeared to be of the appropriate size and location (Chapter 4). Only minimal signs of apoptosis were seen. This is surprising, given the important role of exact ceramide balance during development.

7.3.3.2 Effects in adulthood

At the individual cell level, accumulations could be seen by EM in neurons, glia, and other cells of the brain. The species of lipids that form these accumulations could not be identified with the tools used. There may be different groups of lipids accumulating to form different structures, including dark Farber bodies and light vacuole-like structures.

The likely effects of this accumulation on the individual cell level could be seen in choroid plexus cells, Schwann cells, and macrophages. The extra pressure due to accumulations on choroid plexus cells may cause them to produce excess amounts of CSF,

230 leading to the hydrocephaly seen in Farber mouse brains. Schwann cells with lipid inclusions pinch the axon that they surround, which may result in a reduction in nerve conduction velocity, and may cause the poor grip and strength phenotype seen in Farber mice.

Interestingly, in Farber mice there was no evidence of demyelination, which occurs in other LSDs, such as MLD (Thomas et al., 1977), Krabbe disease (Dunn et al., 1969), and in some instances of Fabry disease (Sheth & Swick et al., 1980). Demyelination is a common cause of impaired movement.

Given the role of ceramide in promoting apoptosis and autophagy, including in neurons (Tong & de la Monte, 2009), apoptosis in the brain of Farber mice was less pronounced than may have been expected. C16:0-Cer and dhC16:0-Cer, species involved in apoptosis and autophagy (Grosch et al., 2012; Cruickshanks et al., 2015; Signorelli et al.,

2009) were elevated in the brains of mice with FD, but only mild indications of apoptosis were seen in neurons.

Increases in C16 ceramides and decreases in C18 ceramides are implicated in cerebellar PC neurodegeneration in the flincher mouse (Zhao et al., 2011), yet in our mouse we did not see significant PC loss, perhaps due to an elevation instead of a reduction of C18.

7.3.3.3 Comparison of nervous system dysfunction to another rodent

model

In another rodent model, poor activity and learning were also seen following increased ceramide exposure (de la Monte et al., 2010). In that study, rat pups were exposed to a ceramide analogue of C2-Cer for the first week of their lives, and in the first few weeks of life performed poorly on the accelerating rotarod test (de la Monte et al., 2010), similar to

231 the Farber mice. They also showed impaired learning on the Morris Water maze test (de la

Monte et al., 2010), while Farber mice did not show impaired learning on the rotarod.

However, in this model, the rats treated with the ceramide analog maintained a normal body and brain weight, and showed a reduction brain lipid despite an increase in ceramides in the brain (de la Monte et al., 2010).

Differences between these two models may be due to differences in the ceramides: the ceramide analog has an extremely short fatty chain length that does not occur naturally and may be able to cross the blood brain barrier. The exposure to excess ceramide was for 1 week immediately after birth and then ended. We have not measured the level of ceramides in neonatal Farber mice and do not know if it accumulates that early.

7.4 Toward improved diagnosis

With the reduced cost of whole-genome sequencing and the ability of the public to pay for their own genetic sequencing there will be new opportunities to identify carriers of rare diseases. This will likely increase the number of LSDs diagnosed in utero or postnatally.

It will be important to accurately identify a mutation in an LSD-related gene as disease causing or benign to help prepare parents with the need for early intervention treatments, or inform them if they choose to terminate the pregnancy. For example, there are three types of Gaucher disease that vary in their severity, including a very mild form which is often undiagnosed due to its mild presentation even in adults. In an Israeli screening program, 84% of the identified carrier mutations would result in mild or asymptomatic

Gaucher children (Eitan et al., 2010). Similarly, when screening high-risk populations for

232

Fabry disease, new mutations were identified and it was unclear whether they would be disease-causing or not (van der Tol et al., 2014).

7.4.1 Position of FD among other LSDs

FD is a LSD that shares similarities with other disorders in the family – children are affected, macrophages are a key feature, hepatosplenomegaly occurs, and neurological symptoms may occur. After describing the first three patients with FD, Dr. Farber suggested,

“this disease may prove to be a bridge between those such as Niemann-Pick and Gaucher disease, which appear to be true disturbances of intracellular lipid metabolism, and those such as Hand Schüller-Christian disease, which are granulomatous and apparently not primary disturbances of metabolism.” (Farber, 1952) We showed that one difference between

FD and Gaucher disease, is in the plasma (Chapter 5). The cytokine profiles of the two diseases shared some common cytokines that are elevated, but not the same degrees.

7.4.2 Misdiagnosed patients

This is an exciting time for FD. As exome sequencing continues to become more affordable, physicians are utilizing it more frequently to screen atypical patients in various diseases to identify the genetic cause of their disease. This has already resulted in two new forms of ACDase deficiency being described in the past 5 years: SMA-PME (Zhou et al.,

2012) and peripheral osteolysis (Bonafe et al., 2016). FD will need to be reclassified as the diverse diseases due to ACDase deficiency are revealed.

In addition to the potential discovery of new ACDase-caused disorders, additional patients in each disorder may be identified through screening. The ability of MCP-1 and

233 other blood cytokines identified in Chapter 5 to differentiate between FD and JIA is a useful tool to easily screen populations of JIA patients for undiscovered FD patients.

It is also possible that the Farber patients misdiagnosed as having JIA may not have classical FD. In the only case report of peripheral osteolysis due to ASAH1 mutation in 3 siblings, the disease presented as rheumatoid arthritis in adolescence, similar to a JIA diagnosis in adolescence (Bonafe et al., 2016). If the cytokine panel developed in Chapter 5 is applied to JIA population to identify misdiagnosed Farber patients, it is likely that a set of patients with a milder form of FD will be found.

7.4.3 Improved prognostication of disease

Due to the limited patient population, there are no reported correlations between

ASAH1 mutation and phenotype (Levade et al., 1995). As more patients will be discovered – through exome sequencing, screening of JIA populations, discovery of patients in other disease categories such as SMA-PME and peripheral osteolysis – this situation will improve.

The ability to identify patterns in patients and categorize them will aid with treatment decision-making. The only pattern identified is a negative logarithmic correlation between sphingomyelin-derived ceramide accumulated in situ and the age of patient death (Levade et al., 1995). No correlation was found between ceramide accumulation in cultured patient fibroblasts and residual AC activity (van Echten-Deckert et al., 1997). In the subset of patients analyzed in this thesis an additional correlation was identified: a negative correlation between the age at sample collection and the level of specific ceramide species (Chapter 5).

234

7.4.3.1 Mutation analysis

Each type of FD differs in its severity and set of symptoms. It is possible that these differences arise from a difference in how ACDase is dysfunctional. For example, one hypothesis is that all patients with SMA-PME have mutations in one region of the protein, while those with classical FD have mutations in another region.

This can be investigated by mapping patient mutations on the predicted structure of

ACDase. Substitution and deleted mutations were curated from various sources (Domain

Mapping of Disease Mutations database (Peterson et al., 2010); The UniProt Consortium,

2014; Bashyam et al., 2014; Filosto et al., 2016; Kim et al., 2016; Bonafe et al., 2016) and from the patients reported in Chapter 5. Benign natural variants were also included (Table 7).

The active site is predicted to be composed of residues C143, M161, D162, E225, and N320 in the human sequence (Alayoubi et al., 2013).

235

Protein Result Protein Result Mutation Mutation Q22H FD V198A FD H23D FD T222K FD Y36C FD R226P FD T42M SMA-PME G235R FD T42A SMA-PME V246A Benign variation A70V Benign variation R254G FD, PO, misdiagnosed JIA patient V72M Benign variation L257P FD V75Afs25 SMA-PME A277D FD V88M Benign variation P278L FD I93V Benign variation G284del SMA-PME V96del FD N320D FD V97E FD N320S FD V97G FD D331N FD D124E Benign variation R333C FD P136L FD R333G FD Y137C FD R333H FD E138V FD D348H FD K152N SMA-PME P362R FD G168W FD P362T FD W169R FD, PO K366Q FD E180K FD V369I Benign variation L182V FD X396L FD

Table 7. Known substitution and deletion mutations in ASAH1. Mutations are colour-coded by category of the result. FD, Farber disease; PO, peripheral osteolysis; SMA-PME, spinal muscular atrophy with progressive myoclonic epilepsy.

236

Figure 70. Mutation reported in ASAH1 visualized on the predicted structure of ACDase. The mutations in Table 7 are plotted on the predicted human ACDase structure (top) and the predicted mouse ACDase structure (bottom). Red amino acids are mutations reported in patients with FD, magenta are in both FD and peripheral osteolysis, light pink are in SMA-PME, and dark blue are benign variations. The alpha chain is in cyan on the left and the beta chain in green on the right. Active site residues are in yellow with pink dots in the predicted active site. The orange residue indicates a FD mutation in the active site.

237

Farber patient mutations were more likely to be found in the beta chain, while benign variations were mostly found on the alpha chain (Figure 70). Most of the Farber patient mutations were in the core of the protein (Figure 70). These mutations may change the conformation of the protein to reduce, but not completely abrogate its activity. Only one mutation was predicted to be in the active site. Active site mutations are not expected to be tolerated in an essential protein, as these are probably embryonic lethal.

Natural variants were mostly found on the alpha chain, suggesting that this chain is more tolerant to changes and mutations there are less likely to be disease-causing.

SMA-PME mutations were found on both chains. The SMA-PME patients reported to date have mutations that are different those in Farber patients. The SMA-PME mutations may preserve more of the activity of ACDase, as these patients survive into at least their twenties and do not have classical FD symptoms.

One mutation for peripheral osteolysis was also previously reported in a patient with

FD who was initially misdiagnosed as having JIA. This is notable, as it is a mutation that appears to result in a nodule phenotype that is milder than classical FD.

As additional patient mutations are identified, it may be the case that FD is like another LSD, Fabry Disease, where more than 450 unique patient mutations are scattered throughout the gene (Filoni et al., 2011).

238

7.5 Toward improved treatment

7.5.1 Importance of early intervention

We performed HSCT on mice that were 4 weeks of age. This represents a life stage where the mice are teenagers and are no longer fed milk by their mother. In patients with FD,

HSCT is performed at an earlier age, within the first few years after birth (Levade et al.,

2009). Treatment at this age was unable to cure FD, but was able to improve symptoms and extend life.

HSCT may have been more effective if it was performed earlier. The added benefits of early intervention have been seen in another LSD, Hurler syndrome (Boelens et al., 2013).

However, this requires that patients be identified early, and misdiagnosis slows this process.

7.5.1.1 Prevention of irreversible damage

Early intervention is critical for successful treatment of LSDs (Boelens et al., 2014).

The alternative screening method identified here could be used to screen for misdiagnosed

FD patients in the JIA and SMA populations. This would likely increase the number of FD cases identified, but it may not result in improved treatment. If the patient is already misdiagnosed with another disease then the symptoms have progressed and may be difficult to reverse (Kingma et al., 2015).

This is the case with Fabry disease, where patients with cardiomyopathy or early cryptogenic stroke or kidney failure of unknown cause are screened for Fabry disease

(Linthorst et al., 2010). These patients would be poor responders to ERT, an available

239 treatment for Fabry disease, because their kidneys have already been irreversibly damaged

(Rombach et al., 2013).

Some FD symptoms, such as those related to the brain or kidneys, are impossible or difficult to reverse, while others, such as organomegaly, are reversible (Ortolano et al.,

2014). In the Farber mice, BMT treatment at age 4 weeks (just after weaning, equivalent to a juvenile age) successfully reversed splenomegaly and other peripheral symptoms, but was less successful at improving neurological symptoms.

7.5.1.2 Prevention of neurological phenotype

Preventing or correcting the CNS phenotype is a problem for FD and other LSDs. Out of all LSDs, two-thirds of untreated patients will develop a brain phenotype (Wraith, 2014).

The question that is addressed in all LSDs is whether early intervention can prevent the neurological signs of disease, and if so how early must this be.

One treatment that has been attempted in infantile neuronal ceroid lipofuscinosis is direct brain injection of allogenic neural stem cells, a treatment that was safely performed

(Selden et al., 2013). This may be a helpful treatment in FD with neurological involvement.

Stem cells could be injected to replace lost neurons, and, more importantly, to replace astrocytes that support the current neurons and metabolize the accumulated substrate (since astrocytes are in an active state and overexpress GFAP). Injection of microglial progenitor cells could also help metabolize the lipid accumulation.

240

7.5.1.3 Perinatal treatment

One way to decrease the likelihood of irreversible damage is to perform a BMT earlier, such as shortly after birth. This has been successful in Krabbe disease, where asymptomatic newborns (age 12-44 days) receiving an umbilical cord blood transplant did not develop the typical neurological symptoms, but symptomatic infants (age 142-352 days) receiving the transplant did not improve (Escolar et al., 2005).

The earliest form of intervention is in utero. A method has been patented for in utero

LSD treatment where human umbilical cord blood cells are injected into the mother

(Garbuzova et al., 2009). The difficulty in treating in utero or shortly after birth is identifying patients with FD. Because the disease is so rare, realistically this intervention could only be pursued in families where a sibling has been diagnosed with FD previously, and the mother is pregnant with another baby who screens positive for FD.

241

Chapter 8 Conclusions

FD is a complex disorder. The ubiquitous ACDase deficiency results in wholesale damage, especially to the hematopoietic system. This is seen by abnormal amounts of hematopoietic cells and a magnification of cytokines. The ability to treat FD by BMT is promising. Transplantation of WT BM drastically increased the lifespan of Farber mice and relieved them of some signs of the disease. Ceramide accumulation was successfully reversed with this treatment, but not all of the effects of prolonged ceramide accumulation were repaired. Hematopoiesis and brain pathology, two gross features of FD were not completely resolved. However, cytokine signaling and macrophage accumulation were reversed.

When mice are young they appear not to be affected by the disease, but as they age the effects become pronounced. Mice with FD progressively lose their lymphocyte progenitor cells and gain foamy macrophages. These macrophages have observable effects: they disrupt organ architecture where they accumulate, including the spleen, BM, thymus,

LN, and brain. During these cell changes, cytokines are being differentially expressed that may be orchestrating these events or a result. The number and complexity of these changes make them increasingly difficult to reverse the longer they are allowed to progress.

Therefore, a treatment for FD would be more successful the earlier in age it is administered.

A key aspect of early treatment is early detection. Identification of FD is not simple, as it is a rare disease that is often not considered by physicians. To identify these patients that

242 fall through the cracks, screening of vulnerable populations could be performed. It is possible that patients with FD who were misdiagnosed as having JIA could be identified by performing cytokine analysis and treated by BMT or another upcoming therapeutic.

Surprisingly, the hallmark of ACDase deficiency – ceramide accumulation – is not a distinctive enough feature in patients with FD to diagnose them. Specific species did not accumulate in all patients. This highlights how far FD progresses beyond ceramide accumulation and into a world of physiological changes, all culminating in a severe disease.

243

Chapter 9 Future Directions

The work presented in this thesis raises many questions about the biological consequences of FD. It also develops a framework for improved treatment.

9.1 The subcellular localization of ceramide accumulation and

ACDase

Initially, ceramide accumulation in FD should theoretically be limited to the lysosome because ACDase is a lysosomal enzyme. However, the ceramide load in the lysosome may become so large that it is shuttled into other subcellular locations. Additionally, it is likely that ACDase is present and active in organelles other than the lysosome, and its deficiency results in other effects elsewhere in the cell. It is also unclear how mutations in ACDase change this subcellular distribution.

To examine the subcellular locations of ceramide and ACDase, organelle fractionation of cells from Farber patients and mice may be performed, followed by mass- spectrometry analysis of each fraction. Immunocytochemistry can also be used, though the reliability of the ACDase antibodies currently available is uncertain.

One extra-lysosomal location where ACDase is likely present is in the nucleus. This has been reported in both normal cells (melanocytes, oocytes) and abnormal cells (melanoma cells and an adrenocortical cell line) (Eliyahu et al., 2007; Lucki et al., 2012; Realini et al.,

244

2016). My attempts at identifying the subcellular localization of ACDase also suggested that

ACDase is in the nucleus, particularly in the nuclear membrane, though the fidelity of the antibody was not confirmed (data not shown). The distribution may change in

Asah1P361R/P361R cells due to the ACDase mutation, although the total amount of ACDase protein in the kidneys of these mice was similar to that in WT mice (Alayoubi et al., 2013).

There are many implications of ceramide accumulation and ACDase presence outside of the lysosome. Excess ceramides in non-lysosomal compartments could stimulate aberrant cell signaling. If ceramide accumulates in the cell membrane there is the possibility of ceramide participating in the formation and stability of lipid rafts that activate inflammatory signaling (Ehlert et al., 2007) or stimulate the apoptosis of lymphocytes (Castro et al., 1996) as discussed above. Activation of these pathways could be assessed by Western Blot analysis of key signal transduction proteins.

Because ACDase activity is reversed in pH environments closer to neutral (Gatt,

1963; Okino et al., 2003), finding ACDase in non-lysosomal compartments may suggest that it is functioning to form ceramide. Additionally, it may reveal other non-enzymatic functions of ACDase, such as transcriptional control if it is found in the nucleus. This role for ACDase has been reported in a human adrenocortical cell line where ACDase bound to a nuclear receptor and regulated nuclear lipid metabolism (Lucki et al., 2012).

9.2 Functionality of the immune system

Given the various changes in hematopoietic cells seen in Hom mice, it is possible that the ability of the immune system to fight infections is compromised in FD. We were unable

245 to investigate this directly due to the short lifespan of the Farber mice; models of in vivo immune challenge we found required several weeks, and from weaning to the earliest death there are only 4 weeks in which to test. A quicker in vivo or in vitro test is needed.

One indication that the immune system is at least somewhat functional is that adult lymphocytes responded to in vitro stimulation (Chapter 3). Next, the ability of macrophages to phagocytose bacteria in vitro should be tested. This function may be compromised in foamy macrophages, as they may be overwhelmed by the excess lipid they are already storing.

Future studies would test the ability of stimulated lymphocytes and macrophages to activate when presented with a challenge, and test their ability to express the appropriate cytokines. This experiment would also be informative in determining whether any of the excess cytokines seen in the PB are from activated cells. Our analysis of chitotriosidase indicates that macrophages are activated to a low level in the normal FD state (Chapter 5), providing an opportunity for a change to be seen following a challenge.

9.3 Investigation of the PNS and muscles

In this thesis we described the motor and behavioural problems seen in the mouse model of FD with an in-depth analysis of the brain. In patients who have neurological manifestations, these symptoms drastically impact the lives of the patients: they may be unable to ambulate on their own and require a wheelchair (Eviatar et al., 1984). Future studies could investigate whether this motor problem is due to problems with the spinal cord and PNS, including the neurons innervating muscles, the neuromuscular junctions, or the muscle fibers themselves.

246

The speed of nerve conduction velocity in Farber mice should also be assessed, as this may be reduced by the pinching of axonal bodies by inclusion-filled Schwann cells. This was the case in Farber patients where Schwann cells were seen compressing the sural nerve

(Pellissier et al., 1986; Zappatini-Tommasi et al., 1992).

Investigation of the intercostal muscles around the ribs may provide insight on the cause of death of the Farber mice – weakness in these muscles and additional features of the lung phenotype not discussed here may make the mice too weak to breathe properly and die of asphyxia.

Finally, the muscles themselves may be all or part of the motor problem. In a 29.5- year-old Farber patient, her arms and legs were atrophied and weak, but she did not appear to have any neurological deficits (good memory, normal nerve conduction velocity) (Fiumara et al., 1993). A suggestion for impaired muscle function is seen in immunohistochemical analysis of muscles from mice with FD. These mice have excess macrophages intercalating between fibers (Alayoubi et al., 2013), which may disrupt their function, as they appear to in the BM where they disrupt normal hematopoiesis (Dworski et al., 2015).

9.4 Expanding the cytokine panel to other patients

Ideally, the cytokine panel identified in this thesis would be confirmed on additional

Farber patients. Given the rarity of the disease it is unlikely that many additional patients will be identified in the near future.

This cytokine set can be tested on blood from other patients with ACDase deficiency

– those with SMA-PME and peripheral osteolysis. Expanding the analysis to include these patients would identify whether this panel is specific to patients with hematopoietic disease

247 with nodules (a more classical type of FD), or whether it is also applicable to patients with neurological-only disease.

Additionally, JIA patients could be screened for FD using this cytokine panel. In addition the panel, each patient should also be tested using the current standards of diagnosis

(ACDase enzyme assay and gene sequencing). These layers of analysis would confirm or reject the ability of the cytokine panel to distinguish between Farber and JIA patients.

9.5 Developing a gene therapy treatment combined with

HSCT

Our previous single dose gene therapy treatment of Farber mice (Alayoubi et al.,

2013) and HSCT treatment described here were both beneficial, but did not result in the normalization of mouse lifespan. Additionally, many features of the disease were not treated.

Combining these two therapies may be synergistic.

HSC would be infected with LV/ACDase, engraft into the BM of Hom mice, and produce progeny that express supraphysiological levels of ACDase (Aubourg, 2016). The daughter cells would be blood cells that circulate through the body, thereby delivering enzyme broadly, consistently, and for the rest of the life of the animal (Boelens et al., 2014).

This could be performed on WT or Hom cells, providing evidence for the use of enhanced allogenic transplantation or autologous stem cell transplantation, or favouring one cell source over the other.

This treatment may also be able to reach the brain to correct the phenotype there.

Corrected monocytes may enter the brain to deliver functional enzyme, though they do not seem to be able to reside in the brain in the long-term (Ajami et al., 2011; Ginhoux et al.,

248

2010). The excess levels of ACDase may also reach the brain in other ways, as seen in HSCT gene therapy of MLD, where functional enzyme was able to reach the brain (Biffi et al.,

2013).

A treatment that is able to penetrate the brain is needed for FD, as HSCT is currently not effective in patients that have or will develop neurological manifestations (Yeager et al.,

2000; Vormoor et al., 2004; Ehlert et al., 2006; Yeager et al., 2000; Cappellari et al., 2016).

249

References

Abbott SK, Li H, Munoz SS, Knoch B, Batterham M, Murphy KE, Halliday GM, Garner B

(2014) Altered ceramide acyl chain length and ceramide synthase gene expression in

Parkinson's disease. Mov Disord 29(4):518-26.

Abenoza P, Sibley RK (1987) Farber's disease: a fine structural study. Ultrastruct Pathol

11(4):397-403.

Adolfsson J, Borge OJ, Bryder D, Theilgaard-Monch K, Astrand-Grundstrom I, Sitnicka E,

Sasaki Y, Jacobsen SE (2001) Upregulation of Flt3 expression within the bone marrow

Lin(−)Sca1(+)c-kit(+) stem cell compartment is accompanied by loss of self-renewal

capacity. Immunity 15(4):659–69.

Ajami B1, Bennett JL, Krieger C, McNagny KM, Rossi FM (2011) Infiltrating monocytes

trigger EAE progression, but do not contribute to the resident microglia pool. Nat

Neurosci 14(9):1142-9.

Alayoubi AM, Wang JCM, Au BCY, Carpentier S, Garcia V, Dworski S, El-Ghamrasni S,

Kirouac KN, Exertier MJ, Xiong ZJ, Privé GG, Simonaro CM, Casas J, Fabrias G,

Schuchman EH, Turner PV, Hakem R, Levade T, Medin JA (2013) Systemic ceramide

accumulation leads to severe and varied pathological consequences. EMBO Mol Med

5(6):827-42.

Aldenhoven M, Boelens JJ, de Koning TJ (2008) The clinical outcome of Hurler syndrome

after stem cell transplantation. Biol Blood Marrow Transplant 14(5):485-98.

Aldenhoven M, Sakkers RJ, Boelens J, de Koning TJ, Wulffraat NM (2009) Musculoskeletal

manifestations of lysosomal storage disorders. Ann Rheum Dis 68(11):1659-65.

Antonarakis SE, Valle D, Moser HW, Moser A, Qualman SJ, Zinkham WH (1984)

250

Phenotypic variability in siblings with Farber disease. J Pediatr 104(3):406-9.

Arana L, Ordoñez M, Ouro A, Rivera I, Gangoiti P, Trueba M, Gomez-Muñoz A (2013)

Ceramide 1-phosphate induces macrophage chemoattractant protein-1 release:

involvement in ceramide 1-phosphate-stimulated cell migration. Am J Physiol Endocrinol

Metab. 304(11):E1213-26.

Aubourg P (2016) Gene Therapy for Rare Central Nervous System Diseases Comes to Age.

Endocr Dev 30:141-6.

Banugaria SG, Prater SN, Ng YK et al (2011) The impact of antibodies on clinical outcomes

in diseases treated with therapeutic protein: lessons learned from infantile Pompe disease.

Genet Med 13:729–736.

Banugaria SG, Prater SN, Patel TT et al (2013) Algorithm for the early diagnosis and

treatment of patients with cross reactive immunologic material-negative classic infantile

pompe disease: a step towards improving the efficacy of ERT. PLoS One 8, e67052.

Barton NW, Furbish FS, Murray GJ, Garfield M, Brady RO (1990) Therapeutic response to

intravenous infusions of glucocerebrosidase in a patient with Gaucher disease. Proc Natl

Acad Sci USA 87, 1913-1916.

Barton NW, Brady RO, Dambrosia JM, Di Bisceglie AM, Doppelt SH, Hill SC, Mankin HJ,

Murray GJ, Parker RI, Argoff CE, Grewal RP, Yu K (1991) Replacement therapy for

inherited enzyme deficiency—Macrophage-targeted glucocerebrosidase for Gaucher’s

disease. N Engl J Med 324, 1464-1470.

Bashyam MD, Chaudhary AK, Kiran M, Reddy V, Nagarajaram HA, Dalal A, Bashyam L,

Suri D, Gupta A, Gupta N, Kabra M, Puri RD, RamaDevi R, Kapoor S, Danda S (2014)

251

Molecular analyses of novel ASAH1 mutations causing Farber lipogranulomatosis:

analyses of exonic splicing enhancer inactivating mutation. Clin Genet 86(6):530-8.

Baudhuin P, Hers HG, Loeb H (1964) An electron microscopic and biochemical study of

type II glycogenosis. Lab Invest 13:1139-52.

Beck M, Moser HW, Sandhoff K. Chapter 35 – Acid Ceramidase Deficiency: Farber

Lipogranulomatosis and Spinal Muscular Atrophy Associated with Progressive Myoclonic

Epilepsy. Rosenberg's Molecular and Genetic Basis of Neurological and Psychiatric

Disease (5th Edition), Rosenberg RN, Pascual JM, eds. Elsevier, 2015.

Bedia C, Casas J, Andrieu-Abadie N, Fabriàs G, Levade T (2011) Acid ceramidase

expression modulates the sensitivity of A375 melanoma cells to dacarbazine. J Biol Chem

286(32):28200-9.

Benoit JB, Attardo GM, Michalkova V, Takác P, Bohova J, Aksoy S (2012)

Sphingomyelinase activity in mother's milk is essential for juvenile development: a case

from lactating tsetse flies. Biol Reprod 87(1):17, 1-10.

Bernardo K, Hurwitz R, Zenk T, Desnick RJ, Ferlinz K, Schuchman EH, Sandhoff K (1995)

Purification, characterization, and biosynthesis of human acid ceramidase. J Biol Chem.

270(19):11098-102.

Bertoncello I, Hodgson GS, Bradley TR (1985) Multiparameter analysis of transplantable

hemopoietic stem cells: I The separation and enrichment of stem cells homing to marrow

and spleen on the basis of rhodamine-123 fluorescence. Exp Hematol 13(10):999-1006.

Bhabak KP, Kleuser B, Huwiler A, Arenz C (2013) Effective inhibition of acid and neutral

ceramidases by novel b-13 and lcl-464 analogues. Bioorg Med Chem 21, 874-882.

Bieberich E (2008) Ceramide signaling in cancer and stem cells. Future Lipidol. 3(3):273-

252

300.

Bielawska A, Perry DK, Hannun YA (2001) Determination of ceramides and diglycerides by

the diglyceride kinase assay. Anal Biochem 298:141-150.

Bierman SM, Edgington T, Newcomer VD, Pearson CM (1966) A disorder of

mucopolysaccharide metabolism with articular, respiratory, and neurologic

manifestations. Arthritis Rheum 9(4):620-30.

Biffi A, Montini E, Lorioli L, Cesani M, Fumagalli F, Plati T, Baldoli C, Martino S, Calabria

A, Canale S, Benedicenti F, Vallanti G, Biasco L, Leo S, Kabbara N, Zanetti G, Rizzo

WB, Mehta NA, Cicalese MP, Casiraghi M, Boelens JJ, Del Carro U, Dow DJ, Schmidt

M, Assanelli A, Neduva V, Di Serio C, Stupka E, Gardner J, von Kalle C, Bordignon C,

Ciceri F, Rovelli A, Roncarolo MG, Aiuti A, Sessa M, Naldini L (2013) Lentiviral

hematopoietic stem cell gene therapy benefits metachromatic leukodystrophy. Science

341(6148):1233158.

Bikman BT (2012) A role for sphingolipids in the pathophysiology of obesity-induced

inflammation. Cell Mol Life Sci. 69(13):2135-46.

Birbes H, El Bawab S, Hannun YA, Obeid LM (2001) Selective hydrolysis of a

mitochondrial pool of sphingomyelin induces apoptosis. FASEB J 15(14):2669-79.

Blitzer MG, McDowell GA (1992) Tay-Sachs disease as a model for screening inborn errors.

Clin Lab Med 463-80.

Boelens JJ, Prasad VK, Tolar J, Wynn RF, Peters C (2010) Current international perspectives

on hematopoietic stem cell transplantation for inherited metabolic disorders. Pediatr Clin

North Am 57(1):123-45.

Boelens JJ, Aldenhoven M, Purtill D, Ruggeri A, Defor T, Wynn R, Wraith E, Cavazzana-

253

Calvo M, Rovelli A, Fischer A, Tolar J, Prasad VK, Escolar M, Gluckman E, O'Meara A,

Orchard PJ, Veys P, Eapen M, Kurtzberg J, Rocha V; Eurocord; Inborn Errors Working

Party of European Blood and Marrow Transplant group; Duke University Blood and

Marrow Transplantation Program; Centre for International Blood and Marrow Research

(2013) Outcomes of transplantation using various hematopoietic cell sources in children

with Hurler syndrome after myeloablative conditioning. Blood 121(19):3981-7.

Boelens JJ, Orchard PJ, Wynn RF (2014) Transplantation in inborn errors of metabolism:

current considerations and future perspectives. Br J Haematol 167(3):293-303.

Boelens JJ, van Hasselt PM (2016) Neurodevelopmental Outcome after Hematopoietic Cell

Transplantation in Inborn Errors of Metabolism: Current Considerations and Future

Perspectives. Neuropediatrics 47(5):285-92.

Bonafé L, Kariminejad A, Li J, Royer-Bertrand B, Garcia V, Mahdavi S, Bozorgmehr B,

Lachman RL, Mittaz-Crettol L, Campos-Xavier B, Nampoothiri S, Unger S, Rivolta C,

Levade T, Superti-Furga A (2016) Brief Report: Peripheral Osteolysis in Adults Linked to

ASAH1 (Acid Ceramidase) Mutations: A New Presentation of Farber's Disease. Arthritis

Rheumatol 68(9):2323-7.

Boot RG, Renkema GH, Strijland A, van Zonneveld AJ, Aerts JM (1995) Cloning of a

cDNA encoding chitotriosidase, a human chitinase produced by macrophages. J Biol

Chem 270(44):26252-6.

Brady RO, Pentchev PG, Gal AE, Hibbert SR, Dekaban AS (1974) Replacement therapy for

inherited enzyme deficiency. Use of purified glucocerebrosidase in Gaucher’s disease. N

Engl J Med 291, 989-993.

254

Breen C, Wynn RF, O'Meara A, O'Mahony E, Rust S, Imrie J, Wraith JE (2013)

Developmental outcome post allogenic bone marrow transplant for Niemann Pick Type

C2. Mol Genet Metab 108(1):82-4.

Brooks DA, Muller VJ, Hopwood JJ (2006) Stop-codon read-through for patients affected by

a lysosomal storage disorder. Trends Mol Med 12(8):367-73.

Bryder D, Rossi DJ, Weissman IL (2006) Hematopoietic stem cells: the paradigmatic tissue-

specific stem cell. Amer J Pathol 169(2):338-46.

Burck U, Moser H, Goebel H, Grüttner R, Held K (1985) A case of lipogranulomatosis

Farber: some clinical and ultrastructural aspects. Eur J Pediatr 143(3):203-8.

Burek C, Roth J, Koch HG, Harzer K, Los M, Schulze-Osthoff K (2001) The role of

ceramide in receptor- and stress-induced apoptosis studied in acidic ceramidase-deficient

Farber disease cells. Oncogene 20(45):6493-502.

Burke W, Tarini B, Press NA, Evans JP (2011) Genetic Screening. Epidemiol Rev

33(1):148-64.

Cappellari AM, Torcoletti M, Triulzi F, Corona F. (2016) Nervous system involvement in

Farber disease. J Inherit Metab Dis 39(1):149-50.

Carlson CG, Rutter J, Bledsoe C, Singh R, Hoff H, Bruemmer K, Sesti J, Gatti F, Berge J,

McCarthy L (2010) A simple protocol for assessing inter-trial and inter-examiner

reliability for two noninvasive measures of limb muscle strength. J Neurosci Methods

186(2):226-30.

Castro JE, Listman JA, Jacobson BA, Wang Y, Lopez PA, Ju S, Finn PW, Perkins DL (1996)

Fas modulation of apoptosis during negative selection of thymocytes. Immunity 5(6):617-

27.

255

Chatelut M, Feunteun J, Harzer K, Fensom AH, Basile JP, Salvayre R, Levade T (1996) A

simple method for screening for Farber disease on cultured skin fibroblasts. Clin Chim

Acta 245(1):61-71.

Chedrawi AK, Al-Hassnan ZN, Al-Muhaizea M, Colak D, Al-Younes B, Albakheet A, Tulba

S, Kaya N (2012). Novel V97G ASAH1 mutation found in Farber disease patients: unique

appearance of the disease with an intermediate severity, and marked early involvement of

central and peripheral nervous system. Brain Dev 34(5):400-4.

Chen CS, Patterson MC, Wheatley CL, O'Brien JF, Pagano RE (1999) Broad screening test

for sphingolipid-storage diseases. Lancet. 354(9182):901-5.

Chen CZ, Li M, de Graaf D, Monti S, Gottgens B, Sanchez MJ, Lander ES, Golub TR, Green

AR, Lodish HF (2002) Identification of endoglin as a functional marker that defines long-

term repopulating hematopoietic stem cells. Proc Natl Acad Sci USA 99(24):15468-73.

Chen H, Chan AY, Stone DU, Mandal NA (2014) Beyond the Cherry-Red Spot: Ocular

Manifestations of Sphingolipid-mediated Neurodegenerative and Inflammatory Disorders.

Survey of ophthalmology 59(1):10.

Choi MJ, Maibach HI (2005) Role of ceramides in barrier function of healthy and diseased

skin. Am J Clin Dermatol 6(4):215-23.

Christensen JL, Weissman IL (2001) Flk-2 is a marker in hematopoietic stem cell

differentiation: a simple method to isolate long-term stem cells. Proc Natl Acad Sci USA

98(25):14541-46.

Clausen J, Rampini S (1970) Chemical studies of Farber's disease. Acta Neurol Scand

46(3):313-22.

Cogan DG, Kuwabara T, Moser H, Hazard GW (1966) Retinopathy in a case of Farber's

256

lipogranulomatosis. Arch Ophthalmol 1966 75(6):752-7.

Coutinho MF, Santos JI, Alves S (2016) Less Is More: Substrate Reduction Therapy for

Lysosomal Storage Disorders. Int J Mol Sci 17(7): pii: E1065.

Cremesti A, Paris F, Grassme H, Holler N, Tschopp J, Fuks Z, Gulbins E, Kolesnick R

(2001) Ceramide enables fas to cap and kill. J Biol Chem 276(26):23954-61.

Cruickshanks N, Roberts JL, Bareford MD, Tavallai M, Poklepovic A, Booth L, Spiegel S,

Dent P (2015) Differential regulation of autophagy and cell viability by ceramide species.

Cancer Biol Ther 16(5):733-42.

Cvitanovic-Sojat L, Hajnzic TF, Fumic K et al.: (1997) Farber lipogranulomatosis type 5:

Report of a new case (abstract). International Congress on Inborn Errors of Metabolism

245.

Cvitanovic-Sojat L, Gjergja Juraski R, Sabourdy F, Fensom AH, Fumic K, Paschke E,

Levade T (2011) Farber lipogranulomatosis type 1-late presentation and early death in a

Croatian boy with a novel homozygous ASAH1 mutation. EJPN 15(2):171-3.

Dahms NM, Lobel P, Kornfeld S (1989) Mannose 6-phosphate receptors and lysosomal

enzyme targeting. J Biol Chem 264:12115-18. de Duve C (1964) From cytases to lysosomes. Fed Proc 23:1045-9. de la Monte SM, Tong M, Nguyen V, Setshedi M, Longato L, Wands JR (2010) Ceramide-

mediated insulin resistance and impairment of cognitive-motor functions. J Alzheimers

Dis 21(3):967-84. de la Monte SM, Tong M (2014) Brain metabolic dysfunction at the core of Alzheimer's

disease. Biochem Pharmacol 88(4):548-59.

Desnick RJ, Ioannou YA, Eng CM. α-Galactosidase A deficiency: Fabry disease. In: Scriver

257

CR, Beaudet AL, Sly WS, Valle D, editors. The metabolic and molecular bases of

inherited disease. 8th edn. New York: McGraw-Hill; 2001. p. 3733-74.

Devi ARR, Gopikrishna M, Ratheesh R, Savithri G, Swarnalata G, Bashyam M (2006)

Farber lipogranulomatosis: clinical and molecular genetic analysis reveals a novel

mutation in an Indian family. J Hum Genet 51(9):811-4.

Diamond EL, Dagna L, Hyman DM, Cavalli G, Janku F, Estrada-Veras J, Ferrarini M,

Abdel-Wahab O, Heaney ML, Scheel PJ, Feeley NK, Ferrero E, McClain KL, Vaglio A,

Colby T, Arnaud L, Haroche J (2014) Consensus guidelines for the diagnosis and clinical

management of Erdheim-Chester disease. Blood 124(4):483-92.

Drobnik W, Liebisch G, Audebert F, Fröhlich D, Glück T, Vogel P, Rothe G, Schmitz G

(2003) Plasma ceramide and lysophosphatidylcholine inversely correlate with mortality in

sepsis patients. J Lipid Res 44(4):754-61.

Dunn HG, Lake BD, Dolman CL, Wilson J (1969) The neuropathy of Krabbe's infantile

cerebral sclerosis (globoid cell leucodystrophy). Brain 92(2):329-44.

Dworski S, Berger A, Furlonger C, Moreau JM, Yoshimitsu M, Trentadue J, Au BC, Paige

CJ, Medin JA (2015) Markedly perturbed hematopoiesis in acid ceramidase deficient

mice. Haematologica 100(5):e162-5.

Dworski S, Lu P, Khan A, Maranda B, Mitchell JJ, Parini R, Di Rocco M, Hugle B,

Yoshimitsu M, Magnusson B, Makay B, Arslan N, Guelbert N, Ehlert K, Jarisch A,

Gardner-Medwin J, Dagher R, Terreri MT, Lorenco CM, Barillas-Arias L, Tanpaiboon P,

Solyom A, Norris JS, He X, Schuchman EH, Levade T, Medin JA (2017) Acid

Ceramidase Deficiency is Characterized by a Unique Plasma Cytokine and Ceramide

Profile that is Altered by Therapy. Biochim Biophys Acta 1863(2):386-94.

258

Dyment DA, Sell E, Vanstone MR, Smith AC, Garandeau D, Garcia V, Carpentier S, Le

Trionnaire E,Sabourdy F, Beaulieu CL, Schwartzentruber JA, McMillan HJ; FORGE

Canada Consortium, Majewski J,Bulman DE, Levade T, Boycott KM (2014) Evidence for

clinical, genetic and biochemical variability in spinal muscularatrophy with progressive

myoclonic epilepsy. Clin Genet 86(6):558-63.

Ehlert K, Roth J, Frosch M, Fehse N, Zander N, Vormoor J (2006) Farber's disease without

central nervous system involvement: bone-marrow transplantation provides a promising

new approach. Ann Rheum Dis 65(12):1665-6.

Ehlert E, Frosch M, Fehse N, Zander A, Roth J, Vormoor J (2007) Farber disease: clinical

presentation, pathogenesis and a new approach to treatment. Pediatr Rheumatol Online J

5:15.

Eitan Y, Abrahamov A, Phillips M, Elstein D, Zimran A (2010) Sixteen years of prenatal

consultations for the N370S/N370S Gaucher disease genotype: what have we learned?

Prenat Diagn 30(10):924-7.

Ekici B, Kürkçü D, Çaliskan M (2012) Farber disease: A clinical diagnosis. J Pediatr

Neurosci 7(2):154–5.

Eliyahu E, Park JH, Shtraizent N, He X, Schuchman EH (2007) Acid ceramidase is a novel

factor required for early embryo survival. FASEB J 21(7):1403-9.

Eliyahu E, Shtraizent N, Shalgi R, Schuchman EH (2012) Construction of conditional acid

ceramidase knockout mice and in vivo effects on oocyte development and fertility. Cell

Physiol Biochem 30(3):735-48.

Erfan M, Haque AU, Ahmed SA (2015) Farber’s disease: a case report. Int J Pathol

13(3):115-9.

259

Escolar ML, Poe MD, Provenzale JM, Richards KC, Allison J, Wood S, Wenger DA,

Pietryga D, Wall D, Champagne M, Morse R, Krivit W, Kurtzberg J (2005)

Transplantation of umbilical-cord blood in babies with infantile Krabbe's disease. N Engl

J Med 352(20):2069-81.

Eviatar L, Sklower SL, Wisniewski K, Feldman RS, Gochoco A (1986) Farber

lipogranulomatosis: an unusual presentation in a black child. Pediatr Neurol 2(6):371-4.

Fahy E, Subramaniam S, Murphy RC, Nishijima M, Raetz CR, Shimizu T, Spener F, van

Meer G, Wakelam MJ, Dennis EA (2009) Update of the LIPID MAPS comprehensive

classification system for lipids. J Lipid Res 50 Suppl:S9-14.

Farber S (1952) A lipid metabolic disorder: disseminated lipogranulomatosis; a syndrome

with similarity to, and important difference from, Niemann-Pick and Hand-Schüller-

Christian disease. AMA Am J Dis Child. 84(4):499-500 (Abstract). In: Society

Transactions, American Pediatric Society, Hugh McCulloch, M.D., President, 62nd Annual

Meeting, Old Point Comfort, Va., May 7, 8, and 9, 1952.

Farber S, Cohen J, Uzman LL (1957) Lipogranulomatosis; a new lipo-glycoprotein storage

disease. J Mt Sinai Hosp NY 24(6):816-37.

Feingold KR (2007) The role of epidermal lipids in cutaneous permeability barrier

homeostasis. J Lipid Res 48(12):2531-46.

Ferlinz K, Kopal G, Bernardo K, Linke T, Bar J, Breiden B, Neumann U, Lang F,

Schuchman EH, Sandhoff K (2001) Human acid ceramidase: processing, glycosylation,

and lysosomal targeting. J Biol Chem 276(38):35352-60.

260

Filippov V, Song MA, Zhang K, Vinters HV, Tung S, Kirsch WM, Yang J, Duerksen-

Hughes PJ (2012) Increased ceramide in brains with Alzheimer's and other

neurodegenerative diseases. J Alzheimers Dis 29(3):537-47.

Filocamo M, Morrone A (2011) Lysosomal storage disorders: molecular basis and laboratory

testing. Hum Genomics 5(3):156-69.

Filoni C, Caciotti A, Carraresi L, Cavicchi C, Parini R, Antuzzi D, Zampetti A, Feriozzi S,

Poisetti P, Garman SC, Guerrini R, Zammarchi E, Donati MA, Morrone A (2010)

Functional studies of new GLA gene mutations leading to conformational Fabry disease.

Biochim Biophys Acta 1802(2):247-52.

Filosto M, Aureli M, Castellotti B, Rinaldi F, Schiumarini D, Valsecchi M, Lualdi S,

Mazzotti R, Pensato V, Rota S, Gellera C, Filocamo M, Padovani A (2016) ASAH1

variant causing a mild SMA phenotype with no myoclonic epilepsy: a clinical,

biochemical and molecular study. Eur J Hum Genet 24(11):1578-1583.

Fiumara A, Nigro F, Pavone L, Moser HW (1993) Farber disease with prolonged survival. J

Inherit Metab Dis 16(5):915-6.

Fletcher JL, Kondagari GS, Vite CH, Williamson P, Taylor RM (2014) Oligodendrocyte loss

during the disease course in a canine model of the lysosomal storage disease fucosidosis. J

Neuropathol Exp Neurol 73(6):536-47.

Franklin KBJ, Paxinos G. The Mouse Brain In Stereotaxic Coordinates. 3rd ed. New York:

Elsevier, Inc. 2008.

Fratantoni JC, Hall CW, Neufeld EF (1969) The defect in Hurler and Hunter syndromes. II.

Deficiency of specific factors involved in mucopolysaccharide degradation. Proc Natl

Acad Sci USA 64, 360–366.

261

Frohbergh M, He X, Schuchman EH (2015) The molecular medicine of acid ceramidase.

Biol Chem 396(6-7):759-65.

Fujiwaki T, Hamanaka S, Koga M, Ishihara T, Nishikomori R, Kinoshita E, Furusho K

(1992) A case of Farber disease. Acta Paediatr Jpn 34(1):72-9.

Fuller M, Meikle PJ, Hopwood JJ (2006) Epidemiology of lysosomal storage diseases: an

overview. Oxford: Oxford PharmaGenesis. Chapter 2.

Futerman AH, Hannun YA (2004) The complex life of simple sphingolipids. EMBO Rep

5(8):777-82.

Gan JJ, Garcia V, Tian J, Tagliati M, Parisi JE, Chung JM, Lewis R, Baloh R, Levade T,

Pierson TM (2015) Acid ceramidase deficiency associated with spinal muscular atrophy

withprogressive myoclonic epilepsy. Neuromuscul Disord 25(12):959-63.

Garanto A, Mandal NA, Egido-Gabas M, Marfany G, Fabrias G, Anderson RE, Casas J,

Gonzalez-Duarte R (2013) Specific sphingolipid content decrease in Cerkl knockdown

mouse retinas. Exp Eye Res 110:96-106.

Garbuzova-Davis S, Sanberg PR, Gografe S (2009) Novel method of prenatal administration

of mammalian umbilical cord stem cells for the intrauterine treatment of mammalian

lysosomal storage disease. Patent US2009016998.

Gatt S (1963) Enzymic hydrolysis and synthesis of ceramides. J Biol Chem 238:3131–3133.

Gaucher PCE. De l'epithelioma primitif de la rate, hypertrophie idiopathique de la rate sans

leucemie. Academic thesis, Paris, France. 1882.

Gault CR, Obeid LM, Hannun YA (2010) An overview of sphingolipid metabolism: from

synthesis to breakdown. Adv Exp Med Biol 688:1-23.

Ginhoux F, Greter M, Leboeuf M, Nandi S, See P, Gokhan S, Mehler MF, Conway SJ, Ng

262

LG, Stanley ER, Samokhvalov IM, Merad M (2010) Fate mapping analysis reveals that

adult microglia derive from primitive macrophages. Science 330(6005):841-5.

Giráldez BG, Rosa Guerrero-López R, Ortega-Moreno L, Verdú A, Carrascosa-Romero MC,

García-Campos O, García-Muñozguren S, Pardal-Fernández JM, Serratosa JM (2014)

Uniparental disomy as a cause of spinal muscular atrophy and progressive myoclonic

epilepsy: phenotypic homogeneity due to the homozygous c.125C>T mutation in ASAH1.

Neuromuscul Disord 25(3):222-4.

Goodell MA, Brose K, Paradis G, Conner AS, Mulligan RC (1996) Isolation and functional

properties of murine hematopoietic stem cells that are replicating in vivo. J Exp Med

183(4):1797-806.

Gomez-Muñoz A, Gangoiti P, Arana L, Ouro A, Riveria I, Ordoñez M, Trueba M (2013)

New insights on the role of ceramide 1-phosphate in inflammation. Biochim Biophys Acta

1831(6):1060-6.

Grassme H, Jekle A, Riehle A, Schwarz H, Berger J, Sandhoff K, Kolesnick R, Gulbins E

(2001) CD95 signalling via ceramide-rich membrane rafts. J Biol Chem 276(23):20589-

96.

Grosch S, Schiffmann S, Geisslinger G (2012) Chain length-specific properties of ceramides.

Prog Lipid Res 51(1):50-62.

Guffon N, Bertrand Y, Forest I, Fouilhoux A, Froissart R (2009) Bone marrow

transplantation in children with Hunter syndrome: outcome after 7 to 17 years. J Pediatr

154(5):733-7.

Guillemin GJ, Brew BJ (2004) Microglia, macrophages, perivascular macrophages, and

pericytes: a review of function and identification. J Leukoc Biol 75(3):388-97.

263

Hammad SM, Pierce JS, Soodavar F, Smith KJ, Al Gadban MM, Rembiesa B, Klein RL,

Hannun YA, Bielawski J, Bielawska A (2010) Blood sphingolipidomics in healthy

humans: impact of sample collection methodology. J Lipid Res 51(10):3074-87.

Hamada Y, Nagasaki H, Fujiya A, Seino Y, Shang QL, Suzuki T, Hashimoto H, Oiso Y

(2014) Involvement of de novo ceramide synthesis in pro-inflammatory adipokine

secretion and adipocyte-macrophage interaction. J Nutr Biochem 25(12):1309-16.

Han X, Holtzman DM, McKeel DW, Jr. (2011) Plasmalogen deficiency in early Alzheimer's

disease subjects and in animal models: molecular characterization using electrospray

ionization mass spectrometry. J Neurochem 77(4):1168-80.

Haruta-Ono Y, Setoguchi S, Ueno HM, Higurashi S, Ueda N, Kato K, Saito T, Matsunaga K,

Takata J (2012) Orally administered sphingomyelin in bovine milk is incorporated into

skin sphingolipids and is involved in the water-holding capacity of hairless mice. J

Dermatol Sci 68(1):56-62.

Hayashida K, Nanki T, Girschick H, Yavuz S, Ochi T, Lipsky PE (2001) Synovial stromal

cells from rheumatoid arthritis patients attract monocytes by producing MCP-1 and IL-8.

Arthritis Res 3(2):118-26.

He X, Li CM, Park JH, Dagan A, Gatt S, Schuchman EH (1999) A fluorescence-based high-

performance liquid chromatographic assay to determine acid ceramidase activity. Anal

Biochem 274(2):264-9.

Hers HG (1963) Alpha-Glucosidase deficiency in generalized glycogen storage disease

(Pompe's disease). Biochem J 86:11-6.

Hers HG (1965) Inborn lysosomal diseases. Gastroenterology 48:625-33.

Hollak CE, van Weely S, van Oers MH, Aerts JM (1994) Marked elevation of plasma

264

chitotriosidase activity. A novel hallmark of Gaucher disease. J Clin Invest 93(3):1288-92.

Hong JH, Kang JW, Kim DK, Baik SH, Kim KH, Shanta SR, Jung JH, Mook-Jung I, Kim

KP (2015) Global changes of phospholipids identified by MALDI imaging mass

pectrometry in a mouse model of Alzheimer's disease. J Lipid Res 57(1):36-45.

Hopkins PV, Campbell C, Klug T, Rogers S, Raburn-Miller J, Kiesling J (2015) Lysosomal

storage disorder screening implementation: findings from the first six months of full

population pilot testing in Missouri. J Pediatr 166(1):172–177.

Hopwood JJ, Vellodi A, Scott HS, Morris CP, Litjens T, Clements PR, Brooks DA, Cooper

A, Wraith JE (1993) Long-term clinical progress in bone marrow transplanted

mucopolysaccharidosis type I patients with a defined genotype. J Inherit Metab Dis

16(6):1024-33.

Hu W, Xu R, Sun W, Szulc ZM, Bielawski J, Obeid LM, Mao C (2010) Alkaline ceramidase

3 (ACER3) hydrolyzes unsaturated long-chain ceramides, and its down-regulation inhibits

both cell proliferation and apoptosis. J Biol Chem 285(11):7964-76.

Huang Y, Tanimukai H, Liu F, Iqbal K, Grundke-Iqbal I, Gong CX (2004) Elevation of the

level and activity of acid ceramidase in Alzheimer's disease brain. Eur J Neurosci

20(12):3489-97.

Hung YH, Chen LM, Yang JY, Yang WY (2013) Spatiotemporally controlled induction of

autophagy-mediated lysosome turnover. Nat Commun 4:2111.

Hurst SM, Wilkinson TS, McLoughlin RM, Jones S, Horiuchi S, Yamamoto N (2001) IL-6

and its soluble receptor orchestrate a temporal switch in the pattern of leukocyte

recruitment seen during acute inflammation. Immunity 14(6):705-14.

Ichi I, Nakahara K, Miyashita Y, Hidaka A, Kutsukake S, Inoue K, Maruyama T, Miwa Y,

265

Harada-Shiba M, Tsushima M, Kojo S, Kisei Cohort Study Grooup (2006) Association of

ceramides in human plasma with risk factors of atherosclerosis. Lipids 41(9):859-63.

Jarisch A, Steward CG, Sörensen J, Porto L, Kieslich M, Klingebiel T, Bader P (2014)

Odontoid infiltration and spinal compression in Farber Disease: reversal by hematopoietic

stem cell transplantation. Eur J Pediatr 173(10):1399-403.

Jones E, Dworski S, Canals D, Casas J, Fabrias G, Schoenling A, Levade T, Denlinger CE,

Hannun YA, Medin JA, Drake RR (2014) On-tissue localization of ceramides and other

sphingolipids by MALDI mass spectrometry imaging. Anal Chem 86(16):8303-11.

Jones EE, Dworski S, Kamani M, Canals D, Wada M, Liu X, Hannun YA, Norris J, Medin

JA, Drake RR. Detection and distribution of sphingolipids in tissue by FTICR MALDI-

imaging mass spectrometry. In Bioactive sphingolipids in cancer biology and therapy, 1st

ed., Eds. Hannun YA, Luberto C, Mao C, Obeid LM. Switzerland: Springer International

Publishing. 2015, p339-58.

Juarez JG, Harun N, Thien M, Welschinger R, Baraz R, Dela Pena A, Pitson SM, Rettig M,

Dipersio JF, Bradstock KF, Bendall LJ (2011) Sphingosine-1- phosphate facilitates

trafficking of hematopoietic stem cells and their mobilization by CXCR4 antagonists in

mice. Blood 119:707-716.

Kaback MM, Nathan TJ, Greenwald S (1977) Tay-Sachs disease: heterozygote screening and

prenatal diagnosis--U.S. experience and world perspective. Prog Clin Biol Res 18:13-36.

Kanzaki S, Yamaguchi A, Yamaguchi K, Kojima Y, Suzuki K, Koumitsu N, Nagashima Y,

Nagahama K, Ehara M, Hirayasu Y, Ryo A, Aoki I, Yamanaka S (2010) Thymic

alterations in GM2 gangliosidoses model mice. PLoS One 5(8):e12105.

Kaplan A, Achord DT, Sly WS (1977) Phosphohexosyl components of a lysosomal enzyme

266

are recognized by pinocytosis receptors on human fibroblasts. Proc Natl Acad Sci USA

74(5):2026–30.

Karaca I, Tamboli IY, Glebov K, Richter J, Fell LH, Grimm MO, Haupenthal VJ, Hartmann

T, Gräler MH, van Echten-Deckert G, Walter J (2014) Deficiency of sphingosine-1-

phosphate lyase impairs lysosomal metabolism of the amyloid precursor protein. J Biol

Chem 289(24):16761-72.

Kattner E, Schäfer A, Harzer K (1997) Hydrops fetalis: manifestation in lysosomal storage

diseases including Farber disease. Eur J Pediatr 156(4):292-5.

Keeley EC, Mehrad B, Strieter RM (2008) Chemokines as mediators of neovascularization.

Arterioscler Thromb Vasc Biol 28(11):1928-36.

Kiel MJ, Yilmaz OH, Iwashita T, Yilmaz OH, Terhorst C, Morrison SJ (2005) SLAM family

receptors distinguish hematopoietic stem and progenitor cells and reveal endothelial

niches for stem cells. Cell 121(7):1109–21.

Kim C, Schneider G, Abdel-Latif A, Mierzejewska K, Sunkara M, Borkowska S, Ratajczak

J, Morris AJ, Kucia M, Ratajczak MZ (2012) Ceramide-1-phosphate Regulates Migration

of Multipotent Stromal Cells (MSCs) and Endothelial Progenitor Cells (EPCs) -

Implications for Tissue Regeneration. Stem Cells 31(3):500-510.

Kim Y, Park S, Park C, Kim S, Lee C (1998) A case of Farber lipogranulomatosis. J Korean

Med Sci 13(1):95-8.

Kim SY, Choi SA, Lee S, Lee JS, Hong CR, Lim BC, Kang HJ, Kim KJ, Park SH, Choi M,

Chae JH (2016) Atypical presentation of infantile-onset farber disease with novel

ASAH1 mutations. Am J Med Genet A 170(11):3023-3027.

Kincade P, Lee G, Watanabe T, Sun L, Scheid M (1980) Antigens displayed on murine B

267

lymphocyte precursors. J Immunol. 127(6):2262-8.

Kingma SD, Bodamer OA, Wijburg FA (2015) Epidemiology and diagnosis of lysosomal

storage disorders; challenges of screening. Best Pract Res Clin Endocrinol Metab

29(2):145-57.

Kishnani PS, Dickson PI, Muldowney L, Lee JJ, Rosenberg A, Abichandani R, Bluestone

JA, Burton BK, Dewey M, Freitas A, Gavin D, Griebel D, Hogan M, Holland S,

Tanpaiboon P, Turka LA, Utz JJ, Wang YM, Whitley CB, Kazi ZB, Pariser AR (2016)

Immune response to enzyme replacement therapies in lysosomal storage diseases and the

role of immune tolerance induction. Mol Genet Metab 117(2):66-83.

Kitatani K, Idkowiak-Baldys J, Hannun YA (2008) The sphingolipid salvage pathway in

ceramide metabolism and signaling. Cell Signal 20(6):1010-8.

Koch J, Gärtner S, Li CM, Quintern LE, Bernardo K, Levran O, Schnabel D, Desnick RJ,

Schuchman EH, Sandhoff K (1996) Molecular cloning and characterization of a full-

length complementary DNA encoding human acid ceramidase. Identification Of the first

molecular lesion causing Farber disease. J Biol Chem 271(51):33110-5.

Kostik MM, Chikova IA, Avramenko VV, Vasyakina LI, Le Trionnaire E, Chasnyk VG,

Levade T (2013) Farber lipogranulomatosis with predominant joint involvement

mimicking juvenile idiopathic arthritis. J Inherit Metab Dis 36(6):1079-80.

Kusano KF, Nakamura K, Kusano H, Nishii N, Banba K, Ikeda T, Hashimoto K, Yamamoto

M, Fujio H, Miura A, Ohta K, Morita H, Saito H, Emori T, Nakamura Y, Kusano I, Ohe T

(2004) Significance of the level of monocyte chemoattractant protein-1 in human

atherosclerosis. Circ J 68(7):671-6.

Lai MK, Chew WS, Torta F, Rao A, Harris GL, Chun J, Herr DR (2016) Biological Effects

268

of Naturally Occurring Sphingolipids, Uncommon Variants, and Their Analogs.

Neuromolecular Med 18(3):396-414.

Lamark T, Johansen T (2012) Aggrephagy: selective disposal of protein aggregates by

macroautophagy. Int J Cell Biol 2012:736905.

Levade T, Moser HW, Fensom AH, Harzer K, Moser AB, Salvayre R (1995)

Neurodegenerative course in ceramidase deficiency (Farber disease) correlates with the

residual lysosomal ceramide turnover in cultured living patient cells. J Neurol Sci 134(1-

2):108-14.

Levade T, Sandhoff K, Schulze H, Medin J. Acid ceramidase deficiency: Farber

lipogranulomatosis. Scriver’s OMMBID (The Online Metabolic & Molecular Bases of

Inherited Disease), Valle, Beaudet, Vogelstein, Kinzler, Antonorakis, and Ballabio, eds.

(McGraw-Hill). 2009.

Levade T, Sandhoff K, Schulze H, Medin JA. Acid Ceramidase Deficiency: Farber

Lipogranulomatosis. In: Valle D, Beaudet AL, Vogelstein B, Kinzler KW, Antonarakis

SE, Ballabio A, Gibson K, Mitchell G. eds. New York, NY: McGraw-Hill; 2009.

Li CM, Hong SB, Kopal G, He XX, Linke T, Hou WS, Koch J, Gatt S, Sandhoff K,

Schuchman EH (1998) Cloning and characterization of the full-length cDNA and genomic

sequences encoding murine acid ceramidase. Genomics 50(2):267-74.

Li CM, Park JH, He X, Levy B, Chen F, Arai K, Adler DA, Disteche CM, Koch J, Sandhoff

K, Schuchman EH. (1999) The human acid ceramidase gene (ASAH): structure,

chromosomal location, mutation analysis, and expression. Genomics 62(2):223-31.

Li CM, Park JH, Simonaro CM, He X, Gordon RE, Friedman AH, Ehleiter D, Paris F,

Manova K, Hepbildikler S, Fuks Z, Sandhoff K, Kolesnick R, Schuchman EH (2002)

269

Insertional mutagenesis of the mouse acid ceramidase gene leads to early embryonic

lethality in homozygotes and progressive lipid storage disease in heterozygotes. Genomics

79(2):218-24.

Liang S, Yoshimitsu M, Poeppl A, Rasaiah V, Cai J, Fowler DH, Medin JA (2007) Multiple

reduced-intensity conditioning regimens facilitate correction of Fabry mice after

transplantation of transduced cells. Molecular Therapy 15(3):618-27.

Linke T, Wilkening G, Sadeghlar F, Mozcall H, Bernardo K, Schuchman E, Sandhoff K

(2001) Interfacial regulation of acid ceramidase activity. Stimulation of ceramide

degradation by lysosomal lipids and sphingolipid activator proteins. J Biol Chem

276:5760–5768.

Linthorst GE, Bouwman MG, Wijburg FA, Aerts JM, Poorthuis BJ, Hollak CE (2010)

Screening for Fabry disease in high-risk populations: a systematic review. J Med Genet

47(4):217-22.

Liu P, Anderson RG (1995) Compartmentalized production of ceramide at the cell surface. J

Biol Chem 270(45):27179-85.

Lloyd-Evans E, Platt FM (2010) Lipids on trial: the search for the offending metabolite in

Niemann-Pick type C disease. Traffic 11(4):419-28.

Lopez-Vasquez L, Dworski S, Pouliot R, Galbraith T, Kamani MA, Lacroix D, Auger FA,

Medin JA (2016) Acid ceramidase deficiency leads to multiple skin abnormalities in a

mouse model of Farber disease. Mol Genet Metab 117:S75-76.

Lucki NC, Li D, Bandyopadhyay S, Wang E, Merrill AH, Sewer MB (2012) Acid

ceramidase (ASAH1) represses steroidogenic factor 1-dependent gene transcription in

270

H295R human adrenocortical cells by binding to the receptor. Mol Cell Biol

32(21):4419-31.

Ma Y, Liu H, Tu-Rapp H, Thiesen HJ, Ibrahim SM, Cole SM, Pope RM (2004) Fas ligation

on macrophages enhances IL-1R1-Toll-like receptor 4 signalling and promotes

chronic inflammation. Nat Immunol 5(4):380-87.

Macauley SL, Sidman RL, Schuchman EH, Taksir T, Stewart GR (2008) Neuropathology of

the acid sphingomyelinase knockout mouse model of Niemann-Pick A disease

including structure-function studies associated with cerebellar Purkinje cell

degeneration. Exp Neurol 214(2):181-92.

Maejima I, Takahashi A, Omori H, Kimura T, Takabatake Y, Saitoh T, Yamamoto A,

Hamasaki M, Noda T, Isaka Y, Yoshimori T (2013) Autophagy sequesters damaged

lysosomes to control lysosomal biogenesis and kidney injury. EMBO J 32(17):2336-

47.

Magnan C, Levin BE, Luquet S (2015) Brain lipid sensing and the neural control of energy

balance. Mol Cell Endocrinol 418 Pt 1:3-8.

Mahdy AE1, Cheng JC, Li J, Elojeimy S, Meacham WD, Turner LS, Bai A, Gault CR,

McPherson AS, Garcia N, Beckham TH, Saad A, Bielawska A, Bielawski J, Hannun YA,

Keane TE, Taha MI, Hammouda HM, Norris JS, Liu X (2009) Acid ceramidase

upregulation in prostate cancer cells confers resistance to radiation: AC inhibition, a

potential radiosensitizer. Mol Ther 17(3):430-8.

Mao C, Obeid LM (2008) Ceramidases: regulators of cellular responses mediated by

ceramide, sphingosine, and sphingosine-1-phosphate. Biochim Biophys Acta

1781(9):424-34.

271

Martin P, Patrick P, Bateson G, Bateson P: Measuring behaviour: an introductory guide:

Cambridge University Press, 1993.

Mason, WT. Fluorescent and Luminescent Probes for Biological Activity, 2nd Edition,

Academic Press. 1999, p136-55.

Matern D, Gavrilov D, Oglesbee D, Raymond K, Rinaldo P, Tortorelli S (2015) Newborn

screening for lysosomal storage disorders. Semin Perinatol 39(3):206-16.

Mechtler TP, Stary S, Metz TF, De Jesús VR, Greber-Platzer S, Pollak A, Herkner KR,

Streubel B, Kasper DC (2012) Neonatal screening for lysosomal storage disorders:

feasibility and incidence from a nationwide study in Austria. Lancet 379(9813):335-41.

Middleton MR1, Grob JJ, Aaronson N, Fierlbeck G, Tilgen W, Seiter S, Gore M, Aamdal S,

Cebon J, Coates A, Dreno B, Henz M, Schadendorf D, Kapp A, Weiss J, Fraass U,

Statkevich P, Muller M, Thatcher N (2000) Randomized phase III study of temozolomide

versus dacarbazine in the treatment of patients with advanced metastatic malignant

melanoma. J Clin Oncol 18(1):158-66.

McGlynn R, Dobrenis K, Walkley SU (2004) Differential subcellular localization of

cholesterol, gangliosides, and glycosaminoglycans in murine models of

mucopolysaccharide storage disorders. The Journal of comparative neurology

480(4):415-26.

Meikle PJ, Hopwood JJ, Clague AE, Carey WF (1999) Prevalence of lysosomal storage

disorders. JAMA 281(3):249-54.

Mencarelli C, Martinez-Martinez P (2013) Ceramide function in the brain: when a slight tilt

is enough. Cell Mol Life Sci 70(2):181-203.

272

Merrill AH Jr, Sullards MC, Allegood JC, Kelly S, Wang E (2005) Sphingolipidomics: high-

throughput, structure-specific, and quantitative analysis of sphingolipids by liquid

chromatography tandem mass spectrometry. Methods 36(2):207-24.

Micsenyi MC, Sikora J, Stephney G, Dobrenis K, Walkley SU (2013) Lysosomal membrane

permeability stimulates protein aggregate formation in neurons of a lysosomal disease.

J Neurosci 33(26):10815-27.

Mielke MM, Bandaru VVR, Haughey NJ, Xia J, Fried LP, Yasar S, Albert M, Varma V,

Harris G, Schneider EB, Rabins PV, Bandeen-Roche K, Lyketsos CG, Carlson MC

(2012) Serum ceramides increase the risk of Alzheimer disease: the Women’s Health

and Aging Study II. Neurology 79(7):633-41.

Molz G (1968) [Farber's disease. Pathologic anatomical findings]. [Article in German].

Virchows Arch A Pathol Pathol Anat 344(1):86-99.

Momoi T, Ben-Yoseph Y, Nadler HL (1982) Substrate-specificities of acid and alkaline

ceramidases in fibroblasts from patients with Farber disease and controls. Biochem J

205(2):419-25.

Mondal RK, Nandi M, Datta S, Hira M (2009) Disseminated lipogranulomatosis. Indian

Pediatr. 46(2):175-7.

Morad SA, Cabot MC (2013) Ceramide-orchestrated signalling in cancer cells. Nat Rev

Cancer. 13(1):51-65.

Moser HW, Prensky AL, Wolfe HJ, Rosman NP (1969) Farber's lipogranulomatosis. Report

of a case and demonstration of an excess of free ceramide and ganglioside. Am J Med

47(6):869-90.

Nigam PK (2011) Serum Lipid Profile: Fasting or Non-fasting? Indian J Clin Biochem 26(1):

273

96–97.

Nishimoto N, Kishimoto T (2004) Inhibition of IL-6 for the treatment of inflammatory

diseases. Curr Opin Pharmacol 4(4):386-91.

Notta F, Doulatov S, Laurenti E, Poeppl A, Jurisica I, Dick JE (2011) Isolation of single

human hematopoietic stem cells capable of long-term multilineage engraftment. Science

333(6039):218-21.

Notta F, Zandi S, Takayama N, Dobson S, Gan OI, Wilson G, Kaufmann KB, McLeod J,

Laurenti E, Dunant CF, McPherson JD, Stein LD, Dror Y, Dick JE (2015) Distinct routes

of lineage development reshape the human blood hierarchy across ontogeny. Science

351(6269):aab2116.

Ogretmen B, Hannun YA (2004) Biologically active sphingolipids in cancer pathogenesis

and treatment. Nat Rev Cancer 4(8):604-16.

Okino N, He X, Gatt S, Sandhoff K, Ito M, Schuchman EH (2003) The reverse activity of

human acid ceramidase. J Biol Chem 278(32):29948-53.

Ortolano S, Viéitez I, Navarro C, Spuch C (2014) Treatment of lysosomal storage diseases:

recent patents and future strategies. Recent Pat Endocr Metab Immune Drug Discov

8(1):9-25.

Osawa M, Hanada K, Hamada H, Nakauchi H (1996) Long-term lymphohematopoietic

reconstitution by a single CD34-low/negative hematopoietic stem cell. Science

273(5272):242–5.

Pacheco CD, Lieberman AP (2008) The pathogenesis of Niemann-Pick type C disease: a role

for autophagy? Expert Rev Mol Med 10:e26.

Pandey MK, Grabowski GA (2013) Immunological cells and functions in Gaucher disease.

274

Crit Rev Oncog 18(3):197-220.

Park JH, Schuchman EH (2006) Acid ceramidase and human disease. Biochim Biophys Acta

1758(12):2133-8.

Partanen S, Haapanen A, Kielar C, Pontikis C, Alexander N, Inkinen T, Saftig P,

Gillingwater TH, Cooper JD, Tyynela J (2008) Synaptic changes in the thalamocortical

system of cathepsin D-deficient mice: a model of human congenital neuronal ceroid-

lipofuscinosis. J Neuropathol Exp Neurol 67(1):16-29.

Patel DD, Zachariah JP, Whichard LP (2001) CXCR3 and CCR5 ligands in rheumatoid

arthritis synovium. Clin Immunol 98(1):39-45.

Pavlova EV, Deegan PB, Tindall J, McFarlane I, Mehta A, Hughes D, Wraith JE, Cox TM

(2011) Potential biomarkers of osteonecrosis in Gaucher disease. Blood Cells Mol Dis

46(1):27-33.

Pavone L, Moser HW, Mollica F, Reitano C, Durand P: Farber's lipogranulomatosis (1980)

Ceramidase deficiency and prolonged survival in three relatives. Johns Hopkins Medical

Journal 147(5):193-196.

Pellissier J, Berard-Badier M, Pinsard N (1986) Farber’s disease in two siblings, sural nerve

and subcutaneous biopsies by fight and electron microscopy. Acta Neuropathol 72:178-88.

Peters A, Palay SL, H.DeF. W Fine structure of the nervous system: neurons and their

supporting cells. 3rd ed. New York: Oxford University Press, 1991.

Peters C, Shapiro EG, Anderson J, Henslee-Downey PJ, Klemperer MR, Cowan MJ,

Saunders EF, deAlarcon PA, Twist C, Nachman JB, Hale GA, Harris RE, Rozans MK,

Kurtzberg J, Grayson GH, Williams TE, Lenarsky C, Wagner JE, Krivit W (1998) Hurler

syndrome: II. Outcome of HLA-genotypically identical sibling and HLA-haploidentical

275

related donor bone marrow transplantation in fifty-four children. The Storage Disease

Collaborative Study Group. Blood 91(7):2601-8.

Peterson TA, Adadey A, Santana-Cruz I, Sun Y, Winder A, Kann MG (2010) DMDM:

Domain Mapping of Disease Mutations. Bioinformatics 26(19):2458-2459.

Petrovic-Rackov L, Pejnovic N (2006) Clinical significance of IL-18, IL-15, IL-12 and TNF-

alpha measurement in rheumatoid arthritis. Clin Rheumatol 25(4)448-52.

Pettus BJ, Baes M, Busman M, Hannun YA, Van Veldhoven PP (2004) Mass spectrometric

analysis of ceramide perturbations in brain and fibroblasts of mice and human patients

with peroxisomal disorders. Rapid Commun Mass Spectrom 18(14):1569-74.

Platt FM, Boland B, van der Spoel AC (2012) The cell biology of disease: lysosomal storage

disorders: the cellular impact of lysosomal dysfunction. J Cell Biol 199(5):723-34.

Praggastis M, Tortelli B, Zhang J, Fujiwara H, Sidhu R, Chacko A, Chen Z, Chung C,

Lieberman AP, Sikora J, Davidson C, Walkley SU, Pipalia NH, Maxfield FR, Schaffer JE,

Ory DS (2015) A murine Niemann-Pick C1 I1061T knock-in model recapitulates the

pathological features of the most prevalent human disease allele. J Neurosci 35(21):8091-

106.

Prensky AL, Ferreira G, Carr S, Moser HW (1967) Ceramide and Ganglioside Accumulation

in Farber's Lipogranulomatosis. Exp Biol Med (Maywood) 126:725-728.

Pressey SN, Smith DA, Wong AM, Platt FM, Cooper JD (2012) Early glial activation,

synaptic changes and axonal pathology in the thalamocortical system of Niemann-Pick

type C1 mice. Neurobiol Dis 45(3):1086-100.

276

Puglielli L, Ellis BC, Saunders AJ, Kovacs DM (2003) Ceramide stabilizes beta-site amyloid

precursor protein-cleaving enzyme 1 and promotes amyloid beta-peptide biogenesis. J

Biol Chem 278(22):19777-83.

Quehenberger O, Armando AM, Brown AH, Milne SB, Myers DS, Merrill AH,

Bandyopadhyay S, Jones KN, Kelly S, Shaner RL, Sullards CM, Wang E, Murphy RC,

Barkley RM, Leiker TJ, Raetz CR, Guan Z, Laird GM, Six DA, Russell DW, McDonald

JG, Subramaniam S, Fahy E, Dennis EA (2010) Lipidomics reveals a remarkable diversity

of lipids in human plasma. J Lipid Res 51(11):3299-305.

Ramsubir S, Nonaka T, Girbes CB, Carpentier S, Levade T, Medin JA (2008) In vivo

delivery of human acid ceramidase via cord blood transplantation and direct injection of

lentivirus as novel treatment approaches for Farber disease. Mol Genet Metab 95(3)133-

141.

Ratajczak MZ, Kim CH, Abdel-Latif A, Schneider G, Kucia M, Morris AJ, Laughlin MJ,

Ratajczak J (2012) A novel perspective on stem cell homing and mobilization: review on

bioactive lipids as potent chemoattractants and cationic peptides as underappreciated

modulators of responsiveness to SDF-1 gradients. Leukemia 26:63-72.

Ratajczak MZ, Lee H, Wysoczynski M, Wan W, Marlicz W, Laughlin MJ, Kucia M,

Janowska-Wieczorek A, Ratajczak J (2010) Novel insight into stem cell mobilization-

plasma sphingosine-1-phosphate is a major chemoattractant that directs the egress of

hematopoietic stem progenitor cells from the bone marrow and its level in peripheral

blood increases during mobilization due to activation of complement cascade/membrane

attack complex. Leukemia 24:976-85.

277

Realini N, Palese F, Pizzirani D, Pontis S, Basit A, Bach A, Ganesan A, Piomelli D (2016)

Acid Ceramidase in Melanoma: expression, localization, and effects of pharmacological

inhibition. J Biol Chem 291(5):2422-34.

Ries M, Schaefer E, Lührs T, Mani L, Kuhn J, Vanier MT, Krummenauer F, Gal A, Beck M,

Mengel E (2006) Critical assessment of chitotriosidase analysis in the rational laboratory

diagnosis of children with Gaucher disease and Niemann-Pick disease type A/B and C. J

Inherit Metab Dis 29(5):647-52.

Rombach SM, Smid BE, Bouwman MG, Linthorst GE, Dijkgraaf MG, Hollak CE (2013)

Long term enzyme replacement therapy for Fabry disease: effectiveness on kidney, heart

and brain. Orphanet J Rare Dis 8:47.

Rossi DJ, Bryder D, Zahn JM, Ahlenius H, Sonu R, Wagers AJ, Weissman IL (2005) Cell

intrinsic alterations underlie hematopoietic stem cell aging. Proc Natl Acad Sci USA

102(26):9194–9.

Rubboli G, Veggiotti P, Pini A, Berardinelli A, Cantalupo G, Bertini E, Tiziano FD,

D'Amico A, Piazza E, Abiusi E, Fiori S, Pasini E, Darra F, Gobbi G, Michelucci R (2015)

Spinal muscular atrophy associated with progressive myoclonic epilepsy: A rare condition

caused by mutations in ASAH1. Epilepsia 56(5):692-8.

Rutsaert J, Tondeur M, Vamos-Hurwitz E, Dustin P (1977) The cellular lesions of Farber's

disease and their experimental reproduction in tissue culture. Lab Invest 36(5):474-80.

Saad AF, Meacham WD, Bai A, Anelli V, Elojeimy S, Mahdy AE, Turner LS, Cheng J,

Bielawska A, Bielawski J, Keane TE, Obeid LM, Hannun YA, Norris JS, Liu X (2007)

The functional effects of acid ceramidase overexpression in prostate cancer progression

and resistance to chemotherapy. Cancer Biol Ther 6(9):1455-60.

278

Saba JD, Nara F, Bielawska A, Garrett S, Hannun YA (1997) The BST1 gene of

Saccharomyces cerevisiae is the sphingosine-1-phosphate lyase. J Biol Chem

272(42):26087-90.

Sachs B (1887) On arrested cerebral development with special reference to cortical

pathology. J Nerv Ment Dis 14:541-554.

Samsel L, Zaidel G, Drumgoole HM, Jelovac D, Drachenberg C, Rhee JG, Brodie AM,

Bielawska A, Smyth MJ (2004) The ceramide analog, B13, induces apoptosis in prostate

cancer cell lines and inhibits tumor growth in prostate cancer xenografts. Prostate

58(4):382-93.

Samuelsson K, Zetterstrom R, Ivemark BI (1972) Studies on a case of lipogranulomatosis

(Farber’s disease) with protracted course, in Volk BW, Aronson SM (eds): Sphingolipids,

Sphingolipidoses and Allied Disorders. New York, Plenum, pp 533.

Sarna JR, Larouche M, Marzban H, Sillitoe RV, Rancourt DE, Hawkes R (2003) Patterned

Purkinje cell degeneration in mouse models of Niemann-Pick type C disease. The Journal

of comparative neurology 456(3):279-91.

Sassa T, Kihara A (2014) Metabolism of Very Long-Chain Fatty Acids: Genes and

Pathophysiology. Biomolecules & Therapeutics 22(2):83-92.

Sawyer SL, Hartley T, Dyment DA, Beaulieu CL, Schwartzentruber J, Smith A, Bedford

HM, Bernard G, Bernier FP, Brais B, Bulman DE, Warman Chardon J, Chitayat D,

Deladoëy J, Fernandez BA, Frosk P, Geraghty MT, Gerull B, Gibson W, Gow RM,

Graham GE, Green JS, Heon E, Horvath G, Innes AM, Jabado N, Kim RH, Koenekoop

RK, Khan A, Lehmann OJ, Mendoza-Londono R, Michaud JL, Nikkel SM, Penney LS,

Polychronakos C, Richer J, Rouleau GA, Samuels ME, Siu VM, Suchowersky O,

279

Tarnopolsky MA, Yoon G, Zahir FR; FORGE Canada Consortium.; Care4Rare Canada

Consortium., Majewski J, Boycott KM (2016) Utility of whole-exome sequencing for

those near the end of the diagnostic odyssey: time to address gaps in care. Clin Genet

89(3):275-84.

Schuchman E. (2014) A132: Farber disease explains subset of Juvenile Idiopathic Arthritis

[abstract]. Arthritis Rheum 66:S173.

Schuchman EH (2016) Acid ceramidase and the treatment of ceramide diseases: The

expanding role of enzyme replacement therapy. Biochim Biophys Acta 1862(9):1459-71.

Scott CR, Elliott S, Buroker N, Thomas LI, Keutzer J, Glass M, Gelb MH, Turecek F (2013)

Identification of infants at risk for developing Fabry, Pompe, or mucopolysaccharidosis-I

from newborn blood spots by tandem mass spectrometry. J Pediatr 163(2):498-503.

Seelan RS, Qian C, Yokomizo A, Bostwick DG, Smith DI, Liu W (2000) Human acid

ceramidase is overexpressed but not mutated in prostate cancer. Genes Chromosomes

Cancer 29(2):137-46.

Ségui B, Bezombes C, Uro-Coste E, Medin JA, Andrieu-Abadie N, Augé N, Brouchet A,

Laurent G, Salvayre R, Jaffrézou JP, Levade T (2000) Stress-induced apoptosis is not

mediated by endolysosomal ceramide. FASEB J 14(1):36-47.

Selden NR, Al-Uzri A, Huhn SL, Koch TK, Sikora DM, Nguyen-Driver MD, Guillaume DJ,

Koh JL, Gultekin SH, Anderson JC, Vogel H, Sutcliffe TL, Jacobs Y, Steiner RD (2013)

Central nervous system stem cell transplantation for children with neuronal ceroid

lipofuscinosis. J Neurosurg Pediatr 11(6):643-52.

280

Selzner M, Bielawska A, Morse MA, Rüdiger HA, Sindram D, Hannun YA, Clavien PA

(2001) Induction of apoptotic cell death and prevention of tumor growth by ceramide

analogues in metastatic human colon cancer. Cancer Res 61(3):1233-40.

Shah C, Yang G, Lee I, Bielawski J, Hannun YA, Samad F (2008) Protection from high fat

diet-induced increase in ceramide in mice lacking plasminogen activator inhibitor 1. J

Biol Chem 283(20):13538-48.

Sharma P, Kar R, Dutta S, Pati HP, Saxena R (2009) Niemann-Pick disease, type B with

TRAP-positive storage cells and secondary sea blue histiocytosis. Eur J Histochem.

53(3):183-186.

Sheth KJ, Swick HM (1980) Peripheral nerve conduction in Fabry disease. Ann Neurol

7(4):319-23.

Shizuru JA, Jerabek L, Edwards CT, Weissman IL (1996) Transplantation of purified

hematopoietic stem cells: requirements for overcoming the barriers of allogeneic

engraftment. Biol Blood Marrow Transplant 2(1):3-14.

Shtraizent N, Eliyahu E, Park JH, He X, Shalgi R, Schuchman EH (2008) Autoproteolytic

cleavage and activation of human acid ceramidase. J Biol Chem 283(17):11253-9.

Shui G, Stebbins JW, Lam BD, Cheong WF, Lam SM, Gregoire F, Kusonoki J, Wenk MR

(2011) Comparative plasma lipidome between human and cynomolgus monkey: are

plasma polar lipids good biomarkers for diabetic monkeys? PLoS One 6(5):e19731.

Signorelli P, Munoz-Olaya JM, Gagliostro V, Casas J, Ghidoni R, Fabrias G (2009)

Dihydroceramide intracellular increase in response to resveratrol treatment mediates

autophagy in gastric cancer cells. Cancer Lett 282(2):238-43.

Sikora J, Leddy J, Gulinello M, Walkley SU (2016) X-linked Christianson syndrome:

281

heterozygous female Slc9a6 knockout mice develop mosaic neuropathological changes

and related behavioral abnormalities. Dis Model Mech 9(1):13-23.

Sikora J, Dworski S, Jones EE, Kamani M, Micsenyi MC, Sawada T, Le Faouder P,

Bertrand-Michel J, Dupuy A, Dunn CK, Xuan I, Casas J, Fabrias G, Hampson DR,

Levade T, Drake RR, Medin JA, Walkley SU (2017) Acid ceramidase deficiency (Farber

disease) in Asah1P361R/P361R mice results in a broad range of central nervous system

abnormalities. Am J Pathol. 187(4):864-883.

Singh RH, Cunningham AC, Mofidi S, Douglas TD, Frazier DM, Hook DG, Jeffers L,

McCune H, Moseley KD, Ogata B, Pendyal S, Skrabal J, Splett PL, Stembridge A, Wessel

A, Rohr F (2016) Updated, web-based nutrition management guideline for PKU: An

evidence and consensus based approach. Mol Genet Metab 118(2):72-83.

Sirrs S, Hollak C, Merkel M, Sechi A, Glamuzina E, Janssen MC, Lachmann R, Langendonk

J, Scarpelli M, Ben Omran T, Mochel F, Tchan MC (2015) The Frequencies of Different

Inborn Errors of Metabolism in Adult Metabolic Centres: Report from the SSIEM Adult

Metabolic Physicians Group. JIMD Rep 27:85-91.

Spada M, Pagliardini S, Yasuda M et al (2006) High incidence of later-onset Fabry disease

revealed by newborn screening. Am J Hum Genet 79(1):31-40.

Spangrude GJ, Heimfeld S, Weissman IL (1988) Purification and characterization of mouse

hematopoietic stem cells. Science 241(4861):58-62.

Sugita M, Dulaney JT, Moser HW (1972) Ceramidase deficiency in Farber's disease

(lipogranulomatosis). Science. 178(4065):1100-2.

282

Sugita M, Connolly P, Dulaney JT, Moser HW (1973) Fatty acid composition of free

ceramides of kidney and cerebellum from a patient with Farber's disease. Lipids 8(7):401-

Sugita M, Iwamori M, Evans J, McCluer RH, Dulaney JT, Moser HW (1974) High

performance liquid chromatography of ceramides: application to analysis in human tissues

and demonstration of ceramide excess in Farber's disease. J Lipid Res 15(3):223-6.

Takenaka T, Murray GJ, Qin G, Quirk JM, Ohshima T, Qasba P, Clark K, Kulkarni AB,

Brady RO, Medin JA (2000) Long-term enzyme correction and lipid reduction in multiple

organs of primary and secondary transplanted Fabry mice receiving transduced bone

marrow cells. PNAS 97(13):7515-20.

Tantawy AAG (2015) Cytokines in Gaucher disease: Role in the pathogenesis of bone and

pulmonary disease. Egyptian Journal of Medical Human Genetics 16(3):207-13.

Tay W (1881) Symmetrical changes in the region of the yellow spot in each eye of an infant.

Trans Opthalmol Soc 1:55–57.

The UniProt Consortium (2014) UniProt: a hub for protein information. Nucl Acids Res

43(D1):D204-D212.

Thomas A, Burant A, Bui N, Graham D, Yuva-Paylor LA, Paylor R (2009): Marble burying

reflects a repetitive and perseverative behavior more than novelty-induced anxiety.

Psychopharmacology (Berl) 204(2):361-73.

Thomas PK, King RH, Kocen RS, Brett EM (1977) Comparative ultrastructural observations

on peripheral nerve abnormalities in the late infantile, juvenile and late onset forms of

metachromatic leukodystrophy. Acta Neuropathol 39(3):237-45.

283

Tohyama J, Oya Y, Ezoe T, Vanier MT, Nakayasu H, Fujita N, Suzuki K (1999) Ceramide

accumulation is associated with increased apoptotic cell death in cultured fibroblasts of

sphingolipid activator protein-deficient mouse but not in fibroblasts of patients with

Farber disease. J Inherit Metab Dis 22(5):649-62.

Tong M, de la Monte SM (2009) Mechanisms of ceramide-mediated neurodegeneration. J

Alzheimers Dis 16(4):705-14.

Torcoletti M, Petaccia A, Pinto RM, Hladnik U, Locatelli F, Agostoni C, Corona F (2014)

Farber disease in infancy resembling juvenile idiopathic arthritis: identification of two

new mutations and a good early response to allogeneic haematopoietic stem cell

transplantation. Rheumatology (Oxford) 53(8):1533-1534.

Tu-Rapp H, Hammermuller A, Mix E, Kreutzer HJ, Goerlich R, Kohler H, Nizze H, Thiesen

HJ, Ibrahim SM (2004) A proinflammatory role for Fas in joints of mice with collagen-

induced arthritis. Arthritis Res Ther 6(5):R404-14. van der Tol L, Smid BE, Poorthuis BJ et al (2014) A systematic review on screening for

Fabry disease: prevalence of individuals with genetic variants of unknown significance. J

Med Genet 51(1):1-9. van Echten-Deckert G, Klein A, Linke T, Heinemann T, Weisgerber J, Sandhoff K (1997)

Turnover of endogenous ceramide in cultured normal and Farber fibroblasts. J Lipid Res

38(12):2569-79. van Lijnschoten G, Groener JEM, Maas SM, Ben-Yoseph Y, Dingemans KP, Offerhaus GJA

(2000) Intrauterine Fetal Death Due to Farber Disease: Case Report. Pediatr Dev Pathol

3(6):597-602.

Vanni N, Fruscione F, Ferlazzo E, Striano P, Robbiano A, Traverso M, Sander T, Falace A,

284

Gazzerro E, Bramanti P, Bielawski J, Fassio A, Minetti C, Genton P, Zara F (2014)

Impairment of ceramide synthesis causes a novel progressive myoclonus epilepsy. Ann

Neurol 76(2):206-12.

Velazquez L, Cheng AM, Fleming HE, Furlonger C, Vesely S, Bernstein A, Paige CJ,

Pawson T (2002) Cytokine signaling and hematopoietic homeostasis are disrupted in Lnk-

deficient mice. J Exp Med 195(12):1599-1611.

Vellodi A, Young E, Cooper A, Lidchi V, Winchester B, Wraith JE (1999) Long-term

follow-up following bone marrow transplantation for Hunter disease. J Inherit Metab Dis

22(5):638-48.

Volland S, Esteve-Rudd J, Hoo J, Yee C, Williams DS (2015) A comparison of some

organizational characteristics of the mouse central retina and the human macula. PLoS

One 10(4):e0125631.

Vormoor, J, Ehlert, K, Groll, AH, Koch, HG, Frosch, M, Roth, J (2004). Successful

hematopoietic stem cell transplantation in Farber disease. J Pediatr 144(1):132-4.

Walia JS, Neschadim A, Lopez-Perez O, Alayoubi A, Fan X, Carpentier S, Madden M, Lee

CJ, Cheung F, Jaffray DA, Levade T, McCart JA, Medin JA (2011) Autologous

transplantation of lentivector/acid ceramidase-transduced hematopoietic cells in

nonhuman primates. Hum Gene Ther 22(6):679-87.

Walkley SU, Baker HJ, Rattazzi MC, Haskins ME, Wu JY (1991) Neuroaxonal dystrophy in

neuronal storage disorders: evidence for major GABAergic neuron involvement. J Neurol

Sci 104(1):1-8.

Wang G, Dinkins M, He Q, Zhu G, Poirier C, Campbell A, Mayer-Proschel M, Bieberich E

285

(2012) Astrocytes secrete exosomes enriched with proapoptotic ceramide and prostate

apoptosis response 4 (PAR-4): potential mechanism of apoptosis induction in Alzheimer

disease (AD). J Biol Chem 287(25):21384-95.

Warnes GR, Bolker B, Bonebakker L, Gentleman R, Liaw WHA, Maechler TLM,

Magnusson A, Moeller S, Schwartz M, Venables B (2011) gplots: Various R

programming tools for plotting data. R package version 2.10.1. Retrieved from

http://cran.r-project.org/package=gplots.

Weidemann F, Krämer J, Duning T, Lenders M, Canaan-Kühl S, Krebs A, Guerrero

González H, Sommer C, Üçeyler N, Niemann M, Störk S, Schelleckes M, Reiermann S,

Stypmann J, Brand SM, Wanner C, Brand E (2014) Patients with Fabry disease after

enzyme replacement therapy dose reduction versus treatment switch. J Am Soc Nephrol

25(4):837-49.

Wittmann J, Karg E, Turi S, Legnini E, Wittmann G, Giese AK, Lukas J, Gölnitz U,

Klingenhäger M, Bodamer O, Mühl A, Rolfs A (2012) Newborn screening for lysosomal

storage disorders in Hungary. JIMD Rep 6:117-25.

Wolf NS, Kone A, Priestley GV, Bartelmez SH (1993) In vivo and in vitro characterization

of long-term repopulating primitive hematopoietic cells isolated by sequential Hoechst

33342-rhodamine 123 FACS selection. Exp Hematol 21(5):614-22.

Wraith EJ (2004) Clinical aspects and diagnosis. F.M. Platt, S.U. Walkley (Eds.), Lysosomal

disorders of the brain, Oxford University Press, pp. 50–77.

Wymann MP, Schneiter R (2008) Lipid signalling in disease. Nat Rev Mol Cell Biol

9(2):162-76.

Yeager AM, Uhas KA, Coles CD, David PC, Krause WL, Moser HW (2000) Bone marrow

286

transplantation for infantile ceramidase deficiency (Farber disease). Bone Marrow

Transplant 26(3):357-63.

Yoshimitsu M, Higuchi K, Ramsubir S, Nonaka T, Rasaiah VI, Siatskas C, Liang SB,

Murray GJ, Brady RO, Medin JA (2007) Efficient correction of Fabry mice and patient

cells mediated by lentiviral transduction of hematopoietic stem/progenitor cells. Gene

Therapy. 14(3):256-65.

Zappatini-Tommasi L, Dumontel C, Guibaud P, Girod C (1992) Farber disease: an

ultrastructural study. Report of a case and review of the literature. Virchows Arch A

Pathol Anat Histopathol 420(3):281-90.

Zetterstrom R (1958) Disseminated lipogranulomatosis (Farber’s disease). Acta Paediatr

47:501.

Zhao L, Spassieva SD, Jucius TJ, Shultz LD, Shick HE, Macklin WB, Hannun YA, Obeid

LM, Ackerman SL (2011) A deficiency of ceramide biosynthesis causes cerebellar

purkinje cell neurodegeneration and lipofuscin accumulation. PLoS Genet 7(5):e1002063.

Zhang P, Liu B, Jenkins GM, Hannun YA, Obeid LM (1997) Expression of neutral

sphingomyelinase identifies a distinct pool of sphingomyelin involved in apoptosis. J Biol

Chem 272(15):9609-12.

Zhou S, Schuetz JD, Bunting KD, Colapietro AM, Sampath J, Morris JJ, Lagutina I,

Grosveld GC, Osawa M, Nakauchi H, Sorrentino BP (2001) The ABC transporter

Bcrp1/ABCG2 is expressed in a wide variety of stem cells and is a molecular determinant

of the side-population phenotype. Nat Med 7(9):1028-34.

Zhou J, Tawk M, Tiziano FD, Veillet J, Bayes M, Nolent F, Garcia V, Servidei S, Bertini E,

Castro-Giner F, Renda Y, Carpentier S, Andrieu-Abadie N, Gut I, Levade T, Topaloglu H,

287

Melki J (2012) Spinal muscular atrophy associated with progressive myoclonic epilepsy is caused by mutations in ASAH1. Am J Hum Genet 91(1):5-14.

288

Chapter 3 Permission to reuse the article “Markedly perturbed hematopoiesis in acid ceramidase deficient mice” in a thesis was granted on September 17, 2015 by Lerella Ripari on behalf of the journal Haematologica under the Ferrata Storti Foundation, Pavia, Italy.

Chapter 4 Permission to reuse the article “Acid Ceramidase Deficiency in Mice Results in a Broad Range of Central Nervous System Abnormalities” in a thesis was granted on March 13, 2017 by the American Journal of Pathology.

Chapter 6 Permission to reuse the article “Acid ceramidase deficiency is characterized by a unique plasma cytokine and ceramide profile that is altered by therapy” in both print and electronic thesis/dissertation was granted on December 20, 2016 by the licensed content publisher Elsevier from the licensed content publication Biochemica et Biphysica Acta (BBA) – Molecular Basis of Disease. License number 4013260677135.