EXPLORING GLUTARIC ACIDEMIA TYPE I and the LYSINE OXIDATION PATHWAY by LEAH EVE VENTURONI BA, University of Colorado, Boulder

Home , ACADSB, Glutaric acid

EXPLORING GLUTARIC ACIDEMIA TYPE I AND THE LYSINE OXIDATION PATHWAY

LEAH EVE VENTURONI

B.A., University of Colorado, Boulder, 2010

A thesis submitted to the

Faculty of the Graduate School of the

University of Colorado in partial fulfillment

of the requirements for the degree of

Doctor of Philosophy

Human Medical Genetics and Genomics

2017

This thesis for the Doctor of Philosophy degree by

Leah Eve Venturoni

has been approved for the

Human Medical Genetics and Genomics Program

David D. Pollock, Chair

Stephen I. Goodman, Advisor

Kimberly Bjugstad

Mark Duncan

Curt R. Freed

Johan L. Van Hove

Date: August 18, 2017

ii Venturoni, Leah Eve (Ph.D., Human Medical Genetics and Genomics)

Exploring Glutaric Acidemia Type I and the Lysine Oxidation Pathway

Thesis directed by Distinguished Professor Stephen I. Goodman

ABSTRACT

Glutaric acidemia type I (GA1) is a recessive inborn error of lysine metabolism caused by a deficiency of glutaryl-CoA dehydrogenase (GCDH) resulting in accumulation of glutaric and 3-hydroxyglutaric acids and acute striatal necrosis. The latter is prevented if detected early and treatment begun before crisis. This thesis aims to further our understanding of GA1 following encephalopathic crisis, improve the ability to diagnose patients, takes an initial step towards testing a new treatment, and explores the use of zebrafish as a model of disease.

The literature describes a loss of medium spiny neurons following crises in patients; however, this has never been tested. Staining postmortem putamen tested for a loss of medium spiny neurons in patients. Instead of observing a loss of this population, I found that the number of medium spiny neurons was the same in both groups suggesting a proportional loss of all cells in the striatum following crisis.

Currently, most patients are diagnosed following identification of two pathogenic mutations in GCDH by sequencing each exon in conjunction with aCGH.

Using new deep sequencing technologies, I describe a method capable of screening all genomic regions of GCDH for mutations and copy number variation using a single procedure.

Current treatment of GA1 prevents striatal necrosis in 80% to 90% of patients; however, there is growing concern about the effects of lifelong exposure to the toxin.

iii One hypothesis is that blocking the lysine oxidation pathway upstream of GCDH would prevent accumulation of the toxin and could treat GA1 patients. By performing exome sequencing on a family with 2-ketoadipic acidemia, I identified DHTKD1 as a candidate for such treatment.

Finally, I characterized a zebrafish line with a splice site mutation in GCDH A for use as a GA1 model. The animals had significantly reduced enzyme activity but only mildly elevated metabolites. These fish are not suitable as a model of disease, but a complete knockout of GCDH A likely will be. It is anticipated that the work described here will further our understanding of GA1 and the lysine oxidation pathway, and lay the ground work for future experiments.

The form and content of this abstract are approved. I recommend its publication.

Approved: Stephen I. Goodman

iv DEDICATION

I would like to dedicate my dissertation first and foremost to my parents, Linda and

Steve Venturoni. With their support and encouragement, I found a field of study that completely thrills, delights, and challenges me. Through all the ups and downs of graduate school, they have been my constant supporters and have helped to give me the drive necessary to complete my classes, experiments, and dissertation.

I am also dedicating my dissertation to my boyfriend and constant companion Steven

Benedict. With his support, I have been able to overcome the difficult moments to be found in graduate work, and he has always been there to celebrate my accomplishments with me, both the small and the large.

Finally, I would like to dedicate my dissertation to two mentors of science without whom I would not have found my love of science nor would I have considered research and graduate school an option. The first mentor was Mr. Charlie Cuba, my science teacher for 9th,

10th, and 12th grades at Summit High School in Summit County, Colorado. Mr. Cuba was the first person to introduce me to this molecular world which I have always found to be so enchanting. He is the first to give me the ability to interact with the scientific process firsthand and to develop scientific inquiries of my own. The second mentor is Andy Roberts, the chief scientific officer of the Keystone Symposia from 2005-2011. Andy acted as a mentor during my

11th and 12th grades and was my supervisor as I worked with the Keystone Symposia to evaluate the success of scientific meetings. Andy was a guiding force in my decision to attend graduate school and was always delighted to hear updates of my work at the University of

Colorado until his passing in 2014.

All those listed above have been guiding and championing forces in my budding career as a scientist, and it is to all of them that I dedicate the following dissertation.

v ACKNOWLEDGMENTS

I would like to thank Dr. Steve Goodman for acting as my mentor for the 6 years I spent in the Goodman laboratory. During my time in the lab, Dr. Goodman has always been exceptionally supportive of my goals and the work that I performed. He helped me to learn techniques that were commonly used in the lab as well as how best to interpret my findings.

Finally, Dr. Goodman was always supportive of my desires to work with techniques that the lab did not have any experience with but were beneficial to my thesis work.

During my time in the Goodman lab, Dr. Mike Woontner also acted as a mentor and taught me many of the techniques that were used during my thesis work. Dr. Woontner taught me the importance of a well-designed experiment, including all appropriate controls, and always challenged me to look only at the facts demonstrated by my experimental results and not to assume facts not discreetly proven by an experimental result.

I would like to thank all members of the Goodman lab for their support and encouragement during my graduate career. During my time with the Goodman lab, they were all happy to teach me about their work to maintain the service lab, showed me how to use their machines and tests for parts of my thesis work, and made me truly feel like a part of the lab family. I will always look back on my time with the Goodman lab and all its members fondly because they all had a hand in making my gradate experience so amazing.

I would also like to thank the Human Medical Genetics and Genomics program, the directors, the administrators, the faculty, and the students for their part in my graduate studies.

All members of the program have helped to create a truly wonderful support network and have helped to make the work described in this thesis possible.

Additionally, I would like to acknowledge the following contributions to specific projects described in my thesis work. I would like to thank the families in the exome sequencing study of 2-ketoadipic and 2-aminoadipic acidemia for their participation and Dr. Bridget

vi Wilcken for her cooperation and support. I would like to acknowledge Dr. Elaine Spector, Dr.

Steve Sperber, and Dr. Yong Shi for their support and expertise in developing the deep sequencing method described in this thesis work and for allowing me to work with their equipment to run and test this method. I would like to acknowledge Dr. Ken Jones and Dr.

Katherine Gowan for their assistance in processing the data from both the exome sequencing and deep sequencing projects. I would like to acknowledge Dr. Kim Bjugstad for teaching me how to perform immunohistology of human putamen samples studied during my thesis work and Dr. Marc Del Bigio for providing the samples, help in interpreting the results, and for providing relevant clinical information on all the samples. Finally, I would like to acknowledge

Dr. Bruce Appel, Dr. Emily Mathews, and Dr. Maria Cattell for sharing their expertise working with a zebrafish model and teaching me the relevant techniques to complete the characterization of a glutaryl-CoA dehydrogenase deficient zebrafish.

vii TABLE OF CONTENTS

CHAPTER

I - INTRODUCTION ______1

Clinical Phenotype of Glutaric Acidemia Type I ______1

Biochemical Phenotype of Glutaric Acidemia Type I ______3

Molecular Description of Glutaric Acidemia Type I ______7

Diagnosis and Treatment of Glutaric Acidemia Type I ______9

Unanswered Questions and the Future of Glutaric Acidemia Type I ______11

Figures______15

II - FATE OF THE MEDIUM SPINY NEURONS: BRAIN HISTOLOGY OF FOUR GLUTARIC

ACIDEMIA TYPE I PATIENTS ______23

Introduction ______23

Materials and Methods ______24

Materials: ______24

Method for 3- dine (DAB) Staining of DARPP-32 Labeled Slides: ______25 Diaminobenzi Method for Crystal Violet Staining: ______26

Imaging and Counting of DARPP-32+ Cells: ______26

Cases ______27

Case 1: ______27

Case 2: ______27

Case 3: ______28

Case 4: ______28

viii Controls: ______28

Results ______29

Discussion ______30

Figures______34

III - DEEP SEQUENCING OF GCDH IN KNOWN GLUTARIC ACIDEMIA TYPE I PATIENTS

______41

Introduction ______41

Materials and Methods ______42

Samples: ______42

Isolation of DNA from Cell Pellets: ______42

Initial Long-Range PCR of GCDH from Genomic DNA: ______43

Generation of DNA Library for MiSeq: ______43

Data Analysis: ______44

GCDH Enzyme Assay: ______45

Results ______45

Discussion ______46

Figures______49

IV - MUTATIONS IN DHTKD1 CAUSE 2-KETOADIPIC ACIDEMIA AND 2-AMINOADIPIC

ACIDEMIA ______53

Introduction ______53

Another Disorder of Lysine Metabolism: 2-Ketoadipic and 2-Aminoadipic Acidemia ______54

ix Abstract ______55

Introduction ______56

Materials and Methods ______57

Case Reports ______57

Discussion and Conclusions ______59

Figure ______63

V - CHARACTERIZATION OF A GCDH DEFICIENT ZEBRAFISH ______64

Animal Models of Glutaric Acidemia Type I ______64

Zebrafish Animal Models ______66

Materials and Methods ______69

Ethics Statement: ______69

Zebrafish Lines and Husbandry: ______69

C-Start Response Testing and Developmental Assessment: ______70

Fish Homogenization for Metabolic Testing: ______70

Quantitative Amino Acid Analysis: ______70

Organic Acid Screen of Whole Zebrafish Homogenate: ______71

Glutaric Acid and 3-Hydroxyglutaric Acid Quantitation by Stable Isotope Dilution: ______72

Glutaryl-CoA Dehydrogenase Enzyme Assay: ______73

Results ______73

C-Start Response Testing: ______73

Quantitative Amino Acid Analysis: ______74

Organic Acid Screen: ______74

x Glutaric Acid and 3-Hydroxyglutaric Acid Quantitation by Stable Isotope Dilution: ______75

Glutaryl-CoA Dehydrogenase Enzyme Assay: ______75

Discussion ______76

Figures______79

VI - CONCLUSIONS AND FUTURE DIRECTIONS ______85

Summary and Future Directions ______85

Striatal Damage in GA1 ______85

Diagnosis of GA1 ______87

DHTKD1 and 2-Ketoadipic and 2-Aminoadipic Acidemia ______89

Towards a Zebrafish Model of GA1 ______90

Conclusion ______92

REFERENCES______94

APPENDIX A: DARPP-32 STAINING IN THE PUTAMEN OF REPRESETITIVE CONTROLS

AND GA1 BRAINS ______111

APPENDIX B: TABLE OF PUBLISHED DESCRIPTIONS OF 2-KETOADIPIC AND

2-AMINOADIPIC ACIDEMIA PATIENTS ______115

xi LIST OF TABLES

TABLE

1: Acyl-CoA Dehydrogenases ______21

2: Clinical Data of Patients with Glutaric Acidemia Type I ______34

3: Long-Range PCR Primer Sets ______49

4: Long-Range PCR Programs ______50

5: Mutations Identified by Deep Sequencing ______51

6: Confirming Previously Identified Mutations ______52

7: Summary of the Clinical Findings of Patients with DHTKD1 Deficiency ______63

8: Quantified Amino Acids in Zebrafish ______80

xii LIST OF FIGURES

FIGURE

1: A 13-Month Old with Glutaric Acidemia Type I ______15

2: Axial Scans of Glutaric Acidemia Type I Patients ______16

3: Saccharopine Pathway of Lysine Oxidation ______17

4: Pipecolate Pathway of Lysine Oxidation ______18

5: Protein Structure of Glutaryl-CoA Dehydrogenase ______19

6: GCDH Enzymatic Reaction ______20

7: Brain Scans Before and After Treatment ______22

8: DARPP-32 Stained Putamen in a 7-Month Old Control ______35

9: 7-Month Old Control Putamen ______36

10: 7-Month Old GA1 Putamen ______37

11: Higher Magnification of 7-Month Old GA1 Putamen ______38

12: Mean Number of DARPP-32 Neurons per 383.5µm2 in Putamen of GA1 Patients vs.

Controls ______39

13: Mean Number of DARPP-32 Neurons per 383.5µm2 in Putamen Separated by Age

______40

14: Structure of GCDH in Humans, Mice and Zebrafish ______79

15: Organic Acid Screen of a Wild Type Fish vs. a Homozygous Fish ______81

16: Quantified Glutaric in Zebrafish ______82

17: Quantified 3-Hydroxyglutaric Acid in Zebrafish ______83

18: Specific Activity of GCDH in Whole Zebrafish ______84

xiii List of Abbreviations

Abbreviations

(BSA) Bovine Serum Albumin

(BSTFA/TMCS) N,O-Bis(trimethylsilyl)trifluoroacetamide/trimethylsilylchloride

(CRISPR) Clustered Regularly Interspaced Short Palindromic Repeats

(C5DC) Glutarylcarnitine

(DARPP-32) Dopamine and cAMP Regulated Neuronal Phosphoprotein

(DHTKD1) Dehydrogenase E1 And Transketolase Domain Containing 1

(Protein)

(DHTKD1) Dehydrogenase E1 And Transketolase Domain Containing 1

(Gene)

(DTT) 1,4-dithiothreitol

(ETF) Electron Transfer Flavoprotein

(ETF-QO) Electron Transfer Flavoprotein Dehydrogenase

(EtOH) Ethanol

(ExAC) Exome Aggregation Consortium

(FAD) Flavin Adenine Dinucleotide

(GA) Glutaric Acid

(GA1) Glutaric Acidemia Type 1

(GA3) Glutaric Acidemia Type 3

(GCDH) Glutaryl-CoA Dehydrogenase (Protein)

(GCDH) Glutaryl-CoA Dehydrogenase (Gene)

(GC/MS) Gas Chromatography-Mass Spectrometry

xiv (H&E) Haematoxylin and Eosin Staining

(JIMD) Journal of Inherited Metabolic Disease

(KAA) 2-Ketoadipic Acidemia

(KOH) potassium hydroxide

(MADD/GA2) Multiple Acyl-CoA Dehydrogenase Deficiency/Glutaric Acidemia

Type 1

(MCAD) Medium-Chain Acyl-CoA Dehydrogenase

(PB) Phosphate Buffer

(PBt) Phosphate Buffer with 0.1% Triton X-100

(PHHI) Familial Hyperinsulinemic Hypoglycemia

(PIPOX) Pipecolic Acid Oxidase (Protein)

(PMS) phenazine methosulfate

(SCHAD) Short-Chain 3-Hydroxyacyl-CoA Dehydrogenase

(TALENs) Transcription-Activator-Like Endonucleases

(TCA) Tricarboxylic Acid Cycle

(UTR) Untranslated Region

(WES) Whole-Exome Sequencing

(ZFNs) Zinc Finger Nucleases

(2-AA) 2-Aminoadipic Acid

(2-HA) 2-Hydroxyadipic Acid

(2-KA) 2-Ketoadipic Acid

(3-OH) 3-Hydroxyglutaric Acid

xv CHAPTER I

INTRODUCTION

Clinical Phenotype of Glutaric Acidemia Type I

Glutaric acidemia type I (GA1) is an inborn error of lysine and tryptophan metabolism caused by a deficiency of glutaryl-CoA dehydrogenase (GCDH). The first description of this disease was a case report in 1975 describing two affected siblings, both of whom had experienced a period of normal development early in life followed by neurological deterioration evidenced by spasticity, loss of head control, and inability to sit without support 1. The siblings had normal amino acids in urine, elevated glutaric acid in urine which could be altered by dietary changes, and the inability to metabolize glutaryl-CoA to CO2 in peripheral leukocytes.

With the identification of more GA1 patients, a clear disease course emerged.

Most GA1 patients have a clinical asymptomatic period before suffering an encephalopathic crisis during or shortly after an illness, vaccination, or surgery 2 4. – Crises typically occur between three months and three years of age and results in acute striatal necrosis causing dystonia, dyskinesia, hypotonia, loss of head control, and seizures (See Figure 1) 2 4. – Several reports have used CT or MRI scans to describe the brains of GA1 patients before and after the encephalopathic crisis. Scans taken before a crisis indicate that there is often aberrant brain development evident as early as two weeks of life 3,4. GA1

incomplete development of the frontal and temporalbrains show lobes. an atrophyWhere brain phenotype matter doeswith not fill the cranial cavity, there is an accumulation of cerebrospinal fluid, primarily around the frontal lobes, the base of the

1 brain and in the expanded Sylvian fissures (See Figure 2). CT and MRI scans taken after encephalopathic crises usually show bilateral atrophy of both striatal structures, the caudate and putamen, in addition to the initial dysmorphology. Damage from the crisis is mostly localized to the caudate and putamen, with the globus pallidus and thalamus usually spared 4. Histological examination of postmortem GA1 brains shows that the loss of the striatum is due to the death of medium-sized neurons 5. Because most medium sized neurons in the striatum are medium-spiny neurons, this is the population thought to be selectively affected 5. This disease course suggests a sudden, toxic neurological event that primarily affects the striatum. Additionally, the striatum appears to have a distinct period of development during which it is most at risk.

There are several, uncommon, atypical presentations of GA1. Another presentation of GA1 is a slow decline of motor function and progressive developmental delay without an acute crisis 6,7. This is the most common atypical presentation of GA1, and in one study this presentation was reported to occur in approximately 33% of all

GA1 patients 8,9. The second most common atypical presentation occurs in approximately 5% of GA1 patients. This group has elevated metabolite levels and a reduced GCDH enzyme activity but remain clinically asymptomatic without treatment

4,8. These patients are usually identified when a child or younger sibling is diagnosed with GA1 and screening is performed for the entire family. Finally, a small number of case reports are appearing in the literature that describe patients who experience a late onset encephalopathic crisis. Most untreated patients experience a crisis before age three, and late onset has been defined as experiencing an encephalopathic crisis after age six 4,8. In the few case reports describing late onset GA1, patients present with

2 symptoms not typically seen in childhood GA1; these include vertigo, headaches, dementia, tremors, epilepsy, and problems with coordination8. All late onset patients are reported to have white matter changes, predominantly periventricular and in the frontal and parietal lobes8. With these atypical presentations, the timing of striatal necrosis can be markedly different to what is seen in a classical presentation, the way the brain is affected can be identical or vastly different, and the acute nature of the encephalopathic crisis can be lacking altogether.

Biochemical Phenotype of Glutaric Acidemia Type I

The lysine oxidation pathway is highly conserved in all vertebrates. This pathway is responsible for the breakdown of L-lysine, hydroxy-L-lysine, and

L-tryptophan into progressively smaller molecules generating energy, and acetyl-CoA to be used in the tricarboxylic acid (TCA) cycle. Both L-lysine and L-tryptophan are essential amino acids; meaning that they cannot be synthesized by the body and must instead come from diet. Of the three amino acids, L-lysine is the most abundant dietary protein andones thus contributes the most to this pathway. In humans, lysine oxidation can occur in either of two tissue-specific pathways. The saccharopine pathway occurs in the liver and kidneys, accounting for the majority of lysine oxidation in a human body 10,11. The enzymes specific to this pathway are targeted to the mitochondria (See Figure 3) 12 15. The pipecolic acid pathway is only active in human – brain tissue, and the enzymes specific to this pathway are targeted to the peroxisomes

(See Figure 4) 14 18. Interestingly, the cellular location of the pipecolic pathway varies – between mammals with pipecolic acid oxidase (PIPOX) displaying different subcellular localization in a species-dependent manner 19. These pathways both make

3 2-aminoadipate-6-semialdehyde which is transported into the cytoplasm. After this step, there is only one pathway to finish converting the semialdehyde to acetyl-CoA. In glutaric acidemia type I (GA1) there is a block in the shared pathway at glutaryl-CoA dehydrogenase (GCDH). Biochemically, GA1 is characterized by elevations of glutaric acid (GA) and 3-hydroxy glutaric acid (3-OH) throughout the body. The GA comes from glutaryl-CoA, but the origin of the 3-OH is unknown.

Biochemically, GA1 patients can be divided into two groups: high metabolite excreters (GA > 100mmol/mol creatinine) and low metabolite excreters

(GA < 100mmol/mol creatinine) 9,20 23. This distinction only refers to the amount of GA – excreted in the urine of a patient and does not consider the amount of GA or 3-OH in the body or brain. Each group is thought to have the same risk of striatal necrosis and the same severity of outcome when a crisis occurs. Treatment for both groups is identical, but with current screening protocols, it is more problematic to detect a low excreter than a high excreter 24. A 2010 study suggests that around 5% of all GA1 patients are low excreters 9. Genotype has not been found to correlate with the severity of the clinical phenotype 7,22,25, but genotype does correlate with the high excreter or low excreter metabolic phenotypes 4,23. Mutations associated with the low excreter biochemical phenotype are found throughout GCDH, and it is not apparent why these mutations cause altered biochemistry. Commonly, low excreters have missense mutations on at least one GCDH allele and a higher residual enzyme activity than high excreters 6,26.

There are several disorders which can cause an elevated GA or an elevated 3-OH, but only in GA1 are both compounds elevated 4,22. Elevated GA is observed in patients

4 with glutaric acidemia type II (MADD/GA2), also known as multiple acyl-CoA dehydrogenase deficiency, in addition to elevated levels of 2-hydroxyglutaric acid, sarcosine, and the C4, C5, C6, C8, C14, C16, and C18 acylcarnitines with a reduction in

C2 acylcarnitine 27,28. GA2 occurs when there is a deficiency of electron transfer flavoprotein (ETF) or electron transfer flavoprotein dehydrogenase (ETF-QO). ETF is composed of alpha and beta subunits, and GA2 patients have been found to have mutations in either subunit as well as in ETF-QO 29 32. ETF functions as an electron – accepter for GCDH (and other related acyl-CoA dehydrogenase enzymes) allowing

GCDH to then bind a new molecule of glutaryl-CoA 33 35. ETF then passes these electrons – to ETF-QO which will then pass them to ubiquinone and the electron transport chain

36,37. 26,28,38

Isolated, elevated GA is also seen in glutaric acidemia type III (GA3) which is caused by mutations in C7orf10 39,40. Initially thought to cause a disease phenotype, GA3 is now known to be a benign metabolic abnormality 39,41 . C7orf10 encodes a succinate-hydroxymethylglutarate CoA-transferase which is responsible for converting glutarate to glutaryl-CoA 40. In rare cases elevated GA has been observed in short-gut syndrome 42, and in patients with an unusual gut microbiome 43.

Elevated 3-OH, originally thought to only occur in GA1, has been observed in familial hyperinsulinemic hypoglycemia (PHHI) when caused by a deficiency of short-chain 3-hydroxyacyl-CoA dehydrogenase (SCHAD) 44. PHHI is a polygenic disorder, but elevated 3-OH has only been observed in patients with mutations in

SCHAD. One family described with PHHI due to SCHAD deficiency did not have clinical findings typically seen with SCHAD deficiency; instead, they display a metabolite profile

5 consistent with hyperinsulinism. The presence of elevated 3-OH in these patients remains an enigma with no clear explanation for its origin.

Because GA1 may involve neuronal toxicity and because GA and 3-OH are elevated in patients, these two compounds are candidates for the neurotoxicity in the brain. Evidence supporting this comes from a disparate set of experiments, all using different methodologies, making them difficult to compare. Cell culture experiments have demonstrated that both GA and 3-OH are toxic to brain cells and often 3-OH is highlighted as the more toxic metabolite 45. For these cell culture experiments the concentrations of GA and 3-OH are much higher than the levels reported in the blood or urine of GA1 patients. Because of this, it is unclear if the toxicity described in these experiments is biologically relevant to GA1 patients. Other cell culture and mouse experiments have shown that striatal astrocytes produce GA and 3-OH in response to lysine exposure. This acute exposure causes astrocyte dysfunction and a subsequent astrocyte-mediated neuronal death 46,47. These findings suggest that GA is not itself a neurotoxin but that elevated GA levels cause astrocytes to release some unknown toxin.

Finally, studies using the GCDH -/- mouse model show that only GA levels increase proportionally to the amount of lysine in the diet and that this accumulation is necessary to cause acute striatal damage 48 50. Taken together, this evidence remains – inconclusive in identifying the neurotoxin. There is evidence that both GA and 3-OH can function as neurotoxins; however, the concentration of each used in these studies often is much higher than peripheral levels of GA and 3-OH in patients. Because of this, these study results do not necessarily reflect what is happening in the brains of GA1 patients.

6 So, the toxin responsible for the striatal necrosis seen in GA1 patients remains unknown.

Molecular Description of Glutaric Acidemia Type I

Glutaric acidemia type I (GA1) is one of the more common organic acidemias in the world with a frequency of 1 in 110,000 live births 2,4. GA1 is an autosomal recessive disease with more than 240 unique mutations documented in glutaryl-CoA dehydrogenase (GCDH). Most mutations are missense changes (80.7%), a handful are small (no larger than 20bp) insertion/deletion events (13.6%) and the remainder are splice site mutations (5.8%) (Personal Communication with Drs. Steve Goodman and

Elaine Spector, University of Colorado Anschutz Medical Campus). Most GA1 patients have two different mutations and are thus compound heterozygotes 25,51. No patient has been reported region, or in thewith introns pathogenic of GCDH mutations. There are in several the promoter genetically region, isolated untranslated populations with an increased incidence of GA1, probably due to founder effects. These populations include the Oji-Cree First Nations in Manitoba and Ontario (IVS1+5 G>T; carrier frequency of 1:8, approximately an incidence of 1:235) 52,53, the Old Order Amish

Community in Pennsylvania (A421V; carrier frequency of 1:12, approximately an incidence of 1:500) 54,55, the Lumbee Indian Tribe of North Carolina (E414K, carrier frequency of 1:16, approximately an incidence of 1:1000) 56, and the Xhosa subgroup of the South African black population (A293T; carrier frequency of 1:36, approximately an incidence of 1:5100) 57.

GCDH maps to chromosome 19p13.2 and encodes an 11-exon transcript spanning 7KB 3,26,54,58,59. The mRNA is translated into a 438-amino acid precursor

7 peptide containing a 44-amino acid mitochondrial localization signal at its amino terminus, and this is cleaved after import into the mitochondrial matrix to a 394-amino acid mature protein of 43.5 kDa 60 62. Individual GCDH subunits combine to form – homodimers, and then into homotetramers as a dimer of dimers, forming the functional enzyme (See Figure 5) 3,63. Each of the four subunits has an active site and a non-covalently bound molecule of flavin adenine dinucleotide (FAD) which acts as a redox cofactor 26,63.

The enzymatic function of GCDH is the oxidative decarboxylation of glutaryl-CoA to crotonyl-CoA, thought to occur in a two-step process involving enzyme-bound glutaconyl-CoA 4,26. The electrons removed from the 2 and 3 carbon atoms of the glutaryl group are transferred first to FAD in GCDH, and then to FAD in electron transfer flavoprotein (ETF), a small (63 kDa) protein in the mitochondrial matrix 64 (See

Figure 6). ETF has its own dehydrogenase, and the electrons continue to pass to ubiquinone, entering the electron transport chain 26,38,65,66. GCDH is expressed in most cells of the human body but has the highest expression in liver and kidney 67. Within the brain, the highest levels of expression are in the cerebellum followed by the striatum and cortex 48,67.

GCDH is part of the gene family of acyl-CoA d (See Table

1). All members of this gene family require FAD as aehydrogenases cofactor, are dependent ACADs on ETF to cycle electrons, and function in the mitochondrial matrix 4. Most function as homotetramers (resembling a dimer of dimers like GCDH) and a few as homodimers.

Five three (including GCDH) function in aminoof the acid ACADs oxidation. function All proteinin fatty acidsequences oxidation and and structures for these enzymes are

8 related, but GCDH has the greatest homology with the medium-chain acyl-CoA dehydrogenase (MCAD)4.

Diagnosis and Treatment of Glutaric Acidemia Type I

Before the adoption of newborn screening, GA1 patients were only identified once they presented with dystonia and dyskinesia after striatal necrosis had occurred.

Today it is possible to diagnosis GA1 patients before an encephalopathic crisis due to elevated levels of glutaric acid (GA) and 3-hydroxy glutaric acid (3-OH) in the blood and urine, which are present from birth 2,68. In many countries throughout the world, screening of acylcarnitines from dried blood spots is performed on all newborns to detect a variety of inborn errors of metabolism 69,70. Using this method, GA1 is suspected by the presence of elevated levels of glutarylcarnitine (C5DC), but this finding must be confirmed by additional diagnostic testing. A diagnosis of GA1 can be bolstered by quantitative analysis of GA and 3-OH by gas chromatography-mass spectrometry

(GC/MS), mutation analysis of glutaryl-CoA dehydrogenase (GCDH), enzyme activity assay, and clinical findings including macrocephaly, encephalopathy, basal ganglia injury, white matter disease, and/or movement disorder 2. All these findings can point to a diagnosis of GA1, but current guidelines state that a diagnosis can only be confirmed by reduced enzyme activity and/or detection of disease-causing mutations on both GCDH alleles. Newborn screening excels at detecting high excreters but often fails to detect low excreters. Many low excreters have C5DC levels more commonly attributed to GA1 carriers making it difficult to set a threshold value of C5DC that can identify all true positives without also flagging an excessive number of false positives.

9 Current newborn screening programs can detect 95% of GA1 newborns with the 5% missed largely comprised of low excreters 2.

While a GA1 diagnosis is being confirmed, it is important to start treatment to prevent any possible crisis 2. Recommended treatment includes a diet low in lysine (or protein), carnitine supplementation, and emergency treatment during catabolism 2,71. A low lysine diet is beneficial because lysine is an essential amino acid and by reducing the total lysine intake, one limits the glutaryl-CoA reaching the metabolic block 2,48,50.

Supplementation of carnitine has two postulated effects in patients. First, it helps remove glutaric acid from the body through urinary excretion of glutarylcarnitine.

Second, receiving extra carnitine is thought to prevent the body from depleting the supply of available carnitine due to the accumulation and loss of glutarylcarnitine. The final aspect of treatment, referred to as an emergency treatment, is used because prior to newborn screening, it was observed that most cases of encephalopathic crisis occurred during or shortly after an illness, vaccinations, or surgery 2. It was suggested that extended periods of catabolism increased usage of the oxidation pathway, increasing the concentration of a toxic product, and thus triggering necrosis of the striatum. An emergency treatment protocol for GA1 patients now includes glucose administration to prevent catabolism. With this combined treatment, the frequency of encephalitic crisis decreases to 10-20% vs. 80-90% without treatment 2.

Several studies have examined the effectiveness of treating GA1 patients before striatal necrosis occurs. Of the three aspects of recommended treatment, adherence to the emergency treatment protocol is thought to be the most important. One study found that only 5% of patients closely following the treatment protocol developed dystonia,

10 while 44% of those who did not strictly follow the dietary treatment developed dystonia, and those who did not follow the emergency treatment protocol all developed dystonia 9. When striatal necrosis is prevented the brains of GA1 patients demonstrate an ability to correct for the dysmorphology observed early in life. MRI and CT scans of

GA1 brains taken early in life reveal that there is abnormal brain development in utero preceding any damage caused by an encephalopathic crisis (See Figure 2). After a few years of adhering to treatment, scans show brain growth of the frontal and temporal lobes, replacing the cerebrospinal-fluid-filled regions with brain tissue (See Figure 7)

22,72,73. While never resembling a normal age matched brain, this phenomenon demonstrates some ability for the brains of patients to heal if a crisis can be prevented.

Unanswered Questions and the Future of Glutaric Acidemia Type I

With more than 40 years of work, clinicians and researchers have answered many questions about glutaric acidemia type I (GA1), but much more remains unknown. Currently, there is a good description of the different pediatric presentations of GA1, but reports of late onset cortical neuropathy 8,74 and kidney damage 2 are just starting to appear in the literature, and it remains unknown for how long patients are at risk of developing symptoms.

There is no clear reason why the caudate and putamen are often solely affected by encephalopathic crisis in patients, or why there is a specific developmental period during which these regions are most vulnerable. This striatal necrosis following encephalopathic crisis has been replicated in a mouse model of GA1 which show that younger animals are more at risk of striatal damage when exposed to known neuronal toxins, elevated levels of glutaric acid (GA) or elevated levels of 3-hydroxyglutaric acid

11 (3-OH). Additionally, it is not known which type or types of cells in the caudate and putamen are damaged by . There are studies asserting that neurons undergo apoptosis as a direct consequencethe toxin of exposure to high GA and/or 3-OH levels and yet other studies show that astrocyte activation results from elevated metabolite levels and this results in neuronal death. To further obfuscate the picture, the identity of the toxin responsible for striatal necrosis is not known.

It is not understood why some mutations cause a low excreter phenotype while others do not. Mutations associated with the low excreter phenotype are dispersed through GCDH, and it is not understood how these mutations affect enzyme activity.

Mutations associated with the high excreter phenotype are also dispersed through

GCDH but appear to reduce the structural stability of the enzyme resulting in a reduced amount of functional enzyme in individuals. Additionally, it is not known how different mutations interact with each other. The GCDH enzyme is composed of four distinct subunits, and each enzyme should be comprised of subunits with different mutations in compound heterozygous patients.

While the elevated GA level is easily explained due to the inability of patients to process glutaryl-CoA into crotonyl-CoA, there is no clear explanation or evidence for the origin of the elevated 3-OH in these patients. There are several variations of a theory which says one of the other acyl-CoA dehydrogenases can act on glutaryl-CoA to generate the 3-OH, but this has never been proven experimentally.

Above all else, the pathogenesis of GA1 remains a mystery. Many theories have been suggested, but there is no substantial evidence for any one theory so far. In the following work, I will describe four studies each designed to answer some unexplored

12 aspect of GA1. Chapter two describes a study using immunohistology to examine the putamen of GA1 patients and controls to determine if medium spiny neurons are selectively lost in GA1 patients following an encephalopathic crisis. Chapter three describes a study of GA1 cell lines analyzed using deep sequencing to determine the mutation spectrum of GCDH untranslated region (UTR) ofand GCDH to .search Chapter for four mutations details ina whole the exomepromoter study and of a family with a different block in the lysine oxidation pathway to identify an upstream enzyme that could be a therapeutic target for GA1 patients. Finally, chapter five details work done to characterize a zebrafish model deficient in GCDH A to determine if zebrafish can be used to model GA1.

In the first project described, I sought to expand our understanding of what occurs to the striatum of a GA1 patient after an encephalopathic crisis. Previous evidence has shown that medium sized striatal neurons are selectively killed during a

GA1 crisis. Using immunohistochemical staining, I sought to expand on this assertation and test if the number of medium spiny neurons is reduced in GA1 patients when compared to age matched controls.

In the second project described, I performed deep sequencing on a section of

DNA that covered the entire GCDH UTR for a group of patients suspected to havegene, GA1 thebut promoterwho did not region, have andtwo the identified mutations.

The goal of this particular study was to identify two pathogenic mutations in each

UTR for pathogenic mutations, and to develop a protocolpatient, to that screen would the allow promoter for robust, and and complete sequencing of multiple samples in a single test.

13 In the third project described, I focus on the enzyme directly preceding GCDH in the lysine oxidation pathway, the 2-ketoadipic acid dehydrogenase complex. Several case reports describe individuals with elevated metabolite levels one would expect from a block of this enzyme complex. While all case reports detail elevated metabolite levels, other aspects of the clinical phenotype differ with each report. This has led to the current conclusion that a deficiency of this enzyme complex causes a biochemical marker, but not a true disease. From this premise, this complex is a potential therapeutic target to be blocked in GA1 patients. To test this hypothesis, however, it is necessary to identify the enzyme which had not been previously done. I performed whole exome sequencing on a family with several siblings known to have a block in this enzyme complex to determine the genetic identity of their specific protein deficiency.

The final aspect of my thesis work was the characterization of a zebrafish with mutations in GCDH. There are already four animal models of GA1, and each is specifically geared to answer distinct questions about the disease. With the generation and validation of a zebrafish model of GA1, a different variety of hypotheses could be asked and answered. For example, a zebrafish model of GA1 could be used to interrogate the origin of the elevated 3-OH. The Sanger zebrafish mutation project 75 can supply mutants, making zebrafish an easy model system in which to generate double mutant animals. This model could then be used to test if blocks upstream of

GCDH (such as a deficiency of the 2-ketoadipic acid dehydrogenase complex) are in fact beneficial to a GA1 animal. Importantly, a zebrafish model of GA1 could be useful to address many more questions about this disease. In the following chapters, I detail all my work on the projects briefly described above.

14 Figures

Figure 1: A 13-Month Old with Glutaric Acidemia Type I. -month-old boy with GA1, showing dystonia of face, tongue, neck, back, arms, and hands. 4 Image from Chapter 95 of The Metabolic and Molecular Bases of InheritedA Disease volume 2 4.

B C

Figure 2: Axial Scans of Glutaric Acidemia Type I Patients. A: MRI showing the horizontal section of an asymptomatic GA1 newborn displaying enlarged cerebral ventricles, expansion of fluid spaces around the brain stem, and fluid collection over the frontal lobe (white arrow) and temporal lobes (blue arrow) in place of brain tissue. Image A from Strauss et al. 2003 3

B: CT showing the horizontal section of an asymptomatic 5-month old displaying a reduced frontal lobes (white arrow) and reduced temporal lobes (blue arrow). The scan shows cerebral spinal fluid occupying these spaces instead of brain matter. The heads of the caudate remains intact (red arrows). C: CT showing the horizontal section of a 3-year old who has experienced a crisis. The reduction of the frontal lobes (white arrow) and temporal lobes (blue arrow) remains in addition to reduced caudate heads (red arrows). Images B and C from Chapter 95 of The Metabolic and Molecular Bases of Inherited Disease volume 2 4.

Figure 3: Saccharopine Pathway of Lysine Oxidation. The saccharopine pathway of lysine oxidation is active primarily in liver and kidney tissues of mammals.

Figure 4: Pipecolate Pathway of Lysine Oxidation. The pipecolate pathway of lysine oxidation is active primarily in the brain tissue of mammals. In humans, the first two steps occur in the peroxisomes.

Figure 5: Protein Structure of Glutaryl-CoA Dehydrogenase. The quaternary structure of GCDH. This depicts four identical subunits of GCDH (one circled) bound to each other to form the functional enzyme. Each subunit also has a non-covalently bound molecule of flavin adenine dinucleotide (FAD) shown here with a wire structure (arrow). Image from the Protein Data Bank.

Figure 6: GCDH Enzymatic Reaction. GCDH functions to oxidatively decarboxylate glutaryl-CoA to crotonyl-CoA. During this reaction, two electrons remain bound to GCDH which are accepted by electron transfer flavoprotein (ETF). This allows GCDH to bind a new molecule of glutaryl-CoA. Figure adapted from Chapter 95 of The Metabolic and Molecular Bases of Inherited Disease volume 2 4.

Table 1: Acyl-CoA Dehydrogenases

Approved Name Gene Pathway Structure Mutations Cause Symbol acyl-CoA dehydrogenase, ACADM Fatty acid beta-oxidation tetramer MCAD Deficiency C-4 to C-12 straight chain isovaleryl-CoA IVD Leucine Catabolism dimer Isovaleric Acidemia dehydrogenase acyl-CoA dehydrogenase, ACADS Fatty acid beta-oxidation tetramer SCAD Deficiency C-2 to C-3 short chain acyl-CoA dehydrogenase, ACADL Fatty acid beta-oxidation unknown LCAD Deficiency long chain glutaryl-CoA GCDH Lysine/Hydroxylysine/ tetramer Glutaric Acidemia Type I dehydrogenase Tryptophan Catabolism acyl-CoA dehydrogenase, ACADVL Fatty acid beta-oxidation unknown VLCAD Deficiency very long chain acyl-CoA dehydrogenase, ACADSB Fatty acid beta-oxidation unknown 2-Methylbutyrlglycinuria short/branched chain and Isovaleric Acidemia acyl-CoA dehydrogenase ACAD8 Branch chain amino acid tetramer Isobutryl-CoA family member 8 catabolism Dehydrogenase Deficiency acyl-CoA dehydrogenase ACAD9 Structural Protein unknown Mitochondrial Complex 1 family member 9 Deficiency Due to ACAD 9 Deficiency acyl-CoA dehydrogenase ACAD10 unknown unknown unknown family member 10 acyl-CoA dehydrogenase ACAD11 unknown unknown unknown family member 11

Figure 7: Brain Scans Before and After Treatment. A: MRI showing the horizontal plane of 5-week old displaying wide Sylvian fissures (white arrows). B: Same patient at 4-years showing resolution of the enlargement. This patient was treated, starting when first identified, and never experienced an encephalopathic crisis or striatal necrosis. Image from Viau et al. al. 2012 22.

22 CHAPTER II

FATE OF THE MEDIUM SPINY NEURONS: BRAIN HISTOLOGY OF FOUR GLUTARIC

ACIDEMIA TYPE I PATIENTS

Introduction

Several reports have used CT or MRI scans to describe the brains of glutaric acidemia type 1 (GA1) patients before and after an encephalopathic crisis. Scans taken before a crisis reveal that there is often aberrant brain development evident as early as two weeks of life 3,4 incomplete development of the . frontalGA brains and temporalshow an atrophylobes and phenotype accumulation with of cerebrospinal fluid in these areas (See Figure 2). Patient brains will often improve with treatment 22, but do not normalize completely. There is a substantial reduction in the cerebrospinal fluid of later scans, with brain matter continuing to grow and fill in the cranial cavity 72 (See

Figure 7).

Scans taken after encephalopathic crisis show the same dysmorphology seen in scans taken before a crisis but with additional bilateral atrophy of the caudate and putamen. New alterations are localized to these two regions of the striatum with the globus pallidus and thalamus typically spared from damage 4. To better understand the pathogenesis of GA1 a 2005 study by Funk et al.5 used haematoxylin and eosin (H&E) stained sections of postmortem brains of several GA1 patients to identify which cells are lost following an encephalopathic crisis. All GA1 samples used in the study came from patients of the Oji-Cree First Nations tribe in Manitoba, all were homozygous for the IVS1+5 G>T mutation 53, and all had striatal atrophy resulting from an acute encephalopathic crisis. Neurons were counted in each sample and grouped by size as

23 small, medium and large. When compared to age matched controls, all GA1 samples displayed a significant loss of medium sized neurons in the caudate and putamen

P<.. The authors concluded thatlikely all occurs cases withinhad near a few a complete weeks, at loss most, of mediumof the initial crisisneurons then from ceases the tostriatum progress that further 5.

Within the striatum, 95% of all neurons are medium-spiny neurons 76 and as such the Funk et al. study findings supported a previous assertion that the medium-spiny neurons are preferentially targeted and destroyed during an encephalopathic crisis. Medium spiny neurons have not been specifically examined in

GA1 striatum samples, so I set out to test this by staining postmortem samples for dopamine and cAMP regulated neuronal phosphoprotein (DARPP-32), a protein expressed by medium spiny neurons in the striatum77. I examined postmortem slides of putamen from GA1 and age matched controls and took several images along the same region of the putamen in all samples. An unpaired t-test was used to determine if the distribution of positively stained neurons differed between GA1 samples and controls.

Materials and Methods

Materials:

Case notes and slides were provided by Dr. Marc Del Bigio, from the University of Manitoba School of Medicine in Winnipeg, a co-author on the Funk et al. paper 5.

Slides bore paraffin-fixed mounted autopsy samples from the putamen; see cases for more information. Anti-DARPP-32 antibody was purchased from Abcam (product ab40801). Secondary goat anti-rabbit antibody (BA-1000) and an ABC-peroxidase kit

24 were purchased from Vector Laboratories. Crystal Violet staining solution was purchased from Fisher Scientific (C581-25).

Method for 3- DAB) Staining of DARPP-32 Labeled Slides:

Slides were deparaffinizedDiaminobenzidine by heating to 45°C to 57°C for 30 minutes followed by immersion in 100% xylene for 10 minutes at room temperature. Samples were then rehydrated by descending serial xylene/alcohol solutions (100% xylene, 1:1 of xylene:ethanol, 100% ethanol (EtOH), 95% EtOH in water, 70% EtOH in water, 50%

EtOH in water) to distilled water. Antigen retrieval was performed by placing slides in a

10mM citric acid buffer (pH 6.0) for 1 hour. The solution was preheated to 90°C, and the slides were maintained at this temperature for 30 minutes. The slides, still in the buffer solution, were then removed from the heat and left at room temperature for another 30 minutes. The final temperature was 45°C. A protein blocker solution consisting of 5% goat serum and 1% bovine serum albumin (BSA) in phosphate buffer with 0.1% Triton

X-100 (PBt) was placed on the slides for 1 hour. After rinsing in phosphate buffer (PB), the slides were placed in a carrier solution (1:1 PB and protein blocker described above) containing the primary antibody at a dilution of 1/7000 which was left on the slides overnight at 4°C. After overnight exposure, slides were rinsed and placed in a carrier solution (1:1 PB and protein blocker) containing the secondary antibody at a dilution of 1/500 which was left on the slides for 1.5 hours. Slides were rinsed and placed in an ABC-peroxidase solution for 2 hours. After another rinse, the slides were exposed to a pH 8.6 sodium acetate solution for one minute, then placed in a sodium acetate solution with added DAB, nickel sulfate, and H2O2 (10 ml sodium acetate buffer with 25 mg nickel sulfate, 4 mg DAB, and 3.3 µL H2O2) for about 40 seconds. The

25 reaction was stopped by placing slides in a sodium acetate solution for 5 minutes. After staining, slides were dehydrated through ascending serial alcohol solutions to xylene

(50% EtOH in water, 70% EtOH in water, 95% EtOH in water, 100% EtOH, 50/50 of

100% xylene and 100% EtOH, 100% xylene) and coverslipped with permount.

Method for Crystal Violet Staining:

For all GA1 samples and the 7-month control sample, a single slide was stained with Crystal Violet to aid in proper identification of the putamen in the DAB stained slides. Slides were deparaffinized by heating to 45°C to 57°C for 30 minutes followed by immersion in 100% xylene for 10 minutes at room temperature. Next, slides were rehydrated by descending serial xylene/alcohol solutions (100% xylene, 1:1 of xylene:ethanol, 100% ethanol (EtOH), 95% EtOH in water, 70% EtOH in water, 50%

EtOH in water) to water, followed by exposure to Crystal Violet stain solution for 30 minutes. Slides were then dehydrated through ascending serial alcohol solutions to xylene (50% EtOH in water, 70% EtOH in water, 95% EtOH in water, 100% EtOH,

50/50 of 100% xylene and 100% EtOH, 100% xylene) and coverslipped with permount.

Imaging and Counting of DARPP-32+ Cells:

Twelve images were taken at 20x magnification (383.5µm x 383.5µm) in the putamen along the external capsule white matter tract (See Figure 8). Six of the images were taken from the most dorsal point in the putamen to the medial putamen, and six were taken from the most ventral point in the putamen to the medial putamen. Four people blinded to patient genotype counted positively stained neurons in each image, and the average cell count was used for each image. For each GA1 patient, a single side was stained and counted. For all age match controls, two slides were stained and

26 counted. An unpaired t-test was used to determine if the number of stained neurons differed between dorsal and ventral regions of GA1 samples and age matched controls.

An unpaired t-test was used to determine if the number of stained neurons differed between a GA1 sample and an age matched control as well as to determine if the number of stained neurons differed between GA1 patients and controls.

Cases

Case 1:

Patient presented at 6 months with bronchiolitis, respiratory failure, increased muscle tone, and dystonic posturing. CT scan showed fluid collections anterior to the temporal lobes with preservation of the caudate nucleus. MRI showed increased signal and restricted diffusion in the caudate and putamen with widened Sylvian fissures and slight enlargement of the lateral ventricles. Clinical diagnosis was acute striatal necrosis. Sequencing showed homozygosity for the IVS1+5 G>T mutation and the patient had a urine glutaric level of 388 micromol/mM creatinine. Death was at 7 ½ months, approximately 10 weeks after the encephalopathic crisis.

Case 2:

This patient is case 4 in Funk et al. 2005 5 and case 3 in Greenberg et al. 2002 78.

Patient was diagnosed with GA1 by genetic screening and was homozygous for the

IVS1+5 G>T mutation. At 5.5 months, patient developed a fever with the onset of dystonia and athetoid limb movements as well as seizure activity. CT showed enlarged frontal horns of the lateral ventricles and widened Sylvian fissures, but no generalized atrophy. Caudate atrophy was worse at 10 months. At 15 months, patient was unable to

27 sit but had some head control and visual interaction. Death was at 18 months during a febrile illness.

Case 3:

Patient was diagnosed with GA1 by newborn screening and was homozygous for the IVS1+5 G>T mutation. At 7 weeks, a right-hand tremor was noted and urine glutaric acid levels were 500 micromol/mM creatinine. At 2 months, patient suffered pneumonia and seizures followed by mild encephalopathic change with regression of neurological function at 9 months. An EEG at 13 months showed epileptiform activity.

At 16 months, no dystonia was noted but by 2.5 years she had axial hypotonia, generalized dystonic posturing following stimulation, and severe developmental delay.

Death was at 5 years of age.

Case 4:

Patient presented at 11 months with fever, vomiting, diarrhea, lethargy,

andopisthotonic bilateral posturingabnormalities and seizures.in the basal MR) ganglia, showed periventricular atrophic anterior white temporalmatter, and lobes cerebral peduncles. Sequencing showed homozygosity for the IVS1+5 G>T mutation and urine glutaric was 33-73 micromol/mM creatinine. Cognitive and motor delay, choreoathetoid movements and dystonia were prominent throughout life. Death was at

12-years of age.

See Table 2 for a summary of all cases.

Controls:

Control 1 is a 7-month-old male who died of sudden infant death syndrome. The post mortem interval was likely 36 hours, and no neuropathologic changes were

28 observed. Control 2 is an 18-month old male who died of sudden infant death syndrome. The post mortem interval was 35 hours, and no neuropathologic changes were observed. Control 3 is a 4.5-year-old female who died from a traumatic chest injury. The postmortem interval was 34 hours, and no neuropathologic changes were observed. Control 4 is a 12-year-old female who committed suicide by hanging. The postmortem interval was 32 hours, and no neuropathologic changes were observed.

Results

Immediately apparent in all GA1 samples is the destruction of the organization and structure within the putamen compared to control samples. Typically, the putamen has a striated appearance due to grey matter interspersed with white matter tracts (See

Figure 8); distinctive myelin tracts can be seen in all control samples in addition to discretely stained medium spiny neurons(See Figure 9). Patient samples lack the interspersed white and gray matter, and landmarks denoting the putamen are destroyed or altered, making it difficult to identify the region (See Figure 10). To ensure that images were taken in the putamen along the external capsule white matter tract for the GA1 samples, a slide containing an immediately adjacent section of putamen was stained with Crystal Violet and examined by a neuropathologist. In one

GA1 sample, a large number of astrocytes were stained by DARPP-32 (See Figure 11).

This phenomenon has been noted in samples from patients who died shortly after encephalopathic crises 5 and is thought to result from ingestion of neurons by reactive astrocytes. The overall volume of the putamen was described as markedly reduced in samples from all GA1 patients by a neuropathologist, and microscopic examination showed that there was severe atrophy of the dorsal regions of the caudate nucleus

29 and/or putamen with less atrophy seen in the ventral regions (personal communication with Dr. Marc Del Bigio).

The average number of DARPP-32 positive neurons was low in all samples, and there was no difference between the number of neurons stained in the dorsal versus ventral putamen for GA1 patients compared to controls when analyzed by an unpaired t-test (t(70)= 0.80, p=0.42 for dorsal, t(70)=0.15 , p=0.88 for ventral). There is no difference in the number of positively stained neurons between GA1 samples and controls when analyzed by an unpaired t-test (t(142)=0.44, p=0.66). The average number of DARPP-32 positive neurons for all controls was 2.1 neurons/ 383.5µm2 compared to 2.3 neurons/ 383.5µm2 for all GA1 samples (See Figure 12). Comparing individual age matched samples, there was no difference in the average number of

DARPP-32 positive neurons in autopsy samples of age matched cases and controls at

7-months, 5-years or 12-years of age (unpaired t-test: t(34)=0.56, p=0.58, t(34)=0.82, p=0.42, and t(34)=1.37, p=0.18 respectively) (See Figure 13). The 18-month old GA1 autopsy sample had a significant increase of DARPP-32 positive neurons compared to an age matched control, determined by an unpaired t-test (t(34)=3.87, p=0.0005). The

18-month old GA1 sample had an average of 2.5 neurons/ 383.5µm2 in the putamen vs. an average of 0.6 neurons/ 383.5µm2 in the control sample (See Figure 13).

See Appendix A for representative images of DARPP-32 positive staining in age matched autopsy cases and controls.

Discussion

The literature describes a loss of medium spiny neurons due to encephalopathic crisis in GA1 patients4,5,79 81. Most of the data supporting this assertion comes from a –

30 study by Funk et al.5, which used H&E staining to identify striatal neurons in GA1 patients and controls and concluded the following:

wide range in duration of survival following encephalopathic crisis, all cases had near complete loss of medium (italics added) neuronsDespite fromgenetic the homogeneity striatum, with and the exception of the nucleus accumbens and tail of the caudate. This likely occurs within, at most, a few weeks of the first encephalopathic crisis. It is important to emphasize the apparent lack of progression over the lifespan because this observation supports the idea that a single severe insult during infancy 5 createsTheir analysis the bulk only of striatal examined injury. neuronal size and found that both medium and large sized neurons were reduced in GA1 patient postmortem samples. Because 65% to 95% of all striatal neurons are medium spiny neurons 76,82,83 Funk et al. concluded by stating,

-1 preferentially targets striatal

Ourmedium-spiny results support neurons, the priorwhereas assertion cholinergic that GA neurons 5.

Medium spiny neurons have not specifically been examinedare affected in GA1 to striatum a lesser samples, extent so I set out to do this by staining postmortem samples with DARPP-32, which has been shown to stain only medium spiny neurons in the striatum 77.

All GA1 samples examined show a large reduction of putamen area (observation was not quantitated) and a reduction of total striatal volume. Striatal landmarks were obliterated because of the severe striatal atrophy in all GA1 samples (See Figure 9 and

Figure 10). In the 7-month old GA1 sample, two populations of cells were positively stained by DARPP-32. One population of cells were medium spiny neurons, and the other were reactive astrocytes. Reactive astrocytes have been observed in postmortem samples of other GA1 patients who have died shortly after encephalopathic crises 5. It was previously hypothesized that reactive astrocytes ingest damaged neurons following encephalopathic crisis in all GA1 patients. Astrocytes do not express DARPP-

31 32, and the most reasonable explanation for positive staining of these cells is that the stained astrocytes had phagocytosed damaged medium spiny neurons. Positively stained reactive astrocytes were only observed in the postmortem 7-month old GA1 sample.

An unpaired t-test showed no difference in the number of positively stained medium spiny neurons between GA1 samples and controls (See Figure 12). This finding counters what is stated in the literature and suggests that medium spiny neurons are not selectively lost following an encephalopathic crisis. There was no difference in the number of DARPP-32 positive neurons in postmortem samples of age matched cases and controls at 7-months, 5-years or 12-years of age; however, the

18-month old GA1 sample shows a significant increase of DARPP-32 positive neurons compared to the age matched control (See Figure 13). This finding is unexpected since at older ages, the counts are similar. However, the number of positively stained neurons in the 18-month old control has the greatest variance of any sample examined and may not accurately represent the number of medium spiny neurons found it a typical 18- month old putamen.

The results of this study are not necessarily inconsistent with the data in Funk et al., but rather suggests that medium spiny neurons are not selectively lost. Two important points bear on this interpretation. First, my data is from a limited sample of the putamen in a few patients. For all samples used in this study, only a single cross section of the putamen was examined. Examination of multiple cross-sections from different positions in the putamen would more accurately describe any difference in the number of medium spiny neurons between cases and controls. Second, the overall

32 volume of the putamen is markedly reduced in GA1 patients; microscopic examination shows that there is severe atrophy of the dorsal regions of the caudate nucleus and/or putamen with less atrophy seen in the ventral regions (personal communication with

Marc Del Bigio). Because patients have a reduced volume of the caudate and putamen, their total number of medium spiny neurons may also be reduced.

Funk et al. suggest that medium sized neurons are selectively lost following encephalopathic crisis in GA1 patients. My findings instead suggest that the proportion of medium spiny neurons is not altered in GA1 patients following a crisis, but that all cell types within the putamen may be equally affected. Funk et al. reports their data as total number of neurons and does not account for the reduced striatal volume. This study provides preliminary data to the effect that medium spiny neurons are not selectively lost during an encephalopathic crisis. An examination of multiple cross sections throughout the caudate and putamen of GA1 patients will definitively show if medium spiny neurons are selectively lost as previously believed or if another neuronal or even a glial cell type may be affected, as suggested by this data. In conclusion,

DARPP-32 staining revealed that the number of medium spiny neurons does not differ between GA1 and control samples per area of the putamen.

33 Figures

Table 2: Clinical Data of Patients with Glutaric Acidemia Type I Case 1 2 3 4

Gender Male Male Female Female

Age at Neurological Crisis 6 Months 5.5 Months No acute 11 Months

crisis

Age at Death 7.5 Months 18.5 Months 5 Years 12 Years

GCDH Mutation Homozygous Homozygous Homozygous Homozygous

IVS1+5 G>T IVS1+5 G>T IVS1+5 G>T IVS1+5 G>T

Urine Glutaric Acid 388 58 500 33-73

(µmole/mmole creatinine,

Normal is <10)

Figure 8: DARPP-32 Stained Putamen in a 7-Month Old Control. A coronal view of the brain. A. Putamen B. Claustrum C. Insular cortex. The arrow between the putamen and claustrum denotes the external capsule white matter tract (along which images were taken for this study). The arrow between the claustrum and insular cortex denotes the extreme capsule white matter tract. Note the distinctive appearance of the putamen compared to the surrounding cortex.

Figure 9: 7-Month Old Control Putamen. A coronal view of the putamen. Darker myelin tracts can be seen throughout the putamen (circled). Also seen are DARPP-32 stained neurons (arrows).

Figure 10: 7-Month Old GA1 Putamen. A coronal view of the putamen. This shows the gross morphological changes that occur in the putamen of a GA1 patient after encephalopathic crisis. The region no longer looks striated, myelin tracts are no longer visible, and most landmarks are obliterated. Points of staining are both neurons and reactive astrocytes.

Figure 11: Higher Magnification of 7-Month Old GA1 Putamen. A coronal view of the putamen. Reactive astrocytes were stained by DARPP-32 in the 7-Month old patient sample. Seen above are DARPP-32+ neurons (circled) and stained reactive astrocytes (arrows).

38 3.0 2.3 2.1 2.5

2.0

1.5

1.0

0.5 Number of Positive NumberPositive of DARPP-32 Neurons 0.0 Age Matched GA1 Patients Controls

Figure 12: Mean Number of DARPP-32 Neurons per 383.5µm2 in Putamen of GA1 Patients vs. Controls. Error bars denote standard error of the mean.

P=. 5 4 4.5 3.6 * 4 3.3 3.5 2.5 2.8 3 2.5 2 0.6 1.5 0.7 1 0.3 0.5 0 Number of Positive NumberofPositive DARPP-32 Neurons 7 7 18 18 5 year 5 year 12 year12 year month month month month Control GA1 Control GA1 Control GA1 Control GA1

Figure 13: Mean Number of DARPP-32 Neurons per 383.5µm2 in Putamen Separated by Age. Error bars denote standard error of the mean.

40 CHAPTER III

DEEP SEQUENCING OF GCDH IN KNOWN GLUTARIC ACIDEMIA TYPE I PATIENTS

Introduction

Glutaric acidemia type I (GA1) is an autosomal recessive disease caused by mutations in glutaryl-CoA dehydrogenase (GCDH). Most patients are confirmed to have

GA1 when two pathogenic mutations are identified, and mutation analysis as done today is thought to detect causative mutations in 98-99% of GA1 patients 2,51. However, historically there have been patients with striatal necrosis and only one identified mutation in GCDH. In the years since these initial reports of affected GA1 patients with one causative mutation, sequencing technology has greatly advanced. One possible explanation of affected GA1 patients with one mutation is that older sequencing technologies were unable to identify the second causative mutation. The other possible explanation is that GA1 can be inherited as a recessive or dominant trait.

The Goodman laboratory at the University of Colorado Anschutz Medical

Campus has amassed a large collection of fibroblast samples from suspected GA1 patients. Many of these fibroblast samples have had both GCDH enzyme analysis and

GCDH sequencing performed. About one-third of this collection consists of cell lines with little or no enzyme activity that do not have two mutant alleles in GCDH. From this collection, I selected twenty GA1 lines to reexamine using a deep sequencing methodology to identify mutations. In addition to the exons that were previously sequenced, I inspected the R) and all introns.

I hypothesize that all cell promoter, lines will haveuntranslated two pathogenic region UT mutant alleles in GCDH and that deep sequencing will reveal mutations missed by the initial sequencing. I also

41 expect to find a few mutations in the could affect the regulation of GCDH. Selected cell lines were promoter all received or UTR before that 2005; the enzyme assay is still done using the same methodology, but the original DNA sequencing had been done by autoradiography on polyacrylamide gels or was manually read from paper printouts of fluorescence tracings.

Materials and Methods

Samples:

DNA from twenty primary fibroblasts cell lines of GA1 patients with clinical disease and GCDH deficiency, but lacking two GCDH mutant alleles on initial sequencing, were used for this study. Three of these cell lines had no mutations identified, and seventeen had one. Cell lines were stored in liquid nitrogen, thawed on dry ice, and grown in minimum essential medium. Cells were treated with trypsin, collected and centrifuged to form a cell pellet. Remaining media was decanted, and cell pellets were stored at -80°C until used for DNA isolation.

DNA from three primary fibroblasts from patients without GA1 were used to ensure that no mutations were added during the long-range PCR steps. Cell lines were stored in liquid nitrogen, thawed on dry ice, and grown in minimum essential medium.

Cells were treated with trypsin, collected and centrifuged to form a cell pellet.

Remaining media was decanted, and cell pellets were stored at -80°C until used for DNA isolation.

Isolation of DNA from Cell Pellets:

Cell pellets were resuspended in lysis solution (10 mM Tris, 20mM Na2EDTA,

10% SDS at pH 7.45) and vortexed. Proteinase K and RNase A solution were added to

42 the lysate and mixed by vortexing before placing samples on ice for one minute. Six molar ammonium acetate was added to lysates, vortexed to mix uniformly, and centrifuged to precipitate protein. DNA in the supernatant was then isolated by isopropanol precipitation, and ethanol washing (70%), dried and finally resuspended in water for long-range PCR use.

Initial Long-Range PCR of GCDH from Genomic DNA:

The primers and long-range PCR conditions used to amplify the entire genomic region around GCDH are listed in Table 3 and Table 4. Using four overlapping PCR fragments ranging from 2.5K to 3K in length, a total of 9.6KB of sequence was amplified.

This spanned 593bp upstream of the first exon and 357bp downstr

PCR products were purified using a Qiagen PCR purification kit (Qiagen,eam ofcatalog the numberUTR.

28106) as recommended by the manufacturer. Each PCR product was normalized to

1µg of DNA as measured by Nanodrop quantitation to ensure equal coverage across the genomic region. Next, the four PCR products from each cell line were combined (i.e. 23 unique cell line samples consisting of equal amounts of all four unique PCR products).

The DNA concentration of each sample was adjusted to 2.5ng/µL by fluorometry, as recommended by the Nextera protocol (Illumina, catalog number FC-121-1031).

Generation of DNA Library for MiSeq:

Library generation for the MiSeq was done using the Nextera DNA Library

Preparation Kit using the conditions specified by the manufacturer; this protocol takes samples through a series of steps to generate a library suitable for NextGen sequencing.

First, the DNA was both enzymatically fragmented into 200bp to 300bp pieces, and an adapter with an amplification sequence was added. The DNA was washed to separate

43 the fragments from the enzyme solution., and clean DNA was amplified using the adapter sequence as a primer. During this amplification process, a unique ID sequence was added to each patient sample. The amplified sequences were cleaned to remove excess PCR reagents, and the quality of the library was assessed for DNA concentration and fragment size. At this point, one GA1 sample was deemed as not suitable for use in the combined library due to sample loss and was removed from the study. Finally, the remaining samples (19 GA1 cell lines and 3 control cell lines) were normalized to 2nM of DNA and combined into a single library sample. 10pM of the library was run on a single channel in the MiSeq run.

Data Analysis:

First, a computer script separated reads by cell line using the unique ID tags.

Trimming scripts were then used to remove the ID tags and adapters from the fragments as well as to remove any bases with poor read quality. The remaining reads were aligned to the genome using gsnap 84. Only reads aligning to the originally amplified region of chromosome 19 were kept. Variants in this region were identified by GATK 85 and annotated by Annovar 86 and dbNSFP 87. The variants were analyzed by

SIFT 88, polyphen2 87, LRT 89, MutationTaster 90, and MutationAssessor 87 to predict the effects. The list of mutations for each cell line was analyzed by hand to identify likely causative mutations. Mutations were deemed likely causative if they had been previously reported in a GA1 patient, or if they were predicted to be damaging by the above prediction algorithms, and had a population frequency less than 5% in the dbSNP, 1000 genomes, or ExAC databases.

44 GCDH Enzyme Assay:

Fibroblasts were homogenized in 0.02M phosphate buffer (pH 7.4) containing

0.208mg/mL FAD and sonicated for 30 cycles at 20% duty. 400µL of the homogenate was incubated in a 1mL reaction with 10µL of [1,5-14C] -glutaryl-CoA in 0.02M potassium phosphate (pH 7.4) containing 0.02M 1,4-dithiothreitol (DTT) and 3.06mg phenazine methosulfate (PMS) for 10 minutes at 37°C. Reactions were stopped by adding 0.5mL of 25% trichloroacetic acid and left at 37°C for 40 minutes to evolve and trap CO2. 14CO2 was trapped in 10% potassium hydroxide (KOH) and counted in a

1900CA Packard liquid scintillation analyzer. Protein concentration in homogenates was determined by Bradford assay and was used to calculate a specific activity of GCDH for each cell line.

Results

Twenty-two cell lines were screened for mutations by deep sequencing; of these, nineteen were suspected GA1 cell lines, and three were from controls. Two pathogenic mutant alleles were identified in seventeen GA1 cell lines, and one mutant allele was identified in each of the remaining two (See Table 5). For both cell lines with only one mutation identified by deep sequencing, the clinical data was reexamined and the cells retested for GCDH enzyme activity. When reexamined, one cell line (#3) had clinical data and an enzyme activity consistent with its being from a GA1 carrier. The other cell line (#5) is clearly from a GA1 patient, both by the clinical description and because it has no GCDH activity.

Six of the mutations identified in this study were novel, i.e., IVS2+1G>A,

IVS2-1G>C, R88S (C>T and C>A), V133L, and R234Q (See Table 5). Previously

45 described, non-pathogenic SNPs were identified in the introns, 5

UTR of all cell lines examined. Additionally, three common exonic promoterpolymorphisms and the were identified in this cohort. Eleven cell lines had the G391G (c.G1173T; rs1060218) variant, which has a gl

UTR A1250G (c.A1250G;obal rs8012) minor variant,allele frequency which has of a%. global Eighteen minor allelecell lines frequency had the of

28%. Finally, one GA1 cell line had the A337A variant (c.A1011G; rs2229460 with a global minor allele frequency of 0.8%) in addition to the two mutant alleles.

Seventeen of the nineteen cell lines had a single previously reported mutation from Sanger sequencing data. Fifteen of these seventeen mutations were confirmed by deep sequencing, and two were found to be erroneous (See Table 6).

Discussion

In the original analysis of these cell lines, exonic regions of GCDH were amplified by PCR and then sequenced by dideoxy Sanger and run on polyacrylamide gels. This method successfully identified two mutant alleles in 109 of 157 suspected GA1 cell lines

(69%). Deep sequencing showed that seventeen of the nineteen GA1 cell lines indeed had two pathogenic mutant alleles; the remaining two cell lines had only one. The first of the latter two, i.e., cell line #3, had the known Oji-Cree First Nations mutation

(IVS1+5G>T) 52,53, and the little clinical data that was provided with the cell line suggested that the patient may have been an unaffected sibling of a child diagnosed with GA1 (slightly increased urine metabolites and no reported striatal necrosis). The other cell line, #5, comes from a patient with striatal necrosis, increased glutaric and

3-hydroxyglutaric acids in urine, and complete GCDH deficiency in cultured fibroblasts, but has only one pathogenic mutation (E365K).

46 All identified pathogenic mutations were either missense or splice site mutations, and none were found in the All cell lines sampled had previously reported common variants in these promoter regions, or but UTR. there were no pathogenic mutations. This finding suggests that, with current sequencing methods, all of these cell lines would now have two mutations identified in GCDH. This study identified six novel mutations in GCDH: IVS2+1G>A, IVS2-1G>C, R88S (C>T and C>A), V133L, and R234Q.

Seventeen of the cell lines had a previously identified mutation determined by dideoxy

Sanger sequencing. Fifteen of the seventeen mutations were confirmed by deep sequencing (88%).

By reanalyzing these samples using newer sequencing methods, I have shown that NextGen methodology is more sensitive and accurate than methods used in the past. With the lack of any deleterious mutations identified in the nonexonic regions of

GCDH, there is no reason to think that the current clinical workup is missing a large group of patients. As shown in this study, deep sequencing has a high degree of accuracy that was missing when these particular samples were initially sequenced; perhaps explaining why 33% of the patients screened did not have two mutations identified. However, current Sanger sequencing methods and the use of computers to align reads and identify mutations, as opposed to reading printouts by hand, is predicted to identify mutations in 98%-99% of GA1 patients 2. Because of this, current diagnostic sequencing focusing on the exonic regions combined with a deletion analysis should suffice to detect mutations in true GA1 patients. However, in cases where GA1 is highly suspected but two pathogenic mutant alleles in GCDH are not identified, deep sequencing is a robust method that can be used to screen for missed mutations.

47 Two of the cell lines in this study did not have two GCDH mutations identified by deep sequencing. One of these proved to be from a GA1 carrier and as such should only have one mutation. The other clearly comes from a GA1 patient but has only one identified mutation. One possible explanation is that this study was not sufficient to detect insertions or deletions in GCDH. When all four long range PCR products were combined by concentration, the amount of DNA for each product was not normalized.

As a result, all samples showed artificial copy number variation in GCDH due to the different amounts of each PCR product. For this cell line, one allele has the E365K mutation, and the other allele may have an insertion/deletion event in GCDH. A second possibility is that a second mutation exists outside of the sequenced region. The human

GCDH promoter region has not been fully characterized, and this study may not have included the entire region. Overall, this study shows that GA1 is a recessive disease. In historical cases of GA1 where patients with striatal necrosis and only one mutation identified by sequencing there were, in fact, two mutations in GCDH and one was simply missed by previous sequencing methods. Deep sequencing is a robust method capable of screening the entire GCDH genomic region in multiple patients with a single sequencing run. Deep sequencing can distinguish carriers from true GA1 patients, and by normalizing the total DNA of each long-range PCR product, deep sequencing would allow for the identification of SNPs and insertions/deletions with a single test.

48 Figures

Table 3: Long-Range PCR Primer Sets

1 2 3 4

Forward CTGCCTCCTTGTG CTTGGAGCTTCGGAGT GTCACAAGTATCTACATG TGGGACCAAGACCTG TGTCCTT TCTG TGCTTCCTGGGACAGACT GTAAG GGCAGAAAGGTTTGC Reverse CTTGTCCTGGTCC GCACTCACCATCTTGT GACTGGGGAACACAGAAA ACCGAGCCCACACTA TTCAAGC TTGTGATCCCCTATAA GC CAAAC CATGGTGTGTCTGAGC TGAG PCR Program Touchdown Program Standard Program Touchdown Program Standard Program

49 Table 4: Long-Range PCR Programs

PCR Conditions for Standard Program PCR Conditions for Touchdown Program Temperature Cycles Temperature Cycles Initial Initial Denaturation 95°C 1 Denaturation 95°C 1 Denature 95°C Denature 95°C 65°C Annealing 65°C 30 Annealing -0.5°C /cycle 10 Extension 72°C Extension 72°C Final Extension 72°C 1 Denature 95°C Hold 4°C Annealing 60°C 20 Extension 72°C ∞ Final Extension 72°C 1 Hold 4°C

∞

50 Table 5: Mutations Identified by Deep Sequencing

Cell Allele 1 Allele 2 Line

1 R161Q R402W 2 IVS1+5 G>T IVS1+5 G>T 3 IVS1+5 G>T 4 IVS4+5 G>A IVS4+5 G>A 5 E365K 6 R402W V400M 7 E414K A298V 8 R383H R355C 9 R402W R234Q 10 A293T T416I 11 A421V R88S 12 M266V IVS10-2 A>C 13 A122V R313W 14 R121Q R402W 15 R402W R88S 16 IVS2+1 G>A IVS2+1 G>A 17 V133L IVS2-1 G>C 18 R227P IVS2+1 G>A 19 G185R M191T Cell lines number 3 and 5 only had one mutation identified (highlighted in red). Cell line number 3 comes from a GA1 carrier, and cell line number 5 comes from a GA1 patient. Bolded items highlighted in blue are mutations that have not previously been described in the literature.

51 Table 6: Confirming Previously Identified Mutations

Cell Previously Mutation Line Reported Identified by Mutation Deep Sequencing 1 IVS1+5 G>T R161Q 2 IVS1+5 G>T IVS1+5 G>T 3 IVS1+5 G>T IVS1+5 G>T 4 none IVS4+5 G>A 5 E365K E365K 6 R402W R402W 7 E414K E414K 8 R383H R383H 9 R402W R402W 10 A293T A293T 11 A421V A421V 12 M266V M266V 13 A122V A122V 14 R121Q R121Q 15 R402W R402W 16 none IVS2+1 G>A 17 T214M V133L 18 R227P R227P 19 G185R G185R Bolded items highlighted in red denote mutations that were corrected by deep sequencing.

52 CHAPTER IV

MUTATIONS IN DHTKD1 CAUSE 2-KETOADIPIC ACIDEMIA AND 2-AMINOADIPIC

ACIDEMIA

Introduction

At present, patients suspected of having glutaric acidemia type 1 (GA1) are treated in an attempt to prevent striatal necrosis during a metabolic crisis. Treatment consists of a diet low in lysine (or protein), carnitine supplementation, and emergency treatment during catabolism 2,71. Such treatment reduces the frequency of encephalopathic crises from 80-90 percent, without treatment, to 10-20 percent2.

Treatment thus prevents striatal necrosis in most patients, but not all. Nor is it clear how long patients need to remain on treatment.

Striatal necrosis usually occurs within the first 3 years of life 2; however, there are reports of late onset GA1 patients diagnosed anywhere between 8 years and 71 years of life 8. In the few case reports describing late onset GA1, patients present with symptoms not typically seen in childhood GA1; these include vertigo, headaches, dementia, tremors, epilepsy, and problems with coordination8. All late onset patients are reported to have white matter changes, predominantly periventricular and in the frontal and parietal lobes8.

It is widely thought that these neurological abnormalities result from prolonged exposure to a toxin, and a treatment to prevent accumulation of the toxin, whatever its nature, might be of benefit. Blocking lysine oxidation upstream of GCDH might prevent the accumulation of such a toxin, and prevent neurological damage. To test this

53 hypothesis, however, one needs an upstream enzyme whose deficiency does not cause disease.

Another Disorder of Lysine Metabolism: 2-Ketoadipic and 2-Aminoadipic Acidemia

One such enzyme might be 2-ketoadipic acid dehydrogenase complex, which when deficient causes 2-ketoadipic acidemia (KAA), and which is immediately upstream of GCDH (See Figure 3 or Figure 4). KAA is characterized by elevations in

2-ketoadipic (2-KA), 2-aminoadipic (2-AA), and 2-hydroxyadipic (2-HA) acids in blood and urine. Pedigree analysis shows that KAA is inherited in a recessive manner 91; and patients with this biochemical profile have been described with a wide range of clinical presentations, from early-onset developmental delay, epilepsy, mental retardation, and immunodeficiency, to being completely normal (See Appendix B) 91 102. – The first report of a proband with elevated 2-KA and 2-AA was published in

1974 and describes two brothers with elevated metabolite levels, one with a learning defect and the other being mentally normal 101. A second report that same year describes a child with elevated metabolite levels who was identified when screening mentally retarded children for inborn errors of amino acid metabolism92. There is controversy over whether KAA represents a true disease state or the biochemical abnormalities found in these patients merely reflects an ascertainment bias. This is due to the broad range of phenotypes described, the identification of multiple normal siblings with elevated metabolite levels, and the biased manner in which many patients were identified. In fact, in the 2014 copy of a Physician’s Guide to the Diagnosis,

Treatment, and Follow-Up of Inherited Metabolic Diseases lists KAA as a biochemical phenotype of questionable clinical significance 103.

54 As of the writing of this thesis, there have been 25 described patients with

2-ketoadipic and 2-aminoadipic acidemia (KAA) (See Appendix B). Presuming that this is a unique biochemical marker but not a disease, KAA and the enzyme whose deficiency causes it was of interest for my work. I hypothesize that a partial or complete block of the lysine oxidation pathway at this point could be used as a GA1 treatment. To this end, I performed whole exome sequencing on DNA from a family known to have

KAA without a clinical phenotype.

Abstract1

2-Ketoadipic aciduria (OMIM 204750), a defect in the catabolic pathway of tryptophan, lysine, and hydroxylysine, is characterized by elevations in 2-ketoadipic,

2-aminoadipic, and 2-hydroxyadipic acids. Patients with the aforementioned biochemical profile have been described with a wide range of clinical presentations, from early-onset developmental delay, epilepsy, ataxia, and microcephaly to completely normal. This broad range of phenotypes has led some to question whether 2-ketoadipic aciduria represents a true disease state or if the biochemical abnormalities found in these patients merely reflect an ascertainment bias. We present four additional individuals from two families, with 2-ketoadipic aciduria with compound heterozygous or homozygous mutations in DHTKD1, three of which remain asymptomatic.

Keywords: 2-Aminoadipic, 2-Ketoadipic aciduria, 2-Oxoadipic aciduria, DHTKD1,

Hydroxylysine, Lysine, Organic acidemia, Tryptophan

1 New Cases of DHTKD1 Mutations in Patients with 2-Ketoadipic Aciduria was published in JIMD Reports in 201599

Introduction

Organic acid analysis has identified elevated 2-ketoadipic acid in patients with a wide variety of symptoms ranging from psychomotor retardation, hypotonia, epilepsy, ataxia, and failure to thrive to no clinical phenotype at all. To date over 20 individuals have been reported, about half of whom were asymptomatic 93,98,102,104. With no known genetic etiology, 2-ketoadipic aciduria was thought to represent ascertainment bias with clinical symptoms being coincidental findings 98,104,105.

In 2012, whole-exome sequencing (WES) of a patient with a biochemical diagnosis of 2-ketoadipic aciduria identified compound heterozygous variants in dehydrogenase E1 and transketolase domain-containing protein 1 (DHTKD1). Sanger sequencing of DHTKD1 in a second unrelated patient identified a missense mutation on one allele and a nonsense variant on the other 98. These patients presented with hypotonia and variable degrees of psychomotor delay, speech delay, and attention deficit hyperactivity disorder with an otherwise unremarkable neurological examination. Due to the phenotypic variability associated with 2-ketoadipic aciduria, functional studies in primary fibroblasts from these patients were employed in order to elucidate the genetic etiology which revealed increased levels of 2-ketoadipic acid in cells and medium that was corrected with expression of wild-type DHTKD1 98.

We report mutations in two additional families and highlight the fact that while genetic abrogation of DHTKD1 can lead to the accumulation of 2-ketoadipic,

2-aminoadipic, and 2-hydroxyadipic acids, this disruption does not always result in an observed clinical phenotype.

Materials and Methods

Amino acid analysis in urine and plasma and urine organic acid analysis was performed in established biochemical genetics laboratories using proprietary methods.

Exome sequencing in both patients was performed prior to the discovery that mutations in DHTKD1 cause 2-ketoadipic aciduria. WES for patient 1 was performed as a trio with both parents at UCLA Molecular Diagnostics Laboratories, Los Angeles, CA.

WES for the second family was performed on the proband and both parents by BGI

Americas and analyzed at the University of Colorado Denver. Identified DHTKD1 mutations were evaluated for possible effects on protein structure and function by

MutationTaster 90, SIFT 88, PROVEAN prediction 106, MutationAssessor 87, and

PolyPhen-2 89. Variant frequencies were derived from the 1,000 genomes database.

Neuropsychological evaluations were performed at C

Psychology using NEPSY-II, Preschool Language Scale, 4th Edition;(OC Childrens Adaptive Division Behavior of

Assessment System, 2nd Ed.; Behavior Assessment System for Children, 2nd Ed.;

Behavior Rating Inventory of Executive Function, Preschool Version; Child

Development Inventory, Sensory Profile, Wechsler Preschool and Primary Scale of

Intelligence, 3rd Ed.; and Woodcock Johnson Tests of Achievement, 3rd Edition.

Case Reports

Our first patient, a girl, was born at full term after a pregnancy that was complicated by maternal hypertension with no postnatal complications to non-consanguineous parents of Filipino and Northern European ancestry. At 15 months of age, she presented with a history of failure to thrive (weight <3rd percentile), seizure-like episodes, and biochemical abnormalities consistent with 2-ketoadipic

aciduria. Her height remained between the 10th and 25th percentile (length/weight ratio <3rd percentile) and head circumference was below the 3rd percentile. Of note, maternal head circumference was also below the 3rd percentile. A metabolic assessment revealed elevated levels of plasma 2- elevated 2-ketoadipate aminoadipate2-hydroxyadipate 7 μmol/L, nl < (28 with

mmol/mol (See creatinine, Table 7). nl < Over time,and she has suffered chronic episodesmmol/mol of creatinine, headaches/migraines nl < in urine with persistent head tilting, nausea, and emesis.

Neurological evaluation was performed and electroencephalogram, magnetic resonance imaging, and magnetic resonance angiogram of the brain were unremarkable. At 24 months of age, she began to show mild developmental delay. Initial neuropsychological evaluations measured less than the 10th percentile in verbal fluency and processing speed. Over the course of a 4-year follow-up, trends of improvement were noted in these areas; however, following the most recent evaluation, she is now below the 1st percentile for auditory attention and was given a diagnosis of a reading disorder.

WES revealed two heterozygous variants, c.2143C>T; p.Arg715Cys and c.2185G>A; p.Gly729Arg, in DHTKD1. While the latter variant, c.2185G>A; p.Gly729Arg, was previously reported in two patients with autosomal recessive 2-ketoadipic aciduria

98, the former variant, c.2143C>T; p.Arg715Cys, is novel. Both variants have been observed in the general population with minor allele frequencies of 0.05% and 0.18%, respectively.

Our second patient, a girl, was born to consanguineous parents of Maltese origin in 1978. Routine urine screening performed at 6 weeks of age detected increased glutamate. Subsequent amino acid and organic acid testing showed elevated urinary

2-aminoadipic, 2-ketoadipic, and glutaric acids in the proband as well as in two elder brothers, ages 5 and 7. All three affected siblings were described as clinically normal when identified and have been followed over time and remain asymptomatic in adulthood 107(Wilcken 2014, personal communication to Steve Goodman).

a novel homozygous variant in

DHTKD1Data, c.915G>C; analysis p of the probands exome identified are also homozygous .Gln(is.for this variant Sanger and sequencingthat each parent showed is a that carrier. the probandsThis variant siblings has been observed in the general population with a minor allele frequency of 0.05%.

No other candidate genes were found during the WES analysis for either patient sequenced. All variants identified are predicted by several algorithms to be disease causing (MutationTaster), damaging to protein function (SIFT), deleterious to protein structure (PROVEAN), and probably damaging to protein structure and function

(PolyPhen-2) and are predicted to have a high functional impact for c.2143C>T; p.Arg715Cys and c.915G>C; p.Gln305His and a medium functional impact for c.2185G>A; p.Gly729Arg (MutationAssessor).

Discussion and Conclusions

2-Ketoadipic acid is formed through three routes: from 2-aminomuconate in the oxidation of tryptophan, from 2-aminoadipic acid formed in the mitochondrial oxidation of lysine via the saccharopine pathway, and from 2-aminoadipic acid formed in the brain-specific, peroxisomal oxidation of lysine via the pipecolic acid pathway 108.

The subsequent conversion of 2-ketoadipic acid to glutaryl-CoA, which is common to all three pathways and has long been assumed to involve a multienzyme complex similar to those that act on pyruvate, branched-chain keto acids and 2-ketoglutarate; indeed,

the activities of the 2-ketoglutarate and 2-oxoadipic dehydrogenases in porcine heart could not be separated 109.

Several subjects have been described with 2-ketoadipic aciduria, including a

14-month-old girl with hypotonia, intermittent metabolic acidosis, and developmental delay 93, a 14-year-old retarded boy and his intellectually normal sister 91, a 9-year-old boy with a mild learning disability and his normal brother 101,102, a 10-year-old retarded girl 95, a 9-year-old retarded boy with a history of seizures 104, a 7-year-old girl with cerebellar ataxia 110, two unrelated children with developmental delay 98, and three apparently normal siblings detected by a newborn screening program in Australia

107(Wilcken 2014, personal communication to Steve Goodman), whose genetic findings and outcome are presented here. Evidence for defects in the metabolism of 2-ketoadipic acid in these patients have included increased excretion of 2-ketoadipic and/or

2-hydroxyadipic acid in urine, often together with 2-aminoadipic acid, and delayed clearance of 2-aminoadipic and 2-hydroxyadipic after an oral lysine load. These studies are not strictly comparable because of the variability in biochemical testing among them. Recent studies on two unrelated patients with 2-ketoadipic aciduria revealed mutations in DHTKD1, a nuclear gene that encodes a protein similar to the E1 component of a 2-ketoglutarate dehydrogenase complex 98. Our data extends these observations, showing mutations in the same gene in several additional patients with

2-ketoadipic aciduria. One of these patients has symptoms, but the other three are asymptomatic adult siblings.

The finding that all three patients with the same mutation, i.e., c.2158G>A; p.Gly729Arg 98,99, are symptomatic may suggest a relation to the clinical phenotype. The

population frequency of this variant may approach one in 650 (188 of 120,740 chromosomes in the Exome Aggregation Consortium (ExAC) cohort as of April 2015).

While this frequency seems high, it could be consistent as an underlying cause of mild developmental delay as observed in these patients. The other known sequence variants appear to be much less frequent; c.915G>C; p.Gln305His was present only once out of

121,316 chromosomes examined, and c.2143C>T; p.Arg715Cys was present in two of

121,298.

While 2-ketoadipic aciduria can occur without apparent clinical manifestations in childhood, and the relation to clinical manifestations may be due only to sampling bias, it is difficult to exclude the possibility that asymptomatic individuals will develop a clinical phenotype later in life or that particular mutations cause clinical disease and others do not. Indeed, knockdown of DHTKD1 expression in a variety of cell lines suggests that this protein plays a role in mitochondrial function and energy production

111; and a proteomic/metabolomic study in mice implicated DHTKD1 in glucose homeostasis through its connection to 2-aminoadipic acid 112. These studies suggest the possibility that further phenotypes, perhaps with a later onset, may be associated with genetic abrogation of DHTKD1. Of particular note is the implication of a c.1455T>G; p.Tyr485* mutation in DHTKD1 as the cause of dominantly inherited

Charcot-Marie-Tooth disease in one family in China 113. The c.1228C>T; p.Arg410* mutation found in a 2-ketoadipic aciduria patient 98 might affect the E1 protein in a similar manner as p.Tyr485*, and it will be of interest to determine if a similar neurological phenotype emerges in this patient or in the parent with the same mutation.

In summary, we present four additional patients with 2-ketoadipic aciduria with variants in DHTKD1. Three of these individuals have remained phenotypically normal in to adulthood, while the other shows clinical characteristics similar to the previously reported patients with 2-ketoadipic aciduria 98. These findings support the genetic etiology of 2-ketoadipic aciduria and continue to highlight the phenotypic variability historically seen in the reported patients.

Figure

Table 7: Summary of the Clinical Findings of Patients with DHTKD1 Deficiency

CHAPTER V

CHARACTERIZATION OF A GCDH DEFICIENT ZEBRAFISH

Animal Models of Glutaric Acidemia Type I

There are four different animal models of glutaric acidemia type I (GA1) described in the literature; one is a naturally occurring bat, and the others are rodents generated in the laboratory. What is known of the bat model allows for some meaningful conclusions of GA1; however, little work has been done with it. The

Egyptian fruit-eating bat R. Aegypticus naturally lacks glutaryl-CoA dehydrogenase

(GCDH) activity in all but the central nervous system. These animals have elevated levels of glutaric (GA) and 3-hydroxyglutaric acids (3-OH) in urine comparable to those seen in human GA1, but have no neurological abnormalities. To date brain levels of GA and 3-OH have not been measured in the bats but the presence of GCDH activity in the brain tissue suggests that levels of these metabolites in the brain would not be elevated.

These animals provide highly suggestive evidence that the neurological effects of GA1 in humans are caused by the accumulation of GA and 3-OH in the brain due to loss of

GCDH function in the brain. 114,115

The various rodent models of GA1 are much easier to work with and care for; additionally, knowledge of the mouse genome aids in the design of experiments and interpretation of results 62. The mouse GCDH gene is very similar to its human ortholog

(79%); in both species, GCDH has 11 exons and 10 introns and, while the intron-exon junctions are identical, the intron lengths are quite different54 (See Figure 14). Studies of the mouse gene were used to provide estimates of the location, structure, and regulation of the human GCDH promoter 67. This work showed that GCDH is a

housekeeping gene and regulation does not respond to increased levels of lysine or GA

67. The GCDH enzyme is 90% similar between mice and humans with complete conservation around the active site of the enzyme. Because of this, rodent models have been used to determine the functional consequence of elevated GA and 3-OH metabolite levels in the presence of functional and deficient GCDH, as well as the effect of GCDH deficiency to various environmental insults. There are three distinctive rodent models of GA1 that have all been used to address various nuances of GA1.

Referred to as the chemical model of GA1, rats have been injected with GA and

3-OH, at various concentrations, to look for behavioral and neurotoxic effects. Several studies have shown that rats injected with GA into the cisterna magna experience a

GA1-like encephalopathic crisis 46,116,117. One such study observed that an acute dose of

GA results in leakage of the blood-brain barrier, in accordance with observations of the

GCDH -/- mouse 118. Another study on rat brain cell cultures observed astrocyte activation by glutaric acid which caused them to release an as yet unidentified neurotoxin 46. Another study demonstrated that exposure to 3-OH causes convulsions, striatal lesions and an imbalance of glutamatergic and GABAergic neurotransmission in rats 119. These studies can examine the effects of acute exposures to these metabolites, in a dose dependent manner, on behavior and general health.

A GCDH -/- mouse model was generated in 2002 by replacing the first seven exons of GCDH -galactosidase reporter regulated by the native promoter 37.

Presence of thewith mutant a β allele did not affect viability, birth weight, neonatal growth or final adult weight. These mice accumulated GA and 3-OH throughout their bodies and excreted the metabolites into their urine at levels similar to those seen in GA1 patients

37. These animals did not spontaneously develop neurological symptoms or display any significant motor deficits. When fed a high lysine diet, GCDH -/- animals suffered hypothermia, dehydration, and hypoactivity, followed by paralysis, seizures and finally death 49. MRI brain scans of animals fed a high lysine diet showed loss of lateral ventricles, bilateral striatal injury, and (often) acute hemorrhage just before death 49.

These animals also displayed a vulnerability period, with younger GA1 animals being more sensitive to a high lysine load or striatal toxins such as 3-nitropropionic acid than wild type or carrier siblings of the same age 120,121. Several studies have shown that cerebral levels of GA and 3-OH positively correlate with increasing amounts of dietary lysine 48,50. The GCDH -/- mouse model has been used in several experiments to understand the pathogenesis of GA1 further, but falls short of being an ideal model because it does not replicate the typical encephalopathic crises seen in human patients.

The final mouse model of GA1 was generated by transplanting the hepatocytes from the GCDH -/- mice into GCDH +/+ animals, producing animals with reduced hepatic GCDH activity but with normal GCDH activity in the brain. This model has been used to show that elevated peripheral levels of GA and 3-OH do not cross the blood-brain barrier122. These results further support the notion that elevated metabolite levels seen in GA1 brains are the result of deficient GCDH activity in the brain and de novo synthesis of GA and 3-OH. These metabolites then accumulate in the brain due to limited flux across the blood brain barrier.

Zebrafish Animal Models

A newer animal model to study inborn errors of metabolism is the zebrafish.

Zebrafish models have several advantages over the mammalian model organisms,

including a rapid development process external to the mother, large clutches of offspring (a single mating can produce 300 embryos in a single clutch), embryos that are optically transparent during development, and substantially lower maintenance costs. Embryonic development during the first 48 hours of life relies on mRNA packaged by the mother which allows for the detection of mutants that would be embryonically lethal in mice. Many disease phenotypes become manifest during embryonic development, i.e., within eight days post-fertilization, allowing for rapid identification and data collection. Zebrafish embryos can be used for high-throughput screening of drug libraries using endpoints such as fluorescent reporters or locomotor behavior changes, and have been suggested as a way to test cell-based drug discoveries in a full organism before using an expensive mammalian model. 123,124

More importantly, though, is that many zebrafish genes have significant sequence homology with their human orthologs. The zebrafish genome is available

(albeit not fully annotated), and zebrafish share many metabolic pathways with higher organisms. It is important to note that a large duplication event in teleost evolution has caused many genes to have two paralogs in the zebrafish 125. In 2011 the Wellcome

Trust Sanger Institute launched the Zebrafish Mutation Project 75, a large-scale effort to generate knockout alleles in every protein-coding gene in the zebrafish genome, and has made all identified mutants available for use. Additionally, the zebrafish genome can readily be manipulated through forward genetics mutagenesis screens, or with the use of site directed manipulation using zinc finger nucleases (ZFNs), transcription-activator-like endonucleases (TALENs) or clustered regularly interspaced short palindromic repeats (CRISPR) systems. A temporary manipulation of the

zebrafish genome can be achieved using morpholinos to knockdown gene expression, or genes can be introduced by injections of recombinant plasmids. 123,124

For my thesis work, I endeavored to generate a GCDH deficient zebrafish and to characterize the effects of that deficiency. My objective was to show that zebrafish could be used as a model for GA1. A zebrafish model would be beneficial due to the decreased animal care cost, a large number of progeny per breeding, and the ease of cross-breeding with other acyl-CoA dehydrogenase mutants. A zebrafish model would allow for the testing of many GA1 hypotheses that are impractical to test in the existing models of GA1. Several zebrafish models of inborn errors have been created and show a remarkable similarity to human disease 71,123,126 135. Most relevant to this study are the – zebrafish models of maple syrup urine disease 136 and multiple acyl-CoA dehydrogenase deficiency (MADD/GA2) 27. In both models, researchers can characterize movement defects, developmental defects, and elevated metabolite levels. The metabolic profile for each resembles that of the human disease and can be demonstrated with standard clinical testing.

A zebrafish model of GA1 does have some unique challenges due to the partial genome duplication in the teleost lineage. There are GCDH A and GCDH B genes in the zebrafish; the two can be distinguished from one another because the GCDH B lacks part of the N-terminus and as a result, does not encode a mitochondrial leader sequence (See

Figure 14). This suggests that the GCDH B cannot be targeted to the mitochondria and thus does not act in lysine oxidation. However, the remainder of the coding sequence is highly conserved between the A and B genes, suggesting that the protein generated by

GCDH B may be functional. Despite this possible pitfall, the zebrafish would still be a

good model of GA1 because they have orthologs of all enzymes in the human lysine oxidation pathway, have a blood-brain barrier 137, and possess a distinct brain region homologous to the striatum 138. In the following study, I characterize a zebrafish line containing a splice site mutation in intron six of GCDH A and evaluated its use as a model of human GA1. A mutation affecting the donor splice site of intron six has not been described in man, but the same mutation (IVS +2T>C) in IVS4 in man causes a truncated protein, elevated urine GA, and clinical symptoms if untreated 25.

Materials and Methods

Ethics Statement:

The animal work in this study was approved by the Institutional Animal Care and Use Committees of the University of Colorado School of Medicine.

Zebrafish Lines and Husbandry:

The GCDH IVS6+2 T>C mutation, discovered in an ENU mutagenesis screen by the Zebrafish Mutation Project 75 (allele name sa12642), was ordered from the

Zebrafish International Resource Center (ZIRC, U. Oregon, Eugene OR). Adult animals (3 months or older) of both sexes were used for organic acid screening, glutaric acid and

3-hydroxyglutaric acid quantitation, quantitative amino acid analysis, and glutaryl-CoA dehydrogenase enzyme assay. Embryos between the ages of 1-days post fertilization and 8-days post fertilization were used for the C-start response testing and developmental assessment. Tanks had between 1 and 50 zebrafish of mixed sex and were separated by genotype and generation. Fish were fed twice a day, and all fish were cold-euthanized. The fish were frozen at -80°C prior to testing. When ready to test, fish were thawed to room temperature and homogenized as described below.

C-Start Response Testing and Developmental Assessment:

When an abrupt sensory stimulus (i.e. a pipette tip tapping the head or tail of an embryo) occurs, a normal fish will respond by bending into a C shape and moving away from the stimulus quickly 139 141. This test evaluates both the motor function and – predator avoidance behavior of the zebrafish and can detect early phenotypes. This was performed daily on all embryos from 4 days post fertilization to 8 days post fertilization

(about 300 embryos). Zebrafish development was assessed using reference images from the ZFIN zebrafish developmental staging series online resource and with help from the Appel Lab at the University of Colorado Anschutz Medical Campus.

Fish Homogenization for Metabolic Testing:

Whole fish homogenates were used for all tests. Fish were thawed, patted dry, weighed, homogenized (ground glass homogenizer) in 10µL water/mg wet weight (or in phosphate buffer for fish to be analyzed for GCDH enzyme assay) and sonicated for

30 cycles at 20% duty. Homogenates in water were mixed with 70mg/mL sulfosalicylic acid, allowed to sit at room temperature for 5 minutes, and centrifuged at 1780 x g for 5 minutes to pellet precipitated protein. The supernatants were then used for metabolic testing. Each homogenate was comprised of a single fish, and the concentration of all homogenates was normalized to fish weight.

Quantitative Amino Acid Analysis:

Quantitation of free amino acids was performed on whole fish of all genotypes.

The amount of each amino acid is not expected to deferrer between genotypes. Fifty µL of each fish homogenate, 150µL of Biochrom lithium citrate loading buffer (pH 2.2) and

50µL of 0.5mM D-glucosaminic acid (internal standard) were combined, mixed by

vortex, and centrifuged at 16,000 x g for 1 minute in a microcentrifuge tube with a

0.2µm filter. The filtrates were then loaded into sample vials for injection into a

Biochrom 30+ amino acid analyzer. The automated Biochrom system separates amino acids by ion exchange chromatography and reacts the column eluate with ninhydrin at high temperature to form dye complexes that are detected by light absorption at 570 nm and 440 nm. Peaks were identified by elution time and quantified by comparing their areas to that of an internal standard. A one-way ANOVA was performed for each amino acid to determine if the amount of an amino acid differed between genotypes.

Organic Acid Screen of Whole Zebrafish Homogenate:

An organic acid screen was performed on fish of all genotypes. The profile of wild type and heterozygous zebrafish was expected to be the same, and the profile of homozygous zebrafish was expected to show increased glutaric acid and

3-hydroxyglutaric acid. One mL of each deproteinized fish homogenate in water was placed in a round bottom glass test tube with 12.5 µL of 0.5mg/mL dimethylmalonic acid (internal standard) and extracted twice with 4.5 mL diethyl ether:ethyl acetate

(2:1). The combined extracts were dried by addition of 0.5 g anhydrous Na2SO4, followed by vortexing for 1 minute, and centrifugation at 725 x g for 3 minutes. After addition of 12.5µL of 0.25mg/mL tetracosane (C-24) as an external standard, the solvent was evaporated under a stream of nitrogen. Twenty µL of 99:1

N,O-Bis(trimethylsilyl)trifluoroacetamide/trimethylsilylchloride (BSTFA/TMCS) was added, and samples were placed in an 80°C heating block for 20 minutes to form trimethylsilyl derivatives. After cooling at room temperature for about 5 minutes, 4µL of each derivatized mixture were added to 50µL of 98:1:1

cyclohexane:pyridine:hexamethyldisilazane dilution solution in a glass autosampler

GC/MS vial. Organic acids were separated from one another and identified by gas chromatography-mass spectrometry using an Agilent 6890N instrument. This is a

of Colorado biochemical genetics lab to analyzestandard non-amino protocol in organic the Childrens acids in (ospitalphysiological fluids. Scans were examined by board certified biochemical geneticists. The presence or absence of glutaric acid was noted for each sample. This is the same practice as used on human clinical samples.

Glutaric Acid and 3-Hydroxyglutaric Acid Quantitation by Stable Isotope

Dilution:

Quantitation of glutaric acid and 3-hydroxyglutaric acid was performed on fish of all genotypes. The amount of each metabolite was not expected to differ between wild type and heterozygous zebrafish and was expected to be elevated in homozygous zebrafish. The same extraction protocol (described above) was used to quantify glutaric acid and 3-hydroxyglutaric acid; the internal standard for the organic acid screen was replaced by 12.5µL of 0.05mg/mL d53-hydroxyglutaric acid and 50 µL of 0.05mg/mL d4glutaric acid internal standards. Once samples were dried, 50µL of BSTFA/TMCS was added, and each was placed in an 80°C heating block for 20 minutes, then allowed to cool at room temperature for about 5 minutes. Once cool, the entire volume was moved into a glass autosampler GC/MS vial. Samples were run using an Agilent 6890N GC/MS.

The abundance of ions unique to glutaric acid and 3-hydroxyglutaric acid were measured by GC/MS and compared to the abundance of ions unique to the deuterated internal standard for each. For each metabolite, the amount of each was calculated as micrograms of glutaric/3-hydroxyglutaric acid per 100 mg wet weight fish. A one-way

ANOVA was used to determine if the amount of glutaric acid or 3-hydroxyglutaric acid differed between genotypes.

Glutaryl-CoA Dehydrogenase Enzyme Assay:

The specific activity of GCDH was measured for fish of all genotypes. The activity of GCDH was expected to be reduced in heterozygous fish compared to wild type fish, and the activity of homozygous zebrafish was expected to be reduced compared to both heterozygous and wild type fish. Fish used for the enzyme assay were homogenized in

0.02M phosphate buffer (pH 7.4) containing 0.208mg/mL FAD and sonicated for 30 cycles at 20% duty. 100µL of the homogenate was incubated in a 1mL reaction with

10µL of [1,5-14C] -glutaryl-CoA in 0.02M potassium phosphate (pH 7.4) containing

0.02M 1,4-dithiothreitol (DTT) and 3.06mg phenazine methosulfate (PMS) for 10 minutes at 37°C. Reactions were stopped by adding 0.5mL of 25% trichloroacetic acid and left at 37°C for 40 minutes to evolve and trap CO2. 14CO2 was trapped in 10% potassium hydroxide (KOH) and counted in a 1900CA Packard liquid scintillation analyzer. Protein concentration in homogenates was determined by Bradford assay and was used to calculate a specific activity of GCDH in each fish sample which is reported as nM CO2/hr/mg protein. A one-way ANOVA was performed to determine if the specific activity of GCDH differed between genotypes.

Results

C-Start Response Testing:

More than 150 zebrafish were tested for C-start response daily between the ages of 4 days post fertilization and 8 days post fertilization. These fish were wild type, heterozygotes, and homozygotes but the exact number of each genotype is unknown.

For all fish tested a normal C-start response was observed each day, and all fish developed normally. The fish were allowed to continue to grow, and all survivors were euthanized after reaching 3-months of age for the metabolic testing described above.

Not all fish tested for a C-start response survived to 3-months of age, but the survival rate was no different than that observed from wild type x wild type mating.

Quantitative Amino Acid Analysis:

Thirteen adult fish (wild type n=5, heterozygous n=4, and homozygous n=4) were used for quantitative amino acid analysis. Amino acid levels were calculated for 27 different amino acids, and all had comparable levels regardless of fish genotype (See

Table 8).

Organic Acid Screen:

Organic acids were examined in seventeen adult fish (wild type n=7, heterozygous n=5, and homozygous n=5). Peaks identified in all zebrafish included lactic acid, phosphate, fumaric acid, lactyl-lactate, salicylic acid, and malic acid (See

Figure 15). Identified in most zebrafish were peaks of oxalic acid, succinic acid, adipic acid, citric acid, azelaic acid, and palmitic acid. There were no qualitative differences noted between organic acid patterns from wild type and heterozygous fish. One heterozygote fish had a small amount of glutaric acid, but no glutaric acid could be detected in any other wild type or heterozygote scans. Of the five homozygous fish, three had trace amounts of glutaric. One homozygous fish had an elevated peak of

2-hydroxyglutaric acid which was not seen in any other fish, and 3-hydroxyglutaric acid was not detected in any fish regardless of genotype.

Glutaric Acid and 3-Hydroxyglutaric Acid Quantitation by Stable Isotope

Dilution:

Fifteen adult fish (wild type n=5, heterozygous n=4, and homozygous n=6) were used for glutaric acid and 3-hydroxyglutaric acid quantitation. The homozygous zebrafish had elevated glutaric and 3-hydroxyglutaric acid compared to wild type and heterozygous zebrafish. The elevated glutaric acid level was not significant

(F(2,12)=1.65, p=0.35) but the elevated 3-hydroxyglutaric acid level is significant

(F(2,12)=10.17, p=0.003). For glutaric acid, wild type fish had an average of 0.13µg glutaric/100mg wet weight (range = 0.07 to 0.19), heterozygotes had an average of

0.12µg glutaric/100mg wet weight (range = 0.07 to 0.19), and homozygotes had an average of 0.28µg glutaric/100mg wet weight (range = 0.10 to 0.72) (See Figure 16).

For 3-hydroxyglutaric acid, wild type fish had an average of 0.02µg

3-hydroxyglutaric/100mg wet weight (range = 0.01 to 0.02), heterozygotes had an average of 0.02µg 3-hydroxyglutaric/100mg wet weight (range = 0.02 to 0.03), and homozygotes had an average of 0.04µg 3-hydroxyglutaric/100mg wet weight (range =

0.02 to 0.06) (See Figure 17). All the 3-hydroxyglutaric acid quantifications were near the limit of detection for the GC/MS.

Glutaryl-CoA Dehydrogenase Enzyme Assay:

A total of twenty-six adult fish (wild type n=9, heterozygous n=10, and homozygous n=7) were used for the glutaryl-CoA dehydrogenase enzyme assay. Wild type fish had an average specific activity of 46.3nM CO2/hr/mg protein (a range = 28.8 to 60.6), heterozygotes had an average of 31.9nM CO2/hr/mg protein (a range = 15.5 to

47.2), and homozygotes had an average of 13.4nM CO2/hr/mg protein (a range of 10.3

to 19.2) (See Figure 18). Heterozygotes had, on average, 68.9% of wild type activity, and homozygotes had an average of 28.3% of wild type activity. The reduction of GCDH activity in homozygous zebrafish was significant (F(2,23)=33.71, p<0.00001), compared to both heterozygous and wild type zebrafish.

Discussion

Homozygous GCDH A IVS6+2 T>C zebrafish did not show any signs of behavioral or developmental changes when compared to heterozygous or wild type siblings.

Behavioral changes were assessed using the C-start test once a day between 4 days post fertilization and 8 days post fertilization. In the MADD zebrafish model, aberrant movement can be detected by the C-start test as early as 2 days post fertilization which progresses into paralysis 27. For the GCDH A IVS6+2 T>C zebrafish, there is no abnormal behavior or movement under normal conditions, however, like the GCDH -/- mouse model additional stress from increased lysine may cause a movement phenotype in the homozygous animals.

In humans, there are no changes in the levels of free amino acids between GA1 patients and normal individuals. Free amino acids were measured in wild type, heterozygous and homozygous GCDH A IVS6+2 T>C zebrafish to determine if there were any differences between the genotypes. As expected, no differences in free amino acids were measured (See Table 8). It is important to note that the number of fish tested with each genotype was small.

In humans, it is possible to differentiate GA1 patients from normal individuals by urine organic acid screening; most GA1 patients excrete greater than normal amounts of glutaric and 3-hydroxyglutaric acid. These elevations were not observed in the whole

fish organic acid screens. The same core group of organic acids was identified in all fish, and the pattern was not drastically changed by genotype. There was no distinct, elevated peak of glutaric acid observed on any screen, but ions for glutaric acid could be found on the leading edge of the salicylic acid peak in four of the seventeen screens, i.e., in one heterozygote and 3 homozygotes.

Glutaric and 3-hydroxyglutaric acids can be measured by adding known amounts of deuterated glutaric or 3-hydroxyglutaric acid as internal standards. When measured in this manner, whole homozygous GCDH A IVS6+2 T>C zebrafish had very slightly elevated levels of glutaric and 3-hydroxyglutaric acid, but only for the latter was the increase statistically significant. On average, homozygous fish had twice as much glutaric acid as wild type or heterozygotes, but it is important to note that the amounts were very low and, in the homozygous fish, very variable. Homozygous fish also had twice as much 3-hydroxyglutaric acid as wild type or heterozygotes, but the levels in all fish were at the limit of detection. Further, the number of animals of all genotypes tested is quite small and underpowered for statistical analysis. In summary, if the homozygous fish do indeed accumulate metabolites, accumulation is far less than in human GA1.

The activity of GCDH in human GA1 is undetectable or very low, with activities in carriers being intermediate between activities seen in homozygotes and normal subjects. Current guidelines state that enzyme activity level of 30% or less is sufficient to cause disease2. In the GCDH A IVS6+2 T>C zebrafish, the presence of one mutant allele reduces enzyme activity by approximately 30% and the presence of two alleles reduces activity by approximately 70% (p<0.0001). Homozygous animals display a

narrow range of specific activity and have a slight overlap with heterozygous animals.

In contrast, the ranges for both the heterozygous and wild type animals display considerable overlap with each other and are three times larger than that of the homozygotes (See Figure 18). In view of the only minimal metabolite accumulation in homozygous zebrafish, their residual enzyme activity appears sufficient to prevent disease. There are no observable behavior or developmental differences in homozygous fish compared to wild type or carrier siblings. Like the GCDH -/- mouse model a phenotype may only occur once the animals are stressed with exposure to high lysine, and this should be tested in the GCDH A IVS6+2 T>C zebrafish at a variate of developmental stages.

In conclusion, the GCDH A IVS6+2 T>C homozygous zebrafish does have significantly reduced GCDH activity and may accumulate small amounts of glutaric acid and 3-hydroxygultaric acid. However, without demonstrating a behavioral or developmental consequence of the reduced enzyme activity and metabolite accumulation it is difficult to claim that the GCDH A IVS6+2 T>C zebrafish is a model of

GA1. Because these animals do show reduced enzyme activity, zebrafish GCDH A is at least in part responsible for the oxidation of glutaryl-CoA, and further reduction of this activity may prove sufficient to generate a GA1 model. It is also possible that a model will only be generated if GCDH B is also deficient.

Figures

Figure 14: Structure of GCDH in Humans, Mice, and Zebrafish. The GCDH gene in humans and mice is highly similar, and the human GCDH gene is also highly similar to GCDH A in the zebrafish. All three of these genes have 11 exons and 10 introns with identical intron/exon junctions. In contrast, GCDH B in the zebrafish only has 9 exons and 8 introns. Looking at the predicted protein product, GCDH B is lacking part of the N- terminus which encodes for the mitochondrial leader sequence in the human GCDH, mouse GCDH, and GCDH A of zebrafish. All four GCDH genes shown encode the active site of GCDH.

Table 8: Quantified Amino Acids in Zebrafish.

Amino Acid Wild Type (TT) Heterozygote (TC) Homozygote (CC) One-way (n=5) (n=4) (n=4) Anova P-value Mean +/- SEM Mean +/- SEM Mean +/- SEM (nmol/mL) (nmol/mL) (nmol/mL)

Taurine 1829 +/- 123 1863 +/- 159 2171 +/- 404 0.575 Aspartic Acid 117 +/- 13 135 +/- 24 104 +/- 15 0.507 Threonine 179 +/- 20 220 +/- 30 207 +/- 62 0.743 Serine 198 +/- 11 226 +/- 34 175 +/- 22 0.341 Glutamic Acid 268 +/- 16 288 +/- 40 276 +/- 65 0.944 Alpha-Aminoadipic 5 +/- 1 8 +/- 2 7 +/- 2 0.543 Acid Glycine 207 +/- 22 209 +/- 17 207 +/- 14 0.997 Alanine 319 +/- 11 343 +/- 34 348 +/- 21 0.617 Citrulline 12 +/- 1 12 +/- 1 12 +/- 1 0.996 Alpha-Aminobutyric 6 +/- 1 7 +/- 1 9 +/- 4 0.651 Acid Valine 144 +/- 14 193 +/- 33 146 +/- 30 0.357 Cystine 2 +/- 0 2 +/- 1 3 +/- 1 0.165 Methionine 64 +/- 7 81 +/- 18 58 +/- 10 0.394 Cystathionine 5 +/- 2 8 +/- 2 8 +/- 5 0.822 Isoleucine 101 +/- 12 137 +/- 28 109 +/- 10 0.352 Leucine 185 +/- 13 259 +/- 53 190 +/- 33 0.285 Tyrosine 77 +/- 5 109 +/- 25 84 +/- 14 0.358 Beta-Alanine 4 +/- 1 4 +/- 1 3 +/- 2 0.771 Phenylalanine 100 +/- 10 137 +/- 36 94 +/- 16 0.384 Ornithine 9 +/- 1 10 +/- 1 12 +/- 3 0.375 Lysine 285 +/- 23 344 +/- 63 299 +/- 77 0.731

1-Methylhistidine 2 +/- 0 3 +/- 1 3 +/- 1 0.246 Histidine 478 +/- 75 559 +/- 65 765 +/- 200 0.276 Tryptophan 12 +/- 1 17 +/- 4 15 +/- 1 0.392 Arginine 211 +/- 29 282 +/- 72 204 +/- 21 0.431 Hydroxyproline 12 +/- 2 11 +/- 3 19 +/- 12 0.679 Proline 159 +/- 22 198 +/- 34 236 +/- 126 0.743

Figure 15: Organic Acid Screen of a Wild Type Fish vs. a Homozygous Fish. A compairison of a typical wild type zebrafish (A) and a typical homozygous zebrafish (B) organic acid screen. All heterozygous scans resembled the wild type scans. Labeled peaks are 1. Lactic acid 2. Internal Standard 3. Phosphate 4. Fumaric acid 5. Lactyl lactate 6. Salicylic acid 7. Malic acid 8. Salicylic acid 9. External Standard C-24. In one heterozygous and three homozygous zebrafish had identifiable ions of glutaric acid on the leading edge of peak 6.

Quantative Glutaric by Genotype 0.40

0.35

0.30

0.25

0.20

0.15

0.10

0.05 mcg Glutaric/100 mg Wet Weight Fish Weight mg Wet Glutaric/100 mcg 0.00 Wild Type (TT) Heterozygote (TC) Homozygote (CC)

Wild Type (TT) Heterozygote (TC) Homozygote (CC)

Number of Fish Tested n = 5 N = 4 N = 6

Average 0.126 0.123 0.277

Standard Error of the Mean 0.03 0.03 0.10

Measured glutaric acid presented as mg glutaric/100mg wet weight fish

0.07 0.07 0.10

0.08 0.08 0.10

0.11 0.15 0.14

0.18 0.19 0.26

0.19 0.34

0.72

Figure 16: Quantified Glutaric in Zebrafish. All measurements are in micrograms glutaric/100mg wet weight fish. On average homozygous zebrafish have twice as much glutaric acid as wild type or heterozygous zebrafish. A: graph of group means +/- SEM. B: table of raw data with, average glutaric and standard error of the mean.

Quantative 3-Hydroxyglutaric by Genotype 0.06

0.05 P = .* 0.04

0.03

Weight 0.02

0.01

0.00 mcg 3-Hydroxyglutaric/100 mg Wet Wet mg 3-Hydroxyglutaric/100 mcg Wild Type (TT) Heterozygote (TC) Homozygote (CC)

Wild Type (TT) Heterozygote (TC) Homozygote (CC)

Number of Fish Tested n = 5 N = 4 N = 6

Average 0.016 0.025 0.042

Standard Error of the Mean 0.00 0.00 0.01

Measured 3-hydroxyglutaric acid presented as mg 3-OH/100mg wet weight fish

0.01 0.02 0.02

0.01 0.02 0.04

0.02 0.03 0.04

0.02 0.05

0.06

Figure 17: Quantified 3-Hydroxyglutaric Acid in Zebrafish. All measurements are in micrograms 3-hydroxyglutaric acid/100mg wet weight fish. On average homozygous zebrafish have twice as much 3-hydroxyglutaric acid as wild type or heterozygous zebrafish. This increase is statistically significant when measured using a 3-way ANOVA (p=0.002). A: graph of group means +/- SEM. B: table of raw data with, average glutaric and standard error of the mean.

P < . *

Wild Type (TT) Heterozygote (TC) Homozygote (CC)

Fish Tested (n) 9 10 7

Standard Error of the 3.3 2.6 1.4

Mean

Average Specific 46.3 nmol CO2/hr/mg 31.9 nmol CO2/hr/mg 13.4 nmol CO2/hr/mg

Activity protein protein protein

% of Average Wild Type 100% 68.90% 28.30%

Activity

Figure 18: Specific Activity of GCDH in Whole Zebrafish. The presence of one mutant allele reduces GCDH activity by 30% (heterozygoes) and the presence of two mutant alleles significantly reduces GCDH activity by 70% (homozygotes). This data is presented in a box plot with x representing the mean, the box defining the 25th and 75th percentiles, and the wiskers indicating the minimum and maximum values.

CHAPTER VI

CONCLUSIONS AND FUTURE DIRECTIONS

Summary and Future Directions

Glutaric acidemia type I (GA1) is a well-described inherited metabolic disease with a known natural history, a typical clinical course and two defined biochemical phenotypes. There are also atypical clinical presentations about which less is known.

GA1 patients have two pathogenic glutaryl-CoA dehydrogenase (GCDH) alleles and are commonly compound heterozygotes. Nearly all described mutations of GCDH are missense mutations. GA1 patients can be identified by newborn screening, and their diagnoses can be confirmed by identification of two pathogenic mutant alleles in GCDH or by an enzyme assay of GCDH demonstrating insufficient activity. Finally, GA1 patients are treated by a combined therapy of reduced dietary lysine/protein, supplemented dietary carnitine, and with an emergency treatment protocol to prevent catabolism during at risk periods. Despite all that is known, there remain gaps and problems with screening, diagnosis, and treatment, and there are many unanswered questions concerning the pathogenesis and biochemistry of GA1. My thesis has addressed many of these gaps and suggests several avenues of approach towards learning more about this disorder.

Striatal Damage in GA1:

Chapter two studies postmortem GA1 brains to test the hypothesis that medium spiny neurons are selectively lost during an encephalopathic crisis. This is an assertion that has been repeated throughout the literature on GA1 but for which there is little evidence. I attempted to address this issue by staining postmortem putamen samples of

GA1 patients and age matched controls immunohistochemically against DARPP-32, a protein unique to medium spiny neurons in the striatum.

In my study, there was no difference in the average number of medium spiny neurons in the putamen when normalized to a putaminal area. The total volume of the entire striatum, caudate and putamen, was reduced in the GA1 patients compared to age matched controls, so GA1 samples do have fewer medium spiny neurons overall; however, this cell type does not appear to be selectively killed following an encephalopathic crisis as suggested by the literature. In Funk et al. a loss of medium sized neurons was observed in GA1 patients but, nowhere in the study is this loss of neurons described in the context of the reduced striatal volume. An additional observation made in my study was the identification of reactive astrocytes stained with

DARPP-32. Previous studies, including Funk et al., have noted the presence of reactive astrocytes in GA1 postmortem brain samples taken shortly following encephalopathic crisis. It has been hypothesized that these reactive astrocytes are acting to remove cells that were damaged in the crisis. Because these cells show positive staining with the

DARPP-32 antibody, they must have phagocytosed damaged medium spiny neurons.

This provides experimental evidence showing that the reactive astrocytes, thought to be present in the brains of all GA1 patients following encephalopathic crisis, are functioning to phagocytose damaged cells.

My work, described in chapter 2, suggests that DARPP-32 positive cells are not preferentially reduced in GA1 patients. Thus, the neuronal loss described by Funk et al. may result from the loss of medium sized neurons other than medium spiny neurons.

However, my work only examines a single cross section of the putamen for GA1

patients and controls. To properly conclude that medium spiny neurons are not selectively lost in GA1 patients after an encephalopathic crisis, serial sections of the putamen and caudate, spanning the entire striatum, should be stained for DARPP-32 expression in multiple autopsy patient samples. With a more encompassing study, one could definitively determine if medium spiny neurons are selectively lost in patients after an encephalopathic crisis or if a different cell population in the striatum is reduced.

Diagnosis of GA1:

Chapter three describes a study using deep sequencing to identify GCDH mutations in a GA1 cohort. This work highlights the fact that current clinical sequencing of GCDH should identify most , and all introns were screened for pathogenicGA patients. mutations, While none the were promoter, identified UTR. After analyzing all GA1 patients, one patient with a single pathogenic allele remains troubling. It has been reconfirmed that fibroblast cells from this patient have no GCDH activity, which suggests that a second pathogenic mutation must exist on the other GCDH allele.

Currently, this cell line is being assayed for duplication or deletion events by microarray and, while it is likely that this will identify a second pathogenic allele, it is also possible that a regulatory mutation exists outside of the region already sequenced.

There is one assumption and one technical problem with the methods described in chapter three that could prevent the identification of a second pathogenic mutation.

First, this study assumes that the minimal GCDH promoter in humans is the same as in mice. The GCDH promoter region has not been described in humans but has been described using the GCDH-/- mouse model. For this study, the analogous region in

humans was identified and sequenced. There is no experimental evidence to suggest that this region contains the entire human GCDH promoter and there may be a mutation upstream of the examined region in the cell line with only one identified mutation. A second is a method problem in which all four initial PCR products were combined by concentration and not the total amount of DNA. This created an artificial copy number variation in my data set and removed any ability to look at true copy number changes in my samples. This is a technical error that was only detected after examining the sequencing data. By adding equal amounts of total DNA for all PCR products, it would be possible to identify duplications or deletions from the sequencing data in a revised method.

While most GA1 patients will be identified by sequencing the exons in GCDH, this improved method would be invaluable for future experiments. It would be interesting to use this approach on a cohort of suspected GA1 patients who have an enzyme activity in the overlap region of patients and carriers, and only one mutation identified in GCDH by clinical sequencing. I hypothesize that this cohort of harbors regulatory changes in GCDH a proposal I would like to test. In conclusion,patients for diagnostic purposes, current– clinical methods should detect most patients. However, the deep sequencing method has been shown to be able to distinguish carriers from true GA1 patients and would allow for identification of SNPs and copy number changes in a single multiplexed test instead of using both sequencing of individual exons and screening for copy number changes by aCGH, a two-step process, as done now clinically.

DHTKD1 and 2-Ketoadipic and 2-Aminoadipic Acidemia:

A growing concern is the development of a treatment that can easily be administered throughout the lifetime of a GA1 patient that would reduce or eliminate exposure to the toxin. This would prevent striatal necrosis from acute toxicity early in life as well as possible white matter disease and kidney failure due to prolonged exposure to a toxin. One hypothesis is that a block in the lysine oxidation pathway upstream of GCDH would prevent accumulation of the toxin and could be used as a treatment for GA1 patients. Looking at the lysine oxidation pathway, a possible target of such a treatment is the 2-ketoadipic acid dehydrogenase complex.

As an initial step to testing this hypothesis, I performed exome sequencing on patients with 2-ketoadipic and 2-aminoadipic acidemia (KAA) to identify the subunit in the 2-ketoadipic acid dehydrogenase complex responsible for their biochemical marker.

I found a homozygous change in exon 5 (c.915G>C p.Gln305His) of dehydrogenase E1 and transketolase domain containing 1(DHTKD1). Two additional studies have identified mutations in DHTKD1 responsible for the same biochemical marker in 11 additional

KAA patients. Together, these studies have confirmed that mutations in DHTKD1 cause

KAA and that DHTKD1 encodes the E1 subunit of the 2-ketoadipic acid dehydrogenase complex. With the identification of DHTKD1, it is now possible to test if blocking this enzyme complex is a viable treatment option for GA1 patients.

Additional studies of DHTKD1 function would further our understanding of the lysine oxidation pathway and may provide new insights into GA1. There are two distinctly different disorders that appear to be caused by mutations in DTHDK1.

Recessive loss of function mutations cause biochemical accumulation of 2-aminoadipic

acid and 2-ketoadipic acid without a clear clinical disease 94,98,99,142. In addition, a particular truncation mutation (c.1455 T>G p.Tyr485*) appears to cause dominant

Charcot-Marie-Tooth type 2 disease, implying that different defects in DHTKD1 may cause very different phenotypes.

Future experiments altering DHTKD1 expression could be done to examine how altered regulation affects the levels of 2-ketoadipic acid and 2-aminoadipic acid. These experiments could be done using knockdown and overexpression assays in cell culture or in a whole organism knockout, such as a mouse or zebrafish, to characterize the effect of no DHTKD1 activity. Another experiment that would help clarify the effect of

DHTKD1 mutations would be to express the mutation associated with dominant

Charcot-Marie-Tooth type 2 (Tyr485*) in cell culture and to look for elevated metabolites in the cell media. This would be one way to test if the dominant mutation in

DHTKD1 affects the lysine oxidation pathway and would suggest that patients with elevated 2-ketoadipic acid and 2-aminoadipic acid are at risk for the later-onset neurological disease Charcot-Marie-Tooth type 2.

Towards a Zebrafish Model of GA1:

Chapter five describes work done to characterize a zebrafish with an IVS6+2 T>C mutation in GCDH A. In zebrafish, there are two orthologous genes that encode for the

GCDH enzyme; GCDH A and GCDH B. The predicted mRNA from GCDH A has a mitochondrial leader sequence and an active site that is identical to the active site in humans and mice. In contrast, the predicted mRNA from GCDH B has the same active site, identical to the active site in humans and mice, but is missing a mitochondrial leader sequence. As part of my thesis work, I generated fish homozygous for a splice

site mutation in GCDH A and compared them to heterozygous and wild type siblings for the following: behavior, development, amino acid profile, organic acid profile, glutaric acid and 3-hydroxygluaric acid level, and GCDH enzyme activity. This was done to determine if the GCDH A (IVS6+2 T>C) zebrafish line should be used as a model of GA1 for future experiments.

I found that homozygous zebrafish did not have any behavioral or developmental differences from heterozygous or wild type siblings. There was no difference in the amino acid profile by genotype either, as expected. Homozygous zebrafish did have significantly reduced enzyme activity (p<0.0001) as measured by a

CO2 release assay and may have accumulated small amounts of glutaric acid and 3- hydroxyglutaric acid. However, despite the reduced enzyme activity, no behavioral consequences were demonstrated. Like the GCDH -/- mouse model, it is possible that these fish will show a phenotype when appropriately stressed, such as exposure to high lysine, which has not yet been performed. With the current characterization, I am hesitant to declare this fish line a model of GA1, but strongly suspect that fish with a complete loss of GCDH A can be used as a model of GA1.

The Zebrafish Mutation Project identified two GCDH A fish mutants (IVS6+2 T>C and Q351*) and one GCDH B mutant. All three were obtained by the Goodman lab at the

University of Colorado Anschutz Medical Campus, and only characterization of the

GCDH A (IVS6+2 T>C) fish has been completed. The other two fish lines should be characterized as well to determine if the GCDH A Q351* fish line can be used as a model of GA1 and to identify the function of GCDH B, or to determine if it is a pseudogene.

The identification of a zebrafish model of GA1 would allow for testing of many hypothesizes of GA1. Zebrafish provide an excellent system with which to test the effects of double mutant animals, and such an approach could be used to look for the origin of 3-hydroxyglutaric acid in GA1. A fish model of GA1 could be used to observe the effects of elevated glutaric acid and 3-hydroxyglutaric acid on the development of the neuronal network. This can be done by crossing a GA1 zebrafish with another zebrafish line expressing a fluorescent neuronal marker. The current study fails to address the effect or function of the GCDH B gene in zebrafish. By comparing this characterization of the GCDH A IVS6+2 T>C fish to a mutant GCDH B fish I could determine if GCDH B functions in lysine metabolism and if any observations of the

GCDH A IVS6+2 T>C fish are the result of a functional GCDH B protein. Despite proving that the GCDH A IVS6+2 T>C fish is not suitable for use as a model of GA1, there remains much that can be learned from this fish as well as other fish lines with mutations in

GCDH A and GCDH B.

Conclusion

In my thesis work, I furthered our understanding of GA1. I provided preliminary evidence suggesting that medium spiny neurons are not selectively killed following an encephalopathic crisis. Instead, my findings suggest that there is a proportional reduction of all cell types within the putamen, and likely the caudate as well.

Additionally, this experiment provided evidence for the roll of reactive astrocytes following encephalopathic crisis. I designed and tested a deep sequencing method of screening the entire GCDH ed regions, for mutations. Withgenomic a small region,change includingto the methodology the promoter described and untranslatin chapter

three, this method would also allow for the identification of copy number variation in

GCDH as well. This method can be used on many patients at once and would change a two-step process, with current clinical techniques, to a single method to identify GA1 patients. Additionally, I showed that this method could distinguish carriers from true

GA1 patients. With growing concern about the consequences of long term exposure to a toxin in GA1 patients, there is a growing need to improve the current treatment for patients to reduce or eliminate exposure to the toxin. One possible way to achieve this is to create a block upstream of GCDH in the lysine oxidation pathway and thus prevent the toxin from being formed. An appealing enzyme target for such treatment is the 2- ketoadpipic acid dehydrogenase complex. Part of my thesis work contributed to the identification of DHTKD1 as the E1 subunit of this complex and allows for future experimentation to test if blocking the lysine oxidation pathway at DHTKD1 is a viable treatment for GA1 patients. Finally, I have characterized a zebrafish with a splice site mutation in GCDH A and have shown that mutations in GCDH A cause reduced enzyme activity. Although the fish line described in my thesis work is not likely to be a good model of GA1 a zebrafish with a complete loss of GCDH A activity, such as the fish line with a nonsense change in exon 9, likely will be. The work completed during my time in the Goodman laboratory has furthered our understanding of GA1 and has provided an alternative diagnostic protocol and has suggested a new animal model of GA1.

REFERENCES

1. Goodman SI, Markey SP, Moe P disorder of amino acid metabolism. Biochem Med. 1975;12(1):12-21. http://www.ncbi.nlm.nih.gov/pubmed/1137568.G, Miles BS, Teng CC. Accessed Glutaric May aciduria; 2, 2017. a new

2. Boy SPN, Mühlhausen C, Maier EM, et al. Proposed recommendations for diagnosing and managing individuals with glutaric aciduria type I: second revision. Journal of Inherited Metabolic Disease. 2016:1-27.

3. Strauss KA, Puffenberger EG, Robinson DL, Morton DH. Type I glutaric aciduria, part 1: natural history of 77 patients. Am J Med Genet C Semin Med Genet. 2003;121C(1):38-52. doi:10.1002/ajmg.c.20007.

4. Goodman SI, Frerman FE. Organic Acidemias Due to Defects in Lysine Oxidation: 2-Ketoadipic Acidemia and Glutaric Acidemia. In: The Metabolic and Molecular Bases of Inherited Disease Volume II. 8th ed. ; 2001:2195-2204.

5. Funk CBR, Prasad AN, Frosk P, et al. Neuropathological, biochemical and molecular findings in a glutaric acidemia type 1 cohort. Brain. 2005;128(4):711- 722. doi:10.1093/brain/awh401.

6. Busquets C, Merinero B, Christensen E, et al. Glutaryl-CoA Dehydrogenase Deficiency in Spain: Evidence of Two Groups of Patients, Genetically, and Biochemically Distinct. Pediatr Res. 2000;48(3):315-322. doi:10.1203/00006450- 200009000-00009.

7. Kölker S, Garbade SF, Greenberg CR, et al. Natural History, Outcome, and Treatment Efficacy in Children and Adults with Glutaryl-CoA Dehydrogenase Deficiency. Pediatr Res. 2006;59(6):840-847. doi:10.1203/01.pdr.0000219387.79887.86.

8. Boy N, Heringer J, Brackmann R, et al. Extrastriatal changes in patients with late- onset glutaric aciduria type I highlight the risk of long-term neurotoxicity. Orphanet J Rare Dis. 2017;12(1):77. doi:10.1186/s13023-017-0612-6.

9. Heringer J, Boy SPPN, Ensenauer R, et al. Use of guidelines improves the neurological outcome in glutaric aciduria type I. Ann Neurol. 2010;68(5):743-752. doi:10.1002/ana.22095.

10. Higashino K, Fujioka M, Yamamura Y. The conversion of L-lysine to saccharopine and alpha-aminoadipate in mouse. Arch Biochem Biophys. 1971;142(2):606-614. http://www.ncbi.nlm.nih.gov/pubmed/4396286. Accessed June 5, 2017.

11. Blemings KP, Crenshaw TD, Swick RW, Benevenga NJ. Lysine-alpha-ketoglutarate reductase and saccharopine dehydrogenase are located only in the mitochondrial matrix in rat liver. J Nutr. 1994;124(8):1215-1221. http://www.ncbi.nlm.nih.gov/pubmed/8064371. Accessed June 5, 2017.

12. Markovitz PJ, Chuang DT, Cox RP. Familial hyperlysinemias. Purification and characterization of the bifunctional aminoadipic semialdehyde synthase with lysine-ketoglutarate reductase and saccharopine dehydrogenase activities. J Biol Chem. 1984;259(19):11643-11646. http://www.ncbi.nlm.nih.gov/pubmed/6434529. Accessed May 30, 2017.

13. Hutzler J, Dancis J. Conversion of lysine to saccharopine by human tissues. Biochim Biophys Acta. 1968;158(1):62-69. http://www.ncbi.nlm.nih.gov/pubmed/4385118. Accessed May 30, 2017.

14. Chang YF. Lysine metabolism in the human and the monkey: demonstration of pipecolic acid formation in the brain and other organs. Neurochem Res. 1982;7(5):577-588. http://www.ncbi.nlm.nih.gov/pubmed/6811962. Accessed May 30, 2017.

15. Hallen A, Jamie JF, Cooper AJL. Lysine metabolism in mammalian brain: an update on the importance of recent discoveries. Amino Acids. 2013;45(6):1249-1272. doi:10.1007/s00726-013-1590-1.

16. Chang YF. Pipecolic acid pathway: the major lysine metabolic route in the rat brain. Biochem Biophys Res Commun. 1976;69(1):174-180. http://www.ncbi.nlm.nih.gov/pubmed/1259753. Accessed May 30, 2017.

17. Chang YF. Lysine metabolism in the rat brain: blood-brain barrier transport, formation of pipecolic acid and human hyperpipecolatemia. J Neurochem. 1978;30(2):355-360. http://www.ncbi.nlm.nih.gov/pubmed/624942. Accessed May 30, 2017.

18. Wanders RJ, Romeyn GJ, Schutgens RB, Tager JM. L-pipecolate oxidase: a distinct peroxisomal enzyme in man. Biochem Biophys Res Commun. 1989;164(1):550- 555. http://www.ncbi.nlm.nih.gov/pubmed/2572224. Accessed May 30, 2017.

19. Mihalik SJ, Rhead WJ. Species variation in organellar location and activity of L- pipecolic acid oxidation in mammals. J Comp Physiol B. 1991;160(6):671-676. http://www.ncbi.nlm.nih.gov/pubmed/2045546. Accessed June 5, 2017.

20. Baric I, Wagner L, Feyh P, Liesert M, Buckel W, Hoffmann GF. Sensitivity and specificity of free and total glutaric acid and 3-hydroxyglutaric acid measurements by stable-isotope dilution assays for the diagnosis of glutaric aciduria type I. J Inherit Metab Dis. 1999;22(8):867-881. http://www.ncbi.nlm.nih.gov/pubmed/10604139. Accessed May 2, 2017.

21. Gallagher RC, Cowan TM, Goodman SI, Enns GM. Glutaryl-CoA dehydrogenase deficiency and newborn screening: Retrospective analysis of a low excretor provides further evidence that some cases may be missed. Mol Genet Metab. 2005;86(3):417-420. doi:10.1016/j.ymgme.2005.08.005.

22. Viau K, Ernst SL, Vanzo RJ, Botto LD, Pasquali M, Longo N. Glutaric acidemia type 1: outcomes before and after expanded newborn screening. Mol Genet Metab. 2012;106(4):430-438. doi:10.1016/j.ymgme.2012.05.024.

23. Christensen E, Ribes A, Merinero B, Zschocke J. Correlation of genotype and phenotype in glutaryl-CoA dehydrogenase deficiency. J Inherit Metab Dis. 2004;27(6):861-868. doi:10.1023/B:BOLI.0000045770.93429.3c.

24. Schillaci L-AP, Greene CL, Strovel E, et al. The M405V allele of the glutaryl-CoA dehydrogenase gene is an important marker for glutaric aciduria type I (GA-I) low excretors. Mol Genet Metab. 2016;119(1-2):50-56. doi:10.1016/j.ymgme.2016.06.012.

25. Mushimoto Y, Fukuda S, Hasegawa Y, et al. Clinical and molecular investigation of 19 Japanese cases of glutaric acidemia type 1. Mol Genet Metab. 2011;102(3):343- 348. doi:10.1016/j.ymgme.2010.11.159.

26. Goodman SI, Stein DE, Schlesinger S, et al. Glutaryl-CoA dehydrogenase mutations in glutaric acidemia (type I): Review and report of thirty novel mutations. Hum Mutat. 1998;12(3):141-144. doi:10.1002/(SICI)1098-1004(1998)12:3<141::AID- HUMU1>3.0.CO;2-K.

27. Song Y, Selak MA, Watson CT, et al. Mechanisms underlying metabolic and neural defects in zebrafish and human multiple acyl-CoA dehydrogenase deficiency (MADD). Riley B, ed. PLoS One. 2009;4(12):e8329. doi:10.1371/journal.pone.0008329.

28. Frerman FE, Goodman SI. Defects of Electron Transfer Flavoprotein and Electron Transfer Flavoprotein-Ubiquinone Oxidoreductase: Glutaric Acidemia Type II. In: The Metabolic and Molecular Bases of Inherited Disease Volume II. 8th ed. ; 2001:2357-2365.

29. Gorelick RJ, Mizzer JP, Thorpe C. Purification and properties of electron- transferring flavoprotein from pig kidney. Biochemistry. 1982;21(26):6936-6942. http://www.ncbi.nlm.nih.gov/pubmed/7159575. Accessed June 4, 2017.

30. McKean MC, Beckmann JD, Frerman FE. Subunit structure of electron transfer flavoprotein. J Biol Chem. 1983;258(3):1866-1870. http://www.ncbi.nlm.nih.gov/pubmed/6822538. Accessed June 4, 2017.

31. Husain M, Steenkamp DJ. Electron transfer flavoprotein from pig liver mitochondria. A simple purification and re-evaluation of some of the molecular properties. Biochem J. 1983;209(2):541-545. http://www.ncbi.nlm.nih.gov/pubmed/6847633. Accessed June 4, 2017.

32. Salazar D, Zhang L, deGala GD, Frerman FE. Expression and characterization of two pathogenic mutations in human electron transfer flavoprotein. J Biol Chem. 1997;272(42):26425-26433. http://www.ncbi.nlm.nih.gov/pubmed/9334218. Accessed June 4, 2017.

33. Ruzicka FJ, Beinert H. A new iron-sulfur flavoprotein of the respiratory chain. A component of the fatty acid beta oxidation pathway. J Biol Chem. 1977;252(23):8440-8445. http://www.ncbi.nlm.nih.gov/pubmed/925004. Accessed June 4, 2017.

34. Beckmann JD, Frerman FE. Electron-transfer flavoprotein-ubiquinone oxidoreductase from pig liver: purification and molecular, redox, and catalytic properties. Biochemistry. 1985;24(15):3913-3921. http://www.ncbi.nlm.nih.gov/pubmed/4052375. Accessed June 4, 2017.

35. Beckmann JD, Frerman FE. Reaction of electron-transfer flavoprotein with electron-transfer flavoprotein-ubiquinone oxidoreductase. Biochemistry. 1985;24(15):3922-3925. http://www.ncbi.nlm.nih.gov/pubmed/2996585. Accessed June 4, 2017.

36. Arigoni F, Kaminski PA, Hennecke H, Elmerich C. Nucleotide sequence of the fixABC region of Azorhizobium caulinodans ORS571: similarity of the fixB product with eukaryotic flavoproteins, characterization of fixX, and identification of nifW. Mol Gen Genet. 1991;225(3):514-520. http://www.ncbi.nlm.nih.gov/pubmed/1850088. Accessed June 4, 2017.

37. Koeller DM, Woontner M, Crnic LS, et al. Biochemical, pathologic and behavioral analysis of a mouse model of glutaric acidemia type I. Hum Mol Genet. 2002;11(4):347-357. http://www.ncbi.nlm.nih.gov/pubmed/11854167. Accessed May 3, 2017.

38. Roberts DL, Frerman FE, Kim JJ. Three-dimensional structure of human electron transfer flavoprotein to 2.1-A resolution. Proc Natl Acad Sci U S A. 1996;93(25):14355-14360. http://www.ncbi.nlm.nih.gov/pubmed/8962055. Accessed June 4, 2017.

39. Sherman EA, Strauss KA, Tortorelli S, et al. Genetic mapping of glutaric aciduria, type 3, to chromosome 7 and identification of mutations in c7orf10. Am J Hum Genet. 2008;83(5):604-609. doi:10.1016/j.ajhg.2008.09.018.

40. Marlaire S, Van Schaftingen E, Veiga-da-Cunha M. C7orf10 encodes succinate- hydroxymethylglutarate CoA-transferase, the enzyme that converts glutarate to glutaryl-CoA. J Inherit Metab Dis. 2014;37(1):13-19. doi:10.1007/s10545-013- 9632-0.

41. Knerr I, Zschocke J, Trautmann U, et al. Glutaric aciduria type III: a distinctive non-disease? J Inherit Metab Dis. 2002;25(6):483-490. http://www.ncbi.nlm.nih.gov/pubmed/12555941. Accessed June 4, 2017.

42. McCabe ER, Goodman SI, Fennessey P V, Miles BS, Wall M, Silverman A. Glutaric, 3-hydroxypropionic, and lactic aciduria with metabolic acidemia, following extensive small bowel resection. Biochem Med. 1982;28(2):229-236. http://www.ncbi.nlm.nih.gov/pubmed/7181872. Accessed May 28, 2017.

43. Wendel U, Bakkeren J, de Jong J, Bongaerts G. Glutaric aciduria mediated by gut bacteria. J Inherit Metab Dis. 1995;18(3):358-359. http://www.ncbi.nlm.nih.gov/pubmed/7474906. Accessed May 28, 2017.

44. Molven A, Matre GE, Duran M, et al. Familial hyperinsulinemic hypoglycemia caused by a defect in the SCHAD enzyme of mitochondrial fatty acid oxidation. Diabetes. 2004;53(1):221-227. http://www.ncbi.nlm.nih.gov/pubmed/14693719. Accessed May 28, 2017.

45. Jafari P, Braissant O, Zavadakova P, Henry H, Bonafé L, Ballhausen D. Ammonium accumulation and cell death in a rat 3D brain cell model of glutaric aciduria type I. Gillingwater TH, ed. PLoS One. 2013;8(1):e53735. doi:10.1371/journal.pone.0053735.

46. Olivera-Bravo S, Fernández A, Sarlabós MN, et al. Neonatal astrocyte damage is sufficient to trigger progressive striatal degeneration in a rat model of glutaric acidemia-I. Ferreira ST, ed. PLoS One. 2011;6(6):e20831. doi:10.1371/journal.pone.0020831.

47. Olivera-Bravo S, Ribeiro CAJ, Isasi E, et al. Striatal neuronal death mediated by Hum Mol Genet. 2015;24(16):4504-4515. doi:10.1093/hmg/ddv175. astrocytes from the Gcdh−/− mouse model of glutaric acidemia type ).

48. Zinnanti WJ, Lazovic J, Housman C, et al. Mechanism of age-dependent susceptibility and novel treatment strategy in glutaric acidemia type I. J Clin Invest. 2007;117(11):3258-3270. doi:10.1172/JCI31617.

49. Zinnanti WJ, Lazovic J, Wolpert EB, et al. A diet-induced mouse model for glutaric aciduria type I. Brain. 2006;129(4):899-910. doi:10.1093/brain/awl009.

50. Sauer SW, Opp S, Hoffmann GF, Koeller DM, Okun JG, Kölker S. Therapeutic modulation of cerebral L-lysine metabolism in a mouse model for glutaric aciduria type I. Brain. 2011;134(1):157-170. doi:10.1093/brain/awq269.

51. Zschocke J, Quak E, Guldberg P, Hoffmann GF. Mutation analysis in glutaric aciduria type I. J Med Genet. 2000;37(3):177-181. http://www.ncbi.nlm.nih.gov/pubmed/10699052. Accessed May 2, 2017.

52. Greenberg CR, Reimer D, Singal R, et al. A G-to-T transversion at the +5 position of intron 1 in the glutaryl CoA dehydrogenase gene is associated with the Island Lake variant of glutaric acidemia type I. Hum Mol Genet. 1995;4(3):493-495. http://www.ncbi.nlm.nih.gov/pubmed/7795610. Accessed May 2, 2017.

53. Haworth JC, Booth FA, Chudley AE, et al. Phenotypic variability in glutaric aciduria type I: Report of fourteen cases in five Canadian Indian kindreds. J Pediatr. 1991;118(1):52-58. http://www.ncbi.nlm.nih.gov/pubmed/1986098. Accessed May 2, 2017.

54. Biery BJ, Stein DE, Morton DH, Goodman SI. Gene structure and mutations of glutaryl-coenzyme A dehydrogenase: impaired association of enzyme subunits that is due to an A421V substitution causes glutaric acidemia type I in the Amish. Am J Hum Genet. 1996;59(5):1006-1011. http://www.ncbi.nlm.nih.gov/pubmed/8900227. Accessed May 2, 2017.

55. Strauss KA, Lazovic J, Wintermark M, Morton DH. Multimodal imaging of striatal degeneration in Amish patients with glutaryl-CoA dehydrogenase deficiency. Brain. 2007;130(7):1905-1920. doi:10.1093/brain/awm058.

56. Basinger AA, Booker JK, Frazier DM, Koeberl DD, Sullivan JA, Muenzer J. Glutaric acidemia type 1 in patients of Lumbee heritage from North Carolina. Mol Genet Metab. 2006;88(1):90-92. doi:10.1016/j.ymgme.2005.12.008.

57. van der Watt G, Owen EP, Berman P, et al. Glutaric aciduria type 1 in South Africa- high incidence of glutaryl-CoA dehydrogenase deficiency in black South Africans. Mol Genet Metab. 2010;101(2-3):178-182. doi:10.1016/j.ymgme.2010.07.018.

58. Greenberg C, Duncan AM, Gregory CA, Singal R, Goodman SI. Assignment of Human Glutaryl-CoA Dehydrogenase Gene (GCDH) to the Short Arm of Chromosome 19 (19p13.2) by in Situ Hybridization and Somatic Cell Hybrid Analysis. Genomics. 1994;21(1):289-290. doi:10.1006/geno.1994.1264.

59. Schwartz M, Christensen E, Superti-Furga A, Brandt NJ. The human glutaryl-CoA dehydrogenase gene: report of intronic sequences and of 13 novel mutations causing glutaric aciduria type I. Hum Genet. 1998;102(4):452-458. http://www.ncbi.nlm.nih.gov/pubmed/9600243. Accessed May 30, 2017.

100

60. Lenich AC, Goodman SI. The purification and characterization of glutaryl- coenzyme A dehydrogenase from porcine and human liver. J Biol Chem. 1986;261(9):4090-4096. http://www.ncbi.nlm.nih.gov/pubmed/3081514. Accessed May 30, 2017.

61. Goodman SI, Kratz LE, DiGiulio KA, et al. Cloning of glutaryl-CoA dehydrogenase cDNA, and expression of wild type and mutant enzymes in Escherichia coli. Hum Mol Genet. 1995;4(9):1493-1498. http://www.ncbi.nlm.nih.gov/pubmed/8541831. Accessed May 30, 2017.

62. Koeller DM, DiGiulio KA, Angeloni S V, et al. Cloning, structure, and chromosome localization of the mouse glutaryl-CoA dehydrogenase gene. Genomics. 1995;28(3):508-512. doi:10.1006/geno.1995.1182.

63. Wang M, Fu Z, Paschke R, Goodman S, Frerman FE, Kim JJ. The Crystal Structure and Mechanism of Human Glutaryl-CoA Dehydrogenase. doi.org. doi:10.2210/pdb1siq/pdb.

64. Schmiesing J, Schlüter H, Ullrich K, Braulke T, Mühlhausen C. Interaction of glutaric aciduria type 1-related glutaryl-CoA dehydrogenase with mitochondrial matrix proteins. Pereira IAC, ed. PLoS One. 2014;9(2):e87715. doi:10.1371/journal.pone.0087715.

65. CRANE FL, BEINERT H. On the mechanism of dehydrogenation of fatty acyl derivatives of coenzyme A. II. The electron-transferring flavoprotein. J Biol Chem. 1956;218(2):717-731. http://www.ncbi.nlm.nih.gov/pubmed/13295225. Accessed May 30, 2017.

66. CRANE FL, GLENN JL, GREEN DE. Studies on the electron transfer system. IV. The electron transfer particle. Biochim Biophys Acta. 1956;22(3):475-487. http://www.ncbi.nlm.nih.gov/pubmed/13382877. Accessed May 30, 2017.

67. Woontner M, Crnic LS, Koeller DM. Analysis of the Expression of Murine Glutaryl- CoA Dehydrogenase: In Vitro and in Vivo Studies. Mol Genet Metab. 2000;69(2):116-122. doi:10.1006/mgme.2000.2962.

68. Lindner M, Ho S, Fang-Hoffmann J, Hoffmann GF, Kölker S. Neonatal screening for glutaric aciduria type I: Strategies to proceed. J Inherit Metab Dis. 2006;29(2- 3):378-382. doi:10.1007/s10545-006-0284-1.

101

69. Loeber JG, Burgard P, Cornel MC, et al. Newborn screening programmes in Europe; arguments and efforts regarding harmonization. Part 1 - From blood spot to screening result. J Inherit Metab Dis. 2012;35(4):603-611. doi:10.1007/s10545-012-9483-0.

70. Pfeil J, Listl S, Hoffmann GF, Kölker S, Lindner M, Burgard P. Newborn screening by tandem mass spectrometry for glutaric aciduria type 1: a cost-effectiveness analysis. Orphanet J Rare Dis. 2013;8(1):167. doi:10.1186/1750-1172-8-167.

71. Boy SPN, Haege G, Heringer J, et al. Low lysine diet in glutaric aciduria type I effect on anthropometric and biochemical follow-up parameters. J Inherit Metab Dis. 2013;36(3):525-533. doi:10.1007/s10545-012-9517-7. –

72. Numata-Uematsu Y, Sakamoto O, Kakisaka Y, et al. Reversible brain atrophy in glutaric aciduria type 1. Brain Dev. 2017;39(6):532-535. doi:10.1016/j.braindev.2017.01.003.

73. Lee C-S, Chien Y-H, Peng S-F, et al. Promising outcomes in glutaric aciduria type I patients detected by newborn screening. Metab Brain Dis. 2013;28(1):61-67. doi:10.1007/s11011-012-9349-z.

74. Kölker S, Koeller DM, Okun JG, Hoffmann GF. Pathomechanisms of neurodegeneration in glutaryl-CoA dehydrogenase deficiency. Ann Neurol. 2004;55(1):7-12. doi:10.1002/ana.10784.

75. Kettleborough RNW, Busch-Nentwich EM, Harvey SA, et al. A systematic genome- wide analysis of zebrafish protein-coding gene function. Nature. 2013;496(7446):494-497. doi:10.1038/nature11992.

76. Yager LM, Garcia AF, Wunsch AM, Ferguson SM. The ins and outs of the striatum: role in drug addiction. Neuroscience. 2015;301:529-541. doi:10.1016/j.neuroscience.2015.06.033.

77. Ouimet CC, Greengard P. Distribution of DARPP-32 in the basal ganglia: an electron microscopic study. J Neurocytol. 1990;19(1):39-52. http://www.ncbi.nlm.nih.gov/pubmed/2191086. Accessed June 6, 2017.

102

78. Greenberg CR, Prasad AN, Dilling LA, et al. Outcome of the First 3-Years of a DNA- Based Neonatal Screening Program for Glutaric Acidemia Type 1 in Manitoba and Northwestern Ontario, Canada. Mol Genet Metab. 2002;75(1):70-78. doi:10.1006/mgme.2001.3270.

79. Soffer D, Amir N, Elpeleg ON, Gomori JM, Shalev RS, Gottschalk-Sabag S. Striatal degeneration and spongy myelinopathy in glutaric acidemia. J Neurol Sci. 1992;107(2):199-204. http://www.ncbi.nlm.nih.gov/pubmed/1564518. Accessed June 6, 2017.

80. Goodman SI, Norenberg MD, Shikes RH, Breslich DJ, Moe PG. Glutaric aciduria: biochemical and morphologic considerations. J Pediatr. 1977;90(5):746-750. http://www.ncbi.nlm.nih.gov/pubmed/856963. Accessed June 6, 2017.

81. Leibel RL, Shih VE, Goodman SI, et al. Glutaric acidemia: a metabolic disorder causing progressive choreoathetosis. Neurology. 1980;30(11):1163-1168. http://www.ncbi.nlm.nih.gov/pubmed/6775244. Accessed June 6, 2017.

82. Kemp JM, Powell TP. The structure of the caudate nucleus of the cat: light and electron microscopy. Philos Trans R Soc Lond B Biol Sci. 1971;262(845):383-401. http://www.ncbi.nlm.nih.gov/pubmed/4107495. Accessed June 6, 2017.

83. Krst the putamen in the adult human neostriatum: a revised classification supported by a onošićqualitative B, Milošević and quantitative NT, Gudović analysis. R, Marić Anat DL, Sci Ristanović Int. 2012;87(3):115-125. D. Neuronal images of doi:10.1007/s12565-012-0131-4.

84. Wu TD, Reeder J, Lawrence M, Becker G, Brauer MJ. GMAP and GSNAP for Genomic Sequence Alignment: Enhancements to Speed, Accuracy, and Functionality. In: Methods in Molecular Biology (Clifton, N.J.). Vol 1418. ; 2016:283-334. doi:10.1007/978-1-4939-3578-9_15.

85. De Summa S, Malerba G, Pinto R, Mori A, Mijatovic V, Tommasi S. GATK hard filtering: tunable parameters to improve variant calling for next generation sequencing targeted gene panel data. BMC Bioinformatics. 2017;18(S5):119. doi:10.1186/s12859-017-1537-8.

86. Wang K, Li M, Hakonarson H. ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. 2010;38(16):e164- e164. doi:10.1093/nar/gkq603.

103

87. Liu X, Jian X, Boerwinkle E. dbNSFP v2.0: A Database of Human Non-synonymous SNVs and Their Functional Predictions and Annotations. Hum Mutat. 2013;34(9):E2393-E2402. doi:10.1002/humu.22376.

88. Kumar P, Henikoff S, Ng PC. Predicting the effects of coding non-synonymous variants on protein function using the SIFT algorithm. Nat Protoc. 2009;4(7):1073-1081. doi:10.1038/nprot.2009.86.

89. Chun S, Fay JC. Identification of deleterious mutations within three human genomes. Genome Res. 2009;19(9):1553-1561. doi:10.1101/gr.092619.109.

90. Schwarz JM, Cooper DN, Schuelke M, Seelow D. MutationTaster2: mutation prediction for the deep-sequencing age. Nat Methods. 2014;11(4):361-362. doi:10.1038/nmeth.2890.

91. Wilson RW, Wilson CM, Gates SC, Higgins J V. Alpha-ketoadipic aciduria: a description of a new metabolic error in lysine-tryptophan degradation. Pediatr Res. 1975;9(6):522-526. doi:10.1203/00006450-197506000-00002.

92. Lormans S, Lowenthal A. Alpha-amino adipic aciduria in an oligophrenic child. Clin Chim Acta. 1974;57(1):97-101. http://www.ncbi.nlm.nih.gov/pubmed/4430147. Accessed May 19, 2017.

93. Przyrembel H, Bachmann D, Lombeck I, et al. Alpha-ketoadipic aciduria, a new inborn error of lysine metabolism; biochemical studies. Clin Chim Acta. 1975;58(3):257-269. http://www.ncbi.nlm.nih.gov/pubmed/1112064. Accessed May 2, 2017.

94. Hagen J, te Brinke H, Wanders RJA, et al. Genetic basis of alpha-aminoadipic and alpha-ketoadipic aciduria. J Inherit Metab Dis. 2015;38(5):873-879. doi:10.1007/s10545-015-9841-9.

95. Casey RE, Zaleski WA, Philp M, Mendelson IS, MacKenzie SL. Biochemical and clinical studies of a new case of alpha-aminoadipic aciduria. J Inherit Metab Dis. 1978;1(4):129-135. http://www.ncbi.nlm.nih.gov/pubmed/117247. Accessed May 2, 2017.

104

96. -aminoadipic aciduria: chemical and enzymatic studies. J Inherit Metab Dis. 1980;2(4):89-92. http://www.ncbi.nlm.nih.gov/pubmed/6796766.Gray RG, ONeill EM, Pollitt RJ. Alpha Accessed May 28, 2017.

97. Takechi T, Okada T, Wakiguchi H, et al. Identification of N-acetyl-alpha- aminoadipic acid in the urine of a patient with alpha-aminoadipic and alpha- ketoadipic aciduria. J Inherit Metab Dis. 1993;16(1):119-126. http://www.ncbi.nlm.nih.gov/pubmed/8487492. Accessed May 21, 2017.

98. Danhauser K, Sauer SWW, Haack TBB, et al. DHTKD1 mutations cause 2- aminoadipic and 2-oxoadipic aciduria. Am J Hum Genet. 2012;91(6):1082-1087. doi:10.1016/j.ajhg.2012.10.006.

99. Stiles AR, Venturoni L, Mucci G, et al. New Cases of DHTKD1 Mutations in Patients with 2-Ketoadipic Aciduria. JIMD Rep. 2015;25:15-19. doi:10.1007/8904_2015_462.

100. Manders AJ, von Oostrom CG, Trijbels JM, Rutten FJ, Kleijer WJ. alpha- Aminoadipic aciduria and persistence of fetal haemoglobin in an oligophrenic child. Eur J Pediatr. 1981;136(1):51-55. http://www.ncbi.nlm.nih.gov/pubmed/6163632. Accessed June 2, 2017.

101. Fischer MH, Gerritsen T, Opitz JM. Alpha-aminoadipic aciduria, a non-deleterious inborn metabolic defect. Humangenetik. 1974;24(4):265-270. http://www.ncbi.nlm.nih.gov/pubmed/4442872. Accessed May 2, 2017.

102. Fischer MH, Brown RR. Tryptophan and lysine metabolism in alpha-aminoadipic aciduria. Am J Med Genet. 1980;5(1):35-41. doi:10.1002/ajmg.1320050106.

103. Goodman SI, Duran M. Biochemical Phenotypes of Questionable Clinical Significance. In: Physician’s Guide to the Diagnosis, Treatment, and Follow-Up of Inherited Metabolic Diseases. Berlin, Heidelberg: Springer Berlin Heidelberg; 2014:691-705. doi:10.1007/978-3-642-40337-8_44.

104. Duran M, Beemer FA, Wadman SK, Wendel U, Janssen B. A patient with alpha- ketoadipic and alpha-aminoadipic aciduria. J Inherit Metab Dis. 1984;7(2):61. http://www.ncbi.nlm.nih.gov/pubmed/6434826. Accessed May 2, 2017.

105

105. Saudubray J-M, Berghe G Van Den, Walter JH. Inborn Metabolic Diseases: Diagnosis and Treatment. 5th ed.; 2012.

106. Choi Y, Sims GE, Murphy S, Miller JR, Chan AP. Predicting the Functional Effect of Amino Acid Substitutions and Indels. de Brevern AG, ed. PLoS One. 2012;7(10):e46688. doi:10.1371/journal.pone.0046688.

107. Wilcken B, Smith A, Brown DA. Urine screening for aminoacidopathies: is it beneficial? Results of a long-term follow-up of cases detected bny screening one millon babies. J Pediatr. 1980;97(3):492-497. http://www.ncbi.nlm.nih.gov/pubmed/7411317. Accessed May 2, 2017.

108. Posset R, Opp S, Struys EA, et al. Understanding cerebral L-lysine metabolism: the role of L-pipecolate metabolism in Gcdh-deficient mice as a model for glutaric aciduria type I. J Inherit Metab Dis. 2015;38(2):265-272. doi:10.1007/s10545- 014-9762-z.

109. Hirashima M, Hayakawa T, Koike M. Mammalian alpha-keto acid dehydrogenase complexes. II. An improved procedure for the preparation of 2-oxoglutarate dehydrogenase complex from pig heart muscle. J Biol Chem. 1967;242(5):902- 907. http://www.ncbi.nlm.nih.gov/pubmed/6020442. Accessed May 2, 2017.

110. Vianey-Liaud C, Divry P, Cotte J, Teyssier G. alpha-Aminoadipic and alpha- ketoadipic aciduria: detection of a new case by a screening program using two- dimensional thin layer chromatography of amino acids. J Inherit Metab Dis. 1985;8 Suppl 2:133-134. http://www.ncbi.nlm.nih.gov/pubmed/3930865. Accessed May 2, 2017.

111. Xu W-Y, Zhu H-B, Gu M-M, et al. DHTKD1 is essential for mitochondrial biogenesis and function maintenance. FEBS Lett. 2013;587(21):3587-3592. doi:10.1016/j.febslet.2013.08.047.

112. Wu Y, Williams EG, Dubuis S, et al. Multilayered genetic and omics dissection of mitochondrial activity in a mouse reference population. Cell. 2014;158(6):1415- 1430. doi:10.1016/j.cell.2014.07.039.

113. Xu W-Y, Gu M-M, Sun L-H, et al. A nonsense mutation in DHTKD1 causes Charcot- Marie-Tooth disease type 2 in a large Chinese pedigree. Am J Hum Genet. 2012;91(6):1088-1094. doi:10.1016/j.ajhg.2012.09.018.

106

114. McMillan TA, Gibson KM, Sweetman L, Meyers GS, Green R. Conservation of central nervous system glutaryl-coenzyme A dehydrogenase in fruit-eating bats with glutaric aciduria and deficient hepatic glutaryl-coenzyme A dehydrogenase. J Biol Chem. 1988;263(33):17258-17261. http://www.ncbi.nlm.nih.gov/pubmed/3182847. Accessed May 2, 2017.

115. Koeller DM, Sauer SW, Wajner M, et al. Animal models for glutaryl-CoA dehydrogenase deficiency. J Inherit Metab Dis. 2004;27(6):813-818. doi:10.1023/B:BOLI.0000045763.52907.5e.

116. Olivera-Bravo S, Isasi E, Fernández A, et al. White matter injury induced by perinatal exposure to glutaric acid. Neurotox Res. 2014;25(4):381-391. doi:10.1007/s12640-013-9445-9.

117. Olivera S, Fernandez A, Latini A, et al. Astrocytic proliferation and mitochondrial dysfunction induced by accumulated glutaric acidemia I (GAI) metabolites: possible implications for GAI pathogenesis. Neurobiol Dis. 2008;32(3):528-534. doi:10.1016/j.nbd.2008.09.011.

118. Isasi E, Barbeito L, Olivera-Bravo S. Increased blood-brain barrier permeability and alterations in perivascular astrocytes and pericytes induced by intracisternal glutaric acid. Fluids Barriers CNS. 2014;11(1):15. doi:10.1186/2045-8118-11-15.

119. de Mello CF, Kölker S, Ahlemeyer B, et al. Intrastriatal administration of 3- hydroxyglutaric acid induces convulsions and striatal lesions in rats. Brain Res. 2001;916(1-2):70-75. http://www.ncbi.nlm.nih.gov/pubmed/11597592. Accessed May 2, 2017.

120. Bjugstad KB, Crnic LS, Goodman SI, Freed CR. Infant mice with glutaric acidaemia type I have increased vulnerability to 3-nitropropionic acid toxicity. J Inherit Metab Dis. 2006;29(5):612-619. doi:10.1007/s10545-006-0102-9.

121. Bjugstad KB, Goodman SI, Freed CR. Age at symptom onset predicts severity of motor impairment and clinical outcome of glutaric acidemia type 1. J Pediatr. 2000;137(5):681-686. doi:10.1067/mpd.2000.108954.

107

122. Sauer SW, Okun JG, Fricker G, et al. Intracerebral accumulation of glutaric and 3- hydroxyglutaric acids secondary to limited flux across the blood-brain barrier constitute a biochemical risk factor for neurodegeneration in glutaryl-CoA dehydrogenase deficiency. J Neurochem. 2006;97(3):899-910. doi:10.1111/j.1471-4159.2006.03813.x.

123. Wager K, Mahmood F, Russell C. Modelling inborn errors of metabolism in zebrafish. J Inherit Metab Dis. 2014;37(4):483-495. doi:10.1007/s10545-014- 9696-5.

124. Seth A, Stemple DL, Barroso I. The emerging use of zebrafish to model metabolic disease. Dis Model Mech. 2013;6(5):1080-1088. doi:10.1242/dmm.011346.

125. Huang P, Zhu Z, Lin S, Zhang B. Reverse Genetic Approaches in Zebrafish. J Genet Genomics. 2012;39(9):421-433. doi:10.1016/j.jgg.2012.07.004.

126. Mahmood F, Fu S, Cooke J, Wilson SW, Cooper JD, Russell C. A zebrafish model of CLN2 disease is deficient in tripeptidyl peptidase 1 and displays progressive neurodegeneration accompanied by a reduction in proliferation. Brain. 2013;136(Pt 5):1488-1507. doi:10.1093/brain/awt043.

127. Bond M, Holthaus S-MK, Tammen I, Tear G, Russell C. Use of model organisms for the study of neuronal ceroid lipofuscinosis. Biochim Biophys Acta. 2013;1832(11):1842-1865. doi:10.1016/j.bbadis.2013.01.009.

128. Follo C, Ozzano M, Mugoni V, Castino R, Santoro M, Isidoro C. Knock-down of cathepsin D affects the retinal pigment epithelium, impairs swim-bladder ontogenesis and causes premature death in zebrafish. Escriva H, ed. PLoS One. 2011;6(7):e21908. doi:10.1371/journal.pone.0021908.

129. Chitramuthu BP, Baranowski DC, Kay DG, Bateman A, Bennett HP. Progranulin modulates zebrafish motoneuron development in vivo and rescues truncation defects associated with knockdown of Survival motor neuron 1. Mol Neurodegener. 2010;5(1):41. doi:10.1186/1750-1326-5-41.

130. Mendelsohn BA, Yin C, Johnson SL, Wilm TP, Solnica-Krezel L, Gitlin JD. Atp7a determines a hierarchy of copper metabolism essential for notochord development. Cell Metab. 2006;4(2):155-162. doi:10.1016/j.cmet.2006.05.001.

108

131. Madsen EC, Gitlin JD. Zebrafish Mutants calamity and catastrophe Define Critical Pathways of Gene Nutrient Interactions in Developmental Copper Metabolism. Mullins M, ed. PLoS Genet. 2008;4(11):e1000261. doi:10.1371/journal.pgen.1000261.–

132. Madsen EC, Morcos PA, Mendelsohn BA, Gitlin JD. In vivo correction of a Menkes disease model using antisense oligonucleotides. Proc Natl Acad Sci. 2008;105(10):3909-3914. doi:10.1073/pnas.0710865105.

133. Savio LEB, Vuaden FC, Rosemberg DB, Bogo MR, Bonan CD, Wyse ATS. Long-term proline exposure alters nucleotide catabolism and ectonucleotidase gene expression in zebrafish brain. Metab Brain Dis. 2012;27(4):541-549. doi:10.1007/s11011-012-9321-y.

134. Savio LEB, Vuaden FC, Piato AL, Bonan CD, Wyse ATS. Behavioral changes induced by long-term proline exposure are reversed by antipsychotics in zebrafish. Prog Neuro-Psychopharmacology Biol Psychiatry. 2012;36(2):258-263. doi:10.1016/j.pnpbp.2011.10.002.

135. Maurer CM, Schönthaler HB, Mueller KP, Neuhauss SCF. Distinct retinal deficits in a zebrafish pyruvate dehydrogenase-deficient mutant. J Neurosci. 2010;30(36):11962-11972. doi:10.1523/JNEUROSCI.2848-10.2010.

136. Friedrich T, Lambert AM, Masino MA, Downes GB. Mutation of zebrafish dihydrolipoamide branched-chain transacylase E2 results in motor dysfunction and models maple syrup urine disease. Dis Model Mech. 2012;5(2):248-258. doi:10.1242/dmm.008383.

137. Jeong J-Y, Kwon H-B, Ahn J-C, et al. Functional and developmental analysis of the blood brain barrier in zebrafish. Brain Res Bull. 2008;75(5):619-628. doi:10.1016/j.brainresbull.2007.10.043. –

138. Xi Y, Noble S, Ekker M. Modeling Neurodegeneration in Zebrafish. Curr Neurol Neurosci Rep. 2011;11(3):274-282. doi:10.1007/s11910-011-0182-2.

139. Kalueff A V, Gebhardt M, Stewart AM, et al. Towards a comprehensive catalog of zebrafish behavior 1.0 and beyond. Zebrafish. 2013;10(1):70-86. doi:10.1089/zeb.2012.0861.

109

140. Roberts AC, Reichl J, Song MY, et al. Habituation of the C-start response in larval zebrafish exhibits several distinct phases and sensitivity to NMDA receptor blockade. Tanimoto H, ed. PLoS One. 2011;6(12):e29132. doi:10.1371/journal.pone.0029132.

141. Roberts AC, Bill BR, Glanzman DL. Learning and memory in zebrafish larvae. Front Neural Circuits. 2013;7:126. doi:10.3389/fncir.2013.00126.

142. Goodman SI, Duran M. Biochemical Phenotypes of Questionable Clinical Significance. In: Physician’s Guide to the Diagnosis, Treatment, and Follow-Up of Inherited Metabolic Diseases. ; 2014:691-704.

110

APPENDIX A:

DARPP-32 STAINING IN THE PUTAMEN OF REPRESETITIVE CONTROLS AND GA1

BRAINS

7-Month Old Control and GA1 Putamen. Images are presented in black and white. A. Dorsal putamen along the external white matter tract in a control sample. Arrows point to all positively stained neurons. B. Dorsal putamen along the external white matter tract in a GA1 sample. Arrows point to some positively stained neurons.

111

18-Month Old Control and GA1 Putamen. Images are presented in black and white A. Medial dorsal putamen along the external white matter tract in a control sample. Arrows point to all positively stained neurons. B. Medial putamen along the external white matter tract in a GA1 sample. Arrows point to all positively stained neurons.

112

5-Year Old Control and GA1 Putamen. Images are presented in black and white A. Medial dorsal putamen along the external white matter tract in a control sample. Arrows point to some positively stained neurons. B. Dorsal putamen along the external white matter tract in a GA1 sample. Arrows point to some positively stained neurons.

113

12-Year Old Control and GA1 Putamen. Images are presented in black and white A. Ventral putamen along the external white matter tract in a control sample. Arrows point to all positively stained neurons. B. Ventral putamen along the external white matter tract in a GA1 sample. Arrows point to some positively stained neurons.

114

APPENDIX B:

TABLE OF PUBLISHED DESCRIPTIONS OF 2-KETOADIPIC AND 2-AMINOADIPIC

ACIDEMIA PATIENTS

Patient Clinical Symptoms DNA DNA Metabolite Ref Sequencing Sequencing Levels Allele 1 Allele 2

1 Learning Defect not done not done n/a 101,102

2 Mentally Normal not done not done n/a 101,102

3 Mentally retarded, deeply not done not done Elevated serum 92 oligophrenic, hypotonic, dysphagia 2-AA

4 Psychomotorical retardation, c.2134C>T c.2134C>T Elevated urine 93,94 subnormal mental development (p.R712X) (p.R712X) 2-AA and 2-KA

5 Mentally retarded not done not done Elevated urine 91 2-AA and 2-KA, elevated serum 2-AA

6 Normal not done not done Elevated urine 91 2-AA and 2-KA, elevated serum 2-AA

7 Mentally retarded, with obesity, not done not done Elevated serum 95 hypotonia, clumsiness and mild 2-AA, no increased ocular abnormalities 2-KA detected in urine

8 Mentally retarded, dysmorphic c.1364G>A not found Elevated serum 94,100 features, Raynaud's phenomenon, (p.R455Q) and urine 2-AA, no hypotonia, petit mal seizures, high elevated 2-KA levels of fetal haemoglobin

9 Immunodeficiency not done not done n/a 96

10 Psychomotor retardation, not done not done Elevated urine 104 abnormal e.e.g., seizures, chronic 2-AA, 2-KA, and upper respiratory infections and 2-HA. Elevated recurrent pneumonia plasma 2-AA after lysine load

11 Mild speech retardation not done not done Elevated urine and 97 serum 2-AA, Elevated urine 2-KA

12 Moderately delayed psychomotor c. 1A>G c.2185G>A Elevated urine 98 development (p.Met1?) (p.Gly729Arg) 2-KA and elevated serum 2-AA

115

Patient Clinical Symptoms DNA DNA Metabolite Ref Sequencing Sequencing Levels Allele 1 Allele 2

13 Microcephaly, mild motor c.1228C>T c.2185G>A Elevated 2-AA and 98 developmental delay, predominant (p.Arg410*) (p.Gly729Arg) 2-KA in urine and speech delay and muscular serum hypotonia

14 Failure to thrive, seizure-like c.2143C>T c.2185G>A Elevated urine 99 episodes, chronic (p.Arg715Cys) (p.Gly729Arg) 2-KA and 2-HA headache/migraine, developmental delay

15-17 Family with 3 affected siblings. All c.915G>C c.915G>C Increased 99 are clinically normal (p.Gln305His) (p.Gln305His) glutamate; elevated urine 2-AA and 2-KA

18 psychomotor retardation and c.2185G>A c.86dup Elevated urine 94 convulsions with EEG (p.G729R) (p.Y63IfsX3) 2-AA, 2-KA abnormalities Elevated serum 2-AA

19 intellectual disability, autism, c.2185G>A c. 2318C>T Elevated urine 94 severe obstipation and epileptic (p.G729R) (p.P773L) 2-AA, 2-KA seizures, 22q11.2 deletion

20 Partial biotinidase deficiency c.2185G>A c.1309G>T Elevated urine 94 (p.G729R) (p.E437X) 2-AA and 2-KA

21 Moderate developmental delay and c.2185G>A not found Elevated urine 94 epileptic features characterized as (p.G729R) 2-AA, 2-KA myoclonic jerks. Elevated serum 2-AA

22 Severe intellectual disability and c.1364G>A c.2329T>C "Strongly elevated 94 behavior disinhibition (p.R455Q) (p.S777P) 2-AA and 2-KA"

23 Delayed gross motor milestones, c.1364G>A c.1159+5G>A Elevated urine 94 mild mental retardation, obsessive (p.R455Q) 2-AA, 2-KA behavior, delayed speech, sleep Elevated serum abnormalities and mild 2-AA dysmorphic features

24 Hirschsprung disease, cerebellar c.1671+1G>A c.1671+1G>A Elevated urine 94 hypoplasia, hexadactylism, 2-AA and 2-KA dysgnathia, psychomotor impairment, behavior problems

25 Hepatomegaly and prolonged c.700_701 c.700_701 Elevated urine 94 icterus, progressive developmental delinsGG delinsGG 2-AA and 2-KA delay, Niemann-Pick type C (p.L234G) (p.L234G) following a lysine challenge

26-33 Family with Autosomal-Dominant c.1455T>G n/a 113 Charcot-Marie-Tooth Type 2 (p.Tyr485*)

116