Lung Defects Contribute to Respiratory Symptoms in a Mecp2-Mutant Mouse Model of Rett Syndrome

Neeti Vashi

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

Lung Defects Contribute to Respiratory Symptoms in a Mecp2- Mutant Mouse Model of Rett Syndrome

Neeti Vashi

Doctor of Philosophy

Molecular Genetics University of Toronto

2021

Abstract

Rett syndrome (RTT) is a progressive neuro-metabolic disorder caused by mutations in the X- linked gene, methyl-CpG-binding protein 2 (MECP2). After a period of seemingly normal postnatal development, RTT patients experience a developmental regression, consisting of loss of acquired verbal and motor skills, stereotypic hand movements, respiratory abnormalities, and seizures. Respiratory impairment causes up to 80% of premature patient death; despite this, lung pathology in RTT is understudied and respiratory symptoms are currently attributed to neuronal loss of MECP2. To study the Mecp2-deficient lung, we utilized a Mecp2-mutant mouse model that recapitulates many features of RTT. I found striking lipid metabolism abnormalities in the lungs of

Mecp2-mutant mice, including increased cholesterol and triglycerides and decreased phosphatidylcholines. My single cell RNA-sequencing and chromatin immunoprecipitation experiments showed that lipogenesis is increased due to decreased binding of the nuclear repressor coreceptor 1/2 (NCOR1/2) complex in the promoters of its target genes in the absence of MECP2, leading to their upregulation. I also showed that lung AE2 cell-specific depletion of

Mecp2 is sufficient to cause lung lipid metabolism abnormalities and respiratory symptoms. In contrast, hindbrain neuron-specific deletion of Mecp2, which removes Mecp2 from the neuronal respiratory control center, imparted a different respiratory phenotype. RNA-sequencing of the

Mecp2-deficient lung revealed decreased expression of key extracellular matrix (ECM) genes; consistently, I found alveolar tissue degradation and bronchiolar enlargement in Mecp2-mutant ii mice. Consistent with these findings, Mecp2-mutant mice have altered pulmonary function.

Finally, we treated whole body metabolism in Mecp2-mutant mice using lipid-modulating compounds, including statins and liver X receptor (LXR) agonists; both improved neurological and respiratory symptoms, suggesting clinical utility. Altogether, these findings implicate key functions of Mecp2 in the lung and highlight the importance of studying non-neuronal aspects of RTT. Our findings will aid in developing treatments and clinical recommendations for RTT patients.

iii

Acknowledgments

I would first like to express my sincere gratitude to my advisor, Dr. Monica Justice. Your contagious enthusiasm, accurate instincts, and thirst for scientific knowledge continue to amaze me. You have always pushed me outside of my comfort zone, and I have grown tremendously as a scientist, and as a person, because of it. it. Thank you for providing me with numerous opportunities that I will always cherish. I am lucky to have been mentored by you and I will always be grateful that you saw my potential.

To my supervisory committee, Dr. Lucy Osborne and Dr. Martin Post: I genuinely enjoyed my committee meetings because of your thoughtful guidance, support, and discussions. Thank you to our collaborators, Dr. Cameron Ackerley, Dr. Pradip Saha, Dr. Gillian Sleep, and individuals at the Center for Phenogenomics, The Center for Applied Genomics, the Analytical Facility for Bioactive Molecules, and the Princess Margaret Genomics Centre, without whom this work would not have been possible.

I have been extremely fortunate to have worked alongside many talented individuals in an incredibly supportive environment. SMK, thank you for your patient mentorship during my first two years of graduate school. To the current members of the Justice lab, correction, family (AE, CT, JR, LH, LT, RT, ZK): thank you for the stimulating scientific discussions and equally stimulating distractions. Each of you has contributed so much to my journey, both scientific and personal, and I am forever grateful. Thank you to the drug study team, JR and CT; I’ll cherish all the hours we spent in fume hoods together. ZK, thank you for always being so willing to help me out, especially during the pandemic. JR, you have been essential to my success – this PhD would have taken an extra three years without you!

I would not have been able to accomplish this without the encouragement of my family. To my parents, I will always be immensely grateful for the sacrifices you made in order for me to have the best educational opportunities. Thank you for being my biggest supporters – I hope I always make you proud. To my brother, PV, thank you for being someone I will always look up to, and to my sister-in-law, RV, for always providing a positive outlook. I am extremely grateful to my friends for endless encouragement and for getting excited about science. Finally, to my incredible partner, KJ, who has listened to me practice my presentations so many times, he probably has my project memorized. Thank you for always encouraging me on the bad days and celebrating all the good ones; this would not have been possible without you.

Table of Contents

Acknowledgments ...... iv Chapter 1 Introduction and Background ...... 1 Introduction and Background ...... 2 1.1 A brief description of Rett syndrome ...... 2 1.1.1 Rett syndrome diagnosis and stages ...... 2 1.1.2 Stage-independent features of Rett syndrome ...... 6 1.1.3 Classic RTT is caused by mutations in MECP2 ...... 8 1.1.4 Atypical Rett syndrome ...... 8 1.1.5 The MECP2 gene ...... 8 1.1.6 The MECP2 protein and its isoforms ...... 10 1.1.7 MECP2 domains ...... 12 1.1.8 RTT-causing MECP2 mutations ...... 15 1.1.9 Phenotypic variation in RTT ...... 17 1.1.10 MECP2 expression and localization ...... 19 1.1.11 MECP2 function ...... 19 1.1.12 Males with MECP2 mutations ...... 25 1.2 Studying MECP2 using animal models ...... 27 1.2.1 Mecp2-null mouse models...... 27 1.2.2 Male vs. female Mecp2-mutant mice ...... 31 1.2.3 Similarities between Mecp2-mutant mice and RTT patients ...... 31 1.2.4 Mecp2-mutant mice with human RTT-causing mutations ...... 33 1.2.5 Temporal deletions of Mecp2 ...... 33 1.2.6 Symptom reversal and suppression in Mecp2-mutant mice ...... 34 1.3 Regulation of breathing ...... 44 1.3.1 Autonomic control of respiratory rhythm ...... 46 1.3.2 Lung development ...... 49 1.3.3 Cellular composition of the lung...... 51 1.3.4 Pulmonary surfactant ...... 54 1.3.5 The lung’s extracellular matrix ...... 57 1.3.6 Respiratory disorders ...... 57 1.3.7 Breathing abnormalities in RTT ...... 61 1.4 Hypothesis ...... 63 Chapter 2 Lung lipid defects contribute to respiratory symptoms in a Mecp2-mutant mouse model of Rett syndrome ...... 64

Lung lipid defects contribute to respiratory symptoms in a Mecp2-mutant mouse model of Rett syndrome ...... 65 2.1 Abstract ...... 65 2.2 Introduction ...... 66 2.3 Methods ...... 68 2.3.1 Animals ...... 68 2.3.2 Electron microscopy and immunohistochemistry ...... 68 2.3.3 Lipid quantification ...... 69 2.3.4 Lung single cell isolation and flow cytometry ...... 70 2.3.5 Single cell RNA-sequencing ...... 70 2.3.6 RNA extraction and quantitative reverse transcription polymerase chain reaction (RT-qPCR)...... 71 2.3.7 Native protein extraction and Western blotting ...... 71 2.3.8 Nuclear protein extraction and immunoprecipitation ...... 73 2.3.9 Chromatin immunoprecipitation ...... 73 2.3.10 Subjective health assessments ...... 74 2.3.11 Rotarod ...... 74 2.3.12 Open field activity ...... 74 2.3.13 Social behavior ...... 77 2.3.14 Plethysmography ...... 77 2.3.15 Statistics ...... 77 2.4 Results ...... 78 2.4.1 Loss of Mecp2 results in abnormal lipid production in the Mecp2/Y mouse lung ...... 78 2.4.2 Heterozygous deletion of Mecp2 is sufficient to cause lung lipid abnormalities in mice...... 81 2.4.3 Single cell RNA-sequencing of mouse AE2 cells reveals altered metabolic gene expression in Mecp2/Y mice ...... 83 2.4.4 Lipid metabolism is altered in Mecp2/Y AE2 cells ...... 85 2.4.5 Loss of Mecp2 from lung AE2 cells alters mitochondrial gene expression ...... 91 2.4.6 MECP2 regulates the expression of lung lipid metabolism enzymes through interaction with the NCOR1/2 co-repressor complex ...... 93 2.4.7 Comparison of mice with an AE2 cell- and hindbrain neuron-specific Mecp2 deletion ...... 97 2.4.8 Hindbrain neuron-specific deletion of Mecp2 imparts neurobehavioral changes that are absent in mice with AE2 cell-specific Mecp2 deficiency ...... 101 2.4.9 Distinct respiratory symptoms in mice with AE2 cell- or hindbrain neuron- specific deficiency of Mecp2 ...... 106 2.5 Discussion ...... 111 vi

2.5.1 Lung lipid abnormalities in Mecp2/Y lungs ...... 111 2.5.2 MECP2, in concert with the NCOR1/2 corepressor complex, directs lung lipid metabolism ...... 113 2.5.3 AE2-cell specific deletion of Mecp2 in mice is sufficient to cause lung lipid abnormalities and respiratory symptoms ...... 114 2.5.4 Mitochondrial genes are impacted in Mecp2/Y AE2 cells ...... 115 2.5.5 Implications for RTT ...... 116 Chapter 3 Structural and functional changes in the lung of a Mecp2-mutant mouse model of Rett syndrome ...... 117 Structural and functional changes in the lung of a Mecp2-mutant mouse model of Rett syndrome ...... 118 3.1 Abstract ...... 118 3.2 Introduction ...... 119 3.3 Methods ...... 121 3.3.1 Animals ...... 121 3.3.2 Plethysmography ...... 121 3.3.3 Immunohistochemistry and histology ...... 121 3.3.4 RNA extraction ...... 122 3.3.5 RNA sequencing and analysis ...... 122 3.3.6 Quantitative reverse transcription polymerase chain reaction (RT-qPCR) ...... 124 3.3.7 Protein extraction and Western blotting ...... 124 3.3.8 Lung function tests ...... 126 3.3.9 Statistics ...... 126 3.4 Results ...... 127 3.4.1 Mecp2-mutant mice have breathing symptoms prior to neurological symptom onset ...... 127 3.4.2 MECP2 is highly expressed in the mouse lung ...... 129 3.4.3 Transcriptional changes in the Mecp2-deficient lung ...... 131 3.4.4 Circadian rhythm is transcriptionally altered in Mecp2/Y lungs ...... 134 3.4.5 Extracellular matrix (ECM) genes are expressed at low levels in Mecp2/Y lungs ...... 137 3.4.6 Alveolar structure is altered in Mecp2/Y mice ...... 141 3.4.7 Bronchiolar enlargement in Mecp2-mutant lungs ...... 143 3.4.8 Pulmonary function is altered in Mecp2-mutant mice ...... 145 3.4.9 AE2 cell-specific deletion of Mecp2 alters pulmonary structure and function ... 147 3.5 Discussion ...... 149 3.5.1 Circadian rhythm is altered in Mecp2/Y lungs ...... 149 3.5.2 Extracellular matrix genes are transcriptionally decreased in Mecp2/Y lungs .. 150 vii

3.5.3 Structural changes in the Mecp2-mutant lung...... 151 3.5.4 Lung function is altered upon loss of Mecp2 ...... 152 3.5.5 Implications for RTT patients ...... 153 Chapter 4 Pharmacological treatment of lipid metabolism perturbations in Mecp2-mutant mice ...... 155 Pharmacological treatment of lipid metabolism in Mecp2-mutant mice ...... 156 4.1 Abstract ...... 156 4.2 Introduction ...... 157 4.3 Methods ...... 160 4.3.1 Animals ...... 160 4.3.2 RNA-sequencing ...... 160 4.3.3 Drug administration ...... 161 4.3.4 Plethysmography ...... 162 4.3.5 Necropsy ...... 162 4.3.6 Lipid quantification ...... 163 4.3.7 Statistics ...... 163 4.4 Results ...... 164 4.4.1 RNA-sequencing of the pre-symptomatic Mecp2/Y brain ...... 164 4.4.2 Lipid biosynthesis and transport misregulation at the transcriptional level in the pre-symptomatic Mecp2/Y brain ...... 164 4.4.3 Treatment with cholesterol-lowering fluvastatin improves health, motor, and respiratory symptoms in male Mecp2/Y mice ...... 167 4.4.4 Treatment with fluvastatin improves motor and respiratory symptoms and lung lipid metabolism in female Mecp2/+ mice ...... 170 4.4.5 Treatment with the LXR-agonist T0901317 improves neurological symptoms but worsens systemic metabolism ...... 173 4.4.6 Treatment with the LXR-agonist LXR-623 improves symptoms and systemic lipid metabolism in Mecp2/Y mice ...... 176 4.4.7 Treatment with the LXR-agonist LXR-623 improves motor coordination and respiratory symptoms in female Mecp2/+ mice ...... 180 4.4.8 Treatment with fluvastatin and LXR-623 fails to improve symptoms in male Mecp2/Y mice ...... 183 4.5 Discussion ...... 186 4.5.1 Brain cholesterol metabolism ...... 186 4.5.2 Cholesterol metabolism is transcriptionally altered in the brain of Mecp2/Y mice ...... 187 4.5.3 Pharmacological targeting of cholesterol biosynthesis in Mecp2/Y mice...... 188 4.5.4 Activation of cholesterol transport through LXR-agonists ...... 189 4.5.5 Treating cholesterol metabolism in RTT patients ...... 190 viii

Chapter 5 Summary and Future Directions ...... 192 Summary and Future Directions ...... 193 5.1 Summary and Significance ...... 193 5.1.1 Aberrant lung lipid metabolism in Mecp2-mutant mice ...... 194 5.1.2 MECP2 regulates transcription of lung lipid metabolism genes with the NCOR1/2 co-repressor complex...... 194 5.1.3 Lung-specific deletion of Mecp2 impairs lung lipid metabolism and causes respiratory symptoms ...... 195 5.1.4 RNA-sequencing of Mecp2/Y lungs reveals an altered transcriptome ...... 196 5.1.5 Lung structure and function are altered in Mecp2-mutant mice ...... 196 5.1.6 Cholesterol metabolism is altered in the brain ...... 197 5.1.7 Lipid metabolism modulators improve symptoms in Mecp2-mutant mice ...... 198 5.1.8 Additional implications for RTT ...... 198 5.2 Future Directions ...... 200 5.2.1 Are lung abnormalities in Mecp2-mutant mice translatable to human RTT patients? ...... 200 5.2.2 How does SREBP1 affect metabolic gene transcription in Mecp2/Y AE2 cells? ...... 200 5.2.3 Do circadian-expressed nuclear receptors have a role in RTT pathology? ...... 203 5.2.4 Are mitochondria impacted by Mecp2 deficiency? ...... 207 5.2.5 Is the Mecp2-deficient lung more susceptible to insults? ...... 208 5.2.6 How is brain lipid metabolism altered in Mecp2/Y mice? ...... 209 5.2.7 Preclinical treatment of RTT symptoms ...... 213 5.3 Future of RTT ...... 214 Bibliography ...... 217

List of Tables Table 1.1: Diagnostic criteria for RTT...... 3 Table 1.2: Atypical Rett syndrome variants...... 9 Table 1.3: Males with MECP2 mutations fall into four categories...... 26 Table 1.4: Mouse models created to study Rett syndrome...... 29 Table 1.5: Preclinical treatments targeting pathways downstream of Mecp2 mutation...... 37 Table 1.6: Comparison between restrictive and obstructive lung disease...... 60 Table 2.1: Primers used for RT-qPCR analysis...... 72 Table 2.2: Primers used for ChIP-qPCR analysis...... 75 Table 2.3: Subjective health score definitions ...... 76 Table 2.4: Lipid metabolism genes are misregulated in Mecp2/Y AE2 cells...... 87 Table 2.5: Mecp2 deficiency alters mitochondrial gene expression in lung AE2 cells...... 92 Table 3.1: Primers used for RT-qPCR...... 125 Table 3.2: ECM genes are altered in Mecp2/Y lungs...... 139

List of Figures Figure 1.1: Timeline of stages in RTT patients...... 5 Figure 1.2: RTT is a multi-systemic disorder...... 7 Figure 1.3: MECP2 is expressed in two isoforms...... 11 Figure 1.4: Functional domains and post-translational modifications of MECP2...... 14 Figure 1.5: RTT-causing MECP2 mutations...... 16 Figure 1.6: RTT symptom severity is influenced by mutation status, XCI pattern, and modifier genes...... 18 Figure 1.7: Possible functions of MECP2...... 20 Figure 1.8: The ‘bridge hypothesis’ of MECP2 function...... 23 Figure 1.9: Animal models used to study MECP2...... 28 Figure 1.10: Similarities between human RTT patients and Mecp2-mutant mice...... 32 Figure 1.11: Preclinical treatments for RTT...... 35 Figure 1.12: Treatment strategies aimed at targeting MECP2 directly...... 43 Figure 1.13: Overview of respiration...... 45 Figure 1.14: Neuronal respiratory centers control breathing...... 48 Figure 1.15: Lung development occurs over five stages...... 50 Figure 1.16: Cell types in the respiratory system...... 53 Figure 1.17: Lung surfactant is essential for breathing...... 56 Figure 2.1: Lipids are altered in the lung and BAL fluid of male Mecp2/Y mice...... 80 Figure 2.2: Lipids are altered in the lung and BAL fluid of female Mecp2/+ mice...... 82 Figure 2.3: Single cell RNA-sequencing of AE2 cells reveals transcriptional metabolic changes in the absence of Mecp2...... 84 Figure 2.4: Metabolic gene expression is altered in Mecp2/Y AE2 cells...... 88 Figure 2.5: Lipid metabolism enzyme expression in +/Y and Mecp2/Y lungs...... 89 Figure 2.6: MECP2 regulates lipogenic gene transcription with NCOR1/2...... 96 Figure 2.7: Conditional removal of Mecp2 in lung AE2 cells or hindbrain neurons using Cre-Lox technology imparts a partial deletion in targeted cells/tissues...... 99 Figure 2.8: Mecp2 expression in conditional deletion littermate controls developed through Cre- Lox breeding strategy...... 100 Figure 2.9: Neurobehavioral changes are present in male mice with a whole-body or hindbrain- specific deletion of Mecp2...... 103 Figure 2.10: Neurobehavioral assessments in Sftpc-CreERT2;Mecp2-flx/Y mice and their littermate controls...... 104 Figure 2.11: Neurobehavioral assessments in Atoh1-Cre;Mecp2-flx/Y mice and their littermate controls...... 105 Figure 2.12: Respiratory features and lung lipids are altered in mice with whole-body and AE2 cell-specific deletions of Mecp2...... 108

Figure 2.13: Respiratory features and lipid quantification assessments in Sftpc- CreERT2;Mecp2-flx compared to their littermate controls...... 109 Figure 2.14: Respiratory features and lipid quantification assessments in Atoh1-Cre;Mecp2-flx compared to their littermate controls...... 110 Figure 3.1: Mecp2-mutant mice develop breathing symptoms prior to neurological symptom onset...... 128 Figure 3.2: MECP2 is expressed in various cell types in the mouse lung...... 130 Figure 3.3: Whole lung RNA-sequencing reveals transcriptomic changes in Mecp2/Y lungs. . 133 Figure 3.4: Core circadian rhythm components are misregulated in Mecp2/Y lungs...... 136 Figure 3.5: Altered expression of structural genes in Mecp2/Y lungs...... 140 Figure 3.6: Emphysema-like changes in adult Mecp2-mutant lungs...... 142 Figure 3.7: Enlarged bronchioles in adult Mecp2-mutant lungs...... 144 Figure 3.8: Lung function is altered in adult Mecp2-mutant mice...... 146 Figure 3.9: Altered lung function in mice with an AE2 cell-specific deletion of Mecp2...... 148 Figure 4.1: Lipid metabolism is misregulated in the pre-symptomatic Mecp2/Y brain...... 166 Figure 4.2: Fluvastatin improves overall health, motor coordination, and respiratory symptoms in male Mecp2/Y mice while normalizing lipid parameters...... 169 Figure 4.3: Fluvastatin treatment improves overall health, motor and respiratory symptoms in female Mecp2/+ mice while normalizing lipid parameters ...... 172 Figure 4.4: Treatment with the LXR-agonist T0901317 improves neurological phenotypes but worsens systemic metabolism in male Mecp2/Y mice...... 175 Figure 4.5: Treatment with the LXR-agonist LXR-623 improves health, motor coordination, and breathing frequency in male Mecp2/Y mice...... 178 Figure 4.6: Treatment with the LXR-agonist LXR-623 lowers serum, liver and lung triglycerides and serum cholesterol in male Mecp2/Y mice...... 179 Figure 4.7: Treatment with the LXR-agonist LXR-623 improves motor coordination and breathing frequency in female Mecp2/+ mice...... 181 Figure 4.8: Treatment with the LXR-agonist LXR-623 increases liver cholesterol and decreases lung triglycerides in female Mecp2/+ mice...... 182 Figure 4.9: Treatment with fluvastatin and LXR-623 has no therapeutic benefit on symptoms in male Mecp2/Y mice...... 184 Figure 4.10: Treatment with fluvastatin and LXR-623 imparts no benefits on lipid metabolism in male Mecp2/Y mice...... 185 Figure 5.1: SREBPs are lipid-sensing transcription factors...... 202 Figure 5.2: The ‘sponge hypothesis’ of NCOR1/2 for NR1D1...... 206 Figure 5.3: Hypothesis on neuron and astrocyte cholesterol metabolism in Mecp2-deficient brain cells...... 212 Figure 5.4: Precision medicine for treating RTT...... 216

xii 1

Chapter 1 Introduction and Background

Portions of this chapter have been published in:

Vashi, N., & Justice, M. J. (2019). Treating Rett syndrome: from mouse models to human therapies. Mammalian Genome, 30(5-6), 90-110.

and

Kyle, S. M., Vashi, N., & Justice, M. J. (2018). Rett syndrome: a neurological disorder with metabolic components. Open Biology, 8(2), 170216.

2 Introduction and Background

1.1 A brief description of Rett syndrome

Rett syndrome (RTT, OMIM #312750) was first described by Andreas Rett, an Austrian pediatric neurologist, after observing two female patients with identical hand-wringing stereotypies in his clinic waiting room. Upon examination, he found that both patients had the same history: normal early development, followed by a period of regression and loss of purposeful hand movements. Intrigued, Dr. Rett documented thirty other female patients in his clinic, all under thirteen years of age, with similar symptoms as well as high blood ammonia levels, abnormal muscle tone, apraxia, and a high incidence of epilepsy. Believing the symptoms to be consistent with a metabolic disorder, he called it ‘cerebroatrophic hyperammonemia’ in a 1966 German publication (1). However, the disorder did not gain general acceptance among the medical community until its description in English by Dr. Bengt Hagberg, 17 years later (Hagberg et al., 1983). Further studies have classified RTT as a progressive neurological disorder that primarily affects girls, occurring in 1:10,000-15,000 live female births (3).

1.1.1 Rett syndrome diagnosis and stages

RTT is one of the most common causes of severe disability in females. The clinical diagnosis of classic RTT is based on a battery of well-defined inclusive and exclusive criteria, most recently updated in 2010 (4,5) (Table 1.1).

3 Table 1.1: Diagnostic criteria for RTT. Patients must fulfill all main criteria and exclusion criteria. Supportive criteria are not required but are common (4). Required for classic 1. Period of regression followed by stabilization RTT 2. All main criteria and all exclusion criteria

Main criteria 1. Partial or complete loss of acquired hand skills 2. Partial or complete loss of acquired spoken language 3. Gait abnormalities: impaired (dyspraxia) or absent 4. Stereotypic hand movements: wringing, clapping/tapping, mouthing, washing, and/or rubbing

Exclusion criteria 1. Brain injury secondary to trauma (peri- or postnatally), neurometabolic disease, or severe infection with neurological problems 2. Grossly abnormal psychomotor development in the first 6 months of life (normal milestones of head control, swallowing, social smiling, etc. are not met)

Supportive criteria 1. Breathing abnormalities 2. Bruxism 3. Impaired sleep patterns 4. Abnormal muscle tone 5. Peripheral vasomotor disturbances 6. Scoliosis/kyphosis 7. Growth retardation 8. Small cold hands and feet 9. Inappropriate laughing/screaming spells 10. Diminished response to pain 11. Intense eye communication (“eye pointing”)

Other notes 1. Since MECP2 mutations can be identified before regression, a diagnosis of “possible” RTT should be given to those under 3 years old who have not lost any skills but have clinical features of RTT. They should be reassessed every 6-12 months for evidence of regression 2. Loss of acquired language is based on acquired spoken language skill, not on higher language skills (ie. learning to babble, then losing this ability, is considered loss of acquired language) 3. Many supportive criteria are age-dependent and will only manifest in older individuals

4 Following a period of normal neurological and physical development during the first 6-18 months of life, the first features of RTT begin to manifest in early childhood and appear progressively over several stages (Figure 1.1):

1. Stagnation stage (age 6 – 18 months): Patients experience subtle developmental delays in motor and language skills such as sitting, crawling and vocalizations. Some patients may experience mild hypotonia and exhibit postural delays. This stage is often overlooked as these subtle changes go unnoticed, leading to a delayed diagnosis.

2. Rapid regression stage (age 1-4 years): Purposeful hand skills are replaced by stereotypic and repetitive hand movements and spoken language is lost. Patients experience motor impairments and develop abnormal breathing including hyperventilation and breath-holding. Patients may develop autistic-like features such as social withdrawal, anxiety, and profound irritability. Seizures may occur.

3. Plateau stage (age 4 – potentially life): Motor problems become more common, but patients may show more interest in their surroundings and exhibit increased alertness. However, breathing abnormalities and seizures persist.

4. Motor deterioration stage (age 5 – life): Patients may experience muscle wasting, dystonia and develop scoliosis and other co-morbidities. Many patients become wheelchair-dependent and may rely on a gastronomy-tube. Most patients will require around-the-clock care in this stage.

Figure 1.1: Timeline of stages in RTT patients. The course of Rett syndrome (RTT) occurs over four progressive stages: 1. Stagnation, 2. Rapid regression, 3. Plateau, and 4. Motor deterioration.

1.1.2 Stage-independent features of Rett syndrome

Many children diagnosed with RTT have a reduced brain volume compared to healthy individuals, with an overall decrease in brain weight up to 35% (6,7). Decreased brain size is not a simple uniform reduction as, for example, the cortical region is reduced in size more substantially than the cerebellum. RTT patient brains also have smaller neuronal bodies and more densely packed neurons, particularly in layers III and V of the cerebral cortex, thalamus, substantia nigra, basal ganglia, amygdala, cerebellum, and hippocampus (8). Patients also have reduced dendritic arborization, indicative of a delay in neuronal maturation (9). Additionally, decreased synaptic density and dysregulated neurotransmitters, neuromodulators and transporters have been seen (9–11). Despite this, no evidence of neuronal or glial cell death has been found in RTT patients or in animal models of the disease, indicating RTT is not a neurodegenerative condition.

A number of patients have altered brain carbohydrate metabolism (12) and neurometabolites associated with cell integrity and membrane turnover (13,14). Dyslipidemia (15,16) and elevated plasma leptin and adiponectin (17,18) have also been reported. Additionally, energy-producing mitochondria have an abnormal structure in patient cells (19–22), consistent with altered electron transport chain complex function (23), increased oxidative stress (24–26), and elevated lactate and pyruvate in blood and cerebrospinal fluid (12,23). Patients may also experience gastrointestinal issues (27), altered QT intervals (28), decreased bone density (29), early-onset osteoporosis (30), gallbladder inflammation (31), scoliosis (31), urological dysfunction (32), and sleep disturbances (33). The involvement of multiple tissues and processes highlights that RTT is a multi-systemic disorder (Figure 1.2).

Treatment for RTT patients is currently limited to symptom control. With attention to orthopedic complications and seizure control, 70% of RTT patients reach the age of 50 (34). However, patients have a sudden and unexpected death rate of 26%, much higher than healthy individuals of a similar age, and typically die due to cardiac instability, lung infection or respiratory failure (31,35).

Figure 1.2: RTT is a multi-systemic disorder. Independent of disease stage, RTT patients develop symptoms that broadly affect various tissues, organs, and cell components. Figure was created using BioRender.

8 1.1.3 Classic RTT is caused by mutations in MECP2

A genetic basis for RTT was hypothesized as early as 1983 based on the preferential involvement of females (2), but as most cases of RTT are sporadic, finding the causative gene was particularly challenging. Multipoint linkage analysis of a family with three affected and three unaffected daughters narrowed the location of the gene to Xq28 of the X chromosome (36). The mother was a carrier of the unknown X-linked mutant gene but presented no symptoms because of her favorable X-inactivation. Using a systematic gene screening approach of nearly 100 candidate genes in Xq28, mutations were identified in the methyl-CpG-binding protein 2 (MECP2) gene in seven patients (36). It is now well-established that de novo mutations in MECP2 account for 95% of typical RTT cases (37). RTT patients are heterozygous for MECP2 mutation, carrying one normal and one mutated copy of MECP2.

1.1.4 Atypical Rett syndrome

When a child presents with RTT-like symptoms but does not fulfill the diagnostic criteria for RTT, they may be diagnosed with atypical RTT, symptoms of which deviate in age of onset, clinical profile, and/or severity (Table 1.2) (4). Most cases of atypical RTT are associated with mutations in X-linked cyclin-dependent kinase-like 5 (CDKL5; OMIM #300203) or Forkhead box G1 (FOXG1; OMIM #164874), but some causative genes remain undefined.

1.1.5 The MECP2 gene

The MECP2 gene was originally mapped to Xq28 region in 1994 and was subsequently shown to undergo X-inactivation (38,39).The MECP2 gene has 4 exons that span 76kb in humans but only 59kb in mice, due to a significantly smaller second intronic region (40). MECP2 has an 8.5kb long 3’UTR, one of the longest known in the human genome, speculated to regulate MECP2’s downstream translation by controlling its mRNA degradation and stability (41). MECP2 protein levels correlate poorly with its transcript levels. This may be related to variable expression of 1.8kb, 5.4kb, 7.5kb, and 10.2kb transcripts that result from four alternative polyadenylation sites (Figure 1.3) (39,42). These transcripts are differentially expressed across tissues and over stages of development. For example, only the longest and shortest mRNAs are detected in neurons while the 1.8kb and 5.4kb transcript are predominantly expressed in the liver and lung (42,43). Multiple polyadenylation sites determine whether any specific MECP2 transcript contains binding sites for proteins and miRNAs that regulate transcript stability and translation. MECP2’s post-translational modifications (discussed in 1.2.3) may also affect transcript to protein ratios.

Table 1.2: Atypical Rett syndrome variants. These variants may be milder or more severe than classical RTT. Severe atypical RTT variants

Type Description

Early-onset seizure type • Can be caused by mutations in X-linked cyclin-dependent kinase-like 5 gene (CDKL5; OMIM #300203) • Seizures in the first months of life • Develop RTT symptoms Congenital variant • Can be caused by mutation in Forkhead box G1 (FOXG1; OMIM #164874) located on chromosome 14 • Born with congenital microcephaly and intellectual disability • Lack of normal psychomotor development • Develop RTT symptoms during first 3 months of life Mild atypical RTT variants

Type Description

Late regression type • Develop RTT symptoms at a preschool age.

Preserved speech “Zapella” • Develop RTT symptoms but recover some verbal skills and can form phrases and type sentences.

“Forme fruste” variant • Well-preserved motor skills and only subtle neurological abnormalities such as mild hand dyspraxia. • The most common atypical variant accounting for 80% of cases

1.1.6 The MECP2 protein and its isoforms

MECP2 is a 486-amino acid protein with a predicted molecular mass of 53 kDa. It was originally identified in a screen for proteins with methyl-DNA specific binding activity (44). MECP2 is an abundant nuclear protein that is ubiquitously expressed in all human tissues, but is particularly abundant within neurons (44). Immunofluorescence studies show MECP2 localizes to heterochromatin, regions with high 5-methyl-cytosine concentrations, indicating it preferentially binds to methylated CpGs.

MECP2 is expressed as two isoforms: MECP2_exon1 (MECP2_e1) and MECP2_exon2 (MECP2_e2) (Figure 1.3) (45). MECP2_e1 employs exons 1, 3 and 4, skipping exon 2. The two isoforms also differ at the N-terminus where MECP2_e1 contains a polyalanine tract and ‘EERL’ motif. The MECP2_e1 isoform is the ancestral form and is conserved across vertebrates, while MECP2_e2 is only present in mammals (46).

The functional difference between protein isoforms remain unknown. Both MECP2 isoforms are nuclear and co-localize with methylated heterochromatin foci in mouse cells (46). However, MECP2_e1 is 10 times more abundant than MECP2_e2 in the human brain as well as in mouse thymus and lung, whereas a 1:1 ratio is seen in mouse testis and liver. This may be partly due to the presence of an upstream open reading frame which causes MECP2_e2 to be translated at a much lower efficiency (46). Interestingly, deletion of the MECP2_e2 isoform in mice results in embryonic lethality due to placental defects but does not result in RTT-like symptoms if removed later in life (47). However, deletion of MECP2_e1 causes an RTT-like phenotype, suggesting it is more functionally relevant in the disease (48). Despite this, amino acid coordinates for mutations and post-translational modifications in the literature refer to the MECP2_e2 isoform, as it is the older isoform in the literature.

MECP2 is present in nearly all vertebrate species and its protein sequence is well conserved throughout evolution. The amino acid sequence of MECP2 is 95% identical between human and mouse, which is a greater level of conservation than average for all proteins (86.4%), suggesting the entire amino acid sequence is functionally important (49). Higher levels of sequence conservation are present between humans, chimpanzees (Pan troglodytes) and dogs (Canis lupus).

Figure 1.3: MECP2 is expressed in two isoforms. The MECP2 gene (middle) is subject to alternate splicing (dashes) to produce two isoforms: MECP2_e1 (above), a 498 amino acid protein, or MECP2_e2 (below), a 486 amino acid protein. Either isoform can use different polyadenylation sites in the 3’UTR (dark grey) resulting in different length transcripts. Arrows indicate ATG start codons for each isoform. Colored boxes indicate different domains: yellow: N-terminus, green: methyl-CpG-binding domain (MBD), blue: transcriptional repression domain (TRD), red: NCOR1/2-interaction domain (NID). Numbers correspond to amino acids.

12 1.1.7 MECP2 domains

The ability of MECP2 to bind methylated CpG dinucleotides has since been shown by a variety of in vitro and in vivo techniques (44,50,51). The binding specificity of MECP2 is dependent on two factors: the presence of methylated DNA, and MECP2’s methyl-CpG-binding domain (MBD) located within a region between amino acids 89-162 and encoded by exons 3-4 (Figure 1.4). Sequence similarity searches revealed ten other proteins carrying the MBD domain (MBD1-6, BAZ2A, BAZ2B, SETDB1, and SETDB2), but only MECP2, MBD1, MBD2, and MBD4 can bind DNA in a methylation-specific manner (52). While deletion of Mecp2 in mice leads to disease and premature death, deletion of other MBD family members results in minimal phenotypes, suggesting MECP2 is the MBD family member with the greatest role in interpreting the DNA methylome (52).

MECP2 has a bipartite nuclear localization signal (NLS) (Figure 1.4), though it is not required for MECP2’s nuclear localization. Rather, an intact MBD is sufficient to bring MECP2 to the nucleus, highlighting its strong affinity for DNA (53). MECP2 also contains several adenine-thymine rich DNA hooks (AT hooks) which bind to the minor groove of AT-rich DNA, providing MECP2 with additional non-specific DNA binding affinity.

Vertebrate genomes are highly CpG methylated at 60-90% compared to their invertebrate counterparts at 10-40% (54). Site-specific DNA methylation is thought to act as a transcriptional repressive mark through recruitment of DNA-binding proteins like MECP2, which then attract transcriptional repressor complexes. A transcriptional repressor domain (TRD) was identified in MECP2 between amino acids 207-310, encoded by exon 4, that is capable of repressing the transcription of genes up to ~2kb away (Figure 1.4) (55,56). Within the TRD lies the NCOR1/2 interaction domain (NID) comprised of amino acids 289-309. As its name suggests, the NID allows MECP2 to interact with the histone-deacetylase (HDAC)-containing NCOR1/2 co-repressor complex. This complex is discussed in greater detail later in this chapter.

MECP2 is an intrinsically disordered protein (IDP), which lacks a native tertiary 3D structure (57,58). Like other IDPs, MECP2 acquires a tertiary structure upon binding to other molecules and post-translational modifications (PTMs) allow MECP2 to bind to its correct partners (Figure 1.5). MECP2 undergoes many PTMs including phosphorylation, ubiquitination, SUMOylation, acetylation, methylation, and GlcNAcylation (59). Additionally, MECP2 undergoes site-specific activity-dependent phosphorylation in neurons suggesting it may function as a neuronal activity- regulated transcriptional regulator. For example, phosphorylation of T308 occurs in neurons following neuronal activity and prevents the interaction of MECP2 with the NCOR1/2 complex, thereby relieving transcriptional repression (60).

13 Within the C-terminal area of the MECP2 protein is a WW-domain-binding region (WDR). WDRs are found in subunits of multiprotein complexes that are involved in signaling pathways for DNA damage and repair, ubiquitin signaling, and protein degradation (61). Additionally, MECP2 contains two proline-glutamic acid, serine, threonine (PEST) rich sequences. PEST sequences are associated with proteins with short intracellular half-lives and are thought to act as a signal peptide for phosphorylation-dependent protein degradation (62). Within the PEST motif, a serine residue can be phosphorylated followed by ubiquitination of a nearby lysine. Ubiquitination targets the protein for rapid degradation by the 26S ubiquitin proteasome system. Thus, these sequences may be important for maintaining an appropriate level of MECP2 within cells.

Figure 1.4: Functional domains and post-translational modifications of MECP2. Amino acid coordinates are for MECP2_e2. MECP2 contains one nuclear localization signal (NLS, purple slashed box), three adenine-thymine hooks (AT-hooks, black boxes), two proline- glutamic acid, serine, threonine-rich sequences (PEST domains, orange slashed boxes), and one WW-domain-binding region (WDR, green slashed box). PTMs occur throughout the MECP2 protein and regulate its interaction with binding partners.

15 1.1.8 RTT-causing MECP2 mutations

To date, 562 RTT-causing mutations have been identified in MECP2, with another 52 that may be linked to atypical RTT or RTT-like symptoms (63). RTT-causing MECP2 mutations include point mutations, deletions, insertions, and whole gene deletions. However, eight mutations are found in approximately 70% of RTT patients (Figure 1.5). These include four missense mutations (R106W, R133C, T158M, and R306C) and four nonsense mutations (R168X, R255X, R270X, and R294X) that result in a premature STOP codon (64). Large deletions in MECP2 account for 5% of RTT-causing mutations, while small C-terminal deletions constitute another 8%.

Approximately 99.5% of MECP2 mutations occur de novo, and 70% of these are C>T transitions. Unlike the X chromosome in oocytes, the X chromosome in sperm is hypermethylated. It has been speculated that spontaneous deamination of methylated cytosine residues may result in a transition to a thymine, increasing the potential for deleterious mutations at hypermethylated CpG sites, such as those within the MECP2 locus. As such, most de novo MECP2 mutations originate from the paternally inherited X chromosome (65).

Figure 1.5: RTT-causing MECP2 mutations. Y-axis represents the percent of RTT patients with the indicated mutation. Missense mutations (blue lines) and nonsense mutations (pink lines) constitute 70% of all RTT-causing mutations. These mutations largely cluster in the MBD and TRD/NID. Large deletions (light blue) generally occur upstream of the NID while small truncations (light green) of 20-500 base pairs often occur in the C-terminal region. MBD: methylated-CpG-binding domain, TRD: transcriptional repression domain, NID: NCOR1/2-interaction domain. Amino acids – R: Arginine, W: tryptophan, C: cysteine, T: threonine, M: methionine, X: stop codon.

17 1.1.9 Phenotypic variation in RTT

The severity of symptoms in classic female RTT patients can vary greatly; some patients may be wheelchair-bound before the age of five, while others may retain the ability to walk and be able to say a few words (66). Given the wide variety of mutation types and phenotype severity, genotype- phenotype studies have correlated mutation status with clinical features of RTT patients (64). Early truncating mutations in MECP2 (eg. R168X, R255X, and R270X) and large deletions cause the most severe phenotype, whereas most missense mutations (eg. R133C and R306C) and late truncating mutations (eg. R294X) are the mildest (64,67) (Figure 1.6A). Despite this, phenotypic variation is also reported in familial cases of RTT where affected sisters present with the same mutation (68); these differences may be due to differences in random X chromosome inactivation (XCI).

The choice of which of the two X chromosomes will be inactivated is random in placental mammals (69). As such, female RTT patients are heterozygous mosaics with roughly half of their cells expressing the mutant MECP2 allele. Occasionally, XCI can be skewed so that the X chromosome carrying the mutant MECP2 allele is more or less expressed throughout the brain and body than the normal allele, influencing their clinical presentation (Figure 1.6B). Tests for skewed XCI are possible in the clinic.

In other cases, phenotypic variation may be due to the presence of modifier mutations. Modifier mutations effect phenotypic outcomes through acting on other genes and thus can alleviate or enhance clinical symptoms through modulating the effects of MECP2 deficiency (Figure 1.6C). Some studies implicated the cilia motility gene CROCC, the neuronal growth regulator GPRIN2, and a protein involved in energy homeostasis, PPYR1, as modifiers of MECP2 mutant phenotypes (70).

Figure 1.6: RTT symptom severity is influenced by mutation status, XCI pattern, and modifier genes. A. The MECP2 missense mutations R133C and R306C and C-terminal deletions cause the least severe clinical presentation, whereas the missense mutations R106W and T158M, nonsense mutations R168X, R255X, R270X, and R294X, and large deletions cause the most severe phenotype. B. Patients with fewer cells expressing the mutant MECP2 allele due to favorable X chromosome inactivation (XCI) patterns will have less severe symptoms than patients with many cells expressing the mutant MECP2 allele. C. Individuals with modifier mutations in genes that suppress pathways downstream of MECP2 mutation have a more favorable clinical presentation than individuals with mutations in genes that enhance detrimental changes.

19 1.1.10 MECP2 expression and localization

MECP2 is expressed ubiquitously across all tissues but is particularly abundant in the brain where it reaches near-histone level abundance in neurons (~16x106 molecules per nucleus) (71). In the brain, MECP2’s expression increases during embryonic and postnatal development, reaching a plateau at 10 years of age in humans and 5 weeks in mice (51,71). In humans, the key developmental processes of neuronal maturation and synaptogenesis occur during embryonic weeks 12 and 20, respectively (72); MECP2 loss within this window may result in microcephaly as seen in patients. Within the brain, MECP2 is primarily expressed in mature post-migratory neurons (73), though it is also found in astrocytes (74), oligodendrocytes (75), and microglia (76).

RTT patients exhibit neurons with decreased dendritic spine density and reduced dendritic arborization across many brain areas including the cortex and hippocampus, suggesting that MECP2 regulates dendritic morphology (77–79). Expectedly, reduced dendritic complexity influences synaptogenesis. Studies in Mecp2-mutant mice have shown decreased and increased spontaneous excitatory synaptic transmission in cortical and hippocampal neurons, respectively, suggesting a shift in the ratio of excitation to inhibition (80,81).

Human Protein Atlas data shows that MECP2 is highly expressed in the lung, liver, heart, kidney, gastrointestinal tract, pancreas, and muscle (82). Many symptoms of RTT including respiratory, cardiac, urological, and gastrointestinal issues could be derived from dysfunctional cellular regulation due to Mecp2 deficiency outside the CNS.

1.1.11 MECP2 function

A predominant view is that MECP2 acts as a multi-functional hub interacting with over 40 different binding partners to participate in a variety of cellular processes. Its roles may include transcriptional repression, transcriptional activation, chromatin remodeling, alternative splicing, and miRNA processing (Figure 1.7).

While MECP2’s role in transcriptional repression was initially reported in 1997, the inability to pinpoint distinct upregulated genes in Mecp2’s absence puzzled the field (55). Instead, mice lacking Mecp2 showed minute changes in gene expression in both directions, leading to the belief that MECP2 may both repress and activate transcription (83,84). However, ChIP-seq and RNA- seq data have shown that MECP2’s occupancy is highest at promoters of genes that are upregulated in a Mecp2-deficient state, suggesting that genes with decreased expression in Mecp2’s absence are an indirect effect of Mecp2 loss (85,86). Thus, recent studies have advocated that MECP2’s most important role is to repress gene transcription. Nonetheless, all of MECP2’s putative roles will be discussed below.

Figure 1.7: Possible functions of MECP2. A. Transcriptional repression: MECP2 binds methylated CpG dinucleotides (red circles) and recruits co-repressor complexes such as NCOR1/2 and SIN3A to regulatory sites around the targeted loci to repress gene transcription. B. Transcriptional activation: MECP2 binds non- methylated DNA (empty red circles) and recruits co-activators such as CREB-1 to increase transcription from the locus. C. Chromatin compaction: MECP2 binds chromatin remodelers such as ATRX to tightly wind chromatin and prevent transcription. D. Alternative splicing: MECP2 binds splicing regulators such as YB-1 to regulate splicing of mRNAs. E. miRNA processing: MECP2, which is phosphorylated at S80, binds to the DGCR8 microprocessor complex, and prevents its interaction with Drosha, thus preventing miRNA processing. Calcium dependent neural activity dephosphorylates MECP2 S80, releasing DGCR8.

21 1.1.11.1 MECP2 as a transcriptional repressor Genome-wide analyses have shown that CpG-dense promoters of actively transcribed genes are unmethylated, implicating CpG methylation in transcriptional repression (87). CpG methylation is thought to physically block transcriptional machinery from binding to gene promoters, and/or may be responsible for recruiting DNA-binding proteins, which attract repressor complexes for chromatin compaction and transcriptional repression. MECP2 binds to methylated CpGs through its methyl-CpG-binding domain (MBD), while its transcriptional repression domain (TRD) allows its interaction with various corepressor proteins (Figure 1.7A). Unlike other methyl-DNA binding proteins, such as MBD1-6, MECP2 can bind to a single, symmetrical mCpG site, making it the most efficient DNA reader of this family.

MECP2 binds to histone deacetylase (HDAC)-containing complexes, NCOR1/2 (88,89) and SIN3 transcription regulator family member A (SIN3A) (90), as well as three co-repressors that interact with these complexes, c-Ski, PU.1 and CBF1 (88,91,92), and thus could repress transcription in a HDAC-dependent manner. MECP2 has also been suggested to mediate transcriptional silencing through interacting with YY1, SOX1 and SP3 transcription factors (93–95), and the histone methyltransferases PRMT6, G9a, and HLCS (96,97). Of these interactions, MECP2’s role in the NCOR1/2 corepressor complex has been best studied. This interaction was first described in 2001 (88), but was not re-examined until over a decade later when mass spectrometry analysis of MECP2-associated proteins revealed an association with five subunits of the NCOR1/2 corepressor complex (89).

NCOR1 and NCOR2 are highly homologous proteins that are recruited to chromatin and act as a scaffold protein for nuclear receptors, DNA-binding proteins, and histone deacetylases (98). Both NCOR1 and NCOR2 are intrinsically disordered proteins, but have two regions that fold into distinct conformations: a SANT-like domain that recruits and activates HDAC3 and is hence called the deacetylase activation domain (DAD, amino acids 412-480 in NCOR2), and a second SANT- like domain that directly interacts with histone tails and is called the histone interaction domain (HID). The DAD domain of NCOR1/2 binds with the N-terminus and C-terminus of HDAC3 and mutation of the DAD or either region in HDAC3, abolishes binding. HDAC3 removes histone acetyl marks and establishes a closed chromatin state, repressing transcription (99). Additionally, NCOR1/2 recruit and bind other components of complex, including G protein pathway suppressor 2 (GPS2), transducin beta-like 1 (TBL1X) and transducin beta-like 1 related (TBL1XR1) (100). TBL1X and TBL1XR1 form a tetramer, and each TBL1X/TBL1XR1 dimer binds to one molecule of NCOR1 or NCOR2. Additionally, TBL1X and TBL1XR1 bind to GPS2, forming a stable three- way complex (101).

22 The interaction between MECP2 and the NCOR1/2 complex is mediated by the NID located at amino acids 289-309 in MECP2, within the TRD (60,89). MECP2 binds to complex members TBL1X and TBL1XR1 at these residues via TBL1X/TBL1XR1’s WD40 domains, unlike other NCOR1/2 recruiters which interact with the NCOR scaffold proteins directly (102). These findings have led to the ‘bridge hypothesis’ wherein MECP2 bridges methylated DNA and the corepressor complex, anchoring NCOR1/2 to DNA via its NID and MBD, respectively (Figure 1.8).

RTT-causing missense mutations are largely confined to MECP2’s MBD and NID, highlighting the importance of these two regions in RTT pathology (103). It is speculated that breaking the ‘bridge’ at either end (ie. MBD or NID mutations) cause RTT. Within the NID is a mutational hotspot between amino acids 302-306 (89). R306C, the most common RTT-causing mutation, produces a protein capable of binding to methylated DNA but unable to recruit the NCOR1/2 complex. This suggests that the interruption of the MECP2-NCOR1/2 complex interaction alone can cause RTT.

The bridge hypothesis implies that mutations in other members of the NCOR1/2 complex could also break the DNA-MECP2-NCOR1/2 chain, resulting in RTT. Mice lacking Ncor1, Ncor2, Gps2, or Hdac3 are embryonic lethal (104–106), which is not surprising since NCOR1/2 complexes have other, MECP2-independent, functions. However, mutations in TBL1X and TBL1XR1 cause developmental delays; many of these mutations are located in the WD40 domain which is required to bind MECP2. Additionally, one patient with a D370N mutation in TBL1XR1 was diagnosed with classic RTT (107).

To test whether the primary role of MECP2 is to recruit NCOR1/2 to DNA, the functional dispensability of MECP2 regions was assessed (108). Mice expressing a MECP2 protein consisting of only the MBD, NID, and short linker regions were generated, while all other amino acid sequences of the protein were removed. Remarkably, mice expressing this truncated protein developed only mild RTT-like symptoms and had a normal life span. Additionally, genetic reactivation of this minimal MECP2 protein in Mecp2-deficient mice prevented or ameliorated symptoms depending on time of reactivation, suggesting that disruption of the MECP2-NCOR1/2 interaction is pivotal in RTT.

Figure 1.8: The ‘bridge hypothesis’ of MECP2 function. A. In normal cells, MECP2 binds to methylated CpG dinucleotides (red circles) through its methyl- CpG-binding domain (MBD) and recruits the NCOR1/2 co-repressor complex by binding to TBL1X/TBL1XR1 through its NID. HDAC3 deacetylates histones (purple triangles), causing a closed chromatin state and transcriptional repression of target genes. B. When MECP2 is absent, the NCOR1/2 co-repressor complex cannot bind to target loci. Histones remain acetylated (purple triangles) and chromatin remains in an open state. Target genes are transcriptionally active. NCOR1/2: Nuclear receptor co-repressor 1/2; HDAC3: histone deacetylase 3, TBL1X: transducin (beta)-like 1X-linked, TBL1XR1: F-box-like/WD repeat-containing protein TBL1XR1, GPS2: G protein pathway suppressor 2, MECP2: methyl-CpG-binding protein 2.

24 1.1.11.2 MECP2 as a transcriptional activator

Although MECP2 is widely considered a transcriptional repressor, there are indications that it participates in transcriptional activation as well. The entire MECP2 protein sequence is extremely conserved in vertebrate species, not just the MBD and NID, suggesting important evolutionary functions outside of these two regions. Additionally, genome-wide analysis of MECP2 binding revealed that a large fraction of MECP2-bound promoters are for genes that are actively expressed (109). In a mass spectrometry analysis, immunopurified MECP2 bound to cAMP response element-binding protein 1 (CREB-1), a major transcriptional activator (Figure 1.7B) (84). Subsequent bisulfite sequencing and ChIP analysis showed that MECP2 bound to promoters of four genes undergoing transcriptional activation (71). Interestingly, the four genes had unmethylated promoters, suggesting a methylation-independent mechanism by which MECP2 activates transcription.

1.1.11.3 MECP2’s role in chromatin remodeling

MECP2’s ability to repress transcription in an HDAC-independent manner prompted studies on its role in chromatin remodeling. In a yeast two-hybrid screen, MECP2 was found to bind the transcriptional regulator ATRX (110). ATRX is a member of the SWitch/sucrose non-fermentable (SWI/SNF) family of chromatin modelers. While ATRX is not well-characterized, SWI/SNF members are known to enforce nucleosome rearrangement in an ATP-dependent manner (111). This facilitates the formation of higher order chromatin structures which provide more or less access to chromatin, allowing genes to be activated or repressed, respectively. MECP2 binds ATRX through amino acids 108-169, and mutations in this region prevent MECP2 from recruiting ATRX to heterochromatin (110). While the significance of this interaction remains unknown, ATRX does in turn interact with chromatin-associated proteins HP1α and EZH2, which modify chromatin in vitro (112–114). Thus, MECP2-mediated transcriptional silencing could also be ascribed to its interaction with ATRX and their combined chromatin reorganization (Figure 1.7C).

1.1.11.4 MECP2’s role in alternative splicing

In another search for MECP2’s binding partners, its interaction with Y box-binding protein 1 (YB- 1) was discovered (115). YB-1 has roles in pre-mRNA splicing and DNA repair. Thus, it was hypothesized that MECP2 may synchronize alternative splicing with gene transcription by recruiting YB-1 to nascent transcripts immediately after MECP2 is released from gene promoters (Figure 1.7D). Subsequently, MECP2 was shown to bind to two mRNA splicing modulators, PC4 and SSFRS1 interacting protein (PSIP1, formerly LEDGF) and DExH-box helicase 9 (DHX9) (116). Interestingly, abnormally spliced genes have been identified in Mecp2-mutant mice (116).

25 Altered splicing is a significant contributor to a number of neurological diseases including spinal muscular atrophy (SMA) and amyotrophic lateral sclerosis (ALS) and could be involved in RTT.

1.1.11.5 MECP2’s role in miRNA processing

While altered gene expression in the absence of MECP2 points to its role in transcriptional regulation, it also opens the possibility for MECP2’s potential role in controlling posttranscriptional regulators, such as microRNAs (miRNAs). miRNAs regulate gene expression by initiating degradation of target mRNAs or physically inhibiting protein translation. The biogenesis of miRNAs begins with their transcription from the genome, followed by processing through Drosha/DiGeorge syndrome critical region 8 (DGCR8) and Dicer complexes. MECP2 binds directly to DGCR8 and interferes with the assembly of its miRNA processing complex (Figure 1.7E) (117). Phosphorylation at MECP2 serine 80 (S80) is crucial for this interaction; interestingly, neural activity induces the calcium-dependent dephosphorylation of S80, releasing MECP2 and allowing miRNA processing to continue. Further, miRNA expression profiles are altered in the brains of mice lacking Mecp2 (118,119). In Mecp2-deficient cells, miRNAs miR-199 and miR-214 which regulate extracellular signal-regulated kinase (ERK/MAPK) and protein kinase B (PKB/AKT) signaling, respectively, are increased (120). Both pathways are involved in neurogenesis and neuronal migration.

1.1.12 Males with MECP2 mutations

As a dominant X-linked disorder, RTT was considered lethal in males. However, males with clinical features resembling classical RTT were reported. Males with MECP2 mutations are currently classified into four categories: severe neonatal encephalopathy and infantile death, typical RTT, less severe neuropsychiatric phenotypes, or MECP2 duplication syndrome (Table 1.3) (121–125). Males in the first group carry a MECP2 mutation passed on by their asymptomatic mothers who usually have favourable X-chromosome inactivation. These males, if born, develop neonatal encephalopathy, severe respiratory arrests and seizures, and typically die shortly after birth (121,122). The second group of males who develop classic RTT are either somatic mosaics for MECP2 mutation or have at least partial Klinefelter’s syndrome (XXY karyotype). These patients have a subset of cells which express normal MECP2, and show a symptom progression similar to female RTT patients. The third group involves males who have non RTT-causing MECP2 mutations that would present with mild or no clinical features in a heterozygous female. These patients have a broad range of symptoms including intellectual disability, autism, and motor abnormalities. Finally, MECP2 duplication syndrome, which occurs almost exclusively in males, is due to a gain in MECP2 dosage. Affected patients exhibit developmental delays, intellectual disability, and recurrent lung infections with almost 50% dying before the age of 25 (126).

26 Table 1.3: Males with MECP2 mutations fall into four categories. Category Genetic Profile Features

Severe neonatal MECP2 mutation passed on by • Spontaneously miscarried encephalopathy mildly symptomatic or asymptomatic • If born, develop neonatal and infantile death mother encephalopathy, respiratory arrest and seizures and die within 2 years

Classical RTT At least partial Klinefelter’s • Symptoms similar to female syndrome (XXY karyotype) or other RTT patients somatic mosaicism

Less severe Non RTT-causing MECP2 mutation • Symptoms overlap with neuropsychiatric Angelman syndrome symptoms (intellectual disability and motor abnormalities)

MECP2 duplication Gain of MECP2 dosage • Hypotonia, intellectual disability, syndrome absent or limited speech and walking, lung infections, and seizures • 50% die before age 25

27 1.2 Studying MECP2 using animal models

Several animal models have been generated to better define MECP2 functions, as well as to understand the progression and pathology of RTT and other MECP2-related disorders. While mice are the most commonly used organism to model RTT, MECP2 has been studied in Drosophila, zebrafish, rats and chimpanzees (Figure 1.9) (127–133).

1.2.1 Mecp2-null mouse models

Developing a mammalian model of RTT was crucial to understanding the mechanistic basis responsible for clinical symptoms. Indeed, mouse models have been instrumental in helping to understand RTT pathogenesis. Shortly after the identification of MECP2 as the causative gene in RTT, two Mecp2-null mouse models were generated which are now the primary models used to study this disease (Table 1.4). Adrian Bird’s laboratory made the Mecp2tm1.1Bird mouse by engineering loxP sites flanking exons 3 and 4 to make a “floxed” conditional mouse line (134). Floxed mice were crossed with transgenic mice carrying a ubiquitously-expressed Cre (CMV- Cre), producing Mecp2tm1.1Bird mice which completely lack MECP2 protein. At the same time, Rudolf Jaenisch’s laboratory used the same basic strategy with loxP sites flanking Mecp2 exon 3 to produce the Mecp2tm1.1Jae mouse (135). Mecp2tm1.1Jae mice maintain low levels of Mecp2 expression. However, both models are considered “null” and display a similar phenotype that recapitulates symptoms of RTT.

Figure 1.9: Animal models used to study MECP2. While there is no homolog for MECP2 in drosophila, transgenic flies overexpressing human MECP2 were used to identify interactors. A mecp2 null zebrafish model revealed roles in neuronal differentiation and migration, skeletal and cardiac function, and energy homeostasis. Mild symptoms in zebrafish are likely due to their continued neurogenesis throughout life. Mice are the most common used animal model to study MECP2 and several null, missense, conditional and overexpression models have been created (summarized in Table 1.4). Rats with a null mutation in Mecp2 develop symptoms similar to Mecp2-null mice and have been used to study neurological and social phenotypes. Recently, primate models of MECP2 were developed using MECP2- targeting transcription activator-like effector nuclease (TALEN) plasmids and showed many symptoms that overlap with RTT patients.

Table 1.4: Mouse models created to study Rett syndrome. Mouse models include null alleles, point mutations designed to recapitulate mutations observed in human RTT patients, whole exon deletions, and conditional alleles used in combination with targeted Cre mice to achieve temporal deletion of MECP2. Male Phenotype Allele Type Allele Description RB BW BR AX M PD Early Death Ref

Null Mecp2tm1.1Bird Null Exon 3-4 Deletion X X^ X X X X 8-10 weeks (134) Mecp2tm1.1Jae Null (*) Exon 3 Deletion X X* X X X X 10 weeks (135) Mecp2tm1Pplt Null MBD Deletion X X NT X X X 8 weeks (136)

Human point mutations Mecp2tm4.1Joez R106W Missense mutation X NT NT NT NT X 10 weeks MGI Mecp2tm1Nlnd Y120A Missense mutation X X NT - X X 14-17 weeks (137) Mecp2tm6.1Bird R133C Missense mutation X X NT X X X 42 weeks (138) Mecp2tm1.1Joez T158A Missense mutation X - NT X X X 16 weeks (139) Mecp2tm4.1Bird T158M Missense mutation X X NT X X X 13 weeks (89,138) Mecp2tm3.1Joez T158M Missense mutation X NT NT NT NT X 14 weeks MGI Mecp2tm1.1Jtc R168X Stop mutation; truncation X X X X X X 12-14 weeks (140–142) Mecp2tm1.1Irsf R255X Stop mutation; truncation X X X X X X 8-10 weeks (143) Mecp2tm5.1Bird R306C Missense mutation X X NT X X X 30 weeks (89,138)

Other mutations Mecp2tm2.1Jae S80A Missense mutation NT X* NT NT X NT NT (144) Mecp2tm1Vnar A140V Missense mutation ------(145) Mecp2tm3Meg T308A Missense mutation X NT NT NT X NT >16 weeks (60) Mecp2tm1Hzo R308X Stop mutation; truncation X - NT X* X X 6-12 weeks (146) Mecp2tm1.1Meg S421A Missense mutation ------(147)

Mecp2tm1.1Mitoh Deletion Isoform 2 deletion ------(47) Mecp2tm1.1Dhy Deletion Isoform 1 deletion X - NT X X X 7-31 weeks (48)

Conditional alleles Mecp2tm1Bird - Exons 3-4 floxed X* - X X X - - (134,148) Mecp2tm1Jae - Exon 3 floxed ------(135) Mecp2tm2Bird - Floxed-stop upstream of exon 3 X X X X X X 8-10 weeks (149) Overexpression Tg(MECP2)1Hzo - ~2x expression of MECP2 NT NT NT X* X - >1 year (150,151) Mapttm1(Mecp2)Jae - ~5-9-fold expression of MECP2 NT NT X NT NT - >1 year (152) Mapttm1(Mecp2)Bird - ~3.8x expression of MECP2 NT NT NT NT NT X <3 weeks (153)

(*): Some protein product is retained; X: present, -: not present, NT: not tested; RB: reduced brain size, BW: reduced body weight, BR: breathing abnormalities, AX: reduced anxiety, MA: motor abnormalities, PD: premature death. In BW category, *: increased body weight, ^:increase or decrease depending on genetic background. In AX category, *: increased anxiety. In Ref, MGI: unpublished, mouse genome informatics submission.

1.2.2 Male vs. female Mecp2-mutant mice

RTT predominantly affects females, so female heterozygous Mecp2-mutant mice are a clinically relevant model of the disease. However, female heterozygous Mecp2-mutant mice display phenotypic variation due to XCI skewing, making it difficult to discern which phenotypes arise through cell autonomous versus non-autonomous pathways. Therefore, initial investigations are often performed in hemizygous male Mecp2-null mice due to their consistent phenotype and complete absence of MECP2 protein, which is advantageous for mechanistic research.

Hemizygous male Mecp2-null mice are phenotypically normal until 3 - 4 weeks of age when they develop hind limb clasping, tremors, breathing irregularities, loss of muscle tone and hypoactivity (134,135). These mice experience a rapid phenotypic regression and die between 6 - 12 weeks of age. Female mice heterozygous for Mecp2 deletion develop the same features at 4-6 months and typically live a normal lifespan.

1.2.3 Similarities between Mecp2-mutant mice and RTT patients

Mecp2-mutant mice recapitulate a broad spectrum of phenotypes seen in human RTT patients (Figure 1.10). Both exhibit a reduced brain volume and neuronal hypotrophy (154). RTT patients and Mecp2-mutant mouse models experience motor abnormalities, including reduced mobility, impaired motor coordination, ataxic gait, and tremors (5). While human patients replace purposeful hand use with stereotypic movements, mouse models clasp their hindlimbs (134,135), which may reflect an analogous phenotype although it is not specific to RTT mouse models (155).

RTT patients experience a neurological regression and loss of speech, two traits which cannot be measured in mouse models. However, learning deficits are a shared symptom. While human RTT patients typically exhibit increased anxiety and social avoidance (156), Mecp2-mutant mice are less anxious and more social (130,157,158). Finally, both experience breathing irregularities (159), cardiac abnormalities, (160,161), seizures, and metabolic disturbances including abnormalities in neurometabolites, increased serum cholesterol, and increased oxidative stress (162). Overall, Mecp2-mutant mice exhibit a broad range of phenotypes that recapitulate symptoms of human RTT patients, making them excellent models to study this disorder.

Figure 1.10: Similarities between human RTT patients and Mecp2-mutant mice. RTT patients exhibit many symptoms that are shared with Mecp2-mutant mice (middle circle). Symptoms unique to humans are shown in the left circle, and phenotypes seen in Mecp2-mutant mice that have not been shown in humans are in the right circle. Symptoms are color-coded based on whether they affect motor function, morphology, behavior, metabolism, or other systems.

1.2.4 Mecp2-mutant mice with human RTT-causing mutations

Despite their face validity, null mouse models do not always represent human cases molecularly since many RTT patients carry missense mutations resulting in a less-efficient or unstable MECP2 protein rather than a complete loss of it. To circumvent this caveat, several mouse lines with point mutations and deletions in MECP2 have been engineered to recapitulate clinically relevant mutations observed in human patients (Table 1.4). Of the eight most common missense mutations in RTT patients, six have an established mouse model which recapitulates symptoms of the disease. These mice have been instrumental in understanding the functional consequences of individual RTT-causing mutations and their correlation to symptom severity. For example, RTT patients carrying MECP2 T158M mutations display severe RTT symptoms, while motor function and speech are preserved in patients with a R133C mutation (67). Consistently, studies in mice determined that while both mutations create MECP2 proteins with reduced DNA-binding ability, the T158M mutation also leads to protein instability, likely resulting in its more severe clinical presentation (138).

Further, mice with Mecp2 mutations have been designed to study specific biological questions of functional relevance. For example, deletion of MECP2_e1, but not MECP2_e2 leads to RTT-like phenotypes and a shortened lifespan in mice, indicating that MECP2_e1 is more functionally relevant in RTT (47,48). PTMs have also been modeled in mice, including mutating phosphorylation sites serine 80 (S80) and 421 (S421). Mutating these residues prevents their neuronal activity-dependent phosphorylation in mice indicating their importance for MECP2- dependent neuronal gene expression (147).

1.2.5 Temporal deletions of Mecp2

One hypothesis in the field is that specific symptoms of RTT are the cumulative result of loss of Mecp2 within specific cell types. Cell-type specific deletion of Mecp2 supports this hypothesis. Conditional-ready alleles were constructed by engineering loxP sites flanking a portion of the Mecp2 gene (134,135). The resulting ‘floxed’ mice can be crossed with transgenic mice carrying a tissue- or cell-type specific Cre to achieve spatial or temporal deletion of Mecp2. Studies using these mice have greatly informed Mecp2 function both within the CNS and in peripheral tissues. For example, mice with a conditional deletion of Mecp2 in forebrain postmitotic neurons were generated by crossing Mecp2-floxed mice with mice expressing Cre under control of the CamK promoter. These mice have a mild, delayed phenotype including elevated anxiety and motor deficits, indicating that these symptoms stem from forebrain neurons. Crossing Mecp2-floxed mice to mice with a glial specific GFAP- Cre transgene achieved targeted deletion of Mecp2 in

34 astrocytes. These mice had decreased body weight, hindlimb clasping, and irregular breathing, demonstrating that glial Mecp2 deficiency could produce distinct RTT symptoms (163). Additionally, liver-targeted deletion of Mecp2 was achieved by crossing Mecp2-floxed mice to mice expressing Cre under the control of an Albumin promoter (Alb-Cre). The resulting mice develop fatty liver disease and dyslipidemia, implicating MECP2 in the regulation of lipid metabolism in non-CNS tissues (164). Since transgenic mice have been engineered to express Cre from many diverse cell types, continued studies will achieve a thorough understanding of Mecp2’s role across tissues and cell types.

1.2.6 Symptom reversal and suppression in Mecp2-mutant mice

A significant question in RTT concerns phenotypic reversibility: To date, neuronal death has not been evidenced in RTT patients or Mecp2-mutant mice. Thus, it is possible that Mecp2-deficient cells could be repaired. In a milestone study, mice were engineered with a transcriptional STOP cassette flanked by loxP sites within the endogenous Mecp2 gene (149). These ‘FloxedStop’ mice were crossed with mice expressing a ubiquitous Cre-ER transgene allowing for Mecp2 to be silenced until Cre activation with tamoxifen. FloxedStop male mice were identical to Mecp2-null mice. Following phenotype onset, tamoxifen administration removed the loxP-STOP-loxP cassette and restored Mecp2 expression to 80% of normal levels. Remarkably, this reversed neurological symptoms and normalized lifespan in the mice (149). This study was key in demonstrating that Mecp2-induced pathology is not permanent, and that symptom reversal may be possible in RTT patients.

Given MECP2’s high expression levels, its global DNA-binding affinity, and its multitude of proposed binding partners and functions, it is unsurprising that Mecp2-deficiency results in a myriad of dysregulated pathways. MECP2 has been implicated in neuronal maintenance (78,165), glial cell function (74), neurotransmitter signaling (139), growth factor signaling (166), lipid metabolism (167), oxidative stress (168), and more. Current preclinical research for RTT treatment follows two approaches: targeting misregulated pathways downstream of Mecp2 mutation or directly targeting the Mecp2 gene and its protein product (Figure 1.11). Both approaches are discussed below.

Figure 1.11: Preclinical treatments for RTT. Treatment development is focused on targeting misregulated pathways downstream of MECP2 mutation (blue) or directly targeting MECP2 and its protein product (red). The former approach includes targeting neurotransmitter signaling, growth factors, metabolism, and other non- pharmacological approaches (green and teal). Studies on the later approach are optimizing readthrough compounds, X-chromosome reactivation, and gene therapy (pink and purple).

1.2.6.1 Preclinical treatment of Mecp2-mutant mice

A significant question in RTT concerns phenotypic reversibility: Since RTT patients Identifying components that lie in pathways downstream of MECP2 has a high potential for safe and efficacious treatment as already-approved pharmacological treatments can be repurposed to treat RTT (169). However, this strategy also comes with several challenges. First, identifying the pathways MECP2 regulates, especially those which may be subtly changed in the absence of MECP2, is difficult. Further, DNA methylation signals are highly cell-specific and the MECP2’s target loci will likely vary across different cell types (170). Consistently, a recent study found that gene misregulation in neuronal subtypes of Mecp2-mutant mice is highly dependent on cell type- specific epigenetic marks (171). Finally, it is difficult to discern which pathways are misregulated due to MECP2 loss and which are secondary effects further downstream that will have less value as therapeutic targets. Nevertheless, advances in molecular biology techniques are making it easier to circumvent these issues. For example, single-cell RNA sequencing can identify genes misregulated in individual Mecp2-deficient cells and ChIP-sequencing could help identify MECP2’s direct transcriptional targets.

Preclinical treatments tested for RTT include targets in neurotransmitter signaling, growth factor signaling, and metabolism pathways, though other non-pharmacological treatments have also been tested (Table 1.5). Notably, a number of these treatments have moved to clinical trials, including desipramine, a norepinephrine reuptake inhibitor, sarizotan, a serotonin 1a agonist and dopamine D2-like receptor, and ketamine, an NMDA receptor (172–175). Dysfunctions in various neurotransmitter pathways have been observed in Mecp2-mutant mice and these treatments aim to correct these deficits (175–183). Brain-derived neurotrophic factor (BDNF) is a direct target of MECP2 and is expressed at lower levels in the brain of Mecp2-mutant mice and RTT patients (184,185). Fingolimod, insulin-like growth factor-1 (IGF), and NNZ-2566 (Trofinetide), which target the growth factor BDNF, have been tested in clinical trials, with the latter currently in phase 3 (186–189). Finally, as lipid metabolism and mitochondrial activity are both altered in Mecp2- mutant mice and RTT patients, lipid-lowering statins and the anaplerotic substance triheptanoin are currently in clinical trials (16,167,190,191).

Table 1.5: Preclinical treatments targeting pathways downstream of Mecp2 mutation. Treatment strategies are divided into categories based on pathways targeted. Clinical Treatment Mechanism Mouse Model Lifespan Phenotype Ref Trial Neurotransmitter Signaling

NO-711 GABA reuptake blocker Mecp2tm1.1Bird NT Reduced apneas (192) - Benzodiazepine GABA reuptake blocker Mecp2tm1.1Bird NT Reduced apneas (192) - diazepam L-838417 GABA reuptake blocker Mecp2tm1.1Bird NT Reduced apneas (192) - Tiagabine GABA reuptake blocker Mecp2tm1.1Bird I No improvement (193) - Improved motor function, social behavior, and THIP GABA receptor agonist Mecp2tm1.1Bird I (194) - reduced apneas Improved overall health, neuronal morphology, Mirtazapine GABA release promoter Mecp2tm1.1Bird NT (195) - dendritic spine number and anxiety Valproate GABA enhancement Mecp2tm1.1Bird NT Reduced apneas (196) - 8-OH-DPAT Serotonin 1a agonist Mecp2tm1.1Bird NT Reduced apneas (192) - tm1.1Bird Citalopram Serotonin reuptake blocker Mecp2 NT Improved sensitivity to CO2 exposure (197) - F15599 Serotonin 1a agonist Mecp2tm1.1Bird NT Reduced apneas and breathing irregularities (198) - Mecp2tm1.1Bird Serotonin 1a agonist & Sarizotan Mecp2tm1.1Jae NT Reduced apneas and breathing irregularities (174) Y dopamine D2-like receptor Mecp2tm1.1Jtc LP-211 Serotonin 7 receptor agonist Mecp2tm1Hzo NT Improved overall health, memory, and anxiety (199) - Levodopa Dopaminergic stimulation Mecp2tm1.1Bird I Improved motor activity (200) - mGlu5 positive allosteric VUO462807 Mecp2tm1.1Bird N Improved motor function and limb clasping (201) - modulator mGluR5 negative allosteric CTEP Mecp2tm1.1Bird I Reduced apneas and improved memory (202) - modulator mGlu7 positive allosteric VU0422288 Mecp2tm1.1Bird NT Reduced apneas and improved memory (203) - modulator D-JHU29 Glutaminase inhibitor Mecp2tm1.1Bird NT Improved motor activity (204) Acetyl-L-carnitine Acetyl group donor Mecp2tm1.1Jae NT Improved motor activity, memory, and weight (141) - Dendrimer-conjugated N-acetyl D-NAC Mecp2tm1.1Bird NT Improved overall health and limb clasping (205) - cysteine Ketamine NMDA receptor antagonist Mecp2tm1.1Jae NT Improved startle response (206) Y Ketamine NMDA receptor antagonist Mecp2tm1.1Bird I Improved limb clasping and motor coordination (183) Y D-cycloserine D-alanine analog Mecp2tm1.1Jae NT No improvement (207) -

Norepinephrine reuptake Desipramine Mecp2tm1.1Bird I Improved breathing irregularities (172) Y inhibitor Norepinephrine reuptake Desipramine Mecp2tm1.1Bird I Reduced apneas (208) Y inhibitor Improved motor coordination and breathing Clenbuterol B2-adrenergic receptor agonist Mecp2tm1.1Bird I (209) - irregularities Growth Factor Signaling CX546 Ampakine (BDNF) Mecp2tm1.1Jae NT Improved breathing irregularities (210) - LM22A-4 TrkB agonist (BDNF) Mecp2tm1.1Jae NT Improved breathing irregularities (211) - Sphingosine-1 phosphate Fingolimod Mecp2tm1.1Bird NT Improved motor activity (187) Y receptor (BDNF) Improved motor activity and breathing 7,8-DHF TrKB agonist (BDNF) Mecp2tm1.1Jae I (212) - irregularities LM22A-4 TrkB agonist (BDNF) Mecp2tm1.1Jae NT Reduced apneas (185) - Reduced limb claspin and improved motor CPT157633 PTP1B inhibitor (BDNF) Mecp2tm1.1Bird I (213) - activity LM22A-4 TrkB agonist (BDNF) Mecp2tm1.1Jae NT Improved memory (214) - I Improved motor activity, breathing irregularities IGF-1 IGF-1 Mecp2tm1.1Jae (188) Y and increased brain size PEG-IGF-1 Slow release IGF-1 Mecp2tm1.1Bird I No improvement (215) - I Improved motor activity, breathing RhIGF01 Recombinant human IGF1-1 Mecp2tm1.1Bird (216) Y irregularities, social behavior, and anxiety Metabolism Choline ACh Mecp2tm1.1Jae NT Improved motor coordination and activity (217) - Choline ACh Mecp2tm1Hzo NT Improved motor activity (218) - Acetyl-L-carnitine Acetyl group donor Mecp2tm1.1Jae NT Improved motor activity and memory (219) - Improved motor coordination, anxiety, and Choline ACh Mecp2tm1.1Bird NT (220) - social behavior Diet Restriction Caloric deficit Mecp2tm1Hzo NT Improved motor activity and anxiety (221) - Mecp2tm1.1Bird I Improved overall health, motor coordination, Statins Cholesterol-lowering medication (167) Y Mecp2tm1.1Jae motor activity and serum and liver lipids Dietary I Improved motor coordination, social behavior, Anaplerotic Mecp2tm1.1Jae (191) Y Triheptanoin insulin sensitivity and metabolic homeostasis Vitamin D supplementation and I Vitamin D Mecp2tm1.1Bird NT (222) - NF-κB inhibition Normalized blood glucose levels and improved Trolox Vitamin E derivative Mecp2tm1.1Bird NT (223) - response to hypoxia Corticosterone Glucocorticoid activation Mecp2tm1.1Bird D No, worsened motor activity (224) - Delayed progression of symptoms and RU486 Glucocorticoid repression Mecp2tm1.1Bird NI (224) - improved motor activity

Corticosterone Glucocorticoid activation Mecp2tm1Hzo NT Improved motor activity (199) - Curcumin Anti-oxidant, anti-inflammatory Mecp2tm1.1Jae NT NT (225) - Insulin Glucose signaling Mecp2tm1.1Bird D No improvement (215) - Target mitochondrial activity and Metformin Mecp2tm1Hzo NT No improvement (226) - oxidative stress Other Zoledronic acid Anti-osteoclastic Mecp2tm1.1Bird NT Increased bone volume and connective density (227) - Improved overall health, motor activity and Cannabidivarin Phytocannibinoid Mecp2tm1Hzo NT (228) - social behavior CNF1 RhoGTPase Mecp2tm1Hzo NT Improved motor activity (229) - Improved mitochondrial dysfunction and CNF1 Rho GTPase Mecp2tm1Hzo NT (230) - memory

Block neuron-microglia tm1.1Bird Improved body weight, overall health, motor CX3CR1 Mecp2 I (231) - interactions coordination and breathing Improved hindlimb clasping, motor Blarcamesine Sigma-1 receptor activator Mecp2tm1.1Bird NT (232) Y coordination and acoustic response Brilliant Blue G P2X7R blocker Mecp2tm1Hzo NT Improved social behavior (233) - KW-2449 KCC2 enhancer Mecp2tm1.1Bird NT Improved locomotion and breathing (234) - Non-Pharmacological Enriched Environment Mecp2tm1Pplt NT Improved motor coordination (235) - Environment Enriched Environment Mecp2tm1.1Jae NT Improved motor activity (236) - Environment Enriched Environment Mecp2tm1.1Jae NI Improved motor coordination and activity (237) - Environment Enriched Environment Mecp2tm1Pplt NT Reduced anxiety (238) - Environment Forniceal deep Neural circuit stimulation Mecp2tm1.1Bird NT Improved memory (239) - brain stimulation Bone Marrow Reduced apneas, improved breathing Brain microglia repopulation Mecp2tm1.1Bird I (76) - Transplantation irregularities and improved motor activity

In lifespan column, I: increase, NI: no increase, D: decrease, NT: not tested. In clinical trial column, Y: yes, –: not started to date.

1.2.6.2 Treatments for RTT directly targeting MECP2

Directly restoring MECP2 gene or protein function is an attractive therapeutic strategy as all of its downstream pathways could also potentially recover. Current approaches include inducing read- through of nonsense mutations, reactivating the silent X chromosome, and gene therapy (Figure 1.12). However, the largest concern of these approaches is dosage: while MECP2 deficiency causes RTT, an abundance of MECP2 causes MECP2 duplication syndrome (126). Some studies have found that 1.6x normal levels of MECP2 cause behavioral impairments in mice (240) while others find that 2.4x normal levels of MECP2 can be tolerated (153). Despite this, slight deviations in MECP2 levels in humans have been linked to autism and other psychiatric conditions. Thus, extreme caution must be taken to provide enough MECP2 per cell to impart a therapeutic benefit while reducing the risk of overexpression.

The use of small molecules known as translational read-through-inducing drugs (TRIDs) is also an attractive treatment strategy for 35-40% of RTT patients who have nonsense mutations in MECP2 (64). TRIDs, such as the aminoglycoside antibiotic gentamicin, permit read-through of premature stop codons by binding to ribosomes and impairing their codon/anticodon recognition, allowing for the insertion of an amino acid in place of a stop codon (Figure 1.12A) (241). This switches the nonsense mutation to a missense mutation, which may impart less of a functional consequence on the final protein. Importantly, this strategy would increase MECP2 in cells with the nonsense mutation, while leaving cells with normal MECP2 unaffected. However, because TRIDs would create missense mutations in MECP2 rather than a normal protein, treatment may only modestly improve symptoms. An additional concern is nonsense mediated decay (NMD), a cellular process facilitating degradation of mRNA transcripts with premature stop codons as their translation could lead to deleterious gain-of-function or dominant-negative effects (242). In mammalian cells, when a premature stop codon is located greater than 50-55 nucleotides away from an exon-exon junction, NMD is activated (243). In vitro studies have shown that the read- through efficiency of gentamicin depends heavily on the context of the stop mutation; gentamicin restored 22% of function MECP2 in cells with R294X mutations and only 10% in cells with R168X mutations (244). In some RTT patients, NMD could naturally reduce the number of transcripts available for TRIDs (245). Best responders to TRIDs will be patients with mutations less susceptible to NMD. Novel TRIDs may impart a higher read-through activity while maintaining a low toxicity (143,244). Ataluren, an oxadiazole, is one such example as it has TRID functions with minimal off-target effects (246). Ataluren restored 20% of normal functional protein in mouse models of DMD and CF, but did not produce enough protein to see a significant benefit in patients

(247,248). Ataluren may be a worthwhile avenue to pursue for RTT since even small increases in MECP2 are associated with improved symptoms.

Through X chromosome inactivation (XCI), mammalian cells silence one X chromosome in each cell, allowing for dosage compensation of X-linked genes (249). XCI is initiated by the long noncoding RNA Xist, which coats the inactive X chromosome (Xi) from which it is produced. On the active X (Xa), Tsix RNA blocks Xist expression. Therefore, in RTT patient cells where the mutated MECP2 is on Xa, a normal copy of MECP2 lies dormant on Xi. The silent state of Xi is temporarily reversed during stem cell remodeling and key molecular players involved in this process could be repurposed to facilitate re-expression of Xi genes (Figure 1.12B). This is an attractive strategy for RTT because patient cells could use their own regulatory functions to control MECP2 expression (250). Ideally, these strategies will aim to target MECP2 alone or its local vicinity on Xi as reactivating the entire chromosome may lead to a pathological level of expression of other loci. However, the Xi can be difficult to reactivate due to multiple mechanisms of epigenetic silencing and attempting to disengage these processes could disrupt epigenetic patterns throughout the genome, which could cause adverse long-term effects. Additionally, because RTT patients are heterozygous mosaics for MECP2 mutation, approximately half of their cells already express a normal copy of MECP2. In patients carrying mutations that only reduce the efficiency of MECP2 rather than destroy its function, Xi reactivation will lead to anywhere from 1-2x the normal amount of MECP2 in some cells, which could lead to toxicity of overexpression. Therefore, Xi reactivation may only be a practical treatment option for patients with loss-of- function MECP2 mutations. Ongoing screens are aimed at identifying new molecules to induce Xi reactivation that can be optimized for human treatment (249,251).

A final potential avenue for RTT treatment involves introducing a normal copy of MECP2 into cells by gene therapy (Figure 1.12C). For RTT, gene therapy approaches must utilize an appropriate vector able to cross the blood-brain-barrier (BBB) and transduce many cells in the CNS. One concern with gene therapy is that high viral titers are often needed to infect a large proportion of cells, and furthermore, that cells which inadvertently receive more than one viral particle would be overburdened. To circumvent this, an siRNA could be supplied to suppress endogenous MECP2 so that only the transgenic MECP2 is expressed (252). However, this would require transgenic MECP2 to possess a seamless promoter to completely mimic endogenous gene expression.

Recombinant adeno-associated virus (AAV) vectors have been used in preclinical gene therapy studies based on their ability to cross the BBB, infect neurons and mediate stable long-term expression of a transgene without inflammation or toxicity (253,254). AAV has a 4.7kb ssDNA genome from which 4.4kb of the viral DNA can be removed and replaced with a human transgene.

Self-complementary (sc) AAV vectors show a 10-100 fold higher transduction efficiency, but their packaging capacity is cut in half to only 2.2kb, making it difficult to package large genes (255). An AAV9-MECP2, injected into the brain of neonatal Mecp2-null mice, produced a transduction efficiency of 7-42% of cells, with the highest infection rate in the hypothalamus and lowest in the striatum (256). However, this low efficiency was sufficient to increase the lifespan of male mice to 16.6 weeks. In contrast, scAAV9-MECP2 under the control of a truncated Mecp2 promoter injected systemically had a very low transduction efficiency in the brain (~2-4%). However, systemic treatment preferentially targeted spleen and liver cells with some receiving 10 copies of the vector, leading to liver damage. A second generation AAV9 vector with a modified 3’UTR and a panel of miRNA binding sites was developed with the goal of biasing transgene expression away from the liver (257). This vector, injected in the cisterna magna, was better tolerated and improved the lifespan of Mecp2 mutant mice, but could not improve behavioral traits without being used at a very high dose (258). Finally, neonatal mice were injected intracranially with an scAAV9 vector encoding a minimal-MECP2 protein lacking all amino acids except those encoding the MBD and NID (108). These mice showed symptom improvement and increased survival, providing a new avenue to pursue in gene therapy studies as a small yet functional MECP2 protein creates room for additional regulatory sequences to be packaged into the limited capacity of scAAV9 vectors. Importantly, this could allow for more precise temporal control of Mecp2 expression. However, a remaining obstacle is scaling dosage for humans. Due to their small size, mice cannot reliably inform effective dosing for clinical applications. Further, gene therapy in Mecp2-null mice is more effective in neonatal mice with smaller brains and less formed neuronal networks; gene therapy must be optimized in older mice and other animal models to even be considered worth pursuing in RTT patients.

Nonsense mutation Mutant MECP2 active Normal MECP2 active

E P A Xi Xa Xa Xa Truncated protein X reactivation CAA TOP 5 mRNA ...AUG UGG UAA GGC ...

Xa Xi Xa Xi Cells treated with TRIDs

Normal MECP2 active Normal MECP2 active E P A Full length protein CAA with missense mutation 5 mRNA ...AUG UGG UAA GGC ...

C Cells with Cells treated with MECP2 deficiency gene therapy

Little to no functional MECP2 protein produced by MECP2 protein exogenous MECP2 gene

Figure 1.12: Treatment strategies aimed at targeting MECP2 directly. A. Translational read-through of nonsense mutations. Nonsense mutations result in a premature STOP codon and a truncated protein product. TRIDs alter ribosomal codon/anticodon recognition allowing an amino acid insertion at the site of a STOP codon. This leads to a full-length protein with a missense mutation which may have less functional consequences. TRID: translational read-through inducing drug. B. Reactivation of the silent X chromosome. Normal cells (blue) express MECP2 from their active X (Xa) while mutated MECP2 is on the inactive X (Xi). Mutant cells (pink) express mutated MECP2 (yellow) from their Xa. Epigenetic reprogramming allows reactivation of silenced X chromosomes making both X chromosomes active. All cells express both normal and mutated MECP2. C. Gene therapy. Cells with MECP2 mutations express little to no functional MECP2 protein. Viral vectors packaged with a functional copy of MECP2 can be administered to cells. Functional MECP2 is transcribed in the nucleus and leads to normal protein production.

1.3 Regulation of breathing

Oxygen is required for cellular respiration, a set of metabolic reactions that convert chemical energy and nutrients into energy. Oxygen passage begins upon inhalation into the nose (Figure 1.13A). Inhaled air then enters the pharynx and continues into the trachea which splits into the two mainstem bronchi (left and right). Simultaneously, the diaphragm and chest muscles contract, pulling the chest cavity open. In contrast, during an exhale, these muscles relax, allowing the lungs to spring back to their normal size. Air then enters the right lung, which has three lobes (upper, middle, and lower) or the left lung, which has two lobes (upper and lower).

The mainstem bronchi divide into smaller bronchi called conducting bronchi. Conducting bronchioles continue to get narrower as they branch, becoming ‘terminal bronchioles’, followed by ‘respiratory bronchioles’. Respiratory bronchioles contain pouches of alveoli, of which there are over 5 million in the lung. Next to alveoli are endothelial cells that comprise the blood-gas barrier, the site of oxygen exchange. Here, carbon dioxide (CO2) diffuses out of deoxygenated blood into the air of the alveoli, which then gets breathed out. Simultaneously, oxygen (O2) from the alveoli diffuses into the blood which travels to the heart, and then is distributed to the body’s tissues (Figure 1.13B).

Figure 1.13: Overview of respiration. A. Schematic of respiration. 1) Air is inhaled through the nose or mouth and 2) enters the pharynx and continues into the trachea, which splits into two mainstem bronchi. 3) Chest muscles contract, pulling the chest cavity open and 4) the diaphragm contracts, pulling the ribs open, allowing for 5) the lungs to expand and fill with air. During expiration, all processes occur in reverse. B. Inhalation brings air to the alveoli, the most distal part of the lung. Here, the alveolar epithelium is near capillary endothelium. Oxygen (O2) from inhaled air diffuses from the alveoli into the pulmonary capillaries, where it can be distributed to the body’s tissues. imultaneously, carbon dioxide (CO2) diffuses from the capillaries into the alveoli where it can be expelled from the body by expiration.

1.3.1 Autonomic control of respiratory rhythm

Three main anatomical regions of the brainstem generate respiratory rhythm: the dorsal respiratory group (DRG), the ventral respiratory group (VRG), and the pontine respiratory group (PRG) (Figure 1.14) (259). Other brain regions can affect respiration, but they are not essential to the respiratory network as their removal does not impair respiratory rhythm (260). Respiration is produced by the autonomic contraction of respiratory muscles that are controlled by three types of respiratory nerves: the phrenic, vagus and the posterior thoracic nerves. The phrenic nerves pass through the heart and stimulate the activity of the diaphragm, the vagus nerve innervates the diaphragm, larynx and pharynx, and the posterior thoracic nerves stimulate the intercostal muscles across the thorax and abdomen.

The DRG and VRG are located in the medulla, whereas the PRG is in the pons. The DRG resides within the nucleus tractus solitarius (NTS), an important integration center for respiratory-related sensory inputs arriving from the PRG as well as from two cranial nerves, the vagus nerve and the glossopharyngeal nerve (259). Most sensory inputs to the NTS use glutamate as a neurotransmitter, while hypoxia activates the tyrosine hydroxylase (TH) positive neurons (261) of the NTS. The DRG within the NTS is responsible for initiating inspiration. It is composed of a collection of neurons that extend along the full length of the medulla, whose firing sets and maintains a steady respiratory rhythm.

The VRG is divided into caudal and rostral sections. The caudal VRG contains expiratory neurons that project to spinal thoracic and lumbar expiratory motor neurons. The rostral VRG primarily contains inspiratory neurons, whose activity initiates inspiration, that project to the intercostal motor neurons, initiating chest expansion (262). Above the VRG lies the Bötzinger complex (BötC) which contains expiratory neurons. The BötC is thought to play a basic role in expiratory pattern generation through its expiratory inhibitory neurons (263). Caudal to the Bötzinger complex is the pre-Bötzinger complex (preBötC), a cluster of interneurons that are essential for regulating respiratory rhythm. Approximately 20% of neurons in the preBötC are ‘pacemakers’ which burst autonomously at a fixed period and burst duration. The preBötC neurons fire rhythmically on their own in vitro while in vivo deletion of this region cause severe disruptions in breathing (264). The cellular composition of the preBötC is complex and includes populations of GABAergic and glycinergic inhibitory neurons that maintain respiratory rhythm (263).

The PRG, located in the pons, includes neuronal populations in the Kölliker-Fuse (KF) nucleus and the parabrachial (PB) complex. The PRG receives signals from the NTS and sends inputs to the medulla, amygdala and hypothalamus (265). While the precise role of the PRG in breathing remains unclear, eliminating the connection between the pons and medulla leads to apneustic

47 breathing (266). Based on its connectivity, pontine circuits are hypothesized to integrate sensorimotor information and coordinate respiratory responses (265).

Despite autonomic regulation of breathing, environmental feedback is essential for maintaining respiratory homeostasis through adjustments in ventilation. The well-characterized CNS chemoreceptor, the retrotrapezoid nucleus (RTN), detects brain CO2 levels and sends excitatory signals to the VRG to increase inspiration (267). The RTN also receives input from peripheral chemoreceptors in the body, such as the aortic body and carotid bodies, which detect changes in blood CO2 and pH. Thus, the RTN is considered a major chemosensory region that adapts to different metabolic states by modulating the activity of the respiratory network (267).

The motor cortex within the cerebral cortex controls voluntary respiration, such as for exercise and speech, by sending signals to the spinal cord to alter breathing. During periods of pain or emotional stress, the hypothalamus can also override normal respiration (268). Additionally, mechanoreceptors regulate breathing: The Hering-Breuer inflation reflex prevents over-inflation of the lungs. Pulmonary stretch receptors on bronchial walls respond to excessive stretching, sending action potentials through the vagus nerve to the pons to inhibit inspiration. Finally, the cough reflex is induced by pulmonary irritant receptors that respond to a variety of allergens and chemicals, initiating coughing and bronchoconstriction (268).

Figure 1.14: Neuronal respiratory centers control breathing. The neuronal respiratory center is in the pons and medulla. The PRG includes the Kölliker-Fuse nucleus and the parabrachial complex and is involved in sensorimotor regulation of breathing. The VRG includes the Bötzinger complex and pre Bötzinger complex which are essential in regulating expiration and maintaining respiratory rhythm, respectively. The DRG includes the nucleus tractus solitarius and is responsible for initiating inspiration and regulating sensory input.

1.3.2 Lung development

Lung development occurs over five stages (Figure 1.15) (269,270). The first, embryonic stage, begins during the third week of gestation in humans and embryonic day 9 (E9) in mice. During this stage, the anterior foregut separates to form the esophagus dorsally and the trachea ventrally (271). The trachea then bifurcates into left and right primary lung buds. Forkhead box A2 (Foxa2) expression is crucial for this process. Around week five in humans, the pseudoglandular stage of development takes place. Pseudoglandular development involves branching morphogenesis, a program of stereotypic divisions to form a tree-like network of airways (272). The two lung buds formed at this time become left and right main stem bronchi, which further divide into secondary bronchi, and subsequently into bronchioles. This process requires reciprocal signaling between the forming lung epithelium and surrounding mesenchyme and is mediated by Wnt, Sonic hedgehog, Bmp, Fgf, and retinoic acid signaling. At week 16, the canalicular stage begins, which is characterized by the development of gas exchange regions. During this stage, existing airways increase in size and mesenchymal vascularization is enhanced. In the saccular stage, beginning at week 26, the lung’s connective tissue grows and thins out and surfactant machinery develops. Saccules at the end of the airways increase in size and are wrapped by a capillary bilayer. Finally, the last stage of alveolarization and microvascular maturation begins in utero in humans but is entirely postnatal in mice. At this time, septae grow from the saccular walls to divide saccules into alveoli, increasing the surface area for gas exchange. In parallel, microvascular maturation occurs and capillaries are fused into a single-layer network so that each capillary is completely surrounded by alveoli (273).

Figure 1.15: Lung development occurs over five stages. During the embryonic stage, the trachea is formed as an endodermal outgrowth of the ventral foregut and bifurcates to form the left and right primary lung buds. Branching morphogenesis begins in the pseudoglandular stage. During the canalicular stage, airways increase in size and mesenchymal vascularization occurs. In the saccular stage, saccules at the end of airways increase and are wrapped by a capillary bilayer. During alveolarization and microvascular maturation, septation of saccules into alveoli occurs to increase the gas exchange surface area, and capillaries fuse into a single-layer network to surround alveoli. W: weeks, E: embryonic day, P: postnatal day.

1.3.3 Cellular composition of the lung

The cellular composition of the lung is complex and varies by region, as each area is specialized to perform distinct functions. Below, the cellular composition of the trachea and bronchi, bronchioles, and alveoli is discussed (Figure 1.16).

1.3.3.1 Trachea and bronchi

The most proximal region of the lung, the tracheobronchial airways, are lined by pseudostratified columnar epithelium allowing each cell to contact the basement membrane. Basal cells comprise up to 30% of the cell population in this region (274,275). Basal cells have adhesive and junctional properties, and are thought to play a structural role in the lung. They also have stem-like properties and can differentiate into other airway epithelial cells upon injury (276). Ciliated cells are a type of columnar epithelial cell and are the most common cell type in this region, making up to 80% of the cell population. Each ciliated cell has approximately 200 constantly beating cilia that drive mucociliary clearance, a process by which inhaled microorganisms and foreign particulates are cleared from the lung in mucus made in the conducting airways (277). Club cells, formerly called Clara cells, are the major secretory cells in the lung. Club cells are dome-shaped cells with short microvilli. They help maintain airway homeostasis by detoxifying xenobiotic and oxidizing substances through cytochrome P-450, monooxygenases and flavin-containing monooxygenases. They also secrete products including uteroglobin and lysozymes which aid in mucociliary clearance (278). In tracheobronchial airways, pulmonary neuroendocrine cells (PNECs) are found individually. PNECs sense stimuli in the airways, such as nicotine and hypoxia, and mediate pulmonary blood flow or modulate the immune response by secreting neurotransmitters such as serotonin, calcitonin and bombesin (279). Finally, a small number of dendritic cells are present and participate in the adaptive immune response against foreign particles and lung infection.

1.3.3.2 Bronchioles (conducting airways)

The bronchiole is made up of much of the same cell types as the trachea and bronchi, though it is characterized by a simple columnar epithelium. Basal cells are present in human conducting airways, but are confined to the trachea and bronchi in mice (280). Ciliated cells, club cells, and dendritic cells have the same function in this region as in the bronchi. However, in bronchioles, PNECs are found in clusters of up to 30 cells called neuroepithelial bodies (NEBs), though their function in groups is poorly understood (279). Unique to this region are goblet cells, cells that secrete mucus glycoproteins (mucins) which facilitate mucociliary clearance (281). Mucins

52 physically entrap inhaled irritants and microorganisms. Ciliated cells then move the mucus back toward the bronchi and trachea, and finally the pharynx for removal by coughing or sneezing. Smooth muscle underlies the airways, the precise role of which remains unclear (282).

1.3.3.3 Alveoli

Alveoli are unique structures in the lung in that their surface area is maximized for optimal gas exchange. The alveolar epithelial consists of alveolar epithelial type 1 (AE1) and type 2 (AE2) cells. Squamous AE1 cells comprise 95% of the surface area of the alveoli, though they represent a minor proportion of the entire cellular population. AE1 cells have a high membrane to cytoplasm ratio and are located adjacent to the lung’s capillary network, making them the primary site of gas exchange. In contrast, cuboidal AE2 cells are the primary source of pulmonary surfactant, a complex mixture of proteins and lipids that line the alveoli and maintain lung compliance. AE2 cells can self-renew and differentiate into AE1 cells upon injury. Alveolar macrophages are responsible for the degradation and clearance of small dust particles and microorganisms that enter the alveoli. They also modulate surfactant recycling and homeostasis. Finally, fibroblasts surround the alveoli and secrete extracellular matrix (ECM) proteins during lung development and injury repair (283).

Figure 1.16: Cell types in the respiratory system. The trachea and bronchi are lined by a pseudostratified epithelium of basal, ciliated, and secretory (Clara/club) cells. Bronchioles consist of a simple columnar epithelium with ciliated and Clara cells, and some mucus-producing goblet cells. Bronchiole cells overlay smooth muscle. Pulmonary neuroendocrine cells (PNECs) and dendritic cells are present in both the bronchi and bronchioles. The alveoli are lined by squamous alveolar epithelial type 1 (AE1) cells and cuboidal AE2 cells. Alveoli also contain macrophages, mesenchymal fibroblasts, and endothelial cells within capillaries.

54 1.3.4 Pulmonary surfactant

In the alveoli, AE2 cells produce and secrete pulmonary surfactant, a lipid-rich complex that forms a surface film. The amount of net pressure required to inflate alveoli is governed by surface tension and LaPlace’s law. Alveoli have a high surface tension at their air-liquid interface. LaPlace’s law states that the pressure within a sphere is inversely proportional to its radius while surface tension is constant:

2푇 푃 = 푟 where P = pressure, T = surface tension, and r = radius

Based on this law, if a sphere’s radius is reduced, its pressure increases (284). Thus, in the lung, air would preferentially enter larger alveoli due to their lower pressure, causing smaller alveoli to collapse (Figure 1.17A). However, pulmonary surfactant reduces surface tension in a manner proportional to alveolar size. This allows alveoli of all sizes to inflate, increasing lung compliance and providing an even distribution of ventilation across the lung (285). Thus, pulmonary surfactant is essential to lower the work of breathing and prevent alveolar collapse, though it also participates in immune surveillance.

1.3.4.1 Surfactant composition

Pulmonary surfactant is composed of 10% protein and 90% lipid (Figure 1.17B). The four major surfactant proteins are designated surfactant protein A (SFTPA), SFTPB, SFTPC, and SFTPD (286). SFTPA and SFTPD are large and hydrophilic, while SFTPB and SFTPC are small and hydrophobic. SFTPA and SFTPD enhance bacterial and viral clearance, and can directly bind allergens to decrease allergen-induced histamine release (287). In contrast, SFTPB is involved in lipid adsorption to the air/liquid interface, formation of tubular myelin, as well as surfactant reuptake, and its deficiency causes lethal respiratory distress (288). Finally, SFTPC is the only surfactant protein exclusively produced by AE2 cells. FTPC’s role overlaps with SFTPB, though it may also have intracellular functions in AE2 cells (289).

The abundance and species-specific composition of surfactant lipids are tightly regulated; phosphatidylcholine (PC) represents 85% of the lipids in pulmonary surfactant. Dipalmitoylphosphatidylcholine (DPPC) is the major PC species in the surfactant, making up approximately 40-60% of its normal composition. DPPC is composed of two 16-carbon saturated chains and is the strongest surface-active molecule in the lung, often credited with generating a near-zero surface tension at the air-liquid interface (290). Other PC molecules in lung surfactant are present in variable quantities and are hypothesized to regulate surfactant fluidity and aid in adsorption.

55 Phosphatidylglycerol (PG) comprises an additional 10% of the surfactant lipid pool, while small amounts of phosphatidic acid (PA), phosphatidylinositol (PI), phosphatidylethanolamine (PE), and cholesterol are also present (285). Cholesterol is thought to increase surfactant fluidity, and excess cholesterol can impair surfactant’s surface tension-reducing properties (291). While, triglycerides are also present in the lung, their role has not been well defined; they may serve as a storage form of lung lipids acting as substrates for phospholipid synthesis when required.

1.3.4.2 Surfactant secretion and recycling

Surfactant is produced in the endoplasmic reticulum of AE2 cells and transported to the Golgi for modification (292). Lamellar bodies then serve as an intracellular storage for surfactant prior to secretion into the alveolar airspace via exocytosis (Figure 1.17C). Upon encountering alveolar space, surfactant forms a complex lattice-like structure called tubular myelin. The factors controlling this switch are unknown, but calcium is required in vitro (293). Tubular myelin is transported to the air-liquid interface, where it again changes form into a film-like monolayer. Here, it performs its essential functions of surface tension reduction and immune modulation.

Pulmonary surfactant has a half-life of 5-10 hours, and approximately 10-40% of the alveolar pool is secreted from the alveolar airspace each hour (294). Surfactant clearance is thought to be occur through two routes (Figure 1.17C): AE2 cells can internalize surfactant components for recycling or degradation (295); this may be an economical method to prevent using cellular energy to synthesize new surfactant components. Alternatively, a small amount of surfactant may be internalized and degraded by alveolar macrophages. Notably, macrophage recycling of surfactant is often altered in disease states (295).

Figure 1.17: Lung surfactant is essential for breathing. A. chematic of Laplace’s law. Without surfactant (left), alveoli with a smaller radius (r) (1) have a high pressure (P) while alveoli with a larger radius (2) have a low pressure. This leads to a pressure gradient dependent on tension (T), where air within the alveolar area enters alveoli 2, causing alveoli 1 to collapse. Surfactant (right) reduces surface tension in a manner proportional to alveolar size. This equalizes the pressure in alveoli of different sizes leading to equal airflow, preventing lung collapse. B. By weight, surfactant is composed of 10% protein and 90% lipid, of which ~36% are dipalmitoylphosphatidylcholine (DPPC) , ~40% are other phosphatidylcholines (PCs), ~8% are phosphatidylglycerol (PG), and 6% are other lipids. C. Surfactant is produced in alveolar epithelial 2 (AE2) cells and stored in lamellar bodies (1). Once needed, it is secreted via exocytosis (2). Upon contact with the alveolar space, the surfactant adopts a lattice-like structure called tubular myelin (3). The surfactant is transported and adsorbed to the air-liquid interface where it becomes a monolayer film (4). Finally, surfactant molecules can be degraded via macrophages (pink cell) or taken up by AE2 cells for recycling and/or degradation (5).

57 1.3.5 The lung’s extracellular matrix

The lung has a unique extracellular matrix (ECM) that provides structural support for cells, but is also involved in lung development, homeostasis, and injury response (296). This ECM constitutes a three-dimensional, interconnected lattice of secreted proteins that form a scaffold for other cells of the lung. The core “matrisome” is made up of almost 300 proteins, including collagens, elastin, glycoproteins and proteoglycans (297). Lung ECM comprises the basement membrane, a thin, fibrous lining underlying the epithelial cells, and the interstitial spaces, the areas surrounding the blood vessels and alveoli.

Cross-linking of ECM molecules contributes to its stiffness, and thus, the lung’s structural integrity. The composition of lung ECM is highly dynamic, and is constantly remodeled during lung development and is heterogenous in different lung regions (298). Collagens are the most abundant protein; to date, 34 genes have been identified as participating in collagen formation, all having the “COL” prefix. Collagen proteins are made of chain-forming helical Gly-X-Y repeat regions where glycine is found at almost every third residue. Three collagen proteins coil to form a single collagen molecule, called tropocollagen, which binds together to make larger collagen aggregates. Fibrillar collagens, types I, II, III, V, and XI, are essential for the lung’s tensile strength and distensibility, while non-fibrillar type IV collagen is present in the basement membrane and is crucial for blood-gas barrier formation (299).

The elastin (Eln) gene encodes a soluble tropoelastin protein that is post-translationally modified, crosslinked, and organized into non-soluble elastin polymers (300). Elastin is required for the elastic recoil of the lung, and elastin fibers are found interspersed with collagen in the ECM (299). Fibronectin, the most common lung glycoprotein, mediates cell-matrix adhesion by binding collagen and cell-surface integrins. Lastly, proteoglycans, which are sulfated (heparan sulfate, chondroitin sulfate) or non-sulfated (hyaluronic acid), contribute to the mechanical stability of the collagen-elastin network (301).

Modifications in the ratios of ECM components alter their relative contribution to ECM stiffness. Additionally, turnover rates for ECM components differ; 3-10% of lung collagen is recycled daily, while elastin is generally not produced in large amounts after lung development (299). Thus, genetic defects and environmental influences that affect the biosynthesis or assembly of lung ECM can both cause respiratory disorders.

1.3.6 Respiratory disorders

A vast number of pathogenic conditions affect respiration. While there are no known breathing disorders caused by specific perturbations to the neural respiratory center, impaired breathing is a common feature of neurological conditions including Alzheimer’s disease, Parkinson’s disease,

58 frontotemporal dementia, multiple sclerosis and amyotrophic lateral sclerosis (302). Here, lung surfactant disorders, restrictive lung disorders and obstructive lung disorders are discussed. Breathing dysfunction in RTT is subsequently reviewed.

1.3.6.1 Lung surfactant disorders

Disorders that affect surfactant metabolism underly respiratory diseases, predominantly in the neonatal and pediatric populations. Pulmonary alveolar proteinosis (PAP) is the most well- characterized disorder of lung surfactant dysfunction, featuring impaired surfactant regulation and clearance, increased risk of lung infection, and respiratory failure (303). Primary PAP is a result of disrupted granulocyte-macrophage colony-stimulated factor (GM-CSF) signaling, which is crucial for surfactant homeostasis and alveolar stability. Alveolar macrophages require GM-CSF for cholesterol efflux and disruption of GM-CSF signaling impairs cholesterol, and subsequently, surfactant clearance, leading to surfactant accumulation in the lungs (304).

Congenital PAP had been reported in infants and children with mutations in surfactant production genes, surfactant-protein B (SFTPB), C (SFTPC), and ATP-binding cassette subfamily A member 3 (ABCA3). Mouse models lacking these genes develop lung disease that recapitulates symptoms seen in humans (305–307). Infants homozygous for loss-of-function SFTPB die shortly after birth due to respiratory failure (308). Over 40 missense mutations in SFTPB have been identified that are associated with abnormal surfactant phospholipid content with increased PIs and decreased PGs, and surfactant that is less effective at lowering surface-tension (309). Mutations in SFTPC have been found in patients with apneas, thickened alveolar septae, collapsed alveoli and interstitial lung disease (310). Finally, mutations in ABCA3, a gene involved in lipid transport into lamellar bodies of AE2 cells, are also fatal if inherited in a homozygous manner. Patients with heterozygous mutations experience surfactant deficiency, low PC levels, and an inability to reduce lung surface tension (311).

1.3.6.2 Restrictive lung diseases

Restrictive lung diseases are characterized by a reduced total lung capacity due to difficulty filling the lungs. In pulmonary function tests, forced vital capacity (FVC) of the lung, the total amount of air exhaled, is decreased, while the forced expiratory volume (FEV), the amount of air exhaled during a forced breath, is generally normal (Table 1.6). Restrictive lung diseases can originate in the lungs, such as interstitial lung disease (ILD), pneumonia, pneumonitis, tuberculosis, and many more. Alternatively, they can be extrinsic, such as in the case of scoliosis and obesity, or caused by neurological factors, such as in muscular dystrophy and ALS.

ILDs encompass over 200 disorders of the lung interstitium and basement membranes (312). Idiopathic pulmonary fibrosis (IPF) is the most common and most aggressive ILD and is

59 characterized by progressive lung scarring, an irreversible decline in lung function, and progressive respiratory failure. IPF is believed to be the result of aberrant wound healing and it commonly features an excessive deposition of disorganized collagen as well as ECM proteins with pathological inflammation, which may modify the architecture of the lung (313). Cigarette smoking is the best recognized risk factor for IPF, while other environmental exposures, such as to metal, wood and coal dusts, silica, mold spores, and other agricultural products, also promote IPF. However, number of genes have also been linked to IPF (314,315). Accelerated telomere shortening is also associated with abnormal tissue repair, and six-telomere related genes have been linked to IPF (TERT, TERC, DKC1, TINF2, RTEL1, PARN) (316).

1.3.6.3 Obstructive lung diseases

Obstructive lung diseases are characterized by slow, shallow exhalation, due to an obstruction within air passages. This can occur when inflammation or swelling that blocks the airways traps air in the lungs. Lungs of patients with obstructive diseases have a normal lung capacity, normal FVC, but decreased FEV as exhalation is difficult (as measured by pulmonary function tests) (Table 1.6). Obstructive lung diseases include chronic obstructive pulmonary disease (COPD), asthma, bronchiectasis, and more.

COPD, formerly called emphysema, is the fourth leading cause of death worldwide (317). The alveolar walls of COPD patients are progressively destroyed. Also, loss of elastic recoil and obstruction of small airways are common, resulting in poor airflow and air trapping in the lungs (318). COPD is thought to result from inflammatory responses to inhaled irritants; thus, smoking is the largest contributor to COPD development. However, increased oxidative stress and connective tissue digestion by proteases are also involved in the disorder. Only one gene, serpin family A member 1 (SERPINA1) has been definitively linked to COPD (319). SERPINA1 encodes alpha-1 antitrypsin (AAT), a serine protease made in the liver which is transported to the lung to function as a major inhibitor of neutrophil elastase (NE). Mutations in AAT that decrease its expression by even 15% prevent its inhibition of NE, leading to lung tissue digestion.

60 Table 1.6: Comparison between restrictive and obstructive lung disease. Total lung capacity refers to the amount of air the lung can hold during a maximum inspiratory effort. FVC refers to the total amount of air exhaled. FEV1 refers to the amount of air exhaled during a forced breath during the first second.

Restrictive Lung Disease Obstructive Lung Disease Characteristics Difficulty filling lungs Difficulty expelling air from (inhalation) lungs (exhalation) Obstruction in airways (usually associated with inflammation) Examples Intrinsic: interstitial lung COPD (emphysema), asthma, disease (ILD), idiopathic bronchiectasis, bronchiolitis, pulmonary fibrosis (IPF), cystic fibrosis pneumonia, pneumonitis

Extrinsic: scoliosis, obesity

Neurological: amyotrophic lateral sclerosis (ALS), muscular dystrophy Total Lung Capacity Decreased Normal Forced Vital Capacity Decreased Normal (FVC) Forced Expiratory Volume Normal Decreased at 1 sec (FEV1)

61 1.3.7 Breathing abnormalities in RTT

RTT breathing symptoms generally include hyperventilation with periods of apneas, breath- holding, and forced breathing (320,321). Although once thought to only occur during wakefulness, breathing abnormalities are also present during sleep, indicating that perturbed breathing is not primarily caused by cortical influences such as arousal (322,323). Breathing irregularities occur in almost all RTT patients and change over the course of disease progression; younger patients often alternate hyperventilation and apneas, while apneas are the most prominent respiratory feature in older patients (324,325). Apneas are a severe feature of RTT and often cause cyanosis and fainting (320). Additionally, up to 80% of premature death in RTT patients is due to respiratory infection, aspiration, and respiratory failure (31).

Early studies proposed brainstem immaturity as the cause of breathing symptoms in RTT patients (324). Subsequent studies performed on patient autopsy samples revealed a reduction of tyrosine hydroxylase (TH) in the brainstem (326), a functional marker for catecholaminergic neurons. Consistently, Mecp2-null mice show a progressive loss of TH-expressing neurons in the medulla, and treatment with desipramine increased the number of TH-expressing neurons and improved breathing (172,179). However, desipramine did not yield improvements in breathing in RTT patients (173). Deficits in brain-derived neurotrophic factor (BDNF) expression and monoaminergic, glutamergic, and GABAergic synaptic transmission were found in RTT patients and were speculated to have roles in breathing anomalies. In Mecp2-mutant mice, brainstem neurons express low levels of BDNF, and brainstem presynaptic GABAergic inhibition is defective while post-synaptic inhibition is reduced (327). However, targeting BDNF (185,212), serotonin (192,198), dopamine (200), or GABA (192,194,328) only transiently or modestly improved respiratory symptoms. Recently, neuronal hypoactivity in the medial prefrontal cortex (mPFC), and an altered excitatory and inhibitory balance, was linked to breathing symptoms. The selective activation of mPFC pyramidal neurons temporarily reduced apneas in Mecp2-null mice, though long term recovery was not tested (329). Thus, while several cellular abnormalities have been found in the neuronal respiratory network, the mechanisms underlying breathing dysfunction in RTT remain unclear.

Temporal deletions have been used to study breathing abnormalities in RTT in mice. A 2006 study aimed to separate the neuronal and systemic contribution of Mecp2 to RTT symptoms by comparing mice with a whole-body deletion to mice with a neuron-specific Nestin-Cre mediated deletion of Mecp2. Interestingly, female Mecp2-mutant mice had a lower breathing frequency and higher tidal volume than wild type mice, while mice with a Nestin-Cre mediated deletion of Mecp2 had an increased breathing frequency and lower tidal volume (330). Additionally, response to

62 hypoxia was more pronounced in Mecp2-mutant females than in mice lacking neuronal Mecp2, suggesting Mecp2 deficiency outside of the CNS contributes to breathing abnormalities.

Another study genetically removed Mecp2 from the brainstem and spinal cord using a Cre enzyme under the control of a HoxB1 promoter. HoxB1 is expressed in the caudal pons, medulla, and spinal cord, encompassing the neural respiratory center. Mice lacking Mecp2 in these areas had no change in breathing frequency, though their tidal volume was modestly increased and they had a heightened response to hypoxia (331). A follow up study removed Mecp2 from the caudal medulla using a HoxA4 promoter to drive Cre expression. The resulting mice had an increased breathing frequency but did not show any incidence of apneas (332). These confusing results suggest a complex cellular network, rather than a distinct brain region, is responsible for breathing abnormalities in RTT.

Despite MECP2’s ubiquitous expression, the RTT lung is relatively uncharacterized. In 2011, CT imaging was conducted on 27 RTT patients. Approximately 55% of patients had abnormal findings in their lung including centrilobular nodules, thickened bronchial walls, ground-glass opacities, and bronchiolecstasis (333), suggesting pulmonary involvement in RTT. Additionally, an epidemiological study of 320 RTT patients found that 17% reported having pneumonia, 6% asthma, 6% bronchitis, and 4% other respiratory illness (31). Recently, evidence of aspiration pneumonia was found in Mecp2-null mice, paired with an increase in inflammatory cell infiltration. A reduction in the number of AE2 cells and enlarged alveolar airspaces were also noted, suggesting emphysema-like changes in the Mecp2-null lung (334). Thus, breathing dysfunction in RTT may be caused by a combination of neuronal and pulmonary deficiencies.

63 1.4 Hypothesis

RTT is a devastating disorder caused by mutations in the ubiquitously-expressed epigenetic regulator, MECP2. Despite immense progress in understanding MECP2’s function, the precise mechanisms underlying RTT remain unknown. While RTT symptoms are likely reversible, treatment options are currently limited. Continued study of MECP2’s functions will identify novel therapeutic targets to ameliorate RTT symptoms.

Among symptoms of RTT, breathing irregularities, including hyperventilation and apneas, are particularly concerning as they lead to 80% of premature patient death. Several studies have attributing breathing symptoms to Mecp2 deficiency in the brainstem’s neurological respiratory control center. However, precise neuronal deficits have not been identified and while a few treatments targeting neurotransmission in these areas have improved breathing in Mecp2-mutant mice, none have improved symptoms in RTT patients.

Conversely, Mecp2 is highly expressed in the human and mouse lung, though its role in pulmonary function is greatly understudied. As such, I hypothesized that genetic loss of Mecp2 would result in lung defects and associated respiratory symptoms in a Mecp2-mutant mouse model of RTT. This dissertation first describes lung lipid metabolism perturbations caused by loss of NCOR1/2-complex-mediated transcriptional regulation in the absence of MECP2 (Chapter 2). The second part describes transcriptomic and structural changes in the Mecp2-null lung that lead to altered pulmonary function (Chapter 3). Finally, the third part of this thesis describes pharmacological analysis of therapeutics used to treat breathing irregularities in Mecp2-mutant mice (Chapter 4).

Chapter 2 Lung lipid defects contribute to respiratory symptoms in a Mecp2- mutant mouse model of Rett syndrome

Portions of this chapter have been submitted for publication in:

Vashi, N., Ackerley, C., Post, M. and Justice, M.J. Aberrant lung lipids cause respiratory impairment in a Mecp2-deficient mouse model of Rett syndrome.

A special thank you to our collaborator, Dr. Cameron Ackerley, for his assistance with the electron microscopy experiments presented in this chapter.

65 Lung lipid defects contribute to respiratory symptoms in a Mecp2-mutant mouse model of Rett syndrome

2.1 Abstract

Pulmonary surfactant is a lipid-rich complex of phospholipids, proteins and neutral lipids that facilitates normal gas exchange. Respiratory impairment is a prominent feature of Rett syndrome (RTT), a severe X-linked disorder caused by mutations in the gene methyl CpG-binding protein 2 (MECP2). MECP2, together with the NCOR1/2 co-repressor complex, represses the expression of target genes, including lipid biosynthesis genes in the mouse liver. Here, a Mecp2-mutant mouse model was used to study effects of Mecp2 loss in the lung. Electron microscopy (EM) and high-performance liquid chromatography (HPLC) analyses showed increased neutral lipids and cholesterol in Mecp2-deficient lungs, and strikingly decreased phospholipids, the most predominant lung lipid, in bronchoalveolar lavage fluid. Single-cell RNA sequencing of lung lipid- producing alveolar epithelial 2 (AE2) cells revealed numerous lipid metabolism and mitochondrial gene expression changes in Mecp2/Y mice. Subsequent ChIP-qPCR assays indicated acyl- coenzyme A thioesterase 1 (Acot1) and 3-hydroxy-3-methylglutaryl-CoA synthase 1 (Hmgcs1) are direct targets of MECP2- NCOR1/2 transcriptional repression. Further, AE2 cell-targeted deletion of Mecp2 is sufficient to cause lipid accumulation and respiratory symptoms, while hindbrain neuron-specific loss of Mecp2 results in distinct respiratory abnormalities, highlighting the importance of MECP2-dependent transcriptional regulation in peripheral tissues. These results will inform future treatment strategies aiming to manage respiratory difficulties in RTT patients.

66 2.2 Introduction

Altered pulmonary lipid metabolism is a feature of several respiratory diseases. Pulmonary surfactant, a lipid-rich complex of phospholipids, proteins and neutral lipids, is essential for lowering surface tension in the lung and preventing alveolar collapse (294). The regulation of abundance and species composition of surfactant lipids requires tight control; phosphatidylcholine represents 85% of lipids in pulmonary surfactant, while phosphatidylglycerol makes up another 11%. Cholesterol and other neutral lipids are present, though the role of triglycerides in the lung has not been well defined (285). Lung triglycerides may serve as substrates for phospholipid synthesis when required.

Preterm infants born before producing enough surfactant develop respiratory distress syndrome (RDS), emphasizing the importance of lung surfactant (310). Additionally, infants born with mutations in surfactant production genes, surfactant-protein B (SFTPB), C (SFTPC), or the lipid transporter ATP-binding cassette subfamily A member 3 (ABCA3) have extremely low quantities of surfactant phospholipids which causes surfactant dysfunction (309,335). Alternatively, excess surfactant is a feature of pulmonary alveolar proteinosis (PAP), caused by mutations in granulocyte-macrophage colony-stimulated factor (GMCSF), which prevents surfactant clearance (304). Further, the deactivation of sterol regulatory element-binding protein 1 (SREBP1) increases lung cholesterol and triglycerides while decreasing phospholipids in mice (336).

Breathing abnormalities, including hyperventilation, apneas, breath holds, and forced breathing, are a prominent feature of Rett syndrome (RTT), a severe neuro-metabolic disorder caused by mutations in the X-linked gene methyl CpG-binding protein 2 (MECP2) (337). Respiratory dysfunction causes up to 80% of RTT patient death (31). As a predominantly neurological disease, breathing symptoms in RTT have been historically attributed to autonomic disturbances and brainstem dysfunction (179,324,326,327). Treatments aimed at dampening the hyperexcitation of brainstem nuclei or modulating neurotransmission through targeting the BDNF, GABA, dopamine and serotonin systems in the respiratory neural network in Mecp2-mutant mice have modestly or transiently improved breathing (185,198,212,328).

MECP2’s primary role is to act as a bridge between methylated DNA and the nuclear receptor corepressor 1/2 (NCOR1/2) complex (89). NCOR1/2 represses transcription by recruiting histone deacetylase 3 (HDAC3) to target genes; HDAC3 removes acetyl marks placed by histone acetyltransferases, resulting in a repressed chromatin state. The interaction between MECP2 and the NCOR1/2 complex is facilitated by MECP2’s amino acids 285-309 which bind to TBL1X and TBL1XR1 via their WD40 domains (60,89). Importantly, when Mecp2 is mutated, this complex binds to chromatin with a lower efficiency, resulting in the upregulation of numerous genes (164).

67 Mecp2-mutant mice have been instrumental in understanding the pathological processes underlying the progression of RTT. Mecp2/Y male mice develop behavioral phenotypes, including hind limb clasping and hypoactivity, at 4 weeks of age, followed by a rapid decline in health that leads to their death by 8-10 weeks (134). Alternatively, Mecp2/+ female mice develop neurological phenotypes between 4-6 months of age with symptom variability due to random X-chromosome inactivation. Like human RTT patients, Mecp2-mutant mice display perturbed breathing which is commonly reported as an increase in breathing frequency or breathing frequency variation and an increased incidence of apneas (134).

Our lab conducted a forward genetic suppressor screen to identify novel therapeutic targets for RTT (167,338). Adult male C57BL/6J mice were injected with N-ethyl-N-nitrosourea (ENU), an alkylating agent that induces single point mutations in spermatogonial stem cells. These mice were mated to 129S6/SvEvTac female mice heterozygous for the Bird null allele (129.Mecp2tm1.1Bird/+). The first generation (G1) of Mecp2/Y offspring were screened for rescue of neurological defects; G1 males with improved health were bred for heritability assessments of putative suppressor mutations. One suppressor mutation was found in squalene epoxidase (Sqle), which encodes a rate-limiting enzyme in the cholesterol biosynthesis pathway. Mecp2/Y mice heterozygous for a nonsense mutation in Sqle had a significantly extended lifespan and drastically improved motor coordination and activity. Subsequent investigations found that cholesterol metabolism is upregulated in the brain and liver of Mecp2-mutant mice. Lipid-lowering statin drugs, which competitively inhibit HMG-CoA reductase (HMGCR) in the cholesterol biosynthesis pathway, improved survival and neurobehavioral symptoms in Mecp2-mutant mice (167). Molecular studies showed that MECP2, with the NCOR1/2 complex, directly regulates the transcription of lipogenic enzymes in the liver. Liver-specific loss of Mecp2 increases liver lipogenesis and causes fatty liver disease (164). Further, a subset of RTT patients have increased serum cholesterol and triglycerides (16,190). These studies implicate MECP2 as an important regulator of lipid metabolism.

MECP2 is highly expressed in the mouse and human lung (51,339). Despite this, the lung is greatly understudied in the RTT field. As the lung represents a major site of de novo lipogenesis, lung metabolism genes may be a target of MECP2-directed transcriptional regulation. Here, I report that Mecp2 deletion causes profound neutral lipid accumulation in the lungs of mice, while phospholipids, which are crucial for normal lung function, are decreased in lavage fluid. Further, MECP2 works in conjunction with NCOR1/2 to repress lipid metabolism enzymes in AE2 cells. AE2 cell-specific deletion of Mecp2 is sufficient to increase lung lipids and cause respiratory symptoms in mice. These findings inform on MECP2’s role as a transcriptional regulator and point to new therapeutic strategies to improve respiratory abnormalities in RTT patients.

68 2.3 Methods

2.3.1 Animals

All animal procedures were approved by the Animal Use Committee at the CCAC-accredited animal facility, The Center for Phenogenomics (TCP). Congenic 129.Mecp2tm1.1Bird/Y mice feature a deletion of the last two exons (exons 3-4) of the Mecp2 transcript, resulting in a null allele. Male Mecp2/Y (null) and +/Y (wild type), and female Mecp2/+ and +/+ mice were obtained by backcrossing Mecp2tm1.1Bird/+ females to males of the 129SvEvS6/Tac strain. Mice were fed a standard diet (Harlan Teklad 2918) ad libitum, consisting of 18% protein, 6% fat, and 44% carbohydrates. Mice were housed in a 13-hour light/dark cycle and were euthanized between the hours of 9AM and 12PM (ZT 2-5) to control for circadian rhythm fluctuations.

Conditional deletions of Mecp2, and respective controls, were obtained by crossing B6J.Mecp2tm1Bird/+ (Mecp2-flx) heterozygous female mice to male mice heterozygous for a Cre transgenes, which were also on a C57BL/6J background. AE2 cell-specific deletion of Mecp2 was achieved by the crossing Mecp2-flx mice to B6J.Sftpctm1(cre/ERT2)Blh mice, which have a tamoxifen- inducible Cre under the control of the surfactant protein C promoter. Sftpctm1(cre/ERT2)Blh;Mecp2tm1Bird/Y (Sftpc-CreERT2/Mecp2-flx) mice were given three 75 mg/kg bodyweight intraperitoneal injections of tamoxifen (Tmx) at 3 weeks of age over a period of 5 days (every other day) to induce Cre excision. Hindbrain-specific deletion of Mecp2 was achieved by crossing Mecp2-flx mice to mice with Cre under the control of the atonal BHLH transcription factor 1 promoter (B6J.Cg-Tg(Atoh1-cre)1Bfri). To achieve whole-body deletion of Mecp2 on the C57BL/6J background, 2-cell embryos were retrieved from pregnant Mecp2tm1Bird/+ mice, incubated in a solution containing His-TAT-NLS tagged Cre recombinase, and transferred to a pseudo-pregnant female recipient, as previously described (340), by the Mouse Model Services Core at TCP.

2.3.2 Electron microscopy and immunohistochemistry

Mice were anesthetized with an intraperitoneal injection of 100 mg/kg bodyweight of ketamine and 10 mg/kg bodyweight xylazine in saline. Once unresponsive to a toe-pinch, a needle was inserted into the left ventricle of the heart and mice were perfused with 10 U/ml heparin in phosphate buffered saline (PBS) at a rate of 3 ml/min for 10 minutes. Adequate perfusion was determined by blanching of the liver. Following perfusion, a 23 G needle was inserted into the trachea and 1ml of 2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) was slowly injected to expand the lung. The trachea was tied with sutures and the lung was dissected out of the chest cavity and immersed in the above fixative for 16 hours at 4°C. Tissues were post-fixed in osmium tetroxide, dehydrated in an ascending series of acetone, and embedded in Epon Araldite prior to

69 polymerization at 60 °C overnight. Super thin sections were cut, mounted on grids, and stained with uranyl acetate and lead citrate prior to microscopy. Images were captured with a charge- coupled device camera (AMT Corp.) and an electron microscope (JEOL JEM1011). A minimum of 15 images were taken from each animal.

For immunohistochemistry, lungs of 3-week-old Mecp2tm3.1Bird mice were perfused with 10 U/ml heparin in PBS as above and fixed via intra-tracheal administration of 4% paraformaldehyde (PFA). Excised lungs were immersed in 4% PFA for 16 hours at 4°C. Tissues were washed in PB , embedded in paraffin, cut at a thickness of 5 μm, and dried on a slide warmer at 37°C. Mounted sections were deparaffinized through a graded series of ethanol washes. Antigen retrieval was performed in 0.01 M citrate buffer, pH 6.0. Slides were blocked with 5% goat serum, 1% bovine serum albumin, and 0.1% Triton X-100 in PBS. Slides were incubated with primary and secondary antibodies: GFP (ab13970, 1:500), SPC (ab90716, 1:500), Alexafluor 488 (Thermo A-11039, 1:400), Alexafluor 594 (Thermo A-11012, 1:400). Slides were mounted using ProLong Gold Antifade Mount with DAPI (Thermo Fisher) and imaged on a Nikon A1R confocal laser microscope equipped with NIS Elements analysis software.

2.3.3 Lipid quantification

Blood was drawn from mice using cardiac puncture between the hours of 3PM and 5PM (ZT 8- 10) and serum was separated in BD serum separation tubes following manufacturer instructions. Serum and tissue samples were stored at -80 °C until analysis by the Diabetes and Endocrinology Center at Baylor College of Medicine (Houston, Texas). Lipids were isolated from tissue using CHCl3:CH3OH extraction, followed by drying of the organic phase under N2 pressure. Serum and tissue cholesterol and triglyceride concentrations were assessed by high performance liquid chromatography (HPLC).

After sacrifice, a blunt tip 20 G needle was inserted into the trachea and the lungs were lavaged with 500 µl - 1 ml of sterile PBS. Bronchoalveolar lavage (BAL) fluid was stored at -80 °C until further processing. Measurements of BAL cholesterol, phosphatidylcholine, and sphingomyelin concentrations were performed by liquid chromatography mass spectrometry (LC-MS/MS) at the Analytical Facility for Bioactive Molecules (AFBM) at The Hospital for Sick Children (Toronto, Canada). BAL samples (200 µl) were transferred to siliconized glass tubes containing 800 µl of ultra-pure water and were spiked with internal standards. Lipid extraction was performed using chloroform/methanol extraction. Organic phases were dried and reconstituted in 100 µl ethanol with 0.2% formic acid and transferred to siliconized vials for analysis. LC/MS-MS was performed on an Agilent 1290 Series binary pump (Agilent Technologies Inc.) coupled to an API4000 triple-

70 quadruple mass spectrometer (SCIEX). Quantitative analyses were based on the calibration curve for each analyte and analyzed by LC/MS-MS in the same conditions.

2.3.4 Lung single cell isolation and flow cytometry

P18 mouse lungs were collected from +/Y and Mecp2/Y mice at 7AM. A single-cell suspension was made as published by Singer, 2014, with modifications (341). Briefly, lungs were excised and incubated in fresh Hank’s balanced salt solution (HB ) with 5 U/ml dispase, 0.1% collagenase I, and 0.002% DNase I for 30 minutes at room temperature. The lung was then disintegrated using forceps in a 6 cm petri dish. The cell suspension was filtered through 100um, 70um, and 30um nylon cell strainers. Red blood cells were lysed using RBC lysis buffer (Miltenyi Biotec). Cells were pelleted and resuspended in 500 ul staining media (SM: HBSS, 2% fetal bovine serum (FBS, Wisent), 10 mM HEPES, pH 7.2). Cells were counted using the TC10/TC20 cell counter (BioRad). Approximately 3 million cells per sample were stained with fluorochrome-conjugated antibodies from BD Biosciences: CD45.2 (558702), CD31 (561814), CD326 (563478), I-A/I-E (553623), and Podoplanin (566390). Cells were sorted using a MoFlo Astrios (Beckman Coulter) cell sorter. AE2 cells were: CD45.2-, CD31-, CD326+, I-A/I-E+, Podoplanin-. Following sorting, cells were pelleted and re-suspended in fresh staining media.

2.3.5 Single cell RNA-sequencing

For single-cell RNA sequencing, AE2 cells were isolated using flow cytometry from one +/Y and one Mecp2/Y mouse. Following cell sorting, over 95% of cells were negative for trypan blue (Invitrogen). Sequencing was carried out at the Princess Margaret Genomics Facility (Toronto, ON) on the 10x Genomics platform. FASTQ sequencing reads were processed, aligned to the mouse genome (mm10), and converted to digital gene expression matrices using STA aligner (STAR v2.5.2b). The CELLRANGER (v3.0.2) pipeline was used to obtain two types of gene- barcode matrices: the first is an unfiltered gene-barcode matrix and the second is the filtered gene-barcode matrix. The matrices were loaded into R (v3.5.1) for the final graphical output of results and statistical analysis using SCATER (v1.2.0), CELLRANGERRKIT (v1.1.0), SCRAN (v1.2.2), RTSNE (v0.11), SC3 (v1.3.14), EDGER (v3.16.5), SEURAT (v2.2.0), and PCAMETHODS (v1.50.0). After cell quality control, 1,559 +/Y cells and 1,524 Mecp2/Y cells remained. Low-abundance genes were filtered out and the data set was normalized as previously published (342). Clustering and differential expression analyses were performed using a K- nearest neighbor algorithm and binary classifiers.

71 2.3.6 RNA extraction and quantitative reverse transcription polymerase chain reaction (RT-qPCR)

RNA qRT-PCR was extracted from lungs of mice dissected at 9:00 – 10:00 am (ZT 2-3). Mice were cervically dislocated and the thoracic cavity was opened. Lungs were carefully removed, flash frozen in liquid nitrogen, and stored at -80 °C until further processing. A 5 mm stainless steel bead (Qiagen, 69989) was added to chilled tubes containing lung tissue, which was then homogenized in 400 uL of Lysis Reagent (Qiazol) using a Qiagen TissueLyser II. RNA extraction was carried out using the RNeasy Lipid Tissue Mini Kit (Qiagen, CA, USA) and stored at -80 °C.

Reverse transcription of RNA was performed using the Superscript VILO cDNA synthesis kit (Invitrogen, CA, USA). Gene primers for RT-qPCR were designed to span exon-exon junctions of the gene of interest. RT-qPCR was performed using Power SYBR Green PCR Master Mix (Invitrogen) and the Viia7 instrument (ABI). PCR conditions were 95 °C for 10 min, followed by 40 cycles of 95 °C for 10 sec and 60 °C for 60 sec. Single product amplification was confirmed by disassociation curves. Each sample was amplified three times for precision. The average Ct of these technical replicates was used to calculate expression. Gene expression was normalized to TATA-binding protein (Tbp) internal loading control and analyzed using the 2-(ΔΔCT) method. The primers used are included in Table 2.1.

2.3.7 Native protein extraction and Western blotting

Lungs were dissected from +/Y and Mecp2/Y mice at postnatal day (P) 21 and 56. Tissues were flash frozen and stored at -80°C until further processing. Frozen tissues were transferred to round bottom tubes with a 5 mm stainless steel bead (Qiagen, 69989) and 300 µl of RIPA buffer (150mM NaCl, 1%NP-40, 0.5% sodium deoxychlorate, 0.1% SDS, 50mM Tris, pH 8.0 and fresh protease inhibitors (Sigma-Aldrich, 11873580001). Tissues were lysed in a TissueLyser II (Qiagen, 85300) at 50 Hz for 2-3 minutes. Protein concentration was determined using a Bradford assay (Thermo Fisher, 23200). Proteins were separated in a 4-15% gradient sodium dodecyl sulfate– polyacrylamide gel electrophoresis (SDS-PAGE) gel (BioRad, 4561084). Proteins were transferred to a polyvinylidene fluoride (PVDF) membrane using the Trans-Blot SD semi-dry transfer cell (BioRad). Following transfer, membranes were blocked in 5% milk-TBST. The following antibodies were used: anti-SREBP1 (Santa Cruz Biotechnology, sc-366; 1:500), anti- SREBP2 (abcam, ab30682; 1:500), anti-GAPDH (Cell Signaling, 5174, 1:2000), and peroxidase- Affinipure goat anti-rabbit IgG (Jackson ImmunoResearch, 111-035-144). Proteins were visualized using Clarity western ECL (BioRad, 170-5060) and imaged on a Chemi-Doc system (BioRad).

72 Table 2.1: Primers used for RT-qPCR analysis. Site Forward Reverse Srebf1 CGACTACATCCGCTTCTTGCAG CCTCCATAGACACATCTGTGCC Hmgcs1 GGAAATGCCAGACCTACAGGTG TACTCGGAGAGCATGTCAGGCT Acly AGGAAGTGCCACCTCCAACAGT CGCTCATCACAGATGCTGGTCA Fasn CACAGTGCTCAAAGGACATGCC CACCAGGTGTAGTGCCTTCCTC Scd1 CCTGCGGATCTTCCTTATCA CTTCTCGGCTTTCAGGTCAG Elovl1 CTGGCTCTTCATGCTTTCCAAGG AAGCACCGAGTGGTGGAAGACA Acot1 AAGAAGCCGTGAACTACCTGCG TGTGATGCCCTTCAGGAAGGAG Plin2 GTGGAAAGGACCAAGTCTGTG GACTCCAGCCGTTCATAGTTG Chka TTGGCGATGAGCCTCGGAAAGT GTGACCTCTCTGCAAGAATGGC Pcyt1a TCCTTCCAAAGTGCAGCGTTGC GCAGGCTTCTTCCATAGTCACC Abca3 CTTCATGGACGAAGCTGACCTG GTGCGGTTCTTTTACCAGCGTC

73 2.3.8 Nuclear protein extraction and immunoprecipitation

Nuclear extraction was performed following the protocol in Lyst, 2013 (89). Briefly, frozen tissue was dounced with 1ml of NE1 buffer (20mM HEPES pH 7.9, 10mM KCl, 1mM MgCl2, 0.1% Triton X-100, 20% glycerol, protease inhibitor). Cells were pelleted and washed twice with NE1. Cells were then resuspended in 500 ul of NE1 with 10 μl of benzonase (Millipore 70746-3). Pellet was gently shaken at room temperature for 15 minutes. An equal volume of NE300 buffer (NE1 buffer with 5 mM NaCl) was added to each tube and gently rotated for 30 minutes at 4 °C. Tubes were spun at 16,000 g for 20 minutes at 4°C. Nuclear lysates (supernatant) were transferred to new tubes.

Immunoprecipitation was performed using the Millipore Catch and Release kit (Millipore). Following manufacturer’s instructions, 500 ug of nuclear lysate was incubated with an anti- TBL1XR1 antibody (Bethyl Laboratories, A300-408A) for 3 hours at 4°C. Following elution, half of the immunoprecipitated sample was loaded into each lane of a 4-20% gradient SDS-PAGE gel. Following transfer to a PVDF membrane, the following antibodies were used for western blotting: anti-MECP2 (Sigma-Aldrich, M7443), anti-HDAC3 (abcam, ab7030), and anti-NCOR1 (Bethyl Laboratories, A301-145A). Input samples represent 25 ug of nuclear extracted protein.

2.3.9 Chromatin immunoprecipitation

Chromatin immunoprecipitation was performed as published in Schmidt, 2009 (343). Briefly, frozen lungs were crosslinked by incubation in Solution A (1% formaldehyde, 50 mM HEPES- KOH, 100 mM NaCl, 1 mM EDTA, 0.5 mM EGTA) for 20 minutes at room temperature, after which 2.5 M glycine was used to quench the reaction. Tissue was washed and dounced in ice-cold PBS with a loose and tight pestle. Dounced cells were filtered through a 100 µm cell strainer and centrifuged. Protein G-Dynabeads (Invitrogen) were pre-blocked with 0.5% BSA in PBS, and bound to an anti-TBL1XR1 antibody (Bethyl Laboratories, A300-408A) by overnight incubation at 4°C. Crosslinked cells were resuspended and washed in exchanges of LB1 (50 mM HEPES-KOH, 140 mM NaCl, 1 mM EDTA, 10% glycerol, 0.5% NP-40, 0.25% Triton-X 100), LB2 (10mM Tris- HCl, pH 8.0, 200mM NaCl, 1mM EDTA, 0.5mM EGTA) and LB3 (10mM Tris-HCl, pH 8.0, 100mM NaCl, 1mM EDTA, 0.5mM EGTA, 0.1% Na-Deoxycholate, 0.5% N-lauroylsarcosine), each mad with fresh protease inhibitors. Chromatin in LB3 was transferred to 1.5 ml Bioruptor Pico microtubes (Diagenode C30010016) and sheared using the Bioruptor (Diagenode). Sheared chromatin was incubated with beads overnight at 4°C. Beads were washed with ice-cold RIPA buffer (50mM Hepes-KOH, pH 7.5, 500mM LiCl, 1mM EDTA, 1% Igepal CA-630, 0.7% Na- Deoxycholate) and TBS (20mM Tris-HCl, pH 7.6, 150mM NaCl). Chromatin was eluted from the beads in elution buffer (50mM Tris-HCl, pH 8, 10mM EDTA, 1% SDS) at 65°C overnight. Input

74 and immunoprecipitated chromatin were treated with RNAse A at 37°C for 30 minutes and Proteinase K at 55°C for 2 hours. DNA was purified using 5PRIME Phase Lock Gel Light tubes (QuantaBio) following the manufacturer’s instructions. DNA pellets were washed with ethanol, dried, and resuspended in 10mM Tris-HCL, pH 8.0. Primers used for downstream qPCR analysis of immunoprecipitated and input DNA are listed in Table 2.1. Primers were designed within predicted CpG methylation sites of gene promoters using sequence data obtained from Ensembl (344) and the UCSC Genome Browser (345). Primers were designed using SnapGene® software.

2.3.10 Subjective health assessments

Mice were assayed for general health once per week from 4 weeks to 10 weeks of age. Scoring was blinded by genotype. Mice were scored using the assessment published in Guy, 2007, with slight modifications (149). In this scoring method, mice are given a score between 0 to 2 based on the severity of the phenotype assessed (Table 2.2). Mice were assessed for limb clasping, tremors, activity, grooming, hypotonia, and body weight, for a combined possible score of 0-12.

2.3.11 Rotarod

Motor coordination was measured using the rotating rod (Stoelting ugo basile mouse rota-rod). Mice were placed on the grooved rotating rod facing the opposite direction of rotation. The revolution rate increased from 4 rotations per minute (RPM) to 40 RPM over the course of 5 minutes. The length of time that each mouse remained on the rod was recorded for eight trials over two consecutive days (four trials per day), with a minimum of 30 minutes between each trial. A trial ended for a mouse when it fell from the rod, stayed stationary on the rod while it spun for two revolutions, or when it successfully stayed on the rod for 5 minutes.

2.3.12 Open field activity

Locomotor activity was measured by free activity in an open field chamber (Accuscan Instruments). Open field chambers were cleaned thoroughly with Clidox prior to use. One mouse was placed in the center of each 40cm x 40cm open field chamber illuminated at 200 lux and left undisturbed for 20 minutes. Movement was tracked using the VersaMax software (Accuscan Instruments). Mice were removed from the open field chamber and each chamber was thoroughly cleaned with Clidox between experiments.

75 Table 2.2: Primers used for ChIP-qPCR analysis.

Site Forward Reverse Product size Acot1 1 GCCAAGGGAGAGATAATCAAGTGTC TGACATCTGGGTCACCCCAT 67 bp Acot1 2 GGCCCATCTCCCACCCCT TATAGAAAAGGAGTGGGGCTGGG 113 bp Acot1 3 TCTTGGCTCCGGGGGT ACTTTGAGCAAAGGGAATGTGG 106 bp Hmgcs1 1 TCCCCAGGAGTGGGACG GCGAGAGCCAACGGGA 100 bp Hmgcs1 2 CCGCAGACTGTGCCTCC CGGGACACTCACCCAAAGGG 91 bp Hmgcs1 3 GGGCTTGGGGACTCCG TGACTATGTCATCAGAACACGGAG 103 bp Prdx6 1 CGCTTCCACGATTTCCTG GAAGCGAGGCAGGTAGCTCT 87 bp Prdx6 2 GGTCCCCTCCCCCTCTG ACAAAATCGGTGCATACCTAGCC 92 bp Prdx6 3 CTAGACTCTCTCTGCTCCGGG CACAGTCCTTGAAATTCATCACC 100 bp

Table 2.3: Subjective health score definitions

Parameter 0 0.5 1 1.5 2 Limb clasping As WT Partial limb clasp Single limb clasped Single limb clasped Both limbs clasped and other limb partially clasped

Tremors As WT Slight tremor Mild intermittent tremor Strong intermittent or Strong continuous mild continuous tremor tremor

Activity As WT Slightly slower than WT Active when prodded Hypoactive Inactive

Grooming As WT Slightly worse Poor grooming Disheveled and/or oily Very disheveled, oily grooming fur fur

Muscle tone/ body As WT Mild loss of muscle One of: flat, poor tone, Two of: flat, poor tone, Flat, poor tone, and condition tone or weak grip strength or weak grip strength weak grip strength

Body weight ≤25g 26-28g 29-31g 32-34g ≥35g Mice were scored once for each trait once per week. Scores were summed with possible values of 0 – 12.

2.3.13 Social behavior

Social behaviour was assessed in a 60 cm x 40 cm three-chamber apparatus with two clear plexiglass partitions. The apparatus was thoroughly disinfected with Clidox prior to use. A novel object (small orange cup) or a stranger mouse (wild type mouse of the same sex and genetic background as the subject) were placed in containment cups in the left or right chambers. Mice were placed in the middle chamber with no access to other chambers for a 5-minute acclimation period. Subjects were then allowed to freely move around the chambers for a period of 10 minutes. Following this, subjects were returned to the middle chamber with no access to the other chambers. The stranger mouse was moved with its containment cup to the opposite chamber. A novel stranger mouse was inserted in a clean containment cup in the chamber where the original stranger mouse was located. Access to all chambers was restored and subjects could explore freely for another 10 minutes. All movements by the subject mouse were recorded using Ethovision XT. The software generated tracking coordinates for each subject, which were analyzed as time spent within each chamber with a custom script.

2.3.14 Plethysmography

Respiration was monitored using a Buxco Whole Body Plethysmography (WBP) apparatus (Data ciences International) according to manufacturer’s instructions. All testing was conducted between the hours of 9AM and 12PM. Mice were placed in plethysmography chambers and allowed to acclimate for 30 minutes, until motionless. Baseline breathing rates were measured for a period of 5 minutes. Following this, mice were exposed to nebulized saline for 2 minutes, and respiratory rates were measured for 5 minutes. Mice were then exposed to simultaneous doses of 6.25, 12.5, 25 and 50 mg/ml of aerosolized methacholine dissolved in saline at a constant rate for 2 minutes, after which readings were taken for 5 minutes at each concentration. Breathing frequency, tidal volume, and enhanced pause (PenH) were analyzed using Buxco FinePoint Software. Apneas were defined as cessation of breathing for over 1 second (2 respiratory cycles) and were assessed manually over a period of baseline breathing.

2.3.15 Statistics

The significance of the differences in mean values across two groups was evaluated by two-tailed Student t-tests. The statistical difference between four groups was evaluated using one-way ANOVA with the Tukey post-hoc test for multiple comparisons. All statistical analyses were performed in GraphPad Prism (Version 7). P-values less than 0.05 were considered statistically significant.

78 2.4 Results

2.4.1 Loss of Mecp2 results in abnormal lipid production in the Mecp2/Y mouse lung

Like RTT patients, Mecp2-mutant mice display respiratory disturbances including altered breathing frequency and increased apneas. To visualize changes in lung structure that could contribute to breathing symptoms in Mecp2/Y mice, transmission electron microscopy (TEM) was performed on perfused, inflation-fixed lungs. TEM was performed in collaboration with Dr. Cameron Ackerley at the Hospital for Sick Children. Strikingly, large lipid droplets were abundant in postnatal day (P) 56 Mecp2/Y mouse lungs (Figure 2.1A). This lipid accumulation was present in the extra-cellular space around the alveoli and was absent in age-matched +/Y mice.

High-performance liquid chromatography (HPLC) and mass spectrometry (LC-MS/MS) were used to quantify lipids in the mouse lungs. Dipalmitoylphosphatidylcholine (DPPC) is the major constituent of pulmonary surfactant, making up approximately 40% of its normal composition. DPPC is composed of two 16-carbon saturated chains and is the strongest surfactant molecule in the pulmonary surfactant mixture. Other phosphatidylcholine (PC) molecules in lung surfactant are hypothesized to regulate the fluidity of the surfactant and aid in adsorption. Cholesterol and other neutral lipids such as triglycerides are present in small quantities in the surfactant.

Consistent with the EM findings, when compared to +/Y mice, Mecp2/Y mice had increased lung triglycerides at P21 (5.31 ± 0.56 mg/g in +/Y; 9.46 ± 0.58 in Mecp2/Y, p=0.002) and P56 (11.15 ± 1.18 mg/g in +/Y; 19.39 ± 2.15, p=0.015) (Figure 2.1B). Mecp2/Y mice also accumulated cholesterol in their lungs at both time points (P21: 9.22 ± 0.93 mg/g in +/Y; 14.30 ± 1.82 in Mecp2/Y, p=0.048; P56: 7.58 ± 0.86 mg/g in +/Y; 11.97 ± 0.97, p=0.015) (Figure 2.1C). Thus, lung lipid accumulation precedes respiratory and neurobehavioral symptom onset. DPPC (PC 32/0) and other PC levels were unchanged in the lung tissue of Mecp2/Y mice at both P21 and P56 (Figure 2.1D,E).

Sphingolipids are important signaling molecules that play a role in immune response, defense from pathogens, and regulation of inflammation. In P21 Mecp2/Y mice, lung sphingomyelins were not altered. However, at P56, a few species of sphingomyelins were increased (Figure 2.1F). This late increase in sphingomyelins could be secondary to the disease state rather than from Mecp2 deficiency. Further, it could reflect an altered immune response in the lung, consistent with previous literature (333,334).

To quantify lipids in the pulmonary surfactant, bronchoalveolar lavage (BAL) fluid was collected through instillation and retrieval of a saline solution through the trachea. DPPC was drastically reduced in Mecp2/Y BAL fluid at both P21 (156.91 ± 12.58 µg/ml in +/Y; 64.80 ± 3.017 in Mecp2/Y,

79 p=0.0024) and P56 (117.58 ± 3.96 µg/ml in +/Y; 86.78 ± 9.38 in Mecp2/Y, p=0.0116) (Figure 2.1G). Other PC species were also detected at lower quantities in Mecp2/Y BAL fluid (Figure 2.1H). PCs other than DPPC are thought to contribute to the viscosity and fluidity of lung surfactant. Such a marked reduction in surfactant DPPC and other PCs likely impacts surfactant function and could greatly affect the ability of the lungs to facilitate normal gas exchange. Finally, cholesterol was increased in Mecp2/Y BAL fluid at P21 (2431.25 ± 283.52 µg/ml in +/Y; 3643.25 ± 381.76 µg/ml in Mecp2/Y, p=0.0436), but was at normal levels by P56 (Figure 1I). Altogether, these findings indicate an increase in cholesterol and neutral lipids in the lung tissue and a parallel decrease in phosphatidylcholines in the BAL fluid.

Figure 2.1: Lipids are altered in the lung and BAL fluid of male Mecp2/Y mice. A) Transmission electron microscopy (TEM) images of +/Y and Mecp2/Y lungs at P56. Scale bars represent 2 µm, *: lipid droplets, n=2. Lung B) triglycerides, C) cholesterol, D) phosphatidylcholine (PC) 32/0 (DPPC), E) other PCs, and F) sphingomyelin (SM) were measured using HPLC or LC- MS/MS. BAL G) PC 32/0 (DPPC), H) other PCs, and I) cholesterol were measured using LC- MS/MS. P: postnatal day. Data are expressed as mean ± SEM. Statistics were performed using tudent’s t-test. *P<0.05, **P<0.01, ***P<0.001, n=4.

81 2.4.2 Heterozygous deletion of Mecp2 is sufficient to cause lung lipid abnormalities in mice

Due to their clinical relevance, lung lipids were assessed in Mecp2/+ female mice. Through TEM, I found that female Mecp2/+ mice also exhibit drastic lung lipid accumulation at 6 and 9 months of age, with a more dramatic progression, likely due to their longer lifespan (Figure 2.2 A,B). Importantly, heterozygous mosaic loss of Mecp2 is sufficient to cause lung lipid accumulation in mice.

Similar to Mecp2/Y mice, female Mecp2/ + mice display elevated lung triglycerides prior to RTT symptom development at 3 months of age (6.87 ± 0.35 mg/g in +/+; 8.80 ± 0.41 in Mecp2/+, p=0.0120), and this elevation remains over the course of disease progression up to 12 months (p<0.05, Figure 2.2C). In contrast, lung cholesterol is only transiently increased at 6 months in Mecp2/+ lungs, during the peri-symptomatic period when neurobehavioral features are beginning to appear (4.48 ± 0.47 mg/g in +/+; 6.38 ± 0.25, p=0.012) (Figure 2.2D). BAL was collected from female mice at 6 and 9 months of age to represent peri-symptomatic and symptomatic periods, respectively. Like male Mecp2/Y mice, DPPC (PC 32/0) was markedly decreased in the lungs at both 6 months (117.58 ± 7.9; 88.14 ± 3.8, p=0.016) and 9 months of age (115.02 ± 3.25; 85.09 ± 11.9, p=0.045) (Figure 2.2E), and again, other PC species were altered as well (Figure 2.2F). Thus, partial loss of Mecp2 decreases lung PC synthesis, a finding that could contribute to pathological changes in human RTT patients. However, BAL cholesterol was unchanged at these time points (Figure 2.2G).

Figure 2.2: Lipids are altered in the lung and BAL fluid of female Mecp2/+ mice. Transmission electron microscopy (TEM) images of +/+ and Mecp2/+ lungs at A. 6 months (mo) and B. 9 mo. Scale bars represent 2 µm, *: lipid droplets, n=2. Lung C. triglycerides and D. cholesterol were measured by HPLC. BAL fluid E. phosphatidylcholine (PC) 32/0, F. other PCs, and G. cholesterol were measured by LC-MS/MS. Data are expressed as mean ± SEM. Statistics were performed using tudent’s t-test. *P<0.05, **P<0.01, n=4.

83 2.4.3 Single cell RNA-sequencing of mouse AE2 cells reveals altered metabolic gene expression in Mecp2/Y mice

The profound lipid abnormalities in the Mecp2-mutant lung suggested a local metabolic defect. In the lung, alveolar epithelial 2 (AE2) cells are metabolically and functionally complex differentiated epithelial cells whose primary role is to synthesize, secrete, and recycle pulmonary surfactant. In producing pulmonary surfactant, AE2 cells produce all subclasses of surfactant lipids, including PCs and neutral lipids. To visualize MECP2 expression, lungs of Mecp2tm3.1Bird mice were perfused, inflation-fixed, and processed for immunohistochemistry. Mecp2tm3.1Bird mice have a green fluorescent protein (GFP) reporter fused to the endogenous Mecp2 locus, allowing visualization of MECP2 through GFP expression. High MECP2 expression was found in AE2 cells (Figure 2.3A) establishing that MECP2 is enriched in lipid-producing cells of the lung, consistent with previous reports in human and mouse (334,339)

MECP2 is a master transcriptional repressor and its loss results in global transcriptional disarray. In the liver, loss of Mecp2 increases the transcription of lipogenesis genes (164). Thus, I hypothesized that MECP2 regulates lipid metabolism enzymes in lung AE2 cells. To identify AE2 cell-expressed genes that are misregulated in Mecp2 deficiency, AE2 cells were isolated from P18 +/Y and Mecp2/Y mouse lungs by flow cytometry for single cell RNA-sequencing analysis. Isolated cells were CD45.2-negative, CD31-negative, CD326-positive, podoplanin-negative and I-A/I-E-positive. Pre-symptomatic mice were used to assess primary transcriptional deviations initiating the onset of lung disease rather than changes induced because of the diseased state.

Approximately 2000 AE2 cells were sequenced from each mouse. Despite using single cell RNA- sequencing, the cells clustered together under a t-distributed stochastic neighbor embedding (t- SNE) algorithm, indicating that the populations of cells across samples were equivalent (Figure 2.3B). Therefore, downstream analysis of differentially expressed genes were conducted as if the cells in each sample were a uniform population. A total of 436 significant differentially expressed genes were identified in Mecp2/Y AE2 cells compared to +/Y. Of these, 113 showed increased expression and 323, decreased expression. The top 20 genes with altered expression are highlighted in Figure 2.3B. The gene with the highest upregulation in Mecp2/Y AE2 cells was regulator of cell cycle (Rgcc), a gene induced by DNA damage that mediates apoptosis in neurons. The most downregulated gene was histocompatibility 2 variant (H2-ab1).

84 Figure 2.3: Single cell RNA-sequencing of AE2 cells reveals transcriptional metabolic changes in the absence of Mecp2.

A. Lungs of Mecp2tm3.1Bird mice with a green fluorescent protein (GFP) reporter fused to the endogenous Mecp2 locus were used to assess MECP2’s localization. Blue: DAPI, green: MECP2, red: surfactant protein C (SPC, marker of alveolar epithelial 2 (AE2) cells). Scale bars represent 40 µm, arrows point to AE2 cells expressing MECP2. B. tSNE plot of sequenced +/Y and Mecp2/Y AE2 cells. C. Volcano plot showing genes with altered expression in Mecp2/Y AE2 cells after single-cell RNA sequencing. The top 20 genes with altered expression are highlighted.

85 2.4.4 Lipid metabolism is altered in Mecp2/Y AE2 cells

Many lipid metabolism genes were misregulated in Mecp2/Y AE2 cells (Table 2.3, Figure 2.4). The altered expression of these genes was confirmed in whole lung tissue using RT-qPCR (Figure 2.5A). The expression of cholesterol biosynthesis enzymes, hydroxymethylglutaryl-CoA synthase (Hmgcs1) and phosphomevalonate kinase (Pmvk), is increased in Mecp2/Y AE2 ells. This is consistent with increased cholesterol levels in Mecp2/Y lungs. HMGCS1 catalyzes an early step in cholesterol biosynthesis, which converts acetoacetyl-CoA to 3-hydroxy-3-methylglutaryl- CoA (HMG-CoA), while PMVK acts downstream from this step. Notably, MECP2, in concert with NCOR1/2, directly represses the cholesterol biosynthesis gene squalene epoxidase (Sqle) in the mouse liver.

Acyl-CoA thioesterase 1 (Acot1) was the second most significantly upregulated gene in Mecp2/Y AE2 cells (FC: 1.13, p-value: 3.12E-123). Consistently, its family member Acot2 was also upregulated. When required for energy or other metabolic processes, fatty acids are converted to acyl-CoAs by long-chain acyl-CoA synthetases (ACSs), and transported to the mitochondria for beta-oxidation. Acyl-CoA thioesterases (ACOTs) catalyze the reverse reaction: the hydrolysis of coenzyme A esters to the free acid and CoA. ACSs and ACOTs are thought to direct the metabolic fate of fatty acids by channeling substrates toward or away from beta-oxidation. The expression of the ACS acyl-CoA synthetase long chain family member 5 (Acsl5) is decreased in Mecp2/Y AE2 cells (FC: -0.470, p-value: 1.78E-14). Together with the increased expression of ACOTs, these results suggest that fatty acids in Mecp2/Y lungs are shuttled away from mitochondrial recycling and pushed toward lipid storage pathways. In support of this, perilipin-2 (Plin2) is significantly elevated in Mecp2/Y AE2 cells (FC: 0.318, p-value: 2.27E-17). PLIN2 promotes lipid accumulation by coating triglycerides and protecting them from lipolysis. Importantly, these gene expression changes may provide an avenue for lipid droplet storage in the Mecp2-mutant lung.

Given the high levels of triglycerides in Mecp2/Y lungs, I expected the fatty acid synthesis pathway to be upregulated in Mecp2/Y AE2 cells. Instead, a global decrease in lipid biosynthesis was found (Figure 2.4). Fatty acid biosynthesis genes including fatty acid synthase (Fasn), stearoyl-CoA desaturase ½ (Scd1,2), and ELOVL fatty acid elongase (Elovl1) are significantly decreased in Mecp2/Y AE2 cells. Fasn encodes a large dimeric enzyme that catalyzes the formation of palmitate from acetyl-CoA and malonyl-CoA. Palmitate is then elongated into mono- (MUFAs) or polyunsaturated fatty acids (PUFAs) through reactions by SCD1/2 and ELOVL1-7, respectively. Notably, Dgat1 and Dgat2, genes encoding diglyceride acyltransferase enzymes which catalyze the formation of triglycerides from diacylglycerol are unchanged, despite triglyceride accumulation in Mecp2/Y mice. Further, ATP citrate lyase (Acly), which encodes the primary enzyme linking

86 carbohydrate metabolism (citric acid cycle) to fatty acid biosynthesis, is expressed at lower levels in Mecp2/Y AE2 cells.

Decreased expression of genes in the fatty acid biosynthesis pathway could be explained by changes to sterol regulatory-element binding proteins (SREBPs). SREBPs are membrane-bound transcriptional regulators of lipid homeostasis. Three isoforms exist, SREBP1a and SREBP1c coded for by Srebf1, and SREBP2 which is coded for by Srebf2 (346,347). SREBP1a stimulates global lipid synthesis in proliferating cells, while SREBP1c and SREBP2 are involved in the regulation of triglyceride and sterol synthesis, respectively (348). SREBPs are synthesized as inactive precursors bound to endoplasmic reticulum (ER) membranes. When low lipid levels are detected, SREBPs are cleaved and migrate to the nucleus, binding sterol response elements (SREs) and activating the transcription of a number of lipogenesis genes (349). Any SREBP can activate each target gene but each has different efficacies (350).

In Mecp2/Y AE2 cells, the expression of Srebf1, which stimulates triglyceride synthesis, is decreased while Srebf2, which stimulates sterol synthesis, is unchanged. A Western blot assay in whole lung tissue confirmed that nuclear SREBP1 expression (antibody recognizes SREBP1a and 1c) is decreased at P21 but returns to normal levels by P56 (Figure 2.5A). In contrast, nuclear SREBP2 expression is unchanged at both time points (Figure 2.5B). Interestingly, direct transcriptional activation by SREBP1c has been shown in SRE-containing genes Fasn (351), Scd1 (352), Scd2 (353), and Acly (354), the fatty acid biosynthesis genes with decreased expression in Mecp2/Y AE2 cells.

87 Table 2.4: Lipid metabolism genes are misregulated in Mecp2/Y AE2 cells.

Category Log FC Gene P-Value (Down/Up) Cebpa ↓ 0.66157 2.75E-32 Transcriptional regulation Srebf1 ↓ 0.48646 1.58E-25 of lipid metabolism Nr1d1 ↓ 0.33904 2.18E-25 Foxa2 ↓ 0.29705 1.31E-11 Pmvk ↑ 0.34146 3.15E-16 Hmgcs1 ↑ 0.28520 2.45E-08 Sterol biosynthesis Soat1 ↓ 0.52903 9.72E-24 Raly ↓ 0.25489 0.000109 Cyb5a ↑ 0.33186 1.88E-10 Scd1 ↓ 1.08181 1.57E-56 Scd2 ↓ 0.74866 2.11E-44 Fatty acid biosynthesis Elovl1 ↓ 0.54825 2.27E-15 Fasn ↓ 0.28478 5.45E-05 Acly ↓ 0.35509 5.37E-11 Acot1 ↑ 1.13401 3.10E-123 Acot2 ↑ 0.28115 6.12E-14 Fatty acid regulation Eci2 ↑ 0.41057 2.34E-33 Acsl5 ↓ 0.46970 1.78E-14 Lipid storage Plin2 ↑ 0.31851 2.27E-17 Chka ↓ 0.54883 5.87E-22 Dgkg ↓ 0.31762 7.60E-13 Phospholipid biosynthesis Lpcat1 ↓ 0.48220 2.39E-06 Pcyt1a ↓ 0.26249 3.77E-05 Sptlc2 ↓ 0.37772 2.04E-14 Sphingolipid synthesis Sgms1 ↓ 0.32411 1.33E-09 Cd74 ↓ 1.31228 9.21E-78 Lipid rafts Cftr ↓ 0.27103 5.74E-11 Mlc1 ↓ 0.36050 4.80E-09 Prdx6 ↑ 0.59145 2.18E-39 Peroxisomal lipid Gpx1 ↑ 0.58170 1.92E-34 metabolism Acoxl ↓ 0.56540 4.54E-27 Genes are classified by major lipid metabolism pathway and sorted by P-values in order of significance. LogFC: Log2 fold change. ↑: Increased expression, ↓: Decreased expression.

Figure 2.4: Metabolic gene expression is altered in Mecp2/Y AE2 cells. Arrows indicate direction of metabolic reaction. Lines indicate dependence on other enzymes. Circular arrows indicate genes involved in regulation or transport of end product. Red: Increased expression, Blue: Decreased expression. PLA2: Phospholipase A2 activity. All genes shown have a P-value < 0.05.

Figure 2.5: Lipid metabolism enzyme expression in +/Y and Mecp2/Y lungs. A. Genes with altered expression in Mecp2/Y AE2 cells were assessed in P21 mouse lungs; n = 3, *p<0.05, **p<0.01. B. SREBP1 expression is decreased in Mecp2/Y lungs at P21 but is unchanged at P56. Note the 68 kDa band (shown) is the active, cleaved protein, while the 120 kDa inactive precursor was not visualized. C. SREBP2 expression is unchanged in Mecp2/Y lungs at P21 and P56. Note the 55 kDa band (shown) is the active, cleaved protein, while the 125 kDa inactive precursor was not visualized. GAPDH was used for normalization. SREBP: Sterol regulatory binding protein.

90 Importantly, phosphatidylcholines (PCs) represent the largest percentage of lipids in pulmonary surfactant. PCs are synthesized by the CDP:choline pathway in AE2 cells. The pathway begins with the uptake of exogenous choline from the bloodstream predominantly by sodium-dependent choline transporters. Choline is phosphorylated by choline kinase (CK) to form phosphocholine, which is activated by CTP:phosphocholine cytidylyltransferase (CCT) to make CDP-choline, which is finally joined to a diacylglycerol backbone by choline/ethanolamine phosphotransferase (CEPT) to form phosphatidylcholine. In Mecp2/Y AE2 cells, the CK, choline kinase alpha (Chka) and the CPT, phosphate cytidylyltransferase 1 choline alpha (Pcyt1a) are significantly underexpressed (Chka FC: -0.549, p-value: 5.87E-22; Pcyt1a FC: -0.262, p-value: 3.77E-5). Reduced expression of any enzyme in the PC synthesis pathway leads to profound reductions in PCs. Thus, the decreased expression of PC-producing enzymes is likely the direct cause for the remarkably low levels of PCs in Mecp2/Y BAL.

While 45% of DPPC, the major constituent of lung surfactant, is produced downstream of CDP:choline pathway described above, the remaining DPPC is produced by a remodeling mechanism whereby unsaturated PCs are deacetylated by phospholipase A2 forming 1-palmitoyl- 2-lysophosphatidylcholine; this species is further reacetylated by lysoPC acetyltransferases (LPCAT). Notably, Lpcat1 is expressed at low levels in Mecp2/Y AE2 cells (FC: -0.482, p-value: 2.39E-6), producing an additional molecular roadblock to normal PC production.

ATP binding cassette subfamily A, member 3 (Abca3) is downregulated in Mecp2/Y AE2 cells (FC: -0.506, p-value: 5.42E-9). ABCA3 is a phosphatidylcholine transporter that is important for surfactant lipid metabolism and lamellar body formation (307). Mutations in ABCA3 are associated with surfactant deficiency and fatal respiratory distress (311). Interestingly, Abca3 is also a major target of SREBP1 (355). In contrast, peroxiredoxin-6 (Prdx6) is overexpressed in Mecp2/Y AE2 cells (FC: 0.591, p-value: 2.18E-39). PRDX6 metabolizes recycled surfactant and modifies newly formed PCs. Deletion of Prdx6 in mice leads to an accumulation of surfactant PCs, while its overexpression decreases lung PCs (356). Thus, both Abca3 reduction and Prdx6 elevation could play an important role in RTT lung pathology.

91 2.4.5 Loss of Mecp2 from lung AE2 cells alters mitochondrial gene expression Intriguingly, loss of Mecp2 impacted both nuclear and mitochondrially-encoded genes for electron transport chain (ETC) components in AE2 cells (Table 2.4). Nuclear-encoded NADH dehydrogenase genes Ndufa6 and Ndufa7, cytochrome c reductase (Uqcrh), cytochrome c oxidase (Cox7a2l) and ATP synthase (Atp5l) were expressed at higher levels in Mecp2/Y AE2 cells. In contrast, 11 of the 13 mitochondrially-expressed ETC components were expressed at lower levels. Previous reports have shown that MECP2 regulates the expression of nuclear- encoded mitochondrial genes (357,358). Abnormal mitochondrial structure, including swollen, elongated mitochondria with enlarged cristae, has been evidenced in RTT patients and Mecp2- mutant mouse models (359). Additionally, compromised ETC function and increased oxidative damage have also been reported in RTT (360). While the mechanism underlying mitochondrial abnormalities in RTT is unclear, evidence suggests a progressive mitochondrial decline.

92 Table 2.5: Mecp2 deficiency alters mitochondrial gene expression in lung AE2 cells.

ETC ETC Complex Expression Gene Log FC P-Value Complex Name I NADH ubiquinone Mitochondrial mt-nd1 ↓ 0.84693 1.21E-11 oxidoreductase/ NADH mt-nd2 ↓ 0.87795 3.75E-23 dehydrogenase mt-nd3 ↓ 0.92690 6.98E-45 mt-nd4l ↓ 0.25441 0.006614 mt-nd4 ↓ 0.82727 3.84E-14 mt-nd5 ↓ 0.57768 2.96E-16 Nuclear Ndufa6 ↑ 0.55456 8.27E-31 Ndufa7 ↑ 0.29279 2.91E-09 Ndufa3 ↓ 0.28295 0.000335 III Cytochrome bc1 Mitochondrial mt-cyb ↓ 0.66476 5.31E-06 complex/ CoQH2- Nuclear cytochrome c Uqcrh ↑ 0.29871 4.72E-11 reductase IV Cytochrome c Mitochondrial mt-co1 ↓ 0.53783 0.046432 oxidase mt-co2 ↓ 0.80265 1.68E-11 mt-co3 ↓ 0.78379 8.45E-09 Nuclear Cox7a2L ↑ 0.29685 6.65E-10 V ATP synthase Mitochondrial mt-atp6 ↓ 0.58252 0.022863 Nuclear Atp5l ↑ 0.29906 1.16E-08

Genes are classified according to the electron transport chain (ETC) complex they contribute to. LogFC: Log2 fold change. ↓: Decreased expression, ↑: Increased expression.

93 2.4.6 MECP2 regulates the expression of lung lipid metabolism enzymes through interaction with the NCOR1/2 co-repressor complex

MECP2’s proposed primary function is to anchor the nuclear receptor co-repressor (NCOR)1/2 co-repressor complex to methylated DNA. The NCOR1/2 complex is a potent regulator of deacetylase-dependent gene transcription. Its other complex members include histone deacetylase 3 (HDAC3), G protein pathway suppressor 2 (GPS2), and transducin beta like 1 X- linked (TBL1X), the latter of which is the direct binding partner of MECP2. In the mouse liver, MECP2 facilitates transcriptional repression of squalene epoxidase (Sqle) and fatty acid synthase (Fasn) with NCOR1/2 through the deacetylase activity of HDAC3 (164). Thus, I hypothesized that this corepressor complex contributes to metabolic perturbations in Mecp2-mutant mice.

To determine if this complex forms in the mouse lung, co-immunoprecipitation (Co-IP) was performed. Co-IP of TBL1XR1 revealed an association with MECP2, as well as with NCOR1/2 corepressor complex partners NCOR1 and HDAC3 in the mouse lung (Figure 2.6A). The formation of this complex in the mouse lung suggests its role in direct regulation of lipid metabolism gene targets. Notably, in the absence of Mecp2, such as in Mecp2/Y mice, other complex components still associate with each other.

To determine if changes in the expression of lung lipid metabolism genes are directly caused by Mecp2 deficiency, chromatin immunoprecipitation (ChIP)-qPCR was performed. As MECP2 primarily acts as a transcriptional repressor with NCOR1/2, genes with increased expression in Mecp2/Y AE2 cells were chosen as putative targets of MECP2-directed transcriptional repression. Notably, ChIP experiments using anti-MECP2 antibodies are often unsuccessful as MECP2 promiscuously binds to DNA and direct targets are difficult to determine. Therefore, ChIP was performed using an anti-TBL1XR1 antibody, as TBL1XR1 is the direct link between MECP2 and NCOR1/2. Additionally, unlike other complex members, TBL1X has not yet been linked to other complexes; thus, this experiment should allow us to find targets of the MECP2-NCOR1/2 complex. Simultaneously, ChIP was performed with an anti-H3K4me3 antibody and IgG as positive and negative controls, respectively.

Primers for ChIP-qPCR analysis were designed within gene promoters. As a methyl-CpG-binding protein, MECP2 has a high affinity for methylated DNA, and DNA methylation signatures alone could predict MECP2 binding with 88% accuracy (361). Therefore, primers were initially designed within the region of CpG methylation in gene promoters surrounding their transcription start sites (TSS). Interestingly, TBL1XR1 binds to the promoter of Acot1 (Figure 2.6B) and Hmgcs1 (Figure 2.6C). Further, loss of Mecp2 significantly hindered the capability of TBL1XR1 to bind to the promoters of these genes, both of which had increased expression in Mecp2/Y AE2 cells. Notably,

94 H3K4me3 marks were unaffected by the loss of Mecp2, as expected. These results suggest that MECP2, in concert with NCOR1/2, directly regulates the expression of Acot1 and Hmgcs2 in the mouse lung. In contrast, TBL1XR1 did not bind to the promoter region of the Prdx6 gene, which was also expressed at high levels in Mecp2/Y AE2 cells, as indicated by low percent yields in ChIP-ed DNA (Figure 2.6D). Thus, the increased expression of Prdx6 is likely a downstream effect of Mecp2 loss rather than its direct target.

96 Figure 2.6: MECP2 regulates lipogenic gene transcription with NCOR1/2. A. TBL1XR1 is associated with NCOR1/2 corepressor complex members NCOR1, MECP2 and HDAC3 in the mouse lung, as shown by co-immunoprecipitation (co-IP) assay. TBL1XR1: TBL1X receptor 1, NCOR1: nuclear receptor corepressor 1, MECP2: methyl-CpG-binding protein 2, HDAC3: histone deacetylase 3. i: input (nuclear protein), ip: immunoprecipitated protein. Anti- TBL1XR1, anti-H3K4me3 (positive control), and anti-IgG (negative control) ChIP-qPCR of sites surrounding the B. Acot1 promoter, C. Hmgcs1 promoter, and D. Prdx6 promoter. Blue line refers to areas of CpG methylation in the gene promoter. Data are expressed as mean ± SEM. Statistics were assessed using tudent’s t-test. *P<0.05, **P<0.01, n=2.

97 2.4.7 Comparison of mice with an AE2 cell- and hindbrain neuron-specific Mecp2 deletion

As MECP2 directly regulates the expression of lipid metabolic enzymes, and its loss causes global lipid metabolism disarray in AE2 cells, I hypothesized that Mecp2 deficiency in lung AE2 cells alone should be sufficient to cause lung lipid abnormalities. Additionally, I reasoned that lipid abnormalities could also contribute to respiratory symptoms in Mecp2-mutant mice. To test this, mice with a lung AE2 cell-specific deletion of Mecp2 were generated. As respiratory symptoms of RTT are currently attributed to neuronal loss of Mecp2, the effects of AE2 cell-specific Mecp2 deficiency were compared to those of a hindbrain neuron-specific Mecp2 deletion. Male mice were used for these experiments because conditional deletion would result in a hemizygous loss of Mecp2 in the targeted cells.

The procedure for generating these conditional deletion mice is shown in Figure 2.7A. Mecp2 was removed from AE2 cells by breeding B6.Mecp2tm1Bird mice, in which exons 3 and 4 of the Mecp2 gene are flanked by loxP sites (Mecp2-flx), to B6.Sftpctm1(cre/ERT2)Blh mice, which express Cre under a tamoxifen-inducible promoter of the surfactant protein C (Sftpc) gene. Sftpc is expressed exclusively in the lung, where it is at its highest levels in AE2 cells. Cre-mediated excision of Mecp2 in resulting “Sftpc-CreERT2;Mecp2-flx” mice occurred upon induction at P21 using three injections of 50mg/kg of tamoxifen.

Simultaneously, Mecp2 was deleted from hindbrain neurons using Cre under the promoter of atonal BHLH Transcription Factor 1 (Atoh1), using Cg-Tg(Atoh1-cre)1Bfri mice. Atoh1 regulatory elements drive constitutive Cre expression in precursors of granule cell neurons of the cerebellum, dorsal hindbrain, and spinal cord. Resulting “Atoh1-Cre;Mecp2-flx” mice lack Mecp2 in the medulla oblongata and pons, the respiratory center of the brain responsible for generating and maintaining the rhythm of respiration.

As a final control, B6.Mecp2-null (Mecp2∆/Y) mice were generated using His-TAT-NLS tagged Cre recombinase induction in embryos. Briefly, 2-cell embryos were retrieved from pregnant Mecp2tm1Bird/+ mice, incubated in a solution containing His-TAT-NLS tagged Cre recombinase, and transferred to a pseudo-pregnant female recipient, as per Ryder, 2014. Importantly, the use of C57BL/6 mice in this experiment allows the assessment of lung lipids on a second genetic background, as the previous experiments described were done in 129SvEvS6 mice.

Using this strategy, PCR showed a complete deletion of Mecp2 in the lungs and hindbrain of Mecp2∆/Y mice, as expected. Additionally, Mecp2 was partially deleted in the lungs of Sftpc- CreERT2;Mecp2-flx but not in the hindbrain, while the hindbrain of Atoh1-Cre;Mecp2-flx mice

98 showed a partial Mecp2 deletion with no change in the lung (Figure 2.7B). Consistently, gene expression analysis confirmed that Mecp2 is not expressed in the hindbrain or lungs of Mecp2∆/Y, and that lung and hindbrain Mecp2 expression is decreased in Sftpc-CreERT2;Mecp2-flx and Atoh1-Cre;Mecp2-flx mice, respectively (Figure 2.7C,D).

Importantly, conditional deletion mice were assessed against wild type (+/Y), Cre-allele (Sftpctm1(cre/ERT2)Blh, or Cg-Tg(Atoh1-cre)1Bfri), and floxed-allele (Mecp2tm1Bird) littermates (Figure 2.8A, D). Tamoxifen injections were given to Sftpc-CreERT2;Mecp2-flx and all control littermates, blindly. The expression of Mecp2 was not altered in the hindbrain or lung of the littermate control animals, with the exception of Mecp2-flx mice, which showed a 20-30% decrease in hindbrain Mecp2 expression (Figure 2.8B, C, E, F), as the allele is hypomorphic (148).

Figure 2.7: Conditional removal of Mecp2 in lung AE2 cells or hindbrain neurons using Cre-Lox technology imparts a partial deletion in targeted cells/tissues. A. Diagram of breeding strategy. Mecp2-flx females were bred to Sftpc-CreERT2 mice to remove Mecp2 from lung AE2 cells. Mecp2-flx females were bred to Atoh1-Cre mice to delete Mecp2 from hindbrain neurons. Mecp2-flx females were bred to +/Y mice and embryos were injected with HTN-Cre at the 2-cell stage to delete Mecp2 ubiquitously in Mecp2-flx embryos, leaving +/Y mice unaffected. B. PCR assessment of the presence of the floxed Mecp2 allele or Mecp2 deletion in the lung and hindbrain of mice. Internal (Int.) control was used to confirm the presence of DNA in samples. Mecp2 mRNA expression was quantified by RT-qPCR in the C. lung and D. hindbrain of conditional deletion mice. Data are expressed as mean ± SEM. Statistics were assessed using tudent’s t-test. *P<0.05, ***P<0.001, n=2.

100

Figure 2.8: Mecp2 expression in conditional deletion littermate controls developed through Cre-Lox breeding strategy. A. Diagram of breeding scheme used to develop Sftpc-CreERT2;Mecp2-flx/Y mice. Sftpc-CreERT2 males were bred to Mecp2-flx females and resulting offspring (listed) were produced in equal proportions. All offspring were blindly injected with tamoxifen for induction of Cre-mediated excision. RT-qPCR was used to detect Mecp2 expression in the B. lung and C. hindbrain of offspring. D. Diagram of breeding scheme to develop Atoh1-Cre;Mecp2-flx mice. Atoh1-Cre males were bred to female Mecp2-flx mice and resulting offspring were produced in equal proportions. RT-qPCR was used to detect Mecp2 expression in the E. lung and F. hindbrain of offspring. Data are expressed as mean ± EM. tatistics were assessed using tudent’s t-test. *P<0.05, **P<0.01, ***P<0.001, n=2.

101 2.4.8 Hindbrain neuron-specific deletion of Mecp2 imparts neurobehavioral changes that are absent in mice with AE2 cell-specific Mecp2 deficiency

Male Mecp2/Y mice quickly develop features of RTT including limb clasping, hypoactivity, and poor muscle tone. Of these features, limb clasping is an overt phenotype; in comparison to male +/Y mice at 10 weeks of age, Mecp2∆/Y mice clasp both hindlimbs to their body, Atoh1- Cre;Mecp2-flx/Y mice have a partial limb clasp, and Sftpc-CreERT2;Mecp2-flx/Y are like wild type (Figure 2.9A). This was an early indication that Atoh1-Cre;Mecp2-flx/Y would have neurobehavioral abnormalities. Blind subjective health scoring (described in Methods) showed that, as previously described, Mecp2∆/Y mice have poor overall health compared to +/Y mice (final score: 7.44 ± 0.3; 0.42 ± 0.1, p=0.0006). Atoh1-Cre;Mecp2-flx/Y also had an increased subjective health score (4.3 ± 0.2, p=0.02) though their overall health was not severely affected.

Health scores of Sftpc-CreERT2;Mecp2-flx/Y were not significantly different than that of +/Y mice (Figure 2.9B). At 10 weeks old, both Mecp2∆/Y and Atoh1-Cre;Mecp2-flx/Y had deficits in rotarod activity compared to +/Y mice (p=0.001 and 0.0446, respectively), indicative of impaired motor coordination (Figure 2.9C). Motor activity was impaired in Mecp2∆/Y and in Atoh1-Cre;Mecp2- flx/Y mice as measured by distance travelled freely in the open field assay (3954 cm ± 232 in +/Y, 1620 ± 294 in Mecp2∆/Y, p<0.0001 and 2797 ± 183 in Atoh1-Cre;Mecp2-flx/Y, p=0.018) (Figure 2.9D). Neither motor coordination nor motor activity were altered in Sftpc-CreERT2;Mecp2-flx/Y mice. Social behavior was assessed using the three-chamber. In the first phase, mice were placed in the apparatus with access to a novel object on one side of the chamber and a stranger mouse on the other. Here, Mecp2∆/Y mice spent more time in the center chamber rather than exploring the three chambers (Figure 2.9E). In the second phase, mice were able to interact with the previous stranger mouse (now familiar) and a new stranger mouse. In this phase, Mecp2∆/Y mice spent more time interacting with the familiar mouse and less time in the chamber with the stranger (Figure 2.9F). This is consistent with previous reports of impaired social behavior in Mecp2-null models (130,362). However, Atoh1-Cre;Mecp2-flx/Y and Sftpc-CreERT2;Mecp2-flx/Y mice did not have altered social behavior compared to +/Y mice. Altogether, these results are as expected: Mecp2∆/Y mice have phenotypes represented in other Mecp2/Y models, Atoh1-Cre;Mecp2-flx/Y mice have slightly poorer overall health and motor impairments consistent with Mecp2 deletion in the hindbrain and dorsal cerebellum, and Sftpc-CreERT2;Mecp2-flx/Y mice do not have any neurobehavioral abnormalities.

Subjective health, rotarod, open field assays, and social behavior assessments were also performed in all littermate controls generated through breeding (+/Y, Mecp2-flx allele, Cre allele, with tamoxifen when necessary). Mecp2-flx mice had a slight increase in subject health score,

102 indicative of poorer overall health (Figure 2.10A, 2.11A). This is unsurprising as previous reports (363), and expression data presented here, show the hypomorphic nature of the Mecp2-flx allele. No other significant neurobehavioral changes were found in +/Y, Sftpc-CreERT2; +/Y, or Mecp2- flx/Y mice treated with tamoxifen (Figure 2.10B-E). In contrast, neurobehavioral changes reported above in Atoh1-Cre;Mecp2-flx/Y mice were also significant when compared to their +/Y, Atoh1-Cre; +/Y and Mecp2-flx/Y littermates (Figure 2.11B-E).

103

Figure 2.9: Neurobehavioral changes are present in male mice with a whole-body or hindbrain-specific deletion of Mecp2. A. Normal hindlimb positioning in +/Y and Sftpc-CreERT2;Mecp2-flx/Y mice, compared to hindlimb clasping to varying degrees in Mecp2∆/Y and Atoh1-Cre;Mecp2-flx/Y mice. B. Subjective health scores (n=10-20), C. performance on a rotating rod (rotarod) as an assessment of motor coordination (n=10-15), D. distance travelled in an open field as a measure of motor activity (n=11), and E-F. social behavior as measured by the three-chamber assay (n=8), were assessed in mice. Data are expressed as mean ± SEM. Statistics were assessed using one-way ANOVA with the Tukey’s multiple comparison test. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

104

Figure 2.10: Neurobehavioral assessments in Sftpc-CreERT2;Mecp2-flx/Y mice and their littermate controls. A. Subjective health scores (n=8-10), B. performance on a rotating rod (rotarod) as an assessment of motor coordination (n=8-10), C. distance travelled in an open field as a measure of motor activity (n=8-11), and D-E. social behavior as measured by the three-chamber assay (n=6), were assessed in mice. Data are expressed as mean ± SEM. Statistics were assessed using one-way ANOVA with the Tukey’s multiple comparison test. *P<0.05.

105

Figure 2.11: Neurobehavioral assessments in Atoh1-Cre;Mecp2-flx/Y mice and their littermate controls. A. Subjective health scores (n=8-10), B. performance on a rotating rod (rotarod) as an assessment of motor coordination (n=8-10), C. distance travelled in an open field as a measure of motor activity (n=10), and D-E. social behavior as measured by the three-chamber assay (n=6- 8), were assessed in mice. Data are expressed as mean ± SEM. Statistics were assessed using one-way ANOVA with the Tukey’s multiple comparison test. *P<0.05, ***P<0.001, ****P<0.0001.

106 2.4.9 Distinct respiratory symptoms in mice with AE2 cell- or hindbrain neuron-specific deficiency of Mecp2

Whole body plethysmography (WBP) is a non-invasive, quantitative method used to measure and analyze respiratory parameters. Unrestrained mice are placed in an enclosed chamber in which minute pressure changes are monitored and converted into respiratory waveforms. Respiratory parameters such as breathing frequency and tidal volume are then mathematically derived from these waveforms.

WBP was used to measure respiratory parameters in conditional deletion mice. Compared to +/Y mice, both Mecp2∆/Y and Sftpc-CreERT2;Mecp2-flx mice have an elevated baseline breathing frequency (378.93 breaths/min ± 3.6 in +/Y, 442.93 ± 17.4 in Mecp2∆/Y, p=0.008, and 452.77 ± 17.4 in Sftpc-CreERT2;Mecp2-flx, p=0.002) (Figure 2.12A). However, only Atoh1-Cre;Mecp2-flx mice show an increase in tidal volume compared to +/Y mice (0.24 ml ± 0.006; 0.27 ± 0.007, p=0.02) (Figure 2.12B). Thus, Mecp2 deficiency in the lung and hindbrain may impart distinct respiratory symptoms. Apneas, the most characteristic respiratory symptom of RTT patients, were also measured. Mecp2∆/Y mice had a drastic elevation in number of respiratory apneas compared to +/Y mice (0.08 apneas/min ± 0.03 in +/Y; 4.82 ± 1.34 in Mecp2∆/Y, p<0.0001). Interestingly, both Sftpc-CreERT2;Mecp2-flx and Atoh1-Cre;Mecp2-flx also showed a significant increase in apneas (1.71 ± 0.24, p=0.0002 and 1.25 ± 0.15, p=0.004, respectively) (Figure 2.12C). Thus, loss of Mecp2 in either the lung or hindbrain is sufficient to cause respiratory apneas. These results are intriguing; it is possible that MECP2 acts in both the lung and hindbrain to influence normal breathing and that the combined loss of Mecp2 in both centers synergistically creates the hallmark respiratory symptom.

To further probe respiratory features, conditional deletion mice were subjected to a respiratory challenge involving methacholine, a synthetic choline ester that acts as a non-selective muscarinic receptor agonist in the parasympathetic nervous system. Due to its charged quaternary amine structure, it is lipid insoluble and does not cross the blood-brain barrier. In the methacholine challenge assay, subjects inhale aerosolized methacholine, which causes airway constriction; airway responsiveness is measured by enhanced pause (Penh), a dimensionless index inferred through plethysmography waveforms. Here, mice were exposed to 0, 12.5, 25 and 50 mg/ml of aerosolized methacholine. Interestingly, Mecp2∆/Y mice consistently displayed an elevated Penh score over +/Y animals at each dose of methacholine, indicating pre-existing airway hyperreactivity (4.65 ± 0.38 at 50 mg/ml methacholine; 11.43 ± 0.85, p<0.0001) (Figure 2.12D). Notably, approximately 25% of Mecp2∆/Y mice had to be removed from the chambers during the 25 or 50 mg/ml dose due to highly arrhythmic breathing, which could lead to death.

107 Sftpc-CreERT2;Mecp2-flx also showed a strong response to high doses of methacholine, suggesting that AE2 cell-specific deficiency of Mecp2 affects airway reactivity (8.51 ± 0.78 at 50 mg/ml methacholine, p=0.0014). The response to methacholine in Atoh1-Cre;Mecp2-flx mice did not differ from +/Y (p>0.05), indicating that airway responsiveness induced by methacholine in Mecp2-mutant mice is independent of Mecp2 deficiency in hindbrain neurons.

Finally, lipid quantification was performed using HPLC and LC-MS/MS. Mecp2∆/Y mice had increased serum cholesterol (107.44 mg/dL ± 15.3 in +/Y; 173.68 ± 7.4 in Mecp2∆/Y, p=0.001) and triglycerides (35.57 ± 1.39 in +/Y; 60.07 ± 5.84 in Mecp2∆/Y, p=0.002) compared to +/Y mice (Figure 2.12D,E). Mecp2∆/Y mice also have increased lung triglycerides (6.82 mg/g ± 0.57 in +/Y; 17.16 ± 1.58 in Mecp2/Y, p=0.001), but not lung cholesterol (Figure 2.12F,G). In contrast, Sftpc-CreERT2;Mecp2-flx did not have an increase in serum triglycerides or cholesterol (p>0.05), but did show a significant increase in lung triglycerides (14.76 ± 3.59, p=0.043) compared to +/Y mice. Atoh1-Cre;Mecp2-flx mice did not show any changes in serum or lung lipids (p>0.05). BAL fluid PC 32/0 (DPPC) was decreased in both Mecp2∆/Y and Sftpc-CreERT2;Mecp2-flx mice (124.59 ± 8.51 μg/ml BAL in +/Y; 72.80 ± 12.37 in Mecp2∆/Y, p=0.0052; 80.13 ± 6.72 in Sftpc- CreERT2;Mecp2-flx, p=0.0087), along with other PCs (Figure 2.12I,J), but not in Atoh1- Cre;Mecp2-flx mice. Interestingly, BAL cholesterol was also decreased in Mecp2∆/Y and Sftpc- CreERT2;Mecp2-flx mice (38.99 ± 3.33 mg/ml BAL in +/Y; 23.44 ± 3.01 in Mecp2∆/Y, p=0.009; 27.14 ± 2.39 in Sftpc-CreERT2;Mecp2-flx, p=0.026) (Figure 2.12K). These results suggest lung lipid abnormalities are associated with respiratory symptoms in Mecp2-mutant mice.

Assessments in littermate controls of Sftpc-CreERT2;Mecp2-flx and Atoh1-Cre;Mecp2-flx mice were performed for respiratory symptoms (Figure 2.13A-D, 2.14A-D) and lipid quantification (Figure 2.13E-H, 2.14E-H). In both cases, no significant changes were found in any parameters across control mice (p>0.05). As above, Sftpc-CreERT2;Mecp2-flx mice had increased breathing frequency, apneas, response to methacholine, and lung triglycerides compared to their littermates. Similarly, Atoh1-Cre;Mecp2-flx mice had an increased tidal volume and apnea count compared to their littermates.

108

Figure 2.12: Respiratory features and lung lipids are altered in mice with whole-body and AE2 cell-specific deletions of Mecp2. Whole body plethysmography (WBP) was used to measure baseline A. breathing frequency, B. tidal volume, and C. apneas. D. A methacholine challenge was administered and response to increasing doses of methacholine were measured by enhanced pause (PenH) scores. HPLC was used to measure serum E. triglycerides (TG), F. cholesterol (chol.) and lung G. TG and H. chol. LC/MS-MS was used to measure BAL I. PC 32/0 (DPPC), J. PCs, and K. cholesterol. bpm: breaths per minute. Data are expressed as mean ± SEM. Statistics were assessed using one- way ANOVA with the Tukey’s multiple comparison test. *P<0.05, ***P<0.001, ****P<0.0001. n=11 for respiratory assays, n=5 for lipid quantification.

109

Figure 2.13: Respiratory features and lipid quantification assessments in Sftpc- CreERT2;Mecp2-flx compared to their littermate controls. Whole body plethysmography (WBP) was used to measure baseline A. breathing frequency, B. tidal volume, and C. apneas. D. A methacholine challenge was administered and response to increasing doses of methacholine were measured by enhanced pause (PenH) scores. HPLC was used to measure serum E. triglycerides (TG), F. cholesterol (chol.) and lung G. TG and H. chol. LC/MS-MS was used to measure BAL I. PC 32/0 (DPPC), J. PCs, and K. cholesterol. bpm: breaths per minute. Data are expressed as mean ± SEM. Statistics were assessed using one- way ANOVA with the Tukey’s multiple comparison test. *P<0.05, ***P<0.001. n=8-11 for respiratory assays, n=5 for lipid quantification.

110

Figure 2.14: Respiratory features and lipid quantification assessments in Atoh1- Cre;Mecp2-flx compared to their littermate controls. Whole body plethysmography (WBP) was used to measure baseline A. breathing frequency, B. tidal volume, and C. apneas. D. A methacholine challenge was administered and response to increasing doses of methacholine were measured by enhanced pause (PenH) scores. HPLC was used to measure serum E. triglycerides (TG), F. cholesterol (chol.) and lung G. TG and H. chol. LC/MS-MS was used to measure BAL I. PC 32/0 (DPPC), J. PCs, and K. cholesterol. bpm: breaths per minute. Data are expressed as mean ± SEM. Statistics were assessed using one- way ANOVA with the Tukey’s multiple comparison test. *P<0.05. n=8-11 for respiratory assays, n=5 for lipid quantification.

111 2.5 Discussion

As a primarily neurological disorder, RTT research has historically focused on the role of Mecp2 in the central nervous system. However, patient symptoms and phenotypes in Mecp2-mutant mice suggest important roles for Mecp2 outside of the brain. Despite respiratory disturbances causing a large portion of patient death, the lung is greatly understudied in the RTT field. In this chapter, I show that Mecp2 deficiency in mice results in lung lipid perturbations including the accumulation of neutral lipids and a drastic decrease in surfactant phospholipids. Further, I show that MECP2, in concert with the NCOR1/2 co-repressor complex, directly regulates the expression of lipid metabolism genes in the lung. Finally, lung AE2 cell-specific deletion is sufficient to cause respiratory symptoms and lung lipid abnormalities in mice.

2.5.1 Lung lipid abnormalities in Mecp2/Y lungs

An unexpected and intriguing finding of this study is that cholesterol and triglycerides are increased in the lungs of Mecp2/Y mice, while PCs, which are required for normal surfactant function, are decreased, indicating a global perturbation of lung lipid metabolism. Notably, these changes were also evidenced in female Mecp2/+ mice, indicating that heterozygous mosaic loss of Mecp2 is sufficient to alter lung lipids. Lipid metabolism abnormalities have been previously evidenced in Mecp2-mutant mouse models and RTT patients. Symptomatic Mecp2/Y male mice have increased brain cholesterol, serum cholesterol and serum triglycerides (167). Mecp2/Y and Mecp2/+ mice also develop fatty liver disease and dyslipidemia, findings that are replicated when Mecp2 is removed only in hepatocytes (164). Further, a subset of RTT patients show increased serum cholesterol and triglycerides (15,16).

While lung cholesterol was increased at P21 and P56 in male Mecp2/Y mice, it was only increased in BAL at P21, and was not altered in the BAL of female Mecp2/+ mice. All cells in the lung are capable of synthesizing cholesterol, while measurements in the BAL reflect only the pool of cholesterol synthesized by AE2 cells. Increased lung cholesterol may represent cholesterol sequestered in lipid droplets throughout the lung. Meanwhile, normalization of BAL cholesterol over time is unsurprising. Multiple feedback mechanisms exist to limit cholesterol overproduction when high cellular cholesterol is detected (364): 1) SREBPs become less active at the sites of their cholesterol biosynthesis target genes; 2) Farnesol, a downstream derivative of the mevalonate pathway, inhibits the translation of the Hmgcr mRNA; and 3) HMGCR becomes increasingly susceptible to ER-associated degradation. Additionally, despite Hmgcs1 being a direct transcriptional target of MECP2, it is upstream of HMGCR, and thus the feedback mechanisms listed allow cholesterol regulation to occur. Finally, cholesterol is excreted from the lung through reverse cholesterol transport, a mechanism that relies on cholesterol transport

112 proteins. In contrast, triglyceride synthesis feedback is only achieved through the activity of SREBPs, and lipolysis occurs through beta-oxidation in the mitochondria. As mitochondrial function may be impaired in Mecp2/Y mice, lung triglycerides likely continue to accumulate despite the need for fatty acid degradation.

In the Mecp2/Y liver, the upregulation of lipid biosynthesis pathways and resulting cholesterol and triglyceride accumulation is independent of SREBP activity (164). SREBPs are essential transcriptional regulators of sterol and fatty acid biosynthesis. Surprisingly, Srebf1, which encodes SREBP1 proteins, was expressed at low levels in Mecp2/Y AE2 cells. Consistently, the expression of SREBP1 protein was decreased in Mecp2/Y lungs at P21 but returned to normal levels by P56. As a lipid-sensing regulator, SREBP activity is dependent on cellular lipid levels; thus, decreased mRNA expression at P18 and protein expression at P21 suggests that lipid accumulation is already rampant in Mecp2/Y lungs prior to these early timepoints. While the decreased expression of fatty acid biosynthesis genes Acly, Fasn, Scd1 and Scd2 in Mecp2/Y AE2 cells was surprising, given the drastic accumulation of triglycerides in Mecp2/Y lungs, these genes are all targets of SREBP activation. Thus, reduced SREBP expression, due to sensing high cellular lipid levels, is the likely mechanism for the decreased expression of its lipid biosynthesis transcriptional targets in Mecp2/Y AE2 cells. Notably, Plin2, a gene involved in lipid droplet formation, is expressed at high levels in Mecp2/Y AE2 cells. While low nuclear SREBP1 levels could act to reduce lung triglyceride synthesis in Mecp2/Y lungs, increased Plin2 expression could facilitate the sequestration of lung lipids into lipid droplets. Consistently, deactivation of SREBP in mice leads to increased lipid storage (365).

PCs are synthesized by the CDP:choline pathway; single cell RNA-sequencing revealed that Chka and Pcyt1a, two enzymes responsible for generating CDP-choline, are remarkably underexpressed in Mecp2/Y AE2 cells. A low pool of CDP-choline would lead to reduced substrate availability for PC synthesis. Consistently, DPPC (PC16:0/16:0), which makes up 60% of lung surfactant and is responsible for lowering lung surface tension, is drastically reduced in Mecp2/Y BAL fluid. Interestingly, Pcyt1a is also a target of SREBP-mediated activation, and mice with a deactivation of Srebp have decreased lung PCs (336). These results further suggest that lipid metabolism is altered earlier than P18 in Mecp2/Y lungs, perhaps in a key developmental window, leading to profound lifelong consequences in surfactant components and lipid storage. RNA- sequencing assessments at earlier and later time points will reveal more insight into metabolic changes occurring in the mouse lung in the absence of Mecp2.

Additionally, Abca3 is significantly underexpressed in Mecp2/Y AE2 cells. ABCA3 transports newly synthesized phospholipids into lamellar bodies to form functional surfactant and, when

113 mutated, alters surfactant composition and function. Over 100 Abca3 gene mutations causing surfactant dysfunction have been identified in human patients, many of which are fatal. Non-fatal surfactant dysfunction causes wheezing, hyperventilation, and decreased blood oxygen levels (366). Changes in phospholipid composition in the lung surfactant of Mecp2-deficient animals alone could be a major driving force in respiratory symptoms in RTT. Notably, surfactant dysfunction also causes ground glass opacities and bronchial thickening in computed tomography (CT) scans, both of which were evidenced in RTT patient lungs (333).

2.5.2 MECP2, in concert with the NCOR1/2 corepressor complex, directs lung lipid metabolism

The proposed primary role of MECP2 is to act as a bridge between methylated DNA and the NCOR1/2 co-repressor complex. RTT-causing missense mutations cluster in its methyl-binding domain (MBD) and its NCOR-interaction domain (NID), highlighting the importance of these two regions (89). NCOR1/2 represses transcription in part by the removal of histone acetyl marks by HDAC3. The NCOR1/2 co-repressor complex cannot readily bind to gene promoters in the absence of MECP2, causing the aberrant transcription of target genes (367). Here, MECP2’s transcriptional regulation of lipid genes in lung AE2 cells with NCOR1/2 were tested by performing ChIP using an antibody against NCOR1/2 complex member, TBLX1R1.

Interestingly, TBLX1R1 binds to the Acot1 and Hmgcs1 gene promoters with a lower affinity in Mecp2/Y lung cells compared to cells from the +/Y lung. Both Acot1 and Hmgcs1 were expressed at higher levels in Mecp2/Y AE2 cells. Fatty acid catabolism, or beta-oxidation, is achieved in the mitochondria. ACS enzymes join fatty acids with CoAs to form acyl-CoA, which can be transported into the mitochondria for degradation. In contrast, ACOTs catalyze the opposite reaction, forming free fatty acids as an end product. Together, ACOTs and ACSs direct the metabolic fate of fatty acids. In the lung, loss of Mecp2 prevents the NCOR1/2 complex from binding to the Acot1 promoter and repressing its transcription. This effectively reduces the efficiency of substrate production for beta-oxidation by shuttling fatty acids away from degradation. Thus, the loss of Mecp2 and its concurrent regulation of Acot1 could provide the means for triglyceride accumulation as seen in Mecp2/Y lungs.

The Hmgcs1 gene encodes the enzyme that catalyzes the reaction of acetyl-CoA and acetoacetyl- CoA to form 3-hydrocy-3-methylglutaryl-CoA (HMG-CoA), the substrate for the rate-limiting step of cholesterol biosynthesis catalyzed by HMGCR. The increased expression of Hmgcs1 in Mecp2/Y AE2 cells is not surprising considering cholesterol is increased in P21 and P56 Mecp2/Y lungs. As a direct target of MECP2-mediated transcriptional repression, Mecp2 deficiency in the lung likely leads to the upregulation of this gene, and a resulting upregulation of the cholesterol

114 biosynthesis pathway. Interestingly, Hmgcs1 is also a target of SREBP2-mediated transcriptional activation in the liver (368) and SREBP2 expression was unchanged in Mecp2/Y lungs. It is important to note that transcriptional regulation is very complex, and each gene can be activated and repressed by multiple enzymes and complexes. Hmgcs1 could be activated by SREBP2 and repressed by MECP2 (and NCOR1/2), or its epigenetic regulation could vary across tissues.

2.5.3 AE2-cell specific deletion of Mecp2 in mice is sufficient to cause lung lipid abnormalities and respiratory symptoms

The comparison of respiratory symptoms in mice with AE2 cell-specific depletion of Mecp2 to mice with a hindbrain-neuron specific deletion of Mecp2 revealed many interesting findings. Most importantly, loss of Mecp2 in lung cells alone can cause respiratory symptoms; mice with an AE2 cell-specific deletion of Mecp2 had an increased baseline breathing frequency, increased number of apneas, and an elevated response to methacholine. Consistently, these mice also had elevated lung triglycerides in the absence of changes to serum lipid levels, indicating that lung triglyceride accumulation is caused by local cellular dysfunction rather than lipids infiltrating the lung from the circulatory system. They also have decreased BAL PC 32/0, suggesting increased lung triglycerides alone can cause a consequent decreased in BAL PCs.

Interestingly, lung cholesterol was not significantly altered in mice with an AE2 cell-specific deletion of Mecp2, while BAL cholesterol was decreased (at P70). Lung cholesterol is likely unchanged because up to 80% of lung cholesterol is derived from the serum (369). In mice where liver Mecp2 expression is normal, serum cholesterol, and therefore lung cholesterol, would not be elevated. In contrast, a late-stage decrease in BAL cholesterol is more intriguing. Strain-specific differences in cholesterol metabolism have been previously reported (370) though these findings could suggest cholesterol biosynthesis is elevated in AE2 cells early in the course of disease and feedback mechanisms lower BAL cholesterol by P70.

The methacholine challenge is used as a diagnostic tool in humans since methacholine causes mild narrowing of the airways, mimicking asthma. A drop in lung function of 20% is considered a positive test score and warrants a respiratory disease diagnosis. In mice, enhanced pause (Penh) is used to evaluate airway reactivity during a methacholine challenge. The finding that mice with a lung-specific deletion of Mecp2 respond to lower doses of methacholine compared to wild type mice provides concrete evidence for lung involvement in RTT. These results are extremely relevant for RTT patient care. A subset of RTT patients are diagnosed with asthma and other respiratory diseases (31,371), and findings using the methacholine challenge here suggest the RTT lung could have an inflammatory component. Consistently, lung inflammation has been evidenced in Mecp2-mutant mice (333). Lung imaging may offer insights into changes in the RTT

115 lung and progress in technology in future years could make it possible to observe lipid accumulation in the lung using non-invasive techniques.

Respiratory apneas are considered the most striking respiratory symptom of RTT as they can lead to cyanosis and fainting (371). Intriguingly, mice with AE2 cell-specific and hindbrain neuron- specific deletions of Mecp2 both displayed an increase in the number of respiratory apneas. However, apneas in Mecp2∆/Y mice far exceeded these numbers. These results suggest that both neuronal- and lung-autonomous processes influence respiratory apneas, and loss of Mecp2 from both areas likely acts in a synergistic manner to produce the dramatic number of apneas seen in RTT patients and Mecp2-mutant mouse models.

Additionally, breathing frequency was increased in mice with an AE2-cell deletion of Mecp2, while tidal volume was changed in mice with a hindbrain neuron-specific deletion of Mecp2, suggesting that loss of Mecp2 from either the lung or hindbrain also imparts distinct respiratory symptoms. These results parallel a 2006 study that compared respiration in mice with a neuron-specific Nestin-Cre deletion of Mecp2 to mice with heterozygous loss of Mecp2 (Mecp2/+), finding that each had distinct respiratory phenotypes (330). However, while these results are intriguing, it is important to note that the brain is a complex network of different cells and the deletion of Mecp2 from neurons of one brain region may not be a true representation of Mecp2’s function in the region when other cells also lack Mecp2.

2.5.4 Mitochondrial genes are impacted in Mecp2/Y AE2 cells

An unexpected finding of this study was a severe decrease in the expression of mitochondrial- encoded components of the electron transport chain (ETC) in Mecp2/Y AE2 cells. These findings add to a growing body of evidence for mitochondrial involvement in RTT. Many features of RTT overlap with mitochondrial diseases, including early symptomatic onset, developmental delay, motor regression, and seizures. Respiratory failure is the leading cause of death in Leigh syndrome, a mitochondrial disease in which 25% of cases are caused by mutations in mitochondrial DNA. Both RTT patients and Mecp2-mutant mice display abnormal mitochondrial structure, altered ETC function, and increased oxidative damage (19,357,372–374). Additionally, Mecp2-null cells have a decreased mitochondrial membrane potential indicative of decreased ETC activity and/or increased proton leaks (359). The drastically decreased expression seen here is likely due to the power of single-cell sequencing which allowed us to assess the transcriptome of metabolically-active AE2 cells alone.

To our knowledge, few single cell RNA-sequencing studies have been carried out on RTT patient or Mecp2-mutant mouse samples; thus, changes in mitochondrial gene expression is likely hidden

116 because different cell types have different energy needs. The decreased expression of mitochondrial genes in Mecp2/Y AE2 cells could represent fewer mitochondria or decreased ETC function, which could lead to decreased ATP production. How mitochondrial underperformance contributes to the pathogenesis of RTT, and specifically lung metabolism, remains to be studied.

2.5.5 Implications for RTT

RTT patient death is most often caused by lung infection, asphyxiation, and respiratory failure (31). Additionally, 17% of RTT patients develop pneumonia, 6% develop asthma, 6% have bronchitis, and 4% have other respiratory illnesses (371). The severe misregulation of surfactant PCs seen here can cause surfactant dysfunction, which can diminish normal lung function and lead to respiratory failure. Accumulation of neutral lipids in the lung could contribute to the increased occurrence of lung infection and pneumonia in RTT patients: lipid droplets can occupy macrophage clearance efforts, thereby decreasing their efficacy in preventing infection, and/or advantage pathogenic bacteria by providing a microenvironment suitable for their proliferation.

Mecp2 is a crucial regulator of lipid metabolism in the lung with the NCOR1/2 corepressor complex. Loss of Mecp2 leads to the increased expression of its target genes, having detrimental effects on respiratory symptoms. Notably, lipid metabolism is highly targetable through pharmaceutical drugs; normalizing the expression of lipid metabolism genes in lung AE2 cells could improve surfactant quality and efficiency and clear lipid accumulation, likely improving respiratory symptoms in RTT. Importantly, lung AE2 cell-specific deletion of Mecp2 is sufficient to cause respiratory symptoms. The lung should therefore be considered by physicians when evaluating respiratory symptoms in RTT patients.

117

Chapter 3 Structural and functional changes in the lung of a Mecp2-mutant mouse model of Rett syndrome

118 tructural and functional changes in the lung of a Mecp2 mutant mouse model of Rett syndrome

3.1 Abstract

Abnormal gas exchange is a feature of many respiratory disorders, including chronic obstructive pulmonary disorder, idiopathic pulmonary fibrosis, and pneumonia. Respiratory symptoms are a major feature of Rett syndrome (RTT), a severe neurological disorder caused by mutations in the X-linked gene, methyl-CpG-binding protein 2 (MECP2). Here, I show that respiratory symptoms precede the onset of overt neurobehavioral symptoms in Mecp2-mutant mice. RNA-sequencing revealed drastically decreased expression of extracellular matrix (ECM) genes in Mecp2/Y lungs including genes encoding elastin, collagen, and proteoglycans essential for normal ECM function. Accordingly, adult Mecp2-mutant lungs have enlarged airways and alveolar airspaces and an emphysema-like phenotype. These structural changes impair pulmonary function in both male and female Mecp2-mutant mice, suggesting that heterozygous mosaic expression of Mecp2 is not sufficient to maintain normal lung mechanics. Finally, AE2 cell-specific loss of Mecp2 also impairs lung structure and pulmonary function. These results show that loss of Mecp2 has consequences on lung structure and impairs lung function, findings with immense implications for RTT patients.

119 3.2 Introduction

The mammalian lung has four requirements for gas exchange: 1) a tree of bronchi and bronchioles that transports air from outside of the body to and from the gas exchange region, 2) a large surface area within the gas exchange (alveolar) region with a thin air-blood barrier, 3) an effective vascular system that is in close contact to the alveoli, and 4) a surfactant-generating system that facilitates breathing (273). The development of these components occurs over five stages: embryonic, pseudoglandular, canalicular, saccular, and alveolar.

Gas exchange is disrupted in several respiratory disorders. Two major classes of respiratory disorders are obstructive and restrictive lung disease. Obstructive lung diseases, such as asthma, bronchiectasis, and chronic obstructive pulmonary disease (COPD, formerly emphysema) are characterized by shallow exhalations due to obstruction in the airways. In contrast, restrictive lung diseases, including idiopathic pulmonary fibrosis (IPF) and pneumonia, are characterized by lung stiffness and difficult inhalation. Both types of respiratory diseases cause difficulty breathing and can lead to death.

COPD and IPF are considered hallmark obstructive and restrictive lung disorders, respectively. COPD affects up to 600 million people worldwide. It is thought to result from chronic inflammation in response to inhaled irritants; this results in the obstruction of small airways, alveolar wall degradation, and loss of elastic recoil of the lung. In contrast, IPF results from aberrant healing after lung injuries. IPF is characterized by lung scarring and an excessive deposition of disordered collagens and other extracellular matrix (ECM) proteins.

The diagnosis of respiratory disease is often aided by pulmonary function tests. In humans, spirometry is the most common method of measuring lung function, specifically, the volume and flow of air moved during respiration. During the test, the patient is asked to complete a series of inhalations and exhalations into a spirometer. The most common parameters measured are total lung capacity, forced vital capacity which represents the volume of air exhaled after one inspiration, and forced expiratory volume which represents the volume of air exhaled in one second. Plethysmographs may also be used to diagnose respiratory disorders. They are generally used to measure residual functional capacity of the lungs, which is the volume of air left in the lungs after expiration.

In preclinical studies, plethysmography is the most common method used to study respiration because it is non-invasive and can be performed on conscious, unrestrained subjects. Common parameters measured by plethysmography include breathing frequency, tidal volume, and enhanced pause, which provide indirect measurements of respiration. Plethysmography allows

120 the detection of abnormal respiration, although the measures may be influenced by neurological processes (ex. anxiety or various disorders) and environmental stimulation (ex. noise, light, etc.). Therefore, to directly measure pulmonary function in animal models, forced oscillation techniques (FOT) are used. FOT is an invasive and terminal procedure performed by applying an oscillatory waveform to the subject’s airway opening. Using this method, respiration is completely controlled by the apparatus, allowing the measurement of lung mechanics, including total lung capacity, airway resistance, pulmonary elastance, compliance, and more. Together, these techniques can determine if unrestricted respiration is altered and if there are changes in lung mechanics that contribute to these changes.

Respiratory symptoms are a major challenge in Rett syndrome (RTT), a neurological disorder caused by mutations in methyl-CpG-binding protein 2 (MECP2). Current estimates suggest nearly all RTT patients will experience breathing dysfunction, including arrhythmic hyperventilation mixed with prolonged apneas and breath-holding (371). These breathing disturbances in RTT patients often lead to cyanosis and fainting (320). Further, up to 80% of RTT patients die due to respiratory abnormalities including asphyxiation and respiratory failure (31). Thus far, only one study has examined RTT patient lungs through high resolution computed tomography (HRCT) imaging. Interestingly, 55% of patients assessed had abnormal lung findings, including centrilobular nodules (67%), thickened bronchial walls (53%), non-specific ground glass opacities (27%), and bronchiectasis (60%). Further, pulse oximetry abnormalities were found in all patients, suggesting altered respiration in RTT breathing (333). Despite these findings, few studies have examined the Mecp2-deficient lung in detail (334).

Here, I found that Mecp2-mutant mice develop respiratory symptoms prior to the onset of other overt neurological symptoms. Because MECP2 is expressed in many different cell types in the mouse lung, I performed RNA-sequencing on postnatal day (P) 21 +/Y and Mecp2/Y mouse lungs. Interestingly, genes encoding essential extracellular matrix (ECM) proteins were expressed at strikingly low levels in Mecp2/Y lungs. Adult male and female Mecp2-mutant mice have enlarged bronchioles and alveolar spaces resulting in abnormal lung function, suggesting bronchiecstasis and emphysema-like changes upon loss of Mecp2. Further, AE2 cell-specific loss of Mecp2 is sufficient to alter lung structure and function. These findings implicate Mecp2 as an essential protein in regulating lung homeostasis.

121 3.3 Methods

3.3.1 Animals

All animal procedures were approved by the Animal Care Committee at the CCAC-accredited animal facility, The Center for Phenogenomics (TCP). Congenic 129.Mecp2tm1.1Bird/Y mice feature a deletion of the last two exons (exons 3-4) of the Mecp2 transcript, resulting in a null allele. Male Mecp2/Y (Mecp2-null) and +/Y (wild type), and female Mecp2/+ and +/+ mice were obtained by backcrossing Mecp2tm1.1Bird/+ females to males of the 129SvEvS6/Tac strain. Mice were fed a standard diet (Harlan Teklad 2918) ad libitum, consisting of 18% protein, 6% fat, and 44% carbohydrates. Mice were housed in a 13-hour light/dark cycle and were euthanized between the hours of 9AM and 12PM (ZT 2-5) to control for circadian rhythm fluctuations.

3.3.2 Plethysmography

Respiration was monitored using a Buxco Whole Body Plethysmography (WBP) apparatus (Data ciences International) according to manufacturer’s instructions. All testing was conducted between the hours of 9AM and 12PM. Mice were placed in plethysmography chambers and allowed to acclimate for 30 minutes, until motionless. Baseline breathing rates were measured for a period of 5 minutes. Following this, mice were exposed to nebulized saline for 2 minutes, and respiratory rates were measured for 5 minutes. Breathing frequency and tidal volume were analyzed using Buxco FinePoint Software. Apneas were defined as cessation of breathing for over 1 second (2 respiratory cycles) and were counted manually over a period of baseline breathing.

3.3.3 Immunohistochemistry and histology

For immunohistochemistry, lungs of 3-week-old Mecp2tm3.1Bird mice were perfused with 10 U/ml heparin in PBS and fixed via intra-tracheal administration of 4% paraformaldehyde (PFA). Excised lungs were immersed in 4% PFA for 16 hours at 4°C. Tissues were washed through a series of PBS washes, embedded in paraffin, cut at a thickness of 5μm, and dried on a slide warmer at

122 37°C. Mounted sections were deparaffinized in a graded series of ethanol washes. Antigen retrieval was performed in 0.01M citrate buffer, pH 6.0. Slides were blocked with 5% goat serum, 1% bovine serum albumin, and 0.1% Triton X-100 in PBS. Slides were incubated with primary antibodies against: GFP (Abcam ab13970, 1:500), prosurfactant protein C (Abcam ab90716, 1:500), uteroglobin (Abcam ab40873, 1:500), podoplanin (R&D Systems AF3244, 1:500), vimentin (R&D Systems MAB2105, 1:500), and elastin (ab21607, 1:300). Secondary antibodies used were: Alexafluor 488 (Thermo A-11039, 1:400) and Alexafluor 594 (Thermo A-11012, 1:400). Slides were mounted using ProLong Gold Antifade Mount with DAPI (Thermo Fisher) and imaged on a Nikon A1R confocal laser microscope equipped with NIS Elements analysis software.

H&E- and Verhoeff’s Van Gieson-stained sections were visualized using a 3DHistech Slide Scanner with a PCO.edge 4.2 cMOS (FL) camera. Images were visualized using CaseViewer software. For morphometric analysis, images were taken using a light microscope. Analysis was performed manually as follows: ten random fields of vision were imaged for each sample (n = 4 per group). Images were coded for blind analysis. A custom cross (5 cm x 5 cm) was placed in the center of each image and the number of intersections with alveolar walls were counted as a single intercept, while intersections with blood vessels were counted as 0.5 intercept. The mean linear intercept (Lm) was calculated using the following equation:

푁 푥 퐿 퐿푚 = 푀 where N = number of traverse lines (2, cross), L = length of traverse (5 cm), and M = sum of all intercepts in a single image.

3.3.4 RNA extraction

RNA for RNA-seq or qRT-PCR was extracted from lungs of mice dissected at 9:00 – 10:00 am (ZT 2-3), unless otherwise specified. Mice were cervically dislocated and the thoracic cavity was opened. Lungs were carefully removed, flash frozen in liquid nitrogen, and stored at -80 °C until further processing. A 5 mm stainless steel bead (Qiagen, 69989) was added to chilled tubes containing lung tissue, which was then homogenized in 400 uL of Lysis Reagent (Qiazol) using a Qiagen TissueLyser II. RNA extraction was carried out using the RNeasy Lipid Tissue Mini Kit (Qiagen, CA, USA) and stored at -80 °C.

3.3.5 RNA sequencing and analysis

RNA from lungs of 3 wildtype and 3 Mecp2 null animals was assessed for integrity using the Agilent RNA 6000 Nano kit (Agilent Technologies) on the Agilent 2100 Bioanalyzer. All RNA integrity numbers (RIN) were above 9.8 with an average score of 9.9, indicating high quality RNA

123 appropriate for sequencing. The RNA libraries were prepped using the NEBNext poly(A) mRNA magnetic isolation module which selects for mRNA and lncRNA with poly(A) tails by binding to oligo d(T)25 paramagnetic beads. The enriched fragment was fragmented for 6 minutes at 94C. Paired-end libraries were prepared and sequenced on an Illumina HiSeq 2500 platform at The Centre for Applied Genomics at The Hospital for Sick Children briefly as follows: fragmented RNA was converted to double-stranded 151 cDNA, end-repaired and adenylated at the 3’ end to create an overhand A to allow for ligation of TruSeq adapters with an overhand T. Library fragments were then amplified under the following conditions: initial denaturation of 98 °C for 10s, 14 cycles of 98 °C for 10s, 60 °C for 30s, 72 °C for 30s, and finally an extension step for 5 min at 72C. Each sample was amplified with a different barcoded adapter to allow for multiplex sequencing. Yield and size distribution were analyzed for purity using the Agilent 2100 Bioanalyzer. RNA libraries were quantified by quantitative PCR using the Kapa Library Quantification Illumina/ABI Prism Kit protocol (KAPA Biosystems). Libraries were pooled in equimolar quantities and paired-end sequenced on an Illumina HiSeq 2500 platform using a Rapid Run Mode flow cell and the V3 sequencing chemistry following Illumina’s recommended protocol to generate pair-end reads of 126 bases in length.

Reads were trimmed to remove adapters and low-quality ends using Trimgalore v0.3.3 resulting in 28,433,880-38,133,514 paired-end reads. Reads were aligned to the “UC C mm10” reference sequence using Tophat v. 2.0.11. Extraction and processing of reads was achieved using htseq- count v.0.6.1p2. Only uniquely mapped reads were counted. Principal component analysis was performed on raw gene counts to assess library distribution. Differential expression was performed using edgeR, R package v.3.8.6. Genes with counts per million reads < 1 in all samples were filtered out.

Gene set enrichment analysis (GSEA) (375) was used to determine whether a priori-defined gene sets would show statistically significant differences in expression between Mecp2/Y and wild type mice. Gene sets were downloaded from the Gary Bader laboratory website (http://baderlab.org/GeneSets) using the most recent available file, which was entitled “Mouse_GOBP_AllPathways_no_GO_iea_May_01_2020_symbol.gmt”, and contained gene sets from GO biological processes excluding annotations with evidence code IEA (inferred from electronic annotation), ND (no biological data available), and RCA (inferred from reviewed computational analysis) and all pathway resources currently available. Enrichment analysis was performed using the GSEAPreranked module using the list of all significantly altered genes, ranked by -log10FDR*sign of the fold change (ie. +1 or -1) with recommended default settings. Gene sets with an FDR <0.05 were considered significantly enriched. Enrichment plot was made in Cytoscape version 3.8.0 using the EnrichmentMap application. The node cut-off was set to q

124 <0.05 and default settings were used for other parameters. Clusters were manually assigned to terms encompassing pathways. Volcano plots were constructed using Galaxy, a web-based platform for computational biology analyses. Genes with an FDR <0.05 were highlighted.

3.3.6 Quantitative reverse transcription polymerase chain reaction (RT-qPCR)

3.3.7 Protein extraction and Western blotting

Lungs were dissected from +/Y and Mecp2/Y mice at postnatal day (P) 21 and 56. Tissues were flash frozen and stored at -80°C until further processing. Frozen tissues were transferred to round bottom tubes with a 5 mm stainless steel bead (Qiagen, 69989) and 300 µl of RIPA buffer (150mM NaCl, 1%NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 50mM Tris, pH 8.0 and fresh protease inhibitors (Sigma-Aldrich, 11873580001). Tissues were lysed in a TissueLyser II (Qiagen, 85300) at 50 Hz for 2-3 minutes. Protein concentration was determined using a Bradford assay (Thermo Fisher, 23200). Proteins were separated in a 4-15% gradient sodium dodecyl sulfate– polyacrylamide gel electrophoresis (SDS-PAGE) gel (BioRad, 4561084). Proteins were transferred to a polyvinylidene fluoride (PVDF) membrane using the Trans-Blot SD semi-dry transfer cell (BioRad). Following transfer, membranes were blocked in 5% milk-TBST. The following antibodies were used: anti-elastin (ab21607; 1:500), anti-GAPDH (Cell Signaling, 5174, 1:2000), peroxidase-Affinipure goat anti-rabbit IgG (Jackson ImmunoResearch, 111-035-144). Proteins were visualized using Clarity western ECL (BioRad, 170-5060) and imaged on a Chemi- Doc (BioRad).

125 Table 3.1: Primers used for RT-qPCR.

Gene Forward Primer Reverse Primer Tbp CCTTGTACCCTTCACCAATGAC ACAGCCAAGATTCACGGTAGA Eln TCCTGGGATTGGAGGCATTGCA ACCAGGCACTAAACCTCCAGCA Col1a1 CCTCAGGGTATTGCTGGACAAC CAGAAGGACCTTGTTTGCCAGG Col1a2 TTCTGTGGGTCCTGCTGGGAAA TTGTCACCTCGGATGCCTTGAG Col3a1 GACCAAAAGGTGATGCTGGACAG CAAGACCTCGTGCTCCAGTTAG Col6a2 TGGTCAACAGGCTAGGTGCCAT TAGACAGGGAGTTGACTCGCTC Clock CACCGACAAAGATCCCTACTGAT TGAGACATCGCTGGCTGTGT Arntl (Bmal1) CAACCTTCCCGCAGCTAACA TCCGCGATCATTCGACCTAT Npas2 GGGTCTGACTTGGCTTAGGG AGAGGCTCTCTTTGCTCTATCCT Per1 CAGGCTAACCAGGAAATATACGAG CACAGCCACAGAGAAGGTGTCCTG Per2 CTGGCTTCACCATGCCTGTT AAGGCCTGAGGCAGGTTTG Per3 GTGTACACAGTGTGCAAGCAAACA ACGGCCGCGAAGGTATCT Cry1 CCTCTGTCTGATGACCATGATGA CCCAGGCCTTTCTTTCCAA Cry2 AGGGCTGCCAAGTGCATCAT AGGAAGGGACAGATGCCAATAG Nr1d1 ACGACCCTGGACTCCAATAA CCATTGGAGCTGTCACTGTAGA

126 3.3.8 Lung function tests

Lung function was measured using the flexiVent FX system (SciReq). Mice were weighed and anesthetized with an intraperitoneal injection of ketamine/xylazine in saline (100mg/kg bodyweight ketamine/10mg/kg bodyweight xylazine). Mice were processed when they were completely unresponsive to a toe pinch stimulus. The throat region of the mouse was cleaned with ethanol, cut open, and the trachea was cannulated with a blunt 20G needle. The cannula was secured in place with sutures. The animal was then connected to the ventilator, avoiding cannular occlusion or tracheal twist. Mechanical ventilation was initiated through the flexiWare 7 software. The experiment proceeded once the mouse produced no spontaneous inspiratory efforts. A ‘deep inflation’ perturbation was performed to recruit closed lung areas and standardize lung volume. A software script initiated a sequence of baseline measurements performed in triplicate, with baseline periods between each measurement. Oscillatory waveforms associated with each measurement were applied and pressure, flow, and volume signals were recorded. Data were analyzed with the flexiWare 7 software.

3.3.9 Statistics

127 3.4 Results

3.4.1 Mecp2-mutant mice have breathing symptoms prior to neurological symptom onset

Mecp2/Y male mice develop neurobehavioral phenotypes, including hind limb clasping, hypoactivity, and tremors, beginning at 4 weeks of age. Alternatively, Mecp2/+ female mice develop neurological phenotypes between 4-6 months of age with symptom variability due to random X-chromosome inactivation. However, the onset of respiratory symptoms in Mecp2- mutant mice is varies depending on the reporting laboratory. Most studies report altered breathing symptoms beginning at 4 weeks in Mecp2/Y male mice (376,377), while others have reported an altered breathing rhythm in a subset of Mecp2/Y mice at postnatal day (P) 15 (378). In Mecp2/+ female mice, breathing symptoms have been reported at 3 months; however, due to skewed X- chromosome inactivation, approximately 30% of Mecp2/+ mice develop breathing symptoms as early as P21, an occurrence similar to male Mecp2/Y mice (379). In each of these cases, B6.Mecp2-mutant mice were tested and breathing symptoms occurred prior to the onset of neurological symptoms.

In our lab, we study Mecp2-deficiency on a 129SvEvS6/Tac strain background. Therefore, I tested whether breathing defects preceded neurological symptoms in 129.Mecp2-mutant mice. Whole body plethysmography (WBP), a non-invasive method used to analyze respiratory parameters, was utilized. Breathing frequency, tidal volume, and apneas were measured in wild-type and Mecp2-mutant mice before and after neurological symptom onset (P21 and P56 in male mice and 3 months and 6 months in female mice, respectively). Consistent with literature published on B6.Mecp2/Y mice, 129.Mecp2/Y mice have an increased breathing frequency at P21 (285.27 bpm ± 9.87 in +/Y; 350.84 ± 12.35 in Mecp2/Y, p=0.001) and at P56 (229.14 ± 8.13 in +/Y; 293.69 ± 9.68 in Mecp2/Y, p<0.005) (Figure 3.1A). However, their tidal volume is not significantly altered compared to +/Y mice at either time point (Figure 3.1B). Finally, while apneas are not present in Mecp2/Y mice at P21, the incidence of apneas increases drastically by P56 (0.09 ± 0.06 apneas/min in +/Y; 0.85 ± 0.13 in Mecp2/Y, p=0.001) (Figure 3.1C). In contrast, 129.Mecp2/+ female mice consistently have an altered breathing frequency at both 3 months (241.73 bpm ± 7.89 in +/+; 315.26 ± 19.1 in Mecp2/+, p=0.004) and 6 months (217.72 ± 12.2 in +/+; 308.02 ± 16.4 in Mecp2/+, p=0.0008). They also have an increased tidal volume at both ages (0.26 ± 0.01; 0.33 ± 0.01 at 3 months, p=0.0005, and 0.19 ± 0.007; 0.24 ± 0.009 at 6 months, p=0.004, in +/+ and Mecp2/+ mice, respectively). Finally, Mecp2/+ mice also had an increased incidence of apneas at both 3 months (0.029 apneas/min ± 0.018 in +/+; 0.37 ± 0.15 in Mecp2/+, p=0.047) and 6 months (0.1 ± 0.04 in +/+; 0.34 ± 0.09 in Mecp2/+, p=0.031) (Figure 3.1D-F). These findings demonstrate early onset breathing abnormalities in Mecp2-mutant mice.

128

Figure 3.1: Mecp2-mutant mice develop breathing symptoms prior to neurological symptom onset. Whole body plethysmography (WBP) was used to measure A. breathing frequency, B. tidal volume, and C. apneas in +/Y and Mecp2/Y mice at postnatal (P) 21 and 56. n=6-8. In female mice, D. breathing frequency, E. tidal volume and F. apneas were measured at 3 months and 6 months of age. n=6-7. Bpm: breaths per minute. tatistics were assessed using the tudent’s t- test. *P<0.05, **P<0.01, ***P<0.001.

129 3.4.2 MECP2 is highly expressed in the mouse lung

The finding that breathing irregularities precede neurological symptoms in Mecp2-mutant mice suggests that Mecp2 loss causes pathological changes in the respiratory process before it has detrimental effects in the brain. MECP2 is highly expressed across the entire brain, however, MECP2 is also highly expressed in the mouse and human lung (380). Thus, MECP2 expression and localization was assessed in the mouse lung.

Lungs of Mecp2tm3.1Bird mice were perfused, inflation-fixed, and processed for immunohistochemistry. Mecp2tm3.1Bird mice have a green fluorescent protein (GFP) reporter fused to the endogenous Mecp2 locus, allowing visualization of MECP2 through GFP expression. MECP2 was hardly expressed in alveolar epithelial 1 (AE1) cells, which were marked by podoplanin (PDPN), a cell-surface marker (Figure 3.2A) at P21. As shown in Chapter 2, MECP2 was most highly expressed in AE2 cells as shown by colocalization with the marker surfactant protein C (SPC) (Figure 3.2B). These findings are consistent with previous reports of MECP2 expression in the mouse (334) and human (339) lung. MECP2 was also expressed in secretory cells in the bronchiole (Figure 3.2C), as marked by expression of secretoglobin family 1A member 1 (SCGB1A1). Finally, MECP2 expression was noted in alveolar macrophages, marked by vimentin (VIM) staining (Figure 3.2D) and MECP2 staining of cells present in the alveolar space. These findings suggest MECP2 is important across various cell types in the mouse lung.

130

Figure 3.2: MECP2 is expressed in various cell types in the mouse lung. Lungs of Mecp2tm3.1Bird mice were stained with DAPI, GFP to visualize MECP2 expression, and markers for A. alveolar epithelial 1 (AE1) cells, B. AE2 cells, C. secretory cells, and D. fibroblasts. Al: alveoli, Br: bronchiole, PDPN: podoplanin, SPC: surfactant protein C, SCGB1A1: secretoglobin family 1A member, VIM: vimentin. Arrows show co-localization. Scale bars represent 50 µm, n=2.

131 3.4.3 Transcriptional changes in the Mecp2-deficient lung

As MECP2 is expressed in many cell types in the mouse lung, it likely has numerous roles in lung homeostasis. To better understand the functions of Mecp2 in the lung, I performed RNA- sequencing on whole lung RNA extracts from +/Y and Mecp2/Y mice at age P21 (n=3). RNA- sequencing was performed by The Center for Applied Genomics (TCAG) at the Hospital for Sick Children. This technique was implemented to gain a broad view of transcriptional changes in the Mecp2-deficient lung and to pinpoint altered processes that could contribute to the onset of breathing symptoms in Mecp2/Y mice. Importantly, a pre-symptomatic time point was chosen to allow us to discern changes that occur due to loss of Mecp2 rather than downstream changes due to the diseased state.

I identified 1,721 genes that were differentially expressed in Mecp2/Y lungs with a false discovery rate (FDR) lower than 0.05. The 20 most significantly altered genes are shown in Figure 3.3A. After Mecp2, the two genes with the most significantly decreased expression were insulin-like growth factor binding protein 2 (Igfbp2, logFC: -1.549; FDR: 0.000025) and elastin (Eln, logFC: - 1.698; FDR: 0.00020687). IGFBP2 is a secreted inhibitor of insulin growth factor (IGF) signaling. Reduced IGF signaling has been evidenced in RTT, and systemic treatment with IGF1 improves neurobehavioral phenotypes and survival in Mecp2-mutant mice (188). In contrast, elastin has never been associated with RTT, likely due to its low expression in the brain. In the lung, elastin is an essential extracellular matrix protein that is independently responsible for the lung’s elastic recoil during exhalation. Elastin will be discussed in more detail below. The gene that was most significantly increased in Mecp2/Y lungs is secreted phosphoprotein 1 (Spp1, logFC: 0.951; FDR: 0.00020687). SPP1 is overexpressed in cancers of the breast, bladder, colorectal tract, liver, lung, and esophagus. SPP1 has also been associated with lung function; Spp1-deficient mice have a decreased total lung capacity and enlarged alveolar airspaces (381).

To better understand changes in Mecp2/Y lungs, I performed gene set enrichment analysis (GSEA) to identify major biological processes with statistically significant differences in gene expression between Mecp2/Y and +/Y mice. Subsequently, Cytoscape was used to design a molecular interaction network of proteins encoded by the differentially expressed genes within enriched gene sets. Each cluster within the network was manually assigned a larger, encompassing term (Figure 3.3B). For genes with decreased expression in Mecp2/Y lungs, gene sets for extracellular matrix proteins were most significantly enriched. Other enriched gene sets include those involved in cell adhesion and growth/differentiation. Additionally, immune response, innate immunity, leukocyte regulation, and injury response terms were also enriched for genes with decreased expression, suggesting altered immune function in Mecp2/Y lungs. Gene sets

132 containing genes with increased expression in Mecp2/Y lungs were most enriched in transcriptional regulation; this is interesting as the loss of Mecp2, an important transcriptional regulator, seemingly results in increased expression of other transcriptional regulators. Circadian rhythm and hormone receptors were also significantly enriched.

133

Figure 3.3: Whole lung RNA-sequencing reveals transcriptomic changes in Mecp2/Y lungs.

A. A total of 1,721 genes were differentially expressed in Mecp2/Y lungs with a false discovery rate (FDR) lower than 0.05. The 20 most significantly altered genes are shown. Colored dots in the Volcano plot represent significantly altered genes. B. Following gene set enrichment analysis (GSEA), Cytoscape was used for protein network analysis. A node cut-off of FDR <0.05 was used. Each node (circle) represents an enriched gene set. Node sizes correlate to the number of significantly altered genes in Mecp2/Y lungs contained within the gene set. Lines highlight related gene sets. Red: pathways enriched with genes with increased expression in Mecp2/Y lungs, blue: pathways enriched with genes with decreased expression in Mecp2/Y lungs.

134

3.4.4 Circadian rhythm is transcriptionally altered in Mecp2/Y lungs

Misregulation of circadian rhythm-related genes in Mecp2/Y lungs was an interesting and unexpected finding. The body’s central pacemaker of circadian timing is located in the suprachiasmatic nucleus (SCN) of the basal hypothalamus; while it synchronizes the entire body’s oscillations, peripheral tissues also have cell autonomous oscillators (382). RTT patients and Mecp2-mutant mice have disturbed sleep-wake cycles and sleeping disturbances (33,383), though the molecular regulation of circadian rhythm has not yet been studied in RTT models.

Core regulators of circadian rhythm operate in a continuous feedback loop (Figure 3.4A). During the wake period, aryl hydrocarbon receptor nuclear translocator-like protein 1 (ARNTL, formerly BMAL1) binds to circadian locomotor output cycles kaput (CLOCK) or its homolog, neuronal PAS domain protein 2 (NPAS2). The ARNTL-CLOCK/NPAS2 heterodimer binds to E-box enhancer elements in the promoters of its target genes. This effectively upregulates the expression of the period circadian protein homolog (PER) genes (Per1, Per2, Per3), cryptochrome circadian regulator (CRY) genes (Cry1, Cry2), basic helix-loop-helix family genes Bhlhe40 and Bhlhe41, and the nuclear hormone receptor Rev-ErbA alpha (NR1D1, formerly Rev-erbα). NR1D1 represses its target genes through binding Rev-erbα response elements (RREs). At the end of the wake period, accumulated PER and CRY proteins form heterodimers that translocate to the nucleus and inhibit the function of ARNTL-CLOCK/NPAS2. Simultaneously, BHLHEs bind to the E-box elements to further prevent transcription. NR1D1 dissociates from RREs, allowing transcription of genes, such as Arntl. By the end of the sleep period, PER and CRY genes are ubiquitinated and targeted for degradation, restarting the cycle.

Intriguingly, all core components of the clock machinery had altered expression in Mecp2/Y lungs. Lung samples for RNA-sequencing were collected at approximately 10:00 am (ZT3), when mice are in the sleep period. Despite this, expression of Arntl, Clock and Npas2 were significantly enhanced in Mecp2/Y lungs compared to +/Y lungs (Figure 3.4B). Additionally, targets of ARNTL- CLOCK/NPAS2 transcriptional activation (Per1, Per2, Per3, Cry1, Cry2, Bhlhe40, Bhlhe41, and Npas2) were expressed at high levels (Figure 3.4B). The expression profile of circadian genes in Mecp2/Y lungs therefore resembles that of the wakeful period, where ARNTL, CLOCK and NPAS2 are actively enhancing transcription of downstream clock components. These results suggest a misregulation of circadian timing, or possible “jetlag” in Mecp2/Y lungs.

To further characterize this misregulation, lungs of P21 mice were collected at two distinct points during the circadian rhythm: ZT0 (7:00am), when mice are entering the sleep period, and ZT12 (7:00pm), when mice are entering the wake period. At ZT0, Arntl, Clock and Npas2 should be

135 expressed at low levels, as NR1D1 is no longer inhibiting their transcription. Targets of ARNTL- CLOCK/NPAS2 activation will also be expressed at low levels as PER-CRY heterodimers begin to block ARNTL-CLOCK/NPAS2 activity. However, in Mecp2/Y lungs, Arntl and Clock expression is decreased compared to +/Y mice. Additionally, while the expression of Per1 and Cry1, two targets of ARNTL-CLOCK/NPAS2 activation, is decreased, Nr1d1 is drastically increased in Mecp2/Y lungs at ZT0 (Figure 3.4C). At ZT12, Arntl, Clock and Npas2 expression is expected to rise, while downstream genes are expected to have a low expression as they are only beginning to be actively transcribed. In Mecp2/Y lungs, Arntl, Clock and Npas2 are transcribed at higher levels than in +/Y lungs at ZT12. Additionally, Per3 and Cry2 are expressed at lower levels than in +/Y mice (Figure 3.4D). This expression profile is more consistent with that of a mouse in mid- sleep period, rather than at the start of the wake period.

136

Figure 3.4: Core circadian rhythm components are misregulated in Mecp2/Y lungs.

A. During the wake period, ARNTL and CLOCK/NPAS2 bind to the E-box of target genes to activate their transcription. Newly transcribed and translated NR1D1 binds to RREs to repress its target genes. PER, CRY and BHLHE proteins accumulate during the day until meeting a threshold level. During the sleep period, PER/CRY heterodimers inhibit ARNTL-CLOCK/NPAS2 binding and BHLHEs bind to E-box elements to prevent target gene transcription. NR1D1 is released from RREs. By the end of the sleep period, PER and CRY are ubiquitinylated for degradation. B. Volcano plot highlighting altered core circadian complex genes in Mecp2/Y lungs, sampled at ZT3. RT-qPCR analysis of circadian clock genes in P21 +/Y and Mecp2/Y lungs at C. ZT0 (7:00am) and D. ZT12 (7:00pm). tatistics were assessed using tudent’s t-test. *P<0.05, **P<0.001. n=3.

137 3.4.5 Extracellular matrix (ECM) genes are expressed at low levels in Mecp2/Y lungs

RNA-sequencing analyses revealed that ECM genes were transcriptionally altered in Mecp2/Y lungs. The lung’s unique three-dimensional ECM is made of an interconnected lattice of proteins that are essential for lung development and homeostasis. The ECM constitutes the lung’s structural architecture and provides it with mechanical stability, while also directing cellular differentiation. The lung ECM comprises the basement membrane, a thin, fibrous lining underlying the epithelial cells, and interstitial spaces, the areas surrounding blood vessels and alveoli.

ECM proteins are strikingly under-expressed in Mecp2/Y lungs (Figure 3.6A, Table 3.2). All the body’s elastin is encoded by the Eln gene, which makes a 70 kDa protein called tropoelastin. Tropoelastin molecules are crosslinked, through the action of lysyl oxidase enzymes, to make insoluble, polymeric elastin proteins (384). Elastic fibers also consist of 10-15 nm thick microfibrils that act as molecular bridges between adjacent cells. Mature elastic fibers, made of elastin and microfibrils, maintain the lung’s elastic recoil, allowing it to stretch during inhalation but return to its normal form during exhalation. Strikingly, Eln expression was decreased by 3.2-fold in Mecp2/Y lungs. Additionally, the expression of lysyl oxidase (Lox1, Lox2, and Lox3) genes, which code for enzymes involved in crosslinking elastin molecules, was also decreased in Mecp2/Y lungs. Finally, fibrillin, the major component of microfibrils, was also expressed at markedly reduced levels in Mecp2/Y lungs.

Elastin characteristically forms a network with collagen proteins in the ECM. While elastin provides elastic recoil to the lung, collagen contributes to its tensile strength and allows the lung to withstand deformation during breathing without suffering tissue damage. Surprisingly, 20 collagen genes were expressed at significantly lower levels in Mecp2/Y lungs. Importantly, this includes the fibrillar collagens (types I, II, III, and V) which make up 15-20% of the lung’s dry weight (Col1a1, Col1a2, Col3a1, Col5a1, Col5a2). Additionally, 10 adamalysins (Adam or Adamts genes) were underexpressed in Mecp2/Y lungs. The adamalysin proteinases mediate cell-ECM interactions through binding to integrins. Several adamalysins (ADAM10, 12, 15, ADAMTS2, 3, 14) are involved in collagen processing, and their decreased expression in Mecp2/Y lungs would further dysregulate the collagen network. The decreased expression of Eln and several collagen genes in Mecp2/Y lungs was validated by qRT-PCR analysis (Figure 3.6B)

Consistent with decreased adamalysin expression, several integrins were also expressed at low levels in Mecp2/Y lungs (Itga1, Itga2 and Itga3). Integrins are responsible for cell adhesion to the ECM and signal transduction. The integrin ligand, fibronectin (Fn1), is also expressed at decreased levels in Mecp2/Y lungs. Laminins, proteins which provide important structural scaffolding for pulmonary cells, were also expressed at low levels in Mecp2/Y lungs (Lama3,

138 Lama4, Lamb3, Lamc1, Lamc2). Finally, proteoglycans are important components in the ECM that interact with chemokines, growth factors, and cell surface receptors. Several proteoglycans, including the prominently expressed members Hspg2, Cspg4 and Hapln3, also showed reduced expression in Mecp2/Y lungs.

Western blot analysis showed that ELN protein was decreased in the lungs of Mecp2/Y mice at both P21 and P56 (Figure 3.6B). Thus, decreased mRNA expression of Eln correlates with low protein expression. Using a Verhoeff’s Van Gieson (EVG) stain, which stains nuclei black, elastin blue/black, collagen pink and muscle and blood cells yellow, I assessed +/Y and Mecp2/Y lungs at P7, P21, and P56 (Figure 3.6C). At P7, less collagen staining was observed around blood vessels and the alveolar area had less blue/black staining, indicative of decreased elastin. Notably, in +/Y lungs, elastin staining was evident throughout the developing septa, but was patchier and restricted in expression in Mecp2/Y lungs. At P21, +/Y lungs had dense collagen networks surrounding blood vessels and bronchioles, which were greatly reduced in Mecp2/Y lungs. This finding carried forward to P56 lungs as well. In the alveoli, elastin staining was markedly decreased at P56.

I used IHC to further study elastin deposition (Figure 3.6D). During lung development, elastin expression increases rapidly at embryonic day (E) 18 and reaches its peak between P10 – P14, within the period of alveologenesis. During this time, elastin forms a scaffolding matrix for fibroblasts to adhere to. Consistently, lungs of Eln-null mice fail to undergo alveolarization, resulting in death shortly after birth. Elastin generally appears in the lung in a “fiber and patch- like” pattern, increasing from P1 to P28 (385). At P7, when elastin levels are highest, +/Y mice show elastin staining along the walls of developing alveoli. In contrast, Mecp2/Y mice show patchy expression of ELN, suggesting abnormal elastin fiber construction or altered elastin deposition. At P21, ELN is seen in fibers lining the alveoli and in patches in +/Y mouse lungs. In Mecp2/Y lungs, the thickness of ELN fibers appears to be reduced. Finally, at P56, ELN was only present in fibers in Mecp2/Y lungs and appeared to be reduced in content.

139 Table 3.2: ECM genes are altered in Mecp2/Y lungs. Category Role Gene FC FDR Elastin Elastin Eln -1.69812 6.17E-08 and Collagens Fibrillar collagens Col1a1 -0.83555 3.14E-07 Col1a2 -0.64359 1.71E-06 Col3a1 -0.86952 4.6E-07 Col5a1 -0.4444 0.00023 Col5a2 -0.45259 5.11E-05 Non-fibrillar collagens Col4a1 -0.54707 9.71E-06 Col4a2 -0.30328 0.00046 Col4a5 -0.3177 0.000429 Col6a1 -0.25914 0.001738 Col6a2 -0.59621 4.11E-06 Col6a3 -0.7147 1.12E-05 Col8a1 0.268136 0.041717 Col12a1 -0.49504 0.003013 Col13a1 -0.40856 4.03E-05 Col14a1 -0.52438 4.42E-06 Col15a1 -0.9145 0.007516 Col16a1 -0.33333 0.000043 Col18a1 -0.35261 4.81E-05 Col20a1 0.298345 0.024461 Col27a1 0.189906 0.021164 Disintegrins Adam8 0.343387 0.028683 and Adam9 -0.15202 0.016975 metalloproteinases Adam10 -0.19796 0.008444 Adam17 -0.41083 7.53E-05 Adamts2 -0.72905 3.8E-07 Adamts9 -0.41679 2.65E-05 Adamts5 -0.24034 0.01064 Adamtsl1 -0.20758 0.01314 Adamtsl2 -0.87407 0.000103 Adamts12 -0.9189 2.85E-05 Lysyl oxidases Loxl1 -1.14922 6.03E-06 Loxl2 -0.1638 0.021139 Loxl3 -0.33367 0.008092 Integrins Itga1 -0.42131 0.002942 Itga3 -0.13629 0.026508 Itga4 -0.31146 0.014662 Itga5 -0.12779 0.017009 Itga6 -0.16374 0.03513 Itga8 -0.14345 0.043583 Itgav -0.19171 0.003236 Itgb2 -0.33778 0.00494 Itgb3 -0.22784 0.030573 Itgb4 0.217746 0.019476 Laminins Lama3 -0.34498 0.02457 Lama4 -0.24932 0.00309 Lamb3 -0.33698 0.001682 Lamc1 -0.20934 0.003916 Lamc2 -0.27417 0.008399 Other glycoproteins Fibrillin Fbn1 -0.78059 3.15E-06 Fibronectin Fn1 -0.72089 2.47E-06 Tenascin C Tnc -1.35395 4.12E-06 Sulfatase 1 Sulf1 -0.57661 0.042879 Proteoglycans Heparan Sulfate (Perlecan) Hspg2 -0.32442 0.000529 Biglycan Bcan -0.55779 0.008304 Chondriotin sulfate proteoglycan 4 Cspg4 -0.31254 0.003219 Hyaluronan and proteoglycan link protein Hapln3 -0.4658 0.018599 Glypican 2 Gpc2 -0.85012 0.001424 Glypican 4 Gpc4 -0.22921 0.034968

140

Figure 3.5: Altered expression of structural genes in Mecp2/Y lungs.

A. RT-qPCR confirmed the decreased expression of elastin and collagen genes in Mecp2/Y lungs. tatistics were assessed using the tudent’s t-test. *P<0.05, **P<0.01, n=4. B. Elastin (ELN) is expressed at low levels in Mecp2/Y lungs at postnatal day (P) 21 and P56. GAPDH was used for normalization. C. Verhoeff’s Van Gieson (EVG) stain highlighting nuclei (black), elastin (blue/black), collagen (pink) and other cells (yellow) in the +/Y and Mecp2/Y lung at P7, P21 and P56. Scale bar represents 10 µm, n=2. D. IHC of elastin (green) in the +/Y and Mecp2/Y lung at P7, P21 and P56. Scale bar represents 50 µm, n=2.

141 3.4.6 Alveolar structure is altered in Mecp2/Y mice

The changes in ECM gene expression in Mecp2/Y lungs at P21 occur during the alveolar stage of lung development, which occurs from birth to approximately P30. Thus, I assessed lung architecture during lung development using histology. Prior to P7, fibroblasts and other smooth muscle cells lay down a network of elastin and collagen fibrils for alveolarization, though septation is at its infancy. As a result, +/Y lungs have large airspaces (saccules) and immature, thick septa. These features are also seen in Mecp2/Y lungs, suggesting embryonic lung development occurs normally (Figure 3.6A). As lung development progresses, new septa are pinched off preexisting septa, dividing the existing airspaces at locations defined by fibroblasts. Clover-like alveoli create a large surface area for gas exchange. In the P21 +/Y lung, this is observed as an increase in airspaces and decrease in tissue surface area. The progressive expansion of the epithelial cells into thin-walled mature alveoli can be seen. In Mecp2/Y lungs, this process is also occurring, though alveolar walls appear to be thicker in some areas than in corresponding +/Y lungs (Figure 3.6B).

I next assessed lung structure in adult mice. In P56 Mecp2/Y mice, alveolar spaces are enlarged compared to +/Y mice (Figure 3.6C). Morphometry analysis indicated an increased mean linear intercept (Lm) in Mecp2/Y lungs, indicating an increased distance between adjacent alveolar tissue in the lung (Figure 3.6D). Alveolar enlargement is also present in 9-month-old female Mecp2/+ mice (Figure 3.7E). Consistently, morphometry analysis indicated these changes were significant (Figure 3.7F). Alveolar changes in adult Mecp2-mutant mice may be due to destruction of alveolar tissue. Alveolar wall degradation and airspace enlargement are characteristic of COPD and related disorders. Emphysema-like changes may occur in Mecp2-mutant lungs. Importantly, these changes in adult Mecp2-mutant mice diminish the gas exchange surface area and could impair normal respiration.

142

Figure 3.6: Emphysema-like changes in adult Mecp2-mutant lungs.

Hematoxylin and eosin (H&E) staining of +/Y and Mecp2/Y lungs at A. P7 and B. P21, during the alveolarization stage of lung development. cale bars represent 100 μm. C. H&E staining of male +/Y and Mecp2/Y lungs at P56, in early adulthood. D. H&E staining of female +/+ and Mecp2/+ lungs at 9 months of age. Scale bars represent 100 µm, n=3. Morphometry analysis was performed manually. Mean linear intercept represents the space between alveolar tissue. tatistics were performed using tudent’s t-test. *P<0.05, **P<0.01.

143 3.4.7 Bronchiolar enlargement in Mecp2-mutant lungs

As alveolar structure was altered in Mecp2-mutant lungs, I next assessed the bronchioles. The bronchioles represent the passages in the lungs that deliver air to the alveoli for gas exchange. Low power magnification of H&E-stained Mecp2/Y lungs revealed profound bronchiolar enlargement compared to +/Y lungs (Figure 3.7A). Through assessing the airways at a higher power magnification, I found some areas of the Mecp2/Y bronchioles had thickened walls with multiple layers of epithelial cells (Figure 3.7B). Intriguingly, despite heterozygous mosaic expression of Mecp2, 9-month-old female Mecp2/+ mice showed more severe bronchiolar enlargement than their male counterparts (Figure 3.7C). Consistently, Mecp2/+ airways also showed thickened epithelial walls (Figure 3.7D).

Airway enlargement is a feature of bronchiectasis, an obstructive lung disorder that occurs with recurrent damage to the airways. It can occur due to impaired mucociliary clearance, lung infection, or autoimmune disease, though a large proportion of cases are in cystic fibrosis (CF) patients (386). Current evidence suggests genetic or environmental triggers initiate an immune response from neutrophils, reactive oxygen species, and inflammatory cytokines that leads to destruction of muscular and elastic components of the bronchial walls. Once damaged, mucociliary clearance is impaired, perpetuating further inflammation and/or infection and subsequent tissue damage in a vicious cycle. Intriguingly, bronchiectasis was seen in a subset of RTT patients by HR-CT (333). However, this is the first report of bronchiectasis in mice.

144

Figure 3.7: Enlarged bronchioles in adult Mecp2-mutant lungs.

H&E staining of male P56 +/Y and Mecp2/Y A. lungs at low power (1x), scale bars represent 1000 µm, and B. bronchioles at a high power (10x) magnification, scale bars represent 50 µm. H&E staining of 9-month-old female +/+ and Mecp2/Y B. lungs, scale bars represent 1000 µm, and C. bronchioles, scale bars represent 50 µm; n=3.

145

3.4.8 Pulmonary function is altered in Mecp2-mutant mice

Since emphysema- and bronchiecstasis-like changes were found in Mecp2-mutant lungs, respiratory mechanics were assessed. In humans, pulmonary function is measured by spirometry, while in mice, forced oscillation technique (FOT) is used. Pulmonary function was assessed in adult male 10-week-old +/Y and Mecp2/Y mice using the SciReq FlexiVent system. Briefly, mice were anesthetized and intubated through the trachea, and lungs were artificially inflated through forced oscillation maneuvers to assess respiratory mechanics.

Pulmonary function is drastically altered in Mecp2/Y mice. Inspiratory capacity, the amount of air that can be inhaled, is measured during a gradual inflation to the lung’s full capacity (30 cmH20). Inspiratory capacity was increased in Mecp2/Y mice compared to +/Y mice (0.599 ml ± 0.01 in +/Y; 0.682 ± 0.02 in Mecp2/Y, p=0.0008, Figure 3.8A), indicating that their lungs can hold more air. Pulmonary resistance, elastance, and compliance are measured during a single frequency forced oscillation maneuver. Lung resistance is a quantitative measure of constriction in the lungs. Elastance and compliance, which are reciprocal measures, capture the elastic stiffness and the ease of extension, respectively. Pulmonary resistance was lower in Mecp2/Y lungs compared to +/Y (0.612 cmH20/s ± 0.02 in +/Y; 0.539 ± 0.02 in Mecp2/Y, p=0.009, Figure 3.8B), consistent with airway enlargement. Lung elastance was decreased in Mecp2/Y mice (31.02 cmH20/ml ± 1.09 in +/Y; 26.29 ± 0.86 in Mecp2/Y, p=0.006, Figure 3.8C) while compliance was increased (0.032 ml/cmH20 ± 0.001 in +/Y; 0.038 ± 0.001 in Mecp2/Y, p=0.003, Figure 3.8D). This suggests that Mecp2/Y lungs can be overextended or hyperinflated during inspiration and that they lack the tensile strength or elastic recoil required for normal exhalation.

Pulmonary function was also assessed in female 6-month-old Mecp2/+ mice. Consistent with male Mecp2/Y mice, female Mecp2/+ mice have an increased inspiratory capacity compared to +/+ mice (0.698 ml ± 0.02 in +/+; 0.757 ± 0.01 in Mecp2/+, p=0.027, Figure 3.8E). However, their pulmonary resistance was like their wildtype counterparts (Figure 3.8F), suggesting heterozygous mosaic expression of Mecp2 can maintain some aspects of pulmonary function. Additionally, in Mecp2/+ mice, pulmonary elastance was decreased (26.04 cm20/ml ± 0.81 in +/+; 23.06 ± 0.98 in Mecp2/+, p=0.042, Figure 3.8G) and compliance increased (0.0389 ± 0.001 in +/+; 0.044 ± 0.002 in Mecp2/+, p=0.034, Figure 3.8H), like in symptomatic male Mecp2/Y mice. These findings suggest that loss of Mecp2 imparts consequences on lung function through reduced lung elastance, increased lung compliance, and by making the lung hyperextendable. Notably, these changes were seen in peri-symptomatic female Mecp2/+ mice; respiratory function could continue to worsen with increasing age and symptom progression.

146

Figure 3.8: Lung function is altered in adult Mecp2-mutant mice.

Forced oscillation technique (FOT) was used to assess inspiratory capacity (A,E), lung resistance (B,F), elastance (C,G), and compliance (D,H) in +/Y, Mecp2/Y, +/+ and Mecp2/+ mice, respectively. tatistics were assessed using the tudent’s t-test. *P<0.05, **P<0.01, ***P<0.001, n = 9 +/Y, 7 Mecp2/Y, 6 +/+, 6 Mecp2/+.

147 3.4.9 AE2 cell-specific deletion of Mecp2 alters pulmonary structure and function

Given the structural and functional changes in Mecp2-mutant lungs, I questioned whether Mecp2 deficiency in lung alveolar epithelial cells alone would alter lung structure function. As in Chapter 2, we bred B6.Mecp2tm1Bird mice (Mecp2-flx) mice to B6.Sftpctm1(cre/ERT2)Blh mice, which express Cre under a tamoxifen-inducible promoter of the surfactant protein C (Sftpc) gene. Sftpc expression is expressed in alveolar epithelial progenitor cells which differentiate into squamous alveolar epithelial 1 (AE1) cells which make up the gas exchange surface, and cuboidal AE2 cells which produce pulmonary surfactant, support AE1 cells, and act as stem cells, differentiating into AE1 cells when needed. Cre-mediated excision of Mecp2 occurred upon induction at P21 using three injections of tamoxifen. Histological analyses and pulmonary function tests were conducted in “Sftpc-CreERT2;Mecp2-flx” mice and their littermate controls.

As epithelial-mesenchymal interactions are imperative for alveolar development and formation, I hypothesized that deletion of Mecp2 in alveolar epithelial cells may disrupt the airspaces, as in Mecp2/Y mice. However, alveoli in 10-week-old Sftpc-CreERT2;Mecp2-flx mice were as wild type (Figure 3.9A), indicating AE2-specific loss of Mecp2 does not cause alveolar wall destruction. However, as Cre-mediated excision of Mecp2 occurred at P21, crucial epithelial-mesenchymal interactions during early lung development were unaffected. Despite this, and to our surprise, bronchiolar enlargement was seen in Sftpc-CreERT2;Mecp2-flx lungs compared to their littermates (Figure 3.9B). Consistent with our results in Mecp2/Y mice, Sftpc-CreERT2;Mecp2-flx lungs also had thickened bronchial walls (Figure 3.9C).

To assess whether bronchial enlargement alone alters pulmonary function, we performed FOT on Sftpc-CreERT2;Mecp2-flx mice and their littermates. Lung inspiratory capacity trended upward in Sftpc-CreERT2;Mecp2-flx mice, though this change was not significant (Figure 3.9D). However, pulmonary resistance was decreased in Sftpc-CreERT2;Mecp2-flx mice compared to control mice (0.545 cmH20/s ± 0.01 in WT; 0.468 ± 0.02 in Sftpc-CreERT2;Mecp2-flx, p=0.045, Figure 3.9E), consistent with bronchial enlargement. Additionally, like in Mecp2/Y mice, lung elastance was decreased (29.21 cmH20/ml ± 1.36 in WT; 24.17 ± 1.10 in Sftpc-CreERT2;Mecp2-flx, p=0.014, Figure 3.9F) and lung compliance increased (0.035 ml/cmH20 ± 0.002 in WT; 0.042 ± 0.002 in Sftpc-CreERT2;Mecp2-flx, p=0.008, Figure 3.9G).

148

Figure 3.9: Altered lung function in mice with an AE2 cell-specific deletion of Mecp2. H&E staining of 10-week-old lungs from male WT, Sftpc-CreERT2, Mecp2-flx, and Sftpc- CreERT2;Mecp2-flx mice in the A. alveolar area, scale bar represents 10 μm, B. bronchioles at low power magnification, scale bar represents 500 μm, and C. bronchioles at high power magnification, scale bar represents 100 μm. Forced oscillation technique was used to measure pulmonary function in Sftpc-CreERT2;Mecp2-flx mice and their control littermates. D: inspiratory capacity, E: pulmonary resistance, F: lung elastance and G: lung compliance was measured. n=10 mice per group. Statistics were assessed using one-way ANOVA with Tukey’s test for multiple comparisons. *P<0.05.

149 3.5 Discussion

While RTT is considered a neurological disease, the ubiquitous expression of MECP2 in humans and mice implies its importance across all tissues (146,339). Prevalent symptoms of RTT outside of the central nervous system (CNS) have emerged in recent years. For example, 60% of RTT patients experience symptoms associated with gastrointestinal dysmotility, and changes in colon length and epithelial histology have been observed in Mecp2-mutant mice (27,387). Additionally, RTT patients experience urological complications and decreased bone mass, features that are also found in Mecp2-mutant mice (29,30,32). Here, RNA-sequencing analyses revealed several misregulated pathways in Mecp2/Y lungs that could lead to pathogenesis. Further I found that lung structure is altered in Mecp2-mutant mice, and that this causes drastic changes in pulmonary mechanics.

3.5.1 Circadian rhythm is altered in Mecp2/Y lungs

To our surprise, genes encoding the core circadian rhythm-generating system were mis- expressed at the transcriptional level in Mecp2/Y lungs. The suprachiasmatic nucleus (SCN) in the basal hypothalamus constitutes the central “master” pacemaker of circadian timing, while peripheral tissues also have cell autonomous oscillators (388). Interestingly, Mecp2 is highly expressed in the SCN (389), suggesting a role in regulating circadian rhythm. Further, cortical MECP2 itself follows a rhythmic expression pattern (390), suggesting circadian-dependent regulation of its target genes. Accordingly, disturbances in the sleep/wake cycle are prevalent in RTT patients and Mecp2-mutant mouse models (33,391).

Transcriptional profiling studies have revealed that at least 10% of all mammalian genes are rhythmically expressed, with most clustering in processes of mitochondrial oxidative phosphorylation, endocrine hormone secretion, carbohydrate metabolism and lipid biosynthesis (392). Circadian rhythms are also important in the onset and severity of diseases, including myocardial infarction, ischemic stroke, and sudden cardiac death, all of which peak in frequency during morning hours (393). Interestingly, circadian rhythm drives daily fluctuations in pulmonary function, mucus secretion, and inflammatory processes. A review study found that approximately 70% of asthma-related deaths occurred between 12 – 6 am, when lung function is reportedly at its lowest (394).

The data here suggest a shift in pulmonary circadian rhythm in Mecp2/Y mice. The precise molecular pathways controlled by pulmonary rhythms are still unclear. One hypothesis is that low cortisol and epinephrine levels during sleep contribute to increased airway inflammation and reactivity during nighttime hours (395); thus, misalignment or sporadically expressed circadian

150 rhythm genes could lead to chronically perturbed lung function. Another hypothesis is that altered SCN rhythms reduce cholinergic activity in the lung at distinct points during development, reducing airway smooth muscle tone (396). As lipid and glucose metabolism are regulated by circadian rhythm in other tissues, it is possible that lung lipid metabolism is also circadian controlled. Nonetheless, many pulmonary-expressed genes are likely mis-expressed due to these changes in core circadian rhythm genes.

3.5.2 Extracellular matrix genes are transcriptionally decreased in Mecp2/Y lungs

The most unexpected finding from this study was drastically decreased expression of ECM genes in Mecp2/Y lungs. The ECM provides structural support for cells while regulating development, homeostasis, and injury response. The ECM matrisome contains approximately 150 proteins that contribute to its unique dynamics and topography. In Mecp2/Y lungs, Eln expression was decreased by 3.2-fold. Mice lacking the Eln gene (Eln-/-) have large air-filled cavities in their lungs and die within 48 hours of birth. In contrast, haploinsufficient mice (Eln+/-) have normal lung morphology, but show a 2-fold increase in collagen-1 and lysyl oxidase proteins, likely as compensatory mechanisms (397). Consistently, patients with Williams-Beuren syndrome (WBS), who are haploinsufficient for ELN due to a chromosome 7q11.23 deletion that encompasses the entire ELN gene, do not have overt respiratory symptoms (398–400). However, upon exposure to cigarette smoke, airway enlargement is increased by 1.8x in Eln+/- mice, suggesting conditional haploinsufficiency predisposes the lung to irreversible injury (401). Furthermore, mice with 30% of wild type Eln expression have large airspaces and highly distended lungs (402), suggesting an Eln threshold level beyond which the lungs cannot compensate. Mecp2/Y lungs also express ~30% of wild type Eln levels. Consistently, ELN protein levels are decreased and elastin deposition is altered in Mecp2/Y mice. Thus, decreased Eln alone could cause the structural and functional changes seen in Mecp2/Y lungs.

Extensive crosslinking is responsible for elastin’s longevity, with estimated turnover rates of ~80 years in humans (384). Here, not only was elastin expression drastically reduced in Mecp2/Y lungs, but fibrillin-1 (Fbn1), the main component of microfibrils, and lysyl oxidase genes (Loxl1- 3), which crosslink elastin molecules, were also expressed at low levels. Mice lacking Fbn1 (Fbn1- /-) die shortly after birth due to cardiopulmonary failure, while heterozygotes (Fbn1+/-) have defective microfibrillar deposition and develop emphysema. Additionally, Loxl1 deletion results in enlarged airspaces and accumulation of the immature elastin precursor, tropoelastin. Decreased expression of these genes in Mecp2/Y mice likely destabilizes the lung elastin network.

Compounding these findings, the expression of 20 collagen genes and several adamalysins were also decreased in Mecp2/Y lungs. Collagen genes protect the lung by providing it with tensile

151 strength, while adamalysins mediate collagen processing and facilitate interactions between cells and ECM through integrin binding. Consistently, several integrins are also expressed at low levels in Mecp2/Y lungs. These expression changes suggest the collagen network, and potentially cell- ECM interactions, are also affected in Mecp2/Y lungs. There are over 40 collagen genes and targeted deletion of single collagen genes does not often lead to drastic phenotypes; however, the diminished expression of 20 collagen genes in Mecp2/Y lungs likely has a cumulative effect, altering the physical properties of the lung. Finally, several glycoproteins and proteoglycans are also expressed at low levels in Mecp2/Y lungs. These proteins have various actions in the lung. For example, tenascin-C (Tnc) mediates epithelial-mesenchymal interactions and the lungs of Tnc-null have fewer smooth muscle cells. This decreased expression of so many ECM components in Mecp2/Y lungs likely has many additive effects, making the ECM unstable and degenerative. Further studies on the protein composition of lung ECM in Mecp2-mutant models will be informative.

3.5.3 Structural changes in the Mecp2-mutant lung

Here, we found that lung development was modestly altered in Mecp2/Y mice. While early postnatal lung development appears normal, at P21, Mecp2/Y mice have thickened alveolar walls. This suggests alveolar wall expansion may be impaired, resulting in diminished septation and a decreased surface for gas exchange. Because alveolarization is mediated by an interconnected elastin network laid down by fibroblasts, it is not surprising that this process is altered. More interesting is that both male and female Mecp2-mutant lungs develop an emphysema-like phenotype in adulthood. A previous study also found enlarged airspaces in adult Mecp2/Y mice on a C57BL/6 background (334). Notably, as female Mecp2/+ mice express normal Mecp2 from roughly 50% of their cells, their lung development is likely normal; this suggests pathological alveolar wall destruction in Mecp2-mutant mice. Importantly, the transcriptional changes in ECM components, such as collagen which provides the lung its tensile strength, could make Mecp2/Y lungs prone to injury. Damage to the ECM is detrimental and virtually irreversible as replacing damaged elastin fibers would require the coordinated re-expression of all molecules making up the microfibril and the enzymes required for crosslinking elastin molecules (403).

Bronchiolar enlargement was seen in adult Mecp2-mutant mice, with females being more affected. Airway enlargement is a prominent feature of bronchiectasis, which develops due to recurrent lung damage. While poorly studied, perhaps due to the absence of mouse models, it is thought that destruction to muscle and elastic components of the lung causes the disorder (386). Inflammation is the most prominent cause of both bronchiectasis and emphysema; inflammatory cells release proteases and cytokines in the lung which contribute to tissue remodeling and

152 destruction. Additionally, excess reactive oxygen species may also lead to tissue damage. Multiple reports have evidenced pathological lung inflammation in Mecp2/Y mice (333,334). It is possible that female Mecp2/+ mice are more severely affected due to prominent or chronic inflammation from an early age, leading to extensive tissue damage. Regardless, the finding that Mecp2/+ mice also have abnormal lung structure has important implications for human RTT patients whose respiratory symptoms are most often attributed to neurological processes.

3.5.4 Lung function is altered upon loss of Mecp2

The changes in pulmonary gene expression and lung structure in Mecp2-mutant mice expectedly impair pulmonary function. A recent study assessed lung function across several well-established mouse models of lung disease, including bleomycin-induced fibrosis and elastase-induced emphysema (404). In their study, inspiratory capacity was decreased in mice with fibrosis and increased in mice with emphysema. Additionally, airway resistance was increased in mice with emphysema, while elastance was increased in mice with fibrosis and decreased in mice with emphysema.

Here, I found that inspiratory capacity was increased in both male and female Mecp2-mutant mice. Further, lung elastance was decreased while lung compliance was increased. These results are consistent with an emphysema-like phenotype and can be explained by our earlier findings. Lung elastance represents the ability of the lung to return to its original form after inhalation. Elastin is solely responsible for the lung’s elastic recoil, and as its mRNA and protein expression was decreased in Mecp2-mutant lungs, it is not surprising that lung elastance was reduced. Accordingly, inspiratory capacity may be increased as more air can be held within the lung without elastic forces driving it out. Lung compliance is the inverse of elastance and represents the lung’s ease of extensibility. Therefore, the Mecp2-mutant is hyperextendable and lacks tensile strength.

Mecp2/Y males, but not Mecp2/+ females, had decreased lung resistance. Normally, lung resistance is increased in patients with emphysema, as elastin and other ECM components that hold bronchioles open are lost; this leads to airway collapse, increasing the amount of work required to pass air through the lung. However, bronchiectasis is associated with decreased lung resistance as large airways allow air to enter the lung freely. Therefore, the pulmonary function profile of Mecp2/Y mice do not model a specific respiratory disease (ie. emphysema) but are entirely consistent with their structural abnormalities. Notably, FOT was performed on 6-month- old peri-symptomatic female Mecp2/+ mice, and while lung resistance was not altered at this stage, it may decrease with disease progression. Regardless, even heterozygous mosaic loss of Mecp2 alters lung mechanics, reducing its elastic recoil and tensile strength and making the lung hyperextendable. As such, air can freely and easily enter the lungs, but exhalation requires

153 physical exertion. Often, air can become trapped in the lung and alveoli may collapse. As a result, inhalation can become difficult as well.

I also assessed whether lung AE2 cell-specific loss of Mecp2 would affect lung structure and function. While fibroblasts produce most ECM proteins, they maintain close contact with AE1 and AE2 cells. I used mice carrying a tamoxifen-inducible Sftpctm1(cre/ERT2)Blh allele to genetically remove Mecp2 from alveolar epithelial cells. Sftpc is expressed primarily in AE2 cells but is also expressed in alveolar epithelial progenitors which differentiate into AE1 cells. Tamoxifen induction was performed at P21. Intriguingly, while mice with an AE2 cell-specific deletion of Mecp2 did not have altered alveolar structures, they did show bronchiolar enlargement with consequent changes in pulmonary function.

While epithelial-mesenchymal interactions are crucial in the lung, Mecp2 was depleted from AE2 cells in our studies at P21, near the end of postnatal lung development. Thus, abnormal lung development likely does not account for the changes we see in lung structure and function. As I showed in Chapter 2, Mecp2 deficiency in AE2 cells causes triglyceride accumulation and decreases surfactant phosphatidylcholines. In this manner, loss of Mecp2 may predispose the lung to inflammatory injury through lipotoxicity-induced inflammation and/or reactive oxygen species generation. This may result in bronchial enlargement and altered lung mechanics. Though future work is required to confirm this hypothesis, it would suggest that early treatment of lung infection and injury is essential for RTT patients to prevent irreversible damage to the lungs.

3.5.5 Implications for RTT patients

The data here largely support a combined emphysema- and bronchiectasis-like phenotype in Mecp2-mutant mouse models. Transcriptomic data suggests decreased ECM gene expression, which is consistent with abnormal lung architecture and impaired pulmonary function. To date, only one group has assessed the lungs of RTT patients. High-resolution computed tomography scans of 27 female RTT patients showed that 55% of patients had abnormal lung findings, including centrilobular nodules (67%), thickened bronchial walls (53%), non-specific ground glass opacities (27%), and bronchiectasis (60%). Further, pulse oximetry abnormalities were found in all patients, suggesting physiological changes in RTT breathing (333). Additionally, RTT patients report diagnoses of pneumonia, asthma, bronchitis, and other respiratory illnesses (31).

Assessing the lung in RTT patients remains a challenge. Lung function in humans is assessed using spirometry, a process largely dependent on patient understanding and cooperation. Lung biopsies are likely avoided as they can temporarily exacerbate respiratory symptoms, which could be lethal in RTT patients. Lung imaging may be helpful, as CT scans can show pockets of air

154 retained in the lung after exhalation as well as evidence of bronchiectasis, but less obvious changes in lung structure, especially in the distal lung, would be difficult to diagnose. As such, studying the lung in Mecp2-mutant models is crucial as it may provide new insights into pathological processes and point us to new therapeutic strategies to help manage respiratory symptoms in RTT.

The similarity between Mecp2-mutant lungs and models of emphysema/COPD is extremely informative. While there are no current treatments for COPD, there are guidelines in place to manage symptoms and prevent exacerbations. Avoiding environmental triggers is crucial, including respiratory viruses, cigarette smoke, and chemical pollutants. Bronchodilators, such as anticholinergic medications, and inhaled corticosteroids can reduce the frequency of acute COPD exacerbations by up to 25%, though further tests would be needed to determine if they would help RTT patients. Altogether, the data presented here suggest that Mecp2 has an important role in lung homeostasis, such that its loss leads to an unstable and degenerative ECM. Further studies are necessary to determine the mechanism responsible for this association, and to assess ECM from other tissues in Mecp2-mutant mice.

155

Chapter 4 Pharmacological treatment of lipid metabolism perturbations in Mecp2-mutant mice

Some of the work in this chapter was performed by Dr. Stephanie Kyle, a previous graduate student in Dr. Monica Justice’s laboratory.

A special thank you to Julie Ruston and Christine Taylor who helped with drug treatments, phenotypic assessments, and tissue collection.

156 Pharmacological treatment of lipid metabolism in Mecp2- mutant mice

4.1 Abstract Defects in cholesterol metabolism are associated with several neurological diseases including Fragile X syndrome, Alzheimer’s disease, and Parkinson’s disease. Recent data has shown aberrant cholesterol metabolism in mouse models of Rett syndrome (RTT), a neuro-metabolic disorder caused by mutations in methyl-CpG-binding protein 2 (MECP2). Cellular cholesterol levels are sensed by two opposing pathways: sterol regulatory-element binding proteins (SREBPs) and liver X receptors (LXRs). Both pathways can be pharmacologically targeted via statins and LXR-agonists, respectively. Here, I employed RNA-sequencing to assess the transcriptome of pre-symptomatic wild type (+/Y) and Mecp2/Y mice. Intriguingly, the expression of genes encoding cholesterol biosynthesis enzymes was significantly increased, while the expression of LXRs and their lipid transporter targets was decreased. I showed that treatment with fluvastatin improved neurological and respiratory symptoms in Mecp2-mutant mice, while normalizing systemic lipid parameters. Treatment with the LXR-agonist T0901317 also improved neurological symptoms, but increased serum and liver lipids in Mecp2/Y mice. However, the LXR- agonist LXR-623 improved neurological symptoms while avoiding detrimental effects to the liver in Mecp2-mutant mice. Finally, simultaneous treatment with fluvastatin and LXR-623 had no beneficial effects in Mecp2/Y mice. These results suggest aberrant cholesterol homeostasis at the transcriptional level in pre-symptomatic Mecp2/Y mice. Further, I show that lipid-lowering therapeutics are an effective treatment for improving symptoms as well as peripheral metabolism, and may be clinically advantageous in RTT patients.

157 4.2 Introduction Cholesterol is an essential component of all cells with important roles in a variety of processes. It regulates membrane fluidity, secures proteins in lipid rafts, and is involved in a number of signaling pathways (405,406). In the brain, it is crucial for myelination and synapse formation (407,408). It also serves as a key metabolic precursor for the synthesis of vitamin D, corticosteroids, and steroid hormones, which interact with nuclear receptors to regulate other aspects of cell function (409). The regulation of cholesterol and other lipids across the body is crucial as excessive cholesterol levels are extremely toxic and a hallmark of diseases such as diabetes, obesity, and atherosclerosis. Cellular cholesterol levels are sensed by two distinct, opposing pathways: sterol regulatory-element binding proteins (SREBPs) and liver X receptors (LXRs) (410).

SREBPs (including SREBP1a, -1c, and -2) are ubiquitously expressed membrane-bound transcription factors encoded by two genes (Srebf1 and Srebf2). Under basal conditions, SREBP binds to SREBP-cleavage activating protein (SCAP) in the endoplasmic reticulum (ER) membrane (411). When cholesterol or oxysterol levels are high, SREBP-SCAP is retained in the ER membrane. When sterol levels are low, the SCAP-SREBP complex is transported to the Golgi where SREBP is cleaved and subsequently translocated to the nucleus. In the nucleus, SREBP binds to sterol response elements (SREs) in gene promoters, activating the transcription of many lipogenic genes (349). For example, SREBP2 induces the expression of the rate-limiting cholesterol biosynthesis enzyme, HMG-CoA reductase (Hmgcr). SREBP1a/1c are major regulators of fatty acid synthesis.

LXRs (including LXRα and LXRβ) are nuclear receptors encoded by Nr1h3 and Nr1h2, respectively. LXRs form heterodimers with retinoid X receptors (RXRs) and bind to the promoters of genes containing LXR response elements (LXREs) (412). LXR-RXR heterodimers recruit corepressors like nuclear coreceptor receptor 1/2 (NCOR1/2) to repress target gene transcription (413). Upon activation by endogenous ligands, such as oxysterols, the heterodimer undergoes a conformational change, releasing co-repressors and recruiting E1A-binding protein P300 (EP300) and coactivators such as nuclear receptor coactivator 6 (NCOA6) (414). LXRs are essential for the process of reverse cholesterol transport (RCT) by which cells and tissues are protected from excessive sterol levels (415). During RCT, cholesterol from peripheral tissues is transferred to high density lipoprotein (HDL) and returned to the liver, where is it ultimately excreted in bile. The ABC family of membrane transporters, including ABCA1, ABCG1, ABCG5 and ABCG8, are the key players in RCT. LXR induces the transcription of the genes encoding these proteins as well as other cholesterol transport genes including Apoe and Abcg4, and is thus considered a transcriptional activator of cholesterol transport (416,417).

158 The liver, lung and brain are three highly active tissues with entirely unique lipid needs. Liver hepatocytes are the body’s major site of endogenous lipid metabolism, supplying 80% of the body’s total daily cholesterol needs, as well as synthesizing large amounts of fatty acids (418). The liver supplies lipids to peripheral tissues through triglyceride-enriched very low-density lipoproteins (VLDL) and cholesterol-enriched low-density lipoproteins (LDL). Lipoprotein particles are organized by apolipoproteins, including apolipoprotein B100 (ApoB) and E (ApoE) (419). In the liver, LXRs regulate sterol and fatty acid synthesis by modulating the expression of SREBPs. The liver is also responsible for cholesterol recycling via RCT for the formation of bile acids.

The lung is also a site of de novo lipogenesis, as discussed in Chapter 2. While phospholipids are the predominant lipid synthesized by alveolar epithelial 2 (AE2) cells, cholesterol is also produced as the major neutral lipid component of pulmonary surfactant (294). In the lung, alveolar macrophages ingest and recycle lipids, making cholesterol and free fatty acids (419). Upon increased expression of LXR-targets, including ABCA1 and ABCG1, cholesterol efflux to HDLs is mediated, facilitating RCT. Consistently, both Abcg1- and Abca1-null mice have a drastic accumulation of cholesterol and other lipids in the lung (420,421), suggesting RCT is essential to prevent pulmonary lipid overload.

The brain contains approximately 25% of the body’s cholesterol. However, because cholesterol cannot cross the blood-brain barrier (BBB), all brain cholesterol is derived by de novo synthesis, forming an isolated but tightly regulated system of cholesterol production, trafficking, and degradation (422). While the majority of the brain’s cholesterol is stored in stable myelin sheaths, neurons and astrocytes also require cholesterol and experience a turnover rate of approximately 20% per day (423). Cholesterol is thought to be important for the neuron’s pre- and post-synaptic functions (424). When brain cholesterol levels increase beyond physiological levels, cholesterol is converted to the oxysterol 24(S)-hydroxycholesterol (24S-OHC) by the enzyme cytochrome P450 46A1 (CYP46A1) (425–427). Lipophilic 24S-OHC can move across the BBB, allowing cholesterol egress from the CNS into the plasma. 24S-OHC is a potent ligand for LXRs causing increased transcription of cholesterol transport genes, while decreasing SREBP-mediated transcription of cholesterol biosynthesis enzymes (428).

The SREBP and LXR pathways can be pharmacologically targeted. SREBP-induced cholesterol biosynthesis can be indirectly manipulated by statins, a commonly prescribed class of lipid- lowering molecules. Statins act by competitively inhibiting HMGCR, the rate-limiting enzyme in cholesterol biosynthesis, thereby reducing global cholesterol biosynthesis (429). Due to the prominent role of cholesterol in cardiovascular disease, statins are most often prescribed to prevent and treat atherosclerosis. Similarly, synthetic LXR agonists can activate LXRs to induce

159 the expression of cholesterol transport genes, mediating cholesterol efflux, RCT, and bile acid secretion (430). This effectively reduces total body cholesterol, particularly in macrophages. Most LXR agonists target both LXRα and LXRβ due to the highly similar sequence of their ligand binding domains.

Our lab previously identified perturbed brain and liver cholesterol homeostasis in a mouse model of Rett syndrome (RTT), a neurological disorder caused by mutations in methyl-CpG-binding protein 2 (MECP2). In the Mecp2/Y brain, Cyp46a1 expression is increased when mice are peri- symptomatic, indicating a heightened need for cholesterol turnover, and total cholesterol is increased when mice are symptomatic, suggesting a general overproduction of brain cholesterol (167). Additionally, Mecp2/Y mice accumulate cholesterol and triglycerides in their livers, ultimately developing fatty liver disease (164). Consistently, serum cholesterol, LDL-cholesterol, and triglycerides are elevated in the serum of Mecp2/Y mice. In Chapter 2, I showed evidence for cholesterol and triglyceride accumulation in Mecp2/Y lungs. These results position aberrant cholesterol metabolism as a major feature of Mecp2-deficiency. Thus, it is possible that RTT patients may be aided by pharmacological interventions designed to modulate lipid metabolism.

Here, we show that the cholesterol biosynthesis pathway is transcriptionally over-activated in the brain of pre-symptomatic Mecp2/Y mice, while LXRs and their cholesterol transport targets are expressed at low levels. I explore lipid metabolism-directed therapies, showing varying effects of fluvastatin, T0901317, and LXR-623 on neurological and respiratory symptoms and lipid parameters in Mecp2-mutant mice.

160 4.3 Methods

4.3.1 Animals

All animal procedures were approved by the Animal Care Animal Care Committee at the CCAC- accredited animal facility, The Center for Phenogenomics (TCP). Congenic 129.Mecp2tm1.1Bird/Y mice feature a deletion of the last two exons (exons 3-4) of the Mecp2 transcript, resulting in a null allele. Male Mecp2/Y (Mecp2-null) and +/Y (wild type), and female Mecp2/+ and +/+ mice were obtained by backcrossing Mecp2tm1.1Bird/+ females to males of the 129SvEvS6/Tac strain. Mice were fed a standard diet (Harlan Teklad 2918) ad libitum, consisting of 18% protein, 6% fat, and 44% carbohydrates. Mice were housed in a 13-hour light/dark cycle and were euthanized between the hours of 9AM and 12PM (ZT 2-5) to control for circadian rhythm fluctuations.

4.3.2 RNA-sequencing

Brain regions (cortex, striatum, hippocampus, thalamus, and cerebellum) were rapidly micro- dissected from P21 +/Y and Mecp2/Y mice between the hours of 3PM and 5PM (ZT 8-10). Briefly, the brain was carefully removed from the skull and placed in ice-cold RNAse-free PBS in an ice-cold petri dish. Under a dissecting microscope, the cerebellum was removed from the inferior and superior colliculus and the brainstem and then frozen. The brain was then placed in an ice-cold stainless-steel brain slicer matrix and serially sectioned. Brain sections (1 mm) were placed in ice-cold PBS and brain regions were freehand dissected under a dissecting microscope. Sections were flash-frozen and stored at -80°C until further processing.

RNA was assessed for integrity using the Agilent RNA 6000 Nano kit (Agilent Technologies) on an Agilent 2100 Bioanalyzer. All RNA integrity numbers (RIN) were above 8.5, with an average score of 9.3. RNA libraries were prepped using the NEBNext poly(A) mRNA magnetic isolation module which selects for mRNA and lncRNA with poly(A) tails by binding to oligo d(T)25 paramagnetic beads. Paired-end libraries were prepared and sequenced on an Illumina HiSeq 2500 platform at The Centre for Applied Genomics (TCAG) at The Hospital for Sick Children. Yield and size distribution were assessed using the Agilent 2100 Bioanalyzer. RNA libraries were quantified by quantitative PCR using the Kapa Library Quantification Illumina/ABI Prism Kit protocol (KAPA Biosystems). Libraries were pooled in equimolar quantities and paired end sequenced on an Illumina HiSeq 2500 platform.

Reads were trimmed to remove adapters and low-quality sequence ends using Trim Galore v0.4.0. Reads were aligned to the “UC C mm10” reference sequence using Tophat v. 2.0.11. Extraction and processing of reads was achieved using htseq-count v.0.6.1p2. ~80% of reads of

161 all samples map to exons. Principal component analysis was performed on raw gene counts to assess library distribution. Differential expression of sections originating from 3 wildtype and 2 Mecp2 null brains were performed using edgeR, R package v.3.8.6.

4.3.3 Drug administration

As lipid metabolism follows a circadian rhythm, each treatment was administered at the same time of day, between 9AM and 10AM (ZT 2-3). Drugs were prepared as follows: Fluvastatin (Sellekchem) was dissolved in 100% DMSO at 6 mg/ml. On the day of administration, aliquots of fluvastatin were dissolved in sterile saline such that the desired dose for a 20 g mouse was given in 100 ul. Male mice were injected intraperitoneally with a twice-weekly 1.5 mg/kg body weight dose from ages P28 to P56. Female mice were injected intraperitoneally with a weekly dose of 3 mg/kg body weight from 6 weeks to 32 weeks of age. Vehicle controls (1:10 DMSO:saline) were administered at the same rate.

T0901317 (Cayman) was dissolved in 100% DMSO at 50 mg/ml. On the day of administration, aliquots were further dissolved in sterile saline to 5 mg/ml. Male mice were injected intraperitoneally twice weekly with a dose of 10 mg/kg from 5 weeks to 10 weeks of age. Vehicle control cohorts were injected at the same rate with a 1:10 DMSO:saline solution.

LXR-623 (Sigma-Aldrich) was dissolved in 100% corn oil at 100 mg/ml. This mixture was rotated overnight at 4°C for complete dissolution. On the day of administration, aliquots were further dissolved in corn oil to 10 mg/ml. Male mice were injected intraperitoneally once weekly with a dose of 50 mg/kg from 5 weeks to 10 weeks of age. Female mice were injected with the same dose weekly from 6 weeks to 32 weeks of age. Vehicle control were administered at the same rate with pure corn oil.

For co-treatment experiments, mice received 2 weekly doses of 1.5 mg/kg fluvastatin and 1 dose of 50 mg/kg LXR-623 or identical vehicle controls (1:10 DMSO:saline and corn oil, respectively). Mice were treated from 5 weeks to 10 weeks of age. Fluvastatin was administered on Tuesday mornings (9:00am - 10:00 am), and fluvastatin/LXR-623 was administered on Friday mornings.

4.3.3.1 Subjective Health

Mice were assayed for general health once per week from 4 weeks to 10 weeks of age. Scoring was blinded to genotype and treatment. Mice were scored using the assessment published in Guy, 2007, with slight modifications (149). Mice were given a score between 0 and 2 based on the severity of the phenotype assessed (Table 2.2). Mice were assessed for limb clasping, tremors, activity, grooming, hypotonia, and body weight, for a combined possible score of 0-12.

162 4.3.3.2 Rotarod

Motor coordination was measured using the rotating rod (Stoelting ugo basile mouse rota-rod) at 6.5-7 weeks of age in male mice and 6 months of age in female mice. Mice were placed on the grooved rotating rod facing the opposite direction of rotation. The revolution rate increased from 4 rotations per minute (RPM) to 40 RPM over the course of 5 minutes. The length of time that each mouse remained on the rod was recorded for eight trials over two consecutive days (four trials per day), with a minimum of 30 minutes between each trial. A trial ended for a mouse when it fell from the rod, stayed stationary on the rod while it spun for two revolutions, or when it successfully stayed on the rod for 5 minutes.

4.3.4 Plethysmography

Respiration was monitored using a Buxco Whole Body Plethysmography (WBP) apparatus (Data ciences International) according to manufacturer’s instructions. All testing was conducted between the hours of 9AM and 12PM. Mice were placed in plethysmography chambers and allowed to acclimate for 30 minutes, until motionless. Baseline breathing rates were measured for a period of 5 minutes. Following this, mice were exposed to nebulized saline for 2 minutes, and respiratory rates were measured for 5 minutes. Mice were then exposed to increasing doses of 6.25, 12.5, and 25 mg/ml of aerosolized methacholine dissolved in saline at a constant rate for 2 minutes, after which readings were taken for 5 minutes at each concentration. Breathing frequency, tidal volume, and enhanced pause (PenH) were analyzed using Buxco FinePoint Software. Apneas were defined as cessation of breathing for over 1 second (2 respiratory cycles) and were calculated manually over a period of baseline breathing.

4.3.5 Necropsy

Following drug studies, male P56-P60 mice, or female P260-270 mice were fasted for 6 hours from 8AM to 2PM. Plethysmography was performed during the fasting period. Necropsy took place between 2PM – 4PM. After recording fasted body weight, mice were individually anesthetized with isoflurane and blood was collected by cardiac puncture. Briefly, a 25 G needle attached to a 1ml syringe was inserted through the pelt into the left ventricle of the heart and a minimum of 0.3 ml of blood was collected. Mice were euthanized by cervical dislocation. Livers, lungs, and brains were cleanly removed from the body and the weight of each organ was recorded. Tissues were snap frozen in liquid nitrogen. Blood samples were transferred to serum separator tubes (BD Vacutainer) and separated according to BD instructions. Serum and tissue samples were stored at -80 °C until analysis.

163 4.3.6 Lipid quantification

Serum and tissue samples were analyzed by the Diabetes and Endocrinology Center at Baylor College of Medicine (Houston, Texas). Lipids were isolated from tissue using CHCl3:CH3OH extraction, followed by drying of the organic phase under N2 pressure. Serum and tissue cholesterol and triglyceride concentrations were assessed by high performance liquid chromatography (HPLC).

4.3.7 Statistics

Values are expressed as mean ± SEM. The statistical difference between the means of four groups were evaluated using one-way ANOVA or two-way ANOVA with Tukey’s test for multiple comparisons. Kaplan-Meier survival curves were assessed using the Log-rank (Mantel-Cox) test. All statistical analyses were performed in GraphPad Prism (Version 7). P-values less than 0.05 were considered statistically significant.

164 4.4 Results

4.4.1 RNA-sequencing of the pre-symptomatic Mecp2/Y brain

Male Mecp2/Y mice develop neurological symptoms, including hypoactivity, tremors, and hind limb clasping, at 4 weeks of age, with symptom progression until death at 6-10 weeks. Most studies of MECP2 function are performed in symptomatic Mecp2-mutant mice, making it difficult to elucidate which molecular pathways are directly affected by Mecp2 deficiency versus those downstream of MECP2’s primary targets. At 4 weeks of age, the expression of Cyp46a1, a gene encoding the major enzyme involved in cholesterol export from the brain, was increased in the Mecp2/Y brain, indicating a heightened need for cholesterol excretion (167). These results suggest cholesterol metabolism is already perturbed in Mecp2/Y brains prior to symptom onset. As such, our lab sought to examine transcriptional changes in the pre-symptomatic Mecp2/Y brain to determine the primary effects of Mecp2 deletion and to better understand changes in brain lipid metabolism that precede neurological symptoms.

RNA-sequencing was employed by a previous graduate student in the laboratory, Dr. Stephanie Kyle, to assess transcriptional changes in the postnatal day (P)21 Mecp2/Y cortex, striatum, hypothalamus/thalamus, hippocampus, and cerebellum. Isolating individual brain regions, rather than sequencing the entire brain, allows the discrimination of local changes in the transcriptional program within discrete clusters of cells, which could be otherwise lost when sequencing the extremely heterogenous brain.

4.4.2 Lipid biosynthesis and transport misregulation at the transcriptional level in the pre- symptomatic Mecp2/Y brain

Cholesterol biosynthesis begins when acetyl-CoA and acetoacetyl-CoA combine to form 3- hydroxy-3-methylglutaryl-coenzyme-A (HMG-CoA). This reaction is catalyzed by the enzyme, hydroxy-3-methylglutaryl-co-enzyme synthase (HMGCS). HMG-CoA is then converted to mevalonate by the rate-limiting enzyme, HMGCR. Mevalonate, through several reactions, forms non-sterol isoprenoids, such as farnesyl pyrophosphate (FPP), which offer substrates for protein prenylation (431). When FPP continues through the cholesterol biosynthesis pathway, it undergoes a series of reactions whereby it is ultimately converted to cholesterol (432).

Of the 23 enzymes in the cholesterol biosynthesis pathway, 22% are overexpressed in the Mecp2/Y cortex, 30% in the striatum, 43% in the hypothalamus/thalamus, 39% in the hippocampus, and 35% in the cerebellum (Figure 4.1A). Ten enzymes show no significant change in expression in any brain region. Overall, these data suggest a general up-regulation of the cholesterol biosynthesis pathway. Interestingly, the gene encoding sterol regulatory-binding

165 protein 2, Srebf2, which activates transcription of cholesterol biosynthesis genes, is expressed at low levels across all brain regions studied (Figure 4.1B). As it regulates its own transcription, reduced Srebf expression suggests that intracellular cholesterol concentrations are high in the CNS as early as 3 weeks of age.

Triglyceride biosynthesis begins when cytosolic acetyl-CoA, originally derived from the mitochondria via ATP citrate lyase (ACLY), is carboxylated by acetyl CoA carboxylase (ACC) into malonyl-CoA. Fatty acid synthase (FAS) then performs a series of condensation reactions to generate the saturated fatty acid, palmitate (C16:0). Stearoyl-CoA desaturases (SCDs) and fatty acid elongases (ELOVLs) desaturate and elongate the fatty acid chain, respectively. Fatty acids can be converted to triglycerides by diglyceride acyltransferases (DGATs).

The expression of Srebf1, which activates genes involved in fatty acid biosynthesis, is reduced in all P21 Mecp2/Y brain regions compared to +/Y mice (Figure 4.1B). Consistently, Acly expression is decreased in the cortex and cerebellum, Acacb expression is decreased in the hippocampus and hypothalamus/thalamus, and Fasn is expressed at low levels in all brain regions (Figure 4.1C). The expression of Dgat1 and Dgat2 is also decreased in Mecp2/Y brains, suggesting decreased triglyceride synthesis. Finally, the expression of Elovl genes, except for Elovl1, are increased in the Mecp2/Y brain and may suggest a preference for long-chain fatty acids.

Interestingly, the expression of Nr1h2, which encodes LXRβ, the most prevalent LXR in the brain, is decreased across all regions of Mecp2/Y brains at P21 (Figure 4.1D). Expression of LXRα- encoding Nr1h3, which is lowly expressed in the brain but is high in the liver, is unchanged. The decreased expression of Nr1h2 corresponds to decreased transcription of the LXR targets Abcg1, Abcg4, and Apoe genes responsible for cholesterol trafficking. Expression of the LXR-target Abca1 is unchanged. These results suggest decreased cholesterol transport in the brain.

166

Figure 4.1: Lipid metabolism is misregulated in the pre-symptomatic Mecp2/Y brain. A. Expression of cholesterol biosynthesis genes in P21 Mecp2/Y brain regions. B. Expression of sterol regulatory-element binding factor genes in Mecp2/Y brain regions. C. Expression of genes involved in fatty acid synthesis, desaturation, elongation, and triglyceride synthesis in Mecp2/Y brain regions. D. Expression of LXRβ (Nr1h2), LXRα (Nr1h3) and LXR-target genes in the Mecp2/Y brain regions. All genes shown have an FDR < 0.05. Cor: cortex, Str: striatum, Hip: hippocampus, H/T: hypo/thalamus, Cer: cerebellum.

167 4.4.3 Treatment with cholesterol-lowering fluvastatin improves health, motor, and respiratory symptoms in male Mecp2/Y mice

The increased expression of cholesterol biosynthesis genes in the pre-symptomatic Mecp2/Y brain suggests it is a primary effect of Mecp2 deficiency. These results add to a growing body of data implicating perturbed lipid metabolism in the early pathogenesis of RTT. Importantly, MECP2, with NCOR1/2, directly represses the expression of cholesterol biosynthesis genes squalene epoxidase (Sqle) and hydroxymethylglutaryl-CoA synthase 1 (Hmgcs1) in the liver (164) and lung (Chapter 2), respectively, and could have similar targets in the brain. Decreased repression of MECP2’s cholesterol biosynthesis targets likely increases cholesterol levels, making statins an ideal treatment for blocking this pathway and restoring lipid metabolism to normal levels. Notably, while the main effect of statins is to decrease LDL cholesterol, statin treatment is also associated with decreased triglyceride accumulation, with many patients experiencing triglyceride reductions of up to 50% (433). Thus, statins can be considered a pan-lipid metabolism treatment.

Our lab previously treated male 129.Mecp2/Y mice with fluvastatin (167). Intriguingly, fluvastatin treatment extended lifespan in Mecp2/Y mice while improving their motor coordination, lowering serum cholesterol and liver triglycerides, and lowering brain lanosterol and desmosterol concentrations, two cholesterol biosynthesis intermediates. Overall, fluvastatin was very effective in improving RTT symptoms in Mecp2/Y mice, suggesting at least some symptoms of RTT are induced by altered lipid metabolism.

I sought to assess whether fluvastatin treatment could pharmacologically correct lung lipid accumulation and improve respiratory symptoms in Mecp2/Y mice. Here, Mecp2/Y mice and their wild type littermates were treated with twice weekly doses of 1.5 mg/kg fluvastatin or a vehicle control from P28 to P56 (Figure 4.2A). As previously published, fluvastatin treatment increased the lifespan of Mecp2/Y mice with 85% (12/14) of fluvastatin-treated Mecp2/Y mice surviving to 60 days compared to only 56% (10/18) of vehicle-treated Mecp2/Y mice (p=0.042, Figure 4.2B). Fluvastatin treatment also improved subjective health scores (Figure 4.2C), indicative of improved overall health, and enhanced motor performance assessed by time spent on a rotating rod (rotarod) compared to Mecp2/Y mice receiving a vehicle control (Figure 4.2D).

Whole-body plethysmography, a non-invasive tool, was used to assess respiratory phenotypes in mice. Strikingly, increased breathing frequency in vehicle-treated Mecp2/Y mice was normalized to wild type levels in Mecp2/Y mice treated with fluvastatin (309 bpm ± 9.24; 262 bpm ± 7.73, p=0.002) (Figure 4.2E). Consistent with previous data, tidal volume was unchanged in Mecp2/Y mice and similar to that seen in +/Y animals, regardless of treatment (Figure 4.2F). Intriguingly, respiratory apneas trended downward after fluvastatin-treatment, though this reduction was not

168 statistically significant (Figure 4.2G). Finally, vehicle-treated Mecp2/Y show a dramatic response to methacholine, a bronchoconstrictive agent, compared to +/Y mice treated with vehicle or fluvastatin (Penh score 10.57 ± 0.67 at 50mg/ml methacholine in vehicle-treated Mecp2/Y; 5.13 ± 0.61 in vehicle-treated +/Y, p<0.0001). This is indicative of airway hyperresponsiveness in Mecp2/Y mice. However, fluvastatin treatment caused a marked improvement in the response to methacholine in Mecp2/Y mice, compared to vehicle-treated mice (Penh score 7.46 ± 0.62 at 50mg/ml methacholine, p=0.001), indicating a drastic improvement in airway resistance with fluvastatin treatment (Figure 4.2H).

Serum cholesterol, but not triglycerides, was elevated in vehicle-treated Mecp2/Y mice as compared to vehicle-treated +/Y mice (138.77 mg/dL ± 6.56 in +/Y; 211.62 ± 8.6 in Mecp2/Y, p=0.0004) (Figure 4.2I-J). Fluvastatin treatment reduced serum cholesterol in Mecp2/Y mice (166.02 mg/dL ± 9.8, p=0.0145), indicating that it was successfully incorporated into cells in an active form. Fluvastatin was also very effective in the lung, where it lowered heightened lung triglycerides in Mecp2/Y mice (5.97 ± 1.0 mg/g tissue in vehicle-treated +/Y; 13.01 ± 2.4 in vehicle- treated Mecp2/Y, p=0.0358) to wild type levels (4.84 ± 0.9, p=0.0149) (Figure 4.2K). Notably, lung cholesterol trended downwards but was not statistically significant (Figure 4.2L). These results highlight the efficacy of fluvastatin in improving overall health, lifespan, motor abnormalities, respiratory disturbances, and global lipid metabolism defects in Mecp2/Y mice.

169

Figure 4.2: Fluvastatin improves overall health, motor coordination, and respiratory symptoms in male Mecp2/Y mice while normalizing lipid parameters. A. +/Y and Mecp2/Y mice were given a 1.5mg/kg dose of fluvastatin or identical vehicle control by intraperitoneal injection twice weekly from 4 weeks to 8 weeks of age. Effects of fluvastatin treatment on B. lifespan, C. subjective health and D. rotarod performance. Whole body plethysmography (WBP) was used to measure E. breathing frequency, F. tidal volume, G. number of respiratory apneas, and H. response to a methacholine challenge. Serum I) triglycerides and J. cholesterol and lung K. triglycerides and L. cholesterol were measured. n= 10 +/Y and 17 Mecp2/Y mice receiving vehicle-control and 10 +/Y and 14 Mecp2/Y receiving fluvastatin. For plethysmography, n=8. For lipid analysis, n=4. Statistics were assessed using one-way or two- way ANOVA. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

170 4.4.4 Treatment with fluvastatin improves motor and respiratory symptoms and lung lipid metabolism in female Mecp2/+ mice

To validate the therapeutic relevance of metabolism-targeted treatment for human RTT patients, the effect of statin treatment on symptoms in female Mecp2/+ mice was assessed. Female Mecp2/+ mice have a much milder phenotype than their male counterparts with symptom presentation between 4-6 months of age, corresponding more closely to the occurrence of symptoms in human RTT patients. Female Mecp2/+ mice also undergo random X chromosome inactivation, making their symptoms more variable. Our lab previously showed that fluvastatin treatment improves rotarod performance in Mecp2/Y mice while also lowering liver triglycerides (167).

Using the same treatment pipeline as published (167), Mecp2/Y female mice and their wild type littermates were treated with a single weekly dose of 3 mg/kg fluvastatin or vehicle control from 6 weeks to 36 weeks of age (Figure 4.3A). Beginning treatment at 6 weeks of age would presumably allow the correction of aberrant lipid metabolism before it contributes to symptom development. Fluvastatin treatment significantly improved subjective health scores (Figure 4.3B) and rotarod performance (Figure 4.3C) at 24 weeks of age. Additionally, like in male Mecp2/Y mice, fluvastatin treatment normalized heightened breathing frequency in Mecp2/+ mice compared to vehicle-treated Mecp2/+ mice (273.01 ± 8.93 in fluvastatin-treated Mecp2/+; 316.81 ± 13.77 in vehicle-treated Mecp2/+, p=0.046, compared to in 273.03 ± 10.42 in vehicle-treated +/+) (Figure 4.3D). Vehicle-treated Mecp2/+ mice had a non-significant increase in tidal volume compared to +/+ mice, but fluvastatin did not influence this respiratory parameter (Figure 4.3E). Interestingly, despite female Mecp2/+ mice having a lower occurrence of respiratory apneas than their male counterparts, fluvastatin treatment remarkably restored respiratory apnea counts (0.2 apneas/min ± 0.06 in vehicle-treated Mecp2/+; 0.04 ± 0.02 in fluvastatin-treated Mecp2/+, p=0.012) to wild type levels (0.023 ± 0.02) (Figure 4.3F). Further, Mecp2/+ had a strong response to methacholine compared to +/+ mice regardless of treatment (p<0.0001). However, Mecp2/+ mice treated with fluvastatin were significantly more resistant to methacholine as compared to vehicle-treated Mecp2/+mice (Penh score=7.23 ± 0.47 at 50mg/ml methacholine; 9.58 ± 1.28, p=0.0001) (Figure 4.3G). Importantly, these results show that fluvastatin treatment can improve respiratory symptoms in female Mecp2/+ mice, a clinically relevant model of RTT, suggesting that lipid-lowering pharmaceuticals could be used to treat respiratory symptoms in RTT patients.

Lipid quantification results showed that at 9 months of age, compared to +/+ mice, Mecp2/+ mice had increased serum triglycerides (39.80 mg/dL ± 2.9; 66.8 ± 5.8, p=0.0076) and cholesterol (153.59 mg/dL ± 11.6; 234.38 ± 24.0, p=0.011) at 36 weeks (Figure 4.3H-I). Fluvastatin treatment

171 reduced serum cholesterol in Mecp2/+ mice (169.93 ± 8.53, p=0.044), but not serum triglycerides. Mecp2/+ mice also have increased lung triglycerides compared to +/+ mice (14.70 mg/g tissue ± 3.01; 4.14 ± 0.9 in +/+, p=0.004); fluvastatin treatment effectively lowered lung triglyceride levels (7.54 ± 0.72, p=0.0229) (Figure 4.3J). Lung cholesterol was unchanged across all four groups (Figure 4.3K). Importantly, these data show that a low dose of fluvastatin does not affect normal cholesterol metabolism in wild type cells, as serum and lung lipids were unaffected after fluvastatin treatment in +/+ mice, but does improve aberrant lipid accumulation in Mecp2/+ mice.

172

Figure 4.3: Fluvastatin treatment improves overall health, motor and respiratory symptoms in female Mecp2/+ mice while normalizing lipid parameters A. Female +/+ and Mecp2/+ mice were given a 3 mg/kg dose of fluvastatin or identical vehicle control by intraperitoneal injection twice weekly from 6 weeks to 36 weeks of age. Effects of fluvastatin treatment on B. subjective health and C. rotarod performance. Respiratory symptoms were evaluated through D. breathing frequency, E. tidal volume F. number of respiratory apneas, and G. response to a methacholine challenge. Serum H. triglycerides and I. cholesterol and lung J. triglycerides and K. cholesterol were measured. n= 7 +/Y and 8 Mecp2/Y mice receiving vehicle-control and 8 +/Y and 7 Mecp2/Y receiving fluvastatin. For plethysmography, n=7. For lipid analysis, n=5. Statistics were assessed using one-way or two-way ANOVA. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

173 4.4.5 Treatment with the LXR-agonist T0901317 improves neurological symptoms but worsens systemic metabolism

Because the expression of the gene encoding LXRβ, Nr1h2, was decreased in the Mecp2/Y brain, as well as its target lipid transport genes, Abcg1, Abcg4 and Apoe, we hypothesized that lipid transport and excretion could be impaired in Mecp2/Y animals. While LXR expression has not been assessed in other tissues of Mecp2/Y mice, cholesterol and triglyceride accumulation in the lung suggests diminished lipid recycling and impaired reverse cholesterol transport (RCT). RNA- sequencing of the whole lung (in Chapter 3), including macrophages which are the major cell type involved in RCT, did not reveal significant changes in Nr1h2 or Nr1h3, but the LXR-target gene Abcg1 was expressed at low levels (FC: -0.2877, FDR: 0.025). However, because lipids accumulate in Mecp2/Y lungs, one would expect lipid transport genes to be upregulated, so the absence of transcriptional changes in these genes is also concerning. Additionally, de novo lipogenesis is increased in the Mecp2/Y liver (164), which would not be the case if RCT were functioning adequately. Thus, we reasoned that treatment with an LXR-agonist could improve symptoms in Mecp2/Y mice by forcing cholesterol efflux from tissues.

The LXR-agonist T0901317 is a non-specific LXR ligand that is prescribed to prevent atherosclerosis. T0901317 binds and activates LXRs, inducing the expression of LXR target genes, including genes essential for lipid transport. LXR-agonists induce cholesterol excretion from tissues for transport to the liver. This causes an overall decrease in body cholesterol, though transiently increases serum cholesterol (434). Additionally, T0901317 increases hepatic triglyceride synthesis by increasing the activity of SREBP1 pathway (435). Thus, while lowering whole body cholesterol, T0901317 also increases serum and liver lipids.

A previous graduate student, Dr. Stephanie Kyle, led a study assessing the efficacy of T0901317 in reducing neurological symptoms in Mecp2/Y mice. Notably, as this was a preliminary study, neither respiratory symptoms, nor a full lipid panel, were assessed. Both +/Y and Mecp2/Y mice were treated with 10 mg/kg of T0901317 or a vehicle control (1:10 DMSO:saline), twice weekly, from 5 to 10 weeks of age (Figure 4.4A). While only 60% (9/15) of vehicle-treated Mecp2/Y mice survived to postnatal day (P) 70, T0901317 treatment brought survival rates up to 91% (10/11), though this improvement was not statistically significant (Figure 4.4B). T0901317 treatment also reduced subjective health scores in Mecp2/Y mice, indicative of better overall health (Figure 4.4C), and improved motor coordination, as measured by time spent on the rotarod (Figure 4.4D). Indicative of stimulated RCT, serum triglycerides were increased in after treatment with T0901317 in +/Y mice (133.83 mg/dL ± 13.5; 63.57 ± 7.3, p=0.0003) and Mecp2/Y mice, though the difference in the latter was not statistically significant (Figure 4.4E). Serum cholesterol was

174 increased in both T0901317-treated +/Y and Mecp2/Y mice compared to their vehicle-treated counterparts (204.83 mg/dL ± 14.9; 153.86 ± 3.5 for +/Y, p=0.028 and 289.57 ± 14.3; 215.86 ± 11.1 for Mecp2/Y, p=0.0007) (Figure 4.4F). These results suggest increased cholesterol efflux from tissues. However, T0901317 treatment also led to liver hypertriglyceridemia in Mecp2/Y mice (41.07 mg/g tissue ± 8.6; 18.15 ± 2.0, p=0.0019) (Figure 4.4G), consistent with previous reports (435). Despite this, improvements in overall health and motor coordination after T0901317 treatment in Mecp2/Y mice suggest that cholesterol efflux yields neurological rescue, though further treatment with T0901317 is not feasible due to its adverse effect on liver health.

175

Figure 4.4: Treatment with the LXR-agonist T0901317 improves neurological phenotypes but worsens systemic metabolism in male Mecp2/Y mice. A. +/Y and Mecp2/Y mice were given a 10 mg/kg dose of T0901317 or identical vehicle control by intraperitoneal injection once weekly from 5 weeks to 10 weeks of age. Effects of T0901317 treatment on B. lifespan, C. subjective health and D. rotarod performance. Serum E. triglycerides and F. cholesterol and liver G. triglycerides were measured. n= 8 +/Y and 13 Mecp2/Y mice receiving vehicle-control and 10 +/Y and 11 Mecp2/Y receiving T0901317. For lipid analysis, n=4- 7. Statistics were assessed using one-way or two-way ANOVA. *p<0.05, **p<0.01, ***p<0.001.

176 4.4.6 Treatment with the LXR-agonist LXR-623 improves symptoms and systemic lipid metabolism in Mecp2/Y mice

While treatment with the LXR-agonist T0901317 improved neurological symptoms, it was recently deemed unsuitable for clinical use due to adverse effects, including its characteristic induction of hepatic steatosis and hypertriglyceridemia (410). LXR-agonists targeting only one of the two LXR proteins, or those that have tissue-specific effects, may be more beneficial for treating RTT. In the quest to identify such LXR-agonists, the compound, LXR-623, was tested. LXR-623 is a more efficient agonist of LXRβ, which is not highly expressed in the liver. It increases expression of LXR target genes and lowers serum LDL levels in mice and primates (436). Importantly, liver hypertriglyceridemia was only observed with elevated doses of LXR-623, suggesting an improved therapeutic window compared to first-generation LXR agonists (437). Therefore, we assessed whether treatment with LXR-623 could improve RTT-like symptoms in Mecp2/Y mice without sacrificing liver health.

Male +/Y and Mecp2/Y mice were treated with 50 mg/kg LXR-623 or vehicle control (1:10 DMSO:corn oil), once weekly, from 5 weeks to 10 weeks of age (Figure 4.5A). As with T0901317 treatment, treatment with LXR-623 increased lifespan with 88% (14/16) of LXR-623-treated Mecp2/Y mice surviving to P70 compared to only 55% (6/11) vehicle-treated Mecp2/Y mice (p=0.0404, Figure 4.5B). Similarly, LXR-623 treatment significantly improved subjective health scores and motor coordination in Mecp2/Y mice, while having no negative effect on these parameters in +/Y mice (Figure 4.5C,D). LXR-623 treatment also modestly reduced breathing frequency in Mecp2/Y mice, though this improvement was not statistically significant (Figure 4.5E). However, LXR-623 did not alter tidal volume, number of respiratory apneas, or influence response to methacholine in any of the treated mice (Figure 4.5F-H).

HPLC analysis was performed to quantify serum, liver, brain and lung triglycerides and cholesterol in mice receiving LXR-623 treatment or a vehicle control. In contrast to T0901317, LXR-623 treatment lowered serum triglycerides (154.04 mg/dL ± 15.7 to 66.9 ± 10.8, p=0.039) and cholesterol (277.46 mg/dL ± 29.1 to 191.24 ± 16.2, p=0.046) in Mecp2/Y mice, but not in +/Y mice (Figure 4.6A,B). This suggests that, like statins, the effect of LXR-623 is amplified by the amount of cellular lipids available. Similarly, liver triglycerides were also lowered in Mecp2/Y mice after LXR-623 treatment (26.41 mg/g tissue ± 4.7 to 13.77 ± 3.3, p=0.0397), implying functioning lipid excretion in bile acids in the absence of increased liver lipogenesis (Figure 4.6C). There were no significant changes in liver cholesterol, though it trended upward in both +/Y and Mecp2/Y mice after LXR-623 treatment (Figure 4.6D). Brain triglycerides and brain cholesterol were unchanged between +/Y and Mecp2/Y mice and were not altered after LXR-623 treatment in either group

177 (Figure 4.6E,F). This is consistent with previous reports showing that while cholesterol is increased during the lifespan of Mecp2/Y mice, it returns to normal levels by P70 (167). However, the improvement in neurological symptoms in LXR-623-treated Mecp2/Y mice implies LXR activation in the brain; LXR-623 treatment may therefore rescue aberrant brain lipids earlier in the course of disease progression. Finally, lung triglycerides were reduced in Mecp2/Y mice after LXR-623 treatment (22.27 mg/g tissue ± 5.4 to 9.08 ± 1.3, p=0.037) (Figure 4.6G) while lung cholesterol was unchanged across the four groups (Figure 4.6H). Altogether, these results suggest that LXR-623 offers neurological rescue while improving serum, liver, and lung lipids in Mecp2/Y mice.

178

Figure 4.5: Treatment with the LXR-agonist LXR-623 improves health, motor coordination, and breathing frequency in male Mecp2/Y mice. A. +/Y and Mecp2/Y mice were given a 50 mg/kg dose of LXR-623 or identical vehicle control by intraperitoneal injection once weekly from 5 weeks to 10 weeks of age. Effects of LXR-623 treatment on B. lifespan, C. subjective health and D. rotarod performance. Whole body plethysmography (WBP) was used to measure E. breathing frequency, F. tidal volume, G. number of respiratory apneas, and H. response to a methacholine challenge. n= 15 +/Y and 14 Mecp2/Y mice receiving vehicle-control and 10 +/Y and 16 Mecp2/Y receiving LXR-623. For plethysmography, n=7. Statistics were assessed using one-way or two-way ANOVA. *p<0.05, **p<0.01, ***p<0.001.

179

Figure 4.6: Treatment with the LXR-agonist LXR-623 lowers serum, liver and lung triglycerides and serum cholesterol in male Mecp2/Y mice.

Serum and tissue analysis of male +/Y and Mecp2/Y mice receiving a 50 mg/kg dose of LXR-623 or identical vehicle control. Serum A. triglycerides and B. cholesterol, liver C. triglycerides and D. cholesterol, brain E. triglycerides and F. cholesterol, and lung G. triglycerides and H. cholesterol were assessed. n = 4 mice per group. Statistics were assessed using one-way ANOVA. *p<0.05, **p<0.01.

180 4.4.7 Treatment with the LXR-agonist LXR-623 improves motor coordination and respiratory symptoms in female Mecp2/+ mice

Since LXR-623 induced significant health benefits in Mecp2/Y mice, we tested its effects in female Mecp2/+ mice. Like the fluvastatin treatment in female Mecp2/+, LXR-623 treatment began prior to symptom onset. We hypothesized that the activation of lipid efflux genes prior to symptom development could prevent lipid accumulation in Mecp2/+ mice.

Female +/+ and Mecp2/Y mice were treated with 50 mg/kg LXR-623 or a vehicle control (1:10 DMSO:corn oil), once weekly, from 6 weeks to 32 weeks of age (Figure 4.7A). LXR-623 treatment led to a consistent reduction in subjective health scores of Mecp2/+ mice over the course of their lifetime, but this improvement was not statistically significant (Figure 4.7B). LXR-623-treated Mecp2/+ mice also displayed a marked improvement in motor coordination (Figure 4.7C). Breathing frequency in LXR-623-treated Mecp2/+ mice improved (314.37 bpm ± 6.8 in vehicle- treated Mecp2/+; 266.77 ± 9.5 in LXR-623-treated Mecp2/+, p=0.0096) (Figure 4.7D). However, like in male mice, LXR-623 had no effect on tidal volume, respiratory apneas, or response to methacholine (Figure 4.7E-G).

Serum triglycerides were increased in vehicle-treated Mecp2/+ mice compared to +/+ mice (62.35 ± 5.23; 44.71 ± 2.54, p=0.027), but neither serum triglycerides nor cholesterol were altered by LXR-623 treatment (Figure 4.8A,B). Liver triglycerides were also increased in vehicle-treated Mecp2/+ mice compared to +/+ mice (25.99 mg/g tissue ± 4.16; 13.61 ± 0.82, p=0.0275); while LXR-623 treatment reduced liver triglycerides, the change was not statistically significant (Figure 4.8C). In contrast, LXR-623 drastically lowered liver cholesterol in +/+ mice (3.56 mg/g tissue ± 0.19; 2.78 ± 0.09, p=0.0136), but unexpectedly increased this parameter in Mecp2/+ livers (3.08 ± 0.03; 3.89 ± 0.20, p=0.0102) (Figure 4.8D). Further, brain triglycerides and cholesterol were not changed across all four groups at 32 weeks (Figure 4.8E-F), suggesting feedback mechanisms in the vehicle-treated Mecp2/+ brain may regulate brain lipid levels. Finally, lung triglycerides were reduced after LXR-623 treatment in Mecp2/+ mice (10.15 mg/g tissue ± 1.45; 5.46 ± 0.61, p=0.0166) (Figure 4.8G). Lung cholesterol was not changed across the groups, regardless of treatment (Figure 4.8H). Thus, LXR-623 treatment significantly improves motor coordination and breathing frequency in Mecp2/+ mice, while having modest effects on lipid metabolism at 32 weeks of age.

181

Figure 4.7: Treatment with the LXR-agonist LXR-623 improves motor coordination and breathing frequency in female Mecp2/+ mice. A. Female +/+ and Mecp2/+ mice were treated with a 50 mg/kg dose of LXR-623 or identical vehicle control once weekly from 6 weeks to 32 weeks of age. Effects of LXR-623 treatment on B. subjective health and C. rotarod performance. Whole body plethysmography (WBP) was used to measure D. breathing frequency, E. tidal volume, F. number of respiratory apneas, and G. response to a methacholine challenge. n= 6 +/+ and 8 Mecp2/+ mice receiving vehicle-control and 7 +/+ and 10 Mecp2/+ receiving LXR-623. For plethysmography, n=6-8. Statistics were assessed using one-way or two-way ANOVA. *p<0.05, **p<0.01, ***p<0.001.

182

Figure 4.8: Treatment with the LXR-agonist LXR-623 increases liver cholesterol and decreases lung triglycerides in female Mecp2/+ mice. Serum and tissue analysis of female +/+ and Mecp2/+ mice receiving a 50 mg/kg dose of LXR- 623 or identical vehicle control. Serum A. triglycerides and B. cholesterol, liver C. triglycerides and D. cholesterol, brain E. triglycerides and F. cholesterol, and lung G. triglycerides and H. cholesterol were assessed. n = 4 mice per group. Statistics were assessed using one-way ANOVA. *p<0.05.

183 4.4.8 Treatment with fluvastatin and LXR-623 fails to improve symptoms in male Mecp2/Y mice

Both statins and LXR-agonists had beneficial effects in Mecp2-mutant mice through their effects on cholesterol biosynthesis and efflux, respectively. Thus, we hypothesized that targeting both arms of cholesterol metabolism simultaneously would impart greater health benefits in Mecp2/Y mice than treating either arm alone. We hypothesized that the concurrent reduction in cholesterol biosynthesis paired with increased cholesterol efflux would rescue the lipid accumulation phenotype in Mecp2/Y tissues and lead to their efficient export from the body.

To test this, we treated male +/Y and Mecp2/Y mice with 1.5 mg/kg of fluvastatin twice per week and 50 mg/kg of LXR-623 once per week, or a vehicle control (1:10 DMSO:saline or corn oil, respectively), from 5 weeks to 10 weeks of age (Figure 4.9A). Treatment with fluvastatin and LXR-623 (herein, combination-treatment) had no significant effects on survival, subjective health scores, or on motor coordination (Figure 4.9B-D). Consistently, breathing frequency, number of respiratory apneas, and response to methacholine were not altered after combination-treatment (Figure 4.9E,G-H). However, strikingly, combination-treatment increased tidal volume in Mecp2/Y mice compared to +/Y and vehicle-treated Mecp2/Y mice (0.20 ml ± 0.013 in vehicle-treated Mecp2/Y; 0.27 ± 0.013 in combination-treated Mecp2/Y, p=0.017), indicating deviation of this respiratory symptom away from wild type scores (Figure 4.9F).

As previously shown, Mecp2/Y mice had increased serum triglycerides compared to +/Y mice; combination-treatment did not improve serum triglycerides and had no effect on serum cholesterol (Figure 4.10A-B). Consistently, increased liver triglycerides in Mecp2/Y mice were not affected by combination treatment and liver cholesterol was unchanged across all four groups (Figure 4.10C-D). Interestingly, combination treatment appeared to reduce brain cholesterol and triglycerides in +/Y and Mecp2/Y mice, though these changes were not statistically significant (Figure 4.10E-F). Finally, lung triglycerides appeared to increase, though not significantly, and lung cholesterol did not change in Mecp2/Y mice after combination treatment (Figure 4.10G-H). These results show that the combination of fluvastatin and LXR-623 does not offer any therapeutic benefits in Mecp2/Y mice.

184

Figure 4.9: Treatment with fluvastatin and LXR-623 has no therapeutic benefit on symptoms in male Mecp2/Y mice. A. +/Y and Mecp2/Y mice received two weekly doses of 1.5 mg/kg fluvastatin and one weekly dose of 50 mg/kg of LXR-623 or identical vehicle control from 5 weeks to 10 weeks of age. Effects of combination treatment on B. lifespan, C. subjective health and D. rotarod performance. Whole body plethysmography (WBP) was used to measure E. breathing frequency, F. tidal volume, G. number of respiratory apneas, and H. response to a methacholine challenge. n= 8 +/Y and 9 Mecp2/Y mice receiving vehicle-control and 7 +/Y and 10 Mecp2/Y receiving combination treatment. For plethysmography, n=6-7. Statistics were assessed using one-way or two-way ANOVA. *p<0.05, **p<0.01, ***p<0.001.

185

Figure 4.10: Treatment with fluvastatin and LXR-623 imparts no benefits on lipid metabolism in male Mecp2/Y mice. Serum and tissue analysis of male +/Y and Mecp2/Y mice receiving a 2x weekly dose of 1.5 mg/kg fluvastatin and a 1x weekly dose of 50 mg/kg LXR-623 or identical vehicle control. Serum A. triglycerides and B. cholesterol, liver C. triglycerides and D. cholesterol, brain E. triglycerides and F. cholesterol, and lung G. triglycerides and H. cholesterol were assessed. n = 4 mice per group. Statistics were assessed using one-way ANOVA. *p<0.05, **p<0.001.

186 4.5 Discussion

4.5.1 Brain cholesterol metabolism

Studies of cholesterol metabolism have highlighted its many important roles in the nervous system. Cholesterol synthesis in the developing mouse brain peaks during the second postnatal week, varying across different brain regions (438). In rodents, the majority of neurogenesis occurs before birth, but synaptogenesis coincides with this period, occurring during the first 2 weeks of life (439). The brain generally overproduces the number of synapses formed by 50%, then selective pruning throughout the brain maturation period brings synapses to their final adult numbers (440). Like other vertebrates, the production of myelin occurs largely postnatally in mice, beginning in the brainstem and spinal cord at ~P0 and advancing to the cerebral cortex by P12. Production of myelin peaks at approximately P21, then gradually declines until peak density is reached (441). Both synaptogenesis and myelination rely heavily on de novo cholesterol biosynthesis in the brain (442).

Current knowledge on the contribution of individual cell types to CNS cholesterol homeostasis is limited. Oligodendrocytes are believed to produce most of the brain’s cholesterol cell- autonomously as they require large amounts of cholesterol for myelin production (442). Neurons and astrocytes also have a high demand for cholesterol. Neurons have large membrane surfaces consisting of axons, dendrites, and synapses, the latter of which have a high cholesterol content in postsynaptic spines and presynaptic vesicles (443). As such, cholesterol depletion severely reduces synaptic vesicle exocytosis and synaptic transmission (424). Blocking cholesterol biosynthesis in mouse neuronal precursor cells leads to the apoptosis of newly generated neurons and perinatal lethality (444), suggesting newborn neurons must synthesize cholesterol cell- autonomously to survive. However, a prevalent hypothesis postulates that mature neurons outsource resource-expensive cholesterol biosynthesis to astrocytes (445). Accordingly, genetic ablation of cholesterol synthesis in mature cerebellar granule cells does not affect neuronal function, synaptogenesis or synaptic function (446).

The mechanism by which astrocytes coordinate the shuttling of cholesterol to neurons is still poorly understood. Astrocytes are thought to excrete cholesterol in APOE-containing lipoproteins which are taken up by neurons through low-density lipoprotein receptor (LDLR) or low-density lipoprotein receptor-related protein (LRP1) (447). Cultured astrocytes express Abca1, Abcg1, and Abcg4 at the transcript and protein level, suggesting they transport cholesterol through similar mechanisms to hepatocytes (447). Consistently, cholesterol efflux from cultured astrocytes is enhanced by treatment with LXR-agonists (448). Surplus neuronal cholesterol is eliminated by conversion of cholesterol into 24S-OHC by the enzyme CYP46A1 (427). Oxysterols, as

187 oxygenated derivatives of cholesterol, can freely pass the blood-brain barrier. 24S-OHC is a potent activator of LXRs, preventing the transcription of SREBP2 and its downstream cholesterol biosynthesis targets (355). Consistently, mice lacking Cyp46a1 experience a 40% reduction in brain cholesterol biosynthesis owing to reduced efflux (449).

4.5.2 Cholesterol metabolism is transcriptionally altered in the brain of Mecp2/Y mice

The transcriptional findings in this study suggest many possibilities for altered cholesterol metabolism. Genes encoding cholesterol biosynthesis enzymes are expressed at high levels in pre-symptomatic P21 Mecp2/Y brains. The decreased expression of genes encoding lipid- sensing transcriptional regulators SREBP1/2 (Srebf1 and Srebf2) in the brain further suggest a global increase in brain lipid content prior to P21. Thus, increased levels of cholesterol in the Mecp2/Y brain coincide with the period of peak cholesterol synthesis in the developing mouse brain (438). At a time of rapid and significant myelination and synaptogenesis, two processes highly dependent on cholesterol, aberrant cholesterol concentrations could have profound detrimental effects. Notably, absolute cholesterol biosynthesis rates in the whole brain of Mecp2/Y mice were not changed at P7 or P14, but were reduced from P21 onward (450). Specific brain regions may be more or less effected by altered cholesterol metabolism. Assessing cholesterol biosynthesis during this crucial developmental window in individual cell types of the Mecp2/Y brain using single-cell RNA sequencing will be extremely informative.

The transcription factors Nr1h2 and Nr1h3, which encode LXRβ and LXRα, respectively, are expressed at low levels in the Mecp2/Y brain. Consistently, the expression of the LXR targets Abcg1, Abcg4, and Apoe is also decreased. Mice lacking both isoforms of LXR (LXRα-/-β-/-) accumulate lipid droplets in their brains, and display a decreased number of neurons, proliferation of astrocytes, and disorganized myelin sheaths (428). Genetic deletion of Abcg1 and Abcg4 leads to cholesterol accumulation in astrocytes (451). The decreased expression of these genes in the Mecp2/Y brain raises a new possibility: perhaps the process of cholesterol transport from astrocytes to neurons is impaired. As such, cholesterol would accumulate in astrocytes, while neurons would have unmet cholesterol needs. This could promote a feedback cycle in astrocytes to limit cholesterol biosynthesis over time due to their high intracellular levels (167,450). Consistently, neurons are thought to be primarily responsible for cholesterol efflux from the brain and adult Mecp2/Y expression of Cyp46a1 is reduced by 40% (167), suggesting reduced neuronal cholesterol levels at later time points.

188 4.5.3 Pharmacological targeting of cholesterol biosynthesis in Mecp2/Y mice

Cholesterol metabolism is highly targetable, and here, we used statins and LXR-agonists to reduce cholesterol biosynthesis and induce cholesterol transport, respectively. Statins are a commonly prescribed class of cholesterol-lowering medications that act through competitive inhibition of HMG-CoA reductase, the rate-limiting enzyme in the mevalonate pathway. Consistent with previous reports, treatment of symptomatic Mecp2/Y mice with fluvastatin twice weekly robustly improved overall health and motor coordination, improved respiratory symptoms, and improved measures of lipid accumulation. Due to this drastic phenotype improvement, female Mecp2/+ mice, a more clinically relevant model of RTT, were treated with fluvastatin once per week. Notably, heterozygous female Mecp2/+ mice do not reach the symptom severity seen in male Mecp2/Y mice and have greater variations in their phenotypes due to random X chromosome inactivation. Yet, treatment with fluvastatin improved health, motor coordination and respiratory symptoms in female Mecp2/+ mice. Remarkably, fluvastatin lowered the number of respiratory apneas in Mecp2/+ mice, a devastating symptom associated with frequent fainting and cyanosis in patients (371). Fluvastatin drastically reduced lung triglycerides in both male and female Mecp2-mutant animals, likely leading to respiratory symptom improvement.

Importantly, statins have pleiotropic effects that could aid symptom improvement in Mecp2-mutant mice. Acting on HMG-CoA reductase in the mevalonate pathway, statins can also inhibit the downstream production of isoprenoid intermediates including farnesyl-pyrophosphate (FPP) and geranylgeranyl-pyrophosphate (GPP). FPP and GPP are isoprenoids that serve as substrates in protein prenylation, an important post-translational modification (PTM) of proteins (452). Members of the Ras and Rho GTPase family are major substrates for PTMs by prenylation and their functions in immune response and endothelial homeostasis are altered after treatment with statins. Inhibition of protein prenylation through statin treatment is being studied as a treatment for Fragile X syndrome (453). Additionally, statins have anti-inflammatory effects through reducing C-reactive protein (CRP), inflammatory cytokines, ROS generation, and macrophage infiltration, though the mechanism underlying this effect is unclear (454).

Notably, Mecp2-mutant mice in our study were treated with 3 mg of fluvastatin per kg of body weight, per week. Children with dyslipidemia may be prescribed low doses of statins at 20 mg daily (for a 40kg child, 3.5 mg/week) or moderate doses of 40 mg daily (for a 40kg child, 7 mg/week). Adults can be treated with up to 80 mg of statins, daily. Statin doses used in this study are similar to or lower than clinical levels, yet still produced beneficial effects. Additionally, statins are effective when cellular cholesterol levels are high, while having no noticeable effects under basal conditions (455) .Thus, RTT patients may benefit from low dose statins.

189 4.5.4 Activation of cholesterol transport through LXR-agonists

A second targetable facet of cholesterol metabolism is mediated through LXR-agonists. As their name suggests, LXR agonists bind ligand-activated LXR transcription factors. Activated LXRs then bind to LXREs in promoter regions of genes involved in cholesterol transport and efflux. Here, treatment with 10 mg/kg of the first-generation LXR-agonist T0901317 improved health and motor phenotypes in Mecp2/Y mice, but increased serum and liver lipids. While used extensively in preclinical research, first-generation LXR-agonists are unsuitable for clinical use due to these pleiotropic effects (410). Through increasing cholesterol efflux from tissues, they simultaneously lower total body cholesterol while increasing serum and liver lipids. T0901317 also induces the expression of SREBP1, leading to increased hepatic lipogenesis (430). Thus, the LXR-agonist T0901317 had beneficial effects on neurological symptoms in Mecp2/Y mice, though at the expense of liver health.

To circumvent this issue, we focused our efforts on a newer LXR-agonist, LXR-623. LXR-623 activates LXR-induced target gene expression in rodents and primates and only causes hypertriglyceridemia with elevated doses. Mecp2/Y mice treated with 50 mg/kg LXR-623 showed an improvement in overall health and motor coordination but saw no improvement in respiratory symptoms. LXR-623 treatment also lowered serum lipids and liver and lung triglycerides. The concurrent lowered lung lipids and lack of improvement in respiratory symptoms is intriguing. While respiratory frequency was reduced after LXR-treatment in Mecp2/Y males, though not significantly, response to the methacholine challenge was unchanged. Methacholine induces constriction of the bronchioles and increased response to the methacholine challenge is indicative of a hyperreactive airway. The finding that treatment with fluvastatin, but not LXR-623, improves airway reactivity suggests that the pleiotropic effects of statins may improve respiratory health in Mecp2/Y mice. For example, it is possible that aberrant lung lipids induce an inflammatory response, leading to airway hyperreactivity. Notably, increased inflammation has been reported in Mecp2/Y lungs (333).

LXR-treatment in female Mecp2/+ mice yielded similar effects to that in male Mecp2/Y mice. Overall health was not significantly improved. Anecdotally, Mecp2/+ mice treated with LXR-623 weighed more than their vehicle-treated counterparts, which increased their subjective health scores, though the cause for this is unknown. Despite increased weight, LXR-623 treated mice showed improved motor coordination, reduced breathing frequency, and reduced lung triglycerides. Interestingly, LXR-623 treatment reduced liver cholesterol in +/+ mice but increased this parameter in Mecp2/+ mice. This result suggests that the effect of LXR-623 may be amplified in conditions of high cholesterol, drastically increasing RCT and thus liver cholesterol levels, like

190 T0901317. Notably, brain cholesterol was unchanged between male and female wild type and Mecp2-mutant mice; as described previously, brain cholesterol returns to normal levels in adult Mecp2-mutant mice, the time of sampling in this study (167).

Intriguingly, concurrent treatment with fluvastatin and LXR-623 offered no significant health benefits in Mecp2/Y mice, while appearing to increase respiratory apneas and lung triglycerides. This was surprising, however, there are a few possible reasons for these negative results. First, inhibition of cholesterol biosynthesis through statins reduces intracellular cholesterol. In response to this reduction, lipid-sensing SREBP can enter the nucleus and activate the transcription of cholesterol biosynthesis enzymes (347). As such, statin treatment that lowers cholesterol beyond physiological levels paradoxically increases cholesterol biosynthesis. In addition, SREBP is a transcriptional target of LXRs and treatment with LXR agonists activates Srebf1/2 expression, further increasing cholesterol biosynthesis (430). Thus, while attempting to lower cholesterol biosynthesis and facilitate its transport, we may have inadvertently upregulated cholesterol biosynthesis. A second possibility is that targeting two arms of cholesterol metabolism simultaneously is not ideal. Drug-drug interactions and/or the ability of these compounds to be recognized by cells may vary when taken together. Additionally, the feedback pathway described above may render these drugs useless when taken together. To this respect, fluvastatin and LXR- 623 may be more effective when treatment is temporally separated. For example, LXR-623 may be more beneficial in early stages of the disease when lipids are in excess and require transport and excretion (ex. Between P14-P28 in male Mecp2/Y mice). Then, statins may be administered to prevent further aberrant cholesterol biosynthesis (ex. P28-P56). Alternatively, statins and LXR- 623 treatment could be separated by time of day. Lipid metabolism is regulated by circadian rhythm and patients are advised to take statins in the evening as lipogenesis is highest in the fasted state. Statins were administered to mice in the morning, corresponding to this period, though LXR-agonists may be more effective if administered at a different time of day.

4.5.5 Treating cholesterol metabolism in RTT patients

A subset of RTT patients experience dyslipidemia (16,190), suggesting that at least a portion of them would benefit from treatment targeting lipid metabolism. Aberrant cholesterol homeostasis has been implicated in a number of other neurological disorders including Fragile X syndrome (456,457), Alzheimer’s disease (458), Parkinson’s disease (459), and Huntington’s disease (460). Mutations in ATP binding cassette protein 1 (ABCA1) cause the neuropathy, Tangier disease, resulting in defective cholesterol efflux, decreased serum cholesterol and hypertriglyceridemia (461). Cholesterol trafficking is affected in Niemann-Pick Disease, type C (NPC), a lysosomal storage disease caused by mutations in NPC1 or NPC2 (462). A small number of NPC patients

191 die within the first months of life from hepatic or pulmonary failure, but surviving patients show progressive neurodegeneration leading to ataxia, dystonia and psychological symptoms (463).

Statins and LXR-agonists are but few of the many metabolic modulators being developed to treat lipid accumulation and cholesterol perturbations. Fibrates, which activate the nuclear receptor peroxisome proliferator-activated receptor α (PPARα), increase fatty acid oxidation and reduce triglyceride synthesis. Fibrates also increase the production of ApoA1, leading to increased RCT. Niacin, through a poorly understood mechanism, lowers liver lipogenesis and directly inhibits DGAT2 to prevent triglyceride biosynthesis. Omega-3 fatty acids also inhibit lipogenesis while also having anti-inflammatory effects. Just earlier this year, the Food and Drugs Administration (FDA) approved a new cholesterol-lowering drug, bempedoic acid. Bempedoic acid inhibits the enzyme ATP citrate lyase (ACLY), which is responsible for converting citrate into acetyl-CoA, an essential substrate for cholesterol and fatty acid synthesis. Further studies are needed to assess the efficacy of these compounds in Mecp2/Y mice, and in other models of neurological disease associated with aberrant lipid metabolism.

192

Chapter 5 Summary and Future Directions

193 ummary and Future Directions

5.1 Summary and Significance

Rett syndrome (RTT) is a rare neurological disorder caused by mutations in the X-linked gene, methyl-CpG-binding protein 2 (MECP2) (36). Patients with RTT undergo seemingly normal postnatal development until 6-18 months of age when they experience a developmental regression; during this time, patients lose acquired verbal and motor skills, halt purposeful hand movements, and develop respiratory symptoms and seizures (4). Therapeutic options for RTT are limited to symptom management; physical and speech therapy are employed to improve motor and verbal skills, while anti-epileptic drugs and gastrostomy tubes may be used to control seizures and optimize nutrition, respectively (27,464). Despite living into middle age, most patients require around-the-clock care.

To date, 562 RTT-causing mutations have been identified in MECP2, though eight mutations account for approximately 70% of cases (63,64). The development of Mecp2-mutant mice has facilitated the study of the gene and has enhanced our understanding of RTT pathogenesis (134,135,169). The MECP2 gene encodes an evolutionarily-conserved nuclear protein that is ubiquitously expressed, though it is most abundant in neurons (51,71). MECP2 is an intrinsically disordered protein that acquires a tertiary structure through interactions with protein partners and post-translational modifications (57,58). MECP2 is considered a multi-functional hub protein associated with transcriptional repression, transcriptional activation, chromatin remodeling, alternative splicing, and miRNA processing, though its major role is thought to be transcriptional repression (52,89). The MECP2 protein contains a methyl-CpG-binding domain (MBD) and transcriptional repression domain (TRD) (89). MECP2 binds to the nuclear receptor corepressor 1/2 (NCOR1/2) complex through a small region within the TRD named the NCOR1/2 interaction domain (NID), acting as a bridge between methylated DNA and the repressor complex. This MECP2-NCOR1/2 interaction is believed to be essential, as mutations disrupting either end of the MECP2 bridge causes RTT symptoms (108).

Respiratory symptoms in RTT include a pattern of irregular hyperventilation with periods of apneas and forced breathing. Up to 80% of RTT patient death is associated with respiratory distress (31). While RTT-associated respiratory symptoms are largely considered to originate in the brainstem, studies have been unable to pinpoint their causative mechanisms or facilitate their long-term rescue (185,192,194,198,212,328,331,332). Despite MECP2’s ubiquitous expression, the RTT lung is relatively uncharacterized. Accordingly, we hypothesized that genetic loss of

194 Mecp2 could result in lung defects and associated respiratory symptoms in a Mecp2-mutant mouse model of RTT.

5.1.1 Aberrant lung lipid metabolism in Mecp2-mutant mice

In Chapter 2, I found aberrant lipid accumulation in the lungs of both male and female 129.Mecp2tm1.1Bird mice through electron microscopy analysis. Lipid quantification assays showed that while triglycerides and cholesterol are strikingly increased in Mecp2/Y lungs, phosphatidylcholines (PCs) are decreased in the bronchoalveolar lavage (BAL) fluid. Phosphatidylcholines, especially dipalmitoylphosphatidylcholine (DPPC), are essential components of lung surfactant and are solely responsible for reducing surface tension in the lung and preventing lung collapse. Importantly, lung triglycerides were also increased in female Mecp2/+ lungs, while BAL PCs were reduced, indicating that heterozygous mosaic expression of Mecp2 is insufficient to prevent aberrant lung lipid metabolism. This is the first time that aberrant lung lipids have been described in Mecp2-mutant mice and provides further support for abnormal lipid metabolism as a pathogenic process in RTT.

As MECP2 is highly expressed in lung alveolar epithelial 2 (AE2) cells, I employed single-cell sequencing to identify transcriptomic changes in Mecp2-deficient AE2 cells at postnatal day (P) 18. Many lipid metabolism genes were misregulated in Mecp2/Y AE2 cells; for example, the expression of hydroxymethylglutaryl-CoA synthase (Hmgcs1) and phosphomevalonate kinase (Pmvk), genes encoding enzymes in the cholesterol biosynthesis pathway, were increased. Likewise, the expression of acyl-CoA thioesterase 1 (Acot1), which channels fatty acids away from beta-oxidation toward storage pathways, was also increased. In contrast, genes involved in fatty acid biosynthesis (ATP citrate lyase (Acly), fatty acid synthase (Fasn), stearoyl-CoA desaturase 1/2 (Scd1,2), and ELOVL fatty acid elongase (Elovl1)), and PC synthesis (choline kinase alpha (Chka), phosphate cytidylyltransferase 1 choline alpha (Pcyt1a)) were drastically decreased. SREBP1 transcript and protein expression was decreased in Mecp2/Y lungs, suggesting altered downstream expression of SREBP1-regulated genes. Notably, expression of 11 of 13 mitochondrially-encoded electron transport chain (ETC) components was decreased, adding to accumulating evidence of mitochondrial involvement in RTT (20,23,357,372,465).

5.1.2 MECP2 regulates transcription of lung lipid metabolism genes with the NCOR1/2 co-repressor complex

MECP2 is an established transcription regulator best known for its role in bridging the NCOR1/2 co-repressor complex to methylated promoters. I determined that MECP2 binds to NCOR1 and its complex members histone deacetylase 3 (HDAC3) and TBL1X receptor 1 (TBL1XR1) in the

195 lung. In support of a direct role for MECP2 in the regulation of lung lipid metabolism, chromatin immunoprecipitation (ChIP) of TBL1XR1 revealed that it, and presumably the NCOR1/2 complex, binds to the promoter region of Hmgcs1 and Acot1. Furthermore, this binding is diminished in Mecp2/Y lungs, suggesting that Mecp2 deficiency prevents NCOR1/2 binding at these loci, allowing their active transcription. Increased expression of the cholesterol biosynthesis enzyme Hmgcs1 and the regulator of fatty acid fate, Acot1, alone could account for the increased accumulation of cholesterol and triglycerides seen in Mecp2-mutant lungs.

To our knowledge, this is the first study of MECP2’s basic molecular function in the lung and we have confirmed its role in transcriptional repression. ome of MECP2’s transcriptional targets have been confirmed in the brain and liver (60,89,164), though MECP2 likely regulates different genes depending in different tissues. Here, an interaction between MECP2 and the NCOR1/2 corepressor complex was confirmed in the lung and shown to be responsible for lung pathogenesis in an RTT model, further implicating the importance of NCOR1/2 in RTT.

5.1.3 Lung-specific deletion of Mecp2 impairs lung lipid metabolism and causes respiratory symptoms

As Mecp2 prevents NCOR1/2 complex binding to lipid metabolism genes, we hypothesized that MECP2 regulates lung lipid metabolism in a cell-autonomous manner. To test this, we generated mice with an AE2 cell-specific deletion of Mecp2 by crossing B6.Mecp2tm1Bird (Mecp2-flx) mice with B6.Sftpctm1(cre/ERT2)Blh mice, which express Cre under a tamoxifen-inducible promoter highly expressed in AE2 cells. Independently, Mecp2 was deleted from hindbrain and cerebellar neurons using Cg-Tg(Atoh1-cre)1Bfri mice. AE2 cell-specific deletion of Mecp2 increased lung triglycerides, increased baseline breathing frequency and respiratory apneas, and elevated response to methacholine, a bronchoconstrictant used for assessing airway reactivity. In contrast, mice with a hindbrain neuron-specific deletion of Mecp2 did not have altered lung lipids but had an increased tidal volume and increased incidence of respiratory apneas. These results indicate that lung-specific loss of Mecp2 is sufficient to alter lung lipid metabolism and cause respiratory symptoms, stressing the importance of lung-expressed Mecp2. Additionally, these results suggest that loss of Mecp2 from either the lung or hindbrain neurons impart distinct respiratory symptoms, where the former alters breathing frequency, and the latter increases tidal volume. In contrast, deletion from both centers independently increased baseline respiratory apneas, though mice with a ubiquitous deletion of Mecp2 had a far greater incidence of apneas than either conditional deletion alone; this suggests that loss of Mecp2 from both the lung and hindbrain have additive effects in producing apneas. These findings demonstrate a cell-autonomous role for AE2 cell- expression Mecp2 in regulating lipid metabolism and normal respiration.

196 5.1.4 RNA-sequencing of Mecp2/Y lungs reveals an altered transcriptome

In Chapter 3, I showed that respiratory symptoms precede the onset of overt neurobehavioral phenotypes in Mecp2-mutant mice, and that MECP2 is expressed in a variety of pulmonary cell types. Thus, we focused our studies on the Mecp2-deficient lung as a whole. RNA-sequencing of the P21 lung revealed 1,721 genes with differential expression in Mecp2/Y mice. This contrasts the subtle transcriptional changes reported in Mecp2/Y brains, which could be attributed to its heterogeneous and complex composition (83,466). Genes with increased expression included many affecting circadian rhythm, and subsequent RT-qPCR analyses confirmed altered clock- regulating gene expression in Mecp2/Y lungs. Genes with decreased expression were enriched for extracellular matrix (ECM) function. To our knowledge, this is the first time RNA-sequencing has been performed in Mecp2-mutant tissues outside of the central nervous system.

The lung ECM plays an essential role in development and homeostasis while also providing structural support for pulmonary cells. Extensive crosslinking of elastin and collagen proteins contribute to the ECM’s stiffness, and therefore, the lung’s structural integrity (298). Additional components, including glycoproteins and proteoglycans, mediate cell-matrix adhesion and contribute to the mechanical stability of the elastin-collagen network (301). Modifications in ECM component ratios therefore alter their relative contributions to ECM integrity.

Elastin (Eln) was one of the most significantly decreased genes in Mecp2/Y lungs. Elastin is an essential component of the lung ECM which facilitates the development of mature alveoli, maintains lung homeostasis as well as injury repair, and establishes the elastic recoil of the lung essential for normal exhalation. In addition to decreased Eln transcript and protein expression, genes encoding proteins that constitute or contribute to mature elastin fibers, including fibrillin (Fbn1) and lysyl oxidases (Loxl1, 2, 3), were also expressed at low levels. Elastin characteristically forms a complex network with collagen, and expression of 20 collagen genes was reduced in Mecp2/Y lungs. Additionally, the expression of numerous proteoglycans essential for ECM homeostasis, cell-ECM adhesion, and signal transduction, such as adamalysins, integrins, and laminins, was also reduced in Mecp2/Y lungs. Altogether, these gene expression changes imply an inadequate and highly unstable lung ECM in Mecp2/Y lungs.

5.1.5 Lung structure and function are altered in Mecp2-mutant mice

Consistent with the reduced expression of ECM genes, lung structure was also altered in Mecp2/Y mice. During postnatal development, Mecp2/Y lung had reduced septation and their alveolar walls remained thick. In adult male and female Mecp2-mutant mice, alveolar walls are further degraded, producing an emphysema-like phenotype. Notably, enlarged alveolar spaces were recently

197 reported in B6.Mecp2/Y mice as well (334). Additionally, bronchiolar enlargement was seen in both male and female Mecp2-mutant mice, suggesting bronchiectasis. Enlarged airways and airspaces likely impair normal gas exchange.

Accordingly, I assessed pulmonary function in Mecp2-mutant mice using forced oscillation technique (FOT). Inspiratory capacity was increased in both male and female Mecp2-mutant mice, indicating that their lungs can hold more air than their wild type counterparts. Consistently, lung elastance was decreased while lung compliance was increased, representative of a lung that can be easily hyperextended, but which lacks the recoil properties to return to its original form. These results were highly consistent with RNA-sequencing findings as elastin is solely responsible for the lung’s elastic recoil while collagens and other ECM proteins contribute to its tensile strength. Reduced expression of these ECM genes, in concert with the structural abnormalities described above, likely impair pulmonary function in Mecp2-mutant mice. Finally, lung resistance was decreased, consistent with bronchial enlargement. Therefore, Mecp2-mutant mice have a distinct pulmonary function profile reflective of their structural changes.

These findings are important for many reasons. First, Mecp2 deficiency imparts structural changes on the lung; therefore, lung imaging should be considered in clinical assessments of RTT patients with severe respiratory dysfunction. This may also suggest that early treatment of RTT will be necessary to avoid irreversible alveolar wall degradation. Second, I show for the first time that Mecp2 deficiency impairs pulmonary function. Lungs with decreased elastance are easily extended during inhalation, though exhalation is troublesome; this can lead to air trapping within the lungs and cause alveolar collapse, further limiting gas exchange. Further, indications for COPD and other respiratory diseases may be useful for RTT.

5.1.6 Cholesterol metabolism is altered in the brain

In Chapter 4, we shifted our focus to remediation of the effects of Mecp2 deficiency. Our lab performed RNA-sequencing, finding that expression of cholesterol biosynthesis enzymes is already increased across five different brain regions in pre-symptomatic Mecp2/Y mice. As in Mecp2/Y lungs, the expression of sterol-sensing Srebf genes was decreased, suggesting cholesterol has already accumulated in the Mecp2/Y brain prior to P21. Additionally, the expression of Nr1h3 and Nr1h2, which encode LXRα and LXRβ, respectively, was decreased, along with their cholesterol transport target genes. These results imply cholesterol biosynthesis is constitutively increased while cholesterol transport is impaired in the Mecp2/Y brain.

198 5.1.7 Lipid metabolism modulators improve symptoms in Mecp2-mutant mice

Altered expression of lipid metabolism genes in the brain, together with abnormal lung lipid metabolism presented in Chapter 2, and previously reported lipid accumulation in the liver of Mecp2-mutant mice (164), suggests that pharmacological treatment of global lipid metabolism could improve symptoms of RTT. As such, we treated mice with fluvastatin, a cholesterol-lowering drug that was previously shown to improve neurological symptoms and extend lifespan in Mecp2- mutant animals (167). Consistent with previous reports, treatment of symptomatic male Mecp2/Y mice, or pre-symptomatic female Mecp2/+ mice, had robust effects, improving overall health and motor coordination, alleviating respiratory symptoms, and reducing lung lipid accumulation. These results are significant as statins are already widely prescribed and safe for use in children (455).

We next treated Mecp2/Y mice with the LXR agonist T0901317 to promote cholesterol efflux from tissues. While this drug improved neurobehavioral symptoms, it increased serum and liver lipids, a common side effect of first generation LXR agonists. Subsequent treatment with the second generation LXR agonist LXR-623 improved neurobehavioral symptoms and lowered peripheral lipid levels, while showing only modest, non-significant effects on respiratory symptoms, without worsening systemic metabolism. LXR-623 treatment in Mecp2/+ females yielded similar results. Thus, we treated mice with both fluvastatin and LXR-623 to determine if combinatorial treatment could provide more significant health benefits. Unfortunately, simultaneous treatment with both drugs had no effect on any symptoms in mice.

A subset of RTT patients experience dyslipidemia (15), suggesting at least some patients will benefit from lipid metabolism modulators. The use of statins for RTT treatment is currently being tested in a clinical trial; however, neurological function, but not lung function, is being assessed. While a combination of fluvastatin and LXR-623 was not effective, it is possible that other combinations of metabolism modulators will yield more beneficial results. Importantly, these therapeutics should be able to act within the CNS to treat neurological symptoms of RTT, while still lowering lipids in peripheral tissues.

5.1.8 Additional implications for RTT

Several links can be made between different chapters of my dissertation that can inform studies on RTT. In Chapter 2, I showed that alterations in pulmonary surfactant and triglyceride accumulation are linked to respiratory symptoms, as AE2 cell-specific deletion of Mecp2 increased breathing frequency and apneas. Surfactant deficiency can impair gas exchange. Changes in blood oxygen or pH are detected by chemoreceptors which send excitatory inputs to the hindbrain, eliciting an increased rate and amplitude of breathing, leading to the increased

199 breathing frequency seen in Mecp2-mutant mice. In Chapter 3, I showed that Mecp2-mutant lungs are hyperextendable. Interestingly, mechanoreceptors, including the Hering-Breuer inflation reflex, prevent overinflation of the lungs by detecting excessive stretching (265). As Mecp2- mutant lungs are more prone to hyperextension, this response may be heightened; in turn, excitatory inputs are sent through the vagus nerve to the hindbrain to inhibit respiration, which may reduce tidal volume and promote respiratory apneas. Consistently, mice with a deletion of Mecp2 in hindbrain neurons alone have an increased tidal volume, a phenotype not seen in Mecp2/Y mice, possibly because the lung is healthy and not overextended. Additionally, AE2- specific deletion of Mecp2 causes surfactant deficiency and changes in pulmonary mechanics, recapitulating breathing frequency and apnea changes in Mecp2/Y mice.

A second link is between abnormal lung lipids and bronchiolar enlargement. In Chapter 2, I showed that mice with an AE2 cell-specific deletion of Mecp2 have a heightened response to methacholine. Methacholine is a bronchoconstrictant that is used to diagnose inflammatory lung conditions, including asthma and COPD (404). In Chapter 3, I showed that AE2 cell-specific deficiency of Mecp2 causes bronchiolar enlargement, a phenotype normally seen after prolonged inflammation and/or exposure to reactive oxygen species (ROS) (386). We therefore hypothesize that Mecp2-mutant lungs may be subject to excess inflammation and/or ROS due to lipotoxicity and related mitochondrial impairment. Importantly, treatment with statins in Chapter 4 lowered lung triglycerides and improved methacholine response in Mecp2-mutant mice, providing further evidence for a link between lung lipids, inflammation, and structural changes. In this manner, early treatment of lung lipid abnormalities in RTT patients could reduce inflammation and prevent irreversible physical changes to the lung.

Finally, over one third of RTT patients die due to respiratory infection (31). Surfactant proteins A and D (SPA and SPD) have important roles in the immune response through binding pathogens and facilitating their clearance by alveolar macrophages (285). Surfactant deficiency, or alterations in surfactant fluidity and spreading through decreased PCs evidenced in Chapter 2, can reduce their antimicrobial properties. Clearance of accumulated lipids could also occupy macrophage clearance efforts, reducing their ability to clear pathogens. Structural changes in Mecp2-mutant lungs shown in Chapter 3 can also contribute to increased lung infection; abnormally enlarged airways are susceptible to bacterial colonization due to impaired mucociliary clearance. Therefore, Mecp2-mutant lungs, through the changes evidenced here, likely have an environment that advantages pathogenic microorganisms. Treating lung lipids and inflammation may reduce the incidence of lung infection in patients.

200 5.2 Future Directions

5.2.1 Are lung abnormalities in Mecp2-mutant mice translatable to human RTT patients?

An obvious next step for our studies is to assess lung lipids, pulmonary function, and lung tissue architecture in RTT patients; however, this comes with many challenges. The assessment of surfactant lipids is achieved through sputum collection, a somewhat invasive and unpleasant procedure requiring forced deep coughing. Pulmonary function is assessed using spirometry, which requires physician-instructed timed breathing. Neither procedure is feasible for RTT patients as they require extensive patient cooperation and could transiently worsen respiratory symptoms or increase the risk of lung infection, either of which could be fatal. Lung lipids and tissue architecture assessments could be achieved through lung biopsy. Again, this is not feasible in RTT patients as their respiratory symptoms could be fatally worsened by this procedure.

Lung imaging can be performed on patients; one study showed thickened bronchial walls, ground- glass opacities, and bronchiectasis in RTT lungs using computed tomography (CT) scans (333). These findings support our results in suggesting lung disease in RTT patients. CT scanning, which is the current gold standard for lung imaging, is used for the diagnosis and monitoring of emphysema (467). However, CT scanning does not have the high resolution needed to detect changes in alveolar structure or the accumulation of lung lipids. Nonetheless, technological advancements in the near future could make it possible to observe these factors using non- invasive techniques.

A final option for verifying our results in humans is the use of autopsy samples. As a predominantly neurological disorder, brain tissue is collected from deceased RTT patients; other tissues are not preserved and, to our knowledge, there have been no assessments on RTT patient post-mortem tissues other than brain. We hope that sharing this work will encourage more researchers, as well as the parents of RTT patients, to consider including the lung and other non-CNS tissues during organ donation. As it is currently not practical to assess pulmonary features in RTT patients, future studies in Mecp2-mutant mice will be crucial for understanding the Mecp2-deficient lung.

5.2.2 How does SREBP1 affect metabolic gene transcription in Mecp2/Y AE2 cells?

An unexpected finding of our studies was that, despite increased triglycerides in the lungs of Mecp2/Y mice, expression of fatty acid biosynthesis genes and other lipogenic enzymes was decreased. Additionally, the expression of Srebf1 was decreased in Mecp2/Y AE2s at P18 and its protein expression was decreased in whole lung lysate at P21. SREBP1 and SREBP2 are self- regulating lipid-sensing transcription factors that, upon detection of low lipids, translocate to the nucleus and activate the transcription of triglyceride or cholesterol biosynthesis genes,

201 respectively (350). SREBPs bind to genes that contain conserved sterol response elements (SREs) in their promoter region.

Our studies suggest that aberrant cholesterol and triglyceride accumulation in Mecp2/Y lungs is mediated by reduced NCOR1/2-mediated transcriptional repression at Hmgcs1 and Acot1 in the absence of Mecp2. Due to its lipid-sensing abilities, the decreased expression of SREBP1 at such early timepoints in Mecp2/Y lungs suggests that lipids are already elevated prior to P18-P21. These results are consistent with the early appearance of respiratory symptoms in male and female Mecp2-mutant mice. Intriguingly, many of the genes with decreased expression in P18 Mecp2/Y AE2 cells are targets of SREBP1. Notably, Acly, Fasn, Scd1 and Scd2, which are SREBP1-activated enzymes in fatty acid synthesis, have low expression in Mecp2/Y AE2 cells, suggesting SREBP1 may downregulate their expression upon detecting high levels of intracellular triglycerides (Figure 5.1). More interesting is the decreased expression of SREBP targets Pcyt1a, Lpcat1 and Abca3, which facilitate the production and packaging of PCs. Low expression of these enzymes likely reduces the pool of PCs available for pulmonary surfactant synthesis; consistently, PCs are reduced in Mecp2-mutant BAL fluid. SREBP could therefore represent a link between Mecp2-deficiency induced lung lipid accumulation and decreased PCs in the pulmonary surfactant of Mecp2-mutant mice. In this manner, an early increase in lung triglycerides and/or cholesterol would decrease SREBP expression, in turn decreasing PC concentrations, leading to further respiratory complications.

To test this hypothesis, we could perform earlier transcriptomic studies in +/Y and Mecp2/Y AE2 cells. In mice, surfactant production begins at E17.5, just before birth (273). Thus, transcriptomic studies within the first week of life in Mecp2/Y AE2s may provide a better picture of the pathogenic processes leading to lung lipid accumulation; I expect that at this early time point, the expression of lipogenic enzymes would be increased as feedback mechanisms have not yet begun to control aberrant lipid production. Subsequently, I would perform ChIP-sequencing in the P21 Mecp2/Y lung using an anti-SREBP1 antibody. This experiment would test whether SREBP binding to its target genes is reduced in Mecp2/Y lungs, as expected, especially at the promoters of PC synthesis enzymes. Such information would be informative for RTT patients as patients with increased lung lipids could be at risk for surfactant deficiency and may benefit from surfactant replacement therapy. In addition, it would provide insight on REBP1’s role as a transcriptional regulator of lung lipid metabolism in AE2 cells.

202

Figure 5.1: SREBPs are lipid-sensing transcription factors.

SREBP is retained in the endoplasmic reticulum (ER) membrane through a tight association with SCAP. In lipid-depleted cells (left), SCAP escorts SREBP to the Golgi apparatus. In the Golgi, proteolytic cleavage releases SREBP from SCAP. Mature SREBP migrates to the nucleus, binds to sterol response elements (SREs), and activates the transcription of genes involved in cholesterol and fatty acid synthesis. In lipid-abundant cells (right), such as in Mecp2/Y alveolar epithelial 2 (AE2) cells, SCAP-SREBP does not leave the ER membrane. This leads to a decreased expression of SRE-containing genes.

203 5.2.3 Do circadian-expressed nuclear receptors have a role in RTT pathology?

An unexpected finding that warrants further research is the altered expression of core circadian rhythm genes in Mecp2/Y lungs. Circadian rhythm is governed by the body’s central pacemaker, located in the suprachiasmatic nucleus (SCN) of the basal hypothalamus, though all tissues have their own pacemakers as well (388). The nuclear receptors NR1D1 and NR1D2 (also called REV- ERBα and REV-ERBβ, respectively) are well-established regulators of circadian rhythm. NR1D1 is broadly expressed across tissues while NR1D2 has higher expression in the brain, thyroid gland and uterus. NR1D1/2 is a potent transcriptional repressor that binds to DNA in a sequence- specific manner. Retinoic acid receptor-related orphan receptors (RORs) bind to the same sites, called ROR elements (ROREs), as transcriptional activators; thus, NR1D1/2 and RORs compete for binding at ROREs.

Intriguingly, NR1D1 recruits NCOR1 and HDAC3 to repress DNA in a deacetylation-dependent manner. Much of NR1D1’s function has been studied in the liver where NR1D1-bound genes are highly enriched for functions related to lipid metabolism, and conversely, Nr1d1-null mice develop fatty liver disease (468). Genome-wide analysis in the mouse liver revealed remarkable overlap in the NR1D1 and HDAC3 cistromes, despite levels of HDAC3 remaining consistent throughout the day, suggesting a strong functional connection between the two proteins (468). Accordingly, liver-specific deletion of HDAC3 also results in fatty liver (469).

NCOR1/2 are prototype nuclear receptor corepressors, though they bind to other transcription factors as well (470). NCOR1/2 act as scaffold proteins, most often recruiting HDAC3 due to their stable association, but can recruit other deacetylases in context-specific manners such as HDAC4, HDAC5 and HDAC7 (471,472). NCOR1/2 have many molecular partners that direct its localization to DNA, including retinoic acid receptors (RARs), thyroid receptors (TRs), peroxisome proliferator-activated receptors (PPARs), and liver X receptors (LXRs) (100). Importantly, MECP2 binds to methylated CpGs and tethers the HDAC3-containing NCOR1/2 corepressor complex to DNA, suppressing transcription of its target genes.

While NCOR1/2 has many interacting partners, there is a possibility that NR1D1 and MECP2 bind NCOR1 in the same complex in certain contexts. There are a few pieces of evidence to suggest this may be the case. First, Mecp2 is highly expressed in the SCN, the region of the brain that is responsible for circadian rhythm regulation (389). Second, NR1D1 peaks at ZT 8-9 in the mouse brain (473), overlapping with MECP2’s peak expression (390). Finally, liver-specific deletion of Mecp2 phenocopies deletions of Nr1d1 and Hdac3, increasing liver lipogenic gene transcription and causing fatty liver disease (164). However, mass spectrometry analyses of MECP2 interactors in the brain have not revealed an association with NR1D1, despite identifying other

204 NCOR1/2 complex members (89), though it could bind indirectly and/or at distinct circadian times. Additionally, if within the same complex, it is unclear whether NR1D1 or MECP2 would dictate complex binding to ROREs or methylated DNA, respectively.

To test this hypothesis, I would begin analyses with density gradient ultracentrifugation experiments to isolate all endogenous protein complexes in +/Y and Mecp2/Y tissue. In this method, protein lysates are loaded onto a sucrose or glycerol gradient and centrifuged for up to 20 hours. The gradient is then fractionated with each subsequent fraction containing protein complexes of increasing molecular weight. The fractions are then purified by SDS-PAGE and the presence of proteins within each fraction detected by Western blot. Through probing for different proteins (ex. NCOR1, MECP2, NR1D1, HDAC3), this assay will allow us to determine whether MECP2 and NR1D1 co-fractionate or are involved in different complexes. If the two proteins are contained in the same fractions, a co-immunoprecipitation (Co-IP) experiment in the fraction using anti-NR1D1 and anti-MECP2 antibodies in wildtype and Mecp2/Y tissues, will allow us to test for a reciprocal interaction. An anti-HDAC3 antibody would represent an ideal positive control as I would expect both NR1D1 and MECP2 to pull down with HDAC3. Importantly, density gradient experiments allow detection of transient and labile interactions that could not be captured by Co- IP alone and would show whether different NCOR1-containing complexes form in Mecp2/Y tissues. The mouse liver would be an ideal place to start for these experiments given its high expression of NR1D1 and the homogenous nature, though the interaction would also be tested in the lung and SCN, as protein interactions can vary dramatically across tissues.

Next, ChIP-sequencing could be employed to compare the cistrome of NR1D1 and MECP2 in wild type and Mecp2/Y tissues, using anti-NR1D1 and TBL1XR1 antibodies, respectively. Again, the liver would be an ideal place to begin as liver ChIP-sequencing data of NR1D1, HDAC3, and NCOR1 have been published (468). However, the lung would be particularly interesting to study as Nr1d1 expression was increased over 5-fold in Mecp2/Y lungs at ZT0. Thus, ChIP-sequencing in Mecp2/Y lungs would be particularly informative.

Regardless of whether NR1D1 and MECP2 are in the same complex or repress transcription independently, the results of these experiments will have important implications. If bound in the same complex, MECP2 could be considered a regulator of circadian rhythm. Loss of MECP2 within the NCOR1/NR1D1/HDAC3 complex would presumably alter its binding profile, making it more or less active at promoters of its target genes. In the SCN, this could cause sleep disturbances, which are common in both RTT patients and Mecp2-mutant mouse models (33,391), and alter the coordinated timing of many physiological processes including hormone production, glucose metabolism, and lipid metabolism.

205 Perhaps NR1D1 and MECP2 participate in independent NCOR1-dependent complexes. One model, termed the ‘sponge hypothesis’, posits that when MECP2 is mutated or absent, its decreased availability for NCOR1/2 recruitment leads to increased binding of NCOR1/2 to its other partners, increasing transcriptional repression at non-MECP2 target loci (Figure 5.2). Consistent with this, NCOR1/2 and HDAC3 concentrations do not change when MECP2 is absent. Both Nr1d1 and Nr1d2 are increased at the transcriptional level in Mecp2/Y lungs at ZT2-3 and RT- qPCR analyses indicated that Nr1d1 was increased by over 5-fold in Mecp2/Y lungs at ZT0, though there was no change at ZT12, indicating it is still rhythmically expressed, albeit its peak expression window is shifted and amplified. ChIP-sequencing of NR1D1-bound DNA in the liver revealed its repression of numerous lipid biosynthesis genes (468); a number of its targets in the liver were decreased in the lung, including the triglyceride synthesis genes, Acly, Fasn, and Scd1, and the PC synthesis gene choline kinase alpha (Chka). The former three genes are also targets of SREBP-mediated activation, as discussed earlier. NR1D1 also represses the expression of PC synthesis genes choline kinase alpha (Chkb), phosphate cytidylyltransferase 2 (Pcyt2), and lysophosphatidylcholine acyltransferase 3 (Lpcat3) in the liver (468); it’s increased expression in the lung at ZT0 suggests increased repression of these crucial PC synthesis genes. If the sponge hypothesis is true, the density gradient experiment proposed above would indicate if ratios of NCOR1-containing complexes are altered in Mecp2/Y tissues. Additionally, ChIP-sequencing experiments would allow us to determine if there is increased binding of NR1D1 to its target loci in Mecp2-deficient cells, and if so, how lipid metabolism and other processes are affected.

Regardless, shifts in NR1D1 expression in Mecp2/Y mice are important to study. Although I have only described altered Nr1d1 expression in the lung, circadian timing is governed by the thalamus’ SCN; thus, circadian rhythm genes are likely altered across many peripheral tissues in Mecp2/Y mice. In addition to regulating lipid metabolism, NR1D1 has important functions in the brain, as Nr1d1-null mice exhibit suppressed axon extension and dendritic arbor formation in cortical neurons (474), a phenotype also observed in Mecp2/Y mice. Accordingly, these mice are hyperactive, aggressive, have increased anxiety, and show depression-associated behaviors (475,476). NR1D1 is also highly expressed in macrophages where it regulates toll-like receptor 4 (TLR4) and the innate immune response, as well as glial cell activation and neuroinflammation (477,478). Loss of Nr1d1 also impairs mitochondrial function through decreased repression of its target autophagy and mitophagy genes (479). Thus, the study of altered NR1D1 expression in the brain and other tissues of Mecp2/Y mice may point to additional pathways that warrant investigation, such as neuroinflammation, mitochondrial function, and altered autophagy.

206

Figure 5.2: The ‘sponge hypothesis’ of NCOR1/2 for NR1D1.

NCOR1 acts as a scaffold protein for many nuclear receptors and transcription factors. In wild type cells, NCOR1 is recruited to DNA by MECP2, NR1D1, and other proteins in unknown stoichiometric ratios. This leads to the repression of MECP2-directed target genes and NR1D1- directed circadian-regulated target genes. In Mecp2-deficient cells, MECP2 is no longer present to recruit NCOR1 to DNA. The ‘sponge hypothesis’ suggests NCOR1 then becomes more available for binding with other NCOR1-containing complexes, such as that of NR1D1. This would lead to increased repression of NR1D1-directed transcriptional targets.

207 5.2.4 Are mitochondria impacted by Mecp2 deficiency?

Before MECP2 was identified as the causative gene in RTT, the disorder was thought to be caused by mutations in a mitochondrial gene (20). Abnormal mitochondrial structures, small muscle mass, and unsatisfactory weight gain despite good nutrition in RTT patients supported this hypothesis. Accordingly, many features of RTT overlap with mitochondrial diseases, such as early symptomatic onset, developmental delay, motor regression, and seizures (359). More recent studies revisiting this idea have found abnormally swollen or elongated mitochondria with fewer than normal cristae, decreased brain ATP levels, compromised mitochondrial function, and increased oxidative damage in patients and Mecp2-mutant mice (19,357,372–374).

Here, I found a strikingly decreased expression of mitochondrial-encoded electron transport chain (ETC) components in Mecp2/Y AE2 cells, with 11 of the 13 ETC genes being significantly underexpressed. Impaired mitochondria have functional consequences in any cell type, but highly metabolic cells are more susceptible to resulting damage. Dysfunctional mitochondria in the brain can cause ataxia, dementia, developmental regression and psychiatric disorders, while mitochondrial impairments in peripheral tissues can cause organ failure (480). Respiratory failure is a major concern of mitochondrial diseases, including Leigh syndrome, in which 25% of cases are caused by mutations in mitochondrial DNA (481).

To define the effects of decreased mtDNA expression in Mecp2/Y AE2 cells, I propose assessment of mitochondrial function. Live flow cytometry-isolated +/Y and Mecp2/Y AE2 cells would be used in the Seahorse assay, a novel method used to assess oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) of live cells, providing information on mitochondrial respiration and glycolysis. Briefly, OCR is measured at baseline by probes in the Seahorse Analyzer (Agilent). Then, oligomycin, an ETC complex V inhibitor, is added, and the resulting OCR is used to calculate ATP-linked respiration capacity. Next, carbonyl cyanide-p- trifluoromethoxy-phenyl-hydrazon (FCCP) is added, which collapses the inner mitochondrial membrane gradient, pushing the ETC to its maximal rate, allowing for calculation of maximal respiratory capacity. Finally, antimycin A and rotenone, which block complex III and I, respectively, are added to shut down ETC function, indicating the mitochondrial reserve capacity.

AE2 cells have abundant mitochondria as an energy source for the production and secretion of surfactant. Testing whether mitochondrial function is impaired in Mecp2/Y AE2 cells will be informative for many reasons. First, compromised mitochondria could accelerate lung lipid accumulation, as mitochondrial beta-oxidation would be impaired, leading to a buildup of fatty acids; in support of this, Acot1 and Acot2, genes which divert fatty acids away from beta-oxidation are increased in Mecp2/Y lungs. In this manner, ACOT1 serves to protect cells from reactive

208 oxygen species (ROS) which build up when mitochondria are overburdened (482). Second, ROS are a common feature of pulmonary diseases and induce emphysema, bronchitis and COPD (483), and altered mitochondrial function in AE2 cells can lead to increased DNA damage and AE2 cell death (484). Thus, impaired mitochondrial function, which leads to increased oxidative stress, could be a driving force in the late-stage tissue degeneration as seen in Mecp2-mutant lungs. Finally, impaired pulmonary mitochondria could contribute to the high incidence of respiratory failure in RTT patients, and mitochondria-targeted treatments could be employed to mitigate this risk.

5.2.5 Is the Mecp2-deficient lung more susceptible to insults?

In our Mecp2/Y model, I found a 3.2-fold decrease in lung elastin (Eln) mRNA expression, with a coordinate decrease in ELN protein. Previous studies suggest a threshold level of Eln expression is required for normal lung development: Eln-null mice die within 48 hours of birth whereas mice with heterozygous mutations in Eln have normal lungs. Further, mice with 30% of normal Eln expression have large airspaces and hyperextendable lungs. Intriguingly, while 50% of normal Eln expression does not cause overt respiratory symptoms in mice or humans, exposure to cigarette smoke enlarges lung airspaces by 1.8x that of wild type mice, suggesting reduced Eln expression predisposes the lung to irreversible injury. In addition to reduced Eln expression, Mecp2/Y mice had decreased expression of numerous collagens, glycoproteins and proteoglycans, all essential components of the lung ECM.

Both pneumonia and aspiration, two leading causes of death in RTT patients (31), are major causes of lung injury. The already structurally-compromised Mecp2-deficient lung could respond to these stresses differently compared to healthy individuals. To study how the Mecp2-deficient lung responds to insults, I would induce lung injury in our mouse model. A number of methods are used to study lung injury, including intravenous injection of lipopolysaccharide (LPS) (485), intrapulmonary administration of bacteria (486), and acid aspiration (487). I expect that, given the lung’s compromised strength and architecture, Mecp2/Y mice will have a stronger response to lung injury than their wild type counterparts. This may lead to a heightened inflammatory response with subsequent tissue degradation. Studying the response to lung injury in Mecp2-deficient mice could inform clinical recommendations for human patients. For example, it may be imperative that RTT patients avoid environmental exposure to chemical irritants such as cigarette smoke and asbestos. Additionally, clinicians and parents could monitor for first signs of respiratory infection to ensure timely treatment.

209 5.2.6 How is brain lipid metabolism altered in Mecp2/Y mice?

My studies on lipid metabolism in Mecp2-mutant lungs can inform studies in the brain. Neurons have a high need for cholesterol as they require it to build the large cell membranes of their axons and dendrites and to perform pre- and post-synaptic functions. A prevalent hypothesis is that while newly synthesized neurons produce cholesterol cell-autonomously, they outsource this process to astrocytes after reaching maturity (445). The mechanisms underlying cholesterol shuttling in the CNS are not fully understood, though astrocytes are thought to excrete cholesterol in APOE-containing lipoproteins mediated by cholesterol transport genes Abca1, Abcg1 and Abcg4, and is taken up by LDLR in neurons. When neurons require cholesterol turnover, the neuron-specific enzyme CYP46A1 converts cholesterol to 24S-OHC, which can be excreted across the blood-brain barrier.

Cholesterol biosynthesis enzymes are increased in the brain of P21 Mecp2/Y mice, despite decreased expression of Srebf1 and Srebf2. I showed that the cholesterol biosynthesis enzyme, Hmgcs1, is a target of MECP2-directed transcriptional repression by the NCOR1/2 complex. Hmgcs1 expression is increased across all brain regions of Mecp2/Y mice at P21 and could be repressed by MECP2 in the brain as well. Thus, in the brains of Mecp2-mutant mice, adult neurons may continue to produce cholesterol due to decreased MECP2-directed repression at cholesterol biosynthesis loci (Figure 5.3). In support of this, 24S-OHC is increased in symptomatic Mecp2/Y mice, indicating a heightened need for cholesterol turnover (488). 24S-OHC is a potent negative regulator of cholesterol biosynthesis and could shut down cholesterol biosynthesis in symptomatic mice (167). In support of this, absolute cholesterol biosynthesis rates are reduced in Mecp2/Y mice from P21 onward (450). Additionally, the expression of the transcription factors Nr1h2 and Nr1h3, which encode LXRβ and LXRα, respectively, and their cholesterol transport targets Abcg1, Abcg4, and Apoe are also decreased.

To further study cholesterol metabolism in the Mecp2/Y brain, we propose a number of studies. We would first assess cholesterol biosynthesis on the cellular level. Squalene epoxidase is the first enzyme in committed cholesterol biosynthesis, making it a good in situ reporter. We plan to use the Tg(Sqle-EGFP)IM104Gensat mouse line (Sqle-EGFP), which has an EGFP-polyA cassette inserted in a Sqle-containing BAC for visualization of Sqle expression. Fluorescent imaging of brain tissues from Sqle-EGFP mice will allow us to determine which cells produce cholesterol in the brain and at what time points. For example, we would expect both neurons and astrocytes to exhibit fluorescence at early time points, but only astrocytes in adult brains. Further, by breeding Sqle-EGFP mice to Mecp2-mutant mice, we could visualize changes in brain cholesterol production in the absence of Mecp2. This would allow us to determine whether

210 neurons continue to produce cholesterol when Mecp2 is absent. It is likely that cholesterol biosynthesis is more affected by Mecp2 loss in certain neuronal populations compared to others. Quantitative assessments of cholesterol-producing cells in different areas of the brain can be achieved using flow cytometry to quantify GFP-positive astrocytes and neurons, using the appropriate cell surface markers. Additionally, lipid stains such as BODIPY 493/503, which label cholesterol and neutral lipids, respectively, could be utilized to study cholesterol localization in the Mecp2/Y brain. This would allow us to determine which cells in the Mecp2/Y brain accumulate cholesterol and at what times, as well as whether cholesterol is successfully cleared at later disease stages.

I also propose single-cell RNA sequencing at two time-points: prior to and after the neuron-to- astrocyte switch in cholesterol biosynthesis. Single-cell sequencing of brain tissue would reveal how cholesterol metabolism is altered in Mecp2-mutant neurons and astrocytes at the transcriptional level. Based on our hypothesis, we expect Mecp2-mutant neurons to have an increased expression of cholesterol biosynthesis genes. How astrocytes respond to this will be interesting: it is possible that MECP2 also regulates cholesterol biosynthesis in astrocytes and that they will continue to produce and transport cholesterol, overburdening neurons and increasing brain cholesterol turnover rates. It is also possible that astrocytes will sense heightened brain cholesterol and in turn, decrease their expression of cholesterol biosynthesis enzymes earlier than neurons. This may also lead to the decreased expression of LXR-regulated cholesterol transport enzymes to limit cholesterol shuttling to neurons. The results of this study will inform future work. ChIP-sequencing of flow-cytometry sorted neurons and astrocytes from +/Y and Mecp2/Y mice using an anti-TBL1XR1 antibody will determine the cholesterol metabolism genes that are directly targeted by MECP2 with the NCOR1/2 complex. This will allow us to assess the differences in how cholesterol metabolism is epigenetically regulated by MECP2, and concurrently in a state of Mecp2-defiency, in neurons and astrocytes.

In support of impaired astrocyte function in RTT, re-expression of Mecp2 in astrocytes improves locomotion, anxiety, and breathing, and extends lifespan in mice (163). Additionally, wild type neurons co-cultured with Mecp2-deficient astrocytes have abnormal dendritic morphology, but Mecp2-deficient neurons have normal dendrites when cultured with wild type astrocytes (74,163). The mechanism underlying these findings is not yet understood, though it is hypothesized that astrocytes’ roles in synapse formation and plasticity could contribute. To study whether cholesterol metabolism is involved in this phenomenon, and to add to our transcriptional studies, we would prepare primary cortical cultures of mouse neurons and astrocytes. We could then compare cholesterol biosynthesis rates and cholesterol concentrations excreted into the culture by wild type or Mecp2/Y astrocytes. Co-culture experiments with wild type or Mecp2/Y neurons

211 would help us understand how neurons and astrocytes interact to fulfill their cholesterol needs as well as if cholesterol metabolism is disrupted on the neuronal or astrocytic end. Further, we could identify if astrocyte or neuron-specific treatment with metabolism modulators could rescue neuronal deficits.

212

Figure 5.3: Hypothesis on neuron and astrocyte cholesterol metabolism in Mecp2-deficient brain cells. In the wild type brain (top), after neuronal maturity, astrocytes (blue) produce cholesterol to fulfill the neuron’s needs. In astrocytes, MECP2 represses cholesterol synthesis while REBPs induce it, maintaining a balance. LXRs promote cholesterol (yellow circles) transport, allowing it to be shuttled from astrocytes to neurons. In contrast, neurons (red) do not produce cholesterol. When neurons require cholesterol turnover, they convert cholesterol to 24S-OHC for excretion across the blood brain barrier, signaling to the astrocyte that new cholesterol is needed. In Mecp2- deficient astrocytes (bottom), loss of MECP2 constitutively increases (red font) cholesterol biosynthesis. This leads to decreased expression of Srebf mRNA and likely, decreased SREBP binding (blue font). Meanwhile, LXR-directed cholesterol transport is decreased. In Mecp2- deficient neurons, cholesterol biosynthesis genes are active, increasing intracellular cholesterol concentrations. This leads to increased conversion to 24S-OHC, signaling cholesterol.

213 5.2.7 Preclinical treatment of RTT symptoms

Here, we described the positive effects of a few metabolism-targeted therapeutics on symptoms in Mecp2-mutant mice. Cholesterol-lowering fluvastatin had robust effects in male and female Mecp2-mutant mice, improving overall health, motor coordination, and respiratory symptoms. Treatment with the LXR-agonist LXR-623 also improved health and motor coordination and breathing frequency trended downward but was not significant. These results implicate misregulated cholesterol metabolism in RTT pathogenesis. Interestingly, fluvastatin, but not LXR- 623, improved the response of Mecp2-mutant mice to methacholine, an agent that causes constriction of the airways. This may be due to the pleiotropic effects of statins; for example, excess lung lipids could induce an inflammatory response which statins are able to moderate. To test this, I propose to treat mice with a non-steroidal anti-inflammatory drug (NSAID). NSAIDs are often prescribed in individuals with lung disorders to slow the progression of lung damage.

Despite the combination of fluvastatin and LXR-623 providing no beneficial results, we are still proponents of combinatorial therapy as a best approach to treat RTT. For example, combined treatment with fluvastatin and an NSAID, to reduce lung lipids and treat lung inflammation, respectively, will likely offer a larger therapeutic benefit than fluvastatin alone. As discussed above, mitochondrial impairments have been evidenced in RTT and treatments aimed at reducing oxidative stress have shown mild benefits in Mecp2-mutant models (26,223). If heightened beta- oxidation due to increased cellular triglycerides contribute to impaired mitochondrial function, treating lipid metabolism and oxidative stress concurrently could alleviate the overburdened mitochondria faster. Additionally, as circadian rhythm may be disrupted in Mecp2-mutant mice, the adequate timing of treatment may be essential and something to consider to future studies.

214 5.3 Future of RTT

Progress in understanding the mechanistic basis for RTT pathology will continue to inform treatment strategies for RTT and other neurological diseases. Here, we showed that aberrant lung lipid metabolism in Mecp2/Y mice is due to loss of NCOR1/2-mediated transcriptional repression, adding to accumulating evidence on the importance of this complex in RTT pathology (89,102,108). Mutations in other components of the complex, TBLX1 and TBLX1R1, have been associated with autism, intellectual disability, Pierpont syndrome, a disorder characterized by developmental delay and abnormal fat distribution in the distal limbs, and West syndrome, a disorder with RTT-like features (489–492). Notably, six of these mutations in TBLX1R1 mapped to the WD40 domain of the protein and disrupted MECP2-binding (102). Therefore, the transcriptional function of the NCOR1/2 complex could represent a shared mechanism for autism spectrum disorders and other neurological conditions.

Recent advances have made treatments that directly target MECP2 a feasible option for RTT in the near future. Restoration of MECP2 gene or protein function could be achieved through read- through of nonsense mutations, reactivation of the silenced X chromosome, or gene therapy. A major concern with these approaches is dosage: as an epigenetic regulator, both too little and too much MECP2 causes adverse symptoms (126). As little as 1.6x normal levels of MECP2 cause behavioral symptoms in mice (240) and slight deviations in MECP2 levels are linked to autism and other psychiatric conditions in humans (125). Systemic injection of an adeno-associated virus (AAV) 9-associated MECP2 transgene in Mecp2/Y mice had a low transduction efficiency in the brain of 2-4%, but accumulated in the spleen and liver, causing liver damage (257). However, injection of a minimal MECP2 protein intracranially was better tolerated and improved neurobehavioral symptoms and lifespan in Mecp2/Y mice. Gene therapy for RTT remains at its infancy as strategies for optimization are still underway, including those to circumvent MECP2 overexpression, effectively scale doses from mice to humans, and infect the large and complex human brain.

While MECP2 has been largely studied in the CNS, some clinically significant aspects of RTT arise independently of MECP2 deficiency in the brain. Both RTT patients and Mecp2 mutant mice present with metabolic syndrome (15,164), oxidative stress (23,25), cardiac defects (160,493), decreased bone density (29,494), and urological dysfunction (32,495). Here, we show that Mecp2 deficiency impairs lung surfactant production and ECM modeling in mice. As gene therapy becomes a more realistic therapeutic approach, an understanding of peripheral deficiency of MECP2 is necessary as some symptoms are likely to persist following targeted genetic treatment to the brain. In this respect, precision medicine for RTT is warranted (Figure 5.4). Patient age,

215 mutation status and level of XCI skewing should be taken into account. Biomarkers are needed to ensure adequate treatment of non-CNS based symptoms, regardless of treatment intervention. For example, high serum cholesterol can be used as an indicator for statin treatment. We anticipate that a combination of mild gene therapy with one or more pharmacological treatments will likely provide the best approach for the treatment of RTT. In this respect, the future of RTT treatment appears promising. Every year, new drug treatments are tested in Mecp2-mutant models that rescue different aspects of RTT phenotypes, many of which move to clinical trials. Continued study of MECP2 will reveal novel targets for the development of treatments and a better understanding of the pathological processes underlying RTT.

216

Figure 5.4: Precision medicine for treating RTT. Individual RTT patients will likely benefit from different combinations of treatment. When treating patients for RTT, their mutation status and unique genetic background should be taken into consideration as they will likely affect symptom presentation. It is likely some patients will not be ideal candidates for gene therapy. Biomarkers, such as serum lipids, can be used to determine which patients will benefit from pharmacological intervention, such as statins. Symptom management of seizures, scoliosis, and other less common features of RTT should resume. Altogether, RTT patients should receive individualized treatment to maximize their symptom improvement.

217 Bibliography

1. Rett A. On a unusual brain atrophy syndrome in hyperammonemia in childhood. Wiener Medizinische Wochenschrift. 1966;1946(116):723–6. 2. Hagberg B, Aicardi J, Dias K, Ramos O. A progressive syndrome of autism, dementia, ataxia, and loss of purposeful hand use in girls: Rett’s syndrome: report of 35 cases. Ann Neurol. 1983 Oct;14(4):471– 9. 3. Burd L, Randall T, Martsolf JT, Kerbeshian J. Rett syndrome symptomatology of institutionalized adults with mental retardation: Comparison of males and females. Am J Ment Retard. 1991;95(5):596–601. 4. Neul JL, Kaufmann WE, Glaze DG, Christodoulou J, Clarke AJ, Bahi-Buisson N, et al. Rett syndrome: Revised diagnostic criteria and nomenclature. Ann Neurol. 2010;68(6):944–50. 5. Hagberg B, Witt-Engerström I. Rett syndrome: a suggested staging system for describing impairment profile with increasing age towards adolescence. Am J Med Genet. 1986;1:47–59. 6. Jellinger K, Seitelberger F, Opitz JM, Reynolds JF. Neuropathology of Rett syndrome. Am J Med Genet. 1986;25(S1):259–88. 7. Jellinger K, Armstrong D, Zoghbi HY, Percy AK. Neuropathology of Rett syndrome. Acta Neuropathol. 1988;76(2):142–58. 8. Armstrong DD. The neuropathology of the Rett syndrome. Brain Dev. 1992 May;14:S89-98. 9. Kaufmann WE, Moser HW. Dendritic anomalies in disorders associated with mental retardation. Cereb cortex. 2000 Oct;10(10):981–91. 10. Taneja P, Ogier M, Brooks-Harris G, Schmid DA, Katz DM, Nelson SB. Pathophysiology of locus ceruleus neurons in a mouse model of Rett syndrome. J Neurosci. 2009;29(39):12187–95. 11. Zoghbi HY, Milstien S, Butler IJ, Smith EOB, Kaufman S, Glaze DG, et al. Cerebrospinal fluid biogenic amines and biopterin in rett syndrome. Ann Neurol. 1989;25(1):56–60. 12. Matsuishi T, Urabe F, Percy AK, Komori H, Yamashita Y, Schultz RS, et al. Abnormal carbohydrate metabolism in cerebrospinal fluid in Rett syndrome. J Child Neurol. 1994 Jan;9(1):26–30. 13. Hanefeld F, Christen HJ, Holzbach U, Kruse B, Frahm J, Hanicke W. Cerebral proton magnetic resonance spectroscopy in Rett syndrome. Neuropediatrics. 1995;26(2):126–7. 14. Horska A, Naidu S, Herskovits EH, Wang PY, Kaufmann WE, Barker PB. Quantitative 1H MR spectroscopic imaging in early Rett syndrome. Neurology. 2000;54(3):715–22. 15. Justice MJ, Buchovecky CM, Kyle SM, Djukic A. A role for metabolism in Rett syndrome pathogenesis: New clinical findings and potential treatment targets. Rare Dis. 2013;1:e27265. 16. Segatto M, Trapani L, Di Tunno I, Sticozzi C, Valacchi G, Hayek J, et al. Cholesterol metabolism is altered in Rett syndrome: a study on plasma and primary cultured fibroblasts derived from patients. PLoS One. 2014 Aug;9(8):e104834. 17. Acampa M, Guideri F, Hayek J, Blardi P, De Lalla A, Zappella M, et al. Sympathetic overactivity and plasma leptin levels in Rett syndrome. Neurosci Lett. 2008;432(1):69–72. 18. Blardi P, De Lalla A, D’Ambrogio T, Vonella G, Ceccatelli L, Auteri A, et al. Long-term plasma levels of leptin and adiponectin in Rett syndrome. Clin Endocrinol (Oxf). 2009;70(5):706–9. 19. Eeg-Olofsson O, Al-Zuhair AG, Teebi AS, Al-Essa MM. Abnormal mitochondria in the Rett syndrome. Brain Dev. 1988 Jan;10(4):260–2. 20. Eeg-Olofsson O, Al-Zuhair AG, Teebi AS, Daoud AS, Zaki M, Besisso MS, et al. Rett syndrome: a mitochondrial disease? J Child Neurol. 1990 Jul;5(3):210–4. 21. Wakai S, Kameda K, Ishikawa Y, Miyamoto S, Nagaoka M, Okabe M, et al. Rett syndrome: findings suggesting axonopathy and mitochondrial abnormalities. Pediatr Neurol. 1990;6(5):339–43. 22. Cardaioli E, Dotti MT, Hayek G, Zappella M, Federico A. Studies on mitochondrial pathogenesis of Rett syndrome: ultrastructural data from skin and muscle biopsies and mutational analysis at mtDNA nucleotides 10463 and 2835. J Submicrosc Cytol Pathol. 1999 Apr;31(2):301–4. 23. Haas RH, Nasirian F, Hua X, Nakano K, Hennessy M. Oxidative metabolism in Rett syndrome: 2 - Biochemical and molecular studies. Neuropediatrics. 1995;26(2):95–9. 24. Sierra C, Vilaseca MA, Brandi N, Artuch R, Mira A, Nieto M, et al. Oxidative stress in Rett syndrome. In: Brain and Development. 2001. 25. De Felice C, Ciccoli L, Leoncini S, Signorini C, Rossi M, Vannuccini L, et al. Systemic oxidative stress in classic Rett syndrome. Free Radic Biol Med. 2009;47(4):440–8. 26. Leoncini S, De Felice C, Signorini C, Pecorelli A, Durand T, Valacchi G, et al. Oxidative stress in Rett syndrome: natural history, genotype, and variants. Redox Rep. 2011;16(4):145–53.

218 27. Motil KJ, Caeg E, Barrish JO, Geerts S, Lane JB, Percy AK, et al. Gastrointestinal and nutritional problems occur frequently throughout life in girls and women with rett syndrome. J Pediatr Gastroenterol Nutr. 2012 Sep;55(3):292–8. 28. Kumar A, Jaryal A, Gulati S, Chakrabarty B, Singh A, Deepak KK, et al. Cardiovascular Autonomic Dysfunction in Children and Adolescents With Rett Syndrome. Pediatr Neurol. 2017;70:61–6. 29. Shapiro JR, Bibat G, Hiremath G, Blue ME, Hundalani S, Yablonski T, et al. Bone mass in Rett syndrome: association with clinical parameters and MECP2 mutations. Pediatr Res. 2010;68(5):446–51. 30. Haas RH, Dixon SD, Sartoris DJ, Hannessy MJ. Osteopenia in Rett syndrome. J Pediatr. 1997;131(5):771–4. 31. Anderson A, Wong K, Jacoby P, Downs J, Leonard H. Twenty years of surveillance in Rett syndrome: what does this tell us? Orphanet J Rare Dis. 2014;9(1):87. 32. Ward CS, Huang TW, Herrera JA, Samaco RC, Pitcher MR, Herron A, et al. Loss of MeCP2 causes urological dysfunction and contributes to death by kidney failure in mouse models of rett syndrome. PLoS One. 2016;11(11). 33. Young D, Nagarajan L, de Klerk N, Jacoby P, Ellaway C, Leonard H. Sleep problems in Rett syndrome. Brain Dev. 2007;29(10):609–16. 34. Tarquinio DC, Hou W, Neul JL, Kaufmann WE, Glaze DG, Motil KJ, et al. The Changing Face of Survival in Rett Syndrome and MECP2-Related Disorders. Pediatr Neurol. 2015; 35. Laurvick CL, de Klerk N, Bower C, Christodoulou J, Ravine D, Ellaway C, et al. Rett syndrome in Australia: a review of the epidemiology. J Pediatr. 2006;148(3):347–52. 36. Amir RE, Van den Veyver IB, Wan M, Tran CQ, Francke U, Zoghbi HY. Rett syndrome is caused by mutations in X-linked MECP2, encoding methyl- CpG-binding protein 2. Nat Genet. 1999;23(2):185–8. 37. Bienvenu T, Carrié A, de Roux N, Vinet MC, Jonveaux P, Couvert P, et al. MECP2 mutations account for most cases of typical forms of Rett syndrome. Hum Mol Genet. 2000 May;9(9):1377–84. 38. Adler DA, Quaderi NA, Brown SDMM, Chapman VM, Moore J, Tate P, et al. The X-linked methylated DNA binding protein, Mecp2, is subject to X inactivation in the mouse. Mamm Genome. 1995;6(8):491–2. 39. D’Esposito M, Quaderi NA, Ciccodicola A, Bruni P, Esposito T, D’Urso M, et al. Isolation, physical mapping, and Northern analysis of the X-linked human gene encoding methyl CpG-binding protein, MECP2. Mamm Genome. 1996 Jul;7(7):533–5. 40. Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, Zahler AM, et al. The Human Genome Browser at UCSC. Genome Res. 2002;12(6):996–1006. 41. Coy JF, Sedlacek Z, Bächner D, Delius H, Poustka A. A complex pattern of evolutionary conservation and alternative polyadenylation within the long 3’ untranslated region of the methyl-CpG- binding protein 2 gene (MeCP2) suggests a regulatory role in gene expression. Hum Mol Genet. 1999;8(7):1253–62. 42. Pelka GJ, Watson CM, Christodoulou J, Tam PPL. Distinct expression profiles of Mecp2 transcripts with different lengths of 3′UTR in the brain and visceral organs during mouse development. Genomics. 2005;85(4):441–52. 43. Coy JF, Sedlacek Z, Bächner D, Delius H, Poustka A. A complex pattern of evolutionary conservation and alternative polyadenylation within the long 3"-untranslated region of the methyl- CpG-binding protein 2 gene (MeCP2) suggests a regulatory role in gene expression. Hum Mol Genet. 1999 Jul;8(7):1253–62. 44. Lewis JD, Meehan RR, Henzel WJ, Maurer-Fogy I, Jeppesen P, Klein F, et al. Purification, sequence, and cellular localization of a novel chromosomal protein that binds to Methylated DNA. Cell. 1992 Jun;69(6):905–14. 45. Mnatzakanian GN, Lohi H, Munteanu I, Alfred SE, Yamada T, MacLeod PJM, et al. A previously unidentified MECP2 open reading frame defines a new protein isoform relevant to Rett syndrome. Nat Genet. 2004; 46. Kriaucionis S, Bird A. The major form of MeCP2 has a novel N-terminus generated by alternative splicing. Nucleic Acids Res. 2004;32(5):1818–23. 47. Itoh M, Tahimic CGT, Ide S, Otsuki A, Sasaoka T, Noguchi S, et al. Methyl CpG-binding protein isoform MeCP2-e2 is dispensable for Rett syndrome phenotypes but essential for embryo viability and placenta development. J Biol Chem. 2012;287(17):13859–67. 48. Yasui DH, Gonzales ML, Aflatooni JO, Crary FK, Hu DJ, Gavino BJ, et al. Mice with an isoform- ablating Mecp2 exon 1 mutation recapitulate the neurologic deficits of Rett syndrome. Hum Mol Genet. 2014;23(9):2447–58.

219 49. Makałowski W, Boguski M . Evolutionary parameters of the transcribed mammalian genome: An analysis of 2,820 orthologous rodent and human sequences. Proc Natl Acad Sci U S A. 1998; 50. Nan X, Meehan RR, Bird A. Dissection of the methyl-CpG binding domain from the chromosomal protein MeCP2. Nucleic Acids Res. 1993 Oct;21(21):4886–92. 51. Shahbazian MDMDD, Antalffy B, Armstrong DL, Zoghbi HY. Insight into Rett syndrome: MeCP2 levels display tissue- and cell-specific differences and correlate with neuronal maturation. Hum Mol Genet. 2002 Jan;11(2):115–24. 52. Tillotson R, Bird A. The Molecular Basis of MeCP2 Function in the Brain. Journal of Molecular Biology. 2020. 53. Lyst MJ, Ekiert R, Guy J, Selfridge J, Koerner M V., Merusi C, et al. Affinity for DNA Contributes to NLS Independent Nuclear Localization of MeCP2. Cell Rep. 2018; 54. Tweedie S, Charlton J, Clark V, Bird A. Methylation of genomes and genes at the invertebrate- vertebrate boundary. Mol Cell Biol. 1997 Mar;17(3):1469–75. 55. Nan X, Campoy FJ, Bird A. MeCP2 is a transcriptional repressor with abundant binding sites in genomic chromatin. Cell. 1997 Feb;88(4):471–81. 56. Nan X, Tate P, Li E, Bird A. DNA methylation specifies chromosomal localization of MeCP2. Mol Cell Biol. 1996 Jan;16(1):414–21. 57. Ausió J, de Paz AM artínez, Esteller M. MeCP2: the long trip from a chromatin protein to neurological disorders. Vol. 20, Trends in molecular medicine. 2014. p. 487–98. 58. Mollica L, Bessa LM, Hanoulle X, Jensen MR, Blackledge M, Schneider R. Binding mechanisms of intrinsically disordered proteins: theory, simulation, and experiment. Front Mol Biosci. 2016;3. 59. Bellini E, Pavesi G, Barbiero I, Bergo A, Chandola C, Nawaz MS, et al. MeCP2 post-translational modifications: a mechanism to control its involvement in synaptic plasticity and homeostasis? Front Cell Neurosci. 2014;8. 60. Ebert DH, Gabel HW, Robinson ND, Kastan NR, Hu LS, Cohen S, et al. Activity-dependent phosphorylation of MeCP2 threonine 308 regulates interaction with NCoR. Nature. 2013;499(7458):341–5. 61. Schapira M, Tyers M, Torrent M, Arrowsmith CH. WD40 repeat domain proteins: A novel target class? Nature Reviews Drug Discovery. 2017. 62. Thambirajah AA, Eubanks JH, Ausió J. MeCP2 post-translational regulation through PEST domains: two novel hypotheses: potential relevance and implications for Rett syndrome. Bioessays. 2009 May;31(5):561–9. 63. Stenson PD, Ball E V, Mort M, Phillips AD, Shiel JA, Thomas NST, et al. Human Gene Mutation Database (HGMD): 2003 update. Hum Mutat. 2003 Jun;21(6):577–81. 64. Neul JL, Fang P, Barrish J, Lane J, Caeg EB, Smith EO, et al. Specific mutations in Methyl-CpG- Binding Protein 2 confer different severity in Rett syndrome. Neurology. 2008;70(16):1313–21. 65. Ishii T, Makita Y, Ogawa A, Amamiya S, Yamamoto M, Miyamoto A, et al. The role of different X- inactivation pattern on the variable clinical phenotype with Rett syndrome. In: Brain and Development. Elsevier; 2001. p. 161–4. 66. Hagberg BA, Skjeldal OH. Rett variants: A suggested model for inclusion criteria. Vol. 11, Pediatric Neurology. 1994. p. 5–11. 67. Cuddapah VA, Pillai RB, Shekar K V, Lane JB, Motil KJ, Skinner SA, et al. Methyl-CpG-binding protein 2 (MECP2) mutation type is associated with disease severity in Rett syndrome. J Med Genet. 2014;51(3):152–8. 68. Zhang Q, Zhao Y, Bao X, Luo J, Zhang X, Li J, et al. Familial cases and male cases with MECP2 mutations. Am J Med Genet Part B Neuropsychiatr Genet. 2017; 69. Reik W, Lewis A. Co-evolution of X-chromosome inactivation and imprinting in mammals. Nat Rev Genet. 2005 Apr;6(5):403–10. 70. Artuso R, Papa FT, Grillo E, Mucciolo M, Yasui DH, Dunaway KW, et al. Investigation of modifier genes within copy number variations in Rett syndrome. J Hum Genet. 2011;56(7):508–15. 71. Skene PJ, Illingworth RS, Webb S, Kerr ARW, James KD, Turner DJ, et al. Neuronal MeCP2 is expressed at near histone-octamer levels and globally alters the chromatin state. Mol Cell. 2010 Feb;37(4):457–68. 72. Marsh R, Gerber AJ, Peterson BS. Neuroimaging studies of normal brain development and their relevance for understanding childhood neuropsychiatric disorders. Journal of the American Academy of Child and Adolescent Psychiatry. 2008. 73. Jung BP, Jugloff DGM, Zhang G, Logan R, Brown S, Eubanks JH. The expression of methyl CpG binding factor MeCP2 correlates with cellular differentiation in the developing rat brain and in cultured cells. J Neurobiol. 2003 Apr;55(1):86–96.

220 74. Ballas N, Lioy DT, Grunseich C, Mandel G. Non–cell autonomous influence of MeCP2-deficient glia on neuronal dendritic morphology. Nat Neurosci. 2009;12(3):311–7. 75. Nguyen MVC, Felice CA, Du F, Covey M V., Robinson JK, Mandel G, et al. Oligodendrocyte lineage cells contribute unique features to Rett syndrome neuropathology. J Neurosci. 2013;33(48):18764–74. 76. Derecki NC, Cronk JC, Lu Z, Xu E, Abbott SBG, Guyenet PG, et al. Wild-type microglia arrest pathology in a mouse model of Rett syndrome. Nature. 2012; 77. Armstrong DD, Dunn K, Antalffy B. Decreased dendritic branching in frontal, motor and limbic cortex in Rett Syndrome compared with Trisomy 21. J Neuropathol Exp Neurol. 1998; 78. Kishi N, Macklis JD. MECP2 is progressively expressed in post-migratory neurons and is involved in neuronal maturation rather than cell fate decisions. Mol Cell Neurosci. 2004;27(3):306–21. 79. Palmer A, Qayumi J, Ronnett G. MeCP2 mutation causes distinguishable phases of acute and chronic defects in synaptogenesis and maintenance, respectively. Mol Cell Neurosci. 2008; 80. Dani VS, Chang Q, Maffei A, Turrigiano GG, Jaenisch R, Nelson SB. Reduced cortical activity due to a shift in the balance between excitation and inhibition in a mouse model of Rett Syndrome. Proc Natl Acad Sci U S A. 2005; 81. Chao H-TT, Zoghbi HYY, Rosenmund C. MeCP2 controls excitatory synaptic strength by regulating glutamatergic synapse number. Neuron. 2007 Oct;56(1):58–65. 82. Uhlén M, Fagerberg L, Hallström BM, Lindskog C, Oksvold P, Mardinoglu A, et al. Tissue-based map of the human proteome. Science (80- ). 2015; 83. Tudor M, Akbarian S, Chen RZ, Jaenisch R. Transcriptional profiling of a mouse model for Rett syndrome reveals subtle transcriptional changes in the brain. Proc Natl Acad Sci U S A. 2002 Nov;99(24):15536–41. 84. Chahrour M, Jung SYY, Shaw C, Zhou X, Wong STCCTC, Qin J, et al. MeCP2, a key contributor to neurological disease, activates and represses transcription. Science (80- ). 2008 May;320(5880):1224– 9. 85. Kinde B, Wu DY, Greenberg ME, Gabel HW. DNA methylation in the gene body influences MeCP2- mediated gene repression. Proc Natl Acad Sci U S A. 2016; 86. Lagger S, Connelly JC, Schweikert G, Webb S, Selfridge J, Ramsahoye BH, et al. MeCP2 recognizes cytosine methylated tri-nucleotide and di-nucleotide sequences to tune transcription in the mammalian brain. PLoS Genet. 2017;13(5). 87. Zemach A, McDaniel IE, Silva P, Zilberman D. Genome-wide evolutionary analysis of eukaryotic DNA methylation. Science (80- ). 2010; 88. Kokura K, Kaul SC, Wadhwa R, Nomura T, Khan MM, Shinagawa T, et al. The Ski protein family is required for MeCP2-mediated transcriptional repression. J Biol Chem. 2001 Sep;276(36):34115–21. 89. Lyst MJ, Ekiert R, Ebert DH, Merusi C, Nowak J, Selfridge J, et al. Rett syndrome mutations abolish the interaction of MeCP2 with the NCoR/SMRT co-repressor. Nat Neurosci. 2013;16(7):898–902. 90. Nan X, Ng H-HHH, Johnson CAA, Laherty CDD, Turner BMM, Eisenman RNN, et al. Transcriptional repression by the methyl-CpG-binding protein MeCP2 involves a histone deacetylase complex. Nature. 1998 May;393(6683):386–9. 91. Suzuki M, Yamada T, Kihara-Negishi F, Sakurai T, Oikawa T. Direct association between PU.1 and MeCP2 that recruits mSin3A-HDAC complex for PU.1-mediated transcriptional repression. Oncogene. 2003; 92. Mann J, Oakley F, Akiboye F, Elsharkawy A, Thorne AW, Mann DA. Regulation of myofibroblast transdifferentiation by DNA methylation and MeCP2: Implications for wound healing and fibrogenesis. Cell Death Differ. 2007; 93. Forlani G, Giarda E, Ala U, di Cunto F, Salani M, Tupler R, et al. The MeCP2/YY1 interaction regulates ANT1 expression at 4q35: Novel hints for Rett syndrome pathogenesis. Hum Mol Genet. 2010; 94. Szulwach KE, Li X, Smrt RD, Li Y, Luo Y, Lin L, et al. Cross talk between microRNA and epigenetic regulation in adult neurogenesis. J Cell Biol. 2010; 95. Hwang CK, Kim CS, Kim DK, Law PY, Wei LN, Loh HH. Up-regulation of the μ-opioid receptor gene is mediated through chromatin remodeling and transcriptional factors in differentiated neuronal cells. Mol Pharmacol. 2010; 96. Dhawan , Georgia , Tschen ing, Fan G, Bhushan A. Pancreatic β Cell Identity Is Maintained by DNA Methylation-Mediated Repression of Arx. Dev Cell. 2011; 97. Xue J, Wijeratne SSK, Zempleni J. Holocarboxylase synthetase synergizes with methyl CpG binding protein 2 and DNA methyltransferase 1 in the transcriptional repression of long-terminal repeats. Epigenetics. 2013;

221 98. Li J, Wang J, Wang J, Nawaz Z, Liu JM, Qin J, et al. Both corepressor proteins SMRT and N-CoR exist in large protein complexes containing HDAC3. EMBO J. 2000;19(16):4342–50. 99. Yoon H-G, Chan DW, Reynolds AB, Qin J, Wong J. N-CoR mediates DNA methylation-dependent repression through a methyl CpG binding protein Kaiso. Mol Cell. 2003 Sep;12(3):723–34. 100. Mottis A, Mouchiroud L, Auwerx J. Emerging roles of the corepressors NCoR1 and SMRT in homeostasis. Vol. 27, Genes and Development. 2013. p. 819–35. 101. Oberoi J, Fairall L, Watson PJ, Yang JC, Czimmerer Z, Kampmann T, et al. Structural basis for the assembly of the SMRT/NCoR core transcriptional repression machinery. Nat Struct Mol Biol. 2011; 102. Kruusvee V, Lyst MJ, Taylor C, Tarnauskaitė Ž, Bird AP, Cook AG. tructure of the MeCP2–TBLR1 complex reveals a molecular basis for Rett syndrome and related disorders. Proc Natl Acad Sci. 2017;114(16):E3243–50. 103. Lyst MJ, Bird A. Rett syndrome: a complex disorder with simple roots. Nat Rev Genet. 2015;16(5):261–75. 104. Jepsen K, Hermanson O, Onami TM, Gleiberman AS, Lunyak V, McEvilly RJ, et al. Combinatorial roles of the nuclear receptor corepressor in transcription and development. Cell. 2000; 105. Bhaskara S, Chyla BJ, Amann JM, Knutson SK, Cortez D, Sun Z-W, et al. Deletion of Histone Deacetylase 3 Reveals Critical Roles in S Phase Progression and DNA Damage Control. Mol Cell. 2008 Apr;30(1):61–72. 106. Guo C, Li Y, Gow CH, Wong M, Zha J, Yan C, et al. The optimal corepressor function of nuclear receptor corepressor (NCoR) for peroxisome proliferator-activated receptor γ requires g protein pathway suppressor 2. J Biol Chem. 2015; 107. Zaghlula M, Glaze DG, Enns GM, Potocki L, Schwabe AL, Suter B. Current clinical evidence does not support a link between TBL1XR1 and Rett syndrome: Description of one patient with Rett features and a novel mutation in TBL1XR1, and a review of TBL1XR1 phenotypes. American Journal of Medical Genetics, Part A. 2018. 108. Tillotson R, Selfridge J, Koerner M V., Gadalla KKE, Guy J, De Sousa D, et al. Radically truncated MeCP2 rescues Rett syndrome-like neurological defects. Nature. 2017;550(7676):398–401. 109. Yasui DH, Peddada S, Bieda MC, Vallero RO, Hogart A, Nagarajan RP, et al. Integrated epigenomic analyses of neuronal MeCP2 reveal a role for long-range interaction with active genes. Proc Natl Acad Sci U S A. 2007; 110. Nan X, Hou J, Maclean A, Nasir J, Lafuente MJ, Shu X, et al. Interaction between chromatin proteins MECP2 and ATRX is disrupted by mutations that cause inherited mental retardation. Proc Natl Acad Sci U S A. 2007 Feb;104(8):2709–14. 111. Sokpor G, Xie Y, Rosenbusch J, Tuoc T. Chromatin remodeling BAF (SWI/SNF) complexes in neural development and disorders. Frontiers in Molecular Neuroscience. 2017. 112. McDowell TL, Gibbons RJ, utherland H, O’Rourke DM, Bickmore WA, Pombo A, et al. Localization of a putative transcriptional regulator (ATRX) at pericentromeric heterochromatin and the short arms of acrocentric chromosomes. Proc Natl Acad Sci U S A. 1999; 113. Cardoso C, Timsit S, Villard L, Khrestchatisky M, Fontès M, Colleaux L. Specific interaction between the XNP/ATR-X gene product and the SET domain of the human EZH2 protein. Hum Mol Genet. 1998; 114. Xue Y, Gibbons R, Yan Z, Yang D, McDowell TL, Sechi S, et al. The ATRX syndrome protein forms a chromatin-remodeling complex with Daxx and localizes in promyelocytic leukemia nuclear bodies. Proc Natl Acad Sci U S A. 2003; 115. Young JI, Hong EP, Castle JC, Crespo-Barreto J, Bowman AB, Rose MF, et al. Regulation of RNA splicing by the methylation-dependent transcriptional repressor methyl-CpG binding protein 2. Proc Natl Acad Sci U S A. 2005 Dec;102(49):17551–8. 116. Li R, Dong Q, Yuan X, Zeng X, Gao Y, Chiao C, et al. Misregulation of alternative splicing in a mouse model of Rett syndrome. Misteli T, editor. PLOS Genet. 2016 Jun;12(6):e1006129. 117. Cheng TL, Wang Z, Liao Q, Zhu Y, Zhou WH, Xu W, et al. MeCP2 Suppresses Nuclear MicroRNA Processing and Dendritic Growth by Regulating the DGCR8/Drosha Complex. Dev Cell. 2014; 118. Wu H, Tao J, Chen PJ, Shahab A, Ge W, Hart RP, et al. Genome-wide analysis reveals methyl- CpG-binding protein 2-dependent regulation of microRNAs in a mouse model of Rett syndrome. Proc Natl Acad Sci U S A. 2010; 119. Urdinguio RG, Fernandez AF, Lopez-Nieva P, Rossi S, Huertas D, Kulis M, et al. Disrupted microrna expression caused by Mecp2 loss in a mouse model of Rett syndrome. Epigenetics. 2010;

222 120. Mellios N, Feldman DA, Sheridan SD, Ip JPK, Kwok S, Amoah SK, et al. MeCP2-regulated miRNAs control early human neurogenesis through differential effects on ERK and AKT signaling. Mol Psychiatry. 2018; 121. Jülich K, Horn D, Burfeind P, Erler T, Auber B, Chahrour M, et al. A novel MECP2 mutation in a boy with neonatal encephalopathy and facial dysmorphism. J Pediatr. 2009 Jul;155(1):140–3. 122. Imessaoudene B, Bonnefont JP, Royer G, Cormier-Daire V, Lyonnet S, Lyon G, et al. MECP2 mutation in non-fatal, non-progressive encephalopathy in a male. J Med Genet. 2001 Mar;38(3):171–4. 123. Schwartzman JS, Bernardino A, Nishimura A, Gomes RR, Zatz M. Rett syndrome in a boy with a 47,XXY karyotype confirmed by a rare mutation in the MECP2 gene. Neuropediatrics. 2001 Jun;32(3):162– 4. 124. Van Esch H, Bauters M, Ignatius J, Jansen M, Raynaud M, Hollanders K, et al. Duplication of the MECP2 region is a frequent cause of severe mental retardation and progressive neurological symptoms in males. Am J Hum Genet. 2005;77(3):442–53. 125. Ramocki MB, Peters SU, Tavyev YJ, Zhang F, Carvalho CMB, Schaaf CP, et al. Autism and other neuropsychiatric symptoms are prevalent in individuals with MeCP2 duplication syndrome. Ann Neurol. 2009 Dec;66(6):771–82. 126. Van Esch H, Bauters M, Ignatius J, Jansen M, Raynaud M, Hollanders K, et al. Duplication of the MECP2 region is a frequent cause of severe mental retardation and progressive neurological symptoms in males. Am J Hum Genet. 2005 Sep;77(3):442–53. 127. Cukier HN, Perez AM, Collins AL, Zhou Z, Zoghbi HY, Botas J. Genetic modifiers of MeCP2 function in Drosophila. PLoS Genet. 2008;4(9). 128. Pietri T, Roman A-CC, Guyon N, Romano SA, Washbourne P, Moens CB, et al. The first mecp2- null zebrafish model shows altered motor behaviors. Front Neural Circuits. 2013;7(July):118. 129. Nozawa K, Lin Y, Kubodera R, Shimizu Y, Tanaka H, Ohshima T. Zebrafish Mecp2 is required for proper axonal elongation of motor neurons and synapse formation. Dev Neurobiol. 2017; 130. Wu Y, Zhong W, Cui N, Johnson CM, Xing H, Zhang S, et al. Characterization of Rett Syndrome- like phenotypes in Mecp2-knockout rats. J Neurodev Disord. 2016;8(1):23. 131. Patterson KC, Hawkins VE, Arps KM, Mulkey DK, Olsen ML. MeCP2 deficiency results in robust Rett-like behavioural and motor deficits in male and female rats. Hum Mol Genet. 2015;25(15):3303–20. 132. Liu H, Chen Y, Niu Y, Zhang K, Kang Y, Ge W, et al. TALEN-mediated gene mutagenesis in rhesus and cynomolgus monkeys. Cell Stem Cell. 2014;14(3):323–8. 133. Chen Y, Yu J, Niu Y, Qin D, Liu H, Li G, et al. Modeling Rett syndrome ising TALEN-Edited MECP2 mutant cynomolgus monkeys. Cell. 2017;169(5):945-955.e10. 134. Guy J, Hendrich B, Holmes M, Martin JEE, Bird A. A mouse Mecp2-null mutation causes neurological symptoms that mimic Rett syndrome. Nat Genet. 2001 Mar;27(3):322–6. 135. Chen RZ, Akbarian S, Tudor M, Jaenisch R. Deficiency of methyl-CpG binding protein-2 in CNS neurons results in a Rett-like phenotype in mice. Nat Genet. 2001 Mar;27(3):327–31. 136. Pelka GJ, Watson CM, Radziewic T, Hayward M, Lahooti H, Christodoulou J, et al. Mecp2 deficiency is associated with learning and cognitive deficits and altered gene activity in the hippocampal region of mice. Brain. 2006; 137. Gandaglia A, Brivio E, Carli S, Palmieri M, Bedogni F, Stefanelli G, et al. A Novel Mecp2 Y120D Knock-in Model Displays Similar Behavioral Traits But Distinct Molecular Features Compared to the Mecp2- Null Mouse Implying Precision Medicine for the Treatment of Rett Syndrome. Molecular Neurobiology. 2018; 138. Brown K, Selfridge J, Lagger S, Connelly J, De Sousa D, Kerr A, et al. The molecular basis of variable phenotypic severity among common missense mutations causing Rett syndrome. Hum Mol Genet. 2016;25(3):558–70. 139. Goffin D, Zhou Z (Joe). The neural circuit basis of Rett syndrome. Vol. 7, Frontiers in Biology. 2012. p. 428–35. 140. Lawson-Yuen A, Liu D, Han L, Jiang ZI, Tsai GE, Basu AC, et al. Ube3a mRNA and protein expression are not decreased in Mecp2R168Xmutant mice. Brain Res. 2007; 141. Schaevitz LR, Gómez NB, Zhen DP, Berger-Sweeney JE. MeCP2 R168X male and female mutant mice exhibit Rett-like behavioral deficits. Genes, Brain Behav. 2013; 142. Wegener E, Brendel C, Fischer A, Hülsmann S, Gärtner J, Huppke P, et al. Characterization of the MeCP2R168X knockin mouse model for Rett syndrome. Cai H, editor. PLoS One. 2014 Dec;9(12):e115444.

223 143. Pitcher MR, Herrera JA, Buffington SA, Kochukov MY, Merritt JK, Fisher AR, et al. Rett syndrome like phenotypes in the R255X Mecp2 mutant mouse are rescued by MECP2 transgene. Hum Mol Genet. 2015;24(9):2662–72. 144. Tao J, Hu K, Chang Q, Wu H, Sherman NE, Martinowich K, et al. Phosphorylation of MeCP2 at Serine 80 regulates its chromatin association and neurological function. Proc Natl Acad Sci. 2009;106(12):4882–7. 145. Jentarra GM, Olfers SL, Rice SG, Srivastava N, Homanics GE, Blue M, et al. Abnormalities of cell packing density and dendritic complexity in the MeCP2 A140V mouse model of Rett syndrome/X-linked mental retardation. BMC Neurosci. 2010;11(1):19. 146. Shahbazian MD, Young JI, Yuva-Paylor LA, Spencer CM, Antalffy BA, Noebels JL, et al. Mice with truncated MeCP2 recapitulate many Rett syndrome features and display hyperacetylation of histone H3. Neuron. 2002;35(2):243–54. 147. Cohen S, Gabel HW, Hemberg M, Hutchinson AN, Sadacca LA, Ebert DH, et al. Genome-wide activity-dependent MeCP2 phosphorylation regulates nervous system development and function. Neuron. 2011;72(1):72–85. 148. Samaco RC, Fryer JD, Ren J, Fyffe S, Chao HT, Sun Y, et al. A partial loss of function allele of Methyl-CpG-binding protein 2 predicts a human neurodevelopmental syndrome. Hum Mol Genet. 2008;17(12):1718–27. 149. Guy J, Gan J, Selfridge J, Cobb S, Bird A. Reversal of neurological defects in a mouse model of Rett syndrome. Science (80- ). 2007 Feb;315(5815):1143–7. 150. Collins AL, Levenson JM, Vilaythong AP, Richman R, Armstrong DL, Noebels JL, et al. Mild overexpression of MeCP2 causes a progressive neurological disorder in mice. Hum Mol Genet. 2004;13(21):2679–89. 151. Sztainberg Y, Chen HM, Swann JW, Hao S, Tang B, Wu Z, et al. Reversal of phenotypes in MECP2 duplication mice using genetic rescue or antisense oligonucleotides. Nature. 2015; 152. Luikenhuis S, Giacometti E, Beard CF, Jaenisch R. Expression of MeCP2 in postmitotic neurons rescues Rett syndrome in mice. Proc Natl Acad Sci U S A. 2004; 153. Koerner M V., FitzPatrick L, Selfridge J, Guy J, De Sousa D, Tillotson R, et al. Toxicity of overexpressed MeCP2 is independent of HDAC3 activity. Genes Dev. 2018;32(23–24):1514–24. 154. Chahrour M, Zoghbi HY. The Story of Rett Syndrome: From Clinic to Neurobiology. Vol. 56, Neuron. 2007. p. 422–37. 155. Lalonde R, Strazielle C. Brain regions and genes affecting limb-clasping responses. Brain Research Reviews. 2011. 156. Mount RH, Charman T, Hastings RP, Reilly S, Cass H. The Rett Syndrome Behaviour Questionnaire (RSBQ): Refining the behavioural phenotype of Rett syndrome. J Child Psychol Psychiatry Allied Discip. 2002; 157. Orefice LLL, Zimmerman ALL, Chirila AMM, Sleboda SJJ, Head JPP, Ginty DDD. Peripheral Mechanosensory Neuron Dysfunction Underlies Tactile and Behavioral Deficits in Mouse Models of ASDs. Cell. 2016; 158. Samaco RC, Mandel-Brehm C, McGraw CM, Shaw CA, McGill BE, Zoghbi HY. Crh and Oprm1 mediate anxiety-related behavior and social approach in a mouse model of MECP2 duplication syndrome. Nat Genet. 2012; 159. Ramirez JM, Ward CS, Neul JL. Breathing challenges in Rett Syndrome: Lessons learned from humans and animal models. Vol. 189, Respiratory Physiology and Neurobiology. 2013. p. 280–7. 160. Hara M, Takahashi T, Mitsumasu C, Igata S, Takano M, Minami T, et al. Disturbance of cardiac gene expression and cardiomyocyte structure predisposes Mecp2-null mice to arrhythmias. Sci Rep. 2015 Jun;5(November 2014):1–17. 161. McCauley MD, Wang T, Mike E, Herrera J, Beavers DL, Huang T-W, et al. Pathogenesis of Lethal Cardiac Arrhythmias in Mecp2 Mutant Mice: Implication for Therapy in Rett Syndrome. Sci Transl Med. 2011;3(113):113ra125-113ra125. 162. Kyle SM, Vashi N, Justice MJ. Rett syndrome: a neurological disorder with metabolic components. Open Biol. 2018;8(2):170216. 163. Lioy DT, Garg SK, Monaghan CE, Raber J, Foust KD, Kaspar BK, et al. A role for glia in the progression of Rett’s syndrome. Nature. 2011 Jun;475(7357):497–500. 164. Kyle SM, Saha PK, Brown HM, Chan LC, Justice MJ. MeCP2 co-ordinates liver lipid metabolism with the NCoR1/HDAC3 corepressor complex. Hum Mol Genet. 2016;25(14):3029–41. 165. Guy J, Cheval H, Selfridge J, Bird A. The Role of MeCP2 in the Brain. Annu Rev Cell Dev Biol. 2011;27(1):631–52.

224 166. Itoh M, Ide S, Takashima S, Kudo S, Nomura Y, Segawa M, et al. Methyl CpG-binding protein 2 (a mutation of which causes Rett syndrome) directly regulates insulin-like growth factor binding protein 3 in mouse and human brains. J Neuropathol Exp Neurol. 2007;66(2):117–23. 167. Buchovecky CM, Turley SD, Brown HM, Kyle SM, McDonald JG, Liu B, et al. A suppressor screen in Mecp2 mutant mice implicates cholesterol metabolism in Rett syndrome. Nat Genet. 2013;45(9):1013– 20. 168. Großer E, Hirt U, Janc OA, Menzfeld C, Fischer M, Kempkes B, et al. Oxidative burden and mitochondrial dysfunction in a mouse model of Rett syndrome. Neurobiol Dis. 2012 Oct;48(1):102–14. 169. Vashi N, Justice MJ. Treating Rett syndrome: from mouse models to human therapies. Mammalian Genome. 2019. 170. Deaton AM, Webb S, Kerr ARW, Illingworth RS, Guy J, Andrews R, et al. Cell type-specific DNA methylation at intragenic CpG islands in the immune system. Genome Res. 2011;21(7):1074–86. 171. Renthal W, Boxer LD, Hrvatin S, Li E, Silberfeld A, Nagy MA, et al. Characterization of human mosaic Rett syndrome brain tissue by single-nucleus RNA sequencing. Nat Neurosci. 2018;21(12):1670– 9. 172. Roux JC, Dura E, Moncla A, Mancini J, Villard L. Treatment with desipramine improves breathing and survival in a mouse model for Rett syndrome. Eur J Neurosci. 2007; 173. Mancini J, Dubus JC, Jouve E, Roux JC, Franco P, Lagrue E, et al. Effect of desipramine on patients with breathing disorders in RETT syndrome. Ann Clin Transl Neurol. 2018; 174. Abdala AP, Lioy DT, Garg SK, Knopp SJ, Paton JFR, Bissonnette JM. Effect of sarizotan, a 5- HT1aand D2-like receptor agonist, on respiration in three mouse models of rett syndrome. Am J Respir Cell Mol Biol. 2014; 175. Katz DM, Menniti FS, Mather RJ. N-Methyl-D-Aspartate Receptors, Ketamine, and Rett Syndrome: Something Special on the Road to Treatments? Vol. 79, Biological Psychiatry. 2016. p. 710–2. 176. Wenk GL. Alterations in dopaminergic function in Rett syndrome. Neuropediatrics. 1995;26(2):123– 5. 177. Ohno K, Saito Y, Ueda R, Togawa M, Ohmae T, Matsuda E, et al. Effect of serotonin 1A agonists and selective serotonin reuptake inhibitors on behavioral and nighttime respiratory symptoms in rett syndrome. Pediatr Neurol. 2016;60:54-59e1. 178. Paterson DS, Thompson EG, Belliveau RA, Antalffy BA, Trachtenberg FL, Armstrong DD, et al. Serotonin transporter abnormality in the dorsal motor nucleus of the vagus in rett syndrome: Potential implications for clinical autonomic dysfunction. J Neuropathol Exp Neurol. 2005;64(11):1018–27. 179. Viemari J-C. Mecp2 Deficiency Disrupts Norepinephrine and Respiratory Systems in Mice. J Neurosci. 2005;25(50):11521–30. 180. Santos M, Summavielle T, Teixeira-Castro A, Silva-Fernandes A, Duarte-Silva S, Marques F, et al. Monoamine deficits in the brain of methyl-CpG binding protein 2 null mice suggest the involvement of the cerebral cortex in early stages of Rett syndrome. Neuroscience. 2010; 181. Abdala AP, Toward MA, Dutschmann M, Bissonnette JM, Paton JFR. Deficiency of GABAergic synaptic inhibition in the Kölliker-Fuse area underlies respiratory dysrhythmia in a mouse model of Rett syndrome. J Physiol. 2016;594(1):223–37. 182. Chao H-TT, Chen H, Samaco RC, Xue M, Chahrour M, Yoo J, et al. Dysfunction in GABA signalling mediates autism-like stereotypies and Rett syndrome phenotypes. Nature. 2010;468(7321):263–9. 183. Patrizi A, Picard N, Simon AJ, Gunner G, Centofante E, Andrews NA, et al. Chronic Administration of the N-Methyl-D-Aspartate Receptor Antagonist Ketamine Improves Rett Syndrome Phenotype. Biol Psychiatry. 2016;79(9):755–64. 184. Klose RJ, Sarraf SA, Schmiedeberg L, McDermott SM, Stancheva I, Bird AP, et al. DNA binding selectivity of MeCP2 due to a requirement for A/T sequences adjacent to methyl-CpG. Mol Cell. 2005 Sep;19(5):667–78. 185. Kron M, Lang M, Adams IT, Sceniak M, Longo F, Katz DM. A BDNF loop-domain mimetic acutely reverses spontaneous apneas and respiratory abnormalities during behavioral arousal in a mouse model of Rett syndrome. DMM Dis Model Mech. 2014; 186. Deacon RMJ, Glass L, Snape M, Hurley MJ, Altimiras FJ, Biekofsky RR, et al. NNZ-2566, a Novel Analog of (1–3) IGF-1, as a Potential Therapeutic Agent for Fragile X Syndrome. NeuroMolecular Med. 2015;17(1):71–82. 187. Deogracias R, Yazdani M, Dekkers MPJ, Guy J, Ionescu MCS, Vogt KE, et al. Fingolimod, a sphingosine-1 phosphate receptor modulator, increases BDNF levels and improves symptoms of a mouse model of Rett syndrome. Proc Natl Acad Sci. 2012;

225 188. Tropea D, Giacometti E, Wilson NR, Beard C, McCurry C, Dong DF, et al. Partial reversal of Rett Syndrome-like symptoms in MeCP2 mutant mice. Proc Natl Acad Sci U S A. 2009; 189. Khwaja O , Ho E, Barnes K V., O’Leary HM, Pereira LM, Finkelstein Y, et al. afety, pharmacokinetics, and preliminary assessment of efficacy of mecasermin (recombinant human IGF-1) for the treatment of Rett syndrome. Proc Natl Acad Sci. 2014;111(12):4596–601. 190. Justice MJ, Buchovecky CM, Kyle SM, Djukic A. A role for metabolism in Rett syndrome pathogenesis. Rare Dis. 2013;1(1):e27265. 191. Park MJ, Aja S, Li Q, Degano AL, Penati J, Zhuo J, et al. Anaplerotic triheptanoin diet enhances mitochondrial substrate use to remodel the metabolome and improve lifespan, motor function, and sociability in MeCP2-null mice. PLoS One. 2014; 192. Abdala APL, Dutschmann M, Bissonnette JM, Paton JFR. Correction of respiratory disorders in a mouse model of Rett syndrome. Proc Natl Acad Sci. 2010; 193. El-Khoury R, Panayotis N, Matagne V, Ghata A, Villard L, Roux JC. GABA and glutamate pathways are spatially and developmentally affected in the brain of Mecp2-deficient mice. PLoS One. 2014; 194. Zhong W, Johnson CM, Wu Y, Cui N, Xing H, Zhang S, et al. Effects of early-life exposure to THIP on phenotype development in a mouse model of Rett syndrome. J Neurodev Disord. 2016; 195. Bittolo T, Raminelli CA, Deiana C, Baj G, Vaghi V, Ferrazzo S, et al. Pharmacological treatment with mirtazapine rescues cortical atrophy and respiratory deficits in MeCP2 null mice. Sci Rep. 2016; 196. Ishiyama M, Tamura S, Ito H, Takei H, Hoshi M, Asano M, et al. Early postnatal treatment with valproate induces gad1 promoter remodeling in the brain and reduces apnea episodes in mecp2-null mice. Int J Mol Sci. 2019; 197. Toward MA, Abdala AP, Knopp SJ, Paton JFR, Bissonnette JM. Increasing brain serotonin corrects CO2 chemosensitivity in methyl-CpG-binding protein 2 (Mecp2)-deficient mice. Exp Physiol. 2013; 198. Levitt ES, Hunnicutt BJ, Knopp SJ, Williams JT, Bissonnette JM. A selective 5-HT 1a receptor agonist improves respiration in a mouse model of Rett syndrome. J Appl Physiol. 2013; 199. De Filippis B, Nativio P, Fabbri A, Ricceri L, Adriani W, Lacivita E, et al. Pharmacological stimulation of the brain serotonin receptor 7 as a novel therapeutic approach for rett syndrome. Neuropsychopharmacology. 2014; 200. Szczesna K, De La Caridad O, Petazzi P, Soler M, Roa L, Saez MA, et al. Improvement of the rett syndrome phenotype in a mecp2 mouse model upon treatment with levodopa and a dopa-decarboxylase inhibitor. Neuropsychopharmacology. 2014; 201. Gogliotti RG, Senter RK, Rook JM, Ghoshal A, Zamorano R, Malosh C, et al. MGlu5positive allosteric modulation normalizes synaptic plasticity defects and motor phenotypes in a mouse model of Rett syndrome. Hum Mol Genet. 2016; 202. Tao J, Wu H, Coronado AA, de Laittre E, Osterweil EK, Zhang Y, et al. Negative Allosteric Modulation of mGluR5 Partially Corrects Pathophysiology in a Mouse Model of Rett Syndrome. J Neurosci. 2016; 203. Gogliotti RG, Senter RK, Fisher NM, Adams J, Zamorano R, Walker AG, et al. MGlu7potentiation rescues cognitive, social, and respiratory phenotypes in a mouse model of Rett syndrome. Sci Transl Med. 2017; 204. Khoury ES, Sharma A, Ramireddy RR, Thomas AG, Alt J, Fowler A, et al. Dendrimer-conjugated glutaminase inhibitor selectively targets microglial glutaminase in a mouse model of Rett syndrome. Theranostics. 2020; 205. Nance E, Kambhampati SP, Smith ES, Zhang Z, Zhang F, Singh S, et al. Dendrimer-mediated delivery of N-acetyl cysteine to microglia in a mouse model of Rett syndrome. J Neuroinflammation. 2017; 206. Kron M, Howell CJ, Adams IT, Ransbottom M, Christian D, Ogier M, et al. Brain activity mapping in Mecp2 mutant mice reveals functional deficits in forebrain circuits, including key nodes in the default mode network, that are reversed with ketamine treatment. J Neurosci. 2012 Oct;32(40):13860–72. 207. Na ES, De Jesús-Cortés H, Martinez-Rivera A, Kabir ZD, Wang J, Ramesh V, et al. D-cycloserine improves synaptic transmission in an animal mode of Rett syndrome. PLoS One. 2017; 208. Zanella S, Mebarek S, Lajard AM, Picard N, Dutschmann M, Hilaire G. Oral treatment with desipramine improves breathing and life span in Rett syndrome mouse model. Respir Physiol Neurobiol. 2008; 209. Mellios N, Woodson J, Garcia RI, Crawford B, Sharma J, heridan D, et al. β2-Adrenergic receptor agonist ameliorates phenotypes and corrects microRNA-mediated IGF1 deficits in a mouse model of Rett syndrome. Proc Natl Acad Sci. 2014;

226 210. Ogier M, Wang H, Hong E, Wang Q, Greenberg ME, Katz DM. Brain-Derived Neurotrophic Factor Expression and Respiratory Function Improve after Ampakine Treatment in a Mouse Model of Rett Syndrome. J Neurosci. 2007; 211. Schmid DA, Yang T, Ogier M, Adams I, Mirakhur Y, Wang Q, et al. A TrkB Small Molecule Partial Agonist Rescues TrkB Phosphorylation Deficits and Improves Respiratory Function in a Mouse Model of Rett Syndrome. J Neurosci. 2012; 212. Johnson RA, Lam M, Punzo AM, Li H, Lin BR, Ye K, et al. 7,8-dihydroxyflavone exhibits therapeutic efficacy in a mouse model of Rett syndrome. J Appl Physiol. 2012; 213. Krishnan N, Krishnan K, Connors CR, Choy MS, Page R, Peti W, et al. PTP1B inhibition suggests a therapeutic strategy for Rett syndrome. J Clin Invest. 2015 Aug;125(8):3163–77. 214. Li W, Bellot-Saez A, Phillips ML, Yang T, Longo FM, Pozzo-Miller L. A small-molecule TrkB ligand restores hippocampal synaptic plasticity and object location memory in Rett syndrome mice. Dis Model Mech. 2017; 215. Pitcher MR, Ward CS, Arvide EM, Chapleau CA, Pozzo-Miller L, Hoeflich A, et al. Insulinotropic treatments exacerbate metabolic syndrome in mice lacking MeCP2 function. Hum Mol Genet. 2013; 216. Castro J, Garcia RI, Kwok S, Banerjee A, Petravicz J, Woodson J, et al. Functional recovery with recombinant human IGF1 treatment in a mouse model of Rett Syndrome. Proc Natl Acad Sci. 2014;111(27):9941–6. 217. Nag N, Berger-Sweeney JE. Postnatal dietary choline supplementation alters behavior in a mouse model of Rett syndrome. Neurobiol Dis. 2007; 218. Ricceri L, De Filippis B, Fuso A, Laviola G. Cholinergic hypofunction in MeCP2-308 mice: Beneficial neurobehavioural effects of neonatal choline supplementation. Behav Brain Res. 2011; 219. chaevitz LR, Nicolai R, Lopez CM, D’Iddio , Iannoni E, Berger-Sweeney JE. Acetyl-L-Carnitine Improves Behavior and Dendritic Morphology in a Mouse Model of Rett Syndrome. PLoS One. 2012; 220. Chin EWM, Lim WM, Ma D, Rosales FJ, Goh ELK. Choline Rescues Behavioural Deficits in a Mouse Model of Rett Syndrome by Modulating Neuronal Plasticity. Molecular Neurobiology. 2018; 221. Mantis JG, Fritz CL, Marsh J, Heinrichs SC, Seyfried TN. Improvement in motor and exploratory behavior in Rett syndrome mice with restricted ketogenic and standard diets. Epilepsy Behav. 2009; 222. Ribeiro MC, Moore SM, Kishi N, Macklis JD, MacDonald JL. Vitamin D supplementation rescues aberrant NF-κB pathway activation and partially ameliorates Rett syndrome phenotypes in Mecp2 mutant mice . eneuro. 2020; 223. Janc OA, Hüser MA, Dietrich K, Kempkes B, Menzfeld C, Hülsmann S, et al. Systemic radical scavenger treatment of a mouse model of Rett syndrome: merits and limitations of the vitamin E derivative Trolox. Front Cell Neurosci. 2016;10. 224. Braun S, Kottwitz D, Nuber UA. Pharmacological interference with the glucocorticoid system influences symptoms and lifespan in a mouse model of Rett syndrome. Hum Mol Genet. 2012; 225. Panighini A, Duranti E, Santini F, Maffei M, Pizzorusso T, Funel N, et al. Vascular Dysfunction in a Mouse Model of Rett Syndrome and Effects of Curcumin Treatment. PLoS One. 2013; 226. Zuliani I, Urbinati C, Valenti D, Quattrini MC, Medici V, Cosentino L, et al. The Anti-Diabetic Drug Metformin Rescues Aberrant Mitochondrial Activity and Restrains Oxidative Stress in a Female Mouse Model of Rett Syndrome. J Clin Med. 2020;9(6). 227. Shapiro JR, Boskey AL, Doty SB, Lukashova L, Blue ME. Zoledronic acid improves bone histomorphometry in a murine model of Rett syndrome. Bone. 2017; 228. Vigli D, Cosentino L, Raggi C, Laviola G, Woolley-Roberts M, De Filippis B. Chronic treatment with the phytocannabinoid Cannabidivarin (CBDV) rescues behavioural alterations and brain atrophy in a mouse model of Rett syndrome. Neuropharmacology. 2018; 229. De Filippis B, Fabbri A, Simone D, Canese R, Ricceri L, Malchiodi-Albedi F, et al. Modulation of RhoGTPases improves the behavioral phenotype and reverses astrocytic deficits in a mouse model of rett syndrome. Neuropsychopharmacology. 2012; 230. De Filippis B, Valenti D, Chiodi V, Ferrante A, de Bari L, Fiorentini C, et al. Modulation of Rho GTPases rescues brain mitochondrial dysfunction, cognitive deficits and aberrant synaptic plasticity in female mice modeling Rett syndrome. Eur Neuropsychopharmacol. 2015; 231. Horiuchi M, Smith L, Maezawa I, Jin LW. CX3CR1 ablation ameliorates motor and respiratory dysfunctions and improves survival of a Rett syndrome mouse model. Brain Behav Immun. 2017; 232. Kaufmann WE, Sprouse J, Rebowe N, Hanania T, Klamer D, Missling CU. ANAVEX®2-73 (blarcamesine), a Sigma-1 receptor agonist, ameliorates neurologic impairments in a mouse model of Rett syndrome. Pharmacol Biochem Behav. 2019;

227 233. Garré JM, Silva HM, Lafaille JJ, Yang G. P2X7 receptor inhibition ameliorates dendritic spine pathology and social behavioral deficits in Rett syndrome mice. Nat Commun. 2020; 234. Tang X, Drotar J, Li K, Clairmont CD, Brumm AS, Sullins AJ, et al. Pharmacological enhancement of KCC2 gene expression exerts therapeutic effects on human Rett syndrome neurons and Mecp2 mutant mice. Sci Transl Med. 2019; 235. Kondo M, Gray LJ, Pelka GJ, Christodoulou J, Tam PPL, Hannan AJ. Environmental enrichment ameliorates a motor coordination deficit in a mouse model of Rett syndrome - Mecp2 gene dosage effects and BDNF expression. Eur J Neurosci. 2008; 236. Nag N, Moriuchi JM, Peitzman CGK, Ward BC, Kolodny NH, Berger-Sweeney JE. Environmental enrichment alters locomotor behaviour and ventricular volume in Mecp21lox mice. Behav Brain Res. 2009; 237. Lonetti G, Angelucci A, Morando L, Boggio EM, Giustetto M, Pizzorusso T. Early Environmental Enrichment Moderates the Behavioral and Synaptic Phenotype of MeCP2 Null Mice. Biol Psychiatry. 2010; 238. Kondo MA, Gray LJ, Pelka GJ, Leang SK, Christodoulou J, Tam PPL, et al. Affective dysfunction in a mouse model of Rett syndrome: Therapeutic effects of environmental stimulation and physical activity. Dev Neurobiol. 2016; 239. Hao S, Tang B, Wu Z, Ure K, Sun Y, Tao H, et al. Forniceal deep brain stimulation rescues hippocampal memory in Rett syndrome mice. Nature. 2015; 240. Jugloff DGM, Vandamme K, Logan R, Visanji NP, Brotchie JM, Eubanks JH. Targeted delivery of an Mecp2 transgene to forebrain neurons improves the behavior of female Mecp2-deficient mice. Hum Mol Genet. 2008;17(10):1386–96. 241. Nagel-Wolfrum K, Möller F, Penner I, Baasov T, Wolfrum U. Targeting Nonsense Mutations in Diseases with Translational Read-Through-Inducing Drugs (TRIDs). BioDrugs. 2016;30(2):49–74. 242. Conti E, Izaurralde E. Nonsense-mediated mRNA decay: Molecular insights and mechanistic variations across species. Vol. 17, Current Opinion in Cell Biology. 2005. p. 316–25. 243. Kuzmiak HA, Maquat LE. Applying nonsense-mediated mRNA decay research to the clinic: progress and challenges. Trends in Molecular Medicine. 2006. 244. Brendel C, Belakhov V, Werner H, Wegener E, Gärtner J, Nudelman I, et al. Readthrough of nonsense mutations in Rett syndrome: Evaluation of novel aminoglycosides and generation of a new mouse model. J Mol Med. 2011;89(4):389–98. 245. Linde L, Boelz S, Nissim-Rafinia M, Oren YS, Wilschanski M, Yaacov Y, et al. Nonsense-mediated mRNA decay affects nonsense transcript levels and governs response of cystic fibrosis patients to gentamicin. J Clin Invest. 2007; 246. Keeling KM, Xue X, Gunn G, Bedwell DM. Therapeutics Based on Stop Codon Readthrough. Annu Rev Genomics Hum Genet. 2014;15(1):371–94. 247. McDonald CM, Campbell C, Torricelli RE, Finkel RS, Flanigan KM, Goemans N, et al. Ataluren in patients with nonsense mutation Duchenne muscular dystrophy (ACT DMD): a multicentre, randomised, double-blind, placebo-controlled, phase 3 trial. Lancet. 2017;390(10101):1489–98. 248. Kerem E, Konstan MW, De Boeck K, Accurso FJ, Sermet-Gaudelus I, Wilschanski M, et al. Ataluren for the treatment of nonsense-mutation cystic fibrosis: A randomised, double-blind, placebo-controlled phase 3 trial. Lancet Respir Med. 2014;2(7):539–47. 249. Bhatnagar S, Zhu X, Ou J, Lin L, Chamberlain L, Zhu LJ, et al. Genetic and pharmacological reactivation of the mammalian inactive X chromosome. Proc Natl Acad Sci. 2014;111(35):12591–8. 250. Vacca M, Della Ragione F, calabrì F, D’Esposito M. X inactivation and reactivation in X-linked diseases. Vol. 56, Seminars in Cell and Developmental Biology. 2016. p. 78–87. 251. Sripathy S, Leko V, Adrianse RL, Loe T, Foss EJ, Dalrymple E, et al. Screen for reactivation of MeCP2 on the inactive X chromosome identifies the BMP/TGF-β superfamily as a regulator of XI T expression. Proc Natl Acad Sci. 2017;114(7):1619–24. 252. Gadalla KKE, Bailey MES, Cobb SR. MeCP2 and Rett syndrome: reversibility and potential avenues for therapy. Biochem J. 2011;439(1):1–14. 253. Foust KD, Nurre E, Montgomery CL, Hernandez A, Chan CM, Kaspar BK. Intravascular AAV9 preferentially targets neonatal neurons and adult astrocytes. Nat Biotechnol. 2009;27(1):59–65. 254. Gonçalves MAFV. Adeno-associated virus: From defective virus to effective vector. Vol. 2, Virology Journal. 2005. 255. McCarty DM, Monahan PE, Samulski RJ. Self-complementary recombinant adeno-associated virus (scAAV) vectors promote efficient transduction independently of DNA synthesis. Gene Ther. 2001;8(16):1248–54.

228 256. Gadalla KK, Bailey ME, Spike RC, Ross PD, Woodard KT, Kalburgi SN, et al. Improved Survival and Reduced Phenotypic Severity Following AAV9/MECP2 Gene Transfer to Neonatal and Juvenile Male Mecp2 Knockout Mice. Mol Ther. 2013 Jan;21(1):18–30. 257. Sinnett SE, Hector RD, Gadalla KKE, Heindel C, Chen D, Zaric V, et al. Improved MECP2 Gene Therapy Extends the Survival of MeCP2-Null Mice without Apparent Toxicity after Intracisternal Delivery. Mol Ther - Methods Clin Dev. 2017;5:106–15. 258. Gadalla KKE, Vudhironarit T, Hector RD, Sinnett S, Bahey NG, Bailey MES, et al. Development of a Novel AAV Gene Therapy Cassette with Improved Safety Features and Efficacy in a Mouse Model of Rett Syndrome. Mol Ther - Methods Clin Dev. 2017; 259. Alheid GF, McCrimmon DR. The chemical neuroanatomy of breathing. Respiratory Physiology and Neurobiology. 2008. 260. Homma I, Masaoka Y. Breathing rhythms and emotions. Experimental Physiology. 2008. 261. Sykes RM, Spyer KM, Izzo PN. Demonstration of glutamate immunoreactivity in vagal sensory afferents in the nucleus tractus solitarius of the rat. Brain Res. 1997; 262. Bianchi AL, Denavit-Saubie M, Champagnat J. Central control of breathing in mammals: Neuronal circuitry, membrane properties, and neurotransmitters. Physiological Reviews. 1995. 263. Smith JC, Abdala APL, Koizumi H, Rybak IA, Paton JFR. Spatial and functional architecture of the mammalian brain stem respiratory network: A hierarchy of three oscillatory mechanisms. J Neurophysiol. 2007; 264. Wang X, Hayes JA, Revill AL, Song H, Kottick A, Vann NC, et al. Laser ablation of Dbx1 neurons in the pre-Bötzinger complex stops inspiratory rhythm and impairs output in neonatal mice. Elife. 2014; 265. Song G, Poon CS. Functional and structural models of pontine modulation of mechanoreceptor and chemoreceptor reflexes. Respir Physiol Neurobiol. 2004; 266. Berger AJ, Herbert DA, Mitchell RA. Properties of apneusis produced by reversible cold block of the rostral pons. Respir Physiol. 1978; 267. Guyenet PG, Stornetta RL, Souza GMPR, Abbott SBG, Shi Y, Bayliss DA. The Retrotrapezoid Nucleus: Central Chemoreceptor and Regulator of Breathing Automaticity. Trends in Neurosciences. 2019. 268. Molkov YI, Bacak BJ, Dick TE, Rybak IA. Control of breathing by interacting pontine and pulmonary feedback loops. Front Neural Circuits. 2013; 269. Burri PH. Fetal and Postnatal Development of the Lung. Annu Rev Physiol. 1984; 270. Rackley CR, Stripp BR. Building and maintaining the epithelium of the lung. Journal of Clinical Investigation. 2012. 271. Nikolić MZ, un D, Rawlins EL. Human lung development: Recent progress and new challenges. Development (Cambridge). 2018. 272. Isaacson D, Shen J, McCreedy D, Calvert M, Cunha G, Baskin L. Dichotomous branching of human fetal lung demonstrated with light sheet fluorescence microscopy. Am J Respir Crit Care Med. 2017; 273. Schittny JC. Development of the lung. Cell and Tissue Research. 2017. 274. Boers JE, Ambergen AW, Thunnissen FBJM. Number and proliferation of basal and parabasal cells in normal human airway epithelium. Am J Respir Crit Care Med. 1998; 275. Evans MJ, Van Winkle LS, Fanucchi M V., Plopper CG. Cellular and molecular characteristics of basal cells in airway epithelium. Experimental Lung Research. 2001. 276. Rock JR, Randell SH, Hogan BLM. Airway basal stem cells: A perspective on their roles in epithelial homeostasis and remodeling. DMM Disease Models and Mechanisms. 2010. 277. Yaghi A, Dolovich M. Airway Epithelial Cell Cilia and Obstructive Lung Disease. Cells. 2016; 278. Rokicki W, Rokicki M, Wojtacha J, Dzeljijli A. The role and importance of club cells (Clara cells) in the pathogenesis of some respiratory diseases. Kardiochirurgia i Torakochirurgia Pol. 2016; 279. Van Lommel A. Pulmonary neuroendocrine cells (PNEC) and neuroepithelial bodies (NEB): Chemoreceptors and regulators of lung development. Paediatr Respir Rev. 2001; 280. Rock JR, Hogan BLM. Epithelial Progenitor Cells in Lung Development, Maintenance, Repair, and Disease. Annu Rev Cell Dev Biol. 2011; 281. Rogers DF. Airway goblet cells: Responsive and adaptable frontline defenders. European Respiratory Journal. 1994. 282. Mitzner W. Airway smooth muscle: The appendix of the lung. American Journal of Respiratory and Critical Care Medicine. 2004. 283. Sirianni N, Naidu S, Pereira J, Pillotto RF, Hoffman EP. Rett syndrome: confirmation of X-linked dominant inheritance, and localization of the gene to Xq28. Am J Hum Genet. 1998 Nov;63(5):1552–8. 284. Prange HD. Laplace’s law and the alveolus: A misconception of anatomy and a misapplication of physics. American Journal of Physiology - Advances in Physiology Education. 2003.

229 285. Veldhuizen EJA, Haagsman HP. Role of pulmonary surfactant components in surface film formation and dynamics. Biochimica et Biophysica Acta - Biomembranes. 2000. 286. Khoor A, Stahlman MT, Gray ME, Whitsett JA. Temporal-spatial distribution of SP-B and SP-C proteins and mRNAs in developing respiratory epithelium of human lung. J Histochem Cytochem. 1994; 287. Kuroki Y, Takahashi M, Nishitani C. Pulmonary collectins in innate immunity of the lung. Cellular Microbiology. 2007. 288. Klein JM, Thompson MW, Snyder JM, George TN, Whitsett JA, Bell EF, et al. Transient surfactant protein B deficiency in a term infant with severe respiratory failure. J Pediatr. 1998; 289. Oosterlaken-Dijksterhuis MA, Haagsman HP, van Golde LMG, Demel RA. Characterization of Lipid Insertion into Monomolecular Layers Mediated by Lung Surfactant Proteins SP-B and SP-C. Biochemistry. 1991; 290. Goerke J, Gonzales J. Temperature dependence of dipalmitoyl phosphatidylcholine monolayer stability. J Appl Physiol Respir Environ Exerc Physiol. 1981; 291. Qua Hiansen J, Keating E, Aspros A, Yao LJ, Bosma KJ, Yamashita CM, et al. Cholesterol- mediated surfactant dysfunction is mitigated by surfactant protein A. Biochim Biophys Acta - Biomembr. 2015; 292. Osanai K, Tsuchihara C, Hatta R, Oikawa T, Tsuchihara K, Iguchi M, et al. Pulmonary surfactant transport in alveolar type II cells. In: Respirology. 2006. 293. Wright JR. Immunoregulatory functions of surfactant proteins. Nature Reviews Immunology. 2005. 294. Han SH, Mallampalli RK. The role of surfactant in lung disease and host defense against pulmonary infections. Annals of the American Thoracic Society. 2015. 295. Wright JR. Clearance and recycling of pulmonary surfactant. American Journal of Physiology - Lung Cellular and Molecular Physiology. 1990. 296. Rishikof DC, Ricupero DA, Goldstein RH. Extracellular matrix. In: Idiopathic Pulmonary Fibrosis. 2003. 297. Naba A, Hoersch S, Hynes RO. Towards definition of an ECM parts list: An advance on GO categories. Matrix Biology. 2012. 298. Zhou Y, Horowitz JC, Naba A, Ambalavanan N, Atabai K, Balestrini J, et al. Extracellular matrix in lung development, homeostasis and disease. Matrix Biology. 2018. 299. Dunsmore SE, Rannels DE. Extracellular matrix biology in the lung. American Journal of Physiology - Lung Cellular and Molecular Physiology. 1996. 300. Patel A, Fine B, Sandig M, Mequanint K. Elastin biosynthesis: The missing link in tissue-engineered blood vessels. Cardiovascular Research. 2006. 301. Kulkarni T, O’Reilly P, Antony VB, Gaggar A, Thannickal VJ. Matrix remodeling in pulmonary fibrosis and emphysema. American Journal of Respiratory Cell and Molecular Biology. 2016. 302. Dick TE, Dutschmann M, Feldman JL, Fong AY, Hülsmann S, Morris KM, et al. Facts and challenges in respiratory neurobiology. Respiratory Physiology and Neurobiology. 2018. 303. Trapnell BC, Nakata K, Bonella F, Campo I, Griese M, Hamilton J, et al. Pulmonary alveolar proteinosis. Nature Reviews Disease Primers. 2019. 304. Stanley E, Lieschke GJ, Grail D, Metcalf D, Hodgson G, Gall JAM, et al. Granulocyte/macrophage colony-stimulating factor-deficient mice show no major perturbation of hematopoiesis but develop a characteristic pulmonary pathology. Proc Natl Acad Sci U S A. 1994; 305. Clark JC, Wert SE, Bachurski CJ, Stahlman MT, Stripp BR, Weaver TE, et al. Targeted disruption of the surfactant protein B gene disrupts surfactant homeostasis, causing respiratory failure in newborn mice. Proc Natl Acad Sci U S A. 1995; 306. Glasser SW, Detmer EA, Ikegami M, Na CL, Stahlman MT, Whitsett JA. Pneumonitis and emphysema in sp-C gene targeted mice. J Biol Chem. 2003; 307. Ban N, Matsumura Y, Sakai H, Takanezawa Y, Sasaki M, Arai H, et al. ABCA3 as a lipid transporter in pulmonary surfactant biogenesis. J Biol Chem. 2007; 308. Nogee LM. Alterations in SP-B and SP-C Expression in Neonatal Lung Disease. Annu Rev Physiol. 2004; 309. Tredano M, Griese M, de Blic J, Lorant T, Houdayer C, Schumacher S, et al. Analysis of 40 sporadic or familial neonatal and pediatric cases with severe unexplained respiratory distress: Relationship toSFTPB. Am J Med Genet. 2003; 310. Whitsett JA, Wert SE, Weaver TE. Alveolar Surfactant Homeostasis and the Pathogenesis of Pulmonary Disease. Annu Rev Med. 2010; 311. Kröner C, Wittmann T, Reu S, Teusch V, Klemme M, Rauch D, et al. Lung disease caused by ABCA3 mutations. Thorax. 2017;

230 312. Cottin V, Hirani NA, Hotchkin DL, Nambiar AM, Ogura T, Otaola M, et al. Presentation, diagnosis and clinical course of the spectrum of progressive-fibrosing interstitial lung diseases. European Respiratory Review. 2018. 313. Werderman DS. Idiopathic Pulmonary Fibrosis. Radiol Technol. 2020; 314. Noth I, Zhang Y, Ma SF, Flores C, Barber M, Huang Y, et al. Genetic variants associated with idiopathic pulmonary fibrosis susceptibility and mortality: A genome-wide association study. Lancet Respir Med. 2013; 315. Fingerlin TE, Murphy E, Zhang W, Peljto AL, Brown KK, Steele MP, et al. Genome-wide association study identifies multiple susceptibility loci for pulmonary fibrosis. Nat Genet. 2013; 316. Barratt S, Creamer A, Hayton C, Chaudhuri N. Idiopathic Pulmonary Fibrosis (IPF): An Overview. J Clin Med. 2018; 317. Vestbo J, Hurd SS, Agustí AG, Jones PW, Vogelmeier C, Anzueto A, et al. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease GOLD executive summary. American Journal of Respiratory and Critical Care Medicine. 2013. 318. Larici AR, Franchi P, Cicchetti G, Bonomo L. Copd. In: Medical Radiology. 2019. 319. Eriksson S. Pulmonary Emphysema and Alpha1‐Antitrypsin Deficiency. Acta Med Scand. 1964; 320. Lugaresi E, Cirignotta F, Montagna P. Abnormal breathing in the Rett syndrome. Brain Dev. 1985; 321. Southall DP, Kerr AM, Tirosh E, Amos P, Lang MH, Stephenson JBP. Hyperventilation in the awake state: Potentially treatable component of Rett syndrome. Arch Dis Child. 1988; 322. Marcus CL, Carroll JL, McColley SA, Loughlin GM, Curtis S, Pyzik P, et al. Polysomnographic characteristics of patients with Rett syndrome. J Pediatr. 1994; 323. Rohdin M, Fernell E, Eriksson M, Albåge M, Lagercrantz H, Katz-Salamon M. Disturbances in Cardiorespiratory Function During Day and Night in Rett Syndrome. Pediatr Neurol. 2007; 324. Kerr AM, Julu POO. Recent insights into hyperventilation from the study of Rett syndrome. Archives of Disease in Childhood. 1999. 325. Julu PO, Kerr AM, Apartopoulos F, Al-Rawas S, Engerström IW, Engerström L, et al. Characterisation of breathing and associated central autonomic dysfunction in the Rett disorder. Arch Dis Child. 2001; 326. Saito Y, Ito M, Ozawa Y, Matsuishi T, Hamano K, Takashima S. Reduced expression of neuropeptides can be related to respiratory disturbances in Rett syndrome. In: Brain and Development. 2001. 327. Medrihan L, Tantalaki E, Aramuni G, Sargsyan V, Dudanova I, Missler M, et al. Early defects of GABAergic synapses in the brain stem of a MeCP2 mouse model of Rett syndrome. J Neurophysiol. 2008; 328. Voituron N, Hilaire G. The benzodiazepine Midazolam mitigates the breathing defects of Mecp2- deficient mice. Respir Physiol Neurobiol. 2011; 329. Howell CJ, Sceniak MP, Lang M, Krakowiecki W, Abouelsoud FE, Lad SU, et al. Activation of the medial prefrontal cortex reverses cognitive and respiratory symptoms in a mouse model of RETT syndrome. eNeuro. 2017; 330. Bissonnette JM, Knopp SJ. Separate respiratory phenotypes in methyl-CpG-binding protein 2 (Mecp2) deficient mice. Pediatr Res. 2006; 331. Ward CS, Arvide EM, Huang TW, Yoo J, Noebels JL, Neul JL. MeCP2 is critical within HoxB1- derived tissues of mice for normal lifespan. J Neurosci. 2011; 332. Huang TW, Kochukov MY, Ward CS, Merritt J, Thomas K, Nguyen T, et al. Progressive changes in a distributed neural circuit underlie breathing abnormalities in mice lacking MeCP2. J Neurosci. 2016; 333. De Felice C, Guazzi G, Rossi M, Ciccoli L, Signorini C, Leoncini S, et al. Unrecognized lung disease in classic Rett syndrome: A physiologic and high-resolution CT imaging study. Chest. 2010; 334. Kida H, Takahashi T, Nakamura Y, Kinoshita T, Hara M, Okamoto M, et al. Pathogenesis of Lethal Aspiration Pneumonia in Mecp2-null Mouse Model for Rett Syndrome. Sci Rep. 2017; 335. Lawson WE, Cheng DS, Degryse AL, Tanjore H, Polosukhin V V., Xu XC, et al. Endoplasmic reticulum stress enhances fibrotic remodeling in the lungs. Proc Natl Acad Sci U S A. 2011; 336. Bridges JP, Schehr A, Wang Y, Huo L, Besnard V, Ikegami M, et al. Epithelial SCAP/INSIG/SREBP signaling regulates multiple biological processes during perinatal lung maturation. PLoS One. 2014; 337. Amir RE, Van den Veyver IB, Wan M, Tran CQ, Francke U, Zoghbi HY. Rett syndrome is caused by mutations in X-linked MECP2, encoding methyl-CpG-binding protein 2. Nat Genet. 1999 Oct;23(2):185– 8. 338. Enikanolaiye A, Ruston J, Zeng R, Taylor C, Schrock M, Buchovecky CM, et al. Suppressor mutations in Mecp2-null mice implicate the DNA damage response in Rett syndrome pathology. Genome Res. 2020;

231 339. Uhlen M, Oksvold P, Fagerberg L, Lundberg E, Jonasson K, Forsberg M, et al. Towards a knowledge-based Human Protein Atlas. Nature Biotechnology. 2010. 340. Ryder E, Doe B, Gleeson D, Houghton R, Dalvi P, Grau E, et al. Rapid conversion of EUCOMM/KOMP-CSD alleles in mouse embryos using a cell-permeable Cre recombinase. Transgenic Res. 2014; 341. inger BD, Mock JR, D’Alessio FR, Aggarwal NR, Mandke P, Johnston L, et al. Flow-cytometric method for simultaneous analysis of mouse lung epithelial, endothelial, and hematopoietic lineage cells. Am J Physiol - Lung Cell Mol Physiol. 2016; 342. Lun ATL, Bach K, Marioni JC. Pooling across cells to normalize single-cell RNA sequencing data with many zero counts. Genome Biol. 2016; 343. Schmidt D, Wilson MD, Spyrou C, Brown GD, Hadfield J, Odom DT. ChIP-seq: Using high- throughput sequencing to discover protein-DNA interactions. Methods. 2009 Jul;48(3):240–8. 344. Yates AD, Achuthan P, Akanni W, Allen J, Allen J, Alvarez-Jarreta J, et al. Ensembl 2020. Nucleic Acids Res. 2020; 345. Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, Zahler AM, et al. The Human Genome Browser at UCSC. Genome Res. 2002; 346. Goldstein JL, DeBose-Boyd RA, Brown MS. Protein sensors for membrane sterols. Cell. 2006. 347. Horton JD, Goldstein JL, Brown MS. SREBPs: activators of the complete program of cholesterol and fatty acid synthesis in the liver. J Clin Invest. 2002 May;109(9):1125–31. 348. Shimano H, Horton JD, Shimomura I, Hammer RE, Brown MS, Goldstein JL. Isoform 1c of sterol regulatory element binding protein is less active than isoform 1a in livers of transgenic mice and in cultured cells. J Clin Invest. 1997; 349. Hua X, Yokoyama C, Wu J, Briggs MR, Brown MS, Goldstein JL, et al. SREBP-2, a second basic- helix-loop-helix-leucine zipper protein that stimulates transcription by binding to a sterol regulatory element. Proc Natl Acad Sci U S A. 1993; 350. Shimano H, Sato R. SREBP-regulated lipid metabolism: Convergent physiology-divergent pathophysiology. Nature Reviews Endocrinology. 2017. 351. Han LQ, Gao TY, Yang GY, Loor JJ. Overexpression of SREBF chaperone (SCAP) enhances nuclear SREBP1 translocation to upregulate fatty acid synthase (FASN) gene expression in bovine mammary epithelial cells. J Dairy Sci. 2018; 352. Bertolio R, Napoletano F, Mano M, Maurer-Stroh S, Fantuz M, Zannini A, et al. Sterol regulatory element binding protein 1 couples mechanical cues and lipid metabolism. Nat Commun. 2019; 353. LeBlanc SE, Srinivasan R, Ferri C, Mager GM, Gillian-Daniel AL, Wrabetz L, et al. Regulation of cholesterol/lipid biosynthetic genes by Egr2/Krox20 during peripheral nerve myelination. J Neurochem. 2005; 354. Kim YM, Shin HT, Seo YH, Byun HO, Yoon SH, Lee IK, et al. Sterol Regulatory Element-binding Protein (SREBP)-1-mediated lipogenesis is involved in cell senescence. J Biol Chem. 2010; 355. Abildayeva K, Jansen PJ, Hirsch-Reinshagen V, Bloks VW, Bakker AHF, Ramaekers FCS, et al. 24(S)-hydroxycholesterol participates in a liver X receptor-controlled pathway in astrocytes that regulates apolipoprotein E-mediated cholesterol efflux. J Biol Chem. 2006 May;281(18):12799–808. 356. Fisher AB, Dodia C, Yu K, Manevich Y, Feinstein SI. Lung phospholipid metabolism in transgenic mice overexpressing peroxiredoxin 6. Biochim Biophys Acta - Mol Cell Biol Lipids. 2006; 357. Kriaucionis S, Paterson A, Curtis J, Guy J, MacLeod N, Bird A. Gene Expression Analysis Exposes Mitochondrial Abnormalities in a Mouse Model of Rett Syndrome. Mol Cell Biol. 2006 Jul;26(13):5033–42. 358. Pecorelli A, Leoni G, Cervellati F, Canali R, Signorini C, Leoncini S, et al. Genes related to mitochondrial functions, protein degradation, and chromatin folding are differentially expressed in lymphomonocytes of rett syndrome patients. Mediators Inflamm. 2013; 359. Shulyakova N, Andreazza AC, Mills LR, Eubanks JH. Mitochondrial dysfunction in the pathogenesis of Rett syndrome: implications for mitochondria-targeted therapies. Front Cell Neurosci. 2017;11. 360. De Felice C, Della Ragione F, Signorini C, Leoncini S, Pecorelli A, Ciccoli L, et al. Oxidative brain damage in Mecp2-mutant murine models of Rett syndrome. Neurobiol Dis. 2014 Aug;68:66–77. 361. Rube HT, Lee W, Hejna M, Chen H, Yasui DH, Hess JF, et al. Sequence features accurately predict genome-wide MeCP2 binding in vivo. Nat Commun. 2016; 362. Ricceri L, De Filippis B, Laviola G. Mouse models of Rett syndrome: from behavioural phenotyping to preclinical evaluation of new therapeutic approaches. Behav Pharmacol. 2008;19(5–6):501–17. 363. Samaco RC, Nagarajan RP, Braunschweig D, LaSalle JM. Multiple pathways regulate MeCP2 expression in normal brain development and exhibit defects in autism-spectrum disorders. Hum Mol Genet. 2004;13(6):629–39.

232 364. Corsini A, Bellosta S, Baetta R, Fumagalli R, Paoletti R, Bernini F. New insights into the pharmacodynamic and pharmacokinetic properties of statins. Pharmacol Ther. 1999 Dec;84(3):413–28. 365. Besnard V, Wert SE, Stahlman MT, Postle AD, Xu Y, Ikegami M, et al. Deletion of Scap in alveolar type II cells influences lung lipid homeostasis and identifies a compensatory role for pulmonary lipofibroblasts. J Biol Chem. 2009; 366. Doan ML, Guillerman RP, Dishop MK, Nogee LM, Langston C, Mallory GB, et al. Clinical, radiological and pathological features of ABCA3 mutations in children. Thorax. 2008; 367. Martinowich K, Hattori D, Wu H, Fouse S, He F, Hu Y, et al. DNA Methylation-Related Chromatin Remodeling in Activity-Dependent Bdnf Gene Regulation. Science (80- ). 2003 Oct;302(5646):890–3. 368. Tao R, Xiong X, DePinho RA, Deng CX, Dong XC. Hepatic SREBP-2 and cholesterol biosynthesis are regulated by FoxO3 and Sirt6. J Lipid Res. 2013; 369. Agudelo CW, Samaha G, Garcia-Arcos I. Alveolar lipids in pulmonary disease. A review. Lipids in Health and Disease. 2020. 370. Jolley CD, Dietschy JM, Turley SD. Genetic differences in cholesterol absorption in 129/Sv and C57BL/6 mice: effect on cholesterol responsiveness. Am J Physiol. 1999;276(5 Pt 1):G1117-24. 371. Tarquinio DC, Hou W, Neul JL, Berkmen GK, Drummond J, Aronoff E, et al. The course of awake breathing disturbances across the lifespan in Rett syndrome. Brain Dev. 2018; 372. Dotti MT, Manneschi L, Malandrini A, De Stefano N, Caznerale F, Federico A. Mitochondrial dysfunction in Rett syndrome An ultrastructural and biochemical study. Brain Dev. 1993;15(2):103–6. 373. Gold WA, Williamson SL, Kaur S, Hargreaves IP, Land JM, Pelka GJ, et al. Mitochondrial dysfunction in the skeletal muscle of a mouse model of Rett syndrome (RTT): Implications for the disease phenotype. Mitochondrion. 2014 Mar;15:10–7. 374. Müller M, Can K. Aberrant redox homoeostasis and mitochondrial dysfunction in Rett syndrome. Biochem Soc Trans. 2014 Aug;42(4):959–64. 375. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, et al. Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A. 2005; 376. Viemari J-CJC, Roux JC, Tryba AK, Saywell V, Burnet H, Peña F, et al. Mecp2 Deficiency Disrupts Norepinephrine and Respiratory Systems in Mice. J Neurosci. 2005;25(50):11521–30. 377. Ren J, Ding X, Funk GD, Greer JJ. Anxiety-related mechanisms of respiratory dysfunction in a mouse model of rett syndrome. J Neurosci. 2012; 378. Voituron N, Zanella S, Menuet C, Dutschmann M, Hilaire G. Early breathing defects after moderate hypoxia or hypercapnia in a mouse model of Rett syndrome. Respir Physiol Neurobiol. 2009; 379. Johnson CM, Cui N, Zhong W, Oginsky MF, Jiang C. Breathing abnormalities in a female mouse model of Rett syndrome. J Physiol Sci. 2015; 380. Shahbazian MDD, Zoghbi HYY. Molecular genetics of Rett syndrome and clinical spectrum of MECP2 mutations. Curr Opin Neurol. 2001 Apr;14(2):171–6. 381. Ganguly K, Martin TM, Concel VJ, Upadhyay S, Bein K, Brant KA, et al. Secreted phosphoprotein 1 is a determinant of lung function development in mice. Am J Respir Cell Mol Biol. 2014; 382. Mohawk JA, Green CB, Takahashi JS. Central and peripheral circadian clocks in mammals. Annual Review of Neuroscience. 2012. 383. Moretti P, Bouwknecht JA, Teague R, Paylor R, Zoghbi HY. Abnormalities of social interactions and home-cage behavior in a mouse model of Rett syndrome. Hum Mol Genet. 2005; 384. Shapiro SD, Endicott SK, Province MA, Pierce JA, Campbell EJ. Marked longevity of human lung parenchymal elastic fibers deduced from prevalence of D-aspartate and nuclear weapons-related radiocarbon. J Clin Invest. 1991; 385. Luo Y, Li N, Chen H, Fernandez GE, Warburton D, Moats R, et al. Spatial and temporal changes in extracellular elastin and laminin distribution during lung alveolar development. Sci Rep. 2018; 386. Athanazio R. Airway disease: Similarities and differences between asthma, COPD and bronchiectasis. Clinics. 2012. 387. Millar-Büchner P, Philp AR, Gutierrez N, Villanueva S, Kerr B, Flores CA. Severe changes in colon epithelium in the Mecp2-null mouse model of Rett syndrome. Mol Cell Pediatr. 2016; 388. Damiola F, Schibler U, Gene C-. A Serum Shock Induces Circadian Gene Expression in Mammalian Tissue Culture Cells. Cell. 1998; 389. Dragich JM, Kim YH, Arnold AP, Schanen NC. Differential distribution of the Mecp2 splice variants in the postnatal mouse brain. J Comp Neurol. 2007; 390. Martínez de Paz A, Sanchez-Mut JV, Samitier-Martí M, Petazzi P, Sáez M, Szczesna K, et al. Circadian cycle-dependent MeCP2 and brain chromatin changes. PLoS One. 2015;

233 391. Li Q, Loh DH, Kudo T, Truong D, Derakhshesh M, Kaswan ZMD, et al. Circadian rhythm disruption in a mouse model of Rett syndrome circadian disruption in RTT. Neurobiol Dis. 2015; 392. Panda S, Antoch MP, Miller BH, Su AI, Schook AB, Straume M, et al. Coordinated transcription of key pathways in the mouse by the circadian clock. Cell. 2002; 393. Truong KK, Lam MT, Grandner MA, Sassoon CS, Malhotra A. Timing matters: Circadian rhythm in sepsis, obstructive lung disease, obstructive sleep apnea, and cancer. Annals of the American Thoracic Society. 2016. 394. Hetzel MR, Clark TJH, Branthwaite MA. Asthma: Analysis of sudden deaths and ventilatory arrests in hospital. Br Med J. 1977; 395. Kraft M, Vianna E, Martin RJ, Leung DYM. Nocturnal asthma is associated with reduced glucocorticoid receptor binding affinity and decreased steroid responsiveness at night. J Allergy Clin Immunol. 1999; 396. Gwilt CR, Donnelly LE, Rogers DF. The non-neuronal cholinergic system in the airways: An unappreciated regulatory role in pulmonary inflammation? Pharmacology and Therapeutics. 2007. 397. Wendel DP, Taylor DG, Albertine KH, Keating MT, Li DY. Impaired distal airway development in mice lacking elastin. Am J Respir Cell Mol Biol. 2000; 398. Collins RT. Cardiovascular disease in Williams syndrome. Circulation. 2013. 399. Jiang X, Luttrell I, Li DY, Yang CC, Chitaley K. Altered bladder function in elastin-deficient mice at baseline and in response to partial bladder outlet obstruction. BJU Int. 2012; 400. Lacolley P, Boutouyrie P, Glukhova M, Daniel Lamaziere JM, Plouin PF, Bruneval P, et al. Disruption of the elastin gene in adult Williams Syndrome is accompanied by a paradoxical reduction in arterial stiffness. Clin Sci. 2002; 401. Shifren A, Mecham RP. The stumbling block in lung repair of emphysema: Elastic fiber assembly. In: Proceedings of the American Thoracic Society. 2006. 402. Hirano E, Knutsen RH, Sugitani H, Ciliberto CH, Mecham RP. Functional rescue of elastin insufficiency in mice by the human elastin gene: implications for mouse models of human disease. Circ Res. 2007;101(5):523–31. 403. Dubick MA, Rucker RB, Cross CE, Last JA. Elastin metabolism in rodent lung. BBA - Gen Subj. 1981; 404. Devos FC, Maaske A, Robichaud A, Pollaris L, Seys S, Lopez CA, et al. Forced expiration measurements in mouse models of obstructive and restrictive lung diseases. Respir Res. 2017; 405. Ohvo-Rekilä H, Ramstedt B, Leppimäki P, Peter Slotte J. Cholesterol interactions with phospholipids in membranes. Progress in Lipid Research. 2002. 406. Riobo NA. Cholesterol and its derivatives in Sonic Hedgehog signaling and cancer. Current Opinion in Pharmacology. 2012. 407. Dietschy JM, Turley SD. Cholesterol metabolism in the central nervous system during early development and in the mature animal. Journal of Lipid Research. 2004. 408. Pfrieger FW. Role of cholesterol in synapse formation and function. Biochimica et Biophysica Acta - Biomembranes. 2003. 409. Zhu R, Ou Z, Ruan X, Gong J. Role of liver X receptors in cholesterol efflux and inflammatory signaling (review). Molecular Medicine Reports. 2012. 410. Hong C, Tontonoz P. Liver X receptors in lipid metabolism: opportunities for drug discovery. Nat Rev Drug Discov. 2014 Jun;13(6):433–44. 411. Sun H, Yuan Y, Sun ZL. Cholesterol contributes to diabetic nephropathy through SCAP-SREBP-2 pathway. Int J Endocrinol. 2013; 412. Willy PJ, Umesono K, Ong ES, Evans RM, Heyman RA, Mangelsdorf DJ. LXR, a nuclear receptor that defines a distinct retinoid response pathway. Genes Dev. 1995; 413. Hörlein AJ, Näär AM, Heinzel T, Torchia J, Gloss B, Kurokawa R, et al. Ligand-independent repression by the thyroid hormone receptor mediated by a nuclear receptor co-repressor. Nature. 1995; 414. Glass CK, Rosenfeld MG. The coregulator exchange in transcriptional functions of nuclear receptors. Genes and Development. 2000. 415. Zhang Y, Breevoort R, Angdisen J, Fu M, chmidt DR, Holmstrom R, et al. Liver LXRα expression is crucial for whole body cholesterol homeostasis and reverse cholesterol transport in mice. J Clin Invest. 2012; 416. Repa JJ, Turley SD, Lobaccaro JMAA, Medina J, Li L, Lustig K, et al. Regulation of absorption and ABC1-mediated efflux of cholesterol by RXR heterodimers. Science American Association for the Advancement of Science; Sep, 2000 p. 1524–9.

234 417. Repa JJ, Berge KE, Pomajzl C, Richardson JA, Hobbs H, Mangelsdorf DJ. Regulation of ATP- binding cassette sterol transporters ABCG5 and ABCG8 by the liver X receptors alpha and beta. J Biol Chem. 2002 May;277(21):18793–800. 418. Nguyen P, Leray V, Diez M, erisier , Le Bloc’H J, iliart B, et al. Liver lipid metabolism. J Anim Physiol Anim Nutr (Berl). 2008; 419. Remmerie A, Scott CL. Macrophages and lipid metabolism. Cell Immunol. 2018; 420. Bates SR, Tao JQ, Collins HL, Francone OL, Rothblat GH. Pulmonary abnormalities due to ABCA1 deficiency in mice. Am J Physiol - Lung Cell Mol Physiol. 2005; 421. Kennedy MA, Barrera GC, Nakamura K, Baldán Á, Tarr P, Fishbein MC, et al. ABCG1 has a critical role in mediating cholesterol efflux to HDL and preventing cellular lipid accumulation. Cell Metab. 2005; 422. Kanungo S, Soares N, He M, Steiner RD. Sterol metabolism disorders and neurodevelopment— an update. Dev Disabil Res Rev. 2013 Jun;17(3):197–210. 423. Andersson M, Elmberger PO, Edlund C, Kristensson K, Dallner G. Rates of cholesterol, ubiquinone, dolichol and dolichyl-P biosynthesis in rat brain slices. Vol. 269, FEBS Letters. No longer published by Elsevier; 1990. 424. Goritz C, Mauch DH, Pfrieger FW. Multiple mechanisms mediate cholesterol-induced synaptogenesis in a CNS neuron. Mol Cell Neurosci. 2005 Jun;29(2):190–201. 425. Lütjohann D, Breuer O, Ahlborg G, Nennesmo I, Sidén A, Diczfalusy U, et al. Cholesterol homeostasis in human brain: evidence for an age-dependent flux of 24S-hydroxycholesterol from the brain into the circulation. Proc Natl Acad Sci U S A. 1996 Sep;93(18):9799–804. 426. Björkhem I, Lütjohann D, Diczfalusy U, Ståhle L, Ahlborg G, Wahren J. Cholesterol homeostasis in human brain: turnover of 24S-hydroxycholesterol and evidence for a cerebral origin of most of this oxysterol in the circulation. J Lipid Res. 1998 Aug;39(8):1594–600. 427. Lund EG, Guileyardo JM, Russell DW. cDNA cloning of cholesterol 24-hydroxylase, a mediator of cholesterol homeostasis in the brain. Proc Natl Acad Sci U S A. 1999 Jun;96(13):7238–43. 428. Wang L, Schuster GU, Hultenby K, Zhang Q, Andersson S, Gustafsson J-ÅÅ. Liver X receptors in the central nervous system: From lipid homeostasis to neuronal degeneration. Proc Natl Acad Sci U S A. 2002;99(21):13878–83. 429. Istvan ES, Deisenhofer J. Structural mechanism for statin inhibition of HMG-CoA reductase. Science (80- ). 2001 May;292(5519):1160–4. 430. Schultz JR, Tu H, Luk A, Repa JJ, Medina JC, Li L, et al. Role of LXRs in control of lipogenesis. Genes Dev. 2000 Nov;14(22):2831–8. 431. Zhang FL, Casey PJ. Protein Prenylation: Molecular Mechanisms and Functional Consequences. Annu Rev Biochem. 1996;65:241–69. 432. Clayton RB, Bloch K. The biological conversion of lanosterol to cholesterol. J Biol Chem. 1956 Jan;218(1):319–25. 433. Maki KC, Bays HE, Dicklin MR. Treatment options for the management of hypertriglyceridemia: Strategies based on the best-Available evidence. Journal of Clinical Lipidology. 2012. 434. Zanotti I, Poti F, Pedrelli M, Favari E, Moleri E, Franceschini G, et al. The LXR agonist T0901317 promotes the reverse cholesterol transport from macrophages by increasing plasma efflux potential. J Lipid Res. 2008 Jan;49(5):954–60. 435. Chisholm JW, Hong J, Mills SA, Lawn RM. The LXR ligand T0901317 induces severe lipogenesis in the db/db diabetic mouse. J Lipid Res. 2003;44. 436. Quinet EM, Basso MD, Halpern AR, Yates DW, Steffan RJ, Clerin V, et al. LXR ligand lowers LDL cholesterol in primates, is lipid neutral in hamster, and reduces atherosclerosis in mouse. J Lipid Res. 2009; 437. Katz A, Udata C, Ott E, Hickey L, Burczynski ME, Burghart P, et al. Safety, pharmacokinetics, and pharmacodynamics of single doses of lxr-623, a novel liver X-receptor agonist, in healthy participants. J Clin Pharmacol. 2009; 438. Quan G, Xie C, Dietschy JM, Turley SD. Ontogenesis and regulation of cholesterol metabolism in the central nervous system of the mouse. Dev Brain Res. 2003 Dec;146(1–2):87–98. 439. Chen VS, Morrison JP, Southwell MF, Foley JF, Bolon B, Elmore SA. Histology Atlas of the Developing Prenatal and Postnatal Mouse Central Nervous System, with Emphasis on Prenatal Days E7.5 to E18.5. Toxicol Pathol. 2017; 440. Semple BD, Blomgren K, Gimlin K, Ferriero DM, Noble-Haeusslein LJ. Brain development in rodents and humans: Identifying benchmarks of maturation and vulnerability to injury across species. Progress in Neurobiology. 2013. 441. Foran DR, Peterson AC. Myelin acquisition in the central nervous system of the mouse revealed by an MBP-Lac Z transgene. J Neurosci. 1992;

235 442. Jurevics H, Morell P. Cholesterol for Synthesis of Myelin Is Made Locally, Not Imported into Brain. J Neurochem. 1995; 443. Takamori S, Holt M, Stenius K, Lemke EA, Grønborg M, Riedel D, et al. Molecular Anatomy of a Trafficking Organelle. Cell. 2006; 444. Saito K, Dubreuil V, Arai Y, Wilsch-Bräuninger M, Schwudke D, Saher G, et al. Ablation of cholesterol biosynthesis in neural stem cells increases their VEGF expression and angiogenesis but causes neuron apoptosis. Proc Natl Acad Sci U S A. 2009; 445. Pfrieger FW. Outsourcing in the brain: Do neurons depend on cholesterol delivery by astrocytes? BioEssays. 2003;25(1):72–8. 446. Fünfschilling U, Saher G, Xiao L, Möbius W, Nave K-A, Incardona J, et al. Survival of adult neurons lacking cholesterol synthesis in vivo. BMC Neurosci. 2007;8(1):1. 447. Pfrieger FW, Ungerer N. Cholesterol metabolism in neurons and astrocytes. Progress in Lipid Research Oct, 2011 p. 357–71. 448. Whitney KD, Watson MA, Collins JL, Benson WG, Stone TM, Numerick MJ, et al. Regulation of cholesterol homeostasis by the liver X receptors in the central nervous system. Mol Endocrinol. 2002; 449. Lund EG, Xie C, Kotti T, Turley SD, Dietschy JM, Russell DW. Knockout of the cholesterol 24- hydroxylase gene in mice reveals a brain-specific mechanism of cholesterol turnover. J Biol Chem. 2003 Jun;278(25):22980–8. 450. Lopez AM, Chuang J-C, Posey KS, Turley SD. Suppression of brain cholesterol synthesis in male Mecp2-deficient mice is age dependent and not accompanied by a concurrent change in the rate of fatty acid synthesis. Brain Res. 2017;1654A:77–84. 451. Wang N, Yvan-Charvet L, Lütjohann D, Mulder M, Vanmierlo T, Kim T-W, et al. ATP-binding cassette transporters G1 and G4 mediate cholesterol and desmosterol efflux to HDL and regulate sterol accumulation in the brain. FASEB J. 2016;22(4):1073–82. 452. Cartagena CM, Burns MP, Rebeck GW. 24S-hydroxycholesterol effects on lipid metabolism genes are modeled in traumatic brain injury. Brain Res. 2010;1319:1–12. 453. Osterweil EKEKEK, Chuang SCS-CC, Chubykin AAA, Sidorov M, Bianchi R, Wong RKSKSRK, et al. Lovastatin corrects excess protein synthesis and prevents epileptogenesis in a mouse model of fragile X syndrome. Neuron. 2013 Jan;77(2):243–50. 454. Weitz-Schmidt G. Statins as anti-inflammatory agents. Vol. 23, Trends in Pharmacological Sciences. 2002. p. 482–6. 455. Bellosta S, Paoletti R, Corsini A. Safety of Statins: Focus on Clinical Pharmacokinetics and Drug Interactions. Circulation. 2004 Jun;109(23_suppl_1):III-50-III–57. 456. Berry-Kravis E, Levin R, Shah H, Mathur S, Darnell JC, Ouyang B. Cholesterol levels in Fragile X syndrome. Am J Med Genet Part A. 2015;167(2):379–84. 457. Carrette LLG, Wang C-Y, Wei C, Press W, Ma W, Kelleher RJ, et al. A mixed modality approach towards Xi reactivation for Rett syndrome and other X-linked disorders. Proc Natl Acad Sci. 2017;201715124, 458. Puglielli L, Tanzi RE, Kovacs DM. Alzheimer’s disease: the cholesterol connection. Nat Neurosci. 2003;6(4):345–51. 459. De Lau LML, Koudstaal PJ, Hofman A, Breteler MMB. Serum cholesterol levels and the risk of Parkinson’s disease. Am J Epidemiol. 2006;164(10):998–1002. 460. Karasinska JM, Hayden MR. Cholesterol metabolism in Huntington disease. Nat Rev Neurol. 2011;7(10):561–72. 461. Oram JF. Tangier disease and ABCA1. Biochimica et Biophysica Acta - Molecular and Cell Biology of Lipids. 2000. 462. Bi X, Liao G. Cholesterol in niemann–pick type C disease. Subcell Biochem. 2010;51:319–35. 463. Tranchant C. Niemann-Pick disease type C. Pratique Neurologique - FMC. 2011. 464. Krajnc N, Župančič N, Oražem J. Epilepsy treatment in Rett syndrome. J Child Neurol. 2011 Nov;26(11):1429–33. 465. Janc OA, Müller M. The free radical scavenger Trolox dampens neuronal hyperexcitability, reinstates synaptic plasticity, and improves hypoxia tolerance in a mouse model of Rett syndrome. Front Cell Neurosci. 2014; 466. Ben-Shachar S, Chahrour M, Thaller C, Shaw CA, Zoghbi HY. Mouse models of MeCP2 disorders share gene expression changes in the cerebellum and hypothalamus. Hum Mol Genet. 2009;18(13):2431– 42. 467. Takahashi M, Fukuoka J, Nitta N, Takazakura R, Nagatani Y, Murakami Y, et al. Imaging of pulmonary emphysema: A pictorial review. International Journal of COPD. 2008.

236 468. Feng D, Liu T, Sun Z, Bugge A, Mullican SEE, Alenghat T, et al. A Circadian Rhythm Orchestrated by Histone Deacetylase 3 Controls Hepatic Lipid Metabolism. Science (80- ). 2011 Mar;331(6022):1315–9. 469. Sun Z, Miller RA, Patel RT, Chen J, Dhir R, Wang H, et al. Hepatic Hdac3 promotes gluconeogenesis by repressing lipid synthesis and sequestration. Nat Med. 2012;18(6):934–42. 470. Jepsen K, Rosenfeld MG. Biological roles and mechanistic actions of corepressor complexes. J Cell Sci. 2002; 471. Fischle W, Dequiedt F, Hendzel MJ, Guenther MG, Lazar MA, Voelter W, et al. Enzymatic activity associated with class II HDACs is dependent on a multiprotein complex containing HDAC3 and SMRT/N- CoR. Mol Cell. 2002 Jan;9(1):45–57. 472. Kao HY, Downes M, Ordentlich P, Evans RM. Isolation of a novel histone deacetylase reveals that class I and class II deacetylases promote SMRT-mediated repression. Genes Dev. 2000; 473. Zhao C, Gammie SC. The circadian gene Nr1d1 in the mouse nucleus accumbens modulates sociability and anxiety-related behaviour. Eur J Neurosci. 2018; 474. Goto M, Mizuno M, Matsumoto A, Yang Z, Jimbo EF, Tabata H, et al. Role of a circadian-relevant gene NR1D1 in brain development: Possible involvement in the pathophysiology of autism spectrum disorders. Sci Rep. 2017; 475. Jager J, Timothy O’Brien W, Manlove J, Krizman EN, Fang B, Gerhart-Hines Z, et al. Behavioral changes and dopaminergic dysregulation in mice lacking the nuclear receptor rev-erbα. Mol Endocrinol. 2014; 476. Chung S, Lee EJ, Yun S, Choe HK, Park SB, Son HJ, et al. Impact of circadian nuclear receptor REV-ERBα on midbrain dopamine production and mood regulation. Cell. 2014; 477. Fontaine C, Rigamonti E, Pourcet B, Duez H, Duhem C, Fruchart JC, et al. The nuclear receptor Rev-erbα is a Liver X Receptor (LXR) target gene driving a negative feedback loop on select LXR-induced pathways in human macrophages. Mol Endocrinol. 2008; 478. Griffin P, Dimitry JM, Sheehan PW, Lananna B V., Guo C, Robinette ML, et al. Circadian clock protein Rev-erbα regulates neuroinflammation. Proc Natl Acad Sci U S A. 2019; 479. Woldt E, Sebti Y, Solt LA, Duhem C, Lancel S, Eeckhoute J, et al. Rev-erb-α modulates skeletal muscle oxidative capacity by regulating mitochondrial biogenesis and autophagy. Nat Med. 2013; 480. Gorman GS, Chinnery PF, DiMauro S, Hirano M, Koga Y, McFarland R, et al. Mitochondrial diseases. Nat Rev Dis Prim. 2016; 481. G. M, M.F. B. The role of mitochondria in the pathogenesis of neurodegenerative diseases. Brain Pathol. 2000;10(3):462–72. 482. Franklin MP, Sathyanarayan A, Mashek DG. Acyl-CoA thioesterase 1 (ACOT1) regulates PPARα to couple fatty acid flux with oxidative capacity during fasting. Diabetes. 2017; 483. Boukhenouna S, Wilson MA, Bahmed K, Kosmider B. Reactive oxygen species in chronic obstructive pulmonary disease. Oxidative Medicine and Cellular Longevity. 2018. 484. Kosmider B, Lin CR, Karim L, Tomar D, Vlasenko L, Marchetti N, et al. Mitochondrial dysfunction in human primary alveolar type II cells in emphysema. EBioMedicine. 2019; 485. Wiener-Kronish JP, Albertine KH, Matthay MA, Osorio O, Neuberger M. Differential responses of the endothelial and epithelial barriers of the lung in sheep to Escherichia coli endotoxin. J Clin Invest. 1991; 486. Fox-Dewhurst R, Alberts MK, Kajikawa O, Caldwell E, Johnson MC, Skerrett SJ, et al. Pulmonary and systemic inflammatory responses in rabbits with gram- negative pneumonia. Am J Respir Crit Care Med. 1997; 487. Raghavendran K, Nemzek J, Napolitano LM, Knight PR. Aspiration-induced lung injury. Critical Care Medicine. 2011. 488. Villani C, Sacchetti G, Bagnati R, Passoni A, Fusco F, Carli M, et al. Lovastatin fails to improve motor performance and survival in methyl-CpG-binding protein2-null mice. Elife. 2016;5(NOVEMBER2016). 489. Chung R-H, Ma D, Wang K, Hedges DJ, Jaworski JM, Gilbert JR, et al. An X chromosome-wide association study in autism families identifies TBL1X as a novel autism spectrum disorder candidate gene in males. Mol Autism. 2011;2(1):18. 490. Pons L, Cordier MP, Labalme A, Till M, Louvrier C, Schluth-Bolard C, et al. A new syndrome of intellectual disability with dysmorphism due to TBL1XR1 deletion. Am J Med Genet Part A. 2015;167(1):164–8. 491. Heinen C a, Jongejan A, Watson PJ, Redeker B, Boelen A, Boudzovitch-Surovtseva O, et al. A specific mutation in TBL1XR1 causes Pierpont syndrome. J Med Genet. 2016;1–8. 492. Saitsu H, Tohyama J, Walsh T, Kato M, Kobayashi Y, Lee M, et al. A girl with West syndrome and autistic features harboring a de novo TBL1XR1 mutation. J Hum Genet. 2014 Oct;59(July):1–3.

237 493. Crosson J, Srivastava S, Bibat GM, Gupta S, Kantipuly A, Smith-Hicks C, et al. Evaluation of QTc in Rett syndrome: Correlation with age, severity, and genotype. Am J Med Genet. 2017;173(6):1495–501. 494. O’Connor RD, Zayzafoon M, Farach-Carson MC, Schanen NC. Mecp2 deficiency decreases bone formation and reduces bone volume in a rodent model of Rett syndrome. Bone. 2009;45(2):346–56. 495. Assadi F, Crowe C, Rouhi O. Hyperkalemic distal renal tubular acidosis associated with Rett syndrome. Pediatr Nephrol. 2006;21(4):588–90.