Global Reactivation and Targeted Preservation of MeCP2 Expression in a Mouse Model of Rett Syndrome

Transcription

1 Global Reactivation and Targeted Preservation of MeCP2 Expression in a Mouse Model of Rett Syndrome By Min Lang A thesis submitted in conformity with the requirements for the degree of Master of Science Graduate Department of Physiology University of Toronto I

2 Ubiquitous Reactivation and Targeted Preservation of MeCP2 Expression in a Mouse Model of Rett Syndrome Min Lang Master of Science, 2012 Department of Physiology University of Toronto Abstract Rett syndrome is a neurodevelopmental disorder that is predominately caused by mutations of the MECP2 gene. As neuronal apoptosis is not observed in RTT patients and MeCP2-deficient mice, the neurological deficits may be reversible. To address this, we reactivated MeCP2 expression ubiquitously in MeCP2-deficient mice after symptom onset. Our results showed that life span, behavioural performances, EEG activity, thermoregulation, and daily rhythmic activity were significantly improved after MeCP2 reactivation. Furthermore, the extent of improvement was dependent upon the efficiency of MeCP2 reactivation. To assess the role of the catecholaminergic system in Rett syndrome pathophysiology, we selectively preserved MeCP2 function within tyrosine hydroxylase expressing cells. We observed a significant improvement in the life span of male rescue mice and reduced sudden unexplained death rates in female rescue mice. Behavioural performances and EEG patterns were also significantly improved. II

3 Acknowledgements I would like to thank Dr. Eubanks for giving me the opportunity to pursue research in his lab and taking the time to teach me and guide me throughout my Master s project. Dr. Zhang for teaching me about the field of EEG and guiding me through my projects. Chiping Wu for implanting all of the mice that was used for the studies. Richard Logan and Guanming Zhang for teaching me laboratory techniques that was used for my research. Dr. Hampson and Dr. Mount for their guidance and advice throughout my project. Elena Sidorova, Natalya Shulyakova, and Robert Wither for their help, advice, and support during my Master s. III

4 Table of Contents Title Page... I Abstract... II Acknowledgements...III Table of Contents... IV List of Figures... IX 1. Introduction General Overview (RTT) Clinical features of Rett syndrome Pathophysiology of Rett syndrome Neuropathology Neurochemistry Electroencephalogram abnormalities and epileptic seizures Autonomic deficits Rett syndrome and MECP MeCP2 function Mouse models of Rett syndrome Gross phenotypes of Rett syndrome mouse models Targeted deletion of MeCP2 expression Reversibility of deficits in mouse models of Rett syndrome Gene therapy...28 IV

5 1.13. Pharmacological treatments Rationale and hypothesis Project1: Delayed ubiquitous reactivation of MeCP Project 2: Preservation of MeCP2 function in catecholaminergic cells Project aims Materials and Methods Mice Western blotting Immunohistochemistry Tamoxifen treatment Electrophysiology data collection Behavioural assessments Phenotypic severity scoring Cell counting Telemetry probe implantation protocol Tethered electrode implantation Electroencephalographic recordings and analysis Statistics Results Rescue of MeCP2 expression in Stop/y,cre mice Restoration of MeCP2 rescues life span and gross phenotypic severity Extent of behavioral rescue is dependent upon MeCP2 reactivation percentage Epileptiform discharges are significantly attenuated after MeCP2 reactivation...57 V

6 Reactivation of MeCP2 in female MeCP2-deificient mice rescues behavioral performances MeCP2 reactivation improves daily rhythmic activity and thermoregulation in adult female MeCP2-deficient mice MeCP2 is selectively preserved in tyrosine hydroxylase-expressing neurons in the Rescue mouse brain Preservation of MeCP2 in catecholaminergic cells extends the lifespan of male MeCP2-deficient mice Preservation of MeCP2 in catecholaminergic cells decreases the rate of sudden unexpected death in female MeCP2-deficient mice Catecholaminergic preservation of MeCP2 improves deficits in ambulatory rate, motor coordination, and anxiety-like behavior in male MeCP2-deficient mice Catecholaminergic preservation of MeCP2 improves the ambulatory and anxiety-like behavioral deficits of adult female MeCP2-deficient mice Preservation of MeCP2 in catecholaminergic cells improves cortical EEG abnormalities in male, but not female, MeCP2-deficient mice Preservation of MeCP2 in catecholaminergic cells improves peak hippocampal theta frequency in male, but not female, MeCP2-deficient mice Preservation of MeCP2 in catecholaminergic cells rescues deficits in hippocampal gamma band oscillatory activity in male, but not female, MeCP2-deficient mice Discussion Part 1: Delayed global reactivation of MeCP2 expression Part 2: Selective preservation of MeCP2 functions in catecholaminergic cells Future Directions References VI

7 List of Figures Introduction Figure 1. Clinical progression of Rett syndrome...4 Figure 2. MeCP2 function...13 Figure 3. Different mouse models of Rett syndrome...19 Figure 4. Different transgenes expressed in different mouse lines...36 Results Figure 5. MeCP2 is reactivated in Stop,cre mice following tamoxifen treatment...47 Figure 6. Survival and gross phenotypic severity is significantly improved in Stop/y,cre mice after tamoxifen treatment...50 Figure 7. Behavioural performances are improved in Stop/y,cre mice following tamoxifen treatment...54 Figure 8. Epileptiform-like discharge incidence rate is significantly improved in Stop/y,cre mice after tamoxifen treatment...58 Figure 9. Behavioural performances are improved in Stop/+,cre mice following MeCP2 reactivation Figure 10. Daily activity is significantly improved in Stop/+,cre mice after MeCP2 reactivation...65 Figure 11. Core body temperature is improved in Stop/+,cre mice after MeCP2 reactivation...67 Figure 12. Temperature and mobility correlation is significantly improved in MeCP2 reactivated female mice...69 Figure 13. MeCP2 is selectively preserved in catecholaminergic neurons of "Rescue" mice...73 Figure 14. MeCP2 expression is not preserved in non-catecholaminergic neurons...75 VII

8 Figure 15. Survival and gross phenotypic behavior are improved in male and female "Rescue" mice...78 Figure 16. The phenotypic severity score of female MeCP2 +/- mice does not correlate with the time of their sudden and unexpected death...81 Figure 17. Behavioral performances are improved in male "Rescue" mice...84 Figure 18. Behavioral performances are improved in female "Rescue" mice...87 Figure 19. The incidence rate of cortical epileptiform discharge activity is reduced in male "Rescue" mice...92 Figure 20. The incidence rate of epileptiform discharge activity is not improved in female "Rescue" mice...94 Figure 21. The peak hippocampal theta frequency and the total hippocampal gamma activity power are significantly improved in male "Rescue" mice...97 Figure 22. The peak hippocampal theta frequency and the total power of hippocampal gamma activity are not improved in female "Rescue" mice VIII

9 Chapter 1 Introduction 1.1. General overview Ich bin Rett, ich bin Rett, ich will jetzt mit Ihnen sprechen (English It s Rett, It s Rett, I want to talk to you ), shouted Andreas Rett as he rushed towards Bengt Hagberg at the International Association Scientific Study Mental Deficiency conference in Toronto. Andreas Rett was the pediatric neurologist who first described the syndrome in It went relatively unnoticed, however, until Bengt Hagberg shared his clinical observation in Rett syndrome (RTT) is now recognized as a leading cause of mental retardation in females with a prevalence rate of 1 in 10,000 live female births (Hagberg et al., 1985; Matijevic et al., 2009; Rett, 1966). RTT is characterized by apparently normal development up to 6-18 months of age, followed by a period of regression and loss of previously acquired skills. The patients lose purposeful hand skills and develop stereotypical hand movements. Additional symptoms include absence of speech, autistic-like features, impaired sleep patterns, cold feet and/or hands, respiratory dysfunction, bruxism, back deformities, motor impairments, and seizures (Smeets et al., 2012; Weaving et al., 2005; Williamson and Christodoulou, 2006). After the stage of regression, which varies from weeks to years, the conditions stabilize and the patients usually survive into adulthood (Matijevic et al., 2009). The rate of sudden unexplained death is significantly greater in RTT patients than controls of the same age. (Matijevic et al., 2009). The reasons of these deaths may be associated with sudden respiratory failure, abnormal cardiac arrhythmia, or seizures (Asthana et al., 1990; Hagberg and Witt-Engerstrom, 1986; Matijevic et al., 2009). 1

10 Mutations in the MECP2 gene cause > 90% of typical RTT cases (Amir et al., 1999; Neul et al., 2008). The protein product, MeCP2 (methyl-cpg binding protein 2), binds to methylated dinucleotides and act as a transcription repressor (Amir et al., 1999; Chahrour and Zoghbi, 2007). However, recent evidence has suggested that MeCP2 may also activate the transcription of several genes (Chahrour et al., 2008). In addition, mutations involving the FOXG1 and CDKL5 gene make up less than 10% of RTT cases. The congenital variant of RTT is related to mutations of the FOXG1 gene and the infantile seizure variant is related to mutations of the CDKL5 gene (Samaco and Neul, 2011; Smeets et al., 2012). It is still unclear whether the gene products of FOXG1 and CDKL5 share a convergent pathway as MeCP2 or whether it induces RTT phenotypes in a different pathway altogether. The lack of observable signs of atrophy, degeneration, or demyelination differentiates RTT from neurodegenerative diseases (Armstrong et al., 1999). The absence of neurodegeneration and the lack of progressive brain weight loss support the hypothesis of RTT is associated with neurodevelopmental arrest. Mouse models of RTT have been generated to explore avenues of treatment and to investigate the reversibility of neurological symptoms. 1.2 Clinical features of Rett syndrome RTT is characterized by normal development for the first 6-18 months of age. The affected individuals meet developmental milestones with no obvious signs of developmental disturbances (Smeets et al., 2003). However, retro-analysis of home videos reveals hypotonia and inadequate hand-eye coordination patterns in some RTT girls before 6 months of age (Smeets et al., 2012). RTT is divided into four developmental stages. During the first 2

11 preregression stage, subtle signs and symptoms begin to manifest, such as reduced eye contact, less demand for attention, and delays in crawling and standing. Overall, gross development appears to be normal. The following regression stage usually occurs between 1 and 4 years of age (Glaze, 2005). The rapid loss of acquired skills occurs acutely and the period of regression can last for days or months. Affected individuals show an obvious decline in communicative and motor skills. Cognitive deficiency also becomes apparent. Exploratory behaviour is lost and sleep pattern becomes disturbed. Panting, hyperventilation, and other respiratory abnormalities also begin to manifest. Seizures may become present during this period. Conditions plateau and stabilize during the third post-regression stage of RTT. RTT girls begin to display a loss of purposeful hand movement and stereotypic hand wringing becomes prominent. Eye contact returns and irritability is slightly improved. Seizures (epilepsy) become a common feature that requires medical treatment. Girls with milder phenotypes may have preserved purposeful hand use and speech. Many affected individuals remain in this stage for the rest of their lives. The final stage of RTT is marked by motor deterioration. RTT patients develop motor weakness, rigidity, and scoliosis. Parkinson-like features such as ataxia, dystonia, and hypomimia becomes prominent. Many RTT patients survive into their 40s and 50s through constant care and assistance, however, sudden unexpected death is often reported (Smeets et al., 2012). 3

12 Figure 1 4

13 Figure 1. Clinical progression of Rett syndrome. RTT girls usually develop normally up to 6-18 months of age. Development then begins to stagnate and learning is delayed. This is followed by the second regression stage, in which previous acquired skills are lost, cognitive deficits become obvious, motor impairments become apparent, and stereotypic hand movements begin to manifest. Symptoms stabilize during clinical stage III and seizures onset is common during this period. The last clinical stage is characterized by motor deterioration. Patients develop Parkinson-like motor impairments, lose the ability to walk, and often become wheel chair bound. 5

14 1.3 Pathophysiology of Rett syndrome Neuropathology The brain anatomy of RTT girls is grossly normal, however, the brain size and weight are well below that of age-matched controls (Armstrong, 2005). At birth, though, the head circumference of RTT girls is relatively normal but decelerated head growth begins at 2-3 months of age (Armstrong, 2005). Atrophy and progressive decline of brain weight is not observed, which is consistent with impaired brain growth (Armstrong et al., 1999). The decrease in the rate of brain growth, however, is not uniform as the cerebral hemispheres are affected more than cerebellar regions. Specifically, brain volume does not increase at normal rates in the prefrontal, anterior, and the posterior temporal regions, while posterior temporal and occipital regions remain relatively preserved (Reiss et al., 1993; Subramaniam et al., 1997). Magnetic imaging studies have also shown a significant decrease of grey matter in RTT brains (Reiss et al., 1993; Subramaniam et al., 1997). Though there are numerous alterations within RTT brains, the effects are generally subtle with no overall decrease in the number of neurons. Neuronal packing density is increased whereas synaptic density, dendritic complexity, and neuronal size are significantly diminished in the RTT brain (Armstrong, 1997; Bauman et al., 1995; Jellinger et al., 1988; Naidu, 1997). Cortical hyperexcitability as well as giant amplitude somatosensory and visual evoked potentials are reported in RTT patients (Glaze, 2005; Guerrini et al., 1998; Yoshikawa et al., 1991). These alterations likely contribute to the EEG abnormalities and clinical seizure onset in RTT (Cooper et al., 1998; Glaze, 2005; Niedermeyer et al., 1997). 6

15 1.3.2 Neurochemistry During neural development, neurotransmitter/neuromodulators are critical for guiding neuronal migration, formation of synapses, and formation of neural networks (Berger-Sweeney et al., 1998; Misgeld et al., 2002). Examinations of the cerebrospinal fluid and autopsy of RTT patients revealed a multitude of neurotransmitter abnormalities, including dopamine, serotonin, noradrenaline, glutamate, GABA, substance P, and acetylcholine (Armstrong, 2005; Gadalla et al., 2011). Cerebrospinal fluid collected from RTT girls shows an increase in glutamate levels and an increase in glutamate receptor density in the cerebral cortex has been reported (Armstrong, 2005). GABAergic dysregulation, including reduced GABA content as well as reduced glutamic acid decarboxylase 1 and 2, is evident in brains from young female RTT patients (Johnston et al., 2005; Chao et al., 2010). Disruption of glutamate and GABAergic systems may underlie the imbalance between excitatory and inhibitory transmission in RTT and contribute to seizure genesis (Gatto and Broadie, 2010). Bioamine deficits underlying RTT-like phenotypes have been suggested since 1985 as many symptoms overlap with disorders involving dopamine or noradrenergic deficits (Nomura et al., 1985). Currently, there are contradicting findings as Zoghbi et al. reported a significant reduction in homo-vanillic acid (catecholamine metabolite) and 3-methoxy-4-hydroxyphenylethylene glycol (noradrenaline metabolite) in the CSF of RTT girls, whereas Perry et al. reported no significant reduction of bioamine levels in RTT girls (Perry et al., 1988; Zoghbi et al., 1989). More recent findings showed a significant reduction of bioamine levels in the substantia nigra of RTT patients and reduced dopamine and its metabolites in the cortex and basal ganglia (Wenk, 1996). Dopamine receptors, D 2 receptors in particular, have also been reported to be increased during early years (4-15 years of age) and decreased during adulthood (15-39 years of 7

16 age) in RTT patients (Chiron et al., 1993; Naidu et al., 2001). The alteration in receptor levels is consistent with the developmental stages of RTT. It is important to note the differential age and phenotypic severity of the patients in the bioamine studies, which may partially explain the conflicting findings (Lekman et al., 1990; Percy, 1992). Further investigation is required to establish the relationship between age, phenotypic severity, and neurochemical changes in RTT (Armstrong, 2005) Electroencephalogram abnormalities and epileptic seizures Epilepsy is a common co-morbidity observed amongst RTT patients and has a prevalence rate of 60-90% (Glaze et al., 2010). Many reported seizure occurrences, however, are nonepileptic in origin as parents and caregiver mistake breath holding, hyperventilation, blank stares or vacant episodes, and motor abnormalities for seizures (Garofalo et al., 1988; Niedermeyer and Naidu, 1998; Witt Engerstrom, 1992). Seizures negatively affect the patient s quality of life and bring difficulty for caregivers (Bahi-Buisson et al., 2008). The onset of seizures is usually between 2 to 3 years of age and during clinical Stages II and III (Witt Engerstrom, 1992). The severity and onset of seizures appears to diminish after puberty (Glaze et al., 2010). Seizures that develop before 1 year of age tends to be more severe and typically results in intractable epilepsy (Steffenburg et al., 2001). Furthermore, early seizure onset occurs less frequently within RTT patients affected by MECP2 mutations, and is more frequent within atypical RTT cases (Glaze et al., 2010). The infantile seizure variant of RTT is characterized by early seizure onset prior to the development of clinical RTT features (Aicardi, 1997). RTT associated seizures are often difficult to treat. Different types of seizures may manifest, including complex partial, tonic-clonic, tonic, 8

17 and myoclonic (Glaze, 2005; Steffenburg et al., 2001). Commonly used antiepileptic drugs include valproate, lamotrigine, and carbamazepine. However, RTT-associated seizures remain poorly controlled (Glaze, 2005). EEG abnormalities are reported in most RTT cases, including individuals without a history of seizures (Glaze et al., 1998). Rhythmic spike wave discharge events associated with absence episodes are characteristics of generalized electrographic seizures in RTT patients (Glaze, 2005). It has been suggested that these seizures originate from limbic structures (Boison, 2012). Hyperexcitability is observed within the hippocampus and the cortex of RTT girls. Alternate hypotheses implicate an imbalance between excitability and inhibitory tone in the RTT brain as the source of seizure genesis (Dani et al., 2005). Further investigation is required to deduce the origin of these EEG abnormalities and epileptic seizure events in order to develop more effective methods of treatment Autonomic deficits Cold extremities, anxiety-like behaviour, respiratory dysfunction, and cardiac abnormalities are prevalent features of RTT (Hagberg et al., 2002). These autonomic dysfunctions may contribute to sudden unexplained deaths in RTT patients (Glaze, 2005). Electrocardiogram studies on RTT girls have shown that the Q-T intervals are prolonged compared to age-matched controls (Ellaway et al., 1999; Sekul et al., 1994). The prevalence of the Q-T interval alteration increases during progressive stages of RTT: 36% of patients in Stage II, 38% of patients in Stage III, and 50% of those in Stage IV (Sekul et al., 1994). Heart rate variability and high frequency power in ECG recordings were also found to be diminished in a 9

18 study of 25 RTT patients (Johnsrude et al., 1995). The autonomic function, including cardiac function and respiration, is modulated by the parasympathetic and sympathetic nervous system through acetylcholine and norepinephrine release, respectively. The observed autonomic impairments in RTT may be caused by an imbalance between parasympathetic and sympathetic input (Glaze, 2005). Hyperventilation, apneas, and breath holding are commonly observed amongst RTT girls (Glaze, 2005). These symptoms are more pronounced during wakefulness and become less severe during sleep (Glaze, 2005). Julu et al. found a decrease in cardiac vagal tone and sensitivity to baroreflex stemming which may contribute to the irregular breathing patterns (Julu et al., 2001). Uncoupling of heart rate and breathing patterns are also observed during day time and night time (Weese-Mayer et al., 2006). The imbalance between sympathetic and parasympathetic tone, as well as compromised autonomic reflexes, are likely the underlying cause of these symptoms and sudden deaths in RTT patients (Glaze, 2005) Rett syndrome and MECP2 MECP2 is critical for normal neurodevelopment and disruption of its expression causes a wide array of neurological deficits. In 1999, Amir et al identified mutations of the gene, MECP2, as the predominate cause of typical RTT (>90%) (Amir et al., 1999). Altered MeCP2 expression is also observed in other neurodevelopmental disorders, including X-linked mental retardation, encephalopathy, Angelman s syndrome, and autism (Gonzales and LaSalle, 2010). MECP2 is located on the X-chromosome, Xq28 region (Guy et al., 2011), and thus exhibits an X-linked inheritance pattern. Most MECP2 mutations, however, transmit paternally through de novo 10

19 mutations in male germ cells (Girard et al., 2001). Although more rare, MEPC2 mutations can be transmitted maternally through mildly affected maternal carriers (Gonzales and LaSalle, 2010). Heterozygous mutations of MECP2 in females leads to RTT, whereas MECP2 mutations in males lead to infantile encephalopathy and early death (Weng et al., 2011). Most RTT females show a mosaic expression of MeCP2 throughout the body due to random X-chromosome inactivation, which contributes to the variability of phenotypic severity in RTT patients (Hoffbuhr et al., 2002). In murine models of RTT, the mutant allele seems to be favourably inactivated (Braunschweig et al., 2004). Recent evidence, however, have shown that X- chromosome inactivation is insufficient to explain the variability in phenotypic severity and therefore, other genetic modifiers must be at play (Takahashi et al., 2008). More than 300 mutations of the MECP2 gene have been related to RTT, and certain types of mutation seem to correlate with the level of phenotype manifestation. Nonsense mutations, which ablate normal MeCP2 function, are found in severe cases of RTT, and truncated mutations, which may preserve partial MeCP2 function, are found in milder manifestations (Bebbington et al., 2010; Huppke et al., 2000; Smeets et al., 2005) MeCP2 function MECP2 gene encodes for MeCP2, a 53 kda nuclear protein belonging to the methyl-cpg binding domain family of proteins (D'Esposito et al., 1996; Quaderi et al., 1994). MeCP2 exists as two isoforms, e1 and e2, with different N-termini (Gadalla et al., 2011). Both isoforms, however, contains a nuclear localization signal, CpG binding domain, and a transcriptional repression domain that is important for formation of the repressor complex. MeCP2 has been 11

20 shown to interact with corepressors such as Sin3a, cski, NcoR, and CoREST (Guy et al., 2011). Histone deacetylases, HDAC1 and HDAC2, can be recruited to the repressor complex which results in condensation of chromatin structure and hindering of gene transcription (Samaco and Neul, 2011). Though MeCP2 is generally considered to be a global repressor, genetic studies have shown only subtle changes in gene transcription profiles. This suggests that MeCP2 may function more than a proximal repressor of methylated genes. Alternate evidence has shown MeCP2 to interact with CREB, a transcription activator (Chahrour et al., 2008). Further, MeCP2 have also been shown to interact with RNA splice site regulators such as RNA-binding protein Y box-binding protein 1 (Young et al., 2005). Though MeCP2 is characterized by a methyl-cpg binding domain, ChIP assays have shown MeCP2 binding to non-methylated DNA sequences, although it remains unclear whether this is affected by other intermediary proteins (Guy et al., 2011). MeCP2 is expressed throughout the body but it is most abundantly found in post-mitotic neurons (Amir et al., 1999). Glia cells of the neural system also express MeCP2, albeit at significantly lower levels (Ballas et al., 2009). MeCP2 expression varies during different stages of development, with low expression levels during embryogenesis and increasing expression levels during neuronal development and synaptogenesis (Guy et al., 2011). Further, the absence of MeCP2 does not seem to affect neuronal precursor formation, migration, or development (Kishi and Macklis, 2004). These findings suggest that MeCP2 s primary function is maintaining neuronal maturity rather than neuronal development (Guy et al., 2011). The observation that RTT patients show normal development up to 6-18 months of age, followed by a period of regression, is consistent with the proposed function and expression profile of MeCP2. 12

21 Figure 2 13

22 Figure 2. MeCP2 function MeCP2 functioning as a transcription repressor. MeCP2 binds to methylated CpG islands and recruits co-repressor complexes, including Sin3a and histone deacetylases. Chromatin condenses and transcription of genes is inhibited. 14

23 1.8. Mouse models of Rett syndrome Our understanding of RTT remained limited due to the variability of phenotypic severity, different forms of MECP2 mutations, different variants of RTT, age-dependent factors, and environmental effects. These factors complicate findings and at times produce conflicting results. Several animal models of RTT have been produced that allow investigators to properly control conditions to reach more definite conclusions (Chen et al., 2001; Guy et al., 2007; Guy et al., 2001; Zoghbi, 2005). The utilization of animal models has provided more insight into the mechanisms underlying the development, pathophysiology, and alterations of RTT. However, it is important to remember there are fundamental difference between the anatomy, biochemistry, and morphology in animal models and human patients (Peters et al., 2007). These differences occur at the molecular level to phenotypic behaviours. Although animal models can increase our understanding of different disorders, there are also limitations to the information that we can extract. After uncovering that MECP2 mutations are the underlying cause of most RTT cases, several mouse models with MECP2 mutations are generated to recapitulate the human condition (Chen et al., 2001; Guy et al., 2007; Guy et al., 2001; Shahbazian et al., 2002). MeCP2-null male mice were generated from both Dr. Jaenisch and Dr. Bird s group (Chen et al., 2001; Guy et al., 2001). The Jaenisch mouse line, MeCP2 Jae was produced by deleting exon 3 of the MECP2 gene, whereas both exon 3 and 4 are both deleted in the Bird s mouse line, MeCP2 tm1bird. Both mouse models recapitulate many features of RTT. These mice develop motor impairments, tremors, respiratory dysfunction, and stereotypic limb movements. Similar to the human condition, these mice seems to develop normally up to 6 weeks of age, after which observable symptoms quickly develop resulting in early lethality (~10-20 weeks of age). Chen et al. also generated a brain- 15

24 specific MeCP2 knockout mouse line through the use of a Nestin-cre transgene (Chen et al., 2001). Mice lacking MeCP2 only in the brain are phenotypically similar to MeCP2-null mice with identical symptom development, suggesting that RTT is mainly a neurological disorder and the function of MeCP2 is primarily within the brain. Mouse models mimicking the different human MECP2 mutations in humans have been generated. Dr. Zoghbi s group produced a transgenic mouse model of RTT (MeCP2 308 ), which expresses a truncated mutation, MeCP2-308, that is commonly found in RTT patients (Shahbazian et al., 2002). In this mouse model, a premature stop codon was inserted after codon 308 of the MECP2 gene. These mice develop milder symptoms than MeCP2-null mice but show RTT-like phenotypes, including stereotypic limb movements, and impairments in social and spatial memory (Moretti et al., 2006; Moretti and Zoghbi, 2006). A mouse model expressing another truncated mutation of MECP2, MeCP2-168, was also produced (Lawson-Yuen et al., 2007). MeCP2 168 male mice develop milder symptoms than MeCP2-null mice but typical RTTlike phenotypes are still present. MeCP2 168 female mice develop normally up to 6 months of age and survive past 1 year of age. Another mouse model is the MeCP2A140V, which possess a point mutation that causes the 140 th alanine to be replaced by valine (Jentarra et al., 2010). Only 0.6% of RTT patients exhibit this form of mutation and show a distinct variant phenotype of RTT that resembles X-linked mental retardation. A more common form of MECP2 mutation, T158M, encompasses 10% of all RTT cases (Goffin et al., 2012). A mouse line is generated with the same mutation, which substitutes a threonine with an alanine at the C-terminus of the MBD, and clinical RTT manifestations are observed in this transgenic mouse model (Goffin et al., 2012). Lastly, a complete MeCP2-null mouse line, MeCP2 Tam is produced by disrupting the 16

25 MBD of MECP2, which essentially ablates the function of the MeCP2 protein (Pelka et al., 2006). RTT-like learning and cognitive deficits are features of this mouse model. In 2007, Guy et al. engineered a new mouse model of RTT, MeCP2 tm2bird, by inserting a floxed-stop cassette between exon 2 and 3 of the MECP2 gene (Guy et al., 2007). These mice are phenotypically similar to MeCP2 tm1bird mice. The floxed-stop cassette enables conditional reactivation of the MECP2 gene through Cre recombinase activity. By crossing MeCP2 tm2bird with mouse lines expressing different transgenes of Cre recombinase driven by various promoters, MECP2 reactivation can be controlled temporally and spatially. This has led to several reactivation and rescue studies that have provided further insight on the pathophysiology and reversibility of RTT. Female counterparts of the different mouse models are also generated. MECP2 expression is disrupted in each model as described previously, however, these female mice still possess a functional MECP2 allele (Guy et al., 2001). X-chromosome inactivation occurs in every cell. Thus, depending upon the inactivation profile, the phenotypic severity can be significantly skewed. Female mouse models of RTT develop, on average, milder symptoms than male counterparts and are phenotypically normal up to 6 months of age (Chen et al., 2001; Guy et al., 2001). Tremors, breathing abnormalities, motor impairments, anxiety-like behaviour, and cognitive deficits develop progressively throughout adulthood. Unlike the human condition, female MeCP2-deficient mice are more prone to becoming obese (Guy et al., 2001). Female mouse models are more clinically relevant, as RTT is a disorder that almost exclusively affects female humans. Most animal work, however, has been conducted on hemizygous male mice with a disrupted MECP2 allele (Ricceri et al., 2008). Male models of 17

26 RTT eliminate the confounding effects of phenotypic variability due to X-chromosome inactivation, and represent the more severe cases of RTT. Furthermore, the more aggressive symptom progression in male mice allows for quicker and more efficient studies, as female mice do not develop overt symptoms until several months of age (Chen et al., 2001; Guy et al., 2001). The use of female mouse models, however, should be incorporated as there are fundamental differences between severe MeCP2 deficiency in male mice and partial MeCP2 deficiency in both female mouse models and RTT patients. 18

27 Figure 3 19

28 Figure 3. Different mouse models of Rett syndrome. MeCP2 expression is knocked out of using different genetic methods. Early lethality is a feature of the male mutant mice. Female mice of the models display normal lifespan and survive well into adulthood. 20

29 1.9. Gross phenotypes of Rett syndrome mouse models The cardinal feature of RTT is the normal development during early life followed by periods of regression and stabilization. Mouse models of RTT recapitulate this feature. Male mouse models, however, develop symptoms significantly earlier than female mice and are more severely (Chen et al., 2001; Guy et al., 2001). Male and female mice of the MeCP2 tm1bird develop RTT-like features at different ages with different rates of symptom progression (Guy et al., 2001). Male MeCP2 tm1bird mice develop symptoms between 4 th and 7 th week of age. The symptoms progress rapidly along with a decline in body weight, which leads to early death at approximately 8 weeks of age. Gait impairment, hind limb clasping, and misalignment of the jaw are commonly observed. In contrast, female MeCP2 tm1bird mice develop symptoms much later, at approximately 12 th weeks of age, and usually survive well into adulthood. Female mice develop milder symptoms, including deficits in inertia and hindlimb clasping. Other features such as respiratory abnormalities and tremor are only present in a portion of the heterozygous females by 9 months of age. The MeCP2 Jae mouse model is very similar to the MeCP2 tm1bird mice (Chen et al., 2001). In male MeCP2 Jae mice, symptoms such as gait impairment can be detected as early as 4 weeks of age. Breathing irregularities, body tremors, and shaking paws are subsequently observed. The body weight of the male mice is decreased compared to wild-type littermates. The heterozygous female mutant mice develop normally up to 4 months of age. Reduced activity, hindlimb clasping, and respiratory deficits are observed after 6 months of age. Female mutant mice also display elevated weight gain. 21

30 In the MeCP2 Tam mouse line, mutant male mice do not survive past 20 weeks of age (Pelka et al., 2006). Gait impairment, dishevelled fur, breathing difficulties, and gait deficits are reported during development. Seizures have also been reported in this model. Female mutant mice show a delayed onset of these symptoms at around 13 weeks of age. Mice expressing the MeCP-308 mutation also recapitulate many RTT-like phenotypes (Shahbazian et al., 2002). Male MeCP 308 mice develop normally until 6 weeks of age. Tremors soon develop and become visibly apparent. Abnormal posture and dishevelling of fur becomes noticeable between 5 and 8 months of age. Myoclonic jerks and seizure events are observed in some mice. Handwringing movement, similar to the stereotypic hand movements in RTT patients, motor deficits, and respiratory dysfunction are reported in mutant MeCP2 308 male mice. Female MeCP2 308 mice develop symptoms at approximately 6 months of age and exhibit milder phenotypes than the mutant male mice Targeted deletion of MeCP2 expression MECP2 is expressed in all tissues but most abundantly in the brain. The brain can be divided into different sectors, each with distinct as well as overlapping functions. Through targeted deletion studies, we gained valuable knowledge as to which brain regions are critical for the pathogenesis in RTT and how different neuronal groups may contribute to different phenotypic impairments. A brain specific deletion of MeCP2 was first produced by Chen et al. This study is consistent with notion that RTT is primarily a disorder of neurodevelopment and MeCP2 function is most critical within the central nervous system. In 2006, Gemelli et al. generated a 22

31 forebrain MeCP2 knockout mouse model through the use of a CamK-Cre93 transgene (Gemelli et al., 2006). These mice developed many RTT-like phenotypes with however, normal life span. Though gross phenotypes were relatively normal in the forebrain knockout (KO) mice, impairments were reported in the rotarod test, anxiety-like behaviour tests (elevated plus-maze and open field test), cue-dependent fear conditioning, and social interaction tests. Taken together, the forebrain likely contributes to a wide spectrum of RTT related symptoms (Gemelli et al., 2006). The basolateral amygdala is known to be central for the regulation of emotion and RTT patients are often reported to exhibit mood swings and anxiety-like behaviour (Sansom et al., 1993). It is hypothesized that MeCP2 mutations disrupt basolateral amygdala function which contributes to the reported symptoms. Adachi et al. selectively ablated MeCP2 expression in the BLA by targeted injection of an adeno-associated virus that express Cre recombinase (Adachi et al., 2009). Consistent with the hypothesis, the targeted MeCP2 KO mice exhibited heightened anxiety-like behaviour and impaired cue-dependent fear learning. Motor and social interaction impairments were not observed. This finding links basolateral amygdala deficits to anxiety and fear learning impairments in RTT, but at the same time, it demonstrates its lack of effect on motor performance and social interaction. Other neuronal populations that may be involved in the pathophysiology of RTT include the catecholaminergic and serotoninergic system (Nomura et al., 1985; Wenk and Hauss- Wegrzyniak, 1999). Clinical features of RTT, such as ataxia, dystonia, and respiratory irregularities, overlaps with disorders involving dopaminergic or noradrenergic deficits (Fahn, 2008; Segawa, 2001). Heightened anxiety, mood swings, and aggression observed in RTT may be attributed to serotonergic abnormalities (Gordon and Hen, 2004; Lucki, 1998; Popova, 2008). 23

32 Samaco et al. selectively ablated MeCP2 function in catecholaminergic and serotonergic cells by crossing transgenic mice that express loxp flanked MECP2 allele with TH-cre and PC12 ets factor 1 (PET1) expressing mouse lines, respectively (Samaco et al., 2008). Tyrosine hydroxylase specific MeCP2 KO mice displayed reduced levels of dopamine and norepinephrine relative to wild-type controls. Reduced locomotive activity was also reported. PET1-specific MeCP2 KO mice displayed reduced levels of serotonin in addition to altered pattern of social interaction and increased aggression. Both KO mouse models did not show early lethality, suggesting that MeCP2 deficiency within these systems is not necessary for normal life span. The findings from this study demonstrate the importance of MeCP2 in the serotonergic and catecholaminergic systems. More generally, the lack of MeCP2 disrupts neuronal systems and causes subsequent alterations in the phenotypes and behaviour which the system normally regulates. Recently it has been shown that MeCP2 is required within the HoxB1-domain as the lack of MeCP2 within the region caused respiratory dysfunction, autonomic irregularities, motor impairment, and premature death (Ward et al., 2011). The HoxB1-specific MeCP2 KO mice displayed impairments in motor coordination tasks, decreased heart rate, and increased respiration rate during hypoxia challenge. The specificity of the deficits confers with the normal function of the brain regions targeted for MeCP2 ablation. Thus the deficits of the HoxB1- domain likely contribute to the motor and autonomic impairments associated with RTT. Excitatory and inhibitory neurotransmission balance is altered in RTT patients (Monteggia and Kavalali, 2009). The GABAergic system is the main source of inhibition in the brain and whether it is compromised by a loss of MeCP2 or if it had any behavioural consequences remained poorly investigated. A mouse line with MeCP2 specifically knocked out 24

33 in GABAergic neurons, viaat-mecp2 -/y, was generated (Chao et al., 2010). These mice develop repetitive behaviours, motor impairment, compulsive grooming, learning deficits, abnormal EEG hyperexcitability, respiratory dysfunction, and premature death. Additionally, mice with MeCP2 deficiency only in forebrain GABAergic neurons develop a subset of the deficits, including altered sensorimotor gating and arousal, impaired motor function, increased sociability, and repetitive behaviours. Quantal size release of GABA is decreased and glutamate decarboxylase 1 and glutamate decarboxylase 2 levels are reduced in MeCP2-deficient GABAergic neurons (Chao et al., 2010). The GABAergic system regulates synaptic signaling throughout the entire brain, thus it is not surprising that MeCP2 deficiency in such a system produces a large array of RTT-like phenotypes. These findings highlight the importance of MeCP2 in GABAergic neurons and that disturbances of this system are central to the pathogenesis of RTT. Though studies of RTT have focused on neuronal alterations, MeCP2 is also expressed, albeit at very low levels, in glia cells (Ballas et al., 2009). MeCP2-deficient glia exerts negative non-cell autonomous effects on neuronal properties, including dendritic morphology. Specifically, astrocytes lacking MeCP2 expression display abnormalities in BDNF and neuronal dendritic regulation (Maezawa et al., 2009). By crossing mice expressing a floxed MECP2 allele with mice expressing hgfapcret2 transgene, Lioy et al. (2011) generated mice that have MeCP2 selectively ablated from astrocytes. These mutant mice develop many RTT-like phenotypes such as breathing irregularities, hindlimb clasping, and decreased body weight. However, lifespan, locomotive activity, and anxiety-like behaviour are not affected. The findings of this study revealed a non-neuronal effect in the pathogenesis of RTT. It remains to be investigated whether MeCP2 is required for proper function in other cell types and whether these deficits contribute to the pathophysiology of RTT. 25

34 All of these targeted knockout studies have demonstrated the cell and tissue-autonomous function for MeCP2 within different brain regions. Local MeCP2 deficiency only produced deficits that are normally related to the affected region. These deletion studies are complemented with reactivation/rescue studies that further describe how different neural systems contribute to the various symptoms of RTT Reversibility of deficits in mouse models of Rett syndrome An important finding in RTT patients and mouse models of RTT is the lack of neuronal death and apoptosis (Johnston et al., 1995). Rather the neurons appear to be preserved in an immature state. These reports lend to the hypothesis that the phenotypes of RTT may be reversible (Guy et al., 2007). As RTT is mainly caused by mutations of the MECP2 gene, it may be possible to achieve phenotypic rescue by re-introduction of the MeCP2 function. In 2007, Adrian Bird s group generated a global inducible rescue model of RTT (Guy et al., 2007). This rescue mouse model was generated by crossing the MeCP2 TM2bird mouse line with a mouse line that expressed the ESRcre construct, which encodes Cre recombinase enzyme fused with a mutated ligand binding domain of the estrogen receptor). Tamoxifen (an estrogen receptor antagonist) treatment allowed the translocation of Cre recombinase into the nucleus, which in turn reactivated MeCP2 expression through excision of the floxed-stop cassette. Guy et al. (2007) reactivated MeCP2 expression in the double transgenic rescue mice after the development of symptoms. MeCP2 reactivation significantly improved life span and gross phenotypic severity in the male rescue mice and reversed the gross phenotypes of the female rescue mice. In addition, long-term potentiation was also significantly improved in the female rescue mice. This study was 26

35 the first to demonstrate the reversibility of RTT and that the alterations of the neural networks are not irremediable (Guy et al., 2007). Rather, the systems may be preserved in an immature state that still consisted of the necessary machinery for further development. Robinson et al. (2012) further extended upon this study by demonstrating that specific sensorimotor performances and neuronal morphology are also improved in the global MeCP2 rescue. However, these studies focused on male MeCP2-deficient mice and the extent of phenotypic rescue in the female mutant mice remains unclear. Following the global rescue studies, several groups have generated targeted rescue models of RTT. Jugloff et al. (2008) reintroduced MeCP2 into the forebrain of MeCP2 deficient female mice. Exploratory and rearing activity was significantly improved. Further, reactivation of MeCP2 expression in glia cells partially rescued behaviour deficits, breathing abnormalities, and neural anatomical abnormalities (Lioy et al., 2011). This was an important finding as it demonstrates that MeCP2 is critical for non-neuronal cells and reactivation of MeCP2 expression within astrocytes can benefit the neuronal network and improve several phenotypes of RTT. Lastly, reactivation of MeCP2 in the HoxB1 domain led to significant autonomic improvements, and most importantly, rescued early lethality (Ward et al., 2011). These genetic rescue mouse models of RTT complement the knockout studies and further demonstrate the role of each brain region in the pathophysiology of RTT. The symptoms of RTT are various and are often affected differentially. Understanding the neural origin of the specific symptoms will allow for more selective treatment strategies tailored to individual patients. 27

36 1.12. Gene therapy Genetic rescue studies in mouse models of RTT suggest gene therapy as a viable option (Cobb et al., 2010). The neuronal network seems to remain intact even in the absence of MeCP2. The methylated DNA targets of MeCP2 are suggested to be preserved, therefore as newly synthesized MeCP2 become present, gene transcription should return to normal (Lewis et al., 1992; Nan et al., 1997). Further, studies have shown that overexpression of MeCP2 leads to the development of motor impairments and tremors, whereas modest overexpression of MeCP2 levels enhances motor performance and reduces anxiety-like behaviour (Collins et al., 2004). The dosage at which MeCP2 is reintroduced becomes critically important. Distinct brain regions have been shown to contribute to different symptoms of RTT, and selective rescue of specific neuronal systems can improve the deficits associated with the network. Genetically targeting specific brain regions is a potential avenue for clinical investigation Pharmacological treatments As current gene therapy methods cannot provide the level of control and specificity required for MECP2 reintroduction, many are seeking alternative pharmacological methods of treatment. Several studies have identified molecules and systems that are affected downstream of MeCP2 function. However, improving downstream pathways is unlikely to completely compensate for the absence of MeCP2. Several candidates for therapeutic treatments are suggested. Brain derived neurotrophic factor (BDNF) is essential for neural development and its levels are significantly decreased in RTT brains (Acheson et al., 1995; Chen et al., 2003; 28

37 Martinowich et al., 2003). Mouse studies support pharmacological methods of enhancing BDNF activity in RTT as increased BDNF level positively affects MeCP2-null mice by delaying onset of symptoms, improving life span, and rescuing electrophysiological abnormalities (Chang et al., 2006). Ogier et al. (2007) tested ampakine CX546, an enhancer of glutamergic AMPA receptor which in turn increases BDNF levels, in MeCP2-null mice. Respiratory function was restored in the mouse, however, other deficits remains unexplored. Another candidate for therapeutic treatment of RTT is the insulin-like growth factor 1. IGF1 is known to be involved in neuronal maturation and synaptic development (Gadalla et al., 2011). Itoh et al. (2007) found elevated levels of IGF3 in RTT patients and MeCP2-null mice. It has been hypothesized that the increase is a consequence of IGF1 reduction. Tropea et al. (2009) investigated whether systemic administration of IGF1 can improve RTT-associated deficits. Indeed, cortical spine density and excitatory current amplitudes are significantly improved in MeCP2-null mice after IGF1 treatment. Life span, locomotion activity, and autonomic functions are improved as well. The severe and potentially fatal respiratory abnormalities of RTT are hypothesized to be associated with noradrenaline deficits (Brucke et al., 1987; Lekman et al., 1989; Nomura et al., 1985). Reduced bioamine levels are reported in both RTT brains and MeCP2-null mice (Samaco et al., 2008). Further, immunohistochemistry studies have shown a significant reduction in the number of tyrosine hydroxylase expressing neurons within the medulla, where respiratory centers are located (Viemari et al., 2005). Following this logic, desipramine, a noradrenaline reuptake inhibitor, was tested in MeCP2-null mice to test whether respiratory abnormalities can be improved (Roux et al., 2007). The treated mice showed a delayed onset of breathing 29

38 abnormalities and life span was improved as well. The rescue was not complete, however, as only a small system was targeted in a wide array of impaired neural networks. Lastly, the cholinergic system is also altered in RTT brains (Gadalla et al., 2011). Studies have shown a reduction of choline acetyltransferase and vesicular transporter binding (Wenk and Mobley, 1996). To test the beneficial effects of improving the cholinergic system in RTT, MeCP2-null mice were provided with choline enriched diet during early development (Nag and Berger-Sweeney, 2007). The effects were subtle as there was only a modest improvement in locomotion activity and motor function. These various pharmacological interventions showed the beneficial effects of targeting different systems in RTT. However, most of the improvements were modest and only a subset of RTT symptoms is improved. As mentioned previously, the most effective treatment would require direct improvement of MeCP2 function. However, treatment methods are limited by our incomplete understanding of MeCP2 function in RTT. 30

39 1.14. Rationale and hypothesis Project 1: Delayed ubiquitous reactivation of MeCP2 Many clinical signs of RTT are recapitulated in MeCP2 deficient mouse models (Guy et al., 2001). MeCP2-deficient male mice develop severe symptoms and have significantly shortened lifespan (Guy et al. 2007). In contrast, MeCP2 deficient female mice resemble clinical RTT progression as they show delayed onset of symptoms at 4-12 months of age (Robinson et al 2012). Phenotypically, male MeCP2 deficient mice show severe behaviour impairments in ambulation, motor coordination, anxiety, and social interaction in addition severe respiratory abnormalities (Samaco et al 2009; Ward et al 2011; Wither et al 2012). At a more cellular and anatomical level, neuron electrical properties, dendritic branching, and synaptic transmission are also impaired in MeCP2 deficient male mice. Milder phenotypes are observed in female MeCP2 deficient mice due to random X-chromosome inactivation profiles. Hemizygous male mouse models of RTT do not exhibit the phenotypic variability as female models and thus, are an efficient model for RTT investigations. However, heterozygous female mutants are more clinically relevant and should be incorporated into investigations to demonstrate the clinically applicability of different treatment methods. Deficits caused by a lack of MeCP2 are not irreversible as reactivation of MeCP2 expression in 3-4 week old male MeCP2-deficient mice rescued early lethality, gross phenotypic severity, behavioural, and neural morphological deficits (Guy et al, 2007; Robinson et al 2012). These previous studies reactivated MeCP2 to approximately 70% of wild-type levels. As work from our lab have recently shown a correlation between MeCP2 levels and behavioural performances in female MeCP2-deficient mice, the extent of improvements in MeCP2-deficient 31

40 male mice may correlate with MeCP2 reactivation efficiency. In addition, little is known about the potential remedial effects of restoring MeCP2 to female RTT mouse models. Only improvements of gross phenotypic severity and LTP have been reported in female mice following MeCP2 reactivation (Guy et al., 2007). The full extent of rescue achieved by MeCP2 reactivation remains unclear. Epileptiform discharge events, impaired daily rhythmic activity, and reduced core body temperature are reported in MeCP2-deficient mice and recapitulate the clinical conditions (D Cruz et al., 2010; Wither et al., 2012). Given that seizures, abnormal biological rhythms, and poor thermoregulation are severe hindrance to RTT girls in everyday life, it is important to determine the reversibility of these symptoms. Here, we tested whether MeCP2 reactivation in the same rescue mouse model described in Guy et al. (2007) can reverse these deficits. Furthermore, we investigated whether MeCP2 reactivation efficiencies correlate with the extent of improvements in MeCP2-deficient male mice. We found that reactivation of MeCP2 post symptom onset leads to significant improvements in sensorimotor tasks, electroencephalography abnormalities, core body temperature, and daily rhythmic activity. In addition, we observed that greater MeCP2 reactivation levels resulted in more pronounced improvements. 32

41 Project 2: Preservation of MeCP2 function in catecholaminergic cells Mutations within the X-linked gene encoding methyl-cpg-binding protein 2 (MeCP2), have been identified as the predominate causes of RTT (Amir et al., 1999). Cardinal RTT phenotypes are recapitulated in both male and female MeCP2 deficient mice (Chen et al., 2001; Guy et al., 2001). By utilizing these models, studies have shown that that global reintroduction of MeCP2 (Guy et al., 2007, Robinson et al., 2012) and targeted reintroduction of MeCP2 into specific populations of neurons or glia (Alvereez-Savaadra et al., 2007; Jugloff et al., 2008; Giacommetti et al., 2005, Ward et al., 2011; Lioy et al., 2011) can improve the behavioural deficits of MeCP2-deficient mice. Collectively, these studies show that the RTT-like phenotype of MeCP2-deficient mice is not irreversible, and raises the possibility that gene reintroduction strategies may have clinical potential. While encouraging, repopulating large regions of the brain with MeCP2 remains a challenging prospect clinically. Thus the next logical step would be to determine whether preserving MeCP2 function within small populations of defined neurons would be sufficient to improve the deficits associated with RTT. The catecholaminergic system has been strongly implicated in the pathophysiology of RTT (Nomura et al., 1985; Zoghbi et al., 1989; Viemari et al., 2005). The majority of norepinephrine projections within the brain originate from neurons residing within the locus ceruleus and lateral tegmental area (Hokfelt et al., 1984), while the majority of dopamine projections arise from neurons within the ventral tegmental area, arcuate nucleus, or substantia nigra (Bjorklund et al., 2007). These regions are well defined anatomically, and while comprised of relatively small numbers of neurons, their functions influence the activity of numerous cell types and neural systems throughout the brain. Many RTT deficits and symptoms are consistent with phenotypes caused by abnormalities in catecholaminergic system. Motor deficits and 33

42 rigidity could be associated with dopaminergic dysfunction and autonomic abnormalities can be attributed to adrenergic impairments (Fahns et al 2008; Tanaka et al 2000). Consistent with this idea, decreased levels of dopamine and norepinephrine have been previous reported in RTT patients as well as MeCP2-null animals (Samaco et al, 2009). The importance of MeCP2 function in catecholaminergic cells has been recently demonstrated, as the selective ablation of MeCP2 from the catecholaminergic system induced RTT-like phenotypic impairments in mice (Samaco et al., 2009). Importantly, no apoptotic neurons are observed within catecholaminergic systems (Roux et al., 2010). Since the targeted deletion of MeCP2 function in catecholaminergic cells can produce RTT-like deficits, we hypothesized that selective preservation of MeCP2 function in catecholaminergic cells of MeCP2 deficient mice will lead to retention of normal phenotypic behaviours. To test this, we selectively preserved MeCP2 function in TH-positive cells of a murine RTT mouse model. Our results show that the preservation of MeCP2 function in this population of cells is sufficient to improve phenotypic deficits in both male and female MeCP2-deficient mice. 34

43 1.15. Project Aims Project 1: to determine whether the delayed ubiquitous reactivation of MeCP2 can improve - behaviour deficits - epileptiform-like discharge events - circadian activity - core body temperature Project 2: to determine whether selective preservation of MeCP2 expression in catecholaminergic cells can improve - gross phenotypic severity - behavioural performances - electroencephalography abnormalities 35

44 Figure 4 36

45 Figure 4. Different transgenes expressed in different mouse lines. Panel A: MeCP2 flox-stop transgene is inserted in the Stop/y or Non-rescue mouse line. In the presence of cre recombinase, the floxed-stop cassette is excised and MECP2 transcription is reestablished. Panel B: THcre transgene is expressed in the TH-cre mouse line. Cre recombinase is only expressed in tyrosine hydroxylase positive cells as it is driven by a TH-specific promoter. Panel C: Rosa26-ESRcre transgene expressed in the Rosa-cre mouse line. Cre recombinase, in this case, is normally sequestered in the cytoplasm. When tamoxifen (an estrogen antagonist) binds to the estrogren receptor, cre recombinase is allowed to enter the nucleus and target loxp flanked sites. 37

46 Chapter 2 Materials and Methods 2.1 Mice. All animal procedures and protocols were approved by the Canadian Council on Animal Care and Toronto Western Research Institute s Animal Facility. Mice were housed in a controlled facility on a 12 hour light/dark cycle and were provided with food and water ad libitum. CreESR and MeCP2 stop/+ mice were obtained from The Jackson Laboratory and maintained on a C57Bl/6 background. MeCP2 stop/+ female mice, which is heterozygous for MeCP2 allele silenced by a neo-stop cassette, is crossed with male CreESR to obtain male MeCP2 stop/y ESRcre and female MeCP2 stop/+ ESRcre mice. TH-cre mice were obtained as a gift from Dr. Joseph Savitt (Johns Hopkins University) and maintained on a pure C57Bl/6 background. Mecp2 tm2bird mice (Guy et al., 2007) were obtained from The Jackson Laboratory (Bar Harbor, Maine) and also maintained on a pure C57Bl/6 background. TH-Cre male mice were crossed with Mecp tm2bird female mice to generate experimental subjects of both genders. Polymerase chain reaction (PCR) was used to identify the genotype of these mice. DNA samples were prepared using the HotSHOT genomic DNA preparation method (Truett et al., 2000) on tissues collected from ear punches. The floxed-stop sequence is identified using the primer set: 5 -CTTCAGTGACAACGTCGAGC and 5 - CATTCTGCACGCTTCAAAAG. The presence of cre recombinase sequence was identified using the primer set: 5 -AAATGTTGCTGCTGGATAGTTTTTACTGC and 5 - GGAAGGTGTCCAATTTACTGACCGTA. 38

47 2.2 Western blotting. Animals were anesthetized under isoflurane and sacrificed for tissue collection. Brain samples were collected on ice and snap frozen in dry ice. Quarter brains were homogenized in approximately 300µl of RIPA buffer (50 mm Tris-HCL[pH 7.5], 150 mm NaCl, 1% NP40, 2 mm EDTA, 0.5% sodium deoxycholate, and 0.1% SDS) supplemented with a mixture of protease inhibitors (PMSF 40 ng/ml, Antipain 2 ng/ml, PepstatinA 2 ng/ml, Leupeptin 20 ng/ml, Aprotinin 20 ng/ml, and MDL ng/ml). Samples were then spun down at 12,000g for 5 minutes and the supernatant was collected. Samples were stored at -80 C until used. Protein concentrations of samples were determined using the Folin method (Bio-Rad, Hercules, CA, Cat # ). Protein samples were resolved by sodium dodecyl sulphate polyacrylamide gel electrophoresis and transferred to a polyvinylidene fluoride membrane in standard transfer buffer (25 mm TRIS, 192 mm glycine, 20% methanol) overnight at 4 C and pre-hybridized for 2 hours at room temperature in blocking solution (TRIS-buffered saline containing 0.05% Tween-20 (v/v) (TBST) and 5% (w/v) non-fat dry milk). Membranes were hybridized overnight at 4 C with primary antibody (1/1000; Cell Signaling Technology, Danvers, MA, Cat # 3456S) in blocking solution. After washes in TBST, membranes were hybridized with HRP-linked secondary antibodies (1/5000; GE Healthcare, Buckinghamshire, England, Cat # NA934 [anti-rabbit], Cat # NA931 [anti-mouse]) for 2 hours at room temperature. After extensive washing in TBST, immunoreactivity was visualized by enhanced chemiluminescence (GE Healthcare, Cat # RPN2106). 39

48 2.3 Immunohistochemistry. Animals were anesthetized through inhalation of 2% isoflurane, and transcardially perfused with 0.9% NaCl saline solution followed by perfusion with an ice-cold 2% paraformaldehyde-pbs solution. Intact brains were dissected from the skull and equilibrated overnight in a 30% sucrose/pbs solution at 4 o C. The brain was then stored at -80 o C until further assessment. For sectioning, the brain was cut at the midline, and one hemisphere was embedded in O.C.T. compound (Sakura, Torrance, California) and coronal sections (15 µm) were collected with a Leica cryostat (model Jung CM 3000, Wetzlar, Germany) at -24 o C. Sections were blocked with 10% NGS + 2% BSA in 0.1% PBS-T for 1 hour, and then incubated with rabbit anti- MeCP2 (1:500 Cell Signaling, #3456S) antibodies and/or mouse anti-tyrosine hydroxylase (1:1000, Millipore, #NG ) in 0.1% PBS-T supplemented with 2% normal goat serum overnight at 4 o C. The sections were then washed using 0.1% PBS-T at room temperature 3 times, and incubated with secondary antibodies conjugated to either DyLight 488 (Invitrogen, goat antimouse, #A11001) and/or DyLight 568 (Invitrogen, goat anti-rabbit, #A11011) for 1 hour at room temperature. Following incubation, the sections were washed with PBS, and then incubated briefly with DAPI (5 µg/ml, Roche Diagnostics, Indianapolis, Indiana # ) for 3 min. Sections were then rinsed with PBS, and mounted atop slides with Dako Fluorescent Mounting Media (Burlington, Ontario, Canada, #S302380). Imaging was done using a Zeiss Axioplan 2 deconvolution microscope (Carl Zeiss, Göttingen, Germany). 2.4 Tamoxifen treatment. Tamoxifen (Sigma) was dissolved in corn oil (6mg/ml) and stored at 4 C until use. Tamoxifen was administered to mice through peritoneal injection for five consecutive days at 100mg/kg or once per week for three weeks at 33mg/kg followed by two booster injections at 67mg/kg. Tamoxifen injections in male Stop/ESRcre mice were delayed 40

49 until phenotypic symptoms are fully developed (~50-60 days of age with a phenotypic score of at least 5). Phenotypic severities of female Stop/ESRcre mice were scored weekly and tamoxifen injections commenced when the mice reached a score of 5 and are at least 250 days of age. Wildtype, MeCP2 stop/y, and MeCP2 -/+ controls were treated with tamoxifen in the same manner. Age matched Rosa male and female mice were also injected with tamoxifen to ensure no confounding effects from the treatment. 2.5 Electrophysiology data collection. 24 hour activity, body temperature, and EEG were collected from telemetrically implant female mice as described previously in Wither et al. (2012). Briefly, the TA11ETA-F10 telemetry probe and a wireless receiver (RPC-1, DSI) were used to collect waveform data. Locomotion activity was collected by calculating the standard deviation of the transmitter signal strength relative to two perpendicularly arranged antennae in the RPC-1 wireless receiver. Body temperature was collected from an internal thermosensor in the TA11ETA-F10 transmitter. Body temperature and locomotion activity data were transmitted at a rate of 50 Hz, using a sampling frequency of 250 Hz. EEG waveform data was transmitted at 200 Hz and sampled at 1 khz. (Implantations were done by Chiping Wu) 2.6 Behavioural Assessments. Animals were assessed in the open field and accelerating rotarod test as described previously (Jugloff et al., 2008). For the open field ambulation test, subjects were placed in a Plexiglass cage (20 x 30 cm 2 ) and an automated movement detection system (AM1053 activity monitors; Linton Instruments, United Kingdom) was used to record the motor activities of the animals for a one-hour period. For the accelerating rotarod test, subjects were placed on a rotating rod (MED Associates Inc., #ENV-575M, St. Albans, Vermont) that 41

50 accelerates linearly from 3.5 rpm to 35 rpm over a 5-minute period. The time at which the subject fell from the rotating rod was recorded via a laser beam sensor. Each subject was assessed on the accelerating rotarod three times a day for four consecutive days. Consecutive trials were separated by at least one hour to allow the animals to recover from physical fatigue. Animals that circumnavigated the rod for three consecutive times were scored as having fallen off the apparatus upon the third rotation. For the light-dark placement preference test, mice were placed into a box consisting of a dark compartment (20 x 14 cm) and a light compartment (20 x 28 cm) connected through a single small opening (4 cm 2 ). The amount of time the animal spent in the dark and light compartments, as well as the number of risk assessment behaviours (head pokes out of the dark compartment) each subject took while in the dark compartment, were recorded from videotaped 10 minute sessions. For the nest-building behaviour test, subjects were placed into a new cage containing a single piece of nestlet, and the volume of each assembled nest was calculated the following day. All behaviour tests were conducted between 9:00am and 13:00pm to minimize circadian effects. Female subjects were assayed after 280 days of age, and male animal subjects were assessed between 50 and 70 days of age. 2.7 Phenotypic severity scoring. Animals were scored using the deficit scoring system described previously (Cheval et al., 2012; Guy et al., 2007; Robinson et al., 2012). In short, mice were scored from 0-2 according to the following scheme: Mobility score: 0 =same as wild-type; 1 = slower movements than wild-type with intermittent freezing periods; 2 = severely reduce or no movement at all. Gait score: 0 = same as wild-type; 1 = hind-limbs are spread wider than wildtypes and slips or double tapping of the same feet is observed; 2 = dragging of hind-limbs or constant slips. Hind-limb clasp score: 0 = hind-limbs are spread out when lifted by the tail, same 42

51 as wild-type; 1 = one hind-limb is pulled towards the body and clasped or both hind-limbs are mildly pulled inwards towards the body; 2 = both hind-limbs are clasped and pulled tightly into the body. Breathing score: 0 = no noticeable respiratory abnormalities, same as wild-type; 1 = occasional respiratory jerks or some irregularity in breathing patterns; 2 = severe respiratory difficulties involving strong body jerks. Tremor score: 0 = no tremor; 1 = intermittent mild tremors; 2 = severe and constant tremors. General condition score: 0 = well-groomed and shiny fur, opened eyes, normal body posture; 1 = slightly dishevelled fur, squinty eyes, and slightly hunched posture; 2 = extremely dishevelled fur, eyes closed, and severely hunched posture. 2.8 Cell Counting. MeCP2 and tyrosine hydroxylase expression was determined through random sampling of every 5 th section in brain slices containing substantia nigra or locus ceruleus. Substantia nigra was identified from Bregma -2.48mm to mm, and locus ceruleus was identified from Bregma mm to mm. MeCP2 expression was assessed only in clearly identified nuclei that displayed DAPI staining. MeCP2 antibody specificity was confirmed by comparing the staining patterns of wild-type and MeCP2 -/y animals. Counts were made for cells displaying either tyrosine hydroxylase or MeCP2 immunoreactivity alone, and for cells expressing both tyrosine hydroxylase and MeCP2. Cell counts were conducted by two independent examiners blinded to condition, and their individual counts averaged for analysis. 2.9 Telemetry Probe Implantation Protocol. Female mice were implanted with a wireless telemetry probe TA11ETA-F10; Data Sciences International (DSI), St. Paul, MN) for long duration electroencephalogram (EEG) and activity recordings as described in (Wither et al., 2012). Animals were allowed to recover for at least 3 weeks prior to any data collection. 43

52 2.10 Tethered Electrode Implantation. Animals were implanted with electrode cap assemblies as described previously (Wu et al., 2008). Briefly, animals were anesthetized under 2-4% isoflurane through inhalation. Electrodes made from polyimide-insulated stainless steel were implanted in the hippocampal CA1 (Bregma, -2.3 mm; lateral, 1.7 mm; depth, 2.0 mm) and contralateral somatosensory cortex (Bregma, -0.8 mm; lateral, 1.8 mm; depth, 1.5 mm). A reference electrode was implanted in the frontal cortex (Bregma, -3.8 mm; lateral 1.8 mm; depth, 1.5 mm). Male mice were implanted between ages 40 and 60 days of age, during which the symptoms begin to. Female mice were implanted post 250 days of age, and after symptom onset. The implanted mice were allowed to recover for at 7 days before any further experiments. Baytril antibiotics (Bayer Healthcare, Toronto, Ontario) were added to the water supply 2 days before surgery and 7 days after surgery to minimize infections. (All implantations were done by Chiping Wu) 2.11 Electroencephalographic recordings and analysis. EEG recordings were collected as described previously (D Cruz et al., 2010). Briefly, the implanted electrodes were connected to two independent head stages (Model-300, AM Systems Inc., Carlsborg, Washington). EEG signals were amplified 1000x, bandpass filtered ( Hz), and digitized (Digidata 1300, Axon Instruments, Weatherford, Texas). EEG data were collected at 60 khz and analyzed using Clampfit software (Axon Instruments). Recording sessions were at least 2 hours in duration, and each subject was recorded for a minimum of two sessions on different days. The EEG recordings were decimated 10x via Clampfit 10.2 software before analysis. Epileptiform discharge-like events were counted manually using the following criteria: frequency between 6-12Hz, minimum duration of 0.5 seconds, and at least 1.5x the baseline amplitude, and high rhythmicity. 44

53 Hippocampal theta epochs during exploratory behaviours were bandpass filtered ( Hz), and spectral plots (50% window overlap and frequency resolution of 0.25 Hz) were generated. A minimum of 10 epochs from at least two recording sessions were averaged to obtain peak theta frequency, total theta power, and total gamma power for each animal. The frequency between 6-12 Hz with the greatest power was taken as the peak theta. Total theta wave power was calculated by taking the area underneath the spectral plot between 6-12 Hz. Similarly, total gamma wave activity is taken as the area underneath the spectral plot between 35 and 60 Hz. All EEG data were calculated and analyzed using Clampfit 10.2 software Statistics. All statistical analysis was performed using PRISM or Microsoft Excel software on a PC. Paired student s t-test was used to compare before and after Stop/+,cre and Stop/y,cre mice. Two-way ANOVA with Bonferroni post hoc correction was used to for multiple group comparison with that had time as a factor. Pearson s product moment correlation was used to compare correlative strength between two groups. All behavioural data were analyzed using oneway ANOVA with Tukey s post hoc comparison, or two-way ANOVA with Bonferroni s posthoc test. Cell counts as well as MeCP2 and tyrosine hydroxylase expression levels were analysed using unpaired student s t-test. Kaplan-Meier survival plot was analyzed using Wilcoxon rank sum tests. Spontaneous death rate was compared using population Chi-squared tests with one degree of freedom. For all cases, the threshold for statistical significance was set at p<

54 Chapter 3 Results Rescue of MeCP2 expression in Stop/y,cre mice. As described in Guy et al. (2007), the MeCP2 Tm2bird mice have disrupted MeCP2 expression as a neo-stop cassette is inserted between exon 2 and exon 3. Consistent with previous reports, we confirmed that the MeCP2 lox-stop allele behaves similar to a null mutation. Stop/y mice expressed MeCP2 at ~1.5-3% of wild-type levels (Figure 5A). By crossing MeCP2+/- mice with mice that expressed ESRcre, we generated rescue mice that allow MeCP2 expression to be reinstated through tamoxifen injections. To determine whether the level of MeCP2 reactivation correlates to the extent of rescue, we employed two tamoxifen injection paradigms. The low reactivation cohort of Stop/y,cre mice received tamoxifen injections once a week at a dosage of 50mg/kg for three weeks, followed by two consecutive booster injections of 100mg/kg (low TMX). The high reactivation cohort of Stop/y,cre mice received five consecutive daily injections tamoxifen at the dosage of 100mg/kg (high TMX). The low reactivation cohort (Stop/y,cre + low TMX) express MeCP2 at 21 +/- 6% of wild-type levels, whereas the high reactivation cohort express MeCP2 at 65 +/- 18% of wild-type levels after tamoxifen injections (Figure 5C). 46

55 Figure 5 47

56 Figure 5. MeCP2 is reactivated in Stop,cre mice following tamoxifen treatment. Panel A: Western blot of whole brain samples from a wild-type and Stop/y mice. Panel B: Western blot of whole brain samples from wild-type (n=3), Stop/y,cre + low TMX (n=3), and Stop/y,cre + high TMX (n=3). Panel C: Level of MeCP2 expression in Stop/y,cre + low TMX (n=3) and Stop/y,cre + high TMX (n=3) mice as a percentage of male wild-type mice (n=3). 48

57 Restoration of MeCP2 rescues life span and gross phenotypic severity. As MeCP2 is essentially silenced in Stop/y mice, early lethality is observed (Figure 6A). The median survival age of Stop/y mice is 77 days. In contrast, survival is significantly improved in Stop/y,cre + low TMX mice compared to Stop/y mice (median survival age of 255 days, p<0.01, Figure 6A). Stop/y,cre + high TMX mice displayed a median survival age of 320 days, which is significantly improved relative to both Stop/y and Stop/y,cre + low TMX mice (p<0.01 and p<0.05 respectively, Figure 6A). Consistent with the improved life span, phenotypic severity scores of both the high and low TMX treated Stop/y,cre mice were significantly lower than Stop/y mice, which showed continuous progression of severity until early death (Figure 6B). In addition, Stop/y,cre + high TMX mice scored on average significantly lower than Stop/y,cre + low TMX mice between 4-23 weeks of age. 49

58 Figure 6 50

59 Figure 6. Survival and gross phenotypic severity is significantly improved in Stop/y,cre mice after tamoxifen treatment. Panel A: Kaplan-Meier survival plot of male Stop/y mice (n=29), Stop/y,cre + low TMX (n=10), and Stop/y,cre + high TMX (n=9). The life span of the Stop/y,cre mice after tamoxifen treatment is significantly longer than Stop/y mice (p<0.01, Wilcoxon ranksum test). The life span of Stop/y,cre + low TMX and Stop/y,cre + high TMX is not significantly different. Panel B: the gross phenotypic severity score of Stop/y mice (n=13), Stop/y,cre + low TMX (n=6), Stop/y,cre + high TMX (n=5). Stop/y,cre + low TMX scored significantly lower than Stop/y mice starting at and after 4 weeks of age (one-way ANOVA with Tukey s post-hoc correction). Stop/y,cre + low TMX scored significantly lower than Stop/y at and after 5 weeks of age, and significantly higher than Stop/y,cre + high TMX between 5 and 20 weeks of age (oneway ANOVA with Tukey s post-hoc correction). 51

60 Extent of behavioural rescue is dependent upon the degree of MeCP2 reactivation. We further examined the reversibility of behavioural impairments by assessing the mice in specific sensory-motor tests open field ambulation test, accelerating rotarod test, light/dark preference test, and nest building test (Figure 7). Stop/y mice were assayed between days of age and the rescue mice are assayed post 100 days of age and at least two months after tamoxifen treatment to avoid confounding effects. As expected, the behavioural performances of Stop/y mice are significantly impaired compared to wild-types in all tests. We further tested low TMX treated and high TMX treated Stop/y,cre mice in the same behavioural tests and found that high TMX treated Stop/y,cre mice displayed greater levels of behaviour rescue than low TMX treated Stop/y,cre mice. In the open field test, activity and rearing counts were significantly improved in Stop/y,cre + low TMX compared to Stop/y (one way ANOVA, p<0.05), whereas ambulation mobility was not improved (one way ANOVA, p>0.05 for each) (Figure 7A). High TMX treated Stop/y,cre mice s performances were significantly improved in every parameter compared to Stop/y (one way ANOVA, p<0.01 for each). Additionally, high TMX treated Stop/y,cre mice displayed significantly more activity and rear counts than low TMX-treated Stop/y,cre mice (one way ANOVA, p<0.01). Motor deficits are cardinal features of RTT and we assayed whether MeCP2 reactivation rescues these impairments by testing the rescue mice on the accelerating rotarod test (Figure 7B). Stop/y mice showed severe motor impairment as their initial latency to fall and their final day latency to fall was 101 +/- 9 seconds and 158 +/- 11 seconds, compared to wild-types 196 +/- 7 and 295 +/- 4 seconds of initial and final latency to fall (two-way ANOVA, p<0.001). Although low TMX treated Stop/y,cre mice remained significantly below wild-type performance levels, they were able to last significantly longer on the rotarod than Stop/y mice on trial days 1,2, and 4 (p<0.05, two-way ANOVA). The 52

61 performances of high TMX treated Stop/y,cre were significantly better than Stop/y mice on all days of trial and not significantly different than wild-type levels (p<0.001, p>0.05, respectively; two-way ANOVA). In the light and dark place preference test, the number of head pokes performed by the mouse while in the dark compartment was taken as an indicator of risk assessment/anxiety-like behaviour. Only high TMX treated Stop/y,cre mice, not low TMX treated, performed significantly more risk assessments per time spent in the dark compartment than Stop/y mice (one way ANVOA, p<0.05) (Figure 7C). Sociability was tested using the nest building behaviour test by measuring the volume of the nest size 24 hours after introducing the mouse to the new cage. Similar to the light/dark preference test, only high TMX treated, Stop/y,cre mice showed a significant improvement in the average nest volume built (high TMX treated versus low TMX treated Stop/y,cre mice versus Stop/y: / cm 3 versus 9.4 +/- 2.5 cm 3 versus 3.5 +/- 0.3 cm 3 ; one way ANOVA, p<0.05 for high TMX treated Stop/y,cre mice versus Stop/y). 53

62 Figure 7 54

63 Figure 7. Behavioural performances were improved in Stop/y,cre mice following tamoxifen treatment. Panel A: Histogram showing the mean and SEM of male Stop/y mice (n=17), Stop/y,cre + low TMX (n=9), and Stop/y,cre + high TMX (n=7) in the open field test. The behavioural performances of the mice are normalized to mean wild-type levels. Stop/y,cre + high TMX mice performed significantly better than Stop/y and Stop/y,cre + low TMX in all parameters (p<0.01, two-way ANOVA with Bonferroni s post-hoc correction). Stop/y,cre + low TMX displayed great total activity and rearing counts than Stop/y mice (p<0.05, two-way ANOVA with Bonferroni s post-hoc correction). Panel B: Motor performance of Stop/y mice (n=27), Stop/y,cre + low TMX (n=11), Stop/y,cre + high TMX (n=9) on the accelerating rotarod. Stop/y,cre + high TMX mice displayed a longer latency to fall than Stop/y mice on all trial days (p<0.01, two-way ANOVA with Bonferroni s post-hoc correction). Stop/y,cre + low TMX displayed a longer latency to fall than Stop/y mice on trial days 1, 2, and 4. No significant difference was detected between Stop/y,cre mice with low TMX and high TMX. Panel C: Anxiety-like behaviour was assessed using the light-dark place preference test. The histogram shows the mean and SEM of the number of risk assessments (head pokes) performed by the different cohort of mice per minute of time spent in the dark chamber. Stop/y,cre + high TMX (n=6) performed significantly more head pokes than Stop/y mice (n=15). Panel D: Social behaviour was assessed using the nest building test. The histogram shows the mean and SEM of the nest volume built by the different cohort of mice 24 hours being introduced into the test cage. Stop/y,cre + high TMX (n=6) built larger nests than Stop/y mice (n=13) (p<0.05, one-way ANOVA with Tukey s post-hoc correction), but still smaller than male wild-type mice (n=14) (p<0.05, one-way ANOVA with Tukey s post-hoc correction). Stop/y,cre + low TMX mice (n=6) did not assemble nests that were significantly larger than Stop/y mice (p>0.05, one-way ANOVA 55

64 with Tukey s post-hoc correction). For panel A, C, and D, * indicates p<0.05. For panel B, * indicates p<0.05 compared between Stop/y,cre + high TMX and Stop/y. # indicate p<0.05 compared between Stop/y,cre + low TMX and Stop/y. 56

65 Epileptiform discharges are significantly attenuated after MeCP2 reactivation Epileptiform-like discharge events were reported in MeCP2 deficient mice (D Cruz et al., 2010, Wither et al., 2012). Here, we observed these events in all of the Stop/y mice after 50 days of age. On average, Stop/y mice exhibited 42 +/- 7.8 discharge events per hour with an average duration of 2.8 +/ seconds per event (Figure 8A,D,E). Similar epileptiform-like discharge events were observed in tamoxifen-treated Stop/y,cre mice. The average duration of individual discharge events, however, is significantly reduced in both low TMX and high TMX treated Stop/y,cre mice (0.96 +/- 0.1 seconds and / seconds, respectively; p<0.05, one-way ANOVA) (Figure 8B,C,D). Low TMX treated Stop/y,cre mice displayed /- 4.2 discharges per hour and high TMX treated Stop/y,cre mice displayed an average of 5.4 +/- 1.4 discharges per hour. The incident rate was only significantly reduced in high TMX treated Stop/y,cre mice compared to Stop/y (p<0.05, one-way ANOVA). There was a strong trend towards significance in low TMX treated Stop/y,cre mice compared to Stop/y (p=0.056, one-way ANOVA). We observed severe long discharge events, which were characterized by at least 5 seconds in duration, in Stop/y mice. These severe long discharge events coincide with movement arrest and blank gaze. 88.9% (8 out of 9) of our Stop/y cohort displayed these severe long discharge events, which was significantly higher than the 11.1% (1 out of 9) observed in low TMX treated Stop/y,cre mice. However, no long discharge events were observed in any high TMX treated mice (0 out 6) (Figure 8F). 57

66 Figure 8 58

67 Figure 8. Epileptiform-like discharge incidence rate is significantly improved in Stop/y,cre mice after tamoxifen treatment. Panel A, B, and C: Representative EEG recording traces (5 seconds) showing a single discharge event. Panel A trace is collected from a Stop/y mice, panel B trace is collected from a Stop/y,cre + low TMX mice, and panel C trace is collected from a Stop/y,cre + high TMX mice. Panel D: Histogram showing the mean and SEM of the average discharge duration of individual epileptiform event in Stop/y mice (n=9), Stop/y,cre + low TMX mice (n=6), and Stop/y,cre + high TMX mice (n=5). The average epileptiform duration is significantly reduced in Stop/y,cre mice with low and high TMX compared to Stop/y mice (p<0.05, one-way ANOVA with Tukey s post-hoc correction). Panel E: Histogram showing the mean and SEM for the number of epileptiform events per hour in Stop/y mice (n=9), Stop/y,cre + low TMX mice (n=8), and Stop/y,cre + high TMX mice (n=5). The number of discharge events per hour is significantly reduced in Stop/y,cre + high TMX mice compared to Stop/y mice (p<0.05, one-way ANOVA with Tukey s post-hoc correction). There is a strong trend in the reduction of epileptiform discharges per hour in Stop/y,cre + low TMX mice compared to Stop/y mice (p=0.056, one-way ANOVA with Tukey s post-hoc correction). Panel F: Histogram showing the percentage of population of Stop/y mice (8 out of 9), Stop/y,cre + low TMX (1 out of 9), and Stop/y,cre mice + high TMX (0 out of 9) that exhibit severe long duration discharge events. For panel E and D, * indicates p<0.05, for panel F, it indicates p<0.05 compared between Stop/y versus Stop/y,cre + low TMX and Stop/y,cre + high TMX. 59

68 Reactivation of MeCP2 in female MeCP2-deificient mice rescues behavioural performances Female Stop/+ and Stop/+,cre display neurological symptoms that recapitulate clinical features of RTT syndrome (Guy et al., 2007). Consistent with symptoms such as impaired inertia, gait, and anxiety behaviour, our Stop/+ female mice showed significant impairment in the open field, rotarod, light/dark placement preference, and nest building behavioural test, compared to wild-type littermates (p<0.01 for all behavioural test) (Figure 9). Stop/+,cre females were similarly impaired before tamoxifen treatments (<270 days of age) and showed no significant differences from Stop/+ females in all the behavioural tests (p>0.05 for all tests). Stop/+,cre females were treated with tamoxifen post 270 days of age and assayed for the same behavioural test 2 months after treatments. Activity, rearing, and mobile counts in the open field test were all significantly improved following MeCP2 reactivation (p<0.05 for all parameters, paired t-test) (Figure 9A). However, no improvements were seen in the light and dark place preference test after treating Stop/+,cre mice with tamoxifen (Figure 9B). Motor coordination was impaired in Stop/,cre mice, as the mice showed a shorter latency to fall on all days of trial compared to wildtype controls (p<0.01, two-way ANOVA). Mild improvements were seen in motor performance as tamoxifen-treated Stop/+,cre mice showed a longer latency to fall on day one of the trials (p<0.05, two-way ANOVA) (Figure 9C). Stop/+,cre females showed a significant impairment in nest building ability compared to wild-type controls (52 +/- 6cm 3 versus 118 +/- 7.1cm 3 in nest volume, respectively; p<0.01 student s t-test) (Figure 9D). Following tamoxifen treatment, nest building behaviour of Stop/+,cre females was significantly improved as the average nest volume was /- 7.9cm 3 (p<0.05, paired t-test) (Figure 9D). 60

69 Figure 9 61

70 Figure 9. Behavioural performances are improved in Stop/+,cre mice following MeCP2 reactivation. Panel A: Histogram showing the mean and SEM of female Stop/+,cre mice before and after tamoxifen treatment in the open field test. The behavioural performances of the mice are normalized to average wild-type levels. Stop/+,cre mice performed significantly better in all parameters after tamoxifen treatment (p<0.05, paired t-test). Panel B: Anxiety-like behaviour was assessed using the light-dark place preference test. The histogram shows the mean and SEM of the number of risk assessments (head pokes) performed per minute of time spent in the dark chamber. Stop/+,cre mice did not show an improvement in risk assessment behaviour after MeCP2 reactivation. Panel C: Motor performance was assayed using the accelerating rotarod. Latency to fall was significantly improved on day one of the rotarod trial in Stop/+ mice after tamoxifen treatment (p<0.05). Panel D: Social behaviour was assessed using the nest building test. The histogram shows the mean and SEM of the nest volumes built by the mice 24 hours being introduced into the test cage. Stop/+,cre mice built significantly larger nest post MeCP2 reactivation. For all panels, n = 7 for Stop/+,cre mice. Paired student s t-test was used for all statistical comparisons between pre and post-tamoxifen treated Stop/+,cre mice. * indicates p<0.05 compared between pre and post tamoxifen treated Stop/+,cre mice. 62

71 MeCP2 reactivation improves daily rhythmic activity and thermoregulation in adult female MeCP2-deficient mice. Impaired movement and activity are cardinal features of RTT syndrome. These impairments were recapitulated in Stop/+ female mice (Wither et al., 2012) and our cohort of Stop/+,cre mice before tamoxifen treatment as total daily (24 hours) activity count is significantly lower in Stop/+,cre female mice than wild-type littermates (139 +/- 8.7 versus /- 20.2, p<0.05, one way ANOVA) (Figure 10C). Activity counts during the light phase (6:00am 6:00pm) and the dark phase (6:00pm-6:00am) were significantly lower in Stop/+,cre than in wild-type female mice (42.8 +/- 3.7 arbitrary units versus /- 6.6 arbitrary units during light phase, and /- 5.8 arbitrary units versus / arbitrary units during dark phase, Stop/+,cre versus wild-type, p<0.05 Student s t-test) (Figure 10A,B). After treating Stop/+,cre female mice with tamoxifen, activity during the light and dark phase as well as total daily activity counts were significantly improved and restored to wild-type levels ( / arbitrary units and / arbitrary units; p>0.05, compared against wild-type, one way ANOVA). The amount of time spent being active was significantly lower in Stop/+,cre mice compared to wild-type controls ( /- 0.74% versus /- 3.29%, p<0.01, Student s t- test). After treating Stop/+,cre with tamoxifen, the percent of day spent being active was significant increased to /- 1.96% (p<0.05, paired t-test) and was not significantly different from wild-type levels (p>0.05, student s t-test) (Figure 10D). Thermoregulation and core body temperature are abnormal in Stop/+,cre female mice (Figure 11C,D). The daily temperature maximum, minimum, and average of Stop/+,cre female mice ( / C, / C, and / C, respectively) were all significantly lower than wild-type levels ( / C, /- 0.2, / C; 63

72 p<0.01 for each, compared between Stop/+,cre and wild-type, Student s t-test) (Figure 11A,B,C,D). Additionally, core body temperature variation is significantly smaller in wild-type than Stop/+,cre mice (2.98 +/ C versus / C, respectively; p<0.05, Student s t- test). After treating Stop/+,cre with tamoxifen, the maximum, minimum, and average daily temperatures were significantly improved to / C, / C, and / C, respectively (p<0.01, p<0.05, and p<0.01, respectively, compared between pre and post tamoxifen treatment in Stop/+,cre mice, paired t-test) (Figure 11A,B,C,D). However, the daily temperature range was not significantly improved (p>0.05, paired t-test) (Figure 11D). Normally, core body temperature increases during activity and decreases during inactive periods. Thus a strong correlation between core body temperature and activity is usually observed. The Pearson s correlation coefficient between activity and body temperature during 24 hour cycle, light phase, and dark phase were all significantly decreased in Stop/+,cre mice before tamoxifen treatment when compared to wild-type controls (0.56 +/ versus /- 0.03, p<0.01, Student s t-test) (Figure 12F). After treating Stop/+,cre mice with tamoxifen, the correlation between core body temperature and activity during light phase, dark phase, and 24 hour period were all significantly improved (p<0.01 for each, paired t-test) and were not significantly different than wild-type controls (p>0.05, Student s t-test). 64

73 Figure 10 65

74 Figure 10. Daily activity is significantly improved in Stop/+,cre mice after MeCP2 reactivation. Panel A: Histograms showing the mean and SEM of activity counts during the dark phase (6:00pm 6:00am) of wild-type mice, Stop/+,cre mice, and Stop/+,cre mice after tamoxifen treatment. Panel B: Histograms showing the mean and SEM of activity counts during the light phase (6:00am 6:00pm) of wild-type mice, Stop/+,cre mice, and Stop/+,cre mice after tamoxifen treatment. Panel C: Histograms showing the mean and SEM of total activity counts during a 24 hours cycle of wild-type mice, Stop/+,cre mice, and Stop/+,cre mice after tamoxifen treatment. Panel D: Histograms showing the percent of the 24 hour duration the wild-type mice, Stop/+,cre mice, and Stop/+,cre mice after tamoxifen treatment spent being active. For all panels, n=8 for wild-types, and n=6 for Stop/+,cre mice (same mice for before and after tamoxifen treatment groups). Paired Student s t-test was used for all comparisons between Stop/+,cre mice before and after tamoxifen treatment. Student s t-test was used for comparisons between wildtype mice and Stop/+,cre mice without tamoxifen. * indicates p<

75 Figure 11 67

76 Figure 11. Core body temperature is improved in Stop/+,cre mice after MeCP2 reactivation. Panel A: Core temperature maximum during 24 hour period of wild-type mice and Stop/+,cre mice before and after tamoxifen treatment. Panel B: Core temperature minimum during a 24 hour period of wild-type mice and Stop/+,cre mice before and after tamoxifen treatment. Panel C: Average core body temperature during a 24 hour period of wild-type mice and Stop/+,cre mice before and after tamoxifen treatment. Panel D: Scatter plot showing the range of core body temperature in in wild-type mice and Stop/+,cre mice before and after tamoxifen treatment. Each point represents the absolute temperature range of an individual mouse. In all panels, n=8 for wild-type, and n=7 for Stop/+,cre mice. Student s paired t-test was used to all statistical comparisons between Stop/+,cre mice before and after tamoxifen. Student s t-test was used for comparisons between Stop/+,cre before tamoxifen with wild-type mice. * indicates p<

77 Figure 12 69

78 Figure 12. Temperature and mobility correlation is significantly improved in MeCP2 reactivated female mice. Panel A and B: Representative 24 hour activity cycle profile of a Stop/+,cre mice before (A) and after (B) tamoxifen treatment. Each spike indicates an instance of activity. Panel C and D: Representative 24 hour temperature cycle of a Stop/+,cre mice before (C) and after (D) tamoxifen treatment. The cycle is plotted from 6:00am to 6:00am of the next day. Panel E: Histograms showing the mean and SEM of Pearson s correlation coefficient (r) for activity and core body temperature in Stop/+,cre mice before and after tamoxifen. Panel F: Scatter plot showing the range of Pearson s correlation coefficient (r) for activity and core body temperature in wild-type mice (n=8), Stop/+,cre before (n=7) and after (n=7) tamoxifen treatment. Each point indicate the daily correlative strength between activity and body temperature of an individual mice. Paired Student s t-test was used for all comparisons between Stop/+,cre mice before and after tamoxifen treatment. Student s t-test was used for comparisons between wild-type mice and Stop/+,cre mice without tamoxifen. * indicates p<

79 Project MeCP2 is selectively preserved in tyrosine hydroxylase-expressing neurons in the Rescue mouse brain. To allow the selective preservation of functional MeCP2 expression within catecholaminergic cells, we crossed female MeCP2 deficient (MeCP2 +/- ) mice containing a stop-flox MeCP2 allele (Guy et al., 2007) with transgenic mice expressing cre recombinase from an exogenous rat tyrosine hydroxylase (TH) promoter (TH-cre) (Savitt et al., 2005). For simplicity, we will refer to MeCP2-deficient mice as Non-Rescue mice, and MeCP2-deficient mice expressing cre recombinase in TH-positive cells as Rescue mice. To confirm the reactivation efficiency of MeCP2 expression in Rescue mice, we employed dual-label immunohistochemistry and quantified MeCP2 expression within the catecholaminergic regions of the adult male Rescue mouse brain (Figure 13A-H). These results revealed that within the substantia nigra, 87.5 ± 5.0% of the MeCP2-positive neurons stained positively for TH, and conversely, 85.4 ± 3.4% of TH-positive neurons expressed MeCP2 (Figure 13C, D). In the locus ceruleus, 81.3 ± 6.2% of MeCP2-positive neurons expressed TH, and 81.9 ± 5.7% of THpositive neurons expressed MeCP2 (Figure 13G, H). These co-expression percentages were comparable to that of adult wild-type littermate mice, where within the substantia nigra, 94.1 ± 3.1% of the MeCP2 positive neurons co-expressed TH, and 96.9 ± 0.9% of TH-positive neurons co-expressed MeCP2, and 92 ± 7.1% of MeCP2-positive neurons in the locus ceruleus coexpressed TH, and 95 ± 4.5% of TH-positive neurons co-expressed MeCP2 (Figure 13C, D, G, H). Few, if any, MeCP2-positive cells were observed within the cortex, hippocampus, or 71

80 cerebellum of Rescue mice (Supplemental Figure 13A-C). However, although the large majority of MeCP2 expression was restricted to TH-positive cells throughout the Rescue mouse brain, reactivated MeCP2 expression was seen infrequently in some TH-negative cells within the periventricular and paraventricular nuclei of the hypothalamus, and in some neurons in the midbrain and brainstem of adult male Rescue mice (Figure 14 D-E). The ectopic expression of the rescue MeCP2 protein in a small cohort of TH-negative cells is not unexpected, however, as the TH-Cre transgenic mouse we employed has been shown previously to activate a stop-flox reporter gene in scattered cells within these same areas (Savitt et al., 2005). 72

81 Figure 13 73

82 Figure 13. MeCP2 is selectively preserved in catecholaminergic neurons of "Rescue" mice. Panels A and B: Dual-label fluorescence micrographs at different magnifications from the substantia nigra region showing immunoreactivity for MeCP2 (red channel), tyrosine hydroxylase (green channel) and the nuclear stain DAPI (blue channel) in wild-type, "Non- Rescue" and "Rescue" male mice. The scale bar for Panel A is 200 microns, and for Panel B is 50 microns. Panel C: The percentage of MeCP2-positive neurons that co-express TH in the substantia nigra of male wild-type (n=3) and male "Rescue" (n=3) mice. Panel D: The percentage of TH-positive neurons co-expressing MeCP2 in the substantia nigra of male wildtype (n=3) and male "Rescue" (n=3) mice. Panels E and F: Dual-label fluorescence micrographs as above showing MeCP2 and tyrosine hydroxylase immunoreactivity in the locus ceruleus of male wild-type, "Non-Rescue", and "Rescue" mice. Scale bar in Panel E is 200 microns, and is 50 microns in Panel F. Panel G: The percentage of MeCP2-positive neurons co-expressing TH in the locus ceruleus of male wild-type (n=3) and male "Rescue" (n=3) mice. Panel H: The percentage of TH-positive neurons co-expressing MeCP2 in the locus ceruleus of male wild-type (n=3) and male "Rescue" (n=3) mice. 74

83 Figure 14 75

84 Figure 14. MeCP2 expression is not preserved in non-catecholaminergic neurons. Panel A: Ectopic MeCP2 immunoreactivity in the cortex regions of a Rescue mouse at different magnifications. Panel B: MeCP2 immunoreactivity is not preserved within the hippocampus region of the Rescue mice. Panel C: Ectopic MeCP2 immunoreactivity is not observed within the cerebellum regions of the Rescue mice brain. Scale bars in all panels indicate 200 µm. 76

85 Preservation of MeCP2 in catecholaminergic cells extends the lifespan of male MeCP2-deficient mice. Analysis of Kaplan-Meyer survival plots revealed life expectancy to be significantly longer in male Rescue mice than male Non-Rescue mice. In contrast to MeCP2-null Nonrescue mice, which displayed a median survival age of 77 days, male Rescue mice displayed a median survival age of 180 days, and more than 40% of the Rescue cohort survived longer than 200 days (Figure 15A). Only one of the male Non-Rescue mice in our cohort lived longer than 100 days (Figure 15A). In addition to increased longevity, male Rescue mice displayed an overall improvement in general phenotypic severity compared to Non-Rescue mice. Using a phenotypic severity scale previously employed for MeCP2-deficient mice (Guy et al., 2007; Robinson et al., 2012; Cheval et al., 2012) we found that male Rescue mice consistently displayed lower severity scores than Non-Rescue mice at and after 5 weeks of age (Figure 15B, two-way ANOVA p<0.05). The increase in lifespan, and attenuation of phenotype severity, occurred in the absence of body mass normalization, however, as male Rescue mice remained significantly underweighted compared to age-matched wild-type mice, and not significantly different from Non-Rescue mice (Figure 15C, two-way ANOVA p<0.001). 77

86 Figure 15 78

87 Figure 15. Survival and gross phenotypic behaviour are improved in male and female TH "Rescue" mice. Panel A: Kaplan-Meier survival plot of male "Non-Rescue" (light grey line, n=29) and male "Rescue" (dark grey line, n=17) mice. The life span of the "Rescue" mice is significantly longer than "Non-Rescue" mice (p<0.01, Wilcoxon rank-sum test). Panel B: The gross phenotypic severity score of male "Rescue" mice (n=10, dark grey line) is significantly lower than male "Non-Rescue mice" (n=13, light grey line) at and after 35 days of age (p<0.05, one-way ANOVA with Tukey's post-hoc test). The severity scores for male wild-type mice were between 0-1 over these ages (shown in closed circles). Panel C: The average body mass of male "Rescue" mice (dark grey line, n=19-8 at different ages) does not significantly differ from male "Non-Rescue" mice (light grey line, n=23-7 at different ages), and both groups remain significantly underweighted compared to male wild-type mice (black line, n=23) (p<0.01, oneway ANOVA with Tukey's post-hoc test). Panel D: The spontaneous death rate in female "Rescue" mice (3 of 33) is significantly lower than in female "Non-Rescue" mice (10 of 29; p<0.05, Chi-Square test with one degree of freedom). Panel E: The gross phenotypic severity score of female "Non-Rescue" mice (light grey line, n=8) and female "Rescue" mice (dark grey line, n=10) does not significantly differ at any time between weeks of age (one-way ANOVA). Wild-type female mice severity scores were between 0-1 over these ages (shown in solid circles). Panel F: The average body mass of female "Rescue" mice (dark grey line, n=12) does not significantly differ from female "Non-Rescue" mice (light grey line, n=13), or female wild-type mice (black line, n=17, one-way ANOVA with Tukey's post-hoc test). 79

88 Preservation of MeCP2 in catecholaminergic cells decreases the rate of sudden unexpected death in female MeCP2-deficient mice Although early lethality is not a typical phenotype of female MeCP2 +/- mice (Chen et al., 2001; Guy et al., 2001), female MeCP2-deficient mice are prone to sudden and unexpected death. In our cohort, 34.5% (10 of 29) of the female MeCP2-deficient mice died suddenly and without indication before reaching one-year of age, which is significantly higher than the rate of 4.8% (2 of 42) observed in female wild-type mice. In contrast, the spontaneous unexplained death rate in female Rescue mice was significantly lower at 9.1% (3 of 33) (Figure 15D, p<0.05), and not significantly different from the rate seen in female wild-type mice. This effect did not correlate with significant improvements in the gross phenotypic severity of the cohort of Rescue mice, however, as the general phenotypic severity score of the female Rescue mice did not significantly differ from the female Non-Rescue mice (Figure 15E). Further, there was no correlation between the phenotypic severity score of a given mouse with its unexpected death; mice displaying low phenotypic severity scores were equally likely to die spontaneously as mice displaying higher cumulative severity scores (Figure 16). Analysis of growth rate in female Rescue mice also failed to reveal any differences from female Non-Rescue mice. It should be noted, though, that on the pure C57Bl/6 background employed for this study, neither the Non-Rescue or Rescue mice displayed significant body mass differences from female wildtype mice (Figure 15F). 80

89 Figure 16 81

90 Figure 16. The phenotypic severity score of female MeCP2 +/- mice does not correlate with the time of their sudden and unexpected death. On the plot, the x-axis denotes the age of spontaneous death, while the y-axis indicates the severity score of the individual mouse one week before its death. Each point on the plot denotes an individual female MeCP2 +/- mouse. Linear regression analysis of these data revealed a Pearson R value of 0.133, indicating a lack of correlation between severity score and time of spontaneous death (p=0.714). 82

91 Catecholaminergic preservation of MeCP2 improves deficits in ambulatory rate, motor coordination, and anxiety-like behaviour in male MeCP2-deficient mice Consistent with previous reports (Samaco et al., 2009; Ward et al., 2011), day old male Non-Rescue mice displayed clear impairments relative to wild-type and TH-cre mice in general activity, balance and coordinated movement, ambulatory rate, risk-assessment behaviour, and in nest-building performance. In the open field test, male Non-Rescue mice displayed on average a 72.3 ± 2% decrease in total activity counts, an 83.2 ± 3% decrease in rearing counts, and a 38.5 ± 3% decrease in ambulatory rate compared to age-matched male wild-type mice. In contrast, although remaining below wild-type levels, the general activity, total rearing behaviour, and ambulatory rate, of male Rescue mice were significantly improved from Non-Rescue mice (Figure 3A; p<0.05 for each behaviour). On the accelerating rotarod, male Non-Rescue mice displayed a shorter latency to fall time compared to wild-type or TH-cre mice on each of the four consecutive trial days. The latency to fall time for male Rescue mice was significantly longer than Non-Rescue mice on days 2-4 of the trial paradigm (Figure 17B, twoway ANOVA p<0.05). In the light/dark place preference test, male Non-Rescue mice conducted fewer risk-assessments while in the dark compartment compared to either wild-type or TH-cre mice (Figure 17C). The risk assessment behaviour of male Rescue mice was significantly improved from Non-Rescue mice (2.34 ± 0.27 verses 1.2 ± 0.25 risk assessments/minute, respectively; p<0.05 one-way ANOVA; Figure 17B), but still below the performance of wild-type mice. Finally, in the nest-building test, male Rescue mice assembled nests with significantly larger volume than Non-Rescue mice (25.5 ± 7.3 cm 3 versus 3.5 ± 0.3 cm 3, respectively), (p<0.05 one-way ANOVA; Figure 17D). 83

92 Figure 17 84

93 Figure 17. Behavioural performances are improved in male "Rescue" mice. Panel A: Histogram showing the mean and SEM of male "Rescue" mice (n=16, black) relative to male "Non-Rescue" mice (n=17, dark grey) and male TH-cre control mice (n=24, light grey) in the open field test. On the histogram, the average performance of male wild-type mice is denoted as 100% (dotted line). The general activity, rearing, and ambulatory rate of male "Rescue" mice was significantly improved from male "Non-rescue" mice. Panel B: Motor coordination was assessed using the accelerating rotarod test. Though remaining below the values of male wild-type (n=26) or THcre (n=21) control mice, male "Rescue" mice (n=20) displayed a significantly longer latency to fall than male "Non-Rescue mice" (n=17) on trial days 2, 3, and 4. Panel C: Anxiety-like behaviour was assessed using the light-dark place preference test. The histogram shows the mean and SEM of the number of risk assessment time (head poke) taken by the different cohorts of male mice per minute of time spent in the dark chamber of the apparatus. The risk-assessment behaviour of male "Rescue" mice (n=12) was significantly above that of male "Non-Rescue" mice (n=12), but remained below the behaviour of male wild-type (n=20) or TH-cre controls (n=7). Panel D: The nest-building test was used as an index of social behaviour. The histogram shows the mean and SEM of the nest volume built by the different cohorts of mice 24 hours after being placed in the test cage with a neslet. Male "Rescue" mice (n=13) assembled significantly larger nests than male "Non-Rescue" mice (n=13), but smaller than male wild-type (n=14) or TH-cre control mice (n=13). One-way ANOVA with Tukey's post-hoc test for multiple comparisons was used for the open field, light-dark place preference, and nesting behaviour tests, and a two-way ANOVA (genotype verses trial) with Bonferroni's post-hoc test was used for the accelerating rotarod test. For each panel, * indicates p<0.05 compared between "Rescue" and "Non-Rescue" mice. 85

94 Catecholaminergic preservation of MeCP2 improves the ambulatory and anxiety-like behavioural deficits of adult female MeCP2-deficient mice Although less well characterized to date, female MeCP2 +/- mice also display significant impairments in overall activity, ambulation rates, motor coordination, and anxiety-like behaviour after 8-12 months of age (Jugloff et al., 2008; Wither et al., 2012; Stearns et al., 2007). Consistent with the results obtained in male Rescue mice, these behavioural deficits were also largely improved in female Rescue mice. In the open field test, the general activity counts, rearing behaviour, and average ambulation rate were each significantly improved compared to female Non-Rescue mice (Figure 18A, one-way ANOVA p<0.01 for each). In the accelerating rotarod, female Rescue mice showed a partial rescue; their performance was significantly improved from female Non-Rescue mice on the 4th day of the trial paradigm (Figure 18B, two-way ANOVA p<0.05). In the light/dark place preference test, female Rescue mice displayed a significant improvement in risk-assessment behaviour compared to age-matched female Non-Rescue mice (Figure 18C, one-way ANOVA p<0.05). In fact, the risk assessment behaviour of female Rescue mice did not significantly differ from that of female wild-type mice. In the nest-building test, female Rescue mice assembled nests that were significantly larger in total volume than those of female Non-Rescue mice (Figure 18D, one-way ANOVA p<0.05), and equivalent in volume to those assembled by female wild-type or TH-cre mice (Figure 18D). 86

95 Figure 18 87