Purkinje-cell-specific MeCP2 deficiency leads to motor deficits and autistic-like behavior due to aberrations in PTP1B-TrkB-SK signaling

Purkinje-cell-specific MeCP2 deficiency leads to

Purkinje-cell-specific MeCP2 deficiency leads to motor deficits and autistic-like behavior due to aberrations in PTP1B-TrkB-SK signaling Purkinje-cell-specific MeCP2 deficiency leads to motor deficits and autistic-like behavior due to aberrations in PTP1B-TrkB-SK signaling

INTRODUCTION
Over the past decades, the relevance of the ''cerebellar connectome'' has been emerging. 1The cerebellum is heavily connected with other brain regions, including the thalamus and cerebral cortex, 2,3 subserving not only motor but also non-motor functions. 4,5[9][10][11][12][13] Rett syndrome patients suffer from loss-of-function mutations of the Mecp2 gene encoding methyl-CpG-binding protein 2 (MeCP2). 14Loss of Mecp2 function causes not only typical ASD features, such as deficits in social interaction and repetitive behavior, but also motor symptoms, such as ataxia and tremors. 15,16In mice, global loss of Mecp2 causes deficits in motor performance, including hindlimb clasping, hypoactivity, and tremors. 17To date, many Rett syndrome symptoms have been attributed to cellular deficits in various brain regions, such as the cerebral cortex and basal ganglia, [18][19][20][21][22] but surprisingly, the impact of Mecp2 mutation in the main output neurons of the cerebellar cortex, Purkinje cells (PCs), has been relatively neglected.Moreover, the pathological consequences of MeCP2 mutation at the molecular and cellular levels that ultimately lead to changes in behavior have yet to be fully established.
Utilizing a mouse model with PC-specific deletion of Mecp2, we here shed light on the impact of loss of function of MeCP2 at both the cellular and behavioral levels.Our results demonstrate that deletion of Mecp2 reduces the excitability of PCs through a signaling pathway, which entails the PTP1B-TrkB (the receptor of brain-derived neurotrophic factor [BDNF]) cascade.These PC-specific cellular changes, in turn, lead to aberrations in both motor and non-motor behaviors.Moreover, increasing the activity of TrkB in PCs at the adult stage is sufficient to rescue PC dysfunction and defective behaviors caused by Mecp2 deficiency.In short, our findings elucidate part of the pathogenetic mechanisms underlying Rett syndrome and highlight how PC dysfunction may contribute to symptoms associated with ASD.(legend continued on next page)

RESULTS
Mecp2 deletion in PCs impairs motor learning Consistent with previous work, 23 we found that MeCP2 is robustly expressed in PCs of post-natal mice (post-natal day 2 [P2]-P90) (Figure 1A).To determine the role of MeCP2 in PCs, mice with a Mecp2 deletion specifically in PCs (pCre;Mecp2 f/f and pCre; Mecp2 f/y ) were generated by crossing pCP2-Cre with Mecp2 f/f (female) or Mecp2 f/y (male) mice.Mecp2 deletion was confirmed by assessing its mRNA in PCs (Figure S1A).Body weight and cerebellar size of Mecp2 conditional knockout (cKO) mice appeared normal at both P30 and P90 (Figures S1B and S1C).
We asked whether PC deletion of Mecp2 leads to motor deficits.During the footprint test, cKO mice exhibited normal gait (Figure 1B).In the rotarod test, cKO mice showed limited improvement and learning over 8 sessions; their latency to fall was shorter than that of control mice at late stages (Figure 1C).Next, we examined compensatory eye movements, which are more specific to the function of the cerebellum.We found that the amplitude (gain) and timing (phase) of the baseline optokinetic reflex (OKR), vestibulo-ocular reflex (VOR), and visually enhanced VOR (VVOR) did not differ between control and cKO mice (Figures 1D-1F), supporting the hypothesis that PC deletion of Mecp2 does not alter basic motor performance.We continued to test a phase-reversal VOR adaptation protocol, a more sensitive test for cerebellar learning. 10,24While both control and cKO mice did not differ in VOR gain decrease, this training resulted in an increase in VOR phase in control mice that could not be matched by cKO mice (day 3, p = 0.02; day 4, p = 0.002; day 5, p < 0.001; n = 12/group; ANOVA repeated measures) (Figure 1G).Together, these results indicate that PC deletion of Mecp2 impairs motor learning.

Abnormal social and repetitive behaviors in cKO mice
][10]13 Thus, we examined psychiatric behaviors in female (pCre;Mecp2 f/f ) and male (pCre;Mecp2 f/y ) mice because functional magnetic resonance imaging (fMRI) scanning in ASD patients shows sex-dependent differences in cortico-cerebellar organization. 25In the three-chamber interaction task, pCre;Mecp2 f/y and pCre;Mecp2 f/f mice displayed a preference for a stranger mouse (S1), similar to Mecp2 f/y and Mecp2 f/f mice (Figure 2A).Moreover, there was no difference in the sniffing time between Mecp2 f/y and pCre;Mecp2 f/y mice or between Mecp2 f/f and pCre;Mecp2 f/f mice (Figure 2A).These results indicate that mutant mice have no deficit in social ability.We next tested social novelty by introducing a second stranger mouse (S2).Both Mecp2 f/y and Mecp2 f/f mice demonstrated an increased preference for S2, whereas neither pCre;Mecp2 f/y nor pCre;Mecp2 f/f mice showed a particular preference for S2 (Figure 2A).
These observations were further examined by performing the resident-intruder test. 26In trial 1, a subject mouse was exposed to an unfamiliar mouse for an interaction of 5 min.After a 1-h inter-trial interval, either the previous familiar mouse or a second novel mouse was introduced to the subject mouse in trial 2 (Figures S2A and S2B).We found a significant reduction in the exploration time for control mice, but not cKO mice, in trial 2. In comparison, both control and cKO mice showed no differences in exploring the novel mouse (Figures S2A and S2B).Combined with the three-chamber test, these results suggested that cKO mice may have impaired social memory.
Repetitive behavior was examined using the grooming, marble burying, and T maze tasks.Water spray 10 induced grooming behaviors in pCre;Mecp2 f/y and pCre;Mecp2 f/f mice, as shown by more interrupted bouts and time spent self-grooming (Figure 2B).In contrast, no difference was observed in the number of buried marbles between control and cKO mice (Figure 2C).Moreover, cKO mice showed no difference in the T maze task test compared with controls (Figure 2D).
Last, we investigated whether the deficit in social function may be due to impaired ability to recognize odorants.Thus, we performed odorant cue recognition experiments and found that cKO mice displayed reduced interest when exposed repeatedly to the same odorant (Figures S2C and S2D).In this test, cKO mice behaved in a manner similar to control mice, suggesting that cKO mice retained normal olfactory discrimination ability.
Together, our results indicate that both male and female mice with a PC-specific deletion of Mecp2 exhibit a social novelty defect.

Mecp2 deficiency in PCs does not affect cytoarchitecture and neurotransmission
As for cerebellar development, Mecp2 deletion did not interrupt the formation of cerebellar lobules at P30 (Figure S3A).Moreover, the PC layer appeared normal in cKO mice (Figure S3B).Finally, Golgi apparatus staining and Sholl analysis showed that both the intersection number (Figures S3C and S3D) and spine formation were normal in cKO PCs (Figures S3E and  S3F).In summary, no evidence supports the hypothesis that Mecp2 deficiency influences PC morphology.
Previous work suggests that synaptic efficacy and excitatoryinhibitory input balance are affected in mice with Mecp2 mutations. 22,27Hence, we investigated whether excitatory and/or inhibitory transmissions of PCs are influenced by Mecp2 deletion.Using whole-cell patch recordings in cerebellar slices, excitatory and inhibitory inputs of PCs were examined (Figure S4A).First, neither frequency nor amplitude of miniature excitatory postsynaptic currents (mEPSCs) was altered in cKO mice (Figure S4B).Second, neither amplitude nor paired-pulse facilitation (PPF) of evoked (parallel fiber) PF-EPSCs was different between control and cKO mice (Figure S4C).Third, no difference in the amplitude or paired-pulse depression (PPD) of climbing fiber (CF)-EPSCs was found between control and cKO mice (Figure S4D).With applying sub-or suprathreshold stimuli, CF-EPSCs were elicited in an all-or-none fashion in both control and cKO mice (Figure S4E), suggesting that CF elimination develops normally in cKO mice.
PCs also receive inhibitory inputs from interneurons (Figure S4A).Thus, we measured miniature and evoked inhibitory postsynaptic currents (mIPSCs and eIPSCs, respectively).Neither the frequency nor amplitude of mIPSCs was affected by Mecp2 deletion (Figure S4F).Mean eIPSC amplitudes at various stimulation intensities were comparable between two groups (Figure S4G).These recordings suggest that GABAergic transmission onto PCs is not affected by Mecp2 deletion.To examine the excitatory-inhibitory input ratio in PCs, constant stimulation was applied to PFs, and eIPSCs or eEPSCs were recorded sequentially from the same PC.No difference in excitatory-inhibitory input ratio was found between control and cKO PCs (Figure S4H).
In addition, we examined whether metabotropic glutamate receptor 1 (mGluR1) is affected in cKO mice.Western blots showed that synaptic expression of mGluR1, GluA2, and excitatory amino acid exchanger 4 (EAAT4) was unaltered in cKO mice (Figure S4I).We found that the peak amplitude of mGluR1-EPSCs was comparable between control and cKO mice (Figure S4J), suggesting that mGluR1 is not affected by PC deletion of Mecp2.

Reduced intrinsic excitability in late adolescent cKO mice
What is the mechanism causing abnormal behaviors of cKO mice?Intrinsic excitability may be a candidate because it is altered in mouse models of ASD and Rett syndrome. 7,10,22,28oreover, PC excitability is critical for the output of the cerebellar cortex, which further regulates cerebello-cerebral loops that give rise to the profile of psychiatric deficits. 6Hence, we examined intrinsic excitability of PCs in mice at P30. Action potentials (APs) of PCs, referred to as simple spikes in vivo, 29 were recorded following a rheobase current (Figure 3A).AP parameters, including threshold, amplitude, half-width, and afterhyperpolarization potential (AHP), were measured (Figure 3B).We found that control and cKO PCs had a similar AP pattern (Figure 3C).In addition, APs of control and cKO PCs showed a comparable threshold, amplitude, half-width and AHP (Figure 3D).Intrinsic excitability was assessed by injecting stepped currents of increasing amplitude on PCs.Our results showed a linear current-to-firing frequency relationship with similar slopes in both groups (Figure 3E).These findings indicate that PC intrinsic excitability is normal in cKO mice at P30.
The examination of the excitability of PCs acquired different results at P90.First, AHP amplitude increased in cKO PCs (Figures 3F and 3G).Second, current injection evoked fewer spikes in cKO PCs compared with control PCs (Figure 3H), demonstrating that PC excitability is reduced in cKO mice at P90. Next, we investigated whether synaptic transmission is affected by MeCP2 deficit at P90.In summary, our results argue against this possibility.First, neither the frequency nor amplitude of mEPSCs was altered in cKO mice (Figure S5A).Second,  (legend continued on next page) neither the amplitude nor PPF of PF-EPSCs was different between control and cKO mice (Figure S5B).Third, no difference in the amplitude or PPD of CF-EPSCs was found between control and cKO mice (Figure S5C).Fourth, all CF-EPSCs were elicited in an all-or-none fashion in both control and cKO mice (Figure S5D).Fifth, neither the frequency nor amplitude of mIPSCs was affected by Mecp2 deletion (Figure S5E).Finally, mean eIPSCs in response to different stimulation intensities were comparable between control and cKO groups (Figure S5F).
Previous work has shown that the inhibition of PC activity in the Crus1/2 area generates ASD-related behaviors. 11Thus, we investigated whether Mecp2 deletion affects the excitability of Crus PCs at P90.Indeed, Crus1/2 PCs in cKO mice fired fewer spikes as well (Figure 3I), indicating that Mecp2 deletion reduces the excitability of Crus1/2 PCs.As a control, we also examined APs and intrinsic excitability of Crus1/2 PCs at P30.Our results showed that AP parameters, including threshold, amplitude, half-width, and AHP, were not affected by MeCP2 deficit in these PCs at this stage (Figure S6A), which is in line with the results obtained from vermal PCs (Figure 3D).We also injected current steps in Crus 1/2 PCs at P30 and examined their intrinsic excitability.Our results indicated similar linear current-to-firing frequency relationships between control and cKO Crus1/2 PCs (Figure S6B).These findings indicate that Crus1/2 cKO PCs also have normal intrinsic excitability at P30. Small-conductance calcium-activated K + (SK) channels are upregulated in cKO mice Next, we investigated which molecular event might influence PC excitability.We focused on SK channels because both pharmacological blockade of SK and deletion of Kcnn2, the gene that encodes SK2, can enhance PC excitability. 24Total and surface expression of SK2 was unchanged in cKO mice at P30 (Figure 4A), but surface SK2 was increased significantly in cKO mice at P90 (Figure 4A).Furthermore, we found that SK2 currents obtained at each holding potential were larger in cKO PCs than those in control PCs, which was also shown by the current-voltage (I-V) relationship plots (Figure 4B).In addition, we found that Mecp2 deletion did not affect SK2 currents at P30, as shown by an unchanged I-V relationship (Figure 4B).
We recorded SK2-mediated tail currents, [30][31][32] which are elicited by a transient Ca 2+ influx responding to a depolarizing voltage step and shown as a resurgent current immediately after returning to the holding voltage. 30For a series of voltage steps from À40 to +30 mV before jumping back to a holding current of À50 mV, SK2-mediated tail currents were markedly larger in cKO PCs at potentials more positive than À20 mV (Figure 4C).Moreover, tail currents were recorded in Crus1/2 PCs, and they were also augmented (Figure 4D).Collectively, our results suggest that Mecp2 deletion enhances SK2 currents.
In addition, we asked whether Mecp2 deletion affects hyperpolarization-activated cyclic nucleotide-gated 1 (HCN1) and HCN2 channels, the major isoforms of the HCN family in PCs 33 that contribute to the hyperpolarization-activated current (I h ) and regulate motor learning. 34Our results indicated that neither HCN1 and HCN2 expression nor I h was affected by PC deletion of Mecp2 (Figure S7).

MeCP2 regulates SK2 through BDNF-TrkB signaling and PTP1B
We continued to investigate the underlying molecular events that link Mecp2 deletion to SK2 up-regulation.Among a large number of molecular signatures driven by Mecp2 mutation, BDNF and its receptor, tyrosine kinase receptor B (TrkB), are regulated by Mecp2. 35,36Moreover, BDNF signaling is considered one of therapeutic targets for treating Rett syndrome. 37Inspired by these findings, we asked whether BDNF signaling participates in the regulation of SK2 by Mecp2 deletion.BDNF (50 ng/mL) was applied to cerebellar slices of control mice for 30 min, and SK2 and TrkB were evaluated immediately.Administering BDNF enhanced the tyrosine phosphorylation of TrkB at 706, and it also significantly decreased surface SK2 (Figure 5A).Subsequently, SK2 currents were recorded in control or BDNF-containing solution, and they were reduced by BDNF application.These results show that BDNF/TrkB signaling negatively regulates SK2 (Figure 5B).
A parallel question was whether Mecp2 deletion affects the phosphorylation of TrkB.We measured the tyrosine phosphorylation of TrkB-Y706, which is essential for the activity of TrkB. 38nterestingly, western blots revealed an age-dependent action of Mecp2 deletion; the phosphorylation of TrkB-Y706 was intact at P30 but inhibited at P90 (Figure 5C).Meanwhile, total TrkB was unaffected at both ages (Figure 5C).These results suggest that Mecp2 deletion reduces TrkB activity and downregulates surface SK2.The effect of BDNF was verified by recording SK2 currents in cKO slices.Our results showed that SK currents decreased with BDNF application (Figure 5D), and the attenuation of SK currents was more pronounced in cKO mice (Figures 5B  and 5D), suggesting that BDNF-TrkB signaling and MeCP2 work through same pathway to control SK2 expression.
However, the question remained how Mecp2 deletion regulates the phosphorylation of TrkB in an age-dependent manner.It has been reported that the expression of PTP1B, a protein tyrosine phosphatase, increases due to Mecp2 disruption and that this increase is sufficient to downregulate tyrosine phosphorylation of TrkB. 39Likewise, we found that there was a development-dependent regulation of Mecp2 on PTP1B expression; PTP1B was unchanged in cKO mice at P30 but increased at P90 (Figure 5E).Furthermore, a series of experiments was conducted to confirm the roles of PTP1B on TrkB signaling and PC excitability.First, an antagonist of PTP1B, 39,40 CPT-157633 (1 mM), was applied to cerebellar slices (P90) of cKO mice for 30 min, and the phosphorylation of TrkB was evaluated.We found that the application of CPT-157633 significantly increased the phosphorylation level of TrkB-Y706 in control slices without changing total TrkB (Figure S8A).Furthermore, CPT-157633 also reversed the phosphorylation of TrkB-Y706 in cKO slices, approximating it to a similar level to PTP1B inhibition on control slices (Figure S8A).Second, we examined the effect of CPT-157633 on APs of control and cKO PCs (P90) by measuring threshold, amplitude, half-width, and AHP.Our results showed that CPT-157633 did not alter the threshold, amplitude, or half-width in control or cKO PCs but reduced AHP in both control and cKO PCs (Figure S8B).Third, intrinsic excitability was assessed by injecting stepped currents on control and cKO PCs (P90) with or without CPT-157633.We found that CPT-157633 increased PC excitability in both control and cKO PCs (Figure S8C).Finally, SK2 currents were recorded in control PCs and cKO PCs with or without CPT-157633.The I-V curves demonstrate that inhibiting PTP1B could reduce SK2 currents under the conditions of both control and MeCP2 deficit (Figure S8D).
Taken together, our results reveal unrecognized PTP1B-TrkB-SK2 signaling in PCs that is independent of MeCP2 deficit.Through this mechanism, PC deletion of Mecp2 can cause a late adolescent elevation in PTP1B expression, which further down-regulates TrkB phosphorylation and enhances membrane expression of SK2.TrkB Y706E expression in PCs recovers SK currents and intrinsic excitability in pCre;Mecp2 f/y mice Given that Mecp2 controls TrkB and SK2, a critical question was whether rescuing the TrkB level was enough to recover SK2 upregulation caused by Mecp2 deletion.Confirming this hypothesis would strengthen the causal link from TrkB to SK2.To address this question, an adeno-associated virus serotype 9 (AAV9)-based double-floxed inverse orientation (DIO)-GFP-TrkB-Y706E-FLAG (TrkB Y706E ) vector was constructed, and an AAV9-based GFP vector acted as the naive control (NC) (Figure 6A), because TrkB-Y706E increases the phosphorylation and activity of TrkB. 38The virus was injected into the middle cerebellar lobules of pCre;-Mecp2 f/y mice (P30).TrkB Y706E was expressed in PCs due to the presence of Cre, shown by a GFP signal abundantly expressed in PCs (Figure 6A).Viral expression was also confirmed by expression of the FLAG tag (Figure 6B).Moreover, both total TrkB and TrkB Y706E were elevated after injection with GFP-TrkB Y706E (Figure 6C), indicating the efficiency of viral infection.
SK2 expression and currents were examined in PCs transduced with NC and TrkB Y706E vectors.TrkB Y706E expression significantly decreased surface SK2 without affecting total expression (Figure 6D).SK2 currents were reduced in PCs transduced with GFP-TrkB Y706E compared with NC (Figures 7A and  7B).Moreover, SK2-mediated tail currents markedly decreased in TrkB Y706E -transduced cells compared with NC (Figure 7C).Collectively, these results indicate that Y706E expression attenuates SK2 currents in cKO PCs.
Finally, APs and intrinsic excitability of cKO PCs were examined.TrkB Y706E expression did not change thresholds but significantly decreased AHP of cKO PCs (Figure 7D), making it similar to control levels (control, 6.1 ± 0.2 mV; TrkB Y706E , 5.9 ± 0.3 mV; p = 0.26).Current injection evoked more spikes in TrkB Y706E cKO PCs than NC PCs, and the plot of firing frequency vs. injected current showed that TrkB Y706E cKO PCs were more excitable than NC (Figure 7E).Moreover, PC excitability was compared between control PCs and PCs receiving TrkB Y706E infection.Again, TrkB Y706E significantly decreased AHP (Figure S9A).Current injection evoked more spikes in TrkB Y706E -expressing control PCs than NC PCs (Figure S9B).Taken together, transducing phosphorylated TrkB can reduce TrkB Y706E expression in PCs rescues motor and non-motor behaviors Given the results above, we asked whether transducing TrkB Y706E could rescue impaired behaviors of pCre; Mecp2 f/y mice.To this end, we injected AAV9 containing GFP-tagged TrkB Y706E or GFP vehicle into contiguous vermal lobules, the flocculus, and Crus1/2 of pCre;Mecp2 f/y mice at P30 and performed behavioral tests 2 months later.
First, we examined locomotion after NC and TrkB Y706E expression in the middle lobules of the vermal cerebellum and the flocculus (Figures S10A and S10B).Compared with NC, TrkB Y706E -treated cKO mice exhibited longer latency to fall from the rotarod after 5 sessions (Figure 7F).In the compensatory eye movement test, TrkB Y706E did not change the gain or phase of the OKR (Figure S11A), VOR (Figure S11B), or VVOR (Figure S11C).During the phase-reversal VOR adaptation test, NC or TrkB Y706E -treated pCre;Mecp2 f/y mice exhibited a similar increase in VOR gain on all days (Figure 7G).However, TrkB Y706E -transduced mice displayed a significant increase in VOR phase compared with NC mice (day 4, p < 0.001; day 5, p < 0.001; ANOVA repeated measures) (Figure 7G).These results reveal that TrkB Y706E expression is sufficient to rescue the impaired motor learning of pCre;Mecp2 f/y mice.
Second, we evaluated whether TrkB Y706E expression in PCs of lobules IV, V, and VI and Crus1/2 could rescue autistic-like behaviors in pCre;Mecp2 f/y mice (Figure S10C).In the social approach (legend continued on next page) test, the mice transduced with NC and TrkB Y706E vectors did not differ in the preference to S1 (Figure 7H).In the social novelty test, GFP-transduced pCre;Mecp2 f/y mice demonstrated impaired social interaction similar to untreated pCre;Mecp2 f/y mice, as shown by low preference to S2 and short sniffing time to S2 (Figure 7H).Interestingly, introducing TrkB Y706E in cKO PCs recovered the response of the mouse to social novelty, as shown by increased preference and sniffing time to S2 (Figure 7H).These findings were further corroborated by the resident-intruder test, in which TrkB Y706E expression successfully rescued the social memory defect of cKO mice, when they interacted with the familiar mice (Figures S2E and S2F).We further evaluated whether TrkB Y706E expression in PCs impacts repetitive behaviors and open field performance of pCre;Mecp2 f/y mice.We found that TrkB Y706E expression ameliorated grooming time, though the number of grooming bouts was unaltered (Figure 7I).Taken together, our results suggest that the introduction of TrkB in PCs might ameliorate part of the autistic-like behaviors in pCre;Mecp2 f/y mice.

DISCUSSION
In the present work, we show that (1) PC deletion of Mecp2 impairs motor learning and causes autistic-like behaviors, (2) Mecp2 deletion in PCs does not change glutamatergic or GABAergic transmission and instead promotes surface expression and currents of SK2 channels and affects the intrinsic excitability, (3) Mecp2 deletion gradually increases PTP1B and ultimately attenuates TrkB activity, (4) BDNF/TrkB signaling regulates the surface expression of SK2, and (5) the expression of TrkB Y706E in adult Mecp2-null PCs is sufficient to restore PC function and rescue both motor and non-motor behavioral deficits.In short, we show a MeCP2 signaling pathway that controls PC excitability by regulating TrkB activity through PTP1B.Accordingly, MeCP2 deficiency augments SK2 currents, reduces intrinsic excitability, and distorts motor and non-motor behaviors.Previous work claims that Mecp2 deletion from the cerebellum, hindbrain, and/or spinal cord causes limited motor deficits. 23Here we expanded on that work through more in-depth experiments and analyses and revealed that MeCP2 in PCs is indeed relevant to a variety of motor and non-motor behaviors.

Distinct functions of MeCP2 in neurons
MeCP2 is widely expressed in neurons and glial cells and regulates their development and function. 41In the cerebral cortex, MeCP2 regulates the strength of excitatory synaptic transmission of pyramidal cells, 42 while its loss in inhibitory neurons reduces GABAergic transmission. 22Here, we show that MeCP2 in PCs affects surface SK2 and intrinsic excitability rather than EPSCs and IPSCs.It is a long-standing question to what extent loss of MeCP2 is involved in neurogenesis and synaptic development. 435][46] While these findings show the participation of MeCP2 in neuronal maturation, our work indicates that MeCP2 does not exert an effect on the development of PCs.Thus, MeCP2 may have different molecular actions in different cell types.These heterogeneous effects are in line with the great variety in the methylation landscape across cell types in Mecp2 T158 M/y and Mecp2 R106 W/y mice. 47CP2 activity, BDNF-TrkB signaling, and SK2 MeCP2 not only represses 48,49 but also promotes 50,51 the expression of BDNF, depending on its phosphorylation status.52 In fact, MeCP2 has versatile roles in transcription, microRNA expression, and RNA splicing 35,41 and dually regulates gene expression by combining repressor and activator effects.53,54 The present work links BDNF/TrkB signaling with upstream PTP1B and downstream SK2, strengthening the notion that BDNF/TrkB signaling is an essential effector of MeCP2 in Rett syndrome.While our result showing that I h is not affected by Mecp2 deletion strengthens the notion that SK2 is the target of TrkB, it cannot be excluded that other channels are also involved in TrkB signaling with Mecp2 deletion.Our data align well with the study by Krishnan et al. 39 demonstrating that disruption of MeCP2 function increases PTP1B levels and thereby decreases the tyrosine phosphorylation of TrkB. Moreover, or observation that increasing TrkB activity is sufficient to rescue PC dysfunction raises a possibility to design therapies for Rett syndrome.A previous study has shown that protein kinase A can regulate surface SK2 through direct phosphorylation.55 Thus, it is possible that TrkB regulates the phosphorylation sites of SK2, leading to the reorganization of SK2 surface expression in response to the condition of either reduced TrkB or TrkB-Y706E expression.

ASD genes and their impacts on cerebellar physiology
More than 1,000 risk genes have been found for ASD, and the number of genes with de novo mutations is still growing.7][58][59] For example, PCs of Tsc1 PC and Pten PC mice show a reduced excitatory/inhibitory input ratio and simple spike firing, 9,10 whereas PC deletion of Shank2 increases inhibitory currents and inhibits intrinsic plasticity. 10We show here that loss of MeCP2 attenuates PC firing frequency through regulating SK2.Thus, Tsc1, Shank2, and Mecp2 may share a common role in stabilizing intrinsic excitability of PCs.Other examples corroborate the concept that altered synaptic plasticity and intrinsic excitability in PCs in general lead to aberrations in social or cognitive behavior. 60,61Thus, altered excitability of PCs together with abnormal cerebellar development may be the key factors for autistic behaviors.

Deterioration and rescue of cerebello-cerebral connectivity
We demonstrate that expressing TrkB Y706E restores the intrinsic excitability of PCs as well as behavioral phenotypes in Mecp2 cKO mice.Similarly, re-expressing fragile X mental retardation protein (FMRP) in young adult mice with fragile X syndrome has been shown to rescue synaptic organization and visual performance. 62However, restoring the function of a gene may not only act on the pre-and/or post-synaptic sites of dendrites and cell bodies, but in young adult mammals, it might, in principle, also rescue aberrations in the related axonal outputs, especially when the rescuing endeavor takes place close to the critical developmental period.Accordingly, it will be interesting to find out to what extent abnormalities in the cerebello-cerebral pathways in Mecp2 cKO mice can also be rescued by overexpressing TrkB Y706E .Indeed, such an approach might eventually offer an attractive general strategy for treating genetic disorders, in which cerebello-cerebral connectivity is impaired from early on.An interesting comparable example in this respect might be tuberous sclerosis (Tsc1), in which the cerebello-cerebral pathway from cerebellar Crus1 to the ventromedial thalamus and medial prefrontal cortex is also affected from early on and in which the genetic defects also lead to social deficits and repetitive behaviors. 11,63Importantly, not only the cerebello-cerebral pathways linked to Crus1 may be subject to ASD pathogenesis but also those comprising other lobules in the anterior and median cerebellar cortex. 64,65We therefore propose that dysfunction of PC excitability and aberrations in the development of cerebello-cerebral connectivity may both contribute to ASD and that both phenomena may have to be rescued to obtain optimal therapeutic effects.

Limitations of the study
While our study sheds light on the underlying cellular mechanisms of PC dysfunctions caused by MeCP2 deficiency, the abnormalities in the cerebello-cerebral pathways in Mecp2 cKO mice remain unclear.We emphasize the roles of PTP1B-TrkB-SK signaling in PC excitability as well as motor and non-motor behaviors, but the evidence linking these events needs to be further strengthened.It should be noted that this signaling may be a parallel mechanistic pathway that improves the disrupted functionality but not necessarily the mechanism driving dysfunc-tion in Mecp2 mutants.The use of specific and developmentdependent PTP1B KO animals may reveal a more complete picture of how this signaling modulates cerebellar development and animal behaviors.In addition, our present work does not answer whether other channels are involved in TrkB signaling and how SK2 is regulated by TrkB, which remains to be uncovered using more techniques.
containing Gabazine (10 mM).PF-EPSCs were evoked using $4 mA pulses (100 ms).To acquire eIPSCs, stimulation electrodes were placed in the inner fourth of molecular layer with aCSF containing NBQX (5 mM).To record CF-EPSCs, a constant current step (3-30 mA) was applied to a patch pipette that was positioned in granule cell layer close to the vicinity of recorded neuron.The recording was performed at a holding potential of À10 mV, while stimulus intensity and electrode position were adjusted so that an all-or-none response was elicited 70.71 .Recordings were excluded from analysis if the series or input resistance varied by > 15% over the course of an experiment.EPSCs and IPSCs were measured with equal driving force, excitatory/inhibitory input ratio was calculated from EPSCs and IPSCs as E/(E + I).mEPSCs and mIPSCs were recorded in whole-cell configuration in the presence of 0.5 mM tetrodotoxin (TTX) plus Gabazine (10 mM) or NBQX, respectively.The offline analysis of mEPSCs and mIPSCs was conducted using a sliding template algorithm (ClampFit 10, Molecular Device).The criteria for inclusion were 1) an amplitude larger than 6 pA and 2) a rise time (10-90%) longer than 1 ms.Overlapping events were rejected.mGluR1 currents were recorded by driving PCs using a burst PF stimulation (100 Hz) under a holding voltage of À70 mV in aCSF plus NBQX. 72o record input-output curves, the membrane voltage of clamped cells was set at approximately À70 mV.A series of current steps was delivered to PCs with an interval of 30 s. AP amplitude was measured from the threshold to the peak.AP width was measured at half amplitude.AHP was measured from AP threshold to the negative peak.AP threshold was measured in the first derivative of action potential (dV/dt) where the velocity was close to 50 mV/ms.SK2 and tail currents were recorded using a protocol according to previous work. 73To record SK currents, slices were perfused with aCSF supplemented with 1 mM TTX, 1 mM TEA, and/or apamin (100 nM) in the bath perfusion, and PCs were held at À80 mV before stepping to potentials.The steady currents within the duration of 2 ms before the end of steps were measured (gray column) as SK2 currents.To record tail currents, slices were perfused with aCSF containing 1 mM TTX and 1 mM TEA and PCs were held at À50 mV for 50 ms before stepping to potentials between À40 mV and 30 mV (10 mV/step, 100-ms duration, 30-s interval).

Footprint test
Walking gait was tested according to previous work. 67,68With hind paws painted with ink, mice were allowed to freely traverse a clear plexiglass tunnel (100 cm 3 10 cm 3 10 cm, length 3 width 3 height), with a sheet of white absorbent paper (100 cm 3 10 cm, length 3 width) placed on the ground.A darkened cage was placed at the end of the tunnel, encouraging mice to run toward the safe environment.The resulting tracks provided a spatial relationship of consecutive foot prints, from which stride length and stance width were measured.Three step cycles were averaged, with each cycle considered as the distance from one pair of prints to the next.

Rotarod test
After habituation at 5 rpm to the rotarod, control and cKO mice were tested for eight consecutive days.In each session, the velocity of the rotation increased with a constant acceleration of 9 rpm/min 2 from 5 rpm to 50 rpm.Only male offspring were used in rotarod and compensatory eye movement tests, because the performance of female mice depends on the phase of estrous cycle, which may affect chronic motor training and confound the results. 74

Three chamber test
The apparatus consisted of a rectangular plexiglas box (60 cm 3 35 cm 3 3 cm, length 3 width 3 height) evenly divided into three chambers.Age-and gender-matched WT target subjects (S1 and S2) were habituated for 5 consecutive days before the test by being placed inside metal-wired cages.On the test day, test mice were placed in the central chamber for 10-min habituation.S1 was introduced into a wire cage in one chamber and the empty chamber served as an inanimate object with no social valence.The dividers were then raised to allow test mice to freely explore all three chambers over a 10-min session.The time spent in each chamber was recorded and the ratio of (S1-E) to (S1+E) was calculated as the preference index (S1-E).Subsequently, S2 was immediately introduced to the other chamber.Again, test mice spent another 10 min in exploring the entire apparatus.Total time spent in each chamber was recorded and the ratio of (S2-S1) to (S2+S1) was measured as the preference index (S2-S1).Behavioral videos were recorded using ANY-maze video tracking system (Stoelting Co.).

Grooming test
Spray-induced grooming was performed according to previous work. 75Each mouse was placed individually into a standard cage (45 3 45 3 50 cm), which was empty to eliminate digging, for a 5 min habituation and then misted from above with water (23 C) for one time using a manual spraying device.The spray adequately coated the dorsal surface of mouse with mist.The activity of mouse was then recorded for 10 min with a high-speed camera (30 Hz frame rate) and analyzed by a trained observer.Cumulative time and bouts of grooming were scored from the videos as previously described. 76rble burying test A clean cage (50 3 26 3 18 cm) was filled with 4-cm-depth bedding material and 20 glass marbles.The marbles were arranged in an equidistant 4 3 5 grid.Animals were given access to the marbles for 30 min.After 30 min, the burying of marbles was scored.The score was 1 when a marble was 100% covered or 0.5 when it was just partially covered.All scores were assessed by two independent researchers.
TrkB-SK signaling cascade controls intrinsic excitability of PCs d MeCP2 deficiency in PCs alters the cascade and causes motor deficits and autistic behaviors d Expressing TrkB Y706E reduces SK2 currents and restores excitability of PCs d Expressing TrkB Y706E in PCs rescues motor and autistic behaviors of Mecp2 mutant mice In brief Xu et al. show that the PTP1B-TrkB-SK signaling cascade controls the intrinsic excitability of Purkinje cells and that MeCP2 deficiency alters this cascade and causes motor deficits as well as autisticlike behaviors.Expressing TrkB Y706E rescues the excitability of Purkinje cells, motor learning, and autistic behavior of Mecp2 mutant mice.
(F) Gains and phases in VVOR.(G) Training sessions for motor learning (cartoons) consisted of in-phase rotation of the vestibular and visual input with the same amplitude (5 , 0.6 Hz) on the first training day and increasing amplitude of visual input afterward.This training induced a reversal of VOR phase.Gray dots indicate individual data points.See Table S1 for statistics.*p < 0.05, ***p < 0.001.

Figure 3 .
Figure 3. Reduced intrinsic excitability of PCs in adolescent cKO mice (A) A diagram of whole-cell recording with current injection in PCs from the cerebellar vermis, which is applicable to the experiments shown in (B)-(H).(B) Representative APs obtained by injecting a depolarizing current (50 ms; range 0.1-0.3nA) in PCs.Arrowheads indicate the measurement of threshold, amplitude, half-width, and AHP.(C) Representative APs induced by a depolarizing current in control and cKO mice (P30).Note that AHP was blocked by apamin (100 nM), a specific inhibitor of SK2 (n = 5).Right: phase-plane plots (V m vs. dV/dt) (V m , membrane potential; dV, delta change of voltage; dt, delta change of time) for APs on the left.(D) No difference was found in threshold, amplitude, half-width, or AHP between control and cKO mice (P30) (n = 20 for each group).(E) Example traces of current pulse-evoked spikes of PCs from control and cKO mice (P30) in response to a 200-ms/200-pA current injection.Bottom: inputoutput curves for injected currents and firing frequency.Input resistance (R input ) was monitored during the course of the experiment, and the percentages of changes are shown after normalization to the first point.(F) Representative APs of PCs in mice (P90).Right: phase-plane plots for APs.(G) Averages of threshold, amplitude, half-width, and AHP (control, n = 13; cKO, n = 11).

Figure 4 .
Figure 4. Increased SK2 currents in adolescent cKO mice (A) Total and surface SK2 in the cerebellum.Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and flotillin were internal controls of total and surface protein, respectively.Histograms show percentage changes of SK2 in cKO mice relative to control.(B) Left: representative SK2 currents induced by voltage steps in control (black) and cKO (red) mice at P90. Right: current-voltage (I-V) curves derived from P30 and P90 mice, with current amplitudes normalized to the maximal value.Recording site: cerebellar vermis.P30: control, n = 8; cKO, n = 7. P90: control, n = 9; cKO, n = 8. (C) Left: stimulation paradigm and a representative current trace with two regions, where #1 represents an inward Ca 2+ current complex, and #2 represents an outward tail current.Center: representative tail currents evoked by the step to À30 mV.Scale bars in the inset, 50 ms/100 pA.Right: peak tail currents were plotted against step potentials.Recording site: cerebellar vermis.Control, n = 9; cKO, n = 8. (D) Tail currents evoked by the step to À30 mV in Crus1/2 PCs from control (black) and cKO (red) mice (P90).They gray trace is the copy of control for comparison.Scale bars in the inset, 50 ms/100 pA.Right: I-V relationship of tail currents.Gray dots indicate individual data points.Recording site: cerebellar vermis.n = 9 for each group.See Table S4 for statistics.*p < 0.05, **p < 0.01, ***p < 0.001.

Figure 5 .
Figure 5.Effect of BDNF-TrkB signaling on SK2 (A) Surface SK2, phosphorylated TrkB Y706 (pY706), and total TrkB in the cerebellum of WT mice (P90) treated with saline (sal) or BDNF.GAPDH and flotillin were internal controls.n = 6 for each group.(B) Left: representative SK2 currents recorded from PCs treated with sal and BDNF, respectively.Right: I-V curves of sal-and BDNF-treated PCs.Recording site: cerebellar vermis.Control+sal: n = 8; cKO+sal: n = 7. (C) Phosphorylated TrkB Y706 and total TrkB in the cerebellum.n = 6 for each group.(D) Left: representative SK2 currents from cKO PCs treated with sal and BDNF.Right: I-V curves.Current amplitudes were normalized to the maximal value; thus, the y axis indicates normalized currents.Recording site: cerebellar vermis.n = 7 for each group.(E) PTP1B expression in the cerebellum.Gray dots indicate individual data points.n = 6 for each group.See Table S5 for statistics.*p < 0.05, **p < 0.01, ***p < 0.001.

(
C) Tail currents evoked by voltage step to À30 mV in pCre;Mecp2 f/y PCs treated with AAV9-EGFP (n = 12) or AAV9-EGFP-TrkB-Y706E-FLAG (n = 10), respectively.Scale bars in the inset, 50 ms/100 pA.Recording site: cerebellar vermis.(D) Representative APs of PCs in NC (n = 16) and TrkB Y706E (n = 17) groups.Bar graphs show the statistics of threshold, amplitude, half-width, and AHP.Recording site: cerebellar vermis.(E) Example traces of PCs in response to a 250-ms/250-pA current injection.Right: input-output curves for injected currents and firing frequency.For the imitation, y 0 values were À12.8 (NC) and À12.0 (TrkB Y706E ); slope (a) values were 0.19 (NC) and 0.24 (TrkB Y706E ).Recording site: cerebellar vermis.NC, n = 11; TrkB Y706E , n = 9. (F) Time spent on the rotarod for pCre;Mecp2 f/y male mice treated with AAV9-EGFP (NC, n = 6) and AAV9-EGFP-TrkB-Y706E-FLAG (TrkB Y706E , n = 8).The black line is copied from Figure 1C, showing control data.(G) Phase-reversal VOR adaptation test in male mice.The black points are copied from Figure 1G, showing control data.(H) Social interaction tests in male mice evaluated by time spent in each chamber and sniffing time.The black columns are copied from Figure 2A, showing control data.(I) Number of bouts and time spent engaged in the grooming test for male mice.The black columns are copied from Figure 2B, showing control data.Gray dots indicate individual data points.See Table S7 for statistics.*p < 0.05, **p < 0.01, ***p < 0.001.