Synaptic Maturation at Cortical Projections to the Lateral Amygdala in a Mouse Model of Rett Syndrome

Rett syndrome (RTT) is a neuro-developmental disorder caused by loss of function of Mecp2 - methyl-CpG-binding protein 2 - an epigenetic factor controlling DNA transcription. In mice, removal of Mecp2 in the forebrain recapitulates most of behavioral deficits found in global Mecp2 deficient mice, including amygdala-related hyper-anxiety and lack of social interaction, pointing a role of Mecp2 in emotional learning. Yet very little is known about the establishment and maintenance of synaptic function in the adult amygdala and the role of Mecp2 in these processes. Here, we performed a longitudinal examination of synaptic properties at excitatory projections to principal cells of the lateral nucleus of the amygdala (LA) in Mecp2 mutant mice and their wild-type littermates. We first show that during animal life, Cortico-LA projections switch from a tonic to a phasic mode, whereas Thalamo-LA synapses are phasic at all ages. In parallel, we observed a specific elimination of Cortico-LA synapses and a decrease in their ability of generating presynaptic long term potentiation. In absence of Mecp2, both synaptic maturation and synaptic elimination were exaggerated albeit still specific to cortical projections. Surprisingly, associative LTP was unaffected at Mecp2 deficient synapses suggesting that synaptic maintenance rather than activity-dependent synaptic learning may be causal in RTT physiopathology. Finally, because the timing of synaptic evolution was preserved, we propose that some of the developmental effects of Mecp2 may be exerted within an endogenous program and restricted to synapses which maturate during animal life.


Introduction
RTT, caused by null mutations of Mecp2 gene [1] is characterized by a period of normal postnatal development followed by a regression of motor language and social skills. This delay in symptoms onset possibly relies to the expression of Mecp2 in mature but not in immature neurons [2], but alternatively, may result from a role of a Mecp2-dependent factor, possibly neurotrophins, in experience-driven synaptic maturation, maintenance and plasticity [3][4][5]. Using a-CamKII promoter-driven recombination, forebrainspecific conditional Mecp2 mutant mice were generated [6] and show that a delayed deletion of Mecp2 in the cortex, the amygdala, the hippocampus and the striatum was sufficient to recapitulate most of behavioral deficits found in constitutive knock-outs (KOs) [6]. Recently, the same group reported that the control of DNA transcription mediated by Mecp2 in the basolateral nucleus of the amygdala (BLA) was responsible for anxiety and fear learning deficits observed in RTT mouse models [7]. A particular role of Mecp2 in amygdala structure is further suggested by its severe volume decrease observed in Mecp2 mutant mice (,40% to compare with ,25% in total brain) [8], a structure which size normally increase during childhood in humans, with the remarkable exception of autistic patients [9].
To date the role of Mecp2 in the key phases of synaptic lifesynaptogenesis, synaptic maintenance and synaptic plasticity -is still conflicting: A dramatic loss of excitatory synapses is observed in Mecp2-deficient primary neuronal cultures, but synaptic marker density appears to be normal in hippocampal slices of symptomatic RTT mice [3]. However, recently, a decrease in the connectivity between layer V cortical neurons was reported in Mecp2 mutant mice [10], but only after symptom occurrence, questioning the role of Mecp2 in the maintenance of synaptic contacts in adult neuronal networks. Also, associative LTP is not systematically affected in Mecp2 deficient preparations [10][11][12], but when defective, is rescued by reintroduction of Mecp2 [13] suggesting that the presence of Mecp2 in the adult is important to maintain synaptic function, possibly due to the activity-dependent control of BDNF production by Mecp2 [14,15].
Here we examined synaptogenesis, synaptic maintenance and synaptic plasticity at cortical and sub-cortical excitatory projections to the LA in a model of RTT, Mecp2 308/Y (KO) mice [16] and their Mecp2 X/Y (WT) littermates. Our data shows that after a period of normal development, synaptic elimination and maturation at cortical-LA synapses were exaggerated in absence of Mecp2. Importantly, the other major excitatory projections to the LA, the so-called Thalamo-LA synapses, were unaffected by the absence of Mecp2. Thus, we propose that some of the developmental effects of Mecp2 are exerted within an endogenous program and restricted to synapses which maturate during animal life.

Synaptic maturation and elimination at excitatory projections to the LA in WT mice
The amygdala is one of the key brain structures for emotional memory acquisition and storage [17,18], and substantial evidence supports the notion that Thalamo-LA and Cortico-LA synapses change during fear acquisition [19,20]. Surprisingly, very few is known about the evolution of the synaptic properties at Cortico-LA and Thalamo-LA synapses during animal life, yet studied in juvenile [21][22][23] or young adult mice [20,24,25]. Thus, because in Mecp2 308/Y animals symptom onset occurs only after 6-12 weeks of age [16], we first characterized synaptic properties at excitatory projections to the LA in pre-symptomatic (1month old) and postsymptomatic (until 12 months) Mecp2 308/Y and their Mecp2 X/Y littermates (Fig. 1).
Synaptic physiology at Cortico-LA and Thalamo-LA synapses was examined in acute coronal brain slices using whole-cell patchclamp recordings from LA principal neurons [21,26] (Fig. 1). Monosynaptic excitatory postsynaptic currents [EPSCs] were elicited by stimulating cortical axons in the external capsule [20][21][22] (Fig. 1A 1 ), and thalamic axons within the internal capsule [26] (Fig. 1B 1 ). We systematically evaluated the global synaptic strength (Input/ Output curves, Fig. S1), the paired pulse ratio (PPR) at 50 milliseconds inter-spike intervals [ Fig. 1], and the postsynaptic responses to 2 second, 20 Hz stimulations trains, allowing the evaluation of the size of the ready releasable pool [RRP] and the refilling rate [27] (Fig. S2) (Fig. 1). Juvenile Cortico-LA and Thalamo-LA synapses differ in their AMPAR/NMDAR ratio (28, but see 22,29), the size and molecular equipment of their dendritic spine [22] and the locus of associative plasticity [21,22,30]. Here we added to this list a different mode -phasic versus tonic -of neurotransmitter release. Indeed, as shown in Fig. 1B, Juvenile Thalamo-LA synapses typically behave as phasic synapses [31] with a small amount of PPR [1.2360.07, n = 18 cells] and a low resistance of neurotransmission to high frequency stimulations due to a low refilling rate [3564%, n = 15, Fig. 1B]. Interestingly, these and other synaptic parameters at these synapses are rather constant during animal life (Fig. 1, Table S1), albeit not excluding some subtle changes such as increased quantal content per axon or decrease in the number of silent synapses which have been reported in juvenile and adult Thalamo-LA synapses in another mouse strain [25].
In contrast, to Thalamo-LA synapses, juvenile Cortico-LA synapses fulfill typical criteria of ''Tonic'' synapses [31], with high PPR (1.6860.08, n = 22 cells, Fig. 1C) and high refilling rate (6167%, n = 27, Fig. 1C). Surprisingly, when assessed in adult animals (.6 months of age), these two parameters evolved in such a way that adult Cortico-LA synapses behave as phasic synapses, similar to their neighboring Thalamo-LA synapses (Fig. 1D). This specific developmental switch in the synaptic modality support an experience-driven maturation of Cortico-LA afferents but not Thalamo-LA synapses during mouse adult life.
In addition to synaptic maturation, we observed a significant synaptic elimination in adult WT animals illustrated by the lower frequency of miniature EPSCs recorded in LA principal cells in presence of the voltage-sensitive sodium channel blocker Tetrodotoxin (TTX, 10 mM) in adult animals (Table S1). As shown in Fig. 1, this may be partially contributed by a loss of functional Cortico-LA synapses as the maximal EPSC in adult is lower than in juvenile animals (Cortico-LA EPSC max : 1 month: 14976225 pA, Adult: 794682 pA, p,0.001, Fig. 1C). Less synaptic elimination is observed at Thalamo-LA afferents (1 month: 11456132 pA, Adult: 882665 pA, p,0.05, Fig. 1C) reinforcing the idea that synaptic maturation at these synapses is almost completed after 1 month of age in mice. All together our findings suggest a strong maturation of Cortico-LA synapses during the mouse adulthood, whereas Thalamo-LA synapses appear globally fixed even in juvenile animals.

Synaptic maturation and elimination at excitatory projections to the LA in Mecp2 deficient mice
We then performed the same analysis of synaptic function at Cortico-LA and Thalamo-LA synapses from Mecp2 308/Y animals (Fig. 2), and compared the results to the synaptic framework built from the Mecp2 X/Y WT littermates (Fig. 1).
First, despite the fact that, in extent, some synaptic parameters change exaggeratedly (see below), the global evolution of Cortico-LA and Thalamo-LA synapses is preserved in absence of Mecp2. In Mecp2 308/Y animals, we still observed a switch in the modality of Cortico-LA synapses (Fig. 2). In addition, as in WT mice, Thalamo-LA synapses showed very little maturation during adulthood (Fig. 2C). However, we gain key information about the role of Mecp2 in amygdala development from this comparative longitudinal study: First, all synaptic parameters collected in juvenile animals were independent to the presence or absence of Mecp2. Indeed, in addition to the synapse specific parameters presented in Fig. 2B and 2C, no changes in the frequency and amplitude of miniature EPSCs, in the AMPA/NMDA ratio and RRP size at both excitatory synapses could be detected (Table S1). Further, associative Long term potentiation (LTP) was similar at juvenile Cortico-LA and Thalamo-LA synapses in Mecp2 308/Y and Mecp2 X/Y animals (Table S1). Thus, our data strongly support the notion that Mecp2 is dispensable for synaptogenesis and synaptic function in the young mouse amygdala. Second, we observed that two of the synaptic parameters developmentally regulated in WT mice were more dramatically affected in adult Mecp2 308/Y mice, albeit evolving with a similar timing (EPSC max Adult: WT: 794682 pA, KO: 392659 pA, p,0.001; PPR: WT: 1.3760.05, KO: 0.9660.03, p,0.001, Fig. 2B). This suggests that both synaptic elimination and maturation at Cortico-LA projections are amplified in absence of Mecp2. Importantly, these findings were confirmed by other sets of data: We first tested minimal stimulations of cortical inputs in old Mecp2 308/Y and Mecp2 X/Y animals (Fig. S3). In absence of Mecp2, the minimal Cortico-LA response was indeed significantly lower than in WT mice (Fig. S3). In adult Mecp2 308/Y mice, the miniature EPSC frequency was also decreased (Table S1) as well as the coefficient of variation at Cortico-LA but not Thalamo-LA synapses (Table  S1). Together with the strong decrease in Cortico-LA PPR in adult Mecp2 308/Y mice ( Fig. 2), we conclude that the prolonged absence of Mecp2 strongly increased release probability at Cortico-LA synapses, possibly compensating for the progressive synaptic elimination.
Intact associative LTP at Cortico-LA synapses in early and late symptomatic Mecp2 308/Y mutant mice Next, we examined associative LTP at Cortico-LA synapses of adult Mecp2 KO mice. Indeed, it is now well established that like its counterparts in other brain areas [32][33][34][35][36], LTP at Cortico-LA synapses is predominantly expressed by a presynaptic, cAMP/ PKA-dependent increase in the probability of release [20,21,23,[37][38]. Thus, if the absence of Mecp2 modified release probability by the mean of a PKA-dependent mechanism, a saturation of presynaptic LTP expression may be expected in adult Mecp2 308/Y animals. Accordingly to earlier studies reporting that changes in release probability at Cortico-LA synapses were consecutive to fear learning [20], the amount of presynaptic associative LTP was lower in adult WT animals as compared to juveniles (WT 1month: 210626% of baseline, WT Adult: 167618% of baseline, p,0.05, Fig. 3). Importantly, at each tested post natal age the absence of Mecp2 never impacts associative presynaptic LTP (KO Adult: 147615% of baseline, p.0.05 as compared to WT adult, and p,0.05 as compared to baseline, Fig. 3) suggesting that the increase in release probability at Mecp2 deficient adult Cortico-LA synapses may not be linked to an over-activation of the cAMP/PKA/Rim pathway [38]. This hypothesis was confirmed by pharmacological activations of adenylate cyclase (AC) by Forskolin (50 mM), a condition which we used previously in juvenile animals to demonstrate the role of the cAMP/PKA/Rim in the control of release probability at Cortico-LA synapses [38]. Indeed, when applied on 10-12 months old adult mice preparations, Forskolin failed in exerting its effect on evoked synaptic transmission in both WT and KO preparations (Fig.S4), providing a plausible explanation to the progressive decrease of presynaptic LTP levels with age in both genotypes. It also indicates that some qualitative aspects of fear learning, such as generalization [39,40], may be preserved in late symptomatic RTT mouse models.

Synaptic elimination is restricted to the AMPA component at Cortico-LA synapses
Next, we examined if the specific Cortico-LA synapse elimination changed the feed-forward recruitment of local GABAergic interneurons, a key physiological parameter for LA function [41]. First, we tested the possibility of a specific loss of AMPAR at Cortico-LA synapses. Indeed, in WT mice we observed a developmental regulation in the NMDA/AMPA ratio at these synapses (Table S1). In Mecp2 mutant mice however, this modulation does not seem to occur, the NMDA/AMPA ratio remaining constant during the animal life (Table S1). Together with the strong decrease of the EPSC max in Mecp2 308/Y animals (Fig. 2), these data support a role for Mecp2 in stabilizing AMPAR within Cortico-LA post-synapses.
Next, we wanted to test if the removal of AMPAR could be generalized to other cortical synapses. We previously reported that the feed-forward recruitment of local interneuron was highly depending on AMPAR function at Thalamo-interneuron synapses [26]. Thus we thought to test for the efficacy of incoming glutamatergic inputs to generate GABA A -mediated currents in LA principal cells (Fig. 4B). To this aim, excitatory and inhibitory current in principal cells were recorded at 270 mV and 210 mV respectively (Fig. 4B). Interestingly, in young animals both thalamic and cortical afferent we found to be efficient in activating local interneurons [42,43]. In adult mice however, both excitatory inputs were not equal in recruiting local interneurons: when the stimulation intensity was set to elicit an EPSC of ,100pA amplitude, a higher IPSC was observed when stimulating within the internal (Thalamo-LA: 304644 pA, Fig. 4C and 4D) as compared to external capsule [139629 pA, p,0.001, Fig. 4C and  4D). In absence of Mecp2, a specific increase in feed-forward efficacy was observed at cortical afferents in the adult (259653 pA, p,0.05, Fig. 4D). These results suggest a specific loss of AMPARmediated currents at contacts between cortical afferents and LA principal cells.

Discussion
Emotional deficits are present in Rett syndrome patients and Mecp2 mutant mice [44], that can be mimicked by specific deletion of Mecp2 within the forebrain [45,46], and, although only partially, by inactivating Mecp2 in the basolateral amygdala [7]. Thus, because Mecp2 appears to affect synaptic function in many different brain areas [4], we thought to characterize synaptic function at excitatory projections to the LA, required for associative fear learning, perturbed in some Mecp2 mutant mice [46]. However, as compared to other monogenetic causes of mental retardation, behavioral and cognitive symptoms in RTT patients and RTT mouse models are significantly delayed [8]. This introduces the notion of one or more critical windows in the physiopathology of the disease, and therefore forces the establishment of longitudinal frameworks to study synaptic and/or neuronal function in absence of Mecp2. It is noteworthy that most of fear learning studies are performed on adult (3-4 months old animals) whereas most of in vitro data are obtained in juvenile animals (,1 month). Thus, in order to reconcile cognitive and physiological data, we examined a broad repertoire of synaptic properties at Cortico-LA and Thalamo-LA synapses in presymptomatic (1month old) and post-symptomatic (until 12 months) Mecp2 308/Y and their Mecp2 X/Y littermates.
Here, we describe a progressive elimination of Cortico-LA starting after 3 months of age and achieved after 6 months (Fig. 1). This process is amplified by the absence of Mecp2, but with a preserved timing (Fig. 2). Other synaptic parameters, such as PPR and refilling rate, evolve similarly, suggesting that this lack of synaptic maintenance may be the end result of synaptic maturation at this excitatory projection. Interestingly, the NMDA/AMPA ratio at these synapses, which progressively decline with age, follows a similar time course (Fig. S5), suggesting that in parallel to synaptic elimination, a reinforcement of AMPAR at subsisting contacts occur. With respect to this, the absence of Mecp2 appears to amplify synaptic elimination without allowing AMPAR redistribution, suggesting a specific role of Mecp2-dependent factors in the maturation of synaptic contacts within the LA.
One intriguing question is the effect of age and of Mecp2 deletion onto Cortico-LA but not Thalamo-LA synapses. Indeed, given the role of Mecp2 in maintaining the integrity of neuronal dendritic arbor [47] one may have expected strong and unspecific rearrangements of synaptic contacts in symptomatic MeCP 308/Y mice. However, as reported by Moretti and colleagues, cortical neurons of 12-15 months old Mecp2 308/Y animals did not exhibit strong modifications of dendritic arbor or distribution of synaptic markers [12]. Several of our results support this preserved architecture: in late symptomatic -over 6 months old -Mecp2 308/Y animals, the EPSC max at Thalamo-LA synapses (Fig. 2), the frequency of spontaneous and amplitude of Thalamo-LA evoked IPSCs were preserved ( Fig. S6 and Fig. 4). We also did not detect any difference in the cell capacitance or membrane resistance of LA principal cells according to age or genotype (Table S1). Thus even if some of these arguments can be discussed [25], we think that no global posts-synaptic defect is present in LA principal cells of Mecp2 mutant mice.
Remains the specific effect observed at cortico-LA synapses. One possible explanation is a presynaptic effect of Mecp2 thus pushing towards a role of Mecp2 in the cortex. Indeed, last years, a crucial role of Mecp2 in supporting the BDNF/TRK pathway has been proposed [14,48]. This hypothesis emerges from several observations linking Mecp2 to BDNF production [14,15,48]. In this discussion frame, we can note that BDNF levels are lowered in the cortex of symptomatic animals [14,48], and that BDNF overexpression can correct for the lack of cortical neuron discharge observed in Mecp2 mutant mice [14]. This is further supported by the fact that changes in PPR and refilling rates are considered to be presynaptic parameters [but see 31,49]. Thus, even in the absence of any study examining the role of Mecp2 within the thalamus, a specific role of Mecp2 at cortical pre-synapses is plausible. Then another important question comes: Is the amygdala defect presented reflects an essential role of Mecp2 in the LA or a secondary consequence of a dysfunction of neurons projecting to the LA? Indeed, when compared to the presymptomatic effects of Mecp2 deletion in the cortex [10,12], the physiological defects observed in the amygdala are delayed for several months. It is however possible that this difference in timing reside in the expression of delayed endogenous programs, in other words, that the cortex and the amygdala have different critical windows for synaptic maturation. Towards this hypothesis, several longitudinal studies showed that in the mouse, cortical synaptic maturation is achieved after 4-5 weeks of life [50].
There is an alternative to a pure presynaptic -cortical -action of both Mecp2 and age. Indeed, some heterogeneity in the morphology and molecular equipment of post-synaptic compartments at Cortico-LA and Thalamo-LA synapses has been reported in juvenile mice [22,28,30]. In a sense, Cortico-LA post-synapses appear rather immature as compared to thalamic ones. Thus one likely possibility is that during animal life cortical dendritic spines maturates, a process which would then depend on the presence of Mecp2.
To conclude, we here show that at excitatory projections to the LA all synaptic effects associated with the absence of Mecp2 were following a developmental time course similar as in WT mice. Thus we propose that some of the developmental effects of Mecp2 are exerted within an endogenous program and restricted to synapses which maturate during animal life.

Ethic statement
All experiments have been conducted using a protocol approved by the European and French guidelines on animal experimentation. All animal experiments were approved by the direction des services veterinaries du Bas-Rhin, Alsace, France (authorization number 67-313 to YH).

Slice preparation
Standard procedures were used to prepare 300 mm thick coronal slices from Mecp2 308/Y and Mecp2 X/Y males on a C57BL/6J background. Briefly, the brain was dissected in ice-cold artificial cerebrospinal fluid (ACSF), mounted on an agar block and sliced with a vibratome (Leica VT1200s; Germany) at 4uC. Slices were maintained for 45 min. at 35uC in an interface chamber containing ACSF equilibrated with 95% O 2 /5% CO 2 and containing (in mM): 124 NaCl, 2.7 KCl, 2 CaCl 2 , 1.3 MgCl 2 , 26 NaHCO 3 , 0.4 NaH 2 PO 4 , 18 glucose, 4 ascorbate, and then for at least 45 min. at room temperature before being transferred to a superfusing recording chamber.

Recordings
Whole-cell recordings from LA principal neurons were performed at 30-32uC in a superfusing chamber as previously described [30]. Neurons were visually identified with infrared videomicroscopy using an upright microscope equipped with a 60x objective. Patch electrodes (3)(4)(5) were pulled from borosilicate glass tubing and filled with a low-chloride solution containing (in mM]: 140 Cs-methylsulfonate, 5 QX314-Cl, 10 HEPES, 10 phosphocreatine, 4 Mg-ATP, and 0.3 Na-GTP [pH adjusted to 7.25 with CsOH, 295 mOsm). All experiments were performed in the presence of picrotoxin [100 mM]. Monosynaptic EPSCs exhibiting constant 10-90% rise times and latencies were elicited by stimulation of afferent fibers with a bipolar twisted platinum/ 10% iridium wire (25 mm diameter).

Data acquisition and analysis
Data were recorded with a Multiclamp700B (Molecular Devices, USA), filtered at 2 kHz and digitized at 10 kHz. In all experiments, series resistance was monitored throughout the experiment, and if it changed by more than 15%, the data were not included in the analysis. Data were acquired and analyzed with pClamp10.2 (Molecular Devices). LTP was induced using a pairing protocol consisting of a single burst of presynaptic activity (80 stim at 10 Hz) applied together with a postsynaptic depolarization (+30 mV) during the same 8 sec period. Changes were quantified by normalizing and averaging EPSC amplitudes during the last 5 min. of the experiments relative to the 5 min. of baseline prior to LTP induction or drug application. All values are given as means 6 standard error of the mean (SEM). Mean values were compared between genotypes using either unpaired Student's t-test or Mann-Whitney (MW) test as appropriate.

Supporting Information
Table S1 Top: Basic cellular and recording parameters were obtained in juvenile and adult WT and KO mice. The unit of a particular measurement is indicated on the left, and the number of recorded cells in brackets. A blue text means a difference between juveniles and adult WT, and red text between WT and KO at a given postnatal age. Methods table S1 The input resistance (Rinput) and cell capacitance (tau/Rinput) were extracted from 10 mV hyper-polarizing steps in voltage clamp mode. Miniature EPSCs (mEPSCs) were automatically detected by a templatebased routine in the ClampFit 10.0 software. Events were then fitted by a standard bi-exponential equation to extract un-noisy amplitude. In general, mEPSC frequency was determined by visual detection (number of peaks) during a 0,5 to 10 minute period depending on the event frequency (a minimum of 200 events was analyzed). The coefficient of variation is determined using at least 30 consecutive EPSCs. No difference between mean EPSC amplitude was detected between considered groups.