A small-molecule TrkB ligand restores hippocampal synaptic plasticity and object location memory in Rett syndrome mice

ABSTRACT Rett syndrome (RTT) is a neurodevelopmental disorder caused by mutations in methyl-CpG-binding protein-2 (MECP2), a transcriptional regulator of many genes, including brain-derived neurotrophic factor (BDNF). BDNF levels are reduced in RTT autopsy brains and in multiple brain areas of Mecp2-deficient mice. Furthermore, experimental interventions that increase BDNF levels improve RTT-like phenotypes in Mecp2 mutant mice. Here, we characterized the actions of a small-molecule ligand of the BDNF receptor TrkB in hippocampal function in Mecp2 mutant mice. Systemic treatment of female Mecp2 heterozygous (HET) mice with LM22A-4 for 4 weeks improved hippocampal-dependent object location memory and restored hippocampal long-term potentiation (LTP). Mechanistically, LM22A-4 acts to dampen hyperactive hippocampal network activity, reduce the frequency and amplitude of miniature excitatory postsynaptic currents (mEPSCs), and reduce the frequency of spontaneous tetrodotoxin-resistant Ca2+ signals in Mecp2 mutant hippocampal neurons, making them comparable to those features observed in wild-type neurons. Together, these observations indicate that LM22A-4 is a promising therapeutic candidate for the treatment of hippocampal dysfunction in RTT. Editors' choice: The brain-penetrant BDNF loop domain mimetic LM22A-4 improves synaptic plasticity and spatial discrimination memory in Rett syndrome mice, making it a promising therapeutic candidate for the treatment of hippocampal dysfunction.


INTRODUCTION
Rett syndrome (RTT) is an X chromosome-linked neurodevelopmental disorder that affects approximately 1:10,000 females worldwide (Neul et al., 2010). RTT individuals develop typically until 6-18 months, when neurological symptoms, including intellectual disability, autistic features, deficits in motor control and sensory perception, breathing irregularities and epilepsy disorders begin (Percy, 2011). The majority of RTT individuals carry loss-of-function mutations in methyl-CpGbinding protein-2 (MECP2), which encodes the transcriptional regulator MeCP2 (Amir et al., 1999). Mecp2-deficient mice recapitulate several neurological features of RTT, including impaired hippocampal-dependent learning and memory (Calfa et al., 2011b;, which makes them useful experimental models for preclinical studies . One prominent target of MeCP2 transcriptional regulation is the gene encoding brain-derived neurotrophic factor (BDNF) (Chahrour et al., 2008;Chen et al., 2003;Martinowich et al., 2003), a neurotrophin that plays critical roles in neuronal survival, differentiation, and synapse formation and plasticity (Park and Poo, 2013). BDNF levels are lower in RTT autopsy brains (Abuhatzira et al., 2007;Deng et al., 2007) and Mecp2-deficient mice Schmid et al., 2012;Wang et al., 2006). Since Bdnf overexpression in male Mecp2 knockout (KO) mice rescues several RTT-like neurological and motor symptoms , enhancement of BDNF signaling is considered a potentially useful therapeutic approach for RTT (Katz, 2014). Owing to its low blood-brain barrier permeability that limits the bioavailability of peripherally administered BDNF, therapeutic approaches have relied on BDNF 'mimetics'. One approach involved the use of ampakines, which are known to increase Bdnf expression by their action on AMPA-type glutamate receptors (Lauterborn et al., 2000). Peripheral treatment with ampakines significantly improved respiratory dysfunction in male Mecp2 KO mice (Ogier et al., 2007). A more direct approach is to activate TrkB receptors with small-molecule mimetics of the BDNF loop domain that are designed in silico to interact with their BDNF binding pocket (Massa et al., 2010). One such TrkB ligand with partial agonist activity, LM22A-4, reduced synaptic hyperactivity within respiratory centers in the brainstem, and improved respiratory function in female Mecp2 heterozygous (HET) mice (Kron et al., 2014;Schmid et al., 2012).
Here, we describe how a 4-week systemic LM22A-4 treatment in symptomatic female Mecp2 HET mice improved general phenotype, motor activity and hippocampal-dependent object location memory by activating the BDNF receptor TrkB. Long-term potentiation (LTP) of hippocampal excitatory synaptic transmission, the cellular substrate of learning and memory, was also restored by LM22A-4 treatment in Mecp2 HET mice. Furthermore, LM22A-4 reduced the spatio-temporal spread of neuronal depolarizations in hippocampal slices from Mecp2 HET mice to levels comparable to wild-type (WT) littermates, thus preventing network hyperactivity. In addition, LM22A-4 reduced the frequency and amplitude of miniature excitatory postsynaptic currents (mEPSCs) in CA1 pyramidal neurons of organotypic slices from Mecp2 KO mice. Finally, LM22A-4 reduced the frequency of spontaneous Ca 2+ signals, reflecting quantal transmitter release in cultured hippocampal Mecp2 KO neurons. Together, these observations indicate that LM22A-4 is a promising therapeutic candidate for the treatment of hippocampal dysfunction in RTT.

RESULTS
LM22A-4 improves general phenotypes, motor activity and hippocampal-dependent memory in female Mecp2 heterozygous mice Because RTT occurs primarily in females, we chose to study the effects of a chronic peripheral treatment with LM22A-4 in female Mecp2 HET mice. In additional experiments, we used male Mecp2 KO mice that are homogenously deficient in MeCP2 expression to investigate the underlying cellular mechanisms of LM22A-4 actions. Following an established and successful dosing regime (Kron et al., 2014;Schmid et al., 2012), 4-month-old female Mecp2 HET mice and age-matched WT littermates received intraperitoneal (i.p.) injections of LM22A-4 (50 mg/kg) twice daily for 1 or 2 months. During the treatment, weight and general phenotypes were assessed every week. At the end of the treatment, animals were examined for motor function and then sacrificed under deep anesthesia for assessment of activation of TrkB receptors and their downstream signaling.
To assess general locomotor activity, we performed an open field test and measured the distance traveled during a 10 min period. We found that Mecp2 HET mice traveled significantly shorter distances than WT mice ( Fig. 1C; WT control, n=9; Mecp2 HET control, n=7; P<0.01, one-way ANOVA), and that LM22A-4 treatment increased it to levels comparable with the WT (n=7; P<0.05 vs Mecp2 HET control). However, LM22A-4 had no effect on the distance traveled by WT mice (n=9). Next, we performed dowel crossings to evaluate motor coordination. Mecp2 HET mice took longer than the WT control to walk off the elevated dowel ( Fig. 1D; P<0.05, one-way ANOVA), which was slightly reduced by LM22A-4 treatment.
In a different set of experiments, we performed the object location task to evaluate hippocampal-dependent spatial memory (Barker and Warburton, 2011). During the acquisition phase, animals normally show equal preference for two objects that are placed in an arena, while they exhibit a higher preference for the moved object during the subsequent test phase. The object location memory can thus be estimated by calculating the location preference index (time spent on moved objects over total time spent on two objects). As expected, the preference index in female WT mice was larger for the moved object during the test phase than the acquisition phase HET mice were exposed to two identical objects for 5 min during the acquisition phase, and 2 h later underwent the test trial in which one object was moved to a new location. Object location test was carried out before (top) and after (bottom) 1 month treatment with vehicle or LM22A-4. (F) Quantitative analyses of the levels of phosphorylated TrkB ( p-TrkB Y816 ) relative to the total full-length (TrkB.FL) protein in hippocampal homogenates. Top panel shows representative examples of western immunoblots for p-TrkB Y816 , TrkB.FL and the internal control β-actin. Data are mean±s.e.m. *P<0.05, **P<0.01. (Fig. 1E,top;n=7;P<0.01,. However, we did not observe any increase in Mecp2 HET mice (n=7), suggesting an impairment of objection location memory. Interestingly, systemic LM22A-4 treatment for 4 weeks significantly enhanced the preference index for the moved object in Mecp2 HET mice (Fig. 1E, bottom; n=7; P<0.05, paired t-test), but was unable to further promote it in WT mice (n=7).
To confirm the central target engagement of peripherally administered LM22A-4, we measured the activation state of TrkB receptors and their downstream signaling. Hippocampal homogenates from mice treated with LM22A-4 for 2 months were subject to western immunoblotting. We found a significant increase in the ratio of phosphorylated TrkB Y816 to full-length TrkB in both WT and Mecp2 HET hippocampi after LM22A-4 treatment ( Fig. 1F; n=8 each; P<0.05, one-way ANOVA). Consistently, the ERK and PLCγ cascades were upregulated in WT samples ( Fig. S1A; P<0.001) and the PLCγ signaling was activated in Mecp2 HET mice (P<0.05). We also measured the acute effects of LM22A-4 on TrkB phosphorylation in hippocampal slices prepared from male Mecp2 KO mice, where every neuron lacks MeCP2. Interestingly, the levels of phosphorylated TrkB increased significantly after a 30 min exposure to LM22A-4 (500 nM), but only in Mecp2 KO slices (Fig. S1B, left; Mecp2 KO control, n=14; Mecp2 KO LM22A-4, n=13; P<0.001) and not in WT slices (WT control and LM22A-4, n=16 each), which is consistent with the actions of a partial agonist (Massa et al., 2010). Unlike LM22A-4, the full agonist BDNF (250 ng/ml) activated TrkB receptors in slices from both WT (Fig. S1B, right; n=4; P<0.05) and Mecp2 KO mice (n=4; P<0.01). The effect of LM22A-4 on TrkB phosphorylation was completely blocked by the pan-Trk inhibitor K252a (200 nM) (n=12; P<0.01 vs LM22A-4). Treatment with K252a alone had no significant effect on the ratio of phosphorylated TrkB to full-length TrkB (Mecp2 KO K252a, n=6). Additionally, acute LM22A-4 application did not alter TrkB downstream signaling in either genotype (Fig. S1C). Collectively, these data suggest that LM22A-4 improves general phenotype, motor activity and object location memory in Mecp2 HET mice by activating TrkB signaling in the brain.

LM22A-4 restores hippocampal LTP in female Mecp2 heterozygous mice
We next evaluated the effect of a 2 month treatment with LM22A-4 on LTP at excitatory synapses in the hippocampus, the cellular substrate of learning and memory. We recorded subthreshold excitatory postsynaptic potentials (fEPSPs) at CA3-CA1 synapses and simultaneously imaged voltage-sensitive dye (VSD) signals throughout the entire hippocampal slice ( Fig. 2A,B) (Chang and Jackson, 2006;Li et al., 2016). In slices from control WT mice, the peak amplitude and spatial spread of VSD signals in area CA1 that were evoked by a single afferent stimulation to the Schaffer collaterals increased significantly after induction of LTP with a theta burst stimulus (TBS), following a time course similar to that of the simultaneously recorded field EPSPs ( Consistent with the improvement of hippocampal-dependent memory, LM22A-4 restored LTP of both VSD signals and fEPSPs in Mecp2 HET mice (n=15/10; P<0.01), which were comparable to those in WT mice treated with LM22A-4. Together, these data demonstrate that systemic LM22A-4 treatment restores LTP at hippocampal synapses in female Mecp2 HET mice.

LM22A-4 reduces synaptic strength and network hyperactivity in the hippocampus of female Mecp2 heterozygous mice
The lack of LTP at hippocampal synapses of male Mecp2 KO mice results from the saturation of their plasticity range by already potentiated synapses (Li et al., 2016). To determine if this is also the case in Mecp2 HET mice, and if LM22A-4 restores LTP by reducing synaptic strength allowing for synaptic potentiation, we performed VSD imaging and obtained the input-output (I-O) relationship of fEPSPs and VSD signals (Fig. 3A). VSD signals in CA1 evoked by increasingly stronger stimulation of Schaffer collaterals are proportional and have similar kinetics to fEPSP (Fig. 3B). VSD signals larger than two-times the standard derivation of background noise were used to calculate the cumulative percentage of VSD signal amplitudes. Compared with WT control slices (n=18/10), the cumulative percentage of VSD signal amplitudes was significantly shifted to larger amplitudes in Mecp2 HET slices, indicating that excitatory synapses are stronger in female Mecp2 HET mice ( Fig. 3C; n=12/9; P<0.001, K-S test), as reported for male Mecp2 KO mice (Calfa et al., 2011a;Li et al., , 2016. LM22A-4 treatment did not affect VSD signals in WT mice (n=20/9), but it significantly shifted the cumulative percentage of VSD signal amplitudes to lower levels in Mecp2 HET slices, making it comparable to WT slices (n=16/10; P<0.001 vs Mecp2 HET control; P>0.05 vs WT LM22A-4). Similar results were obtained for the cumulative percentage of the spatial spread of VSD signals (Fig. 3D).
To directly test the effects of LM22A-4 on synaptic strength, we recorded mEPSCs from CA1 pyramidal neurons in organotypic slice cultures treated with LM22A-4 for 2 days. Whole-cell intracellular recordings were performed in the presence of the sodium channel blocker TTX and the GABA A R antagonist picrotoxin (Fig. 4A). To confirm the effects of LM22A-4 on spontaneous neuronal activity, we imaged spontaneous Ca 2+ transients (SCTs) in cultured hippocampal neurons [day in vitro (DIV) 7-13] filled with the Ca 2+ indicator Fura-2 (Fig. 4D, Fig. S2A). To measure the effects of LM22A-4 on quantal excitatory synaptic transmission, Ca 2+ imaging was performed in the presence of TTX (1 µM). Under these conditions, SCTs reflect the activation of voltage-gated calcium channels (VGCCs) during membrane depolarizations caused by temporally summing mEPSPs. Similar to whole-cell recordings of mEPSCs, the inter-event interval (IEI) of SCTs in Mecp2 KO neurons was shorter compared with that in WT neurons (Fig. 4E,F; P<0.05, Student's t-test). Also consistent with its effects on mEPSCs, a single localized application of LM22A-4 to Mecp2 KO neurons significantly increased the IEI of SCTs (n=12 cells/5 coverslips; P<0.05), but had no effect on WT neurons (Fig. 4E,F; n=9/3). This effect of LM22A-4 on SCT frequency is similar to that of locally applied BDNF to both WT and Mecp2 KO neurons ( Fig. S2B; WT BDNF, n=47/3; Mecp2 KO BDNF, n=12/3; P<0.05, one-way ANOVA). Because these SCTs are due to all-or-none Ca 2+ spikes driven by VGCCs, there were no differences between WT and Mecp2 KO neurons or any effects of LM22A-4 or BDNF in SCT amplitude (360/380 nm ratio) (Fig. S2C). Further analysis showed that the half-width of SCTs was significantly shorter in Mecp2 KO neurons, which was increased to WT levels after LM22A-4 or BDNF puffing ( Fig. S2D; P<0.05). The effects on SCT duration are due to differences in decay time, but not rise time (Fig. S2E,F). Taken together, these results suggest that LM22A-4 reduces synaptic strength and tones down network hyperactivity, restoring the optimal range for synaptic plasticity.

DISCUSSION
Several recent studies have indicated that the TrkB partial agonist LM22A-4 is a potentially useful therapeutic agent for several neurological diseases. LM22A-4 treatment was shown to promote motor recovery after hypoxic-ischemic stroke in mice (Han et al., 2012), improve motor impairment in the mouse model of Huntington's disease (Simmons et al., 2013), enhance the recovery of limb function after spinal cord injury in mice (Yu and Wang, 2015), reduce alcohol intake in the mouse model of BDNF polymorphism-related compulsive alcohol drinking (Warnault et al., 2016) and ameliorate respiratory abnormalities in a mouse model of RTT (Kron et al., 2014;Schmid et al., 2012). Our study provides additional support for the potential use of LM22A-4 in RTT by demonstrating that it reverts deficits in hippocampaldependent object location memory and long-term synaptic plasticity in female Mecp2 HET mice. LM22A-4 also reduces atypically stronger excitatory synaptic transmission and network hyperactivity in the hippocampus of female Mecp2 HET mice, which is likely to restore the proper dynamic range of synaptic plasticity. BDNF induces TrkB phosphorylation at mouse Tyr 515 (equivalent to Tyr 490 in mouse TrkA), Tyr 705/706 and Tyr 816 (equivalent to Tyr 817 in human TrkB) in neurons (Minichiello, 2009). TrkB Y515 provides a docking site for Shc that leads to activation of PI3K/AKT and MAPK/ERK signaling pathways, TrkB Y705/706 is located in the activation loop that is responsible for TrkB autophosphorylation and TrkB Y816 creates a docking site for the PLCγ/PKC signaling cascade. As a partial agonist of TrkB receptors, LM22A-4 activates TrkB Y515 , TrkB Y705/706 and TrkB Y816 both in control brains and in several mouse models of human neurological disorders (Han et al., 2012;Kajiya et al., 2014;Massa et al., 2010;Schmid et al., 2012;Simmons et al., 2013;Yu and Wang, 2015). Demonstrating central target engagement during our peripheral treatments with LM22A-4, TrkB Y816 is activated in the hippocampus of both WT and Mecp2 HET mice, consistent with a robust increase in phosphorylation levels of its downstream target PLCγ. Interestingly, although TrkB phosphorylation can be induced by LM22A-4 in the WT hippocampus (Massa et al., 2010), it fails to do so in the WT brainstem (Schmid et al., 2012), suggesting that different brain regions have distinct sensitivity of TrkB phosphorylation sites, or that higher extracellular BDNF levels outcompete a partial agonist (see below).
The significant activation of PLCγ signaling in the hippocampus following LM22A-4 treatment may contribute to the restoration of spatial memory and LTP in female Mecp2 HET mice, because PLCγ is necessary for LTP maintenance in area CA1 (Minichiello et al., 2002); however, LM22A-4 did not affect hippocampal function in WT mice. There are several possibilities why LM22A-4 could act differently in WT and Mecp2 HET mice. First, BDNF levels are lower in Mecp2-deficient mice than in WT mice Schmid et al., 2012;Wang et al., 2006). In WT mice, the efficacy of LM22A-4 may be mitigated by the presence of the full TrkB agonist BDNF, while it can function in Mecp2 HET mice. Second, TrkB receptors may have different sensitivity in response to LM22A-4 treatment. Increased neuronal activity in hippocampal cultures results in prolonged activation of TrkB and its downstream signaling (Guo et al., 2014). As in male Mecp2 KO mice (Calfa et al., 2011a;Li et al., 2016), the hippocampus of female HET mice is hyperactive, which may alter the kinetics of TrkB signaling. Indeed, only acute slices from Mecp2 KO mice showed increased TrkB phosphorylation after 30 min exposure to LM22A-4 (Fig. S1B). In addition, only hippocampal neurons from Mecp2 KO mice responded to LM22A-4 by reducing the frequency of spontaneous Ca 2+ transients. Third, altered network activity (hyperor hypoactivity) may affect not only TrkB receptor sensitivity, but also local downstream signaling at synapses.
The short-term actions of BDNF on excitatory synaptic transmission vary depending on brain regions; for example, it: (1) increases evoked field EPSPs at CA3-CA1 synapses (Kang and Schuman, 1995;Ji et al., 2010); (2) decreases evoked EPSCs in sensory neurons within the nucleus of the tractus solitarius in the brainstem (Balkowiec et al., 2000); (3)  and (5) has no effect at all on field EPSPs and evoked EPSCs at CA3-CA1 synapses (Patterson et al., 1996;Frerking et al., 1998;Gottschalk et al., 1998). Similarly, the long-term effects of BDNF vary; it: (1) increases EPSCs in autaptic cultures of hippocampal pyramidal neurons (Sherwood and Lo, 1999); (2) decreases AMPAR expression in medium spiny neurons of nucleus accumbens (Reimers et al., 2014); (3) increases spontaneous mEPSC frequency, synaptic vesicle docking at active zones, and spine density in hippocampal pyramidal neurons (Tyler and Pozzo-Miller, 2001); and (4) has no effect on evoked EPSCs in pyramidal neurons of rat visual cortical cultures (Rutherford et al., 1998). Such different actions of BDNF on excitatory synaptic transmission may be due to different levels of network activity in each brain region or experimental condition. For example, during prolonged neuronal activity induced by the GABA A receptor antagonist bicuculine, the BDNF scavenger TrkB-Fc prevents scaling down of surface expression of AMPARs (Reimers et al., 2014). Furthermore, BDNF prevents the scaling up of mEPSC amplitude induced by prolonged activity blockade with TTX, while the BDNF scavenger TrkB-Fc mimics scaling-up of mEPSC amplitude during homeostatic synaptic plasticity (Rutherford et al., 1998). Similar to the higher levels of AMPAR during the homeostatic synaptic plasticity, Mecp2-deficient mice have higher surface levels of AMPARs at synapses, which saturates the dynamic range of synaptic plasticity (Li et al., 2016). The actions of LM22A-4 at hippocampal synapses of female Mecp2 HET mice may reflect the restoration of activitydependent AMPAR endocytosis, resulting in smaller evoked EPSPs, mEPSCs and spontaneous Ca 2+ transients driven by mEPSPs. Alternatively, LM22A-4 may enhance TrkB-dependent maturation of GABAergic neurons and synapses, resulting in improved synaptic inhibition and hippocampal network stability.
In conclusion, we present evidence that chronic peripheral LM22A-4 treatment in female Mecp2 HET mice improves motor function and hippocampal-dependent object location memory, and restores hippocampal long-term synaptic plasticity deficits. LM22A-4 exerts these effects by subduing excitatory synaptic transmission and network activity to levels amenable for the induction of synaptic plasticity and behavioral learning and memory. Our findings add to the growing body of literature supporting the high therapeutic potential of the TrkB ligand LM22A-4 for the treatment of RTT and other diseases associated with lower levels of BDNF.

Animals
Breeding pairs of mice lacking exon 3 of Mecp2 (B6.Cg-Mecp2 tm1.1Jae , Jaenisch strain in a pure C57BL/6 background) (Chen et al., 2001) were purchased from the Mutant Mouse Regional Resource Center at the University of California, Davis. A colony was established at the University of Alabama at Birmingham by mating WT C57BL/6 male mice with heterozygous Mecp2 tm1.1Jae female mice (Mecp2 HET), as recommended by the supplier. Genotyping was performed by PCR of DNA samples from tail clips. Hemizygous Mecp2 tm1.1Jae males (Mecp2 KO), develop typically until 5-6 weeks of age, when they begin to exhibit RTT-like motor symptoms, such as hypoactivity, hind limb clasping and reflex impairments. Acute slices prepared from male symptomatic Mecp2 KO mice were used for in vitro treatment. Female Mecp2 HET mice, which develop RTT-like symptoms between 2-3 months of life (Samaco et al., 2013), were used for in vivo treatment because they represent the best model for preclinical studies . Animals were handled and housed according to the Committee on Laboratory Animal Resources of the National Institutes of Health; all experimental protocols were reviewed annually and approved by the Institutional Animals Care and Use Committee of the University of Alabama at Birmingham. In vivo LM22A-4 treatment LM22A-4 was prepared fresh daily. Female Mecp2 HET mice and their agematched WT littermates (∼4 months old) received intraperitoneal (i.p.) injections of either sterile LM22A-4 (50 mg/kg) or vehicle (0.9% NaCl) twice daily for 1 or 2 months, following an established dosing regime (Schmid et al., 2012). Mice were randomly assigned to each treatment.

General phenotype
General phenotype was evaluated according to an established protocol (Guy et al., 2007). We visually scored RTT-like general symptoms consisting of illgroomed condition, reduced motility, imbalanced gait, hind limb clasping, irregular breathing and tremor. Each category was scored every week during the treatment with LM22A-4 or vehicle according to their symptom severity (0: absent; 1: moderate; 2: severe), and the aggregated value was then calculated to reflect general phenotypes during disease progression. We scored the following features: (A) General condition; 0: clean and sleek hair, limpid eyes; 1: ungroomed hair, opaque eyes; 2: severe piloerection, narrowing eye. (B) Motility; 0: free and steady movement; 1: less movement, frequent freezing; 2: almost no voluntary movement. (C) Gait; 0: normal stance; 1: wider spread during the movement of the hind legs; 2: severe walking abnormality, low pelvic elevation. (D) Hind limb clasping; 0: legs spreading outwards; 1: one leg frequently drawing close to the body; 2: two legs always tightening close to the body. (E) Breathing; 0: normal breathing; 1: occasional breathing stops and gasping; 2: frequent breathing stops and gasping. (F) Tremor; 0: no tremor; 1: occasional mild tremor; 2: continuous tremor.

Open field test
WT and Mecp2 HET mice that received either LM22A-4 or vehicle for 2 months were placed in the center of a 60×60 cm open arena. Following 10 min habituation, mice were imaged for 10 min with an IR-sensitive Gigabit Ethernet video camera (ace acA780-75gm, Basler). Recorded activity was analyzed for the distance that these mice traveled. The open field arena was cleaned with 70% ethanol between each trial.

Dowel crossings
Control or LM22A-4-treated mice were placed on the center of a dowel suspended between two platforms. The time to successfully cross the dowel without falling off was recorded for each mouse. If the mouse failed, the mouse was placed back to the center of the dowel. The test lasted 2 min beginning when the mouse was placed on the dowel.

Object location test
The object location test for mice was adapted from an established protocol (Murai et al., 2007). Mice were habituated for 3 consecutive days before testing, by allowing them to freely explore the empty test field (30×40 cm) for 5 min. During the acquisition phase, one mouse was allowed 5 min to freely explore 2 identical objects placed equidistant from each other and from the walls. The mouse was then returned to its home cage for a 2 h consolidation phase. During the test trial, the mouse was returned to the open field where one object had been moved to a new location, and allowed to freely explore the objects for 5 min. The new location of the displaced object was counterbalanced for each mouse to prevent the use of spatial cues outside the test field. The acquisition and test phases were imaged with an IRsensitive Gigabit Ethernet video camera and movies saved directly to a hard disk for offline analyses (ImageJ). The time spent exploring each object in the test phase was scored to calculate a location preference index as T d ×100/ T nd +T d , where T d is the time spent exploring the displaced object, and T nd the time spent exploring the non-displaced object. Exploration was defined as pointing the nose towards the object at a distance of <1 cm and/or touching with the nose; circling or climbing the object was not considered exploration.

Western immunoblotting
For short-term drug treatment in acute brain slices, mice were anesthetized with isoflurane, and the brain was rapidly removed and placed in ice-cold cutting artificial cerebrospinal fluid (aCSF) (87 mM NaCl, 2.5 mM KCl, 0.5 mM CaCl 2 , 7 mM MgCl 2 , 1.25 mM NaH 2 PO 4 , 25 mM NaHCO 3 , 25 mM glucose and 75 mM sucrose, bubbled with 95% O 2 /5% CO 2 ). The brain was cut transversely at 300 μm using a vibrating blade microtome (VT1200S, Leica) and slices were transferred to standard aCSF (125 mM NaCl, 2.5 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , 1.25 mM NaH 2 PO 4 , 25 mM NaHCO 3 and 25 mM glucose, bubbled with 95% O 2 /5% CO 2 ) at 32°C for 30 min, and then allowed to recover at room temperature (24°C) for 1 h prior to drug exposure. Slices were randomly assigned to treatment groups: control, LM22A-4 (500 nM), recombinant human BDNF (250 ng/ml, Promega, G1491), K252a (200 nM, Calbiochem, 420298), LM22A-4 +K252a, and BDNF+K252a. Two to four slices per mouse were perfused under each treatment condition at 4-6 ml/min for 30 min. Following treatment, the hippocampus was quickly dissected from each slice and rapidly frozen on dry ice. For sample preparation from female WT and Mecp2 Het mice receiving in vivo treatment, hippocampi were dissected after the last injection and rapidly frozen on dry ice.

Organotypic slice cultures
Slice cultures were prepared from male Mecp2 KO mice and WT littermates at postnatal day 5-7 (P5-7), as described (Chapleau et al., 2009). Hippocampal slices 500 µm thick were plated on tissue culture plate inserts (Millicell-CM) in 6-well plates with serum-containing culture medium consisting of Neurobasal-A without Phenol Red (Invitrogen), 20% heat-inactivated equine serum (Invitrogen), B27 supplement (Invitrogen), 0.5 mM L-glutamine (Invitrogen) and placed in an incubator at 36°C, 5% CO 2 , 90% relative humidity. Serum was titrated out over 3 days in vitro and treatments were made in serum-free culture medium (Chapleau et al., 2008). At DIV 10-12, medium was aspirated from culture wells and replaced with serum-free medium containing LM22A-4 (500 nM), followed by gentle application of 50 μl of drug-containing medium on top of each hippocampal slice. After 2 days of LM22A-4 treatment, slices were transferred to an immersion chamber for whole-cell patch clamping.

High-speed VSD imaging and electrophysiology
Acute hippocampal slices from female WT and Mecp2 HET mice that received treatment for 2 months were prepared as described above. After recovery, individual slices were stained with the voltage-sensitive fluorescent dye RH414 (30 µM in aCSF, Anaspec) for 1 h at room temperature, and transferred to an immersion chamber continuously perfused (2 ml/min) with aCSF at 32°C and saturated with 95%O 2 /5% CO 2 . VSD signals and extracellular field EPSPs were evoked in CA1 stratum radium by stimulation of Schaffer collaterals. RH414 was excited at 530±50 nm with a phosphor-pumped LED (Heliophor, 89North), and its filtered fluorescence (535±50 nm band-pass, 580 nm beam-splitter, 594 nm long-pass, Semrock) imaged in an inverted microscope (IX71, Olympus) through a 10×0.5NA objective (Fluar, Zeiss) and acquired with a scientific CMOS camera running at 2500 frames per second in full 128×128 pixel resolution (NeuroCMOS-SM128, RedShirt Imaging). Field EPSPs were acquired with Axopatch-2A amplifier (Molecular Devices) in current-clamp mode, filtered at 2 kHz, and digitized at 10 kHz with ITC-18 A/D-D/A interface (Instrutech) controlled by custom-written software on a G5 PowerMac computer (TI-WorkBench, provided by Dr Takafumi Inoue, Waseda University, Tokyo). The I-O relationship of VSD signals and field EPSPs was obtained by delivering three different stimulus intensities with 30 µA increments. LTP of VSD signals and field EPSPs was induced by TBS of afferent fibers, which consisted of 4 trains of 10 bursts of 5 pulses at 100 Hz, with 200 ms between bursts (5 Hz), and 5 s between trains. The peak amplitude of VSD signals was measured in a 3×3 pixel region. The cumulative percentage of VSD amplitude and VSD spatial spread was obtained by measuring the area showing ΔF/F levels 2-times the baseline noise. mEPSCs were recorded in the whole-cell configuration from CA1 pyramidal neurons within hippocampal slice cultures at a holding membrane voltage of −60 mV in the presence of the sodium channel blocker TTX (1 µM) and the GABA A R antagonist picrotoxin (50 µM). mEPSCs were analyzed using MiniAnalysis (Synaptosoft), with a detection threshold of 6 pA. The cumulative probability distribution of mEPSC amplitudes and inter-event intervals were calculated.

Primary cultures of hippocampal neurons and intracellular Ca 2+ imaging
Primary neuronal cultures were obtained from anesthetized postnatal day 0 or 1 (P0-1) male Mecp2 KO mice and WT littermates. Both hippocampi were dissociated in papain (20 U/ml) plus DNase I (Worthington, Lakewood, NJ, USA) for 20-30 min at 37°C, as described (Amaral and Pozzo-Miller, 2007). The tissue was then triturated to obtain a single-cell suspension, and the cells were plated at a density of 50,000 cells/cm 2 on 12 mm poly-L-lysine/laminin-coated glass coverslips, and immersed in Neurobasal medium supplemented with 2% B27 and 0.5 mM glutamine (Life Technologies, Carlsbad, CA). Neurons were grown in 37°C, 5% CO 2 , 90% relative humidity incubators (Thermo-Forma), with half of the fresh culture medium changed every 3-4 days.
Fura-2 AM (Invitrogen) was dissolved in 20% pluronic acid/DMSO, and Fura-2 AM stock solution (5 mM) was then diluted in aCSF to obtain the working solution (5 µM). Neurons (DIV 7-13) on the coverslips were incubated with Fura-2 AM at room temperature for 30 min. The coverslips were then transferred to a recording chamber and continuously perfused with aCSF (1 ml/min). Neurons were imaged in a fluorescence microscope with a 40×0.75NA water immersion objective. A 16 min movie with 4 s frame intervals was taken by exciting Fura-2 AM with 360 nm and 380 nm light from a monochromator (Polychrome-II, TILL Photonics), and its emission at >510 nm was detected with a cooled CCD camera (CoolSNAP, Photometrics). Ca 2+ signals were imaged in the presence of TTX (1 μM) to block voltage-gated Na + channels. LM22A-4 (500 nM) or BDNF (250 ng/ml) were pressure-applied (2 min, 2 pounds per square inch) to WT or Mecp2 KO neurons during imaging. Regions of interest (ROIs) were drawn on individual neuronal cell bodies for Ca 2+ signals and a cell-free region for background subtraction. Spontaneous Ca 2+ transients (SCTs) in the 360/380 nm ratio were automatically detected and analyzed using custom-written codes in MATLAB (MathWorks). The inter-event interval, amplitude, half-width, rise time and decay time of SCTs were calculated.

Statistical analyses
All statistical analyses were performed blinded to the genotype and treatment groups using Prism (GraphPad), MATLAB, and G*Power. Data were compared using Student's t-test for two groups and one-way ANOVA with Bonferroni's post hoc test for more than two groups. The time courses of mouse weight and general phenotype score were analyzed using two-way repeated measures ANOVA. The Kolmogorov-Smirnov (K-S) test was used for comparisons of cumulative percentage distributions. P<0.05 was considered significant.