Abstract
Mutations in the X-linked MECP2 gene are responsible for Rett syndrome (RTT), a severe neurological disorder for which there is no treatment. Several studies have linked the loss of MeCP2 function to alterations of brain-derived neurotrophic factor (BDNF) levels, but non-specific overexpression of BDNF only partially improves the phenotype of Mecp2-deficient mice, which suggests we need to identify cellular pathways that selectively enhance BDNF signaling in appropriate neuronal circuits. We and others have previously shown that Huntingtin (HTT) scaffolds molecular motor complexes, transports BDNF-containing vesicles, and is under-expressed in Mecp2 knock-out brains. Here we demonstrate that promoting HTT phosphorylation at Ser421, either by a phospho-mimetic mutation or inhibition of the phosphatase calcineurin, restored endogenous BDNF axonal transport in vitro, increased striatal BDNF availability and synaptic connectivity in vivo, and improved the phenotype and the survival of Mecp2 knockout mice—even though treatments were initiated only after the mice had already developed symptoms. Stimulation of endogenous cellular pathways may thus be a promising approach for the treatment of RTT patients.
Introduction
MeCP2 (Methyl CpG binding protein 2) is one of the most abundant proteins in the brain, yet the precise nature of its activities remains controversial. It was originally discovered as a DNA methylation-dependent transcriptional repressor (Meehan et al, 1992), but has also been shown to play various roles in chromatin compaction, global gene expression, alternative splicing, and miRNA processing (Skene et al, 2010; Chahrour et al, 2008; Young et al, 2005; Cheng et al, 2014). Interest in this protein rose sharply after it was discovered that mutations in MECP2 cause Rett syndrome (RTT), a severe developmental disorder (Amir et al, 1999; Lyst & Bird, 2015). Females with RTT begin life apparently healthy, but in their second year undergo regression of early milestones, resulting in the deterioration of motor skills, eye contact, speech, and motor control; they develop a range of neurological symptoms, including anxiety, respiratory dysrhythmias, and seizures (Lyst & Bird, 2015; Katz et al, 2016). Whereas loss of MeCP2 function leads to RTT, duplications or triplications of the locus lead to intellectual disability, autistic features, and motor dysfunction, as observed in males with MECP2 duplication syndrome (Van Esch, 2012).
Disease-causing mutations in MECP2 alter the expression of thousands of genes (Chahrour et al, 2008). Among these, brain-derived neurotrophic factor (BDNF), a neuronal modulator that plays a key role in neuronal survival, development, and plasticity (Cheng et al, 2011), is one of the best studied (Chang et al, 2006; Chen et al, 2003, 2015; Sampathkumar et al, 2016). Studies have consistently shown that BDNF is strongly involved in the appearance and progression of the RTT phenotype in mouse models (Chang et al, 2006; Martinowich et al, 2003; Chen et al, 2003, 2015). Mecp2 knock-out mice (KO) present lower BDNF levels and conditional BDNF deletion in Mecp2 KO mice accelerates onset of RTT-like symptoms (Chang et al, 2006). Conversely, conditional BDNF overexpression in the brain of Mecp2 knock-out mice leads to an improvement of certain locomotor and electrophysiological deficits (Chang et al, 2006). Although it is possible that the incomplete rescue is due to the broader regulatory effect of MeCP2 than that of BDNF, it is possible that BDNF overexpression fails to restore synaptic and neuronal function because it does not target the appropriate neurons. In support of this hypothesis, recent evidence suggests that MeCP2 deficiency leads to the disruption of cell-autonomous and autocrine BDNF signaling in excitatory glutamatergic neurons, and that increasing BDNF levels in diseased neurons restores their growth and ability to form synapses (Sampathkumar et al, 2016).
BDNF homeostasis and transport involve huntingtin (HTT) and HTT-associated protein 1 (Hap1) (Saudou & Humbert, 2016; Roux et al, 2012). We previously found that levels of both these proteins are lower than normal in excitatory cortical neurons deficient for Mecp2 (Roux et al, 2012). Here we tested whether activating HTT, by a genetic or pharmacological approach, could improve BDNF homeostasis in Mecp2-deficient neuronal circuits and in Mecp2 KO mice.
Methods
Mouse breeding and genotyping
All mouse lines were on a C57BL/6J genetic background. The Mecp2tm1-1Bird Mecp2-deficient mice were obtained from the Jackson Laboratory and maintained on a C57BL/6 background. The HTT knock-in mice were previously generated by inserting a point mutation in exon 9 (AGC>GAC, Ser>Asp called S421D or AGC>GCC, Ser>Ala called S421A) (Thion et al, 2015). To obtain double mutant mice (Mecp2 KO and S421D or S421D), heterozygous females (Mecp2+/−) were crossed with homozygous S421D or S421A males.
Hemizygous mutant males (Mecp2−/y also called Mecp2 KO) were generated by crossing heterozygous females (Mecp2+/−) with C57BL/6 males. Genotyping was performed by routine PCR technique following a previously described protocol (Roux et al, 2012). Animals were housed under standard conditions of temperature (21 ± 2°C) and humidity (55 ± 5%), with food and water ad libitum in a 12:12h day/night cycle.
Experimental protocols were approved by the ethical committee of the Aix Marseille University and the French M.E.N.E.S.R. minister (Permit Number: 02910.02).
The experimental procedures were carried out in keeping with the European guidelines for the care and use of laboratory animals (EU directive 2010/63/EU), the guide for the care and use of the laboratory animals of the French national institute for science and health (INSERM). All experiments were made to minimize animal suffering. In order to reduce animal suffering, endpoints were fixed as weight loss limit (below 80% of maximum weight), obvious breathing defects or severe injury. In these cases, animals were euthanized with an overdose of pentobarbital (100mg/kg BW i.p., Ceva Santé Animale, La Ballastiere, France).
FK506 chronic in vivo treatment of Mecp2-deficient mice (KO and KO/HTTSA)
From P30, animals received an i.p. injection 3 times a week of either 10mg/kg FK506 in 17% DMSO or vehicle alone (17% DMSO).
Behavioral testing
All mice were weighed every 5 days and assessed for survival. At P35, P45 and P55, each animal underwent a set of behavioral tests to assess motor function, activity and breathing pattern (Figure 2). All testing occurred during the light phase of the light-dark cycle (except for the PhenoRack monitoring).
Open Field
an arena made of clear perspex (38×30 cm), under controlled light conditions (300lx), was used for the FK506 study. For the genetic study, mice were placed in a 1 meter diameter arena under controlled light conditions (300lx). Activity was recorded using the Videotrack software (Viewpoint, Lyon, France) for 20 minutes. Activity and average velocity (cm/s) were determined from the total distance moved and activity duration. Vertical activity (rearing, leaning and grooming) was noted by an experimenter blind of the genotype.
Rotarod
sensorimotor abilities were assessed by the accelerating Rotarod apparatus (Panlab LE-8200, Harvard Apparatus). Each trial started at 4rpm and reached 40rpm speed after 300s. Mice underwent 3 trials, with 5 min rest time in between. The trial ended when the mouse fell off the rod or after 300 s. Latency to fall (in second) was measured and only the best trial was recorded. In the case of mice clinging to the rod, the trial was stopped and the passive rotation was considered a failure in performance like falling (Brown et al, 2005).
Grip strength
A Bioseb grip strength meter (Panlab) was used to measure forelimb strength. Five measures for each mouse was taken and means were calculated from the three best trials.
Whole body plethysmography
To assess apneas, mice were placed in a clear plexiglass chamber (200ml) and allowed to breathe naturally under conscious and unrestrained conditions. After a ~30 min adaptation, breathing was recorded: the spirogram was obtained by recording the pressure difference between the two chambers, then the signal was amplified, filtered and fed to an analog-to-digital converter (sampling frequency, 1 kHz) and finally analyzed by the Spike2 interface and software (v.5.04, Cambridge Electronic Design Ltd., Cambridge, UK). Apneas were defined by more than 1s without breathing, as previously described (Roux et al, 2007). Breathing cycles were divided into four groups according to their duration: hyperventilation (including cycles in the range 0–0.3 s range); ventilation (0.3–0.7 s); hypoventilation (0.7–1 s) and apneas (1–∞ s). The breathing variability was calculated as the mean standard variability. Breathing parameters were obtained from the analysis of quiet period of at least 100 consecutive cycles (Supplementary Fig. 6).
Home cage activity
To analyze spontaneous activity and circadian rhythm, mice were put in individual cages and monitored by the PhenoRack system (Viewpoint, Lyon, France). Mice locomotion was tracked by infrared light during 48h, after a 24h adaptation phase only the last 24h activity was analyzed. The recorded data allowed us to analyze activity, distance (cm) and velocity (cm/sec).
BDNF Immunoassay
P55 (n=4 Wt; n=4 Mecp2 KO; n=4 KO HTTSD; n=5 KO HTTSA) male mice were euthanized by cervical dislocation, and their brains were dissected out within the first 2 min post-mortem. The cortex and the striatum were microdissected using a punching needle (0.5 mm in diameter). Briefly, brain area dissection was performed on cryostat brain sections with the help of a 5X magnifying lens, following their stereotaxic coordinates (Paxinos and Franklin, 2001). We dissected cortical and striatal samples only coming from the same slice in the same rostro-caudal level. Tissue samples were freshly isolated and lysed in 200 μ l of the extraction buffer (100 mM Tris-pH7.5, 125 mM NaCl, 0.1 mM EGTA, 0.1% TritonX-100, Roche® protease inhibitors cocktail), sonicated, centrifugated. The supernatant was stored at − 80 °C until assay. Total protein concentration was determined by using the bicinchronic acid (BCA) protein assay (Thermofisher scientific) and measured with a spectrophotometer (Glomax, Promega). The level of BDNF protein from tissue extracts was determined with the BDNF Emax® ImmunoAssay System (Promega) using the manusfacturer’s instruction. In the present study we measured only free mature BDNF, therefore we proceeded directly to the ELISA protocol avoiding any acid treatment. In each assay, duplicate wells were assigned for each sample. A Victor 4 Perkin Elmer microplate reader was used to measure signal intensity from the wells at 450 nm. A linear standard curve was generated with standard BDNF from 5.8 to 500 pg/ml. The total amount of BDNF per well was calculated based on the standard curve and each sample value was within the linear range. The Relative BDNF value was then calculated by normalizing the amount of BDNF against the total amount of protein input.
Western Blotting
Adult (P55) male mice were euthanized by cervical dislocation, and their brains were dissected out within the first 2 min post-mortem. Brains were dissected on ice and proteins were extracted by sonication and isolated in a lysis buffer containing 50mM Tris–HCl (ph = 7.5), 150 mM NaCl, 2 mM EGTA, 2mM EDTA, 1% Triton X-100, 10nM betaglycerophosphate, 5mM Sodium pyrophosphate, 50mM sodium fluoride and Halt™ proteases and phosphatases inhibitor cocktail (Pierce Thermofisher). Proteins were extracted from neuronal culture with lysis buffer containing 4 mM Hepes, pH 7.4, 320 mM sucrose and protease inhibitor cocktail (Roche). Protein concentrations were determined by the bicinchoninic acid method. After a denaturation step at 96 °C for 5 min, proteins (100 μg) were separated on 4-20% SDS-polyacrylamide gel (Life technology) and transferred onto a nitrocellulose membrane by electroblotting using the Trans-Blot turbo transfer system (Bio-Rad). The membrane was blocked with blocking buffer (Millipore, WBAVDFL01) for 1 h at room temperature. Primary antibodies for HTT (1:1000, rabbit, clone D7F7, CST(Lunkes et al, 2002)), HTT-pS421 (1:500. rabbit homemade previously described(Humbert et al, 2002)), Mecp2 (1:1000, rabbit, CST), TRKB-p816 (Millipore ABN1381), PSD95 (1:1000, Neuromab), tubulin (1:5000, mouse, Sigma) GAPDH (1:5000, rabbit, Sigma), Actin (1:10000, mouse, Millipore), Calnexin (1:1000, rabbit, Sigma C4731) were diluted in the same solution and incubated overnight at 4 °C. The membrane was incubated using appropriate HRP secondary antibodies (donkey anti-rabbit 711-035-152, donkey anti-mouse 715-035-150, Jackson Immunoresearch Laboratory and Biorad ChemiDoc XRS System) or fluorescent secondary antibodies (IRDye 800 CW, IRDye 680 RD, LI-COR and LI-COR odyssey Imager). Quantitative analyses of signal intensity were performed using ImageJ software. For quantification, total (phosphoindependent) protein signals were used to normalize the phospho-protein signal.
Videomicroscopy
Embryonic (E15.5) neuronal cultures were prepared as previously described (Liot et al, 2013). Ganglionic eminences and cortex were dissected and dissociated cortical neurons were nucleofected with ON-TARGET plus mouse Mecp2 siRNA or Non-targeting siRNA 1 (Dharmacon) according to the protocol of Amaxa Nucleofection (Lonza). Then neurons were plated into microchambers coated with poly-D-lysine (0.1 mg/ml) in the cortical and synaptic compartment or poly-D-lysine and laminin (10 µg/ml, Sigma) into the striatal compartment and cultured at 37°C in a 5% CO2 incubator for 5 days. After 24h in culture, cortical neurons were transduced with lentivector coding for BDNF-mCherry into presynaptic neuron chamber for axonal transport analysis as previously described (Virlogeux et al, 2018). Acquisitions were done on microgrooves, at the limit of the synaptic compartment, at 5 Hz for 30 seconds on inverted microscope (Axio Observer, Zeiss) coupled to a spinning-disk confocal system (CSU-W1-T3; Yokogawa) connected to an electron-multiplying CCD (charge-coupled device) camera (ProEM+1024, Princeton Instrument) at 37□°C and 5% CO2. Quantifications of vesicle velocity, linear flow rate and vesicle number was done on 100 µm of axon using KymoTool Box ImageJ plugin as previously described (Zala et al, 2013; Virlogeux et al, 2018). Vesicle velocity corresponds to segmental anterograde or retrograde velocity. Directional net flux is the anterograde cumulative distance minus the retrograde cumulative distance. Regarding vesicle number, a vesicle is considered anterograde when the distance travelled by one vesicle is more anterograde than retrograde.
Statistical analysis
All analyses were performed using GraphPad Prism for Windows/MacOS (GraphPad Software, La Jolla, California, USA, www.graphpad.com). The results are reported as mean +/-standard error of the mean (S.E.M). A p-value < 0.05 was considered to be statistically significant. For group comparisons normality distribution of the datasets were tested. 1-way ANOVAs were performed as indicated with Dunnett’s post-hoc analysis for pairwise comparisons when normal distribution. Kruskal-Wallis test with Dunn’s multiple comparison were performed on datasets without normal distribution. When appropriate, means of two groups were compared with an unpaired two-way t test.
The Kaplan–Meier log-rank test was used for survival studies.
Data availability
The data that support the findings of this study are available from the corresponding author, upon reasonable request.
Results
BDNF transport is slowed in Mecp2-deficient axons
To examine BDNF transport in Mecp2-deficient neurons, we took advantage of the recent development and validation of microfluidic devices to reconstitute a neuronal network and monitor intracellular dynamics (Taylor et al, 2010; Virlogeux et al, 2018). We focused here on the corticostriatal network, which is altered in RTT and Mecp2 KO mice (Roux et al, 2012; Xu et al, 2014). The device consists of a presynaptic compartment containing cortical neurons, a postsynaptic chamber containing striatal neurons, and a synaptic chamber containing corticostriatal contacts. Cortical neurons plated in the first compartment extend axons that connect to striatal dendrites within the synaptic compartment (Fig. 1A), thus creating an oriented corticostriatal network on-a-chip (Virlogeux et al, 2018). This network reproduces the physiological network as we previously showed that cortical neurons from embryonic day (E)15.5 are enriched in CTIP2/TBR1 neurons that correspond to the deepest layers of the cortex that send axons to the striatum. Also, most striatal neurons correspond to enkephalin-positive neurons that are the output-projecting neurons of the striatum (Virlogeux et al, 2018). Cortical neurons were transduced with mCherry-tagged BDNF lentivirus (BDNF-mCherry) and we used using spinning disk confocal videomicroscopy to record the dynamics of BDNF-mCherry containing vesicles within microchannels in the axons (Fig. 1A and Supplement Movie 1). The dynamics of vesicles are in agreement with our recent studies using this technology (Virlogeux et al, 2018; Moutaux et al, 2018).
We transfected neurons with Mecp2 siRNA, which reduced Mecp2 protein levels by 63% compared to WT (Supplementary Fig. 1A). This reduced the mean velocity and overall linear flow of BDNF vesicles reaching the corticostriatal contacts (Fig. 1B). The number of moving vesicles was unchanged (Supplementary Fig. 1B).
HTT phosphorylation status influences anterograde and retrograde BDNF transport
Overexpression of phosphorylated HTT at S421 leads to increased outward transport in cells (Colin et al, 2008) but the role of endogenous HTT phosphorylation in axons is unknown. We isolated cortical neurons from homozygous knock-in mice in which S421 of HTT was replaced by an alanine (HTTS421A/S421A or HTTSA), mimicking the absence of phosphorylation, or by an aspartic acid (HTTS421D/S421D or HTTSD), mimicking constitutive phosphorylation (Thion et al, 2015). Importantly, there is no difference in HTT protein expression between WT, HTTSD, and HTTSA mice (Supplementary Fig.1). We then plated the neurons in microfluidic devices. Phospho-mimetic HTT (HTTSD) significantly increased the speed of BDNF vesicles moving in the anterograde direction (Fig. 1C, Supplementary Movie 1) without affecting the number of moving BDNF vesicles (Supplementary Fig. 2C). In contrast, the absence of HTT phosphorylation (HTTSA condition) significantly increased the retrograde velocity of BDNF vesicles (Fig. 1C). Thus, HTT phosphorylation status influences the velocity of BDNF vesicles in axons.
We next investigated whether HTT phosphorylation status influences the observed dysregulation of BDNF transport in Mecp2 siRNA-transfected neurons (Fig. 1D and Supplementary Fig. 2 D and E). Phospho-mimetic HTT (HTTSD) rescued both anterograde and retrograde transport of BDNF, along with mean velocity of BDNF vesicles and linear flow, back to control values (Fig. 1D). Preventing HTT phosphorylation (HTTSA) restored only the retrograde velocity of BDNF and linear flow rate to control levels (Fig. 1D). The overall effect of HTT phosphorylation on BDNF transport under normal or low-Mecp2 conditions was not due to a change in the number of moving BDNF vesicles (Supplementary Fig. 2, C and E). These results demonstrate that genetically promoting HTT phosphorylation at S421 rescues the transport of BDNF vesicles in projecting corticostriatal siMecp2 neurons.
Constitutive phosphorylation of HTT rescues corticostriatal BDNF transport, increases postsynaptic TRKB phosphorylation and markers of postsynaptic density in vivo
Mecp2 KO mice show altered corticostriatal connections and reduced BDNF levels in the striatum (Roux et al, 2012). We therefore assessed the impact of huntingtin phosphorylation on BDNF transport in vivo. We crossed Mecp2 KO mice with either HTTSA or HTTSD mice. The resulting double mutant male mice, deficient for the Mecp2 gene (KO) and homozygous for the S421A (HTTS421A/S421A) or S421D mutation (HTTS421D/S421D), will from hereon be referred to as KO/HTTSA or KO/HTTSD, respectively. Most of the BDNF protein located in the striatum comes from the cortex by anterograde transport within corticostriatal afferences (Altar et al, 1997). We therefore quantified the level of BDNF proteins using ELISA at the site of translation (the cortex) and the target site (the striatum) of 55-day-old WT mice, KO, KO/HTTSD and KO/HTTSA mice (Fig. 2A). The ratio of striatal and cortical BDNF is an indicator of the in vivo efficacy of BDNF axonal transport to the corticostriatal synapses. The ratio we observed in KO/HTTSD mice (1.57±0,3) was equivalent to what we observed in WT mice (1,57±0.6), and was significantly higher than that in Mecp2 KO mice (1.14±0.1) or KO/HTTSA mice (1.1±0.2). These results suggest that huntingtin phosphorylation rescues corticostriatal BDNF transport in vivo.
We also found that improved corticostriatal BDNF transport in KO HTTSD mice increases levels of TRKB phosphorylation (+ 83.4%, p<0.05) at the postsynaptic level, showing that the BDNF release is stimulated in vivo (Fig. 2A, Supplementary Fig. 3). Finally, as a consequence the postsynaptic marker PSD95 is increased in KO HTTSD striatum compared to KO WT striatum (+ 67%, p<0.05) (Fig. 2A, Supplementary Fig. 3). Together we conclude that promoting HTT phosphorylation stimulates the striatal BDNF pathway and corticostriatal synapse homeostasis in vivo.
Constitutive phosphorylation of HTT reduces the loss of body weight of Mecp2 KO mice and extends their lifespan
We next investigated whether manipulating HTT phosphorylation in vivo could have an effect on Mecp2 KO mouse symptoms. We first extensively assessed the behavior of HTTSA and HTTSD homozygous mice using a modified SHIRPA primary screen (Supplementary Table 1) and various behavioral assays (Supplementary Fig. 4, A-D) and found no significant differences between WT and HTTSA or HTTSD mice at six months in motor activity, strength, coordination, exploratory behavior, or body weight. Then, Mecp2 KO mice carrying the S421D mutation (KO/HTTSD) had a longer lifespan than Mecp2 KO mice, whereas KO/HTTSA mice showed no improvement over that of Mecp2 KO mice (Fig. 2, B and C). The constitutive HTT phosphorylation (KO/HTTSD) mice also had greater body weight than the Mecp2 KO mice, whereas the absence of phosphorylation in the KO/HTTSA mice had no effect on weight (Fig. 2D and Supplementary Fig. 4E).
Constitutive phosphorylation of HTT reduces apneas in Mecp2 KO mice
Breathing disturbances are prominent and deleterious in RTT patients (Kerr et al, 1997) and Mecp2 KO mice (Viemari et al, 2005). Indeed, apnea frequency increased with age in Mecp2 KO mice, from P35 to P55 (Fig. 2E). The KO/HTTSD mice had significantly fewer apneas than Mecp2 KO mice, at both time points (Fig. 2E). The absence of HTT phosphorylation slightly worsened this phenotype.
These data suggest that promoting HTT phosphorylation in vivo largely rescues the apneas observed in Mecp2 KO mice.
Constitutive phosphorylation of HTT improves motor function of Mecp2 KO mice
We next investigated motor coordination on the accelerating rotarod test (Pratte et al, 2011). In agreement with previous studies, Mecp2 KO mice showed a progressive, significant decrease in the latency to fall relative to WT mice (Fig. 2F). Promoting HTT phosphorylation delayed the appearance of motor incoordination until P55, but ablating HTT phosphorylation worsened motor coordination at both time points (Fig. 2F). HTT phosphorylation had no effect on the overall exploration pattern of Mecp2 KO mice in the open-field (OF) test (Supplementary Fig. 2F), but the time before initiation of the first movement was significantly longer for the Mecp2 KO than WT mice (Fig. 2G). The latency to explore the OF arena was further increased in KO/HTTSA mice, whereas it was reduced in the KO/HTTSD mice towards the values observed for WT mice (Fig. 2G).
We next monitored circadian activity and found no deregulation in the circadian rhythm in the different genotypes during the day (Supplementary Fig. 2G). There was, however, a striking suppression of spontaneous locomotion in the absence of Mecp2 during the dark phase (7:00 PM - 7:00 AM) (Supplementary Fig. 4G). Promoting HTT phosphorylation enhanced spontaneous night activity, increasing the distance travelled by the KO/HTTSD mice, although it did not reach the value of the WT mice (Supplementary Fig. 4H). Conversely, the distance travelled by the KO/HTTSA mice was significantly less than that covered by the KO/HTTSD mice. Overall, these results show that promoting HTT phosphorylation improves sensorimotor coordination and locomotor activity in Mecp2 KO mice.
Inhibition of calcineurin by FK506 restores BDNF transport in Mecp2-silenced axons
Since chronic phosphorylation rescues BDNF trafficking in vitro and improves symptoms in mice, we investigated whether pharmacological induction of HTT phosphorylation could be of therapeutic value in RTT. We previously reported that HTT phosphorylation can be inhibited by the protein phosphatase calcineurin (Pardo et al, 2006; Pineda et al, 2009). We therefore evaluated whether FK506, a calcineurin inhibitor, can restore BDNF transport in Mecp2-silenced neurons by increasing HTT phosphorylation. Cortical neurons connected to striatal neurons within the microfluidic devices were transduced with the BDNF-mCherry lentiviral vector and a siRNA targeting either a control sequence or that of Mecp2 (Supplementary Fig. 5A). Five days after plating, we incubated the microfluidic chambers for 1 h with 1 µM FK506 or vehicle and recorded BDNF axonal transport.
FK506 treatment rescued the reduced BDNF trafficking measured after Mecp2 silencing (Fig. 3A, Supplementary Movie 2). Both mean anterograde and retrograde vesicle velocities were increased in siMecp2-transfected neurons, with a significant overall effect on mean velocity and linear flow (Fig. 3B). The number of moving vesicles did not change. FK506-induced calcineurin inhibition thus mitigates deficits of BDNF transport observed in Mecp2-silenced neurons.
FK506 increases HTT phosphorylation in Mecp2 KO mice
To determine whether calcineurin inhibition via FK506 treatment could improve Mecp2 KO mouse symptoms, we treated Mecp2 KO, WT and WT HTTSA mice with FK506 to induce HTT phosphorylation. Mice were first injected intraperitoneally with FK506 (5 mg/kg) and sacrificed 2 h after administration, as previously described (Pardo et al, 2006). We analyzed HTT phosphorylation at S421 by immunoblotting of fresh whole-brain protein extract (Fig 4A). As previously described, administration of FK506 in WT mice increased HTT phosphorylation, by about 1.3-fold (Pardo et al, 2006). In Mecp2 KO mice, FK506 doubled HTT phosphorylation relative to vehicle. FK506 had no effect on HTT phosphorylation in WT HTTSA mice.
FK506 treatment improves respiration and motor function of Mecp2 KO mice and extends their lifespan via HTT phosphorylation
To determine whether FK506 could improve the Mecp2-deficient phenotype, we treated Mecp2 KO mice by intraperitoneal FK506 injection (10 mg/kg), three times a week, starting at P30, the onset of the RTT phenotype. FK506 treatment significantly increased the lifespan of Mecp2 KO mice (Fig. 4B) and induced a significant increase in body weight by P55 (Fig. 4C). The number of apneas was significantly lower in the FK506-treated group than the vehicle group at P35 and remained lower at P55 (Fig. 4D). FK506-treated Mecp2 KO mice performed better on the accelerating rotarod at P50 (Fig. 4E) and showed better forelimb muscle strength than the vehicle group (Fig. 4F).
To establish that the beneficial effect of FK506 is mediated by HTT phosphorylation at S421, we treated KO HTTSA mice by intraperitoneal FK506 injection as previously done (Fig. 4). Interestingly, the lifespan of Mecp2 KO/HTTSA mice was significantly shorter than that of Mecp2 KO mice (Fig. 4B). Moreover, FK506 treatment affected neither motor function nor apneas of KO HTTSA mice (Fig. 4D, E and F). These results demonstrate that HTT phosphorylation at S421 is essential for the therapeutic effect of FK506.
Discussion
Here, we demonstrate that genetically or pharmacologically inducing HTT phosphorylation at S421 rescues BDNF vesicular transport in Mecp2-silenced projecting corticostriatal neurons and improves several pathophysiological features of Mecp2 KO mice. The ability of FK506 as a proof-of-principle candidate to improve Mecp2 symptoms highlights the feasibility of pharmacological stimulation of HTT S421-P in a mouse model of RTT. Importantly, our findings show that HTT phosphorylation can stimulate endogenous machinery to promote BDNF trafficking in the corticostriatal network.
There has been considerable interest in modulating BDNF expression and signaling as a treatment for RTT. Unfortunately, BDNF itself has very low blood-brain-barrier (BBB) permeability, precluding its peripheral administration as a potential therapy. Several studies have used indirect stimulation of BDNF metabolism via fingolimod (Deogracias et al, 2012) or ampakine treatment (Ogier et al, 2007) to circumvent this limitation, but these pharmacological treatments only partially improved the phenotype of Mecp2 KO mice. Daily injection of IGF1, another neurotrophic factor that can cross the BBB and is known to induce Akt phosphorylation (Humbert et al, 2002), has been found to improve survival, locomotor activity and respiratory rhythm in Mecp2 KO mice (Tropea et al, 2009). Finally, potential agonists of the TrkB receptor, such as 7,8-dihydroxyflavone (Johnson et al, 2012) and LM22A-4 (Schmid et al, 2012; Kron et al, 2014; Li et al, 2017), improved breathing patterns in Mecp2 KO mice. It is important to note, however, that the improvements observed in these different studies required that pharmacological treatments be initiated before the appearance of the first symptoms.
Given that RTT is not diagnosed until long after symptoms have begun, we searched for a more practical translational approach based on observed improvements in the RTT phenotype by constitutive phosphorylation of HTT. We selected FK506 because Pardo et al. (Pardo et al, 2006) found that this drug increases phosphorylation of mutant HTT. Importantly, FK506 also induces HTT phosphorylation in Mecp2 KO mice. Our results showed that FK506 treatment, starting at an already-symptomatic stage, improved the lifespan, motor strength and coordination, exploratory behavior, and reduced the frequency of apneas in Mecp2 KO mice in a manner that requires HTT phosphorylation at S421. Thus, part of the beneficial effect of FK506 treatment is due to the stimulation of the HTT-dependent transport of BDNF. It is possible that FK506 also modulates the trafficking of other cargo, such as mitochondria, which could contribute to the in vivo improvement we saw in the Mecp2 KO mice. Future studies investigating such additional effects would be of interest, as they would increase the therapeutic relevance of FK506 treatment for Mecp2 symptoms (Reddy et al, 2012).
We conclude that BDNF trafficking and supply are diminished in the absence of Mecp2 and can be effectively stimulated by promoting HTT phosphorylation. Our results are in accord with a recent study that demonstrated that BDNF acts cell-autonomously in an autocrine loop, as wild-type neurons were unable to rescue growth deficits of neighboring Mecp2-deficient neurons (Sampathkumar et al, 2016). Thus, HTT phosphorylation may increase the bioavailability of BDNF at the synapse through autocrine and paracrine mechanisms in the brains of Mecp2 KO mice and likely represents a strategy of therapeutic interest versus a general, non-synapse-specific increase of BDNF levels in the brain.
Author contributions
YE, JB, NP, YSA, BD, LV, FS and JCR designed experiments. YE, JB, NP, YSA, LS, VM, SC and HV performed experiments. YE, JB, NP, YSA, LS, VM, SC, HV, BD, LV, FS and JCR analyzed the data and YE, JB, FS and JCR wrote the manuscript.
Supplementary Movie 1
BDNF-mCherry-containing vesicles trafficking within WT cortical axon transfected with siControl (siCtl) or siMecp2. HTTSD rescues BDNF-mCherry trafficking in siMecp2 cortical axons.
Supplementary Movie 2
BDNF-mCherry-containing vesicles trafficking within cortical axons transfected with siMecp2 or siControl (siCtl) and treated with 1 μM FK506 or vehicle for 1 h. FK506 rescues BDNF-mCherry trafficking in siMecp2 cortical axons.
Acknowledgements
We thank S. Humbert for sharing the HTTS421A and HTTS421D mice and for advice; the GIN imaging facility (PIC-GIN) for help with image acquisitions; G. Froment, D. Nègre, and C. Costa from the lentivirus production facility of SFR Biosciences (UMS3444/CNRS, US8/Inserm, ENS de Lyon, UCBL); This work was supported by grants from Agence Nationale pour la Recherche (ANR-2012-BSV1-0003-03 ANTARES, J.C.R. & F.S., ANR-14-CE35-0027-01 PASSAGE, F.S.; ANR-15-JPWG-0003-05 JPND CIRCPROT, F.S.), Fondation pour la Recherche Médicale (FRM, équipe labellisée, F.S.), Fondation Bettencourt Schueller (F.S.), Fédération pour la Recherche sur le Cerveau (F.S.), Inserm (J.C.R. & F.S.), Aix-Marseille Université (J.C.R.), AFSR (J.C.R.) and NeuroCoG in the framework of the “Investissements d’avenir” program (ANR-15-IDEX-02, F.S.). F.S. laboratory is member of the Grenoble Center of Excellence in Neurodegeneration (GREEN).