Histone methyltransferase Ash1L mediates activity-dependent repression of neurexin-1α

Activity-dependent transcription is critical for the regulation of long-term synaptic plasticity and plastic rewiring in the brain. Here, we report that the transcription of neurexin1α (nrxn1α), a presynaptic adhesion molecule for synaptic formation, is regulated by transient neuronal activation. We showed that 10 minutes of firing at 50 Hz in neurons repressed the expression of nrxn1α for 24 hours in a primary cortical neuron culture through a transcriptional repression mechanism. By performing a screening assay using a synthetic zinc finger protein (ZFP) to pull down the proteins enriched near the nrxn1α promoter region in vivo, we identified that Ash1L, a histone methyltransferase, is enriched in the nrxn1α promoter. Neuronal activity triggered binding of Ash1L to the promoter and enriched the histone marker H3K36me2 at the nrxn1α promoter region. Knockout of Ash1L in mice completely abolished the activity-dependent repression of nrxn1α. Taken together, our results reveal that a novel process of activity-dependent transcriptional repression exists in neurons and that Ash1L mediates the long-term repression of nrxn1α, thus implicating an important role for epigenetic modification in brain functioning.


Identification of the nrxn1α promoter binding proteins by the synthetic zinc finger protein.
To identify the transcriptional regulators for activity-dependent repression of nrxn1α in neurons, we designed a synthetic zinc finger protein that targeted the nrxn1α promoter (GST-ZFP-pnrxn1α ) to directly pull down the chromatins around nrxn1α in the mouse brain. Zinc finger proteins bind to a specific DNA sequence via the interaction between amino acids in the recognition domain of the protein and the sequence specific DNA (Table 1). To test the binding capacity of the designed GST-ZFP-pnrxn1α , an electrophoretic mobility shift assay was performed as previously reported 31 . GST-ZFP-pnrxn1α showed dose dependent binding to the DNA fragment of the nrxn1α promoter. As a control, GST alone did not show binding activity ( Fig. 2A). To further test the binding specificity, the DNA fragments with different tags were used for the competition assay. Only the DNA fragment of the nrxn1α promoter sequence could compete with the TagA-nrxn1α promoter sequence to bind to the GST-ZFP-pnrxn1α protein (Figs 2B and 3C). Thus, GST-ZFP-pnrxn1α specifically binds to the nrxn1α promoter.
With the designed GST-ZFP-pnrxn1α , we pulled down the proteins binding to the nrxn1α promoter in the mouse brain (Fig. 2B). The whole brains of three 3-month-old C57BL/6 mice were fixed with formaldehyde to cross link DNA with the adjacent proteins associated with the chromatin in vivo. The chromatin was sheared to ~1500 base pair sections and then used to be affinity purified by the GST-ZFP-pnrxn1α collection. The DNA fragments and proteins enriched by GST-ZFP-pnrxn1α were then analyzed by PCR and mass spectrometry.
To test the specificity of the chromatins enriched by GST-ZFP-pnrxn1α pull down, elution was subjected to PCR-based analysis. The promoter region of nrxn1α was enriched in the GST-ZFP-pnrxn1α pull-down samples compared to the GST control pull-down samples (Fig. 2C). Other chromatin regions did not show such enrichment, such as the promoter of β -actin, β -tubulin, nrxn2α , nrxn3α , NR2A, GAPDH (Fig. 2C). Importantly, the enrichment of precipitated chromatin increased significantly near the TSS site of nrxn1α (Fig. 2D, bottom), indicating the specificity of this method.
Proteins associated with the nrxn1α promoter were enriched by the GST-ZFP-pnrxn1α pull down, detected by sliver staining, and identified by high-resolution mass spectrometry ( Fig. 2E and Table 2).
Next, NG108-15 cells were co-transfected with pCAG-ChR2 and shRNAmir-Ash1L and then stimulated with LEDs for 15 minutes at 50 Hz (Fig. 4D). The expression of egr1 was induced 3 hours after optical stimulation and returned to basal levels 24 hours later, thus indicating the cells were activated ( Ash1L-deficient mice brains exhibit normal morphology. To further confirm the effect of Ash1L in activity-dependent nrxn1α repression, we generated Ash1L-knockout mice using the CRISPR/Cas9 system 36 . sgRNA and Cas9-coding mRNA were co-injected into C57BL/6 zygote pronuclei. We obtained 28 neonates, 4 of which showed a mutation at the target locus that was revealed by DNA sequencing. One founder was a heterozygously mutated clone with an 11 bp deletion in exon 2 leading to a null allele of ash1L (Fig. 5A). Such mutations were detectable in each animal using PCR (Fig. 5B). The homozygous offspring was absent at 1 week of age from intercrosses of the Ash1L(− /+ ) mice (Fig. 5B,C), indicating that the homozygous mutant of Ash1L had embryonic lethality. The Ash1L(− /+ ) mice were backcrossed to wild-type C57BL/6 mice for more than 5 generations to eliminate any potential off-target effects. Meanwhile, the expression of the Ash1L protein was significantly reduced in the hippocampus of 3-month-old Ash1L (− /+ ) mice (Fig. 5D, right panel; P < 0.05, Student's t-test). Furthermore, the transcription of Ash1L in the hippocampus was relatively high compared to other brain areas (Fig. 5E).
Consistent with the neuronal cultures studies, Ash1L(− /+ ) mice exhibited up-regulation of nrxn1α mRNA in the hippocampus ( Fig. 6A; P < 0.05, Student's t-test), suggesting that Ash1L is involved in the regulation of nrxn1α transcription. Such regulation might not be due to the overall changes in the epigenetic landscape, as the H3K36me2 and H3K4me3 in the hippocampus were not significantly changed (Fig. 6B). In the hippocampus of Ash1L(− /+ ) mice (3 months of age), the ChIP experiments revealed that the enrichment of H3K36me2 at the nrxn1α promoter region was significantly decreased ( Fig. 6C; P < 0.05, Student's t-test). In contrast, other histone Synthetic zinc finger protein GST-ZFP-pnrxn1α that targeting the nrxn1α promoter is shown. The target sequence in the nrxn1α promoter is listed in the 3′ to 5′ direction. The amino acid sequence of each zinc finger for binding helices is also listed.
modifications in the nrxn1α promoter were unchanged, including H3K4me3, H3R17me2 and H3K9ac (Fig. 6C). Thus, Ash1L was essential for the regulation of H3K36me2 at the nrxn1α promoter, but not the total level of H3K36me2.  Fig. S2). (G) Proteins associated with the nrxn1α promoter were resolved by SDS-PAGE and silver staining (full-length gels in Fig. S2). (H,I) ChIP analysis showed the identified proteins specifically enriched at the nrxn1α promoter in the hippocampus of 3-month-old C57BL/6 mice were quantified according to the real-time PCR signal (full-length gels in Fig. S2). The relative enriched signal at the promoter region was normalized against the signal obtained at 3′ UTR. (n > 3 biological samples for each group; two-way ANOVA with Bonferroni post hoc test, * P < 0.05, * * P < 0.01, * * * P < 0.001).

Activity-dependent repression of nrxn1α is abolished in Ash1L−/− cortical neuronal cultures.
To confirm the role of Ash1L in the activity-dependent repression of nrxn1α , we further identified a few homozygous embryos from E14 by intercrossing of the Ash1L(− /+ ) mice (Fig. 7A,B). Ash1L− /− primary cortical neuron cultures derived from embryos (E14, Ash1L(− /+ ) mouse) showed low levels of Ash1L mRNA ( Fig. 7C; P < 0.001,  Student's t-test), which might be due to the mRNA surveillance mediated by nonsense-mediated decay. The 11 bp deletion in exon 2 might lead to instability of missense transcripts in Ash1L− /− culture neurons. When tested at the protein levels, Ash1L− /− primary cortical neuron cultures showed almost no expression of the intact Ash1L protein ( Fig. 7D; P < 0.05, Student's t-test). The cultured neurons showed no activity-dependent repression of nrxn1α 24 hours after high K + stimulation (two-way ANOVA, genotype × KCl stimulation interaction:  (Fig. 7G,H). These results indicated that Ash1L mediates the neuronal activity-dependent repression of nrxn1α .

Discussion
We have identified a unique process of activity-dependent repression in the gene transcription of nrxn1α . Furthermore, using a newly developed tool to isolate chromatin-associated proteins at specific genomic loci, we identified an epigenetic regulator required for the neuronal activity-dependent long-lasting repression of nrxn1α . We found that transient membrane depolarization recruits a histone modifier, Ash1L, to the nrxn1α promoter to enrich H3K36me2 in this region, leading to a long-term repression of nrxn1α transcription.
Since the finding of neuronal activity-dependent gene transcription in the late 1980s, many activity-dependent genes have been identified, which showed essential functions in regulating neuronal plasticity 37 . Most of the studies have focused on the activity-dependent increase of expression, and very few studies have examined activity-dependent repression. In fact, activity-dependent repression and elimination are important processes for normal brain function. A key step in the refinement of neuronal wiring during the late stage of development in the postnatal brain is mediated by the activity-dependent elimination of synapses [38][39][40] . Activity-dependent transcription factors, such as MEF2, may regulate synapse elimination and suppression 41 . Such a process coordinates the expression of a broad program of gene expression, including Bdnf 42 , Arc and Homer1 43 . Here, we extended this finding by showing that neurexin-1α , an essential protein for synapse formation, undergoes activity-dependent repression. Furthermore,  Fig. S5; n = 6 for WT mice, n = 9 for Ash1L (− /+ ) mice; Student's t-test, * P < 0.05). (E) Ash1L expression in brain regions of 3-monthold WT mice. The amount of Ash1L mRNA were quantified by q-PCR relative to GAPDH. Data are presented as the amount relative to the retrosplenial area (n = 5). (F,G) Ash1L (− /+ ) mice aged 3 months showed normal brain weights (F) (n = 12 for WT mice; n = 13 for Ash1L (− /+ ) mice) and neuron density in the hippocampus (G) (n = 3 for WT mice; n = 4 for Ash1L (− /+ ) mice; blue, DAPI; red, NeuN). Student's t-test, * P < 0.05, ns, no significance. Scale bar = 200 μ m.
Scientific RepoRts | 6:26597 | DOI: 10.1038/srep26597 by discovering the novel role of Ash1L in this process, we suggested that the epigenetic regulation is involved in the repression of synapse formation, implicating the function of epigenetic regulation in the long-term maintenance of memory circuits 18 .
Some methods have been established to identify the molecules bound to specific genomic loci in vivo [44][45][46] . Insertional chromatin immunoprecipitation (iChIP) was shown to be useful for this purpose. By inserting repeats of exogenous binding sites in the genome, this method utilizes an engineered DNA-binding protein to enrich the targeted genome loci to purify proteins associated with the inserted loci. However, the insertion of an exogenous DNA sequence might affect the transcriptional regulation of the endogenous genes. Here, we developed a new method. By using synthetic zinc finger proteins, we were able to enrich the targeted chromatin as the native state in vivo. Although the sensitivity of this method still needs to be improved, we have successfully identified a novel transcriptional regulator for neurexin-1α . Our screening also detected MeCP2 as a regulator for the neurexin-1α expression. The regulation of neurexin-1α expression by MeCP2 has been reported 47 , suggesting the specificity of this method.
Ash1 was originally identified as one of the epigenetic regulators in Drosophila. The ash1 gene encodes a member of the trithorax group (TrxG) of proteins that maintain active transcription by competing with Polycomb proteins 48,49 . The mammalian ortholog, Ash1L, acts as histone methyltransferase targeting H3K36me2 33,50,51 . However, H3K36me2 also recruits histone deacetylase to repress spurious transcripts within the gene body 16,17 . Similar repression mechanisms might also take place in the H3K36me2-dependent repression of the alternative promoter in nrxn1α . Therefore, Ash1L could both enhance and repress gene expression, depending on the genomic environments on the regulated gene targets 16,52 .
Ash1L is widely expressed in multiple organs and enriched in the brain 51,53,54 . Its expression is enriched in the hippocampus so that the protein level of Ash1L in the hippocampus is sensitive to the genomic deletion in one allele. However, the roles of Ash1L in the brain remain poorly understood. Here, we identified a novel role of Ash1L in activity-induced repression of neurexin-1α expression. Further studies on the regulation of Ash1L and its role in the adult brain might help to improve our understanding of the roles that epigenetic modifications play in regulating brain function, especially in activity-dependent network rewiring. Such studies might also uncover the neural basis of cognitive diseases, such as autism spectrum disorder.

Materials and Methods
Plasmid Construction. The creation of a six-finger Zinc Finger protein targeting nrxn1α promoter (GST-ZFP-pnrxn1α ) was carried out according to a previously described procedure 55 . DNA sequences encoding the first zinc finger were digested with XhoI and SpeI, and subsequently cloned into a XhoI/SpeI-digested pSCV vector to create one-finger ZF insertion pSCV-1ZF. The last five zinc finger-encoding sequences were digested with XmaI and SpeI and then cloned into AgeI/SpeI-digested pSCV-1ZF successively to create six-finger ZF insertion pSCV-6ZF. Finally, the 6ZF encoding sequence was digested with XmaI/SpeI and cloned into XmaI/SpeIdigested pGEX-4T-1, which has a modified multiple cloning site, creating pGEX-6ZF. Insertion was confirmed by sequencing.
The 900 bp nrxn1α promoter was cloned into the KpnI and XhoI site of pGL3-enhancer luciferase reporter vector to create pnrxn1-Luc plasmid. The primers used to amplify nrxn1α promoter from mouse genomic DNA are listed in Table 3.

Glutathione S-transferase (GST)-tagged zinc finger protein expression and purification.
For synthetic ZFP-pnrxn1α purification, E. coli BL21 (DE3) transformed with pGEX-6ZF were grown in Luria-Bertani (LB) medium supplemented with ampicillin (100 μ g/ml) at 37 °C, and then IPTG (final concentration, 1 mM) was added when the A 600 of the culture reached 0.6. After incubation for 12 hours at 16 °C, the bacteria were harvested, re-suspended in lysis buffer (20 mM Tris-HCl, 150 mM NaCl, pH 8.0), and lysed via sonication. Lysates were cleared by centrifugation at 13000 rpm for 20 min. The GST-ZFP-pnrxn1α was purified using a Pierce GST spin purification kit (Thermo Scientific) according to the manufacturer's instructions. Ultrafiltration centrifugation (Millipore) was conducted to remove GSH and concentrate protein.  Electrophoretic mobility shift assay. The nrxn1α promoter probe was amplified by standard PCR from the C57BL/6 mouse genome. Purified proteins were incubated with probes at 37 °C for 30 min in buffer (50 mM Tris-HCl, 150 mM NaCl, 5% glycerol, 1 mM dithiothreitol, 0.1% NP-40, pH 8.0), followed by electrophoresis in 8% native polyacrylamide gels in 0.5 × TBE buffer, at 80 V for 2 h. The gels were stained with ethidium bromide and visualized under ultraviolet transillumination. For EMSA competition assay, DNA fragments with paired-end tags were prepared by PCR amplification. A 15-fold excess of competitor was added to the incubation system. The gels containing the shift DNA-protein complexes were purified followed by PCR amplification against the paired-end tags.
Chromatin pull-down assay. C57BL/6 mice at the age of 3 months were deeply anesthetized with an intraperitoneal injection of 2% sodium pentobarbital (400 mg/kg body weight) and subsequently perfused with 4% paraformaldehyde solution in 0.1 M phosphate buffer (pH 7.4). The brains were removed rapidly, and 1-mm-thick coronal sections were prepared using brain matrix. Fresh tissue from the cortex and hippocampus was cut into small pieces, homogenized, cross-linked with 4% formaldehyde for 15 min, and quenched with 0.125 M glycine. Cells were fragmented by sonication (Micro-tip, Branson sonicator). The chromatin segments were collected by centrifugation at 16000 g for 15 min at 4 °C and incubated with purified GST-ZFP-pnrxn1α for 2 hours. Glutathione agarose resin (Thermo Scientific) was added for 40 min, and the chromatin segments were then washed with washing buffer until the absorbance at 280 nm was stabilized at the baseline level. Bound complexes were eluted in 50 mM Tris-HCl containing 15 mM glutathione. Protease inhibitor phenylmethyl sulfonyl fluoride (1 μ M, PMSF) was added to all buffers. Elution with 200 μ l was reversed at 65 °C with 8 μ l of NaCl (5 M), 2 μ l of RNaseA (20 mg/ml) and 1 μ l of proteinase K (10 mg/ml) for at least for 6 hours. DNA was purified using the phenol/chloroform extraction and ethanol precipitation method. The relative enrichment ratios within 3.8 kb regions upstream of TTS were calculated by using GST-ZFP-pnrxn1α pull-down signals normalized to GST control signals. The primers are listed in Table 3.

Western blot analysis and silver staining of SDS-PAGE.
For western blot analysis, the samples were loaded on SDS-PAGE and then transferred to nitrocellulose membranes (Bio-Rad). After blocking with blocking buffer (5% non-fat dry milk and 0.1% Tween-20 in PBS) for 1 hour at room temperature, membranes were incubated with probed antibodies overnight at 4 °C. Membranes were washed three times, incubated with HRP conjugated secondary antibodies for 1 hour at room temperature, and the signals were detected with chemiluminescence solution (Thermo Scientific). The following antibodies were used: antibody specific to Ash1L (Novus Biologicals), H3K9me3 (Abcam), H3K4me3 (Abcam), H3K36me2 (Abcam), Histone3 (Santa Cruz Biotechnology), and GAPDH (Cell signaling). Silver staining was performed using a modified protocol of Mortz 56 . Briefly, gels were fixed with 10% acetic acid/40% methanol for 20 min and then washed with deionized distilled water. After incubation in 0.2% Na 2 S 2 O 3 for 30 min, gels were washed 3 times again and incubated in 0.25% AgNO 3 for 20 min. Gels were developed with 0.04% paraformaldehyde and 2.5% Na 2 CO 3 until the desired staining had occurred. Development was stopped by the addition of 5% acetic acid.

Identification of Nrxn1 promoter associated proteins by mass spectrometry. GST-ZFP-pnrxn1α
pull-down samples were separated on SDS-PAGE gel. Each gel lane, except zinc finger protein, was entirely sliced into 4 bands, reduced with 10 mM dithiothreitol, and alkylated with 55 mM iodoacetamide. In-gel digestion was then carried out with sequence-grade modified trypsin (Promega) in 50 mM ammonium bicarbonate at 37 °C overnight. The peptides were extracted twice with 0.1% trifluoroacetic acid in 50% acetonitrile aqueous solution for 30 min. High-performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS) analysis was performed as described previously 57   incubator equilibrated with 5% CO 2 in air at 37 °C. Passages 4 through 20 were used for experience. NG108-15 cells were differentiated by allowing the cells to reach approximately 60% confluence and then replacing the serum-free neuronal medium (neurobasal medium (Life Technologies) containing 2% B27 supplement (Gibco), 2 mM glutamax (Gibco) and penicillin/streptomycin (Gibco)) for at least 36 hours. Cells were transfected with cationic lipids VigoFect (Vigorous Biotechnology). Primary cortical neuron cultures were prepared from embryonic day 16 (E16) for ICR mice and from E14 for Ash1L (− /+ ) intercrosses. Cells were seeded onto culture dishes at a density of 5 × 10 5 per square centimeter and grown in serum-free neuronal medium. After DIV 1, cells were transfected using Lipofectamine 2000 reagent (Invitrogen) according to the manufacturer's protocol.
For KCl depolarization, primary cortical neuron cultures were treated with culture medium containing 51 mM KCl for 10 min on DIV9, followed by washing and medium exchange with fresh neuronal medium (without KCl).
RNA extraction, reverse transcription, and quantitative real-time PCR. Total RNA was isolated using TRIzol reagent (Invitrogen) according to the manufacturer's protocol, and 1.5 μ g RNA was used for reverse transcription by the one step first strand cDNA synthesis kit (Transgen). Quantitative real-time PCR (qRT-PCR) was performed using SYBR Green PCR master mix (Bio-Rad) in a CFX96 machine (Bio-Rad). The relative fold-change in each mRNA expression was calculated using the ddCt method relative to the expression of GAPDH. The primers used for qRT-PCR are listed in Table 3.
Cell transfection and Luciferase assays. Primary cortical neuron cultures from E16 for ICR mice were seed in a 24-well plate. DNA plasmids were delivered into cells 24 hours after plating. LV: CamKIIα -ChR2 infection was carried out on DIV2. Luciferase activities were measured 36 hours after optogenetic stimulation using Dual-Luciferase Reporter Assay System (Promega).
Microinjection for generation of Ash1L-deficient mice. CRISPR/Cas9 system was employed to generated Ash1L-deficent mouse. Target sequence within the second exon of ash1L was chosen according to the sgRNA recognition guidelines described previously 59,60 . Three sgRNAs were designed. According to the initial tests, we choose one of the sgRNAs for the injection. In vitro transcription of customized sgRNAs was performed using a RiboMAX Large Scale RNA Production Systems-T7 Kit (Promega). Cas9-mRNA was synthesized in vitro using a mMESSAGE mMACHINE T7 Ultra Kit (Life Technologies). The sgRNA and Cas9-coding mRNA were mixed to final concentrations of 50 ng/μ l and 250 ng/μ l, respectively. Injection of C57BL/6 zygotes pronuclei was performed with an established setup at the Laboratory Animal Facility at the Tsinghua University. 1-week-old founder mice were identified by PCR, using template of DNA isolated form tail biopsies. The primers of Ash1L-Wt-F/R, which was used for genotype, are listed in Table 3.
Immunohistochemistry. Immunohistochemical analysis was performed as described previously 11 .
Anti-NeuN antibody (Abcam) was used at a dilution of 1:500. Confocal images (1 μ m) were scanned and subjected to three-dimensional reconstruction. Brain sections with the strongest intensity were scanned first. All other images included in the analysis were scanned with the same settings (Zeiss LSM780). ImageJ software (National Institutes of Health, Bethesda, MD, USA) was used to calculate the mean neuronal density.
Ethics statement. All animals were kept in animal research facility of Tsinghua University. The Laboratory Animal Facility at the Tsinghua University is accredited by AAALAC (Association for Assessment and Accreditation of Laboratory Animal Care International). All experimental protocols involving mice were approved and conducted in accordance with the guidelines by the Institutional Animal Care and Use Committee (IACUC) of Tsinghua University. Data analysis. The statistical significance between two groups was determined by Student's t-test. One-way ANOVA followed by Dunnett post hoc test or Tukey post hoc test for three or more groups. Two-way ANOVA accompanied by a Bonferroni post hoc test was used for multiple comparisons using Prism 6 (Graphpad Software Inc., La Jolla, CA, USA). All data are presented as mean ± SEM; P values of less than 0.05 were considered statistically significant (* P < 0.05, * * P < 0.01, * * * P < 0.001).