Conserved higher-order chromatin regulates NMDA receptor gene expression and cognition

Abstract

Three-dimensional chromosomal conformations regulate transcription by moving enhancers and regulatory elements into spatial proximity with target genes. Here we describe activity-regulated long-range loopings bypassing up to 0.5 Mb of linear genome to modulate NMDA glutamate receptor GRIN2B expression in human and mouse prefrontal cortex. Distal intronic and 3' intergenic loop formations competed with repressor elements to access promoter-proximal sequences, and facilitated expression via a "cargo" of AP-1 and NRF-1 transcription factors and TALE-based transcriptional activators. Neuronal deletion or overexpression of Kmt2a/Mll1 H3K4- and Kmt1e/Setdb1 H3K9-methyltransferase was associated with higher-order chromatin changes at distal regulatory Grin2b sequences and impairments in working memory. Genetic polymorphisms and isogenic deletions of loop-bound sequences conferred liability for cognitive performance and decreased GRIN2B expression. Dynamic regulation of chromosomal conformations emerges as a novel layer for transcriptional mechanisms impacting neuronal signaling and cognition.

Figure 1. Brain-specific Higher Order Chromatin at GRIN2B (chr. 12p31.1)

(A) Linear map for 800kb surrounding GRIN2B TSS, including GRIN2B gene body with exons and CpG islands as indicated. Browser tracks for (top to bottom) H3K27ac in PFC and cingulate cortex (CG)(Zhu et al., 2013), H3K4me3 from PFC neurons(Cheung et al., 2010) and PFC-RNAseq(Bharadwaj et al., 2013)(B) (left) X–Y graphs (mean ± S.D.) present chromosome conformation capture (3C) profiles anchored on restriction fragment with TSS (green). (top) adult PFC (N=4) and (bottom) skin fibroblasts (N=3). Notice much stronger interaction of red-marked fragments, peak 1 (distal arm of GRIN2B^TSS+348kb loop) and peak 2 (distal arm of GRIN2B^TSS+449kb) with the TSS specifically in brain but not fibroblast (Two-way ANOVA cell type × 3C primer F(42,210)=157.7, P<0.001, Newman-Keuls posthoc P < 0.001) (right) Representative 3C PCR gels for four PFC specimens PFC 1–4 and cultured fibroblasts (FIB), showing 3C long-range interactions of TSS with +348 kb (peaks 1) and +449kb (peak 2) sequences specifically in PFC samples, and, as control, sequences within first intron. (right) Bar graph (mean ± S.D; N=3–4/group) comparing GRIN2B RNA in PFC and FIB. (C) 3C PCR gels from iPS and iPS-derived differentiated neuronal culture (NEU-iPS), nl = 3C assays without ligase. Bar graphs (mean ± S.D; N=3–4/group) quantify TSS-bound loopings with distal (+348kb and +449kb sequences) and proximal intron, as indicated. Significant changes after differentiation, GRIN2B RNA: F_(1,8) = 714.85, P < 0,001 and GRIN2B^TSS+449kb, F_(1,8) = 507.35, P < 0,001; Newman-Keuls post hoc p < 0.001. See also Figure S1.

Figure 2. Coordinated regulation of multiple chromosomal loopings targeting GRIN2B TSS

(A) Browser tracks from ENCODE data collection (Bernstein et al., 2012), showing highest enrichments for (top) CTCF and (bottom) NRF-1 proteins in sequences +449kb downstream from TSS (arrows). Bottom gels, PCR from 3C-ChIP-loop with anti-CTCF antibody (HEK293 cells), showing CTCF enrichment specific for GRIN2B^TSS+449kb loop (B) (top) 4kb of GRIN2B^TSS+449kb (chr12:13,680,308-13,684,308) showing (bars) elevated density of AP-1 motifs and increased AP-1 and NRF-1 occupancy in ENCODE ChIP-seq tracks (Bernstein et al., 2012). (bottom) Reporter assay for three 100–125 bp sequences (C (control) and 2a, 2b from red box no. 2 (=GRIN2B^TSS+449kb) shown in panel A, showing up to 6-fold increase in minimal promoter- TATA box luciferase activity. (C) (top) Immunoblots (left) anti-SETDB1, (right) anti-FLAG antibody recognizing inducible FLAG-SETDB1 protein and (bottom) quantitative RT-PCR assays from stable HEK293 clone for inducible SETDB1 expression. Note robust induction of SETDB1, and parallel decline in GRIN2B RNA, 72 hours after addition of doxycycline (+ DOX), compared to untreated culture (−DOX). (D) ChIP-PCR, with (left) repressive SETDB1, HP1and H3K9me3 and (right) facilitative H3K27ac, AP-1 and NRF-1 in chromatin across GRIN2B locus, as indicated (gray box, −250kb (upstream from TSS) control sequence; green box, TSS; red box no. 1, GRIN2B^TSS+348kb and no. 2, GRIN2B^TSS+449kb; yellow box, intronic repressor and SETDB1 target site). Striped (white) bars, + (−) DOX. (E) 3C PCR expressed as fold change (+DOX/−DOX). Notice decrease in long-range loopings, GRIN2B^TSS+348kb and GRIN2B^TSS+449kb, in conjunction with (D) localized, SETDB1-mediated H3K9 hypertrimethylation and HP1 binding at proximal intronic sequences < 40kb from TSS. All bar graphs N=3-group, mean ± (B) S.E.M, (C, D, E) S.D. *, **, *** P < 0.05, 0.01, 0.001, Two way ANOVA (ChIP) and unpaired t-tests after Bonferroni correction (3C, qRT-PCR, luciferase assays). (F) Dynamic model; long-range GRIN2B^TSS+348kb and GRIN2B^TSS+449kb loops are absent in cells not expressing GRIN2B. Upon GRIN2B expression, AP-1 and NRF-1 enriched enhancer elements, positioned in the distal arm of GRIN2B^TSS+449kb, undergo relocation and are moved into close spatial proximity with the TSS. This is counterbalanced by shorter-range TSS-bound intronic loopings enriched with repressive chromatin. See also Figure S1.

Figure 3. Higher order Grin2b chromatin in mouse cerebral cortex and hippocampal neurons

(A) 500kb of linear genome (mm9, chromosome 6:135,623,529-136,123,529), encompassing Grin2b, including TSS (green), proximal intronic sequences targeted by SETDB1 (yellow), and two loop loopings (red) (1) Grin2b^TSS+378kb and (2) Grin2b^TSS+471kb, with sequences homologue to human GRIN2B^TSS+348kb and GRIN2B^TSS+449kb (Fig. S1) (B) (Top) Browser tracks for histone marks H3K4me3 and H3K27ac in adult mouse cerebral cortex(Dixon et al., 2012). (C) Fold-change (CK-Setdb1 transgenic (Tg) /wildtype littermate (Wt) of 3C PCR from adult Tg and Wt cortex (see also Fig. S2). Notice increased physical interactions of (yellow) TSS-bound +15 to +40kb intronic sequence, and significant decrease in (red box no. 1) Grin2b^TSS+378kb and (red box no. 2) Grin2b^TSS+471kb. Two-way ANOVA, 3C interaction × genotype F(7,32)=54.905(p<0.001). Newman-Keuls post-hoc *, **; P< 0.05, 0.01. (D) Quantification of Grin2b (left) RNA and (right) protein in adult (6–8 week) CK-Setdb1 Tg and wildtype (Wt) littermate control cortex. mean ± S.D, N=5 (RNA) and N=3 (immunoblot)/group. (E) Activity-dependent regulation of Grin2b higher order chromatin in hippocampal neurons. (top, left to right) Grin2b RNA, 3C quantification of Grin2b^TSS+378kb and Grin2b^TSS+471kb and SETDB1 occupancy across four regulatory sequences at Grin2b locus. Note significant increase at SETDB1 target site after 15 hours of picrotoxin (PTX) or vehicle control (DMSO). N=3–6 experiments/group, data shown as mean ± S.D..(F) (left) Grin2b-TALE-VP64-GFP specifically targets mouse Grin2b loop no. 2 (Grin2b^TSS+471kb), 471 kb downstream of TSS. Images show Neuro2A cells and cultured hippocampal neurons expressing GFP-tagged TALE-VP64. RT-PCR for Grin2b and Gapdh control showing specific expression in cultured neurons. (right) anti-TALE-VP64 ChIP in N2A and NG108 cells and primary cortical neurons, expressed (y-axis) as fold change compared to non-transfected condition (N=3/group; (mean ± S.E.M., * Two-way ANOVA cell type × TALE-VP64 binding F(2,12)=4.04, P<0.05, Bonferroni posthoc P < 0.05).RT-PCR Grin2b, Grin2a and Setdb1 RNA levels in neurons transfected with TALE-VP64-EGFP (TALE), compared to mock-transfected neurons (Con). Notice specific TALE-VP64 mediated increase in Grin2b RNA (N = 6 per group, mean ± S.E.M, *P < 0.05, t-test). See also Figures S1 and S2.

Figure 4. Intergenic sequences affect cognition and GRIN2B expression

(A) Regional association plot for 100kb surrounding GRIN2B 3’end, from PGC2 dataset(Consortium, 2014). SNP rs117578877 is within (red box no. 2) distal arm of GRIN2B^TSS+449kb. (B) GRIN2B RNA in PFC from N = 12 SCZ and N = 12 control subjects biallelic for major (C) allele and N=5 SCZ and N=5 control minor (T)-allele carriers. *, P < 0.05 (two-tailed t-test) (C) (Left) Spatial working memory strategy and (Right) N-back continuous performance task scores in N=31 C/T heterozygotes compared to N = 794 C/C subjects. Sole T/T subject marked by arrow (mean ± S.E.M., *, **, **** = P < 0.05, 0.005, 00005 (Mann-Whitney). (D) (top) 248 bp from distal arm of GRIN2B^TSS+449kb with a total of 15 (N)GG PAM leader sequences for CRISPR/CAS editing. (middle from left to right) Transfected HEK293(FT) cells, showing high transfection efficiency (>95%) by GFP reporter and ~180kDa Cas9 protein immunoblot. Size bar = 100 micron. Representative 5% acrylamide gels showing predicted (Table S3) banding patterns of SURVEYOR nuclease cleavage products in Cas9 nuclease-treated cultures exposed to 13/15 sgRNAs (one sgRNA/transfection) and Cas9 nickase treated cultures exposed to combined sgRNAs 2+7 and sgRNA 2/nuclease-treated cells. (Bottom) DNA sequence and PAM positions (sgRNA 2 and 7, black triangles mark bp position predicted to be the primary target in nickase assays), and electropherograms from (top to bottom) DNA clones of wildtype, sgRNA2/nuclease-exposed and sgRNA2+7/nickase-exposed cells, confirming small 2bp (GT) deletion 5’ to PAM in sgRNA2/nuclease-exposed and much larger deletions and mutations in sgRNA2+7/nickaseexposed cells. Red triangles mark site of mutations. Bar graph (N=3/group, mean ± S.E.M.) showing lower GRIN2B RNA expression after sgRNA2+7/nickase, compared to mock transfected (CRISPR/Cas9 transfected without sgRNA) as control. See also Tables S1–S4 and Figure S3.

Figure 5

Spatial working memory showing significant increase in repetitive errors, measured as repeat entries in 8-arm radial maze on 3^rd (final) day of testing in (left to right) C57BL6/J mice treated with GRIN2B antagonist RO 25–6981 (10mg/kg) (N=14–18/treatment group), drug-naive CK-Setdb1 transgenic (Tg) mice compared to wildtype (Wt) littermate control (N=12–17/genotype) and CK-Cre, Mll1^2lox/2lox (M) mice (N=12–16/genotype). Data shown as mean ± S.E.M. *; P < 0.05. See also Figure S4.