Project 2. Methyl-Binding Proteins in Rodent Depression Models (Schahram Akbarian, Mount Sinai; with Hanno Hock, MGH)
Project 2 builds upon stress regulation of H3K9me2 by examining a family of proteins—the malignant brain tumor domain proteins (MBTs)—that serve, among other actions, as readers of this repressive histone methylation mark. Each Lys residue that undergoes methylation can exist in a mono-, di-, or tri-methylated state (me1, me2, and me3, respectively). H3K9me3, like H3K9me2, is a potent repressor of gene expression, however, it is associated primarily with highly repressed genomic regions. Indeed, we found recently by ChIP-Seq that H3K9me3 binding in brain occurs predominantly at non-genic regions. H3K9me2 binding, in contrast, is high across the genome and associated with large numbers of dynamically regulated genes. Moreover, we have seen selective regulation of H3K9me2, and not H3K9me1 or H3K9me3, in the NAc and PFC after several forms of chronic stress. MBTs selectively bind to the me1 and me2 state of methylated histones, including H3K9me2, and indeed are the most intensely studied methyl-reader proteins known; MBT binding then promotes nucleosome aggregation and gene repression. However, virtually nothing is known about these mechanisms in brain, let alone in psychiatric disease models. In recent months, we have obtained considerable evidence to support our hypothesis that MBTs, as would be expected from our data on H3K9me2, are critically important in depression-related phenomena. We have shown that 3 of 9 known MBT proteins, L3MBTL1, L3MBTL2 and SFMBT1, are expressed at high levels in adult mouse and human brain, including PFC and NAc, and that expression levels of all three MBTs are robustly regulated in PFC in several chronic stress models, including chronic social defeat stress (CSDS) and maternal separation stress. We have evidence that these proteins form transcription regulatory complexes with specific histone-modifying enzymes, including G9a. Importantly, we have generated mutant mouse lines for each of these MBTs; mice lacking L3MBTL1 constitutively display pro-depression-like behaviors in initial assays. Depressed humans display reduced levels of L3MBTL1 expression in PFC and NAc as well. These observations provide a powerful rationale to embark on the proposed studies of MBTs and related proteins in mouse models of depression and antidepressant action. Several complementary approaches are being used. First, to gain initial insight into the role of these proteins in depression models, we are characterizing conditional knockouts of L3MBTL1, L3MBTL2, or SFMBT1 in a behavioral screen. Second, these broad deletions are being compared with more highly targeted deletions, as well as overexpression models, focusing on PFC and NAc. Third, we are extending the standard ChIP-Seq studies of several di-methylated histone residues carried out in the Chromatin and Gene Analysis Core with the genome-wide analysis of these same sites, and of MBT binding, selectively in neurons, focusing initially on PFC due to its larger size. This utilizes cutting-edge methodologies, shown in the Figure below, developed in the Akbarian laboratory. Such analyses include examination of H3K9me2 as well as H4K20me2, another major form of repressive histone methylation that is targeted by MBTs. By complementing these analyses with parallel studies of human postmortem brain samples in Project 4, the goal of our coordinated behavioral and chromatin analyses is to construct a genome-wide map of “epigenetic risk loci” highly relevant for depression. The studies outlined in Project 2 thus provide a template by which many other reader proteins can be characterized in depression models.
FIGURE. Isolating neuronal nuclei from adult brain for ChIP. A. Summary of method. B. Nuclei extracted from adult forebrain of (a-d) CaMKII-H2B-GFP transgenic mice and (e-j) wildtype mice immunolabeled with NeuN, as indicated pre- and post-fluorescent sorting and after pelleting. Green channel for (a,c) GFP or (e,g,i) NeuN; blue channel for DAPI. Note that, post-sorting, samples are comprised entirely of neuronal nuclei. Bar = 20 μm. C. Representative sorting scatter plots from (a) negative control (NC) processed without NeuN antibody, and (b) sample processed with NeuN antibody (NeuN). D. Images from 8 week old CaMKII-H2B-GFP mice. (a) Cerebral cortex. Note GFP+ nuclei across layers II-VI, but not in layer I or white matter (WM). (b) Caudate-putamen and (c) hippocampus. (d-f) Nuclei extracted from forebrain and processed for NeuN (red), and nucleophilic dye, DAPI (blue). Note that GFP expression is limited to 5 of the 6 NeuN+ nuclei. Arrows label GFP-, NeuN- nuclei, * labels GFP-, NeuN+ nucleus. (g) Section from cerebral cortex labeled with anti-GAD67 antibody (red). Note absence of GFP signal in GABAergic neuron. Bar = (a,c) 100, (b) 300, (d-g) 5 μm. From Jiang et al., BMC Neuroscience, 9:42 (2011).