Role of the histone variant H2A.Z.1 in memory, transcription, and alternative splicing is mediated by lysine modification

Abstract

Creating long-lasting memories requires learning-induced changes in gene expression, which are impacted by epigenetic modifications of DNA and associated histone proteins. Post-translational modifications (PTMs) of histones are key regulators of transcription, with different PTMs producing unique effects on gene activity and behavior. Although recent studies implicate histone variants as novel regulators of memory, effects of PTMs on the function of histone variants are rarely considered. We previously showed that the histone variant H2A.Z suppresses memory, but it is unclear if this role is impacted by H2A.Z acetylation, a PTM that is typically associated with positive effects on transcription and memory. To answer this question, we used a mutation approach to manipulate acetylation on H2A.Z without impacting acetylation of other histone types. Specifically, we used adeno-associated virus (AAV) constructs to overexpress mutated H2A.Z.1 isoforms that either mimic acetylation (acetyl-mimic) by replacing lysines 4, 7 and 11 with glutamine (KQ), or H2A.Z.1 with impaired acetylation (acetyl-defective) by replacing the same lysines with alanine (KA). Expressing the H2A.Z.1 acetyl-mimic (H2A.Z.1^KQ) improved memory under weak learning conditions, whereas expressing the acetyl-defective H2A.Z.1^KA generally impaired memory, indicating that the effect of H2A.Z.1 on memory depends on its acetylation status. RNA sequencing showed that H2A.Z.1^KQ and H2A.Z.1^KA uniquely impact the expression of different classes of genes in both females and males. Specifically, H2A.Z.1^KA preferentially impacts genes involved in synaptic function, suggesting that acetyl-defective H2A.Z.1 impairs memory by altering synaptic regulation. Finally, we describe, for the first time, that H2A.Z is also involved in alternative splicing of neuronal genes, whereby H2A.Z depletion, as well as expression of H2A.Z.1 lysine mutants influence transcription and splicing of different gene targets, suggesting that H2A.Z.1 can impact behavior through effects on both splicing and gene expression. This is the first study to demonstrate that direct manipulation of H2A.Z post-translational modifications regulates memory, whereby acetylation adds another regulatory layer by which histone variants can fine tune higher brain functions through effects on gene expression and splicing.

Fig. 1. Acetylated H2A.Z occurs over accessible regions in hippocampal neurons.

A Left: Measurements of peak distance to nearest TSS across each H2A.Z enrichment quartile (Peaks for H2A.Z were separated into quartiles according to acetylated H2A.Z levels). All AcH2A.Z peaks are relatively close to TSS, as compared with a random sampling of the genome, and regions within Q3 or Q4 tend to be further from genes than Q1 or Q2. Right: Measurements of AcH2A.Z peaks intersection with promoters or intergenic CpG Islands, a proxy for putative enhancers. Fisher odds ratios are displayed as a bar chart with p-value indicated. P values generated from two tailed hypergeometry testing. ***indicates p < 0.0001. B Enrichment for chromatin modifications across quartiles defined in panel A. Regions within Q3 or Q4 tend to have higher levels of H3K4me3, H3K27ac, and chromatin accessibility, as compared with Q1 and Q2, but lower levels of H3K4me1. C Gene ontology analysis indicates a significant enrichment for gene pathways associated with neuron and brain function across all quartiles, with the most significant enrichment occurring at the most highly enriched AcH2A.Z loci, those within Q3 and Q4. D Genome browser snapshot depicts example genes where AcH2A.Z levels are enriched. AcH2A.Z ChIP seq N = 3.

Fig. 2. H2A.Z.1 acetyl-mimic promotes memory and acetyl-defective H2A.Z.1 impairs memory.

A Male and female mice were trained on the object location task for either 10 min (standard training protocol) or 5 min (subthreshold training protocol) and memory was tested 24 h after each training session. On the 5 min task, mice expressing the acetyl-mimic H2A.Z.1^KQ had improved memory compared to all other groups (left). On the 10 min task, mice expressing acetyl-defective H2A.Z.1^KA had impaired memory compared to all other groups. Sex was not a significant factor, so data are combined for males and females. N/group for females: Myc:13, H2A.Z^KO:12, H2A.Z.1^WT:12, H2A.Z.1^KA:8; H2A.Z.1^KQ:8. N/group for males 5 min training: Myc:12; H2A.Z^KO:8, H2A.Z.1^WT:10, H2A.Z.1^KA: 10; H2A.Z.1^KQ:11. N/group for males 10 min training: Myc:19; H2A.Z^KO:13, H2A.Z.1^WT:16, H2A.Z.1^KA:17; H2A.Z.1^KQ:16. B Mice were trained on contextual fear conditioning using a single 0.3 mA foot shock. Female mice expressing acetyl-defective H2A.Z.1^KA had impaired memory compared to all other groups. Male mice had lower freezing overall, but freezing did not differ between treatment groups. N/group for females: Myc:13, H2A.Z^KO:12, H2A.Z.1^WT:11, H2A.Z.1^KA:8; H2A.Z.1^KQ:8. N/group for males: Myc:7, H2A.Z^KO:5, H2A.Z.1^WT:6, H2A.Z.1^KA:6; H2A.Z.1^KQ:6. C A subgroup of male mice were trained with a stronger protocol (two 0.3 mA shocks instead of one 0.3 mA shock). Under these conditions, H2A.Z deletion improved memory, whereas expression of acetyl-defective H2A.Z.1^KA impaired memory. * signifies p < 0.05 for comparisons identified by lines between bars; **signifies p < 0.05 for H2A.Z.1^KA vs. all other groups N/group = Myc:11; H2A.Z^KO:10, H2A.Z.1^WT:11, H2A.Z.1^KA:10; H2A.Z.1^KQ:12. Data are expressed as Mean ± SEM. *q < 0.05.

Fig. 3. H2A.Z deletion and expression of lysine mutants impact gene expression.

Volcano plots showing effects of H2A.Z deletion alone (H2A.Z^KO) and H2A.Z deletion combined with H2A.Z.1^WT, H2A.Z.1^KA, or H2A.Z.1^KQ replacement on transcript levels in (A) female and (B) male mice. Red and green circles indicate differentially expressed genes (DEGs) with FDR < 0.05 and log₂FC > 1 that are significantly increased or decreased relative to the Myc control group. Gray circles represent significant DEGs with FDR < 0.05 and log₂FC < 1. Black circles indicate genes that did not meet significance criteria. Number of DEGs shown in brackets reflect FDR < 0.05 and log₂FC > 1. Example gene tracks are shown for (C) female and (D) male mice. Data in gene tracks are expressed as Counts Per Million (CPM). N/group is as follows. Males: Myc 4, H2A.Z^KO 4, H2A.Z.1^WT5, H2A.Z.1^KA 4, H2A.Z.1^KQ 5. Females: Myc 3, H2A.Z.1^KO4, H2A.Z.1^WT 4, H2A.Z.1^KA 4, H2A.Z.1^KQ 3.

Fig. 4. H2A.Z.1 replacement rescues gene expression changes caused by H2A.Z deletion.

A Whisker box plots comparing DEGs from Myc controls vs. H2A.Z^KO sorted into quintiles by their Log₂FC value in increasing order. ***Denotes significant difference between H2A.Z^KO and all other groups. *Any additional differences between individual groups are denoted with lines drawn between significantly distinct groups. P < 0.05. B Heat map comparing gene expression in the three H2A.Z.1 replacement groups: H2A.Z.1^KA, H2A.Z.1^WT and H2A.Z.1^KQ. Heat maps represent rlog(DESeq2) normalized raw counts that were then z-scaled. C Left: Gene track showing an example of H2A.Z.1^KQ promoting gene expression compared to H2A.Z.1^KA in both sexes. Right: Gene track showing an example of H2A.Z.1^KQ resulting in lower gene expression than H2A.Z.1^KA only in male mice. Data in gene tracks are expressed as Counts Per Million (CPM).

Fig. 5. H2A.Z deletion and expression of lysine mutants impact alternative splicing.

Volcano plots showing differential isoform usage for mice with H2A.Z deletion and H2A.Z.1 mutants relative to Myc controls in A female and B male mice. dIF = Differential Isoform Fraction. (FDR < 0.05).