(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



KAT6A is essential for developmental control gene expression in neural stem and progenitor cells [1]

['Anne K. Voss', 'The Walter', 'Eliza Hall Institute Of Medical Research', 'Melbourne', 'Victoria', 'Department Of Medical Biology', 'University Of Melbourne', 'Samantha Eccles', 'Johannes Wichmann', 'Waruni Abeysekera']

Date: 2026-05 

Heterozygous variants in the KAT6A gene encoding the histone lysine acetyltransferase KAT6A (MOZ, MYST3) cause Arboleda-Tham syndrome, a cognitive impairment syndrome. Histone acetylation is generally associated with active gene transcription. Genetic deletion of both alleles of the Kat6a gene in mice causes developmental defects including anterior homeotic transformation, cleft palate, interrupted aortic arch and cardiac septal defects. Loss of KAT6A impairs expression of HOX, DLX and TBX genes, which are essential for body segment identity specification, palate, heart and aortic arch development. However, the effects of loss of KAT6A on chromatin modifications and gene expression in neural cells, which are relevant to normal brain development and function, is still poorly understood. In this study, we used an automated high-throughput chromatin profiling method and RNA sequencing in mouse neural system and progenitor cells to assess the effects of loss of one or two alleles of Kat6a on gene expression, histone acetylation and methylation. We also assessed occupancy by a trithorax group protein and RNA polymerase II. Our data suggests two modes of action for KAT6A: (1) acetylation of histone H3 on lysine 23 at promoters and enhancers and (2) recruitment of the trithorax group protein MLL1 (KMT2A) to promote the expression of developmental genes, including SOX and homeodomain genes. Together, these two functions appear to be required for normal gene expression in neural progenitors and essential for proliferation and neuronal differentiation.

During embryonic development, specific families of transcription factors pattern the early embryo to lay down and define the body and organ structure. However, the mechanisms governing the onset of the expression of these developmental transcription factors in less well understood. KAT6A is thought to promote gene expression by acetylation histone proteins. Here we determine the effects of KAT6A on histone acetylation and gene expression in mouse neural stem and progenitor cells. Our data are relevant for the understanding of pathogenic genetic variants in one allele of the human KAT6A gene, which cause the Arboleda-Tham cognitive impairment syndrome.

Competing interests: I have read the journal’s policy and the authors of this manuscript have the following competing interests: AKV and TT are inventors on patent WO2016198507A1. AKV and TT have received research funding from the Cancer Therapeutics CRC (CTX). AKV and TT have served on a clinical advisory board for Pfizer. The other authors have declared that no competing interests exist.

Funding: This work was supported by the Lorenzo and Pamela Galli Medical Research Trust to AKV; the Valda Klaric Foundation to TT; the Australian National Health and Medical Research Council through project grant 1160517 to TT, Ideas Grant 2010711 to TT, Research Fellowships 1081421 to AKV and 1154970 to GKS, and Investigator Grants 1176789 to AKV and 2025645 to GKS; through the Independent Research Institutes Infrastructure Support Scheme; and by the Victorian Government through an Operational Infrastructure Support Grant. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Pathogenic variants in only one allele of KAT6A have profound effects on cognitive development and function. This raises the question of how heterozygous and homozygous loss of KAT6A affect gene expression and chromatin modifications, specifically in a neural cell type derived from the developing brain that is functionally affected by loss of KAT6A. To address this question, we established that murine embryonic forebrain neural stem and progenitor cells form fewer neurons after loss of one and two alleles of Kat6a and used these cells to conduct automated chromatin profiling. We determined within the same cell preparation acetylation levels at histone H3 lysines 9, 14 and 23, as well as RNA polymerase II (POLR2A) and MLL1 occupancy and methylation levels at histone H3 lysine 4 (H3K4), the MLL1 target. We compared these chromatin profiling results to the effects of loss of KAT6A on gene expression.

The deletion of the Kat6a gene results in a reduction of histone H3 lysine 9 acetylation (H3K9ac) at HOX, DLX and TBX genes, among other genes [ 25 , 38 , 39 ], but no global genome-wide changes in H3K9 or H3K14 acetylation levels [ 25 ]. In addition, a reduction in the occupancy of HOX genes by the trithorax group protein mixed lineage leukaemia I [(MLL1), also known as histone-lysine N-methyltransferase 2A (KTM2A)] was observed in the absence of KAT6A [ 25 ]. Deletion of one or two alleles of Kat6a results in a global reduction in H3K23ac in mouse embryos ex vivo [ 42 ], in adult mouse brain cells [ 8 ], in adult mouse brain and in adult mouse peripheral white blood cells [ 7 ]. Knockdown of KAT6A causes a reduction in H3K23ac in cultured human cells [ 43 , 44 ], while lowerH3K14ac has also been reported [ 12 ]. Introduction of Arboleda-Tham syndrome mutations in KAT6A gene exons 2–15 in cultured human cells cause a reduction in H3K23ac [ 7 ]. Together, these studies suggest that KAT6A may promote gene expression by acetylating histone H3 at lysines 9, 14 and 23. Alternatively, some of the changes in histone acetylation levels may be secondary effects.

Homozygous null mutation of the Kat6a gene in mice causes an extensive anterior homeotic transformation of 19 skeletal and neural body segments and a posterior shift in HOX gene expression as well as lowered HOX gene expression [ 25 ]. Body segment identity is conferred by the sequential, collinear activation and nested expression of HOX genes in a specific anterior to posterior time course and with specific anterior HOX gene expression boundaries [reviewed in [ 31 – 33 ]]. Activation of HOX gene expression by treating pregnant females with retinoic acid rescues the Kat6a knockout body segment identity defect [ 25 ]. The time course and expression boundaries of HOX genes are controlled by the antagonistic action of polycomb repressor complexes and trithorax group proteins [reviewed by [ 34 – 36 ]]. For example, mutation of the polycomb repressor complex 1 protein, BMI1 (also known as PCGF4 or RNF51) leads to a posterior homeotic transformation [ 37 ]. Congruent with a role for KAT6A in activating HOX genes, combined loss of KAT6A and BMI1, rescues both the Kat6a and the Bmi1 knockout body segment identity defects and restores the anterior HOX gene expression boundaries to normal [ 23 ]. Furthermore, loss of KAT6A causes cleft palate and reduced expression of DLX genes [ 38 , 39 ] and aortic arch and cardiac septal defects along with reduced expression of TBX genes [ 39 , 40 ]. Transgenic overexpression of Tbx1 rescues the Kat6a knockout aortic arch defects [ 39 ]. Moreover, deletion of Kat6a in the germline or in adult haematopoietic cells leads to the complete absence of haematopoietic stem cells [ 19 , 24 , 41 ]. These observations demonstrate the critical roles that KAT6A plays during development of different organ systems.

Histone acetylation is generally more abundant at transcriptionally active gene loci [ 9 ] and thought to promote gene expression [ 10 , 11 ]. KAT6A is a member of the MYST family of histone acetyltransferases, which comprises five of the nine nuclear histone acetyltransferases [ 1 , 2 ]. KAT6A occurs as the catalytic component in a large multi-protein complex [ 12 ] that, besides KAT6A or its paralogue KAT6B, contains the adaptor proteins BRPF1, 2 or 3 and ING4 or 5, as well as MEAF6 [ 12 ]. BRPF1, 2 and 3 bind H2AK5ac, H4K12ac and H3K14ac via bromodomains and unmodified histone H3 tails and nucleosomal DNA via plant homeodomain (PHD) fingers [ 13 , 14 ]. ING4 and 5 bind histone H3 methylated on lysine 4 via their ING domains [ 15 – 17 ]. Comparison of the loss of function phenotypes between the protein complex members [ 18 – 28 ] and protein-protein interactions studies involving overexpression in cultured cells [ 29 , 30 ] suggest that the most prominent ‘KAT6’ protein complex composition during development is KAT6A (or alternatively KAT6B) with BRPF1, ING5 and MEAF6.

The lysine (K) acetyltransferase KAT6A is one of nine nuclear histone lysine acetyltransferases encoded by the mammalian genome that have structurally defined acetyltransferase domains [ 1 , 2 ]. Heterozygous pathogenic variants in the KAT6A gene cause Arboleda-Tham syndrome, characterised by global developmental delay and cognitive dysfunction (93% penetrance), speech delay (73%), visual impairment (77%), gastrointestinal dysfunction (69%), sleep disruption (65%), congenital heart defects (51%; including septal defects), frequent infections (47%) and autism-like behaviour (32%), among other less common anomalies [ 3 – 6 ]. Most pathogenic variants (90% to 92%) in KAT6A are truncating mutations [ 5 , 6 ]. Variants truncating the KAT6A protein in more carboxy-terminal regions appear to be associated with more severe anomalies [ 5 ]. Consistent with heterozygous pathogenic variants in the KAT6A gene causing the Arboleda-Tham cognitive impairment syndrome, Kat6a +/– mice display learning and memory defects [ 7 , 8 ], as well as hyperactivity and autism-like sociability defects [ 7 ].

(A,B) CUT&Tag results of NSPCs isolated from N = 3 Kat6a +/+ , 4 Kat6a +/– and 3 Kat6a –/– E12.5 embryos. Data were analysed as described in the methods section under Automated CUT&Tag sequencing data analysis. (A,B) Total number of CUT&Tag reads mapped for the histone marks and proteins indicated in proliferating (A) and differentiating (B) Kat6a +/+ , Kat6a +/– and Kat6a –/– NSPCs. Each circle in represents NSPCs isolated from an individual mouse embryo. Data are presented as mean ± SEM and were analysed by two-way ANOVA with Tukey’s multiple comparison test.

Similarly, in differentiating Kat6a –/– vs. Kat6a +/+ NSPCs, changes at promoters in H3K9ac (R 2 = 0.61), H3K4me3 (R 2 = 0.56), H3K14ac (R 2 = 0.42) and POLR2A (R 2 = 0.41) correlated positively with change in RNA levels of individual genes, while H3K23ac (R 2 = 0.10) correlated only weakly ( S5 Fig ). Few changes in MLL1 occupancy reached the FDR < 0.05 cutoff in differentiating NSPCs and so in the case of MLL1 the relationship between MLL occupancy and RNA levels is merely suggestive of a possible correlation in the promoter regions of genes ( S5G Fig ). In the gene bodies of differentiating Kat6a –/– vs. Kat6a +/+ NSPCs, changes in H3K9ac (R 2 = 0.62), H3K4me3 (R 2 = 0.56), POLR2A (R 2 = 0.56) and to a lesser extent H3K14ac (R 2 = 0.29) correlated positively with RNA levels, while H3K23ac and MLL1 did not suggest a prominent correlation ( S5 Fig ).

(A-H) Comparison of results from RNA sequencing and CUT&Tag results of proliferating NSPCs isolated from N = 3–4 Kat6a +/+ and 3–4 Kat6a –/– E12.5 embryos. Data were analysed as described in the methods section under RNA sequencing data analysis and Automated CUT&Tag sequencing data analysis. Differences in gene expression or occupancy with a false discovery rate (FDR) < 0.05 were considered significant. (A-H) Correlation between log 2 fold-change in RNA levels and log 2 fold-change in H3K9ac levels (A,B), H3K14ac levels (C,D), POLR2A occupancy (E,F) and H3K23ac levels (G,H) in proliferating Kat6a –/– vs. Kat6a +/+ NSPCs at promoters (A,C,E,G) and bodies (B,D,F,H) of protein coding genes differentially expressed (FDR < 0.05) and differentially occupied (FDR < 0.05).

(A-D) Comparison of results from RNA sequencing and CUT&Tag results of NSPCs isolated from N = 3–4 Kat6a +/+ and 3–4 Kat6a –/– E12.5 embryos. Data were analysed as described in the methods section under RNA sequencing data analysis and Automated CUT&Tag sequencing data analysis. Differences in gene expression or occupancy with a false discovery rate (FDR) < 0.05 were considered significant. (A-D) Correlation between log 2 fold-change in RNA levels and log 2 fold-change in MLL1 occupancy (A,C) or H3K4me3 levels (B,D) of proliferating Kat6a –/– vs. Kat6a +/+ NSPCs at promoters (A,B) and bodies (C,D) of protein coding genes differentially expressed at FDR < 0.05 and differentially occupied at FDR < 0.05. (E) Read depth plots of CUT&Tag results for MLL1 occupancy at the Sox1, Otx1, Emx2 and Dmrta2 locus. (F) Log 2 fold-change in RNA levels of Sox1, Otx1, Emx2 and Dmrta2 in proliferating and differentiating Kat6a –/– vs. Kat6a +/+ and Kat6a +/– vs. Kat6a +/+ NSPCs.

We observed the tightest positive correlation between the change in MLL1 occupancy at promoters and change in RNA levels per gene in proliferating Kat6a –/– vs. Kat6a +/+ NSPCs Fig 4A ). MLL1 occupancy (R 2 = 0.94) and H3K4me3 (R 2 = 0.75) were decreased at the promoters of neural stem cell and early brain development genes ( Fig 4A and 4B ) and in the gene bodies (R 2 = 0.62 and 0.81, respectively), including at the Sox1, Otx1, Emx2 and Dmrta2 genes in the absence of KAT6A ( Fig 4C – 4E ). Congruently, mRNA output of these genes was downregulated significantly in Kat6a –/– vs. Kat6a +/+ NSPCs. In Kat6a +/– vs. Kat6a +/+ NSPCs, these genes only showed a tendency for downregulation ( Fig 4F and S1 to S4 Tables ). The other chromatin marks that showed a positive correlation between changes in the chromatin mark and changes in RNA levels in proliferating Kat6a –/– vs. Kat6a +/+ NSPCs at specific gene promoters and bodies were H3K9ac (R 2 = 0.69 and 0.69, respectively), H3K14ac (R 2 = 0.82 and 0.28) and POLR2A (R 2 = 0.55 and 0.85; Fig 5A to 5F ). Interestingly, changes in H3K23ac did not correlate with changes in RNA levels in proliferating Kat6a –/– vs. Kat6a +/+ NSPCs ( Fig 5G and 5H ).

Overall, we observed a skewing towards a decrease in H3K23ac in the absence of KAT6A. This was observed across a large number of individual genes in both proliferating and differentiating NPSCs, in particular at promoters and enhancers ( Fig 3H ). This, however, does not preclude that other, less numerous changes are functionally relevant.

Like proliferating NSPCs, differentiating Kat6a –/– vs. Kat6a +/+ NSPCs also showed the greatest skewing towards a reduction in H3K23ac in promoters (1396 down, 889 up; Fig 3G and S6 and S7 Tables ) and active enhancers (241 down, 137 up; S6 and S7 Tables ). While significant changes were observed for other chromatin marks in differentiating Kat6a –/– vs. Kat6a +/+ NSPCs, these were not obviously skewed with respect to the number of genes showing an increase vs. a decrease in individual chromatin marks ( S6 and S7 Tables ).

The largest number of significant changes at the level of individual genes were observed in H3K23ac levels, changed in the body of the genes of 7448 individual genes in proliferating Kat6a –/– vs. Kat6a +/+ NSPCs (3683 increased and 3765 decreased) and at 6873 individual genes in differentiating Kat6a –/– vs. Kat6a +/+ NSPCs (3434 increased and 3439 decreased; S5 , S6 , and S7 Tables ). However, a skewing towards a reduction in the acetylation mark is what one would expect for results including direct targets of KAT6A. A predominant decrease in proliferating Kat6a –/– vs. Kat6a +/+ NSPCs was observed for H3K23ac levels at active enhancers (312 genes decreased in H3K23ac and 110 genes increased; S5 and S7 Tables ), promoters (2039 down, 909 up; Fig 3F and S5 and S7 Tables ) and all enhancers (1402 down, 835 up; S5 and S7 Tables ). In addition, H3K9ac, H3K14ac and H3K4me3 were skewed towards a reduction particularly at active enhancers, albeit in a smaller number of genes ( S5 and S7 Tables ).

In addition to examining the read count aggregated over all genes ( Figs 3A – 3E and S3A - S3M ), we assessed the read count aggregates separately over genes that were upregulated and genes that were downregulated in proliferating and differentiating Kat6a –/– vs. Kat6a +/+ NSPCs ( S4 Fig ). The results over upregulated and downregulated genes ( S4A - S4P Fig ) were similar to the read count aggregated over all genes ( Figs 3A – 3E , S3A , S3B , S3D – S3F , and S3I ). Only a modestly stronger genotype effect of the reduction in histone acetylation, MLL1 occupancy, H3K4me3 and POLR2A occupancy was observed in the downregulated as compared to the upregulated genes ( S4Q – S4T Fig ).

(A-H) CUT&Tag results of NSPCs isolated from N = 3 Kat6a +/+ , 4 Kat6a +/– and 3 Kat6a –/– E12.5 embryos. Data were analysed as described in the methods section under Automated CUT&Tag sequencing data analysis. Differences in occupancy with a false discovery rate (FDR) < 0.05 were considered significant. Data in (A) were analysed by Kruskal-Wallis test followed by Dunn’s correction for multiple testing. (A) Log 2 of CUT&Tag read count per kilobase normalised to library size (RPKM) accrued over gene bodies for H3K9ac, H3K14ac, H3K23ac, MLL1, H3K4me3 and POLR2A in differentiating Kat6a +/+ , Kat6a +/– and Kat6a –/– NSPCs. (B-E) Read depth aggregates over all protein coding genes for H3K9ac (B), H3K23ac (C), MLL1 (D) and H3K4me3 (E) over the interval from -1 kb to +1 kb of the transcription start site (TSS) in differentiating Kat6a +/+ , Kat6a +/– and Kat6a –/– NSPCs. (F,G) MA plot of genes differentially decorated with H3K23ac in promoters of proliferating (F) and differentiating (G) Kat6a –/– vs. Kat6a +/+ NSPCs. (H) Number of genomic regions displaying differences in levels of H3K23ac in proliferating and differentiating Kat6a –/– vs. Kat6a +/+ NSPCs.

We performed automated CUT&Tag chromatin profiling [ 47 ] on NSPCs isolated from the dorsal telencephalon (the cerebral cortex precursor) of Kat6a +/+ , Kat6a +/– and Kat6a –/– E12.5 embryos, to detect multiple chromatin marks in the same biological cell sample. These included the acetylation targets proposed for KAT6A, namely H3K9ac, H3K14ac and H3K23ac, as well as H3K4me3, MLL1 and RNA polymerase II occupancy. We had previously observed MLL1 occupation of HOX genes to be dependent on the presence of KAT6A in whole E10.5 embryos [ 25 ]. Thus, we included MLL1. Within its four to five subunit protein complex, MLL1 has H3K4me2/3 methyltransferase activity [ 48 – 51 ]. We therefore assessed H3K4me3, which is the target of MLL1 that is most closely associated with mRNA levels [ 52 ]. In addition, we included RNA polymerase II subunit A phosphorylated on Ser2 (POLR2A; also known as Phospho-Rpb1 CTD). Phosphorylation of POLR2A at Ser2 is required for elongation and allowed us to determine if the absence of KAT6A affected POLR2A progression through a gene locus.

The analysis of proliferating and differentiating Kat6a +/– vs. Kat6a +/+ NSPCs did not return any differentially expressed individual genes that reached FDR < 0.05 significance ( S3 and S4 Tables ). However, gene set analysis showed that gene expression changes in Kat6a +/– vs. Kat6a +/+ NSPCs correlated positively and significantly with genes expression changes in Kat6a –/– vs. Kat6a +/+ NSPCs, both, for the proliferating and the differentiating condition (p = 0.0002 and 0.0001, respectively; Fig 2K and 2L ), suggesting that a multitude of modest changes in gene expression might contribute to the dysfunctions observed in individuals with heterozygous pathogenic variants in the KAT6A gene and in Kat6a +/– mice.

(A-L) RNA sequencing data of NSPCs isolated from N = 4 Kat6a +/+ , 4 Kat6a +/– and 4 Kat6a –/– E12.5 embryos. Data were analysed as described in the methods section under RNA sequencing data analysis. Differences in gene expression with a false discovery rate (FDR) < 0.05 were considered significant. (A,B) M (log 2 ratio) and A (mean average) plot of genes differentially expressed in Kat6a –/– vs. Kat6a +/+ NSPCs grown under proliferating (A) and differentiating conditions (B) for 3 days. The number of upregulated (red) and downregulated genes (blue) are indicated. Genes not significantly changed are indicated in black. (C,D) Heatmaps showing hierarchical clustering of samples and genes differentially expressed in the contrast of proliferating (C) and differentiating (D) Kat6a –/– vs. Kat6a +/+ NSPCs but also displaying results for Kat6a +/– NSPCs. (E,F) Log 2 fold-change in RNA levels of the top 20 genes (by fold-change amplitude, average expression > 1 CPM) downregulated in proliferating (E) and differentiating (F) Kat6a –/– vs. Kat6a +/+ NSPCs. FDRs shown below the bars. (G,H) Log 2 fold-change in RNA levels of SOX genes (G) and homeodomain protein genes (H) in differentiating Kat6a –/– vs. Kat6a +/+ NSPCs. FDRs shown below the bars. (I,J) Top 20 gene ontology terms (biological process) associated with genes downregulated in proliferating (I) and differentiating (J) Kat6a –/– vs. Kat6a +/+ NSPCs. (K,L) Barcode plots showing significant positive correlations between gene expression changes in proliferating (K; p = 0.0002) and differentiating (L; p = 0.0001) Kat6a –/– vs. Kat6a +/+ and Kat6a +/– vs. Kat6a +/+ E12.5 NSPCs. The horizontal axis shows t statistics for all genes in the Kat6a –/– vs. Kat6a +/+ dataset. The long vertical lines represent genes and the shorter vertical lines the t-statistics of genes in the Kat6a +/– vs. Kat6a +/+ dataset. Worms show the relative enrichment of the genes up (red) and downregulated (blue) in the Kat6a +/– vs. Kat6a +/+ dataset.

(A) H3K23 acetylation levels and pan histone H3 levels assessed by Western blotting (A) and densitometry (B) in Kat6a +/+ wildtype and Kat6a –/– homozygous mutant NSPCs. Each lane was loaded with 0.5 µg of acid extracted protein from NSPCs isolated from an individual mouse embryo. H3K23ac levels were normalised to pan-H3 levels. (C) Representative phase contrast images of proliferating Kat6a +/+ wildtype, Kat6a +/– heterozygous mutant and Kat6a –/– homozygous mutant NSPCs. (D) Percentage of cells expressing the neural stem cell (NSC) marker protein SOX2 assessed by intranuclear immunofluorescence staining and flow cytometry in proliferating Kat6a +/+ , Kat6a +/– and Kat6a –/– NSPCs. (E,F) Cumulative cell count of Kat6a +/+ , Kat6a +/– and Kat6a –/– NSPCs grown over 5 passages (P1 to P5) in culture (E) and cumulative cell number of cells at passage 5 (F). (G-J) Percentage of cells expressing the neuronal marker protein ßIII-tubulin (G), the astrocyte marker glial acidic protein (GFAP; H), the oligodendrocyte epitope O4 (I) and the NSC marker protein SOX2 but not GFAP (J) assessed by immunofluorescence staining and flow cytometry of Kat6a +/+ , Kat6a +/– and Kat6a –/– NSPCs after three days of culture in differentiation conditions. (K) Representative epifluorescence images of Kat6a +/+ , Kat6a +/– and Kat6a –/– NSPCs after five days of differentiation stained for ßIII-tubulin (red) and GFAP (blue). NSPC isolates from N = 3 (A-C,E,F) and 4 (D,G-K) E12.5 embryos per genotype. Each circle in (B,D,F-J) represents NSPCs isolated from an individual mouse embryo. Data are presented as mean ± SEM and were analysed by unpaired, two-tailed Student’s t-test (B), one-way ANOVA with Tukey’s multiple comparison test (D,F-J) and two-way ANOVA with Tukey’s multiple comparison test (E). Scale bars in (C) equals 110 µm and in (K) 67 µm.

We determined the effects of loss of KAT6A on histone acetylation at histone H3 lysine 9, 14 and 23 by comparing acetylation in Kat6a +/+ and Kat6a –/– NSPCs. These are the previously published targets in a variety of human cell types and mouse embryos [ 12 , 25 , 39 , 43 , 44 ]. Global histone H3 lysine 23 acetylation (H3K23ac) was moderately, but consistently reduced in Kat6a –/– vs. Kat6a +/+ NSPCs ( Fig 1A and 1B ), whereas genome-wide acetylation levels at H3K9 and H3K14 were unchanged ( S1 A - S1D Fig ). Examination of NSPCs cultured under proliferation conditions in vitro revealed the expected appearance of Kat6a +/+ wildtype NSPCs as spherical, floating colonies (neurospheres) ( Fig 1C ). In contrast, Kat6a +/– NSPCs formed smaller colonies that tended to attach and Kat6a –/– NSPCs formed few floating colonies and instead attached to the uncoated tissue culture plastic ( Fig 1C ). However, NSPCs of all three genotypes robustly expressed the neural stem cell marker protein SOX2 ( Fig 1D ). Kat6a –/– NSPCs proliferated more slowly than the other two genotypes ( Fig 1E and 1F ).

Discussion

In this study, we have shown that the absence of KAT6A causes a reduction in NSPC proliferation and neuronal differentiation. A proliferation defect has also been observed in embryonic forebrain NSC isolated from mice with a point mutation in the KAT6A histone acetyltransferase domain in a previous report by Perez-Campo and co-workers [45]. Perez-Campo and colleagues did not assess differentiation potential of the NSCs or effects on histone acetylation or gene expression. We observed changes in the expression of thousands of genes, a global reduction in H3K23ac and gene-specific reductions in H3K9ac, H3K14ac, H3K4me3, as well as reductions in occupancy by MLL1 in the absence of KAT6A. Unexpectedly, not changes in H3K23ac, but changes in H3K4me3, H3K9ac, H3K14ac, MLL1 and POLR2A were positively correlated with changes in gene expression levels when KAT6A was lacking.

The loss MYST histone acetyltransferase family members causes profound reductions in specific histone acetylation marks; loss of KAT7 (HBO1) causes the absence of H3K14ac in neural stem cells [53]; H4K16ac is eliminated in the absence of KAT8 (MOF) in mouse preimplantation embryos [54,55], and H2AZ acetylation at lysine 7 is abolished in the absences of KAT5 (TIP60) in mouse embryonic fibroblasts and human 293T and U2OS cells [56]. Congruent with our data reported here, loss of KAT6A does not result in the abolition of any histone mark, but rather a global reduction of around 50% in H3K23ac [42–44], suggesting that other acetyltransferases also mediate H3K23ac. Loss of KAT6B causes a global reduction in H3K9ac in NSPCs, E12.5 dorsal telencephalon, E16.5 cortical neurons and adult cortex and affects H3K23ac to a lesser extent [57,58].

H3K23ac is the most abundant histone acetylation mark in the mammalian genome. It occurs at 20% to 50% of all nucleosomes [59,60] and is present in genic, as well as intergenic regions [44]. Its levels within genes are positively correlated with gene expression levels [44]. In our data presented here, H3K23ac is clearly the most profoundly and the only globally affected histone acetylation target in the absence of KAT6A. While it is not reduced to zero, the reduction is highly reproducible and detectable at a genome-wide level both by western blotting and CUT&Tag. However, the reduction in H3K23ac did not appear to correlate significantly with changes in gene expression. In contrast, the changes in MLL1 occupancy and changes in its target, H3K4me3, correlated highly and positively with changes in gene expression in the absence vs. presence of KAT6A. The absence of a correlation between changes in H3K23ac and RNA expression levels and the presence of a correlation between changes in MLL1, as well as H3K4me3, and RNA expression levels leads us to hypothesise that KAT6A has at least two functions: (1) a global function in acetylation H3K23, which we hypothesise may be required to maintain regions of the genome accessible for transcription factors and transcriptional co-regulators, and (2) facilitating the recruitment of MLL1 (and possibly other SET domain proteins) to developmental control genes such as HOX, DLX, TBX and SOX genes. Function (1) appears to be more essential in differentiating NSPCs, where we observed a Kat6a gene dose dependency, compared to proliferating NSPCs, possibly because another histone acetyltransferase contributes to this role in proliferating cells. Function (2), namely affecting MLL1 occupancy at specific genes, appears to be exceptionally tightly aligned with changes in RNA levels in proliferating NSPCs. Interestingly, histone H3 poly-acetylation has been shown to enhance MLL1-mediated methylation of nucleosomal H3K4 [61], contributing to a possible mechanism. MLL1 is essential for neurogenesis and positional information in postnatal neural stem cells [62,63]. Deletion of the Mll1 gene in the developing forebrain results in cognitive deficits and decreased expression of the homeodomain transcription factor gene Meis2 along with decreased levels in H3K4me3 at the Meis2 gene [64]. Congruently, we observed reduced MLL1 occupancy of the Meis2 locus and reduced Meis2 mRNA levels in proliferating Kat6a–/– NSPCs. Suggesting a close interaction of KAT6A and MLL1, endogenous KAT6A and the endogenous C-terminal fragment of MLL1 have been reciprocally co-immunoprecipitated in HEK293T and K562 cells [65] and MLL1 occupancy of HOX genes is reduced in Kat6a–/– mouse embryos [25].

Alternative to a specific effect of KAT6A on histone acetylation levels and MLL1 activity, all chromatin changes observed to correlate strongly and positively with gene expression changes could be a consequence of a yet to be discovered effect of KAT6A. However, since KAT6A has been shown to acetylate histone substrate [43,44] and MLL1 has been shown to be an order of magnitude more effective in methylating H3K4 when histone H3 is poly-acetylated [61], a mechanistic connection between KAT6A and MLL1 is likely. Remarkably, it has recently been established that the KAT6A target H3K23ac (along with H3K18ac) most robustly promotes H3K4me3 by MLL1 [66]. In this context, it is worth noting that the developmental anomalies of observed in Kat6a–/– mice, namely anterior homeotic transformation, cleft palate, interrupted aortic arch and cardiac septal defects [25,39], would be curiously specific for a co-activator that acetylates H3K23 at 20% to 50% of all nucleosomes, obviously not restricted to the genes involved in these anomalies, HOX, DLX and TBX genes. In contrast, a requirement for KAT6A in recruiting MLL1 would explain this level of specificity. The role of MLL1 in regulating HOX genes is well established [67]. A role for MLL1 in DLX gene regulation has also been reported [63]. Moreover, combined mutation of Kat6a and the polycomb repressor complex protein encoding gene Bmi1 rescue the homeotic transformation and HOX gene expression pattern defects observed in the individual gene mutations [23], just as combined mutation of Mll1 and Bmi1 do [68]. Taken together, a direct or indirect requirement for KAT6A for MLL1 recruitment and function would be congruent with data resulting from a number of studies.

KAT6A is a large multi-domain protein. Besides the centrally located MYST family-type histone lysine acetyltransferase (HAT) domain, which contains an acetyl-co-enzyme A binding pocket and a histone binding region [69], it has two amino-terminal winged helix domains, one of which has been shown to bind DNA [WHD; [70,71]], followed by two plant-homeodomain (PHD) fingers, which to bind to unmodified H3R2 and acetylated or crotonylated H3K14 [44,72,73]. Carboxyterminal of the HAT domain, KAT6A has an acidic region, a serine- and a methionine rich region. The serine- and a methionine rich region has been shown to interact with the RUNX family transcription factor, RUNX2 [74]. Given the number of protein domains, it is possible that KAT6A may directly interact with MLL proteins, which has been shown for another MYST family histone acetyltransferase, KAT8 [MOF; [75]]. Indeed, endogenous KAT6A and the endogenous C-terminal fragment of MLL1 have been reciprocally co-immunoprecipitated in HEK293T and K562 cells [65]. Alternatively, or in addition, KAT6A may interact with MLL indirectly. A physical interaction between KAT6A and the chromosomal translocation fusion partner of MLL1, ENL, has previously been suggested in the context of leukaemia [76,77]. A strong functional cooperation between the native KAT6A and MLL1 proteins has been shown in leukaemia [78].

Loss of one allele of Kat6a caused no changes in gene expression that reached transcriptome-wide significance after correction for multiple testing in our NSPC datasets and in adult hippocampus [8]. Expression changes were observed after loss of one allele of Kat6a using single nucleus RNA sequencing of the CA3 regions of the hippocampus [8]. We observed the differential expression of 3718 and 5691 genes caused by the loss of two alleles of Kat6a in proliferating and differentiating NSPCs, respectively. Our observation that the Kat6a heterozygous expression profiles were intermediate between wild-type and homozygous profiles suggests that a multitude of subtle changes may exist in the Kat6a heterozygous state, which could be relevant to individuals with pathogenic variants in one allele of KAT6A. The effects of the complete loss of KAT6A on gene expression indicates that KAT6A is essential for the normal expression of a large number of genes, as one would expect since KAT6A is one of only 9 nuclear histone acetyltransferases [1,2], representing the capacity of the human genome to regulate approximately 20,000 protein coding genes, and non-coding genes in addition, at the level of histone acetylation. Accordingly, the defects in brain development and function that accompany Arboleda-Tham syndrome are likely to be caused by modest deregulation of many of genes, with possible contribution of multiple genes to one functional defect. For example, even just among the top 20 genes downregulated in NSPCs lacking KAT6A, four are required for neuron survival (Esco2, Eomes, Insm1, Elavl4), three for neuronal precursor proliferation (Esco2, Eomes, Mycn) and three for a normal brain size (Eomes, Kif14, Mycn) [79–84].

While severe brain anomalies are generally reported in mice that have lost both alleles of a specific gene and the experimental examination of the compound molecular effects of the subtle reduction in expression of thousands of genes may not be feasible, multiple subtle changes in gene expression likely underly the anomalies observed in brain function in individuals with heterozygous pathogenic variants in the KAT6A gene.

[END]
---
[1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1012133

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/