Mammalian H4K16ac regulates the spatiotemporal order of genome replication rather than gene expression

Abstract

Histone acetylation is widely assumed to directly instruct gene activation. Among acetylated residues, H4K16ac is one of the most abundant modifications, conserved across all eukaryotes. Despite its established role in X-chromosome hyperactivation in Drosophila, its function in mammalian cells has remained elusive. Here, we show that in human somatic cells, H4K16ac does not substantially affect gene expression, but instead controls the spatiotemporal program of genome replication. By combining a meta-analysis of public datasets and perturbation experiments designed to minimize confounding effects, we found that H4K16ac is neither associated with nor required for transcriptional activity. Rather, H4K16ac depletion resulted in premature replication of heterochromatic regions and widespread alterations in replication timing across the genome. These defects were driven by the aberrant activation of cryptic replication origins at long terminal repeats-repetitive elements typically marked by H4K16ac and whose sequence context resembles that of canonical origins in euchromatic regions. Our findings reveal an unexpected role for one of the most prevalent chromatin modifications and uncover a new regulatory mechanism that safeguards genome replication fidelity.

Graphical Abstract

Figure 1.

Meta-analysis of H4K16ac distribution and relationship with gene activity in public datasets. (A) Schematic of the strategy used in this study. Created in BioRender. Milan, M. (2025), https://BioRender.com/d1vcdqf. (B) Summary of public H4K16ac profiling datasets analyzed in this study. Cell lines profiled in multiple studies are color-coded. (C) H4K16ac tracks showing high variability in distribution among datasets. (D). H4K16ac metaprofiles at coding genes from different datasets. Dataset number in the upper right corner: same as panel (B). TTS, transcription termination site. (E) Distribution of ChIP-seq signal at quartiles of gene expression in the indicated cell lines. Reads per kilobase per million (RPKM) are the sum of signal from promoter [1 kb upstream of transcription start site (TSS)] and gene body. P-values from one-way Analysis of Variance (ANOVA). (F) Tracks showing high level of expression of the housekeeping β-actin gene in both activated (aB) and resting (rB) B cells but lack of H4K16ac enrichment in rB. (G) Visualization of H4K16ac and RNA signal at promoter of coding genes in the indicated cell lines. (H) Visualization of H4K16ac signal at differentially expressed genes (DEGs; False discovery rate (FDR) < 0.01, |log₂FC| ≥ 1) in the transition from resting (rB) to activated (aB) B cells, sorted by fold change. (I) Relationship between H4K16ac signal (RPKM at promoter—1 kb upstream of TSS—and gene body) and fold change of DEGs. Each dot is a gene. N = 5781.

Figure 2.

Background-corrected H4K16ac distribution across the genome and impact on transcriptional regulation. (A) Immunofluorescence microscopy of HME1 cells showing complete loss of the H4K16ac mark in a MSL3 KO clone. Cntr: KO of the nonexpressed gene TNP2. Scale bars: 20 μm. (B) Experimental strategy used in this study to control for possible technical variability affecting H4K16ac profiling. (C) ChIP-seq tracks of H4K16ac signal detected in the indicated samples using three different antibodies (Ab) and corresponding background-corrected signal. H3K36me3 is shown as a marker of actively transcribed genes. (D) Metaprofiles of H4K16ac ChIP-seq signal (antibody #1) at coding genes, as detected in the indicated cells (left) or after subtraction of KO-derived signal from Cntr signal (right). (E) Percentage of the genome (10 kb bins) showing enrichment of the indicated ChIP-seq signal in HME1 cells (see the ‘Materials and methods’ section). (F) Percentage of peptides containing the indicated modifications as detected by mass spectrometry analysis of HME1 cells. (G) Signal distribution at HiC-defined chromatin compartments of the indicated histone modifications. The corrected signal from antibody #1 is shown for H4K16ac. A, euchromatin; B, heterochromatin. P-values from two-sided Wilcox test. (H) Distribution of ChIP-seq signal at quartiles of gene expression in HME1 cells. RPKM signal from gene body of corrected H4K16ac ChIP-seq Ab#1 (left) and H3K36me3 ChIP-seq (right) is shown as relative to the mean of the highest gene quartile to enable comparison of the two signals on a common scale. P-values from one-way ANOVA. (I) Visualization of corrected H4K16ac signal distribution (antibody #1) at repetitive elements. LTR and SINE enriched for the mark are shown. FL L1: full length L1. (J) Metaprofiles of H4K16ac signal at different classes of repetitive elements. Corrected signal for antibody #1 is shown. (K) Quantification of gene expression changes upon acute inactivation of MSL-complex specific subunits. Genes with |log₂FC| ≥1 and FDR ≤ 0.01 relative to the control (acute inactivation of the not-expressed gene TNP2) were considered differentially expressed. (L) Quantification of c-MYC target genes by RT-qPCR data in the indicated samples upon c-MYC overexpression. N = 5 biological replicates. P-value from Mann–Whitney test. (M) Comparison of global mRNA levels in control and MSL KO cells by spike-in-normalized RNA-seq analysis. Every dot is a synthetic ERCC spike-in RNA. Human genes are represented as density plot.

Figure 3.

Replication defects in H4K16ac-deficient cells. (A) Representative images and quantification of interphase HME1 cells containing double positive pRPA/γH2AX foci in the indicated cell lines, detected by immunofluorescence microscopy. Scale bar: 2 μm. From left to right, n = 3162, 3066, 1285, and 718 cells. Values are the mean ± SD from n = 3 well areas. P-values from two-tailed Fisher’s exact test. (B) Aphidicolin sensitivity assay. Values are the mean ± SD from n= 3 biologically independent samples after 5 days of growth. P-values from unpaired t-test with Welch correction. Legend: same as in panel (A). UT: untreated. (C) Limiting dilution clonogenic assays quantifying the long-term proliferative capacity of the indicated cell lines. The percentage of populated wells (see the ‘Materials and methods’ section) relative to the corresponding control condition is shown. Values are the mean ± SD from n = 3 biologically independent samples. P-values from unpaired t-test with Welch correction. Legend: same as in panel (A). Empty: cells transduced with the empty control vector. (D) Flow cytometry visualizing DNA replicating EdU + cells in the indicated conditions. After synchronization at the G1/S border and release, cells synchronously incorporate EdU into nDNA. In the presence of HU, active origins are arrested soon after firing enabling mapping of replication initiation sites. (E) Large-scale distribution of H4K16ac and EdU-HU signals in control HME1 cells. The corrected signal is shown for all H4K16ac antibodies. (F) Quantification of replication initiation sites marked by H4K16ac. Data from antibody #1 are shown. Similar percentages were observed with antibody #2. (G) Visualization of EdU-HU signal in the indicated cells at replication initiation sites. (H). Distribution of EdU-HU signal at the top 50% constitutive replication initiation sites in the indicated cell lines. Mean of the RPM of the 100 kb surrounding the peak summit from two replicates is plotted. P-values from two-sided Wilcox test. (I) Principal component analysis (PCA) plot showing sample variance between the indicated conditions. The first two components and their percentage contribution are shown. (J) Example of differential replication sites showing highly reproducible patterns in biological replicates. (K) Overlap of activated replication initiation sites in the indicated conditions. Activated sites are defined as log₂ fold change ≥ than 0.7, FDR ≤ 10⁻¹⁰. (L) Relative gene abundance at the indicated groups of replication initiation sites. (M) Distribution of the indicated histone modification signals at constitutive and activated (red: MSL KO; green: c-MYC) replication initiation sites. RPM of the 100 kb surrounding the replication peak summit are plotted. P-values from two-sided Wilcox test. See the ‘Materials and methods’ section for ChIP-seq data sources. (N) DNA fiber assay. Top: Representative DNA fiber images showing CldU (green) and IdU (red) incorporation in Cntr and MSL KO cells. Bar = 5 μm. Bottom: Quantification of fork speed (kb/min). Overlaid dots: Averages from three independent experiments. Horizontal bars: Median of all data points. Ns, nonsignificant (unpaired Student’s t-tests assessing the difference in means of two data groups). (O) Metaprofiles of EdU signal showing unaffected replication fork progression in MSL KO cells. Merged replicate tracks are visualized as mean ± standard error of the mean.

Figure 4.

LTRs are cryptic replication origins repressed by H4K16ac. (A) Enrichment analysis of repetitive elements at replication initiation sites activated upon MSL KO (top) or c-MYC overexpression (bottom). Enrichment is calculated through a permutation test comparing activated initiation sites to all the other replication initiation sites. (B) Track showing LTR enrichment at a replication initiation site activated in MSL-KO cells. (C) Metaprofiles of EdU and H4K16ac (corrected ChIP-seq with antibody #1) signals in the indicated conditions at LTR overlapping replication initiation sites. Signals are stranded based on LTR orientation. (D) Schematic illustrating the relationship between initiation sites and individual replication origins at different scales, and the nucleotide compositions of origin-flanking regions that favors their activation. Based on data from Tubbs et al. [44]. (E, F). Metaprofiles of the indicated features at LTRs. The profiles of AT content (E) and poly(dA)/poly(dT) tracts (F) are shown relative to the EdU-HU signal detected in H4K16ac-depleted cells. Signals are stranded based on LTR orientation. (G) AT content profiles at the indicated LTR subsets. (H) Distribution of the indicated histone modification signals at LTR subsets. P-values from two-sided Wilcox test. See the ‘Materials and methods’ section for ChIP-seq data sources.

Figure 5.

Altered RT in H4K16ac-deficient cells. (A) Flow cytometry of asynchronous cells illustrating the subset of cells analyzed by REPLI-seq upon cell sorting. (B) Overlay of REPLI-seq tracks from control and H4K16ac-deficient HME1 cells (see also Supplementary Fig. S6A). (C) Quantification of the indicated features in regions displaying differential RT in H4K16ac-deficient HME1 cells (see the ‘Materials and methods’ section for definitions). (D) Enrichment analysis of repetitive elements at anticipated regions upon H4K16ac loss. Enrichment is calculated through a permutation test comparing anticipated regions to random regions in the genome. (E) Distribution of RT index of 50 kb genomic bins overlapping the indicated element, as measured in control HME1 cells. P-values from two-sided Wilcox test. (F) Distribution of RT index of LTRs in the indicated cell lines. P-values from two-sided paired Wilcox test. (G) Track showing the RT of chromosome 12’s long arm in the indicated cell lines, highlighting the global effect of H4K16ac loss. (H) Distribution of RT index at 50 kb bins across the genome in the indicated cell lines. Bins are classified as early or late based on the RT index measured in control cells. P-values from two-sided paired Wilcox test. (I) Genome-wide relationship between RT indexes measured in control and H4K16sc-depleted cells. Each dot is a 50 kb genomic bin. Note the overall tilted pattern relative to the diagonal, and the greater differences observed in late vs early regions (red arrows). (J) Model of how H4K16ac preserves genome stability by repressing LTR-associated AT-flanked cryptic replication origins. In its absence, premature replication of LTR enriched heterochromatic regions alters the temporal control of genome duplication. Created in BioRender. Milan, M. (2025), https://BioRender.com/n87p237. (K) S phase duration in individual cells of the indicated FUCCI(CA)2-expressing cell lines. P-values from one-way ANOVA. (L) Time-lapse images of representative FUCCI(CA)2-expressing cells showing prolonged S phase in H4K16ac-deficient cells in the second cycle after Palbociclib release.