Aerobic glycolysis and lactate regulate histone H3K18Lactylation occupancy to fine-tune gene expression in developing and mature retina

Abstract

High aerobic glycolysis in retinal photoreceptors, as in cancer cells, is implicated in mitigating energy and metabolic demands. Lactate, a product of glycolysis, plays a key role in epigenetic regulation through histone lactylation in cancer. Here, we demonstrate that increased ATP production during retinal development is achieved primarily through augmented glycolysis. Histone lactylation, especially H3K18La, parallels enhanced glycolysis and lactate in developing retina and in retinal explants. Multi-omics analyses, combined with confocal imaging, reveal the localization of H3K18La near H3K27Ac in euchromatin at promoters of active retinal genes. H3K18La and gene expression are also concordant with glucose metabolism in retinal explants. Evaluation of accessible chromatin at H3K18La marked promoters uncovers an enrichment of GC-rich motifs for transcription factors of SP, KMT and KLF families, among others, indicating specificity of H3K18La-mediated gene regulation. Our results highlight glycolysis/lactate/H3K18La as a regulatory axis in fine-tuning gene expression in developing and mature retina.

Keywords: Histone modification; cellular metabolism; epigenome; gene regulation; photoreceptor function; retinal development.

Figure 1:. Enhanced contribution of aerobic glycolysis and lactate to ATP generation during retinal development

A) Graphical representation of glycolytic flux profile (left) and mitochondrial respiration profile (right) using Seahorse XFe24 Analyzer. The left graph represents extracellular acidification rate (ECAR), which is largely proportion to glycolytic activity of cells and the right graph represents oxygen consumption rate (OCR), which is an indicator for mitochondrial respiration. The X-axis shows duration of the assay, and the timing of sequential inhibitor additions are indicated by vertical lines. The initial baseline (0 to 36 min) represents normal glycolysis level and respiration level before any metabolic interventions. With Oligomycin (an inhibitor of mitochondrial complex V), oxygen consumption drops drastically whereas glycolytic flux displays a sharp increase followed by a slight decline. The addition of Rotenone/Antimycin A (inhibitors to complex I and III) further shuts down mitochondrial respiration but with minimal influence on glycolysis flux. The final addition of the glycolytic inhibitor 2-Deoxy-D-glucose (2-DG) completely abrogates glycolysis. B) and C) Basal GlycoECAR and basal mitoOCR were quantified from Seahorse assays using retinal punches isolated from mice at different developmental age groups (n = 13–35 retinal punches per age group from 4 mice). Data are presented to compare the metabolic activity between age groups, with glycoECAR reflecting glycolytic activity and mitoOCR reflecting mitochondrial respiration. D) ATP production rates calculated from collated MitoOCR and GlycoECAR data from B and C. The number on the bar indicates percentage. E) Ratio of glycolysis derived ATP to mitochondrial derived ATP was calculated from data collated for D. F) Lactate concentration was quantified for different age groups in retina tissue samples using Lactate-Glo Luciferase Assay, n= 4 except for P28, n = 3. Data plotted as mean luminescence values (relative light units – RLU) from each replicate per age group. G) Lactate Dehydrogenase (LDH) Assay Kit (ab102526, Abcam) was used to measure LDH activity in the mouse retina from different age groups (P6, P10, P28; n=4 per age group). H) Pictorial representation of ATP production during retinal development, showing enhanced glycolysis derived ATP production (represented as red rectangle), however proportional percentage contribution from mitochondrial derived ATP remains largely constant with age (represented as green rectangle). All data in the figures are presented as mean ± SEM. Statistical significance is indicated by p-values displayed above the bars. One-way ANOVA with Tukey’s post hoc comparison was used to determine the significance.

Figure 2:. Histone lactylation in the retina

A) Graphical illustration of non-metabolic role of high lactate production in retinal photoreceptors cells because of Warburg’s effect. The lactate produced can act as chemical tag “La” for histone which further modulates gene transcription in retina. B) Illustration of histone lysine (K) lactylation (La) sites identified by mass spectrometry in derivatized tryptic-digested peptides from adult (P28) mouse retina. The identified lactylation sites on lysine residues are highlighted in red and marked by yellow circle icons with “La” written inside. See also Figure S1A for annotated HCD Spectrum and Table S1 for a complete list of identified lactylation sites, represented as carboxyethyl in the modification list. C) Immunoblot-based quantification of histone KLa in purified histone samples at different ages of mice for Pan-lactyl (n = 3), H3K18La (n=4) and H4K12La (n=3) with the experiment repeated twice. ImageStudio software was used for quantification. Statistical significance was analyzed using one-way ANOVA with post hoc Tukey’s test. Immunoblots for Pan-lactyl, H3K18La, and H4K12La show molecular weights of 15 kDa. All data are presented as mean ± SEM. *, p<0.05. D) Immunohistochemical analysis of Pan-KLa or H3K18La in retina tissue samples from different age groups of mice. Inset represents a zoomed-out view with DAPI staining in the ONL on the left. Abbreviations: OS, outer segment; IS, inner segment; ONL, outer nuclear layer; INL, inner nuclear layer; GCL, ganglion cell layer. Scale bar = 50 μm.

Figure 3:. Glycolysis regulates histone lactylation in the retina

(A) Representation of the retinal explant workflow. Key steps in the experiment are shown. The isolated retina was cultured in explant media for 48 hours before proceeding for subsequent experiments as outlined in the figure. (B, D) Extracellular lactate levels using bioluminescence assay and (C, E) histone lactylation (KLa) levels using immunoblot were measured in retinal explants cultured for 48 hours in different concentrations of glucose (5 mM, 15 mM, and 25 mM) or 2-Deoxy-D-glucose (2-DG) (10 mM and 20 mM) with 5 mM glucose as a baseline control in 2-DG experiments. Extracellular lactate, the amount of lactate secreted into the culture media was quantified, and the amount of lactate was correlated with luminescence levels, expressed in relative light units (RLU) on Y axis as shown. For (C, E) immunoblots, histones were extracted from retinal tissues cultured under the same conditions for 48 hours. Pan anti-KLa, H3K18La, and H3K27Ac antibodies detected protein bands at approximately 15 kDa. Data are presented as mean ± SEM (n = 3 biological replicates, *, p<0.05; **, p<0.01). Statistical significance was determined using one-way ANOVA followed by Sidak’s multiple comparisons test.

Figure 4:. Genome-wide profiling of H3K18La

(A) The number of consensus peaks passing a 1x10⁻⁶ FDR for each replicate per postnatal timepoint. The red dot indicates the number of genes containing a peak (17,647, 11,511, and 20,087 respectively), whereas the blue dot represents the number of genes containing a peak in the proximal promoter (6,646, 5,627, and 12,822 respectively). (B) Upper panel: Percentage of H3K18La bound consensus peak loci distance relative to TSS. Lower panel: Percentage of H3K18La bound consensus peak loci to gene annotation feature. (C) Heatmap of H3K18La-bound peak signal enrichment for all genes and their flanking 3kb region. (D) Genomic histogram traces of H3K18La at each timepoint for genes involved in glycolysis (Aldoa and Ldha) and the phototransduction cascade in rod photoreceptors (Rho). The histogram traces are group scaled for each gene. The bars under each histogram represent consensus peaks. (E) Colocalization of H3K18La peaks with H3K27Ac bound regions. (F) Confocal immunofluorescent images showing colocalization of H3K27Ac and H3K18La in nuclear periphery of photoceptor cells. (G) Dynamic active histone mark (H3K4me3 and H3K27Ac ) co-occupancy with H3K18La in the proximal promoter of genes containing H3K18La during development. (H) Selection of GO Biological Process gene sets enriched for genes from panel (G) containing all three histone marks at all timepoints or genes that acquired H3K18La loci at P10 or P28. (I) Genomic histogram traces of histone marks during development for representative selected genes found in panel H. H3K18La marks at P4, P10, and P28 (Red), H3K27Ac at P3, P10, P21 (Green), and H3K4me3 at P3, P10, and P21 (Purple). (J) The UpSet plot illustrates the differential H3K18La co-occupancy at promoter regions of protein coding genes between retina and other metabolically distinct tissues . See also Figure S2. Abbreviations: TSS, transcriptional start site; TES, transcriptional end site; GO, gene ontology; GAS, gastrocnemius; MT, post-mitotic end-state myotubes; PIM, post-ischemia macrophages; MB, myoblasts; ADIPO, adipose tissues; BMDM, bone marrow-derived macrophages.

Figure 5:. Differential binding of H3K18La over development

(A) Upper panel: Principal component analysis of quantitative peak binding for all H3K18La samples. Lower panel: Principal component analysis of RNA expression for the same genes having H3K18La peaks. (B) Heatmap of H3K18La quantitative binding for differentially bound peaks in the promoter region of protein coding genes. (C) Left panel: Positive correlation of quantitative peaks of H3K18La found in panel (B) with retina RNA-seq expression . Right panel: Negative correlation of quantitative peaks of H3K18La found in panel (B) with retina RNA-seq expression . (D) Dynamic active histone mark (H3K4me3 and H3K27Ac ) co-occupancy with H3K18La in the positive correlation set of genes from panel (C). (E) Selection of GO Biological Process gene sets enriched for genes from panel (D) containing all three histone marks at all timepoints or genes that acquired H3K18La loci at P10. (F) Left: Genomic histogram traces of histone marks during development for representative selected genes found in panel (E). H3K18La marks at P4, P10, and P28 (Red), H3K27Ac at P3, P10, P21 (Green), and H3K4me3 at P3, P10, and P21 (Purple). Right: Boxplot of quantitative binding and RNA-seq expression and CUT&Tag analysis. Abbreviations: PCA, principal component analysis; PC1, principal component 1; PC2, principal component 2; Pearson correlation coefficient (PCC); FPKM, fragments per kilobase per million reads; CPM; counts per million; GO, gene ontology; BP, biological process

Figure 6:. Genome-wide profiling of H3K18La in retinal explant cultures with different glucose concentration

(A) The number of consensus peaks passing a 1x10⁻⁶ FDR for each replicate per explant glucose concentration and histone mark. The red dot indicates the number of genes containing a peak, whereas the blue dot represents the number of genes containing a peak in the proximal promoter. (B) Upper panel: Percentage of H3K18La and H3K27Ac bound consensus peak loci distance relative to TSS. Lower panel: Percentage of H3K18La bound consensus peak loci to gene annotation feature. (C) Heatmap of H3K18La-bound and H3K27Ac-bound peak signal enrichment relative to distance from TSS/TES for all genes and their flanking 3kb region. (D) Principal component analysis of quantitative peak binding for all H3K18La and H3K27Ac samples. (E) Heatmap of peaks found from time course H3K18La differential binding analysis (Figure 5B) plotted with peak quantitation from explant CUT&Tag analysis of H3K18La and H3K27Ac. (F) Comparison of significantly, differentially expressed genes from RNA-seq explant analysis (25 mM versus 5 mM) with explant CUT&Tag differential binding results. (G) Genomic histogram traces of explant histone marks for representative genes of several affected pathways observed. The histogram traces are group normalized for each gene. The bars under each gene track represent consensus peaks. Abbreviations: TSS, transcription start site; TES, transcription end site; PCA, principal component analysis; PC1, principal component 1; PC2, principal component 2; CPM, counts per million; DB, differential bound; FPKM, fragments per kilobase per million reads; logFC, log fold-change.

Figure 7:. H3K18La promoters are enriched for GC rich binding motifs.

(A) Number of enriched accessible motifs in H3K18La enhancers, promoters and gene bodies, at P4, P10 and P28. (B) Motif rank vs. p-value for motifs found at enhancers, promoters or gene bodies. (C) Top 5 accessible motifs enriched at H3K18La promoters at P28. (D-F) Examples of genes with H3K18La promoters containing accessible enriched motifs. Tracks represent H3K18La signal for P4, P10 and P28, H3K18La peaks (black) at either enhancers (D), promoters (E) or gene bodies (F) and accessible footprints containing enriched motifs at the given peaks (red) at P4, P10 and P28, and a selection of Refseq genes. One accessible footprint can contain multiple enriched motifs. (G) Number of enriched accessible motifs in H3K18La or H3K27Ac enhancers, promoters and gene bodies, for 5 or 25 mM glucose in explants. (H) Motif rank vs. p-value for motifs found at H3K18La or H3K27Ac enhancers, promoters or gene bodies for 5 or 25 mM glucose in explants. See also Figure S3 and S4