METTL3-mediated chromatin contacts promote stress granule phase separation through metabolic reprogramming during senescence | Nature Communications

Nature Communications volume 15, Article number: 5410 (2024) Cite this article

4096 Accesses

1 Citations

28 Altmetric

Metrics details

METTL3 is the catalytic subunit of the methyltransferase complex, which mediates m6A modification to regulate gene expression. In addition, METTL3 regulates transcription in an enzymatic activity-independent manner by driving changes in high-order chromatin structure. However, how these functions of the methyltransferase complex are coordinated remains unknown. Here we show that the methyltransferase complex coordinates its enzymatic activity-dependent and independent functions to regulate cellular senescence, a state of stable cell growth arrest. Specifically, METTL3-mediated chromatin loops induce Hexokinase 2 expression through the three-dimensional chromatin organization during senescence. Elevated Hexokinase 2 expression subsequently promotes liquid-liquid phase separation, manifesting as stress granule phase separation, by driving metabolic reprogramming. This correlates with an impairment of translation of cell-cycle related mRNAs harboring polymethylated m6A sites. In summary, our results report a coordination of m6A-dependent and -independent function of the methyltransferase complex in regulating senescence through phase separation driven by metabolic reprogramming.

Cellular senescence is a stress response that is characterized by a stable cell growth arrest1. Senescence can be induced by triggers such as activation of certain oncogenes like RAS1. The stable growth arrest of senescent cells is accompanied by downregulation of cell cycle-promoting genes2. However senescent cells are metabolically active as exemplified by secretion of chemokines and cytokines, known as the senescence-associated secretion phenotype (SASP)3,4. These coordinated senescent alterations are essential for maintaining cellular homeostasis and facilitating physiological functions5.

The catalytic methyltransferase-like 3 (METTL3) and the associated METTL14 are the core subunits of the methyltransferase complex (MTC) that mediates the RNA N6-methyladenosine (m6A) modification6. m6A modification regulates gene expression in several different ways, including facilitating a liquid–liquid phase separation (LLPS)7,8,9. In addition, MTC regulates gene transcription in a m6A-independent manner through three-dimensional (3D) chromatin organization10. However, how these functions of MTC are coordinated at the molecular level remain unknown.

3D chromatin organization such as chromatin loops plays an important role in gene transcription11. For example, studies have demonstrated that physical convergence of enhancer and promoter activates gene transcription12,13,14. Interestingly, a genome-wide redistribution of the MTC promotes the expression of SASP genes10. Specifically, METTL3 is predominantly redistributed to pre-existing NFκB sites within the promoters of SASP genes. Although METTL3’s localization to the enhancers is also increased, METTL14 is preferentially redistributed to these sites. Nevertheless, direct evidence of whether MTC affects 3D chromatin organization and its functional impact during senescence remains poorly understood.

Here we show that the MTC complex coordinates the senescence phenotypes by integrating its enzymatic activity-dependent and -independent functions. Specifically, METTL3-mediated chromatin loops induce Hexokinase 2 (HK2) expression independently of m6A modification, subsequently promoting phase separation in the form of stress granules during senescence. This phase separation inhibits the translation of cell-cycle related polymethylated m6A mRNAs, which contributes to senescence-associated stable growth arrest by preventing the death of senescent cells.

To map genome-wide changes in chromatin interaction during senescence, we conducted H3K27Ac HiChIP analysis in control and senescent primary human lung fibroblasts IMR90 cells induced by an inducible oncogenic RAS (Fig. 1a and Supplementary Fig. 1a, b)15. The analysis revealed an overall decrease in the number of intra-chromosomal interactions in senescent compared with control cells (Supplementary Fig. 1c). To determine the potential role of MTC in the observed changes, we knocked down METTL3 or METTL14 in senescent cells (Supplementary Fig. 1d). Notably, knockdown of either METTL3 or METTL14 increased the number of interactions determined by H3K27Ac HiChIP analysis (Supplementary Fig. 1d). This suggests that MTC contributes to the observed decrease in chromatin contacts at a global level during senescence. We also classified the chromatin contacts into short-distance (<5Kb), medium-distance (5kb-2Mb), and long-distance (>2 Mb) categories. We found a decline in medium-distance interactions in senescence, which was largely rescued by knockdown of either METTL3 or METTL14 (Supplementary Fig. 1e and Supplementary Table 1).

a Schematic diagram showing genomic approaches employed: HiChIP-seq for monitor chromatin looping and fastGRO-seq for tracking nascent RNA transcription. b Pearson correlation between differences in chromatin contact numbers and alterations in nascent transcription in senescent cells relative to proliferating cells. n = 2 biologically independent fastGRO-seq experiments. The bottom and top edges of the box plot respectively represent the 25th and 75th percentiles, while the center refers to the median values of fold changes in gene expression. The whiskers extend to the minimum and maximum values within 1st and 99th percentiles range, respectively. c Heatmap showing profiles of nascent RNA transcription in proliferating and senescent cells with or without knockdown (KD) of METTL3 or METTL14. d Venn diagram showing overlapping of genes under indicated conditions. e KEGG pathway enrichment analysis showing enriched 40 genes depicted in (d) with pathways associated with SASP and metabolic process. f Mean density profiles of fastGRO reads for metabolic genes under indicated conditions. g Pearson correlation coefficient of 10 MTC-regulated genes exhibiting enhanced chromatin contacts was calculated between changes in chromatin contacts and differences in nascent transcription. P value was calculated using two-sided Pearson r analysis.

We next determined whether alterations in chromatin contacts correlated with changes in transcription during senescence. Toward this goal, we performed fast global run-on sequencing (fastGRO-seq) to profile global nascent RNA transcription in proliferating and senescent cells (Fig. 1a)16. Indeed, there is a positive correlation between changes in chromatin contacts and gene transcription during senescence (Fig. 1b). This is particularly evident for those upregulated genes with increased chromatin contacts (Fig. 1b). Accordingly, we focused on the genes that are upregulated during senescence but downregulated upon knockdown of METTL3 or METTL14 (Fig. 1c, d). The analysis identified a list of 38 genes (Supplementary Table 2). We next subjected these genes to KEGG pathway enrichment. Consistent with the known role of MTC in regulating SASP10, the analysis showed the enrichment of crucial pathways associated with the SASP, including NFκB signaling, chemokine signaling, and cytokine-cytokine receptor interactions (Fig. 1e and Supplementary Fig. 1f), which validated the experimental strategy we used in the present study.

Interestingly, the analysis revealed an enrichment of metabolic pathway, exemplified by HK2 expression (Fig. 1f and Supplementary Fig. 1g). Indeed, HK2 is upregulated in senescent compared with control cells and the observed upregulation was rescued by knockdown of METTL3 or METTL14 (Supplementary Fig. 1d). This was not a consequence of overcoming senescence because markers of senescence such as expression of p21 and markers of cell cycle exit such as downregulation of cyclin A were not affected by knockdown of METTL3 or METTL14 (Supplementary Fig. 1d). As a negative control, knockdown of METTL3 or METTL14 did not affect the expression of 40 highest expressed genes in senescent cells based on fastGRO-seq analysis (Supplementary Fig. 1h), highlighting the specific regulation of the SASP and metabolic genes by the MTC during senescence.

To identify those MTC-regulated genes that are mediated by chromatin contacts, we cross-referenced the 38 genes with H3K27Ac HiChIP dataset. The analysis revealed a list of ten genes with enhanced chromatin contacts in senescent cells whose upregulation was METTL3/14 dependent (Fig. 1g). The list includes five SASP genes and five metabolic genes including HK2. This is consistent with previous reports that senescence is associated with metabolic reprogramming and enhanced energetic needs17,18. Strikingly, we observed a significant positive correlation between the observed increase in these chromatin contacts and fold changes in gene expression (Fig. 1g). Together, these data support that an increase in chromatin contacts promotes expression of MTC-regulated SASP and metabolic genes such as HK2.

HK2 mediates the rate-limiting step of glucose metabolism19. Despite HK2’s functional importance in senescence17,18, its regulation in senescence remains poorly understood. To unravel the regulatory mechanism by which the MTC promotes HK2 expression, we first cross-referenced publicly available H3K27Ac ChIP-seq, METTL3, and METTL14 Cut&Run-seq10,20. The analysis revealed an increased intensity of bound peaks along the HK2 genomic locus for H3K27Ac, METTL3, and METTL14 at either promoter or putative distal enhancer sites in senescent compared with control cells (Fig. 2a), suggesting that the observed upregulation occurs at the transcription levels mediated by enhancer-promoter interaction. To validate the observed HK2 upregulation occurs at the transcription level, we performed the KAS-seq, a chemical-based genetic approach that captures transcriptional activation with high sensitivity21. When integrating KAS-seq with fastGRO-seq and pol II ChIP-seq, the peak signals were consistently increased at the HK2 locus in senescent compared with control cells, which was decreased by knockdown of METTL3 or METTL14 (Fig. 2b). This result further supports that MTC is required for transcriptionally activation of HK2.

a Tracks of publicly available H3K27Ac ChIP-seq (GSE74328), METTL3, and METTL14 Cut&Run-seq (GSE141992) showing peaks along HK2 genomic region in RAS-induced senescent cells and proliferating control cells. Arrow indicates the transcription start site (TSS) of HK2. b Tracks of fastGRO-seq, KAS-seq, and publicly available Pol II ChIP-seq (GSE141992) showing peaks surrounding HK2 genomic locus in proliferating and senescent cells with vector, METTL3, or METTL14 KD. Senescence was induced by 4-hydroxy-tamoxifen (4-OHT) in ER: RAS-expressing IMR90 cells. Arrow indicates the TSS of HK2. c Chromatin loops surrounding the HK2 locus indicated by H3K27ac HiChIP in proliferating and senescent cells with either vector, METTL3, or METTL14 KD. Tracks of publicly available ChIP-seq of H3K27Ac (GSE74328) in proliferating and senescent cells displayed peaks aligned the loop location as shown in H3K27Ac HiChIP. d Chromatin loops at the HK2 locus revealed by METTL3 HiChIP in proliferating and senescent cells, aligned with tracks of publicly available METTL3 Cut&Run-seq (GSE141992). Arrow indicates the loop from METTL3 HiChIP coinciding with the loop shown in (c). e 3C-qPCR analysis of the ligation frequency on the indicated HK2 gene loci in proliferating and senescent cells with either vector, METTL3 or METTL14 KD. 3C-qPCR primers were designed as illustrated in Fig. 2c with one forward primer anchoring at the promoter of HK2 and the reverse primers targeting the downstream enhancer region of HK2 according to the ChIP-seq peaks for H3K27Ac in control and senescent cells. NPC, non-peak control. Data represent the mean from n = 2 biologically independent experiments. Source data are provided as a Source Data file.

We next determined whether chromatin loops play a role in the observed upregulation of HK2. Toward this goal, we analyzed significant chromatin loops within −200 kb upstream and +100 kb downstream of HK2 gene body using H3K27Ac HiChIP. Two strong loops were newly formed in senescent cells spanning the HK2 genomic locus (Fig. 2c). Notably, knockdown of METTL3 or METTL14 restored the loop distribution as observed in control cells (Fig. 2c). This suggests that MTC may play a direct role in the formation of observed chromatin loops in senescent cells. To test this possibility, we conducted METTL3 HiChIP-seq. Similar to H3K27Ac HiChIP, we observed a reduction in intra-chromosomal loops during senescence (Supplementary Fig. 2a). Likewise, we noted a decrease in the percentage of medium-distance loops and an increase in the percentage of long-distance loops (Supplementary Fig. 2b and Supplementary Table 3). A detailed examination of loop formation based on METTL3 HiChIP analysis demonstrated an identical loop spanning HK2 genomic locus as shown in H3K27Ac HiChIP that was newly formed in senescent cells (Fig. 2c, d). We have previously shown that METTL3 regulates SASP gene expression by localizing to the pre-existing NF- κB genomic sites through the interaction with p65, which is known as a transcription factor that directly binds to DNA10. To explore the mechanism of how METTL3 affects high-order chromatin structure, we cross-referenced previously published p65 Cut&Run -seq and METTL3 Cut&Run-seq along the HK2 genomic locus and revealed an increased p65 bound peaks at putative distal enhancer in senescent compared to control cells, indicating that METTL3 may regulate high-order chromatin structure through its association with p65 (Supplementary Fig. 2c). In addition, to explore whether chromatin RNA methylation was involved in the loop formation of HK2, we analyzed chromosome-associated regulatory RNAs (carRNAs) in control and senescent cells with or without METTL3 and METTL14 knockdown using our previously published carRNA m6A-seq10. We showed that HK2 was not subject to the regulation of chromatin RNA methylation at putative distal enhancer anchors (Supplementary Fig. 2d).

Functional topologically associated domains (TADs), which govern chromatin interactions within specific regions, have been shown to spatially regulate gene expression including those mediated by chromatin loops22. By leveraging previously published chromosome conformation capture (Hi-C) studies, we detected a gain of two sub-TAD domains surrounding HK2 genomic locus with enhanced interaction frequencies in senescent compared with control cells (Supplementary Fig. 2e)23. The location of acquired sub-TADs aligns consistently with the loops formed in the H3K27Ac HiChIP, indicating enhanced interactions in restricted chromatin domains around HK2.

To directly examine chromatin looping for HK2 gene locus guided by the common new loop formed in senescent cells based on H3K27Ac and METTL3 HiChIP, we performed chromosome conformation capture quantitative PCR (3C-qPCR) analysis in control and senescent cells24. We observed robust interactions between the anchored transcription starting sites (TSS) and the putative enhancer of HK2 in senescent cells (Fig. 2e). Notably, knockdown of METTL3 or METTL14 reduced the association to the levels observed in control cells (Fig. 2e). In addition, restoration of enzymatic dead mutant METTL3 was sufficient to rescue HK2 expression, further supporting the notion that METTL3 regulates HK2 in a m6A independent manner (Supplementary Fig. 2f). Indeed, we found no significant change in m6A levels within the HK2 genomic locus (Supplementary Fig. 2g). Consistently, treatment of senescent cells with STM2457, a highly selective inhibitor of METTL3 methyltransferase activity25, did not decrease HK2 expression (Supplementary Fig. 2h, i). Moreover, knock down of YTHDC1, a nuclear reader of m6A, did not affect HK2 expression26 (Supplementary Fig. 2j, k). Together, these findings suggest that HK2 is transcriptionally activated independently of m6A modification, which correlates with METTL3-mediated chromatin looping at the HK2 locus in senescent cells.

Senescent cells accumulated in aged tissues and contributed to tissue aging27. Thus, we sought to examine whether Hk2 is regulated during tissue aging. Toward this goal, we employed tissue transposase-accessible chromatin with sequencing (ATAC-seq) to evaluate chromatin accessibility in liver tissues from young and aged mice (Fig. 3a). The analysis revealed that Hk2 exhibited a more open chromatin status in the liver of aged compared with young mice (Fig. 3a). And a detailed examination of the ATAC-seq peaks demonstrated enhanced peak signals specifically at the promoter region of Hk2 (Fig. 3b). Consistently, hepatocytes RNA-seq confirmed an elevated expression level of Hk2 in aged compared to young mice, which was further validated by RT-qPCR (Fig. 3c and Supplementary Fig. 3a). Likewise, the protein level of Hk2 was upregulated in aged compared to young mice in both whole liver tissues and isolated hepatocytes (Fig. 3d–g). In addition, the expression level of Mettl3 and Mettl14 remains no change in aged mice hepatocytes (Supplementary Fig. 3b). Finally, mining the Aging Atlas database confirmed HK2 upregulation in multiple other aged tissues (Supplementary Fig. 3c)28. Together, we conclude that HK2 is upregulated during tissue aging.

a Heatmaps showing quantitative analysis of ATAC-seq data in young and aged whole liver tissues with a false discovery rate (FDR) cutoff of <0.1. Red represents higher expression, while blue represents lower expression. Data represent two technical replicates of two biological replicates. b Tracks displaying ATAC-seq peak signals in aged liver tissues compared to the young. Arrows indicate the transcription start site of Hk2. c RT-qPCR analysis of Hk2 expression in young (n = 5) and aged (n = 5) hepatocytes. Data represent mean ± SD using a two-tailed Mann Whitney test. d Western blot analysis showing the expression level of Hk2 in young (n = 5) and aged (n = 5) whole liver tissues. Ponceau S staining was used as the internal control. e The band intensity of Hk2 compared to Ponceau S, as in (d), was quantified. Data represent mean ± SD using a two-tailed Mann Whitney test. f Western blot analysis showing the expression level of Hk2 in isolated hepatocytes from young (n = 5) and aged (n = 4) mice. Ponceau S staining was used as the internal control. g The band intensity of Hk2 compared to Ponceau S, as in (f), was quantified. Data represent mean ± SD values using a two-tailed Mann-Whitney test. Source data are provided as a Source Data file.

The two HK2 catalyzing domains convert glucose to Glucose-6-phosphate (G-6-P) that is a rate-limiting manner in glycolysis29,30. Point mutations at both N-terminal and C-terminal catalytic domains (D209A and D657A) of HK2 effectively diminish the enzymatic activity of HK229,31. To explore the metabolic consequences of HK2 upregulation in senescence, we conducted 13C glucose isotope tracing in control and senescent cells with or without HK2 knockdown, and then rescued with ectopic expression of either wildtype or enzymatic activity inactive mutant HK2 (Fig. 4a). Notably, metabolites involved in the glycolysis pathway, such as G-6-P, and Fructose 1,6-bisphosphate, displayed high labeling rates across each group (Supplementary Fig. 4a). Interestingly, the central intermediate metabolite Fructose 1,6-bisphosphate in the glycolysis pathway did not exhibit reduced labeling (M + 6) by HK2 knockdown (Supplementary Fig. 4a). This could be due to the compensatory increase in the isoenzyme HK1 level in response to HK2 KD (Supplementary Fig. 4b).

a Five groups were analyzed by 13C glucose isotope tracing. 1, control proliferating cells. 2, RAS-induced senescent cells via 4-OHT with control shRNA (shControl). 3, senescent cells with shRNA targeting HK2 (shHK2). 4, senescent cells with shRNA targeting HK2 (shHK2) and rescued with wildtype Flag-HK2. 5, senescent cells with shRNA targeting HK2 (shHK2) and rescued with mutant Flag-HK2. Note that the HK2-overexpressing plasmid is resistant to shHK2 due to silent mutation. IMR90 cells with or without 4-OHT induction of RAS expressing shControl or shHK2 with or without rescue by ectopic expression of wildtype or catalytic-mutant HK2 (D209AD657A) were analyzed for expression of the indicated proteins by western blot on Day 9. b Distribution of 13C isotopologues of Inosine monophosphate (IMP), Adenosine monophosphate (AMP), Adenosine diphosphate (ADP), Adenosine diphosphate (ATP), Adenosine, and Inosine at indicated groups. Data represent mean ± SEM of n = 3 biologically independent experiments. c Boxplots showing the labeled fractions of IMP (M + 5), AMP (M + 5), ADP (M + 5), ATP (M + 5), Adenosine (M + 5), and Inosine (M + 5) under indicated conditions. Data represent mean ± SEM of n = 3 biologically independent experiments. P values were calculated using a two-tailed Student’s t test. d Distribution of 13C isotopologues of Ribose-5-phosphate (R-5-P) at indicated groups. Data represent mean ± SEM of n = 3 biologically independent experiments. P values were calculated using a two-tailed Student’s t test. e Diagram showing labeled metabolites through 13C glucose isotope tracing. Black circles represent 13C atoms. G-6-P Glucose-6-phosphate, R-5-P Ribose-5-phosphate, IMP inosine monophosphate, AMP Adenosine monophosphate, ADP Adenosine diphosphate, ATP Adenosine triphosphate. Source data are provided as a Source Data file.

To unbiasedly identify metabolite products potentially regulated by the enzymatic function of HK2 during senescence, we employed the following criteria: a) showing more extensive labeling in senescent compared to control cells; b) exhibiting decreased labeling in senescent cells by HK2 knockdown; c) rescuing the decrease in labeling by expressing of wildtype HK2 but not by the mutant HK2. This analysis revealed that HK2 plays a major role in regulating metabolites involved in purine metabolism such as inosine monophosphate (IMP), adenosine monophosphate (AMP), adenosine diphosphate (ADP), adenosine triphosphate (ATP), adenosine and inosine (Fig. 4b-c). In addition, our results showed that Ribose-5-phosphate (R-5-P) displayed lower 13C labeling in senescent compared to control cells (Fig. 4d). Consistently, the pentose phosphate pathway is known to provide nucleotide precursors for purine metabolism by branching from G-6-P and producing the intermediate product R-5-P32. Indeed, the level of R-5-P labeling was increased by HK2 knockdown, which was rescued by wildtype HK2, but not the mutant HK2 (Fig. 4d). In addition, the level of ATP and inosine that were reduced by METTL3 knockdown could be rescued by HK2 restoration, suggesting that METTL3’s function in purine metabolism depends on HK2 (Supplementary Fig. 4c–e). Together, our data support a model whereby HK2 promotes purine metabolism and ATP production during senescence in an enzymatic activity-dependent manner (Fig. 4e).

ATP is required for stress granules formation and dynamics33. These membrane-less granules can result from LLPS that is promoted by polymethylated m6A RNAs along with m6A readers such as YTHDF proteins7,8,9,34. MTC is essential for the formation of these endogenous phase-separated compartments, which correlates with m6A modifications35. Inspired by HK2-depenent increase in ATP levels observed during senescence, we proceeded to investigate whether the assembly of phase-separated stress granules occur during senescence. Accordingly, we stained for YTHDF1-3 and TIAR as phase separation and stress granules markers, respectively8. We found that 5–8% of senescent cells undergo phase separation in the form of stress granules as determined by colocalization of YTHDFs and TIAR (Fig. 5a–c). Comparable percentages of phase-separated senescent cells were observed for all three YTHDF family members (Fig. 5c and Supplementary Fig. 5a, b). Supporting the notion that the observed colocalized YTHDF and TIAR indeed represent phase-separation. 1,6-hexanediol, an organic chemical known to reverse phase separation36, significantly decreased the number of senescent cells with the colocalization (Supplementary Fig. 5c, d). As a negative control, no phase separation was observed in RAS-transformed NIH3T3 cells (Supplementary Fig. 5e). This indicates that the observed phase separation resulted from senescence induction instead of RAS expression.

a Immunofluorescence (IF) of control and oncogene (RAS or BRAF)-induced senescent cells with antibodies targeting YTHDF3 and TIAR. 4′,6-diamidino-2-phenylindole (DAPI) counterstaining was used to visualize the nuclei. b Quantification of the percentage of cells with phase separation events at indicated conditions, as in a, was determined (n > 200 cells over three independent experiments). P values were calculated using a two-tailed Student’s t test. Error bars represent mean with SD. c Quantification of the percentage of cells with phase separation events by antibodies targeting control IgG, YTHDF1, YTHDF2, and YTHDF3 in RAS-induced senescent cells was determined (n > 200 cells over three independent experiments). P values were calculated using a two-tailed Student’s t-test. Error bars represent mean with SD. d IF of control and RAS-induced senescent cells with antibodies targeting YTHDF3 and TIAR at indicated conditions. DAPI counterstaining was used to visualize the nuclei. Rectangle represented the area that was shown on the right with 5X magnification, showing cases of phase separation. e Quantification of the percentage of cells at indicated conditions, as in (d), was determined (n > 200 cells over three independent experiments). P values were calculated using a two-tailed Student’s t test. Error bars represent mean with SD. f IF of RAS-induced senescent cells with antibodies targeting YTHDF3 and TIAR at indicated conditions. DAPI counterstaining was used to visualize the nuclei. g Quantification of the percentage of cells with phase separation events at indicated conditions, as in (f), was determined (n > 200 cells over three independent experiments). P values were calculated using a two-tailed Student’s t test. Error bars represent mean with SD. h IF of RAS-induced senescent cells with antibodies targeting YTHDF3 and TIAR when treated with DMSO or METTL3 inhibitor (2.5 μM STM2457) for 48 h. i Quantification of the percentage of cells with phase separation events at indicated conditions, as in (h), was determined (n > 200 cells over three independent experiments). P values were calculated using a two-tailed Student’s t test. Error bars represent mean with SD. Scale bars, 20 μm. Source data are provided as a Source Data file.

We next sought to determine whether the observed phase separation depends on METTL3-regulated HK2 upregulation in senescent cells. We knocked down HK2 or METTL3 in senescent cells. Indeed, knockdown of either HK2 or METTL3 led to a marked reduction in the phase separation observed in senescent cells (Fig. 5d, e). Similar observation was also made using a BRAF-induced senescent model (Supplementary Fig. 6a, b). Additionally, we exposed control proliferating cells to NaAsO2, a known inducer of phase-separated stress granules in nearly 90% of cells8. Consistently, knockdown of HK2 or METTL3 was sufficient to reduce the phase-separated events induced by NaAsO2 (Supplementary Fig. 6c, d). Thus, METTL3-regulated HK2 is required for formation of phase-separated stress granules.

To determine whether the observed phase separation depends on the enzymatic activity of the upregulated HK2 in senescent cells, we compared phase separation events in HK2 knockdown cells restored with either wildtype or an enzymatic activity inactive mutant HK2. Indeed, wildtype HK2 was significantly more efficient in restoring the reduction in stress granule formation induced by HK2 knockdown in senescent cells compared with the mutant HK2 (Fig. 5f, g). Consistently, treatment of senescent cells with 2-Deoxy-D-glucose, a HK2 inhibitor37, effectively reduced stress granules formation in senescent cells (Supplementary Fig. 6e, f). In addition, we showed that there was no colocalization of HK2 with stress granules during senescence (Supplementary Fig. 6g). Human hexokinase isoenzymes include four genes named HK1 to HK419. While HK3 and HK4 expression levels were barely detectable in senescent cells, we asked whether Hexokinase 1 (HK1) inhibition could diminish phase separation in senescent cells. Consistent with the notion that HK2-regulated phase separation is its enzymatic activity dependent, HK1 knockdown also reduced the phase-separated stress granules in senescent cells, albeit to a much lesser degree compared to HK2 knockdown (Supplementary Fig. 6h, i). This could be due to the fact that HK1 exhibits activity with only its C-terminal domain, whereas HK2 contains two catalytically active domains29. Collectively, these findings indicated that HK2 promotes phase separation during senescence in an enzymatic activity-dependent manner.

Since phase separation observed in the form of stress granules depends on MTC activity35, we determined whether inhibition of METTL3 enzymatic activity affects stress granule formation. Indeed, treatment of senescent cells with STM2457, a selective inhibitor of METTL3 enzymatic activity, significantly reduced stress granules formation in senescent cells (Fig. 5h, i). This result indicates that phase separation observed in senescent cells in the form of stress granules depends on m6A modification.

Stress granules impact mRNA fate in various ways, including a reduction in their stability and translation7. mRNAs that contain multiple, but not single, m6A sites have been shown to promote the formation of phase-separated stress granules7,9. To determine the functional consequences of phase separation during senescence, we focused on mRNAs with polymethylated m6A signals in senescent cells (Fig. 6a). The analysis of m6A-seq revealed 5223 unique m6A sites among genomic regions including 5′ untranslated region (5′ UTR), coding sequence, and 3′ UTR in senescent cells (Supplementary Fig. 7a). Notably, 523 mRNAs containing more than three m6A sites were identified, which we defined as polymethylated m6A mRNAs in senescent cells. Since stress granules reduce mRNA stability7, we focused on those polymethylated m6A transcripts that are downregulated in senescent compared with control cells. This analysis revealed a list of 35 polymethylated m6A mRNAs. Pathway analysis showed that genes encoded by these transcripts are significantly enriched for pathways such as cell cycle, cell division, and mitosis (Fig. 6b, c and Supplementary Fig. 7b).

a Venn diagram illustrates the enrichment of RNAs with increased polymethylated m6A signals from publicly available m6A-seq when comparing senescent to control cells (GSE141993), overlapping with downregulated transcripts from publicly available RNA-seq (GSE141991). The highest m6A peak signal among all polymethylated m6A signals was used for comparison. b Pathway enrichment analysis showing enriched 35 genes depicted in (a) with pathways associated with cell cycle, cell division, and mitosis. c, Tracks of publicly available m6A-seq (GSE141993) from control (Cont) and senescent (Sen) cells showing peaks within the genomic locus of CDCA8, CENPE, ERCC6L, SMC2, SMC3, KIF16A and CCAR1, indicating polymethylated m6A signals. The m6A signal was normalized to the corresponding input. d Tracks (upper panel) of publicly available m6A-seq (CRA005942) in primates showing the median m6A peak signals within the SMC2 genomic locus from either the young or aged group. Heatmap (lower panel) revealed the distribution of the m6A peak signals from young (n = 8) and aged (n = 8) liver tissue in primates. The m6A signal was normalized to the corresponding input. e UMAP plot of publicly available scRNA-seq (GSE197017) from mouse skeletal muscle tissue showed clustering of cellular components of control (Non Sen) and senescent cells that were differentiated by CD45 marker (CD45+ Sen and CD45- Sen). UMAP projections were shown by Feature Plots depicting Cxcl1, Hk2, and Cdca8 expression in single cells. f Translational efficiency of mRNAs including CDCA8, CENPE, and SMC2 was detected by polysome profiling. Translational efficiency is normalized to free mRNA level for each gene. Data represent mean ± SD of n = 3 biologically independent experiments. P values were calculated using a two-tailed Student’s t test. g Senescent cells were co-stained by SPiDER SA-β Gal, and the LIVE/DEAD Near-IR dead cell stain kit then analyzed by flow cytometry under indicated conditions. h Quantification of dead senescent cells at indicated conditions, as in (g). Data represent mean ± SD of n = 3 biologically independent experiments. P values were calculated using a two-tailed Student’s t test. Source data are provided as a Source Data file.

To explore the potential physiological relevance of the newly identified polymethylated m6A mRNAs, we analyzed the published m6A-seq from liver or heart tissues of young and aged primates38. Similar polymethylated m6A signals were detected for cell-cycle related genes such as SMC2, SMC3, and CCAR1 (Fig. 6d and Supplementary Fig. 7c, d). In addition, using publicly available single-cell RNA sequencing (scRNA-seq) dataset from mouse skeletal muscle tissue39, we found that the transcriptional abundance of cell-cycle related genes such as Cdca8, Cenpe, and Smc2 exhibit lower levels in CD45- senescent cells with high HK2 expression compared with non-senescent cells (Fig. 6e and Supplementary Fig. 7e). As positive controls, SASP genes including Cxcl1, Mmp2, showed a similar pattern as HK2 expression (Fig. 6e and Supplementary Fig. 7e). As a negative control, we found that HK1 transcriptional level did not correlate with the senescent status (Supplementary Fig. 7e). Together, these results suggest that HK2 upregulation is associated with downregulation of cell cycle-related polymethylated m6A mRNAs.

We next determined the role of HK2 in regulating translation of cell-cycle related polymethylated m6A mRNAs by performing polysome profiling in senescent cells with or without HK2 knockdown. Indeed, HK2 knockdown significantly increased the translational efficiency of polymethylated m6A mRNAs such as CDCA8, CENPE, and SMC2 (Fig. 6f and Supplementary Fig. 7f). Since HK2 upregulation promotes phase-separated stress granules that impair the translation of polymethylated m6A mRNAs7, we further knocked down YTHDFs to disrupt the phase separation in senescent cells (Supplementary Fig. 7g). Consistently, knockdown of YTHDFs increased translational efficiency for polymethylated m6A mRNAs such as SMC2 and CENPE in senescent cells (Supplementary Fig. 7h). As a control, knockdown of YTHDF3 did not increase the translational efficiency for transcripts such as CDCA8, CENPE, and SMC2 before HK2 upregulation (Supplementary Fig. 7i, j). These findings support that HK2 suppresses the translation of cell-cycle related polymethylated m6A mRNAs through phase-separated stress granules.

We next determined the consequence of increased translation efficiency of cell-cycle related mRNAs induced by HK2 knockdown. We found that HK2 knockdown increased the percentage of senescent cells in the G2/M phase (Supplementary Fig. 8a) and induced a concurrent decrease in cell viability (Supplementary Fig. 8b, c). Supporting the notion that the observed effects are due to HK2’s role in regulating phase-separated stress granules, similar observation was also made by disrupting stress granules through knockdown of YTHDFs (Supplementary Fig. 8d, e). To directly determine whether the observed decrease in cell viability was due to the death of senescent cells, we stained these cells with SPiDER-β-Gal to identify senescent cells together with a live/dead cell marker. Indeed, both knockdown of HK2 and YTHDFs increased the percentage of dead senescent cells compared to controls (Fig. 6g, h and Supplementary Fig. 9). Together, these data support a model whereby HK2 contributes to senescent-associated stable cell growth arrest by preventing the death of senescent cells through phase-separated stress granules.

We observed a global decrease in chromatin contacts as determined by both H3K27Ac and METTL3 HiChIP analysis. Knockdown of METTL3 restored the overall decrease in chromatin contacts. However, an increase in enhancer-promoter interaction positively correlates with changes in transcription in senescent cells. This suggests a redistribution instead of a simple decrease of chromatin contacts during senescence. Indeed, in the present studies, we showed that an increase in chromatin contacts correlates with upregulation of SASP and metabolic genes. These findings are consistent with previously reported genome-wide redistribution of MTC during senescence10.

HK2 is functionally important in modulating glycolysis19. Here, we unveiled a direct link between METTL3-mediated chromatin interactions and the upregulation of HK2 during senescence. HK2 upregulation was also observed in aged hepatocytes and other aged tissues. Mechanistically, HK2 metabolically increases purine metabolism during senescence, which is consistent with previously published evidence that senescent cells have higher HK2 activity, exhibit enhanced glycolysis and produce more ATP17,18. This finding also aligns with previous observations of elevated purine metabolism in aged drosophila40. Notably, senescent cells are not dividing. This suggests that the observed increase in purine metabolism and the associated production of ATP may play a role in maintaining the senescent status. Consistent with this notion, inhibition of HK2 triggers the death of senescent cells. This finding indicates that HK2-regulated purine metabolism could be a target for developing senolytics to eliminate senescent cells. In addition, ATP plays a key role in regulating the formation and dynamics of stress granules33,41. Indeed, we showed that the enzymatic activity of HK2 is required for formation of stress granules in senescent cells. Similar to HK2 inhibition, disruption of stress granules genetically by knocking down YTHDFs induces the death of senescent cells. Notably, the regulation of polymethylated m6A mRNAs by phase-separated stress granules only occurs in senescent but not proliferating cells. This is because despite the presence of polymethylated m6A mRNA, there is no stress granules in proliferating cells without senescence-associated metabolic reprogramming. Collectively, these findings support the key role of HK2-dependent metabolic reprogramming in preventing the death of senescent cells through inducing phase-separated stress granules.

We observed 5–8% phase-separated stress granules in senescent cells. There are several possibilities that may account for this relatively low percentage. For example, this could be due to the dynamic nature of stress granules assembly and the inherent heterogeneity among senescent cells1,42. In addition, it is possible that a higher frequency of these phase-separated events could be observed using more sensitive techniques8,33. However, the mechanism we discovered in the present study may be broadly applicable. For example, we showed that METTL3 and HK2 are necessary for formation of stress granules that are chemically induced in ~90% of cells. Regardless, these findings advanced our conceptual understanding by which senescent cells leverage phase-separated stress granules to counteract stress via metabolic reprogramming.

We showed that MTC transcriptionally upregulates HK2, which correlated with chromatin looping during senescence. Likewise, MTC transcriptionally promotes the SASP genes in an enzymatic activity-independent manner10. In addition, HK2 prevents the death of senescent cells by inducing the formation of phase-separated stress granules in an m6A-dependent manner. Thus, MTC coordinates two key phenotypes of senescence, namely the senescence-associated growth arrest and the SASP, by integrating its enzymatic activity -dependent and -independent functions. These results are consistent with the notion that metabolic needs and SASP are intimately linked17 and the MTC may represent one layer of the integration between these two phenotypes of senescent cells. In summary, our findings show that senescent cells employ phase-separated stress granules via METTL3-mediated HK2 upregulation to govern the senescence-associated stable growth arrest by preventing the death of senescent cells. This mechanism attributes to the coordination of both m6A-independent and -dependent functions of METTL3 during senescence.

All animal procedures were performed by protocols and guidelines approved by the Institutional Animal Care and Use Committee (IACUC) at the Sanford Burnham Prebys (SBP) Medical Discovery Institute.

IMR90 primary human diploid lung embryonic fibroblasts (ATCC, CCL-186) were cultured in DMEM medium supplemented with 10% fetal bovine serum (FBS), l-glutamine, sodium pyruvate, non-essential amino acids, sodium bicarbonate and 1% penicillin–streptomycin under low oxygen tension (2%). For the experiments, only IMR90 cells that had undergone 25–36 population doublings were used unless otherwise stated. Viral packing cells 293T (ATCC, CRL-3216) or Phoenix (a kind gift from Dr. Gary Nolan, Stanford University) was cultured in DMEM medium supplemented with 10% FBS and 1% penicillin–streptomycin. Mouse NIH 3T3 fibroblasts (a kind gift from Dr. Bin Tian of The Wistar Institute) were cultured in DMEM medium supplemented with 10% calf serum, sodium pyruvate, and 1% penicillin–streptomycin. Cell lines were authenticated at The Wistar Institute’s Genomics Facility using short-tandem-repeat DNA profiling. Regular mycoplasma testing was performed using PCR detection as previously described.

Induction of senescence by oncogenic BRAFV600E was carried out using the previously described protocol43. For ER: RAS-induced senescence, IMR90 cells were infected with retrovirus containing a 4-hydroxytamoxifen (4-OHT)-inducible ER: RAS construct (pLNC-ER: Ras)44. After 2 weeks of selection with G418 (400 μg ml−1, Gibco), cells were maintained with a lower dose of G418 (200 μg ml−1) and treated with 4-OHT at a final concentration of 100 nM. Cells were then harvested on Day 6 to examine the expression of RAS and other senescence markers.

METTL3 inhibitor STM2457 was purchased from MCE (HY-134836), Cycloheximide was purchased from Thermofisher (J66901.03), and NAC was ordered from Sigma (A9165).

The pBABE-puro- BRAFV600E and pBABE-puro-Empty plasmids were obtained from Addgene (#15269 and #1764). The template for HK2-3XFlag was a gift from Jason Cantor (Addgene plasmid # 163451). A PCR reaction was performed using primers that contained complementary sequences to the BamHI/EcoRI restriction enzyme sites, and the resulting PCR products were purified using a Zymo DNA purification kit (D4013). The purified PCR products were then ligated into a BamHI/EcoRI-digested PCDH-CMV vector (Addgene #72265) using the ClonExpress Ultra One Step Cloning Kit (Collagen C115-01) to obtain PCDH-CMV-HK2-3XFlag according to the manufacturer’s protocols. To generate the HK2-expressing plasmid that is resistant to HK2 KD, Mut Express II Fast Mutagenesis Kit V2 (C214) was used according to the manufacturer’s instructions. Site-specific mutation of HK2 enzymatic pocket (D209AD657A) was generated using the same kit. All plasmid construction and mutation sites were validated by Sanger sequencing. The following antibodies were purchased from the indicated suppliers. For HiChIP: anti-H3K27Ac (Abcam #4729); anti-METTL3 (Abcam #195352). For immunofluorescence, anti-YTHDF1 (Proteintech, #17479-1-AP; 1:150); anti-YTHDF2 (Abcam, #245129; 1:150); anti-YTHDF3 (Proteintech, #25537-1-AP; 1:150); anti-TIAR (BD biosciences, #610352; 1:150); Alexa Fluor 488 conjugated Goat anti-Rabbit IgG (H + L) Cross-Adsorbed Secondary Antibody (Thermo, A-11008; 1:200);, Alexa Fluor 568 conjugated Goat anti-Mouse IgG (H + L) Cross-Adsorbed Secondary Antibody (Thermo, A11004; 1:200). For western blots: anti-RAS (Becton Dickinson, # 610001; 1:1000); anti-p21 (Abcam, # 7960; 1:1000); anti-β-actin (CST #4967; 1:5000); anti-METTL3 (Abcam # 195352; 1:1000); anti-METTL14 (Sigma, #HPA038002; 1:1000); anti-HK2 (abcam, #209847; 1:1000), and anti-Flag (Sigma, #F1804; 1:2000).

Retrovirus was packaged using Phoenix cells as previously described10. Lentivirus particles were generated for genetic knockdown, over-expression, and rescue experiments. Briefly, HEK293T cells were co-transfected with lentivector-containing plasmids, including pLKO.shscramble (Addgene, #1864), pLKO.shMETTL3 (TRCN0000034715), pLKO.shMETTL14 (TRCN0000015936), pLKO.HK2-1 (TRCN0000037670), pLKO.HK2-2 (TRCN0000037671), pLKO.YTHDF1 (TRCN0000062768), pLKO.YTHDF2 (TRCN0000167813), pLKO.YTHDF3 (TRCN0000370250), pLKO.HK1-1 (TRCN0000037655), pLKO.HK1-2 (TRCN0000037657), shYTHDC1-1 (TRCN0000243987), shYTHDC1-2 (TRCN0000243989) along with the packaging plasmids pMD2.G and psPAX2 (Addgene #12260 and Addgene #12259) using the cationic polymer polyethylenimine (PEI) at a ratio of 1:3. After 48 and 72 h of transfection, the lentiviral particles were harvested from the culture supernatant by filtering through a 0.45 µm filter (Corning, 431220). The harvested lentivirus was aliquoted and stored at −80 °C until further use. Senescent IMR90 cells were infected with shRNA-encoding lentivirus with 8 μg/ml polybrene (Sigma, #28728-55-4). 2 μg/ml Puromycin selection was used to select knockdown cell lines for 48–72 h after transduction with lentiviral particles.

HiChIP was performed mainly according to an established procedure15. Briefly, 15 million of IMR90 cells were fixed in freshly made 1% formaldehyde at room temperature for 10 min followed by quenching with 125 mM glycine. Samples were lysed in ice-cold Hi-C lysis (10 mM Tris, pH 8, 10 mM NaCl, 0.2% IGEPAL CA-630 with protease inhibitors) buffer and digested with MboI (NEB, R1047) for 2 h at 37 °C. After 1 h of biotin dATP (Thermo, 19524016) incorporation, samples were ligated with T4 DNA ligase (NEB, M0202) at room temperature and ChIPped with H3K27Ac antibody (abcam, ab4729) or METTL3 antibody (abcam, ab195352) after sonication. IP samples were reversed-crossed, pulled down with streptavidin C1 beads (Invitrogen, 65001) and went through tagmentation by transposaseTn5 (Illumina, 20034198). Samples were amplified with certain cycle numbers according to qPCR with 5-cycle pre-amplified library. Libraries were purified through two-sided size selection with Ampure XP beads (Beckman Coulter, A63880) and were sequenced with 75-base pair-end reads on the Illumina NextSeq 500 platform. Raw fastq files were demultiplexed and HiC-Pro 3.1.0 was used to align the reads to the hg19 human genome as previously described. Reads were assigned to MboI restriction site and duplicated reads were removed. The following parameters were set: BOWTIE2_GLOBAL_OPTIONS = –very-sensitive, –end-to-end, –reorder, BOWTIE2_LOCAL_OPTIONS = –very-sensitive, –end-to-end, –reorder, GENOME_FRAGMENT = hg19_MboI.bed, LIGATION_SITE = GATCGATC. All other parameters were set as default. hichipper 0.7.7 was used for loop calling as previously described45. Library quality was determined, and chromatin loops were called using a restriction fragment-aware approach based on pre-determined narrow peaks. The narrow peak file of H3K27Ac ChIP-seq and METTL3 Cut&Run-seq were obtained from Gene Expression Omnibus (GEO; accession number GSE74328 and GSE141992). Following parameters were used: –peak-pad 700 –make-ucsc. All other parameters were set as default. Loops that passed FDR < 0.01 with at least 4 reads and FDR < 0.05 were considered significant for H3K27Ac HiChIP and METTL3 HiChIP, respectively.

For loop mapping to promoters and enhancers, promoters were defined as 1 kb upstream and downstream from TSS. All significant HiChIP interactions overlapped with promoters for either side were defined as Enhancer-Promoter interaction. For correlation analysis of gene expression with looping, the total number of significant HiChIP interactions in growing (EP#(growing)) and senescent IMR90 cells (EP#(senescent)) were counted for each gene. Log fold-change of the fastGRO-seq scores were compared with the difference of significant interactions (EP#(growing) - EP#(senescent)). To avoid the effect of outliers, 70 genes were randomly selected from the same EP#(growing) - EP#(senescent) group 200 times and the average log fold-change was plotted.

The Hi-C data for both growing and senescent IMR90 cells were downloaded from GSE118494 and processed as previously described23. Briefly, the paired reads were individually mapped to the hg19 human genome using Bowtie2 (version 2.2.9) with the iterative alignment approach. The aligned reads were assigned to MboI fragments. Redundant paired reads derived from PCR biases, aligned to repetitive sequences, or with low mapping quality (MapQ <30) were filter out. Additionally, paired reads originated from self-ligation and undigested products were also eliminated. The human genome was divided into non-overlapped 10-kb bins. Raw contact matrices were constructed by counting paired reads assigned to each pair of bins, and Hi-C biases in the contact matrices were corrected using the Iterative Correction and Eigenvector decomposition (ICE) method46. The ICE normalization was repeated 30 times. The ICE-normalized contact scores were used for heat maps, where the maximum intensity corresponded to the top 5% score.

fastGRO-seq was performed mainly based on the published procedure16. Briefly, 20 million IMR90 cells at each condition were washed with ice-cold PBS and incubated with cold swelling buffer (10 mM Tris, pH 7.5, 2 mM MgCl2, and 3 mM CaCl2 with 2 U/ml Superase-in) for 5 min on ice. Cells were washed with GSB buffer (swelling buffer + 10% glycerol with 2 U/ml Superase-in) and lysed in lysis buffer (10 mM Tris pH 7.5, 2 mM MgCl2, 3 mM CaCl2, 10% glycerol, 1%l Igepal (NP-40), 2 U/ml Superase-in). Isolated nuclei were washed twice with cold lysis buffer and were frozen in freezing buffer (50 mMTris-HCL pH 8.0, 40% glycerol, 5 mM MgCl2, 0.1 mM EDTA) at −80 °C. Drosophila spike-in nuclei were added to thawed nuclei before use. Freshly prepared and pre-warmed 2X Nuclear run-on buffer (10 mM Tris-HCl pH 8, 5 mM MgCl2, 300 mM KCl, 1 mM DTT, 500 mM ATP, 500 mM GTP, 500 mM 4-thio-UTP, 2 mM CTP, 200 U/ml Superase-in, 1% Sarkosyl (N-Laurylsarcosine sodium salt solution)) was added and incubated for 7 min at 30 °C for the nuclear run-on. RNA was extracted and purified with TRIzol LS reagent (Invitrogen) and ethanol precipitation. Nuclear run-on RNA was then fragmented with a UCD-200 Bioruptor for 1–5 cycles of 30 s ON/30 s OFF at high settings. Fragmented RNA was incubated at 65 °C for 10 min then on ice for 5 min to ensure full exposing of 4-thio-UTP. Biotinylated solution (20 mM Tris pH 7.5, 2 mM EDTA pH 8.0, 40% dimethylformamide, 200 mg/ml EZ-link HPDP Biotin (Thermo Scientific)) was added to the fragmented RNA for 2 h at room temperature with shaking at 100 g. Followed by ethanol precipitation, the biotinylated RNA was pulled down by streptavidin M280 beads (Invitrogen). 4-thio-UTP labeled RNA was eluted in 100 mM DTT and purified through RNA Clean and Purification kit (Zymo Research). Genomic DNA was eliminated through in-column DNase procedure. Purified RNA was used to constructed RNA sequencing libraries using the NEBNext Ultra II Directional RNA Library Prep kit (New England Biolabs) and was sequenced on the Illumina NextSeq 500 system. Reads were aligned to the hg19 human genome and were processed and analyzed as previously described16. Bam files were normalized based on the drosophila spike-in RNA among each group and bigwig files were generated with deeptools 3.3.1. For average density analysis, deeptools was used to extract read densities, then mean density profiles were generated with R 4.2.0. using ggplot247. For differential gene expression analysis, HOMER 4.10 was used to calculated RPKM (Reads Per Kilobase of transcript per Million fragments mapped) and DESeq2 1.12.3 was used to estimate the significance of differential gene expression among groups.

Low-input KAS-seq was performed mainly based on the published protocol21. Briefly, 10,000 IMR90 cells were incubated in media with 5 mM N3-kethoxal for 10 min at 37 °C. Cells were pelleted and labeled genomic DNA (gDNA) was extracted by using Quick gDNA mini plus kit (Zymo, D4068). DNA was eluted with 25 mM K3BO3 (pH 7.0) and went through click reaction with 20 mM DBCO-PEG4-biotin (sigma 760749) in PBS solution. RNase A (Thermo, 12091039) was added and biotinylated gDNA was fragmented by Tn5 transposase (Illumina) at 37 °C for 30 min. Fragmented DNA was purified and eluted by DNA Clean & Concentrator-5 kit (Zymo, D4013) and was pulled down by streptavidin C1 beads (Invitrogen, 65001). DNA-conjugated beads and corresponding inputs were used for library construction by using index primers and purified by two-sided size selection with Ampure XP beads (Beckman Coulter, A63880). Generated library was sequenced in a 75-bp single-end mode on the Illumina NextSeq 500 platform and raw reads were demultiplexed. Reads were trimmed using trim-galore and low-quality reads or duplicates were removed. Aligned reads were extended to 150 bp to fit the average length of DNA fragments from the KAS-seq libraries. Bedtools were used to convert Bam files to bed files and bedGraph files. Bigwig files were generated by using ucsc-bedgraphtobigwig tools. MACS2 was used to call KAS-seq peaks using the following parameter: –broad -g hs –broad-cutoff 0.01 -q 0.01. All other parameters were set as default.

All procedures were performed by protocols and guidelines approved by the IACUC at the SBP Medical Discovery Institute. The mice were housed in a controlled environment with a 12-h light/dark cycle, while the ambient temperature was maintained at 22–23 °C with 40–60% humidity. Male young C57BL/6N (5-month-old, n = 5) and aged C57BL/6J (21-month-old, n = 5) mice were obtained from Charles River facility and fed by Teklad Gloabal 18% Protein Rodent diet (Inotiv/Envigo, ref 2018). Liver ATAC-seq were performed as previously described48. Briefly, whole liver was excised and maintained on ice cold DMEM and nuclei were isolated while tissue was fresh. Tagmentation was conducted for 30 min at 37 °C followed by DNA isolation and library preparation. For data analyses, reads were mapped to the mm10 mouse genome with Ensemble gene information. HOMER was used for peak calling, bigwig generation and peak counts quantification. 146,684 unique sites with significant peak in at least one sample were identified. DEseq2 was used with raw counts to estimate the significance between groups. Sites with at least one group showing peaks in all four replicates were considered.

All procedures were performed by protocols and guidelines approved by the IACUC at the SBP Medical Discovery Institute. Hepatocytes RNA from male young C57BL/6N (5-month old, n = 5) and aged C57BL/6J (21-month-old, n = 5) mice were extracted as previously described49,50. For data analyses, raw read quality was checked using FastQC v0.10.0. and reads were trimmed with a threshold quality score >20 using Trimgalore v0.3.0. Trimmed reads were then aligned to mm9 using bowtie2. Duplicated reads were marked using Picard Tools v1.98. Peaks were called using macs2 with parameter –nomodel.

Total protein extracts were prepared from indicated samples using RIPA buffer (50 mmol/L Tris-HCl (pH 8.0), 150 mmol/L NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, and 1 mmol/L phenylmethylsulfonyl fluoride) on ice for 20 min. Protein concentration was determined using a BCA protein assay kit (Pierce #23225). Equal amounts of protein samples (10–30 μg) were separated by SDS-PAGE using 8–15% polyacrylamide gels and transferred onto nitrocellulose membranes (Millipore). The membranes were blocked with 4% BSA (Sigma-Aldrich) in TBST buffer (Tris-buffered saline with 0.1% Tween-20) for 1 h at room temperature, and then incubated with specific primary antibodies against the target proteins of interest overnight at 4 °C. The membranes were washed three times with TBST buffer for 8 min each, followed by incubation with horseradish peroxidase-conjugated secondary antibodies (CST, anti-rabbit #7074 and anti-mouse #7076) for 1 h at room temperature. Protein bands were detected using the Dura Extended Duration Substrate (Thermofisher, # 34075).

Total RNA was extracted from cells using the Zymo Quick-RNA™ MiniPrep Kit (R1054) according to the manufacturer’s instructions. RNA concentration and purity were determined using a NanoDrop One spectrophotometer (Thermo Fisher Scientific). Complementary DNA (cDNA) was synthesized from 1 μg of total RNA using the SuperScript™ VILO™ cDNA Synthesis Kit (Thermo Fisher Scientific, #11754050). qPCR was performed using iTaq Universal SYBR Green Supermix (Bio-Rad, #1725121) and run on the QuantStudio 7 Flex Real-Time PCR System. The cycling conditions included an initial denaturation at 95 °C for 3 min, followed by 40 cycles of denaturation at 95 °C for 10 s, annealing at 60 °C for 30 s, and extension at 72 °C for 30 s. The specificity of the PCR products was confirmed by melting curve analysis. Gene expression levels were normalized to the geometric mean of reference gene (GAPDH) using the comparative Ct (ΔΔCt) method, and fold changes were calculated using the 2−ΔΔCt method. List of primers used in the present study is included in Supplementary Data 1.

Stable isotope tracer analysis using [13C6]-glucose was conducted with three biologically independent experiments at groups including control proliferating cells, RAS-induced senescent cells via 4-OHT with control shRNA (shControl), senescent cells with shRNA targeting HK2 (shHK2), senescent cells with shRNA targeting HK2 (shHK2) and rescued with wildtype Flag-HK2, and senescent cells with shRNA targeting HK2 (shHK2) and rescued with mutant Flag-HK2. The experiments were performed by seeding cells at a density of 3 × 105 cells per 6 cm dish and incubating them with 25 mM [13C6]-glucose tracer in glucose-free medium for 30 min. Following incubation, the cells were washed with chilled PBS and incubated with 500 µl of extraction solution (80:20 v/v methanol/water) at 4 °C for 5 min. Next, cells were scraped with a polypropylene cell scraper and the extraction solution from each sample was collected, vortexed, and incubated on dry ice for at least 30 min. Then, each sample was centrifuged at maximum speed at 4 °C for 10 min, and the resulting supernatant was used for analysis. Metabolite measurements were normalized based on the protein concentration in the protein pellets (Pierce BCA protein assay kit; # 23250). A sample pool was generated by mixing equal volumes of all extracts from unlabeled samples. LC–MS metabolite flux analysis was performed on a Thermo Scientific Q Exactive HF-X mass spectrometer equipped with heated electrospray ionization (HESI) II probe and coupled to a Thermo Scientific Vanquish UHPLC system. A Millipore SeQuant ZIC-pHILIC 2.1-mm i.d × 150 mm column with a gradient elution method was employed to separate the metabolites. The MS analysis of all samples was performed in full-scan, polarity-switch mode. The unlabeled sample pool was also analyzed by data-dependent MS/MS in positive and negative modes. The mass spectrometer was operated with a spray voltage set at 3.5 kV in positive ion mode and 3.2 kV in negative ion mode. The heated capillary and HESI probe were set at 325 °C and 350 °C, respectively, with the S-lens RF level set at 40. The gas settings for sheath, auxiliary, and sweep were 40, 10, and 2 units, respectively. Full MS scans were acquired for the top 10 highest abundance ions at a resolution of 15,000 with an AGC target of 5E4, a maximum IT of 50 ms, an isolation width of 1.0 m/z, and a stepped normalized collision energy of 20, 40, and 60. Annotation and quantitation of metabolites and carbon isotopologues with natural isotope abundance correction were performed using Compound Discoverer 3.3 software (Thermo Scientific).

For immunofluorescence staining, cells on coverslips were washed by phosphate-buffered saline (PBS) twice and fixed with 3.7% paraformaldehyde for 12 min at room temperature. The samples were washed twice with PBS and permeabilized with 0.2% Triton X-100 in PBS for 15 min. Following the PBS wash, the samples were blocked with 3% bovine serum albumin (BSA) in PBS for 20 min at room temperature and then incubated with the primary antibody diluted in dilution buffer (1% BSA in PBS) overnight at 4 °C. The next day, the samples were washed three times with PBST and probed with an Alexa Fluor-conjugated secondary antibody diluted in dilution buffer for 1 h at room temperature. The cells or tissues were washed and stained with 1 μg/ml 4,6-diamidino-2-phenylindole (DAPI) for 5 min at room temperature. After three washes with PBST, the coverslips or slides were mounted with an Prolong Gold anti-fade mountant (Thermofisher, #P36930) and imaged using a high-resolution confocal microscope (Leica TCS SP5 II).

Cells were seeded in 24-well plates and cultured for 10 days before staining with 0.05% crystal violet for visualization. The cells were then washed with PBS and air-dried before being imaged using a high-resolution scanner. Images were processed using ImageJ software to quantify the integrated density of the stained cells.

The cell-permeable SPiDER β-gal (Dojindo) was used for detection of senescent β-gal activity as instructed by the manufacturers’ manual. Briefly, 5 × 105 cells of each sample were washed with Hanks’ HEPES buffer twice and incubate at 37 °C for 15 min with 1 μmol/l SPiDER β-gal working solution. Following the incubation, LIVE/DEAD™ Fixable Near-IR Dead Cell Stain Kit (ThermoFisher L34975) was used to stain live or dead cells at RT for 15 min with protection from light. Cell suspensions were analyzed by flow cytometry using channels of FITC and Near-IR respectively.

The 3C-qPCR assays were performed following a previously described protocol10. Briefly, five million cells were fixed with 1% formaldehyde in fresh medium for 8 min at room temperature and quenched with 0.2 M glycine for 10 min. The cells were lysed in cold Hi-C lysis buffer (10 mM Tris pH 8.0, 10 mM NaCl, 0.2% IGEPAL CA-630, cOmplete protease inhibitor cocktail (Roche, # 11697498001)) for 15 min, followed by a wash with 500 μl cold lysis buffer. The cell pellet was resuspended in 500 μl 0.5% SDS and incubated at 62 °C for 10 min with shaking. Then, 150 μl water and 25 μl of 10% Triton X-100 were added to quench the SDS, and the samples were incubated at 37 °C for 15 min with shaking. Next, 25 μl 10× NEBuffer2 and 100 U Mbo I were added to digest the chromatin overnight at 37 °C with shaking. Following inactivation of the Mbo I at 62 °C for 20 min, the ligation was performed at room temperature for 4 h with shaking by adding 750 μl ligation master mix (100 μl of 10×NEB T4 DNA ligase buffer, 80 μl of 10% Triton X-100, 10 μl BSA (10 mg ml−1), 5 μl T4 DNA ligase (400 U μl−1), and 555 μl water). After centrifugation at 10,000 r.p.m. for 5 min, reverse cross-linking was performed at 65 °C for at least 4 h. The DNA was then purified by phenol–chloroform extraction and ethanol precipitation. Finally, qPCR was performed using iTaq Universal SYBR Green Supermix (Bio-Rad, #1725121) and run on the QuantStudio 7 Flex Real-Time PCR System. The data were analyzed using the comparative Ct method. List of primers used in the present study is included in Supplementary Data 1.

Polysome profiling was performed according to the protocol described previously51. Briefly, 3 × 107 cells were washed with ice-cold PBS containing 100 mg ml−1 cycloheximide, harvested, and suspended in hypotonic buffer (10 mM HEPES pH 7.9, 1.5 mM MgCl2, 10 mM KCl, 200 U ml−1 SUPERaseIn RNase inhibitor, 1× protease inhibitor cocktail, and 0.5 mM dithiothreitol). The cells were then incubated on ice for 30 min and centrifuged at 14,000 g for 10 min to remove the nuclear pellet. The resulting supernatant was loaded onto a 10 ml 10–50% sucrose gradient in a polyallomer tube containing 20 mM HEPES–KOH pH 7.5, 15 mM MgCl2, 80 mM KCl, 2 mM dithiothreitol, and 100 mg ml−1 cycloheximide, and centrifuged in a TH-641 rotor (Thermo Fisher) at 19,000 g at 4 °C for 2 h. The gradient was fractionated using a system consisting of a syringe pump (Harvard Apparatus model 11), a density gradient fractionator (Brandel), and an ISCO UA-6 UV/VIS detector. Polysome-free, monosome, and polysome fractions were separated based on their UV-light absorbance. The monosome and polysome fractions were incubated with 25 mM EDTA (pH 8.0), 10 mM Tris–HCl (pH 7.0), and 1% SDS at 65 °C for 5 min. The RNA was then extracted with phenol–chloroform and ethanol precipitation. cDNA was obtained and qPCR was performed following the abovementioned protocols.

The m6A RNA methylation status was detected using total RNA (200 ng) isolated from samples with or without treatment of the METTL3 inhibitor STM2457 (2.5 μM) by the EpiQuik m6A RNA Methylation Quantification Kit (Colorimetric) according to the manufacturer’s protocols. Absorbance signals were measured on a microplate reader at 450 nm, and m6A RNA methylation levels were determined through background extraction.

ATP levels were detected in the indicated conditions using the ATP Luminescence Assay Kit (Dojindo laboratories) following the manufacturer’s protocols. Inosine levels were measured under the indicated conditions by Inosine Quantification Assay Kit (ab126286) based on the manufacturer’s procedures. Fluorescent signals were read at Ex/Em = 535/587 nm on a microplate reader.

Results are representative of a minimum of three biologically independent experiments with data expressed as mean ± standard deviation of the mean (SD) unless stated otherwise. Statistical analyses were conducted using GraphPad Prism 7 (GraphPad) or R 4.2.0. The significance of differences between groups was determined using a two-tailed unpaired Student’s t test, and P values < 0.05 were considered statistically significant unless otherwise stated.

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

HiChIP-seq, fastGRO–seq, KAS-seq, ATAC-seq, and RNA-seq data and processed files generated in this study have been deposited in the GEO under the accession number GSE243047 (HiChIP-seq, fastGRO–seq, and KAS-seq), GSE243630 (ATAC-seq), and GSE243906 (RNA-seq). In addition, publicly available dataset such as GSE74328 (H3K27Ac ChIP-seq), GSE141992 (METTL3 and METTL14 Cut&Run-seq), GSE141993 (m6A-seq), GSE141994 (RNA-seq), GSE197017 (mouse skeletal muscle tissue scRNA-seq), GSE118494 (Hi-C), and CRA005942 (primates m6A-seq) were obtained from GEO. Metabolomics raw data has been deposited to Metabolomics Workbench (study_id: ST003182 [https://doi.org/10.21228/M8NQ82])52. Spectral data for relevant metabolomics can be found in Supplementary Data 2. Source data are provided with this paper.

Kumari, R. & Jat, P. Mechanisms of cellular senescence: cell cycle arrest and senescence associated secretory phenotype. Front. Cell Dev. Biol. 9, 645593 (2021).

Article PubMed PubMed Central Google Scholar

Narita, M. et al. A novel role for high-mobility group a proteins in cellular senescence and heterochromatin formation. Cell 126, 503–514 (2006).

Article CAS PubMed Google Scholar

Coppe, J. P., Desprez, P. Y., Krtolica, A. & Campisi, J. The senescence-associated secretory phenotype: the dark side of tumor suppression. Annu. Rev. Pathol. 5, 99–118 (2010).

Article CAS PubMed PubMed Central Google Scholar

Hao, X., Wang, C. & Zhang, R. Chromatin basis of the senescence-associated secretory phenotype. Trends Cell Biol. 32, 513–526 (2022).

Article CAS PubMed PubMed Central Google Scholar

Ritschka, B. et al. The senescence-associated secretory phenotype induces cellular plasticity and tissue regeneration. Genes Dev. 31, 172–183 (2017).

Article CAS PubMed PubMed Central Google Scholar

Liu, J. et al. A METTL3-METTL14 complex mediates mammalian nuclear RNA N6-adenosine methylation. Nat. Chem. Biol. 10, 93–95 (2014).

Article ADS CAS PubMed Google Scholar

Ries, R. J. et al. m(6)A enhances the phase separation potential of mRNA. Nature 571, 424–428 (2019).

Article CAS PubMed PubMed Central Google Scholar

Fu, Y. & Zhuang, X. m(6)A-binding YTHDF proteins promote stress granule formation. Nat. Chem. Biol. 16, 955–963 (2020).

Article CAS PubMed PubMed Central Google Scholar

Gao, Y. et al. Multivalent m(6)A motifs promote phase separation of YTHDF proteins. Cell Res. 29, 767–769 (2019).

Article CAS PubMed PubMed Central Google Scholar

Liu, P. et al. m(6)A-independent genome-wide METTL3 and METTL14 redistribution drives the senescence-associated secretory phenotype. Nat. Cell Biol. 23, 355–365 (2021).

Article CAS PubMed PubMed Central Google Scholar

Oudelaar, A. M. & Higgs, D. R. The relationship between genome structure and function. Nat. Rev. Genet. 22, 154–168 (2021).

Article CAS PubMed Google Scholar

Deng, W. et al. Controlling long-range genomic interactions at a native locus by targeted tethering of a looping factor. Cell 149, 1233–1244 (2012).

Article CAS PubMed PubMed Central Google Scholar

Schoenfelder, S. & Fraser, P. Long-range enhancer-promoter contacts in gene expression control. Nat. Rev. Genet. 20, 437–455 (2019).

Article CAS PubMed Google Scholar

Wang L., et al. Dynamic enhancer interactome promotes senescence and aging. bioRxiv, (2023).

Mumbach, M. R. et al. HiChIP: efficient and sensitive analysis of protein-directed genome architecture. Nat. Methods 13, 919–922 (2016).

Article CAS PubMed PubMed Central Google Scholar

Barbieri, E. et al. Rapid and scalable profiling of nascent RNA with fastGRO. Cell Rep. 33, 108373 (2020).

Article CAS PubMed PubMed Central Google Scholar

Dorr, J. R. et al. Synthetic lethal metabolic targeting of cellular senescence in cancer therapy. Nature 501, 421–425 (2013).

Article ADS PubMed Google Scholar

Kaplon, J. et al. A key role for mitochondrial gatekeeper pyruvate dehydrogenase in oncogene-induced senescence. Nature 498, 109–112 (2013).

Article ADS CAS PubMed Google Scholar

Guo, D., Meng, Y., Jiang, X. & Lu, Z. Hexokinases in cancer and other pathologies. Cell Insight 2, 100077 (2023).

Article PubMed PubMed Central Google Scholar

Tasdemir, N. et al. BRD4 connects enhancer remodeling to senescence immune surveillance. Cancer Discov. 6, 612–629 (2016).

Article CAS PubMed PubMed Central Google Scholar

Wu, T., Lyu, R., You, Q. & He, C. Kethoxal-assisted single-stranded DNA sequencing captures global transcription dynamics and enhancer activity in situ. Nat. Methods 17, 515–523 (2020).

Article CAS PubMed PubMed Central Google Scholar

Lupianez, D. G. et al. Disruptions of topological chromatin domains cause pathogenic rewiring of gene-enhancer interactions. Cell 161, 1012–1025 (2015).

Article CAS PubMed PubMed Central Google Scholar

Iwasaki, O. et al. Involvement of condensin in cellular senescence through gene regulation and compartmental reorganization. Nat. Commun. 10, 5688 (2019).

Article ADS CAS PubMed PubMed Central Google Scholar

Hagege, H. et al. Quantitative analysis of chromosome conformation capture assays (3C-qPCR). Nat. Protoc. 2, 1722–1733 (2007).

Article CAS PubMed Google Scholar

Yankova, E. et al. Small-molecule inhibition of METTL3 as a strategy against myeloid leukaemia. Nature 593, 597–601 (2021).

Article ADS CAS PubMed PubMed Central Google Scholar

Roundtree, I. A. & He, C. Nuclear m(6)A reader YTHDC1 regulates mRNA Splicing. Trends Genet. 32, 320–321 (2016).

Article CAS PubMed Google Scholar

Jeyapalan, J. C. & Sedivy, J. M. Cellular senescence and organismal aging. Mech. Ageing Dev. 129, 467–474 (2008).

Article CAS PubMed PubMed Central Google Scholar

Consortium A. A. Aging Atlas: a multi-omics database for aging biology. Nucleic Acids Res. 49, D825–D830 (2021).

Ferreira, J. C. et al. Linker residues regulate the activity and stability of hexokinase 2, a promising anticancer target. J. Biol. Chem. 296, 100071 (2021).

Article CAS PubMed Google Scholar

Nawaz M. H. et al. The catalytic inactivation of the N-half of human hexokinase 2 and structural and biochemical characterization of its mitochondrial conformation. Biosci Rep. 38, BSR20171666 (2018).

Blaha, C. S. et al. A non-catalytic scaffolding activity of hexokinase 2 contributes to EMT and metastasis. Nat. Commun. 13, 899 (2022).

Article ADS CAS PubMed PubMed Central Google Scholar

Stincone, A. et al. The return of metabolism: biochemistry and physiology of the pentose phosphate pathway. Biol. Rev. Camb. Philos. Soc. 90, 927–963 (2015).

Article PubMed Google Scholar

Jain, S. et al. ATPase-modulated stress granules contain a diverse proteome and substructure. Cell 164, 487–498 (2016).

Article CAS PubMed PubMed Central Google Scholar

Lin, Y., Protter, D. S., Rosen, M. K. & Parker, R. Formation and maturation of phase-separated liquid droplets by RNA-binding proteins. Mol. Cell 60, 208–219 (2015).

Article CAS PubMed PubMed Central Google Scholar

Wang, J. et al. Binding to m(6)A RNA promotes YTHDF2-mediated phase separation. Protein Cell 11, 304–307 (2020).

Article PubMed Google Scholar

Kroschwald S., Maharana S. & Simon A. Hexanediol: a chemical probe to investigate the material properties of membrane-less compartments. Matters 201702000010 (2017).

Bertoni, J. M. Competitive inhibition of rat brain hexokinase by 2-deoxyglucose, glucosamine, and metrizamide. J. Neurochem. 37, 1523–1528 (1981).

Article CAS PubMed Google Scholar

Wu, Z. et al. m(6)A epitranscriptomic regulation of tissue homeostasis during primate aging. Nat. Aging 3, 705–721 (2023).

Article CAS PubMed Google Scholar

Moiseeva, V. et al. Senescence atlas reveals an aged-like inflamed niche that blunts muscle regeneration. Nature 613, 169–178 (2023).

Article ADS CAS PubMed Google Scholar

Wang, R. et al. Global stable-isotope tracing metabolomics reveals system-wide metabolic alternations in aging Drosophila. Nat. Commun. 13, 3518 (2022).

Article ADS CAS PubMed PubMed Central Google Scholar

Hilliker, A., Gao, Z., Jankowsky, E. & Parker, R. The DEAD-box protein Ded1 modulates translation by the formation and resolution of an eIF4F-mRNA complex. Mol. Cell 43, 962–972 (2011).

Article CAS PubMed PubMed Central Google Scholar

Wheeler J. R., Matheny T., Jain S., Abrisch R. & Parker R. Distinct stages in stress granule assembly and disassembly. Elife 5, e18413 (2016).

Vredeveld, L. C. et al. Abrogation of BRAFV600E-induced senescence by PI3K pathway activation contributes to melanomagenesis. Genes Dev. 26, 1055–1069 (2012).

Article CAS PubMed PubMed Central Google Scholar

Young, A. R. et al. Autophagy mediates the mitotic senescence transition. Genes Dev. 23, 798–803 (2009).

Article CAS PubMed PubMed Central Google Scholar

Lareau, C. A. & Aryee, M. J. hichipper: a preprocessing pipeline for calling DNA loops from HiChIP data. Nat. Methods 15, 155–156 (2018).

Article CAS PubMed PubMed Central Google Scholar

Imakaev, M. et al. Iterative correction of Hi-C data reveals hallmarks of chromosome organization. Nat. Methods 9, 999–1003 (2012).

Article CAS PubMed PubMed Central Google Scholar

Villanueva RAM & Chen, Z. J. ggplot2: elegant graphics for data analysis (2nd ed.). Meas. Interdiscip. Res. Perspect. 17, 160–167 (2019).

Article Google Scholar

Corces, M. R. et al. An improved ATAC-seq protocol reduces background and enables interrogation of frozen tissues. Nat. Methods 14, 959–962 (2017).

Article CAS PubMed PubMed Central Google Scholar

Charni-Natan, M. & Goldstein, I. Protocol for primary mouse hepatocyte isolation. STAR Protoc. 1, 100086 (2020).

Article PubMed PubMed Central Google Scholar

Jung, Y., Zhao, M. & Svensson, K. J. Isolation, culture, and functional analysis of hepatocytes from mice with fatty liver disease. STAR Protoc. 1, 100222 (2020).

Article PubMed PubMed Central Google Scholar

Hao, X. et al. ADAR1 downregulation by autophagy drives senescence independently of RNA editing by enhancing p16(INK4a) levels. Nat. Cell Biol. 24, 1202–1210 (2022).

Article CAS PubMed PubMed Central Google Scholar

Sud, M. et al. Metabolomics Workbench: an international repository for metabolomics data and metadata, metabolite standards, protocols, tutorials and training, and analysis tools. Nucleic Acids Res. 44, D463–D470 (2016).

Article CAS PubMed Google Scholar

Download references

This work was supported by US National Institutes of Health grants R01CA160331 and R01CA276569 to R.Z., and P01AG031862 to K.N., P.D.A., S.L.B and R.Z. R.Z. is a CPRIT Scholar in Cancer Research. Support of Core Facilities was provided by Cancer Centre Support Grant (CCSG) CA010815 to The Wistar Institute and P30CA016672 to The University of Texas MD Anderson Cancer Center. Support of metabolomic study was provided by NIH grant U2C-DK119886 and OT2-OD030544 grants.

Department of Experimental Therapeutics, University of Texas M.D. Anderson Cancer Center, Houston, TX, 77030, USA

Chen Wang, Liping Liao, Xue Hao, Hao Nie & Rugang Zhang

Institute of Molecular Biology, University of Oregon, Eugene, OR, 97403, USA

Hideki Tanizawa & Ken-ichi Noma

Institute for Genetic Medicine, Hokkaido University, Sapporo, 060-0815, Japan

Hideki Tanizawa & Ken-ichi Noma

Gene Expression and Regulation Program, The Wistar Institute, Philadelphia, PA, 19104, USA

Connor Hill, Qiang Zhang, Bin Tian & Alessandro Gardini

Sanford Burnham Prebys Medical Discovery Institute, San Diego, CA, USA

Aaron Havas, Xue Lei & Peter D. Adams

Penn Epigenetics Institute, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA, USA

Lu Wang & Shelley L. Berger

Department of Bioinformatics & Computational Biology, University of Texas MD Anderson Cancer Center, Houston, TX, 77054, USA

Yuan Qi

Immunology, Microenvironment and Metastasis Program, The Wistar Institute, Philadelphia, PA, USA

Andrew V. Kossenkov

Molecular and Cellular Oncogenesis Program, The Wistar Institute, Philadelphia, PA, USA

Aaron Goldman

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

C.W., H.T., C.H., A.H., Q.Z., L.L., X.H., X.L., L.W., H.N., Y.Q., A.V.K., and A.G. (Aaron Goldman) performed the experiments and analyzed data. C.W., R.Z. designed the experiments. A.H., Q.Z., X.L., L. W., A. G. (Alessandro Gardini), and B.T. contributed key experimental materials. S.B., P.D.A., K.N., and R.Z. supervised studies. C.W. and R.Z. wrote the manuscript. R.Z. conceived the study.

Correspondence to Rugang Zhang.

The authors declare no competing interests.

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

Wang, C., Tanizawa, H., Hill, C. et al. METTL3-mediated chromatin contacts promote stress granule phase separation through metabolic reprogramming during senescence. Nat Commun 15, 5410 (2024). https://doi.org/10.1038/s41467-024-49745-5

Download citation

Received: 20 November 2023

Accepted: 18 June 2024

Published: 26 June 2024

DOI: https://doi.org/10.1038/s41467-024-49745-5

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative