The expression of histone Kcr positively correlates with the severity of CKD in patients and mice
To explore the roles of protein crotonylation in human kidneys, we first collected kidney biopsy slides and determined the clinical characteristics of patients with CKD, and performed immunohistochemistry (IHC) using antibodies targeting crotonyl lysine (anti-Kcr) (Fig. 1A and Fig. S1A). The results demonstrated that Kcr was mainly present in TECs, and was significantly higher in the kidneys of patients with than without CKD (Fig. 1A). The IHC data showed that the intensity of Kcr staining positively correlated with disease progression, especially samples with nuclear staining (Fig. 1B and Fig. S1B). Kcr participated in CKD as well as contributed to fibrosis in the obstructed kidneys of mice. The immunofluorescence (IF) staining data showed that unilateral ureteric obstructions (UUOs) induced the increasing expression of Kcr, which was mainly colocalized in the nuclei (Fig. 1C).
To confirm the changes in the histone Kcr, we extracted total histones from the fibrotic kidneys of mice and evaluated the histone purification using Coomassie brilliant blue staining and liquid chromatography-mass spectrometry (LC-MS/MS) analysis. The LC-MS/MS analysis confirmed that all the proteins were extracted from the nucleus and functioned with chromatin structure and dynamics (Fig. S2A-C). To compare the levels of histone crotonylation in kidneys, western blotting analysis using anti-Kcr antibodies was performed (Fig. 1D and Fig. S1C), demonstrating that the increase in histone crotonylation was accompanied with kidney fibrosis in both UUO and folic acid nephropathy (FAN) mice (Fig. S2D-G).
As previously reported, different histone lysine modifications might have diverse roles. To identify histone Kcr and Kac at several important histone H3 residues reported (K9, K14, K18, and K27), we used different antibodies to blot the histone extraction. Particularly, H3K9cr increased remarkably in two experimental models of kidney fibrosis, while H3K9ac remained stable (Fig. 1E and Fig. S1D), proposing that H3K9cr may have distinct effects relative to H3K9ac in renal fibrosis. However, H3K18cr and H3K27cr, corresponding with H3K18ac and H3K27ac, both increased in fibrotic kidneys, suggesting that these two markers are coregulated. Most interestingly, H3K14cr increased in kidneys with FAN and UUO; H3K14ac showed opposite trends, recommending that the roles of histone Kcr and Kac are both complicated and comprehensive (Fig. S2H-K).
Genetic deletion of ACSS2 decreased H3K9 crotonylation and alleviated kidney fibrosis
As histone Kcr and Kac are reversible, dynamic processes mediated by multiple identical enzymes, it has been challenging to distinguish between their individual functions. Based on the abovementioned data, the alterations of H3K9cr appear to be independent of those of H3K9ac in kidney fibrosis. We examined known enzymatic and metabolic regulation variables, including Sirtuin (SirT) 1/2/3/4/5/6 8,10, histone deacetylase (HDAC) 1/2/3 11,12, acyl-CoA synthetase short chain family member 2 (ACSS2) 13 and crotonate in vitro, to identify critical factors that could regulate H3K9cr more than H3K9ac (Fig. 2A) in both mouse renal tubular epithelial (TCMK-1) cells, and human embryonic kidney (HEK-293T) cells. By screening these factors, ACSS2 overexpression by plasmid transfection (Fig. S3A-B) dramatically increased H3K9cr expression and did not change H3K9ac, indicating that ACSS2 may have a greater impact on H3K9cr than H3K9ac (Fig. S4B).
Crotonate, a recourses of histone Kcr, also increased the expression of H3K9cr and H3K9ac in TECs (Fig. S3C-D, Fig. S4C-D), demonstrating that crotonate has an impact on the histone Kac process via effects on ACSS2. SirT and HDAC were successfully overexpressed in TCMK-1 and HEK-293T cells after transfection with various plasmids (Fig. S3E-P). Overexpression of SirT1/2/3 and HDAC1/2/3 reduced the expression of H3K9cr and H3K9ac in kidney TECs (Fig. S4E-H); however, overexpression of SirT4/5/6 only affected H3K9ac modification without affecting H3K9cr (Fig. S4I-J). Consequently, we purposed that ACSS2, notwithstanding H3K9ac interference, could be the ideal instrument for examining the functions of H3K9cr in TECs.
To understand the role of ACSS2-mediated H3K9cr in renal fibrosis, we generated mice with genetic deletion of ASCC2, taking advantage of the CRISPR/Cas9 knock-out system (Fig. S5A-C). The results of western blotting confirmed ACSS2 reduction in the kidneys of knockout mice (ACSS2−/−) compared with littermate controls (Fig. S5D-E). The ACSS2−/− mice were born at the expected Mendelian ratio, with no birth or growth defects and no signs of kidney function defects.
First, two experimental kidney fibrosis mice (with UUOs and FAN) were created using ACSS2−/− mice and littermate controls. After evaluating the expression of H3K9cr by western blotting, we confirmed that genetic knockout of ACSS2 could reduce the expression of H3K9cr in fibrotic kidneys, which was consistent with the results of cell experiments. Similarly, the expression of H3K9ac remained unchanged (Fig. 2B and Fig. S6A-B). Next, we analyzed the phenotype of these animals to explore the function of ACSS2 in kidney fibrosis (Fig. 2A). Histological changes, such as tubule atrophy and interstitial fibrosis, were alleviated in ACSS2−/− compared with wild-type (WT) UUO mice (Fig. 2C). Protein markers of fibrosis, including levels of fibronectin (FN1), collagen type 6 (COL6), and smooth muscle actin (α-SMA), were higher in fibrotic kidneys, but lower in ACSS2−/− kidneys in UUO mice (Fig. 2D). Transcript levels of fibrotic markers—including Fn1, collagen type 1a1 (Col1a1) and smooth muscle alpha (α)-2 actin (Acta2)—were altered similarly to the levels of fibrotic markers at the protein level (Fig. 2E). Genetic deletion of ACSS2 was confirmed to decrease H3K9cr and alleviate kidney fibrosis in FAN mice (Fig. S6C-F). All these data confirmed that global genetic deletion of ACSS2 would influence H3K9cr expression and thus alleviate kidney fibrosis.
Tubular-specific deletion of ACSS2 decreased H3K9 crotonylation to delay the progression of kidney fibrosis
TECs play a core role in kidney fibrosis 14. The IHC results confirmed that H3K9cr and ACSS2 was mainly expressed in TECs in fibrotic kidneys (Fig. 3A). To investigate the contribution of ACSS2 in TECs to kidney fibrosis, we crossed ACSS2 flox mice with Ksp-Cre mice to selectively delete ACSS2 in TECs in the kidneys (Fig. S7). The Ksp-Cre ACSS2fl/fl (ACSS2tecKO) mice and Cre-negative littermate controls (ACSS2tecWT) were subjected to UUO or injected with folic acid (FA). The kidneys of ACSS2tecKO and ACSS2tecWT mice exhibited the same extent of H3K9ac when compared with sham-operated kidneys. The H3K9cr expression increased dramatically in ACSS2tecWT UUO kidneys, and was alleviated in ACSS2tecKO UUO kidneys (Fig. 3C). The changes in H3K9cr and H3K9ac were also similar in the kidneys of FAN ACSS2tecKO mice (Fig. S8A).
H&E and Masson staining showed that tubule injury and collagen deposition were alleviated in the kidneys of ACSS2tecKO UUO mice compared with ACSS2tecWT UUO mice (Fig. 3D). The western blot and qPCR results also confirmed that kidney fibrosis in UUO and FAN mice were ameliorated in ACSS2tecKO mice compared with ACSS2tecWT mice, demonstrated by the decreased expression of fibrotic markers (Fig. 3E-F and Fig. S8B-D). In summary, the data indicated that specific deletion of tubular ACSS2 decreased H3K9 crotonylation to delay the progression of kidney fibrosis.
H3K9cr promoted cytokine production and regulated cytokine-cytokine receptor interaction in fibrotic kidneys
Since the genomic locations of H3K9cr have not been previously mapped in kidneys, it is difficult to explore the possible regulatory genes of H3K9cr through the public database. Therefore, we sought to determine the effects of H3K9cr ourselves using chromatin immunoprecipitation sequencing (ChIP-seq). For comparison and as a control, we also performed H3K9ac ChIP-seq using the same samples (Fig. 4A). We found that both H3K9ac and H3K9cr are enriched at transcriptional start sites (Fig. S9). The location of H3K9cr at transcriptional start sites is consistent with previous findings 15,16. Remarkably, the ChIP signal did not differ between H3K9cr and H3K9ac for control mice, while the ChIP signal of H3K9cr was obviously stronger than that of H3K9ac in the kidneys of UUO mice (Fig. 4C).
We also performed RNA-seq on these samples, as the combination of ChIP-seq and RNA-seq data can be used to decipher the transcriptional regulation network (Fig. 4A). Whether in control mice or mice with kidney fibrosis, the results of individual analyses revealed a strong correlation between H3K9cr and H3K9ac regarding gene expression during the process of kidney fibrosis. The highest quartiles of gene expression displayed the highest occupancy of histone Kac and Kcr, suggesting that both H3K9cr and H3K9ac can activate gene transcription (Fig. 4B).
To further examine the relationship between H3K9cr and H3K9ac, we examined the relative ratio of H3K9cr to H3K9ac when both acylations can be detected (Fig. 4D). This analysis demonstrated that gene expression in UUO kidneys were among those with the highest H3K9 crotonylation/ acetylation ratios. When deleting ACSS2 in UUO kidneys, the decrease in H3K9cr leads to lower H3K9cr/ ac ratios. Nevertheless, deletion of ACSS2 had an insignificant effect on genes in control kidneys, as the H3K9cr levels remained unchanged (Fig. S10A). According to these findings, H3K9cr may activate gene expression similarly to H3K9ac, while having a greater impact on renal fibrosis-related genes.
To understand the specific regulatory genes of H3K9cr in kidney fibrosis, we investigated the RNA-seq data between control and UUO kidneys. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway studies revealed considerably enriched pathways of cytokine-cytokine receptor interaction, and a cytokine-mediated signaling pathway in fibrotic kidneys (Fig. S10B-C). After ACSS2 deletion in mice, these two pathways decreased (Fig. S10D-E). Furthermore, to determine whether the reduction of the two signaling pathways produced by ACSS2 deletion is regulated by H3K9cr or H3K9ac, we investigated the results of GO pathway analysis from ChIP-seq data. It is noteworthy that the deletion of ACSS2 had a greater impact on the genes controlling cytokine production by H3K9cr than H3K9ac in UUO kidneys (Fig. 4E), which raises the possibility that the renoprotection offered by the deletion of ACSS2 mainly depends on H3K9cr-mediated cytokine production.
Within these pathways, interleukin-1 (IL-1β) was found to be a core and altered proinflammatory cytokine, as mentioned in a previous study 17. Excitingly, Il1b and Il1r1 displayed increased H3K9cr levels at their proximal promoter regions, and downregulated expression with ACSS2 deletion in UUO kidneys. However, these two genes displayed lower H3K9ac levels and fewer alterations after ACSS2 deletion (Fig. 4F and Fig. S10G). The increase in Il1b and Il1r1 enrichment of H3K9cr following transfection of the ACSS2 plasmid in HEK-293T cells was further confirmed by ChIP-qPCR (Fig. S10H). Overall, our findings indicate that H3K9cr—not H3K9ac—controlled the interaction between cytokines and cytokine receptors (particularly Il1b, which was crucial for fibrotic kidneys).
H3K9 crotonylation promoted IL-1β production in both kidney cells and fibrotic kidneys
As abovementioned, ChIP investigations indicated that IL-1β could be directly regulated by H3K9 crotonylation. We independently evaluated H3K9cr-mediated changes in IL-1β production in animals and cells (Fig. 5A). First, the increase in IL-1β was accompanied by an increase H3K9cr in the fibrotic kidneys of UUO and FAN mice (Fig. 5B and Fig. S11A-B). When global or tubular specific knockout of ACSS2 were applied to decrease H3K9cr modification in mice with kidney fibrosis, both the protein levels and transcription of IL-1β were suppressed (Fig. 5D-E and Fig. S11C-D). Remarkably, it was clear that the deletion of ACSS2 reduced the amount of IL-1β produced in UUO kidneys, without influencing serum concentration (Fig. 5C).
After demonstrating the alterations of H3K9cr-mediated IL-1β in vivo, we used two different kidney cell lines to confirm the link between H3K9cr and IL-1β in vitro. Both the protein and mRNA levels of IL-1β noticeably increased following crotonate stimulation (Fig. S12A-D). The expression of IL-1β was further boosted by transfecting TCMK-1 cells with the ACSS2 plasmid, which raises H3K9cr. In addition to intracellular IL-1β, ELISA kit analysis revealed that the concentration of IL-1β in the supernatant of ACSS2-overexpressed TCMK-1 cells was also elevated (Fig. S12E-G); HEK-293T cells showed similar effects (Fig. S12H-J). When SirT1/2/3 and HDAC1/2/3 were overexpressed in TCMK-1 and HEK-293T cells to block H3K9cr modification, the levels of IL-1β decreased (Fig. S13). Surprisingly, IL-1β expression was unaffected by SirT4/5/6 plasmids, which had no impact on H3K9cr alteration (Fig. S14). Collectively, our findings show that whether in vivo or in vitro, H3K9cr expression is significantly associated with IL-1β production; additionally, ACSS2 may be a key factor to regulate H3K9cr-mediated IL-1β production.
H3K9cr-derived IL-1β promoted macrophage activation in cells and fibrotic kidneys
As known, macrophages (particularly M1 macrophages) are considered deleterious in kidneys, as they sustain the proinflammatory environment, leading to the progression of renal injury and development of fibrosis 18. IL-1β is an important cytokine that can induce proinflammatory or activated M1 macrophages; thus, it is possible that H3K9cr-mediated IL-1β triggers macrophage activation to promote kidney fibrosis progression.
To test whether IL-1β could promote macrophage polarization, we first stimulated RAW264.7 with different doses of IL-1β; according to the Cell Counting Kit-8 assay, low doses of IL-1β can promote macrophage proliferation (Fig. S15A). As expected, microscopic images depicted changes in cell morphology from a rounded M0 to flat M1 phenotype after IL-1β simulation (Fig. S15B). Additionally, there were significantly higher levels of M1 macrophage markers (Tnf-α, iNOS, and Il1b), and lower levels of M2 markers (CD206) in IL-1β-simulated versus M0 cells; this was consistent with the observed morphological changes (Fig. S15C).
To test whether H3K9cr-derived IL-1β exerted similar effects, we collected the supernatant of HEK-293T cells transfected with ACSS2 plasmids ahead of time to increase H3K9cr, and subsequently stimulated RAW264.7 macrophages (Fig. 6A). After supernatant stimulation, the microscopic images and mRNA levels of M1 macrophage markers—including iNOS, Tnf-α, Mcp1, and Il6—changed, indicating that macrophage polarization occurred (Fig. 6B and Fig. S16A). We repeated the experiments with TCMK-1 cells, and observed similar results (Fig. S16B). In addition to supernatants from ACSS2-overexpressed cells, we collected supernatants from HEK-293T and TCMK-1 cells transfected with SirT1/2/3 and HDAC1/2/3 plasmids, which inhibited H3K9cr-mediated IL-1β production. Microscopic images revealed that the flat M1 phenotype in the H3K9cr inhibition group was less than in the control group (Fig. S17A). As the previous qPCR tests revealed, H3K9cr inhibition can reduce M1 macrophage markers (Fig. S17B-C).
When observing the M1 macrophage marker in vivo, we found that regardless of whether global or tubular epithelial-specific knockout of ACSS2 was performed to decrease H3K9cr modification in mice with kidney fibrosis, Cd68 and Tnf-α were suppressed (Fig. S16C-D). Overall, these data suggest that regulating H3K9cr modification in TECs could influence the secretion of IL-1β, thereby influencing macrophage activation.
H3K9cr-derived IL-1β accelerated tubular cell senescence in cells and fibrotic kidneys
Other than mediating macrophage polarization, recent studies report that IL-1β may be involved in the senescence of several types of cells, including vascular smooth muscle cells 19, astrocytes 20, bovine oviduct epithelial cells 21, mature chondrocytes 22, etc. Recently, several studies have shown positive correlations between senescent cell accumulation and fibrosis in kidneys during ageing 23–25 and disease 26–28. Based on known results, we suspected that IL-1β—the key mediator identified by H3K9cr—could also regulate TECs senescence in kidney fibrosis.
First, we administered IL-1β directly to TECs at different concentration to determine whether it affected tubular cell senescence. High doses of exogenous IL-1β did not cause any toxicity in TCMK-1 cells, as shown in Fig. S18A; however, even low doses of IL-1β triggered senescence-associated β-galactosidase (SA-β-gal) positive cells (Fig. S18B). IL-1β also increased the cellular senescence marker P53, and several senescence-associated secretory phenotype (SASP) markers (Il6, Mmp9, and Il1b) (Fig. S18C).
To further confirm whether H3K9cr-derived IL-1β has similar effects, we collected supernatants from TCMK-1 cells or HEK-293T cells transfected with ACSS2 plasmids, which are IL-1β enriched. Thereafter, we treated TCMK-1 cells to observe changes in senescence markers. Similarly, the senescence phenotype increased in vitro, as evidenced by increased P53, Il6, and Mcp1 levels (Fig. S19A-B). When we evaluated the data in vivo, the expression of P53, Il6, and Mcp1 increased in the kidneys of UUO and FAN mice (Fig. S19C-F). Global knockout of ACSS2 in UUO and FAN mice to decrease H3K9cr-mediated IL-1β production resulted in a decrease in P53, Il6, and Mcp1 (Fig. S19G-J). Even when we specifically knocked out ACSS2 in renal TECs, we discovered that the kidney cellular senescence marker decreased significantly (Fig. S19K-L). These data therefore suggest that regulating H3K9cr modification in TECs could influence the secretion of IL-1β and thus influence senescence in TECs.
Anti-IL-1β IgG treatment alleviated macrophage activation and tubular cell senescence in kidney cells and fibrotic kidneys
We attempted to identify approaches to suppress IL-1β after confirming that H3K9cr-generated IL-1β might drive macrophage activation and tubular cell senescence. When anti-IL-1β IgG was added to RAW264.7 cells treated with IL-1β, IL-1β-induced M1 macrophage polarizations and cellular senescence markers P53 and SASP were also diminished (Fig. 6C and E). When treating cells with IL-1β antibodies to neutralize IL-1β-enriched supernatant stimulation, the changes in morphology and mRNA expression of M1 phenotypic markers and cellular senescence could be alleviated (Fig. 6D and F).
To examine the effects of IL-1β in vivo, we also administered anti-IL-1β IgG to UUO mice (Fig. 7A). Anti-IL-1β IgG dramatically reduced renal fibrosis in UUO mice, as demonstrated by Masson's trichrome staining and the decreased mRNA and protein expression of kidney fibrosis markers (Fig. 7B-D and Fig. S20A). Anti-IL-1β IgG treatment also suppressed key markers of M1 macrophage and cellular senescence (Fig. 7E-G and Fig. S20B-C). To summarize the data presented above, anti-IL-1β IgG can alleviate M1 macrophage polarization and tubular cellular senescence caused by H3K9cr-mediated IL-1β both in vivo and in vitro, and thus improve kidney fibrosis.
Pharmacological inhibition of ACSS2 repressed H3K9cr-mediated IL-1β production to protect kidney fibrosis
Although anti-IL-1β IgG can help treat renal fibrosis, monoclonal antibodies are prohibitively expensive for patients with CKD who require long-term treatment. Another option is to use small molecule enzyme inhibitors, which are less expensive and easier to use 29. Unfortunately, no small molecule inhibitor of IL-1β is available on the market. Importantly, the newly developed ACSS2 inhibitor could serve as an alternative to inhibit IL-1β by reducing H3K9cr modification. In our search to discover a potential therapy, we identified an ACSS2 inhibitor (S8588) 30 that can treat kidney fibrosis mice (Fig. 8A). ACSS2 protein production was inhibited in control mice after seven days of ACSS2 inhibitor administration (Fig. S21A). Inhibiting ACSS2 suppressed H3K9cr, and had no effect on H3K9ac (Fig. 8B and Fig. S21B). Remarkably, the ACSS2 inhibitor (S8588) significantly reduced kidney fibrosis in UUO and FAN mice, as evidenced by improvements in pathological staining, western blot, and qPCR for fibrosis markers (Fig. 8C-D and Fig. S21C-E). As expected, IL-1β decreased after inhibitor treatment, accompanied by a decline in SA-β-gal-positive senescent cells (Fig. 8E-F and Fig. S22A-E). Additionally, the cellular senescence markers P53 and SASP were suppressed (Fig. 8G and Fig. S22G-J). Interestingly, these SASPs could also be identified by M1 macrophages, specifically Il1b, Il6, and Mcp1. The decrease in these inflammatory markers was accompanied by a decrease in the M1 macrophage marker Cd86 (Fig. 8G).