Set7 Methyltransferase and Phenotypic Switch in Diabetic Glomerular Endothelial Cells

Visual Abstract Key Points Set7 knockout improves diabetic glomerular structure and function and prevents diabetes-induced endothelial–mesenchymal transition (EDMT) by regulating Igfbp5. Set7 knockdown prevents, and (R)-PFI-2 hydrochloride reverses, diabetes-induced EDMT by regulating insulin growth factor binding protein 5. Set7 regulates the phenotypic EDMT switch, and inhibiting the methyltransferase attenuates glomerular injury in diabetic kidney disease. Background Hyperglycemia influences the development of glomerular endothelial cell damage, and nowhere is this more evident than in the progression of diabetic kidney disease (DKD). While the Set7 lysine methyltransferase is a known hyperglycemic sensor, its role in endothelial cell function in the context of DKD remains poorly understood. Methods Single-cell transcriptomics was used to investigate Set7 regulation in a mouse model of DKD, followed by validation of findings using pharmacological and short hairpin RNA inhibition inhibition of Set7. Results Set7 knockout (Set7KO) improved glomerular structure and albuminuria in a mouse model of diabetes. Analysis of single-cell RNA-sequencing data showed dynamic transcriptional changes in diabetic renal cells. Set7KO controls phenotype switching of glomerular endothelial cell populations by transcriptional regulation of the insulin growth factor binding protein 5 (IGFBP5). Chromatin immunoprecipitation assays confirmed that the expression of the IGFBP5 gene was associated with mono- and dimethylation of histone H3 lysine 4 (H3K4me1/2). This generalizability was investigated in human kidney and circulating hyperglycemic cells exposed to TGFβ1. We showed that the highly selective Set7 inhibitor (R)-PFI-2 hydrochloride attenuated indices associated with renal cell damage and mesenchymal transition, specifically (1) reactive oxygen species production, (2) IGFBP5 gene regulation, and (3) expression of mesenchymal markers. Furthermore, renal benefit observed in Set7KO diabetic mice closely corresponded in human glomerular endothelial cells with (R)-PFI-2 hydrochloride inhibition or Set7 short hairpin RNA silencing. Conclusions Set7 regulates the phenotypic endothelial–mesenchymal transition switch and suggests that targeting the lysine methyltransferase could protect glomerular cell injury in DKD. Podcast This article contains a podcast at https://dts.podtrac.com/redirect.mp3/www.asn-online.org/media/podcast/JASN/2024_04_25_ASN0000000000000345.mp3


Introduction
Diabetic kidney disease (DKD) is the leading cause of kidney failure, 1,2 and chronic hyperglycemia is an important contributing factor to glomerular endothelial cell injury. 3Diabetes-induced glomeruli lesions can lead to mesangial expansion with podocyte loss and glomerular endothelial cell damage that influence hemodynamic indices early in the progression of DKD. 4,5In addition to maintaining glomerular homeostasis, endothelial cells undergo a process of mesenchymal transition (endothelial-mesenchymal transition or EDMT) that is a distinguishing event in the development of diabetic kidney fibrosis. 6,7While extracellular stimuli such as TGFb1, inflammatory molecules, and metabolic dysfunction can promote the transition, 8,9 the mechanism of glomerular EDMT remains poorly understood in DKD.
signaling by TGFb1 activation under hyperglycemic conditions.This article characterizes the hyperglycemic sensor Set7 with its role in glomerular endothelial cell function and diabetic kidney injury.

Methods
Further details regarding the materials and procedures, including Supplemental Tables 1 and 2, are provided in the Supplemental Material.

Diabetic Mouse Model
The Set7 deletion (Set7 2/2 , deletion of the functional membrane occupation and recognition nexus domain in exon 2) mice were generated by genOway (Lyon, France).The Set7 2/2 mice were crossed with the ApoE 2/2 mouse strain to generate double knockout animals (Set7 2/2 ApoE 2/2 ).Sixweek-old male Set7 1/1 ApoE 2/2 and Set7 2/2 ApoE 2/2 mice were rendered diabetic by five daily intraperitoneal injections of streptozocin (Sigma-Aldrich) at a dose of 55 mg/ kg.At 10 weeks after the induction of diabetes, the levels of blood glucose and glycated hemoglobin (hemoglobin A1c) including other metabolic parameters were measured to confirm progression of diabetes.At 10 weeks, the animals were killed by CO 2 overload.The kidneys were rapidly dissected and processed for subsequent analyses.All animal experiments were approved by the Alfred Medical Research and Education Precinct Animal Ethics Committee.

Single-Cell RNA Sequencing
Preparation of single-cell suspension from fresh mouse kidney tissue was performed as described in the Online Data Supplement.Single-cell capture and transcriptomic profiling were performed using a droplet-based (Drop-Seq) technology and single-cell microfluidics platform (Dolomite Bio).Barcoded libraries were generated using the Nextera XT DNA Library Preparation Kit (Illumina) in house and sequenced by Novogene (Singapore).Single-cell data were processed and quantified with Drop-seq tools version 2.1 and further analyses using the R package Seurat version 3. 22,23 Cell Culture Immortalized human glomerular endothelial cells were cultured and differentiated as previously described. 24For target validation experiment using the Set7 inhibitor, human immortalized cells were cultured with or without 15 mM (R)-PFI-2 hydrochloride (PFI-2) (PFI-2; Cayman Chemicals) dissolved in DMSO for 24 hours before exposing them to 5.5 or 25 mM D-glucose in the presence or absence of 5 ng/ml TGF-b1 (R&D Systems) for 48 hours at 37°C.

RNA and Protein Analysis
RNA and protein analyses were performed as described previously. 12Total RNA from the cells was extracted using TRIzol (Life Technologies) and the Direct-zol RNA Mini prep kit (Zymo Research) according to the manufacturer's instructions.cDNA synthesis and quantitative real-time reversetranscription PCR analysis of gene expression were performed as previously described. 12,25,26Real-time quantitative PCR was performed by ABI Prism 7500 using the primers shown in Supplemental Table 1.Whole-cell extract preparation and histone extraction were performed as described previously. 12Histone methyltransferase activity assay was performed as described previously. 12,25,26Protein blotting signals were quantified by an infrared imaging system (Odyssey; LI-COR).An expanded description and the other assays are provided in the Data Supplement.

Chromatin Immunoprecipitation
Chromatin immunoprecipitation assay was performed as described previously 11,12 using anti-H3K4me1 (Active Motif) and anti-H3K4me2 (Millipore) antibodies.The promoter regions of the human insulin growth factor binding protein 5 (IGFBP5) gene were amplified by quantitative real-time PCR, and the primers are provided in Supplemental Table 2.

Statistical and Bioinformatic Analyses
Data are represented as mean6SEM.Statistical significance was determined using Student t-tests and one-or two-way ANOVA with Tukey post hoc test for multiple comparisons as necessary in Prism 9 (Graphpad).A P value , 0.05 was considered statistically significant.

Data Availability
All raw sequencing data are available in NCBI's Gene Expression Omnibus database GSE158626.

Set7KO Improved Diabetic Glomerular Structure and Function
To assess the role of Set7 in DKD development, we generated the Set7-deficient (Set7 2/2 ) mouse by the deletion of exon 2 of the Setd7 gene.The Set7 2/2 mice were crossed with apolipoprotein E-deficient (ApoE 2/2 ) mice to accelerate DKD and vascular complication. 27We confirmed that Set7 and ApoE were absent at the mRNA level in the kidney cortex isolated from Set7 2/2 ApoE 2/2 (referred to as Set7KO) mice (Supplemental Figure 1).To induce diabetes, streptozocin was administered over 5 consecutive days in the Set7KO mice.At 10 weeks, renal parameters were monitored, including the regulatory effect of Set7KO on the transcriptome using single-cell RNA-sequencing (scRNA-seq).Furthermore, the pharmacological Set7 inhibitor PFI-2 28 was used for gene target validation in human kidney cells (Figure 1A).We observed an increase in blood glucose and glycated hemoglobin (hemoglobin A1c) and decreased body weights of diabetic mice when compared with nondiabetic controls (Table 1).In contrast to the diabetic animals, water intake and urine output were improved in Set7KO mice.Kidney parameters, including elevated urinary albumin excretion and albumin-creatinine ratio (Figure 1, B and C), kidney injury molecule 1 expression in the kidney cortex (Figure 1D), glomerular collagens I and IV (Figure 1, E and F), and mesangial area expansion (Figure 1G), were similarly improved in the diabetic Set7KO mice.

Set7 Regulates Pathways Associated with Diabetic Kidney Injury
To identify the complete transcriptomic changes regulated by Set7, we performed high-depth scRNA-seq in diabetic groups.Analysis identified proximal tubule cells, glomerular endothelial cells, macrophages, mesenchymal cells, and podocytes as the major cell types in the kidney cortex (Figure 2A and Supplemental Figure 2).While the number of proximal tubule cells were reduced in the diabetic group (P 5 0.05), we observed an increase in the glomerular endothelial cell cluster (P 5 0.07) in the diabetic Set7KO group (Figure 2B).To assess diabetes-associated pathways regulated by Set7, we performed gene set enrichment analysis (GSEA) on the scRNA-seq clusters and identified four major reactome pathways involving respiratory electron transport (abbreviated as OXPHOS), ribosomal RNA (rRNA) processing, extracellular matrix organization (EMO), and peroxisome proliferator-activated receptor alpha (PPARa) activation (Figure 2C).Representative genes from these pathways were assessed for expression, specifically Ndufb2 and Mdh2 (OXPHOS), Acox1 (PPARa activation), Nhp2 (rRNA processing), and the gene group comprising Col4a3 and Pdgfa (EMO).Ndufb2, Mdh2, and Acox1 were highly expressed in proximal tubule cells (Figure 2D).We observed the expression of Col4a3 was highest in podocytes, whereas Pdgfa was distributed between the glomerular endothelial and tubule cell clusters.These single-cell data suggest that Set7KO may regulate major pathways that are associated with diabetic kidney injury.

Set7 Controls the Glomerular Endothelial Transcriptome in DKD
Single-cell RNA-seq showed dynamic transcriptional regulation by Set7 in the kidney cortex.Consistent with previous studies, 20,29 we observed increased Set7 mRNA expression in the diabetic kidney (Figure 3A) consequentially in the podocytes (24%), mesenchymal (MSC, 10%), and glomerular endothelial cell (8%) clusters that were barely detectable in proximal tubule cells (0.8%) (Figure 3B).Despite the increase in glomerular endothelial cell numbers (shown previously in Figure 2B), Set7KO reduced the expression of genes that define glomerular endothelial cell identity in diabetic animals (Figure 3C).Indeed, we observed segregation of the glomerular endothelial cell    population from the cell cluster defining mesenchymal identity with reduced gene expression in the diabetic Set7KO mice.This influence on the transcriptome was consistent with a phenotypic switch involving the glomerular endothelial cell population.Gene set enrichment analysis of the glomerular endothelial and mesenchymal clusters found the principal pathway regulating epithelial-to-mesenchymal transition (EMT) was inversely correlated: while the EMT pathway was upregulated in the diabetic group, EMT was downregulated by Set7KO in diabetic mice (Figure 3D).These results suggest that Set7 is closely associated with diabetic glomerular endothelial-to-mesenchymal transition.
Unsupervised clustering of cell populations (Figure 4A) found two glomerular endothelial subpopulations on the basis of Igfbp5 expression levels, designated as glomerular endothelial (Igfbp5 high) and glomerular endothelial (Igfbp5 low) (Figure 4, B and C).The regulatory dosage of renal Igfbp5 is critical because the transcription factor is a marker of endothelial identity 30,31 and is closely associated with the development of DKD. 32,33While Igfbp5 expression was elevated in diabetic Set7WT mice, mRNA levels remain unchanged in the kidney cortex in Set7KO mice (Supplemental Figure 3A).Functionally divergent, Igfbp5 was predominantly expressed in the glomerular endothelial cell population (Supplemental Figure 3B).While the glomerular endothelial (Igfbp5 high) cells account for more than 75% of the endothelial population in diabetic mice, this is significantly reduced (P 5 0.03) in the diabetic Set7KO group (Figure 4D).By contrast, the glomerular endothelial (Igfbp5 low) cell cluster (P 5 0.01) was remarkably elevated in the kidney cortex of diabetic Set7KO mice (Figure 4, C and D).
Whereas glomerular endothelial Igfbp5 high and low cells clustered separately from their respective control groups, scRNA-seq confirmed glomerular endothelial cell identity with strong expression of Plat (plasminogen activator tissue type), Ehd3 (Eps15 homology domain-containing protein 3), and Emcn (endomucin) in glomerular endothelial (Igfbp5 high) and glomerular endothelial (Igfbp5 low) groups.Importantly, GSEA identified pathways involved with EMT, and reactive oxygen species (ROS) were upregulated in the glomerular endothelial (Igfbp5 high) cell population but downregulated in the glomerular endothelial (Igfbp5 low) cell cluster (Figure 4E).These results suggest that Set7 regulates dosage of Igfbp5 expression in the glomerular endothelial transcriptome influencing DKD.

PFI-2 Prevented Expression of Diabetes-Associated Renal Genes
To assess whether inhibition of the hyperglycemic sensor Set7 could prevent EDMT and was consistent with a model in which histone methylation is coupled with transcriptional output, we stimulated clinically relevant human cells with PFI-2 28 as shown in Figure 5A.To assess the conformational changes in the C-terminal domain upon substrate binding, the crystal structure of SET7 in complex with S-adenosyl methionine or S-adenosyl homocysteine, the histone H3 peptide and PFI-2 were compared.Molecular docking analysis indicated PFI-2 bound to the substrate peptide-binding groove with an affinity of 29.5 kcal/mol (Figure 5B).PFI-2 was predicted to form hydrogen bonds with G336 and the polar residue S268.On the basis of the per-residue root-mean-square deviation analysis, the post-SET loop (G336-A349) displayed the highest degree of variability (Figure 5C).We confirmed specific inhibition of H3K4 methyltransferase activity by PFI-2 (Figure 5D).To investigate the binding mode of the unmethylated histone H3 peptide to SET7 in the absence and presence of PFI-2, protein-peptide docking was performed using the HPEPDOCK algorithm.In the absence of PFI-2, the unmethylated histone H3 peptide was found to preferentially bind to the substrate peptide-binding groove and the orientation coincided with the experimentally determined structure of the methylated histone H3 peptide (Figure 5E).Although the unmethylated histone H3 peptide maintained its preference for the peptide-binding groove in the presence of PFI-2, the target lysine residue (K4) was displaced from the lysinebinding channel (Figure 5E, right).Because hyperglycemia activates the TGFb response 34,35 and we demonstrated that Set7KO regulates key renal cell clusters, we stimulated human podocytes, glomerular endothelial cells, macrophage monocytes, and proximal tubular cells with PFI-2 before highglucose and TGFb1 stimulation (Figure 5F).We confirmed that PFI-2 inhibited methyltransferase activity in these cells by quantitative analysis of the ribosomal Rpl29me2 protein, 36 a reliable biomarker of Set7 activity (Figure 5G).Next, we assessed the expression of genes identified by scRNA-seq in diabetic Set7KO mice (as shown previously in Figure 2D).PFI-2 attenuated the activation of ANGPTL4, which is involved in TGFb1-dependent glucose homeostasis 37 (Figure 5H).Moreover, PFI-2 enhanced expression of the fatty-acid oxidation target gene ACOX1 in the glomerular endothelial cell and proximal tubule cell clusters.These results closely correspond with the transcriptome changes observed in the Set7KO mice.Because diabetic fibrosis is also characterized by excessive deposition of extracellular matrix proteins, 2 we show the activation of PDGFA was attenuated by PFI-2 in human podocyte and glomerular endothelial cells.Furthermore, PFI-2 reduced COL4A3 mRNA levels in podocytes despite elevated expression in human glomerular endothelial cells.The activation of the core genes associated with the OXPHOS and rRNA processing pathways-NDUFB2, MDH2, and NHP2-were also attenuated by PFI-2 in glomerular endothelial cells.These results suggest that pharmacological Set7 inhibition in human cells closely corresponded with the expression of genes implicated in the four major pathways in the kidney cortex of diabetic mice.

Set7 Regulation of IGFBP5
Hyperglycemia-induced oxidative stress plays a critical role in the development of DKD.Hyperglycemic glomerular endothelial cells treated with PFI-2 attenuate TGFb1-   FLAG tagged Set7 protein (FLAGSet7) was incubated with [H 3 ]-S-adenosyl-methionine and histone H3 peptide in the presence of PFI-2.Tritiated histone H3 peptide was measured by liquid scintillation.Activity is presented as percentage, 100% corresponding to FLAGSet7 with DMSO vehicle control.Data are presented as mean6SEM, n53.(E) Protein-peptide docking was performed using the HPEPDOCK server.The histone H3 peptide was modified to contain an unmethylated lysine residue (K4) and was docked to structures of SET7 in the absence (left and middle) and presence of PFI-2 (right) within the peptide-binding groove.(F) Schematic of the experimental design to assess the Set7 inhibitor PFI-2 in human cell lines under diabetic conditions.Cells were treated with PFI-2 for 24 hours before hyperglycemic stimulation with TGFb1 for 48 hours.(G) Set7 inhibition by PFI-2 in human cell lines.Demethylation of Rpl29 and Set7 levels were quantified by Li-COR Odyssey.Gapdh was used as loading control.(H) PFI-2 attenuates the expression of TGFb1-induced genes in human cell types.The expression of genes shown in brackets belong to the four core pathways: OXPHOS (MDH2, NDUFB2), PPARa induced cellular and mitochondrial ROS production (Figure 6A).If Set7 was responsible for EDMT in diabetic mice, then PFI-2 should prevent the expression of key EDMT genes.We stimulated human glomerular endothelial cells with TGFb1 to confirm reduced expression of the endothelial markers CDH5 (cadherin 5), PECAM (platelet endothelial cell adhesion molecule), and PLAT, while upregulating the expression of mesenchymal gene markers VIM (vimentin), EDN1 (endothelin-1), TAGLN (transgelin), and THBS2 (thrombospondin-2) (Figure 6B).PFI-2 significantly improved the expression of EDMT genes in TGFb1-induced hyperglycemic glomerular endothelial cells.These results suggest that pharmacological Set7 inhibition may prevent the endothelial-mesenchymal phenotype switch.While we previously showed that IGFBP5 activation in diabetic glomerular endothelial cells was attenuated by Set7KO in mouse (Figure 4D), we also show PFI2 prevented the expression of core EDMT genes in glomerular endothelial cells (Figure 6B).To validate Set7-mediated regulation of IGFBP5 gene expression, we performed short hairpin RNA Set7 knockdown (shSet7) in human glomerular endothelial cells.We confirmed knockdown of Set7 protein was associated with methyltransferase activity by RPL29me2 (Figure 6C).Furthermore, shSet7 significantly reduced IGFBP5 but not PLAT (Figure 6D).TGFb1 and high glucose did not influence IGFBP5 expression in shSet7 glomerular endothelial cells (Supplemental Figure 4), suggesting that Set7 activity was required for IGFBP5 expression.Because Set7 is known to methylate histone H3 lysine K4, 12,13 we investigated mono-and di-methylation of H3K4 at the promoter region (R1) and enhancer regions (R2 and R3) of the IGFBP5 gene directly by chromatin immunoprecipitation (Figure 6E).We demonstrated that elevated H3K4 methylation at the IGFBP5 R2 site was significantly reduced by PFI-2 in hyperglycemic glomerular endothelial cells stimulated with TGFb1 (Figure 6F).Moreover, we showed reduced H3K4 methylation of the IGFBP5 gene in shSet7 glomerular endothelial cells (Figure 6G).We also investigated chromatin accessibility at the IGFBP5 promoter/enhancer region from publicly available single cell assay for transposase-accessible chromatin using sequencing (scATAC-seq) datasets for the human kidney (Supplemental Figure 5).The IGFBP5 R2 site intersects with overlapping chromatin accessibility peaks between two independent scATAC-seq datasets generated from the human kidney.We observed altered chromatin accessibility in glomerular endothelium between control and CKD groups for the IGFBP5 gene at R1 and R2 sites.Taken together, these results suggest that Set7 regulates H3K4 methylation at the enhancer regions of the IGFBP5 gene in human glomerular endothelial cells.

Preventing Diabetic Phenotypic Switch in Mouse and Human Glomerular Endothelial Cells
To understand the influence of Set7 on the transcriptome, we profiled human glomerular endothelial cells using the small molecule inhibitor PFI-2 or by the viral shSet7 knockdown using mRNA sequencing (Supplemental Figure 6A).Genes upregulated in expression by high glucose and TGFb1 were attenuated by targeting Set7 activity (Supplemental Figure 6B) involving the EMO and IGFBP signaling pathways (Supplemental Figure 6C).Consistent with our previous studies in diabetic mice, we confirm that IGFBP5 expression was dependent on Set7-mediated regulation in glomerular endothelial cells.
To study the generality that the Set7 methyltransferase could regulate the phenotype switch, we compared the transcriptome changes in mouse and human glomerular endothelial cells (Figure 7, A and C).We first examined pathways that were directly regulated by diabetes in glomerular endothelial cells and reversed by Set7KO in mice.Transcriptome changes assessed by scRNA-seq identified the regulation of the IGFBP pathway, including TGFb receptor signaling, extracellular matrix organization (EMO), smooth muscle contraction, and IL-12 signaling.Enrichment scores for pathways regulated by diabetes and prevented by Set7KO, including statistical significance, are shown in the left panel of Figure 7B.RNA-seq shows the transcriptional expression index (TEI) of genes shift toward recovery in Set7KD diabetic mice by preventing EDMT.Endothelial Set7 is a critical hyperglycemic sensor that regulates gene expression in primary human vascular cells.To determine whether the phenotypic switch observed in diabetic mice was consistent with a model in which Set7 was coupled with IGFBP5 control, we assessed the transcriptome of hyperglycemic human glomerular endothelial cells stimulated by TGF (Figure 7C).We observed a parallel association of EDMT pathways in human Set7KD glomerular endothelial cells that was consistent with diabetic Set7KO mice preventing the phenotype switch (Figure 7D).39][40][41] Consistent with these observations, we also report that pharmacological inhibition of Set7 by PFI-2 can regulate DKD pathways associated with the phenotype switch.These results suggested that targeting IGFBP5 regulation by Set7 prevents diabetes-induced EDMT in mouse and human glomerular endothelial cells.

Discussion
In the context of DKD, hyperglycemia plays a significant role in the progression of EDMT; however, the molecular mechanisms that regulate this phenotype switch remain poorly understood.The role of the glucose sensor Set7 is largely unknown.We have combined mouse genetics, Figure 5. Continued.(ACOX1, ANGPTL4), rRNA (NHP2), and EMO (PDGFA, COL4A3).Gene expression was assessed by qRT-PCR.The assessment of ANGPTL4 expression was used as a TGFb1 response gene.n56 per group.Data are represented as mean6SEM.*P , 0.05, **P , 0.01, ***P , 0.001 versus normal glucose control; #P , 0.05, ##P , 0.01, ###P , 0.001 versus diabetic condition (HG1TGFb1).HG, high glucose; RMSD, root-mean-square deviation; SAH, S-adenosyl homocysteine; SAM, S-adenosyl methionine.single-cell transcriptome, and biochemical approaches to show that the Set7 methyltransferase advances the phenotypic switch in diabetic kidney injury.Renal fibrosis is a pathological feature of excessive deposition of extracellular matrix in the diabetic kidney, and our studies show that Set7 regulates the expression of genes critical to fibrosis, including endothelial barrier function and mitochondrial ROS production (Figure 8).Glomerular endothelial cells express the highest Set7 levels, and this corresponds with EDMT that is implicated in microalbuminuria generation in DKD. 7Our studies also show that the mesenchymal gene markers upregulated by diabetes in mice were attenuated in human glomerular endothelial cells by PFI-2.These results are novel for several reasons.First, albuminuria and glomerular structure were dramatically improved in the diabetic Set7KO mice.While albuminuria is a clinical manifestation of glomerular filtration barrier destruction, 5 the molecular mechanisms that drive this process in diabetic kidney injury remain poorly defined.Our studies show that Set7 is predominantly expressed in podocyte and glomerular endothelial cell types that are considered to be functionally important in glomerular filtration.Furthermore, scRNA-seq analyses revealed the glomerular endothelial cell populations were dynamically altered in the diabetic Set7KO kidney cortex.Nowhere is this more evident than the effect on core pathways, such as extracellular matrix organization, IL-12 signaling, and TGFb receptor signaling observed in mouse and human glomerular endothelial cells.Previous studies have shown a context-dependent role for vascular Set7 in proinflammatory gene activation, 10,11,13 and the experimental results presented here suggest that diabetic kidney injury associated with hyperglycemia regulate endothelial inflammation and progressive damage of the glomerular filtration barrier.

A
A key challenge in endothelial biology is explaining how glomerular cells acquire distinct mesenchymal fates in response to hyperglycemic signaling cues.For example, extracellular glucose does not carry intracellular instructions that mediate epigenetic change unless there are methyltransferase sensors that perceive hyperglycemia to regulate gradient-dependent gene expression.Endothelial Set7 is a glucose-dependent sensor that regulates histone code changes and the activation of vascular gene expression in response to transient and chronic hyperglycemia. 12Furthermore, hyperglycemia promotes nuclear Set7 localization to activate proinflammatory gene expression that is distinguished by histone methylation.This could explain the nucleosomal H3K4 preference we observe of Set7 methylation associated with IGFBP5 chromatin to directly regulate gene expression in response to hyperglycemia.In this study, we show Set7 was correlated in mouse and human cell studies using genetic and pharmacological approaches that attenuated the mesenchymal markers of expression influencing ROS overproduction.This is consistent with the role of Set7 in mesenchymal differentiation 25 and does not exclude the immunometabolic changes associated with trained immunity. 36Set7 enzyme methylates not only lysine 4 residue of histone H3, which is implicated in gene-activating transcriptional events, but can also methylate lysine residues of nonhistone proteins.7][18] We have previously demonstrated that Set7 controls the expression of mesenchymal markers by methylation of histone H3 and serum response factor transcription factor. 25These observations support a role for Set7 regulation of EDMT in glomerular endothelial cells in response to diabetic stimuli.Consistent with its role as a hyperglycemic sensor, we show that the expression of vascular cell proliferation and differentiation markers-PDGFA, PDGFB, and BGN 42,43 were significantly reduced by PFI-2.
In diabetic glomerular endothelial cells, we have shown that Igfbp5 is a primary Set7 gene target in mice using scRNA-seq analyses.IGFBP5 is an important secretory protein that is related to inflammation and fibrosis in DKD 44 and a kidney-specific endothelial cell marker. 30,31,45While upregulation of IGFBP5 has been associated with DKD, 32,33,46 the mechanism of regulation remains poorly understood.Recent studies have shown that IGFBP5 activates inflammation signaling directly by influencing endothelial glycolysis in diabetic mice. 478][49] We have shown that Set7 plays an important role for IGFBP5 human gene regulation mediated by H3K4me1/2 on the enhancer region in glomerular endothelial cells.Because IGFBP5 expression was not elevated by TGFb1 in Set7KD glomerular endothelial cells, we hypothesize that lysine methyltransferase is required for transcriptional expression of the gene.Furthermore, the IGFBP5 protein can translocate to the nucleus to  exert its activity in an IGF-independent manner. 50The subcellular localization of Set7 is also induced by high glucose levels and associated with transcriptional changes of the inflammatory and extracellular matrix genes pathways. 12Because Set7KO reduced collagen IV expression in peritubular capillaries, it is plausible that Set7 could play a role in peritubular capillary endothelial cells in the diabetic setting.Future studies examining protein interaction and endothelial cell types may provide new insights of the phenotype switch in early development of diabetes-related complications.
The experimental results in this article suggest the dynamic transcriptional changes by PFI-2 attenuated EMO, IL-12 signaling, and EMT pathways, as well as IGFBP5 in human glomerular endothelial cells, closely correspond with the pathways improved by Set7KO in diabetic mice.While progress with single-cell technology has meant that the capacity to uncover the physiological targets in the kidney is now possible, a limitation of our preclinical studies is applied translation.Although PFI-2 is not licensed for diabetic complications, including DKD, the findings provide a proof of concept to translate these studies and the continued development of novel Set7 inhibitors. 51aken together, our findings have provided new insights into Set7 regulating EDMT and IGFBP5.Our studies highlight the prospect of targeting Set7 lysine methyltransferase and the phenotype switch in diabetic kidney injury.

Disclosures
Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/JSN/E623.

Figure
Figure2.Influence of Set7 deletion in diabetic kidney transcriptome revealed by scRNA-seq.(A) scRNA-seq cell clusters in the diabetic kidney were visualized by t-distributed stochastic neighbor embedding (tSNE) after analysis using Seurat and R package.(B) Composition of the cell clusters in each experimental group.Significant changes in cell composition between experimental groups (ANOVA) included the proximal tubule (P , 0.05) and GEN (P 5 0.066) clusters, as calculated using the propeller method.(C) GSEA identifies four reactome pathways that belong to OXPHOS, rRNA processing, EMO, and PPARa activation.These pathways were significantly changed by Set7 in diabetes.(D) Distribution of diabetic Set7-dependent gene expression was visualized by tSNE.Color gradient indicates the number of transcripts detected per cell.BL, B-lymphocyte; CDI, collecting duct intercalated; DCT, distal convoluted tubule; EMO, extracellular matrix organization; ER, erythroid; GEN, glomerular endothelial; GSEA, gene set enrichment analysis; LH, loop of Henle; MAC, macrophage; MG, mesangial; MSC, mesenchymal; NK, natural killer; OXPHOS, respiratory electron transport; PDC, podocyte; PPARa, peroxisome proliferator-activated receptor alpha; PTC, proximal tubule cell; rRNA, ribosomal RNA; TL, T-lymphocyte.

Figure 3 .
Figure 3. Set7 regulates the GEN transcriptome.(A) Increased Set7 expression in the kidney cortex derived from diabetic mice.Expression of Set7 mRNA in the kidney cortex was assessed by qRT-PCR.n55 per group.Data are represented as mean6SEM.***P , 0.001 versus control Set7WT mice.(B) Cell distribution of Set7 expression in the kidney cortex by scRNA-seq.(C) Set7KO regulates the expression of GEN and mesenchymal markers in the diabetic kidney.Color gradient indicates significant (P , 0.05) log fold changes for each gene: red (elevated),

Figure 3 .
Figure 3. Continued.blue (reduced), and white (no significant change; P .0.05).(D) The top 15 reactome pathways that are significantly changed by diabetes (control versus diabetic Set7WT) and Set7 deletion in diabetes (Diab Set7KO; diabetic Set7WT versus diabetic Set7KO) are illustrated in GEN and mesenchymal cell clusters using GSEA.qRT-PCR, quantitative real-time reversetranscription PCR.

Figure 4 .Figure 5 .
Figure 4. Set7-dependent regulation of the GEN transcriptome in diabetic mouse kidney.(A) Identification of the novel GEN cell population in diabetic Set7KO mouse kidney.Reclustering of the GEN and mesenchymal populations in our scRNA-seq data.(B) Top 40 genes significantly changed in the GEN (Igfbp5 high) and GEN (Igfbp5 low) clusters.(C) Distribution of Igfbp5 and Plat gene markers in each experimental group is illustrated.(D) Composition of the cell clusters in each experimental group.Significant changes in cell composition between experimental groups (ANOVA) included GEN (Igfbp5 high, P , 0.05) and GEN (Igfbp5 low, P , 0.01) and calculated using the propeller method.(E) Reactome pathway exchanges observed in the GEN (Igfbp5 high) and GEN (Igfbp5 low) cell clusters.EMT, epithelial-tomesenchymal transition; IGFBP5, insulin growth factor binding protein 5; UMAP, uniform manifold projection.

Igfbp5BFigure 7 .
Figure 7. Targeting Set7 in diabetic GEN transcriptome.Set7-mediated endothelial transcriptomes in the GEN population derived from (A) diabetic mice and (B) human GEN cells stimulated with TGFb1 and HG (TGF/HG) conditions.(C) scRNA-seq analysis demonstrates the transcriptional expression index (TEI) of Set7-dependent diabetic pathways in diabetic mice include the following: IGFBPs regulation, TGFb receptor signaling, Rho GTPases-activated ROCKs, EMO, smooth muscle contraction, JAK-STAT signaling after IL-12 stimulation, cooperation of prefoldin and TriC/CCT in actin and tubulin folding, and NR1H2-and NR1H3-mediated signaling pathways are identified.*P , 0.05,

Figure 8 .
Figure 8. Schematic model of Set7-mediated regulation of phenotypic switch.In diabetic GEN cells, Set7 activates IGFBP5 expression by regulating H3K4me1 and H3K4me2.The mesenchymal-like cells are hypothesized to transition and activate ROS, EMO, JAK-STAT, and EMT pathways.Genetic knockout and pharmacological inhibition of Set7 attenuates ROS production and improves fibrosis and proinflammatory pathways and the expression of EDMT genes implicated in DKD.