Krϋppel‐like factor 15 suppresses renal glomerular mesangial cell proliferation via enhancing P53 SUMO1 conjugation

Abstract Mesangial cell (MC) proliferation is a key pathological feature in a number of common human renal diseases, including mesangial proliferative nephritis and diabetic nephropathies. Knowledge of MC responses to pathological stimuli is crucial to the understanding of these disease processes. We previously determined that Krϋppel‐like factor 15 (KLF15), a kidney‐enriched zinc‐finger transcription factor, was required for inhibition of MC proliferation. In the present study, we investigated the direct target gene and the underlying mechanism by which KLF15 regulated mesangial proliferation. First, we screened small ubiquitin‐related modifier 1 (SUMO1) as the direct transcriptional target of KLF15 and validated this finding with ChIP‐PCR and luciferase assays. Furthermore, we demonstrated that overexpressing KLF15 or SUMO1 enhanced the stability of P53, which blocked the cell cycle of human renal MCs (HRMCs) and therefore abolished cell proliferation. Conversely, knockdown of SUMO1 in HRMCs, even those overexpressed with KLF15, could not inhibit HRMC proliferation rates and increase SUMOylation of P53. Finally, the results showed that the levels of SUMOylated P53 in the kidney cortices of anti‐Thy 1 model rats were decreased during proliferation periods. These findings reveal the critical mechanism by which KLF15 targets SUMO1 to mediate the proliferation of MCs.

sclerosis and interstitial fibrosis; further aggravate renal damage; and cause progression to irreversible end-stage renal disease. [5][6][7] Knowledge of the MC response to pathological stimuli is crucial for the development of treatments to reduce the kidney damage caused by abnormal proliferation of MCs and delay progression into endstage renal disease.
The Krüppel-like factors (KLFs) are a group of zinc-finger DNAbinding transcription factors. Increasing evidence indicates that this family is involved in a variety of cellular processes, such as cell differentiation, cardiac remodelling, haematopoiesis and stem cell fate determination. [8][9][10] The KLF family consists of eighteen members, among which KLF15 is widely distributed in the kidney, pancreas, heart, liver and skeletal muscles. Previous studies have demonstrated that KLF15 is a key transcriptional regulator in diverse physiological processes, including gluconeogenesis, immune responses and adipocyte differentiation. [11][12][13][14] Our previous study has demonstrated that KLF15 is required for inhibiting the proliferation of MCs. 15 This study aimed to clarify the direct target gene and the downstream mechanism by which KLF15 regulates mesangial proliferation.

| Anti-Thy1 nephritis model
Male Wistar rats (Beijing Vital River Laboratory Animal Technology Co., Ltd.) weighing between 200 and 220 g were randomly allocated to the control and anti-Thy1 groups. All rats were housed in an animal care facility under a light/dark cycle of 12/12 hours with free access to food and water. Welfare-related assessments and interventions were performed throughout the experiment. In total, 28 rats were injected with a mouse anti-Thy1 monoclonal antibody, as described previously, 16 and seven were injected with saline (controls). Glomeruli were purified from the renal cortex tissue using a sieving method. Sera were collected for measurements of serum urea nitrogen and creatinine levels. Anti-Thy1 model rats were killed on days 0, 3, 5, 7 and 10, and their glomeruli were harvested. All animal welfare and experimental procedures were performed in strict accordance with the Guide for the Care and Use of Laboratory Animals (USA National Research Council, 1996).

| Immunohistochemistry (IHC)
Paraffin-embedded kidney tissue was sectioned at 4 μm thickness, and the sections were stained for IHC. The slices were incubated with 3% H 2 O 2 to block endogenous peroxidase following deparaffinization. After antigen retrieval and blocking, the slices were incubated with anti-KLF15 (ab81604, Abcam), anti-small ubiquitin-related modifier 1 (SUMO1, ab32058, Abcam) and anti-PCNA (ab92552, Abcam) antibodies at 4°C. The slices were then incubated with a secondary antibody labelled with horseradish peroxidase at 37°C for 30 minutes, and the immunohistochemical reaction was observed according to the manufacturer's instructions with 3,3'-diaminobenzidine (DAB, ORIGENE, CN) as the chromogenic agent. Subsequently, the slices were counterstained with haematoxylin-eosin and imaged with a light microscope (Olympus).
Immunohistochemical staining was evaluated with Image J software according to the average optical density (AOD) value.

| Chromatin immunoprecipitation (ChIP) with parallel sequencing (ChIP-Seq)
ChIP was performed using a Simple ChIP Plus Enzymatic Chromatin IP Kit (Magnetic Beads; Cell Signaling Technology) according to the manufacturer's protocol. Briefly, HRMCs were overexpressed with either KLF15 or scramble at a cell density of 2.0 × 10 5 cells/ mL and cross-linked with 1.7% formaldehyde in PBS for 5 minutes at room temperature (RT). Cross-linked cells were lysed to obtain the nuclear fraction, which was sonicated in 1 × ChIP buffer. The lysates were clarified by centrifugation at 9300 × g for 10 minutes at 4°C. Chromatin samples were incubated with either an anti-KLF15 antibody (ab81604, Abcam) or an anti-IgG antibody (background control) overnight at 4°C. The antibody-bound complexes were captured by incubation with protein G magnetic beads.
ChIP-Seq libraries for sequencing were prepared using a TruSeq ChIP Sample Prep Kit (Illumina). The libraries were subjected to parallel sequencing with a HiSeq2500 sequencer (Illumina) using the single-end 50 bp sequencing length protocol. Next-generation sequencing (NGS) raw data were converted into FASTQ files using CASAVA software (version 1.8.2), and each data set was aligned to the human reference genome (UCSC hg19) using the Burrows-Wheeler Aligner (version 0.7.12). 17 ChIP-Seq peak calling was performed using the MACS2 program (version 2.0.1) with the default parameters but a -q value of 0.05, with the input data for subtraction. 18 Peak comparison with the scramble control (the parental cells) and overexpression samples was performed with Diffbind, an R package, using the DeSeq2 algorithm, and a false discovery rate of 0.1 was considered to indicate significance. 19 Superenhancers were identified from the set of peaks detected in DMSO-treated HRMCs with the superenhancer software ROSE. 20 The ChIP-Seq peaks were annotated with the R package ChIPpeakAnno, and promoters were defined as KLF15-enriched regions within 1 kb of the transcription start site. 21 The gene with the transcription start site closest to the centre of each superenhancer was defined as the target gene of that superenhancer.

| ChIP-quantitative real-time PCR (ChIP-PCR)
The ChIP-PCR procedure was similar to the procedure described above. ChIP DNA from anti-KLF15 antibody-treated cells was used to detect the association between KLF15 and SUMO1.
DNA from anti-IgG antibody-treated cells served as the control. Purified DNA was used for analysis of the SUMO1 proximal promoter region by real-time PCR on an ABI PRISM 7600 using SYBR GreenER qPCR Supermix. The SUMO1 primers were as follows: forward: 5′-AGCAGCGGTCTTTAGCATCA-3′; reverse: 5′-TCGGTTAACCAGCACCCTTG-3′. The relative amplification of the promoter sequence of the gene was calculated using the 2 −ΔΔCT method, and normalization was performed against a 1:100 dilution of the input DNA.

| Cell labelling and stable isotope labelling by amino acids in cell culture (SILAC) analysis
Cells at 15% confluency were seeded in the appropriate complete medium (for HRMCs: MsCM). The labelling and SILAC procedures have been described in previous research. 22 Briefly, after being labelled with light ( 13 C 6 -labelled) or heavy ( 15 N 2 -labelled) lysine and light ( 13 C 6 -labelled) or heavy ( 15 N 4 -labelled) arginine, cells were transfected with a KLF15 overexpression plasmid or a control plasmid.
After treatment, the cells were trypsinized and counted to obtain a cell pellet of 2 × 10 7 cells/condition, and the cells were subjected to SILAC analysis using mass spectrometry. The heavy and light cell pellets were lysed in radioimmunoprecipitation buffer spiked with protease and phosphatase inhibitors using short 15 second sonication bursts. The lysates were centrifuged at 14,000 rpm for 20 minutes. After centrifugation, the supernatants were collected, and the protein concentration was measured using a Hitachi L-8900 amino acid analyser. Aliquots of 200 μg of protein were prepared from the samples, combined and precipitated using a methanol-chloroform precipitation method. The protein pellets were resuspended in

| RNA preparation, cDNA synthesis and RT-qPCR analysis
Total RNA was extracted using TRIzol (Invitrogen) and purified using a RNeasy Mini Kit (Qiagen) according to the manufacturer's instructions. cDNA was generated using a ProtoScript first-strand cDNA synthesis kit (New England Biolabs). Quantitative PCR was performed using Power SYBR Green Master Mix (Life Technologies).
The oligonucleotide sequences used for RT-qPCR are provided in Table 1.

| Western blot analysis and immunoprecipitation (IP)
Cells were lysed using RIPA buffer (Thermo Scientific), and the total protein in the supernatants was quantified using a BCA protein assay kit (Thermo Scientific). Western blot analysis and IP were performed as described previously. 23,24 Anti-KLF15 antibody (AV32587, Merk), anti-SUMO1 antibody(S8070, Merk) and antiβ-actin (A5316, Sigma) were used for Western blot analysis. Anti-P53 antibody (P5813, Merk) was used for IP analysis.

| Bioinformatics analysis
Candidates from ChIP-Seq or SILAC showed at least twofold en-

| Statistical analysis
All experiments were performed in triplicate, and the results are expressed as the means ± SEs. All data were analysed using SPSS software (ver. 20.0; SPSS Inc) and compared using Student's t test or one-way ANOVA. A P-value <.05 was considered to reflect statistical significance.

| Screening of KLF15-binding genes through ChIP-Seq in primary renal glomerular MCs
To identify the direct binding partner genes of KLF15, we performed ChIP-Seq analysis and ultimately screened 2478 genes. GO analysis of these genes through the tool at the website www.unipr ot.org ( Figure 1A,B) revealed molecular function terms associated with 1941 genes, cellular component terms related to 1662 genes and biological process terms related to 1766 genes. Since our aim was to find out how KLF15 affects MCs, we focused on cell process-related genes. Among the 1315 cell process-related genes, 74 genes were found to participate in cell cycle processes; specifically, these genes participated in the mitotic cell cycle, the cell cycle phase transition and other processes ( Figure 1C,D). Furthermore, we analysed the genes involved in growth and found that they were related to developmental growth, cell growth and other types of growth ( Figure 1E).

| Screening of differentially regulated genes in KLF15-overexpressing renal glomerular MCs using SILAC and LC/MS
SILAC and LC/MS analysis of HRMCs overexpressing KLF15 compared to parental cells led to the identification of 1357 proteins.
We used the DAVID and IPA to acquire the GO domains and enriched pathways of the quantified proteins identified by SILAC.
Interestingly, many proteins biological function-related terms were among the top 30 significantly enriched GO terms, including the regulation of cellular amino acid metabolic process, proteasome complex, protein binding, and transcription and translation terms, all of which are closely related to ubiquitination (Figure 2A). Figure 2B shows the top 30 significantly enriched pathway terms. Several biological process terms, including the proteasome and neurodegenerative disease terms, were also related to ubiquitination.

| Bioinformatics analysis of KLF15-binding genes that are potentially associated with renal glomerular MC proliferation
To explore the KLF15-binding genes that are potentially associated with renal glomerular MC proliferation, we performed bioinformatics analysis of the ChIP-Seq data and the SILAC-LC/MS data. Fifty-two genes were screened (Table 2  Increasing amounts of evidence have indicated that HG is one of the major factors inducing the development of diabetic nephropathy, and it promotes MC proliferation and increased matrix synthesis in vitro. 25 A previous study has shown that PDGF-BB is essential for MC proliferation preceding the development of glomerulosclerosis in experimental glomerulonephritis. 26 After confirming that MCs were stimulated by HG or PDGF-BB in vitro, we detected changes in the expression of KLF15 and SUMO1 at both the RNA and protein levels and found that these molecules had similar trends ( Figure 2D-I). Finally, we determined SUMO1 to be the target gene of KLF15.

| KLF15 up-regulates SUMO-1 gene expression by binding to a 16 bp CACCCA-SUMO-1 promoter region and a C+A-rich motif
We used the MEME suite (http://meme-suite.org/tools/ meme) to characterize the KLF15-binding motifs of the 52 screened genes from the KLF15 ChIP-Seq data and identified the KLF-binding motif ( Figure 3A). In addition, we further analysed more than one thousand bases upstream of the SUMO1 promoter and found that KLF15 could recognize and bind to a 16 bp sequence (nucleotides −205 ~ −199) including -CACCCA-( Figure 3B). ChIP-PCR confirmed that KLF15 bound to the SUMO1 promoter, and the results showed significantly higher expression of SUMO1 in the anti-KLF15 antibody group than in the background control group ( Figure 3C). Furthermore, we performed a dual-luciferase reporter assay to confirm the targeting relationship between SUMO1 and KLF15. The wild-type SUMO1 promoter group showed higher luciferase activity than the pGL3 vector and mutant groups in HRMCs, and the activity was significantly increased by overexpression of KLF15. The enhancement in luciferase activity was reversed by transfection with a plasmid expressing the mutant promoter region ( Figure 3D). Taken together, these data suggest that SUMO1 is a direct transcriptional target of KLF15. To explore the downstream signalling molecules directly conjugated by SUMO1, we performed network analysis of SUMO1 using the IPA database, and the data showed that SUMO1 could directly conjugate to P53, APP, JUN and AKT, among other proteins ( Figure 4A-C). We initially selected P53 as a downstream molecule of SUMO1

| Screening of P53 as the
because it is a well-known protein that is closely related to cell proliferation. 28,29 Furthermore, we used SUMOsp software to predict the possible SUMOylation sequences of P53 in various species and found that P53 had the SUMOylation sequence Ψ-K-x-D/E ( Figure 4D). To determine whether P53 is indeed modified by SUMOylation, we transiently transfected HRMCs with a SUMO1 overexpression plasmid or SUMO1 siRNA. Both the IP and Western blot results revealed trends in expression changes of SUMO1-P53 and P53 that were consistent with the changes in SUMO1 ( Figure 4E-J). These data indicate that P53 is a SUMOylation substrate of SUMO1.

| KLF15 inhibits MC proliferation by promoting SUMO1 expression and P53 SUMOylation
To establish the regulation of P53 SUMOylation and MC prolifera-  Figure 6G,H). We therefore conclude that KLF15 suppresses MC proliferation by enhancing P53 SUMOylation.

| Global SUMO1 and P53 expression in glomerular MCs is negatively correlated with MC proliferation in rat Thy-1 nephritis
Anti-Thy1 nephritis is a classical model of self-limited mesangial proliferative glomerulonephritis with a proliferative phase and a recovery phase. We injected a Thy1 antibody into Wistar rats to create this model. Both serum urea nitrogen and creatinine have no significant change between control rats and the model rats ( Figure S1).
Marked mesangial proliferation and ECM accumulation were observed during the proliferative phase (days 5 and 7) in the model rats, and the number of MCs decreased during the recovery phase on day 10 ( Figure 7A). We also detected the expression changes in the cell proliferation marker PCNA by IHC ( Figure 7A,B). PCNA levels were increased on day 5, peaked on day 7 and were decreased on day 10 ( Figure 7F). Western blot analysis showed that the protein expression of P53, SUMO1 and KLF15 in isolated glomeruli was lower in  the model groups than in the control group ( Figure 7C,D), consistent with the immunohistochemical results ( Figure 7E,F). These results indicate that the abnormal proliferation of MCs in anti-Thy1 model rats is related to the low-level expression of KLF15 and SUMO1.
Interfering with the expression of these molecules is expected to alleviate the pathological phenotype in rats.

| D ISCUSS I ON
The whole-genome analysis with higher resolution, higher detection sensitivity and lower sample size demand. 35 We used ChIP to obtain the DNA fragments directly bound by KLF15, and after comparison and analysis with GenBank, we screened 2478 possible target genes.
Through GO and pathway analyses, we identified many target genes involved in cell cycle and proliferation processes.
ChIP-Seq experiments require PCR for amplification of the detection signal, and some degree of bias during the amplification process is inevitable. In addition, ChIP-Seq obtains only the genes that In addition to ubiquitin, increasing numbers of UBL proteins, 36,37 including SUMO, 38 neural precursor cell-expressed, developmen- To identify the substrates of SUMO1, we performed network analysis of SUMO1 using the IPA database and SUMOsp software.

ACK N OWLED G EM ENTS
Thanks to Dr Fei Peng for her contribution in the revision of the manuscript.

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.