Inhibition of the processing of miR-25 by HIPK2-Phosphorylated-MeCP2 induces NOX4 in early diabetic nephropathy

Phosphorylated methyl-CpG binding protein2 (p-MeCP2) suppresses the processing of several microRNAs (miRNAs). Homeo-domain interacting protein kinase2 (HIPK2) phosphorylates MeCP2, a known transcriptional repressor. However, it is not known if MeCP2 and HIPK2 are involved in processing of miRNAs implicated in diabetic nephropathy. p-MeCP2 and HIPK2 levels were significantly increased, but Seven in Absentia Homolog1 (SIAH1), which mediates proteasomal degradation of HIPK2, was decreased in the glomeruli of streptozotocin injected diabetic mice. Among several miRNAs, miR-25 and its precursor were significantly decreased in diabetic mice, whereas primary miR-25 levels were significantly increased. NADPH oxidase4 (NOX4), a target of miR-25, was significantly increased in diabetic mice. Protein levels of p-MeCP2, HIPK2, and NOX4 were increased in high glucose (HG)- or TGF-β-treated mouse glomerular mesangial cells (MMCs). miR-25 (primary, precursor, and mature) and mRNA levels of genes indicated in the in vivo study showed similar trends of regulation in MMCs treated with HG or TGF-β. The HG- or TGF-β-induced upregulation of p-MeCP2, NOX4 and primary miR-25, but downregulation of precursor and mature miR-25, were attenuated by Hipk2 siRNA. These results demonstrate a novel role for the SIAH1/HIPK2/MeCP2 axis in suppressing miR-25 processing and thereby upregulating NOX4 in early diabetic nephropathy.

Methyl-CpG binding protein 2 (MeCP2) is a transcriptional repressor by binding to methylated DNA and recruiting histone deacetylase complex proteins [23][24][25][26] . Interestingly, MeCP2 also regulates gene expression by posttranslational mechanisms involving suppression of nuclear miRNA processing 27,28 . They found that phosphorylated MeCP2 (p-MeCP2) binds directly to DiGeorge syndrome critical region 8 (DGCR8), a critical component of the nuclear miRNA-processing machinery [29][30][31][32] and interferes with the assembly of the Drosha and DGCR8 complex. On the other hand, homeo-domain interacting protein kinase 2 (HIPK2), which is a conserved serine/ threonine nuclear kinase and controls gene expression by phosphorylating transcription factors has been shown to phosphorylate MeCP2 at Ser 80, and p-MeCP2 mediated by HIPK2 was suggested to contribute to apoptosis 28,33 . HIPK2 plays a role in kidney fibrosis in mice with human immunodeficiency virus (HIV), and HIPK2 expression is higher not only in kidneys of HIV transgenic mice and patients with HIV associated nephropathy, but also in kidneys of patients with focal segmental glomerulosclerosis (FSGS), DN and immunoglobulin A nephropathy (IgA nephropathy) 34 . We previously reported that let-7 miRNA family members are downregulated under diabetic conditions through changes in lin-28b which mediates the processing of let-7 35 . However, it is not known if MeCP2 and HIPK2 are involved in processing and expression of candidate miRNAs that are downregulated in DN.
In this study, we evaluated whether the downregulation of key protective miRNAs, such as miR-25, under diabetic conditions in the kidney are mediated by the mis-regulation of factors mediating the biogenesis and processing of these miRNAs. Specifically, we assessed whether changes in HIPK2 and p-MeCP2 are observed in glomeruli from diabetic mice, and can affect the processing of miR-25 in renal mesangial cells under diabetic conditions. Results p-MeCP2, HIPK2, and NOX4 expressions are upregulated in glomeruli of diabetic mice. We first examined whether the expression of p-MeCP2, HIPK2, and NOX4 are altered in the glomeruli of diabetic versus control mice. Samples were obtained four weeks after the onset of diabetes in streptozotocin (STZ) or vehicle injected C57BL6 mice. Immunohistochemistry showed that p-MeCP2-, HIPK2-, or NOX4-stained cells were significantly increased in the glomerulus from diabetic mice compared with control non-diabetic mice ( Fig. 1A to D). Glomerular size was significantly larger in these diabetic mice versus control mice indicating increased glomerular hypertrophy (Fig. 1E). Moreover, western blotting also showed that the protein expressions of p-& total (t)-MeCP2, HIPK2, and NOX4 were also higher in the diabetic mice compared to controls (Fig. 1F). These results suggest that glomerular p-MeCP2, HIPK2, and NOX4 are related to the pathology of DN.
Nox4 mRNA expression is increased, but Siah1 mRNA expression is decreased in diabetic conditions. We found that mRNA levels of Nox4 were significantly increased in the glomeruli of the diabetic mice compared to control, but Seven in Absentia Homolog 1 (Siah1) mRNA expression was significantly decreased in diabetes compared to control ( Fig. 2A and B). However, there were no significant differences in the mRNA expressions of Mecp2 and Hipk2 between the two groups ( Fig. 2C and D), suggesting that the expression of these proteins may be regulated by post-translational mechanisms.
Primary miR-25 expression is increased, while precursor and mature miR-25 expressions are decreased in diabetic conditions. The levels of several miRNAs have been reported to be decreased in DN 15,[35][36][37][38][39][40][41][42][43][44][45] . Among these miRNAs and also decreased miRNAs identified by small RNA sequencing in glomeruli from STZ diabetic mice 22 , we confirmed that miR-25 and miR-93 expressions were significantly downregulated in TGF-β treated mouse mesangial cells (MMCs), but the expression of only miR-25 was consistently and significantly decreased even in HG-treated MMCs relative to control. Since we previously showed that let-7 family (lin28-mediated) 35 and miR-130b 44 were downregulated under diabetic conditions by mechanisms not involving p-MeCP2, we focused on miR-25 in this study. The expression of primary miR-25 was significantly upregulated in the glomeruli of diabetic mice compared with that in the control (Fig. 2E), whereas precursor and mature miR-25 levels were significantly downregulated in the diabetic mice compared to control ( Fig. 2F and G). p-MeCP2, HIPK2, and NOX4 expressions are upregulated, but SIAH1 expression is downregulated in HG and TGF-β treated MMCs. To further clarify whether the glomerular expressions of p-MeCP2, HIPK2, and NOX4 are regulated by diabetic conditions in vitro, MMCs were cultured in either normal glucose (NG, 5.5 mM) or high glucose (HG, 25 mM) or equimolar osmotic control mannitol for 72 hrs. The mRNA levels of Mecp2 and Hipk2 (Fig. 3A to C) and protein expressions of p-MeCP2, HIPK2, and NOX4 (Fig. 3E,F,H and I) were significantly upregulated in the HG treated MMCs compared to those in the control group, while mRNA expression of Siah1 was significantly decreased in the HG treated MMCs compared with controls (Fig. 3D). Protein levels of t-MeCP2 were not significantly different between the three groups ( Fig. 3E and G).
We found that these factors were also similarly regulated by TGF-β . Thus, in TGF-β (10 ng/ml) treated MMCs, the mRNA levels of Mecp2 and Hipk2 were significantly increased by 6 hr after treatment, and returned below normal by 24 hr compared with the control group ( Fig. 4A and B). mRNA expression of Nox4 was significantly upregulated in the 24 hr TGF-β treated MMCs compared to control (Fig. 4C). On the other hand, Siah1 mRNA levels were significantly decreased in the 6 & 24 hr TGF-β treated MMCs compared to control (Fig. 4D). Western blots showed that the protein levels ( Fig. 4E,F,H and I) of p-MeCP2, HIPK2, and NOX4 were also significantly increased in the 24 hr TGF-β treated MMCs relative to those in the control group, whereas there was no significant difference in t-MeCP2 protein expression among the groups (Fig. 4G). Taken together, p-MeCP2, HIPK2, and NOX4 expressions are increased, but mRNA expression of Siah1 is decreased in mesangial cells under diabetic conditions in vivo and in vitro. Immunohistochemistry showed that p-MeCP2-, HIPK2-, or NOX4-stained cells (black arrows) were significantly increased in the glomerulus from streptozotocin (STZ) injected-diabetic mice (4 weeks post diabetes injection) compared with those in non-diabetic control mice (A to D), Representative immunostains of glomerular p-MeCP2, HIPK2, and NOX4). Moreover, the glomerular size was significantly larger in these diabetic mice than that in control mice (E). In each group (n = 5), more than 30 pictures were used for quantification of immunostaining (Scale bar, 20 um, Mean ± SEM, ***p < 0.001 vs. Control). Western blotting showed that the protein expressions of p-& t-MeCP2, HIPK2, and NOX4 were also higher in the diabetic mice compared to controls (F). We used 2 pooled protein lysates collected separately from 2 to 3 sets of mice (non-diabetic control and diabetic mice), respectively. Western blottings were performed 2 to 3 times with these 2 pooled protein lysates and results shown are representative blots. Uncropped scans are presented in Supplementary Fig. 1 (Fig. 6F). Taken together, the downregulated HIPK2 (by Hipk2 siRNA) in HG treated mesangial cells has functional outcomes to downregulate MeCP2 phosphorylation and NOX4 expression, while siHIPK2 had no effect on SIAH1, confirming that SIAH1 is upstream of HIPK2. We next examined the effects of Hipk2 siRNA on TGF-β actions. Significant reductions of Hipk2 mRNA and protein (basal and 24hr TGF-β induced) by Hipk2 siRNA in MMC were confirmed ( The mRNA expression of Nox4 was significantly upregulated in the glomeruli of the diabetic mice compared with non-diabetic control mice (A), while Siah1 mRNA expression was significantly downregulated in diabetes compared to control (B). However, there were no significant differences in the mRNA expressions of Mecp2 and Hipk2 between the two groups (C and D). The expression of primary miR-25 was significantly increased in the glomeruli of diabetic mice compared with that in control mice (E), whereas precursor and mature miR-25 expressions were significantly decreased in the diabetic mice compared to control (F and G). We used RNA extracted from five glomeruli in each of the non-diabetic control and diabetic group, respectively. qPCR experiments for RNA expression were conducted 3 to 4 times with these respective 5 separate glomerular samples in each experiment (Mean ± SEM, *p < 0.05, and **p < 0.01 vs. Control). PCR data shown is normalized to cyclophilin A (Cypa) or U6 RNA Abbreviations; STZ, streptozotocin; NOX4, NADPH oxidase 4; SIAH1, seven in absentia homolog 1; MeCP2, methyl-CpG binding protein 2; HIPK2, homeo-domain interacting protein kinase 2.
transfected with Hipk2 siRNA. Moreover, the significant increases in protein expressions of p-MeCP2 and NOX4 that were evident in NTC transfected MMCs treated for 24 hr with TGF-β were also attenuated in the MMCs by knockdown of Hipk2 (Fig. 7E,G and I). However, the decrease in Siah1 mRNA after TGF-β treatment (Fig. 7D) was not affected by the siRNA mediated knockdown of Hipk2. Taken together, these results demonstrate that HIPK2, which is upregulated in HG and TGF-β treated mesangial cells, is also involved in MeCP2 phosphorylation and the expression of NOX4, whereas, because Hipk2 siRNA had no effect on SIAH1, it again confirms that SIAH1 is upstream of HIPK2.
Increased primary miR-25 expression, but decreased precursor and mature miR-25 expression levels are also attenuated by knockdown of HIPK2 even in HG and TGF-β treated MMCs. The primary miR-25 was significantly upregulated, but expressions of precursor and mature miR-25 were significantly decreased after HG and 24 hr TGF-β treatment in NTC transfected MMCs. However, these changes were also attenuated in Hipk2 siRNA transfected MMCs relative to NTC (Fig. 8A to F). These results suggest that MeCP2 phosphorylated by HIPK2 stabilized under diabetic conditions can block the first step of miR-25 processing and thus reduce the levels of precursor and mature miR-25.

Discussion
In this study, we showed for the first time that miR-25 expression can be downregulated under diabetic conditions due to the inhibition of the first step of miR-25 processing by p-MeCP2. The levels of p-MeCP2 were enhanced by the kinase HIPK2, whose expression is augmented due to stabilization by the downregulation of SIAH1 under diabetic conditions in vitro and in vivo. We also observed that NOX4, a validated target of miR-25 and a known inducer of oxidative stress 37,46,47  , and NOX4 were revealed to be significantly increased in the 24 hr TGF-β treated MMCs relative to control, whereas there was no significant difference in t-MeCP2 protein expression among the groups (E and G). qPCRs were performed 3 to 5 times with RNA isolated from 3 independent cell culture experiments. Western blotting was also performed 3 to 4 times with protein lysates derived from 3 independent cell culture experiments, and representative blots are shown (Mean ± SEM *p < 0.05, **p < 0.01, and ***p < 0.001 vs. Control). Uncropped scans are presented in Supplementary Fig. 2 in mouse models [48][49][50] , our results further support the significance of this SIAH1/HIPK2/MeCP2/miR-25/NOX4 pathway in DN. Besides NOX4, several other targets of miR-25 have been reported [51][52][53][54] , which might also contribute to signaling pathways in DN. However, as supported by several reports [46][47][48][49][50] . NOX4 is one of the most relevant targets of miR-25 that is related to DN pathology. Glomerular mesangial hypertrophy and ECM accumulation induced by HG, TGF-β , oxidant stress and related stimuli are relatively early events in the pathogegesis of DN, which can trigger pathological effects in other renal cells (podocytes, tubular, endothelial cells) and ultimately lead to end stage of renal disease. Inhibition of such early features of DN can prevent disease progression, proteinuria and renal failure. Therefore, increased understanding of the mechanisms mediating the early stages of DN is important 12 . Alterations in processing of miRNAs such as miR-25 by p-MeCP2 could be one such mechanism.
Recently miRNAs have been widely studied as novel mechanistic regulators of DN progression. Several miR-NAs are dysregulated in early DN and can promote the expression of extracellular matrix proteins and other genes associated with the initial stages of DN 12 . Moreover, several studies using renal cells in vitro, and in vivo animal models have shown functional relationships between aberrant expression of miRNAs and genes as well as pathways related to renal fibrosis and DN 12,13,18,20,22,39,40,45,[55][56][57][58] . However, it is not still clear whether changes in miRNA processing mechanisms are associated with the observed changes in the expression of key miRNAs implicated in DN. We hypothesized that MeCP2 phosphorylated by HIPK2 (p-MeCP2) may affect miRNA processing in diabetic kidney disease based on recent reports showing the role of p-MeCP2 in suppressing nuclear miRNA processing 27 and that HIPK2 can phosphorylate MeCP2 at Ser 80 33 . Moreover, evidence shows that HIPK2 is stabilized and activated after disruption of HIPK2-SIAH1 complex in response to DNA damage 59,60 . Our results revealed associations between SIAH1, HIPK2 and p-MeCP2 levels and processing of miRNAs, such as miR-25 which are downregulated by factors related to DN. Although loss-of-function mutations in the MeCP2 gene itself are found in the Rett syndrome 61,62 , the pathogenesis of DN is different, since p-MeCP2 regulates miRNA processing contributing to the expression of detrimental factors like NOX4.
As indicated above, p-MeCP2, HIPK2, and NOX4 were expressed at higher levels under diabetic conditions (in vivo and in vitro) compared with normal conditions, whereas the mRNA expression of Siah1 was decreased.    A and B). Moreover, the protein expressions of p-MeCP2 and NOX4 that were significantly upregulated in the NTC transfected MMCs with 24 hr TGF-β treatment were also attenuated by knockdown of Hipk2 in MMCs (E,G, and I). However, Siah1 mRNA expression level was still significantly decreased after TGF-β treatment even after knockdown of Hipk2 (D). qPCRs were performed 3 to 5 times with RNA isolated from 3 independent cell culture experiments. Western blotting was also performed 3 to 5 times with protein lysates derived from 3 independent cell culture experiments, and representative blots are shown (Mean ± SEM *p < 0.05, **p < 0.01, and ***p < 0.001 vs. Control). Uncropped scans are presented in Supplementary Fig. 3  Moreover, the increases in p-MeCP2 and NOX4 expressions after HG and 24 hr TGF-β treatments were attenuated in MMC transfected with siRNAs targeting Hipk2, whereas the decreases in Siah1 mRNA levels under these conditions were not altered by these treatments. These results suggest that SIAH1 suppressed by HG and/or TGF-β can stabilize HIPK2, and the subsequent increases in HIPK2 (via stabilization) can augment the phosphorylation of MeCP2 via the kinase activity of HIPK2. We also observed that the expression levels of mature and precursor miR-25 were decreased, but primary miR-25 levels were increased in vivo in the diabetic mice. Similar trends were also seen in MMC treated with HG or TGF-β (24 hr), and moreover these changes were attenuated by siRNAs targeting Hipk2 relative to NTC. These data further substantiate the role of HIPK2 as a critical kinase of MeCP2 as reported 33 and that increased p-MeCP2 mediated by HIPK2 stabilized under the diabetic conditions can block miR-25 processing from the primary to precursor conversion steps. Importantly, since NOX4 is a direct target of miR-25 37 , the resultant decreases in miR-25 can lead to enhance NOX4 expression (and related oxidative stress) as seen in diabetes. Conversely, knockdown of Hipk2 can prevent the induction of NOX4 by restoring miR-25 processing and mature miR-25 levels. Based on these findings, a schematic model for miR-25 processing and NOX4 expression in the early stage of DN is depicted in Fig. 8G. Our studies do not fully address the functional in vivo role of this pathway from SIAH1 to miR-25 and NOX4 via HIPK2 and p-MeCP2. This could be assessed in the future by treating diabetic mice with Hipk2 siRNAs. As suggested earlier that MeCP2 and HIPK2 protein expressions may be regulated at the posttranslational level 63 , we also observed that the mRNA expressions of Mecp2 and Hipk2 were not significantly different in the glomeruli of diabetic versus control mice ( Fig. 2C and D), although the protein levels of p-MeCP2 and HIPK2 were significantly upregulated in vivo in the glomeruli of diabetic versus control mice (Fig. 1F). Interestingly however, mRNA expressions of Mecp2 and Hipk2 were significantly increased in vitro in MMCs treated only for 6 hr TGF-β , but this increase was lost and even reduced below control after 24 hr of TGF-β treatment compared to control ( Fig. 4A and B). Expressions of these two factors were also significantly upregulated in MMCs treated with HG relative to control ( Fig. 3A and B). These results suggest that exogenous short-term treatment with stimuli like TGF-β (e.g. 6 hr) may upregulate mRNA expression of Mecp2 and Hipk2 by transcriptional regulation, possibly via Smads, and that these effects may be lost at later time points (24 hr treatment with exogenous TGF-β ). In cells treated with HG for 72 hr, endogenous TGF-β can be increased 18 which can account for the increases in Hipk2 mRNA levels. However, in vivo (4 week-STZ injected diabetic mice), the duration of hyperglycemia is relatively longer and hence upregulation of endogenous TGF-β is also longer, suggesting that HIPK2 protein expression under these in vivo conditions is likely to be upregulated at the posttranslational level (stabilization through decreases in SIAH1). Taken together, these data show that TGF-β can modulate HIPK2 expression by two independent mechanisms. Since we found three repeats of consensus Smad binding sites (CAGA) 3 kb upstream of the Hipk2 promoter, one mechanism for Hipk2 mRNA upregulation by TGF-β may be enhanced transcription via rapid Smad activation as reported previously 18,22,35 and binding to the Hipk2 promoter. In the second mechanism, HIPK2 protein can be stabilized and enhanced posttranslationally via decrease of SIAH1 as seen under diabetic conditions in vivo (Fig. 8G) 34,59,60,63 .
Notably, the mechanism described in this study is likely also relevant to human DN and other human kidney diseases, because MeCP2 expression has been reported to be increased in the kidneys of patients with chronic kidney disease, lupus nephritis, FSGS, IgA nephropathy as well as DN (https://www.nephroseq.org/resource/ main.html). In summary, these data demonstrate that MeCP2 regulated by HIPK2 stabilized by decreases in SIAH1 under diabetic conditions plays an important role in suppressing miR-25 processing and expression. As a consequence, NOX4, a target of miR-25, can be upregulated and this leads to oxidant stress associated with the pathology of DN (Fig. 8G). Together, these results reveal a novel mechanism for downregulation of key protective miRNAs in the diabetic kidney and also new therapeutic targets for the prevention of DN.

Methods
Animals. All animal studies were conducted according to a protocol approved by the Institutional Animal Care and Use Committee at the Beckman Research Institute of City of Hope. C57BL/6 mice (The Jackson Laboratory) were injected with 50 mg/kg of streptozotocin (STZ) intraperitoneally on 5 consecutive days. Mice injected with diluent served as controls. Diabetes was confirmed by tail vein blood glucose levels (fasting glucose > 300 mg/dl). Each group was composed of five mice. All mice were sacrificed at 4 weeks post-induction of diabetes. Glomeruli were isolated from freshly harvested kidneys by a sieving technique 58 . Enriched glomerular tissue below the sieve was collected and transferred to another sieve with a pore size of 75 μ m. After several washes with cold PBS, the glomerular tissue remaining on top of the sieve was collected. The glomeruli were collected for protein and RNA extraction. RNA samples were isolated from glomeruli of individual mice. However, for Western blots, because protein amounts obtained from glomeruli of single mice are not sufficient, we pooled glomeruli from two or three mice for protein extraction.
Cell Culture Experiments. MMCs were obtained and cultured as described previously in RPMI 1640 medium supplemented with 10% FBS 55 . Passages 6-8 were used for experiments. Recombinant human TGF-β 1 was from R&D Systems (Minneapolis, MN).