A microRNA circuit mediates Transforming Growth Factor-ß1 autoregulation in renal glomerular mesangial cells

Abstract

Enhanced transforming growth factor-β1 (TGF-β1) expression in renal cells promotes fibrosis and hypertrophy during the progression of diabetic nephropathy. The TGF-β1 promoter is positively controlled by the E-box regulators, Upstream Stimulatory Factors (USFs), in response to diabetic (high glucose) conditions; however, it is not clear whether TGF-β1 is autoregulated by itself. Since changes in microRNAs (miRNAs) have been implicated in kidney disease, we tested their involvement in this process. TGF-β1 levels were found to be upregulated by microRNA-192 (miR-192) or miR-200b/c in mouse mesangial cells. Amounts of miR-200b/c were increased in glomeruli from type 1 (streptozotocin) and type 2 (db/db) diabetic mice, and in mouse mesangial cells treated with TGF-β1 in vitro. Levels of miR-200b/c were also upregulated by miR-192 in the mesangial cells, suggesting that miR-200b/c are downstream of miR-192. Activity of the TGF-β1 promoter was upregulated by TGF-β1 or miR-192, demonstrating that the miR-192-miR-200 cascade induces TGF-β1 expression. TGF-β1 increased the occupancy of activators USF1 and Tfe3, and decreased that of the repressor Zeb1 on the TGF-β1 promoter E-box binding sites. Inhibitors of miR-192 decreased the expression of miR-200b/c, Col1a2, Col4a1 and TGF-β1 in mouse mesangial cells, and in mouse kidney cortex. Thus, miRNA-regulated circuits may amplify TGF-β1 signaling accelerating chronic fibrotic diseases such as diabetic nephropathy.

Introduction

Diabetes mellitus is associated with several debilitating complications including kidney disease or diabetic nephropathy (DN), a main cause for patients requiring painful and costly dialysis. Accumulation of extracellular matrix (ECM) proteins such as collagen in the kidney mesangium and tubulointerstitium is one of the major hallmarks of DN and contributes to renal failure^{1, 2}. Transforming growth factor-beta1 (TGF-β1) levels and signaling are enhanced in renal cells during the progression of DN. TGF-β1 plays a key role in mesangial cell fibrosis under diabetic conditions by inducing the expression of ECM proteins such as collagen^2-8. TGF-β1 is upregulated by high glucose (HG) in mesangial cells (MC) via the binding of Upstream Stimulatory Factors (USFs) (positive regulators) at the glucose-response element (CACGTG, also a typical E-box motif) in its promoter^9-11. On the other hand, TGF-β1 induces the expression of the Collagen type I alpha2 (Col1a2) gene by inhibiting the expression of the E-box repressors, Zeb1/2, while increasing Tfe3, another positive regulator of E-boxes^{4, 12}. Under basal conditions, Zeb1/2 repressors negatively regulate Col1a2 expression by binding to E-box elements in the far upstream region of the Col1a2 promoter^{4, 12}. ZEB1/2 are now widely recognized as general E-box repressors that bind to E-box elements in the promoters of genes such as E-Cadherin and collagens resulting in their repression^13-17.

microRNAs (miRNAs) are short (∼22 nucleotides) non-coding RNAs that are important regulators of gene expression^{18, 19}. miRNAs induce post-transcriptional gene repression by blocking protein translation via binding to the 3′UTR of their target genes, or by inducing mRNA degradation, and therefore have the potential to play central roles in gene expression under physiological and pathological conditions. Their widespread and distinct expression patterns under normal and disease states make miRNAs attractive molecular therapeutic targets for human diseases especially due to the recent advances in the development of chemically modified inhibitors of miRNAs like antagomirs²⁰ and locked nucleic acid (LNA) antimiRs^{21, 22}.

miRNAs are also involved in progressive kidney diseases²³. miR-192 is up-regulated by TGF-β1 in mouse MC (MMC)⁴ and by HG in human MC²⁴, and in glomeruli of diabetic mice, demonstrating that diabetic conditions induce miR-192. Zeb2 is targeted and negatively regulated by miR-192 in response to TGF-β1 in MMC and this leads to increased collagen expression due to relief of repression⁴. TGF-β1 and miR-192 levels are increased along with enhanced fibrosis in the kidneys of diabetic FXR knockout (KO) mice²⁵. miR-192 is also upregulated in the kidneys of other models of renal fibrosis (unilateral ureteral obstruction in mice and a rat model of remnant kidney disease) and in tubular epithelial cells treated with TGF-β1 in a Smad3-dependent manner²⁶. On the other hand, one study showed that TGF-β1 treatment decreased miR-192 expression in a tubular epithelial cell line²⁷. Targeted deletion of Dicer, a key enzyme involved in miRNA biogenesis, from proximal tubules could protect against renal ischemia-reperfusion injury²⁸. Since miR-192 levels were decreased in these tubular-specific Dicer KO mice, these data suggest that miR-192 inhibitors might be beneficial in models of kidney injury and disease. miR-192 and miR-200 family members regulate Zeb1/2 in epithelial-to-mesenchymal transition (EMT) in cancer cells and other established cell lines since Zeb1/2 are also repressors of E-cadherin^{27, 29-35}. miR-200 family members are also auto-regulated by Zeb1/2 through E-boxes in their promoters^{31, 36}. Although TGF-β1 expression is induced by HG in MC via the binding of USFs to E-boxes in the TGF-β1 promoter^9-11, it is not clear if the TGF-β1 promoter is autoregulated by TGF-β1 itself or miRNAs. Here we report that TGF-β1 is upregulated by TGF-β1 through miR-192 and miR-200b/c which target Zeb1/2. Furthermore, we observed that miR-200b/c are also directly induced by miR-192, implicating miR-192 as a key upstream regulatory renal miRNA. These miRNA-dependent positive feedback amplifying circuits may play major roles in accelerating TGF-β1 expression and signaling under diabetic conditions as proposed previously^{22, 37}.

Results

Autoregulation of TGF-β1 is mediated by miR-192 in MMC

TGF-β1 mRNA levels were significantly increased in MMC treated with TGF-β1 for 24 hr in a dose-dependent manner (5∼10ng/ml)(Figure 1a). Time course showed no significant increase of TGF-β1 mRNA levels at earlier time points (1 or 6 hour)(Figure1b). To test whether the TGF-β1 promoter is also activated by TGF-β1, MMC were transfected with pA835luc (luciferase reporter plasmid driven by the mouse TGF-β1 promoter containing glucose response element, CACGTG, also known as E-box)^{9, 38}. A significant increase (about 2-fold) in luciferase activity was observed in MMC treated with TGF-β1 (Figure 1c). These results demonstrate that TGF-β1 induces its own expression and promoter activity (autoregulation) in MMC.

Figure 1. Auto-upregulation of TGF-β1 in MMC.

(a) Dose-response curve of TGF-β1 mRNA levels (assessed by quantitative real time PCR) in MMC treated with exogenous recombinant TGF-β1 for 24 hr (mean and S.E. n=4). Significant increase was detected at 5∼10ng/ml of TGF-β1. (b) Time course study of TGF-β1 mRNA levels in MMC treated with 10ng/ml of TGF-β1 (mean and S.E. n=4). Significant increase was detected only at 24 hr. (c) Mouse TGF-β1 promoter (including upstream glucose response element, CACGTG, also known as E-box) Luc reporter (pA835luc) activity was significantly increased by TGF-β1(10ng/ml for 24hr) (mean and S.E. n=4). SD, serum depletion. (c) miR-192 mimic oligonucleotides (oligos) increased TGF-β1 promoter Luc reporter (pA835luc) activity, while miR-192 inhibitor oligos inhibited the TGF-β1-induced reporter activity (mean and S.E. n=4). NC, negative control oligo. ***, **, and *indicate p<0.001, p<0.01 and p<0.05, respectively.

Next we tested whether miR-192 can regulate the TGF-β1 promoter, because miR-192 increases gene expression through E-box enhancer elements by downregulating the E-box repressors Zeb1/2^{4, 22}, and the TGF-β1 promoter is also regulated by an E-box in its promoter^9-11. Transfection of MMC with miR-192 mimic significantly increased reporter activity of pA835luc in MMC compared to negative control (NC)(Figure 1d). Conversely, miR-192 inhibitor significantly decreased reporter activity under basal as well as TGF-β1-treated conditions. These results suggest that inhibition of Zeb1/2 by miR-192 may enhance TGF-β1 promoter activity in response to TGF-β1.

miR-200b and miR-200c are increased in MMC treated with TGF-β1 and in glomeruli of diabetic mice

Since miR-200 family members target both Zeb1 and Zeb2^29-34 and are also self-regulated by Zeb1/2 through E-boxes in their promoters^{31, 36}, miR-200 family members may mediate TGF-β1 signaling in MC. The miR-200 family consists of two groups based on their seed sequences: group 1 (miR-141 and miR-200a) and group 2 (miR-200b, miR-200c and miR-429)(Figure 2a). miR-200b and miR-200c levels as well as miR-192 were significantly increased by TGF-β1 (Figure 2b). miR-200b and miR-200c levels were also increased in the glomeruli of both type 1 (STZ injected C57BL6 mice) and type 2 diabetic mice (db/db mice) compared to their respective controls (Figure 2c). Increased levels of miR-192 and miR-200b were also observed in glomeruli of STZ-injected mice on the DBA/2J background (Supplementary Figure S1). These results suggest that miR-200b and miR-200c may also play a role in MC dysfunction associated with DN.

Figure 2. The expression of miR-200 family members in MMC treated with TGF-β1 and in the glomeruli of diabetic mice.

(a) Mature sequences of miR-200 family members. miR-200 family members are divided to two groups, Group 1 (miR-141 and miR-200a) and Group 2 (miR-200b, miR-200c and miR-429), based on their seed sequences underlined. (b) Expression levels of miR-200 family member and miR-192 and miR-215 in MMC treated with or without TGF-β1 (10ng/ml for 24hr). miR-200b and miR-200c levels were significantly increased by TGF-β1, with no significant change in miR-141 or miR-429 (mean and S.E. n=3). miR-192 levels were increased and miR-215 levels decreased, as reported previously⁴. (c) miR-200b and miR-200c levels were increased in diabetic mice glomeruli relative to respective control non-diabetic mice (mean and S.E. n=4). CTR, control; STZ, Streptozotocin. ***, **, and *indicate p<0.001, p<0.01 and p<0.05, respectively.

miR-200b and miR-200c are upregulated by miR-192 in MMC

Because miR-200 family members can be self-regulated by Zeb1/2^{31, 36}, we tested whether miR-192 upregulates miR-200b/c by targeting Zeb1/2 at E-boxes in miR-200b/c promoters. miR-192 mimic significantly increased the miR-200b (about 2.5 fold) and miR-200c (about 1.5 fold) levels compared to control (NC)(Figure 3a and b). TGF-β1 again significantly increased the miR-200b (about 3 fold) and miR-200c (1.6 fold), and this was reversed by the miR-192 inhibitor (Figure 3a and b). These results indicate that TGF-β1-induced miR-200b/c upregulation is mediated by miR-192 in MC.

Figure 3. miR-200 family members and TGF-β1 are upregulated by miR-192 in MMC.

(a, b) miR-192 mimic or inhibitor oligos were transfected into MMC and miR-200b/c levels examined by realtime qPCR. Effects of miR-192 mimic and inhibitor on the expression of miR-200b (a) and miR-200c (b). miR-192 significantly increased miR-200b and miR-200c levels in MMC compared to negative control (NC). miR-192 inhibitor ameliorated the TGF-β1(10ng/ml for 24hr)-induced increases in miR-200b and miR-200c levels (mean and S.E. n=3) (c) Effects of miR-192 and miR-200b mimics and inhibitors on the expression of TGF-β1. miR-192 or miR-200b significantly increased TGF-β1 levels in MMC. Conversely, miR-192 or miR-200b inhibitor significantly attenuated the TGF-β1–induced increase in TGF-β1 levels. (mean and S.E. n=3). ***, **, and *indicate p<0.001, p<0.01 and p<0.05, respectively.

TGF-β1 is upregulated by miR-192 and miR-200b in MMC

Since TGF-β1 is upregulated by HG in MC through the actions of USFs at the E-box in its promoter^9-11, we tested if miR-192 or miR-200b affects TGF-β1 mRNA expression. miR-192 or miR-200b mimic significantly increased TGF-β1 mRNA levels while their inhibitors reversed the stimulatory effect of TGF-β1 (Figure 3c). Because miR-200b is upregulated by miR-192 (Figure 3a), these results suggest that the observed autoregulation of TGF-β1 is mediated by a miRNA cascade from miR-192 to miR-200b.

miR-200b/c suppress Zeb1 through its 3′UTR

Treatment of MMC with miR200b or miR-200c significantly decreased Zeb1 protein levels relative to NC (Figure 4a and b ). This was similar to the inhibitory effect of TGF-β1 on Zeb1 (Fig. 4c, d). Conversely, inhibitors of miR-200b reversed the inhibitory effect of TGF-β1 on Zeb1 protein expression (Figure 4e-f). These results further support the role of TGF-β-induced miR-200b in Zeb1 repression.

Figure 4. Zeb1 is inhibited by miR-200b and miR-200c.

(a) Zeb1 protein levels are significantly decreased by miR-200b or miR-200c (or miR-200a) in MMC. NC, negative control of mimic oligo. (b) Quantification of Zeb1 protein levels (mean and S.E. n=3). (c) Decreased levels of Zeb1 in TGF-β1 (10ng/ml for 24hr)-treated MMC (d) Quantification of Zeb1 protein levels (mean and S.E. n=3). (e) Increase of Zeb1 protein levels by miR-200b inhibitor. All samples were treated with TGF-β1 (10ng/ml for 24hr). (f) Quantification of Zeb1 protein levels (mean and S.E. n=3). ***, **, and *indicate p<0.001, p<0.01 and p<0.05, respectively.

Next we tested if mouse Zeb1 is a direct target of miR-200b and miR-200c also in MMC. Luciferase (Luc) reporters were constructed containing the 3′UTR sequence of the mouse Zeb1 gene with two potential target sites (and mutations in these sites) for miR-200b/c identified by TargetScan³⁹(Figure 5a). miR-200b or miR-200c significantly inhibited the Zeb1 3′UTR Luc reporter activity whereas the NC had no effect (Figure 5b). These results are consistent with our findings that miR-200b or miR-200c inhibited Zeb1 protein levels in MMC (Figure 4a). Given that miR-200b and miR-200c belong to the same group and share the same seed sequence, they likely share the same targets (Figure 2a). Single mutations in the Zeb1 3′UTR (M1 or M2) resulted in an attenuated response to miR-200b compared to the WT construct (Figure 5C). A construct containing both mutations M1 and M2 completely lost the response to miR-200b (Figure 5C), confirming that these two sites in the Zeb1 3′UTR are bona-fide targets of miR-200b.

Figure 5. Zeb1 3′UTR is a bona-fide target of miR-200b.

(a) Luciferase reporter constructs of Zeb1 3′UTR. Two potential miR-200b/c target sites (1 and 2) were found in this 3′UTR sequence (WT, wild type). Mutations were introduced into these sites by site direct mutagenesis. Zeb1 3′UTR MT1 and MT2 are single mutants of site 1 (M1) and site 2 (M2), respectively. Zeb1 3′UTR MT1&2 is a double mutant of both sites (M1 M2). (b) miR-200b and miR-200c significantly inhibited the reporter activity of WT Zeb1 3′UTR. (mean and S.E. n=4) (c) The constructs with single mutation in either site1 or site2 in the Zeb1 3′UTR showed less inhibitory response to miR-200b mimic while the double mutant of both sites completely lost the response to miR-200b, indicating that these two sites are real targets of miR-200b. (mean and S.E. n=4). ***indicates p<0.001.

miR-192 and miR-200b upregulate Col1a2 via E-box enhancer region

Collagens are regulated through E-box sites in the upstream regions^{4, 12, 16, 17}. Since our current results showed that miR-200b and miR-200c inhibit Zeb1 expression in MMC through the Zeb1 3′UTR (Figure 4 and 5), we examined the effects of miR-200b on Col1a2 gene expression and whether it co-operates with miR-192. A significant increase in Col1a2 expression was detected in MMC transfected with a combination of miR-192 and miR-200b, while a slight but not significant increase was noted with either miR-192 or miR-200b alone (Figure 6a). In addition, we also observed a co-operation between miR-192 and miR-200b in increasing the activity of a Luc reporter containing the E-box-Col1a2 promoter¹² (Figure 6b). As shown previously, the inhibition of both Zeb1 and Zeb2 is necessary for significant induction of Col1a2⁴. Therefore, the simultaneous transfection of miR-200b (to inhibit Zeb1) and miR-192 (to inhibit Zeb2) may be necessary for significant induction of Col1a2. On the other hand, miR-200b inhibitor alone was sufficient to significantly repress Col1a2 expression (Figure 6c). Consistent with the above observations, inhibitors of miR-192 and miR-200b each attenuated TGF-β1-induced E-box-Col1a2 promoter Luc activity (Figure 6d). Together, these results indicate that miR-200b/c upregulate Col1a2 gene via inhibition of Zeb1 in collaboration with miR-192 (inhibitor of Zeb2).

Figure 6. miR-200b regulates Col1a2 expression through the upstream E-boxes.

(a) Col1a2 mRNA levels were significantly increased by a combination of miR-192 and miR-200b mimics. (mean and S.E. n=3) (b) Mouse Col1a2 promoter (including upstream E-boxes) Luc reporter activity was increased by miR-192 +miR-200b mimics (mean and S.E. n=4) (c) miR-200b inhibitor inhibited the TGF-β1-induced expression of Col1a2 (mean and S.E. n=3). (d) miR-192 or miR-200b/c inhibitors inhibited TGF-β1 (10ng/ml for 24hr)-induced Col1a2 promoter reporter activity (mean and S.E. n=4). (e) Col4a1 mRNA levels were significantly increased by miR-192 or miR-200b mimics as well as TGF-β1 in MMC, while inhibitors of miR-192 or miR-200b attenuated Col4a1 mRNA levels in TGF-β1-treated MMC (mean and S.E. n=3). ***and ** indicate p<0.001 and p<0.01, respectively.

TGF-β1, miR-192 or miR-200b upregulates Col4a1

Since an increase of type IV collagens occurs in early stages of DN^{7, 8}, we also tested the expression of Col4a1gene. Col4a1 mRNA levels were significantly increased by miR-192 or miR-200b as well as TGF-β1 in MMC (Figure 6e). Inhibitors of miR-192 or miR-200b significantly attenuated TGF-β1-induced Col4a1mRNA expression. Interestingly, we identified at least twelve potential E-boxes in the promoter region of Col4a1gene (Supplementary Figure S2). Thus Col4a1 gene may be regulated by TGF-β1 through these E-boxes via miR-192 and miR-200b in MMC.

TGF-β1 increases USF1 and Tfe3 occupancies, but decreases Zeb1 occupancy on the TGF-β1 promoter E-box

Next, we used chromatin immunoprecipitation (ChIP) assays to examine the occupancy of E-box regulators, USF1 and Tfe3 (positive) and Zeb1 (negative), on the TGF-β1 promoter in MMC treated with TGF-β1 or HG (25mM). Occupancy was determined by amplification of TGF-β1 promoter E-box region (Figure 7a) using quantitative real-time PCR of ChIP-enriched DNA. USF1 occupancy was significantly increased by HG treatment compared to normal glucose (5.5mM)(NG) or Mannitol (19.5mM)(Man)(Figure 7b), as reported⁹. Interestingly, TGF-β1 treatment showed significant enrichment of USF1 as well as Tfe3, with reciprocal significant depletion of Zeb1 compared to untreated control (SD)(Figure 7c). We also examined occupancies of these E-box regulators on the Col1a2 enhancer E-box region and observed that TGF-β1 increased Tfe3 and decreased Zeb1 enrichments at this Col1a2 enhancer as also noted in our previous studies^{4, 12}. However, no significant change of USF1 occupancy on Col1a2 E-box region was detected (Figure 7d & e ). These results suggest that the downregulation of Zeb1 by miR-200b/c induced by TGF-β1 can enhance both TGF-β1 and Col1a2. But the recruitment of USF1 to TGF-β1 promoter E-box region may be different from Col1a2 enhancer E-boxes.

Figure 7. ChIP analysis of E-box regulators (USF1, Tfe3 and Zeb1) on the TGF-β1 promoter and Col1a2 enhancer E-box regions.

(a) Upstream promoter region of the TGF-β1 gene. Arrows indicate positions of primers used for ChIP analysis. (b) ChIP analysis of USF1 on the TGF-β1 promoter E-box region. Significant increase in USF1 occupancy at 24 hr after HG treatment. NG, normal glucose (5.5mM); Man, Mannitol (19.5mM); HG, high glucose (25mM). mean ±SEM (n=3). (c) ChIP analysis of USF1, Tfe3 and Zeb1 on the TGF-β1 promoter E-box region. Results show significant increases in the occupancies of USF1 and Tfe3, but significant decrease in that of Zeb1 at 24 hr of TGF-β1 (10ng/ml) treatment relative to control (serum depletion, SD). mean ±SEM (n=3). (d) Upstream enhancer region of the Col1a2 gene. Arrows indicate positions of primers used for ChIP analysis. (e) ChIP analysis of USF1, Tfe3 and Zeb1 on Col1a2 far-upstream E-box region. Significant increase in the occupancy of Tfe3 and reciprocal decrease in that of Zeb1 were detected at 24 hr of TGF-β1 treatment. No significant change was detected in USF1 occupancy. mean ±SEM (n=3). ***, **, and *indicate p<0.001, p<0.01 and p<0.05, respectively.

miR-192 inhibitor decreases the expression of miR-200 family members as well as Col1a2, Col4a1 and TGF-β1 in MMC in vitro and in mouse kidney cortex in vivo

miR-192 is a key candidate upstream regulator that controls the expression of downstream genes such as the miR-200 family, Col1a2, Col4a1 and TGF-β1through E-boxes in their promoters. Thus, we tested whether the inhibition of miR-192 could have protective effects. miR-192 inhibitor significantly inhibited the expression of miR-141, miR-200b, miR-200c, Col1a2, Col4a1 and TGF-β1as well as miR-192 in vitro in cultured MMC (Figure 8a).

Figure 8. miR-192 inhibitors attenuate the expression of miR-200 family members, TGF-β1 and Col1a2 genes in MMC (in vitro) and mouse kidney cortex (in vivo).

(a) miR-192 inhibitor significantly inhibited the expression of miR-141, miR-200b, miR-200c, Col1a2, Col4a1and TGF-β1 in MMC in vitro (mean and S.E. n=3). (b) In vivo effects of injecting normal mice with LNA-antimiR-192: 21mer, 5′-gGctGtcAatTcaTagGtcAg-3′(uppercase: LNA; lowercase: DNA)²². LNA-antimiR-192 significantly inhibited the expression of miR-141, miR-200b, miR-200c, Col1a2, Col4a1and TGF-β1 in mouse kidney cortex relative to those in saline-injected control mice (mean and S.E. n=3). ***, **, and *indicate p<0.001, p<0.01 and p<0.05, respectively.

To evaluate the in vivo relevance, we tested the effects of miR-192 inhibition in mice using locked nucleic acids (LNA)-modified antisense miR-192 (LNA-antimiR-192)²². Injection with LNA-antimiR-192 decreased the expression of miR-192, miR-141, miR-200b, miR-200c, Col1a2, Col4a1and TGF-β1 in the mouse renal cortical tissues in vivo (Figure 8b).

Discussion

In the current study, we observed for the first time that TGF-β1 can be auto-upregulated in MMC by the activation of a miRNA cascade including miR-192 and key miR-200 family members. It is likely that, initially, diabetic conditions (HG) induces TGF-β1 through various mechanisms including binding of USFs (E-box activators) to its promoter^9-11. Next, miR-192 can be upregulated by TGF-β1 possibly via Smad3 and p53 in MMC and diabetic kidneys since the miR-192 promoter has Smad and p53 binding sequences and is upregulated by Smad3 or p53^{26, 40}. Then miR-192 can target and downregulate Zeb1/2 (E-box repressors)^{4, 27, 35, 41} and thereby enhance the expression of TGF-β1 through the recruitment of USF1 and Tfe3 but dissociation of Zeb1 from the E-box region in the TGF-β1 promoter to create an amplifying circuit (Figure 9). Interestingly, we found that key miR-200 family members were also upregulated by TGF-β1 in MMC and in the glomeruli of diabetic mice. In addition, miR-192 also increased miR-200b/c (Figure 3) probably through E-boxes in their promoters^{31, 36}. miR-200 family targets Zeb1/2 and also enhances the expression of miR-200 family itself though E-boxes and this results in another positive feedback circuit to accelerate the downstream signaling (Figure 9). We identified at least two miR-200b target sites in the mouse Zeb1 3′UTR and confirmed that miR-200b targets Zeb1 in MMC (Figure 5). These results in MMC are consistent with other reports showing that miR-200 family members suppress Zeb1expression by targeting its 3′UTR^29-33 and suggest that Zeb1/2 inhibition is a major function of the miR-200 family. Further decrease of Zeb1/2 by miR-200 family in MMC can augment the expression of TGF-β1. These signaling loops may be a key feature of chronic diseases like DN^{12, 22, 37}.

Figure 9. Auto-upregulation of TGF-β1mediated by a miRNA-cascade.in MMC.

Diabetic conditions increase the expression of TGF-β1 which induces miR192 (inhibitor of Zeb1/2) probably via mechanisms involving the actions of Smad3 and p53. miR-200b/c are also upregulated due to the downregulation of Zeb1/2 by miR-192. Because miR-200b/c target Zeb1/2, this can lead to a further decrease of Zeb1/2 and further augment the expression of miR-200b/c, TGF-β1, Col1a2and Col4a1through loss of repression, along with gain of activation (via Tfe3 and/or USF1) at their E-boxes. These signaling loops accelerate TGF-β1 signaling and downstream gene expression involved in the progression of chronic kidney diseases like DN.

miR-200b also increased Col1a2 expression and promoter activity by collaborating with miR-192 (Figure 6). Simultaneous transfection of miR-192 (inhibitor of Zeb2) and miR-200b (inhibitor of Zeb1) was necessary for significant induction of Col1a2 or its promoter activity (Figure 6). These results are consistent with previous data showing that Zeb2 inhibition by miR-192 alone was not sufficient but also needed Zeb1 inhibition (by shRNAs) to elicit a significant increase Col1a2 expression in MC⁴. On the other hand, single transfection of an inhibitor of either miR-192 or miR-200b was sufficient to inhibit Col1a2 expression and its promoter activity (Figure 6). Thus, inhibition of both Zeb1 and Zeb2 is necessary for significant induction of Col1a2, while the presence of just one of these repressors (Zeb1 or Zeb2) is sufficient to suppress Col1a2. However, unlike Col1a2, miR-200b/c, Col4a1 and TGF-β1 were upregulated by either miR-192 or miR-200b alone (Figure 3 & 6). Since Col1a2 upstream E-box region is regulated by multiple factors like Tfe3, Tsc-22 and Ybx1¹² besides Zeb1/2, induction of Col1a2 might involve additional or more complex mechanisms compared to TGF-β1, Col4a1 or miR-200 family promoters.

ChIP experiments showed significant decrease of Zeb1 occupancy on E-box regions of TGF-β1 promoter and Col1a2 enhancer in MMC treated with TGF-β1 (Figure 7). These results further support the notion that suppression of E-box repressor Zeb1 by miR-200b/c induced by TGF-β enhances the expression of TGF-β1 and Col1a2. Although E-box activator Tfe3 occupancy was increased by TGF-β1 at E-box regions of both of TGF-β1 and Col1a2 genes, USF1 occupancy was increased only at E-box region of TGF-β1 gene. These results demonstrate similarities in the regulation of these two genes with respect to decrease in Zeb1, but differences in the recruitment of positive factors. Since we noted at least twelve potential E-boxes in the Col4a1 gene promoter, Col4a1may be regulated by TGF-β1, miR-192 or miR-200b through these E-boxes similar to the regulation of TGF-β1, miR-200b and Col1a2, although more studies are required to clarify this.

miR-200 family members also regulate Zeb1 and Zeb2 in EMT in cancer^29-34 and were decreased during EMT of the canine kidney cell line (MDCK) treated with TGF-β1 for long time periods (2∼3 weeks)²⁹. Since type I collagen induces EMT^{42, 43}, it is possible that there is a biphasic regulation in which miR-192 and 200 family are upregulated by TGF-β1 at early time points (6-24 hours) which decreases Zeb1 and Zeb2 to induce type I collagen in MMC. This increase in collagen may subsequently induce EMT (maybe by decreasing these miRNAs) at later time points. In the NRK52E epithelial cell line, Chung et al reported that miR-192 levels were increased by 2ng/ml TGF-β1 at 1-24 hr²⁶, while Wang et al. reported that miR-192 levels were decreased by 10ng/ml TGF-β1 at 3 days with no data at earlier time points³⁵. Although the dose of TGF-β1 might dictate its response, it is possible that initially TGF-β1 increases miR-192 and miR-200 family to promote type I collagen expression (up to 24hr), while miR-192 levels may be reduced at later time periods (3days to weeks). Furthermore, there is no EMT in mesangial cells, and hence, it is highly likely that the response of miRNAs to TGF-βmay be cell-type specific and context dependent. On the other hand, fibrosis and accumulation of ECM proteins in MC and tubular cells are well-established features of DN in both human and mice^{1, 2}. Thus, the upregulation of collagen expression by TGF-β1 via increases in miR-192 and miR-200b and concomitant decrease in Zeb1/2 in MC may be a key feature of DN and possibly more relevant during the early stages of diabetic kidney injury. Interestingly, increased levels of miR-192 and miR-200 family were observed in human kidneys from patients with hypertensive nephrosclerosis, IgA nephropathy and lupus nephritis^44-46 and some animal models of kidney injury^{25, 26, 28}, further supporting the miRNA-mediated mechanism of TGF-β1 auto-upregulation and enhanced renal fibrosis. On the other hand, it was also reported that miR-192 expression decreased with increased severity of DN and fibrosis, however, normal levels of miR-192 in healthy kidneys were not provided for comparison²⁷. Clearly more work is needed to further evaluate these aspects in renal cells.

Notably, miR-192 is regulated by Smad3 and p53^{26, 40}. Numerous mutations in p53 and Smad genes have been found in cancer cells and immortalized cell lines^47-49 which may affect the miR-192 response to TGF-β1 in these cells. In other words, p53 and Smad genes may also be decision makers of TGF-β1 response. In fact, mutant p53 changes the response to TGF-β1 via Smad interaction⁵⁰. Kidney injury and fibrosis were inhibited in p53 or Smad3 knockout mice^{26, 51}. In the current study, we used primary mouse MC and glomeruli which have normal intact p53 and Smad genes. Therefore, the response of miR-192 and miR-200 family to TGF-β1 in primary MMC may be different from those reported in cancer or immortalized kidney cell lines, and might explain some of the discrepancies in TGF-β1 responses. They also indicate the importance of using primary cells (at early passage) for studying models of renal fibrosis or injury and especially the mediatory role of miRNAs.

miR-200 family was also reported to control phosphatidylinositol-3-kinase (PI3K) signaling by targeting Fog2, an inhibitor of PI3K⁵². As PI3K/Akt activation in diabetic kidney cells induces hypertrophy and fibrosis^{12, 22, 53, 54}, upregulation of miR-200 family members may also activate Akt through inhibition of Fog2 to promote mesangial hypertrophy. Since miR-192 also activates Akt kinase and fibrosis by inducing miR-216a and miR-217 which target Pten²², this cascade of miRNAs (miR-200 family, miR-216a and miR-217) initiated by miR-192 may enhance Akt activation, hypertrophy and fibrosis in MC.

In summary, these novel miR-mediated circuits demonstrated in the current study can lead to the auto-upregulation of TGF-β1, amplification of TGF-β1 signaling and thereby accelerate chronic fibrotic kidney diseases including DN (Figure 9). The ability of the miR-192 inhibitor to repress the expression of miR-200 family members, TGF-β1, Col4a1and Col1a2 both in vitro (MMC treated with TGF-β1) and in vivo (mouse kidney cortex), suggests that miR-192 could be a novel target for the prevention or treatment of fibrotic disorders like DN as well as other diseases where TGF-β signaling is hyperactive.

Methods

Mice

All animal studies were conducted according to a protocol approved by the Institutional Animal Care and Use Committee. Induction of diabetes by streptozotocin injections in C57BL/6 mice was carried out as described⁴. Mice were used 7-8 weeks after the onset of diabetes. Type 2 diabetes db/db mice and genetic control db/+mice (10-12 weeks old) were obtained from Jackson Laboratories (Bar Harbor, Maine). Glomeruli were sieved from renal cortical tissue as described^{4, 53}.

Cell culture

Primary mouse mesangial cells (MMC) were isolated and cultured as described⁴. Recombinant human TGFβ1 was from R&D Systems, Inc. (Minneapolis, MN, USA).

Real time quantitative PCR (qPCR)

Real-time qPCR was performed using SYBR Green PCR Master Mix and 7300 Realtime PCR System (Applied Biosystems, Foster City, CA USA), as reported^{4, 12, 22}. PCR primer sequences are: mouse TGF-β1, 5′-GGACTCTCCACCTGCAAGAC-3′and 5′-GACTGGCGAGCCTTAGTTTG-3′; mouse Col4a1, 5′-GCCTTCCGGGCTCCTCAG-3′and 5′-TTATCACCAGTGGGTCCG-3′. miRNA expression levels were also confirmed by miScript System (Qiagen).

Western blot analyses

These were performed as described⁴. Antibodies against ZEB1 (Santa Cruz Biotechnology, Santa Cruz, CA) and beta-actin (Cell Signaling, Beverly, MA, USA) were used at 1:1000 dilution.

Plasmids and 3′-UTR reporters

pA835luc was a generous gift from Dr. Kumar Sharma (University of California, San Diego)^{9, 38}. Mouse Col1a2 promoter luciferase plasmid (E-box-Col1a2P-luc) was described previously¹². 3′UTR of mouse Zeb1 was amplified using primers, Zeb1-3′UTR-F, 5′-ccaaCTCGAGTTCTTCTAAAAGGAAATTCTACTTGG-3′and Zeb1-3′UTR-R, 5′-gagaGCGGCCGCGGGAATTCTGTAGTGCAGAAGTTCTC-3′.

The PCR fragment was digested with Xho I and Not I and cloned into psiCheck2 (Promega). Potential miR200b/c site mutants were made by site directed mutagenesis as described²² using primers, for Site1, 5′- GCATGGCTAGCAGCTGTATCACTCTTA-3′and 5′- TAAGAGTGATACAGCTGCTAGCCATGC-3′and for Site2, 5′- AAATCCGCTTCAGCTGTTTATTATGTT-3′and 5′-AACATAATAAACAGCTGAAGCGGATTT -3′. Bases substituted are underlined.

Luciferase assays

MMC were transfected with plasmids and/or miRNA mimics or inhibitors using Nucleofector (Lonza). Basic Nucleofector Kit (Lonza) was used. After 48 h, Luciferase activities were measured as described^{4, 22}.

miRNA Oligonucleotides (Oligos)

Oligos representing the miRNA mimics, the negative control mimics, miRNA inhibitors, and negative control inhibitors were obtained from Dharmacon (Lafayette, CO). LNA modified anti-miR-192 was from Exiqon (Vedbaek Denmark)²².

Chromatin immunoprecipitation (ChIP) assays

ChIP assays were performed as reported^{4, 9, 12, 22}. Briefly, MMC were treated with 10 ng/ml TGF-β1 or high glucose (25mM) condition for 24 hr and then formaldehyde-fixed. The crosslinked chromatin was sheared and immunoprecipitated using antibodies against USF1 (Santa Cruz Biotechnology), Tfe3 (BD Bioscience), or Zeb1. ChIP-enriched DNA was purified and used as template for real time qPCRs using primers spanning Col1a2 or TGF-β1 E-box regions as described^{4, 9}.

LNA-antimiR injections in mice

LNA anti-miR-192 was subcutaneously injected (2.5 mg/kg) into C57BL/6 mice and renal cortical tissues harvested 6 h later as previously described²².