Enhanced myostatin expression and signalling promote tubulointerstitial inflammation in diabetic nephropathy

Myostatin (MSTN), a family member of the transforming growth factor (TGF)-β super family, has been detected in the tubuli of pig kidney, but its role in the human kidney is not known. In this study we observed upregulation of MSTN mRNA (~8 to 10-fold increase) both in the glomeruli and tubulointerstitium in diabetic nephropathy (DN). In DN, immunoreactive MSTN was mainly localized in the tubuli and interstitium (∼4–8 fold increase), where it colocalized in CD45+ cells. MSTN was also upregulated in the glomeruli and the arterial vessels. Tubulointerstitial MSTN expression was directly related to interstitial fibrosis (r = 0.54, p < 0.01). In HK-2 tubular epithelial cells, both high (30 mmol) glucose and glycated albumin upregulated MSTN mRNA and its protein (p < 0.05–0.01). MSTN-treated HK-2 cells underwent decreased proliferation, together with NF-kB activation and CCL-2 and SMAD 2,3 overexpression. In addition, MSTN induced intracellular ROS release and upregulated NADPH oxidase, effects which were mediated by ERK activation. In conclusion, our data show that MSTN is expressed in the human kidney and overexpressed in DN, mainly in the tubulointerstitial compartment. Our results also show that MSTN is a strong inducer of proximal tubule activation and suggest that MSTN overexpression contributes to kidney interstitial fibrosis in DN.


Results
MStn is expressed in the normal kidney and is upregulated both in the glomeruli and in the tubulointerstitium of patients with Dn. Figure 1A,B shows MSTN mRNA levels in the glomerular and tubulointerstitial compartments of microdissected biopsies of normal controls and DN. In the normal kidney MSTN mRNA was expressed at low level in both in the tubular and glomerular compartments and its protein was very faintly expressed in the tubuli and interstitium (Figs. 1C,D and 2A). MSTN mRNA was markedly overexpressed in DN in the glomeruli and tubulointerstitium (a ~20 to 40-fold increase vs. controls, and a ~4 to 8-fold increase vs. glomerular diseases, respectively, p < 0-05-0.01). In addition, MSTN protein was upregulated in DN (p < 0.01; Fig. 1A,B). In DN, MSTN immunostaining was observed in glomerular cells (Fig. 2C,D), mainly Activin-type IIB receptor is upregulated in the fibrotic tubulointerstitial areas and in tubulointerstitial infiltrates in DN. MSTN binds to Act RIIB in skeletal muscle and adipose tissue 6 . Act RIIB was expressed in both glomeruli and tubuli of kidneys in control subjects (Fig. 4A,B). In DN, Act RIIB immunostaining was downregulated when the values were collectively evaluated (Fig. 4A,B). However, Act RIIB was expressed in fibrotic areas along with upregulated MSTN (Fig. 4C). Of note, infiltrating cells were highly positive for Act RIIB (Fig. 4D). Therefore in DN ActRIIB is expressed not only in tubular cells (at a lower intensity as compared to normal kidney), but also in interstitial areas.
MSTN expression correlates with glomerulosclerosis and interstitial fibrosis in DN. When MSTN immunostaining was related to individual structural kidney changes (Table 1), we observed that In addition, logMSTN immunostaining in the tubulointerstitium was directly associated with interstitial fibrosis (R = 0.54, p < 0.01) and interstitial inflammation (R = 0.40, p < 0.05). clinical determinants of kidney MStn expression. As a next step, we studied whether MSTN expression in the diabetic kidney could be predicted by clinical findings. Table 2 shows the associations between individual clinical data and logMSTN expression in DN. LogMSTN expression in the glomeruli, tubuli and tubulointerstitium was directly related to serum C-reactive protein, but not to proteinuria, nor to eGFR. MSTN decreases HK-2 cell proliferation rate. In analogy with the MSTN inhibitory effect on cell growth and differentiation shown in skeletal muscle cells, we studied if MSTN in PTECs can negatively act on proliferation. A 24-hour MSTN treatment showed a trend in slow cell-cycle progression (control, 45 ± 3%; MSTN, 48.5 ± 1.8%; p = NS; Supplementary Fig. S2A). The HK-2 proliferation rate was significantly reduced after a 48-hour MSTN treatment ( Supplementary Fig. S2B,C). This observation is consistent with an upregulation of P16 ink4a at the same time point, as shown in Supplementary Fig. 3A. No differences in apoptosis rate were observed ( Supplementary Fig. S3B).

Effects of MSTN on ROS production. Recently, it has been shown that MSTN is a pro-oxidant and causes
ROS generation in muscle cells via the NADPH system 26 . CellRox was used to investigate the effects of MSTN on ROS production in HK-2 cells. Cytofluorimetric analysis showed that intracellular ROS increased ∼1.5-fold after 5-hour stimulation with MSTN (p < 0.01; Fig. 6A) and quenched after prolonged exposure (−25% with respect to untreated cells). These data indicate that MSTN can immediately induce intracellular ROS production in HK-2. To further investigate the mechanisms underlying the induction of ROS release, we examined the effects of MSTN on the expression of Nox4, a member of the renal nicotinamide adenine dinucleotide phosphate reduced form (NADPH) oxidase, and a major source of oxidative stress in DN. As shown in Fig. 6B, both Nox4 mRNA and its protein were markedly upregulated (by ∼six-folds, p < 0.05 vs controls) following stimulation with MSTN. As a next step, we investigated the downstream pathways involved in MSTN signaling. We found that MSTN induced p44/42 (ERK1/2) phosphorylation (p < 0.01 vs basal T0; Fig. 6C), while p38 and JNK were unchanged (data not shown). When we studied the effects of ERK inhibition by PD 98059 on fibronectin, cytokine, and Nox4 mRNA  www.nature.com/scientificreports www.nature.com/scientificreports/ expression, we observed that only Nox4 mRNA was blunted ( Fig. 6D; −50% with respect to MSTN-treated cells, p < 0.01) suggesting a role for ERK in regulating ROS production by NADPH oxidase. Lastly, in renal tubular cells, as in skeletal and adipose tissue, 48-hour exposure to MSTN promoted pAKT suppression (−40% with respect to untreated cells, p < 0.01; Fig. 6E).

Discussion
MSTN affects multiple pathways of glucose and protein metabolism, yet its role in the human kidney has not been studied so far. Three issues are addressed in this study, which bear discussion. The first is MSTN expression in the human kidney, its upregulation in type 2 DN, and its association with glomerulosclerosis, interstitial inflammation and fibrosis, findings that collectively suggest that MSTN, similar to other members of the TGF-β super family, is involved in kidney fibrogenesis. The second is the proinflammatory and profibrotic action that MSTN exerts on kidney tubular cells, an effect that is similar to that previously shown in skeletal muscle. The third is the upregulation of MSTN by the diabetic milieu in proximal tubular cells, which suggests that the MSTN response in native kidney cells is a feature of the upregulated innate immunity in DN. (A) HK-2 were exposed to LG or HG, Alb or Gly-Alb for 48 hours and MSTN mRNA was measured by rt-PCR and its protein by western blot. After 48 hours, HG upregulated both MSTN mRNA and its protein (by ∼6 and ∼1.3 folds, respectively p < 0.05-0.01 HG vs. LG). Also Gly-Alb (500 ng/ml) increased either MSTN mRNA and protein expression (by ∼6.4 and ∼3 folds, respectively p < 0.05-0.01 vs Alb). Photos are representative of MSTN expression. Each experiment was carried out 3 times. (B) The graph shows Act RIIB mRNA levels, measured by rt-PCR in cells exposed to HG or Gly-Alb. The diabetic milieu increased either Act RIIB mRNA and protein expression. Pictures are representative of Act RIIB staining by immunofluorescence. www.nature.com/scientificreports www.nature.com/scientificreports/ The overexpression of MSTN in human DN is a novel finding. In our study, MSTN was expressed in both infiltrating tubulointerstitial cells and native kidney cells; furthermore, MSTN expression was related to glomerulosclerosis and tubulointerstitial fibrosis, observations that suggest a role for MSTN in mechanisms mediating CKD. Of note, the MSTN Act RIIB receptor was downregulated in kidney tubules with normal morphology, suggesting an adaptive response to raised MSTN exposure. However, concurrent with MSTN upregulation, Act RIIB was expressed in areas of atrophy/fibrosis and in inflammatory infiltrates. Therefore in DN ActRIIB is expressed not only in tubular cells (at a lower intensity as compared to normal kidney), but also in interstitial areas.
These observations suggest that MSTN in the DN kidney acts mainly in the tubulointerstitium and in inflammatory/tubular atrophy lesions, by acting locally and/or by boosting chemiotaxis.
Both immunological and inflammatory mediators play a role in initiating and extending glomerular and tubular damage in DN 27 . We previously observed that innate immunity is upregulated in native kidney cells already at the stage of diabetic microalbuminuria 21 ; at a more advanced stage, tubulointerstitial kidney cell infiltration is associated with albuminuria and fibrosis 21 , suggesting that kidney infiltration from circulating cells can accelerate tissue damage. Tubulointerstitial injury is a major feature of DN and an important predictor of renal dysfunction 28 . Several cell types, including leukocytes, monocytes, and macrophages 29,30 , are implicated in processes related to DN. In our study, MSTN was expressed more in tubulointerstitial infiltrating cells in diabetic kidney disease compared to nondiabetic proteinuric diseases, suggesting that the kidney MSTN response is induced by diabetes per se. In several tissues, the MSTN response is a component of the innate immune response to endogenous signals, such as free radicals 4 and high interleukin-6 31,32 . In addition, in high-fat-induced obesity, MSTN is overexpressed in leucocytes and spleen, suggesting its role as a mediator of inflammation 32 .
In a previous study, we observed MSTN expression in infiltrating cells in atherosclerotic aorta. Colocalization studies showed that these cells expressed CD45, a marker of hematopoietic lineage. In a monocyte cell line (THP-1 cells) and in freshly isolated human monocytes, MSTN increased CCL-2 and α-SMA mRNA expression 20 . In turn, monocytes stimulated with CCL-2 displayed increased MSTN gene expression, suggesting that MSTN participates in a feed-forward inflammatory loop. Monocyte CCR-2 membrane expression was also significantly upregulated following MSTN treatment. In THP-1 cells, MSTN also acted as a chemoattractant 20 . Taken together, these data indicate that in circulating CD45+ cells and monocytes, MSTN plays an inflammatory and chemoattractive role.
The reason(s) why MSTN is more upregulated in diabetic infiltrates that in other glomerular diseases is not completely understood. In analogy with the finding that high IL-6 upregulates MSTN in muscle, diabetes-induced low grade inflammation might upregulate MSTN in circulating cells. In our study, kidney MSTN expression was neither related to the diabetes duration nor to HbA1c, while it was directly associated with serum C-reactive protein (CRP) levels.
A major question is whether the upregulation of MSTN in DN derives from MSTN + /CD45 + infiltrating cells or by MSTN overexpression by native kidney cells. In our study MSTN was already markedly upregulated (by about 3 folds) in tubules showing no interstitial infiltration, while it was even more expressed (by about 8 folds) in the presence of tubulointerstial infiltrates. In addition, high glucose was able per se to upregulate MSTN in HK-2 cells. Taken together, our data suggest that both cell infiltration and native kidney cells response contribute to abnormal MSTN expression in DN.
In our study, MSTN expression in the renal tubulointerstitium correlated with cell infiltration, in accordance with MSTN role in mediating cell recruitment. In addition MSTN in the tubulointerstitium, but not in tubuli, correlated with interstitial fibrosis. This finding suggests that MSTN overexpression in infiltrating cells, but not in tubule cells, play a role in kidney fibrosis. As a matter of fact the downregulation of Act RIIB in kidney tubules might have conferred protection from high MSTN levels. Such an effect has been observed in different cell types, when an excessive Activin/MSTN ligand level causes the internalization and degradation of Act RI/II receptor 33 , There is increasing evidence that the immune activation in kidney cells in response to hyperglycemia or to other endogenous ligands that are upregulated by diabetes plays a major role in tissue damage 21,27 . To demonstrate the contribution of kidney tubules to MSTN overexpression, we used a human PTEC culture system (HK-2). We showed that both high glucose and glycated albumin caused an overexpression of MSTN and its receptor in human proximal tubular cells, suggesting that the hyperglycemic milieu per se or downward signals produced by hyperglycemia induce tubular MSTN. To recognize the transcriptional pathways that are activated by MSTN in the kidney, we studied the expression profiles of selected MSTN downward genes. In HK-2 cells, MSTN caused a decrease in replication and enhanced NF-κB activation and enrichment of several members of the NF-κB inflammatory pathway. In addition, exposure of tubular cells to glucose or glycated albumin upregulated MSTN, CCL-2, and fibronectin, effects that were blunted by MSTN silencing. All together, our findings support the hypothesis that the diabetic milieu increases MSTN production by renal cells, which results in pro-inflammatory and profibrotic effects. This is a new mechanism, linking hyperglycemia and MSTN in the pathogenesis of diabetic nephropathy.
Our findings also have other implications for the mechanisms of damage in DN. The observation that in HK-2 cells MSTN enhances ROS production through NADPH oxidase suggests that MSTN may potentiate the mechanisms of injury and cell loss already known to be active in DN [34][35][36] . Another finding that needs discussion is that the inhibition of the MAPK-ERK cascade downregulated the MSTN-induced NOX4 upregulation, a finding in keeping with MSTN action in muscle 37 . Therefore, the inhibition of the MAPK-ERK cascade may be another strategy to blunt MSTN effects in kidney tubular cells.
The absence of association between MSTN expression and proteinuria, and the lack of altered regulation of MSTN in renal tissues of nondiabetic kidney disease suggests that the observed MSTN activation in DN was not consequence of protein excretion.
Consistent with prior work in atherosclerotic lesions 20  www.nature.com/scientificreports www.nature.com/scientificreports/ chemotaxis, in vascular smooth muscle cells (VSMCs) MSTN induces both cytoskeletal rearrangement and increases cell migratory rate 20 . Accordingly, our results indicate that MSTN is upregulated both in progressive abdominal aortic atherosclerosis 20 and in the kidney vessels of patients with DN, suggesting a similar role of MSTN on vascular damage.
The present study suggests the activation of a MSTN-dependent pathway of fibrosis in DN. This hypothesis raises several issues, including a possible interaction between MSTN and other TGF-β superfamily proteins 35 . TGF-β1 and TGF-β2 have been identified as inducers of fibrosis due to their ability to recruit monocytes and myofibroblasts, activate the EMT program, and promote inflammation and apoptosis 36,37 . TGF-β mediates fibrosis via Smad-dependent and -independent pathways. TGF-β SMAD-independent fibrotic signaling follows activation of MEK/Erk, Rho-like GTPases, and p38 mitogen-activated protein kinase (MAPK) 36,37 . The activation of extracellular-regulated kinases (ERK) and p38 MAPK is also necessary for collagen synthesis and accumulation 37 . In our model, we observed a MSTN-induced increase in the expression of ERK and P-38 MAPK phosphorylation that may promote the development of renal fibrosis through the SMAD-independent pathway.
It is also important to consider that several activities of MSTN overlap with those of activin A, which is upregulated in mouse models of chronic kidney disease 38 . The activation of the Act RIIA in PTECs promotes apoptosis and inhibits cell growth 39 . In addition, renal interstitial fibroblasts are activated by activin A produced by tubular cells 40 .
This study has some limitations. First, we studied MSTN in patients with clinical diabetic disease. Therefore, additional work is needed to understand the time course of MSTN regulation at different stages of DN. In addition, although our data show a strong association between kidney MSTN upregulation and interstitial fibrosis, the effects of MSTN inhibition have been addressed only in vitro.
In summary, we demonstrated that the expression of MSTN is upregulated in both infiltrating and native kidney cells in patients with DN, and that it is associated with glomerulosclerosis and tubulointerstitial fibrosis. We observed also that MSTN decreases cell proliferation, induces the expression of NFkB, and enhances the expression of CCL-2 and fibronectin mRNA in renal proximal tubular cells. In addition, the diabetic milieu upregulates MSTN, and the blockade of MSTN signaling reduces CCL-2 and fibronectin overexpression in kidney proximal tubule cells. Our results suggest that MSTN participates in the mechanisms of kidney inflammation and is a potent inducer of proximal tubule activation in the kidney. All together, our findings suggest that MSTN overexpression contributes to kidney interstitial fibrosis in DN.

Methods
Twenty-six patients with type 2 diabetes and albuminuria were recruited for this study from the Department of Internal Medicine, Nephrology Division, University of Genoa. The study was part of a larger study in patients with type 2 DN approved by the Ethical Committee of the Department of Internal Medicine, Genoa University 21 . The inclusion and exclusion criteria were defined to select a cohort of type 2 diabetic patients whose albuminuria was the result of DN 21 . The indications for renal biopsy were proteinuria greater than 0.5 g/d or atypical DN, and therefore, all biopsies were for clinical assessment. All subjects were informed about the nature, purposes, procedures, and possible risks of the renal biopsy before their informed consent was obtained. The procedures were in accordance with the Helsinki declaration. The clinical and laboratory characteristics of patients diabetic subjects are represented in Supplementary Table 1. Diabetic subjects (age 60 ± 4 years, 15 M/11 F) had overt DN (proteinuria 4.1 ± 2.5 g/day, eGFR= 31 ± 3 ml/min). Angiotensin-converting enzyme inhibitors and/or angiotensin receptor blockers had been withdrawn at least two weeks prior to the renal biopsy 20 .
As a control group we examined kidney tissue obtained from the healthy pole of kidneys removed because of small and localized tumors (n = 13, 8 M/5 F, 62 ± 2 yrs, eGFR= 77 ± 4 ml/min). All subjects had normal blood pressure and urinary protein excretion and were nondiabetic. To further check the specificity of MSTN expression in DN, MSTN gene and protein expressions were also studied in kidney biopsies of 22 subjects with nondiabetic proteinuria (FSGS, n = 11, IgA nephropathy, n = 11). Their demographic and clinical data are shown in Supplementary Table 1.
All kidney biopsies were analyzed by the same pathologist (G. Salvidio) who was unaware of study results. The morphological changes in kidney biopsies were considered in glass slides stained with hematoxylin and eosin, periodic acid-Schiff, trichrome and silver stain. Morphological changes, including interstitial fibrosis and tubular atrophy were classified as previously described 41 . Histological preparation and immunohistochemical staining. Paraffin sections (5 µm) of 2% paraformaldeyde-fixed tissue were analyzed for MSTN (Myostatin polyclonal antibody, Proteintech, LaboSpace s.r.l., Milan, Italy), its receptor Act RIIB (H-70 and G7) (Santa Cruz Biotechnology, D.B.A. Italia s.r.l., Seregno, Italy) and CD45 (Novocastra, Leica Biosystem, Milan, Italy), Immunostaining was performed as previously described 20 . MSTN protein expression was evaluated by Leica Qwin Image Analysis System (Leica, Cambridge, UK). Constant optical threshold and filter combination were set to select only the positive areas and both positive and negative tubuli or glomeruli were evaluated. For immunofluorescence, frozen tissue sections (5 µm) were fixed in cold methanol. MSTN detection was performed by Alexa Fluor ® 594 Goat Anti-Rabbit IgG (Thermo Fisher Scientific, Milan, Italy) and CD45 and Act RIIB by FITC Goat Anti-Mouse IgG (Sigma Aldrich, Milan, Italy) Megalin/LRP2 was used to characterize MSTN localization in proximal tubuli. Isotype-matched antibodies corresponding with the primary antibodies were used as negative controls (Supplementary Fig. 1B). Glomeruli and the tubulointerstitial specimens were microdissected as previously described 20 . Briefly, frozen biopsy sections (7 µm thick) of kidney underwent laser capture microdissection (LCM) performed with a Veritas apparatus (Arcturus Bioscence, Mountain View, California, U.S.A.). Total RNA was extracted by Arcturus PicoPure Isolation Kit (Applied Biosystem, Life Technologies, Monza, Italy).