MMP-10 is Increased in Early Stage Diabetic Kidney Disease and can be Reduced by Renin-Angiotensin System Blockade

Matrix metalloproteinases have been implicated in diabetic microvascular complications. However, little is known about the pathophysiological links between MMP-10 and the renin-angiotensin system (RAS) in diabetic kidney disease (DKD). We tested the hypothesis that MMP-10 may be up-regulated in early stage DKD, and could be down-regulated by angiotensin II receptor blockade (telmisartan). Serum MMP-10 and TIMP-1 levels were measured in 268 type 2 diabetic subjects and 111 controls. Furthermore, histological and molecular analyses were performed to evaluate the renal expression of Mmp10 and Timp1 in a murine model of early type 2 DKD (db/db) after telmisartan treatment. MMP-10 (473 ± 274 pg/ml vs. 332 ± 151; p = 0.02) and TIMP-1 (573 ± 296 ng/ml vs. 375 ± 317; p < 0.001) levels were significantly increased in diabetic patients as compared to controls. An early increase in MMP-10 and TIMP-1 was observed and a further progressive elevation was found as DKD progressed to end-stage renal disease. Diabetic mice had 4-fold greater glomerular Mmp10 expression and significant albuminuria compared to wild-type, which was prevented by telmisartan. MMP-10 and TIMP-1 are increased from the early stages of type 2 diabetes. Prevention of MMP-10 upregulation observed in diabetic mice could be another protective mechanism of RAS blockade in DKD.

Diabetic kidney disease (DKD) is a leading cause of end-stage renal disease. Matrix metalloproteinases (MMPs) have been found to play a role in various pathogenic mechanisms involved in microvascular complications of diabetes mellitus (DM). These endopeptidases are implicated in the degradation and remodelling of the extracellular matrix (ECM), as well as proteolytic modification of growth factors and cytokines 1 . MMP activity is regulated by specific tissue inhibitors of matrix metalloproteinases (TIMPs).
Higher circulating MMP-2 and MMP-9 levels have been found in diabetes 2,3 . An association between high serum levels of MMP-10 and a greater risk of DKD has been previously shown in type 1 diabetes (T1DM), and the absence of Mmp10 in a murine model protected against renal macrophage infiltration and mesangial expansion 4 . We and others have demonstrated that serum concentration of MMP-10 is elevated in chronic kidney disease (CKD) associated with vascular complications 5,6 . Previous studies on the endogenous inhibitor of MMP-10, TIMP-1, show inconclusive data, demonstrating elevated circulating levels on DM 5 , while others observed similar levels as compared to healthy subjects 7 .

Experimental methods
All procedures performed in studies involving animals were conducted according to the European Community guidelines for the ethical animal care and use of laboratory animals (2010/63/EU) and approved by the University of Navarra Animal Research Review Committee. Five weeks-old male db/db (BKS.Cg-+Lepr db /+Lepr db /OlaHsd) and db/m (BKS.Cg-+Lepr db /OlaHsd) mice were purchased from Envigo (Venray, Netherlands). Mice were maintained on a 12-hour light/12-hour dark cycle, had free access to water and were fed with standard chow diet (db/m) or 2018-chow diet (db/db) (Teklad Global 18%-Protein Rodent Diet, Envigo, Madison, WI, USA). Animals were divided into four groups (n = 6), db/db (diabetic); db/db telmisartan-treated (5 mg/kg/day starting at 8 weeks of age); db/m (non-diabetic); db/m telmisartan-treated. Two days before being sacrificed, blood pressure was measured using a system adapted to the mouse tail, following manufacturer's instructions (Mouse/Rat Tail Cuff Blood Pressure System, IITC Life Science, Woodland Hills, CA, USA), and 24-hour urine was collected in metabolic cages. Mice were sacrificed by CO 2 inhalation at 8, 12 and 16 weeks of age, and kidneys and blood were collected.
Analytical measurements. Measurements of plasma creatinine and glucose concentrations, as well as urine creatinine and albumin, were performed using specific kits on a Cobas C311 biochemical analyser (Roche, Mannheim, Germany). Plasma insulin concentration was determined by ELISA (Merck-Millipore EZRMI-13K, Billerica, MA, USA).
Kidney histological analysis. Sections (3-μm-thick) cut from 10% formalin-fixed, paraffin-embedded kidney samples were used for periodic acid-Schiff (PAS) and Masson trichrome staining. Using coronal sections of the kidney, 10-15 consecutive glomeruli per mouse (6 mice per group) were examined for evaluation of glomerular mesangial expansion. The index of the mesangial expansion was defined as the ratio of mesangial area to glomerular tuft area. The mesangial area was determined by assessment of the PAS positive and nucleus-free area in the mesangium using ImageJ software.
Statistical analysis. Parametric tests were used for statistical analyses in the clinical study due to the large sample size. Data are expressed as means (SD), median (IQR) or proportions, as appropriate. Differences in demographic and clinical variables between the patient and control groups were evaluated using unpaired Student's t tests for quantitative variables, and the Chi-Square tests for categorical variables. Since the groups were found to differ in age, gender frequency and BMI, group differences in MMP-10 and TIMP-1 were evaluated using a generalized linear model, including: group (patients with diabetes or healthy subjects), age, gender and BMI to control for the possible confounding effect. Associations between variables were examined using Pearson correlation. Changes in MMP-10 and TIMP-1 across different stages of CKD or albuminuria were evaluated using two way ANCOVA, introducing age, gender and BMI as co-variables of confusion. When appropriate, post-hoc contrasts of the adjusted means were performed using Tukey correction. For the experimental study, differences between groups of mice were analysed using ANOVA, followed by appropriate post-hoc tests. Associations between renal Mmp10 expression and other variables were analysed by Pearson correlation, and adjusted per group by multiple linear regression. P-values were penalized by Bonferroni and statistical significance was established at p < 0.05. Statistical analysis was performed using R 3.0.3 11 .

Results
Clinical Study. Population characteristics. Table 1 shows the demographic and clinical characteristics of the participants. There was a higher proportion of males among patients with diabetes (38% vs. 73%), who were also older and had a higher BMI compared to controls. Prevalence of hypertension among diabetics was 77%, and 70% of them received RAS inhibitors. Diabetic subjects using RAS inhibitors had a greater albuminuria compared to those without RAS blockade (40.58 ± 40.58 vs 22.64 ± 74.62 mg/mmol, p = 0.009).
The univariate analysis showed a readily noticeable increase in MMP-10 with GFR and albuminuria (Fig. 2e), which was not as striking for TIMP-1 (Fig. 2f). This implies that the serum levels of both increase as DKD progresses, a process characterized by macroalbuminuria and a greater deterioration in the filtration rate.
The increase in MMP-10 and TIMP-1 throughout the different stages of CKD remained significant (p < 0.0001 both) in the two way ANCOVA (GFR and albuminuria) after adjusting for age, gender and BMI. This statistical model explained variations in MMP-10 (adjusted R = 0.56) better than in TIMP-1 (adjusted R = 0.42). However, the differences between stages of albuminuria disappeared in this statistical analysis. When performing the ANCOVA analysis, we found an interaction (p < 0.05) between RAS inhibitors treatment and GFR that led us to study the association between MMP-10 and renal function separately in the two subgroups of patients. The slope of the regression line was more steep in the group of patients who received RAS inhibitors, as shown by the coefficients [−0.0090 (−0.0111, −0.0069), p = 1.5 10 −14 vs −0.0053 (−0.0080, −0.0026), p = 1.94 10 −4 ]. These two regression lines cross at MDRD = 76 ml/min/1.73 m 2 , so that patients with better renal function treated with RAS inhibitors tend to have lower serum MMP-10 concentration. Conversely, patients with worst renal function (MDRD < 76) treated with RAS inhibitors tend to have higher serum MMP-10 concentration than non-treated ones. However, no interaction was found and no differences were detected when comparing circulating levels (2020) 10:26 | https://doi.org/10.1038/s41598-019-56856-3 www.nature.com/scientificreports www.nature.com/scientificreports/ of TIMP-1 (Fig. 2g,h) (see Supplementary Fig. S2). However, the magnitude of this effect was not big enough to find statistically significant differences when comparing circulating levels of MMP-10 between patients on RAS treatment and patients without RAS inhibitors, despite comparing groups with similar renal function (see Supplementary Fig. S2). No interaction was found and no differences were detected when comparing circulating levels of TIMP-1 ( Fig. 2h; see Supplementary Fig. S2).

Experimental study.
A study was conducted in db/db mice to explore whether renal expression of Mmp10 and Timp1 may be up-regulated in this diabetic model, recapitulating our observations in the clinical setting. Moreover, since prior data has supported a link between RAS and other MMPs, we evaluated if renal Mmp10 and Timp1 expression could be modulated through RAS blockade with telmisartan. At 8 weeks of age, all db/db mice became obese (40.1 ± 2.7 vs. 24.9 ± 1.3 g; p < 0.001) and developed T2DM, demonstrated by hyperglycaemia and hyperinsulinemia (Table 2). Following the onset of diabetes, there was a progressive increase in albuminuria between 8 and 16 weeks of age, which was not observed in controls. After 8 weeks of treatment, those diabetic animals treated with telmisartan displayed a significant reduction in albuminuria in comparison with non-treated diabetic mice. Diabetic mice at 16 weeks of age presented higher plasma creatinine than controls, but no changes were observed between db/db mice treated with telmisartan and those without treatment (Table 2). No changes were observed with blood pressure either (db/db + T: 123.5 ±11.3 vs db/db: 127.7 ±15.5 mmHg, p = 0.23).
At 16 weeks of age, glomerular hypertrophy and mesangial matrix expansion were significantly increased in db/db mice (Fig. 3a-d). In spite of the noticeable glomerular hypertrophy, the mesangial matrix expansion was www.nature.com/scientificreports www.nature.com/scientificreports/ not significantly different in db/db compared with db/m mice at age 8 weeks (Fig. 3). Altogether, these results confirmed the suitability of the model to analyse early stage of DKD changes at 8 weeks and mild changes at 16 weeks.  weeks-old, db/db mice showed increased renal gene expression of Mmp10 compared to age and sex-matched db/m mice. The highest Mmp10 gene expression was observed at 16 weeks of age, with an 8-fold increase versus controls (Fig. 4a). A non-significant increment on Timp1 gene expression was noted in diabetic mice (p = 0.21), as well as a down-regulation after RAS inhibition which did not achieve statistical significance (see Supplementary  Fig. S3). Immunohistochemical analysis was performed to evaluate the localization of MMP-10 expression in the kidney. Positive staining was observed in podocytes and juxtaglomerular apparatus from diabetic mice (Fig. 4be). No specific staining was observed in tubules or interstitium. Treatment with telmisartan significantly downregulated Mmp10 gene expression after 4 (p < 0.01) and 8 weeks (p < 0.05), to nearly the normal expression in control mice (Fig. 4a). A significant positive correlation was observed in diabetic mice between renal Mmp10 gene expression and albuminuria (R = 0.792; p = 0.019).

Discussion
Our study demonstrates that circulating MMP-10 is increased in T2DM, together with its inhibitor TIMP-1, starting at the earliest stages of DKD, and their concentration increases with the severity of kidney disease. Moreover, we demonstrated overexpression of renal Mmp10 in an experimental murine model of T2DM, which was prevented by RAS inhibition.
Increased serum levels of MMP-10 and TIMP-1 were found in patients with diabetes compared to healthy subjects. The increment was observed in those with glomerular filtration rates greater than 60 and 90 ml/min/1.73 m 2 for MMP-10 and TIMP-1, respectively. In addition, a progressive elevation occurred across different stages of CKD with the highest values in patients with advanced kidney disease. A strong correlation was demonstrated between both proteins and glomerular filtration rate estimated by traditional equations, especially with cystatin C in comparison with serum creatinine. An experimental murine model was carried out, with the aim of studying the renal expression of Mmp10 and Timp1 during hyperglycaemia. We were able to demonstrate glomerular overexpression of Mmp10 in diabetic mice from the earliest stages of DKD. Mmp10 downregulation occurred after RAS blockade with telmisartan during 4 and 8 weeks of treatment, indicating an involvement of the RAS system.
The role of other MMPs in DKD has been previously studied 2,[12][13][14] . Two recent publications have demonstrated association between MMP-10 and cardiovascular risk factors in T1DM 4 www.nature.com/scientificreports www.nature.com/scientificreports/ lacking about the association between MMP-10 and T2DM. We also saw an increase in TIMP-1 levels in the population with diabetes. Several studies confirm our results in which the elevation of TIMP-1 is demonstrated in patients with DM both in the presence or absence of overt vascular complications 5,15 . A few studies have shown divergent results compared to ours but had either a different study population (mean age: 19 years-old) 2 or a low www.nature.com/scientificreports www.nature.com/scientificreports/ sample size 7 . The observed correlation between serum MMP-10 and TIMP-1 in DM reflects the coordinated up-regulation of MMPs and TIMPs that has been reported in other studies 16 .
We further evaluated MMP-10 and TIMP-1 according to renal function as recent studies have demonstrated that DKD can exist despite the absence of biochemical indicators [17][18][19] , and elevation of other MMPs may precede the development of microalbuminuria 7 . A parallel increase was observed between both proteins in relation to the progression of renal disease. Of note is the increase in MMP-10 and TIMP-1 in T2DM despite the absence of a significant decrease in eGFR: TIMP-1 begins to increase when the eGFR is still above 90 ml/min/1.73 m 2 , while MMP-10 increases with an eGFR lower than 90, but greater than 60 ml/min/1.73 m 2 . Previous data has shown an imbalance between other MMPs and TIMP-1 in DM, even in the absence of overt nephropathy 7,20 , however this vs. db/m (Cont). Treatment with telmisartan (+T) resulted in down-regulation of Mmp10 expression. Renal MMP10 immunostaining (panels b-e) has been found to be mainly located in podocytes (short arrows) and juxtaglomerular apparatus (long arrows) of db/db mice. DM: db/db; Cont: db/m; DM + T: db/db treated with telmisartan; Control + T: db/m treated with telmisartan. *** p < 0.001 vs. db/m; ## p < 0.01 vs. 12 weeks-old db/db without telmisartan; # p < 0.05 vs. 16 weeks-old db/db without telmisartan. n = 6 on each group. (2020) 10:26 | https://doi.org/10.1038/s41598-019-56856-3 www.nature.com/scientificreports www.nature.com/scientificreports/ is the first study showing an elevation of serum MMP-10 in early-stage DKD, followed by a progressive increase towards late-stage disease. In addition, an inverse correlation was observed between MMP-10 and TIMP-1 with eGFR, in line with previous results from our group that described higher levels of MMP-10 and TIMP-1 in patients with advanced CKD 5 . The previous study estimated GFR by MDRD-4, while in the present work we evaluated four different formulas (MDRD-4, CKD-EPI creatinine, CKD-EPI cystatin C, CKD-EPI creatinine-cystatin C), showing that the strongest correlation is that based on cystatin C.
Despite not being a sine qua non condition, the presence of albuminuria is one of the characteristic signs of DKD. In our study, we found an association between the level of albuminuria and higher circulating levels of MMP-10 and TIMP-1. However, this association disappears after adjusting for GFR, but was not modified after adjusting for RAS treatment. The reason of this may be due to the fact that the level of albuminuria may not be fully dependent on the degree of GFR reduction (i.e. patients with low GFR may have either high or low uACR). On the other hand, in our cohort the patients using RAS inhibitors had a greater albuminuria compared to those without treatment, which may explain the variability in MMP-10 levels in this population. The relationship between albuminuria and MMP-10 has also been described by Peeters et al. who observed a significant association between the degree of albuminuria and circulating MMP-10 and TIMP-1 in patients with T1DM 15 . This association is interesting as their study population had a mean eGFR of 95.7 ml/min/1.73 m 2 and the association held good for normo-, micro-and macroalbuminuric patients.
To assess the role of renal tissue as a possible source of MMP-10 and TIMP-1 overexpression, and coupled to data showing an association between other MMPs and RAS 12,21,22 , we analysed a murine model to test whether the renal expression of both proteins could be up-regulated during hyperglycaemic state and if their expression could be downregulated with telmisartan. Higher Mmp10 renal expression was observed from early stage of DKD in db/db as compared to db/m, even in the absence of significant histological lesions (8 weeks-old), and a down-regulation was observed after RAS blockade. Moreover, a correlation between glomerular Mmp10 expression and albuminuria was demonstrated. Immunohistochemical analysis showed MMP-10 localization in podocytes and juxtaglomerular apparatus known to have high RAS activity 23 . These findings may support the relationship between MMP-10 and RAS, and a pathogenic role of MMP-10 in glomerular DKD. In agreement with data from other studies, reporting increased glomerular Timp1 expression in diabetic rats 24 , we also observed higher Timp1 expression in renal tissue in the diabetic mice, however our results did not achieve statistical significance probably due to the limited sample size.
The role of MMP-10 in kidney disease has not been extensively studied. Toni et al. demonstrated that diabetic mice with Mmp10 deficiency showed minor renal morphologic alterations, as well as improved renal function, compared to those expressing Mmp10 4 . Keeping this in mind, our results would suggest a deleterious glomerular effect for MMP-10 in DKD and show that its expression can be down-regulated by blocking RAS. The mechanism involving MMP-10 in the pathophysiology of DKD is not yet clear. Two recent studies have analysed MMP-10 in primary glomerular disease 25,26 . The authors demonstrated the presence of MMP-10 in podocytes using immunofluorescence and suggest a regulation of the metalloproteinase through focal adhesion kinase (FAK) activation. Since FAK activation is at the beginning of the angiotensin II signalling pathway 27,28 , we propose a possible link between MMP-10 and RAS in DKD, where angiotensin II, induced by hyperglycaemia, could stimulate FAK activation resulting in Mmp10 overexpression. In this scenario, the significant down-regulation of renal Mmp10 expression achieved following RAS blockade by telmisartan would be explained. However, this hypothesis must be demonstrated by complementary mechanistic studies. In our experimental model, telmisartan dosage was chosen so that RAS inhibition did not produce changes in blood pressure, indicating that reduction in Mmp10 expression was independent of hemodynamic factors other than DKD.
Certain limitations of our study must be acknowledged. The clinical study is a cross-sectional observational study and prospective long-term studies are needed to stablish a causal link between MMP-10 and TIMP-1 dysregulation and the onset of DKD. Prospective long-term studies are also necessary to identify whether MMP-10 and TIMP-1 could be used as diagnostic and prognostic biomarkers. Another limitation is that we did not reach enough statistical power to detect statistically significant differences between patients with or without RAS inhibitors, with similar GFR, even when we compared GFR > 90 or GFR < 60 ml/min/1.73 m 2 . In addition, confounding factors such as markers of chronic inflammation or subclinical atherosclerosis were not analysed as they did not form part of the original objectives for the study. In the experimental model, an important limitation is the difficulty in measuring the concentration of circulating MMP-10 in mice, as we have found no ELISA kits with the required specificity (especially, differentiating MMP-10 from MMP-3), sensitivity and reliability. Information is lacking about urinary concentration of both proteins, as well as a comparison of the results with other glomerular diseases.
Interestingly, the findings of the present study are relevant for identifying an up-regulation of MMP-10 and TIMP-1 in T2DM, even before overt kidney disease is noted, and opens the door for future studies aiming at elucidating the mechanistic role of this MMP and its inhibitor on DKD.
In conclusion, we demonstrate for the first time an association between MMP-10 and TIMP-1 and the different stages of renal disease, as well as glomerular overexpression of Mmp10 in T2DM, which is prevented by blocking RAS. The novelty of these results lies in the participation of MMP-10 in diabetic glomerular disease, as well as its possible role as a new therapeutic target.

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.