Mesangial matrix expansion in a novel mouse model of diabetic kidney disease associated with the metabolic syndrome

1Nephrology and Renal Transplantation Research Group, Department of Microbiology, Immunology and Transplantation, KU Leuven, Leuven, Belgium 2University Hospitals Leuven, Department of Nephrology and Renal Transplantation, Leuven, Belgium 3Centre for Molecular and Vascular Biology, Department of Cardiovascular Sciences, KU Leuven, Leuven, Belgium 4Department of Imaging and Pathology, Division of Translational Cell and Tissue Research, KU Leuven, Leuven, Belgium 5Division of Nephrology and Hypertension, Mayo Clinic, Rochester, Minnesota, USA 6William J von Liebig Center for Transplantation and Clinical Regeneration, Mayo Clinic, Rochester, Minnesota, USA


Introduction
Diabetic nephropathy is a major cause of chronic kidney disease (1). The evolution of diabetic nephropathy is hallmarked by functional changes (proteinuria, hyperfiltration) and structural changes (glomerular basement membrane thickening, mesangial matrix expansion, glomerulosclerosis) (2)(3)(4). In humans with diabetes, kidneys are typically enlarged, and associated with worse renal outcome (5). The mechanism behind this enlargement is not well understood and could be related to compensatory hypertrophy, or excessive accumulation of abnormal substrates, or renotrophic factors (6). The clinical diagnosis of diabetic nephropathy in humans does not require histological confirmation. Since only few histological studies were performed in native kidneys of patients with a variable duration of (mainly type 1) diabetes (7,8), the dynamics of the structural changes of diabetic nephropathy in native kidneys are not well documented.
In a previous study in renal allografts, with the unique opportunity to study the histological evolution of diabetic nephropathy in sequential protocol biopsies, mesangial matrix expansion occurred more frequently in patients with pre-transplant (primarily type 2) diabetes mellitus, but independently of glycated haemoglobin levels (9). We hypothesized that the rapid onset of mesangial matrix expansion could be due to effects of diabetes on advanced glycation end product formation, inflammation, hemodynamics, and oxidative stress (10)(11)(12). Also, dyslipidemia, hemodynamic factors, insulin resistance and other metabolic features of the metabolic syndrome could play a role in the development of mesangial matrix expansion (13). Our previous study and other studies also showed mesangial matrix expansion to be related to ageing (9,(14)(15)(16). In the human setting, where many of these factors co-occur, especially in patients with type 2 diabetes and the metabolic syndrome, it remains unclear which factors are most contributing to the mesangial matrix changes.
Since there is a considerable difference in the metabolism of glucose and fructose (17), it was hypothesized that fructose may have a unique contribution to the etiology of obesity, type 2 diabetes mellitus, and cardiovascular disease (18,19). High-fructose-containing diets have also been linked to renal injury and exaggeration of diabetic nephropathy in both humans and animals (20)(21)(22)(23)(24)(25).
To study this in more controlled settings, we recently established a model of type 2 diabetes mellitus induced by a high-sucrose/high-fat (HSHF) diet in female LDLreceptor knockout C57BL/6 J mice (26). These mice were given a HSHF diet to induce the metabolic syndrome and type 2 diabetes mellitus, or a standard chow diet (SC). The HSHF diet induced a metabolic syndrome with gain of weight, hyperinsulinemia, insulin resistance, type 2 diabetes mellitus, and hyperlipidemia (26). Also, diabetic cardiomyopathy with increased heart weight and more specifically features of heart failure with preserved ejection fraction developed upon the HSHF diet (26). Additionally, a gain in kidney weight was observed, consistent with the enlargement in humans with diabetic nephropathy (26). This finding led us to hypothesize that this mouse model also had histopathological features of diabetic nephropathy.
Recently, Denic et al (27) demonstrated that manual morphometry performed on surveillance biopsies five years post-transplantation better predicted subsequent renal allograft failure than the visual scoring by a pathologist. More importantly, in a subset of patients without chronic glomerulopathy, morphometric measure of mesangial matrix expansion (but not the semiquantitative histological mm score) was identified as predictor of allograft failure. This suggests that visual estimation likely underestimates the evolution of mesangial expansion associated with diabetes mellitus.

Objectives
In this study, we aimed to evaluate whether our recently described model of diabetes mellitus type 2 induced by an HSHF diet in female LDL-receptor knockout C57BL/6 mice could be used as valid model for preclinical studies on diabetic nephropathy, using manual morphometry to examine the renal histological changes in this model.

Animal model
At the semi-specific pathogen free facility of KU Leuven, C57BL/6J low-density lipoprotein receptor deficient (LDLr -/-) mice (purchased from Jackson Laboratories (Bar Harbor, ME, USA)) were bred. All mice were female. All mice were fed a standard chow (SC) diet (Sniff Spezialdiäten GMBH, Soest, Germany) or an HSHF western diet (Test Diet 58Y1, TestDiet ® , London, United Kingdom). Metabolizable energy from the SC diet is 13.5 MJ/kg (9 kJ% fat, 24 kJ% protein, 67 kJ% carbohydrates), and from the HSHF diet is 19.5 MJ/kg (46.4 kJ% fat, 17.6 kJ% protein, 36.0 kJ% carbohydrates). In the HSHF diet, mono-and disaccharides consisted of high-fructose corn syrup-55 (17.5%) and sucrose (17.5%) (28). The SC diet was continued in N=10/30 of the mice whereas in the other 20/30 mice the HSHF diet was initiated at the age of 12 weeks and continued for 16 weeks. At the age of 28 weeks (=end of study), all mice were alive and included for analysis (26). and anticoagulated with trisodium citrate. Plasma was then isolated by centrifugation at 1100× g for 10 minutes and stored at -20°C. Methods for quantification of plasma levels of free fatty acids, insulin and tumor-necrosis factor alfa (TNF-α) were previously described (26). In a subset of mice (n=10 of the 20 HSHF diet LDLr -/mice) data on insulin levels, TNF-α and adiponectin were measured, and fatty acids levels were measured in n=18 of the 20 HSHF diet LDLr -/mice, using the techniques previously described. The homeostasis model assessment-insulin resistance (HOMA-IR) index was calculated as measure for insulin resistance using the formula: HOMA-IR = Insulin 0 (µU/mL) x Glucose 0 (mmol/L)/22.5 (29).

Morphology/Morphometry
Paraffin-embedded left and right kidney sections were stained with periodic acid-Schiff and Jones methenamine silver stain. All light microscopic images were evaluated by one pathologist (EL), and the histological changes were described in the glomerular, tubulo-interstitial and vascular renal compartments. In addition, morphometric measurements were performed at Mayo Clinic, Rochester, USA. First, all biopsy slides were scanned into highresolution digital images (Aperio ® AT2 system scanner, Leica Microsystems, Inc., Buffalo Grove, IL; http:// www.aperio.com). Biopsy images were morphometrically analyzed by investigators masked to study group to minimize bias. The scanned digital images were magnified with ImageScope software (version 12.2.2.5015 Aperio ® ) onto a large pen sensitive screen. Only one kidney per mouse was selected for morphometric analysis. Mesangial expansion was assessed in glomeruli in a sampled cortex (approximately 20-25% of the total cortex) ( Figure 1A). At higher magnification, every glomerular cross-sectional profile in the sampled cortex was traced to obtain the number and area of glomeruli ( Figure 1B). Then, in each traced glomerulus, areas of mesangial expansion were also traced ( Figure 1C). Mesangial hypercellularity was defined as ≥3 cells in the mesangium. The percent of mesangial matrix expansion was calculated as the sum of all mesangial matrix expansion areas over the area of all traced glomeruli (including non-affected), as previously reported (27). Furthermore, as additional measure we calculated percent of glomeruli affected with mesangial expansion. The mean cross-sectional tubular area was obtained in a 1 mm 2 of sampled cortex, approximately equidistant between the capsule and cortico-medullary junction (Supplementary file 1, Figure S1A), using the same approach previously applied to kidney donor biopsies (30). For glomerular morphometry measures, we traced the whole cortex and all glomeruli on a single kidney (Supplementary file 1, Figure S1B). From these measurements we calculated mean glomerular area and mean glomerular area density, and using stereology models (31) we calculated mean glomerular volume and glomerular volumetric density.

Ethical issues
All experimental procedures in animals were executed in accordance with protocols approved by the Institutional Animal Care and Research Advisory Committee of KU Leuven (Approval number P191/2015).

Statistical analysis
The investigators who performed the analyses were blinded to group allocation. Variables are displayed as mean ± standard deviation. Morphometric differences between the two groups were tested with Kruskal-Wallis tests. Pearson correlation analysis was used to assess associations between continuous variables and histological lesions. For correlation analysis, samples with missing values were excluded. A correlation matrix was constructed using Pearson correlation coefficients. The SAS 9.4 software (SAS Institute) and R (version 3.6.3, R core team, 2014) were used for statistical analysis. For data presentation GraphPad Prism (version 8; GraphPad Software, San Diego, CA) was used.

Study population characteristics
Biological characteristics of the C57BL/6J LDLr -/mice at 28 weeks (end of study) were described previously (26). At the end of the study (28 weeks), mice on the HSHF diet (n=20) developed features of the metabolic syndrome with gain of weight, hyperglycemia, hyperinsulinemia with increased HOMA-IR index and hyperlipidemia compared to mice fed the SC diet (n=10). Increased levels of TNF-α levels were observed in mice fed the HSHF diet compared to the SC diet. No significant differences were found in free fatty acid or adiponectin levels ( Figure 2A, Table 1).

Morphological findings
Kidney weights were different between the two groups ( Table 1). Morphological evaluation of kidneys from C57BL/6J LDLr -/mice fed the HSHF (n=20) or SC (n=10) diet showed discrete changes. The intraglomerular nuclei (mesangium, visceral epithelium and endothelium) of mice are somewhat larger compared to humans, which gave the impression of crowded glomeruli. On a higher magnification, using criteria for assessment of human glomeruli, discrete mesangial matrix expansion and hypercellularity was seen in some slides of mouse kidneys. These changes were more often found in the mice fed the HSHF diet (n=11/20; 55%) compared to 30% (n=3/10) in mice fed the SC diet. By light microscopy, no glomerulosclerosis, thickening of the glomerular basement membrane, interstitial or vascular lesions were observed.

Morphometry findings
Morphometric analysis confirmed these mesangial matrix changes in the HSHF fed mice more robustly. There were no differences in size or density of the glomeruli. There was no difference in glomerulosclerosis between the two groups; neither in tubular size measurement ( Table 1). The percentage of mesangial matrix expansion was significantly greater in mice fed the HSHF diet compared to the SC  Figure 2B).

Discussion
In this study, we present a novel mouse model of the metabolic syndrome with mesangial matrix expansion compatible with diabetic nephropathy. This model better reflects the human disease with an accumulation of metabolic risk factors in type 2 diabetic patients than e.g. the streptozocin-induced mouse model of pancreatic beta cell toxicity. Patients often fulfil criteria of the metabolic syndrome with increased abdominal obesity together with elevated fasting glucose levels, insulin resistance, elevated blood pressure, and plasma lipids, as was also the case in our mouse model. We demonstrated that the glomerular changes in this mouse model associated with insulin resistance and TNF-α levels, but not with plasma glucose levels, free fatty acid levels or adiponectin. Mesangial matrix expansion has been associated with changes in diabetic nephropathy, such as glomerulosclerosis or arteriolar hyalinosis, were not observed in this study, in contrast to the mesangial matrix effects. This makes this model ideally suited to study the glomerular changes caused by type 2 diabetes, rather than the general ageing changes. Many rodent models for type 2 diabetes and obesity have been described (32), ranging from monogenic models defective in leptin signalling (Lep ob/ob mice, Lepr db/db mice and ZDF rats) to polygenic mouse models of diet-induced obesity, glucose intolerance and diabetes (C57BL/6J mouse strain (33), C57BL/6N , New Zealand Obese (NZO), TALLYHO/Jng and KK mice and also non genetic chemical-induced animal models of diabetes (like the high-fat diet/streptozotocin-treated (HFD/STZ) animal model). For an overview of mouse models of diabetic nephropathy we refer to the detailed reviews by Alpers et al and Breyer et al (34,35). In contrast to our model, most mouse models for type 2 diabetes and diabetic nephropathy report glomerulosclerotic changes. Mouse models of diabetes mimicking early morphological changes of human diabetic nephropathy, mostly mesangial matrix expansion and sometimes, podocyte loss, include db/db and Akita mice (35). However, these models have limitations. db/db mice with leptin deficiency are infertile, making the breeding of mice in sufficient numbers for interventional studies expensive and labor intensive. Akita mice on the other hand are more suitable as a model for observations. Pearson correlation coefficients are given. Blank boxes indicate non-significant correlations (P > 0.05). All cases with missing data were omitted from the analysis. HOMA-IR: homeostasis model assessment-insulin resistance index; TNF-α: tumor-necrosis factor α; FFA: free fatty acids; ME: mesangial matrix expansion. type 1 diabetes and demonstrate IgA deposition (36), which is a confounding the interpretation of the observed mesangial matrix expansion. There are only few models that mimic morphological features of both early and late diabetic nephropathy; the most robust seem to be those with deficiency of eNOS, OVE26 FVB mice, and BTBR ob/ob (35). All mouse models available to date are subject to important limitations and better models of diabetic nephropathy are still needed for deepening our mechanistic insight or for testing novel therapeutic interventions.
The mouse models of diabetes that mimic early morphological changes of human diabetic nephropathy, mainly mesangial matrix expansion, include db/db and Akita mice (35). Interestingly, C57BL/6 mice have been found to be less susceptible to diabetic nephropathy changes (34,35). Balmer et al found that B6 mice display less renal hypertrophy (kidney weight: body weight) response to diabetes compared to mice from another strain (FVB). When they introgressed B6 regions to the FVB strain, these regions protected against diabetes-induced increased kidney weight and glomerulosclerosis (37).
C57BL/6 is a common inbred strain of mice, widely used due to the availability of congenic strains, easy breeding, and robustness.
In our novel mouse model, we found strong correlations between the mesangial matrix measurements and TNF-α levels, a pro-inflammatory cytokine. The association between mesangial matrix expansion and oxidative stress/ inflammation was illustrated before (13,38). Also, studies in mice using glucose-sodium transporter blockade with sodium-glucose cotransporter-2 (SGLT2) inhibitors demonstrated prevention and reversal of mesangial matrix expansion by dampening hyperglycemia-induced oxidative stress and inflammation (39,40). This effect was even demonstrated to be independent of its glucoselowering effect (41,42). In a previous study in kidney transplant recipients, we found apparent independence of glycemic control in developing mesangial matrix expansion (9). Reflecting these findings in humans, also in our novel mouse model, mesangial matrix expansion was not significantly correlated with glucose levels but rather with the metabolic measures of insulin resistance and inflammation. Insulin might indeed be a more important determinant of mesangial matrix expansion than plasma glucose levels alone, as suggested by the higher mesangial matrix expansion incidence in insulin-dependent diabetic patients (9), and by the finding that insulin prevents diabetes-induced hyperfiltration and proteinuria, but could not prevent glomerular growth, and even induced mesangial expansion (43). This might reflect an effect of insulin treatment itself, with more hypoglycemic hyperinsulinemic episodes, a trophic effect of insulin, and more fluctuating glucose profiles.
By assessing both morphological and morphometric features, we demonstrated the increased sensitivity and more detailed evaluation of manual morphometry over visual estimation by light microscopy. These findings are in agreement with the recent study of 175 kidney transplant patients in which semiquantitative lesion scores were compared to morphometric surrogates on 5 year surveillance biopsies (27). Although semiquantitative lesion scores and morphometry overall had significant correlation with each other, at best this correlation was modest. However, most importantly, morphometry performed better in predicting subsequent graft loss better than semiquantitative pathological scoring in terms of model discrimination (c-statistic). Moreover, when analysis was performed in a subset of patients whose biopsies did not feature chronic glomerulopathy, manual morphometry not only had a better c-statistic, but also identified additional lesions that were predictive of graft loss (e.g. mesangial expansion and luminal stenosis) (27). Finally, this study also showed how compared to visual estimation, manual morphometry was more reproducible. This is in agreement with recent studies that showed better reproducibility and reliability when scanned slides were analyzed compared to traditional glass slides on light microscopy (44,45).
There are some limitations to this study. Notably, the morphological findings cannot fully explain the gain in weight in these mouse kidneys. Steatosis might possibly be playing a role in this weight gain, however this could not be examined due to the fixation technique (paraffin embedded slides). Furthermore, there are considerable confounders in this study, especially the collinearity of the glucose metabolism parameters with very high cholesterol levels in this LDL-receptor knockout mouse model. Iacobini et al described mesangial matrix expansion in response to lipid toxicity (13), and our data show correlation of mesangial matrix expansion with total cholesterol, however we found no significant correlations between free fatty acid or adiponectin levels and mesangial matrix expansion measures. Also, Bornfeldt et al described no effect of hyperlipidemia, induced by LDL-R degradation, on mesangial expansion or podocyte numbers in a BTBR mouse strain with leptin deficiency, which has emerged as one of the best models of advanced human diabetic kidney disease (46). Next, there are no standardized definitions of the metabolic syndrome or the HOMA-IR calculation in mice. There were also some missing values for some biological parameters in the HSHF diet group due to the initial study design. Moreover, no data on microalbuminuria was available since no urine samples were collected. Finally, plasma creatinine levels were not measured; therefore, no statements could be made on renal function/hyperfiltration.
In conclusion, we describe a novel, accessible mouse model with features of the metabolic syndrome and development of mesangial matrix expansion. This model is comparable to the human setting and could therefore serve as a relevant experimental model for nephropathy associated with type 2 diabetes mellitus.

Conflicts of interest
The authors declare that they have no competing interests.

Ethical considerations
Ethical issues (including plagiarism, data fabrication, double publication) have been completely observed by the authors.

Data sharing statement
All of the data that underlie the results reported in this article (text, tables, figures, and appendices) can be made available on a collaborative basis following institutional review board approval, immediately following publication, without end date and for any purpose. Proposals should be directed at the corresponding author.

Funding/Support
EVL holds a fellowship grant (1143919N) from the Research Foundation Flanders (FWO). MN is a senior clinical investigator of the FWO (1844019N).

Supplementary Materials
Supplementary file 1 contains Figure S1A-B.