Excessive cholecalciferol supplementation increases kidney dysfunction associated with intrarenal artery calcification in obese insulin-resistant mice

Diabetes mellitus accelerates vascular calcification (VC) and increases the risk of end-stage renal disease (ESRD). Nevertheless, the impact of VC in renal disease progression in type 2 diabetes mellitus (T2DM) is poorly understood. We addressed the effect of VC and mechanisms involved in renal dysfunction in a murine model of insulin resistance and obesity (ob/ob), comparing with their healthy littermates (C57BL/6). We analyzed VC and renal function in both mouse strains after challenging them with Vitamin D3 (VitD3). Although VitD3 similarly increased serum calcium and induced bone disease in both strains, 24-hour urine volume and creatinine pronouncedly decreased only in ob/ob mice. Moreover, ob/ob increased urinary albumin/creatinine ratio (ACR), indicating kidney dysfunction. In parallel, ob/ob developed extensive intrarenal VC after VitD3. Coincidently with increased intrarenal vascular mineralization, our results demonstrated that Bone Morphogenetic Protein-2 (BMP-2) was highly expressed in these arteries exclusively in ob/ob. These data depict a greater susceptibility of ob/ob mice to develop renal disease after VitD3 in comparison to paired C57BL/6. In conclusion, this study unfolds novel mechanisms of progressive renal dysfunction in diabetes mellitus (DM) after VitD3 in vivo associated with increased intrarenal VC and highlights possible harmful effects of long-term supplementation of VitD3 in this population.

To further investigate the effects of VitD 3 protocol in renal dysfunction, we assessed 24h-urine volume output, albuminuria and calculated ACR to estimate glomerular damage, which also depicts progressive renal disease in diabetes mellitus 16 . As expected, the diabetic model showed increased 24h-urine volume in comparison to C57BL/6 (1.49 ± 0.06 mL vs. 1.14 ± 0.17 mL) at baseline. After 21 days of VitD 3 treatment, both ob/ob and C57BL/6 strains presented decreased 24h-urine volume. This was more pronounced in ob/ob (0.31 ± 0.11 mL) than in C57BL/6 mice (0.74 ± 0.21 mL), Fig. 1B.

BMP-2 is highly expressed on calcified intrarenal artery from Vitamin D 3 -treated diabetic mice.
As we consistently demonstrated increased VC induced by VitD 3 in diabetic mice (Figs. 2 and 3), we evaluated BMP-2 expression, which has a pivotal role in smooth muscle cells osteochondrogenic dedifferentiation and ectopic calcification 17,18 . We found that BMP-2 was highly expressed in intrarenal arteries exclusively from ob/ob mice subjected to VitD 3 protocol (Fig. 4), but not in paired VitD 3 -injected C57BL/6 samples.

Severe mesangial expansion associated with acute tubular necrosis in VitD 3 -treated ob/ob mice.
In diabetic ob/ob mice, mesangial expansion was more severe when compared to wild type mice, Fig. 5A. Of note, VitD 3 accelerated extracellular matrix deposition in mesangial compartment of C57BL/6, demonstrated by statistically similar values of fractional mesangial area in VitD 3 -treated C57BL/6 compared to VitD 3 -treated ob/ ob mice, Fig. 5B. Both VitD 3 -treated C57BL/6 and VitD 3 -treated ob/ob mice's fractional mesangial area did not show statistical difference when compared to paired saline-treated C57BL/6 and ob/ob mice respectively, Fig. 5B. These findings may suggest a possible effect of inappropriate high-dose of VitD 3 in regulating glomerular damage in C57BL/6 mice, since VitD 3 promoted increased matrix deposition in mesangial compartment that was able to reach ob/ob levels. However, in pre-existing diabetic nephropathy settings, in which several deleterious pathways www.nature.com/scientificreports www.nature.com/scientificreports/ have already been activated, VitD 3 did not promote additional damage to the mesangial compartment. In parallel, we showed increased glomerular area in ob/ob mice. There was no additive effect of VitD 3 in ob/ob mice, but VitD 3 was able to equalize glomerular area in C57BL/6 and in ob/ob mice, Fig. 5C. Importantly, VitD 3 also induced acute kidney injury, e.g., ATN both in ob/ob and in wild type mice. Cortical and corticomedular tubules showed widespread degenerative changes with luminal dilation and loss of brush border, as well as tubular atrophy, Fig. 5D. These findings may explain the observation of higher serum creatinine levels and lower creatinine clearance and diuresis volume in VitD 3 -treated animals. Importantly, increased mesangial area in ob/ob mice combined with ATN lesions may explain higher levels of albuminuria, impairing kidney function in these animals.
Collagen I and III deposition in tubule-interstitial area, as demonstrated by Picro Sirius Red staining, was mild in all groups, but slightly increased in VitD 3 -treated ob/ob mice, Supplemental

Discussion
There is no doubt that the association of DM, VC and ESRD increases morbidity and mortality 19 . In order to investigate kidney-specific VC mechanisms that may implicate in diabetes mellitus-related kidney dysfunction and pathophysiology, we studied the effect of high dose VitD 3 i.p. injection using an experimental model that mimics T2DM and insulin resistance, the ob/ob mouse. Recently, we demonstrated that VitD 3 increased expansive vascular remodeling associated with accelerated VC in ob/ob mice, which occurred by the convergence of increased oxidative stress, matrix metalloproteinase activation and absence of VDR downregulation in the vascular wall from these animals 14 . Conversely, VitD 3 and its analogous have been reported as renoprotective, resulting in an attenuation of proteinuria, inflammation, glomerulosclerosis and interstitial fibrosis and an improvement of glomerular ratio 20,21 . Reciprocal benefits include renin-angiotensin-aldosterone system regulation, anti-inflammatory, anti-oxidative stress and anti-apoptosis properties, podocyte protection, autophagy activation, immunomodulatory effects, hepatocyte growth factor induction, mitochondrial function regulation, and www.nature.com/scientificreports www.nature.com/scientificreports/ tubular epithelium preservation, via epithelial-mesenchymal transition blockade [20][21][22] . Vitamin D 3 prescription has been increasing and its inappropriate supplementation may be associated with toxicity and hypercalcemia 23 . Recently, investigators reported an interesting ESRD murine model 24 , using phosphorus-rich diet, but it requires twice more time (6 weeks) to develop compared to our model, besides not describing intrarenal arteries calcification. In our study we demonstrated that: 1) VitD 3 additionally impacted in diabetic mouse model's renal function and early markers of glomerular filtration damage, by decreasing creatinine clearance, 24 h urine volume output and by increasing albumin/creatinine ratio in ob/ob in comparison to paired C57BL/6 mice; 2) renal dysfunction occurred in parallel to exaggerated effect of VitD 3 in increasing intrarenal VC in ob/ob mice vs. C57BL/6 mice; 3) Increased BMP-2 expression in the vascular wall of calcified intra-renal arteries from ob/ob mice, but not from C57BL/6 animals after VitD 3 protocol; 4) increased mesangial expansion in ob/ob mice, without an additive effect of VitD 3 , associated with severe acute tubular necrosis in VitD 3 -treated ob/ob mice, contributing to higher levels of albuminuria, and decreased kidney function.
Hypercalcemia, due to increased intestinal and renal calcium resorption is an expected effect of VitD 3 25,26 supplementation at high serum levels 27 . Hypercalcemia has been extensively studied in this context, since this is a mechanism involved in VC due to high calcium x phosphorus product 14,28 . To further understand increased VC response of ob/ob mice after our protocol, we performed serum biochemical analysis and assessed bone histomorphometric parameters. We found a significant, but similar increase in serum calcium concentration (≈60% increase), both in ob/ob and in C57BL/6 mice after VitD 3 administration when compared to saline-injected mice 14 . Moreover, both VitD 3 and kidney dysfunction-induced effect on bone tissue damage, observed in our study, are demonstrated by increased osteoid matrix and resorption parameters, which characterizes a mixed bone disease 29 . Bone disease, as shown by parameters of bone resorption (e.g. osteoblastic surface, resorption surface, and diminished number of osteoclasts) was similar in both strains. This implies the reproducibility of the effect of VitD 3 in bone/calcium metabolism both in ob/ob and in C57BL/6 mice, conceiving the idea that augmented VC in ob/ob mice translates an individual response from the obese insulin resistance model. A limitation of our study is that we were not able to precisely determine absolute values of serum 25-hidroxyvitamin D both in C57BL/6 and in ob/ob mice after VitD 3 stimulation. In this setting, we found that 25-hidroxyvitamin D serum concentration increased to above 100 ng/mL (the limit of the standard curve used, data not shown) in both strains after excessive VitD 3 supplementation. In fact, although we used less VitD 3 in ob/ob mice versus C57BL/6 mice, these animals showed increased calcification of intrarenal arteries associated with greater kidney dysfunction in comparison to C57BL/6 mice. VC, which is a condition without specific medical treatment, positively associates with coronary artery disease and cardiovascular events, especially in patients with diabetes mellitus 30 . Interestingly, now we showed an association of augmented intra-renal arteries calcification and renal dysfunction in ob/ob mice, demonstrated by positive correlation between intrarenal arteries calcification and increased albumin/creatinine ratio, and decreased creatinine clearance. This experiment suggests a direct relationship between arterial calcification and decreased renal function induced by hypervitaminosis D. Furthermore, we demonstrated that high-dose VitD 3 stimulation promoted a paradoxical increase in mesangial area in wild type mice. Thus, our work unveils an important effect of VitD 3 on mesangial compartment in a toxic dose-dependent manner. We postulate that this effect may www.nature.com/scientificreports www.nature.com/scientificreports/ include VDR aberrant down-regulation in that compartment and probably in podocytes as well. In ob/ob mice, VitD 3 treatment was not implicated in an additive effect in mesangial compartment expansion and in glomerular area augmentation, which may be explained at least by the fact that animals were not treated with insulin and hyperglycemia may continuously aggravated mesangial expansion and abrogated VitD 3 -mediated renoprotective effects. Furthermore, VitD 3 toxicity may induce renal hemodynamic dysfunction, which can be complicated by ATN. Tubular damage could also be explained by diabetes-induced lysosomal dysfunction in proximal tubules 31 . Therefore, as a surrogate marker of diabetic kidney disease progression, the augmentation in exocytosis-mediated urinary megalin excretion creates a vicious cycle of tubular damage and may also contribute to higher values of albuminuria found in VitD 3 -treated ob/ob mice. Other possible mechanisms include acute hypercalcemia induced by VitD 3 toxicity, which may lead to acute kidney injury by decreasing extracellular fluid volume due to anorexia, nausea, vomiting, and decreased ability to concentrate urine, besides a direct renal vasoconstriction effect, as observed in VitD 3 -treated animals from our protocol, and in vitamin D intoxication in humans 13 . VitD 3 effects are mediated by VDR. Previously, we demonstrated that VDR downregulation was abrogated in ob/ob mice after VitD 3 stimulation, which corroborated to increased VC (aorta) in this mouse model 14,15 . In the kidneys, VDR is mainly expressed in proximal and distal tubular epithelial cells, podocytes, macula densa of the juxtaglomerular apparatus, and collecting duct epithelial cells 32 . However, VDR expression is low in glomerular mesangial cells. Despite the low expression of VDR in mesangial compartment, VitD 3 may reduce mesangial cell Bars = 20 µm. (B) VitD 3 promoted increased matrix deposition in mesangial compartment of C57BL/6 mice that was able to reach ob/ob levels (no statistical difference among VitD 3 -treated C57BL/6 and VitD 3 -treated ob/ob mice). (C) Increased glomerular area in ob/ob mice. There was no additive effect of Vitamin D 3 in ob/ ob mice, but VitD 3 was able to equalize glomerular area in C57BL/6 and in ob/ob mice. α, β, γ, δ = P < 0.05, in comparison to saline-treated C57BL/6 (n = 9), VitD 3 -treated C57BL/6 (n = 7), saline-treated ob/ob (n = 7), and VitD 3 -treated ob/ob mice (n = 8), respectively. (D) Vitamin D 3 induced acute tubular necrosis (ATN) both in C57BL/6 and in ob/ob mice, as shown by flattening of the renal tubular cells due to tubular dilation (asterisks), loss of brush border (arrows), and degenerative changes characterized by diffuse denudation of the renal cells, presence of necrotic cells, and cellular debris. Insert depicts degenerative changes in a distal tubule. Bars = 20 µm. For saline-injected mice n = 4-6, and for Vitamin D 3 -treated mice, n = 5-8. www.nature.com/scientificreports www.nature.com/scientificreports/ proliferation induced by hyperglycemia in diabetic rat via mTOR pathway modulation and decreasing glomerular volume 33 . In addition, VitD 3 suppressed Monocyte Chemoattractant Protein-1 (MCP-1) expression in mesangial cells by blocking Nuclear factor kappa B (NF-κB) activation, which indicates that vitamin D may protect the kidney by reducing macrophage infiltration 34 . Moreover, VitD 3 per se was able to increase VDR expression at mRNA and protein levels in mesangial cells cultured with either low or high glucose 34 .
Diabetes mellitus may activate specific osteochondrogenic signaling involved in vascular smooth muscle cells dedifferentiation into osteoblast-like cells, thus increasing VC progression. To further investigate this specifically in intrarenal arteries, we assessed BMP-2 expression in the kidney. BMP-2 is a protein secreted by smooth muscle vascular cells, endothelial cells and inflammatory macrophages 35 . BMP-2 binds to its receptor on the plasma membrane, triggers a specific intracellular signaling cascade, activating osteochondrogenesis and VC. Of note, BMP-2 expression is also regulated by the BMP signaling pathway itself, since BMP-2 is a self-regulatory protein 36 . In diabetic preclinical animal models, VC is governed by increased MSH homeobox 2 (Msx2) and Tumor Necrosis Factor Alpha (TNF-alpha) expression in the middle layer and in the adventitia of vessel wall, finally activating osteochondrogenic transcription program in smooth muscle cells. Previous reports demonstrated that increased macrophage infiltration in the adipose tissue and in the kidney, as well as augmented TNF-alpha expression in animal models of obesity, play a pivotal role in inducing kidney disease 37,38 . Accordingly, inflammatory cytokines e.g. C-C chemokine ligand 2 (CCL2) initiates inflammation by binding to C-C chemokine receptor 2 (CCR2) to induce diabetic glomerular sclerosis 39 . Moreover, investigators showed that blocking CCL2/ CCR2 signaling pathway can ameliorate renal injury and proteinuria in a mouse model of obesity and insulin resistance 37 . BMP-2 activation has also been well characterized in aorta from diabetic mice 40 . In our laboratory, vascular smooth muscle cells isolated from ob/ob mice aortae showed increased calcification in vitro after BMP-2 stimulation with concurrent upregulation of other osteogenic proteins 41 . Accordingly, we showed augmented baseline MSX2, BMP-2 expression and Smad 1,5 phosphorylation as well as increased BMP-2-induced osteocondrogenic signaling and dedifferentiation in ob/ob vascular smooth muscle cells in comparison to paired C57BL/6 vascular smooth muscle cells 41 . In the present study, we identified a significant increase in BMP-2 expression in calcified intrarenal arteries of VitD 3 -treated ob/ob mice, which also demonstrated renal dysfunction. On the contrary, paired VitD 3 -treated C57BL/6 mice demonstrated that BMP-2 expression was very low in intrarenal arteries. We speculate that VitD 3 induces BMP-2 expression, through the activation of its receptor. To assess the role of leptin-deficiency in vascular smooth muscle cells calcification from ob/ob mice, we demonstrated that calcification increased in leptin-incubated ob/ob vascular smooth muscle cells, but not in leptin-incubated C57BL/6 vascular smooth muscle cells (Supplemental Fig. II). On the contrary of C57BL/6 vascular smooth muscle cells, co-incubation of leptin in BMP-2-treated cells further augmented mineralization in ob/ob vascular smooth muscle cells. These data suggest mechanisms of increased susceptibility of ob/ob mice to develop accelerated vascular calcification, independently of leptin supplementation. A putative explanation for the exaggerated response of ob/ob mice could be that CYP27B1 (1-α-hydroxylase) and CYP24A1 (24-hydroxylase), which are enzymes responsible for regulating active VitD 3 metabolites, exhibit altered mRNA and activity levels, thus favoring increased local concentration of bioactive VitD 3 and consequently potentiating BMP-2 expression in leptin-deficient diabetic mice. Of note, non-hepatic 25-hydroxylases (Cyp2R1 and Cyp27A1), enzymes responsible for the conversion of cholecalciferol into 25-hydroxyvitamin D and for maintaining VDR levels 42 , may have contributed to changes in VDR expression levels and the phenotype demonstrated in our model, because 25-hydroxylases lack the tight control that exists for 1-hydroxylase and 24-hydroxylase during excessive cholecalciferol supplementation. Altogether, we postulate that augmented intra-renal calcification potentiates progression of renal dysfunction in ob/ob mice. Moreover, a mechanism involved could be worsening of Windkessel effect, because of increased arterial rigidity and decreased vascular elasticity, which impairs tissue perfusion due to VC 24,43 . In addition, vascular mineralization may act as an adjunct etiology of this process or even in its progression 44,45 , adding importance to this study. Nonetheless, we did not find glomerular calcification in ob/ob mice, but we cannot rule out the hypothesis that decreased tissue perfusion and increased BMP-2 expression could have impacted in glomerular dysfunction, by altering podocytes number and/or function [46][47][48] and increasing inflammation 19,49 . A limitation of our study is that we can't distinguish whether systemic and local VitD 3 effects, e.g. increased serum calcium levels, augmented BMP-2 expression and intrarenal arteries calcification influenced renal dysfunction alone or whether this occurred together with hemodynamic imbalance due to vascular mineralization. This needs further clarification by exploring respective individual impact on renal disease progression. In conclusion our results demonstrate that high-dose VitD 3 administration in ob/ob mice, but not in C57BL/6, induce increased intrarenal VC associated with kidney dysfunction. These conditions are common in diabetic patients, bringing high morbidity and mortality 50,51 . Moreover, pathophysiological and molecular mechanisms investigated in this study represent an important contribution both to understand VitD 3 biological properties in the kidneys and to clarify specific aspects of renal disease in diabetic patients, which usually present with VC 7,8 . Our model may be instrumental for the investigation of new therapeutic targets and/or to develop compounds that attenuate renal dysfunction in diabetes mellitus, especially in the context of increased VC.

Materials and Methods
Animals. We used 16 to 20-week-old male homozygous leptin-deficient ob/ob mice (C57BL/6 background) from Jackson Laboratory (Bar Harbor, ME). A total of 30 ob/ob and 30 C57BL/6 littermates were used to perform the experiments. This study was conducted after approval of the protocol #2242-14 by Sociedade Beneficiente Israelita Albert Einstein's ethics committee and undertook according to guidelines for the care and use of laboratory animals, which conforms to Guide for the Care and Use of Laboratory Animals (NIH Publication, 8 th edition). After 21 days, animals were euthanized with 1 mg/kg IM xylazine chlorohydrate (Bayer, São Paulo, Brazil, Cat#1002181) and 100 mg/kg IM ketamine chlorohydrate (Cristália, São Paulo, Brazil, Cat#404800). Of note, not all 60 animals were represented in all experiments, due to technical difficulties as follow: (i) we were (2020) 10:87 | https://doi.org/10.1038/s41598-019-55501-3 www.nature.com/scientificreports www.nature.com/scientificreports/ not able to draw blood samples from all animals (dehydration, samples with clots, unsuccessful vein puncture); (ii) automated Abbott ® i-STAT Clinical Analyzer failed to give a result in some samples (equipment error) and repeated measurement were not possible due to insufficient sample/blood volume; (iii) we did not have as many metabolic cages as the number of animals used during protocol, in order to collect 24h-urine from all animals; (iv) some animals were anuric, so we were not able to determine albuminuria in some samples/animals, (v) we did not use all animals to perform all the experiments.
Vitamin D 3 administration protocol. Ob/ob and C57BL/6 male mice were injected with VitD 3 6.4 × 10 4 IU/day and 4.4 × 10 4 IU/day respectively i.p. for 18 days and another 3 days of sodium chloride 0.9% i.p. This corresponds to a daily dose of 1.46 × 10 3 IU/g/day administered to C57BL/6 mice and 1.06 × 10 3 IU/g/day administered to ob/ob mice, considering a body weight of 30 g and 60 g respectively. Sodium chloride 0.9% (saline) i.p. only was used for 21 days in controls. Of note, we did not use the same proportional dose (calculated by body weight) in ob/ob and in C57BL/6 mice, because 80% of ob/ob mice died when we used the aforementioned proportional dose, due to Vitamin D intoxication after 7 to 10 days of protocol. Consequently, we reduced the dose in approximately 30% in ob/ob mice to be able to complete the animal protocol.
Serum and urine analysis. Blood was collected from the animals before and after the protocol in order to assess serum glucose, urea, creatinine and calcium levels using Abbot ® i-STAT Clinical Analyzer and the I-STAT CHEM8+ cartridge (Abbott Laboratories, Illinois, USA, Cat#AB-9P3125). For urine analysis, animals were maintained in a metabolic cage to collect 24-hour urine before and after the protocol. Urine albumin levels were determined by Albumin Mouse ELISA Kit (Abcam, Cat#ab108792). Urine creatinine levels were evaluated by a colorimetric assay Labtest Diagnostics kit (Vista Alegre, Brazil, Cat#10009010034) and quantified by an automatic biochemical analyzer Cobas Mira Plus (Roche, Switzerland).

Bone histomorphometric analysis.
After euthanasia, left femurs were dissected, fixed in 70% ethanol, dehydrated, embedded in methyl methacrylate, and sectioned longitudinally with a Policut S microtome (Reichert-Jung, Heidelberg, Germany) in 5 µm-thick sections. Samples were stained with 0.1% toluidine blue (pH 6.4) for histomorphometric analysis which was performed using semiautomatic method with a Labophot-2A Ex vivo calcification assessment of the kidneys using Osteosense 680 EX. 24 h before sacrifice, mice were injected with 0.2 µM Osteosense 680 EX (NEV10020EX, Perkin Elmer, USA). After euthanasia, cardiovascular system from mice was perfused with saline, followed by radical nephrectomy. Isolated kidneys were analyzed with a fluorescence detector IVIS ® Lumina LT Series III (PerkinElmer, USA), and fluorescence signals were normalized to estereoradian and ROI of 4.0 cm 2 and converted to photon/s 53 .

Histological quantification of kidneys' vascular calcification. Kidneys were fixed in 10%
phosphate-buffered formalin (pH 7.4), embedded in paraffin and processed for Von Kossa (silver nitrate Sigma, Cat# S1179) and Alizarin Red S (Sigma, Cat#A5533) analysis using a FSX100 microscope (Olympus Life Sciences) and Olympus CellSens software. Kidneys from animals previously labeled with Osteosense 680 EX, were extracted, incubated with sucrose (Sigma, Cat#S9378) 30% overnight, frozen at −80 °C in O.C.T. (optimum cutting temperature) compound (Sakura Finetek, California, USA, VWR Cat# 25608-930) and sections were obtained using a cryostat microtome (Leica Biosystems, Germany). Samples were incubated with Hoechst 33342 (Thermo Fischer Scientific, Cat#H1399) for nuclei staining, and analyzed by Zeiss LSM 710 laser scanning microscope for confocal imaging and the respective software Zen Image (Zeiss ® ). Osteosense fluorescence quantification was calculated by using the fluorescence intensity divided by vascular area and normalized by C57BL/6 control.
Immunofluorescence analysis of Bone Morphogenetic Protein-2 (BMP-2) expression in the vascular wall. 10 µm O.C.T. kidney's sections were fixed with 4% paraformaldehyde and permeabilized with 0.25% Triton X-100 and PBS for 15 min. Samples were washed, incubated with primary antibody overnight (anti-BMP-2, Abcam, ab:6285) 5 µg/mL, and finally incubated with secondary anti-mouse IgG AlexaFluor 488 (Thermo Fischer Scientific, Cat#A10680) 10 µg/mL for 1 hour, and coverslipped with Hoechst 33342 (Thermo Fischer Scientific, Cat#H1399). Fluorescence image analysis was performed in parallel with controls, using the same settings for all samples in a Zeiss LSM 710 confocal laser scanning microscope and the respective software Zen Image (Zeiss ® ). BMP-2-derived fluorescence quantification was calculated by using the fluorescence intensity divided by vascular area.
Mesangial and histological assessment of the kidneys. Kidney sections were stained with periodic acid-Schiff (PAS) trichrome staining in each experimental group. Sections were then analyzed by light microscopy (magnification, x400). A quantitative analysis of mesangial expansion was performed. The increase in mesangial matrix was determined by the presence of PAS-positive area in the mesangium, and was expressed in percentage. The glomerular area (μm 2 ) was also traced along the outline of capillary loops using CellSens software (Olympus) in 28-35 randomly selected glomeruli in each animal. Acute tubular necrosis (ATN) was identified by