Diabetic nephropathy associates with deregulation of enzymes involved in kidney sulphur metabolism

Abstract Nephropathy is a major chronic complication of diabetes. A crucial role in renal pathophysiology is played by hydrogen sulphide (H2S) that is produced excessively by the kidney; however, the data regarding H2S bioavailability are inconsistent. We hypothesize that early type 1 diabetes (T1D) increases H2S production by a mechanism involving hyperglycaemia‐induced alterations in sulphur metabolism. Plasma and kidney tissue collected from T1D double transgenic mice were subjected to mass spectrometry‐based proteomic analysis, and the results were validated by immunological and gene expression assays.T1D mice exhibited a high concentration of H2S in the plasma and kidney tissue and histological, showed signs of subtle kidney fibrosis, characteristic for early renal disease. The shotgun proteomic analyses disclosed that the level of enzymes implicated in sulphate activation modulators, H2S‐oxidation and H2S‐production were significantly affected (ie 6 up‐regulated and 4 down‐regulated). Gene expression results corroborated well with the proteomic data. Dysregulation of H2S enzymes underly the changes occurring in H2S production, which in turn could play a key role in the initiation of renal disease. The new findings lead to a novel target in the therapy of diabetic nephropathy. Mass spectrometry data are available via ProteomeXchange with identifier PXD018053.

changes occurring in diabetic nephropathy, we chose to employ double transgenic diabetic mice (dTg) that, between the 3rd and the 8th week of age, develops gradually T1D with somatic, metabolic and vascular disorders comparable to human diabetes. [3][4][5] These characteristics allow the study of the etiopatogenic mechanisms of early diabetic nephropathy, the understanding of which could open up the possibility of finding new targets for the therapy of this disease.
Hydrogen sulphide (H 2 S) is produced redundantly by the kidney and has a crucial role in renal physiology (ie regulation of the excretory function of the kidney, release of renin and oxygen sensor) and pathology (ie renal ischaemia/reperfusion, obstructive or hypertensive nephropathy). H 2 S is a colourless, water-soluble and membrane-permeable gas, considered the third gaseous signalling molecule after nitric oxide and carbon monoxide. It is generated through the cysteine metabolism by cystathionine b-synthase (Cbs) and cystathionine g-lyase (Cth) enzymes or by the concerted action of cysteine aminotransferase, D-amino acid oxidase (Dao) and 3-mercaptopyruvate sulphurtransferase (Mpst). Recent studies reveal a complex role for H 2 S in many biochemical processes taking place during angiogenesis and carcinogenesis. H 2 S has a protective effect being a potent antioxidant, anti-apoptotic, anti-inflammatory and oxidative stress regulator. 6 In streptozotocin-treated murine animal models, H 2 S has been reported as a protective molecule, playing a key role in the reduction of endothelial dysfunction, cardiomyopathy and nephropathy. [7][8][9][10][11] Addressing the role of H 2 S in diabetes, Xie et al 7 demonstrated for the first time that H 2 S suppresses diabetes-induced accelerated atherosclerosis. 12,13 However, there are no consistent data on the H 2 S role and its concentration in the plasma and different tissues of the diabetic patients and experimental animal models. 14,15 Some studies argue that the plasma H 2 S level in streptozotocin-induced diabetic rats is reduced 16 while others, on similar animal models report unchanged plasma levels of H 2 S, although insulin administration results in increased plasma H 2 S concentration. 15 Thus, due to its importance, we examined the changes occurring in the plasma and kidney concentration of H 2 S level in early T1D and questioned the underlying mechanisms of this alteration. We hypothesize that hyperglycaemia deregulates the enzymes involved in H 2 S metabolism, which in turn change the equilibrium of H 2 S production that could play an important role in the onset of early kidney dysfunction in diabetes.
To test this hypothesis, we designed experiments using a double transgenic diabetic mouse model and high-performance mass spectrometry-based proteomic analysis. Hereby, we present data showing that in diabetes there's an increase in circulating H 2 S and the diabetic kidney exhibits significantly differentially expressed enzymes implicated in sulphur metabolism. The results provide new markers of kidney dysfunction that could become novel therapeutic targets in T1D complications.

| MATERIAL S AND ME THODS
All chemicals used for liquid chromatography (LC) and mass spectrometry (MS) experiments were of LC-MS grade. Trypsin Gold was purchased from Promega. Urea, sodium deoxycholate (DOC), Trizma hydrochloride (Tris), DL-dithiothreitol (DTT), iodoacetamide (IAA), N-acetyl-L-cysteine (NAC), ethylenediaminetetraacetic acid (EDTA), bovine serum albumin (BSA), ammonium bicarbonate and all solvents were provided by Sigma-Aldrich, unless otherwise specified. Complete protease inhibitor cocktail was offered by Roche. C18 solid-phase extraction (SPE) columns were acquired from Waters Corporation. Protein concentration was determined by amido black assay using Amido Black 10B (JT Baker Chemical Co.). Cholesterol, glucose and triglyceride plasma levels were determined using the specific assays from DIALAB GMBH. The concentration of renal plasma and renal tissue H 2 S were determined by the H 2 S Assay Kit E-BC-K355 (Elabscience) according to the manufacturer's instructions.

| Experimental animal models
Twenty transgenic mice were from the 'Cantacuzino' National Institute of Research and Development for Microbiology and Immunology (Bucharest, Romania) and housed in suitable facility under controlled temperature, humidity and 12 hours light cycle. The Ins-HA ± /TCR-HA ± double transgenic Balb/c mice (dTg) were obtained as previously described. 3,4 Briefly, the mice express simultaneously the hemagglutinin (HA) of influenza virus under the rat insulin promoter (Ins-HA) and T cell receptor (TCR) specific for the immunodominant CD4 T cell epitope of HA (HA110-120) in pancreatic β-cells. Since the Ins-HA +/+ Tg mice were homozygous and the TCR-HA ± Tg mice were heterozygous, the offspring were either Ins-HA ± /TCR-HA ± (dTg) mice (which developed diabetes) or Ins-HA ± /TCR-HA -/-(sTg) mice (which did not develop diabetes). As controls, we employed both, the single transgenic, Ins-HA ± / TCR-HA -/-(sTg) and the wild-type Balb/c mice (WT). The animals were divided in three experimental groups, ten animals per group: (a) double transgenic mice (dTg) that developed autoimmune T1D; (b) single transgenic mice (sTg); and (c) BALB/c mice (WT). During the experiments, the animals had free access to standard diet and freshwater. After 12 weeks, the animals were evaluated for changes in the bodyweight and plasma glucose concentration. When hyperglycaemia reached high level in the dTg group (435 ± 108 mg/dL), the mice were anesthetized via an i.p. injection with a ketamine (100 mg/kg bodyweight)/xylazine (10 mg/kg bodyweight) cocktail for mice. After the blood was collected on 5 mmol/L EDTA through a ventricular puncture, the mice were perfused with phosphate buffered saline (PBS) and the kidneys were

| Histological studies
Immediately after the standard cardiac perfusion with PBS, the left kidney of each animal was fixed by immersion in 10% formalin for paraffin embedding. The 6-µm-thick sections, obtained with a

| Protein extraction
From each animal, thirty milligrams of renal cortex tissue was mechanically homogenized (1 minutes) on ice in 0.3 mL buffer containing 8 M urea, 1% DOC and 100 mmol/L Tris-HCl (pH 7.5) and the samples were thereafter centrifuged (11 000 ×g) at 4°C for 10 minutes. The supernatants were collected and stored at −30°C until use.
All protein extracts were subsequently used for biochemical, mass spectrometry and Western blot (WB) quantification experiments.

| Preparation of kidney tissue samples for mass spectrometry analysis
After solubilization, 50 µg of proteins from each sample was further purified by acetone precipitation and incubated for 60 min- Proteolysis was performed overnight, at 37°C, using a 1:20 trypsin to substrate quantity ratio. Formic acid was added to the resulted peptide mixtures up to pH 2.5 for enzyme inhibition and DOC precipitation. The sedimented ionic detergent was discarded by centrifugation for 20 minutes, 20 000 ×g at room temperature.
The peptides were desalted using SPE and concentrated with the Concentrator plus system (Eppendorf).

| Database protein identification, label-free quantification and data mining
Protein identification was performed using Proteome Discoverer 1.4 (Thermo Scientific) and Mascot 2.4.1 (Matrix Science), and the F I G U R E 1 Validation of the diabetic double transgenic mice (dTg) model compared to controls, the non-diabetic single transgenic (sTg) and BALB/C wildtype (WT) mice. Plasma glucose (A) and cholesterol measurements (B) performed for each experimental group. H 2 S levels in the plasma (C) and the kidney tissue homogenate (D) from WT, sTg and dTg mice, respectively. Note that, type 1 diabetes is accompanied by changes of H 2 S concentration in the plasma and the renal tissue. Data are expressed as means ± standard deviation (SD); *P ˂ .05, **P ˂ .01 and ***P ˂ .001 taxonomy was set on Mus musculus organism in UniProtKB/Swiss-Prot fasta database, build 11.2017. A maximum of 2 missed cleavage sites was allowed. Oxidation of methionine and deamidation of asparagine and glutamine were enabled as variable modifications while carbamidomethylation of cysteine was set as fixed modification. False discovery rate target was set below 0.05. SIEVE 2.1 software (Thermo Scientific) was used for label-free relative quantification of all identified proteins. A t test was employed to calculate p-value and to detect significant regulations (dTg/WT ˃1.5 or sTg/ WT ˃1.5 or and dTg/WT ˂0.67 or sTg/WT ˂0.67 and P-value ˂.05) over ten biological replicates. Protein Center v 3.12 bioinformatics platform was utilized for projection of the quantitative data onto Kyoto Encyclopedia of Genes and Genomes (KEGG) signalling pathways and determine their over-representation (FDR P-value <0.05).
Proteins identified with a high Mascot score (>100) in the overrepresented KEGG signalling pathways (FDR P-value <2.04E-02) in diabetic samples were further used for validation using as alternative methods, Western blot (WB) and gene expression assays.

| Western blotting assay
Equal amounts of the renal tissue protein were loaded and separated by electrophoresis using 12.5% SDS polyacrylamide gels (SDS-PAGE). The proteins were subsequently transferred onto nitrocellulose membrane and analysed by Western blotting (WB) assay.
The membranes were first exposed (2 hours) to the following pri- β-actin (mouse monoclonal, ab6276-Abcam, dilution 1:1000) in TBS with 1% BSA. This was followed by the appropriate secondary antibodies (1 hour), IgG coupled with horseradish peroxidase (IgG-HRP). Beta-actin was used to normalize the semi-quantitative data. The reaction product was detected with the ECL Western blotting Substrate kit, and the protein bands were digitally detected (ImageQuant LAS4000) and quantified by densitometry with Scion Image software.

| Gene expression analysis
Total RNA was extracted from kidney tissues using the RNeasy Mini Kit (Qiagen). Its quality was assessed using Agilent 2100 Bioanalyzer (Agilent Technologies) and quantified by NanoDrop ND 1000 spectrophotometer (Thermo Scientific). The cDNA was generated from 1 μg total RNA using Transcriptor First Strand cDNA Synthesis Kit (Roche). Light Cycler 480 SYBR Green I Master mix was used to perform real-time PCR analysis in the Light Cycler System (Roche). All reactions were performed in triplicate, and the product specificity was validated by melting-curve analysis. Amplification of housekeeping gene b-actin was used for normalization. The primer sequences used were as follows: for Cbs 5′-GCTGGAACCTGCTCCTTTTC-3′ (forward) and 5′-TGCATGTCCAAGTGCTGGAA-3′ (reverse); for Soux 5′-GTCATGGGGACCCTGTTAGG-3′ (forward) and.

| Statistics
All the data were generated by running samples in triplicate and expressed as a mean ± SD. The results were analysed by Student's unpaired t test using GraphPad Prism 5.0 statistical software (GraphPad Software). Significance was defined as a Pvalue <.05.

| Type-1 diabetes induces subtle structural changes of kidney tissue
Staining of renal tissue sections with haematoxylin-eosin showed no histological abnormalities in the kidney isolated from the control groups (WT and sTg). Diabetic kidney tissue from the dTg group exhibited an apparently normal structure of the glomeruli and the Bowman's capsules, as well. However, the Masson's trichrome staining (Figure 2A-C) evidenced minor variations in the tubulointerstitial structure of the diabetic kidney. Thus, in the dTg kidney sections we detected the presence of interstitial fibrosis surrounding the uriniferous tubules ( Figure 2C, blue staining) and a ~10% (P ≤ .001) increase of the mean area of glomeruli compared to control samples ( Figure 2D).

| Modifications of kidney proteome in T1D revealed by global shotgun proteomic analysis
Nano-liquid chromatography coupled with tandem mass spectrometry (nLC-MS/MS) experiments were applied on samples of homogenates of renal cortex isolated from dTg, sTg and WT mice. The data led to the identification of 351 proteins that were uniquely identified in dTg, 467 were found only in sTg, while 549 were uniquely attributed to the WT group ( Figure 3A). Among all the identified proteins (4261), we also found (average Mascot Score >100) nine of the most frequently studied urinary biomarkers for diabetic nephropathy, 17 supporting the reliability of our method (listed in Table S1 as supplementary data).
The dTg and sTg proteomes were normalized over the WT mice and the resulting differentially abundant proteins were retained for further comparative analysis ( Figure 3B). Thus, 106 proteins (7 up-regulated and 99 down-regulated) were found to be uniquely and differentially expressed in sTg/WT, while 144 (120 up-regulated and 24 down-regulated) exhibited a significantly altered abundance only in diabetic kidney group, dTg/WT. Finally, 313 proteins (210 up-regulated and 103 down-regulated) were common to both groups ( Figure 3B). Worth mentioning is that, compared to controls,  Table S2 as supplementary data.
The quantitative analysis comparing dTg and sTg with WT sets indicated 56 statistically significant differentially abundant proteins (Table S3 as supplementary data) involved in nine statistically over-represented KEGG pathways (FDR P-value <2.04E-02) just in the diabetic samples (Table 1). This analysis evidenced ten enzymes related to the sulphur metabolism. Six of them were found to be involved in the first KEGG pathway (mmu00920-highest FDR corrected p-value) and are implicated in sulphate activation and degradation and in H 2 S clearance and four of them were responsible for the H 2 S production. Figure 4A  In contrast, 3(2)5 bisphosphate nucleotidase 1 (Bpnt1), cystathionine β-synthase (Cbs), 3-mercaptopyruvate sulphur transferase F I G U R E 2 Representative images of Masson's trichrome staining of kidney tissue sections (A-C). Note that the diabetic kidney tissue exhibits marked interstitial fibrosis (blue stained) surrounding the uriniferous tubules (C, arrows). Morphometric analysis (D) shows that the mean area of glomeruli was significantly increased in diabetic kidney tissue compared to non-diabetic controls. Data are expressed as mean ± SD; ***P < .001 (Mpst) and D-amino acid oxidase (Dao) were found to be significantly down-regulated in diabetic kidney ( Figure 4A). These enzymes were previously shown to have specific metabolic functions and to be responsible for the bioavailability of sulphate and the H 2 S homeostasis in different experimental models and biological mechanisms (see

| D ISCUSS I ON
This study was designed to uncover the debated role of H 2 S in kidney dysfunction that is characteristic for T1D. To this purpose,  (Table S1 as supplementary data). The latter are two important protein markers of early kidney injury, 17 indicating an uncomplicated diabetes without major damage of renal tissue.
The global gene ontology characterization of differentially abundant proteins revealed their association with mitochondrion and oxidoreductase activity (see Table S2 as supplementary data). These data extend the observations made in diabetic animal models and humans, which state that mitochondrial dysfunction is at the centre of renal disease development and progression. 21 Using a T1D mouse model and relative quantitative proteomic analysis, we unambiguously revealed the deregulated tissue protein levels of Papss2, Bpnt1, Sqrdl, Tst, Suox and Ethe1 enzymes. These were associated with minor structural changes in the kidney tissue There are reports indicating a tight correlation between the protein expression of Cth and Cbs with renal tissue H 2 S production induced by diabetes. 3 Indeed, in our experiments the two enzymes, as well as Mpst and Dao, were found to be significantly down-regulated in diabetic condition ( Figure 4A). Under physiological conditions, the level of renal Cth protein was reported to be ~20-fold higher than Cbs, suggesting that the former may be the main H 2 S-forming enzyme in the kidney. 25 From our data, the differential abundance of selected enzymes (Cth, Cbs, Mpst and Dao), associated with kidney H 2 S production, correlates well with the increased H 2 S level detected in plasma and kidney homogenate (Figures 4 and 1C,D). These data are in good agreement with previous findings stating that H 2 S production is sustained by overexpression of tissue Cth in response to diabetes. 26 There are reports recognizing that in streptozotocin-induced diabetic rats, renal H 2 S-producing enzymes Cbs and Cth are down-regulated. 10 17,40 One can envisage the need for therapeutic interventions for the early stages of diabetic nephropathy to delay or halt its progression.
In summary, these experiments showed that the mechanism underlying the elevated renal and plasma H 2 S content in early T1D nephropathy relates to the deregulated-differential expression of specific renal enzymes involved in sulphate activation and degradation, H 2 S production and clearance mechanisms.
Based on these novel findings, we can safely assume that the dysregulation of the enzymes involved in H 2 S metabolism plays a significant part in the aetiology of nephropathy in diabetes and may recommend them as new targets for the therapy of T1D.

| FAIR DATA S HARING
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE 41-43 partner repository with the data set identifier PXD018053.
F I G U R E 5 mRNA gene expression level for: (A) cystathionine β-synthase (Cbs) and (B) sulphite oxidase, mitochondrial (Suox) in kidney homogenate from wild-type (WT), single transgenic mice (sTg) and double transgenic mice (dTg) normalized to b-actin gene. Note that RT-PCR analysis revealed the down-regulation of Cbs gene expression and the up-regulation of Suox molecule in diabetic renal tissue. The results support the mass spectrometry quantification and immunoassay data. Statistical analysis was performed using Student's t test, and the results show means ± SD; *P < .05; **P < .01; n = 5

ACK N OWLED G EM ENTS
The present work was supported by the Romanian Academy and par-

DATA AVA I L A B I L I T Y S TAT E M E N T
Data openly available in a public repository that issues datasets with DOIs.