Hydrogen sulfide deficiency and diabetic renal remodeling: role of matrix metalloproteinase-9

Abstract

Matrix metalloproteinase-9 (MMP-9) causes adverse remodeling, whereas hydrogen sulfide (H₂S) rescues organs in vascular diseases. The involvement of MMP-9 and H₂S in diabetic renovascular remodeling is, however, not well characterized. We determined whether MMP-9 regulates H₂S generation and whether H₂S modulates connexin through N-methyl-d-aspartate receptor (NMDA-R)-mediated pathway in the diabetic kidney. Wild-type (WT, C57BL/6J), diabetic (Akita, C57BL/6J-Ins2^Akita), MMP-9^−/− (M9KO), double knockout (DKO) of Akita/MMP-9^−/− mice and in vitro cell culture were used in our study. Hyperglycemic Akita mice exhibited increased level of MMP-9 and decreased production of H₂S. H₂S-synthesizing enzymes cystathionine-β-synthase and cystathionine-γ-lyase were also diminished. In addition, increased expressions of NMDA-R1 and connexin-40 and -43 were observed in diabetic kidney. As expected, MMP-9 mRNA was not detected in M9KO kidneys. However, very thin protein expression and activity were detected. No other changes were noticed in M9KO kidney. In DKO mice, all the above molecules showed a trend toward baseline despite hyperglycemia. In vitro, glomerular endothelial cells treated with high glucose showed induction of MMP-9, attenuated H₂S production, NMDA-R1 induction, and dysregulated conexin-40 and -43 expressions. Silencing MMP-9 by siRNA or inhibition of NMDA-R1 by MK801 or H₂S treatment preserved connexin-40 and -43. We conclude that in diabetic renovascular remodeling MMP-9 plays a major role and that H₂S has therapeutic potential to prevent adverse diabetic renal remodeling.

diabetic renal remodeling is a progressive pathological condition ultimately leading to end-stage renal disease. Kidney dysfunction results from a combination of lesions in the glomerular, tubulointerstitial, and vascular space and is characterized by extensive extracellular matrix (ECM) remodeling (12, 36). Increased matrix metalloproteinase (MMP) activity, particularly MMP-9, plays a pivotal role in ECM alteration in diabetic kidney (28).

MMP-9 is a zinc-dependent endopeptidase, which is activated by reactive oxygen species (ROS) (30). Antioxidants as well as endogenous antioxidant enzymes are reported to protect cellular integrity in several disease states by modulating MMPs either by scavenging oxidative radicals or by diminishing its generation (38, 40). Emerging evidence suggests that hydrogen sulfide (H₂S) is an antioxidant that protects the tissue against pathological matrix turnover, in part by scavenging ROS (39, 45). However, the ability of MMP-9 to inhibit enzymes that generate H₂S and thus potentially aggravate oxidative stress in diabetic kidney is underestimated.

Physiologically, H₂S is produced by three enzymes, namely cystathionine-β-synthase (CBS), cystathionine-γ-lyase (CSE), and 3-mercaptopyruvate sulfurtransferase (3MST) (18). With few exceptions, the enzymatic pathways for H₂S production are tissue specific. For example, CBS is the main H₂S-generating enzyme in the nervous system (16), CSE in the vascular system (43). On the other hand, 3MST is primarily located in the mitochondria (18). Interestingly, all the three enzymes are present in the kidney (17). However, H₂S production and its function in diabetic kidney are yet to be defined completely.

In 1996, Abe and Kimura (1) demonstrated that H₂S acts as a neuromodulator mediating response via N-methyl-d-aspartate receptor (NMDA-R). NMDA-R is a heterotetrameric glutamate receptor(s) first reported in the brain and is known to control synaptic plasticity and memory function (24). Subsequently, it has also been reported in other mammalian tissues, including heart and kidney (23); however, very little is known about its function. The presence of NMDA-R1 subunit has been reported in the medulla and cortex of rodent kidney (23). In a recent report by Deng and Thomson (9), NMDA-R was found to regulate renal absorption and glomerular filtration rate in male Wister rats. Abundant NMDA-R1 has also been reported in the renal pelvis, where it contributes to regulate body fluids involving renorenal reflex mechanism (25). Despite these early studies, the role of H₂S in NMDA-R-mediated diabetic renal remodeling is largely unknown and is a subject of recent interest due to its therapeutic potential.

Although the relationship between H₂S and NMDA-R in diabetic kidney is incomplete, NMDA-R has been shown to regulate connexins (Cxs) (6, 46). Cxs are integral transmembrane proteins that form gap junctions and are critical for cell-cell communication including signal transduction and vascular function (7, 8). The distribution and functions of Cxs are heterogeneous and vary from cell to cell, even within similar cell types. Renal Cxs in general are thought to regulate blood pressure and may be involved in diabetic renal remodeling (11); however, their exact mechanisms are still not clear. In addition, to our knowledge, no study has been done to determine whether H₂S regulates connexin.

Therefore, the goal of our study was to determine whether MMP-9 regulated H₂S generation by triggering CBS and CSE enzymes in diabetic kidney. We also tested the hypothesis whether H₂S-mediated Cxs remodeling in diabetic kidney was NMDA-R1 dependent.

MATERIALS AND METHODS

Animal models and treatments.

C57BL/6J wild-type (WT) mice, diabetic mice of WT background (Akita, Ins2^Akita) and MMP-9 knockout (M9KO) mice were procured from Jackson Laboratory (Bar Harbor, ME). Animals were maintained on a 12:12–h light-dark cycle with free access to rodent chow and water ad libitum in the animal care facility of the University of Louisville. Akita and M9KO mice were crossbred to obtain double-knockout (DKO) mice. Mice aged 14–16 wk were used for experiments. The animal care protocol was approved by the IACUC of the University of Louisville.

We treated WT and Akita mice with N-acetyl-cysteine (NAC) at 100 mg·kg⁻¹·day⁻¹ ip for 7 days.

Cell culture and treatments.

Mouse glomerular endothelial cells (MGEC) and proximal tubular epithelial cells (MPTEC) were purchased from Cell Biologics (Chicago, IL), maintained, and passaged in laboratory conditions following supplier's instructions. Wherever used, appropriate cells were pretreated with NAC (10 mM), MK-801 (50 μM), transfected with siRNAs (80–100 p mol), and/or supplemented with NaHS (30 μM as a source of H₂S) prior to incubation under normoglycemic conditions (NG, 5 mM glucose) or hyperglycemic conditions (HG, 25 mM glucose) for 24 h.

Antibodies and reagents.

Rabbit polyclonal antibodies to CBS, CSE, NMDA-R1, Cxs-40 and -43, and MMP-9 were purchased from Abcam (Cambridge, MA). Anti-GAPDH antibody was purchased from Cell Signaling Technology (Danvers, MA). Anti-nitrotyrosine antibody was from Millipore (Temecula, CA). NaHS and other analytic reagents were purchased from Sigma-Aldrich (St. Louis, MO). Horseradish peroxidase-linked anti-rabbit IgG antibody was from Santa Cruz Biotechnology (Santa Cruz, CA). PVDF membrane was from Bio-Rad (Hercules, CA).

H₂S measurement.

To measure H₂S concentration in the tissue or cell extracts, 100-μl aliquots were mixed with potassium phosphate buffer (pH 7.4, 350 μl) and zinc acetate (1% wt/vol, 250 μl) in a microcentrifuge tube and sealed. This was followed by adding N,N-dimethyl-p-phenylenediamine sulfate (20 mmol/l, 133 μl) in 7.2 mol/l HCl and FeCl₃ (30 mmol/l, 133 μl) in 1.2 M HCl. After this, the tubes were sealed again and incubated at 37°C for 20 min for color development. To precipitate protein from the extracts, reactions were terminated by TCA (10% wt/vol, 250 μl). The mixture was then centrifuged at 10,000 g for 10 min. The resulting supernatant solutions (200 μl) were transferred to a 96-well plate, and absorbance (670 nm) was measured using a spectrophotometer. All samples were assayed in duplicate in each experiment, and H₂S was calculated against a calibration curve of known concentrations of NaHS (0.01 to 250 μmol/l).

Western immunoblotting.

Protein from kidney tissue homogenates MGEC and MPTEC were extracted using protein extraction buffer (Boston BioProducts, Ashland, MA) containing 1 mM PMSF and 1% protease inhibitor cocktail (Sigma). Samples were centrifuged at 12,000 g for 10 min at 4°C and supernatant was collected. Equal amounts of protein (50 μg) were analyzed by SDS-PAGE, electrophoretically transferred to PVDF membrane, and immunoblotted as described earlier (33).

Immunostaining.

Immunostaining was performed on 5-μm-thick frozen tissue sections according to a standard protocol (Abcam). Primary antibodies were applied overnight including NMDA-R1, Cx-40, and Cx-43 antibodies. A secondary antibody labeled with Alexa fluor 488 (Invitrogen) was applied for immunodetection of these proteins. Stained slides were analyzed for fluorescence intensity under a laser scanning confocal microscope (Olympus FluoView1000) using appropriate filter.

RNA extraction and quality assessments.

Isolated kidney samples were processed for RNA extraction using the TRIzol isolation method following the manufacturer's protocol (Invitrogen, Grand Island, NY). The quality of total RNA was determined by NanoDrop ND-1000, and only high-quality RNA (260/280-2.00 and 260/230-2.00) was used for RT-PCR.

Semiquantitative RT-PCR.

Total RNA (200–500 ng) was reverse transcribed by a two-step process using a Promega RT-PCR kit. The primer sequences are described in Table 1. Incubation of RNA with oligo(dT) at 70°C was for 6.00 min. The RT cycle was set at 25°C for 2.00 min, 42°C for 50.00 min, 75°C for 5.00 min, and 4°C until analyzed. The PCR program for amplification of cDNA was at 95°C for 10.00 min followed by (95°C for 00.30 min, 58°C for 1.00 min, 72°C for 00.30 min) x 40 cycles, 95°C for 1.00 min, 55°C for 00.30 min, and 95°C for 00.30 min.

Table 1.

Primer sequences of mRNAs

mRNA	Orientation	Primers (5′-3′)
MMP-9	Forward	TGTTTTTGATGCTATTGCTG
	Reverse	AGAAGAAAACCCTCTTGGTC
CBS	Forward	TGCGGAACTACATGTCCAAG
	Reverse	TTGCAGACTTCGTCTGATGG
CSE	Forward	GACCTCAATAGTCGGCTTCGTT
	Reverse	CAGTTCTGCGTATGCTCCGTAA
NMDA-R1	Forward	ACACAGGAGCGGGTAAACAAC
	Reverse	CTTGTCTGAGGGGTTTCTGAG
Cx-40	Forward	ATTCTGATCCGCACCACCAT
	Reverse	CATGCAGGGTATCCAGGAAGA
Cx-43	Forward	AAAATCGAATGGGGCAGGC
	Reverse	TGCTTGTTGTAATTGCGGCA
GAPDH	Forward	TAAATTTAGCCCGTGTGACCT
	Reverse	AGGGGAAAGACTGAGAAAAC

MMP-9, matrix metalloproteinase-9; CBS, cystathionine-β-synthase; CSE, cystathionine-γ-lyase; NMDA-R, N-methyl-D-aspartate receptor; Cx, connexin; GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

Gene silencing by small interfering RNA transfection.

Predesigned small interfering RNA (siRNA) targeting MMP-9 was purchased from Invitrogen, and CBS and CSE were from Santa Cruz Biotechnology. The transfection procedure was performed using DNA transfection reagent (jetPRIME, France) according to the manufacturer's instructions. Briefly, MGEC were seeded in six-well culture plates and allowed to grow to 50% confluence. Transfection reagents were prepared by adding 200 μl of jetPRIME buffer, 100 pmol of siRNA, and 4 μl of jetPRIME reagent. After mixing, the final solution containing 10 nM siRNA was added to the wells and allowed to grow for 24 h before the experiment was started.

Gelatin zymography.

Gelatin zymography was performed using 0.1% gelatin in gel. The detailed protocol was published earlier (30).

Statistical analysis.

Data are expressed as means ± SE of “n” number of animal samples in each group or experiments, as stated in the figure legends. Following tests for normality and homogeneity, significance was determined by one-way ANOVA within and across different effectors or inhibitors. Individual comparisons between treatments were made using Bonferroni's multiple comparison tests with SPSS (Chicago, IL), and a P value of ≤0.05 was considered statistically significant.

RESULTS

Expression and activity of MMP-9 in kidney tissue.

Blood glucose was measured in all animals and results are shown in Table 2. WT and M9KO mice were normoglycemic (NG), and Akita and DKO mice were hyperglycemic (HG). Immunoblotting assay was then performed to measure protein expression of MMP-9 in the kidney samples. A strong band was detected for MMP-9 in Akita mice (Fig. 1A), which was significantly increased compared with WT (Fig. 1B). The expression of MMP-9 in M9KO mice appeared as a very thin band and did not differ in DKO mice (Fig. 1, A and B).

Table 2.

Gravimetric data and glucose level

	WT (n =6)	Akita (n =5)	M9KO (n =6)	DKO (n =5)
Age, days	98	109	102	106
Body weight, g	27.34 ± 0.37	25.08 ± 0.46	28.06 ± 0.06	25.28 ± 0.21
Glucose level, mg/dl	124.8 ± 2.63	554.2 ± 11.91*	143.24 ± 3.66	550.4 ± 12.74*

Values are means ± SE. Male mice were used,

^*P < 0.01 vs. WT. WT, wild type C57BL/6J; Akita, Ins2^Akita; M9KO, MMP-9 knockout; DKO, Akita and MMP-9 double knockout.

Fig. 1.

Hyperglycemia induces expression and activity of matrix metalloproteinase-9 (MMP-9) through oxidative stress. A: 50 μg of protein was analyzed by Western blot using anti-MMP-9 antibody; anti-GAPDH antibody was used as loading control (top). RT-PCR was performed using 1 μg of mRNA extracted from kidney. GAPDH mRNA was used as loading control to normalize the abundance of MMP-9 mRNA (bottom). B: band intensity was quantified using Image J software (NIH). Values are presented as means ± SE; n = 5–6 animals per group. P values are with their respective protein and mRNA expression. C: in-gel gelatin zymography showing pro- and active forms of MMP-9 activity in kidney protein extracts. Zymography was performed as described in materials and methods. N-acetyl-cysteine (NAC) was given by ip injection with daily dose of 100 mg kg body wt in saline for 7 days. D: bar diagram represents relative activity of active form of MMP-9. Values are means ± SE; n = 5–6 animals per group. E: immunostaining of kidney sections against N-tyrosine antibody (green fluorescence). Nucleus was stained with DAPI (blue).

To determine whether protein expression of MMP-9 also correlated with mRNA level, we performed RT-PCR analysis and data are shown in Fig. 1A (bottom). The mRNA results corresponded with protein levels in all groups (Fig. 1, A and B).

Since the activity of MMP is crucial for remodeling process, we measured enzymatic activity of MMP-9 using in-gel gelatin zymography in the protein extracted from the kidneys. MMP-9 activity in the WT animals was taken as baseline (Fig. 1C). Compared with baseline values and corroborating protein expression (Fig. 1A), the activity of both pro- and active MMP-9 was significantly increased in Akita mice (Fig. 1, C and D). As expected, the activity of MMP-9 was very low in M9KO and DKO mice (Fig. 1, C and D).

MMP-9 is considered to be upregulated by cellular oxidative stress (30). Therefore, we determined whether oxidative blocker N-acetylcysteine (NAC) mitigates MMP-9 induction. Results indicated that NAC significantly inhibited MMP-9 activity both in WT and in Akita mice (Fig. 1, C and D).

Furthermore, we also examined the level of renal glomerular and tubular oxidative stress, using the cellular nitrosative stress marker nitrosylated tyrosine (N-tyrosine). Results indicated that WT mice had basal level of N-tyrosine in the kidney (Fig. 1E). In Akita mice, immunolabeling indicated a robust increase of N-tyrosine in the glomerular as well as in the tubular areas (Fig. 1E). The intensity of N-tyrosine in M9KO mice was similar to that in WT; however, in DKO kidney the N-tyrosine was higher, which was largely comparable to the Akita group (Fig. 1E).

Urinary protein excretion remains unchanged, but H₂S production is decreased in diabetic kidney.

We measured urinary protein excretion, as a marker of renal damage, to determine whether oxidative stress and MMP-9 led to renal pathophysiology in diabetes. Although there was a slight increase of proteinuria in Akita mice, the change was nonsignificant compared with baseline levels in WT (Fig. 2A). Similarly, urinary protein levels were not significantly changed in M9KO or DKO vs. WT. NAC also did not alter levels of urinary protein in WT or in Akita mice (Fig. 2A).

Fig. 2.

Protein excretion remains unaltered in diabetes, but H₂S production reduces due to decreased renal expression of cystathionine-β-synthase (CBS) and cystathionine-γ-lyase (CSE). A: urine collected for 24 h and total urinary protein were measured by Bradford method. B: kidney samples were homogenized in 1× PBS (pH 7.2), and H₂S production was measured. Values represent means ± SE; n = 5–6 animals per group. C: 50 μg of protein from kidney was separated by SDS-PAGE, transferred to PVDF membrane, and immunoblotted using anti-CBS and anti-CSE antibodies as shown. GAPDH was used as a loading control. D: pixel densities of bands were quantified using Image J software (NIH) and are presented in the bar diagram as fold change. Data represent means ± SE; n = 5–6 animals per group. P values are with their respective CBS and CSE protein expression. E: RT-PCR amplifications of CBS and CSE genes were performed using 1 μg of mRNA extracted from kidney. GAPDH mRNA was used as loading control to normalize the abundance of CBS and CSE mRNA. F: relative expressions of CBS and CSE gene are shown from densitometric analyses. Each bar represents the normalized mean expression ± SE; n = 5–6 animals per group. P values are with their respective CBS and CSE mRNA expression.

Next, we determined whether H₂S production was decreased in diabetes and whether MMP-9 and oxidative stress have any role on its production. Results summarized in Fig. 2B show that, in Akita mice, H₂S production was significantly reduced compared with WT mice. Although the change of H₂S production in M9KO was negligible compared with WT, the level of H₂S production was significantly improved in DKO mice vs. Akita (Fig. 2B). Interestingly, NAC also significantly improved H₂S production in Akita mice (Fig. 2B). The production of H₂S in NAC-treated WT mice remained unchanged, which was comparable to WT baseline level.

Renal expression of CBS and CSE.

Since H₂S level was low in the diabetic kidney (Fig. 2B) and its production is dependent on CBS and CSE enzymes, we measured the expression of these two enzymes in kidney samples. Figure 2, C and E, represents Western blot and RT-PCR analyses, respectively. Western blot analysis recognized a band at 63 kDa and another at 45 kDa corresponding to CBS and CSE, respectively (Fig. 2C). CBS and CSE protein expressions were significantly decreased in the Akita group compared with WT (Fig. 2D), but there was no significant difference between WT and M9KO. DKO mice showed significantly increased protein expression compared with Akita mice but was slightly attenuated compared with WT (Fig. 2, C and D). There was no difference in their expressions between M9KO and DKO groups (Fig. 2D). RT-PCR analyses revealed decreased mRNA expressions for both CBS and CSE in Akita kidney compared with WT (Fig. 2E). Data normalization with GAPDH mRNA indicated that CBS and CSE levels were marginally attenuated in M9KO compared with WT. Interestingly, in DKO mice both mRNAs were significantly increased compared with Akita group, which was almost to WT levels (Fig. 2, E and F).

Oxidative blocker NAC normalizes MMP-9, CBS, and CSE expression as well as H₂S production in HG condition.

To determine whether oxidative stress activates MMP-9, which in turn downregulates CBS and CSE and thus H₂S production, we used NAC, an antioxidant, to inhibit MMP-9 in HG. Western blot analysis of kidney samples showed that upregulated MMP-9 in Akita mice was mitigated by NAC (Fig. 3, A and B). In addition, the expression of CBS and CSE in Akita samples also showed improvement with NAC treatment (Fig. 3, A and B). There was no change in the expressions of MMP-9, CBS, and CSE in WT mice without or with NAC (Fig. 3, A and B). Similar results were obtained in in vitro NG or HG conditions in MGEC without or with NAC (Fig. 3, C and D). Corroborating with protein expressions, mRNA expressions of MMP-9, CBS and CSE revealed a similar pattern in WT and Akita kidneys as well as in in vitro MGEC in NG and HG conditions without or with NAC (Fig. 3, E–H).

Fig. 3.

NAC normalizes MMP-9, CBS, and CSE in hyperglycemia. Protein from kidney (A) and 24-h in vitro culture of mouse glomerular endothelial cells (MGEC) (C) was separated by SDS-PAGE and immunoblotted using anti-mouse MMP-9, CBS, and CSE antibodies. GAPDH was used as a loading control. Pixel densities of bands from kidney samples and MGEC were quantified and plotted in bar diagram (B and D, respectively) as relative intensity. Data represent normalized mean expression ± SE; n = 5–6 animals per group (B), and n = 4 independent experiments (D). P values are with their respective MMP-9, CBS, and CSE protein expression. RT-PCR was performed using 1 μg of mRNA extracted from kidney (E) and from in vitro MGEC culture (G). GAPDH mRNA was used as loading control to normalize abundance of MMP-9, CBS, and CSE mRNAs. Semiquantitative pixel densities of bands from kidney and MGEC samples were plotted in bar diagram (F and H, respectively) as relative intensity. Data represent normalized mean expression ± SE; n = 5–6 animals per group (F), and n = 4 independent experiments (H). P values are with their respective MMP-9, CBS, and CSE mRNA expression. MMP-9 activity was measured by gelatin zymography from in vitro MGEC cultured lysates (I); band intensities were quantified and plotted in bar diagrams (J). K: cells were homogenized in PBS, and H₂S production was measured. Data represent normalized mean expression ± SE; n = 4 independent experiments (J and K).

The activity of MMP-9 in in vitro MGEC was increased almost threefold in HG (Fig. 3, I and J). This increase was attenuated and normalized with NAC treatment. Interestingly, H₂S production in MGEC was also diminished in HG, which was normalized by NAC treatment (Fig. 3K).

Expression and localization of NMDA-R1 in mouse kidney.

Next, we sought to determine the expression levels of NMDA-R1 among study groups, and results are shown in Fig. 4. Western blot analysis of kidney tissue samples indicated induction of NMDA-R1 in the Akita group compared with WT, which was significantly decreased in DKO (Fig. 4, A and B). The expression level of NMDA-R1 in M9KO mice, however, did not change significantly vs. WT. Figure 4, A and B, also shows the RT-PCR result of NMDA-R1 gene expression. Corroborating with the protein expression, NMDA-R1 gene expression in Akita kidney was significantly higher than in WT (Fig. 4, A and B). In DKO mice, a lower level of NMDA-R1 was measured, and the difference was significant compared with Akita (Fig. 4, A and B), whereas in M9KO mRNA expression remained similar to WT.

Fig. 4.

N-methyl-d-aspartate receptor (NMDA-R1) expression is upregulated in diabetes. A: protein sample (top) from each group was separated by SDS-PAGE and immunoblotted using mouse anti-NMDA-R1 antibody. GAPDH was used as a loading control. RT-PCR (bottom) amplification and transcript level of NMDA-R1 were performed using 1 μg of mRNA extracted from kidney. PCR amplification was performed as described in materials and methods. GAPDH mRNA was used as loading control to normalize abundance of NMDA-R1 mRNA. B: pixel densities of bands were quantified and plotted in bar diagram as fold change. Data represent normalized mean expression ± SE; n = 5–6 animals per group. P values are with their respective protein and mRNA expression. C: representative immunostained kidney sections (5 μm) showing expression of NMDA-R1 in tubules (yellow arrow) and glomeruli (red arrow). (×60 magnification).

We also determined the localization of NMDA-R1 within the kidneys, and the results are shown in Fig. 4C. The baseline expression of NMDA-R1 was localized both in tubules and in glomeruli in WT mice (Fig. 4C). Interestingly, in Akita mice, NMDA-R1 was predominantly expressed in tubules (Fig. 4C), which was more intense than in WT. Although NMDA-R1 expression in M9KO mice did not vary from baseline and was comparable to WT, in DKO mice its expression was diminished to WT level (Fig. 4C).

Expression of Cxs-40 and -43 in mouse kidney.

To determine whether gap junction proteins, in particular Cxs-40 and -43 are involved in diabetic renal remodeling, we performed immunoblotting, RT-PCR, and immunostaining assays. The results indicated that protein (Fig. 5, A and B) and mRNA (Fig. 5, C and D) levels of Cxs-40 and -43 in diabetic kidney were significantly higher compared with WT. There was a nonsignificant increase of Cxs-40 and -43 protein expressions in M9KO vs. WT mice (Fig. 5, A and B). Unlike protein, differential mRNA expression of Cxs-40 and -43 was detected in M9KO mice, where Cx-40 mRNA was higher and Cx-43 mRNA was lower in M9KO mice vs. WT (Fig. 5, C and D). Interestingly, both protein and mRNA levels were diminished in DKO mice compared with Akita (Fig. 5, A–D).

Fig. 5.

Expressions of connexins (Cxs)-40 and -43 are upregulated in diabetes. A: kidney protein sample was separated by SDS-PAGE and immunoblotted using rabbit anti-Cxs-40 and -43 antibodies. GAPDH was used as a loading control. B: pixel densities of bands were quantified and presented in bar diagram as fold changes. Each bar represents normalized mean expression ± SE; n = 5–6 animals per group. P values are with their respective protein expression. C: RT-PCR amplification of Cxs-40 and -43 genes from kidney mRNA were performed. GAPDH mRNA was used as loading control to normalize abundance of Cxs-40 and Cx-43 mRNA. D: bar diagram showing relative pixel densities of mRNA. Data represent normalized mean mRNA expression ± SE; n = 5–6 animals per group. †P < 0.05 vs. WT and P values on bars are with their respective mRNA expression. E: representative immunostained image showing expression of Cx-40 in glomeruli (red arrows) as well as in the tubules (yellow arrows) in all groups studied (original magnification ×60; n = 5–6 animals per group). F: glomerular (red arrows) and tubular (yellow arrows) expressions of Cx-43 in diabetic kidney. Image: original magnification ×60; representative image of n = 5–6 animals per group.

Immunofluorescence microscopy images revealed Cx-40 expression mainly in the glomeruli and periglomerular tubules in the WT kidney (Fig. 5E). Its expression was robust in Akita kidney (Fig. 5E). Corroborating with protein expression, Cx-40 immunostaining was similar in M9KO mice vs. WT (Fig. 5E). Interestingly, in DKO mice Cx-40 was detected only in the periglomerular area, which was relatively lower than in the Akita group (Fig. 5E). A similar trend for Cx-43 expression was observed in WT, Akita, and M9KO mice, where the highest expression was noticed in Akita mice compared with WT (Fig. 5F). In M9KO mice, immunodetection of Cx-43 was similar to that in WT, whereas in DKO mice the expression of Cx-43 was relatively lower compared with Akita mice (Fig. 5F). The most interesting finding was that, unlike Cx-40, Cx-43 was mostly localized in the glomerulus of DKO mice.

Expression of Cxs-40 and -43 in MGEC vs. MPTEC cells.

The level of NMDA-R1 was more strongly increased in the renal tubules than in the glomerulus (Fig. 4C), whereas Cxs-40 and -43 were upregulated mostly in the glomerulus (Fig. 5, E and F). Therefore, to determine which cell type, glomerular vs. tubular, is the main target of NMDA-R1 on Cxs-40 and -43 expression in the kidney, we used MGEC and MPTEC. Results indicated that both cell types expressed Cxs-40 and -43 at the basal level in NG (Fig. 6). The expressions of Cxs-40 and -43 were increased in HG; however, the magnitudes of their expressions were higher in MPTEC compared with MGEC (Fig. 6). NMDA-R inhibitor MK-801 inhibited and normalized Cxs-40 and -43 expressions in both cell types in HG. MK-801 did not have differential effects on Cxs-40 and-43 expressions in NG (Fig. 6).

Fig. 6.

Proximal tubular epithelial cells are the main target of NMDA-R1 on Cxs-40 and -43 expressions. Mouse glomerular endothelial cells (MGEC) and mouse proximal tubular epithelial cells (MPTEC) were incubated without or with MK-801 in normoglycemic (NG) and hyperglycemic conditions (HG) as described in materials and methods. At the end of incubation, cells were immunostained with Cx-40 (green fluorescence) and -43 (red fluorescence); nucleus was stained with DAPI. Fluorescence images were taken under confocal microscope.

Expressions of CBS and CSE in MGEC vs. MPTEC.

We also determined by Western blot whether CBS and CSE are expressed differentially in MGEC vs. MPTEC. Representative immunoblotting shows that MGEC expressed both enzymes (Fig. 7A), whereas these enzymes were nondetectable in MPTEC (Fig. 7B). Under HG, the expressions of both CBS and CSE were attenuated in MGEC (Fig. 7A), and interestingly, MK-801 had no effect on their expression (Fig. 7A).

Fig. 7.

Expressions of CBS, CSE, and H₂S production are MGEC specific. MGEC and MPTEC were incubated without or with MK-801 in NG and HG. At the end of incubation, cells were lysed and immunoblotted with CBS and CSE antibodies. One representative blot from each cell type of 4 independent experiments is shown here (A and B). In separate experiments, cells were homogenized in PBS, and production of H₂S was measured (C and D). Data represent means ± SE; n = 5 independent experiments.

NMDA-R1 regulates H₂S production in HG condition.

We further determined H₂S generation in both cell types under similar experimental conditions. Corroborating with CBS and CSE expression, H₂S production was decreased in HG condition compared with NG condition in MGEC (Fig. 7C). MK-801 did not alter H₂S production neither in NG nor HG conditions. The production of H₂S in MPTEC was extremely low in NG and was nondetectable in HG (Fig. 7D). As expected, MK-801 did not change H₂S production either in NG or HG condition compared with their respective controls (Fig. 7D).

In vitro expression of MMP-9 and NMDA-R1 and role of H₂S in HG condition.

This experiment was performed to determine whether H₂S regulated MMP-9 and NMDA-R1 in HG. The results shown in Fig. 8A are in accord with our earlier experiments presented in this study (Fig. 3, C, G, and I) and indicated that MMP-9 expression was ∼2.4-fold higher in HG compared with the NG. Although MK-801 marginally increased MMP-9 in NG compared with control, MMP-9 was surprisingly significantly decreased in HG with MK-801 than without compared with the HG alone (Fig. 8A). MMP-9 knockdown by siRNA decreased its expression by 45% in NG compared with control, which further diminished in HG (Fig. 8A). H₂S treatment also attenuated increased MMP-9 expression in HG. Similar to in vivo expression (Fig. 4C), NMDA-R1 expression was significantly higher in the in vitro HG condition, which was attenuated either by MK-801 or by MMP-9 siRNA (Fig. 8A). Interestingly, treatment with H₂S normalized NMDA-R1 expression in the in vitro HG condition (Fig. 8A).

Fig. 8.

H₂S normalizes NMDA-R1 and Cxs-40 and -43 but not MMP-9 expression in HG. Equal amounts of cell lysate protein were immunoblotted with antibody against NMDA-R1 and MMP-9 (A), and Cxs-40 and -43 (B). The same GAPDH immunoblot was used as loading control. Representative blots and composite bar diagrams are shown. Values represent means ± SE; n = 5 independent experiments.

Cxs in HG condition and their regulation by MMP-9, H₂S, and NMDA-R1.

To determine whether MMP-9, H₂S, and NMDA-R1 have roles in Cxs-40 and -43 regulation during HG condition, we performed experiments as shown in Fig. 8B. The results indicated that Cxs-40 and -43 were upregulated in HG. Preincubation of MGEC with the NMDA-R blocker MK-801 significantly attenuated their induction in HG (Fig. 8B). Most interestingly, silencing MMP-9 by siRNA completely normalized Cxs-40 and -43 expressions in HG. Although similar attenuation of Cxs' expressions were seen in the H₂S-pretreated experimental HG condition, the attenuation effect was lesser compared with MMP-9 siRNA (Fig. 8B).

H₂S is critical for NMDA-R1 and cxs-40 and -43 expressions.

To ascertain whether H₂S is critical for NMDA-R1 regulation as well as Cxs-40 and -43 expressions, CBS and CSE were directly inhibited by siRNAs in HG followed by downstream gene expression. Results depicted in Fig. 9 indicated that silencing CBS and CSE with respective siRNAs did not significantly mitigate NMDA-R1 mRNA expression in HG (Fig. 9A). In fact, its expression remained similar to that of HG without siRNA transfection. The most striking result was increased mRNA expression of NMDA-R1 in NG transfected with CBS or CSE siRNAs (Fig. 9, A and B). Similar to NMDA-R1, gene expressions of Cxs-40 and -43 remained upregulated without further change in HG even though CBS or CSE was silenced by siRNAs. Moreover, in NG CBS and CSE siRNAs upregulated Cxs-40 and -43 gene expressions to a level similar to that in HG (Fig. 9, A and B).

Fig. 9.

CBS or CSE gene silencing upregulates NMDA-R1 and Cxs-40 and -43 gene expression. Gene silencing experiments and RT-PCR amplification of NMDA-R1 and Cxs-40 and -43 genes were performed as described in materials and methods. A: representative DNA gels are shown. GAPDH mRNA was used as loading control to normalize abundance of targeted mRNAs. B: bar diagram represents fold changes of mRNAs expression with their respective NG control. Values represent means ± SE; n = 4 independent experiments.

DISCUSSION

In the present study, we investigated whether H₂S was involved in diabetic renal remodeling and the role of MMP-9 in modulating the H₂S pathway. We also investigated whether altered NMDA-R1 expression had a causative role on Cxs remodeling in HG. Our results indicated that diabetic kidneys express high levels of MMP-9 with increased activity, resulting in downregulation of H₂S production by diminishing CBS and CSE enzymes. Low levels of H₂S in turn induced NMDA-R1 and upregulated Cxs-40 and -43 in the downstream pathway. Taken together, these results suggest that H₂S is a key mediator in diabetic renal remodeling (Fig. 10).

Fig. 10.

Schema of overall hypothesis. In hyperglycemia, ROS induces MMP-9, causing downregulation of H₂S by suppressing CBS and CSE. Low levels of H₂S in turn induce NMDA-R1 expression, probably as a compensatory mechanism where low H₂S balances its signaling cascade with an increase in NMDA-R1. This mechanism is perhaps different from the neuronal mechanism of NMDA-R1 induction by H₂S (1), which leads to connexin remodeling in renal tissue. H₂S mitigates NMDA-R1 expression and ameliorates diabetic Cxs-40 and -43 remodeling.

In our study, we used two genetic mice models, diabetic Akita and homozygous M9KO mice. The heterozygous spontaneous mutation of Akita (Ins2^Akita) has many similarities with diabetogen-induced, such as the alloxan- or streptozotocin-induced diabetic model in terms of depleted β-cells, hypoinsulinemia, and hyperglycemia (44), but no toxicity. Hence, it is a superior model over the diabetogen-induced diabetic model. The M9KO mice were used to demonstrate whether MMP-9 plays a significant role in diabetic renovascular remodeling. We and others have previously reported that MMP-9 plays a critical role in diabetic renal remodeling of diabetogen-induced kidney (19, 32), where MMP-9 induction was especially associated with glomerular-tubulointerstitial fibrosis. In the present study, we cross-bred Akita and M9KO to obtain Akita and MMP-9 double knockouts (herein reported as DKO) to further understand the mechanisms of the remodeling process in diabetic kidney. Also, since MMP-9 is involved in ECM remodeling (22), and remodeling is directly linked to H₂S generation (47), we further explored the importance of H₂S in diabetic kidney.

Remodeling by its very nature is a dynamic process that could be deleterious in disease states and may end in organ malfunction. In diabetic kidney, proteinuria occurs secondarily to podocyte injury and glomerulosclerosis, but this is not universal. Variations in excretion of protein are known to occur across different strains, and it may also depend on the stage of disease, i.e., early vs. established, and method of detection (3). Our results indicated that, although there was no overt proteinuria, H₂S production was significantly decreased in Akita kidney, suggesting functional impairment at the cellular level to maintain normal physiological levels of H₂S. In addition, upregulation of MMP-9 in diabetic kidney indicated an active remodeling process due to increased oxidative radicals as suggested by increased levels of N-tyrosine, which was mitigated by NAC treatment. Moreover, in DKO and in NAC-treated Akita mice, decreased levels of H₂S were normalized, indicating that MMP-9, perhaps in concert with oxidative stress, regulated H₂S production in Akita kidney.

Progressive renal remodeling in diabetes can lead to deterioration of renal function, where excessive deposition of ECM in the kidney plays a crucial role (10, 29). In this regard, a number of factors are involved in ECM degradation and deposition, including ROS, TGF-β1, plasminogen inhibitor and plasmin, and MMPs (10, 20, 27, 34). MMP-9 is widely reported to be associated with ECM deposition and endothelial and kidney dysfunction (4, 32). In earlier studies from our laboratory, we reported MMP-9 induction in diabetogen-induced (32) as well as one kidney hypertensive mice models (30). H₂S supplementation mitigated MMP-9 induction in the latter animal model (30), but it remained unknown whether H₂S treatment mitigates MMP-9 induction similarly in diabetic condition, despite the fact that H₂S inhibits high-glucose-induced matrix accumulation (21). In the present study, we report that high glucose induces MMP-9 and decreases H₂S generating enzymes, CBS and CSE, resulting in diminished H₂S production. This result corroborates a recent in vitro study (21). Interestingly, although H₂S production did not differ significantly in M9KO mice compared with WT, its production was significantly improved in DKO mice vs. Akita, suggesting that MMP-9 plays a critical role in regulating H₂S production in the diabetic condition. In vitro studies revealed that H₂S supplementation although marginally attenuated, did not normalize MMP-9 in HG condition. Taken together, these results suggest that H₂S supplementation does not regulate MMP-9, at least, in the present experimental condition. This is in contrast to our previous in vivo findings that H₂S regulated MMP-9 in one kidney hypertension model (30, 31). We expected a similar result in the present study, but the contrast may be due to in vivo vs. in vitro and hypertension vs. hyperglycemic models and remains to be elucidated. In addition, whether CBS and CSE inhibition is due to cleavage or conformational change by active MMP-9 or altered matrix turnover, which may affect the enzyme function, would be of further interest to investigate.

The enzymatic pathways for H₂S production are tissue specific. For example, CBS is the prevalent and main enzyme in the nervous system (16), whereas CSE is the primary enzyme for H₂S production in the vascular system (43). Despite their tissue specificity, both CBS and CSE have been reported in the kidney (13). In a recent report by Yamamoto et al. (41), the authors demonstrated decreased levels of CSE but normal levels of CBS in transgenic diabetic mice. In addition, they also described the presence of CBS and CSE in the brush border and cytoplasm of proximal tubules in normal kidney. In contrast, in the present in vivo studies we found the expression of CBS and CSE markedly reduced in diabetes. In in vitro studies, although MGEC showed the presence of both enzymes, MPTEC did not. This difference in observation could be due to variation in experimental condition and animal models.

It is reported that CSE knockdown induces hypertension by downregulating H₂S production (43). Alteration in H₂S levels are known to occur in diabetes (2, 37); however, the consequences of diminished H₂S in diabetic renovascular tissue are still far from clear. In this study, we report attenuated expressions of CBS/CSE and diminished production of H₂S in Akita kidney as well as in in vitro MGEC in HG. In a recent report, Lee et al. (21) also reported that reduction of CBS and CSE enzymes manifested in clinical expression of diabetic nephropathy. Since H₂S is a functional physiological gaseous molecule that regulates a variety of vascular functions spanning from vascular tone to endothelial and smooth muscle function (5, 14, 43), it is no surprise that any malfunction or low expression of CBS/CSE enzymes in diabetes may lead to renovascular complications due to loss of H₂S-dependent functions. This study thus delineates the underlying mechanism in diabetic renal remodeling that occurs due to H₂S deficiency.

A growing body of evidence suggests that H₂S regulates a number of receptors, including NMDA-R, in a variety of physiological and pathophysiological states, such as induction and long-term potentiation of hippocampus (1, 15), ischemic brain injury (26), and nociception (35). In the kidney, NMDA-R1 is abundantly expressed in the outer medulla, cortical brush border membrane, and glomeruli, and intrarenal NMDA agonist is known to suppress glomerular and tubular function significantly (42). However, to our knowledge it was not known whether depletion of H₂S has a causative role in NMDA-R1 expression that may play a crucial role in diabetic renal remodeling. The present report suggests that, in diabetic kidney, expression of NMDA-R1 is high both in mRNA and in protein level. This result was further confirmed by in vitro studies where high glucose induced NMDA-R1 in MGEC. Additionally, we also show that although it expresses both in glomerular and tubular spaces, it is highly expressed in tubular cells in Akita kidney. The induction of NMDA-R1 was H₂S dependent and inversely related in diabetic condition. This was supported by the result that supplementation of H₂S mitigated NMDA-R1 expression in HG in vitro. NMDA-R1 induction was also mitigated by MMP inhibitor and by NMDA-R1 blocker MK-801. These results further confirmed that induction of NMDA-R1 in the diabetic condition is MMP-9 mediated and dependent on H₂S pathway. We also noticed that MK-801 inhibited upstream MMP-9 in HG. Although this does not fit with the overall mechanism of MMP-9-mediated H₂S and NMDA-R1 regulation, a possible and perhaps the best explanation is that MK-801 frees endogenous H₂S molecules to bind with NMDA-R1, and unbound H₂S scavenges oxidative radicals and thus reduces oxidative stress and MMP-9 induction in HG through a feedback mechanism.

The level of NMDA-R is more strongly increased in renal tubules than in the glomerulus, whereas Cxs-40 and -43 are enhanced in the glomerulus. Therefore, to delineate which cell type, glomerular (MGEC) vs. tubular (MPTEC), is the main target of NMDA-R on Cxs-40 and -43 expression in the kidney, we used both cell types to address whether MK-801 is or is not sensitive to H₂S production and its enzymes (CBS/CSE) are reduced. First of all, in MPTEC we could not detect CBS or CSE enzymes, and thus H₂S production was at a negligible range compared with MGEC. Although NMDA-R1 was expressed highly in tubular cells in Akita mice and may be targeted to NMDA-R1 blocker, the production of H₂S was mainly from MGEC. It was also observed that both cell types were targets of NMDA-R1 blocker; however, the blocker had a stronger effect in MPTEC than in MGEC. This was probably because MPTEC expresses more Cxs-40 and -43 than MPTEC in HG.

We also performed experiments to delineate whether NMDA-R1 regulates Cxs-40 and -43 during hyperglycemia and whether H₂S modulates these gap junction proteins that may contribute to diabetic renal remodeling. Our results indicate that NMDA-R1 induction upregulated Cxs-40 and -43 in HG. This result further supports that gap junction proteins contribute to NMDA-R function (6). Interestingly, when treated with NMDA-R1 blocker, the expressions of Cxs-40 and -43 were downregulated, suggesting a possible role for NMDA-R1 in Cxs-40 and -43 modulations. To our knowledge, this is the first report of NMDA-R1 induction, and Cxs-40 and -43 upregulation in HG. In addition, we also confirmed that H₂S is indeed critical for NMDA-R1 regulation as well as Cxs-40 and -43 expressions by siRNA modulation of CBS and CSE followed by downstream gene expression. Since CBS and CSE are the main enzymes responsible for H₂S production (17) (the third one is 3-mercaptopyruvate sulfurtransferase), silencing these enzymes with siRNA would result in decreased H₂S production. Thus, we were not surprised with the result that CBS and CSE siRNAs transfection did not mitigate NMDA-R1 mRNA induction, since H₂S is necessary for NMDA-R1 inhibition. Similarly, consistent and upregulated Cxs-40 and -43 gene expression with CBS and CSE siRNAs further strengthened our hypothesis that H₂S is critical for Cxs-40 and -43 regulation. Therefore, the involvement of H₂S and NMDA-R1 in Cxs-40 and -43 regulations in HG was postulated (Fig. 10).

Limitations.

The findings that upregulation of MMP-9 is involved in downregulation of CBS/CSE and upregulation of NMDA-R1/connexin are supported by in vivo data in DKO mice. Our in vitro study demonstrated that H₂S deficiency is the cause of NMDA-R1 and connexin upregulation. In the in vivo condition, whether their upregulation is the consequence of MMP-9 excess independently of H₂S remains to be answered. Additionally, it is possible that MMP-9 may stimulate cells through its putative receptors. The effect of exogenous MMP-9 on CBS/CSE and thus H₂S production may also help in better understanding of the underlying mechanism.

In conclusion, our study demonstrates that hyperglycemia induces MMP-9, which diminishes CBS/CSE and thus H₂S production. Diminished H₂S in turn induces NMDA-R1 and mediates Cxs-40 and -43 remodeling, suggesting that H₂S is a player in this pathway.

GRANTS

This study was supported, in part, by National Institutes of Health Grant HL-104103.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: S.K., S.B.P., and U.S. conception and design of research; S.K., S.B.P., A.T., D.C., and U.S. performed experiments; S.K. and S.B.P. analyzed data; S.K., S.B.P., and U.S. interpreted results of experiments; S.K. and S.B.P. prepared figures; S.K. and U.S. drafted manuscript; S.B.P. and U.S. edited and revised manuscript; U.S. approved final version of manuscript.