An intracellular matrix metalloproteinase-2 isoform induces tubular regulated necrosis: implications for acute kidney injury

Abstract

Acute kidney injury (AKI) causes severe morbidity, mortality, and chronic kidney disease (CKD). Mortality is particularly marked in the elderly and with preexisting CKD. Oxidative stress is a common theme in models of AKI induced by ischemia-reperfusion (I-R) injury. We recently characterized an intracellular isoform of matrix metalloproteinase-2 (MMP-2) induced by oxidative stress-mediated activation of an alternate promoter in the first intron of the MMP-2 gene. This generates an NH₂-terminal truncated MMP-2 (NTT-MMP-2) isoform that is intracellular and associated with mitochondria. The NTT-MMP-2 isoform is expressed in kidneys of 14-mo-old mice and in a mouse model of coronary atherosclerosis and heart failure with CKD. We recently determined that NTT-MMP-2 is induced in human renal transplants with delayed graft function and correlated with tubular cell necrosis. To determine mechanism(s) of action, we generated proximal tubule cell-specific NTT-MMP-2 transgenic mice. Although morphologically normal at the light microscopic level at 4 mo, ultrastructural studies revealed foci of tubular epithelial cell necrosis, the mitochondrial permeability transition, and mitophagy. To determine whether NTT-MMP-2 expression enhances sensitivity to I-R injury, we performed unilateral I-R to induce mild tubular injury in wild-type mice. In contrast, expression of the NTT-MMP-2 isoform resulted in a dramatic increase in tubular cell necrosis, inflammation, and fibrosis. NTT-MMP-2 mice had enhanced expression of innate immunity genes and release of danger-associated molecular pattern molecules. We conclude that NTT-MMP-2 “primes” the kidney to enhanced susceptibility to I-R injury via induction of mitochondrial dysfunction. NTT-MMP-2 may be a novel AKI treatment target.

acute kidney injury (AKI) continues to represent a major complication for hospitalized patients, leading to unacceptably high levels of morbidity and mortality. Furthermore, it has been demonstrated convincingly that even a single episode of AKI can lead to the subsequent development of chronic kidney disease (CKD) (12, 13, 28, 42). The severity of an individual AKI episode is greatly increased in the setting of advanced age, diabetes mellitus, preexisting CKD, and underlying cardiovascular disease (14, 16, 18, 28, 51, 56). The pathophysiology of AKI is remarkably complex and has been the focus of intense laboratory and clinical investigation for more than four decades (1, 5, 6, 10, 23, 30). Despite these efforts, the successful development of either preventative or therapeutic modalities has remained elusive.

A common feature of AKI is the development of enhanced oxidative stress, particularly in the setting of ischemia-reperfusion (I-R) injury and sepsis (1, 6, 10, 50). Multiple cellular pathways contribute to enhanced renal oxidative stress in AKI, and there has been increased experimental attention paid to the role of mitochondrial dysfunction in this process (1, 5, 6, 11, 49, 52, 53).

Specific matrix metalloproteinases, particularly matrix metalloproteinase-2 (MMP-2), have been shown to play important roles in the development of renal tubular injury in rodent models of I-R AKI (22, 35, 47). These studies have focused on the actions of the secreted, full-length isoform of MMP-2 (FL-MMP-2), with the assumption that the deleterious actions of this enzyme are directed to the tubular basement membrane and extracellular space. Recently, we characterized a second novel intracellular isoform of MMP-2 generated by oxidative stress-mediated activation of an alternate promoter located in the distal first intron of the MMP-2 gene (43). This event generates an NH₂-terminal truncated MMP-2 (NTT-MMP-2) transcript with an open reading frame starting at M⁷⁷ in the second exon. The NTT-MMP-2 isoform lacks the secretory sequence and part of the inhibitory prodomain. NTT-MMP-2 remains intracellular, is enzymatically active, and is concentrated within the mitochondrial intramembranous space (43). NTT-MMP-2 degrades inhibitory IκB present in the mitochondrial intramembranous space, with subsequent activation of mitochondrial/nuclear NF-κB and nuclear factor of activated T cell signaling cascades (43). This event results in the induction of a highly restricted innate immunity transcriptome composed almost exclusively of chemokines and cell death-associated genes (39). De novo expression of the NTT-MMP-2 isoform was detected in cardiac mitochondria from 14-mo-old mice and in a model of accelerated atherogenesis (43). Cardiac-specific expression of the NTT-MMP-2 isoform results in severe systolic failure associated with cardiomyocyte necrosis and mononuclear inflammation (44). In addition, these mice have significantly enhanced cardiomyocyte necrosis in response to I-R injury (44).

In parallel studies described herein, we determined that the NTT-MMP-2 isoform was also expressed de novo in proximal tubular epithelial cells of 14-mo-old wild-type mice as well as in the setting of a murine model of chronic heart failure and CKD. This paper describes in detail the histopathological and functional phenotype of transgenic mice expressing the NTT-MMP-2 isoform in the renal proximal tubule. We demonstrate that proximal tubular NTT-MMP-2 expression triggers an innate immune response associated with tubular epithelial cell-regulated necrosis and severe interstitial inflammation. Furthermore, we show that NTT-MMP-2 expression sensitizes the tubular epithelial cell to enhanced injury in the setting of limited I-R injury.

MATERIALS AND METHODS

This investigation was approved by the Animal Care and Use Subcommittee of the San Francisco Veterans Affairs Medical Center (Protocols 09-053-03 and 13-038) and conformed to the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH Publication 85-23, revised 1996). This institution is accredited by the American Association for the Accreditation of Laboratory Animal Care (Institutional PHS Assurance no. is A3476-01).

Construction of NH₂-terminal truncated NTT-MMP-2 cDNA and generation of transgenic mice.

Assembly of a human NTT-MMP-2/enchanced green fluorescent protein (eGFP) cDNA was performed as reported in detail (15). The expression cassette was placed under the control of the rat type I γ-GT promoter in tandem with a human growth hormone intron and polyadenylation tail, as reported in detail (15).

The expression cassette was excised, purified, and microinjected into FVB/N strain fertilized pronuclei (Xenogen Biosciences, Cranbury, NJ). Transgenic mice were identified by PCR of tail genomic DNA using a 5′ primer for NTT-MMP-2 (5′-TGGATCCTGGCTTCCCC-AAGCTCATCG-3′) and a 3′ primer for eGFP (5′-GCTGAAGCACTGCACGCCGTAGGTCA-3′).

Characterization of NTT-MMP-2 renal proximal tubule-specific transgenic mice.

NTT-MMP-2/eGFP transgene copy numbers were determined by quantitative PCR (qPCR) of genomic DNA using serial dilutions of a known quantity of expression cassette standard DNA. Lines expressing between one and two transgenic inserts per genome or two to three inserts per genome were chosen for evaluation, and each line manifested a similar phenotype. The transgenic lines were maintained as heterozygotes in the FVB/N background. All experiments were performed by comparing the transgenic mice with their age-matched, wild-type littermate controls. For ischemia-perfusion studies, all experiments were performed with male wild-type litter mate controls and transgenic mice.

NTT-MMP-2/eGFP transgene protein expression was confirmed by Western blot for the eGFP epitope tag using methods reported in detail (44). In brief, a NTT-MMP-2/eGFP protein positive control was generated by transfection of the NTT-MMP-2/eGFP expression plasmid into Chinese hamster ovary (CHO) cells (ATCC) using standard methodology. The NTT-MMP-2/eGFP protein from CHO lysates was recovered by affinity capture on gelatin-Sepharose beads (Sigma-Aldrich), as reported (44). Kidneys of euthanized mice were perfused in situ with 4°C PBS until they were free of blood. Excised renal cortices were homogenized in lysis buffer (50 mM Tris·HCl, pH 7.4, 150 mM NaCl, 0.5% Triton X-100, 0.5% CHAPS, and 0.5% sodium deoxycholate plus protease inhibitor cocktail; Thermo Fisher Scientific) and sonicated briefly on ice, and the supernatant was collected after centrifugation at 10,000 g for 20 min. Extracts (150 µg/sample) were incubated overnight at 4°C with 100-µl gelatin-Sepharose beads in 500 µl of binding buffer (50 mM Tris·HCl, pH 7.4), followed by washing three times in binding buffer and elution in an equal volume of 2× SDS-PAGE sample buffer. Western blots used a 1:1,000 dilution of rabbit polyclonal anti-eGFP (Ab 6556; Abcam), followed by horseradish peroxidase-conjugated goat anti-rabbit IgG (Zymed) and detection with ECL Plus reagent (GE Life Sciences).

In situ zymography to localize transgene enzymatic activity was performed on 5-µm renal cortical frozen sections from wild-type littermate controls and NTT-MMP-2/eGFP transgenic mice using DQ-gelatin (Thermo Fisher) according to the manufacturer’s instructions.

Plasma blood urea nitrogen (BUN) from wild-type litter mate controls and NTT-MMP-2/eGFP transgenic mice was determined with a commercial BUN quantitation kit (K024-H1; Arbor Assays).

Analysis of NTT-MMP-2/eGFP transgenic protein association with renal cortical tubule mitochondria.

Kidneys of 4-mo-old NTT-MMP-2/eGFP euthanized mice were perfused in situ with 4°C PBS. Excised renal cortices were homogenized in 0.25 M sucrose, 10 mM HEPES, pH 7.5, 5 mM EDTA, and protease inhibitor cocktail at 4°C, followed by centrifugation at 700 g for 5 min. The supernatants were centrifuged for 5 min at 700 g and the resulting supernatants centrifuged at 10,000 g for 5 min to pellet mitochondria. Pelleted mitochondrial preparations were suspended in 4°C buffer A (2 mM HEPES, 70 mM sucrose, and 0.21 M D-mannitol, pH 7.4) at a concentration of 500 µg mitochondrial protein/500 µl buffer A according to Anholt et al. (3). Digitonin (Sigma-Aldrich) was added to the mitochondrial suspension in buffer A to achieve final concentrations ranging from 0 to 0.6%. The suspensions were incubated at 4°C for 15 min on a rotating shaker, after which the suspensions were diluted threefold with 4°C buffer A and centrifuged for 30 min at 10,000 g at 4°C. The resulting supernatants (20 µg/sample) were analyzed by SDS-PAGE/Western blot using the anti-eGFP antibody detailed above.

Murine model of accelerated atherosclerosis.

As reported in detail, mice were generated expressing the hypomorphic apoE allele in the setting of knockout of the SR-B1 receptor (ApoER61^h/h/SR-B1^−/− or “HypoE mice”) crossed to Mx1-Cre mice (24, 26, 45). Injection of polyinosinic:polycytidylic (pI-pC) acid induces rapid repair of the HypoE allele, normalization of plasma lipids, and stabilization of atherosclerotic plaque burden. For this study, HypoE mice (3 mo old) were placed on a high-fat diet (HFD; 16% fat/1.25% cholesterol; Research Diets) for 22 days. Thereafter, the mice were injected with pI-pC and returned to normal chow. Mice were maintained on normal chow for 6 wk. Hearts and kidneys were removed after 21 days of HFD and following the 6-wk period of lipid normalization and processed for histology and qPCR of MMP-2 isoform transcript abundance. BUN was measured as above at the 6-wk time point.

Histological methods.

At the time of euthanasia, kidneys were perfused with 4°C PBS, followed by 4% buffered paraformaldehyde (4°C). Periodic acid-Schiff (PAS) and Picrosirius Red (PSR) staining were performed using standard methodology. For electron microscopy, kidneys were perfused with 4°C phosphate-buffered saline and fixed in modified Karnovsky’s solution. Toluidine blue sections were prepared using standard methodology. Ultrathin sections were stained with lead citrate-uranyl acetate using standard methodology.

Immunohistochemical methods.

Immunohistochemistry was performed on paraformaldehyde-fixed, 5-µm, paraffin-embedded sections. Sections were deparaffinized with xylene, followed by stepped rehydration. For immunostaining of the full-length MMP-2 (FL-MMP-2), the sections were incubated for 30 min at room temperature with a prediluted [phosphate-buffered saline, pH 7.4, containing 0.1% bovine serum albumin (PBS-BSA)] monoclonal anti-mouse antibody against the NH₂-terminal sequence of the MMP-2 protein (ab54401; Abcam), followed by a 3-min incubation with biotinylated anti-mouse IgG (Vector). For immunostaining of the NTT-MMP-2 isoform, the sections were incubated overnight at 4°C with an affinity-purified goat IgG (5 µg IgG/ml PBS-BSA) targeting the S1′ substrate binding sequence (P³⁴⁷YTYTKNFRLSQDD³⁶¹) located adjacent to the catalytic site (57). This was followed by 30 min of incubation with biotinylated anti-goat IgG (Vector). For immunostaining of the eGFP component of the chimeric NTT-MMP-2/eGFP expression product, sections were incubated for 30 min at room temperature with 1 µg/ml rabbit polyclonal anti-eGFP in PBS-0.1% BSA, (Ab6556; Abcam), followed by 30 min of incubation with affinity-purified biotinylated goat-anti-rabbit IgG (Vector). Development for all three immunohistochemical (IHC) stains was performed by incubation with Vectastain ABC complex, followed by VIP peroxidase substrate and counterstaining with methyl green (Vector).

Measurement of mitochondrial membrane potential (Δψm).

The HK2 human proximal tubular cell line was obtained from ATCC and cultured in complete growth medium (Invitrogen). Subconfluent cultures in 24-well dishes were transiently transfected using FuGENE 6 (Promega) with either a control pcDNA3.1 vector (200 ng/well) or the pcDNA3.31-NTT-MMP-2 expression plasmid at either 100 or 200 ng/well. After 48 h, the cells were harvested into PBS-Ca²⁺- and Mg²⁺-free saline solution (CMF) with 10 mM EDTA and incubated with 25 µM TMRE (tetramethylrhodamine ethyl ester; Abcam) diluted 1:1,000 at 37°C for 20 min. Cells were then pelleted, washed in PBS-CMF, and analyzed by FACS (FACScan; BD Biosciences) with histograms of 10,000 counts using excitation at 488 nm and emission at 590 nm.

TUNEL staining and quantitation.

Fragmentation of tubular epithelial cell nuclear DNA was determined using the TACS-XL Blue kit according to the manufacturer’s instructions (R & D Systems), using renal cortical sections from six wild-type and six NTT-MMP-2-eGFP transgenic mice (8 mo of age). The numbers of TUNEL-positive tubular epithelial cells were counted in 10 transversely cut, randomly chosen tubules in each specimen.

Reactive oxygen species detection.

Fresh renal cortical frozen sections (8 µm) were incubated for 30 min at 37°C with the fluorescent dye, 2′-7′-DCF-diacetate (20 µM, Invitrogen/Molecular Probes). Sections were rinsed and incubated for an additional 15 min at 37°C. Following rinsing the sections were mounted in anti-fade/DAPI medium (Invitrogen) and examined by confocal microscopy. DCF detects several reactive oxygen species, and the fluorescent signal localizes primarily to mitochondria (49).

Kidney qPCR.

Total RNA from renal cortex was isolated using TriZol (Invitrogen). RNA was quantified and normalized to synthesize cDNA using a Transcriptor First Strand cDNA Synthesis Kit (Roche Applied Bioscience). qPCR was performed to quantify the expression of the MMP-2 isoforms and a panel of innate immunity transcripts using a LightCycler 480 SYBR Green I Master Kit (Roche Applied Bioscience). Each sample was plated in triplicate in a 384-well PCR plate (Thermo Fisher). Primer sequences are listed in Table 1. Amplification reactions were performed with 40 cycles (95°C for 15 s, 58°C for 45 s, and 72°C for 1 min) and normalized to β₂-microglobulin. Melt curves were used to verify the absence of primer dimers and other nonspecific products in the amplification reactions. Fold change in mRNA expression was calculated by using the 2^−ΔΔC_T method. Values for each study group were determined (n = 6/group) and expressed as means ± SD.

Table 1.

Primer sequences for qPCR studies

Gene	Forward (5′→3′)	Reverse (5′→3′)
Full-length MMP-2	GACCTCTGCGGGTTCTCTGC	TTGCAACTCTCCTTGGGGCAGC
NTT MMP-2	GGCTCTGGAGCATGACCGCTT	TTGCAACTCTCCTTGGGGCAGC
β2-microglobulin	TAAGCATGCCAGTATGGCCG	AGAAGTAGCCACAGGGTTGG
OAS1A	ATTACCTCCTTCCCGACACC	CAAACTCCACCTCCTGATGC
IFIT1	TGTTGAAGCAGAAGCACACA	TCTACGCGATGTTTCCTACG
IRF-7	CAGCGAGTGCTGTTTGGAGAC	AAGTTCGTACACCTTATGCGG
CXCL10	GCTGCCGTCATTTTCTGC	TCTCACTGGCCCGTCATC
IL-6	GGGAAATCGTGGAAATGAGAAA	AAGTGCATCATCGTTGTTCATACA
COX III	ACCAATGATGGCGCGATGTA	GGCTGGAGTGGTAAAAGGCT
CytB	AACTTCGGCTCACTCCTTGG	GCGTCTGGTGAAGTAGTGCAT

MMP-2, matrix metalloproteinase-2; NTT-MMP-2, NH₂-terminal truncated MMP-2; OAS1A 2′-5′-oligoadenylate synthetase 1A; IFIT1, interferon-induced protein with tetratricopeptide repeats 1; IRF-7, interferon regulatory factor 7; CXCL10, chemokine (C-X-C) ligand 10; IL-6, interleukin-6; COX III, cytochrome c oxidase III; CytB, cytochrome B.

Plasma mitochondrial DNA quantitation.

At the time of euthanization (n = 6 mice for each study group), 0.5 ml of whole blood was placed in heparinized Eppendorf tubes and centrifuged at 1,200 g for 10 min at 4°C. The plasma fraction was recovered and DNA isolated using the DNAEasy Blood and Tissue kit (Qiagen). Isolated DNA was quantified and used as a template for qPCR. Controls consisted of plasma DNA samples from sham-operated mice. Primers for the mitochondrial genes cytochrome c oxidase III (COX III) and cytochrome B (CytB) are listed in Table 1. Amplification reactions were performed with 40 cycles (95°C for 15 s, 60°C for 45 s, and 72°C for 1 min) and normalized to β-microglobulin. Fold change in plasma mitochondrial DNA content was calculated by using the 2^−ΔΔC_T method.

Surgical methods.

FVB/N mice aged 8–10 wk (Charles River Laboratories) were subjected to unilateral renal ischemia-reperfusion injury (n = 6/study group). Flank incisions were made bilaterally in a longitudinal fashion. The kidneys were identified and delivered through the incisions. Blunt dissection was used to skeletonize the renal vessels and separate the ureter. The right kidney was then replaced into the retroperitoneum. The left renal vessels were occluded using an atraumatic vascular clip (S & T vascular clamp; Fine Science Tools). The kidney was placed back into its native position for 40 min. The clip was then removed and reperfusion confirmed. The incisions were closed in two layers. Sham-operated mice were subjected to the identical surgical procedure without vascular clamping. Blood and kidneys were collected for analysis at the time of euthanasia.

Assessment of kidney redox capacity by hyperpolarized ¹³C-dehydroascorbate MR spectroscopy.

Dehydroascorbic acid (DHA) is an oxidized form of vitamin C. The degree of DHA reduction to vitamin C (VitC) has been used as an index of cellular redox capacity in vivo (9). ¹³C-DHA signal was enhanced using dissolution dynamic nuclear polarization (DNP) and injected into mice as a solution containing the probe, as described previously (32–34). NTT-MMP-2 transgenic mice (average age 5 mo; n = 7) and age-matched litter mate controls (n = 4) were fasted 8 h before the experiments. A 2.2-M solution of [1-¹³C]DHA in dimethyacetamide containing 15 mM OX063 trityl radical (Oxford Instruments) was hyperpolarized on a HyperSense DNP instrument (Oxford Instruments). Imaging was performed using a 3T MRI scanner (GE Healthcare). Prior to ¹³C-DHA studies, coronal and axial T2-weighted images were acquired for anatomic localization using a standard fast-spin echo sequence. ¹³C-MR spectroscopic images were acquired 10 s after the injection of 250 μl of 15 mM HP ¹³C-DHA over 15 s. Spectra were analyzed with the custom-developed Sivic software (17). Results are reported as metabolite peak height ratios VitC/(VitC + DHA) and reflect the reduction of DHA to vitamin C. Differences between groups were analyzed using an unpaired t-test and assuming a two-tail Gaussian distribution. Values are presented as means ± SD.

Statistical methods.

Each data set was tested for normality, and either a parametric or nonparametric test was selected. If the comparison was from the same mouse, then the comparison was paired (paired t-test in the case of normal distribution and Wilcoxon matched-pairs and signed-rank test for nonparametric). If from different mice the comparison was unpaired, a t-test for normally distributed data or a Mann-Whitney test for nonparametric data was performed. Significance was set at P < 0.05.

RESULTS

De novo expression of NTT-MMP-2 in renal proximal tubules with increased age.

We examined the kidneys of young (4 mo) and older (14 mo) wild-type FVB/N mice for expression of the FL-MMP-2 isoform using an antibody directed against the NH₂ terminus. We used an antibody raised against the S1′ substrate-binding loop of MMP-2 to stain for the NTT-MMP-2 isoform. Representative results are shown in Fig. 1. There was a low basal level of FL-MMP-2 expression in the renal cortex of 4-mo-old wild FVB/N mice that was not changed in 14-mo-old mice (Fig. 1I, images A and B). No IHC signal was observed using the α-MMP-2 S1′ antibody in 4-mo-old wild-type FVB/N mice, whereas strong IHC staining was readily detected in the proximal tubules of 14-mo-old FVB/N mice (Fig. 1II). IHC staining was concentrated within the basolateral compartment of the tubular epithelial cells in a filamentous pattern consistent with the previously reported mitochondrial association of the NTT-MMP-2 isoform (43, 57). The relative transcript abundance of the FL-MMP-2 and NTT-MMP-2 isoforms in the renal cortex of 4- and 14-mo-old FVB/N mice was determined using an established qPCR method that permits isoform-specific transcript quantitation (57). As shown in Fig. 1III, there was no significant increase in FL-MMP-2 transcript abundance as a function of increasing age. In contrast, there was an approximately threefold statistically significant increase in NTT-MMP-2 transcript abundance as a function of age. The qPCR results confirm the results obtained with the IHC staining and support the conclusion that the α-MMP-2 S1′ antibody is recognizing the NTT-MMP-2 isoform in this setting.

Fig. 1.

The NH₂-terminal truncated matrix metalloproteinase-2 (NTT-MMP-2) isoform, in contrast to the full-length MMP-2 (FL-MMP-2) isoform, is induced by increasing age. I: immunohistochemical (IHC) staining for the FL-MMP-2 isoform of renal cortex from 4- and 14-mo-old wild-type FVB/N mice. There is a low level of FL-MMP-2 cytoplasmic staining of 4-mo-old renal cortex that is not increased at 14 mo (images A and B). II: IHC staining of renal cortex from 4- and 14-mo-old wild-type FVB/N mice, using an antibody directed against the S1′ substrate-binding loop of MMP-2 (α-MMP-2 S1′). There is no detectable IHC signal in the 4-mo renal cortex (image A). There is a strong IHC signal present in the renal cortex of 14-mo-old wild-type FBV/N mice (image B). The staining is concentrated in a filamentous pattern in the basolateral compartment of the tubular epithelial cells (image C, arrow). The filamentous staining pattern, consistent with a mitochondrial association, is particularly prominent in image D, which used pseudocolor enhancement of an image obtained with Nomarski optics (arrow). III: quantitative PCR (qPCR) of FL-MMP-2 and NTT-MMP-2 isoform transcript abundance from renal cortices of 4- and 14-mo-old wild-type FVB/N mice. In contrast to FL-MMP-2, there is a statistically significant increase in NTT-MMP-2 transcript abundance as a function of age. (I: images A and B, ×200; II: images A and B, ×200; image C, ×600; image D, ×1,200; *P < 0.05, n = 6 for each study group). TBM, tubular basement membrane.

Enhanced expression of MMP-2 isoforms in a murine model of diet-induced coronary atherogenesis and myocardial infarction that displays features of heart failure and CKD.

Previously, we have reported in extensive detail on the development and characterization of a murine model of diet-induced occlusive coronary atherogenesis and chronic heart failure using hypomorphic apoE mice deficient in the scavenger receptor type BI, also termed HypoE/SR-B1^−/− Mx1-Cre mice (24, 26, 45). In addition to developing coronary artery atherogenesis associated with myocardial infarction and heart failure, these mice develop diffuse atherosclerosis in all major arterial beds. For the purposes of this study, HypoE/SR-B1^−/− Mx1-Cre mice were given a high-fat diet for 22 days, followed by correction of the hypomorphic apoE allele through inducible Mx1-Cre-mediated recombination and returned to a standard chow diet for 6 wk with normalization of plasma lipids (45). This protocol was chosen to maximize survival and permit the development of CKD, as more prolonged periods of high-fat diet are associated with very high levels of mortality due to heart failure (45).

Compared with hearts from HypoE/SR-B1^−/− Mx1-Cre mice maintained on normal chow, 22 days of feeding a high-fat diet led to extensive, acute myocardial infarction with widespread cardiomyocyte necrosis and inflammatory cell infiltration (Fig. 2I). At 6 wk following correction of the hypomorphic apoE allele, coupled with a normal-fat diet, the hearts showed healed infarcts and cardiomyocyte hypertrophy consistent with chronic ischemic cardiomyopathy. We have reported previously that mice in this setting have severely impaired ventricular systolic function (45).

Fig. 2.

MMP-2 isoform expression is increased in the HypoE model of human type II cardiorenal syndrome. HypoE mice were placed on a high-fat diet (HFD) for 22 days, followed by HypoE allelic correction and return to normal chow (NC) for 6 wk. I, top: left ventricular sections from controls (image A), after 22 days of HFD (image B), and at 6 wk following allelic correction and return to normal chow (image C). There is evidence for extensive acute myocardial infarction with cardiomyocyte necrosis (arrow) and inflammatory infiltration in image B. Image C demonstrates areas of healed myocardial infarction with fibrosis (arrow) and cardiomyocyte hypertrophy characteristic of ischemic cardiomyopathy. I, bottom: renal cortical sections of controls (image D), at 22 days of HFD (image E), and at 6 wk following allelic correction and return to normal chow (image F). There are foci of tubular epithelial cell necrosis and inflammation at 22 days of HFD (image E, arrows). Image F demonstrates a typical wedge-shaped cortical infarct characteristic of atherosclerotic arteriolar occlusion (outlined by arrows) (hematoxylin-eosin; images A–F, ×220) II: immunohistochemical staining with anti-FL-MMP-2 and α-MMP-2 S1′ antibodies of control renal cortices (images A and C) and at 6 wk following HypoE allelic correction and return to normal chow (images B and D). Immunohistochemical signal for both isoforms is increased in images B and D. The immunohistochemical signal is most prominent within dilated tubules (arrows) (images A–D, ×200). III: qPCR of FL-MMP-2 and NTT-MMP-2 isoform transcript abundance in control renal cortices and at 6 wk following HypoE allelic repair and return to normal diet. There are statistically significant increases in the transcript abundance of both MMP-2 isoforms at 6 wk following HypoE allelic repair and return to normal diet (*P < 001; n = 8 for each study group).

Kidneys from HypoE/SR-B1^−/− Mx1-Cre mice fed a high-fat diet for 22 days had foci of tubular epithelial cell necrosis associated with inflammation (Fig. 2I). At 6 wk following allelic correction and return to a normal diet, there were multiple wedge-shaped areas of cortical infarction, which was consistent with arterial and arteriolar atherosclerotic occlusion. Compared with controls, the plasma BUN levels were significantly elevated at 6 wk following allelic correction in the experimental group (control: BUN 22 ± 2.5 mg/dl vs. Hypo E: 68 ± 5 mg/dl; n = 6/group, P < 0.05).

Immunohistochemical staining using antibodies directed against the NH₂ terminus (FL-MMP-2) and the MMP-2 S1′ substrate-binding loop revealed increased staining of renal cortical tubular epithelial cells at 6 wk following allelic correction and return to normal diet (Fig. 2II). Staining with both the FL-MMP-2 and MMP-2 S1′ antibodies was particularly prominent within dilated and atrophic tubular segments.

We used qPCR to confirm the results obtained with immunohistochemistry. As shown in Fig. 2III, transcript abundance of both the FL-MMP-2 and NTT-MMP-2 isoforms was significantly increased at 6 wk following allelic correction and return to normal diet. Our findings demonstrate enhanced renal cortical expression of both MMP-2 isoforms in a model of atherosclerotic heart failure and CKD that clinically simulates human type II cardiorenal syndrome (severe heart failure with associated CKD).

Generation and characterization of NTT-MMP-2/eGFP transgenic mice.

We generated transgenic mice using the proximal tubule-specific type I γ-GT promoter driving expression of a NTT-MMP-2/eGFP transgene. The type I γ-GT promoter is relatively weak and not active until 6 wk of age, at which time murine nephrogenesis is complete (15). A schematic of the transgenic construct is shown in Fig. 3I, and the relative location of the S1′ substrate binding loop is indicated. Founders were identified by genomic PCR, as outlined in materials and methods, and transgene insert copy numbers determined by qPCR of genomic DNA. Expression of the 92-kDa NTT-MMP-2/eGFP fusion protein was confirmed by Western blot of mitochondrial preparations from the renal cortices of three founder lines containing between one and two and two to three transgene copies/genome (Fig. 3II). Enzymatic activity of the transgenic NTT-MMP-2/eGFP fusion protein was confirmed by in situ zymography of renal cortical frozen sections, using DQ-gelatin as a substrate. Enzymatic activity cleaves gelatin and unquenches DQ fluorescence (Fig. 3III). Enzymatic activity is concentrated in the expected basolateral pattern.

Fig. 3.

Characterization of NTT-MMP-2 renal proximal tubular-specific transgenic mice is outlined. I: the NTT-MMP-2 transgene consists of the type I γ-GT promoter driving expression of the NTT-MMP-2/eGFP expression cassette. The relative location of the S1′ substrate-binding loop is depicted. II: Western blot of renal cortical extracts for the NTT-MMP-2/eGFP transgenic fusion protein from transgenic founders and wild-type littermate controls shows expression of the 92-kDa NTT-MMP-2/enhanced green fluorescent protein (eGFP) fusion protein. III: gelatin in situ zymography of control (image A) and NTT-MMP-2/eEGF transgenic renal cortices (image B) shows enzymatic activity of the NTT-MMP-2/eGFP transgene. Gelatinase activity is concentrated in the basolateral aspects of tubular epithelial cells (arrows) (images A and B, ×200).

Confirmation of NTT-MMP-2/eGFP fusion protein trafficking and mitochondrial association.

Figure 4I demonstrates similar patterns of tubular epithelial cell staining when comparing IHC staining with the α-MMP-2 S1′ antibody and an α-eGFP antibody. In both cases, there is no detectable staining in the wild-type controls. With each antibody, the staining is concentrated in the basolateral aspects of the tubular epithelial cells of the NTT-MMP-2 transgenic kidneys, indicating that the presence of the eGFP cassette does not alter cellular trafficking compared with NTT-MMP-2 alone.

Fig. 4.

Cellular trafficking of the NTT-MMP-2/eGFP transgenic (Tg) fusion protein; mitochondrial association shows correct cellular processing. I: there is no detectable IHC signal in the renal cortex of wild-type (WT) controls using the α-MMP-2 S1′ and α-eGFP antibodies (images A and C). IHC signal for both antibodies is present in identical basolateral distributions (images B and D, arrows) in the renal cortices of the NTT-MMP-2/eGFP Tg mice. These findings are consistent with a normal cellular trafficking of the NTT-MMP-2/eGFP fusion protein (images A–D, ×300). II: mitochondrial association of the NTT-MMP-2/eGFP fusion protein is demonstrated by digitonin solubilization. Mitochondria were isolated as detailed in materials and methods and incubated with increasing concentrations of digitonin. Supernatants of pelleted mitochondria were examined by Western blot using a rabbit antibody against eGFP. There is a digitonin concentration-dependent release of the 92-kDa NTT-MMP-2/eGFP fusion protein consistent with a mitochondrial localization with the outer membrane/intramembranous space (left). Quantitative densitometry of the NTT-MMP-2/eGFP fusion protein is shown on the right.

To further confirm proper cellular trafficking of the NTT-MMP-2/eGFP fusion protein to mitochondria, we performed graded digitonin-mediated solubilization of isolated mitochondria from transgenic kidneys according to the method of Anholt et al. (3). The outer membrane of mitochondria is much more sensitive to detergent solubilization than the detergent-resistant inner membrane/matrix (3). Mitochondria were incubated with increasing concentrations of digitonin, as detailed in materials and methods, followed by centrifugation. Western blots of the supernatants, which represent solubilized outer membrane/mitochondrial intramembranous space, show a concentration-dependent release of the 92-kDa NTT-MMP-2/eGFP fusion protein (Fig. 4II). Thus, the NTT-MMP-2/eGFP fusion protein is trafficked to mitochondria in the same manner as that reported previously for NTT-MMP-2 alone (43, 44).

Renal transgenic expression of NTT-MMP-2 induces early proximal tubule epithelial cell necrosis.

Although conventional histology of the NTT-MMP-2 transgenic kidneys was normal at 4 mo of age, examination of toluidine blue-stained semi-thin (0.5 µm) sections of the transgenic kidneys revealed foci of proximal tubule epithelial cells with the typical features of necrosis not seen in the WT controls (Fig. 5). Cellular necrosis has a characteristic morphology that is distinct from apoptosis and is characterized by cytoplasmic and nuclear swelling associated with plasma membrane rupture and loss of organelles (46).

Fig. 5.

Proximal tubule transgenic expression of the NTt-MMP-2 isoform induces epithelial cell necrosis. Semithin (0.5 µm) toluidine blue-stained sections of 4-mo-old wild-type littermate control renal cortex (image A) and NTT-MMP-2/eGFP transgenic renal cortex (image B) are shown. In the transgenic kidneys there are foci of tubular epithelial cells with the typical morphological features of necrosis with cytoplasmic and nuclear swelling and loss of organelles (image B, arrows). (images A and B, ×1,200).

Prolonged NTT-MMP-2 transgene expression induces severe tubular atrophy and mononuclear cellular inflammation in the absence of fibrosis.

Histological analysis of NTT-MMP-2 transgenic kidneys at 8 mo of age demonstrated severe structural abnormalities. Representative examples are depicted in Fig. 6I. Compared with age-matched littermate controls (Fig. 6I, image A), there was extensive necrosis of tubular epithelial cells associated with tubular atrophy (Fig. 6I, images B and C). In addition, there was evidence for dysfunctional tubular basement membrane repair, with thickening and lamination, as well as the occasional development of tubular cyst formation (Fig. 6I, images D and E). Glomerular changes were insignificant other than some mild to moderate mesangial expansion typically seen in mice of this age. The tubular epithelial cell changes were associated with prominent mononuclear cell infiltration (Fig. 6I, image F). In contrast, there was no evidence for a significant induction of interstitial fibrosis in the NTT-MMP-2 transgenic mice as assessed by PSR staining (Fig. 6II). The extensive histological abnormalities of the NTT-MMP-2 transgenic kidneys at 8 mo of age were matched with a corresponding decrease in renal function as determined by quantitation of plasma BUN. Plasma BUN levels in 8-mo-old wild-type littermate controls were 19 ± 2 mg/dl, whereas plasma BUN levels in 8-mo-old NTT-MMP-2 transgenic mice were 92 ± 10 mg/dl (n = 6/group, P < 0.05).

Fig. 6.

Prolonged expression of the NTT-MMP-2/eGFP transgene (8 mo) induces extensive tubular epithelial cell necrosis, tubular atrophy, and mononuclear inflammation. I: periodic acid-Schiff (PAS)-stained cortical sections of 8-mo-old wild-type kidney show normal tubular epithelial cell structure (image A). Eight-month-old NTT-MMP-2 transgenic kidneys demonstrate extensive tubular epithelial cell necrosis with shedding into the tubular lumen (image B, arrows) and formation of atrophic tubular structures (image C, arrow). There is tubular basement membrane replication with thickening (image D) and with occasional acellular cyst formation (image E, arrow) associated with intense interstitial mononuclear cell infiltration (image F, arrow). (images A–D, ×200; image E, ×300; image F, ×400). II: Picrosirius Red staining of NTT-MMP-2 transgenic kidneys does not demonstrate an increase in interstitial collagen. Cortical sections from 8-mo-old wild type (image A) and NTT-MMP-2 transgenic mice (image B) stained with Picrosirius Red show minimal amounts of interstitial collagen (arrows, ×200).

NTT-MMP-2 expression is associated with tubular epithelial cell DNA fragmentation.

Although originally ascribed to apoptosis, DNA fragmentation also occurs during the process of regulated necrosis (39–41). Consistent with this, TUNEL staining of the NTT-MMP-2 transgenic kidneys demonstrated extensive DNA fragmentation within tubular epithelial cell nuclei at 8 mo of age (Fig. 7). We quantified DNA fragmentation in wild-type and NTT-MMP-2/eGFP transgenic kidneys by counting TUNEL-positive nuclei in 10 randomly chosen renal cortical tubular cross sections/kidney. Wild-type kidneys had 0.5 ± 0.3 TUNEL-positive nuclei/10 tubular cross-sections, whereas the NTT-MMP-2 transgenic kidneys had 88 ± 15 TUNEL-positive nuclei/10 tubular cross-sections (n = 6/group, P < 0.05).

Fig. 7.

TUNEL staining of NTT-MMP-2/eGFP transgenic kidneys demonstrates nuclear DNA fragmentation in tubular epithelial cells. Renal cortical sections from 8-mo-old mice were stained using the TUNEL method to detect fragmented DNA. There was very rare TUNEL staining in wild-type kidneys (image A), whereas there was extensive TUNEL staining of tubular epithelial cells in the NTT-MMP-2eGFP transgenic kidneys (image B). This frequently involved entire tubular structures (image C, arrows) (images A and B, ×200; image C, ×400).

NTT-MMP-2 expression induces the mitochondrial permeability transition and mitochondrial membrane depolarization.

Ultrastructural examination of 4 mo old NTT-MMP-2 transgenic kidneys revealed prominent alterations in renal proximal tubule mitochondrial structure. As compared with the age-matched litter mate controls, mitochondria in the NTT-MMP-2 transgenic mice were grossly swollen with loss of organized cristae (Fig. 8I, images A and B). These ultrastructural features are characteristic of the mitochondrial permeability transition. The mitochondrial permeability transition occurs as a consequence of the collapse of the mitochondrial membrane potential, with the resulting influx of solute into the mitochondria (4, 27). This process ultimately leads to mitochondrial swelling, rupture of the mitochondrial membranes, and release of proinflammatory mitochondrial components into the cytosol and extracellular space (2, 8, 23, 48, 52, 61).

Fig. 8.

NTT-MMP-2 induces the mitochondrial permeability transition (MPT). I: transmission electron microscopy of wild-type renal proximal tubule cells reveals normal mitochondrial ultrastructure with intact mitochondrial membranes and highly organized cristae (image A). In contrast, the mitochondria in proximal tubule cells of NTT-MMP transgenic mice are grossly distorted in shape, visibly swollen with disorganized cristae, and display evidence of rupture (image B, arrows) (images A and B, ×1,500). II: direct demonstration that NTT-MMP-2 induces the MPT in vitro; human proximal tubular epithelial HK2 cells preloaded with tetramethylrhodamine ethyl ester (TMRE) were transiently transfected with the cDNA encoding NTT-MMP-2. NTT-MMP-2 induced a concentration-dependent reduction in TMRE signal, consistent with mitochondrial depolarization and the MPT.

To confirm the ultrastructural demonstration of NTT-MMP-2-mediated mitochondrial permeability transition, we determined the effects of transient transfection of cDNA constructs encoding NTT-MMP-2 on TMRE (tetramethylrodamine ethyl ester)-loaded HK2 human proximal tubular epithelial cells. TMRE dye is taken up by normally respiring mitochondria and emits a bright fluorescent signal. However, in the setting of the mitochondrial permeability transition, TMRE is no longer retained within mitochondria and the fluorescent signal decays. Representative results of these experiments are shown in Fig. 8II. It can be seen that transfection with the NTT-MMP-2 cDNA leads to a loss of TMRE signal in a concentration-dependent manner, consistent with NTT-MMP-2-mediated loss of mitochondrial membrane potential and induction of the mitochondrial permeability transition.

Detailed ultrastructural analysis of NTT-MMP-2 transgenic kidneys demonstrates necrosis and autophagy (mitophagy).

Although light microscopic analysis of 4-mo-old NTT-MMP-2 transgenic mice was unremarkable, ultrastructural analysis at this time point revealed a number of pathological abnormalities compared with age-matched littermate controls. Figure 9A shows a typical proximal tubule epithelial cell in the wild-type controls. Characteristic features include the abundant mitochondria arranged in filamentous arrays concentrated in the basolateral compartment of the cell along with a compact nucleus containing dense heterochromatin. In contrast, obviously necrotic cells were frequently observed in the NTT-MMP-2 transgenic kidneys (Fig. 9B). The ultrastructural morphological features of necrosis include disruption of cellular membranes with loss of plasma membrane integrity, nuclear swelling, and loss of mitochondrial structure. Confirming the nature of the mononuclear cell infiltration described at the light level in Fig. 6, we demonstrate mononuclear (lymphocyte) cell attachment to a proximal tubular cell undergoing regulated necrosis (Fig. 9C). There was extensive tubular luminal accumulation of necrotic debris (Fig. 9D) coupled with the development of autophagic vesicles (Fig. 9E). In some cases, these autophagic vesicles contained lamellar mitochondrial remnants characteristic of mitophagy (Fig. 9F).

Fig. 9.

Ultrastructural analysis of 4-mo-old NTT-MMP-2 transgenic mice reveals regulated necrosis, inflammation, and autophagy (mitophagy). Image A: wild-type kidney with normal proximal tubular epithelial cell ultrastructure. There are abundant basolateral mitochondria arranged in linear arrays. There is a compact nuclear structure with prominent heterochromatin. Image B: proximal tubular epithelial cells undergoing regulated necrosis in NTT-MMP-2 transgenic mice. Features include loss of plasma membrane integrity, nuclear expansion, loss of cytosolic organization, and mitochondrial disruption (arrow). Image C: mononuclear cell (arrow) adherent to a tubular epithelial cell in an early phase of regulated necrosis. Image D: tubular lumen filled with necrotic tubular epithelial cell debris (white arrow). Autophagic vesicles present in tubular epithelial cell (black arrow). Image E: higher power image of typical autophagic vesicles in tubular epithelial cells (black arrow). Image F: autophagic vesicles containing lamellar inclusions (black arrows) characteristic of mitochondria (mitophagy). (Images A and B, ×1,400; image C, ×2,500; image D, ×900; image E, ×4,600; image F, ×6,000).

NTT-MMP-2 expression induces mitochondrial reactive oxygen species production.

Loss of the mitochondrial permeability transition interrupts the oxidative phosphorylation pathway and leads to enhanced mitochondrial generation of reactive oxygen species (ROS). We used the fluorescent dye, 2′-7′-DCF-diacetate (DCF) to evaluate the consequences of NTT-MMP-2-mediated loss of the mitochondrial permeability transition for ROS generation. DCF staining of 8-µm frozen renal cortical sections was performed as detailed in materials and methods, and the sections were examined by confocal microscopy. DCF is concentrated primarily within mitochondria and detects several ROS species, including O₂⁻ and H₂O₂ (49). Representative results for these experiments are shown in Fig. 10. There was no detectable DCF staining in 4-mo-old wild-type renal cortices, whereas occasional foci of punctuate DCF staining were seen in age-matched NTT-MMP-2 transgenic kidneys (Fig. 10, A and B). By 8 mo of age, there was intense cortical staining in the NTT-MMP-2 transgenic kidneys, with punctuate DCF staining characteristic of a mitochondrial localization (Fig. 10, C and D) Thus, NTT-MMP-2 expression results in enhanced mitochondrial ROS production.

Fig. 10.

NTT-MMP-2 induces renal tubular epithelial reactive oxygen species (ROS). Renal cortical frozen sections from 4- (image A: wild type; image B: NTT-MMP-2 transgenic) and 8-mo-old (image C: wild type; image D: NTT-MMP-2 transgenic) mice were stained with the ROS detection agent 2′-7′-DCF-diacetate (DCF), as detailed in materials and methods. No ROS signal was detected in the wild-type kidneys at either 4 or 8 mo of age (images A and C). Foci of DCF signal were present in the cortices of the 4-mo-old NTT-MMP-2 transgenic mice (image B). Diffuse DCF signal in a characteristic punctuate patterns characteristic of a mitochondrial localization was present in the renal cortices of 8-mo-old NTT-MMP-2 transgenic mice (×200).

NTT-MMP-2-mediated oxidative stress can be detected by rea- time hyperpolarized ¹³C-dehydroascorbic acid in vivo.

We conducted hyperpolarized ¹³C-DHA MR experiments to investigate the effect of NTT-MMP-2 expression and enhanced ROS production on the in vivo redox capacity in the kidneys of transgenic mice (average age 5 mo). Hyperpolarized ¹³C-MR is a new molecular imaging technique that allows rapid and noninvasive monitoring of dynamic metabolic and physiological processes in real time. DHA is rapidly taken up by cells and reduced to vitamin C via a glutathione (GSH)-dependent mechanism. It follows that the degree of reduction of DHA to vitamin C thereby reflects cellular redox capacity (32–34). Following the injection of hyperpolarized ¹³C-DHA, localized ¹³C spectroscopy showed a peak at 174 ppm, specific for DHA, with an excellent signal/noise ratio. The peak detected 3.8 ppm downfield corresponds to vitamin C (Fig. 11, A–C). In the kidneys, which are rich in GSH, the vitamin C peak was detected as a result of the intracellular reduction of DHA. We found that NTT-MMP2 transgenic kidneys had a significantly lower redox capacity, as shown by the lower VitC/(VitC + DHA) ratio of 0.23 ± 0.04 compared with 0.30 ± 0.03 in WT (P = 0.002; Fig. 11D). As expected, the voxel containing aorta and inferior vena cava showed only a ¹³C-DHA signal without visible vitamin C signal (Fig. 11C).

Fig. 11.

Kidney redox capacity measured by hyperpolarized ¹³C-dehydroascorbate (DHA) MR spectroscopy in vivo shows ongoing oxidative stress in NTT-MMP-2 transgenic kidneys. A: coronal T2-weighted image of a mouse showing both kidneys and typical voxel placement for 3-dimensional chemical shift acquisitions. B: spectrum from a voxel placed in a kidney, showing conversion of DHA into vitamin C (VitC; red voxel). C: in blue, voxel predominantly containing signal from a blood vessel, showing no metabolism of DHA. D: average results obtained in wild-type (WT; n = 4) and NTT-MMP2 transgenic kidneys (NTT-MMP-2; n = 7). Renal VitC-to-DHA + VitC ratios showed a significantly lower reduction of DHA to VitC in NTT-MMP2 mice (P = 0.002). This measure is an index of renal redox capacity and reflects decreased glutathione concentration.

Renal proximal tubule expression of NTT-MMP-2 sensitizes kidneys to enhanced ischemia/reperfusion injury.

Mitochondria are intimately involved in the both the defense and response to renal ischemia-reperfusion injury. We hypothesized that NTT-MMP-2-mediated mitochondrial dysfunction, with enhanced underlying oxidative stress, would sensitize kidneys to more severe ischemia-reperfusion injury. For these studies we utilized the unilateral renal ischemia-reperfusion injury model, as we wished to interrogate the nonclamped contralateral kidney for systemic inflammatory responses. We determined in pilot studies that an ischemic period of 40 min was sufficient to induce mild to moderate injury in FVB/N mice according to the injury scale of Day et al. (19). Groups of 4-mo-old male wild-type littermate controls and NTT-MMP-2 transgenic mice (n = 6/group) were subjected to limited unilateral ischemia-reperfusion injury and evaluated at 96 h and 3 wk after surgery. Representative results of PAS-stained cortical sections at 96 h after ischemia-reperfusion injury are shown in Fig. 12I. The contralateral kidneys of the wild-type mice (Fig. 12I, image A) have normal morphology, whereas the wild-type kidney subjected to ischemia-reperfusion injury has evidence of mild to moderate injury based on the degree of tubular dilatation, cast formation, and inflammation (Fig. 12I, image B). In contrast, the kidneys of the NTT-MMP-2 transgenic mice subjected to ischemia-reperfusion injury showed evidence of much more extensive injury, with massive cast formation, tubular dilatation, and cellular inflammation (Fig. 12I, image D). Significantly, the contralateral NTT-MMP-2 transgenic kidneys also showed mild to moderate degrees of injury, with both tubular dilatation and cast formation, consistent with a systemic inflammatory response to unilateral renal ischemia-reperfusion injury.

Fig. 12.

Renal proximal tubule expression of NTT-MMP-2 sensitizes kidneys to enhanced ischemia-reperfusion (I-R) injury. I: PAS-stained images at 96 h following I-R injury of contralateral (CL) and kidneys subjected to I-R injury from wild-type (WT) and NTT-MMP-2 transgenic (Tg) mice. Image A: contralateral wild-type kidney after unilateral I-R injury with normal histology. Image B: wild-type kidney after I-R injury with mild to moderate tubular dilation, occasional cast formation, and minimal mononuclear cell infiltration. Image C: contralateral NTT-MMP-2 transgenic kidney after I-R injury with occasional tubular dilation and cast formation (arrows). Image D: NTT-MMP-2 transgenic kidney subjected to I-R injury; there is extensive tubular dilation, cast formation, and cellular infiltration. II: PAS-stained images at 3 wk following I-R injury: Image A: there is mild to moderate tubular dilatation with rare cast formation in the contralateral kidney of the wild-type mice. Image B: there is moderate tubular dilatation, cast formation, and cellular infiltration in the wild-type kidney subjected to I-R injury. Image C: contralateral kidney of the NTT-MMP-2 Tg mouse has moderate to severe tubular dilation and extensive cellular infiltration. Image D: the renal cortex of the NTT-MMP-2 Tg mouse subjected to I-R injury shows extensive loss of tubular structures with cast formation and intense cellular infiltration. III: Picrosirius Red-stained images at 3 wk following I-R injury. Image A: the contralateral wild-type renal cortex has no significant increase in interstitial collagen deposition. Image B: there are occasional patchy foci of interstitial collagen deposition (arrows) in the renal cortices of wild-type kidneys subjected to I-R injury. Image C: the contralateral renal cortices of the NTT-MMP-2 transgenic mice have dense foci of interstitial collagen deposition along with evident tubular basement membrane thickening. Image D: there is extensive collagen deposition in the renal cortices of NTT-MMP-2 kidneys subjected to I-R injury. (I–III, ×200).

When examined at 3 wk following injury, the kidneys of wild-type mice subjected to ischemia-reperfusion injury showed persistent mild to moderate degrees of injury, with some tubular dilatation, cast formation, and inflammatory cell infiltration (Fig. 12II, image B), whereas the contralateral kidney showed only very limited injury (Fig. 12II, image A). At this time point the kidneys of the NTT-MMP-2 mice subjected to ischemia-reperfusion injury revealed very severe changes, including extensive tubular epithelial cell dropout and mononuclear cell infiltration (Fig. 12II, image D). The contralateral kidneys in the NTT-MMP-2 transgenic mice also revealed moderate to severe changes, with tubular dilatation, tubular epithelial cell dropout, and cellular inflammation (Fig. 12II, image C), consistent with a sustained systemic inflammatory response.

We examined the degree of renal cortical fibrosis using Picrosirius Red (PSR) staining (Fig. 12III). Interstitial collagen deposition at 96 h following ischemia-reperfusion injury was not detectable in either the wild-type or NTT-MMP-2 transgenic kidneys (not shown). At 3 wk following ischemia-reperfusion injury, there were patchy areas of interstitial collagen deposition noted in the wild-type kidneys (Fig. 12III, image B) as compared with the contralateral kidneys (Fig. 12III, image A). In contrast, there was prominent deposition of interstitial collagen in the renal cortices of NTT-MMP-2 transgenic mice subjected to ischemia-reperfusion injury (Fig. 12III, image D). Furthermore, the contralateral transgenic kidneys had dense patches of collagen deposition along with PSR staining that was characteristic of peritubular collagen deposition (Fig. 12III, image C). Thus, renal proximal tubule expression of the NTT-MMP-2 isoform greatly enhances the injury response in the kidneys subjected to ischemia-reperfusion injury and is associated with a sustained systemic inflammatory response, as manifested by the extent of injury in the contralateral kidney.

NTT-MMP-2 enhances a systemic inflammatory response to ischemia-reperfusion injury.

Previously, we reported that NTT-MMP-2 induced a discrete transcriptome consisting of genes characteristic of an innate immune response (43). These transcripts included IL6, OAS-1A, IFIT-1, IRF-7, and CXCL-10. OAS-1A and IFIT-1 promote RNA degradation and cell death, whereas IRF-7 is the dominant transcription factor for the expression of innate immunity genes (43). CXCL-10 is a cytokine that induces monocyte and T cell infiltration (43). We compared the expression of this panel of innate immunity genes in wild-type and NTT-MMP-2 kidneys subjected to unilateral ischemia-reperfusion injury and with the respective contralateral kidneys. The results of these studies are summarized in Fig. 13. Figure 13A shows significantly enhanced expression of OAS-1, IRF-7, and CXCL-10 in NTT-MMP-2 kidneys 96 h after ischemia-reperfusion injury compared with wild-type controls. Consistent with an enhanced systemic inflammatory response, the contralateral kidneys of NTT-MMP-2 mice also showed highly significant increases in OAS-1, IRF-7, and CXDL-10 expression at 96 h following ischemia-reperfusion injury (Fig. 13B). The enhanced innate immune response seen in the NTT-MMP-2 transgenic mice was persistent out to 3 wk following ischemia-reperfusion injury. Enhanced expression of OAS-1, IRF-7, and CXCL-10 was evident in the kidneys of NTT-MMP-2 transgenic mice subjected to ischemia-reperfusion injury and in the contralateral kidneys (Fig. 13, C and D). In contrast, expression of the innate immunity panel genes at 3 wk in the wild-type kidneys was not significantly increased compared with sham-operated controls. Thus, NTT-MMP-2 expression results in an enhanced and durable innate immune response to ischemia-reperfusion injury and a sustained systemic inflammatory response not observed in the wild-type kidneys.

Fig. 13.

NTT-MMP-2 enhances the innate immune response to ischemia-reperfusion injury. Transcript levels of 5 innate immunity genes (IL6, OAS-1A, IFIT-1, IRF-7, and CXCL-10) were measured by qPCR in the renal cortex of wild-type (WT) and NTT-MMP-2 transgenic mice (Tg) at 96 h following ischemia-reperfusion injury (A) and the contralateral kidneys (B). Innate immunity transcripts were quantified as detailed in materials and methods at 3 wk in WT and Tg mice following ischemia-reperfusion-treated kidneys (C) and the contralateral kidneys (D) (n = 6/treatment group; *P < 0.05 and **P < 0.01).

NT-MMP-2 expression results in sustained release of danger-associated molecular patterns following ischemia-reperfusion injury.

Cellular injury leads to the release of multiple molecules with danger-associated molecular pattern (DAMP) features, including histones, phospholipids, and mitochondrial DNA. Mitochondrial DNA is of prokaryotic origin and is a prototypic DAMP that binds to TLR-9, thereby triggering an innate immune response (48, 52, 61). We established a quantitative PCR assay to measure plasma levels of two mitochondria-encoded genes, COX III and CytB. The results of these studies are shown in Fig. 14. Levels of both mitochondrial genes were equally and significantly elevated in plasma from wild-type and NTT-MMP-2 transgenic mice at 96 h following unilateral ischemia-reperfusion injury. At 3 wk following ischemia-reperfusion injury, plasma levels of the mitochondrial genes in the wild-type mice had returned to levels equivalent to sham-operated controls. In contrast, the levels of the two mitochondrial genes remained elevated in the plasma of the NTT-MMP-2 transgenic mice, consistent with ongoing tubular epithelial cell necrosis.

Fig. 14.

NTT-MMP-2 induces a prolonged release of mitochondrial DNA danger-associated molecular patterns following ischemia-reperfusion injury. Plasma levels of mitochondrial COX III and CytB DNA were quantified by qPCR as detailed in materials and methods and compared with plasma from sham-operated controls. A: COX III and CytB DNA levels were similarly elevated at 96 h following I-R injury of both wild-type and NTT-MMP-2 Tg mice. B: COX III and Cyt B DNA levels returned to basal levels at 3 wk following I-R injury in the wild-type mice, whereas mitochondrial DNA levels remained significantly elevated in the NTT-MMP-2 transgenic mice (n = 6/treatment group; *P < 0.05).

DISCUSSION

The current study has five principal findings. First, we have demonstrated that a novel NH₂-terminal truncated isoform of MMP-2 (NTT-MMP-2) is generated de novo in the renal proximal tubules as a function of age in wild-type FVB/N mice. Furthermore, we report that HypoE/SR-B1^−/− Mx1-Cre mice, a model of accelerated atherogenesis, express NTT-MMP-2 and develop CKD in association with ischemic cardiomyopathy, thereby closely simulating human type II cardiorenal syndrome. Second, we show that renal proximal tubule expression of the NTT-MMP-2 isoform directly induces the mitochondrial permeability transition. Third, we show that the NTT-MMP-2 isoform induces regulated necrosis of tubular epithelial cells with resultant tubular atrophy and inflammation. Fourth, we demonstrate that altered renal redox capacity in the NTT-MMP-2 transgenic mice can be detected by real-time in vivo hyperpolarized ¹³C-DHA imaging. Finally, and perhaps clinically most significant, we present data demonstrating that the NTT-MMP-2 isoform sensitizes or “primes” renal tubular epithelial cells to more severe ischemia-reperfusion injury.

De novo expression of NTT-MMP-2 in aging and progressive renal disease.

We initially reported expression of the NTT-MMP-2 isoform in cardiac mitochondria isolated from transgenic mice expressing the FL-MMP-2 isoform (7, 43). The complex cardiac phenotype of this mouse included severe inflammation, oxidative stress, and enhanced sensitivity to ex vivo ischemia-reperfusion injury (7). In addition, we observed de novo cardiac expression of the NTT-MMP-2 isoform in 14-mo-old wild-type mice and in the Hypo E model of accelerated atherogenesis, both of which are associated with enhanced oxidative stress (43). As reported herein, our parallel renal studies also demonstrated the de novo expression of the NTT-MMP-2 isoform in the renal proximal tubules of 14-mo-old wild type mice and in a model of CKD induced by accelerated atherogenesis. We are currently mapping the promoter and enhancer regions of the NTT-MMP-2 alternate promoter regulated by oxidative stress and believe that these studies will possibly provide a mechanistic link between the inflammatory state associated with aging, NTT-MMP-2 transcription, and progressive renal disease (14, 16, 51).

The NTT-MMP-2 isoform directly induces the mitochondrial permeability transition.

The ultrastructural studies of the NTT-MMP-2 transgenic kidneys demonstrated mitochondrial morphological changes consistent with the mitochondrial permeability transition. These features included loss of the highly defined structure of the cristae, which is associated with mitochondrial swelling and rupture. Our transfection studies of HK2 cells with the NTT-MMP-2 cDNA provide direct evidence for mitochondrial depolarization resulting in the mitochondrial permeability transition.

Degradomic studies indicate that MMP-2 cleaves multiple intracellular proteins, including at least eight mitochondrial or mitochondrial-associated proteins (36). Further studies have determined that MMP-2 degrades the mitochondrial ATP synthase β₁-subunit, cytochrome C oxidase subunit 5A, and the NADH dehydrogenase-1α subcomplex, among others (38). Within the context of isolated cardiomyocytes, MMP-2 inhibition resulted in a 30% increase in ATP synthase activity and ATP production (38). Although more detailed studies are needed, we believe it is reasonable to hypothesize that NTT-MMP-2-mediated cleavage of critical components of mitochondrial energy metabolism results in uncoupling of oxidative phosphorylation, enhanced oxidative stress and ultimately the mitochondrial permeability transition.

Mitochondrial associations with MMP-2: a comparison of cardiomyocytes and renal proximal tubular epithelial cells.

Our initial studies of NTT-MMP-2 were focused on cardiomyocyte expression of this isoform in the setting of enhanced oxidative stress, wherein NTT-MMP-2 was not expressed under basal conditions (43). Schulz (54) has demonstrated in a series of studies that high levels of the FL-MMP-2 isoform are retained within cardiomyocytes in association with the sarcomeric apparatus under physiological conditions. In addition, under physiological conditions, significant quantities of the FL-MMP-2 isoform are found in contact with the mitochondrial-associated membrane, an extension of the endoplasmic reticulum that attaches to the mitochondrial outer membrane (29, 55). These observations contrast with our current observations with renal proximal tubular epithelial cells. There is a low, minimally detectable basal level of FL-MMP-2 expression in tubular epithelial cells with a diffuse cytoplasmic localization not consistent with a mitochondrial or mitochondrial-associated membrane distribution. The cytoplasmic localization of FL-MMP-2 in renal proximal tubular epithelial cells described in the current study replicates the cytoplasmic localization of the FL-MMP-2 isoform in human delayed graft function kidneys (57). These observations strongly suggest that there are major differences between cardiomyocytes and renal proximal tubular epithelial cells in terms of the levels of expression and subcellular localization of the MMP-2 isoforms.

The NTT-MMP-2 isoform induces regulated necrosis of tubular epithelial cells with resultant tubular atrophy and inflammation.

We used transgenic mice with low levels of NTT-MMP-2 transgene expression to simulate those observed with aging or progressive chronic renal disease. The resultant phenotype of these mice was complex and evolved over 8–12 mo before the mice died of uremia. Although actually a continuum, we focused on pathological changes at an early time point (4 mo), when the kidneys have normal light microscopic histology, and at a later time point (8 mo), at which time there is overt tubular atrophy and inflammation. In contrast to our earlier findings with proximal tubular expression of the FL-MMP-2 isoform, there was no increase in interstitial fibrosis in the NTT-MMP-2 transgenic kidneys (15). A probable explanation for these differences lies in the sites of MMP-2 enzymatic action (extracellular for FL-MMP-2, intracellular for NTT-MMP-2).

We believe that the pathological changes observed at the early time points are the most informative in terms of understanding the mechanism(s) of action of the NTT-MMP-2 isoform.

The most notable pathological alternations in the NTT-MMP-2 transgenic mice at the early time point were defined by the ultrastructure. There was widespread evidence for the mitochondrial permeability transition associated with the physical rupture of mitochondrial membranes. Concurrently, autophagocytic vesicles were frequently present, some of which contained lamellar remnants of mitochondria (mitophagy). Autophagy in the setting of renal disease and ischemia-reperfusion injury has generally been considered as a cellular defense mechanism to remove damaged organelles, including mitochondria (21, 31, 46). There is a limit to the autophagic capacity to remove damaged organelles, and it is likely, as a function of time, that the progressive accumulation of damaged mitochondria in the NTT-MMP-2 mice overwhelms this defensive capacity. Consistent with this concept is our observation of major increases in mitochondrial oxidative stress in 8-mo-old NTT-MMP-2 transgenic mice, as detected by DCF fluorescence. Accumulation of damaged mitochondria, with intracellular and extracellular release of danger-associated molecular pattern (DAMP) molecules, leads to activation of innate immunity and subsequent regulated necrosis and inflammation. Several cell death pathways have recently been determined to contribute to regulated necrosis in the setting of renal ischemia-reperfusion injury. These include necroptosis, ferroptosis, parthanatos, and the mitochondrial permeability transition (25, 39–41). Based on our current results, we conclude that the regulated necrosis induced by the NTT-MMP-2 isoform is mediated by the latter.

NTT-MMP-2-mediated oxidative stress can be detected by real-time hyperpolarized ¹³C-DHA imaging in vivo.

An unmet clinical need in nephrology is the ability to noninvasively visualize in real time meaningful metrics that reflect underlying disease processes. For the current study, we used hyperpolarized ¹³C-dehydroascorbate (¹³C-DHA) as a probe for renal reduction/oxidation processes. DHA is taken up by renal cells following intravenous injection and can be reduced to vitamin C through glutathione-dependent mechanisms (9, 32–34). We examined wild-type and NTT-MMP-2 transgenic mice with an average age of 5 mo and observed a highly significant decrease in the intracellular reduction of DHA to vitamin C, which is consistent with ongoing oxidative stress. Given that the transgenic mice at this age have an ongoing mitochondrial permeability transition with uncoupling of oxidative phosphorylation, it can be argued that we are measuring mitochondrial-mediated oxidative stress in real time in an intact animal.

The NTT-MMP-2 isoform sensitizes or “primes” renal tubular epithelial cells to more severe ischemia-reperfusion injury.

For this series of experiments, we utilized the unilateral model of renal ischemia-reperfusion injury as developed by Zager and colleagues (59, 60). This model has several advantages over the more widely employed bilateral model of ischemia-reperfusion injury, primarily as the mice do not die from uremia in the setting of more severe ischemia-reperfusion injury. Importantly for our studies, these mice also develop progressive tubular atrophy, inflammation, and fibrosis in the injured kidney over time, offering experimental parallels to the natural progression of human AKI to CKD (37, 60). An additional advantage is the ability to interrogate the contralateral kidney for injury as a metric of the potential systemic effects of unilateral ischemia-reperfusion injury.

Because the degree of ischemia-reperfusion injury in mice is strain dependent (58), we determined experimentally in pilot experiments that 40 min of ischemia-reperfusion was sufficient to induce mild to moderate degrees of injury in FVB/N mice using the assessment scale of Day et al. (19).

We found that NTT-MMP-2 transgenic kidneys were much more sensitive to a given degree of ischemia-reperfusion exposure when examined at 96 h. The degrees of tubular dilation, cast formation, and inflammatory cell infiltration were greatly enhanced in the NTT-MMP-2 transgenic kidneys, and these changes were particularly marked when examined at 3 wk following injury. Significantly, we noted evidence for progressive injury in the contralateral kidneys of the NTT-MMP-2 mice that was not present in the contralateral kidneys of the wild-type controls. Furthermore, plasma levels of two mitochondrial DAMPs (COX III and Cyt B) remained elevated at 3 wk following ischemia-reperfusion injury, consistent with ongoing tubular cell necrosis and systemic inflammation.

The loss of mitochondrial membrane polarization and the subsequent mitochondrial permeability transition are key determinants of the severity of ischemia-reperfusion injury. Reduction in the extent of ischemia-reperfusion injury of several organs, including the heart and kidney, can be provided by ischemic preconditioning, a process that reduces the extent of the mitochondrial permeability transition (4, 27). We reported previously that cardiac-specific transgenic mice expressing MMP-2 had latent mitochondrial dysfunction before the development of functional or histological abnormalities (62). These mice failed to precondition and had much larger infarcts than wild-type controls after ex vivo ischemia-reperfusion injury. In addition, ischemic preconditioning of transgenic hearts failed to improve the oxygen consumption rate and respiratory control ratio, which are basic metrics of mitochondrial oxidative phosphorylation (62). Our ultrastructural analyses of NTT-MMP-2 transgenic kidneys showed an ongoing mitochondrial permeability transition that precedes the development of overt structural injury. Based on these observations, we propose that NTT-MMP-2 reduces mitochondrial reserve, thereby rendering or “priming” tubular epithelial cells to enhanced necrosis following a period of ischemia-reperfusion injury that would otherwise result in more limited damage.

Conclusions, limitations, and clinical implications.

As recently discussed in key position papers, the translation to humans of experimental AKI work, done primarily in rodents, has not resulted in effective preventative or therapeutic approaches (1, 20, 50). Our current transgenic mouse studies were informed by the results of our detailed analysis of human renal transplant delayed graft function (57). We noted a very strong association of NTT-MMP-2 expression, and to a lesser extent FL-MMP-2, with tubular epithelial cell necrosis in transplants with delayed graft function. We determined that MMP-2 isoform expression was limited to cells with the morphological features of necrosis, which was consistent with the observations of NTT-MMP-2 transgenic mice detailed in this report.

We believe that our finding of de novo NTT-MMP-2 expression in the setting of aging and underlying CKD may be of considerable clinical significance, as NTT-MMP-2 greatly enhances the injury response to ischemia-reperfusion. Taken together, our human and murine studies identify NTT-MMP-2 as a potential therapeutic target. However, it should be recognized, given the complexity of AKI pathophysiology, that targeting a single molecular entity, in this case NTT-MMP-2, is unlikely to be sufficient to prevent or treat AKI. We suggest that more expanded therapeutic approaches based on human observations, which may include targeting NTT-MMP-2, will be needed.

GRANTS

D. H. Lovett was supported by Department of Veterans Affairs (VA) Merit Review Award 1-BX000593 and National Institute of Diabetes and Digestive and Kidney Diseases Grant RO1-DK-39776. R. L. Raffai was supported by Department of Veterans Affairs Merit Review Award 1-BX000532 and National Heart, Lung, and Blood Institute Award RO1-HL-133575–01. A. J. Barker was supported by Department of Veterans Affairs Merit Review Award 1BX000740 and American Heart Association Grant-in-Aid Award 15GRNT25550041. D. H. Lovett’s, R. L. Raffai’s, and A. J. Barker’s contributions are the result of work supported with resources and use of the facilities at the San Francisco VA Medical Center. S. Wanga was supported by a Kolff Student Fellowship Abroad Award from the Dutch Kidney Foundation. J. P. Walker was supported by National Institute of Diabetes and Digestive and Kidney Diseases T32 Training Grant DK-007219 (D. H. Lovett). S. H. Song was supported by a grant from the Pusan National University School of Medicine. Z. J. Wang was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant RO1-DK-1097357. P. M. Cowley was supported by American Heart Association Post-Doctoral Fellowship 16POST30970031.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

C.S.C., C.B., S.K.J., S.W., P.M.C., J.P.W., R.M., A.J.B., R.L.R., Z.J.W., and D.H.L. performed experiments; C.S.C., C.B., S.K.J., S.W., P.M.C., J.P.W., S.H.S., A.J.B., and D.H.L. analyzed data; C.S.C., S.K.J., P.M.C., J.P.W., S.H.S., R.L.R., Z.J.W., and D.H.L. interpreted results of experiments; C.S.C. and D.H.L. drafted manuscript; C.S.C., C.B., S.K.J., S.W., P.M.C., J.P.W., S.H.S., R.M., A.J.B., R.L.R., Z.J.W., and D.H.L. approved final version of manuscript; C.B., S.K.J., P.M.C., J.P.W., A.J.B., Z.J.W., and D.H.L. prepared figures; A.J.B., R.L.R., Z.J.W., and D.H.L. edited and revised manuscript; D.H.L. conceived and designed research.