Matrix Metalloproteinases Contribute to Insulin Insufficiency in Zucker Diabetic Fatty Rats

We obtained male ZDF (ZDF/Gmi, fa/fa) and ZLC (ZDF/Gmi, +/?) rats from Charles River Genetic Models (Indianapolis, IN). The animals were shipped 1 week before the experiments to acclimate them to the local environment. Rats were housed three to five per cage under a 12-h light/dark cycle in a pathogen-free environment and with free access to water and Purina 5008 chow (Purina Mills, Richmond, IN). They were studied at 6 (pre-diabetic) and 9 (early diabetic stage) weeks of age. All of the experiments using animals in this study conformed to state and federal guidelines and the published guidelines of the American Association of Laboratory Animal Science.

Female ZDF and ZLC rats were also supplied by Charles River Genetic Models. They were initially fed Purina 5008 but then given either the Gmi-13004 (58% of fat on the basis of kcal %; Research Diet, New Brunswick, NJ) or regular Purina 5008 chow (16.7 kcal % fat) for 5 weeks starting at 8 weeks of age. We monitored body weight, food intake, blood glucose, and serum insulin once a week.

PD166793 [(S)-2-(4′-bromo-biphenyl-4-sulfonylamino)-3-methyl-butyric acid], a broad-spectrum MMP inhibitor (25), was kindly supplied by Drs. Walter Soeller, John Thompson, and Ralph Stevenson (Pfizer, Groton, CT). Female ZDF rats were fed with a mixture of the Gmi-13004 diet and PD166793 custom made by Research Diet for 4–5 weeks starting at 8 weeks of age. The mixture was produced by mixing powders of 72.9 mg PD166793 with 1 kg Gmi-13004 and reformatting to standard rodent pellets to administer 5 mg · kg body wt⁻¹ · day⁻¹ of the compound. The compound-to-food ratio was based on food intake data obtained in a similar feeding study previously completed at our facility. The treatment is expected to maintain a plasma drug concentration of ∼100 μmol/l (27). Nonfasting blood glucose and serum insulin were measured weekly, and the pancreata were isolated at the termination of the feeding study for pancreatic insulin content and islet morphology studies.

Islets from ZLC and ZDF rats at 6 and 9 weeks of age (two animals per condition) were isolated by collagenase digestion and Ficoll gradient separation as previously described (27). After a 1-h culture in RPMI-1640 medium (11 mmol/l glucose), islets were homogenized in 50 mmol/l Tris-HCl (pH 7.4), 150 mmol/l NaCl, 1% Nonidet P-40, 0.5% deoxycholate, 0.1% SDS, and protease inhibitor cocktail (Roche). The homogenates were centrifuged at 14,000g for 10 min at 4°C. Protein concentrations in the supernatant were measured with the Bradford assay (Bio-Rad, Hercules, CA). Two micrograms each of the supernatant sample or the gelatinase zymography standard (human MMP-2 and -9 mix; Chemicon International, Temecula, CA) were mixed with zymogram sample buffer (Bio-Rad), fractionated by electrophoresis on a gel containing 10% gelatin (Bio-Rad), renatured, and assayed according to the manufacturer’s instruction.

Custom rat islet Affymetrix Genechips (Metabolex Rat Islet Oligonucleotide Arrays) containing 22,787 probe sets representing at least 90–95% of the mRNAs expressed in rat islets were designed by Metabolex and manufactured by Affymetrix (17). The Affymetrix standard protocol for chip hybridization was used for RNA hybridization experiments (Affymetrix Genome Array Expression Analysis Protocol [P/N 701119]). Briefly, 5 μg total RNA was used to synthesize double-stranded cDNA with the Superscript Choice System (Invitrogen, San Diego, CA), followed by an in vitro transcription reaction to yield biotin-labeled cRNA. A total of 10 μg cRNA was used to hybridize each individual chip. The Affymetrix Genechip Analysis Suite (version 3.2) was used to convert the image signal of each probe set to an average difference value; the latter is the mean of fluorescence intensity differences between the 20 pairs of perfect-match and mismatch features (probes). The expression levels of each gene were measured as the mean of the average difference values from the four to five replicate samples for each group and compared with control-cultured islets using the Mann-Whitney test in GraphPad Prism.

Islets were isolated from 9-week-old male ZDF and ZLC rats as above. Total RNA was extracted, and cDNA was synthesized from 1 μg islet RNA using Superscript II (Invitrogen). Gene-specific PCR primers and FAM-labeled probes (Assay on Demand) were purchased from Applied Biosystems (ABI, Foster City, CA; ABI probe/primer assay ID nos.: MMP-2 Rn01538167_m1, MMP-12 Rn 00588640_m1, MMP-14 Rn00579172_m1, CTSK Rn00580723_m1, CTSS Rn00580723_m1, β-Actin Rn00667869_m1). Samples were run on an ABI Prism 7700 Sequence Detector and normalized to β-actin levels in each sample. Changes in gene expression were calculated by using the comparative threshold cycle method (available at http://docs.appliedbiosystems.com/pebiodocs/00105622.pdf) and compared for statistical significance with Student’s t test.

Three sections of the whole pancreas from each rat were imaged after staining with insulin antibody. The relative cross-sectional area of β-cells in these images was determined by quantification of the cross-sectional area occupied by β-cells and the total pancreatic cross-sectional area with Metamorph (Universal Imaging, Dowingtown, PA) image analysis software.

Animals were anesthetized with sodium pentobarbital (50 mg/kg i.p.), and the pancreas was quickly separated from adjacent tissues (fat and lymph nodes, etc.) and placed in Bouin’s fixative solution. After 4 h fixation, tissue samples were washed with running tap water, placed in 50% ethanol, and subsequently embedded in paraffin. Pancreas sections (5 μm thick) were cut using a Surgipath microtome (Surgipath Medical Industries, Richmond, IL). All primary antibodies were used at the dilution and incubation time optimized individually in our laboratory. Polyclonal anti-insulin and -glucagon antibodies were from Zymed Laboratories (Foster City, CA). Fluorescein isothiocyanate–and red Alexa-Fluor–labeled secondary antibodies (Molecular Probes, Eugene, OR) were used to visualize the presence of insulin and glucagon, respectively.

Collagen deposition and fibrosis in islets were visualized by Picrosirius Red staining (19). Direct Red 80 and picric acid were obtained from Sigma-Aldrich (St. Louis, MO). Pancreatic sections were deparaffinized and rehydrated, followed by a 2-h incubation in the dark at room temperature with an aqueous solution of saturated picric acid containing 0.1% Direct Red 80. Stained slides were washed slowly under running distilled water for 6 min, dehydrated, mounted, and examined by light microscopy for red-stained collagen fibers.

The male ZDF rats used in this study spontaneously developed overt hyperglycemia at 8 weeks, whereas their heterozygous lean littermates (ZLC) remained normoglycemic throughout the study (Fig. 1A). Since we studied the male ZDF and ZLC rats at ages of 6 and 9 weeks, the ZDF rats were in either a pre-diabetic (6-week) or early diabetic (9-week) stage of the disease, respectively.

The chow-fed female ZDF rats were normoglycemic throughout the observation period in the study (Fig. 1B), yet they were severely obese and had elevated serum insulin levels compared with their heterozygous (ZLC) female littermates (Fig. 1C and D). The high-fat (Gmi-13004) feeding of these rats was initiated at 8 weeks of age and led to hyperglycemia by 2–3 weeks (Fig. 1B). The plasma insulin levels of the female ZDF rats increased in the first 2 weeks of the high-fat feeding but declined sharply after the 3rd week of feeding (Fig. 1D). Similar high-fat feeding in female ZLC rats only induced a slight increase in serum insulin levels and did not affect blood glucose (Fig. 1E).

To elucidate the molecular mechanism of β-cell failure in type 2 diabetes, we obtained expression profiles of pancreatic islets before and after the development of diabetes in both male and female ZDF rats using custom rat islet oligonucleotide arrays (17). Of the 12,000 genes or their splicing variants profiled in these studies, the most striking change we observed in the islets of diabetic ZDF rats was the large increase in expression of a set of genes known to be involved in tissue remodeling (Table 1). The expression of several ECM-degrading proteases, including MMP-2, -12, and -14 and cathepsin-S and -K, were increased by 3.5- to 55-fold in the diabetic islets. MMP-23 mRNA was elevated 10-fold in both male and female ZDF models, and the change was significant in the fat-fed females (P = 0.000277) but not in the males (P = 0.064). The other matrix-degrading proteases that are represented on the custom rat islet arrays (i.e., MMP-7, -9, -11, -13, -16, and -24 and cathepsin-B, -D, -H, and -L) did not display significant changes of at least twofold in either diabetes model. The mRNAs for specific protein inhibitors of MMPs (TIMPs [tissue inhibitor of MMPs]), which are often increased in parallel with the proteases in activated tissues, were also significantly upregulated in the islets of male and female ZDF rats relative to islets from ZLC rats (Table 1).

The large increases in mRNA levels of the proteases observed to increase by oligonucleotide arrays in ZDF rat islets were also observed by quantitative RT-PCR (Table 2). In this experiment, the mRNA levels of MMP-2 and cathepsin-K and -S were increased by 300–500% in 9-week-old ZDF rat islets relative to age-matched islets from ZLC rats. MMP-12 and -14 were increased 4,200 and 9,700%, respectively, in the ZDF islets relative to the ZLC islets.

Tissue remodeling is often associated with angiogenesis and fibrosis. These phenomena were also detected by our expression profiling in islets from both male and female ZDF rats. Many fibrillar collagens and other ECM species (type I, II, III, and XI collagen, fibronectin, etc.) were profoundly increased in islets from 9-week male ZDF and high-fat–fed female ZDF rats relative to their nondiabetic controls (Table 1). In addition, molecular markers of angiogenesis (VEGFR2 [vascular endothelial growth factor receptor 2], VCAM [vascular cell adhesion molecule 1b], etc.) were also significantly increased in the diabetic islets. The induction of the ECM components and proteases in ZDF islets was progressive in both the male and female ZDF rats, starting before the onset of hyperglycemia, but escalating substantially afterward (Fig. 2). The increased collagen mRNA is consistent with the collagen fibers found within diabetic ZDF islets (30) (Fig. 5).

To determine whether MMP activity is increased in the islets in parallel with MMP mRNAs (Fig. 3A), we performed gelatin zymography on islets isolated from male ZDF and ZLC rats at 6 and 9 weeks of age. Gelatinase activity was detected in 6- and 9-week ZLC and 6-week ZDF islets, but the activity migrates much more slowly than either MMP-2 or -9; there was no detectable MMP-2 in these samples (Fig. 3B). In contrast, the 9-week ZDF islet samples display a large amount of gelatinase activity at the mobility of MMP-2, which parallels the large increase in MMP-2 mRNA we observed at this time point in islets from ZDF male rats.

In contrast to the large changes in the expression of the remodeling genes described above, there was little or no change in genes that are known to be selectively expressed in the β-cell (Table 1). The mRNAs for the β-cell transcription factors pancreatic duodenal homeobox-1 and NKX6.1 and the glucose transporter GLUT2 were not substantially different between the ZDF and ZLC islets in both males and high-fat–fed females. Thus, the β-cell dysfunction observed in ZDF islets is probably not due to an overall loss of the β-cell character, either by dedifferentiation of β-cells or by loss of β-cells relative to other islet cell types.

To further evaluate the role of MMPs in β-cell dysfunction, we studied the effect of a broad-spectrum MMP inhibitor on the development of diet-induced diabetes in high-fat–fed female ZDF rats. Eight-week-old female ZDF rats were fed either a high-fat diet (Gmi-13004, 48% kcal of fat) or the same high-fat diet mixed with PD166793 (25,26) for 4 weeks. The mixture was designed to deliver ∼5 mg/kg of the compound daily.

The female ZDF rats were normoglycemic and hyperinsulinemic before the high-fat feeding started. The high-fat feeding induced overt hyperglycemia and a progressive decline in serum insulin levels in 3 weeks. Adding PD166793 to the high-fat diet substantially reduced the rise in blood glucose (Fig. 4A) and the decline of plasma insulin levels (Fig. 4B) induced by the diabetogenic diet. This occurred without affecting food intake and body weight (Fig. 4C and D). In further support of the improvement of glucose homeostasis by the MMP inhibitor, the blood glucose excursion (or glucose area under the curve values) during an intraperitoneal glucose tolerance test (IPGTT) in high-fat + PD166973–fed female ZDF rats was significantly lower than those fed with high-fat diet alone (Fig. 4E).

The improvement of glucose homeostasis in PD166793-treated ZDF rats was a result of the enhancement of β-cell function. The MMP inhibitor–treated animals had significantly greater insulin secretory response during an IPGTT (Fig. 4F), higher pancreatic insulin content (Fig. 4G), and substantially greater β-cell mass (Fig. 4H) relative to the female ZDF rats on high-fat diet alone or chow-fed female ZDF rats (Fig. 4E and F).

Consistent with previous reports (20–23) and in contrast to 9-week ZLC islets, islets in 9-week ZDF rats have lost the normal spherical shape and display finger-like projection into the exocrine pancreas. Also, α-cells were often present in the center of islets, and β-cells were separated irregularly by fibrotic material (Fig. 5). High-fat feeding of female ZDF rats caused increased islet disruption that was similar to that seen in 9-week-old male ZDF rat pancreata, but concomitant treatment with PD166793 with the high-fat diet resulted in islets that appeared more like the islets of the chow-fed rats (Fig. 5).

Consistent with the finding of increased remodeling and fibrosis-associated mRNAs during the development of spontaneous diabetes in male ZDF rats and the diet-induced diabetes of female ZDF rats, we observed deposition of collagen fibers within islets of both pre-diabetic and diabetic ZDF rats. In ZLC islets, collagen fibers were restricted to the islet capsule (Fig. 5). In contrast, a network of collagen fibers infiltrated the majority of the islets in pre-diabetic, chow-fed female ZDF rats, and the nested collagen fibers became denser in the diabetic islets and often occupied over one-third to one-half of the total islet area; PD166973 treatment of high-fat–fed female ZDF rats appeared to prevent this increased density of fibers (Fig. 5).

ZDF rats have a leptin receptor signaling defect that leads to obesity and insulin resistance. These animals also have islet dysfunction, which precipitates diabetes in both males and high-fat–fed females. It has been demonstrated previously that ZDF islets become progressively star shaped, fibrotic, and disorganized during the onset of diabetes (28–31). Our comprehensive profiling studies have now uncovered several molecular components that are substantially increased in ZDF islets contemporaneously with the decline in islet function and morphology. In these studies, we examined >12,000 rat islet–expressed genes for expression changes, but the largest changes occurred in only a few mRNAs, and almost all of these fit in the category of genes involved in tissue remodeling. ZDF islets displayed very large increases in the cysteine proteases cathepsin-K and -S; the metalloproteases MMP-2 (gelatinase A), MMP-12 (macrophage metalloelastase), and MMP-14 (MT1-MMP); and ECM components such as collagens, fibronectin, and fibrillin at the same time these islets are showing evidence of pervasive intraislet collagen deposition.

A large increase in MMP-2 in 9-week-old ZDF males was confirmed by gelatin zymography. MMPs are translated as zymogens, and zymography of homogenates does not indicate whether the proteases are active in the islet. However, since MMP-2 is activated by MMP-14, which is also induced in parallel, and since these MMPs are being induced in parallel with other proteins that normally participate in tissue remodeling, it is reasonable to assume that these large increases in mRNAs for MMP-2, -12, and -14 and, at least, MMP-2 protein result in an increase in overall MMP activity within ZDF islets.

The remodeling genes whose expression is increased in ZDF islets undoubtedly underlie a major part of changes in islet morphology observed in these models. Further, the ability of a class-selective MMP inhibitor to prevent islet dysfunction and diabetes in the fat-fed female ZDF rats suggests that MMPs contribute directly to islet dysfunction in these animals. Since the mRNAs for the β-cell transcription factors pancreatic duodenal homeobox-1 and NKX6.1 and the β-cell glucose transporter were not substantially changed in abundance relative to the total mRNA content of the islet between ZDF and ZLC during the onset of diabetes, the fraction of the islet that is occupied by β-cells is probably also little changed (Table 1).

MMPs are normally expressed in islets, and they may be involved in islet development (31–34). The expression of MMPs and other remodeling genes in islets of these insulin-resistant animals might be triggered as a requisite for islet expansion that normally occurs in response to an increase in insulin demand. MMP activity has also been implicated in pathologies associated with cancer, fibrosis, arthritis, and atherosclerosis, and inhibitors of MMP activity represent a potential therapeutic approach for modification of these pathologies. MMPs have not previously been implicated in islet dysfunction or as etiologic factors in diabetes.

PD166973 prevents β-cell dysfunction and diabetes in female ZDF rats on a high-fat diet. PD166793 is a broad spectrum, but class-selective, MMP inhibitor that has been shown to block left ventricular remodeling and dysfunction in a rat model of heart failure (26). Preservation of normoglycemia and normal glucose tolerance in compound PD166793-treated animals is accompanied by preservation of the high serum insulin levels that are a consequence of the insulin resistance present in these animals. In fact, the high serum insulin levels that are maintained over several weeks of the study in the PD166973-treated animals indicates that the compound is not reversing insulin resistance in these animals. The most striking effect of the compound is on pancreatic insulin content and β-cell volume. High-fat–fed rats treated with the MMP inhibitor were also more morphologically normal than those of untreated high-fat–fed animals. Others have shown that insulin-resistant, but nondiabetic, ZDF rats show a greater than threefold expansion of β-cell mass relative to control lean (and non–insulin-resistant) animals (8). The sharply contrasting insulin insufficiency observed in the ZDF models is probably the result of both a failure to expand β-cell mass and an increase in β-cell apoptosis in the face of insulin resistance and increased insulin demand. Since the MMP inhibitor allows for a marked increase in both pancreatic insulin content and β-cell volume, MMPs appear to contribute to the failure of β-cell expansion in ZDF islets. MMP inhibition increased islet area well above that of even the chow-fed animals but did not completely restore pancreatic insulin content to that found in the chow-fed rats. This difference between effects on β-cell area and insulin content in PD166973-treated animals could be due to the increased demand for insulin in high-fat–fed ZDF female rats.

How might MMPs limit expansion of β-cell mass? MMPs avidly cleave matrix proteins, and this might interfere with critical cell-matrix interactions. MMPs may also degrade growth factors or receptors that are required for β-cell survival and expansion. MMPs appear to play a role in neuronal damage after trauma or ischemia, and in analogy to our studies on islet dysfunction, treatment of mice with a broad-spectrum MMP inhibitor (BB-94) reduced hippocampal neuronal damage after global cerebral ischemia (35). The angiogenesis marker genes VCAM-1b and VEGFR2 are upregulated three- to sixfold in ZDF islets relative to ZLC islets (Table 1), and this may reflect a response to transient ischemia in the islets brought on by initial expansion in the face of an increased demand for insulin.

Our studies demonstrate that islet MMPs contribute to the development of diabetes in ZDF rats. Since human islets also express high levels of MMPs and other remodeling protease mRNAs (J.D.J., A.S., unpublished observations), these studies also suggest that selective protease inhibition may represent a novel approach to preserving islet function in the face of insulin resistance and thereby prevent or reverse the progression to diabetes.