Tissue- and Cell-Specific Co-localization of Intracellular Gelatinolytic Activity and Matrix Metalloproteinase 2

Abstract

Matrix metalloproteinase 2 (MMP-2) is a proteolytic enzyme that degrades extracellular matrix proteins. Recent studies indicate that MMP-2 also has a role in intracellular proteolysis during various pathological conditions, such as ischemic injuries in heart and brain and in tumor growth. The present study was performed to map the distribution of intracellular MMP-2 activity in various mouse tissues and cells under physiological conditions. Samples from normal brain, heart, lung, liver, spleen, pancreas, kidney, adrenal gland, thyroid gland, gonads, oral mucosa, salivary glands, esophagus, intestines, and skin were subjected to high-resolution in situ gelatin zymography and immunohistochemical staining. In hepatocytes, cardiac myocytes, kidney tubuli cells, epithelial cells in the oral mucosa as well as in excretory ducts of salivary glands, and adrenal cortical cells, we found strong intracellular gelatinolytic activity that was significantly reduced by the metalloprotease inhibitor EDTA but not by the cysteine protease inhibitor E-64. Furthermore, the gelatinolytic activity was co-localized with MMP-2. Western blotting and electron microscopy combined with immunogold labeling revealed the presence of MMP-2 in different intracellular compartments of isolated hepatocytes. Our results indicate that MMP-2 takes part in intracellular proteolysis in specific tissues and cells during physiological conditions.

The matrix metalloproteinases (MMPs) are a family of 23 structurally related enzymes of which MMP-2 and MMP-9 constitute the gelatinase subgroup. MMP-2 is the most widely expressed of all MMPs and is found in most tissues and cells (Sariahmetoglu et al. 2007). In contrast, expression of MMP-9 is normally limited to a few cell types, including monocytes, tissue macrophages, polymorphonuclear cells, and keratinocytes (Van den Steen et al. 2002). All MMPs contain an N-terminal signal sequence that directs them to the secretory pathway. Activation of MMP-2 occurs mainly after its release from the cells by proteolytic removal of a pro-peptide but can also occur through binding to macromolecules that induce conformational changes that expose the active site without cleaving off the pro-domain (Sela-Passwell et al. 2010; Hadler-Olsen, Fadnes, et al. 2011). MMP-2 has broad substrate specificity. In addition to gelatin, MMP-2 can degrade several other extracellular matrix (ECM) substrates, including collagen types III, IV, V, VII, X, and XI; fibronectin; vitronectin; laminin; and versican, as well as several non-matrix substrates such as α1-antichymotrypsin, insulin-like growth factor binding protein (IGFBP)-3, IGFBP-5, substance P, interleukin (IL)1-α, pro-tumor necrosis factor (TNF)-α, proMMP-1, proMMP-2, proMMP-9, proMMP-13, fibroblast growth factor receptor 1, and latent transforming growth factor (TGF)α (Visse and Nagase 2003; Butler and Overall 2009).

Lately, various proteins normally localized intracellularly have been reported as substrates for MMP-2 (Butler and Overall 2009; Cauwe and Opdenakker 2010; Ali et al. 2011). These proteins could be released to the extracellular compartment by cell necrosis, but it has also been shown that MMP-2 is active inside some cells. Active MMP-2 has been localized intracellularly in cardiac myocytes after ischemic injuries (Kwan et al. 2004) and in the nucleus of cigarette-exposed apoptotic endothelial cells (Ruta et al. 2009). Recently, we found strong gelatinolytic activity, probably mediated by MMP-2, intracellularly in invading tumor cells (Hadler-Olsen, Wetting, et al. 2011).

Little is known about MMP-2–mediated intracellular proteolysis during physiological conditions. Immunohistochemistry has shown intracellular MMP-2 staining in a number of cell types, but this has usually been interpreted as pro-enzymes on their way through the secretory pathway. The present study was conducted to assess possible intracellular MMP-2 activity in normal tissues and cells. Such information offers new insight into mechanisms of intracellular proteolysis and is crucial to understand the role of intracellular MMP-2. Because of its involvement in a number of pathological conditions, MMP-2, together with other MMPs, is considered as an appealing drug target (Gialeli et al. 2011). However, to be able to develop safe, predictably working MMP inhibitors, all aspects of their functions during physiological as well as pathological conditions should be known.

In situ gelatin zymography is a method for detection and localization of gelatinolytic activity in tissue sections. Recently, we refined this method in a way that allows a much more detailed localization of the cell-associated proteolytic activity (Hadler-Olsen et al. 2010). By use of this method, combined with immunohistochemical staining and western blotting, we show a co-localization between intracellular immunohistochemical staining of MMP-2 and intracellular gelatinase activity in a number of normal mouse tissues and cells. These results suggest that MMP-2 takes part in intracellular proteolysis in several types of tissues and cells during physiological conditions.

Materials and Methods

Tissues and Cells

Eight- to 10-week-old, healthy C57BL/6J mice were sacrificed by an intraperitoneal injection (10 mg) of sodium pentobarbital. Tissue samples from brain, heart, lung, liver, spleen, pancreas, kidney, adrenal gland, thyroid gland, gonads, oral mucosa, salivary glands, esophagus, intestine, and skin were obtained. For each organ, samples were obtained from at least three different animals, and they were immediately fixed in a zinc-based fixative (ZBF) (Beckstead 1994). After fixation, samples were dehydrated and paraffin embedded. The mice were treated in accordance with the guidelines on accommodation and care of animals formulated by the European Convention for the Protection of Vertebrate Animals for Experimental and Other Scientific Purposes, and the project was approved by the Animal Welfare Committee, University of Tromsø. Hepatocytes were isolated from mice as previously described (Hansen et al. 2002). Some of the isolated hepatocytes were fixed in ZBF, dehydrated, and paraffin embedded.

Subcellular Fractionation

A cell suspension of hepatocytes was fractionated into a cytosolic fraction, membranous fraction, and nuclear fraction using the Qproteome cell compartment kit (Qiagen; Hilden, Germany) according to the protocol.

Hepatocyte cell lysate was prepared by incubating hepatocytes with ice-cold lysis buffer (0.1 M Hepes, 10 mM CaCl₂, 1% Triton X-100, and Sigmafast Protease Inhibitor Cocktail Tablets, EDTA-free [Sigma-Aldrich; Steinheim, Germany]) for 15 min. The cells were then centrifuged for 5 min at 6800 × g to pellet the nuclei of the cells. The remaining supernatant (lysate) was collected.

Protein content in the hepatocyte fractions and lysate was determined using the Bio-Rad DC Protein Assay Kit (Bio-Rad Laboratories; Hercules, CA), according to the protocol.

The hepatocyte fractions were concentrated by use of Amicon Ultra 30 K centrifugal filter units (Millipore Corporation; Cork, Ireland), according to the protocol, or a Savant Speed Vac (Thermo Scientific; Waltham, MA). Gelatin-binding proteins in the hepatocyte fraction and lysate were isolated by performing gelatin affinity chromatography. In brief, Gelatin-Sepharose columns were prepared (Gelatin Sepharose 4B; GE Healthcare Bio-Sciences AB, Uppsala, Sweden) and equilibrated (0.1 M Hepes [pH 7.5], 0.05% Brij, 0.5 M NaCl, 10 mM EDTA). Gelatin-binding proteins were eluted using a buffer containing 0.1 M Hepes (pH 7.5), 0.05% Brij, 5% DMSO, and 10 mM EDTA.

The purity of the different fractions was confirmed by western blotting using antibodies against antigens specific to the different subcellular organelles (data not shown). The antibodies used to verify the purity of the fractions are listed in Table 1.

Table 1.

Antibodies Used for Western Blotting

Antibody	Concentration/Dilution
Anti–α-tubulin, Abcam, Cambridge, UK (ab4074)	1 µg/ml
Anti–Lamp-2, Abcam, Cambridge, UK (ab37024)	1 µg/ml
Anti–Laminin-B, Abcam, Cambridge, UK (ab16048)	0.2 µg/ml
Anti–MMP-2, Novus Biologicals, Littleton, CO (NB200-193)	1 µg/ml
Anti–MMP-2, R&D Systems, Minneapolis, MN (AF902)	0.1 µg/ml
Anti–MMP-9, Abcam, Cambridge, UK (ab38898)	1:1000
Anti-biotin, Cell Signaling Technology, Danvers, MA (#7075)	1:500
Anti–Rabbit IgG–horseradish peroxidase, Southern Biotech, Birmingham, AL (#4050-05)	1:2000
Anti–Goat/Sheep IgG, peroxidase conjugate, Sigma-Aldrich, St. Louis, MO (A 9452)	1:50,000

Immunohistochemistry

Immunohistochemistry was performed on 4-µm-thick sections of ZBF-fixed, paraffin-embedded hepatocytes and tissue of the different organs with primary antibodies against MMP-2 (NB200-193, diluted 1:300; Novus Biologicals, Littleton, CO) and MMP-9 (Ab 38898, diluted 1:800; Abcam, Cambridge, MA). Horseradish peroxidase (HRP)–labeled secondary antibody and diaminobenzidine substrate were used for visualization (EnVision+ system-HRP for rabbit primary antibodies; DAKO, Glostrup, Denmark) according to product manuals. Sections were counterstained with Harris hematoxylin (Chemiteknikk; Oslo, Norway). Staining was performed as previously described. Sections in which the primary antibody was replaced by 1.5% normal goat serum were used as negative controls.

Immunogold Analyses

Isolated hepatocytes were fixed in 8% formaldehyde in 200 mM Hepes, dehydrated, and embedded in epoxy resin. Electron microscopic (EM) immunogold detection of MMP-2 was performed with the same MMP-2 primary antibody as used for immunohistochemistry (Novus Biologicals, diluted 1:100) after blocking non-specific binding sites with 1% bovine serum albumin and 0.1% gelatin from cold water fish skin (Sigma-Aldrich; Steinheim, Germany). Gold-conjugated anti–rabbit antibodies (10 nm) were used for detection (CMC; Utrecht, Netherlands). The endometrial adenocarcinoma cell line Ishikawa (Sigma-Aldrich; St. Louis, MO), which expresses very low levels of MMP-2, was used as a negative control. Immunogold labeling was studied in a JEOL JEM-1010 transmission electron microscope (JEOL Ltd.; Tokyo, Japan) equipped with a Morada digital camera (Olympus Corporation; Hamburg, Germany).

In Situ Zymography

Gelatinolytic activity was studied by in situ zymography with DQ-gelatin as substrate on ZBF-fixed tissues and isolated cells as previously described. To analyze the contribution of metalloproteases and cysteinproteases, 20 mM EDTA or 10 µM E-64, respectively, was added to the substrate in control sections. The level of autofluorescence in the tissue was evaluated by substrate incubation on control sections from each tissue at −20C. Fluorescence was studied using a Leica TSC SP5 confocal laser microscope with Leica Application Suite Advanced Fluorescence software (Leica; Wetzlar, Germany). For most of the tissues, at least three independent experiments were performed.

Gelatin Zymography

SDS-PAGE gelatin zymography was used to detect gelatin-degrading enzymes in hepatocytes, as described previously (Malla, et al. 2008). Gels contained 0.1% (w/v) gelatin in both the separating and stacking gel. The stacking gel contained 4% and the separating gel 7.5% (w/v) polyacrylamide. To determine whether the enzymes responsible for the observed gelatinolytic activity were metalloproteinases or not, parallel gels were run without or with 10 mM EDTA (a metalloprotease inhibitor) in the wash and incubation buffers. Human proMMP-2 (72 kDa) and proMMP-9 (92 kDa), from serum-free culture medium of human skin fibroblasts and THP-1 cells, respectively, were mixed and used as molecular weight standards (herein referred to as “S1” when combined, “S2” for proMMP-2 and “S3” for proMMP-9).

Western Blotting

Unpurified as well as Gelatin-Sepharose purified hepatocyte subcellular fractions and whole-cell lysate were mixed with 4× NuPAGE LDS sample buffer (Invitrogen; Carlsbad, CA) and DL dithiothreitol and boiled for 3 to 5 min. About 20 µg of protein was loaded to each well. Human proMMP-2 (72 kDa) and proMMP-9 (92 kDa), from serum-free conditioned medium of human skin fibroblasts and THP-1 cells, respectively, were used as positive controls. Biotinylated Protein Ladder (Cell Signaling Technology Inc, Danvers, MA) was used as molecular mass standard. Samples were then subjected to reducing SDS-PAGE (NuPAGE Novex 4–12% Bis-Tris gels; Invitrogen) before transfer to PVDF membranes. The membranes were blocked in 5% skim milk in TBS-T (150 mM NaCl, 0.25% Tween-20, 20 mM Tris-HCL, pH 7.4) at room temperature for 2 hr and incubated overnight at 4C with primary antibodies as listed in Table 1. Two different antibodies against MMP-2 were used (Table 1) and gave similar results. After primary antibody incubation, membranes were washed in TBS-T 3 × 5 min and incubated with HRP-conjugated goat anti–biotin secondary antibody (Cell Signaling Technology; Danvers, MA), goat anti–rabbit secondary antibody (Southern Biotech; Birmingham, AL), or HRP-labeled mouse anti–goat/sheep secondary antibody (Sigma-Aldrich; St. Louis, MO) for 1 hr and washed with TBS-T 3 × 5 min before visualization using western Blotting Luminol reagent from Santa Cruz Biotechnology (Santa Cruz, CA). The intensity of immunoblot bands was measured using a Luminescent Image Analyzer LAS-3000 with MultiGauge software version 3.0 (Fujifilm; Tokyo, Japan).

Results

To study intracellular MMP-2 activity in normal cells and tissues, we performed a combination of in situ gelatin zymography and immunohistochemical staining on samples from normal mouse heart, lung, liver, spleen, pancreas, kidney, adrenal gland, thyroid gland, gonads, oral mucosa, salivary glands, esophagus, small and large intestine, skin, and brain. Localization and expression level of the two major gelatinases, MMP-2 and MMP-9, were analyzed by immunohistochemistry. On sequential sections of the same tissues, in situ zymography was performed to assess the level of intracellularly active gelatinolytic enzymes. The metalloprotease inhibitor EDTA was added to parallel sections to evaluate the contribution of metalloproteases to the observed gelatinolytic activity. In addition, the cysteine protease inhibitor E-64 was used to evaluate the contribution of cysteine proteases because such enzymes may also have gelatinolytic activity. E-64 had no noticeable influence on the gelatinolytic activity in any of the tissues, suggesting that cysteine proteases did not contribute to the observed gelatinolytic activity (data not shown). Intracellular MMP-2 staining was seen in almost all tissues analyzed (Figs. 1–7), whereas MMP-9 staining was confined to a few, scattered cells, most likely inflammatory cells (data not shown). The results from the immunohistochemical and in situ gelatin zymography analyses are summarized in Table 2.

Figure 1.

Zinc-based fixative (ZBF)–fixed sections of heart (A–C), kidney cortex (D–F), and kidney medulla (G–I). Immunohistochemical (IHC) staining for matrix metalloproteinase 2 (MMP-2) is shown in the left lane (A, D, and G) and is brown; cell nuclei are stained blue by hematoxylin. Gelatin in situ zymography without inhibitors added is shown in the middle lane (B, E, and H) and gelatin in situ zymography with 20 mM EDTA in the right lane (C, F, and I). Gelatinolytic activity is seen as green fluorescence; cell nuclei are stained blue by DAPI. In D to F, glomeruli are labeled gl. In G to I, the renal papilla is labeled rp. Scale bars: 50 µm.

Figure 7.

Zinc-based fixative (ZBF)–fixed sections of the pancreas (A–C) and of the skin (D–F). Immunohistochemical (IHC) staining for matrix metalloproteinase 2 (MMP-2) is shown in A and is brown; cell nuclei are stained blue by hematoxylin. Gelatin in situ zymography without inhibitors is shown in B and gelatin in situ zymography with 20 mM EDTA added in C. Gelatinolytic activity is seen as green fluorescence; cell nuclei are stained blue by DAPI. In D to F, a sebaceous gland is labeled sg. The frizzled layer at the top of the pictures (demarked by a strongly fluorescent line in E and F) is the keratinized epithelial layer. Scale bars: 50 µm.

Table 2.

MMP-2 Staining and In Situ Gelatin Zymography Summarized

Tissue	Co-localized MMP-2 Staining and Gelatinolytic Activity	Correlated MMP-2 Staining Intensity and Gelatinolytic Activity	Gelatinolytic Activity Inhibited by EDTA
Heart****	+++	+++	+++
Salivary glands****	+++	+++	+++
Small intestine***	+++	++	+++
Adrenal gland***	+++	++	+++
Kidney***	+++	+++	++
Liver***	+++	+++	++
Oral mucosa***	+++	+++	++
Thyroid gland***	+++	+++	++
Large intestine**	++	+	++
Trachea**	+	+	+++
Lung**	+	+	+++
Pancreas**	+	+	++
Skin*	+	++	+
Testicle*	+++	+++	+
Esophagus*	+++	+++	+
Ovaries*	+++	++	—
Fallopian tube*	+++	+++	—
Brain*	+	—	+

A summary and comparison of results from matrix metalloproteinase 2 (MMP-2) immunohistochemical staining and in situ gelatin zymography analyses performed with and without the addition of 20 mM EDTA. Tissues with very strong correlation between intracellular localization, staining intensity, and gelatinolytic activity inhibited by EDTA are labeled ****. Tissues with strong correlation between intracellular localization and staining intensity, but with less strong correlation between staining intensity and gelatinolytic activity, or where the EDTA did not inhibit all the gelatinolytic activity, are labeled ***. Tissues with weaker correlation between intracellular localization and staining intensity, as well as with less strong correlation between staining intensity and gelatinolytic activity, are labeled **. Tissues where EDTA had little or no inhibitory effect on gelatinolytic activity are labeled*.

Tissues with Strong Correlation between MMP-2 Staining and Gelatinolytic Activity, Inhibited by EDTA

Heart

Cardiac myocytes showed a pronounced intracellular MMP-2 staining (Fig. 1A) as well as strong intracellular gelatinolytic activity (Fig. 1B). The activity was almost completely eradicated in the presence of EDTA (Fig. 1C), confirming that the gelatinolytic activity was due to one or more intracellular active metalloproteinases.

Kidney

In the kidneys, the proximal and distal tubular cells in the cortex as well as the tubular cells of the outer medulla showed strong MMP-2 staining (Fig. 1D) along with strong gelatinolytic activity (Fig. 1E). Glomeruli showed only weak MMP-2 staining (Fig. 1D) and hardly any gelatinolytic activity (Fig. 1E). Tubular cells in the renal papilla showed a distinctly weaker MMP-2 staining (Fig. 1G) along with weaker gelatinolytic activity (Fig. 1H) compared with tubular cells in the more peripheral medulla and cortex. EDTA abolished most of the observed gelatinolytic activity except in the cell nuclei (Fig. 1F, I), indicating that most of the activity was due to metalloproteases. This is in accordance with our previous reports of MMP-2 staining and gelatinolytic activity (Hadler-Olsen et al. 2010; Hadler-Olsen, et al. 2011).

Liver

Immunohistochemistry revealed strong intracellular MMP-2 staining of the hepatocytes (Fig. 2A). The cytoplasmic staining had a granular appearance. In situ zymography showed moderately strong gelatinolytic activity co-localized with the MMP-2 staining (Fig. 2B). In the presence of EDTA, the gelatinolytic activity was significantly reduced but not completely abolished (Fig. 2C), suggesting that the activity was due to a mixture of metalloproteases and non-metalloproteases.

Figure 2.

Zinc-based fixative (ZBF)–fixed sections of liver (A–C), oral mucosa (D–F), and salivary glands (G–I). Immunohistochemical (IHC) staining for matrix metalloproteinase 2 (MMP-2) is shown in the left lane (A, D, and G) and is brown; cell nuclei are stained blue by hematoxylin. Gelatin in situ zymography without inhibitors added is shown in the middle lane (B, E, and H) and gelatin in situ zymography with 20 mM EDTA in the right lane (C, F, and I). Gelatinolytic activity is seen as green fluorescence; cell nuclei are counterstained blue by DAPI. In D to F, the epithelial lining is labeled ep, the keratinized layer of the epithelium is marked by stippled lines, and lamina propria is labeled lp. In G to I, examples of mucus acini are labeled ma and excretory ducts ed. Scale bars: 50 µm.

Oral Mucosa

In the tongue, all layers of the epithelium, except the keratinized layer, showed strong MMP-2 staining (Fig. 2D) as well as a marked gelatinolytic activity (Fig. 2E). Notably, the keratinized layer showed neither MMP-2 staining nor any gelatinolytic activity. EDTA caused a marked but not total inhibition of the gelatinolytic activity in the tongue epithelium (Fig. 2F). Also, scattered cells in the lamina propria showed MMP-2 staining along with gelatinolytic activity that was inhibited by EDTA (Fig. 2D–F). Thus, the gelatinolytic activity in oral mucosa was mainly due to metallo-dependent proteases.

Salivary Glands

In salivary glands, moderately strong MMP-2 staining was seen in excretory ducts, whereas the acinar cells were negative (Fig. 2G). Accordingly, the gelatinolytic activity was confined to the excretory ducts (Fig. 2H), and this was almost totally abolished with the addition of EDTA (Fig. 2I). Thus, the intracellular gelatinolytic activity in the excretory ductal cells was due to metalloproteases. The pattern of MMP-2 staining and gelatinolytic activity was similar in serous and mucus salivary glands.

Tissues with Moderate Correlation between MMP-2 Staining and Gelatinolytic Activity

Respiratory System

The respiratory epithelium in the trachea showed strong MMP-2 staining (Fig. 3A) but only weak intracellular gelatinolytic activity (Fig. 3B). This activity was completely abolished by the addition of EDTA (Fig. 3C), indicating a metalloprotease-dependent activity. The relatively weak gelatinolytic activity compared with the strong MMP-2 staining suggests that only a fraction of the intracellular MMP-2 was active. Cartilage and glandular cells in the tracheal wall were devoid of both MMP-2 staining and gelatinolytic activity (Fig. 3A,B), whereas smooth muscle cells showed some weak MMP-2 staining and a rather weak but distinct gelatinolytic activity that was completely inhibited by EDTA (Fig. 3A–C). In the lungs, most of the cells showed marked MMP-2 staining (Fig. 3D), whereas the gelatinolytic activity was confined to the respiratory epithelium in the bronchioles (Fig. 3E), indicating that the MMP-2 in the alveolar pneumocytes and endothelial cells was inactive. EDTA partly inhibited the bronchiolar gelatinolytic activity, suggesting that a fraction of the intracellular MMP-2 was active (Fig. 3F).

Figure 3.

Zinc-based fixative (ZBF)–fixed sections of trachea (A–C), lung (D–F), and small intestine (G–I). Immunohistochemical (IHC) staining for matrix metalloproteinase 2 (MMP-2) is shown in the left lane (A, D, and G) and is brown; cell nuclei are stained blue by hematoxylin. Gelatin in situ zymography without inhibitors added is shown in the middle lane (B, E, and H) and gelatin in situ zymography with 20 mM EDTA in the right lane (C, F, and I). Gelatinolytic activity is seen as green fluorescence; cell nuclei are stained blue by DAPI. In A to C, the respiratory epithelium is labeled re, glands are labeled gl, cartilage is labeled cl, and the smooth muscle is labeled sm. In D to F, the bronchioles are labeled br, and in G to I, the intestinal epithelium is labeled ep and the lamina propria lp. Scale bars: 50 µm.

Small Intestine

In the small intestine, very strong MMP-2 staining was seen in all epithelial cells except the goblet cells, as well as in the lamina propria (Fig. 3G). The epithelial cells showed some gelatinolytic activity, whereas the stromal compartment was almost totally negative (Fig. 3H). The addition of EDTA completely inhibited the activity (Fig. 3I). The discrepancy between the strong MMP-2 staining and the relatively low gelatinolytic activity indicates that only a fraction of the intracellular MMP-2 was active.

Adrenal Gland

The adrenal gland showed very strong intracellular MMP-2 staining in the outer part of the cortex and somewhat weaker staining in the inner parts (Fig. 4A). In contrast, the gelatinolytic activity was most prominent in the inner part of the cortex, corresponding to the zona reticularis (Fig. 4B). The gelatinolytic activity was completely abolished by EDTA (Fig. 4C). Cells of the medulla showed weak MMP-2 staining (Fig. 4A) and hardly any gelatinolytic activity (Fig. 4B). The correlation between MMP-2 staining and gelatinolytic activity, which was inhibited by EDTA, strongly suggests that a substantial fraction of MMP-2 in the adrenal cortex cells was active.

Figure 4.

Zinc-based fixative (ZBF)–fixed sections of adrenal gland (A–C), thyroid gland (D–F), and a single seminiferous tubulus of the testicle (G–I). Immunohistochemical (IHC) staining for matrix metalloproteinase 2 (MMP-2) is shown in the left lane (A, D, and G) and is brown; cell nuclei are stained blue by hematoxylin. Gelatin in situ zymography without inhibitors added is shown in the middle lane (B, E, and H) and gelatin in situ zymography with 20 mM EDTA added in the right lane (C, F, and I). Gelatinolytic activity is seen as green fluorescence; cell nuclei are counterstained blue by DAPI. In A to C, the outer cortex of the adrenal gland is labeled oc, the inner cortex is labeled ic, and medulla is labeled med. Scale bars: 50 µm.

Thyroid Gland

Follicular cells of the thyroid gland showed strong MMP-2 staining (Fig. 4D) and moderately strong gelatinolytic activity (Fig. 4E). The addition of EDTA partly inhibited this activity (Fig. 4F), suggesting that some of the intracellular MMP-2 was active along with other types of gelatinolytic enzymes.

Testicles

In the testicles, strong cytoplasmic MMP-2 staining was seen in the entire seminiferous tubulus (Fig. 4G). Moderately strong gelatinolytic activity was seen in cells of the outer part of the tubulus (Fig. 4H), whereas strong gelatinolytic activity was seen in the center of the tubulus, probably corresponding to the most differentiated sperm cells, the spermatozoas. The gelatinolytic activity was partly inhibited by EDTA (Fig. 4I), indicating that some of the activity was due to MMP-2. Notably, the activity was not inhibited in the spermatozoas.

Tissues with Strong Correlation between MMP-2 Staining and Gelatinolytic Activity, Not Inhibited by EDTA

Esophagus

The epithelial lining of the esophagus showed rather strong MMP-2 staining in all layers as well as prominent gelatinolytic activity (Fig. 5A, B). However, the gelatinolytic activity was only slightly inhibited by EDTA (Fig. 5C), indicating that most of the intracellular gelatinolytic activity was due to enzymes other than MMP-2.

Figure 5.

Zinc-based fixative (ZBF)–fixed sections of esophagus (A–C), ovaries (D–F), and fallopian tubes (G–I). Immunohistochemical (IHC) staining for matrix metalloproteinase 2 (MMP-2) is shown in the left lane (A, D, and G) and is brown; cell nuclei are stained blue by hematoxylin. Gelatin in situ zymography without inhibitors added is shown in the middle lane (B, E, and H) and gelatin in situ zymography with 20 mM EDTA in the right lane (C, F, and I). Gelatinolytic activity is seen as green fluorescence; cell nuclei are counterstained blue by DAPI. In D to F, the subcapsular area is labeled sca. Scale bars: 50 µm.

Ovaries and Fallopian Tubes

In the ovaries, strong MMP-2 staining was seen in the subcapsular area, whereas the rest of the stroma and follicles were almost completely negative (Fig. 5D). The gelatinolytic activity showed a similar pattern but was not inhibited by EDTA (Fig. 5E, F). This indicates that the protease activity seen was not metal dependent, and hence MMP-2 was not active in this organ. The epithelial lining of the fallopian tubes showed a rather strong MMP-2 staining (Fig. 5G) accompanied by a prominent gelatinolytic activity that was not inhibited by EDTA (Fig. 5H, I), indicating that MMP-2 was inactive.

Tissues with Poor Correlation between MMP-2 Staining and Gelatinolytic Activity

Brain

The brain showed a very diverse pattern of both gelatinolytic activity as well as MMP-2 expression. In the cerebellum, MMP-2 staining was seen in both the molecular layer as well as in the granular layer of the cortex, whereas the medulla was almost negative (Fig. 6A). In contrast, gelatinolytic activity was mostly confined to the medulla and to scattered cell nuclei in the molecular layer (Fig. 6B). The gelatinolytic activity was partly inhibited by EDTA (Fig. 6C) but was not co-localized with MMP-2 and hence must have been caused by other gelatinolytic enzymes. In the cerebrum, the hippocampus showed strong MMP-2 staining, except in the cell nuclei (Fig. 6D), whereas the gelatinolytic activity was restricted to cell nuclei (Fig. 6E). EDTA had no inhibitory effect on this activity (Fig. 6F), indicating that it was due to non-metalloproteases. The diencephalon showed no MMP-2 staining but focal, weak gelatinolytic activity (data not shown).

Figure 6.

Zinc-based fixative (ZBF)–fixed sections of the cerebellum (A–C) and hippocampus (D–F) of the brain as well as the large intestine (G–I). Immunohistochemical (IHC) staining for matrix metalloproteinase 2 (MMP-2) is shown in the left lane (A, D, and G) and is brown; cell nuclei are stained blue by hematoxylin. Gelatin in situ zymography without inhibitors added is shown in the middle lane (B, E, and H) and gelatin in situ zymography with 20 mM EDTA in the right lane (C, F, and I). Gelatinolytic activity is seen as green fluorescence; cell nuclei are stained blue by DAPI. In A to C, the molecular layer of the cerebellum is labeled ml, the granular layer is labeled gl, and the medulla is labeled m. In G to I, the lumen in the large intestine is labeled lu. Scale bars: 50 µm.

Large Intestine

In the large intestine, moderate to weak MMP-2 staining was seen in the epithelial cells facing the intestinal lumen, whereas epithelial cells in the crypts were unstained (Fig. 6G). In contrast to the MMP-2 staining, the gelatinolytic activity was strongest in the crypts (Fig. 6H). Addition of EDTA inhibited the activity of the epithelial cells facing the lumen completely and most of the activity in the crypts (Fig. 6I). Thus, the inhibitory effect of EDTA was most prominent in the MMP-2–positive cells toward the lumen. The gelatinolytic activity in the crypts is probably caused by a mixture of metalloproteases other than MMP-2 and non-metalloproteases.

Pancreas

The exocrine cells of the pancreatic gland showed strong cytoplasmic MMP-2 staining (Fig. 7A) but very weak gelatinolytic activity (Fig. 7B). This activity was inhibited by EDTA (Fig. 7C), but because it was weak, most of the intracellular MMP-2 in the pancreas was most likely inactive.

Skin

In the skin, very strong MMP-2 staining was seen in the sebaceous glands and in scattered cells in the dermis; otherwise, only weak staining was seen (Fig. 7D). Strong gelatinolytic activity was seen in the sebaceous glands, in scattered cells of the dermis, and in the keratinized epithelial layer, whereas the more basal epithelial cells showed only weak gelatinolytic activity (Fig. 7E). EDTA did not inhibit the activity of the sebaceous glands or the keratinized epithelial layer but completely inhibited the activity in the remaining epithelium and also reduced the gelatinolytic activity in the dermis (Fig. 7F).

Spleen

In the spleen, only scattered MMP-2 staining was seen, and therefore no further analysis was performed (data not shown). In contrast to the other tissues tested, some more MMP-9–positive cells were seen in the spleen.

Isolated Hepatocytes

To study the intracellular MMP-2–mediated proteolysis in more detail, we isolated hepatocytes from the liver of mice. Hepatocytes were chosen because of the strong correlation between MMP-2 staining and gelatinolytic activity in the liver and because the liver is a large, homogeneous organ allowing isolation of a substantial number of cells. Some of the hepatocytes were ZBF fixed and paraffin embedded for immunohistochemical and in situ gelatin zymographic analyses. The isolated hepatocytes showed strong intracellular MMP-2 staining (Fig. 8A), whereas MMP-9 staining was weak (Fig. 8B). Gelatinolytic activity showed the same pattern as the MMP-2 and MMP-9 staining (Fig. 8C). The addition of EDTA markedly reduced the gelatinolytic activity (Fig. 8D), confirming that the activity seen was mainly due to metalloproteases. Thus, both the immunohistochemical staining pattern and the gelatinolytic activity were similar in isolated hepatocytes and in hepatocytes in situ (Fig. 2A–C). In addition, ultra-structural localization of MMP-2 was determined by immunogold EM analyses (Fig. 8E). This revealed that MMP-2 was found in various intracellular compartments such as in the nucleus, in mitochondria in the endoplasmic reticulum, and in the cytosol. The endometrial adenocarcinoma cell line Ishikawa, which expresses very low levels of MMP-2, was used as a control. The Ishikawa cells showed only very scattered MMP-2 labeling (data not shown), indicating that the labeling seen in the hepatocytes was specific.

Figure 8.

(A-D) Sections of zinc-based fixative (ZBF)–fixed, paraffin-embedded isolated hepatocytes. Immunohistochemical staining for matrix metalloproteinase 2 (MMP-2) and matrix metalloproteinase 9 (MMP-9) is brown in A and B, respectively. Nuclei are stained blue by hematoxylin. In situ gelatin zymography without and with 20 mM EDTA is seen in C and D, respectively. Gelatinolytic activity is seen as green fluorescence, whereas nuclei are stained blue by DAPI. Scale bars in A–D: 50 µm. (E) Electron microscopic immunogold MMP-2 labeling of formalin-fixed isolated hepatocytes. MMP-2 labeling is seen as small, round, black spots. Nuclei are labeled nu, examples of mitochondria are labeled m, and endoplasmic reticulum is labeled er. Scale bar: 0.5 µm.

Isolated hepatocytes were fractionated into cytosolic, membranous, and nuclear fractions. In addition, whole-cell lysates were made. The different fractions and the cell lysate were passed through a Gelatin-Sepharose column and the gelatin binding proteins were eluted. Both unpurified whole-cell lysate, various hepatocyte subcellular fractions, and flow-through and fractions bound to a Gelatin-Sepharose column were subjected to SDS-PAGE gelatin zymography and western blotting. Gelatin zymography gels (Fig. 9A) showed bands at various positions in the unpurified cytosolic and membranous fractions, as well as in the whole-cell lysate. The intensity of these bands varied between the different samples, with the strongest bands seen in the whole-cell lysate and the weakest in the membrane fraction. The positions of the bands from these samples were identical or almost identical, indicating that they corresponded to the same enzymes. Rather weak bands could be seen in positions corresponding to molecular masses of about 78 and 88 kDa, and a somewhat stronger band was seen in the position corresponding to about 100 kDa. In the lanes loaded with samples of cytosol and whole-cell lysate, a weak band corresponding to about 92 kDa was also seen. Two additional distinct bands could be seen at about 120 and 140 kDa. The nuclear fraction showed no bands. The samples that had been eluted from the Gelatin-Sepharose column showed no clear bands, whereas weak bands corresponding to about 120 and 140 kDa could be seen in the lanes loaded with the flow-through fraction of the cytosolic and membranous fractions as well as the whole-cell lysate. The addition of EDTA in the washing and incubation buffer abolished all gelatinolytic bands; only protein bands were left. Western blot for MMP-2 (Fig. 9B, left panel) with unpurified samples revealed bands corresponding to 72 kDa in the cytosolic and membranous subcellular fractions as well as in the whole-cell lysate. The band in the cytosol fraction was much weaker than those seen in the membrane fraction and in the whole-cell lysate. The bands corresponded to a molecular mass slightly higher than the MMP-2 standard. This may be because of alternative splice variants, or the phosphorylation of MMP-2 may be different in the standard and the hepatocyte samples, causing charge differences that affect the electrophoresis. The Gelatin-Sepharose purified samples revealed very faint bands corresponding to 72 kDa in all the Gelatin-Sepharose purified fractions as well as in the whole-cell lysate (Fig. 9B, right panel). Two different MMP-2 antibodies were tested, and they both gave the same results. MMP-9 western blots of unpurified subcellular fractions and whole-cell lysate revealed a weak band with molecular mass corresponding to proMMP-9 (92 kDa) in the cytosolic fraction as well as in the whole-cell lysate (Fig. 9C, left panel). In addition, bands corresponding to about 130 kDa were seen. This may be an unspecific band or a complex of MMP-9 with another macromolecule. No bands with molecular mass corresponding to pro- or active MMP-9 could be seen on western blot gels of Gelatin-Sepharose purified fractions (Fig. 9C, right panel).

Figure 9.

Hepatocyte fractions and whole-cell lysate assessed by SDS-PAGE gelatin zymography in A and by western blotting, reducing conditions, for matrix metalloproteinase 2 (MMP-2) in B and matrix metalloproteinase 9 (MMP-9) in C. Labeling of wells: S1: combined MMP-2 and MMP-9 standard; S2: MMP-2 standard (two different batches are used in right and left panels); S3: MMP-9 standard; C: cytosolic fraction; M: membranous fraction; N: nuclear fraction; L: whole-cell lysate; and Mr: molecular weight ladder. The gels labeled “Unpurified” were loaded with samples that had not been purified in a Gelatin-Sepharose column; the gels labeled “Unpurified + EDTA” were loaded with the same samples, but 10 mM EDTA was added to the washing and incubation buffers; the gels labeled “Gelatin binding” were loaded with the samples that had bound to the Gelatin-Sepharose column; and the gel labeled “Flow-through” was loaded with the flow-through fractions that did not bind to the Gelatin-Sepharose column.

Discussion

So far, intracellular MMP-2 activity has mostly been studied in pathological conditions, such as myocardial and brain ischemic injuries and cancer. In the present study, we have focused on intracellular MMP-2 activity in normal tissues and cells. Because MMP-2 is a gelatinase, we used in situ gelatin zymography to detect active MMP-2. A strong correlation was demonstrated between MMP-2 immunohistochemical staining and gelatinolytic activity in a number of normal mouse cells. In hepatocytes, cardiac myocytes, smooth muscle cells, kidney tubuli cells, and epithelial cells of oral and intestinal mucosa, as well as in excretory ducts of salivary glands and adrenal cortical cells, we found that the intracellular gelatinolytic activity was markedly reduced by the metalloprotease inhibitor EDTA but not by the cysteine protease inhibitor E-64. In addition to MMP-2, MMP-9 is known to have strong gelatinolytic activity. We found MMP-9 staining mainly confined to scattered cells in the tissues analyzed. This staining was strong and distinct, suggesting that the antibody worked well. MMP-9 is known to be expressed in various inflammatory cells such as macrophages and neutrophils but less in other cell types. These results therefore suggest that MMP-2, and to a little extent MMP-9, contributed to the intracellular gelatinolysis described above.

Many tissues showed strong intracellular MMP-2 staining, whereas the co-localized gelatinolytic activity was relatively weak, indicating that only a fraction of the intracellular MMP-2 was active. This enzyme may also have intracellular functions that are not associated with proteolysis. There are a few reports of other MMPs being involved in cell signaling through domains other than the catalytic domain (D’Alessio et al. 2008; Mori et al. 2009; Eisenach et al. 2010).

Although MMP-2 and MMP-9 are the most effective gelatinolytic MMPs, other MMPs/metalloproteases are capable of degrading this substrate. However, these enzymes have a much poorer gelatinolytic activity and are not as widely expressed as MMP-2. Cysteine proteases such as caspases, calpains, and some of the cathepsins are known to be important intracellular proteolytic enzymes (Victor and Sloane 2007; Ono and Sorimachi 2012). To investigate if the gelatinolytic activity could be due to any of these enzymes, we added the cysteine protease inhibitor E-64 to control sections. Because E-64 had no detectable effect, we conclude that cysteine proteases did not contribute to the gelatinolytic activity.

The mechanisms directing active MMP-2 to the intracellular compartment, as well as the intracellular activation mechanisms, are not fully understood. A recently published study showed that 40% of the MMP-2 found in human embryonic kidney cells was located in the cytosolic fraction of the cells, whereas 60% was found in the membranous fraction, representing MMP-2 in the secretory pathway. The cytosolic MMP-2 was partly explained by the existence of an alternative MMP-2 splice variant lacking the N-terminal signal sequence and partly by the fact that the signal sequence of canonical MMP-2 was found to be leaky, meaning that it did not target all newly synthesized MMP-2 to the secretory pathway (Ali et al. 2012). The same mechanisms were also shown to localize MMP-2 to the cytosolic fraction in HeLa cells and in heart myocytes (Ali et al. 2012), indicating that this may be a general feature of MMP-2 sorting. Active MMP-2 may also be recruited to the intracellular environment by endocytosis of extracellularly activated enzymes. Endocytosed MMP-2 may be further directed to specific intracellular compartments where it can contribute to proteolytic processing or be degraded or recirculated to the extracellular compartment. These events are believed to take place for MMP-14 (Osenkowski et al. 2004; Ip et al. 2007), another member of the MMP family, which is able to form complex with and activate MMP-2. It has also been demonstrated that MMP-2 can be activated intracellularly either proteolytically by MMP-14 or by MMP-14–independent mechanisms (Hadler-Olsen, Fadnes, et al. 2011). When exposed to oxidative or nitrosative stress, reactive oxygen and nitrogen species can react with the thiol group of the conserved cysteine residue in the pro-peptide, which interacts with the active site Zn-ion and is responsible for the latency of MMPs. This disrupts the Cys-Zn complex and is often followed by an autocatalytic removal of the pro-peptide (Cauwe and Opdenakker 2010). It has also been reported that MMP-2 can be activated by reactive nitrogen species without proteolytic cleavage and removal of the pro-domain and hence gives rise to an allosterically activated MMP-2 with the same molecular weight as the proenzyme (Viappiani et al. 2009). Furthermore, oxidative stress was recently found to induce generation of an intracellular MMP-2 isoform lacking both the N-terminal signal sequence as well as the inhibitory pro-domain (Lovett et al. 2012). In cardiac myocytes, caspases have been found to be potential activators of MMP-2 after myocardial ischemia, and finally, MMP-2 activity may be regulated by phosphorylation (Cauwe and Opdenakker 2010). Five phosphorylated sites have been demonstrated on the surface of recombinant human MMP-2. In vitro de-phosphorylation of MMP-2 with alkaline phosphatase enhanced enzyme activity, whereas phosphorylation by protein kinase C had the opposite effect (Sariahmetoglu et al. 2007). To our knowledge, MMP-2 regulation by phosphorylation has not yet been demonstrated in vivo.

In the present study, MMP-2 and MMP-9 western blots of hepatocyte fractions showed mainly the pro-forms of the enzymes. Although these are often denoted as inactive forms, they may be responsible for part of the intracellular gelatinolytic activity. As mentioned above, reactive nitrogen species can activate MMP-2 without proteolytic removal of the pro-domain. Furthermore, binding of various protein substrates to both MMP-2 and MMP-9 has been demonstrated to induce conformational changes that expose the active site without cleavage of the pro-domain (Bannikov et al. 2002; Fedarko et al. 2004; Jain et al. 2008; Freise et al. 2009).

To our surprise, no bands corresponding to the MMP-2 monomer (72 kDa) could be detected in the zymography gels of the hepatocyte fractions and whole-cell lysate. However, bands corresponding to approximately 140, 120, 100, 92, 88, and 78 kDa could be detected in the unpurified fractions. (The Gelatin-Sepharose purification procedure seemed to dilute the samples beyond the detection limit for most of the enzymes.) These results are in contrast with the western blot analyses that revealed MMP-2 bands of 72 kDa with two different MMP-2 antibodies. As zymography is known to be a much more sensitive method than western blotting, these results were quite unexpected. A possible explanation for the lack of bands of 72 kDa in the zymography gel could be that the isolation and fractionation procedures disrupted the enzymatic activity of MMP-2. Some of the inhibitors in the protease inhibitor cocktails used in the cell fractionation and cell lysate protocols could, for instance, bind irreversibly to the protein. We tested this possibility by adding different inhibitor cocktails to the MMP-2/MMP-9 standard prior to SDS-PAGE zymography analyses. This did not affect the intensity of the MMP-2 or MMP-9 bands (data not shown), thereby making this theory less feasible. All of the bands with higher molecular weight than proMMP-2 in the zymography gel were inhibited by EDTA and are therefore metalloproteases. It is possible that one or several of them represents a complex between MMP-2 and other proteins (Malla, et al. 2008; Koo et al. 2012). We cannot offer any definitive explanation for the discrepancy between results from immunohistochemistry, immunogold EM analyses, Western blotting, and SDS-PAGE gelatin zymography for MMP-2 in hepatocytes. Further analyses are warranted to explore this further.

Intracellular substrates and functions have been demonstrated for MMP-2 after ischemic injuries in brain neuronal cells and cardiac myocytes (Kwan et al. 2004; Yang et al. 2010; Ali et al. 2011; Hill et al. 2012), but intracellular functions of the enzyme during physiological homeostasis are still elusive. We found intracellular gelatinolytic activity, probably mediated by MMP-2, in many different types of cells and tissues. The role of MMP-2 may not be the same in all these cell types, and the enzyme may have several different functions within a single cell. A number of putative intracellular MMP-2 substrates have been recognized by various degradomic approaches. Among these substrates are proteins involved in carbohydrate metabolism, cytoskeletal components and regulators, proteins involved in signal transduction, transcriptional and translational regulation and biosynthesis of proteins, and proteins involved in protein chaperoning, ubiquitination, lysosomal degradation, apoptosis, and redox regulation (Cauwe and Opdenakker 2010). As these potential substrates are involved in normal physiological processes, this can explain the wide, constitutive expression of MMP-2.

Most of the cells that had a substantial amount of putative intracellular MMP-2 activity were of epithelial origin, such as oral squamous epithelial cells, epithelial cells of the intestine, kidney tubular cells, excretory duct cells of salivary glands, and respiratory epithelial cells in bronchioles. These cells share some characteristics such as a high cell turnover and involvement in secretion and/or absorption. In cultured human epidermal cells, the addition of Ca²⁺, which stimulates keratinocyte differentiation, induced MMP-9 expression, whereas retinoic acid, known to stimulate proliferation rather than differentiation, induced the expression of MMP-2 (Kobayashi et al. 1998). We detected hardly any MMP-9 staining in the epithelial cells, whereas the expression level of MMP-2 in tongue epithelium was prominent in the basal layers where there is proliferative activity, which gradually diminished, along with the gelatinolytic activity, in the more differentiated layers. The epithelial lining of the esophagus showed a rather homogeneous MMP-2 staining throughout all layers. This indicates that there are regional differences in the regulation of MMP-2 production and activation in squamous epithelial cells. There was also a striking difference in MMP-2 expression and gelatinolytic activity between epithelial cells of the small and large intestine.

We found strong intracellular MMP-2 staining in heart myocytes that was co-localized with EDTA-inhibited gelatinolytic activity. Active MMP-2 has previously been demonstrated in ischemic cardiac myocytes, where the enzyme is thought to contribute to the reversible loss of contractile force during the postischemic phase (Ali et al. 2011). It has been found that MMP-2 degrades the cytoskeletal protein α-actinin in peroxynitrite-mediated myocardial injury (Sung et al. 2007). The role of MMP-2 in healthy myocardium is not known, but the enzyme might be involved in the regulation of contractile forces also during normal homeostasis.

In several of the organs analyzed, we found nuclear localization of MMP-2, along with cytoplasmic staining/activity, most prominent in kidney tubular cells. Nuclear activity of MMP-2 has previously been observed in heart myocytes, pulmonary endothelial cells, and brain neuronal cells during apoptosis (Kwan et al. 2004; Ruta et al. 2009; Yang et al. 2010; Hill et al. 2012). MMP-2 has a typical nuclear localization sequence close to the C-terminus, which might be involved in the nuclear localization of the enzyme (Kwan et al. 2004). Several factors of the DNA repair machinery are sensitive to MMP-2–mediated degradation (Yang et al. 2010), supporting a role of MMP-2 in apoptosis. Several nuclear proteins have been identified as potential MMP-2 substrates, such as histones and high-mobility group proteins, and MMP-2–mediated processing of these may have a role in physiological conditions (Cauwe and Opdenakker 2010).

Active MMP-2 has been found in exocytic vesicles in cells such as polymorphonuclear leucocytes, endothelial cells, chondrocytes, and various cancer cells, and this is thought to represent a way of achieving rapid directional proteolysis of the ECM upon release (Dean et al. 1996; Ginestra et al. 1997; Taraboletti et al. 2002; Vilen et al. 2008). This mechanism might be employed by other cell types during normal turnover of the ECM, and some of the intracellular MMP-2 activity we observed may be due to such storage vesicles.

Also, intracellular uptake and degradation of ECM components, such as collagen, have been demonstrated in fibroblasts, endothelial cells, and some cancer cells (Ten Cate et al. 1976; van der Zee et al. 1995; Coopman et al. 1998; Cavallo-Medved et al. 2009). Some of the intracellular gelatinolytic activity we observed may therefore be due to endocytosis and intracellular degradation of ECM components surrounding the cells.

The present study clearly indicates that MMP-2 is part of the intracellular proteolytic machinery in many different tissues and cells under physiological conditions. The restricted distribution of putatively active MMP-2 may give clues to which substrates it processes and thereby provide new insight in MMP-2 biology.