Matrix Metalloproteinase 2 (MMP2) and MMP9 Are Produced by Kidney Collecting Duct Principal Cells but Are Differentially Regulated by SV40 Large-T, Arginine Vasopressin, and Epidermal Growth Factor*

We analyzed the expression and regulation of matrix metalloproteinase 2 (MMP2) and MMP9 gelatinases in a rabbit kidney collecting duct principal cell line (RC.SVtsA58) (Prié, D., Ronco, P. M., Baudouin, B., Géniteau-Legendre, M., Antoine, M., Piedagnel, R., Estrade, S., Lelongt, B., Verroust, P. J., Cassingéna, R., and Vandewalle, A. (1991) J. Cell Biol. 113, 951–962) infected with the temperature-sensitive (ts) SV40 strain tsA58. At the permissive temperature (33 °C), cells produced only MMP2. Shifting cells to a nonpermissive temperature (39.5 °C) induced a marked increase in total gelatinolytic activity due to an increase of MMP2 and an induction of MMP9 synthesis. This effect was attributed to large-T inactivation at 39.5 °C because it was abolished by re-infecting the cells with wild-type SV40 strain LP. Run-on experiments showed that negative regulation of MMP2 and MMP9 by large-T was transcriptional and posttranscriptional, respectively. MMP2 and MMP9 were also produced by primary cultures of collecting duct cells. In rabbit kidney, both MMP2 and MMP9 were almost exclusively expressed in collecting duct cells, where an unexpected apical localization was observed. Arginine vasopressin and epidermal growth factor, which exert opposite hydroosmotic effects in the collecting duct, also exhibited contrasted effects on MMP9 synthesis. Epidermal growth factor increased but arginine vasopressin suppressed MMP9 at a posttranscriptional level, whereas MMP2 was not affected. These results suggest a specific physiological role of MMP2 and MMP9 in principal cells of renal collecting duct.

Gelatinases matrix metalloproteinase 2 (MMP2) 1 and MMP9 belong to the broad family of MMPs that contain Zn 2ϩ , require Ca 2ϩ for activity, and are active at neutral pH. MMPs are traditionally subdivided into four classes based on their substrate specificity: (i) interstitial collagenases; (ii) gelatinases or type IV collagenases; (iii) stromelysins, matrilysin, and met-alloelastase; and (iv) membrane-type MMPs. This gene family taken as a whole can degrade all extracellular matrix (ECM) components. Therefore, they play a critical role in tissue remodeling during development and in pathophysiological processes, including inflammation, tissue repair, tumor invasion, and metastasis. However, a growing body of evidence suggests that gelatinases may degrade non-ECM components, such as myelin basic protein (1) and interleukin-1␤ (2), may process the tumor necrosis factor-␣ precursor (3), and may be involved in a variety of physiological processes, such as platelet aggregation (4). Except for MMP2, which is considered to be constitutively expressed, MMPs are highly regulated by growth factors, cytokines, hormones and extracellular matrix components. This regulation occurs at transcriptional and posttranscriptional levels and also involves natural tissue inhibitors of MMPs (TIMPs) (5).
In kidney, MMP2 and MMP9 are present in early stages of development because the mesenchyme of 11-day embryonic kidney already synthesize both gelatinases (6). MMP2 had no effect on kidney morphogenesis in organotypic culture. In contrast, blocking MMP9 activity with specific antibodies or TIMP1 impaired renal morphogenesis by inhibiting growth and branching of the ureteric bud, the embryonic epithelial precursor of the collecting duct (6). In adult kidney, gelatinase activity was detected in glomeruli (7), in epithelial cells that synthesized MMP2 and MMP9 (8,9), and in mesangial cells that produced only MMP2 (10). The phenotype of glomerular mesangial cells (11) can be modified by MMP2 gene expression that is constitutively regulated at a high level by a specific cell-type enhancer promoter element (12,13). In contrast, synthesis of gelatinases by renal tubules is poorly documented. In vitro cultures of proximal tubule and collecting duct epithelia showed secretion of both MMP2 and MMP9 (14,15). However, the collecting duct epithelium is composed of three different cell types including principal cells and ␣and ␤-intercalated cells, making it difficult to identify the cell type responsible for MMP secretion.
We had previously established a rabbit collecting duct epithelial cell line (RC.SVtsA58) infected with the temperaturesensitive SV40 strain tsA58 (16). When cultured at the permissive temperature of 33°C (functional large-T antigen), cells display characteristics of transformed cells. After shifting the cells to the restrictive temperature of 39.5°C (nonfunctional large-T antigen), cells stop dividing and acquire a differentiated phenotype characteristic of collecting duct principal cells including expression of functional type-2 arginine-vasopressin (AVP) receptors (16,17). We had also shown that cells shifted to 39.5°C synthetized a well organized basement membrane. Profound alterations of the ECM were observed in dedifferen-tiated cells at 33°C, including a marked transcriptional decrease of perlecan, a basement membrane proteoglycan (18). Because the basement membrane of renal tubules is mainly composed of type-IV collagen, we asked whether gelatinases could be produced by collecting duct cells and whether their expression could be increased by large-T.
We took advantage of the duality of the RC.SVtsA58 cell line, that is the possibility to switch on or turn off SV40 large-T at each passage level, to determine the gelatinolytic profile of differentiated collecting duct principal cells and to analyze the regulation of MMPs by large-T and by AVP and epidermal growth factor (EGF), two physiological ligands of collecting duct principal cells. In addition, we established the in vivo expression of MMP2 and MMP9 at the apical pole of collecting duct cells, suggesting a specific role of these MMPs in the physiology of principal cells, irrespective of ECM remodeling.
Cell Culture and Experimental Protocol-Primary cultures of rabbit collecting duct cells were obtained as described (21) and provided by M. Tauc (CNRS, Sophia Antipolis, Nice, France). The RC.SVtsA58 renal collecting duct cell line was generated by infection of a primary culture of isolated rabbit renal cortical cells with the SV40 temperature-sensitive mutant tsA58 (22). Experiments were performed between the 30th and 60th passages following the protocol previously described (16). Briefly, cells were seeded in uncoated Petri dishes or in 12-well plastic trays at a concentration of 2 ϫ 10 4 cells/cm 2 and cultured to confluency (day 5) at the permissive temperature (33°C) to allow functional expression of the large-T oncogene. The medium used was a serum-free hormonally defined medium (Dulbecco's modified Eagle's medium-Ham's F-12 1:1 (v/v); transferrin, 5 g/ml; sodium selenate, 30 nM; glutamine, 2 mM; dexamethasone, 5 ϫ 10 Ϫ8 M; insulin, 5 g/ml; HEPES, 20 mM, pH 7.4). On day 5, the medium was changed, and cell cultures were separated in two batches for the next 48 h: one batch was maintained at the permissive temperature, whereas the other one was transferred to the restrictive temperature of 39.5°C. In some experiments, confluent cells were incubated for the last 24 h of culture with EGF (15 ng/ml) or AVP (10 Ϫ7 M). Cells and 48-h conditioned media were separately sampled on day 7 and kept at Ϫ20°C until further analysis.
To analyze the effects of temperature per se irrespective of the functional status of SV40 large-T oncogene on metalloproteinase synthesis, we used as control a rabbit collecting duct cell line of the same cell type origin (RC.SV3), previously established in our laboratory after infection of renal cells with the wild-type SV40 strain LP (23). This cell line was studied under the same culture conditions as RC.SVtsA58 and analyzed at 33 and 39.5°C.
Finally, the effects of SV40 large-T oncogene on the gelatinase profile were also analyzed by re-infecting RC.SVtsA58 cells with a wild-type SV40 strain at the time of temperature shifting. Cells were seeded as described above in 12-well plastic trays, and after 5 days of culture, they were washed and either incubated with 300 l of a stock solution containing SV40 strain LP (multiplicity of infection, 100 plaque-forming units/cell) or mock-infected (control cell population). After 3 h at 37°C, 800 l of hormonally defined medium were added, and cells were transferred for 48 or 72 h at 33 or 39.5°C with one medium change at 24 h. Gelatinolytic activity was analyzed 48 and 72 h after re-infection or mock infection.
Substrate Gel Electrophoresis (Zymography)-Cell-associated and secreted proteinases were detected by zymography. 48-h conditioned media collected from cultures grown at 33 or 39.5°C were centrifuged to remove cell debris. Cell layers of the same culture were lysed in SDSpolyacrylamide gel electrophoresis sample buffer (50 mM Tris, 1% SDS, 5% glycerol, 0.002% bromphenol blue). Both media and cell layers were then stored at Ϫ20°C until further analysis. Lysates and conditioned media of ϳ10,000 cells were subjected to electrophoresis under nonreducing condition in 8% SDS-polyacrylamide gels copolymerized with 1 mg/ml gelatin or type IV collagen. Gels were washed twice for 30 min in 2.5% Triton X-100 to remove SDS, incubated in substrate buffer (50 mM Tris-HCl, 5 mM CaCl 2 , 1 M ZnCl 2 , 0.01% NaN 3 , pH 7.5) overnight at 37°C, stained in 0.5% Coomassie Blue G in 40% methanol, 10% acetic acid for 30 min at room temperature and destained in 40% ethanol, 1% acetic acid. Clear proteolytic zones indicated the presence of gelatinases at their respective molecular weights.
In some experiments, the active forms of gelatinases were induced by incubating the samples for 3 h at 37°C with 1 mM p-aminophenyl mercuric acetate (APMA).
All experiments were done at least 10 times.
14 C-Acetylated Gelatin and TIMP Assays-14 C acetylated gelatin assay and measurement of TIMP activity were performed exactly as described previously (24). The unit of TIMP activity is the amount that will cause 50% inhibition of 2 units (g collagen degraded/min) of collagenase.
Immunoblotting-Conditioned media from cells cultured at 33°C and 39.5°C were concentrated approximately 70 times using Amicon microconcentrators with a 30-kDa cutoff. Samples were submitted to SDS-polyacrylamide gel electrophoresis in a 8% polyacrylamide gel under nonreducing conditions and electrotransferred to nitrocellulose for 90 min at a constant current of 190 mA. Afterward, the nitrocellulose sheet was saturated with 5% dry milk in 0.1% PBS-Tween and 1 mM levamisole for 1 h at 37°C, washed in Tris-buffered saline with 0.1% Tween, and incubated overnight at 4°C with anti-human MMP2 (2 g/ml) or anti-pig MMP9 (2 g/ml) sheep IgGs. This step was followed by a 2-h incubation at room temperature with an anti-sheep IgG antibody (0.2 g/ml) conjugated to alkaline phosphatase. Alkaline phosphatase activity was revealed by adding the nitro blue tetrazolium substrate (nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate complex in 100 mM Tris-HCl, 100 mM NaCl, and 5 mM MgCl 2 , pH 9.5). The reaction was stopped in 20 mM Tris-HCl, 5 mM EDTA, pH 8.0.
Immunoblotting experiments were done at least three times.
Morphological Studies-Expression of gelatinases was analyzed on cryostat rabbit kidney sections. Briefly, a New Zealand White rabbit was anesthetized, and its kidneys were perfused with PBS and snapfrozen in liquid nitrogen. Tissue sections (4 m) were fixed in 4% paraformaldehyde in PBS, pH 7.4, for 10 min at room temperature, for 30 min in 50 mM NH 4 Cl, and further incubated for 30 min in a peroxidase suppressor reagent (Pierce) to neutralize endogenous peroxidases. They were then saturated with 10% dry milk diluted in PBS. Afterward, sections were incubated overnight at 4°C with sheep anti-MMP2 or anti-MMP9 IgG (10 g/ml) diluted in PBS supplemented with 5% dry milk. They were then incubated for 2 h at room temperature with horseradish peroxidase-conjugated rabbit anti-sheep IgG antibody (dilution, 15 g/ml). Enzyme activity was revealed with the 3-amino-9ethylcarbazol substrate dissolved in 50 mM acetate buffer, pH 5.0, supplemented with 0.015% H 2 O 2 . Preparations were counterstained with hematoxylin and examined in a Leitz microscope.
RNA Isolation and Analysis by Northern Blot-Confluent cells at 33 or 39.5°C, cultured in standard conditions or stimulated with EGF or AVP, were lysed in 4 M guanidium thiocyanate, 25 mM sodium citrate, pH 7.0, 0.5% sarkosyl, 100 mM ␤-mercaptoethanol. Total RNAs were then extracted as described by Chomczynski and Sacchi (25). 20 g of RNA for each condition were then analyzed by Northern blot using a NorthernMax TM kit (Ambion, Austin, TX). Briefly, RNAs were electrophoresed in 1% agarose gel under denaturing conditions and transferred for 2 h onto Hybond-N nylon membrane. The bluescript plasmid containing the rabbit MMP9 cDNA (26) was digested with smaI to excise a 1971-base pair cDNA fragment of MMP9. The resulting plasmid containing a 352-base pair 3Ј fragment of rabbit MMP9 cDNA was used as a template to produce an antisense riboprobe. Transcription of MMP9 antisense RNA was carried out with T7 polymerase using a RNA transcription kit (Ambion). Transferred RNAs were then hybridized overnight at 65°C in a hybridization buffer provided with the kit, washed, and exposed to Fuji medical x-ray films at Ϫ80°C.
Run-on experiments were performed at least three times.

RESULTS
Cultured Principal Cells Synthesize and Secrete MMP2 and MMP9 -At the permissive temperature (33°C), gelatin zymograms revealed a major zone of lysis at 68 kDa in conditioned media (Fig. 1A, lane 1) and in cell extracts (Fig. 1B, lane 1). This zone corresponded to the migration of MMP2 identified by immunoblotting (Fig. 1D, lane 1). MMP2 activity normalized per cells number was markedly increased in conditioned media compared with cell extracts (Fig. 1B, lane 1), suggesting that the enzyme was predominantly secreted. In conditioned media, the 68-kDa band of lysis was associated with a faint band at 62 kDa that became predominant upon incubation with APMA (Fig. 1A, lane 2), an organomercurial compound that initiates conversion of the proenzymatic forms of MMPs into their active lower molecular weight forms. Therefore, the upper band of the doublet is most likely the latent form, whereas the lower is the active form of the enzyme.
At the restrictive temperature (39.5°C), cell-associated MMP2 activity increased (Fig. 1B, lane 2), and the profile of secreted gelatinases was profoundly modified (Fig. 1A, lanes 3  and 4). In addition to a 2-fold increase of MMP2 confirmed by Western blotting (Fig. 1D, lane 2), zymograms showed the appearance of a lytic zone at 90 kDa (Fig. 1A, lane 3). The anti-MMP-9 antibody identified a doublet at 90 and 88 kDa (Fig. 1D, lane 4). APMA treatment induced lower molecular mass bands, corresponding most likely to active forms of MMP9 as shown in Fig. 1A, lane 4. The gelatinolytic pattern of RC.S-VtsA58 conditioned media at 39.5°C was similar to that of collecting duct cells in primary culture (Fig. 1C), a finding that is in keeping with the differentiation process induced by shifting RC.SVtsA58 cells from 33 to 39.5°C.
Gelatinase activity was also measured in conditioned media using the 14 C-gelatin assay. No activity was detected in samples from RC.SVtsA58 cells kept at 33°C, most likely because of lower sensitivity compared with zymography. By contrast, enzymatic activity was readily measurable in the media of cells shifted to 39.5°C (0.75 units/ml/10 6 cells, n ϭ 2). TIMP activity was 2-3-fold lower at 33°C (0.32 units/ml/10 6 cells, n ϭ 2) than at 39.5°C (0.73 units/ml/10 6 cells, n ϭ 2). This result was confirmed by Western blot analysis of conditioned media with an antibody to TIMP1, the main inhibitor of MMP9 (data not shown).
Taken as a whole, these results show that both transformed (RC.SVtsA58 at 33°C) and differentiated (RC.SVtsA58 at 39.5°C and primary cultures) collecting duct cells secrete MMP2, but MMP9 is produced only when cells are differentiated. The differentiation process is associated with a marked increase in gelatinolytic activity despite higher TIMP activity.
Appearance of MMP9 at 39.5°C in RC.SVtsA58 Cells Is Due to Inactivation of SV40 Large-T-To rule out an effect of the increase in temperature (irrespective of large-T functional activity) on MMP expression, we first studied the control cell line RC.SV3 derived from the same renal tubule cell population, but infected with a wild strain of SV40. Gelatinase activity in conditioned media of cells grown at 33°C (Fig. 2A, lane 1) or at 39.5°C (Fig. 2A, lane 2) was restricted to the proteolytic zone of MMP2, and it was not influenced by the culture temperature. In particular, shifting RC.SV3 cells to 39.5°C did not induce the expression of MMP9, in keeping with the sustained activity of wild-type large-T oncogene at 39.5°C.
To confirm the inhibitory effect of large-T on MMPs, we re-infected RC.SVtsA58 cells with the wild-type SV40 strain LP at the time of temperature shifting. In cells transferred to 39.5°C, induction of MMP9 was totally inhibited as early as 48 h after the temperature shift in SV40 strain LP-infected cells (Fig. 2B, lane 4) compared with mock-infected cells (Fig.  2B, lane 3). In parallel, the increase in MMP2 activity was markedly blunted (Fig. 2B, lane 3 versus lane 4). Re-infection did not affect the gelatinase pattern in cells maintained at 33°C (Fig. 2B, lanes 1 and 2). Similar data were observed at 72 h (not shown). These results thus indicate that the changing pattern of gelatinase activity in RC.SVtsA58 cells does not result from a direct effect of temperature but is induced by SV40 large-T functional status.
MMP2 and MMP9 Are Expressed in Collecting Ducts of Normal Rabbit Kidney-To determine whether expression of MMP2 and MMP9 by differentiated RC.SVtsA58 cells and collecting duct primary cultures was induced by culture conditions or could be considered as an additional marker of differentiation, we analyzed their distribution in the normal kidney by immunohistochemistry with the antibodies previously used for immunoblotting. Anti-MMP2 as well as anti-MMP9 antibodies stained tubule sections identified as collecting ducts because of the absence of brush border on phase contrast examination and their association by bunches of 4 -5 units (Fig.   FIG. 1. Gelatin zymograms (A-C) and Western blot (D) of conditioned media (A, C, and D) and cells lysates (B) from the collecting duct cell line RC .SVtsA58 (A, B, and D) 1 and 3, and D, lanes 1 and 2) and induced MMP9 (A, lanes 1 and 3, and D, lanes 3 and 4). 3A). All other nephron segments including glomeruli did not show appreciable labeling. However, a faint reactivity was observed in the kidney interstitium with anti-MMP2 antibody (Fig. 3B). Thus, MMP2 and MMP9 are essentially expressed in the tubule segment from which the RC.SVtsA58 cell line originates. On higher magnification (Fig. 3, C and D), the two MMPs, although mostly implicated in ECM remodeling, were unexpectedly localized at the apical pole of collecting duct cells, suggesting the presence of receptors and/or substrates for these enzymes at the luminal pole of the cells.
MMP9 Is Differentially Regulated by Physiological Ligands of Collecting Duct Cells-EGF and AVP are physiological ligands of principal cells that exert antagonistic hydroosmotic effects on the collecting duct (30,31). We therefore tested the ability of these factors to regulate the expression of MMP2 and MMP9 in RC.SVtsA58 cells. MMP2 expression analyzed by zymography (Fig. 4A) and Western blotting (Fig. 4B) in conditioned media of 33 and 39.5°C cultures was not altered either by 15 ng/ml EGF (Fig. 4, A, lanes 2 and 5, and B, lanes 2 and 4) or by 10 Ϫ7 M AVP (Fig. 4A, lanes 3 and 6), which induces a peak cyclic AMP response (16).
In sharp contrast, EGF and AVP had dramatic effects on MMP9 expression. EGF induced MMP9 secretion by transformed cells (33°C) (Fig. 4, A, lane 2, and C, lane 2). It also increased enzyme expression in differentiated cells at 39.5°C (Fig. 4, A, lane 5, and C, lane 5). Concentration-response experiments indicated that MMP9 induction or stimulation in the two populations of cells (33 and 39.5°C) occurred from 0.5 ng/ml (not shown). Cells cultured at the permissive temperature (33°C) did not express AVP receptors (16) and, as expected, the effect of AVP was only observed in differentiated cells (39.5°C) in which 10 Ϫ7 M AVP markedly inhibited MMP9 activity and antigen shown by zymography (Fig. 4A, lane 6) and Western blotting (Fig. 4C, lane 6), respectively.

Posttranscriptional Regulation of MMP9 Gene Expression by SV40 Large-T and Physiological Ligands of Collecting Duct
Cells-The mechanisms whereby MMP9 was differentially regulated by large-T, AVP and EGF were further investigated by Northern blotting and run-on experiments. We used a riboprobe to detect MMP9 mRNA because of lack of sensitivity of cDNA probe. Northern blot data were in total accordance with those obtained by zymography and Western blot analysis. The riboprobe failed to hybridize with RNAs isolated from transformed cells (33°C) (Fig. 5, lane 1), whereas it strongly hybridized with an approximately 2.5-kilobase mRNA in differentiated cells (39.5°C) (Fig. 5, lane 4). When cells were incubated with EGF (15 ng/ml), MMP9 mRNA hybridization signal was induced in transformed cells (33°C) and increased by about 3-fold in differentiated cells (39.5°C) (Fig. 5, lanes 2 and 5). By contrast, AVP (10 Ϫ7 M) strongly reduced the amount of MMP9 mRNA in differentiated cells (Fig. 5, lane 6).
To investigate whether MMP9 mRNA levels were transcriptionally regulated, we performed run-on assays on isolated nuclei of cells cultured at 33 and 39.5°C under standard conditions or in the presence of EGF (15 ng/ml) or AVP (10 Ϫ7 M) (Fig. 6). In addition to actin, we used GAPDH as housekeeping gene because we had previously shown that actin gene transcription was increased by about 2-fold in transformed RC.S-VtsA58 cells maintained at 33°C (18). Such modulation was confirmed in the present experiments, in which only GAPDH signal was unaltered by large-T functional activity, EGF, and AVP (Fig. 6). In contrast, actin signal was reduced by 2-3-fold compared with GAPDH in differentiated cells (39.5°C), but it was not appreciably altered by EGF and AVP.
In transformed cells, the MMP9 gene was clearly transcribed (Fig. 6) although the enzyme was not detected in conditioned media and cell extracts (Fig. 1, A, lane 1, B, lane 1, and D, lane  3). Surprisingly, in differentiated cells, the level of MMP9 transcription was not significantly increased (Fig. 6), in contrast with MMP9 antigen (Fig. 1D, lanes 3 and 4), enzymatic activity (Fig. 1A, lanes 1 and 3), and mRNA levels (Fig. 5, lanes  1 and 4). Moreover, EGF did not significantly modify the level of MMP9 gene transcription (Fig. 6B) although it induced or increased MMP9 antigen (Fig. 4C, lanes 2 and 5), activity (Fig.  4A, lanes 2 and 5), and mRNA (Fig. 5, lanes 2 and 5). AVP, which had the opposite effects, did not alter either MMP9 transcription (Fig. 6B). These results suggest that in principal cells of the renal collecting duct, regulation of MMP9 expression occurs predominantly at a posttranscriptional level.
On the other hand, run-on analysis showed that transcription of the MMP2 gene was stimulated by 2-fold after large-T antigen inactivation at 39.5°C (Fig. 6A), in accordance with the 2-fold increase of MMP2 antigen (Fig. 1D, lanes 1 and 2) and activity (Fig. 1A, lanes 1 and 3). Thus in contrast to MMP9, MMP2 was essentially regulated at the transcriptional level. DISCUSSION We took advantage of the duality of the RC.SVtsA58 cell line to analyze the gelatinolytic profile of collecting duct cells and its regulation (i) by SV40 large-T at the permissive temperature (33°C) and (ii) by physiological ligands of collecting duct cells (AVP and EGF) at the nonpermissive temperature (39.5°C). We first showed that differentiated collecting duct cells in culture (39.5°C) and in vivo produced the two gelatinases MMP2 and MMP9. Second, we demonstrated that functional expression of large-T antigen reduced MMP2 and suppressed MMP9 at transcriptional and posttranscriptional levels, respectively. Third, we provided evidence that MMP9, but not MMP2, was posttranscriptionally regulated by EGF and AVP, suggesting that MMP9 could play a physiological role in principal cells.
It is generally admitted that MMP2 is regulated differently from the other MMPs and that it is constitutively expressed at low level in normal tissues (32). We showed that cultured RC.SVtsA58 cells secreted large amounts of MMP2 and that, in vivo, the collecting duct was the segment of the renal tubule to express substantial amounts of MMP2 antigen. In other nephron segments, MMP2 specific staining was absent or very faint, although cultured cells from these segments were shown to produce MMP2 (14). The discrepancy between the in vivo and in vitro data may be due to rapid secretion of the enzyme in the extracellular milieu. In culture, the secreted enzyme can accumulate in the medium, being therefore easily detectable by zymography. Unlike MMP2, MMP9, which was originally iden-  2 and 5; B, lanes 2 and 4) or AVP (A, lanes 3 and 6). tified in polymorphonuclear leukocytes and macrophages (33,34), is produced by a small number of cell types. However, its expression can be readily regulated by a number of agents, including growth factors, hormones, cytokines, and extracellular matrix molecules (35), which explains why MMP9 is often associated with inflammation, tissue injury and tumor invasion. In normal rabbit kidney, we showed that MMP2 and MMP9 were only detected at the apical pole of collecting duct cells. In addition, both MMPs were secreted partially in active forms as attested by the presence of a lower molecular weight band on zymograms and Western blots. This finding suggests the expression of a MT-MMP required for MMP2 activation (36,37) at the apical pole of principal cells, as well as of enzymes activating the pro-enzymatic form of MMP9. Further studies are required to identify the molecular mechanisms involved in MMP activation by collecting duct cells.
To investigate the effects of SV40 large-T on MMPs, RC.S-VtsA58 cells were either kept at 33°C or shifted to 39.5°C. At the permissive temperature, functional large-T antigen induced a down-regulation of MMP2 and suppressed MMP9 expression. At first glance, these results were surprising because transformed cells are usually thought to exhibit high proteolytic activity responsible for their invasiveness potential. They cannot be accounted for by cell selection or genetic drift because studies were conducted in parallel on cells originating from the same passage maintained at 33°C or shifted to 39.5°C at various passage levels. Neither can they be explained by the change in culture temperature, which did not affect the pattern of gelatinolytic activity in a control cell line transformed with a wild-type strain of SV40 and grown at 33 or 39.5°C. Moreover, reinfection of RC.SVtsA58 cells at the time of cell shifting to 39.5°C with a wild strain of SV40 suppressed within 48 h the increase of MMP2 and the induction of MMP9, thus demonstrating that when large-T is functional, it has a negative effect on MMPs expression irrespective of culture temperature.
Decreased expression of MMP2 by SV40 had previously been reported in human skin fibroblasts (38) and, at the permissive temperature, in human placenta cells transformed with the temperature-sensitive SV40 strain tsA30.1 (39). We further show that this inhibition is transcriptional. Like large-T, adenovirus E1a oncogene represses MMP2 gene transcription in human tumor cell lines (40). The effect of E1a is mediated by the AP-2 transcription pathway (41). Binding of AP-2 to the 5Ј-flanking region of MMP2 gene seems to be essential for gene activation. Consequently the interaction of E1a protein with the DNA binding/dimerization region of AP-2 inhibits MMP2 gene transcription (42). Large-T also prevents AP-2-mediated activation of gene transcription by inhibiting AP-2 binding to DNA (43). However, the mechanisms of the negative transcriptional effect of large-T on MMP2 may be more complex. Indeed in addition to AP-2, a p53 binding site located in the enhancer region containing the AP-2 regulatory sequence of the MMP2 promoter was recently described by Bian and Sun (44), who demonstrated a dual direct and indirect effect of p53 on MMP2 transcription resulting in a net activation of MMP2. In RC.S-VtsA58 cells, cascade immunoprecipitation experiments with anti-large-T and anti-p53 antibodies revealed that p53 was totally bound to large-T at 33°C, whereas it was released from the mutated large-T protein at 39.5°C (45). 2 Thus, increased MMP2 transcription at 39.5°C could be the consequence of both the sudden release of free p53 and the loss of inhibitory effect of the mutated large-T on AP-2 binding to DNA. Conversely, the negative effect on MMP2 transcription of p53 quenching by large-T at 33°C is corroborated by preliminary experiments showing a substantial increase in MMP2 activity in RC.SVtsA58 cells transfected with the full-length rabbit p53 cDNA. 3 In contrast to MMP2, MMP9 was essentially regulated at a posttranscriptional level as shown by Northern blot and run-on experiments. However, we cannot exclude a minor effect of large-T on MMP9 transcription as well because AP-2 was shown to induce cell-type specific transcription of MMP9 in rabbit corneal epithelial cells (46).
Gelatinolytic activity was not detectable at 33°C, and this was associated with a marked decrease in TIMP activity and TIMP1 antigen. Decrease in gelatinolytic activity in transformed cells (33°C) apparently contradict previous literature reporting that transfection with oncogenes induces protease activity, including MMP9 activity (38,47,48), together with increased invasiveness and metastatic potential (review in Ref. 49). But they are in agreement with a growing number of reports showing that in human tumoral tissues, MMPs are secreted by leukocytes, stroma cells, or both, rather than by tumor cells themselves (50).
Similarly to our kidney principal cell line, shifting of the temperature-sensitive human placenta cell line to the restrictive temperature increased production of MMPs in parallel with an induction of invasiveness, a characteristic of differentiated placenta trophoblast-like cells (39). Therefore, we can postulate that high gelatinolytic activity in principal cells cultured at 39.5°C could participate in the expression of a differentiated phenotype.
Because MMP2 and MMP9 are expressed in collecting duct cells in vivo, we used RC.SVtsA58 cells as a model to test the ability of EGF and AVP, two physiological ligands of collecting duct, to influence MMPs production. AVP is the principal hormone regulating water and electrolyte transport in the collecting duct via cAMP production and protein kinase A activation (Ref. 51 and review in Ref. 52). It increases water reabsorption by inducing insertion of AQP2 water channels in the apical membrane of principal cells (53). It also raises sodium reabsorption by apical epithelial sodium channels (54). EGF inhibits the hydroosmotic effects of AVP by acting at a post-cAMP level (30). Conversely, AVP inhibits the EGF-stimulated mitogen-activated protein kinase cascade (55), indicating that the signal transduction pathways of AVP and EGF are closely connected in collecting duct cells. We found that MMP2 was not regulated by either AVP or EGF. In sharp contrast, MMP9 was markedly but differentially modulated by the two ligands: AVP decreased whereas EGF increased MMP9 protein, activity, and mRNA in collecting duct principal cells cultured at 39.5°C. This regulation occurred at a posttranscriptional level. Regulation of MMPs by cAMP is poorly documented. cAMP upregulated TIMP1, TIMP2, and MMP2 expression both at mRNA and protein levels in the human fibrosarcoma cell line HT1080, but MMP9 was not affected (56). Stimulation of MMP9 by EGF has been reported in other cell types (57-62), but the regulation was not posttranscriptional. On the other hand, EGF induced MMP1 and MMP3 in cultured human fibroblasts by increasing mRNA stability (63). Rat, rabbit, and human MMP1 mRNA have a 3Ј-untranslated region that contains three repeats of the AUUUA motif (64) that are implicated in the regulation of mRNA stability (65). Similar sequences might be involved in the regulation of MMP9 mRNA stability by EGF and AVP and during the cell differentiation process induced by temperature shift.
The strong expression of MMP2 and MMP9 at the apical pole of collecting duct cells in vivo, as well as the opposite posttranscriptional regulation of MMP9 by physiological ligands with 2 M. L. Cittanova, unpublished results. 3 F. Le Goas and R. Piedagnel, unpublished results.
antagonistic hydroosmotic effects, suggests that MMP9 plays a role in the physiology of principal cells, irrespective of its collagenase activity. A variety of nonextracellular matrix macromolecules are cleaved by MMP9, including myelin basic protein (1), galactoside-binding proteins CBP30 and CBP35 (66,67), and interleukin-1␤ (2). In the collecting duct, other proteases have been implicated in important physiological functions. It was shown that a glycosylphosphatidylinositol-anchored serine protease named CAP1 regulated apical epithelial sodium channel activity (68) and that a membrane metalloendopeptidase could degrade the V2-type AVP receptor (AVP-R2) and thus regulate its function (69). Considering the apical localization of MMP2 and MMP9, and MMP9 regulation by AVP and EGF, potential substrates include apical ion and water channels, apical receptors and their ligands, and urine constituents. To identify these substrates, it is essential to establish whether MMPs are located in the plasma membrane or in submembranous vesicles as shown for MMP9 in human microvascular endothelial cells (70).
In conclusion, we have demonstrated that MMP2 and MMP9 are produced by cultured collecting duct cells and are specific markers of this nephron segment in vivo. Large-T dramatically reduces MMP2 and MMP9 expression at transcriptional and posttranscriptional levels, respectively. Only MMP9 is regulated by AVP and EGF, which induce opposite posttranscriptional effects. Further studies are needed to determine the renal physiological consequences of MMP9 and MMP2 gene invalidation and the physiological substrates of these enzymes in principal cells.