(Received for publication, August 13, 1996, and in revised form, October 16, 1996)
From the Department of Anatomy and Cellular Biology,
Tufts University School of Medicine, Boston, Massachusetts 02111 and
the § Departments of Research and Medicine, Veterans Affairs
Medical Center, Northport, New York 11768
Many of the tumor-associated matrix metalloproteinases that are implicated in metastasis are produced by stromal fibroblasts within or surrounding the tumor in response to stimulation by factors produced by tumor cells. In this study we transfected Chinese hamster ovary cells with putative cDNA for human extracellular matrix metalloproteinase inducer (EMMPRIN), a transmembrane glycoprotein that is attached to the surface of many types of malignant human tumor cells and that has previously been implicated in stimulation of matrix metalloproteinase production in fibroblasts. We show that these transfected cells synthesize EMMPRIN that is extensively post-translationally processed; this recombinant EMMPRIN stimulates human fibroblast production of interstitial collagenase, stromelysin-1, and gelatinase A (72-kDa type IV collagenase). We propose that EMMPRIN regulates matrix metalloproteinase production during tumor invasion and other processes involving tissue remodeling.
Successful tumor metastasis requires many steps, one of which is local proteolytic destruction of extracellular matrix at sites of tumor invasion. A major class of proteinases associated with tumor invasion is the matrix metalloproteinases (MMPs)1 (1, 2). Although it was initially thought that these enzymes were mainly produced by malignant tumor cells themselves, it is now clear that interstitial collagenase (MMP-1), gelatinase A (MMP-2, a 72-kDa type IV collagenase), and stromelysin-1 (MMP-3) are produced in vivo by stromal fibroblasts associated with several types of tumors (2, 3, 4, 5, 6, 7). MMP-2 synthesized and secreted by these fibroblasts has been shown to adhere to the surface of tumor cells, facilitating tissue invasion (8, 9, 10). Because quiescent fibroblasts generally produce relatively low amounts of MMPs (11, 12), tumor-associated fibroblasts must be influenced in some way to give rise to the elevated levels of MMPs usually present in malignant tumors. One possibility that we have investigated is that tumor cells interact with fibroblasts via soluble or cell-bound factors, stimulating fibroblast MMP production (11, 12, 13, 14, 15). Our studies have led to characterization of a tumor cell surface protein, extracellular matrix metalloproteinase inducer (EMMPRIN; previously termed tumor cell-derived collagenase stimulatory factor or TCSF), that stimulates fibroblast production of MMP-1, MMP-2, and MMP-3 (12, 13, 14). We recently obtained cDNAs for human EMMPRIN and verified their identity by recognition of recombinant EMMPRIN by activity-blocking monoclonal antibody and by sequence identity with amino acid sequences of peptides isolated from EMMPRIN (15). However, recombinant EMMPRIN produced by bacteria is much smaller than native EMMPRIN isolated from tumor cells because it is not post-translationally processed. This form of recombinant EMMPRIN is inactive, thus leaving some doubt regarding the identity of the cDNAs. In this study, we use the cDNAs to transfect CHO cells and show that EMMPRIN produced by these transfected cells is post-translationally processed and stimulates fibroblasts to produce elevated levels of MMP-1, MMP-2, and MMP-3.
EMMPRIN cDNA (15) was subcloned into an expression vector, pcDNA/Neo (Invitrogen, San Diego, CA), and purified by CsCl gradient centrifugation and phenol/chloroform extraction. CHO cells (American Type Culture Collection, Bethesda, MD) were seeded at 106 cells/100-mm tissue culture dish and incubated overnight, at which stage they were 50-70% confluent. The cells then were transfected in 5 ml of serum-free Ham's F-12 medium containing lipofectamine-DNA complex (10 µl of lipofectamine (Life Technologies, Inc.) mixed with 10 µg of DNA with or without the EMMPRIN insert). After 6 h of incubation at 37 °C, 5 ml of medium containing 20% fetal bovine serum was added to the transfection mixture, which was then cultured at 37 °C for a further 72 h. The cells then were treated with trypsin-EDTA (Life Technologies, Inc.) and subcultured in medium containing 400 mg/liter of Geneticin (Life Technologies, Inc.) for 2-3 weeks. Successful transfection was assessed by immunocytochemistry using monoclonal antibody E11F4 raised against EMMPRIN, as described previously (13).
Purification of EMMPRINEMMPRIN was purified from detergent extracts of cell membranes from LX-1 cells or stably transfected CHO cells by immunoaffinity chromatography using monoclonal antibody E11F4 against EMMPRIN as described previously (13). Briefly, the cell membranes were extracted with 10 mM Tris-HCl buffer (pH 8.2), containing 0.5% Nonidet P-40, 2 mM phenylmethylsulfonyl fluoride, and 1 mM EDTA. The supernatant of the extract was then applied to a 5-ml anti-EMMPRIN affinity column and recirculated through the column for 12 h at 4 °C. The column was washed with buffer several times, and EMMPRIN was then eluted from the column with 50 mM diethylamine, 30 mM octylglucoside (pH 11.5). The eluted protein was neutralized with 0.5 M NaH2PO4, dialyzed against 0.1 M acetic acid, concentrated, and dissolved in 0.1 M acetic acid.
Assays for EMMPRIN ActivityHuman fibroblasts (isolated from human skin in our laboratory) were cultured for 24 h in 24-well plates in 1 ml of DMEM medium supplemented with 10% fetal bovine serum, after which the medium was replaced with 0.5 ml of DMEM containing 2% fetal bovine serum in the presence or the absence of EMMPRIN or TPA, and the cultures were further incubated at 37 °C for 3 days. Media from these cultures were used for zymographic assay of MMP-3 (16) and ELISA of MMP-1, MMP-2, and MMP-3 (12, 17).
In initial attempts to demonstrate recombinant EMMPRIN activity we tested purified, pGEX bacterial expression protein. However, EMMPRIN produced in the pGEX system had a molecular mass of only ~29 kDa (equivalent to that expected from the cDNA open reading frame (15)), compared with native EMMPRIN from tumor cells, which is ~58 kDa (12, 13, 14). This bacterially produced recombinant EMMPRIN protein was inactive in stimulating MMP production by human fibroblasts. Next, COS and CHO cells were transfected with EMMPRIN cDNA under a variety of conditions, but in most cases the EMMPRIN produced was either of similar molecular mass to bacterial recombinant protein, i.e. ~29 kDa, or was partially post-translationally processed with molecular masses ranging from 30-45 kDa. EMMPRIN isolated from these cells was also inactive.
However we found that after stable transfection (see "Experimental
Procedures"), CHO cells could be selected that synthesize high levels
of EMMPRIN of similar molecular mass to that of native EMMPRIN isolated
directly from LX-1 human carcinoma cells. Fig. 1A shows detection of this recombinant
EMMPRIN in the transfected CHO cells by immunocytochemistry using
monoclonal antibody raised against native EMMPRIN from LX-1 cells (13).
Because the transfected cells are very flat, it is difficult to discern
the precise cellular distribution of EMMPRIN. However, if the cells are
fixed shortly after plating (~4 h), i.e. before they have
flattened, it is clear that EMMPRIN is located at the surface of the
transfected cells (Fig. 1B). Untransfected cells or cells
that are mock-transfected with vector only show no reactivity with the
antibody (Fig. 1C).
Fig. 2A shows a silver-stained SDS-PAGE gel
of recombinant EMMPRIN purified by immunoaffinity chromatography from
membrane extracts of the stably transfected CHO cells; this gel was
deliberately overloaded to reveal potential contaminants in the
preparation. A single broad band at ~58 kDa was detected, as
previously obtained for tumor cell-derived EMMPRIN (13, 14). Direct
comparison of purified recombinant and LX-1 carcinoma cell-derived
EMMPRIN by Western blotting showed that they were identical in size
(Fig. 2B), indicating that the recombinant EMMPRIN was fully
or almost fully post-translationally processed. Untransfected CHO cells or cells transfected with vector only did not produce any EMMPRIN detectable by immunoaffinity chromatography and Western blotting (not
shown).
We then tested purified recombinant EMMPRIN from transfected CHO cells
for its ability to stimulate MMP production by human fibroblasts in
culture. We first measured the effect of recombinant EMMPRIN on MMP-3
production by zymography, using two separate substrates, casein (Fig.
3A) and carboxymethylated transferrin (Fig. 3B). A clear-cut increase in active MMP-3 was observed
in fibroblasts treated with the recombinant EMMPRIN (Fig. 3,
A and B, lane 1 versus 2). The amount
of MMP-3 was similar to that induced by TPA treatment (Fig. 3,
A and B, lane 3).
To ensure that stimulation of MMP-3 production was not due to minor
contaminants in the recombinant EMMPRIN preparation, we tested the
effect of blocking antibody raised against native EMMPRIN (13) on the
stimulation by recombinant EMMPRIN, using two different approaches. In
the first approach, antibody was included in the culture medium
together with EMMPRIN throughout the 3-day incubation period (Fig.
4, lane 7). In the second approach, the
antibody was mixed with EMMPRIN and then removed by binding to protein A; the supernatant from this reaction, depleted of antigen, was then
added to the culture for the 3-day incubation (Fig. 4, lane 6). As shown in Fig. 4, the stimulation of stromelysin production in cells treated with EMMPRIN (lanes 4 and 5 versus
lanes 3 and 8) was completely reversed by either of the
two different treatments with antibody to EMMPRIN (lanes 6 and 7).
Finally, we measured the effect of recombinant EMMPRIN on MMP-1, MMP-2, and MMP-3 production by ELISA. In two separate experiments, treatment of fibroblasts with the EMMPRIN gave rise to significant increments in production of MMP-1 (~6- and ~11-fold), MMP-2 (~1.5- and ~16-fold), and MMP-3 (~2- and ~4-fold) (Table I). In most cases the degree of stimulation of MMP by recombinant EMMPRIN was similar to that caused by TPA (Table I).
|
The results presented here constitute final proof that the cDNAs that we have obtained genuinely encode EMMPRIN, as defined by its MMP stimulatory activity. The degree of stimulation of MMP-1, MMP-2, and MMP-3 by recombinant EMMPRIN obtained herein is similar to that obtained previously with native EMMPRIN purified from LX-1 carcinoma cells (12). It should be pointed out, however, that the degree of stimulation by either native or recombinant EMMPRIN varies in different fibroblast preparations. First, some fibroblast preparations are not significantly responsive to EMMPRIN stimulation, whereas others are very responsive (e.g. see Ref. 12). Second, among populations of fibroblasts that are responsive to EMMPRIN, an important variable is the extent to which a batch of fibroblasts already produces a given MMP without addition of stimulatory agents. This can be readily appreciated for MMP-3 production in Fig. 3B, lane 2, versus Fig. 4, lanes 3 and 8, and for MMP-2 production in Table I, experiment 1 versus experiment 2. Our overall experience has been that EMMPRIN stimulation of MMP-1 and MMP-3 production is usually in the range of 3-10-fold. Stimulation of MMP-2, however, is even more variable (e.g. see Table I, experiment 1 versus 2) but is usually rather modest, e.g. 1.5-2-fold. One reason for this variability is clearly the significant amounts of MMP-2 that many fibroblast preparations produce without treatment with exogenous agents, e.g. experiment 1 in Table I. Also, however, the mechanism of stimulation of MMP-2 may be more complex than for MMP-1 and MMP-3. As well as stimulating overall MMP-2 production, EMMPRIN apparently enhances activation of MMP-2, and this sometimes leads to underestimation of MMP-2 stimulation in ELISA assays (12).
The experiments described here indicate that EMMPRIN activity is dependent on post-translational processing. However, it is not yet clear whether processing is required to attain the appropriate conformation for activity or whether specific side groups, e.g. carbohydrate or phosphate, are involved in EMMPRIN receptor recognition.
We have shown by immunocytochemistry that EMMPRIN is present on the surface of tumor cells but not fibroblasts and several other normal adult cell types (13, 18, 19). Prior studies have also shown that tumor cells shed EMMPRIN (13) and that EMMPRIN appears in the urine of bladder carcinoma patients (18, 20). Thus tumor cell EMMPRIN, in soluble or membrane-bound form, is likely to be responsible for at least part of the stimulation of fibroblast MMP production observed in vivo in association with a variety of malignant tumors (3, 4, 5, 6, 7). In preliminary experiments, we have found that production of the tissue inhibitor of matrix metalloproteinases, TIMP-1, is not stimulated by EMMPRIN; if this proves to be true in vivo, an imbalance of active versus inactive MMP production may result from EMMPRIN action on stromal fibroblasts.
The amino acid sequence of EMMPRIN is identical to that of human M6 antigen (21) and basigin (22), for which no function had previously been ascribed until EMMPRIN was cloned (15). Several proteins with high levels of homology to EMMPRIN, i.e. neurothelin, HT7, OX47, and gp42, have also been characterized in other species (21, 22, 23), but again their function has not been determined. These proteins may be species homologues of one another or closely related members of a protein family; it remains to be seen whether all of these proteins have EMMPRIN activity.
The ability of EMMPRIN to stimulate MMP production has led us to propose that it may be involved in a wide range of physiological and pathological processes where tissue remodeling takes place (24). For example, its presence in the epidermis (25) and in several embryonic epithelia2 suggests that EMMPRIN may participate in epithelial-mesenchymal interactions leading to changes in tissue architecture during development and wound healing. Also, EMMPRIN on the surface of activated lymphocytes and monocytes (21) may contribute to elevated MMP levels found in arthritis. However, association of EMMPRIN-like material with endothelium during formation of the blood-brain barrier and with highly organized epithelia such as retina and kidney tubules (23) suggests that it may have additional functions involving cell-cell interactions.
We thank Diane Silva and Michelle Drews for technical help.