Processing of CD44H by MT-MMPs
To examine whether MT-MMPs can shed CD44H, CD44H was coexpressed with different MT-MMPs in human breast carcinoma ZR-75-1 cells that express undetectable levels of both endogenous CD44H and MT1-MMP. Expressed CD44H was detected as a 95-kD protein ( A, Cell, lane 2) and did not show soluble fragment sCD44H in the medium (Med, lane 2). On the other hand, coexpression of MT1-MMP or MT3-MMP resulted in shedding of a 70-kD sCD44H into the media (lanes 3 and 5, respectively), whereas MT2, MT4, and MT5-MMP did not (lanes 4, 6, and 7, respectively). To ensure that the lack of CD44H processing by MT2, MT4, and MT5-MMP is not the result of inefficient delivery of the enzymes to the cell surface, immunoreactivity of FLAG-tagged MT-MMPs on the surface was examined. Relative intensities of cell surface signals were as follows: MT1-MMP (1.0); MT2-MMP (0.32); MT3-MMP (0.36); MT4-MMP (1.08); MT5-MMP (0.29); and that of mock-transfected cells was negligible. Thus, the amount of MT2, MT4, and MT5-MMP on the cell surface is almost comparable to that of MT3-MMP that can cleave CD44H. The cells also showed gelatin-degrading activity upon expression of MT-MMPs in a BB94-sensitive manner (synthetic hydroxamate MMP inhibitor). Relative gelatin-degrading activities by the cells were as follows: MT1-MMP (+3), MT2-MMP (+1), MT3-MMP (+2), MT4-MMP (+1), MT5-MMP (+1).
Figure 1 Shedding of CD44H by MT-MMPs. (A) CD44H was coexpressed with each of the MT-MMPs, as indicated by transient transfection of the expression plasmids into ZR-75-1 cells, and incubated in the serum-free media. After 48 h, cell lysates and medium fractions (more ...)
The shedding by MT1-MMP was inhibited by TIMP-2 and BB94, but not by TIMP-1 or a serine proteinase inhibitor, AEBSF ( B). TIMP-2 but not TIMP-1 is known to inhibit MT1-MMP, whereas all soluble MMPs including MMP-2 and MMP-13 can be inhibited by both TIMPs (Nagase and Woessner 1999
; Seiki 1999
). Also, endogenous MMP-2 was not detected in the culture supernatant of ZR-75-1 by zymography (data not shown). Thus, CD44H is thought to be processed directly by MT1-MMP rather than by some other soluble MMPs activated by MT1-MMP. Similar results were obtained with MT3-MMP.
Upon coexpression of either MT1-MMP or MT3-MMP with CD44H, CD44H with a lower molecular mass (80 kD) was detected in the cell fraction in addition to the 95-kD CD44H ( A). To examine the integrity of the NH2- and COOH-terminal ends of the molecule, we constructed CD44H, tagged with c-Myc epitope at the NH2 terminus and FLAG tag at the COOH terminus, and subjected it to Western Blotting using specific antibodies against these tags. Consequently, the 80-kD CD44H in the cell lysate appeared to retain both tags ( C), suggesting that it is not a processing product of 95-kD CD44H and the polypeptide core is intact. Thus, posttranslational modifications of CD44H, most likely glycosylations, might be affected by the coexpression of these MT-MMPs, although the reason is not clear. Also, the 70-kD sCD44H detected in culture medium was confirmed to be the NH2-terminal part of CD44HMF retaining the NH2-terminal Myc tag but not the COOH-terminal FLAG tag ( C).
Effect of CD44H Processing on Ligand-binding Capacity and Morphology of the Cells
HA is the major ligand for CD44 on the cell surface. Therefore, we examined the effect of processing of CD44H by MT1-MMP on the ability of the cells to bind HA. Transfected ZR-75-1 cells were incubated with increasing amount of FITC-HA, and amount of the bound HA was quantified by fluorescence scanning using a confocal laser microscope ( A). The expression of CD44H greatly increased the binding of FITC-HA compared with the mock-transfected cells. The binding was saturable, and an addition of 100-fold excess nonlabeled HA competed the binding and decreased it to the level of nontransfected cells (Mock). Coexpression of MT1-MMP with CD44H reduced the HA binding activity significantly by 40% ( B, compare CD44H+MT1 with CD44H). On the other hand, when the transfected cells were cultured in the presence of BB94 to inhibit the shedding of CD44H, the amount of bound HA was significantly increased compared with samples without BB94 ( B, compare CD44H+MT1/BB94 with CD44H+MT1). Thus, expression of MT1-MMP surely downregulates the net HA-binding activity as a result of CD44H processing.
Figure 2 Effect of CD44 shedding on HA-binding activity and cell morphology. (A) ZR-75-1 cells were transfected with expression plasmid for CD44. After 24 h, the cells were incubated with increasing concentrations of FITC-labeled HA for 60 min at 37°C. (more ...)
Expression of CD44H in ZR-75-1 significantly altered cell morphology, possibly by changing the adherent nature of the cells. As shown in C, expression of CD44H resulted in the formation of numerous small and large protrusions (arrowheads) at the adherent edge of the cells compared with the mock-transfected cells ( K). When MT1-MMP was coexpressed with CD44H, on the other hand, these protrusions were not formed ( and ). The expression of MT1-MMP alone did not change cell shape ( H) compared with the mock-transfected cell ( K). The effect of MT1-MMP on the CD44H-expressing cells ( and ) appeared to be the result of CD44H processing by MT1-MMP, as many protrusions were formed when the cells were cultivated in the presence of BB94 ( and ).
Effect of CD44H and MT1-MMP on Cell Migration
Next, we examined the effect of CD44H and MT1-MMP on the migration of the cells. As shown in and , the expression of either CD44H or MT1-MMP alone had no effect on the cell migration. However, coexpression of CD44H and MT1-MMP significantly increased motility of the cells. Addition of BB94 to these transfectants, which inhibits the processing of CD44H by MT1-MMP ( C), suppressed the increased cell migration. A similar result was also obtained with the MG-63 osteosarcoma cell line, which expresses low levels of CD44H ( and ).
Figure 3 Effect of the shedding of CD44H by MT1-MMP on the cell motility. (A) ZR-75-1 cells were transfected with the expression plasmids for CD44H and/or MT1-MMP together with the one for GFP. The motility of GFP-positive cells was analyzed by phagokinetic track (more ...)
Since the expression plasmids were introduced into the cells by transient transfection together with green fluorescent protein (GFP) plasmid as a transfection marker, we could compare motility of both transfected and nontransfected cells in the same field. Stimulation of cell migration was observed only with the transfected cells expressing CD44H and MT1-MMP, but not with the surrounding nontransfected cells, indicating that the shed CD44H fragment itself does not have activity to stimulate cell motility ( D). We also examined the cell motility using coloidal gold glass coverslips coated with HA instead of serum-coated coverslips, but there are no differences in the data.
When cell surface localization of both CD44H and MT1-MMP were analyzed by confocal microscope, both molecules appeared to distribute over the cell surface ( E, Integrated Image). However, colocalization was prominent at the adherent edge of the transfected cells ( E, Adherent Layer).
Determination of the Cleavage Site of CD44H by MT1-MMP
According to the molecular size of sCD44H, the cleavage site of CD44H by MT1-MMP was expected to locate within the stem region of CD44H (Thr130-Gln265) that is between the globular and transmembrane domain ( A). Therefore, we expressed the fragment corresponding to that region (rCD44HS) in E. coli and incubated it with recombinant MT1-MMP catalytic domain (rMT1-CAT) to determine the putative cleavage site. Incubation of rCD44HS with rMT1-CAT generated 28- and 26-kD bands in a time-dependent manner on SDS-PAGE (data not shown). To isolate these fragments, the reaction mixture was subjected to reverse phase chromatography, and four fragment peaks were recovered ( B). These peak fractions were collected and subjected to NH2-terminal amino acid sequence analyses. The NH2-terminal sequence of peak 2 corresponded to that of the original fragment, and those of peaks 1, 3, and 4 were mapped to 163Thr (TNPED), 193Tyr (YIFYT), and 187Ser (SSTSG), respectively ( C). From their molecular sizes and height of the peaks, cleavage between the Gly192–Tyr bond (corresponding to peak 3) and Arg186–Ser bond (corresponding to peak 4) are likely to generate 26- and 28-kD fragments, respectively. A mutant CD44H fragment that lacks the 40 internal amino acids, which include all the three cleavage sites (158Lys to 197Thr), was constructed, and it was confirmed to be resistant to cleavage by rMT1-CAT (data not shown).
Figure 4 Processing of CD44H by MT1-MMP in vitro. (A) Schematic illustration of CD44H, the stem fragment (rCD44HS) expressed in E. coli, and the mutant lacking the processing sites by MT1-MMP (CD44HM). Fragments expressed in E. coli were tagged with FLAG at the (more ...)
CD44H Processing Is Critical for Cell Migration
Although BB94 inhibits cell migration driven by CD44H and MT1-MMP, it does not necessarily mean that the processing of CD44H is the absolute requirement for cell migration. To examine this, we constructed the mutant CD44H that can not be processed by MT1-MMP. Deletion of the 40 internal amino acids (158Lys–197Thr) was introduced into CD44H ( A, CD44HM), and it was expressed in ZR-75-1 cells. CD44HM was not shed by MT1-MMP under the condition where the wild-type CD44H was shed into the media ( A). To confirm that the mutant CD44HM retains comparable ligand-binding ability to the wild-type CD44H, binding of FITC-HA to the cells was analyzed. The cells expressing CD44HM showed comparable HA-binding activity ( D) and similar morphological change of the cells to that of the wild-type CD44H-expressing cells ( and ). On the other hand, coexpression of MT1-MMP with CD44HM did not promote migration of the cells ( E, CD44HM+MT1). Furthermore, CD44HM inhibited migration of the cells promoted by CD44H and MT1-MMP ( E, compare CD44H+ CD44HM+MT1 with CD44H+MT1), suggesting a dominant-negative effect against the wild-type CD44H. These data indicate that processing of CD44H is an essential step for CD44H and MT1-MMP–promoted cell migration.
Figure 5 Dominant-negative effect of CD44HM on cell migration stimulated by CD44H and MT1-MMP. (A) Either wild-type CD44H or the mutant CD44HM was expressed in ZR-75-1 cells together with MT1-MMP. Cell lysate and medium fractions were subjected to Western blot (more ...)
Proteinases Responsible for CD44H Shedding in a Human Pancreatic Tumor Cell Line
A human pancreatic tumor cell line, MIA PaCa-2, expresses high levels of CD44, which is spontaneously shed into the culture medium ( A). The major form of CD44 expressed in MIA PaCa-2 cells was confirmed to be CD44H by RT-PCR using a set of specific primers that can amplify all the splicing variants ( B).
Figure 6 Shedding of endogenous CD44H in human pancreatic tumor cell line, MIA PaCa-2. (A) MIA PaCa-2 cells (3 × 105) were cultured in a six-well plate in serum-free medium in the presence or absence of the proteinase inhibitors as indicated. After 48 (more ...)
Using an anti-CD44 mAb, 95-kD CD44H was detected in the cell lysate as a major form ( A, lane 1). The cells shed 90- and 70-kD sCD44H spontaneously into the medium ( A, lane 1). To examine the types of proteinases that are responsible for the shedding, we tested proteinase inhibitors selective for metalloproteinase (BB94), serine proteinases (AEBSF), and cystein proteinases (E-64) ( A). BB94 completely inhibited the shedding of the 70-kD fragment, selectively ( A, lane 5). In contrast, AEBSF inhibited the shedding of the 90-kD but not the 70-kD fragment ( A, lane 7). The combination of BB94 and AEBSF inhibited the shedding of the both fragments ( A, lane 8). Thus, the 70-kD fragment is processed by metalloproteinases, and the 90-kD fragment, by serine proteinases. Since inhibition of serine poroteinase did not inhibit the shedding of the 70-kD form, the shedding event by these different types of proteinase occurred independently rather than in a sequential manner. The shedding of the 70-kD fragment was inhibited by TIMP-2 but not by TIMP-1 ( A, lanes 2 and 3) and the size of the fragment is similar to that of the one processed by MT1-MMP upon coexpression with CD44H. Such different sensitivity to TIMP-1 and TIMP-2 is characteristic of MT-MMPs (Nagase and Woessner 1999
; Seiki 1999
). Thus, the metalloproteinase responsible for the shedding of the 70-kD fragment in MIA PaCa-2 cells is likely to be either MT1-MMP or MT3-MMP as they have the ability to process CD44H (). Transcripts for MT-MMPs were analyzed by RT-PCR, and MIA Paca-2 cells were found to express MT1-MMP but not MT3-MMP ( B). Expression of MT1-MMP was also confirmed by Western blotting (data not shown).
Processing Site–deleted Mutant, CD44HM, Inhibits Migration of the Pancreatic Tumor Cells
MIA PaCa-2 cells showed spontaneous motile activity on the slide glass coated with coloidal gold (see below). Thus, we asked whether this motility is attributed to the shedding of endogenous CD44H by MT1-MMP ( and ). BB94, which inhibits shedding of 70-kD sCD44H, suppressed the motility by 65%, whereas the serine proteinase inhibitor (trypsin inhibitor) had no effect. In addition, expression of CD44HM inhibited the motility by 40%, whereas expression of full-length CD44H had no effect. Since CD44HM plasmid was transfected transiently, both transfected and nontransfected cells were found in the same field (transfected cells with arrow). In spite of the spontaneous shedding of 70 kD sCD44 from surrounding cells, inhibition of the motility was observed with the cells expressing CD44HM ( B, CD44HM). This suggests that processing event but not its product (70-kD sCD44) is important to promote migration.
Figure 7 BB94 and CD44HM inhibits motile activity of MIA PaCa-2. (A) MIA PaCa-2 cells were transfected with the expression plasmid for the FLAG-tagged CD44H or CD44HM together with the one for GFP. After 48 h, cells were subjected to the migration assay as described (more ...)
CD44HM is resistant to MT1-MMP–dependent processing but still susceptible to the serine proteinase, as it was shed as a 78-kD fragment that was inhibited by AEBSF (data not shown) or trypsin inhibitor ( C, lane 4). Since CD44HM inhibited the migration of the cells, this further strengthened the idea that MT1-MMP–dependent shedding of CD44H promotes cell migration, whereas the shedding by a serine proteinase has no effect on the motile phenotype of the cells.