Expression of tetraspanins in human lung cancer cells: frequent ...

Oncogene (2003) 22, 674–687

& 2003 Nature Publishing Group All rights reserved 0950-9232/03 $25.00 www.nature.com/onc

Expression of tetraspanins in human lung cancer cells: frequent downregulation of CD9 and its contribution to cell motility in small cell lung cancer Toshiki Funakoshi1, Isao Tachibana*,1, Yoshihiko Hoshida2, Hiromi Kimura1, Yoshito Takeda1, Takashi Kijima3, Kazumi Nishino1, Hiroyuki Goto1, Tsutomu Yoneda1, Toru Kumagai1, Tadashi Osaki1, Seiji Hayashi1, Katsuyuki Aozasa2 and Ichiro Kawase1 1

Department of Molecular Medicine, Osaka University Graduate School of Medicine, 2-2 Yamada-oka, Sutia, Osaka 565-0871, Japan; 2Department of Pathology, Osaka University Graduate School of Medicine, 2-2 Yamada-oka, Sutia, Osaka 565-0871, Japan; 3 Division of Thoracic and Adult Oncology, Dana-Farber Cancer Institute, Harvard Medical School, 44 Binney Street, Boston, MA 02115, USA

Small cell lung cancer (SCLC) invades locally and metastasizes distantly extremely early when compared with nonsmall cell lung cancer (NSCLC). The underlying molecular mechanisms, however, have not been elucidated. Accumulating evidence suggests that downregulation of several members of tetraspanins is associated with progression of solid tumors, thus indicating poor prognosis. Here we screened 30 lung cancer cell lines for expression of tetraspanins, CD9, CD63, CD81, CD82, CD151, and NAG-2. Flow cytometry revealed that, among these proteins, CD9 is broadly expressed in NSCLC lines, but is absent or highly reduced in most SCLC lines (Po0.0001). Using the Boyden chamber and videomicroscopic cell motility assays, we showed that stable transfection of CD9 into an SCLC line, OS3-R5, reduced cell motility on fibronectin. Furthermore, by transient transfection of green fluorescent protein (GFP)tagged CD9 into three other SCLC lines, we observed that SCLC cells expressing GFP-CD9 were uniformly less motile than untransfected cells. CD9 or GFP-CD9 was associated with b1 integrins and distributed at the tumor cell periphery and cell–cell contacts, suggesting that CD9 modifies b1 integrin function to reduce motility. These findings suggest that low expression of CD9 may contribute to the highly invasive and metastatic phenotype of SCLC. Oncogene (2003) 22, 674–687. doi:10.1038/sj.onc.1206106 Keywords: small cell lung cancer (SCLC); tetraspanin; CD9; motility

Introduction Lung cancer has become the most common fatal tumor in the developed countries. It is classified into small cell *Correspondence: I Tachibana; E-mail: [email protected] Received 18 October 2001; revised 30 September 2002; accepted 4 October 2002

lung cancer (SCLC) and nonsmall cell lung cancer (NSCLC). SCLC is characterized by several neuroendocrine features, as evidenced by the presence of dense core granules, high enzymatic activities of L-dopa decarboxylase, production of hormones and neuropeptides, and expression of neural cell adhesion molecule (N-CAM) (Carney et al., 1985). Clinically, SCLC is distinct from NSCLC in that most patients are inoperable at diagnosis, because the tumor locally invades and distantly metastasizes extremely early (Mountain, 1978). The underlying molecular mechanisms of this early progression, however, are not clarified. Tetraspanin, also called tetraspan or transmembrane4 superfamily (TM4SF), comprises more than 20 cellsurface proteins including CD9, CD37, CD53, CD63, CD81, CD82, and CD151. All of these molecules share a characteristic structure that spans the membrane four times thereby forming two extracellular loops. Although the precise biological functions of tetraspanins remain elusive, experiments using monoclonal antibodies, cDNAs, and knockout mice have suggested its involvement in cell proliferation, activation, motility, and fusion (Hemler et al., 1996; Maecker et al., 1997; Tachibana and Hemler, 1999; Le Naour et al., 2000; Miyado et al., 2000). Based on coimmunoprecipitation and immunofluorescence results, tetraspanins are believed to function as molecular facilitators. This family of proteins connects other surface molecules including CD4, CD8, CD19, CD21, major histocompatibility complex (MHC) class I and II, and integrins as well as cytoplasmic signaling molecules such as phosphatidylinositol 4-kinase and protein kinase C into a tetraspan web (Rubinstein et al., 1996; Maecker et al., 1997; Hemler, 1998). Among tetraspanins, CD9, CD63, and CD82 were considered to be metastasis-inhibitory factors. Transfection of these molecules into certain tumor cells inhibited in vitro cell motility and in vivo tumor metastasis (Hemler et al., 1996). In addition, clinical and pathological findings suggest that downregulation of CD9, CD63, or CD82 is associated with the progression of

Tetraspanins in human lung cancer cells T Funakoshi et al

675

solid tumors. Reduced expression of these molecules is more frequently observed in metastatic tumors than primary sites, and patients with tumors lacking these molecules tend to be at advanced stages and thus show poor survival. Downregulation in CD9 or CD82 is associated with the progression of differentiated carcinomas such as breast (Miyake et al., 1995, 1996; Huang et al., 1998; Yang et al., 2000), pancreas (Guo et al., 1996; Sho et al., 1998), colon (Cajot et al., 1997; Lombardi et al., 1999), esophagus (Uchida et al., 1999), and NSCLC (Higashiyama et al., 1995; Adachi et al., 1996, 1998), whereas CD63 reduction associated with that of melanoma (Atkinson et al., 1985; Hotta et al., 1988). Intriguingly, CD63 expression had no correlation with the spread of NSCLC and pancreatic cancer (Adachi et al., 1998; Sho et al., 1998). Recently, the loss of another tetraspanin, CD81, was reported to associate with differentiation and metastasis of hepatocellular carcinomas (Inoue et al., 2001). Accumulating clinical and pathological findings have indicated coincidence of the downregulation of tetraspanins with the progression of differentiated carcinomas. However, in vitro studies that account for these findings still remain to be done. Moreover, little information on tetraspanins has been obtained from less differentiated yet more malignant cases such as SCLC. In the present study, we examined the expressions of six tetraspanins, which are ubiquitously expressed in most organs including the lung, in 30 lung cancer cell lines. Among these tetraspanins, CD9 showed highly biased expression between SCLC and NSCLC. Most SCLC lines expressed much lower levels of CD9 than NSCLC lines. We further showed that CD9 expression into SCLC lines suppressed cell motility on fibronectin (FN).

Results Expression of tetraspanins in lung cancer cell lines Since little information has been reported on tetraspanin expression of lung cancer lines, we first examined the expressions of CD9, CD63, CD81, CD82, CD151, and NAG-2 in SCLC lines and compared with those in NSCLC lines. Since a portion of SCLC lines might have lost neuroendocrine features and acquired NSCLC-like phenotypes during culture (Mabry et al., 1991), we excluded such variant SCLC lines from this study by testing for the expression of N-CAM (see Materials and methods), which is known as a cluster 1 SCLC antigen and is a differential marker between SCLC and NSCLC (Souhami et al., 1987). Consequently, all SCLC lines and NSCLC lines tested in this study (Table 1) are NCAM-positive and N-CAM-negative, respectively. Flow cytometry of representative cell lines are shown in Figure 1. Three NSCLC lines, A549, EBC-1, and Lu99, expressed CD9 and CD63 but not N-CAM. On the other hand, three SCLC lines, OS3-R5, OS2-RA, and CADO LC6, expressed N-CAM and CD63 but not CD9. Summary of the expressions of six tetraspanins in

30 lung cancer lines is shown in Table 1. CD63, CD81, CD82, and CD151 were broadly expressed in both SCLC and NSCLC lines. NAG-2 was clearly positive in several NSCLC lines, while it was absent or only marginally stained in most SCLC lines (Po0.01). Remarkably, whereas all NSCLC lines but VMRCLCD were CD9-positive (12 of 13), only a minor population of SCLC lines expressed CD9 (three of 17) (Po0.0001). We also stained these lung cancer lines with anti-MHC class I monoclonal antibody (MoAb). Consistent with a previous report (Doyle et al., 1985), expression of this molecule was less frequent in SCLC than in NSCLC (Po0.05). All CD9-positive SCLC lines also expressed class I molecules. To clarify the mechanisms for CD9 downregulation, eight CD9-negative SCLC lines were randomly selected and CD9 gene expression was evaluated by reverse transcription–polymerase chain reaction (RT–PCR). As shown in Figure 2, SCLC lines except OS2-RA displayed absent or reduced levels of CD9 in contrast to abundant expression in NSCLC lines. OS2-RA appeared to have a moderate level of full-length CD9 mRNA despite no reactivity against anti-CD9 MoAb (Figure 1 and Table 1). We also examined CD63 gene expression and observed no difference between SCLC and NSCLC. Together, among the six tetraspanins, CD9 showed the highly biased expression between SCLC and NSCLC, and the CD9 reduction in SCLC lines primarily occurs at or prior to the transcription level. A possibility, however, could not be excluded that the reduced levels of CD9 may be because of the instability of CD9 mRNA in SCLC lines. Ectopic expression of CD9 into OS3-R5 reduces cell motility on FN To investigate whether ectopic expression of CD9 alters SCLC cell behavior, we selected OS3-R5 and established OS3-R5-pZCD9, a CD9-overexpressing stable transfectant of OS3-R5. Successful CD9 introduction was confirmed by flow cytometry (Figure 3a) and immunoblotting (Figure 4a, left panel). Expressions of another tetraspanin, CD81, and b1 integrin of OS3-R5-pZCD9 were comparable to that of a mock-transfectant (OS3R5-pZSV) or the parent (Figure 3a). We observed no significant difference in the in vitro proliferation rates among these three lines (data not shown). In addition, these three lines displayed similar strong adherence to FN, but not to bovine serum albumin (BSA) or laminin (LM) (Figure 3b). Next, we investigated whether CD9 expression affects cell motility of OS3-R5 on FN. In a conventional Boyden chamber assay, a substantial number of parent cells and mock-transfectants transmigrated through the FN-coated membrane into lower chambers (Figure 3c). This motility is b1 integrindependent because it was almost completely abolished by anti-b1 integrin MoAb but not affected by control IgG or anti-MHC class I MoAb, and few cells transmigrated through the uncoated or poly-l-lysine (PLL)-coated membranes. Remarkably, transmigration of OS3-R5-pZCD9 cells was reduced to only 20% of Oncogene


676 Table 1

Expression of tetraspanins in LC cell lines

Cell lines

Histology

Morphology

Class I

CD9

CD63

CD81

CD82

CD151

NAG-2

SCLC NCI-H69 NCI-N231 NCI-N857 SBC-1 CADO LC6 OC10 OS1 OS2-RA OS3-R5 OS4 OS5 OS6 OS7 OS10 OS15 OS17 OS18A

Sma Sm Sm Sm Sm Sm Sm Sm Sm Sm Sm Sm Sm Sm Sm Sm Sm

M/Fb F F F M F F M M F M M/F F F F F M

+c + 7 + ++ ++ ++ + + ++ ++ + +

++ + +

+ ++ ++ + ++ ++ + ++ ++ ++ + + + + + ++ ++

++ ++ ++ ++ ++ ++ ++ ++ ++ ++ + + + ++ + ++ ++

+ ++ + 7 + + ++ + + + + + 7 + +

++ ++ ++ + ++ ++ ++ ++ + + 7 + + ++ + ++ ++

7 7 + 7 7 7 7

NSCLC A549 OAD2 CADO LC9 RERF-LC-MS PC-3 PC9 VMRC-LCD QG56 OSQ1 EBC-1 HARA Lu65 Lu99

Ad Ad Ad Ad Ad Ad Ad Sq Sq Sq Sq La La

M M M/F M M M/F M M M M M M/F M

++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++

++ ++ ++ ++ ++ + ++ ++ ++ ++ ++ ++

+ ++ ++ + ++ + + ++ ++ ++ ++

++ + ++ + + + ++ ++ ++ ++ ++ + ++

++ + + 7 ++ 7 + ++ + +

++ ++ + ++ + + ++ ++ ++ ++ ++ + ++

+ 7 + + + + ++

Po0.05

Po0.0001

NSe

NS

NS

NS

Po0.01

P-valued a

Sm: small cell carcinoma; Ad: adenocarcinoma; Sq: squamous cell carcinoma; La: large cell carcinoma LC cell lines grew as monolayer cultures (M), floating aggregates (F), or a mixture (M/F) c Class I and tetraspanin expressions were evaluated by flow cytometry. Staining was scored as ‘’, ‘7’, ‘+’, or ‘++’ if fluorescence intensity was otwofold, 2–3-fold, 3–10-fold, or >10-fold higher than staining observed for the control mouse IgG, respectively. Two additional experiments showed similar results. For CD9 expression, two MoAbs (MM2/57 and BU16) gave similar results d Differences in the number of positive (+ or ++) lines between SCLC and NSCLC were analysed by the w2 test e NS, not significant b

that of the mock transfectant or the parent. We further studied the motility of OS3-R5-pZCD9 using a timelapse videomicroscopic motility assay and obtained similar results. The parent and the mock transfectant randomly migrated on FN, while the motility of OS3R5-pZCD9 cells was again reduced to 20% of the parent. All these three lines were static on PLL (Figure 3d). These data suggested that CD9 introduction suppresses b1 integrin-dependent motility of OS3-R5. CD9 associates with b1 integrins but not with N-CAM SCLC reportedly expresses adhesion proteins, b1 integrins and N-CAM, and both of these molecules are likely to regulate motility and metastasis of SCLC (Scheidegger et al., 1994; Bredin et al., 1998). To explore the mechanisms involved in the reduction of OS3-R5pZCD9 motility, coimmunoprecipitation experiments using MoAbs against these molecules were performed (Figure 4b, left panel). Although N-CAM expression was more intense than b1 integrins or CD81 in flow Oncogene

cytometry (data not shown), CD9 coprecipitated with anti-b1 integrin and anti-CD81 MoAbs but not with control or anti-N-CAM MoAbs. Thus, CD9 specifically associates with b1 integrins but not with N-CAM in OS3-R5-pZCD9. Transfection of green fluorescence protein (GFP)-tagged CD9 into OS3-R5 To exclude the possibility that the reduction of OS3-R5pZCD9 cell motility resulted from clonal selection, we planned to again analyse CD9 effect by transient transfection with GFP-tagged CD9 (GFP-CD9) cDNA (see below). Before transient transfection, we established a stable OS3-R5 transfectant that expressed GFP-CD9 (OS3-R5-GFP-CD9) to confirm successful introduction of GFP-CD9. Immunoblotting against both GFP and CD9 clearly showed the prominent existence of 50-kDa GFP-CD9 protein in OS3-R5-GFP-CD9 lysate (Figure 4a). Minor additional bands were also stained and these were presumably degraded proteins. On the


677

Figure 1 Expression of CD9, CD63, and N-CAM in SCLC and NSCLC lines. Tumor cells were stained with MM2/57 (anti-CD9), 6H1 (anti-CD63), or ITK-2 (anti-N-CAM), and then labeled with FITC-conjugated goat anti-mouse immunoglobulin (open histograms). Normal mouse IgG was used as a control (filled histograms). Stained cells were analysed on a FACSort

Figure 2 Reduced CD9 gene expression in SCLC lines. Total RNA was extracted from SCLC and NSCLC lines. A measure of 1 mg of the total RNA was reverse-transcribed and subjected to 25 PCR amplification cycles each consisting of 40 s at 941C, 40 s at 601C, and 90 s at 721C. The amplified DNA samples were electrophoresed on 1% agarose gels, and the bands were visualized with ethidium bromide. Primers used for CD9 (Higashiyama et al., 1995; Miyake et al., 1995) and CD63 (Adachi et al., 1998) have been previously described. b-actin cDNA amplification was used as the internal control

other hand, a control line stably transfected with GFP cDNA alone (OS3-R5-GFP) showed a single 27-kDa GFP band that was not stained with anti-CD9 MoAb. We further confirmed cell-surface expression of GFPCD9 using anti-CD9 MoAb by flow cytometry and by immunoprecipitation from surface-biotinylated OS3R5-GFP-CD9 (data not shown). We next investigated if the GFP-CD9 protein associates with b1 integrins as well as untagged CD9 (Figure 4b, right panel). As expected, the chimeric GFP-CD9 coprecipitated with b1 integrins and CD81 but not with N-CAM. Several degraded proteins also appeared to coprecipitate with b1 integrins and CD81. Similar results were obtained when cell lysates of transient transfection were used (data not shown).

To investigate the intracellular distribution of GFPCD9, OS3-R5 cells were transiently transfected with GFP or GFP-CD9, and fluorescent cells were analysed (Figure 5a). Consistent with the previously reported distribution of CD9 in fixed and permeabilized cells (Nakamura et al., 1995; Berditchevski and Odintsova, 1999), GFP-CD9 was distributed at the cell periphery and at cell–cell contacts in viable transfected cells. This was in clear contrast to the homogenous distribution of GFP. Despite its presence at cell–cell contacts, GFPCD9 overexpression did not appear to promote the aggregation of tumor cells (Figure 5a). We further carried out detailed analysis of the distribution of GFP-CD9 and b1 integrins after fixation and permeabilization. Although these treatments might remove Oncogene


678

Figure 3 CD9 expression in OS3-R5 reduces cell motility on FN. (a) Flow cytometry of OS3-R-5, a mock transfectant, OS3-R5pZSV, and a CD9-overexpressing transfectant, OS3-R5-pZCD9. These cells were stained with anti-CD9 MoAb, BU16, anti-CD81 MoAb, M38, anti-b1 integrin MoAb, SG/19, or mouse control IgG, labeled with FITC-conjugated goat anti-mouse immunoglobulin, and then analysed on a FACSort. (b) Adhesion assay of OS3-R5, OS3-R5-pZSV, and OS3-R5-pZCD9. Cells (1.5 104) were allowed to adhere to the FN-, LM-, or BSA-coated wells. Unattached cells were washed away and then attached cells were quantified using an MTT assay. Values are shown as percentages of the absorbance obtained from unwashed cells on PLL. (c) Boyden chamber motility assay of OS3-R5, OS3-R5-pZSV, and OS3-R5-pZCD9. Cells (1.5 104) were applied to the upper chamber of Transwells that were uncoated or precoated with FN or PLL. Cells migrating through the membrane to the lower chamber were counted. Motility of the parent OS3-R-5 cells was also evaluated in the presence of 10 mg/ml of control IgG (*), anti-MHC class I MoAb, IOT2 (**), or anti-b1 integrin MoAb, SG/19 (***). For (b) and (c) each bar represents the mean and standard deviation of determinations made on triplicate cultures run in parallel. (d) Time-lapse videomicroscopic cell motility assay of OS3-R5, OS3-R5-pZSV, and OS3-R5-pZCD9. One thousand cells were plated on FN- or PLL-coated wells and monitored using an inverted microscope equipped with a video recorder. Cell motility was traced and analysed with an imaging device, MCID, using the program Image Analyzer. The tracks and distances of random motility of at least 30 cells were determined for each cell line. Each bar represents the mean and standard deviation

substantial amounts of GFP-CD9 from the cell surface (Berditchevski et al., 1997), Figure 5b nonetheless showed colocalization of GFP-CD9 and b1 integrins at the cell periphery and the cell–cell contact. These findings suggested that the introduced GFP-CD9 protein was targeted to the membrane properly, and it would function to inhibit motility of SCLC cells in the same way as untagged CD9. To confirm this, we performed a Boyden chamber motility assay using OS3-R5 and its stable transfectants, OS3-R5-GFP and OS3-R5-GFP-CD9. The transmigration through the FN-coated membrane of OS3-R5-GFP-CD9 was reduced to the similar level to OS3-R5-CD9, while that of OS3-R5-GFP was comparable to the parent, OS3-R5. None of these cell lines showed significant migration through the uncoated or PLL-coated membranes (data not shown). Oncogene

Transfection of GFP-CD9 into other SCLC lines uniformly reduces cell motility on FN To confirm the motility–inhibitory effect of CD9 on SCLC, we performed transient transfection of GFPCD9 into multiple SCLC lines including OS3-R5. We selected three other CD9-negative SCLC lines that are adherent to FN. Cell motility of these lines was compared with NSCLC lines by the time-lapse videomicroscopic motility assay (Figure 6a). Notably, these SCLC cell lines including OS3-R5 were more motile on FN than NSCLC lines. All the SCLC and NSCLC cells were static on PLL, suggesting that their motility on FN was mediated by integrins. In Figure 6b, OS3-R5 cells were transiently transfected with GFP or GFPCD9, and motility of fluorescent cells was analysed. When cells were traced and their tracks were visualized,


679

Figure 3 Continued

OS3-R5 cells expressing GFP alone are as motile as untransfected cells, whereas cells expressing GFP-CD9 turned less motile than untransfected cells (Figure 6b and c, top panel). Remarkably, transfection into the three other SCLC lines, CADO LC6, NCI-H69, and OC10, exhibited similar results. When transfected with GFP-CD9, these tumor cells became uniformly less motile than untransfected cells, whereas their motility was not affected when transfected with GFP alone (Figure 6c). The inhibition of motility was not complete, but was statistically significant in all the SCLC lines

(data not shown). We also performed transfection of GFP-tagged NAG-2 (GFP-NAG-2) cDNA into OS3R5 cells, and found that GFP-NAG-2 slightly reduced the tumor cell motility, but the difference was not statistically significant (data not shown). CD9 is absent in SCLC tissues To confirm downregulation of CD9 in SCLC tissues, we stained tumor tissues with anti-CD9 mAb. Representative tissue specimens were shown in Figure 7. The left Oncogene


680

Figure 4 CD9 and GFP-CD9 associate with CD81 and b1 integrins in OS3-R5 transfectants. (a) Whole cell lysate from OS3-R5 and its transfectants (OS3-R5-pZSV, OS3-R5-pZCD9, OS3-R5-GFP, and OS3-R5-GFP-CD9) were separated on 11% SDS–PAGE and transferred to an Immobilon-P membrane. Western blotting was performed with anti-CD9 MoAb, MM2/57 (left panel). After stripping, the membrane was reprobed with anti-GFP antibodies (right panel). As a control, lysate of CADO LC9, a NSCLC line, was run in parallel. (b) Immunoprecipitates from lysates of OS3-R5-pZCD9 (left panel) and OS3-R5-GFP-CD9 (right panel) were obtained using anti-CD9 MoAb, BU16, anti-CD81 MoAb, M38, anti-integrin b1 subunit MoAb, A1A5, anti-N-CAM MoAb, ITK-2, or control IgG. Immunoprecipitated proteins were separated on 11% SDS–PAGE, transferred to a membrane, and Western blotted with biotinylated MM2/57

panel shows typical submucosal invasion of SCLC cells. Notably, these invasive SCLC cells did not express CD9. Normal epithelium and bronchial glands adjacent to the tumors were CD9-positive (Figure 7, left panel). In contrast to SCLC, noninvasive, moderately differentiated adenocarcinoma cells were clearly CD9-positive (Figure 7, right panel). We stained six more SCLC specimens, and all these specimens showed negative staining of CD9 (data not shown).

Discussion Although molecules responsible for the invasion and metastasis of SCLC have been explored, few, if any, have been identified (Shtivelman, 1997; Lantuejoul et al., 1998). In the present study, we explored the possible involvement of tetraspanins in SCLC progression. Accumulating clinical evidence indicates that downregulation of several tetraspanins correlates with proOncogene

gression of solid tumors including NSCLC (Higashiyama et al., 1995; Adachi et al., 1996, 1998). Since SCLC and NSCLC are proposed to originate from a common stem cell or an early proliferative cell in the bronchial epithelium. (Mabry et al., 1991), we expected that SCLC may also downregulate CD9 or CD82 in malignant transformation. The present study, for the first time, screened multiple SCLC cell lines for tetraspanin expressions. All tetraspanins tested in this study (CD9, CD63, CD81, CD82, CD151, and NAG-2) are present in multiple organs including the lung (Nagira et al., 1994; Fitter et al., 1995; Oritani et al., 1996; Tachibana et al., 1997; Lantuejoul et al., 1998). The flow cytometry data indicated that, among these molecules, expressions of CD9 and NAG-2 were less frequent in SCLC compared with NSCLC. Almost all NSCLC cell lines are CD9-positive although the mean fluorescence intensity was variable (Table 1 and data not shown). This observation is consistent with a previous report that examined CD9, CD81, and CD82 expressions in various cancer cell lines, and showed that most NSCLC


681

Figure 5 (a) Distribution of GFP or GFP-CD9 after transfection into OS3-R5. OS3-R5 cells placed on FN were transiently transfected with GFP or GFP-CD9 and photographed under an inverted fluorescence microscope. GFP-CD9 was distributed at the cell periphery and at the cell–cell contact (right panel) in contrast to homogenous distribution of GFP (left panel). Magnifications, 400 for the larger panels. Enlarged images of transfected cells were shown in the insets. (b) Colocalization of GFP-CD9 and b1 integrins. After the transient transfection, cells were fixed and permeabilized. After treatment of anti-b1 MoAb, localization of GFPCD9 (left panel), and b1 integrins (middle panel) was analysed using a confocal laser scanning fluorescence microscope. The merged image was shown in the right panel. Colocalization of GFP-CD9 and b1 integrins were observed at the cell periphery (arrowheads) and the cell–cell contact (arrow). Magnification, 640

lines (10 of 11) expressed CD9 (White et al., 1998). Among the NSCLC lines tested in the present study, only VMRC-LCD lacked CD9. This cell line, however, might be a variant because N-myc is amplified as seen in SCLC (Kashii et al., 1992). In contrast to NSCLC, CD9 was absent or highly reduced in most SCLC lines. The three CD9-positive SCLC lines grew in floating aggregates in usual culture condition, but CD9 reactivity did not correlate with cell morphology, because the other floating SCLC lines were CD9-negative (Table 1). Also, absence of CD9 does not seem to be a common feature in cells bearing neuronal characteristics, for this protein was expressed in normal neuronal tissues (Kaprielian et al., 1995) and in two neuroblastoma cell lines, IMR32 and NB1 (I Tachibana, unpublished observations). Interestingly, all CD9-positive SCLC lines were also MHC class I-positive (Table 1). This may reflect associated expression between CD9 and class I antigens on these cell lines (Lagaudriere-Gesbert et al., 1997). Regarding the absence of another tetraspanin, NAG-2,

the significance of difference between SCLC and NSCLC was smaller than CD9. Ten of 17 SCLC lines were not stained with anti-NAG-2 MoAb, but six of 13 NSCLC lines were also clearly negative for NAG-2 (Table 1). Thus, it seems unlikely that the absence of this protein accounts for clinical aggressiveness of SCLC. The precise mechanisms of CD9 downregulation in SCLC are not known. However, CD9 gene expression was absent or markedly reduced in many SCLC lines, while there was no difference in another tetraspanin, CD63, when compared with NSCLC (Figure 2). Among SCLC lines tested, OS2-RA had a moderate level of CD9 gene expression despite no reactivity against antiCD9 MoAb (Figure 1 and Table 1). This line may have defects in translation or processing of the premature form of CD9. In the other seven SCLC lines, CD9 reduction probably occurs at or prior to the transcription level. Allelic loss of 12p13, at which the CD9 gene is located (Boucheix et al., 1985), was not reported in SCLC. Also, mutations in the CD9 gene or Oncogene


682

hypermethylation in its promoter region have not been demonstrated in cancers. Interestingly, a recent study reported that p53 tumor suppressor protein directly activates CD82 gene expression in an NSCLC line, A549

(Mashimo et al., 2000). Since tumor suppressor genes such as p53 and Rb are more frequently inactivated in SCLC than in NSCLC (Salgia and Skarin, 1998), correlation between inactivation of these tumor

Figure 6 GFP-CD9 expression uniformly suppresses cell motility of SCLC lines. (a) Four SCLC lines (open bars) and four NSCLC cell lines (closed bars) were plated onto FN (upper panel)- or PLL (lower panel)-coated wells. The tracks and distances of random motility of at least 30 cells were determined. Each bar represents the mean and standard deviation. (b) OS3-R5 cells were plated onto FN and transiently transfected with GFP (left panel) or GFP-CD9 (right panel). Motility of transfected (fluorescent) and untransfected (nonfluorescent) cells were traced, and tracks of transfected (white line) and untransfected (black line) were visualized using the imaging device, MCID, and the program Image Analyzer. The end point of each cell track is indicated by a filled circle. Magnification, 40. The results of assay is shown in the top panel of c. (c) Each SCLC line was transiently transfected with GFP (left panel) or GFPCD9 (right panel). Cell motility of fluorescent ((+), closed bars) and nonfluorescent ((), open bars) was analysed on FN. Each bar represents the mean and standard deviation Oncogene


683

Figure 6 Continued

suppressors and downregulation of tetraspanins in lung cancer should be explored. In this respect, it is of note that reconstitution of the Rb gene into a Rb-defective NSCLC cell inhibited tumor cell invasion in vitro (Li et al., 1996). Although a previous study showed that overexpression of CD9 into an NSCLC line suppressed tumor cell motility (Ikeyama et al., 1993), another study presented contradictory evidence that CD9 transfection into a Bcell line upregulated cell motility on FN and LM (Shaw et al., 1995). In addition, other studies reported that transfection of tetraspanins, CO-029 (Claas et al., 1998)

and CD151 (Testa et al., 1999), into a pancreatic cancer cell line and a cervical cancer cell line, respectively, resulted in enhanced cell migration or metastasis. Possibly, the effects on tumor cell motility and metastasis may depend on which tumor cell expresses which tetraspanin. This is compatible with the clinical evidence that CD63 reduction is associated with metastasis of melanoma (Hotta et al., 1988) but not with that of NSCLC (Adachi et al., 1998) or pancreatic cancer (Sho et al., 1998). Here we presented evidence that CD9 functions to inhibit cell motility of an SCLC line, OS3-R5, by two motility assays. Moreover, we performed transient transfection of GFP-CD9 into multiple SCLC lines to exclude the possibility that the reduced motility was a result of clonal selection. Consequently, we observed similar motility-inhibitory effects of this molecule in all SCLC lines, although the inhibition was not complete. These results suggest that, at least partially, decreased expression of CD9 may be responsible for the highly invasive property of SCLC. We assumed that GFP-CD9 would function in the same way as untagged CD9 for the following reasons. First, GFP-CD9 as well as untagged CD9 associated with b1 integrins and CD81 (Figure 4b). Although we used mild detergent Brij 99 to show this association, absence of CD9 or GFP-CD9 in control or anti-NCAM immunoprecipitation, at least assures that this association is not nonspecific. Second, GFP-CD9 was distributed and colocalized with b1 integrins at the cell periphery and cell–cell contacts (Figure 5). This distribution of GFP-CD9 was consistent with previously reported intracellular distribution of CD9 (Nakamura et al., 1995; Berditchevski and Odintsova, 1999). Third, the stable transfectant, OS3-R5-GFP-CD9, was as static as OS3-R5-pZCD9 when compared with the parent OS3-R5 or OS3-R5-GFP on FN-coated Transwells (data not shown). Moreover, the time-lapse videomicroscopic cell motility assays also revealed that GFP-CD9 inhibits cell motility (Figure 6c) as well as untagged CD9 in OS3-R5 (Figure 3d). A videomicroscopic assay after transfection of a fluorescein-tagged cDNA similar to the present one might be useful to screen various tumor lines to see if a given tetraspanin protein is motility-inhibitory or not. It is well accepted that tetraspanin forms complexes with integrins and modulates integrin functions to regulate cell motility (Hemler et al., 1996; Hemler, 1998). SCLC is surrounded by an extensive stroma of extracellular matrix including FN at both primary and metastatic sites (Sethi et al., 1999). Most lung cancer cell lines as well as normal bronchial epithelial cells express b1 integrins, and among known a and b integrins, a3b1 was uniformly expressed in SCLC lines in previous studies (Feldman et al., 1991; Sethi et al., 1999). Although a3b1 does not mediate strong adhesion to FN, a3b1-dependent cell migration on FN has been previously reported (Yauch et al., 1998). SCLC also expresses another adhesion molecule, N-CAM, which mediates homophilic cell–cell adhesion, and may be involved in SCLC metastasis (Scheidegger et al., 1994). In the present study, however, the exogenous CD9 and Oncogene


684

Figure 7 Immunohistochemical staining of CD9 in lung cancer tissues. Primary tumor tissues of SCLC (left panel) and lung adenocarcinoma (right panel) were stained with a monoclonal antibody to CD9. The left panel shows typical submucosal invasion of SCLC cells. These SCLC cells were CD9-negative, while normal epithelium and bronchial glands adjacent to the tumor were CD9positive. Moderately differentiated adenocarcinoma cells in the right panel were CD9-positive. Magnification, 50

GFP-CD9 was associated with b1 integrins, but not with N-CAM in the OS3-R5 transfectants. Thus, when introduced into SCLC cells, CD9 and GFP-CD9 presumably complex with and regulate b1 integrins to reduce cell motility on FN. In contrast to NSCLC, SCLC easily spreads into submucosal tissues and metastasizes to regional lymph nodes. Our in vitro data have raised the hypothesis that downregulation of CD9 may play a role in the early invasion of SCLC. We further considered in vivo spontaneous metastasis experiments using stable transfectants in nude mice. However, none of these SCLC lines formed metastatic foci when inoculated subcutaneously. The immunohistochemistry data may instead support our hypothesis. While CD9 is expressed in adenocarcinoma of the lung, all seven primary SCLC tumors tested were CD9-negative (Figure 7 and data not shown). It was reported that CD9 and CD82 were variably expressed in resected primary tumors of NSCLC, and that low expression of these tetraspanins correlated with poor prognosis of the patients (Higashiyama et al., 1995; Adachi et al., 1996, 1998). Moreover, CD9 reduction was more frequently observed in metastatic lymph nodes than primary tumors in breast cancers (Miyake et al., 1995). Thus, it is probable that a portion of cells loses CD9 expression and acquires advantages for invasion and metastasis in the late stage of progression in NSCLC, whereas downregulation of CD9 occurs in the much earlier stage of malignant transformation of SCLC. These two types of lung cancers may have distinct mechanisms of CD9 downregulation, which lead to distinct clinical behaviors. Since SCLC initially responds well to chemoradiotherapy but later most of the tumors recur around the primary sites and metastasize to distant organs, suppression of the early invasion is important to overcome the spread of SCLC. In a recent study, adenovirus-mediated CD9 delivery into an inoculated murine melanoma markedly inhibited pulmonary metastasis and prolonged survival of the mice (Miyake et al., 2000). Thus, CD9 may be a promising candidate for a novel gene-delivery strategy against SCLC. Oncogene

In summary, we have shown that, among tetraspanins, CD9 expression is selectively downregulated in SCLC but not in NSCLC, and that recovery of CD9 expression suppresses cell motility of SCLC. The difference in clinical aggressiveness between SCLC and NSCLC may be, at least in part, because of the absence of this motility-related protein in SCLC.

Materials and methods Cell lines NCI-H69, NCI-N231, NCI-N857, Lu65, and Lu99, were gifts from Dr Y Shimosato, National Cancer Research Institute, Tokyo, Japan. SBC-1, RERF-LC-MS, PC-3, VMRC-LCD, and EBC-1 were obtained from the Japanese Collection of Research Bioresources (Tokyo, Japan). OC10, CADO LC6, and CADO LC9 were provided by Osaka Medical Center for Cancer and Cardiovascular Diseases, Osaka, Japan. Seven SCLC cell lines, OS1, OS2-RA, OS3-R5, OS4, OS5, OS6, and OS7, were established in our laboratory, and their biological properties were previously characterized (Tanio et al., 1992; Saito et al., 1994). Four additional SCLC lines, OS10, OS15, OS17, and OS18A, a lung adenocarcinoma line, OAD2, and a lung squamous cell carcinoma line, OSQ1, were also established in our laboratory. A549, QG56, and neuroblastoma lines, IMR-32 and NB1, were obtained from the American Type Culture Collection, Rockville, MD, USA. PC9 was donated by Dr Y Hayata, Tokyo Medical College, Tokyo, Japan. HARA was a kind gift from Dr H Iguchi, Kyusyu Cancer Center, Fukuoka, Japan. All cell lines were maintained in RPMI 1640 medium supplemented with 10% heatinactivated fetal calf serum (FCS), 100 U/ml penicillin, and 100 mg/ml streptomycin, and were free of Mycoplasma infection when tested with Bacto pleuropneumonia-like organism agar (Difco Laboratories, Detroit, MI, USA). All SCLC or NSCLC lines described above were examined for N-CAM expression by flow cytometry prior to experiments, and confirmed to be N-CAM-positive or -negative, respectively (data not shown). Antibodies Mouse anti-CD9 MoAbs, BU16 and MM2/57, were purchased from Biodesign International (Kennebunk, ME, USA) and Biosource International (Camarillo, CA, USA), respectively.


685 M38 and M104, mouse MoAbs against CD81 and CD82, respectively, were gifts from Dr O Yoshie (Kinki University School of Medicine, Osaka, Japan). 6H1, 5C11, and NAG-1, mouse MoAbs against CD63, CD151, and NAG-2, respectively, and A1A5 and SG/19, mouse anti-b1 integrin MoAbs, were previously described (Tachibana et al., 1996; Berditchevski et al., 1997; Tachibana et al., 1997; Yauch et al., 1998). A mouse anti-N-CAM MoAb, ITK-2, was established in our laboratory and characterized previously (Saito et al., 1991). A mouse anti-MHC class I MoAb, W6/32, was provided by Dr ME Hemler (Dana-Farber Cancer Institute, Boston, MA, USA). Another mouse anti-MHC class I MoAb, IOT2, was purchased from Immunotech, ME, USA. Rabbit anti-GFP antibodies and normal mouse IgG were purchased from Clontech (Palo Alto, CA, USA) and Serotec (Kidlington, England), respectively. Flow cytometry Ten thousand cells were stained with ITK-2 (anti-N-CAM), W6/32 (anti-MHC class I), MM2/57 and BU16 (anti-CD9), 6H1 (anti-CD63), M38 (anti-CD81), M104 (anti-CD82), 5C11 (anti-CD151), or NAG-1 (anti-NAG-2) at a concentration of 10 mg/ml, washed twice, and then labeled with fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse immunoglobulin (Biosource International). Normal mouse IgG was used as a control. Stained cells were analysed on a FACSort (Becton-Dickinson, San Jose, CA, USA).

lines were transiently transfected with GFP-tagged CD9 or GFP alone, and analysed in a time-lapse videomicroscopic cell motility assay. Western blotting Cells were lysed at 41C for 1 h in analysis buffer containing 1% Brij 99, 25 mm HEPES, pH 7.5, 150 mm NaCl, 5 mm MgCl2, 2 mm phenylmethylsulfonyl fluoride,10 mg/ml aprotinin, and 10 mg/ml leupeptin. Samples containing an equal protein concentration were mixed with an equivalent volume of sample buffer (50 mm Tris/HCl, pH 6.5, 10% glycerol, 2% SDS, and 0.1% bromophenol blue), boiled for 5 min, electrophoresed on 11% SDS–polyacrylamide gels under nonreducing conditions, and transferred to Immobilon-P membranes (Millipore, Bedford, MA, USA). After washes, nonspecific binding sites of the membranes were blocked in PBS containing 30 mg/ml nonfat skimmed milk for 30 min. Then the membranes were probed with an anti-CD9 MoAb, MM2/57, diluted at 1 mg/ml by shaking overnight at 41C, followed by 1-h incubation with peroxidase-conjugated goat anti-mouse IgG (H+L) (BIO-RAD, Hercules, CA, USA) diluted 1 : 2000. Immunoreactive bands were visualized by treating with Renaissance Chemiluminescent Reagent (NENTM Life Science Products, Boston, MA, USA). After MM2/57 was stripped, the membrane was reprobed with rabbit anti-GFP antibody (Clontech). Immunoprecipitation

RT–PCR analysis Total RNA was extracted from cultured lung cancer cell lines with Isogen (Nippon Gene, Tokyo, Japan). A volume of 1 mg of total RNA was reverse transcribed using One Step RNA PCR Kit (Takara, Kyoto, Japan) following the manufacturer’s instructions. The generated cDNAs were amplified using primers for CD9 (50 -TGCATCTGTATCCAGCGCCA-30 and 50 -CTCAGGGATGTAAGCTGACT-30 ) (Higashiyama et al., 1995; Miyake et al., 1995) and CD63 (50 CCCGAAAAACAACCACACTGC-30 and 50 -GATGAGGAGGCTGAGGAGACC-30 ) (Adachi et al., 1998). b-actin cDNA amplification was used as the internal control. Primers used for b-actin were 50 -CAAGAGATGGCCACGGCTGCT30 and 50 -TCCTTCTGCATCCTGTCGGCA-30 . These primer pairs for CD9, CD63, and b-actin amplify 799-, 347-, and 274bp fragments, respectively. The reaction mixtures were subjected to 25 PCR amplification cycles each consisting of 40 s at 941C, 40 s at 601C, and 90 s at 721C. The amplified DNA samples were electrophoresed on 1% agarose gels and visualized with ethidium bromide. We confirmed that parameters on these RT–PCR conditions yielded amplification of template DNAs within a linear range. cDNA transfection The coding region of CD9 cDNA was generated by RT–PCR and cloned in frame into expression vectors, pZeoSV (Invitrogen, Carlsbad, CA, USA) and pEGFP-C1 (Clontech). In the latter, CD9 cDNA was fused to the carboxyl terminus of GFP with the addition of a nine-amino-acid polylinker (SGLRSRAQA). OS3-R5-pZCD9 and OS3-R5-GFP-CD9 were established by transfection of these constructed vectors into OS3-R5 using LipofectAMINE 2000 Reagent (GibcoBRL, Rockville, MD, USA) followed by selection in Zeocin (Invitrogen) and G-418 sulfate (GibcoBRL), respectively. OS3-R5-pZSV and OS3-R5-GFP were similarly established by transfection with the vectors alone. In other experiments, several other SCLC

Whole cell lysates were precleared by incubation with protein A-Sepharose beads (Sigma, St Louis, MO, USA) for 1 h at 41C. Then the lysates containing equal amounts of protein were rotated for 1 h at 41C with mouse anti-CD9 MoAb, BU16, anti-CD81 MoAb, M38, anti-b1 integrin MoAb, A1A5, anti-N-CAM MoAb, ITK-2, and normal mouse IgG. Immune complexes were collected by an additional rotation with protein A-Sepharose beads for 1 h at 41C. After washing three times with lysis buffer and boiling for 5 min in sample buffer, immunoprecipitated proteins were separated on 11% SDS– polyacrylamide gel electrophoresis (PAGE) under nonreducing conditions and transferred to an Immobilon-P membrane. Western blotting was performed by incubation with biotinylated anti-CD9 MoAb, MM2/57, followed by peroxidaseconjugated streptavidin (Zymed, San Francisco, CA, USA). Bands were visualized with Renaissance Chemiluminescent Reagent as described above. Cell adhesion assay Human plasma FN (GibcoBRL), murine EHS sarcoma laminin-1 (LM) (Sigma), and BSA (Sigma) were diluted in 0.1 m sodium bicarbonate at a concentration of 20 mg/ml. Wells in a 96-well nontissue-culture-treated plate (Linbro, McLean, VA, USA) were precoated with 100 ml of the solution overnight at 41C. Nonspecific binding sites were then blocked with 100 ml of PBS, 0.1% BSA at 371C for 2 h. After washing three times with PBS, 1.5 104 cells resuspended in 100 ml of RPMI 1640 were allowed to adhere to the FN-, LM-, or BSA-coated wells for 1.5 h at 371C. Unattached cells were eliminated by shaking, and then the remaining adherent cells were evaluated using a 3(4, 5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT) assay as previously described (Tachibana et al., 1996). Boyden chamber cell motility assay Tissue-culture-treated Transwell membranes (pore size, 8 mm, Costar, Cambridge, MA, USA) were precoated with FN or Oncogene


686 PLL as described above. Cells (5 104) resuspended in 100 ml of RPMI 1640, 10% FCS were applied to the upper chamber of Transwells. A volume of 600 ml of RPMI 1640, 10% FCS were added to the lower chamber. After incubation for 18 h at 371C, cells migrating through the membrane to the lower chamber and adhering to the bottom were counted. For antibody blocking experiments, OS3-R5 cells were preincubated with 10 mg/ml of antibodies for 30 min, and then directly applied to the upper chamber of Transwells. Time-lapse videomicroscopic cell motility assay One thousand cells suspended in 100 ml of RPMI 1640, 10% FCS were plated onto an FN-, or PLL-precoated wells in a 96well plate and cultured for 24 h before monitoring. The cells were then incubated at 371C, 10% CO2, and monitored using an inverted fluorescence microscope, Diaphoto (Nikon, Tokyo, Japan) equipped with an SVHS video recorder (Sony, Tokyo, Japan) for 2 h (Kashii et al., 1992; White et al., 1998). The tracks and distances of random motility of at least 30 cells were obtained by the microcomputer imaging device (MCID) (Imaging Research, Ontario, Canada) using the program, Image Analyzer (Ver. 3) (Imaging Research). Immunofluorescence OS3-R5 cells cultured on FN were transiently transfected with GFP-tagged CD9 as described above, and incubated for 24 h. After washes, cells were fixed with PBS, 4% paraformaldehyde for 15 min, permeabilized with PBS, 0.1% Triton X-100, and blocked with PBS, 3% BSA for 30 min at room temperature. Immunofluorescent staining was conducted with anti-b1 integrin MoAb, SG/19, followed by rhodamine-conjugated

goat anti-mouse IgG. Localization of GFP-CD9 and b1 integrins were analysed using a confocal laser scanning fluorescence microscope (Carl Zeiss, Thornwood, NY, USA). Immunohistochemistry The immunoperoxidase procedures (ABC method) using a monoclonal antibody to CD9 (Novocastra, Newcastle, UK; diluted 1 : 20) was carried out in transbronchial biopsy specimens of seven cases of SCLC and one case of lung adenocarcinoma as previously described (Hsu et al., 1981). For antigen retrieval, sections were pretreated with microwaving for 15 min in 1 mm EDTA (pH 8.0) before incubation. Statistical analysis Statistical significance of differences in the number of positive lines for each tetraspanin expression between SCLC and NSCLC was determined by w2 test. Acknowledgments This work was partially supported by a grant from Sankyo Foundation of Life Science, Tokyo, Japan, and by Grants-inAid for Cancer Research from the Ministry of Education, Science and Culture, Japan. The authors are grateful to Dr Martin E Hemler (Dana-Farber Cancer Institute, Boston) and Dr Osamu Yoshie (Kinki University School of Medicine, Osaka, Japan), for providing antibodies to tetraspanins. We also thank Yasuo Katsuki and Eiji Oiki (Osaka University Graduate School of Medicine) for technical assistance on the time-lapse videomicroscopic cell motility assay and Yoko Habe for secretarial assistance.

References Adachi M, Taki T, Konishi T, Huang CI, Higashiyama M and Miyake M. (1998). J. Clin. Oncol., 16, 1397–1406. Adachi M, Taki T, Ieki Y, Huang CL, Higashiyama M and Miyake M. (1996). Cancer Res., 56, 1751–1755. Atkinson B, Ernst CS, Ghrist BF, Ross AH, Clark WH, Herlyn M, Herlyn D, Maul G, Steplewski Z and Koprowski H. (1985). Hybridoma, 4, 243–255. Berditchevski F and Odintsova E. (1999). J. Cell. Biol., 146, 477–492. Berditchevski F, Tolias KF, Wong K, Carpenter CL and Hemler ME. (1997). J. Biol. Chem., 272, 2595–2598. Bredin CG, Sundqvist KG, Hauzenberger D and Klominek J. (1998). Eur. Respir. J., 11, 400–407. Boucheix C, Nguyen-van-Cong Perrot JY, Foubert C, Gross MS, Weil D, Laisney V, Rosenfeld C and Frezal J. (1985). Ann. Genet., 28, 19–24. Cajot JF, Sordat I, Silvestre T and Sordat B. (1997). Cancer Res., 57, 2593–2597. Carney DN, Gazdar AF, Bepler G, Guccion JG, Marangos PJ, Moody TW, Zweig MH and Minna JD. (1985). Cancer Res., 45, 2913–2923. Claas C, Seiter S, Claas A, Savelyeva L, Schwab M and Zoller M. (1998). J. Cell Biol., 141, 267–280. Doyle A, Martin WJ, Funa K, Gazdar A, Carney D, Martin SE, Linnoila I, Cuttitta F, Mulshine J and Bunn P. (1985). J. Exp. Med., 161, 1135–1151. Feldman LE, Shin KC, Natale RB and Todd RF. (1991). Cancer Res., 51, 1065–1070. Fitter S, Tetaz TJ, Berndt MC and Ashman LK. (1995). Blood, 86, 1348–1355. Oncogene

Guo X, Friess H, Graber HU, Kashiwagi M, Zimmermann A, Korc M and Buchler MW. (1996). Cancer Res., 56, 4876–4880. Hemler ME. (1998). Curr. Opin. Cell Biol., 10, 578–585. Hemler ME, Mannion BA and Berditchevski F. (1996). Biochim. Biophys. Acta., 1287, 67–71. Higashiyama M, Taki T, Ieki Y, Adachi M, Huang CL, Koh T, Kodama K, Doi O and Miyake M. (1995). Cancer Res., 55, 6040–6044. Hotta H, Ross AH, Huebner K, Isobe M, Wendeborn S, Chao MV, Ricciardi RP, Tsujimoto Y, Croce CM and Koprowski H. (1988). Cancer Res., 48, 2955–2962. Hsu SM, Raine L and Fanger H. (1981). Am. J. Clin. Pathol., 75, 816–821. Huang CI, Kohno N, Ogawa E, Adachi M, Taki T and Miyake M. (1998). Am. J. Pathol., 153, 973–983. Ikeyama S, Koyama M, Yamaoko M, Sasada R and Miyake M. (1993). J. Exp. Med., 177, 1231–1237. Inoue G, Horiike N and Onji M. (2001). Int. J. Mol. Med., 7, 67–71. Kaprielian Z, Cho KO, Hadjiargyrou M and Patterson PH. (1995). J. Neurosci., 15, 562–573. Kashii T, Mizushima Y, Nakagawa K, Monno S and Yano S. (1992). Anticancer Res., 12, 621–624. Lagaudriere-Gesbert C, Lebel-Binay S, Wiertz E, Ploegh HL, Fradelizi D and Conjeaud H. (1997). J. Immunol., 158, 2790–2797. Lantuejoul S, Moro D, Michalides RJ, Brambilla C and Brambilla E. (1998). Am. J. Surg. Pathol., 22, 1267–1276.


687 Le Naour F, Rubinstein E, Jasmin C, Prenant M and Boucheix C. (2000). Science, 287, 319–321. Li J, Hu SX, Perng GS, Zhou Y, Xu K, Zhang C, Seigne J, Benedict WF and Xu HJ. (1996). Oncogene, 13, 2379–2386. Lombardi DP, Geradts J, Foley JF, Chiao C, Lamb PW and Barrett JC. (1999). Cancer Res., 59, 5724–5731. Mabry M, Nelkin BD, Falco JP, Barr LF and Baylin SB. (1991). Cancer Cells, 3, 53–58. Maecker HT, Todd SC and Levy S. (1997). FASEB J., 11, 428–442. Mashimo T, Bandyopadhyay S, Goodarzi G, Watabe M, Pai SK, Gross SC and Watabe K. (2000). Biochem. Biophys. Res. Commun., 274, 370–376. Miyado K, Yamada G, Yamada S, Hasuwa H, Nakamura Y, Ryu F, Suzuki K, Kosai K, Inoue K, Ogura A, Okabe M and Mekada E. (2000). Science, 287, 321–324. Miyake M, Inufusa H, Adachi M, Ishida H, Hashida H, Tokuhara T and Kakehi Y. (2000). Oncogene, 19, 5221–5226. Miyake M, Nakano K, Itoi SI, Koh T and Taki T. (1996). Cancer Res., 56, 1244–1249. Miyake M, Nakano K, Ieki Y, Adachi M, Huang CL, Itoi S, Koh T and Taki T. (1995). Cancer Res., 55, 4127–4131. Mountain CF. (1978). Semin. Oncol., 5, 272–279. Nagira M, Imai T, Ishikawa I, Uwabe KI and Yoshie O. (1994). Cell Immunol., 157, 144–157. Nakamura K, Iwamoto R and Mekada E. (1995). J. Cell Biol., 129, 1691–1705. Oritani K, Wu X, Medina K, Hudson J, Miyake K, Gimble JM, Burstein SA and Kincade PW. (1996). Blood, 87, 2252–2261. Rubinstein E, Le Naour F, Lagaudriere-Gesbert C, Billard M, Conjeaud H and Boucheix C. (1996). Eur. J. Immunol., 26, 2657–2665. Saito S, Tanio Y, Tachibana I, Hayashi S, Kishimoto T and Kawase I. (1994). Lung Cancer, 10, 307–318. Saito S, Inoue T, Kawase I, Hara H, Tanio Y, Tachibana I, Hayashi S, Watanabe M, Matsunashi M and Osaki T. (1991). Cancer Immunol. Immunother., 33, 165–170.

Salgia R and Skarin AT. (1998). J. Clin. Oncol., 16, 1207–1217. Scheidegger EP, Lackie PM, Papay J and Roth J. (1994). Lab Invest., 70, 95–106. Sethi T, Rintoul RC, Moore SM, MacKinnon AC, Salter D, Choo C, Chilvers ER, Dransfield I, Donnelly SC, Strieter R and Haslett C. (1999). Nat. Med., 5, 662–668. Shaw AR, Domanska A, Mak A, Gilchrist A, Dobler K, Visser L, Poppema S, Fliegel L, Letarte M and Willett BJ. (1995). J. Biol. Chem., 270, 24092–24099. Sho M, Adachi M, Taki T, Hashida H, Konishi T, Huang CL, Ikeda N, Nakajima Y, Kanehiro H, Hisanaga M, Nakano H and Miyake M. (1998). Int. J. Cancer, 79, 509–516. Shtivelman E. (1997). Oncogene, 14, 2167–2173. Souhami RL, Beverley PC and Bobrow LG. (1987). Lancet, 2, 325–326. Tachibana I and Hemler ME. (1999). J. Cell Biol., 146, 893–904. Tachibana I, Bodorova J, Berditchevski F, Zutter MM and Hemler ME. (1997). J. Biol. Chem., 272, 29181–29189. Tachibana I, Mori M, Tanio Y, Hosoe S, Sakuma T, Osaki T, Ueno, K, Kumagai T, Kijima T and Kishimoto T. (1996). Exp. Cell Res., 227, 230–239. Tanio Y, Watanabe M, Osaki T, Tachibana I, Kawase I, Kuritani T, Saito S, Masuno T, Kodama N and Furuse K. (1992). Jpn J. Cancer Res., 83, 736–745. Testa JE, Brooks PC, Lin JM and Quigley JP. (1999). Cancer Res., 59, 3812–3820. Uchida S, Shimada Y, Watanabe G, Li ZG, Hong T, Miyake M and Imamura M. (1999). Br. J. Cancer, 79, 1168–1173. White A, Lamb PW and Barrett JC. (1998). Oncogene, 16, 3143–3149. Yang X, Wei L, Tang C, Slack R, Montgomery E and Lippman M. (2000). Clin. Cancer Res., 6, 3424–3429. Yauch RL, Berditchevski F, Harler MB, Reichner J and Hemler ME. (1998). Mol. Biol. Cell, 9, 2751–2765.

Oncogene