Oncogene (2000) 19, 1772 ± 1782 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc Down-regulation of the extracellular matrix protein SPARC in vSrc- and vJun-transformed chick embryo ®broblasts contributes to tumor formation in vivo

Emmanuel Vial1 and Marc Castellazzi*,1

1Unite de Virologie Humaine, Institut National de la Sante et de la Recherche MeÂdicale (INSERM-U412), Ecole Normale SupeÂrieure, 46 alleÂe d'Italie, 69364 Lyon Cedex 07, France

In vitro transformation of primary cultures of chick cytoskeleton and the extracellular matrix (Jove and embryo ®broblasts by the membrane-bound vSrc or the Hanafusa, 1987). nuclear vJun oncoproteins is correlated with a down- Transcription factor AP1 is a nuclear target of the regulation of the secreted glycoprotein SPARC (also vSrc-induced oncogenic pathway (Muramaki et al., called BM-40 or osteonectin). This protein is a non- 1997; Suzuki et al., 1994) and is thought to trigger structural component of the extracellular matrix that is part of the phenotypic spectrum induced by vSrc thought to regulate cell ± matrix interaction during (Bos et al., 1990; Castellazzi et al., 1990; Muramaki development, wound repair, and carcinogenesis. Its et al., 1997; Suzuki et al., 1994; van Dam et al., precise function remains unclear. To estimate the 1998). This transcription factor consists of a contribution of SPARC down-regulation to the major collection of homo- and heterodimers between aspects of the transformed phenotype, we have re- members of the Jun, Fos and ATF/CREB family expressed this protein from a self-replicating retrovirus (Karin et al., 1997). The vJun oncoprotein, a Rcas, designated R-SPARC, in the transformed cultures. mutated version of cellular cJun, initially isolated These R-SPARC-infected cultures display the following from retrovirus ASV17 (Maki et al., 1987), is main properties: (i) they accumulate the SPARC protein thought to transform by promoting aberrant tran- to a level identical to or only slightly higher than the scription of oncogenically relevant target genes level in normal chick embryo ®broblasts; (ii) they retain (Vogt, 1994). the main phenotypic properties characteristic of in vitro In the search for such target genes whose activity transformed cells, that is, altered morphology, capacity is modulated in vSrc- and/or vJun- transformed to grow in a reduced amount of serum, and capacity to CEFs, we became interested in a particular compo- develop colonies from single cells in agar; (iii) they nent of the extracellular matrix, called SPARC for display a clearly reduced capacity to develop local secreted protein, acidic and rich in cysteine (Mason ®brosarcomas in vivo. Taken together, these data et al., 1986a,b). This protein was initially designated strongly suggest that down-regulation of SPARC as osteonectin (Termine et al., 1981) and later as contributes to the transformed phenotype triggered by BM-40 for basement membrane-40 (Dziadek et al., vSrc and vJun in primary avian ®broblasts, by 1986; reviewed in Reed and Sage, 1996; Timpl and facilitating in vivo tumorigenesis. Oncogene (2000) 19, Aumailley, 1993). SPARC displays a high degree of 1772 ± 1782. sequence conservation from Caenorhabditis elegans to humans, indicating a basic functional role in animal Keywords: vSrc; vJun; AP1; SPARC; chick embryo tissues (Lankat-Buttgereit et al., 1998). Although ®broblasts; cell transformation; tumorigenesis SPARC level is highest in bone and in basement membranes, it is also expressed in a variety of cell types and associated with remodelling tissues and Introduction sites of high cellular turn-over. In vitro studies with SPARC peptides (Funk and Sage, 1993) and with Transformation of chick embryo ®broblasts (CEFs) primary cultures from SPARC null mice (Bradshaw as induced by the membrane-associated vSrc onco- et al., 1999) indicate that this protein regulates cell protein from Rous sarcoma virus, results in the proliferation. SPARC has been shown to be acquisition of four major, abnormal phenotypic repressed by cJun in rat embryo ®broblasts at the properties: (i) an alteration in cell morphology; (ii) mRNA and protein level (Mettouchi et al., 1994) a reduced serum requirement for proliferation; (iii) an and by vSrc in CEFs at the mRNA level (Young et anchorage-independent growth capacity; and (iv) in al., 1986), suggesting a role for this protein in cell vivo tumorigenicity. Numerous reports in this avian transformation. model have associated transformation with complex In the present study, we asked whether SPARC changes in gene expression, a€ecting among other down-regulation contributes to the transformed aspects the composition and organization of the phenotype established by vSrc or vJun in primary avian cells. Therefore CEFs stably transformed by these oncoproteins were generated. After having shown that SPARC was indeed repressed in these *Correspondence: M Castellazzi cultures, we analysed the consequence of its re- Received 4 October 1999; revised 3 December 1999; accepted 25 expression with respect to the major oncogenic January 2000 properties induced by these oncoproteins. Anti-tumoral effect of SPARC in chickens E Vial and M Castellazzi 1773 Results

Steady-state level of SPARC mRNA and protein in CEFs stably transformed by vSrc and vJun Uninfected CEFs or CEF cultures chronically infected by R, R-vSrc, R-vJun, or R-vJun-m1 (a mutant of vJun that exhibits enhanced tumorigenicity in vivo; Huguier et al., 1998) were generated. Total RNA was subjected to Northern blot analysis with an avian SPARC radioactive probe (Figure 1). The level of SPARC mRNA was estimated in the transformed CEFs in comparison with the non-transformed CEFs (referred to as 100%). A reduction of 50% was found with R-vJun, and of 80% with R-vSrc. The strongest reduction (over 90%) was observed with R-vJun-m1. Reprobing of the same blot with GAPDH cDNA Figure 2 Schematic representation of the retroviral vector indicated the same loading for the di€erent lanes. Rcas-envD-SPARC sig+ designated RD-SPARC in the text. Protein extracts from the same cultures were The SPARC sig+ gene has been cloned at the unique ClaI site subjected to Western blot analysis followed by present in the vector. sd: splice donor; sa: splice acceptor. immunodetection of the SPARC protein. As expected Retroviral genes gag, pol, and envD are shown from the mRNA levels, SPARC accumulation was reduced in the transformed CEFs (Figure 3a). With non-transformed CEFs arbitrarily set at 100%, vSrc- and vJun-m1-transformed CEFs showed almost un- could be resolved into two distinct, closely-migrating detectable SPARC levels, i.e. 510% and 55%, bands. Although the relative intensity of these bands respectively. These data show that transformation of varied from one extract to another, the upper band CEFs by either vJun or vSrc is accompanied by a clear- was usually more intense. Both bands seemed to cut reduction in the accumulation of the SPARC correspond to the mature form of SPARC, devoid of mRNA and protein. the peptide signal (see right lane `in vitro translated' SPARC sig+ and SPARC sig7 for comparison). As shown in Figure 3b, a treatment of the extracts with Re-expression of SPARC in vSrc- and vJun-transformed N-glycosidase shifted the slower-migrating upper band CEFs to a single, fast-migrating band. The two closely- An Rcas envD derivative carrying the entire coding migrating bands are therefore likely to correspond to sequence of SPARC with the N-terminal peptide two isoforms of mature SPARC with distinct signal, designated RD-SPARC, was constructed glycosylation levels. This explanation is consistent (Figure 2). CEFs infected by RD or RD-SPARC with the existence of potential N-linked glycosylation were generated and superinfected with R, R-vJun, R- sites at Asn 67 and Lys 95 in the avian protein vJun-m1, or R-vSrc. Protein cell extracts were (Bassuk et al., 1993). subjected to Western blot analysis and probed Detection of the SPARC protein was also performed successively with antibodies speci®c to SPARC, Jun, at the single cell level by immuno¯uorescence. In and Src. As shown in Figure 3a (upper panel), agreement with previous reports (Mettouchi et al., infection with RD-SPARC correlated with an en- 1994; Reed and Sage, 1996), SPARC could be detected hanced accumulation of the SPARC protein in all intracellularly in the cytoplasm. The staining was cultures, reaching a level identical to, or slightly homogeneously distributed within the cell population. (about twofold) higher than the level in normal CEFs. As expected, SPARC accumulation was strongly By short exposure (Figure 3a lower panel), SPARC reduced in m1- and Src-transformed CEFs, and returned to a wild-type level as a consequence of RD-SPARC superinfection (data not shown). In the Jun-transformed CEFs (Figure 3c, upper panel), accumulation of the vJun and vJun-m1 oncoproteins was clearly detected, reaching about ®vefold the level of the endogenous cJun. There was no signi®cant di€erence in the presence or absence of exogenous SPARC. In the vSrc-transformed CEFs (Figure 3c, lower panel), a strong accumulation of the oncoprotein was observed, reaching over 20-fold the level of the endogenous cSrc. Again, there was no signi®cant di€erence with or without exogenous SPARC. In these vSrc-transformed cells endogenous cJun accumulated as a consequence of the activation of Figure 1 Level of accumulation of SPARC mRNA from CEFs AP1 by vSrc (Suzuki et al., 1994). chronically infected by the retroviruses R, R-vJun, R-vJun-m1 Taken together, these data show that SPARC can be (designated m1), or R-vSrc by Northern blot analysis. Total RNA was probed with either the complete SPARC probe or a GAPDH eciently re-expressed in the RD-SPARC-infected, probe as indicated transformed CEFs, and accumulates to levels identical

Oncogene Anti-tumoral effect of SPARC in chickens E Vial and M Castellazzi 1774

Figure 3 Level of accumulation of the SPARC (a and b), Jun and Src (c) proteins from CEFs chronically infected by the empty vector Rcas, or by Rcas vectors expressing vJun, vJun-m1 or vSrc with or without SPARC as indicated. Western blotting was followed by detection of the accumulated proteins with speci®c antibodies using the ECL detection system. (a) Long and short exposures are shown (upper and lower panels, respectively), and in vitro translated SPARC sig+ and SPARC sig7 proteins are presented (right lane). (b) Shows the various extracts treated or untreated with N-glycosidase; the arrows on the right indicate the two SPARC isoforms. In (c) the arrows point to the endogenous cJun, or the de novo expressed oncoprotein vJun, vJun-m1, and vSrc, as indicated

to or slightly higher than those observed in normal Effect of SPARC re-expression on Jun- and Src-mediated cultures. Furthermore the de novo expression of transformed phenotype in vitro SPARC does not a€ect the levels of accumulation of exogeneously introduced and constitutively expressed RD-SPARC did not signi®cantly alter the morphology oncoproteins vJun and vSrc. These various cultures of either non-transformed or transformed CEFs. vJun- therefore constitute an appropriate tool to investigate a and vSrc-transformed cells displayed a typical fusiform possible contribution of SPARC down-regulation to (Bos et al., 1990; Maki et al., 1987) or slightly rounded carcinogenesis, and to investigate whether SPARC re- and refractile morphology (Jove and Hanafusa, 1987; expression may reverse the transformed phenotype Mayer et al., 1986), regardless of their SPARC content. established by vJun and vSrc. Moreover RD-SPARC did not a€ect the main growth

Oncogene Anti-tumoral effect of SPARC in chickens E Vial and M Castellazzi 1775 parameters in normal medium supplemented with 6% Effect of SPARC re-expression on Jun- and Src-mediated serum. Thus the doubling time of exponentially tumorigenesis in vivo growing cultures (Figure 4a), and the plating ecien- cies at very low density (16103 cells/100 mm plate; In a ®rst series of experiments, the development of data not shown) remained unchanged. local, primary ®brosarcomas was tested by injecting Surprisingly, RD-SPARC-infected CEFs, slightly CEFs stably transformed by vJun-m1 (26106 cells per and reproducibly, stimulated proliferation in the animal), into the wing web of 1-day-old chicks. vJun- transformed cells. When plated in medium with 0.6% m1 was chosen instead of vJun because this mutant is a serum (conditions that allow growth of the trans- formed CEFs but not of the normal CEFs), and after a 12 day incubation, the vJun-, vJun-m1-, and vSrc- Table 1 Colony formation of CEF cultures fully infected with the 5 transformed cultures reached a density of 17610 , various Rcas derivatives 246105,and116105 cells per plate, respectively with Number of colonies per 5 5 excess SPARC, compared to 9610 ,10610 ,and CEFs chronically plate after plating with 46105, respectively in the absence of SPARC (Figure infected with Rcas 16104 56103 26103 4b). A small enhancement of plating eciency with derivatives expressing cells cells cells SPARC was also observed for CEFs transformed by R±±± vJun, vJun-m1, and vSrc when plated as single cells in SPARC ± ± ± agar (by a 1.1-, 1.3-, and 2.3-fold factor, respectively) (Table 1). RD-SPARC could not in itself induce vJun + 644/532 260/248 (12.7%) growth in low-serum (Figure 4b, left) or colony SPARC+vJun + 800/896 290/250 (14.7%) formation in agar (Table 1) in the non-transformed m1 + 382/458 162/188 (8.7%*) cultures. SPARC+ml + 580/556 222/250 (11.8%*) These results were unexpected. They showed that, although RD-SPARC cannot transform CEFs, this vSrc + 1060/926 278/398 (16.9%) virus stimulates the transforming activity mediated by SPARC+vSrc + 1736/2032 806/792 (39.9%) vJun(-m1) or vSrc, with respect to proliferation in low Colonies were counted under the microscope after 2 weeks. Each serum and in agar. Thus, down-regulation of SPARC number represents the number of colonies in a given 60 mm petri dish. Numbers in brackets refer to the percentage of single cells that is not needed for in vitro transformation by Jun and developed into colonies. *In this particular experiment, ml cells still Src; on the contrary, excess SPARC even constitutes a gave rise to a signi®cant number of colonies (van Dam et al., 1998); stimulatory factor in this oncogenic process. 7 = absence of colonies; + = growth of colonies

Figure 4 Growth curves of CEF cultures chronically infected by the empty Rcas vector, or by Rcas vectors expressing vJun, vJun- m1, or vSrc with or without SPARC. Cultures were grown either in normal medium (supplemented with 6% serum; (a)) or in medium with reduced amount of serum (0.6%; (b)). Cell numbers correspond to the mean for two 100 mm petri dishes

Oncogene Anti-tumoral effect of SPARC in chickens E Vial and M Castellazzi 1776

Figure 5 Tumor formation in young chicks after injection of CEFs transformed by either vJun-m1 (a) or vSrc (b) with or without SPARC as indicated. Cells (26106 and 0.56106 for m1- and Src-transformed CEFs, respectively) were injected subcutaneously into the wing web of 1-day-old chicks. Data represent tumor size after a given number of weeks as indicated. In a chicks no. 7, 8, 9 and 10 did not develop any visible tumors and were sacri®ced after 6 weeks

Oncogene Anti-tumoral effect of SPARC in chickens E Vial and M Castellazzi 1777 more potent tumor-inducer (Huguier et al., 1998). The of such a Jun mutant might have contributed to the formation of tumors was monitored over a 6-week development of the unique tumor in chick no. 6 that period. As shown in Figure 5a, vJun-m1-transformed arose from a culture co-expressing m1 and SPARC; CEFs induced tumors in all birds, i.e. in ®ve birds out and ®nally (iii) the level of vSrc was the same in all of ®ve injected. Strikingly, exogenous SPARC strongly cultures in which this oncoprotein had been intro- reduced the number of tumors, since only one bird (no. duced. 6) out of ®ve developed a tumor. Furthermore this Taken together, the data showed that re-expression tumor developed late, between weeks 5 and 6. Birds of SPARC in vJun-m1 and vSrc-transformed CEFs injected with RD-SPARC-infected CEFs, i.e. without antagonizes tumorigenesis induced by these oncopro- Jun or Src, did not develop any tumor at all (data not teins, and consequently, that down-regulation of shown). A histopathological analysis of the m1 and the SPARC facilitates in vivo tumorigenesis. This view is m1+SPARC tumors was performed. These tumors reinforced by the recovery of ®ve out of six tumor lines were identically diagnosed as low-grade, highly cellular that exhibit reduced level of SPARC (including cell typical ®brosarcomas. In contrast to the m1 tumors lines in which de novo SPARC had not been re- whose cellular compartment was largely disseminated expressed). within a very abundant matrix, the m1+SPARC tumor showed a more condensed, well delimited cellular compartment, with a reduced amount of Discussion extracellular matrix (Figure 6). In a second series of experiments (Figure 5b), tumor In this paper we have con®rmed a previous report development was tested by injecting vSrc-transformed (Young et al., 1986) showing strong down-regulation CEFs with or without compensatory SPARC re- of SPARC mRNA in vSrc-transformed CEFs (480% expression. Tumorigenesis was more ecient with vSrc reduction). We have further shown that such a down- compared to vJun-m1 in that: (i) less cells were needed regulation also takes place in vJun-transformed CEFs per animal for 100% tumor induction (i.e. 0.56106 (40 ± 60% reduction) and, to an even greater extent, in cells rather than 26106 cells); and that (ii) the period CEFs transformed by vJun-m1 (480% reduction), a of latency was clearly shorter. Therefore these experi- highly tumorigenic mutant of vJun (Huguier et al., ments were conducted over a 3-week period. The 1998). This down-regulation parallels a similar reduc- presence of re-expressed SPARC had a clear-cut tion in the accumulation of the SPARC protein in the inhibitory e€ect on tumor growth because: (i) not all transformed cultures. animals (e.g. only 11 out of 12) developed a tumor; and The reduction of SPARC expression is not restricted (ii) the rate of development and the ®nal size of the to transformation by vSrc and vJun in primary avian tumors were signi®cantly reduced. The average size cells. We have shown that CEFs transformed by R- with vSrc+SPARC was only one-third the size reached Ras, an Rcas derivative that expresses an activated by vSrc-induced tumors after the 3 weeks. Even with form of the avian cRas (mutant ras 12 glu; Givol et al., such a powerful oncoprotein, SPARC displayed an 1992) also down-regulates SPARC to a level identical anti-tumoral e€ect, essentially by reducing the rate of to R-vSrc (data not shown). Recently strong repression tumor development. Histological analysis of the vSrc of SPARC mRNA has also been found in quail and vSrc+SPARC tumors did not reveal any clear embryo ®broblasts transformed by vMyc (Oberst et al., di€erence between them. All were diagnosed as 1999). In mammalian cells, independent reports also aggressive ®brosarcomas with a high mitotic index, show repression of SPARC mRNA and protein in regions of necrosis and a poor collagen matrix (Figure primary rat embryo ®broblasts which transiently 6). express cJun (Mettouchi et al., 1994) as well as in The steady-state level of the SPARC, Jun, and Src immortalized rodent ®broblasts stably transformed by proteins was analysed in cell lines expanded in vitro cJun, Ha-ras (Mettouchi et al., 1994), vAbl, vSrc from several tumors and compared to the original (Mason et al., 1986a,b), and Ki-Ras (Colombo et al., cultures before injection (Figure 7). The tumor-derived 1991). In contrast, the simian virus 40 and bovine cell lines (noted T) were numbered according to the papilloma virus 1 oncogenes exert little or no repressive chick numbers in Figure 5a,b. Western blotting e€ect on SPARC (Colombo et al., 1991; Kraemer et followed by immunodetection gave the following al., 1999). Together these data strongly suggest a results: (i) The level of SPARC was clearly reduced functional link between the transformed state as in ®ve out of six tumor-derived cultures (compare T2 established in cultured cells by certain oncoproteins with m1; T6 with m1+SPARC; T5 and T4 with vSrc; and the down-regulation of SPARC. To further test and T13 with Src+SPARC; only T17 exhibited a high this hypothesis we have introduced an exogenous level of SPARC identical to the original culture). This SPARC gene from a retrovirus in the vSrc- and reduction was also observed in cultures in which vJun-transformed CEFs, looking for a possible rever- SPARC had not been re-expressed (i.e. T2, T5 and sion of the transformed phenotype. T4); (ii) vJun accumulated at the same level in the De novo SPARC does not signi®cantly alter the original cultures transformed by vJun-m1 +/7 morphology of either non-transformed CEFs or CEFs SPARC and in tumor derived cultures (T2 and T6). transformed by vSrc and vJun. The absence of a An additional high molecular weight product of about morphological response to SPARC re-expression in 70 kDa was observed in culture T6 that might stably transfected clones from two human cell lines correspond to a Gag ± Jun fusion product; such fusion otherwise de®cient in endogenous SPARC has already events have already been described in this particular been reported (Nischt et al., 1991). In contrast, stably test (Morgan et al., 1994) and have been shown to transfected teratocarcinoma F9 stem cells overexpres- generate a more potent vJun oncoprotein; the presence sing SPARC became permanently aggregated and

Oncogene Anti-tumoral effect of SPARC in chickens E Vial and M Castellazzi 1778

Figure 6 Cross-section of tumors derived from CEFs expressing vJun-m1, vJun-m1+SPARC, vSrc, and vSrc+SPARC, as indicated. The hematoxylin-eosin-sa€ron mixture has been used to stain the cell bodies in red/violet and the extracellular matrix in orange/yellow. The bar represents 0.04 mm

Figure 7 Level of accumulation of the SPARC, vSrc, and Jun proteins from tumor-derived cell lines (noted T2, T6, etc). The numbering of the tumor-derived cell lines refers to individual birds from Figure 5a and b. Cell extracts from m1, m1+SPARC, vSrc, and vSrc+SPARC were those from the original CEF cultures before inoculation, and are independently presented in Figure 3. Western blotting was followed by detection of the accumulated proteins with speci®c antibody using the ECL detection system

Oncogene Anti-tumoral effect of SPARC in chickens E Vial and M Castellazzi 1779 rounded (Everitt and Sage, 1992). Discrepancies mitogenic e€ect of vascular endothelial growth factor between these results might re¯ect cell-type speci®cities, (Kupprion et al., 1998), a downstream target of Jun medium conditions, and/or di€erences in the level of (Kraemer et al., 1999) and Src (Mukhopadhyay et al., exogenous SPARC accumulation. 1995). Although no change in the vascularization of Re-expression of SPARC does not prevent growth in the tumors was evident (data not shown), a detailed low serum and colony formation in agar. On the histological analysis at early stages of tumor develop- contrary, exogenous SPARC enhances the growth ment would allow this to be resolved. The reduced level capacity of transformed cultures, without a€ecting the of SPARC protein in tumor-derived cultures might non-transformed cultures. It has recently been sug- have resulted from the in vivo selection of a sub- gested that SPARC might participate in the recruit- population of injected CEFs that harbor a reduced ment of growth factors to the cell surface, and thus number of RD-SPARC proviruses, and/or from host modulate cell proliferation; for example the platelet- ®broblast cells that were secondarily infected by R- derived growth factor may bind within the follistatin- vJun or R-vSrc, and not by RD-SPARC. Whatever the like structure of SPARC (Hohenester et al., 1997). It is exact reason is, it is nevertheless clear that a low level therefore possible that SPARC stimulates proliferation of SPARC favors tumor development, at least in this in agar and in low serum by facilitating the interaction particular test used, that is, the induction of local, between certain growth factors and their surface primary ®brosarcomas after subcutaneous injection of receptors. Such a mechanism might contribute to Src- or Jun-transformed CEFs. In this test there are no carcinogenesis, for example during late stages of cancer secondary, internal metastatic foci visible, and no progression with tumor cells expressing high amounts animal death within the 2 months after inoculation; of SPARC. sometimes tumors even cease to grow and regress. Re-expression of SPARC antagonizes tumorigenesis High levels of SPARC have repeatedly been reported induced by vSrc and vJun and most cell lines derived to be associated with neoplastic progression of breast from tumors display a reduced accumulation of the (Bellahcene and Castronovo, 1995) and colon cancer SPARC protein. Excess SPARC might act by inhibit- (Porte et al., 1995), and to correlate with later, more ing neovascularization of the growing tumor. This is an aggressive stages of melanoma (Ledda et al., 1997) in attractive hypothesis because SPARC has recently been humans. It is tempting to speculate that a secondary shown to directly interact with, and inhibit the enhancement of SPARC levels in the cancer cells might help the dissemination of these primary tumors, and that SPARC down-regulation and up-regulation might contribute to distinct stages in carcinogenesis. In line with this speculation, we have recently obtained some evidence that SPARC can be up- and down-regulated by distinct vJun-containing, AP1 heterodimers (Vial and Castellazzi; to be published). Moreover a biphasic regulation has been reported during transformation of rat ®broblasts, both for the cJun-controlled tenascin-C (Mettouchi et al., 1997) and for the vSrc-controlled thrombospondin-1 (Slack and Bornstein, 1994). These two extracellular proteins, like SPARC, are thought to regulate cell-matrix interaction (Sage and Bornstein, 1991). Finally, the re-expression of SPARC induces a reversion of the transformed phenotype that is only partial, and restricted to the situation in vivo.Itis likely that other components of the extracellular matrix contribute to a more complete restoration of a normal phenotype. Many genes coding for such components are known to be repressed by vSrc in CEFs. As shown in Figure 8, some of these genes are simultaneously repressed by vJun and/or by m1 (e.g. ®bronectin, thrombospondin 2, a2 (I) collagen, and collagen XII), others are speci®cally repressed by vSrc (e.g. a1 (I) collagen and lysyl oxidase) or by vJun (e.g. clusterin and decorin). Some of these genes have already been shown to individually exhibit transformation-suppres- sing activity when stably re-expressed in malignant tumor cells of mammalian origin (e.g. ®bronectin; (Akamatsu et al., 1996); decorin (Santra et al., 1995); Figure 8 Level of accumulation of mRNA encoding various thrombospondin (Sheibani and Frazier, 1995); a2 (I) extracellular matrix components from CEFs chronically infected collagen (Travers et al., 1996)). It would therefore be by the retroviruses R, R-vJun, R-vJun-m1 (noted m1), or R-vSrc worthwhile to re-express these genes in our trans- by Northern blot analysis. Arrows indicate the relevant formed CEFs, alone and in combination with SPARC, transcript(s). Radioactive probe for extracellular matrix compo- nents (i.e. ®bronectin, thrombospondin-2, a2 (I) collagen, collagen in an attempt to reveal a functional cooperation XII, lysyl oxidase, a1 (I) collagen, clusterin, decorin) and between these extracellular matrix components in GAPDH were used as indicated relation to oncogenesis.

Oncogene Anti-tumoral effect of SPARC in chickens E Vial and M Castellazzi 1780 SPARC-expressing cultures were obtained by chronic Materials and methods infection with the replication-competent retrovirus Rcas (Hugues et al., 1987). Co-infections were performed with DNA constructs Rcas vectors with two di€erent envelope speci®cities, envA The cDNA sequence encoding the secreted form of the avian and envD, and designated R and RD respectively. SPARC, noted rSPARC, has been recovered from the Routinely, transfections with RD (no insert) and RD- original plasmid rSPARC-pCR II (Bassuk et al., 1993) as a SPARC, R-vJun, R-vJun-m1, R-vSrc ts72 plasmid DNAs 928 base pair, EcoRI ± EcoRI fragment, and inserted at were performed after the ®rst passage using the DMSO- EcoRI into the pH plasmid to generated pH-rSPARC. pH is polybrene technique (Kawai and Nishizawa, 1984), and a pBluescript II SK+ derivative (Stratagene) in which the viruses were allowed to spread through the entire population SacItoSalI fragment from the original polylinker has been over the following week. Doubly infected cultures were then modi®ed (Baguet and Castellazzi; unpublished data); the new generated by superinfection of RD-derivatives with culture polylinker contains the following primer/promoter sequences, supernatant from CEFs chronically infected by R-deriva- restriction sites, and polyA sequence in the order: RP ± tives, and allowed to grow for another week. Colony BssHII ± T3 ± SP6 ± ClaI±HindIII ± PstI±SalI/ AccI±XbaI± formation in agar and growth in low serum medium were BamHI ± SmaI±SacI± EcoRI ± ClaI-polyA30 ± NotI±XhoI± performed as described (Castellazzi et al., 1990). To generate ApaI±KpnI ± T7 ± BssHII ± M13. This pH plasmid is used for cultures from tumor cells, tumoral tissues were sliced into DNA sequencing, in vitro protein synthesis, and as an small pieces and incubated overnight in regular medium adaptor plasmid replacing the CLA12 plasmids (Hugues et supplemented with collagenase H (1 mg/ml ®nal concentra- al., 1987) for introduction of the coding sequences as a ClaI± tion, Boehringer). Cells were subsequently passaged in ClaI fragment into the replication competent, retroviral medium routinely used to grow CEFs. vector Rcas at the unique ClaI site (Hugues et al., 1987). A pH-rSPARC derivative was constructed that contains the natural peptide signal sequence (Met 717 to Ala 71; see Antibodies and Western blotting Figure 1 in Bassuk et al., 1993) in front of Ala+1 from the rSPARC sequence and designated pH-SPARC sig+. To do A GST-SPARC fusion protein was prepared from bacteria this, two double-stranded, synthetic oligonucleotides linked carrying the pGEX-2T SPARC sig7 plasmid as described in together by means of an arti®cial S®I site have been inserted the `GST-gene fusion' kit (Pharmacia). Rabbits were into pH-rSPARC open at the restriction sites HindIII from immunized by repeated intradermal injection of the puri®ed the pH polylinker and BglII from the coding sequence protein following standard technique (Covalab, Lyon, (covering amino acids Glu+14 to Leu+16). The sequence of France). Preparation of cell extracts for Western blotting the HindIII ± BglII coding strand is the following: 5'- were performed as described (Castellazzi et al., 1990). For AAGCTTGCTGCAAG ATG AGA ACC TGG ATT TTC Western blotting, 10 mg protein from total cell extract was TTC TTC CTC TGC CT G GCC GCC AA G GCC CTG resolved by SDS/10%-polyacrylamide gel electrophoresis GCA GCTCCGCAAGAGGCTCTGGCTGATGAG (PAGE) and blotted onto nitrocellulose membranes that ACG GAG GTG ATT GAA GAT CT. Underlined codons were further incubated with speci®c anti-SPARC antibody. ATG, GCA, GCT and GAT encode Met717, Ala71, The peroxidase-coupled secondary anti-rabbit antibody was Ala+1, and Leu+16, respectively. The arti®cial site S®I purchased from Amersham (ECL detection system). does not modify the amino acid sequence of the peptide For enzymatic removal of SPARC N-linked carbohy- signal and is shown in italics. A pH-SPARC derivative was drates, N-glycosidase F (1.5 units; Roche) was added to also constructed that encodes the secreted form, devoid of the denatured protein extracts (10 mg) in 10 ml of 100 mM peptide signal, designated pH-SPARC sig7, with a BamHI Na2HPO4/NaH2PO4 pH 8, 30 mM EDTA, 1% Triton X- site in front of the ATG. To do this, a PCR product was 100, 1.5% SDS, 10 mM O-phenanthroline, 1% b-mercap- cloned into pH-rSPARC open at the restriction sites BamHI toethanol and incubated overnight at 378C. Samples from the polylinker and SphI from the coding sequence without enzyme were treated identically to samples contain- (covering amino acids Arg+144 to Arg+146). PCR sense ing N-glycosidase F. The oligosaccharide removal was and anti-sense primers were 5'-CTGCTCAGA GGA TCC assessed by SDS ± PAGE. GGT ATG GCT CCG CAA GAG GC and 5'-GCA CAT Commercial mouse monoclonal anti-SPARC and anti-Src TCT TCA GCC AGT CCC GCA TGC GCA GG respectively. were purchased from Haematologic Technologies Inc. (Essex, Restriction sites for cloning are underlined. pH-SPARC sig7 VT, USA; cat no. AON-5031), and from Upstate Biotechnol- was used for in-frame cloning of the secreted form as a ogy (Lake Placid, USA; cat no. 05-185), respectively. Rabbit BamHI ± EcoRI fragment into pGEX-2T (Pharmacia) to anti-Jun antibody that recognize both avian cJun and vJun generate a glutathione S-transferase (GST)-SPARC sig7 were generated against a GST-vJun fusion protein (Baguet fusion protein for antibody production. The vJun coding and Castellazzi; unpublished data). sequence was introduced in-frame into pGEX-2T as a BamHI ± EcoRI for antibody production. In all cases changes in the pH derivatives were con®rmed by DNA sequencing. Northern blotting The vSrc ts72 coding sequence was recovered as a 1674 base Total RNA was extracted essentially according to a pair, NcoI±NruI fragment from the pXD72 plasmid (Mayer et published procedure (Piu et al., 1993). Brie¯y, exponentially al., 1986), inserted into the CLA12 NCO adaptor plasmid growing cells were lysed in GET solution (5 M guanidium (Hugues et al., 1987) between NcoI and SmaI, to generate isothyocyanate, 50 mM Tris pH 7.5, 10 mM EDTA pH 8, CLA12 NCO vSrc ts72 for subsequent cloning into Rcas. Rcas 0.1 mM b mercaptoethanol), then precipitated overnight at envA viruses expressing vJun and vJun-m1 have already been 08C after addition of eight volumes of LiCl 4 M. After described (Huguier et al., 1998; van Dam et al., 1998). An Rcas centrifugation for 1 h at 48C , the pellet was resuspended in envD virus expressing SPARC sig+ was also generated. TESP solution (10 mM Tris pH 7.5, 1 mM EDTA pH 8, 0.1% SDS, 5% phenol), phenol/chlorophorm extracted Cell culture twice, then ethanol precipitated, and ®nally resuspended in 10 mM Tris pH 8, 10 mM EDTA, 1% SDS. Northern Primary CEF cultures were prepared from 8-day-old C/E blot analysis was performed using conventional methods. SPAFAS chicken embryos (Merial, Lyon, France) and Radioactive probes were coding sequences of avian SPARC grown in normal medium supplemented with 6% serum as and glyceraldehyde 3-phosphate dehydrogenase gene described (Castellazzi et al., 1990). vJun-, vSrc-, and (GAPDH).

Oncogene Anti-tumoral effect of SPARC in chickens E Vial and M Castellazzi 1781 Histological study Fondation pour la Recherche Me dicale (to E Vial). It was also funded by grants from the Association pour la Tumors were ®xed in Bouin's solution and embedded in Recherche sur le Cancer, from the Mutuelle Ge ne rale de paran. Consecutive 4 mm-thick sections were cut. Evalua- l'Education Nationale, as well as from the European tion of in¯ammation, necrosis, and ®brosis was performed Economic Community Biomedicine and Health Program using a conventional hematoxylin-eosin-sa€ron staining (Biomed II contract no BMH4 CT98 3505 to M Castel- followed by light microscope examination. lazzi). We thank JoeÈ l Baguet for the construction of the pH, GST-SPARC, and GST-vJun plasmids and for the recovery of the fusion proteins. We also thank Germain Abbreviations Gillet for the gift of the monoclonal anti-Src antibody. We The abbreviations used are: AP1, activating protein 1; thank the following persons for the gift of various avian ASV17, avian sarcoma virus 17; CEF, chicken embryo cDNA probes: Brigitt Hogan for SPARC, Gilbert Brun for ®broblast; SPARC, secreted protein, acidic and rich in clusterin, Beat Trueb for collagen XII, Sherrill L Adams cysteine. for a2 (I) collagens, Jack Lawler for thrombospondin 2, and Richard O Hynes for ®bronectin. We are indebted to Sylviane Guerret (Novotec, Lyon, France) and MicheÁ le Lechevallier (Laboratoire Marcel Me rieux, Lyon, France) Acknowledgments for the processing of the tumor samples, and the This work was supported by fellowships from the MinisteÁ re histopathological diagnosis, respectively. We also thank de la Recherche et de la Technologie and from the Edmond Derrington for critical reading of the manuscript.

References

Akamatsu H, Ichihara-Tanaka K, Ozono K, Kamiike W, Mason IJ, Murphy D, Munke M, Francke U, Elliott RW and Matsuda H and Sekiguchi K. (1996). Cancer Res., 56, Hogan BL. (1986a). EMBO J., 5, 1831 ± 1837. 4541 ± 4546. Mason IJ, Taylor A, William JG, Sage EH and Hogan BLM. Bassuk JA, Iruela Arispe ML, Lane TF, Benson JM, Berg (1986b). EMBO J., 5, 1465 ± 1472. RA and Sage EH. (1993). Eur. J. Biochem., 218, 117 ± 127. Mayer BJ, Jove R, Krane JF, Poirier F, Calothy G and Bellahcene A and Castronovo V. (1995). Am.J.Pathol.,146, Hanafusa H. (1986). J. Virol., 60, 858 ± 867. 95 ± 100. Mettouchi A, Cabon F, Montreau N, Dejong N, Vernier P, Bos TJ, Monteclaro FS, Mitsunobu F, Ball ARJ, Chang Gherzi R, Mercier G and Bine truy B. (1997). Mol. Cell. CHW, Nishimura T and Vogt PK. (1990). Genes Dev., 4, Biol., 17, 3202 ± 3209. 1677 ± 1687. Mettouchi A, Cabon F, Montreau N, Vernier P, Mercier G, Bradshaw AD, Francki A, Motamed K, Howe C and Sage Blangy D, Tricoire H, Vigier P and Bine truy B. (1994). EH. (1999). Mol. Biol. Cell, 10, 1569 ± 1579. EMBO J., 13, 5668 ± 5678. Castellazzi M, Dangy JP, Mechta F, Hirai SI, Yaniv M, Morgan IM, Havarstein IS, Wond WY, Luu P and Vogt PK. Samarut J, Lassailly A and Brun G. (1990). Oncogene, 5, (1994). Oncogene, 9, 2793 ± 2797. 1541 ± 1547. Mukhopadhyay D, Tsiokas L, Zhou XM, Foster D, Brugge Colombo MP, Ferrari G, Biondi G, Galasso D, Howe CC JS and Sukhatme VP. (1995). Nature, 375, 577 ± 581. and Parmiani G. (1991). Eur. J. Cancer, 27, 58 ± 62. Muramaki M, Sonobe MH, Ui M, Kabuyama Y, Watanabe Dziadek M, Paulsson M, Aumailley M and Timpl R. (1986). H, Wada T, Handa H and Iba H. (1997). Oncogene, 14, Eur. J. Biochem., 161, 455 ± 464. 2435 ± 2444. Everitt EA and Sage EH. (1992). Exp. Cell Res., 199, 134 ± Nischt R, Pottgiesser J, Krieg T, Mayer U, Aumailley M and 146. Timpl R. (1991). Eur. J. Biochem., 200, 529 ± 536. Funk SE and Sage EH. (1993). J. Cell. Physiol., 154, 53 ± 63. Oberst C, Hartl M, Weiskirchen R and Bister K. (1999). Givol I, Greenhouse JJ, Hughes SH and Ewert DL. (1992). Virology, 253, 193 ± 207. Oncogene, 7, 141 ± 146. PiuF,JurdicP,BrunG,SamarutJandCastellazziM. Honenester E, Maure P and Timpl R. (1997). EMBO J., 16, (1993). C.R. Acad. Sci. Paris, 316, 772 ± 779. 3778 ± 3786. Porte H, Chastre E, Prevot S, Nordlinger B, Empereur S, Hugues SH, Greenhouse CJ, Petropoulos CJ and Sutrave PS. Basset P, Chambon P and Gespach C. (1995). Int. J. (1987). J. Virol., 61, 3004 ± 3012. Cancer, 64, 70 ± 75. Huguier S, Baguet J, Perez S, van Dam H and Castellazzi M. Reed MJ and Sage EH. (1996). Curr. Top. Microbiol. (1998). Mol. Cell. Biol., 18, 7020 ± 7029. Immunol., 213, 81 ± 94. Jove R and Hanafusa H. (1987). Annu. Rev. Cell Biol., 3, 31 ± Sage EH and Bornstein P. (1991). J. Biol. Chem., 266, 56. 14831 ± 14834. Karin M, Liu ZG and Zandi E. (1997). Curr. Opin. Cell Biol., Santra M, Skorski T, Calbretta B, Lattime EC and Iozzo 9, 240 ± 246. RV. (1995). Proc. Natl. Acad. Sci. USA, 92, 7016 ± 7020. Kawai S and Nishizawa M. (1984). Mol. Cell. Biol., 4, 1172 ± Sheibani N and Frazier WA. (1995). Proc. Natl. Acad. Sci. 1174. USA, 92, 6788 ± 6792. Kraemer M, Tournaire R, Dejong V, Montreau N, Briane D, Slack JL and Bornstein P. (1994). Cell Growth Di€er., 5, Derbin C and Bine truy B. (1999). Cell Growth Di€er., 10, 1373 ± 1380. 193 ± 200. Suzuki T, Murakami M, Onai N, Fukuda E, Hashimoto Y, Kupprion C, Motamed K and Sage EH. (1998). J. Biol. Sonobe MH, Kameda T, Ichinose M, Miki K and Iba H. Chem., 273, 29635 ± 29640. (1994). J. Virol., 68, 3527 ± 3535. Lankat-Buttgereit B, Mann K, Deutzmann R, Timpl R and Termine JD, Kleiman HK, Whitson SW, Conn KM, Krieg T. (1998). FEBS Letts., 236, 352 ± 356. McGarvery ML and Martin GR. (1981). Cell, 26, 99 ± 105. Ledda F, Bravo AI, Adris S, Bover L, Mordoh J and Timpl R and Aumailley M. (1993). Molecular and Cellular Podhajcer OL. (1997). J. Invest. Dermatol., 108, 210 ± 214. Aspects of Basement Membranes. Rohrbach DH and Maki Y, Bos TJ, Davis C, Starbuck M and Vogt PK. (1987). Timpl R. (eds). Academic Press, Inc., pp. 211 ± 235. Proc. Natl. Acad. Sci. USA, 84, 2848 ± 2852.

Oncogene Anti-tumoral effect of SPARC in chickens E Vial and M Castellazzi 1782 Travers H, French NS and Norton JD. (1996). Cell Growth Vogt PK. (1994). The Fos and Jun Families of Transcription Di€er., 7, 1353 ± 1360. Factors. CRC Press: Boca Raton, FL, USA, pp. 203 ± 219. van Dam H, Huguier S, Kooistra K, Baguet J, Vial E, van Young MF, Bolander ME, Day AA, Ramis CI, Robey PG, der Eb AJ, Herrlich P, Angel P and Castellazzi M. (1998). Yamada Y and Termine JD. (1986). Nucleic Acids Res., Genes Dev., 12, 1227 ± 1239. 14, 4483 ± 4497.

Oncogene