- Email: [email protected]
Differentiation and antiproliferation effects of retinoic acid receptor β in hepatoma cells
Cancer Letters 124 (1998) 205–211
Differentiation and antiproliferation effects of retinoic acid receptor b in hepatoma cells Chen Li, Yu-Jui Yvonne Wan* Department of Pathology, Harbor-UCLA Medical Center, 1000 West Carson Street, Torrance, CA 90509, USA Received 8 October 1997; received in revised form 17 October 1997; accepted 31 October 1997
Abstract Evidence indicates that the retinoic acid receptor b (RARb) gene might be a tumor suppressor gene. Previously, we have shown that the expression of the RARb gene was either inhibited or downregulated in tumorigenic hepatoma cell lines such as McA-RH8994. McA-RH8994 cells expressed RARa and g and three types of retinoid X receptor (RXRa, b and g), but not RARb mRNA. To further analyze the molecular mechanisms which might account for RARb gene inactivation, the rat RARb gene promoter was cloned from McA-RH8994 cells and no mutation was detected. By transient transfection, McA-RH8994 cells contained the necessary factors to activate the RARb gene. To study the possible roles of RARb in hepatoma cells, the expression of the RARb gene was restored in McA-RH8994 cells by stable transfection. A RARb positive cell line named McA-RH8994b was characterized. The results demonstrated that expression of the RARb gene resulted in increased sensitivity of the hepatoma cells to the antiproliferative effect of retinoic acid (RA). Furthermore, expression of RARb resulted in a spontaneous differentiation of the hepatoma cells. These data indicate that RARb plays important roles in differentiation and antiproliferation. 1998 Elsevier Science Ireland Ltd. Keywords: Retinoic acid; Retinoic acid receptor; Hepatoma; Differentiation; Antiproliferation
1. Introduction Retinoic acid (RA) regulates essentially all important biological processes including growth, development, differentiation, reproduction, morphogenesis, metabolism and homeostasis [1]. RA exerts its effects by modulating gene expression through its receptors. The highly pleiotropic effects of RA can be explained in part by the multiplicity of retinoic acid receptors (RARs) and retinoid X receptors (RXRs). The a, b and g subtypes of both RARs and RXRs have distinct * Corresponding author. Tel.: +1 310 2223876; fax: +1 310 7826649; e-mail: [email protected]
and conserved sequence domains and exhibit distinct ligand-binding properties [2–4]. The genes coding for RARa and RARb are localized in chromosomal regions in which deletions frequently occur in a number of tumor types (17q21 and 3p24, respectively) [5]. Such chromosomal deletions are generally regarded as strong indicators for the presence of a tumor suppressor gene in that region. Because of the important role of these receptors in inducing differentiation and inhibiting growth, the RAR coding genes are regarded as potential tumor suppressor genes [5]. In breast cancer, deletions and allele losses frequently are observed at chromosomes 3p and 17q [6,7], which could affect RARa and b
0304-3835/98/$19.00 1998 Elsevier Science Ireland Ltd. All rights reserved PII S0304-3835 (97 )0 0475-8
206
C. Li, Y.-J.Y. Wan / Cancer Letters 124 (1998) 205–211
expression. In epidermoid lung cancer, a deletion is detected at 3p where RARb is coded [8]. When RARb mRNA was expressed in lung cancer cells, the cells became less tumorigenic both in morphology and behavior [8]. In hepatoma, the RARb gene was found to be a hepatitis B viral DNA integration site [9]. In hepatoma cell lines, the expression of the RARb gene is either inhibited or reduced [10]. Previously, we have shown that only the RARb gene, and not the RARa and g genes, is inactivated in McA-RH8994 and McA-RH7777 rat hepatoma cells. Expression of the RARb gene is also reduced in other hepatoma cell lines including H4IIE and HepG2 [10]. Since downregulation of the RARb gene is also found in lung cancer cells, head and neck tumors and mammary tumors [11–13] and upregulation of the RARb gene is usually associated with RA-induced differentiation [10,14,15], the RARb gene might also play an important role in hepatocarcinogenesis. In this study, we further investigate the possible molecular mechanisms underlying RARb gene inactivation and the role of RARb in hepatoma cells.
2. Materials and methods
CAGCTCACTTCC-3′ (+39 to +16) [20]. Nucleotides indicating HindIII and XbaI restriction endonuclease cut sites are shown in italics. PCR was performed as described previously [10]. PCR products were digested with XbaI and HindIII and subsequently subcloned into the promoterless pCAT-Enhancer (Promega, Madison, WI), which gave the rRARb-CAT construct. DNA sequencing was performed by Sanger’s dideoxy sequencing method [21]. 2.3. Establishment of the McA-RH8994b cell line The McA-RH8994 cell line was obtained from American Type Culture Collection (Rockville, MD) [22]. Stable transfection was performed using a standard calcium phosphate precipitation method [23]. Cells were exposed to pSV2-neo and RARb2 expression plasmids PECE-RARb [24] at a 1/5 molar ratio for 6 h and then selected by G418 (GIBCO, Bethesda, MD) at a final concentration of 500 mg/ml. After 30 days of treatment, neomycin-resistant clones were selected and expanded individually. The expression of the RARb gene was analyzed by Northern blot hybridization. For transient transfection assay, cells were harvested 72 h after the initiation of transfection. CAT activity was assayed by the phase extraction method [25] and corrected by b-gal activity.
2.1. Northern blot hybridization Northern blot hybridization was performed according to the method described previously [11]. The probes used in the hybridization procedure included a full-length cDNA insert of B1-RARb [16], KpnI fragment of RXRa (∼1.6 kb), 5′ end to AccI site of RXRb (∼1.5 kb), 5′ end to HindIII site of RXRg [17], HindIII fragment of albumin [18] and PstI fragment of PRAF6 [19] for detecting AFP mRNA. The membranes were rehybridized with rat b actin probes for RNA quantitation purposes. 2.2. PCR amplification and DNA sequence analysis Genomic DNA was prepared from McA-RH8994 and male Buffalo rat livers and then amplified with primers complementary to sequences flanking the promoter region of the mouse RARb gene, including 5′cccaagcttGTGAGAATCCTGGGAGTTGGTGA-3′ (−120 to −97) and 5′-attctagaCCTGCCTCGGAG-
3. Results 3.1. Expression of the RXR genes in the McARH8994 cell line Previously, we have shown that by Northern blot hybridization and reverse transcription PCR, McARH8994 and McA-RH7777 hepatoma cells expressed RARa and g mRNA, but not RARb mRNA [10]. To examine the role of RXRs in hepatoma, the expression of the RXRa, b and g genes was examined by Northern blot hybridization (Fig. 1). McA-RH8994 and McA-RH7777 cells expressed 5.6 kb RXRa mRNA and 2.7 and 3.0 kb RXRb mRNA, respectively. Different RXRg transcripts were detected in these two hepatoma cell lines. McA-RH8994 cells expressed a single RXRg transcript of 2.3 kb and McA-RH7777 cells expressed two RXRg transcripts of 2.3 and 3.8 kb. The hybridization was performed under stringent
C. Li, Y.-J.Y. Wan / Cancer Letters 124 (1998) 205–211
207
Fig. 1. Expression of RXR mRNAs in McA-RH8994 and McA-RH7777 rat hepatoma cells. RNA was extracted from McA-RH8994 (1) and McA-RH7777 (2) cells. Total RNA (20 mg) was fractionated by formaldehyde agarose gel, electroblotted onto nylon membrane and then hybridized with 32P-labeled RXRa, b and g and b-actin cDNA probes. Arrows indicate the specific transcripts.
conditions. Thus, the RARb gene is the only RA receptor gene inactivated in these hepatoma cell lines. 3.2. Analysis of the promoter of RARb2 in McARH8994 cells Human and mouse RARb genes have been cloned and sequenced [26,27]. However, the rat RARb gene has never been studied. Since the mouse RARb cDNA clone can be used to hybridize rat RARb mRNA, this suggests that they are highly homologous to each other. To unravel the molecular mechanism involved in inactivation of the RARb gene, the promoter of RARb2/4 was cloned from normal adult livers and McA-RH8994 cells by PCR. The mouse RARb2 sequence was used to synthesize the primers. The results indicated that the RARb2 promoter sequence was identical in McA-RH8994 cells and rat and mouse livers from nucleotides −96 to +15 (data not shown). Within the promoter, it contains the RA response element (bRARE), as reported in the mouse gene [28]. 3.3. The effect of RA on the RARb gene in McARH8994 cells Unlike many other systems, the expression of the RARb gene cannot be induced by RA in McARH8994 and McA-RH7777 cells [10]. RA-induced
RARb expression is mediated through the bRARE present in the RARb promoter [28]. To test if inactivation of RARb expression is due to impaired transcriptional regulation of the RARb gene, reporter constructs containing either promoter region (−96 to +15, rRARb-CAT) or the entire regulatory region (−2800 to +480, RARb2-H3-CAT) of the RARb gene were used for transient transfection assays. The data demonstrated that in McA-RH8994 cells, both constructs responded to RA (Fig. 2). There were seven- and two-fold inductions when RARb2-H3CAT and rRARb-CAT were used as reporter genes,
Fig. 2. RARb gene activity in McA-RH8994 cells. CAT activity was measured in McA-RH8994 cells transiently transfected with rRARb-CAT and RARb2-H3-CAT and treated with RA (10−5 M) for 64 h. pCAT-Enhancer (pCAT-E, Promega) was included as a control plasmid. The data presented are averages of three experiments.
208
C. Li, Y.-J.Y. Wan / Cancer Letters 124 (1998) 205–211
Fig. 3. Expression of the RARb gene in McA-RH8994b cells. The expression of RARb mRNA was studied in rat livers and McA-RH8994 and McA-RH8994b cells treated with or without RA (10−5 M) for 3 days by Northern blot hybridization. Ethidium bromide staining of ribosomal RNA demonstrated that each lane had a similar amount of RNA. The arrow indicates the specific transcript.
respectively. These data suggest that besides the promoter, there are other cis-acting elements involved in RA-mediated RARb gene regulation and McARH8994 cells contain the necessary factors to transactivate the RARb gene. 3.4. Analysis of the role of RARb in McA-RH8994 cells
Fig. 4. Analysis of the function of RARb in McA-RH8994b cells. McA-RH8994 and McA-RH8994b cells were transiently transfected by rRARb-CAT, treated with or without RA (10−6 and 10−5 M) (a) and CD2497 (2 × 10−7 M) (b) for 48 h and then CAT assays were performed. The data presented are averages of three experiments.
To analyze the role of RARb in McA-RH8994 cells, RARb was restored in McA-RH8994 cells by stable transfection. Twenty clones were screened for RARb mRNA expression. Among them, 12 expressed detectable amounts of RARb mRNA of variable sizes. One clone, designated as McA-RH8994b, was selected for further characterization because this cell line and adult rat livers expressed similar amounts and sizes of RARb mRNA. It is unlikely that the establishment of the McA-RH8994b clone was due to the selection of McA-RH8994 cells since RARb mRNA could not be detected in McA-RH8994 cells at all by RT-PCR [10]. As shown in Fig. 3, rat livers expressed 3.1 and 2.8 kb RARb transcripts. No RARb mRNA was detected in McA-RH8994 cells. McA-RH8994b cells expressed a single RARb transcript of 2.8 kb
C. Li, Y.-J.Y. Wan / Cancer Letters 124 (1998) 205–211
209
Fig. 5. The expression and regulation of the AFP and albumin genes in McA-RH8994 and McA-RH8994b cells. McA-RH8994 (b−) and McARH8994b (b+) cells were grown in the absence or presence of RA (10−6 M) for 3 days. Northern blot hybridization was performed to examine the expression of the AFP (a), albumin (b) and rRNA (c) genes. Arrows indicate the specific transcripts.
since no alternative splicing and/or differential usage of promoter would occur. In addition, RARb mRNA in McA-RH8994b cells could not be induced by RA because the RARb expression plasmid used in transfection does not contain bRARE. To ensure that RARb expressed in McA-RH8994b cells was functioning, RARb (+) and RARb (−) McARH8994 cells were transiently transfected with rRARb-CAT. These cells were then treated with or without all-trans-RA (RA) at 10−6 and 10−5 M and RARb selective ligand (CD2497) at 2 × 10−7 M. At 2 × 10−7 M, CD2497 maintains its RARb selectivity
Fig. 6. The antiproliferative effect of RA on McA-RH8994 and McA-RH8994b cells. McA-RH8994 and McA-RH8994b cells were seeded in 10-cm culture dishes and treated with or without RA (10 − 6 and 10 − 5 M) for 3 days. The cell number was then determined for the triplicated samples. The data presented are averages of three experiments.
[29]. The data indicated that the basal CAT activity and the fold increase of CAT activity mediated by RA were higher in RARb (+) than in RARb (−) cells (Fig. 4a). In RARb (−) cells, CAT activity was not induced by RARb selective ligand CD2497. In contrast, CD2497 increased CAT activity in RARb (+) cells (Fig. 4b). These data indicate that RARb is actually expressed and functioning in McA-RH8994b cells. Previously, we have demonstrated that RA induced the expression of the AFP and albumin genes in McARH8994 cells, which resulted in a phenotype similar to well differentiated fetal hepatocytes [30]. The phenotype of McA-RH8994b cells was also analyzed by examining the expression and regulation of the AFP and albumin genes by Northern blot hybridization. RARb (+) and (−) cell lines were treated with or without RA (10−6 M) for 3 days. Compared with the RARb (−) cell line, McA-RH8994b cells expressed high basal levels of AFP as well as albumin mRNA, which was similar to the phenotype of RA-treated McA-RH8994 cells. In addition, the levels of both AFP and albumin mRNA were increased upon RA treatment (Fig. 5). The data suggest that expression of the RARb gene results in spontaneous differentiation of hepatoma cells. As demonstrated in different cell systems, RA profoundly inhibits the growth of tumor cells [1]. The antiproliferative effect of RA was also tested on RARb (+) and (−) hepatoma cells. McA-RH8994 and McA-RH8994b cells were treated with RA for 3
210
C. Li, Y.-J.Y. Wan / Cancer Letters 124 (1998) 205–211
days and then the cell number and total protein were determined by cell counting and Bio-Rad protein assay kit, respectively. Fig. 6 shows that RA had no significant effect on the growth inhibition of McARH8994 cells. However, RA (10−5 M) inhibited the growth of McA-RH8994b cells by 60%. Protein data were similar to the cell count data (data not shown). At 10−5 M, RA was not toxic to the cells since no significant amount of dead cells was detected by the Trypan Blue exclusion test. These results indicate that expression of RARb not only changes the hepatoma cells to a more differentiated phenotype, but also enhances the sensitivity of the cells to the antiproliferative effect of RA.
4. Discussion Besides hepatoma cells, inactivation of the RARb gene is found in lung cancer cells, head and neck tumors and mammary tumors [11–13]. Most of the neoplastic cells and tissues that express little or no RARb mRNA still express RARa and g and at least one of the RXRs, which might mediate the transactivation of retinoid responsive genes. This observation raises the possibility that RARb regulates specific genes that are important for suppression of carcinogenesis. This notion is supported by our findings and those of others which show that expression of RARb increases the antiproliferative effect of RA. Furthermore, overexpression of RARb decreases the tumorigenicity of human lung cancer cells [11], transient transfection of the RARb gene results in RA-dependent suppression of cell proliferation [31] and loss of RARb expression in vivo might play an important role in the expansion of premalignant clones for oral lesions [12]. The mechanisms underlying the loss of RARb mRNA in cancer cells including hepatoma are not clear. In McA-RH8994 and McA-RH7777 cells, we have shown that no RARb gene rearrangement is detected [10]. In addition, no mutation is present in the RARE of the RARb2 promoter. By transient transfection, RA was able to activate the RARb gene promoter in McA-RH8994 cells. Furthermore, when the CAT gene was driven by the full-length RARb gene regulatory region, the RA-induced CAT activity was further increased. These data indicate that McA-
RH8994 cells contain the necessary factors to transactivate the RARb gene and the sequence beyond the RARb promoter is also involved in RA-mediated RARb gene activation. Therefore, one possible cause for the reduction in the expression of the RARb gene in hepatoma cells is the presence of mutation in other regulatory regions of the RARb gene which we did not examine. Since we did not examine the rate of transcription of the endogenous RARb gene in these cells from both RARb promoters, it is also possible that the gene is transcribed. However, there is a problem with the production of a full-length pre-RARb mRNA or processing of the RARb mRNA and thus RARb mRNA was not detected in the cells. We have shown that RA induces the expression of AFP and albumin mRNA in McA-RH8994 cells and results in a phenotype similar to well differentiated fetal hepatocytes [30]. Since the RARb gene is inactivated in McA-RH8994 cells, the induction of the AFP and albumin genes is not mediated directly through RARb. However, this should not be interpreted to mean that RARb is not important for differentiation. In fact, our data have shown that RARb causes spontaneous differentiation of hepatoma cells, suggesting that when RARb is absent, other types of RA receptors can also mediate the differentiation effect of RA. However, when RARb is present, the hepatoma cells can undergo spontaneous differentiation. In other words, the differentiation effect of RA receptors might be redundant and RARb does have a differentiation action. In addition, RARb might play a unique role in antiproliferation since expression of RARb increases the sensitivity of McA-RH8994 cells to the antiproliferative effect of RA. However, since RARs have been found to be functionally redundant [4], it is also possible that a similar response will be obtained if RARa and g are overexpressed in the McA-RH8994 cells.
Acknowledgements This work was supported by NIH grant CA53596. We thank Dr R. Evans for the RXR cDNA clones and mRARb2-H3-CAT, Dr M. Pfahl for the RARb cDNA clones and Drs U. Reichert and B. Shroot for CD2497. We also thank Ms A. Flores for assistance in the preparation of this manuscript.
C. Li, Y.-J.Y. Wan / Cancer Letters 124 (1998) 205–211
References [1] M.B. Sporn, A.B. Robers, D.S. Goodman (Eds.), The Retinoids, second ed., Raven Press, New York, 1994. [2] D.J. Mangelsdorf, C. Thummel, M. Beato, P. Herrlich, G. Schutz, K. Umesono, B. Blumberg, P. Kastner, M. Mark, P. Chambon, R.M. Evans, The nuclear receptor superfamily: the second decade, Cell 83 (1995) 835–839. [3] D.J. Mangelsdorf, R.M. Evans, The RXR heterodimers and orphan receptors, Cell 83 (1995) 841–850. [4] P. Chambon, A decade of molecular biology of retinoic acid receptors, FASEB J. 10 (1996) 940–954. [5] R. Lotan, J.L. Clifford, Nuclear receptors for retinoids: medicators of retinoid effects on normal and malignant cells, Biomed. Pharmacother. 45 (1991) 145–156. [6] I.U. Ali, R. Lidereau, R. Callahan, Presence of two members of c-erb A receptor gene family (c-erb Ab and c-erb A2) in the smallest region of somatic homozygosity on chromosome 3p21–p25 in human breast carcinoma, J. Natl. Cancer Inst. 81 (1989) 1815–1820. [7] J.M. Hall, M.K. Lee, B. Newman, J.E. Morrow, L.A. Anderson, B. Huey, M.C. King, Linkage of early onset familial breast cancer to chromosome 17q21, Science 250 (1990) 1684–1689. [8] F.J. Gebert, N. Moghal, J.V. Frangioni, D.J. Sugarbaker, High frequency of retinoic acid receptor b abnormalities in human lung cancer, Oncogene 6 (1991) 1859–1868. [9] H. de The, A. Marchio, P. Tiollais, A. Dejean, A novel steroid thyroid hormone receptor-related gene inappropriately expressed in human hepatocellular carcinoma, Nature 330 (1987) 667–670. [10] Y.-J.Y. Wan, L. Wang, T.-C.J. Wu, Expression of retinoic acid receptor genes in developing rat liver and hepatoma cells, Lab. Invest. 66 (1992) 646–651. [11] B. Houle, C. Rochette-Egly, W.E.C. Bradley, Tumor-suppressive effect of the retinoic acid receptor b in human epidermoid lung cancer cells, Proc. Natl. Acad. Sci. USA 90 (1993) 985–989. [12] R. Lotan, X.-C. Xu, S.M. Lippman, J.Y. Ro, I.S. Lee, J.J. Lee, W.K. Hong, Suppression of retinoic acid receptor b in premalignant oral lesions and its up-regulation by isotretinoin, N. Engl. J. Med. 332 (1995) 1405–1410. [13] K. Swisshelm, K. Ryan, X. Lee, H.C. Tsou, M. Peacocke, R. Sager, Down-regulation of retinoic acid receptor b in mammary carcinoma cell lines and its up-regulation in senescing normal mammary epithelial cells, Cell Growth Differ. 5 (1994) 133–141. [14] H. de The, A. Marchio, P. Tiollais, A. Dejean, Differential expression and ligand regulations of the retinoic acid receptor a and b genes, EMBO J. 8 (1989) 429–433. [15] Y.-J.Y. Wan, L. Wang, T.-C.J. Wu, Different response to retinoic acid of two teratocarcinoma cell lines, Exp. Cell Res. 219 (1995) 392–398. [16] D. Benbrook, E. Lernhardt, M. Pfahl, A new retinoic acid receptor identified from a hepatocellular carcinoma, Nature 333 (1988) 669–672. [17] D.J. Mangelsdorf, U. Borgmeyer, R.A. Heyman, J.Y. Zhou,
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
211
E.S. Ong, A.E. Oro, A. Kakizuka, R.M. Evans, Characterization of three RXR genes that mediate the action of 9-cis retinoic acid, Genes Dev. 6 (1992) 329–344. T.D. Sargent, M. Yang, J. Bonner, Nucleotide sequence of cloned rat serum albumin, Proc. Natl. Acad. Sci. USA 78 (1981) 243–246. L.L. Jagodzinski, T.D. Sargent, M. Yang, C. Glackin, J. Bonner, Sequence homology between RNAs encoding rat a-fetoprotein and rat serum albumin, Proc. Natl. Acad. Sci. USA 78 (1981) 3521–3525. A. Zelent, C. Mendelsohn, P. Kastner, A. Krust, J.M. Garmier, F. Ruffenach, P. Leroy, P. Chambon, Differentially expressed isoforms of the mouse retinoic acid receptor b are generated by usage of two promoters and alternative splicing, EMBO J. 10 (1991) 71–81. F. Sanger, S. Nicklen, A.R. Loulson, DNA sequencing with chain-terminating inhibitors, Proc. Natl. Acad. Sci. USA 74 (1977) 5463–5467. J.E. Becker, B. De Nechaud, V.R. Potter, Two new rat hepatoma cell lines for studying the unbalanced ontogeny hypothesis, in: W.H. Fishman, S. Sell (Eds.), Oncodevelopmental Gene Expression, Academic Press, New York, 1976, p. 259. C. Chen, H. Okayama, Calcium phosphate-mediated gene transfer: a highly efficient system for stable transforming cells with plasmid DNA, Biotechniques 6 (1988) 632–638. G. Graupner, K.N. Wills, M. Tzukerman, X.K. Zhang, M. Pfahl, Dual regulatory role for thyroid-hormone receptors allows control of retinoic-acid receptor activity, Nature 340 (1989) 653–656. D.A. Nielson, T.C. Chang, D.J. Shapiro, A highly sensitive mixed phase assay for chloramphenicol acetyltransferase activity in transfected cells, Anal. Biochem. 179 (1989) 19–23. A. Zelent, A. Krust, M. Petkovich, P. Kastner, P. Chambon, Cloning of murine retinoic acid receptor a and b cDNAs and of a novel third receptor g predominantly expressed in skin, Nature 339 (1989) 714–717. N. Brand, M. Petkovich, A. Krust, P. Chambon, H. de The, A. Marchio, P. Tiollais, Identification of a second human retinoic acid receptor, Nature 332 (1988) 850–853. H.M. Sucov, K.K. Murakami, R.M. Evans, Characterization of an autoregulated response element in the mouse retinoic acid receptor type b gene, Proc. Natl. Acad. Sci. USA 87 (1990) 5392–5396. B.A. Bernard, J.M. Bernardon, C. Delesduse, B. Martin, M.C. Lenoir, J. Maignan, B. Charpentier, W.R. Pilgrim, U. Reichert, B. Shroot, Identification of synthetic retinoids with selectivity for human nuclear retinoic acid receptor g, Biochem. Biophys. Res. Commun. 186 (1992) 977–983. Y.-J. Wan, T.-C. Wu, The effects of retinoic acid on the expression of a-fetoprotein and albumin genes in rat hepatoma cell lines, Differentiation 50 (1992) 107–111. J.U. Frangioni, N. Moghal, A. Stuart-Tilley, B.G. Neel, S.L. Alper, The DNA binding domain of retinoic acid receptor b is required for ligand-dependent suppression of proliferation: application of general purpose mammalian co-expression vectors, J. Cell Sci. 107 (1994) 827–838.
Differentiation and antiproliferation effects of retinoic acid receptor b in hepatoma cells Chen Li, Yu-Jui Yvonne Wan* Department of Pathology, Harbor-UCLA Medical Center, 1000 West Carson Street, Torrance, CA 90509, USA Received 8 October 1997; received in revised form 17 October 1997; accepted 31 October 1997
Abstract Evidence indicates that the retinoic acid receptor b (RARb) gene might be a tumor suppressor gene. Previously, we have shown that the expression of the RARb gene was either inhibited or downregulated in tumorigenic hepatoma cell lines such as McA-RH8994. McA-RH8994 cells expressed RARa and g and three types of retinoid X receptor (RXRa, b and g), but not RARb mRNA. To further analyze the molecular mechanisms which might account for RARb gene inactivation, the rat RARb gene promoter was cloned from McA-RH8994 cells and no mutation was detected. By transient transfection, McA-RH8994 cells contained the necessary factors to activate the RARb gene. To study the possible roles of RARb in hepatoma cells, the expression of the RARb gene was restored in McA-RH8994 cells by stable transfection. A RARb positive cell line named McA-RH8994b was characterized. The results demonstrated that expression of the RARb gene resulted in increased sensitivity of the hepatoma cells to the antiproliferative effect of retinoic acid (RA). Furthermore, expression of RARb resulted in a spontaneous differentiation of the hepatoma cells. These data indicate that RARb plays important roles in differentiation and antiproliferation. 1998 Elsevier Science Ireland Ltd. Keywords: Retinoic acid; Retinoic acid receptor; Hepatoma; Differentiation; Antiproliferation
1. Introduction Retinoic acid (RA) regulates essentially all important biological processes including growth, development, differentiation, reproduction, morphogenesis, metabolism and homeostasis [1]. RA exerts its effects by modulating gene expression through its receptors. The highly pleiotropic effects of RA can be explained in part by the multiplicity of retinoic acid receptors (RARs) and retinoid X receptors (RXRs). The a, b and g subtypes of both RARs and RXRs have distinct * Corresponding author. Tel.: +1 310 2223876; fax: +1 310 7826649; e-mail: [email protected]
and conserved sequence domains and exhibit distinct ligand-binding properties [2–4]. The genes coding for RARa and RARb are localized in chromosomal regions in which deletions frequently occur in a number of tumor types (17q21 and 3p24, respectively) [5]. Such chromosomal deletions are generally regarded as strong indicators for the presence of a tumor suppressor gene in that region. Because of the important role of these receptors in inducing differentiation and inhibiting growth, the RAR coding genes are regarded as potential tumor suppressor genes [5]. In breast cancer, deletions and allele losses frequently are observed at chromosomes 3p and 17q [6,7], which could affect RARa and b
0304-3835/98/$19.00 1998 Elsevier Science Ireland Ltd. All rights reserved PII S0304-3835 (97 )0 0475-8
206
C. Li, Y.-J.Y. Wan / Cancer Letters 124 (1998) 205–211
expression. In epidermoid lung cancer, a deletion is detected at 3p where RARb is coded [8]. When RARb mRNA was expressed in lung cancer cells, the cells became less tumorigenic both in morphology and behavior [8]. In hepatoma, the RARb gene was found to be a hepatitis B viral DNA integration site [9]. In hepatoma cell lines, the expression of the RARb gene is either inhibited or reduced [10]. Previously, we have shown that only the RARb gene, and not the RARa and g genes, is inactivated in McA-RH8994 and McA-RH7777 rat hepatoma cells. Expression of the RARb gene is also reduced in other hepatoma cell lines including H4IIE and HepG2 [10]. Since downregulation of the RARb gene is also found in lung cancer cells, head and neck tumors and mammary tumors [11–13] and upregulation of the RARb gene is usually associated with RA-induced differentiation [10,14,15], the RARb gene might also play an important role in hepatocarcinogenesis. In this study, we further investigate the possible molecular mechanisms underlying RARb gene inactivation and the role of RARb in hepatoma cells.
2. Materials and methods
CAGCTCACTTCC-3′ (+39 to +16) [20]. Nucleotides indicating HindIII and XbaI restriction endonuclease cut sites are shown in italics. PCR was performed as described previously [10]. PCR products were digested with XbaI and HindIII and subsequently subcloned into the promoterless pCAT-Enhancer (Promega, Madison, WI), which gave the rRARb-CAT construct. DNA sequencing was performed by Sanger’s dideoxy sequencing method [21]. 2.3. Establishment of the McA-RH8994b cell line The McA-RH8994 cell line was obtained from American Type Culture Collection (Rockville, MD) [22]. Stable transfection was performed using a standard calcium phosphate precipitation method [23]. Cells were exposed to pSV2-neo and RARb2 expression plasmids PECE-RARb [24] at a 1/5 molar ratio for 6 h and then selected by G418 (GIBCO, Bethesda, MD) at a final concentration of 500 mg/ml. After 30 days of treatment, neomycin-resistant clones were selected and expanded individually. The expression of the RARb gene was analyzed by Northern blot hybridization. For transient transfection assay, cells were harvested 72 h after the initiation of transfection. CAT activity was assayed by the phase extraction method [25] and corrected by b-gal activity.
2.1. Northern blot hybridization Northern blot hybridization was performed according to the method described previously [11]. The probes used in the hybridization procedure included a full-length cDNA insert of B1-RARb [16], KpnI fragment of RXRa (∼1.6 kb), 5′ end to AccI site of RXRb (∼1.5 kb), 5′ end to HindIII site of RXRg [17], HindIII fragment of albumin [18] and PstI fragment of PRAF6 [19] for detecting AFP mRNA. The membranes were rehybridized with rat b actin probes for RNA quantitation purposes. 2.2. PCR amplification and DNA sequence analysis Genomic DNA was prepared from McA-RH8994 and male Buffalo rat livers and then amplified with primers complementary to sequences flanking the promoter region of the mouse RARb gene, including 5′cccaagcttGTGAGAATCCTGGGAGTTGGTGA-3′ (−120 to −97) and 5′-attctagaCCTGCCTCGGAG-
3. Results 3.1. Expression of the RXR genes in the McARH8994 cell line Previously, we have shown that by Northern blot hybridization and reverse transcription PCR, McARH8994 and McA-RH7777 hepatoma cells expressed RARa and g mRNA, but not RARb mRNA [10]. To examine the role of RXRs in hepatoma, the expression of the RXRa, b and g genes was examined by Northern blot hybridization (Fig. 1). McA-RH8994 and McA-RH7777 cells expressed 5.6 kb RXRa mRNA and 2.7 and 3.0 kb RXRb mRNA, respectively. Different RXRg transcripts were detected in these two hepatoma cell lines. McA-RH8994 cells expressed a single RXRg transcript of 2.3 kb and McA-RH7777 cells expressed two RXRg transcripts of 2.3 and 3.8 kb. The hybridization was performed under stringent
C. Li, Y.-J.Y. Wan / Cancer Letters 124 (1998) 205–211
207
Fig. 1. Expression of RXR mRNAs in McA-RH8994 and McA-RH7777 rat hepatoma cells. RNA was extracted from McA-RH8994 (1) and McA-RH7777 (2) cells. Total RNA (20 mg) was fractionated by formaldehyde agarose gel, electroblotted onto nylon membrane and then hybridized with 32P-labeled RXRa, b and g and b-actin cDNA probes. Arrows indicate the specific transcripts.
conditions. Thus, the RARb gene is the only RA receptor gene inactivated in these hepatoma cell lines. 3.2. Analysis of the promoter of RARb2 in McARH8994 cells Human and mouse RARb genes have been cloned and sequenced [26,27]. However, the rat RARb gene has never been studied. Since the mouse RARb cDNA clone can be used to hybridize rat RARb mRNA, this suggests that they are highly homologous to each other. To unravel the molecular mechanism involved in inactivation of the RARb gene, the promoter of RARb2/4 was cloned from normal adult livers and McA-RH8994 cells by PCR. The mouse RARb2 sequence was used to synthesize the primers. The results indicated that the RARb2 promoter sequence was identical in McA-RH8994 cells and rat and mouse livers from nucleotides −96 to +15 (data not shown). Within the promoter, it contains the RA response element (bRARE), as reported in the mouse gene [28]. 3.3. The effect of RA on the RARb gene in McARH8994 cells Unlike many other systems, the expression of the RARb gene cannot be induced by RA in McARH8994 and McA-RH7777 cells [10]. RA-induced
RARb expression is mediated through the bRARE present in the RARb promoter [28]. To test if inactivation of RARb expression is due to impaired transcriptional regulation of the RARb gene, reporter constructs containing either promoter region (−96 to +15, rRARb-CAT) or the entire regulatory region (−2800 to +480, RARb2-H3-CAT) of the RARb gene were used for transient transfection assays. The data demonstrated that in McA-RH8994 cells, both constructs responded to RA (Fig. 2). There were seven- and two-fold inductions when RARb2-H3CAT and rRARb-CAT were used as reporter genes,
Fig. 2. RARb gene activity in McA-RH8994 cells. CAT activity was measured in McA-RH8994 cells transiently transfected with rRARb-CAT and RARb2-H3-CAT and treated with RA (10−5 M) for 64 h. pCAT-Enhancer (pCAT-E, Promega) was included as a control plasmid. The data presented are averages of three experiments.
208
C. Li, Y.-J.Y. Wan / Cancer Letters 124 (1998) 205–211
Fig. 3. Expression of the RARb gene in McA-RH8994b cells. The expression of RARb mRNA was studied in rat livers and McA-RH8994 and McA-RH8994b cells treated with or without RA (10−5 M) for 3 days by Northern blot hybridization. Ethidium bromide staining of ribosomal RNA demonstrated that each lane had a similar amount of RNA. The arrow indicates the specific transcript.
respectively. These data suggest that besides the promoter, there are other cis-acting elements involved in RA-mediated RARb gene regulation and McARH8994 cells contain the necessary factors to transactivate the RARb gene. 3.4. Analysis of the role of RARb in McA-RH8994 cells
Fig. 4. Analysis of the function of RARb in McA-RH8994b cells. McA-RH8994 and McA-RH8994b cells were transiently transfected by rRARb-CAT, treated with or without RA (10−6 and 10−5 M) (a) and CD2497 (2 × 10−7 M) (b) for 48 h and then CAT assays were performed. The data presented are averages of three experiments.
To analyze the role of RARb in McA-RH8994 cells, RARb was restored in McA-RH8994 cells by stable transfection. Twenty clones were screened for RARb mRNA expression. Among them, 12 expressed detectable amounts of RARb mRNA of variable sizes. One clone, designated as McA-RH8994b, was selected for further characterization because this cell line and adult rat livers expressed similar amounts and sizes of RARb mRNA. It is unlikely that the establishment of the McA-RH8994b clone was due to the selection of McA-RH8994 cells since RARb mRNA could not be detected in McA-RH8994 cells at all by RT-PCR [10]. As shown in Fig. 3, rat livers expressed 3.1 and 2.8 kb RARb transcripts. No RARb mRNA was detected in McA-RH8994 cells. McA-RH8994b cells expressed a single RARb transcript of 2.8 kb
C. Li, Y.-J.Y. Wan / Cancer Letters 124 (1998) 205–211
209
Fig. 5. The expression and regulation of the AFP and albumin genes in McA-RH8994 and McA-RH8994b cells. McA-RH8994 (b−) and McARH8994b (b+) cells were grown in the absence or presence of RA (10−6 M) for 3 days. Northern blot hybridization was performed to examine the expression of the AFP (a), albumin (b) and rRNA (c) genes. Arrows indicate the specific transcripts.
since no alternative splicing and/or differential usage of promoter would occur. In addition, RARb mRNA in McA-RH8994b cells could not be induced by RA because the RARb expression plasmid used in transfection does not contain bRARE. To ensure that RARb expressed in McA-RH8994b cells was functioning, RARb (+) and RARb (−) McARH8994 cells were transiently transfected with rRARb-CAT. These cells were then treated with or without all-trans-RA (RA) at 10−6 and 10−5 M and RARb selective ligand (CD2497) at 2 × 10−7 M. At 2 × 10−7 M, CD2497 maintains its RARb selectivity
Fig. 6. The antiproliferative effect of RA on McA-RH8994 and McA-RH8994b cells. McA-RH8994 and McA-RH8994b cells were seeded in 10-cm culture dishes and treated with or without RA (10 − 6 and 10 − 5 M) for 3 days. The cell number was then determined for the triplicated samples. The data presented are averages of three experiments.
[29]. The data indicated that the basal CAT activity and the fold increase of CAT activity mediated by RA were higher in RARb (+) than in RARb (−) cells (Fig. 4a). In RARb (−) cells, CAT activity was not induced by RARb selective ligand CD2497. In contrast, CD2497 increased CAT activity in RARb (+) cells (Fig. 4b). These data indicate that RARb is actually expressed and functioning in McA-RH8994b cells. Previously, we have demonstrated that RA induced the expression of the AFP and albumin genes in McARH8994 cells, which resulted in a phenotype similar to well differentiated fetal hepatocytes [30]. The phenotype of McA-RH8994b cells was also analyzed by examining the expression and regulation of the AFP and albumin genes by Northern blot hybridization. RARb (+) and (−) cell lines were treated with or without RA (10−6 M) for 3 days. Compared with the RARb (−) cell line, McA-RH8994b cells expressed high basal levels of AFP as well as albumin mRNA, which was similar to the phenotype of RA-treated McA-RH8994 cells. In addition, the levels of both AFP and albumin mRNA were increased upon RA treatment (Fig. 5). The data suggest that expression of the RARb gene results in spontaneous differentiation of hepatoma cells. As demonstrated in different cell systems, RA profoundly inhibits the growth of tumor cells [1]. The antiproliferative effect of RA was also tested on RARb (+) and (−) hepatoma cells. McA-RH8994 and McA-RH8994b cells were treated with RA for 3
210
C. Li, Y.-J.Y. Wan / Cancer Letters 124 (1998) 205–211
days and then the cell number and total protein were determined by cell counting and Bio-Rad protein assay kit, respectively. Fig. 6 shows that RA had no significant effect on the growth inhibition of McARH8994 cells. However, RA (10−5 M) inhibited the growth of McA-RH8994b cells by 60%. Protein data were similar to the cell count data (data not shown). At 10−5 M, RA was not toxic to the cells since no significant amount of dead cells was detected by the Trypan Blue exclusion test. These results indicate that expression of RARb not only changes the hepatoma cells to a more differentiated phenotype, but also enhances the sensitivity of the cells to the antiproliferative effect of RA.
4. Discussion Besides hepatoma cells, inactivation of the RARb gene is found in lung cancer cells, head and neck tumors and mammary tumors [11–13]. Most of the neoplastic cells and tissues that express little or no RARb mRNA still express RARa and g and at least one of the RXRs, which might mediate the transactivation of retinoid responsive genes. This observation raises the possibility that RARb regulates specific genes that are important for suppression of carcinogenesis. This notion is supported by our findings and those of others which show that expression of RARb increases the antiproliferative effect of RA. Furthermore, overexpression of RARb decreases the tumorigenicity of human lung cancer cells [11], transient transfection of the RARb gene results in RA-dependent suppression of cell proliferation [31] and loss of RARb expression in vivo might play an important role in the expansion of premalignant clones for oral lesions [12]. The mechanisms underlying the loss of RARb mRNA in cancer cells including hepatoma are not clear. In McA-RH8994 and McA-RH7777 cells, we have shown that no RARb gene rearrangement is detected [10]. In addition, no mutation is present in the RARE of the RARb2 promoter. By transient transfection, RA was able to activate the RARb gene promoter in McA-RH8994 cells. Furthermore, when the CAT gene was driven by the full-length RARb gene regulatory region, the RA-induced CAT activity was further increased. These data indicate that McA-
RH8994 cells contain the necessary factors to transactivate the RARb gene and the sequence beyond the RARb promoter is also involved in RA-mediated RARb gene activation. Therefore, one possible cause for the reduction in the expression of the RARb gene in hepatoma cells is the presence of mutation in other regulatory regions of the RARb gene which we did not examine. Since we did not examine the rate of transcription of the endogenous RARb gene in these cells from both RARb promoters, it is also possible that the gene is transcribed. However, there is a problem with the production of a full-length pre-RARb mRNA or processing of the RARb mRNA and thus RARb mRNA was not detected in the cells. We have shown that RA induces the expression of AFP and albumin mRNA in McA-RH8994 cells and results in a phenotype similar to well differentiated fetal hepatocytes [30]. Since the RARb gene is inactivated in McA-RH8994 cells, the induction of the AFP and albumin genes is not mediated directly through RARb. However, this should not be interpreted to mean that RARb is not important for differentiation. In fact, our data have shown that RARb causes spontaneous differentiation of hepatoma cells, suggesting that when RARb is absent, other types of RA receptors can also mediate the differentiation effect of RA. However, when RARb is present, the hepatoma cells can undergo spontaneous differentiation. In other words, the differentiation effect of RA receptors might be redundant and RARb does have a differentiation action. In addition, RARb might play a unique role in antiproliferation since expression of RARb increases the sensitivity of McA-RH8994 cells to the antiproliferative effect of RA. However, since RARs have been found to be functionally redundant [4], it is also possible that a similar response will be obtained if RARa and g are overexpressed in the McA-RH8994 cells.
Acknowledgements This work was supported by NIH grant CA53596. We thank Dr R. Evans for the RXR cDNA clones and mRARb2-H3-CAT, Dr M. Pfahl for the RARb cDNA clones and Drs U. Reichert and B. Shroot for CD2497. We also thank Ms A. Flores for assistance in the preparation of this manuscript.
C. Li, Y.-J.Y. Wan / Cancer Letters 124 (1998) 205–211
References [1] M.B. Sporn, A.B. Robers, D.S. Goodman (Eds.), The Retinoids, second ed., Raven Press, New York, 1994. [2] D.J. Mangelsdorf, C. Thummel, M. Beato, P. Herrlich, G. Schutz, K. Umesono, B. Blumberg, P. Kastner, M. Mark, P. Chambon, R.M. Evans, The nuclear receptor superfamily: the second decade, Cell 83 (1995) 835–839. [3] D.J. Mangelsdorf, R.M. Evans, The RXR heterodimers and orphan receptors, Cell 83 (1995) 841–850. [4] P. Chambon, A decade of molecular biology of retinoic acid receptors, FASEB J. 10 (1996) 940–954. [5] R. Lotan, J.L. Clifford, Nuclear receptors for retinoids: medicators of retinoid effects on normal and malignant cells, Biomed. Pharmacother. 45 (1991) 145–156. [6] I.U. Ali, R. Lidereau, R. Callahan, Presence of two members of c-erb A receptor gene family (c-erb Ab and c-erb A2) in the smallest region of somatic homozygosity on chromosome 3p21–p25 in human breast carcinoma, J. Natl. Cancer Inst. 81 (1989) 1815–1820. [7] J.M. Hall, M.K. Lee, B. Newman, J.E. Morrow, L.A. Anderson, B. Huey, M.C. King, Linkage of early onset familial breast cancer to chromosome 17q21, Science 250 (1990) 1684–1689. [8] F.J. Gebert, N. Moghal, J.V. Frangioni, D.J. Sugarbaker, High frequency of retinoic acid receptor b abnormalities in human lung cancer, Oncogene 6 (1991) 1859–1868. [9] H. de The, A. Marchio, P. Tiollais, A. Dejean, A novel steroid thyroid hormone receptor-related gene inappropriately expressed in human hepatocellular carcinoma, Nature 330 (1987) 667–670. [10] Y.-J.Y. Wan, L. Wang, T.-C.J. Wu, Expression of retinoic acid receptor genes in developing rat liver and hepatoma cells, Lab. Invest. 66 (1992) 646–651. [11] B. Houle, C. Rochette-Egly, W.E.C. Bradley, Tumor-suppressive effect of the retinoic acid receptor b in human epidermoid lung cancer cells, Proc. Natl. Acad. Sci. USA 90 (1993) 985–989. [12] R. Lotan, X.-C. Xu, S.M. Lippman, J.Y. Ro, I.S. Lee, J.J. Lee, W.K. Hong, Suppression of retinoic acid receptor b in premalignant oral lesions and its up-regulation by isotretinoin, N. Engl. J. Med. 332 (1995) 1405–1410. [13] K. Swisshelm, K. Ryan, X. Lee, H.C. Tsou, M. Peacocke, R. Sager, Down-regulation of retinoic acid receptor b in mammary carcinoma cell lines and its up-regulation in senescing normal mammary epithelial cells, Cell Growth Differ. 5 (1994) 133–141. [14] H. de The, A. Marchio, P. Tiollais, A. Dejean, Differential expression and ligand regulations of the retinoic acid receptor a and b genes, EMBO J. 8 (1989) 429–433. [15] Y.-J.Y. Wan, L. Wang, T.-C.J. Wu, Different response to retinoic acid of two teratocarcinoma cell lines, Exp. Cell Res. 219 (1995) 392–398. [16] D. Benbrook, E. Lernhardt, M. Pfahl, A new retinoic acid receptor identified from a hepatocellular carcinoma, Nature 333 (1988) 669–672. [17] D.J. Mangelsdorf, U. Borgmeyer, R.A. Heyman, J.Y. Zhou,
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
211
E.S. Ong, A.E. Oro, A. Kakizuka, R.M. Evans, Characterization of three RXR genes that mediate the action of 9-cis retinoic acid, Genes Dev. 6 (1992) 329–344. T.D. Sargent, M. Yang, J. Bonner, Nucleotide sequence of cloned rat serum albumin, Proc. Natl. Acad. Sci. USA 78 (1981) 243–246. L.L. Jagodzinski, T.D. Sargent, M. Yang, C. Glackin, J. Bonner, Sequence homology between RNAs encoding rat a-fetoprotein and rat serum albumin, Proc. Natl. Acad. Sci. USA 78 (1981) 3521–3525. A. Zelent, C. Mendelsohn, P. Kastner, A. Krust, J.M. Garmier, F. Ruffenach, P. Leroy, P. Chambon, Differentially expressed isoforms of the mouse retinoic acid receptor b are generated by usage of two promoters and alternative splicing, EMBO J. 10 (1991) 71–81. F. Sanger, S. Nicklen, A.R. Loulson, DNA sequencing with chain-terminating inhibitors, Proc. Natl. Acad. Sci. USA 74 (1977) 5463–5467. J.E. Becker, B. De Nechaud, V.R. Potter, Two new rat hepatoma cell lines for studying the unbalanced ontogeny hypothesis, in: W.H. Fishman, S. Sell (Eds.), Oncodevelopmental Gene Expression, Academic Press, New York, 1976, p. 259. C. Chen, H. Okayama, Calcium phosphate-mediated gene transfer: a highly efficient system for stable transforming cells with plasmid DNA, Biotechniques 6 (1988) 632–638. G. Graupner, K.N. Wills, M. Tzukerman, X.K. Zhang, M. Pfahl, Dual regulatory role for thyroid-hormone receptors allows control of retinoic-acid receptor activity, Nature 340 (1989) 653–656. D.A. Nielson, T.C. Chang, D.J. Shapiro, A highly sensitive mixed phase assay for chloramphenicol acetyltransferase activity in transfected cells, Anal. Biochem. 179 (1989) 19–23. A. Zelent, A. Krust, M. Petkovich, P. Kastner, P. Chambon, Cloning of murine retinoic acid receptor a and b cDNAs and of a novel third receptor g predominantly expressed in skin, Nature 339 (1989) 714–717. N. Brand, M. Petkovich, A. Krust, P. Chambon, H. de The, A. Marchio, P. Tiollais, Identification of a second human retinoic acid receptor, Nature 332 (1988) 850–853. H.M. Sucov, K.K. Murakami, R.M. Evans, Characterization of an autoregulated response element in the mouse retinoic acid receptor type b gene, Proc. Natl. Acad. Sci. USA 87 (1990) 5392–5396. B.A. Bernard, J.M. Bernardon, C. Delesduse, B. Martin, M.C. Lenoir, J. Maignan, B. Charpentier, W.R. Pilgrim, U. Reichert, B. Shroot, Identification of synthetic retinoids with selectivity for human nuclear retinoic acid receptor g, Biochem. Biophys. Res. Commun. 186 (1992) 977–983. Y.-J. Wan, T.-C. Wu, The effects of retinoic acid on the expression of a-fetoprotein and albumin genes in rat hepatoma cell lines, Differentiation 50 (1992) 107–111. J.U. Frangioni, N. Moghal, A. Stuart-Tilley, B.G. Neel, S.L. Alper, The DNA binding domain of retinoic acid receptor b is required for ligand-dependent suppression of proliferation: application of general purpose mammalian co-expression vectors, J. Cell Sci. 107 (1994) 827–838.