The retinoblastoma susceptibility gene product/Sp1 signalling ... - Nature

4 downloads 25 Views 185KB Size Report
The retinoblastoma susceptibility gene product/Sp1 signalling pathway is modulated by Ca2+/calmodulin kinases II and IV activity. FreÂdeÂric Sohm1,2 ...
ã

Oncogene (1999) 18, 2762 ± 2769 1999 Stockton Press All rights reserved 0950 ± 9232/99 $12.00 http://www.stockton-press.co.uk/onc

The retinoblastoma susceptibility gene product/Sp1 signalling pathway is modulated by Ca2+/calmodulin kinases II and IV activity FreÂdeÂric Sohm1,2, Christian Gaiddon1,2, Muriel Antoine1, Anne-Laurence Boutillier1 and Jean-Philippe Loe‚er*,1 1

IPCB, UMR 7519 CNRS, 21 rue Rene Descartes, 67084 Strasbourg Cedex, France

To investigate the possible link between Ca2+ signalling and cell cycle control we analysed Ca2+/calmodulin kinases (CamK) interaction with the retinoblastoma susceptibility gene product/SP1 pathway. CamK II and IV activate c-fos transcription through a short promoter region (799 to 753) containing the retinoblastoma control element (RCE) and a cAMP response element (CRE) related sequences. Deletion analysis revealed that the RCE is a major CamK responsive element and is sucient to confer CamK and Ca2+ regulation to a minimal promoter. Direct interactions between SP1 and RCE were con®rmed by gel shift experiments. Using transient transfection experiments, we show that CamKdependent transcription is regulated by the retinoblastoma (Rb) susceptibility gene product and the p107 Rb related protein. However, the stimulatory e€ects of CamKs and Rb on c-fos are blocked by overexpression of both proteins. These e€ects appear to be directly mediated by SP1 as shown by the use of a Gal4/SP1 fusion proteins. In conclusion, CamK II and IV, two major Ca2+-dependent intracellular e€ectors, may represent a molecular link between this second messenger transduction pathway and e€ectors that control cell cycle progression through Rb/SP1 signalling pathway. Keywords: Ca2+ signalling pathway; di€erentiation; retinoblastoma; SP1

cell

cycle;

Introduction The Ca2+ signalling pathway is implicated in the control of both cell growth and di€erentiation. However, the molecular mechanisms integrating these processes at the transcriptional level are poorly understood. We used the c-fos gene as a model and a marker of cell cycle initiation to analyse the interactions between the Ca2+/calmodulin pathway and that of the retinoblastoma susceptibility gene product/SP1, another key element in cell cycle control (Angel and Karin, 1991; Horowitz, 1993). Indeed, the immediate early gene c-fos is regulated both in cycling and post mitotic cells by a variety of regulatory pathways (Karin, 1992). It is well documented that the Ca2+ signalling pathway is a major regulatory system which couples bioelectrical activity and c-fos transcription (Sheng and Greenberg, 1990; Wang and

*Correspondence: J-P Loe‚er 2 The ®rst two authors contributed equally to this study Received 6 July 1998; revised 23 November 1998; accepted 15 December 1998

Simonson, 1996). These regulations have been studied extensively and it is now clear that, within the c-fos promoter, multiple cis acting elements respond to Ca2+ modulation. Although the CRE/CaRE (cAMP response element/calcium response element) (Fisch et al., 1989) located at 760 bp prior to the transcription start site is the most studied (Sheng et al., 1990; Ghosh et al., 1994; Barthel and Loe‚er, 1995; Johnson et al., 1997; Hardingham et al., 1998), additional control elements: SRE (serum response element) and FAP (Fos AP1-like element), both located around 7300 bp prior to the transcription initiation site, are also able to activate depolarization-dependent c-fos transcription (Misra et al., 1994; Barthel and Loe‚er, 1995; Antoine et al., 1996; Johnson et al., 1997). Ca2+/calmodulin dependent kinases (CamKs) are major e€ectors of the Ca2+ signalling pathway. In a recent study, we have shown, by expressing dominant active forms of CamKs II and IV, that the SRE, the FAP element and a proximal region of the c-fos promoter (799 to 753) are also the direct targets of these kinases in AtT20 cells (Antoine et al., 1996). This proximal region of c-fos promoter contains the CRE/ CaRE element. At this level, transcriptional stimulation appears to be mediated by the CRE-binding protein (CREB) transcription factor (Dash et al., 1991; Sheng et al., 1991). Although CREB is a direct phosphorylation target of CamKs, CamK II has been shown to strongly repress CREB mediated transcription through CRE while CamK IV activates CREB activity (Sheng et al., 1991; Matthews et al., 1994; Sun et al., 1994). Since CamK II and IV are equipotent in stimulating transcription from the proximal region (799 to 753 bp) of the c-fos promoter (Antoine et al., 1996), it is likely that additional CamK-dependent cis activating sequence(s) exists in this promoter area. One possible candidate is the retinoblastoma control element (RCE) (Robbins et al., 1990; Kim et al., 1992). The RCE is located between 799 and 772 bp prior to the transcription initiation site. This element can be positively or negatively regulated in a cell type-speci®c manner by the retinoblastoma gene product and three nuclear proteins of 115, 95 and 80 kD (retinoblastoma control proteins, RCPs) which bind RCE in vitro (Robbins et al., 1990; Kim et al., 1992; Udvadia et al., 1992). This opens up the interesting possibility that the Ca2+/CamK pathway directly interacts with the Rb proteins family and/or RCP(s). The RCPs speci®cally bind to the RCEs within the c-fos, c-myc and TGF-b1 promoters (Udvadia et al., 1992). RCPs were characterized recently and shown to be members of the SP1 transcription factor family. The 95 and 115 kD proteins correspond respectively to SP1 and SP3 transcription factors and activate the RCE

Retinoblastoma and Ca2+ calmodulin kinase cross talk F Sohm et al

mediated transcription, whereas the 80 kD RCP, which is produced by internal initiation of translation from SP3 mRNA, acts as a potent inhibitor of SP1/SP3 mediated transcription (Udvadia et al., 1993, 1995; Kennett et al., 1997). RCP mediated transcription is controlled by Rb proteins (Kim et al., 1992). It seems that Rb regulates RCPs-dependent transcription rather by interaction with RCPs than binding to the RCE DNA sequence (Udvadia et al., 1992, 1995). The RB gene was ®rst characterized as a tumour suppressor gene (Friend et al., 1986; Fung et al., 1987; Lee et al., 1987a,b). The Rb protein is an ubiquitously expressed nuclear protein with non-speci®c binding capacity for DNA (Lee et al., 1987a,b). In normal cells, Rb activity is regulated by cell cycle-dependent phosphorylations (Mihara et al., 1989; Xu et al., 1989; Buchkovich et al., 1989; Chen et al., 1989; DeCaprio et al., 1989). During G0 or G1 phases Rb is underphosphorylated whereas it is phosphorylated at multiple sites during the G1/S transition and S phase. The cyclin-dependent phosphorylations of Rb protein are well documented (reviewed in Hollingsworth et al., 1993). There is also some evidence for the implication of Ca2+/calmodulin kinases in Rb dependent pathways (Takuwa et al., 1993). The phosphorylation level of Rb is thought to control Rb interaction with several transcription factors implicated in cell cycle progression. Rb has been shown to form cell cycle-regulated complexes with E2F, a transcription factor that regulates expression of cellular and viral genes and to regulate E2F activity (Nevins, 1992). In the present study, we show that the Rb control element (RCE) located in the proximal promoter region of the c-fos gene is fully responsive to CamK II and IV. Furthermore, CamK-dependent transcription is regulated by Rb, through the SP1 transcription factor. These ®ndings provide a molecular basis linking bioelectrical activity to cell cycle progression and cellular di€erentiation.

et al., 1994; Matthews et al., 1994). With a reporter gene containing solely the RCE (deletion of the CRE from the proximal promoter that generates pFC99DCRE, Figure 1b), the potency of CamK II is not signi®cantly changed (t-test, P50.01), whereas the activity of CamK IV is reduced by about 50% (t-test, P50.05) as compared to the parental plasmid pFC99. Since both CamK II and IV stimulate pFC99DCRE in an almost equipotent manner, our results show that the RCE is a major site of transcriptional regulation by CamKs in the c-fos promoter proximal region. The RCE confers CamK responsiveness in AtT20 cells To verify that the RCE element of the c-fos promoter is sucient to confer Ca2+ responsiveness, the e€ects of depolarization were tested on AtT20 cells, transfected with a reporter gene containing two wild-type RCE sequences in front of the thymidine kinase (tk) minimal promoter, coupled to the CAT reporter gene (RCE-tkCAT). After K+ treatment, a threefold increase in CAT activity is observed (Figure 2a). A mutated RCE construct (DRCE-tk-CAT) was transfected as negative control and did not show any response to K+ treatment. In co-transfection experiments, both CamK II and IV are able to increase RCE-driven CAT activity (Figure 2b), suggesting that CamKs directly regulate transcription factors, acting through the RCE. We then analysed the nucleoprotein complexes that bind the RCE sequence by mobility gel shift assay. Figure 3a shows three speci®c retarded complexes (lane 1) that are competed for by the cold probe (RCE, lane

Results Activation of the 799 c-fos promoter by CamKs recruits the Rb responsive element To determine the c-fos promoter element(s) implicated in CamK II and IV-mediated induction we cotransfected either active forms of CamK II or IV with partial deletions of the c-fos promoter proximal region (99 bp) (Figure 1a). As expected, both calmodulin kinases type II and IV activate transcription of 799 c-fos promoter and are unable to activate the 53 bp minimal promoter (Figure 1b: pFC99 and pFC53). We then tested separately the two elements that have been characterized between base pair 799 and 753 of the c-fos promoter: the CRE/CaRE located at 760 bp, and the retinoblastoma control element (RCE) located at 790 bp. When compared to pFC99 that gives the full transcriptional response to both CamKs, the induction is largely (for CamK IV) or entirely (for CamK II) inhibited by the deletion of the RCE (Figure 1b: pFC72). This result is in line with previous reports that show that CamK IV but not CamK II stimulates c-fos transcription through the CRE by recruiting the transcription factor CREB (Sun

Figure 1 Delineation of CamKs-responsive elements in the proximal region of the c-fos promoter. (a) The di€erent promoter constructs are depicted: pFC700: full length c-fos promoter (SRE: serum responsive element; FAP: fos AP1 like element; RCE: retinoblastoma control element; CRE: cAMP responsive element). PFC99, pFC72, pFC99DRCE: partial deletions of the c-fos promoter as indicated; pFC53: minimal c-fos promoter. (b) Induction of CamK mediated transcription. The reporter genes were co-transfected together with expression vectors coding the bglobin (negative control), CamK II or IV. The results are presented as fold induction of CAT activity. Fold induction is de®ned as CAT activity in the presence of CamKs versus CAT activity in the presence of b-globin. Histogram show means+s.e.m. (n=4), *P50.001 compared to controls with bglobin (open bars) (Student's t-test)

2763

Retinoblastoma and Ca2+ calmodulin kinase cross talk F Sohm et al

2764

2) and not by an unrelated sequence (CRE, lane 3). The binding of these proteins to the RCE does not change upon K+ treatment when compared to untreated control (lanes 4 and 5). These complexes were further characterized by preincubation of the nuclear extracts with anti-SP1 and SP3 antibodies. Supershifted material (lanes 6, 7, 8 and 9) revealed that the c-fos RCE binds SP1 and two forms of SP3 (immunologically related proteins). The role of CamKs on the regulation of SP1 transcription factor-mediated transcription was further investigated, using chimerical SP1-Gal4 fusion protein and the Gal4 promoter fused to the luciferase gene (Gal4-luc) (Sif et al., 1993). The SP1-Gal4 fusion protein encodes the SP1 transcription factor, lacking the DNA-binding zinc ®nger domain, fused to the Gal4 DNA binding and dimerization domains (amino acids 1 ± 147). When the Gal4-SP1 expression vector is co-

Figure 2 The RCE mediates depolarization and CamKs dependent transcription. Reporter genes containing two wild type (RCE-tk-CAT) or mutated (negative control: DRCE-tkCAT) RCE fused to the thymidine kinase minimal promoter were tested. (a) Depolarization (K+, 30 mM for 8 h) induces RCEdriven transcription. The mutated construct (DRCE-tk-CAT) is unresponsive. (b) Both CamK II and IV activate the wild type RCE directed transcription. These e€ects are selective as demonstrated by the absence of induction of the reporter gene containing the mutated element. b-globin was taken as an internal control, to keep the total amount of transfected DNA constant. Histograms are means+s.e.m. (n=4), *P50.01 compared to controls with b-globin (open bars) (Student's t-test)

transfected with a Gal4-luc reporter gene in the presence of CamK II or CamK IV, both kinases induce a sixfold increase of luciferase activity compared to the control without CamK (b-globin) (Figure 3b). The control B17-Gal4 that encodes an acidic activation domain, unrelated to SP1 (Sif et al., 1993) is unable to regulate the transcription of the Gal4-luciferase gene in response to CamKs (Figure 3b). Taken together, this set of experiments clearly demonstrates that CamKs modulate the transcriptional activity of SP1. CamKs-mediated c-fos and RCE driven transcriptional activation is inhibited by Rb and p107 proteins co-expression Di€erent studies indicate that Rb stimulates c-fos transcription in some cell types (Kim et al., 1992),

Figure 3 CamKs stimulate transcription by recruiting the SP1 transcription factor. (a) Gel mobility shift assays. The labelled RCE probe generates three speci®c complexes (lane 1) as indicated by arrows on the left side. They are competed by the cold probe (RCE, lane 2) but not by an unrelated sequence (CRE, lane 3). No changes in binding activity was observed when nuclear extracts from control (lane 4) or depolarized (K+, 30 mM, 30 min) (lane 5) cells were used. Preincubation of the nuclear extracts with anti-SP1 and SP3 directed antibodies generates supershifted material (indicated by arrows on right side, lanes 6, 7, 8 and 9), showing that the c-fos RCE binds SP1 and two forms of SP3. (b) CamK II and IV stimulates SP1-dependent transcription. A reporter gene driven by Gal4 binding sites was co-transfected with a expression vector encoding a Gal4-SP1 fusion protein (see Materials and methods) and CamKs expression vectors (as in Figure 1). An SP1 unrelated fusion protein (B17-Gal4) is not a€ected by expression of CamKs. Histograms show means of fold induction versus control+s.e.m. (n=4), *P40.01 compared to controls with bglobin (open bars) (Student's t-test)

Retinoblastoma and Ca2+ calmodulin kinase cross talk F Sohm et al

whereas it acts as an inhibitor in others (Robbins et al., 1990). Here, we tested the e€ects of Rb or p107 proteins on c-fos-driven transcription (pFC700) by transient co-transfection assays. Cells were transfected with di€erent amounts of the p107 protein expression vector (Figure 4a, left). This experiment shows that CAT activity ¯uctuates depending on the amounts of transfected expression vector, but stimulation of c-fosdriven transcription by p107 clearly remained significant relative to the non stimulated situation (control vector). The same results were obtained with Rb (not shown). The same approach was used, to study the e€ect of CamKs on Rb/p107-stimulated c-fos transcription (Figure 4a, right). Here again, CAT activity varies relative to the quantities of expression vector added in the transfection medium. However, the major e€ect of both CamKs (II and IV) is an inhibition of Rb/p107-stimulated c-fos transcription, the maximum response being obtained at 0.2 mg of Rb/p107 expression vectors. For further experiments, the amount of Rb/p107 expression vector was held constant at 0.2 mg (see Materials and methods). The next step was to investigate whether the stimulation of c-fos transcription by Rb and p107 was dependent on the RCE. As shown in Figure 4b, while the full length c-fos promoter (pFC700) is clearly induced by Rb and p107 up to 12 ± 13-fold, deletion of the RCE (p700DRCE) abolished this stimulatory response. We found the same e€ects when the RCE/ tk-CAT and its control (DRCE-tk-CAT) were used as reporter genes. This series of experiments implies that the RCE element is necessary to activate c-fos transcription by Rb or p107 proteins. Having shown that the CamKs -mediated induction of c-fos transcription is downregulated when pRb family of proteins are co-expressed, it was then important to understand if this inhibitory e€ect was acting through the same regulatory sequence, the RCE. A series of co-transfection experiments was performed with the full length c-fos promoter (pFC700), the full length but RCE-deleted c-fos promoter (pFC700DRCE) and the RCE-tk-CAT reporter genes, together with CamKs (II or IV) and either p107 or Rb expression vectors (quantities de®ned as in Figure 4a). Results are shown on Figure 4c. The inhibitory e€ects of the Rb proteins family on CamKs-induced c-fos transcription (basal) is clearly lost when the RCE sequence is deleted from the full length c-fos promoter, indicating that Rb and p107 speci®cally inhibit RCEmediated CamKs e€ects. The greater inhibitory e€ect of Rb and p107 on the isolated RCE than on pFC700 probably re¯ects the presence of additional CamKs inducible (FAP, SRE, CRE), but Rb insensitive elements, in the 5' ¯anking region of the c-fos gene (Antoine et al., 1996). The mutated DRCE construct was not responsive to any stimulation. Rb and p107 proteins induce SP1 transcriptional activity, but repress it when co-expressed with CamKs We next veri®ed the e€ects of p107 and Rb proteins on SP1 transcriptional activity, using the SP1-Gal4 fusion proteins. As shown in Figure 5a, both Rb and p107 activate SP1-Gal4 mediated transcription. As CamKs also activate SP1-Gal4 mediated transcription (Figure

3b), we tested the in¯uence of Rb and p107 on this response. In the presence of CamK II or IV both Rb and p107 showed a strong inhibition of CamKsinduced SP1-Gal4 transcription (Figure 5b), suggesting that the repressive e€ects observed on c-fos and RCE-mediated transcription may be the re¯ection of SP1 transcriptional inhibition.

Figure 4 CamKs modulate the Rb signalling pathway. (a) E€ects of p107 on c-fos expression. Increasing quantities of p107 expression vectors were co-transfected with the pFC700 reporter vector in the absence (left) or presence (right) of CamKs. In the absence of CamK, p107 activates transcription of the full length promoter. *P50.01 compared to basal expression (Student's ttest). In the presence of CamK II or CamK IV, p107 inhibits CamKs-mediated transcription. The results are presented as fold induction in the presence of p107 and/or CamKs compared to basal transcription in the absence of p107 and CamKs. *P40.01 compared to basal expression induced by CamKs (Student's ttest). (b) CamKs and Rb regulate transcription through the RCE. Both p107 and Rb activate transcription of the full length c-fos and RCE-tk promoters. The RCE de®cient c-fos promoter (PFC700DRCE) and the DRCE-tk-CAT are not regulated by p107 or Rb. Histograms show means of fold induction versus control+s.e.m. (n=4), *P40.01 compared to controls (control vector, open bars) (Student's t-test). (c) E€ects of Rb and p107 on CamKs-dependent transcription. The results are presented as percentage of inhibition of the CamKs induced transcription (represented as 100%) by Rb and p107. Both Rb and p107 repress CamKs induced transcription of the full length c-fos promoter and RCE-tk reporter gene. The activation of transcription by CamKs of the pFC700DRCE construct which does not contain the RCE sequence is una€ected by Rb and p107, indicating that the RCE is necessary for Rb repression. Means+s.e.m. (n=4) are given, *P40.01 compared to the basal obtained for each vector (represented as 100%, open bars) (Student's t-test)

2765

Retinoblastoma and Ca2+ calmodulin kinase cross talk F Sohm et al

2766

Discussion Our starting point in this series of experiments was to establish the key elements conferring CamK responsiveness in the c-fos promoter. Previous studies had demonstrated that CamK II and IV regulate c-fos transcription through several responsive elements: the fos AP1-like element (FAP), the serum responsive element (SRE) and one or more element(s) placed on proximal region of the c-fos promoter as yet unidenti®ed (Antoine et al., 1996). In the present work, we show that the cis-acting elements located in the proximal region of the c-fos promoter between bp 799 and 753 bp, confer the CamKs response. This

Figure 5 The transcriptional e€ects of Rb/p107 and CamKs are mediated by SP1 (a) Rb and p107 stimulate transcription mediated by an SP1-Gal4 fusion protein. Rb or p107 were cotransfected with SP1-Gal4 expression vector and the Gal4Luciferase reporter gene as in Figure 3b. The results are presented as fold induction in the presence of Rb or p107 as compared to non stimulated transcription levels (absence of Rb/ p107). (b) Both Rb and p107 are able to inhibit CamKs induced SP1-Gal4-dependent transcription. The results are presented as percentage of inhibition of maximal stimulation (represented as 100%) by CamK II and IV. Histograms are means+s.e.m. (n=4) are given, *P40.001 compared to respective controls (a control vector; b basal induced by CamKs) (Student's t-test)

region encompasses a RCE and a CRE. A detailed analysis reveals that, the RCE is more responsive to the CamK II than the CRE whereas the CamKIV activate both the RCE and the CRE. Indeed an isolated RCE drives CamKs-induced transcription of the thymidine kinase promoter. We further demonstrate that CamK-induced RCE-mediated transcription involves the SP1 protein family. Indeed, the physical interaction between SP1 and SP3 proteins and RCE was con®rmed by gel shift analysis. These data are in line with previous work from other laboratories (Udvadia et al., 1992, 1993, 1995; Kennett et al., 1997), as is the result showing that overexpression of Rb/p107 also induces RCE-driven transcription. However, an unexpected result was the inhibition of RCE-mediated and SP1-Gal4 mediated transcription when CamK and Rb/p107 are co-expressed. This suggests that the RCE is the site of integration of several pathways through the SP transcription factors, ®rst the Ca2+ and biolelectrical activity pathways via the CamKs and second, the Rb/p107 pathway. If we now turn to the molecular mechanisms that underlie the interactions between CamKs and Rb at the level of SP1-dependent transcription we can start from three clear cut observations: ®rst, Rb protein potentiates the transcriptional e€ects of SP1, second CamK II and IV share this activation property and third, when expressed together, instead of having additive or synergistic e€ects, CamKs and Rb coexpression inhibits SP1 dependent transcription. Considering the action of Rb protein ®rst, hypotheses can be forwarded for both its stimulatory e€ects seen when expressed alone and for the inhibitory e€ects induced on co-expression with CamK. Indeed, the observation that Rb stimulates SP1 corroborates published data showing that Rb liberates SP1 from an inhibitory partner SP1-I thus enabling it to activate transcription (Chen et al., 1994). Using CCL-64 cells these authors demonstrated that Rb releases SP1 from SP1-I and enhances SP1-mediated transcription of cjun by increasing SP1 binding to the RCE. Moreover, the RCE-mediated transcription is dependent on the type of SP1 protein family bound to this element. SP1 and the longer isoform of SP3 are both activators of transcription whereas the short form of SP3 is an inhibitor of transcription (Udvadia et al., 1992, 1993, 1995; Kennett et al., 1997). The inhibition of SP1 activity by CamK and Rb together could be explained by the phosphorylation of Rb by CamK, as phosphorylation of Rb increases its sensitivity to proteolytic degradation (Nishinaka et al., 1997). There is experimental evidence for this mechanism. First, Rb is a substrate for CamK phosphorylation in serum-stimulated human fibroblasts (Takuwa et al., 1995) and second, phosphorylated Rb is rapidly degraded by a speci®c protease, the Spase, that is able to degrade both SP1 and Rb (Nishinaka et al., 1997). Most importantly, this SPase failed to degrade the underphosphorylated form of Rb, an observation that led the authors to suggest that phosphorylation possibly induces conformational changes that facilitate proteolytic degradation. Moreover, SPase is found in both the nuclear and the cytoplasmic compartments covering the separate subcellular localizations of the CamK II and IV (Hanson and Schulman, 1992). Thus, the activation

Retinoblastoma and Ca2+ calmodulin kinase cross talk F Sohm et al

of a such protease could account for the inhibition of transcription occurring in the presence of both CamKs and Rb molecules. The one observation for which there is currently no straightforward mechanistic model is the activation of SP1 by CamK. In our experiments the binding of the SP1 proteins (RCPs) to the RCE does not change when the cells are stimulated by high K+ (K+, 30 mM). However, overexpression of CamKs directly induces SP1-Gal4-driven transcription, and this data clearly shows that SP1 is a target for CamKs regulation. Hypotheses can thus be built either around phosphorylation of SP1 itself by CamK, or an indirect action involving phosphorylation of inhibitor/activator proteins. Little relevant data is available, but some recent work has shown that SP1-induced transcription is regulated by phosphorylation and so far two kinases have been shown to phosphorylate SP1. First, DNAdependent kinase can be involved, but this phosphorylation does not seem to alter SP1 activity (Jackson et al., 1990). Second, SP1 is also phosphorylated by Casein Kinase II which provokes an inhibition of DNA binding activity (Armstrong et al., 1997). Neither of these models ®t with our observations. So the question of whether SP1 is a direct or indirect (maybe through Rb and their relatives) target for the CamKs remains open. To address this question one would need to take into account the possibility of tissue speci®c regulations of the RCP family by second messengers such as Ca2+. Indeed, activation of the Rb/SP1 pathway will have important functional consequences according to the gene activated and the cellular context. Kim et al. (1991) have demonstrated that the transcriptional response of a given RCE to Rb expression is cell-type dependent and may be inhibitory or stimulatory. This brings us back to our initial consideration of the link between cell cycling and di€erentiation. Both pathways examined in this work, Rb and CamK, have been identi®ed as majors players in each process (Takuwa et al., 1995; Hollingsworth et al., 1993). The negative e€ect of Ca2+/calmodulin activity on Rb-dependent SP1 activity could thus provide a molecular explanation for a switch from cycling to di€erentiation in excitable cells such as neurones or secretory cells. Indeed, expression of voltage-dependent Ca2+ channels is a developmentally controlled process in both cell types (Spitzer, 1994). Thus, in non-di€erentiated cells with no voltage-dependent Ca2+ channels the Rb/Sp1 pathway will be able to activate c-fos and other genes directly involved in cell cycling (such as cyclin D1 which contains SP1 binding sites (Albanese et al., 1995)). In contrast, when the developmental switch to expression of voltage-dependent Ca2+ channels occurs the subsequent activation of CamKs will put a molecular brake on Rb/Sp1 transcription. Thus, according to promoter and cell context, integration of these signalling pathways can modify cell commitment. Preliminary experiments from our laboratory are in line with this interpretation since co-transfection studies with constitutively active CamKs repress transcription of genes such as c-Myc and cyclin D1 which are directly implicated in cell cycle progression (Loe‚er, Sohm, Antoine, unpublished). Conversely, inhibition of Ca2+ signalling that triggers apoptosis in neurones may also operate through this pathway.

Indeed, induction of CD1 (and other cell cycle genes controlled by Rb as well) has been proposed as an early step of neuronal apoptosis (Rubin et al., 1994). However, whether bioelectrical activity and subsequent Ca2+ entry allows survival of terminally di€erentiated neurones by repressing the Rb signalling pathway remains to be investigated.

Materials and methods Cell culture Mouse AtT-20/16-16 pituitary tumour cells were cultured in Dulbecco's Modi®ed Eagle Medium/F12 (DMEM/F12) supplemented with glutamine (286 mg/l), kanamycin (50 mg/ ml), penicillin (100 mg/ml), streptomycin (100 mg/ml) and 10% foetal calf serum (FCS). Cells were seeded in 6-well multi-dishes at a density of 2x106 cells/well and grown in a 5% CO2 humidi®ed incubator at 378C until 60% con¯uence. Cells were then serum-deprived for 18 ± 24 h prior experiments. Plasmids Reporter genes pFC700, pFC700DRCE, pFC99, pFC99DCRE, pFC72 and pFC53 contain the progressively deleted human c-fos promoter linked to the chloramphenicol-acetyl-transferase (CAT) reporter gene (Fisch et al., 1989). RCE-tk-CAT and the mutated RCE-tk-CAT (RCPtk-CAT) are described in Robbins et al. (1990). Expression vectors RSV-CamK II and RSV-CamK IV encode the constitutively active Calcium/calmodulin independent form of CamK II and IV respectively. RSVb-globin expresses the b-globin protein and is used as an internal control. All these constructs have been previously described (Sun et al., 1994). Gal4-SP1 contains the coding sequence for SP1, lacking the zinc ®nger-binding domain, fused to the GAL4 DNA-binding domain (1 ± 147). In B17Gal4 expression vector, the SP1 coding sequence is replaced by the bacterial B17 transcription factor coding sequence (Sif et al., 1993). The p107 and Rb expression vectors are described in Zhu et al. (1993). Statistical analysis Statistical signi®cance of data was assessed by means of analysis of variance (ANOVA), followed by the StudentNewman-Keuls test for multiple comparisons, using `Graphpad's Instat2' software. Data shown in ®gures are means of 3 ± 4 determinations performed in triplicate or quadruplicate. Transfection and enzyme assays Transfection experiments have been performed using a lipopolyamine-based method (TRANSFECTAM2, Promega) as previously described (Loe‚er and Behr, 1993). In co-transfection experiments, cells were incubated for 8 h with 1.5 mg of reporter gene and 1.5 mg of expression vector. Cells were then switched to fresh serum-free medium (DMEM/F12) for 12 h before exposure to appropriate pharmacological agents for 8 h or switched to fresh serumfree medium for 20 h in absence of treatment. In dose response experiments, cells were incubated with 1.5 mg of the reporter vector, and 0.5 mg of b-globin or CamK expressing vectors. Rb or p107 expression vector amounts varied between 0.2 ± 1 mg. Total DNA quantity of 3 mg/well was achieved with the corresponding empty expression vector. In order to avoid non speci®c e€ects of CamK or Rb/p107 proteins, the same protocol was realised

2767

Retinoblastoma and Ca2+ calmodulin kinase cross talk F Sohm et al

2768

using a CMV-luciferase reporter (+RB or p107 and/or +CamK II or IV). Results obtained with the CMV promoter were not signi®cantly di€erent between all conditions (n=3). For SP1-Gal4 experiments, in order to keep the amount of DNA constant, the quantities were as follows: Gal4-luciferase reporter vector, 0.75 mg; CamKs/b-globin: 0.5 mg; SP1-Gal4/ B17-Gal4: 0.75 mg; p107/Rb: 0.2 mg. CAT activity was determined by the method of Seed and Shee (1988). Luciferase activity was determined with the luciferase assay kit (Boehringer Manheim) following the manufacturer's instructions. The data are presented in fold induction relative to non stimulated controls. Nuclear extracts and gel mobility shift assays AtT20 were grown in 10 cm tissue culture dishes (Falcon) until 70 ± 80% con¯uence. Cells were serum-deprived for 24 h and treated with 30 mM potassium for 1 h. Nuclear extracts were isolated according to Dignam et al. (1983). Final protein concentration were typically 2 ± 3 mg/ml as determined by the Bio-Rad protein assay (Bio-Rad, Germany). Equivalent amounts of proteins (10 mg), 1 mg of poly (dI-dC) and a 23base pair 32P-labelled oligonucleotide containing the human cfos RCE sequence were incubated in binding bu€er (20 mM

HEPES, pH 7.9, 6 mM KCl, 2 mM DTT, 5 mM spermidine, 2% Ficoll and 8% glycerol) in a ®nal volume of 20 ml. The binding reaction lasted 20 min at room temperature. The DNA protein complexes were resolved on non denaturing 5% polyacrylamide gels. Supershift experiments were carried out using antibodies directed against SP1 (anti-SP1 polyclonal goat antibody (PEP-2) TransCruz2 SC-59X, Santa Cruz Biotechnology) and SP3 (anti-SP3 rabbit polyclonal antibodies (D-20) TransCruz2 SC-644X, Santa Cruz Biotechnology) and were added to the binding mixture 2 h prior adding the probe.

Acknowledgements The authors thank Dr S Brooks for critical review of the manuscript, C Nelson, L Le Personnic and F Herzog for their ecient technical support. The authors are grateful to Dr G Roeder (New York, USA), Dr RC Mulligan (Boston, USA), Dr TD Gilmore (Boston, USA), Dr ME Ewen (Boston, USA), and Dr R Maurer (Portland, USA) for the generous gift of recombinant material. This work was supported by ARC (grants n8 6089 and 9821 to JP Loe‚er).

References Albanese C, Johnson J, Watanabe G, Eklund N, Dzuy V, Andrew A and Pestell RG. (1995). J. Biol. Chem., 270, 23589 ± 23597. Angel P and Karin M. (1991). Biochem. Biophys. Acta., 1072, 129 ± 157. Antoine M, Gaiddon C and Loe‚er JP. (1996). Mol. Cell. Endocrinol., 120, 1 ± 8. Armstrong SA, Barry DA, Leggett RW and Mueller CR. (1997). J. Biol. Chem., 272, 13489 ± 13495. Barthel F and Loe‚er JP. (1995). J. Neurochem., 1, 41 ± 51. Buchkovich K, Du€y LA and Harlow E. (1989). Cell, 58, 1097 ± 1105. Chen LI, Nishinaka T, Kwan K, Kitabayashi I, Yokoyama K, Fu YH, Grunwald S and Chiu R. (1994). Mol. Cell. Biol., 14, 4380 ± 4389. Chen PL, Scully P, Shew JY, Wang JY and Lee WH. (1989). Cell, 58, 1193. Dash PK, Karl KA, Colicos MA, Prywes R and Kandel ER. (1991). Proc. Natl. Acad. Sci. USA, 88, 5061 ± 5065. DeCaprio JA, Ludlow JW, Lynch D, Furukawa Y, Grin J, Piwnica-Worms H, Huang CM and Livingston DM. (1989). Cell, 58, 1085 ± 1095. Dignam JD, Lebovitz RM and Roeder RG. (1983). Nucleic Acids Res., 11, 1475 ± 1489. Fisch TM, Prywes R, Simon MC and Roeder RG. (1989). Genes Dev., 3, 198 ± 211. Friend SH, Bernards R, Rogelj S, Weinberg RA, Rapaport JM, Albert DM and Dryja TP. (1986). Nature, 323, 643 ± 646. Fung YK, Murphree AL, T'Ang A, Qian J, Hinrichs SH and Benedict WF. (1987). Science, 236, 1657 ± 1661. Ghosh A, Ginty DD, Bading H and Greenberg ME. (1994). J. Neurobiol., 25, 294 ± 303. Hanson PI and Schulman H. (1992). Annu. Rev. Biochem., 61, 559 ± 601. Hardingham GE, Cruzalegui FH, Chawla S and Bading H. (1998). Cell calcium, 23, 131 ± 134. Hollingsworth RE, Chen Jr PL and Lee WH. (1993). Curr. Op. Cell Biol., 5, 194 ± 200. Horowitz JM. (1993). Genes Chrom. Cancer, 2, 124 ± 131. Jackson SP, MacDonald JJ, Lees-Miller S and Tjian R. (1990). Cell, 63, 155 ± 165. Johnson CM, Hill CS, Chawla S, Treisman R and Bading H. (1997). J. Neurosci., 17, 6189 ± 6202.

Karin M. (1992). Trends Biochem. Sci., 10, 418 ± 422. Kennett SB, Udvadia AJ and Horowitz JM. (1997). Nucleic Acids Res., 25, 3110 ± 3117. Kim SJ, Lee HD, Robbins PD, Busam K, Sporn MB and Roberts AB. (1991). Proc. Natl. Acad. Sci. USA, 88, 3052 ± 3056. Kim SJ, Onwuta US, Lee YI, Li R, Botchan MR and Robbins PD. (1992). Mol. Cell. Biol., 12, 2455 ± 2463. Lee WH, Bookstein R, Hong F, Young LJ, Shew JY and Lee EY. (1987a). Science, 235, 1394 ± 1399. Lee WH, Shew JY, Hong FD, Sery TW, Donoso LA, Young LJ, Bookstein R and Lee EY. (1987b). Nature, 329, 642 ± 645. Loe‚er JP and Behr JP. (1993). Methods Enzymol., 217, 599 ± 618. Matthews RP, Guthrie CR, Wailes LM, Zhao X, Means AR and McKnight GS. (1994). Mol. Cell. Biol., 14, 6107 ± 6116. Mihara K, Cao XR, Yen A, Chandler S, Driscoll B, Murphree AL, T'Ang A and Fung YK. (1989). Science, 246, 1300 ± 1303. Misra RP, Bonni A, Miranti CK, Rivera VM, Sheng M and Greenberg ME. (1994). J. Biol. Chem., 269, 25483 ± 25493. Nevins JR. (1992). Science, 258, 424 ± 429. Nishinaka T, Fu YH, Chen LI, Yokoyama K and Chiu R. (1997). Biochem. Biophys. Acta, 1351, 274 ± 286. Robbins PD, Horowitz JM and Mulligan RC. (1990). Nature, 346, 668 ± 671. Rubin LL, Gatchalian CL, Rimon G and Brooks SF. (1994). Curr. Opin. Neurobiol., 4, 696 ± 702. Seed B and Shee J. (1988). Gene, 67, 271 ± 277. Sheng M, McFadden G and Greenberg ME. (1990). Neuron, 4, 571 ± 582. Sheng M and Greenberg ME. (1990). Neuron, 4, 477 ± 485. Sheng M, Thompson MA and Greenberg ME. (1991). Science, 252, 1427 ± 1430. Sif S, Capobianco AJ and Gilmore TD. (1993). Oncogene, 8, 2501 ± 2509. Spitzer NC. (1994). Trends Neurosci., 17, 115 ± 118. Sun P, Enslen H, Myung PS and Maurer RA. (1994). Genes Dev., 8, 2527 ± 2539. Takuwa N, Zhou W, Kumada M and Takuwa Y. (1993). J. Biol. Chem., 268, 138 ± 145.

Retinoblastoma and Ca2+ calmodulin kinase cross talk F Sohm et al

Takuwa N, Zhou W and Takuwa Y. (1995). Cell Signal., 7, 93 ± 104. Udvadia AJ, Rogers KT and Horowitz JM. (1992). Cell Growth Di€er., 3, 597 ± 608. Udvadia AJ, Rogers KT, Higgins PD, Murata Y, Martin KH, Humphrey PA and Horowitz JM. (1993). Proc. Natl. Acad. Sci. USA, 90, 3265 ± 3269. Udvadia AJ, Templeton DJ and Horowitz JM. (1995). Proc. Natl. Acad. Sci. USA, 92, 3953 ± 3957.

Wang Y and Simonson MS. (1996). Mol. Cell. Biol., 16, 5915 ± 5923. Xu HJ, Hu SX, Hashimoto T, Takahashi R and Benedict WF. (1989). Oncogene, 4, 807 ± 812. Zhu L, Van den Heuvel S, Helin K, Fattaey A, Ewen M, Livingston D, Dyson N and Harlow E. (1993). Genes Dev., 7, 1111 ± 1125.

2769