Human cyclin D1 encodes a labile nuclear protein whose synthesis is directly induced by growth factors and suppressed by cyclic AMP

Cyclins were originally identified as proteins that accumu-late during the cell cycle and are degraded in mitosis (for reviews see Nurse, 1990; Pines and Hunter, 1990; Lewin, 1990). These cyclins, also referred to as mitotic cyclins, control entry of the cell into M-phase by modulating as a regulatory subunit the activity of the serine-threonine pro-tein kinase p34cdc2 in Schizosaccharomyces pombe and mammalian cells and the homologous CDC28 in Saccharo-myces cerevisiae. Another group of cyclins does not show the pattern of expression and destruction typical of the mitotic cyclins, and some members of this group have been shown to play a key role in the regulation of S-phase entry, most notably the CLN1, CLN2 and CLN3 genes in S. cere -visiae (for reviews see Reed, 1991; Nasmyth, 1990, 1991). Mammalian cells contain, besides the B-type mitotic cyclins required for the formation of the maturation promoting factor (MPF), another set of proteins which are structurally related to cyclins (for reviews see Hunter and Pines, 1991; Sherr, 1991; Lew and Reed, 1992). Because of their induc-tion in G1/S after growth factor stimulation of resting cells, these cyclins have been suggested to figure in G1⟶S pro-gression (Matsushime et al., 1991; Lew et al., 1991). The best-characterized protein of this catagory seems to be cyclin A, which forms complexes with both CDC2 kinase in S-phase and with the CDC2-related kinase CDK2 in G2 (Pagano et al., 1992). Cyclin A also interacts with the tumor suppressor gene product pRB and its kin p107 and appears to be directly involved in the regulation of E2F/DRTF-dependent transcription (Bandara et al., 1991; Mudryj et al., 1991; Devoto et al., 1992; Ewen et al., 1992; Faha et al., 1992; Shirodkar et al., 1992). In addition, expression of the nuclear cyclin A protein seems to be indispensable for the entry of mammalian cells into DNA synthesis and into mitosis (Pagano et al., 1992; Girard et al., 1991; Walker and Maller, 1991). In contrast, the function of the other putative G1-cyclins, including cyclins C, D1 and E, remains elusive (Matsushime et al., 1991; Lew et al., 1991; Xiong et al., 1991; Koff et al., 1991; Motokura et al., 1991).

Human cyclin D1 was identified by complementation cloning through expression of a human Ewing sarcoma cDNA library in S. cerevisiae cells that were deficient in the G1-cyclins CLN1, CLN2 and CLN3 (Koff et al., 1991). The gene encoding the murine homologue of cyclin D1, cyl-1, has been identified independently as a gene induced by colony-stimulating factor-1 (CSF-1) in macrophages (Matsushime et al., 1991). The kinetics of cyl-1 induction by CSF-1 suggest a function in S-phase commitment. On the basis of immunological cross-reactivity data, the cyl-1 protein has been suggested to associate with a CDC2-related protein kinase, perhaps a G1/S-specific CDK, rather than with p34CDC2 itself (Matsushime et al., 1991). In addition, cyclin D1 seems to be of particular interest because of its potential role in human tumorigenesis (Motokura et al., 1991; Lammie et al., 1991). It is part of a gene cluster (11q13) that is amplified in human breast and squamous cell carcinomas and translocated in parathyroid adenoma (bcl-1 breakpoint, PRAD1).

In the present study, we have addressed the question as to whether cyclin D1 induction is CSF-1 specific or also observed in a different growth factor-regulated system, i.e., stimulated human diploid fibroblasts, and which signal transduction pathways might be involved in the regulation of cyclin D1 expression. We were also interested to study whether cyclin D1 expression might be cell cycle-depen-dent or rather directly regulated by growth factors. In addition, we have measured the turnover of cyclin D1 pro-tein and determined its subcellular location as a first step to elucidate its function as a putative cell cycle regulator.

WI-38 and RDES cells were cultured in Dulbecco-Vogt modified Eagle’s minimum essential medium (DMEM) supplemented with 10% fetal calf serum (FCS), 0.5% glucose, penicillin (100 i.u./ml) and streptomycin (100 μm/ml). EGF, IGF-1, TPA, dexamethasone, forskolin and dibutyryl-cAMP were obtained from Sigma (St. Louis, MO), PDGF from Gibco-BRL/Life Technologies (Eggen-stein, Germany) and IGF-1 from Serva (Heidelberg, Germany).

WI-38 cells were seeded at a density of 10₄ cells/cm₂ 24 h prior to starvation in serum-free medium for 72 h and stimulated with 10% FCS for the indicated times. In some experiments protein synthesis was inhibited by the addition of cycloheximide (5μg/ml; Sigma) to the cells 1 h prior to serum stimulation. To synchro-nise cells in S-phase, semi-confluent WI-38 cells were blocked with 2 mM thymidine for 15 h, released from the block by wash-ing in phosphate buffered saline (PBS) three times and grown for an additional 9 h in complete medium containing 0.24 × 10_-4 M thymidine and deoxycytidine prior to the addition of aphidicolin (Sigma, St. Louis, MO) at a concentration of 5 μm/ml for 15 h (Heintz et al., 1983). The cells were released from the block by washing in PBS three times and grown for the indicated times in complete medium.

WI-38 cells, grown on 25 cm₂ dishes in parallel to the cells for RNA isolation, were trypsinised, washed once with PBS and fixed in ice-cold 75% ethanol overnight at 4_°C. After washing once with PBS, fixed cells were stained for 15 min in Hoechst 33258 stain-ing buffer (100 mM Tris-HCl pH 7.4, 154 mM NaCl, 1 Mm CaCl2, 0.5 mM MgCl2, 0.1%, (v/v) Nonidet-P40, 0.2%, (w/v) BSA, 2 μm/ml Hoechst 33258) (Rabinovitch et al., 1988). Stained cells were analysed in a FACS-STAR Plus (Becton Dickinson) using an UV-laser excitation at 325 nm. The fluorescence was amplified linearly.

RNA was isolated according to a modified method of Chirgwin et al. 1979). cDNA synthesis was done with 2-4 μg of total RNA in a reaction volume of 40 μl containing 1.25 μg of oligonu -cleotide (dT)15, 1 mM each dATP, dGTP, dCTP, dTTP, 30 units RNasin, 50 mM Tris-HCl, pH 8.3, 75 mM KCl, 3 mM MgCl2, 10 mM DTT, 200 units MMLV reverse transcriptase (Gibco-BRL/Life Technologies, Eggenstein, Germany), and incubated for 1 h at 37°C. PCR reactions were performed with 1–2 μl of the cDNA reaction mixture in the presence of 0.25 μCi of [α-32P]dCTP in a volume of 50 μl containing 200 μM each dATP, dGTP, dCTP, dTTP, 1 μM each 5′-and 3′-primer and 1.25 units Taq DNA polymerase (Boehringer, Mannheim, Germany; Saiki et al., 1988). The annealing temperature was 60°C for all primer pairs, the number of PCR cycles used was 20 for cyclin D1, 25 for cyclin B, 26 for ODC, 22 for c-fos and 16 for GAPDH. The oligonucleotide primer pairs had the following sequences (num-bers refer to nucleotide positions in the respective sequences quoted; some primers contain in addition restriction sites at their 5′-end): Cyclin D1 (Xiong et al., 1991): (5′-primer), 5′-GCC-GAATTCTGGATGCTGGAGGTCTGCGAGGAA-3′ (341 to 364); (3′-primer), 5′-GCCGAATTCGGCTTCGATCTGCTC-CTGGCAGGC-3′ (880 to 905). Cyclin B (Pines and Hunter, 1989): (5′-primer): 5′-CCATTATTGATCGGTTCATGCAGA-3′ (817 to 840); (3′-primer): 5′-CTAGTGGAGAATTCAGCTGTG-GTA-3′ (1378 to 1401). ODC (Hickok et al., 1987): (5′-primer): 5′-ACTGGATCCTCATGAACAACTTTGGTAAT-3′ (86 to 105); (3′-primer): 5′-ACTGAATTCTGAAAGCTGATGCAACATAG-3′ (921 to 940). c-fos (Van Straaten et al., 1983): (5′-primer): 5′-ACGCAGACTACGAGGCGTCA-3′ (908 to 927); (3′-primer): 5′-TTCACAACGCCAGCCCTGGA-3′ (1960 to 1979). GAPDH Arcari et al., 1984): (5′-primer): 5′-CGTCTTCACCACCATG-GAGA-3′ (360 to 379); (3′-primer): 5′-CGGCCATCACGC-CACAGTTT-3′ (640 to 659). A 10 μl sample of the PCR reac-tion mixture were separated on a 5% polyacrylamide gel, dried and exposed to Kodak X-ray film overnight. Quantitation was per-formed by b-scanning of the gels using a Molecular Dynamics PhosphorImager. Data were corrected using the GAPDH signal as the standard.

A full-length human cyclin D1 clone was isolated from a HepG2 cDNA library (Stratagene) by hybridisation with a synthetic oligonucleotide derived from the published sequence and verified by nucleotide sequencing. A human cyclin B clone (Pines and Hunter, 1991) was kindly provided by J. Pines and T. Hunter (Salk Institute, San Diego, CA). Human cyclin E, C and A cDNAs were cloned after PCR amplification of reverse-transcribed HeLa cell RNA and verified by DNA sequencing. The primer pairs used for the PCR reactions were as follows (numbers refer to nucleotide positions in the respective cDNA): Cyclin E (Koff et al., 1991): (5′-primer): 5′-ATGGCTCGAGCCATGAAGGAG-GACGGCGGC-3′ (-2 to 18). (3′-primer): 5′-AACGGAATTCG-GTG GTCACGCCATTTCCGG-3 ′ (1174 to 1195). Cyclin C (Xiong et al., 1991): (5′-primer): 5′-CTTCAGGATCCTATG-GTCGCTCCGCGGCCG-3′ (-2 to 16). (3′-primer): 5′-GAC-TAAGCTTCTATGGAATTCTTCGGAATG-3′ (915 to 935). Cyclin A (Wang et al., 1990): (5′-primer): 5 ′-GTCAGGATCCT-GATGTTGGGCAACTCTGCG-3′ (106 to 125); (3′-primer): 5′-GTCACGGATCATGTTGGGCAACTCTGCG-3′ (1389 to 1406).

In vitro translations were carried out with the Promega TNT kit using 1 μm of template DNA.

Human cyclin D1 was expressed in bacteria using the GST-expression system (Pharmacia). Briefly, a BamHI fragment encod-ing the entire open reading frame was subcloned into the BamHI site of pGEX-3X. Expression and induction in E. coli (JM-101) were performed according to the instructions of the manufacturer. Bacteria were collected by centrifugation, pellets were resus-pended in PBS/1% Triton X-100 and lysed by sonication. After centrifugation, the pellet was washed twice with 6 M urea in PBS and resuspended in 8 M urea in PBS. After centrifugation the supernatant was diluted with 9 volumes of PBS and dialysed against PBS/1% Triton X-100 for 10-16 h. Finally, the sample was loaded onto a GST-affinity column and washed with PBS/1% Triton X-100. Bound GST-Cyclin D1 was eluted with a buffer containing 50 mM Tris-HCl, pH 7.5, and 10 mM glutathione.

Bacterially expressed fusion-protein was used for the production of antisera. Castor rex rabbits were immunised with 200 μm of protein emulsified in ABM-S (Linaris, Germany) and boosted at three-week intervals.

A total of 4×10₅ cells were seeded in a 25 cm2 dish 24 h prior to labelling. Cells were washed twice in PBS and incubated in methionine-free DMEM with 0.2 mCi of [35S]methionine (Amer-sham, ∼1400 Ci/mmol) for 60 min. Cells were then washed twice in chilled PBS and lysed by the addition of 0.8 ml RIPA-buffer (1% Triton X-100, 1% sodium deoxycholate, 0.1% sodium dodecyl sulfate (SDS), 0.15 M NaCl and 0.01 M Tris-HCl, pH 7.4) (Curran et al., 1984). Lysates were clarified by centrifugation at 45,000 g for 30 min at 4°C. Samples were incubated with 8 ml of rabbit anti-CyclinD1 serum overnight at 4°C, followed by incu-bation for another 60 min with 40 ml Protein A-Sepharose CL-4B (Pharmacia). Immune complexes were centrifuged, pellets were washed eight times in RIPA-buffer containing 500 mM NaCl, resuspended, boiled in SDS sample buffer and analysed on dis-continuous 12.5% SDS-polyacrylamide slab gels followed by fluorography. Cell fractionation into a nuclear and cytoplasmic fraction was carried out as described (Curran et al., 1984). The purity of the subcellular fractions was examined by measuring lac-tate dehydrogenase activity as a cytoplasmic marker (Bergmeyer, 1983) and the DNA content as a nuclear marker (Richardson, 1974).

A 100 μm sample of protein per lane was separated on 12.5% SDS-polyacrylamide gels and transferred onto nitrocellulose. Fil-ters were blocked with 5% non-fat dry milk in PBS for 30 min at room temperature and then incubated with a 1:750 dilution of the antibody for 2 h. After several washes with 0.1% Tween 20 in PBS the filter was incubated with a biotinylated swine a-rabbit IgG diluted 1:750 (Dako, Germany) for 30 min, washed several times in the same buffer as before and incubated with a 1:500 dilution of a streptavidin/horseradish peroxidase complex (Dako, Germany). After extensive washing the filter was developed using diaminobenzidine with metal ion enhancement (Harlow and Lane, 1988).

Cells on coverslips were fixed with 3.7% formaldehyde in PBS. Fixed cells were washed twice with cold PBS, incubated in 100 mM NH4Cl in PBS for 10 min and permeabilised for 5 min with 0.5% Triton X-100 in PBS. After 30 min incubation in 3% BSA in PBS the coverslips were exposed to a 1:150 dilution of the a-cyclin D1 antibody for 1 h at ambient temperature. Cells were washed as before and incubated with a 1:250 dilution of a cy3-conjugated a-rabbit antibody (Dianova, Germany). Counterstain-ing of DNA was carried out for 5 min with Hoechst 33258 (2 μm/ml in PBS). The cells were observed with a Leitz Diavert microscope and photographed using a Kodak TMX 400 film.

The human fibroblast line WI-38 was chosen for this study because it closely resembles normal fibroblasts. These cells possess a diploid karyotype, have the ability to enter a G0 state and re-enter the cell cycle, and also to cease prolifer-ation after reaching the senescent state. WI-38 cells kept in serum-depleted medium for 3 days are practically com-pletely arrested in a G0/G1 state (see FACS analysis in Fig. 1a). After stimulation with 10% FCS, the cells enter DNA synthesis at 18-21 h and reach the G2-phase at approxi-mately 28 h post-stimulation as shown both by FACS analy-sis (Fig. 1a) and by 5-bromodeoxyuridine-incorporation experiments (our unpublished observations).

RNA was isolated from cells at different times after stimulation, transcribed into cDNA and amplified by PCR in the presence of 32P-labelled precursor, so that a low number of cycles was sufficient to detect the amplified product (Mumberg et al., 1991), thus allowing a quanti-tative evaluation of the results by b-radiation scanning. To verify that the assay conditions (e.g. enzyme, primers, nucleotides) were indeed non-limiting, different numbers of PCR cycles were performed in each experiment (not shown). In all experiments, we also determined the levels of glyceraldehyde phosphate dehydrogenase (GAPDH) mRNA as a standard to correct for fluctuations in exper-imental conditions. The long-term kinetics shown in Fig. 1b and the short-term kinetics depicted in Fig. 1c demon-strate that cyclin D1 mRNA is induced after serum-stim-ulation of serum-deprived cells. Increased mRNA levels are clearly detectable after 2 h and reach a maximum at approximately 10 h post-stimulation ("8-fold induction). This induced level is similar to the expression of cyclin D1 mRNA seen in normally cycling WI-38 cells (Fig. 2). We also performed the stimulation experiment in the pres-ence of cycloheximide (5 μg/ml) with virtually identical results (Fig. 2), indicating that cyclin D1 mRNA induc-tion does not require protein synthesis and is therefore directly induced by serum growth factors. In the same experiment, other growth factor and cell cycle regulated genes were included for comparative purposes. The data displayed in Fig. 1b and c show that cyclin D1 mRNA expression peaks at a similar time as ornithine decar-boxylase (ODC), following the induction of c-fos (maximum at 1 h) and preceding the maximum expression of the G2-specific cyclin B (maximum at ≤ 28 h).

To address the question as to whether cyclin D1 expression might be regulated during the cell cycle, we synchronized cells in early S-phase by a thymidine-aphidicolin block and measured mRNA levels after release from the block. The FACS data in Fig. 3a show that the blocked cells (0 h) had, as expected, a 2C DNA content that rapidly increased after release from the block. At 3 h most cells were in mid-late S-phase, reached G2 after 6-9 h, re-entered a new cell cycle after "12 h and were found in the second S-phase after "18 h. The RNA expression data exhibited in Fig. 3b show that under these conditions the levels of cyclin D1 mRNA remained very similar with <2-fold fluctuations. In the same experiment, cyclin B mRNA levels showed a clear induc-tion in G2, reaching peak levels between 9 and 12 h after release from the thymidine-aphidicolin block. From these observations we conclude that cyclin D1 mRNA expression is not subject to any significant regulation during the cell cycle.

To be able to identify and characterise cyclin D1 protein in WI-38 cells, we decided to raise antibodies against recombinant cyclin D1 expressed as a glutathionine S-trans-ferase (GST)-fusion protein in E. coli. The antiserum obtained from one rabbit (K 44) proved to be highly reactive and specific for human cyclin D1 in all immunoassays used. Fig. 4a shows the immunoprecipitation of 35S-labelled in vitro translated cyclin D1 by the K44 antiserum, but not by the corresponding preimmune serum. In addition, the precipitation of the 35S-labelled cyclin D1 was completely blocked by the addition of an excess of recombinant GST-cyclin D1. The bottom panel of Fig. 4a demonstrates the specificity of the antiserum, since it does not precipitate in vitro translated cyclin A, C or E. A single prominent protein of ≈ 39 kDa (p39) was also precipitated by the K44 antibodies from [35S]methionine-labelled WI-38 cells, while no precipitation was observed with the preimmune serum (Fig. 4b). Again, precipitation of p39 was completely abolished by an excess of recombinant GST-cyclinD1, indi-cating that the precipitated protein is indeed cyclin D1. Finally, the antibodies were tested by Western blotting using extracts from the Ewing sarcoma cell line RDES (Fig. 4c). Glioblastoma cells contain very high levels of cyclin D1 mRNA and were originally used to clone the human cyclin D1 cDNA. In agreement with the immunoprecipita-tion data, a single prominent ≈ 39 kDa protein was detected by the K44 antibodies, and the immunoreaction was specif-ically abolished by recombinant GST-cyclin D1, but not by GST alone. The determined size of 39 kDa is in close agree-ment with the predicted size of 38.9 kDa (Xiong et al., 1991). We therefore conclude that the K44 antiserum specifically detects cyclin D1 in human cells.

The K44 antiserum was then used to measure the synthe-sis of cyclin D1 after serum stimulation of resting WI-38 cells in an experimental setup identical to that in Fig. 1. The data depicted in Fig. 5a and the quantitation in Fig. 5b show that the induction of cyclin D1 protein synthesis is very similar to that of the mRNA induction (compare with Fig. 1), the peak synthesis seen at 9 h post-stimulation. Fig. 5a also shows an immunoprecipitation from RDES Ewing sarcoma cells to illustrate the high rate of cyclin D1 syn-thesis in the WI-38 fibroblasts. We also studied the turnover of cyclin D1 in serum-stimulated WI-38 cells by perform-ing pulse-chase experiments (Fig. 5c). Cells were metabol-ically pulse-labelled with [35S]methionine for 60 min and then incubated in the presence of unlabelled methionine for various times. Fig. 5c displays a quantitative evaluation of this experiment. In the graph depicted the respective loga-rithmic (ln) values were plotted against the time in order to obtain a straight line, which is a prerequisite for an accu-rate calculation of protein half-life. The data show a con-stant exponential decline of labelled cyclin D1 levels over the 4-h chase period with a half-life of "38 min, indicating that cyclin D1 is a labile protein in WI-38 cells.

A significant induction of cyclin D1 protein synthesis was not only observed after serum stimulation, but also by treat-ing quiescent WI-38 cells with purified growth factors (Fig. 6a), such as epidermal growth factor (EGF, 5-fold induction), basic fibroblast growth factor (bFGF, 5-fold), platelet-derived growth factor (PDGF, 5-fold) and insulin-like growth factor-1 (IGF-1, 3-fold). We also tested the effect of various compounds that are known to stimulate certain intracellular signal transduction pathways. The data dis-played in Fig. 6a show that the phorbol ester, phorbol myristate acetate (TPA), a well-characterised inducer of protein kinase C, gave a 2.5-fold induction on cyclin D1 synthesis, while dexamethasone had no detectable effect. Increasing the intracellular level of cAMP, on the other hand, resulted in a repression of cyclin D1 synthesis, in both untreated and EGF-stimulated cells. This repression was observed after exposing the cells to the membrane-per-meable cAMP analogue dibutyryl-cAMP (dBcAMP) as well as after treatment with the adenylate cyclase activator forskolin (Seamon et al., 1981). These results indicate that both the basal expression and the induced expression of cyclin D1 are under negative control of cAMP-mediated signals.

We finally addressed the question of the subcellular local-isation of cyclin D1 in WI-38 cells. For this purpose, nor-mally cycling cells were metabolically labelled with [35S]methionine and fractionated into a nuclear and a cyto-plasmic fraction prior to immunoprecipitation (Fig. 6). The subcellular fractionation was controlled by determining the distribution of lactate dehydrogenase (LDH) as a cytoplas-mic protein and of DNA as a nuclear marker. These analy-ses showed that 93.9% of LDH activity were found in the cytoplasmic fraction, while DNA was detected exclusively in the nuclear fraction (the amount of DNA in the cyto-plasmic fraction was below the detection limit of the assay; see bottom of Fig. 7). Under these conditions, cyclin D1 cofractionated with the DNA, i.e. it was found specifically in the nuclear fraction. We cannot decide at the moment whether the minor amount of antibody-reactive protein in the cytoplasmic fraction reflects the in vivo situation or, what appears more likely, is due to the presence of a small amount of nuclear material in the cytoplasmic fraction.

We also studied the subcellular localisation of cyclin D1 by immunostaining. Cycling WI-38 cells were double-stained with the K44 antiserum and Hoechst 33258. Fig. 8 shows the results of this experiment at two different mag nifications. It is obvious that the immunostaining of cyclin D1 coincides with the DNA staining by Hoechst 33258, providing further evidence for a nuclear localisation of the protein. We generally observed some differences in the staining of individual cells, as can also be seen in Fig. 8. These differences do not seem to be related to cell cycle progression, since the analysis of cells released from an S-phase block did not show any significant fluctuation of the level of cyclin D1 protein throughout the cell cycle (measured as in Fig. 8; data not shown). Fig. 9 shows a number of controls that prove the specificity of the observed staining. In Fig. 9a the corresponding preimmune serum was used and this gave rise to background staining only. Furthermore, the staining could be abolished by the addition of an excess of recombinant GST-cyclin D1 (Fig. 9c), but not by competition with recombinant cyclin A, C or E fusion proteins (Fig. 9b, d, e). Taken together, the data derived by subcellular fractionation as well as by immunos-taining provide strong evidence that cyclin D1 is a nuclear protein in WI-38 cells.

The human cyclin D1 gene was originally identified by virtue of its ability to rescue S. cerevisiae cells with a defect in G1 ⟶ S progression (Xiong et al., 1991). Although this complementation has proved very useful to identify and clone new mammalian cyclin-related genes, it cannot pro-vide conclusive information about their function in mam-malian cells. This is demonstrated, for instance, by the fact that complementation of a CLN deficiency, leading to a G1-arrest, can be achieved by the mitotic cyclin B (Lew et al., 1991). Apart from the high expression of cyclin D1 in a human Ewing sarcoma cell line and a considerably lower level of expression in HeLa cells, no data are available regarding the expression and regulation of cyclin D1. Its murine counterpart, cyl-1, has been shown to be expressed at high levels in macrophages, where its expression is dependent on and regulated by CSF-1 (Matsushime et al., 1991). No cyl-1 mRNA was detectable in other cell lines of the lymphoid and myeloid lineages, suggesting that cyl-1 expression may be specific for CSF-1-stimulated macrophages (Matsushime et al., 1991). As a first step to elucidate the function of cyclin D1 in human cells we there-fore analysed its expression in a different growth factor-regulated system, i.e. serum-stimulated human diploid fibroblasts and analysed a potential cell cycle dependence of cyclin D1 expression in the same cell system. We fur-thermore studied the metabolism of cyclin D1 protein and its subcellular localisation.

Our data clearly show that both cyclin D1 mRNA and protein are regulated by serum and a number of purified growth factors, including EGF, bFGF, PDGF and IGF-1. In the absence of growth factors, cyclin D1 expression drops and is reinduced by the addition of FCS (Figs 1, 2, 5), even in the presence of cycloheximide (Fig. 2) at a concentra-tion that blocks protein synthesis by >90% (our unpublished observation). These observations indicate that cyclin D1 is directly induced by a signal transduction cascade triggered by the interaction of the cell with a growth factor, and that the continuous presence of growth factors is required to maintain cyclin D1 expression at a high level. This is sup-ported by the observation that cyclin D1 synthesis was also serum-inducible in growth factor-deprived cells arrested in S-phase by thymidine/aphidicolin treatment (data not shown). In addition, our conclusion is in agreement with the findings made with CSF-1-stimulated murine macrophages cited above (Matsushime et al., 1991). Our studies also suggest that cyclin D1 expression in not sub-ject to significant regulation during the cell cycle (Fig. 3). Obviously, these results have to be interpreted with some caution because of possible synchronisation-related influ-ences. The observed expression of cyclin B in G2, how-ever, is in perfect agreement with the published data (Pines and Hunter, 1989), supporting the validity of the results obtained. It would thus seem that these two cyclins are reg-ulated in different ways: cyclin B expression is clearly linked to the cell cycle, while cyclin D1 is directly con-trolled by growth factors, as are the genes of the immedi-ate early response to growth factors (for a review see Her-shman, 1991). This result is intriguing, as it places cyclin B and D1 into different categories not only on structural grounds, but also by way of fundamental differences in the regulation of their expression. In addition, this mechanism of cyclin D1 regulation may establish a direct link between growth factor stimulation and the cell cycle, provided a function of cyclin D1 in cell cycle progression can be proven. Both, its induction ≈ 8 h prior to S-phase entry (Figs 1, 5; Matsushime et al., 1991), as well as its association with a p34CDC2-related protein rather than p34CDC2 itself (Matsushime et al., 1991), point to a potential function in G1 ⟶ S progression.

In contrast to our results, Motukura et al. (1991) have reported an elevated level of PRAD-1 mRNA in HeLa S3 cells after release from a thymidine-aphidicolin block. We could not reproduce this result with WI-38 cells in several independent experiments, although both FACS analyses and cyclin A and B expression (Fig. 3 and data not shown) showed the expected results. On the other hand, it is known that the cell cycle control mechanisms in HeLa cells are not normal, e.g. due to the expression of papilloma virus protein E7 forming a complex with pRB. We therefore con-clude that our results obtained with WI-38 cells reflect the natural situation more closely.

Another question addressed in this study concerns the signal transduction pathways involved in cyclin D1 regula-tion. As a first step in this direction we have analysed the effect of several compounds that are able to trigger certain signal transduction pathways. In this context, the induction of cyclin D1 by both TPA and EGF deserves particular attention. TPA is a well-known inducer of protein kinase C (PKC), which is physiologically stimulated by the phos-pholipid breakdown product diacylglycerol (Castagna et al., 1982; Nishizuka, 1988). Although it cannot be ruled out that TPA may also have other effects, the observed induc-tion by TPA is a good indication that PKC is involved in the induction of cyclin D1. On the other hand, EGF does not lead to the breakdown of phosphatidyl inositol phos-phate and the ensuing stimulation of PKC (Besterman et al., 1986), pointing to the existence of a second signal trans-duction pathway that can induce cyclin D1 expression. The clarification of these questions requires the performance of a number of different experiments and has therefore to remain the subject of future studies. A particularly inter-esting observation is the repression of cyclin D1 synthesis by compounds that increase intracellular cAMP levels either directly, i.e. dBcAMP, or indirectly via the activa-tion of adenylate cyclase, i.e. forskolin (Seamon et al., 1981). It remains to be shown whether this effect is medi-ated via the cAMP-inducible protein kinase A. Our find-ings suggests that multiple, in part antagonistic, pathways (as in the case of EGF and forskolin) act in concert to con-trol cyclin D1 expression. Analysis of the mechanisms underlying this regulatory network will be another intrigu-ing subject for future investigation. p34CDC2 and cyclin A have both been localised to the nucleus. The same is true for a number of other proteins that have been shown or are likely to have regulatory func-tions around the G1/S-boundary, such as pRB, p107, p53 or E2F/DRTF. To address the question as to whether cyclin D1 may also be a component of this regulatory apparatus, its subcelluar location has to be known. We therefore devel-oped an antibody against a recombinant fusion protein that is suitable for the specific detection of cyclin D1 in differ-ent assays, including immunoprecipitation and immunos-taining (Figs 4, 8). The specificity of the K44 antiserum is not doubted, since its reaction in all assays used can be abolished by the addition of recombinant cyclin D1 fusion protein, but not by other competitors, e.g. cyclin A, C and E. These antibodies were used to determine the subcelluar location of cyclin D1 by cell fractionation and immunos-taining. Both assays gave unambiguous results and showed that most, if not all, cyclin D1 protein is nuclear (Figs 6-8). Cyclin D1 is therefore located in the same compartment as the other proteins mentioned above and might thus be involved in related regulatory events in the nucleus, which may either directly or indirectly affect gene transcription. Such a putative regulatory role of cyclin D1 is also sup-ported by its short half-life of <60 min, which is a hall-mark of regulatory proteins. If this is true there should be another mechanism apart from the regulation of expression that modulates the function of cyclin D1, such as post-tran-scriptional modification, since its level remains largely con-stant during the cell cycle. Alternatively, one could envis-age a function for cyclin D1 exclusively in G1, even though its expression remains high and its state of activity unchanged throughout the cell cycle. The short half-life of the protein may only come into play when the cell has to react quickly to the withdrawal of growth factors by the rapid destruction of cell cycle-promoting molecules. Future work will have to establish whether cyclin D1 falls into this class. The data presented in this study should provide a useful basis for these investigations.

We are grateful to T. Hunter and J. Pines for providing the human cyclin B cDNA clone. This work was supported by the Deutsche Forschungsgemeinschaft (Mu601/5-2 and Mu601/7-1) and the Dr. Mildred Scheel-Stiftung für Krebsforschung. A.S. and S.B. are the recipients of fellowships from the Graduiertenkolleg “Zell-und Tumorbiologie” at the Philipps-Universität Marburg.