¹Cancer Research Division, Lilly Research Laboratories, A Division of Eli Lilly and Company, Indianapolis, IN 46285-0424, USA

²Research Technologies and Proteins, Lilly Research Laboratories, A Division of Eli Lilly and Company, Indianapolis, IN 46285-0424, USA

³Sphinx Pharmaceuticals, A Division of Eli Lilly and Company, Durham, North Carolina, NC 27707, USA

Correspondence to: R Nagaraja Rao, Lilly Corporate Center, Drop Code 0424, Indianapolis, IN 46285-0424, USA

Cyclin D1 gene overexpression is a frequent event in a number of human cancers. These observations have led to the suggestion that cyclin D1 alterations might play a role in the etiology of cancer. This possibility is supported by the finding that transfection of mammalian cells with cyclin D1 can accelerate progression through the G1 phase of the cell cycle. Moreover, cyclin D1 can function as an oncogene by cooperating with activated Ha-ras to transform primary rat embryo fibroblasts (REFs). In addition, cyclin D1 transgenics develop hyperplasia and neoplasia of the thymus and mammary gland. We have constructed a novel fusion gene consisting of full-length human cyclin D1 and cdk4 genes. This fusion gene was expressed in insect cells and the fusion protein was shown to be enzymatically active. The fusion gene was expressed in mammalian cells under the control of tet-repressor. This fusion gene immortalized primary REFs, and cooperated with activated Ha-ras to transform primary REFs, in terms of anchorage-independent growth in vitro and formation of tumors in vivo. Utilizing a tet-regulated gene expression system, we have shown that proliferation of stably transfected primary REFs in vitro and in vivo is dependent on the continued expression of the cyclin D1-cdk4 fusion gene. These cell lines could be useful in the discovery of novel cancer therapeutics to modulate cyclin D1.cdk4 activity.

cyclin D1; cdk4; transformation; cell models

cdk, cyclin-dependent kinase; D1.k4, cyclin D1 complexed with cdk4; D1-k4, cyclin D1 fused with cdk4; IP, immunoprecipitation; IB, immunoblot; REFs, rat embryo fibroblasts; Rb, retinoblastoma protein; FCS, fetal calf serum; tet, tetracycline

Introduction

Progression of cells through the cell cycle is controlled, in part, by cyclins complexing with cyclin-dependent kinases (cdk's) resulting in enzymatically active kinases. These complexes ensure the fidelity of DNA synthesis and cell division, and their activity is regulated at the level of synthesis, destruction, inhibitory/activating phosphorylation, and inhibitory/activating proteins (Morgan, 1995; Sherr and Roberts, 1995). These complexes include cyclins D1,2,3/cdk4,6 which are involved in G1 progression; cyclin E/cdk2 which acts at the G1 to S transition; cyclin A/cdk2 which operates during S; cyclin A/cdk1 which functions in late S and G2 stages; and cyclin B/cdk1 which functions during G2-M stages (Elledge, 1996; Morgan, 1995). There is considerable evidence suggesting that dysregulation of cyclins and cdk's, particularly those controlling G1 progression, may be at the root of many forms of human cancer (Hall and Peters, 1996; Sellers and Kaelin, 1997; Sherr, 1996).

Overexpression of the G1 cyclin, D1, is common in human cancers (Donnellan and Chetty, 1998; Hall and Peters, 1996; Sellers and Kaelin, 1997; Sherr, 1996). Cyclin D1 overexpression has been seen in breast, colorectal, bladder, soft tissue, uterine, melanoma, hepatocellular, and small cell lung cancers (Hall and Peters, 1996; Sherr, 1996). Overexpression of cdk4 has been reported to occur in Ewing's sarcoma (Ladanyi et al., 1995), gliomas and soft tissue sarcomas (Rollbrocker et al., 1996; Sherr, 1996) and astrocytomas (Petronio et al., 1996). In addition, p16, which is a negative regulator of cdk4,6 is frequently mutated or deleted in a large variety of human tumors (Biggs and Kraft, 1995; Hirama and Koeffler, 1995; Liggett and Sidransky, 1998).

The primary in vivo substrate of cdk4 appears to be the retinoblastoma protein (Rb) (Dowdy et al., 1993; Ewen et al., 1993; Kato et al., 1993). Phosphorylation determines its ability to regulate its downstream effector, the E2F transcription factor (Dyson, 1998; La Thangue, 1996; Neuman et al., 1994; Nevins, 1992, 1998). E2F exerts its influence by controlling the transcription of a number of genes that are involved in cell cycle progression and in initiation of DNA synthesis (DeGregori et al., 1995; Dyson, 1998; La Thangue, 1996; Nevins, 1998). The Rb gene is inactivated in a large variety of human cancers (Hall and Peters, 1996; Sellers and Kaelin, 1997). The overall incidence of alterations in the p16-I cyclin D1.cdk4-I Rb pathway in lung and esophageal carcinomas is nearly 100% (Sherr, 1996), highlighting the importance of this pathway for the development of novel cancer therapies.

Transfection of the cyclin D1 gene into mammalian cells alters the G1 phase of the cell cycle (Jiang et al., 1993; Quelle et al., 1993; Resnitzky et al., 1994). In addition, cyclin D1 and cyclin D2 genes can function as oncogenes (Hinds et al., 1994: Kerkhoff and Ziff, 1995; Lovec et al., 1994b). Furthermore, cyclin D1 transgenic mice develop mammary hyperplasia and carcinoma (Wang et al., 1994), or B-cell lymphoma, in cooperation with c-myc (Lovec et al., 1994a), or thymic hyperplasia (Robles et al., 1996), or dysplasia of the tongue, esophagus and forestomach (Nakagawa et al., 1997). Significance of the p16-I cyclin D1.cdk4-I Rb pathway in cancer is enhanced by the recent observations showing that Ras-mediated growth inhibition is abrogated in Rb-null mouse embryo fibroblasts (Peeper et al., 1997) and that Ras-mediated tumorigenesis is dramatically decreased in cyclin D1-null mice (Robles et al., 1998).

The importance of cdk4 kinase activity in human cancer has led to a search for inhibitors of this pathway (Arber et al., 1997; Arguello et al., 1998; Carlson et al., 1996; Gray et al., 1998; Kim and Miller, 1998; Sandig et al., 1997). Proper evaluation and development of cdk inhibitors is facilitated by having appropriate cell/animal models. Instead of coexpressing cyclin D1 and cdk4 genes from the same vector, we constructed a fusion gene (D1-k4) consisting of full-length human cyclin D1 and human cdk4 genes. Utilizing the tet-regulated gene expression system (Gossen and Bujard, 1992; O'Brien et al., 1997), we show that the fusion gene immortalizes/transforms primary REFs and that the proliferation is dependent on the continued expression of the D1-k4 fusion gene.

Results

Construction of a cyclin D1-cdk4 fusion gene and its cloning in a tet-controlled expression system

Human cyclin D1 (Lovec et al., 1994b) and cyclin D2 (Kerkoff and Ziff, 1995) genes, presumably by increasing the kinase activity of cdk4/6, cooperate with activated Ha-ras in transforming primary rat embryo fibroblasts. In order to study the contribution of human cyclin D1 and human cdk4 in the immortalization of primary REFs, we chose to introduce both these genes simultaneously into the same cell under the control of tet repressor (Gossen and Bujard, 1992; O'Brien et al., 1997). To achieve this, we constructed a cyclin D1-cdk4 (D1-k4) translational fusion resulting in a D1-k4 fusion protein (as described in Materials and methods). When this work was initiated in 1995, the crystal structure of cdk2 was known (De Bondt et al., 1993) but there was no crystal structure of other cdk's, or cyclins (Andersen et al., 1996, 1997; Brown et al., 1995; Kim et al., 1996), or cyclin.cdk complexes (Jeffrey et al., 1995). In the absence of structural information, our gene fusion strategy was influenced by single-chain antibody (ScFv) experiments, and we chose GGGGSGGGGSGGGGS (Huston et al., 1988) as a hinge connecting the cyclin D1 carboxy-terminus with the amino-terminus of cdk4. To aid immunological detection, a myc-epitope (EQKLISEEDL) (Evan et al., 1985) was added to the amino-terminus. To facilitate protein purification, a hexa-histidine(H6)-tag (Smith et al., 1988) was added at the amino-terminus and a streptavidin binding sequence (SAWRHPQFGG, strep-tag) (Schmidt and Skerra, 1994) was added at the carboxy-terminus constructing plasmids K415 and K485.

To evaluate the role of D1-k4 in the immortalization of primary cells, it is advantageous to control the expression of the fusion gene. We have previously described a one-plasmid system to control gene expression by the tet repressor (O'Brien et al., 1997) that is suitable for the introduction of genes into primary cells. The D1-k4 fusion gene construct, from plasmid K415, was cloned into the tet-repressible single plasmid vector, K255 (O'Brien et al., 1997), for transfection into primary REFs. Before we could use the D1-k4 fusion gene construct in mammalian cell transformation experiments, it was necessary to show that the D1-k4 fusion protein retained enzymatic activity.

Cyclin D1-cdk4 fusion protein produced in the baculovirus expression system is enzymatically active

Experiments were carried out to determine whether the D1-k4 fusion protein possessed normal levels of enzymatic activity for its natural substrate, the retinoblastoma protein (Rb). The D1-k4 fusion protein was produced in the baculovirus expression system using Sf9 insect cells. This protein, which contained six histidines at the N-terminus, was isolated using Ni-NTA affinity chromatography to ~10% purity. The unfused D1.k4 (D1.k4) complex was prepared by co-expression of the cyclin D1 and cdk4 genes in baculovirus-infected insect cells to ~25% purity. These two preparations were compared for their ability to phosphorylate full-length Rb in vitro. Briefly, various amounts of D1.k4 complex or D1-k4 fusion protein were incubated for 30 min at 30°C in an assay buffer that included [-³²P]ATP and Rb. The reaction mixtures were fractionated on SDS - PAGE gels, transferred to nitrocellulose, and examined by autoradiography. The data show that both the D1.k4 complex and the D1-k4 fusion protein phosphorylated Rb, and the level of phosphorylation was dependent on the amount of enzyme extract added to the reaction mixture (Figure 1a). To compare the level of activity between the unfused complex and the fusion protein, the nitrocellulose membrane was subjected to immunoblot analysis using an antibody specific to cyclin D1 (Figure 1b) or cdk4 (Figure 1c). Cyclin D1 and cdk4 proteins used in these assays were quantitated by densitometric analysis of the exposed Hyperfilm-ECL film. When the intensity of the bands obtained in the immunoblot (Figure 1b and c) was used to normalize the intensity of the bands seen in the radioactive phosphorylation assay (Figure 1a), the complex and fusion proteins were found to have similar levels of enzymatic activity. In addition, both the unfused complex and the fusion protein were similarly inhibited, in vitro, by human p15 protein partially purified from bacteria, or by a p16 peptide (aa 90 - 99) (Fahraeus et al., 1998, data not shown).

To show that the D1-k4 fusion protein is responsible for the kinase activity, we introduced mutations (Kato et al., 1994; Matsuoka et al., 1994; van den Heuvel and Harlow, 1993) in cdk4 (D158N or T172E), that are known to inactivate them, and assayed enzymatic activity of the fusion protein. As a control, we introduced Y17F (Terada et al., 1995) and R24C mutations (Wolfel et al., 1995; Zuo et al., 1996) in cdk4, that are not known to inactivate cdk4, and assayed enzymatic activity of the fusion protein. Mutant fusion proteins with D158N or T172E mutations were enzymatically inactive, whereas a Y17F,R24C double mutant was enzymatically active (Figure 2).

Immortalization of rat embryo fibroblasts with the cyclin D1-cdk4 fusion gene

Having established that the D1-k4 fusion protein was functional in cell-free enzymatic assays, experiments were performed to derive stable cell lines by transfection of primary REFs with the fusion gene construct, in the presence or absence of activated Ha-ras. Primary REFs were transfected with eukaryotic tet-repressible expression vectors (Gossen and Bujard, 1992; O'Brien et al., 1997) expressing the following genes: either cyclin D1 alone or D1-k4 fusion construct alone, Ha-rasG12V alone, both the D1-k4 fusion construct and Ha-rasG12V, or both c-myc and Ha-rasG12V, or the empty vector. Ha-rasG12V and c-myc were expressed constitutively, whereas expression of cyclin D1 and the D1-k4 fusion protein were under the control of the tet-repressible system. All cell lines were isolated in the absence of tet, under conditions allowing the expression of the fusion protein. Table 1 summarizes the stable cell lines derived from three separate transfections. No stable cell lines were obtained after single transfection with the plasmids expressing cyclin D1 or Ha-rasG12V, consistent with the findings of others (Hinds et al., 1994; Lovec et al., 1994b). In contrast, stable cell lines were derived after transfection with plasmids expressing either the D1-k4 fusion protein (immortalization), or the D1-k4 fusion protein+Ha-rasG12V (transformation). In addition, spontaneously immortalized cell lines were also obtained following transfection with the empty vector. Also shown in Table 1 is the highest passage number reached after approximately 18 months of continuous in vitro culture among the various stable cell lines derived in each group. As shown in Figure 3a, cell lines derived following co-transfection with the D1-k4 fusion construct+Ha-rasG12V vectors (panels 2 - 6) appear morphologically transformed and are spindle-shaped and refractile. These transfectants are similar in general appearance to those derived by co-transfection with c-myc+Ha-rasG12V vectors (Figure 3b). In contrast, the empty vector transfected control cells are flat and non-refractile, of which one example is shown in Figure 3a (panel 1). The cell lines obtained by transfection with the D1-k4 fusion construct, in the absence of Ha-rasG12V, appear to be immortalized but non-transformed. These cells are also flat and non-refractile (Figure 3c, panels 2 - 4) and are indistinguishable from the empty vector transfected cells (Figure 3c, panel 1).

Cyclin D1-cdk4 +Ha-rasG12V transfected REF cell lines form colonies in soft agarose

Cell lines derived after transfection with D1-k4 and Ha-rasG12V appeared to be morphologically transformed. Many of these cell lines (18/22) formed colonies in soft agarose. Based on the number and size distribution of colonies, the 22 cell lines were grouped into four arbitrary categories from `-' to `+++', ranging from no colony formation to 16% colony forming efficiency (Table 2). Average colony size fell between <0.08 mm² to 0.59 mm². An example of colony formation by each of the four categories is shown in Figure 4a - d.

Cyclin D1-cdk4 fusion protein isolated from stably transfected REF cell lines is enzymatically active

The D1-k4 fusion protein expressed in the baculovirus system was shown to be enzymatically active in phosphorylating full-length Rb in vitro (Figure 1). To demonstrate if fusion protein from a FPr-5 cell line, a transformed cell line obtained after transfecting REFs with D1-k4 fusion construct+Ha-rasG12V, has enzymatic activity, the fusion protein was immunoprecipitated with a -cyclin D1 antibody and an IP kinase assay was performed using [-³²P]ATP and Rb protein as the substrate. The reaction products were separated on a SDS - PAGE gel and detected by autoradiography. The experiment indicates that there was a high level of phosphorylation of Rb (Figure 5a). This radioactive band was not seen in the absence of added Rb substrate. Also, a similar band was not seen when -cyclin D1 immunoprecipitates from cells grown in the presence of doxycycline or when a control IgG2a antibody immunoprecipitates was used. The accompanying panel (Figure 5b) shows that a similar amount of D1-k4 fusion protein was used in both the -Rb and +Rb lanes of the autoradiogram in Figure 5a. We conclude that this cell line expresses an enzymatically active fusion protein whose synthesis is repressed by doxycycline.

Cell proliferation of cyclin D1-cdk4 gene transfected REF cell lines is inhibited by tetracycline

The previous experiment demonstrated that doxycycline repressed the synthesis of the D1-k4 enzyme. To examine whether tet inhibited cell proliferation, FPr-5 (passage 21) cells were incubated with or without tet (1 g/ml) and viable cell numbers were determined by trypan blue dye exclusion. As shown in Figure 6a and c there was a complete inhibition of cell division over the 3 day period. From the same experiment, we also determined the level of fusion protein present by blotting with a cyclin D1 antibody. The data in Figure 6b illustrate that after 6 h of exposure, the level of the D1-k4 fusion protein was dramatically reduced, and this level decreased further with longer incubation with tet. These data demonstrate that addition of tet inhibits expression of the D1-k4 fusion protein, followed by a profound inhibition of cell proliferation. The in vitro growth of each of the 22 cell lines derived by cotransfection with the D1-k4 fusion gene and Ha-rasG12V (Table 1) was similarly inhibited (90%) by tet (data not shown). There was no morphological change associated with the inhibition of cell proliferation in the FPr-5 cell line (Figure 6c). With other cell lines that grew attached to the plastic, cells became flattened and enlarged in size, as their proliferation was inhibited. With some other cell lines that grew unattached to the plastic, cells appeared to increase in size, as their proliferation was inhibited. The basis for these clonal differences has not been determined. Growth of the cell lines obtained by transfection with the D1-k4 fusion construct alone (without Ha-rasG12V) was also inhibited (»90%) by tet.

Inhibition of cell proliferation of stably transfected REF cells by tetracycline is specific for cell lines carrying the tetracycline-repressible gene expression system

To eliminate the possibility that the inhibition of cell division by tet (Figure 6a and c) was due to a non-specific effect, FPr-1 and mr-5 cell lines were analysed for inhibition of cell proliferation by tet. The FPr-1 cell line was derived by stable transfection with vectors expressing the D1-k4 fusion protein+Ha-rasG12V, in which expression of the fusion protein was controlled by the tet-repressible system. The mr-5 cell line was derived by stable transfection with plasmids constitutively expressing c-myc+Ha-ras-G12V, in which neither gene was under the control of the tet-repressible system. With the mr-5 cell line, tet would not be predicted to inhibit growth unless it was acting non-specifically. These cell lines were incubated with or without tet (1 g/ml). At 0, 3, 6, 9 and 12 days following addition of tet, triplicate cultures of each cell type were harvested and viable cell counts (trypan blue dye exclusion) were done (Figure 7). Addition of tet inhibited the proliferation of the FPr-1 cell line over the 12 day course of the study. In the absence of tet, this cell line continued its proliferation. Tet inhibited the expression of the fusion protein, as determined by immunoblot analysis of the lysates prepared on day 9 after tet addition, with an -cyclin D1 antibody. Significantly, there was no detectable reduction in the expression of the endogenous rat cyclin D1. Cell proliferation of all the nine cells derived by cotransfection with c-myc and Ha-rasG12V (Table 1), including mr-5, was not inhibited by tet (Figure 7 and data not shown). These results suggest that the effect of tet on cell proliferation is due to its inhibition of the expression of D1-k4 fusion protein, rather than to a non-specific effect.

Characterization of the in vivo tumorigenicity of REF cell lines obtained after cyclin D1-cdk4 fusion construct+Ha-rasG12V transfection

Colony formation of a cell line in soft agar is usually an indication of its tumorigenicity (Tucker et al., 1977). We tested five of the 22 cell lines, derived by transfection with the D1-k4 construct and Ha-rasG12V, for tumorigenicity in nude mice. Each cell line tested formed tumors, ranging from aggressively tumorigenic, forming large tumors (2300 mg) within 3 weeks, to weakly tumorigenic, forming small tumors (200 - 600 mg) 4 - 6 weeks following implantation (Figure 8). The variation in tumorigenicity and its relationship with colony formation in soft agarose is presented in Table 2. Several additional cell lines including a cell line derived from an aggressive tumor (FPr-5T, passage 6), a low passage number of the parental cell line (FPr-5, passage 29), and an immortalized cell line derived by transfection with the cyclin D1-cdk4 fusion construct vector alone (FP-E5, passage 25) were tested for tumorigenicity. The results indicate that FPr-5T, passage 6 was highly tumorigenic, whereas FPr-5, passage 29, but not passage 35, was somewhat less tumorigenic (Figure 8). A cell line derived by transfection with the D1-k4 construct (FP-E5, passage 25) was non-tumorigenic even at 42 days post-implantation. The normal parental REFs (passage 5) remaining from the non-transfected pool of primary cultures, were found to be non-tumorigenic under the same conditions (Figure 8).

Inhibition of cyclin D1-cdk4 fusion protein expression inhibits the growth of established tumors

Inhibition of D1-k4 fusion protein expression by tet inhibited cell proliferation in vitro (Figure 6). We wanted to extend these results to tumorigenicity in vivo. Mice implanted subcutaneously with the FPr-5 cell line (D1-k4 fusion construct+Ha-rasG12V) were either untreated or treated with tet (either given in the drinking water, 0.1 mg/ml, or as a subcutaneous pellet, 3.3 mg release per day) starting 7 days prior to tumor implantation and measurements of tumor size were made weekly. Large tumors (2200 mg) formed in untreated control mice after 21 days and there was a greater than 99% inhibition of tumor formation in both groups of tet-treated mice after 21 days (data not shown). This result suggests that both modes of tet administration were equally effective and that continued expression of the D1-k4 fusion protein was necessary for tumorigenicity in vivo, analogous to the inhibition of cell proliferation observed in vitro.

Neither tet, nor its analog doxycycline, significantly inhibited tumor formation by mr-5, a cell line derived after transfection with c-myc+Ha-rasG12V in which neither c-myc nor Ha-rasG12V was under tet control (data not shown). This result demonstrates that the in vivo anti-tumor effect of these drugs is due to specific inhibition of gene expression, rather than being the result of non-specific inhibition.

In the preceding experiment, tet was administered to mice beginning 7 days prior to tumor implantation. The present study examined the effect of delaying tet administration to determine if growth of the established tumors could be inhibited. Mice were implanted subcutaneously on day 0 with the FPr-5 cell line (D1-k4 fusion construct+Ha-rasG12V). These mice were treated with tet given in the drinking water beginning either on day -7, day 0, day +7, or day +14, and this treatment continued until day 70. As observed previously, tet given beginning on day -7 led to complete inhibition of tumor formation (Figure 9). In addition, administration of tet beginning on day 0, the same day as tumor implantation, led to a similar degree of inhibition. Delaying tet administration until day 7, when untreated control tumors had reached a size of ~300 mg, resulted in approximately 93% tumor inhibition on day 28 (compared to day 28 in untreated controls, when the large tumor size necessitated euthanasia). Delaying tet until day 14, when untreated control tumors had reached a size of ~1000 mg, resulted in approximately 64% tumor inhibition on day 28, and 90% on day 70 (compared to day 28 control tumors) (Figure 9).

Tumor regrowth was assayed by discontinuing tet treatment on day 70, followed by weekly tumor measurements. Five out of the seven mice initially treated with tet starting on day -7 continued to have undetectable tumors (0 - 10 mg) until euthanized on day 112. In the other two mice, tumors regrew to ~1888 mg and 80 mg, respectively. All mice initially treated with tet beginning on day 0 continued to have undetectable tumors (0 - 10 mg) for the duration of the experiment. The majority of mice treated with tet beginning on day +7 or day +14 had tumors that regrew. These tumors were highly heterogeneous in size, ranging from ~70 - 3000 mg. Our results show that achieving sustained tumor inhibition requires continuous inhibition of cdk4 kinase activity.

Discussion

We constructed a novel fusion gene consisting of full-length human cyclin D1 joined at the carboxy-terminus by a short linker to the amino-terminus of full-length human cdk4. These constructs also included (H)6, c-myc, and streptavidin tags. The results presented in this study demonstrate that the fusion protein is enzymatically active; the specific activity of D1.k4 complex being similar to that of D1-k4 fusion protein (Konstantinidis et al., 1998). Expression of the fusion protein was sufficient to immortalize REFs. These cell lines were morphologically indistinguishable in appearance from the parental REFs, did not form colonies in soft agarose, and were non-tumorigenic in nude mice. In previous work showing oncogenicity of cyclin D1, transfection of cyclin D1 alone did not result in focus formation (Haas et al., 1997; Lovec et al., 1994b). Focus formation required that cyclin D1 be cotransfected with activated Ha-rasG12V or c-myc (Lovec et al., 1994b) or by an E1A mutant (pm928) deficient in Rb binding but which by itself did not result in focus formation (Hinds et al., 1994). Transfection with the D1-k4 fusion gene construct was sufficient to immortalize REFs. We do not know if similar results could have been obtained by cotransfection with cyclin D1 and cdk4 genes on separate transcripts. Our experiments also show that immortalization is dependent on the continued expression of D1-k4 fusion protein. Transformation of primary cells requires the expression of two oncogenes (Land et al., 1983; Newbold and Overell, 1983; Ruley, 1983). We were able to transform REFs by cotransfection with D1-k4 fusion gene construct and activated Ha-rasG12V. These cell lines were morphologically transformed, refractile, grew loosely attached to plastic, formed colonies in soft agarose, and were highly tumorigenic in nude mice.

Cyclin D1 cooperates with activated Ha-rasG12V to transform primary rodent cells (Hinds et al., 1994; Lovec et al., 1994b; Zwicker et al., 1999). Transformation is dependent on the ability of cyclin D1 to interact with cdk4 (Hinds et al., 1994; Zwicker et al., 1999). Interaction of cyclin D1 with cdk4 is facilitated, in vivo, by cdc37 protein (Dai et al., 1996; Stepanova et al., 1996). Besides cyclin D1, cdk4 is also able to transform REFs in cooperation with activated Ha-rasG12V (Haas et al., 1997). This transformation is seen even with a kinase-inactive cdk4 mutant (K35M) but not with a cdk4 mutant (R24C), which cannot interact with p16 (Terada et al., 1995; Wolfel et al., 1995; Zuo et al., 1996), suggesting that p16 titration by cdk4 can result in cdk4-mediated transformation. Both cyclin D1-mediated and cdk4-mediated transformation are thought to result from the inactivation of Rb, through phosphorylation, by cdk4 kinase. However, this interpretation has recently been questioned (Zwicker et al., 1999) and it was suggested that cdk4 may have other substrates, besides Rb, whose phosphorylation is necessary for transformation.

Unlike the work discussed so far, our work used a cyclin D1-cdk4 fusion gene construct to immortalize/transform REFs. A fusion gene construct would synthesize equal levels of cyclin D1 and cdk4 producing an effective high local concentration. This may make the assembly of functional cdk4 kinase less dependent on cdc37 protein (Dai et al., 1996; Stepanova et al., 1996). These differences make it difficult to assess the relative contribution of cyclin D1 and cdk4 genes in our immortalization/transformation experiments. The probability that cdk4 participated in immortalization/transformation is supported by our ability to isolate cell lines after transfection with D1-k4 construct alone but not after cyclin D1 construct alone. In preliminary experiments, cyclin D1-cdk4 Y17F, R24C construct, whose kinase activity cannot be inhibited by the p16 peptide (Fahraeus et al., 1998, unpublished observations) was able to extend the life span of primary human fibroblasts from seven to 15 population doublings (J Sedivy, personal communication). These observations suggest, but do not prove, that immortalization/transformation requires cdk4 kinase activity of the fusion gene construct.

Proliferation of immortalized/transformed cells was inhibited by the addition of tet. Inhibition of fusion protein synthesis by tet could be seen by 6 h and it reached a maximum by 27 h. The reduction in the fusion protein levels preceded the inhibition of proliferation, first observed at 27 h (Figure 6a and b). Since neither the cyclin D1 nor Ha-rasG12V alone led to the establishment of immortalized clones (Table 1), we conclude that cdk4 is critical for the continued proliferation of REFs in our studies. Tet at 1 g/ml did not affect the proliferation of c-myc+Ha-rasG12V transfected REFs whereas it inhibited the proliferation of D1-k4+Ha-rasG12V transformed cells showing that tet-inhibition of cell proliferation is a specific consequence of inhibiting D1-k4 expression. Furthermore, the IC50 for the growth inhibitory effect of tet on D1-k4+Ha-rasG12V transfectants was approximately 25 ng/ml (data not shown), which is below the concentration reported to inhibit growth of mammalian cells in vitro (Fife and Sledge, 1995; Fife et al., 1998; Sauter, 1998; van den Bogert et al., 1986). Inhibition of proliferation of immortalized cells by tet is explained by the decrease in the amount of fusion protein. However, in the case of inhibition of transformed cells, tet decreases the expression of the fusion protein, leaving constitutive expression of activated Ha-rasG12V. Although ras has been reported to activate cyclin D1 expression in established cell lines (Albanese et al., 1995; Filmus et al., 1994; Liu et al., 1995; Weber et al., 1997), it leads to growth inhibition in primary mouse embryo fibroblasts, by the accumulation of p53 and p16 (Serrano et al., 1997). Immortalization of primary mouse embryo fibroblasts requires inactivation of both p53 and Rb pathways. Most of the spontaneously immortalized mouse embryo fibroblasts have alterations in p53 locus (Harvey and Levine, 1991) or in Ink4a-ARF locus (Kamijo et al., 1997; Serrano et al., 1996; Zindy et al., 1997). D1-k4+Ha-rasG12V transformed REFs have a functional Rb, as shown by p16-mediated growth arrest (unpublished observations). However, we have no data on the p53 status of these cell lines. In these cell lines, growth-inhibitory signals from activated Ras led to an increase in the level of p16 (unpublished observations), but these signals are probably overcome by the overexpression of D1-k4 fusion protein (Latham et al., 1996; Lukas et al., 1995; Medema et al., 1995). Repression of the D1-k4 fusion gene transcription by tet would restore Rb function, rendering the cells sensitive to growth suppression by the accumulated p16. Growth-suppressed cells had elevated levels of p16 (unpublished observations) and adopted large, flat morphology, reminiscent of p16-associated senescence (Serrano et al., 1997).

The ability of a cell line to form soft agar colonies has been correlated with its ability to grow as a tumor xenograft (Freedman and Shin, 1974; Tucker et al., 1977). However, in our cell lines this correlation was not absolute, suggesting that other factors also influence tumor formation (Tucker et al., 1977).

Continued expression of the D1-k4 fusion protein was also shown to be required for the growth of tumors in nude mice. Administration of tet to mice, beginning 7 days prior to tumor implantation, greatly inhibited the ability of cell lines to grow as tumors, when cells were implanted subcutaneously. Furthermore, considerable inhibition was observed when tet administration was delayed until day 0, 7, or 14 relative to tumor implantation. It is important to note that when tet treatment was begun on day 14, tumors had already reached a size of ~1 g, yet these tumors regressed by 90% (measured at day 70, compared to untreated controls measured when euthanized at day 28). We do not know if the tumor regression results from an inhibition of cdk4 kinase activity or from an immunological response. Where tet administration was delayed until day 7 or day 14 post-implantation (a regimen that resulted in lack of complete tumor inhibition), tumor regrowth was greater and occurred in a higher percentage of mice than when tet was begun at earlier times (day -7 or day 0). The inhibitory effect of tet on the tumorigenicity of REFs was not due a non-specific anti-tumor effect of the drug (Kroon et al., 1984) since there was no inhibition of REFs stably transfected with c-myc+Ha-rasG12V, in which neither gene was under tet control (data not shown).

There is considerable interest in the cyclin-dependent kinases as potential therapeutic targets in human cancer because of the frequent alterations in the cell cycle pathways regulated by these kinases (Gray et al., 1998). It is likely that specific inhibitors of selected members of the cyclin-dependent kinase family may be useful oncolytic agents. The cell lines described in this study, because of their dependence for proliferation on human D1-k4 genes, may serve as useful experimental tools for the in vitro and in vivo characterization of cdk4 inhibitors.

Materials and methods

Construction of cyclin D1-cdk4 fusion gene

Modifications of pNEB193 vector: A number of modifications were introduced into pNEB193(K329) (New England Biolabs, Beverly, MA, USA) to help clone different fragments of cyclin D1 and/or cdk4 for further manipulation. Plasmid K329 (pNEB193) was cut with HindIII - EcoRI and oligonucleotides 11562(5'-AGCTTACGGCGCG CCGCCGCA CCATG GCGCTA GCATCGATG AAGGAGGACG GCG GCGTC GCGA CGG AGG GG CCC-3') and 11563(5'-AATTGGGCCCCTCCGTCGCGACGCCGCCGTCCTC CTTCATCGATG CTAGCCCAT GGTGGG-GCGGCGCGCCGTA-3') were cloned, introducing AscI, NheI, ClaI, NruI and ApaI sites (K335). Plasmid K335 was cut with ClaI - ApaI and oligonucleotides 11566 (5'-CGA TGGCTAC CTCTCGATATGAGCCAGTGGCTGAA ATTGGT GTCG GTG CCTATGGGACAGTGTACAGATATCAATCCGGAGTGAGGGCC-3') and 11567(5' -CTCA CTCCGGAT TGATA TCTGTACACTGT CCCATAG GCACCGACACC AATTTCAGCCACTGGCTCATATCGAGAGGTAGCCAT-3')were cloned, introducing a ClaI site at the N-terminus of cdk4 and reconstructing the N-terminus of cdk4 (1 - 21 amino acids), followed by BsrGI, EcoRV and BspEI sites. (K338). Plasmid K335 was cut with ClaI-ApaI, and oligonucleotides12072 (5'-CGATGATATCGGATCCACCCGAGGCCGGCCCTCGAGAA GTC GA CTCA TAG TAC CTGCAG GG CC -3')and 12073 (5'-CTGCAGGTACTAGTGAGTCGACTTCTCGAG GG CC GG CC TC GG GTGG AT CCGAT ATC AT-3') were cloned introducing EcoRV, BamHI, NgoAIV, FseI, SalI, SpeI and Sse8387I sites (K381).

Cloning of the cyclin D1 gene: Human cyclin D1 was cloned by PCR from a cDNA library of humanplacenta using 5'-CCCCAGCCATGGAACACCAGCTCC-3' and5'-CCCCTCAGATGTCCACGTCCCGCA-3' as primers. The amplified fragment was cloned into pCR - SCRIPT (Stratagene cat # 211190) at the unique SrfI site, and a full-length, cyclin D1 clone was reconstructed by assembling AscI - BspMI and BspMI - NotI fragments of cyclin D1 into AscI - NotI cut plasmid K255 (O'Brien et al., 1997 (K344).

Introduction of a 5'-NheI site into the cyclin D1 gene: Plasmid K381 was cut with PstI, the ends were dephosphorylated with calf intestinal alkaline phosphatase, cut again with NheI and was ligated with a AlwNI - PstI fragment which included the 5'-end of cyclin DI with oligonucleotides 12105 (5'-CTAGCAATGGAACACCAGCTC-3') and 12106 (5'-CTGGTGTTCCATTG-3') to introduce a NheI site upstream of the cyclin D1 gene (K391).

Introduction of a 3'-ClaI site into the cyclin D1 gene: Plasmid K381 was cut with PstI, the ends were dephosphorylated with calf intestinal alkaline phosphatase, then cut again with NdeI before being ligated with a PstI-EarI fragment which included the 3' end of cyclin D1 with oligonucleotides 12107 (5'-CGAGGAGGAGGAAGAGGAGGAAG AGGAGGTGGA CCTGGCTTGCACACCCACCGACGTGC GGGACGTGGACATCG CATCG ATGCCTGCAG-3')and 12108 (5'-TACTGCAGGCATCGATGCGATGTCCACGTCCCGCACGTCGGTGGGTGTGC-AAGCCAGGTCCACCT CCTCTTCCTCCTCT TCCTCCTCC-3') to introduce a ClaI site immediately after the last amino acid (C-pool). This cloning step inadvertently introduced two extra glutamic acid residues (E), resulting in a total of 11 contiguous glutamic acid residues.

Cloning of the cdk4 gene: The full-length cdk4 gene was cloned by RT - PCR from RNA isolated from SKBR3 cell line using a reverse transcriptase kit (Perkin-Elmer). The primers used were 5'-CCCGGATCCAATATGGCTACCTCTCGATATGAGCC-3' and5'-CATAAGGATGAAGGTAATCCGGAGTGAGCGGCCGCGG-3' which contain a BamHI site and a NotI restriction site for cloning purposes. PCR amplified fragments were cloned into pUC119 and verified to be cdk4 by sequence analysis (K328). A BsrGI - BspEI cdk4 fragment from plasmid K328 was cloned into plasmid K338 to generate the cdk4 gene fused with LacZ (K380).

Introduction of a myc-tag into the cdk4 gene: Plasmid K380 was cut with NcoI - ClaI and oligonucleotides 12099 (5'-CAT GGCGGAG GAGCAGAAGCTGATA TCCGAGGAG GACCT-GCTGCTAGCTGG TCGCGATTCGAAGGG TGGTGGAG GTTCTGGAGGTGGAGG ATCCG-GTGGTGGAGGTT-3') and 12100 (5'-CGAACCTCCACCACCGGATCCTCCACCTCCAG-AACC TCCACCACCCTCGAATCGC GACCAGCTAGCAGCAGGTCCTCCTCGGATATCAGCTTCTGCTCCTCCGC-3') were cloned introducing a myc-tag (EQKLISEEDL) at the N-terminus of cdk4 followed by EcoRV, NheI, NruI, and NspV sites followed by an inter-domain linker (GGGGSGGGGSGGGGS) which included a BamHI site (K390). The NheI and NspV sites were subsequently used to clone the cyclin D1 gene.

Introduction of a strep-tag into the cdk4 gene: Plasmid K390 was cut with BspEI and BstXI and oligonucleotides 12103 (5'-CCGGAGGGCGGCAGCGCTTGGCGCCACCCACAGTT CGGTGGTTGAATA AATAGATGA ATGACCTGCAGG-3' and 12104 (5'-TGAACCTGCAGGT-CATTCATCTATTTATTCAACCACCGAACTGTGGGTGGCGCCAAGCGCTGCCGCCCT-3') were cloned introducing a strep-tag (SAWRHPQFGG) and Eco47III, NarI and Sse8387I sites at the C-terminus of the cdk4 gene (K396).

Construction of the D1-k4 fusion gene: Plasmid K396 was cut with NheI - NspV and was ligated to the 5' end of cyclin D1 on a NheI - PstI fragment along with 3' end of cyclin D1 PstI - ClaI fragment (K415).

Construction of the D1-k4 fusion gene in baculoviral vectors: Plasmid pVL1393-BP (see Expression of proteins in the Baculovirus system) was cut with AscI - Sse8387I and the D1-k4 fusion gene from plasmid K415 on an AscI - Sse8387I fragment was cloned (K433).

Construction of a tet-repressible D1-k4 fusion gene: Plasmid K255 (O'Brien et al., 1997) was cut with AscI - Sse8387I and the D1-k4 fusion gene from plasmid K415 on an AscI - Sse8387I fragment was cloned (K429).

Construction of a His-tagged D1-k4 fusion gene: The NcoI - NheI region coding for the myc tag in plasmid K415 was replaced with oligonucleotides 12393 (5'-CATGGCGCATCATCATCATCATCATGGAGGTGGAGGTT CGG AGCAGAAGC TTAT TTCCGAGGAG GATCTGCTGGTGCCACGCGGTTCCCTG-3') and12394 (5'-CTAG CAGGG AACCGCGTGG CACCA GCAGATCCTCCTCGGAAATA AGCTTCTG CTCCGAACCTC CACCTCCATGAGATGATGATGATGCGC-3')to introduce a His-tag, a myc-tag and a thrombin cleavage site amino-terminal to cyclin D1 (plasmid K485). Correctness of all the plasmids was verified by DNA sequence analysis of the appropriate regions.

Other DNA constructs

Plasmids pSVC-myc1 and pUC EJ76.6, that express human c-myc or human EJ-activated Ha-rasG12V in mammalian cells, were obtained from Dr R Swift. Mutations in cyclin D1 and cdk4 were constructed by oligonucleotide cassette mutagenesis and were verified by DNA sequencing.

Expression of proteins in the Baculovirus system

The multiple cloning sites of the insect vectors pVL1393 and pFastBac1 (Gibco - BRL) were modified by inserting the following sequence between the BamHI and PstI sites, 5'-GATCC GGCGCGCC GCCGCCACCATGGCGCTA GCGTCGCC CGGGCCT GCAGGGCCC CCTAGGAGCGCTTGG CGCCACCCACAGTTCGGTGGTCACTAGTTGATTGAGCGGCC-GCTGCA-3' and 5'-GCGGCCGCTCAATCAAC TAGTGACCACCGAACTGTGGGTGGCGC-CAAGCGCTCCTAGGG GGCCCTGC AGGCCCGGGCGACGCTAGGCCAGGTGGCGGCGGCGCGCCG-3'. This modification introduces unique AscI and Sse8387I sites into these vectors to readily insert the fusion fragments. The modified pVL1393 is designated pVL1393-BP and the modified pFastBacI is designated plasmid K491. The pVL1393-BP generated vectors were co-transfected into Sf9 insect cells using the linear transfection module (Invitrogen) for production of recombinant virus. The Bac-to-Bac Baculovirus expression system (Gibco - BRL) was used for the production of recombinant virus when fusion fragments were cloned into plasmid K491. Sf9 insect cells were grown in Sf900 II SFM media (Gibco - BRL). Baculovirus stocks encoding the D1-k4 fusion gene were produced and used to infect Sf9 cells using standard techniques. Typically, a multiplicity of infection of three or more was used and the Sf9 cells were harvested at 48 h post infection.

Purification of cyclin D1-cdk4 fusion protein

D1-k4 fusion protein was partially purified from either insect or mammalian cells. Baculovirus infected Sf9 cells were collected by centrifugation, resuspended at 1´10⁷ cells/ml were lysed in IP lysis buffer (50 mM HEPES pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM DTT, 2.5 mM EGTA, 0.1% Tween 20, 10% Glycerol, 0.1 mM PMSF, 500 M ATP, 10 mM -glycerophosphate, 1 mM NaF, and 0.1 mM orthovanadate) on ice for 30 min. Cells were sonicated three times on ice for 20 s each, and clarified by centrifugation at 15 000 g for 15 min at 4°C. Immunoprecipitations were carried out by the addition of 20 l (100 g/ml) of -c-myc (Ab-1) antibody (Oncogene Research Products, Cambridge, MA, USA) to 500 l of clarified cell lysate. The mixture was incubated with agitation for 3 h at 4°C, followed by addition of 50 l of 50% Protein-G Agarose per sample (Boehringer Mannheim, Indianapolis, IN, USA), which had been equlibrated with IP lysis buffer, followed by incubation with agitation for 2 - 5 h at 4°C (Matsushime et al., 1994). The Protein-G-Agarose-bound immunoprecipitates were pelleted and washed 4´ with IP Lysis Buffer and then 2´ with 50 mM HEPES pH 7.4 and 1 mM DTT. The washed precipitates were then resuspended in buffer compatible with the enzymatic assay (see Assay of cdk4 kinase activity). For streptavidin purification, 500 l of the insect lysate was incubated for 45 min at room temperature with 200 l of Streptavidin Paramagnetic Beads (Promega, Madison, WI, USA). The streptavidin paramagnetic beads were pelleted at room temperature using a MagneSphere Technology magnetic separation stand (Promega). The streptavidin beads were washed three times with 1 ml of 1´PBS/0.1% Tween 20 at room temperature. The fusion protein was eluted from the streptavidin beads in 120 l of elution buffer (25 mM HEPES pH 7.5, 0.1 mM PMSF, 0.1 mM DTT, 10 mM -glycerophosphate, 1 mM NaF, 10 mM sodium orthovanadate, 5 mM d-biotin, and 10% glycerol) for 30 min at room temperature. Eluted protein was stored at -70°C.

For Ni-NTA agarose purification, 1 ml of the insect extract was combined with 1.0 ml of Ni-NTA agarose (Qiagen, Chatsworth, CA, USA) which had been equilibrated with wash buffer (50 mM HEPES pH 7.5, 300 mM NaCl, and 20 mM imidazole, 0.1 mM PMSF). The insect lysate was incubated with Ni-NTA agarose for 4 h at 4°C with agitation. Ni-NTA agarose was pelleted by centrifuging at 2000 g for 2 min. After washing the Ni-NTA agarose three times with 5.0 ml of 50 mM HEPES pH 7.5, 300 mM NaCl, and 1 mM DTT at 4°C, the fusion protein was eluted from the agarose in 750 l of elution buffer B (50 mM HEPES pH 7.5, 300 mM NaCl, 250 mM imidazole, 0.1 mM PMSF, 10 mM sodium orthovanadate, 1 mM NaF, and 20 mM -glycerophosphate) for 1 h at 4°C with agitation. The eluted fusion protein was dialyzed overnight at 4°C in 3.0 l of dialysis buffer (25 mM HEPES pH 7.5, 10% glycerol, 0.01% Triton X-100, 0.1 mM PMSF, 0.1 mM DTT, 20 mM -glycerophosphate, 1 mM NaF, and 10 mM sodium orthovanadate), and then stored at -70°C. Cyclin D1-cdk4 fusion protein from mammalian cells was isolated and assayed for enzymatic activity according to published procedures (LaBaer et al., 1997; Matsushime et al., 1994).

Purification of co-expressed cyclin D1-cdk4 complex

A 1 : 10 suspension of insect cell pellet was homogenized in 50 mM HEPES pH 7.5, 320 mM sucrose, 1 mM DTT, 0.1 mM PMSF, 1 mM EGTA, 1 mM EDTA and 20 g/ml leupeptin (equilibration buffer). The lysed cells were centrifuged for 1.5 h at 100 000 g to remove the cytosol. A Poros Q column was equilibrated in 25 mM Tris pH 8.0, 10% glycerol, 1 mM DTT, 0.1 mM PMSF, 1 mM EDTA, and 20 g/ml leupeptin, and loaded with lysate at 5 ml/l of infected insect cells. The column was then washed with 10-column volumes of equilibration buffer, eluted with a 0 - 1 M NaCl gradient, and 2 ml fractions were collected. Column fractions were assayed for activity, and peak fractions were pooled. The resulting pool was diluted to give a final NaCl concentration of 100 mM, and was loaded onto a hydroxyapatite column equilibrated with 25 mM Tris pH 8.0, 0.1 mM PMSF, 1 mM EDTA, and 20 g/ml leupeptin. This column was washed with 10-column volumes of equilibration buffer and the D1.k4 complex was eluted with a gradient of 0 - 400 mM potassium phosphate, pH 7.5. Column fractions were assayed for kinase activity, and the peak fractions were pooled and stored at -70°C.

Assay of cdk4 kinase activity

Partially purified co-expressed D1.k4 or fused D1-k4 proteins were assayed for kinase activity as follows. Kinase reactions with various amounts of partially purified D1.k4 or D1-k4 from insect cells contained: 35 mM HEPES pH 7.5, 10 mM MgCl₂, 10.0 Ci [-³²P]ATP (Amersham Life Science, Arlington Heights, IL, USA) (6000 Ci/mmol), 0.2 g Rb (full-length protein) (QED Bioscience, Inc., San Diego, CA, USA), 1 mM DTT, 2.5 mM EGTA, 0.1 mM sodium orthovanadate, 10 mM -glycerophosphate and 1 mM NaF in a total of 100 l. Reactions were incubated at 30°C for 30 min, boiled for 5 min, and half of the reaction was loaded onto a 12.5% SDS - polyacrylamide gel. The gel was transferred to Hybond-ECL nitrocellulose (Amersham Life Science) and exposed to Hyperfilm-ECL (Amersham Life Science) for detection of phosphorylated retinoblastoma protein.

Immunoblots

-cyclin D1 antibody (PRAD1 or HD-11) or -cdk4 antibody (H-22) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) was used to detect cyclin D1 or cdk4 proteins according to published procedures (Harlow and Lane, 1999).

Cell culture

Cells were cultured at 37°C with 10% CO₂ in Dulbecco's Minimal Essential Medium (DMEM) (Gibco - BRL, Gaithersburg, MD, USA) supplemented with 10% defined fetal bovine serum (HyClone, Logan, UT, USA), 1 mM sodium pyruvate (Gibco - BRL), 0.1 mM non-essential amino acids (Gibco - BRL), 10 mM HEPES buffer (Gibco - BRL), and 50 g/ml gentamycin sulfate (Gibco - BRL).

Derivation of transfected cell lines

Whole embryo cell suspensions were prepared from CD rats (Charles River Laboratories, Wilmington, MA, USA) at day 15 of gestation by dissociation of minced tissue fragments in 0.25% trypsin at 37°C. Single cell suspensions were prepared by low speed centrifugation (260 g) and removal of debris by filtration through cell strainers with a porosity of 70 microns (Becton Dickinson, Franklin Lakes, NJ, USA). Cells were resuspended in supplemented medium and seeded at 1´10⁷ cells per 10 cm plastic dish and cultured at 37°C in 10% CO₂. Cells were transfected in 6-well plates at passage 2 or in 10 cm dishes at passage 3 as follows: (1) Six-well plates were seeded with 2.1´10⁵ cells per well and transfected 1 day later at 20 - 30% confluence by addition of 2.5 g of DNA per well using a calcium phosphate precipitation method (Wigler et al., 1979). After 6 h the cells were exposed to 12.5% glycerol in phosphate buffered saline for 30 s, washed and then cultured with complete medium. After 2 days, cells were subjected to drug selection in Hygromycin-B at 50 g/ml, and fresh drug was added twice per week. One set of cultures was not subjected to drug selection until there were visible colonies, which yielded similar results to drug selected set. (2) 10 cm dishes were seeded with 5´10⁵ cells per dish and transfected 1 day later at 20% confluence by addition of 20 g of DNA per dish after precipitation with calcium phosphate. Six hours later, cultures were shocked with glycerol as described above. Two days following transfection, cultures were trypsinized and split 1 : 2 into 10 cm dishes with 50 g/ml of Hygromycin-B which was replinished twice weekly. Cell lines were established from several independent colonies in supplemented DMEM with 50 g/ml Hygromycin-B. To measure expression levels of the D1-k4 fusion protein, cells were lysed in a high salt buffer (100 l per 10⁶ cells) on ice for 10 min (lysis buffer: 50 mM Tris-Cl-pH 8.0, 0.5% NP-40, 420 mM NaCl, 1 mM PMSF, 25 g/ml leupeptin, 25 g/ml aprotinin, 1 mM benzamidine, 10 g/ml trypsin inhibitor). Insoluble debris was removed by centrifugation in a Beckman Microfuge at 12 000 g for 15 min at 4°C, and then stored at -70°C. Protein concentrations of the lysates were determined using the Biorad (Hercules, CA, USA) DC Protein Assay kit, following the manufacturer's procedure. Immunoblots were done according to published procedures (Harlow and Lane, 1999).

Assay for growth in soft agarose

Bottom layers of 4 ml of 1.2% agarose (SeaPlaque, FMC BioProducts, Rockland, ME, USA) in water were prepared in 5 cm dishes and then overlaid with 2 ml of cell suspension containing 900 cells in 0.6% agarose prepared in supplemented DMEM with 50 g/ml of Hygromycin-B. The dishes were fed once per week with 2 ml of medium. The first feeding was with 0.6% agarose prepared in supplemented DMEM, whereas subsequent feedings were with supplemented DMEM without agarose. Colony number and size were quantitated on day 16 by capturing images with a video camera (Cohu, Inc., San Diego, CA, USA) and image analysis was done using NIH Image (V1.57) and custom macros.

In vitro growth curves in 6-well plates

Cells were cultured in 6-well plates at a starting density of 1´10⁴ or 3´10⁵ per well in triplicate in a volume of 2 ml in supplemented DMEM. At the indicated time points, individual wells were harvested by gentle pipetting, spun down at 260 g for 5 min, and resuspended in a small volume of medium. Dilutions were made in 0.4% trypan blue in saline (Gibco - BRL) and live and dead cells were enumerated using a hemocytometer counting chamber. Cell counts are expressed as the average number of cells per 6-well culture±the standard deviation.

In vivo tumor studies

CD1 nu/nu mice (Charles River Laboratories, Wilmington, MA, USA) were implanted subcutaneously in the axillary region with 1.0 - 2.5´10⁶ cells (see figure legends). Each week following tumor implantation (each group contained seven mice), two-dimensional measurements (width and length in mm) of all tumors were taken using digital electronic calipers interfaced to a microcomputer and converted into tumor weight (Worzalla et al., 1990).

In some studies, the effect of controlled expression of the D1-k4 fusion protein was determined by in vivo treatment of mice with tet or doxycycline (Sigma Chemical Co., St. Louis, MO, USA). Mice were treated with tet by administration as a 0.1 mg/ml solution in the drinking water (dissolved in 5% sucrose), or as a 200 mg pellet (Innovative Research of America, Sarasota, FL, USA) implanted subcutaneously in the nape of the neck, which gave a sustained release of approximately 3.3 mg/day. Doxycycline (0.1 mg/ml in 5% sucrose) was given in the drinking water as described above. Drinking water (in brown bottles) containing the drug was changed twice weekly and administered to mice ad libitum.

Acknowledgements

We thank Bruce Glover, for oligonucleotides; Pam Rockey, for DNA sequencing; Julia Enkema, for assistance with the surgical implantation of the tetracycline pellets; Mike Esterman, for help with capture and analysis of video images of soft agarose colony formation assays and for preparation of custom NIH Image macros; Jeff Dixon (Sphinx Pharmaceuticals), for purified cyclin D1.cdk4 complex; and Robert Swift for Ha-rasG12V and c-myc plasmids; and Beverly Teicher, for reviewing the manuscript. Special thanks are due to Mei Lai for a thorough and a critical review of the manuscript. We thank John Sedivy for permission to cite his unpublished results.

Albanese C, Johnson J, Watanabe G, Eklund N, Vu D, Arnold A and Pestell RG. (1995). J. Biol. Chem. 270, 23589-23597. Article MEDLINE

Andersen G, Busso D, Poterszman A, Hwang JR, Wurtz JM, Ripp R, Thierry JC, Egly JM and Moras D. (1997). EMBO J. 16, 958-967. Article MEDLINE

Andersen G, Poterszman A, Egly JM, Moras D and Thierry JC. (1996). FEBS Lett. 397, 65-69. MEDLINE

Arber N, Doki Y, Han EK, Sgambato Z, Zhou P, Kim NH, Delohery T, Klein MG, Holt PR and Weinstein IB. (1997). Cancer Res. 57, 1569-1574. MEDLINE

Arguello F, Alexander M, Sterry JA, Tudor G, Smith EM, Kalavar NT, Greene Jr JF, Koss W, Morgan CD, Stinson SF, Siford TJ, Alvord WG, Klabansky RL and Sausville EA. (1998). Blood 91, 2482-2490. MEDLINE

Biggs JR and Kraft AS. (1995). J. Mol. Med. 73, 509-514. MEDLINE

Brown NR, Noble ME, Endicott JA, Garman EF, Wakatsuki S, Mitchell E, Rasmussen B, Hunt T and Johnson LN. (1995). Structure 3, 1235-1247. MEDLINE

Carlson BA, Dubay MM, Sausville EA, Brizuela L and Worland PJ. (1996). Cancer Res. 56, 2973-2978. MEDLINE

Dai K, Kobayashi R and Beach D. (1996). J. Biol. Chem. 271, 22030-22034. MEDLINE

De Bondt HL, Rosenblatt J, Jancarik J, Jones HD, Morgan DO and Kim SH. (1993). Nature 363, 595-602. MEDLINE

DeGregori J, Kowalik T and Nevins JR. (1995). Mol. Cell. Biol. 15, 4215-4224. MEDLINE

Donnellan R and Chetty R. (1998). Mol. Pathol. 51, 1-7. MEDLINE

Dowdy SF, Hinds PW, Louie K, Reed SI, Arnold A and Weinberg RA. (1993). Cell 73, 499-511. MEDLINE

Dyson N. (1998). Genes Dev. 12, 2245-2262. MEDLINE

Elledge SJ. (1996). Science 274, 1664-1672. Article MEDLINE

Evan GI, Lewis GK, Ramsay G and Bishop JM. (1985). Mol. Cell. Biol. 5, 3610-3616. MEDLINE

Ewen ME, Sluss HK, Sherr CJ, Matsushime H, Kato J and Livingston DM. (1993). Cell 73, 487-497. MEDLINE

Fahraeus R, Lain S, Ball KL and Lane DP. (1998). Oncogene 16, 587-596. MEDLINE

Fife RS and Sledge Jr GW. (1995). J. Lab. Clin. Med. 125, 407-411. MEDLINE

Fife RS, Sledge Jr GW, Roth BJ and Proctor C. (1998). Cancer Lett. 127, 37-41. MEDLINE

Filmus J, Robles AI, Shi W, Wong MJ, Colombo LL and Conti CJ. (1994). Oncogene 9, 3627-3633. MEDLINE

Freedman VH and Shin SI. (1974). Cell 3, 355-359. MEDLINE

Gossen M and Bujard H. (1992). Proc. Natl. Acad. Sci. USA 89, 5547-5551. MEDLINE

Gray NS, Wodicka L, Thunnissen A-MWH, Norman TC, Kwon S, Espinoza FH, Morgan DO, Barnes G, LeClerc S, Meijer L, Kim S-H, Lockhart DJ and Schultz PG. (1998). Science 281, 533-538. Article MEDLINE

Haas K, Staller P, Geisen C, Bartek J, Eilers M and Moroy T. (1997). Oncogene 15, 179-192. MEDLINE

Hall M and Peters G. (1996). Adv. Cancer Res. 68, 67-108. MEDLINE

Harlow E and Lane D. (1999). Using Antibodies. Cold Spring Harbor Press, Cold Spring Harbor, NY, pp. 495.

Harvey DM and Levine AJ. (1991). Genes Dev. 5, 2375-2385. MEDLINE

Hinds PW, Dowdy SF, Eaton EN, Arnold A and Weinberg RA. (1994). Proc. Natl. Acad. Sci. USA 91, 709-713. MEDLINE

Hirama T and Koeffler HP. (1995). Blood 86, 841-854. MEDLINE

Huston JS, Levinson D, Mudgett-Hunter M, Tai MS, Novotny J, Margolies MN, Ridge RJ, Bruccoleri RE, Haber E, Crea R and Oppermam H. (1988). Proc. Natl. Acad. Sci. USA 85, 5879-5883. MEDLINE

Jeffrey PD, Russo AA, Polyak K, Gibbs E, Hurwitz J, Massague J and Pavletich NP. (1995). Nature 376, 313-320. MEDLINE

Jiang W, Kahn SM, Zhou P, Zhang YJ, Cacace AM, Infante AS, Doi S, Santella RM and Weinstein IB. (1993). Oncogene 8, 3447-3457. MEDLINE

Kamijo T, Zindy F, Roussel MF, Quelle DE, Downing JR, Ashmun RA, Grosveld G and Sherr CJ. (1997). Cell 91, 649-659. MEDLINE

Kato J, Matsushime H, Hiebert SW, Ewen ME and Sherr CJ. (1993). Genes Dev. 7, 331-342. MEDLINE

Kato JY, Matsuoka M, Strom DK and Sherr CJ. (1994). Mol. Cell. Biol. 14, 2713-2721. MEDLINE

Kerkhoff E and Ziff EB. (1995). EMBO J. 14, 1892-1903. MEDLINE

Kim HG and Miller DM. (1998). Biochemistry 37, 2666-2672. MEDLINE

Kim KK, Chamberlin HM, Morgan DO and Kim SH. (1996). Nature Struct. Biol. 3, 849-855.

Konstantinidis AK, Radhakrishnan R, Gu F, Nagaraja Rao R and Yeh WK. (1998). J. Biol. Chem. 273, 26506-26515. MEDLINE

Kroon AM, Dontje BH, Holtrop M and Van den Bogert C. (1984). Cancer Lett. 25, 33-40. MEDLINE

La Thangue NB. (1996). Biochem. Soc. Trans. 24, 54-59. MEDLINE

LaBaer J, Garrett MD, Stevenson LF, Slingerland JM, Sandhu C, Chou HS, Fattaey A and Harlow E. (1997). Genes Dev. 11, 847-862. MEDLINE

Ladanyi M, Lewis R, Jhanwar SC, Gerald W, Huvos AG and Healey JH. (1995). J. Pathol. 175, 211-217. MEDLINE

Land H, Parada LF and Weinberg RA. (1983). Nature 304, 596-602. MEDLINE

Latham KM, Eastman SW, Wong A and Hinds PW. (1996). Mol. Cell. Biol. 16, 4445-4455. MEDLINE

Liggett Jr WH and Sidransky D. (1998). J. Clin. Oncol. 16, 1197-1206. MEDLINE

Liu JJ, Chao JR, Jiang MC, Ng SY, Yen JJ and Yang-Yen HF. (1995). Mol. Cell. Biol. 15, 3654-3663. MEDLINE

Lovec H, Grzeschiczek A, Kowalski MB and Moroy T. (1994a). EMBO J. 13, 3487-3495. MEDLINE

Lovec H, Sewing A, Lucibello FC, Muller R and Moroy T. (1994b). Oncogene 9, 323-326. MEDLINE

Lukas J, Parry D, Aagaard L, Mann DJ, Bartkova J, Strauss M, Peters G and Bartek J. (1995). Nature 375, 503-506. MEDLINE

Matsuoka M, Kato JY, Fisher RP, Morgan DO and Sherr CJ. (1994). Mol. Cell. Biol. 14, 7265-7275. MEDLINE

Matsushime H, Quelle DE, Shurtleff SA, Shibuya M, Sherr CJ and Kato JY. (1994). Mol. Cell. Biol. 14, 2066-2076. MEDLINE

Medema RH, Herrera RE, Lam F and Weinberg RA. (1995). Proc. Natl. Acad. Sci. USA 92, 6289-6293. MEDLINE

Morgan DO. (1995). Nature 374, 131-134. MEDLINE

Nakagawa H, Wang TC, Zukerberg L, Odze R, Togawa K, May GH, Wilson J and Rustgi AK. (1997). Oncogene 14, 1185-1190. MEDLINE

Neuman E, Flemington EK, Sellers WR and Kaelin Jr WG. (1994). Mol. Cell. Biol. 14, 6607-6615. MEDLINE

Nevins JR. (1992). Science 258, 424-429. MEDLINE

Nevins JR. (1998). Cell Growth Diff. 9, 585-593. MEDLINE

Newbold RF and Overell RW. (1983). Nature 304, 648-651. MEDLINE

O'Brien K, Otto K and Nagaraja Rao R. (1997). Gene 184, 115-120. MEDLINE

Peeper DS, Upton TM, Ladha MH, Neuman E, Zalvide J, Bernards R, DeCaprio JA and Ewen ME. (1997). Nature 386, 177-181. MEDLINE

Petronio J, He J, Fults D, Pedone C, James CD and Allen JR. (1996). J. Neurosur. 84, 1020-1023.

Quelle DE, Ashmun RA, Shurtleff SA, Kato JY, Bar-Sagi D, Roussel MF and Sherr CJ. (1993). Genes Dev. 7, 1559-1571. MEDLINE

Resnitzky D, Gossen M, Bujard H and Reed SI. (1994). Mol. Cell. Biol. 14, 1669-1679. MEDLINE

Robles AI, Larcher F, Whalin RB, Murillas R, Richie E, Gimenez-Conti IB, Jorcano JL and Conti CJ. (1996). Proc. Natl. Acad. Sci. USA 93, 7634-7638. MEDLINE

Robles AI, Rodriguez-Puebla ML, Glick AB, Trempus C, Hansen L, Sicinski P, Tennant RW, Weinberg RA, Yuspa SH and Conti CJ. (1998). Genes Dev. 12, 2469-2474. MEDLINE

Rollbrocker B, Waha A, Louis DN, Wiestler OD and von Deimling A. (1996). Acta Neuropathol. 92, 70-74. MEDLINE

Ruley HE. (1983). Nature 304, 602-606. MEDLINE

Sandig V, Brand K, Herwig S, Lukas J, Bartek J and Strauss M. (1997). Nature Med. 3, 313-319. MEDLINE

Sauter C. (1988). Br. J. Cancer 57, 514-515. MEDLINE

Schmidt TG and Skerra A. (1994). J. Chromato. 676, 337-345.

Sellers WR and Kaelin Jr WG. (1997). J. Clin. Oncol. 15, 3301-3312. MEDLINE

Serrano M, Lee H, Chin L, Cordon-Cardo C, Beach D and DePinho RA. (1996). Cell 85, 27-37. MEDLINE

Serrano M, Lin AW, McCurrach ME, Beach D and Lowe SW. (1997). Cell 88, 593-602. Article MEDLINE

Sherr CJ. (1996). Science 274, 1672-1677. Article MEDLINE

Sherr CJ and Roberts JM. (1995). Genes Dev. 9, 1149-1163. MEDLINE

Smith MC, Furman TC, Ingolia TD and Pidgeon C. (1988). J. Biol. Chem. 263, 7211-7215. MEDLINE

Stepanova L, Leng X, Parker SB and Harper JW. (1996). Genes Dev. 10, 1491-1502. MEDLINE

Terada Y, Tatsuka M, Jinno S and Okayama H. (1995). Nature 376, 358-362. MEDLINE

Tucker RW, Sanford KK, Handelman SL and Jones GM. (1977). Cancer Res. 37, 1571-1579. MEDLINE

van den Bogert C, Dontje BH, Holtrop M, Melis TE, Romijn JC, van Dongen JW and Kroon AM. (1986). Cancer Res. 46, 3283-3289. MEDLINE

van den Heuvel S and Harlow E. (1993). Science 262, 2050-2054. MEDLINE

Wang TC, Cardiff RD, Zukerberg L, Lees E, Arnold A and Schmidt EV. (1994). Nature 369, 669-671. MEDLINE

Weber JD, Hu W, Jefcoat Jr SC, Raben DM and Baldassare JJ. (1997). J. Biol. Chem. 272, 32966-32971. Article MEDLINE

Wigler M, Pellicer A, Silverstein S, Axel R, Urlaub G and Chasin L. (1979). Proc. Natl. Acad. Sci. USA 76, 1373-1376. MEDLINE

Wolfel T, Hauer M, Schneider J, Serrano M, Wolfel C, Klehmann-Hieb E, De Plaen E, Hankeln T, Meyer zum Buschenfelde KH and Beach D. (1995). Science 269, 1281-1284. MEDLINE

Worzalla JF, Bewley JR and Grindey GB. (1990). Inv. New Drugs 8, 241-251.

Zindy F, Quelle DE, Roussel MF and Sherr CJ. (1997). Oncogene 15, 203-211. MEDLINE

Zuo L, Weger J, Yang Q, Goldstein AM, Tucker MA, Walker GJ, Hayward N and Dracopoli NC. (1996). Nature Genet. 12, 97-99. MEDLINE

Zwicker J, Brusselbach S, Joos KU, Sewing A, Behn M, Lucibello FC and Muller R. (1999). Oncogene 18, 19-25. MEDLINE

Figure 1 Enzymatic activity of the D1.k4 complex and D1-k4 fusion protein. (a) Autoradiograms of SDS - PAGE gels depicting phosphorylation of full-length Rb protein (indicated by the arrow). The protein molecular weight markers are shown on the left and list the size in kilodaltons. The heading indicates the enzyme being tested: D1.cdk4=cyclin D1.cdk4 enzyme complex and D1-cdk4=cyclin D1-cdk4 fusion protein. Also indicated are the l of partially purified extract from infected Sf9 cells added to the enzyme reaction mixture which included Rb and [-³²P]ATP. Phosphorylation reactions were carried out as described in Materials and methods; reaction mixtures were fractionated on SDS - PAGE gels, transferred to nitrocellulose and then exposed to Hyperfilm-ECL film. (b) Immunoblot analysis of Rb phosphorylation reaction for cyclin D1. This is the same experiment as that described in (a) above, except that the nitrocellulose membrane was analysed by immunoblotting with a -cyclin D1 antibody. (c) Immunoblot analysis of Rb phosphorylation reaction for cdk4. This is the same experiment described in (a), except that the nitrocellulose membrane was analysed by immunoblotting with a -cdk4 antibody

Figure 2 Enzymatic activity of the D1-k4 fusion protein requires functional cdk4 gene. Enzymatic assays and immunoblot analysis was done as described in Figure 1 legend. Lysates are from baculoviral infected insect cells; uninfected (lane 1), wild-type (lane 2), T172E (lane 3), D158N (lane 4), and Y17F, R24C (lane 5)

Figure 3 Photomicrographs of stably transfected REFs. All micrographs are at 20´magnification. (a) REFs stably transfected with the D1-cdk4 fusion construct and Ha-rasG12V. (1) Empty vector control - V-3, passage 4. (2) - (6) Different cell lines obtained after D1-k4+Ha-rasG12V transfections. (2) FPr-9, passage 4. (3) Fpr-4, passage 7. (4) FPr-51, passage 19. (5) FPr-2, passage 11. (6) FPr-1, passage 17. (b) REFs stably transfected with c-myc and Ha-rasG12V. (1) mr-5, passage 19. (2) mr-6, passage 17. (c) REFs stably transfected with the D1-k4 fusion construct alone. (1) Empty vector control - V-B21, passage 8. (2) - (4) Different cell lines obtained after D1-k4 transfections. (2) FP-E5B, passage 11. (3) FP-E3A, passage 15. (4) FP-E5, passage 11

Figure 4 Colony formation by cell lines derived after transfection of REFs with the D1-k4 fusion construct and Ha-rasG12V in soft agarose. Colony number and average colony size was determined 16 days after plating 900 cells in 5 cm dishes by capturing images with a video camera (Cohu, Inc.) and image analysis was done using NIH Image (V1.57) and custom macros. Examples are shown of the different degrees of colony formation observed with the 22 cell lines derived by. (a) FPr-9; -; (b) FPr-5, +; (c) FPr-2, ++; (d) FPr-51, +++

Figure 5 Enzymatic activity of the cyclin D1-cdk4 fusion protein isolated from REFs transfected with the D1-k4 fusion construct and Ha-rasG12V. REFs were grown either in the presence or absence of 0.5 g/ml doxycycline for 72 h. Lysates from REFs stably transfected with the D1-k4 fusion construct and Ha-rasG12V (FPr-5 cell line) were immunoprecipitated with a -cyclin D1 (DCS-11) antibody or a control mouse IgG2a antibody. (a) Immunoprecipiates were tested for their ability to phosphorylate full-length Rb by incubation with [-³²P]ATP, the reaction products were separated on SDS - PAGE gels, transferred to nitrocellulose, and exposed to Hyperfilm-ECL film. (b) Same experiment as above except that the nitrocellulose membrane was analysed by immunoblotting with a -cdk4 antibody to compare the amount of loading of D1-k4 immunoprecipitates in lanes with and without added Rb substrate

Figure 6 Effect of controlled expression of the D1-k4 fusion protein on the growth and morphology of REF transfectants. REFs stably transfected with the D1-k4 fusion construct and Ha-rasG12V (FPr-5, passage 21) were cultured at 3´10⁵ cells/well in 6-well plates in DMEM+10% FCS, either without or with tet (1 g/ml) for various times. (a) Triplicate wells were harvested at 0, 27, 48, or 72 h and the number of viable cells were determined by trypan blue dye exclusion. (b) Triplicate wells were harvested at 0, 3, 6, 9, 27, 48, and 72 h, lysed and analysed by immunoblotting using a -cyclin D1 antibody. The fusion protein runs at approximately 70 kD, and can be distinguished from the endogenous rat cyclin D1. ECL film images were captured with a video camera (Cohu, Inc., San Diego, CA, USA) using NIH Image (V1.57). (c) Cultures were examined and photographed before harvest at 0, 3, 6, 9, 27, 48 and 72 h. All micrographs are at 20´magnification and were made from separate dishes on different days

Figure 7 Specificity of the inhibitory effect of tet on the growth of REF cell lines. REFs stably transfected either with the D1-k4 fusion construct and Ha-rasG12V (FPr-1, passage 3), or with c-myc and Ha-rasG12V (mr-5, passage 4) were cultured at 1´10⁴ cells/well in 6-well plates either without or with tet (1 g/ml) for various times. Triplicate wells were harvested on days 0, 3, 6, 9, 12 and the number of viable cells were determined by trypan blue dye exclusion. The insert shows immunoblotting of a lysate prepared from the FPr-1 transfectants on day 9 of culture using a -cyclin D1 antibody. The protein molecular weight markers are shown on the left of the autoradiogram and list the size in kilodaltons. On the right are shown the locations of the D1-k4 fusion protein (upper band) and the endogenous rat cyclin D1 (lower bands). ECL film images were captured with a video camera (Cohu, Inc., San Diego, CA, USA) using NIH Image (V1.57)

Figure 8 Tumorigenicity of transfected REF cell lines in nude mice. Independently derived cell lines, following transfection of REFs with either the D1-k4 fusion construct or the fusion construct in combination with Ha-rasG12V were injected subcutaneously with 2.5´10⁶ cells per mouse, into seven nude mice per group. Tumor measurements (length and width) were made weekly. These measurements were used to estimate tumor mass with the following formula: tumor weight (in mg)=tumor length (in mm)´[tumor width (in mm)]²/2. The inability of non-transfected and D1-k4 fusion construct transfected REFs to form tumors under these conditions are also shown

Figure 9 Effect of delayed administration of tet on the tumorigenicity of FPr-5 cell line (D1-k4 fusion gene construct+Ha-rasG12V, passage 46) in nude mice. Tet was administered to mice in the drinking water (0.1 mg/ml in 5% sucrose) beginning either 7 days prior to tumor implantation, or was delayed until day 0 (the day of tumor implantation) or until 7 or 14 days following tumor implantation. A total of seven mice per treatment group were implanted subcutaneously with 2´10⁶ cells (FPr-5, passage 46). Mice were taken off drug on day 70 following tumor implantation and put on regular drinking water. Tumor measurements were continued in these groups until the experiment was terminated

Table 1 Summary of stably transfected rat embryo fibroblast (REF) cell lines

Table 2 Relationship between the formation of colonies in soft agarose and tumor formation in nude mice