Regulation of Cellular Genes in a Chromosomal Context by the ...

MOLECULAR AND CELLULAR BIOLOGY, Aug. 1998, p. 4565–4576 0270-7306/98/$04.0010 Copyright © 1998, American Society for Microbiology. All Rights Reserved.

Vol. 18, No. 8

Regulation of Cellular Genes in a Chromosomal Context by the Retinoblastoma Tumor Suppressor Protein ANN MARIE BUCHMANN,† SATHYAMANGALAM SWAMINATHAN,‡ AND BAYAR THIMMAPAYA* Robert H. Lurie Cancer Center and Department of Microbiology and Immunology, Northwestern University Medical School, Chicago, Illinois Received 31 March 1998/Accepted 6 May 1998

The retinoblastoma tumor suppressor gene product (pRb) is involved in controlling cell cycle progression from G1 into S. pRb functions, in part, by regulating the activities of several transcription factors, making pRb involved in the transcriptional control of cellular genes. Transient-transfection assays have implicated pRb in the transcription of several genes, including c-fos, the interleukin-6 gene, c-myc, cdc-2, c-neu, and the transforming growth factor b2 gene. However, these assays place the promoter in an artificial context and exclude the effects of far 5* upstream regions and chromosomal architecture on gene transcription. In these experiments, we have studied the role of pRb in the control of cell cycle-related genes within a chromosomal context and within the context of the G1 phase of the cell cycle. We have used adenovirus vectors to overexpress pRb in human osteosarcoma cells and breast cells synchronized in early G1. By RNase protection assays, we have assayed the effects of this virus-produced pRb on gene expression in these cells. These results indicate that pRb is involved in the transcriptional downregulation of the E2F-1, E2F-2, dihydrofolate reductase, thymidine kinase, c-myc, proliferating-cell nuclear antigen, p107, and p21/Cip1 genes. However, it has no effect on the transcription of the E2F-3, E2F-4, E2F-5, DP-1, DP-2, or p16/Ink4 genes. The results are consistent with the notion that pRb controls the transcription of genes involved in S-phase promotion. They also suggest that pRb negatively regulates the transcription of two of the transcription factors whose activity it also represses, E2F-1 and E2F-2, and that it plays a role in downregulating the immediate-early gene response to serum stimulation. that pRb indirectly controls the transcription of these genes (6, 9, 33). Complicating the picture, however, is the fact that there are at least five E2F family members which complex with at least two heterodimeric binding partners, DP-1 and DP-2 (13, 18, 22, 56, 60). In addition, there are two other members of the pRb family, p107 and p130, which bind members of the E2F family and seem to play a role in cell cycle control or differentiation. p107 also arrests cells in G1 and represses transcription from E2F sites (45, 61); it may be involved in cell cycle progression through G1 and S phases. Less is known about the role of p130, although it seems to be acting in the G0/G1 phase transition or in the process of differentiation (4, 50). It is unclear at present whether different members of the E2F and pRb families control different sets of genes or whether they are functionally redundant in transcriptional control. Previous transient-transfection studies have shown that pRb is able to downregulate expression from the c-fos, c-myc, cdc-2, neu, and Rb promoters and to upregulate expression from the transforming growth factor-b2 and insulin-like growth factor promoters (6, 15, 26, 35a, 39, 43, 58). However, such assays place the promoter of the gene of interest in an artificial setting, one that may be missing crucial 59 regulatory sequences or which ignores the effects of chromosomal architecture or chromosome position on cellular transcription. Also, pRb is often expressed out of context, in a period of the cell cycle when it would normally be inactive for E2F or other transcription factor binding. Thus, the results may be contradictory or misleading. Experiments which have attempted to identify pRb-regulated genes in a chromosomal context have relied on two approaches: overexpression of pRb under an inducible promoter and measurement of changes in gene expression in pRb2/2 cells. One experiment, in which Rb was expressed under the

The retinoblastoma tumor suppressor protein (pRb) is a cell cycle-regulated protein which is phosphorylated at specific points during the cell cycle. The first of these phosphorylations occurs near the G1/S phase boundary and is mediated by cyclin D1/cdk4 or cdk6 and cyclin E/cdk2 (8, 19, 38). Before this series of phosphorylations, pRb acts as a master switch for cell cycle progression, holding cells in G1 until they are ready to progress into S phase. This is the result of pRb binding to several key regulators of cell cycle progression including cyclin D, c-abl, and members of the E2F family of transcription factors as well as transcription factors ATF-2, Sp1, and Sp3 (3, 10, 11, 14, 53). By binding members of the E2F family and other transcription factors, pRb plays an indirect role in controlling cell cyclerelated transcription. When pRb is bound to E2F early in G1, it actively represses transcription from E2F sites (17, 52). Several genes containing E2F sites are important in cell cycle progression, including the cdc-2, c-myc, b-myb, dihydrofolate reductase (DHFR), thymidine kinase (TK), thymidylate synthase, and E2F-1 genes (6, 21, 23, 44). At the G1/S phase boundary, pRb is phosphorylated and releases the E2F species bound to it, allowing these family members to transcribe E2Fcontrolled genes. Indeed, protein levels of the TK, thymidylate synthase, and cdc-2 genes increase after this point, suggesting * Corresponding author. Mailing address: Robert H. Lurie Cancer Center and Department of Microbiology and Immunology, Northwestern University Medical School, 303 East Chicago Ave., Chicago, IL 60611-3088. Phone: (312) 503-5224. Fax: (312) 908-1372. E-mail: bayar @casbah.acns.nwu.edu. † Present address: Department of Adult Oncology, Dana Farber Cancer Institute, Boston, MA 02115. ‡ Present address: ICGEB, Aruna Asaf Ali Marg, New Delhi 110067, India. 4565

4566

BUCHMANN ET AL.

tetracycline-responsive promoter, revealed that several genes were upregulated, including those encoding proteoglycans and endothelin-1 (40). Cells from pRb-negative mice show higher levels of thrombospondin, proliferating-cell nuclear antigen (PCNA), prohibitin, and ST2 but lower levels of EIF-5, suggesting that the genes encoding these proteins may be regulated by pRb (40). However, overexpression or loss of pRb for long periods has profound effects on tumorigenicity, growth characteristics, morphology, and state of differentiation. Therefore, it is possible that some of the changes seen in gene expression result from the effect of pRb on the cellular processes and not from direct regulation by pRb. pRb2/2 cells synchronized in G1 were found to have deregulated expression of cyclin E and p107 but were unchanged in the expression patterns of several other genes including the B-myb, cdc2, E2F-1, thymidylate synthase, cyclin A2, DHFR, TK, PCNA, and DNA polymerase a genes (20). In these experiments, we have examined the effect of pRb on the transcription of cell cycle-related genes in a chromosomal setting and in the context of the G1 phase of the cell cycle. To do this, we have constructed an adenovirus (Ad) vector which is able to overexpress pRb in most cell types. We have expressed this gene in SAOS cells (pRb-negative human osteosarcoma cells) and MCF-10A cells (immortalized human breast cells) in the early G1 phase of the cell cycle and have examined the effects of pRb on the transcription of cell cyclerelated genes. The genes assayed were either E2F controlled (the E2F-1, TK, c-myc, DHFR, and PCNA genes), cyclin inhibitors (the p21/CIP1 and p16/INK-4 genes) or members of the E2F and pRb families (the E2F-1 through E2F-5, DP-1, DP-2, and p107 genes). pRb downregulated the expression of the E2F-1, E2F-2, and p107 genes. It also downregulated the expression of the DHFR, c-myc, and PCNA genes in both SAOS and MCF-10A cells and the TK gene in MCF-10A cells. Additionally, it downregulated the expression of the cyclin inhibitor p21 gene, which at low levels serves as a focal point for the construction of active cyclin-cdk-PCNA complexes and at high levels is a general inhibitor of cyclin activity. The assay described can be easily adapted for the discovery of new genes regulated by pRb. MATERIALS AND METHODS Cell culture and viruses. SAOS cells were a kind gift of Pradip Raychaudri (University of Illinois) and were grown in Dulbecco’s modified Eagle’s medium (DME) containing 10% donor calf serum. MCF-10A cells were a gift from Sigmund Weitzman (Northwestern University) and were grown in a 1:1 mixture of Ham’s F12–DME containing 5% horse serum, 10 mg of insulin per ml, 100 ng of cholera toxin per ml, 0.5 mg of hydrocortisone per ml, and 20 ng of epidermal growth factor per ml. AdRb was constructed by ligating the expression cassette of vector pCMVHARb into the E1A region (map units [m.u.] 0 to 4) of Ad5 309/356. pCMVHA consists of the full-length human Rb (huRb) cDNA (nucleotides [nt] 1 to 2798) (12) under the control of the cytomegalovirus (CMV) promoter, a strong constitutive promoter. The 12-amino-acid influenza virus hemagglutinin (HA) epitope was inserted in frame into the 59 end of the Rb gene, and the plasmid contained the simian virus 40 (SV40) polyadenylation signal at the 39 end of the gene. Ad5 309/356 contains a 5-m.u. deletion in the E3 region of the Ad and a 2-bp deletion in the E4 open reading frame 6/7 region which inactivates the E4 open reading frame 6/7 gene. AdDRb was a kind gift of Jeffrey Leiden (University of Chicago) (2) and consists of an unphosphorylatable form of the mouse Rb gene (in which nine phosphorylation sites have been mutated; Dp34 in reference 15) under the control of the EF1 alpha promoter with an HA epitope tag at the 39 end of the gene. This Rb expression cassette was cloned into the E1A/E1B region of Ad5. Adb-gal was a gift of Aviand Ayer (Northwestern University) and contains the b-galactosidase (b-gal) gene under the control of the CMV promoter cloned into the E1A/E1B region of Ad5. Viral titers were determined by standard plaque assays on 293 cells. Western blots. Protein was extracted from SAOS or MCF-10A cells at various times after infection by placing the cells in 13 RIPA buffer (0.15 M NaCl, 0.1% sodium dodecyl sulfate [SDS], 1% sodium deoxycholate, 1% Triton X-100, 1 mM EDTA, 20 mM Tris z Cl [pH 7.5]) on ice for 15 min. The solution was then centrifuged at 17,000 rpm with a Sorvall SS34 rotor for 30 min to remove

MOL. CELL. BIOL. impurities, and the protein concentration was determined by the Bradford method (Bio-Rad). A 50-mg sample of protein was run on a 7.5% polyacrylamide–SDS gel and transferred to Hybond C nitrocellulose membranes (Amersham) with the Bio-Rad protein transfer apparatus. The membrane was then blocked in phosphate-buffered saline (PBS)–5% dry milk–0.1% Tween for 1 h to overnight and probed for 1 h with either anti-pRb antibody (Ab-5 [Oncogene] at a dilution of 1:100 in PBS-Tween) or anti-HA antibody (16B12 [BAbCO] at a dilution of 1:1,000). The blots were washed, placed for 1 h in the second antibody (horseradish peroxidase-linked anti-mouse immunoglobulin) at a concentration of 1:1,000, and then developed with the ECL Western blotting kit (Amersham). Cell cycle synchronization. SAOS cells were synchronized as described previously with slight modifications (14). Briefly, the cells were first arrested in early S phase with 5 mg of aphidicolin per ml for 16 h. The cells were then released from the aphidicolin block by a brief wash with DME and allowed to progress into G2 phase for 7 h. After this, they were infected at approximately 40 PFU/cell with one of three viruses (AdRb, AdDRb, or Adb-gal) or mock infected with PBS for 1 h. After infection, the cells were placed for 10 h in fresh medium containing 100 ng of nocodazole per ml to arrest the cells in late M phase. At the end of 10 h, the rounded, loosely attached mitotic cells were harvested by mitotic shake-off and replated. The cells were harvested for fluorescence-activated cell sorter (FACS) analysis, Western blotting, or RNase protection assays at various times thereafter. MCF-10A cells were synchronized by starvation in 1:1 Ham’s F12–DME for 48 h. At 12 h before serum stimulation, the cells were infected with 40 PFU of AdRb, AdDRb, or Adb-gal per cell or were mock infected with PBS. The cells were stimulated with the growth medium described above and harvested for FACS analysis, Western blotting, or RNase protection assays at various times thereafter. FACS analysis. SAOS cells or MCF-10A cells synchronized and infected by the above methods were harvested at various times after replating or serum stimulation, washed with PBS containing 0.1% glucose, and resuspended in 70% ethanol for overnight fixation. They were then centrifuged, and the pellet was resuspended in approximately 1 ml of a solution containing 200 mg of propidium iodide per ml, 180 U of RNase A per ml, and 0.1% glucose in PBS. The cells were incubated at 4°C overnight and analyzed on a FACSort apparatus (BectonDickinson) the next day. RNase protection assays. SAOS cells synchronized by the double-drug method were harvested 6 h after replating; serum-starved MCF-10A cells were harvested 12 h after serum stimulation. RNA from infected synchronized cells was isolated and purified by the guanidinium isothiocyanate (GIT) purification method. The cells were washed twice with PBS and lysed with GIT buffer (4 M GIT, 20 mM sodium acetate [pH 5.2], 0.1 mM dithiothreitol [DTT], 0.5% N-lauroylsarcosine). The GIT-RNA solution was overlaid on 1.5 ml of 5.7 M CsCl and centrifuged at 35,000 rpm with a Beckman SW50.1 rotor for 18 to 20 h. The RNA pellet was then resuspended, extracted once with phenol-chloroform, and precipitated with ethanol. For RNase protection assays, partial cDNAs of several cell cycle-related genes were subcloned in antisense orientation into Bluescript T3/T7 vectors (Stratagene) and transcribed into 32P-labeled probes. The probes used were the 39 activation domain of E2F-1 (provided by P. Raychaudri), nt 1576 to 1291 of E2F-2 (E. Harlow, Massachusetts General Hospital), nt 730 to 340 of E2F-3 (E. Harlow), nt 1220 to 961 of E2F-4 (R. Weinberg, Whitehead Institute), nt 1057 to 768 of E2F-5 (R. Weinberg), nt 860 to 585 of DP-1 (E. Harlow), nt 1260 to 1013 of DP-2 (E. Harlow), nt 600 to 440 of p21 (D. Beach, Cold Spring Harbor Laboratory), nt 1140 to 780 of cyclin D1 (D. Beach), nt 632 to 446 of PCNA (D. Beach), nt 1343 to 1094 of c-myc (L. Lau, University of Illinois), nt 3172 to 3052 of p107 (E. Harlow), nt 321 to 133 of DHFR (Thimmapaya laboratory collection), nt 425 to 1 of p16 (D. Beach), nt 1343 to 958 of TK (Thimmapaya laboratory collection), and nt 660 to 256 of glyceraldehyde-3-phosphate dehydrogenase (N. Bouck, Northwestern University). For each RNase protection assay, 15 mg of total RNA was hybridized to 3 3 105 cpm of a 100- to 500-base single-stranded antisense 32P-labeled RNA probe for 16 to 18 h at 60°C in 10 ml of hybridization buffer {80% formamide, 40 mM PIPES [pH 6.7; piperazine-N,N9-bis(2-ethanesulfonic acid)], 0.4 M NaCl, 1 mM EDTA}. The hybridization product was then digested with 1 to 5 U of RNase ONE (Promega) in 250 ml of 0.2 M sodium acetate–5 mM EDTA–10 mM Tris z Cl (pH 7.5) for 1 h at 32°C. The RNase-resistant RNA-RNA complex was then extracted once with phenol-chloroform, ethanol precipitated, and analyzed on a 4% polyacrylamide–8 M urea gel. Nuclear run-on assays. MCF-10A cells were synchronized and infected with AdDRb or Adb-gal as described in “Cell cycle synchronization.” Nuclei from these infected cells were harvested by the following method. The cells were washed twice with ice-cold PBS and were then lysed for 10 min on ice in a buffer consisting of 0.3 M sucrose, 10 mM Tris z Cl (pH 7.4), 5 mM MgCl2, 0.4% Nonidet P-40, and 0.5 mM DTT. They were then disrupted by 20 strokes in a Dounce homogenizer. The lysed-cell suspension was gently layered over a sucrose cushion consisting of 0.88 M sucrose, 10 mM Tris z Cl (pH 7.4), 5 mM MgCl2, 0.4% Nonidet P-40, and 0.5 mM DTT. This mixture was centrifuged at 2,500 rpm with a Sorvall tabletop centrifuge for 5 min. The pellet, which consisted of the intact nuclei, was then frozen in storage buffer (40% glycerol, 50 mM Tris z Cl [pH 8.0], 5 mM MgCl2, 0.1 mM EDTA) at a concentration of 108 cells per ml.

VOL. 18, 1998

pRb REGULATION OF CELL CYCLE-RELATED GENES

FIG. 1. Viruses used in cell cycle experiments. AdRb contains an HA-tagged full-length human cDNA under the control of a constitutive CMV promoter inserted into the E1A region of Ad. AdDRb contains an HA-tagged mutant mouse Rb cDNA under the control of the EF1 alpha promoter inserted into the E1A/E1B region of Ad. Asterisks indicate mutations in nine phosphorylation sites in the Rb cDNA: Thr-246, Ser-601, Ser-605, Ser-780, Ser-786, Ser-787, and Ser-800 were changed to Ala; Thr-350 was changed to Arg; and Ser-804 was changed to Ala. These mutations render the protein unphosphorylatable. Adbgal, used as a control in these experiments, contains the b-gal cDNA under the control of the CMV promoter inserted into the E1A/E1B region of Ad.

Nuclear run-on probes were created by incubating 50 ml of nuclei with 10 ml of [a-32P]UTP (3,000 Ci of NEG BLU-013H per mmol), 25 ml of 43 reaction mix (100 mM HEPES [pH 7.4], 10 mM MgCl2, 10 mM DTT, 300 mM KCl, 20% glycerol), 12.5 ml of 83 triphosphate mix (2.8 mM each ATP, GTP, and CTP, plus 3.2 mM UTP), and 1 ml of RNasin. This mixture was incubated at room temperature for 30 min, 10 ml of RQ1 RNase-free DNase (1,000 U/ml) was added, and the mixture was incubated for an additional 30 min. The probe was purified by the addition of 300 ml of Trizol and 100 ml of chloroform. This mixture was vortexed and centrifuged at 15,000 rpm with a Sorvall SS34 rotor for 15 min. The aqueous layer was then removed and ethanol precipitated overnight. Equal counts of run-on probe were hybridized at 42°C for approximately 72 h to cDNAs spotted at concentrations ranging from 4 to 0.1 mg per slot. After hybridization, the membranes were washed several times and then were dried and exposed to autoradiography film overnight.

RESULTS Ectopic expression of pRb via Ad vectors arrests cells in G1. To overexpress pRb in every cell of a given population, we constructed an Ad vector, AdRb, which contains the fulllength human Rb gene cDNA under the control of the CMV promoter in the E1A region of Ad5 (Fig. 1). To confirm the results seen with AdRb, a second Rb-expressing Ad, AdDRb, was obtained from Jeffrey Leiden. AdDRb contains a nonphosphorylatable mouse Rb cDNA (Dp34 in reference 15) under the control of the EF1 alpha promoter cloned into the E1A/ E1B region of Ad5 (Fig. 1). The control for all of our experiments was Adb-gal, which contains the b-gal gene under the

4567

control of the CMV promoter cloned into the E1A/E1B region of Ad5 (Fig. 1). SAOS cells were used in these experiments because they lack endogenous pRb and because their response to exogenously added pRb has been well characterized in earlier studies (14, 19). MCF-10A cells are a nontransformed line of human breast cells. These cells arose as a spontaneous immortalization of a normal breast epithelial cell line, MCF-10 (49). They were used as a nontransformed secondary cell line to establish that the changes seen in SAOS cells were not contaminated by any effects of transformation, genomic instability, or long-term loss of pRb or p53 on the SAOS cells. To examine the level of protein expression produced by AdRb and AdDRb, SAOS or MCF-10A cells were infected with 40 PFU of AdRb, AdDRb, or Adb-gal per cell. The cells were synchronized as described in Materials and Methods and harvested at various times after infection. Samples (50 mg) of collected protein were run on 7.5% acrylamide– SDS gels and transferred to nitrocellulose for Western blot analysis. Western blots were probed with either anti-Rb (Ab-5) or anti-HA (16B12) antibodies and developed with the ECL kit. In SAOS cells, AdRb began to produce detectable levels of pRb at about 10 h after infection (Fig. 2A, lanes 7, 15, and 20). The 110-kDa protein was seen both with anti-HA antibodies and with anti-pRb antibodies (lanes 7, 15, 17, and 23). Neither b-gal-infected nor mock-infected cells showed any levels of the protein with either antibody (Fig. 2A, lanes 16 and 21, and data not shown). SAOS cells do produce a truncated form of the pRb protein, but this was not detected by the Oncogene Ab-5 antibody used in these assays. None of the protein produced is phosphorylated in SAOS cells, as expected since SAOS cells normally do not phosphorylate exogenously added pRb (19). In MCF-10A cells, the Rb protein produced by AdRb and AdDRb was clearly detected by 24 h after infection when the HA antibody was used (Fig. 2B, lanes 3 and 4). At this point, the cells had been serum stimulated for 12 h and were mainly in the G1 phase of the cell cycle (see Fig. 4C). At 36 h after infection (24 h after serum starvation), pRb produced by AdRb started to show several phosphorylation forms, although at least some of the protein remained in the lowest phosphorylation state (data not shown). In our protein experiments, human pRb was consistently slightly larger than mouse pRb for unknown reasons. As expected, all of the protein produced by AdDRb remained in the lowest phosphorylation state, since the phosphorylation sites of the Rb gene in AdDRb had been mutated (data not shown). Immunofluorescence of AdRb-infected SAOS cells by using the anti-HA antibody showed that about 80 to 90% of cells expressed the introduced pRb in their nuclei (data not shown). SAOS cells were synchronized in G1 by a double-drug method, as shown in Fig. 3A. The cells were first arrested in S phase by the addition of aphidicolin, an inhibitor of DNA polymerase ], for 16 h. The cells were then allowed to progress into G2, where they were infected and arrested in late M phase by nocodazole, an inhibitor of microtubule formation. At 10 h after the addition of nocodazole, the loosely attached mitotic cells were washed off in a mitotic shake-off, replated, and harvested at time intervals thereafter. FACS analysis of infected SAOS cells synchronized by the aphidicolin-nocodazole method showed that nearly 90% of the cells were in the G1 phase of the cell cycle 6 h after replating (Fig. 4A). Mock-infected and b-gal-infected cells began to enter the S phase of the cell cycle around 18 h after replating and had begun to leak into the G2/M phase by 24 h (Fig. 4B and data not shown). By 30 h after replating, b-gal-infected and uninfected cells were actively cy-

4568

BUCHMANN ET AL.

MOL. CELL. BIOL.

FIG. 2. Western blot analysis of pRb produced upon AdRb or AdDRb infection of SAOS and MCF-10A cells. The human pRb protein band (approximately 110 kDa) is marked by a solid arrow; the mouse pRb is marked by an open arrow. (A) SAOS cells infected with AdRb. The protein is first seen in AdRb cells about 10 h after infection (lane 7), and its levels increase at times thereafter. (B) MCF-10A cells infected with AdDRb. pRb is seen as a single band at 24 h after infection in both AdRb- and AdDRb-infected cells in Western blots probed with an anti-HA antibody.

cling, with many of the cells reentering the G1 phase for a second cycle (data not shown). A majority of AdRb-infected cells, however, remained arrested in the G1 phase of the cell cycle, never progressing into S and G2/M. Figure 4B shows the results for the 24-h time point. At this time point, about 50% of Adb-gal-infected cells and 55% of uninfected cells had exited G1; however, only about 15% of the AdRb-infected cells had done so. Over 80% of AdRb-infected cells remained in the G1 phase of the cell cycle at all time points assayed. These results mimic those seen with pRb added to SAOS cells by other methods (transfection or microinjection), in which the added pRb arrests the cells in G1 (14, 36). In addition, cells infected with AdRb or AdDRb display the distinctive SAOS “large-cell” phenotype (36) at about 72 h after infection; that is, the cells became flat and more spread out on the plate with a higher cytoplasm-to-nucleus ratio (data not shown). Thus, the pRb provided to the cells via AdRb or AdDRb acts the same as other exogenously added wild-type pRb in SAOS cells. MCF-10A cells were synchronized by 48 h of serum starvation and were infected with the Ad vectors 12 h prior to serum stimulation, as shown in Fig. 3B. At 12 h after serum stimulation, the cell cycle profiles of all populations were similar, with 87 to 94% of the cells in G1 (Fig. 4C). Mock-infected cells and cells infected with Adb-gal began to enter the S phase of the cell cycle about 18 h after serum stimulation and reached peak S phase about 21 h after serum stimulation (Fig. 4D and data not shown). Cells infected with AdDRb or AdRb tended to

remain in the G1 phase even after Adb-gal-infected cells had entered S phase. At 21 h after infection, about 84% of AdDRbinfected cells and 76% of AdRb-infected cells remained in G1 while only 36% of Adb-gal-infected cells had not progressed into S or G2. The G1 phase arrest produced by AdRb infection was less strong in MCF-10A cells than in SAOS cells, most probably because at least some of the pRb introduced into MCF-10A cells had been phosphorylated and thus inactivated. AdRb- and AdDRb-infected cells progress through the early stages of G1. Although FACS analysis showed nearly identical profiles for AdRb-, AdDRb-, and Adb-gal-infected cells at the point at which we harvested the cells for RNA analysis (Fig. 4A and C), there remained the possibility that the AdRb- and AdDRb-infected cells were arrested in G0 or very early in G1 while Adb-gal-infected cells progressed into mid-G1. Such a difference would not be seen in FACS analysis and would influence the expression levels of several of the genes tested, particularly the immediate-early genes. To address this question, we analyzed the expression level of one of the immediateearly genes, c-jun, in MCF-10A cells. The level of expression of c-jun rises dramatically within the first hour after serum stimulation and then falls back to basal levels at mid-G1 (41). MCF-10A cells were serum starved and infected as shown in Fig. 3B. Immediately before and at various times after serum stimulation, these cells were harvested and 25 mg of protein was run on SDS-polyacrylamide gels. The gels were then probed with the anti-jun antibody (sc-44X; Santa Cruz Bio-

VOL. 18, 1998


4569

FIG. 3. Time course of cell synchronization and viral infection of SAOS and MCF-10A cells. (A) SAOS cells were first arrested in S phase with aphidicolin. At 16 h later, the cells were released from the S-phase block by being refed with fresh medium. The cells were infected 7 h after the refeeding and were then arrested in late M phase with nocodazole. At 10 h later, the M-phase cells were collected by mitotic shake-off and replated. RNA was harvested for RNase protection assays 6 h after the replating. (B) MCF-10A cells were synchronized in G1 by a 48-h serum starvation followed by serum stimulation. The cells were infected 12 h before serum stimulation, and the RNA was harvested 12 h after serum stimulation.

technology). As shown in Fig. 5, AdRb-, AdDRb-, and Adbgal-infected populations of cells all showed similar profiles of c-jun expression. Before serum stimulation, the level of c-Jun protein was quite low (Fig. 5, lanes 1 to 3). The level of protein began to rise about 15 min after serum stimulation (lanes 4 to 6), reached a peak about 30 min after serum stimulation (lanes 7 to 9), and fell by the end of the first hour after serum stimulation (lanes 10 to 12). The timing of the rise and fall of c-jun expression was identical in all three populations, indicating that the three populations enter the G1 phase of the cell cycle at about the same time and begin to progress through G1 at roughly similar rates. All three populations continued to show similar levels of c-jun expression through late G1, when the cells were harvested (Fig. 5, lanes 13 to 24). This shows that the AdRb-, AdDRb-, and Adb-gal-infected cells do progress through the very early stages of the cell cycle at roughly the same rate. There is, of course, still the possibility that the AdRb- or AdDRb-infected cells arrest before the time point at which we harvested the cells while Adb-gal-infected cells continue to progress through G1. We consider this to be unlikely since MCF-10A cells progress through the restriction point only at about 15 to 16 h after serum stimulation as measured by a rise in the level of cyclin E protein and cyclin E-associated kinase activity (data not shown). The Ad vector-infected MCF10A cells were harvested at least 3 to 4 h before this point. pRb regulates transcription of genes containing E2F sites. To assess the effect of pRb on cell cycle-related genes, RNase protection assays were performed on RNAs harvested from AdRb-, AdDRb-, and Adb-gal-infected SAOS cells 6 h after replating and from infected MCF-10A cells 12 h after serum stimulation, when nearly 90% of the cells in both populations were in early G1 (Fig. 4A and C). A 15-mg sample of harvested RNA was hybridized to 100- to 500-nt antisense RNA probes. The hybridization products were then digested with small amounts of RNase and run on 4 to 6% denaturing gels. For a summary of the mRNAs assayed and the degree of regulation by pRb, see Fig. 9A. Glyceraldehyde-3-phosphate dehydrogenase, a constitutively expressed housekeeping gene, was used side by side as a control in these assays and never

showed more than a 10% variation among RNA samples. For the results of a typical experiment, see Fig. 8D. pRb is known to actively repress transcriptional activity from E2F sites; therefore, several of the genes we chose to assay were known to be regulated by E2F. These were the E2F-1, TK, c-myc, PCNA, and DHFR genes. All of these genes showed some degree of downregulation by pRb. Although none of the differences were overwhelming, this level of downregulation was enough to hold the majority of cells in G1 (Fig. 4B and D). Each of the assays was performed at least three times, with consistent results. E2F-1 gene expression was consistently downregulated fourto fivefold in both MCF-10A and SAOS cells. Figure 6A shows the results of a representative experiment from SAOS (top panel) and MCF-10A (bottom panel) cells. In each case, lane 1 shows the 280-nt undigested probe. This probe contains a nonspecific 30-nt tail, which remains unhybridized and is digested by the RNase treatment; thus, the protected RNA fragments in lanes 2 through 5 are about 30 nt smaller than the probe shown in lane 1. Lanes 2 and 3 show the amount of RNA present in AdRb- and AdDRb-infected cells, respectively, whereas lane 4 shows the darker E2F-1-specific band in the Adb-gal-infected cells. RNA from Adb-gal-infected cells was used as a control in each of these assays to take into account perturbations in gene expression, if any, as a result of Ad infection. In nearly every case, the level of gene expression in Adb-gal-infected cells was comparable to that in mock-infected synchronized cells (data not shown). As expected under these conditions, hybridization of the probes with tRNA, the negative control, failed to show the probe-specific band in all experiments (lane 5). PCNA and DHFR mRNAs both showed a 2- to 3.5-fold downregulation in response to pRb (Fig. 6B and C, compare lanes 2 and 3 with lane 4). TK gene expression was repressed two- to threefold by pRb in MCF-10A cells; however, no repression was seen in several experiments with SAOS cells (Fig. 6D). Presumably, TK, which is expressed at higher levels in SAOS cells, has lost the ability to be regulated by pRb in these cells or the repression was simply too low to be seen in these

4570

BUCHMANN ET AL.

MOL. CELL. BIOL.

FIG. 4. FACS analysis of infected and uninfected cells. (A) SAOS cells 18 h after virus infection and 6 h after replating following mitotic shake-off. Nearly 90% of the cells in all populations are synchronized in G1. Numbers inside each grid refer to the percentage of cells in each stage of the cell cycle. (B) SAOS cells 36 h after virus infection and 24 h after replating. AdRb-infected cells remain arrested in G1, while the majority of Adb-gal- and mock-infected cells have progressed into S and G2. (C) MCF-10A cells 24 h after infection and 12 h after serum stimulation. Nearly 90% of the cells in all populations are synchronized in G1. (D) MCF-10A cells 33 h after virus infection and 21 h after serum stimulation. A majority of AdRb- and AdDRb-infected cells remain arrested in G1, while Adb-gal- and mock-infected cells exit G1 and progress into S and G2.

assays. Expression of the c-myc gene showed a 3.5- to 5-fold downregulation in response to pRb (Fig. 6E). pRb downregulates the transcription of E2F-2 and p107. Since pRb repressed the transcription of the E2F-1 gene, we decided to see if it also repressed other members of the E2F gene family. The promoters of genes encoding these family

members have not been analyzed for transcription factor binding sites; however, E2F-2 levels increase dramatically near the end of G1, coincidentally with the rise in E2F-1 levels (22). Thus, it seemed possible that the promoters of at least some of these genes were regulated by pRb. Along with the E2F-1 gene, E2F-2 gene expression was down-

VOL. 18, 1998

FIG. 5. Western analysis of c-jun expression in AdRb-, AdDRb-, and Adbgal-infected MCF-10A cells. c-jun expression is quite low in cells before serum stimulation (0 h) (lanes 1 to 3) but rises within 15 min of serum stimulation (115 min) (lanes 4 to 6) and peaks at about 30 min after serum stimulation (130 min) (lanes 7 to 9). The protein level then declines to near basal levels by 1 h after serum stimulation (160 min) (lanes 10 to 12). All populations of infected cells show similar levels of c-jun at all time points tested. The size of c-jun was larger than expected, most probably because the gel conditions used were not rigorous enough to dissociate c-jun/c-fos complexes.

regulated by pRb expression, showing a 2.5- to 3-fold inhibition in AdRb-infected SAOS and MCF-10A cells and about a 4.5to 5.5-fold inhibition in AdDRb-infected cells (Fig. 7A). The remainder of the mRNAs from genes encoding E2F family members E2F-3 (Fig. 7B), E2F-4 (Fig. 7C), E2F-5 (Fig. 7D), DP-1 (Fig. 7E), and DP-2 (Fig. 7F) showed no response to pRb.


4571

The Rb family member p107 is also able to arrest some cells in G1 and to inhibit the activity of some of the E2F family members (45, 61). The promoter of the p107 gene contains two E2F sites (62); therefore, it seemed possible that the p107 gene would be regulated by pRb. In our experiments, p107 mRNA was downregulated about five- to sevenfold in both SAOS and MCF-10A cells (Fig. 8A). pRb represses transcription of p21 but not p16. pRb arrests cells in G1 when overexpressed or when maintained in an underphosphorylated state. Since one of the primary mechanisms by which cell cycle arrest can occur is by an inhibition of cdk activity, we examined the effects of pRb on two inhibitors of cdk activity, p16/Ink4 and p21/Cip1. p21 is a general inhibitor of cdk activity which is known to arrest cells in G1 when upregulated by p53 (16). It is also upregulated in the cellular response to TGF-b1 through Sp1 sites (7). Since one of the effects of TGF-b is to maintain the Rb protein in an underphosphorylated state, it seemed possible that pRb would upregulate p21 when in the underphosphorylated state. To our surprise, p21 was repressed rather than activated by pRb, showing a two- to threefold downregulation in both SAOS and in MCF-10A cells (Fig. 8B; compare lanes 2 and 3 with lane 4). p16 is an inhibitor of the cyclin D-cdk4 or cyclin D-cdk6 complex (42). It arrests cells in G1 when overexpressed; however, this arrest is dependent on the presence of wild-type pRb (30). It is presumed, therefore, that the primary action of p16 is to prevent cyclin D1 from phosphorylating pRb, keeping cells from progressing into the S phase. Since the level of p16 influences the phosphorylation state of pRb, we wished to determine whether pRb plays a role in the transcriptional control of the gene encoding this protein. In SAOS cells, however, the levels of p16 mRNA were quite high and did not alter significantly upon infection with pRb-expressing Ad (Fig. 8C). In MCF-10A cells, the level of p16 was so low that it was un-

FIG. 6. Riboprobe analysis of mRNAs encoding E2F-controlled genes. In each case, lane 1 contains the full-length probe, lane 2 contains the protected RNA fragment from AdRb-infected cells, lane 3 contains the protected RNA from AdDRb-infected cells, lane 4 contains the protected RNA from Adb-gal-infected cells, and lane 5 contains tRNA, used as a negative control. The solid arrow marks the location of the undigested probe; the open arrow marks the location of the RNA-RNA hybrid. The upper half of each segment shows data from virus-infected SAOS cells; the lower half shows data from MCF-10A cells.

4572

BUCHMANN ET AL.

MOL. CELL. BIOL.

FIG. 7. Riboprobe analysis of mRNAs encoding E2F family members. Lane numbers and arrows are as described in the legend to Fig. 5.

detectable in RNase protection assays; therefore, the results seen in SAOS cells could not be confirmed in these cells. As mentioned above, Fig. 9A is a summary of the results obtained with SAOS cells, showing the mean and standard error of the mean for at least three separate experiments. Of the 14 cell cycle-related genes tested, 7 showed some response to the presence of pRb. However, the level of repression by pRb was modest for many of these genes. The E2F-1, E2F-2, c-myc, and p107 genes showed the greatest downregulation by pRb, with between four- and eightfold repression. The DHFR, p21, and PCNA genes showed about a two- to threefold downregulation. The TK, E2F-3, E2F-4, E2F-5, DP-1, and DP-2 genes showed no response to pRb in SAOS cells, although the TK gene did show a slight downregulation in response to pRb in MCF-10A cells. Repression of gene expression by pRb occurs at the level of transcription. RNase protection assays measure changes in the levels of RNAs; however, such changes may reflect alterations either in transcription levels or in RNA stability. To determine whether the repression in RNA levels of cell cycle-related genes by pRb occurs at the level of transcription, we performed nuclear run-on assays with AdDRb and Adb-gal in MCF-10A cells synchronized by serum starvation and harvested 12 h after serum stimulation, as described above. As shown in Fig. 9B, cells infected with AdDRb showed a lower rate of transcription for all the genes which showed lower levels of RNA in Fig. 5 to 7. This experiment was repeated three times with three separate sets of nuclei, and the results were normalized to the E2F-4 and E2F-5 genes, whose mRNA levels did not change in RNase protection assays. The level of transcriptional repression for most of the genes measured was similar to the decrease in the amounts of RNA. That is, the TK, DHFR, and PCNA genes whose mRNA levels decreased only about 2- to 3-fold, showed only a slight de-

crease in transcription rates (2- to 3.7-fold); the E2F-1, E2F-2, c-myc, and p107 genes showed a greater decrease in transcription rates, about 5- to 6-fold, which roughly corresponds to the fold decrease in mRNA levels. The only exception to this general rule was the transcription rate of p21, which had only a slight decrease in mRNA levels but showed a sixfold decrease in transcription rates. We are uncertain why such a relatively large decrease in the transcription rate would not be reflected in a larger decrease in the RNA levels. These experiments suggest that the E2F-1, E2F-2, c-myc, p107, p21, TK, DHFR, and PCNA genes are all controlled by pRb at the level of transcription. DISCUSSION The cell cycle is an intimately regulated process, responding to an abundance of positive and negative signals for growth. The Rb protein is an integrator of such signals, holding the cell in G1 in response to negative signals, such as transforming growth factor-b, and becoming phosphorylated and allowing the cell to enter S phase in response to positive signals. The genes that are transcriptionally regulated by pRb are therefore expected to be important in cell cycle progression and tumorigenesis. We have examined the effect of pRb on several such genes, including genes which are regulated by E2F and genes involved in the early part of the cell cycle. Our results suggest that pRb maintains cells in G1 by downregulating at least two sets of cellular genes, genes involved in S phase progression, and at least two of the immediate-early genes. The purpose of these experiments was to systematically examine the effects of pRb on the transcription of cell cyclerelated genes within a chromosomal context and within the context of the G1 phase of the cell cycle. While transienttransfection assays have indicated that pRb plays a role in the

VOL. 18, 1998


4573

histone H4 promoter (reviewed in reference 46). In addition, architectural transcription factors such as upstream binding factor (UBF), lymphoid enhancer binding factor 1 (LEF-1), and high-mobility group protein HMG-I/Y directly bend DNA, allowing transcription factor sites up to thousands of base pairs apart to cooperatively interact, leading to an increase in transcription (reviewed in reference 54). These far-upstream regions may not be included in vectors used in transient-transfection assays. Besides architectural transcription factors that bend DNA, several proteins of the SWI/SNF2 family which seem to affect cellular transcription by displacing or disrupting histone binding have recently been cloned. These proteins include p300/ CBP, GCN5, hBRM, TAFII 250/CCG, and hBRG1 (reviewed in reference 55). pRb binds two of these proteins, hBRM and hBRG1, and overexpression of these proteins produces a growth suppression phenotype similar to that seen with overexpression of pRb (47). Thus, it is likely that transcriptional control by pRb is greatly influenced by promoter structure. Our own experiments have seen discrepancies in transcriptional regulation between transient-transfection assays and in assays in which the gene is located in its normal chromosomal

FIG. 8. Riboprobe analysis of mRNAs encoding p107 and cyclin inhibitors. Lane numbers and arrows are as described in the legend to Fig. 5. In panel C, only SAOS data are shown; the level of p16 in MCF-10A cells was too low to measure in these assays. Due to the large amount of glyceraldehyde-3-phosphate dehydrogenase RNA present in cells (D), autoradiographs were exposed for about 30 min, so the undigested probe is not visible. GAD, glyceraldehyde-3phosphate dehydrogenase.

transcription of some of these cell cycle-related genes, such assays can be unreliable and misleading because they may include only part of the entire transcriptional unit and ignore the effects that chromosome structure and the chromosomal position of the gene may have on its transcription. Recent studies have shown that the chromosomal and nuclear milieu of a gene may play an important role in determining the level of gene transcription. Promoter and enhancer regions are not simply arranged linearly but are compacted into complex three-dimensional structures, which may change during the course of the cell cycle. Transcription factor access to cellular promoter regions may be limited because these regions are tightly looped around histone cores; for the region to become activated, this arrangement of histones must be acetylated or otherwise destabilized (reviewed in reference 55). Thus, the structural arrangement of the chromatin in the cell may serve as a barrier to transcription factor and initiation complex binding and gene expression. On the other hand, the nuclear architecture may actually serve to increase the level of cellular transcription by increasing local concentrations of transcription factors and directing these transcription factors to their chromosomal contacts. YY1 and an activating transcription factor-related transcription factor have been found bound to the nuclear matrix; these may be directed to specific nuclear matrix binding sites on cellular promoters such as the

FIG. 9. (A) Graph of the changes in mRNA levels in SAOS cells infected with AdRb or AdDRb in comparison with cells infected with Adb-gal, as measured by RNase protection assays. The mRNA level in Adb-gal-infected cells was taken as 1 unit. (B) Variations in the rate of transcription in MCF-10A cells infected with AdDRb in comparison with those infected with Adb-gal, as measured by nuclear run-on assays. The transcription rate in Adb-gal cells was taken as 1 unit.

4574

BUCHMANN ET AL.

context. SV40 small-t and large-T antigens and Ad DNA binding protein transactivate the Ad E2 promoter located on a plasmid but not the same promoter located on the viral chromosome (37, 48). Other experiments which have attempted to identify pRbregulated genes in a chromosomal context have relied on two approaches: (i) overexpression of pRb via transient transfection or under an inducible promoter or (ii) measurement of changes in gene expression in pRb2/2 cells. In one such experiment, overexpression of pRb via a tetracycline-responsive promoter in Bt549 cells led to the upregulation of several genes including those encoding endothelin-1 and proteoglycans PG40 and versican (40). The upregulation of these genes may contribute to the suppression of tumorigenicity and change in morphology that occurs in these cells when pRb is reexpressed. Cells from pRb-negative mice have higher levels of thrombospondin, PCNA, prohibitin, and ST2 but lower levels of EIF-5 (40), suggesting that some of the genes encoding these proteins may be regulated by pRb. However, overexpression or loss of pRb for long periods has profound effects on the tumorigenicity, growth characteristics, morphology, and state of differentiation of a cell. Therefore, it is possible that at least some of the changes seen in gene expression levels are secondary results of the effect of pRb on the cellular processes and are not due to a direct regulation by pRb. Also, the reintroduced pRb is expressed out of context, in a period of the cell cycle when the protein would normally be inactive for E2F or other transcription factor binding. The Ad vector system is ideal for examination of the shortterm effects of proteins on cellular transcription and cell cyclerelated events. Ad vectors are able to infect nearly every cell of a given population and produce the protein of interest at high levels within a few hours of infection. Unlike permanent cell lines, the system is not leaky; before Ad infection, the cells do not overexpress the gene in question and there is no need to consider the possible effects of long-term, low-level overexpression. pRb binds E2F in G1 and actively represses transcription from E2F sites. Our studies show that pRb regulates the transcription of several genes which contain E2F sites, i.e., the E2F-1, PCNA, DHFR, TK, and c-myc genes. While we have not shown that the regulation of any of these genes by pRb occurs through the E2F sites, these results are consistent with earlier observations that E2F-1 upregulates transcription from its own promoter and that this upregulation is further increased upon cotransfection with cyclin D-cdk4, the complex which phosphorylates pRb and releases E2F-1 (23). Overexpression of E2F-1 induces quiescent cells to enter S phase as long as the protein retains the ability to induce the TK and DHFR promoters (24), indicating that repression of E2F-1 transcription and activity by pRb may be the most crucial event holding the cells in G1. The PCNA, TK, and DHFR genes are all necessary for S-phase replication of DNA, and repression of these genes by pRb would tend to maintain the G1-phase arrest of cells as long as pRb remains in the underphosphorylated state. The levels of E2F1, PCNA, TK, and DHFR mRNA all increase near the transition from the G1 to the S phase (9, 23, 27, 44), around the same time that pRb is phosphorylated and E2F is released, and the DHFR gene is present at a higher level in pRb-negative mouse embryo fibroblasts than in those which contain functional pRb (1, 44); therefore, regulation of these genes by pRb is consistent with most earlier reports. However, one recent examination of pRb2/2 mouse fibroblasts synchronized in G1 showed that none of these genes were derepressed in the absence of pRb, suggesting that pRb had no role in regulating these genes (20). One possible explanation

MOL. CELL. BIOL.

for the discrepancy between these two sets of data is that pRb, p130, and p107 play redundant roles in regulating these essential cell cycle genes. In p1302/2 or p1072/2 fibroblasts, the E2F-1 and DHFR genes were not upregulated; however, in p130/p107-double-negative cells, these genes were depressed in G1. Thus, in pRb2/2 cells, p107 and/or p130 may have retained the ability to control these essential S-phase genes. In support of this is the finding that in pRb2/2 cells, p107 is greatly upregulated in G1 and binds to members of the E2F family that are normally bound by pRb (20). Of course, it is also possible that the overexpression of pRb in our experiments produces some artificial results. We cannot, at this time, rule out the possibility that overexpression of pRb causes the protein to be in complexes that are not found under physiological conditions. However, it is almost impossible to study the function of a protein under normal physiological conditions, and it seems probable that the short period of pRb overexpression and the rather small degree of pRb overexpression (less than threefold for AdRb-infected MCF-10A cells) would limit the number of artifacts seen. The fact that the results are roughly similar for two separate viruses in two distinct cell lines in two separate assays (RNase protection assays and nuclear run-on assays) argues that the downregulation of these genes by pRb is physiologically relevant. c-myc has also been shown to be controlled transcriptionally by E2F, and transient-transfection assays have implicated pRb in the transcriptional regulation of the gene (15, 34, 35a). In addition, pRb plays a role in the repression of c-myc transcription during the differentiation of HL60 cells (21). The pattern of c-myc expression during the cell cycle, however, differs from that of other E2F/pRb-controlled genes, and it has been uncertain whether pRb plays a role in the control of c-myc levels during the course of the normal cell cycle. c-myc is an immediate-early gene; its RNA and protein levels increase dramatically a few hours after the start of G1 (5) and then rapidly decline within hours after this early rise (5). The repression of c-myc seen in our experiments may indicate that pRb is involved in this rapid decline. To our surprise, pRb also repressed the transcription of a second immediate-early gene, the p21 gene, which encodes a general inhibitor of cyclin activity. The levels of p21 mRNA and protein increase shortly after serum stimulation, an increase that may be mediated by growth factors (31, 32). The amount of protein then declines to basal levels by mid-G1 (31). This rise in the p21 level early in the cell cycle may actually facilitate cell cycle progression. Studies have demonstrated the importance of the stoichiometric ratio of p21 relative to the levels of cyclin, cdk, and PCNA. When p21 levels are roughly equal to the levels of these other proteins, p21 acts as an anchor for the formation of active cyclin-cdk-PCNA complexes, with one molecule of each protein per complex (59). Thus, at low levels, p21 actually increases the amount of cdk activity by promoting the formation of active cyclin-cdk complexes. The increase in p21 mRNA and protein in early G1 correlates with an increase in p21-linked cyclin-cdk activity (31). Only when the level of p21 exceeds those of the other components of the cyclin-cdk-PCNA-p21 complex does the protein become growth inhibitory and cdk activity decline abruptly (59). One explanation for the role of pRb in p21 control may be to mitigate the immediate-early gene response, decreasing the p21 level in mid-G1 to below that necessary for efficient cyclin-cdk complex formation. An alternative explanation is that the transcriptional regulation of p21 by pRb may be a mechanism by which pRb indirectly promotes its own phosphorylation, by the repression of a negative regulator of the cyclin D-cdk4 or cyclin E-cdk2

VOL. 18, 1998

complexes that phosphorylate pRb. Although we consider this scenario less likely, considering the low level of repression by pRb, it is certainly possible. It will be interesting to see whether the repression of p21 mRNA levels by pRb has an effect on p21 protein levels and whether it has a functional effect on the cyclin D- or cyclin E-related cdk activity in SAOS and MCF10A cells. If pRb causes the levels of p21 to decline below that needed for cyclin-cdk complex assembly, the kinase activity may decline, contributing to the G1 phase arrest. If pRb is serving to rid the cells of excess inhibitory p21, however, the kinase activity may rise, leading to an increase in the ability of cyclin D-cdk4 or cyclin E-cdk2 to phosphorylate pRb. The repression of c-myc and p21 by pRb in mid-G1 seems to establish a role for pRb in facilitating the sharp decline in the levels of the immediate-early gene products that occurs in mid-G1. Transient-transfection assays have implicated pRb in the control of c-fos as well (39), and it will be interesting to examine whether pRb is involved in the rapid decline in the levels of other immediate-early genes in mid-G1 as well. One of the genes that pRb failed to regulate was p16, an inhibitor of the cyclin D-cdk4 or cyclin D-cdk6 complex. Previous observations have suggested that pRb downregulates this gene, indirectly promoting its own phosphorylation. The protein is present at high levels in Rb-negative cell lines (35), and expression of a temperature-sensitive SV40 large-T antigen or human papillomavirus E7 in primary human fibroblast cells leads to an increase in the amount of p16 protein present (25). In our experiments, however, p16 mRNA levels were quite high in SAOS cells, which do not express pRb; this level did not alter significantly upon infection with pRb-expressing Ad. Unfortunately, we were unable to detect any p16 in MCF-10A cells, and so we could not confirm this result in these cells. Therefore, it is possible that pRb does play a role in p16 regulation in more normal cell lines but that this regulation requires additional factors that have been lost in SAOS cells. SAOS cells do not normally phosphorylate exogenously added pRb. It is possible that the high levels of p16, coupled with the inability of pRb to downregulate the transcription of this gene, contribute to the failure of SAOS cells to phosphorylate pRb. In addition to its repression of S-phase genes and immediate-early genes, pRb downregulated the transcription of E2F-1 and E2F-2 genes but not the genes of the remainder of the members of the E2F family. At present we do not know if E2F-2 is regulated by its own transcription factor or even through E2F sites, as is the case with E2F-1; however, E2F-2 binds pRb specifically in vivo (28). Therefore, it is possible that this defines a second autoregulatory loop in the E2F family, with E2F-2 regulating its own transcription. Since pRb is suspected to bind at least E2F-1, E2F-2, and E2F-3 in most cells (28), this would seem to define an additional method by which pRb can control the activity of E2F, by controlling the levels of certain E2F family members. The downregulation of E2F-2 by pRb may define an additional autoregulatory loop within the family; it is quite possible that the pRb/E2F-2 complex downregulates transcription from the E2F-2 promoter and that, conversely, E2F-2 upregulates its own transcription, and it will be interesting to see which member of the E2F family is involved in this regulation. Since E2F-4 and E2F-5 are probable binding partners for p130 and p107 (13, 18, 50), it will be interesting to see if the promoters of these proteins are influenced by these other members of the Rb family. Perhaps controlling the transcription of these transcription factors is a generalized secondary method for pRb family members to hold E2F activity in check during the early period of the cell cycle. Finally, pRb downregulated the expression of p107 in these experiments. The p107 promoter contains two tandem E2F


4575

sites, one of which was shown to respond to both pRb and p107 in previous transient-transfection assays (62). p107 was also one of the cell cycle-related genes shown to be derepressed in G1 in pRb2/2 fibroblasts (20). p107, like pRb, causes cell cycle arrest in G1 and represses transcription from E2F sites (45, 61). Although it is not known whether the genes regulated by pRb and by p107 are the same, there is clearly some functional overlap between the two proteins. It is likely, therefore, that the transcriptional repression of p107 represents a self-attenuation of the antiproliferation signals sent by pRb. Transienttransfection assays have suggested that pRb also represses the transcription of its own gene (15); this form of negative autoregulation may be necessary to ensure that pRb and p107 do not cause a permanent cell cycle arrest by rising to levels beyond which they can be effectively phosphorylated and deactivated by cyclin-cdk complexes. The genes found to be repressed by pRb overexpression in these assays represent only a small minority of genes involved in cell cycle progression and the processes of differentiation and tumorigenesis. It is likely that pRb is involved in the regulation of several other genes, at least some of which have not yet been isolated or extensively studied. Recently, several powerful techniques, including differential display (29) and the serial analysis of gene expression technique (51), have been developed to isolate genes which are differentially expressed between two sets of conditions. The system described in this paper may be used with such techniques to isolate other pRbcontrolled genes. Since pRb has been shown to be so crucial for cell cycle regulation and tumorigenesis, genes regulated by it may provide useful targets for future cancer therapies. ACKNOWLEDGMENTS We thank E. Harlow, D. Beach, R. Weinberg, L. Lau, and N. Bouck for cDNAs used in these experiments; J. Leiden and A. Ayer for AdDRb and Adb-gal; and S. Weitzman for the MCF-10A cells. We are indebted to P. Raychaudri for SAOS cells, for the E2F-1 clone, and for assistance in setting up the RNase protection experiments. We thank K. Rundell for assistance with the FACS technique and for helpful discussions. This work was funded by National Institutes of Health grant AI20156 (currently CA74403) to B.T., NIH carcinogenesis training grant T32CA09560 to A.B., and U.S. Army Medical Research and Materiel Command Grant DAMD17-94-J-4466 to A.B. REFERENCES 1. Almasan, A., Y. Yin, R. E. Kelly, E. Y.-H. P. Lee, A. Bradley, W. Li, J. R. Bertino, and G. M. Wahl. 1995. Deficiency of retinoblastoma protein leads to inappropriate S-phase entry, activation of E2F-responsive genes, and apoptosis. Proc. Natl. Acad. Sci. USA 92:5436–5440. 2. Chang, M. W., E. Barr, J. Setzer, Y.-Q. Jiang, G. J. Nabel, M. S. Parmacek, and J. M. Leiden. 1995. Cytostatic gene therapy for vascular proliferative disorders with a constitutively active form of the retinoblastoma gene product. Science 267:518–522. 3. Chellappan, S. P., S. Hiebert, M. Mudryj, J. M. Horowitz, and J. R. Nevins. 1991. The E2F transcription factor is a cellular target for the RB protein. Cell 65:1053–1061. 4. Cobrinik, D., P. Whyte, D. S. Peeper, T. Jacks, and R. A. Weinberg. 1993. Cell cycle-specific association of E2F with the p130 E1A-binding protein. Genes Dev. 7:2392–2404. 5. Cosenza, S. C., R. Carter, A. Pena, A. Donigan, M. Borrelli, D. R. Soprano, and K. J. Soprano. 1991. Growth-associated gene expression is not constant in cells traversing G-1 after exiting mitosis. J. Cell. Physiol. 147:231–241. 6. Dalton, S. 1992. Cell cycle regulation of the human cdc2 gene. EMBO J. 11: 1797–1804. 7. Datto, M. B., Y. Yu, and X.-F. Wang. 1995. Functional analysis of the transforming growth factor b responsive elements in the WAF1/Cip1/p21 promoter. J. Biol. Chem. 270:28623–28628. 8. De Caprio, J., Y. Furukawa, F. Ajchenbaum, J. D. Griffin, and D. M. Livingston. 1992. The retinoblastoma-susceptibility gene product becomes phosphorylated in multiple stages during cell cycle entry and progression. Proc. Natl. Acad. Sci. USA 89:1795–1798. 9. Dou, Q.-P., P. J. Markell, and A. B. Pardee. 1992. Thymidine kinase tran-

4576

10. 11. 12.

13.

14. 15. 16. 17. 18. 19. 20. 21. 22.

23. 24. 25.

26.

27. 28. 29. 30. 31.

32. 33. 34.

BUCHMANN ET AL.

scription is regulated at G1/S phase by a complex that contains retinoblastoma-like protein and a cdc2 kinase. Proc. Natl. Acad. Sci. USA 89:3256–3260. Dowdy, S. F., P. W. Hinds, K. Louie, S. I. Reed, A. Arnold, and R. A. Weinberg. 1993. Physical interaction of the retinoblastoma protein with human cyclins. Cell 73:499–511. Ewen, M. E., H. K. Sluss, C. J. Sherr, H. Matsushime, J. Y. Kato, and D. M. Livingston. 1993. Functional interactions of the retinoblastoma protein with mammalian D-type cyclins. Cell 73:487–497. Friend, S. H., M. R. Horowitz, X.-F. Wang, P. Bogenmann, F. P. Li, and R. A. Weinberg. 1987. Deletions of a DNA sequence in retinoblastomas and mesenchymal tumors: organization of the sequence and its encoded protein. Proc. Natl. Acad. Sci. USA 84:9059–9063. Ginsberg, D., G. Vairo, T. Chittenden, Z. X. Xiao, G. Xu, K. L. Wydner, J. A. DeCaprio, J. B. Lawrence, and D. M. Livingston. 1994. E2F-4, a new member of the E2F transcription factor family, interacts with p107. Genes Dev. 8: 2665–2679. Goodrich, D. W., N. P. Wang, Y.-W. Qian, E. Y.-H. Lee, and W.-H. Lee. 1991. The retinoblastoma gene product regulates progression through the G1 phase of the cell cycle. Cell 67:293–302. Hamel, P. A., R. M. Gill, R. A. Phillips, and B. L. Gallie. 1992. Transcriptional repression of the E2-containing promoters EIIaE, c-myc, and RB1 by the product of the RB1 gene. Mol. Cell. Biol. 12:3431–3438. Harper, J. W., G. R. Adami, N. Wei, K. Keyomarsi, and S. J. Elledge. 1993. The p21 Cdk-interacting protein Cip1 is a potent inhibitor of G1 cyclindependent kinases. Cell 75:805–816. Hiebert, S. W., S. P. Chellappan, J. M. Horowitz, and J. R. Nevins. 1992. The interaction of RB with E2F coincides with an inhibition of the transcriptional activity of E2F. Genes Dev. 6:177–185. Hijmans, E. M., P. M. Voorhoeve, R. L. Beijersbergen, L. J. van’t Veer, and R. Bernards. 1995. E2F-5, a new E2F family member that interacts with p130 in vivo. Mol. Cell. Biol. 15:3082–3089. Hinds, P. W., S. Mittnacht, V. Dulic, A. Arnold, S. I. Reed, and R. A. Weinberg. 1992. Regulation of retinoblastoma protein functions by ectopic expression of human cyclins. Cell 70:993–1006. Hurford, R. K., Jr., D. Cobrinik, M. H. Lee, and N. Dyson. 1997. pRB and p107/p130 are required for the regulated expression of different sets of E2F responsive genes. Genes Dev. 11:1447–1463. Ishida, S., K. Shudo, S. Takada, and K. Koike. 1995. A direct role of transcription factor E2F in c-myc gene expression during granulocytic and macrophage-like differentiation of HL60 cells. Cell Growth Differ. 6:229–237. Ivey-Hoyle, M., R. Conroy, H. E. Huber, P. J. Goodhart, A. Oliff, and D. C. Heimbrook. 1993. Cloning and characterization of E2F-2, a novel protein with the biochemical properties of transcription factor E2F. Mol. Cell. Biol. 13:7802–7812. Johnson, D. G., K. Ohtani, and J. R. Nevins. 1994. Autoregulatory control of E2F1 expression in response to positive and negative regulators of cell cycle progression. Genes Dev. 8:1514–1525. Johnson, D. G., J. K. Schwarz, W. D. Cress, and J. R. Nevins. 1993. Expression of transcription factor E2F1 induces quiescent cells to enter S phase. Nature (London) 365:349–352. Khleif, S. N., J. DeGregori, C. L. Yee, G. A. Otterson, F. J. Kaye, J. R. Nevins, and P. M. Howley. 1996. Inhibition of cyclin D-cdk4/cdk6 activity is associated with an E2F-mediated induction of cyclin kinase inhibitor activity. Proc. Natl. Acad. Sci. USA 93:4350–4354. Kim, S.-J., S. Wagner, F. Liu, M. A. O’Reilly, P. D. Robbins, and M. R. Green. 1992. Retinoblastoma gene product activates expression of the human TGF-b2 gene through transcription factor ATF-2. Nature (London) 358:331–334. Kurki, P., M. Vanderlaan, F. Dolbeare, J. Gray, and E. M. Tan. 1986. Expression of proliferating cell nuclear antigen (PCNA)/cyclin during the cell cycle. Exp. Cell Res. 166:209–219. Lees, J. A., M. Saito, M. Vidal, M. Valentine, T. Look, E. Harlow, N. Dyson, and K. Helin. 1993. The retinoblastoma protein binds to a family of E2F transcription factors. Mol. Cell. Biol. 13:7813–7825. Liang, P., and A. B. Pardee. 1992. Differential display of eukaryotic messenger RNA by means of the polymerase chain reaction. Science 257:967–971. Lukas, J., D. Parry, L. Aagaard, D. J. Mann, J. Bartkova, M. Strauss, G. Peters, and J. Bartek. 1995. Retinoblastoma-protein-dependent cell-cycle inhibition by the tumour suppressor p16. Nature (London) 375:503–506. Macleod, K. F., N. Sherry, G. Hannon, D. Beach, T. Tokino, K. Kinzler, B. Vogelstein, and T. Jacks. 1995. p53-dependent and independent expression of p21 during cell growth, differentiation, and DNA damage. Genes Dev. 9: 935–944. Michieli, P., M. Chedid, D. Lin, J. H. Pierce, W. E. Mercer, and D. Givol. 1994. Induction of WAF1/CIP1 by a p53-independent pathway. Cancer Res. 54:3391–3395. Nagarajan, M., and L. F. Johnson. 1989. Regulation of thymidylate synthase gene expression in mouse fibroblasts synchronized by mitotic selection. Exp. Cell Res. 181:289–297. Oswald, F., H. Lovec, T. Moroy, and M. Lipp. 1994. E2F-dependent regulation of human MYC: trans-activation by cyclins D1 and A overrides tumour suppressor protein functions. Oncogene 9:2029–2036.

MOL. CELL. BIOL. 35. Parry, D., S. Bates, D. J. Mann, and G. Peters. 1995. Lack of cyclin D-cdk complexes in Rb-negative cells correlates with high levels of p16INK4/MTS1 tumour suppressor gene product. EMBO J. 14:503–511. 35a.Pietenpol, J. A., K. Munger, P. M. Howley, R. W. Stein, and H. L. Moses. 1991. Factor-binding element in the human c-myc promoter involved in transcriptional regulation by transforming growth factor b1 and by the retinoblastoma gene product. Proc. Natl. Acad. Sci. USA 88:10227–10231. 36. Qin, X.-Q., T. Chittenden, D. M. Livingston, and W. G. Kaelin. 1992. Identification of a growth suppression domain within the retinoblastoma gene product. Genes Dev. 6:953–964. 37. Rajan, P., V. Dhamankar, K. Rundell, and B. Thimmapaya. 1991. Simian virus 40 small-t does not transactivate RNA polymerase II promoters in virus infections. J. Virol. 65:6553–6561. 38. Resnitzky, D., and S. I. Reed. 1995. Different roles for cyclins D1 and E in regulation of the G1-to-S transition. Mol. Cell. Biol. 15:3463–3469. 39. Robbins, P. D., J. M. Horowitz, and R. C. Mulligan. 1990. Negative regulation of human c-fos promoter by the retinoblastoma gene product. Nature (London) 346:668–671. 40. Rohde, M., P. Warthoe, T. Gjetting, J. Lukas, J. Bartek, and M. Strauss. 1996. The retinoblastoma protein modulates expression of genes coding for diverse classes of proteins including components of the extracellular matrix. Oncogene 12:2393–2401. 41. Ryseck, R. P., S. I. Hirai, M. Yaniv, and R. Bravo. 1988. Transcriptional activation of c-jun during the G0/G1 transition in mouse fibroblasts. Nature (London) 344:535–537. 42. Serrano, M., G. J. Hannon, and D. Beach. 1993. A new regulatory motif in cell-cycle control causing specific inhibition of cyclin D/CDK4. Nature (London) 366:704–707. 43. Shan, B., C. Y. Chang, D. Jones, and W. H. Lee. 1994. The transcription factor E2F-1 mediates the autoregulation of RB gene expression. Mol. Cell. Biol. 14:299–309. 44. Slansky, J. E., Y. Li, W. G. Kaelin, and P. J. Farnham. 1993. A protein synthesis-dependent increase in E2F1 mRNA correlates with growth regulation of the dihydrofolate reductase promoter. Mol. Cell. Biol. 13:1610–1618. 45. Starostik, P., K. N. B. Chow, and D. C. Dean. 1996. Transcriptional repression and growth suppression by the p107 pocket protein. Mol. Cell. Biol. 16: 3606–3614. 46. Stein, G. S., A. J. van Wijnen, J. Stein, J. B. Lian, and M. Montecino. 1995. Contributions of nuclear architecture to transcriptional control. Int. Rev. Cytol. 162A:251–278. 47. Strober, B. E., J. L. Dunaief, S. Guha, and S. P. Goff. 1996. Functional interactions between the hBRM/hBRG1 transcriptional activators and the pRB family of proteins. Mol. Cell. Biol. 16:1576–1583. 48. Swaminathan, S., and B. Thimmapaya. 1996. Transactivation of adenovirus E2-early promoter by E1A and E4 6/7 in the context of viral chromosome. J. Mol. Biol. 258:736–746. 49. Tait, L., H. D. Soule, and J. Russo. 1990. Ultrastructural and immunocytochemical characterization of an immortalized human breast epithelial cell line, MCF-10A. Cancer Res. 50:6087–6094. 50. Vairo, G., D. M. Livingston, and D. Ginsberg. 1995. Functional interaction between E2F-4 and p130: evidence for distinct mechanisms underlying growth suppression by different retinoblastoma protein family members. Genes Dev. 9:869–881. 51. Velculescu, V. E., L. Zhang, B. Vogelstein, and K. W. Kinzler. 1995. Serial analysis of gene expression. Science 270:484–487. 52. Weintraub, S. J., K. N. Chow, R. X. Luo, S. H. Zhang, S. He, and D. C. Dean. 1995. Mechanism of active transcriptional repression by the retinoblastoma protein. Nature (London) 375:812–815. 53. Welsh, P. J., and J. Y. J. Wang. 1993. A C-terminal protein-binding domain in the retinoblastoma protein regulates nuclear c-Abl tyrosine kinase in the cell cycle. Cell 75:779–790. 54. Wolffe, A. P. 1994. Architectural transcription factors. Science 264:1100–1101. 55. Wolffe, A. P., and D. Pruss. 1996. Targeting chromatin disruption: transcription regulators that acetylate histones. Cell 84:817–819. 56. Wu, C. L., L. R. Zukerberg, C. Ngwu, E. Harlow, and J. A. Lees. 1995. In vivo association of E2F and DP family proteins. Mol. Cell. Biol. 15:2536–2546. 57. Xiong, Y., G. J. Hannon, H. Zhang, D. Casso, R. Kobayashi, and D. Beach. 1993. p21 is a universal inhibitor of cyclin kinases. Nature (London) 366:701– 704. 58. Yu, D., A. Matin, and M.-C. Hung. 1992. The retinoblastoma gene product suppresses neu oncogene-induced transformation via transcriptional repression of neu. J. Biol. Chem. 267:10203–10206. 59. Zhang, H., G. J. Hannon, and D. Beach. 1994. p21-containing cyclin kinases exist in both active and inactive states. Genes Dev. 8:1750–1758. 60. Zhang, Y., and S. P. Chellappan. 1995. Cloning and characterization of human DP2, a novel dimerization partner of E2F. Oncogene 10:2085–2093. 61. Zhu, L., S. van den Heuvel, K. Helin, A. Fattaey, M. Ewen, D. Livingston, N. Dyson, and E. Harlow. 1993. Inhibition of cell proliferation by p107, a relative of the retinoblastoma protein. Genes Dev. 7:1111–1125. 62. Zhu, L., L. Zhu, E. Xie, and L.-S. Chang. 1995. Differential roles of two tandem E2F sites in repression of the human p107 promoter by retinoblastoma and p107 proteins. Mol. Cell. Biol. 15:3552–3562.