INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

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 cycle-related 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 E2F-controlled genes. Indeed, protein levels of the TK, thymidylate synthase, and cdc-2 genes increase after this point, suggesting 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-2 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 5' 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 pRb^/ cells. One experiment, in which Rb was expressed under the 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. pRb^/ 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  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 cycle-related 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

Top Abstract Introduction Materials & Methods Results Discussion References

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 µg of insulin per ml, 100 ng of cholera toxin per ml, 0.5 µg 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 5' end of the Rb gene, and the plasmid contained the simian virus 40 (SV40) polyadenylation signal at the 3' 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. AdRb 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; p34 in reference 15) under the control of the EF1 alpha promoter with an HA epitope tag at the 3' end of the gene. This Rb expression cassette was cloned into the E1A/E1B region of Ad5. Ad-gal was a gift of Aviand Ayer (Northwestern University) and contains the -galactosidase (-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 1× 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 · 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 impurities, and the protein concentration was determined by the Bradford method (Bio-Rad). A 50-µg 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 µg 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, AdRb, or Ad-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.

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 µg 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 (Becton-Dickinson) 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.

Nuclear run-on assays. MCF-10A cells were synchronized and infected with AdRb or Ad-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 · 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 · 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 · Cl [pH 8.0], 5 mM MgCl₂, 0.1 mM EDTA) at a concentration of 10⁸ cells per ml.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

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 full-length 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, AdRb, was obtained from Jeffrey Leiden. AdRb contains a nonphosphorylatable mouse Rb cDNA (p34 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 Ad-gal, which contains the -gal gene under the control of the CMV promoter cloned into the E1A/E1B region of Ad5 (Fig. 1).

View larger version (20K): [in this window] [in a new window]   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. AdRb 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. Ad-gal, used as a control in these experiments, contains the -gal cDNA under the control of the CMV promoter inserted into the E1A/E1B region of Ad.

View larger version (60K): [in this window] [in a new window]   FIG. 2.   Western blot analysis of pRb produced upon AdRb or AdRb 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 AdRb. pRb is seen as a single band at 24 h after infection in both AdRb- and AdRb-infected cells in Western blots probed with an anti-HA antibody.

View larger version (12K): [in this window] [in a new window]   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.

View larger version (23K): [in this window] [in a new window]   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 Ad-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 AdRb-infected cells remain arrested in G1, while Ad-gal- and mock-infected cells exit G1 and progress into S and G2.

AdRb- and AdRb-infected cells progress through the early stages of G₁. Although FACS analysis showed nearly identical profiles for AdRb-, AdRb-, and Ad-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 AdRb-infected cells were arrested in G₀ or very early in G₁ while Ad-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 immediate-early 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 µg of protein was run on SDS-polyacrylamide gels. The gels were then probed with the anti-jun antibody (sc-44X; Santa Cruz Biotechnology). As shown in Fig. 5, AdRb-, AdRb-, and Ad-gal-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-, AdRb-, and Ad-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 AdRb-infected cells arrest before the time point at which we harvested the cells while Ad-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 MCF-10A cells were harvested at least 3 to 4 h before this point.

View larger version (60K): [in this window] [in a new window]   FIG. 5.   Western analysis of c-jun expression in AdRb-, AdRb-, and Ad-gal-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 (+15 min) (lanes 4 to 6) and peaks at about 30 min after serum stimulation (+30 min) (lanes 7 to 9). The protein level then declines to near basal levels by 1 h after serum stimulation (+60 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.

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-, AdRb-, and Ad-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-µg 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.

View larger version (50K): [in this window] [in a new window]   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 AdRb-infected cells, lane 4 contains the protected RNA from Ad-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.

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.

View larger version (68K): [in this window] [in a new window]   FIG. 7.   Riboprobe analysis of mRNAs encoding E2F family members. Lane numbers and arrows are as described in the legend to Fig. 5.

View larger version (52K): [in this window] [in a new window]   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-3-phosphate dehydrogenase.

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.

View larger version (46K): [in this window] [in a new window]   FIG. 9.   (A) Graph of the changes in mRNA levels in SAOS cells infected with AdRb or AdRb in comparison with cells infected with Ad-gal, as measured by RNase protection assays. The mRNA level in Ad-gal-infected cells was taken as 1 unit. (B) Variations in the rate of transcription in MCF-10A cells infected with AdRb in comparison with those infected with Ad-gal, as measured by nuclear run-on assays. The transcription rate in Ad-gal cells was taken as 1 unit.

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 AdRb and Ad-gal in MCF-10A cells synchronized by serum starvation and harvested 12 h after serum stimulation, as described above.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

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-, 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 cycle-related genes within a chromosomal context and within the context of the G₁ phase of the cell cycle. While transient-transfection assays have indicated that pRb plays a role in the 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 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, TAF_II 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 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 pRb-regulated 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 pRb^/ 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 short-term effects of proteins on cellular transcription and cell cycle-related 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 pRb^/ 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 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 p130^/ or p107^/ 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 pRb^/ 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 pRb^/ 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 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 MCF-10A 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 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 pRb^/ 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. Transient-transfection 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 pRb-controlled 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.