Analysis of the proliferating cell nuclear antigen promoter and its response to adenovirus early region 1.

The levels of the mRNA for the proliferating cell nuclear antigen (PCNA), a DNA replication factor, increase upon growth stimulation of quiescent cells. To study the transcriptional aspect of this response, we have cloned a PCNA gene fragment from size-fractionated human placental DNA. This fragment contains 1269 nucleotides upstream from the PCNA transcriptional start site and includes an Alu sequence that is transcribed in vitro. The PCNA genomic DNA promotes transcription of a linked heterologous reporter gene in HeLa and 293 cells. Transient expression assays and in vitro transcription analyses showed that 249 nucleotides of upstream sequence are sufficient for full promoter activity in HeLa cells, whereas only 172 nucleotides are needed in 293 cells. Co-transfection with a plasmid expressing the adenovirus E1 gene transactivates the PCNA promoter in HeLa cells. An E1-responsive element maps in the 85-nucleotide region immediately upstream of the site of transcription initiation.

The levels of the mRNA for the proliferating cell nuclear antigen (PCNA), a DNA replication factor, increase upon growth stimulation of quiescent cells. To study the transcriptional aspect of this response, we have cloned a PCNA gene fragment from size-fractionated human placental DNA. This fragment contains 1269 nucleotides upstream from the PCNA transcriptional start site and includes an Alu sequence that is transcribed in vitro. The PCNA genomic DNA promotes transcription of a linked heterologous reporter gene in HeLa and 293 cells. Transient expression assays and in vitro transcription analyses showed that 249 nucleotides of upstream sequence are sufficient for full promoter activity in HeLa cells, whereas only 172 nucleotides are needed in 293 cells. Co-transfection with a plasmid expressing the adenovirus E 1 gene transactivates the PCNA promoter in HeLa cells. An El-responsive element maps in the 85-nucleotide region immediately upstream of the site of transcription initiation.
Consistent with its role in DNA replication, expression of the proliferating cell nuclear antigen (PCNA),' also known as cyclin (Mathews et al., 1984) and the DNA polymerase-b auxiliary factor (Tan et al., 1986;Prelich et al., 1987a) is linked to cell growth, (Almendral et al., 1987;Matsumoto et al., 1987;Jaskulski et al., 1988;Shipman et al., 1988;Wold et al., 1988; see Mathews, 1989 for a recent review). PCNA functions in concert with DNA polymerase-6 to replicate the leading strand of an SV40 DNA template in vitro (Prelich et al., 1987b;Prelich and Stillman, 1988). The importance of its function to the replication machinery is highlighted by the high degree of evolutionary conservation of the protein. Homologs have been identified in plants (Suzuka et al., 1989), animals (Celis et al., 1987) including insects (Yamaguchi et al., 1990), yeast (Bauer and Burgers, 1990), and even a bacteriophage (Tsurimoto and Stillman, 1990).
Although PCNA protein and mRNA levels change relatively little during the cell cycle (Wold et ul., 1988;Liu et al., 1989;Morris and Mathews, 1989) upon growth stimulation of quiescent cells by such agents as serum, growth factors, and viral infection (Almendral et al., 1987;Matsumoto et al., 1987;Zerler et al., 1987;Jaskulski et al., 1988). Like other genes involved in DNA metabolism, the induction of PCNA mRNA levels by these agents requires protein synthesis and occurs after a delay of several hours (Almendral et al., 1987;Jaskulski et al., 1988). Genes such as thymidine kinase, dihydrofolate reductase, and thymidylate synthase are transcriptionally activated during this secondary response to growth stimuli (Johnson, 1984;Jenh et al., 1985;Ito and Conrad, 1990), but it is not known whether the PCNA gene is similarly activated at the transcriptional level. However, as little as 210 nucleotides of upstream sequence from the PCNA gene are sufficient to give substantial transcriptional activity in stably transformed cells (Ottavio et al., 1990). Adenovirus infection can cause cells to bypass normal controls of growth in culture and to produce tumors in animals . The El region of the virus, which encodes both ElA and ElB, is necessary and sufficient for this process . Viral infection activates many, if not all, of the genes activated during the secondary response to growth factors (Liu et al., 1985), including the PCNA gene (Zerler et al., 1987). The effects of ElB on transcription are probably indirect and possibly result from increasing ElA levels Yoshida et al., 1987;Jochemsen et al., 1987;Herrmann and Mathews, 1989). ElA, on the other hand, is well characterized as a transcriptional activator of gene expression (Flint and Shenk, 1989). Most of the positive transcriptional effects of the ElA protein map to conserved region 3, which is unique to the ElA 13 S transcript (Moran and Mathews, 1987), but a virus expressing only the 12 S ElA transcript, which lacks conserved region 3, can induce expression of the cellular genes for PCNA (Zerler et al., 1987) as well as hsp 70 (Simon et al., 1987), cdc 2 (Draetta et al., 1988), and brain creatine kinase (Kaddurah-Daouk et al., 1990). The ability to induce these cellular genes might be related to the transforming properties shared by the 13s and 12s ElA transcripts (Moran and Mathews, 1987).
To define the interactions of ElA with the PCNA promoter, we have isolated human PCNA promoter sequences. General features of the PCNA promoter were characterized by sequence comparison and in vitro transcription, and the effects of deletions on promoter activity were assayed by transfection of PCNA-CAT constructs into HeLa cells and 293 cells. Transactivation of the PCNA-CAT fusion constructs by the adenovirus El gene was investigated by co-transfection into HeLa cells, and an El responsive region in the PCNA promoter was identified.

Isolation of PCNA Gemonic
Clones-Methods were as described in Sambrook et al. (1989) unless otherwise indicated. Genomic DNA was isolated from human placenta with sodium dodecyl sulfate in place of sarkosyl. To enrich for PCNA promoter fragments, 200 pg of genomic DNA was digested in concert with EcoRI and PstI prior to electrophoresis in an agarose gel. DNA was recovered from horizontal gel slices by maceration, freeze-thawing, and phenol extraction. Selected gel-purified DNA fractions were ligated into EcoRI-and PstI-digested pUC118, and high density colony screening was performed with PCNA cDNA fragments labeled bv random miming. Cloned PCNA gene fragments-were sequenced on both strands by dideoxynucleotide chain termination reactions with a Sequenase Kit (Stratagene).

PCNA-CAT
Constructs-Deletions of the PCNA promoter were prepared by digestion with an appropriate restriction enzyme or by ExoIII digestion (Nested Deletion Kit, Pharmacia LKB Biotechnology Inc.)and cloned into the HindIII site or between the Xhol and Hind111 sites of nBACAT. The seauence at the unstream border of each insert, from the common X&I site, is as follows (with PCNA promoter sequences printed in upper case): In all cases except for the -560 to -2 clone, the downstream border was formed at the NruI site (+60) in the PCNA sequence with addition of a Hind111 linker. The upstream border of the -560 to -2 clone is identical with the -560 clone above and its downstream border is at -2 in the PCNA gene sequence. Downstream sequences are as follows: -560-2 GGCGGgcatgcaagctt all others CGTCGCaagctt RNA Analyses-Cytoplasmic RNA was isolated after lysis of cells with Nonidet P-40, and poly(A+) RNA was isolated by oligo(dT)cellulose chromatography. Sl nuclease mapping was performed with a 467-nucleotide fragment (-246 to +221) prepared by digesting the 1.5-kb PCNA genomic clone with DraII and 5'-end-labeling with [y-"*P]ATP and polynucleotide kinase. Primer extension reactions were conducted with 3 pg of poly(A+) RNA and 0.1 pmol of end-labeled primer. The 51-nucleotide DNA primer (+221 to +170) was prepared by digesting the 5'-end-labeled &a11 fragment with BssHII followed by purification on a polyacrylamide gel. For sequencing, dideoxynucleotides were included in the primer extension reaction at a ratio of 1:l with deoxynucleotides (100 PM each). RNase protection experiments were performed with DNase I-treated cvtonlasmic RNA from transfected cells. The PCNA-CAT riboprobe (425 nucleotides) was synthesized by SP6 polymerase transcription of an EcoR1/P~~u11 fragment from the -172 PCNA-CAT construct cloned into EcoRI/ HincII-digested pGEM-1. The riboprobes were purified by gel electrophoresis. Controls verified that the probe was in excess and that the protection was dependent on RNA. As a control for transfection efficiency and RNA recovery, the same RNA preparations were assayed simultaneously or in a parallel set of reactions for transcriuts from a co-transfected-U2-globin clone with a globin riboprobe piepared from pMRA.2 3'U6 (Lob0 and Hernandez. 1989). Primer extension products and Sl and RNase protection products were analyzed in 8% polyacrylamide, 7 M urea gels, which were fixed, dried, and exposed to x-ray film.

Transcription in Nuclear
Extracts-Nuclear extract preparation from HeLa and 293 cells and in vitro transcription reactions with [a-"P]UTP were essentially as described by Price et al. (1987), except that incubations were carried out at 30 "C. Optimal concentrations of KC1 (60-80 mM), extract 3.5-4 mg/ml), and template DNA (15-25 pg/ml) were determined experimentally. To detect RNA polymerase III transcripts, DNA concentrations higher than 30 pg/ml were used.
Temnlate DNAs were digested with EcoRI: a radiolabeled product of 303 nucleotides was expected for transcripts initiating at-+l. RNA was recovered by phenol extraction and ethanol precipitation and was analyzed by electrophoresis in 8% polyacrylamide, 7 M urea gels. RNA recoveries were monitored by ethidium bromide staining of the gel to detect endogenous nuclear RNAs in the extract.
Transfections and CAT Assays-Transfection experiments were performed by the calcium phosphate precipitation technique  except that the medium was not changed prior to addition of the precipitate. 5 rg of plasmid was employed for each transfected plasmid, and salmon sperm DNA was added to bring the total to 20 pg of DNA per 6-cm plate. CAT activity was assayed as previously described  and quantified using an AMBIS Beta Scanner. were grown in Dulbecco's modified Eagle's medium plus 10% fetal calf serum and 100 rg/ml penicillin and streptomycin. 293 cell monolayers were grown in the same medium with 10% calf serum instead of 10% fetal calf serum.

RESULTS
Cloning of the PCNA Promoter-As a first step toward cloning the PCNA promoter, we analyzed total genomic DNA from human placenta by digestion with the restriction endonuclease EcoRI. The resulting fragments were resolved and probed by hybridization to the entire PCNA cDNA clone or to a fragment containing the 5' quarter of the cDNA (probe 1, Fig. 1B). Three genomic fragments of about 4.14, 2.75, and 1.56 kb hybridized to the complete cDNA ( . Consistent with the cDNA sequence, the 2.75-kb fragment was cleaved by PstI, producing two smaller fragments of about 1.5 kb and 1.0 kb that hybridized to a fragment representing the 5'-half of the cDNA (probe 2, Fig. 1B; data not shown). To clone these fragments, genomic DNA was digested with EcoRI and PstI on a preparative scale and was fractionated by agarose gel electrophoresis (Fig. 1C). The fractions containing the desired PCNA gene fragments were identified by hybridization to probe 2 ( Fig. 10). Enriched plasmid libraries were prepared from these two fractions and colonies corresponding to the l.O-kb fragment and the 1.5-kb fragment were selected by hybridization with probe 2. Sequence of the PCNA Promoter-DNA sequence analysis proved that the 1.5-kb fragment contains the 5' end of the cDNA and upstream sequences, and that it is contiguous with the l.O-kb clone which contains adjacent downstream cDNA sequences as depicted in Fig. 1B. The upstream 1.5-kb fragment is uninterrupted, whereas the downstream l.O-kb fragment is interrupted twice by introns of approximately 0.7 kb and 0.1 kb. The sequence of the larger fragment is shown in Fig. 2. While this work was in progress, Travali et al. (1989) published the sequence of a human genomic clone comprising the entire PCNA coding sequence. Relative to their sequence, our sequence contains 6 insertions, 3 deletions, and 15 substitutions. Four of the changes have been confirmed by restriction analysis of our clone (marked in Fig. 2). Several of the changes are clustered and presumably represent polymorphisms in the PCNA gene, but others are isolated differences. The upstream region of the PCNA gene displays numerous homologies with consensus sequences for RNA polymerase II transcription factor binding sites. Some of these are summarized in Table I to facilitate consideration of their potential role in the expression of the PCNA gene in light of the data presented below.
An Upstream Alu Sequence-Hybridization of the 1.5-kb genomic clone to total genomic DNA revealed the presence of a repetitive DNA element (data not shown). In uitro assays at high template concentrations, which favor transcription by RNA polymerase III, gave evidence of RNA polymerase III transcription of the upstream sequence. Transcription of a PCNA construct containing 1265 nucleotides of upstream sequence produced several RNA species in the 300-500 nucleotide range (Fig. 3, lane 2). Their synthesis was resistant to a low concentration of cu-amanitin (Fig. 3, lane I), consistent with RNA polymerase III transcription, and they were not transcribed from truncated templates containing 560 nucleotides or less of upstream sequence (Fig. 3, lane 3). These data place a RNA polymerase III transcription unit between -1265 and -560. A homology search of the entire 1.5-kb clone with a member of the human Ah family, located upstream of the c-globin gene (DiSegni et al., 1981), revealed an Ah family member between -1164 and -844 in the PCNA promoter (underlined in Fig. 2). The region of homology includes consensus sequences for RNA polymerase III promoter elements (DiSegni et al., 1981), and, by analogy to its homolog in the e-globin promoter, the Ah sequence in the PCNA promoter should be transcribed in the same direction as the PCNA gene itself. The c-globin Alu sequence is transcribed in uiuo (Allan and Paul, 1984) as well as in vitro (DiSegni et al., 1981) and is co-regulated with the c-globin promoter (Wu et al., 1990).  Beato. 1989 It is not yet known if the Alu sequence functions in PCNA gene expression. The Transcription Initiation Site-Preliminary attempts to map the site of transcription initiation by the Sl nuclease technique suggested a location about 40 nucleotides upstream from the 5' end of the nearly complete human cDNA clone of Almendral et al. (1987). In the experiment shown in Fig.  4B, a broad DNA band of approximately 217 nucleotides was protected efficiently by poly (A+) or total cellular RNA (lanes 2 and 3), but not by poly(A-)RNA (lane I). The same endlabeled DNA probe was digested with a second enzyme to produce a 51-nucleotide fragment for 5' end mapping by the primer extension method (Fig. 4-4). Dideoxy chain termination reactions (lanes A, C,G, 7') were carried out in parallel with the unblocked primer extension reaction (lane P) and gave a partial sequence that matched the cDNA sequence of Almendral et al. (1987). This verified that the primer was specific for PCNA mRNA. The unblocked primer extension product was about 4 nucleotides longer than the fragment protected against nuclease Sl, probably because of overdigestion by the nuclease in this AT-rich region. The position deduced for the 5' end of the PCNA mRNA, indicated in Fig. 2 by conversion to lower case, lies 160 nucleotides upstream of the translation initiation codon. Similar results have been obtained by Travali et al. (1989), who located the site of transcription initiation by primer extension to the same position as the 5' terminus of a full length PCNA cDNA clone (Jaskulski et al., 1988). Initiation at the A residue designated here as +1 is consistent with the transcriptional start site being fixed by a surrounding initiator element (see Table I and below). Transient Expression in HeLa Cells-To discover whether the 1.5-kb clone contains a functional promoter, the region from -1265 to +60 was fused to a reporter gene (chloramphenicol acetyltransferase, CAT) and transfected into HeLa cells. The upstream sequence from the PCNA gene promoted the synthesis of CAT (Fig. 6), albeit 20-40 times less effectively than the SV40 early promoter (data not shown). To identify regions of the PCNA promoter important for its function, we produced a series of upstream deletion mutants of the PCNA promoter, all fused at +60 to the CAT reporter  Table I). Removal of sequences from -172 to -87 reduced the activity of the promoter further, to levels only slightly higher than the basal levels observed with the parent clone lacking inserted PCNA promoter sequences, pBACAT. The region from -172 to -87 contains two CCAAT boxes and two Spl sites (Table I). The largest upstream deletion tested, a -46 to +60 CAT fusion construct, removed a site homologous to the ATF consensus sequence (Table I). CAT activity generated by the -46 to +60 construct did not routinely exceed that from the pBACAT control, except in the most sensitive experiments conducted with more extract and longer assay incubation times (Fig. 6B). A summary of the relative CAT activities for each of the constructs, averaged over several experiments, is shown in Fig. 5. We also performed RNase protection assays to examine the PCNA-CAT transcripts generated in transfected HeLa cells. RNA from the transient expression of each upstream deletion construct gave rise to a protection product of the correct size for a transcript that initiated at +l on the PCNA genomic sequence (Fig. 6C). Control assays, with tRNA or RNA from untransfected cells or cells transfected with pBACAT, did not produce a protection product that mapped to this site. For these constructs, the levels of correctly initiated transcripts correlated well with the amount of CAT activity detected (compare Fig. 6, A and C), and a simultaneous RNase protection assay for a globin clone co-transfected with each deletion produced similar signals in each case. Longer PCNA-CAT protection products were observed in each lane (Fig. 6C A, primer extension reactions. Reactions were performed with a 51-nucleotide end-labeled primer in the absence (P) or presence of dideoxynucleotides (A, C, G, and 7'). Sizes of DNA markers are shown at left. Lane P was exposed with an intensifying screen for 13 h. Lanes A, C, G, and T were exposed for 1 week with an intensifying screen. B, Sl nuclease protection. An end-labeled DNA probe (467 nucleotides) was hybridized with poly(A-) (lane I), poly(A+) (lane 2), or total cellular RNA (lane 3) and digested with Sl nuclease. The products were fractionated with end-labeled DNA markers (shown at right). The probe primer, mRNA, and primer extension product are illustrated diagrammatically.
1-4) and probably correspond to transcripts from cryptic promoters upstream of the inserted PCNA genomic fragment. These read-through transcripts were more prominent than the PCNA-CAT mRNA protection product with the -87 and -46 deletions. Read-through transcription might lower the amount of correctly initiated PCNA-CAT mRNA by promoter occlusion (Adhya and Gottesman, 1982;Vales and Darnell, 1989;Wu et al., 1990); perhaps the inverted CCAAT motif at -145 in the PCNA promoter serves to reduce the amount of read-through transcription from upstream promoter sequences, as is the case in the adenovirus major late promoter (Connelly and Manley, 1989).
Possible Initiator Element-Some promoters that lack TATA elements possess a sequence encompassing the transcription initiation site, termed the initiator sequence @male and Means and Farnham, 1990), that serves to designate the site of transcription initiation. The PCNA promoter lacks a suitably placed TATA box and displays some homology near its cap site to the published initiator sequences (Table I). To assess the role of sequences near the cap site in the function of the PCNA promoter, sequences from -560 to -2 nucleotides upstream of the cap site were fused to the CAT reporter gene (Fig. 5, bottom line). This cap site deletion construct gave rise to 30% less CAT activity in HeLa cells than the -560 to +60 construct (Figs. 5 and 6A), suggesting that sequences between -1 and +60 play some role in gene expression. On the other hand, no decrease in CAT Cloned PCNA genomic fragments were fused to a CAT reporter gene in pBACAT as detailed under "Materials and Methods." Each construct is shown diagrammatically, except the -397 to +60 construct which has been assayed in only a limited number of experiments. The CAT activities (relative to the activity of the -1265 construct) from transient expression assays in HeLa and 293 cells were averaged over 5-10 or 3-5 experiments, respectively, and are shown with standard deviations. CAT CODING REGION r-j mRNA was evident when the major RNase protection product (band 5) arising from the cap site deletion construct was compared with the product derived from the longer clone (Fig.  6C). However, the RNase protection analysis is likely to overestimate the production of mRNA by the -560 to -2 clone; since the sequence of this construct diverges from that of the probe at the 5' end of the CAT gene, any transcripts initiating upstream of this point, including the read-through transcripts referred to above, would be scored in band 5. It is probable that some of the upstream transcription initiations scored in this assay do not give rise to a functional CAT mRNA, thereby accounting for the apparent discrepancy between the CAT assay and the RNase protection experiment. To obtain more direct transcriptional information, each of the CAT constructs described above was tested for its template activity in a HeLa nuclear extract. The predicted (Yamanitin-sensitive run-off product was observed with all of the templates carrying upstream deletions (Fig. 6D). The yields of run-off product observed approximated the activity of the promoters in the transient expression assays described above. The -560 to -2 clone did not produce a prominent amanitin-sensitive run-off product, but a greater amount of amanitin-sensitive transcription was observed within the region 50 to 100 nucleotides longer than the predicted run-off transcript (bracketed in Fig. 6D). Presumably, these arise from heterogeneous upstream starts on the -560 to -2 template, reflecting deletion of an element (such as an initiator element) involved in determining the site of transcription initiation. Alternatively, the deletion of sequences including the cap site might simply produce a more deleterious effect on transcription in vitro than in vivo.
Transient Expression in 293 Cells-Zerler et al. (1987) demonstrated that PCNA gene expression is induced by the adenovirus ElA gene. To characterize the effects of ElA on the PCNA promoter, we first assayed the function of the PCNA promoter in 293 cells, which express both the adenovirus ElA and ElB gene products. Fig. 7 recapitulates in 293 cells the experiments illustrated in Fig. 6 for HeLa cells. The site of transcription initiation was identical in the two cell types (Figs. 6C and 7C), and many of the results obtained in the two cell types were similar, so only the differences will be mentioned here. The PCNA promoter was more active in 293 cells than in HeLa cells; thus, in these cells, CAT expression from the -1265 clone approached that of the SV40 promoter (data not shown). In HeLa cells, the -172 construct exhibited less promoter strength than the longer constructs, but its activity in 293 cells was undiminished both in transient expression assays and in nuclear extracts (Figs. 5 and 7,A and D). The deletion of sequences at the cap site (in the -560 to -2 construct) had a much more deleterious effect on CAT activity in 293 cells than in HeLa cells (Figs. 5, 6A, and 7A), suggesting that the region from -1 to +60 is responsive to transactivation by El. As with HeLa cells, the cap site deletion clone produced disparate RNase protection and CAT assay results (Fig. 7, A and C). As noted above, the discrepancy is probably due to the detection of transcripts that do not produce a CAT protein in the RNase protection assay. In 293 cell nuclear extracts, the -560 to -2 template also did not produce a prominent run-off product consistent with transcription initiation immediately following the promoter sequences (Fig. 70). These observations support the idea that the region from -1 to +60 determines the site of transcription initiation.
Transactivation by the Adenovirus El Region-The elevated expression of PCNA-CAT constructs in 293 cells compared to HeLa cells suggested that the PCNA promoter is transactivated by the El region of adenovirus.
In an attempt to demonstrate this directly, we co-transfected HeLa cells with PCNA-CAT constructs and an ElA expression plasmid, but did not consistently observe the expected transactivating effect. However, the simultaneous expression of both transforming genes of adenovirus (ElA and ElB, from plasmid pE1) reproducibly elicited a substantial transactivation of the PCNA-CAT construct; about 6-fold in the experiment of Fig.  8. To define the region of the PCNA promoter that is responsive to El, each of the deletion mutants illustrated in Fig. 5 was tested by co-transfection with pE1 into HeLa cells. pE1 increased the expression of all the promoter fusion constructs except the -46 to +60 construct and the pBACAT control (Fig. 8). In a limited number of experiments, a slight response was seen with the -46 clone, but the effect never approached the magnitude of the response observed with the other clones. Consistent with results obtained in 293 cells, the -560 to -2 clone was usually transactivated by El about half as well as the control (-560 to +60), possibly indicating a role for the initiator element in the El response. It is clear, however, that at least one El-responsive element lies downstream of -87 since the -87 to +60 construct was transactivated by El (Fig.  8). Because the -560 to -2 clone was also El-responsive, at least one El-responsive element lies upstream of -2. These show that an El-responsive element exists between -87 and -2, but do not preclude the possibility that the PCNA promoter contains more than one such element. DISCUSSION Since PCNA synthesis correlates closely with cell growth, further understanding of its expression might provide a basis for a better appreciation of the cellular mechanisms involved in controlling cell growth. We have cloned the PCNA promoter so that factors regulating its activity might be studied in more detail; here we have examined the response of the PCNA promoter to the transforming genes of adenovirus.
Transcription Factor Binding Sites for Basal Transcription-potential transcription factor binding sites in the PCNA promoter are highlighted in Table I. Since sequences upstream of -249 had no effect on PCNA promoter activity in HeLa or 293 cells, sites further upstream would appear to be inconsequential although they may be important for PCNA expression in other cell types or during different cellular responses. For example, there is good homology to an octamer motif in the PCNA promoter at -350 (8/8) and -400 (7/8) and this motif is known to respond to the herpes viral transactivator VP-16 in HeLa cells (Stern et al., 1989) and have both positive (Scheidereit et al., 1987) and negative (Leonardo et al., 1989) effects depending on cell type. In contrast to results obtained by Travali et al. (1989) with baby hamster kidney cells transformed with the human PCNA gene, we did not observe a negative effect on PCNA promoter activity of sequences between -560 and -397. Parenthetically, a negative effect of the fourth intron of the PCNA gene on its expression in the absence of serum has been reported (Ottavio et al., 1990), but is not addressed here.
The two homologies with the CCAAT element at -95 and -145 may be functionally relevant, since removal of the region encompassing these sites severely impairs expression from the PCNA promoter in both 293 cells and HeLa cells (Fig. 5). Promoters for other secondary response genes required for DNA metabolism possess CCAAT boxes; for example, a CCAAT motif is involved in the serum response of the human thymidine kinase promoter (Lipson et al., 1989) which possesses two functionally important CCAAT homologies (Arcot et al., 1989) and in the S phase specific transcription of the human histone Hl gene (LaBella et al., 1989) and a rat H2b globin clone. One-fifth of a freeze-thaw extract from the transfection of a 6-cm plate was incubated with ["Clchloramphenicol for 20 min. The acetylated ['*C]chloramphenicol was separated by ascending thin layer chromatography. Nontransfected cell extracts were assayed in parallel (-). B, CAT activity of the weaker promoters; as in A, except that one-half of each extract was assayed for 60 min. C, RNase protection assay for CAT mRNA. Cytoplasmic RNA was isolated from two additional 6-cm plates from the transfection experiment shown in A. One-half of the RNA sample was assayed by RNase protection with PCNA-CAT and globin riboprobes. As negative controls, equal amounts of RNA from untransfected cells (-) or calf liver tRNA (tRNA) were assayed in parallel. Bands l-4 are readthrough transcripts while band 5 is of the size expected for PCNA-CAT transcripts of the -560 to -2 construct. To ensure that the probe was in excess, an equal amount of RNA from the -1265 transfection was assayed with twice as much probe (2 x probe). Note that the RNA mobilities of the RNA are consistently slower than that of the DNA markers, but a plot of the log of the predicted molecular weight for bands 1-5 and PCNA-CAT mRNA uersus distance migrated defines a line that parallels a similar plot of the DNA markers. D, transcription of the mutant constructs in a HeLa cell nuclear extract in the presence (+) and absence (-) of 2 rg/ml (Yamanitin. The bracket indicates presumptive heterogeneous start sites on the -560 to -2 template. Little incorporation of [w~*P]UTP was observed if no template was added (-DNA). End-labeled DNA marker sizes are shown at left.  Fig. 6A, except using one-tenth of the extract per assay. B, CAT activity of the weaker promoters in 293 cells. One-half of the extract from a different transfection experiment, performed in duplicate, was assayed as in Fig. 6B. C, RNase protection assay for CAT mRNA: as in Fig. 6C. D, transcription of the mutant constructs in 293 cell nuclear extracts: as in Fig. 6D except that only the -560 to -2 construct was incubated with Lu-amanitin (2 rg/ml).
-a* c t -3 't tttt??????: FIG. 8. Stimulation of PCNA-CAT expression by El. PCNA-CAT constructs were co-transfected into HeLa cells with (+) or without (-) pE1. One-half of the extract from each plate was incubated for 60 min with ["C-Jchloramphenicol to assay CAT activity. Two other constructs, E3CAT and SVSCAT, were assayed with and without El in parallel. The lane designated CAT is an assay with bacterial CAT enzyme.
gene (Hwang et al., 1990). The mouse DHFR gene possesses two CCAAT motifs downstream of the cap site that are important for DHFR transcription in vitro  although their role in uiuo is not known. The CCAAT motif and the family of proteins that bind this element have been implicated in developmental control of gene expression, as a serum response element, in the regulation of energy metabolism, and in S phase specific transcription (McKnight et al., 1989), and it will be important to address their participation in transcription of the PCNA promoter.
There are four potential binding sites for Spl in the PCNA promoter. Removal of the distal Spl site has little effect in either HeLa or 293 cells, but removal of the three other sites at -190, -165, and -125 might be involved in the decline of PCNA-CAT expression observed with the larger deletions. Repeated Spl sites are a recurrent theme of late response genes including the genes for DHFR (Dynan et al., 1986), thymidine kinase (Arcot et al., 1989), and thymidylate synthase (Deng et al., 1986). The PCNA promoter of Drosophila, which paradoxically may not produce Spl (Courey et al., 1989), also possesses homology with Spl sites (Suzuka et al., 1989). Presumably the synergistic properties of Spl (Courey et al., 1989) and the repeated arrangement of Spl sites in late response genes would produce a large enhancement of transcription during periods of rapid growth. The recent observation that a negative regulator of transcription can bind Splrelated sequences (Kageyama and Pastan, 1989) could add another layer of complexity to the PCNA promoter.

El-responsive
Elements in the PCNA Promoter-The differences that we observe in relative CAT activity between 293 cells and HeLa cells for the various deletion constructs could be relevant to the interaction of El with the PCNA promoter. From the HeLa 293 cell comparison one might predict that an El-responsive element lies downstream of -249, since the -249 to +60 construct is fully active in both cell types. CO-transfection experiments in HeLa cells indicate that El responsivity resides in the -87 to +60 region. Removal of sequences between -1 and +60 reduces the El response slightly, but does not abrogate it, suggesting that the main El response element lies between -87 and -2.
Since ElB is not known to function at the transcriptional level, and ElA does, we will consider ElA-responsive elements in the -87 to -2 region. A likely candidate is the ATF homology at -50 (Table I). Each of the adenovirus early promoters except ElB contains at least one ElA-responsive ATF site. The ATF site can bind a large family of proteins that may include the immunologically related AP-1 family (Hai et al., 1988). Taylor and Kingston (1990) found that an ATF site requires specific core promoter elements (TATA motifs) for activity, but an ATF site (12/13 homology to the PCNA ATF site, Table I) in the DNA polymerase /3 promoter functions in the absence of an identifiable TATA element (Widen et al., 1988). The DNA polymerase /3 promoter is also transactivated by co-transfection of El (Widen et al., 1988). Although a &-acting El-responsive element has not been identified, mutation of the ATF palindrome reduces DNA polymerase /3 promoter activity in 293 cells by 70% (Widen et al., 1988).
Another potential ElA-responsive target in this region is the sequence TTGCGAC at -45 (Table I) which is the upstream half of an imperfect inverted repeat with the second half 25 nucleotides downstream. This sequence is part of a direct repeat that confers El responsiveness on the adenoassociated virus P5 promoter (Chang et al., 1989). This inverted repeat arrangement is also reminiscent of the ElAresponsive E2F sites of the adenovirus E2A gene. E2F can bind to adjacent sites in the promoters of the human myc and hamster DHFR genes and may be involved in the response of these promoters to serum (Blake and Azizkhan, 1989;Hiebert et al., 1989). No additional ElA-responsive elements can be identified by sequence comparisons of the -87 to -2 region.
This report maps sequences in the PCNA promoter required for basal activity and localizes a region that responds to cotransfection of El and less well to ElA. It is probable that the ElB effects are mediated through increasing ElA levels, but a cooperative interaction of ElA and ElB cannot be discounted. We are presently examining the effects of each of the transcripts of the El region individually and in combination on the activity of the PCNA promoter.