The influence of the proliferating cell nuclear antigen-interacting domain of p21(CIP1) on DNA synthesis catalyzed by the human and Saccharomyces cerevisiae polymerase delta holoenzymes.

In eukaryotes, processive DNA synthesis catalyzed by DNA polymerases δ and ε (pol δ and ε) requires the proliferating cell nuclear antigen (PCNA). It has recently been shown that in humans (h), the PCNA function, required for both DNA replication and nucleotide excision repair, can be inactivated by p21CIP1 due to a specific interaction between hPCNA and the carboxyl terminus of p21CIP1. In this report, we show that Saccharomyces cerevisiae (S. cerevisiae) PCNA-dependent pol δ-catalyzed DNA synthesis was inhibited less efficiently than the human system by the intact p21CIP1 protein and was unaffected by the p21CIP1 carboxyl-terminal peptide (codons 139-160). This species-specific response of PCNA to p21CIP1-mediated inhibition of DNA synthesis results from a marked difference in the ability of h and S. cerevisiae PCNA to interact with p21CIP1. As shown by binding studies using the surface plasmon resonance technique, hPCNA binds both full-length p21CIP1 and the p21CIP1 peptide-(139-160) stoichiometrically with a similar affinity (KD∼ 2.5 nM) while S. cerevisiae PCNA binds p21CIP1 with ∼10-fold less affinity and does not interact with the p21CIP1 peptide-(139-160).

The DNA replication machinery is functionally conserved from bacteria to mammals. The mechanism underlying DNA synthesis of primed templates in T4 bacteriophage (1), Escherichia coli (2), Saccharomyces cerevisiae (3), and humans (4,5) depends on a protein complex (the "clamp-loader") that binds to a primer-template junction. In each case, the multisubunit clamp loader complex loads a toroidal shaped protein (the "clamp") (6 -8) onto a primer end in an ATP-dependent reaction. The DNA polymerases that catalyze deoxynucleotide incorporation are tethered to DNA by a direct interaction with the clamp. This interaction, which occurs independently of the clamp loader, converts DNA polymerases from dispersive into highly processive enzymes capable of catalyzing extensive DNA synthesis (9 -12). Furthermore, chromosomal replicases of E. coli (pol 1 III core, ␥ complex, and ␤), phage T4 (T4gp43, T4gp44⅐62 complex, and T4gp45) and eukaryotes (pol ␦ or ⑀, RF-C, and PCNA) not only possess functional similarities but, in many cases, show marked sequence conservation (13,14).
In eukaryotes, insight into chromosomal DNA synthesis has been derived from in vitro studies of simian virus 40 replication. With the exception of one viral encoded protein, the SV40 T antigen, this replication pathway is dependent on the cellular replication machinery (15)(16)(17). The essential host proteins have been isolated, and in most cases their detailed mechanisms have been elucidated (9,10,18,19). The DNA polymerase ␣⅐DNA primase complex is responsible for the initiation of replication. DNA primase generates oligoribonucleotide primers that are elongated by pol ␣ for about 30 nucleotides resulting in the accumulation of pre-Okazaki fragments. In the absence of proteins that bind to 3Ј-hydroxyl primer ends, pol ␣ is capable of rebinding and elongating chains to mature Okazaki fragments on the lagging strand and generating DNA chains that are approximately one-half the length of the circular duplex DNA template on the leading strand (20). This non-physiological mechanism, referred to as the monopolymerase system, requires relatively high levels of the pol ␣⅐primase complex. When RFC and PCNA are present, they interfere with pol ␣ rebinding by binding to the 3Ј-hydroxyl primer ends where they facilitate the tethering of pol ␦ (or pol ⑀) (3,5,21). As a result, both leading strand synthesis and the completion of lagging strand pre-Okazaki fragments depend on pol ␦ (or pol ␦ and pol ⑀).
In eukaryotes, PCNA has been shown to be the target of a number of factors that control cell growth. One such factor, p21 CIP1 , is a checkpoint protein that acts as an antimitogenic signal by binding to and inhibiting cyclin-dependent kinases as well as by binding to PCNA and inhibiting in vitro PCNA-dependent DNA replication (22)(23)(24)(25)(26)(27)(28)(29). These two disparate inhibitory functions have been shown to reside in separate domains of p21 CIP1 , a 164-amino acid protein. The cyclin-dependent kinase inhibitory activity of p21 CIP1 is located within the amino-terminal domain while the PCNA binding region resides in the carboxyl-terminal domain. These distinct inhibitory activities have been demonstrated both in vitro and in vivo using the overexpressed amino-or carboxyl-terminal domains of p21 CIP1 (30,31).
A peptide derived from amino acids 141-160 within the carboxyl-terminal domain of p21 CIP1 that binds PCNA and inhibits PCNA-dependent DNA synthesis in vitro has previ-ously been described (32). In this report, we have examined the species specificity involved in the pol ␦ holoenzyme system using enzymes isolated from S. cerevisiae and HeLa cells and have shown that a chemically synthesized peptide that spans amino acid residues 139 -160 of p21 CIP1 specifically inhibits reactions dependent on human PCNA. This p21 CIP1 peptide-(139 -160) had no effect on elongation reactions dependent on S. cerevisiae PCNA, although both the full-length p21 CIP1 protein and a truncated p21 CIP1 protein (truncated p21 CIP1 protein-(70 -164)) inhibited the S. cerevisiae pol ␦ holoenzyme system albeit to a lesser extent than that observed with the human pol ␦ holoenzyme system.
Binding studies using surface plasmon resonance demonstrated that hPCNA binds p21 CIP1 and the p21 CIP1 peptide-(139 -160) stoichiometrically (ϳthree molecules bound per hPCNA trimer) with a similar affinity (K D ϳ2.5 nM). The affinity between S. cerevisiae PCNA and p21 CIP1 was 10-fold less than the affinity of hPCNA for p21 CIP1 , and there was no detectable interaction between the p21 CIP1 peptide-(139 -160) and S. cerevisiae PCNA. Recently the crystal structure of the p21 CIP1 peptide-(139 -160) complexed with hPCNA has been elucidated (33). Based on this structure and knowledge of the specific amino acid differences between S. cerevisiae and hPCNA, we propose a model explaining the differential effects of p21 CIP1 and its derivatives on S. cerevisiae and hPCNA-dependent reactions.
PCNA containing a cAMP-protein kinase recognition site at its amino-terminal domain was phosphorylated as follows. A reaction mixture (40 l) containing 20 mM Tris-HCl (pH 7.5), 12 mM Mg(OAc) 2 , 2 mM dithiothreitol, 0.1 M NaCl, 16.5 pmol of [␥-32 P]ATP (1.1 ϫ 10 7 cpm/ pmol), 65 pmol of amino-terminal tagged PCNA, and 6.66 units of cAMP-protein kinase was incubated at 37°C for 30 min. The reaction was halted by the addition of 2 l of 0.5 M EDTA, and aliquots, pre-and post-acid precipitation (filtered through GFC glass filters in the latter case), were counted by liquid scintillation. Approximately 60% of the 32 P was recovered in PCNA, and its specific activity was ϳ1500 cpm/ fmol. SDS-polyacrylamide gel electrophoresis analysis indicated that PCNA was the only labeled protein formed (data not shown).
Preparation of DNAs-Singly-primed DNA was isolated following hybridization of a 34-mer oligonucleotide to nucleotides 6300 -6333 of circular M13 mp 7 (7.2 kilobases) DNA (containing ϳ10% linear molecules). The annealed product was 32 P-labeled following the incorporation of a single dCMP residue (residue 6299) by Klenow fragment (Boehringer Mannheim). After phenol-CHCl 3 extraction, the reaction mixture was filtered through a G-50 Sephadex column, and the excluded labeled singly-primed M13 DNA (1000 -2000 cpm/fmol) was stored in buffer containing 10 mM Tris-HCl (pH 8.0) and 1 mM EDTA at 4°C.
Singly-nicked pBS DNA was prepared in a reaction mixture (1 ml) containing 300 g/ml ethidium bromide, 5 mM Tris-HCl (pH 7.5), 125 mM NaCl, 0.1 mg/ml bovine serum albumin, 20 mM MgCl 2 , 400 g of pBS DNA, and 12 g of pancreatic DNase I that was incubated for 25 min at 37°C. An equal volume of a solution of buffered-saturated phenol (Life Technologies, Inc.) was added followed by extraction of the aqueous phase with an equal volume of phenol-CHCl 3 /isoamyl alcohol (24:1, v/v). The aqueous phase was adjusted to 0.1 M NaCl, and 2 volumes of ethanol were added. The solution was centrifuged, dried in vacuo, and dissolved in 0.2 ml of 0.1 M NaCl in TE buffer (10 mM Tris-HCl (pH 8.0) ϩ 1 mM EDTA (pH 8.0)). Electrophoresis of this DNA through a 1.2% alkaline agarose gel (30 mM NaOH, 1 mM EDTA) indicated the presence of an equal mixture of single-stranded circular DNA molecules and linear DNA fragments, reflecting the presence of a single nick in the duplex circular DNA. Consistent with this, agarose gel electrophoresis in TAE buffer (0.04 M Tris acetate ϩ 1 mM EDTA (pH 8.0)) containing ethidium bromide indicated that the DNA had been quantitatively converted from a RF I structure to a RF II structure.
Protein Interaction Analyses and Determination of Kinetic Parameters-The immobilization of hPCNA, S. cerevisiae PCNA, p21 CIP1 , and the p21 CIP1 peptide-(139 -160) to sensor chips was carried out using the carbodiimide covalent linkage protocol specified in the manufacturer's instructions (Pharmacia Biotech Inc.). The interaction between each immobilized ligand and these proteins in solution was then followed by monitoring changes in surface concentration on the sensor chip using the BIAcore 2000. The equilibrium dissociation constants (K D ) for hPCNA binding to p21 and the p21 CIP1 peptide-(139 -160) were determined following passage of several concentrations of hPCNA over sensor chip surfaces to which each of these proteins had been covalently coupled. Both the BIAcore kinetic evaluation software version 1.2 and Scatchard analysis techniques were employed to determine K D values.
Singly-primed M13 Elongation and PCNA Loading Assays-The elongation of singly-primed M13 DNA and the subsequent analysis by alkaline agarose gel electrophoresis were carried out as described (5) with modifications as outlined in the figure legends. PCNA loading experiments were carried out as described previously for ␤ (40).

RESULTS
Species Specificity of the Pol ␦ Holoenzyme-The elongation of a [ 32 P]dCMP-labeled oligonucleotide primer (35 nucleotides) hybridized to circular single-stranded M13 mp 7 (7.2 kilobases) DNA by both the h and S. cerevisiae pol ␦ holoenzymes was compared (Fig. 1). Three different DNA binding proteins, HSSB, E. coli SSB, and the T4 gp 32, were compared in each system. HSSB was more effective than the other SSBs in supporting elongation of the primed template by the human pol ␦ holoenzyme (Fig. 1, lanes 1-3), whereas all three SSBs supported elongation of the labeled primer by the yeast pol ␦ holoenzyme to a similar extent (lanes 4 -6). For this reason all subsequent experiments were carried out using HSSB as the DNA binding protein.
The six different permutations possible for reconstituting the pol ␦ holoenzyme from combinations of h and S. cerevisiae PCNA, RFC and pol ␦ proteins were examined ( Fig. 1, lanes 9 -14). In all cases, the efficiency of full-length product formation with interspecies mixtures of these proteins was reduced compared with the full-length products formed in reactions containing proteins from a single species. The most pronounced species specificity was observed with mixtures containing hpol ␦, hRFC, and S. cerevisiae PCNA (lane 10) and reactions carried out with S. cerevisiae pol ␦, hRFC, and S. cerevisiae PCNA (lane 14). The finding that hRFC substituted poorly for S. cerevisiae RFC in the presence of S. cerevisiae pol ␦ is consistent with the data of Fien and Stillman (37). The experiments described by Fien and Stillman (37) were carried out with S. cerevisiae SSB, whereas the experiments described in Fig. 1 contained HSSB. Thus, the marked specificity noted is most likely independent of the SSB used. Limited elongation of primer chains was observed in reactions containing hPCNA, S. cerevisiae RFC, and S. cerevisiae pol ␦ (lane 9). The other combinations of human and S. cerevisiae proteins supported chain elongation (lanes [11][12][13], although a substantial number of products accumulated that had not been maximally extended. The formation of discrete shorter products in reactions containing heterologous mixtures of proteins indicated that the processive action of pol ␦ was reduced compared with reactions carried out with proteins from homologous species. Primer extension assays were repeated with an unlabeled primed template in the presence of [␣-32 P]dNTPs (Table I).
Qualitatively, the level of nucleotide incorporation mirrored the results shown in Fig. 1, i.e. reactions containing S. cerevisiae PCNA, hRFC, and hpol ␦ were virtually inactive. The combination of S. cerevisiae pol ␦, hRFC, and hPCNA resulted in substantial nucleotide incorporation (Table I), but the resulting DNA chains were highly heterogeneous in length (data not presented).
Elongation reactions catalyzed by either the h or S. cerevisiae pol ␦ holoenzyme were absolutely dependent on addition of all three holoenzyme components (pol ␦, RFC, and PCNA) as well as an SSB protein. Thus, in the absence of PCNA no detectable elongation of labeled primer was observed with either the human proteins (Fig. 1, lane 7) or the S. cerevisiae proteins (Fig.  1, lane 8). Likewise, omission of S. cerevisiae pol ␦ or S. cerevisiae RFC ( Fig. 2A, lanes 15 and 16, respectively) prevented DNA synthesis. Similar results were obtained following omission of any one of the pol ␦ holoenzyme components or SSB protein (data not presented).
The influence of the p21 CIP1 peptide-(139 -160) on the elongation of a 32 P-labeled primed M13 single-stranded DNA was examined ( Fig. 2A). As previously observed, high levels of fulllength p21 CIP1 blocked the elongation reaction catalyzed by the hpol ␦ holoenzyme ( Fig. 2A, lane 3). Elongation catalyzed by the S. cerevisiae pol ␦ holoenzyme was also inhibited but, in contrast to the human system, the labeled primer was elongated without the accumulation of full-length material ( Fig. 2A, lane  4). Interestingly, high levels of the p21 CIP1 peptide-(139 -160) had no effect on DNA elongation by the S. cerevisiae pol ␦ holoenzyme (lane 6). As previously noted, this p21 CIP1 peptide-(139 -160) quantitatively inhibited DNA synthesis by the hpol ␦ holoenzyme (Fig. 2A, lane 5). Elongation catalyzed by any combination of human and S. cerevisiae proteins that included hPCNA was also completely inhibited by the peptide (lanes 8, 10, and 12). The combination of hpol ␦, S. cerevisiae RFC, and S. cerevisiae PCNA was the only heterologous pol ␦ holoenzyme complex that included S. cerevisiae PCNA and supported elongation (lane 13). This reaction was not altered by addition of the p21 CIP1 peptide-(139 -160) (lane 14).
In the presence of the hpol ␦ holoenzyme, the extent of inhibition of DNA elongation by p21 CIP1 and its carboxyl-terminal derivatives was influenced by the ratio of PCNA to the inhibitor (27,28,38). For this reason, the influence of the p21 CIP1 peptide-(139 -160) on the S. cerevisiae pol ␦ holoenzyme was examined at various PCNA concentrations (Fig. 2B). DNA elongation became more efficient with increasing levels of S. cerevisiae PCNA (Fig. 2B, lanes 1, 4, and 7). Addition of the p21 CIP1 peptide-(139 -160) did not affect the elongation of the labeled primer irrespective of the level of S. cerevisiae PCNA present (Fig. 2B, compare lanes 1 and 3; lanes 4 and 6; lanes 7 and 9). In contrast, the influence of the p21 CIP1 peptide-(139 -160) on hpol ␦ holoenzyme-catalyzed nucleotide incorporation was dependent upon the level of PCNA added (Fig. 2C). In the presence of 170 and 68 nM hPCNA, 120 and 45 nM of the p21 CIP1 peptide-(139 -160) were required to inhibit DNA elongation by 50%, respectively.
Using overlapping 20-amino acid peptides Warbrick et al. (32) showed that the critical region of p21 CIP1 required for its interaction with PCNA spans amino acids 144 -151 (see Table  II). Consistent with this finding high levels of a 12-mer peptide derived from amino acids 139 -160 of p21 CIP1 containing a deletion of amino acids 146 -156 (see Table II) did not affect DNA synthesis catalyzed by the hpol ␦ holoenzyme (Fig. 2C).
Full-length p21 CIP1 and Truncated p21 CIP1 Protein-(70 -164) Inhibit Human and, Less Efficiently, Yeast PCNA-dependent DNA Synthesis-Full-length p21 CIP1 inhibited the S. cerevisiae pol ␦ holoenzyme reaction, and the length of DNA formed was dependent upon the ratio between the inhibitor and PCNA. Thus, reducing the level of PCNA in the presence of a fixed amount of p21 CIP1 resulted in the production of less full-length material (Fig. 2B, compare lanes 2, 5, and 8).
The influence of p21 CIP1 on DNA synthesis carried out in the presence of higher levels of singly-primed M13 DNA substrate  Further experiments investigating the influence of p21 CIP1 on the incorporation of labeled nucleotides confirmed that the extent of inhibition of hpol ␦ holoenzyme-catalyzed elongation by p21 CIP1 was dependent upon the level of hPCNA present (Fig. 3B). Nucleotide incorporation was reduced by 50% in the presence of 0.37, 0.18, and 0.10 M p21 CIP1 in reactions containing 170, 68, and 34 nM hPCNA (monomer), respectively. In the presence of the S. cerevisiae pol ␦ holoenzyme, 50% inhibition was observed with 1.2 M p21 CIP1 in the presence of either 170 or 17 nM S. cerevisiae PCNA, whereas reactions containing a low level of S. cerevisiae PCNA (1.7 nM) required the addition of 0.8 M p21 CIP1 to inhibit DNA synthesis by 50%. Altering the amount of S. cerevisiae RFC or S. cerevisiae pol ␦ added to reactions containing a limiting concentration of S. cerevisiae PCNA did not significantly alter the level of p21 CIP1 required to inhibit DNA synthesis by 50% (data not shown).
The inhibition of the S. cerevisiae pol ␦ holoenzyme by p21 CIP1 and the lack of inhibition by the p21 CIP1 peptide-(139 -160) prompted us to examine the effects of the addition of a FIG. 2. The p21 CIP1 peptide-(139 -160) inhibits human but not yeast PCNA-dependent DNA synthesis. A, reaction mixtures were as described in Fig. 1 except that HSSB was used as the DNA binding protein in all cases. Where indicated, 12.5 pmol of p21 CIP1 or 12.5 pmol of the p21 CIP1 peptide-(139 -160) was added. Reaction mixtures were incubated for 10 min at 0°C prior to the addition of RFC and pol ␦ to permit an interaction between p21 CIP1 or the peptide and PCNA to take place. Following addition of RFC and pol ␦, the incubation was continued at 37°C for 20 min before being processed as described in Fig. 1. Dried gels were exposed for autoradiography for 4 h. B, influence of S. cerevisiae PCNA concentration on the effects of p21 CIP1 and p21 CIP1 peptide-(139 -160). Reactions (10 l) were as described in Fig. 1 except that 4.6 fmol of singly-primed M13 DNA (9.31 ϫ 10 3 cpm) was added. Reaction mixtures were incubated at 37°C for 5 min in the presence of 1.25 M of p21 CIP1 or p21 CIP1 peptide-(139 -160) prior to the addition of S. cerevisiae pol ␦ and RFC. Following addition of these proteins, reactions were incubated for a further 20 min at 37°C before being processed as described in Fig. 1. C, quantitative effects of the p21 CIP1 carboxyl-terminal-derived peptides on h and sc (S. cerevisiae) pol ␦ holoenzyme-catalyzed deoxynucleotide incorporation as a function of PCNA concentration. Reaction mixtures (10 l) were as described in Fig. 1    peptide that spans an additional 70 amino acids within the carboxyl terminus of p21 CIP1 (Fig. 3C). Previous studies have shown that a truncated p21 CIP1 protein derived from amino acids 70 -164 within the carboxyl-terminal of p21 CIP1 was capable of binding to hPCNA as well as inhibiting hpol ␦ holoenzyme-catalyzed elongation of primed templates (31). In con- FIG. 3. Full-length p21 CIP1 and truncated p21 CIP1 peptide-(70 -164) inhibit human and less efficiently yeast PCNA-dependent DNA synthesis. A, influence of S. cerevisiae (sc) PCNA and p21 CIP1 concentrations on S. cerevisiae pol ␦ holoenzyme activity. Reactions (12 l) were as described in A except that 11 fmol of singly-primed M13 DNA was added with 20 mM [␣-32 P]dCTP (26,000 cpm/pmol). Reactions were incubated for 10 min at 37°C and processed as described in Fig. 1. B, quantitative effects of p21 CIP1 on h and S. cerevisiae pol ␦ holoenzyme catalyzed deoxynucleotide incorporation as a function of PCNA concentration. Reaction mixtures (10 l) were processed and quantitated as described in trast to the effects of the p21 CIP1 peptide-(139 -160), the truncated p21 CIP1 protein-(70 -164) inhibited the elongation reaction catalyzed by the S. cerevisiae pol ␦ holoenzyme (Fig.  3C, lanes 7-9). These results suggest that amino acids within codons 70 -139 are capable of interaction with S. cerevisiae PCNA. Like the full-length p21 CIP1 protein, the effects of the truncated p21 CIP1 protein-(70 -164) were more pronounced in the presence of low levels of PCNA than at higher PCNA concentrations (compare lanes 9 and 7).
The Binding of p21 CIP1 and the p21 CIP1 Peptide-(139 -160) to Human and S. cerevisiae PCNA-In order to explain the different effects of the p21 CIP1 and p21 CIP1 peptide-(139 -160) on S. cerevisiae and hPCNA-dependent reactions, we determined whether these proteins exhibit differential binding affinities for PCNA. p21 CIP1 was shown previously to directly interact with hPCNA by coimmunoprecipitation, gel filtration, and surface plasmon resonance binding (27,28). The interaction detected using the latter procedure indicated that ϳ2.3 molecules of p21 CIP1 monomer bound to each PCNA trimer. Consistent with this, passage of solutions of p21 CIP1 over a sensor surface on which 110 fmol of hPCNA (3320 RU) was immobilized resulted in an immediate increase in mass (Fig. 4A). In this experiment, 140 fmol of p21 CIP1 (3009 RU) was retained on the hPCNA sensor chip, corresponding to a stoichiometry of ϳ1 molecule of p21 CIP1 monomer bound per molecule of PCNA monomer. Fourteen-fold less p21 CIP1 (10 fmol, 222 RU) bound to a sensor surface to which 92 fmol (2748 RU) of S. cerevisiae PCNA had been covalently coupled (Fig. 4B). The background values have been subtracted from these data and were calculated following passage of p21 CIP1 solutions over a blank sensor chip surface (Fig. 4C). These data indicate that the affinity of S. cerevisiae PCNA for p21 CIP1 is 14-fold less than the affinity of hPCNA for p21 CIP1 . These results were confirmed following passage of S. cerevisiae or hPCNA solutions over a sensor surface chip to which p21 CIP1 was immobilized (data not shown).
A 100-fold difference in the association between S. cerevisiae and hPCNA with the p21 CIP1 peptide-(139 -160) was observed.
In this experiment, 100 fmol (281.4 RU) and 1 fmol (30 RU) of the p21 CIP1 peptide-(139 -160) were retained by sensor chips to which hPCNA (110 fmol, 3320 RU) or S. cerevisiae PCNA (92 fmol, 2748 RU) had been covalently coupled, respectively (Fig.  4, D and E). As described above, background values have been subtracted and were calculated following passage of the p21 CIP1 peptide-(139 -160) over a blank sensor chip surface (Fig. 4F). These data indicate that approximately 1 molecule of the p21 CIP1 peptide-(139 -160) binds to each monomer molecule of hPCNA, similar to the stoichiometry observed with the fulllength p21 CIP1 protein. In contrast, no significant interaction was observed between the p21 CIP1 peptide-(139 -160) and S. cerevisiae PCNA. Identical results were obtained following p21 CIP1 peptide-(139 -160) immobilization on a sensor chip over which solutions of h or S. cerevisiae PCNA were continuously flowed (data not shown).
Sensorgrams recorded using different hPCNA concentrations passing over the surface of chips to which p21 CIP1 or the biotinylated p21 CIP1 peptide-(139 -160) had been coupled are shown in Fig. 5, A and B The Influence of p21 CIP1 and the p21 CIP1 Peptide-(139 -160) on the RFC-dependent Loading of PCNA onto DNA-Previous experiments suggested that p21 CIP1 inhibits DNA synthesis as a result of its effect on DNA chain elongation rather than RFC-catalyzed PCNA loading. At a fixed concentration of hPCNA, the addition of increasing levels of p21 CIP1 resulted in the concomitant decrease in the length of DNA chains synthesized (27).
hPCNA containing a cAMP protein kinase recognition site at its amino terminus was labeled with 32 P. The direct loading of hPCNA onto DNA by RFC could then be followed after A15m agarose gel filtration which separated 32 P-PCNA on DNA from free 32 P. As shown in Fig. 6, incubation of hRFC and 32 Plabeled PCNA with singly-nicked pBR322 DNA and ATP resulted in the elution of a peak of 32 P-labeled hPCNA from the sizing column in the excluded volume. When reactions were carried out in the presence of p21 CIP1 or the p21 CIP1 peptide-(139 -160) (in amounts that quantitatively inhibited DNA synthesis), the formation of an excluded peak of 32 P-labeled PCNA complexed to DNA was inhibited by only 50%. This finding indicates that the inhibitory effects of p21 CIP1 and the p21 CIP1 peptide-(139 -160) on hpol ␦ holoenzyme-catalyzed DNA synthesis are not predominantly due to inhibition of RFC-catalyzed PCNA loading but to inhibition of the subsequent interaction between PCNA and pol ␦ and/or the ability of PCNA to slide along DNA. DISCUSSION The experiments presented here investigated the interactions between p21 CIP1 and a p21 CIP1 carboxyl-terminal derived peptide with PCNA and the subsequent effect on pol ␦-catalyzed DNA elongation. Full-length p21 CIP1 inhibited DNA synthesis catalyzed by h or S. cerevisiae pol ␦ holoenzymes to an extent dependent upon the ratio between p21 CIP1 and PCNA.
In contrast, addition of the p21 CIP1 -derived peptide-(139 -160) only inhibited reactions dependent upon hPCNA, not S. cerevisiae PCNA. Thus, reactions containing S. cerevisiae pol ␦ and S. cerevisiae RFC with hPCNA were inhibited by the p21 CIP1 peptide-(139 -160), whereas reactions containing h or S. cerevisiae pol ␦, S. cerevisiae RFC, and S. cerevisiae PCNA were unaffected by the peptide. The species origin of the pol ␦ and RFC did not influence DNA synthesis inhibition by p21 CIP1 and its derivatives.
Real time interaction analysis presented here demonstrated that the affinity of the p21 CIP1 for hPCNA was ϳ15-fold greater than its affinity for S. cerevisiae PCNA. This is consistent with the biochemical data presented indicating that 12 times more p21 CIP1 was required to inhibit S. cerevisiae PCNA-dependent nucleotide incorporation by 50% than was required to inhibit hPCNA-dependent incorporation by 50%. Furthermore, the affinity of the p21 CIP1 peptide-(139 -160) for hPCNA was 100-fold greater than its affinity for S. cerevisiae PCNA. These data are also consistent with the singly-primed M13 elongation assays presented which indicated that addition of the p21 CIP1 peptide-(139 -160) in amounts stoichiometric with hPCNA inhibited nucleotide incorporation by 50%, whereas the addition of excess peptide did not affect reactions dependent upon S. cerevisiae PCNA.
The crystal structure of the p21 CIP1 peptide-(139 -160) complexed with hPCNA has recently been determined (33). This analysis demonstrated that the p21 CIP1 peptide-(139 -160) is bound to the inter-domain connector loop that links the aminoand carboxyl-terminal domains of each PCNA monomer (residues 119 -133). Additional interactions are formed with both domains of PCNA. A schematic model of the hPCNA-p21 CIP1 peptide-(139 -160) complex derived from crystal structure analyses is presented in Fig. 7. Although the ␣-carbon backbone traces of h and S. cerevisiae PCNA are very similar (8,41), significant differences in the backbone conformation are located in the regions where the p21 CIP1 peptide-(139 -160) interacts with hPCNA. These differences may account for the distinct effects of both p21 CIP1 and the p21 CIP1 peptide-(139 -160) in these two systems.
First, and of particular interest, the configuration of the hPCNA interconnector loop that forms part of the p21 CIP1 peptide-(139 -160) binding surface differs in S. cerevisiae PCNA. A number of residues in this interconnector loop contribute to the formation of two hydrophobic pockets in hPCNA that bind the Met-147, Phe-150, Tyr-151, and Ile-158 residues of the p21 CIP1 peptide-(139 -160). There is also an extensive array of hydrogen bonding-mediated contacts between hPCNA and the p21 CIP1 peptide-(139 -160) in this region. Thus both the conformation and sequence of the interconnector loop of hPCNA are important in forming the interface with the p21 CIP1 peptide-(139 -160). Specific differences in the amino acid sequence between the h and S. cerevisiae PCNA are likely to exclude some favorable contacts with the p21 CIP1 peptide-(139 -160) and S. cerevisiae PCNA. For example, Ile-158 of the p21 CIP1 peptide-(139 -160) fits snugly into a small hydrophobic pocket adjacent to the hPCNA connector loop. Substitution of a cysteine residue (Cys-27) for asparagine at this position in S. cerevisiae PCNA would significantly alter the nature of the binding site. Additionally, Arg-156 of the p21 CIP1 peptide- (139 -160) participates in hydrogen bonding interactions with Asp-29 and Gln-125 of hPCNA, whereas in the S. cerevisiae PCNA these residues are replaced by glutamine and phenylalanine, respectively, precluding these interactions.
In the carboxyl-terminal region of the hPCNA connector loop, residues Leu-126, Gly-127, Ile-128, and Pro-129 form one edge of a binding cleft into which the methionine and tyrosine side chains of the p21 CIP1 peptide-(139 -160) are inserted. The conformation of the S. cerevisiae PCNA connector loop in this region greatly differs to that of the hPCNA and probably precludes insertion of these p21 CIP1 peptide side chains. The p21 CIP1 peptide-(139 -160) tyrosine residue also interacts with the carboxamide group of Gln-131 in hPCNA, which is replaced by a leucine (Leu-131) in S. cerevisiae PCNA, although the backbone of the connector loop has shifted such that the carboxylate moiety of Glu-130 in S. cerevisiae PCNA occupies the same relative position. In addition to these changes, subtle shifts in the position of a ␤-strand and a short loop which border this hydrophobic cavity optimize the complementarity of the p21 CIP1 peptide-(139 -160) and hPCNA surfaces. The presence of the Met-40 in hPCNA (instead of a valine residue in S. cerevisiae PCNA) augments these interactions in tailoring the shape of the binding site.
Thus, a number of the specific residues that dictate the conformation of the connector loop of hPCNA and its high affinity interaction with the p21 CIP1 peptide-(139 -160) are not conserved in S. cerevisiae PCNA, and these differences proba-bly account for the inability of the p21 CIP1 peptide-(139 -160) to interact with or inhibit S. cerevisiae PCNA.
Results from Warbrick et al. (32) also suggest that a region that includes the interdomain connector of hPCNA is essential for its interaction with p21 CIP1 . Using the yeast 2-hybrid system, they demonstrated that the interconnector loop of each PCNA monomer interacts with p21 CIP1 , whereas PCNA with a deletion of amino acids 100 -150 could not form this complex. The connector loop of hPCNA also appears to be critical for its interaction with pol ␦ (42). A monoclonal antibody against hPCNA that blocks the interaction between hpol ␦ and hPCNA was competed by a peptide derived from amino acids 121-135 within the connector loop of PCNA. Further evidence for the importance of this region of PCNA comes from cold-sensitive mutations of the S. cerevisiae PCNA gene (POL30) that are clustered in the interconnector domain region and exhibit a cell division cycle phenotype at the restrictive temperature arresting with a 2C DNA content (43).
The effects of p21 CIP1 and the p21 CIP1 peptide-(139 -160) on hPCNA-dependent DNA synthesis, the complex stoichiometry, and dissociation constants between each of these agents and hPCNA are all quantitatively similar. This suggests that the biological consequence of the p21 CIP1 interaction with hPCNA is quantitatively due to the region contained within the 22amino acid p21 CIP1 peptide-(139 -160). In contrast, the effects of full-length p21 CIP1 and the p21 CIP1 peptide-(139 -160) on S. cerevisiae PCNA are different. Furthermore, p21 CIP1 was much less effective in blocking S. cerevisiae PCNA-dependent reactions than hPCNA-dependent reactions. These findings indicate that p21 CIP1 may affect the S. cerevisiae pol ␦-holoenzyme by a mechanism distinct from that observed with the hpol ␦-holoenzyme. Since S. cerevisiae PCNA has a low affinity for p21 CIP1 and no detectable affinity for the p21 CIP1 peptide-(139 -160), other sites within p21 CIP1 (present in the p21 CIP1 peptide-(70 -164)) may contribute to its interaction with S. cerevisiae PCNA and subsequent effects on the S. cerevisiae pol ␦-holoenzyme activity.
The loading of PCNA onto DNA by RFC was maximally inhibited by 50% following addition of concentrations of p21 CIP1 that completely blocked the elongation of primed DNA. Hü bscher's laboratory (44) has shown that PCNA loaded onto circular DNA can still slide off the DNA in the presence of p21 CIP1 after linearization of the DNA. These results suggest that p21 CIP1 may act to inhibit the interaction between the polymerase and PCNA through its ability to bind to the interconnector domain of PCNA. This would suggest that all reactions dependent upon PCNA and pol ␦ (or pol ⑀) would be affected by p21 CIP1 including DNA repair reactions in which PCNA plays the same role as in replication, acting to tether the polymerase to DNA (45,46).