Clamp loading, unloading and intrinsic stability of the PCNA, ¯ and gp45 sliding clamps of human, E. coli and T4 replicases

Nina Yao1, Jennifer Turner1, Zvi Kelman1,a, P. Todd Stukenberg2,b Frank Dean1, David Shechter1, Zhen-Qiang Pan2, Jerard Hurwitz2 and Mike O’Donnell1, * 1Microbiology Department and Hearst Research Foundation, *Howard Hughes Medical Institute, Cornell University Medical College, 1300 York Avenue, New York, HY 10021, USA, and 2Graduate Program in Molecular Biology, Memorial Sloan-Kettering Cancer Centre, 1275 York Avenue, New York, NY 10021, USA

Communicated by: Martin Gellert

Abstract

Background: The high speed and processivity of oligomeric state. The T4 gp45 clamp is a much replicative DNA polymerases reside in a processivity less stable trimer than PCNA (Kd % 250 nM) and factor which has been shown to be a ring-shaped requires association with the polymerase to stabilize protein. This protein (‘sliding clamp’) encircles it on DNA as observed previously. The consequence DNA and tethers the catalytic unit to the template. of this cooperation between clamp and polymerase Although in eukaryotic, prokaryotic and bacter- is that upon finishing a template and dissociation of iophage-T4 systems, the processivity factors are the polymerase from DNA, the gp45 clamp sponta-

ring-shaped, they assume different oligomeric neously dissociates from DNA without assistance.

¯ states. The Escherichia coli clamp (the ¯ subunit) is However, the greater stability of the PCNA and

active as a dimer while the eukaryotic and T4 phage clamps on DNA necessitates an active process for clamps (PCNA and gp45, respectively) are active as their removal. The clamp loaders (RF-C and ° trimers. The clamp can not assemble itself on DNA. complex) were also capable of unloading their Instead, a protein complex known as a clamp loader respective clamps from DNA in the presence of ATP. utilizes ATP to assemble the ring around the Conclusions: The stability of the different clamps in primer-template. This study compares properties solution correlates with their stability on DNA. of the human PCNA clamp with those of E. coli and Thus, the low stability of the T4 clamp explains the T4 phage.

inability to isolate gp45 on DNA. The stability of the

Results: The PCNA ring is a stable trimer down to a PCNA and ¯ clamps predicts they will require an concentration below 100 nM (Kd % 21 nM). On DNA, unloading factor to recycle them on and off DNA

the PCNA clamp slides freely and dissociates from during replication. The clamp loaders of PCNA and

¯ ¯

DNA slowly (t1/2 % 24 min). is more stable in double as clamp unloaders presumably for the solution (Kd <60pM) and on DNA (t1/2 % 1 h) than purpose of clamp recycling. PCNA which may be explained by its simpler

Introduction

* Corresponding author: Fax: +1 212 746 8587. The sliding clamps of chromosomal replicases are ring- a Present address: Department of Molecular Biology and Genetics, shaped proteins that encircle DNA and tether the John Hopkins University, School of Medicine, 725 N. Wolfe replicase to the template for highly processive chain Street, Baltimore, MD 21205-2195, USA. elongation (reviewed in Kuriyan & O’Donnell 1993). b Present address: Department of Cell Biology, Harvard Medical The clamp loader recognizes a primed template

School, Boston, MA 02115, USA junction and couples ATP hydrolysis to assemble the 5 Blackwell Science Limited Genes to Cells (1996) 1, 101–113 101

N. Yao et al. clamp around DNA. The eukaryotic clamp is the similar to  in its stability on DNA and requirement for a

proliferating cell nuclear antigen (PCNA), the pro- clamp unloader for its removal. karyotic clamp is the  subunit and the T4 phage clamp is the product of gene 45 (gp45) (reviewed in Kuriyan

Results

& O’Donnell 1993). The E. coli  sliding clamp is a dimer (Stukenberg et al. 1991; Kong et al. 1992) while To directly follow the properties of PCNA,  and gp45 the eukaryotic and T4 phage sliding clamps are formed in these studies we have radiolabelled them. All three by trimers (Krishna et al. 1994; Jarvis et al. 1989). In each were tritiated by reductive methylation, a modification case, the clamps have been shown to slide on that results in a 3H-methyl group on one or two lysine DNA freely (Stukenberg et al. 1991; Gogol et al. residues of each molecule (the charge is retained at 1992; Burgers & Yoder 1993; Tinker et al. 1994). The physiological pH) (Kelman et al. 1995a). Additionally,

clamp loaders in each of these systems are composed some studies were performed using 32P-labelled clamps

of multiple subunits. The E. coli clamp loader is the using  and PCNA containing a kinase recognition five-subunit  complex, the eukaryotic clamp loader is motif on the C- or N-terminus, respectively. the five protein RF-C (also called activator-1) and the T4 clamp loader is the gene protein 44/62 complex In solution, the trimer clamps are less stable (also a five subunit structure, reviewed in Kelman & oligomers than the dimer clamp

O’Donnell 1994). A major difference among these systems is that the  and PCNA clamps can be Previous studies have shown that upon transfer of sliding

isolated on DNA by gel filtration whereas the T4 clamps to DNA by their respective clamp loaders, the gp45 clamp requires association with its DNA PCNA and  clamps can be isolated on DNA by gel polymerase, the product of gene 43 (gp43), to be filtration, but the T4 gp45 clamp can not (Wickner

isolated on DNA. 1976; O’Donnell 1987; Maki & Kornberg 1988; Lee &

In E. coli, the  complex acts catalytically to assemble Hurwitz 1990; Burgers 1991; Capson et al. 1991; the  clamp on DNA and can be removed from the Richardson et al. 1991). One possible explanation for clamp-DNA complex (Wickner 1976; Maki & Korn- the difference (in ability to form stable clamps on DNA)

berg 1988; Stukenberg et al. 1991). The DNA amongst the three systems could be the inherent stability polymerase III (Pol III core) then associates with the  of the oligomeric structure of the clamps themselves. In

clamp on DNA for processivity. However, an additional Fig. 1 the oligomeric structure of these clamps were

‘connector’ protein called ( binds two molecules of Pol examined by gel filtering a mixture of all three in

III core and one  complex to produce a tightly solution. They were analysed at a high concentration associated particle called Pol III* (Onrust et al. 1995). (2.5 "M, Fig. 1A) and then at a 50-fold lower concen- The two Pol III core polymerases are thought to tration (50 nM, Fig. 1B). To follow these proteins we replicate the leading and lagging strands concurrently added a small amount of 3H-protein and analysed their

(Sinha 1980; Kornberg & Baker 1992). elution patterns by fluorography after SDS-polyacryl- In eukaryotes and T4 phage, no such particle amide gel electrophoresis. At 2.5 "M, each clamp eluted

containing both polymerase and clamp loader has at a position consistent with its native oligomeric state: been identified. Although it is presumed that the PCNA as a trimer (86.7 kDa, fractions 28–31), ˜ as a clamp loader and polymerase assemble together with dimer (81.2 kDa, fraction 31–34) and gp45 as a trimer the clamp on the DNA in these systems, it remains (74.1 kDa, fractions 31–34). However, at 50 nM, the

possible that the clamp loader acts catalytically resulting oligomeric states of the trimeric clamps begin to change, in only the clamp-polymerase complex, as observed in but  remained a stable dimer. At 50 nM, PCNA showed the E. coli system. Indeed, the catalytic action of the T4 two distinct peaks; one correlated with its trimeric form, gp44/62 complex has been documented (Kaboord and the other was a smaller species that eluted in

& Benkovic 1995). In eukaryotes, both DNA poly- fractions 37-40. At this low concentration, gp45 eluted

32

    merase  (pol ) and DNA polymerase (pol ) interact entirely as a smaller species in fractions 37–43. P- of

with the PCNA clamp (Burgers 1991; Lee et al. 1991a). high specific activity was used to examine the behaviour

32

 In this report the inherent strength of the oligomeric of  at even more dilute conditions. P- remained a

structure of the clamps in solution, and their stability on stable dimer even upon dilution to 67 pM and incubation DNA have been examined. Despite the apparent at 37 VC with 1 M NaCl in buffer D for 6 h (J.

similarities of PCNA and T4 gp45 (both are trimers Andjelkovic & M. O’Donnell, unpublished data). The



and less stable oligomers than the  dimer), PCNA is possibility that the monomer elutes at the same 102 Genes to Cells (1996) 1, 101–113 5 Blackwell Science Limited

Comparison of DNA sliding clamps

the  dimer, as expected, supporting the conclusion that the wild type  dimer remains a stabile oligomer at the lowest concentrations tested. Ring shaped trimer oligomers are expected to be less stabile oligomers that ring shaped dimers (Kelman et al. 1995b). In a protein ring each protomer has two contacts, one with each adjacent protomer. In the case of the trimer, upon dissociation of one protomer, the remaining two protomers now have only one contact, which should lead to cooperative disassembly of the entire ring. Thus, it seems reasonable to expect that the smaller species observed upon dilution of PCNA is the monomer. Further dilution did not result in a third, smaller species, as would be expected if PCNA disassembled in stages from trimer-to-dimer- to-monomer. A recent study has shown that the PCNA trimer dissociates into a putative dimer form upon dilution (Zhang et al. 1995). In light of the above reasoning, a dimeric form of PCNA that is more stabile than the trimer would require a conformational change leading to a tighter contact between the monomers of the dimer than exists in the trimer (e.g. perhaps a dimer ring in which the two protomers form a smaller central

cavity). 100 nM radioactive PCNA was included in the

gel filtration analysis of the monomeric  mutant of Fig. 1C. A PCNA monomer should elute later than a 

monomer, and a PCNA dimer should elute earlier than a  monomer. However, the autoradiogram of the gel

filtration fractions showed that the disaggregated form of PCNA comigrated with the monomer  in this column (not shown). This result is consistent with the smaller species of PCNA as a dimeric form, but a monomer can not be ruled out. A linear form of the PCNA

Figure 1 Trimer clamps (PCNA and gp45) dissociate at low trimer (i.e. an open ring) can be excluded on the basis concentration while the ˜ ring remains a dimer. A mixture of the

that this would be an intramolecular reaction which three clamps (PCNA, ˜ and gp45) was gel filtered on Superose 12

should be concentration independent and not affected " at 4 VC at two different concentrations: (A) 2.5 M and (B) 50 nM.

In both experiments, 5 pmol of 3H-clamp was present to follow by dilution.

˜ protein by autoradiography. (C) A mixture of 2.5 "M monomer The ability to observe the PCNA trimer and its mutant (˜ Leu108Pro) and 100 nM radiolabelled PCNA was gel disaggregated form in one sample provided the means to filtered and a SDS polyacrylamide gel of the column fractions was approximate the dissociation constant of the PCNA stained with Coomassie Blue. ring. Figure 2A shows that as PCNA is diluted from 250 nM to 10 nM, the proportion of the smaller form of position as a dimer due to the unique shape of these ring PCNA increases at the expense of the trimer. Figure 2B proteins was examined. For this purpose, a mutation in shows a double reciprocal plot of the fraction of PCNA

the dimer interface was introduced (Leu108Pro) and the in trimeric and disaggregated form as a function of mutant was purified. This mutant , which exists only as PCNA concentration. The x-intercept corresponds to a

a monomer, was soluble but inactive in replication Kd value of 21 nM.

assays. The Leu108Pro mutant of  was gel filtered at a The gp45 trimers started to dissociate at concentra- concentration of 2.5 "M. The Coomassie Blue stained tions greater than 300 nM. However, unlike PCNA, the

SDS-polyacrylamide gel (Fig.1C) showed that the gel filtration profile of gp45 showed a single peak which

monomer  eluted in fractions 38–42. Hence, the shifted in a continuous fashion to the lower molecular

monomer form of  elutes in a different position than weight species as its concentration was decreased 5 Blackwell Science Limited Genes to Cells (1996) 1, 101–113 103 N. Yao et al.

(data not shown). This suggests a rapid equilibrium between dissociation and reassociation of the gp45 trimer. For this reason, the dissociation constant for

gp45 could not be accurately determined, but can be estimated to be % 200–400 nM from the shift in the peak.

In summary, these experiments demonstrate that in solution, the  dimer is more stable than the PCNA

trimer, which in turn is more stable than gp45. E. coli  and human PCNA form stable preinitiation complexes on DNA, but T4 gp45

does not Next, PCNA, gp45 and  were compared for their ability to be placed on singly primed M13mp18 ssDNA during gel filtration. This is a non-equilibrium technique and thus only stable protein–DNA com- plexes survive this procedure. In these experiments, each clamp was assembled onto the primed template by its respective clamp loader and ATP. The ssDNA was coated with E. coli SSB except in the case of T4 proteins in which the ssDNA was coated with T4 gp32. Figure 3 shows a comparison of the gel filtration profiles resulting from the clamp loading reactions in the three systems. The clamps were radiolabelled, therefore enabling us to track their distribution during gel filtration over a Bio-Gel A15 m column which resolves large DNA molecules, and 3H-proteins bound to them

(fractions 9–13), from proteins not bound to DNA

 (fractions 20–30). The results show that % 1.5–2 clamps were assembled onto each DNA molecule. Slightly less PCNA, 1–1.5 clamps per DNA molecule, was observed in the human system. Association of the T4 gp45 with DNA was too weak to be detected by gel filtration [Fig.3C (circles)]. However, when the poly- merase, gp43, was included in the loading reaction some of the gp45 remained with the DNA during gel filtration. The amount of gp43 in this experiment was stoichiometric to the DNA template and the gp45 was supplied in a 15-fold molar excess over the DNA.

Under these conditions of excess gp45 and limiting gp43, % 0.25–0.5 gp45 clamps were retained per DNA molecule. Due to the inability to isolate gp45 on DNA in the absence of the T4 DNA polymerase, we could not analyse the dynamic behaviour of gp45 in the Figure 2 Kd measurement for disaggregation of the PCNA ring.

(A) The indicated concentrations of 3H-PCNA as trimers were experiments of Figs 4–6 described below.

incubated at 37 VC for 1.5 h before gel filtration on Superose-12 at 4 VC as described in Experimental proceedures. (B) Double Human PCNA has similar sliding dynamics

reciprocal plot of PCNA trimer and ‘monomer’ as a function of on DNA as the E. coli  clamp protein concentration. 1/r represents the ratio of total fmol of

‘monomer’ PCNA to fmol of trimer PCNA. Previously we developed DNA sliding assays for clamp 104 Genes to Cells (1996) 1, 101–113 5 Blackwell Science Limited Comparison of DNA sliding clamps

3

H-) was assembled onto a plasmid containing a single nick using RF-C (or  complex) and ATP, then the reaction was gel filtered to remove clamps not bound to

3 3 DNA (not shown). The H-PCNA (or H-) on DNA was then divided into separate tubes and one half was treated with BamHI to linearize the plasmid. The two reactions were then gel filtered a second time. The results, in panel A, show that 3H-PCNA remained bound to the circular DNA, but dissociated from linear

DNA. These results are comparable to those observed

with the  clamp [bottom plot in panel A and in previous work on  (Stukenberg et al. 1991)]. To further test that 3H-PCNA dissociated from linear DNA by sliding over the ends, a protein block experiment was performed (panel B). 3H-PCNA was assembled onto a nicked plasmid containing oriP, the latent origin of replication of the Epstein-Barr virus. The oriP sequence contains two elements, each of which bind several molecules of the viral encoded EBNA-1 protein (Frappier & O’Donnell 1991). Sufficient EBNA-1 was added to saturate both sites. Then the reaction was gel filtered and the 3H-PCNA bound to DNA was collected from the excluded fractions of a gel filtration column. The pool of DNA-bound clamps was divided into two tubes and one was treated with EcoRV to linearize the DNA between the EBNA-1 sites, then the two reactions were gel filtered side-by-side. The result shows that just as much 3H-PCNA was retained on the linear DNA by protein bound near the ends, as was retained on the circular DNA. This result supports the conclusion that PCNA slides off the ends of DNA, and is comparable to

3 the result of similar experiments using H- (panel B, lower plot) and observed in our previous study (Stukenberg et al. 1991). This sliding behaviour of PCNA is consistent with

earlier reports. Briefly, yeast PCNA was shown to confer

processivity to pol  on a linear template, in the absence Figure 3 PCNA and ˜ form stable clamps on circular DNA but of clamp loader and ATP, by slipping onto the ends of gp45 requires DNA polymerase for stability. Clamps were loaded the DNA and sliding to the primer-template junction onto 800 fmol (8 nM) singly primed M13mp18 ssDNA using

3 3 3 (Burgers & Yoder 1993). Further, a protein-DNA 12 pmol (120 nM) of either H-˜, H-PCNA, or H-gp45; and

crosslinking study showed that PCNA could be cross-

1 pmol (10 nM)  complex, RF-C or gp44/62 complex, respectively. The reactions were incubated at 37 VC for 10 min. linked to DNA as it tracked along the DNA (Tinker et before applying to Bio-Gel A15 m columns equilibrated in buffer al. 1994). Additionally, it was found that PCNA could 3

E containing 100 mM NaCl. The H-gp45 reaction was repeated be stably isolated on circular but not linear DNA (Podust in the presence of 864 fmol gp43, and 60 "M each of dGTP and et al. 1995).

dCTP. Human PCNA and E. coli  are slow to proteins (Stukenberg et al. 1991), and these assays have

been applied to human PCNA in Fig. 4. Results dissociate from DNA

˜ obtained with in side-by-side experiments are The time required for the PCNA and  rings to

shown for comparison. In panel A, 3H-PCNA (or dissociate from DNA was measured in the experiment 5 Blackwell Science Limited Genes to Cells (1996) 1, 101–113 105 N. Yao et al.

3

Figure 4 Dynamics of the PCNA ring on DNA. (A) ‘Sliding assay’: H-PCNA (1 "g, 11.5 pmol as trimer) was incubated with nicked

" V plasmid (3.4 "g, 914 fmol as circles) and 200 ng (675 fmol) RF-C in a 100 L reaction for 30 min. at 37 C and then gel filtered on a Bio-

" Gel A15 m column at 4 VC equilibrated with buffer E containing 50 mM NaCl. The excluded fractions (nos 11 and 12, 280 L)

3

containing H-PCNA on DNA were pooled and divided into two tubes. One tube was treated with 6 "L (120 U) BamHI, and the other

V tube was treated with 6 "L of BamHI storage buffer. The reactions were incubated for 1 min. at 37 C followed by gel filtration as

3

"  described above. In the bottom panel, the experiment was repeated using H-˜ (1 g, 12 pmol as dimer) and complex (336 ng, 1680

3 3 3

" fmol) in place of H-PCNA and RF-C. (B) ‘Protein-block assay’: Either H-PCNA or H-˜ was assembled onto 2 g (615 fmol) of nicked pGEMoriP plasmid containing the latent origin (oriP) of the Epstein-Barr virus as described for the ‘sliding assay’. Then 4.3 "g (179 pmol) of EBNA-1 (DNA binding domain) was added to saturate the two elements of oriP and the reaction was gel filtered to remove

3 3

free protein. The excluded fractions containing H-PCNA (or H-˜) on DNA were pooled and divided into two tubes. One tube was

" treated with 10 "L (200 U) of EcoRVand the other tube was treated with 10 L of EcoRV storage buffer [10 mM Tris-HCl (pH 7.4),

V 0.1 mM EDTA, 1 mM DTT, 50 mM NaCl, 50% glycerol, 200 "g/mL BSA]. After 1 min. at 37 C the reactions were gel filtered and

fractions were analysed for 3H-protein as described in Experimental proceedures. shown in Fig. 5. Either radiolabelled PCNA, or , were DNA (as will be shown later). Hence, if RF-C is assembled onto a plasmid containing a single nick using present, it may speed the dissociation of PCNA from their respective clamp loader and ATP. Radioactive DNA. In fact, PCNA dissociation from DNA appears to clamps bound to DNAwere isolated by gel filtration and level off with time which may indicate an approach to a

divided into several tubes. Each tube was then incubated new equilibrium of PCNA on and off DNA as catalysed at 37 VC for a different length of time ranging from 0 to by residual RF-C. An alternative explanation for the 60 min before analysis through a second gel filtration biphasic dissociation of PCNA from DNA is that there

column. The results show a 72 min half-life for are two different populations of clamps on the DNA dissociation of  from DNA, and a 24 min half-life which differ in stability. For example, some PCNA

for PCNA dissociation from DNA. It is important to clamps may still be associated with RF-C. The observed note that the E. coli  complex is extracted from the half-life of PCNA dissociation from DNA is similar to

reaction by the first gel filtration column (Stukenberg et that of another study (t1/2 = 22 min.) (Podust et al.

al. 1991), but the human RF-C may not have been 1995). In this latter study, ATP--S was used to halt RF- removed. This is vital because these clamp loaders can C catalysed loading of PCNA, but the effect of ATP--

also catalyse the removal of their respective clamps from S on RF-C clamp unloading is uncharacterized and 106 Genes to Cells (1996) 1, 101–113 5 Blackwell Science Limited Comparison of DNA sliding clamps

Figure 6 Clamp loaders of E. coli and human replicases are also

clamp unloaders. (A) Radiolabelled PCNA was loaded onto

˜ Figure 5 Rate of decay of PCNA and rings from DNA at singly nicked plasmid in a 100 "L reaction containing 12 pmol

32 32

37 VC. (A) P-PCNA was loaded onto a nicked plasmid in a (120 nM) P-PCNA, 0.9 units of RF-C, and 1.2 pmol (12 nM)

32

"

200 L reaction containing 24 pmol (120 nM) P-PCNA, 1.8 singly nicked plasmid. After 10 min. at 37 VC, the reaction was gel

units of RF-C and 1.66 pmol (8 nM) singly nicked plasmid DNA. filtered on a Bio-Gel A15 m column at 4 VC equilibrated with

The mixture was incubated at 37 VC for 15 min. before applying buffer E containing 100 mM NaCl. Excluded peak fractions (10-

V to a Bio-Gel A15 m column at 4 C equilibrated with buffer E 13) were combined and then divided into two tubes (160 "L

containing 100 mM NaCl. The excluded peak fractions (nos 10– each). The two reactions were incubated 6 min at 37 VC either in

13) were pooled and then divided into five tubes of 100 "L each the presence or absence of 0.45 units of RF-C and then analysed

V and incubated at 37 C for the indicated amounts of time before for clamp removal on a second Bio-Gel A15 m column at 4 VC analysis on the second Bio-Gel A15 m column. (B) The equilibrated in buffer E containing 100 mM NaCl. (B) The

3 32

˜ ˜ experiment was repeated using 24 pmol (120 nM) H- and experiment was repeated using 384 nM  complex to load P-

1 pmol (5 nM)  complex in place of PCNA and RF-C. clamps on DNA, gel filtered, divided into two tubes, followed by

incubation for 10 min. at 37 VC either in the presence or absence

may not halt RF-C catalysed release of PCNA from of 192 nM  complex prior to a second gel filtration analysis at DNA. 4 VC.

containing 32P-PCNA on DNA were divided into

RF-C and  complex are also clamp unloaders

two tubes and one was treated with RF-C and ATP.

 We have observed that the  complex, besides loading Then both reactions were gel filtered a second time. on DNA, also unloads  clamps from DNA (Stukenberg The result showed that RF-C removed most of the

1993; Stukenberg et al. 1994). We have examined RF-C 32P-PCNA clamps from DNA. Panel B shows a similar

32 3

 for PCNA clamp unloading activity in Fig. 6. First, P- experiment in the E. coli system using H- and PCNA was loaded onto a nicked plasmid by RF-C complex. and ATP, and then gel filtered to remove unbound In both systems, this clamp unloading process is

32P-PCNA (data not shown). The column fractions dependent on the presence of ATP (data not shown). 5 Blackwell Science Limited Genes to Cells (1996) 1, 101–113 107 N. Yao et al.

A possible explanation of how the seemingly opposing actions of clamp loading and unloading can be performed by one protein complex, and the physio- logical relevance of this reaction are presented in the Discussion section.

T4 gp45 spontaneously cycles off DNA after replication is complete As was shown in Fig. 3, T4 DNA polymerase is required to stabilize the gp45 clamp on DNA through analysis by gel filtration. In the experiment described in Fig. 7A, the 3H-gp45 clamp was placed on a plasmid with a 1 kb gap of ssDNA coated with gp32 in the presence of the gp44/62 complex and the DNA polymerase. A complete set of dNTPs was added to to one half of this reaction to allow the gap to be filled and then both reactions were gel filtered side-by-side. The results show that the 3H-gp45 clamp was efficiently released upon completion of the gap.

These findings are in contrast to our previous demonstration that the  clamp is left on DNA after the polymerase finishes a template, although the polymerase itself dissociates into solution (Stukenberg et al. 1994). For comparison with the T4 system, an

32 experiment using P- and Pol III* on gapped DNA Figure 7 Fate of the T4 clamp after completing replication of a before and after replication is shown in Fig. 7B. template. (A) The T4 holoenzyme was assembled onto gapped

3 plasmid DNA using H-gp45 in a 100 "L reaction containing 3 12 pmol (120 nM) H-gp45, 3.6 pmol gp44/62 complex, 570

Discussion fmol gapped plasmid, 149 pmol gp32, 864 fmol (8.6 nM) gp43 and

V 60 "M each of dGTP and dCTP. After 5 min at 37 C, the

Stability of clamps in solution correlates with reaction was filtered. The excluded peak fractions (10–12) were their stability on DNA combined and 200 "L withdrawn into each of two tubes, one for

analysis of gp45 stability during idling, and the other for analysis This study shows that , a ring shaped protein composed after replication. Idling reaction: The T4 holoenzyme on DNA was

of two identical subunits, is a highly stable oligomer in incubated for 1 min at 37 VC before analysis on a second gel fil- solution and remains associated at concentrations as tration column. Termination reaction: The T4 holoenzyme on

low as 67 pM. The PCNA and gp45 clamps, composed DNA was supplemented with 60 "M each of dATP and dTTP,

V of three identical subunits, dissociate as their con- incubated for 1 min at 37 VC, and then analysed at 4 C on a second centration is reduced, but the PCNA trimer is an order Bio-Gel A15 m column equilibrated in buffer E containing all 4

3 of magnitude more stable than the gp45 trimer (Kd dNTPs. (B) The experiment was repeated using 3.6 pmol H-˜,

3

values, 21 nM and 200–400 nM, respectively). This 3.2 pmol H-Pol III*, 12.6 "g E. coli SSB in place of the T4

difference in stability probably underlies the ability proteins, and 1.34 pmol gapped plasmid DNA in a 200 "L reaction. to isolate ˜ and PCNA, but not gp45, on DNA by the non-equilibrium technique of gel filtration. This process may involve complete protomer dissociation

result is consistent with previously described results from the rings. of yeast PCNA (Burgers & Yoder 1993) and human Unlike PCNA and , the T4 gp45 ring is unstable

PCNA (Lee & Hurwitz 1990; Podust et al. 1995). both in solution and on DNA and requires the DNA

% The half-life of the  clamp on DNA is 72 min and polymerase to stabilize it on DNA through a gel

that of the PCNA ring is % 24 min. Spontaneous filtration column as observed previously (Richardson et dissociation of these clamps from DNA requires at least al. 1991). The relative instability of gp45 relative to 

one subunit–subunit interface to open, although the and PCNA suggests differences among these systems as 108 Genes to Cells (1996) 1, 101–113 5 Blackwell Science Limited Comparison of DNA sliding clamps

Figure 8 Models of clamp assembly and release in different systems. (A) In the E. coli and human system, the clamp loader assembles the clamp onto DNA (Step I). The clamp tethers the polymerase to DNA for processive synthesis (Step II). After replication is complete, the

polymerase dissociates from the clamp (Step III) and the clamp loader removes the clamp from DNA (Step IV). It is not yet clear whether RF-C remains associated with the PCNA clamp in Step I, or whether pol  dissociates from the clamp in Step III. (B) The gp44/62 complex assembles the gp45 trimer onto DNA (Step I), but the gp45 clamp requires association with the gp43 polymerase for enhanced stability on DNA (Step II). Upon termination of replication the polymerase dissociates from DNA and gp45 looses its gel filterable

stability on DNA (Step III).

illustrated in Fig. 8. The  and PCNA clamps are stable of three protomers. A trimer oligomer is entropically less

oligomers:  remains stably bound to DNA in the favoured than a dimer. Hence if the subunit-subunit absence of  complex (Stukenberg et al. 1991) and interfaces of dimer and trimer rings were of equivalent PCNA likely remains stable without RF-C as indicated strength, a trimer would tend to dissociate at higher

by its ability to slide off DNA upon linearizing the DNA concentration than a dimer. However, the crystal

(if RF-C were bound to PCNA and the 3’ terminus, structure shows the interface of  is significantly PCNA would not be released for sliding over the end of different from that of yeast PCNA. The  interfaces DNA). This is reflected in step I of Fig. 8A. Association are composed of an antiparallel sheet four residues in of DNA polymerase could occur even several minutes length whereas the sheet at the interface of yeast PCNA later (step II) as observed in these systems (O’Donnell is twice as long, presumably to counteract the effects of 1986; Burgers & Yoder 1993). The gp45 clamp is not as entropy (Krishna et al. 1994). It may be predicted that stable on DNA although it can exist for short periods of the forces that bind the gp45 trimer together will be less time without the clamp loader as indicated by DNA extensive than those in PCNA. crosslinking studies (Tinker et al. 1994) and a cryoelec- Why eukaryotes and T4 utilize trimer clamps rather

tron microscopy study which observed a structure on than a dimer like prokaryotes remains to be elucidated. DNA the size of  (presumably gp45) that disappeared In both T4 and eukaryotes the clamp is involved in seconds after removal of ATP (Gogol et al. 1992). processes other than replication. In T4 phage, the gp45 Hence, gp45 can exist alone on DNA, but must be clamp activates late gene transcription (Herendeen et al. quickly captured by the polymerase as reflected in 1989) and in mammals the p21 protein interacts with Fig. 8B, step II. The instability of the gp45 trimer in PCNA as a means of stopping long chain synthesis solution suggests that it may be assembled from (Flores-Rozas et al. 1994; Waga et al. 1994). PCNA is protomers to encircle the DNA as illustrated in also found with p21 in complex with cdk and cyclin D, Fig. 8B, step I. implying a role in cell cycle control (Xiong et al. 1992). PCNA has recently been shown to interact with DNase IV (also called FEN-1) in higher eukaryotes and RTH1 Are dimers inherently more stable than trimers?

in S. cerevisiae and stimulates its nucleolytic activity (Li et The one feature that most distinguishes the stable  ring al. 1995). DNase IV has been shown to play a critical

from the less stable PCNA and gp45 rings is its native role in maturation of Okazaki fragments in lagging oligomeric state;  forms a ring composed of two strand DNA synthesis (Robbin et al. 1994). These

protomers while PCNA and gp45 form rings composed observations of trimer clamps utilized by other proteins 5 Blackwell Science Limited Genes to Cells (1996) 1, 101–113 109 N. Yao et al.

may indicate a particular use of trimers in other cellular by the same protein complex. We presume that this processes such as transcription, repair and cell cycle apparent contradiction is explained by a dynamic control. For example, perhaps a metastabile trimer- equilibrium of clamps on and off the DNA catalysed monomer equilibrium is more amenable to control by these clamp loaders. If one starts with all the clamps processes than a clamp that exists as a stable dimer. off the DNA, the forward reaction of clamp loading is observed. If on the other hand, one starts with all the Two different molecular strategies of clamp clamps on the DNA (i.e. by placing clamps on DNA and then removing those left in solution by gel filtration) recycling then clamp unloading is observed. In both cases the end In replication of the leading strand, a DNA polymerase result is an equilibrium of clamps on and off the DNA, held tightly to DNA for continuous synthesis is but the direction in which the equilibrium is advantageous. However, the lagging strand is synthe- approached determines whether clamp loading or sized as a series of numerous fragments, and upon unloading is observed. completing each fragment the polymerase must rapidly In contrast to the E. coli and human system, the T4 cycle off the DNA and back onto the next primed site. phage system may use a different strategy of clamp This rapid cycling off and on the DNA is a process that recycling. The gp45 clamp spontaneously dissociates would conceptually be hindered by too tight a grip on from DNA after the template has been fully replicated, the DNA such as incurred by a protein ring. The suggesting that active clamp removal by the gp44/62 mechanism for this rapid polymerase cycling has been clamp loader may not be necessary as illustrated in elucidated in the E. coli and T4 phage systems. The Fig. 8B, step III. This result could be anticipated from DNA polymerase is tightly held to the sliding clamp the instability of gp45 clamps on DNA in the absence of during chain extension, but upon completing a template polymerase, and the fact that T4 polymerase rapidly the polymerase rapidly dissociates from its clamp dissociates from DNA upon completing a template

(Hacker & Alberts 1994a,b; Stukenberg et al. 1994). (Hacker & Alberts 1994b). However, the gel filtration

% In the E. coli system, the  clamp remains on the technique utilized in this report requires 15 min at completed DNA after the polymerase departs as 4 VC. This is a long time relative to the speed of fork

illustrated in step III of Fig. 8A. Although the stoichio- movement. If gp45 were bound to DNA for a fraction

metric use of  clamps for lagging strand fragments is of this time it may still require assistance for efficient consistent with the intracellular abundance of  over Pol removal from DNA. Hence, dissociation of gp45 clamps

III, there are still 10 times more Okazaki fragments than from DNA assisted by the gp44/62 complex in vivo can

 clamps. Hence these clamps must be recycled. The not be rigorously excluded.

stability of  on DNA implies clamp unloading will be

 an active process. Indeed, the  complex removes clamps from DNA as illustrated in step IV of Fig. 8A. Experimental Procedures The observed stability of PCNA clamps on DNA suggests that clamp recycling may be an active process in Sources eukaryotes as well. A simple calculation supports this Radioactive chemicals, DuPont-New England Nuclear; idea. The amount of PCNA in the cell nucleus (HeLa) unlabelled ATP Pharmacia LKB; Bio-Gel A-15 m, Bio-Rad;

has been estimated by immunoblot to be in the 3– DNA modification enzymes and the T4 DNA polymerase, New

5 5

 

6 10 range (as monomer) [1–2 10 trimers] England BioLabs (NEB) (specific activity = 3 units/"g where 1 (Morris & Mathews 1989). The total number of unit supports incorporation of 10 nmol dNTP in 30 min at 37 VC) Okazaki fragments in replication of four billion base and GIBCO BRL; gp32, Boehringer Mannheim. (Indianapolis,

7

IN). Other proteins were purified as described:  (Kong et al.

pairs, assuming a 200 nucleotide length, is 2  10

PK

 fragments. Hence PCNA must be reused about 100 1992),  (Stukenberg et al. 1994), complex (Onrust et al.

7 5 1995), E. coli single-stranded DNA binding protein (SSB)  times per cell cycle (2  10 /2 10 ). Assuming an S phase of 6 h, the time of a PCNA cycling event is (Weiner et al. 1975), PCNA (Kelman 1995), PCNA with an

N-terminal kinase site (Kelman et al. 1995c) and EBNA-1 DNA

% 3.6 min, much faster than the observed stability of

binding domain (Kelman et al. 1995a). RF-C and Pol  were

PCNA on DNA (t1/2 % 24 min). This report shows that purified from HeLa cells as described (Lee et al. 1991b). 3H- RF-C, like  complex, efficiently removes PCNA PCNA (520 cpm/fmol) and 3H-gp45 (75 cpm/fmol) were clamps from DNA. prepared by reductive methylation using formaldehyde and [3H]- 3 It seems a dilemma that the seemingly opposed NaBH4 (Kelman et al. 1995a). H-gp45 was within 90% the

actions of clamp loading and unloading are performed activity of wild-type gp45 in replication of gp32-coated primed 110 Genes to Cells (1996) 1, 101–113 5 Blackwell Science Limited

Comparison of DNA sliding clamps

ssDNA using gp44/62 and gp43. Modified versions of  resuspended in buffer A [with Tris-HCl (pH 8.8)] and dialysed

(Stukenberg et al. 1994) and PCNA (Kelman et al. 1995c) against 2  4 L buffer A [with Tris-HCl (pH 8.8)] overnight at containing protein kinase recognition motifs at the C- and N- 4 VC. The solution (108 mg in 25 mL) was applied to an 8 mL

32 terminus, respectively, were labelled with - P-ATP using mono Q column equilibrated in buffer A (but at pH 8.8), and cAMP-dependent protein kinase to specific activities of 65-387 eluted with a 160 mL gradient of 25-250 mM NaCl in buffer A cpm/fmol and 225 cpm/fmol, respectively, as described (pH 8.8). Fractions 7-66 containing > 95% pure gp45 were

(Kelman et al. 1995a). 3H-PCNA and 32P-PCNA were within pooled (120 mg in 62 mL), dialysed against 4 L buffer A 50% and 99%, respectively, the activity of wild-type PCNA in overnight, aliquoted and stored at -70 VC.

replication of gp32 coated poly(dA)5000: oligo(dT)20 and the The gp44/62 complex was further purified upon resuspending elongation of singly primed M13 DNA using RF-C and pol . the ammonium sulphate pellet in 250 mL buffer B (pH 7.5),

3 32 PK

˜  H- and P- were within 95% the activity of wild-type in dialysed against 2 changes of 4 L each of buffer B and then applied replication of SSB-coated primed M13mp18 ssDNA using to a 100 mL heparin-Sepharose column. The gp44/62 was eluted pol III*. with a 1.4 L linear gradient of 0-0.5 M NaCl gradient in buffer B. Fractions 76-105 were pooled (100 mg in 250 mL) and precipitated upon addition of ammonium sulphate to 70% Buffers saturation. Protein was dissolved in buffer A, dialysed against 4 L

Buffer A is 20 mM Tris-HCl (pH 7.5), 0.5 mM EDTA (pH 7.5), of buffer A overnight, and applied to an 8 mL MonoQ column 2mM dithiothreitol (DTT) and 20% glycerol. Buffer B is 10 mM equilibrated in buffer A. The gp44/62 complex was eluted with a sodium acetate (pH 7.5), 20% glycerol, 0.5 mM EDTA and 2 mM 160 mL linear gradient from 25 to 250 mM NaCl in buffer A.

DTT. Buffer C is 50 mM sodium acetate, 0.3 M NaCl, 10 mM zinc Fractions 17-36 containing gp44/62 complex were pooled sulphate. Buffer D is 20 mM Tris-HCl (pH 7.5), 0.5 mM EDTA, (61.7 mg, 45 mL), dialysed against buffer A and stored at -70 VC.

100 mM NaCl and 10% glycerol. Buffer E is 20 mM Tris-HCl (pH 7.5), 0.1 mM EDTA, 40 "g/mL bovine serum albumin (BSA), DNA substrates

5mM DTT, 8 mM MgCl2, 4% glycerol. Plasmid DNAs [pET-˜16 (a pET3c vector containing the 1 kb dnaN gene) (5.7 kb), pGEMoriP (5 kb) (Frappier & O’Donnell Purification of gp45 and gp44/62 complex 1991) and pBluescript SK+ (Stratagene, LaJolla, CA) (2.96 kb)] The plasmid, pTL151 W containing genes 45, 44 and 62, under were purified through Qiagen columns (Qiagen Inc., Chats- worth, CA), and then were further purified by ultracentrifuga- control of a PL promoter, was the gift of Dr William Konigsberg,

Yale University. HB101 cells (24 L) harboring pTL151 W were tion in cesium chloride as described (Maniatis et al. 1982). Nicked

grown in VAT (per litre: 10 g tryptone, 4 g yeast extract, 1 g DNA was prepared by treating 24.2 "g of supercoiled (RFI)

plasmid with 60 units of S1-nuclease in 400 "L of buffer C at

KH2PO4, and 10 g K2HPO4) supplemented with glucose and

37 VC until 50% of the RFI DNA was converted to RFII DNA (7 V ampicillin at 30 VC and induced at 42 C as described (Rush et al.

1989) Each litre was grown to OD = 0.8 and then brought to min) and then was purified by phenol extraction. Plasmid DNA

+

with a ssDNA gap was made using 390 "g of pBluescript SK V 42 VC rapidly upon addition of 250 mL of VATat 62 C followed

plasmid DNA followed by treatment with gpII protein as by continued incubation at 42 VC for 3 h. The following

described (Meyer & Geider 1979) to produce a site specific procedures were performed at 4 VC. Cells were harvested by

centrifugation and lysed with lysozyme by heat-lysis as described nick in 60% of the plasmid. The nicked DNA was then treated

(Wickner 1974). The clarified lysate was dialysed against buffer A with 1000 units exonuclease III at 37 VC for 10 min. A gap size of (3 g of protein in 250 mL) and then applied to a 150 mL fast-flow 1 kb was estimated by analysis of 3 "g in an alkaline agarose gel. Q-Sepharose column equilibrated in buffer A and eluted with a 2.1 L gradient of 0-0.5 M NaCl in buffer A. Presence of the Replication assays proteins was followed in 10% SDS-polyacrylamide gels stained with Coomassie blue. The gp44/62 complex and gp45 co-eluted M13mp18 ssDNA was purified as described (Turner & (at about 0.25 M NaCl) and were pooled (12 mL), dialysed against O’Donnell 1995b) and then uniquely primed with a synthetic buffer B, and applied to a 100 mL heparin-Sepharose column DNA 30-mer oligonucleotide (M13mp18 map position 6816-

equilibrated in 10 mM sodium acetate (pH 7.5). gp45 was 6847) as described (Stukenberg et al. 1991). Assays contained collected in the flow through and gp44/62 complex was eluted 80 ng primed ssDNA, 2.5 "g gp32, 10 ng gp44/62, 270 ng gp45

3

" with 2 M NaCl in 10 mM sodium acetate (300 mg). The flow or H-gp45, 0.1-0.3 units of gp43, 60 "M dCTP and 60 M

V through fraction, containing gp45 was slowly adjusted to pH 5.5 dGTP in 23 "L of buffer E. Reactions were incubated at 37 C

32

"  with 10 mM glacial acetic acid and then applied to a 100 mL for 1 min., and then 60 "M dATP and 20 M [ - P]TTP were heparin-Sepharose column equilibrated in buffer B (pH 5.5), and added to initiate a 30-s pulse of replication before quenching the then eluted with a 1.4 L linear gradient of 0-0.5 M NaCl in Buffer reaction by spotting onto DE81 paper and quantitating the B. Fractions containing gp45 (fractions 25-45, 150 mg in 180 mL) amount of DNA synthesis by scintillation counting.

. were pooled and precipitated upon addition of ammonium Replication assays of PCNA contained 0.125 "g (dA)4500 oli-

sulphate to 70% saturation. After centrifugation, the pellet was go(dT)12-18 annealed at nucleotide ratio of 20:1, 40 mM creatine 5 Blackwell Science Limited Genes to Cells (1996) 1, 101–113 111

N. Yao et al.

 phosphate, 0.03 mg/mL creatine phosphate kinase, 1 "g gp32, catalyzed by polymerase holoenzyme. Proc. Natl Acad. Sci.

32

 " 33.3 "M [ - P]dTTP,0-0.1 g PCNA, 0.09 unit RF-C and 0.2 USA 91, 8655–8659.

" units of pol  in 30 Lof40mMTris-HCl (pH 7.8), 7 mM MgCl2, Frappier, L., and O’Donnell, M. (1991) Overproduction, 1mM DTT, 4 mM ATP and 0.15 mg/mL BSA. The reaction was purification, and characterization of EBNA-1, the origin

binding protein of Epstein-Barr virus. J. Biol. Chem. 266, incubated at 37 VC for 1 h before quenching the reaction by 7819–7826. spotting on DE81 paper and quantitating the amount of DNA Gogol, E.P., Young, M.C., Kubasek, W.L., Jarvis, T.C. & von synthesis by scintillation counting. Hippel, P.H. (1992) Cryoelectron microscopic visualization of functional subassemblies of the bacteriophage T4 DNA

replication complex. J. Mol. Biol. 224, 395–412. Gel filtration of PCNA,  and gp45 Hacker, K.J., and Alberts, B.M. (1994a) The rapid dissociation of the T4 DNA polymerase holoenzyme when stopped by a

Radiolabelled proteins (amounts indicated in legends to the DNA hairpin helix. J. Biol. Chem. 269, 24221–24228.

V figures) were incubated in 100 "L of buffer D at 37 C for 1.5 h. Hacker, K.J. & Alberts, B.M. (1994b) The slow dissociation of the The protein mixture was analysed by gel filtration on a Fast T4 DNA polymerase holoenzyme when stalled by nucleotide

Protein Liquid Chromatography FPLC HR 10/30 Superose 12 omission. An indication of a highly processive enzyme. J. Biol.

column (Pharmacia-LKB) equilibrated in buffer D at 4 VC. After Chem. 269, 24209–24220.

" the first 6 mL, fractions of 150 "L were collected and 100 Lof Herendeen, D.R., Kassavetis, G.A., Barry, J., Alberts, B.M. & the indicated fractions were analysed in a 12% SDS polyacryla- Geiduschek, E.P. (1989) Enhancement of bacteriophage T4 mide gel either stained with Coomassie Brilliant Blue or late transcription by components of the T4 DNA replication apparatus. Science 245, 952–958. subjected to fluorography using EN3HANCE (NEN, Boston, Jarvis, T.C., Paul, L.S. & von Hippel, P.H. (1989) Structural and MA) and then exposed to X-ray film. enzymatic studies of the T4 DNA replication system. J. Biol. Chem. 264, 12709–12716.

Kaboord, B.F. & Benkovic, S.J. (1995) Accessory proteins Gel filtration of PCNA,  and gp45 on DNA function as matchmakers in the assembly of the T4 DNA polymerase holoenzyme. Curr. Biol. 5, 149–157.

Proteins and DNA (amounts are specified in the figure legends) Kelman, Z. (1995) Interaction Between Dna Polymerase and Other

V were incubated in 100 "L of buffer E + 1 mM ATP at 37 C for the Cellular Proteins. New York, NY: Cornell University Graduate

indicated time before applying to columns of agarose Bio-Gel School of Medical Sciences.

A15 m (5 mL bed volume) equilibrated in buffer E at 4 VC. Kelman, Z. and O’Donnell, M. (1994) DNA replication- Fractions of 180 "L were collected and the amount of clamp in enzymology and mechanisms. Curr. Opin. Genet. Dev. 4, each fraction was quantifyd by liquid scintillation. When the T4 185–195.

Kelman, Z. & O’Donnell, M. (1995) Structural and functional DNA polymerase (gp43) was present, 60 "M each of dCTP and

dGTP was added to the column buffer to prevent removal of the similarities of prokaryotic and eukaryotic DNA polymerase

H H sliding clamps. Nucl. Acids Res. 23, 3613–3620. DNA primer by the proofreading 3 –5 exonuclease of gp43. Kelman, Z., Naktinis, V. & O’Donnell, M. (1995a) Radiolabel- ling of proteins for biochemical studies. Meth. Enzymol. 262, 430–442. Acknowledgements Kelman, Z., Finkelstein, J. & O’Donnell, M. (1995b) Why have We are grateful to Dr William Konigsberg for providing us with six-fold symmetry? Curr. Biol. 5, 1239–1242. the pTL151 W plasmid. This work was supported by a grant from Kelman, Z., Yao, N. & O’Donnell, M. (1995c) Esherichia coli the National Institutes of Health, GM38839. expression vectors containing a protein kinase-recognition motif, His6-tag and hemagglutinin epitope. Gene 166, 177–178 Kong, X.-P., Onrust, R., O’Donnell, M. & Kuriyan, J. (1992) References Three-dimensional structure of the b subunit of E. coli DNA polymerase III holoenzyme: a sliding DNA clamp. Cell 69, Burgers, P.M.J. (1991) Saccharomyces cerevisiae Replication Factor 425–37. C: II. Formation and activity of complexes with the Kornberg, A. & Baker, T. (1992) DNA Replication, 2nd edn. New

proliferating cell nuclear antigen and with DNA polymerase York: W.H. Freeman & Co.

  and . J. Biol. Chem. 266, 22698–22706. Krishna, T.S., Kong, X.P., Gary, S., Burgers, P.M. & Kuriyan, J. Burgers, P.M.J. & Yoder, B.L. (1993) ATP-independent loading (1994) Crystal structure of the eukaryotic DNA polymerase of the proliferating cell nuclear antigen requires DNA ends. J. processivity factor PCNA. Cell 79, 1233–43. Biol. Chem. 268, 19923–19926. Kuriyan, J. & O’Donnell, M. (1993) Sliding clamps of DNA Capson, T.L.,Benkovic, S.J. & Nossal, N.G. (1991) Protein-DNA polymerases. J. Mol. Biol. 234, 915–25. cross-linking demonstrates stepwise ATP-dependent assembly Lee, S.-H. & Hurwitz, J. (1990) Mechanism of elongation of of the T4 DNA polymerase and its accessory proteins on the primed DNA by DNA polymerase d, proliferating cell nuclear primer-template. Cell 65, 249–258. antigen, and activator-1. Proc. Natl Acad. Sci. USA 87, 5672– Flores-Rozas, H., Keiman, Z., Dean, F.B., Pan, Z.-Q., Harper, 5676. J.W., Elledge, S.J., O’Donnell, M. & Hurwitz, J. (1994) Cdk- Lee, S.-H., Pan, Z.-Q., Kwong, A.D.,Burgers, P.M.J. & Hurwitz, interacting protein-1 (Cip1, WAF1) directly binds with J. (1991a) Synthesis of DNA by DNA polymerase e in vitro.

proliferating cell nuclear antigen and inhibits DNA replication J. Biol. Chem. 266, 22707–22717. 112 Genes to Cells (1996) 1, 101–113 5 Blackwell Science Limited

Comparison of DNA sliding clamps Lee, S.-H., Kwong, A.D., Pan, Z.-Q. & Hurwitz, J. (1991b) DNase IV and the 5H-nuclease domain of Esherichia coli DNA Studies on the activator 1 protein complex, an accessory factor polymerase I. J. Biol. Chem. 269, 28535–28538. for proliferating cell nuclear antigen-dependent DNA poly- Sinha, N.K., Morris, C.F.& Alberts, B.M. (1980) Efficient in vitro merase d. J. Biol. Chem. 266, 594–602. replication of double-stranded DNA templates by a purified Li, X., Li, J., Herrington, J., Licker, M.R. & Burgers, P.M.J. T4 bacteriophage replication system. J. Biol. Chem. 263, 6561– (1995) Lagging strand DNA synthesis at the Eukaryotic 6569. Replication Fork involves binding and stimulation of Stukenberg, P.T.(1993) The dynamics of E. coli DNA polymerase III FEN-1 by PCNA. J. Biol. Chem. 270, 22109–22112. holoenzyme in an in vitro lagging stand model system. PhD thesis, Maki, S. & Kornberg, A. (1988) DNA polymerase III Cornell University Medical College, New York, NY. holoenzyme of Escherichia coli. II. A novel complex including Stukenberg, P.T., Studwell-Vaughan, P.S. & O’Donnell, M. the g subunit is essential for processive synthesis. J. Biol. Chem. (1991) Mechanism of the sliding b clamp of DNA polymerase 263, 6555–6560. III holoenzyme. J. Biol. Chem. 266, 11328–11334. Maniatis, T., Fritsch, E.F. & Sambrook, J. (1982) Molecular Stukenberg, P.T., Turner, J. & O’Donnell, M. (1994) An Cloning: A Laboratory Manual. Cold Spring Harbor, NY: Cold explanation for lagging strand replication: polymerase hopping Spring Harbor Laboratory. among DNA sliding clamps. Cell 78, 877–87. Meyer, T.F. & Geider, K. (1979) Bacteriophage fd gene II- Tinker, R.L., Kassavetis, G.A. & Geiduschek, E.P. (1994) protein. II. Specific cleavage and relaxation of supercoiled Detecting the ability of viral, bacterial and eukaryotic FR from filamentous phages. J. Biol. Chem. 254, 12642– replication proteins to track along DNA. EMBO J. 13, 12646. 5330–5337. Morris, G.F.& Mathews, M.B. (1986) Regulation of proliferating Turner, J.T. & O’Donnell, M. (1995b) Cycling of E. coli DNA cell nuclear antigen during the cell cycle. J. Biol. Chem. 264, polymerase III from one sliding clamp to another-A model for 13856–13864. the lagging strand. Meth. Enzymol. 262, 426–430. O’Donnell, M. (1987) Accessory proteins bind a primed template Weiner, J.H., Bertsch, L.L. & Kornberg, A. (1975) The and mediate rapid cycling of DNA polymerase III holoenzyme deoxyribonucleic acid unwinding protein of Escherichia coli: from Escherichia coli. J. Biol. Chem. 262, 16558–16565. Properties and functions in replication. J. Biol. Chem. 250, Onrust, R., Finkelstein, J., Naktinis, V., Turner, J., Fang, L. & 1972–1980. O’Donnell, M. (1995a) Assembly of a chromosomal replica- Wickner, S. (1976) Mechanism of DNA elongation catalyzed by tion machine: Two DNA polymerases, a clamp loader, and Escherichia coli DNA polymerase III, DnaZ protein, and DNA sliding clamps in one holoenzyme particle, I. Organization of elongation factors I and III. Proc. Natl Acad Sci USA 73, 3511– the clamp loader. J. Biol. Chem. 270, 13348–13357. 3515. Onrust, R., Finkelstein, J., Turner, J., Naktinis, V. & O’Donnell, Wickner, W. & Kornberg, A. (1974) A holoenzyme form of M. (1995b) Assembly of a chromosomal replication machine: deoxyribonucleic acid polymerase III: Isolation and properties. two DNA polymerases, a clamp loader, and sliding clamps in J. Biol. Chem. 249, 6244–6249. one holoenzyme particle III. Interface between two poly- Waga, S., Hannon, G.J., Beach, D. & Stillman, B. (1994) The p21 merases and the clamp loader. J. Biol. Chem. 270, 13366– inhibitor of cyclin-dependent kinases controls DNA replica- 13377. tion by interaction with PCNA. Nature 369, 574–578. Podust, V.N., Podust, L.M., Muller, F. & Hu¨bscher, U. (1995) Xiong, Y., Zhang, H. & Beach, D. (1992) D type cyclins associate DNA polymerase delta holoenzyme: action on single-stranded with multiple protein kinases and the DNA replication and DNA and on double-stranded DNA in the presence of repair factor PCNA. Cell 71, 505–514. replicative DNA helicases. Biochemistry 34, 5003–5010. Zhang, P., Zhang, S.-J., Zhang, Z., Woessner, J.F. & Lee, Richardson, R.W., Ellis, R.L. & Nossal, N.G. (1991) Protein- M.Y.W.T. (1995) Expression and physiochemical characteriza- protein interactions within the bacteriophage T4 DNA tion of human proliferating cell nuclear antigen. Biochemistry replication complex. In: Molecular Mechanisms in DNA 34, 10703–10712. Replication and Recombination (eds A.B. Author). Alan R. Liss Inc. Robins, P., Pappin, D.J.C., Wood, R.D. and Lindahl, T. (1994) Received: 28 September 1995

Structural and functional homology between mammalian Accepted: 16 October 1995 5 Blackwell Science Limited Genes to Cells (1996) 1, 101–113 113