Maturation of Bacteriophage T4 Lagging Strand Fragments Depends on Interaction of T4 RNase H with T4 32 Protein Rather than the T4 Gene 45 Clamp*

In the bacteriophage T4 DNA replication system, T4 RNase H removes the RNA primers and some adjacent DNA before the lagging strand fragments are ligated. This 5′-nuclease has strong structural and functional similarity to the FEN1 nuclease family. We have shown previously that T4 32 protein binds DNA behind the nuclease and increases its processivity. Here we show that T4 RNase H with a C-terminal deletion (residues 278–305) retains its exonuclease activity but is no longer affected by 32 protein. T4 gene 45 replication clamp stimulates T4 RNase H on nicked or gapped substrates, where it can be loaded behind the nuclease, but does not increase its processivity. An N-terminal deletion (residues 2–10) of a conserved clamp interaction motif eliminates stimulation by the clamp. In the crystal structure of T4 RNase H, the binding sites for the clamp at the N terminus and for 32 protein at the C terminus are located close together, away from the catalytic site of the enzyme. By using mutant T4 RNase H with deletions in the binding site for either the clamp or 32 protein, we show that it is the interaction of T4 RNase H with 32 protein, rather than the clamp, that most affects the maturation of lagging strand fragments in the T4 replication system in vitro and T4 phage production in vivo.

DNA synthesis on the lagging strand of the replication fork is accomplished by the rapid repetition of a cycle in which primase makes a short RNA primer that is elongated by polymerase; the primer is removed from the previous fragment, and the two fragments are joined by DNA ligase. The efficient sealing of adjacent fragments is essential to maintain the accuracy of DNA replication. The accumulation of nicks and gaps on the lagging strand ultimately results in double-stranded breaks, increased mutation frequencies, and cell lethality (reviewed in Refs. 1 and 2). The lagging strand cycle must be repeated every few seconds because the discontinuous fragments are so short, 1-2 kb in prokaryotes and less than 200 bases in eukaryotes. In each of the replication systems studied, the primers are removed by a member of a family of 5Ј-nucle-ases with conserved sequences and similar structures (Fig. 1). It is important that the process of primer removal from one fragment is coordinated with the elongation of the next fragment to ensure rapid and accurate replication. Our studies indicate that the mechanism by which lagging strand polymerization and primer removal are coordinated in bacteriophage T4 replication is different from that used in eukaryotes.
The phage T4 member of the FEN1 nuclease family was called T4 RNase H because it hydrolyzed the RNA strand in an RNA:DNA hybrid, as expected for an enzyme removing the RNA primers. However, it also acts as a 5Ј-nuclease on DNA duplexes (3). Genetic studies indicate that either T4 RNase H or the 5Ј-to 3Ј-exonuclease of the Escherichia coli host DNA polymerase I is necessary for phage production (4). A T4 mutant with a large deletion (⌬118 -305) in the rnh gene gives a burst size of 50% of wild type T4 phage in a wild type host, but a burst of only a few phage per infected cell under nonpermissive conditions in E. coli PolA12, which has a conditionally lethal mutation in the host nuclease. Short DNA fragments accumulate, consistent with a defect in removing the primers from lagging strand fragments that prevents ligation of adjacent fragments. Phage production is restored by supplying T4 RNase H on a plasmid. The T4 rnh deletion mutant is also hypersensitive to UV irradiation and to antitumor agents that induce T4 topoisomerase cleavage products (5) and is defective in DNA homing (6). Processing of the RNA transcript that serves as the primer for leading strand synthesis at the T4 uvsY origin appears to be impaired in the T4 rnh mutant (7).
The 5Ј-nucleases in this family have both a 5Ј-exonuclease activity that degrades RNA:DNA and DNA:DNA duplexes, giving short oligonucleotide products, and a flap endonuclease activity that cuts close to the junction of single-and doublestranded DNA on fork and flap substrates (reviewed in Ref. 1). The relative strength of these two activities differs within the family, with the exonuclease stronger in the phage enzymes (T4 RNase H (8) and T5 5Ј to 3Ј-exonuclease (9)) and the flap endonuclease stronger in the FEN proteins (10). The 5Ј-exonuclease activity of T4 RNase H is nonprocessive, removing a single oligonucleotide (predominantly dimers and trimers) each time it binds its substrate. On substrates where the T4 gene 32 ssDNA 1 -binding protein can bind behind the nuclease, its processivity is increased, so that a total of about 10 short oligonucleotides are hydrolyzed at each binding. However, the flap endonuclease of T4 RNase H is inhibited when 32 protein binds to the single-stranded flap (8). Similarly, FEN1 cutting of flaps long enough to bind RPA, the eukaryotic counterpart of 32 protein, is inhibited when RPA is present (1).
In the bacteriophage T4 replication system, the polymerase is held on the primer by the gene 45 clamp, which is loaded by the 44/62 clamp loader complex. T4 32 protein plays a major role in orchestrating the lagging strand cycle by increasing primer synthesis, promoting loading of the clamp by the clamp loader, and increasing the processivity of both polymerase and RNase H (11,12). We have previously shown that T4 RNase H removes the RNA primers and about 30 nucleotides of adjacent DNA from each lagging strand fragment during DNA replication in vitro, and that it is the 5Ј-exonuclease activity, rather than the flap endonuclease, that is responsible for most of this digestion (13). Our studies indicated that, on most molecules, polymerase filled in the gap between adjacent fragments before the nuclease could bind for a second round of degradation. The amount of DNA removed along with the primers was similar to the DNA removed during a single binding by T4 RNase H, when 32 protein was behind it. Thus our studies were consistent with a model in which the extent of degradation was controlled by the difference in rates of digestion by T4 RNase H and synthesis by polymerase, when 32 protein covered the single-stranded DNA between them.
In eukaryotic DNA replication, both FEN1 nuclease and the nuclease activity of the Dna2 helicase-nuclease have roles in removing primers from eukaryotic lagging strand fragments (reviewed in Refs. 1 and 14). Recent studies (15)(16)(17)(18) indicate that, in contrast to T4 RNase H, FEN1 uses its flap endonuclease to remove the primer and adjacent DNA, after a flap is created by the polymerase extending the upstream fragment. The rate of strand displacement synthesis by Saccharomyces cerevisiae pol ␦ with the PCNA clamp and replication factor C clamp loader is increased by FEN1, so that these four proteins together catalyze efficient nick translation. The interaction between FEN1 and the PCNA replication clamp is clearly important, because there was less strand displacement synthesis with a FEN1 protein with a mutation in the C-terminal PCNA interaction site (16).
In this paper we show that T4 RNase H, like FEN1, is stimulated by its replication clamp, the T4 gene 45 protein.
However, in contrast to FEN1, T4 RNase H interaction with the 45 clamp is not required for the normal processing of lagging strand fragments. Instead, it is the T4 RNase H interaction with 32 protein that is essential. T4 RNase H interacts with the clamp through a conserved clamp-binding motif at the N terminus of the nuclease. A C-terminal helical bundle at the C terminus of the nuclease is needed for its stimulation by 32 protein. An N-terminal deletion in T4 RNase H that prevents stimulation by the clamp does not decrease fragment sealing in the T4 replication system in vitro. Plasmid encoding this mutant T4 RNase H can replace the wild type in restoring production of T4 phage with a disrupted rnh gene. C-terminal deletions that abolish T4 RNase H interaction with 32 protein strongly impair fragment maturation in vitro and fail to restore T4 ⌬rnh mutant phage production in vivo.

EXPERIMENTAL PROCEDURES
DNA Substrates-Oligonucleotides were made and reverse phasepurified by Sigma-Genosys, except that oligonucleotides longer than 50 bases were gel-purified. The 3Ј or 5Ј end-labeled partial duplexes were made by annealing an 84-mer DNA complementary to nucleotides 6198 -6281 of M13mp19 to the viral single-stranded DNA, as described previously (13). Nicked or gapped molecules were made as described (13) (20). Wild type 32 protein was purified as described (21). The truncated 32 proteins, 32-A and 32-B, were the generous gift of David Geidroc (22,23). T4 DNA ligase was obtained from U. S. Biochemical Corp.
Nuclease and Polymerase Assays-Unless otherwise indicated, reaction mixtures (10 l) contained 1.0 nM substrate, 25 mM Tris acetate, pH 7.5, 63 mM potassium acetate, 6 mM magnesium acetate, 20 mM dithiothreitol, 1 mM EDTA, and 200 g/ml bovine serum albumin. The concentrations of wild type and mutant T4 RNase H are indicated in the figure legends. When present, gene 32 single-stranded DNA-binding protein was 1 M; T4 DNA polymerase, gene 45 clamp protein (trimer), and gene 44/62 (4:1 complex) clamp loader were 60, 160, and 240 nM, respectively; and T4 DNA ligase was 200 Weiss units/ml, unless otherwise indicated. In experiments that included polymerase, clamp, and clamp loader, ATP was present at 1 mM and each dNTP at 250 M. Unless otherwise indicated, reaction mixtures without T4 RNase H, ligase, or DNA polymerase were incubated for 2 min at 30°C, and the reaction was begun by the addition of the nuclease, or a mixture of the nuclease, ligase, and polymerase, as noted in the figures. Aliquots were taken at the times indicated, and the reaction was stopped by addition of 1.5 volumes of a solution of 83% (v/v) formamide, 0.01% xylene cyanol and bromphenol blue, and 33 mM EDTA. Products were heated for 3 min at 95°C before electrophoresis on polyacrylamide (19:1), 7 M urea gels of the percentage indicated in the figure legends. Gels were exposed to Kodak Biomax MR film or were scanned and quantified with a Fujifilm FLA 3000 PhosphorImager and Fuji Multigauge software.
In the experiments shown in Figs. 2B and 10, 5-l aliquots of the reaction mixtures were removed at the indicated times and then heated for 20 min at 60°C to inactivate the enzymes. BstNI endonuclease (2 units) (New England Biolabs) was then added, and the incubation at 60°C continued for 30 min, before adding 7 l of the formamide stop solution.
Coupled Leading and Lagging Strand Synthesis-The reaction mixtures (10 l) contained 1.6 nM singly nicked pUCNICK plasmid (2.7 kb), 2 mM ATP, 250 M of each dNTP including [␣-32 P] dTTP (ϳ1800 cpm/ pmol), 250 M CTP, GTP, and UTP, 25 mM Tris acetate, pH 7.5, 60 mM potassium acetate, 6 mM magnesium acetate, 10 mM dithiothreitol, and 20 g/ml bovine serum albumin. The protein concentrations were 2 M 32 ssDNA-binding protein, 328 nM 41 helicase, 30 nM wild type DNA polymerase, 242 nM 44/62 clamp loader, 162 nM 45 clamp, 95 nM 59 helicase loading protein, and 64 nM 61 primase. When indicated, RNase H was 100 nM, and DNA ligase was 200 Weiss units/ml. Reaction mixtures without polymerase, primase, helicase, RNase H, and DNA ligase were incubated for 2 min at 37°C, and synthesis was begun by the addition of polymerase, primase, and helicase. RNase H and DNA ligase were added 1 min later. At the times indicated, aliquots of the reaction mixtures were mixed with an equal volume of 0.2 M EDTA to stop the synthesis, and the products were analyzed by 0.6% alkaline agarose gel electrophoresis (24) and trichloroacetic acid precipitation (20).
Point mutations in the N-terminal region of T4 RNase H were made by site-directed mutagenesis of the wild type gene in the plasmid pV- Purification of the Mutant Proteins-All of the plasmids were transformed into E. coli BL21(DE3)pLysS (26) for expression of the proteins. With the exception of ⌬C-(278 -305), the mutant proteins were induced with isopropyl thioglucoside for 2 h at 37°C, as described for the wild type protein (20), and partially purified by the small scale procedure described in Ref. 27. For ⌬C-(278 -305), plasmid pMB5002 in BL21(DE3)pLysS was grown in Luria Broth with 50 g/ml of carbenicillin (Invitrogen) at 24°C to A 600 ϭ 0.4 in a 20-liter New Brunswick Scientific BioFlo 3000 fermentor, and protein synthesis was induced by addition of 1 mM isopropyl thioglucoside. The cells (45 g) were harvested after overnight induction, resuspended in 50 mM Tris-Cl, pH 7.5, 500 mM NH 4 Cl, 10 mM MgCl 2 , 2 mM dithiothreitol, and 0.1 mM 4-(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride, and broken by sonication, and the cell lysate was clarified by centrifugation at 100,000 ϫ g at 4°C. T4 RNase H was purified from the supernatant by chromatography first on SP-Sepharose (Amersham Biosciences) and then on Poros-S (Perspective Biosystems) by using linear gradients formed from PC buffer A (50 mM Tris-Cl, pH 8.0, 100 mM NH 4 Cl, 10 mM MgCl 2 , 1 mM dithiothreitol, 1 mM EDTA, and 25 g/ml 4-(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride) and PC buffer B (PC, buffer A containing 750 mM NaCl).
Complementation of Bacteriophage T4 ⌬rnh by Plasmids Encoding Wild Type and Mutant T4 RNase H-T4 ⌬rnh (4) has a deletion of residues 118 -305 in the gene encoding T4 RNase H. The host E. coli MIC2003, an rnhA339::catpolA12 derivative of E. coli FB2 (28), has an interruption in the gene for RNase HI and a temperature-sensitive mutation in DNA pol I affecting its 5Ј-nuclease. The host also contained pGP1.2 (29), a plasmid with the gene for T7 RNA polymerase under the temperature controlled P L promoter, as well as a compatible plasmid with T4 RNase H under control of the T7 promoter as follows: pNQ1004 (wild type), pNQ1101 (⌬N-(⌬2-10)), pMB5002 (⌬C-(⌬278 -305)), pGO1701 (⌬C-(⌬286 -305)), pGO1801(⌬C-(⌬295-305)), or the pVex11 vector (see Refs. 3 and 4 and this paper). Cells were grown in LB media with 60 g/ml kanomycin and 50 g/ml carbenicillin to 1 ϫ 10 8 /ml at 30°C, shifted to 43°C for 15 min to induce production of T7 RNA polymerase and disrupt the polA12 DNA polymerase, and then infected with wild type T4D or T4 ⌬rnh at a multiplicity of 0.5 phage/bacteria. Infective centers were determined by plating on E. coli CR63 after 5 min. Total phage were measured at 60 min, after lysis with chloroform.

T4 RNase H Exonuclease Activity Is Stimulated When the T4 Gene 45 Clamp Protein Is Loaded Behind It-
The T4 gene 45 replication clamp protein is loaded preferentially at the 3Ј end of a junction between single-and double-stranded DNA and at the 3Ј side of a nick (11,12), whereas T4 RNase H is loaded at the 5Ј end (3) (see Fig. 1C and diagrams at the top of Fig. 2A). The clamp protein can move in both directions on the duplex DNA. It can also track for short distances on single-stranded DNA but falls off single-stranded DNA more rapidly than double-stranded DNA (30). In Fig. 2, the nicked, gapped, and partial duplex substrates were made by annealing oligonucleotides to single-stranded circular DNA and were labeled at the 3Ј end of the downstream 86-base fragment. Addition of the T4 clamp, clamp loader, and ATP stimulates the 5Ј-to 3Ј-exonuclease on the nicked substrate ( Fig. 2A, lane 9) and on the substrate with the 28-base gap (lane 14), where the clamp can be loaded behind the nuclease but does not stimulate on the partial duplex (lane 4). The nuclease, by itself, has similar activity on the partial duplex and the gapped molecule ( Fig. 2A, lanes 2 and 12) but has very little activity on the nicked substrate (lane 7), where there is no single-stranded DNA behind the 3Ј-labeled duplex. As we have shown previously (8), 32 protein increases the processivity of the nuclease when there is single-stranded DNA behind the nuclease ( Fig. 2A, lanes 3 and  13). However, it does not increase the nuclease activity on the nicked substrate ( Fig. 2A, lane 8). There will be more clamp loaded in front of the nuclease in reactions with 32 protein because 32 protein increases the loading of the clamp by the clamp loader (31,32). Interference by this increased clamp in front of the nuclease is the likely reason that there is less activity on the nicked substrate when 32 protein is present in addition to the clamp, clamp loader, and ATP (compare Fig. 2A, lanes 9 and 10, and Fig. 2B, reactions 4 and 5). When polymerase is present to extend the duplex available to the clamp ahead of the nuclease, there is no longer inhibition by 32 protein (Fig. 2B, compare reactions 12 and 13). Note that in Fig.  2B, the reaction products were cut with BstNI restriction nuclease to allow determination of the extent of hydrolysis at the 5Ј end.
ATP and the T4 gene 44/62 clamp loader are needed in addition to the clamp for stimulation of T4 RNase H activity on nicked substrates. This is shown, using 5Ј end-labeled substrates, in Fig. 3. Addition of the clamp did not change the size of the oligonucleotides removed by the nuclease, which are predominantly dimers and trimers (8). The same size distribution is observed when a 5Ј-labeled partial duplex is cut by the nuclease alone (Fig. 3, reaction 2), and when the nicked substrate is cut in the presence of the clamp, clamp loader, and ATP (reaction 10).
The T4 clamp can move along single-stranded DNA, but it falls off more rapidly than on a duplex (30,33). We detected some stimulation of RNase H when the gap between the nucleotides was increased from 28 to 100 bases, but none with a gap of 200 bases (Fig. 4).
The Clamp Does Not Increase the Processivity of T4 RNase H-Circular replication clamps surrounding the DNA tether their respective polymerases on the template, thus increasing their processivity (number of nucleotides added per polymerase binding). We have shown previously that T4 RNase H by itself is nonprocessive, hydrolyzing a single small oligonucleotide with each binding, but that the nuclease becomes processive when the T4 gene 32 protein is added (8). In that study, hydrolysis of a partial duplex like that in Fig. 2A (left) by T4 RNase H alone was halted by addition of a fork DNA trap, unless 32 protein was present (8). Likewise, the limited hydrolysis of the nicked substrate by RNase H alone was stopped by addition of fork DNA after 20 s (Fig. 5, compare reactions 4 and 6). The increased rate of hydrolysis of the nicked substrate with the clamp (Fig. 5, reaction 7) is not the result of an increase in processivity. This hydrolysis stopped completely as soon as the fork DNA trap was added (Fig. 5, reaction 8).

Mutations in a Conserved Clamp Interaction Motif at the N terminus of RNase H Eliminate Stimulation by the Clamp-A
short conserved motif essential for interaction with replication clamps has been identified in bacteria, archaea, and eukaryotes (reviewed in Refs. 34 and 35). The N terminus of T4 RNase H has a sequence similar to the C-terminal clamp interaction motifs identified previously in T4 DNA polymerase (36), and the T4 gene 33 and 55 late transcription activators (37) (Fig. 6A). Shamoo and Steitz (38) have determined the crystal structure of the clamp from the closely related phage RB69, complexed with a C-terminal peptide from RB69 DNA polymerase. The most highly conserved RB69 polymerase residues (Fig. 6A, shaded letters) directly contacted the clamp in their structure. The analogous shaded residues in p21 contact human PCNA clamp (39), and those from Archaeoglobus fulgidu FEN1 contact the archaeal clamp (40).
The N-terminal region of RNase H was disordered in the crystal structure of Mueser et al. (41) (Fig. 1A) and is distant from the essential residues that surround the two magnesiums in the active site of the enzyme. To evaluate the importance of the motif in T4 RNase H, we have constructed a protein with a deletion of this disordered region (⌬2-10, hereafter ⌬N) and several point mutants (Fig. 6B). The mutant RNase H proteins were expressed and purified as described under "Experimental Procedures." Each of the mutant proteins had exonuclease activity similar to the wild type on the partial duplex substrate (Fig.  6B, right panel), and like the wild type had very little activity on the nicked substrate (Fig. 6B, center panel). Addition of the clamp and clamp loader did not stimulate hydrolysis by the deletion mutant ⌬N or the proteins with mutations of two residues, L3A and M6A, corresponding to RB69 DNA polymerase residues contacting the clamp (Fig. 6B, center panel). Mutations in two other residues, M5A and D8A, which correspond to RB69 DNA polymerase residues that did not contact the clamp, were stimulated by the clamp like the wild type nuclease.
A C-terminal Domain of T4 RNase H Is Required for Its Interaction with 32 Protein-T4 gene 32 ssDNA-binding pro-tein increases the processivity of T4 RNase H and controls how much DNA is removed from lagging strand fragments by the nuclease (see Introduction) (8,13). The C terminus of RNase H has a pair of helices (H12 and H13) that protrude away from the catalytic site in the direction where we expected 32 protein should be located (Fig. 1A) (see "Discussion"). We find that T4 RNase H with a deletion of the last helix (⌬295-305), both helices (⌬286 -305), or the entire region from 278 -305 (⌬C) retain exonuclease activity on the partial duplex substrate but are not stimulated by 32 protein (Fig. 7). The C-terminal region is not required for stimulation by the clamp protein (Fig. 6B,  left panel). Conversely, the N-terminal mutant (⌬N), not affected by the clamp (Fig. 6B), is stimulated normally by 32 protein (Fig. 7A).  (41)) with the regions deleted or mutated in this study shown in purple. The active site residues (red) surround the magnesium (shown as yellow spheres) (27). Residues are denoted as circles and were disordered in the structure. Residues structurally aligned (46) with A. fulgidus FEN1 residues that contacted the template strand in B are shown in magenta (see "Discussion"). B, ribbon model of the crystal structure of A. fulgidus FEN1 with DNA (Protein Data Bank 1RXW (40)). The template strand (5Ј-CGATGCT) is in magenta, and the primer strand (5Ј-AGCATCGG) is in blue. Residues contacting the template and primer strands are shown in magenta and blue, respectively (see "Discussion"). The helices containing these residues (␣2, ␣3, ␣14, and ␣15) and the corresponding helices in T4 RNase H (H2, H3, H12, and H13) are noted on the structures. C, diagrams of the substrates for the exonuclease and flap endonuclease activities of the 5Ј-nuclease family that includes T4 RNase H and FEN1. Top, exonuclease substrate. The region in magenta, which corresponds to the magenta template strand in B and its extension in the 3Ј direction, is the predicted binding site for T4 32 protein and T4 gene 45 clamp, behind T4 RNase H. Bottom, flap substrate with an unpaired 3Ј end nucleotide. Regions colored magenta and blue correspond to the same colored strands in the crystal structure in B. 32 protein has a central core (residues 22-253) that contains the DNA-binding cleft, a short N-terminal "B-domain" (residues 1-21) that is thought to bind to the adjacent 32 protein when it is cooperatively bound to DNA, and an acidic C-terminal "A-domain" (residues 254 -302) that has profound effects on the T4 replication system (reviewed in Ref. 42). Both helicasedependent leading strand synthesis (43,44) and the elongation of primers made by the primase-helicase (43) are greatly decreased when 32 protein is replaced by 32 protein without the A-domain (32-A). We find that the A-domain is not required to increase the processivity of T4 RNase H. 32-A protein stimulated the nuclease to the same extent as full-length 32 protein (Fig. 8). 32 protein missing the B-domain (32-B) failed to stimulate when added either with (Fig. 8, lanes 15-18)    we have used the ⌬N and ⌬C RNase H deletion proteins in the T4 replication system (Fig. 9). The template is the 2.7-kb pUC-NICK plasmid, nicked at the single recognition site for the N.BbvC IA nicking enzyme, as described under "Experimental Procedures." The reactions contained T4 DNA polymerase, clamp, clamp loader, 32 protein, 41 helicase, 59 helicase loader, 61 primase, and DNA ligase in addition to the RNase H indicated on the figure. In the absence of RNase H (Fig. 9, reaction 1) the short lagging strand fragments are not joined together because the RNA primer with a 5Ј-triphosphate at the end of each fragment prevents ligation. Lagging strand fragments are efficiently sealed with the wild type or ⌬N RNase H (Fig. 9, reactions 2 and 4). Sealing adjacent fragments is impaired with the ⌬C RNase H (Fig. 9, reaction 3), as shown by the products migrating with unligated fragments and intermediate length DNA migrating between the normal leading and lagging strand products. We conclude that reducing the RNase H interaction with 32 protein (⌬C), but not the clamp (⌬N), interferes with the sealing of lagging strand fragments.
The defect of T4 RNase H ⌬C in preparing fragments for sealing is also evident in a model lagging strand system in which a 3Ј-labeled 86-base oligonucleotide and an upstream 43-mer are separated by 1479 bases, a gap length similar to the length of an Okazaki fragment (Fig. 10). During DNA synthesis, the downstream 3Ј-labeled fragment is extended past the BstNI site, giving a 143-base restriction product, when BstNI is added after the replication reaction (Fig. 10, reaction 7). Simultaneous elongation of the unlabeled upstream fragment creates a nicked DNA that cannot be sealed by DNA ligase without RNase H (Fig. 10, reaction 8), because there is a hydroxyl group rather than a phosphate group on the 5Ј end of the downstream fragment. 5Ј to 3Ј digestion by T4 RNase H provides the 5Јphosphate, so that after gap filling the adjacent fragments can be sealed by ligase, giving a 191-base BstNI restriction fragment (Fig. 10, reaction 2). The number of nucleotides removed by T4 RNase H, under conditions needed for ligation, is measured by the length of products shorter than the 143-base fragment in reactions in which ligase is omitted (Fig. 10, reactions  1, 3, and 5). If digestion continues for 86 bases, the two labeled nucleotides are removed as dimers or trimers, and the 191-base ligated product will not be labeled. With the wild type and ⌬N RNase H, all of the labeled fragments except the dimers and trimers are ligated (Fig. 10, compare reaction 1 with 2 and reaction 5 with 6), showing that the nuclease has made at least one cut on each fragment to provide the required 5Ј-phosphate. Within 30 s, in the absence of ligase, there is extensive hydrolysis as shown by fragments shorter than 143 bases. There is little further hydrolysis at later times, because polymerase has filled the gap, creating a nick that limits access to the nuclease. When wild type RNase H is replaced by the C-terminal deletion (⌬C), there is significantly less hydrolysis. In the absence of ligase there are few products shorter than 123 bases (Fig. 10,  reaction 3), and at 2 min 43% of the labeled fragments could not be ligated (reaction 4), showing that the original 5Ј-hydroxyl group was still present. This suggests that the interaction between 32 protein and the wild type RNase H helps to load the nuclease at the junction between the 5Ј end of the fragment and the 32 protein-covered ssDNA behind it, in addition to increasing the processivity of the nuclease once it is loaded.

T4 Phage with a Deletion in the rnh Gene Is Not Complemented by T4 RNase H with a C-terminal Deletion-Either T4
RNase H or the 5Ј-nuclease that is part of E. coli DNA pol I is essential for T4 phage production (4) (see Introduction). Production of T4 phage with a large deletion in the gene encoding T4 RNase H (T4 ⌬rnh (⌬118 -305)) was only 1-2% of the wild type on a host with the polA12 mutation in E. coli pol I. Phage production was substantially restored by providing wild type T4 RNase H on a plasmid (4). We have used this system to test the effect of the ⌬C and ⌬N deletions in T4 RNase H in vivo ( Table I). The host E. coli MIC2003 (rnhA339::cat polA12) has an interruption in the gene for RNase HI and a temperaturesensitive mutation in DNA pol I affecting its 5Ј-nuclease. The cells were grown at 30°C, and then shifted to 43°C before infection, to disrupt the PolA12 5Ј-nuclease and induce production of T7 RNA polymerase from the pGP1.2 plasmid, as described under "Experimental Procedures." Plasmid encoding the ⌬N-(⌬2-10) deletion in T4 RNase H was as effective as the wild type plasmid in supporting growth of T4 ⌬rnh, showing that interaction between the clamp and the nuclease is not essential for phage production. In contrast, E. coli MIC2003 with plasmids encoding the ⌬C deletion (⌬278 -305), or the shorter deletions ⌬286 -305 or ⌬295-305, produced significantly less T4rnh than the wild type T4 RNase H plasmid. Interestingly, the yield of wild type T4 phage was lower with plasmids encoding the ⌬286 -305 or ⌬295-305 deletions than with the wild type or vector, raising the possibility that the truncated RNase H was a dominant negative inhibitor of the wild type enzyme. We conclude that T4 RNase H that is not stimulated by 32 protein cannot support the production of T4rnh in vivo.

DISCUSSION
T4 RNase H, which removes the RNA primers and some adjacent DNA from lagging strand fragments, is stimulated by both the T4 gene 45 replication clamp (Figs. 2 and 3) and 32 ssDNA-binding protein (8). It interacts with the clamp via a conserved clamp binding motif at the N terminus of the nuclease (Fig. 6). A C-terminal helical bundle at the C terminus is needed for its stimulation by 32 protein (Fig. 7). By using mutant T4 RNase H with deletions in the binding site for either the clamp or 32 protein, we have shown that it is the 32 protein, rather than the clamp, that most affects the maturation of lagging strand fragments during DNA replication in vitro (Figs. 9 and 10), and phage production in vivo (Table I).
Binding Sites for the Clamp, 32 Protein, and DNA-Although the binding sites for the clamp and 32 protein are on separate termini of T4 RNase H (Fig. 1A, shown in purple), they are close together in the crystal structure of the nuclease (41), away from the active site residues (shown in red), surrounding the magnesium in a large cleft at the top of the protein (27). The N-terminal residues 1-11, which include the clamp binding site, were disordered in the original structure (41) but are ordered in a more recent structure of the metal-free enzyme. 2 The analogous PCNA-binding motif is close to the C terminus of FEN1 (40), but is in a similar location relative to the active site cleft, as the N-terminal clamp binding site on T4 RNase H (compare Fig. 1, A and B).
There is unfortunately no crystal structure of a member of this 5Ј-nuclease family with DNA present in the active site of the enzyme. The recent structure (40) of the co-crystal of A. fulgidus FEN1 with the DNA that would be behind the 2 T. C. Mueser, unpublished data.

FIG. 6. Mutations in a conserved clamp interaction sequence at the N terminus of T4 RNase H eliminate stimulation of the exonuclease by the clamp.
A, alignment of selected clamp interaction sequences in phage, eukaryotic, and archaeal proteins. Consensus residues are shown in boldface. Shaded residues in RB69 DNA polymerase contact the RB69 gene 45 clamp in a crystal structure of the clamp and polymerase peptide (38). Shaded residues in p21(WAF1/CIP1) contact the human PCNA clamp (39), and those in the A. fulgidus FEN1 contact the archaeal clamp (40). References for the clamp interaction sequences are as follows: T4 DNA polymerase (36); T4 gp33 and gp55 transcription factors (37); RB69 DNA polymerase (38); human p21 (39) active site shows that this DNA is bound on the surface of FEN1 close to the PCNA-binding site (primer strand blue and template strand magenta as shown in Fig. 1, B and C). Biochemical analysis of mutant human FEN1 proteins is also consistent with this location for the upstream DNA (45). The FEN1 residues making contact with the template strand (Thr 63 , Lys 317 , Asn 67 , Lys 321 , Arg 64 , Phe 35 , and Arg 64 as shown in magenta in Fig. 1B) are located on helices ␣2, ␣3, and the C-terminal helices ␣14 and ␣15, and the loops connecting these helical pairs. Structural alignment of T4 RNase H and A. fulgidus FEN1 using the Combinatorial Extension (CE) program (46) shows that these correspond to T4 RNase H helices H1 and H2 and the C-terminal helices H12 and H13. The T4 RNase H residues that align structurally with the A. fulgidus FEN1 template-binding residues (Lys 52 , Asn 300 , Lys 56 , Glu 304 , Lys 53 , Thr 27 , and Thr 30 , respectively) are shown in magenta on Fig. 1A. Deletion of helix H13 (T4 RNase H ⌬295-305) was sufficient to prevent 32 protein stimulation of the nuclease (Fig. 7). In the absence of DNA, we observed only a weak interaction between N-terminal His or glutathione S-transferase-tagged T4 RNase H and 32 protein in pull-down experiments (data not shown). However, a tight complex of T4 RNase H and 32 protein on a gapped DNA, like that diagramed in Fig.  1C, top, can be shown in gel mobility shift experiments. 3 Assuming that T4 RNase H binds DNA as shown for the A. fulgidus FEN1, the binding sites for both 32 protein and the 45 clamp are located close to the predicted position for the DNA behind the catalytic site surrounding the magnesium (Fig. 1). This is consistent with the expected positions of the clamp and ssDNA-binding protein during lagging strand synthesis (Fig.  1C), as well as our experimental finding that both 32 protein (8) and the clamp (Fig. 2) stimulate only when they can be loaded behind the nuclease.
The PCNA, E. coli ␤, and T4 45 protein DNA replication clamps have similar circular structures (47,48) and recognize a conserved clamp-interaction sequence on the DNA polymerases, 5Ј-nucleases, ligases, and mismatch repair proteins that are affected by the clamp (reviewed in Refs. 34 and 35). There are now several crystal structures of clamp proteins bound to peptides with the clamp-binding sequences from interacting proteins (see Fig. 6A). In all of these structures, including that of the phage RB69 gene 45 clamp with the peptide from RB69 DNA polymerase (38), the peptide is found in the same location in a hydrophobic pocket on the interdomain loop. However, there is evidence that the clamp-interacting peptide from T4 polymerase can also bind at the subunit interface of the 45 clamp in solution (49). When the structure of T4 RNase H was docked on the phage RB69 clamp by putting its N-terminal clamp interacting residues in the position of the RB69 DNA polymerase peptide, it was clear that the nuclease could not bind the clamp, if 32 protein was bound near the C terminus (not shown). On a gapped substrate, 32 protein stimulates the nuclease to a greater extent than the clamp, and there is no further stimulation by adding the clamp, as well as 32 protein (Fig. 2).
The preferred flap substrate for FEN1 is a so-called double flap with a single nucleotide displaced at the 3Ј end, in addition to the 5Ј flap cut by the nuclease (Fig. 1C, bottom)  cuts the double flap predominantly in the duplex one nucleotide beyond the junction, thus making a nicked duplex product that can be sealed by ligase. In contrast, T4 RNase H cuts the double flap substrate at a significantly slower rate than a single flap of the same sequence, although like FEN1, T4 RNase H cuts the double flap most often at the position giving a nicked duplex product. 3 The primer strand in the DNA in the A. fulgidus FEN1 structure had an unpaired 3Ј nucleotide, corresponding to the 3Ј flap (Fig. 1, B and C, bottom). Three residues in the loop between ␣14 and ␣15 that bind this unpaired 3Ј nucleotide (His 308 , Phe 310 , and Ser 311 ) have no match in T4 RNase H, in the structural alignment between these two proteins, because helices H12 and H13 in T4 RNase H are shorter than helices ␣14 and ␣15 in A. fulgidus FEN1.
Different Mechanisms for Coordinating Maturation of Lagging Strand Fragments-Phage T4, E. coli, and eukaryotic cells each have efficient, but different, mechanisms for coordinating the reactions that remove the RNA primers, and those that elongate the discontinuous fragments, on the lagging strand. This coordination is essential to prevent the accumulation of recombinogenic nicks and gaps as a result of the discontinuous nature of lagging strand synthesis. The T4 replication system is the simplest (11,12). A single clamped polymerase is used to extend the RNA primer made by the phage-encoded primase and to complete the synthesis of the lagging strand fragment.
While polymerase is extending the fragment, a separate nuclease, T4 RNase H, acting as a 5Ј-exonuclease rather than a flap endonuclease, removes the RNA pentamers and about 30 nucleotides of adjacent DNA. 32 protein, which coats the ssDNA between the nuclease and polymerase, coordinates their activities by increasing the rate and processivity of each of these enzymes (13). The large difference in their rates ensures that in most cases polymerase will complete the upstream fragment before the nuclease can bind a second time (13). T4 DNA polymerase dissociates rapidly when it reaches an annealed duplex (51,52). Thus the nuclease must act before polymerase finishes, so that a ligatable nick is formed when polymerase leaves. In contrast to the eukaryotic system discussed below, the polymerase and nuclease act separately, although each is affected by 32 protein. The rate of hydrolysis by the nuclease is not changed when polymerase is present (13), and the rate of lagging strand synthesis is the same, with or without the nuclease (Fig. 9).
Our experiments suggest that the interaction between 32 protein and RNase H promotes the binding of the nuclease on 32 protein-coated DNA, in addition to increasing its processivity. We used a model gapped lagging strand substrate, where removing a single nucleotide was sufficient to expose the 5Јphosphate needed for ligation. A C-terminal deletion mutant of RNase H, defective in the 32 protein interaction, was unable to  The E. coli MIC2003 host contains the rnhA339::cat interruption in the gene encoding the host RNase HI and the temperature-sensitive PolA12 mutation affecting the 5Ј-to 3Ј-exonuclease of DNA polymerase I. All host cells also contained a T7 promoter expression plasmid encoding wild type (WT) or mutant T4 RNase H, or the pVex11 vector, as well as a compatible pGP1.2 plasmid, which contains the gene for T7 RNA polymerase under the temperature-controlled P L promoter. The cells were grown to 1 ϫ 10 8 /ml at 30°C, shifted to 43°C for 15 min to induce production of T4 RNA polymerase and disrupt the polA12 DNA polymerase-associated 5Ј-nuclease, and then infected with T4D or T4 ⌬ rnh (⌬118 -305) at a ratio of 0.5 phage/cell. Infective centers, the number of cells containing at least one phage, were determined by plating on E. coli CR63, 5 min after infection. Total phage were measured at 60 min, after lysis with chloroform (see "Experimental Procedures"). a "Phage/infective center" is the total phage yield per original infective center, rather than the burst size of phage from a single cell, because the infected bacteria were not diluted to avoid additional rounds of infection. make a single cut on 43% of the molecules in the time needed for polymerase to fill in the 1.5-kb gap (Fig. 10). Under the same conditions, the wild type nuclease and an N-terminal deletion, which could interact with 32 protein but not the clamp, made the cut needed for a ligatable nick on all of the molecules. Control experiments showed that the wild type and both mutant nucleases hydrolyzed the DNA at the same rate in the absence of 32 protein. Fragment sealing during coupled leading and lagging strand synthesis by the complete T4 replication system was also impaired when the C-terminal deletion mutant of RNase H replaced the wild type (Fig. 9). Finally, although plasmids encoding wild type T4 RNase H can restore production of T4 phage with a mutation in its rnh gene (see Introduction), there was no increase in phage production with plasmids encoding the C-terminal deletions (Table I). Collectively, these studies provide strong evidence that the interaction between T4 RNase H and 32 protein is essential for lagging strand maturation in vivo and in vitro.
The interaction between T4 RNase H and the clamp could serve a back-up function, allowing the nuclease to be loaded on the nicked molecules that are formed when polymerase completes the upstream fragment, before the downstream primers are removed by the nuclease. T4 RNase H by itself degrades the 5Ј end of a nick very slowly; its activity at a nick is stimulated by the clamp but not by 32 protein (Figs. 2A and 4). The clamp increases loading of the nuclease at a nick, but it does not increase the processivity of the nuclease after it is loaded (Fig.  5). Our finding that a mutant of RNase H, not stimulated by the clamp, can function like the wild type in promoting fragment sealing in vitro, and phage production in vivo, shows that the primers are normally removed before the nick is formed by the upstream polymerase. A role for T4 RNase H in DNA repair is suggested by the increased sensitivity of the T4 rnh mutant to DNA-damaging agents (see Introduction). It seems likely that the interaction of the clamp with RNase H may prove to be important for this function.
In E. coli DNA replication, DNA pol III holoenzyme extends the primer made by primase and remains tightly associated with the replication clamp ␤, until the fragment is extended to form a nick. When the downstream DNA is reached, the subunit of the clamp loader, which is part of the holoenzyme, actively promotes separation of the ␣ polymerase subunit of the core polymerase from the clamp by binding to the clamp interaction motif at the C terminus of this subunit (53,54). E. coli polymerase I and the 5Ј-nuclease that is part of the same polypeptide catalyze a nick translation activity that is responsible for removing the primer (55). The preferred substrate for this 5Ј-nuclease is also the double flap with a single nucleotide displaced at the 3Ј end (Fig. 1C, bottom). Coupling of the polymerase and nuclease activities is suggested by the finding that the majority of molecules cleaved by the nuclease had been extended previously by the polymerase (55). However, most molecules extended by the polymerase were released, rather than transferred to the nuclease. These studies with pol I were carried out in the absence of other proteins. There is evidence that the polymerase activity of pol I is stimulated by the ␤-clamp (56) and may bind the clamp after the pol III has dissociated (57). It would be interesting to determine whether this clamp affects how pol I removes the primers, perhaps increasing the coupling between polymerase and the 5Ј-nuclease.
In eukaryotic replication, the RNA primer is made by pol ␣ polymerase-primase, which also adds the first DNA nucleotides to the primer. There is then a switch to pol ␦ (reviewed in Ref. 58). In contrast to T4 DNA polymerase and E. coli pol III, pol ␦ does not dissociate when it reaches a downstream duplex but instead catalyzes a limited strand displacement reaction. As summarized in the Introduction, FEN1 nuclease has the primary role of removing the RNA primers. The nuclease of the Dna2 nuclease-helicase is needed to cut longer flaps (30 bases), because RPA binding to these displaced strands inhibits cutting by FEN1 but stimulates cutting by Dna2 (reviewed in Refs.1 and 14). Recent studies demonstrate that coordination of the 5Ј-nuclease of FEN1 with the strand displacement activity of pol ␦ results in a nick translation reaction that efficiently removes the 5Ј RNA and a short region of adjacent DNA (15,16). In a coupled lagging strand model system with pol ␦, PCNA, replication factor C clamp loader, RPA ssDNA-binding protein, and FEN1, FEN1 nuclease activity increased with pol ␦, and strand-displacement synthesis by pol ␦ increased when FEN1 was present. When FEN1 is absent, the 3Ј-5Ј-nuclease of pol ␦ removes enough nucleotides to maintain a ligatable nick, if the RNA primer has already been removed. Genetic studies showing synthetic lethality of mutations in FEN1, the 3Ј-nuclease activity of pol ␦, and the 5Ј-nuclease of ExoI suggest that these enzymes have overlapping functions with alternate pathways for forming ligatable nicks on the lagging strand (59 -61).
In summary, most of the RNA primers are removed and replaced by nick translation reactions in E. coli and eukaryotes, catalyzed by the 5Ј-nuclease and polymerase of activities of pol I in the former and by the combined activities of FEN1 nuclease and pol ␦ in the latter. T4 DNA polymerase and RNase H do not carry out nick translation together. Instead, the stimulation of the nuclease by the 32 protein on the lagging strand ensures that the primers will be removed in time for a nick to be formed by the polymerase completing the upstream fragment. In each of these systems, the clamp present on the lagging strand can interact with the polymerase extending the fragment, the 5Ј-nuclease removing the primer, and the ligase joining adjacent fragments. Because of the subunit composition and symmetry of the clamp structures, there are three of the interdomain binding sites for these proteins on each clamp. An important unanswered question is whether the lagging strand clamp binds more than one of these proteins simultaneously. In this regard, the recent finding that the Sulfolobus solfataricus clamp is composed of three nonidentical subunits that contact either polymerase, ligase, or FEN1 (62) is intriguing.