Protein Determinants of RNA Binding by DNA Polymerase of the T4-related Bacteriophage RB69*

DNA polymerase (gp43) of phage T4 plays two biological roles, one as an essential DNA binding replication enzyme and the other as an mRNA-specific autogenous translational repressor. Binding of T4 gp43 to its mRNA target (translational operator RNA) interferes with gp43-DNA interactions, but it is unclear how the protein determinants for binding DNA are affected by the dynamics of gp43-mRNA interactions. We have used RB69 gp43, a natural variant of the T4 enzyme whose crystal structure has been determined to identify protein sites that respond to the interaction with specific RNA. We used protein phosphorylation markers, photocross-linking studies, protease sensitivity assays, and mutational analyses to examine the effects of operator RNA on the enzyme's five structural domains (N, exo, palm, fingers, and thumb). Our studies suggest that this RNA affects gp43-DNA interactions through global effects on protein structure that occlude DNA-binding sites but leave the enzyme accessible to interactions with the sliding clamp (RB69 gp45) and possibly other polymerase accessory proteins. We discuss the possible biological significance of putative RNA-binding motifs in the N and palm domains of RB69 gp43.

The replicative DNA polymerase of bacteriophage T4, product of phage gene 43 (gp43), is a pol ␣-like (B-family) enzyme that displays a number of activities, all resident in the same 898-residue protein molecule (reviewed in Ref. 1). In addition to its two distinct catalytic functions (i.e. the polymerase (pol) 1 and 3Ј-exonuclease (exo or editing) activities, T4 gp43 bears specific binding activities to deoxyribonucleoside triphosphates (dNTPs)), other proteins of the T4 DNA replication complex, and its own mRNA. In vivo, T4 gp43 functions as an autogenous translational repressor that interacts with a nucleotide sequence, termed "translational operator," which overlaps the ribosome-binding site in the translation-initiation region of gene 43-encoded mRNA (2,3). It is known that the RNA determinants of operator specificity to the protein include a propensity of the nucleotide sequence to form a specific stem-loop (hairpin) structure (2)(3)(4)(5)(6). Nucleotide sequence-independent interactions with the protein are also required (6). In contrast, little is known about the protein determinants required for recognition of the RNA determinants of specificity. In the work described here, we identified some of the protein sites for translational operator recognition through studies that utilized a phylogenetic variant of T4 gp43, the enzyme from phage RB69 (7,8). The crystal structure of RB69 gp43 has been determined (8) and subsequently refined through structural studies of gp43-DNA cocrystals (9,10). RB69 gp43 consists of five modules, or domains (N, exo, fingers, palm, and thumb), the orientations of which differ slightly between the two DNA-binding configurations of the enzyme (i.e. the pol or "closed" form (9) and the exo or "open" form (10)). The apoenzyme crystallizes in the "open" configuration (8). There is no similar structural information available for the complex between RB69 gp43 and its operator RNA.
We have used the known structural features of RB69 gp43 as a framework for a multifaceted analysis of complexes between this protein and its mRNA target. In one type of analysis, we introduced protein phosphorylation markers at specific RB69 gp43 sites and measured the ability of operator RNA to alter the level of phosphorylation at these sites. We observed a mix of responses to the RNA, including inhibition and stimulation of phosphorylation, depending on site location. In some cases (e.g. the gp43 site for interaction with the sliding clamp protein (RB69 gp45)), operator RNA had no effect on phosphorylation. Collectively, the results suggest that operator RNA induces specific global changes in the structure of RB69 gp43 that interfere with access to the DNA binding grooves of the protein but not with sites of gp43 interaction with certain polymerase accessory proteins. Results of other studies that utilized protein-RNA photocrosslinking, protease sensitivity assays, and site-directed mutagenesis also showed that operator RNA binding affects multiple sites on the same face of RB69 gp43 that binds DNA. Overall, our studies provide a rationale for proposing that specific RNA evolved as a physiological regulator of the access of DNA to sites on this DNA polymerase. We note and discuss the presence in RB69 gp43 of amino acid sequences that resemble RNP motifs of known RNA-binding proteins.

EXPERIMENTAL PROCEDURES
Bacteria, Phages, and Plasmids-The Escherichia coli strain DH5␣ (F Ϫ 80dlacZ⌬M15⌬(lacZYA-argF) U169 deoR recA1 hsdR17(r Ϫ k ,m Ϫ k ) phoA supE44 -thi-1 gyrA96 relA1) was used as host for transformations with recombinant DNA in all protocols involving the cloning of DNA fragments in expression vectors and the introduction of specific mutations in cloned DNA through site-directed mutagenesis. DH5␣ was purchased as a frozen stock from Invitrogen (catalog no. 18265-17) and used as recommended by this supplier. The E. coli strain BL21(DE3) was used as the bacterial host for recombinant plasmids in experiments requiring in vivo transcription of cloned RB69 or T4 genetic sequences under control of a plasmid-borne T7 10 promoter. BL21(DE3) carries a chromosomally integrated T7 RNA polymerase gene that can be activated through lac operon inducers (11). The inducer used here was isopropyl-␤-D-thiogalactoside. E. coli strain CAJ70 was used as host for RB69 phage infections as previously described (7,12).
Many of the plasmids we used were constructed as part of this study and are described below and under "Results." Other plasmids used have been described in previous reports (7,12,13).
Purification of gp43-The purified T4 gp43 and RB69 gp43 used for photocross-linking and, in the case of RB69 gp43, protease sensitivity assays were prepared by ion exchange and DNA-cellulose chromatography, as described previously (12,14). For RB69 gp43 phosphorylation assays, we used N-terminally or C-terminally histidine-tagged derivatives of this protein. Optimal conditions for protein purification with such derivatives were obtained with three histidines added to the C-terminal end or six histidines added to the N-terminal end of the protein. The histidines were introduced through PCR-driven site-directed mutagenesis of pCW19R, a recombinant plasmid that carries the wild-type RB69 gene 43 under control of the T7 10 promoter (12,13). Stratagene's QuikChange protocol and reagents were used for the in vitro mutagenesis. We fused three histidine codons to the 3Ј-terminal end of the open reading frame of the gene by amplifying pCW19R in the presence of a synthetic DNA duplex in which the sense strand had the sequence 5Ј-gttcgatatgttcgattttCACCACCACtgaTATAATGTCGACCT-GC-3Ј (RB69 gp43 codons in lowercase, histidine codons in uppercase, and vector residues italicized). This resulted in the isolation of pVPR43C-His3 (Table I). To add six histidine codons to the 5Ј-terminal side of the RB69 gene 43 open reading frame, we constructed pVPR43N-His 6 (Table I) by amplifying pCW19R in the presence of a synthetic DNA duplex in which the sense strand had the sequence 5Ј-GTCtaaa-ggaaattaaatgCACCACCACCACCACCACaaagaattttacttaacgg-3Ј (RB69 gene 43 sequence in lowercase, initiation codon underlined, histidine codons in uppercase, and vector residues italicized). The two resulting plasmid constructs, pVPR43C-His3 and pVPR43N-His, were isolated, and their nucleotide sequences were verified after transforming products of the respective QuikChange mutagenesis protocol into E. coli DH5␣.
His-tagged RB69 gp43 was purified from crude extracts of the appropriate clones following induction of transcription of recombinant plasmids in vivo. In each case, about 10 ml of E. coli BL21(DE3) culture (OD ϳ0.5) containing the recombinant plasmid of interest was aerated with isopropyl-␤-D-thiogalactoside (at 1 mM) for 2 h at 30°C to effect elevated synthesis of gp43 under control of the plasmid-borne T7 promoter. Induced cells were harvested by centrifugation at 5000 ϫ g for 10 min and washed in 5 ml of sonication buffer containing 20 mM sodium phosphate (pH 7.5), 400 mM NaCl, and 5 mM imidazole before being resuspended in 0.5 ml of the same buffer. The concentrated cell suspension was then sonicated (model W-225R sonicator; Heat Systems-Ultrasonics) and clarified by centrifugation at 15,000 ϫ g for 30 min. The resulting 0.4 -0.5 ml of crude extract was applied to a Ni 2ϩ -nitrilotriacetic acid spin column (Qiagen; catalog no. 30210). After washing the column with 20 volumes of 20 mM sodium phosphate buffer (pH 7.5) containing 20 mM imidazole and 400 mM NaCl, bound material was eluted with buffer containing 100 mM imidazol. The eluate was dialyzed against buffer containing 20 mM Tris-Cl (pH 7.0), 2 mM dithiothreitol, and 50% glycerol and then stored at Ϫ20°C until used. Typical yields using this protocol were 100 -150 g of purified gp43.
Protein Phosphorylation Assays-The sequence R(R/K)X(S/T) in proteins can be phosphorylated by the catalytic subunit (␣ isoform, PKA) of mouse cAMP-dependent protein kinase. RB69 gp43 contains no such sequence but does contain several short sequences that differ from the site for PKA by one or two amino acids. We used site-directed mutagenesis to generate a set of clones from VPR43C-His3 and pVPR43N-His6, whose protein products could be phosphorylated at designated sites by PKA. These clones and their encoded proteins (each bearing a histidine tag and a target for PKA) were each compared with the wild-type construct (pCW19R and its encoded gp43) for biological and in vitro properties by using previously described methods (12). The constructs used here (see "Results") were all similar to the wild-type construct in replication proficiency and RNA-binding properties (results not shown). Table I lists the specific nucleotide and amino acid substitutions used to generate the phosphorylation markers in RB69 gp43 for this study.
For the conduct of gp43 phosphorylation assays (see "Results"; Fig.  2), 2.5 pmol of the desired gp43 construct (purified on a Ni 2ϩ -nitrilotriacetic acid column) were incubated at 37°C with or without operator RNA (at 50 nM when used) and 1 unit of a commercial PKA that does not require cAMP for activation (catalog no. 6000S; New England Biolabs). The reaction mixture (25 l) also contained 20 mM Tris-Cl (pH 7.5), 10 mM MgCl 2 , 20 Ci of ␥-32 P-labeled ATP, 20 M cold ATP, 500 nM E. coli tRNA (catalog no. 109550; Roche Molecular Biochemicals), and 8 units of the RNase inhibitor RNasin (catalog no. 4769401; Promega). Samples (5 l) were removed from each reaction mixture at 1-min intervals and diluted into 10 l of 50 mM Tris-Cl (pH 6.8) buffer containing 2.3% (w/v) SDS and 50 mM mercaptoethanol. The samples were subsequently analyzed by SDS-PAGE and phosphorimaging.
Preparation of RNA Substrates for Photocross-linking Experiments-Nucleotide sequences of the RNA substrates used for photocross-linking experiments are depicted in Fig. 1. The RNAs were synthesized by in vitro transcription of the appropriate DNA duplexes that had been cloned into the BglII-SmaI interval of the T7/SP6 expression vectors pSP73 and pSP72, sold by Promega (Madison, WI) (catalog no. P2191 and P2221, respectively). Nucleotide sequences of the sense strands of the cloned DNAs were as follows: (i) 5Ј-GATCTATAATATATCAAGAG-CCTAATAACTCGGGCTATAAACTAAGGAATATCTATGAAAGAATT-3Ј, for cloning the T4 "CWTop" operator sequence ( Fig. 1) under control of the SP6 promoter of pSP73, and (ii) 5Ј-GATCTAAAACAAAAGACA-ATAACTCGTCTAAAGGAAATTAAATGAAAGAAUUUCCC-3Ј, for cloning of the RB69 "RBop4" operator sequence ( Fig. 1) under control of the T7 10 promoter of pSP72.
For in vitro transcription reactions, recombinant plasmids containing the desired DNA sequences were linearized by BamHI digestion, repurified, and then used in the RiboMax transcription mixture sold by Promega (catalog no. P1300) with either T7 RNA polymerase or SP6 RNA polymerase. Reaction mixtures (50 l) contained 5 g of DNA template, the four ribonucleoside triphosphates (ATP, CTP, UTP, and GTP) at 300 M each, and 90 units of the designated RNA polymerase. The mixtures were incubated for 1.5 h at 37°C and then quenched by the addition of 5 units of RQ1 RNase-free DNase (catalog no. P1300; Promega). After an additional incubation of 15 min at 37°C, RNA products were purified through two phenol/chloroform (pH 4.5) extractions and one ethanol precipitation, dissolved in 100 l of nuclease-free 10 mM Tris-Cl (pH 8.0) buffer, and separated from small DNA and RNA fragments and nucleotides on "Quick Spin G-50" columns (catalog no. 100406; Roche Molecular Biochemicals). Purity of the RNA products was evaluated by electrophoretic analysis on 10% polyacrylamide gels containing 7 M urea, and RNA concentrations were determined spectophotometrically. For radiolabeling RNA, [␣-33 P]UTP (catalog no. NEG-307H; PerkinElmer Life Sciences) was added to transcription mixtures at 50 Ci/reaction (specific activity of 3000 Ci/mmol). To prepare photoreactive RNA products for cross-linking experiments, the unlabeled CTP in these mixtures was replaced with 5-iodo-CTP (catalog no. I5628; Sigma). All manipulations with photoreactive RNAs were performed under reduced light conditions or in the dark. The iododerivatives of operator RNAs exhibited a diminished affinity to gp43 (by ϳ10-fold) as compared with the unmodified counterparts; however, binding remained specific (i.e. 7-10-fold higher affinity than with RNA of nonspecific nucleotide sequence (results not shown). Radiolabeled RNA substrates were examined for purity by electrophoresis on 10% polyacrylamide gels containing 7 M urea and visualized by the use of a phosphor imager (Fuji model FLA-3000). RNA stocks were stored at Ϫ80°C and used within 3-4 days of preparation for radiolabeled RNAs or 4 -5 weeks of preparation for unlabeled RNAs.
RNA-Protein Cross-linking Methods-The methods we used to identify RNA-linked gp43 peptides (Figs. 3 and 4) were based on the demonstrated utility of methylene blue as an enhancer of photocross-linking between proteins that bind specific RNA and base-paired residues in the RNA (15) as well as the utility of iodosubstituted pyrimidines as chromophores (16). In our experiments, we used 5 ng of methylene blue (catalog no. M9140; Sigma) per pmol of RNA. The reaction mixtures for cross-linking (25 l) were deposited in the wells of a microtiter plate that was placed on an ice surface and subsequently exposed to a source of high intensity visible light (20-watt, 1200-lumen lamp, EarthLight; Philips) for 20 -40 min. Samples of the irradiated solutions were subsequently made 7 M in urea, incubated at 42°C for 15 min, and then analyzed by electrophoresis in 10% polyacrylamide gels containing 7 M urea. Resolved radioactive gel bands were visualized and quantitated by using the phosphor imager.
Irradiated mixtures of RB69 gp43 (100 pmol) and 33 P-labeled iodocytosine-containing RBop4 RNA (20 pmol), were treated with RNases A and T1 (1 l of 2 units/reaction each) for 30 min at 37°C to effect complete hydrolysis of the RNA. The RNase-treated material was then incubated with 0.5 g (in 2 l) of preactivated clostripain (Sigma; catalog no. C0888) for 1 h at the same temperature. Proteolysis was stopped by the addition of an equal volume of electrophoresis loading buffer containing 2% SDS and 100 mM dithiothreitol, followed by a 15-min incubation at 60°C. Electrophoretic analysis of the hydrolysis products of photocross-linked material was carried out by SDS-PAGE (12.5% gels), and resolved bands were transferred to polyvinylidene difluoride membranes ("Trans-Blot"; Bio-Rad catalog no. 162-0180) using a semidry electroblot system (Novablot, model 2117-250; Amersham Biosciences). The membranes were then stained for 30 min in 0.2% Coomassie Brilliant Blue R-250 (made in 50% methanol) and partially destained in 50% methanol. A radiolabeled gp43 fragment (ϳ29 kDa) was localized on the stained membrane, excised, and completely destained in methanol. Subsequently, N-terminal sequence determination was carried out on this peptide through the core facility at Louisiana State University (New Orleans, LA), which uses an Applied Biosystems model 447 pulsed liquid protein sequencer.
Limited Proteolysis of RB69 gp43 with Clostripain-In experiments that assessed the effect of operator RNA on hydrolysis of RB69 gp43 by clostripain (protein protection by unmodified RNA; see Fig. 4B), 2.5 g of the protein were incubated for specified time periods at 25°C with 0.25 g of preactivated clostripain in the presence or absence of 0.5 g of RBop4 RNA in 25 l of solution containing 20 mM sodium phosphate (pH 7.5), 1 mM CaCl 2 , and 20 g/ml E. coli tRNA. Proteolysis was arrested by the addition of 25 l of electrophoresis sample buffer containing 2% SDS and 0.1 M mercaptoethanol. Then this solution was heated in a boiling water bath for 2 min, and aliquots were subjected to SDS-PAGE (10% gel) in duplicate sets. One set was stained for protein with Coomassie Brilliant Blue, and the other set was transferred to a polyvinylidene difluoride membrane for subsequent N-terminal sequencing of the separated peptides, as described above for samples that were subjected to both RNase and protease treatments following photocross-linking.

RESULTS
Operator RNA Affects RB69 gp43 Structure-Previous studies seemed to suggest that operator RNA makes multiple contacts with gp43, some at locations outside the protein regions that bind primer-template DNA (4,12). We introduced phosphorylation markers at various positions in RB69 gp43 and scanned the protein molecule for those sites whose access to the protein kinase PKA could be affected by the interaction with operator RNA. Results are shown in Fig. 2. We observed two types of effects by operator RNA on phosphorylation within this set of RB69 gp43 markers: inhibition (site 382 RRRS 385 ; Fig. 2) and stimulation (sites 25 RRRT 28 , 240 RRLS 243 , and 245 RRKT 248 ; Fig. 2). Inhibition of phosphorylation at the 382 RRRS 385 site may reflect either a direct or indirect occlusion of the site by the RNA. We note that this site is within ϳ25 Å of the primertemplate junction in the crystal structure of RB69 gp43 (8 -10). The RNA-mediated stimulatory responses at the other three gp43 sites are likely to be the result of RNA-induced conformational changes in the protein.
For some of the PKA targets that we introduced into RB69 gp43, we observed no effect by the RNA on phosphorylation of the respective gp43 construct. One of the unaffected sites, 893 RRAS 895 (Fig. 2), is located in the C-terminal tail segment of this 903-residue polymerase (i.e. in a gp43 segment known to bind the phage-induced polymerase processivity factor gp45 (sliding clamp protein) (10)). As also shown in Fig. 2B, and perhaps as expected, phosphorylation of the 893 RRAS 895 site is inhibited by purified RB69 gp45. The addition of operator RNA did not alter the inhibitory effect by RB69 gp45, suggesting that gp43-operator interactions do not overlap, or affect, the gp43 determinants for binding the sliding clamp. Interestingly, FIG. 1. The T4 and RB69 RNA substrates used in this study. The RNA substrates were prepared by in vitro transcription of cloned synthetic DNA duplexes as described under "Experimental Procedures." Structure of the RNA hairpin for the T4 operator has been inferred from genetic, phylogenetic, biochemical (RNase sensitivity), and NMR studies (2,5,31,32). Existence of the RNA hairpin for RB69 has been inferred from comparative studies with the T4 counterpart (6,12). 2 Position of nucleotides on the RNA sequences shown are marked in relation to the start codons of the respective mRNAs. Cloning vector-derived nucleotides in RNA products of in vitro transcription are displayed in lowercase letters.
the three sites whose phosphorylation is stimulated by operator RNA (Fig. 2) are clustered in the vicinity of a gp43 region that has been proposed to interact with the phage-induced single strand-binding protein, gp32 (8); however, it is not yet known if operator RNA can affect gp32-gp43 interactions.
Specific Cross-linking between RB69 gp43 and the RB69 Operator-Recently, we demonstrated that iodocytosine-substituted RNA targets corresponding to the T4 and RB69 operator hairpins can be photocross-linked to the respective gp43 at high efficiency (up to 50%) in the presence of the dye methylene blue. 2 Such cross-linking between proteins and RNA or DNA is known to be specific to base-paired regions of the nucleic acid (15). We also showed that the base-paired C at position Ϫ14 of RBop4 RNA (Fig. 1) can be photocross-linked to T4 gp43 as well as to RB69 gp43. 2 In Fig. 3A, we compare the levels of gp43-RNA photocross-linked (XL) products obtained in experiments that utilized the two types of operator RNA substrates diagrammed in Fig. 1 as ligands for wild-type RB69 gp43 and T4 gp43. Conditions for photocross-linking were optimized for maximal yields of XL products, as described elsewhere, 2 in order to attempt to isolate these products for further analysis. We also examined photocross-linking of the same RNA targets with an in-frame deletion mutant of T4 gp43 (T4 DelM; Fig. 3A) that we had previously observed to possess the specific RNA binding activity of the wild-type protein (7). Although we have been able to visualize products of photocross-linking between gp43 and operator RNA substrates from both the T4 and RB69 phage systems, we have only been able to subsequently analyze products containing RB69-derived components. For unknown reasons, cross-linked products between RB69 gp43 and RBop4 could be solubilized in buffer lacking denaturants (SDS or urea), whereas similar products containing T4 gp43 were not similarly soluble. Fig. 3B shows electrophoretic separations of the XL products we used in the analyses described below.
In one type of analysis, photocross-linked material from the mixture of RB69 gp43 and RBop4 RNA (XL in Fig. 3B, lane 2) was incubated with a mixture of RNases T1 and A to effect complete hydrolysis of the RNA in the complex. Electrophoretic analysis of the hydrolysate (not shown), yielded an estimate of 1 mol of 33 P-labeled nucleotide/mol of cross-linked product. This estimate is consistent with our previous findings, 2 which also showed that the point of cross-linking is the iodocytosine-[ 33 P]U dinucleotide at positions Ϫ14/Ϫ13 of the RNA target depicted in Fig. 1.
In another type of analysis, a sample of the XL material in Fig. 3B, lane 2, was subjected to partial hydrolysis by clostripain, in addition to the complete prior hydrolysis of the RNA component of the complex with the mixture of RNases. These treatments resulted in the production of several nucleotidebound ( 33 P-labeled) protein fragments ranging in size between ϳ29 and ϳ70 kDa, as estimated from SDS-PAGE analysis (Fig.  3B, lane 4). The smallest of these fragments (ϳ29 kDa) was subsequently isolated (after transfer to a polyvinylidene diflu- Effect of operator RNA on phosphorylation of markers in RB69 gp43. RB69 gp43 mutants bearing specific PKA phosphorylation sites were constructed, purified, and phosphorylated as described under "Experimental Procedures." A, effects of RBop4 RNA (opRNA) on phosphorylation of these sites; "Cleft T" refers to the N-exo interdomain groove that binds the template DNA strand (see Fig. 6). B, the inhibitory effect of RB69 gp45 on phosphorylation of PKA site 893RRAS895 in RB69 gp43. The purified RB69 gp45 used was prepared as described by Nossal et al. (17) for T4 gp45. Note that the RB69 gp45 effect is not altered by opRNA.

FIG. 3. An analysis of gp43-operator interactions by methylene blue-mediated photocross-linking.
A, comparison of photocrosslinked products from several experiments that utilized different RNA substrates with either T4 gp43 or RB69 gp43 (lanes 1-4). Lanes 5 and 6 of A compare products of photocross-linking from experiments that utilized a deletion mutant of T4 gp43 (T4 DelM) that retains specific RNA-binding activity (7). B, lanes 1 and 2, separation of products of photocross-linking (on a urea gel) from an experiment that utilized RB69 gp43 and an iodocytosine-substituted RBop4 RNA substrate. oride membrane) and subjected to N-terminal amino acid sequencing (see "Experimental Procedures"). It was found to contain the sequence (N)Tyr-Lys-Tyr-Val-Met-Ser-Phe-Asp-Leu(C), suggesting that it corresponds to the RB69 gp43 segment extending from Tyr 404 to approximately Arg 668 of the RB69 gp43 primary structure (7). Sequence of this ϳ270-residue peptide and its location in RB69 gp43 are shown in Fig. 4A. We were able to rule out involvement in the cross-link of the middle ϳ100 residues of this peptide through the following observations. (i) Deletion of residues Glu 501 -Ala 555 of RB69 gp43 (unshaded sequence in Fig. 4A) or the corresponding sequence in T4 gp43 (Asp 498 -Asn 552 ) has been shown not to alter the RNA binding properties of the respective protein (7). Also, photocross-linking assays that utilized the deletion mutant from T4 yielded similar efficiencies of cross-linking to RNAs as did wild-type T4 gp43 (Fig. 3A, lanes 7-9). (ii) Binding mixtures were prepared between RB69 gp43 and its RBop4 substrate or E. coli tRNA under the same conditions used for photocross-linking, but without the addition of methylene blue and without exposing the mixtures to high intensity visible light. Samples were then subjected to partial hydrolysis with clostripain, and the products of hydrolysis were resolved by SDS-PAGE. The results, shown in Fig. 4B, revealed a protection of the protein from the protease by RBop4 RNA. Three peptide fragments (ϳ80, ϳ50, and ϳ28 kDa, respectively) were detected by Coomassie Blue staining of products from incubations containing tRNA that were either seen at much lower levels or not observed with treatments in the presence of RBop4. N-terminal sequencing of the 28-and 50-kDa proteolytic products recovered from gels indicated that cleavage of the unprotected RB69 gp43 occurred at Arg 249 (Fig. 4B). The cleavage at Arg 707 (or possibly Arg 719 ) is inferred from the estimated peptide fragment sizes. In the crystal structure of RB69 gp43 (8), all three of the putative RBop4-protected sites map on the face of the protein molecule opposite from the location of the fingers domain (which includes amino acid residues 471-572 of RB69 gp43; Fig. 4A). In other words, the results of this experiment implicate the two palm domain peptide segments highlighted in Fig. 4A in the cross-link with operator RNA. The results also suggest that access to clostripain of the gp43 sites near amino acid positions Arg 249 and Arg 707 (or Arg 719 ) is diminished in the presence of operator RNA. Arg 249 resides in the RB69 gp43 N-exo interdomain groove, which binds the single-stranded DNA template, and Arg 707 /Arg 719 reside in the pol-thumb interdomain groove, which binds double-stranded DNA (8,9). We note that results of the protease sensitivity assays contrast with results of experiments that used PKA as a gp43-modifying reagent (Fig. 2), where no RNA-mediated inhibition of phosphorylation by PKA was observed near Arg 249 . The difference in responses may be related to differences in dimensions of the two gp43-modifying enzymes used, which could affect their access to Arg 249 .
Mutational Analysis of RNA Binding by RB69 gp43-Since all of the PKA site-bearing RB69 gp43 constructs we have described in this report (Fig. 2) were observed to bind operator RNA normally, we presume that none of the amino acid positions that were altered in these constructs are critical for operator recognition. We have targeted additional RB69 gp43 residues for mutational analysis of this protein, focusing on amino acid residues that have been implicated in utilization of primer-template DNA and nucleotide precursors for DNA replication (9,13). Alanine-scanning mutagenesis did not reveal any effects on the translational repressor function of RB69 gp43 when residues in the polymerase and 3Ј-exonuclease catalytic centers of this protein were substituted with alanine. Examples of results are shown in Fig. 5, and a summary of the analysis of a larger set of RB69 gp43 mutants is presented in Table II. We have not yet encountered any single (or double) amino acid substitutions in gp43 that eliminate RNA binding without also affecting some aspects of the DNA replication function of the protein. In contrast, replication defective gp43 mutants that bind operator RNA normally are common and typically exhibit dominant lethal phenotypes when tested in plasmid-phage complementation assays (Table II). This phenotype ensues when a plasmid-encoded mutant gp43 represses the biosynthesis of wild-type gp43 from a T4 or RB69 phage that infects the plasmid-bearing E. coli host. Fig. 6 summarizes our findings from the current study on a reproduction of the RB69 gp43 structure that appeared in the work of Franklin et al. (9). In this figure, we have highlighted the locations of selected landmarks on the ribbon diagram of the gp43 structure and also included a representation of the RNA hairpin structure for the gene 43 translational operator FIG. 4. A, amino acid sequence of cross-linked 29-kDa peptide. Nine N-terminal residues of the 29-kDa peptide produced in the experiment for Fig. 3B, lane 4, were sequenced (see "Experimental Procedures"). The rest of the sequence was inferred from the approximate size of the peptide and cleavage site specificity of clostripain. Boldface residues are identical to corresponding positions in T4 gp43 (7). Shaded segments were implicated in the cross-link with RNA (see "Results"). Italicized residues correspond to a segment of the fingers domain that can be deleted (DelM mutants; Fig. 3) without loss of the RNA-binding function. Underlined residues were substituted with alanine and tested for biological activity in plasmid-phage complementation assays as described previously (Ref. 12; see Table II and Fig. 5). B, RNA-mediated protection of RB69 gp43 from cleavage with clostripain. The upper part shows electrophoretic separation (SDS-PAGE) of products from incubations of RB69 gp43 with clostripain in the presence (ϩ) and absence (Ϫ) of RBop4 RNA (see "Experimental Procedures"). Interpretations of the results are diagrammed in the lower part, with the photocross-linked 29-kDa gp43 fragment marked in the middle of the bar representing the length of the RB69 gp43 molecule. The cleavage at position Arg 249 was determined by N-terminal sequencing of the ϳ50-kDa fragment. The cleavage at Arg 707 /Arg 719 was inferred from the sizes of peptide products separated by SDS-PAGE after partial hydrolysis with clostripain and the specificity (Arg-X) of this protease. from RB69. We propose that the RNA hairpin binds RB69 gp43 within the protein cavity that harbors the intersection between the dNTP binding site and the primer-template junction near the polymerase catalytic center (10, 18 -20). RB69 gp43 residues that are essential for dNTP binding and the two catalytic functions of this enzyme do not appear to be essential for operator recognition, since they can be substituted with alanine without loss of repressor activity (Fig. 5 and Table II). Assays for protection of RB69 gp43 from clostripain by operator RNA (Fig. 4B) suggest that the interaction with specific RNA affects access of the protease to sites in the binding grooves for both double-stranded DNA (near Arg 707 ; Fig. 6) and singlestranded DNA (near Arg 249 ; Fig. 6) of this DNA polymerase. Previously, in binding studies that utilized T4 gp43 and a variety of RNA targets, we estimated that an RNA length of at least 26 nucleotides of nonspecific sequence, in addition to the sequence-specific hairpin (ϳ18 nucleotides), is included in the gp43-operator complex during translational repression (6). An RNA length of ϳ26 nucleotides 3Ј distal to the operator RNA hairpin structure may correspond to as much as 88 Å, which would be sufficient to reach any other part of the gp43 molecule if the RNA hairpin were to be anchored in the region bounded by the sites mapped in our current study. Thus, it is possible that all five domains of this protein contribute determinants for RNA binding. In essence, we can rule out the possibility that gp43 harbors a structural domain exclusively for the RNA binding function, although our photocross-linking studies implicate the enzyme's palm domain in contacts with the basepaired segment of the RNA hairpin component of the operator (Fig. 4). Possibly, the palm domain bears a cluster of determinants for RNA sequence recognition.

DISCUSSION
The multiplicity of protein-RNA contacts (this work) and degeneracy of RNA sequences that can be repressed by a gp43 (5, 6) make it impractical to probe a multifunctional protein of this size for mutations that selectively or preferentially affect the RNA binding function. Nevertheless, it has been possible to demonstrate that knocking out the catalytic activities of the RB69 gp43 does not knock out either the RNA or DNA binding function of this protein (Table II). In contrast, all of the known T4 gp43 missense mutants that exhibit defects in RNA binding (i.e. exhibit derepressed gp43 synthesis) appear to be also defective in aspects of the DNA binding function (e.g. fidelity of DNA synthesis or protein stability (21,22)). A crystal structure of the gp43-RNA complex would be helpful in establishing whether DNA (or nonspecific nucleic acid) and operator RNA share contact points on the protein but may be insufficient to elucidate dynamic aspects of the gp43-operator interaction.
Phages T4 and RB69 are phylogenetically related to each other and encode similar sets of DNA replication proteins from similarly organized and regulated genes on the two phage genomes (7,(23)(24)(25). Nevertheless, several of the homologous gene products from the two phage systems can be distinguished from each other, not only by sequence, but also on the basis of biological specificity of the respective gene functions (7,12,25). In particular, the T4 and RB69 DNA polymerases, which are ϳ61% identical (ϳ74% similar) in amino acid sequence (7), differ in their relative abilities to service genomes of the two phages when tested in crossspecies complementation assays (12). On the other hand, although T4 gp43 appears to be more narrowly specific toward its own genome and T4 genome products, the RB69 enzyme can replicate T4 genomes and interact with the T4 gene 43 translational operator just as effectively as T4 gp43 (7,12,13,26). 2 Also, a variety of biologically active RB69-T4 gp43 chimeras can be constructed through segmental exchanges between the two related proteins (7,12). Thus, we presume that the structures of T4 gp43 and RB69 gp43 are very similar to each other in the global context but differ at the atomic level, especially with regard to the protein determinants that recognize the specificity elements of translational operators. Unlike operator RNA, primer-template DNA appears to be recognized solely through nucleotide sequence-independent interactions (9,10). In vitro, the affinity of T4 gp43 to operator RNA (K d ϭ 1-2 nM) is 30-100-fold higher than to DNA or RNA of nonspecific sequence (K d ϭ 70 -100 nM) (4,6,12). In addition, the specific RNA is a potent inhibitor of DNA binding by the protein (4), suggesting that the RNA and DNA binding activities of gp43 are functionally linked to each other. Based on the work described here, we surmise that the inhibition of gp43-DNA binding by operator RNA is mediated through two types of effects, RNA-induced conformational changes in the protein and direct occlusion of some DNA-binding sites by the specific RNA. RNA-induced effects on global structure of the protein are perhaps best reflected in the results of experiments that utilized protein phosphorylation markers in RB69 gp43. The phosphorylation of some of these markers was stimulated by operator RNA (Fig. 2; see also Fig. 6), suggesting that the respective gp43 sites were made more accessible to PKA through the gp43operator interaction. Also, our photocross-linking studies (Fig. 3A) 2 do not directly implicate the DNA binding grooves of this DNA polymerase in operator binding (Fig. 6). Thus, it is possible that occlusion of DNA-binding sites by operator RNA occurs entirely as a consequence of RNA-induced effects on the structure of RB69 gp43. Interestingly, these effects by the RNA do not seem to interfere with interactions between the polymerase and its sliding clamp (Fig. 2).
The level of amino acid sequence identity between T4 gp43 and RB69 gp43 (ϳ61%) contrasts with the 90 -94% identity that has been observed among the catalytic subunits of the pol ␣-like DNA polymerase ␦ in mammals and ϳ58% identity between the mammalian ␦ subunits and their counterparts in Drosophila. Also, we note in comparison that the pol I-like (A-family) replicative DNA polymerases of phages T7 and T3 are ϳ97% identical. These comparisons (based on alignments from GenBank TM ) may underscore the importance of a conserved RNA binding function in the two diverged (by sequence) gp43 variants we have studied here. The RB69 gp43 structural framework and many of the protein's amino acid sequence motifs are shared by at least three other B-family DNA polymerases whose crystal structures have also been determined  (27)(28)(29). The high degree of divergence between the mRNA targets for T4 gp43 and RB69 gp43, 2 in contrast to conservation of the gp43 structural framework in nature (12,(27)(28)(29), may mean that most gp43-like DNA polymerases have had opportunities to evolve specific RNA binding functions. It will be important to find out whether these other polymerases interact with naturally occurring specific RNA ligands and whether such ligands have diverse physiological roles or are always associated with translational control.
In studies of RB69 gp43, the protein consistently co-crystallized with a guanosine bound to a site in the N-terminal domain (8 -10). The putative rGMP-binding site bears some structural similarity to a module of protein secondary structure, ␤␣␤␤␣␤ (RNP motif), which is seen in some known RNA-  diagram to the right). Orientations of the five gp43 domains (N, exo, thumb, palm, and fingers) are shown in the "closed" form of the protein structure (9). The primer (yellow) and template (gray) DNA strands are shown occupying the Palm-Thumb interdomain groove (for double-stranded DNA) and the N-Exo interdomain groove (for the single-stranded DNA template). Marked onto the gp43 structure are the approximate locations of sites protected from clostripain by operator RNA (Arg 249 and Arg 707 ; Fig. 4), the polymerase catalytic residues (pol active site) and the locations of putative RNA-binding motifs (RNP1 and RNP2) of the palm. The fingers (blue) domain of RB69 gp43 protrudes behind the structure (relative to the orientation shown in the figure) and includes the dNTP-binding residues near the fingers-palm junction (9,20). Also marked are the PKA target sites that we engineered (Table I) to test for gp43 phosphorylation in the presence and absence of operator RNA (Table II). For these sites, the effect of operator RNA on PKA-mediated phosphorylation is marked either in green (for stimulation), red (for inhibition), or blue (for lack of an effect). binding proteins (16). Recently, it was pointed out that the corresponding region in the N-terminal domains of the archaeal polymerases also bear this architectural feature (28,29). Whereas these observations are intriguing, it remains to be seen whether the archaeal enzymes also bind RNA or whether the putative rGMP binding region in RB69 gp43 is involved in RNA binding by this protein. Amino acid substitutions that disrupt the putative RNP motif of the N domain do not seem to affect gp43-operator interactions (K48A, Y49A, and D95A; Table II). We have examined the structure of RB69 gp43 for additional clues to the existence of protein folds that resemble the RNA-binding motifs of other proteins that bind specific RNA. In Figs. 4 and 6, we point out the approximate location of the sequence 612 EGEAFVLY 619 in one of the two gp43 palm segments that we implicated in the photocross-link to RNA (Figs. [3][4][5]. This sequence resembles the RNP1 motif (KG-FAFVXY) of certain RNA-binding proteins (30), and interestingly, it is included in a ␤20-strand that lies in close proximity, and antiparallel, to another ␤14-strand in RB69 gp43 bearing the sequence 394 AFVKEP 399 , which resembles a second known RNP motif, RNP2 (LFVKGM). On the other hand, the RNP1/ RNP2 structural arrangement usually occurs within modules consisting of 30 -40 amino acids (30), whereas the RNP1-and RNP2-like sequences of the gp43 palm are separated by ϳ200 residues, including the entire fingers domain of this polymerase (Fig. 4). Amino acid substitutions in RB69 gp43 that disrupted the similarity to the RNP1 consensus did not affect the replication and repressor functions of the protein (E614A and Y619A; Table II). Mutations in the RNP2-like sequence of RB69 gp43 knocked out both the replication and repressor functions of the protein (F395A and V396A; Table II). There are yet additional folds in RB69 gp43 that exhibit similarity to RNA binding motifs. In particular, a structural motif in a looped segment of the thumb domain (residues 711 NVWD-MEGTRY 720 ) resembles motif E of some RNA replicases and protrudes within 10 -18 Å of the pol catalytic center of RB69 gp43. A crystal structure of the gp43-operator complex is needed to shed more light on the relevance of these putative RNA-binding motifs to translational operator recognition by this DNA polymerase.