A Domain of the Gene 4 HelicaseIPrimase of Bacteriophage T7 Required for the Formation of an Active Hexamer*

The bacteriophage T7 gene 4 protein, like a number of helicases, is believed to function as a hexamer. The amino acid sequence of the T7 gene 4 protein from res idue 475 to 491 is conserved in the homologous proteins of the related phages T3 and SP6. In addition, part of this region is conserved in DNA helicases such as Esch erichia coli DnaB protein and phage T4 gp41. Mutations within this region of the T7 gene 4 protein can reduce the ability of the protein to form hexamers. The His 475 --> Ala and As p 485 --> Gly mutant proteins show decreases in nucleotide hydrolysis, single-stranded DNA binding, double-stranded DNA unwinding, and primer synthesis in proportion to their ability to form hexamers. The mutation Arg487 --> Ala has little effect on oligomeriza tion, but nucleotide hydrolysis by this mutant protein is inhibited by single-stranded DNA, and it has a higher affinity for dTTP, suggesting that this protein is defec tive in the protein-protein interactions required for ef ficient nucleotide hydrolysis and translocation on sin gle-stranded DNA. Gene 4 protein can form hexamers in the absence of a nucleotide, but dTTP increases hex amer formation, as does dTDP, to a lesser extent, dem onstrating that the protein self-association affinity is influenced by the nucleotide stop buffer containing 40 mM EDTA and 25% glycerol. The reaction samples were analyzed by nondenaturing 20% PAGE and phosphorimage analysis. The decrease with time in the amount of radiolabeled 25-mer migrating as duplex DNA with the 75-mer was measured, and the percent oligonucleotide displaced was calculated as described previously (35). RNA-primed DNA Synthesis-Analysis of gene 4 protein stimulation of DNA synthesis on ssDNA by T7 DNA polymerase was performed as described previously (15) with the following specific changes: the 20-/-LI reactions contained dTTPase buffer plus 50 roM potassium glutamate, 0.3 mM NTP, 0.3 mM d(G,A,C)TP, and 2.0 mM [a- 32 PJdTTP. The reac tions were incubated at 30°C for 20 min, and the amount of radioac tivity incorporated into DNA was determined by scintillation counting. The specific activity of the [a- 32 PJdTTP solutions varied from 4 to 11 cpm/pmol.

II To whom correspondence and reprint requests should be addressed.
An internal translation initiation sequence in the gene 4 transcript results in expression of the encoded protein as two colinear forms: the 63-kDa gene 4A protein and the 56-kDa gene 4B protein 0). The 63-kDa gene 4 protein exhibits helicase activity and, by virtue of an amino-terminal zinc binding motif, catalyzes the template-directed synthesis of tetraribonucleotides that are used as primers by the T7 DNA polymerase (10,11). The 56-kDa protein lacks the 63 amino-terminal residues that form the zinc binding motif and consequently catalyzes only helicase activity (12). Since the 63-kDa protein (primase) can provide both primase and helicase activities, it is both necessary and sufficient for productive infection by T7 phage (13,14).
While the gene 4 primase is sufficient for T7 DNA replication, there is considerable evidence that the helicase and primase proteins interact cooperatively. For example, the 56-kDa helicase stimulates template-dependent tetraribonucleotide synthesis by the 63-kDa primase and enhances DNA replication in vivo (14,15). Further evidence of this interaction was derived from studies (6) with a gene 4 protein containing a defective nucleotide binding site (NBS). The NBS mutant protein inhibits nucleotide hydrolysis by wild-type gene 4 proteins through direct protein-protein interactions, demonstrating the importance of cooperation between the gene 4 proteins in order to translocate on ssDNA and to catalyze the unwinding of double-stranded DNA. Also, the low level of primer synthesis catalyzed by the NBS mutant primase is increased by wild-type helicase, suggesting that the wild-type protein forms a complex with the mutant primase enabling it to translocate along the template DNA to a primase recognition site (17).
The precise mechanism of strand separation by a DNA helicase is not known. However, all helicases examined thus far function as multimeric proteins and use the energy of nucleotide hydrolysis to unwind dsDNA (18)(19)(20). Two forms of helicase multimers have been identified: dimer and hexamer. The E. coli Rep protein is the best characterized example of a dimeric helicase (21). The reported group ofhelicases that form hexamers currently includes proteins such as the T7 gene 4 protein (22), E. coli proteins DnaB (23), Rho (24), and RuvB (25), SV40 large T antigen (26), and the bacteriophage T4 gp41 (18).
The T7 gene 4 protein is one of a group of bacterial and bacteriophage helicases known as the "DnaB helicase family" that share multiple regions of amino acid sequence similarity (27). Notably, this group includes E. coli DnaB, bacteriophage T4 gp41, and the gene 4 protein of phage T3. Four regions of sequence similarity were identified in the DnaB helicase family; the first two motifs are known to be associated with nucleotide binding, while the roles of the third and fourth motifs are not yet known. We compared the amino acid sequence of the T7 gene 4 protein with its homologs in the closely related phages T3 and SP6. Other than the NBS, the only continuous region of conserved amino acid sequence among these three proteins occurs toward their carboxyl terminus and corresponds to T7 gene 4 residues 475-491 (refer to Table I). This region of the T7 gene 4 protein overlaps the fourth DnaB helicase family motif, corresponding to T7 gene 4 protein residues 481-500. Interestingly, the phage SP6 gene 4 protein has sequence similarity to the fourth motif of the DnaB family ofhelicases only within the common overlapping region (corresponding to T7 gene 4 protein residues 481-491).
Since the function of this highly conserved carboxyl-terminal region is not known, we have investigated its role in T7 gene 4 protein through site-directed mutagenesis. We show that mutations within this region affect both the formation of gene 4 protein hexamers and cooperative protein-protein interactions within the hexamer that are required for nucleotide hydrolysis and translocation on ssDNA.
DNA, Nucleotides, and Enzymes-Single-stranded M13mp6 DNA was purified as described (29). T7 DNA polymerase was provided by S. Tabor (Harvard Medical School). Wild-type gene 4A protein was prepared by B. Beauchamp (Harvard Medical School) using the protocol described under "Methods." Restriction enzymes were obtained from Amersham and New England Biolabs, Inc. DNA sequencing reagents and l3;y-methylene dTTP were obtained from United States Biochemical Corp. "Ultma" DNA polymerase from Perkin Elmer was used for the mutagenic polymerase chain reaction. Oligonucleotides were purchased from Oligos, Etc. Radiolabeled nucleotides were purchased from Amersham and DuPont NEN.
Complementation Analysis-E. coli DH5a with plasmids encoding wild-type or mutant gene 4A proteins were used to titer T7 WT and T7 4-1 phage following a standard protocol (32). Plaque size and number were assessed after incubation for 18 h at 37°C.
Protein Purification-T7 gene 4A overexpression, cell harvest, and cleared lysate production were performed as described previously (16). Gene 4A is overexpressed to levels where the protein production and purification are readily monitored by SDS-PAGE and Coomassie Brilliant Blue staining. The gene 4A protein R487 A was precipitated from 47 ml of cleared lysate (3.2 mg/ml total protein, fraction I; Fig. 1, lane 1) by adjusting the NaCI concentration to 0.5 M and the polyethylene glycol (PEG4000) concentration to 10%, followed by incubation on ice for 60 min and centrifugation at 12,000 x g for 15 min at 4°C. The gene 4A protein-PEG precipitate was then resuspended in Buffer P (40 mM potassium phosphate, pH 6.8, 5 mM DTT, 5 mM EDTA, 10% glycerol) plus 20 mM KCl (fraction II; Fig. 1, lane 2). Fraction II (20 ml at 2.7 mg/ml protein) was loaded onto a phosphocellulose (Whatman PH) column, 2.5 em" x 5 em, that was equilibrated in Buffer P. The column was then washed with 10 column volumes of Buffer P plus 20 mM KCl, and the gene 4A protein eluted with a 150-mllinear gradient of 20-1000 mMKCl in Buffer P. The fractions containing gene 4A protein were pooled (fraction III; Fig. 1, lane 3), and MgCl2 was added to 10 mM. Fraction III (18 ml at 0.9 mg/ml protein) was loaded onto a 2-ml agarose-hexane-ATP type 3 (Pharmacia Biotech Inc.) affinity column equilibrated in Buffer T (Buffer P plus 500 mM KCl and 10 mM MgCI2)' In the presence of magnesium, the gene 4A protein is tightly bound to the affinity resin, even at 500 mMKCl, allowing the effective removal of contaminating proteins by thorough washing of the column with 15 ml of Buffer T. The gene 4A protein was then eluted with Buffer P plus 500 mMKCl and 20 mMEDTA. The fractions containing gene 4 protein were pooled (fraction IV; Fig. 1, lane 4) and dialyzed extensively at 4°C against storage buffer (40 mx potassium phosphate, pH 7.0, 20 mMKCl, 5 mM DTT, 5 mM EDTA, 50% glycerol).
Native·PAGE and ssDNA Binding-The electrophoresis samples contained 2 fJoM gene 4A protein, 40 mM Tris-HCl, pH 7.0, 50 mM NaCl, 10 mM DTT, 10 mM MgCI 2, 20% glycerol, and 1 mM nucleotide, when present. The reaction mixtures were incubated for 10 min at room temperature before being applied to the gel. The electrophoresis buffer contained 25 mM Tris, 190 mM glycine, 10 mM magnesium acetate, and 0.1 mM of the same nucleotide present in the reaction mixtures. Nondenaturing electrophoresis was performed using 4-15% linear gradient polyacrylamide gels (Bio-Rad) that were soaked in the electrophoresis buffer for at least 25 min before the samples were loaded. Electrophoresis was at 7.1 V/cm for 30 min and then 14.3 V/cm for 2 h. The gels were then fixed and silver-stained for analysis. The molecular weights of the gene 4 protein oligomers were estimated by comparison with a curve of the R, versus log(M r ) of native PAGE protein standards of 67,000-669,000 (Pharmacia Biotech Inc.), Scanning densitometry of the silverstained gels was performed using a Personal Densitometer SI (Molecular Dynamics).
DNA binding experiments were performed using the same gel and electrophoresis buffer system described for the native PAGE. The 10-fJol binding reactions contained 40 mM Tris-HCl, pH 7.0, 50 mM NaCl, 10 mM MgCI 2, 10 mM DTT, 0.1 fJoM oligonucleotide, 1 mM l3;y-methylene dTTP, and gene 4A protein at various concentrations. The samples were incubated at room temperature for 5 min, then loaded onto the gel for electrophoresis. The 35-base oligonucleotide used for ssDNA binding experiments had the sequence: 5'-CAGATGCGCGCCTCCTGGCT-TATCGGTGTACTTGG-3' and was end-labeled in a standard reaction with [y_ 32PIATP and T4 polynucleotide kinase; unincorporated label was removed using a spin-column (Microspin S300, Pharmacia Biotech Inc.), After electrophoresis, the gels were fixed, stained with Coomassie Brilliant Blue, and dried. The amount of DNA that co-migrated with the gene 4 protein was determined by PhosphorImager (Molecular Dynamics) analysis. UV-mediated cross-linking of gene 4 proteins to radiolabeled (dT)2o was performed as described previously for UV-cross-linking of gene 4 proteins to dTTP (16). The lO-fJol reactions contained 0.2 fJoM gene 4A protein, 0.01 fJoM radiolabeled (dT)2o, 40 mM Tris-HCl, pH 7.0, 10 mM DTT, 100 mM NaCl, 50 fJog/ml bovine serum albumin, and 10% glycerol. The (dT)2o was 5'-end labeled with [y_ 32PIATP and T4 polynucleotide kinase. When included in the reactions, MgCl2 was at 10 mM, and nucleotides, dTTP or l3,y-methylene dTTP were at 2 mx, The reactions were incubated for 10 min at 30°C, then placed on ice and exposed to the UV source (1.0 milliwatts/cm'') for 15 min. The reaction mixtures were then subjected to SDS-PAGE, and the amount of labeled DNA bound to the gene 4 protein was determined by phosphorimage analysis.
Electron Microscopy-In preparation for electron microscopy, the gene 4A proteins at a concentration of 10 fJog/ml (175 nM) were incubated for 10 min at 20°C in a buffer of 10 mM HEPES, pH 7.0, 50 mM NaCl, 10 mM MgCI 2, 1 mM DTT, and 0.1 mM EDTA. When present, the nucleotide concentration was 0.6 mM. The samples were adsorbed onto a carbon film supported by a copper grid and stained with 1% uranyl acetate (33). The photomicrographs were taken on a Philips 400 TLG electron microscope at magnifications of 30,000 and 60,000.
Nucleotide Hydrolysis Assay-The assay for examining nucleotide hydrolysis by gene 4 proteins was performed essentially as described previously (16,34). The reactions were carried out in a volume of 20 fJol at 30°C for 20 min and analyzed by thin layer chromatography on polyethyleneimine-cellulose followed by scintillation counting of the isolated dTDP spot. The reaction buffer (dTTPase buffer) contained 40 mM Tris-HCl, pH 7.0, 50 mM NaCl, 10 mM MgCI 2, 10 mM DTT, and 50 fJog/ml bovine serum albumin and varying concentrations of [3HldTTP. Gene 4A proteins were at a concentration of 200 nM, and M13mp6 ssDNA, when present, was used at 50 fJoM (expressed as nucleotide equivalents). To ensure an adequate magnesium concentration in the The corresponding amino acid sequences of the T3 and SP6 gene 4 proteins and the E. coli DnaB protein are aligned with T7 gene 4 protein amino acids 475 to 500. The T7 gene 4 protein residues mutated for this study are in bold. The region of sequence similarity among T7, T3, and SP6 gene 4 proteins is indicated by the upper bracket. The region of sequence similarity among bacteriophage T7, T3 and the fourth motif of the E. coli DnaB protein helicase identified by Ilyina et al. (27) is indicated by the lower bracket. The dots within the brackets above and below the amino acid sequences indicate residues common to each group, and the numbering corresponds to the T7 gene 4 protein amino acid sequence. <s. coli DnaB protein amino acids 389-413 aligned with T7 gene 4 protein sequence as in Ilyina et al. (27).
reactions, the magnesium acetate concentration of the [3HJdTI'P solutions was equal to the nucleotide concentration. The kinetic constants were derived with the aide of the Enzyme Kinetics program for the Macintosh computer (Trinity Software).
Helicase Assay-The helicase substrate consisted of a 75-base oligo. nucleotide with a partially complementary radiolabeled 25-base oligonucleotide annealed to its 3'-end. The 25-mer (10 pmol) was 5'-end labeled with ['Y. 32 PJATP and diluted with unlabeled 25-mer (120 pmol) for the annealing reaction. The 75-and 25-base oligonucleotides, 4.5 /-LM final concentration each, were annealed in a 27-/-LI reaction containing 40 roM Tris-HCI, pH 7.0, 50 roM NaCI, and 10 roM MgCI 2 • The annealing mixture was incubated at 70°C for 2 min and cooled slowly to 30 DC, 5 /-Liloading buffer was added, and the reaction mixture was loaded onto a 3% agarose gel (MetaPhor Agarose, FMC BioProducts) in TAE buffer and electrophoresed at 14 V/cm for 1 h. The gel was briefly exposed to film, and the autoradiograph was used to locate and excise the labeled substrate band. The helicase substrate was eluted from the agarose slice by freezing the slice at -80°C, thawing at 37°C, and then filtering the slice through a microcentrifuge filter device (Bio-Rad). The gel slice was rewet with TE buffer and incubated at room temperature for 30 min, filtered, and combined with the first filtrate. The concentration of the helicase substrate was determined spectroscopically by measuring the absorbance at 260 nm.
The helicase assay measures the ability of the gene 4 protein to separate the partially complementary oligonucleotides of the helicase substrate. The helicase reactions (50 /-Lll contained dTTPase buffer, 2 mM dTI'P, 44 nM helicase substrate, and 20 nM gene 4A protein. The reaction mixtures were assembled on ice, and the reaction was started by the addition of gene 4 protein and incubation at 30 "C, At timed intervals, 7.5-/-LI samples were removed from the reaction and added to 7.5 /-Ll of stop buffer containing 40 mM EDTA and 25% glycerol. The reaction samples were analyzed by nondenaturing 20% PAGE and phosphorimage analysis. The decrease with time in the amount of radiolabeled 25-mer migrating as duplex DNA with the 75-mer was measured, and the percent oligonucleotide displaced was calculated as described previously (35).
RNA-primed DNA Synthesis-Analysis of gene 4 protein stimulation of DNA synthesis on ssDNA by T7 DNA polymerase was performed as described previously (15) with the following specific changes: the 20-/-LI reactions contained dTTPase buffer plus 50 roM potassium glutamate, 0.3 mM NTP, 0.3 mM d(G,A,C)TP, and 2.0 mM [a-32 P JdTTP . The reactions were incubated at 30°C for 20 min, and the amount of radioactivity incorporated into DNA was determined by scintillation counting. The specific activity of the [a-32 P JdTTP solutions varied from 4 to 11 cpm/pmol.

Comparison of Conserved Helicase Motifs
Ilyina et al. (27) observed that the T7 gene 4 protein shares four conserved amino acid sequence motifs with a number of procaryotic DNA helicases. This group ofhelicases includes the E. coli DnaB protein and was termed the DnaB helicase family (27). Among the other helicases in this group are phage T4 gp41 and the T3 gene 4 protein. The T7 and T3 gene 4 proteins share 90% sequence identity and have the lowest overall similarity to the other members of the DnaB helicase family. We performed a pairwise comparison of the gene 4 protein amino acid sequences of the closely related phages T7 (1), T3 (36), and SP6. 2 Of the four DnaB helicase motifs only motif 1, the NBS (T7 gene 4 protein residues 313-319) and the region just preceding and overlapping the fourth helicase motif (T7 gene 4 protein residues 475 to 481, refer to Table I) were conserved among the three gene 4 protein homologs. The fourth and most carboxylterminal DnaB helicase motif of T7 gene 4 protein has 94% and 65% identity with the same region of the T3 and SP6 gene 4 proteins, respectively. The T7 and T3 gene 4 proteins have 8 of the 11 most highly conserved residues in this helicase motif, whereas SP6 has only 4 of these residues in common, all within the central region of similarity (residues 481 to 491, Table I). Conversely, the amino acid sequence from 475 to 480 that is conserved in the T7, T3, and SP6 gene 4 proteins are not conserved in any of the other DnaB-related helicases. This sequence analysis has further refined a highly conserved helicase motif represented here by T7 gene 4 residues 485 to 491 and demonstrates the uniqueness of the T7-related gene 4 proteins.

Mutagenesis and Complementation Analysis
The observation that similar sequence motifs are shared between the T7, T3, and SP6 gene 4 homologs and other members of the DnaB family of helicases suggests that this domain may play an important role in gene 4 protein functions. Consequently, we examined the effect of mutations within this region on the properties of the gene 4 proteins. The mutations were constructed in a plasmid, pGP4-G64 s 10, which encodes a copy of the T7 gene 4 protein with a mutation of Met 6 4~G ly (M64G) (17). This mutation allows expression of the 63-kDa gene 4 protein and prevents expression of the 56-kDa protein.
The M64G mutation has no detectable effect on the helicase or primase activities of the 63-kDa gene 4 protein (14,17). A similar mutation in the gene 4 protein ofM64L was also shown to not change the function ofthe protein (13). Therefore, in the interest of clarity, in this report we refer to the gene 4A protein with the M64G mutation as "wild-type." The amino acid changes His 4 7 5~A la (H475A) and Arg 4 8 7~A la (R487A) were const ructed by oligonucleotide-directed mutagenesis . In the process of constructing t he mutants for thi s a nd other gene 4 protein studi es, we discovered t hat the original gene 4 clone, pGP4-6, carried a mu ta tion of A to G a t pha ge T7 nucl eotide 13,018, changing gen e 4 protein amino acid resi due 485 from Asp to Gly. This mu tan t pr otein D485G has been incor porated into t his study, a nd t he gene 4 plasm ids were reconstr ucte d with the correct seq uence.
The effect t hese mu tations ha ve on the ability of gene 4A protein to su pport T7 bacterioph age repr oduction in vivo wa s examined by complementation a na lysis. Gene 4-delete d T7 ph age (T7 il4-1) will not lyse E. coli unl ess a fun cti onal copy of gene 4 is prov ided in tran s. Previous stud ies ha ve shown t hat the 63-kDa form of gene 4 protein is sufficie nt for T7 ph age replic ation (13 , 14). E. coli DH 5a -ca r rying plasmids encoding the wild-type a nd mutant gene 4A proteins were infected wit h wild-ty pe T7 or T7 M -1 ph age, an d the number a nd size of the plaques produ ced were determ ined. Wh en in fecte d with T7 il4-1, t he strains carrying the mu tati ons H475A an d D485G produced 20 .8-and 3.3-fold fewer plaq ues , respectively, t ha n wild-ty pe gene 4A (Table II). Also, bot h of t hese mu tati ons resulted in plaqu es t hat were on average sma ller in diam eter (pinpoint to 2.5 mrn) than thos e pr oduced by wild -typ e gene 4 (2.5 to 5 rnm ) under the sa me conditions . No plaques were produced by ph age T7 il4-1 whe n plated on cells containing the plasmid with gene 4A mu ta tion R487A.

Gene 4 Pr otein Oligom erization
To determ ine the basis of the defects, the mutan t gene 4 protein s were purified to homogen eity for biochemic al a nd biophysical analysis. We chose to purify the 63-kDa vers ion of the gene 4 pr otein since it possesses bot h he licase a nd prim ase activities.
Protein Purification-We have improve d and streamli ne d our ea rlier procedures for t he purifica tion of the gene 4 protein (11,12,16,37) by combi ning a n effecti ve pr ecipita tion and en richment step with a very efficient a ffinity chromatography resin (see "Methods" for detail s). Th e gene 4 pr otein is se lectiv ely precipitated from t he cleared lysate with polyeth ylen e glycol 4000 (PEG) and NaC I (Fig. 1, lane 2 ). Th is step pr ovides a n enri chment of the gene 4 protein, an d the PE G does not interfere with sub seque nt procedures. Th e resus pended gene 4 protein-PEG pellet wa s loa ded onto a ph osph ocellulose column an d eluted with a KCl gradient; t he gene 4 pr otein eluted between 300 a nd 400 m xt KC!. Ma gnesium was a dded to 10 rna , and the pooled fractions were load ed onto a n agarose -he xa ne -ATP column . In the pr esence of magnesium, the gene 4 pr otein bin ds tightly to the a ffinity resin a t KCl concentrati ons of at least 500 mM. After t he column wa s wa shed, th e gene 4 pr otein (DE3 )/pGP4-G64 s IQR 487A; 2 , fraction II , resu sp ended polyet hy len e glycol pr ecipita te; 3 , fraction III , phosphocellu lose chroma tography pool; 4, puri fied ge ne 4A R487A protein , fraction IV, agarose-hexane-ATP column pool; 5, purified R487A protein overloaded to demonstrate purity. was elute d with bu ffer containing EDTA (Fig. 1, lane 4). Based on a na lysis of silve r-stained gels (not shown), we estimate that t he gene 4 protein s purified following t his protoco l are gr eater than 99.8% pure wit h a yie ld for t he R487A pr otein of 3.4 mg from 152 mg of total cell protei n. The yield and purity of the wild-ty pe pr otein wa s similar, 10.3 mg from 390 mg of total cell pr otein.
Native PAGE A nalysis-Recently, t he gen e 4 pr otein was shown to for m hexam ers (22,38), confir ming studies indicating th at the gene 4 protein was a ctive as a mu ltimer (16,37,39). We exa mine d th e effect of t he mu tati ons within t he conserved dom ain on t he oligomeri zati on of gen e 4 pr oteins usin g na tive PAGE . Th e a ppa re nt molecula r weights of t he bands form ed by t he gene 4 pr otein s were estima te d by comparison wit h k nown standa rds . Th e gene 4 pr oteins migr ate to positi ons that correspond to the following form s with estimated molecul ar mass, from th e top to t he bot tom of th e gel: hexam er , 408 -kDa (Fig. 2, open arrows); pen tam er, 322-kDa ; tet ram er , 263-kDa; trimer, 215-kDa ; and dim er, 121-kDa (Fig. 2, closed arrows ). In ea ch cas e, t he estimated molecul ar mass is very close to th at of t he calculated mass for t hat form of gene 4 63-k Da pr otein oligome r . No gene 4 protein was det ected at the position in the gels that would corres pond to a monomer (63-kDa) . Th e migr ation pattern of t he pr otein s was not a ffected by the pr esen ce ( Fig. 2A) or a bsence of magnesium (not shown). In the absence of a nu cleotid e, the protein s form ed prima rily hexam ers a nd dimers with some minor interme diate oligomers. Sca nning den sitometry of the silver-stained gels showe d th at wild-ty pe pro tein migr ated pr edomin antly as a hexam er , whe reas prote in R487A was evenly divided bet ween dimers a nd hexam ers (Ta ble II I). Th e H475A and D485G pr oteins, however , migr ated pr edom in antly as dim ers. This direct comparison of the pr otei ns dem onstrated that in the abse nce of a nu cleotid e ligand eac h of th ese mutati ons wea ke ns the gene 4 protein-protein in ter actions re quire d for hexam er form ation .
Th e effect of various nucl eotid es on the as sociation state of th e gene 4A pr otein s can be observed by including the nu cleotide in the rea ction a nd elect rophore sis bu ffer s. In t he presence of dTTP, t he preferred nucl eotide substrate for T7 gene 4 pr ote in, eac h of th e gene 4A pr otein s form predomin an tly hexamers ( Fig. 2D and Tabl e III ). However, a sign ifica nt portion of the H4 75A and D485G pr oteins still migr ate as dimer s (Fig. 2D,   closed arrowhead ). Also, wit h /3,v-methylene dTTP, the wildty pe and mu tant R487A pr otein s form hexam ers almost exclusively (Fig. 2, C a nd D). In contrast, in the presence of dTDP, none of the pr otein s form ed hexam ers as readily as they did with dTTP (Fig. 2B ). Th e H475A an d D485G protein s are  pri ncipally dimers in t he presen ce of dTDP and f3;y-m ethylen e dTT P, with few distinct hi gh molecul a r weight bands vis ible, in dicating eithe r t hat t hey do not form discrete compl exes or that the complexes are unstab le under t he conditions u sed for elect rophoresis (Fig. 2, B and C). In addit ion to further demonstrating t he effect of t he mut ati ons in this r egion on oligom er formation, these resul t s indicat e that the gene 4 pro t ein-protei n affinity (a ssocia t ion affinity) will va ry dep ending on whethe r a nucleosid e di-or triphosp h ate is bound.
Visua lizat ion of the T7 Gene 4 Protein by Electron Microscopy-To assess hexam er formation by a n alte r native method, the wild -typ e an d mutant ge ne 4 protein s were exa mine d by EM (Fig. 3). Th e proteins were in cub ated with a nd without nu cleotides a nd th en nega tively staine d wit h uranyl acetate. In the presenc e of dTTP, most of the visi ble compl exes form ed by th e wild -typ e a nd mutant protein s we re hexamers , in confirmation of the results r ep ort ed by Ege lman et at. (38) . Th e per cent of t he popul ation of ea ch gene 4 pr otein that form ed h examers in t he absen ce a nd presence of various nucleotides was determin ed by countin g fields of mol ecul es (Table IV). These est imates are largely cons istent with the results of the native PAGE analysis. However , because of the difficul ty involve d in distingui shing an d scoring the smalle r multimers, it is possible tha t t he values represen t a n overest imat ion of the per cent h exam er s form ed . Alternatively, the time required a nd t he stresses imposed by electro pho resis may have disrupted the weake r prot ein-protein association of these mutants, thus decreasing our ability to det ect h exam ers by native PAGE . This is es pe cia lly evide nt for the H47 5A a nd D485G pro t ein s which sh ow little to no hexamer formation in the abse nce of dTTP by n ati ve PAGE , bu t , as score d by EM , al most 50% of t he complexes were hexam ers . Nevertheless, both the results of the n ati ve PAGE an a lysis a nd the EM show t hat in t he abse nce of nucleotide th e mutant protein s form h exam er s less r eadily than the wild-typ e pro t ein . Moreover , t he enla rge d views of individual h exam er s (Fig. 38) form ed by th e mutan t gene 4 protein s are in distinguish able from those formed by the wildtype protein , demonstrating that t hese mutations do not cause a ny gross morphological defects in th e pr otein s at this level of resolu tion.

B iochemical Analy sis of Wild-typ e and Mu tant Gene 4 Protein s
Earlier st udies indica te d that the active form of gene 4 prote in is oligomeric , a nd recent evide nce h a s show n that th e protein form s h examers (22), bu t t he dep endenc e of gene 4 protein ca talytic activities on hexam er form ation ha s not been exa mine d. To determine if the defects in h examer form a tion have an effect on the var ious enzymatic activi ties of t he gene 4 protein , we compa re d t he biochemi cal properti es of the wildtyp e a nd mut ant pr oteins. Th ese activities include nu cleotide hydrolysis, h elica se activity, a nd primer sy nt hesis .
N ucleotide Hydrolysis-The enzymatic activities of T7 gene 4 pro t ein r equire a hydrolyzabl e NTP, with the preferred subst rate in vitro bein g dTTP; mor eover , nucl eotid e hydrolysis is stimula te d greatly by ss DNA (34). Accordingly, we assayed the conve rs ion of dTTP to dTDP by t he wild-ty pe a nd mutant protein s in th e pr esenc e a nd abse nce of ss DNA (Fig. 4). For ease of comp arison , th e K", and V IO U K for each protein were determined from t he dat a show n in Fig. 4 and are give n in Tabl e V. Th e wild-typ e, H47 5A, an d D485G gene 4 protein s hydrolyze dTTP in a react ion th at is st imu la te d by ss DNA. Th e act ivity of the R487A protein , however, is inhibited by ss DNA, a result that will be discu ssed below . In t he abse nce of ss DNA, the VIOU K of the three gene 4A protein s , wild-typ e, H475A , a nd D485G are similar. Th e a ddition of ss DNA stimula tes the act ivity of the wild -typ e protein a pproximate ly 25-fold wh erea s the H47 5A and D485G protein s are sti mula te d only 7.6-and 18-fold, respecti vely. Furthermore, the K", values for dTTP of t he H47 5A and D485G protein s are close to that of the wildtyp e protein an d do not ch ange significa ntly between rea cti ons with and without ss DNA. Th e simila rity in the va lues for the K", a nd the VIOUK in the abse nce of ss DNA, together with the lower VIOUK in the presenc e of ss DNA, indicat e th at th ese two DNA . In the presenc e of f3,y-me thylene dTT P, the hi gh est concentration of wild-typ e gene 4A protein (2.0 JLM) was able to retain all of t he lab eled 35-me r presen t in th e assay (Fig. 5). However, th e hi gh est concen tra tion s of t he H47 5A and D485G pro t eins (a lso 2.0 JLM) bound on ly 30 a nd 55% of the ss DNA, resp ecti vely. Binding of ssDNA by the R487A pr ot ein was FIG . 4. E ffect of increasing su bstrate co ncent ration on nucleotide h ydrol ysis by t he w il d -type a n d mut a n t T7 ge ne 4A p r oteins in the p r e s enc e a n d absence of ssDNA. Th e reactions wer e performed as described under "Experi mental Procedur es." All react ions conta ined gen e 4A prot ein a t 200 nM, an d the nucleotide concentrations were as indi cated . A , th e nucleotid e hyd rolysis reacti ons cont a ined 50 11M M13mp6 ssDNA (nucleotide equivalents). B, reactions wer e perform ed as in A bu t in the absence of ss DNA. The curves a re labeled: WT , wild-type; HlA , H475A; DIG, D485G; and RIA , R487A. Each cur ve repr esents th e average of two experime nts , a nd each experiment was perform ed in duplicate. mutati on s, H475A a nd D485G , do not directly affect the ability of th e ge ne 4 protein s t o bind and hydrolyze dTTP. Rath er , consi de ring the effect t hese mut a tions h a ve on oligome riza t ion , the y suggest t h at the r educed hydrolysis activity in th e presence of ss DNA is th e r esult of fewe r active h examer s. The velocity of th e dTTP hydro lysis r eactio n catalyzed by t he R487A protein decr ea ses in t h e pr esence of ss DNA, the opposit e of the reaction observed with wild-type protein. Thi s apparent in hibition of activity by ss DNA wa s confirmed in experime nts wh ere nucleoti de hydrolysi s wa s mea sured befor e and after the a ddition of ssDNA to r ea ctions wit h t he R487A protein. Th e rate of hydrolysis by t he R487A protein decrea sed immediately a fte r t he a ddition of ssDNA to the r eaction mi xtures (da t a not sh own ). In addition , this mutant protein h as an almost 5-fold hi gh er affinity for dTTP than t he wild-ty pe protein a n d form s hexa mer s as well as t he wild-ty pe protein . Th e resu lt s show that th e ability of th e mutant protein to hydro lyze NTP is in tact, but the R487 A mutation in t erfer es with the mech anism by wh ich the gene 4 protein s within t he h exam er interact in t he presenc e of ss DNA t o coordinately hydro lyze nucleotides .
ssDNA Binding-DNA bindin g by t he gene 4 protein is depen dent on the ability of the protein to bind nucleotides (40) a nd form hexamers (4 1). We used a gel-sh ift assa y with 32p _ lab eled oligonucleotides to mea sure ssDNA bin ding beca use it als o reveals the oligom eric nature of the protein bound to the T AB LE IV Percent of T7 gene 4A wild-type and mutant proteins as hexamers versus smaller oligonucleotides a s determ in ed by EM Th e wild-type and mu tant 63-kDa gene 4 pr otein s wer e pr ep ar ed for EM as described under "Experime ntal Procedu res." Th e perc ent of th e negat ively sta ined pro teins formi ng hexam ers was det ermin ed by count ing the number of hexam ers versus sma ller mu ltimers; a pproxima tely 300 protein comp lexes wer e counte d for each se t (refer to Fig. 3). Th e proced ur e was rep eated at least two ti mes , and th e sta nda rd error of th e mean is given in paren th eses.
Nucleotide" a Th e prot eins wer e a t 10 Ilglml (175 nxt), " Each nu cleotide was presen t at 0.6 mxt. Th e K m for dTTP a nd the Vm a x for th e conver sion of dTTP to dTDP plu s P; wer e det ermined from th e data shown in Fig. 4

Gene 4 Protein (liM)
t he R487A pr otein bound ra diola beled (dT)2o approximately 75% as well as t he wild-type gene 4 protein (indica te d by the arrowheads in Fig. 6). Cross -lin king occurred only when both nucl eotide a nd ma gn esium are pre se nt in th e reaction mixt ures, th us dem onstrating a specific in teraction bet ween the ge ne 4 protein a nd t he oligon ucleoti de. Th e lab eled oligonucleotide prim arily reacted wit h a single monomer of gene 4A pr ote in (closed arrow, Fig . 6); however, a fraction of t he oligon ucleotide was cross -lin ked to multiple gene 4 pr otein monomers (open arrow , Fig. 6). It is uncl ear if this res ult was du e to mul tipl e pr otein s cross -linke d to a single oligonucleoti de or cross-linked pr otein s bound to a single oligonucleot ide. The nucl eotide presen t in the assay has a slight influence on ssDNA binding; both wild-ty pe a nd R487A pr otein s were cross -linked to th e lab eled oligon ucleoti de only 86% as well wit h dTTP as with {3,v-methylene dTTP. Thi s result confir ms that t he R487A pr otein interacts wit h ssDNA, bu t does not reveal t he relative strength of the in ter acti on. It may be that this mu tan t has a te n uous hold on ss DNA t ha t is not sufficiently strong to mai ntain contact during t he gel-shift assay, but is strong eno ugh to inhibit nu cleotide hydrolysis. Helicase Acti vity-To determine if helicase activity is intact in the mu tan t pr oteins, we used a n oligonucleoti de substrate to measure dsDN A unwinding. Th e helicase substrate cons ists of a 75-base oligonucleot ide with a pa rti ally compleme ntary radiolab eled 25-ba se oligonucleot ide a nnealed to its 3'-end (see inset to Fig. 7). Th e 5'-17 bases of the 25-mer base pai r with t he 3 '-17 bases of the 75-mer, leaving t he 3'-8 bases of the 25-mer and the 5 '-58 bases of t he 75-mer as ss DNA. Th e 25-mer is 32P-labeled at its 5 '-end so that its migr ation in th e gel can be detected by autora diography. Strand sepa ration is determined by measuring t he cha nge in the amount of radio lab eled 25-base oligonu cleotide t hat migr ates as ss DNA. Th e relati ve amount of un winding act ivity show n by the wild-ty pe a nd mu tan t pr ote ins (Fig. 7) is pr oporti onal to the activities observed in th e ss DNA bindi ng (Fig. 5) a nd nu cleotide hydro lysis reac tions (Fig. 4A ). Th e H475A a nd D485G pr otein s wer e not able to se pa rate t he oligonucleoti des as efficiently as th e wild-ty pe pr otein . Thi s is cons istent with the fact th at the mu tan t protein s do not bind ss DNA as well as the wild-type protein.
Unwinding activity by th e wild-ty pe pr otein does not reach 100% becau se th e conce ntration of ss DNA in creases as the reaction pr oceeds a nd eve ntually exceeds t he concentration of  barely detectabl e. Comparison of the auto ra diogr aphs and t he Coomassie Blue-stained gels revea led th at t he rad iolab eled ssDNA migr ated togeth er with gene 4 protein complexes of a hexamer or greater in size (dat a not shown). Th e rela tive binding ability of the H475A a nd D485G proteins is proporti onal to t he ir ability to form hexam er s, suggesting that th ese mu tant pro te ins bind DNA poorly du e to defects in protein-protein interaction an d not to a re duction in DNA binding affinity . Th e fact that the R487A pr otein bound ss DNA so poorly as measured by t he gel-shift as say was pu zzlin g. Thi s mu tant protein form s hexam ers as well as the wild -type pr otein a nd must interact with DNA since its dTTPase activity is inhibited by ss DNA. One possible explanation for th e poor binding is th at {3,v-rnet hylene dTT P im parts a confor mation to t his mu tan t protein that differs somewhat from that imp arted by dTTP. However , this gel-shift assay can not be used wit h dTTP, becaus e t he gene 4 protein t rans locates off the ss DNA a nd is th en sepa rated from the DNA du ring electrophoresis . Conseque ntly, to exa mine DNA bind ing by t he R487A pr otein with greate r se nsitivity an d determine if there are differen ces attri buta ble to t he nu cleotide pr esent, we employed a UV-me dia te d crosslinking as say (Fig. 6). Wit h this assa y we demonstrated t hat the helicase substrate, effectively competing for the enzyme. The R487 A protein cannot hydrolyze dTTP in the presence of ssDNA and, as anticipated, could not separate the strands of the helicase substrate (Fig. 7).
Primase Activity-In addition to the activities examined thus far, the gene 4A protein catalyzes the synthesis of tetraribonucleotide primers essential for the replication of the bacteriophage genome. To assess the effect the mutations described here have on primase activity, we used an assay that couples primer synthesis to DNA synthesis catalyzed by the T7 DNA polymerase. Consequently, this assay measures both the ability of the proteins to synthesize oligonucleotides and to provide functional primers for the DNA polymerase. Primase activity catalyzed by the wild-type protein reaches peak activity at approximately half the maximum protein concentration used in the assay (Fig. 8). None of the mutant proteins stimulated the same level of DNA synthesis even at the highest protein concentrations used in the assay. Each mutant protein, however, is able to prime DNA synthesis indicating that they are all capable of catalyzing the synthesis of tetraribonucleotides that can be extended by T7 DNA polymerase. The level of DNA synthesis measured in reactions with each protein is proportional to the levels of activity observed in each of the biochemical assays previously presented. This result is consistent with the loss of activity as a result of changes in the ability of the mutant proteins to form hexamers and in turn interact with ssDNA. DISCUSSION The hexameric nature of the T7 gene 4 protein was previously demonstrated by EM, gel filtration analysis, and chemical cross-linking experiments (22,38). Studies with a T7 gene 4 protein having mutations in its NBS support this physical evidence (16). In these latter studies, we exploited the ability of the NBS mutant gene 4 protein to inhibit ssDNA-dependent nucleotide hydrolysis by the wild-type protein to investigate the stoichiometry of the gene 4 protein complex on ssDNA. Both this inhibition reaction and the stoichiometry of ssDNA binding indicated that the gene 4 protein was active as a hexamer. The ability of the NBS mutant protein to inhibit completely the activity of the wild-type gene 4 protein also demonstrated the importance of cooperative nucleotide hydrolysis within the hexamer.
In the present study, we have identified mutations in the gene 4 protein that affect its ability to form hexamers and to interfere with coordinated interactions between the subunits of the hexamer. The mutations lie within a conserved domain in the carboxyl-terminal region of the protein. This domain spans amino acid residues 475 to 500 and shares sequence similarities with the gene 4 proteins of phages T3 and SP6, the DnaB proteins of E. coli and Salmonella typhimurium, the phage T4 gp41 helicase and others (refer to Table I of this report and Ilyina et al. (27». Moreover, the core of this region, T7 gene 4 protein residues 481 to 491, is highly conserved in each protein of the DnaB helicase family, many of which form hexamers. Our finding that this conserved domain is responsible for oligomerization and protein-protein interactions required for nucleotide hydrolysis on ssDNA suggests that this related group of NTP-dependent hexameric helicases may share a common mechanism for translocation on ssDNA and unwinding dsDNA.
Each of the mutations within this domain affected the ability of the gene 4A protein to complement a T7 phage lacking gene 4. The H475A and D485G mutations decreased T7~4-1 plaque size and number, and the R487 A mutation prevented growth of this phage. EM analysis revealed that each of the mutant gene 4 proteins form hexagonal rings, and, morphologically, these hexamers were indistinguishable from those formed by the wild-type protein. Further analysis of the mutant proteins revealed that the amino acid substitutions cause two distinct yet related changes in the properties of the gene 4 protein. The H475A and D485G mutations decrease the ability of the proteins to form hexamers, and the R487 A mutation affects the ability of the gene 4 protein hexamer to use the energy of nucleotide hydrolysis for translocation on ssDNA.
The H475A and D485G proteins have significantly lower ssDNA-dependent dTTPase activity than does the wild-type protein. Since nucleotide hydrolysis catalyzed by the gene 4 protein is stimulated by ssDNA, and since only gene 4 protein hexamers bind ssDNA (data not shown, refer to Ref. 41), it appears that the reduced dTTPase activity of these mutants is due to less efficient hexamer formation. Consistent with this is the finding that the loss of nucleotide hydrolysis activity resulting from these mutations is proportional to their decreased ability to form hexamers as measured by native PAGE (refer to Table III). The results of the ssDNA binding and helicase assays also support the conclusion that the primary defect caused by the H475A and D485G mutations is a lower proteinprotein binding affinity and not a defect in nucleotide binding or hydrolysis. Based on these findings, we conclude that the activities of the gene 4 protein, ssDNA-dependent nucleotide hydrolysis, ssDNA binding, translocation on ssDNA, and DNA strand separation are all dependent on hexamer formation.
The R487 A mutation affects the activity of the gene 4 protein in a different manner. This mutant protein forms hexamers as well as wild-type protein and hydrolyzes nucleotides better than wild-type in the absence of ssDNA. However, nucleotide hydrolysis by this protein is inhibited, instead of stimulated by ssDNA. In addition, the R487A protein does not bind ssDNA tightly or exhibit any ability to unwind dsDNA. Together, these findings strongly suggest that the R487 A mutation affects the ability of the monomers within a hexamer to interact properly during nucleotide hydrolysis and translocation on ssDNA.
The 63-kDa gene 4 protein, by virtue of its unique aminoterminal domain, also catalyzes template-dependent synthesis of tetraribonucleotides (10). Each mutant protein is able to synthesize primers for T7 DNA polymerase, indicating that the mutations in this conserved carboxyl-terminal region of the protein do not directly affect its ability to function as a primase. Similar results were observed in our analysis of a NBS mutant gene 4A protein (17). This latter mutant protein could not hydrolyze dTTP, and therefore could not translocate on ssDNA, but it could synthesize template-directed primers, presumably through random interaction with DNA.
The D485G mutation found in our original clone of gene 4 protein can be attributed to the lethality of the wild-type protein to E. coli. In fact, in the process ofrecloning wild-type gene 4, we found that a high frequency of clones contained mutations, at least one of which was defective in nucleotide hydrolysis (data not shown). The primary defect caused by this mutation resulted in our initial inability to demonstrate the oligomeric nature ofthe gene 4 protein by gel filtration (12,15). In this report, we have shown that the D485G protein retains all of the catalytic properties of the wild-type protein, but in each assay the specific activity of this mutant protein is lower than that of the wild-type protein. Nonetheless, the overall observations made with this cloned mutant protein are consistent with those made with gene 4 protein purified from phageinfected cells. This is not a surprising result since the mutant protein does support T7 replication and growth (14,37). Therefore, we do not believe that any of our earlier results obtained with this protein will differ significantly from those of the wild-type protein. Rather, we believe that certain reactions requiring tight protein-protein interactions, such as the coupling of DNA polymerase and helicase/primase activities at the T7 DNA replication fork (35) will be augmented.
A recent EM analysis of the gene 4 protein revealed that ssDNA passes through the center of the gene 4 protein hexamer (38). It was also demonstrated that the E. coli RuvB branch migration protein, which has helicase activity, forms double hexameric rings around DNA (25). The ring structure of these protein complexes raises the intriguing question of how these hexagonal rings load onto the DNA. Since the gene 4 protein tightly binds circular ssDNA (40) we can rule out mechanisms requiring loading via free ends. Therefore, in order to bind ssDNA, the gene 4 hexamer must either assemble around the DNA or the preformed hexameric ring must open to load onto the ssDNA. As observed with other hexameric helicases, such as DnaB and T4 gp41 (18,42), nucleoside triphosphate binding facilitates hexamer formation by the gene 4 proteins (22). In this study, we also observed an increase in hexamer formation by the gene 4 protein upon binding dTTP ( Fig. 2D and Tables III and IV). The product of the hydrolysis reaction, dTDP, does not induce hexamer formation to the same extent as does dTTP (Fig. 2B and Table IV). This result suggests that NTP binding leads to a conformational change in the gene 4 protein that increases the protein-protein binding affinity and that conformational changes during the hydrolysis of the NTP to NDP decreases the association affinity. When bound to ssDNA, these conformational changes are rapid and result in translocation. In the absence of bound ssDNA, the hexamer will have more time to partially dissociate due to the relatively slow rate of the hydrolysis reaction. The decreased protein self-association affinity that occurs as dTTP is hydrolyzed to dTDP may therefore be a necessary step for the hexameric ring formed by the gene 4 protein to open and bind ssDNA.
The ssDNA binding experiments demonstrate that nucleotide binding and hexamer formation are not sufficient for tight DNA binding, nor is nucleotide hydrolysis required for the gene 4 protein to bind ssDNA, since the nonhydrolyzable analog {:l,y-methylene dTTP promotes strong binding (40). The R487A protein forms hexamers as well as wild-type protein and binds ssDNA almost as well in the cross-linking assay. It may be that the hexamers formed by the R487 A protein interact with DNA, but cannot undergo the conformational changes required for the hexamer to "grip" the DNA and translocate. Consequently, it slips off the end of the oligonucleotide during electrophoresis, and, thus, binding cannot be detected in the gel-shift assay.
We find it of interest that the two closely spaced mutations D485G and R487 A have such differing effects on the oligomerization and enzymatic activities of the gene 4 protein. The residues Asp485 and Arg 4 8 7 are at the center of the core region of homology with the DnaB family of helicases and so suggest a common mechanism of action for hexameric helicases. Thus, we speculate that interactions mediated by this domain between the subunits in a hexamer may be responsible for the presence of three high and three low affinity NBS observed in studies of the DnaB hexamer (43). This hypothesis is supported by the close proximity of the residues directly involved in the protein-protein interactions. It is very likely that the protein domain responsible for communication of conformational changes between the hexamer subunits during nucleotide hydrolysis and the domain responsible for protein-protein interactions would be integrated.
It should be noted that the gene 4 proteins H475A and D485G, while defective in hexamer formation, do not migrate as monomers in native PAGE analysis. Although it is possible that the fastest migrating protein band visible on the silverstained native gels is actually a monomer with aberrant migration, this seems unlikely since a "Ferguson" analysis (44) of both the protein standards and the estimated molecular weight of each of the visible gene 4 protein bands generates a linear plot. Additionally, small zone gel filtration analysis of gene 4 protein at various concentrations in the presence and absence of nucleotide detected only two species of oligomer, estimated to be dimer and hexamer (22). If the proteins interact in a "headto-tail" configuration, it is difficult to imagine how mutations that influence hexamer formation would not affect dimerization. This head-to-tail configuration predicts a single interface, the disruption of which would lead to the appearance of monomers (Fig. 9). On the other hand, if dimer formation is repre- sented as a "head-to-head" interaction of the monomers, and the dimers then interact through "tail-to-tail" contacts to form hexamers, we would predict two types of protein-protein interface (Fig. 9). Mutations at one interface would affect hexamer formation, but would have no influence on dimer formation. Precisely such a model was proposed by Dong et at. (18) for nucleotide-induced hexamer formation by the phage T4 gp41 helicase. They observed the formation of dimers at low protein concentrations and hexamers at higher concentrations and proposed the existence of two interfaces with different protein association strengths. The gene 4 protein-protein interactions appear to be much stronger than those ofT4 gp41 as hexamers of gene 4 protein were detected in the absence of nucleotide by native PAGE at protein concentrations of2 J-LM and by electron microscopy at very low protein concentrations (175 nsr), whereas T4 gp41 does not form hexamers at low protein concentrations unless a nucleotide is present (18).