Structural and Enzymatic Studies of the T4 DNA Replication System I. PHYSICAL CHARACTERIZATION OF THE POLYMERASE ACCESSORY PROTEIN COMPLEX*

In this study, we have investigated the structural and physical properties of the bacteriophage T4 DNA polymerase accessory proteins. We find that T4 gene 44 and 62 proteins associate to form a tight, highly homogeneous complex, containing four gene 44 protein subunits and one gene 62 protein subunit. The molec- ular mass of the complex is 163,700 daltons. Sedimentation results suggest that the complex is quite asym- metric, with a prolate ellipsoid axial ratio of about 5: 1. This protein complex is known to carry a DNA-de-pendent ATPase activity; we show by photoaffinity labeling that the ATP-binding sites reside in the gene 44 protein subunits of the complex. Equilibrium sedi- mentation and chemical cross-linking studies indicate that the T4 gene 46 protein self-associates to form a trimer in solution. This trimer species also appears to be quite asymmetric, showing an axial ratio for a pro- late ellipsoid of about 6:1, assuming normal hydration. The bacteriophage T4 gene 43, 44, 62, 45, and 32 proteins, known to be required for DNA replication in uiuo, can be formed into a reconstituted complex that is able to carry out leading strand DNA synthesis in vitro (1,Z). The “polymerase accessory proteins,” encoded by genes 44, 62, and 45, show a DNA-dependent ATPase activity (3, 4), and in the presence of ATP these proteins greatly stimulate the enzymatic activity of the polymerase (5-7). Mechanistic studies of functional replication complexes require a structural understanding of the component (22) and silver-stained by the method of Morrisey (23). Azido-ATP Photoaffinity Labeling-Photoaffinity labeling was performed using the ATP analog 8-azidoadenosine 5"triphosphate (NaATP). [T-~'P]N~ATP was obtained from Du Pont-New England Nuclear, and unlabeled NSATP was purchased from Schwarz-Mann. NsATP-containing solutions were irradiated with an ultraviolet lamp with a 302-nm emission peak (Spectronics model BLE-1T158). Sam- ples were mixed in a microtiter dish (Falcon) in volumes of approximately 20 pl and irradiated for 4 min at a distance of about 3 cm. The labeled products were analyzed by SDS-PAGE, with the addition of 6 M urea to the gel and sample buffer.

accessory proteins interact with the DNA template and the other proteins within the replication complex.
T4 gene 44 and 62 proteins associate to form a tight complex, dissociable only under denaturing conditions (3, 9,10). Both of the genes have been cloned (11,12), and the monomer molecular masses are known to be 35,584 and 21,347 daltons, respectively. Estimates in the literature for the size and composition of the complex vary (9,11,13). Thus, in order to understand the mechanism we must first define the stoichiometry of the complex. We have approached this by molecular weight determination, using velocity sedimentation, dynamic laser light scattering, and equilibrium sedimentation techniques. In addition, we have determined subunit composition directly by using reverse-phase HPLC' to separate the gene 44 and 62 protein subunits from the complex.
The gene 45 protein has also been cloned (14) and has a monomer molecular mass of 24,710 daltons. In order to understand how this protein interacts with the gene 44/62 protein complex, it is important to determine the association state of gene 45 protein both in solution and in association with the gene 44/62 protein complex. We have used equilibrium sedimentation and chemical cross-linking to determine the molecular weight of the associated species of gene 45 protein.

MATERIALS AND METHODS
Preparation of Proteins and Nucleic Acids"T4 gene 45 protein was prepared according to the method of Morris et al. (10) with the following modification. After the norleucine-Sepharose chromatography step, the protein was dialyzed into 50 mM NaC1, 1 mM EDTA, 10% glycerol, 1 mM /3-mercaptoethanol, and 20 mM Tris-HC1, pH 8.1, and run over single-stranded DNA-cellulose in series with Bio-Rex-70 (Bio-Rad), followed by a second DEAE-cellulose column (Whatman DE52). Under these conditions, the gene 45 protein passes through the first two columns and binds to DE52. It was then batch eluted from DE52 with a similar buffer containing 400 mM NaC1.
T 4 gene 44/62 protein was also purified according to Morris et al. (101, except that the protein was eluted from the hydroxylapatite column with a linear salt gradient (13). In addition, the final pool was dialyzed into I buffer (as defined by Morris et al. (lo)), loaded onto single-stranded DNA-cellulose, and eluted with a linear gradient from I buffer to I buffer plus 100 mM NaC1. The gene 44/62 protein complex eluted a t about 40 mM NaCl.
T4 gene 43 protein was purified according to Morris et al. (10). Escherichia coli rho protein and NusA protein were gifts from Johan- 12709 nes Geiselmann and Stanley C. Gill, respectively (this laboratory). All protein preparations were judged >98% pure based on SDS-PAGE and were shown to he free of contaminating endonuclease by incubation with supercoiled pBR322.
Poly(dA) was purchased from Pharmacia LKB Biotechnology Inc. Partial Specific Volume and Extinction Coefficient-Amino acid sequence information (11,12,14) allows us to calculate theoretical values for various physical properties of the accessory proteins, such as the molar extinction coefficient and the partial specific volume. In general, the implicit assumption in these calculations is that the physical characteristics of the individual amino acid residues will remain essentially the same in the native, folded protein as they are free in solution. While there are clearly examples of proteins for which this is not the case, the assumption proves surprisingly valid for the majority of proteins, and thus such calculations can provide valuable estimates of the physical properties of proteins of known sequence. The theoretical partial specific volume, 6, of gene 45 protein is 0.743 g/ml. This is based on known partial specific volumes of individual amino acid residues (17). Gene 44 and 62 proteins have calculated values of B of 0.739 and 0.746 g/ml, respectively.
Theoretical molar extinction coefficients for the proteins were calculated based on the number of tryptophans and tyrosines they contain, and using the molar extinction coefficients determined by Edelhoch (18) for these residues cz80 = N~,5690 + M~,,1280. (Procedures for carrying out such calculations and their limits of error, based on an extensive analysis of a basis set of experimentally determined values for a variety of proteins, have been described by Gill and von Hippel.)' N and M represent the numbers of tryptophan and tyrosine residues/protein monomer, respectively. The molar extinction coefficients are 1.91 X lo4, 2.29 X lo4, and 3.17 X 10' I/molcm for gene 45, 44, and 62 proteins respectively.
Velocity Sedimentation-Sedimentation coefficients for gene 45 and 44/62 proteins were determined by velocity sedimentation in a Beckman model E Ultracentrifuge. A 12-mm double sector cell with quartz windows was filled with 0.9 ml of sample and a slightly greater amount of reference buffer. The buffer for both proteins consisted of 200 mM KCI, 25 mM Tris-HCI, pH 7.4, and 0.4 mM dithiothreitol. The sedimentation boundary was monitored using a UV scanner set at a wavelength of 280 nm, and the boundaries were analyzed as described by Van Holde and Weischet (19). The initial gene 45 protein concentration was 0.38 mg/ml, and the rotor speed was 36,000 rpm. Gene 44/62 protein was sedimented at 34,000 rpm, with an initial concentration of 0.40 mg/ml. Light Scattering-The diffusion coefficient for the gene 44/62 protein complex was determined by dynamic laser light scattering using the 514.4-nm line of a n argon ion laser (Spectra Physics model 5). An intracavity etalon was used to ensure single frequency output. Scattered light was detected by a photomultiplier at scattering angles of 60", go", 120", and 135". Translational diffusion coefficients were obtained using a Langley-Ford model 1096 autocorrelator. The analysis was performed as described by Bloomfield and Lim (20). The sample buffer contained 2.5% (w/v) glycerol, 25 mM Tris-OAc, pH 7.5, 60 mM KOAc, and 0.5 mM dithiothreitol and protein at 1.1 mg/ ml. Sedimentation Equilibrium-Equilibrium ultracentrifugation was performed in a Beckman model E Ultracentrifuge, using the meniscus depletion method (21). The sample buffer for both gene 45 and 44/ 62 proteins contained 100 mM KCl, 25 mM Tris-HCI, pH 7.4, and 0.4 mM dithiothreitol. 0.13 ml of sample and 0.14 ml of reference buffer were placed in a double-sector cell with sapphire windows and interference window holders, and the cell was placed in an AN-D rotor. Interference optics were used, with 0.75-mm slits. Interference fringe photos were taken on Kodak Spectrograde 11-G photographic plates, and the fringes were analyzed on a Nikon model 6C Profile Projector. In general, after centrifugation for 40 to 48 h, the fringe patterns showed no further changes with time, and we judged that equilibrium had been attained. In all cases, the meniscus depletion condition was clearly achieved, as judged by the levelness of the fringes in the upper half of the cell image.
Reuerse-phase HPLC-Reverse-phase HPLC separation of the gene 44 and 62 proteins was performed on an Aquapore Butyl (C4) column (330 X 4.6 mm, 7 pm), using Beckman llOA Pumps, 421 Controller, 163 Variable Wavelength Detector, and a Spectra Physics SP4270 Integrator. The mobile phases contained 0.05% (v/v) trifluoroacetic acid (Applied Biosystems) and 0.05% (v/v) triethylamine (Fluka). The gene 44/62 protein was diluted from storage buffer (10) S. C. Gill and P. H. von Hippel, submitted for publication. 5-fold in 0.1% trifluoroacetic acid, and a total of 50 pg of protein were loaded onto the column at 15% (v/v) acetonitrile (UV grade, Brand). The subunits were eluted from the column over a 30-min period, using a linear 15-62% acetonitrile gradient with a flow rate of 0.35 ml/min.
Chemical Cross-linking-The chemical cross-linking of gene 45 protein was done with DMS (Pierce Chemical Co). The proteins were dialyzed into 40 mM HEPES, pH 8.1, 100 mM KCI, 0.2 mM dithiothreitol, and 10% (w/v) glycerol. The DMS was freshly prepared each time by dissolving in 40 mM HEPES, pH 8.1, and adjusting the pH to approximately 8.5 with NaOH. Reactions contained protein at 0.5-1.0 mg/ml and DMS at 5 mg/ml. Reactions were performed a t room temperature and quenched by addition of ethanolamine to a final concentration of 0.6 M. The products were analyzed by SDS-PAGE as described by Laemmli (22) and silver-stained by the method of Morrisey (23).
[T-~'P]N~ATP was obtained from Du Pont-New England Nuclear, and unlabeled NSATP was purchased from Schwarz-Mann. NsATP-containing solutions were irradiated with an ultraviolet lamp with a 302-nm emission peak (Spectronics model BLE-1T158). Samples were mixed in a microtiter dish (Falcon) in volumes of approximately 20 pl and irradiated for 4 min at a distance of about 3 cm. The labeled products were analyzed by SDS-PAGE, with the addition of 6 M urea to the gel and sample buffer.

Determination of the Molecular Weight and Subunit
Composition of the Gene 44/62 Protein Complex As noted above, the products of T4 genes 44 and 62 associate to form a tight complex. Subunit exchange studies (9) have failed to detect any dissociation and reassociation of the complex under native conditions, indicating that the complex, once formed, is quite stable. Although the molecular weights of the monomers are accurately known from the DNA sequence, the precise molecular weight of the total complex has only been estimated. Various stoichiometries have been reported for the subunits of this complex, ranging from a ratio of four gene 44 to two 62 subunits (9) based on Coomassie staining of the proteins eluted from denaturing gels, to a 5:l ratio (13), obtained by densitometry of a Coomassie-stained gel. Most recently, Spicer et al. (11) have reported a 3.6 (k 0.6): 1 ratio, based on quantitation of the size of the peaks obtained at each step of protein sequencing. This is probably the best estimate currently available, but leaves some uncertainty as to the exact subunit stoichiometry, and hence as to the molecular weight of the complex. T o firmly establish the composition of this multiprotein complex, we have re-examined the issue of subunit stoichiometry and determined the molecular weight of the complex by two different methods.
Molecular Weight Determination Using the Svedberg Equation-When a macromolecular experiences a centrifugal field, the rate at which it sediments will be directly proportional to its reduced mass and inversely proportional to its frictional coefficient. The frictional coefficient is influenced by factors such as shape and hydration and these same frictional forces will modulate the rate at which the macromolecule diffuses.  (1) allows the calculation of molecular weight using both the sedimentation and diffusion coefficients. Here M equals the molecular weight of the solute, R is the gas constant, T is the temperature, S is the sedimentation coefficient, D is the diffusion coefficient, d is the partial specific volume, and p is the solute density. The frictional factors cancel, and we can obtain a molecular weight that is independent of shape and hydration.
The molecular weight of the gene 44/62 protein complex was determined by the sedimentation and diffusion method. A sedimentation coefficient of 7.1 f 0.2 S was obtained by velocity sedimentation in an analytical ultracentrifuge.
A diffusion coefficient of 3.9 X cm2/s was determined by dynamic laser light scattering. The sedimentation and diffusion constants are corrected to 20 "C in water. The data were not obtained over a wide concentration range and thus are not corrected to infinite dilution. The samples are already quite dilute, however, and as noted by Van Holde (25), the S and S o (sedimentation coefficient extrapolated to infinite dilution) values for most globular proteins differ by only about 0.5% a t protein concentrations of 1 mg/ml.
In addition, several lines of evidence support the homogeneity of the species, including laser light scattering and sedimentation equilibrium. Therefore, we feel justified in combining the sedimentation and diffusion data using the Svedburg equation, to give a molecular mass of 170,000 daltons for the complex. This relies on theoretical values for the partial specific volumes of the proteins, based on amino acid sequence (see "Materials and method^").^ The sedimentation constant of 7.1 S for the gene 44/62 protein complex is in good agreement with that determined by Barry and Alberts (9) using sucrose density gradient sedimentation. Consistently low values of the "quality parameter", p/y2, for the light-scattering data show that the autocorrelation function fits well to a single exponential decay, providing a sensitive indication of sample homogeneity (20).
Determination of Molecular Weight by Equilibrium Sedimentation-The molecular weight of the gene 44/62 protein complex was further studied by equilibrium sedimentation, a technique capable of high precision. At sufficiently low rotor speeds, the centrifugal force that results in transport of the macromolecule by sedimentation will be balanced by diffusion transport in the opposite direction. An equilibrium condition is established, generating a concentration gradient throughout the cell. A homogeneous species, in an ideal two component system (i.e. solute and solvent) will show a concentration distribution described by: where c is the concentration of the solute, r is the distance from the center of rotation, and w is the angular velocity (for a review see Schachman (24)). The molecular weight values for the gene 44/62 protein complex, obtained by equilibrium sedimentation at two different rotor speeds, are shown in Table I. The results are in excellent agreement with those obtained by the sedimentation and diffusion method (see above). Calculation of the molecular weights relies on theoretical values for the partial specific volumes of the proteins, obtained from the known amino acid sequences. Since the experimentally determined partial specific volume of most proteins falls within 2% of the value predicted from sequence: this assumption is expected to contribute an uncertainty of up to &6% in the final calculated molecular weight, thus constituting the largest single source  of error in the molecular weight determination. Experimental error, such as uncertainties in rotor speed, temperature, and quantitation of the interference fringe patterns, can be estimated and brings the total uncertainty in molecular weight to a maximum of about +lo%.
The In c uersus r2 plots for equilibrium sedimentation experiments run at 14,000 rpm are shown in Fig. 1A. Since the fringe displacement is proportional to solute concentration, the slope of such a plot is proportional to the apparent molecular weight. The straightness of the line is indicative of the homogeneity of the complex. As noted by Van Holde (25), In c uersus r2 curves can be relatively insensitive to heterogeneity in some cases because the upward curvature produced by heterogeneity can be compensated by a downward curvature due to nonideality, making the line appear straight. The of the Accessory Protein Complex laser light-scattering experiments, however, are quite sensitive to heterogeneity and argue strongly in favor of a high degree of homogeneity for the gene 44/62 protein complex. In view of these results, the data of Fig. 1A can be most simply interpreted as representing a homogeneous complex behaving in a relatively ideal fashion. It should also be noted that the Yphantis (21) meniscus depletion method used here, although quite sensitive to low molecular weight contaminants, can entirely miss very high molecular weight species. Again, because the light scattering technqiue is very sensitive to high molecular contaminants, the results give us confidence that gene 44/62 protein in solution exists as a single species of molecular mass 170 f 20 kDa.
Determination of Subunit Ratios by Reverse-phase HPLC- The molecular mass studies described above have sufficient error to make it difficult to distinguish, for example, a complex containing four gene 44 subunits and one gene 62 subunit (molecular mass 164 kDa) from one containing four gene 44 subunits and two gene 62 subunits (molecular mass 185 kDa). Therefore, we have also investigated the stoichiometry of the subunits by using reverse-phase HPLC to separate and quantitate the individual protein components of the complex. The gene 44 and 62 proteins separate on a C4 column with a gradient of increasing acetonitrile concentration. The elution profiles are shown in Fig. 2. At least 90% separation of the subunits is achieved under these conditions. Gel electrophoresis of fractions collected from this column identify the species that elutes first to be gene 62 protein; the larger peak with the longer retention time corresponds to gene 44 protein.
The elution of the proteins from the column was monitored by a UV detector at two different wavelengths. The dominant chromophore at 220 nm is expected to be the peptide bond. Therefore, the relative areas at 220 nm have been corrected by the molecular masses of the gene 44 and 62 protein monomers (35,584 and 21,347 daltons, respectively) to give the calculated subunit stoichiometry. At 280 nm, the absorbance is principally due to tryptophan and tyrosine residues. Thus, the theoretical extinction coefficients for the gene 44 and 62 proteins at this wavelength were used to convert the relative areas into subunit stoichiometries. The resulting stoichiometries are shown in Table 11. Note that higher subunit ratios show incrementally smaller changes in the relative peak areas,  Integration of the absorbance peaks gives the relative areas corresponding to the gene 44 and 62 proteins. protein peak. This corresponds to 3.7 k 0.6 and 4.5 -+ 0.5, respectively, for the subunit ratios. These relatively small errors demonstrate the reproducibility of the data. There is, however, an uncertainty inherent in detection when monitoring the steep absorbance shoulder at 220 nm, increasing the total error in that measurement. The interpretation of the peak area determined at 280 nm relies on theoretical extinction coefficients (see "Materials and Methods"). This method of molar extinction coefficient calculation has been applied to proteins of known sequence and shown to yield an average standard deviation of less than k5% from the extinction coefficients of the proteins as determined by a variety of other methods.' Therefore, we have incorporated that uncertainty in estimating the final error in the 280-nm measurement.
making it difficult to distinguish, for example, a 4:l from a 5:l complex. This accounts for the asymmetric errors shown in Table 11. Although the errors are too large to allow us uniquely to specify the subunit stoichiometry, this technique is particularly good at distinguishing between complexes with low subunit ratios; the difference between a 2:l and a 3:l complex is quite large. Thus, we can feel confident in ruling out the 2:1 gene 44 to 62 protein ratio originally suggested by Barry and Alberts (9).
The Gene 44 and 62 Proteins Form a 4:l Complex-The molecular weight and subunit stoichiometry data described above for the gene 44/62 protein complex are summarized in Table 111. Several hypothetical subunit compositions are listed along with predicted molecular weights that span the range of compositions previously reported in the literature. The results of the HPLC experiments (determining subunit stoichiometry of the complex), and the sedimentation experiments (yielding complex molecular weight), are scored in terms of agreement with the proposed subunit composition. While the molecular weight determination could not unambiguously rule out the 3:2 and 4:2 stoichiometries, the HPLC data is clearly inconsistent with such low subunit ratios. Only the 4:l ratio is corroborated by both the sedimentation and HPLC data. Therefore, the combined evidence indicates that T4 gene 44 and 62 proteins form a tight, homogeneous species consisting of four gene 44 subunits, and one gene 62 protein subunit, with a total molecular mass of 163,700 d a l t~n s .

I11
Summary of evidence on the molecular weight of the native gene 44/62 protein complex Several hypothetical subunit ratios of the gene 44/62 protein complex are listed, along with predicted molecular weights. The list focuses on those ratios most consistent with the current data as well as estimates in the literature of the molecular weight of the native complex. The results from the HPLC separation of the subunits (Table 11), which yields subunit stoichiometries, are compared with those from the sedimentation and light scattering experiments, which provides molecular weight information. For each type of experiment, (+) indicates that the results were consistent with the subunit composition in question, (+) indicates that the composition is unlikely based on the experiment, and (-) indicates that the results of the experiment clearly rule out the subunit composition. Thus, it can be seen that a complex consisting of four gene 44 protein subunits and one gene 62 protein subunit, having a molecular mass of 163,700 daltons, is the only one consistent with both the HPLC and the sedimentation data. Shape Factors Indicate an Asymmetric Complex-We can make some predictions about the shape of the gene 44/62 protein complex on the basis of the sedimentation coefficient. A frictional coefficient for the complex can be calculated using the relationship: We can also calculate the frictional coefficient expected for an anhydrous sphere of the same size. Using this approach we obtain a maximum value of 1.45 for the Perrin shape factor, F (for a general discussion see Ref. 27). If we assume that the hydration of the gene 44/62 protein complex falls in the range of 0.3-0.4 g of water/g of protein, as is typical for most proteins, then the shape factor, F, is about 1.25. This corresponds to a prolate ellipsoid with an axial ratio of about 5:l. This agrees with the shape factor reported by Barry and Alberts (9) for this complex, based on sucrose density gradient sedimentation and gel filtration. Thus, it appears that the complex either has a very asymmetric shape or an unusually high degree of hydration. Since such a degree of hydration is very unlikely, a relatively asymmetric complex seems the most plausible explanation of this shape factor.

Gene 45 Protein Associates to Form a Trimer
Gene 45 protein has a monomer molecular mass, based on the DNA sequence, of 24.7 kDa (14). Previous literature reports had suggested that the protein associates to form a dimer in solution (10, 28) corresponding to a calculated molecular mass of 49 kDa. Gene 45 protein elutes with the void volume on a Bio-Gel P-60 column,'j for which the exclusion limit is approximately 60 kDa. This could be consistent with an asymmetric dimer model but is more likely to be indicative of higher states of gene 45 protein association. We have undertaken a more rigorous analysis of the physical state of gene 45 protein in order to ascertain whether the protein exists principally as a dimer, or as some higher associated species.

Sedimentation Equilibrium
Yields a Trimer Molecular Weight for the Associated Species of Gene 45 Protein-The results of the sedimentation equilibrium experiments on gene 45 protein are displayed in Table I. As was the case for gene 44/62 protein, the In c versus ? plot (shown in Fig. 1B) is quite linear. This indicates that the protein species is probably homogeneous, judging by the fact that the molecular weights obtained a t three different rotor speeds and two different initial protein concentrations are quite consistent. Any heterogeneity or nonideality of the solute should be revealed by comparison of runs made at different speeds and concentrations. The average molecular mass of the gene 45 protein from the three runs is 77,000 daltons, which is very close to the predicted trimer molecular mass of 74,100 daltons. Therefore, we conclude that the gene 45 protein exists primarily as a trimer in solution.
Chemical Cross-linking Shows Associated Species-The association state of gene 45 protein was additionally probed by chemical cross-linking, using a bifunctional reagent (DMS) that reacts with primary amines. Such diimidoesters have been used extensively as structural probes of multimeric proteins (29). It must be noted that a negative result from chemical cross-linking (ie. no cross-linkedproducts observed) is inconclusive because the spatial positions of lysine residues on adjacent subunits of an oligomer may not be favorable for cross-link formation. The formation of discrete higher molecular weight species, however, provides reasonably good evidence for the existence of multimeric protein species, presuming one can establish that the reaction conditions are unlikely to favor collisional cross-linking (ie. cross-linking between initially free protein monomers, rather than within pre-existing oligomers).
Samples of gene 45 protein were treated with DMS and the cross-linking reaction was quenched at various times by the addition of excess ethanolamine. The products of such a time course, separated on a denaturing gel, are shown in Fig. 3A. Clearly, distinct higher molecular weight bands can be seen after only a few minutes of cross-linking. Duplicate samples were analyzed on a 7.5% gel, and the mobilities compared with those of protein standards to obtain the apparent molecular weights of the products (not shown). The approximate molecular masses were 89, 76, 55, and 44 kDa, respectively, for bands A, B, C, and D. These values can only be used as rough indicators of the size of the cross-linked products since DMS cross-linked proteins frequently show anomalous migration on denaturing gels, presumably as a consequence of the branched chain structures that are formed by the intersubunit cross-links.
Clearly, species larger than dimers of gene 45 protein are being covalently linked by the DMS. Formation of crosslinked species stops abruptly at an apparent molecular mass of about 89 kDa, providing a rough estimate of the upper size limit for the gene 45 protein oligomer. The absence of yet higher molecular weight species cannot conclusively rule out larger species. However, as a positive control to determine that the cross-linking reagent is active, we have cross-linked E. coli rho protein, and shown that products as large as hexamers are formed. This protein associates.to form a hexamer under these buffer condition^,^ and has been shown by Finger and Richardson (30) to form a ladder of cross-linked products, from monomer to hexamer, when subjected to DMS.
It is also important to establish that the formation of crosslinked products reflects covalent linkage of subunits within the same oligomer, rather than collisional cross-linking. The J. Geiselmann, T. Yager, and P. H. von Hippel, manuscript in preparation. protein concentrations used in this study are low enough ( 4 mg/ml) to make the probability of collisional cross-linking very low (29). In addition, we have tested E. coli NusA protein, which has been shown to exist as a monomer in solution? Under identical reaction conditions, NusA shows no tendency to form collisional cross-links. Thus, we can feel fairly confident that the presence of higher molecular weight cross-linked species of gene 45 protein accurately reflects the association state of the species pre-existing in the solution? Although it is difficult to obtain accurate molecular weights of crosslinked species, the approximate molecular weight of the highest molecular weight band we observed falls between that expected for a trimer and that expected for a tetramer, clearly demonstrating associated species larger than a dimer. Thus, the cross-linking results corroborate the molecular weight determination by equilibrium sedimentation, which shows the associated species of gene 45 protein to be a trimer.
The cross-linking time course presented in Fig. 3A shows the appearance of a t least four major cross-linked species. We * S. C. Gill and P. H. von Hippel, manuscript in preparation.
No change in the cross-linking pattern or efficiency was seen on addition of DNA or ATP, indicating that the association state of the protein is probably not affected by binding of either of these potential cofactors (data not shown).  0.125,0.250,0.375,0.500,0.625,0.750, and 0.875 (see text). The least squares fitted lines for each value of w converge on an uncorrected S value of 4.2 f 0.1. This is corrected to 3.9 S for 20 "C. in water.
can postulate structures that might give rise to such bands on a gel, based on the trimer model for the gene 45 protein (diagrams in Fig. 3B). The uncross-linked monomer gradually disappears with time. Within 2 min, a significant amount of band C is seen, which probably results from a single crosslink between two adjacent subunits to form a dimer. Bands A and D form on a slightly slower time scale, appearing within 5 min, and might each result from the formation of two crosslinks. In the case of peak A, these cross-links could form a covalently linked trimer, while the two cross-links in the peak D species could join the same subunits, forming a doubly linked dimer species with altered mobility from that of the peak C form. Finally at long times (10 and 40 min) we see that the amount of peak A is depleted in favor of the more rapidly migrating peak B, which might result from formation of a third cross-link.
The schematic cross-linking diagrams of Fig. 3B are merely intended to represent subunit cross-linking connectivity and not to suggest that the overall trimer is quite compact. In fact (see following section) the actual trimer is quite asymmetric, and perhaps one simple way to visualize this in the context of Fig. 3B is to consider the cross-linking species as representing end-on views of quite asymmetric (cylindrical?) subunits.
Shape Information from the Sedimentation Coefficient-Velocity sedimentation of gene 45 protein yields a sedimentation coefficient, S20,w, of 3.9 -t 0.1 S. Fig. 4 shows the apparent (uncorrected) S values plotted uersus tlh for seven values of w = C(r)/Cp, the ratio of the concentration of protein in the ultracentrifuge cell at radius r to the plateau concentration. This analysis, described by Van Holde and Weischet (19), results in a "fan plot" for a homogeneous sample, because at infinite time, the apparent S, values converge to the same limit, S. Therefore, it appears that the trimer of gene 45 protein exists as a fairly homogeneous species.'O If we assume lo It should be noted that if the subunits of the trimer associate and dissociate on a rapid time scale, the resulting sedimentation boundary could appear to be homogeneous. In this case, the sedimentation constant obtained would be somewhat lower than that for the true trimer species, being an intermediate value between, for example, monomers and trimers. Since no indication of lower molecular weight species was seen in equilibrium sedimentation, however, we have no reason to believe that any significant fraction of the gene 45 protein exists as monomers or dimers under the solution conditions we have studied.
A. Coumassie-stained gel that the hydration is between 0.3 and 0.4 g water/g protein, then the Perrin shape factor for gene 45 protein would be about 1.32, corresponding to an axial ratio of 6 1 for a prolate ellipsoid. Thus, it appears that the trimer gene 45 protein may also be quite asymmetric in shape. Since the literature reports identifying gene 45 protein as a dimer were presumably based on sucrose density gradient sedimentation (28), this may explain the discrepancy between our results and those previously reported in the literature. Molecular weight determinations based only on sedimentation transport will tend to err on the small side when the complex in question exhibits a high degree of asymmetry.
The ATP-binding Site Resides in the Gene 44 Subunit The T4 polymerase accessory proteins have been shown to have an ATPase activity that is stimulated by DNA (for details and references, see (8)). Piperno et a1. (3), examined the identity of the ATPase protein by treating gene 44/62 protein and gene 45 protein in turn with 6-mercaptopurine ribonucleoside 5'-triphosphate, which reacts at the ATPbinding site of various ATPases and inhibits ATPase activity (31). Gene 45 protein treated in this manner is able to stimulate the ATPase activity of normal gene 44/62 protein, while modified gene 44/62 protein shows no ATPase activity in the presence of normal gene 45 protein. Thus, the functional site for ATP hydrolysis appeared to reside in the gene 44/62 protein complex.
We have further investigated the locations of potential ATP-binding sites on the replication proteins by covalently labeling the proteins with an ATP photoaffinity analog, NATP. Aromatic azido compounds such as N3ATP form highly reactive nitrenes when exposed to ultraviolet light. Photoactivation is possible outside the normal absorption range of most biological macromolecules (i.e. above 300 nm). Thus, absorption of the incident light responsible for photoactivation is likely to be minimal and photochemical reaction of the macromolecules themselves to be slight. Covalent bond formation between the affinity probe and a macromolecule is likely if the probe is actually bound at the moment of photoactivation. Ideally, unbound affinity probe will react with solvent, although earlier estimates of solution lifetimes on the order of milliseconds (33) have been challenged by the work of Staros (34), suggesting that the lifetimes of reactive intermediates may be on a many millisecond to second time scale. However, under proper reaction conditions photoaffinity labeling can nonetheless be made quite specific.
Gene 44 Protein Reacts Specifically with Azido-ATP-Samples of each of the five T4 replication proteins (gene 44, 62, 32, 45, and 43 proteins), alone and in combination, were irradiated with 300 nm light in the presence of [y-32P]N3ATP. The results are shown in Fig. 5. The autoradiogram shows that gene 44 protein reacts specifically with NSATP. Since the gene 44/62 protein complex is the only one among the group for which ATPase activity has been demonstrated, it is not surprising that one of its subunits binds ATP. This specific labeling is notably absent when no Mg+ is present. Since the substrate for ATP hydrolysis is Mg-ATP, this lends credence to the idea that N3ATP is binding to the true ATPbinding site in a manner similar to the normal substrate. Gene 62 protein shows virtually no labeling. Thus, it appears that the ATP hydrolysis activity resides in the gene 44 subunits of the complex. This agrees with the results of Lin et al.," showing that gene 44 protein purified from the cloned T4 gene has a low level DNA-dependent ATPase activity, while gene 62 protein (also purified from a clone) does not.
The only other protein to show specific photoaffinity labeling under these conditions is the polymerase (gene 43 protein). Although it lacks any detectable ATPase activity, the polymerase clearly must contain a deoxynucleoside triphosphatebinding site, and this site probably has considerable affinity for N3ATP. None of the other proteins show any significant labeling by the photoaffinity probe. In particular, the lack of labeling of gene 45 protein supports the conclusion of Piperno et al.
(3), that the gene 44/62 protein complex is responsible for the ATPase activity of the accessory proteins.
Evidence That Photoaffinity Labeling Occurs at the Normal ATP-binding Site-In order to demonstrate unequivocally that the observed labeling of gene 44 protein occurs specifically at (or near) the normal ATP-binding site, several criteria should be met. The most important criterion is that ATP can compete for the binding site, and reduce the amount of  Covalently hound N3ATP was separated from unbound material using spin columns (32). Radioactive incorporation was determined by scintillation counting in aqueous mixture. Each reaction contained gene 44/62 protein a t 1.1 PM complexes, and 4 pM N3ATP (3.3 pCi/ ml), in a buffer consisting of 10 mM HEPES, pH 7.5,50 mM NaC1,6 mM MgCl,, and 1 mM 0-mercaptoethanol. The poly(dA) concentration was 60 U M and the ATP concentration was 1 mM. labeling. We have addressed this by photoaffinity labeling the gene 44/62 protein complex in the presence and abesence of ATP (Table IV). The amount of N3ATP covalently incorporated upon photolysis is reduced by inclusion of ATP in the reaction mixture. This incorporation is not decreased when adenosine is added (not shown) suggesting that the effect is specific and not due to an artifact such as absorption of the photoactivating light by the added nucleotide. Inclusion of DNA in the reaction mix also does not inhibit covalent incorporation of N3ATP, so it is unlikely that a DNA-binding site is being labeled. These results suggest that the azido-ATP analog is binding in the normal ATP-binding site."

DISCUSSION
The T4 polymerase accessory proteins are an essential part of a fully functional DNA replication complex. We have shown that the gene 44 and 62 proteins form a complex consisting of four gene 44 protein subunits and one gene 62 protein subunit, Under our buffer conditions, the complex appears to be quite homogeneous. The anomalous sedimentation properties of the complex are indicative of an asymmetric ( i e . non-spherical) configuration. Although the current study does not address the number of ATP molecules bound/ complex, it is clear from photoaffinity labeling that the ATP-'' The argument for the specificity of the labeling is strengthened when certain additional criteria are met. First, when the level of analog is raised, the amount covalently incorporated should reach saturation and thereafter remain relatively constant. In addition, the analog should be usable as an enzyme substrate and show values of K , and VmaX similar to those of ATP. N3ATP has been shown to meet all of these criteria for several types of ATPases, but not in other cases, possibly due to the fact that the bulky azido group tends to force the molecule into the cis-conformation rather than the characteristic trans-conformation (35). In the gene 44/62 protein labeling studies, we have been unable to observe saturation of the photoaffinity labeling a t concentrations up to 1 mM N3ATP. N3ATP is also hydropresence of gene 45 protein and DNA. Although the K,,, for ATP of lyzed very poorly by the gene 44/62 protein complex, even in the gene 44/62 protein alone is not known, the K , in the presence of gene 45 protein and DNA is quite high (about 0.2 mM). In this concentration range, nonspecific labeling by N3ATP of proteins that do not normally bind ATP is prevalent (L. Paul, unpublished results). If the K,,, for N,ATP is similar to that of ATP, then saturation may not be observed due to nonspecific labeling. Although labeling may be specific at low concentration levels, nonspecific labeling is prevalent at the concentrations expected for saturation. binding sites reside in the gene 44 protein subunits. Thus, the complex has the potential capacity to bind up to four ATP molecules at once.
The gene 45 protein is seen to exist primarily as a trimer in dilute solution, contrary to previous reports in the literature (10,28). Since the association constant between the gene 44/ 62 protein complex and gene 45 protein is relatively weak (8) it has not been possible to isolate the complete accessory protein complex. Therefore, the stoichiometry of binding of gene 45 protein trimers to gene 44/62 protein complexes can only be surmised from enzymatic studies. This will be discussed in greater detail in the accompanying paper. In addition, the interaction of the accessory protein complex with primer-template junction DNA (the functional site for replication) will be discussed.