Overexpression, Purification, and Characterization of Recombinant T4 Gene 32 Protein22,301 (g32P-B)*

Gene 32 protein (g32P), the replication accessory protein from bacteriophage T4, is a zinc metalloprotein which binds with high cooperativity to single-stranded (ss) nucleic acids. The basic N-terminal 2 1 amino acids (termed the “B” domain) is required for highly coop- erative (W = 500) binding of g32P monomers to ss nucleic acids. As part of our studies to systematically evaluate the structural features of the B domain important for cooperative binding, a homogeneous source of g32P which binds noncooperatively to nucleic acids (Cd = 1) and is devoid of contamination by native g32P is needed. we N-terminal

Gene 32 protein (g32P),' encoded by gene 32 of bacteriophage T4, is a single-stranded (ss) nucleic acid binding protein which binds to regions of ssDNA formed transiently during replication and repair processes (1,2). At intermediate binding densities and nucleic acid excess, gene 32 protein monomers will tend to cluster on both naturally occurring and synthetic homopolymeric ssDNA and RNA lattices, a binding mode thought to be important for preparing the ssDNA in a conformation suitable for the enzymatic machinery in uiuo, e.g. DNA replication, as well as afford protection of the single strand against degradation by intracellular nucleases. Such cluster formation at equilibrium is a consequence of the high cooperativity of binding, and is contained within the cooperativity parameter, w, reported to range from about 200-2000 (1). The apparent equilibrium association constant (I&,,,) of g32P for a polynucleotide lattice is described as Kapp = Kobs. w, where Kobs is the intrinsic association constant of a g32P monomer for an isolated lattice binding site of n (where n = 7-10) nucleotides (3).
G32P (301 amino acids, M, = 33,487) is a multidomain protein of known primary (4) and undefined tertiary structure. Three functional domains become apparent from limited trypsinolysis studies (5,6). The C-terminal "A" domain (residues 254-301) makes heterologous contacts with other proteins in an active replication complex, including the DNA polymerase and accessory proteins associated with both leading strand and lagging strand DNA synthesis (7,8). Tryptic cleavage of this domain from g32P to form g32P-A also removes a kinetic ' N-terminal Deletion of T4 Gene 32 Protein block in the ability of g32P to destabilize natural doublestranded DNAs; equilibrium (thermodynamic) binding parameters remain relatively unchanged (9). The N-terminal basic or "B" domain (residues 1-21) is required for g32P to bind with high cooperativity (10). This conclusion has been reached on studies carried out with g32P lacking both the A and B domains, termed the trypsin-resistant DNA-binding core fragment or g32P-(A + B); g32P-(A + B) has been shown to bind noncooperatively (w = 1) to poly(dT) (11). The core fragment coordinates an intrinsic Zn(I1) ion, which a variety of spectroscopic studies indicate is coordinated by the side chain S-atoms of three Cys (residues 77, 87, and 90) and a fourth non-sulfur liganding donor atom, proposed to be His" on the basis of recent 'H NMR studies (12)(13)(14). The Zn(I1) ion provides structural stabilization to the core domain of the molecule, as shown by the reduced thermal stability (15) and greatly enhanced susceptibility of the core domain to proteolysis (12, 16).
The molecular mechanism of cooperative binding by this prototype ss nucleic acid binding protein remains obscure, aside from the fact that the N-terminal 21 amino acids (and probably as few as residues l-9) (17,18)  for molecular cooperativity, we require a well characterized g32P species completely lacking this region of the molecule and devoid of even trace amounts of native intact g32P. In this paper, we describe an overexpression and purification strategy which provides a rich source of homogeneous tryptic N-terminally deleted g32P, g32P22-301, or g32P-B, heretofore unavailable.
We verify that the recombinant protein has the expected tertiary structure on the basis of limited proteolysis studies and 'H NMR spectroscopy. We show that the protein is monomeric over a wide range of protein concentrations.
Finally, we describe for the first time quantitative ss polynucleotide binding studies with tryptic g32P-B which provides insight on the energetic contribution that the C-terminal A domain makes to the protein-nucleic acid complex in this macromolecular system.  1. A, construction of the expression plasmid for g32P-B, termed pT7g32-B.wt (5.5 kb). As described under "Materials and Methods," the g32P coding sequence cloned into a ss phage vector was first submitted to oligonucleotide-directed mutagenesis which introduces a unique NdeI (..CATATG..) restriction site at nucleotides +58-63 which correspond to codons 20 and 21. Upon subcloning the so-generated l-kilobase pair NdeI/BamHI fragment into vector PET-3b (4.5 kilobase pairs), the ATG specifies the new initiation codon of the N-terminally deleted g32P-B gene product. G32P-B is under transcrintional control of the T7 RNA nolvmerase promoter. B, SDSpolyacrylamide gel electrophoresis anaiysi of the time course profiles of total cellular protein taken from liquid cultures at the indicated times following induction (0 min) of the following gene products: g32P, E. coli HBlOl transformed with the plasmid pPLg32.N20H:K21M, which expresses the double mutant full-length g32P obtained after introduction of the described (A) NdeI site. In this construct, the l.l-kilobase pair NheI/BamHI fragment is cloned into the same sites of the expression vector, pTL9W (46). Expression is from the phage X PL promoter induced by temperature jump from 30 to 40 "C exactly as described previously (13); g32P-B, E. cob BL21(DE3) transformed with pT7g32-B.wt; lpE7'-361, E. coli BL21(DE3) transformed with the parent vector, PET-3b. The electrophoretic migration of purified g32P is indicated in the left-most he.

RESULTS
transcribe its own promoter sequences. Such derepression is conveniently attained by addition of isopropyl-P-D-thiogalactopyranoside to the exponentially growing culture. Fig. lB, lanes 7-10, shows the profile of cellular proteins from BL21(DE3) transformed with pT7g32-B.wt obtained prior to (0 min) and 30, 60, and 120 min following addition of isopro-pyl-P-D-thiogalactopyranoside. This is compared with the same cells which harbor the parent vector, PET-3b (23) (lanes [11][12][13][14]. Also shown for comparison (lanes 2-6) are the induction profiles of Escherichia coli HBlOl transformed with a plasmid which inducibly expresses a full-length g32P from the phage X PI, promoter (13). Note that the cells which contain pT7g32-B.wt specifically express a protein with a relative molecular mass some 2500 daltons less than that of the full-length g32P, consistent with the predicted molecular mass of 30,834 daltons for authentic g32P-B. We note the level of induction can be significantly improved upon this experiment, with some experiments approaching g32P-B levels upwards of 30% of the cellular protein and 50-60% of the soluble protein by weight (cf. "Materials and Methods" and Fig. 2B).
As described in the Miniprint, recombinant g32P-B is highly soluble in cells which express it. Upon lysis of cells by the resident T7 lysozyme after a single freeze-thaw cycle, the protein is readily purified to homogeneity from the low-speed supernatant by a combination of anion exchange (Fig. 2), ssDNA-cellulose ( Fig. 3), and hydrophobic interaction (e.g. phenyl-Sepharose) ( Fig. 4) chromatographies.
All g32P-B protein preparations following the phenyl-Sepharose step are 299% homogeneous as shown by gel electrophoresis of large quantities of protein (-25 pg) and are free of detectable singlestranded DNA endonucleases (see "Materials and Methods"). The only visible contaminant in g32P-B preparations, if present at all, comigrates with the g32P-(A + B) core fragment, likely obtained by in situ proteolysis of g32P-B during purification (see below). Automated N-terminal sequencing of g32P-B reveals NH?-Gly"-Phe-Ser-Ser-Glu-Asp-Lys-Gly-Glu-Trp3', the expected primary structure (4) provided the initiator Met from the cloning vector is cleaved upon/after biosynthesis.
Typical yields of highly purified g32P-B range from 4 to 10 mg/g induced cells. The purification is not especially noteworthy (see "Miniprint"), as we and others utilize the precise sequence of chromatographic steps to purify intact g32P (24). The ssDNA-cellulose column step, however, requires some comment (see below).

Recombinant g32P-B Possesses a Native Globular Structure
Limited Proteolysis-The N terminus of any given protein may potentially influence the folding pathway from the nascent polypeptide chain in vivo, especially if the reaction involves the concerted folding of distinct structural domains in a multidomain protein.
Although nothing is known of the folding pathway for g32P, we do know that the native structure contains distinct structural domains, the B domain being one. Considerable effort was therefore expended to substantiate that the recombinant N-terminally deleted g32P-B possesses a wild-type structure, both with regard to specific features of the DNA-binding core and the A domain, as well as the physical relationship of these two domains to one another.
A simple, albeit low resolution, indication of the expected tertiary globular structure in g32P-B takes advantage of the resistance to proteolysis of the core domain of the molecule (5, 6). The electrophoretic migration of the g32P-(A + B) core fragment is shown in the left-most lane. Each incubation (100 ~1) contained 7.16 I.~M g32P (or g32P-B) and 0.01 mg/ml trypsin (-25:l w/w g32P:trypsin ratio) in 10 mM Tris-HCI, pH 8,0.1 M NaCI, 5% v/v glycerol, 16 "C. lo-p1 aliquots were withdrawn at the indicated times and added to ice-cold SDS-polyacrylamide gel electrophoresis gel loading buffer, mixed, and immediately heated at 75 'C for 5 min and returned to ice until electrophoresis. These are conditions known to quench the proteolysis reaction (12).
'H NMR Spectroscopy-Further evidence for a native structure is provided by 'H NMR spectroscopy of g32P-B. The 'H NMR spectrum of wild-type g32P is dominated by resonances of the C-terminal A domain (25, 26). This is due to the fact that g32P aggregates extensively which severely broadens most of the resonances; the A domain on the other hand experiences faster average motion than that described by the aggregate, resulting in an NMR spectrum comprised of some rather sharp resonances superimposed on a broad featureless spectrum. In contrast to this situation with the intact protein, the g32P-B derivative should show reduced if not eliminated aggregation (see below) which should enhance the overall resolution of the 'H NMR spectrum.
The 400 MHz 'H NMR spectrum of g32P-B (Fig. 6A) is compared with that of g32P-(A + B) (Fig. 6B). Even with g32P-B, there exists a subset of narrow resonance lines superimposed on less well resolved resonances. This is clearly seen, for example, in the aromatic region at 7.2-7.3 ppm where a set of sharp overlapping resonances assignable to protons of the three A domain Phe ring systems, and therefore specific to g32P-B, are found. Similar groups of A domain resonances, e.g. at -4.45, ~3.85, -2.90, ~2.60, and = 0.85 ppm, can easily be seen throughout the aliphatic region as well. This spectrum provides compelling evidence that in g32P-B, the A domain is conformationally more mobile than the rest of the molecule, giving rise to a relatively small set of narrow lines, whose chemical shifts exactly correspond to those found with the A domain of the intact protein (26), albeit better resolved here. In addition, careful inspection of the remaining spectral features associated with the core domain in both g32P-B and g32P-(A + B) spectra indicate common single resonance or groups of overlapping resonances on the chemical shift axis. For example, the single peak at 5.52 ppm, which likely corresponds to the 3,5 protons of TyrR in Pan et al. (26), is clearly present in both spectra. However, this and all other common core domain resonances generally appear broader in the g32P-B sample (cf. the aromatic envelope, from 6 to 8 ppm), consistent with its 23% greater mass, but also perhaps reflecting a greater overall globular asymmetry, and thus larger effective reorientation time, as revealed by hydrodynamic measurements (see below). In any case, Both spectra were recorded at a protein concentration of 0.4 mM in 50 mM NaPi, pH 8, 30 mM NaCl at 22 "C. The sharp peak at ~4. 8  and intact proteins (25-27). Analytical Gel Filtration of g32P-B-It has been reported previously that both g32P-B' (obtained by limited proteolysis by Staphylococcal protease following amino acid Glu' in the B domain)3 and g32P-(A + B) obtained by limited proteolysis of g32P migrate on gel filtration columns (e.g. Sephadex G-100) as monomers at low salt and near neutral pH (17). However, there are no reports on the effects of protein concentration or salt concentration, data essential to correctly interpret the equilibrium binding studies described below. To quantitatively assess the solution aggregation properties of the recombinant g32P-B protein, we performed analytical HPLC gel filtration studies with a Waters Protein Pak 300s~ column. In Fig. 7, we plot the observed elution coefficients (K,,) of g32P, g32P-B, and g32P-(A + B) obtained at ambient temperature in 10 mM Tris-HCl, 0.1 mM EDTA, 0.1 M NaCl, pH 7.24 as a function of protein concentration as described ' Conditions can be found which utilize staphylococcal protease to preferentially cleave after Glu' to give the previously described g32P-B' (17,18 under "Materials and Methods." In the protein concentration range extending from 2 X 1Om8 to 5 X 10M6 M, we document that g32P-B and the core g32P-(A + B) fragment give the same K,,, indicating a constant molecular species. In contrast, as already demonstrated by previous sedimentation experiments (28), the intact protein begins detectable apparent aggregation at 20.2 pM. Knowing that at protein concentrations sloe7 M, the intact protein is a monomer (28), the observed KBV value (0.420 -t 0.004) must represent a limiting value, i.e. that of a monomer. Since the monomer molecular weight is known precisely (4), one can estimate the apparent Stokes radius a, and frictional coefficient f/f,, for the monomeric g32P, by assuming an average partial specific volume (u = 0.73 cm3/g) as outlined under "Materials and Methods." Since the g32P-B protein also exhibits a K,, similar to the intact protein at very dilute concentrations, we assume that this value also incorporates the hydrodynamic properties of monomeric g32P-B. We adopt 0   Derived from the data presented in Fig. 7 as outlined under "Materials and Methods." Conditions: 10 mM Tris-HCI, 0.1 mM NaZEDTA, 0.1 M NaCl, pH 7.2, ambient temperature. ' Assuming a prolate ellipsoidal globular shape. similar reasoning for the core g32P-(A + B) fragment. Compiled in Table I are estimates of the apparent molecular weights, Stokes radii, frictional coefficients, and calculated axial ratios for g32P and derivatives, the latter assuming a prolate ellipsoidal globular structure as was done previously for the intact protein (29).
These data show that recombinant g32P-B migrates on gel filtration with an apparent particle size and shape characteristics similar to those of native g32P under conditions where both proteins are monomeric. All g32P derivatives are considerably asymmetric (Table I) (11)). In addition, KObs (= K.,) for g32P-B appears independent of input g32P-B concentration (from 0.2 to 1.6 pM) with the values incorporated in the range given in Table II. [NaCU Dependence of KobS for g32P-B and g32P-(A + B)-Proteins which bind to nucleic acids derive much of their binding free energy from an increase in entropy which occurs upon release of cations thermodynamically associated with the nucleic acid lattice resulting from electrostatic interactions (31). Analysis of the salt dependence of protein-nucleic acid binding equilibria provides molecular information on the 6 Table  II. We found no evidence that n for the g32P-(A + B) core fragment is as small as 5.0 2 0.5 as reported previously (11). We are presently exploring the question of site size for both g32P derivatives with short single-site oligonucleotides of varying length (1) by directly determining stoichiometries according to Bujalowski and Lohman (33). net number of ions (cation and anion) released concomitant with formation of the complex according to -8 log K&d log[NaCl] = m'0 + a (31). The first term encompasses cation release and screening effects primarily associated with the nucleic acid, whereas the second term incorporates preferential anion effects as the net number of anions displaced from the protein (31). With such an approach, we can ascertain if the lower Kobs of g32P-B for poly(dT) at low salt relative to the core protein is due to a differential net release of ions concomitant with complex formation.
We have obtained the [NaCl] dependence of the binding equilibria, ie. a log K,b.le log[NaCl], for both g32P derivatives via analysis of individual "salt-back" titrations where the protein-nucleic acid complex is formed at low salt and is gradually dissociated upon incremental increases in solution [NaCl], monitored by an increase in protein fluorescence (32). At each NaCl concentration, La and LF can be calculated from Equations 1-3 under "Materials and Methods," permitting calculation of Kobs at each NaCl concentration, according to Equation 4. This salt-back analysis also requires that Qma., n, and the cooperativity parameter w = 1 not change as a function of solution NaCl concentration (32). Analysis of reverse titrations carried out with g32P-B and g32P-(A + B) and poly(dT) at various [NaCl] reveal that all titrations can be fit satisfactorily with an o = 1 and n within the range indicated, whereas best-fit Q,., values do not vary systematically by &5% in either case (data not shown). Fig. 9, A   represent the theoretical curve obtained when those parameters derived from an analysis of the data cast in the linear format of Fig. 9C are recast, indicating an acceptable fit to the non-transformed experimental data, given the indicated uncertainty.  . (ll), -3.8 + 0.4. The new data for the g32P-B protein indicate a reduced salt dependence and thus a net release of fewer ions upon complex formation by this molecule. We see, however, that the different [NaCl] dependences for both species reduce the relative difference in Kobs for both proteins as the [NaCl] is increased such that the extrapolated log Kabs (1 M Na') values are quite similar to one another within experimental error. In other words, the unfavorable effect of the A domain on core domain nucleic acid binding in g32P-B at low salt is clearly relieved at higher monovalent ion concentrations.
In Table II, we also show preliminary data on the binding of g32P-B and g32P-(A + B) to the ribohomopolymer, poly(U), to which native g32P binds with much less affinity (20). The trends for poly(U) are qualitatively as outlined for poly(dT) binding. Although both molecules bind less tightly to poly(U) than poly(dT) (Table II), g32P-B is characterized by a greatly reduced binding affinity (ZIOO-fold) at low salt and a salt dependence considerably less than g32P-(A + B) core fragment under the same conditions of pH and temperature. Thus, the qualitative trends we see with poly(dT) appear independent of the base and sugar composition of the nucleic acid molecule and thus must reflect generic features of the noncooperative linear lattice binding mode of g32P.

DISCUSSION
Cooperative binding by g32P to ss nucleic acids requires the N-terminal B domain as evidenced by quantitative equilibrium binding experiments carried out with the tryptic core g32P-(A + B) fragment on the polynucleotide, poly(dT), i.e. w = 1 or noncooperative binding (11). Analogous quantitative measurements of equilibrium binding parameters for g32P-B' remain unreported, although qualitatively the binding affinity of g32P-B' for poly(dT) (17,18)

and poly[d(A-T)]
(16) is of lower affinity and generally nonstoichiometric under low salt conditions, relative to the intact protein. As the mechanism of cooperative binding by this prototype ss nucleic acid binding protein remains obscure, we intend to elucidate the functional importance of individual amino acids in the B domain, by systematically creating a large library of B domain point mutations to be functionally characterized in the context of the entire molecule. In order to initiate these experiments, we require a predictable and large-scale source of a g32P molecule completely lacking only this domain and one devoid of contamination by the intact protein.
In this report, we describe overexpression, purification, and characterization of the N-terminal tryptic deletion fragment of T4 gene 32 protein, g32PZ2-301 or g32P-B. The purified protein has the expected and desired primary structure in the N-terminal region, NH*-Gly2*-etc. and contains stoichiometric Zn(I1). We provide evidence from 'H NMR and hydrodynamic experiments that the recombinant protein has tertiary and quaternary structural properties indistinguishable from the proteolytic fragments obtained in the usual way via partial proteolysis of intact g32P. The described means of obtaining homogeneous g32P-B represent a considerable improvement over the techniques of protein chemistry which rely on partial and specific proteolysis by trypsin of only the N-terminal B domain after Lys*l with the linkage connecting the core domain and the C-terminal A domain (after LYS*~~) remaining intact. In fact, this particular molecule, g32P22-301 is impossible to make with trypsin, as the A domain is preferentially removed prior to the B domain.4 In all cases, the possibility persists that these preparations can be contaminated with either trace amounts of intact g32P or the respective core fragment, all of which exhibit widely disparate binding affinities (Ref. 11; this work). Reflecting these limitations, at least in part, the physical and nucleic acid binding properties of g32P lacking only all or part of amino acids 1-21 are only qualitatively described (17,18).
We demonstrate quantitatively that g32P-B binds noncooperatively to polynucleotides. However, the most striking result from the current studies is that the equilibrium binding affinity of g32P-B for poly(dT) and poly(U) is significantly reduced relative to core fragment derived from it via proteolytic cleavage of the C-terminal A domain. Thus, the A domain contributes unfavorably to the binding free energy of the core nucleic acid binding domain to an isolated lattice site. We also show that this unfavorable contribution is diminished as the solution salt concentration is raised, such that the nonelectrostatic component of the binding free energy for both g32P derivatives would appear to be similar (Table II). However, since the anion component to the salt dependences thus far remain undefined for both proteins, we cannot as yet conclude that the unfavorable influence of the C-terminal A domain is exclusively of electrostatic origin.
Our results with g32Ps (with or without the A domain) which lack the B domain and therefore bind noncooperatively topolynucleotides (Table II), are qualitatively consistent with previous studies which examined the noncooperative binding of intact g32P versus g32P-A to short oligonucleotides of maximum length (1) nucleotides less than that able to span more than one protein binding site (e.g. d(pT)s) (11,19). With oligonucleotides of length, 1 < 6 nucleotides, both g32P and g32P-A exhibited similar affinities and salt dependences (11). However, with 1 2 6, g32P-A showed a detectably greater affinity at low salt concentration and enhanced salt dependence relative to the native protein (11). For example, g32P-A binds to d(pA)s at 0.1 M NaCl with a Kobs = 5.0 x lo6 M-' and a d log &,Jd log [NaCl] = -1.3 + 0.1, whereas the same values for the intact g32P are ~1.6 X 10' M-l and z-0.3 + 0.1, respectively.7 Evaluation of K,,bs at various higher NaCl concentrations permitted construction of a log-log plot which upon tenuous extrapolation to 1 M Na' standard state revealed a similar nonelectrostatic contribution to the binding free energy for both proteins. Since the anion component to the binding free energy appeared negligible in this system, Lonberg et al. (11) suggest that the effect of removing the A domain is to increase the binding affinity for these oligonucleotides manifest entirely through electrostatic interactions. This would further imply a greater number of cations (~2) displaced from the oligonucleotide in g32P-A, relative to the native protein (~1). Our preliminary results with the singlesite oligonucleotide, d[T(pT)7] (1 = 8), also indicate similar disparities in equilibrium affinities of g32P-B and g32P-(A + B), assuming a 1:l oligonucleotide:protein stoichiometry. 8 Quantitative determinations of equilibrium binding parameters of g32P and g32P-A, on polynucleotides mirror the results with oligonucleotides in that Kobs becomes 2-3-fold higher upon proteolysis of the A domain (11) with no detect-' Kobs values were obtained assuming an oligonucleotide:protein monomer stoichiometry of 13. a D. Giedroc and R. Khan, unpublished observation.
able change in the cooperativity parameter (11); however, a log K&d log[NaCl] values are indistinguishable for both proteins and significantly more negative (m-6.5 + 1.0) than those obtained with the B domain deletions (Table II). From these studies, a structural model was developed in which the negatively charged A domain forms all or part of an arm or "flap" that forms electrostatic interactions with a cluster of positively charged side chains on the core domain (l-3, 19), which prevents g32P from adopting a highly saltdependent polynucleotide-type binding mode with oligonucleotides 1 I 8. Some or all of this cluster of cationic side chains, previously occluded by the A domain, are destined to interact electrostatically with the acidic phosphodiester backbone when bound to polynucleotide. Thus, a structural transition in g32P from the so-called oligonucleotide to polynucleotide conformation can be conceptually visualized as movement of all or part of the A domain such that an electrostatic sub-site and thus the entire nucleic acid binding groove becomes uncovered. In this model, the binding affinity of A domain containing g32Ps is simply weaker because fewer electrostatic interactions are formed in the complex and/or energy must be expended to "move" the flap to one or more environments. Structural evidence for such a conformational change is the observation that the A domain becomes more susceptible to proteolytic cleavage upon binding cooperatively to polynucleotides (6); recent 'H NMR experiments are also consistent with a faster average mobility of A domain resonances in the cooperative binding conformation (26).
As pointed out by Burke et al. (8), this modular feature of the A domain with respect to the rest of the bound g32P molecule would appear to be well suited to hold the helixdestabilizing activity of g32P in check by providing a kinetic block to uncontrolled DNA duplex melting (9). Such inhibition could be selectively relieved only when needed, e.g. nearest the point of duplex unwinding in an active assembled replication complex carrying out fully coordinate DNA synthesis. Here, the energy required to move the A domain on a small subset of g32P molecules might come from heterologous protein-protein interactions with accessory proteins, e.g. the gene 41/61 primosome complex and the DNA polymerase holoenzyme on the lagging and leading strands, respectively (47, 48).
Since the binding of g32P to homopolynucleotide lattices is expected to model many of the fundamental features of the productive binding conformation in the replicative complex, our data would indicate that g32Ps which specifically lack the B domain experience considerable difficulty in adopting the conceptualized "polynucleotide binding conformation," in clear contrast to native g32P. We thus provide evidence consistent with the scenario that some of the additional binding free energy derived from B domain-dependent cooperative interactions on polynucleotides be used to drive or facilitate an A domain conformational change such that the respective structural transitions or roles of the A and B domains are energetically coupled. Again, this makes good biological sense from the standpoint that as a result of high affinity, cooperative binding by g32P monomers (which requires the B domain), the A domain becomes liberated to appropriately enable or disable any number of different activities. As to why the marked unfavorable influence of the A domain at equilibrium was not fully appreciated in previous studies, it could simply be that by removing only the Nterminal B domain, we are able to quantify polynucleotide binding of g32P at salt concentrations low enough such the effect is quite dramatic and easily measured. At these low concentrations of salt, both g32P and g32P-A simply bind too tightly (or stoichiometrically) to poly(dT) and poly(U) such that &,,, cannot be measured accurately with fluorescence techniques. For example, at more moderate [NaCl] (e.g. 0.4 M), the reported difference in Kobs for g32P and g32P-A bound in the cooperative binding mode on polynucleotides (11) would appear to be comparable to the difference in affinities which we measure for g32P-B and g32P-(A + B) bound noncooperatively under similar solution conditions. Watanabe (21) has recently suggested that A domain function may, in fact, be intimately incorporated into the cooperative binding mode of g32P. Data were presented which showed that the cooperativity parameter exhibits a weak salt dependence, becoming slightly larger (by =2-fold) as the solution [NaCl] is raised. Titrations carried out with NaF gave indistinguishable values for w, whereas MgClz seemed to facilitate cooperative interactions.
He suggested that the negative charge of the A domain must be effectively "screened" by cations in order for g32P to bind tightly (cooperatively) to polynucleotides.
Thus, although not physically associated in the thermodynamic sense, solution cations may enhance the equilibrium binding affinity of g32P by reducing the repulsion of negatively charged A domains on contiguously bound g32P monomers through a simple ionic strength effect (21). In this line of thinking, a trivial interpretation of our binding data thus far might be that the binding energetics of g32P-B simply represent the sum of such a screening effect (or weak cation uptulze component) specific to the A domain and the particular energetics which characterize the core fragment-polynucleotide complex.
Clearly, additional experiments beyond the scope of the current work are required to further elucidate the role of the A domain in molecular terms with regard to linear lattice binding by g32P. Such experiments are facilitated by the B domain deletion molecules since cooperativity is eliminated from the binding equilibria.
As we show, this permits direct determination of affinities and salt-dependences of the isolated site binding mode under a wide variety of conditions, vastly reducing the complexity of the system. We point out that very recent high resolution 'H NMR experiments which carefully examined the isolated site (mimicked by g32P-(A + B) binding to d(pA)40-60) and the cooperative binding modes (intact g32P complexed with d(pA)40-60) reveal that the protein-nucleic acid interfacial region must be very similar in both conformations (26). Studies are underway which systematically alter the nature of the cation and anion species of the dissociating salt, providing molecular information as to whether the observed differences in Kob between g32P-B and only marginally reduced relative to that obtained in NaCl (11). Our experiments with g32P-B as well as g32P-(A + B) should extend these observations as well as provide additional detail concerning the generic mechanisms of high affinity linear lattice binding by this prototype ss nucleic acid binding protein.
Acknowledgments-We thank Dr. T. M. Lohman and members of his laboratory for assistance in analyzing the fluorescence experiments in the early stages of this work as well as their ongoing interest in these studies.
We also thank Drs.