32 Protein Trypsin-generated Fragments FLUORESCENCE MEASUREMENT OF DNA-BINDING PARAMETERS*

The intrinsic fluorescence of the T4 helix-destabiliz-ing protein specified by gene 32 (32P) is not altered by the proteolytic removal of either the 6200-dalton COOH-terminal “A” region (32P*-A) or both the A and the 2300-dalton NHz-terminal ‘?B” region (32P*-(A+B)). The intrinsic fluorescence of 32P, 32P*-A, and 32P*-(A+B) is decreased 23% by the addition of d(pTh and 34% by the addition of poly(dT). Saturation binding curves of the percentage of change in protein fluores- cence as a function of nucleotide concentration show that the intact 32P as well as the two proteolysis-gen- erated fragments all have association constants of -lo6 M-’ for d(pT),+ This demonstrates that the DNA binding site is not contained within either the A or B regions of 32P. Both 32P and 32P*-A bind cooperatively to poly(dT) as evidenced by a 400- to lOOO-fold increase in association constant for poly(dT) compared to d(pTb. Since within the limits of our measurements 32P and 32P*-A equally to


T4 Gene 32 Protein
Trypsin-generated Fragments The intrinsic fluorescence of the T4 helix-destabilizing protein specified by gene 32 (32P) is not altered by the proteolytic removal of either the 6200-dalton COOH-terminal "A" region (32P*-A) or both the A and the 2300-dalton NHz-terminal '?B" region (32P*-(A+B)). as evidenced by a 400-to lOOO-fold increase in association constant for poly(dT) compared to d(pTb. Since within the limits of our measurements 32P and 32P*-A bind equally well to poly(dT) (K,,,,, -5 l 10' M-l), the enhanced helix-destabilizing properties previously reported for 32P*-A cannot be accounted for by a significant increase in binding affinity of 32P*-A for single-stranded DNA. The binding constant for the 32P*-(A+B):poly(dT) complex is only 3-fold higher than that for the 32P*-(A+B):d(pT)R complex, which confirms our proposal that the B region is essential for cooperative 32P:32P protein interactions.
In the preceding paper (1) differential scanning microcalorimetry was used to examine the role of the COOH-terminal "A" region and the NH*-terminal "B" region of the bacteriophage T4 gene 32 protein. ever, a question remained as to whether the B region also contributes directly to the binding of 32P to DNA as suggested by Tsugita and Hosoda (2) or only indirectly by enhancing 32P:32P interactions in the presence of DNA. We have tried to resolve this question by measuring the binding affinity of 32P and the proteolytic fragments derived from it to d(pT)* and poly(dT) using fluorescence-quenching measurements.
This technique which has been used by Kelly et al. (3) in studying the binding of intact 32P to a variety of oligo-and polynucleotides requires only small quantities of protein and ligand and is thus particularly suited to the problem.
We report here the results of these studies and discuss their relationship to the various models proposed for T4 gene 32 protein action.

MATERIALS AND METHODS
Gene 32 Protein-The purification of gene 32 protein and its partial proteolytic cleavage products is discussed in detail in the preceding paper (1 constants we report are averages of three experiments and varied by a maximum of a factor of two. The error range was largely a function of differences in the activity of the protein preparations.

Proteolytic
Removal of the A and B Regions From Gene 32 Protein Does Not Change Its Intrinsic Fluorescence-Kelly and von Hippel (4) have demonstrated that the intrinsic fluorescence of gene 32 protein is characterized by an excitation maximum at 280 nm and an emission maximum at approximately 350 nm whose intensity and spectral position is a function of the immediate environment of the tryptophan residues. Since all five tryptophan residues are located in the 32P fragment that has had the COOH-terminal A and NHzterminal B regions removed (32P*-(A+B)), it was expected that the fluorescence properties of 32P, 32P*-A, and 32P*-(A+B) would be similar. This is indeed the case (Fig. 1). The fluorescence emission intensity (at 350 nm) of the two trypsingenerated fragments was the same as that of the intact protein (on a molar basis), and the similarity of peak shape and maxima position for 32P, 32P*-A, and 32P*-(A+B) suggests that the tryptophan residues are in similar environments in each protein species. Hence, by this criterion, the removal of the A and B regions does not alter the conformation of the core region of gene 32 protein.
Removal of the A and B Regions Does Not Affect the Affinity of the Proteolytic Fragments for d(pT)"-The partial quenching of gene 32 protein fluorescence produced by nucleotide binding (4) was used to measure the affinity of 32P, 32P*-A, and 32P*-(A+B) for d(pT)+ The choice of d(pT)" as a ligand was dictated by the desire to distinguish the domains of 32P which are involved in DNA binding from those involved in the cooperative protein:protein interactions which lead to a tighter binding to DNA. The nucleotide-binding site size for 32P has been reported to be approximately six nucleotides (3,8), and the binding of 32P to d(pT)", unlike the binding to shorter oligonucleotides, results in a large quenching of protein fluorescence (4). Fig. 2  measuring the percentage of change in protein fluorescence (100 AF/F') as a function of d(pT)" concentration.
The maximum fluorescence quenching was 21 to 24% for both 32P and the proteolytically derived fragments (Table I), suggesting that d(pT)x interacts with or influences the environments of the tryptophan residues of each protein in the same manner. This is additional evidence that the structural integrity of the core region of 32P remains unchanged after partial proteolysis.
To ensure that the quenching of protein fluorescence which we have measured results solely from the binding of d(pT)R to the proteins, parallel experiments were performed in standard buffer containing 2 M NaCl, conditions which do not allow 32P to bind d(pT)". As anticipated, under these conditions the addition of saturating amounts of d(pT), to 32P did not result in a decrease in fluorescence.
The association constants for d(pT)h binding to 32P and its proteolytic derivatives were calculated from the fluorescence quenching data by analysis of double reciprocal and Scatchard plots, as detailed under "Materials and Methods." The double reciprocal plots for 32P, 32P*-A, and 32P*-(A+B) derived from the binding curves are given in Fig. 3. The results of the double reciprocal analysis, summarized in Table II  M-' iv-1 32P 6.0 x 10" 7.3 x 10" 6.2 32P*-A 9.8 x 10" 3.7 x 10" 6.4 32P*-(A+B) 1.2 x 10" 3.4 x IO" 6.3' " n, stoichiometry of binding (nucleotides/protein).
h Since we were unable to determine n for 32P*-(A+B) due to the low K,,, we have assumed a value based on n for 32P and 32P*-A.
for d(pT)* is not reduced by removal of either the A or both the A and B regions. binding to poly(dT).
Poly(dT) was chosen as a ligand because previous data have shown that 32P binds more tightly to poly(dT) than it does to other synthetic polynucleotides (9). The average maximum fluorescence quenching produced by poly(dT) binding to gene 32 protein and its derived fragments, as summarized in Table I, was similar to that reported for the binding of 32P to d(pT)lc and to denatured calf thymus DNA (4).
The saturation binding curves for the association of 32P, 32P*-A, and 32P*-(A+B) with poly(dT) are shown in Fig. 4. The stoichiometry of poly(dT) binding for 32P and 32P*-A was 6.3 + 0.1 nucleotides per protein monomer, which is in accord with the findings of Kelly et al. (3,8), and Peterman and Wu (lo), but is smaller than that reported by Anderson and Coleman (ll), who determined binding stoichiometry by titrating changes in nucleic acid conformation induced by 32P. Exact determination of intrinsic association constants from the fluorescence titration data is difficult due to the tight cooperative binding of 32P to poly(dT). For this reason we have employed the approximation of Kelly et al. (3) for deriving "apparent" association constants (see under "Materials and Methods").
Association constants determined from titration curves in this manner are approximations and represent the lower limits of their values. The apparent binding constants for 32P and 32P*-A binding to poly(dT) are 7.3 x 10" M-' and 3.7 X 10H M-', respectively (Table II). These values are in good agreement with the reported apparent association constants 2.4 x 10' M-' for 32P binding to poly(dA) (3) and 1.9 x 10R Mm' for binding to fd DNA (10). The apparent association constants for 32P and 32P*-A binding to poly(dT) are, therefore, essentially equivalent, thus indicating that removal of the A region does not significantly affect the strength of this interaction.
The cooperativity component of the binding to DNA can be evaluated by expressing the association constant as the product of the intrinsic association constant, which is the affinity of the 32P monomer for a single site consisting of 6 nucleotides, and the cooperativity parameter (0) which reflects the increase in the observed association constant upon contiguous binding of the protein to DNA (7). Comparison of the association constants for binding to poly(dT) with those for binding to d(pT)R indicates that the cooperativity factor (w) for both 32P and 32P*-A is 0.4 + 1 X lo". This value of o is in good agreement with that reported by Kelly  stantially to the affinity for DNA. In contrast, the association constant for 32P*-(A+B) binding to poly(dT) is 3.4 x lo6 M -I, two orders of magnitude less than the association constants for the 32P:poly(dT) and 32P*-A:poly(dT) complexes, with the result that the w value for the 32P*-(A+B):poly(dT) complex is only 3. This confirms quantitatively that removal of the B region virtually eliminates the ability of 32P*-(A+B) to bind cooperatively to DNA.

DISCUSSION
Moise and Hosoda have previously shown (12) that the interaction of gene 32 protein with DNA is profoundly affected by the proteolytic removal of the NH;?-terminal B region or the COOH-terminal A region. Loss of the B region from 32P results in decreased affinity for DNA while loss of the A region from 32P enables the resulting protein to bind dsDNA cellulose (12) and to denature T4 DNA (13), properties not exhibited by the intact 32P.
We have proposed that the B region is essential for cooperative binding of 32P to ssDNA (l), based on the results of differential scanning microcalorimetry of 32P, 32P*-A, and 32P*- (A+B) in the presence of poly(dT). In order to quantitatively estimate the relative contribution of the B region to the 32P:32P and to 32P:DNA interactions it was necessary to compare the binding constants of 32P*-A and 32P*-(A+B) for d(pT), and for poly(dT).
We have shown that 32P*-A and 32P*-(A+B) have similar association constants for d(pT)" which confirms that the B region does not contribute to the strength of 32P:DNA interactions.
In contrast, 32P*-A binds two orders of magnitude better than 32P*-(A+B) to poly(dT). This loo-fold difference in binding affinity is best accounted for by the contribution of the B region to 32P:32P interactions. The exact nature of the involvement of the A region of 32P in DNA binding is not clear. Since 32P*-A but not 32P can denature double-stranded T4 DNA (12), Moise and Hosoda suggested the A region is important for controlling the helixdestabilizing ability of 32P. Presumably, interaction of the A region with other proteins in the replication complex might have the same effect as the proteolytic removal of the A region and thus provide a means for localizing the helix-destabilizing activity of 32P to the region just in front of the replication complex (12). Based on its ability to denature dsDNA and on the calorimetry data in the preceding paper (l), we expected 32P*-A to have a higher affinity than 32P for poly(dT). However, using fluorescence titration no significant difference was found in the association constants for the 32P:poly(dT) and 32P*-A:poly(dT) complexes. The ability of 32P*-A to denature dsDNA is, therefore, either due to an increase in affinity for ssDNA which is too slight to detect using the approximations described by Kelly et al. (3) to determine "apparent" binding constants or it results from some other property of 32P*-A. Two other possible explanations for the enhanced helix-destabilizing activity of 32P*-A compared to 32P are that unlike 32P, 32P*-A binds to double helical DNA and subsequently forces the two strands apart or, more likely, that proteolytic removal of the A region of 32P removes the "kinetic block" that Jensen et al. suggest keeps 32P from denaturing T4 dsDNA in uiuo (8). That is, 32P*-A has a faster "on rate" for binding ssDNA than 32P and for this reason 32P*-A can trap transiently "open" single-stranded regions of native DNA which 32P is unable to do for kinetic reasons. We plan to examine the kinetics of 32P and 32P*-A binding to DNA, to test the validity of this hypothesis and its relation to an in uiuo mechanism for the control of the helix-destabilizing activity of 32P.