| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Molecular and Cell Biology, The University of Texas at Dallas, Richardson, Texas 75083-0688, USA
2 Department of Biochemistry and Molecular Biology, Colorado State University, Fort Collins, Colorado 80523, USA
3 Biosciences Division, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USA
Reprint requests to: Donald M. Gray, Department of Molecular and Cell Biology, Mail Stop FO31, The University of Texas at Dallas, P.O. Box 830688, Richardson, TX 75083-0688; e-mail: dongray{at}utdallas.edu; fax: (972) 883-2409.
(RECEIVED July 24, 2001; FINAL REVISION November 22, 2001; ACCEPTED November 22, 2001)
Article and publication are at http://www.proteinscience.org/cgi/doi/10.1110/ps.30002.
 |
Abstract |
|---|
 TOP Abstract IntroductionResultsDiscussionMaterials and methodsReferences |
|---|
Keywords: Circular dichroism; Ff gene 5 protein; tyrosine CD bands; protein–protein cooperative interactions; secondary structure analysis
Abbreviations: CD, circular dichroism • DSSP, Define Secondary Structures of Proteins computer program (Kabsch and Sander, 1983) • 
, L - R (or molar CD) • Ff, filamentous phages (M13, fd, and f1) that require F pili to infect E. coli • g5p, gene 5 protein • K
, the intrinsic binding constant K times a cooperativity factor • n, the number of nucleotides bound per g5p monomer • OB fold, oligonucleotide/oligosaccharide binding fold • PDB, Protein Data Bank • [P]/[N], [protein monomer]/[nucleotide] molar ratio • SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis • ssDNA, single-stranded DNA • WT, wild-type protein
|
Introduction |
|---|
|
TOPAbstractIntroductionResultsDiscussionMaterials and methodsReferences |
|---|
) that could be as high as 5000 (Terwilliger 1996) to coat newly synthesized ssDNA genomes during viral morphogenesis. In addition, g5p binds to leader sequences and represses translation of five of the viral mRNAs (Zaman et al. 1990, 1991). Three binding modes have been observed for the g5p, with stoichiometries of n = 4, 3, or 2–2.5 nucleotides per g5p monomer. The n = 2–2.5 binding mode is weak, and is dissociated at salt concentrations of 0.1 M or less (Kansy et al. 1986; Sang and Gray 1989; Thompson et al. 1998). The binding affinity of g5p is dependent on the ssDNA sequence, and is inversely related to the nearest-neighbor base stacking tendencies in (A + C)-containing ssDNA sequences (Mou et al. 1999).
In the absence of nucleic acid, the three-dimensional structure of g5p has been determined by X-ray crystallography and NMR spectroscopy (Folkers et al. 1994; Skinner et al. 1994). The Ff g5p monomer is a relatively compact OB-fold ß-structure of only 87 amino acids, and it exists as a stable dimer in solution even at a concentration as low as 1 nM (Murzin 1993; Terwilliger 1996). Aromatic residues of the protein consist of five Tyr, three Phe, and no Trp. The five Tyr are conserved or conservatively substituted in the related IKe and Pf3 ssDNA binding proteins (de Jong et al. 1989). Two Tyr (Tyr56 and Tyr61) are largely buried in the interior of the g5p. The other three Tyr (Tyr26, Tyr34, and Tyr41) are located on the protein surface. Tyr26 is directly involved in binding ssDNA, and Tyr34 and Tyr41 are involved in protein–protein cooperative interactions. Tyr41, located in a flexible dimer–dimer contact loop, is particularly important in stabilizing dimer–dimer interactions. The Y41H mutant protein has improved solubility, making possible high-resolution NMR structural studies (Folkers et al. 1994). The locations of the five Tyr within the g5p dimer are shown in Figure 1
.
|
Phe mutants, Y26F, Y34F, Y41F, and Y61F, showed that each has a positive 229-nm CD band, but with a slightly (<10%) reduced magnitude compared with the wild-type g5p. Our initial inference was that Tyr56 might dominate this CD band, but this was without taking into account the CD contributions of the substituted Phe. It was also inferred that Tyr56 should have a strong positive Lb band at longer wavelengths, if tyrosyl contributions are additive (Mark et al. 1995; Thompson et al. 1998). In this paper we present new CD data for a Y56H mutant, and show that the La band of the g5p consists of approximately additive contributions of the five Tyr, while, at the same time, tyrosyl contributions to the Lb band are not additive. This insight has implications for the interpretation of the decreased La band intensity associated with the binding of g5p to ssDNA, as will be discussed. The structure of the g5p•ssDNA complex, which forms a left-handed superhelix with two antiparallel binding channels for ssDNA, has been studied by electron microscopy, NMR, CD spectroscopy, solution scattering techniques, and molecular modeling (Gray 1989; Folmer et al. 1994; Skinner et al. 1994; Olah et al. 1995). Of particular importance to our current study, two models of the g5p•ssDNA complex have been proposed for constructing the polymeric protein shell of the g5p complex on the basis of the X-ray structures, in the absence of ssDNA, of the wild-type protein and two g5p mutants, Y41H and Y41F (Guan et al. 1994). A superhelical complex based on the crystal structure of the Y41H mutant was more satisfactory in its dimer–dimer contacts than that based on the wild-type or Y41F proteins. The model, based on the crystal structure of the Y41H protein, was similar to that independently derived from the solution structure of the Y41H protein (Folmer et al. 1994). Because NMR results showed that the solution structure of the Y41H mutant was very nearly identical to that of the wild-type or Y41F mutant proteins (Prompers et al. 1995), differences in the crystal structures of Y41H and the other two proteins were apparently the result of crystal packing forces. It has been puzzling why the crystal packing interactions of the mutant Y41H protein, which is defective in cooperative interactions in solution, appeared to mimic the dimer–dimer arrangement of wild-type g5p complexed with ssDNA in solution.
By simultaneously monitoring the CD changes of the nucleic acid CD at 270 nm and the protein tyrosyl CD band at 229 nm, we show that Y41H and Y41F mutants have significantly different binding properties when added to poly[d(A)] at low ionic strength, although they have similar CD spectral characteristics when free in solution. In the presence of 50 mM Na+, the CD binding properties observed for Y41F and Y41H mutants were both like those of the wild-type g5p. Our results suggest that these effects reflect different defects in the cooperative binding of the two Tyr41 mutants. Moreover, our results show that the Y41H protein can undergo coordinated changes in its DNA binding and dimer–dimer cooperative interactions with a change of salt concentration in solution. We conclude that it is not unexpected that dimer–dimer packing interactions of the wild-type and Y41F proteins in crystals without ssDNA do not represent those that occur in solution when the g5p is bound to ssDNA.
|
Results |
|---|
|
TOPAbstractIntroductionResultsDiscussionMaterials and methodsReferences |
|---|
. In this wavelength region, the g5p mutants retained the large tyrosyl La CD band of the wild-type g5p at 229 nm, but the magnitude at 229 nm was about 10% smaller for Y41H and Y41F and 19% smaller for Y56H. Because the 229-nm CD band is completely abolished by denaturation (Liang and Terwilliger 1991), and because there was an overall similarity in the far-UV CD spectra of the mutant and wild-type gene 5 proteins, these data provided evidence that the mutant proteins maintained a native structure in solution.
|
and previously published data (Mark et al. 1995; Thompson et al. 1998), are reflected in the difference spectra in Figure 3A. These spectra were obtained by subtracting the spectra of the TyrPhe/His mutants from the spectrum of wild-type g5p. The sum of these difference spectra is compared with the measured spectrum of the wild-type g5p in Figure 3B. The maximum in the summed difference spectra is considerably smaller than the maximum in the wild-type CD spectrum and is shifted to the red. These differences do not necessarily reflect deviations from additivity of the Tyr La CD bands, because difference spectra are being compared with an absolute spectrum. If Phe contributions to four of the difference spectra are estimated and subtracted as described in Materials and Methods, the sum of the contributions of the individual Tyr side chains is close to the experimental (229) of the wild-type protein, as shown in Figure 3B and Table 1. All five Tyr residues make positive contributions to the La band, ranging in magnitude from 8.5 M-1 cm- 1 for Tyr26 to 26.9 M-1 cm-1 for Tyr61. This analysis of the difference spectra therefore yields a value of +83 M-1 cm-1 for the sum of the individual Tyr contributions to (229),which is within 8% of the directly observed value of +90 M-1 cm-1 for the wild-type protein. Thus, the far-UV Tyr contributions are, to a good approximation, additive. There is no one Tyr that makes a dominant contribution.
|
|
shows CD spectra of the wild-type and mutant proteins in the near-UV region (300–250 nm). At wavelengths above 260 nm, where Phe and His make little or no CD contributions, the CD spectrum is dominated by tyrosyl Lb bands (Strickland 1974). CD spectra of the Y41H and Y41F mutants were almost identical, but were different from that of the wild-type g5p in having somewhat more negative magnitudes. This showed that the two mutant proteins were essentially identical, and their spectra in this region reflected only the removal of Tyr41. The CD spectrum of the Y56H mutant was only slightly changed from that of the wild-type protein, showing that Tyr56 does not make a large contribution to the near-UV spectrum of the wild-type protein, contrary to expectations from the long-wavelength spectra of the other mutants (Thompson et al. 1998).
Empirical CD analyses
As assigned by the DSSP method (Kabsch and Sander 1983), the secondary structures in the crystallographic structures of wild-type, Y41F, and Y41H proteins (Protein Data Bank [PDB] codes 1GVP, 1YHB, and 1YHA, respectively) are listed in Table 2A. The mutant proteins have structural features similar to those of wild-type g5p, with a predominant ß-structure content (44–48%). Moreover, all three proteins have 7% of distorted 
-helix (two short segments of three residues in 310 helices), 15–17% of turn, and 30–33% of unordered structures. The experimental protein CD spectra were analyzed by three CD analysis programs (CONTIN [Provencher and Glöckner 1981], SELCON3 [Sreerama et al. 1999], and CDSSTR [Johnson 1999]) for secondary structure assignments, and the averaged results of these three analyses are shown in Table 2B. All three mutant proteins (Y41F, Y41H, and Y56H) had secondary structure fractions similar to those of the wild-type g5p with less than 1% -helix and significant percentages (38–41%) of ß-structure. These results for wild-type and mutant (Y41F and Y41H) gene 5 proteins were comparable to the results derived from crystallographic data (Table 2A). For example, the fractions of regular ß-sheet were 24–26% and 26–30% for data from CD and crystallographic structures, respectively. In general, the protein secondary structure assignments from the CONTIN and the CDSSTR analyses were more consistent with the crystal structures than were those from the SELCON3 analysis. Specifically, the SELCON3 analysis gave less ß-structure (by 7–10%) and more unordered structure (by 7–17%) than given by the CONTIN and CDSSTR analyses (data not shown).
|
|
Protein•poly[d(A)] complexes
CD titrations and nucleic acid CD changes at 270 nm.
Figure 4A and B shows the CD spectral changes in the near UV of poly[d(A)] upon additions of Y41F or Y41H mutant proteins in 2 mM Na+ (phosphate buffer, pH 7.0). CD spectral changes during titrations of poly[d(A)] with the Y41F protein were qualitatively the same as those published for titrations of poly[d(A)] with wild-type, Y34F, and Y61F proteins (Mark et al. 1995; Thompson et al. 1998), as well as with the Y56H protein (not shown). Upon additions of g5p at [P]/[N] ratios <0.25, shown as open symbols, the positive CD bands of poly[d(A)] decreased in magnitude and became negative. Because the CD spectrum of g5p is negligible in comparison with that of the ssDNA in this wavelength region (Thompson et al., 1998), this type of CD change is attributable to changes in the nucleic acid and includes the effects of dehydration, partial base–base unstacking, and the conformational alterations needed to form a left-handed superhelical structure when complexed with g5p (Gray 1996).
|
show the reduction in CD magnitude at 270 nm as a function of the [P]/[N] molar ratio for titrations with the four proteins. Solid symbols show data for titrations in 2 mM Na+ (phosphate, pH 7.0). Binding of the wild-type, Y41F, and Y56H gene 5 proteins to poly[d(A)] was essentially stoichiometric, and the first titration endpoints were reached at [P]/[N] ratios of 0.25 to 0.30, which indicated binding in n = 4 or 3 binding modes, as summarized in Table 3. Also, at higher [P]/[N] ratios with wild-type, Y41F, or Y56H proteins, the CD at 270 nm became less negative until a second titration end point was reached at [P]/[N] ratios of 0.39 to 0.44, indicative of a weaker n 
2.5 binding mode (Fig. 5A). Full spectra are shown in this [P]/[N] range for the Y41F protein in Figure 4A (solid symbols). This weak binding mode has also been observed when titrating poly[d(A)] with the three other Tyr mutants, Y26F, Y34F, and Y61F (Mark et al. 1995; Thompson et al. 1998). In such a binding mode where there are only a few nucleotides per protein monomer, structural considerations led to the inference that the ssDNA is probably partially released from the protein binding sites, which are held in juxtaposition by cooperative interactions, accounting for the reduced CD effect at 270 nm (Mark et al. 1995; Thompson et al. 1998).
|
|
; solid symbols for Y41H). Moreover, there was only one titration breakpoint, so that the CD did not change at [P]/[N] molar ratios above this breakpoint (Figs. 4B,5A), which was consistent with a loss of binding cooperativity. Finally, the total CD change at 270 nm of poly[d(A)] was only about -4 M-1•cm-1 when titrated with the Y41H mutant. Separate experiments were performed to test whether the data for the Y41H•poly[d(A)] complex were time-dependent, and no changes in measured CD values were found between those taken within minutes after mixing and those taken after 24 h (not shown).
When titrations were performed in 50 mM Na+ (phosphate buffer, pH 7.0), the binding of the Y41H protein was stoichiometric, and titrations with the Y41H, Y41F, and the wild-type proteins were similar (Fig. 5A, open symbols). There were single breakpoints in the titrations in 50 mM Na+, because the weaker binding mode of g5p is dissociated at the higher salt concentration (Mark et al. 1995).
Tyrosyl CD changes at 229 nm.
The panels of Figure 5B show the tyrosyl CD changes at 229 nm as a function of the [P]/[N] ratio during CD titrations of poly[d(A)] with the wild-type g5p and the Tyr41-substituted mutants. Dashed lines represent the linear increase in CD with increasing the [P]/[N] ratio that would be observed from adding the free proteins to buffer. Symbols show the experimental data obtained during titrations, after subtracting the contribution of the nucleic acid as described in Materials and Methods. In 2 mM Na+ (phosphate, pH 7.0), the tyrosyl CD band of wild-type and Y41F proteins respectively decreased 37 and 46%, relative to the respective free proteins, as the proteins were added to poly[d(A)] up to the breakpoint. (The apparent decrease would be even greater without a correction for the decreased CD of the nucleic acid when bound by the protein.) Breakpoints in CD titration plots at 229 nm are more difficult to discern than in plots at 270 nm, but, for the wild-type and Y41F proteins, there was at least one breakpoint in a [P]/[N] range consistent with the first (n = 4 or 3) binding mode detected at 270 nm. A weak mode of binding at n 2.5 was not detected in the 229-nm CD titrations. Similar CD perturbations and breakpoints have been observed before in the 229-nm CD titrations of genomic ssDNA at 2 mM Na+ with Y26F and Y61F, but not with Y34F, which led to the conclusion that Tyr34 is responsible for the 229-nm CD perturbation during cooperative binding (Mark et al. 1995; Thompson et al. 1998). Consistent with this conclusion, we found in the present work that the 229 nm band of the Y56H mutant also decreased by 31–32% during the titration of poly[d(A)] (not shown), close to the decrease found for the wild-type protein. That is, the CD change at 229 nm for the wild-type and Y41F proteins is most simply explained as being the result of cooperative interactions and a perturbation of Tyr34 at the dimer–dimer interface under these low salt conditions.
In the case of titrations of poly[d(A)] with Y41H at 2 mM Na+, there was no change in CD magnitude at 229 nm throughout the full range of protein additions (Fig. 5B), although the nucleic acid perturbation at 270 nm (Fig. 5A) clearly showed that Y41H bound to, and perturbed, the ssDNA. This result can be interpreted as indicating that the Y41H mutation caused a more drastic reduction in the binding cooperativity than did the Y41F mutation so that, indirectly, the Tyr34 was not perturbed to the same extent in titrations of poly[d(A)] with the two mutant proteins.
On the other hand, when titrations with Y41H were performed in 50 mM Na+ (phosphate, pH 7.0), the CD band at 229 nm did exhibit a decrease and showed two breakpoints at [P]/[N] ratios at which the poly[d(A)] became saturated in the n = 4 and n = 3 binding modes (Fig. 5B, open symbols). Below [P]/[N] 0.25, the 229-nm CD magnitude was reduced by about 23% from that expected for addition of free protein. Then, as the protein was added to give [P]/[N] values between 0.25 and 0.30, the tyrosyl band was more fully perturbed to be decreased a total of 29%. Above a [P]/[N] ratio of 0.3, the 229-nm CD band paralleled the slope expected for the addition of free protein. This pattern of change in the 229-nm CD band that was exhibited by the Y41H protein was very close to the patterns observed for the wild-type and Y41F proteins at this higher salt concentration. The two breakpoints could arise from different perturbations of Tyr34 when binding is in the n = 4 and n = 3 modes, or possibly from the added perturbation of another Tyr such as Tyr26 when binding is in the n = 3 mode. In any case, the CD titrations left no doubt that differences in the binding characteristics of the two Tyr41 mutants that were apparent in 2 mM Na+ almost disappeared in 50 mM Na+.
Salt-dissociation of the complexes.
Figure 4C and D shows representative CD spectra during the dissociation of saturated Y41F•poly[d(A)] or Y41H•poly[d(A)] complexes by increasing the salt concentration. Spectra for complexes with wild-type or Y56H g5p at various salt concentrations (data not shown) were similar to those shown for the complex with Y41F. The complexes were formed in 2 mM Na+ (phosphate buffer, pH 7.0) at [P]/[N] ratios of 0.25 or 0.33. As the NaCl concentration was increased to 0.03 M, the CD bands above 250 nm of the Y41H•poly[d(A)] complex became more negative, indicating a greater perturbation of the poly[d(A)]. As the [NaCl] increased to >0.1 M, the complexes dissociated, as shown by the fact that the CD spectra above 260 nm became less negative and eventually were identical to the spectrum of free poly[d(A)] at a high NaCl concentration of 1.0 M.
Dissociation curves (not shown) were obtained by monitoring the CD change, relative to the maximum difference in CD value at 270 nm, as a function of NaCl concentration (see Materials and Methods).
Dependence of binding constant on NaCl concentration.
The binding affinity (K) was determined from the free g5p dimer concentration required for 50% dissociation of each complex. The salt concentration needed for 50% dissociation increased as the concentration of the g5p•poly[d(A)] complex increased. Figure 6 shows that log[K] was linearly related to the log[NaCl] for each type of complex. The slopes of the log[K] versus log[NaCl] plots indicate the number of ions released per g5p dimer when forming a complex (Record et al. 1976; Terwilliger 1996) and are listed in Table 3. Binding of wild-type or TyrPhe/His mutant gene 5 proteins to poly[d(A)] generally resulted in the release of about four ions per g5p dimer, consistent with results of prior CD and fluorescence work (Bulsink et al. 1985; Stassen et al. 1992a; Mark et al. 1995; Thompson et al. 1998). Exceptions are the Y41H and Y34F mutants, for which the measured slopes are -4.9 ± 0.6 (this work) and -6.6 ± 1.7 (Mark et al. 1995), respectively, indicating the possible additional release of one to two ions per protein dimer–dimer interface for these mutants.
|
values for the various proteins were extrapolated to 0.2 M NaCl for comparison; these are listed in the last column in Table 3. The mutant proteins all showed lower binding affinities compared with the wild-type protein. The Tyr41 mutants, which are defective in protein–protein interactions and have a lowered cooperativity factor, had binding affinities for poly[d(A)] much lower than those of the Y56H mutant or of the Y26F or Y61F mutants previously studied (Mark et al. 1995; Thompson et al. 1998). The binding affinities of Y41H and Y41F were, respectively, about 90 and 35 times lower than that of wild-type g5p. Binding affinities of the Y41F and Y34F mutants are almost identical at 0.2 M NaCl (Mark et al. 1995). These data further showed that Y41H has a more defective cooperative interface than do the other two mutants, Y41F and Y34F, that have a substitution at the protein–protein interface. |
Discussion |
|---|
|
TOPAbstractIntroductionResultsDiscussionMaterials and methodsReferences |
|---|
showed that the three mutant gene 5 proteins in this study (Y41F, Y41H, and Y56H), like previously studied mutant proteins (Y26F, Y34F, and Y61F; Mark et al. 1995; Thompson et al. 1998), retained the unique positive band at 229 nm that is characteristic of the native OB-fold structure of the wild-type g5p.
For the first time, it is possible to consider whether the atypical CD spectrum of g5p, which has no Trp or disulfides, has Tyr contributions that are approximately additive. We found that the sum of the La bands of the individual Tyr side chains agrees well with the observed intensity of the 229-nm band for the wild-type protein (Fig. 3B, Table 1). If exciton interactions among the Tyr La excited states played a significant role in determining the CD of the 229 nm band, this band should exhibit deviations from additivity. For example, if two Tyr La transitions were coupled to give an exciton CD couplet, mutation of either side chain would eliminate the couplet. This contribution would then be counted twice when the spectra derived from the mutants are combined. The nearly additive Tyr La contributions demonstrate that exciton interactions among the Tyr La excited states play at most a minor role in determining the 229-nm CD. The amplitude of an exciton couplet is proportional to max2 and inversely proportional to the square of the interchromophoric distance (Woody and Dunker 1996). Whereas the fully allowed B bands of Tyr, Phe, and Trp have max2 
35,000 M-1 cm-1, the La bands have max2 
5000–10,000 M-1 cm-1. Exciton couplets can be expected for B bands at separations of 5–10 Å, but such a couplet in the Tyr La band is only likely to be observable if the rings are nearly in van der Waals contact. The closest pair of Tyr side chains in g5p is Tyr56 and Tyr61, separated by 5.5 Å. Therefore, it is not surprising that exciton contributions to the Tyr La band of g5p are not significant. The CD of the La band for each Tyr derives primarily from coupling of the La transition with the 
* transitions of nearby peptide groups and with the fully allowed, short-wavelength Ba and Bb transitions of nearby Tyr and Phe side chains (unpublished results).
The absence of a change in (229) when Y34F binds to poly[d(A)] at [P]/[N] 
0.25, implies that Tyr34 plays a major role in this ligand-induced effect. The contribution of Tyr34 to the 229-nm band (+16 M-1 cm-1) can be compared with the change in (229) upon g5p binding to poly[d(A)], which is -37% of +90 M-1 cm-1 or -33 M-1 cm-1 for the wild-type protein. The effect of binding to ssDNA is about twice as large as the Tyr34 contribution to the wild-type CD spectrum. Therefore, the effect is not simply an abolition of the Tyr34 contribution. If the effect is solely due to Tyr34, the results imply that the Tyr34 contribution is reversed in sign upon binding to ssDNA, presumably due to conformational changes induced in the g5p. Alternative explanations include: (1) coupling of Tyr34 with chromophores in neighboring dimers upon binding to ssDNA, or (2) transmission by Tyr34 to other Tyr residues of conformational changes induced by ssDNA binding. In our previous paper (Thompson et al. 1998) it was suggested that coupling of Tyr34 with the Tyr56/Tyr61 pair may account for the absence of perturbation of the 229 nm band when Y34F binds ssDNA. Our present results show that Tyr56 does not dominate the 229-nm band, so although this coupling still could provide part of the explanation, it is no longer considered to be as probable.
Above 270 nm, where tyrosyl Lb bands dominate, Tyr34 and Tyr61 make significant negative contributions and the other three Tyr have nearly zero or small positive signals (Mark et al. 1995; Thompson et al. 1998). Comparison with the near-UV CD spectrum of the wild-type protein, which has positive bands at 276 and 283 nm (Fig. 2B), led to the prediction that Tyr56 should make a significant positive CD contribution near 280 nm to compensate for the negative CD contributions of Tyr34 and Tyr61. As seen in Figure 2B, this is not the case, and, therefore, the tyrosyl CD contributions are not additive in the near UV. The deviations from additivity of the Tyr Lb CD bands may result from relatively minor conformational changes induced in the remaining Tyr side chains by replacement of a given Tyr by Phe (or His). Because the Lb transition is polarized perpendicular to the twofold axis of the phenolic ring, variations in 
2 (the torsional angle about the Cß—C
bond) will affect the Lb rotational strength. By contrast, the La transition is polarized along the twofold axis and is insensitive to variations in 2.
The very similar near-UV CD spectra of Y41H and Y41F in Figure 2B indicate that neither replacement of Tyr41 detectably changes the average CD signals of the remaining aromatic residues. These CD data are consistent with NMR work that shows Y41F and Y41H to have nearly identical structures to each other and to the wild-type g5p in solution (Prompers et al. 1995). In their crystal structures, Y41H differs from the Y41F and wild-type proteins in the DNA-binding loop and dimer-contact regions (Guan et al. 1994; Skinner et al. 1994). Calculations suggest that these aspects of the Y41H crystallographic structure are energetically unstable in solution, and, therefore, differences between the solution and crystal structures may well be due to crystal packing forces (Prompers et al. 1995).
Empirical CD analyses of the ß-structure in gene 5 protein
Because the CD spectra of g5p have a unique large positive 229-nm CD band resulting from tyrosyl contributions, it might have been difficult to perform a successful analysis of the g5p secondary structures (Woody 1978). However, we found that CD analyses of wild-type and mutant (Y41F and Y41H) gene 5 proteins were consistent with high-resolution crystallographic and NMR studies in showing that the protein has a predominant ß-structure. Of the three programs used, CONTIN (Provencher and Glöckner 1981), SELCON3 (Sreerama et al. 1999), and CDSSTR (Johnson 1999), CDSSTR gave the best overall analysis in terms of the agreement between the secondary structure fractions estimated from CD and those from the X-ray structures. However, the results for a given gene 5 protein from all three programs were similar, all being within about 5% for a given secondary structure, except for the regular ß-sheet, for which the results were all within 10% (not shown). The successful analysis of CD spectra of wild-type and mutant gene 5 proteins, despite substantial aromatic contributions, is attributable to the improvements in protein CD analysis achieved by the use of a larger reference protein set (Sreerama and Woody 2000a).
Structural implications of differential binding properties of Y41H and Y41F
The Y41F mutant has characteristics of binding to poly[d(A)] that are similar to those of the wild-type and four other mutant proteins (Y26F, Y34F, Y56H, and Y61F) (Mark et al. 1995; Thompson et al. 1998; this work). These proteins exhibit two modes of binding to poly[d(A)] when the nucleic acid CD at 270 nm is monitored during titrations in 2 mM Na+. The Y41H protein, however, showed only one mode of binding, as monitored by the perturbation of the nucleic acid CD. Other differences from the Y41F protein were that the 270-nm binding curve for the Y41H mutant was nonstoichiometric (at low [P]/[N] ratios) and poly[d(A)] was less perturbed when it was complexed with Y41H than with the wild-type or any of the other mutant proteins studied to date, because the CD at 270 nm only reached a negative value of -4 M-1 cm-1 in the low salt buffer (Fig. 5A).
Moreover, the Y41H and Y41F mutants differed in that no 229-nm CD perturbation was found during titrations of poly[d(A)] with Y41H in 2 mM Na+ (Fig. 5B). The titration of poly[d(A)] with Y41F at low ionic strength resulted in a decrease of 46% in the magnitude of the 229-nm tyrosyl CD band, similar to that observed during titrations with the wild-type g5p and three other mutant proteins (Y26F, Y56F, and Y61H). As described in Results, this effect on the Tyr band at low ionic strength is probably caused by the perturbation of Tyr34 that results from protein–protein interactions during cooperative binding at [P]/[N] 0.25.
These observations were undoubtedly a reflection of the decreased cooperativity in Y41H dimer–dimer interactions (Folkers et al. 1991), and were in accord with the fact that the Y41H mutant differs from the Y41F and wild-type proteins in its crystal packing (Guan et al. 1994). The Y41H and Y41F proteins were originally described as both having a reduced tendency to aggregate, presumably because of reduced dimer–dimer interactions relative to the wild-type protein (Folkers et al. 1991). We now have shown that the binding affinity of Y41H for poly[d(A)] is significantly less than that of Y41F, by a factor of about 2.6 when extrapolated to 0.2 M NaCl (Fig. 6, Table 3).
Although the Y41H protein has a reduced cooperativity of binding and reduced binding affinity in solution, Guan et al. (1994) proposed that the dimer–dimer packing of Y41H in crystals represents the interaction of neighboring dimers when the wild-type g5p is bound to ssDNA and that the contacts in the Y41F and wild-type crystals represent those of the protein when it aggregates in solution in the absence of ssDNA. These authors modeled the g5p•ssDNA superhelix on the basis of the Y41H crystal packing, and the neighboring dimers in their model were similar to those in a superhelical complex independently modeled by Folmer et al (1994). The success of this model, despite the effect of the Y41H41 substitution on the solution binding properties of g5p, may be rationalized on the basis of an interplay between (1) the protein loops involved in dimer–dimer contacts and (2) the DNA-binding wings that directly contact the ssDNA. Protein–protein cooperative interactions could indirectly constrain (or be constrained by) contacts with the ssDNA. Alternatively, there could be coupled structural changes in the DNA binding and cooperativity loops of the protein. Correlated motions of the DNA-binding wings and the dimer–dimer interaction loops have recently been described for the closely related Pf3 DNA-binding protein (Horstink et al. 1999) and for the dynamic ssDNA-binding mechanism described for human replication protein A (Bochkareva et al. 2001). That is, it is possible that the dimer–dimer crystal packing of the Y41F or wild-type proteins differs from that when these proteins are bound to ssDNA in solution, but the Y41H mutation permits "native" interdimer contacts in the crystal structure in the absence of ssDNA. Guan et al. (1994) observed that, in the crystal structure of the Y41H mutant, the ssDNA binding hairpins were affected and even adopted different conformations within the same g5p dimer. Structural perturbations permitted by the mutation need be only relatively small, given that a Raman study showed that the g5p secondary structure does not change significantly upon binding to poly[d(A)] (Benevides et al. 1996).
In solution, our combined CD titration data from monitoring the nucleic acid at 270 nm and the protein at 229 nm are supportive of correlated changes in the DNA-binding site and the cooperativity domain of the Y41H mutant during formation of a Y41H•poly[d(A)] complex. As shown in Figure 5A, in 50 mM Na+, the ssDNA binding of Y41H was stoichiometric throughout the titration, and the poly[d(A)] was as fully perturbed as with Y41F or the wild-type protein. In addition, Figure 5B shows that, at the higher salt concentration, the 229-nm tyrosyl CD band decreased to approximately the same extent and with the same dependence on the binding mode as for the Y41F and wild-type proteins. Figure 7 schematically illustrates the simultaneous change in ssDNA binding and protein–protein interactions with an increase in salt concentration, which conceivably helps overcome repulsion between positive charges on His41 on neighboring dimers. Because the pKa of His41 is 6.5 (Folkers et al. 1991), His41 side chains will be only partially positively charged at pH 7.0. Nevertheless, the charge on His41 appears to be important in the binding properties of Y41H to poly[d(A)], because CD measurements of Y41F•poly[d(A)] and Y41H•poly[d(A)] complexes ([P]/[N] = 0.25; 2 mM Na+) as a function of pH showed that the CD (270 nm) of the Y41H•poly[d(A)] complex was even less negative at pH 6.1 than at pH 7.0 (Fig. 5A), while the CD of the Y41F•poly[d(A)] complex was unchanged (not shown).
|
). |
Materials and methods |
|---|
|
TOPAbstractIntroductionResultsDiscussionMaterials and methodsReferences |
|---|
(276) of 7074 and 5660 M-1•cm-1 for the wild-type and Y41F mutant proteins, respectively (Day 1973; Mark et al. 1995). Because His and Phe have essentially no absorption above 270 nm, the same (276) was used for all three mutants. All g5p stocks were dialyzed into and stored in a buffer of 2 mM Na+ (phosphate buffer, pH 7.0). Poly[d(A)] (Sigma) had an average length of 260 nucleotides according to the manufacturer. The poly[d(A)] was dissolved in 2 mM or 50 mM Na+ (phosphate buffer, pH 7.0), and its concentration was determined from absorption spectra using an (260) of 9650 M-1•cm-1, per mole of nucleotide (Bollum 1966).
UV absorption and CD measurements
UV absorption spectra were measured with an Olis-modified Cary 14 spectrophotometer (On-Line Instrument Co.). CD spectra were measured with a Jasco Model J715 spectropolarimeter (Jasco Incorporated). Far-UV CD spectra (250–180 nm) of free g5p at concentrations of 90–150 µM were obtained using water-jacketed cylindrical quartz cells having a light path of 0.1 mm. Near-UV CD spectra (320–250 nm) of free g5p and spectra during CD titrations and salt dissociations were taken in 10- or 20-mm cuvettes. Spectropolarimeter parameters and processing were as in earlier work (Mou et al. 1999), except that the near-UV spectra, collected at 0.1-nm resolution, were smoothed over 13 points instead of 99 points to preserve the resolution of vibronic bands. All spectra were taken at 20 ± 0.5°C. The smoothed CD data were plotted at 1-nm intervals as in units of M-1•cm-1, per mole of g5p monomer for free protein spectra or per mole of nucleotide for CD titrations and salt dissociations.
Calculation of Tyr contributions to the 229-nm CD band
To calculate the contribution of a given Tyr side chain from the difference CD spectrum [CD(wild-type) - CD(TyrPhe mutant)], different procedures are required, depending on the shape of the difference spectrum. The difference spectra for the five mutants exhibit three different shapes. Those for Y34F, Y41F, and Y61F proteins show the classic first-derivative shape expected for difference spectra resulting from a shift in wavelength with little or no change in intensity (Donovan et al. 1961). If the TyrPhe substitution does not lead to a conformational change, the La band of the Phe is expected to have a rotational strength of the same sign and comparable magnitude to that of the Tyr that it replaces. However, the Phe La band is shifted by about 10 nm to shorter wavelengths, that is, from 230 to 220 nm. Therefore, when the CD difference spectrum of the mutant is subtracted from that of the wild-type, the La band of the replacement Phe will contribute a band of opposite sign, centered at 220 nm. The resultant CD difference spectrum will have the appearance of a CD couplet (Schellman 1968; Bayley 1973) with a crossover wavelength at the average of the La wavelengths for Phe and Tyr (225 nm). The long-wavelength maximum will be red-shifted relative to 
max for the Tyr La band, and the short-wavelength maximum will be blue-shifted relative to max for the Phe La band. Moreover, the amplitude for the long-wavelength branch of the couplet will be smaller than that of the Tyr contribution.
For this type of difference spectrum, the value of max for the 229 nm band can be estimated from the maximum in the difference spectrum, max, assuming that the substitution of Phe for Tyr leads to a shift of max without a change in amplitude:
 |
max is the wavelength of the maximum in the difference spectrum (232 nm), Tyr = 229 nm, Phe = 220 nm, and is the Gaussian band width for the La bands of Tyr and Phe, assumed to be the same and taken to be 9 nm. This leads to the relationship:
 |
The mutant Y26F also gives a CD difference spectrum with a first-derivative shape, but this is shifted about 7 nm to the blue, with a positive maximum at 225 nm, a crossover near 220 nm, and a negative maximum at 212 nm. Because of this shift in wavelength, it is not clear that a procedure analogous to that used for the previous mutants should be used. Instead, the observed (229) value for the Y26F mutant has been corrected for the contribution of a Gaussian band of 9 nm bandwidth, with the sign, magnitude, and wavelength of the negative maximum at 212 nm. Finally, the CD difference spectrum for Y56H has the shape of a simple Gaussian band centered at 228 nm, indicating that the substituted His makes little contribution to the CD in this region. Therefore, the value of (229) from the difference CD spectrum for this mutant was used for the (229) of Tyr56.
Empirical CD analyses for protein secondary structure
Protein CD spectra were analyzed by CDPro software (CONTIN, SELCON3, and CDSSTR, available at the website http://lamar.colostate.edu/~sreeram/CDPro) for determining the fractions of secondary structure in the individual proteins. The input experimental CD data ranged from 240 to 190 nm at 1-nm intervals, and were in units of per mole of amino acid residue. The largest reference protein set, containing 48 proteins, was selected for all three CD analyses (Sreerama and Woody 2000a). Reference protein sets that include the poly(Pro)II fraction were not used for the analyses because g5p has no poly(Pro)II structure. CD analysis results were compared with the fractions of secondary structures in the crystallographic structures as assigned by the DSSP method (Kabsch and Sander 1983). These secondary structures were the regular -helix, distorted -helix, regular ß-sheet, distorted ß-sheet, turn, and unordered structures.
Titrations of poly[d(A)] with g5p
CD titrations were performed by adding weighed volumes of concentrated g5p (1 x 10-4 M) to poly[d(A)] (60 µM) in buffers of 2 mM or 50 mM Na+ (phosphate, pH 7.0). For plots of protein CD changes at 229 nm during the titrations, the nucleic acid contribution was subtracted from the measured CD. The CD value at 229 nm of poly[d(A)] when saturated with g5p (at a [protein monomer]/[nucleotide] molar ratio, [P]/[N], of 0.25) has been shown to be close to that of free poly[d(A)] at about 76°C (Mark and Gray 1997). Thus, at [P]/[N] 0.25, this constant value for the poly[d(A)] component was subtracted from the measured CD. For samples with [P]/[N] molar ratios between 0 and 0.25, the measured CD at 229 nm was corrected by subtracting a proportional contribution due to the poly[d(A)] (Thompson et al. 1998).
Determination of binding affinities
The binding affinities (K) of individual proteins for single-stranded poly[d(A)] were determined by the NaCl-induced dissociation of g5p•poly[d(A)] complexes. Saturated g5p•poly[d(A)] complexes were formed at [P]/[N] molar ratios of 0.25 or 0.33 in a buffer of 2 mM Na+ (phosphate, pH 7.0). The NaCl concentration was increased by adding weighed aliquots of a 4.0 M NaCl solution. Light scattering was minimal, as in previous work (Mou et al. 1999). Dissociation of the complexes as the salt concentration was increased was determined by the change in the CD at 270 nm of poly[d(A)], calculated as: % CD change = 100 x [CD(ssDNA) - CD(complex at given [NaCl])]/[CD(ssDNA) - CD(complex with maximum negative CD)], where the CD(ssDNA) was the value at 270 nm for poly[d(A)] at 1 M NaCl. The salt concentrations required for 50% dissociation of the complex were determined by fitting the dissociation curves with a nonlinear, rational, four-parameter regression function (SigmaPlot 5.0, SPSS Inc.). The binding affinities (K) for each complex were determined as K = 1/(2L), where L was the free dimer concentration (taken to be one-half of the total protein concentration) at 50% dissociation (Thompson et al. 1998).
|
Acknowledgments |
|---|
The publication costs of this article were defrayed in part by payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 USC section 1734 solely to indicate this fact.
|
References |
|---|
|
TOPAbstractIntroductionResultsDiscussionMaterials and methodsReferences |
|---|
Benevides, J.M., Terwilliger, T.C., Vohnik, S., and Thomas, Jr., G.J. 1996. Raman spectroscopy of the Ff gene V protein and complexes with poly(dA): Nonspecific DNA recognition and binding. Biochemistry 35:9603–9609.[CrossRef][Medline]
Bochkareva, E., Belegu, V., Korolev, S., and Bochkarev, A. 2001. Structure of the major single-stranded DNA-binding domain of replication protein A suggests a dynamic mechanism for DNA binding. EMBO J. 20:612–618.[CrossRef][Medline]
Bollum, F.J. 1966. Biosynthetic polydeoxynucleotides. In Procedures in nucleic acids research (Eds. G.L. Cantoni and D.R. Davies), pp. 577–583. Harper and Row, New York.
Bulsink, H., Harmsen, B.J., and Hilbers, C.W. 1985. Specificity of the binding of bacteriophage M13 encoded gene-5 protein to DNA and RNA studied by means of fluorescence titrations. J. Biomol. Struct. Dyn. 3:227–247.[Medline]
Day, L.A. 1973. Circular dichroism and ultraviolet absorption of a deoxyribonucleic acid binding protein of filamentous bacteriophage. Biochemistry 12:5329–5339.[CrossRef][Medline]
de Jong, E.A., van Duynhoven, J.P., Harmsen, B.J., Konings, R.N., and Hilbers, C.W. 1989. Two-dimensional 1H nuclear magnetic resonance studies on the gene V-encoded single-stranded DNA-binding protein of the filamentous bacteriophage IKe. I. Structure elucidation of the DNA-binding wing. J. Mol. Biol. 206:119–132.[CrossRef][Medline]
Donovan, J.W., Laskowski, Jr., M., and Scheraga, H.A. 1961. Effects of charged groups on chromophores of lysozyme and of amino acids. J. Am. Chem. Soc. 83:2686–2694.[CrossRef]
Folkers, P.J., Nilges, M., Folmer, R.H., Konings, R.N., and Hilbers, C.W. 1994. The solution structure of the Tyr41His mutant of the single-stranded DNA binding protein encoded by gene V of the filamentous bacteriophage M13. J. Mol. Biol. 236:229–246.[CrossRef][Medline]
Folkers, P.J., Stassen, A.P., van Duynhoven, J.P., Harmsen, B.J., Konings, R.N., and Hilbers, C.W. 1991. Characterization of wild-type and mutant M13 gene V proteins by means of 1H-NMR. Eur. J. Biochem. 200:139–148.[Medline]
Folmer, R.H., Nilges, M., Folkers, P.J., Konings, R.N., and Hilbers, C.W. 1994. A model of the complex between single-stranded DNA and the single-stranded DNA binding protein encoded by gene V of filamentous bacteriophage M13. J. Mol. Biol. 240:341–357.[CrossRef][Medline]
Gray, C.W. 1989. Three-dimensional structure of complexes of single-stranded DNA-binding proteins with DNA. IKe and fd gene 5 proteins form left-handed helices with single-stranded DNA. J. Mol. Biol. 208:57–64.[CrossRef][Medline]
Gray, D.M. 1996. Circular dichroism of protein–nucleic acid interactions. In Circular dichrosim and the conformational analysis of biomolecules (Ed. G.D. Fasman), pp. 469–500. Plenum Press, New York.
Guan, Y., Zhang, H., Konings, R.N., Hilbers, C.W., Terwilliger, T.C., and Wang, A.H. 1994. Crystal structures of Y41H and Y41F mutants of gene V protein from Ff phage suggest possible protein–protein interactions in the GVP–ssDNA complex. Biochemistry 33:7768–7778.[CrossRef][Medline]
Guan, Y., Zhang, H., and Wang, A.H.-J. 1995. Electrostatic potential distribution of the gene V protein from Ff phage facilitates cooperative DNA binding: A model of the GVP-ssDNA complex. Protein Sci. 4:187–197.[Abstract]
Horstink, L.M., Abseher, R., Nilges, M., and Hilbers, C.W. 1999. Functionally important correlated motions in the single-stranded DNA-binding protein encoded by filamentous phage Pf3. J. Mol. Biol. 287:569–577.[CrossRef][Medline]
Johnson, W.C. 1999. Analyzing protein circular dichroism spectra for accurate secondary structures. Proteins 35:307–312.[CrossRef][Medline]
