Two Conformations of Archaeal Ssh10b

The DNA-binding protein Ssh10b from the hyperthermophilic archaeon Sulfolobus shibatae is a member of the Sac10b family, which has been speculated to be involved in the organization of the chromosomal DNA in Archaea. Ssh10b affects the DNA topology in a temperature dependent fashion that has not been reported for any other DNA-binding proteins. Heteronuclear NMR and site-directed mutagenesis were used to analyze the structural basis of the temperature-dependent Ssh10b-DNA interaction. The data analysis indicates that two forms of Ssh10b homodimers co-exist in solution, and the slow cis-trans isomerization of the Leu61-Pro62 peptide bond is the key factor responsible for the conformational heterogeneity of the Ssh10b homodimer. The T-form dimer, with the Leu61-Pro62 bond in the trans conformation, dominates at higher temperature, whereas population of the C-form dimer, with the bond in the cis conformation, increases on decreasing the temperature. The two forms of the Ssh10b dimer show the same DNA binding site but have different conformational features that are responsible for the temperature-dependent nature of the Ssh10b-DNA interaction.

Proteins of the Sac10b family are highly conserved among thermophilic and hyperthermophilic Archaea, and homologous sequences have also been identified in eukaryal proteins from higher plants, protists, and vertebrates (1)(2)(3). The members of this family have been postulated to play a role in chromosomal organization in Archaea since the initial isolation of Sac10b from the hyperthermophile Sulfolobus acidocaldarius in the mid-1980s. Sac10b exists as a dimer of two 10-kDa subunits in solution and binds to DNA nonspecifically (4,5). Electron microscopic studies have shown that Sac10b binds to DNA cooperatively and forms different protein-DNA complexes depending on protein/DNA ratios but does not induce DNA supercoiling or compact DNA (6).
Recently, Bell et al. (2) discovered that Alba (also named Sso10b, a member of Sac10b family) forms a specific complex with a Sir2 homolog in Sulfolobus solfataricus cell extracts. They found that Sir2 in the presence of NAD ϩ can regulate the DNA binding affinity of Alba by deacetylation of Lys 16 of the protein. More recently, the crystal structure of Alba has been solved. Interestingly, the protein shares structural homology to the C-terminal domain of the Escherichia coli translation factor IF3 and the N-terminal DNA binding domain of DNase I. A model for the Alba-DNA interaction has been proposed (7).
Ssh10b, another member of the Sac10b family, was isolated from Sulfolobus shibatae (1). The protein is highly abundant and basic and binds double-stranded DNA without apparent sequence specificity. Gel retardation assays have shown that Ssh10b has two modes of DNA binding with distinctively different binding densities. In the low binding density mode, Ssh10b exhibits a binding size of ϳ12 bp of DNA, whereas in the high binding density mode, the protein appears to bind shorter stretches of DNA. Interestingly, Ssh10b affects DNA topology in a temperature-dependent fashion; it is capable of significantly constraining DNA in negative supercoils at temperatures higher than 318 K, but this ability is drastically reduced at 298 K (1). A previous NMR study revealed the co-existence of two forms of Ssh10b dimers at temperatures between 283 and 320 K, with one dominating at lower temperatures and the other at higher temperatures (8). However, the structural basis for the conformational heterogeneity of the Ssh10b dimer and the temperature dependence of the interaction of Ssh10b with DNA remained to be clarified.
In the present study, we investigated the heterogeneous conformations of Ssh10b and the structural factors influencing the interaction of Ssh10b with DNA by heteronuclear NMR spectroscopy. We found that the cis-trans isomerization of the Leu 61 -Pro 62 peptide bond of Ssh10b is the primary determinant of the conformational heterogeneity of the Ssh10b dimer. We also found that the equilibrium between the cis-and transforms of Ssh10b is sensitive to temperature. Our data suggest that the effect of temperature on the capacity of the protein to constrain negative DNA supercoils is related to the temperaturedependent conversion between the two Ssh10b conformations.

EXPERIMENTAL PROCEDURES
Expression and Purification of Ssh10b and Its Mutants-Ssh10b was produced from a synthetic gene with codon usage optimized for expression in E. coli. The gene was created from 12 overlapping oligonucleotides primers that were ligated and then cloned into the EcoRI and BamHI sites of vector pBV220. The genes of ⌬8 (deletion of the Nterminal eight residues), ⌬8P18A (Pro 18 replaced by Ala of the ⌬8 mutant), and P62A (Pro 62 replaced by Ala) mutants of Ssh10b were obtained by primer-directed mutagenesis. Each gene was cloned into expression vector pET11c, and the products were used to transform E. coli BL21(DE3) cells. The transformed cultures were grown at 37°C in 1 liter of LB broth containing 50 mg/liter ampicillin until A 600 ϭ 0.8 -1.0, and expression was induced for 2 h by adding isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 1 mM. The harvested cells were resuspended in 20 ml of buffer containing 30 mM potassium phosphate, pH 6.6, 0.1 mM EDTA, and 1 mM phenylmethylsulfonyl fluoride and sonicated. The lysate was centrifuged at 150,000 ϫ g for 2.5 h at 4°C and then the supernatant was heated for 20 min at 80°C to precipitate the E. coli proteins. After centrifugation, the supernatant was applied to a Resource-S column. Bound proteins were eluted with a 50-ml KCl gradient (0 to 0.75 M). Fractions containing Ssh10b proteins were pooled, dialyzed against distilled de-ionized water, and finally lyophilized. The purity of the proteins was confirmed by SDS-PAGE to be more than 95% and by matrix-assisted laser desorption ionization time-of-flight to be free of nucleic acid contaminants.
NMR Sample Preparation-15 N or 13 C singly labeled and 15 N/ 13 C doubly labeled Ssh10b proteins were expressed in E. coli strain BL21(DE3) grown in M9 minimal medium using 15 NH 4 Cl and/or 13 Cglucose as the sole nitrogen and carbon sources. All protein samples for NMR measurements were dissolved in 500 l of 90% H 2 O/10% D 2 O containing 20 mM deuterated acetate buffer, pH 4.8, 50 l of NaN 3 , 1 M 2,2-dimethyl-2-silapentanesulfonic acid and 20 mM KCl to a final protein concentration of about 1 mM, unless otherwise indicated. The sample for determination of the dimer interface of Ssh10b was 15 N/ 13 C asymmetrically labeled.
NMR Spectroscopy-All NMR experiments were carried out on a Bruker DMX 600 spectrometer equipped with a triple resonance probe and an actively shielded three-axis gradient unit. The experimental temperature was set to 310 K except for the temperature-dependent experiments. 1 H chemical shifts were referenced to the internal standard 2,2-dimethyl-2-silapentanesulfonic acid at 0 ppm. 15 N and 13 C chemical shifts were calculated indirectly using the corresponding consensus ⌶ ratios (9).
Although the assignment of the T-form Ssh10b has already been published (8), assignment of the remaining resolved resonances of the C-form Ssh10b was achieved by further exploring existing spectra: 3D 1 1 H-13 C-15 N HNCA, HN(CA)CO, CBCA(CO)NH, HNCACB, and HNCO experiments for backbone assignments and HBHA(CBCA)NH, HB-HA(CBCACO)NH, and CC(CO)NH, as well as 3D 1 H-13 C HCCH-TOCSY experiments for side chain assignments. Most of the backbone assignments of [P62A]Ssh10b was obtained by comparison with Ssh10b; the remaining ambiguities were resolved with an HNCA experiment. A 3D NOESY-1 H, 13 C-HMQC experiment was carried out to distinguish the cis and trans conformations of the X-prolyl bond of Ssh10b. Conformational exchange of Ssh10b was monitored using a series of 2D 1 H-15 N correlation experiments for simultaneous measurements of 15 N longitudinal decay and chemical exchange rates at exchange delays of 12, 52, 152, 302, 452, 552, 652, 902, 1102, 1302, and 1602 ms (10). The temperature dependence of 2D 1 H-15 N HSQC spectra of Ssh10b was measured at temperatures ranging from 283 to 330 K at increments of 3.5 K. To determine residues involved in the dimer interface, a 2D version of the four-dimensional 1 H, 13 C-HMQC-1 H, 1 H-NOESY-1 H, 15 N-HSQC experiment (11) was carried out on the 15 N/ 13 C asymmetrically labeled Ssh10b sample, and 15 N was allowed to evolve in the indirect dimension. The mixing time of the experiment was 150 ms. 2D intensity modulated HSQC (12) was used to measure the 3 J NH␣ coupling constants on 15 N singly labeled proteins. The delay for 3 J coupling evolution was set to 30 ms. 2D 1 H-15 N HSQC experiments were also used for exploring the binding behavior of Ssh10b with synthesized 16-bp double-strand DNA fragments (5Ј-GGCAGA-CGCGTCTGCC).
All NMR data were processed and analyzed using Felix 98 (Accelrys Inc.). The data points in each indirect dimension were usually doubled by linear prediction and zero-filled. A 90 to 60°shifted square sine bell apodization was used for all dimensions prior to Fourier transformation.
Nick Closure Assay-The single-nick plasmid pUC18 was prepared as described previously (14). The nicked plasmid (1 g) was incubated with Ssh10b or [P62A]Ssh10b at various mass ratios for 5 min at temperatures of 298 or 330 K. The ligating reactions were then per-formed as described previously (15). T4 DNA ligase (3 Weiss units) and Pfu DNA ligase (4 Weiss units) were used for the reactions carried out at 298 and 330 K, respectively. After the ligation reaction, the samples were analyzed by 1.4% agarose electrophoresis in 0.5ϫ Tris-Phosphate-EDTA (13).

Co-existence of Two Ssh10b
Homodimers with Different Conformations-As shown in the previous study, two sets of crosspeaks (for simplicity denoted as "doublets" hereafter) are observed for most residues of Ssh10b in the 2D 1 H-15 N HSQC spectrum ( Fig. 1A) (8). The signal intensities of the resonance doublets are not equal, with one signal stronger than the other in all doublets. Although the results of chemical cross-linking (1) and the related crystal structure (7) suggest that Ssh10b is a dimer, the resonance doublets with unequal signal intensities (  (7), the ⌬␦ values make it clear that the Ssh10b molecule does not exist as a mixture of dimeric and monomeric forms in solution. If this were the case, then the resonance doublets should be observed only for residues located at the dimer interface. The cross-peaks shown in Fig. 3 correspond to the residues at the Ssh10b dimer interface, detected by X-nucleus edited NOESY. When mapped to the sequence, the data indicate that helix ␣ 2 , a portion near the C-terminal of strand ␤ 3 , and the N-terminal part of strand ␤ 4 are involved in the dimeric surface (Fig. 4A). However, ⌬␦ values ( Fig. 2) reveal that the residues of the N-terminal part of strand ␤ 4 give only "singlet" signals, whereas residues in helix ␣ 1 and strands ␤ 1 and ␤ 2 , which are distant from the interface, generate doublet signals (Fig. 4B). The line widths for each pair of doublets are the same. In addition, the ratios of the signal intensities of the doublets are independent of the concentrations (0.05-1.5 mM) of Ssh10b as revealed by NMR experiments (data not shown). These therefore also exclude the possibility of an oligomerization equilibrium. Because Ssh10b is dimeric as confirmed by size-exclusion chromatography (data not shown), it was considered whether two forms of Ssh10b dimer might co-exist in solution. The form with higher signal intensity in doublets was assigned as the T-form and the other with lower signal intensity as the C-form. No "multiplets" other than doublets were observed for most residues of Ssh10b (Fig. 1A), consistent with both the T-form and the C-form as homodimers, with the monomeric subunits arranged symmetrically in each dimer.
The main-chain torsion angle is closely related to the backbone conformation of proteins and can be calculated from the 3 J NH␣ scalar coupling constants. The J-coupling constants of Ssh10b were measured by 2D intensity modulated HSQC (12). The differences between the 3 J NH␣ value for the T-form and for the C-form (⌬ 3 J NH␣ ) at each residue position is shown in Fig. 2. Residues with an absolute ⌬ 3 J NH␣ value greater than 1 Hz are found in segments spanning the whole molecule: the N-terminal region, the loop linking strand ␤ 1 and helix ␣ 1 , helices ␣ 1 and ␣ 2 , and the C-terminal of strand ␤ 4 . This suggested that the main-chain conformations are different in the two forms of the Ssh10b dimer.
Cis and Trans Conformations of Ssh10 - Fig. 5 shows portions of a 1 H-15 N heteronuclear chemical exchange spectrum at an exchange delay of 1.3 s (10). Two categories of doublets, classified by the exchange features, were observed for Ssh10b. Residues Thr 5 , Thr 7 , and Ser 9 gave exchange cross-peaks (Fig.  5A) when the exchange delays were set in the range of 0.1 to 1.6s. However, the remaining doublets shown in Fig. 1A did not give any exchange cross-peaks at the same exchange delays but gave a result like that shown in Fig. 5B for residue Lys 97 . Thr 5 -Pro 6 -Thr 7 -Pro 8 -Ser 9 is an unstructured N-terminal segment of Ssh10b as determined by the chemical shift index (8). The appearances of the exchange cross-peaks of Thr 5 , Thr 7 , and Ser 9 are clearly because of the cis-trans isomerization of the X-prolyl bonds in this segment. Thr 7 is between Pro 6 and Pro 8 and therefore showed two minor peaks in Fig. 1A and four exchange cross-peaks in Fig. 5A. The intensity ratios of the major auto-peaks to the minor ones were about 11:1 at 310 K, a value much higher than that for the remaining doublets in Fig. 1A. Therefore, the remaining doublets were not caused by the cis-trans isomerization of Thr 5 -Pro 6 or Thr 7 -Pro 8 peptide bonds. This was confirmed by the 2D 1 H-15 N HSQC spectrum of [⌬8]Ssh10b (spectrum not shown), in which all resonance doublets remained the same as those in Fig. 1A, except for the absence of the cross-peaks for residues Gly 4 , Thr 5 , Thr 7 , Ser 9 , Met 10 , and Val 11 Fig. 1B. Only a single set of cross-peaks was observed for all residues except the residues Gly 4 , Thr 5 , Thr 7 , Ser 9 , and Asn 10 of [P62A]Ssh10b. Substitution of Pro 62 by Ala 62 eliminated the cis-trans isomerization, so that the mutant Ssh10b dimer was found in a single conformational state. Overlay of Fig. 1, B and A shows that the crosspeaks of [P62A]Ssh10b can be mapped to the cross-peaks of the T-form of Ssh10b. This indicates that the T-form of the Ssh10b homodimer adopts the same conformation as that of the [P62A]Ssh10b homodimer. Fig. 6A shows 1 H-1 H slices at the 13 C ␣ frequencies of Leu 61 and Pro-62 extracted from the 3D NOESY-1 H, 13 C-HMQC (lower strip) and the 3D HCCH-TOCSY (upper strip) spectra for the T-form Ssh10b dimer, whereas Fig. 6B shows those for the C-form Ssh10b dimer. Two inter-residue NOE cross-peaks between 1 H ␣ of Leu 61 and 1 H ␦1 and 1 H ␦2 of Pro 62 could be observed in the 3D NOESY-HMQC strip for the T-form Ssh10b dimer (Fig. 6A). The appearance of these two NOEs characterizes the trans conformation of the Leu 61 -Pro 62 peptide bond in the T-form Ssh10b dimer. In the strips corresponding to the C-form Ssh10b dimer (Fig. 6B), only one NOE cross-peak between the 1 (Fig. 7). The ratio of C-form to T-form was around 0.26 at 318 K and greater than 0.6 below 298 K. Clearly, the T-form of the Ssh10b dimer is dominant, although the population of the C-form increases on decreasing the temperature.
The 1 H N chemical shifts of all resolved cross-peaks in the 2D 1 H-15 N HSQC spectra were measured at different temperatures for the T-form and the C-form Ssh10b dimers. The differences between the 1 H N chemical shifts at 298 K (␦ 298 ) and at 318 K (␦ 318 ) for the T-form (⌬␦ T ϭ ␦ T298 Ϫ ␦ T318 ) and the C-form (⌬␦ C ϭ ␦ C298 Ϫ ␦ C318 ) of a Ssh10b sample containing a high concentration of salt (200 mM KCl), and ⌬␦ T ϭ ␦ T300 Ϫ ␦ T320 and ⌬␦ C ϭ ␦ C300 Ϫ ␦ C320 of a Ssh10b sample containing a low concentration of salt (20 mM KCl), were obtained (data not shown). On increasing the temperature, the 1 H N chemical shifts of the majority of the cross-peaks were shifted upfield (ϩ⌬␦) for both the T-form and the C-form Ssh10b dimers. However, this was not the case for residues Ala 25 , Leu 48 , Val 53 , Arg 57 , and Leu 61 of the C-form of Ssh10b in the sample containing 20 mM KCl and for residues Asn 58 and Asp 63 of the C-form of Ssh10b in the sample containing 200 mM KCl, which all showed downfield shifts (Ϫ⌬␦) on increasing the temperature.
The extent of upfield movements of the 1 H N chemical shifts (ϩ⌬␦) varied for the residues in the Ssh10b dimer that generated resonance doublets. The differences between the changes in the 1 H N chemical shifts of the resonances for the T-form and for the C-form of Ssh10b (⌬⌬␦ T-C ϭ ⌬␦ T Ϫ ⌬␦ C ), obtained from the data of the Ssh10b sample containing 200 mM KCl at temperature of 298 and 318 K, are shown in Fig. 2 and also mapped onto 3D structure (Fig. 4C). For residues Tyr 22 , Val 23 , Ala 25 , Ala 26 , and Leu 27 , located in helix ␣ 1 , and residues Lys 48 , Asp 51 , Val 53 , Glu 54 , Arg 57 , and Asn 58 , located in helix ␣ 2 , the ⌬⌬␦ T-C values are larger than ϩ0.01 ppm. Residues Val 34 , located in the turn between helix ␣ 1 and strand ␤ 2 , and Asp 63 , in the loop linking helix ␣ 2 and strand ␤ 3 , also showed ϩ⌬⌬␦ T-C (Ͼ0.01 ppm). However, negative ⌬⌬␦ T-C with absolute values larger than 0.01 ppm were observed for residues Ser 35 and Ile 37 at the N terminus of strand ␤ 2 , residues Lys 64 , Glu 66 , Gly 73 , Ser 74 , and Gln 75 in strand ␤ 3 , and residues Ile 92 , Ile 94 , Arg 95 , Lys 96 , and Lys 97 at the C terminus of strand ␤ 4 (Fig. 2). Thus, the upfield shifts of the 1 H N resonances of the residues in helices ␣ 1 and ␣ 2 of the T-form were larger than those of the The first region includes helix ␣ 1 and the loop linking strand ␤ 1 and helix ␣ 1 , the second one is the C-terminal of strand ␤ 2 and helix ␣ 2 , and the third one consists of the C terminus and the N terminus of strands ␤ 3 and ␤ 4 , respectively, and the ␤-turn between them (Fig. 2). ⌬␦ T (DNA) 298 , and ⌬␦ C (DNA) 298 (data not shown) showed similar histograms to that of ⌬␦ T (DNA) 318 and ⌬␦ C (DNA) 318 . ⌬␦(DNA) values provided information about the location of DNA binding sites and the local conformational changes of the Ssh10b dimer induced by binding of DNA.
Compare the values of ⌬␦ T (DNA) 318 and ⌬␦ T (DNA) 298 with ⌬␦ C (DNA) 318 and ⌬␦ C (DNA) 298 , respectively, and the differences in the 1 H N chemical shift perturbations by DNA binding between two forms of Ssh10b dimer can be noticed. The residues showing the observable differences are indicated in the 3D structure (Fig. 4D). The differences were found mainly in two helices and in three ␤ strands. However, upon interaction with DNA, 1 H N resonances of the residues in the segments, Ala 41 -Ser 47 and Val 76 -Ile 90 , still remain as singlet. It seems that DNA binding produces similar conformational changes to these two polypeptide segments in the two forms of the Ssh10b molecule, although different changes in the 1 H N resonances between the two forms of the Ssh10b dimer are observed in other regions of the polypeptide chain of the molecule.
Temperature-dependent Interaction of DNA with Ssh10b and [P62A]Ssh10b-The interaction of DNA with the Ssh10b dimer has been found to be temperature-dependent (1). To further investigate the temperature-dependent features of DNA binding, EMSA and nick closure assays were performed on both Ssh10b and the P62A-mutant Ssh10b ([P62A]Ssh10b) under identical experimental conditions.
In the EMSA assay, a 32 P-labeled 108-bp dsDNA fragment (0.5-1 ng) was mixed with different amounts of Ssh10b or [P62A]Ssh10b and analyzed by gel electrophoresis at 293 and 320 K. The bands show the distribution of the products of the DNAprotein interaction by complex size (Fig. 8). The apparent K d was  Fig. 8). However, when the protein concentration was higher, such as 0.16 or 0.32 M, the nature of cooperative binding of Ssh10b to DNA makes the migration patterns of the DNA-Ssh10b complexes at 320 K totally different from those at 293 K, but resemble those of the DNA-[P62A]Ssh10b complexes at both temperatures (lanes 3 and 4 in Fig. 8). Therefore, the mode of binding of the Ssh10b dimer to DNA in the complex at 293 K must be different from that at 320 K or in the DNA-[P62A]Ssh10b complex at either temperature.
Nick closure assays were carried out to detect the capabilities of the proteins to constrain DNA in supercoils at 298 or 330 K. In the nick closure assay, a single-nick plasmid pUC18 was ligated in the absence and presence of the proteins. The results for different protein:DNA ratios is shown in Fig. 9. The assay in the absence of protein (lane 1 in Fig. 9) was performed as a control. At 298 K, addition of Ssh10b to the reaction mixture produced only a weak CCC (covalently closed circular plasmid) band with a high supercoil density at a protein:DNA mass ratio of 2:1 (Fig. 9, upper left panel, lane 4). However, at 330 K the ability of the bound Ssh10b to introduce supercoils into the plasmid increases dramatically. [P62A]Ssh10b showed similar ability to introduce supercoils into the plasmid at 330 K. However, unlike bound Ssh10b at 298 K, [P62A]Ssh10b was capable of introducing supercoils into the plasmid over the whole range of the protein concentrations at 298 K (Fig. 9, lower left panel). Clearly, the abilities of Sshb10b and [P62A]Ssh10b to affect the  4). The gel of Ssh10b at 320 K was slightly under-exposed to x-ray film, and that of [P62A]Ssh10b at 320 K was over-exposed. topology of DNA are similar at high temperature (330 K) but different at low temperature (298 K).
The results of EMSA and nick closure assays indicate that [P62A]Ssh10b, existing in a single trans conformation in solution, shows the same features upon interaction with DNA at both high and low temperatures. Therefore, the temperaturedependent nature of the interaction of Ssh10b with DNA correlates with the two conformations of the Ssh10b dimer.

Different Conformational Features of the T-form and C-form Ssh10b
Homodimer-The conformational features of a protein are strongly influenced by factors that affect the strength of intramolecular hydrogen bonds. Changes in the chemical shifts of 1 H N resonance are a sensitive indicator of changes in the strength of hydrogen bonding. The chemical shifts of 1 H N resonances are affected by hydrogen bond acceptors, particularly carbonyl groups (20,21). Wagner and co-workers (22,23) demonstrated that 1 H N chemical shifts depend on the inverse third power of the distance between the 1 H N and the hydrogen bond acceptor. In the case of hydrogen bonding with CϭO, large downfield shifts are observed for strongly hydrogen-bonded amide protons. Within protein secondary structure, the 1 H N of the ith residue forms a hydrogen bond with the CϭO of the (i-4)th residue in an ␣-helix, and in an antiparallel ␤-sheet, the 1 H N and the CϭO of a residue in one ␤ strand forms hydrogen bonds with the CϭO and the 1 H N of a residue in the opposite strand. When the temperature is increased, the thermal fluctuations of an ␣-helix or a ␤-sheet are enlarged, and the average distance between the 1 H N and the CϭO increases. As a consequence of weakened hydrogen bonding, the chemical shifts of the 1 H N resonances will tend to move upfield.
The 1 H N resonances of almost all cross-peaks in the 2D 1 H-15 N HSQC spectrum of the Ssh10b molecule were shifted upfield on increasing the temperature. This is thus consistent with a general weakening of the hydrogen bonding in both the T-form and the C-form of the Ssh10b dimer on increasing the temperature. However, the temperature-dependent shifts of the 1 H N resonances were different in size for the T-form (⌬␦ T ) and the C-form (⌬␦ C ). This then suggests a difference between the hydrogen bonding strengths within the secondary structure of the two forms of the Ssh10b dimer. The values of ⌬⌬␦ T-C ϭ ⌬␦ T Ϫ ⌬␦ C for each residue are shown in Fig. 2. On increasing the temperature, the residues in helices ␣ 1 and ␣ 2 show larger upfield shifts of the 1 H N resonances for the T-form (ϩ⌬⌬␦ T-C ), whereas the residues involved in the antiparallel ␤-sheet show larger upfield shifts for the C-form (Ϫ⌬⌬␦ T-C ) (Fig. 2). In the antiparallel ␤-sheet formed by ␤ 2 -␤ 4 -␤ 3 strands, the 1 H N and CϭO groups of residues Ile 94 , Arg 95 , Lys 96 , and Lys 97 form hydrogen bonds with CϭO and 1 H N of Ile 37 , Glu 66 , Ser 35 , and Lys 64 , respectively. Residues Ile 94 , Arg 95 , Lys 96 , and Lys 97 are located in the C-terminal of strand ␤ 4 . Residues Ile 37 and Ser 35 are located in the N-terminal region of strand ␤ 2 , and Glu 66 and Lys 64 are located in the N-terminal region of strand ␤ 3 . Thus, on increasing the temperature, the lengthening of the hydrogen bond distances in the portion of antiparallel ␤-sheet near to the C-terminal of the Ssh10b molecule is greater for the C-form than that for the T-form Ssh10b dimer. Conversely, the lengthening of the hydrogen bond distances in the ␣-helices is greater for the T-form. These results suggest that the spatial packing of the residues is tighter in the ␣-helices and looser in the portion of the antiparallel ␤-sheet near the C-terminal of the molecule for the C-form than that for the T-form of Ssh10b molecule.
Further support for different temperature-dependent features of the secondary structure of the T-form and the C-form of the Ssh10b dimer was from the observation that for a few 1 H N resonances, the direction of the temperature-dependent shift is different in the two forms. The 1 H N resonances of residues Ala 25 , Leu 48 , Val 53 , Arg 57 , and Leu 61 for the sample containing 20 mM KCl, and of residues Asn 58 and Asp 63 for the sample containing 200 mM KCl, shifted downfield for the C-form and upfield for the T-form of Ssh10b molecule. Residues Ala 25 , Leu 48 , Val 53 , Arg 57 , and Asn 58 are located in ␣-helices. Leu 61 and Asp 63 are the nearest neighbors of Pro 62 , located in the loop linking helix ␣ 2 and strand ␤ 3 . The magnitudes of downfield shifts of the 1 H N resonances of these residues upon increasing temperature were in the range 0.6ϳ2.8 ppb/K. In addition, the ⌬ 3 J NH␣ values for residues Leu 24 , Leu 27 , Lys 48 , Val 53 , and Glu 54 were all greater than Ϯ1 Hz (Fig. 2), corresponding to a change in backbone torsion angle, ⌬ Ͼ Ϯ10°, between the two forms of Ssh10b molecule. Residues Ala 25 and Asn 58 also showed small variations of ⌬ 3 J NH␣ (Fig. 2). Thus, differences in the hydrogen bonding strengths in the secondary structure of the two forms correlate with main-chain conformational differences between the T-form and the C-form Ssh10b molecule.
DNA Binding Sites on the T-form and the C-form ssh10b Dimers-Proteins bind in the major or minor grooves of DNA and some protein structures have contacts with DNA in both grooves simultaneously (24). In the model proposed for the DNA-Alba complex (7), the Alba dimer interacts simultaneously with a major groove and the two flanking minor grooves of the DNA. Residues Lys 16 , Lys 17 , and Arg 42 in the central "belly" of Alba dimer are involved in DNA binding at the major groove, and the ␤-hairpin of Alba interacts with the minor grooves.
Ssh10b lacks the first three N-terminal amino acid residues of Sso10b but is otherwise identical in sequence. In fact, the protein used to solve Alba crystal structure is identical with Ssh10b (7). Thus, Ssh10b is supposed to have the same DNA binding sites shown by the DNA-Alba complex. Examination of the interaction of the Ssh10b dimer with DNA in solution by NMR spectroscopy revealed the location of the DNA binding sites on both the T-form and the C-form Ssh10b molecule. The DNA binding regions of the