Specificity and Zn2+ Enhancement of the S100B Binding Epitope TRTK-12*

The calcium-binding protein S100B (an S100 dimer composed of two S100β monomers) is proposed to act as a calcium-sensory protein through interactions with a variety of proteins. While the nature of the exact targets for S100B has yet to be defined, random bacteriophage peptide mapping experiments have elucidated a calcium-sensitive “epitope” (TRTK-12) for S100B recognition. In this work, interactions of TRTK-12 with S100B have been shown to be calcium-sensitive. In addition, the interactions are enhanced by zinc binding to S100B, resulting in an approximate 5-fold decrease in the TRTK-12/S100B dissociation constant. Moreover, Zn2+ binding alone has little effect. TRTK-12 showed little evidence for binding to another S100 protein, S100A11 or to a peptide derived from the N terminus of S100B, indicating both a level of specificity for TRTK-12 recognition by S100B and that the N-terminal region of S100B is probably not involved in protein-protein interactions. NMR spectroscopy revealed residues most responsive to TRTK-12 binding that could be mapped to the surface of the three-dimensional structure of calcium-saturated S100B, revealing a common region indicative of a binding site.

The S100s are a group of proteins that belong to the EF-hand calcium-binding protein family (1)(2)(3). This family includes such mechanistically well understood proteins as the calcium-dependent muscle sensor troponin-C, the ubiquitous enzyme regulator calmodulin, and the visual signaling molecule recoverin. Signaling by these molecules is controlled through calcium binding to the EF-hand protein and subsequent induction of a conformational change that modifies protein-protein interactions with a target protein. For example, troponin-C undergoes a calcium-induced conformational change, allowing a strengthening of its interactions with a second member of the troponin complex, troponin-I (4,5). An analogous mechanism has been proposed for several S100 proteins, allowing them to control such diverse processes as protein phosphorylation, cytoskeletal protein assembly, neurite outgrowth, and cell cycle regulation through a variety of calcium-sensitive interactions with other proteins (3,6,7). S100B (an S100 dimer composed of two S100␤ monomers) is one member of the S100 protein family for which several potential cellular targets have been identified. For example, po-lymerization of cellular architecture molecules such as glial fibrillary acidic protein and tubulin can be inhibited through a calcium-dependent interaction with S100B (8 -10). Alternatively, S100B can inhibit the phosphorylation of proteins such as myristoylated alanine-rich C kinase substrate (11), the Alzheimer protein Tau (12,13), and p53 (14,15) through interaction with these proteins rather than with the kinase responsible. This action is different from that of the S100 protein S100A1 2 , which has been shown to inhibit phosphorylation of the myosin protein kinase twitchin through interaction with its regulatory region (16) in a fashion reminiscent of calmodulin's action with myosin light chain kinase (17).
In order to clarify potential target proteins, a bacteriophage random peptide library has recently been used to define a recognition sequence for S100B (18). These studies showed that a 12-residue sequence containing the consensus motif (K/R)(L/ I)XWXXIL was sufficient to bind to S100B in a calcium-sensitive manner. Further studies have shown this peptide (TRTK-12) successfully competes with other proteins such as glial fibrillary acidic protein and CapZ for calcium-sensitive S100B binding (10,18). Similar approaches have identified several 15-residue sequences, analogous to the sequences for myosin light chain kinase, melittin, and mastoparan, as representative samples of the calcium-dependent target proteins for calmodulin (19). A clear distinction exists between these binding "epitopes" for S100B and calmodulin as the sequences of TRTK-12 and the calmodulin peptides are not related by alignment. However there is a similarity in composition in that both peptides have a preponderance of hydrophobic and basic residues.
Recently, NMR spectroscopy and x-ray crystallography have been used to determine the three-dimensional structures of calcium-saturated human (20), rat (21), and bovine S100B (22). The structures revealed the S100B dimer has two symmetric monomers each comprising two EF-hand calcium binding sites. The N-terminal EF-hand is formed from helices I and II where helix I (and IЈ from the other monomer) are integral to the maintenance of the dimer. Likewise the C-terminal calcium binding site (site II) is composed of helices III and IV where helix IV (and IVЈ) interact at the dimer interface. Calcium binding to S100B has a minimal effect on the conformation of the N-terminal site I but has a pronounced effect on the canonical C-terminal EF-hand. This results in a reorientation of helix III with respect to other helices in the protein (20 -23). Further, it has been observed that a hydrophobic region composed of several residues near the C terminus of helix IV and in the linker between sites I and II is present on the surface of Ca 2ϩ -S100B 1 (20). Based on this structural data, these regions have been proposed to form a possible recognition site for calcium-sensitive protein-protein interactions in S100B. The composition of this site, primarily hydrophobic and acidic residues, is reminiscent of the protein recognition surface in calmodulin (17). In addition, the amino acid sequences in the C-terminal helix and linker show the least sequence similarity among the S100 protein family, suggesting a rationale for target protein specificity (3,24,25).
Several groups have reported that some of the S100 proteins including S100B, S100A1 2 , S100A6, S100A11, calgranulin C, and S100A3 are not only calcium-binding proteins but also bind Zn 2ϩ with high affinity (26 -30). Further, zinc binding has the pronounced effect of increasing calcium affinity in S100B (26) and calgranulin C (29) by at least 10-fold. Such an observation is unique in the EF-hand calcium-binding protein family and may indicate a new mode for calcium regulation and signaling in the S100 proteins. The impact of these observations was recently demonstrated for the giant protein kinase twitchin, where the addition of Zn 2ϩ to calcium-bound S100A1 2 increased the kinase activation by more than 30-fold over that with calcium alone (16). With this in mind, the current work studies the interaction of the 12-residue peptide TRTK-12 with S100B as a function of the divalent metal ions Mg 2ϩ , Ca 2ϩ , and Zn 2ϩ in order to understand the influence of each metal ion on TRTK-12 binding. We have used the S100 protein S100A11 and an N-terminal peptide from S100B to determine whether, at least in this case, TRTK-12 binding is specific for S100B. Further, we have used NMR spectroscopy and the three-dimensional structure of human Ca 2ϩ -S100B to highlight residues that may be important for TRTK-12 interaction. Recombinant human S100B was expressed in Escherichia coli (strain N99) and purified to homogeneity as described previously (32). The backbone amides of alanine and phenylalanine residues of S100B were selectively 15 N-labeled and purified as described previously (33). TRTK-12 (TRTKIDWNKILS) peptide was synthesized and purified as reported (18). Bacterially expressed and purified S100A11 (34) was a kind gift of Dr. Michael Walsh (University of Calgary, Calgary, Alberta, Canada). The hs1bI peptide (residues 1-46) from human S100B (SELE-KAMVALIDVFHQYSGREGDKHKLKKSELKELINNELSHFLEE) was custom synthesized by Chiron Mimotopes (Clayton, Victoria, Australia) using Fmoc (N-(9-fluorenyl)methoxycarbonyl)chemistry and purified by reversed-phase high pressure liquid chromatography.
Fluorescence Spectroscopy-Spectra were obtained using a Hitachi F-4010 fluorescence spectrophotometer equipped with a stirred cell holder. Tryptophan fluorescence was excited at 295 nm and emissionscanned from 305 to 450 nm using an emission band pass of 5 nm. Titrations of the TRTK-12 peptide with S100B were followed by monitoring the increase in fluorescence at 332.8 nm. The concentrations of the TRTK-12 peptide and S100B stock solutions were determined from absorbance spectra using extinction coefficients of ⑀ 280 ϭ 5600 cm Ϫ1 M Ϫ1 for TRTK-12 and ⑀ 280 ϭ 3400 cm Ϫ1 M Ϫ1 for S100B. Samples of TRTK-12 were typically made in 50 mM KCl, 50 mM Tris buffer at pH 7.2. Additions of S100B or hs1bI were made using 1-2-l volumes of the proteins in 50 mM KCl, 50 mM Tris buffer at pH 7.2 using a calibrated Hamilton 10-l syringe. Total sample volumes did not change by more than 3%.
NMR Spectroscopy-All NMR spectra were acquired on a Varian Unity 500 MHz spectrometer equipped with a triple-resonance, pulsedfield gradient probe. Carrier frequencies used were centered at 120.0 ( 15 N) and 4.73 ( 1 H) ppm. One-dimensional 1 H NMR spectra for TRTK-12 were collected at 25°C. TRTK-12 (ϳ1 mg) was dissolved in 0.5 ml of 20 mM CD 3 CO 2 Na, 50 mM KCl, pH 6.5. Typically, a spectral width of 6000 Hz was used with the transmitter set on the H 2 O resonance. Water suppression was accomplished during a 2.0-s relaxation delay between transients using a weak presaturation pulse. All spectra were referenced to the trimethylsilyl resonance of sodium 2,2-dimethyl-2silapentane-5-sulfonate at 0.00 ppm, zero-filled to 65,536 points, and processed using line broadening of 0.5 Hz. Two-dimensional 1 H- 15 N HSQC experiments were acquired on a 0.5 mM S100B sample at 35°C using the sensitivity-enhanced method (35) as described previously (33).

RESULTS
The calcium-binding protein S100B has been suggested to interact with a variety of cellular targets in a calcium-sensitive fashion. Rather than utilizing a speculative target, we have used a synthetic peptide, TRTK-12, to probe the calcium and zinc-sensitive binding to S100B. With several three-dimensional structures of S100B now in hand, we have used this data to identify potential sites for TRTK-12 interaction with Ca 2ϩ -S100B and correlate this with the identification of a potential biological target.
Interaction of TRTK-12 with S100B-The intrinsic fluorescence spectrum of TRTK-12 is shown in Fig. 1A. The peptide displays an emission maximum at 354 nm for the single Trp 7 in the sequence. This wavelength of emission is consistent with the tryptophan located in a polar aqueous environment. For interaction with S100B, this tryptophan provided a convenient marker, since S100B itself contains no tryptophan residues. The addition of two equivalents of apo-S100B to TRTK-12 resulted in a minimal change in the TRTK-12 tryptophan fluorescence, indicating that any interaction between the apo-S100B calcium-binding protein and TRTK-12 is very weak. The addition of physiological levels of Mg 2ϩ (5 mM) to the apo-S100B/TRTK-12 solution had little effect on the tryptophan fluorescence of TRTK-12 despite previous observations that Mg 2ϩ may bind to S100B (36). In contrast, the addition of saturating amounts of calcium to the apo-S100B/TRTK-12 solution resulted in a 40% enhancement and a blue shift to 332.8 nm for tryptophan fluorescence. These observations are consistent with previous results showing that TRTK-12 is able to interact with the calcium form of S100B (18,37). Further, the blue shift indicates the tryptophan residue moves to a more nonpolar environment. S100B has been shown to tightly bind 2 mol of Zn 2ϩ per monomer, and binding of this metal causes a significant change in the surface hydrophobicity of the protein (26). As shown in Fig. 1A, binding of Zn 2ϩ to apo-S100B resulted in little enhancement or shift of the TRTK-12 fluorescence. These observations indicate that calcium is the primary metal responsible for TRTK-12 binding to S100B.
To determine whether TRTK-12 fluorescence and binding to Ca 2ϩ -S100B could be enhanced or reduced by other metal ions, we examined the additional influences of Mg 2ϩ and Zn 2ϩ on tryptophan fluorescence (Fig. 1B). In the presence of Ca 2ϩ -S100B the addition of 5 mM Mg 2ϩ had a small negative effect on the fluorescence intensity of TRTK-12. This finding was similar to the observed small effect in Fig. 1A, where the addition of Mg 2ϩ alone had only a weak effect on Trp 7 fluorescence. In contrast, the addition of Zn 2ϩ to the Ca 2ϩ -S100B solution resulted in an approximate 10% increase in tryptophan fluorescence with no further change in the emission maximum. These results are consistent with Zn 2ϩ binding to S100B and enhancement of the interaction of TRTK-12 with Ca 2ϩ -S100B. The addition of 5 mM Mg 2ϩ to this sample resulted in a small decrease in tryptophan fluorescence similar to that observed in the absence of Zn 2ϩ .
While the above work and other studies have indicated that TRTK-12 interacts with S100B in a calcium-sensitive manner, calcium-saturated S100B, S100␤, and S100A11, respectively; HSQC, heteronuclear single quantum coherence spectroscopy. the specificity of TRTK-12 for other S100 proteins has not been examined. We investigated this using an S100 protein that has some sequence differences from S100B, especially in the linker and C-terminal regions. S100A11 is an S100 protein (also called S100C) originally isolated from cardiac muscle (34), which, like S100B, has been shown to bind both Ca 2ϩ and Zn 2ϩ (38). Fig. 1C shows the tryptophan fluorescence spectra obtained for the addition of Ca 2ϩ -S100B and Ca 2ϩ -S100A11 to a TRTK-12 solution. The spectra show the characteristic shift in TRTK-12 fluorescence in the presence of Ca 2ϩ -S100B. However, the addition of Ca 2ϩ -S100A11 yielded little change in the TRTK-12 fluorescence emission wavelength or amplitude, indicating that Ca 2ϩ -S100A11 does not perturb TRTK-12 fluorescence. This result is consistent with little or no interaction between TRTK-12 and Ca 2ϩ -S100A11.
In an effort to localize the region of Ca 2ϩ -S100B that interacts with TRTK-12, we studied the effect of a 46-residue peptide comprising a single EF-hand from the N-terminal region of S100B (hs1bI) on TRTK-12 fluorescence. This peptide has been shown to be mostly ␣-helical by circular dichroism spectroscopy, similar to S100B, and is able to form a tetramer analo-gous in structure to the arrangement of the four EF-hand motifs in the S100B dimer (39). As shown in Fig. 1D, the addition of hs1bI in the presence of calcium resulted in little change in TRTK-12 fluorescence compared with TRTK-12 alone. This observation indicates that the N terminus of S100B alone is not sufficient to interact with TRTK-12.
TRTK-12 Affinity for S100B-The interaction of TRTK-12 with Ca 2ϩ -S100B was measured by following the tryptophan fluorescence emission at 332.8 nm. Since the previous section showed that TRTK-12 does not interact with apo-S100B, it was important to ensure that Ca 2ϩ -S100B was the major populated species during these titrations. To address this, two titrations of TRTK-12 with S100B were done in the presence of differing amounts of excess calcium (1 and 10 mM). Under these conditions and given the calcium dissociation for S100B (ϳ7-200 M) (40), Ca 2ϩ -S100B should be the dominant species. Thus, the interaction of TRTK-12 should not be affected by the apo-S100B to Ca 2ϩ -S100B equilibrium. Fig. 2 shows three representative titrations of TRTK-12 with Ca 2ϩ -S100B. Fig. 2 shows that TRTK-12 fluorescence at 332.8 nm increases as a function of added Ca 2ϩ -S100B, a result of TRTK-12 binding to Ca 2ϩ - . C shows TRTK-12 alone (a) and with 1.02 M S100␤ and 1 mM CaCl 2 (b) or with 1.18 M S100A11 and 1 mM CaCl 2 (c). D shows TRTK-12 alone (a) and with 1.02 M S100␤ and 1 mM CaCl 2 (b) or 1.0 M hs1bI and 1 mM CaCl 2 (c). In all cases, fluorescence is expressed in relative units, and background buffer has been subtracted. S100B. There is excellent agreement between the two curves, indicating that Ca 2ϩ -S100B must be the predominant form of the protein at both calcium concentrations and that the interaction of TRTK-12 with Ca 2ϩ -S100B is the major event being monitored by these titrations. The shape of the curves does not reveal the stoichiometry of TRTK-12 binding to Ca 2ϩ -S100B, which has been suggested to be either 1 or 2 molecules of TRTK-12/S100B dimer (i.e. 1 molecule of TRTK-12/dimer or 1 molecule of TRTK-12 for each S100␤ monomer) (37). Indeed, fitting of the above data for either a single TRTK-12 or two TRTK-12 molecules binding to Ca 2ϩ -S100B yielded very similar results. However, examination of titration data in the presence of Zn 2ϩ or as monitored by NMR spectroscopy (see below) clearly indicated a stoichiometry of 1 TRTK-12 molecule/Ca 2ϩ -S100␤ monomer. A further titration was done using a 40% more dilute TRTK-12 solution. As expected, the change in fluorescence was correspondingly lower, and a subtler hyperbolic curve was obtained. From these data, a dissociation constant of 0.91 Ϯ 0.17 M for TRTK-12 binding to Ca 2ϩ -S100B was determined.
Fluorescence experiments shown in Fig. 1 indicated that tryptophan emission of TRTK-12 was enhanced in the presence of both Zn 2ϩ and Ca 2ϩ compared with Ca 2ϩ only. To determine whether this was a direct result of increased affinity caused by Zn 2ϩ , titrations to determine the binding affinity of TRTK-12 for Ca 2ϩ -S100B were done in the presence of Zn 2ϩ . As with the previous calcium titration experiments, a Zn 2ϩ concentration of 20-fold the S100␤ concentration was chosen based on the reported dissociation constants of S100B for Zn 2ϩ . The data plotted in Fig. 2 show that Zn 2ϩ binding to S100B in addition to Ca 2ϩ increased the response of TRTK-12 toward S100B. From these data, a stoichiometry of 1:1 TRTK-12:S100␤ monomer is clearly evident. Iterative curve fitting of this data yielded a dissociation constant of 0.18 Ϯ 0.01 for TRTK-12 binding to Zn 2ϩ /Ca 2ϩ -S100B, an approximate 5-fold tighter binding than to Ca 2ϩ -S100B alone.
Regions of TRTK-12 Affected by Binding to S100B-The residues particularly influenced by TRTK-12 binding to Ca 2ϩ -S100B were examined by 1 H NMR spectroscopy. Fig. 3 shows the aromatic, tryptophan indole NH, and methyl regions of a series of 1 H NMR spectra of TRTK-12 with increasing amounts of Ca 2ϩ -S100B added. The complete assignment of TRTK-12 1 H resonances, in the absence of S100B, was accomplished using standard two-dimensional methods (data not shown). In the absence of Ca 2ϩ -S100B, the resonances for TRTK-12 were sharp, with 1 H chemical shifts close to those representative of a random coil structure. As the concentration of Ca 2ϩ -S100B was increased, several of the resonances in TRTK-12 shift and broaden dramatically, while others are less affected. For example the indole NH of Trp 7 in TRTK-12 broadens and shifts downfield by 0.25 ppm. In addition, this resonance becomes similar in line width to those observed for other amide resonances in Ca 2ϩ -S100B, indicating that this residue now has similar relaxation properties as the Ca 2ϩ -S100B protein. The magnitude of the shift of the Trp 7 indole NH indicates that the k off for the TRTK-12⅐Ca 2ϩ -S100B complex is on the order of 125 s Ϫ1 . In the methyl region, Ile 5 from TRTK-12 experiences similar line broadening as Trp 7 but has a 0.3 ppm upfield chemical shift change. The figure also indicates that line broadening is not consistent for all residues in the TRTK-12 peptide. An interesting observation is the differential changes in Thr 1 and Thr 3 ␥-CH 3 groups when Ca 2ϩ -S100B is added. In the early additions of Ca 2ϩ -S100B, the ␥-CH 3 group of Thr 3 is broadened more so than that of Thr 1 . Together, these observations provide evidence that the TRTK-12 peptide is binding to Ca 2ϩ -S100B.
Regions of S100B Affected by TRTK-12 Binding-Fluorescence studies showed that the TRTK-12 tryptophan fluorescence was affected very little in the presence of apo-S100B. This was reinforced by 1 H-15 N HSQC spectra of apo-S100B in the absence and presence of TRTK-12, which indicated the backbone resonances of apo-S100B are not influenced by equimolar amounts of TRTK (data not shown). Qualitatively, this would indicate that the affinity of TRTK-12 for apo-S100B is conservatively 10-fold larger than the apo-S100B concentration (1 mM S100␤).
In previous work, we determined that calcium binding to S100B results in excessive line broadening, making studies of the Ca 2ϩ -S100B species more difficult than apo-S100B. We had shown that this line broadening was a result of aggregation of Ca 2ϩ -S100B in the absence of a target protein such as TRTK-12 (32). To reduce the possibility of Ca 2ϩ -S100B aggregation in this work, we studied the interaction of TRTK-12 with Ca 2ϩ -S100B by the addition of incremental amounts of Ca 2ϩ to apo-S100B/TRTK-12 solutions. This yielded identical results compared with direct TRTK-12 addition to Ca 2ϩ -S100B. As shown in Fig. 4A, the 1 H-15 N HSQC spectrum of Ca 2ϩ -S100B and Ca 2ϩ -S100B/TRTK-12 show some significant differences. In particular, the largest changes in the N and NH chemical shifts (weighted average Ͼ0.25 ppm) were noted for residues Asp 12 , Ser 41 , Phe 43 , Ile 47 , Val 53 , Val 56 , Thr 59 , Asp 61 , Ala 78 , Ala 83 , Cys 84 , Phe 87 , Phe 88 , and His 90 . In the case of alanine and phenylalanine residues, these assignments were confirmed using a specifically 15 N-labeled sample (Fig. 4B). Interestingly, we observed little change in the position of Ile 11 upon binding of TRTK-12. This was consistent with observations for binding of a p53 peptide to S100B (41) but different from a previous study of TRTK-12 binding to bovine S100B, where Ile 11 was observed to undergo the largest shift of any residue (37). DISCUSSION The focus of this work was to determine how the interaction of the TRTK-12 binding epitope for S100B varies with metalion binding and whether this binding showed a degree of specificity for S100B. These aspects of S100B interaction have not been dealt with in previous studies with TRTK-12. In addition, we have used our data to probe a region on the Ca 2ϩ -S100B three-dimensional structure that may be responsible for peptide interaction.
The affinity of TRTK-12 for Ca 2ϩ -S100B is approximately 1 M. This value is higher than previously measured in the absence of added salts (0.15 M) and may reflect a sensitivity of peptide binding to ionic strength which is known to have a significant negative influence on S100B calcium affinity. The magnitude of the TRTK-12 dissociation constant is consistent with those found for other calcium-binding protein-peptide complexes such as TnI peptides binding to skeletal muscle troponin C (24 M) (4) and caldesmon peptides binding to calmodulin (1 M) (42, 43). The interaction of TRTK-12 with S100B is Ca 2ϩ -specific, since neither Mg 2ϩ nor Zn 2ϩ binding to FIG. 3. Series of 500-MHz 1 H NMR spectra showing the effects of Ca 2؉ -S100␤ on TRTK-12 peptide. The lower spectrum shows 1.1 mM TRTK-12 in 90% H 2 O, 10% D 2 O with 20 mM CH 3 CO 2 Na-d 3 , 50 mM KCl, pH 6.5, at 25°C. Assignments of the TRTK-12 resonances were done using standard two-dimensional methods and are indicated. The sample was titrated with Ca 2ϩ -S100␤, giving the Ca 2ϩ -S100␤/TRTK-12 ratios shown at the right to a final concentration of 1.2 mM S100␤, 3.08 mM Ca 2ϩ for 1.1 mM TRTK-12 peptide. The dotted lines represent resonances for Trp 7 ⑀, Thr 3 ␥-CH 3 , and Ile 5 ␥-CH 3 , which shift and broaden as a function of added Ca 2ϩ -S100␤. The spectrum of apo-S100␤ is identical to that obtained in the absence of calcium (33). The arrows show the change in resonance position upon the addition of calcium resulting from both calcium binding to S100␤ and binding of TRTK-12. S100B could stimulate peptide binding. The finding that Zn 2ϩ binding alone to S100B is unable to promote TRTK-12 binding is in contrast to results using the hydrophobic probe TNS, where a large increase in TNS fluorescence is observed in the presence of Zn 2ϩ -S100B (26,27,44). This observation occurs as a result of the interaction of TNS with the Zn 2ϩ form of S100B. Since TNS is known to be a probe for hydrophobic surface rather than a specific binding site, these differences indicate that the hydrophobic surface exposed in S100B by Zn 2ϩ binding is not specific for TRTK-12 binding. In turn, this probably indicates that the protein does not adopt a proper conformation in the Zn 2ϩ form to allow target protein binding.
The most dramatic effect on peptide binding to S100B is in the presence of both Zn 2ϩ and Ca 2ϩ , where an increase in peptide affinity of about 5-fold is noted compared with the presence of Ca 2ϩ only. This observation is consistent with results for S100A1 2 , where a 30-fold increase in twitchin kinase activity was observed upon the addition of Zn 2ϩ to the calcium form of the protein (16). Together with previous observations that S100B is able to bind two Zn 2ϩ ions per monomer (26,44) this indicates that binding of TRTK-12 is enhanced by Zn 2ϩ binding to S100B in the presence of Ca 2ϩ only. Since TRTK-12 does not appear to bind to Zn 2ϩ -S100B as judged by the present experiments, it would appear the calcium binding to S100B is critical. This indicates that Zn 2ϩ plays more of a structural role in proteins such as S100B and S100A1 2 rather than a regulatory role. Similar proposals have been suggested for the S100 proteins S100A3 (30) and calgranulin C (29). Consistent with this idea, the affinity of Zn 2ϩ for S100B (10 Ϫ7 M) is near the physiological intracellular concentrations of free Zn 2ϩ , which are thought to be in the micromolar range. It has been previously proposed that Zn 2ϩ can act as a structural ion in other S100 proteins based on the observation of a His-X 3,4 -His pattern found near the C termini (30). This pattern is similar to that observed for Zn 2ϩ -catalytic sites that frequently display a His-X 3 -His motif (45). It is intriguing to note that the threedimensional structure of human Ca 2ϩ -S100B shows that one possible Zn 2ϩ binding site might include residues His 85 and His 90 based on their proximity in the structure (20). This region ultimately appears to be important for peptide binding, since several changes in chemical shift and exposed surface area are noted for residues in the C terminus.
The measured affinity of TRTK-12 for Ca 2ϩ -S100B represents at least a 5-fold increase in affinity compared with that for a 23-residue peptide from the tumor suppressor protein, p53 (residues 367-388) (41). Further, Zn 2ϩ binding to S100B did not have a notable effect on p53 peptide binding. These observations indicate there are some clear differences between these two peptide species. As indicated by Wilder et al. (15), the p53 peptide studied previously fits some of the TRTK-12 consensus sequence but lacks residues Ile 10 and Leu 11 . Further, sequence analysis of nine unique p53 sequences from the PIR data base reveals that all forms of p53 do not resemble the TRTK-12 binding motif for these last two residues. 2 This may explain the decreased affinity of p53 for S100B. It is intriguing that secondary structure prediction methods show that TRTK-12 should form an amphipathic ␣-helix between residues Lys 4 and Leu 11 . If this structure exists in the S100B-protein complex, it would place residues Thr 3 , Trp 7 , Leu 10 , and Ile 11 on one side of the helix, allowing integral contact of Leu 10 and Ile 11 with S100B. On the surface, this would provide differences between the interactions of TRTK-12 with Ca 2ϩ -S100B compared with p53.
The results presented in this work allow an initial overview of the TRTK-12 interaction with Ca 2ϩ -S100B previously attempted in the absence of a Ca 2ϩ -S100B three-dimensional structure (37). Residues in the N terminus (Asp 12 ), linker and N terminus of helix III (Ser 41 , Phe 43 , Ile 47 , Val 53 ), and C terminus (Ala 78 , Ala 83 , Cys 84 and Phe 88 ) undergo the most significant changes in chemical shift upon TRTK-12 binding. Since TRTK-12 has little interaction with the N-terminal peptide, hs1bI, from S100B this would indicate that the linker and C terminus are more important. Using the structure-activity relationship (46), it can be proposed that residues change chemical shift as a direct result of interaction with TRTK-12. Fig. 5 shows the structure-activity relationship for Ca 2ϩ -S100B based on changes in chemical shift. The S100B structure clearly shows these residues comprise a large localized hydrophobic surface on the protein. Further, several residues including Phe 43 , Ala 83 , Phe 87 , and Phe 88 are included in or near this region and increase their accessible surface area more than 20% upon calcium binding (20). These observations are consistent with a region for interaction of TRTK-12 with S100B. S100B has been proposed to interact with a large number of potential target proteins including the head domain of glial fibrillary acidic protein that bears some sequence similarity to TRTK-12 (10). Most recently, studies have probed this interaction with p53 (15,41), guanylate cyclase, and p80 (47,48). During the course of the current work, a series of synthetic peptides were used to map the regions of S100B that stimulate guanylate cyclase activity and phosphorylation of p80. In agreement with our current findings, these studies indicated that the C-terminal region of S100B encompassing residues Thr 81 -Glu 91 was most effective for guanylate cyclase activation. It was also apparent that residues Leu 32 -Leu 40 could elicit a similar response. This region is just N-terminal to residue Ser 41 observed here. Interestingly, our peptide hs1bI (residues 1-46) does not bind to TRTK-12, perhaps suggesting a weaker contribution from the linker region when TRTK-12 is bound and a stronger contribution for guanylate cyclase. Supporting this possibility, it has also been noted that guanylate cyclase does not contain a region similar in sequence to TRTK-12, so this may indicate a fine difference between interaction of S100B and its target proteins. The surface was generated using InsightII (MSI) for the coordinates of human S100B (20).