Identification of contact sites in the actin-thymosin beta 4 complex by distance-dependent thiol cross-linking.

Binding sites of actin and thymosin β4 were investigated using a set of bifunctional thiol-specific reagents, which allowed the insertion of cross-linkers of defined lengths between cysteine residues of the complexed proteins. After the cross-linkers were attached to actin specifically at either Cys, Cys, or to the sulfur atom of the ATP analog adenosine 5′-O-(thiotriphosphate) (ATPS), the actin derivatives were reacted with synthetic thymosin β4 analogs containing a cysteine at one of the positions 6, 17, 28, 34, and 40. Immediate cross-linking as followed by UV spectroscopy was found for Cys of actin and Cys6 of thymosin β4, indicating that the N terminus of thymosin β4 is in close proximity (≤9.2 Å) to the C terminus of actin. In contrast, only insignificant reactivity was measured for all thymosin β4 analogs when the cross-linkers were anchored at Cys of actin. A second contact site was identified by cross-linking of Cys and Cys in thymosin β4 with the ATPS derivative bound to actin, indicating that the hexamotif of thymosin β4 (positions 17-22) is in close proximity (≤9.2 Å) to the nucleotide. The importance of the amino acids 17 and 28 in thymosin β4 for the interaction with actin was emphasized by the finding that thymosin analogs containing cysteine in these positions exhibited strongly reduced abilities to inhibit actin polymerization.

One of the binding sites of T␤4 on the actin molecule seems to be located in subdomain 1 as suggested by cross-linking studies (13,14). In order to gain more knowledge about contact sites in the actin⅐T␤4 complex, we performed a structural analysis using bifunctional thiol-specific reagents of the type alkylene-bis-[5-dithio-(2-nitrobenzoic acid)] for intermolecular crosslinking of two cysteine residues. Such reagents were successfully used for cross-linking two distinct cysteines in muscle actin as well as for preparing a defined disulfide-linked actin dimer (15,16). By varying the length of the cross-linkers (as well as using Ellman's reagent for zero-length cross-linking), information can be obtained about the distance up to which two thiol groups in the complexed proteins can approach. In a first reaction, the cross-linkers (9.2Å to 18.4Å) were anchored monovalently at one of three thiols in monomeric actin. Since native T␤4 does not contain any cysteine, T␤4 analogs were synthesized, each containing cysteine at one of the positions 6, 17, 28, 34, and 40. Thus, the substitutions were distributed over the whole protein but were restricted to hydrophobic amino acids. After adding the T␤4 analogs to the actin derivatives, the kinetics and extents of cross-linking were followed by spectrophotometric analysis of the 2-nitro-5-thiobenzoate released.

MATERIALS AND METHODS
Protein Purification-Actin was prepared from rabbit muscle as described by Spudich and Watt (17) and further purified by a gel filtration step on a Fractogel TSK HW 55 column (3 ϫ 120 cm) (E. Merck, Darmstadt) in buffer G (2 mM Tris, 0.2 mM ATP, 0.1 mM CaCl 2 , 0.02% NaN 3 , pH 7.8). Thymosin ␤4 was isolated from bovine lungs according to Spangelo et al. (18) or obtained by synthesis.
Preparation of Actin Derivatives-Actin 374 SS-(CH 2 ) n -SSAr was prepared by reacting G-actin (3.8 ϫ 10 Ϫ5 M) in buffer G with a 3 molar excess of the reagent ArSS-(CH 2 ) n -SSAr (n ϭ 3, 6 or 9) (see Fig. 1). The cross-linkers were prepared according to Refs. 15 and 16, and their corresponding lengths were 9.2, 13.8, or 18.4 Å, respectively. The mixture was kept at 4°C until one equivalent of 2-nitro-5-thiobenzoate (ArS Ϫ ) was released (⑀ 412 ϭ 14,150 M Ϫ1 cm Ϫ1 ). By exhaustive dialysis in buffer G, the major part of excess reagent was removed together with ArS Ϫ before the protein was purified on a Bio-Rad P2 column (2 ϫ 45 cm) equilibrated with buffer G. Labeling of the actin derivative was 80 -90% as determined from protein concentration of the purified derivative, and the amount of ArS Ϫ detected at 412 nm after cleavage with excess of dithiothreitol (DTT). For zero-length cross-linking, we prepared actin 374 SSAr by reacting G-actin with 3-fold excess of Ellman's reagent under the same conditions. * This work was supported by a grant from the Deutsche Forschungsgemeinschaft. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
To introduce the cross-linkers into position 10 of actin, the cysteine residue in position 374 was blocked by incubating G-actin in buffer G with a 100-fold excess of N-ethylmaleimide (NEM) at 4°C for 30 min. The reaction was quenched with excess DTT, and the protein was separated on a Bio-Rad P2 column equilibrated with buffer G. NEMactin was polymerized by the addition of 0.2 mM EGTA, 1 mM MgCl 2 , and removal of ATP was achieved by incubation with hexokinase (5 units/ml actin solution, Sigma) and 0.4 mM glucose for 90 min at room temperature (19). After centrifugation at 100,000 ϫ g, the pellets were allowed to soften on ice in ADP buffer (2 mM Tris, 1 mM ADP, 0.02% NaN 3 , pH 7.8) for 30 h. After that time, Cys 10 was completely accessible (20) and could be reacted with one end of the cross-linking reagent. The 374 NEM-actin 10 SS-(CH 2 ) n -SSAr was purified as described above. Yield of the labeling reaction was ϳ80 -90%.
Incorporation of ATP␥S-S-(CH 2 ) n -SSAr was performed only with actin blocked at Cys 10 and Cys 374 in order to exclude any unspecific reaction. For this, ADP⅐G-actin with both cysteines exposed was prepared as described above. Reaction with NEM and removal of excess reagent were achieved as described for 374 NEM-actin. The resulting 10/374 (NEM) 2 -actin was polymerized, and ATP was removed by the hexokinase reaction. The 10/374 (NEM) 2 -actin pellet was allowed to soften on ice in the preformed softening buffer in order to incorporate the labeled ATP␥S. Directly before use, excess of ATP␥S-S-(CH 2 ) n -SSAr was removed on a Bio-Rad P2 column (1 ϫ 18 cm) equilibrated with ADP buffer 2 (2 mM Tris, 0.2 mM ADP, 0.1 mM CaCl 2 , 0.02% NaN 3 , pH 7.8) yielding a fraction of actin that contained nearly one equivalent (95%) of the labeled nucleotide.
Cross-linking Studies-In the thymosin ␤4 analogs, cleavage of the S-protecting isopropylthiol residue was achieved by incubating 1 mg of the T␤4 analog in a 200-fold excess of 2-mercaptoethanol in 2 mM Tris, pH 7.5, for 2 h at 4°C. Excess reagent was removed on a Bio-Rad P2 column (1 ϫ 15 cm) equilibrated with the same buffer. The concentration of Cys⅐T␤4 was determined by titrating an aliquot with Ellman's reagent and measuring the released ArS Ϫ at 412 nm. The actin derivatives were mixed with the Cys⅐␤4 analogs at ratios of 2:1 or 1:1 (concentrations of both proteins were in the range of 1-3 ϫ 10 Ϫ5 M) and allowed to react at room temperature. At time intervals of 10, 30, and 90 min, the amount of ArS Ϫ released due to the cross-linking reaction was measured, and the extent of cross-linking was determined (100% values were defined by the Cys⅐T␤4 concentration added). To verify that the extinctions measured at 412 nm reflected the cross-linking events quantitatively, aliquots of the reaction mixtures were taken after 60 min and applied to SDS-PAGE. The amount of actin⅐T␤4 complex formed in the cross-linking reaction was determined by integrating the 47 kDa band.
All steps were performed in an argon atmosphere in order to minimize oxidation of the unprotected cysteine residue in the thymosin analogs.
Viscosimetric Measurements-Actin polymerization was monitored in a Cannon capillary viscosimeter, using a 10 M concentration of G-actin in buffer G. Polymerization conditions were established by the addition of KCl to a final concentration of 100 mM. To investigate the inhibitory capacities of the different thymosin ␤4 analogs, the peptides were added to the actin to a final concentration of 15 M and allowed to incubate for 30 min at room temperature before polymerization was started.
For measuring the effects of the different thymosins on the nucleotide exchange rate, the T␤4 analogs (8 M) were added to the ⑀-ATP buffer. For comparing the nucleotide exchange rates of 10/374 (NEM) 2actin containing either the ATP␥S derivative or ATP as the bound nucleotide, both actin derivatives were purified on a Bio-Rad P2 column (1 ϫ 18 cm) equilibrated with ADP buffer 2, assayed for concentration, and applied to the nucleotide exchange measurements just after elution from the column.

RESULTS
A method was developed that allowed the investigation of contact sites between actin and thymosin ␤4 by assessing whether two thiol groups in the protein complex could approach sufficiently close to allow cross-linking by thiol-specific cross-linkers of different lengths.
Anchoring of the Cross-linkers to Actin-The actin derivatives of the type actin 374 SS-(CH 2 ) n -SSAr (n ϭ 3, 6, 9) were obtained in a nearly quantitive reaction of G-actin with a 3-fold excess of the cross-linkers (Fig. 1). The stoichiometry of the actin derivatives was proved by the release of approximately 1 equivalent of ArS Ϫ during the cross-linking reaction, as well as by analysis of the purified actin derivative, which released 0.8 -0.9 equivalents of ArS Ϫ on the addition of DTT. It was shown that actin derivatives of this type polymerized similar to normal actin apart from a slightly increased critical concentration. In SDS-PAGE, these actin derivatives were indistinguishable from G-actin (Fig. 2a). Likewise we found that binding to native T␤4 was not altered by the modification, as concluded from cross-linking experiments with EDC/NHS yielding the typical 47 kDa band (Fig. 2b). Densitometric evaluation yielded an extent of 25 Ϯ 1% cross-linking for all actin derivatives and G-actin.
For specific labeling of Cys 10 , ATP in 374 NEM-actin was exchanged for ADP, a reaction that initiates a slow unfolding reaction and results in selective and quantitative exposure of this cysteine residue (20). After reaction with a 3-fold excess of reagent, the resulting 374 NEM-actin 10 SS-(CH 2 ) n -SSAr (n ϭ 3, 6, 9) was purified and shown to contain 0.8 -0.9 equivalents of cross-linker as assessed by spectrophotometry in the presence of DTT. The actin derivatives of this type were again indistinguishable from G-actin in SDS-PAGE (Fig. 2a) as well as with respect to their binding capacities for thymosin ␤4 (Fig. 2b).
In order to prepare the actin derivative with the cross-linker anchored at the actin-bound nucleotide, the cross-linking reagent had first to be attached to ATP␥S. The modified ATP␥S was identified by its 1 H NMR spectrum 2 as well as by UV spectrometry (Fig. 3a). The presence of a 1:1 adduct of ATP␥S and nonylene-5-dithio-2-nitrobenzoate was proved by evaluating the amount of ArS Ϫ (⑀ 412 ϭ 14,150 M Ϫ1 cm Ϫ1 (25)) released after treatment with DTT ( Fig. 3b), which corresponds to the amount of cross-linker present in the modified nucleotide. (The molar extinction coefficient of the cross-linker part is ⑀ 338 ϭ 9400 Ϯ 50 M Ϫ1 cm Ϫ1 , a value that agrees with the extinction coefficient previously reported for n-octyl-5-dithio-2-nitrobenzoate (⑀ 338 ϭ 9050 M Ϫ1 cm Ϫ1 (26))). Considering the contribution of the cross-linking part to the absorbance at 259 nm (0.4 ϫ E 338 (26)) the absorption of the adenosine part at that wavelength (⑀ 259 of ATP ϭ 16,415 M Ϫ1 cm Ϫ1 (27)) ( Fig. 3a) reveals a ratio of 1:0.97 for the ATP␥S part and the cross-linking part. The modified nucleotide was exchanged for ADP in 10/374 (NEM) 2 -actin, which was prepared in order to avoid intramolecular cross-linking of the modified nucleotide with the two potentially reactive thiol groups in actin, yielding 10/374 (NEM) 2 -actin⅐ATP␥SS-(CH 2 ) n -SSAr (n ϭ 3, 9). Since the affinity of the modified ATP␥S for actin is lower than that of ATP (see below), loading with the labeled nucleotide was optimized by separating the excess of unbound, labeled ATP␥S just before use. Incorporation of the labeled nucleotide into actin at the time of the experiment was then as high as 95%.
For making sure that ATP␥S could indeed be used as an anchoring point in actin, the affinity of the modified nucleotide to actin was assayed by determining the exchange rate of the modified ATP␥S bound to 10/374 (NEM) 2 -actin for ⑀-ATP. This exchange rate was found to be accelerated 5-fold in comparison with normal ATP bound to 10/374 Preparation of the Thymosin ␤4 Analogs-Since thymosin ␤4 does not contain any cysteine residue, five different T␤4 analogs were synthesized, each containing one cysteine in a defined position (Fig. 4). The distribution of the cysteines in the T␤4 sequence was such as to replace hydrophobic residues only. To ensure that these substitutions did not influence binding of T␤4 to actin, cross-linking studies using EDC/NHS were performed with all analogs. These studies showed that all analogs were still able to bind actin similar to normal T␤4 as indicated by the occurrence of the 47 kDa band in SDS-PAGE (Fig. 5). To confirm this result, and to ensure that the substitution even in the hexamotif of T␤4 had no significant effect on the affinity to actin, the K D value of Cys 17 ⅐T␤4 was determined according to Ref. 13. It was shown to be 0.8 M Ϯ 0.1 M and thus in the same range as the K D value of native T␤4, which was reported to be 0.7-2.0 M (4, 7).
Cross-linking Studies-Cross-linking reactions were detected by measuring changes in absorbance at 412 nm that occur when actin derivatives and thymosin ␤4 analogs were allowed to form a complex. In order to prove that the ⌬OD really reflected cross-link formation between the two proteins, the reaction mixtures were investigated in parallel by SDS-PAGE (Fig. 6). The amounts of the 47 kDa bands, representing the covalently linked actin-thymosin ␤4 complex, were measured by integration of the gel bands and compared with the absorbance values detected at the same time. From the good agreement of the two sets of data, it was concluded that the release of ArS Ϫ measured by UV spectroscopy at 412 nm indeed reflected the formation of cross-links. Proof of the disulfide nature of the linkage between the two proteins was obtained by SDS-PAGE where the 47 kDa band disappeared in the presence of DTT. Since each of the two proteins in the complex was exposing only one thiol group, a positive cross-linking reaction could be taken as evidence that the two thiols had approached to a distance that could be bridged by the length of the crosslinker. In total, more than 40 kinds of cross-linking experiments were performed, the results of which are compiled in Table I  when added to the actin derivatives did not induce the release of ArS Ϫ .
Based on extent and kinetics of the ArS Ϫ release, three types of reactions could be distinguished. In the first type, the reaction proceeded rapidly reaching its end point (Ͼ50%) within less than 10 min (Fig. 7, a-c). For one of these reactions, a complete kinetic analysis was performed showing that the halfmaximal value was actually reached after about 1 min (data not shown). Reaction kinetics of this type were taken as indicating the close proximity of the two thiols in the protein complex. Based on this type of kinetics it was possible to identify three sites of very close contact (Յ9.2 Å) between the two proteins. One of these contacts is between Cys 374 ⅐actin and Cys 6 ⅐T␤4. Cross-linking at this site was almost independent of the length of the cross-linker as yields and kinetics of actin 374 SS-(CH 2 ) n -SSAr were similar when n was 3, 6, or 9. The proximity of Cys 374 ⅐actin and Cys 6 ⅐T␤4 was even close enough to allow for zero-length cross-linking as shown for actin 374 SSAr when allowed to complex with Cys 6 ⅐T␤4. However, zero-length cross-linking was distinctly slower than the cross-linking reactions described first, and thus belongs to the second type of kinetics described below. The two other sites of close contact were identified from the rapid reactions of the cross-linkers attached to the actin-bound ATP␥S with Cys 17 ⅐T␤4 and, to a lower extent, with Cys 28 ⅐T␤4.
In the second type of kinetics, yield of cross-linking was low at the beginning (ϳ10% after 10 min) but became extensive with time (Fig. 7d). It appears that in this type of cross-linking reaction, the two thiols are not in close proximity but can come close to each other due to the mobility of one, or both, of the partners. Examples of this second type of kinetics are, besides the reaction already mentioned, the cross-links between actin 374 SS-(CH 2 ) n -SSAr (n ϭ 3, 6, 9) and the cysteines located in the central part of thymosin ␤4. Particularly Cys 17 ⅐T␤4, and to a much lower extent also Cys 28 ⅐T␤4 showed considerable extents of cross-linking with cysteine 374 of actin, although with low reaction rates. Cross-linking reactions of this type were not regarded as identifying sites of strong contact.
The third type of cross-linking reactions comprises those with very low amounts (Ͻ10%) of ArS Ϫ released during the first 10 min followed by an only slight increase within 90 min (Fig. 7e). This reaction pattern was the most frequent one and, in contrast to the two other types of kinetics, was taken as an indication that the two thiol groups were remote from each other. This type of kinetics was found e.g. in all experiments involving Cys 40 ⅐T␤4, suggesting that this position in thymosin ␤4 must be located distant from both cysteine residues in subdomain 1 of actin as well as from the actin-bound nucleotide. This type of kinetics was likewise found in all crosslinking experiments involving 374 NEM-actin 10 SS-(CH 2 ) n -SSAr (n ϭ 3, 6,9).
Functional Interactions between the Thymosin ␤4 Analogs and Actin-All five thymosin ␤4 analogs obtained by peptide synthesis were able to bind actin as shown from the crosslinking studies illustrated above (Fig. 5). In order to assay the influence of the substitutions in T␤4 on the polymerizationinhibiting capacity, polymerization of G-actin was monitored in the presence of each of the analogs. Generally, we found that all T␤4 analogs, which proved positive in one of the cross-linking reactions, also showed a reduced inhibitory capacity on actin polymerization (Fig. 8). Lowest inhibitory capacities were found for those two analogs that showed the highest yields in the cross-linking reactions with the actin-bound nucleotide (Cys 17 ⅐T␤4 and Cys 28 ⅐T␤4). In line with this, even Cys 34 ⅐T␤4, which reacted with the actin-bound nucleotide only to a small amount, had likewise lost part of its inhibitory capacity. The substitution in position 6 of T␤4 resulted in an only slight alteration in the polymerization-inhibiting capacity, whereas the substitution in position 40 of T␤4 had no influence at all. In accordance with this, the capacity of the latter to inhibit actin polymerization was virtually indistinguishable from that of native T␤4. Yield of cross-linking was determined by densitometric evaluation of the gel bands and was in good agreement with the corresponding spectrophotometric values representing the amount of ArS Ϫ released due to the cross-linking reactions. For lanes 1-4, calculations of the yields of cross-linking took into account that the actin derivatives were present in excess (2:1) over T␤4. 1, actin 374 SS-(CH 2 ) 9 -SSAr ϩ Cys 6 ⅐T␤4 mixed at a ratio of 2:1 (yield of cross-linking, 60%); 2, actin 374 SS-(CH 2 ) 6 -SSAr ϩ Cys 6 ⅐T␤4 mixed at a ratio of 2:1 (yield of cross-linking, 55%); 3, actin 374 SS-(CH 2 ) 3 -SSAr ϩ Cys 6 ⅐T␤4 mixed at a ratio of 2:1 (yield of cross-linking, 54%); 4, actin 374 SSAr ϩ Cys 6 ⅐T␤4 mixed at a ratio of 2:1 (yield of cross-linking, 29%); 5, actin; 6, 374 NEM-actin 10 SS-(CH 2 ) 9 -SSAr ϩ Cys 6 ⅐T␤4 mixed at a ratio of 1:1 (yield of cross-linking, 11%); 7, 374 NEM-actin 10 SS-(CH 2 ) 6 -SSAr ϩ Cys 6 ⅐T␤4 mixed at a ratio of 1:1 (yield of cross-linking, 9%); 8, 374 NEM-actin 10 SS-(CH 2 ) 3 -SSAr ϩ Cys 6 ⅐T␤4 mixed at a ratio of 1:1 (yield of cross-linking, 6%).
Encouraged by the good correlation found between polymerization inhibiting capacities and cross-linking data, we assayed the retardation of the nucleotide exchange rate of actin as another functional parameter of T␤4. The influence of Cys 17 ⅐T␤4 on the nucleotide exchange was examined in comparison with native T␤4 and Cys 40 ⅐T␤4, the latter as an example of a T␤4 analog, which is ineffective in the thiol-specific cross-linking reactions as well as in the polymerization-inhibiting assay. The retardation effect of Cys 17 ⅐T␤4 was found to be indeed partly abolished. While the k value of Cys 40 ⅐T␤4 (k ϭ 2.8 ϫ 10 Ϫ4 Ϯ 0.2 ϫ 10 Ϫ4 s Ϫ1 ) was almost indistinguishable from that of native T␤4 (k ϭ 2.9 ϫ 10 Ϫ4 Ϯ 0.2 ϫ 10 Ϫ4 s Ϫ1 ), the nucleotide exchange rate of Cys 17 ⅐T␤4 was found to be accelerated to a value of k ϭ 4.7 ϫ 10 Ϫ4 Ϯ 0.2 ϫ 10 Ϫ4 s Ϫ1 , a value that approaches the k value of pure actin (k ϭ 6.2 ϫ 10 Ϫ4 Ϯ 0.3 ϫ 10 Ϫ4 s Ϫ1 ) under these conditions (Fig. 9).

DISCUSSION
In order to identify contact sites between actin and T␤4 we successfully used a method of selective cross-linking between thiols that is able to measure the closest approach of two cysteines in the protein complex. By using a set of cross-linking reagents of different lengths, or the procedure of direct activation of one of the thiols with Ellman's reagent, we were able to assay distances between two thiols in the range from 0 to ϳ18 Å.
In obtaining reliable results from this kind of study it was essential that the cross-linking reaction was absolutely thiolspecific and that each of the two proteins exposed only one thiol group. The first condition was assured by the fact that the FIG. 7. Typical reaction kinetics of Cys 6 ⅐T␤4 with actin derivatives carrying cross-linking reagents of different length at Cys 374 or Cys 10 , as monitored by the release of ArS ؊ . The two proteins were mixed as described above, and the amount of ArS Ϫ was determined after 10, 30 and 90 min. Curves a, b, and c represent reaction kinetics defined as type 1 (see "Results"); curve d represents the kinetics of a type 2 reaction; curve e represents the kinetics of a type 3 reaction. a, actin 374 SS-(CH 2 ) 9 -SSAr ϩ Cys 6 ⅐T␤4; b, actin 374 SS-(CH 2 ) 6 -SSAr ϩ Cys 6 ⅐T␤4; c, actin 374 SS-(CH 2 ) 3 -SSAr ϩ Cys 6 ⅐T␤4; d, actin 374 -SSAr ϩ Cys 6 ⅐T␤4; e, 374 NEM-actin 10 SS-(CH 2 ) 9 -SSAr ϩ Cys 6 ⅐T␤4

TABLE I Extents of thiol-specific cross-linking reactions between three types of actin derivatives and five different T␤4 analogs using cross-linkers of different lengths
Values are expressed as percentage of ArS Ϫ released due to the cross-linking reaction (100% values given by the Cys ⅐ T␤4 concentration) as measured by spectrophotometry at 412 nm. Yields of cross-linking were independent on whether one of the compounds was used in excess (2:1). High yields (ϳ50%, or more) of ArS Ϫ obtained after 10 min were taken as indicating sites of closest contact between the two proteins. All values are the average of five experiments (for experimental details see "Materials and Methods"). disulfide-exchange reaction runs with thiols only (28). The second requirement was met for T␤4 in that the synthetic T␤4 analogs used contained only one cysteine each. As for actin, we made use of the fact that actin in buffer G exposes only cysteine 374 (29), which could either be reacted with the cross-linking reagents or be blocked with NEM. By exchanging ATP for ADP in 374 NEM-actin, cysteine 10 could be selectively uncovered (20), thus providing another distinct thiol group to be reacted with the cross-linking reagents. In order to obtain a third well defined anchoring point in actin, the cross-linkers were attached to the ATP analog ATP␥S. The modified nucleotide was characterized by UV and 1 H NMR spectroscopy and shown to contain adenosine and the cross-linking reagent at a ratio of 1:1. Although the exchange rate of the modified actin-bound nucleotide was increased by a factor of five over that of ATP, binding of the labeled ATP␥S was regarded as tight enough to provide the third point of attachment for the cross-linkers. For all actin derivatives, it was shown that they behaved similarly, or even identically, to G-actin with respect to polymerization, appearance in SDS-PAGE, and binding to native T␤4. For all T␤4 analogs, it was proven that the substitutions did not impair complex formation with actin. As in titrations using Ellman's reagent, formation of a crosslink between an actin derivative and a T␤4 analog was accompanied by the release of ArS Ϫ detectable at 412 nm, which allowed easy determination of the extent and kinetics of crosslink formation by UV spectrometry. Since these data were in very good agreement with those obtained by integrating the corresponding gel bands in SDS-PAGE, it was concluded that the release of ArS Ϫ reflected the cross-link formation quantitatively. The extent of cross-linking as determined by spectrophotometry was independent on whether one of the components was used in excess (2:1) and never exceeded 75%. The incompleteness of the reaction may be explained by the K D value of the actin⅐T␤4 complex (ϳ1 M) limiting complex formation. In addition, the extent of cross-linking may be lowered by the fact that all actin derivatives were labeled only up to 80 -90%. Finally, it cannot be excluded that the unprotected cysteine in the T␤4 analogs was partially oxidized during the cross-linking reaction. On the other hand, in all reactions classified as negative, the release of ArS Ϫ was never zero. We suppose that the small amounts of ArS Ϫ (Ͻ10%) detected in these experiments were released by unspecific reactions in which the small T␤4 reacted to some extent in a way similar to a low molecular weight thiol.
Three major reaction types could be distinguished on the basis of kinetics and the extent of cross-linking. Fast reactions with high extents of cross-linking (50 -75%) within a few minutes were taken as indicating close proximity of the two cysteines in the protein complex. According to this classification, one major contact was identified between the C terminus of actin and the N terminus of T␤4. In particular, there is evidence that the thiols of Cys 374 in actin and Cys 6 in T␤4 approach to within 9.2 Å. This finding confirms previous data that identified Cys 374 as a part of a short distance cross-link with T␤4 (13). Contact in this region must indeed be very close since it was even possible to form a zero-length cross-link between Cys 374 of actin and Cys 6 ⅐T␤4, although at a low rate. As a second major contact site the hexamotif of T␤4 (position [17][18][19][20][21][22] was identified as located near the actin-bound nucleotide, since the distance of Cys 17 ⅐T␤4 and the sulfur atom of ATP␥S could be bridged by a cross-linker of 9.2 Å in length. Lower, but still significant yields of cross-linking were found also between Cys 28 ⅐T␤4 and the modified ATP␥S, suggesting that the whole central part of T␤4 is in proximity to the ␥-phosphate of the nucleotide. Considering the different yields of these two crosslinking reactions, Cys 17 may be located closer to the nucleotide than Cys 28 , provided sterical influences can be excluded. Fig.  10 illustrates the position of the two major contact sites within a space-filling model of G-actin according to Kabsch et al. (30). Due to the mobility of the cross-linkers, only spheres of contact can be defined with dimensions determined by the length of the cross-linkers.
Different from the first type of reaction, a second one was distinguished from its distinctly slower kinetics. In this type, rather high yields of cross-linking were obtained but only after a prolonged reaction time (ϳ40% yield after 90 min). Examples were the formation of the zero-length cross-link mentioned above as well as the reaction of the Cys 374 of actin and Cys 17 ⅐T␤4. It appears that cross-linking in these cases depends on processes that proceed at a low rate, as for example mobility of the C terminus of actin (30). For this reason, cross-linking reactions proceeding slowly were not used in identifying sites of closest contact. Finally, more than 50% of all cross-linking experiments proceeded not only at very slow kinetics but on the average reached yields of only 9% of cross-linking (highest yield, 18%) even after prolonged incubation. This third type of reaction kinetics was taken as excluding a proximity of the two thiols concerned. However, the possibility still exists that a limited reactivity in some of the positions in the T␤4 analogs may rely on a partial occlusion of the corresponding cysteine residue due to binding to actin.
The N terminus of T␤4 has been described as tending to form an ␣-helix between the residues 5 and 16 (10). Given that this structure is stabilized in contact with actin as hypothesized by Sun et al. (3), the helix might stretch along subdomain 1 with contacts between the N terminus of T␤4 and the C terminus of actin on one hand, and the hexamotif of T␤4 and the actinbound nucleotide on the other. According to Kabsch et al. (30) the distance between Cys 374 and the ribose unit is around 28 Å, a distance that can easily be spanned by the length of the two cross-linkers plus the 11 amino acids between the residues 6 and 17 of the T␤4 molecule, even when they are in helical conformation. The yield of cross-linking with the actin-bound nucleotide decreased when the position of the cysteine residue in the T␤4 analog approached the C terminus. This correlation indicates that the highly flexible C terminus of T␤4 is most probably directed away from the actin-bound nucleotide. Since no thiol-specific cross-link formation was found in any experiment involving Cys 40 ⅐T␤4, this part of T␤4 appears to be located in a domain of actin that is different from subdomain 1 and remote from the nucleotide region.
Reactions of cysteine 10 of actin with all T␤4 analogs followed the third type of reaction kinetics. According to our classification, we believe that T␤4 is not in direct contact with that side of actin bearing Cys 10 , the latter known to be part of a ␤-sheet (30). Nevertheless, low yield cross-linking reactions with Cys 10 were measured by spectrophotometry, and confirmed by SDS-PAGE. The existence of these reactions between the cross-linker attached to Cys 10 and T␤4 may be understood on the basis of the maximal possible reaction range of the cross-linking reagent (Ն9.2Å) that reaches far beyond the thiol of Cys 10 . The reaction range of the cross-linker may be comparable with that of the first four amino acids of actin, which are believed to form a mobile structure (30) and have been reported to be involved in EDC cross-links with T␤4 (14,31).
Finally, we assayed whether the replacement of five hydrophobic amino acids in T␤4 had caused any functional deficits. For all T␤4 analogs, a clear correlation was found between the extents of cross-linking with the actin-bound nucleotide and the decrease in the inhibitory capacities on actin polymerization. The strongest reduction of inhibitory capacities was found for substitutions by cysteine in or near by the hexamotif. This observation is in line with the results of Vancompernolle et al. (9) who showed the hexamotif to be most important for binding and function. Interestingly, in the case of Cys 17 ⅐T␤4, the greatest extent of cross-linking was paralleled not only by the most strongly reduced inhibitory capacity, but also by a significant decrease of the retardation effect on the nucleotide exchange rate in comparison to native T␤4. FIG. 10. Schematic representation of the two major contact sites identified in actin for T␤4 as illustrated by two spheres fitted into the structure of G-actin in a space-filling model according to Kabsch et al. (30). The spheres are centered either at the ␥-phosphate of ATP (sphere A), or at C ␣ 372 (sphere B), and both have a radius of 10 Å representing the maximal possible reaction range for the cross-linkers of the type ArSS-(CH 2 ) 3 -SSAr. Sphere B was centered at C ␣ 372 as the last defined position in G-actin as determined by x-ray analysis. Cys 374 , to which the cross-linker was attached, can be assumed to be ca. 3Å apart from C ␣ 372 , given an ␣-helical conformation at the C terminus of actin. The numbers in the illustration denote the four subdomains of the actin molecule. This figure was prepared by the PLUTO program written by Sam Motherwell at the Chemical Laboratory, Cambridge, UK.