Tubulin sequence region beta 155-174 is involved in binding exchangeable guanosine triphosphate.

Assembly-competent microtubule protein was directly photoaffinity labeled with [alpha-32P]guanosine triphosphate by UV irradiation. The labeled tubulin was digested with trypsin. The radioactive fragments were isolated and sequenced, revealing beta-tubulin residues 155-174 to be the major labeled region. An antibody to a synthetic peptide comprising residues beta 154-165 inhibits GTP incorporation and tubulin polymerization.

Guanosine triphosphate is required for the assembly of microtubules (1). Two molecules of GTP are bound to the tubulin heterodimer (2,3). One binds to an N site, where it is not exchangeable and is not involved in assembly in vitro (4). Nonexchangeable GTP labeled with 32P in cells in uiuo remains unhydrolyzed after several cycles of assembly-disassembly in the presence of unlabeled GTP (5). At the other site, the exchangeable or E site, GTP is readily exchanged with added nucleotide and is hydrolyzed during incorporation into the microtubule (2, 4, 6).
However, tubulin can also polymerize in the presence of the nonhydrolyzable GTP analog guanyl-5'-yl-imidodiphosphate (6, 7). The resulting microtubules are more resistant toward depolymerizing agents. Furthermore, GTP hydrolysis lags behind polymerization (8-10). Thus, the energy of hydrolysis is not required for polymerization. Possibly, GTP binding causes a tubulin conformation favorable for polymerization, whereas hydrolysis may be a consequence of cooperative interactions of the dimer with its neighbors and favor subsequent disassembly (11,12).
Previously, the exchangeable site has been localized on the @-subunit with the photoaffinity analog 8-azido-GTP (13,14) and by UV cross-linking with unmodified GTP (15,16). Upon limited hydrolysis with chymotrypsin, the label is found on the amino-terminal fragment, comprising residues 1-281 (17, 18). Here we report that the p-tubulin sequence region 155-174 contributes to the binding site for exchangeable GTP.
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. $Recipient of a grant from the Studienstiftung des Deutschen Volkes.
8 To whom correspondence should be addressed.

RESULTS
Neither microtubule-associated proteins nor bound GTP was removed before covalent binding of [cx-~'P]GTP in order to minimize a possible denaturation of tubulin. This restricted the covalent incorporation of added [cx-~'P]GTP to 2% of the tubulin dimers. However, this reflects only the exchange of added radioactive GTP for already bound unlabeled nucleotide and not the total amount of nucleotide covalently incorporated into tubulin upon irradiation. The percentage of cross-linking is not known, but is probably much higher. To assess polymerizability of tubulin, one cycle of assemblydisassembly was carried out with the whole sample immediately before photoaffinity labeling, and with aliquots thereafter ( Fig. 1). Irradiated tubulin polymerizes more rapidly than nonirradiated tubulin, but we observed the same final extent of assembly under all conditions tested, i.e. after 60 min of irradiation of tubulin without additional GTP, irradiation in the presence of a 20-fold excess of GTP over tubulin, as well as without irradiation.
Immediately after photoaffinity labeling, tubulin was separated from microtubule-associated proteins (MAPS)' and unbound GTP by chromatography on DEAE-cellulose. The elution profile is shown in Fig. 2. While all of the MAPS and a small fraction of tubulin were found in the flow through, tubulin eluted at 0.4 M NaCl just behind the bulk of unbound GTP.
Gel electrophoresis and autoradiography of the labeled tubulin revealed that the label was bound exclusively to the psubunit as demonstrated previously in more detail (15). Excess unlabeled GTP inhibited incorporation of the radioactive compound. Therefore, only fragments of the @-chain were considered as possible carriers of the label. Tubulin was then digested with trypsin, and the peptides were separated by gel filtration on Sephadex G-50 (Fig. 3). Two radioactive peaks were obtained at the very end of the absorption profile. Peak T1 contained 46% and peak T2 47% of the initial radioactivity. Further separation by HPLC anion exchange chromatography yielded one single radioactive peak originating from peak T1 (Fig. 4a), while the radioactivity of peak T2 was split into four fractions (Fig. 4b). By sequence analysis of peak Tla, containing 44% of the radioactivity, we found residues p155-162 of the known amino acid sequence of porcine brain tubulin (26,27). In peak T2a (16% of the total radioactivity) only a-peptide-305-308 could be identified, while peak T2c (8% of the radioactivity) contained residues 163-174 of ptubulin. No peptides were identifiable in peaks T2b and T2d (5 and 18% of the radioactivity, respectively). In summary, these experiments indicate that sequence region 155-174 of @-tubulin forms part of the GTP-binding site.
The exact labeled residue was not identified. The radioactivity remained on the filter upon degradation while the phenylthiohydantoins were extracted and identified. The bond between GTP and peptide was apparently sensitive to the degrading chemicals.
T o confirm these findings by an immunological procedure, we generated an antiserum to a synthetic peptide comprising P-tubulin residues 154-165. The IgG fraction of the serum was isolated on protein A-Sepharose and was found to bind to native tubulin (data not shown). The IgG fractions of the preimmune serum and of an antiserum generated to a-tubulin residues 399-412 were used as controls throughout. The antibodies were adjusted to equal titers to native tubulin and incubated in various concentrations with tubulin prior to photoaffinity labeling with GTP. Antibody to p154-165 inhibited GTP incorporation into @-tubulin a t low concentrations (Fig. 5 ) , whereas the control antibodies were inhibitory only at higher concentrations. Neither antibody incorporated GTP, illustrating the specificity of the photoaffinity labeling procedure.
Only the antibody to p154-165 and none of the control antibodies impaired tubulin polymerization (Fig. 6). This confirms the results obtained by sequencing photolabeled peptides.

DISCUSSION
From our studies of direct photoaffinity labeling of tubulin with [CY-~~PIGTP, we conclude that the @-tubulin sequence region 155-174 directly interacts with GTP and forms part of the exchangeable GTP-binding site. An antibody generated to a synthetic peptide corresponding to a part of this region specifically reduces incorporation of the photolabel and inhibits tubulin polymerization in vitro. This is in agreement with most of the previous less specific experimental evidence (13)(14)(15)(16)(17)28). Only in chemical cross-linking studies with GTP analogs, carrying highly reactive groups in the ribose moiety, were a-and @-tubulin found to be labeled to an approximately equal extent (29).
Two caveats should be kept in mind when considering and comparing photoaffinity labeled sequences. 1) Usually several sequence regions of a polypeptide are involved in the binding of a nucleotide (30-32). It is, therefore, not surprising that a different labeling method quite correctly yields another contact region (33). 2) Not all amino acid side chains are activated by UV irradiation to form reactive radicals. Moreover, the radicals may not be as short lived as anticipated, and as a consequence the final covalent bond may not necessarily mark the immediately contacting residue but rather a particularly reactive one nearby.
We could not identify the exact labeled residue within a sequence of six, but it is tempting to speculate that the bond between GTP and tubulin is formed with tyrosine p-159. Tyrosine, like other aryl compounds, can form long lived radicals upon UV irradiation (34). Moreover, in studies of Manser and Bayley (35) removal of E site nucleotide caused specific changes in the near-UV circular dichroism of tubulin. These were attributed to changes in protein conformation involving the interaction between tyrosine residues and the guanine moiety.
Predictions of secondary structure by the methods of Chou and Fasman (36) and Garnier et al. (37) indicate that this tyrosine is the last residue of a short helix and the first of a reverse turn or loop. One should expect to see the essential residues of the binding site conserved in all species. However, c a -,/"" " " " " " " _ of the sequences available to date (apart from pig brain tubulin, they were all established on the nucleotide level), two appear to have the tyrosine exchanged for Phe (human 5p) (38) and Leu (yeast) (39), respectively. On the basis of conservation, the important invariant residues seem to be Glu-157, Pro-160, Asp-161, and Arg-162. They may be in close proximity in the predicted helix-turn conformation.
Interestingly, there is another invariant region, HSLGGGTGSG at residues 137-146, ending only 13 residues upstream of Tyr-159. On the basis of sequence similarities with the Rossmann fold of dinucleotide-binding proteins (40), we have predicted it to be a triphosphate binding loop (27). Recent data, however, indicate that the function of the typical glycine loop resides rather in its high flexibility than in direct interaction with the phosphate groups, at least in two wellexamined cases (30, 31), thereby controlling the access of the nucleotide to its binding site.
Kim et al. (33) have used 8-azido-GTP as a photoaffinity label to elucidate the sequence regions in P-tubulin involved in binding the exchangeable GTP and have found peptide-63-77 to be labeled by the photogenerated nitrene at position 8 of the purine base. From the combined results of the two investigations, a more complete image of the binding site emerges: residues in regions p-63-77 are binding to the purine moiety, whereas side chains in the region 155-174 may contact the ribose and a flexible loop in 137-146 may regulate the access to the binding site.