Cross-linking site in Azotobacter vinelandii complex.

The Fe-protein and the MoFe-protein of the Azotobacter vinelandii nitrogenase complex can be chemically cross-linked by 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (Willing, A., Georgiadis, M.M., Rees, D. C., and Howard, J. B. (1989) J. Biol. Chem. 264, 8499-8503). In this reaction, one of the identical subunits of the Fe-protein dimer is linked by an isopeptide bond to each beta-subunit of the MoFe-protein tetramer. The reaction has been found to be highly specific with greater than 85% of amino acid residues Glu-112 (Fe-protein) and Lys-399 (MoFe-protein) cross-linked to each other. Although Glu-112 is located in a highly conserved amino acid sequence, it is found in only half of the known Fe-protein sequences. Likewise, Lys-399 is not a conserved residue in the MoFe-protein. Glu-112 appears to be part of an anionic cluster of nine carboxylic acids which is located between the proposed thiol ligands for the Fe:S center. In contrast, the basic residue cluster which includes Lys-399 has been found in only in the Azotobacter MoFe-protein. Thus, this crosslinking reaction either is unique to Azotobacter nitrogenase or must involve other residues in the MoFe-protein of other species. Because Lys-399 and Glu-112 form a specific cross-link, it is probable that they are part of the interaction site leading to productive complex formation. This information should be useful for the model building of the complex from the crystallographic structures of the individual components.

In this reaction, one of the identical subunits of the Fe-protein dimer is linked by an isopeptide bond to each ,&subunit of the MoFe-protein tetramer. The reaction has been found to be highly specific with ~85% of amino acid residues Glu-112 (Fe-protein) and and has Thr as its amino-terminal residue (7,8); the Avl P-subunit is the faster moving subunit and has Ser as its aminoterminal residue (7,8). Numbering of the amino acid residues is based on the protein sequences which are one less than the DNA-encoded sequence (8). For the Azotobacter proteins, the amino-terminal methionines have been removed in uivo by processing (7,9). multiple times during catalysis (3,4). The formation of the complex is highly specific. For example, the Fe-protein is the obligate electron donor for nitrogenase turnover; other low potential electron donors cannot replace Fe-protein. Likewise, the Fe-proteins isolated from various microbial species have different rates of nitrogenase turnover when used with a common MoFe-protein (5,6). Thus, the interaction sites must provide the correct orientation between the components for efficient electron transfer and coupled ATP hydrolysis.
Recently, we reported that the two components from Azotobacter uinelandii could be covalently cross-linked by the water-soluble carbodiimide EDC' (10). The cross-linking reaction was dependent upon several factors known to affect the enzyme activity levels and, therefore, was considered to occur during the formation of the active complex. The reaction appeared to be highly specific with only one of the two identical Av2 subunits cross-linked to each Avl P-subunit (Avl is an a& tetramer). One important conclusion from the proposed stoichiometry of the cross-linking reaction was that once Av2 was bound to Avl, the two Av2 subunits had different points of contact with the P-subunit.
That is, the Av2 binding site is asymmetric with respect to the two subunits. Because EDC cross-linking involves carboxyl and amino groups, and because the enzyme activity is significantly inhibited by moderate salt concentrations, ionic interactions are likely to be important elements of the protein-protein binding site (10,11). In this communication, the specific amino acid residues involved in the EDC cross-linking reaction are identified.

MATERIALS AND METHODS
Nitrogenase components from A. uinelandii, Avl and Av2, were isolated and characterized as described previously (12,13 to elute the peptides. Reverse phase chromatography was performed on either C-4 or C-8 resin using a buffer of 0.1% trifluoroacetic acid and a gradient of acetonitrile.

AND DISCUSSION
The EDC cross-linking of the nitrogenase complex is most likely between one of the two identical Av2 subunits and each of the Avl p-subunits (9). However, the experimental approach used in our earlier report could not exclude the possibilities that both Av2 subunits were cross-linked or that, in addition, the cross-linking involved the a-subunit which was blocked at its amino terminus. To resolve these questions, mixed peptide maps were prepared which included 2-3Hcarboxylmethylated complex and either 2-'4C-carboxymethylated Av2, a-Avl, or P-Avl subunits. Because the tryptic digests of the carboxymethylated subunits of Avl and Av2 are well characterized, the cysteinyl peptides derived from the individual components can be identified readily in the peptide map of the complex (7,12). The tryptic peptide maps for those mixtures are shown in Fig. 1.
Three important conclusions were obtained from inspection of the maps: First, only the cysteine-containing tryptic peptides from Av2 and P-Avl were found in the tryptic map of the cross-linked complex. That is, the complex was composed only of Av2 and /3-Avl. This confirms our earlier results. Second, from the specific radioactivity of the individual peptides in the map, the ratio of the two subunits in the complex was calculated to be 1:l. Thus, as previously suggested, only one of the two identical Av2 subunits was cross-linked to the P-Avl subunits. Third, except for one peptide in each subunit, all of the cysteine-containing tryptic peptides from P-Avl and Av2 were accounted for in the map of the complex. The missing peptide in the P-subunit, designated a-6 (residues 381-400 of the /3 subunit), contains 2 cysteines (7,8). Because the peptide terminates with the sequence, -Lys-Arg, there are two trypsin cleavage sites, both of which were observed in the map of the subunit (indicated by arrows in Fig. 1C) but were missing in the map of the complex. Likewise, five of six cysteine-containing tryptic peptides from Av2 were present in the complex. The only cysteinyl peptide missing was T-11 (residues 101-140 of Av2, indicated by the arrow in Fig. lA).
However, two new radioactive peptides were identified in the maps. The peptides (a major and a minor peak, ~85 and (15% of the radioactivity in this region of the map, respectively) are identified as A and B in Fig. 1C and together they contained the amount of radioactivity expected for the missing a-6 and T-11 peptides (2X1/3.00, based upon the specific radioactivity for 1 cysteine). Thus, we conclude that Avl and Av2 were cross-linked through the regions of peptides T-11 and a-6 and that the resulting peptides were eluted later in the solvent gradient.
To confirm the identity of the peptides and to determine which specific residues composed the cross-linking site(s), a preparative scale separation of the peptides was performed. For this purpose, only the cross-linked complex was used (the marker peptides from purified subunits were omitted). Peptides comparable to those indicated by bars A and B in Fig.  1C were pooled. The combined yield of the major and minor radioactive peaks was ~60% based upon the assumption of 3 cysteines in the cross-linked peptides. A single radioactive peptide (55% yield) from pool B was separated from small quantities of other, nonradioactive peptides by reverse phase chromatography on a C-4 resin. The nonradioactive peptides were identified by amino acid sequencing as predominantly tryptic peptides T-8 (residues 53-77) and T-27 (residues 251-284) of Av2 and residues 481-503 of the /3-Avl subunit (data not shown) (8,12). These peptides are known to elute in this For each map, the complex, labeled by reductions and carboxymethylations with ["Hliodoacetic acid, was mixed with either Av2, Ser-Avl subunit, or Thr-Avl subunit, labeled with ['4C]iodoacetic acid. The mixtures were digested with trypsin and the resulting peptides were separated as described under "Materials and Methods." 1.0.ml fractions were collected. The linear salt gradient was begun at fraction 15. A, complex and Av2 subunit. B, complex and Avl a-subunit (Thr-Avl). C, complex and Avl P-subunit (Ser-Avl). region of the ion exchange chromatogram.
Because a single, unique sequence was found for these peptides, they did not appear to contain the sites of cross-linking.
The amino acid composition for the major radioactive peptide, B, is given in Table I and the results of Edman degradation are given in Table II. The composition was that expected for a 1:l mixture of T-11 and a-6. For the first 20 cycles, except for cycles 12 and 19, each cycle of Edman degradation contained 2 residues of nearly equal yield. These residues exactly matched the sequences for T-11 and a-6. For cycle 12, the expected glutamic acid in T-11 was absent; likewise, the expected lysine at cycle 19 for a-6 was absent. It should be noted that none of the other acidic residues in either peptide was below the expected yield given the overall sequencing repetitive yield. The absence of the penultimate lysine is significant because the carboxyl-terminal arginine was found. In addition, there is only 1 lysine in the composition of the putatively cross-linked peptide. Therefore, the missing lysine at cycle 19 must be one of the two moieties of the isopeptide bond. Beyond cycle 20 only the sequence of T-11 was observed because of the difference in length between a-6 and T-11. The positions of the 3 carboxymethylcysteines were confirmed by their radioactivity.
The sequence and composition for the minor radioactive peptide, A, was less clear due to the substantially higher concentration of the non-cysteinyl peptides and smaller quantities of the radioactive peptide for purification. To find the cross-linking site in this material, a portion of pool A was digested with chymotrypsin and the resulting peptides were separated by reverse phase chromatography on C-8 resin. One of the isolated peptides had the amino-terminal sequences expected for a cross-linked peptide between T-11 and a-6 (data not shown). Although this peptide is consistent with peptide B, the limited amount of material available was insufficient to determine unambiguously which glutamyl residue between 110 and 112 was in the isopeptide bond. The elution of peptide A at lower ionic strength than B suggests that A has a more positive charge than B. A tentative explanation is that peptide B contains an extension at the carboxyl terminus due to incomplete trypsin hydrolysis at the arginine next to   (14) ASP (371, Gly (10) Asp (41), Asn (7) Leu (29), -' ASP W), Arg (4) Phe (19) Val (14) Phe (14) Tyr (12) Asp (12) Val (8) Leu (6) Gly (6) ASP (7) Val (5)  the cross-linking site; additional trypsin cleavage sites are located just beyond the terminal arginine of a-6 (see Fig. 2).
The results are consistent with a primary site (X35%) of cross-linking between Avl and Av2 at residues Glu-112 (Av2) and Lys-399 (P-Avl). This is shown in Fig. 2. No evidence was found for alternate cross-linking involving the other 8 glutamyl or aspartyl residues in T-11 (residues 101-140 of Av2). (Cross-linking at <lo% would not be detected from the yield of the Edman degradation.) Although we cannot exclude sites of cross-linking located in non-cysteinyl tryptic peptides, the cross-linking between Glu-112 and Lys-399 is sufficient to account for full cross-linking of the two proteins. Namely, peptides T-11 and a-6 are quantitatively cross-linked (see peptide maps in Fig. 1). Indeed, it was fortuitous that both the amino and carboxyl groups involved in the isopeptide are located in cysteinyl tryptic peptides which allowed for ready identification and quantification of the modified and unmodified peptides.
The most unexpected aspect of our results is that neither