Proc. Natl. Acad. Sci. USA Vol. 84, pp. 7066-7069, October 1987 Biochemistry Site-directed mutagenesis of the nitrogenase MoFe protein of Azotobacter vinelandii (iron-molybdenum cofactor/nitrogen fixation/niJD gene/protein engineering) KEVIN E. BRIGLE*, ROBERT A. SETTERQUIST*, DENNIS R. DEAN*, JOHN S. CANTWELLt, MARY C. WEISSt, AND WILLIAM E. NEWTONtt§ *Department of Anaerobic Microbiology, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061; tWestern Regional Research Center, U.S. Department of Agriculture-Agricultural Research Service, Albany, CA 94710; and tDepartment of Biochemistry and Biophysics, University of California, Davis, CA 95616 Communicated by Martin D. Kamen, July 8, 1987

ABSTRACT A strategy has been formulated for the site- Mossbauer and electron paramagnetic resonance spectros- directed mutagenesis of the Azotobacter vinelandii nifJK genes. copies (2-6), as well as biochemical and genetic studies (10), These genes encode the a and .3 subunits of the MoFe protein have been used to demonstrate that the metal centers located of nitrogenase, respectively. Six mutant strains, which produce within the MoFe protein are likely to participate in the MoFe proteins altered in their a subunit by known single amino binding and catalytic reduction of dinitrogen. However, little acid substitutions, have been- produced. Three of these direct information is available concerning the structures, transversion mutations involve cysteine-to-serine changes (at redox properties, and functions of the individual metal residues 154, 183, and 275), two involve glutamine-to-glutamic centers within the MoFe protein. Furthermore, there is no acid changes (at residues 151 and 191), and one involves an information regarding the spatial distribution of the metal aspartic acid-to-glutamic acid change (at residue 161). All three centers among or between the four subunits of the MoFe possible phenotypic responses are observed within this group- protein. One approach toward addressing these issues is to i.e., normal, slow, and no growth in the absence of a fixed- alter specifically the polypeptide environments of the indi- nitrogen source. Two-dimensional gel electrophoresis indicates vidual clusters and to determine the spectroscopic, redox, that all mutants accumulate normal levels of the subunits of and catalytic consequences that result from such alterations. both nitrogengse component proteins. Whole-cell and crude- For example, specific amino acid substitutions at key metal- extract acetylene-reduction activities indicate substantial levels center ligation points within the MoFe protein should pro- of Fe protein activity in all strains. In contrast, MoFe protein duce altered proteins with strengthened, weakened, or elim- activities do not parallel the diazotrophic growth capability for inated metallocenter interactions. Observations of these all strains. Two strains appear to exhibit altered substrate effects could indicate appropriate cluster-binding amino acid discrimination. Such analyses should aid in the identification of residues and also provide information on the mechanistic metallocluster-binding sites and subunit-subunit interaction roles of the individual clusters. For this purpose, we isolated domains of the MoFe protein and also provide insight into the the nifHDK gene cluster from Azotobacter vinelandii and mechanistic roles of the various prosthetic groups in catalysis. determined its complete nucleotide sequence (11). Here, we describe the formulation of a strategy for the site-directed Biological nitrogen fixation is catalyzed by nitrogenase (EC mutagenesis of the genes that encode the MoFe protein 1.18.6.1), a complex metalloenzyme composed of two sepa- subunits and describe the isolation and characterization ofsix rately purifiable component proteins called the Fe protein mutant A. vinelandii strains that produce altered MoFe and the MoFe protein. The Fe protein is encoded by the nifH a-subunit proteins with known amino acid substitutions. gene and acts as a specific ATP-binding, one-electron reductant of the MoFe protein, which contains the site for substrate binding and reduction (1-3). The MoFe protein is an MATERIALS AND METHODS a2f32 protein of approximately 220,000 daltons and it contains 2 Mo atoms and approximately 32 Fe and 32 S2- atoms per Oligonucleotide-Directed Mutagenesis. Oligonucleotides molecule. The a and f3 subunits are encoded by the nifD and used for mutagenesis were synthesized on a Beckman System nifK genes, respectively. Mossbauer spectroscopy (4-6) and 1 Plus DNA synthesizer. After deprotection, the oligonucle- quantitative extrusion studies indicate that the metal atoms otides were purified by HPLC (12). Oligonucleotide-directed contained within the MoFe protein are organized into at least mutagenesis was performed by the method of Zoller and six metal-containing prosthetic groups of at least two distinct Smith (13). Template DNA used for mutagenesis was a types. Approximately 16 Fe atoms can be extruded from each 1.4-kilobase A. vinelandii niJ-specific EcoRI fragment (11) MoFe protein molecule in the form of [4Fe-4S] clusters by cloned into the EcoRI site of bacteriophage vector Ml3mpl8 treatment of the native protein with thiols in a denaturing (14). Single-lane dideoxy sequencing (15) was used to screen organic solvent (7, 8). All or most of the remaining Fe and for the desired mutations. Further DNA sequence analysis both Mo atoms constitute the two identical iron-molybde- using all four reactions confirmed the mutations and estab- num cofactors (FeMoco), and these can be isolated from the lished that there were no other mutations within 400 nucle- native MoFe protein by anaerobic acid/base treatment otides surrounding the planned mutation. For each mutant (which destroys the Fe-S clusters) followed by extraction construction, replicative-form DNA carrying a known single- with N-methylformamide (9). The mutual exclusivity ofthese base-pair transversion mutation within the nifD coding region two cluster-releasing processes indicates different binding was prepared and used for transformation of A. vinelandii modes for the two cluster types. cells. Such DNA, which carries a single transversion muta-

The publication costs of this article were defrayed in part by page charge Abbreviation: FeMoco, iron-molybdenum cofactor. payment. This article must therefore be hereby marked "advertisement" §To whom reprint requests should be addressed at: Western Regional in accordance with 18 U.S.C. §1734 solely to indicate this fact. Research Center, USDA-ARS, Albany, CA 94710.

7066 Downloaded by guest on September 26, 2021 Biochemistry: Brigle et al. Proc. Natl. Acad. Sci. USA 84 (1987) 7067 tion within the niflJ coding sequence, will hereafter be called Burk medium was used to transfer cells back to the fermentor mutation-vector DNA. and incubated as above for 2.5 hr. Derepressed cells were Isolation of Mutant Strains. Transversion mutations carried harvested as above except that the cells were washed with on mutation-vector DNA were transferred to the A. vine- chilled 0.05 M Tris (pH 8.0), then centrifuged at 10,000 x g landii chromosome in either one or two steps (Fig. 1), for 10 min and stored at -80'C until needed. depending on the resultant Nif phenotype of the mutant Two-Dimensional Gel Electrophoresis. Preparation of ex- strain. Mutations resulting in a single amino acid substitution tracts for two-dimensional gel electrophoresis was performed within the nifD-encoded polypeptide may exhibit one of three as described by Bishop et al. (18). Two-dimensional gel Nif phenotypes. Namely, such mutant strains will grow analysis was performed as described by O'Farrell (19). normally, at a reduced rate, or not at all in the absence of a Enzyme Assay. Frozen cells (=13 g) were suspended in 39 fixed-nitrogen source. Mutations that cause slow growth or ml of cold 0.05 M Tris (pH 8.0) containing 1 mM Na2S204 and that have no effect on diazotrophic growth can be introduced passed through a chilled, argon-flushed French pressure cell into the A. vinelandii genome by a marker-rescue procedure at 12,000 psi (82.7 MPa), followed by centrifugation at 17,000 where the mutation-vector DNA is used to transform a strain x g for 20 min at 40C. The extracts were then degassed under deleted for nifD to prototrophy (see Fig. 1). Mutations that argon and the MoFe and Fe protein activities were assayed result in a strictly Nif- phenotype can be introduced into the by acetylene reduction (20), with and without saturating A. vinelandii chromosome by the simultaneous transforma- amounts of the purified complementary protein. Specific tion of wild-type cells with mutation-vector DNA and puri- activities of these purified MoFe and Fe proteins were 2020 fied A. vinelandii chromosomal DNA that carries an and 1700 nmol of acetylene reduced per min per mg of uncharacterized rifampicin-resistance marker. Rifrtransform- protein, respectively. Protein concentrations were deter- ants are recovered by selection on rifampicin-containing mined by the biuret method (21). medium, and the coincident transfer (congression) ofthe Nif marker is identified by scoring transformants in the absence RESULTS AND DISCUSSION or presence of a fixed-nitrogen source (see ref. 16 for a detailed description of this procedure). Rationale for Amino Acid Replacements. Several indirect Cell Growth and Nitrogenase Derepression. Wild-type and approaches have been used in attempts to identify structur- mutant strains of A. vinelandii were cultured on a modified ally important regions within the MoFe protein. These Burk medium (17) that was supplemented with filter-steril- include (i) comparison of MoFe protein sequences to other ized urea to a final concentration of 10 mM when a fixed- metal-center-containing proteins of known structure, in par- nitrogen source was required. For derepression of nitrogen- ticular ferredoxins (11); (ii) comparison of MoFe protein ase synthesis, 11 liters of urea-supplemented Burk medium sequences from widely diverse diazotrophic species (see for was inoculated with a 500-ml culture (=200 Klett units, no. example refs. 11 and 22); (iii) comparison ofthe MoFe protein 54 filter) in a New Brunswick Microgen SF116 fermentor of a and p subunits (22); and (iv) comparison of MoFe protein 12-liter working capacity. All cultures were stirred at 300 subunit sequences to sequences of polypeptides that are rpm at 30TC with an aeration rate of 12 liters/min. When the required for metal-center assembly [for example, the nilE and culture density reached =110 Klett units, cells were concen- nifN gene products, which are required for FeMoco biogen- trated to 1 liter with a Millipore tangential-flow ultrafiltration esis (20)]. apparatus. Two liters of sterile nitrogen-free Burk medium The above comparisons have shown that there are no fermentor, and the culture was typical ferredoxin-like sequences located within the A. were then added to the vinelandii MoFe protein, an observation that is consistent concentrated again. A similar second wash was taken to with the unusual redox and spectroscopic properties of the 500-ml final volume. Eleven liters of sterile, nitrogen-free MoFe protein Fe-S centers when they are compared to ferredoxin Fe-S clusters. Interspecies comparisons of MoFe A H AD K Ni protein subunits reveal homologies that are greatest within chromosome D K Nif the amino-terminal half of homologous subunit polypeptides, and these homologies are largely centered around cysteine M utatZIioD4-mutation vector residues. There are five cysteine residues that can be con- sidered conserved among all known a-subunit sequences chromosome Nif (residues 62, 88, 154, 183, and 275, using the A. vinelandii sequence in ref. 11) and three conserved cysteine residues among all known P-subunit sequences (residues 70, 95, and 153, using the A. vinetandii sequence in ref. 11). Significant B sequence homology is also found surrounding a-subunit chromosome H 0 K Nif + cysteine residues 62, 88, and 154 when they are respectively compared to P-subunit cysteine residues 70, 95, and 153. 4mutation vector These latter homologies have led to the suggestion that there are structural prosthetic group-binding features common to chromosome Nif both subunits of the MoFe protein (22). Finally, the products of the FeMoco-specific biosynthetic genes, nifE and nifN, FIG. 1. Strategy for site-directed mutagenesis. Introduction of a share structural homology when compared to the MoFe single point mutation, indicated by a dot on the mutation vector, into protein a and p subunits, respectively (20). Importantly, the A. vinelandii nifD coding sequence leads to either a Nif+ or a NiU these homologies are also centered around those cysteine phenotype in the resultant mutant strain. (A) Those mutations that do residues that exhibit interspecies and, in some cases, not cause a Nif- phenotype are introduced into the A. vinelandii intersubunit sequence conservation. genome by transformation of strain DJ100 (16), which carries a Insights regarding potential metal-center ligands within the defined deletion within the nifD gene (indicated by cross-hatching), the extrusion to prototrophy. (B) Those mutations that result in a Nif- phenotype MoFe protein can also be gained by considering are introduced into the A. vinelandii genome by transforming the requirements of the metal centers. The Fe-S clusters, which wild-type strain to the Nif- character. Reciprocal recombination are extruded by thiol treatment, are likely to exhibit cysteine- events are indicated by crosses between the chromosome segments thiol-type ligation within the MoFe protein. However, the and the mutation vector. unusual redox and spectroscopic properties of the Fe-S Downloaded by guest on September 26, 2021 7068 Biochemistry: Brigle et al. Proc. Natl. Acad. Sci. USA 84 (1987) centers within the MoFe protein raise the possibility that there are also other ligand modes. Thiol reactivity experi- ments using purified FeMoco have also indicated the poten- tial for a single cysteine-thiol ligand for each FeMoco species (23). In addition, the requirement for N-methyformamide or formamide for the extraction of FeMoco indicates a potential role for amide groups in binding FeMoco to the MoFe protein. Construction and Expression of Altered MoFe Proteins. In Klebsiella pneumoniae, the nif genes are clustered and are organized into at least seven transcriptional units (24). The promoter regions for these nifgene clusters share a charac- teristic structure and their activation requires the presence of a positive regulatory element, the product of the nifA gene (24). It has been shown that elevation of the copy number of the nilf gene promoter unbalances nifgene expression (25), presumably by sequestering the available activator mole- cules. Recent studies (20, 26) indicate that the genetic organization, as well as the mechanisms for regulating nif gene expression in A. vinelandii, is similar to that determined for K. pneumoniae. Thus, to avoid potential deleterious effects that could unbalance the synthesis of nif-specific components, it is important that altered nif genes are rein- corporated into the chromosome in single copy rather than FIG. 2. Diazotrophic growth characteristics of the mutant residing in the form of multicopy plasmids. Such gene- strains. Cells were streaked onto Burk's nitrogen-free medium and replacement strategies are described in Materials and Meth- incubated at 30'C for 5 days. A, wild type; B, DJ45; C, DJ55; D, ods and are depicted in Fig. 1. Strains constructed in this way DJ56; E, DJ62; F, DJ63; G, DJ64. Note growth of DJ55; this growth are particularly useful, since all of the ancillary nif-specific was not evident after 2 days, although good growth was recognized genes and their products are unaffected by this procedure. after 2 days for the other Nif+ strains. Consequently, any alterations in the properties of the MoFe proteins prepared from the various mutant strains can be protein a-subunit polypeptide as judged by two-dimensional attributed to the introduced mutation rather than as arising gel. electrophoresis (Fig. 3). Two-dimensional-gel and/or from indirect effects on processing or metallocenter-assem- immunological analysis of crude extracts prepared from bly functions. nitrogenase derepressed cultures was also used to demon- Characterization ofMutants. The site-directed mutagenesis strate that each of the mutant strains accumulates normal strategy described above was used to isolate six mutant amounts of Fe protein and MoFe protein subunits (data not strains, each of which contains a known transversion muta- shown). Thus, active MoFe protein is not required for the full tion resulting in an altered MoFe protein a subunit containing expression ofnitrogenase components in A. vinelandii. These a known single amino acid substitution (Table 1). All three of results further indicate that inactive MoFe protein subunits the anticipated Nif phenotypes are represented in this group. Namely, mutant strains were isolated that are altered within the MoFe protein a subunit and that individually exhibit a A w Nif phenotype (DJ45, DJ64, and DJ56), a Nif+ phenotype D ! # (DJ62 and DJ63), or slow growth in the absence of a K X

fixed-nitrogen source (DJ55). The diazotrophic growth prop- war ."W MO- erties of these mutants are shown in Fig. 2. Strain DJ62 was . _ useful because it exhibits normal diazotrophic growth yet carries a mutation that should lead to a charge change within ItH the MoFe protein a subunit (Glu-191 substituted for Gln-191, Table 1). This feature permitted the proof that strains that carry phenotypically silent mutations are readily recovered by our mutagenesis scheme. This proof was accomplished by B . .:' . demonstrating the appropriate charge shift in the MoFe

Table 1. Mutant strains isolated and characterized in this study .

niJD codon Amino acid w-w.. . Strain altered* Codon changet substitutions -, . H DJ45 154 TGC-*TCC Cys--Ser . DJ55 183 TGC--TCC Cys-*Ser DJ56 275 TGC-)TCC Cys-*Ser DJ62 151 CAG--GAG Gln-*Glu FIG. 3. Two-dimensional gel electrophoresis of protein extracts DJ63 161 GAC-*GAG Asp-+Glu of wild-type strain and mutant strain DJ62. For each gel the first DJ64 191 CAG--GAG Gln- Glu dimension is the isoelectric-focusing dimension (left to right, basic to acidic) and the second dimension is the size dimension (top to *Indicates the position ofthe altered codon within nifD in the mutant bottom, large to small). Positions of the Fe protein (nif'H product), strain. The initiation ATG codon is number 1. the MoFe protein a subunit (nifD product), and the MoFe protein ,3 tIndicates the specific nucleotide change (wild type--mutant) within subunit (nifK product) have been determined previously (16) and are the altered codon. indicated. (A) Wild type. (B) DJ62. Notice a shift in the mobility of tIndicates the amino acid substitution resulting from the transversion the MoFe a subunit toward the acidic region in strain DJ62 when mutation. compared to wild type. Entire gels are not shown. Downloaded by guest on September 26, 2021 Biochemistry: Brigle et al. Proc. Natl. Acad. Sci. USA 84 (1987) 7069 Table 2. Component protein activities in mutant and that normal levels of both MoFe protein subunits accumulate wild-type strains in strains altered within the a subunit, indicating that inactive Acetylene-reduction activity subunit polypeptides are not rapidly degraded in A. vinelandii. Purification and exhaustive biochemical charac- Whole cell, Crude extract, spec. act.* terization of MoFe protein purified from mutant strains Strain % wild type Fe proteinf MoFe proteint described here should help to identify the role each mutation Wild type 100 43.3 53.7 has in altering enzyme activity. Specifically, we believe that DJ45 0 14.7 0.1 such analyses will provide insights into the roles that each DJ55 4 23.4 3.1 cluster type plays in the catalytic activity of this extraordi- DJ56 0 20.9 0.2 narily complex metalloenzyme. Finally, studies using the DJ62 70 33.6 38.2 mutagenesis strategy described here could identify catalyti- DJ63 64 18.1 41.7 cally important residues within the MoFe protein 3 subunit and the Fe protein, and in those gene products involved in DJ64 12 27.3 10.0 FeMoco biogenesis. All values represent the average of at least three determinations. *Specific activity: nmol of ethylene formed per min per mg of crude This work was supported by the National Science Foundation extract protein. (Grant DMB 85-10029). tDetermined in the presence of saturation levels of purified A. vinelandii MoFe protein. 1. Hageman, R. V. & Burris, R. H. (1978) Proc. Natl. Acad. Sci. tDetermined in the presence of saturation levels of purified A. USA 75, 2699-2702. Oinelandii Fe protein. 2. Smith, B. E., Lowe, D. J. & Bray, R. C. (1973) Biochem. J. 135, 331-341. 3. Orme-Johnson, W. H., Hamilton, W. D., Jones, T. L., Tso, are not rapidly degraded in A. vinelandii as was suggested for M.-Y., Burris, R. H., Shah, V. K. & Brill, W. J. (1972) Proc. K. pneumoniae (27). Natl. Acad. Sci. USA 69, 3142-3145. Whole-cell and crude-extract activity measurements (Ta- 4. Smith, B. E. & Lang, G. (1974) Biochem. J. 137, 169-180. ble 2) show that alteration of the interspecifically conserved 5. Zimmerman, R., Munck, E., Brill, W. J., Shah, V. K., Henzl, a-subunit cysteine residues 154 and 275 results in a complete M. T., Rawlings, J. & Orme-Johnson, W. H. (1978) Biochim. loss of both MoFe protein activity and the capacity for Biophys. Acta 537, 185-207. diazotrophic growth. In contrast, alteration of a-subunit 6. Dunham, W. H., Hagen, W. R., Braaksma, A., Grande, H. J. & Haaker, H. (1985) Eur. J. Biochem. 146, 497-501. Cys-183 results in substantially lowered MoFe protein activ- 7. Kurtz, D. M., McMillan, R. S., Burgess, B. K., Mortensen, ity, yet the mutant strain retains the capacity for diazotrophic L. E. & Holm, R. H. (1979) Proc. Natl. Acad. Sci. USA 76, growth. These comparisons are in line with the notion that 4986-4989. conserved cysteine residues 154 and 275 play an essential 8. Watt, G. D., Burns, A. & Tennent, D. L. (1981) Biochemistry structural or functional role in the MoFe protein, whereas 20, 7272-7277. conserved cysteine residue 183 does not. These results do 9. Shah, V. K. & Brill, W. J. (1977) Proc. Natl. Acad. Sci. USA not, however, prove a proposed role for cysteine residues 154 74, 3249-3253. and 275 as metallocluster-binding ligands within the MoFe 10. Hawkes, T. R., McLean, P. A. & Smith, B. E. (1984) Bio- protein. Rather, the data provide a basis for the further chem. J. 217, 317-321. 11. Brigle, K. E., Newton, W. E. & Dean, D. R. (1985) Gene 37, biophysical analyses of MoFe protein prepared from these 37-44. mutant strains, as well as provide some insight regarding 12. Itakura, K., Rossi, J. J. & Wallace, R. B. (1984) Annu. Rev. other potential targets for site-directed mutagenesis. Biochem. 53, 323-356. Comparison of mutant strains DJ55 and DJ64 is of partic- 13. Zoller, M. J. & Smith, M. (1983) Methods Enzymol. 100B, ular interest. Although strain DJ64 is incapable ofdiazotroph- 468-500. ic growth (Fig. 2), it retains substantial levels ofMoFe protein 14. Messing, J. & Vieira, J. (1982) Gene 19, 269-276. activity as judged by its acetylene-reduction activity (Table 15. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl. Acad. Sci. USA 74, 5463-5467. 2). In contrast, DJ55 exhibits a much lower MoFe protein 16. Robinson, A. C., Burgess, B. K. & Dean, D. R. (1986) J. acetylene-reduction activity (Table 2), yet it is still capable of Bacteriol. 166, 180-186. diazotrophic growth (Fig. 2). The difference in substrate 17. Strandberg, G. W. & Wilson, P. W. (1968) Can. J. Microbiol. utilization shown by these two strains must reside in their 14, 25-31. ability to recognize or reduce different substrates. Conse- 18. Bishop, P. E., Jarlenski, D. M. L. & Hetherington, D. R. quently, the further characterization of these strains and the (1980) Proc. Natl. Acad. Sci. USA 77, 7342-7346. 19. O'Farrell, P. H. (1975) J. Biol. Chem. 250, 4007-4021. isolation of other mutants with similar phenotypes should 20. Brigle, K. E., Weiss, M. C., Newton, W. E. & Dean, D. R. prove useful in elucidating the substrate-discrimination and (1987) J. Bacteriol. 169, 1547-1553. -reduction properties ofnitrogenase. In this regard, analytical 21. Gornall, A. G., Bardawall, C. J., & David, M. M. (1948) J. measurements of the Fe and Mo content and spectroscopic Biol. Chem. 177, 751-766. characterization ofMoFe protein prepared from these mutant 22. Thony, B., Kaluza, K. & Hennecke, H. (1985) Mol. Gen. organisms should be particularly useful. Genet. 198, 441-448. 23. Burgess, B. K., Stiefel, E. 1. & Newton, W. E. (1980) J. Biol. In summary, the above results demonstrate the potential Chem. 255, 353-356. for introducing any site-specific mutation within the genes 24. Beynon, J., Cannon, M., Buchanan-Wollaston, V. & Cannon, encoding nitrogenase components. Such mutant organisms F. (1983) Cell 34, 665-671. are readily constructed irrespective of their Nif phenotype. 25. Buchanan-Wollaston, V., Cannon, M. C. & Cannon, F. C. Importantly, mutants constructed by the procedure de- (1981) Mol. Gen. Genet. 184, 102-106. 26. Kennedy, C. & Robson, R. (1983) Nature (London) 301, scribed here do not alter the expression ofany of the ancillary 626-628. nif-specific components required for metallocenter assembly 27. Roberts, G. P., MacNeil, T., MacNeil, D. & Brill, W. J. (1978) or component maturation. Another important observation is J. Bacteriol. 136, 267-279. Downloaded by guest on September 26, 2021