JOURNAL OF BACTERIOLOGY, Mar. 1987, p. 1127-1136 Vol. 169, No. 30021-9193/87/031127-10$02.00/0Copyright © 1987, American Society for Microbiology

Genetic and Structural Analysis of the Rhizobium melilotifixA, fixB,fixC, and fixX Genes

CHRISTOPHER D. EARL,"2t CLIVE W. RONSON,"3t AND FREDERICK M. AUSUBEL1*

Department of Genetics, Harvard Medical School, and Department of Molecular Biology, Massachusetts GeneralHospital, Boston, Massachusetts 021141; Department of Cellular and Developmental Biology, Harvard University,

Cambridge, Massachusetts 021382; and Grasslands Division, Department of Scientific and Industrial Research, PrivateBag, Palmerston North, New Zealand3

Received 18 August 1986/Accepted 25 November 1986

The fixA, fixB, fixC, and flxX genes of Rhizobium meliloti 1021 constitute an operon and are required fornitrogen fixation in alfalfa nodules. DNA homologous to the R. meliloti fixABC genes is present in all otherRhizobium and Bradyrhizobium species examined, butJlxABC-homologous sequences were found in only onefree-living diazotroph, Azotobacter vinelandii. To determine whether the fixABCX genes share sequencehomology with any of the 17 KlebsieUla pneumoniae nif genes, we determined the entire nucleotide sequence ofthe fixA, fixB, fixC, and fixX genes and defined four open reading frames that code for polypeptides ofmolecular weights 31,146, 37,786, 47,288, and 10,937, respectively. Neither DNA nor amino acid sequencehomology to the R. melilotifixA, -B, -C, and -X genes was found in the K. pneumoniae nifoperon. The fixX genecontains a cluster of cysteine residues characteristic of ferredoxins and is highly homologous to an Azotobacterferredoxin which has been shown to donate electrons to nitrogenase. ThefixABC operon contains a promoterregion that is highly homologous to other nifA-activated promoters. We also found a duplication of the 5' endof thefixABCX operon; a 250-bp region located 520 bp upstream of theJixABCX promoter bears more than65% homology to the 5' end of the transcribed region, including the first 32 codons of fixA.

Several genes have been identified as essential for symbi-otic nitrogen fixation by the bacterium Rhizobium meliloti.These genes are characterized as nif and fix genes. Bydefinition, the nif genes of R. meliloti are those which bearstructural or functional homology to the well-characterizednifgenes of the free-living, nitrogen-fixing species Klebsiellapneumoniae. Thefix genes, on the other hand, are essentialfor nitrogen fixation by virtue of the Fix- phenotype ofnodules elicited by strains which contain mutations in thesegenes (1, 4, 15, 29, 31) but have not yet been assigned abiochemical function.

In R. meliloti three fix genes, fixA, fixB, and fixC, werepreviously identified which are closely linked to a cluster ofnif genes located on a large endogenous plasmid, the so-called Sym plasmid (29, 31) (Fig. 1). The fixABC genes arelocated in a single operon (6, 31) and are transcriptionallyactivated coordinately with the nitrogenase structural genesnifH, -D, and -K by the nifA gene product (38).

In the work reported here, we examined the fine structureof the fixABC genes and attempted to determine their evo-lutionary conservation by searching for physical homologybetween the fixABC genes and the genomes of otherdiazotrophs. We obtained the DNA sequence of the 4,500-base-pair (bp) fixABC region and report here the first com-plete nucleotide sequence of fix genes unrelated to K.pneumoniae nif genes. Interestingly, a previously unidenti-fiedfix gene (flxX) was found to be situated betweenflxC andnifA. The new gene is transcribed as part of the fixABCoperon and codes for a protein that is highly homologous toan Azotobacter ferredoxin. We found hybridization between

* Corresponding author.t Present address: Plant Resources Venture Funds, Cambridge,

MA 02138.t Present address: BioTechnica International, Cambridge, MA

02140.

R. melilotifixABCX and DNA from all other Rhizobium andBradyrhizobium species examined. Of the three free-livingdiazotrophs, only genomic DNA from Azotobactervinelandii hybridized to fixABCX probes. The R. melilotifixABCX genes neither showed homology to the 17 nifgenesof K. pneumoniae nor complemented mutations in any of 12K. pneumoniae nif genes tested.

MATERIALS AND METHODSBacterial strains and plasmids. Strains, plasmids, and

bacteriophage vectors used are shown in Table 1. K. pneu-moniae nifmutant strains contain point mutations created bynitrosoguanidine or diethylsulfonic acid mutagenesis andwere kindly provided by the laboratory of W. Brill. UNF706is a his nif host for pMF250, a low-copy plasmid derivedfrom pRD1 (24), which contains the entire his-nif region ofK. pneumoniae.

Media. Rich medium was L broth, and minimal mediumwas M9 (26). Antibiotic concentrations were: ampicillin, 50,ug/ml for Escherichia coli or 200 jig/ml for K. pneumoniae;kanamycin, 20 ,ug/ml; and spectinomycin, 100 ,ug/ml.DNA manipulations. Standard procedures such as restric-

tion digests, agarose gels, ligations, cloning, nick transla-tions, and large- and small-scale alkaline lysis plasmid prep-arations were as described by Maniatis et al. (23). DNA wastransferred by the Southern procedure (34) onto GeneScreen nylon membranes from New England Nuclear Corp.(Boston, Mass.). DNA-DNA hybridization reactions weredone as described in Gene Screen protocol no. 3, using 50%formamide and 10% dextran sulfate at 42°C, and includedwashes at medium stringency, 55°C in 0.1 x SSC (1 x SCC is0.15 M NaCl plus 0.015 M sodium citrate). Low-stringencyhybridizations were the same except that the reaction mix-tures contained 30% formamide and were done at 37°C.

Plasmid constructions. Plasmid pCE101 was created bycloning the 4.5-kilobase (kb) HindIII fragment of pRmBE1

1127

1128 EARL ET AL.

RH Bg R RH H1i I I I 1

Bg H RI

HHB R BM I I

Bg H BI j

<InifK nifD nf/H

A M

I I IZ>=D ZZDfixA flxB fixC nifA nifB8

N S N SM N PI I1 I Il

Transcripts1052

Open ReadingFrames 1168 2046 2078

f4A fixO

, 1

3139 3151

fix C

4931

4458 4471 4767 4985fAX nifA

FIG. 1. Map of the R. meliloti nif-fix region, showing coding regions and direction of transcription for the nif and fix genes. The hatchedregion in the top line represents the limits of the DNA sequence data reported in this paper and is expanded in the lower portion of the figure.Numbers indicate coordinates of transcriptional and translational start and termination sites in the sequence, with 1 being the first A in theleft-hand Hindlll site. Restriction enzymes are abbreviated as follows: R, EcoRI; H, Hindlll; A, AvrII; B, BamHI; Bg, BglII; M, MluI; N,Narl; P, PstI; S, Sall.

(11) (see hatched region of Fig. 1) and was the source ofDNA for sequencing. Two plasmids were constructed withthe objective of placing the fixABC genes under the controlof the chloramphenicol acetyltransferase (CAT) promoter.Plasmid pCE116 contains the 3.8-kb EcoRI-HindIII frag-ment of pCE101, which carries thefixABC genes, cloned intoEcoRI- and HindIII-cleaved pBR325 (2,7). This constructioneliminates portions of the pBR325 genes coding for chloram-phenicol and tetracycline resistance and places the fixABCgenes behind the CAT promoter without creating a transla-tional fusion between CAT andfixA (CAT terminates beforereaching the fixA gene).The second plasmid, pCE117, is an EcoRI-to-AvrII dele-

tion of pCE116 designed to remove the fixA promoter.Plasmid pCE116 was cleaved with restriction enzyme AvrII(kindly provided by E. Rosenvold ofNew England BioLabs,Inc., Beverly, Mass.) and EcoRI. Sticky ends were filled outwith Klenow polymerase, and the plasmid was religated. Acytosine at the 5' end of the cleaved AvrII site allowedregeneration of the EcoRI site.

Sequencing strategy. We used the shotgun cloning anddideoxy sequencing strategy of Bankier and Barrell (3) tosequence the 4.5-kb HindIII fragment from pCE101. A 10-,gportion of the fragment was isolated by electroelution from a1% agarose gel, self-ligated overnight at 0°C, and sonicated(three 40-s bursts at full power) in a cell disruptor (W-375;Heat Systems-Ultrasonics, Plainview, N.Y.) equipped witha cup horn. The DNA was phenol extracted and ethanolprecipitated. Ragged ends were repaired with Klenow poly-merase or T4 DNA polymerase in the presence of fourdeoxynucleoside triphosphates. The fragments were sizefractionated on a 1.5% agarose minigel; fragments from 400to 1,500 bp were isolated by electroelution and purified byphenol extraction and precipitated. Between 25 and 100 ng ofinsert DNA was ligated into 20 ng of M13mplO that had beencleaved with SmaI restriction endonuclease and treated withcalf intestine phosphatase (Boehringer MannheimBiochemicals, Indianapolis, Ind.). Bankier and Barrell (3)used E. coli JM103 as a transformation host; however,because JM103 and related strains transform poorly and are

hsdR +, we instead transformed into the high-frequencytransforming strain MC1061 (hsdR) and plated onto a lawn ofstrain G-157 (JM103 with the cryptic P1 lysogen deleted).Phage produced in transformed MC1061 infected G-157 cellsthrough their F pili, producing plaques overnight (25).Plaques were picked into 2 ml of G-157 cells diluted 1:100from a saturated culture. After 5 to 6 h of growth, the cellswere centrifuged, and the supernatants were used to preparephage DNA (3).To sequence the region to the right (Fig. 1) of the 4.5-kb

HindIIl fragment, the 4.4-kb PstI fragment containing part offixC, fixX, and nifA was isolated from pRmWB536 (11) andthen either digested with BamHI and cloned into M13mpl9cut with PstI plus BamHI or digested with HindIII andcloned into M13mpl8 cut with Hindlll. This enabled us toobtain sequence data between the BamHI site at nucleotideposition 4826 in Fig. 1 and the 4.5-kb HindIII fragment.Additional sequence data to the right of the HindlIl siteadjacent to BamHI 4826 was obtained previously by W.Buikema (unpublished data; Ph.D. thesis, Harvard Univer-sity, Cambridge, Mass., 1985). The sequence data presentedin this paper adjoin the sequence data presented in reference12.Sequencing was done in 0.5-ml microcentrifuge tubes

inserted in racks designed to be spun in microtiter adaptersof a Sorvall RC6000 or Jouan C3000 centrifuge. Sequencingreactions were exactly as described by Bankier and Barrell(3), using the 15-base primer 5'-TGT-AAA-ACG-ACG-GCC-3' synthesized in the laboratory of J. Smith, Department ofMolecular Biology, Massachusetts General Hospital. Com-pressions were resolved by resequencing appropriate cloneswith dITP substituted for dGTP.Sequence data were entered into a PDP11/780 VAX/VMS

computer with a GTCO (Rockville, Md.) digitizing tablet andthe GELREAD program of W. Buikema (this laboratory).The random sequence was arranged in coherent form withthe DBAUTO, DBUTIL, DBCOMP, and related programsof Staden (35). An average of 5.20 bases of raw data wasentered for each base in the final consensus; after somedirected cloning and sequencing of pCE101 fragments in

H- P RNI, I IIu

HSH B

4826

xxxxxxxx I

I

J. BACTERIOL.

RHIZOBIUM fixABCX GENES 1129

TABLE 1. Bacterial strains and plasmidsStrains and plasmids Characteristics Source or reference

ara leu lac gal hsdR rpsLAlac-pro thi rpsL supE44 endA sbcBlS hsdR4 (F' traD36 ProAB+

lacIq AlacZMJ5)thr leu pro lac ara galK xyl mtl supE44 thi recA13 hsdR hsdM rpsLrecAl uvrA6 phr-Jtrp recA his Spcr

HB101CSR603JC5466

M. CasadabanG. Gussin

H. Boyer32F. Cannon and N. Willets

Klebsiella pneumoniaeaUN2138UN1647UN1656UN27UN1990UN1011UN1688UN1639UN195UN1649UN1678UN1691UNF706Kp5614

Other bacterial speciesRhizobium meliloti 1021Rhizobium leguminosarum 128C53Rhizobium trifolii Rld164Bradyrhizobium japonicum 192Bradyrhizobium sp. (Parasponia)Rp501

Azotobacter vinelandii OPBacillus subtilis 168Anabaena strain 7021Agrobacterium tumefaciens C58

PlasmidspMF250pPC922pCE101pCE116pCE117M13mpl0, M13pll

nifQ4969nifA4683nifF4692nifM4027nifV4944nifS4359nifN4l724niJE4675nifK4195nifD4685nifH44714nifJ4727hsdR recA Ahis-nifrecA hsdR

J. ImperialJ. ImperialJ. ImperialJ. ImperialJ. ImperialJ. ImperialJ. ImperialJ. ImperialJ. ImperialJ. ImperialJ. ImperialJ. ImperialF. Cannon30

11B. KneenS. KumarK. WilsonJ. Tjepkema

W. Orme-JohnsonR. LosickR. HaselkornJ. Schell

F. CannonC. ElmerichThis workThis workThis workJ. Messing

Gnd+ His' Nif' ShiA+ Tra+ (Tn5)NifF AprpBR322 FixABC+pBR325 FixABC+pBR325 FixABC+

a All UN strains are rec+ and wild type except for the nif point mutation.

M13, the sequence was confirmed in its entirety with datafrom both strands.

Maxicell experiments. Expression of pCE116 and pCE117in maxicells, labeling, and polyacrylamide gel analysis ofproteins was as described by Sancar et al. (32) with minormodifications by de Bruijn and Ausubel (17).K. pneumoniae complementation experiments. In an at-

tempt to complement K. pneumoniae nif point mutationswith the R. melilotifixABC genes, we transformed pCE117into each strain by the standard CaCl2 procedure (23).Plasmid pBR325, the vector for pCE117, was transformedinto these strains as a negative control. Because all the K.pneumoniae nif strains are hsdR +, the DNA was firsttransformed into the hsdR strain Kp5614 (30), and alkalinelysis miniprep DNA was used to transform the nif strains.The positive control plasmid pMF250 contains the entire K.pneumoniae his-nifregion on a broad-host-range vector withTn5 inserted in a nonessential region as a selectable marker.pMF250 was conjugated from UNF107 into E. coli JC5466by plating mating mixtures on minimal medium containingtryptophan, spectinomycin, and kanamycin. JC5466 contain-

ing pMF250 was then used as a donor for the recipient K.pneumoniae strains, and exconjugants from these matingswere selected on minimal kanamycin plates.

Derepression of the K. pneumoniae strains was done asdescribed previously (30) in NFDM medium (13) containing20 ,ug of serine per ml as the nitrogen source and with orwithout (NH4)2SO4 as a control for inhibition of nif activa-tion. Acetylene reduction assays were as described previ-ously (30).

RESULTS AND DISCUSSION

Conservation offixABCX gene structure. To determine theextent offixABCX conservation among a variety of nitrogen-fixing species, we used pCE117 (which contains the R.melilotifixABC genes and the 5' end offixX; see Fig. 1) as ahybridization probe against EcoRI digests of genomic DNAfrom Rhizobium leguminosarum, Rhizobium trifolii,Bradyrhizobium japonicum, Bradyrhizobium sp. (Para-sponia), Azotobacter vinelandii, Anabaena strain 7021, Agro-bacterium tumefaciens, and Bacillus subtilis (the latter two

Escherichia coliMC1061G-157

VOL. 169, 1987

1130 EARL ET AL.

were included as presumptive negative controls). As ex-pected from previous experiments which showed thatfixABC genes are conserved among Rhizobium andBradyrhizobium species (18, 29; C. Earl and B. T. Nixon,unpublished data), R. meliloti fixABCX DNA hybridizedstrongly to DNA from R. leguminosarum and R. trifolii andweakly to DNA from B. japonicum and Bradyrhizobium sp.(Parasponia), presumably reflecting the evolutionary diver-gence between Rhizobium and Bradyrhizobium species (datanot shown). Among the other species tested, only DNA fromAzotobacter vinelandii showed hybridization to the R.melilotifixABCX region (data not shown).To determine whether the R. meliloti flxABC genes are

structurally homologous to any of the 17 K. pneumoniae nifgenes, individualfixA,fixB, andfixC probes were hybridizedunder medium and low stringencies (see Materials andMethods) to SalI digests of plasmid pPC922 which containsthe entire K. pneumoniae nifgene cluster. No hybridizationwas observed (data not shown). Fuhrmann et al. (18) alsofound that thefix genes ofB. japonicum failed to hybridize toK. pneumoniae nif DNA. The Azotobacter vinelandii andRpS01 DNA served as positive controls in the K. pneumo-niae hybridization experiments.Complementation of K. pneumoniae nif mutants with the

fixABC genes. Although R. meliloti fixABCX DNA failed tohybridize to any of the 17 K. pneumoniae nif genes, wetested the possibility that the fixABC genes might encodeproteins with functions similar or identical to one or more ofthe K. pneumoniae nif gene products. Plasmid pCE117 wasintroduced into K. pneumoniae nifmutant strains (Materialsand Methods; Table 1). Since pCE117 contains a CATpromoter fused to the fixABC genes but lacks the fixApromoter, we eliminated the possibility of multicopy nifinhibition (30). However, no evidence of complementationas monitored by acetylene reduction assays on derepressedcultures was obtained. In contrast, the nif mutant strainswere fully complemented by pMF250, a Nif+ plasmid con-taining the entire his-nif region of K. pneumoniae.One explanation for the failure of pCE117 to complement

K. pneumoniae nif mutants is that pCE117 does not containan intact fixX gene. Unfortunately, fixX was only identifiedrecently, long after the complementation tests with pCE117had been performed (see next section).

Nucleotide sequence offixABCX genes. In an attempt to findhomologies at the amino acid sequence level between the R.melilotifixABC genes and the K. pneumoniae nifgenes, wedetermined the nucleotide sequence of the 4,500-bp HindIIIfragment shown in Fig. 1. The sequence data are presentedin Fig. 2 and are summarized in the expanded map of Fig. 1.Three open reading frames (ORFs) which code forpolypeptides containing 293, 354, and 436 amino acids withmolecular weights of 31,146, 37,786, and 47,288 were iden-tified by using the ANALYSEQ program of Staden (37) (Fig.3).Subsequent to sequencing the 4.5-kb HindIII fragment,

the presence of an ORF in the region between fixC and nifAin R. trifolii was brought to our attention by S. Iismaa and J.Watson (personal communication). This led to the identifi-cation of a 73-bp HindIlI fragment between fixC and nifA inR. meliloti 1021 which was cloned and sequenced (seeMaterials and Methods for details). This analysis showedthat R. meliloti contains a 297-bp ORF between flxC andnifA. It is likely that this new ORF corresponds to apreviously unidentified essential R. meliloti fix gene, whichwe have termed flxX, because interruption of this ORF bythe endogenous transposable element ISRm2 in R. meliloti

41 results in a Fix- phenotype (I. Dusha, S. Kovalenko, Z.Banfalvi, and A. Kondorosi, submitted for publication).ATG start codons for the four ORFs described above were

chosen as follows: For fixA, there is only one ATG in thefirst 300 bp of the ORF; forfixB,fixC, andfixX, we chose thefirst ATG of each ORF after the preceding coding region.Translational initiation sites resembling the consensus E.coli ribosome-binding sites (33, 37) were found in front ofeach ATG start codon; these are shown as underlinedsequences in Fig. 2.Evidence that the three ORFs identified by the

ANALYSEQ program correspond to thefixA,fixB, andfixCgenes is shown in Fig. 4. Maxicell analysis was performed onplasmid pCE117, a plasmid that contains a transcriptionalfusion between the CAT promoter ofpBR325 and thefixABCgenes (see Materials and Methods). The purpose of thisexperiment was to compare the proposed molecular weightsof the fixA, fixB, and fixC proteins, based on the DNAsequence, with actual polypeptides produced under consti-tutive transcription in E. coli. The upstream noncodingregion between the AvrII site of pCE117 and the fixA startcodon contains at least one stop codon in each of the threereading frames, thereby eliminating the possibility of prob-lems arising from translational fusion of the CAT polypep-tide to the fixA polypeptide or downstream polarity effectson fixB or fixC. The observed proteins are approximately30,000, 37,000, and 47,000 molecular weight. These dataagree with the results obtained by minicell analysis by Puhleret al. (29) and are very close to the molecular weightspredicted by the DNA sequence.Fuhrmann et al. (18) sequenced the 5' end of the

BradyrhizobiumjaponicumfixA gene and the region contain-ing the 3' end offfixB and the 5' end offixC of B. japonicum.The deduced amino acid sequences from R. meliloti and B.japonicum are identical over more than 90% of the comparedregions of fixA, 50% of fixB, and 75% of fixC. The DNAsequence of the promoter region and the 5' end offixA fromR. meliloti 1021 are nearly identical to the sequence data ofBetter et al. (6), who examined thefixA promoter of a relatedstrain of R. meliloti, Rm102F34. Preliminary DNA sequenc-ing in our laboratory of a fixB-homologous sequence fromBradyrhizobium sp. (Parasponia) (B. T. Nixon, unpublisheddata) also showed extensive homology with our sequence offixB from R. meliloti.

The fixABCX products are not homologous to K. pneumo-niae nif gene products. The R. meliloti fixABCX DNA se-quences were compared with published sequences for thenifH, nifB, nifA, nifD, nifK, and nifF genes of K. pneumo-niae. In addition, the fixABCX sequences were comparedwith extensive but preliminary sequence data from shotgundideoxynucleotide sequencing of K. pneumoniae nifM, nil',nil'S, nifU, nifX, nifN, and nifE genes (C. Earl, unpublisheddata) and the nij'L gene (W. Buikema, unpublished data).Both the GAP and BESTFIT programs from the Universityof Wisconsin failed to reveal any homology above randombackground, even for short segments.

Analysis offixABCX promoter region. The promoter regionof the fixABCX operon has been thoroughly analyzed byBetter et al. (5, 6) in R. meliloti 102F34. The promoter-regionsequences that we report here and those of Better et al. (6)are 97% homologous over 400 bases and identical in theirmain features, including the nifA-activated consensus se-quences of TTTTGCA beginning at -17 and CTGG begin-ning at -27. Both sequences also showed that the fixApromoter differs from the nifH promoter by the addition of a9-bp sequence, CTCTGCGAC, directly in front of the CTGG

J. BACTERIOL.

RHIZOBIUM fixABCX GENES 1131

1 AAGCTTAGAGCCTTTTTTCAGATAGTTAGGTCTGGWTTGACCATATGGTCGAACGTACAACTTGTGTCGTTAAGCCGi 80

81 CAAAGGTGAOTAACATTGTOGTATGCGTGGTCGTCACGT0TCCGGTCGGTAAAGGTCCACGATAGAAGGGGCCTGCAi 160

181 TAAGGCCAAGCTCATGCATTCGCACCCATCACACCCGCGGCWTACGTCCCATCACCGAAGTCTCTCTCCAAGCGATCi 240

241 ACACCTGCTGCGCCAGAACCGTGCAGCGGGTTGGCTATTCCGATGGTCTTCAGCCAGCGATAATGAGGAGCGACTTGCCO 320

321 GCCTGGACGACGGCGTTGAGGATCGGGAGGAAGCATCGCCTGCAGGTTGGAGAGCTTCTTCTCGAGCAGGCGCGCATGAT 400

401 GCACGTCCGACAGACGCAGGGGTGTCCCGTCCCCCACATAGCAGCCGCAGACTGTCCGGTTCCACAGGCCGTGT 480

481 TGGGAGGTGACGATGGGCCGACACGACAAAGACTAAAACACGCCGGAATCTCCAAGAACAATACGTCGATCACATTAi 660

661 TGCTTTGTGATGACGTCAGTGGTTGGCACGGTTGATGCTGGAAGAGACAGAAACGCTGGCACACATATGjTTGGAGCCTG 640

841 GAGAGATACCATGCACCTCGTAGTCTGTATCAAGCAGGTGCCGGATTTCTGCGCGAATTCACGTCGCTTCAGTGACGAGC 720

721 ACGATCATGCGCCTTGGGAAGGCGCCGCACGCGCCTTGCOTTCCGATCGCGCAGACAATCCCCGTGCGAGTTGACATGC6 800

801 CCGCCCTCTCAAAAAATGCCCGCCATTGCAACTGCGGCATTCATCACCTCCGCGCCCTATACGCAAAGAAACCCGCCGTC 880

881 GAGCGGATGATTGCCGCAGCCATATGACATTGTCCGTCGCCTCTGTCGGCCCCTCGACAGATTGTTCCTTCAAGCATGCA 960

961 GCCAATTTCCCGATCTAACTATTTGAAAGGAAGCAATTAGCATTATTTCAGTCACCTCTGCGACCTGGCACGACTTTTGC 1040

1041 ACGATCATCCCdCTAGGAAGCGGTAAGAAGACACAATATCGCGCAAGACGGTCAGGACCATTCCGGTTG6TGCTTAACGC 1120

1121 TTGCCAGCOACGCTGGCACACATGTGATTGGAGCCTGGAGAGACACCATGCACCTTGTAGTCTGTATCAiACAGGTGCCG 1200M H L V V C I K Q V P

1201 GATTCCGCGOAAATACGCGTCCATCCGGTGACCAACACGATCATGCGACAGGGCGTGCCGACCATTATCMACCCCCATGi 1280D S A Q I R V H P V T N T I M R Q G V P T I I N P H D

1281 TCTGGCCGCOCTCGAAGAAGCACTGAAGTTGTGCGACACGTATGGAGGCIAGGTTACCGTGGTGACCATGGGCCCTAAGi 1360L A A L E E A L K L C D T Y G G E V T V V T M G P K M

1361 TGGCCGAGGACGCGCTGCGCAAGGCACTCACGTTTGGCGCACACCGCGCCGTTCTCTTGACCGACCGCCATTTTGCAGGC 1440A E D A L R K A L T F G A H R A V L L T D R H F A G

1441 TCGGATACGCTCGCGACCTCCTTCGCCCTTGCTCAAGCAATCGCGGAGATCGGCGAGACCTTCGGCACGCCTGATGTTGi 1620B D T L A T 8 F A L A Q A I A E I G E T F G T P D V V

1621 GTTCACCGGCAAGCAGACGATCGACGGCGACACTGCCCAAGTTGGACCTGGAATTGCCAAGCGCCTCGACCTACAGCAGC 1600F T G K Q T I D G D T A Q V G P G I A K R L D L Q Q L

1801 TCACTTACGTGGCGAAGATTCTCTCCATTGATGCCGCTTCGCGCGAGATCACTGTTGAGCGGCGCGCGCGiAGGCGGCTCG 1680T Y V A K I L S I D A A S R E I T V E R R A E G G S

1681 CAGATCCTGAGAACCGGACTACCATGCCTTGTCACTATGCTGGACGGCGCCGACGCCATCCGTCGCGGGCGTCTCGACGi 1760Q I L R T G L P C L V T M L D G A D A I R R G R L D D

1761 CGCCCTTCGCGCCGCGCGCACCAAGGTCGTCAAGTGGAGTGCGGCGGACGCCGGCATTGCAGAACCGGCCAACTGTGGCi 1840A L R A A R T K V V K W S A A D A G I A E P A N C G L

1841 TGCGAGGATCTCCGACGGTCGTAAAGCGCGTGTTTGCCCCAACTTCCCGCGAACAAAAGGCAAGGCAGA+CGACACCACi 1920R G 8 P T V V K R V F A P T S R- E Q K A R Q I D T T

1921 AATAAGCCACTGCGTGAGATCGCGGACGGTCTGATCGCG0CAATCTTTGCCGACCGGCCGGCCTTGAAACGATCTCGG 2000N K P L R E I A D G L I A A I F A D R P A L K H D L G

*K *I*G*A*A*A*H L2001 CAGCACGGGACAGCAAGGAGCACCAGATGTCGACCGAGAATCGTGAAGCCTGTTCGCCACCCTCCGGACGCGCCGCCATG 2080

8 T G Q Q G A P D V D R E 8 * -M*TG*Q*A*D*D*s**;2081 AAGAAGGGGCTACCCAAGCAATTCCAGGACTATCGGAACGTATGGGTCTTCATCGAACTGGAGCACGGTCAGGTGCATCC 2160

K K G L P K Q F Q D Y R N V W V F I E L E H G Q V H P

2181 CGTCTCCATCGAACTACTCGGTGAAGGCCGCAAGCTCGCCGACAAGCTGGGTGTCCATCTTGCGGGCGTGGTCATTGGGC 2240V S I E L L G E G R K L A D K L G V H L A G VV I G P

2241 CGCCGGGAGGACAGGGCACGGCAAATGCTATTGCGGACGCCTTCGCTTACGGCGCCGACCTCAGCTACCTCGTCGAGTCG 2320P G G Q G T A N A I A D A F A Y G A D L S Y L V E 8

2321 CCGCTACTCGCCCACTACCGUAACGAGCCTTTCACCAAAUCCCTGACGGATCTCGTCTTUGCWCAAGCCGGAAATCCT 2400P L L A H Y R N E P F T K A L T D L V L A N K P E I L

2401 GCTTCTCGGCGCGACTACGCTCGGCCGCGATCTCGCCGGATCGGTGGCAACGACATTGAAGACAGGACTCACGGCCGACT 2480L L G A T T L G R D L A G S V A T T L K T G L T A D C

2481 GCACTGAGCTAAATGTTGATTCAGATGGTTCCCTCGCTGCGACCAGGCCAACCTTCGGAUGTTCCTTGCTCTGCACGATC 2560T E L N V D S D G S L A A T R P T F G G S L L C T I

FIG. 2. Sequence of the fixABCX operon, including 1 kb of upstream noncoding sequence. The promoter and putative upstream nifAbinding sites are underlined, and the arrow indicates the transcriptional start site (6). The likely ribosome-binding site for each gene isindicated by underlining in front of the ATG start codon. The fixA coding region begins at nucleotide position 1168, fixB begins at 2068, fixCbegins at 3151, and fixX begins at 4471. (Continued on next page.)

VOL. 169, 1987

1132 EARL ET AL.

2681 TATACACTCAAGTGTCGACCGCAGATGGCAACCGTACGGCCGAGTGTAATGGCCACGCCGCAACGCGTGAATAGACCAAC 2840Y T L K C R P Q M A T V R P S V M A T P Q R V N R P T

2641 TGGAAGCATCATCCGGCACGATCTGAAAATGCTTGAGGAGGAGATCGCGACCAAAGTCCTTGCCTTTTTTTCCGATTGCG 2720G 8 I I R H D L K M L E E E I A T K V L A F F 8 D C D

2721 ACTCGACCATAGCCAATCTCGCCTACGCCGACGTCGTGGTTGCCGGGGGACTCGGTCTTGGAGCAGTGCAGAACCTTCA0 2800S T I A N L A Y A D V V V A G G L G L G A V Q N L Q

2801 CTCTTGAAG0ATCTCGCACGAACGCTCGGCGGGGATTTTGGGTGTTCGCGGCCGCTAGTCCAAAAGGGGTGGATGCCGTT 2880L L K D L A R T L G G D F G C 8 R P L V Q K G W U P F

2881 TGACCGGCAGATCGGCCAAACGGGCAACACGATCCGGCCAAGCTCTATATAGCGGCTGGGATCTCGGGCGCCGTTCAGO 2980D R Q I G Q T G N T I R P K L Y I A A G I S G A V Q H

2961 ACCGCGTTGGCGTCGAAGGATCTGACCTCATTGTTGCTATCAACACCGATCCGAACGCGCCCATCTTCGATTTCGCCCAC 3040R V G V E G 8 D L I V A I N T D P N A P I F D F A H

3041 CTCGGCGTTGTGGCCGATGCGATCAGTTTTCTGCCTGCCCTGACGGAAGTCTTCACTAAACGGTTGGAGCCACGCAATCT 3120L G V V A D A I S F L P A L T E V F T K R L E P R N L

3121 TGAGAAGTTTGT0CAGTGAGACAAGGACCMATGACAAAGGAAAAATTTGACGCCATCGTCGTTGGTGCCGGTATGTCCGG 3200E K F V Q * T K R K F D A I V V G A G U B GEK V* *KE*DA* V* AG*AS

3201 GAATGCGGCOGCTTACGCGATGGCTAGCCGTGGGCTAAAOGTGCTGCAGTTGGAGCGCGGAGAGTATCCGGGTTCAAAGA 3280N A A A Y A Y A 8 R G L K V L Q L E R G E Y P G 8 K N

3281 ACGTCCAAGGCGCTATAATGTACGCCAACATGCTGGAGGCAATCATTCCAGATTTCCGGiATGACGCACCGCTCGAGCGG 3360Vr q G A I U Y A N U L E A I I P D F R N D A P L E R

3381 CATCTCGTCGAGCAGCGATTCTGGATAATOGACGATACGTCTCACACCGGGATGCACTACGOGTCAGACGATTTCAATGA 3440H L V E Q R F W I U D D T 8 H T G U H Y R S D D F N E

3441 GGTGACGCCAAACCGCTACACGATCATTCGCGCTCAATTCGATAAATGGTTATCGCGCAAGGTGTGCGAGGCTGGCGGAA 3620V T P N R Y T I I R A Q F D K W L S R K V C E A G G T

3621 CGGTCCTGTGAAACAACGCTACAGGACTGGAATGGGACAGTGCCGGGAAGGCGATAGGCGTCCGCACTGACCGCGCT 3800V L C E T T A T G L B W D 8 A G K A I G v R T D R A

3801 GGCGACGTCGTTCTTGCAGATGTGGTCGTGCTCGCTGAAGGGOTCAACGGACTCCTTGGCACGCGCGCCGGCTTACGTGA 3880G D V V L A D V V V L A E G V N G L L G T R A G L R E

3681 GATGCCGAAOTCGAAAAACGTAGCCCTCGCGGTAAAGGAATTGCATTTCCTCCCCGAAGAGGTGATCGCAGAGCGCTTCO 3760U P K 8 K N V A L A V K E L H F L P E E V I A E R F G

3761 GACTGACAGGCGATGAAGGATGCGTGATTGAGGCGGGCGOGACAATTTCOCGCGGOAATGGCAGGATTGGOCTTCCTCTAC 3840L T G D E G C V I E A G G T I S R G U A G L G F L Y

3841 ACCAATAAGOAGTCAATCTCACTCGGCATCGGCTGCCTCATCTCGAACTTCGCCGAGACCATGGAGAGACCTTACGCGCT 3920T N K E 8 I 8 L G I G C L I 8 N F A E T M E R P Y A L

3921 TCTCGACGCCTTTAAACGCCACCCTTCAATCCAGCCGCTAATAGCAGGGTC0OAGTAAAGGAATATGCOGCCCATCTCA 4000L D A F K R H P 8 I Q P L I A G 8 E V K E Y A A H L I

4001 TTCCGGAGG0CGGTTTCAACGCAATACCG0GGCTCTGCG0TAACGGCTGOGTCGTCGTTOGAGACGCGG0GCAACTTAAC 4080P E G G P N A I P R L C G N G w v v v G D A A Q L N

4081 AATGCCGTGCACCGCGAAG0GTCTAACCTTGCCATGGCATCTGGGAGGATGGCGGGAGAOGCGATTTCAATCATAAAGAO 4160N A V H R E G 8 N L A U A 8 G R M A G E A I S I I K 8

4181 CCGCGGGGG0GTAATGGATMAGGCGAGCCTCTCTCTCTACAAAACGATGCTAGACAAGTCCTTCGTTGTTGAGGACCTG0 4240R G G v Ul D K A 8 L 8 L Y K T U L D K 8 F V V E D L 8

4241 GCAAACAAAGGACATGCCTTCCCTGCTCCATACCAATTCCCCAAACTTCTTCACGACGTACCCGCAGCTGATATCGCAT 4320K Q K D Y P 8 L L H T N 8 P N F F T T Y P Q L I 8 H

4321 GCCGCGCAGOACTTCGTGCOCGTGGACGGCACGCCTAAGATTGAGAGGGAATTGCGACCACCGCCGCCTTTCTCAGGGC 4400A A Q N F V R V D G T P K I E R E I A T T A A F L R A

4401 ACGATCCCGATGGGGTCTCOTCAGTGACG0GGTTCGCCTTGCCTCCGCCTGGCGCTAAAOGAGAATGATATGAAGACGO 4480R 8 R W G L V 8 D A V R L A 8 A W R * - U K T AR *R*OV*DAV L*8A* *

MKT

4481 CAATCGCGGAGCGTATCGAAGACAAGCTTTACCAGAACCOGTATCTGGTCGACGCAOGGCGTCCACACATTACAGTGCGG 4680I A E R I E D K L Y Q N R Y L V D A G R P H I T V R

4661 CCACACCGGTCGCCAAGCTTAAACCTGCTGCGOCTCACGCGAGTCTGCCCGGCCAAATGCTACGAGTTGAATGAAACTG0 4840P H R 8 P 8 L N L L A L T R V C P A K C Y E L N E T G

4841 GCAAGTGGAAGTCACTGCCOATGGCTGCATGGAGTGCGG0ACATGCAGAOTGTTGTGCGAGGCAAACGGGTACGTCGAGT 4720Q V E V T A D G C U E C G T C R V L C E A N G D V E W

4721 GGAGCTATCCACGAGGT0GGTTCGGTGTCCTCTTCAAGTTCGGATGAGCCACTCTAAGGTCGATT 47858 Y P R O O F G V L F K 0G *

FIG. 2-(Continued)

J. BACTERIOL.

RHIZOPIUM fixABCX GENES 1133

fixA I fixB .1 fixC l

ktb 1 2 3 4

FIG. 3. ANALYSEQ plot of thefixABC region. For a 67-bp window surrounding each codon, ANALYSEQ measures the probability thatthe codon is part of a protein-coding region. The program examines observed amino acid preference and codon preference and compares themwith what is expected for a protein-coding region with that base composition. The y axis is a probability scale from 0 to 100%. Protein-codingregions are much less random in their codon usage than are noncoding regions and appear as high peaks. Blips are placed on the midline ofthe reading frame with the highest probability at each location in the sequence; contiguous horizontal lines indicate likely protein-codingregions. ORFs are indicated by blips on the top and bottom of each plot, which mark stop codons and ATGs, respectively.

sequence. This sequence interrupts the larger consensusACGGCTGG sequence found in the nifH promoters of bothR. meliloti and K.- pneumoniae (Fig. 2). Transcription beginsat or near base 1052 (6).

Recently, an additional regulatory sequence, TGT-Nlo-ACA, has been identified at a location approximately 100 bp

1 2 3

42 -228 -~ A

FIG. 4. Maxicell products of pCE117 (lane 3), pBR325 (lane 2),and no plasmid control (lane 1) electrophoresed on a 12% polyacryl-amide gel. Major polypeptide bands are indicated. In lane 2, the42-kilodalton band is the product of the tetracycline resistance gene,the 28-kilodalton band is the product of the ampicillin resistancegene, and the 25-kilodalton band is the product of the chloramphen-icol resistance gene. C, B, and A, fixC, fixB, and fixA products,respectively.

upstream of nifpromoters of K. pneumoniae regulated by K.pneumoniae nifA (9). Activation by nifA and inhibition of nifgene expression by the presence of the niJH promoter regioncloned on a multicopy vector (so-called nifinhibition) (8, 30)depends on this upstream element. The sequence is con-served in the R. meliloti promoters of the nifHDK (6) andnifB (10) operons. The sequence TGT-N1o-ACA is locatedfrom positions 924 to 939 on thefixABCX sequence, exactly100 bases in front of the fixA promoter (Fig. 2).The fixC product contains a signal sequence. Comparison

with E. coli signal peptides (42) supports the hypothesis thatthe amino-terminal end of fixC appears to be a signalsequence for membrane insertion. The requisite features ofsuch a sequence include: (i) a charged residue arnong the firstfive residues, usually a lysine, (ii) followed by at least ninehydrophobic or uncharged residues; (iii) a helix-breakingresidue such as glycine or proline or a large polar residuesuch as glutamine from four to eight residues before thecleavage site; (iv) termination of the signal sequence withalanine or glycine. These criteria are all fulfilled by the fixCs,equence. Further investigation is required to determinewhether the signal sequence is cleaved and whether fixCprotein is located on the inner membrane, the outer mem-brane, or in the periplasmic space.

Relationship between fixABCX and nifA operons. The fixXand nifA coding regions are separated by 202 bp and aretranscribed in the same direction (12, 43). Kim et al. (22) andBuike,ma et al. (12) have found that approximately 60% ofthe nifA transcripts in alfalfa nodules are actuallyreadthrough products of thefixABCX operon. After compar-ing the individual activities of the fixA and nifA promoters,Kim et al. (22) conclude that approximately 70% of thefixABCX transcripts do not terminate after fixX, but extendthrough the 202-bp intergenic region into nifA. Althoughseveral weak stem-loop structures could potentially be

VOL. 169, 1987

1134 EARL ET AL.

499 C CACGCCGG AAT;CCAAACGT 548li1 111111 1IIi1 1 11 11 I

1021 CGACCT=.CGACTTTTACG.M-ATCACCCTACCAGCGGTAAGAL 1068_ I

549 C 59811 1111 11 11 1 1111 111111 111

1069 AGA. ..CTATCGC....GCACGTcAGGcCAn'CG ;GC 1113

599 TGGAAG...ATCA 646I I 11111111111 111111111

1114 TTWCC mGG 1163

647 TACCP CACCAT AGCAGGTGCGGT''1TCGCGA 69611111111111 11111111111111 11111111111 111III

1164 ACC CGGA.TTCCWGCGA 1212

M H L V V C I K Q V P D S A Q

697 WTTAGGCG 74611 1 1111 I I 11111 I 111111111111 I 11 I 11

1213 T G CCGC 1262

I R V H P V T N T I M R Q G V P ...

FIG. 5. Homology between the fixA promoter region and a duplication 500 bp upstream.

formed near the transcription initiation site of nifA, none ofthese resembles a rho-independent termination site of E. coli(40).

Duplication of noncoding region upstream offixA. Approx-imately 500 bp upstream of the beginning of the fixABCXoperon is a 250-bp region which bears 65% homology to theregion of DNA from the fixA promoter through codon 32 offixA (Fig. 1 and 5). ThefixA promoter itself is not duplicated.Better et al. (6) have found an upstream duplication of the R.meliloti nifH promoter with an initial section of the niJIcoding region. The two duplicated regions are separatedfrom one another by approximately 500 bp. An attractivehypothesis is that the niJH and fixA promoters, originallyadjacent to each other, were separated by a duplication ofthe 5' regions of the two operons. It is interesting to note thatthe duplicated fixA region has lost its promoter and that theregion of homology terminates precisely at the expectedposition of a promoter duplication. The nifjH duplication, onthe other hand, is strongly transcribed in nodules (6).Comparison with other protein-coding sequences. A com-

puter search of the protein sequence library of Dayhoff et al.(16) was done with FASTP with the fixA,fixB,fixC, andfixXcoding sequences as bases for comparison. This searchrevealed no significant homologies for thefixA,fixB, andfixCproteins with other proteins. However, thefixX product wasclearly related to several ferredoxins sharing the C-x-x-C-x-x-C-x-x-x-C pattern characteristic of these proteins. Thealignment of the fixX product to the four most homologousferredoxins is shown in Fig. 6. The fixX product is unusual

among the ferredoxins in that its six cysteine residues areclustered within the C-terminal half of the protein, whereasother ferredoxins of a similar size have their cysteinesclustered in the N-terminal half.

Conclusions and potential functions of the R. melilotifixABCX genes. The results presented here suggest that theR. meliloti fixA, fixB and fixC genes are not homologous,either structurally or functionally, to K. pneumoniae nifgenes. Nevertheless, the fixABCX genes are required forsymbiotic nitrogen fixation in R. meliloti, and homologousgenes have been observed in all other rhizobia examined sofar. Homology was also detected with genomic DNA fromAzotobacter vinelandii. Recent results from our laboratory(C. Earl, unpublished data) indicate that the fixB and fixCgenes of Bradyrhizobium sp. (Parasponia) Rp501 (identifiedby homology with the sequences described here) are re-quired for ex planta acetylene reduction.Taken together, these results suggest that the fixA, fixB,

and fixC genes are involved with the process of nitrogenfixation per se and are not merely required for nodulemaintenance or bacteroid maturation. Though no homologywith flavodoxins or ferredoxins was discovered in the aminoacid sequence comparisons, and patterns of cysteine resi-dues in the fixA, fixB and fixC proteins do not match thosetypically observed for iron-sulfur-binding proteins (27, 28), itis possible that the fixABC gene products are involved inelectron transport. This possibility is made more likely bythe unexpected discovery of fixX and its high degree ofhomology to ferredoxin I from Azotobacter vinelandii (Fig.

FIXX 45 V AKCYELNE G Q V EV TA DIGICM E C GTCR VLCEANGD E83FEDV2N 21 V C P V ElMJY E L Q G K A V P V N E L[CfSCIEV C NCVE 59FEAV 19 V C P V D C F-Y E GP N F L V I H PD E C I D C ALC E CP A E A I F 56FEPSFV 19 V C P V D C F - Y E G P N F L V I H PD E C I D C A L C E CP A Q A I F S 56FECLCT 17 V EA I- G K Y Q V D CID C G CQA CP A55

* * * * * *

FIG. 6. Homology between fixX and various ferredoxins. Alignment of conserved portion of R. meliloti fixX product to the four mostclosely related ferredoxins identified by using FASTP from the National Biomedical Research Foundation data base. Functionally conservedresidues are boxed, and the conserved cysteines are marked with asterisks. FIXX, R. melilotifixX product; FEDV2N, ferredoxin II fromDesulfovibrio desulfuricans (19); FEAV, ferredoxin I from Azotobacter vinelandii (21); FEPSFV, ferredoxin from Pseudomonas putida (20);FECLCT, ferredoxin from Clostridium tartarivorum (39).

Duplication

Promoterregion

fixA

J. BACTERIOL.

RHIZOBIUM fixABCX GENES 1135

6). Interestingly, the Azotobacter vinelandii ferredoxin I isanalogous to ferredoxin I from Azotobacter chroococcumwhich has been shown to donate electrons directly to nitro-genase (41). A ferredoxin purified from Bradyrhizobiumjaponicum bacteroids has also been shown to transfer elec-trons to nitrogenase (14). However, on the basis of size andamino acid composition, this ferredoxin clearly does notcorrespond to the fixX product. Because Rhizobium, Brady-rhizobium, and Azotobacter species are all obligate aerobes,it is possible that the fixABCX operon codes for a nitroge-nase-specific electron transport system which is unique tothose bacteria that fix nitrogen aerobically.

ACKNOWLEDGMENTS

We thank W. Buikema for crucial assistance with shotgun clon-ing, E. Richards for help with maxicell experiments, J. Beynon, V.Buchanan-Wollaston, and W. Herr for troubleshooting our earlyattempts at sequencing, A. Bankier for providing protocols andexcellent advice, S. Iismaa and J. Watson for pointing out to us theexistence of a small HindIII fragment between fixC and nifA, A.Kondorosi for providing results before publication, and R. Hyde forexpert assistance in preparing the manuscript.

This work was supported by a grant from Hoechst AG to theMassachusetts General Hospital.

LITERATURE CITED1. Agifilar, 0. M., D. Kapp, and A. Puhler. 1985. Characterization

of a Rhizobium meliloti fixation gene (fixF) located near thecommon nodulation region. J. Bacteriol. 164:245-254.

2. Alton, N K., and D. Vapnek. 1979. Nucleotide sequence analy-sis of the chloramphenicol resistance transposon Tn9. Nature(London) 282:864-869.

3. Bankier, A. T., and B. G. Barrell. 1983. Shotgun DNA sequenc-ing, p. 1-34. In R. A. Flavell (ed.), Techniques in life sciences,B5: nucleic acid biochemistry. Elsevier/North-Holland Publish-ing Co., Limerick, Ireland.

4. Batut, J., B. Terzaghi, M. Gherardi, M. Huguet, E. Terzaghi,A. M. Garnerone, P. Boistard, and T. Haguet. 1985. Localiza-tion of a symnbiotic fix region on Rhizobium meliloti pSymmegaplasmid more than 200 kilobases from the nod-nif region.Mol. Gen. Genet. 199:232-239.

5. Better, M., G. Ditta, and D. R. Helinski. 1985. Deletion analysisof Rhizobium meliloti symbiotic promoters. EMBO J. 4:2419-2424.

6. Better, M., B. Lewis, D. Corbin, G. Ditta, and D. R. Helinski.1984. Structural relationships among Rhizobium meliloti symbi-otic promoters. Cell 35:479-485.

7. Bolivar, F. 1978. Construction and characterization of new

cloning vehicles. Part 3. Derivatives of plasmid pBR322 carry-ing unique EcoRI sites for selection of EcoRI generated recom-binant DNA molecules. Gene 4:121-136.

8. Brown, S. E., and F. M. Ausubel. 1984. Mutations affectingregulation of the Klebsiella pneumoniae nifH (nitrogenasereductase) promoter. J. Bacteriol. 157:143-147.

9. Buck, M., S. Miller, M. Drummond, and R. Dixon. 1986.Upstream activator sequences are present in the promoters ofnitrogen fixation genes. Nature (London) 320:374-378.

10. Buikema, W. J., J. A. Klingensmith, S. L. Gibbons, and F. M.Ausubel. 1986. Conservation of structure and location of Rhizo-bium meliloti and Klebsiella pneumoniae nifB genes. J. Bacte-riol. 169:1120-1126.

11. Buikema, W. J., S. R. Long, S. E. Brown, R. C. van den Bos,C. D. Earl, and F. M. Ausubel. 1983. Physical and geneticcharacterization of Rhizobium meliloti symbiotic mutants. J.Mol. Appl. Genet 2:249-260.

12. Buikema, W. J., W. W. Szeto, P. V. Lemley, W. H. Orme-Johnson, and F. M. Ausubel. 1985. Nitrogen fixation specificregulatory genes of Klebsiella pneumoniae and Rhizobiummeliloti share homology with the general nitrogen regulatorygene ntrC of K. pneumoniae. Nucleic Acids Res. 13:4539-4555.

13. Cannon, F. C., R. A. Dixon, and J. R. Postgate. 1976. Deriva-

tion and properties of F-prime factors in E. coli carryingnitrogen fixation genes from Klebsiella pneumoniae. J. Gen.Microbiol. 93:111-125.

14. Carter, K. R., J. Rawlings, W. H. Orme-Johnson, R. R. Becker,and H. J. Evans. 1980. Purification and characterization of aferredoxin from Rhizobium japonicum bacteroids. J. Biol.Chem. 255:4213-4223.

15. Corbin, D., L. Barran, and G. Ditta. 1983. Organization andexpression of Rhizobium meliloti nitrogen fixation genes. Proc.Natl. Acad. Sci. USA 80:3005-3009.

16. Dayhoff, M. O., W. C. Barker, and L. T. Hunt. 1983. Establish-ing homologies in DNA sequences. Methods Enzymol. 91:524-545.

17. de Bruijn, F. J., and F. M. Ausubel. 1983. The cloning andcharacterization of the glnF (ntrA) gene of Klebsiella pneumo-niae: role of glnF (ntrA) in the regulation of nitrogen fixation(nif) and other nitrogen assimilation genes. Mol. Gen. Genet.192:342-353.

18. Fuhrmann, M., H.-M. Fischer, and H. Hennecke. 1985. Mappingof Rhizobium japonicum nijB-, fixBC- and fixA-like genes andidentification of the fixA promoter. Mol. Gen. Genet. 199:315-322.

19. Guerlesquin, F., M. Bruschi, G. Bovier-Lapierre, J. Bonicel, andP. Couchoud. 1983. Primary structure of the two (4 Fe-4 S)clusters in ferredoxin from Desulfovibrio desulfuricans (strainNorway 4). Biochimie. 65:43-47.

20. Hase, T., S. Wakabayashi, H. Matsubara, D. Ohmori, and K.Suzuki. 1978. Pseudomonas ovalis ferredoxin similarity toAzotobacter vinelandii and chromatium ferredoxins. FEBSLett. 91:315-319.

21. Howard, J. B., T. W. Lorsbach, D. Ghosh, K. Melis, and C. D.Stout. 1983. Structure ofAzotobacter vinelandii 7 Fe ferredoxin:amino acid sequence and electron density maps of residues. J.Biol. Chem. 258:508-522.

22. Kim, C.-H., G. Ditta, and D. R. Helinski. 1985. Transcriptionalorganization of a Rhizobium meliloti niflfix gene cluster down-stream from P2-overlapping transcription, p. 186, In H. J.Evans, P. J. Bottomley, and W. E. Newton (ed.), Nitrogenfixation research progress. Martinus Nijhoff, Dordrecht, TheNetherlands.

23. Maniatis, T., E. F. Fritsch, and J. Sambrook. 1982. Molecularcloning, a laboratory manual. Cold Spring Harbor Laboratory,Cold Spring Harbor, N.Y.

24. Merrick, M., M. Filser, C. Kennedy, and R. Dixon. 1978.Polarity of mutations induced by insertion of transposon TnS,Tn7 and TnJO into the nif gene cluster of Klebsiella pneumo-niae. Mol. Gen. Genet. 165:103-111.

25. Messing, J. 1983. New M13 vectors for cloning. MethodsEnzymol. 101:20-78.

26. Miller, J. H. 1972. Experiments in molecular genetics, p. 37-46.Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.

27. Newton, W. E. 1984. Sulfide and other sulfur-containing ligandsin metalloproteins and enzymes, p. 409-477. In A. Muller and E.Krebs (ed.), Studies in inorganic chemistry, vol. 5.Elsevier/North-Holland Publishing Co., Amsterdam.

28. Orme-Johnson, W. H. 1973. Iron-sulfur proteins: structure andfunction. Annu. Rev. Biochem. 42:159-204.

29. Puhler, A., M. 0. Aguilar, P. Hynes, W. Klipp, U. Priefer, R.Simon, and G. Weber. 1984. Advances in the genetics offree-living and symbiotic nitrogen fixation bacteria, p. 609-619.In C. Veeger and W. E. Newton (ed.), Advances in nitrogenfixation research. Nijhoff-Junk/Pudoc, The Hague.

30. Riedel, G. E., S. E. Brown, and F. M. Ausubel. 1983. Nitrogenfixation in Klebsiella pneumoniae inhibited by certain multicopyhybrid nif plasmids. J. Bacteriol. 153:45-56.

31. Ruvkun, G. B., V. Sundaresan, and F. M. Ausubel. 1982. Di-rected transposon TnS mutagenesis and complementation anal-ysis of Rhizobium meliloti symbiotic nitrogen fixation genes.Cell 29:551-559.

32. Sancar, A., A. M. Hack, and W. D. Rupp. 1979. Simple methodfor identification of plasmid-coded proteins. J. Bacteriol. 137:692-693.

33. Shine, J., and L. Dalgarno. 1974. The 3'-terminal sequence of

VOL. 169, 1987

1136 EARL ET AL.

Escherichia coli 16S ribosomal RNA: complementarity to non-

sense triplets and ribosome binding sites. Proc. Natl. Acad. Sci.USA 71:1342-1346.

34. Southern, E. M. 1975. Detection of specific sequences amongDNA fragments separated by gel electrophoresis. J. Mol. Biol.98:503-517.

35. Staden, R. 1982. Automation of the computer handling ot gelreading data produced by the shotgun method ofDNA sequenc-

ing. Nucleic Acids Res. 10:4731-4751.36. Staden, R. 1984. Graphic methods to determine the function of

nucleic acid sequences. Nucieic Acids Res. 12:521-538.37. Stormo, G. D., T. D. Schheider, and L. M. Gold. 1982. Charac-

terization of translational initiation sites in E. coli. NucleicAcids Res. 10:2971-2996.

38. Szeto, W. W., J. L. Zimmerman, V. Sundaresan, and F. M.Ausubel. 1984. A Rhizobium meliloti symbiotic regulatory gene.

Cell 36:1035-1043.

39. Tanaka, M., M. Haniu, G. Matsueda, K. T. Yasunobu, R. H.Hines, J. M. Akagi, E. M. Barnes, and T. Devanathan. 1971.The primary structure of the Clostridium tartarivorum ferre-doxin, a heat stable ferredoxin. J. Biol. Chem. 246:3953-3960.

40. von Hippel, P. H., D. G. Bear, W. D. Morgan, and J. A.McSwiggen. 1984. Protein-nucleic acid interactions in trAnscrip-tion. Annu. Rev. Biochem. 53:389-446.

41. Yates M. G. 1979. Physiological aspects of nitrogen fixation, p.218-270. In W. Newton, J. R. Postgate, and C. Rodriguez-Barrueco (ed.), Recent developments in nitrogen fixation. Aca-demic Press, Inc. (London), Ltd. London.

42. Watson, M. E. E. 1984. Compilation of published signal se-

quences. Nucleic Acids Res. 12:5145-5164.43. Weber, G., H. Reilander, and A. Puhler. 1985. Mapping and

expression of a regulatory nitrogen fixation gene (fixD) ofRhizobium meliloti. EMBO J. 4:2751-2756.

J. BACTERIOL.