Proc. Nati. Acad. Sci. USAVol. 84, pp. 1319-1323, March 1987Genetics

Rhizobium symbiotic genes required for nodulation of legume andnonlegume hosts

(Parasponia Rhizobium/symbiosis/gene replacement)

DEBORAH J. MARVEL*tt, JOHN G. TORREYt, AND FREDERICK M. AUSUBEL*§*Department of Genetics, Harvard Medical School, and Department of Molecular Biology, Massachusetts General Hospital, Boston, MA 02114;and tDepartment of Organismal and Evolutionary Biology, Harvard University. Cambridge, MA 02138

Contributed by John G. Torrey, November 3, 1986

ABSTRACT Parasponia, a woody member of the elmfamily, is the only nonlegume genus whose members are knownto form an effective nitrogen-fixing symbiosis with Bradyrhi-zobium or Rhizobium species. The Bradyrhizobium strainRp501, isolated from Parasponia nodules, also nodulates thelegumes siratro (Macroptilium atropurpureum) and cowpea(Vigna unguiculata). To test whether some of the same genesare involved in the early stages of Igume and nonlegumenodulation, we generated transposon TnS insertions in theregion of three evolutionarily conserved genes (nodA, nodB,and nodC) required for legume nodulation in several Rhizobiumand Bradyrhizobium species. Assays of these mutant Rp501strains on legume hosts and Parasponia seedlings establishedthat nodABC are required for nodulation of legume andnonlegume hosts, indicating that nonlegumes and legumes canrespond to the same bacterial signal(s). In addition, a straincarrying a TnS insertion adjacent to the nodABC genes vigor-ously nodulated RpSO1 legume hosts but was incapable ofnodulating Parasponia, possibly identifying a nonlegume-specific nodulation function.

Symbiotic associations between nitrogen-fixing bacteria andplants substantially reduce host plant requirements for ex-ogenous fixed nitrogen. Almost exclusively, symbiotic nitro-gen-fixing bacteria of the family Rhizobiaceae infect the rootsof host plants of the legume family. Parasponia, a woodymember of the elm family, is the only nonlegume genuswhose members are known to form an effective nitrogen-fixing symbiosis with rhizobia (1).The rhizobial infection pathway of the nonlegume host

Parasponia differs substantially from the infection pathwaysof most legume hosts, in particular with respect to earlyinfection events (2, 3). Root hair curling and infection threadformation are not primary events in Parasponia infection;rather, initial infection stages have been characterized aspathogenic in appearance, involving intercellular penetrationof the epidermis, often accompanied by degradation ofcortical cells (2). Eventually, infection threads do develop,and plant cell infection is achieved solely by way of theseinvading threads. Parasponia rhizobia fix nitrogen whileretained within these threads and, in contrast to legumenodules, do not lie within the host cell cytoplasm as differ-entiated membrane-bound bacteroids. Vasculature of legumeand nonlegume nodules also differs (3).

Previously, we identified three "common" nodulationgenes, nodA, nodB, and nodC, in Rp501, a slow-growingBradyrhizobium strain isolated from Parasponia rigida nod-ules and demonstrated structural and functional conservationbetween the Bradyrhizobium sp. (Parasponia) nodABCgenes and the corresponding nodulation genes of the alfalfasymbiont Rhizobium meliloti (4, 5). Analogues of the R.

meliloti nodABC genes have been shown to be present in avariety of Rhizobium and Bradyrhizobium strains, indicatingthat the products of these genes play a common role in allRhizobium-legume symbioses (6-9). In R. meliloti, expres-sion of nodABC is regulated by a plant-derived flavoneinducer molecule, luteolin (10).Because Bradyrhizobium sp. (Parasponia) strain Rp501

also nodulates the legumes siratro (Macroptilium atropurpu-reum) and cowpea (Vigna unguiculata), it is likely that theRp501 nodABC genes are required for legume nodulation;however, it is of interest to determine whether nodABC arealso required for nodulation of Parasponia. In this report wedescribe Rp501-derived mutant strains, each of which con-tains a TnS insertion in a different location in the Rp501nodulation region. Rp501 nodABC::TnS insertions displayeda nodulation-defective phenotype on both legume hosts andon Parasponia.

MATERIALS AND METHODSBacterial Strains and Plasmids. Bacterial strains and plas-

mids used in these experiments are listed in Table 1. MutantpPRC6 plasmids used in construction ofRp501 mutant strainshave been described (5), and locations of TnS insertions areshown in Fig. 1.Media. LB, TY, and M-9 sucrose bacterial media and Nod

plant medium have been described (13). Antibiotics werepurchased from Sigma.

Bacterial Genetics. Rp501 nodABC::Tn5 strains were con-structed as follows. Mutant pPRC6: :TnS plasmids weremobilized from E. coli strain MM294 into Rp501 in triparentalmatings with MM294/pRK2013 (14) and exconjugants wereselected on solid TY medium containing streptomycin (Sm)at 1000 ,ug/ml and kanamycin (Km) at 150 ,4g/ml at 32°C.SmrKmr (where r = resistant) exconjugant colonies weregrown in liquid TY medium containing streptomycin at 1000,ug/ml and kanamycin at 150 ,g/ml at 32°C until cultureswere nearly saturated (5-7 days). Exconjugant cultures werepelleted and resuspended in TY medium prior to matings onTY plates with E. coli strain 2174 containing the gentamycin(Gm) resistance plasmid pPH1IJ. Exconjugants of this matingwere scraped from plates and suspended in S ml of liquid TYmedium containing streptomycin at 1000 ,ug/ml, kanamycinat 150 ,ug/ml, and gentamycin at 70 ,g/ml. After an overnightincubation at 32°C, cultures were plated on TY platescontaining streptomycin at 1000 ,ug/ml, kanamycin at 150,.ug/ml, and gentamycin at 70 ,ug/ml. Colonies grew in 7-10days. One hundred SmrKmrGmr colonies from each of twomatings were screened in colony hybridizations (per speci-

Abbreviations: Sm, streptomycin; Km, kanamycin; Gm, gentamy-cin; r, resistant.tPresent address: Department of Cellular and Developmental Biol-ogy, Harvard University, Cambridge, MA 02138.§To whom reprint requests should be addressed.

1319

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Apr

il 25

, 202

0

Proc. Natl. Acad. Sci. USA 84 (1987)

Table 1. Bacterial strains and plasmids

Source orRelevant markers ref.

StrainEscherichia coliMM294 endA, thi-1, hsdR17, M. Meselson

supE44, pro2174 met, pro A. Johnston

Bradyrhizobiumsp (Parasponia)Rp501 str-r J. Tjepkema

PlasmidpPRC6 tet-r, pLAFR1 (inc-P) with 5

21.3-kb insertpRmSL42 ap-r, pBR322 with R. S. R. Long

meliloti BamHI-HindIIIinsert (nodABC)

pRmJ30 ap-r, pBR322 with 8.7-kb S. R. LongR. meliloti insertincluding nodDABC

pPH1IJ gent-r, inc-P 11pRK2013 repcolEl, nm-r, km-r 12

kb, Kilobases.

fications of New England Nuclear, manufacturer of Gene-ScreenPlus) for retention of vector sequences. None of thesecolonies retained vector sequences. Under the conditionsdescribed, we saw no instances of cointegration into thegenome ofpPRC6 plasmids carrying cloned and mutagenizedDNA, such as have been observed using other gene replace-ment techniques in Bradyrhizobia (15). Frequency ofhomog-enotization was -1 x 10o--comparable to frequenciesachieved in fast-growing rhizobia (16).DNA Biochemistry. Total DNA of Bradyrhizobium sp.

(Parasponia) strain 501 and derivatives was prepared asdescribed (14). Restriction endonucleases were purchasedfrom Boehringer Mannheim or Bethesda Research Labora-tories and were used according to manufacturers' specifica-tions. Agarose gel electrophoresis was performed as de-scribed (17). Southern blotting and hybridizations wereperformed as described by Marvel et al. (5). Washes werecarried out at 65°C for 2 hr in 0.1% NaDodSO4/15 mMNaCl/1.5 mM sodium citrate. Nick-translations were per-formed according to Maniatis et al. (18). Restriction frag-ments for nick-translation were isolated from agarose gelsusing a glass powder DNA purification protocol (19).R. meliloti Sequences Used as Hybridization Probes. Radio-

labeled probes used to identify putative Rp501 nodD se-quences were derived from the coding sequences of the nodDgene ofR. meliloti strain 1021 and include a 700-base-pair (bp)Bgl II-Ssp I fragment from the plasmid pRMJ30 (Table 1). Asecond, smaller R. meliloti nodD probe, carrying sequencesfrom the 5' end of the nodD gene, hybridized primarily to a600-bp Xho I-Cla I pPRC6 fragment (data not shown). Theprobe comprised a 240-bp Bgl II-BamHI restriction fragmentfrom plasmid pRmSL42 (Table 1). Sizes of restriction frag-ments used as hybridization probes were deduced frompublished R. meliloti nodD sequence data (20).

Nodulation Assays. Siratro (M. atropurpureum) seeds weresterilized in concentrated H2SO4 for 12 min and then rinsedsix times in sterile distilled H20 prior to sowing on Nodmedium agar slants (13). One seed was sown per tube. Siratroseeds were inoculated with Rhizobium 2-4 days after sowing.Nodules were visible 10 days to 2 wk after inoculation andplants were scored for nodules 3-4 wk after inoculation.Cowpea seeds (V. unguiculata) were sterilized in 0.2%

mercuric chloride for 3 min and then were rinsed six timeswith sterile distilled H20. Seeds were sown on H20/1% agarplates. Upon germination, sprouted seeds were transferred to

sterile growth pouches containing 20 ml of 1x Bergersen'smedium (21). Pouches in growth chambers were covered withplastic bags to minimize cross-contamination. Cowpeas wereinoculated with Rhizobium cultures 3-5 days after transfer togrowth pouches. Two or three seedlings were grown in eachpouch and plants were scored for nodules 3-4 wk afterinoculation.Parasponia rigida fruits were hydrated overnight with

distilled H20 so that the fleshy fruit could be peeled awayfrom the seed. Peeled seeds were scarified with sand paperand then sterilized in H2SO4. Sterilized seeds were washedthoroughly with distilled H20 and sown on H20/1% agarplates. When seeds germinated, seedlings were transferred toPetri dishes containing Nod medium. Seedlings were inocu-lated 1-2 wk after transfer to Nod medium. Plates werewrapped with Parafilm. Plants were scored for nodules 3 wkand 6 wk after inoculation.

Inocula were prepared fresh by pelleting logarithmic-phase(-1 x 108 cells) Rp5O1 cultures grown in selective media.Cells were resuspended in 5 ml of H20. Suspensions werediluted 1:10 in 5 ml of H20, and 1/10th of the dilution volumewas used to inoculate tubes or pouches. Smaller volumes ofthe diluted cultures were used to inoculate Parasponiaseedlings.

After nodules had been scored, acetylene reduction assaysof siratro plants were performed as described (13). Cowpeanodules and roots were removed from growth pouches andfloated in 15 ml of sterile distilled H20 in 250-ml sterile flasks,which were sealed with Suba seals. Flasks were injected with5 ml of acetylene and incubated overnight in the dark,shaking. The following day, acetylene reduction assays wereperformed as above. Prior to assay, Parasponia roots andnodules were incubated overnight on circles of H20-saturat-ed Whatman 3MM paper, in Suba-sealed glass stab bottlesthat were injected with 0.5 ml of acetylene.

RESULTSFig. 1 is a physical and genetic map of a 13.4-kb EcoRIfragment from Bradyrhizobium sp. (Parasponia) strainRp501, which carries the nodABC region and is contained inplasmid pPRC6. Subsequent to the identification ofthe Rp501nodABC genes (4, 5), a putative homologue of the R. melilotinodD locus was also mapped to the 13.4-kb EcoRI fragment

M-125 M l 14 M-1I28M1*29 M*2I11 M;1*13

M2.9 M 2 5MI-F21 Ml-24M-212 M-2 14

Bg Bg B

EtS BgBX xlCsI IIII(

SBI I

S c X C CX X B EI I I 1 I I

nodC nodAB nodD

FIG. 1. Physical map of Rp501 nodulation loci. Physical map ofthe 13.4-kb EcoRI fragment from Bradyrhizobium sp. (Parasponia),which complements nodABC mutants of R. meliloti. Shaded regionsindicate restriction fragments, portions of which are homologous tonodAB, nodC, or nodD probes from R. meliloti strain 1021, asrevealed in Southern hybridization analyses. Hybridization andcomplementation data suggest the order of the nodABC genes isconserved between Rp501 and R. meliloti. Also shown are mappedlocations ofplasmid-borne TnS insertions created in E. coli. Plasmidscontaining Tn5 insertions at the locations shown, except insertionM-2-11, were used to construct mutant Rp501 derivatives. E, EcoRI;B, BamHI; S, Sal I; Bg, Bgl II; C, Cla I; X, Xho I.

1320 Genetics: Marvel et al.

Dow

nloa

ded

by g

uest

on

Apr

il 25

, 202

0

Proc. Natl. Acad. Sci. USA 84 (1987) 1321

(Fig. 2) by Southern blotting and hybridization analysis, usingradiolabeled probes derived from the coding sequence of R.meliloti nodD. In fast-growing rhizobia, nodD has a demon-strated role in regulation of the nodABC genes (8, 22). Thefunction of the nodD homologue in Bradyrhizobium sp.(Parasponia) has not been established as yet.Transposon TnS insertion mutations in the Rp501 nod-

ABCD region were previously generated in E. coli in thecloned 13.4-kb EcoRI fragment from Rp501 (Fig. 1) (5). Todefine the functions of these TnS mutated nod genes in thenatural Rp501 background, we replaced wild-type Rp501genes with sequences mutated with TnS using a proceduremodified from the gene replacement protocol of Ruvkun andAusubel (14) and described in detail in Materials and Meth-ods. A Southern blot of Xho I-digested genomic DNA from11 putative gene replacement mutants was probed withradiolabeled pPRC6 DNA, verifying that, in each case, thewild-type Xho I fragment had been replaced by a homologousfragment containing TnS (Fig. 3).Bradyrhizobium sp. (Parasponia) strains carrying the 11

nod region::Tn5 insertions were assayed on siratro plants, alegume host infected through root hairs, for their capacity toelicit nodules, to curl root hairs, and to reduce acetylene toethylene in nitrogenase assays (see Table 2). Three Tn5insertion strains, Rpl-21, Rp2-12, and Rpl-28, which containTnS insertions in the nodABC homologous region, did notnodulate siratro. Mutant strain Rp2-9 elicited very few siratronodules, which appeared later and which did not reduceacetylene. All of the remaining mutant strains, includingthose in which TnS mapped to nodD homologous sequencesof Rp501, elicited Fix' nodules on siratro plants within 2½2wk after inoculation. Mock-inoculated plants had no nodules.Bacteria were reisolated from surface sterilized nodules (fivenodules per mutant strain) and four colonies from eachnodule were tested for kanamycin resistance. All reisolatedcolonies were Kmr, indicating that the nodules were elicitedby the TnS insertion mutant strains and not by wild-typeBradyrhizobium contaminants.

Mutants of Rp501 that could not elicit siratro nodules alsodid not elicit root hair curling, although some branching ofsiratro root hairs was observed. All of the mutants that

Tz_ c4)0 ILI) -Cy1..

*- 5.2kbFIG. 2. Structural conservation of

nodD sequences between Bradyrhizo-bium sp. (Parasponia) and R. meliloti.

2.9 k b Southern blot hybridization of the 700-bp- 2.4 kb Bgl II-Ssp I nodD fragment from R.meliloti to Xho I-digested DNA fromplasmid pPRC6 (labeled Rp501). The

* -1.4kb nodD probe hybridizes to contiguous2.4-kb and 5.2-kb Xho I fragments of thewild-type plasmid (lane 1). When digestsof mutant plasrnids M-2-5 and M-1-13 areprobed (lanes 2 and 3), nodD sequenceshybridize to either the 2.4-kb Xho I wild-type fragment (lane 2) or the 5.2-kb XhoI fragment (lane 3) and to one TnS-p-

1 2 3 PRC6 junction fragment.

1 2 3 4 5 6 7 8 9 10 11

-5.2 kb

-2.4 kb

* -1.k__~~~~~PW * w 1.5 kb

mk

LaneII-12345678910ns

StrainRp5011-242-91-141-291-251-211-132-51-282-122-14

- 0.8 kb

X/10I Fragments5.2-+ 2.45.2 + 2.7 + 0.74.3 + 1.9 + 2.44.6 + 1.6 + 2.45.6 + 0.6 + 2.45.4 + 0.8 + 2.44.1 + 2.1 + 2.44.8 + 1.4 + 2.45.2 + 2.9 + 0.53.8 + 2.4 + 2.44.0 + 2.2 + 2.45.2 + 2.5 + 0.9

FIG. 3. Rp501 fragments containing TnS have replaced wild-typefragments in mutant Rp501 genomes. Southern blot hybridization ofa radiolabeled pPRC6 probe to Xho I-digested DNAs from 10Bradyrhizobium sp. (Parasponia) Rp501 strains in which wild-typefragments have been replaced by homologous fragments containingTnS insertions as a consequence of a double homologous recombi-nation event. All insertions map to either a 5.2-kb or a 2.4-kbwild-type Xho I fragment. Since Xho I cleaves within TnS, in eachcase (lanes 1-10), one ofthese wild-type fragments is replaced by twonew fragments. The table below the autoradiogram of the hybridizedblot shows expected sizes of newly generated fragments in eachstrain. Note that vector sequences from pPRC6::Tn5 plasmids usedto create these strains are not retained (such vector sequences wouldgenerate a >20-kb Xho I fragment). Size markers alongside theautoradiogram denote Xho I fragments cloned entirely in pPRC6. ns,Not shown.

elicited nodules, including Rp2-9, also elicited a typical roothair curling response with shepherds' crooks. All of thenonnodulating mutants could be complemented with wild-type pPRC6 and, in all cases, Fix' nodules were elicited.When the mutant Rp501 derivatives were assayed on

cowpea, another root hair infection host, for their capacity toelicit effective nodules, results were similar to those observedin siratro nodulation assays, with two exceptions (Table 2).Mutant Rp2-9, which weakly nodulated siratro, did not elicitnodules on cowpea. More strikingly, one of the putative

Genetics: Marvel et al.

Dow

nloa

ded

by g

uest

on

Apr

il 25

, 202

0

Proc. Natl. Acad. Sci. USA 84 (1987)

Table 2. Nodulation phenotypes of Bradyrhizobium sp.(Parasponia) mutants tested on different plant hosts

Siratro* Cowpeat ParasponiatMutant Homologue§ Nod Hac Fix Nod Fix Nod

Rpl-29 ND + + + + + -Rpl-25 ND + + + + + +Rpl-14 ND + + + + + +Rp2-9 nodC ± + - - - +Rpl-21 nodCRp2-12 nodCRpl-28 nodABRpl-13 ND + + + + + +Rp2-5 nodD + + + + + +Rpl-24 nodD + + + + - +Rp2-14 ND + + + + + +Rp501 + + + + + +

Nod, nodulation; +, wild-type nodulation frequency; -, nonodules observed; ±, <20% wild-type nodulation frequency. Fix,nitrogen fixation measured in acetylene reduction assays; +,90-100% of wild-type acetylene reduction levels; -, <10% ofwild-type acetylene reduction levels. Hac, root hair curling; +,wild-type root curling; -, absence of root hair curling. ND, notdetermined.*Ten plants per mutant inoculum were assayed in each trial. Eachmutant was assayed in at least two independent trials.

tSix plants per mutant inoculum were assayed in each trial. Eachmutant was assayed in at least two trials.fAt least two plants per mutant inoculum were assayed in each trial.Each mutant was assayed in at least two trials.§Location of Tn5 insertion in Rp501 sequences homologous toindicated R. meliloti nodulation genes as determined by hybridiza-tion and/or complementation analyses (5, 23).

nodD mutants, Rpl-24, was unable to nodulate cowpeasefficiently. As in Rp2-5 (see Fig. 2), in Rpl-24 the region ofhomology to the nodD probe is interrupted by the TnSinsertion. In several assays, this strain elicited only one ortwo nodules per plant, which developed up to 2 wk later thanRp501-elicited nodules. Wild-type strains commonly elicited20 or more nodules per plant. Mock-inoculated controls werenot nodulated.

Since Parasponia is infected by a different pathway thancowpeas or siratro, nodulation assays on this host using themutant strains were ofgreat interest. Each mutant Rp501 strainwas assayed on at least two aseptically grown Parasponiaseedlings per trial, and the results presented in Table 2 werederived from two separate inoculation trials. Results of thenodulation assays were consistent between both trials. Mutantstrains Rpl-21, Rp2-12, and Rpl-28 were unable to elicitnodules on Parasponia seedlings, as they had been unable toelicit cowpea and siratro nodules. Mutant Rp2-9 weaklynodulated Parasponia seedlings, as it had siratro.The mutant strain Rpl-29, which vigorously generated

nodules on the other Rp501 hosts, was incapable of elicitingParasponia nodules. The mutation in Rpl-29 may inactivatea unique function involved specifically in nonlegume nodula-tion. All of the other mutants listed in Table 2 nodulatedParasponia seedlings and elicited an average of one noduleper seedling. Rp501 derivatives with mutations in the nodDhomologous locus readily nodulated Parasponia. Noduleselicited by the mutant strains reduced acetylene at variablelevels, perhaps because Parasponia nodules develop asyn-chronously, and fixation levels may be related to nodule age.

DISCUSSIONOur studies with Bradyrhizobium sp. (Parasponia) Rp501nodABC::TnS mutants and Parasponia seedlings show thatthe common nodulation loci defined by mutant strainsRpl-28, Rp2-12, and Rpl-21 are required for legume andnonlegume nodulation. Therefore, the role of these genes in

nodulation seems essential and universal. In addition, theRp501 nodABC::Tn5 mutants failed to elicit root hair defor-mations when assayed on siratro and cowpeas, hosts that areinfected primarily through root hairs. This substantiatesresults obtained with certain mutants of a slow-growingcowpea Rhizobium strain, IRC78, which nodulate neithercrack entry nor root hair entry hosts (9).The mutant Rp2-9, a nodC mutant according to physical

mapping and to interspecies complementation tests (5, 23),elicited a few ineffective nodules on siratro, although thisstrain did not nodulate cowpeas. Promoter activity within theTn5 insertion in Rp2-9, and partial activity of a truncatedprotein may account for these observations.Two mutant Rp501 strains containing Tn5 insertions in a

region of homology to the R. meliloti nodD gene failed todefine a clear role for this locus in Rp501-elicited nodulation.Although the two mutants exhibited wild-type phenotypeswhen assayed on siratro and Parasponia seedlings, interest-ingly, one of these mutant strains, Rpl-24, exhibited a muchless efficient, delayed nodulation phenotype, relative toRp501, when assayed on cowpeas. Thus, this locus may havehost-specific effects, and perhaps this mutation differentiallyaffects cowpea nodulation by diminishing a factor cowpeasrequire at a higher level than do other host species. However,the possibility that a second, cowpea specific mutation existsat another locus in Rpl-24 has not yet been eliminated.Alternatively, nodD may be essential for nodulation byRp501, but, as postulated for R. meliloti, extra functionalcopies of nodD may reside elsewhere in the Rp501 genome(20). In a closely related Bradyrhizobium sp. (Parasponia)strain (ANU289), Scott has shown that the cloned Bradyrhi-zobium sp. (Parasponia) nodD-like gene complements Rhi-zobium trifolii nodD mutants (24).When these studies were initiated we hoped to learn if

legume-specific or nonlegume-specific nodulation genes ex-isted in the Rp501 genome. The mutant Rpl-29 nodulatessiratro and cowpea, but not Parasponia, which suggests thatits TnS lesion affects a nonlegume-specific nodulation factor.Although a single TnS insertion mutation may behave in ananomalous fashion, we have defined a genetic differencemanifested in this mutant strain relative to wild type, whichindicates that a Parasponia-specific Bradyrhizobium nodula-tion function exists.Because the common nodulation genes (nodABC) are

required for all modes ofRhizobium infection of plant roots-root hair invasion, crack entry, and Parasponia infection-and because the nodDABC genes of R. meliloti in anAgrobacterium background are sufficient to elicit an orga-nized nodule, albeit devoid of bacteria (25), it is likely thatnodDABC products are involved in signaling the plant host toinitiate nodule development. Similar plant developmentalprocesses are likely to be involved in nodulation of aspectrum ofwoody dicots by the actinomycete Frankia, sincethe Parasponia infection pathway also prevails in somewoody nonlegumes nodulated by Frankia (26). Also, excep-tional legumes, such as Sesbania (stem nodulated), appear toshare the Parasponia infection pathway (27). Thus, noduledevelopment in Parasponia is not an isolated phenomenon;instead, Parasponia nodules resemble both Rhizobium-in-duced nodules of legumes and Frankia-induced nodules ofnonlegumes and are considered to represent an intermediateform with respect to these symbiotic associations (28). Thegenetic basis for this alternative infection pathway in any ofthese systems has hitherto been unexplored.That the nodABC genes are involved in Parasponia infec-

tion, as well as typical legume nodule development, mightseem a surprising result in view of differences in noduledevelopmental processes between most legumes and non-legumes and lack of any data defining a genetic basis foratypical infection phenomena. We do not know yet if legume

1322 Genetics: Marvel et al.

Dow

nloa

ded

by g

uest

on

Apr

il 25

, 202

0

Proc. Natl. Acad. Sci. USA 84 (1987) 1323

and nonlegume nodulation processes are mechanisticallyidentical. Defining a role for nodABC in the Parasponiainfection pathway does not preclude the involvement of otherRhizobium genes specifically in the nonlegume nodulationprocess-in fact, our results indicate that nonlegume-specificfactors do exist. Clearly, common features are shared by hostplants of widely divergent genera, which permit a response tothe same Rhizobium nodulation functions. We can ask whichcommon features shared by Parasponia trees and legumesfacilitate the nodulation response. The fact that this atypicalassociation exists may offer a key to the perplexing questionof why nitrogen fixation in Rhizobium symbioses is almostentirely confined to legumes.

We thank R. Hyde for graciously preparing the manuscript and M.Honma and K. Wilson for helpful discussion. This work wassupported by a grant from Hoechst AG to Massachusetts GeneralHospital.

1. Trinick, M. J. (1973) Nature (London) 244, 459-460.2. Lancelle, S. A. & Torrey, J. G. (1984) Protoplasma 123,

26-37.3. Lancelle, S. A. & Torrey, J. G. (1985) Can. J. Bot. 63, 25-35.4. Marvel, D. J., Kuldau, G., Hirsch, A. M., Park, J., Torrey,

J. G. & Ausubel, F. M. (1984) in Advances in Nitrogen Fixa-tion Research, eds. Veeger, C. & Newton, W. E. (Nijhoff, TheHague), p. 691.

5. Marvel, D. J., Kuldau, G., Hirsch, A., Richards, E., Torrey,J. G. & Ausubel, F. M. (1985) Proc. Natl. Acad. Sci. USA 82,5841-5845.

6. Long, S. R., Buikema, W. J. & Ausubel, F. M. (1982) Nature(London) 298, 485-488.

7. Schofield, P. R., Djordjevic, M. A., Rolfe, B. G., Shine, J. &Watson, J. M. (1983) Mol. Gen. Genet. 192, 459-465.

8. Downie, J. A., Knight, C. D., Johnston, A. W. B. & Rossen,L. (1985) Mol. Gen. Genet. 198, 255-262.

9. Noti, J. D., Dudas, B. & Szalay, A. A. (1985) Proc. NatI.

Acad. Sci. USA 82, 7379-7383.10. Peters, N. K., Frost, J. W. & Long, S. R. (1986) Science 233,

977-980.11. Hirsch, P. R. (1978) Dissertation (University of East Anglia,

Norwich, England).12. Ditta, G., Stanfield, S., Corbin, D. & Helinski, D. R. (1980)

Proc. Natl. Acad. Sci. USA 77, 7347-7351.13. Meade, H. M., Long, S. R., Ruvkun, G. B., Brown, S. E. &

Ausubel, F. M. (1982) J. Bacteriol. 49, 114-122.14. Ruvkun, G. B. & Ausubel, F. M. (1981) Nature (London) 289,

85-89.15. Jagadish, M. N. & Szalay, A. A. (1984) Mol. Gen. Genet. 196,

290-300.16. Ruvkun, G. B., Sundaresan, V. & Ausubel, F. M. (1982) Cell

29, 551-559.17. Buikema, W. J., Long, S. R., Brown, S. E., van den Bos,

R. C., Earl, C. D. & Ausubel, F. M. (1983) J. Mol. Appl.Genet. 2, 249-260.

18. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) MolecularCloning: A Laboratory Manual (Cold Spring Harbor Labora-tory, Cold Spring Harbor, NY).

19. Vogelstein, B. & Gillespie, D. (1979) Proc. Natl. Acad. Sci.USA 76, 615-619.

20. Egelhoff, T. T., Fisher, R. F., Jacobs, T. W., Mulligan, J. T.& Long, S. R. (1985) DNA 4, 241-248.

21. Gibson, A. H. (1980) in Methods for Evaluating BiologicalNitrogen Fixation, ed. Bergersen, F. J. (Wiley, New York),pp. 139-184.

22. Mulligan, J. T. & Long, S. R. (1985) Proc. Natl. Acad. Sci.USA 82, 6609-6613.

23. Marvel, D. J. (1986) Dissertation (Harvard University, Cam-bridge, MA).

24. Scott, K. F. (1986) Nucleic Acids Res. 14, 2905-2919.25. Hirsch, A. M., Drake, D., Jacobs, T. W. & Long, S. R. (1985)

J. Bacteriol. 161, 223-230.26. Miller, I. M. & Baker, D. D. (1985) Protoplasma 128, 107-119.27. Tsien, H. C., Dreyfus, B. L. & Schmidt, E. L. (1983) J.

Bacteriol. 156, 888-897.28. Newcomb, W. E. (1981) Int. Rev. Cytol. Suppl. 13, 247-298.

Genetics: Marvel et al.

Dow

nloa

ded

by g

uest

on

Apr

il 25

, 202

0