y . Cell Sci. Suppl. 2, 347-354 (1985)Printed in Great Britain © Company of Biologists Limited 1985 347

R H IZ O B IU M LE G U M IN O SA R U M GENES INVOLVED

IN EARLY STAGES OF N O D U L A T IO N

J. A. DO W N IE*, L. R O SSE N , C. D . K N IG H T f, J. G. R O B E R T SO N !, B. W ELLS a n d A. W. B. JO H N ST O N John Innes Institute, Colney Lane, Norwich NR4 7UH, U.K.

SUMMARYNodulation genes IromRhizobium leguminosarum have been subcloned and transferred to a strain

of R. phaseoli with its symbiotic plasmid deleted (and therefore its nodulation and nitrogen fixation genes). Normal infection and nodule development occurred when these strains were added to the roots of Pisum sativum (peas) and Vicia hirsuta. The pea nodules were examined by electron microscopy; bacteroid forms were seen surrounded by peribacteroid membranes and using immuno-gold labelling it was shown that the nodule cells contained leghaemoglobin within the cytoplasm. By subcloning and analysing the nodulation region it was found that a small (3-2 X 103 bases) fragment of DNA contained three genes involved in root hair curling, the first observed step in the interaction between R. leguminosarum and the root hair cells of these legumes.

I N T R O D U C T I O NAs described by Robertson et al. (1985, this volume) the overall process of nodula­

tion of leguminous plants by Rhizobium can be described in terms of an interaction between the cell surfaces of two very different organisms. This programme of joint and complex differentiation extends from the early stages of infection through to the mature nitrogen-fixing root nodule and even at this stage the bacteria can be con­sidered to be topologically outside the plant nodule cells. It is self-evident that bacteria of the genus Rhizobium have genes that are specifically required for the induction of nodules, but within the rhizobia there is a level of specificity that limits the number of plant species nodulated by any one species of Rhizobium. This specificity may occur at two stages at least: in some cases plants inoculated with a heterologous species of Rhizobium do not form nodules (e.g. R. meliloti does not nodulate peas); in other cases nodules can be induced but do not fix nitrogen (e.g. R. leguminosarum forms non­fixing nodules on subterranean clover). At present we do not understand the biochemistry of the interaction of rhizobia with legumes, but there has been sig­nificant progress in the identification of the bacterial genes involved in the initial steps of this interaction. In this paper we will describe the genetic approach that has been used to elucidate the early steps involved in the nodulation of peas or Vicia hirsuta by R. leguminosarum; for a more general review of the genetics of nodulation in Rhizobium, see Long (1984).

* Author for correspondence.fPresent address: Department of Genetics, University of Leeds, U.K.JPresent address: DSIR, Palmerston North, New Zealand.

348 jf. A. Downie and othersN O D U L A T IO N G E N E S OF R. LEGUMINOSARUM ARE ON P L A S M ID S

It has been established by both physical and genetic methods that in the fast- growing species of Rhizobium large indigenous plasmids carry the nitrogen fixation and nodulation genes (see Long, 1984). One such ‘symbiotic’ plasmid, pRLlJI, has been studied extensively in this laboratory and was shown to confer the ability to nodulate peas upon strains of R. trifoli and R.phaseoli (Johnston et al. 1978). Mutant strains of R. leguminosarum (and R. trifolii and R. phaseoli) have been isolated that have lost their symbiotic plasmid and all have lost the ability to form nodules or to induce any of the early steps such as root hair curling. Such strains, ‘cured’ of their symbiotic plasmids should provide ideal material for testing whether particular assay- able biochemical functions that have been proposed to be important for nodulation have been lost along with the plasmid. Thus, for example, several strains have been examined with regard to production of plant growth hormones (Badenoch-Jones et al. 1982; Wang, Wood & Brewin, 1982a,b), pectinases and cellulases. To date such studies have been unsuccessful; the phenotypes of the mutants have been identical to those of the parent strains.

These cured strains of Rhizobium have been used in genetic studies to show that the particular host range of a given strain is specified by its symbiotic plasmid. Using strains of Rhizobium with their symbiotic plasmids deleted it is possible to construct a series of strains that have an identical chromosomal genome but differ only in that they have inherited a symbiotic plasmid specific for R. leguminosarum, R. phaseoli or R. trifolii (Brewin et al. 1982; Lamb, Hombrecher & Johnston, 1982; Djordjevic, Zurkowski, Shine & Rolfe, 1983). In each case normal nitrogen-fixing nodules have been formed on the appropriate host. However, it does not appear that the genetic background of the recipient strain is also important in determining normal phenotypic expression of the introduced plasmid since a strain of Agrobacterium tumefaciens containing a symbiotic plasmid from R. trifolii formed aberrant nodules on clover (Hooykaas et al. 1981).

The basis of the genetic and biochemical nature of host specificity may be understood by comparison with otherwise isogenic strains of Rhizobium that can nodulate different host legumes. One such comparison studied recently involves the comparison of two strains of R. leguminosarum, only one of which can nodulate the primitive pea cultivar Afghanistan. This host is not nodulated by conventional strains of R. leguminosarum from Western Europe or North America. However, a strain called TOM from Turkey can nodulate this primitive pea in addition to commercial pea cultivars (Lie, 1978). Strain TOM contains symbiotic plasmid, pRL5JI, which when transferred to conventional R. leguminosarum strains, conferred upon the recipient strain the extra ability to nodulate cv. Afghanistan peas (Brewin, Beringer & Johnston, 1980). One attractive feature of this system is that the genetics of the ‘resistance’ to nodulation is determined by a single recessive gene in cv. Afghanistan, and it is therefore possible to study this aspect of host range using a genetic approach in both R. leguminosarum and the plant host (Holl, 1975).

Nodulation genes from Rhizobium 349Although symbiotic plasmids in different Rhizobium species determine the par­

ticular host range of the strain, it is now clear that certain nodulation genes are functionally equivalent in Rhizobium species with different host ranges. This was first shown by Banfalvi et al. (1981), who transferred the R. leguminosarum symbiotic plasmid pRLlJI into non-nodulating mutant strains of R. meliloti. The resulting transconjugants regained the ability to nodulate alfafa, although the introduced sym­biotic plasmid came from a species that nodulates peas. This functional equivalence is reflected in the striking similarity in the DNA sequences of certain nod genes found in R. leguminosarum and R. meliloti (see below).

One further point to emerge from studies in which a symbiotic plasmid is trans­ferred from one species of Rhizobium to another, is that in the ‘hybrid’ the presence of one symbiotic plasmid can prevent the phenotypic expression of the host range specified by the other. When the R. leguminosarum plasmid pRLlJI was transferred to a strain of R. phaseoli, the transconjugants could maintain both symbiotic plasmids through many generations. When the bacteria containing both plasmids were inoculated onto peas, nodules were formed but, significantly, the bacteria isolated from such pea nodules had, without exception, either lost the resident R. phaseoli symbiotic plasmid entirely or there were large deletions in that plasmid (Beynon, Beringer & Johnston 1980; A. W. B. Johnston, unpublished observations). Thus it would appear that the presence of the resident R. phaseoli symbiotic plasmid prevented the functional expression of pRLlJI, the R. leguminosarum symbiotic plasmid. It is not clear if this phenomenon is due to actual repression of the R. leguminosarum nodulation genes mediated by the R. phaseoli symbiotic plasmid. Alternatively, it is possible that these ‘hybrid’ strains express host-range determinants for the infection of both peas and Phaseolus beans; however, in this model the production of the biochemical determinants normally required for development of Phaseolus bean nodules would cause the infections on pea roots to abort. Therefore, there would be selection for those strains that had lost the Phaseolus bean nodulation determinants.

/ i ........mi........... pi J 1089pIJ 1085 ' ' ................ / ,

nodulation genesI I I I£coRI Bgl II EcoRI EcoRI ' pIJ1216

-------------- pIJ1389► ► — ►

nod A B C

Fig. 1. Map of the nodulation region of R. leguminosarum. Plasmids pIJ 1085 and pljlt)89 were described previously (Downie et al. 1983a) and each contains about 3 X 104 base- pairs of DNA from pRLlJI the native R. leguminosarum plasmid. The nodulation genes are contained within the 104-base-pair region of DNA common to both plasmids. The restriction endonuclease sites iicoRI and Bglll used to subclone the region of DNA in plasmids pIJ 1216 and pIJ 1389 are indicated. The extent and location of nodAB and C genes are indicated.

350 J . A. Downie and others

fni eg6.

Fig. 2. Root hair curling and infection threads induced by strains of Rhizobium contain­ing cloned DNA fragments. Roots of P. sativum ( a ) and V. hirsuta ( b , c ) were examined as described by Fahraeus (1957) by light microscopy using Nomarski optics. In a and B, roots were inoculated with Rhizobium strain 8401 containing p l j 1089 and inB an infection thread can be observed in the rightmost root hair. In c, the roots were inoculated with Rhizobium strain 8401 containing p lj 1389 and root hair deformation can be seen, a and c, x 160; b , x 320.

It is clear from these studies on Rhizobium plasmids that several plasmid-borne genes exist that are involved in the formation of nodules: some of these genes appear to be common to different Rhizobium species, some genes are specific for the types of host plants to be nodulated and others may in some way regulate or block the ex­pression of nodulation-specific determinants. In the following section we will outline current approaches in the molecular biology of Rhizobium, which are aimed at defin­ing these genes. In particular we will focus on genes that are specifically involved in root hair curling, the first observed step in the interaction between/?, leguminosarum and its legume hosts.

C L O N I N G T H E N O D U L A T I O N A N D H O S T - R A N G E G E N E S

With the advent of cloning vectors that are transmissible from Escherichia coli to Rhizobium (Bagdasarian et al. 1981; Friedman et al. 1982) the genetic and physical analysis of the nodulation genes has been greatly facilitated. Coupled with the use of transposons as mutagens, nodulation-specific genes have been identified and the corresponding wild-type DNA cloned and analysed. The nodulation genes from a strain of R leguminosai~um were cloned on a broad host-range plasmid and shown to be able to confer upon a strain of R. phaseoli (strain 8401) cured of its symbiotic plasmid, the ability to nodulate peas (Downie et al. 1983a). As shown in Fig. 1, two clones (called pi J 1085 and pi J 1089) were identified and these contained only about 104 base-pairs of DNA in common; within this 104-base-pair region of DNA the nodulation genes have been identified.

It appeared that the initial events of infection and nodulation occurred normally in plants inoculated with Rhizobium strain 8401 containing p lj 1085 or p l j 1089. Fig. 2a

Nodulation genes from Rhizobium 351

Fig. 3. Transmission electron micrograph (X 35 000) of a thin section of a nodule from P. sativum (var. Wisconsin Perfection) formed by infection of the root by Rhizobium strain 8401 containing p lj 1085 (a similar result was observed with p lj 1089). Nodules were fixed in glutaraldehyde and osmium, dehydrated in ethanol according to Beringer, Johnston & Wells (1977) and embedded in LR white resin. Thin sections were treated with periodate/H Cl (Craig & Goodchild, 1984) before immuno-gold staining for leghaemo- globin (Robertson et at. 1984). Specific staining for leghaemoglobin occurred only in the plant cytoplasm of nodule cells, be, bacterial cytoplasm; pbm, peribacteroid membrane; pc, plant cytoplasm.

shows a normal curled root hair formed on pea roots by the strain of Rhizobium containing plasmid p lj 1089. The development of a normal infection thread on the roots of V. hirsuta is shown in Fig. 2b and nodules on peas or V. hirsuta were formed at about the same frequency as with the wild-type strain. However, since the Rhizobium strain 8401 was cured for its symbiotic plasmid (and therefore had lost genes involved in nitrogen fixation) and pi J 1089 lacks several fix genes (Downie et al. 1983b) the nodules formed did not fix nitrogen. Nevertheless, the release of bacteria from the infection thread appeared to be normal and bacteroid forms were found to be enclosed within the peribacteroid membrane (Fig. 3). The major dif­ference in these nodules at the ultrastructural level was that they accumulated starch granules and senesced early, within about 7 days of formation of the nodules (Downie

352 J. A. Downie and otherset al. 1983a). Early in the development of the nodules induced by strain 8401 contain­ing pIJ1085 or pIJ1089 they were observed to be slightly pink indicating that the plant-encoded nodulin, leghaemoglobin, may be present. Spectrophotometric analysis of extracts of these nodules confirmed that leghaemoglobin was present although at only about 40% of the normal level. The presence and normal location of leghaemoglobin in nodules induced by strain 8401 containing p lj 1085 or p lj 1089 was confirmed by immuno-gold labelling (Robertson et al. 1984) of thin sections of nodules fixed with glutaraldehyde and embedded in LR white resin. As shown in Fig. 3, the leghaemoglobin is confined to the plant cell cytoplasm, as has been observed previously by Robertson et al. (1984).

For all the complexity of the nodulation process, it is now clear that relatively few genes in the symbiotic plasmid of R. leguminosarum are required for host-range specificity, development of nodules and bacteroids and the induction of at least some plant-encoded nodulins. Clustering of nodulation genes has also been found in other Rhizobium species. Thus, transfer of a relatively small region (1-4 X 104 base-pairs) of the symbiotic plasmid of R. trifolii to other species allowed the transconjugants to nodulate clover (Schofield et al. 1984).

ROOT HA IR C U R L IN G G E N E SAs seen above, the nodulation genes of pRLlJI were located within a 104-base-pair

region of the symbiotic plasmid, which was defined by the overlap between recom­binant plasmids p lj 1085 and pIJ1089. We have concentrated on characterizing the genes that are required for root hair curling, one of the first observable steps in the infection process. This 10+-base-pair nodulation region of pRLlJI comprises two adjacent iscoRI fragments of 6-6 and 3-2 X 103 base-pairs (Fig. 1). Each of these was separately cloned in the wide host-range cloning vector pKT230 and introduced into the strain of Rhizobium that lacked its native symbiotic plasmid. The strain containing the cloned 6-6 X 103 base-pair fragment (pIJ1216) was capable of inducing root hair curling on peas but true nodules were not formed. Thecloned3-2 X 103 base-pair £coRI fragment induced no root hair deformation on peas. The finding that the 6-6 X 103 base-pair iscoRI fragment contained the root hair curling genes was consistent with other studies in which it was found that Tn5 mutations, which abolished the ability to nodulate and to curl root hairs were located within this 6-6 X 103 base-pair fragment (Downie et al. 1983a). From this region a 3-2 X 103 base-pair Bgl\\-EcoY!A fragment was subcloned (Fig. 1) into the vector pKT230 to form plasmid p lj 1389. When this plasmid was transferred to the strain olR.phaseoli cured of its symbiotic plasmid, the resulting transconjugant acquired the ability to induce root hair curling/deformation on peas (Fig. 2c).

Therefore, it seems probable that very few genes are involved in this initial step in the interaction of Rhizobium with its host plant. In order to define the genes involved, the region of DNA on p lj 1389 was sequenced and three genes were identified (Rossen, Johnston & Downieb1984). These genes were called nodA, nodB and nodC and the extent of these genes is indicated in Fig. 1. In parallel with these studies in

R. leguminosarum, similar work has been carried out with R. meliloti nod genes and three similar genes have been identified by DNA sequencing (Torok et al. 1984). It appears that these genes are highly conserved between different species of Rhizobium and can be considered to be at least some of the common nodulation genes.

Thus at this stage we have a very detailed picture of the Rhizobium genes required for root hair curling. This information does not by itself lead to a further understand­ing of the mechanism of induction of root hair curling by Rhizobium, but an analysis of the mode of action of these gene products should give a considerable insight into the mechanism of the induced plant response.

We thank Dr T. Bisseling (University of Wageningen) for the gift of antibody to leghaemoglobin. This work was supported in part by the EEC (contract no. GBI-5-070-UK); L.R. was supported by a Scientific and Technical Training Grant from the commission of European Communities and Danish Agricultural and Vetinary Research Council; J.G.R. was the recipient of an award from the John Innes Trust and a study award from DSIR, New Zealand.

Nodulation genes from Rhizobium 353

R E F E R E N C E SB a d e n o c h -J o n e s , J., S im m o n s , R . E., D j o r d je v il , M. A., S h i n e , J., L e t h a m , D . D . & R o l f e ,

B. G. (1982) Mass spectrometric quantification of indole-3-acetic acid in Rhizobium culture supernatants: relation to root hair curling and nodule initiation. Appl. environ. Microbiol. 44, 275-280.B a g d a s a r ia n , M., L u r z , R . , R u c k e r t , B ., F r a n k l in , F . H. C., B a g d a s a r ia n , M. M., F r e y ,

J. & Timmis, K. N. (1 9 8 1 ) . Specific purpose cosmid cloning vectors. II. Broad host-range high copy number R S F 1 0 1 0 derived vectors and a host-vector system for gene cloning in Pseudomonas. Gene 16, 237—247.

B a n f a l v i , Z., S a k a n y a n , V., K o n c z , C., K i s s , A., D u s h a , I. & K o n d o r o s i , A. (1981). Location of nodulation and nitrogen fixation genes on a high molecular weight plasmid of R. meliloti. Molec. gen. Genet. 184, 318-325.

B e r in g e r , J . E., J o h n s t o n , A. W . B . & W e l l s , B . (1977). The isolation of conditional ineffective mutants of Rhizobium leguminosarum. J . gen. Microbiol. 98, 339-343.B e y n o n , J. L., B e r in g e r , J. E. & J o h n s t o n , A. W. B . (1980). Plasmids and host-range in

Rhizobium leguminosarum and Rhizobium phaseoli. jf. gen Microbiol. 120, 421-429.B r e w in , N. J., B e r in g e r , J. E. & J o h n s t o n , A. W. B . (1980). Plasmid mediated transfer of host-

range specificity between two strains of Rhizobium leguminosarum. J . gen. Microbiol. 120, 413-420.B r e w in , N. J., W o o d , E. A., J o h n s t o n , A. W . B ., D ib b , N. J . & H o m b r e c h e r , G. (1982).

Recombinant nodulation plasmids in Rhizobium leguminosarum. J . gen. Microbiol. 128, 1817-1827.C r a ig , S. & G o o d c h il d , D. J. (1984). Periodate-acid treatment of sections permits on-grid

immunogold localization of pea seed vicillin in ER and Golgi. Protoplasma 122, 35-44.D jo r d je v ic , M. A., Z u r k o w s k i, W., S h i n e , J. & R o l f e , B. G. (1983). S y m p la s m id t r a n s f e r to

v a r io u s sy m b io t ic m u ta n ts o f Rhizobium trifolii, R. leguminosarum a n d R. meliloti. J . Bact. 156, 1035-1045.D o w n ie , J . A., H o m b r e c h e r , G., M a , Q-S., K n i g h t , C. D . , W e l l s , B. & J o h n s t o n , A. W .

B. (1983a). Closed nodulation genes of Rhizobium leguminosarum determine host-range specific­ity. Molec. gen. Genet. 190, 359-365.D o w n ie , J. A., M a , Q-S., K n i g h t , C. D . , H o m b r e c h e r , G. & J o h n s t o n , A. W. B. (19836).

Cloning of the symbiotic region of Rhizobium leguminosarum; the nodulation genes are between the nitrogenase genes and a nifA-Vikc gene. EMBOJ. 2, 947-952.F a h r a e u s , G. (1957). The infection of clover root hairs by nodule bacteria. Studies by a simple glass slide technique. J. gen. Microbiol. 16, 374-381.

354 y. A. Downie and othersF riedman , A . M., L ong , S. R., B row n , S. E ., B uikema , W. J. & A usu bel , F . M. (1982).

Construction of a broad host-range cosmid cloning vector and its use in the genetic analysis of Rhizobium mutants. Gene 13, 227-237.

H oll, F. B. (1975). H ost plant control of the inheritance of di-nitrogen fixation in the Pisum-Rhizobium symbiosis. Euphytica 3, 767-770.

H ooykaas, P. J. J., V an B russel , A. A. N., D en D ulk -R as , H ., V an Slogteren, G., M. S. & Schilperoort, R . A. (1981). Sym plasmid of Rhizobium trifolii expressed in different Rhizobial species and Agrobacterium tumefaciens. Nature, Lond. 291, 351-353.

J o h n s to n , A. W. B ., B ey n o n , J . L., B u c h a n a n -W o lla s to n , A. V., S e tc h e l l , S. M., H irsc h , P. R . & B e r in g e r , J . E. (1978). H igh frequency transfe r of nodu la tion ability be tw een strains an d species o f Rhizobium. Nature, Lond. 276, 635-636.

L amb, J . W., H ombrecher, G. & J ohnston , A. W. B. (1982). Plasmid-determined nodulation and nitrogen-fixation abilities in Rhizobium phaseoli. Molec. gen. Genet. 186, 449-452.

L ie , T. A. (1978). Symbiotic specialisation in pea plants. Ann. appl. Biol. 88, 462-465.L ong , S. R. (1984). Genetics of Rhizobium nodulation. In Plant-Microbe Interactions:

Molecular and Genetic Perspectives, vol. 1 (ed. T. Kasuge & E. W. Nester), pp. 265-306. New York, London: Macmillan.

R obertson, J . G., W el ls , B ., B isseling , T., F a rn den , K. J. F . & J ohnston , A. W . B . (1984). Immunogold localization of leghaemoglobin in cytoplasm in nitrogen-fixing root nodules of pea. Nature, Lond. 311, 254-256.

R obertson, J. G., W el ls , B . ,B r ew in , N .J., W ood , E., K night, C. D . & D o w n ie , J. A. (1985). The legume-Rhizobium symbiosis. A cell surface interaction. J. Cell Sci. Suppl. 2, 317-331.

R o ssen , L., J ohnston , A. W. B. & D ow n ie , J . A. (1984). DNA sequence of the Rhizobium leguminosarum nodulation genes nodAB and C required for root hair curling. Nucl. Acid Res. 12, 9497-9508.

S c h o f ie ld , P. R ., R id g e , R . W., R o lf e , B. G ., S h in e , J. & W a tso n , J. M. (1984). Host-specific nodulation is encoded on a 14kb DNA fragment in Rhizobium trifolii. PI. molec. Biol. 3 ,3-11.

T orok, I., K ondorosi, E., Stepkowski, T ., P osfai, J. & K ondorosi, A. (1984). Functional delimitation and nucleotide sequence of Rhizobium meliloti common nod genes. Nucl. Acids Res. 12, 9509-9524.

W a n g , T. L., W ood , E. A. & B r ew in , N. J. (1982a). Growth regulators, Rhizobium and nodulation in peas. Indole-3-acetic acid from the culture medium of nodulating and non- nodulating strains of R. leguminosarum. Planta 55, 345-349.

W a n g , T, L., W o od , E. A. & B r ew in , N. J. (1982b). Growth regulators, Rhizobium and nodulation in peas. The cytokinin content of a wild-type and a Ti-plasmid containing strain of R. leguminosarum. Planta 155, 350-355.