Symbiotic and Genetic Diversity of Rhizobium galegae Isolates Collected from the Galega orientalis Gene Center in the Caucasus

Abstract

This paper explores the relationship between the genetic diversity of rhizobia and the morphological diversity of their plant hosts. Rhizobium galegae strains were isolated from nodules of wild Galega orientalis and Galega officinalis in the Caucasus, the center of origin for G. orientalis. All 101 isolates were characterized by genomic amplified fragment length polymorphism fingerprinting and by PCR-restriction fragment length polymorphism (RFLP) of the rRNA intergenic spacer and of five parts of the symbiotic region adjacent to nod box sequences. By all criteria, the R. galegae bv. officinalis and R. galegae bv. orientalis strains form distinct clusters. The nod box regions are highly conserved among strains belonging to each of the two biovars but differ structurally to various degrees between the biovars. The findings suggest varying evolutionary pressures in different parts of the symbiotic genome of closely related R. galegae biovars. Sixteen R. galegae bv. orientalis strains harbored copies of the same insertion sequence element; all were isolated from a particular site and belonged to a limited range of chromosomal genotypes. In all analyses, the Caucasian R. galegae bv. orientalis strains were more diverse than R. galegae bv. officinalis strains, in accordance with the gene center theory.

Rhizobium galegae (14) is a species that forms an effective symbiosis with plants of Galega orientalis and Galega officinalis, the only species in the genus Galega (Fabaceae) that have been studied for symbiosis. This symbiotic system represents a sharply defined cross-inoculation group. However, there are some differences in symbiotic performance. R. galegae strains are able to infect both Galega species, but strains isolated from G. officinalis form effective nodules on that plant and ineffective nodules on G. orientalis, while the converse is true for strains from G. orientalis (16). This finding, along with a wide range of phenotypic and genotypic approaches using numerical taxonomy (15), phage typing, DNA homology (13, 40), lipopolysaccharide and protein patterns (16), plasmid profiling (29), phylogeny of ribosomal genes (25, 35), randomly amplified polymorphic DNA, and repetitive PCR (21, 28), led to the proposal of two biovars for strains forming an effective symbiosis with G. officinalis (R. galegae bv. officinalis) and G. orientalis (R. galegae bv. orientalis) (24). Symbiosis-related genetic traits were found to be the main factor in genetic divergence between the biovars (24). G. orientalis is a good fodder plant with some widely used cultivars, whereas G. officinalis is rather poisonous because of its high alkaloid content and has no agricultural importance.

The interesting symbiotic properties, together with the extensive taxonomic knowledge about R. galegae, prompted us to use the species to answer important questions related to the diversity and evolution of rhizobia and their symbiosis with legumes. A recent study indicated that R.galegae bv. orientalis strains have lower genetic diversity than R. galegae bv. officinalis strains (37). However, very few biovar orientalis strains were included in that study. In order to obtain a larger sample, R. galegae strains belonging to both biovars were collected during an expedition to the Caucasus, one of the centers of diversity of G. orientalis but not of G. officinalis. The collection was used for two different purposes.

First, we wanted to investigate the influence of the host plant on the genetic diversity of rhizobial populations. The host plant is an important factor shaping the genetic structure of a natural population of rhizobia (1, 2, 4, 9, 42, 45). The genomic diversity of the collected rhizobia was assessed by amplified fragment length polymorphism (AFLP) fingerprinting and by restriction fragment typing of the rRNA gene internal transcribed spacer (ITS). It is generally believed that insertion sequence (IS) elements in rhizobia promote genetic diversification through genomic rearrangements and recombination (8, 11, 17). The use of IS elements as probes for Southern hybridization can provide high-resolution fingerprints of rhizobial strains (26, 27, 31) and some information for understanding the function of these elements in rhizobial populations (2). An IS-related sequence, closely related to an Agrobacterium transposase from the widely studied plasmid pTiA6NC, has been found in the symbiotic region of R. galegae bv. orientalis strain HAMBI540 (24, 33), so we also used this IS sequence for fingerprinting the R. galegae strains.

Second, the collection was used to get a picture of the roles that the different parts of the symbiotic genome play in processes related to the evolutionary adaptation of rhizobia to different host plants. The genetic diversity in the symbiotic gene regions of R. galegae strains belonging to the two biovars was studied by PCR-restriction fragment length polymorphism (RFLP) analysis of five different symbiotic loci containing previously identified nod box sequences (32).

MATERIALS AND METHODS

Expedition and plant sampling.

G. officinalis and G. orientalis plants were sampled during the INTAS expedition to the Caucasus (Russia) in the summer of 1999 (Fig. 1). The plants, with roots, were taken from 25 sites (∼1 ha each) along a 300-km route through a mountainous part of the Krasnodar region in the northwest of the Caucasus (Fig. 1). An interesting observation was that the Galega species were found growing separately, with the exception of only two sites (Fig. 1, sites 104 and 126). Galega species in the area surveyed were readily distinguished by morphological traits of inflorescences and of the root system. G. orientalis was always part of well-established plant associations, whereas G. officinalis preferred disturbed or destroyed ecotopes and was part of the pioneer flora. The G. officinalis population was homogeneous morphologically, while the G. orientalis population was differentiated into distinct ecotypes, as is typical for the species. At least four well-differentiated ecotypes of G. orientalis were present in the region: mountain ecotypes A, from the Lagonaki tract (site 161), and B, from the mountain Otrub (site 97), were characterized by great height, good foliation, and a large inflorescence (up to 40 to 50 cm). The mountain ecotype C from Mezmay (site 160) was distinguished by dark foliage, and ecotype D from Khadyzhensk (site 176) and from Kamenoostrovsky (sites 104, 124, 126, and 127) was not as tall as the mountain ecotypes; had paler, shorter inflorescences; and sometime grew together with G. officinalis plants (sites 104 and 126).

FIG. 1.

Sampling locations. Solid circles, sites where G. orientalis grew; open circles, sites where G. officinalis grew; split circles, sites where both host plants grew.

Bacterial strains.

The plants sampled were brought to the laboratory and nodules were cut from the roots and washed in sterile water for 2 min in 70% ethanol, 4 min in 4% Ca(OCl)₂, and three times in sterile water. The nodules were crushed, suspended in 100 μl of liquid TY medium (3), and streaked out on TY agar. Single colonies were picked and restreaked. In total, 47 strains were isolated from G. orientalis sampled from 10 sites and 54 strains were isolated from G. officinalis sampled from 17 sites (Table 1). Two reference strains, HAMBI540 (R. galegae bv. orientalis) and HAMBI1141 (R. galegae bv. officinalis), were included in the study. All strains are available from the HAMBI collection, University of Helsinki.

TABLE 1.

R. galegae strains used in the study

AFLP groupa	Host plant	Strain(s)b	Site(s)	RFLP type revealed in rRNA ITS and symbiotic locid
				ITS	nb1	nb2	nb3	nb4	nb5
1	G. officinalis	G003	72	A	A	−c	A	A	A
2	G. officinalis	G004	72	C	A	−	A	A	B
3	G. officinalis	G005∗, G007	72	B	A	−	A	A	B
4	G. officinalis	G006	72	B	A	−	A	A	A
5	G. officinalis	G008	72	B	A	−	A	A	B
6	G. officinalis	G009	72	B	A	−	A	A	A
7	G. officinalis	G010∗, G011, G027, G028	72, 77, 127	A	A	−	A	A	A
8	G. officinalis	G013, G014∗, G015, G018, G021	108	B	A	−	A	A	B
9	G. officinalis	G017	108	A	A	−	A	A	A
10	G. officinalis	G019∗, G020	108	A	A	−	A	A	B
11	G. officinalis	G022	110	B	A	−	A	A	−
12	G. officinalis	G023	110	B	A	−	A	A	A
13	G. officinalis	G024	110	D	A	−	A	A	−
14	G. officinalis	G025	124	B	A	−	A	A	B
15	G. officinalis	G026	127	A	A	−	A	A	A
16	G. officinalis	G029	127	B	−	−	A	A	A
17	G. officinalis	G030	127	C	A	−	A	A	A
18	G. officinalis	G031	127	B	A	−	A	A	A
19	G. officinalis	G032∗, G035	130	A	A	−	A	A	A
20	G. officinalis	G033∗, G038	130	B	A	−	A	A	B
21	G. officinalis	G034	130	B	A	−	A	A	B
22	G. officinalis	G036	130	B	A	−	A	A	A
23	G. officinalis	G037	130	B	A	−	A	A	B
24	G. officinalis	G039	133	B	A	−	A	A	A
25	G. officinalis	G040∗, G041, G042, G043, G046, G050	139, 169, 176	B	A	−	A	A	B
26	G. officinalis	G045	173	B	A	−	A	A	B
27	G. officinalis	G047	173	E	B	−	A	A	B
28	G. officinalis	G048	173	E	A	−	A	A	B
29	G. officinalis	G049	175	F	C	−	A	A	B
30	G. officinalis	G053	176	A	A	−	A	A	A
31	G. officinalis	G054	177	B	A	−	A	A	B
32	G. officinalis	G055	177	A	A	−	A	A	A
33	G. officinalis	G056	177	B	A	−	A	A	A
34	G. orientalis	G093∗, G097	104	A	A	−	A	A	A
35	G. orientalis	G098	126	B	A	−	A	A	A
36	G. orientalis	G110∗, G112	161	B	A	−	A	A	B
37	G. orientalis	G057	80	G	D	A	B	B	C
38	G. orientalis	G058	80	H	−	A	B	B	C
39	G. orientalis	G060	85	I	E	A	B	B	C
40	G. orientalis	G061∗, G062	85	K	−	A	B	B	C
41	G. orientalis	G063	85	L	−	A	B	B	C
42	G. orientalis	G064∗, G083, G084	85, 97	M	E	A	B	B	C
43	G. orientalis	G065	85,	K	E	A	B	B	C
44	G. orientalis	G066	85	I	E	A	B	B	C
45	G. orientalis	G067	86	N	E	A	B	B	C
46	G. orientalis	G068∗, G071, G078, G079	86	O	D	A	B	B	C
47	G. orientalis	G069∗, G070, G076, G077	86	G	D	A	B	B	C
48	G. orientalis	G072∗, G073, G074, G075	86	G	D	A	B	B	C
49	G. orientalis	G080	86	G	D	A	B	B	C
50	G. orientalis	G081	97	M	E	A	B	B	C
51	G. orientalis	G086∗, G090	97	M	E	A	B	B	C
52	G. orientalis	G087∗, G089	97	P	F	A	−	C	D
53	G. orientalis	G091	97	Q	E	A	C	B	C
54	G. orientalis	G092	97	R	G	A	B	B	C
55	G. orientalis	G094	111	S	D	A	B	B	C
56	G. orientalis	G095	111	G	D	A	B	B	C
57	G. orientalis	G096	126	T	E	A	B	B	C
58	G. orientalis	G099	126	M	E	A	B	B	C
59	G. orientalis	G100∗, G101	142, 146	I	G	A	B	B	C
60	G. orientalis	G102	146	U	H	B	D	B	C
61	G. orientalis	G103∗, G104	146	W	E	A	B	B	C
62	G. orientalis	G105∗, G106	146	M	E	A	B	B	C
63	G. orientalis	G107	146	X	G	A	B	B	C
64	G. orientalis	G108	160	R	G	A	B	B	C
65	G. orientalis	G109	160	W	−	A	B	B	C
66	G. orientalis	G111	161	M	−	−	−	−	−
	R. galegaebv. officinalis	H1141 (reference strain)		X	A	−	A	A	B
	R. galegaebv. orientalis	H540 (reference strain)		O	D	A	B	B	C

^aAFLP groups were revealed in initial AFLP analyses and comprised strains with identical AFLP patterns (see explanations in the text).

^bStrains printed in boldface contain IS element copies. , representative strain from each AFLP group that was tested.

^cNo or weakly amplified fragment.

^dRFLP types are composite types revealed in restriction analyses of PCR-amplified fragments by using three enzymes, HaeIII, MspI, and AluI.

Plant test.

The symbiotic properties of the five R. galegae bv. officinalis isolates recovered from G. orientalis nodules were studied on G. orientalis cv. Nadezhda and G. officinalis var. E-103 (Maykop region) from the VIR (N. I. Vavilov Russia Research Institute of Plant Genetic Resources) collection for 30 days under nitrogen-deficient sterile conditions in glass tubes containing vermiculite (30). Each strain was tested in two replicates with each host plant.

DNA isolation.

Total genomic DNAs of all R. galegae strains were obtained after lysozyme-sodium dodecly sulfate lysis followed by phenol-chloroform extraction and ethanol precipitation (12).

AFLP analyses.

The AFLP procedure was performed as described by Vos et al. (39), with some modifications. One hundred to 500 ng of DNA was digested with EcoRI and TruI (MseI) (MBI Fermentas) restriction enzymes in a PTC-200 thermal cycler (MJ Research, Watertown, Mass.) and, in the same step, ligated with double-stranded adapters specific for each restriction half-site. (7). Two sets of primers were used in separate PCRs: one set with two selective nucleotides (boldface) (MseI-gc [GAT GAG TCC TGA GTA AGC] and EcoRI-gc [GAC GTC GTA CCA ATT CGA GC]) and another set with a mixture of two and three selective nucleotides (MseI-gc and EcoRI-gag [GAC GTC GTA CCA ATT CGA GAG]). The PCR conditions were as described previously (37). Silver-stained polyacrylamide gels, prepared as described by Dresler-Nurmi et al. (7), were scanned with an AGFA scanner and analyzed using Bionumerics software, version 2.0 (Applied Maths, Kortrijk, Belgium). An unweighted pair group method with averaging (UPGMA) dendrogram was generated from all strains by using Pearson correlation coefficients. For further analyses, including PCR-RFLP and IS fingerprinting, a representative strain (Table 1) was chosen from each AFLP group. Based on AFLP data, a UPGMA dendrogram was constructed for each biovar. We then chose representatives from each AFLP cluster and loaded them on the same gel; there were 20 R. galegae bv. orientalis and 21 R. galegae bv. officinalis strains with distinct fingerprints. Two different sets of primers were used and two gels, containing all 41 chosen strains, were obtained. The final dendrogram was constructed from the combined gel. The HAMBI1141 and HAMBI540 strains were excluded from the final dendrogram construction to avoid distorting the comparison of the branching depths of the two biovars studied.

PCR amplification.

The PCRs were carried out with 5 to 10 ng of total DNA. Dynazyme DNA polymerase (Finnzymes) was used with the supplied buffer (1.5 mm MgCl₂) and deoxynucleotide triphosphates. The primers fD1 and rD1 were used to amplify 16S rRNA genes (41). For ITS amplification, the primers FGPS1490-72 and FGPL132′ (22) were used. The locations of primers for amplifying regions containing nod boxes are shown in Fig. 2. Primers for nod box regions were constructed using Jellyfish software (Biowire): nb1f, TCC ATC TCC TAG ATG CCT CA; nb1r, GGC TAC GCA GTG ATT TGA TG; nodDf, CAG ATG TTG CCT GGG TCG; nodDr, ATC AAC CTC AGT CAA CCG GC; nb2f, TCG ATT GTT TGA ATG CCA TA; nb2r, GGC AAC ACG TCC AAC TTT CT; nb3f, GCG TGA ATT GGT GTC ATC CT; nb3r, AAG CTC GTC GCC AAA ATA GA; nb4f, GCC GCG ATA TAC TGA CCA TT; nb4r, TAT TCG AGG CGT GGA TGT CT; nb5f, GGT CAT GCT CTC GAT GGT TT; and nb5r, CTT CAA ATC CAC GCA ATC AA. All PCRs were performed using a PTC-200 Peltier Thermal Cycler (MJ Research) with the following standard temperature profile: an initial denaturation at 95°C for 3 min; 30 cycles of denaturation (1 min at 94°C), annealing (1 min at 55°C), and extension (1.5 min at 72°C); and final extension at 72°C for 3 min. When amplification was weak, the PCR was repeated with a lower annealing temperature (52 or 53°C). The amplified DNA was examined by electrophoresis in 1% agarose gels.

FIG. 2.

Maps of symbiotic gene regions of Rhizobium galegae. The locations of primers used in the study and of amplified fragments are shown.

RFLP analyses.

Aliquots of PCR products were digested with the restriction endonucleases HaeIII, MspI, and AluI (MBI Fermentas) using the Y⁺/Tango buffer supplied. The restricted DNA was analyzed by electrophoresis in 4% agarose. pGEM DNA markers (Promega) were used. The gels were stained with ethidium bromide and photographed with the Kodak EDAS290 system. The RFLP type for each particular endonuclease and combined RFLP types were defined for each strain and for each locus. The genetic diversity at each locus was calculated as H = n(1 − Σx_i ²)/(n − 1), where x_i is the frequency of the ith RFLP type and n is the number of RFLP types (5). Similarities between the HAMBI540 and HAMBI1141 strains in symbiotic loci were estimated from the proportion of shared and unshared restriction fragments by using a similarity coefficient (20).

Southern hybridization.

An aliquot (0.5 to 1 μg) of the total DNA sample was digested to completion with HindIII (MBI Fermentas). Restriction fragments were separated by electrophoresis in 1% agarose. DNA of phage λ digested with HindIII was included on the gel as a size marker. Southern blotting of DNA, hybridization, and detection were done according to instructions provided with the nonradioactive labeling and detection kit (Roche). The digoxigenin (DIG) labeling of the IS probe was performed using PCR amplification with the ISf (CGA TTG CTA CAA TGG CTT CA) and ISr (TCT TTC TTC CAC ATG CAA CC) primers, flanking an inner fragment of the unnamed IS element, constructed from sequence data of R.galegae strain HAMBI540 (34).

RESULTS AND DISCUSSION

A number of studies suggest that the host plant is the most important factor that shapes the structure of rhizobial populations (1, 2, 4, 5, 10, 42, 45; M. L. Roumiantseva, N. A. Provorov, and B. V. Simarov, Proc. VIII Eastern Eur. Symp. Biol. Nitrogen Fixation, p. 53, 1992). Moreover, the process of fine tuning symbiosis can be to some extent inferred from the contemporary diversity of both host plants and their rhizobial microsymbionts. It appears that the greater the genetic diversity of the host plant population, the more diverse is the rhizobial population associated with it. This can best be shown by comparing rhizobia belonging to the same species or even biovar but isolated from different host plants (4, 5, 42). Additional evidence comes from the higher diversity of rhizobia in gene centers of host plants (46).

One hundred and one R. galegae strains were isolated from nodules of wild-growing G. officinalis and G. orientalis in this work. All the strains had 16S rRNA MspI restriction profile identical to that of the reference R. galegae strains HAMBI540 and HAMBI1141 (data not shown). AFLP fingerprinting (Fig. 3), ITS PCR-RFLP, and nod box PCR-RFLP (Table 1) clearly distinguished the strains isolated from each host plant. However, five strains originally isolated from G. orientalis nodules belong by their AFLP and ITS and nod box PCR-RFLP patterns to the G. officinalis cluster (Table 1). The symbiotic performances of these five isolates were studied. They produced relatively large pink nodules on G. officinalis plants and small white or light-green nodules on G. orientalis plants. Thus, we can conclude that these strains are R. galegae bv. officinalis even though they were isolated from nodules of the nontypical host plant G. orientalis.

FIG. 3.

UPGMA dendrogram constructed from AFLP data. The relative positions of R. galegae bv. orientalis HAMBI540 and R. galegae bv. officinalis HAMBI1141 are indicated by dashed lines.

The Caucasus region is known as the gene center (or a center of diversity) for many legumes (38, 47) and in particular for Galega species (18). The results of our expedition confirm this. However, there is a significant difference between G. orientalis and G. officinalis. The G. orientalis populations in the region explored are exceptionally diverse and have a longer history than the G. officinalis populations.

The data obtained in the study by molecular characterization of rhizobial isolates correlates well with host plant diversity. AFLP analysis, which is based on whole-genome variability, shows that the Caucasian populations of R. galegae bv. orientalis are more diverse than the R.galegae bv. officinalis populations. The branching depth in the UPGMA dendrogram is deeper in the case of R. galegae bv. orientalis (Fig. 3). Interestingly, the R. galegae bv. orientalis strains group into two different AFLP clusters, A and B, the latter being closer to the R. galegae bv. officinalis cluster than to the main R. galegae bv. orientalis AFLP cluster (Fig. 3). Another interesting finding is that HAMBI1141 lies outside the main AFLP cluster of R. galegae bv. officinalis. This strain also has a unique ITS RFLP type, X (Table 1), and we believe that it is a rare genotype not typical of this region.

The results of the PCR-RFLP analysis, which allows the study of variability at particular loci (Table 1), are in agreement with those of the AFLP analyses. The corresponding values of heterogeneity in the ITS region for the two biovars were 0.87 for R. galegae bv. orientalis and 0.56 for R. galegae bv. officinalis. The average heterogeneity in the symbiotic regions analyzed was 0.42 for R. galegae bv. orientalis and 0.13 for R. galegae bv. officinalis. The only symbiotic region in which R. galegae bv. officinalis strains are more diverse is nod box 5. Overall, the strains of R. galegae bv. orientalis are more diverse than the R. galegae bv. officinalis strains in symbiotic regions as well as in the ITS region. This finding, together with the AFLP results, provides strong evidence that within the Caucasus region studied R. galegae bv. orientalis strains are more diverse than R. galegae bv. officinalis strains.

Interestingly, the results appear to contradict a previous study (37) in which G. officinalis strains were shown to be more diverse than G. orientalis strains. However, we believe that there is no contradiction: the population of G. officinalis studied by Terefework et al. (37) was geographically heterogeneous, representing five different locations (New Zealand, United Kingdom, Bulgaria, Italy, and Argentina). At least two of them (New Zealand and Bulgaria) are places where native G. officinalis is widely represented. At the same time, the population of G. orientalis analyzed by Terefework et al. represented two locations, Russia and Finland (where G. orientalis for the most part is an introduced plant) and included few strains isolated in the Caucasus. Therefore, the observations of Terefework et al. (37) are in concordance with the results reported here, emphasizing the importance of host plant diversity for the diversity of rhizobia.

Analysis of the symbiotic region is of prime interest when subtle genetic adaptation of rhizobia to particular host plants is under investigation. Five different fragments containing nod box sequences were studied in this work. The presence of nod boxes itself points to the great importance of these regions in symbiotic interaction. We constructed five pairs of primers (see Materials and Methods) to amplify the region adjacent to each of five nod boxes, which were identified by Suominen et al. (32) in the symbiotic regions of the R. galegae bv. orientalis reference strain HAMBI540 (Fig. 2). Regions nb3-nodD2, nb4-nodU, and nb5-nodE were amplified successfully for almost all R. galegae bv. officinalis and R. galegae bv. orientalis strains (Table 1). However, the nb2-dctA region was successfully amplified only for R. galegae bv. orientalis and not for R. galegae bv. officinalis strains (Table 1). The same was true for the nb1-nodB region (not shown), but when the nb1f primer was replaced by the nodDf primer, located inside the nodD1 gene (Fig. 2), the corresponding fragment comprising the nodD1-nodB region (which includes nb1) was amplified successfully for most of the strains of both biovars (Table 1). The last finding means that there is some difference between the nucleotide sequences of nod box 1 in strains belonged to different biovars.

The most conserved symbiotic genes seem to be in the nod box 3 and nod box 4 regions. The nod box 1 region is highly conserved among R. galegae bv. officinalis strains, and the nod box 5 region is highly conserved among R. galegae bv. orientalis strains. We expected to find high levels of divergence in the nod box regions because of the high proportion of noncoding sequence. However, they proved to be highly conserved despite the high noncoding content, e.g., 56% of noncoding sequence in the case of the most conserved nod box 3 region (unpublished sequence data for strain HAMBI540). We assume that there may be strict functional constraints on the noncoding regions adjacent to nod boxes (6).

The Nei similarity coefficient was calculated by comparing the restriction patterns of strains HAMBI540 and HAMBI1141 to give a measure of differences between different biovars in particular symbiotic loci. The coefficients were 0.77 for the nod box 3 region, 0.76 for nod box 4, 0.52 for nod box 5, and 0.47 for nod box 1. Thus, the greatest differences were found in the nod box 1 and nod box 5 regions, and the smallest differences were found in the nod box 3 and nod box 4 regions. The nod box 2 region has not been taken into account, as there was no amplified fragment for the R. galegae bv. officinalis strains. Thus, we can suggest that the nod box 1 and 5 regions with adjacent genes might play a more important role in host plant adaptation and symbiotic tuning. This suggestion is quite plausible because we know that the nodA and nodEF genes (located adjacent to nod boxes 1 and 5) are involved in synthesis of unsaturated fatty acids and their attachment to the Nod factor backbone (23). It is established that the unsaturated fatty acid strategy is the main factor for Rhizobium-Galega recognition (reviewed in reference 36). The presence and importance of different allelic forms of these host range genes (46) in different biovars of R. galegae cannot be excluded. However it seems there is no clear correlation between the genetic background and the Nod factors produced. It was shown that R. galegae strains belonging to different biovars produce nearly identical Nod factors, which are a mixture including a series of C₁₈ and C₂₀ fatty acids with carbonyl-conjugated double bonds (44).

To obtain the whole picture of genome adaptation to particular host plants, this comparative study should be expanded to include more symbiotic genes. Thus, further investigation is needed to complete the picture of symbiotic adaptation in the Rhizobium-Galega symbiosis. This work demonstrates an approach to trace evolutionary pressure along the symbiotic genomes of two closely related rhizobium biovars. All the analyses in this study, AFLP and RFLP of ITS and symbiotic regions, delineated the two biovars. The absence of shared restriction patterns or combined profiles from the ITS and symbiotic regions strongly suggests that there has been no recombination between the biovars. Thus, the major factor contributing to the divergence of this part of the symbiotic machinery could be the host plant.

We used a pair of primers which allowed us to amplify the inner fragment of an IS element closely related to an Agrobacterium transposase, which was revealed in the symbiotic region of R. galegae bv. orientalis strain HAMBI540 (33). Only 16 strains of R. galegae bv. orientalis were found to contain copies of the IS sequence (Table 1). Furthermore, we noticed that all IS-positive strains, including the reference strain HAMBI540, had closely similar rRNA ITS types: patterns O and G (Table 1) differ only in their HaeIII restriction patterns. Strains of these genotypes formed a clearly separated cluster (B) in the AFLP dendrogram (Fig. 3). It is particularly remarkable that 13 of these strains were isolated from the same site (Table 1). By using a DIG-labeled IS probe for Southern hybridization, it was shown that the strains contain from three to nine copies of the IS element; HAMBI540 contains two copies (Fig. 4). Each hybridizing band corresponds to at least one copy of the IS element, because the IS sequence does not contain HindIII restriction sites. Strains with identical IS hybridization patterns had identical AFLP fingerprints.

FIG. 4.

IS fingerprinting by Southern hybridization of R. galegae strains containing the IS element.

The distribution of the IS-like sequences in the population raises some interesting questions. Why is the IS element found only in genetically similar backgrounds? Why were most strains containing this IS element found in the same site? A likely answer to the first question is a limitation of horizontal transfer between different chromosomal groups, i.e., a clonal structure of the population (19, 43). Our results show no evidence for any recombination event between the biovars. Furthermore, it appears that the strains that belong to AFLP group B and contain the IS element also represent a clonal group. The second question could be explained by limitations on the transport of strains between different locations. The finding that it was rare for the same AFLP type to be found at more than one site provides support for this view. On the other hand, according to Beijerinck's principle that “everything is everywhere, the environment selects,” the limitations could be assigned to either specific soil conditions or selection by plant genotype.

Our work shows that the study of rhizobial diversity in gene centers of the host is a way to answer some interesting questions in plant-microbe interaction and can shed some light on patterns and constraints of evolution in rhizobia.