Genetic diversity of \u003ci\u003eSinorhizobium\u003c/i\u003e populations recovered from different \u003ci\u003eMedicago\u003c/i\u003e varieties cultivated in Tunisian soils

Genetic diversity of Sinorhizobium populationsrecovered from different Medicago varietiescultivated in Tunisian soils

M. Jebara, R. Mhamdi, M.E. Aouani, R. Ghrir, and M. Mars

Abstract: A collection of 468 rhizobial isolates was obtained from different ecological areas of Tunisia by trappingthem onMedicago sativacv. Gabes,Medicago scutelletacv. Kelson,Medicago truncatula, andMedicago ciliaris. Asubsample of 134 rhizobia was chosen to determine their plasmid profile, and 89 isolates were subjected to multilocusenzyme electrophoresis (MLEE) and PCR/RFLP analysis using 16S, IGS (inter genic spacer), andnifKD probes.Twenty-five representatives from these isolates were evaluated for their nodulation and nitrogen fixation capacities.MLEE studies revealed two groups with highly heterogeneous host specificity and geographical origin. The discrimina-tory power was found to be slightly better with the amplified ribosomal intergenic region, than thenifKD genes. Divi-sions detected bynifKD amplified DNA analysis matched those established by ribosomal PCR- RFLPs. Thecomparison between different analyses revealed that MLEE illustrated better phenotypic properties of isolates thanPCR-RFLP or plasmid content analysis. Clear distinction betweenSinorhizobium melilotiand Sinorhizobium medicaewere observed by analysis of the IGS symbiotic regions betweennifD and nifK genes. Were able to distinguish threeinoculation groups; isolates trapped fromM. sativacv. Gabes andM. scutelletacv. Kelson formed one inoculationgroup which was more closely related to isolates trapped fromM. truncatula than those trapped fromM. ciliaris.

Key words: Sinorhizobium, Medicago, diversity, MLEE, PCR, symbiotic effectiveness.

Résumé: Une collection de 468 isolats de rhizobiums fut obtenue à partir de diverses régions écologiques de Tunisie,en les captant dansMedicago sativacv. Gabes,Medicago scutelletacv. Kelson,Medicago truncatulaet Medicago ci-liaris. Une échantillon représentatif de 134 rhizobiums a été choisi afin de définir leur profil plasmidique. 89 isolats fu-rent analysés par technique de mobilité électrophorétique des enzymes ou MLEE et par PCR-RFLP en utilisant dessondes spécifiques aux régions 16S, IGS etnifKD. Par ailleurs, 25 représentants des différents groupes ayant été déga-gés des études moléculaires ont été évalués pour leur capacité de nodulation et de fixation de l’azote. Les analyses deMLEE ont décelé deux groupes ayant des spécificités pour l’hôte et des origines géographiques grandement hétérogè-nes. Le pouvoir discriminant des analyses s’avéra légèrement supérieur avec l’usage des régions intergéniques riboso-males pour l’amplification, comparativement aux gènesnifKD. Les divisions détectées par l’analyse d’amplificationd’ADN nifKD se sont appariées à celles établies par le PCR-RFLP ribosomal. La comparaison des différentes analysesrévéla que le MLEE mettait mieux en évidence les caractéristiques phénotypiques des isolats que les analyses de PCR-RFLP ou de contenu plasmidique. Une nette distinction entreSinorhizobium melilotiet Sinorhizobium medicaeapparutlors de l’analyse des régions IGS symbiotiques situées entre les gènesnifD et nifK. Nous avons pu distinguer troisgroupes d’inoculation; les isolats captés parM. sativacv. Gabes etM. scutelletacv. Kelson ont constitué un grouped’inoculation qui s’apparentait davantage aux isolats captés parM. ciliaris.

Mots clés: Sinorhizobium, Medicago, diversité, MLEE, PCR, efficacité symbiotique.

147

Jebara et al.Introduction

Soil bacteria forming nitrogen-fixing nodules on legumesare divided into five generaRhizobium, Mesorhizobium,Bradyrhizobium, Azorhizobium,and Sinorhizobium. However,the continued application of molecular approaches has led totheir division into additional genera (Martinez-Romero andCaballero-Mellado 1996).

Sinorhizobium(formerly Rhizobium) meliloti is defined byits ability to form symbiotic relationships with members ofthree related genera of legumes,Medicago (perennial andannual medics),Melilotus (sweet clover), andTrigonella(fenugreek) (Jordan 1984). Although medics are indigenous

Can. J. Microbiol. 47: 139–147 (2001) © 2001 NRC Canada

139

DOI: 10.1139/cjm-47-2-139

Received November 23, 1999. Revision received October 25,2000. Accepted November 15, 2000. Published on the NRCResearch Press Web site on January 30. 2001.

M. Jebara,1 R. Mhamdi, M.E. Aouani, R. Ghrir, and M.Mars. Laboratoire de Biochimie Végétale et Symbiotes,Institut National de Recherche Scientifique et Technique, BP95, CP 2050, Hammam-lif Tunisie.

1Author to whom all correspondence should be addressed(e-mail: [email protected]).

to all parts of Tunisia, inoculation was applied only in thenorth twenty years ago (F.A.O. 1988).

The economic significance of theRhizobium–legume as-sociation has resulted in extensive research designed to im-prove the efficiency of the symbiosis. Many attempts havebeen made to select naturally occurring strains that are bothsymbiotically effective and able to persist in the environ-ment.

The perennial speciesM. sativa is the most widely culti-vated species of lucerne in the world. Symbiotic strains iso-lated from this legume have been studied intensively andhave been included in many genetic and taxonomic studiesof the Rhizobiaceae.

Annual Medicagospecies are well adapted and native toareas in the Mediterranean basin (Lesins and Lesins 1979).They are candidates for use in sustainable agriculture sys-tems. To take advantage of biological nitrogen input in semi-natural systems of economic and environmental importance,suitable rhizobia must be present in the soil capable of es-tablishing an efficient symbiosis withMedicagospecies.

Thus, it is essential to characterise the nativeRhizobiumpopulations and the factors which influence their composi-tion and population dynamics.

Various tools have been used to explore the phenotypicand genotypic variation among rhizobial populations and to as-sess their diversity. Phenotypic characterisations ofS. melilotistrains indicated that the composition of field populationsvaries greatly between soils of different geographical areas.However, theS. melilotipopulation obtained from nodules isnot representative of the total population present in the soil(Bromfield et al. 1995).

The most widely used method to measure genotypic diver-sity of prokaryote populations has been multilocus enzymeelectrophoresis (MLEE) (Selander et al. 1986). This methodhas proved to be a powerful tool in genetic studies of a widerange of organisms. The genetic relatedness of bacterialstrains obtained by MLEE strongly correlated with estima-tions of divergence obtained from DNA–DNA hybridisationexperiments and with phylogenetic trees derived from nucle-otide sequence analysis (Milkman and Bridges 1990; Nelsonet al. 1991).

MLEE has been applied to the analysis of genetic popula-tions of soil bacteria (Pinero et al. 1988) and it has providedinteresting new information on genetic relationships in thefamily of Rhizobiaceae, as well as revealed important diver-sity at the species level (Rinaudo et al. 1993). More recently,various other methods based on PCR have been used tocharacteriseRhizobiumstrains and to examine the geneticrelationship between them (Laguerre et al. 1996).

The aim of the present investigation was to assess the di-versity of field populations of rhizobia nodulating the com-mon Medicago in Tunisia by means of plasmid profiling,MLEE, PCR/RFLP analysis, examining symbiotic effective-ness, and by carrying out phenotypic tests. Therefore, com-parisons were made between different methods and theirability to demonstrate the characteristics and symbiotic ef-fectiveness of strains.

The final goal of this study was to develop a collection ofappropriate strains that are adapted to plant host and soils.

Materials and methods

Bacterial strainsRhizobial isolates were collected from root nodules harvested

from different Medicago species grown on different soil types.Four Medicagospecies were used as hosts to trap rhizobia. Seedswere surface sterilised and scarified to improve seed germination,and placed in pots (Vincent 1970) filled with soil from 60 differentsites located in different geographical regions of Tunisia.

Bacterial isolates were obtained using the crushed-nodulemethod (Vincent 1970) from nodules removed 50 day old plants.Pure cultures were obtained after single colonies were streakedthree times. When stable colony morphology variants were ob-tained, the identity of the nodulating strains was examined by per-forming plant infection tests with the original host plants. A totalof 468 isolates were obtained and stored on YEM slants at 4°C andin 40% glycerol at –80°C. All isolates were tested for their abilityto form nodules onMedicagoas previously described (Beck et al.1993).

Sinorhizobium melilotiSU47, synonymous with RCR 2011, hasbeen used in commercial inoculants forM. sativa. Other strainsused includeAgrobacterium tumefaciensPD67 (Knauf and Nester1982), CFN42 the type strain ofR. etli (Pinero et al. 1988), andS. frediiUSDA 205, which was kindly provided by Laguerre et al.1996.

Phenotypic tests

pH toleranceThe ability of the 468 isolates to grow in alkaline or acid media

was tested using YEM broth in which pH was adjusted to 4, 4.5, 5,6, 8, 9, 9.5, or 10 by adding HCl or NaOH. Control cultures weregrown at pH 7.0

NaCl toleranceThe tolerance of isolates was tested by using YEM broth con-

taining 0.15, 0.3, 0.5, and 0.7 M NaCl. Concentrations of 0.15, 0.3,and 0.5 M NaCl were used for comparing rhizobial species (Jordan1984), and 0.7 M NaCl was used to select highly resistant strains.The standard YEM medium with 0.1 M NaCl was used as control.

Plant testsFour host plants were used:M. sativa cv. Gabes, which is the

most widely cultivated perennial species ofMedicagoin Tunisia, theannualMedicagospeciesM. scutelletacv. Kelson, M. truncatula,and M. ciliaris.

Effectiveness trials were carried out on plants grown in plasticpots (one plant in each pot) filled with a heat sterilised mixture ofgravel and perlite. Inoculation was performed with 1 mL of a sus-pension containing approximately 109 bacteria. Fourteen replicatesper treatment were examined. Plants were grown in a greenhouseunder natural light with a daily minimum-maximum temperature of18°C–24°C. A sterilised nitrogen free solution (Broughton andDilworth 1971) was supplied when needed.

The plants were harvested after 50 days. Shoots were separatedfrom roots, dried at 80°C for three days, weighed, and analysed forN content by Kjeldahl digestion.

Symbiotic effectiveness was estimated by shoot dry matter pro-duction according to the mean of 14 plants inoculated with thesame isolate.

Sinorhizobium melilotiSU47 was used for symbiotic effective-ness tests (Bromfield et al. 1985).

The ANOVA program involving analysis of variance and thestandard deviation of the means was used to determine the signifi-cance of apparent differences in symbiotic effectiveness betweenstrains.

© 2001 NRC Canada

140 Can. J. Microbiol. Vol. 47, 2001

Plasmid profilesUsing agarose gel electrophoresis, 134 isolates were analysed

for plasmid contenthe Eckhardt (1978) and as modified byRosenberg et al. (1982), except that cells were grown on YSM me-dium (Lesley 1982). Estimates of molecular size ofSinorhizobiumplasmids relative to molecular mass markers were obtained by lin-ear regression of log10 molecular mass (MDa) on log10 mobility us-ing standard plasmids of 267, 177, 117, and 35 MDa fromA. tumefaciensPD67. The megaplasmid ofS. meliloti SU47 wasused as a standard to detect the presence of megaplasmids ofequivalent mobility. Each isolate was tested in three separate gels.

Multilocus enzyme electophoresis (MLEE)

Preparation of lysatesBacterial cultures were grown for 36 h on an orbital shaker at

28°C in 150 mL of TY broth (Beringer 1974), supplemented with5.3 mM calcium chloride. Cells were harvested by centrifugation at6000 ×g for 10 min at 4°C. After suspension in 1.5 mL of 50 mMTris hydrochloride buffer containing 5 mM EDTA (pH 7.5), thebacteria were sonicated for 45 s. The sonicated suspension wascentrifuged at 20 000 ×g for 20 min at 4°C, and the clear lysatewas stored at –80°C.

ElectrophoresisThe MLEE technique of Selander et al. (1986) was used. The

electrophoretic buffer systems and the enzymes assayed were asfollows: (i) Tris citrate pH 8: glucose-6-phosphate dehydrogenase(G6P), phosphoglucose isomerase (PGI), isocitrate dehydrogenase(IDH), leucyl-alanine peptidase (PEP). (ii ) Tris citrate pH 6.7:hydroxybutyrate dehydrogenase (HBD), 6-phosphogluconatedehydrogenase (6PG), NAD-malate dehydrogenase (MDH),phosphoglucomutase (PGM). (iii ) Borate pH 8.2: hexokinase(HEX), adenylate kinase (ADK), and indophenol oxidase (IPO).

Distinctive mobility variants of each enzyme, numbered in orderof decreasing anodal mobility, were equated with alleles at the cor-responding structural gene locus, and allele profiles for the lociwere equated with multilocus genotypes (Selander et al. 1986).

The extent of linkage disequilibrium among electrophoretictypes was evaluated by comparison of frequencies of allele combi-nations between pairs of polymorphic enzyme loci (Hedrick andThomson 1986). Genetic distance between each pair of electropho-retic types (ETs) was estimated as the proportion of loci at whichdissimilar alleles occurred, and clustering using a matrix ofpairwise genetic distances was performed by the average-linkagemethod (Sneath and Sokal 1973).

PCR /RFLP analysis

PCR amplificationThe PCR was carried out with 2.5µL of bacterial cell suspen-

sion as template DNA. PCR amplification was performed in stan-dard 50µL PCR mixtures containing 200µM of each of dNTP(Pharmacia LKB), 0.1µM of each of the two primers, templateDNA (2.5 µL); with the polymerase reaction buffer (10 mM Tris-HCl (pH 9.0) at 25°C, 50 mM KCl, 1% Triton X-500, 1.5 mMMgCl2) and 2.5 units of Taq-polymerase (Gibco-BRL). The ampli-fication mixture was overlaid with one drop of mineral oil. In allcases, the negative controls contained all compositions for PCR,excluding the template DNA. The reactions were run on a DNAthermal cycler (Biometra TRIO- Thermoblock) using a 35-cycleamplification series; after initial denaturation of the reaction mix-ture at 95°C for 3 min, each cycle included denaturation at 95°Cfor 1 min, reannealing at 55°C for 1 min and extention at 72°C for1 min. The final extention was carried out at 72°C for 3 min. Am-plified DNA was examined by horizontal electrophoresis in 0.9%

agarose with 5µL aliquots of PCR product. Amplified DNAs weresubsequently digested for RFLP analysis. DNA samples (1 to 2µg)were digested completely with 10 U of restriction endonucleases.The restriction enzymesMspI, RsaI, NdeII, CfoI, AluI, HaeIII, andHinf I were used according to the manufacturer’s instructions. Therestricted fragments were separated by horizontal electrophoresison 3% agarose (Nusieve 3: 1). Electrophoresis was carried out at80 V for 3 h 30 with 11 × 14-cm gels. A 100 pb DNA ladder wasused as a standard for molecular size determinations. The gelswere stained in an aqueous solution of ethidium bromide(1 µg/mL) and photographed under 312 nm ultraviolet light withPolaroid type 667 films.

Oligonucleotide primersThe 16S rDNA gene amplifications were carried out with FGPS

6: 5′- GGA GAG TTA GAT CTT GGC TC -3′ and PGPS 1509: 5′-AAG GAG GTG ATC CAG CC -3′. The 16S rDNA sequence am-plified between primers FGPS 6 and PGPS 1509 is 1478 nucleo-tides long according to the analysis of twoSinorhizobium16 SrDNA sequences (Willems and Collins 1993; Yanagi and Yamasoto1993).

One DNA segment containing the IGS (inter genic spacer) be-tween the 16Srrn (gene encoding 16S rRNA) and 23Srrn genes(approximately 1.35 kb) was amplified from eachRhizobiumiso-late. The two oligonucleotide sequences used as primers were: 5′-TGC GGC TGG ATC ACC TCC TT -3′ (FGPS1490) derived fromconserved sequences in the 3′ part of 16S rDNA genes (Navarro etal. 1992) and 5′- CCG GGT TTC CCC ATT CGG -3′ (FGPS132′)characterised according to Normand et al. (1992) corresponding tothe 5′ part of the 23S rDNA gene adjacent to the IGS (Ponsonnetand Nesme 1994).

Another DNA fragment (approximately 1.35 kb) containing theIGS between thenifD and nifK was amplified from the followingtwo primers: primer FGPD 807: 5′- CAC TGC TAC CGG TCGATG AA -3′, corresponding to anifD sequence conserved amongnitrogen-fixing bacteria (Jamann et al. 1993) and FGPK 597: 5′-GTG GCT GCC CAC GAA GAA GCT TGG NGT GTG-3′. Synthe-sised and purified oligonucleotides were obtained from GENSETSA (Paris, France).

Results

Bacterial collectionA collection of 468 rhizobial isolates were obtained from

different ecological areas of Tunisia by trapping assay from60 different sites: 123 isolates were fromM. sativacv. Gabes,113 from M. scutelletacv. Kelson, 123 fromM. truncatula,and 109 fromM. ciliaris nodules.

The isolates obtained were first subjected to phenotypictests. Then, a subsample of 134 rhizobia was chosen forplasmid content analysis. A group of 89 isolates was subse-quently subjected to MLEE, PCR/RFLP analysis of 16SrDNA, IGS 16S–23S andnifKD regions. Finally a represen-tative sample of 25 isolates was analysed with respect to

© 2001 NRC Canada

Jebara et al. 141

Fig. 1. Schematic representation of plasmid profiles found amongthe Medicagoisolates.

their nodulation and nitrogen fixation abilities on differenthost plants.

Phenotypic characteristics of isolatesAll of the 468 isolates studied were rod-shaped bacteria.

On YEM medium, they were mucous, and they acidified themedium, changing the pH from 7 to 5. The generation timeranged between 3 and 5 h. All of the isolates grew at 28°Cin the range of pH 5 to 10.

Analysis of NaCl tolerance led to the distinction of threegroups: four resistant isolates (0.7 M) were found to have ashorter generation time (3 h), eight sensitive strains (0.15 M)with a longer generation time (5 h), and an intermediateclass. Strains from four sites were chosen to represent eachof the three NaCl tolerance groups. Thus plasmid contentanalysis was conducted using 134 isolates from 12 represen-tative sites.

Plasmid profilesThe 134 isolates were analysed for their plasmid content

by the procedure of Eckhardt (1978). The results showed aremarkable plasmid heterogeneity, and one to six bands weredetected in each isolate with molecular masses varying be-tween 5 and >267 MDa. At least 18 different plasmid typeswere observed among the 134 analysed isolates (Fig. 1). Themore dominant (profile type 5) was encountered in 35 iso-lates. Moreover, the two bands characteristic of this plasmidtype were common with nine other plasmid profile typescorresponding to 76 isolates.

Based on the distribution of plasmid types, the sites couldbe divided in three groups: the first group yielded the mostheterogeneous plasmid type, and in each of these sites morethan six different plasmid profiles were found. These sitesare marginal and cannot be cultivated because of their salin-ity. Less heterogeneity was found in the second group than

in the first group. In the third group, homogeneous siteswere usually cultivated by cereal culture. So, this distribu-tion may be explained by the soil properties, agriculturepractice, and related to the environmental conditions.

Multilocus enzyme electrophoresis (MLEE)Of the 134 isolates, we obtained quantitative and repeti-

tive results in MLEE with 89. Therefore MLEE was con-ducted on this subset of 89 isolates, which originated fromdifferent Tunisian sites.

All eleven enzymes examined were represented by a rangeof 1 to 7 profiles for each metabolic enzyme (Fig. 2). Theleast polymorphic enzyme was IPO with an enzymatic diver-sity of h = 0.181 followed by HBD, HEX, MDH, G6P, IDH,PGM, PEP, 6PG, ADK, and PGI with h = 0.57, 0.58, 0.60,0.61, 0.63, 0.68, 0.70, 0.73, 0.75, and 0.79 respectively.

Genetic relatedness among EtsIn the collection of 89 isolates tested, 10 of 11 enzymes

assayed were polymorphic. Only indophenol oxidase was in-variable. The comparison of the allele profiles of the ten re-maining loci revealed 72 distinct multilocus genotypes orETs (Fig. 3).

Cluster analysis of these ETs revealed two primary divi-sions (i and ii) at a genetic distance of 0.84, and two otherbranches represented by two and four ETs diverging respec-tively from the other ETs at a genetic distance of 0.90 and0.92. TheS. fredii strain USDA205 was found in a branchdistant from all of theMedicagoisolates tested.

Division i contained 29 of the identified ETs. These ETswere represented by 38 of the 89 isolates examined includ-ing the reference strain SU47.

Division ii included 45 ETs corresponding to 51 isolatesincluding S. medicaestrain M1. At a genetic distance of0.81, three subdivisions of division ii could be identified.

© 2001 NRC Canada

142 Can. J. Microbiol. Vol. 47, 2001

Fig. 2. Gels illustrating electrophoretic variation in seven enzymes. (A) Phosphoglucose isomerase (PGI), (B) 6-phosphogluconatedehydrogenase (6PG), (C) Leucyl-alanine peptidase (PEP), (D) Glucose-6-phosphate dehydrogenase (G6P), (E) Isocitrate dehydrogenase(IDH), and (F) NAD-malate dehydrogenase (MDH).

PCR/RFLP analysis

PCR-RFLP analysis of 16S rRNA genesThe 16S rRNA genes from DNA of the above 89 isolates

were amplified by PCR. Electrophoretic analysis of undi-gested PCR products revealed a single band of about 1500 bpfor all strains. This band corresponded to the expected size ofthe 16S rRNA genes among bacteria (Weisburg et al. 1991).PCR products were digested separately byMspI, RsaI,NdeII, and HaeIII, and the combined data of the four endo-nucleases used allowed the distinction of two ribosomaltypes that differed only by digestion withRsaI. These two ri-bosomal types were also found withS. melilotistrain SU47and S. medicaestrain M1, respectively. The results werefound to be in agreement with the distinction of the in twogroups determined by MLEE (not shown).

PCR-RFLP analysis of 16S–23S rDNA IGS regions16S–23S rDNA IGS regions of the 89 isolates were am-

plified with primers FGPS 1490 and FGPS 132. All strainsproduced single bands of approximately 1350 bp as esti-mated by summing the sizes of the restriction fragments af-ter digestion with restriction enzymes.

Four restriction endonucleases (CfoI, AluI, HaeIII, andMspI) were used to digest the 16S–23S rDNA IGS amplifi-cation product. Ten to eighteen distinct restriction patterns,with three to seven restricted fragments per pattern, were de-tected with each restriction enzyme. The combined data ofthe four endonucleases used yielded sixty-two DNA IGStypes. To estimate the phylogenetic relationships betweenthese 62 types, a similarity matrix was determined by ana-lysing the total restriction fragment patterns. This matrixwas used to construct a dendrogram based on the UPGMAalgorithm (Fig. 4). The resulting dendrogram indicate thatthe Medicago isolates were separated into two diverginggroups, at a genetic distance of approximately 0.84.

Group i containedS. melilotiSU47 and 37 isolates identi-fied as S. meliloti. Group ii contained 43 isolates and thestrain M1 identified as aS. medicae. The two groups arecompletely divergent.

The first group includes 25 distinct IGS profiles repre-sented by 38 isolates, which are grouped in three subdivi-sions at a genetic distance of 0.78. The first subdivision wasformed by 6 distinct profile types associated with 18 iso-lates, the second subdivision is consisted by 7 profile types,each one represented by only one isolate and the thirdsubdivision consisted of 13 isolates with 12 different profiletypes.

The second group contained 43 isolates with 37 distinctprofile types, which may be divided into four subdivisions.

© 2001 NRC Canada

Jebara et al. 143

Fig. 3. Genetic relationships among 72 electrophoretic types ofSinorhizobiumnodulatingMedicagoin Tunisian soils based onelectrophoretically detectable allelic variation at 10 enzyme loci.

Fig. 4. Dendogram (UPGMA) of genetic relationships betweenisolates ofSinorhizobiumnodulatingMedicagoin Tunisian soilsestimated by PCR-RFLP of 16S–23S rDNA IGS.

The first subdivision diverges at 0.82 and contains five pro-files represented by 8 isolates. The second one, which di-verges at 0.76, is composed of 7 isolates, each one with itsown profile. The third subdivision also diverges at 0.76, andwas formed by 6 distinct profile types for 7 isolates. Thefourth subdivision included 17 isolates with 15 different pro-files. The analysis of the relation between defined groupsand the geographic origin of isolates permitted the distinc-tion of three types of sites: a homogeneous site where allisolates are members only of one group; a second type hasone dominant subdivision which is a part of the group of thesite; a third more heterogeneous type has representatives ofeach one of the four defined subdivisions of groups i or ii.Our analysis of the chromosomal region containing the 16SrDNA plus the IGS nested between 16S and 23S rDNAgenes permitted an easy differentiation among the two spe-cies of Sinorhizobiumisolates from the same geographicalorigin. By this analysis we found two groups as already de-fined by MLEE studies. However, the first group describedhere includes the more divergent isolates classified neitherin group i nor in group ii defined by MLEE analysis.

PCR-RFLP analysis of symbiotic gene regionsPrimers FGPD 807 and FGPK 597 were used to amplify

IGS symbiotic regions betweennifD and nifK genes.

Four restriction endonucleases were used to digest the am-plified fragment from the 89 isolates. One to four distinct re-striction patterns, with three to seven restricted fragmentsper pattern, were detected with each restriction enzyme. Sta-tistical analysis of these profiles revealed less divergencethan that found in the 16S–23S IGS studies (Fig. 5). Thedendrogram, obtained by analysis ofnif patterns, had fourgroups: groups 1, 2, and 3, defined at 0.78 diversity, and amore divergent one represented by 13 isolates with 10 dis-tinct profiles. In fact, 33 profile types were obtained amongstthe isolates analysed, which may be subdivided in fourgroups. The first group is formed by 30 isolates having ninedifferent profile types. All of these isolates are from group i,already defined by MLEE or analysis of 16S–23S IGS. Thesecond group is represented by 27 isolates grouped in 9 dis-tinct profiles, but all of these isolates make part of group ii.In the third group, 13 isolates could be clustered in five dif-ferent profiles, and the majority of isolates were from groupi. Finally, the fourth group, the more heterogeneous one,contains 13 isolates, which are part of group ii with ten dis-tinct profiles.

Phylogenetic analysis of the different symbiotic typesshowed that the symbiotic types associated with chromo-somal group i were closely related, and were divergent fromthose found with chromosomal group ii. However, a uniqueexception was detected with 4 isolates from chromosomalgroup ii which were found to be associated to symbiotictypes related to members of chromosomal group i.

Our results reflected the variability of chromosomal DNAregions and the variability of symbiotic gene regions in alocal collection ofSinorhizobiumspecies. A dominant symbi-otic type was usually found in each of the twelve sites stud-ied.

Symbiotic effectivenessThe symbiotic effectiveness of fourMedicago species

with a sample of 25Sinorhizobiumisolates belonging to the18 identified plasmid types and to different subdivisions de-tected with MLEE and PCR/RFLP analysis was estimated.Analysis of plant dry matter yield showed that better resultswere obtained withM. scutelletacv. Kelson (Table 1). Bycontrast,M. ciliaris showed the smallest yield. Response ofM. sativa, to inoculation with different isolates seems to bemore homogeneous.

The tested isolates were found to nodulate the three otherMedicagospecies from which they were not isolated, thoughvariations in effectiveness were observed.

None of the strains were effective with all theMedicagospecies (Table 1). Three groups of inoculation, isolatestrapped fromM. sativa cv. Gabes andM. scutelleta cv.Kelson formed one inoculation group, which is more closelyrelated to isolates trapped fromM. truncatula than thosetrapped fromM. ciliaris.

Discussion

The plasmid content of 134 isolates from a local collec-tion of rhizobia nodulatingMedicagoshows high plasmidicdiversity with 18 profile types. The dominant type (type 5),represented by 37.3% of isolates, contained two mega-plasmids characteristic ofS. melilotistrains described earlier

© 2001 NRC Canada

144 Can. J. Microbiol. Vol. 47, 2001

Fig. 5. Dendogram (UPGMA) of genetic relationships betweenisolates ofSinorhizobiumnodulatingMedicagoin Tunisian soilsestimated by PCR-RFLP ofnif KD rDNA IGS.

by Rosenberg et al. (1982). However, the plasmid profilegroupings did not appear to be correlated with the pheno-typic traits of the bacteria analysed, but more with physicaland chemical proprieties of the soils from which they werederived because a higher diversity was found in the marginalsites.

Bromfield et al. (1987) and Broughton et al. (1987), ex-amined the diversity of plasmid content among field popula-tions of S. meliloti and found a wide variety of distinctplasmid profiles. Hartmann and Amarger (1991), showedthat DNA restriction patterns and IS fingerprints werestrongly correlated with the plasmid profiles of the strains,confirming the validity of plasmid grouping in assessing di-versity of natural populations ofS. meliloti.

Other studies have also suggested that symbiotic plasmidand chromosomal genotypes are correlated (Young andWexler 1988) and that plasmid sequences might be morevariable than chromosomal sequences (Paffetti et al. 1996).However, Laguerre et al. (1993) observed that the same typeof chromosomal background may harbour different plasmidtypes. This may be explained by lateral transfer of symbioticand (or) cryptic plasmid as suggested by Velazquez et al.(1995).

Therefore, plasmid analysis appeared to be a highly dis-criminating method forRhizobiumisolates. But in our case,

plasmid content seems not to be correlated to the other mo-lecular criteria used.

MLEE have allowed the distinction of two groups withhighly heterogeneous specificity from host and geographicalorigin of the rhizobial isolates. These two groups, designatedgroups i and ii, were defined on the basis of the analysis ofplasmidic and chromosomal markers, and of the symbioticproperties of different isolates. This result indicates thatstrains of Sinorhizobiumnodulating Medicago in Tunisiansoils are strongly heterogeneous in their chromosomal back-ground and show a high level of genetic diversity.

Groups i and ii apparently correspond to division A andB, as defined by Eardly et al. (1990) and Brunel et al.(1996), who concluded thatSinorhizobiumoriginating fromMedicago species were divided into two highly diverginggroups, which could coincide with different hostmicrosymbiont affinities.

Thus, this two-group classification is confirmed byanalysis of our local collection and is therefore validated by acombination of different techniques, as required for a poly-phasic taxonomic approach (Vandamme et al. 1996). Group ifrom our collection (group A described by Eardly et al.1990; group I of Brunel et al. 1996 and group 1 of Rome etal. 1996a) may correspond toS. meliloti. Group ii (group Bof Eardly et al. 1990; group II of Brunel et al. 1996 and

© 2001 NRC Canada

Jebara et al. 145

Plant dry matter ofMedicagospecies (mg/plant ± S.D.;P < 0.01)

Strains M. sativa M. scutelleta M. truncatula M. ciliaris

3*G2 188 ± 17 422 ± 98 171 ± 36 118 ± 117G**2 198 ± 16 331 ± 88 178 ± 31 51 ± 77T3 152 ± 15 350 ± 78 341 ± 121 202 ± 2113G1 167 ± 25 515 ± 181 208 ± 47 31 ± 813K1 137 ± 11 482 ± 131 209 ± 49 36 ± 419G1 153 ± 21 415 ± 108 188 ± 41 48 ± 219T3 161 ± 20 352 ± 91 363 ± 102 129 ± 2424K3 157 ± 22 502 ± 173 301 ± 88 42 ± 325G3 159 ± 23 493 ± 141 45 ± 4 34 ± 1025T2 111 ± 24 373 ± 121 358 ± 97 218 ± 2529C1 44 ± 11 117 ± 24 277 ± 57 252 ± 4329C3 41 ± 8 116 ± 12 273 ± 53 269 ± 3729T1 121 ± 10 362 ± 143 352 ± 101 209 ± 3132C1 38 ± 8 138 ± 12 301 ± 79 281 ± 4432K1 178 ± 25 526 ± 190 321 ± 83 46 ± 440C3 151 ± 21 195 ± 25 281 ± 67 265 ± 3340K1 131 ± 27 498 ± 143 211 ± 43 47 ± 443C1 46 ± 4 111 ± 26 288 ± 65 273 ± 5143T1 182 ± 19 206 ± 41 356 ± 102 210 ± 2954G3 167 ± 32 462 ± 129 190 ± 23 56 ± 854K2 171 ± 24 473 ± 138 248 ± 37 118 ± 3060C1 102 ± 19 368 ± 103 291 ± 83 275 ± 4760C3 109 ± 28 343 ± 91 264 ± 43 276 ± 4360G3 227 ± 22 381 ± 112 63 ± 9 104 ± 1260T1 42 ± 8 318 ± 82 328 ± 83 232 ± 25SU47 185 ± 25 321 ± 79 279 ± 69 188 ± 36Control with N (20 mM) 249 ± 39 541 ± 207 370 ± 133 291 ± 43Control non inoculated 21 ± 2 106 ± 13 27 ± 4 23 ± 2

*Characteristic number of site**Host plant strain isolated from (C,Medicago ciliaris; G, Medicago sativacv. Gabes; K,Medicago scutelletacv. Kelson;

T, Medicago truncatula).

Table 1. Symbiotic effectiveness of 25 isolates ofSinorhizobium.

group 2 of Rome et al. 1996a) represents a separate genomicspecies identified asS. medicaeby Rome et al. (1996b).

Group ii could be a lower competitor than group i on themedics tested. Although group i of rhizobia was able to ef-fectively nodulate the perennial lucerneM. sativaand someother medics, they were not recovered from them. This sug-gestion of a difference in competition ability is not in contra-diction with characterisation by MLEE (Eardly et al. 1990).For strains collected fromM. sativa and M. scutelleta,itshould be noted that grouping does not correlate stronglywith geographical origins, since strains obtained from widelyseparated areas were found in the same group, and strainsfrom the same area were included in both group i and ii.

Cross-inoculation data have shown that annualMedicagospecies exhibit various degrees of specificity in their symbi-oses (Provorov et al. 1994; Brunel et al. 1996). Our resultsconfirmed these observations and this specificity is better il-lustrated by isolates trapped fromM. ciliaris, which may bea discriminating criteria to distinguishS. meliloti fromS. medicae. Relationship analysis between geographic originof isolates and their genomic structure seems to indicate thatone dominant symbiotic type exists in ten of twelve studiedsites. Yet, all molecular polymorphisms detected were notrelated to symbiotic properties. Strains within the i and iigroupings could not be related to the nodulation and fixationtest results, but could correspond to other biological proper-ties.

Paffetti et al. (1996), investigating genetic diversityofRhizobiumnodulating four varieties ofM. sativa in Italiansoil, showed that the genetic diversity of the population ofS. melilotiresides on the whole bacterial genome, but seemsto be distributed either on the chromosome (IGS) or on thesymbiotic plasmid.

A plasmidic marker might have been more informativethan a chromosomal one for screening symbiotic properties.Nevertheless, the discriminating power between the twomolecular markers was slightly better with the amplified ribo-somal intergenic region than those ofnifKD genes, and divi-sions detected bynifKD amplified DNA analysis matchedthose established by ribosomal PCR–RFLPs. The correlationbetween thenif and the rDNA IGS matrix was relativelylow, although with significant value. Our results confirmedthe observation by Nour et al. (1994), that intergenic rDNAregions are excellent targets to discriminate closely relatedspecies among rhizobia nodulatingMedicagospecies. Alsowe noted clear distinction between the two genomic speciesof S. meliloti and S. medicaeby analysis of IGS symbioticregions betweennifD and nifK genes.

In this study the divergence between groups i and ii canreflect geographic and environmental differences. Harrisonet al. (1989), analysed genetic variation in populations ofR. leguminosarumbv. trifolii and demonstrated the effect ofsoil pH as a determinant factor. Thus, distribution of rhizo-bial populations can be influenced by other factors.

The high diversity observed among genus ofMedicagoseems to be associated with a high level of diversity ofSinorhizobiumisolates. However, it should be noted that iso-lates were obtained only from nodules. These isolates mayrepresent only a fraction of the total rhizobial population.

Our results showed that all field isolates demonstrated ahigh level of symbiotic effectiveness with a relative affinity

to their origin host specificity. Eighty-four percent of iso-lates were cultured from annual medics in Tunisian soil.This result is in agreement with observation of Eardly et al.(1990) suggesting specificity of those isolates to annualMedicagospecies.

The comparison of different techniques revealed thatMLEE was the most powerful method to investigate intrinsiccharacteristics of isolates as determined by phenotypic tests.However PCR/ RFLP classification is more influenced bysites of origin, therefore plasmid content analysis better re-flects the nature and characteristics of the soil.

Acknowledgments

This work was supported by Secrétariat d’Etat à la Re-cherche Scientifique et à la Technologie PNM (P96BAP25)and International foundation for Science IFS (C/2237-2). Weare grateful to Dr. G. Laguerre for correction and helpfuldiscussion.

References

Beck, D.P., Materon, L.A., and Afandi, F. 1993. PracticalRhizobium-legume technology manual. ICARDA, Aleppo, Syria.

Beringer, J.E. 1974. R-factor transfer inRhizobium leguminosarum.J. Gen. Microbiol.84: 188–198.

Bromfield, E.S.P., Lewis, D.M., and Barran, L.R. 1985. Crypticplasmid and rifamicin resistance inRhizobium melilotiinfluenc-ing competitiveness. J. Bacteriol.164: 410–413.

Bromfield, E.S.P., Thurman, N.P., Whitwill, S.T., and Barran, L.R.1987. Plasmids and symbiotic effectiveness of representativephage types from two indigenous populations ofRhizobiummeliloti. J. Gen. Microbiol.133: 3457–3466.

Bromfield, E.S.P., Barran, L.R., and Wheatcroft, R. 1995. Relativegenetic structure of a population ofRhizobium melilotiisolateddirectly from soil and from nodules of alfalfa (M. sativa) andsweet clover (Melilotus alba). Mol. Ecol. 4: 183–188.

Broughton, W.J., and Dilworth, M.J. 1971. Control ofleghaemoglobin synthesis in snake beans. Biochem. J.125:1075–1080.

Broughton, W.J., Samrey, U., and Stanley, J. 1987. Ecological genet-ics of Rhizobium meliloti: Symbiotic plasmid transfer in theMedicago sativarhizosphere. FEMS Microbiol. Lett.40: 251–255.

Brunel, B., Rome, S., Ziani, R., and Cleyet-Marel, J.C. 1996. Com-parison of nucleotide diversity and symbiotic properties ofRhizobium meliloti populations from annual species ofMedicago. FEMS Microbiol. Ecol.19: 71–82.

Eardly, B.D., Materon, L.A., Smith, N.H., Johnson, D.D.A.,Rumbaugh, M.D., and Selander, R.K. 1990. Genetic structure ofnatural populations of the nitrogen-fixing bacteriumRhizobiummeliloti. Appl. Environ. Microbiol.56: 187–194.

Eckhardt, T. 1978. A rapid method for the identification of plasmiddeoxyribonucleic acid in bacteria. Plasmid,1: 584–588.

F.A.O. 1988. Programme de coopération technique. Programme dedéveloppement des productions fouragères et de l’elevage. Rap-port de synthèse, volume n°1.

Harrison, S.P., Jones, D.G., and Young, J.P.W. 1989.Rhizobiumpop-ulation genetics: Genetic variation within and between popula-tions from diverse locations. J. Gen. Microbiol.135: 1061–1069.

Hartmann, A., and Amarger, N. 1991. Genotypic diversity of an in-digenous Rhizobium meliloti field population assessed byplasmid profiles, DNA fingerprinting, and insertion sequencetyping. Can. J. Microbiol.37: 600–608.

© 2001 NRC Canada

146 Can. J. Microbiol. Vol. 47, 2001

© 2001 NRC Canada

Jebara et al. 147

Hedrick, P.W., and Thomson, G. 1986. A two-locus neutrality test:Applications to humans,E. coli, and lodgepole pine. Genetics,112: 135–156.

Jamann, S., Fernandez, M.P., and Normand, P. 1993. Typingmethod for N2-fixing bacteria based on PCR-RFLP- applicationto the characterization ofFrankia strains. Mol. Ecol.2: 17–26.

Jordan, D.C. 1984. Family III.Rhizobiaceae. In Bergey’s Manualof Systematic Bacteriology.Edited byN.R. Krieg and J.G. Holt.Williams and Wilkins, Baltimore, Maryland. pp. 234–242.

Knauf, V.C., and Nester, E.W. 1982. Wide host range cloning vec-tors: A cosmid clone bank of anAgrobacteriumTi plasmid.Plasmid,8: 45–54.

Laguerre, G., Geniaux, E., Mazurier, S.I., Rodriguez Casartelli, R.,and Amarger, N. 1993. Conformity and diversity amongRhizobium leguminosarumbv. viciae, bv. trifolii , and bv.phaseolirevealed by DNA hybridization using chromosome andplasmid probes. Can. J. Microbiol.39: 412–419.

Laguerre, G., Mavingui, P., Allard, M.R., Charnay, M.P., Louvrier,P., Mazurier, S.I., Rigottier-Gois, L., and Amarger, N. 1996.Typing of rhizobia by PCR DNA fingerprinting and PCR- re-striction fragment length polymorphism analysis of chromo-somal and symbiotic gene regions: Application toRhizobiumleguminosarum and its different biovars. Appl. Environ.Microbiol. 62: 2029–2036.

Lesins, K.A., and Lesins, I. 1979. GenusMedicago(Leguminosae):A taxogenetic study. Dr. W. Junk bv Publishers, The Hague, TheNetherlands.

Lesley, S.M. 1982. A bacteriophage typing system forRhizobiummeliloti. Can. J. Microbiol.28: 180–189.

Martinez-Romero, E., Caballero-Mellado, J. 1996.Rhizobiumphylogenies and bacterial genetic diversity. Crit. Rev. Plant Sci.15: 113–140.

Milkman, R., Bridges, M.M. 1990. Molecular evolution of theEscherchia colichromosome. III. Clonal frames. Genetics,126:505–517.

Navarro, E., Simonet, P., Normand, P., Bardin, R. 1992. Character-ization of natural populations ofNitrobacter spp. usingPCR/RFLP analysis of the ribosomal intergenic spacer. Arch.Microbiol. 31: 1730–1734.

Nelson, K., Whittam, T.S., and Selander, R.K. 1991. Nucleotidepolymorphism and evolution in the glyceraldehyde-3-phosphatedehydrogenase gene (gapA) in natural populations ofSalmo-nella and Escherchia coli. Proc. Natl. Acad. Sci. U.S.A.88:6667–6671.

Normand, P., Cournoyer, B., Simonet, P., and Nazaret, S. 1992.Analysis of a ribosomal RNA operon in the actinomyceteFrankia. Gene,111: 119–124.

Nour, S.M., Cleyet-Marel, J.C., Beck, D., Effosse, A., andFernandez, M.P. 1994. Genotypic and phenotypic diversity ofRhizobiumisolated from chickpea (Cicer arietinumL.). Can. J.Microbiol. 40: 345–354.

Paffetti, D., Scotti, C., Gnocchi, S., Fancelli, S., and Bazzicalupo,M. 1996. Genetic diversity of an italianRhizobium melilotipop-ulation from differentMedicago sativavarieties. Appl. Environ.Microbiol. 62: 2279–2285.

Pinero, D., Martinez, E., and Selander, R.K. 1988. Genetic diversityand relationships among isolates ofRhizobium leguminosarumbv.phaseoli. Appl. Environ. Microbiol.54: 2825–2832.

Ponsonnet, C., and Nesme, X. 1994. Identification ofAgrobacteriumstrains by PCR/RFLP analysis of pTi and chro-mosomal regions. Arch. Microbiol.161: 300–309.

Provorov, N.A. 1994. The interdependence between taxonomy oflegumes and specificity of their interaction with rhizobia in rela-tion to evolution of the symbiosis. Symbiosis,24: 183–200.

Rinaudo, G., Fernandez, M.P., Effosse, A., Picard, B., and Bardin,R. 1993. Enzyme polymorphism ofAzorhizobiumstrains andother stem and root nodulating bacteria isolated fromSesbaniarostrata. Res. Microbiol.144: 55–67.

Rome, S., Brunel, B., Normand, P., Fernandez, M.A., and Cleyet-Marel, J.C. 1996a. Evidence that two genomic species ofRhizobium are associated withMedicago truncatula. Arch.Microbiol. 165: 285–288.

Rome, S., Fernandez, M.A., Brunel, B., Normand, P., and Cleyet-Marel, J.C. 1996b. Sinorhizobium medicaesp. nov., isolatedfrom annualMedicagospp. Int. J. Syst. Bacteriol.4: 972–980.

Rosenberg, C., Casse-Delbart, F., Dusha, I., David, M., andBoucher, C. 1982. Megaplasmids in the plant-associated bacteriaRhizobium meliloti and Pseudomonas solanacearum. J.Bacteriol.150: 402–406.

Selander, R.K., Caugant, D.A., Ochman, H., Glimour, M.N., andWhittam, T.S. 1986. Methods of multilocus enzyme electropho-resis for bacterial population genetics and systematics. Appl.Environ. Microbiol. 51: 873–884.

Sneath, P.H.A., and Sokal, R.R. 1973. Numerical taxonomy. Theprinciples and practice of numerical classification. W.H. Free-man & Co., San Francisco.

Vandamme, P., Pot, B., Gillis, M., De Vos, P., Kersters, K. andSwings, J. 1996. Polyphasic taxonomy, a consensus approach tobacterial systematics. Microbiol. Rev.2: 407–438.

Velazquez, E., Mateos, P.F., Pedrero, P., Dazzo, F.B., and Martinez-Molina, E. 1995. Attenuation of symbiotic effectiveness byRhizobium melilotiSAF22 related to the presence of a crypticplasmid. Appl. Environ. Microbiol.61: 2033–2036.

Vincent, J.M. 1970. A manual for the practical study of the root-nodule bacteria. Blackwell Scientific Publications, Oxford.

Weisburg, W.G., Barns, S.M., Pelletier, D.A., and Lane, D.J. 1991.16S ribosomal DNA amplification for phylogenetic study. J.Bacteriol.43: 305–313.

Willems, A., and Collins, M.A. 1993. Phylogenetic analysis ofrhizobia and agrobacteria based on 16S rDNA gene sequence.Int. J. Sys. Bacteriol.43: 305–313.

Yanagi, M., and Yamasoto, K. 1993. Phylogenetic analysis of thefamily Rhizobiaceaeand related bacteria by sequencing of 16SrRNA gene using PCR and DNA sequencer. FEMS Microbiol.Lett. 107: 115–120.

Young, J.P.W., Wexler, M. 1988. Sym plasmid and chromosomalgenotypes are correlated in field populations ofRhizobiumleguminosarum. J. Gen. Microbiol.134: 2731–2739.