APPLIED AND ENVIRONMENTAL MICROBIOLOGY, May 1993, p. 1310-13170099-2240/93/051310-08$02.00/0Copyright © 1993, American Society for Microbiology

Correlative Association between Resident Plasmids and theHost Chromosome in a Diverse Agrobacterium

Soil PopulationHACENE BOUZAR,lt* DJAOUIDA OUADAH,1 ZOULIKHA KRIMI,' JEFFREY B. JONES,2

MAURIZIO TROVATO,3t ANNIK PETIT,3 AND YVES DESSAUX3

Institut d'Agronomie, Universite des Sciences et Technologies de Blida, 09000 Blida, Algeria'; Gulf CoastResearch and Education Center, University ofFlorida, Bradenton, FL 342032; and Institut des Sciences

Vegetales, Centre National de la Recherche Scientifique, F-91198 Gif-sur-Yvette, France3

Received 4 January 1993/Accepted 25 February 1993

Soil samples collected from a fallow field which had not been cultivated for 5 years harbored a populationofAgrobacterium spp. estimated at 3 x 107 CFU/g. Characterization of 72 strains selected from four differentisolation media showed the presence of biovar 1 (56%) and bv. 2 (44%) strains. Pathogenicity assays on fivedifferent test plants revealed a high proportion (33%) of tumorigenic strains in the resident population. Alltumorigenic strains belonged to bv. 1. Differentiation of the strains by restriction fragment length polymor-phism analysis, sodium dodecyl sulfate-polyacrylamide gel electrophoresis of cellular proteins, and utilizationpatterns of 95 carbon substrates (Biolog GN microplate) revealed a diversified bv. 1 population, composed offive distinct chromosomal backgrounds (chr A, C, D, E, and F), and a homogeneous bv. 2 population (chr B).chr A, B, C, and D were detected at similar levels throughout the study site. According to opine metabolism,pathogenicity, and agrocin sensitivity, chrA strains carried a nopaline Ti plasmid (pTi), whereas chr C strainshad an octopine pTi. In addition, four of six nontumorigenic bv. 1 strains (two chr D, one chr E, and one chrF) had distinct and unusual opine catabolism patterns. chr B (bv. 2) strains were nonpathogenic andcatabolized nopaline. Although agrocin sensitivity is a pTi-borne trait, 14 chr B strains were sensitive to agrocin84, apparently harboring a defective nopaline pTi similar to pAtK84b. The other two chr B strains wereagrocin resistant. The present analysis of chromosomal and plasmid phenotypes suggests that in thisAgrobacterium soil population, there is a preferential association between the resident plasmids and theirbacterial host.

Agrobacterium species are gram-negative soil bacteriabelonging to the family Rhizobiaceae (29). This genus iscomposed of at least three basic chromosomal groups whichhave been allocated the rank of biovar (29). Each biovar isidentified by an array of biochemical and physiological tests(36). Some Agrobacterium strains can induce a neoplasticdisease, known as crown gall, on a large number of plantfamilies (15). This tumorigenic ability is borne on largeextrachromosomal DNA elements, called Ti plasmids (pTi),which are transferable by conjugation to nontumorigenicAgrobacterium strains (56). During the infection process, aspecific DNA fragment (the T-DNA) is transferred from thepTi to wounded plant cells and integrated into the nucleargenome (12). Transfer of T-DNA is directed by a set ofvirulence (vir) genes located in the nontransferred region ofthe pTi (54). The T-DNA carries both oncogenes, whichperturb plant cell division (1, 50), and genes coding for thesynthesis of low-molecular-weight compounds, termedopines, specifically metabolized by the tumor-inducing bac-terium for its growth (4, 44). Some opines also induceconjugal transfer of the pTi to nontumorigenic agrobacteria(46). However, different opines are synthesized in tumorsincited by different pTi, and induction of conjugation is

* Corresponding author.t Present address: Gulf Coast Research and Education Center,

University of Florida, 5007 60th St. E., Bradenton, FL 34203.Present address: Dipartimento di Genetica e Biologia Molecu-

lare, Universita di Roma "La Sapienza," 00185 Rome, Italy.

relatively specific. Octopine induces conjugation by strainscarrying an octopine-type pTi, whereas nopaline strains are

not induced to conjugate by octopine or even nopaline but byanother opine, agrocinopine A, which is also present intissue transformed by T-DNA of nopaline-type pTi. Thisselective opine induction contributes to the propagation ofpathogenicity genes and, consequently, plays a subtle role inthe ecology of this bacterium and the epidemiology of thedisease it causes (16).Crown gall infection may have significant detrimental

effects on the host plant, generating major economic losses(25, 37, 51). To evaluate the potential for crown gall epidem-ics and the possible efficacy of control measures such as theintroduction of antagonists (e.g., strain K84 [27]), moreinformation on the behavior of this bacterium and of its pTioutside the tumor environment is needed. At present, infor-mation on survival potential and genetic stability of tumori-genic agrobacteria in the soil environment is, at best, limited.This paucity of information is due in part to the raredetection of tumorigenic strains in the soil and on roots, eventhough nontumorigenic agrobacteria are commonly found (2,7, 10, 35, 52). These apparently low levels of tumorigenicbacteria in soil are in sharp contrast to the regular occur-rence of crown gall epidemics and the efficacious protectionagainst soil infections by the biocontrol agent K84 (5, 35).The need for a better understanding of the ecology of

agrobacteria and resident pathogenic plasmids led us toexamine the composition of an Agrobacterium soil popula-tion and the type of plasmids associated with the hostbacterium. This bacterial population was indigenous to a soil

1310

Vol. 59, No. 5

on July 6, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

CORRELATIVE ASSOCIATION BETWEEN PLASMIDS AND CHROMOSOME 1311

which was fallow for 5 years after 2 years of stone-fruit treenursery production.(A portion of this work was presented at the Annual

Meeting of the American Phytopathological Society, St.Louis, Mo. [8].)

MATERIALS AND METHODS

Isolation and culture of bacteria. Bacteria were isolatedfrom soil samples collected at the experiment station of theInstitut Technique d'Arboriculture Fruitiere et de Viticul-ture located in the Mitidja plain (Algeria). Four samplingsites were randomly chosen from a plot (32 by 38 m).Samples for analysis were taken at a soil depth of 5 to 10 cm.Soil analysis revealed a loamy texture and an alkalinereaction (pH 8.5). From each sample, 1.0 g of soil wassuspended by shaking in 10 ml of sterile distilled water, andthe suspension was allowed to settle for 5 min. Tenfold serialdilutions were made, and a 100-pl aliquot of selected dilu-tions was spread onto duplicate plates containing the follow-ing media: 1A (selective for bv. 1) (9), 2E (selective for bv.2) (9), RS (selective for bv. 3) (48), and MG (nonselective)(24). Plates were incubated for 5 days at 27°C before colonycounts were made. From each soil sample, 10 coloniesresemblingAgrobactenum spp. (36) were selected from eachmedium and purified by successive streakings on potatodextrose agar (PDA; Difco Laboratories, Detroit, Mich.)supplemented with 0.05% (wt/vol) calcium carbonate andKing's medium B (30). Gram-negative strains (20), nonfluo-rescent on King's medium B, were stored at 4°C on PDA-CaCO3 slants and in sterile distilled water (36).

Bacterial identification. The biovar affiliation of each puta-tive Agrobacterium strain was determined by classical bio-chemical and physiological tests (36). To determine thetaxonomic structure of this Agrobactenum community be-low the biovar level, subpopulations were differentiated bymorphology of colonies growing on PDA-CaCO3 at 27°C,profiles of cellular proteins separated by sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE),and profiles of HindIII- and BamHI-digested genomic DNAfrom individual strains. In addition, strains representative ofeach subpopulation were compared by their ability to utilizeas a sole carbon source 95 substrates from the Biolog GNmicroplates (Biolog Inc., Hayward, Calif.); incubation, in-oculation, data acquisition, and analysis were done as de-scribed in the manufacturer's recommendations. Utilizationprofiles were compared by using a newly developed Agro-bacterium data base (6a) and subjected to cluster analysis byusing the Biolog MLCLUST program for phenotypic group-ing.

Cellular proteins were extracted from bacteria grown tothe stationary phase in 1.5 ml of AT minimal medium (45)supplemented with mannose (2.0 g/liter), ammonium sulfate(2.0 g/liter), and yeast extract (0.1 g/liter). One milliliter ofbacterial suspension was transferred to a microcentrifugetube, and cells were recovered by centrifugation for 2 min at13,500 x g. Bacteria were washed once with 0.9% NaCl andtreated with 100 ,ul of lysis buffer (31) for 5 min at 100°C.Five microliters of lysate was electrophoretically separated(35 mA for 5 h) in a 12% acrylamide gel, and proteins werevisualized by using Coomassie blue staining.Genomic (total) DNA was extracted as described previ-

ously (18), endonuclease digested, and separated in an 0.8%agarose gel by using standard procedures (49). Total DNArestriction patterns were visualized by ethidium bromidestaining. When DNA hybridization was required, digested

genomic DNA was transferred to nylon membranes (Hy-bond N+; Amersham Co., Arlington Heights, Ill.) as de-scribed in the manufacturer's instructions. Chromosomalbackgrounds were compared by using a phosphomannoseisomerase (PMI) probe and a proline utilization (Put) probe.Both probes were cosmid clones selected from a genomicDNA library of the bv. 1 Agrobacterium strain C58. PMIconsisted of the labeled pCD1 clone, obtained by comple-menting a PMI mutation of Escherichia coli CD1 (23a). Putwas obtained by labelling the gel-purified insert of a cosmidclone complementing a mutation that abolished proline deg-radation in strain C58. Probes were labelled by randompriming by using a digoxygenin-modified nucleotide and adetection kit (Boehringer Mannheim Biochemicals, India-napolis, Ind.). Hybridization and detection of DNA duplexwere performed by using the X-pho-nitroblue tetrazoliumdye system of the digoxygenin detection kit. When neces-sary, filters were destained and dehybridized as described inthe manufacturer's instructions.

Characterization of plasmid phenotype. Several phenotypicfeatures known to be associated with pTi, such as pathoge-nicity, the presence of oncogenes and vir genes, opinesynthesis and catabolism, and agrocin sensitivity, wereidentified in the present Agrobacterium population.

Tumorigenicity of each strain was determined by inocu-lating scalpel-wounded tissue of tomato (Lycopersicon escu-lentum cv. Marmande) seedling stems and Kalanchoe dai-gremontiana leaves with 24-h-old bacterial growth asdescribed previously (36). Strains that did not induce tumorson these two plants were inoculated on sunflower (Helian-thus annuus cv. Nick 89-1), zucchini (Cucurbita pepo),tobacco (Nicotiana tabacum cv. Xanthi), Kalanchoe tubi-flora, Datura stramonium, and castor bean (Ricinus commu-nis). Plants were grown under long-day conditions (light, 16h) at 24°C (day) and 17°C (night). Tumor formation wasassessed visually 6 to 8 weeks after inoculation.The presence of opines in tumors was detected after

extraction from plant tissues, concentration, and separationby high-voltage paper electrophoresis at pH 1.9, as previ-ously described (43). Opines were revealed on the electro-pherograms by using various reagents. Mannityl opines weredetected by silver nitrate staining. Octopine and nopalinewere detected by using phenanthrene quinone. Cucumopinewas detected by using the Pauly reagent (for a review, seereference 16).

Opine utilization was tested by inoculating individualstrains into three degradation cocktails which differed inopine composition. Experiments were performed in a finalvolume of 100 ,ul. The cocktails consisted of AT minimalmedium supplemented with ammonium sulfate (1.0 g/liter),yeast extract (100 mg/liter), and the appropriate opines(provided by P. Guyon, Laboratoire de Biologie Cellulaire,Institut National de la Recherche Agronomique, Versailles,France). Degradation cocktail A contained the opines oc-topine (5 mM), nopaline (5 mM), and cucumopine (4 mM).Cocktail B was supplemented with agropine (2.5 mM) andmannopine (2.5 mM). Cocktail C contained the opine man-nopinic acid (5 mM). The inoculated cocktails were incu-bated at 28°C for up to 7 days. Opine utilization was assessedby investigating their disappearance from the degradationcocktails by high-voltage paper electrophoresis as describedabove.A feature associated with nopaline pTi is sensitivity of the

host bacterium to agrocin 84, a bacteriocin produced by thebiological control agent Agrobacterium radiobacter K84(23). Strains were tested for agrocin sensitivity on agar plates

VOL. 59, 1993

on July 6, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

1312 BOUZAR ET AL.

TABLE 1. Chromosome and plasmid phenotypes of 24 pathogenic and 22 nonpathogenic Agrobacterium strainsisolated from four soil samples

No. of strains from soil Chromosome Plasmid

Strain sampietype I III IV Biovar Subgroupb Agrocin Tumor Opinec

sensitivity induction Catabolized Synthesized

B17 (16)d 5 4 3 4 1 A + + N, Oe NB14 (8) 3 2 1 2 1 C - + O, M, Ma, A, Aa O, M, A, AaE35 (2) 1 1 1 D - - 0 NAfB11 (2) 1 1 1 D - - NAE34 (1) 1 1 E - - Aa NAF13 (1) 1 1 F - - C, N,O NAB16 (14) 4 4 4 2 2 B + - N, oe NAE44 (2) 2 2 B _ - N, oe NA

a Number of strains sharing the same phenotype recovered from each soil sample.b Chromosomal subgrouping by DNA and protein profiles (see Materials and Methods).Abbreviations: N, nopaline; 0, octopine; M, mannopine; Ma, mannopinic acid; A, agropine; Aa, agropinic acid; C, cucumopine.

d The number of strains with the same phenotype are shown in parentheses.e Octopine was used only in the presence of nopaline.f NA, not applicable.

by the method of Stonier (55). The agrocin-sensitive strainC58 and the agrocin-resistant derivative C58C1 (our collec-tion) were used as references.To determine whether nontumorigenic opine-utilizing

strains carried oncogenes or vir genes, genomic DNA digestswere hybridized, as described above, with gel-purified seg-ments of the T-DNA region (HindIlI fragment 1 whichencompasses tmsl, tmr, 6a, 6b, and ocs) and the vir region(HindIII fragment 3 which includes virG and most of the virBoperon) of the octopine pTi15955, respectively. Both probeswere obtained from HindIII digests of cosmid clones ofpTi15955 (17).To determine whether nopaline degradation in nontumor-

igenic agrobacteria was plasmid borne, plasmids present inthese strains were transferred by conjugation to the recipientstrain C58C1RSpAt-, a rifampin- and streptomycin-resistantmutant of a plasmid-free derivative of strain GMI9017 (47).Bacterial conjugation was performed by using a drop-matingtechnique. Donors were grown in AT medium supplementedwith nopaline (10 mM), whereas the recipient strain wasgrown in AT medium supplemented with mannitol (2.0g/liter), ammonium sulfate (2.0 g/liter), rifampin (0.1 mg/ml),and streptomycin (0.5 mg/ml). Drops of the donor log-phaseculture were spotted onto a freshly plated lawn of therecipient strain. The cross was performed directly on theselective medium which was AT supplemented with no-paline (10 mM), rifampin (0.1 mg/ml), and streptomycin (0.5mg/ml). Transconjugants were picked up after 6 days at28°C, purified under the same conditions on the same me-dium, and analyzed for the concomitant presence of aplasmid and the ability to catabolize nopaline. The plasmidcontent ofAgrobacterium strains was visualized in an 0.8%agarose gel by the Eckhardt procedure (22).

RESULTS

Agrobacterium-like colonies were recovered from all foursoil samples and at density levels estimated at about 3 x 107CFU per g of soil. Biochemical and physiological testsperformed on gram-negative nonfluorescent isolates ob-tained from purified Agrobactenum-like colonies identified40 bv. 1 and 32 bv. 2 isolates. Most bv. 1 strains wererecovered on medium 1A (24 of 40) and medium RS (13 of

40), whereas most bv. 2 strains were isolated from medium2E (19 of 32) and medium MG (9 of 32). Several Agrobacte-num-like colonies isolated on RS medium were fluorescentpseudomonads (data not shown).

Testing for pathogenicity and sensitivity to agrocin 84 ledto the identification of 24 tumorigenic and 30 agrocin-sensitive strains (Table 1). All 24 strains which induced thedevelopment of tumors on test plants belonged to bv. 1, ofwhich 16 were sensitive to agrocin. The 14 remaining agro-cin-sensitive strains were nonpathogenic bv. 2. These patho-genic and/or agrocin-sensitive strains were recovered fromall soil samples and in about the same proportions (Table 1).An assay for the presence of opines in tumors induced by

the 24 pathogenic bv. 1 strains revealed the presence of bothnopaline- and octopine-producing tumors. Nopaline wasdetected in tumors incited by 16 of the strains. Octopine wasdetected in tumors of the remaining eight strains. In additionto synthesizing octopine, these tumors synthesized the man-nityl opines mannopine, agropine, and agropinic acid (Table1).

Characterization below the biovar level by DNA digestionprofile analysis of the 24 tumorigenic bv. 1, the 14 nonpatho-genic and agrocin-sensitive bv. 2 strains, and 8 randomlyselected nonpathogenic and agrocin-resistant strains (6 bv. 1and 2 bv. 2 strains) revealed six distinct genomic fingerprints(Fig. 1). Five different chromosome backgrounds (chr A, C,D, E, and F) were identified among bv. 1 strains, and one(chr B) was identified among bv. 2 (Table 1; Fig. 1). Southernblot analysis with the bv. 1 chromosome probe PMI con-firmed the distinctiveness of each chromosome group withinbv. 1 strains. Strains with the same DNA digestion profileshared a unique hybridization signal (data not shown).Similar observations were made when the Put probe wasused on a limited number of strains representative of eachchromosomal group (Fig. 2).

Interestingly, neither the PMI (data not shown) nor the Put(Fig. 2) probe hybridized to chr B (i.e., bv. 2) strains. This isconsistent with a previous study which showed that DNA ofbv. 2 strains does not usually react with bv. 1 chromosomeprobes (33). When whole-cell proteins from these strainswere separated by SDS-PAGE, the resulting electrophoreticprofiles were identical for strains of the same chromosomebackground; each chromosome group had a unique profile

APPL. ENVIRON. MICROBIOL.

on July 6, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

CORRELATIVE ASSOCIATION BETWEEN PLASMIDS AND CHROMOSOME

FIG. 1. Agarose (0.8%) gel electrophoresis of HindIII-digestedDNA from Agrobacterium strains representative of chromosomalgroups A through F. Lanes: S, X phage DNA digested with HindIII;A, chr A strains B32 (lane 1) and G14 (lane 2); B, chr B strains E43(lane 1) and G22 (lane 2); C, chr C strains G42 (lane 1) and G44 (lane2); D, chr D strains E35 (lane 1) and Bl1 (lane 2); E, chr E strainE34; and F, chr F strain F13; C58, the reference nopaline strain; Cl,pTi-free derivative of C58, termed C58C1.

(Fig. 3). Cluster analysis of utilization patterns of 95 carbonsubstrates confirmed the segregation of the strains by theaforementioned chromosome grouping (Fig. 4); each clusterconsisted of strains that had distinct metabolic patterns.Colony morphology was also correlated to chromosomalbackground. Although colonies of all strains were cream-white in color, circular (i.e., round with smooth edges), andhad entire margins, several chromosome groups were distin-guishable after 48 h on PDA-CaCO3. Colonies of chr Astrains were 3 mm in diameter, umbonate, and presented theunusual property of producing a pellicle which held togetherand could not be easily dispersed. Chr D, E, and F strainshad the typical convex Agrobacterium colony morphology,and these colonies were 3 to 4 mm in diameter. Chr C strainsformed pulvinate (i.e., raised) colonies approximately 1.5mm in diameter. Chr B strains produced punctiform (i.e.,less-than-i-mm) colonies. After 4 days, chr B colonies werepulvinate and cleared the CaCO3 from the medium. Clearingof the medium is typical of bv. 2 strains and is the result ofacid production from glucose (6). Analysis of chromosomephenotypes showed the predominance among the 30 bv. 1strains of 16 chr A strains, followed by 8 chr C strains and 4chr D strains; chr E and F types were each identified onlyonce (Table 1). Strains with chr A, B, C, or D wererecovered at similar levels from all soil samples. The twounique chromosome types (chr E and F) were recoveredfrom different samples (Table 1).

Opine utilization analysis revealed the presence of differ-

FIG. 2. Southern blots of HindIII-digested DNA of representa-tive chr A, B, C, D, E, and F strains hybridized with the Put probe.This blot was prepared from the agarose gel shown in Fig. 1, and thelanes in this figure are the same as those described in the legend toFig. 1. White arrows on the left indicate the positions of theHindIII-digested A DNA fragments.

ent opine catabolic patterns (Table 1). The tumorigenic,agrocin-sensitive chr A strains utilized both nopaline andoctopine, but the latter was utilized only in the presence ofnopaline, a property common to nopaline-type pTi. Theremaining eight tumorigenic bv. 1 strains had patterns typi-

FIG. 3. SDS-polyacrylamide gel of whole-cell proteins fromAgrobacterinum strains representative of chromosomal groups Athrough F (the arrow indicates the direction of migration). The Tlanes contained the following protein molecular size markers (fromthe top): phosphorylase B (94 kDa), bovine serum albumin (67 kDa),ovalbumin (43 kDa), carbonic anhydrase (30 kDa), trypsin inhibitor(21 kDa), and lactalbumin (14.4 kDa).

VOL. 59, 1993 1313

on July 6, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

APPL. ENVIRON. MICROBIOL.

I.... ... I. .... I.... :....

A

I1Jc

uD3FJE

I53

B

K84

A

0

1 2 3 4 512345v 6 7 I 9 1s 11

v LVY12 13 14 IS 1IIVY v v

1

3

2

30 20 10 0DENDROGRAM DISTANCES

FIG. 4. Dendrogram of reference Agrobacterium strains andrepresentative soil strains based on utilization of 95 carbon sources(Biolog GN microplate). Letters (A to F) and numbers (1 to 3) to theright of the brackets represent chromosome subgroups and biovars,respectively. The following strains were included as references:C58, B6, and FACH (biovar 1); AB2/73, AR5/K, and K84 (biovar 2);and, CG660, 3/2, and CG1005 (biovar 3).

cal of octopine pTi: they catabolized octopine and themannityl opines (i.e., mannopine, mannopinic acid, agro-pine, and agropinic acid) and were resistant to agrocin. Alleight octopine strains were of the chr C type. Interestingly,opine utilization by the six nonpathogenic bv. 1 agrobacteriarevealed the presence of strains diverse in opine metabolism.Two chr D strains degraded octopine but differed fromregular octopine-utilizing strains by their inability to utilizeany of the mannityl opines. The other two chr D strains wereunable to utilize any of the opines used in the present study.The strain with chr E catabolized agropinic acid but none ofthe other mannityl opines. The other bv. 1 strain with aunique chromosome background (chr F) had the unusualability to metabolize cucumopine, nopaline, and octopine.Fourteen of the bv. 2 (chr B) strains had utilization patternstypical of nopaline pTi. These strains were agrocin sensitiveand utilized nopaline and octopine, the latter only in thepresence of nopaline. However, these strains were unable toinduce tumors on any of the plants used in pathogenicityassays. The last two chr B strains were undistinguishablefrom the other chr B strains, except for their resistance to

I v v v vI

ILsC$4F

*,A:6-v _ _e_ +, R .FIG. 5. HindIII-digested DNA hybridized with the vir probe. (A)

chr A strains (lanes 1 to 16); (B) chr C strains (lanes 1 to 8), X DNA(lane 9), and strain C58 (lane 10). The symbol 0 and the arrowindicate the origin and direction of migration, respectively.

agrocin. Southern blot hybridization of digested DNA fromstrains representative of the different chr groups withT-DNA and vir probes corroborated the pathogenicity tests.On the basis of the position of the hybridization signal,nopaline pTi present in chr A strains were identical or veryclosely related and the same held true for octopine pTi of chrC strains (Fig. 5). DNA from the nonpathogenic opine-utilizing strains did not hybridize to either the vir or T-DNAprobes (not shown), confirming the nontumorigenic nature ofthe strains with chr B, D, E, or F. To determine whether theability of chr B strains to catabolize nopaline was plasmidborne, selected strains were conjugated with the plasmid-free strain C58C1RSpAt- that does not utilize opines. Theresulting transconjugants were selected on nopaline andfound to harbor one of two plasmids detected in the selecteddonor strains (data not shown), clearly demonstrating thatnopaline utilization by chr B strains was plasmid borne.Strains carrying a nopaline pTi, an octopine pTi, or anontumorigenic nopaline plasmid were recovered from thedifferent sampling sites at similar levels (Table 1).

DISCUSSIONThe present study assessed the distribution and diversity

of the Agrobactenum population of an alkaline soil whichwas continuously fallow for 5 years and where there wasapparently no galled host. The soil in this fallow field

1314 BOUZAR ET AL.

on July 6, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

CORRELATIVE ASSOCIATION BETWEEN PLASMIDS AND CHROMOSOME 1315

harbored a population of agrobacteria estimated at 3 x 107CFU/g. One in three Agrobacterium strains induced tumorson at least one of the test-plant species. To our knowledge,this is the first report of a soil with a tumorigenic populationestimated at 107 CFU/g. Previously, populations of agrobac-teria found in soil have been estimated to vary between 103and 106 CFU/g, with pathogenic strains seldom being de-tected except in the vicinity of infected roots (7, 10, 52).Although artificially introduced tumorigenic strains survivepoorly in soil devoid of susceptible hosts (3, 19, 35, 41),naturally occurring pathogenic agrobacteria have been re-ported in some instances to persist for years in the soil undernatural conditions (14, 52). The apparent maintenance ofsuch a large population of tumorigenic strains in the presentfallow field may be due to properties intrinsic to thatparticular soil and/or to the rhizosphere of some of the weedsgrowing in that field. One edaphic factor which may havecontributed to the survival of thisAgrobactenium populationis the loamy texture of the soil. Because of its highercapacity for water retention, such a texture has been shownto contribute to the multiplication and propagation of bac-teria (32). Another parameter that may have promotedsurvival of the tumorigenic strains in this soil is pH. Acid pHhas been reported to reduce survival of introduced patho-genic agrobacteria in the absence of a host (19), and crowngall has been reported to be more serious in regions wherethe soil is alkaline (34, 53). For instance, the disease is rarelyobserved in the Willamette Valley of Oregon where soils areacidic, whereas infections are common in alkaline soils ofthe Sacramento River valley of California (53) and centralWashington (34). It has also been shown that alkaline pHpositively influences survival and nodulation ability of Rhi-zobium spp. (21), soil bacteria closely related to Agrobacte-rium spp.On the basis of biochemical tests, the resident Agrobac-

terium population was composed predominantly of bv. 1isolates. In addition to bv. 1, bv. 2 appears to be the onlyother biovar present in this fallow soil. The higher rate ofrecovery of bv. 1 and bv. 2 strains on medium 1A and 2E,respectively, confirmed the selectivity of these two media.Recovery from RS medium of bv. 1 strains and of a largenumber of fluorescent pseudomonads, which had coloniesthat appeared identical on that medium to those ofAgrobac-terium spp., indicates that this medium should be used withcaution in studies targeting recovery of bv. 3 strains fromsoil. Selective isolation of bv. 1 and/or pseudomonads bythis medium has been reported elsewhere (5, 10). It shouldbe noted, however, that RS and MG media were useful incomplementing the other two media for the isolation ofadditional strains, some of which had peculiar chromosomeand plasmid phenotypes. Surprisingly, all pathogenic strainsin this soil were bv. 1, whereas bv. 2 pathogens predomi-nated in tumors of nursery trees cultivated in the sameregion (5), apparently confirming the fitness of bv. 1 in soil(38, 39). The prevalence of bv. 1 in our soil may have beenthe result of the differential motility reported among agro-bacteria; indeed, bv. 1 strains are more motile in alkaline pHthan bv. 2 strains, whereas bv. 3 strains lack motility at highpH (11). It should be noted, however, that when tworepresentative bv. 1 strains from this soil were tested underlaboratory conditions, they grew better at pH 7 than at pH8.5 (hla).

Cellular protein patterns correlated well with digestedgenomic DNA patterns; such a correlation between DNAand protein fingerprints was not surprising since proteinsresult from the expression of genetic information. The same

groupings were obtained from a comparison of utilizationpatterns of 95 carbon substrates. By using these techniques,six distinct chromosome backgrounds were identified in thiscollection of strains. The bv. 1 population was diverse, andits strains were segregated into five distinct chromosomalbackgrounds (chr A, C, D, E, and F), whereas bv. 2 strainsformed a homogeneous cluster (chr B). In addition, therewas a good relationship between colony morphology andchromosomal background. With the exception of the six chrD, E, and F strains which produced identical colonies onPDA-CaCO3, strains of other chromosome groups had col-onies with distinct features. Such differences in colonymorphology reflect apparently fundamental differences be-tween strains and may serve as indicators of genetic diver-sity. Differences in protein profiles, genomic DNA, carbonmetabolism, and colony morphology suggest that thesesubpopulations are distinct and probably differ in a numberof physiological characteristics of unknown importance forsurvival fitness and pathogenic ability. chr A, B, C, and Dwere the most common chromosome groups and wereevenly distributed throughout the field. The rare occurrenceof strains with chr E and F did not allow us to determinewhether these atypical bv. 1 strains were present in othersites; interestingly, these two strains have rare or uniqueopine catabolic patterns. More work is needed to determinewhether all of these unusual opine degradation properties areplasmid borne or chromosomally encoded.On the basis of opine metabolism, pathogenicity, reaction

to agrocin 84, and hybridization with T-DNA and vir probes,three plasmid types were found in relative abundance in allsampling sites. The most common one was the nopaline pTi,followed by both the octopine pTi and a plasmid similar topAtK84b (also called pNoc), which is a nonpathogenic,nopaline-agrocinopine plasmid present in strain K84 (13). Inaddition to nopaline utilization, this nonpathogenic plasmidshares with pAtK84b a similar-size plasmid and plasmidtransfer induced by nopaline (15a); transfer of nopaline pTi isgenerally induced by agrocinopine A or B. The relativeabundance of the pAtK84b-type plasmid was surprising andremains unexplained, although it may be correlated with therelatively high abundance of the pathogenic nopaline-typeplasmid in the population. All analyzed bv. 2 strains isolatedin this study harbored a second plasmid, the function ofwhich is unknown. Preliminary analysis performed on alimited number of strains showed that this plasmid does notencode for synthesis of an agrocin (data not shown).

In this study of indigenous agrobacteria and their opinecatabolic plasmids, the type of plasmid maintained in abacterial cell was correlated with the chromosomal back-ground. Indeed, bv. 1 strains with chr A carried a nopaline-type pTi, while those with chr C had an octopine-type pTi;bv. 2 (chr B) strains harbored a pAtK84b-type plasmid(Table 1). Although this population could reflect the types ofstrains that were introduced to the field during the years ofstone-fruit nursery production, it seems improbable thatplasmid exchange would not have occurred in the presenceof such highly susceptible hosts unless some type of selec-tion pressure was occurring. The affinity of a pTi for aparticular chromosomal background was previously sus-pected in a survey of pathogenic strains, where it wasnoticed that bv. 2 strains were three times more likely tocarry a nopaline-agrocinopine-type pTi than bv. 1 strains(5). Also, when pTi from a collection of bv. 3 strains wereanalyzed for the presence of two related insertion elements,it was suggested that vitopine, octopine-cucumopine, andnopaline pTi are not or only rarely transmitted to strains with

VOL. 59, 1993

on July 6, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

APPL. ENVIRON. MICROBIOL.

a different chromosomal background (42). In apparent con-

tradiction to the findings described above, plasmid exchangewas suggested when analysis of strains from a forest nursery

revealed the presence of the same plasmids in differentchromosomal backgrounds (33). These apparently conflict-ing results may reflect different behaviors ofAgrobacteriumpathogenic plasmids in different environmental conditions.In a tumor, plasmids conjugate between bacteria, as seen

experimentally in the Kerr cross (26, 28); in such an envi-ronment, opines would induce the conjugal transfer of plas-mids and would likely provide valuable carbon and energy

sources to account for the maintenance of the plasmids indifferent chromosomal backgrounds. In a gall-free, opine-free environment, there is no apparent selective pressure fortransfer and propagation of the pathogenic plasmids; how-ever, under such conditions, some plasmid-chromosomebackground combinations are probably more stable thanothers. The biological reasons underlying this apparentbetter stability or fitness are not yet understood.

Finally, the present phenotypic diversity observed in thissoil may reflect the agricultural history of the field. The even

distribution among samples of the different chromosomegroups and their associated plasmids may have occurredbecause of the regular annual plowing of the land. Limitedanalysis of four additional soil samples from this study siteconfirmed the distribution of both plasmids and chromosomepopulations reported in the other samples (40). The concom-

itant presence at the time of isolation of phenotypicallydistinct Agrobactenium populations suggests successful col-onization of this soil and stable cohabitation among membersof a relatively diverse community with apparently little or noplasmid exchange. Additional work is required to determinethe ecological significance of the phenotypic diversity seen

in this field. Identification of the factors influencing compo-

sition ofAgrobacterium communities and the distribution oftheir pTi in the soil will help advance our understanding ofthe epidemiology of crown gall disease.

ACKNOWLEDGMENTS

This research was supported by grant BlA to H.B. as part of thethird contract between the Algerian Ministere des Universites andthe European Community.We thank T. J. Burr, M. L. Canfield, R. N. Goodman, and L. W.

Moore for supplying reference strains of Agrobacterium spp., L.Jouanin and C. M. Brasilero for helpful discussions, D. Clerot andR. Esnault for advice on labelling and detection of DNA by usingdigoxygenin-modified nucleotides, E. Kondorosi for help perform-ing the Eckhardt procedure, N. Froger for technical assistance, andD. Meur for photographic artwork.

REFERENCES1. Akiyoshi, D. E., H. Klee, R. M. Amasino, E. W. Nester, and

M. P. Gordon. 1984. T-DNA of Agrobacterium tumefaciensencodes an enzyme of cytokinin biosynthesis. Proc. Natl. Acad.Sci. USA 81:5994-5998.

2. Bell, C. R., and W. D. Ramey. 1991. Hybrid biovars of Agro-bacterium species isolated from conifer roots. Can. J. Micro-biol. 37:34-41.

3. Bishop, A. L., B. H. Katz, and T. J. Burr. 1988. Infection ofgrapevines by soilborne Agrobacterium tumefaciens biovar 3and population dynamics in host and nonhost rhizospheres.Phytopathology 78:945-948.

4. Bomhoff, G. H., P. M. Klapwik, H. C. Kester, R. A. Schilpero-ort, J. P. Hernalsteens, and J. Schell. 1976. Octopine andnopaline synthesis and breakdown genetically controlled by a

plasmid ofAgrobactenum tumefaciens. Mol. Gen. Genet. 145:177-181.

5. Bouzar, H., N. Daouzli, Z. Krimi, A. Alim, and E. Khemici.

1991. Crown gall incidence in plant nurseries of Algeria, char-acteristics ofAgrobacterium tumefaciens strains, and biologicalcontrol of strains sensitive and resistant to agrocin 84. Agron-omie 11:901-908.

6. Bouzar, H., and J. B. Jones. 1992. Distinction of biovar 2 strainsof Agrobactenum from other chromosomal groups by differen-tial acid production. Lett. Appl. Microbiol. 15:83-85.

6a.Bouzar, H., and J. B. Jones. Unpublished data.7. Bouzar, H., and L. W. Moore. 1987. Isolation of different

Agrobacterium biovars from a natural oak savanna and tallgrassprairie. Appl. Environ. Microbiol. 53:717-721.

8. Bouzar, H., D. Ouadah, Y. Dessaux, A. Petit, Z. Krimi, and M.Trovato. 1991. Composition of the Agrobacterium populationfrom a fallow field. Phytopathology 81:1222.

9. Brisbane, P. G., and A. Kerr. 1983. Selective media for threebiovars of Agrobacterium. J. Appl. Bacteriol. 54:425-431.

10. Burr, T. J., B. H. Katz, and A. L. Bishop. 1987. Populations ofAgrobacterium in vineyard and nonvineyard soils and graperoots in vineyards and nurseries. Plant Dis. 71:617-620.

11. Bush, A. L., and S. G. Pueppke. 1991. A rapid and efficient newassay for determination of the three biotypes ofAgrobacteriumtumefaciens. Lett. Appl. Microbiol. 13:126-129.

11a.Canfield, M. L., and L. W. Moore. Personal communication.12. Chilton, M. D., M. H. Drummond, D. J. Merlo, D. Sciaky, A. L.

Montoya, M. P. Gordon, and E. W. Nester. 1977. Stable incor-poration of plasmid DNA into higher plant cells: the molecularbasis of crown gall tumorigenesis. Cell 11:263-271.

13. Clare, B. G., A. Kerr, and D. A. Jones. 1990. Characteristics ofthe nopaline catabolic plasmid in Agrobacterium strains K84and K1026 used for biological control of crown gall disease.Plasmid 23:126-137.

14. Cochran, L. C. 1941. A case of the crown gall organismpersistent in grain land. Plant Dis. Rep. 25:73.

15. De Cleene, M., and J. De Ley. 1976. The host range of crowngall. Bot. Rev. 42:389-466.

15a.Dessaux, Y. Unpublished data.16. Dessaux, Y., A. Petit, and J. Tempe. 1992. Opines in Agrobac-

terium biology, p. 109-136. In D. P. S. Verma (ed.), Molecularsignals in plant-microbe communications. CRC Press, Inc.,Boca Raton, Fla.

17. Dessaux, Y., J. Tempe, and S. K. Farrand. 1987. Geneticanalysis of mannityl opine catabolism in octopine-type Agro-bacterium tumefaciens strain 15955. Mol. Gen. Genet. 208:301-308.

18. Dhaese, P., H. De Greve, H. Decraemer, J. Schell, and M. VanMontagu. 1979. Rapid mapping of transposon insertion anddeletion mutations in the large Ti plasmids of Agrobacteriumtumefaciens. Nucleic Acids Res. 7:1827-1849.

19. Dickey, R. S. 1961. Relation of some edaphic factors to Agro-bacterium tumefaciens. Phytopathology 51:607-614.

20. Doetsch, R. N. 1981. Determinative methods of light micros-copy, p. 21-33. In P. Gerhardt (ed.), Manual of methods forgeneral bacteriology. American Society for Microbiology,Washington, D.C.

21. Dughri, M. H., and P. J. Bottomley. 1983. Effect of acidity onthe composition of an indigenous soil population of Rhizobiumtrifoldi found in nodules of Tnfolium subterraneum L. Appl.Environ. Microbiol. 46:1207-1213.

22. Eckhardt, T. 1978. A rapid method for the identification ofplasmid deoxyribonucleic acid in bacteria. Plasmid 1:584-588.

23. Ellis, J. G., and A. Kerr, M. Van Montagu, and J. Schell. 1979.Agrobactenium: genetic studies on agrocin 84 production andthe biological control of crown gall. Physiol. Plant Pathol.15:311-319.

23a.Farrand, S. K. Unpublished data.24. Keane, P. J., A. Kerr, and P. B. New. 1970. Crown gall of stone

fruit. II. Identification and nomenclature of Agrobacteniumisolates. Aust. J. Biol. Sci. 23:585-595.

25. Kennedy, B. W., and S. M. Alcorn. 1980. Estimates of U.S. croplosses to prokaryotic plant pathogens. Plant Dis. 64:674-676.

26. Kerr, A. 1969. Transfer of virulence between isolates of Agro-bacterium. Nature (London) 223:1175-1176.

27. Kerr, A. 1980. Biological control of crown gall through produc-

1316 BOUZAR ET AL.

on July 6, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from

CORRELATIVE ASSOCIATION BETWEEN PLASMIDS AND CHROMOSOME 1317

tion of agrocin 84. Plant Dis. 64:25-30.28. Kerr, A., P. Manigault, and J. Tempe. 1977. Transfer of

virulence in vitro and in vivo in Agrobactenium. Nature (Lon-don) 65:560-561.

29. Kersters, K., and J. De Ley. 1984. Agrobacterium, p. 244-254.In N. R. Kreig and J. G. Holt (ed.), Bergey's manual ofsystematic bacteriology, vol. 1. The Williams & Wilkins Co.,Baltimore.

30. King, E. O., M. K. Ward, and D. E. Raney. 1954. Two simplemedia for the demonstration of pyocyanin and fluorescein. J.Lab. Clin. Med. 44:301-307.

31. Laemmli, U. K. 1970. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature (London)227:680-685.

32. Lynch, J. M., and M. H. Ebben. 1986. The use of micro-organisms to control plant disease. J. Appl. Bacteriol. (Symp.Suppl.):115S-126S.

33. Michel, M. F., C. M. Brasileiro, C. Depierreux, L. Otten, F.Delmotte, and L. Jouanin. 1990. Identification of differentAgro-bacterium strains isolated from the same forest nursery. Appl.Environ. Microbiol. 56:3537-3545.

34. Moore, L. W. 1979. Practical use and success ofAgrobacteriumradiobacter strain 84 of crown gall control, p. 553-568. In B.Schippers and W. Gams (ed.), Soil-borne plant pathogens.Academic Press, Inc., New York.

35. Moore, L. W., and D. A. Cooksey. 1981. Biology ofAgrobacte-rium tumefaciens: plant interactions. Int. Rev. Cytol 13(Suppl.):15-46.

36. Moore, L. W., C. I. Kado, and H. Bouzar. 1988. Agrobactenium,p. 16-36. In N. W. Schaad (ed.), Laboratory guide for identifi-cation of plant pathogenic bacteria, 2nd ed. APS Press, St. Paul,Minn.

37. Nesme, X., T. Beneddra, and E. Collin. 1990. Importance ducrown gall chez les hydrides Populus tremula L. x P. alba L. enpepiniere forestiere. Agronomie 10:581-588.

38. Nesme, X., M.-F. Michel, and B. Digat. 1987. Population heter-ogeneity of Agrobacterium tumefaciens in galls of Populus L.from a single nursery. Appl. Environ. Microbiol. 53:655-659.

39. Ophel, K., and A. Kerr. 1987. Ecology of Agrobacterium:plasmid and biovars, p. 3-5. In D. P. S. Verma and N. Brisson(ed.), Molecular genetics of plant-microbe interactions. Marti-nus Nijhoff Publishers, Dordrecht, The Netherlands.

40. Ouadah, D. 1992. Caracterisation phenotypique d'une popula-tion tellurique d'Agrobacterium. Magister thesis. Universite desSciences et Technologies de Blida, Algeria.

41. Patel, M. K. 1929. Viability of certain plant pathogens in soils.Phytopathology 19:295-300.

42. Paulus, F., M. Ride, and L. Otten. 1989. Distribution of two

Agrobactenum tumefaciens insertion elements in natural iso-lates: evidence for stable association between Ti plasmids andtheir bacterial hosts. Mol. Gen. Genet. 219:145-152.

43. Petit, A., C. David, G. A. Dahl, J. G. Ellis, P. Guyon, F.Casse-Delbart, and J. Tempe. 1983. Further extension of theopine concept: plasmids in Agrobacterium rhizogenes cooper-ate for opine degradation. Mol. Gen. Genet. 190:204-214.

44. Petit, A., S. Delhaye, J. Tempi, and G. Morel. 1970. Recherchessur les guanidines. des tissus de crown gall. Mise en evidenced'une relation biochimique specifique entre les souches d'Agro-bacterium tumefaciens et les tumeurs qu'elles induisent. Phys-iol. Veg. 8:205-213.

45. Petit, A., and J. Tempe. 1978. Isolation of Agrobacterium Tiplasmid regulatory mutants. Mol. Gen. Genet. 167:147-155.

46. Petit, A., J. Tempi, A. Kerr, M. Holsters, M. Van Montagu, andJ. Schell. 1978. Substrate induction of conjugative activity ofAgrobacterium tumefaciens Ti plasmids. Nature (London) 271:570-572.

47. Rosenberg, C., and T. Huguet. 1984. The pAtC58 of Agrobac-terium tumefaciens is not essential for tumor induction. Mol.Gen. Genet. 196:533-536.

48. Roy, M., and M. Sasser. 1983. A medium selective forAgrobac-terium tumefaciens biotype 3. Phytopathology 73:810.

49. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecularcloning: a laboratory manual, 2nd ed. Cold Spring HarborLaboratory Press, Cold Spring Harbor, N.Y.

50. Schroder, G., S. Waffenschmidt, E. W. Weiler, and J. Schroder.1984. The T-region of Ti plasmids codes for an enzyme synthe-sizing indole-3-acetic acid. Eur. J. Biochem. 138:387-391.

51. Schroth, M. N., A. H. McCain, J. H. Foott, and 0. C. Huisman.1988. Reduction in yield and vigor of grapevine caused by crowngall disease. Plant Dis. 72:241-246.

52. Schroth, M. N., A. R. Weinhold, A. H. McCain, D. C. Hilde-brand, and N. Ross. 1971. Biology and control ofAgrobacteriumtumefaciens. Hilgardia 40:537-552.

53. Seigler, E. A. 1938. Relation between crown gall and pH of thesoil. Phytopathology 28:858-859.

54. Stachel, S. E., and E. W. Nester. 1986. The genetic and tran-scriptional organization of the vir region of the A6 Ti plasmid ofAgrobacterium. EMBO J. 5:1445-1454.

55. Stonier, L. 1960. Agrobacterium tumefaciens Conn. II. Produc-tion of an antibiotic substance. J. Bacteriol. 79:880-898.

56. Van Larebeke, N., C. Genetello, J. Schell, R. A. Schilperoort,A. K. Hermans, J. P. Hernalsteens, and M. Van Montagu. 1975.Acquisition of tumor-inducing ability by non-oncogenic agro-bacteria as a result of plasmid transfer. Nature (London) 255:742-743.

VOL. 59, 1993

on July 6, 2017 by guesthttp://aem

.asm.org/

Dow

nloaded from