Vol. 161, No. 3

Identification and Genetic Analysis of an Agrobacteriumtumefaciens Chromosomal Virulence Region

CARL J. DOUGLAS,' ROBERTO J. STANELONI,2 ROBERT A. RUBIN,2 AND EUGENE W. NESTER2*Department ofBotany1 and Department of Microbiology and Immunology,2 University of Washington, Seattle,

Washington 98195

Received 21 September 1984/Accepted 26 November 1984

A genetic analysis of Agrobacterium tumefaciens chromosomal functions required for virulence wasundertaken. Large Tn5-containing cosmid clones were isolated frokn DNA of avirulent A. tumefaciens mutantshaving chromosonal TnS insertions and exhibiting defective attachment to plant cells. The clones from severaldifferent mutants each contained overlapping segments of a 30-kilobase A. tumefaciens chromosomal region,which were physically mapped. All chromosomal TnS insertions leading to the avirulent, attachment-defectivephenotype were localized within an 11-kilobase portion of this chromosomal virulence region. TransposonTn3::HoHol (Tn3 containing lacZ) was used to simultaneously mutagenize and create lac fusions within thevirulence region. This analysis demonstrated the presence of two distinct chromosomal virulence loci, whichwere 1.5 and 5 kilobases long; transposon insertions into these loci led to avirulence and defective attachment.The 1-galactosidase activit associated with various Tn3::HoHol-created lac fusions indicated that the loci aretranscribed in opposite directions, aid complementation studies suggested that each locus consists of a singletranscriptional unit. A cosmid clone of the chroomosomal virulence region containing a iac fusion in the extreme3' portion of the 5-kilobase locus was used to demonstrate that expression of this region is dependent on thepresence of sequences in the 5' portion of the locus, confirming its operon-like nature.

Agrobacterium tumefaciens is a plant pathogen whichcauses crown gall tumors when it is present on woundedtissue of dicotyledonous plants. Virulent bacteria containlarge tumor-inducing (Ti) plasmids which are necessary fortumorigenicity (44, 47). A portion 'of Ti plasmid DNA istransferred to plant cells during tumor formation and isstably maintained in tumor cells (9, 16, 43, 50, 51). Tiplasmid DNA genes are transcribed in crown gall cells andencode enzymes for the synthesis of plant growth regulators(19, 40; D. E. Akiyoshi, H. J. Klee, R. M. Amasino, E. W.Nester, and M. P. Gordon, Proc. Natl. Acad. Sci. U.S.A., inpress).

Although the mechanism by which A. tumefaciens trans-fers plasmid DNA into plant cells is unknown, the bacterialgenetics of tumorigenesis have been extensively studied. Inaddition to the Ti plasmid DNA, a 35-kilobase (kb) portion ofthe Ti plasmid containing the vir genes is required forvirulence (21). These genes are thought to encode functionsthat are necessary for the transfer of Ti plasmid DNA intoplant cells, but are not themselves found in tumor cells.Mutational analysis has shown that A. tumefaciens chromo-somal genes are also required for virulence, since TnSinsertions in chromosomal DNA can lead to avirulence (14).An early step in crown gall tumor formation is the attach-

ment of bacteria to plant cells (24). Numerous studies haveshown that A. tumefaciens attaches to plant tissue culturecells (11, 30, 32, 34) and freshly isolated plant cell suspen-sions (11). This attachment is thought to be a prerequisite forTi plasmid DNA transfer to plant cells. Preinoculation ofplant tissue with bacteria which are avirulent but able toattach inhibits tumor formation by subsequently added vir-ulent bacteria, presumably by blockage of attachment siteson the plant cells by the avirulent bacteria (17, 24).There is controversy concerning the nature of the bacte-

rial cell surface molecules which participate in the attach-

* Corresponding author.

ment process. A. tumefaciens lipopolysaccharide has beenimplicated in this process by its ability to inhibit tumorformation on pinto bean leaves (48) and its ability to blockattachment to tobacco tissue culture cells (32). However, A.tumefaciens lipopolysaccharide has been found not to inhibittumorigenesis on potato disks (36). Cellulose fibrils pro-duced by A. tumefaciens have been shown to anchor bacte-ria to plant cells and to each other during attachment (31),but do not seem to be involved in the initial attachment ofA.tumefaciens to plant cells, since mutants blocked in cellu-lose fibril synthesis retain virulence and attachment ability(29).Most evidence suggests that chromosomally encoded func-

tions are primarily responsible for the attachment ability ofA. tumefaciens. Although in some cases Ti plasmids havebeen found to affect attachment (32), aviruient strains lack-ing Ti plasmids have been demonstrated to attach to plantcell walls (24), tissue culture cells (11; N. Neffand A. Binns,submitted for publication), leaf mesophyll cells (11, 12), andregenerating protoplasts (28) equally as well as do isogenicstrains harboring Ti plasmids. In previous studies, we dem-onstrated that six avirulent A. tumefaciens mutants contain-ing TnS insertions in chromosomal DNA are defective intheir ability to attach to plant cells (11), providing directevidence that chromosomally encoded functions are re-quired both for attachment ability and for virulence. Thesechromosomal genes are distinct from the genes necessary forthe synthesis of cellulose fibrils, since the avirulent mutantsstill produce such fibrils (11).To better understand the chromosomally encoded func-

tions which are required for attachment and tumor forma-tion, we undertook a genetic analysis of these chromosomalmutants. In this paper we describe the isolation of additionalchromosomal mutants and show that all TnS insertionsleading to the avirulent, attachment-defective phenotype areclustered within an 11-kb chromosomal virulence region.Additionally, we describe the isolation of a large number of

850

JOURNAL OF BACTERIOLOGY, Mar. 1985, p. 850-8600021-9193/85/030850-11$02.00/0Copyright © 1985, American Society for Microbiology

on August 18, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

A. TUMEFACIENS CHROMOSOMAL VIRULENCE REGION 851

TABLE 1. Bacteria and plasmidsStrain or Genotype or phenotype' Reference or sourceplasmid

BacteriaHB101 E. coli thr leu thi recA

hsdR hsdM pro StrrMM294 E. coli hsdR thi endAC2110 E. coli Nalr polA I rha

hisJM83 E. coli ara A(lac-pro)

rpsL thi 4)80 d A lacZM15

A723 A. tumefaciens C58chromosome,pTiB6806

A348 A. tumefaciens C58chromosome,pTiA6NC

A1011 A. tumefaciens C58chromosome: :Trn5vsir

A1020 A. tumefaciens C58chromosome: :TnSvir

A1038 A. tumefaciens C58chromosome: :TnSvir

A1045 A. tumefaciens C58chromosome: :TnSvir

A2501 A. tumefaciens C58chromosome: :TnSvir

A2505 A. tumefaciens C58chromosome: :Th5vir

A2506 A. tumefaciens C58chromosome: :TnSvir-

A2507 A. tumefaciens C58chromosome: :Tn5vir-

PlasmidspPHlJIpJB4JI

pLAFRIpVK102pRK2073ColE1::TnSpUC13-CampTJS140

pGA254F'::Tn3pHoHol

pSShepCD521

IncP Gmr StrrIncP Gmr Strr Mu

c: :Tn5IncP TcrIncP Tcr KmrIncP tra+KmrCamrTcr Cbr RK2 and ColEl

repliconsTcr Cbr ColEl repliconCbrTn3::lacZ Cbr tnpA

tnp+Camr tnpR tnpTcr

pCD522 Kmr

pCD523 Tcr

pCD526 Kmr

pRAR201 Cbr

5

M. Messelson23

46

15

15

TABLE 1-ContinuedStrain or Genotype or phenotype" Reference or sourceplasmid

A. tumefaciens chro-mosome (Bam2.3)

pRAR205 Cbr This study; pTJS140-A. tumefaciens chro-mosome (Bam6.7 +2.3)

pRAR213 Cbr This study; pTJS140-A.tumefaciens chromo-some (Bam6.7)

a Camr, Chloramphenicol resistance; Cb', carbenicillin resistance; Gmr,gentamicin resistance; Kmr, kanamycin resistance; Nalr, nalidixic acid resist-ance; Str', streptomycin resistance; Tcr, tetracycline resistance; vir-, aviru-lent.

bStachel and Nester, manuscript in preparation.

14

14

14

14

11

This study

This study

This study

22

132223

D. Berg46

T. Schmidhauser

G. An49

Stachel and Nesterb

Stachel and NesterbThis study; pLAFRI-A. tumefaciens chro-mosome

This study; pVK102-A. tumefaciens chro-mosome

This study; pLAFR1-A. tumefaciens chro-mosome

This study; pLAFR1-A. tumefaciens chro-mosome (Sal3.1)

This study; pTJS140-

transposon insertions in this region, which define two dis-tinct chromosomal loci required for virulence.

MATERIALS AND METHODS

Bacterial strains, plasmids, and media. The strains andplasmids of A. tumefaciens and Escherichia coli used arelisted in Table 1. E. coli strains were maintained on L-agar(33) and grown in liquid L-broth. Agrobacterium strains weremaintained on AB agar (8) and grown in MG/L broth (14) orliquid AB medium. The antibiotics used in media for E. coliwere tetracycline (10 ,ug/ml), kanamycin (50 ,ug/ml), carben-icillin (50 jigIml), chloramphenicol (50 ,ug/ml), and nalidixicacid (50 j.Lg/ml); the antibiotics used in media for Agrobac-terium were kanamycin (100 j.g/ml), carbenicillin (100 jig/ml),gentamicin (100 ,ug/ml), and nalidixic acid (50 ,ugIml).DNA isolation. Total DNA was prepared by the method of

Marmur (27) or by a modification of this method, using 2-mlcultures. The method of Birnboim and Doly (3) was used toprepare plasmid DNA. In somne cases, plasmid DNA wasfurther purified by centrifugation through cesium chloride-ethidium bromide density gradients.

Restriction endonucleases. Restriction endonucleases werepurchased from Bethesda Research Laboratories, Rockville,Md., and New England Biolabs, Beverly, Mass., and usedaccording to the specifications of the suppliers.

Gel electrophoresis. Horizontal gels containing varyingagarose concentrations (0.7 to 1.2%) were used to analyzerestriction digests, and vertical slab gels containing 0.7%agarose were used to analyze Ti plasmids. The staining andphotographic conditions used have been described previ-ously (14).

Molecular cloning and plasmid construction. For cosmidcloning, large fragments of Agrobacterium chromosomalDNA were prepared by partial digestion with EcoRI orHindIll and isolated by agarose gel electrophoresis in pre-parative gels containing 0.2 ,ug of ethidiun bromide per ml.DNA approximately 20 kb long was electroeluted, ethanolprecipitated, and ligated into the EcoRI site of pLAFRI (13)or the HindlIl site of pVK102 (22). Ligated niolecules werepackaged in vitro into phage X particles (4) and were used totransduce E. coli HB101. When DNA with Tn5 insertionswas cloned, transductants were selected on media contain-ing tetracycline (pLAFR1) and kanamycin (Tn5). To con-struct a clone bank of total Agrobacterium DNA, HindIll-cleaved pVK102 was treated with alkaline phosphatasebefore ligation with partially Hindlll-digested, size-fraction-

VOL. 161, 1985

on August 18, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

852 DOUGLAS ET AL.

U')0 oco) U)6o< <f ll

3-7 12-05 1is5 1 3.8 | 32 2.1 h25FB 1.65 | 4.0 |1.3 |1.1 N 3.0 1 I2.8 | 2.4 4.0 134 | 5.0 Ii

81.81 9-2

POCD110

pCD120UpCD135 v

1kb

pCD144pCD148

pCD152-

pUvLL1.

FIG. 1. Restriction endonuclease map of a portion of the A. tumefaciens C58 chromosome, showing the location of Tn5 insertions whichaffect virulence ahd attachment ability. The vertical lines above the map show the location of the TnS insertion in each strain, and thearrowhead indicates the approximate location of the 2.2-kb insertion in strain A2506. The sizes of restriction fragments are indicated (inkilobases). the structures of cosmids containing cloned chromosomal DNA from various avirulent, attachment-defective mutants are shownbelow the restriction map; the arrowheads indicate the locations of TnS in these cosmids. pCD521 is a clone of wild-type DNA isolated froma clone bank of chromosomal DNA.

ated A. tumefaciens DNA, and bacteria harboring recombi-nant plasmids were selected on media containing tetracy-cline. Cosmid pCD521 was identified in this bank by colonyhybridization (18).

Plasmid pCD522 was constructed by partial SalI digestionof Tn5-containing cosmid pCD148 (Fig.1), ligation into theSall site of pVK102, and screening Kmr Tcs transformantsfor the presence of the 3.1- and 6.4-kb Sall fragments of thatplasmid. Recombination between pCD148 and a plasmidcontaining homologous A. tumefaciens DNA without TnSwas used to remove the TnS insertion from pCD148 in orderto construct pCD523. The 13.4-kb HindIII fragment ofpCD521 was recloned into pGA254, a plasmid containing aColEl replication origin and carbenicillin and tetracyclineresistance (G. An, unpublished data). This resulted in theloss of the tetracycline resistance gene from pGA254 and itsreplacement with the 13.4-kb HindlIl fragment to yieldpGA254-Hind 13.4. This plasmid, which was compatiblewith pCD148, was cotransformed with pCD148 into E. coliMM294, and Cbr Tcr Kmr transformants harboring bothplasmids were grown for several generations in liquid broth.Cointegrate plasmids which formed due to a single recombi-nation event between the homologous sequences of A.tumefaciens DNA in the two plasmids were detected bymobilizing plasmid DNA from E. coli MM294 into E. coliC2110, a Nalr polA strain in which ColEl replicons cannotreplicate (20). Nalr transconjugants were selected for theantibiotic resistance properties of both plasmids and wereobtained at a frequency of approximately 10-5. Bacteria inwhich a second recombination event had taken place, result-ing in the breakdown of the pCD148::pGA254-Hind 13.4cointegrates and the loss of TnS, were identified by screen-ing colonies for the loss of kanamycin resistance and wererecovered at a frequency of 102. This method yieldedpCD523, which was identical to pCD148 except for the lossof TnS. Plasmids pRAR201, pRAR205, and pRAR213 wereconstructed by BamHI digestion of pCD523 and ligation intothe BamHI site of cloning vector pTJS140, inactivating the

a-complementation peptide of ,-galactosidase (46) encodedby the vector. Digestion of pCD523 with Sall, ligation intoSall-cleaved pVK102, and screening Kmr Tcs transformantsfor the 3.1-kb Sall fragment yielded pCD526.Mapping of genomic TnS insertions. The locations of TnS

elements in mutants for which TnS-containing cosmid cloneswere available were deduced by digestion of purified plasmidDNA with a combination of restriction enzymes, includingHindIII and BglII (which cut once in each of the terminalrepeats of TnS) and EcoRI and KpnI (which do not cutwithin TnS), followed by agarose gel electrophoresis. Inser-tions in mutants for which no cosmid clones were availablewere mapped by Southern hybridization analysis. TotalDNA from the mutants was cut with EcoRI, electrophor-esed, blotted, and probed with (i) 32P-labeled ColEl::TnSDNA or (ii) a clone covering most of the 30-kb chromosomalvirulence region. The TnS element was localized to a partic-ular EcoRI fragment by (i) the size of the novel EcoRIfragment containing TnS or (ii) the particular wild-typefragment present as a larger TnS-containing fragment. Tofurther map the location of the TnS, total DNA was digestedwith EcoRI and Hindlll and probed with a subclone of thewild-type EcoRI fragment containing the insertion, and thehybridization pattern was compared with that of similarlytreated DNA from the parental strain.Transposon mutagenesis. Suicide plasmid pJB4JI was used

to introduce TnS insertions into the genome of A. tumefaci-ens as described by Garfinkel and Nester (14). Kanamycin-resistant colonies were screened for virulence on Kalanchoeleaves, and the cellular location (plasmid or chromosome) ofthe TnS insertion in avirulent isolates was determined bySouthern hybridization of plasmid screens as describedpreviously (14). Mutants containing Tn5 in chromosomalDNA were also tested for virulence on Kalanchoe stems,tobacco (Nicotiana tabacum var. Xanthi), sunflower, andtomato.Cosmid clones ofAgrobacteriurn chromosomal DNA were

used to introduce site-specific transposon insertions into the

Eco RIHind IIIKpn I

J. BACTERIOL.

on August 18, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

A. TUMEFACIENS CHROMOSOMAL VIRULENCE REGION 853

Agrobacterium chromosome. Insertions into cosmid cloneswere generated by using Tn3 and the method of White et al.(49) or by using Tn3: :HoHol, a derivative of Tn3 containinga promoterless lacZ gene and lacking tnpA (S. Stachel andE. Nester, manuscript in preparation). Tn3::HoHol con-tained on a plasmid with a ColEl replicon (pHoHol) trans-poses when tnpA is supplied by a helper plasmid (pSShe).Cosmid clones containing Tn3::HoHol insertions were iso-lated by using a procedure described elsewhere (Stachel andNester, manuscript in preparation). Plasmid DNAs from E.coli strains containing cosmid clones with transposon inser-tions were prepared by the rapid screening method ofBirnboim and Doly (3). When insertions were present incloned DNA, the locations of the insertions relative toknown restriction sites were determined by using new re-striction sites introduced by Tn3 or Tn3::HoHol. Since bothBamHI and EcoRI sites are located asymetrically inTn3::HoHol, digestion with these enzymes also alloweddetermination of the orientation of this element with respectto the lacZ gene.

Introduction of mutagenized DNA into genomic chromo-somal DNA. Cosmid clones containing transposon insertionsof interest were introduced into the Agrobacterium chromo-some by the marker exchange technique (39). Merodiploidstrains containing individual cosmids with transposon inser-tions were constructed by using strain A348 (Table 1). Afterconjugation of pPHlJI into these strains, transconjugantswere plated onto AB media containing gentamicin and eitherkanamycin (TnS) or carbenicillin (Tn3 and Tn3::HoHol), asdescribed previously (15). The colonies obtained were puri-fied and tested for virulence and attachment ability.

Southern blot analysis. Marker exchanges were verified bySouthern blot analysis of putative recombinants. Total DNAwas digested with restriction endonucleases, subjected toagarase gel electrophoresis, and transferred to nitrocellulosefilters (Schleicher & Schuell, Inc., Keene, N.H.) by themethod of Southern (42). DNA on the filter was hybridizedwith nick-translated (26) DNA homologous to the region ofthe transposon insertions. Marker exchange was assumed tohave worked properly when the appropriate wild-type re-striction fragment was absent in the mutants and was re-placed by new fragments, the sizes of which depended onthe location of the transposon insertion. In all cases, thesizes of the new fragments corresponded to the sizes pre-dicted from restriction mapping of the transposon insertions.

Construction of merodiploid strains. To construct merod-iploid strains for complementation analysis and markerexchange, broad-host-range plasmids containing Agrobac-terium chromosomal DNA were mobilized into Agrobacte-rium by using the triparental system described by Ditta et al.(10). Transconjugants were selected on AB media containingantibiotics appropriate for the selection of the plasmid inquestion.

,B-Galactosidase assays. 1-Galactosidase activity was as-sayed in Agrobacterium strains containing plasmids withlacZ fusions, as described by Miller (33). Bacteria weregrown in Mg/L liquid media. The bacteria were pelletted,suspended in 500 pll of Z-buffer, mixed with a Vortex mixerwith 20 ,ul of 0.05% sodium dodecyl sulfate and 20 ,u1 ofCHC13 for 10 s, and incubated for 10 min at 28°C. Then 100iil of o-nitrophenyl-,-D-galactopyranoside (4 mg/ml) wasadded, and the 3-galactosidase activity was determined.

Bacterial attachment to plant cells. The ability of Agrobac-terium strains to attach to plant cells was tested by usingfreshly isolated Zinnia leaf mesophyll cells or tobacco sus-pension culture cells, as described previously (11).

Virulence assays. Virulence of Agrobacterium strains wasassayed on Kalanchoe daigremontiana leaves as describedpreviously (14).

RESULTS

Isolation of chromosomal TnS insertion mutants. A numberof mutants affected in virulence with TnS insertions inchromosomal DNA have been isolated (14; D. Puetz andD. J. Garfinkel, unpublished data). These were obtained byusing Tn5-containing plasmid pJB4JI, which cannot repli-cate in Agrobacterium (45), to deliver TnS. Approximatelyone-third of these chromosomal mutants were avirulent onall plants tested, whereas the rest had altered host ranges.The avirulent chromosomal mutants which have been shownto be defective in attachment to plant cells (11) are listed inTable 1.

In order to obtain as many independent avirulent chromo-somal mutants as possible, the above-described method ofmutagenizing the entire A. tumefaciens genome was re-peated. A total of 8,000 kanamycin-resistant colonies fromfilter matings between A. tumefaciens and E. coli1830(pJB4JI) were screened for virulence on Kalanchoeleaves. From these, 22 avirulent mutants from separate filtermatings were obtained. These were tested for the cellularlocation of TnS by hybridization of 32P-labeled ColEl::TnSDNA to blots of total plasmid DNA fractionated on agarosegels, as described by Garfinkel and Nester (14). Strainswhich showed hybridization to linear DNA but not to Tiplasmid DNA contained TnS in the chromosomal DNA. Ofthe 22 TnS insertion mutants avirulent on Kalanchoe leaves,8 contained chromosomal TnS insertions. Three of thesechromosomal mutants (mutants A2505, A2506, and A2507)(Table 1) were avirulent on all plants tested, and five werevirulent on one or more plants other than Kalanchoe. StrainsA2505, A2506, and A2507 were tested for their ability toattach to plant cells and were found to have phenotypessimilar to those previously described for avirulent chromo-somal mutants defective in attachment (11).

Identification of a chromosomal virulence region. DNAfrom several Tn5-induced avirulent chromosomal mutantswere used to construct a series of cosmid clones. Thepresence of Tn5 marked the DNA into which it was insertedby virtue of the kanamycin resistance encoded by theelement, and cosmid cloning allowed the isolation of largeregions of DNA flanking the TnS insertions. By comparingcosmid clones from various mutants we hoped to detect anyclustering of insertions from different mutants on the chro-mosome.

Large fragments of DNA from various mutants, whichwere generated by partial EcoRI digestion of total DNA andsize fractionation on preparative agarose gels, were clonedinto the EcoRI site of cosmid vector pLAFR1, and bacteriaharboring recombinant plasmids containing fragments withTnS insetions were identified by selection on media contain-ing kanamycin. A comparison of the restriction patterns ofclones from different mutants revealed that the clones con-tained a number of EcoRI fragments in common. Furtheranalysis showed that the cosmid clones isolated representeda series of overlapping DNA fragments spanning a region ofthe chromosome about 30 kb long, and a physical map of theregion was constructed by using several restriction enzymes(Fig. 1). The identification of this region suggested that theavirulent, attachment-defective phenotype of the chromo-somal mutants was due to the inactivation of genes clusteredon the Agrobacterium chromosome.

VOL. 161, 1985

on August 18, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

854 DOUGLAS ET AL.

The location of the Tn5 insertion in each avirulent chro-mosomal mutant was determined by restriction enzymeanalysis, using the restriction sites introduced by TnS. Inmutants for which cosmid clones of DNA containing TnSwere available, purified plasmid DNA was analyzed; inser-tions in mutants for which no cosmid clones were availablewere mapped by Southern hybridization analysis. The loca-tions of the elements were deduced from the sizes of thenovel restriction fragments due to TnS insertion and areindicated by strain designation on Fig. 1. With one excep-tion, all TnS insertions were clustered within an 11-kbportion of the chromosomal virulence region.

Strain A2506 contained a single chromosomal TnS inser-tion in a 15-kb EcoRI fragment not located in the previouslydefined 30-kb chromosomal region. Further analysis showedthat an additional 2.2-kb insertion was present in the 3.2-kbEcoRI fragment of the virulence region in this strain. Theapproximate location of this insertion (Fig. 1) was deter-mined by Southern blot analysis of strain A2506 DNAdigested with EcoRI and HindlIl (which did not cut withinthe insertion), using the subcloned 3.2-kb EcoRI fragment asa probe (data not shown). The nature of this 2.2-kb insertionis unknown, but it is likely to be an endogenous Agrobac-terium IS element (14).

Genetic linkage between TnS insertions and the mutantphenotype. Genetic linkage between the Tn5 insertion instrain A1045 and the attachment-defective phenotype waspreviously demonstrated (11) by using the marker exchangetechnique (39) to correlate acquisition of the TnS insertionwith acquisition of the mutant phenotype in a wild-typestrain. In order to be certain that the TnS insertions in theadditional mutants were linked to the observed phenotypes,similar experiments were carried out by using cosmid DNAcontaining TnS insertions isolated from strains A1011, A1020,A1038, and A2501 (Fig. 1). In each case, acquisition ofmutant, TnS-containing DNA by marker exchange into thewild-type strain resulted in acquisition of the mutant pheno-type, demonstrating that the chromosomal region defined byoverlapping cosmid clones contains genes which are inacti-vated by Tn5 insertions in the mutants, are necessary forvirulence, and affect the ability of A. tumefaciens to attachto plant cells.

Isolation of wild-type clones of the virulence region. Ourstrategy for further genetic analysis of the chromosomalvirulence region was to use site-directed transposon mutage-nesis of cloned chromosomal DNA to delineate the geneticloci present. In order to isolate wild-type clones spanningthe chromosomal virulence region for these studies, a clonebank of total A. tumefaciens strain A348 DNA was con-structed by using cosmid vector pVK102 (22). The clonebank consisted of 2,000 E. coli HB101 colonies containingpVK102 molecules with insertions of HindlII-digested A.tumefaciens DNA. Cosmid clone pCD521 was obtained fromthis clone bank by colony hybridization (18). This clonecontains a 13.4-kb HindlIl fragment which partially overlapsthe chromosomal virulence region (Fig. 1 and 2). PlasmidspCD522 and pCD526, which also span portions of thevirulence region (Fig. 2), were constructed as describedabove.To obtain a clone which spans the entire chromosomal

virulence region, we utilized a strategy relying on recombi-nation (see above) to remove the TnS insertion from pCD148,a cosmid clone of the virulence region isolated from strainA1038 (Fig. 1). The wild-type clone obtained in this mannerwas designated pCD523 and spans the entire chromosomalvirulence region, as defined by TnS insertions (Fig. 2). It is

identical to pCD148 except for the loss of TnS. This methodof obtaining a cosmid clone of the desired size proved to bemuch less laborious than further screening of a clone bankfor such a cosmid. In subsequent studies (see below), thesame strategy of selecting for cointegrate formation andscreening for cointegrate breakdown was successfully usedto create a small deletion in a large plasmid. This methodseems to be generally applicable to creating specific changesin large plasmids, providing that adequate DNA homologyfor recombination is present.

Organization of the chromosomal virulence loci. To distin-guish distinct loci within the Agrobacterium chromosomalvirulence region which are necessary for virulence andefficient attachment to plant cells, we used site-directedmutagenesis of cloned Agrobacterium chromosomal DNA.Over 70 transposon insertions into plasmids pCD521,pCD522, pCD523, and pCD526 were physically mapped byrestriction analysis and introduced into the chromosome byhomologous recombination.The results of the transposon mutagenesis are summarized

in Fig. 2. Most insertions were obtained by using Tn3: :HoHol(a Tn3 derivative containing lacZ [Stachel and Nester,manuscript in preparation]); insertions 1 and 2 were obtainedby using Tn3. Our analysis showed that there are twoclusters of insertions which lead to avirulence; these areseparated by large regions of DNA into which transposoninsertions have no effect on virulence. The left cluster isabout 1.5 kb long and is defined by seven Tn3::HoHolinsertions. The Tn5 insertion of strain A2505 is also presentin this region. Since the 1.5-kb segment of DNA is notinterrupted by insertions giving a wild-type response, it isprobable that a single locus is present. The right cluster ofinsertions extends over 5 kb of DNA and is defined by 17Tn3::HoHol insertions. In addition, the TnS insertions ofseven of the original chromosomal mutants are present here.This cluster contains two insertions (insertions 2 and 116)which give a wild-type virulence response. On the basis ofthe complementation and deletion studies discussed below,we believe that a single transcriptional unit is present withinthis right cluster and that insertions 2 and 116 do not defineseparate genetic loci within this cluster.Mutants with insertions in both the right and left virulence

loci were tested for their ability to attach to plant cells andwere found to have attachment-defective properties similarto those previously described for avirulent chromosomalTnS mutants (11). We named these loci chvA and chvB (Fig.2) (for chromosomal virulence). Insertions outside chvA andchvB had no detectable effect on virulence, and mutants withsuch insertions were found to be indistinguishable from thewild-type strain in their attachment ability. Thus, the onlyinsertions in this chromosomal region which affect the abilityto attach to plant cells are those which lead to avirulence.

Direction of transcription. The direction of transcription ofchvA and chvB was determined by using the 3-galactosidaseactivity encoded by the lacZ gene within transposonTn3::HoHol. Significant expression of lacZ from within thetransposon is dependent on transcriptional or translationalfusions with the gene into which the element is inserted, sothat expression is under control of the promoters of theAgrobacterium genes. This expression only occurs if thereading frame of lacZ is in the same orientation as thedirection of transcription of the Agrobacterium gene, inwhich case lacZ is expressed at a relatively high level (inproportion to the activity of the Agrobacterium gene). In theopposite orientation, insertions have little or no lacZ expres-sion. The orientation of each Tn3::HoHol insertion with

J. BACTERIOL.

on August 18, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

A. TUMEFACIENS CHROMOSOMAL VIRULENCE REGION

chvA1 kb,

chvB

+++-=-- ++ ++ +++ =-- --+=+ +++ H+*++++++62 107 105 113111 3 1 11116 863 1023 620 3126

|205 1.55 | 3.8 3.2 |2.1 1.25 .85| 1.65 | 4.0 1.3 1.0 Eco RI

6.4 3.1 0.9 13 6.8 Sal

2.8 2.4 4.0 13.4 Hind III

2.3 6.7 Bam HIl85 185 9.2 Kpn

-pCD521

pCD526FIG. 2. Transposon mutagenesis of the chromosomal virulence region. The central portion of the figure shows a restriction map of the

chromosomal virulence region; the sizes of restriction fragments are indicated (in kilobases). Clones of the chromosomal virulence region(pCD521, pCD522, pCD523, pCD526) were subjected to transposon mutagenesis by using Tn3 or Tn3::HoHol. The location of each insertionis indicated by a vertical line above the restriction map. Insertions 1 and 2 are Tn3; the remainder are Tn3::HoHol. The arrowheads indicatethe locations of the genomic Tn5 insertions shown in Fig. 1. Tn3::HoHol insertions above the horizontal line are the result of transpositionsin which the lacZ gene is oriented from left to right; those below the line are the result of transpositions in which lacZ is oriented from rightto left. The virulence properties of A. tumefaciens strains obtained after marker exchange of plasmids containing each insertion into awild-type strain are indicated by plus and minus signs. The transposon insertions define two virulence loci, chvA and chvB, insertions intowhich lead to avirulence and defective attachment to plant cells. See the text for an explanation of anomolous insertions 2 and 116.

respect to the lacZ gene was determined by using asymmet-rical BamHI and EcoRI restriction sites within the element.The P-galactosidase activities of merodiploid strains contain-ing plasmid pCD523 or pCD522 with Tn3::HoHol-generatedlac fusions in both orientations within and outside thevirulence loci were assayed. Figure 3 shows that chvA wastranscribed from left to right since only those chvA inser-

chvA

tions with lacZ in that orientation gave high ,-galactosidaseactivity. The chvB locus appeared to be transcribed fromright to left, since only those chvB insertions in which thelacZ gene is oriented from right to left gave significant levelsof expression. The low levels of expression associated withinsertions in which the lacZ gene is oriented opposite theapparent direction of transcription were only slightly higher

chv B

3 12 863 7313 713-galactosidase activity

7 45i 106 1 ,0l26 611232 66*616967 43

E ES E E S E S SE Sy E S E E ESIII I I

FIG. 3. Direction of transcription of chvA and chvB. The direction of transcription of chromosomal virulence loci chvA and chvB wasdetermined by assaying the P-galactasidase activity of plasmids with lac fusions created by Tn3::HoHol insertions into chvA and chvB. Theinsertions in which the lacZ gene is oriented from left to right are shown above the line; those with the lacZ gene oriented from right to leftare shown below the line. P-Galactosidase activities in merodiploid A. tumefaciens strains containing the plasmids were assayed by themethod of Miller (33). Symbols: 0, high ,B-galactosidase activity (>28 U); 0, low P-galactosidase activity (<6 U); and O), intermediate activity(12 U). Transcription of chvA occurs from left to right since only lac fusions in that orientation had high activity. Similarly, chvB is transcribedfrom right to left. The expression of ,-galactosidase from insertions outside the virulence loci differed sharply from the pattern of expressionwithin the loci. A restriction map of the chromosomal virulence region is shown at the bottom. E, EcoRI; S, SalI.

6242 1054114

855VOL. 161, 1985

on August 18, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

856 DOUGLAS ET AL.

chvA9In0

E H ES E

chvB

H E B: isse Ev

*T,-W,-W, pRAR205

pRAR201

$ +

72

20+ + + + + pLS20

39+ 3?6 pLS39

60+ - pLS60

84

12 pLS 64+ -- - pLS 2

32+ 4 -- - pLS32

45+ + + + pLS45

FIG. 4. Complementation analysis of the chromosomal virulence loci. Plasmids containing portions of the chromosomal virulence regionwith or without transposon insertions were tested for their ability to complement in trans avirulent mutants containing chromosomaltransposon insertions. The locations of the genomic insertions in chvA and chvB mutants used in complementation experiments are shownabove the restriction map. Also shown are the locations of virulent insertions 2 and 116, which occur within chvB. The structures of theplasmids introduced into the mutants are also shown. In some of these plasmids Tn3: :HoHol insertions were present; the locations of theseare indicated by arrows and the number designations of the insertions. Merodiploid mutant strains were tested for virulence on Kalanchoeleaves. A plus sign above a plasmid map indicates complementation (virulence) of the mutant containing a genomic transposon insertion atthat location; a minus sign indicates a lack of complementation (avirulence). Our analysis showed that mutants with insertions within chvBcan be complemented only by plasmids containing chvB in its entirety. B, BamHI; E, EcoRI, H, HindIll; S, SalI.

(1 to 2 U) than the endogenous 0-galactosidase activity inwild-type Agrobacterium (data not shown). The pattern ofP-galactosidase expression from plasmids with Tn3::HoHolinsertions outside the defined virulence loci differed sharplyfrom that of insertions within the loci (Fig. 3). For example,insertions 62 and 4, which were located just outside chvA,and insertion 67, which was located to the right of chvB, hadlow levels of activity, in contrast to the high activity ofadjacent insertions with the same orientations but within thevirulence loci (Fig. 3). The results support the interpretationthat the lacZ gene fusions created by Tn3: :HoHol insertionsinto chromosomal DNA are under the control of specificAgrobacterium promoters and that the P-galactosidase ac-tivities associated with them are a measure of the expressionof chromosomal genes.Complementation analysis. From transposon mutagenesis

studies, it remained unclear whether the right 5-kb portion ofthe chromosomal virulence region (chvB) consisted of a

single locus or three or more closely linked loci. Althoughtranscription of the entire 5 kb ocurred from right to left (Fig.3), two wild-type insertions (insertions 116 and 2) (Fig. 2)were isolated within the 5-kb region. Even with the use of asmall clone (pCD526), DNA sequences surrounding inser-tion 116 were refractory to Tn3::HoHol insertion, making itdifficult to characterize this region of DNA by transposonmutagenesis. To determine whether the chvB cluster ofgenes is organized into a single operon or whether insertions

116 and 2 delineate separate transcriptional units withinchvB, we performed genetic complementation studies byusing wild-type clones of the chromosomal virulence regionand TnS or Tn3: :HoHol insertion mutants. We chose cloneswhich contained sequences spanning the entire chvB clusteror which lacked the right (5') or left (3') portion of chvB.Complementation experiments were performed by mobiliz-ing the broad-host-range plasmids containing cloned DNAinto strains with genomic transposon insertions in variouslocations. The resulting merodiploid strains were tested forvirulence on Kalanchoe leaves.The results of the complementation analysis are summa-

rized in Fig. 4. Plasmid pRAR205 contains the 2.3- and6.7-kb BamHI fragments from pCD523 which together spanthe entire chvB region, whereas pRAR201 and pRAR213contain the single 2.3- and 6.7-kb fragments covering onlythe 3' and 5' ends of chvB respectively (Fig. 4). Whenintroduced into mutants with TnS insertions in chvB,pRAR205 restored virulence in all cases, but pRAR201 wasunable to complement the genomic TnS insertion in strainA1011 present in the left portion of chvB (Fig. 4). PlasmidpCD526, which contains a single 3.1-kb SalI fragment cov-ering a slightly larger portion of this region, was also not ableto complement mutants containing genomic transposon in-sertions in the 3' portion of chvB (Fig. 4). Finally, plasmidpCD522 was tested for complementation ability. This plasmidcontains the 3.1-kb Sall fragment covering part of chvB, as

B E ES EHE

J. BACTERIOL.

on August 18, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

A. TUMEFACIENS CHROMOSOMAL VIRULENCE REGION 857

TABLE 2. Restoration of virulence and attachment ability bypRAR205

Strain Virulence % of bacteria attachedto Zinnia cellsa

A348 + 24 ± 1A1011 - 5 ± 0.2A1011(pRAR205) + 24 ± 2

a The percentages of radiolabeled bacteria attached to Zinnia leaf meso-phyll cells after 2 h of incubation were determined as described previously(11). Values are means ± standard errors of the mean of three measurements.

well as the contiguous 6.4-kb SalI fragment encompassingthe chvA locus. Although pCD522 was also not able tocomplement mutants in the 3' portion of chvB, it fullycomplemented genomic insertions in the chvA locus (Fig. 4).These results show that complementation of mutants withtransposon insertions in chvA and chvB is possible. Al-though plasmids pCD522, pCD526, and pRAR201 containDNA sequences spanning the left portion of chvB, includingthe region in which wild-type insertion 116 is located, theyare not able to complement mutants with insertions in thisregion. This suggests that these plasmids do not containcomplete transcriptional units and that sequences further tothe right of insertion 116 are required for the trancription ofthe left portion of chvB. Similarly, pRAR213 and pLS72(pCD521 with wild-type Tn3::HoHol insertion 72; used tofacilitate selection), which span the right portion of chvB andlack only the 3' end of this locus, were not able to comple-ment mutants with transposon insertions in chvB (Fig. 4).These results show that only plasmids containing chvB in itsentirety can complement chvB chromosomal mutants.

Further complementation experiments were performed toconfirm that the entire chvB locus behaved as a singleoperon. Since TnS and Tn3 normally cause poplar mutations(38, 41), plasmids with insertions within chvB would not beexpected to complement mutants with genomic insertions inchvB if the locus contained a single operon. Selected plas-mids with Tn3::HoHol insertions were mobilized into mu-tants with genomic Tn5 insertions to construct a series ofmerodiploid strains, which were then tested for virulence.The plasmnids used in constructing these strains were allderived from pCD523, differing only in the locations oftransposon insertions. Plasmid pLS20, containing an inser-tion outside both chvA and chvB, complemented all mutantswith TnS insertions in these loci (Fig. 4), but plasmidspLS39, pLS60, pLS64, pLS12, and pLS32, containing inser-tions spanning chvB, were not able to complement any TnSchvB mutants (Fig. 4). However, each of these plasmnidsrestored virulence to strain A2505, which contained a TnLSinsertion in chvA. Plasmid pLS45, which contained aninsertion in chvA, failed to complement strain A2505 butrestored virulence to all mutants with TnS insertions in chvB(Fig. 4).Taken together, the complementation data are consistent

with the interpretation that chvA and chvB are single tran-scriptional units. However, since the analyses describedabove were performed in a rec+ Agrobacterium background,it is possible that apparent complementation could haveresulted from recombination events. The fact that apparentcomplementation was not observed in many cases evenwhen large regions of homology with chromosomal DNAwere present on introduced plasmids makes this an unlikelyexplanation. In some experimnents, the attempted comple-mentation of chvB mutants with an incomplete chvB operonor a chvB operon with transposon insertions allowed a very

weak virulence response on Kalanchoe leaves, but thisresponse was clearly different from wild-type virulence andoccurred only when the complementing DNA extended >1kb on both sides of the mutation. These occasional weakresponses were assumed to be due to recombinational res-cue and were scored as avirulent. If recombination were thecause of virulence in putatively complemented strains, thepercentage of bacteria which underwent recombination andthus were both virulent and wild type in attachment abilitywould be expected to be low, since recombination is aninfrequent event. However, the attachment ability of strainA1011 complemented with pRAR20S was identical to that ofwild-type strain A348 (Table 2). Other mutants comple-mented with pRAR205 gave similar results (data not shown),suggesting that the strains were homogeneous populations ofvirulent bacteria with wild-type attachment ability and thusthe result of genetic complementation rather than recombi-nation.

Deletion of the 5' end of chvB. From the complementationstudies described above, it appeared that the chvB locusconsisted of a single operon about 5 kb long. Complementa-tion analysis suggested that the expression of genes in the 3'distal portion of the locus depended on the presence of apromoter in the 5' region of the locus. To test this directly,we determined the effect of deletion of about 1 kb of the 5'portion of chvB on the expression of a gene in the 3' end ofthe locus. The lac fusion caused by Tn3: :HoHol insertion 32in the extreme 3' end of chv B (Fig. 2) provided a convenientassay for the expression of such a distal gene.A derivative of plasmid pLS32 (pCD523 with insertion 32

[Fig. 2]) containing a deletion of the 5' end of chvB wasconstructed. B3ecause of the large size of pLS32 and its manyrestriction sites, it was necessary to first create the deletionin a smaller plasmid and then recombine the deletion intopLS32, relying on a strategy similar to that used to constructpCD523 (Fig. 2). Plasmid pCD521 contains two PstI sites,one located 1 kb within the 5' end of chvB and the otherlocated 2 kb outside the locus in a region in which insertionshad no effect on virulence. Digestion of pCD521 with PstIand religation created a plasmid with a 3-kb deletion, includ-ing 1 kb of the 5' end of chvB. This PstI deletion (APst) wassubcloned in a BamHI fragment into pUC13 and cotrans-formed with pLS32 into E. coli MM294. Cointegrate plasm-ids of pLS32 and pUC13-APst were selected by mobilizationinto E. coli C2110, in which pUC13 cannot replicate, andselection for the chloramphenicol resistance carried onpUC13. Cointegrate breakdown due to a second recombina-tion event was detected by screening for loss of chloram-phenicol resistance in the transconjugants. Among the ex-pected products of this breakdown were derivatives ofpLS32 in which the second recombination event took placeon the opposite side of the PstI deletion from the recombi-nation event which created the cointerate, leading to thetransfer of the PstI deletion to pLS32. Among the chloram-phenicol-resistant transconjugents a strain harboring pLS32

TABLE 3. ,-Galactosidase activity of pLS32 and pLS32APstStrain 3-Galactosidase activity (U)"

A348 2.6 ± 0.5A348(pLS32) 30.5 ± 3.8A348(pLS32APst) 4.7 ± 0.7A348(pLS12) 3.5 ± 1.2

a Means ± standard errors of at least three separate assays performed bythe method of Miller (33).

VOL. 161, 1985

on August 18, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

858 DOUGLAS ET AL.

with the PstI deletion was found. Restriction analysis indi-cated that the only change in this plasmid was the loss of the3-kb PstI fragment including 1 kb of the 5' portion of chvB.The ,-galactosidase activity of wild-type Agrobacterium

sp. strain A348 containing pLS32APst was compared withthat of strain A348 containing pLS32 and pLS12, whichcontains the lacZ gene in an orientation opposite that oftranscription of chvB (Table 3). The level of ,-galactosidaseexpression from pLS32APst was reduced from about 30 U inthe pLS32-containing strain to about 4 U. This level wasonly slightly higher than the basal level found in strain A348and was similar to the level of activity expressed by pLS12.Thus, deletion of the 5' end of the chvB locus eliminated theexpression of the most distal gene in chvB, supporting theinterpre-tation that a single promoter is responsible for thetranscription of the entire chvB locus.

DISCUSSION

Our results demonstrate that TnS and Tn3::HoHol inser-tions which cause an avirulent, attachment-defective pheno-type are clustered within an il-kb portion of the Agrobac-terium chromosome and define a chromosomal virulenceregion. Two distinct chromosomal loci are required forvirulence. These loci, designated chvA and chvB, are sepa-rated by 4 kb of DNA, insertions into which have no effecton viruIence, and are transcribed in opposite directions, Weintroduced an additional 19 insertions in a 9-kb fragment tothe left of the chvA locus. None of the insertions had aneffect on virulence. Thus, it appears that there are noadditional regions in the immediate vicinity of chvA andchvB required for virulence. Both chvA (about 1.5 kb) andchvB, (about 5 kb) appear to be single transcriptional units,since mutations within either locus could not be comple-mented by plasmids having insertions in the same locus.Furthermore, appropriate chvB mutants could not be com-plemented with plasmids pCD522, pCD526, and pRAR201,which lack the 5' end of chvB, or plasmids pRAR213 andpLS72, which lack the 3' end of chvB. Only plasmidscontaining chvB in its entirety can complement chvB chro-mosomal mutants.Two insertions, one Tn3 and the other Tn3::HoHol,

within the chvB locus did not eliminate virulence. Theseinsertions may be located intergenically or near the ends ofgenes within the locus and have nonpolar effects because ofpromoter activity within the Tn3 or Tn3::HoHol element.Similar examples of nonpolar effects of transposon inser-tions have been reported previously for TnS (i) and for someTn3 and Tn3:.HoHol insertions in Ti plasmid vir genes (21;S. Stachel, unpublished data).Tn3::HoHol was used to mutagenize chromosomal genes

and simultaneously create lac fusions with the genes, therebyproviding a method for the assay of their activity. Theexpression of the most distal portion of chvB, which wasassayed by the f-galactosidase activity of Tn3: :HoHolinsertion 32, was abolished by deletion of the 5' end of chvB,demonstrating conclusively that a single promoter controlsthe activity of the entire locus. Based on the mapping oftransposon insertions in the 0.85-kb EcoRI fragment contain-ing the 5' end of chvB (Fig. 2), this presumptive promotermust lie within a 3.0- to 0.5-kb portion of this fragment, tothe left of wild-type insertion 71 (Fig. 2). Because of its largesize, chvB is likely to be an operon containing several genes.Our analysis of lac fusions within the chvA and chvB loci

made it possible to investigate the control of the expressionof these loci. Preliminary results indicate that both chvA and

chvB are expressed constitutively and that this expression isnot affected by the presence of plant cells (C. Douglas and C.Thienes, unpublished data).The defective attachment of the chromosomal mutants

implies that they have alterations in their cell surfaces, butthe nature of the changes responsible for defective attach-ment are unknown. Whatever the nature of the cell surfacedefect(s) in these strains, it is likely that it alters thearchitecture of the outer membrane of the mutants, sincemutants in chvB have been shown to lack flagella, resultingin resistance to two phages (6). However, the lack of flagellaalone has no effect on virulence attachment ability. Althoughit seems likely that defective attachment due to mutations inchvA and chvB is directly responsible for the virulence ofthese strains, we cannot exclude the possibility that a cellsurface change in the chromosomal mutants pleiotropicallyaffects both attachment ability and other required virulencefunctions, such as the activity of vir gene products.A minimum of two chromosomal loci, one of which is an

operon large enough to comprise several genes, is necessaryfor conferring the cell surface characteristics required forvirulence. The function(s) of these loci is unknown. Theycould be involved in the synthesis and export of cell surfacepolysaccharidde or proteins, processes which often requirethe functions of many genes (25, 35).

In this and two previous studies (14; Puetz and Garfinkel,unpublished data), random TnS mutagenesis with suicideplasmid pJB4JI identified a total of eight avirulent chromo-somal mutants defective in attachment. With one exception,all TnS insertions mapped to an 11-kb region. One strain,strain A2506, contained a TnS insertion outside this region;however, an additional 2.2-kb insertion was detected in thisstrain near the Tn5 insertion of strain A1011. The presenceof the 2.2-kb insertion in close proximity to other avirulentinsertions within this region suggests that it, not the TnSelement located elsewhere, causes the avirulent phenotype.Although the 2.2-kb insertion has not been characterized, itis likely to be an endogenous Agrobacterium IS element.Such elements have been detected in other studies (14).Serious problems have been encountered with the use ofpJB4JI in Rhizobium meliloti, including cotransposition ofMu sequences and lack of linkage between TnS and mutantphenotypes (7), but these problems generally did not occurin our studies of Agrobacterium. Mu sequences were neverfound contiguous to Tn5, and linkage between TnS and amutant phenotype was demonstrated in most cases. Anexception to this (in strain A2506) is discussed above.Additionally, two avirulent mutants originally described ashaving chromosomal TnS insertions but retaining wild-typeattachment ability (strains A2502 and A2503 [11]) wereshown subsequently to be avirulent due to defective Tiplasmids and not chromosomal TnS insertions (Douglas,unpublished data).

All bona fide avirulent chromosomal mutants which wehave identified by suicide mutagenesis have insertions lo-cated within the two chromosomal virulence regions identi-fied in this study. There are other chromosomal mutantswhich have altered host ranges (14), but the locations of TnSinsertions in these strains are not known. The randomchromosomal TnS mutants were obtained from screening forvirulence of approximately 23,000 kanamycin-resistant col-onies. Although it seems likely that there are other chromo-somal genes required for virulence and attachment to plantcells, either such genes must be less susceptible to TnSinsertion, or the insertions may result in bacteria that areeither not viable or grow very poorly.

J. BACTERIOL.

on August 18, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

A. TUMEFACIENS CHROMOSOMAL VIRULENCE REGION 859

ACKNOWLEDGMENTS

We thank Scott Stachel for helpful suggestions in cloning exper-iments and Corty Thienes for technical assistance.

This work was supported by grant 82-CRCR-1-1102 from the U.S.Department of Agriculture and by Public Health Service grant 5 R01GM32618-12 from the National Institutes of Health. R.J.S. wassupported by a fellowship from the Consejo Nacional de Investigaci-ones Cieutificas y Tecnicas, Argentina. R.A.R. was supported by apostdoctoral fellowship from the National Institutes of Health.

LITERATURE CITED1. Berg, D. E., A. Weiss, and L. Crossland. 1980. The polarity of

TnS insertion mutants in Escherichia coli. J. Bacteriol.142:439-446.

2. Beringer, J. E., J. L. Beynon, A. V. Buchanon-Wollaston, andA. W. B. Johnston. 1978. Transfer of the drug-resistance tran-sposon Tn5 to Rhizobium. Nature (London) 276:633-634.

3. Birnboim, H. C., and J. Doly. 1979. A rapid extraction proce-dure for screening recombinant plasmid DNA. Nucleic AcidsRes. 7:1513-1525.

4. Blattner, F. R., B. G. Williams, A. E. Blechl, K. Denniston-Thompson, H. E. Farber, L. A. Furlong, D. J. Cyrunwa1d, D. D.Kiefer, J. W. Schumm, E. L. Sheldon,' and 0. Smithies. 1977.Charon phages: safer derivatives of bacteriophage lambda forDNA cloning. Science 196:161-169.

5. Boyer, H. W., and D. Roulland-Dussoix. 1969. A complementa-tion ,analysis of the restriction and, modification of DNA inEscherichia coi. J. Mol. Biol. 41:459-472.

6. Bradley, D. E., C. J. Douglas, and J. Peschon. 1984. Flagella-specific bacteriophages of Agrobacterium tumefaciens: demon-stration of virulence of nonmotile mutants. Can. J. Microbiol.30:676-681.

7. Buikema, W. J., S. R. Long, S. E. Brown, R. C. van den Bos, C.Earl, and F. M. Ausubel. 1983. Physical and genetic character-ization of Rhizobium meliloti symbiotic mutants. J. Mol. Appl.Genet. 2:249-260.

8. Chilton, M.-D., T. C. Currier, S. K. Farrand, A. J. Bendich,M. P. Gordon, and E. W. Nester. 1974. Agrobacterium DNAand PS8 bacteriophage DNA not detected in crown gall tumors.Proc. Natl. Acad. Sci. U.S.A. 71:3672-3676.

9. Chilton, M.-D., M. H. Drummond, D. J. Merlo, D. Sciaky, A. L.Montoya, M. P. Gordon, and E. W. Nester. 1977. Stableincorporation of plasmid DNA into higher plant cells: themolecular basis of crown gall tumorgenesis. Cell 11:263-271.

10. Ditta, G., S. Stanfield, D. Corbin, and D. R. Helinski. 1980.Broad host range DNA cloning system for Gram-negativebacteria: construction of a gene bank of Rhizobium meliloti.Proc. Natl. Acad. Sci. U.S.A. 77:7347-7351.

11. Douglas, C., W. Halperin, and E. W. Nester. 1982. Agrobacte-rium tumefaciens mutants affected in attachment to plant cells.J. Bacteriol. 152:1265-1275.

12. Draper, J., I. A. MacKenzie, M. R. Davy, and J. P. Freeman.1983. Attachment of Agrobacterium tumefaciens to mechani-cally isolated Asparagus cells. Plant Sci. Lett. 29:227-236.

13. Friedman, A. M., S. R. Long, S. E. Brown, W. J. Buikema, andF. M. Ausubel. 1982. Construction of a broad host range cosmidcloning vector and its use in the genetic analysis of Rhizobiummutants. Gene 18:289-296.

14. Garfinkel, D. J., and E. W. Nester. 1980. Agrobacterium tume-faciens mutants affected in crown gall tumorigenesis and oc-topine catabolism. J. Bacteriol. 144:732-743.

15. Garfinkel, D. J., R. B. Simpson, L. W. Ream, F. F. White, M. P.Gordon, and E. W. Nester. 1981. Genetic analysis of crown gall:fine structure map of the T-DNA by site-directed mutagenesis.Cell 27:143-153.

16. Gelvin, S. B., M. F. Thomashow, J. C. McPherson, M. P.Gordon, and E. W. Nester. 1982. Sizes and map positions ofseveral plasmid-DNA-encoded transcripts in octopine-typecrown gall tumors. Proc. Natl. Acad. Sci. U.S.A. 79:76-80.

17. Glogowski, W., and A. G. Galsky. 1978. Agrobacterium tume-faciens site attachment as a necessary prerequisite for crowngall tumor formation on potato discs. Plant Physiol. 61:

1031-1033.18. Grunstein, M., and D. S. Hogness. 1975. Colony hybridization: a

method for the isolation of cloned DNAs that contain a specificgene. Proc. Natl. Acad. Sci. U.S.A. 72:3961-3965.

19. Inze, D., A. Follin, M. Van Lijsebettens, C. Simoens, C. Gene-tello, M. Van Montagu, and J. Schell. 1984. Genetic analysis ofthe individual T-DNA genes of Agrobacterium tumefaciens;further evidence that two genes are involved in indole-3-aceticacid synthesis. Mol. Gen. Genet. 194:265-274.

20. Kingsbury, D. T., and D. R. Helinski. 1973. Temperature-sensi-tive mutants for the replication of plasmids in Escherichia coli:requirement for deoxyribonucleic aid polym,erase I in the repli-cation of the plasmid ColEl. J. Bacteriol. 114:1116-1124.

21. Klee, H. J., F. F. White, V. N. Iyer, M. P. Gordon, and E. W.Nester. 1983. Mutational analysis of the virulence region of anAgrobacterium tumefaciens Ti plasmid. J. Bacteriol. 153:878-883.

22. Knauf, V. C., and E. W. Nester. 1983. Wide host range cloningvectors: a cosmid clone bank of an Agrobacterium Ti plasmid.Plasmid 8:45-54.

23. Leong, S., G. S. Ditta, and D. R. Helinski. 1982. Heme biosyn-thesis in Rhizobium. Identification of a cloned gene coding forb-aminolevulinic acid synthetase from Rhizobium meliloti. J.Biol. Chem. 257:8724-8730.

24. Lippincott, B. B., and J. A. Lippincott. 1969. Bacterial attach-ment to a specific wound site as an essential stage in tumorinitiation by Agrobacterium tumefaciens. J. Bacteriol. 97:620-628.

25. Makela, P. H., and A. D. Stocker. 1981. Genetics of the bacterialcell surface, p. 219-264. In S. Glover and D. Hopwood (ed.),Genetics as a tool in microbiology. Cambridge University Press,Cambridge.

26. Maniatis, T., A. Jeffrey, and D. G. Kleid. 1975. Nucleotidesequence of the rightward operator of phage X. Proc. Natl.Acad. Sci. 'U.S.A. 72:1184-1188.

27. Marmur, J. 1961. A procedure for the isolation of deoxyribo-nucleic acid from micro-organisms. J. Mol. Biol. 3:208-218.

28. Marton, J., G. J. Wullems, L. MolendiJk, and R. A. Schilperoort.1979. In vitro transformation of cultured cells from Nicotianatabacum by Agrobacterium tumefaciens. Nature (London)277:129-131.

29. Matthysee, A. G. 1983. Role of bacterial cellulose fibrils inAgrobacterium tumefaciens infection. J. Bacteriol. 154:906-915.

30. Matthysse, A. G., and R. H. G. Gulitz. 1982. Plant cell range forattachment of Agrobacterium tumefaciens tissue culture cells.Physiol. Plant Pathol. 21:381-387.

31. Matthysse, A. G., K. V. Homes, and R. H. Gurlitz. 1981.Elaboration of cellulose fibrils by Agrobacterium tumefaciensduring attachment to carrot cells. J. Bacteriol. 145:583-595.

32. Matthysse, A. G., P. M. Wyman, and K. V. Holmes. 1978.Plasmid-dependent attachment of Agrobacterium tumefaciensto plant tissue culture cells. Infect. Immun. 22:516-522.

33. Miller, J. H. 1972. Experiments in molecular genetics. ColdSpring Harbor Laboratory, Cold Spring Harbor, N.Y.

34. Ohyama, K., L. E. Pekher, A. Schaefer, and L. C. Fowke. 1979.In vitro binding of Agrobacterium tumefaciens to plant cellsfrom suspension culture. Plant Physiol. 63:382-387.

35. Osborn, M. J., and H. C. P. Wu. 1980. Proteins of the outermembrane of gram negative bacteria. Annu. Rev. Microbiol.34:369-422.

36. Pueppke, S. G., and U. K. Benny. 1983. Agrobacterium tumor-igenesis in potato: effect of added Agrobacterium lipopolysac-charides and the degree of methylation of added plant galcatur-ans. Physiol. Plant Pathol. 23:439-446.

37. Ream, L. W., and M. P. Gordon. 1982. Crown gall disease andprospects for the genetic manipulation of plants. Science218:854-859.

38. Rubens, C., F. Heffron, and S. Falkow. 1976. Transposition of aplasmid deoxyribonucleic acid sequence that mediates ampicil-lin resistance: independence from host rec functions and orien-tation of insertion. J. Bacteriol. 128:425-434.

39. Ruvkin, G. B., and F. M. Ausubel. 1981. A general method forsite-directed mutagenesis in prokaryotes. Nature (London)

VOL. 161, 1985

on August 18, 2019 by guest

http://jb.asm.org/

Dow

nloaded from

860 DOUGLAS ET AL.

289:85-88.40. Schroeder, G., S. Waffenschmidt, E. W. Weiler, and J. Schroe-

der. 1984. The T-region of Ti plasmids codes for an enzyme

synthesizing indole-3-acetic acid. Eur. J. Biochem. 138:387-391.41. Shaw, K. J., and C. M. Berg. 1979. Escherichia coli autotrophs

induced by the insertion of transposable element Tn5. Genetics92:741-747.

42. Southern, E. M. 1975. Detection of specific sequences among

DNA fragments separated by gel electrophoresis. J. Mol. Biol.98:503-527.

43. Thomashow, M. F., R. Nutter, A. L. Montoya, M. P. Gordon,and E. W. Nester. 1980. Integration and organization of Ti-plasmid sequences in crown gall tumors. Cell 19:729-739.

44. Van Larebeke, N., G. Engler, M. Holsters, S. Van der Elsocker,I. Zaenen, R. A. Schilperoot, and J. Scheil. 1974. Large plasmidin Agrobacterium tumefaciens essential for crown gall inducingactivity. Nature (London) 252:255-264.

45. Van VUet, F., B. Silva, M. Van Montagu, and J. Schell. 1978.Transfer of RP4::Mu plasmids to Agrobacterium tumefaciens.Plasmid 1:446-455.

46. Viera, J., and J. Messing. 1982. The pUC plasmids, an M13mp7-derived system for insertion mutagenesis and sequencing withsynthetic universal primers. Gene 19:259-268.

47. Watson, B., T. C. Currier, M. P. Gordon, M.-D. Chilton, andE. W. Nester. 1975. Plasmid required for virulence of Agrobac-terium tumefaciens. J. Bacteriol. 123:255-264.

48. Whatley, M. H., J. S. Bodwin, B. B. Lippincott, and J. A.Lippincott. 1976. Role forAgrobacterium cell envelope lipopoly-saccharide in infection site attachment. Infect. Immun. 13:1080-1083.

49. White, F. F., H. J. Klee, and E. W. Nester. 1983. In vivopackaging of cosmids in transposon-mediated mutagenesis. J.Bacteriol. 153:1075-1078.

50. Willmitzer, L., G. Simons, and J. Schell. 1982. The TL-DNA inoctopine crown-gall tumors codes for seven well-defined polyad-enylated transcripts. EMBO J. 1:139-146.

51. Zambryski, P., M. Holsters, K. Kruger, A. Depicker, J. Schell,M. Van Montagu, and H. M. Goodman. 1980. Tumor DNAstructure in plant cells transformed by A. tumefaciens. Science209:1385-1391.

J. BACTERIOL.

on August 18, 2019 by guest

http://jb.asm.org/

Dow

nloaded from