Proc. Natl. Acad. Sci. USAVol. 92, pp. 230-234, January 1995Plant Biology

Intracellular Agrobacterium can transfer DNA to the cell nucleusof the host plantJESUS ESCUDERO*, GUNTHER NEUHAUSt, AND BARBARA HOHN*t*Friedrich Miescher-Institut, Postfach 2543, CH-4002 Basel, Switzerland; and tlnstitut fuir Pflanzenwissenschaften, Eidgenossiche Technische Hochschule-Zentrum, Universitatstrasse 2, CH-8092 Zurich, Switzerland

Communicated by Frederick M. Ausubel, Massachusetts General Hospital, Boston, MA, September 29, 1994 (received forreview June 23, 1994)

ABSTRACT Agrobacterium tumefaciens is a Gram-negative, soil-borne bacterium responsible for the crown galldisease ofplants. The galls result from genetic transformationof plant cells by the bacteria. Genes located on the transferredDNA (T-DNA), which is part of the large tumor-inducing (Ti)plasmid of Agrobacterium, are integrated into host plantchromosomes and expressed. This transfer requires virulence(vir) genes that map outside the T-DNA on the Ti plasmid andthat encode a series of elaborate functions that appear similarto those of interbacterial plasmid transfer. It remains a majorchallenge to understand how T-DNA moves from Agrobacte-rium into the plant cell nucleus, in view of the complexity ofobstacles presented by the eukaryotic host cell. Specificanchoring of bacteria to the outer surface of the plant cellseems to be an important prelude to the mobilization of theT-DNA/protein complex from the bacterial cell to the plantcell. However, the precise mode of infection is not clear,although a requirement of wounded cells has been docu-mented. By using a microinjection approach, we show herethat the process of T-DNA transfer from the bacteria to theeukaryotic nucleus can occur entirely inside the plant cell.Such transfer is absolutely dependent on induction ofvir genesand a functional virB operon. Thus, A. tumefaciens can func-tion as an intracellular infectious agent in plants.

Crown gall is a neoplastic disease caused by Agrobacteriumtumefaciens in dicotyledonous plants. The engineering of thisgenetic change in the eukaryotic host cell represents a highlyevolved case of microbe-plant interaction (for review, see refs.1-4). For the plant cell, the major consequence of this naturalinterkingdom DNA transfer is its transformation into a tu-morigenic state in which, as is the case in animal tumorigen-esis, new mechanisms govern growth and differentiation (5).DNA transferred to the host plant cell, known as T-DNA for

transferred DNA, represents a defined segment on a large (200kb) extrachromosomal element originally isolated from viru-lent agrobacteria and named pTi for tumor-inducing plasmid(6). Transformation occurs by a process in which bacteria arerequired for the establishment of neoplastic development, buttheir continued presence in tumor cells is not needed (7). Uponintegration into the plant-cell genome T-DNA overproducesplant growth hormones (auxin and cytokinin), which results inthe cancerous phenotype, as well as other compounds (opines)believed to serve as nutrients for the infecting bacteria (8, 9).Besides T-DNA, the Ti plasmid also contains an essentialgenetic component (vir) known as virulence region. vir genesare designed to provide the cellular machinery that T-DNAneeds both in bacteria and in the plant. In addition to the virgenes located on the Ti plasmid, Agrobacterium requires somechromosomal genes for efficient plant colonization (1-4).

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

On the basis of homology between Agrobacteiiwn virulencegenes and the transfer genes of several conjugative plasmids, it nowappears that T-DNA is transferred to plant cells by a mechanismanalogous to bacterial conjugation (10), in which the recipientbacterial cell is replaced by a eukaryotic cell. The current model forplant infection by Agrobacteriwn involves several steps. Amongthem, chemotaxis to plant wounds and attachment to the plant cellsurface are the earliest critical events allowing the cell-cell contactneeded for DNA transfer. Transcriptional induction of the viroperons by phenolic compounds and sugars (11, 12), released fromwounded plant cells, results in the processing of T-DNA molecules(the so-calledT strands), and it has been suggested that the proteinsencoded by the virB operon provide the structural apparatus thatallows the export of the T-DNA complex through the bacterialenvelope (1-4).DNA delivered from the bacteria has been recentlyshown to reach the plant-cell nucleus in single-stranded form (13),in analogy to bacterial conjugation. However, the mechanism bywhich this T-DNA complex passes from the bacterium to theplant-cell nucleus remains enigmatic, especially when the eukary-otic cell barriers that the T-DNA has to cross are considered.

After early microscopic observations (for review, see ref. 14)and some microinjection studies using a tumorigenesis assay(15), Agrobacterium has been described as an obligately ex-tracellular pathogen. However, in these experiments bacteriacultured in standard broth medium were used, and the neces-sity for transcriptional induction of bacterial virulence genesby plant-released compounds was unknown at that time. Herewe addressed the question whether the early stage of bacterialbinding to plant cells is required for T-DNA transfer. For thispurpose we analyzed the behavior of intracellular agrobacte-ria. With the help of a very sensitive assay for transientexpression of a chimeric reporter gene (16), we studied thetransfer of T-DNA from Agrobacterium cells deposited withinplant cells by microinjection. Our results suggest that transferof T-DNA does occur under these intracellular conditions andthat it is absolutely dependent on both vir gene activation andVirB functions. In addition, we show that most likely bacteriadelivered into a plant cell by microinjection release T-DNAcopies into the nucleus of the very same plant cell.

MATERIALS AND METHODSBacteria and Plasmids. A. tumefaciens strains used are

listed in Table 1. Strain C58 is a wild-type virulent bacteriumthat harbors the corresponding nopaline Ti plasmid (17).Bacterial strains A136 and A348 (18), as well as LBA1100 andLBA1143 (19), are all C58 derivatives cured of their originalpTi. A136 contains no pTi plasmid and consequently is avir-ulent; A348 contains a wild-type octopine pTiA6 conferringvirulence; LBA1100 contains the virulent vir-helper plasmidpAL1100, an octopine pTiB6 in which the T-DNA, conjuga-tion transfer, and octopine catabolism regions have been

Abbreviations: T-DNA, transferred DNA; GUS, ,3-glucuronidase; AS,acetosyringone; CaMV, cauliflower mosaic virus.tTo whom reprint requests should be addressed.

230

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0

Proc NatL Acad Sci USA 92 (1995) 231

Table 1. Agrobacterium strains and plasmids

Strain Ref. Characteristics GUS plasmid*C58 17 Wild-type, nopaline pTiC58 pBG5A136 18 pTi cured of C58 pBG5A348 18 A136, octopine pTiA6 pCG5, pC-90GLBA1100 19 Octopine pTiB6 derivative pCG5LBA1143 19 pTiB6(virB4::Tn3) derivative pCG5A6.1h 20 Attachment defective, pTiA6 pCG5*Two different plasmid vectors (binaries) containing either nopaline(pBG5) or octopine (pCG5, pC-90G) T-DNA border sequences wereused, depending on the kind of native Ti plasmid in the bacterial strain.

deleted; LBA1143 contains pAL1100 with the transposonTn3Hohol inserted in the virB4 gene, creating a polar andavirulent mutation (21). Strain A6.1h is derived from theoctopine type A6 and contains a TnS insertion in the pscA(exoC) chromosomal locus, resulting in the inability of thebacteria to attach to plant cells and in loss of virulence (20).Both nopaline TiC58 and octopine TiA6 or TiB6-derived

plasmids were tested in combination with the appropriatebinary vector plasmid carrying a T-DNA encoding 3-glucuron-idase (GUS; ref. 22). pBG5 (nopaline borders) and pCG5(octopine borders) contain the N-terminal start of cauliflowermosaic virus (CaMV) gene V fused to the GUS codingsequence (23) and driven by a complete CaMV 35S promoter.This GUS fusion has been previously reported as a T-DNAmarker in plants (24, 25). pC-90G was constructed using theEcoRI-BamHI fragment from plasmid X-GUS-90 (26), con-taining the domain A (-90 to +8) of the CaMV 35S promoter(here named "Mini" promoter), the GUS coding sequence,and the 3' end from the small subunit of the ribulose bisphos-phate carboxylase 3C gene of pea. This fragment was clonedinto the polylinker of pCGN1589 (27). Plasmid pGUS23 (28)contains the same CaMV-GUS fusion as pBG5 and pCG5. Forstandard cloning techniques Escherichia coli DH5a andNM522 strains were used (29), and purified plasmids wereintroduced into Agrobacterium by electroporation (30).

vir-Gene Induction Conditions.Agrobacterium was grown at28°C in YEB medium (31) with appropriate antibiotics (ri-fampicin, 20 mg/liter; kanamycin, 50 mg/liter; neomycin, 40mg/liter; gentamycin, 20 mg/liter). Overnight shaking (250rpm) cultures were washed and diluted in liquid M9 minimalmedium (29) to an OD600 of 0.1. Two kinds of M9 bacterialcultures were prepared: (i) Preinduced, in the presence of 0.2mM acetosyringone (AS) at pH 5.5; (ii) nonpreinduced,without AS at pH 7. After 12 hr at 28°C, 250 rpm, bacteria werecollected, washed with 10 mM MgSO4, and adjusted to OD6wo= 1 (=109 colony-forming units per ml) in 10 mM MgSO4,before use. Induction conditions were assayed by using thebacterial strain A348(pSM219) carrying a virH::lacZ fusionand measuring 13-galactosidase activity as described (32).

Injection of Plant Cells and GUS Assay. Seeds of Nicotianatabacum cv. Petit Havana, SR1 line, were germinated in plasticdishes with moist filter paper under sterile conditions in agrowth chamber. Cotyledons of 7- to 10-day-old tobaccoseedlings were used in all injection experiments. Normally,three single-cell injections were performed per cotyledon. Themesophyll layer immediately underneath the epidermis waschosen as target to facilitate microscopic observation. Theinjection capillary (0.7-1 ,Lm in diameter) was coupled to anEppendorf 5242 microinjector, and the pressure was adjustedto allow delivery of 10 pl. Cytoplasmic microinjections wereoptimized by using fluorescent dyes (lucifer yellow) as has beenreported (33, 34).TGA-la, a transcription factor specifically binding to a 21-bp

sequence in domain A of the CaMV 35S promoter (35), wasoverproduced in E. coli and purified by DNA-affinity columnchromatography. The protein was mixed withAgrobacmrium in 10mM MgSO4 before injection (-3 x 104 molecules per injected cell).

pGUS23 was injected as supercoiled pure DNA in water (105molecules per injection). Alternatively to the cell injection, seedlingswere infiltrated (24) by application of moderate vacuum (=0.4atmosphere, 5 min) with bacterial suspensions identical to thoseused for injection. Injected and infiltrated tobacco seedlings werecultivated for 3 days on MS medium (36)/agar plates understandard-growth chamber conditions. GUS activity was then as-sayed by a histochemical procedure using 2mM 5-bromo-4-chloro-3-indolyl (3-D-glucuronic acid (X-Gluc; Biosynth, Staad, Switzer-land) in 100mM phosphate buffer, pH 7/10mM Na2 EDTA/0.1%NaN3, supplemented with 3 mM K3 [Fe(CN)6] to avoid productdiffusion (37). Seedlings were dipped and infiltrated under mod-erate vacuum in the staining solution for 10 min, and the reactionwas continued for 2.5-16 hr at37C in the dark. The plant tissue wasthen fixed overnight at room temperature with 4% (vol/vol)formaldehyde/0.8% NaCl/0.1% NaN3 in 100 mM phosphatebuffer. Bleaching with ethanol and transfer to phosphate-bufferedsaline facilitated scoring for GUS-expressing cells under the ste-reomicroscope. A Zeiss Axiophot microscope was used for detailedexamination and photography. All GUS plasmids were tested forinduction ofGUS activity inAgrobacterium under conditions similarto the transient assays in plants, with negative results.

RESULTSAgrobacterium Cells Injected into Single Tobacco Cells

Mediate T-DNA Transfer. To study Agrobacterium-plant in-teraction at the unicellular level in the plant, single mesophyllcells of otherwise intact tobacco seedlings were injected witha low number (10 on average) of bacterial cells and tested forthe expression of a T-DNA-encoded GUS gene, which hadbeen modified to avoid expression in bacteria (23). As acontrol, similar tobacco seedlings were infiltrated with iden-tical bacterial suspensions, a process that introduces bacteriainto intercellular spaces. Both injected and infiltrated tobaccocells exhibited GUS activity that was strictly dependent on thepresence of the bacterial Ti plasmid. Plant-cell injection withA136(pBG5), a bacterial strain that lacks the Ti plasmid andtherefore the whole vir gene region, did not result in T-DNAgene expression (Table 2). With microinjection, a positiveresult was strictly dependent on conditions favoring transcrip-tional induction of bacterial virulence genes-i.e., preinduc-tion in AS-containing medium at acidic pH. Infiltration ob-viated this requirement because the manipulation of theseedlings caused wounding of plant cells and thereby vir geneinduction. Failure to detect T-DNA transfer from nonprein-duced bacteria after microinjection was probably a conse-quence of the minimal injury of the plant cell brought aboutby injection, which apparently produced insufficient com-pounds to activate bacterial virulence genes. These resultsrepresent a positive correlation between bacterial cultureconditions that induce vir genes on the Ti plasmid and T-DNAactivity in plant cells injected with agrobacteria.GUS gene expression after injection with preinduced Agrobac-

teriwn could result from bacteria transferring T-DNA within anintact plant cell (Fig. 1, route A). However, at least two alternativeexplanations are possible. (i) The "bacterial lysate" hypothesis: theinjected bacterial suspension might contain active T-DNA com-plexes resulting from preinduction during culture and lysis duringmanipulation (Fig. 1, route B). (u) The "neighboring cell" hypoth-esis: injected bacteria might escape one plant cell, attach to anotherplant cell in the vicinity, and only then transferT-DNA (Fig. 1, routeC).Agrobacterium Cells Remain Metabolically Active After

Injection into Plant Cells. When plant cells were injected withbacteria cultured under noninducing conditions, reproducibly,T-DNA transfer was undetectable (see Table 2). As a test forthe bacterial lysate hypothesis, we studied this time the be-havior of nonpreinduced wild-type agrobacteria coinjectedwith 0.2 mM AS. Our expectation was to restore functionality

Plant Biology: Escudero et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0

232 Plant Biology: Escudero et al.

Table 2. T-DNA transfer in cells of tobacco seedlings injected withAgrobacterium

Bacteria not preinduced* Bacteria preinduced*

Injected plant GUS-positive Injected plant GUS-positiveInjected strain cells, no. plant cells, no. Efficiency,t % cells, no. plant cells, no. Efficiency,t %

A348(pCG5) 60 0 <1 130 18 14A348(pCG5)t 60 6 10 NAC58(pBG5) 60 0 <1 173 19 11C58(pBG5)t 80 7 9 NALBA1100(pCG5) ND 200 15 7LBA1143(pCG5) ND 480 0 <1A6.lh(pCG5) ND 144 20 14A136(pBG5) 60 0 <1 140 0 <1

NA, not applicable; ND, not determined. Data represent three independent experiments.*Bacteria were cultured under either noninducing or inducing conditions before injection.tEfficiency was calculated as the number of tobacco cells showing GUS activity (positives) per total number of injected cells after histochemicalstaining.tBacteria cultured under noninducing conditions coinjected with 0.2 mM of AS.

in the bacteria due to. the vir gene activation by the inducerafter injection. A high frequency of GUS-positive plant cellswas obtained. Thus, bacteria responded in planta to AS withefficient T-DNA transfer [Table 2, compare the A348(pCG5)data and also the C58(pBG5) data]. Because of the short time(a few minutes) between addition of the vir-inducer AS to thebacteria and injection of this suspension (pH of -7) intotobacco cells, we conclude that the Agrobacterium cells weremetabolically active after injection and that de novo geneexpression steps in the bacteria led to DNA transfer inside theplant cells. Because the combined action of AS and acidic pHseems essential for efficient activation of virulence genes (38),the bacteria might be exposed to low pH intracellularly.

Intracellular Agrobacterium in the Plant Needs the VirBProteins for Transfer of T-DNA. To test a relevant function forintraplant cellular T-DNA transfer, we chose to inject apreinduced Agrobacterium strain mutated in the virB operon,which is essential for T-DNA transfer. This mutant strain(LBA1143) has been previously described as DNA-transferdeficient but proficient in T-DNA processing (19, 39). Plantcells injected with virB-deficient agrobacteria had undetect-able levels of GUS expression, whereas its wild-type progen-itor (LBA1100) was transfer proficient [Table 2, compareLBA1100(pCG5) and LBA1143(pCG5) data]. This resultagrees with the proposed role of the VirB proteins to transport

I-Ncomplex

%,,~~~~~~~..,,............

V 4V E -b

_Agrobacferlum _T-DNA complex

FIG. 1. Schematic representation of the plant-cell microinjectionand three simple ways to explain the GUS activity arising fromAgrobacterium. Route A, bacteria injected into the plant cell deliverT-DNA, which is targeted to the plant cell nucleus; route B, bacteriaare injected as lysate; route C, bacteria injected into the plant cellattach and deliver T-DNA to juxtaposed cell.

DNA across bacterial membranes, even in the absence ofbacterial attachment to the external surface of the plant cell.Thus, the absence of T-DNA expression in the injected plantcells with the mutant strain LBA1143 suggests again that thebacterial lysate hypothesis is unlikely.Upon Injection into Plant Cells Attachment-Defective

Agrobacterium Very Efficiently Transfers T-DNA. To test theneighboring cell hypothesis, we used an Agrobacterium pscA-strain (A6.1h, see Table 1) defective in attachment to plantcells and severely attenuated in virulence. In plant-inoculationassays strain A6.1h was tested repeatedly for tumorigenesis,both in turnip and in tobacco plants, and failed (data notshown). When tested by infiltration into young tobacco seed-lings, the highest numbef of GUS positives recorded fromA6.lh(pCG5) was 3% of the value seen for wild-type A348(data not shown), in agreement with the tumorigenesis test.However, after injection of the mutant into mesophyll cells,T-DNA transfer occurred at wild-type levels (Table 2). Thisresult implies that attachment per se is not required forAgrobacterium T-DNA transfer, and it is, therefore, unneces-sary to postulate invasion of neighboring plant cells by theinjected bacteria. Because GUS expression was generallydetected only in single plant cells after injection of bacteria(see Fig. 2a) T-DNA transfer was limited to injected cotyledoncells.

Is the Injected Plant Cell Expressing T-DNA Genes? Afurther experiment was devised to test whether injected plantcells could express the T-DNA marker. For this purpose, weused a bacterial strain containing a defective GUS gene, theexpression of which absolutely depends on the presence of anactivator protein in the same plant cell. As T-DNAwe used theplasmid pC-90G, carrying the GUS coding sequence fused tothe A domain ("Mini") of the CaMV 35S promoter. StrainA348 carrying this T-DNA was coinjected into tobacco cellswith TGA-la protein, the cognate plant transcriptional acti-vator (35). Because of its large size (40 kDa), this protein is notbelieved to move from cell to cell. It has indeed been shownto function only within the injected cell in transgenic tobaccocarrying the corresponding binding sequence (34). In addition,TGA-la protein has been shown to possess nuclear targetingfunctions in- plants (40). When strain A348(pC-90G) wasinjected together with purified TGA-la, single 6ells histo-chemically stained blue were observed. Therefore, nuclei ofplant cells injected with Agrobacterium are direct T-DNArecipients (Table 3). Expression of the defective gene wasstrictly dependent on coinjection with the TGA-la activatorprotein. The efficiency of T-DNA transfer under these con-ditions was equivalent to that obtained with the strain carryingthe T-DNA construct coupled to a complete CaMV 35Spromoter (pCG5) whose activity is TGA-la independent.Tobacco cells injected with 105 plasmid pGUS23 DNA

Proc Natt Acad Sci USA 92 (1995)

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0

Proa Nati Acad Sci USA 92 (1995) 233

FIG. 2. Representative tobacco mesophyll cells showing GUS activity after microinjection with a few Agrobacterium cells (a) or with plasmidpGUS23 DNA (-105 molecules) encoding the GUS marker gene (b). Note the indigo blue precipitate that was observed in single cells after5-bromo-4-chloro-3-indolyl )3-D-glucuronic acid staining (arrows). (Bars = 10 Am.)

molecules carrying the complete 35S promoter-GUS reporteralso resulted in single-cell GUS activity (Fig. 2b), although ata lower efficiency.

DISCUSSION

F'rom the results reported above we conclude that individual

plant cells injected with Agrobacterium can act as primaryT-DNA recipients. With our experimental set-up we cannotexclude that trafficking of T-DNA complexes occurs between

plant cells in a natural infection. However, the expression ofT-DNA in single plant cells injected with agrobacteria does notsupport this cell-to-cell T-DNA movement in the plant.Agrobacterium anchorage to the surface of plant cells is most

probably a prerequisite for crown gall disease. Nevertheless,we present definitive evidence here that at least the specificsteps responsible for T-DNA transfer from the bacterium tothe eukaryotic nucleus can occur without external binding-i.e., from inside the plant cell. Our results also point to therequirement of the VirB apparatus for the transfer of T-DNA

Table 3. Recovery of T-DNA gene expression by coinjected transcriptional activator

CaMV 35Spromoter Injected plant GUS-positive

Injected specimen* driving GUS cells, no. plant cells, no. Efficiency, %

A348(pCG5) Full 76 15 20A348(pC-90G) Mini 75 0 <1A348(pC-90G) + TGA-lat Mini 72 18 25pGUS 23 DNAt Full 88 4 4

*Agrobacteria were preinduced before use.tCoinjected (-3 x 104 molecules) into the plant cell with Agrobacterium.tA 5-5-kb-long plasmid encoding the CaMV gene V-GUS fusion used in pBG5 and pCG5 (_105molecules per injected plant cell).

Plant Biology: Escudero et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0

234 Plant Biology: Escudero et al.

from Agrobacterium within the plant cell. Although the virBgenes have been repeatedly reported to be critical for virulence(for recent data, see ref. 41 and references therein), our studiesreveal that VirB proteins likely facilitate the translocation ofthe T-DNA complex out of the bacterial cell. Thus, themechanism by which the bacteria transfer T-DNA from withinremains unclear but triggering of the transfer by the nuclearmembrane might be one possibility.Our evidence of infection within the plant cell clearly

conflicts with the conclusions of Hildebrand (15), although hisresults agree with the present study in that nonpreinducedagrobacteria are deficient in T-DNA transfer when injectedinto plant cells. Agrobacterium, therefore, may have the po-tential to be an intracellular plant parasite, as suggested bySmith (14). This property may be related to the evolutionaryorigin of Agrobacterium, in agreement with the homologiesfound for the 16S rRNAs among agrobacteria, rhizobacteria,rickettsias, and plant mitochondria (42).We do not know whetherAgrobacterium invades or becomes

internalized in plant cells as part of its repertoire of naturalinfection. The occasional survival of agrobacteria in antibiotic-treated plant tissue during transformation experiments (R. A.Ludwig, personal communication) should perhaps be reeval-uated in the light of our findings. In mammalian systems,invasive bacteria are protected from a variety of antibiotics,whereas noninvasive strains are killed (43).TheAgrobacterium-plant interaction consists of a large and

yet partially known series of events in which many bacterialgenes need to be activated. The microinjection techniqueapplied here may be useful for studying the function of certainbacterial genes (or discovering new ones) that are required forpathogenesis. For instance, virulence functions could be iden-tified that are required in the early stage of bacterium-plantcell recognition, which are essential for mobilization of T-DNA within the plant cell and which are involved in integra-tion of T-DNA.Although the results presented here are based on T-DNA

transient expression, our approach may have potential forplant transformation in cases in which the host-cell recognitionis not efficient under conditions of classical bacterial infection.

We thank the members of our groups, especially Zdena Kouko-lf'kovi-Nicola for useful discussion and for encouragement, and IngoPotrykus for his support. Holger Puchta and Bruno Tinland providedplasmids pBG5 and pCG5. We are indebted to Paul Hooykaas, EugeneNester, and Michael Thomashow for their gifts of LBA1100 andLBA1143, A348, and A6.1h bacterial strains, respectively. FumiakiKatagiri and Nam-Hai Chua are acknowledged for their gifts ofTGA-la and pX-GUS-90. We also thank Thomas Boller, Mary-DellChilton, Patrick King, Jacques Tempe, and Bruno Tinland for usefulsuggestions and critical comments on this manuscript. The technicalexpertise of Cynthia Ramos and Alessandro Galli is also very muchappreciated. J.E. was partially supported by a fellowship from theSpanish Ministerio de Educacion y Ciencia, while on leave of absencefrom Instituto Nacional de Investigaciones Agrarias, Madrid.

1. Hooykaas, P. J. J. & Schilperoort, R. A. (1992) Plant Mol. Bio.19, 15-38.

2. Winans, S. C. (1992) Microbiol. Rev. 56, 12-31.3. Zambryski, P. C. (1992)Annu. Rev. Plant Physiol. Plant Mol. Biol.

43, 465-490.4. Hohn, B. (1992) in Development: The Molecular Genetic Ap-

proach, eds. Russo, V. E. A., Brody, S., Cove, D. & Ottolenghi,S. (Springer, Berlin), pp. 206-216.

5. Chilton, M.-D., Drummond, M. H., Merlo, D. J., Sciaky, D.,Montoya, A. L., Gordon, M. P. & Nester, E. W. (1977) Cell 11,263-271.

6. Zaenen, I., Van Larebeke, N., Teuchy, H., Van Montagu, M. &Schell, J. (1974) J. Mol. Biol. 86, 109-127.

7. Braun, A. (1982) inMolecular Biology ofPlant Tumors, eds. Kahl,G. & Schell, J. S. (Academic, New York), pp. 155-210.

8. Willmitzer, L., Debeuckeleer, M., Lemmers, M., Van Montagu,M. & Schell, J. (1980) Nature (London) 287, 359-361.

9. Tempe, J. & Goldmann, A. (1982) in Molecular Biology ofPlantTumors, eds. Kahl, G. & Schell, J. S. (Academic, New York), pp.427-449.

10. Lessl, M. & Lanka, E. (1994) Cell 77, 1-4.11. Stachel, S. E., Nester, E. W. & Zambryski, P. C. (1986) Proc.

Natl. Acad. Sci. USA 83, 379-383.12. Cangelosi, G. A., Ankenbauer, R. G. & Nester, E. W. (1990)

Proc. Natl. Acad. Sci. USA 87, 6708-6712.13. Tinland, B., Hohn, B. & Puchta, H. (1994) Proc. Natl. Acad. Sci.

USA 91, 8000-8004.14. Rodgers, A. D. (1952) Erwin F. Smith, A Story ofNorthAmerican

Plant Pathology (Am. Philos. Soc., Philadelphia), pp. 565-577.15. Hildebrand, E. M. (1942) J. Agric. Res. 65, 45-59.16. Rossi, L., Escudero, J., Hohn, B. & Tinland, B. (1993) Plant Mol.

Biol. Rep. 11, 220-229.17. Watson, B., Currier, T. C., Gordon, M. P. & Nester, E. W. (1975)

J. Bacteriol. 123, 255-264.18. Garfinkel, D. J., Simpson, R. B., Ream, W., White, F. F. &

Gordon, M. P. (1981) Cell 27, 143-153.19. Beijersbergen, A., Dulk-Ras, A. D., Schilperoort, R. A. & Hooy-

kaas, P. J. J. (1992) Science 256, 1324-1327.20. Thomashow, M. F., Karlinsey, J. E., Marks, J. R. & Hurlbert,

R. E. (1987) J. Bacteriol. 169, 3209-3216.21. Stachel, S. E. & Nester, E. W. (1986) EMBO J. 5, 1445-1454.22. Jefferson, R. A. (1987) Plant Mol. Biol. Rep. 5, 387-405.23. Schultze, M., Hohn, T. & Jiricny, J. (1990)EMBO J. 9, 1177-1185.24. Shen, W. H., Escudero, J., Schlappi, M., Ramos, C., Hohn, B. &

Koukolikova-Nicola, Z. (1993) Proc. Natl. Acad. Sci. USA 90,1488-1492.

25. Koukolikova-Nicola, Z., Raineri, D., Stephens, K., Ramos, C.,Tinland, B., Nester, E. W. & Hohn, B. (1993) J. Bacteriol. 175,723-731.

26. Benfey, P. N. & Chua, N.-H. (1989) Science 244, 174-181.27. McBride, K E. & Summerfelt, K R. (1990) Plant Mol. Biol. 14,

269-276.28. Puchta, H. & Hohn, B. (1991) Nucleic Acids Res. 19, 5034-5040.29. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989) Molecular

Cloning: A Laboratory Manual (Cold Spring Harbor Lab. Press,Plainview, NY), 2nd Ed.

30. Cangelosi, G. A., Best, E. A., Martinetti, G. & Nester, E. W.(1991) Methods Enzymol. 204, 384-397.

31. Lichtenstein, C. P. & Draper, J. (1985) in DNA Cloning: APractical Approach, ed. Glover, D. M. (IRL, Washington, DC),Vol. 2, p. 78.

32. Stachel, S. E., An, G., Flores, C. & Nester, E. W. (1985) EMBOJ. 4, 891-898.

33. Schnorf, M., Neuhaus-Url, G., Galli, A., Iida, S., Potrykus, I. &Neuhaus, G. (1991) Transg. Res. 1, 23-30.

34. Neuhaus, G., Neuhaus-Url, G., Katagiri, F., Seipel, K. & Chua,N.-H. (1994) Plant Cell 6, 827-834.

35. Katagiri, F., Lam, E. & Chua, N.-H. (1989) Nature (London) 340,727-730.

36. Murashige, T. & Skoog, F. (1962) Physiol. Plant. 15, 473-493.37. De Block, M. & Debrouwer, D. (1992) Plant J. 2, 261-266.38. Melchers, L. S., Regensburg-Tuink, T. J. G., Bourret, R;B.,

Sedee, N. J. A., Schilperoort, R. A. & Hooykaas, P. J. J. (1989)EMBO J. 8, 1919-1925.

39. Stachel, S. E., Timmerman, B. & Zambryski, P. (1987) EMBO J.6, 857-863.

40. van der Krol, A. & Chua, N.-H. (1991) Plant Cell 3, 667-675.41. Berger, B. R. & Christie, P. J. (1994)J. Bacteriol. 176,3646-3660.42. Yang, D., Oyaizu, Y., Oyaizu, H., Olsen, G. J. & Woese, C. R.

(1985) Proc. Natl. Acad. Sci. USA 82, 4443-4447.43. Isberg, R. R. (1991) Science 252, 934-938.

Proa Natl Acad Sci USA 92 (1995)

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0