Nonreciprocal homologous recombination between Agrobacterium transferred DNA and a plant chromosomal locus

Genetics

Nonreciprocal homologous recombination between Agrobacterium

transferred

DNA

and

a

plant chromosomal locus

(Niotianatbcum/protoplasts/gene targeting/geneconversion/neomycin phosphotransferase II)

REMKOOFFRINGA*t, MARRY E. I. FRANKE-VAN DUIK*, MARCEL J. A. DE GROOTt, PETERJ. M. VAN DEN ELZENt, AND PAUL J. J. HOOYKAAS*

*Institute of Molecular Plant Sciences, Leiden University, Clusius Laboratory, Wassenaarseweg 64, 2333 AL Leiden, The Netherlands; and

tMogen

Internationalnv,Einsteinweg97,2333CBLeiden,TheNetherlands

Communicated by Mary-Dell Chilton, March 8, 1993

ABSTRACT Previously, we demonstrated the occurrence ofgene targeting intobaccoceLls afterAgrobacterium-mediated transformation. In these experiments a defective kanamycin resistance (Kmr) gene residing at achromosomal location was restored via homologous recombination with an incoming transferred DNA (T-DNA) repair construct(pSDM101) con-taining a different defective Kmr gene. In this article we describegene targeting experiments with the sametarget line, butusing an improved repairconstruct, pSDM321. In one of the Kmr calli obtained after transformation with pSDM321 (line A) the product of homologousrecombination was detected using PCR. Further molecular analysis revealed that the defectiveKmr gene present on the incomingT-DNAhad been restored via homologousrecombination with the target locus. The target locus was left unchangedand the corrected T-DNA was foundto be insertedon the samechromosome but not close to thetarget locus. This paper presents molecular evidence in plantsfor the conversionofanintroduced DNAmolecule(in

thiscase, T-DNA) by a homologous chromosomal locus. The integration of foreign DNA in higher eukaryotic cells

occurs atrandomlocithroughaprocessthat is referredto as

illegitimate recombination (1-3). Genetargeting, defined as

the integration ofintroduced DNA viahomologous recom-bination into the genome, occurs only at a relatively low frequency. Nonetheless, it has become a well-established tool for the specific inactivation or modificationofgenes in

somemammaliansystems(4).Paszkowskietal.(5)werethe first to report that homologous recombination between a

target locus and an incoming purified DNA molecule does

occurinplant cells, albeitat averylowfrequency.

Inprevious experimentsweinvestigatedthepotentialuse

ofAgrobacterium transferred DNA (T-DNA) for gene

tar-geting in plants. The transgenic tobacco line T, which is hemizygousforaT-DNAwith adefectivekanamycin resis-tance(Kmr) gene (target locus), was retransformed via Agro-bacterium with aT-DNA containing adefective Kmr gene with acomplementing nonoverlapping mutation(repair con-struct).Among 213kanamycin-resistant calliselected froma

total of105 transformants, onerecombinantline was identi-fiedbyPCRanalysis. Inthis line the defectiveKmrgeneat

thetarget locus had beenproperly restored (6).

To study the process of Agrobacterium-mediated gene targetinginmoredetailweattemptedtoreducebackground

events andtoincrease the detection sensitivityfor homolo-gous recombination events. Two important modifications were made to theoriginal repairconstructpSDM101: (i)the aux-2 gene was inserted as a segment ofnonhomologous

DNAbetween therightT-DNAborder and thepromoterless

Thepublicationcostsof this articleweredefrayedinpartby pagecharge payment.Thisarticle must therefore beherebymarked "advertisement" in accordance with 18U.S.C. §1734solelytoindicate this fact.

defective neomycinphosphotransferase II (NPTII) gene to reduce theformation of genefusions afterintegrationof the repair T-DNA; (ii) a 137-bp deletion was introducedinto the 3'noncoding region of the defective Kmr genein therepair construct, thusallowing clear distinction between aproduct ofhomologous recombinationand the wild-type construct.

Protoplasts ofplant line T, which had been used in the previous experiments (6), were cocultivated with an Agro-bacteriumcarrying the modified T-DNA. From these exper-iments a recombinant line (A) was isolated, of which a detailed molecularanalysis ispresentedhere.

MATERIALS AND METHODS

Constructs.Thebinaryvectors werederivedfromplasmids pSDM100 and pSDM101 (6) using standard molecular tech-niques (7). A base-pair substitution in the NPTII coding region (8) was corrected in pSDM100 and pSDM101 (rein-troduces an Xho IIsite). In this waypSDM300 (not shown) was derived from pSDM100. Following this base-pair ex-change in pSDM101, the 137-bp Sma I/Pst Ifragmentwas

deleted from the 3' noncodingregion oftheKmrgene,which resulted inpSDM301 (not shown). A 2543-bpHindIIIpartial thatcontained the aux-2 gene and the 5' part of the aux-l gene ofpTiAch5 (positions 3390-5933) (9) was cloned into the HindIIIsite ofpIC20R (10).The 5' partofaux-Jwasremoved up to the HincII site atposition 5721 (9) by digestionwith HincIland PstI(oneof themultiplecloning sites ofpIC20R) and religation. Subsequently, the Xho I and Sal I sites of pIC20R were removed by digestion with these enzymes followed byligationof the fragment in inverse orientation. Finally,the aux-2 genewasclonedas a

BamHI/Bgl

IIpartial

attheBcl Isite ofpSDM300 or as aBamHI/EcoRIpartialat

the EcoRI site of pSDM301. The resulting plasmids are

referred toaspSDM320 and pSDM321, respectively (seeFig. 1). The plasmids were mobilized (11) into Agrobacterium strain GV2260 (12) to form strains SDM320 and SDM321, respectively.

Plant Tissue Culture. The method for cocultivation of tobaccoprotoplasts (Nicotiana tabacum Petit Havana line SR1),theoriginof tobaccoplantline T(originallyreferredto asline 104),and planttissue culture mediawere described previously(6).

PCR,InversePCR,andSequence Analysis. Plant DNA for PCRanalysis wasisolated accordingto Lassneretal. (13). PCR was performed in a Perkin-Elmer thermocycler 480 usingastandardprotocolof 30cycles: 1min, 95°C denatur-ation; 1 min, 57°C annealing; 2

min,

72°C

elongation.

The

Abbreviations: Kmr/Hmr, kanamycin/hygromycin resistance; NPTII, neomycin phosphotransferase II; DSBR: double-strand breakrepair; T-DNA, transferred DNA; HPT, hygromycin

phos-photransferase; NOS, nopaline synthase. tTowhomreprintrequestsshould beaddressed.

reaction mixture contained 0.1 unit Taq DNA polymerase (HTBiotechnology,Cambridge, U.K.), 100

,uM

dNTPs (Am-ersham), and 25 pmol of each primer.

InversePCR was performed as described (14) except that restriction endonucleases Hpa II and Sac II and primers 7 and 8 were used (see Fig. 4A). Amplification occurred in a 35-cycle reaction and annealing at 58°C. Sequence analysis wasperformed using the Sequenase 2.0 DNA sequencing kit (UnitedStates Biochemical).

DNA Isolation and Southern Analysis. Isolation of plant DNAand Southern analysis were performed essentially as described (6). DNA was blotted onto a Hybond N+ mem-brane(Amersham) that was used according to the manufac-turer'srecommendations. Hybridization of Hybond N+ was performed in flasks in a Hybaid oven at 65°C. DNA probes, labeled with [a-32P]dCTP (specific activity: 0.7-2.0 x 109

dpm/,ug),

were obtained using the mixed primer method (Boehringer Mannheim).

RESULTS

Homologous Recombination Between Target Locus and Re-pair Construct pSDM321. Cocultivation of 3.6 x 107 proto-plastsof plant line T, which is hemizygous for the target locus depicted in Fig. 1, with Agrobacterium strain SDM321

re-sulted in 109kanamycin-resistantcalli.Thisis a number 6000-to7000-fold lower than that obtained inparallel cocultivation experiments withcontrol strain SDM320(containsa T-DNA with an intact Kmr gene, Fig. 1), which provided an indication for the total numberof transformedsurvivors in thetargeting experiments, =7.0 x 105. The 109kanamycin-resistant calli werescreened by PCRanalysiswithprimerset 1and 10 for the occurrence of homologous recombination. With this primercombination a 1301-bp fragment isamplifiedwhenan

intact recombinant Kmrgene is present, whereasthe pres-enceofaputativecontaminatingcontrolconstructwill result inamplification of a 1438-bp fragment (Fig. 1). In 1ofthe109 kanamycin-resistant calli transformed with construct pSDM321 arecombination productwasdetected.

Regener-ation of shoots from thisrecombinant callus resulted in plant line A.

The Corrected Kmr Gene in Line A Is Not Present at the Target Locus. ForSouthern analysis genomic DNA was first digested with EcoRI and Bcl I and then hybridized with the NPTII probe(Fig. 2). The target locus in plant line T showed a NPTII hybridizing fragment of 0.96 kbp (Fig. 3). An additional 2.0-kbp NPTIIhybridizing EcoRI/Bcl I fragment, indicative of the presence of the recombinantKmr gene, was detected in plantline A and confirmed our conclusions from the PCR analysis (see above). The presence of a 1.7-kbp EcoRI fragment indicated the insertion of one or more unalteredrepair T-DNAs in this line (Figs. 2 and 3).

SubsequentdigestionwithHindIlI andhybridization with the NPTII probe revealed that the 2.5-kbp fragment of the target locus wasleft unchanged in plant line A (Fig. 3). In case ofatargetingevent(T.E.inFig.2)thisfragmentshould have been convertedinto a3.6-kbp fragment. Instead, two extra NPTII hybridizing Hindlll fragments were observed, indi-catingtheinsertion ofextraT-DNAs atchromosomal posi-tions other than the target locus. One of the HindIII frag-ments in line A was 2.1 kbp insize, which is indicative of the presenceof an intact copy of repair construct pSDM321(Fig. 2). Thesubstoichiometryof the extra bands can be explained by thefact that the 2.5-kbp HindIII fragment from the target locus carries two NPTII hybridizing regions, whereas the extrafragments contain only one copy of this region. Hy-bridization ofHindIII-digested DNA with the HPT probe indicatedthat also thejunctions between theartificialtarget locus andplantchromosomalDNA, inFigs.2and 3referred to asJJ andJ2,hadbeen leftunchangedand confirmed that two extra T-DNAcopieshad beeninsertedinto the genome of line A(Fig. 3).

The exactnumber of integration sites(including the target locus) was determined bydigestion with Xho I(for which no cleavagesitesarepresent in the T-DNAs used) and hybrid-ization with the NPTII probe. The target line (T) and the targeted line (R) isolated in our previous experiments (6) showedone NPTII hybridizing fragmentof -20kbp

repre-LB PS X RB i I I

aux-2I

5SHPT3'NOSW3'OC/NPTlII/5'NOSI

pSDM 320

10

1

-~1438

X E A

5'A

aux-2

s~~~~

I

pSDM 321

NTI

10 4

A

1099

1301 2279 or

1195

NTI 1

FIG. 1. T-DNA constructspSDM320andpSDM321,theartificialtarget locus inplantlineT,and theprimercombinations used for PCR analysis.Thesizes of thefragments expectedtobeamplifiedareindicated inbp.Thearrowin aux-2 indicates the direction oftranscription.

p35S, promoterregionof the cauliflower mosaic virus 35S transcript;HPTorNPTII,region encoding hygromycinorneomycin

phospho-transferase,respectively; 5'NOSor3'NOS,promoterortranscriptionterminationareaof thenopaline synthasegene;3'OCS, transcription

Eco

RI+BcI I

Hindl l

2.1 pSDM 321 AUX-2Probe -J1orJ2--_H- ---2.5----H Jl orJ2-T: _.... ---21.7> 0.96>, T. E.: _-.--Jl orJ2. _0 H-* -- -3.6 - - -0H _-Jf-i orJ2 -_-E2.0 0.96 E -_- BB - - -L_ I nt HPT NPTII NPTII HPT

FIG. 2. Schematicrepresentation of a genetargetingevent(T.E.) after transformation ofprotoplasts ofplant line T with construct pSDM321 (see also Fig. 1 and legend). The predicted sizes of restriction fragments are indicated in kbp and the direction of transcriptionof the aux-2 geneand the 3' deleted or correctedKmr geneisindicated byan arrow.TheHPThybridizing junction frag-mentsbetween the artificial target locus and plant genomic DNA (thick lines) are referred to asJJ and J2. E, EcoRI;B, BclI; H, HindIll.

senting the original target locus and the targeted locus, respectively. Both bands run at approximately the same

location as adifference of1.1 kbp in fragments of this large size isnot resolvedin thegel system used. Inplant line A

additional hybridizing fragments were detected (Fig. 3),

which indicated that thetwo extraT-DNAshad been inserted

atseparate locations eitheronthe samechromosome oron

different chromosomes. Finally, a blot containing HindIII-digested DNA of line A was hybridized with the AUX-2 probe. The 2.1-kbp fragment confirmed thepresence ofan

unalteredcopyof the repair T-DNA (Figs. 2 and 3) and an extraAUX-2hybridizingfragment of -3.7 kbp suggested the integration of one additional T-DNA segment containing aux-2 sequences (Fig. 3).

Thus, homologous recombination between the incoming

T-DNAand thetargetlocus didoccurin line A, but thetarget

locuswasleftunchanged. Apparently, itwasnotthe defec-tive Kmr gene at the target locus that was corrected after recombination but rather the defectivegeneof theincoming repair construct. The corrected construct was inserted at

anotherchromosomallocation.Thefact that thetargetlocus remained unaltered indicated that recombination occurred viaanonreciprocalgeneconversion-likeprocess.Besides the

targetlocus and the correctedKmr T-DNA copy,lineA was

foundtocontainoneunchangedcopyof therepairconstruct

pSDM321 at adifferent chromosomal location andanextra

insertcontaining aux-2 sequences.

Sequence Analysis of the Recombination Product. The 1301-bp recombination product obtained from plant lineAby PCRamplification with primer combination 1 and 10 was

cloned andsequenced.Nodeletions, insertions,orbase-pair mutationsweredetected,indicatingthathomologous

recom-bination had resulted in theperfectrestoration of the defec-tive Kmrgene onthe incomingDNA.

Toanalyze the upstream regionof the corrected T-DNA weperformed inverse PCR as depicted inFig. 4A. Inverse

PCRonDNAfromplantline A showedamplificationofone

fragment of -1 kbp (IPCRA), whereas no amplification products were found with DNAfrom line T orline R (not shown). Fragment IPCRAwascloned and sequenced (Fig.

±20 > nt T R B A

> ..

w NPTII 2.1 o nt T B A Il--AUX-2

FIG. 3. Southern analysis of plant line A. Restriction endonu-cleases used fordigestion and probes used for hybridization are indicated above and below theblots,respectively. Sizes ofexpected fragments (see Fig. 2) are indicated in kbp. The±indicates that the size of a fragment was estimated using a DNA size marker as reference. nt, Wild-type tobacco; T, original target plant line; R, targeted plant line from previous experiments (6) was used as reference for a targeting event (T.E. in Fig. 2); +, plant line transformed with control constructpSDM300; A,recombinant plant line A; B, recombinantplantline that will bereported elsewhere.

4B). The entire 5' region comprising the nopaline synthase (nos) gene promoter is present upstream of the restored T-DNAcopy,lacking only the first4bp from the nick site of theright borderrepeat.Thissequencecontinues withashort (5bp) invertedrepetition of the end of thenosgenepromoter

(arrow 3) followed by a 15-bp DNA segment of unknown origin. Next comes aperfect43-bpinvertedrepetitionof the right end of thenosgenepromoter up totheBcl Irestriction site (arrow1).Theinvertedrepeat structureis reminiscentof theinvertedrepeat structure attheoriginaltargetlocus. Its

presence suggests that in addition to the 5' end of the resistance gene partof the inverted repeat structure of the

target locus was copied to the incoming repair T-DNA. Remarkably, the 43-bp repetition is followed by a short

stretchofmultiple cloning sites and the 3' noncodingregion of the aux-2genefrom theHindIII sitetotheHpaIIsite[base pairs 3390-4076, according to Barker et al.

(9)].

This se-quence(including the 43-bprepetition)isidenticaltothatof

theright border end ofrepairconstructpSDM321,suggesting

thatthispart of theincoming constructincluding theaux-2

gene wasinvertedduringconversion of the repairconstruct

bythetargetlocus.

Thesesequencedataarecompletelyin accordance with the Southernblot data. The sizeof the NPTIIhybridizing HindIII fragment from the correctedconstruct,whichwasestimated

R T A + I J R nt T A 3.7> _ .... Probes: 2.5,'. 2.1 0 nt R T A . . HPT NPTII NPTII

Xho

I

Hind

III

A 5'A H2 FS///,/NNP-T-I5N0,IS, ( (3)-(2) B H2 .0- -..-8 7 digestion S ligation S digestion 7 8 PCR Z

~~~~~~~(1)

79. . BcII . . . 141 LineA: TTCTCCGCTCATGATCAGATTGTCGTTTCCCGCCTTCAGTTTAAACTATCAGTGT-AACACTAT-1111111111111111111111111111111111111111111111111111111I

T-DNA:ttctccgctcatgatcagattgtcgtttcccgccttcagtttaaactatcagtgtttgacaggat

R.B. (2) (1) 142 . 197 LineA: -AAGGAGCGCTATA-ACACTGATAGTTTAAACTGAAGGCGGGAAACGACAATCTGATC-198 HindIIl HpaII904 LineA: -CTCTAGAGTCGAGCTCGCGAAAGCTT--AUX-2-AACCGG C 13 24---0 Hm 3 ±+3.7< H3

10000

_ 0 2 1

12

NPTII AUX-2

FIG. 4. Sequence analysis of theupstreamregion of the corrected T-DNA insert in lineA.(A)Depiction ofthe inversePCRmethod.Primers 7and 8arepositioned within the 5' region deleted from the repairconstructpSDM321. The position of this region is indicated.H2, HpaII;S, Sac II.(B) Sequence of the inverse PCRfragmentfromplantlineAisaligned with thesequenceof the right border end of T-DNAconstruct

pSDM104 (6) that ispresent atthetargetlocus(T-DNA). The first eight bases of the 25-bp right borderrepeat(R.B.)areunderlined.Base

numbering 79-904 is from the first base of primer 7. Thearrowsnumbered 1-3 indicate the positions ofrepeatsin thesequence.AUX-2 between

the HindIII and the Hpa II sites indicates thepresenceof the 3' noncoding region of theaux-2genefrom position3390toposition 4076 (9). (C) Structure of the correctedincoming repair T-DNAasdeduced fromtheSouthern blot andsequencedata. Sizes of the HindIII (H3) fragments

hybridizingtothe NPTIIorAUX-2probeareindicated in kbp. H2,HpaII.

tobe 2.4kbp from the Southern analysis,was nowcalculated tobeexactly 2418 bp (Fig. 4C). Moreover, the finding that

aux-2sequencesarepresent upstreamof the corrected con-struct does explain the detection of one additional band

hybridizingtothe AUX-2probe in the Southern analysis. The factthatthe length of the hybridizing fragment is -3.7 kbp

andnot2.1kbp (Fig. 2) indicates that the 5'partof the aux-2 fragment is deleted.

LinkageAnalysis of theT-DNAInsertionsinLineA.Seeds

obtained after selfing ofplant line A were germinated on

mediumcontainingkanamycinorhygromycin. Resistanceto

kanamycin segregated at a3:1 ratio (Table 1), as was

ex-pected from the presence of one recombinant Kmr gene.

Hygromycin resistance(Hmr) showedasegregation ratio that

suggested the presence of two unlinked loci. Molecular analysis oflineA, however, clearly predicted threeseparate

T-DNAinsertion locicontainingaHmrgene:thetargetlocus,

the corrected incomingconstruct, andoneadditionalrepair construct.

In a more detailed study ofthe linkage between these

insertions weperformed PCR analysis onDNAsof21

kan-amycin-resistant seedlings using different primer combina-tions,eachspecificforaT-DNA insertionlocus(Fig. 1).With

primercombination 1 and10thecorrected Kmrgene(1301-bp

Table 1. Analysisofprogenyfromaselfing ofline A

Selection Res./sens.* Ratiot X2

Kanamycin 242:65 3:1 2.39

Hygromycin 286:17 15:1 0.21

*Ratioof resistanttosensitiveseedlings.

tThedata obtainedweretested forgoodnessof fit with thepresented ratio in the x2testandwerefoundtobesignificant(P>0.1 and P >0.6, respectively).

fragment)wasdetected in each of the 21seedlings(Table2). The 1195-bp fragment that is specifically amplified with primer combination 1 and NTI from the target locus was obtainedin 18ofthe 21 Kmrseedlings (genotypes I andII, Table 2). The observed ratio of6 PCR positive to 1 PCR negative seedlings (Table2) suggests that the assortmentof the targetlocus and the corrected T-DNA is notcompletely independent, which is in accordance with the segregation dataofHmr.Thus,the targetlocus and thecorrectedT-DNA

seem tobe presentonthe same chromosome. However, in this case the twolocimustbeseparated byalarge distance

as segregationis detectable in 3 of the 21 Kmrseedlings. PCRanalysis with primer combination A and 4 showed amplification ofa 1099-bp fragment in 15 ofthe 21 tested seedlings, indicating the presence of an unchanged repair

constructin theseplants (genotypesIandIV, Table2).This insert and the recombinant locus showed an independent

assortmenttypical for unlinked loci (3+:1-;significantin the binomialtest atthe5%level).The targetlocus and theextra

insert of the repair T-DNA also showed an independent

assortment (14 ofgenotype I:4 of genotype II, Table 2), suggestingthat there isnolinkagebetween theselocieither.

Table 2. PCRanalysisof the genotype of 21 kanamycin-resistant seedlingsobtainedafterselfingof line A

Genotype Ratio

Primers Locus I II III IV +:

-1+ 10 Recombinant + + + + 21:0

1+ NT Target + + - - 6:1

A+4 Repair(321) + - - + 2.5:1

No. ofseedlingsper

genotype 14 4 2 1

Sincethe target locus andthe corrected construct are not completely linked, one would expect theratioof hygromycin-resistant tohygromycin-sensitive seedlings from a selfingof line A to behigher than 15:1(ratio fortwounlinkedloci).We calculated the expected ratio by incorporating the extra

separationin 1 of 7seedlingsandcomparedthe data in Table 1with thisnewratio (20.8:1) in the

X2

testfor goodnessof fit. The fitof the data was found to besignificant(P> 0.3).

DISCUSSION

Previously we showed that a defective locus in the tobacco genomecanbecorrectedviahomologous recombinationwith aT-DNArepair construct that was introducedvia Agrobac-terium(6). Here wedescribearecombination eventresulting in the opposite. An incoming T-DNA was accurately cor-rected vianonreciprocal homologousrecombinationwith the T-DNAinsert at the target locus. The corrected T-DNAwas

present at the same chromosome but separated by a large distance from the target locus. Most likely the process of homologousrecombinationoccurredpriortoinsertion of the incomingT-DNAinto theplantgenome.

Correction of a targeting vectorby the target locus has been observedin earlygene targeting experimentsin

mam-malian systems. In these experiments homology with the targetlocuswaspresentatbothsides of themutantsequence in the targeting vectorand the recombination products ob-tained fit the predictions of the double-strand break repair (DSBR) model forrecombination (15, 16). In contrast, the recombination event presented in this paper is not easily explained bythe DSBR model since thehomologybetween thetargetlocusand theregion righttothemutant sequence inconstructpSDM321comprises only50bp. Moreover,this small region of homology is separated from the

600-bp

homologousregiontothe left of themutationbyan-2.3-kbp nonhomologous sequence (the aux-2 gene, Fig. 1). Data obtainedmore recently in mammalian systems suggest that sometimes sequences beyond the region of homology are

copied from thetarget locusto the

incoming

construct(17, 18). An alternative mechanism has beenproposedto

explain

this recombination process. The mechanism resembles DSBRin that therecombinationisinitiatedbyinvasionofone

ofthe3' OH strands of the acceptor DNA molecule into the homologous donor duplex.

Subsequently,

the

invading

strandis elongated using the complementary chromosomal strand as template (17, 18). In view of the structure ofthe corrected T-DNAinsertand the

homology

distribution

be-tweenrepair constructpSDM321 and the target

locus,

a 3' OH strand invasion and

elongation

model could

provide

a

plausible explanation for the initiation of

repair

of the pSDM321Tregion bythe target locus.

However,

the T-DNA transferintermediateisassumedtobea

single-stranded

linear DNAmoleculewith its 3' OH endatthe left border site

(19),

and thus 3' OH

elongation

toward the

right

border sitecan occuronly during or after

synthesis

ofthe second T-DNA strand. Theinversion of the

right

endofthe

repair

T-DNA containing aux-2 sequences could then be duetothe inter-actionof the50-bpsequence atthe

right

endofthe Tstrand with the inversely orientated homologous sequence at the targetlocusduringtherecombinationprocess.

Accordingtothis model the inverted

repetition

upstreamof the corrected T-DNA represents the

junction

between the invertedlyinserted T-DNAsatthe target locus in

plant

line T. MultipleT-DNAinserts presentat onelocusareoften found

tobeorganizedashead-to-head

(inverted)

concatemers asin target line T (20-22). The DNA sequences between such

invertedlylinked T-DNAs havenotyetbeendeterminedand thuswedonotknowbywhich processes these repeatsarise. Our sequence data suggest that "filler" DNAs may be presentbetweeninvertedly repeatedT-DNAcopiesand that thesestructuresdonotarise viaplainend-to-endligation.

Our

findings clearly

have

implications

for future gene targeting experimentsinplants.Inaccordancewith

mamma-lian systems,experimentswill havetobe

designed

in sucha

way that genetargetingcanbe proven anddistinguished from correctionof the

incoming

construct.

We thankProf. M.vanMontaguforgenerouslyprovidingstrain GV2260,John Clement forprovidinga subclone ofpTiAch5 con-tainingthe aux-2 gene,Robbert Rottier for technicalassistance,and Peter Hock for skillful drawing. In addition, we thank Nicholas Harphamforhelpfulcommentsonthe manuscript.Thisworkwas

supported byTheNetherlands FoundationforBiologicalResearch (BION)withfinancial aid from The NetherlandsOrganizationfor Scientific Research(NWO)aswellasbyThe Netherlands Founda-tion for Chemical Research (SON) with fmancial aid from The NetherlandsTechnologyFoundation(STW).

1. Roth,D. B.&Wilson,J. H.(1988)in GeneticRecombination,

eds. Kucherlapati, R. &Smith, G. R. (Am. Soc. Microbiol., Washington, DC),pp. 621-653.

2. Mayerhofer,R.,Koncz-Kalman, Z.,Nawrath,C., Bakkeren, G.,Crameri, A., Angelis, K.,Redei,G.P.,Schell,J.,Hohn,B. &Koncz,C. (1991)EMBOJ. 10,697-704.

3. Gheysen, G., Villarroel,R. &vanMontagu,M. (1991)Genes Dev. 5,287-297.

4. Capecchi,M. R.(1989)Science244, 1288-1292.

5. Paszkowski,J., Baur, M.,Bogucki,A. &Potrykus, I.(1988)

EMBOJ. 7, 4021-4026.

6. Offringa, R., de Groot, M.J.A., Haagsman, H.J., Does,

M. P., van den Elzen, P. J. M. & Hooykaas, P.J. J. (1990)

EMBOJ. 9,3077-3084.

7. Sambrook,J.,Fritsch,E. F. &Maniatis,T.(1989)Molecular

Cloning: A Laboratory Manual (Cold Spring Harbor Lab.

Press,Plainview, NY).

8. Yenofsky,R. L.,Fine, M. &Pellow,J. W.(1990)Proc.Natl.

Acad. Sci. USA87,3435-3439.

9. Barker,R. F.,Idler, K.B.,Thompson,D. V. &Kemp,J. D.

(1983)Plant Mol. Biol.2, 335-350.

10. Marsh, J.L., Erfle, M. & Wijkes, E. J. (1984) Gene 32,

481-485.

11. Ditta, G., Stanfield, S., Corbin, D. & Helinski, D. R. (1980)

Proc. Natl.Acad. Sci. USA77, 7347-7351.

12. Deblaere, R.,Bytebier, B.,deGreve,H.,Deboeck,F.,Schell,

J.,vanMontagu,M.&Leemans,J.(1985)Nucleic Acids Res. 13,4777-4788.

13. Lassner,M. W., Peterson,P.&Yoder,J. I.(1989)Plant Mol. Biol.Rep. 7,116-128.

14. Does, M.P., Dekker, B. M. M., de Groot, M.J. A. &

Of-fringa,R. (1991)PlantMol. Biol. 17,151-153.

15. Thomas,K. R.,Folger, K. R. &Capecchi, M. R.(1986)Cell 44, 419-428.

16. Song, K.-Y., Schwartz, F., Maeda, N., Smithies, 0. &

Kucherlapati,R. (1987)Proc.Natl.Acad. Sci. USA84, 6820-6824.

17. Adair, G.M., Naim, R. S., Wilson, J. H., Seidman,M.M., Brotherman,K.A.,MacKinnon, C. &Scheerer,J. B. (1989)

Proc.Natl. Acad. Sci. USA86,4574-4578.

18. Aratani, Y.,Okazaki,R.&Koyama,H.(1992)NucleicAcids Res.20,4795-4801.

19. Stachel,S.E.,Timmerman,B.&

Zambrisky,

P.(1987)EMBO J.6,857-863.

20. Jorgensen, R.,Snyder,C. &Jones,J. D.C.(1987)Mol.Gen. Genet.207,471-477.

21. Jones, J. D.C., Gilbert, D.E., Grady, K. L. &Jorgensen,

R.A.(1987)Mol. Gen. Genet.207,478-485.

22. DeBlock,M.&Debrouwer,D.(1991)Theor.Appl.Genet.82,