Plant Physiol. (1985) 77, 87-940032-0889/85/77/0087/08/$0 1.00/0

Genotypic Variability of Soybean Response to AgrobacteriumStrains Harboring the Ti or Ri Plasmids

Received for publication July 23, 1984

LoWELL D. OWENS* AND DEAN E. CRESSUnited States Department ofAgriculture, Agricultural Research Service, Beltsville, Maryland 20705

ABSTRACT

Twenty four diverse cultivars of soybean (Glycine max [L.] Merrill)and three lines of its annual wild progenitor Glycine soja Sieb and Zucc.were tested for their response to Agrobacterium strains harboring eitherthe Ti (tumor-inducing) plasmid (pTi) from Agrobacterium tumefaciensor the Ri (root-inducing) plasmid (pRi) from Agrobacterium rhizogenesfollowing uniform wounding and inoculation. Based upon gall weight at 8weeks postinfection, three G. max cultivars (Biloxi, Jupiter, and Peking)and one G. soja line, Plant Introduction (PI) 398.693B, were judgedhighly susceptible to A. tumefaciens strain A348 (pTiA6), ten genotypesmoderately susceptible, 11 weakly susceptible, and two nonsusceptible.Of 26 genotypes inoculated with strain R1000 (pRiA4b), only sevenresponded in a clearly susceptible fashion by forming small, fleshy rootsat internodal infection sites. Cotyledons excised from 1- or 3-day oldseedlings of Peking and Biloxi cultivars also formed galls when infectedin vitro with agrobacteria carrying either the Ti or Ri plasmid. Tumorlines established from cotyledon and stem galls induced byA. tumefacinxsA348 (pTiA6) exhibited the T-DNA borne traits of phytohormone-independent growth and octopine synthesis. Additionally, DNA isolatedfrom cultured tumors hybridized with labeled T-DNA probe.

The Ti (tumor-inducing) and Ri (root-inducing) plasmids ofthe pathogenic bacteria Agrobacterium tumefaciens and Agro-bacterium rhizogenes, respectively, are known to cause the trans-fer ofDNA into plant cells during the course of infection (4, 7).In the host cell a part ofthe plasmid DNA, the T-DNA, becomesintegrated into the chromosomes (4, 7, 24). In certain instances,whole plants may be regenerated from cells transformed by T-DNA (2, 7, 16, 18, 22), and genes normally carried on T-DNAor, alternatively, foreign genes inserted into T-DNA have beenexpressed in various tissues of the regenerated plants (7, 16, 18).Furthermore, genes introduced into these plants by the Ti plas-mid were inherited through seed in a Mendelian fashion (16,18).The above research has demonstrated that the Ti, and probably

the Ri, plasmids potentially can be used as gene vectors for thosehigher plants that are hosts to the respective bacterium. Becauseof the economic importance of soybean worldwide, we investi-gated whether soybean is a host to agrobacteria carrying eitherof these plasmids. Reviews of extensive data on the host rangeofA. tumefaciens, the crown-gall inducing bacterium (9), and A.rhizogenes the 'hairy root' inducing bacterium (11), were ambig-uous as to whether soybean is susceptible to either Agrobacteriumspecies. Becasue of this uncertainty, we undertook a study of 24genetically diverse cultivars of Glycine max and three lines of itsannual wild progenitor G. soja in relation to Agrobacteriumpathogenesis.

We found a wide range in susceptibility to agrobacteria har-boring Ti or Ri plasmids and identified highly susceptible geno-types of Glycine max and G. soja to strains of either plasmidtype. While this work was in progress, Pedersen et al. (20)reported the induction of crown gall on three cultivars of G.max, including Mandarin tested in our experiments, by A. tu-mefaciens. Our results both confirm and extend their findings.

MATERIALS AND METHODS

Plants. The 24 Glycine max (L.) Merr. cultivars and three G.soja Sieb. and Zucc. lines (indicated by PI numbers) used in theexperiments are listed in Figure 1. Seeds of these genotypes,obtained from Dr. Perry Cregan, were germinated in moist papertowels, and selected uniform seedlings were planted in Ver-miculite (21 x 21 cm pots, seven seedlings per pot) wetted with1 L ofa complete nutrient solution. Two pots were used for eachgenotype. In the greenhouse experiment (Fig. la), three plants ineach pot were infected with A. tumefaciens strain A348 (pTiA6),three with strain R1000 (pRiA4b), and one with A. tumefaciensA136, a control strain lacking a Ti or Ri plasmid. In the subse-quent growth chamber experiment (Fig. lb), only strains A348and A136 were tested. The greenhouse experiment was con-ducted in late summer with temperatures ranging to 32°C to35°C maxima during the day and about 22°C minimum at night,light intensities up to 1750 jsmol m-2 s-' PAR, and daylengthsdiminishing from about 15 to 12 h. Growth chamber conditionswere 25 ± 1.5°C, and 18-h photoperiod of cool white fluorescentlight (250 ALmol mr2 s-' PAR). Humidity was uncontrolled inboth environments and ranged from 60% to over 95%.

Bacteria. All Agrobacterium strains used are derivatives of A.tumefaciens C58. Strain A136 (C58 cured of its Ti plasmid) wasobtained from Dr. Andrew Binns. Strains A348 (A136 harboringpTiA6 from A. tumefaciens A6NC) (14) and R1000 (A 136harboring pRiA4b from A. rhizogenes A4T) were obtained fromDr. Milton P. Gordon.

Stock cultures were maintained on YEB agar medium (yeastextract, 1 g/l; beef extract, 5 g/l; peptone, 5 g/l; MgSO4. 7H20,0.5 g/l; sucrose, 5 g/l; and agar, 15 g/l). Strain A348 wasoccasionally streaked on minimal medium containing octopineas the sole N source (15).

Inoculations. Bacteria were grown on AB minimal agar me-dium (6) for 2 to 3 d, scraped off, washed, and suspended in0.9% NaCl to a concentration of 5 x 100' cells/ml. Soybeanplants at the third trifoliate stage of growth (2-3 weeks old) werewounded between nodes two and three (internode about 3.5 cm),and 10 gl of the bacterial suspension was spread over the woundarea. Uniform wounding was achieved by use ofa wounding toolconsisting of a small, flat piece of wood through which 12hypodermic needles (26 gauge) were arranged along a 1-cm line.The needle tips, protruding about 1 mm through the wood, weregently pressed into the stem to the maximum extent possible.

For cotyledon experiments, seeds were surface-sterilized by87

www.plantphysiol.orgon April 16, 2020 - Published by Downloaded from Copyright © 1985 American Society of Plant Biologists. All rights reserved.

Plant Physiol. Vol. 77, 1985

00 FISKEBY V.MAPLE ARROW

O BEESON'MANDARIN'

II MANCHURICHLAND'SENECA

III KANRICHVERDEWILLIAMS

IV CLARKCLARK rhjEMERALDKENTPEKINGSOOTY

V PI 342.434PI 407.287ESSEXFORREST

VI LEE

VII PI 378.89RANSOM

VIII BILOXI3COB

IX JUPITER

3/86/E0/58/88/BB/B0/64/58/65/8

5/55/5B/B8/B5/8B/B

B/B4/85/58/BB/8

6/84/5

B/I.5/B5/5

-..3--

i.no dith4

.3 i :

* I;.|.

GREENHOUSEEXPERIMEN

ia

I : : _ _

5 10 15 20 Du6 70 au

MANDARIN 2 3/83 GROWTH CHAMBERIV MITCHELL2 5/B EXPERIMENT

PEKINS 5/8 ibibVII RANSOM 5/8VIII BILOXI 4/8

COBB 4/BIX JUPITER 4/11

.~~~~~~~~~~~~f AO..

Gall fresn weight - mg

FIG. 1. Response of soybean genotypes to infection with A. tumefa-ciens A348 (pTiA6). Bars represent the mean gall fresh weight ± SE perinoculated plant that survived the 8-week experimental period. Thenumber of inoculated plants that formed galls is listed following thegenotype name as a fraction of the inoculated plants that survived theexperimental period. Galls were excised and weighed at 8 weeks postin-fection. The genotype maturity group classification, based upon theability of plants to mature productively at a given latitude, is indicatedas a roman numeral preceding the genotype name or number. G. max

cultivars are listed by name, and G. soja lines by Plant Introduction (PI)number. Dashed vertical lines delimit susceptibility classes: weak, lessthan 5 mg; moderate, 5 to 15 mg; and high, more than 15 mg gall freshweight. Footnote 1, Genotypes that set seed and senesced considerablybefore end of experiment. Footnote 2, Genotypes that flowered under18-h photoperiod (b). Flowers were removed from these genotypes toextend the period of vegetative growth and delay senescence. Footnote3, Scale beyond break is doubled for gall fresh weight. Footnote 4, Gallswere removed at 5 weeks for culturing in vitro.

immersing in commercial bleach diluted 1:5 with water contain-ing a few drops of detergent and sonicating for 5 min in anultrasonic cleaner. After five rinses in water, seeds were germi-nated on 0.8% water agar containing Benomyl (5 ,ug/ml activeingredient, Science Products, Chicago). At specified times, thecotyledons were excised and wounded on their flat surface bymaking two 1-cm scratches (about 0.5 mm deep) parallel to thecentral vein, and each was inoculated with 2.5 ,l of bacterialsuspension (10"/ml prepared as above). After incubating 3 d onwater agar (0.8%), each cotyledon was transferred to 3 ml (24-well Costar dish) of water agar (0.8%) containing Benomyl (5,gg/ml active ingredient), Vancomycin (Sigma) (200 /g/ml), andcarbenicillin (Beecham) (200 ,ug/ml) or, in the case of Biloxi, toB5 agar (0.8%) medium (13) supplemented with 0.5 g/l casaminoacids (B5 cas) containing the above antibiotics. Cotyledons werecultured at 27°C with a 16-h photoperiod (30 ,umol m 2 s-', coolwhite fluorescent).

In Vitro Tumor Culture. Tumor cultures ofgreenhouse-growncrown galls were established by the procedures of White et al.(24) except that B5 cas medium minus phytohormones (B5 cas-hor) containing the above antibiotics was used throughout.Cotyledon tumors were removed when 1.5 to 2.0mm in diameterand cultured initially on 50 ,l of B5 cas-hor with the above

antibiotics.Octopine Assay. About 75 mg of tumor callus was incubated

24 h or, for weak octopine-producing lines, 88 h in 50 ,l of B5cas-hor medium supplemented with L-arginine (100 mM). Tissueextracts were subjected to electrophoresis as described ( 19).

Southern Blot Hybridization Analysis of T-DNA. DNA wasextracted and purified as described (12) from axenic culturedtumor line Cr1O derived from G. max cv Clark rj1 crown gallsincited by A. tumefaciens A348, or from normal suspensioncultures of G. max cv Peking. Plasmid pTiA6 was purified byCsCl-ethidium bromide equilibrium centrifugation followingchemical lysis of strain A348 (5). Ten jsg of plant DNA or 3 ngofpTiA6 (equivalent to two copies per diploid soybean genome)were digested to completion with Bam HI (Bethesda ResearchLabs) and electrophoresed in horizontal 1.0% agarose slab gelsin TBE buffer (89 mm Tris, 89 mM boric acid, 2 mm EDTA, pH8.3) at 2 v/cm for 18 h. DNA was transferred to nitrocellulose(BA 85, Schleicher and Schuell) by blotting (21). Probe DNA,plasmid pTRC 188 (1), a cosmid partial-digest clone containingEcoRl fragments 23 through 20 (homologous to those in pTiA6)of pTi B6S3, was labeled to specific activity of 8 x 107 cpm/4ogby nick translation (17) with [a-3p]dCTP (New England Nu-clear) and a commercial kit (Amersham). Blots were hybridizedin 5x SSPE (lx SSPE is 0.18 M NaCl, 10 mm sodium phosphate1 mM EDTA, pH 7.7), 50% formamide, lx Denhardt's reagent,and 100 gg/ml denatured salmon sperm DNA at 42'C for,70 h.Probe concentration was 25 ug/ml. After hybridization, blotswere washed four times at room temperature with 250 ml -2xSSPE, 0.1% SDS and four times with 0.1x SSPE, 0.1% SDS.Autoradiography was carried out on Kodak RP X-omat film at-80°C with an intensifying screen.

RESULTSResponse to Stem Inoculation with A. tumefaciens. In two

experiments, a total of 24 cultivars of the domesticated soybeanG. max and three lines of its wild progenitor G. soja were testedfor their response to Agrobacterium strains A348 (pTiA6) andR1000 (pRiA4b) following uniform wounding and inoculation.With the most susceptible genotypes, galls began to appear inthe intemodal wounds at about 3 weeks postinoculation. Onecriterion of genotypic response to strain A348, crown gall freshweight at 8 weeks postinfection, is presented in Figure 1. Themean ± SE of gall weights in Figure 1 were calculated on thebasis of number of plants inoculated, rather than on the numberof inoculated plants that formed galls. This basis was chosen sothat the data would reflect not only the growth rate of galls, onceinitiated, but also the frequency of gall formation. The actualfrequency with which plants of each genotype responded toinoculation by forming galls is listed after each genotype namein Figure 1 as a fraction of the total number of plants inoculatedthat survived the experimental period. In order to classify theresponses, we have arbitrarily defined four classes based uponthe mean gall fresh weight per plant: highly susceptible (morethan 15 mg); moderately susceptible (5-15 mg); weakly suscep-tible (less than 5 mg); and nonsusceptible (response not differentfrom inoculation with control strain A 136). The four classes arepictorially represented in Figure 2. According to this classifica-tion scheme, three G. max cultivars, Biloxi, Jupiter, and Peking,and one G. soja line, PI 378.693B, werejudged highly susceptible,10 genotypes moderately susceptible, 11 weakly susceptible, andtwo nonsusceptible, taking into account both experiments (Fig.1, a and b). No galls were ever obtained from inoculation withthe Ti-plasmidless strain A136.

There was a notable genotype x environment interaction withcv Biloxi; it was highly responsive to A. tumefaciens infection inthe greenhouse (Fig. la), but only weakly so in the growthchamber (Fig. lb). Additionally, the frequency of crown gall

)- - &X

88 OWENS AND CRESS

lU 10 ZU t;> UV tu Ov

www.plantphysiol.orgon April 16, 2020 - Published by Downloaded from Copyright © 1985 American Society of Plant Biologists. All rights reserved.

SOYBEAN RESPONSE TO AGROBACTERIUM

RESPONSEHIGH MODERATE WEAK NONE

E0

BILOXI WILLIAMS VERDE CONTROLSx x x x

A348 A348 A 348 A136FIG. 2. Response of three cultivars to infection with A. tumefaciens A348 (pTiA6) or with the control strain A136 (lacking Ti plasmid) illustrating

high (Biloxi), moderate (Williams), and weak (Verde) susceptibility classes and the nonsusceptible response (X A136).

formation on the six cultivars grown in both environments was

somewhat lower in the growth chamber, 72 + 7% compared to91 ± 4% for the same six genotypes in the greenhouse (calculatedfrom data in Fig. 1).Two genotypic differences, in addition to gall weight, were

noted; the number of crown galls per wound area, and growthof galls beyond the 8-week experimental period. Cultivars Pekingand Biloxi tended to form masses of small galls over the entirewound area caused by the 12 needle punctures (Fig. 2), whileJupiter, Williams, and Verde (Fig. 2), for example, formed onlya few. These differences in the incidence of gall formation perneedle puncture did not always correlate with gall weight perplant at 8 weeks, since Jupiter and Peking were equal in thisrespect (Fig. 1). Additionally, galls on Biloxi and Peking ceasedgrowing at about 8 weeks and commenced senescing. In contrast,galls on Clark rib, for example, continued to enlarge considerablyin succeeding weeks.We investigated whether the genotypic differences described

above might be an artifact ofthe mode of infection. For example,could differences in stem pubescence have influenced the amountof liquid inoculum retained on the wound? In a growth chamberexperiment, we inoculated stem wounds on Jupiter, Cobb, Biloxi,and Ransom cultivars with approximately the same number ofstrain A348 bacteria carried either in a mucoid scraping from an

agar plate or in a liquid suspension. We found that the mucoidinoculum increased the overall frequency with which inoculatedplants responded by forming galls, from 50 to 70% compared to

the liquid suspension, but that the gall fresh weights per plantthat formed galls were not significantly different for the twoinocula (data not shown). From these results, together with thehigh frequencies of gall response generally obtained, we con-cluded that the differences in gall weights observed among gen-otypes were not artifacts of the mode of inoculation but due togenotypic variation.

Response to Stem Infection with Agrobacteria Carrying the RiPlasmid. All of the 26 genotypes listed in Figure la were infectedwith strain R1000 (pRiA4b). Of these 26 genotypes, seven re-sponded in a way recognizably characteristic of hairy root galls,i.e. they displayed small roots emerging from internodal wounds(Table I). The roots were fleshy in appearance (Fig. 3) as is typicalof hairy roots caused by A. rhizogenes on other species. Biloxigave the most prolific response, but the longest roots wereobtained with G. soja line PI 407.287 and G. max cv Peking(Table I).In Vitro Culture of Tumors. Tumor lines were established in

vitro from crown galls incited by A. tumefaciens A348 on green-house-grown plants of G. max cv Clark ii' and G. soja PI378.693B. These tumor lines produced octopine (Fig. 4, lanes 1and 2) and grew on medium free of phytohormones with adoubling time (7 d) equal to nontransformed soybean callus onhormone-containing medium. Attempts to culture hairy rootsinduced by strain RI000 have thus far not met with success.Tumor Induction In Vitro. Peking cotyledons excised from

seeds that had germinated for 1 d were maximally responsive to

89

www.plantphysiol.orgon April 16, 2020 - Published by Downloaded from Copyright © 1985 American Society of Plant Biologists. All rights reserved.

Plant Physiol. Vol. 77, 1985

rl

I

I CONTROL L 1 CONT+:4fX RI00 X A136 X R1000 x A41 J..

X1... i. sA.. PEKIN -

FIG. 3. Response of Biloxi and Peking to Ri-plasmid strain RO000 infection at 9 weeks postinfection.

Table I. Response ofSoybean Genotypes to Infection withAgrobacterium Strain RJOOO(pRiA4b)

One or more replicates of the following genotypes formed small gallswithout roots: Emerald, Kanrich, Lee, Maple Arrow, Manchu, Kent,and Sooty. The remainder of the 26 genotypes listed in Figure la did notrespond to RIOOO infection.

Hairy RootResponse Hairy LengthGenotype Frequency' Roots"

Mean Rangemm

Biloxi 3/6 12.0 2.9 14Cobb j/6 1.0 3.0 3Forrest 1/6 3.0 2.2 1-5Peking 3/6 2.7 3.8 2-6PI 378.693B 2/6 2.0 1.5 1-2PI 407.287 3/6 1.7 3.0 1-6Ransom 2/6 2.5 1.1 1-2'Number of plants forming hairy roots/number of plants infected."Mean number/plant that formed hairy roots.

tumor induction by the Ti-plasmid strain A348 (Table II). Pu-tative tumor calli appeared in wounds about 12 d after inocula-tion (Fig. 5a) and upon removal at 15 to 25 d grew on hormone-free medium and synthesized octopine (Fig. 4, lane 3). At about17 d postinoculation, roots began to emerge from some of thesetumor calli (Fig. 5b), and by 25 d rooting was observed on 67%of the tumor calli (Table II). Some of these roots exhibitedcallusing when about 1 cm in length. These roots, when excised

and cultured on hormone-free medium, proliferated unorganizedcallus that synthesized octopine (Fig. 4, lane 4) and eventuallyregenerated new roots. The callus lost this rhizogenic capabilityafter several subcultures, however.

In contrast to the above results with the Ti-plasmid strainA348, cotyledons excised from Peking seedlings germinated for1 d were not sensitive to infection by the Ri-plasmid strainsRIOOO. However, those excised from 3-d seedlings were maxi-mally sensitive (Table I). Oddly, only 5% of these wound calliformed roots. Neither calli nor roots have been characterized.

Biloxi cotyledons cultured on water agar without phytohor-mones failed to form tumors in response to infection with strainA348. Nevertheless, when they were cultured on B5 cas mediumcontaining 2,4-D (1 mg/l), call intermixed with roots proliferatedin most of the infected wounds. From these growths, severaltumor lines were extablished that exhibited hormone-independ-ent growth and octopine production (Fig. 4, lane 5).

Cultured Tumors Contain T-DNA. Southern blot hybridizationwas used to assay for the presence of Ti-homologous fragmentsin DNA isolated from cultured soybean tumor line Cr1O (fromClark rjI stem gall incited by strain A348) and from normalPeking soybean cultures. Plant DNA samples were digested withBam HI, fractionated on 1% agarose gels, and transferred tonitrocellulose. The filters were probed with a cosmid clonecontaining Ti DNA spanning the entire T-DNA region normallyfound in tobacco plants transformed with this plasmid (23). Asa control, a two-copy plasmid reconstruction mixture was elec-trophoresed in parallel (Fig. 6, lane c). The autoradiogram of theblot (Fig. 6) displays several bands in the lane containing tumorDNA, but no hybridization to normal soybean DNA. Two bands

90 OWENS AND CRESS

i if

www.plantphysiol.orgon April 16, 2020 - Published by Downloaded from Copyright © 1985 American Society of Plant Biologists. All rights reserved.

SOYBEAN RESPONSE TO AGROBACTERIUM< By:.. __:X_sX -Ad r q%--

ctFIG. 4. Octopine production by soybean tumor

lines cultured in vitro. Fluorograph was made of apaper electrophoretogram on which 2.5 jul of tumorcalli extracts were analyzed. Tumor calli had beenincubated for 24 h or, for weak octopine-producinglines, 88 h (lanes 2 and 4) on medium containing 100mm L-arginine. Lane 1, line Cr1O from Clark dji stemgall; 2, PI 378.693B from stem gall; 3, Peking fromcotyledon tumor, 4, Peking from cotyledon tumorroot; 5, Biloxi from cotyledon tumor, 6 and 7, octo-pine and arginine (I and 5 nmol each, respectively);8, control-normal Peking callus.

1 2 3 4 5 6 7 8

Table II. Effect ofPeking Cotyledon Age on Response to Infection withAgrobacterium Harboring the Ti or Ri Plasmids

Response to infection is recorded as the average number (± SE) percotyledon of putative tumor calli that grew from inoculated wounds.Each value is the average from 10 replicate cotyledons. Data wererecorded at 25 d postinoculation. The percentage of calli that had a rootemerging from them is shown in parenthesis for those treatments thatproduced an appreciable number of calli.

Agrobacterium Cotyledon Agea (d)Strain 1 3 5 9 12

A348 (pTi) 2.0 ± 0.7 (67) 1.1 ± 0.3 (18) 0.1 0.1 0.2A136b 0 0.1 0 0 0RI000 (pRi) 0.2 2.1 ± 0.7 (5) 0.1 0.2 0.1

a Peking seeds were germinated for the indicated days prior to cotyle-don excision and inoculation.

b Control strain that lacks either a Ti or Ri plasmid.in the tumor DNA co-migrated with the contiguous Bam HIfragments 17a and 28 (10), indicating that these are internalfragments wholly contained within the T-DNA. The prominentregion of hybridization to high mol wt DNA band probablyrepresents one or more border fragments containing Ti andsoybean DNA covalently joined. Several additional faint bandsmay reflect abnormal T-DNA insertions or rearrangements. The

intensity ofhybridization ofthe two internal fragments comparedwith the plasmid reconstruction lane suggested that two to fourcopies ofT-DNA were inserted per soybean genome.

DISCUSSION

Soybean, both the domesticated G. max and its wild annualprogenitor G. soja, displays wide genotypic variation in suscep-tibility to infection with either Ti- or Ri-plasmid bearing agro-bacteria. With the Ti-plasmid strain A348, three G max and oneG. soja genotypes were judged highly susceptible, 10 moderatelysusceptible, 11 weakly susceptible, and two nonsusceptible (Fig.1). These classifications were based on gall fresh weight at 8weeks which, in turn, reflected genotypic variability in the inci-dence of inoculated plants that formed galls, the incidence ofgall formation per stem puncture, the growth rate of galls, andthe length of time during which the galls continued to grow.Because of variability among genotypes in these several responseparameters, this single-parameter classification scheme for sus-ceptibility is accurate only in a general sense. Also it should benoted that, compared to crown galls on many other dicots,soybean galls were small and slow-growing even on the mostsusceptible genotypes.

In selection of germplasm for this investigation, we usedmaturity group classification as an indicator ofgenetic variability.While the selection criterion did indeed provide a broad range

91

www.plantphysiol.orgon April 16, 2020 - Published by Downloaded from Copyright © 1985 American Society of Plant Biologists. All rights reserved.

Plant Physiol. Vol. 77, 1985

*S.. ML

W.*... ..

I Fig. 5. Tumor induction on Peking cotyledons ex-cised from 1-d seedlings. Photograph taken 12 d (a) and17 d (b) after wounding and inoculation with A. tume-faciens A348 (pTiA6).

of genotypic variation, it also introduced the possible confound-ing problem of early maturation and senescence of plants inmaturity groups 00, 0, and II, the earlier maturing groups. Thisdid not seem to be the case, however, since similar results wereobtained with four of five cultivars when they were retested inthe growth chamber under LD conditions especially chosen todelay maturation oflower-number maturity groups (Fig. lb). Weconclude, therefore, that crown gall induction and growth werenot related to the maturation characteristics of the genotype, butrather reflected some other genetic trait.

Neither was the weak response observed with Biloxi in thegrowth chamber (Fig. 1 B) related to maturation, since plants inthat experiment matured to about the same degree as in thegreenhouse experiment (Fig. la). The specific environmentalfactor causing this genotype x environment interaction has notbeen identified. Growth chamber conditions differed from thosein the greenhouse in longer photoperiod, much reduced lightintensity, and lower temperature. Pedersen et al. (20) also ob-served a strong genotype x environment interaction in crowngall formation on G. max. In experiments with Mandarin cv, themost susceptible ofthe three cultivars they tested, very few, smallgalls (1-2 mm) were obtained unless they employed a combina-tion of certain controlled environmental conditions and enclo-sure of the inoculation site in a small tube. We did not employinoculation-site enclosures in any of our experiments, but wealso observed that with no protection ofthe wound site Mandarin

responded to A. tumefaciens infection in a weakly susceptiblefashion (Fig. 1).Some soybean genotypes, for example Peking and Biloxi, that

were highly susceptible to the Ti-plasmid strain were also highlyresponsive to the Ri-plasmid strain. Nevertheless, response toone strain was not a reliable indicator of response to the other.G. soja line PI 407.287 responded weakly to A348 (pTi) butstrongly to R1000 (pRi). Conversely, Jupiter, a cultivar highlyresponsive to A348, showed no reaction to R100 0. These differ-ential responses of soybean genotypes to the type of oncogenicplasmid carried by the Agrobacterium strain are striking andprobably reflect the expression of several traits that affect theinteraction of the plant with the bacterium. Included might betraits that affect the initial survival ofagrobacteria in the wounds;the subsequent binding of bacteria to the plant cell wall; thetransfer ofDNA from the bacteria into the host cell; the integra-tion of DNA into the plant chromosome; and the endogenousphytohormone levels in the stem that could enhance or inhibittumor growth and possibly determine the age at which galls beginto senesce. The biochemistry of most of these interactions hasnot been elucidated.Genotypic differences were further observed in the response

of excised cotyledons to infection with strain A348 (pTi). Pekingcotyledons formed wound tumors when cultured on phytohor-mone-free medium, while Biloxi cotyledons required the addi-tion of 2,4-D to the medium for tumor induction. This finding

92 OWENS AND CRESS

www.plantphysiol.orgon April 16, 2020 - Published by Downloaded from Copyright © 1985 American Society of Plant Biologists. All rights reserved.

SOYBEAN RESPONSE TO AGROBACTERIUM

a b c

17a

28

FIG. 6. Hybridization of nick-translated cosmid probe containing T-DNA to Southern blots of BAM Hl-digested soybean tumor line Cr1ODNA (a), normal soybean DNA (b), and pTiA6 DNA (c). The internalT-DNA fragments 17a and 28 (10) are indicated.

is in accord with those of Bouckaert-Urban and Vendrig (3) whoreported that plant growth regulators could, in certain instances,enhance tumor initiation by A. tumefaciens B6 on excised coty-ledons of sunflower.

In our experiments the physiological age ofsoybean cotyledonsexerted great influence on both susceptibility to tumor initiationand tumor morphology (Table II). The rooty morphology ofcotyledon tumors incited by the Ti-plasmid strain A348 wasespecially unexpected, because roots were never observed onstem galls incited by A348 on any of the 27 genotypes tested.Also unexpected was the low incidence (5%) ofrooty morphologyof cotyledon tumors incited by the Ri-plasmid strain Rl000(Table II), since stem galls induced by this strain on Peking andother cultivars did exhibit rooty morphology (Fig. 3). The influ-ence of cotyledon age on tumor initiation and morphologyprobably reflects rapidly occurring changes in phytohormonelevels in cotyledons during germination. Our observation thatcotyledons excised from seeds that were germinated for only 1 dgenerated roots much faster from their basal meristem than didcotyledons excised from 3-d seedlings (unpublished data) maybe further indication of such changes. Different phytohormoneratios in cotyledons of different genotype and physiological agecould account for their differential behavior in regard to bothtumor initiation and subsequent tumor rhizogenesis.We have established more than a dozen tumor lines from galls

induced by A. tumefaciens A348 on plants grown in the green-

house or on cotyledons cultured in vitro. All display the T-DNAborne traits of phytohormone-independent growth and octopinesynthesis. In vitro tumor lines varied in appearance from thedark green, hard calli derived from Peking cotyledon galls; to thelight tan, friable calli from Biloxi and Clark ijI stem galls; to thesegregating mixture of the two types in cultures derived from G.soja PI 378.693B stem galls. Octopine-synthesizing ability wassimilar in dark green and tan friable segregants ofthe PI 378.693Btumor line (data not shown).DNA isolated from a Clark rdi x A348 tumor line contains

several copies of T-DNA as shown in the Southern blot hybrid-ization (Fig. 6). This finding, together with the demonstrationthat the T-DNA borne gene for octopine synthase is expressedin cultured tumor lines, provides incontrovertible evidence thatsoybean cells are capable of being genetically transformed by Ti-plasmid transfer from A. tumefaciens.The results reported here open the way to gene transfer exper-

iments with selected soybean genotypes and the Ti plasmid. Theultimate goal of such experiments, the regeneration of trans-formed plants from transformed cells, poses a serious problem.With normal tissue cultures of several Glycine species, the mostnotable efforts have succeeded only in regenerating small shootsor plantlets that failed to develop further (8, 25). Recently, certainTi plasmid mutants have proven capable of affecting the mor-phogenetic potential of tobacco cells transformed by these mu-tants (14). We have found these tms and tmr mutants to beweakly oncogenic on G. max (unpublished data). Whether theycan also effect morphogenesis is currently under investigation.

Acknowledgments-We thank Dr. Perry B. Cregan for generously providing thesoybean seeds and Drs. Milton P. Gordon and Andrew N. Binns for the Agrobac-terium strains. We also acknowledge the excellent technical assistance of Aria Bush,Barbara Smith, and Denise O'Donnell.

LITERATURE CITED

1. AP RHYS CMJ 1983 Molecular cloning of large fragments of the Ti-plasmid ofAgrobacterium tumefaciens. MS Thesis. University of Maryland-BaltimoreCounty

2. BARTON KA, AN BINNs, AJM MATZKE, M-D CHILTON 1983 Regeneration ofintact tobacco plants containing full length copies of genetically engineeredT-DNA, and transmission ofT-DNA to RI progeny. Cell 32: 1033-1043

3. BOUCKAERT-URBAN A-M, JC VENDRIG 1981 The influence of plant growthregulators on crown-gall initiation on cotyledonary leaves of Helianthusgiganteus L. in vitro. Z. Pflanzenphysiol 103: 75-81

4. CAPLAN A, L HERRERA-ESTRELLA, D INzE, E VAN HAUTE, M VAN MONTAGU,J SCHELL, P ZAMBRYSKI 1983 Introduction of genetic material into plantcells. Science 222: 815-821

5. CASSE F, C BOUCHER, JS JULLIOT, M MICHEL, J DENARE 1979 Identificationand characterization of large plasmids in Rhizobium meliloti using agarosegel electrophoresis. J Gen Microbiol 113: 229-242

6. CHILTON M-D, TC CURRIER, SK FARRAND, AJ BENDICH, MP GORDON, EWNESTER 1974 Agrobacterium tumefaciens DNA and P58 bacteriophage DNAnot detected in crown gall tumors. Proc Natl Acad Sci USA 71: 3672-3676

7. CHILTON M-D, DA TEPFER, A PETIT, C DAVID, F CASSE-DELBART, J TEMPE1982 Agrobacterium rhizogenes inserts T-DNA into the genomes of the hostplant root cells. Nature (Lond) 295: 432-434

8. CHRISTIANSON ML, DA WARNICK, PS CARLSON 1983 A morphogeneticallycompetent soybean suspension culture. Science 222: 632-634

9. DE CLEENE M, J DE LEY 1976 The host range ofcrown gall. Bot Rev 42: 389-486

10. DE Vos G. M DE BEUCKELEER, M VAN MONTAGU, J SCHELL 1981 Restrictionendonuclease mapping of the octopine tumor-inducing plasmid PTiAch5 ofAgrobacterium tumefaciens. Plasmid 6: 249-253

1 1. ELLIOT C 1951 Manual of Bacterial Plant Pathogens, Ed 2 (revised). ChronicaBotanica Co., Waltham, MA, pp 4-12

12. FISCHER RL, RB GOLDBERG 1982 Structure and flanking regions of soybeanseed protein genes. Cell 29: 651-660

13. GAMBORG 0 1970 The effects of amino acid and ammonium on the growth ofplant cells in suspension cultures. Plant Physiol 45: 372-375

14. GARFINKEL DJ, RB SIMPSON, LW REAM, FF WHITE, MP GORDON, EW NESTER1981 Genetic analysis of crown gall: Fine structure map of the T-DNA bysite-directed mutagenesis. Cell 27: 143-153

15. HOOYKAAS PJJ, C ROOBOL, RA SCHILPEROORT 1979 Regulation of the transferof Ti plasmids ofAgrobacterium tumefaciens. J Gen Microbiol 1 10: 99-109

16. HORSCH RB, RT FRALEY, SG ROGERS, PR SANDERS, A LLOYD, N HOFFMAN1984 Inheritance of functional foreign genes in plants. Science 223: 496-498

93

www.plantphysiol.orgon April 16, 2020 - Published by Downloaded from Copyright © 1985 American Society of Plant Biologists. All rights reserved.

94 OWENS AD;

17. MANIATIS T, A JEFFREY, DG KLEID 1975 Nucleotide sequence ofthe rightwardoperator of phage lambda. Proc Natl Acad Sci USA 72: 1184-1188

18. OrrEN L, H DEGREVE, JP HERNALSTEENS, M VAN MONTAGU, 0 SCHIEDER, JSTRAUB, J SCHELL 1981 Mendelian transmission of genes introduced intoplants by the Ti plasmid of Agrobacterium tumefaciens. Molec Gen Genet183: 209-213

19. OWENs LD 1982 Characteristics of teratomas regenerated in vitro from octo-pine-type crown gall. Plant Physiol 69: 37-40

20. PEDERSON HC, J CHRISTIANSEN, R WYNDAELE 1983 Induction and in vitroculture of soybean crown gall tumors. Plant Cell Rep 2: 201-204

21. SOUTHERN EM 1975 Detection of specific sequences among DNA fragments

{D CRESS Plant Physiol. Vol. 77, 1985

separated by gel electrophoresis. J Mol Biol 98: 503-51822. SPANO L, P CONSTANTINO 1982 Regeneration of plants from callus cultures of

roots induced by Agrobacterium rhizogenes on tobacco. Z Pflanzenphysiol106: 87-92

23. THOMASHOW MF, R NUTTA, A MONTOYA, MP GORDON, EW NESTER 1980Integration and organization of Ti plasmid sequences in crown gall tumors.Cell 19: 729-739

24. WHITE FF, G GHIDOSSI, MP GORDON, EW NESTER 1982 Tumor induction byAgrobacterium rhizogenes involves the tranfer ofplasmid DNA to the plantgenome. Proc Natl Acad Sci USA 79: 3193-3197

25. WIDHOLM JM, S RICK 1983 Shoot regeneration from Glycine canescens tissuecultures. Plant Cell Rep 2: 19-20

www.plantphysiol.orgon April 16, 2020 - Published by Downloaded from Copyright © 1985 American Society of Plant Biologists. All rights reserved.