Tn5 insertions in the agrocin 84 plasmid: The conjugal nature of pAgK84 and the locations of determinants for transfer and agrocin 84 production

PLASMID

13, 106-117(1985)

Tn5 Insertions in the Agrocin 84 Plasmid: The Conjugal Nature of pAgK84 and the Locations of Determinants for Transfer and Agrocin 84 Production STEPHENK.FARRAND,*,' JOHN E. SLOTA,* J.-S. SHIM,? AND ALLEN KERR-~ *Department

of Microbiology, Stritch School of Medicine, Loyola University of Chicago, Maywood, and fDepartment of Plant Pathology, Waite Agricultural Research Institute, University of Adelaide, Glen Osmond, South Australia 5064

Illinois 60153,

Received August 13, 1984; revised November 8, 1984 The kanamycin-resistance transposon Tn5 was randomly introduced into pAgK84, a 47.7-kb plasmid coding for agrocin 84 production in Agrobacterium. Using such marked plasmids, pAgK84 was found to be conjugal. It could be transferred to several Agrobacterium strains including those harboring octopine- or nopaline-type Ti plasmids. Its presence has no effect on Ti plasmid functions such as opine utilization and tumorigenicity, but it does confer agrocin 84 immunity upon previously sensitive strains. The plasmid could also be conjugally transferred to a Nod+ Fix+ strain of Rhizobium meliloti. The production of agrocin 84 is expressed in all Agrobacterium and Rhizobium transconjugants tested. The agrocin plasmid could not be introduced into restrictionless Escherichia coli or Pseudomonas aeruginosa recipients by conjugation or transformation. The sites of 92 independent Tn5 insertions were mapped on pAgK84. These insertions are dispersed over the entire length of the plasmid. Analysis of the sites and effects of the Tn5 insertions has allowed us to construct a functional map of pAgKg4. Forty-three of these insertions, spanning a 20-kb segment of the plasmid, abolished or greatly reduced the production of agrocin 84. The presence of two insertions within this segment having an effect on agrocin production suggeststhat at least three regions of the plasmid are involved in agrocin 84 biosynthesis. Fourteen of the Tn5 insertion derivatives are no longer conjugally transferable. These insertions all map to a single region of the plasmid and define about 3.5kb as being associatedwith transfer functions. o 1985 Academic ores, inc.

Agrobacterium tumefaciens has gained attention over the past few years for its ability to induce tumorous growths in many dicotyledonous plant species. Initiation of these crown gall tumors is dependent on the presence of a tumor-inducing plasmid in the bacteria and is the consequence of transfer from bacterium to host plant of a small piece of the Ti plasmid (Chilton et al., 1977). Because any fragment of DNA artificially introduced into this T-DNA can be transferred and incorporated into plant genomic DNA, the A. tumefaciens system holds promise as a vector system for plant genetic engineering (Herr&teens et al., 1980). Unfortunately, A. tumefaciens remains a plant pathogen of considerable economic im’ To whom reprint

requests

should be addressed.

0147-619X/85 $3.00 Copyright 0 1985 by Academic Press. Inc. All rights of reproduction in any form reserved.

portance (Kennedy and Alcom, 1980). It is responsible for nursery and field lossesamong a wide variety of plants including stone-fruit trees (Alconero, 1980), apple and pear trees, grape vines (Perry and Kado, 1982), and omamentals (Bazzi and Rosciglione, 1982). The diseaseis widely distributed in temperate countries and both pathogenic and nonpathogenic A. tumefaciens strains can be readily isolated from soil. A biological mechanism for the field control of crown gall has been developed (Htay and Kerr, 1974; Kerr and Htay, 1974) and appears to be effective in several parts of the world (reviewed by Moore, 1979). The basis of the mechanism is the biosynthesis and secretion by an avirulent A. radiobacter strain, K84, of an antibiotic-like substance which specifically inhibits many tumorigenic strains

106

GENETIC

ANALYSIS

of this phytopathogen (Kerr and Htay, 1974; Kerr and Roberts, 1976). The active substance, called agrocin 84, is a disubstituted fradulent analog of adenine nucleoside (Murphy et al., 1981). Its production is under the control of a 47.7-kb plasmid, pAgK84 (Ellis et al., 1979; Slota and Farrand, 1982). Sensitivity to agrocin 84 appears to require the presenceof certain types of Ti plasmids (Kerr and Roberts, 1976) and evidence to date indicates that all such plasmids specify the capacity to catabolize a class of opines called the agrocinopines (Ellis and Murphy, 1981). Agrocin 84 apparently enters sensitive cells via- the Ti plasmid-encoded agrocinopine transport system (Murphy and Roberts, 1979; Ellis and Murphy, 198I). Agrobacterial strains lacking such a Ti plasmid-encoded permease are invariably resistant to agrocin 84. As effective as the agrocin control system is, failures in field control have occurred. In addition to the existence of intrinsically resistant, agrocinopine noncatabolizing isolates, sensitive agrobacteria can give rise to fully tumorigenic agrocin 84-resistant strains. This can occur by mutations mapping to the Ti plasmid (Holsters et al., 1980) or by acquisition of pAgK84 by the sensitive strains and subsequent expression of plasmid encoded immunity functions (Ellis and Kerr, 1979; Ellis et al., 1979). Relevant to this latter point are the reports that pAgK84 is transmissible to other agrobacteria, apparently by plasmid mobilization (Ellis and Kerr, 1979; Ellis et al., 1979). This threat to the continued successof the biological control of crown gall could be reduced if pAgK84 was suitably engineered with regard to certain critical traits. For example, the use of transfer- or mobilizationdefective derivatives should circumvent the problem of resistance due to immunity functions associated with acquisition of pAgK84 by previously sensitive tumorigenic strains. A rational approach to such engineering requires an understanding of the physical and genetic characteristics of the agrocin plasmid. To this end Slota and Farrand (1982) have isolated pAgK84 in a genetic background

107

OF pAgK84

free from other Ti or opine-catabolic plasmids and have constructed a map of this plasmid with respect to seven restriction endonucleases.We now report the addition of a genetically selectable trait to this plasmid, and experiments which show that it is itself a conjugative plasmid. We also present a functional map of pAgK84 locating regions involved in agrocin production and conjugative transfer. MATERIALS

AND METHODS

Bacterial strains. Bacterial strains used in this study and their relevant characteristics are listed in Table 1. A. tumefaciens strain C58C 1RS- 1 was constructed by transforming pTiC58 into strain C58ClRS (Ellis et al., 1982). Rhizobium meliloti strain AK63 1-1 is strain AK63 1 made resistant to rifampin (25 pg/ml) and 5-fluorouracil (10 &ml). Medium and buffers. L Broth, L agar, AB minimal medium, mannitol glutamate medium (MG), and nutrient agar (NA) have been previously described (Farrand et al., 1981) as has Stonier’s medium (Stonier, 1960), Agrobacterium basal medium (Petit et al., 1978) and minimal medium for Pseudomonas aeruginosa (Olexy et al., 1979). Rhizobium complete medium (CM) and GTS medium were prepared as described by Kiss et al. (1979). TE and TES buffers were prepared as reported by Farrand et al. (1981). LTE buffer contains 10 mM Tris-HCl (Sigma Chemical Co.), 1 mM NazEDTA, pH 8.0. Agrocin 84 assays. Production of agrocin 84 by agrobacteria was determined on either MG medium or Stonier’s medium as described by Slota and Farrand (1982). R. meliloti strains were assayed on GTS medium. In all cases, overlays containing the appropriate indicator strain (see text) were prepared using 0.7% agar buffered with 20 IIIM KP04 buffer, pH 7.2. Conjugations. Filter matings were performed as described by Farrand et al. (198 1). Transconjugants from matings involving the A. tumefaciens recipient strains A 136, SA 101, and C58ClRS were selected on NA contain~.~~~

108

FARRAND ET AL. TABLE 1 BACTERIAL STRAINS~

Strain

Relevant plasmids

Other relevant traits

Source or reference

Agrobacterium

Al36 K84

pAtC58 pAgK84, pAtK84b

RifR, NalR Nontumorigenic, agrocin producer Strain 15955 $101 RiP, StrR RiP, Str” Tra constituitive pTiC58 in C58ClCE. Supersensitive to agrocin 84 Ti plasmid-cured C58

(Watson, et al., 1975) (Slota and Farrand, 1982)

SAlOl C58ClRS C58ClRS1 439

pTi15955 pAtC58 pTiC58, pAtC58 pTiC58 Tra’

NT1

pAtC58

(Beringer et al., 1978) (Maniatis et al., 1982)

-

met-63, pro-22, nal, KmR pro, leu-6, thi-2, hsdR, recA, rpsL serB, k-6, thi-2, hsdR. lacy

-

trp-54, res-I, chl-13, sm-6, mod-1

A. Chakrabarty

-

RifR, Nod’, Fix+, 5-FuR

A. Kondorosi

This laboratory (Ellis et al., 1982) This laboratory (Ellis et al., 1982)

(Watson et al., 1975)

Escherichia coli

pJB4JI

1830 HBlOl

-

1231

(Pi&l

and Farrand, 1983)

Pseudomonas aeruginosa

PA0403 Rhizobium

AK631-1

meliloti

a Abbreviations: chl, chloramphenicol; Fix, nitrogen fixation; 5-Fu, 5-fluorouracil; hsd, host specificity determinant; Km, kanamycin; Zac, lactose; leu, leucine; met, methionine; mod, modification; Nal and nal, nalidixic acid Nod, nodulation; pro, proline; rec. recombination; res, restriction; Rif and rif rifampin; rps, ribosome small subunit protein; ser, wine, Str and sm, streptomycin; thi thiamine; Tra, conjugal transfer; trp, trytophan.

ing kanamycin (kan, 50 pg/ml) and rifampin (r-if, 25 &ml). Mating mixtures containing P. aeruginosa strain PA0403 were plated on NA supplemented with kan (500 &ml) and streptomycin (str, 750 &ml), while those containing Escherichia coli strain HB 101 were plated on NA with kan (50 &ml) and str (150 &ml). Mating mixtures containing R. meliloti strain AK63 1-1 were plated on CM supplemented with kan ( 100 &ml) and rif (25 a/ml). Transformations. Agrobacterium strains were transformed by the freeze-thaw technique of Holsters et al. (1978). E. coli strain 1231 was transformed essentially as described by Pischl and Farrand (1983). P. aeruginosa strain PA0403 was transformed as described by Bagdasarian et al. ( 1981). Plasmid isolations. Plasmid DNA was partially purified by the alkaline miniprep pro-

cedure described by Maniatis et al. (1982). For preparations from agrobacteria, the cell pellets from 1.5 ml of culture were washed with 1.0 ml 0.5 M NaCl, 50 mM Tris-HCl, 10 mM Na2-EDTA, pH 8.0, containing 0.1% Na sarkosyl (J. Ellis, personal communication) before resuspension in the glucoseTris-EDTA-lysozyme lysis solution. When large amounts of plasmid DNA were required, the method of Casseet al. (1979) was used. Such plasmid preparations were further purified by two successive equilibrium centrifugations in CsCl-ethidium bromide (Farrand et al., 1981). Covalently closed plasmid DNA was electrophoretically analyzed on 0.7% agarosegels as previously described (Farrand et al., 1981). Plasmid sizes were determined from such gels by comparing electrophoretic mobilities with those of standard plasmid size markers (Meyers et al., 1976). The stan-

GENETIC

ANALYSIS

dard plasmids used were Rldrdl9, 102 kb; RP4, 58 kb; and pSa, 40 kb (Meyers et al., 1976). Restriction endonuclease analysis. Digestion of purified plasmid DNA samples and the electrophoretic separation and analysis of the fragments on agarose gels were performed as previously described (Slota and Farrand, 1982). In general, plasmid DNA prepared from all strains by the miniprep technique could be digested and analyzed by the same procedure. The sites of Tn5 insertions were determined by separately cleaving plasmid DNA preparations with at least two enzymes. Where ambiguities still existed, a third enzyme was used. Pathogenicity testing, Agrobacterium strains were tested for tumorigenicity on carrot disks as described by Hamada and Fart-and ( 1980). RESULTS

Introduction of a Selectable Marker Although pAgK84 has been physically characterized (Slota and Farrand, 1982), lack of a selectable marker associated with the plasmid has made genetic studies difficult. To supply such a marker, the kanamycinresistance transposon Tn5 was introduced onto pAgK84 by the following method. Twenty independent filter matings were simultaneously performed between A. tumefaciens strain Al36 harboring pAgK84 and E. coli strain 1830 harboring pJB4JI. This latter TnS-containing plasmid can be transferred to A. tumefaciens strains but because of the presence of a defective Mu insert, it cannot replicate in this phytopathogen (Beringer et al., 1978). Transconjugants containing Tn5 as a result of transposition from the nonreplicating pJB4JI were selected on NA containing rifampin and kanamycin. To minimize the number of siblings, one plate containing 200-300 colonies was chosen from each of the 20 independent matings. From each of these 20 plates the colonies were resuspended in 5.0 ml of LTE. The resuspended cells from all the plates were pooled (ca. 100 ml) and washed once with LTE.

OF pAgK84

109

Total plasmid DNA was isolated from these cells, purified on CsCl-EtBr gradients and used to transform A. tumefaciens strain NT 1. Two hundred ninety-one kanamycin-resistant transformants were isolated. When tested against A. tumefaciens strain C58, 202 of these isolates had acquired the ability to produce agrocin 84. Twenty kanamycin-resistant transformants were screened for plasmid DNA by agarose gel electrophoresis. Half of the isolates chosen produced agrocin 84 (Agr+). In all twenty, a new plasmid species was observed with an electrophoretic mobility slightly less than that of pAgK84 (data not shown). Calculations based on standard curves constructed from the electrophoretic mobilities of marker plasmids (see Materials and Methods) indicated that each of these new plasmids is about 53 kb in size. This is the expected size of pAgK84 harboring a Tn5 insertion. To confirm the presence of Tn5, plasmid DNA was purified from one Agr’ and one Agt- transformant and subjectedto restriction endonuclease analysis. Although we have not yet constructed a Hind111 map of pAgK84, we chose this enzyme because it cuts Tn5 twice, generating a characteristic 3.3-kb fragment from the transposon (Jorgensen et al., 1979). As can be seen in Fig. 1, digestion patterns of the plasmids isolated from the two transformants are similar to each other (lanes D and E) and also to the wild-type parent, pAgK84 (lane C). However, each of the transformant plasmids differs from pAgK84 in the absence of one parental fragment and in the presence of three new fragments. In the case of isolate III-1 in lane E, one of the new fragments forms a doublet with parental fragment B. In each digest one of these new fragments comigrates with the pJB4JI-derived 3.3-kb Hind111 fragment internal to Tn5 (lane B). These results are consistent with the notion that the restriction pattern alterations have resulted from insertion of Tn5 into pAgK84. One can also see in Fig. 1 that the insertion event occurring in III-2 (lane D) is different from that occurring in III-1 (lane E). In the former plasmid

110

FARRAND

ABCDE

FIG. I. Endonuclease Hind111 digestion of plasmid DNAs. Plasmid DNA was isolated and purified from bacterial strains as described under Materials and Methods. The DNA samples were digested with Hind111 and the fragments separated by electrophoresis in a 0.7% agarose gel. The lanes contain: (A), X DNA; (B), pJB4JI; (C), pAgK84; and pAgK84 Tn5 insertion derivatives (D) III-2 (Agr’) and (E), III-I (Agr-).

Tn5 has inserted in Hind111 fragment D; in the latter, Hind111 fragment C is the target site. Conjugal Transfer by pAgK84

Acquisition of pAgK84 by tumorigenic agrobacteria may represent one pathway for the failure of field control by agrocin 84 (Panagopoulos et al., 1979). It has been reported that pAgK84 can be transferred among agrobacteria via mobilization by other conjugal agrobacterial plasmids such as pAtK84b (Ellis and Kerr, 1979; Ellis et al., 1979). However, it is also possible that pAgK84 is conjugative in its own right. With Tn5 inserted into the plasmid, we now had a genetically selectable marker by which this char-

ET AL.

acter could be tested. Two A. tumefaciens strain NT1 transformants, one Agr+ the other Agr-, were mated on nitrocellulose filters with several recipients including octopineand nopaline-utilizing tumorigenic agrobacteria, a Nod+ Fix+ strain of R. meliloti and restrictionless mutants of E. coli and P. aeruginosa. Following overnight incubations on NA or CM plates, the mating mixtures were harvested, diluted, and aliquots plated on selective media as described under Materials and Methods. Results presented in Table 2 show that, while the frequencies were low, pAgK84 is transferable to eachAgrobacterium recipient tested. Furthermore, the progeny exhibited the agrocin production character of the donor with which they were crossed. It is also clear that the presenceof the agrocin plasmid has no effect on the two classical Ti plasmid-encoded functions, opine catabolism and tumorigenicity. The agrocin plasmid was also found to be transferable to R. meliloti strain AK63 l-l. When tested against A. tumefaciens strain C58, the rhizobial progeny were found to have acquired the ability to produce agrocin 84 (Fig. 2). In contrast to these results, we were unable to demonstrate conjugal transfer of pAgK84 to restrictionless E. coli or P. aeruginosa recipients (Table 2). Furthermore, no kanamycin-resistant transformants were isolated when competent E. coli or P. aeruginosa cells were incubated with TnS-marked pAgK84 DNA. Control experiments with pDP37, a broad-host-range TnS-containing derivative of R68.45 (Pischl and Farrand, 1983) yielded transformants at frequencies ranging from lo5 per microgram of DNA for E. coli to lo4 per microgram of DNA for P. aeruginosa (data not shown). Function Mapping

The sites of 92 independent Tn5 insertions were determined by restriction endonuclease analysis (Fig. 3). Our strategy involved first digesting each plasmid with HpaI which located the transposon at one of two positions on one of the two HpaI fragments. The

111

GENETIC ANALYSIS OF pAgK84 TABLE 2 CONJUGAL TRANSFER OF pAgK84 AND CHARACERISTICS OF THE TRANSCONJUGANTS Agrocin

Donor”

phenotype

Recipient

NT1 (M-2)

Agr+

A. tumefaciens Al36 SAlOl CMClRS-1 R. meliloti AK631-I E. coli HBlOl 1231 P. aeruginosa PA0403

NT1 (III-I)

A. tumefaciens Al36 SAlOl C58ClRS1 R. meliloti AK631-1 E. coli HBIOl 1231 P. aeruginosa PA0403

Km Transfer frequency*

Agrocin phenotype of transconjugant

Opine utiIizationc

None O&opine Nopaline

Tumorgenicityd +++ ++

IO-’ IO-’ IO-’

AIT+ &T+ &T+

IO-*

A@+

NT

NT

<1o-9 <10-g

-

-

-

<1o-9

-

-

-

10-s lo-’ 10-s

AW &FAIS-

None Octopine Nopaline

+++ ++

lo-*

AIT-

NT

NT


-

-

-


-

-

-

(1The numbers in parenthesesindicate the Tn5 insertion derivative of pAgK84 present in the donor strain. b Expressed as the number of kanamycin-resistant transconjugants selected per input donor. c Determined on solid basal medium with octopine or nopaline (2 mg/ml) as sole source of carbon and nitrogen. NT = not tested. d Determined on carrot disks as described by Farrand et al. (1981). NT = not tested.

choice as to which of these two positions was correct was made on the basis of cleavage by a second enzyme, usually BumHI, &I, or SacI. In the rare caseswhere ambiguities still existed, cleavagewith a third enzyme resolved the location. In many cases we were also able to orient the polarity of the insertion with respect to the asymmetry of the BarnHI site within Tn5. The observation that almost one third (89/ 29 1) of the kanamycin-resistant Tn5 insertion derivatives fail to produce detectable agrocin 84 suggestedto us that the transposon may have inserted into structural or regulatory loci for agrocin synthesis. Figure 4 shows the Tn5 map locations for 43 of the Agr- derivatives. These insertions span a 20-kb segment of the plasmid from coordinate 32.8 (with

respect to the single XbaI site) to coordinate 4.9. The initial assayswere performed using our standard indicator, A. tumefaciensstrain 08. However, when these 43 isolates were tested against the supersensitive A. tumefaciens strain 439, 5 were found to produce low but detectable amounts of the agrocin (Figs. 4 and 5). A number of Tn5 derivatives were found to be nonconjugal to A. tumefuciensstrain C58ClRS. The sites of the Tn5 elements were mapped in 14 of these plasmids. The insertions were all located in a small region between coordinates 27.8 and 31.2 (Fig. 6). All Tra- derivatives were Agr+ and all Agrderivatives were Tra+. Insertions having no effect on agrocin 84 production or conjugal transfer mapped from

112

FARRAND

ET AL.

FIG. 2. Production of agrocin 84 by R. meliloti transconjugants. Matings between A. tum&ciens strain NT1 harboring the agrocin insertion plasmid III-2 and R. mdiluti strain AK63 1-I were performed as described in the text. Assays for the production of agrocin 84 were performed on GTS medium using strain C58 as the indicator. Strains tested are: 1 and 5, A. tumefaciens NT1 harboring insertion derivative 111-2; 2 and 6, A. tumefaciens NT I ; 3 ‘and I, R. meliloti AK63 I- I transconjugant; 4 and 8, R. meliloti AK631-1.

coordinate 6.2 clockwise to 27.2. The Tra+, Agr’ insertion A-56, mapping at coordinate 3 1.6 separates the transfer region from the region involved in agrocin 84 biosynthesis (Fig. 3). In addition, two Agr+ insertion derivatives, A-60 (40.9 kb) and B-58 (1.2 kb), harbor Tn5 elements mapping within the agrocin biosynthetic region (Figs. 3 and 4).

C58 survive to become recipients. However, our analyses show that all transconjugants tested still harbor pTiC58 and express virulence functions (Table 2). These strain C58 transconjugants now show immunity to exogenously added agrocin 84. Such progeny are similar in all respects to those derived from matings with strain K84 (Ellis et al., 1979). That pAgK84 is conjugal to R. meliloti and that the transconjugants express agrocin DISCUSSION 84 production is not surprising (Table 2, Fig. Results from our studies, made possible 2). Hooykaas et al. (1977) have reported the by the introduction of a selectable marker transfer of Ti plasmids to the rhizobia. Such onto the plasmid, show that pAgK84 is itself transconjugants express Ti plasmid encoded a conjugative plasmid, transferable to several traits including opine utilization and tumoragrobacteria and at least one strain of Rhi- igenicity. In addition, Thomson and Hendson zobium (Table 2). Following such transfer, (1983) reported that a plasmid encoding recipients gain the ability to produce agro- agrocin 84-like activity could be mobilized from A. tumefaciens to Rhizobium strains by tin 84. It is interesting to note that pAgK84 is RP4. The rhizobial progeny were found to transferable to the agrocin-sensitive tumori- produce the agrocin 84-like agent. Nor is it genie strain C58. Even though the donor surprising that pAgK84 could not be introstrain is secreting agrocin in the high density duced into E. coli or P. aeruginosa. The Ti filter-bound mating mix, sufficient numbers plasmids also show similar host range limiof C58 cells evidently survive to acquire the tations. Our observation that E. coli and P. plasmid. It might be argued that only agrocin- aeruginosa recipients could not be transresistant Ti plasmidcured derivatives of strain formed with purified pAgK84 DNA indicates

113

GENETIC ANALYSIS OF pAgK84

FIG. 3. Map of pAgK84 showing the sites of 92 independent Tn5 insertions. Mapping was performed as described in the text. Where determined, the crossbar on each insertion shows the polarity with respect to the 2.8-kb BumHI arm of Tn5.

that this plasmid is not capable of replication in these two organisms. The ability of pAgK84 to transfer to Agrobacterium hosts has implications for crown Indicator Strain C58 439

---

++ ++

-

----= ----=

gall field control by strain K84. There have been recent reports that satisfactory control is not always achieved (Moore, 1979; Moore and Warren, 1979; Panagopoulos et al., 1979;

Agrocin 84 Production + -----+ -;---I---

-

I-,Is:;;III1;;1:;:~

1

Sma I t -1 I

kb

,,,,,,I1

,I

35

40

(,(,,,,,,

45

470

I

5

FIG. 4. Map positions of Tn5 insertions affecting agrocin 84 production. The insertion mutants were all assayed against A. tumefuciens strain C58 and its agrocin supersensitive derivative, 439 as described in the text. +, Produces agrocin 84; -, no agrocin 84 detected.

114

FARRAND ET AL,

FIG. 5. Decreased production of agrocin 84 by certain Tn5 insertion mutants. Following inoculation with mutants of interest and incubation for 48 h, the plates were overlaid with A. fumefuciens strains C58 (A) or 439 (B). Insertion derivatives assayedare: I, H-64; 2, C-19; 3, C-21; 4, D-l 1; 5, D-12; and 6, the wild-type producer, A- I

Alconero, 1980). In two cases, resistant A. agrobacteria. They suggested that pAgK84 tumefaciens strains appeared following treat- was being mobilized by pAtK84b, the nopalment with strain K84 (Panagopoulos et al., ine degradative plasmid coresident in strain 1979). These resistant agrobacteria retained K84 (see, also, Ellis and Kerr, 1979). Our tumorigenicity and had gained the ability to results show pAgK84 to be self conjugal. synthesize agrocin 84. Slota and Farrand Both mechanisms allow for the introduction (1982) have also reported that the tumorigenic of pAgK84 into the tumorigenic agrobacterial strain, Bo542, synthesizesan agent with agro- population. Because acquisition of the plastin 84-like activity. They were able to show mid confers immunity to the agrocin, such that this strain harbors a plasmid closely transfer into previously sensitive agrobacteria related to pAgK84. These studies suggestthat would result in the appearance of strains no pAgK84 can transfer from one Agrobacterium longer controllable by strain K84. A mechato another. Ellis et al. (1979) reported that nistic basis therefore exists to explain the such transfer can occur between strain K84 findings of Panagopoulos et al. (1979). The and other tumorigenic and nontumorigenic self conjugal nature of pAgK84 also brings into question whether pAtK84b really does act to mobilize the agrocin plasmid. Experi-Tra Phenotype ments are currently underway to determine +--:--------++ + +++ +++ + ++ if the nopaline catabolic plasmid acts as a mobilizing agent. Our strategy involving the use of pJB4JI resulted in the isolation of a large number of independent Tn5 insertions into pAgK84. SmaI The results presented in Fig. 3 show that these insertions map with fairly even distribution over the entire plasmid. However, FIG. 6. Map positions of Tn5 insertions affecting there are several regions where adjacently conjugal transfer to A. tumefaim tin NTl. Mappings and matings were performed as described in the text. mapping Tn5 elements are separated by con-

GENETIC ANALYSIS OF pAgK84

siderable distances. Most notably, the consecutive insertions A-7 (coordinate 18.1) and A- 10 (coordinate 22.0) are almost 4 kb apart. These regions may represent areas refractory to the transposon. Alternatively, one or more such regions may encode functions required for plasmid maintenance or replication. It should be noted that no alterations due to the defective Mu in pJB4JI (Beringer et al., 1978) were observed. In mapping the insertions every alteration could be explained by a single Tn5 transposition. Almost one-third of the transformants harboring pAgK84 with Tn5 insertions do not secrete detectable agrocin 84. When a series of these plasmids was analyzed, the Tn5 insertions were found to map over a 20-kb segment from coordinate 32.8 clockwise to coordinate 4.7 (Figs. 3 and 4). Interspersed within this region are two insertions having no demonstrable effect on agrocin production (Fig. 4). In addition, a third set of insertions, also mapping to this region, greatly reduces the production of agrocin 84 (Figs. 4 and 5). This latter class was detectable only when we used an indicator strain supersensitive to the antiagrobacterial agent. This set of insertions may lie in intercistronic regions but still exert polar effects on agrocin synthesis loci. Alternatively, they may lie within coding sequences but at positions in which they only partially effect the activities of the translational products of the affected genes. On the assumption that B-58 and A-60, the two insertions having no effect on agrocin production, map to intergenic regions, we have divided the 20-kb segment into three subsections (Fig. 7). The largest of these, spanning coordinates 32.8 to 40.9, is relatively devoid of insertions (Fig. 3) and there may be additional small regions in this section not involved in agrocin biosynthesis. We are screening remaining isolates for Tn5 insertions mapping to this region of the plasmid. At any rate, it is interesting to note that a large segment of DNA is necessaryfor agrocin 84 production. This would suggest that the biosynthetic pathway for this antibiotic is complex.

115

FIG. 7. Functional map of pAgK84. Regions defined by consecutive Tn5 insertions are shown in solid lines. Dotted lines representdistancesbetween the last insertion affecting a phenotype and the nearest insertion having no effect on that trait. ApAgBo542 representsthat portion of pAgK84 absent in the closely related agrocinogenic plasmid, pAgBo542 (Slota and Farrand, 1982).

General screening yielded a number of isolates no longer able to conjugally transfer the agrocin plasmid. The insertions in 14 of these mutants all map to one region (Figs. 6 and 7). In addition, the Tn5 elements in the remaining Tra- mutants examined have been located to this region only (data not shown). Since the precise locations of these later insertions have not been determined, they are not shown on the map. The region defined by the mapped insertions indicates that about 3.5 kb of the plasmid encodes conjugal transfer functions. This is in marked contrast to other conjugative plasmids; in the best studied systems, Tra genes usually occupy in excess of 10 kb. For exampie, the Tra region of F encompassessome 33 kb (Willetts and Skurray, 1980) while those of RP4 and R46 span about 12 kb (Barth, 1979) and 20 kb (Brown and Willetts, 1981), respectively. The Tra regions of the octopine and nopaline Ti plasmids have not been sufficiently characterized to make comparisons with pAgK84.

116

FARRAND ET AL.

It is possible that one or more regions of the agrocin plasmid refractory to Tn5 insertions encodes additional transfer functions. The isolation of a large number of independent mutants unable to synthesizeagrocin 84 should be of value in determining the biosynthetic pathway for this novel antibacterial agent. In addition, analysis of Tn5 insertions having no effect on agrocin biosynthesis will allow us to identify and map mutations affecting other plasmid functions such as replication and stability. Finally, definition of the transfer region by Tn5 mutagenesis presents us with strategies for the construction of stable, Tra- deletion derivatives of pAgK84. Such constructs no longer transferable to tumorigenic strains may prove valuable as crown gall field control strains. ACKNOWLEDGMENTS This work was supported in part by Grant CA19402 from the National Cancer Institute and by Grant 82CRCR-I-1092 from the U. S. Department of Agriculture to S. K. Farrand. A. Kerr received support from the Australian Research Grants Scheme. Note added in proof: NT1 contains the cryptic megaplasmid pAtC58 and it could be argued that this element mobilizes the transfer of pAgK84. A derivative of C58 lacking both pTiC58 and pAtC58 has recently become available [C. Rosenberg and T. Huguet (1984) Mol. Gen. Genef. 196, 533-5361. When Tra+ Tn5 insertion derivatives of pAgK84 were introduced into this strain, the resultant transformants transferred the agrocin plasmids to recipient agrobacteria at frequencies indistinguishable from those obtained with NT1 donors. This shows that conjugal transfer of pAgK84 is not mediated by pAtC58.

REFERENCES AEONERO, R. (1980). Crown gall Agrobacterium tumefaciens of peaches Prunus persica from Maryland, South Carolina and Tennessee USA and problems with biological control. Plant Dis. 64, 835-838. BAGDASARIAN,M., Luaz, R., RUCKERT,B., FRANKLIN, F. C. H., BAGDASARIAN,B. B., FREY,J., AND TIMMIS, K. N. (198I). Specific-purposeplasmid cloning vectors. II. Broad-host-range, high copy number RSFIOIOderived vectors and a host-vector system for gene cloning in Pseudomonas. Gene 16, 237-241. BARTH, P. (1979). RP4 and R3OOBas wide host range plasmid cloning vehicles. In “Plasmids of Medical, Environmental and Commercial Importance” (K. Timmis and A. Puhler, eds.), pp. 399-410. Elsevier/ North-Holland, Amsterdam.

BAZZI, C., AND ROSCIGLIONE, 8. (1982). Agrobacterium lumeficiens biotype 3 causal agent of crown gall on Chrysanthemum in Italy. Phytopathol. Z. 103, 280284.

BERINGER,J. E., BEYNON,J. L., BUCHANON-WOLLASTON,A. V., AND JOHNSTON,A. W. B. (1978). Transfer of the drug resistance transposon Tn5 to Rhizobium. Nature (London) 276, 633-634.

BROWN, A. M. C., AND WILLETTS, N. S. (1981). A physical and genetic map of the IncN plasmid R46. Plasmid 5, 188-20 I. CASSE,F., BOUCHER,C., JULLIOT, J. S., MICHEL, M., AND DENARI~,J. (1979). Identification and characterization of large plasmids in Rhizobium meliloti using agarose gel electrophoresis. J. Gen. Microbial. 113, 229-242.

CHILTON, M.-D., DRUMMOND, M. H., MERLO, D. J., MONTOYA, A. L., GORDON, M. P., AND NESTER, E. W. (1977). Stable incorporation of plasmid DNA into higher plant cells: The molecular basis of crown gall tumorigenesis. Cell. 11, 263-271. ELLIS, J. G., AND KERR, A. (1979). Transfer of agrocin 84 production from strain 84 to pathogenic recipients: A commment on the previous paper. In “Soil-Borne Plant Pathogens” (B. Schippers and W. Gams, Eds.), pp. 579-583. Academic Press,New York. ELLIS, J. G., KERR, A., VAN MONTAGU, M., AND SCHELL,J. (1979). Agrobacterium: Genetic studies on agrocin 84 production and the biological control of crown gall. Physiol Plant Pathol. 15, 31 I-319. ELLIS, J. G., AND MURPHY, P. J. (1981). Four new opines from crown gall tumors, their detection and properties. Mol. Gen. Genet. 181, 36-43. ELLIS, J. G., MURPHY, P. J., AND KERR, A. (1982). Isolation and properties of transfer regulatory mutants of the nopaline Ti plasmid, pTiC58. Mol. Gen. Genef. l&275-28 1. FARRAND, S. K., KADO, C. I., AND IRELAND, C. R. (198 I). Suppression of tumorigenicity by the IncW Rplasmid pSa in Agrobacterium tumefaciens. Mol. Gen. Genet. 181, 44-5 I.

HAMADA, S. E., AND FARRAND,S. K. (1980). Diversity among B6 strains of Agrobacterium tumefaciens. J. Bacterial. 141, 1127-l 133. HERNALSTEENS, J.-P., VAN VLIET, F., DE BEUCKELEER, M., DEPICKER,A., ENGLER,G., LEMMERS,M., HOLSTERS,M., VAN MONTAGU,M., and SCHELL,J. (1980). The Agrobacterium tumefaciens Ti plasmid as a host vector system for introducing foreign DNA in plant cells. Nature (London) 287, 654-656.

HOLSTERS,M., DE WAELE,D., DEPICKER,A., MESSENS, E., VAN MONTAGU, M., AND SCHELL, J. (1978). Transfection and transformation of Agrobacterium tumefaciens. Mol. Gen. Genet. 163, 181-187. HOLSTERS,M., SILVA, B., VAN VLIET, F., GENETELLO, C., DEBLOCK, M., DHAESE,P., DEPICKER,A., INzfi, D., ENGLER,G., VILLARROEL,R., VAN MONTAGU, M., AND SCHEL~J. (1980).The functional o@attizatiOn of the nopaline A. tumefaciens plasmid pTiC58. Plasmid 3, 2 12-230.

GENETIC ANALYSIS OF pAgK84 HOOYKAAS,P. J. J., KLAPWIJK, P. M., NUTI, M. P., QHILPEROORT, R. A., AND KORSCH, A. (1977). Transfer of the Agrobacterium tumefaciens Ti-plasmid to avirulent agrobacteria and to Rhizobium ex planta. J. Gen. Microbial. 98, 477-484. HTAY, K., AND KERR, A. (1974). Biological control of crown gall: Seedand root inoculation. J. Appl. Bacferiol. 37, 525-530.

117

MURPHY, P. J., TATE, M. E., AND KERR, A. (1981). Substituents at N6 and C-5’ control selective uptake and toxicity of the adenine-nucleotide bacteriocin, agrocin 84, in agrobacteria. Eur. J. Biochem. 115, 539-543.

OLEXY, W. M., BIRD, T. J., GRIEBLE, H. G., AND FARRAND,S. K. (I 979). Hospital isolates of Serratia marcescens transferring ampicillin, carbenicilin and JORGENSEN,J. S., ROTHSTEIN,S. J., AND REZNIKOFF, gentamicin resistance to other gram-negative bacteria including Pseudomonas aeruginosa. Antimicrob. Agents W. S. (1979). A restriction enzyme cleavage map of Tn5 and location of a region encoding neomycin Chemother. 15, 93- 100. PANAGOPOULOS, C. G., PSALLIDAS,P. G., AND ALIVIresistance.Mol. Gen. Genet. 177, 65-72. KENNEDY,B. W., AND ALCORN, S. M. (1980). Estimates ZATOS,A. S. (1979). Evidence of a breakdown in the effectiveness of biological control of crown gall. In of USA crop losses to prokaryotic plant pathogens. Plant Dis. 64, 674-676. “Soil-Borne Plant Pathogens” (B. Schippers and W. KERR, A., AND HTAY, K. (1974). Biological control of Gams, eds.), pp. 569-578. Academic Press,New York, crown gall through bacteriocin production. Physiol. N. Y. Plant Pathol. 4, 37-44. PERRY,K. L., AND KADO, C. I. (1982). Characteristics of Ti plasmids from broad host range and ecologically KERR, A., AND ROBERTS,W. P. (1976). Agrobacterium: specific biotype 2 and 3 strains of Agrobacterium Correlations between and transfer of pathogenicity, tum&aciens. J. Bacterial. 151, 343-350. o&opine, nopaline metabolism and bacteriocin 84 PETIT, A., TEMPT, J., KERR, A., HOLSTERS,M., VAN sensitivity. Physiol. Plant Pathol. 9, 205-2 1 I. MONTAGU, M., AND SCHELL, J. (1978). Substrate KISS, G. B., VINCZE, E., KALMAR, Z., FORRAI,T., AND induction of conjugative activity of Agrobacterium KONDOROSI,A. (1979). Genetic and biochemical ans!tumtlfaciens Ti plasmids. Nature (London) 271, 570ysis of mutants affected in nitrate reductasc in Rhizo571. bium meliloti. J. Gen. Microbial. 113, 105-l 18. PISCHL, D. L., AND FARRAND,S. K. (1983). TransposonMANIATIS, T., FRITSCH, F. F., AND SAMBROOK,J. facilitated chromosome mobilization in Agrobacterium (1982). “Molecular Cloning, A Laboratory Manual,” tumefaciens. J. Bacterial. 153, 145 1- 1460. pp. 368-369. Cold Spring Harbor Laboratory, Cold SLOTA, J. E., AND FARRAND, S. K. (1981). Genetic Spring Harbor, N. Y. isolation and physical characterization of pAgK84, the MEYERS, J. A., SANCHEZ, D., ELWELL, L. P., AND plasmid responsiblefor agrccin 84 production. Plasmid FALKOW,S. (1976). Simple agarosegel electrophoretic 8, 175-186. method for the identification and characterization of STONIER,T. (1960). Agrobacterium tumefaciens Conn. plasmid deoxyribonucleic acid. J. Bacterial. 127, 1529II. Production of an antibiotic substance.J. Bacterial. 1537. MOORE, L. W. (1979). Practical use and success of Agrobacterium radiobacter strain 84 for crown gall control. In “Soil-Borne Plant Pathogens” (B. Chippers and W. Gams, eds.), pp. 553-561. Academic Press, New York, N. Y. MOORE,L. W., AND WARREN,G. (1979). Agrobacterium radiobacter strain 84 and biological control of crown gall. Annu. Rev. Phytopathol. 17, 163-179. MURPHY, P. J., AND ROBERTS,W. P. (1979). A basis for agrocin 84 sensitivity in Agrobucterium radiobacter. J. Gen. Microbial.

114, 207-213.

79, 889-898.

THOMSON,J., AND HENDSON,M. (1983). Transfer of the plasmid encoding agrocin 84 from Agrobacterium tumefaciens to Rhizobium. In “Abstr. Annu. Meeting Amer. Sot. Microbiology,” H52., p. 114. WATSON,B., CURRIER,T. C., GORDON,M. P., CHILTON, M.-D., AND NESTER,E. W. (1975). Plasmid required for virulence of Agrobacterium tumefaciens. J. Bacteriol. 123, 255-264.

WILLETTS,N., AND SKURRAY,R. (1980).The conjugative system of F-like plasmids. Annu. Rev. Genet. 14, 4179.