The txtAB genes of the plant pathogen Streptomyces acidiscabies encode a peptide synthetase required for phytotoxin thaxtomin A production and pathogenicity


  • Frank G. Healy,

    1. Department of Plant Pathology, 334 Plant Science Building, Cornell University, Ithaca, NY 14853, USA.
    Search for more papers by this author
    • †Present address: Department of Microbiology and Cell Science, Box 110700, Museum Road, University of Florida, Gainesville, FL 32611-0700, USA.

  • Michael Wach,

    1. Department of Plant Pathology, 334 Plant Science Building, Cornell University, Ithaca, NY 14853, USA.
    Search for more papers by this author
  • Stuart B. Krasnoff,

    1. USDA Agricultural Research Service, US Plant, Soil, and Nutrition Laboratory, Tower Road, Ithaca, NY 14853, USA.
    Search for more papers by this author
  • Donna M. Gibson,

    1. USDA Agricultural Research Service, US Plant, Soil, and Nutrition Laboratory, Tower Road, Ithaca, NY 14853, USA.
    Search for more papers by this author
  • Rosemary Loria

    Corresponding author
    1. Department of Plant Pathology, 334 Plant Science Building, Cornell University, Ithaca, NY 14853, USA.
    • *For correspondence. E-mail rl21@; Tel. (+1) 607 255 7831; Fax (+1) 607 255 4471. †Present address: Department of Microbiology and Cell Science, Box 110700, Museum Road, University of Florida, Gainesville, FL 32611-0700, USA.

    Search for more papers by this author


Four Streptomyces species have been described as the causal agents of scab disease, which affects economically important root and tuber crops worldwide. These species produce a family of cyclic dipeptides, the thaxtomins, which alone mimic disease symptomatology. Structural considerations suggest that thaxtomins are synthesized non-ribosomally. Degenerate oligonucleotide primers were used to amplify conserved portions of the acyladenylation module of peptide synthetase genes from genomic DNA of representatives of the four species. Pairwise Southern hybridizations identified a peptide synthetase acyladenylation module conserved among three species. The complete nucleotide sequences of two peptide synthetase genes (txtAB) were determined from S. acidiscabies 84.104 cosmid library clones. The organization of the deduced TxtA and TxtB peptide synthetase catalytic domains is consistent with the formation of N-methylated cyclic dipeptides such as thaxtomins. Based on high-performance liquid chromatography (HPLC) analysis, thaxtomin A production was abolished in txtA gene disruption mutants. Although the growth and morphological characteristics of the mutants were identical to those of the parent strain, txtA mutants were avirulent on potato tubers. Moreover, introduction of the thaxtomin synthetase cosmid into a txtA mutant restored both pathogenicity and thaxtomin A production, demonstrating a critical role for thaxtomins in pathogenesis.


Streptomycetes are soil-borne filamentous eubacteria with GC-rich genomes that produce the vast majority of commercially important antibiotics as well as a number of extracellular hydrolases and other compounds. Among the hundreds of described Streptomyces species, only four species to date have been described as plant pathogens. These species, Streptomyces scabies, Streptomyces acidiscabies, Streptomyces turgidiscabies and Streptomyces ipomoeae, are the causal agents of scab diseases of economically important root and tuber crops (Loria et al., 1997; Miyajima et al., 1998).

The mechanisms of pathogenicity of these species remained elusive until the identification of thaxtomins [cyclo-(l-4-nitrotryptophyl-l-phenylalanyl) and related congeners] (Fig. 1) found in extracts of S. scabies-infected plant tissue. Subsequently, purified thaxtomins were shown to elicit symptoms on host tissue identical to those of infection (King et al., 1989; 1992; Lawrence et al., 1990); in addition, phytotoxic activity was observed on seedlings of both monocot and dicot plant species (Leiner et al., 1996). The importance of thaxtomins in plant pathogenicity was further corroborated through the isolation of chemically mutagenized S. scabies strains that were deficient or reduced in virulence; all of these mutants produced reduced or undetectable levels of thaxtomin A relative to the parent strain (Healy et al., 1997; Goyer et al., 1998).

Figure 1.

Structure of major thaxtomins produced by plant pathogenic streptomycetes.

Although S. scabies, S. acidiscabies, S. turgidiscabies and S. ipomoeae are clearly distinct based on numerous taxonomic criteria (Healy and Lambert, 1991; Takeuchi et al., 1996; Loria et al., 1997), they all produce one or more congeners of the thaxtomin family. Whereas thaxtomin A is the predominant compound produced by S. scabies, S. acidiscabies and S. turgidiscabies, thaxtomin C is the predominant compound produced by S. ipomoeae (King et al., 1994; Loria et al., 1997; Bukhalid et al., 1998). The production of thaxtomins by genetically diverse plant pathogens, along with the previously described sequence conservation of the ORFtnpnec1–IS1629 region among the three potato pathogens (S. scabies, S. acidiscabies and S. turgidiscabies) (Bukhalid et al., 1998; Healy et al., 1999), strongly suggests that pathogenicity factors and pathogenicity-associated loci have been acquired through horizontal gene transfer. Based on these findings, it has been postulated that pathogenicity-associated factors may reside within an as yet uncharacterized pathogenicity island that has played a critical role in the evolution of these plant pathogens (Bukhalid et al., 1998; Healy et al., 1999).

Many small microbially derived peptides are synthesized on multifunctional peptide synthetases and often share certain characteristics, e.g. the incorporation of ‘non-proteinogenic’ amino or hydroxy acids, which may be epimerized, N-methylated, hydroxylated or cyclized in the final products (Hermann et al., 1996; von Dohren et al., 1997; Marahiel, 1997; Marahiel et al., 1997). A minimal peptide synthetase contains adenylation and thiolation (thioesterification, also referred to as peptidyl carrier protein) domains encoded by ‘autonomous’ modules that occur in collinear arrangement within the peptide synthetase gene encoding them (von Dohren et al., 1997; Marahiel, 1997; Marahiel et al., 1997). The structures of thaxtomins (Fig. 1) strongly suggest that they too are non-ribosomally derived, given that they are cyclized dipeptides composed of the unusual N-methylated l-4-nitrotryptophyl and variously hydroxylated, often N-methylated l-phenylalanyl moieties.

The occurrence of conserved stretches of amino acids within peptide synthetase enzymatic domains has been exploited for the identification and cloning of peptide synthetase modules and genes through degenerate oligonucleotide-primed polymerase chain reaction (PCR) amplification from genomic DNA of peptide producers (Turgay and Marahiel, 1994; Burmester et al., 1995; Nikolskaya et al., 1995; Meissner et al., 1996; Panaccione, 1996; Chong et al., 1998). One potential obstacle in cloning a specific synthetase is that a given organism may produce numerous synthetase-derived peptides and, consequently, may harbour multiple peptide synthetase genes, all of which may be co-amplified during PCR. Alternatively, a given degenerate oligonucleotide may preferentially anneal to a specific synthetase because of higher nucleotide sequence complementarity between the oligonucleotide primer and DNA template; in this case, other synthetases would be under-represented in the PCR products.

The production of thaxtomins by taxonomically distinct plant pathogenic streptomycetes coupled with our current model of pathogen evolution via horizontal gene transfer suggests that thaxtomin structural genes would probably be conserved across species boundaries, i.e. ‘thaxtomin synthetase’ genes would be highly similar, if not identical, from one species to the next. If these genes were conserved, then, through a pairwise hybridization strategy using radiolabelled putative peptide synthetase PCR amplification products as probes, it should be possible to bypass the cloning/characterization of ‘species-specific’ synthetases and identify directly those synthetase(s) shared by diverse species yet which catalyse the production of similar non-ribosomally derived peptides, such as thaxtomins. The identification of the thaxtomin structural genes is essential to an evaluation of the role of thaxtomins in plant pathogenicity as well as the further exploration of the role of pathogenicity gene transfer in pathogen evolution. This work describes the utilization of a pairwise hybridization strategy to identify and characterize the thaxtomin synthetase of S. acidiscabies 84.104. Whole potato tuber pathogenicity assays using thaxtomin synthetase null mutants as well as mutants carrying thaxtomin synthetase genes demonstrate that the pathogenicity of S. acidiscabies is thaxtomin dependent.


Identification of a peptide synthetase conserved among three out of four thaxtomin-producing plant pathogenic streptomycetes

The probable non-ribosomal production of the cyclic dipeptide thaxtomins, as well as the production of thaxtomins by genetically diverse plant pathogenic streptomycetes, suggested a pairwise hybridization strategy to identify the thaxtomin synthetase. Pairwise high-stringency Southern hybridizations (Fig. 2) using radiolabelled putative peptide synthetase probes derived from genomic DNA of four phylogenetically distinct thaxtomin-producing plant pathogens against genomic DNA of the same four species revealed that a putative synthetase fragment derived from PCR amplification of S. turgidiscabies 94.09 (HiC-13) DNA not only hybridized to probable peptide synthetase genes present in S. turgidiscabies, but also to specific 3.5 kb KpnI fragments found in S. acidiscabies and S. scabies genomic DNA. The probe, however, did not hybridize to genomic DNA from the fourth thaxtomin producer, S. ipomoeae (possible reasons for this are discussed below).

Figure 2.

Pairwise Southern hybridization analysis of KpnI-digested genomic DNA of the four representative thaxtomin-producing plant pathogenic streptomycetes using purified radiolabelled putative peptide synthetase (PS) fragment probes derived from PCR amplification of genomic DNA from the same strains using degenerate oligonucleotide primers based on conserved A2 and A3 motifs of PS adenylation domains (see text). i, S. ipomoeae 81.45; s, S. scabies 87.22; a, S. acidiscabies 84.104; t, S. turgidiscabies 94.09. Putative PS probes are listed above the brackets. Arrows indicate conserved 3.5 kb KpnI fragments identified in S. scabies 87.22 and S. acidiscabies 84.104 DNA using a radiolabelled S. turgidiscabies PS probe.

The S. turgidiscabies probe was used to identify hybridizing S. acidiscabies 84.104 cosmids containing possible peptide synthetase genes. Colony hybridizations revealed three such cosmids, SACOS1, SACOS5 and SACOS9 (not shown). These were analysed by restriction mapping and standard subcloning techniques. Nucleotide sequence analysis of hybridizing subclones indicated that we had identified a peptide synthetase, based on similarities to a number of bacterial and fungal peptide synthetases deposited in nucleotide and protein databases (not shown). All of the cosmids contained the 3.5 kb KpnI peptide synthetase fragment found in pairwise hybridization experiments; this fragment was cloned and used as a probe in hybridizations against genomic DNA from the four thaxtomin producers. As seen in Fig. 3, the cloned 3.5 kb KpnI peptide synthetase fragment hybridized with genomic DNA from S. scabies and S. turgidiscabies, but failed to hybridize with genomic DNA from S. ipomoeae.

Figure 3.

Southern hybridization analysis of KpnI-digested genomic DNA of S. ipomoeae 81.45 (i ), S. scabies 87.22 (s), S. acidiscabies 84.104 (a) and S. turgidiscabies 94.09 (t ) using a radiolabelled 3.5 kb KpnI ‘txtA–txtB’ (truncated) fragment derived from S. acidiscabies cosmid SACOS1.

The conserved peptide synthetase is required for thaxtomin A production by S. acidiscabies 84.104

A 4.9 kb PstI fragment present in all three of the S. acidiscabies cosmids, which hybridized with the S. turgidiscabies probe, was cloned into pUC19, and this plasmid was designated pFGH201. Preliminary nucleotide sequence and restriction mapping data revealed a 1.3 kb NcoI fragment within the 4.9 kb PstI insert, which itself comprised an incomplete open reading frame (ORF) with homology to peptide synthetases (Figs 4 and 6). The peptide synthetase gene disruption construct pFGH303 was created by deleting the internal NcoI fragment of pFGH201 and replacing it with the Tn5-based Kmr gene cassette. This marker exchange construct was cloned into the streptomycete gene disruption vector pOJ260, which lacks a Streptomyces plasmid origin of replication, but carries the ColE1 replicon as well as the broad-host-range RK2 plasmid origin of conjugal transfer (Bierman et al., 1992). pFGH303 was used to transform the mobilizing strain E. coli S17-1. S17-1(pFGH303) donor cells were then used in conjugal matings with S. acidiscabies 84.104 recipients. S. acidiscabies 84.104 peptide synthetase deletion mutants were selected for Kmr and screened for Ams. Two such mutants, 303.1 and 303.2, were chosen for further study. PstI-digested genomic DNA extracted from the parent strain and the two transconjugants was immobilized on hybridization membranes and used for Southern hybridization analysis with radiolabelled plasmid pFGH201 as well as the 2.2 kb HindIII Kmr cassette. The results shown in Fig. 4 demonstrate that, although DNA from the parent strain contains the 4.9 kb PstI fragment, the mutants 303.1 and 303.2 contain copies of the Kmr gene integrated within the parental PstI DNA fragment. Morphological characteristics as well as growth of the mutant strains on solid medium and in liquid culture [≈ 2.0 g dry weight l−1 oatmeal broth, (OMB, thaxtomin production medium)] were indistinguishable from those of the parent strain (data not shown).

Figure 4.

Southern hybridization analysis of PstI-digested genomic DNA of parent strain S. acidiscabies 84.104 and mutants 303.1 and 303.2 using radiolabelled pFGH201 (PS probe) and Kmr gene probes, demonstrating disruption of peptide synthetase through insertion of the Kmr gene in txtA (left) and graphic interpretation of a recombination-mediated disruption event (right). Relevant restriction endonuclease recognition sites are as follows: P, PstI; N, NcoI; N/H, NcoI–HindIII junction created in the construction of pFGH303.

S. acidiscabies 84.104 and mutant derivatives 303.1 and 303.2 were cultured in OMB medium to evaluate thaxtomin production. The results of the high-performance liquid chromatography (HPLC) analyses are shown in Fig. 5. The presence of thaxtomin A in crude ethyl acetate extracts of the parent strain can be seen, whereas thaxtomin A production was abolished in the peptide synthetase gene disruption mutants.

Figure 5.

HPLC chromatograms of purified thaxtomin A and 100 µg each of culture filtrate extracts of parent strain S. acidiscabies 84.104 and mutant strains 303.1 and 303.2 grown in thaxtomin production medium.

Organization of the txtAB (thaxtomin) synthetase genes

Complete nucleotide sequence analysis of cosmid subclones covering the synthetase region was carried out, and the genetic organization of this region is shown in Fig. 6. Two peptide synthetase genes, designated txtA and txtB (thaxtomin synthetase), were identified. The 4377 bp txtA and 4518 bp txtB genes exhibit high GC content and codon bias typical of streptomycete ORFs (≈ 71% GC for both txtA and txtB; not shown). The most probable start codons for the deduced txtA and txtB gene products were identified based on the occurrence of adjacent upstream sequences resembling Shine–Dalgarno boxes. The absence of any intergenic sequences between txtA and txtB, along with the location of the probable txtB ribosome-binding sequence within the 3′ terminus of txtA (Fig. 6), suggests that the expression of txtA and txtB is translationally coupled.

Figure 6.

Genetic organization of the txtAB region identified in S. acidiscabies 84.104. Relevant cosmid subclones are shown as bracketed solid lines at the top, including NcoI sites (N) in plasmid pFGH201 used for the construction of gene disruption construct pFGH303. The conserved 3.5 kb KpnI (K) fragment identified through ‘pairwise’ Southern hybridization is shown as a bracketed dashed line. Other restriction sites are shown on the partial restriction map below plasmid subclones: P, PstI; S, SacI; R, EcoRI. Modules within txtA and txtB are shown below the restriction map indicating proposed enzymatic domains: A, adenylation (solid); M, N-methyltransferase (horizontally hatched); T, thiolation (stippled); C, condensation (vertically hatched). Proposed Shine–Dalgarno sequence within the 3′ terminus of txtA is boxed; the txtA stop codon is overlined with an asterisk; the probable txtB start codon is doubly underlined.

The deduced amino acid sequences of TxtA and TxtB comprise enzymes of 1458 amino acids (157 274 Da) and 1505 amino acids (162 590 Da), respectively, and reveal requisite adenylation (A), thiolation (T) and condensation (C) domains (Figs 6 and 7). Additionally, both synthetases contain S-adenosylmethioninedependent N-methyltransferase (M) domains, integrated between the A and T domains, similar to other fungal and streptomycete peptide synthetases that catalyse the production of peptides containing N-methylated constituents (Haese et al., 1993; Weber et al., 1994; De Crecy-Lagard et al., 1997a, b).

Figure 7.

Alignment of proposed TxtA and TxtB enzymatic domains with those of other peptide synthetases activating known substrates (given above each). CssA, cyclosporin synthetase and substrates l-methylleucine (l-MeLEU); l-methylvaline (l-MeVAL); (4R)-4-[(E)-2-butenyl]-4-methyl-l-threonine (l-BMT); SnbDE, pristinamycin synthetase and substrates l-proline (l-PRO); l-di(mono)methylparaaminophenylalanine (l-DMPAPA); 4-oxo-l-pipecolic acid (l-PIP); l-phenylglycine (l-PG); ACVS, ACV synthetase and substrates l-δ-(α-aminoadipate) (l-αAA); l-cysteine (l-CYS); l-valine (l-VAL) (note ACVS does not carry N-methyltransferase domains). Enzymatic domains are shown at the right.

Amino acid alignments of the deduced TxtA and TxtB sequences exhibited ≈ 44% identity (647 out of 1460 amino acids) and ≈ 56% similarity (831 out of 1460 amino acids) to one another. The deduced TxtA and TxtB sequences also exhibited high similarities to the l-4-di(mono)methylparaaminophenylalanine (l-DMPAPA) incorporation domains of the cyclohexadepsipeptide pristinamycin SnbDE synthetase of Streptomyces pristinaespiralis (≈ 42% identity over ≈ 1460 amino acids for both TxtA and TxtB) (De Crecy-Lagard et al., 1997a); similar results were found with the related l-Phe incorporation domains of the Streptomyces virginiae virginiamycin SnbDE synthetase (De Crecy-Lagard et al., 1997b). High similarities were also identified with the lN-methylvaline (l-MeVAL) incorporation domains of the Tolypocladium inflatum cyclosporin synthetase, CssA (29% identity over 1450 amino acids) (Weber et al., 1994), and the A, M and T domains of the calcium-dependent antibiotic (CDA) peptide synthetases I, II and III of Streptomyces coelicolor (≈ 51% identity over 472 amino acids) (Kieser et al., 1992; Chong et al., 1998).

The most notable feature of the deduced amino acid sequences of TxtA and TxtB is the organization of the conserved domains. As shown in Figs 6 and 7, the C domains are located at the C-termini of TxtA and TxtB. This is in contrast to the N-terminal localization of C domains characteristic of those peptide synthetases exhibiting the highest similarities to TxtA and TxtB, based on extensive analyses of available sequences of peptide synthetases and peptide synthetase genes found in protein and nucleotide databases. This ‘A(M)TC’-type domain organization resembles that of the ACV synthetases (ACVS) of various Streptomyces, Penicillium and Aspergillus spp. (Banko et al., 1987; van Liempt et al., 1989; Smith et al., 1990; Yu et al., 1994). ACVS catalyses the condensation of three amino acids into the penicillin and cephalosporin precursor tripeptide l-δ-(α-aminoadipoyl)-l-cysteinyl-d-valine (ACV) (Schofield et al., 1997). This A(M)TC domain organization was also described recently for the peptide synthetase portion of the lipopeptide mycosubtilin synthetase of Bacillus subtilis (Duitman et al., 1999). The domain organizations of these ‘CA(M)T’- and ‘A(M)TC’-type synthetases, exemplified by SnbDE/CssA and ACVS/TxtA/TxtB, respectively, is illustrated in Fig. 7. Although the deduced amino acid similarities between TxtA/TxtB and ACVS were lower (≈ 30% identity) than the similarities between TxtA/TxtB and SnbDE, the alignments of TxtA/TxtB and ACVS show that they share the same organization of functional domains. In Fig. 7, the A, M and T domains of TxtA and TxtB are aligned with the A, M and T domains of the l-di(mono)methylparaaminophenylalanine (l-DMPAPA) and l-methylvaline (l-MeVAL) incorporation domains of SnbDE and CssA, respectively, whereas the TxtA and TxtB C domains are aligned with those of the C-terminal adjacent 4-oxo-l-pipecolic acid (l-PIP) and (4R)-4-[(E)-2-butenyl]-4-methyl-l-threonine (l-BMT) incorporation domains of SnbDE and CssA respectively. Conversely, each of the three substrate incorporation regions of ACVS exhibit C domains localized to their respective C-termini, similar to that of TxtA and TxtB (N-methyltransferase domains are absent in ACVS as ACV does not contain N-methylated constituents).

S. acidiscabies pathogenicity is thaxtomin dependent

The results of the whole potato tuber assays are shown in Fig. 8. Within 72–96 h after inoculation, brown–black necrotic lesions were visible on the surfaces of tubers treated with the parent strain, accompanied by hyphal growth in and around the lesions. Tubers treated with the mutant strains 303.1 (not shown) and 303.2, however, were devoid of lesions on the tuber periderm surfaces, although some mycelial development was evident. The overall appearance of the tubers inoculated with the mutant strains did not differ from that of the negative control tubers dipped in OMB alone. SACOS1, carrying the thaxtomin synthetase genes, was conjugated into txtA mutant 303.2, thereby restoring the pathogenicity of 303.2 to wild-type levels. Similarly, thaxtomin A production was restored to the complemented mutant, as illustrated in Fig. 9.

Figure 8.

Whole potato tuber pathogenicity assay scored 96 h after inoculation. Tubers were inoculated with: OMB medium (A); OMB culture of parent strain S. acidiscabies 84.104 (B); OMB culture of thaxtomin synthetase mutant 303.2 (C); and OMB culture of thaxtomin synthetase mutant 303.2 carrying thaxtomin synthetase cosmid SACOS1 (D).

Figure 9.

Silica gel TLC of purified thaxtomin A (lanes 1 and 6); ethyl acetate extract of S. acidiscabies 84.104 (lanes 2 and 5); ethyl acetate extract of 303.2 (lane 3); ethyl acetate extract of 303.2(SACOS1) (lane 4).


Progress in understanding the molecular genetic basis of pathogenicity among Gram-positive plant pathogenic bacteria has lagged behind that of Gram-negative pathogens. This has, at least in part, been caused by a lack of efficient methods and tools for their genetic manipulation. The finding that recombinant S. acidiscabies strains could be created through conjugation-mediated gene transfer, along with the use of peptide synthetase-based oligonucleotide primers to amplify putative fragments of peptide synthetase modules from four representative thaxtomin-producing plant pathogenic species, allowed the direct identification of a peptide synthetase conserved among three of these species. Targeted disruption of the first peptide synthetase gene, txtA, in S. acidiscabies and the absence of thaxtomins in extracts of these mutants confirmed the role of this conserved peptide synthetase in thaxtomin biosynthesis. The creation of thaxtomin synthetase null mutants arising from targeted gene disruption of txtA has allowed us to determine the role played by thaxtomins in the pathogenicity of this organism. The results presented here demonstrate that thaxtomin production is required for the infection of intact host tuber tissue by S. acidiscabies.

The genetic organization of the deduced peptide synthetase gene products is depicted in Fig. 6 and is consistent with the chemical structures of thaxtomin congeners identified in extracts of infected plant tissue, as well as those found in in vitro production medium (King et al., 1989; Babcock et al., 1993). Two distinct acyladenylation, thioesterification and condensation domains are present, as well as two integral N-methyltransferase domains, corresponding to the condensation of two N-methylated substrate amino acids. The two genes encoding the proposed peptide synthetases have been designated txtA and txtB, and nucleotide sequence data suggest that their expression is translationally coupled, as seen with the snbC/snbDE pristinamycin synthetase genes of S. pristinaespiralis (De Crecy-Lagard et al., 1997a) and calcium-dependent antibiotic (CDA) synthetases of S. coelicolor (Kieser et al., 1992; Chong et al., 1998). This type of co-ordinated gene expression could provide stoichiometric amounts of TxtA and TxtB, which we propose would form a heterodimeric multifunctional complex, post-translationally modified through the covalent attachment of the 4′-phosphopantetheinyl groups to conserved serine residues (not shown) within the respective thiolation domains. The amino acid substrates for the proposed TxtA and TxtB enzymes have not been identified, although the high overall amino acid identities between the proposed enzymes probably reflects similarities between the two aromatic amino acid substrates.

It is noteworthy that neither the S. scabies- nor the S. acidiscabies-derived probes significantly cross-hybridized with genomic DNA from the other species (Fig. 2), suggesting that nucleotide sequences corresponding to the A2 and A3 motifs of these putative ‘species-specific’ peptide synthetases exhibited greater heteroduplex stability with the PS1F and PS1R oligonucleotide primers under the PCR amplification conditions used than the sequences corresponding to the same motifs of txtB found using the S. turgidiscabies-derived probe and were thus preferentially amplified over the corresponding fragment found within txtB (Fig. 6). Indeed, using degenerate oligonucleotide PCR primers and S. acidiscabies genomic DNA templates, we have identified an as yet uncharacterized peptide synthetase from S. acidiscabies cosmid libraries; Southern hybridization data suggest that this synthetase is not present in the other plant pathogenic species (F. G. Healy and R. Loria, unpublished).

The lack of hybridization between the thaxtomin synthetase probe and S. ipomoeae genomic DNA is interesting in several respects. Of the four described thaxtomin-producing plant pathogenic Streptomyces species, S. ipomoeae exhibits a different aetiology and host range from the other three species (Person and Martin, 1940; Clark and Matthews, 1987; Clark and Moyer, 1988). Moreover, S. ipomoeae strains, rather than producing thaxtomin A as the primary thaxtomin congener, produce the 15-de-N-methyl derivative thaxtomin C (Fig. 1), and this is not produced in thaxtomin production medium but, rather, is found in extracts of diseased host tissue (King et al., 1994). Further, S. ipomoeae strains do not carry the 54% GC necrogenic factor nec1, which is found in strains of S. scabies, S. acidiscabies and S. turgidiscabies (Bukhalid et al., 1998).

The identification of the pathogenicity-associated IS element IS1629 and its linkage with nec1 and the ORFtnp transposase pseudogene among some S. scabies strains and all S. acidiscabies and S. turgidiscabies strains available supports our current model of unidirectional horizontal pathogenicity gene transfer from S. scabies to S. acidiscabies and S. turgidiscabies (Healy et al., 1999). Although the presence or absence of IS1629 has not been firmly established in S. ipomoeae strains, hybridization data presented here, along with the different production profiles and apparent differences in the physiological control of the biosynthesis of S. ipomoeae-derived thaxtomin structural variants, suggest that the S. ipomoeae thaxtomin synthetase structural genes and mechanism(s) regulating their expression may be fundamentally different from those of the other three species. If this is true, then the S. ipomoeae thaxtomin synthetase may not have been acquired through the same proposed horizontal transfer mechanism as S. scabies, S. acidiscabies and S. turgidiscabies. Moreover, the conservation of the ‘txtA–txtB’ fragment (Fig. 3) at the level of high-stringency hybridization supports our model for the recent emergence of S. acidiscabies and S. turgidiscabies plant pathogens as the result of horizontal gene transfer phenomena (Bukhalid et al., 1998; Healy et al., 1999). Future work directed towards the physical analysis of streptomycete plant pathogen genomes will allow us to explore further the structural, mechanistic and evolutionary aspects of horizontal gene transfer among Gram-positive plant pathogens. Further biochemical and genetic analysis of the txt region may also be useful in the development of rational strategies directed towards engineering thaxtomin analogues, which could aid in probing the requirements for its phytotoxic activity.

Experimental procedures

Bacterial growth conditions, genomic DNA extraction and PCR amplification

S. scabies 84.34, S. acidiscabies 84.104, S. turgidiscabies 94.09 (HiC-13) and S. ipomoeae 81.45 were grown and genomic DNA extracted as described previously (Rao et al., 1987; Healy et al., 1999). High-GC biased degenerate oligonucleotides corresponding to peptide synthetase A2 and A3 domains (using the nomenclature of Marahiel et al., 1997), with terminal EcoRI adaptors, were provided by Integrated DNA Technologies; the oligonucleotide sequences were as follows: PS1F, 5′-GGAATTCCTSAAGDCSGGCG GIGCCTACGTSCC; PS1R, 5′-GGAATTCCCTTSGGCIKSC CGGTSGISCCGGAGG (where D = A, G or T; K = G or T; S = G or C). ‘Touchdown’ PCR amplification (Don et al., 1991) conditions were established for optimal amplification of genomic DNA templates using degenerate oligonucleotides: 95°C for 2.5 min, one cycle; 95°C for 30 s, 45°C to 41.6°C (temperature decrement of ≈ 0.1°C/cycle) for 45 s, 72°C for 1 min, 35 cycles; 95°C for 30 s, 40°C for 30 s, 72°C for 1 min, six cycles; 72°C for 6 min, one cycle. PCR amplification products from each of the four genomic templates were size fractionated by electrophoresis through 1.5% agarose gels, and product(s) that would be predicted to have arisen from the amplification of portions of peptide synthetase acyladenylation modules based on their expected size (≈ 300–400 bp between the A2 and A3 conserved regions) were excised from gels and purified as described previously (Healy et al., 1999).

Southern hybridization and nucleotide sequencing

Purified putative peptide synthetase fragments obtained from PCR amplification of each of the four genomic templates were radiolabelled and used as probes in high-stringency ‘pairwise’ Southern hybridizations as follows. Four separate hybridization membranes were prepared for use with the four radiolabelled probes. Equal amounts of KpnI-digested genomic DNA samples from the four representative pathogens were electrophoresed through a single 0.8% agarose gel (16 gel lanes) and transferred to the membranes. Therefore, although electrophoresis conditions for each membrane were essentially identical, the possibility of residual background isotope signals resulting from membrane washing and reprobing was eliminated. High-stringency hybridization conditions were similar to those described previously (Bukhalid and Loria, 1997).

A radiolabelled probe derived from amplification of S. turgidiscabies 94.09 (HiC-13) genomic DNA was used to identify hybridizing clones from a S. acidiscabies 84.104 cosmid library constructed in cosmid vector pOJ446 (Bierman et al., 1992) using the Gigapack III Gold Packaging Extract kit (Stratagene) and standard methods (Sambrook et al., 1989). The hybridizing cosmids were subcloned in pUC19, and nucleotide sequencing was performed by the BioResource Center, Cornell University, using an ABI 377XL automated sequencer and M13 forward/reverse and custom oligonucleotide primers provided by the BioResource Center or Integrated DNA Technologies. The txtAB nucleotide sequence has been submitted to the GenBank database under accession number AF255732.

Southern hybridizations using radiolabelled pFGH201 and kanamycin resistance (Kmr) gene cassette probes against PstI-digested genomic DNA of S. acidiscabies parent strain 84.104 and mutants 303.1 and 303.2 were performed as described previously (Bukhalid and Loria, 1997).

Gene disruption of a S. acidiscabies 84.104 peptide synthetase

Plasmid pFGH201 [pUC19 carrying a 4.9 kb PstI fragment from S. acidiscabies cosmids SACOS1, SACOS5 and SACOS9, which hybridized with the S. turgidiscabies 94.09 (HiC-13) peptide synthetase probe] was digested with NcoI, releasing a 1.3 kb fragment within the insert. The ends of the resulting linearized plasmid were filled in with T4 DNA polymerase according to the manufacturer's recommendations (New England Biolabs). A 2.2 kb HindIII fragment containing the Kmr gene cassette from mini-Tn5-Km (kindly provided by Dr K. N. Timmis) was similarly treated and ligated with the NcoI-deleted pFGH201 plasmid using T4 DNA ligase (New England Biolabs). The PstI insert of pFGH201 containing the Kmr gene cassette in place of the NcoI fragment was excised and ligated into the PstI site of the streptomycete gene disruption vector pOJ260 (Bierman et al., 1992). The resulting construct, pFGH303 (ΔtxtA-Kmr), was used to transform E. coli S17-1; transformants were then used for RK2/RP4-mediated conjugal transfer of the plasmid from S17-1 donors to S. acidiscabies 84.104 recipients (Bierman et al., 1992).

S. acidiscabies 84.104 was grown in tryptic soy broth (TSB) medium, and transconjugants were selected on AS1 medium containing apramycin sulphate (Am, 25 µg ml−1) and nalidixic acid (40 µg ml−1; both from Sigma Chemical) as described previously (Bierman et al., 1992). S. acidiscabies double recombinants were identified by growing Amr transconjugants non-selectively both in TSB (to mid-log phase) and, subsequently, on ISP2 solid medium (Bukhalid et al., 1998), from which spores were harvested. Spores were then plated on ISP2 medium containing Km (25 µg ml−1); colonies were replica plated to ISP2 medium containing Am (25 µg ml−1) to screen for mutants resulting from double cross-over events (Kmr Ams). E. coli DH5α and E. coli S17-1 (containing a chromosomally integrated copy of the conjugal transfer functions of the broad-host-range plasmid RP4; Simon et al., 1983) strains were grown in LB medium or on LB agar, supplemented with Km (50 µg ml−1, Sigma Chemical), Am (100 µg ml−1) or ampicillin (100 µg ml−1) as needed.

HPLC and thin-layer chromatography (TLC) analysis of thaxtomin production

The S. acidiscabies 84.104 parent strain and the thaxtomin synthetase gene disruption mutants 303.1 and 303.2 (ΔtxtA-Kmr) were grown in thaxtomin production medium as described previously (Loria et al., 1995) to evaluate in vitro thaxtomin A production. Culture filtrates were extracted with ethyl acetate; this fraction was dried and the resulting material was dissolved in methanol and analysed via HPLC using a modification of the method of King et al. (1992). Samples were injected onto a Phenomenex Prodigy 5 µm ODS 3 column (4.6 × 250 mm) and eluted at a flow rate of 1 ml min−1 with the following solvent gradient: CH3CN (25%):H2O (75%) to CH3CN (100%), 10 min; CH3CN (100%), 5 min. Thaxtomin A was monitored by UV absorbance at 225 nm. Silica gel TLC analyses were performed as described previously (Loria et al., 1995).

Whole potato tuber pathogenicity assays

Approximately 5 × 104 cfu ml−1 parent strain S. acidiscabies 84.104 and mutant strains 303.1 and 303.2 were inoculated into 250 ml of OMB medium (Loria et al., 1995) in 1 l Erlenmeyer flasks and grown for 4–6 days at 28°C. Samples (30 ml) of the 250 ml culture were centrifuged at 3000 r.p.m. for 10 min and washed three times with sterile distilled water. Mycelial pellets, resuspended in 30 ml of sterile water, were used as inoculum. Chippewa potato tubers were planted in pots and grown in the greenhouse for 3–4 weeks. Plants were then transferred to a growth chamber set for a 9 h photoperiod, ≈ 21°C during the day and ≈ 15°C at night. After 12–15 days, leaf bud cuttings were transplanted to trays of moist vermiculite and placed in a mist chamber for 12–15 days for tuber production. Tubers of varying sizes recovered from trays were washed thoroughly in distilled water and immersed in inoculum suspensions for 10 min. Tubers were then removed and placed in plastic boxes lined with moist paper towels, covered and incubated at ≈ 25°C. Pathogenicity reactions were scored after 72–96 h. Twelve replicates were included per treatment, and assays were repeated once.


The authors would like to thank Dr Timmis for providing mini-Tn5, Kent Loeffler for the preparation of figures, and David W. Bauer for providing S. acidiscabies 84.104 cosmid libraries. This work was supported by USDA NRI grant 97-35303-4487.


  1. †Present address: Department of Microbiology and Cell Science, Box 110700, Museum Road, University of Florida, Gainesville, FL 32611-0700, USA.