Occurrence of sep Insecticidal Toxin Complex Genes in Serratia spp. and Yersinia frederiksenii

Abstract

Some strains of Serratia entomophila and S. proteamaculans cause amber disease of the grass grub Costelytra zealandica (Coleoptera: Scarabaeidae). Three genes required for virulence, sepABC, are located on a large plasmid, pADAP. Sequence analysis suggests that the sepABC gene cluster may be part of a horizontally mobile region. This study presents evidence for the putative mobility of the sep genes of pADAP. Southern blot analysis showed that orthologues of the sep genes reside on plasmids within S. entomophila, S. liquefaciens, S. proteamaculans, and a plasmid from Yersinia frederiksenii. Three plasmids hybridized to the pADAP sep virulence-associated region but not the pADAP replication and conjugation regions. Subsequent DNA sequence analysis of the Y. frederiksenii sep-like genes, designated tcYF1 and tcYF2, showed that they had 88% and 87% DNA identity to sepA and sepB, respectively. These results indicate that the sep genes are part of a discrete horizontally mobile region.

Amber disease of the New Zealand grass grub Costelytra zealandica (Coleoptera: Scarabaeidae) is caused by some strains of Serratia entomophila and Serratia proteamaculans (Enterobacteriaceae) (11). First identified in 1981 (38), amber disease has a unique pathology. After ingestion of S. entomophila, larvae of C. zealandica cease feeding within 1 to 3 days; clearance of the midgut then occurs, resulting in the characteristic amber coloration of the larvae. The bacteria do not invade the hemocoel until the final stage of the disease (more than 3 weeks after infection), which then results in the death of the larvae (21). Host range testing of S. entomophila against a range of closely related Coleoptera has failed to find any other species that develop symptoms of amber disease, suggesting that the pathogen is highly specific (22).

The disease determinants are located on a plasmid designated pADAP (amber disease-associated plasmid) (11). Hurst et al. (18) identified three genes on pADAP, termed sepABC (for S. entomophila pathogenicity), that are required for the initiation of amber disease. Transposon insertions in sepA, sepB, or sepC completely abolished both gut clearance and cessation of feeding (18). Through the mapping of pADAP and the use of sequences derived from the peripheral region of pADAP subclones, the locations of putative insertion elements and genes involved in the replication and conjugative transfer of pADAP were also determined (17). More recently, an additional pADAP gene cluster encoding antifeeding determinants was described (20).

While amber disease is unique in its pathology, the predicted products of the sep genes show similarity to other insecticidal proteins from bacteria such as Photorhabdus luminescens (2, 3) and Xenorhabdus nematophila (28). These bacteria are symbionts of entomopathogenic nematodes of the families Heterorhabditidae and Steinernematidae. This new family of insecticidal proteins, known as toxin complex (Tc) proteins (3), has been shown to be of high molecular weight and to have strong insecticidal activity toward a large number of insects, including species of Coleoptera, Dictyoptera, Hymenoptera, and Lepidoptera (2). Analysis of the genome sequences of Yersinia pestis C092 (30), Pseudomonas syringae pv. tomato DC3000 (5), P. syringae pv. syringae B728a (gi:28876514), Pseudomonas fluorescens PfO-1 (gi:48732052), and Chromobacterium violaceum ATCC 12472 (39), among others, has also revealed the presence of putative insecticidal tc genes. However, the target insect species against which the products of these genes may act have yet to be identified.

Waterfield et al. (41) and ffrench-Constant et al. (10) suggested that the tc genes of P. luminescens are located on a pathogenicity island (PAI) (defined as a large, chromosomally located segment of horizontally acquired DNA that encodes virulence genes [13]), adjacent to an Asp-tRNA gene. While the origin of most PAIs is unclear, it is possible that some were originally plasmid-located gene clusters that integrated into chromosomal sites and subsequently lost their mobility functions (8).

Further sequencing of pADAP DNA flanking the sep genes provided evidence that the genes may be carried on a discrete mobile DNA element similar to a pathogenicity island. Hurst et al. (19) sequenced the 50-kb region upstream and the 10-kb region downstream of the sepABC genes and noted that a 44.7-kb region was bordered by degenerate 785-nucleotide (nt) direct repeats. This region encompasses the sepABC genes, three genes with similarity to hypothetical genes from Bacillus subtilis, a fimbrial gene cluster, and several residual insertion elements. Hurst et al. (19) speculated that the direct repeats represent the boundaries of a virulence-encoding region that may be horizontally mobile.

Little is known about the diversity of sep-like genes in Serratia strains capable of causing amber disease, since most studies have been conducted on one strain, S. entomophila A1MO2. The present study investigated the occurrence of sep-like genes in soil bacteria isolated from New Zealand pastures, in particular S. entomophila and S. proteamaculans. The aim was to test the hypothesis that the sep-containing region is horizontally mobile separately from pADAP itself. Serratia plasmids were examined for variation, and selected plasmids were probed for the presence of sep-like genes as well as pADAP replication and conjugative transfer regions. In addition, pADAP-related plasmids that did not hybridize with the sep region were further assessed for their ability to hybridize to the regions flanking the sep genes. Selected environmental bacteria were also screened for sep gene orthologues by using DNA hybridization.

MATERIALS AND METHODS

Bacterial strains, plasmids, and media.

The plasmids used in this study are described in Table 1. All Serratia isolates were obtained from the AgResearch (Lincoln, New Zealand) Insect Pathogen culture collection and were collected from 1985 to the present. Most of the strains were originally isolated from soil or C. zealandica larvae throughout New Zealand, while the remaining strains were obtained from P. A. D. Grimont, Pasteur Institute, France. Serratia species were identified using a combination of differential media (Adonitol, DNase, and Itaconate) (29) and API 20E (bioMerieux, France) analysis. Other gram-negative bacteria were identified using API 20E and 20NE analysis. Bacteria were cultured in LB broth or on LB agar (31) and on the Serratia-specific medium caprylate thallous agar (CTA) (33) or 10% TSA (3 g/liter tryptic soy broth [Gibco BRL], 15 g/liter agar). Antibiotics, where used, were 100 μg/ml ampicillin, 30 μg/ml chloramphenicol, and 30 μg/ml tetracycline.

TABLE 1.

Bacterial plasmids used or identified in this study

Plasmid	Description	Reference
p8L	2-kb BamHI fragment resulting from cleavage of the mini-Tn10 insert pBM32-8	16
pADAP	Amber disease-associated plasmid, 155 kb, from S. entomophila strain A1MO2	17
pACΔ8	17-kb deletion derivative resulting from the cleavage of the mini-Tn10 insert pBM32-8 and containing 8 ORFs including sepABC	16
pAY2-4	Arabinose expression vector	32
pBM32-8	23-kb pADAP BamHI fragment in pBR322 containing a mini-Tn10 insertion flanked by BamHI sites	16
pGLA2	11-kb pADAP HindIII fragment in pUC19	12
pGLA3	7-kb HindIII fragment of pADAP in pLAFR3	6
pLAFR3	Tcr; pRK290 with cos lacZ and multicloning site from pUC8	34
pLYF2	15-kb EcoRI fragment of pYF49 in pGLA3	This study
paraA	7,131-bp sepA gene in pAY2-4	M. Hurst, unpublished data
paraB	4,287-bp sepB gene in pAY2-4	Hurst, unpublished
paraC	2,922-bp sepC gene in pAY2-4	Hurst, unpublished
pUB15	15-kb pADAP BamHI fragment in pHSG398 containing conjugation-related genes	17
pUC19	Apr; lacZ multicloning site	42
pHSG398	Cmr; lacZ multicloning site	36
pUH1.8	1.8-kb pADAP HindIII fragment in pUC19 containing partial replication and plasmid partition genes	17
pUH5.6	5.6-kb pADAP HindIII fragment in pUC19	17
pSG348	Plasmid of S. grimesii strain 348 that hybridizes to conjugation and replication genes on pADAP but lacks the virulence region	This study
pSL377	Plasmid of S. liquefaciens strain 377 containing sepABC orthologues and hybridizing to the conjugation and replication genes on pADAP	This study
pSP1137	Plasmid of S. proteamaculans strain 1137 that hybridizes to conjugation and replication genes on pADAP but lacks the virulence region	This study
pSP149	Plasmid of S. proteamaculans strain 149 containing sepABC orthologues and hybridizing to the conjugation and replication genes on pADAP	This study
pSP591	Plasmid of S. proteamaculans strain 591 containing sepABC variants	This study
pYF49	Largest plasmid of Y. frederiksenii isolate 49, containing sepABC variants	This study

Isolation of bacteria from larvae and soil.

Bacteria were isolated from pasture soil (Wakanui silt loam from Canterbury, New Zealand) by dilution plating onto 10% TSA, and non-Serratia isolates were selected by screening on DNase, MacConkey, and CTA agars. In addition, 16 C. zealandica larvae from three different sites in Hawke's Bay, New Zealand, with various phenotypes (healthy, feeding; healthy, nonfeeding; diseased, feeding; diseased, nonfeeding; and dead) were macerated in phosphate buffer and the macerate plated on 10% TSA. Four hundred bacterial isolates (25 per larva) were randomly selected for further analysis.

Plasmid visualization.

Large plasmids were visualized using a method modified from that of Kado and Liu (25). Protein was removed by phenol-chloroform extraction, and a 40-μl aliquot of the aqueous phase was loaded onto an 0.8% Tris-borate-EDTA agarose gel for electrophoresis. The sizes of plasmids from other strains were estimated by comparison with the λ HindIII marker and by their relative migration to pADAP.

Isolation and digestion of plasmid DNA.

Plasmids from Escherichia coli strains were isolated using the alkaline lysis method (31). For isolation of large plasmids, 20-ml overnight cultures of strains of Serratia species and Yersinia frederiksenii were used in conjunction with a QIAGEN (Hilden, Germany) plasmid midi kit. Purified plasmid DNA was resuspended in 80 to 120 μl of sterile distilled H₂O, of which 40 μl was used for restriction digestion. Total genomic DNA was isolated from a 1.5-ml volume of an overnight culture using a Wizard genomic DNA purification kit (Promega, Madison, Wis.). DNA was typically resuspended in 600 μl of sterile distilled H₂O. The method of Heery et al. (15) was used to elute DNA fragments from agarose gels.

Southern hybridization.

Standard methods were used for Southern hybridizations (31). ³²P-labeled probes were prepared by random priming (RTS RadPrime DNA labeling system; Gibco BRL, MD). DNA blots were hybridized for 24 h at 65°C; washed at 65°C sequentially in 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)-0.1% SDS, 1× SSC-0.1% SDS, and 0.1× SSC-0.1% SDS; and then exposed to film at −80°C.

Probing of pADAP regions against other strains.

Three pADAP subclones were used as probes (Table 1; Fig. 1): (i) pACΔ8, which encompasses the sepABC virulence-associated region; (ii) pUB15, which encodes 16 open reading frames (ORFs) with high similarity to components of the type IV pilus and other conjugative transfer components of the E. coli plasmid R64 (19); and (iii) pUH1.8, the cloned DNA of which has translated similarity to two proteins involved in plasmid replication and partitioning and is the predicted replication region of pADAP (unpublished data) from several members of the Enterobacteriaceae (17). Gel-eluted gene-specific probes for sepA, sepB, or sepC were derived from the plasmid clones paraA, paraB, and paraC, respectively (Table 1).

FIG. 1.

BamHI (inner circle), EcoRI (middle circle), and HindIII (outer circle) restriction enzyme map of pADAP (from the work of Hurst and Glare [17]). Putative conjugation (pUB15), replication (pUH1.8), and virulence-encoding (pACΔ8) segments used as probes are indicated. p8L and pACΔ8 are BamHI fragments derived from cleavage of the mini-Tn10 insert pBM32-8. Solid regions represent clones containing the fragments used as probes to identify plasmids orthologous to pADAP.

Cloning of plasmid DNA encoding sep-like genes.

Total plasmid DNA obtained using a QIAGEN plasmid midi kit was digested with either BamHI, EcoRI, HindIII, or PstI. Fragments smaller than 15 kb were ligated into pHSG398 (36) and transformed into E. coli DH10B (14). Fragments greater than 15 kb were ligated into pLAFR3 (34) and packaged, using a GigapackIIIXL packaging extract (Stratagene, La Jolla, Calif.), into E. coli XL1-Blue MRA. Cloned fragments were sequenced on both strands using universal primers and primer walking.

PCR amplification and sequencing.

PCR was carried out in a Perkin-Elmer model 480 thermal cycler, and PCR products were purified using a QIAquick kit (QIAGEN). For bacterial identification, a portion of the 16S rRNA gene was amplified using primers 16a (5′-AGAGTTTGATCCTGGCTC) and 16b (5′-TACGGYTACCTTGTTACGACTT) (26) and sequenced.

PCR primers were designed to conserved regions of the sepA, sepB, and sepC genes, based on protein similarities (18), to amplify analogous DNA regions from pSP591. PCR was undertaken with the following primer pairs: pSP591sepAIF (5′-GGATAATCAGGTGTCAGCCAAG) and pSP591sepAIR (5′-GGGCATCATTGAGTCTTGGAT), pSP591sepAIIF (5′-CGGAGCCAATGCCCTCTA) and pSP591sepAIIR (5′-AGTAGCCTTTGAGCACCTCGT), pSP591sepBF (5′-TTTGGCTGATGGAGTCCTCG) and pSP591sepBR (5′-CGTCCCGTCCGACTCATAAG), and pSP591sepCF (5′-CGGCGTGGTAACCTCGTA) and pSP591sepCR (5′GGTGCTTCTGGTGACCTCCT). Thirty cycles of 1 min at 94°C, 1 min at 50°C, and 2 min at 72°C were used.

DNA was sequenced by automated sequencing using an Applied Biosystems 377 autosequencer (University of Waikato DNA sequencing facility; http://sequence.bio.waikato.ac.nz). Sequences were assembled using SEQMAN (DNAStar Inc., Madison, Wis.). Databases at the National Center for Biotechnology Information were searched using BlastN, BlastX, and BlastP (1). Searches for open reading frames were initiated using EDITSEQ (DNAStar Inc., Madison, Wis.).

Bioassay of bacterial strains against C. zealandica larvae.

The virulence of a given strain toward C. zealandica was determined using the standard bioassay described by Jackson and Saville (24). Healthy second- or third-instar larvae, collected from the field, were individually fed cubes (∼3 mm³) of carrot that had been rolled in bacterial colonies grown overnight on solid medium, resulting in ∼10⁷ CFU/carrot cube. Twelve larvae were used for each treatment, and the experiment was carried out in a randomized block design with two blocks of six larvae. Known virulent and avirulent strains were included in all bioassays. Larvae were fed treated carrot at day 1 and were transferred to fresh trays containing untreated carrot at days 6 and 9. The occurrence of gut clearance and cessation of feeding (23) were monitored at days 3, 6, and 12. Strains were considered virulent if at least nine (75%) of the treated larvae showed disease symptoms by day 12 and if at least nine of the larvae from the negative control remained healthy.

Nucleotide sequence accession numbers.

The sequences obtained in this study have been deposited in GenBank as follows. The tcYF1, tcYF2, and tcYF3 sequences have been assigned accession numbers AY220302, AY220492, and AY220493, respectively. The 16S rRNA gene of Y. frederiksenii strain 49 has been assigned accession number DQ486901. The S. proteamaculans sepABC-like sequences have been assigned accession numbers AY297168 (sepA), AY297169 (sepA), AY297170 (sepB), and AY297171 (sepC).

RESULTS

Plasmids of Serratia species.

A collection of Serratia spp. (predominantly isolated from scarab larvae or soils in New Zealand) were screened for the presence of large plasmids, sep-like genes, and virulence toward grass grub larvae by a standard bioassay. The plasmid contents of 94 strains from 10 Serratia spp. were examined. Analysis showed that of the 21 S. entomophila strains assessed, all strains that contained plasmids were virulent, with the exception of 1 strain (440) that was avirulent (Table 2). Analysis of 27 S. proteamaculans strains showed that 7 were plasmid free and 8 strains containing one or more large plasmids were avirulent, while the other 12 strains that contained one or more large plasmids were virulent (Table 2). All strains of species other than S. entomophila and S. proteamaculans were avirulent toward grass grub larvae. No plasmids were visualized in Serratia plymuthica, Serratia fonticola, Serratia rubidaea, or Serratia odorifera strains.

TABLE 2.

Plasmid content and virulence toward C. zealandica larvae of S. entomophila, S. proteamaculans, S. liquefaciens, and S. grimesii strains used in this study

Strain tested	Phenotypea	Estimated plasmid size(s)b (kb)	Plasmid hybridizationc using a cloned fragment of the following probe:
			pACΔ8 (virulence)	pUB15 (conjugation)	pUH1.8 (replication)
S. entomophila
167, 168, 190, 291, 292, 626, 203, 207, 213, 210, 307, 402, 465, 1100, 1347, 369, 480, 673	Virulent	155	+	+	+
176	Virulent	155, 120	+, −	+, −	+, −
345	Virulent	155, 140, 120	+, −, −	+, −, −	+, −, −
398	Virulent	155, 140	+, −	+, −	+, −
192, 204, 220,d 314, 328, 329, 396, 400, 481, 625	Avirulent	−	Not applicable
440	Avirulent	120	−	−	−
S. proteamaculans
341, 145, 299, 301, 142, 387, 1475, 146, 1129	Virulent	155	+	+	+
1048, 1071, 376	Virulent	140	+	+	+
143	Avirulent	140, 120	+, −	+, −	+, −
200	Avirulente	160	−	+	+
336	Avirulente	155	+	+	+
187, 193, 202, 332, 387, 462, 356	Avirulent	−	Not applicable
163, 1137	Avirulent	140	−	+	+
149	Avirulent	155, 120	+, −	+, −	+, −
495	Avirulent	160, 155	−, −	−, −	−, −
591	Avirulent	160	+	−	−
S. grimesii 348	Avirulent	155	−	+	+
S. liquefaciens 377	Avirulent	150	+	+	+
Yersinia frederiksenii 49	Avirulent	3, 5, 80, 210	−, −, −, +	−, −, −, +	−, −, −, −

^aAs defined by Jackson and Saville (24). Strains were considered virulent if at least 75% of the treated larvae became diseased and if 75% of the negative-control larvae remained healthy.

^bEstimated by comparison with pADAP, which is 155 kb (19). −, no plasmid was detected.

^c+, hybridization; −, no hybridization.

^dOriginally from P. A. D. Grimont, Institute Pasteur, France.

^eCaused an antifeeding effect but not gut clearance in bioassays.

Occurrence of pADAP gene orthologues in plasmids of Serratia spp.

Plasmid-bearing Serratia strains were assessed for components of pADAP-derived sequences (Fig. 1) by Southern blot analysis. The results showed that among the 21 S. entomophila strains assessed, the plasmids of 1 avirulent strain did not hybridize to any of the pADAP-derived probes (Table 2). Analysis of other Serratia strains showed that an avirulent Serratia liquefaciens strain, 377, that contained the plasmid (pSL377) hybridized to all of the three regions probed (Table 2). The avirulent S. proteamaculans strains 1137 and 163 and Serratia grimesii strain 348 contained plasmids (Table 1) that hybridized to pUB15 and pUH1.8 but not pACΔ8, indicating that these plasmids were from the same family as pADAP but lacked the sepABC region. In contrast, the avirulent plasmid of S. proteamaculans strain 591 harbored a plasmid (pSP591) larger than pADAP that hybridized to pACΔ8 but not pUB15 or pUH1.8, suggesting that DNA similar to at least part of the virulence region resided on the plasmid, which was unrelated to pADAP. For this reason the DNA content of the sep-like region of pSP591 was assessed. PCR products were generated to a component of each of the three sep genes, and the amplicons were sequenced. The results showed that relative to the DNA of the pADAP virulence region sequence (accession number AF135182), the DNA sequence of the 483-bp sepAI amplicon was 100% identical to sepA between nucleotides 52272 and 52627, while the amplicon obtained using sepAII primers (577 bp) was 99% identical to sepA between nucleotides 54718 and 54218. The 523-bp PCR product obtained using sepBI primers was 96% identical to sepB between nucleotides 57065 and 57561, while that obtained using the sepCI primers (529-bp amplicon) was 95% identical to sepC between nucleotides 62288 and 62777. These results indicated that the sepABC orthologues of S. proteamaculans pSP591 were very similar to their pADAP counterparts.

To assess whether the avirulence phenotype of Serratia strains containing plasmids pSP149, pSP591, and pSL377 (Table 2), which probed positive with the pACΔ8 DNA probe, was a result of absence of one or more of the sep genes, the plasmids of these strains were probed with gene-specific probes for sepA, sepB, or sepC. All three plasmids hybridized to sepA, sepB, and sepC, although their hybridization profiles to restriction digests differed from those of pADAP (data not shown).

Genetic variation within plasmids of virulent strains of S. entomophila and S. proteamaculans.

The DraI restriction enzyme profiles of eight S. entomophila and nine S. proteamaculans strains from around New Zealand were examined for variation in their pADAP-like plasmids. Seven of the eight S. entomophila plasmids had DraI profiles identical to that of pADAP, while the eighth strain, designated 626, contained an additional ∼4.2-kb fragment and the loss of one member of a doublet at 4.0 kb (Fig. 2). All strains exhibited similar hybridization profiles when probed with the fragments of pACΔ8, pUB15, or pUH1.8. In contrast, analysis of the DraI profiles of nine virulence-encoding plasmids of S. proteamaculans showed seven distinct DraI restriction profiles (data not shown). Six unique hybridization profiles were obtained when the plasmids were probed with the fragments of pACΔ8, four when they were probed with pUB15, and three when they were probed with pUH1.8 (data not shown).

FIG. 2.

Comparison of the DraI plasmid restriction profiles of plasmids pADAP (lane 2) and p626 (lane 3) from virulent strains of Serratia entomophila and electrophoretic profiles of undigested S. entomophila strain 190 pADAP (lane 4) and Y. frederiksenii strain 49 plasmid (lane 5) obtained by plasmid visualization. Plasmids were visualized by the method of Kado and Liu (25). The arrow indicates the position of the 4.2-kb DraI fragment. The asterisk indicates the pADAP doublet. Sizes of HindIII-cut λ fragments (lane 1) are given on the left (in kilobases).

To assess differences internal to the sep virulence-encoding region, gene-specific probes for sepA, sepB, or sepC were used to probe the DraI-digested plasmids. The S. entomophila plasmids pADAP and p626, and the S. proteamaculans plasmid p142, exhibited identical hybridization profiles when probed with each of the three sep genes. The remaining six S. proteamaculans plasmids exhibited different hybridization patterns with sepA, but no variation was detected within the sepB and sepC orthologues of these six plasmids, which were all located on single large DraI fragments (data not shown).

Screening of bacteria from C. zealandica larvae for the presence of sep orthologues.

Biochemical testing of the 400 isolates from grass grub larvae found that 197 did not belong to the genus Serratia and that only 9% of these non-Serratia isolates hybridized strongly to the cloned insert of pACΔ8. Most of these isolates were from the macerate of a C. zealandica larva that displayed a “diseased-but-feeding” phenotype. Two of the 24 visually identical isolates that hybridized to the cloned insert of pACΔ8 were randomly selected and identified using a commercial identification kit (API20E; bioMerieux, France); both were identified as Yersinia spp. Despite being isolated from larvae visually assessed as “diseased but feeding,” the Yersinia sp. isolates did not cause disease when tested by a standard bioassay. Sequencing of a portion of the 16S rRNA gene identified both isolates as Y. frederiksenii (accession number DQ486901). One isolate, designated YF49, was found by the method of Kado and Lui (25) to contain four plasmids. Southern blot analysis undertaken using the three pADAP-based probes showed that, of the four plasmids, only the largest (named pYF49) hybridized to the insert of pACΔ8, and none of the plasmids hybridized to pUB15 or pUH1.8. A 15-kb EcoRI fragment of pYF49 was cloned into pLAFR3 to give pLYF2 (Table 1), which hybridized to gene-specific probes for sepA and sepB but not to sepC. Cloning of the adjacent EcoRI fragments of pYF49 into pHSG398 facilitated the identification of the DNA of the full complement of sepABC orthologues.

Translation of the resultant DNA sequence identified three potential ORFs that had high identity at both the DNA and protein levels to the Sep virulence-associated proteins. The ORFs were designated tcYF1 (similar to sepA), tcYF2 (sepB), and tcYF3 (sepC), for toxin complex of Y. frederiksenii. A BLASTP search of the translated ORFs identified high amino acid similarity with many insecticidal Tc proteins of S. entomophila, P. luminescens, X. nematophila, and Y. pestis for all three genes (Table 3).

TABLE 3.

Similarities of products of putative TcYF ORFs to selected protein sequences in NCBI databases detected using BLASTP^a

ORF (size of product [aa])	Protein homologue (size [aa])	Degree of similarityb	Function of orthologue's protein	Organism	E value, accession no.
TcYF1 (2,374)	SepA (2,376)	71/80 (1-2376)	Insecticidal toxin	S. entomophila	0.0, NP065276
	XptA1 (2,504)	66/80 (1777-2523), 36/53 (422-1643), 47/65 (1-203)	Insecticidal toxin	X. nematophila	0.0, CAC38401
	XptA2 (2,538)	70/84 (1777-2523), 39/56 (1-1643)	Insecticidal toxin	X. nematophila	0.0, CAC38404
	TcdA1 (2,516)	62/76 (1758-2515), 39/56 (21-1239), 39/55 (1530-1640)	Toxin complex protein	P. luminescens	0.0, AAF05542
TcYF2 (1,434)	SepB (1428)	88/92 (1-1423)	Insecticidal toxin	S. entomophila	0.0, NP065277
	TcdB1 (1,476)	53/66 (1-1471)	Toxin complex protein	P. luminescens	0.0, AAL18487
TcYF3 (979)	SepC (973)	64/75 (5-677)	Insecticidal toxin	S. entomophila	0.0, CAC34921
	YPO2312 (874)	61/73 (5-682)	Putative insecticidal toxin	Y. pestis	0.0, CAC93143
	XptB1 (1,016)	61/74 (8-678)	Insecticidal toxin	X. nematophila	0.0, CAC38402

^aOnly the most similar proteins for each species are included.

^bAmino acid similarity (percent identity/percent similarity over the amino acid residues given in parentheses) in relation to a sequence generated in this study.

DNA similarities between pYPF49, pSP591, pSP1137, and pADAP.

To further define the relationships between the plasmids that hybridized to at least one but not all three of the probes, Southern blots of restriction-digested pADAP, pYF49 (total plasmid DNA [Fig. 2; Table 2]), pSP591, and pSP1137 DNA (Fig. 3A) were probed with DNA from each plasmid individually (Fig. 3B). The results were used to construct plasmid maps of regions of pADAP-related DNA on pYF49, pSP591, and pSP1137 (Fig. 3C).

FIG. 3.

(A) Agarose gels of restriction digests of total plasmid DNAs isolated from S. entomophila (pADAP), S. proteamaculans (pSP591 and pSP1137), and Y. frederiksenii (pYF49, with three additional plasmids). (B) Southern blot analyses of gels in panel A probed with either pSP591, pYF49 (and the three additional plasmids), or pSP1137. (C) Schematic diagram showing the restriction fragments of pADAP DNA that hybridized to pSP591, pYF49 (and the three additional plasmids), or pSP1137 DNA based on the restriction map of pADAP from the work of Hurst and Glare (17). Inner circle, BamHI; middle circle, EcoRI; outer circle, HindIII. Hatched regions indicate pADAP fragments that hybridized to pSP591, pYF49, or pSP1137. Asterisks denote the cloned fragments of pACΔ8 and p8L.

Approximately 45 kb of pADAP DNA encompassing the sepABC region hybridized to pYF49 and pSP591 DNA (Fig. 3B and C). The amount of pADAP-similar DNA on the plasmid of strain 1137 was investigated by probing Southern blots of BamHI, EcoRI, and HindIII restriction digests of pADAP (Fig. 3A) with the entire pSP1137 DNA. It was evident that the contiguous pADAP HindIII fragments of 7 kb, 11 kb, and 5.6 kb (17) did not hybridize to pSP1137 DNA (Fig. 3B). The relatively weak hybridization of pSP1137 DNA to the largest BamHI fragment of pADAP was attributed to similarity between DNA at the left of this large fragment, since the fragment of p8L, but not that of pACΔ8, hybridized to pSP1137 DNA (Fig. 3). It could not be ascertained whether the lack of hybridization to small (<500-bp) fragments in Fig. 3 was due to insufficient DNA or to a lack of similarity between this DNA and the probe DNA. Taken together, the data indicate that the region of pADAP not present in pSP1137 is similar to the region of pADAP-homologous DNA present in pSP591 and pYF49 (Fig. 3C).

DISCUSSION

Previously, only S. entomophila strain A1MO2 has been assessed for the presence of genetic determinants of amber disease that are carried on the 155-kb plasmid pADAP. In this study, we found sep gene orthologues on plasmids from S. entomophila, S. proteamaculans, and an avirulent strain of S. liquefaciens. Studies on the native plasmids in 94 Serratia strains found that, in most cases, genes orthologous to sepABC resided on plasmids with replication and conjugation regions similar to those of pADAP. However several S. proteamaculans strains contained large plasmids that hybridized to the pADAP replication and conjugation regions but lacked the sep gene cluster, while another strain contained a plasmid that hybridized only with the sep genes. These observations provide strong support for the suggestion that the sep virulence-associated region has been acquired by horizontal gene transfer separately from the plasmids on which it is located.

To date, the predicted products of most tc-like genes identified in different species and genera exhibit high amino acid similarity, but the genes show little similarity at the DNA level. The high nucleotide identities of the Y. frederiksenii genes to sepA and sepB of S. entomophila are suggestive of horizontal gene transfer between the two species. While tcYF1 and tcYF2 of Y. frederiksenii showed 88% and 87% DNA identities to sepA and sepB, respectively, tcYF3 had no significant DNA similarity to sepC. Nevertheless, the translated product of tcYF3 had significant amino acid similarity from the amino terminus to amino acid residue 678; this is in accordance with TcYF3 belonging to the Rhs family of elements (18, 40). The absence of DNA similarity between tcYF3 and sepC suggests that these genes were acquired independently by their hosts and may explain the absence of virulence of Y. frederiksenii toward C. zealandica. Alternatively, the tc-like genes from Y. frederiksenii may not be expressed under the bioassay conditions used.

The presence of tc-like genes in Y. frederiksenii is intriguing, since the species is closely related to Yersinia enterocolitica but is generally considered nonpathogenic (9). It has most often been isolated from fresh water or sewage and only rarely from soil. Yersinia spp. have been recovered from both healthy and amber disease-affected grass grub larvae previously (35). A tc gene cluster has recently been identified in Y. enterocolitica strain T83 (37), but it is thought to be avirulent. It is possible that these genes and the Y. frederiksenii tc genes code for specific toxin complexes for which a target insect species has yet to be identified.

The tc-like genes identified in P. luminescens, X. nematophila, and Y. pestis are chromosomally borne, while the sep genes of S. entomophila (18) and the tc-like genes of Y. frederiksenii strain 49 are plasmid borne. This raises the possibility that the tc genes could be transmitted between species via plasmids and that the chromosomally located tc genes could also once have been plasmid borne or mobile in the other bacteria.

The S. proteamaculans strain 591 plasmid pSP591 and pYF49 were the only two plasmids identified in the present study that were unrelated to pADAP but harbored sep-like genes. Given the difference in replication regions, these plasmids could potentially coexist within the same cell, facilitating the acquisition of sep genes from pADAP. Zhang et al. (43) reported a two-way genetic exchange between a symbiotic and a nonsymbiotic plasmid in Rhizobium leguminosarum, involving symbiotic genes. Recent experiments have demonstrated that pADAP was transferred by conjugation at a higher rate inside grass grub larvae than in surrounding soil (S. Dodd et al., unpublished data). Hence, such an environment may have facilitated the acquisition of the sepA/tcYF1 and sepB/tcYF2 genes in either Y. frederiksenii or S. entomophila. Two 785-nt degenerate direct repeats encompass the sepABC genes and the fimbrial cluster of pADAP (19). It will be of interest to probe the plasmids identified in this study with the 785-nt repeat, since it may define a mechanism behind the spread of the sep-like genes.

Numerous studies have reported the association of phage genes or partial phage sequences with PAIs of gram-negative bacteria (7, 27, 41). An ORF (ORF4) residing between sepB and sepC on pADAP encodes a bacteriophage-like protein (18), but no analogous ORF was present between tcYF2 and tcYF3 on pYF49. Furthermore, P. luminescens strains Hb (ATCC 29999) and W14 and Y. pestis strain CO92 also contain phage-like genes between their tcaC-like and tccC-like genes, whereas X. nematophila strain ATCC 19061 does not (40). These may be relics of a previously mobile phage element that was responsible for the acquisition of tc genes. A similar scenario has been postulated for the transfer of other virulence-encoding genes in Enterobacteriaceae (4).

In summary, the findings of sep-like genes on plasmids in Y. frederiksenii, a sep gene-containing region on a non-pADAP type plasmid in S. proteamaculans strain 591, and a large sep gene-containing region missing from the plasmids from four S. proteamaculans isolates (Table 2) together suggest that the region encompassing amber disease-encoding genes is mobile among certain strains of bacteria. Further sequencing of the novel regions will reveal the extent of similarity along the entire plasmid regions and will help determine if the sep-associated region is part of a pathogenicity island.