From insects to human hosts: Identification of major genomic differences between entomopathogenic strains of Photorhabdus and the emerging human pathogen Photorhabdus asymbiotica

ARTICLE IN PRESS

International Journal of Medical Microbiology 296 (2006) 521–530 www.elsevier.de/ijmm

From insects to human hosts: Identification of major genomic differences between entomopathogenic strains of Photorhabdus and the emerging human pathogen Photorhabdus asymbiotica Slim Tounsia, Mark Blightb, Samir Jaouaa, Andre´a de Lima Pimentac, a

Laboratoire des Biopesticides, Centre de Biotechnologie de Sfax, B.P.’’K’’, 3038 Sfax, Tunisie Centre de Ge´ne´tique Mole´culaire, Baˆt. 26, CNRS, F-91198 Gif-sur-Yvette, France c Universite´ de Cergy-Pontoise, Dept. Biologie, Lab. ERRMECe, F-95302 Cergy-Pontoise Cedex, France b

Received 14 November 2005; received in revised form 10 March 2006; accepted 8 June 2006

Abstract Pathogenic bacteria of the genus Photorhabdus are naturally found in symbiotic association with soil entomopathogenic nematodes, and are of increasing economic interest in view of their potential for the development of novel biopesticides. This bipartite natural system is currently used for the biological control of crop pests in several countries. However, an increasing number of Photorhabdus strains have recently been isolated from human clinical specimens in both the United States and Australia, associated with locally invasive soft tissue infections and disseminated bacteraemia. In view of their growing use in biological control, which increases the potential rate of exposure of humans to these pathogens, we decided to undertake a comparative study of the genomic differences between insect and human pathogenic strains of Photorhabdus, in an attempt to understand the genetic mechanisms involved in the apparent change of host specificity, presumably responsible for their recently acquired capacity to infect humans. The data presented here demonstrates that major genomic differences exist between strains of Photorhabdus exhibiting virulence against insects or humans. Several individual genes, coding for virulence factors, were isolated and shown to be specific to the Photorhabdus asymbiotica human pathogens. One of these genes, sopB, encoding a host cell invasion factor translocated via the type III secretion system, has been cloned and the comparison of its genomic context in different pathogens strongly indicates that horizontal gene transfer is implicated in the acquisition of these virulence factors specific to the human pathogens. The precise role of this and other virulence factors identified here in the pathogenicity of P. asymbiotica towards humans is currently under investigation. r 2006 Elsevier GmbH. All rights reserved. Keywords: Photorhabdus; Host–pathogen interaction; Host adaptation

Introduction Bacteria of the genus Photorhabdus, belonging to the Enterobacteriaceae, are the only terrestrial bacteria Corresponding author. Tel.: +33 1 3425 6612; fax: +33 1 3425 6694. E-mail address: [email protected] (A. de Lima Pimenta).

1438-4221/$ - see front matter r 2006 Elsevier GmbH. All rights reserved. doi:10.1016/j.ijmm.2006.06.004

exhibiting bioluminescence (Richardson et al., 1988). The classification within the genus is complex, with the first isolates being classified as Xenorhabdus luminescens (Szallas et al., 1997), further re-named as Photorhabdus luminescens, and currently subdivided into three recognised species: P. luminescens, P. temperata and P. asymbiotica (Fischer-Le Saux et al., 1999), with several subspecies identified to date.

ARTICLE IN PRESS 522

S. Tounsi et al. / International Journal of Medical Microbiology 296 (2006) 521–530

The first species Photorhabdus (P. luminescens) were isolated from soil nematodes belonging to the genus Heterorhabditis, with which they establish a symbiotic association. This symbiotic stage, during which the bacteria colonise the intestinal tract of the nematodes, is followed by a pathogenic stage, when the susceptible insect larva preys are killed by the combined action of the nematode and their associated bacteria (Derzelle et al., 2004). After several rounds of reproduction, a new generation of infective juvenile nematodes, carrying the symbiotic bacteria inside their intestinal tract, emerge from the cadaver and actively seek new insect larva hosts in the soil. Due to their production of novel toxins with insecticidal activity, and to the specificity of the host infected, which is determined by the nematode, Photorhabdus spp. have been the subject of intensive studies for the development of biopesticides (Ehlers, 2001). The use of these bacteria, or their purified insecticidal toxins, for the biological control of insect pests is a reality in many countries, such as Australia and the United States (ffrench-Constant and Bowen, 1999, 2000). Recent reports have acknowledged the identification of Photorhabdus strains obtained from human clinical specimens. These reports describe Photorhabdus as a virulent primary pathogen, not a benign secondary coloniser of existing wounds (Gerrard et al., 2004). The first human clinical isolates, at the time identified as Xenorhabdus luminescens, were reported by the Center for Disease Control (CDC), in the United States in 1986. More recently, the Microbiological Diagnostic Unit, in Australia, has identified six other clinical isolates of Photorhabdus. To date, 12 strains of this species are described in total, isolated from skin, upper or lower limbs, and less frequently from blood, of patients generally presenting symptoms of fever, acute dermal abscess, infected cutaneous or subcutaneous disseminative lesions or recurrent non-healing ulcers. Interestingly, all patients either worked in outside activities and/ or had lesions originating from recent spider bites (Peel et al., 1999; Farmer et al, 1989). A polyphasic approach conducted for the taxonomic study of these new isolates indicated that all human clinic isolates of Photorhabdus known to date fall into two different subspecies of P. asymbiotica. The American isolates (Farmer et al., 1989) grouped as P. asymbiotica subsp. asymbiotica subsp. nov., and the Australian clinical isolates (Gerrard et al., 2003a, b) as P. asymbiotica subsp. australis subsp. nov. (Akhurst et al., 2004). No invertebrate nematode vector for P. asymbiotica has yet been identified, nevertheless, curiously, all P. asymbiotica strains so far tested have proven to be highly virulent in insect models, with LT50 values sometimes equivalent to those of nematode symbiotic strains (Gerrard et al., 2004). Whether specific virulence factors associated with human infection

account for this high degree of insect pathogenicity, or vice versa, is not known. There is currently no experimental vertebrate model for P. asymbiotica, and little is known regarding the pathogenic mechanisms specifically associated with human infections. In view of their unique biological, physiological and ecological traits, and of their growing use in biological control, which increases the rate of exposure of humans to these pathogens, we decided to undertake a comparative study of the genomic differences between symbiotic and asymbiotic strains of Photorhabdus, in the attempt to understand the genetic mechanisms involved in the change in host specificity, responsible for its recent acquired capacity to infect humans. Here we have applied the suppression subtractive hybridisation (SSH) technique (Diatchenko et al., 1996) to the analysis of the major genomic differences between two Australian human isolates of P. asymbiotica, SN98-1 and 9800946 (herein after referred to as 946 and SN), and two geographically related insect pathogens P. luminescens strains, HV-16/2 and Q617/2 (herein after referred to as HV and Q). We were able to identify genes present in the asymbiotic strains that do not show similarities to, or that have greatly diverged from, their symbiotic counterparts. Such genes are candidates for virulence factors involved in the pathogenic mechanism specifically associated with human infection.

Materials and methods Bacterial strains and cultures All strains used in the SSH experiments were isolated in Australia (see Table 1 for details). Human isolates of P. asymbiotica, strains 9800946 and SN98-1 (herein after referred to as 946 and SN), and P. luminescens strains HV16/2 and Q617/2 (herein after referred to as HV and Q), specific to insect infections, were kindly supplied by Dr. Ray Akhurst, CSIRO, Canberra, Australia. P. temperata strain K122 was kindly supplied by Dr. D. Clarke, and P. luminescens strain W14 by Prof. Richard ffrench-Constant, University of Bath, UK. Details about all Photorhabdus strains used in this study are indicated in Table 1.

Molecular biology techniques Molecular biology protocols are as previously described (Ausubel et al., 1999; Sambrook et al., 1989). All restriction endonucleases and DNA-modifying enzymes were purchased from Promega with the exception of high-concentration T4 DNA ligase (400 U/ml) from New England Biolabs. Plasmid and genomic DNA were prepared using the appropriate Wizard kit systems

ARTICLE IN PRESS S. Tounsi et al. / International Journal of Medical Microbiology 296 (2006) 521–530

Table 1.

523

Photorhabdus strains used in this study

Strain

Species

Geographic origin

Clinical

Alleged vector

Source of isolate

Case number and Reference

9800946

P. asymbiotica

Victoria

Cough and fever

Spider

Blood

SN98-1

P. asymbiotica

New South Wales

None

Blood, sputum, pus and tissue

HV16/2

P. luminescens subsp. laumondii

Victoria

Multifocal soft tissue infections (upper and lower limbs, abdomen, pneumonia) n.a.

2 Peel et al. (1999) 4 Peel et al. (1999)

Soil

Akhurst et al. (2004)

Q167/2

Photorhabdus sp

Queensland

n.a.

Soil

K122

P. temperata

Ireland

n.a

R. Akhurst, pers. commun. Clarke (1993)

W14

P. luminescens subsp. akhurstii

Florida, USA

n.a

Heterorhabditis bacteriophora HV16 Unidentified nematode Heterorhabditis nematode Heterorhabditis nematode

Soil Soil

Bowen and Ensign (1998)

n.a ¼ does not apply.

(Promega). PCR reactions were performed in a GeneAmp PCR system 9700 machine (Perkin Elmer) using 50 pmol of each appropriate oligonucleotide primer (Genosys), 200 mM of each dNTP (Promega), 2.5 mM MgCl2 and Taq DNA polymerase (Promega) with the manufacturer’s supplied buffer.

Genomic subtraction Genomic subtraction reactions were performed according to Diatchenko et al. (1996) upon genomic DNA of tester (946 and SN) and driver (HV and Q) strains digested with AluI. Generation of the subtracted library was achieved by cloning the PCR-amplified subtracted fragments into the EcoRV site of pGEM-T (Promega), followed by transformation of competent Escherichia coli DH5a cells (F
, deoR, recA1, endA1, hsdR17 (rk
, mk+), supE44, l
, thi
1, gyrA96, relA1).

Nucleic acid hybridisation Southern blotting was performed following 0.7% TAE agarose gel electrophoresis of 3 mg HindIIIdigested genomic DNA and transfer to Hybond N+ nylon membranes (Amersham Pharmacia Biotech) with a semi-dry blotter (Biorad, model 785VA). Dot Blot membranes were arrayed with a 96-well vacuum manifold blotter (Bioblock) with 50 ng of PCR products per well. DNA hybridisations were performed using 32PdCTP (Amersham Pharmacia Biotech) labelled DNA probes prepared with the kit Rediprime II (Amersham Pharmacia Biotech) according to the manufacturer’s

instructions. Probes were purified on MicroSpin S200 HR columns (Amersham Pharmacia Biotech) and denatured for 2 min at 100 1C prior to hybridisation. Membranes were prehybridised for 3 h in 440 mM sodium phosphate buffer (pH 7.5), 1 mM EDTA, 7% (w/v) SDS, 1% BSA followed by the addition of denatured probe and hybridisation for 16 h at 65 1C. Following hybridisation, membranes were washed twice for 15 min in 2  SSC, 0.5% SDS (w/v) at 65 1C, and exposed to autoradiography film. For cloning of the genomic insert of clone 214, genomic DNA of P. asymbiotica strain 946 was prepared, digested with HindIII and analyzed by Southern blot, using the PCR-amplified genomic fragment of clone 214 as radioactive probe. A unique band corresponding to a 3-kb HindIII restriction fragment was visualised on the autoradiogram. A library enriched for 3-kb HindIII fragments was constructed from P. asymbiotica strain 946 genomic DNA and screened in stronger stringency conditions using the same probe as above.

DNA sequence analysis DNA sequencing was performed using the universal forward and reverse M13 primers. Elongated sequences were determined using internal primers based on the sequences obtained with the universal M13 primers for each sequenced fragment. Analysis of sequences was carried out with the BLAST program from NCBI. Alignments with the P. luminescens TT01 and P. asymbiotica ATCC43949 genome sequences

ARTICLE IN PRESS 524

S. Tounsi et al. / International Journal of Medical Microbiology 296 (2006) 521–530

were performed on the Photolist search service (http://genolist.pasteur.fr) and Wellcome Trust Sanger Institute sequencing projects Blast search services (http://www.sanger.ac.uk/DataSearch/blast.shtml), respectively.

1

(a)

2

3

4

5

6

7

8

9

1 0 1 1 12

A B C D E

Results Construction of the subtracted library containing sequences specific to the human-infecting Photorhabdus strains Genomic DNA prepared from human clinical isolates SN and 946 (tester) was subtracted against that of insect specific strains (driver) HV and Q by SSH (Diatchenko et al., 1996), as described in Materials and methods. In order to minimise the isolation of strain-specific sequences and maximise the identification of global differences between insect and human infective strains, both tester and driver samples were combinations of genomic DNA of more than one strain, as mentioned above. Fig. 1 illustrates the hybridisation of PCR products from 77 individual clones resulting from the ligation of subtracted genomic DNA fragments into pGEM-T vector. As expected, all clones analysed hybridised with the labelled tester DNA (Fig. 1a). Hybridisation results obtained with the insect pathogen DNA (Fig. 1b) indicates the presence of possibly three types of clones in the subtracted library: Type A, clones failing to hybridise, indicating DNA sequences that are absent from the insect pathogen genome (e.g. Fig. 1, G12); type B clones showing weak hybridisation signals, representing DNA sequences that are probably greatly divergent from their equivalent in the human pathogens, such that under the hybridisation conditions used, a weak or null signal is detected (e.g. Fig. 1, A1), and finally, type C, clones giving positive hybridisation signals equally intense with both probes, which were discarded as false positives (e.g. Fig. 1, E3). Using this approach, 218 subtracted fragments, with sizes ranging between 200 and 1200 bp, were cloned and analysed. This allowed the generation of a subtracted library containing 81 cloned fragments that were potentially either specific to the Photorhabdus strains capable of infecting humans, or that have sufficiently diverged from their insect-infecting counterparts. In order to confirm the differential nature of individual clones obtained in the subtracted library, 12 clones were randomly selected to be individually tested by Southern blotting against total genomic DNA of the tester (human) and driver (insect) strains. These 12 clones selected represented either type A or type B classes described above.

F G

(b) A B C D E F G

Fig. 1. Dot-blots of 77 individual clones of SN+946 subtracted genomic DNAs probed with either SN+946 (a, tester) or HV+Q (b, driver) genomic DNA 32P-labelled probes. PCR amplification products from these clones were arrayed onto Hybond N+ membranes in duplicate and subsequently hybridised with 32P-labelled probes prepared by random priming total genomic DNA from either strains 946 and SN (tester) or HV and Q (driver). G1–G5, empty wells; G6 and G7, vector DNA control.

Identification of genomic differences between Photorhabdus strains infecting insects and humans Fig. 2 shows typical Southern blot results obtained for each type of clones mentioned above (A and B), clearly illustrating the specificity of clone 214 for the P. asymbiotica genome, which shows no hybridisation with the driver strain DNAs (lanes 3 and 4), nor with the insect pathogen strains P. luminescens W14 and P. temperata K122, used here as controls (lanes 1 and 2). On the other hand, results obtained for clone 162 illustrate a divergence between human and insect genomes, as determined not only by the weaker hybridisation signal obtained with the driver DNA (lanes 3 and 4) or with the insect pathogens K122 and W14 genomes (lanes 1 and 2), when compared to the bands obtained with the human pathogen DNA (lanes 5 and 6), but also by the differences in size between the hybridised bands, indicating that fragment 162 is present in a different genomic context in each of the genomes analysed. Indeed, BLAST results further indicated that clone 162 shows significant similarity

ARTICLE IN PRESS S. Tounsi et al. / International Journal of Medical Microbiology 296 (2006) 521–530

Probe: Genomic

DNA:

Clone 162 1

2

3

4

525

Clone 214 5

6

1

2

3

4

5

6

Fig. 2. Southern blots of genomic DNA of different strains digested with HindIII and hybridised against 32P-labelled PCR fragments of clones 162 and 214. Lanes 1–4: insect pathogen strains K122, W14, and driver strains HV and Q, respectively; lanes 5 and 6: human pathogen strains (driver) SN and 946, respectively.

(e-value 3  10
72; 94% identity) to a P. luminescens gene coding a PgE1 protein, and this clone was not further analysed. Southern blotting analysis of genomic DNA probed with individual clones confirmed the dot blot results of individual clones tested against genomic DNA random probes. Therefore, the choice of clones subsequently submitted to DNA sequencing and analysis was made based directly on the dot blot results, without further analysis by Southern blotting. A total of 81 clones, either showing positive hybridisation on dot blots only with the tester DNA, or showing weak signals with the driver DNA, were chosen for further sequence analysis.

DNA sequencing and in silico analysis of the subtracted library To further screen the subtracted library for fragments specific to the human pathogen Photorhabdus strains, we compared the sequence of the 81 fragments selected above with the genome of two Photorhabdus strains recently available in the databases: P. luminescens subsp. laumondii TT01, an insect pathogen, and P. asymbiotica subsp. asymbiotica subsp. nov. ATCC43949, a human pathogen isolated in the USA. Results of the BLAST searches are summarised in Table 2. These results allowed the classification of the subtracted fragments analysed into 3 categories: (1) those found in the ATCC43949 genome, but showing no other homologues in the databank (1 out of 81); (2) those showing significant sequence similarity with the P. asymbiotica genome, but no significant similarity with the insect pathogen TT01 genome (10 out of 81); (3) those showing no similarity with either the insect

pathogen P. luminescens TT01 or the human pathogen P. asymbiotica genomes (2 out of 81 sequences analysed). It should be noticed that 68 additional P. asymbiotica sequences identified in subtracted DNA fragments still showed significant similarities to P. luminescens genes, having apparently diverged from the latter, and will not be further discussed. The accession number of the sequences identified here showing similarity to known genes are presented in Table 2. When analysed for their relative positions in the P. asymbiotica ATCC43949 genome, all subtracted fragments cloned showed similarity to genes located in the first quarter (1.4 Mb) of the ATCC43949 genome, with the majority of them clustering in the first 700 kbp of the P. asymbiotica chromosome. Such clustering of P. asymbiotica-specific genes may be indicative of a pathogenicity island.

sopB, a gene potentially implicated in the adaptation of Photorhabus to human hosts As summarised in Table 2, in silico analysis of the subtracted clones allowed the selection of 13 fragments from the subtracted library as possible candidates for genes directly implicated in the transition of Photorhabdus from insects to human hosts. Amongst the clones present uniquely in the P. asymbiotica genome, clone 214 showed a strong similarity with Salmonella enterica virulence factor sopB. In the human pathogen strain S. enterica, sopB, located within SPI5, encodes an inositol phosphate phosphatase, a bacterial effector protein that is delivered into the host cell by a type III secretion system (TTSS),

ARTICLE IN PRESS 526

S. Tounsi et al. / International Journal of Medical Microbiology 296 (2006) 521–530

Table 2. Similarity of subtracted fragments identified by BLAST against available genomes in the database pathogen strain P. asymbiotica ATCC 43949 (see text for details) Clone

Accession no.

First BLASTX hit

Organism

e-valuea

% Idb

Similarity with ATCC43949c

Fragments showing significant similarityd with the human pathogen genome (ATCC43949) but no significant similarityd with the insect pathogen genome (TT01) 6 AM231533 Putative type III secretion Vibrio 5e
25 48 Phot3-160g07.p1k system EscV protein parahaemolyticus 1.1e
83–94% RIMD 2210633 74 AM231534 ORF_12; similar to Pseudomonas 9e
07 50 Phot3-145g12.p1k glycosyl transferases group aeruginosa 1.3e
51–58% 1 83 AM231535 Transcriptional regulator, Pseudomonas 1e
25 59 Phot3-160g07.p1k ArsR family/rhodanesefluorescens Pf-5 662863 bp like domain protein 1.8e
53–92% 87 AM231536 ORF_12; similar to Pseudomonas 3e
07 51 Phot3-145g12.p1k glycosyl transferases group aeruginosa 1.3e
51–58% 1 92 AM231537 Low-affinity inorganic Yersinia pestis KIM 0.003 51 Phot3-220e08.q1k phosphate transport 4.3e
39–92% protein 113 AM231538 Putative membrane Bordetella 2e
21 37 Phot3-160g07.p1k protein parapertussis 12822 662863 bp 1.5e
112–90% 117 AM231539 Hypothetical protein Vibrio 2e
13 32 Phot3-160g07.p1k VPA1352 parahaemolyticus 662863 bp 1.1e
116–94% RIMD 2210633 167 AM231540 Conserved hypothetical Pseudomonas 4e
14 37 Phot5-492b06.p1k protein fluorescens Pf-5 327677 bp 0.92–57% 170 AM231541 Unknown Vibrio cholerae 3e
08 65 Phot2-2g07.p1k 308975 bp 1.1e
32–92% 214 AM231542 SopB Salmonella enterica 1e
25 53 Phot3-220e08.q1k subsp. houtenae 372317 bp 8.8e
53–88% Fragments showing no similarity with either the human pathogen genome (ATCC43949) or with the insect pathogen genome (TT01) 166 AM231543 Putative maturase-related Escherichia coli 6e
21 60 protein CFT073 193 AM231544 COG1418: Predicted HD Pseudomonas 5e
25 59 superfamily hydrolase fluorescens PfO-1 Fragments showing significant similarityd with the human pathogen genome (ATCC43949) and no significant similarityd with any other sequences available in the NCBI database 112 None — — — Phot2-119£06 1.6e
36 79% a

Expected value. % Identity. c e-value, % identity. d e-valueo1e
05. Genes are dived into three subgroups according to their specificity to the genome of the human pathogenic strain. b

mediating invasion of epithelial cells by the pathogen. In view of its clear role in virulence and host–pathogen interactions, we investigated the genomic context surrounding the sopB gene in the genome of the human pathogens, P. asymbiotica and S. enterica, in comparison to that of the insect pathogen P. luminescens TT01. Fig. 3 shows the results of the comparative analysis of the genomic context of the sopB gene in different pathogenic bacterial strains. As indicated, the sopB homologue is bordered upstream by the pykF gene,

encoding a pyruvate kinase, in both human pathogenic strains of P. asymbiotica, ATCC43949 (position 665 kb) and 946. The pykF gene is also present upstream of sopB in the human pathogen S. enterica (position 1499 kb), but in this case these two genes are separated by a region of 340.2 kb. In S. enterica the gene immediately upstream of sopB, ppiD (position 1160.6 kb), codes for the pathogenicity island encoded protein SPI3, probably belonging to the peptidase family, and has no homologues in the P. asymbiotica genome. Again in

ARTICLE IN PRESS S. Tounsi et al. / International Journal of Medical Microbiology 296 (2006) 521–530

TT01:

[3054224 norM

ATCC4 3949:

1kb

[3059799

plu2612

pykF

p lu 2 6 1 4

lpp [661700

[652000 norM

527

plu2612

pykF

P. asymbiotica 946:

integrase

sopB

pykF

plu2614 lpp

sopB [1499082

[1158843 S. enterica: sopB

pipD

340.2kb

lpp

pykF

SPI5

Fig. 3. Map of the chromosomal region encoding the effector protein SopB showing the genomic context of this gene in the genome of the insect pathogen TT01 (http://genolist.pasteur.fr/PhotoList/), the human pathogen strains P. asymbiotica ATCC43949 (http:// www.sanger.ac.uk/Projects/P_asymbiotica/) and 946 and S. enterica (Chiu et al., 2005). In the case of strain 946, only the 3-kb region cloned and sequenced here is represented. Vertical arrows indicate genes encoding tRNA-Val. plu2612 and plu 2614, first identified in the TT01 genome, code for an unknown protein and for a probable enoyl-CoA hydratase, respectively. Dotted line in the S. enterica map represents a gap of 340.2 kb not shown here for simplification. SPI5: Salmonella pathogenicity island 5.

S. enterica, the gene found immediately downstream of pykF is lpp (position 1497.1 kb), coding for a murein lipoprotein, a major outer membrane lipoprotein precursor, linking the outer and inner membranes. BLAST searches on the P. asymbiotica genome identified a homologue of lpp gene (position 661.4 kb), in this case separated from the pykF by a 5.6 kb region, containing sopB, as well as an integrase and a plu2614 homologues. plu2614 has been originally identified in the P. luminescens TT01 genome, and encodes a putative enoyl-CoA hydratase. In the P. asymbiotica ATCC43949 genome the couple sopB–pykF is flanked upstream by a gene similar to plu2612 of P. luminescens TT01, encoding an unknown protein, preceded by two tandem repetitions of a valine tRNA gene and by norM (position 3054.2 kb), coding for a predicted multidrug resistance protein. These four genes (pykF, plu2612, RNA-Val and norM) are also present in the genome of the insect pathogenic strain P. luminescens TT01, where the sopB and the integrase genes are missing.

Discussion Historically, bacteria belonging to the genus Photorhabdus are entomopathogenic, generally found in symbiotic association with soil nematodes belonging to the genus Heterorhabditis, which transport the bacteria inside their intestines as they invade insect host larvae. As the use of the couple Photorhabdus-nematode, or the

bacteria alone, increases in agriculture for the biological control of insects, so does the concern about possible future environmental and human health impacts of spreading such microorganisms in the field. It has been shown, for example, that insecticidal toxins from Photorhabdus spp. are encoded by genes homologous to those occurring naturally in the genome of Yersinia pestis, the causing agent of plague (ffrench-Constant et al., 2000; Tennant et al., 2005), and it is currently proposed that horizontal transfer of genes between these two species might have resulted from their common association with insects as bacterial pathogens (Parkhill et al., 2001). Within the variety of molecular techniques available for the analysis of genomic differences, subtractive genomic DNA hybridisation provides insights both at the level of strain-specific genes and those that have become highly divergent. In an attempt to identify some of the genomic determinants responsible for the recently acquired ability of P. asymbiotica to infect human hosts, we have therefore performed a subtractive hybridisation analysis of the major genomic differences between two Australian human isolates of P. asymbiotica, SN98-1 and 9800946, and two geographically related insect pathogenic P. luminescens strains, HV-16/2 and Q617/2. A mixture of DNA from two tester and two driver strains was used to avoid amplification of sequences specific to one isolate and not necessarily related to the host specificity. Equally, the choice of geographically related strains was based on our previous experience, showing that the use of geographically distinct isolates in the SSH experiments results in the amplification of a

ARTICLE IN PRESS 528

S. Tounsi et al. / International Journal of Medical Microbiology 296 (2006) 521–530

large number of transposon and DNA mobilisationassociated sequences. Moreover, to take into account the role of temperature adaptation in the change of insect pathogenic strains to human hosts, we have chosen to include in the mixture of driver DNAs the genome of an insect pathogen strain capable of growing at 37 1C, so as to avoid amplification of genes involved in temperature adaptation, and not directly correlated with acquisition of host-specific virulence factors. Using this approach a library of subtracted fragments was obtained, composed of sequences cloned from the genome of the P. asymbiotica strains which were shown, by direct hybridisation analysis, to be either specific to the human pathogen strains or to have highly diverged between the human and the insect pathogens. These two classes of clones were DNA sequenced and, with the availability of the sequence of two Photorhabdus genomes, were compared to the genome of the P. asymbiotica strain ATCC43949 and the P. luminescens strain TT01. The first interesting result that emerges from this analysis is that all fragments cloned after subtracted hybridisation showed similarity to genes located in the first 1.4v of the ATCC43949 genome, with the majority of them clustering in the first 700 kb of the P. asymbiotica chromosome. Since the analysis of Photorhabdus genomes does not indicate a bias towards AluI sites distribution, one possible explanation for this clustering could be that this region of the P. asymbiotica genome serves as a hotspot for the insertion and accumulation of sequences acquired through horizontal transfer, which explains why genes localised outside this 1.4-Mb region are not amplified after the subtractive hybridisation procedure. If this hypothesis is correct, this result should guide future work aiming at the study of horizontal transfer of genes and adaptation of this bacteria species. When analysed individually, BLAST results showed intriguing differences amongst the fragments cloned. Interestingly, as shown in Table 2, one subtracted fragment (clone 112), similar to Phot2-119 in the P. asymbiotica ATCC43949 genome, failed to present any significant similarity to any other sequence currently deposited in the databases. This fragment most probably represents a novel gene, present only in the genome of the human pathogen P. asymbiotica strains isolated both in Australia and in the USA, and may possibly serve as a marker for the geographic speciation of these Photorhabdus strains. Amongst all the analysed sequences, only 2 subtracted fragments showed no similarity to the P. asymbiotica ATCC43949 genome. Moreover, these sequences (clones 166 and 193), are not only absent from the genome of the human pathogen P. asymbiotica ATCC43949 but also from the P. luminecens TT01 genome. These sequences might represent genes that are unique to the

P. asymbiotica strains isolated in Australia used in this study (946 and SN), not present in the North American clinical isolates. BLASTX results showed that these sequences present both significant similarity to proteins involved in the maturation of tRNAs: RNase Z (de la Sierra-Gallay et al., 2005) in the case of clone 166, and template-independent nucleotidyltransferase (ATP(CTP):tRNA nucleotidyltransferase) (Yakunin et al., 2004), in the case of clone 193 (see Table 2). We speculate that these fragments may represent markers of DNA regions recently acquired in the evolution of the P. asymbiotica Australian strains. Most interestingly, Table 2 shows the identification of 10 subtracted sequences present uniquely in the P. asymbiotica genome. Those fragments (clones 6, 74, 83, 87, 92, 113, 117, 167, 170 and 214) potentially represent genes that are specific to the human pathogenic strains, with no counterparts in the insect pathogen genomes. Interestingly, all fragments classified in this category show similarity to genes present in other pathogenic bacteria, and most of them are acknowledged virulence factors, as discussed below. Clone 6, for example, is similar to a putative protein of Vibrio parahaemolyticus involved in type III secretion, which is homologous to the inner membrane protein EscV of the enteropathogenic E. coli, responsible for the injection of virulence factors into the eukaryotic cell during infection (Gauthier et al., 2003). It is important to note that the apparently relatively restricted number of P. asymbiotica specific sequences identified here is typically in agreement with results of similar studies performed with different pathogens (Janke et al., 2001; Parsons et al., 2002; Walker and Verma, 2002), as well as with the findings of Marokhazi et al. (2003), who showed, using a microarray analysis to investigate the distribution of virulent factors in strains of Photorhabdus, that most of the genes coding for the classic virulence factors are present in all strains of Photorhabdus studied, including P. asymbiotica. Clones 74 and 87 are both similar to a glycosyltransferase of the opportunistic pathogen Pseudomonas aeruginosa. In the case of clone 87, part of its sequence also shows similarity to genes coding for enzymes of the epimerase family. These two enzymes (glycosyltransferase and epimerase) are both involved in the biosynthesis of lipopolysaccharide and alginates in different strains of Pseudomonas and Salmonella (Campa et al., 2004), molecules with well-established functions in pathogenicity and virulence. Moreover, these two genes are generally localised together on the genome of different human pathogenic bacteria, which reinforces the hypothesis of their acquisition by horizontal transfer. Clone 117 shows similarity to a putative membrane protein NP_800863 (VPA1352) encoded by a filamentous phage associated with pandemic V. parahaemolyticus O3:K6 strain (Nasu et al., 2000). Interestingly, in

ARTICLE IN PRESS S. Tounsi et al. / International Journal of Medical Microbiology 296 (2006) 521–530

Vibrio this gene is localised upstream of VPA1354 which is similar to type III secretion inner membrane protein EscV, a protein also identified in this study as being specific to P. asymbiotica strains (clone 6). Since the role of filamentous phages in the intra- and interspecific exchange of genetic material is notorious, this finding seems to corroborate the hypothesis that these sequences were acquired by horizontal transfer during evolution of Photorhabdus strains, and are probably involved in their adaptation to human hosts. Finally, clone 214 was shown to be similar to the sopB gene, coding for a protein effector (SopB), injected by Salmonella inside the eukaryotic cell during invasion. SopB/SigD is an effector with phosphoinositide phosphatase activity required for TTSS-mediated invasion of epithelial cells by Salmonella (Hernandez et al., 2004; Knodler et al., 2005; Raffatellu et al., 2005). As a hostcell invasion factor, sopB plays a definitive and important role in the pathogenicity of Salmonella strains, and since our results indicated this gene as one of those present exclusively in the genome of human pathogen P. asymbiotica strains, we were interested in investigating the genetic context surrounding this gene in P. asymbiotica compared to the insect pathogen P. luminescens strain. In contrast to the other effectors secreted via TTSS, sopB is not localised in the vicinity of the genes coding for its secretory apparatus. In Salmonella, sopB is found in pathogenicity island SPI5 while the TTSS is located elsewhere on the genome, in SPI1. When analysing the P. asymbiotica genome, the same gene distribution is found, with sopB localising around 0.5 Mb distant from the cluster of genes encoding the TTSS backbone. More interestingly, the analysis of the genomic context of sopB indicates the presence of markers often related to genetic mobility and horizontal gene transfer, such as tRNA and integrase genes. When comparing the genomic region containing the sopB gene in the genome of the insect pathogen P. luminescens TT01 with that of the human pathogen P. asymbiotica, it is clear that in the adaptation to the human host P. asymbiotica have specifically acquired sopB and the pykF genes, since those are the only genes missing in this genomic region in TT01 (see Fig. 3). With these two exceptions, all the other genes are maintained in both strains and in the same orientation. It is also interesting to note that the genomic region cloned here from the Australian P. asymbiotica strain also shows the presence of these two genes (sopB and pykF), allowing the conclusion that the same genomic organisation also occurs in this strain. Taken together, these elements strongly indicate that sopB could be a factor acquired by P. asymbiotica through horizontal transfer. The acquisition of this gene, together with others identified and discussed here, might be the basis of the genetic arsenal that enabled this bacterium to broaden its host range to include

529

humans. The precise role of this and other virulence factors identified here in the pathogenicity of P. asymbiotica towards humans is currently under investigation.

Acknowledgements We are grateful to acknowledge the support of the C.N.R.S. and the French Ministe`re de l’Education et de la Recherche. We are also grateful to Ray Akhurst and CSIRO, and to Dr. D. Clarke (University of Bath), for providing us with some of the Photorhabdus strains used in this work.

References Akhurst, R.J., Boemare, N.E., Janssen, P.H., Peel, M.M., Alfredson, D.A., Beard, C.E., 2004. Taxonomy of Australian clinical isolates of the genus Photorhabdus and proposal of Photorhabdus asymbiotica subsp. asymbiotica subsp. nov. and P. asymbiotica subsp. australis subsp. nov. Int. J. Syst. Evol. Microbiol. 54, 1301–1310. Ausubel, F.M.B., Kingston, R., Moore, R.E., David, D., Seidman, J.G., Smith, J.A., Struhl, K. (Eds.), 1999. Short Protocols in Molecular Biology, fourth ed. Wiley, New York. Bowen, D.J., Ensign, J.C., 1998. Purification and characterization of a high-molecular-weight insecticidal protein complex produced by the entomopathogenic bacterium Photorhabdus luminescens. Appl. Environ. Microbiol. 64, 3029–3035. Campa, C., Holtan, S., Nilsen, N., Bjerkan, T.M., Stokke, B.T., Skjak-Braek, G., 2004. Biochemical analysis of the processive mechanism for epimerization of alginate by mannuronan C-5 epimerase AlgE4. Biochem. J. 381, 155–164. Chiu, C.H., Tang, P., Chu, C., Hu, S., Bao, Q., Yu, J., Chou, Y.Y., Wang, H.S., Lee, Y.S., 2005. The genome sequence of Salmonella enterica serovar Choleraesuis, a highly invasive and resistant zoonotic pathogen. Nucleic Acids Res. 33, 1690–1698. Clarke, D.J., 1993. An investigation into cold activity in the insect pathogenic bacterium Xenorhabdus luminescens. Ph.D. Thesis. National University of Ireland, Maynooth. de la Sierra-Gallay, I.L., Pellegrini, O., Condon, C., 2005. Structural basis for substrate binding, cleavage and allostery in the tRNA maturase RNase Z. Nature 433, 657–661. Derzelle, S., Ngo, S., Turlin, E., Duchaud, E., Namane, A., Kunst, F., Danchin, A., Bertin, P., Charles, J.F., 2004. AstR-AstS, a new two-component signal transduction system, mediates swarming, adaptation to stationary phase and phenotypic variation in Photorhabdus luminescens. Microbiology 150, 897–910. Diatchenko, L., Lau, Y.F., Campbell, A.P., Chenchik, A., Moqadam, F., Huang, B., Lukyanov, S., Lukyanov, K., Gurskaya, N., Sverdlov, E.D., Siebert, P.D., 1996.

ARTICLE IN PRESS 530

S. Tounsi et al. / International Journal of Medical Microbiology 296 (2006) 521–530

Suppression subtractive hybridization: a method for generating differentially regulated or tissue-specific cDNA probes and libraries. Proc. Natl. Acad. Sci. USA 93, 6025–6030. Ehlers, R.U., 2001. Mass production of entomopathogenic nematodes for plant protection. Appl. Microbiol. Biotechnol. 56, 623–633. Farmer 3rd., J.J., Jorgensen, J.H., Grimont, P.A., Akhurst, R.J., Poinar Jr., G.O., Ageron, E., Pierce, G.V., Smith, J.A., Carter, G.P., Wilson, K.L., 1989. Xenorhabdus luminescens (DNA hybridization group 5) from human clinical specimens. J. Clin. Microbiol. 27, 1594–1600. ffrench-Constant, R., Bowen, D., 1999. Photorhabdus toxins: novel biological insecticides. Curr. Opin. Microbiol. 2, 284–288. ffrench-Constant, R.H., Bowen, D.J., 2000. Novel insecticidal toxins from nematode-symbiotic bacteria. Cell. Mol. Life Sci. 57, 828–833. ffrench-Constant, R.H., Waterfield, N., Burland, V., Perna, N.T., Daborn, P.J., Bowen, D., Blattner, F.R., 2000. A genomic sample sequence of the entomopathogenic bacterium Photorhabdus luminescens W14: potential implications for virulence. Appl. Environ. Microbiol. 66, 3310–3329. Fischer-Le Saux, M., Viallard, V., Brunel, B., Normand, P., Boemare, N.E., 1999. Polyphasic classification of the genus Photorhabdus and proposal of new taxa: P. luminescens subsp. luminescens subsp. nov., P. luminescens subsp. akhurstii subsp. nov., P. luminescens subsp. laumondii subsp. nov., P. temperata sp. nov., P. temperata subsp. temperata subsp. nov. and P. asymbiotica sp. nov. Int. J. Syst. Bacteriol. 49, 1645–1656. Gauthier, A., Puente, J.L., Finlay, B.B., 2003. Secretin of the enteropathogenic Escherichia coli type III secretion system requires components of the type III apparatus for assembly and localization. Infect. Immun. 71, 3310–3319. Gerrard, J.G., McNevin, S., Alfredson, D., Forgan-Smith, R., Fraser, N., 2003a. Photorhabdus species: bioluminescent bacteria as emerging human pathogens? Emerg. Infect. Dis. 9, 251–254. Gerrard, J.G., Vohra, R., Nimmo, G.R., 2003b. Identification of Photorhabdus asymbiotica in cases of human infection. Commun. Dis. Intell. 27, 540–541. Gerrard, J., Waterfield, N., Vohra, R., ffrench-Constant, R., 2004. Human infection with Photorhabdus asymbiotica: an emerging bacterial pathogen. Microbes Infect. 6, 229–237. Hernandez, L.D., Hueffer, K., Wenk, M.R., Galan, J.E., 2004. Salmonella modulates vesicular traffic by altering phosphoinositide metabolism. Science 304, 1805–1807. Janke, B., Dobrindt, U., Hacker, J., Blum-Oehler, G., 2001. A subtractive hybridisation analysis of genomic differences between the uropathogenic E. coli strain 536 and the E. coli K-12 strain MG1655. FEMS Microbiol. Lett. 199, 61–66. Knodler, L.A., Finlay, B.B., Steele-Mortimer, O., 2005. The Salmonella effector protein SopB protects epithelial cells from apoptosis by sustained activation of Akt. J. Biol. Chem. 280, 9058–9064. Marokhazi, J., Waterfield, N., LeGoff, G., Feil, E., Stabler, R., Hinds, J., Fodor, A., ffrench-Constant, R.H., 2003. Using a DNA microarray to investigate the distribution of

insect virulence factors in strains of Photorhabdus bacteria. J. Bacteriol. 185, 4648–4656. Nasu, H., Iida, T., Sugahara, T., Yamaichi, Y., Park, K.S., Yokoyama, K., Makino, K., Shinagawa, H., Honda, T., 2000. A filamentous phage associated with recent pandemic Vibrio parahaemolyticus O3:K6 strains. J. Clin. Microbiol. 38, 2156–2161. Parkhill, J., Wren, B.W., Thomson, N.R., Titball, R.W., Holden, M.T., Prentice, M.B., Sebaihia, M., James, K.D., Churcher, C., Mungall, K.L., Baker, S., Basham, D., Bentley, S.D., Brooks, K., Cerdeno-Tarraga, A.M., Chillingworth, T., Cronin, A., Davies, R.M., Davis, P., Dougan, G., Feltwell, T., Hamlin, N., Holroyd, S., Jagels, K., Karlyshev, A.V., Leather, S., Moule, S., Oyston, P.C., Quail, M., Rutherford, K., Simmonds, M., Skelton, J., Stevens, K., Whitehead, S., Barrell, B.G., 2001. Genome sequence of Yersinia pestis, the causative agent of plague. Nature 413, 523–527. Parsons, Y.N., Panagea, S., Smart, C.H.M., Walshaw, M.J., Hart, C.A., Winstanley, C., 2002. Use of subtractive hybridization to identify a diagnostic probe for a cystic fibrosis epidemic strain of Pseudomonas aeruginosa. J. Clin. Microbiol. 40, 4607–4611. Peel, M.M., Alfredson, D.A., Gerrard, J.G., Davis, J.M., Robson, J.M., McDougall, R.J., Scullie, B.L., Akhurst, R.J., 1999. Isolation, identification, and molecular characterization of strains of Photorhabdus luminescens from infected humans in Australia. J. Clin. Microbiol. 37, 3647–3653. Raffatellu, M., Wilson, R.P., Chessa, D., Andrews-Polymenis, H., Tran, Q.T., Lawhon, S., Khare, S., Adams, L.G., Baumler, A.J., 2005. SipA, SopA, SopB, SopD, and SopE2 contribute to Salmonella enterica serotype Typhimurium invasion of epithelial cells. Infect. Immun. 73, 146–154. Richardson, W.H., Schmidt, T.M., Nealson, K.H., 1988. Identification of an anthraquinone pigment and a hydroxystilbene antibiotic from Xenorhabdus luminescens. Appl. Environ. Microbiol. 54, 1602–1605. Sambrook, J., Fritsch, E.F., Maniatis, T., 1989. Molecular Cloning – A Laboratory Manual, second ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Szallas, E., Koch, C., Fodor, A., Burghardt, J., Buss, O., Szentirmai, A., Nealson, K.H., Stackebrandt, E., 1997. Phylogenetic evidence for the taxonomic heterogeneity of Photorhabdus luminescens. Int. J. Syst. Bacteriol. 47, 402–407. Tennant, S.M., Skinner, N.A., Joe, A., Robins-Browne, R.M., 2005. Homologues of insecticidal toxin complex genes in Yersinia enterocolitica biotype 1A and their contribution to virulence. Infect. Immun. 73, 6860–6867. Walker, J.C., Verma, N.K., 2002. Identification of a putative pathogenicity island in Shigella flexneri using subtractive hybridisation of the S. flexneri and E. coli genomes. FEMS Microbiol. Lett. 213, 257–264. Yakunin, A.F., Proudfoot, M., Kuznetsova, E., Savchenko, A., Brown, G., Arrowsmith, C.H., Edwards, A.M., 2004. The HD domain of the Escherichia coli tRNA nucleotidyltransferase has 20 ,30 -cyclic phosphodiesterase, 20 -nucleotidase, and phosphatase activities. J. Biol. Chem. 279, 36819–36827.