Veterinary Microbiology 168 (2014) 403–412

Contents lists available at ScienceDirect

Veterinary Microbiology journal homepage: www.elsevier.com/locate/vetmic

Population structure and virulence content of avian pathogenic Escherichia coli isolated from outbreaks in Sri Lanka D.R.A. Dissanayake, Sophie Octavia, Ruiting Lan * School of Biotechnology and Biomolecular Sciences, University of New South Wales, Sydney, NSW 2052, Australia

A R T I C L E I N F O

A B S T R A C T

Article history: Received 20 February 2013 Received in revised form 27 October 2013 Accepted 22 November 2013

Avian pathogenic Escherichia coli (APEC) causes economically significant infections in poultry. The genetic diversity of APEC and phylogenetic relationships within and between APEC and other pathogenic E. coli are not yet well understood. We used multilocus sequence typing (MLST), PCR-based phylogrouping and virulence genotyping to analyse 75 avian E. coli strains, including 55 isolated from outbreaks of colisepticaemia and 20 from healthy chickens. Isolates were collected from 42 commercial layer and broiler chicken farms in Sri Lanka. MLST identified 61 sequence types (ST) with 44 being novel. The most frequent ST, ST48, was represented by only six isolates followed by ST117 with four isolates. Phylogenetic clusters based on MLST sequences were mostly comparable to phylogrouping by PCR and MLST further differentiated phylogroups B1 and D into two subgroups. Genotyping of 16 APEC associated virulence genes found that 27 of the clinical isolates and one isolate from a healthy chicken belonged to highly virulent genotype according to previously established classification schemes. We found that a combination of four genes, ompT, hlyF, iroN and papC, gave a comparable prediction to that of using five and nine genes by other studies. Four STs (ST10, ST48, ST117 and ST2016) contained APEC isolates from this study and human UPEC isolates reported by others, suggesting that these STs are potentially zoonotic. Our results enhanced the understanding of APEC population structure and virulence association. ß 2013 Elsevier B.V. All rights reserved.

Keywords: APEC MLST Phylogroup Virulence factors

1. Introduction Avian pathogenic Escherichia coli (APEC), a subgroup of extraintestinal pathogenic E. coli (ExPEC) enters through different routes including respiratory and genital tracts causes various extra intestinal diseases collectively termed as colibacillosis in chickens. So far the virulence mechanisms of APEC are not well defined. A large array of virulence genes encoding iron acquisition, adhesion, toxin/cytotoxin production and serum resistance, have been investigated to identify virulence determinants in APEC (Picard et al.,

* Corresponding author. Tel.: +61 2 9385 2095; fax: +61 2 9385 1591. E-mail address: [email protected] (R. Lan). 0378-1135/$ – see front matter ß 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.vetmic.2013.11.028

1999; Rodriguez-Siek et al., 2005; Moulin-Schouleur et al., 2007). Recent studies showed that the APEC virulence types are better classified using combinations of virulence associated genes as strains with combinations of certain virulence genes are more likely to be associated with diseases (Ewers et al., 2007; Johnson et al., 2008a; Schouler et al., 2012). It is also well established that the APEC virulence factors are associated with particular genetic backgrounds (Escobar-Paramo et al., 2004; Ewers et al., 2007). Several studies reported that APEC is mainly derived from ECOR phylogroup B2 and occasionally from phylogroup D (Johnson and Russo, 2005). However, APEC strains are now found among all four phylogroups with the majority in phylogroup A (Rodriguez-Siek et al., 2005;

404

D.R.A. Dissanayake et al. / Veterinary Microbiology 168 (2014) 403–412

Moulin-Schouleur et al., 2007; Dissanayake et al., 2008). Here, we present phylogenetic analysis and virulence gene profiling of 75 avian E. coli strains from 42 layer and broiler chicken farms in Sri Lanka. We used multilocus sequence typing (MLST) to determine the diversity and phylogenetic relationships of the isolates. The MLST data obtained were also compared with the E. coli reference collection (ECOR) and other APEC and ExPEC data available in the MLST database (http://mlst.ucc.ie/mlst/dbs/Ecoli), to obtain a comprehensive understanding of the population structure of APEC. Sixteen APEC associated virulence genes were typed for these isolates to determine their pathogenic potential. 2. Materials and methods 2.1. Bacterial isolates E. coli isolates analysed in this study were from intensively managed commercial broiler and layer chickens in Sri Lanka between the period 2001–2006. Dissanayake et al. (2008) have previously characterised these E. coli isolates according to phylogroup, LPS core type, serum resistance and for the presence of three virulence genes; tsh, iss and cvaC. Of the 196 isolates used in the previous study, a total of 75 isolates were selected for the present study based on the dendrogram constructed using previous phylogenetic and virulence genotype data (Supplementary Fig. 1). From each of the nine major clusters of the dendrogram, more than 40% isolates were selected to represent the diversity of phylogroup, LPS core type, virulence gene profiles and source of isolation (clinical or healthy chickens). If the isolates in a particular cluster were identical in terms of phylogroup, LPS core type and virulence profile, they were assumed to represent a single clonal strain and a representative sample was selected. Supplementary material related to this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.vetmic.2013.11.028. Accordingly, there were 55 strains categorised as clinical and 20 as commensals. Clinical isolates were from 55 different outbreaks in 42 farms, including 31 broiler farms and 11 layer farms. Of those, 40 were small scale farms whereas the other two were large scale broiler farms integrated with two different hatcheries. These farms were geographically unrelated but some farms might have obtained the day old chicks from the same hatchery. Clinical isolates were collected from liver, heart-blood or spleen of chickens with lesions of colibacillosis. Commensal E. coli isolates were from cloacae of apparently healthy chickens. Isolates were cultured on nutrient agar (Oxoid, UK) and incubated overnight at 37 8C. DNA was extracted using phenol chloroform method as described previously (Octavia and Lan, 2006). 2.2. Multilocus sequence typing Multilocus sequence typing (MLST) was performed according to Wirth et al. (2006). PCR amplification and sequencing of seven housekeeping gene fragments, i.e.

adenylate kinase (adk), fumarate hydratase (fumC), DNA gyrase (gyrB), isopropylmalate dehydrogenase (icd), malate dehydrogenase (mdh), adenylosuccinate dehydrogenase (purA) and ATP/GTP binding motif (recA), were performed following the protocols specified at the E. coli MLST website, http://mlst.ucc.ie/mlst/dbs/Ecoli. We developed a new set of primers (F50 -AGCCAGCGGACATCATTG30 , R50 GACGGAGAAAATGCCCAC-30 ) to amplify the fumC gene fragment. The other genes were amplified using the published primer sequences (Wirth et al., 2006). The mdh gene fragment was amplified using a nested PCR. Briefly, the gene fragment was first amplified using the outer primer pair and then 1 ml of resulting products was amplified by the inner primer pair. The nested inner primer pair PCR product was used for sequencing. The PCR reaction mixtures (50 ml) contained 0.5 ml of 30 mM for each forward and reverse primer (Invitrogen), 0.5 ml of 10 mM deoxynucleoside triphosphates (dNTPs, New England BioLabs, Beverly, MA, USA), 5 ml of 10 PCR buffer New England BioLabs, Beverly, MA, USA), 1.25 U of Taq polymerase (New England BioLabs, Beverly, MA, USA), 20 ng of template DNA and sterile ultrapure (type 1) water commonly known as MilliQ water (Millipore, Darmstadt, Germany). The PCR conditions were: initial denaturation for 4 min at 94 8C, followed by 35 cycles of 15 s DNA denaturation at 94 8C, 30 s primer annealing at 55 8C (60 8C for gyrB and mdh gene fragments) and 30 s extension at 72 8C with a final extension of 5 min at 72 8C. PCR products were verified on gel red (Biotium, Hayward, CA) stained agarose gels before purification using sodium acetate– ethanol precipitation. The PCR sequencing reaction mixtures contained BigDyeTM version 3.1 (Applied Biosystems, Foster city, CA, USA) and PCR was performed as recommended by the manufacturer using appropriate forward or reverse primer (note that mdh was sequenced using one of the inner primers). The reaction products were separated and detected by capillary electrophoresis using ABI3730 automated DNA sequence analyzer (Applied Biosystems, Darmstadt, Germany) at the Ramaciotti Centre, University of New South Wales, Sydney, Australia. 2.3. In silico MLST analysis A collection of 316 sequence types (ST) representing APEC and UPEC available online (http://mlst.ucc.ie/mlst/ dbs/Ecoli) and 61 STs identified in the present study were used for in silico MLST analysis. 2.4. Virulence genotyping and O78 serogrouping The presence of 13 virulence genes and O78 serogrouping were analysed as described previously by single PCR reactions with primers detailed in Table 1. Each 25 ml reaction mixture contained, 0.5 U Taq DNA polymerase (New England BioLabs, Beverly, MA, USA) 2.5 ml of 10 Taq DNA polymerase buffer (New England BioLabs, Beverly, MA, USA), 0.5 ml of 10 mM dNTP (New England BioLabs, Beverly, MA, USA), 0.15 ml (30 mM) of each oligonucleotide primer pair (Table 1, Invitrogen), 2 ml (30 ng) of template DNA and sterile ultrapure (type 1) water (Millipore, Darmstadt, Germany).

D.R.A. Dissanayake et al. / Veterinary Microbiology 168 (2014) 403–412

405

Table 1 Primer sequences and the distribution of virulence genes among clinical and commensal E. coli isolates. Gene

Primer sequence (30 –50 )

Size (bp)

Temp Ref.

Distribution of genes (%)

P value

Clinical

Commensal

78

55

0.0917

69

40

0.04342*

72

65

0.7164

67

65

1

55

50

0.9305

58

45

0.4522

25

40

0.8746

25

10

0.01057*

45

10

0.01057*

74

40

0.01222*

72

60

0.439

49

20

0.03322*

55

30

0.1052

55

35

0.2164

issa

47

0

0.00041*

cvaCa

42

20

0.5084

Iron acquisition iroN F AAGTCAAAGCAGGGGTTGCCCG iutA

sitAC

F GGCTGGACATCATGGGAACTGG

302 608

R TACCGGGCCGTTTTCTGTGC

Rodriguez-Siek et al. (2005)

F ATGCACTCGATAAAAAAAGT

860

F ACAAAAAGTTCTATCGCTTCC

F GTGGCAGTATGAGTAATGACCGTTA R ATATCCTTTCTGCAGGGATGCAATA

63 8C 63 8C

Schouler et al. (2012) 714

R CCTGATCCAGATGATGCTC Adhesins papC

63 8C

Johnson et al. (2008b)

F AGGGGGCACAACTGATTCTCG

R TTAAGAAGGTCGATATACGT iucD

63 8C

Rodriguez-Siek et al. (2005)

R CGTCGGGAACGGGTAGAATCG sitAP

667

R GATCGCCGACATTAAGACGCAG

58 8C Ewers et al. (2007)

205

63 8C

Johnson et al. (2008b)

Pap-

F GGCCTGCAATGGATTTACCTGG

258

GIII

R CCA CCAAATGACCATGCCAGAC

fel A

F GGTCAASCAGCTAAAAACGGTAAGG

Moulin-Schouleur et al. (2007) 239 61 8C

R CCTTCAGAAACAGTACCGCCATTCG

63 8C

Moulin-Schouleur et al. (2007)

tsha Protectins ompT

F TCATCCCGGAAGCCTCCCTCACTACTAT R TAGCGTTTGCTGCACTGGCTTCTGATAC

hlyF

F GGCCACAGTCGTTTAGGGTGCTTACC R GGCGGTTTAGGCATTCCGATACTCAG

frz

F TCAGTAAGAACGAAAGTGTG

(orf4)

R ACAGGAACAATCCCGTGGAT

aec26

F ATGAGCGATATGAGTGAAGC R TTATCGGAGTAATTTATTGA

aec4

F AT CGTACCGCCTTCATTATC R CTTTACCGTCCTCACC

O78

F GGTATGGGTTTGGTGGTA R AGAATCACAACTCTCGGCA

496

63 8C

Johnson et al. (2008b) 450

63 8C

Morales et al. (2004) 565

53 8C

Schouler et al. (2012) 760

58 8C

Schouler et al. (2012) 135

53 8C

Schouler et al. (2012)

983

50 8C Liu et al. (2010)

Virulence genes significantly (P < 0.05) associated with clinical isolates are shown with a * mark. Virulence genes tested in the previous study (Dissanayake et al., 2008) are shown with an a mark.

2.5. Bioinformatic analyses Sequences were edited using the PHRED-PHRAPCONSED package (Gordon et al., 1998). Allele numbers for seven gene fragments of each isolate were obtained by comparing with corresponding allele available in MLST E. coli database (http://mlst.ucc.ie/mlst/dbs/Ecoli) and ST of each isolate was determined by combining seven allelic profiles (Wirth et al., 2006). The multiplesequence alignments were done using MULTICOMP (Reeves et al., 1994). PHYLIP package (Felsenstein, 1993) was used to generate neighbour-joining (NJ) tree, unweighted pair group method with arithmetic means (UPGMA) dendrogram and bootstrap values. Salmonella enterica serovar Typhi strain Ty2 (Accession No.: NC_004631) was used as an outgroup (Deng et al., 2003). STRUCTURE version 2.2 (Pritchard et al., 2000), a

Bayesian approach for analysis of MLST data, was used to assess the possibility that each isolate has derived all of its ancestry from only one population with varying degrees of admixture. The ‘‘admixture’’ model was used to determine the number of populations (K) and in each run, the Markov Chain Monte Carlo (MCMC) simulation of 30,000 iterations approximated the posterior probability of K, following a burn-in of 10,000 iterations. Different values of K were run multiple times, and the K value that generated the highest posterior probability was used as the number of possible populations. Based upon related ST algorithm, eBURST V3, available at http://eburst.mlst.net, was used to cluster STs into clonal complexes (CCs) with single locus variants (SLV) (Feil et al., 2004). A Minimum spanning tree (MST) using pairwise difference was generated using Arlequin v. 3.1 (Excoffier et al., 2005).

406

D.R.A. Dissanayake et al. / Veterinary Microbiology 168 (2014) 403–412

Significance of association between virulence genes among clinical and commensal E. coli isolates was determined by Pearson’s Chi-square test with Yates continuity correction using ‘R’ software (version 2.15.2). Fisher’s exact test was used to determine the significance of association of virulence genes with phylogroup. Differences were considered statistically significant at 5% (P  0.05). 3. Results 3.1. Allelic profiles, sequence types and clonal complexes of the strains We sequenced the internal fragments of seven housekeeping genes (adk, fumC, icd, gyrB, purA and recA) from 75 avian E. coli strains isolated from commercial chickens in order to understand the genetic diversity and phylogenetic relationship of APEC. Of the seven genes, the most variable gene was fumC with maximum and average pair wise differences of 24.3% and 2.89%, respectively and the most conserved gene was recA with corresponding values of 9.5% and 0.75%, respectively. The 75 avian E. coli isolates were differentiated into 61 STs. Out of the 61 STs, seven STs were represented by more than one isolate while the remaining 54 STs were each represented by only one isolate, 17 were known STs in the MLST database while 44 were novel STs (ST3949-ST3992). Of the 17 known STs, ST48 was the most common and represented by six strains; ST117 by four strains; ST155 by three strains; ST162 and ST949 by two strains. Twelve known STs, ST10, ST101, ST206, ST616, ST865, ST1464, ST1634, ST2016 ST2207, ST2732, ST271, ST2750, were each represented by one isolate. With the exception that ST117 was isolated twice in two different outbreaks of the same farm, all other STs represented by more than one isolate were collected from different farms. Since the clinical isolates were obtained from outbreaks, the MLST analysis showed high diversity of APEC in causing the outbreaks. Of the 75 isolates analysed in the present study, 18 were collected from commercial layer chickens and 57 from broiler chickens. There was only one ST (ST48) isolated from both layers and broilers with two and four isolates respectively. Therefore, we cannot determine whether there is any association of particular ST causing infections in broiler or layer chickens. We then performed eBURST analysis to identify CC, which is a group of closely related genotypes that share six loci to at least one other ST and descended from a common recent ancestor. Accordingly, there were nine CCs and 28 singletons. The largest CC contained 10 STs and the predicted founder of the CC was ST10. ST48, the most common ST represented by six strains belonged to this CC. The second and third largest CCs were represented by eight (founder = ST155) and five STs (founder = ST117). Three CCs were represented by three STs with founder STs being ST3967, ST206 and ST3972 respectively while the remaining CCs were represented by only two STs. Clonal complexes with three or more STs are hereafter referred by the predicted founder of each complex with the prefix ‘‘CC’’ i.e. CC10, CC155, CC117 and CC206, CC3967 and CC3972.

3.2. Phylogenetic relationships of the isolates The NJ tree of 75 E. coli isolates, based on the concatenated sequences of all seven genes, is shown in Fig. 1. However, the clustering pattern and the topology of individual NJ trees constructed using sequences of each gene were not congruent to each other (data not shown). It may indicate that the recombination is not uncommon among avian E. coli. We then reconstructed NJ tree incorporating the concatenated sequences of ECOR strains (Supplementary Fig. 2). The avian E. coli isolates clustered under ECOR phylogroups A, B1, B2, D and E, respectively. There were some inconsistencies between PCR and MLST based phylogrouping. As described previously (Clermont et al., 2011), there was a group closely related to B1 but the majority of the isolates fell into this group was identified as phylogroup A by triplex PCR (Clermont et al., 2000). To exclude the technical errors we repeated the triplex PCR for all isolates and eight isolates previously identified as phylogroup A were corrected as phylogroup B1 (phylogroup of those isolates were highlighted in red in Supplementary Fig. 1). According to our results, inconsistencies between PCR and MLST phylogrouping were mainly associated with misidentification of isolates as phylogroup A by PCR. Supplementary material related to this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.vetmic.2013.11.028. In the in silico MLST analysis that is performed including the sequences of ECOR strains, two subgroups were identified within the strains identified as phylogroup D by PCR (D and D-I). As shown in Supplementary Fig. 1, three strains (labelled as 41, 111 and 112) clustered together with ECOR48, ECOR49 and ECOR50 (D-I). The remaining nine strains identified as phylogroup D by PCR (D) were separately grouped. 3.3. Population structure of avian E. coli isolates We used Bayesian statistics tool, STRUCTURE, to determine the population structure of the 75 avian E. coli isolates. These isolates were divided into five subpopulations (Fig. 2). The ancestry of each isolate was estimated as the sum of probabilities from each subpopulation over all informative bases. As shown in Fig. 2, 38.6% (29/75) of the strains possessed mosaic blocks, which were likely to be resulted from gene flows between subpopulations. The proportion of nucleotide imported from a single or multiple populations varied from 1.4% to 55%. When a Q-value (probability of membership) of >0.67 was used as the cut-off point, 93% of the strains were unambiguously assigned into one of the five subpopulations. Four isolates were designated as ADB (marked with * in Fig. 2) and one isolate was designated as AB1 (marked with ** in Fig. 2) following Wirth et al. (2006). As shown in Fig. 2, subpopulations identified by STRUCTURE were mostly correlated with the clonal clusters identified by phylogenetic analysis. However, three strains (41, 111, 112) identified as phylogroup D by PCR were clustered together with phylogroup B2 isolates. It is noteworthy that the proportion of genes imported by the isolates belonging to B1-I was considerably higher

D.R.A. Dissanayake et al. / Veterinary Microbiology 168 (2014) 403–412

407

Fig. 1. Neighbour-Joining tree combining nucleotide sequences of seven housekeeping genes: adk, fumC, icd, mdh, recA and purA. Phylogroup (PCR), sequence type, lipopolysaccharide core types and virulence genes, were displayed alongside the strain. Black square indicates the presence of the virulence gene. Virulence clusters (I–IV) were determined by UPGMA (see Supplementary Fig. 2). Numbers at the tip of the terminal branches are strains names and the numbers on the nodes are bootstrap values (only 50% are displayed). *Type: 1. Clinical isolates 2. Commensal isolates. LPS core types (R1–R4). (Dissanayake et al., 2008)

D.R.A. Dissanayake et al. / Veterinary Microbiology 168 (2014) 403–412

408

1.00

0.60

0.80

0.20

0.40

0.00 20 (ST117) 18 (ST3949) 36 (ST117)

16 (ST117) 40† ( ST3117) 58† (ST3951) 38 (ST3950) 103 (ST3952) * 61† (ST3983) *8 (ST3974)

D-I

3.4. Distribution of virulence associated factors in phylogroups

33 (ST616) 56 † (ST155) 42 (ST155) 130† (ST3962) 2 (ST3953) 66 (ST1464) 116 (ST865) 53† (ST155) 35 (ST3955) 185 (ST3967) 13 (ST3954) 46† (ST949) 15 (ST949)

B1

115 (ST3961) 78 (ST3959) 55† (ST3957) 49 (ST3991) 167 (ST3966) 133 (ST3963) 147 (ST3964) 188 (ST3968) 164 (ST3965) 52† (ST3956) 141† (ST206) 101 (ST3960) 59 (ST3958) **140 † (ST3984) 156 (ST3972) 190 (ST3972) 196 (ST3973) 19 (ST3969) 41† (ST3970) 112 (ST3971) *111 (ST2772) 76 (ST2207)

B2

D-II

22† (ST2750) 54† (ST3976) 23 (ST48) 29 (ST48) 30† (ST48) 50† (ST48) 83 (ST3987)

A

93 (ST8) 10 (ST3985) 27(ST48) 178† (ST1634) 31† (ST3975) 3† (ST10) 175 (ST3977) 34 (ST3986) 21 (ST3990) 24 (ST3989) 12 (ST3989) 25 (ST162) 92 (ST162) 90 (ST3992) 7 (ST3988) 6 (ST3978) 32† (ST2741) 45 (ST3979) 48 (ST3980) 142 (ST3982) 60 (ST3981)

than that of other subpopulations. Subpopulation A has shown the least amount of gene flow. The majority of commensal isolates (marked as y in Fig. 2) were in phylogroups A and B1 respectively.

B1-I

138 (ST101) *70 (ST2016)

Fig. 2. STRUCTURE analysis of 75 avian E. coli isolates. Five subpopulations (1–5) identified by STRUCTURE are shown in different colours. Phylogroup (A, B1, B2 and D) comparable to each subpopulation and the subgroups of phylogroup B1 (B1-I) and D (D-1) are shown in parenthesis. Mosaic colours signify the isolates with mixed ancestries and colours of the mosaic boxes represent the colour of the respective population where the gene is acquired from. Identification number and sequence type of the strain are shown in the left. The * sign and ** sign before the strain name respectively indicate the stains identified as ABD and AB1 based on Wirth et al., 2006). The y sign after strain name indicates commensal isolates.

LPS core types and the presence of three virulence encoding genes (cvaC, tsh and iss) of these isolates have been determined previously (Dissanayake et al., 2008). In the present study, the isolates were screened for 13 more virulence associated genes namely, iroN, iutA, sitAP (plasmid), sitAC (chromosomal), papC, papG (III), felA, ompT, hlyF, frz, aec4, aec26 and iucD since the presence of these genes have been reported as high predictors for APEC. As shown in Table 1, of the 16 genes tested, iutA, felA, tsh, ompT, iss and frz were significantly associated with clinical isolates (P < 0.05). To visualise the distribution of virulence genes in different phylogroups, virulence gene profiles of each isolate is shown alongside the NJ tree (Fig. 1). E. coli isolates of phylogroups B2 and D showed high percentage of virulence genes tested. As shown in Fig. 1 isolates representing same ST had different virulence gene profiles. The virulence gene profiles, LPS core types, and phylogroups of 75 E. coli isolates were used to construct UPGMA tree (Supplementary Fig. 3). Virulence gene profiles and phylogroups were superimposed on the tree to visualise the distribution of virulence genes in each cluster. The UPGMA tree revealed four major clusters with different virulence profiles. As shown in Table 2, the vast majorities of commensal isolates were in cluster I and cluster III respectively. None of the isolates of cluster I (20 isolates) possessed cvaC gene and only 15%, 20% and 30% isolates possessed iss, felA and tsh gene, respectively. However, the majority of isolates (90%) in cluster I contained at least three iron related genes. Almost all isolates of cluster II harboured cvaC gene (93%) together with a minimum of three iron related genes. Isolates of cluster III lacked many of the virulence genes tested and only 11% carried at least three iron related genes. Cluster IV consisted almost entirely (96%) of clinical E. coli isolates (92%) and carried four or more iron related genes. Of the 24 isolates belonged to this cluster, 19 (79%) possessed more than 10 virulence associated genes tested. Among the virulence genes tested, the iss and felA genes were strictly associated with clinical avian E. coli isolates. Further, it was very clear that the sitAP, ompT, hlyF and iroN genes are closely associated with clinical avian E. coli isolates and more often absent in commensals. Supplementary material related to this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.vetmic.2013.11.028. 3.5. Comparison of APEC and ExPEC STs To display the relationship of our avian E. coli isolates with APEC, commensal E. coli and ExPEC from other studies, we extracted MLST data of 316 STs representing APEC, UPEC, avian commensal isolates from the publicly available MLST database (http://mlst.ucc.ie/mlst/dbs/ Ecoli). We identified all STs that were related to our STs

D.R.A. Dissanayake et al. / Veterinary Microbiology 168 (2014) 403–412

409

Table 2 Distribution of isolates with different virulence gene combinations among four UPGMA clusters. Cluster

Number of strains Clinical commensal

Gene combinations

(UPGMA)

Johnsona

Schoulerb

sitAP, ompT, hlyF, iroNc

I II III IV

13 10 9 23

7 4 8 1

1 – – 15

7 (1*) 5 – 16

8 (2*) 4 1 (1*) 17

Total

55

20

16

28 (1*)

30(3*)

The * mark indicates the number of commensal showing the particular gene combination. a iss, iroN, iutA, hlyF and ompT gene combination described in Johnson et al. (2008a). b Any of the four gene combinations described in Schouler et al. (2012). c Gene combination described in the present study.

at clonal complex level by eBURST and displayed in Fig. 3 by minimum spanning tree. The largest clonal complex, CC10 (phylogroup A) containing 24 STs, was mostly consisted of commensal STs. ST48, the most common ST in our strain collection and one new ST (ST3987) identified in our study belonged to this CC. CC155 was also comprised of APEC, UPEC and commensal STs and contained eight STs identified in the present study. Of the remaining two large clonal complexes, CC93 consisted of avian pathogenic and commensal STs whereas the CC88 entirely consisted of avian pathogenic STs. As shown in Fig. 3, ST48, ST117 ST2016 and ST 1567 were represented by both APEC and UPEC isolates or isolates closely associated with both groups. ST3987, a novel ST identified in the present study was closely associated with ST48. Another four novel STs, ST3949, ST3951, ST3952 and ST3974, were closely related to ST117. 4. Discussion 4.1. Genetic diversity and phylogenetic relationship of clinical and commensal avian E. coli strains The dataset analysed in the present study revealed substantially high genetic diversity within the subset of APEC. The 75 APEC isolates we studied contained 61 STs with 44 being novel. The high genetic diversity observed may in part be due to geographical location where APEC has not been sampled previously. This is reflected by the fact that the nearly half of the novel STs (21 out of 44) do not belong to any of the known clonal complexes. However, for the novel STs that belong to known clonal complexes or form clonal complexes among themselves, recombination seems to be a main driver to generate novel STs. We examined the nature of the allelic difference (recombination or mutation) of each novel ST to its closest ST of the same clonal complex. We used the rules devised by Feil et al. (2004) which have been used in other studies (Ch’ng et al., 2011). If the allelic difference is due to a single base difference, we count it as a mutation whereas if the allelic difference is due to more than one base difference, we count it as a recombination. We found that of the 28 possible allelic differences between a novel ST and other STs with one allele difference (see Fig 3), 20 were due to recombination and eight were due to mutation. Thus, it is

2.5 times more likely that a novel ST was generated by recombination than by mutation. Comparison of our data with the data in the MLST database revealed that there were certain STs and CCs more frequently represented by pathogenic E. coli than commensals. As an example, ST117 was one of the STs identified in the present study as highly virulent avian E. coli. There were 17 other ST117 APEC isolates in the MLST database (http://mlst.ucc.ie/mlst/dbs/Ecoli) confirming ST117 as a pathogen of chickens. Our virulence genotyping revealed that all four strains representing ST117 in the present study had the majority of virulence traits. The presence of smaller pathogenic units within major phylogroup was further evidenced by the CC10 and CC93 of phylogroup A. Strains belonging to phylogroup A are more likely to be commensal (Johnson and Russo, 2005; Jaureguy et al., 2008). CC10, the largest CC of phylogroup A was almost exclusively consisted of avian commensals. Of the 10 Sri Lankan isolates belonging to CC10, six did not posses any of the APEC virulence gene combinations tested. In contrast, CC93 from phylogroup A was consisted of a number of pathogenic strains including one of our clinical isolates. The phenomenon that particular STs or CCs rather than the entire phylogroup as more appropriate units of pathogenicity has been previously shown in a study of human bacteraemic E. coli strains (Jaureguy et al., 2008). Our study supports this observation. 4.2. Association of virulence profiles with phylogenetic groups The avian E. coli strains used in the present study mainly belonged to phylogroups B1 and A, and to a lesser extent to phylogroups D and B2. In contrast, it has been shown in some studies that the avian pathogenic clones are mainly derived from phylogroup B2, and some from phylogroup D (Johnson and Russo, 2005). We screened the isolates for 16 virulence associated genes in order to identify whether they are true pathogens or commensals causing secondary infections. Virulence profile analysis (16 genes) revealed that the isolates belonged to phylogroup A, B1, B2 and D possessed an average of 6, 9, 14 and 11 virulence genes, respectively. However, the number of genes present in phylogroup A and B1 strains varied between 1 and 15 while phylogroup B2 and D strains ranged between 6 and 16. Cluster analysis

410

D.R.A. Dissanayake et al. / Veterinary Microbiology 168 (2014) 403–412

Fig. 3. Minimum Spanning Tree (MST) showing the relationship between APEC, UPEC and commensal E. coli strains. Each circle represents a sequence type (ST) and the number within the circle is the ST number. Black lines connect the single locus variant STs (STs with one allele different). A cluster of linked STs corresponds to a clonal complex (CC). Note: Others: APEC, UPEC and commensal, STs reported on MLST database (http://mlst.ucc.ie/mlst/dbs/Ecoli).

by UPGMA on virulence profiles, phylogenetic groups, LPS profiles identified four main clusters. Cluster I and II contained majority of phylogroup B1 isolates and some isolates from phylogroup A and D. Cluster III was largely composed of phylogroup A (82%). All four isolates of phylogroup B2 and seven isolates of phylogroup D belonged to the cluster IV and remaining isolates were from phylogroups B1 and A. Four clusters showed remarkable variations in the distribution of virulence genes. Certain genes (tsh, iucD, iroN, cvaC, and iss) located on large transmissible R or ColV plasmids are known to be associated with APEC strains (Ewers et al., 2007; Johnson et al., 2006b). Most of the strains in the cluster IV harboured tsh, iucD, iroN and iss genes whereas these genes were rarely present among other clusters particularly, cluster III. Only half of the isolates of cluster IV possessed ColV operon encoding gene (cvaC). However, there was a high (93%) association of cvaC gene with cluster II isolates. A previous study (Johnson et al., 2006b) has shown that the genes contained within ColV plasmid are highly conserved and significantly associated with APEC. Acquisition of those plasmids by commensals could enhance their virulence potential. In addition, considerable differences in the presence were observed for the genes encoding for iron regulation. Cluster IV isolates harboured four or more plasmid encoded

iron related genes while there were only one or two of those genes in the isolates from cluster III. Clusters I and II isolates possessed a minimum of three iron related genes. Ironacquisition genes are considered to be important for the pathogenesis of avian colibacillosis particularly to evade humoral immunity (Rodriguez-Siek et al., 2005). We confirmed that there are gene combinations with a greater association with virulence to chickens as described previously (Johnson et al., 2008a; Schouler et al., 2012). The majority of the isolates in cluster IV were identified as highly virulent E. coli by the said gene combinations. As shown in Table 2, 16 of the 75 isolates had the combination of five genes identified as predictors of high virulence APEC by Johnson et al. (2008a) and (28) isolates had at least one of the gene combinations described by Schouler et al. (2012). Based on the distribution of these gene combinations (Johnson et al., 2008a; Schouler et al., 2012), we observed the presence of sitAP, ompT, hlyF and iroN can be used as the minimum predictor to identify high virulence APEC. Of the 55 clinical isolates, 30 possessed this gene combination including 14 isolates harboured the combination of five genes described in Johnson et al. (2008a). However, further research is warranted to confirm this scenario. Of the 55 clinical isolates we studied, only 35 carried at least one gene combination described previously as being associated with high or moderately virulent APEC (Johnson

D.R.A. Dissanayake et al. / Veterinary Microbiology 168 (2014) 403–412

et al., 2008a; Schouler et al., 2012). The designation of isolates into clinical and commensal is often depending on the site of isolation of the organism. It is not unusual for commensal E. coli to invade tissues as opportunistic pathogens or as post-mortem invaders and when isolated they are identified as clinical E. coli. There are a number of primary infections such as infectious bursal disease, infectious laryngotracheitis, Newcastle disease and Marek disease that can compromise immune system of chickens (Balamurugan and Kataria, 2006). Immunosuppression caused by these infections can pave the way to secondary infection caused by commensal E. coli. High number of phylogroup A strains reported in many studies including our own probably reflect the high incidence of secondary infections in poultry. 4.3. Disease association and public health risks of avian E. coli We expanded the phylogenetic assessment by incorporating STs representing APEC, UPEC and avian commensals listed in the MLST database, in order to understand the disease association of avian E. coli. A previous study (Moulin-Schouleur et al., 2007) has shown that some APEC strains are closely related to clones that include human ExPEC strains. Our in silico MLST analysis further confirmed this and there were a number of CCs (data not shown) including CC155 that contained both APEC and UPEC STs. The absence of STs and CCs exclusively consisting of APEC suggests that the APEC may not be host-specific. MLST study on E. coli O78 strains associated with variety of diseases in humans, animals and poultry has also shown that the closely related clones can reside in different hosts and are potentially zoonotic (Adiri et al., 2003). As shown in Mora et al. (2009), ST117 is a highly virulent emerging APEC with zoonotic potential. However, as mentioned earlier, E. coli strains with same ST may differ in the virulent profiles and the ability of those organisms to invade different hosts may also be different. As suggested previously (Clermont et al., 2011; Ron, 2006), subtle genetic changes (i.e. changes in adhesions) may help to develop host specificity even among the E. coli strains with the same ST. 4.4. Phylogenetic structure of avian E. coli: MLST vs. PCR based phylogrouping We used both triplex PCR described by Clermont et al. (2000) and MLST to identify phylogenetic structure of avian E. coli. As shown in previous studies, subpopulations identified by MLST were correlated with ECOR phylogroups (Johnson et al., 2006a; Wirth et al., 2006; Walk et al., 2007). However, there were some inconsistencies in PCR based phylogrouping. High proportion of isolates (45%) identified as phylogroup A by PCR were clustered together with phylogroup B1 strains. This was consistent with the findings of Gordon et al. (2008). We further observed that some of the PCR phylogroup A strains clustered with ECOR37, a previously known group E strain which was not assigned by Clermont triplex PCR. Previous studies have also indicated that many of the strains identified as group A organisms are actually not the

411

members of that group (Gordon et al., 2008; Walk et al., 2007). Further, the strains of group E have six of the eight virulence genes combinations (chuA, yjaA, TSEP4;   ,  + ,   +, + + +, +  + and +  ) described in Clermont et al. (2000). However, the existence of phylogroup E is still contentious. In the structure analysis we performed, those strains identified as phylogroup E clustered together with the subgroup of phylogroup B1(B1-I). MLST analysis divided the strains identified as phylogroup D by triplex PCR into two groups. Three isolates identified as phylogroup D by PCR were clustered together with the phylogroup B2 strain in the STRUCTURE analysis. Our results confirmed the study carried out by Gordon et al. (2008) who found that a number of strains identified as Clermont phylogroup D (chuA+, yjaA, and TSPE4+) belonged to phylogroup B2 based on MLST data. They suggest that all strains exhibiting the phylogroup D should be screened for the ibeA gene and if it is present, the strains are actually belonged to phylogroup B2. The present study confirmed that MLST is a better approach in identifying phylogroups of E. coli. Due to simplicity and inexpensive nature, Clermont triplex PCR is still useful in large scale screening of isolates. However, strains identified as phylogroup A by this method need further confirmation by MLST or other means. 5. Conclusions Avian E. coli isolates analysed in this study show high genetic diversity. Absence of dominant ST suggests that there is no predominant genotype of E. coli causing colibacillosis in chickens. Using both MLST and virulence genotyping, we have shown that most of phylogroup B2 and D strains were highly virulent avian E. coli. There is a small number of strains with high virulence gene profiles within phylogroup B1 and A. Strains of phylogroup B1 and a few phylogroup A strains possessed a number of iron related genes and ColV plasmid associated genes that enable them to become potential pathogens. The majority of avian E. coli isolates in phylogroup A lacked most of the virulence genes tested and may cause opportunistic infections in chickens. However, further studies are warranted to understand the nature of infections caused by E. coli with different gene combinations. We further identified the combination of sitAP, ompT, hlyF and iroN as being commonly associated with the E. coli causing septicaemia in chickens. The high diversity of APEC presents us significant challenges in controlling infections due to APEC. Conflict of interest None declared. Acknowledgments We would like to thank Professor Ian R Poxton, University of Edinburgh, UK for all his support. We acknowledge the use of the E. coli MLST database (http://mlst.ucc.ie/) in this study and Prof Mark Achtman

412

D.R.A. Dissanayake et al. / Veterinary Microbiology 168 (2014) 403–412

for assistance in MLST data submission. We gratefully acknowledge the funding support of Australian Research Council. References Adiri, R.S., Gophna, U., Ron, E.Z., 2003. Multilocus sequence typing (MLST) of Escherichia coli O78 strains. FEMS Microbiol. Lett. 222, 199–203. Balamurugan, V., Kataria, J., 2006. Economically important non-oncogenic immunosuppressive viral diseases of chicken – current status. Vet. Res. Commun. 30, 541–566. Ch’ng, S.H., Octavia, S., Xia, Q., Duong, A., Tanaka, M.M., Fukushima, H., Lan, R., 2011. Population structure and evolution of pathogenicity of Yersinia Pseudotuberculosis. Appl. Environ. Microbiol. 77 (3) 768–775. Clermont, O., Bonacorsi, S., Bingen, E., 2000. Rapid and simple determination of the Escherichia coli phylogenetic group. Appl. Environ. Microbiol. 66, 4555–4558. Clermont, O., Olier, M., Hoede, C., Diancourt, L., Brisse, S., Keroudean, M., Glodt, J., Picard, B., Oswald, E., Denamur, E., 2011. Animal and human pathogenic Escherichia coli strains share common genetic backgrounds. Infect. Genet. Evol. 11, 654–662. Deng, W., Liou, S.R., Plunkett III, G., Mayhew, G.F., Rose, D.J., Burland, V., Kodoyianni, V., Schwartz, D.C., Blattner, F.R., 2003. Comparative genomics of Salmonella enterica serovar Typhi strains Ty2 and CT18. J. Bacteriol. 185, 2330–2337. Dissanayake, D., Wijewardana, T., Gunawardena, G., Poxton, I., 2008. Distribution of lipopolysaccharide core types among avian pathogenic Escherichia coli in relation to the major phylogenetic groups. Vet. Microbiol. 132, 355–363. ˜Escobar-Paramo, P., Clermont, O., Blanc-Potard, A.B., Bui, H., Le BouguA nec, C., Denamur, E., 2004. A specific genetic background is required for acquisition and expression of virulence factors in Escherichia coli. Mol. Biol. Evol. 21, 1085–1094. ˜ o, E.M., Laturnus, C., Ewers, C., Li, G., Wilking, H., Kiessling, S., Alt, K., AntA Diehl, I., Glodde, S., Homeier, T., 2007. Avian pathogenic, uropathogenic, and newborn meningitis-causing Escherichia coli: how closely related are they? Int. J. Med. Microbiol. 297, 163. Excoffier, L., Laval, G., Schneider, S., 2005. Arlequin (version 3.0): an integrated software package for population genetics data analysis. Evol. Bioinform. Online 1, 47. Feil, E.J., Li, B.C., Aanensen, D.M., Hanage, W.P., Spratt, B.G., 2004. eBURST: inferring patterns of evolutionary descent among clusters of related bacterial genotypes from multilocus sequence typing data. J. Bacteriol. 186, 1518–1530. Felsenstein, J., 1993. PHYLIP (Phylogeny Inference Package), 3.5 c. Department of Genetics, University of Washington, Seattle. Gordon, D., Abajian, C., Green, P., 1998. Consed: a graphical tool for sequence finishing. Genome Res. 8, 195–202. Gordon, D.M., Clermont, O., Tolley, H., Denamur, E., 2008. Assigning Escherichia coli strains to phylogenetic groups: multi-locus sequence typing versus the PCR triplex method. Environ. Microbiol. 10, 2484– 2496. Jaureguy, F., Landraud, L., Passet, V., Diancourt, L., Frapy, E., Guigon, G., Carbonnelle, E., Lortholary, O., Clermont, O., Denamur, E., 2008. Phylogenetic and genomic diversity of human bacteremic Escherichia coli strains. BMC Genomics 9, 560. Johnson, J.R., Owens, K.L., Clabots, C.R., Weissman, S.J., Cannon, S.B., 2006a. Phylogenetic relationships among clonal groups of extraintestinal pathogenic Escherichia coli as assessed by multi-locus sequence analysis. Microbes Infect. 8, 1702.

Johnson, J.R., Russo, T.A., 2005. Molecular epidemiology of extraintestinal pathogenic (uropathogenic) Escherichia coli. Int. J. Med. Microbiol. 295, 383–404. Johnson, T.J., Siek, K.E., Johnson, S.J., Nolan, L.K., 2006b. DNA sequence of a ColV plasmid and prevalence of selected plasmid-encoded virulence genes among avian Escherichia coli strains. J. Bacteriol. 188, 745–758. Johnson, T.J., Wannemuehler, Y., Doetkott, C., Johnson, S.J., Rosenberger, S.C., Nolan, L.K., 2008a. Identification of minimal predictors of avian pathogenic Escherichia coli virulence for use as a rapid diagnostic tool. J. Clin. Microbiol. 46, 3987–3996. Johnson, T.J., Wannemuehler, Y., Johnson, S.J., Stell, A.L., Doetkott, C., Johnson, J.R., Kim, K.S., Spanjaard, L., Nolan, L.K., 2008b. Comparison of extraintestinal pathogenic Escherichia coli strains from human and avian sources reveals a mixed subset representing potential zoonotic pathogens. Appl. Environ. Microbiol. 74, 7043–7050. Liu, B., Wu, F., Li, D., Beutin, L., Chen, M., Cao, B., Wang, L., 2010. Development of a serogroup-specific DNA microarray for identification of Escherichia coli strains associated with bovine septicemia and diarrhea. Vet. Microbiol. 142, 373–378. Mora, A., Lo´pez, C., Dabhi, G., Blanco, M., Blanco, J.E., Alonso, M.P., Herrera, A., Mamani, R., Bonacorsi, S., Moulin-Schouleur, M., 2009. Extraintestinal pathogenic Escherichia coli O1: K1: H7/NM from human and avian origin: detection of clonal groups B2 ST95 and D ST59 with different host distribution. BMC Microbiol. 9, 132. Morales, C., Lee, M.D., Hofacre, C., Maurer, J.J., 2004. Detection of a novel virulence gene and a Salmonella virulence homologue among Escherichia coli isolated from broiler chickens. Foodborne Pathog. Dis. 1, 160–165. Moulin-Schouleur, M., Reperant, M., Laurent, S., Bree, A., Mignon-Grasteau, S., Germon, P., Rasschaert, D., Schouler, C., 2007. Extraintestinal pathogenic Escherichia coli strains of avian and human origin: link between phylogenetic relationships and common virulence patterns. J. Clin. Microbiol. 45, 3366–3376. Octavia, S., Lan, R., 2006. Frequent recombination and low level of clonality within Salmonella enterica subspecies I. Microbiology 152, 1099–1108. Picard, B., Garcia, J.S., Gouriou, S., Duriez, P., Brahimi, N., Bingen, E., Elion, J., Denamur, E., 1999. The link between phylogeny and virulence in Escherichia coli extraintestinal infection. Infect. Immun. 67, 546–553. Pritchard, J.K., Stephens, M., Donnelly, P., 2000. Inference of population structure using multilocus genotype data. Genetics 155, 945–959. Reeves, P.R., Farnell, L., Lan, R., 1994. MULTICOMP: a program for preparing sequence data for phylogenetic analysis. Comput. Appl. Biosci. 10, 281–284. Rodriguez-Siek, K.E., Giddings, C.W., Doetkott, C., Johnson, T.J., Fakhr, M.K., Nolan, L.K., 2005. Comparison of Escherichia coli isolates implicated in human urinary tract infection and avian colibacillosis. Microbiology 151, 2097–2110. Ron, E.Z., 2006. Host specificity of septicemic Escherichia coli: human and avian pathogens. Curr. Opin. Microbiol. 9, 28–32. Schouler, C., Schaeffer, B., Bre´e, A., Mora, A., Dahbi, G., Biet, F., Oswald, E., Mainil, J., Blanco, J., Moulin-Schouleur, M., 2012 b. Diagnostic strategy for identifying avian pathogenic Escherichia coli based on four patterns of virulence genes. J. Clin. Microbiol. 50, 1673–1678. Walk, S.T., Alm, E.W., Calhoun, L.M., Mladonicky, J.M., Whittam, T.S., 2007. Genetic diversity and population structure of Escherichia coli isolated from freshwater beaches. Environ. Microbiol. 9, 2274–2288. Wirth, T., Falush, D., Lan, R., Colles, F., Mensa, P., Wieler, L.H., Karch, H., Reeves, P.R., Maiden, M.C.J., Ochman, H., 2006. Sex and virulence in Escherichia coli: an evolutionary perspective. Mol. Microbiol. 60, 1136–1151.

!