Genetic Comparison of Enterococcus Species Isolated from Osteomyelitis Lesions and the Barn Environment of Successive Broiler Chicken Flocks Open Access

Enterococcal osteomyelitis/spondylitis is an emerging disease affecting broiler chickens worldwide (1,2,3,4,5,6). The disease is characterized by an infection of the free thoracic vertebral body (T4), which leads to compression of the spinal cord, lameness, reduced mobility, and typical paralysis. Affected birds often die of dehydration or inanition when not culled (2). Osteomyelitis in various bones as well as septicemic lesions such as polyserositis are also observed, with all clinical presentations resulting in lack of flock uniformity, variable mortality, and increased condemnations rates (7).

Initially identified as Streptococcus cecorum in 1983 (8) and reclassified in 1989 to the Enterococcus genus based on 16S ribosomal RNA (rRNA) gene sequencing (9), Enterococcus cecorum has been the bacterium most frequently implicated in outbreaks of vertebral and nonvertebral osteomyelitis in broiler and broiler breeders (2,3,5,10,11). Other species such as Enterococcus faecium, Enterococcus faecalis, Enterococcus hirae, Enterococcus durans, Enterococcus gallinarum, Enterococcus casseliflavus, and Enterococcus avium have also been isolated from broiler chickens affected with osteomyelitis and other septicemic lesions (6). Enterococcus spp. are Gram-positive bacteria, and they are facultative anaerobic, nonmotile, non-spore-forming organisms with diverse biochemical properties (7). They are ubiquitous in nature, with a worldwide distribution among avian species, and they are considered part of the normal gut microbiota of chickens commonly found in poultry house environments.

The epidemiology of transmission is unknown. There is no current evidence of vertical transmission (4,12), and experimental infection models suggest early infection of young chicks via an oral route (10,13,14). A contaminated environment, and hence horizontal transmission, would seem to play an important role, one that could be tempered by the cleaning and disinfection of the poultry house (15).

A few comparative genomic analyses of pathogenic E. cecorum strains have been published. Six genes associated with clinical E. cecorum isolates were identified in poultry by Laurentie et al. (16). The presence of two or more of these six genes discriminated 94% of their isolates associated with disease in chickens from other E. cecorum isolates. Borst et al. (17) also described other potential virulence-associated determinants such as a capsular locus as well as cell wall–associated proteins.

Interestingly, recurring outbreaks in the same house have been reported by poultry practitioners, which could suggest persistence of a pathogenic E. cecorum isolate in the environment. Such a house was initially visited in July 2020, and E. cecorum and E. faecalis were isolated from osteomyelitis lesions. Surprisingly, the next flock revealed the presence of E. faecalis and E. raffinosus in the lesions, but no E. cecorum was isolated.

To better understand the epidemiology of these outbreaks, verify the persistence of pathogenic isolates, and identify the possible horizontal transfer of genetic material between Enterococcus species in successive flocks, our objectives were 1) to assess whether the enterococci isolates isolated during the first visit were clonal to those isolated during the second visit, 2) to identify the presence of mobile genetic elements (MGEs), virulence factor genes (VFs), and antimicrobial resistance genes (AMGs) in the enterococci isolates isolated, and 3) to determine whether there were MGEs common to the different enterococci.

The farm comprises four three-story broiler barns housing between 15,000 and 33,000 birds each. All buildings are fully automated, and litter is taken out between each lot with a downtime period of approximately 2 wk. Chicks are from the same hatchery, and they are delivered to the farm and shipped to slaughter on the same day. A first episode of E. cecorum was diagnosed in one barn in May 2019. All barns were affected as of September 2019. Since then, penicillin or amoxicillin has regularly been administered in the first 2 wk of rearing via in-feed supply or via drinking water to control the infection, but any attempt to stop antimicrobial use is eventually met with a resurgence of the disease.

To identify a potential on-farm reservoir of E. cecorum, we collected multiple samples, using the same protocol, on Visits A and B from a single barn (Barn 4) on July 13, 2020, and August 24, 2020, when birds were 21 days of age. Series of swabs from lesions (vertebra or tibia) and dust from various environmental surfaces (walls, feeders, fans, litter) were submitted on ice to the Chair in Poultry Research Laboratory at the Faculty of Veterinary Medicine, University of Montreal. On Visit A, one E. cecorum isolate (n = 1) and two E. faecalis isolates (n = 2) were isolated from lesions, whereas one E. faecalis isolate (n = 1) was isolated from the feeder. On Visit B, 12 E. faecalis isolates (n = 12) were isolated from four tibias presenting lytic foci in the proximal physis, and one E. raffinosus isolate (n = 1) was isolated from a vertebral abscess, whereas one E. raffinosus isolate (n = 1) was isolated from dust collected on the wall.

For the detection of E. cecorum, multiple broth enrichment/selection steps followed by culture on a medium selective for Gram-positive bacteria were done. Briefly, all samples were enriched in 50 ml of Tryptic Soy Broth (Thermo Fisher Scientific, Nepean, Ontario, Canada) and incubated overnight at 37 C with 5% CO₂. Two-milliliter aliquots were then put into 18 ml of Todd Hewitt Broth (Thermo Fisher Scientific) with 1% yeast extract (Thermo Fisher Scientific) and amikacin (Sigma-Aldrich Canada Ltd., Toronto, Ontario, Canada) at a final concentration of 64 μg/ml, incubated overnight, and transferred onto Columbia colistin nalidixic acid supplemented with 5% sheep blood agar plates before an overnight incubation in the same conditions. Characteristic colonies were streaked onto brain heart infusion agar plates and incubated overnight at 37 C with 5% CO₂, and three catalase-negative colonies per agar plate were transferred onto blood agar plates for species confirmation using matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry after incubation.

All isolates confirmed by MALDI-TOF to be Enterococcus species were selected. The genomic DNA of one isolate of E. cecorum, 15 isolates of E. faecalis, and two isolates of E. raffinosus was extracted using the Lucigen MasterPure™ Gram Positive DNA purification kit (LGN-MGP04100; Mandel Scientific Company Inc., Guelph, Ontario, Canada) following the manufacturer’s instructions. After adjusting all samples at 50 ng/μl in a final volume of 20 μl, samples were sent to the Genome Quebec Centre (Montreal, Quebec, Canada) sequencing center and sequenced on a NovaSeq6000 SP instrument (Illumina, San Diego, CA, USA) to generate 150-bp paired-end reads.

Genomes were assembled using the following method. Briefly, reads were trimmed to remove Illumina adapters and poor-quality reads using Scythe v0.994 and Sickle v1.33 (18,19). Read quality was assessed before and after trimming using FastQC v0.12.0 (20) to ensure high-quality reads were obtained prior to assembly. Genome assembly was performed using SPAdes v3.15.2 (21), and protein annotation was done using Prokka v.1.14.6 with default parameters (22). Genome size, GC content (guanine-cytosine content), N50 values (average length of a set of sequences), and number of coding sequences (CDS) were evaluated using Quast v5.2.0 (23) and are summarized in Table 1.

Multilocus sequence typing (MLST) of E. faecalis isolates was achieved using the typing tool of PubMLST (24) with the E. faecalis typing database. Briefly, the nucleotide sequences of seven housekeeping genes (gdh, gyd, pstS, gki, aroE, xpt, and yqiL) from the genome of E. faecalis isolates were compared to the typing database to determine their allelic profile. Sequence types (STs) of isolates were determined according to their allelic profile. Only exact matches (7/7) were considered.

A pangenome analysis of the E. faecalis genomes was performed using the bioinformatic tool Roary v3.12.0 (25). Roary’s pipeline was run using the GFF3 files generated during the protein annotation by Prokka as input with -s (do not split paralogs) and -n (generate a core genome alignment) options. The core, accessory, and unique genomes were defined by genes present in ≥99% isolates, 98% ≥ isolates >1%, and a single isolate of E. faecalis. The alignment of the core genome was done using MAFFT v7.407 (26). A maximum-likelihood tree was then constructed using a general time reversible model of nucleotide substitution with a gamma model of rate heterogeneity based on the core genome alignment mentioned above using the tool raxml-ng v1.2.0 (27). Briefly, 25 random and 25 parsimony trees were constructed by raxml-ng. Then, the tree with the best likelihood value was selected and supported by a bootstrap value of 1000 replicates with a cutoff value of 0.03. The phylogenetic tree was visualized using iTOL v6.8 (28).

The genomes of E. cecorum, E. faecalis, and E. raffinosus were screened for the presence of ARGs, VFs, and MGEs. Detection of ARGs among the genomes was performed using the bioinformatic tool Abricate v0.8.10 (29) with the highly curated MEGARes database (30). An additional screening was done at the protein level using the online Resistance Gene Identifier (RGI) tool from the Comprehensive Antibiotic Resistance Database (CARD) (31), and detection of mutational resistance was performed using the webtool ResFinder v.4.6.0 with thresholds of 90% and 80% for homology and coverage parameters (32,33). Detection of VFs was achieved by comparing the proteomes of all isolates with the protein sequence of VFs from the Virulence Factor Database (VFDB) (34). Briefly, a BLASTp alignment of the annotated proteomes against the protein database of VFDB was performed for all isolates. Another BLASTp analysis was performed to screen the genes associated with avian clinical isolates of E. cecorum identified by Laurentie et al. (16) and a gene part of the capsular locus identified by Borst et al. (17) and referred as the cpsO gene (35) within the genome of E. cecorum isolate CECO0008. For both analyses, only hits with percent identity >90% with a query coverage percentage >80% and an e value <1e⁻¹⁰ were considered as VFs. Detection of MGEs including insertion sequences (IS), plasmids, and prophages was done. Genomes were screened for IS using the bioinformatic tools ISEScan v1.7.2.3 (36) and MobileElementFinder v1.1.2 (37) with default parameters. Plasmid detection was achieved by first separating plasmids from chromosome contigs using the bioinformatic tool Plasmer v0.1 (38) with default parameters. Plasmid contigs were then submitted to the search tool of PLSDB (39) using Mash screen as the search strategy with default parameters and the winner-takes-all option to remove redundancy. Prophage detection was achieved using the command line version of the PHAge Search Tool with Enhanced Sequence Translation (PHASTEST) (40), and only intact prophages (score >90) were considered. The presence of ARGs, the VFs, and the prevalence of MGEs for each isolate are indicated in Fig. 1.

The total number of reads per isolate ranged between 4.16 and 6.40 million with an average number of 5.02 million reads per isolate. Two E. faecalis isolates were withdrawn from the study because the sequencing reads and the resulting assemblies were of low quality. E. cecorum isolate CECO0008 had a genome of 2.22 Mbp with a GC content of 36.56% and 2161 CDS (Table 1). The average genome size of E. faecalis isolates was 2.34 ± 0.23 Mbp, with the smallest and largest genomes having a genome size of 2.27 Mbp and 3.22 Mbp, respectively. The average GC content of E. faecalis genomes was 37.35 ± 0.43%, whereas the minimum and maximum GC contents observed were 36.89% and 38.57%, respectively. The average CDS observed within E. faecalis genomes was 2863 ± 130 CDS, whereas the least and the greatest numbers of CDS detected were 2697 and 3101 CDS, respectively. The average genome size of E. raffinosus isolates was 4.14 ± 0.28 Mbp, while its average GC content was 40.1 ± 0.64%. The average number of CDS within E. raffinosus genomes was 4448 ± 349 CDS.

According to the phylogenetic analysis, isolates recovered during Visit A were distinct from those of Visit B as no cluster was composed of isolates from both visits was observed (Fig. 1). Moreover, the ST profile of isolates from Visit A (ST 1500) was different from the ST profiles of isolates from Visit B (ST 207, 227, 300, 314, and 841). Nonetheless, isolates from Visit B with the profiles ST 207 and 841 were the closest relatives to isolates from Visit A. Isolates ENT0382, ENT0383, and ENT0384 isolated from the tibia of Bird 7 were clonal based on their core genome, ST profile, and accessory genome (ARGs, VFs, and MGEs). Isolates ENT0380 and ENT0381 isolated from the tibia of Bird 6 were closely related, but ENT0380 lacked the VF fss2. Although the isolates isolated from different sources of Visit A seemed clonal at the core genome level, differences at the accessory genome level were observed, notably the presence of the VFs fliE and fss1, as well as the total number of MGEs detected. Isolates ENT0376, ENT0377, and ENT0379 were clonal based on their core genome, ARGs, and VFs, but the number of identified MGEs among the genomes of these isolates was different. The population of E. faecalis isolates isolated from lesions was mostly heterogenous as nonclonal isolates were isolated from the same lesion. For example, isolates of ST 207 (n = 1) and ST 841(n = 2) recovered from the tibia of Bird 5 and isolates of ST 300 (n = 2) and ST 841 (n = 1) recovered from the tibia of Bird 6 were not clonal.

Two plasmids (NZ_CP064416.1 and LR962096.1) were identified in E. cecorum CECO0008, whereas seven plasmids (NZ_CP086561.1, LR962096.1, LR962516.1, NZ_CP135067.1, NZ_KY513281.1, NZ_MW647491.1, and NZ_OP046174.1) and six plasmids (LR962516.1, NZ_CP081847.1, NZ_MW647491.1, LR962096.1, NZ_CP135067.1, and CP101347.1) were identified in E. faecalis and E. raffinosus isolates, respectively. The plasmid LR962096.1 was found in all three Enterococcus species, although it does not appear to encode any genes other than a putative transposase previously identified in E. faecalis (GenBank accession no. AAL05546.1). The plasmids LR962516.1, NZ_CP135067.1, and NZ_MW647491.1 were shared among our E. faecalis and E. raffinosus isolates. Although no ARG was encoded by the plasmid LR962516.1, the plasmids NZ_CP135067.1 and NZ_MW647491.1 encoded the ARGs erm(B) and bla_TEM-116, respectively. A single intact phage was detected within the genome of CECO0008, whereas 10 and two intact phages were detected within the genomes of E. faecalis and E. raffinosus, respectively. One phage (PHAGE_Strept_5093_NC_012753) was detected in both E. cecorum and E. faecalis species, and one phage (PHAGE_Entero_vB_IME197_NC_028671) was detected in both E. faecalis and E. raffinosus species.

In total, three ARGs (lnuC, tetL, and tetM) were detected in isolate CECO0008, potentially conferring resistance to lincomycin and tetracycline, whereas 19 ARGs [ant(6)-Ia, aph(3’)-IIa, aph(3’)-III, bla_TEM-116, ble_Tn5, catA1, dfrE, dfrG, efrA, efrB, ermB, gyrA, lsa(A), narA, narB, parC, tet(L), tet(M), tet(O)] were detected among E. faecalis isolates. These ARGs are associated with resistance to aminoglycosides, beta-lactams, glycopeptides, phenicols, disinfectants, macrolides, rifamycin, fluoroquinolones, lincosamides, streptogramins, ionophores, and tetracyclines. Eight ARGs [aph(3’)-IIa, bla_TEM-116, ble_Tn5, catA1, ermB, marA, narA, and narB] were detected among E. raffinosus isolates. These ARGs are associated with resistance to aminoglycosides, beta-lactams, glycopeptides, phenicols, macrolides, lincosamides, streptogramins, and ionophores. Interestingly, isolate ENT0071 also carried marA, a gene encoding a transcription regulator of the AraC family that regulates the expression of several proteins leading to an overall reduction of the sensitivity to antibiotics (41,42,43).

Two VFs, CIRMBP1228_00573 and CIRMBP1228_00586, both identified by Laurentie et al. (16) in their poultry clinical isolates, as well as cpsO, identified in pathogenic strains by Borst et al. (17,35), were detected within the genome of the E. cecorum isolate CECO0008. In total, 39 VFs were detected among the E. faecalis isolates. The average number of VFs detected per E. faecalis isolate was 25 ± 7 VFs, with isolates ENT0373, ENT0383, ENT0383, and ENT0384 having the least VFs (n = 18) and isolates ENT0376, ENT0377, ENT0379, and ENT0381 having the most VFs (n = 34). Two clusters of isolates on the phylogenetic tree with more than 30 VFs were observed (Fig. 1). The first cluster was composed of isolates ENT0380 and ENT0381, whereas the second cluster was composed of isolates ENT0376, ENT0377, and ENT0379. In total, seven VFs were detected among E. raffinosus isolates. The average number of VFs detected per E. raffinosus isolate was 6 ± 1 VFs, with isolate ENT0007 having the least VFs (n = 5) and isolate ENT0071 having the most VFs (n = 7). No VF was common to all Enterococcus species.

E. cecorum osteomyelitis lesions are commonly observed in broiler chicken flocks in Québec, Canada. In one episode of vertebral and nonvertebral osteomyelitis in a broiler chicken flock, E. cecorum and E. faecalis were both isolated from the same vertebral abscess. Surprisingly, during a recurring episode of osteomyelitis in the following lot, housed in the same barn, E. faecalis was the bacteria most isolated from tibial lytic foci, while E. raffinosus was isolated from a vertebral abscess. To verify whether VFs as well as AMGs originating from the original E. cecorum isolate could have been horizontally transferred to other Enterococcus species via MGEs, whole-genome sequencing of 16 Enterococcus species isolates (1 E. cecorum, 13 E. faecalis, and 2 E. raffinosus isolates) from osteomyelitis lesions, feeder surfaces, and litter surfaces from these two consecutive broiler chicken flocks was carried out and compared using various bioinformatic tools.

Genomic analysis revealed that more than one Enterococcus species could be isolated from the same lesion site. Surprisingly, isolates of the same species isolated from the same lesion were genetically distinct. Indeed, E. faecalis isolates from lesions were mostly heterogeneous, as nonclonal isolates were isolated from the same lesion. This observation suggests that several E. faecalis isolates may be involved in lesion development rather than one dominant pathogenic isolate. E. faecalis is known to be involved in polymicrobial infections and has been shown to enhance the virulence of other pathogens (44). In this study, its presence without other pathogens suggests that E. faecalis might be an opportunistic bacterium when predisposing factors support its colonization and growth.

Although the isolates obtained from different sources during Visit A appeared to be clonal at the core genome level, some differences were observed at the accessory genome level, in particular the VFs fliE and fss1 and the total number of MGEs detected. Genes surrounding fliE are involved with the conjugation machinery, which suggests that this VF is encoded onto a plasmid. Consequently, the differences observed in the accessory genome of these isolates may be explained by a different number of MGEs.

The presence of plasmids common to the enterococci species in this study suggested that the horizontal transfer of genetic material between them can happen, but further research is necessary to verify this hypothesis. For example, it has been shown that E. faecium and E. faecalis can exchange conjugative plasmids such as the pheromone-responsive plasmid pMG2200 that encodes vancomycin resistance (45). E. faecalis is notorious for its ability to share genetic content within and across species, which also explains the genomic diversity observed among our isolates (44). Moreover, the fact that the plasmids identified in this study were first identified in bacterial genera other than Enterococcus, such as Salmonella enterica, Streptococcus suis, and Bacillus safiensis, also points to the evidence that enterococci species can acquire foreign DNA from the various species found in the intestinal microbiota of chickens and the barn environment.

Phages were also detected in all three Enterococcus species. Phages are a known source of horizontal gene transfer through transduction, a process during which nonviral DNA is trapped inside a virus particle and transferred from the bacterial host cell to another cell via viral infection (46). Although phages normally infect a narrow host range, some can infect different species and even different genera of bacteria (47), as observed in this study. Phages are likely another passive means by which to exchange genetic material between enterococci species and other organisms present in the intestinal microbiota or the barn environment. Because no ARG or VF was associated with the phages detected in this study, it was not possible to confirm this hypothesis.

Multidrug resistance has been defined as the acquired nonsusceptibility to at least one agent in three or more antimicrobial categories (48). According to that definition, all E. faecalis and E. raffinosus isolates of this study were multidrug-resistant forms, whereas the E. cecorum isolate CECO0008 was not. It is somewhat concerning to observe that enterococci isolated not only from osteomyelitis lesions but also from the barn environment showed up to 11 ARGs in total. The early and preventive administration of penicillin or amoxicillin is regularly used in Quebec, Canada, as a means to control E. cecorum outbreaks. The presence of the bla_TEM-116 gene, encoded on plasmid and associated with the resistance to beta-lactams in Clostridium perfringens (49), could potentially see the end of this practice. For example, this plasmid-encoded ARG could be transferred to a larger enterococci population as observed in this study, with the E. faecalis and E. raffinosus isolates sharing this plasmid. However, further studies are necessary to verify if the presence of this gene indeed results in phenotypic resistance.

More worrisome, however, is the large number of ARGs detected in both the pathogenic and environmental E. faecalis isolates. Furthermore, except for dfrE, efrAB, and lsa(A), all the ARGs detected in this study were associated with MGEs and could therefore potentially be disseminated in the environment. The impact of this observation is yet to be assessed, but enterococci species as gram-positive markers for antimicrobial resistance should be considered as an interesting ARGs reservoir, since mostly gram-negative bacteria, i.e., Salmonella spp., Campylobacter spp., and Escherichia coli, are monitored in national antimicrobial resistance surveillance programs (50,51).

The numbers of VFs did vary from 1 to 34 between E. faecalis and E. raffinosus isolates. E. cecorum was only isolated once from a lesion swabbed during the first visit. No VF from the VFDB database was detected for E. cecorum CECO0008. However, two of the six genes identified by Laurentie et al. (16), CIRMBP1228_00586 and CIRMBP1228_00573, as well as cpsO identified by Borst et al. (17,35), all associated with clinical poultry E. cecorum isolates, were detected in this study. E. faecalis is a well-characterized pathogen in human medicine, and its extensively studied genome has led to the discovery of multiple VFs (52). Therefore, it was expected that a higher number of VFs would be detected among E. faecalis isolates. Given that E. faecalis, E. raffinosus, and likely E. cecorum are equipped with various virulence and survival mechanisms, it is not surprising to see them adapting to and colonizing various tissues.

In conclusion, persistence of a pathogenic E. cecorum isolate and horizontal transmission of VFs to other Enterococcus species from one broiler chicken lot to a following one were not demonstrated. However, given the number of different MGEs shared between the Enterococcus isolates sequenced and because numerous VFs involved in the pathogenesis of vertebral and nonvertebral osteomyelitis have yet to be identified, this possibility cannot be entirely dismissed.

We wish to thank the Innov’Action Agroalimentaire program from the Quebec Department of Agriculture, Fisheries and Food (MAPAQ) for financial support (grant no. IA120627).

BioProject ID: PRJNA1208540

ARGs =

antimicrobial resistance genes;

CDS =

number of coding sequences;

GC content =

guanine-cytosine content;

IS =

insertion sequences;

MALDI-TOF =

matrix-assisted laser desorption/ionization time-of-flight;

MGEs =

mobile genetic elements;

MLST =

multilocus sequence typing;

rRNA =

ribosomal RNA;

ST =

sequence type;

VFs =

virulence factor genes;

VFDB =

virulence factor database