Occurrence of Putative Pathogenicity Islands in Enterococci from Distinct Species and of Differing Origins

Abstract

Enterococci isolated from ewe's milk and cheese, clinical isolates of human and veterinary origins, and reference strains obtained from culture collections were screened for the occurrence of putative pathogenicity island (PAIs). Results obtained after PCR amplification and hybridization point toward PAI dissemination among enterococci of diverse origins (food/clinical) and species (Enterococcus faecalis/non-E. faecalis).

Enterococci are ubiquitous microorganisms traditionally viewed as commensal bacteria but now established as major community-acquired and nosocomial pathogens causing a variety of infections. They are opportunistic pathogens and generally exhibit a low level of pathogenicity; however, in recent years, several potential enterococcal virulence factors have been identified but their exact role in the pathogenesis mechanisms is still not fully established (2, 12, 13).

Genomic islands are large genomic regions which frequently harbor mobility loci and gene clusters with specific functions. Genomic islands harboring genes that contribute, directly or indirectly, to the pathogenic potency of the harboring microorganism are termed pathogenicity islands (PAIs). The generalized sequencing of genomes revealed that PAIs are much more widespread than previously thought (3).

In the genus Enterococcus, two PAIs have been described: one in a vancomycin-resistant Enterococcus faecalis strain responsible for a nosocomial outbreak (15) and the other in nosocomial E. faecium strains (4). More recently (9), the worldwide dissemination of the E. faecalis PAI among isolates from diverse locations and different origins was demonstrated. However, the majority of the studies previously performed analyzed clinical isolates of the species E. faecalis. So, in the present investigation we broadened the range of origins (food/clinical) and species (E. faecalis/non-E. faecalis) with the aim of understanding if the genus Enterococcus constitutes a risk or if the capacity to infect humans and cause disease is limited to specific species or origins.

In this study, 40 enterococcal isolates previously identified (1) and representing distinct virulence potential (12, 13) were screened for the presence of putative PAIs by a sequential polyphasic approach.

First, all enterococci were screened by PCR for PAI-related genes (see Table S1 in the supplemental material). The analyzed loci represent diverse regions of the E. faecalis PAI (15) and include the esp and araC-like genes also present in the putative PAI identified in E. faecium (4). All isolates showed between 11 and 22 genes (Table 1 and data not shown), with food strain QSE123 presenting the lowest incidence and four strains harboring the majority of the regions under analysis (22 genes). Eight of the analyzed loci were detected in more than 90% of the enterococci, pointing to a high conservation of these regions. Only locus ef0571 was detected in less than 10% of the isolates.

TABLE 1.

Incidence of PAI-related loci among the food, clinical, and reference enterococci tested in this study

Locus (gene)	No. (%) of strains with locus				Locus presencea
	Reference strains	Food strains	Clinical strains	Total
ef0481	20 (100)	13 (93)	6 (100)	39 (98)	+
ef0482	20 (100)	13 (93)	6 (100)	39 (98)	+
agg	20 (100)	14 (100)	6 (100)	40 (100)	+
ef0495	20 (100)	13 (93)	6 (100)	39 (98)	+
cbh	10 (50)	3 (21)	6 (100)	19 (48)	d
cylLl	1 (5)	4 (29)	1 (17)	6 (15)	d
cylLs	2 (10)	6 (43)	1 (17)	9 (23)	d
cylM	9 (45)	8 (57)	3 (50)	20 (50)	d
cylB	9 (45)	10 (71)	5 (83)	24 (60)	d
cylA	9 (45)	9 (64)	5 (83)	23 (58)	d
cylI	15 (75)	9 (64)	4 (67)	28 (70)	d
esp	15 (75)	6 (43)	3 (50)	24 (60)	d
araC (ef0530)	17 (85)	10 (71)	4 (67)	31 (78)	d
ef0534	20 (100)	14 (100)	6 (100)	40 (100)	+
ef0539	19 (95)	12 (86)	6 (100)	37 (93)	+
ef0543	19 (95)	11 (79)	6 (100)	36 (90)	+
ef0551	11 (55)	9 (64)	5 (83)	25 (63)	d
xylA (ef0556)	10 (50)	9 (64)	4 (67)	23 (58)	d
ef0571	1 (5)	0 (0)	1 (17)	2 (5)	−
ef0590	20 (100)	14 (100)	6 (100)	40 (100)	+
gls24-like (ef0604)	16 (80)	10 (71)	6 (100)	32 (80)	d
ef0609	16 (80)	8 (57)	4 (67)	28 (70)	d
ef0617	13 (65)	7 (50)	6 (100)	26 (65)	d
ef0628	7 (35)	2 (14)	3 (50)	12 (30)	d

^a+, locus present in ≥90% of the strains; d, locus present in 11 to 89% of the strains; −, locus present in ≤10% of the strains.

In 2005, Nallapareddy and coworkers (9) analyzed 454 E. faecalis isolates and identified isolate-specific deletions in four PAI regions: one included cylM, araC, and ef0534 and was also previously described for E. faecalis V583 (15); the second included ef0530 and ef0534; the third corresponded to region ef0571; and the fourth included loci ef0604 to ef0609. PAI variability was also reported by other authors (7, 8) and corroborates the findings observed in the present study, where deletions were found throughout the PAI, with major incidence among loci cbh to esp, ef0551 to ef0571, and ef0628. After chi-square analysis of contingency tables, a statistically significant association was found between the presence of the gene cbh and clinical origin (P = 0.00525). Distinct results were observed in previous studies (9) where the esp, xylA, and gls24-like genes were enriched in infection-derived isolates. The differences observed are probably due to sample size effects (454 isolates of E. faecalis versus 40 of enterococci).

In conclusion, after PCR screening for PAI-related genes, 13 putative PAI-harboring enterococci were selected, including reference, clinical human/veterinarian, and food isolates.

Subsequently, a genome-walking strategy was applied to establish the genetic organization of the putative PAIs by comparison with the positive controls included in this study (MMH594, E. faecalis complete PAI; V583, E. faecalis PAI with a deletion [15]; E300, E. faecium putative PAI [4]). The results obtained with 33 primer combinations (see Table S2 in the supplemental material) are summarized in Fig. 1 and show that the majority of the strains possess more than 20 of the analyzed loci, with the least-conserved regions being located between the genes cbh and esp, pointing to this region as a “deletion hot spot” and suggesting that some deletions are more favored than others, as can be inferred from the results reported by other authors (7, 8, 9, 15). Excluding the PAI-harboring control strains, the isolate harboring more PAI-related genes was LN11, an E. casseliflavus ewe's milk isolate with an amplification profile almost 100% identical to that of MMH594. The diversity of genetic organizations observed in the putative PAIs is probably a reflection of the enterococcal ability to acquire and share material with other bacteria inhabiting the same ecological niche.

FIG. 1.

Genetic organization of the putative enterococcal PAIs (dendrogram obtained with the BioNumerics software, version 4.61 [Applied Maths, Kortrijk, Belgium], Pearson's correlation coefficient, and the unweighted-pair group method using average linkages). Gray fill represents positive amplification of the corresponding PAI region (PAI-harboring E. faecalis MMH594 is in black for comparison). The numbers 1 to 33 correspond to the regions or loci in Table S2 in the supplemental material.

Subsequently, to test for physical linkage of the virulence-related genes in the same region of the bacterial chromosome, total cell DNA was cleaved with restriction enzymes (I-Ceu-I, SfiI, and SmaI), the macrorestriction fragments were separated by pulsed-field gel electrophoresis (PFGE; contour-clamped homogeneous electric field with pulse times increasing linearly from 5 to 35 s for 22 h at 220 V), transferred to nylon membranes by Southern blotting (11), and hybridized under high-stringency conditions (probes and signals detected with Dig-High Prime DNA labeling and detection starter kit II as recommended by the manufacturer (Roche Diagnostics).

The restriction enzyme I-Ceu-I recognizes a unique sequence within the 23S rRNA (5), so cleavage of total DNA, followed by hybridization with a 16S rRNA gene probe (6), highlights the fragments corresponding to genomic DNA, allowing the distinction between chromosomal DNA and extrachromosomal elements. When rehybridization of the same membrane with a PAI gene as the probe reveals a positive hybridization signal in the same fragment, this demonstrates its chromosomal nature. Subsequent hybridizations with other PAI-related probes confirm the physical linkage of those loci in the same region of the bacterial chromosome, pointing to the presence of putative PAIs.

Figure 2 shows the results obtained after DNA cleavage with I-Ceu-I (Fig. 2, left panel), followed by hybridization with the 16S rRNA gene (Fig. 2, center panel) and cylMBA (Fig. 2, right panel) probes, which demonstrate the chromosomal location of the cyl operon in veterinary clinical strain V434 and food isolates LN11, LA78, and QCB54. Concerning the other cyl-positive enterococci included in this investigation, very weak or no hybridization signals were obtained (data not shown), suggesting that such traits must be carried by plasmids.

FIG. 2.

Macrorestriction profile obtained with the endonuclease I-Ceu-I (left panel) and corresponding hybridization with 16S rRNA gene (center panel) and cylMBA (right panel) probes. Lanes 1 and 18, λ ladder PFGE marker (New England BioLabs); lane 2, DSMZ 20680^T (the superscript T indicates the type strain); lane 3, DSMZ 5633^T; lane 4, V583; lane 5, MMH594; lane 6, E300; lane 7, H1881; lane 8, V95; lane 9, V434; lane 10, LN9; lane 11, LN11; lane 12, QSE123; lane 13, LA78; lane 14, LA160; lane 15, QA29a; lane 16, QA40; lane 17, QCB54.

Subsequently, Southern blot assays of SfiI and SmaI macrorestriction profiles were hybridized with esp and cylMBA (Fig. 3, left and right panels, respectively). Hybridization of the two probes within the same-size macrorestriction fragment was observed for strains MMH594 and LN11, pointing to the physical linkage of these PAI-related genes in ewe's milk E. casseliflavus isolate LN11. For V583, with a deletion including this region, as expected, no hybridization signal was observed.

FIG. 3.

Hybridization signals obtained by using the PAI-related genes esp (left panel) and cylMBA (right panel) as probes. Left panel: lanes 1 and 17, λ ladder PFGE marker (New England BioLabs); lanes 2 to 8, MMH594, E300, V434, LN9, LN11, LA78, and QA40 cleaved with SfiI; lanes 9 to 16, V583, E300, V434, LN9, LN11, LA78, QA40, and QCB54 cleaved with SmaI. Right panel: lanes 1 and 19, λ ladder PFGE marker (New England BioLabs); lanes 2 to 10, MMH594, E300, V434, LN9, LN11, LA78, QA40, QCB 54, and V583 cleaved with SfiI; lanes 11 to 18, MMH594, E300, V434, LN9, LN11, LA78, QA40, and QCB54 cleaved with SmaI.

Positive amplification with primer combination EF-1/ER-2, directed to specific regions of E. faecalis V583 located before and after its PAI (10), demonstrated that in strains V95 and QSE123, either the PAI is absent or its insertion occurred at a distinct location. For the other 11 putative PAI-harboring enterococci, primer combinations EF1/ER1 (before the PAI/at the beginning of the PAI) and EF2/ER2 (at the end of the PAI/outside of the PAI) confirmed a PAI insertion site similar to that of MMH594 and V583, as also reported for other enterococci (7, 8, 9, 14).

Although there are many differences within this genomic region of the isolates under analysis, the overall results point toward PAI dissemination among enterococci from diverse origins (ewe's milk/cheese) and species (E. faecalis, E. faecium, E. casseliflavus, E. raffinosus, E. durans, and E. hirae), suggesting a wide pathogenicity potential of the genus Enterococcus.