Clostridium difficile erm ( B )-containing elements and the burden on the in vitro fitness

In Clostridium difficile, resistance to the macrolide-lincosamide-streptogramin B group of antibiotics generally relies on erm(B) genes. In this study, we investigated elements with a genetic organization different from Tn5398, the mobilizable non-conjugative element identified in C. difficile strain 630. Our results suggested that the elements most frequently found in strains isolated during the European surveillance study in 2005 were related to Tn6194, the conjugative transposon recently detected in different C. difficile types, including PCR-ribotype 027. We characterized a Tn6194-like and a novel element rarely found in clinical isolates. A burden on the in vitro fitness of C. difficile was observed after the acquisition of these elements as well as of Tn5398.


INTRODUCTION
Clostridium difficile is recognized as the leading cause of hospital-acquired diarrhoea.Development of C. difficile infection (CDI) is a consequence of antibiotic treatment.Historically, the majority of the epidemic isolates were resistant to the macrolide-lincosamide-streptogramin (MLS B ) family.Resistance to erythromycin and/or clindamycin is still the most common phenotype in C. difficile strains isolated in Europe (Huang et al., 2009).
Resistance to MLS B in this pathogen is usually conferred by erm(B) genes.In C. difficile 630, the mobilizable nonconjugative element Tn5398 contains two copies of the gene (Farrow et al., 2001).However, erm(B) genes are located on elements different from Tn5398 in the majority of clinical isolates.In total, 17 different genetic organizations of such elements have been identified so far (Farrow et al., 2001;Spigaglia et al., 2005Spigaglia et al., , 2011)).Their genetic organizations were investigated using a PCR-mapping assay based on the sequence of Tn5398 and named from E1 to E17 (Spigaglia et al., 2005(Spigaglia et al., , 2011)).In particular, 14 different organizations were identified in 31 clinical strains isolated during the first European surveillance study conducted in 2005 (Barbut et al., 2007;Spigaglia et al., 2011).The most frequent organization detected was E4 (29 %), whereas 9.7 % of the strains contained an element E1, corresponding to Tn5398.
In this study, we demonstrate that the erm(B)-containing elements with an organization E4 are related to Tn6194, a conjugative transposon recently identified in C. difficile 2007855, and previously referred as CTnCD11 (He et al., 2010(He et al., , 2013)).We characterized a Tn6194-like transposon and another element, showing the genetic organization E3 that is rarely detected among clinical isolates.The effect of the acquisition of these elements as well as of Tn5398 on C. difficile fitness was investigated.

METHODS
Bacterial strains, culture conditions and antibiotic susceptibility.All strains used in this study are listed in Table 1.The recipient strain C. difficile CD13 is a clinical isolate collected in 1993 at the University Hospital of Parma, Italy.C. difficile was grown on brain heart infusion (BHI) agar plates or in BHI broth (Oxoid Ltd) at 35 uC in anaerobic conditions (85 % N 2 , 10 % H 2 , 5 % CO 2 ).Escherichia coli was grown on Luria-Bertani (LB) agar plates or in LB broth at 37 uC (Oxoid Ltd).
Minimal inhibitory concentrations (MICs) to erythromycin, clindamycin and rifampicin were determined using the Etest, as already described (Spigaglia et al., 2011).The breakpoints used for erythromycin and clindamycin were 8 mg ml 21 , in accordance with the guidelines established by the Clinical and Laboratory Standards Institute (CLSI, 2007).The breakpoint for rifampicin was 4 mg l 21 , in accordance with the CLSI interpretive categories approved for Staphylococcus aureus, since no values are provided for anaerobes (CLSI, 2008).
Cloning in E. coli and DNA sequencing.Overnight-digested chromosomal DNA was ligated within digested and dephosphorylated plasmid vector pUC19 (Invitrogen).E. coli transformants were selected on LB agar supplemented with ampicillin (100 mg ml 21 ) and erythromycin (150 mg ml 21 ).A primer walking strategy was employed to sequence the inserts using the primers listed in Table S1 (available in JMM Online).
PCR-mapping assay.Strains carrying an erm(B)-containing element with a genetic organization E4 were analysed by PCR mapping using overlapping primers designed on the Tn6194 sequence of C. difficile strain 2007855 (accession number FN665654).A total of 13 PCRs, detailed in Table S2, were used to amplify products overlapping for more than 26 kb, internal to the putative conjugative transposon.PCR conditions consisted of an initial denaturation step of 5 min at 94 uC followed by 30 cycles of 30s at 94 uC, 30s at 52 uC, 2 min at 72 uC, and a final elongation step of 10 min at 72 uC.
Pulsed field gel electrophoresis (PFGE).PFGE was performed as previously described (Spigaglia et al., 2001) with the following modifications.After digestion with PstI (Roche Applied Science) overnight at 37 uC, PFGE was performed with a CHEF-Mapper apparatus (Bio-Rad Laboratories) at 6V cm 21 for 18 h at 14 uC with an included angle of 120u and linear ramping from 1 to 12 s.When plugs were digested with I-CeuI (NEB), PFGE was run at 6 V cm 21 for 14h with an included angle of 120u and linear ramping from 5 to 70 s.
Southern blot hybridization.Digested DNA was transferred to a Hybond N+ membrane (Amersham Biosciences) by Southern blotting.The erm(B) internal probe of 268 bp was obtained by PCR using the primers 3140 and E5 (Farrow et al., 2001;Spigaglia et al., 2005).The 23S rRNA probe was obtained by PCR using the primers 6F and 8R (Schmidt et al., 2007).Labelling of the probes, hybridization to the membrane and signal detection were carried out as described in the ECL Direct Nucleic Acid Labelling and Detection System (Amersham Biosciences).
Determination of bacterial fitness in vitro.Bacterial fitness was estimated comparing growth rates of resistant and susceptible isogenic strains.Bacterial cultures of 18 h were diluted at OD 600 0.05 in 25 ml of BHI.Absorbance was measured every hour for 12 h at OD 600 .Each growth rate was determined at the beginning of the exponential phase, and the mean values were calculated from at least three independent experiments.Relative growth rates were calculated as the ratio of the growth rates of the resistant versus susceptible strains.
For growth competition assays, susceptible and resistant strains were grown separately in BHI for 24 h, mixed in a 1 : 1 ratio, and diluted 250-fold in 25 ml of BHI.One hundred microlitres were transferred to fresh 25 ml broth every 24 h over three cycles.The number of viable cells was determined at the end of every cycle by plating aliquots on BHI agar plates containing 20 mg erythromycin ml 21 or without antibiotic.The competition index (CI) was calculated as the c.f.u.ratio of the resistant and susceptible strains at time t 1 divided by the same ratio at time t 0 , and the selection coefficient s was calculated as s5ln(CI)/[t6ln(2)], where t is the number of generations (Foucault et al., 2010).Fitness of the susceptible strain was set to 1, and the relative fitness of the resistant transconjugants was determined as 1+s.
Statistical analyses.Statistical analyses were performed using GraphPad software.strains CII7 and F17 contain elements with genetic organization E4 and E3, respectively.A schematic representation of the two cloned fragments and their homology with sequences already deposited in GenBank is shown in Fig. 1.

RESULTS AND DISCUSSION
A region of 6.5 kb of the HindIII fragment from strain CII7 showed 100 % identity with the 39 end of Tn6194 from C. difficile 2007855 (GenBank accession number FN665654, position: 3131167-3137705) (He et al., 2010(He et al., , 2013)).Furthermore, 3.2 kb of the fragment, containing the 39 end of the CII7 Tn6194-like element and a partial CDS homologous to CD630_31070 showed 100 % identity with a region of the genome of C. difficile BI-9 (GenBank accession number FN668944, position 3292473-3295633).
Recently, a Tn6194-like genetic element has also been identified in C. difficile M68 (GenBank accession number FN668375) (Brouwer et al., 2012).To determine the ends of Tn6194 and its insertion site in strains 2007855, M68, BI-9 and 630 (GenBank accession number AM180355), a genome comparison was performed.We found that this conjugative transposon is located within a gene homologous to CD630_28310 and CD630_33170 in strains 2007855 and M68, respectively.Instead, sequence analysis indicates that in BI-9 the ends of a Tn6194 element are located between CD630_31060 and CD630_31070.The same localization of Tn6194 found in BI-9 was also observed in CII7.Since both BI-9 and CII7 are PCR-ribotype 001, Tn6194 seems to integrate into the same genomic region in strains belonging to the same PCR-ribotype.
PCR mapping demonstrated that the elements with a genetic organization E4 were Tn6194-like (data not shown), suggesting that these elements are widely distributed among European C. difficile clinical isolates.
About 5.2 kb of the 7.3 kb NdeI fragment from strain F17, including erm(B), showed 99.9 % identity with part of the plasmid pAMb1 from E. faecalis DS5 (GenBank accession number GU128949.1)(Fig. 1).Since reciprocal genetic exchanges between C. difficile and E. faecalis can occur in vitro (Jasni et al., 2010), it is possible that the mosaic element of F17 has been originated by a partial integration of the plasmid pAMb1 into the genome of C. difficile.

Transfer of ermB determinants
Conjugation assays were performed to determine if transfer of elements from strain CII7 and F17 to the MLS Bsusceptible recipient strains CD37 and CD13 was possible.Assays were also performed to transfer Tn5398 from 630 to the same recipient strains.
All elements were successfully transferred to both CD37 and CD13 (Table 2).All transconjugants were obtained from independent filter-mating assays to exclude the possibility of analysing siblings.Three independent transconjugants were selected for each transfer, except when F17 was used as donor and CD13 as recipient.
Southern blots and PCR assays showed that Tn5398 inserts into the genome of all transconjugants (630xCD13 and 630xCD37) into the site previously described in CD37 (Farrow et al., 2001) (data not shown).Transconjugants CII7xCD13 A and B showed the same hybridizing band at 44 kb and CII7xCD13 C at 42 kb.A hybridizing band at 42 kb was also observed when the donor strain CII7 was analysed.Transconjugants CII7xCD37 A, B and C had hybridizing bands at 128, 68 and 61 kb, respectively (Fig. 2a).Transconjugants F17xCD13 and F17xCD37 each showed one hybridizing band at 122 and 128 kb, respectively, while F17 had a 110 kb hybridizing band (Fig. 2b).These results suggested that the Tn6194-like element from CII7 could integrate into the recipient genome at different sites, while the element from F17 seems to have a unique integration site.
Localization of erm(B) was analysed by Southern hybridization of an erm(B) and a 23S probe on I-CeuI PFGE fragments.This enzyme specifically targets genes coding for rRNA (Liu et al., 1993).Hybridization patterns indicated that the resistance gene was located on the chromosome in all transconjugants (data not shown).
Information about the presence of these elements was available only for strain 630 but not for CII7 and F17.To determine if any transposon was cotransferred with Tn5398, transconjugants 630xCD13 and 630xCD37 were investigated by PCR.All transconjugants were negative for other elements except 630xCD37B that, in addition to Tn5398, acquired CTn2 (data not shown).

Fitness cost of erm(B)-containing elements
Few studies have investigated the fitness cost associated to transposons (Andersson & Hughes, 2010).It is estimated that their carriage would not be costly, since they are relatively small in size and integrated within the chromosome (Enne et al., 2005).To our knowledge, nothing is known concerning the fitness cost associated with antibiotic resistance in C. difficile.
The growth rates of all transconjugants derived from CD13 were significantly reduced, compared to those of their isogenic ancestors, regardless of the element acquired (P¡0.05) (Fig. 3).Similarly, all transconjugants derived  from CD37 showed a significant reduction in growth rates except those containing the element from F17.The fitness cost for each element acquired was also assessed by competition assays between transconjugants and recipient strains (Table 3).As observed in growth assays, all transconjugants from CD13 were disadvantaged compared to the susceptible ancestor and showed relative fitness between 0.864±0.012and 0.911±0.012(Table 3).Different behaviours were observed for transconjugants from CD37 (Table 3).While transconjugants 630xCD37A and 630xCD37C were not significantly disadvantaged, 630xCD37B outcompeted CD37, with a relative fitness of 1.195±0.056.Among transconjugants CII7xCD37, CII7xCD37A was the most disadvantaged with a relative fitness of 0.818±0.013.Finally, in agreement with growth assay results, transconjugants F17xCD37 were not disadvantaged compared to CD37.
Since Tn5398 has a unique integration site (Farrow et al., 2001), it cannot be ruled out that disruption of the insertion sequence is responsible for the fitness cost observed rather than the carriage of the element itself.Among transconjugants acquiring Tn5398, only those from CD13 were disadvantaged.Thus, it can be hypothesized that Tn5398 imposes a burden depending on the recipient strain.Interestingly, the transconjugant 630xCD37B, containing both Tn5398 and CTn2, prevailed over CD37 in the competition assays.So, it will be very useful to deeply investigate this transconjugant to understand the effect of the CTn2 acquisition.
While all CII7xCD13 transconjugants were similarly affected by the acquisition of Tn6194, CII7xCD37 transconjugants were not, probably because of the different insertion sites of the element in the two recipients.The element from strain F17 imposed a fitness burden only for transconjugants from CD13, underlining the importance of both insertion site and recipient strain type in the fitness cost.
In conclusion, we demonstrated that Tn6194-like elements are distributed among European C. difficile clinical isolates belonging to different PCR-ribotypes and that the carriage of an erm(B)-containing element frequently has a cost on the in vitro fitness.Our results also suggest that, independent of the burden that the element imposes on fitness, other factors, such as the capacity of in vivo transfer, the different insertion sites and the intrinsic characteristics of the different C. difficile types, are involved in the successful spreading of an element among C. difficile clinical isolates, although in vivo studies will be necessary to confirm these observations.

Fig. 1 .
Fig. 1.Organization of erm(B)-containing elements in the strains analysed in this study.Top to bottom and left to right: position 3244 to 10 796 from pAMb1; erm(B)-containing NdeI fragment from C. difficile F17; position 2 325 159 to 2 310 625 from C. difficile 630; erm(B)-containing HindIII fragment from CII7; position 3 139 326 to 3 131 167 from C. difficile 2007855; position 3 292 473 to 3 299 673 from C. difficile BI-9.Accession numbers of deposited sequences are shown within parentheses.Tn5398 and part of Tn6194 are indicated on sequences of 630 and 2007855, respectively.Grey boxes indicate regions of homology.Homologous genes found in all sequences are shown in grey.

Fig. 3 .
Fig. 3. Relative growth rates of transconjugants derivative of CD13 (a) or CD37 (b).Relative growth rate represents the ratio of growth of each transconjugant compared to that of its wild-type ancestor (set at 1).Values are the means of at least three separate experiments with error deviations.*Significant difference (P,0.05) of the mean value.

Table 2 .
Characterization of transconjugants obtained in this study