Characterisation of a new group of Francisella tularensis subsp . holarctica in Switzerland with altered antimicrobial susceptibilities , 1996 to 2013

Molecular analysis of Francisella tularensis subsp. holarctica isolates from humans and animals revealed the presence of two subgroups belonging to the phylogenetic groups B.FTNF002-00 and B.13 in Switzerland. This finding suggests a broader spread of this group in Europe than previously reported. Until recently, only strains belonging to the Western European cluster (group B.FTNF002-00) had been isolated from tularaemia cases in Switzerland. The endemic strains belonging to group B.FTNF002-00 are sensitive to erythromycin, in contrast to the strains of the newly detected group B.13 that are resistant to this antibiotic. All the strains tested were susceptible to ciprofloxacin, streptomycin, gentamicin, nalidixic acid and chloramphenicol but showed reduced susceptibility to tetracycline when tested in a growth medium supplemented with divalent cations. The data show a previously undetected spread of group B.13 westwards in Europe, associated with changes in the antibiotic resistance profile relevant to treatment of tularaemia.


Introduction
Francisella tularensis is a Gram-negative bacterium causing the zoonotic disease tularaemia.The two clinically relevant subspecies are F. tularensis subsp.tularensis and F. tularensis subsp.holarctica.Of the two, only the latter subspecies is present in Europe.Human infections mainly occur through inhalation, ingestion, or by direct contact with infected animal species and contaminated animal tissues, water and aerosols [1].In general, it is difficult to trace the source of infection [2].Analyses of molecular genetic markers of the strains circulating in the environment provide valuable information on the dynamics of infection in people and in animals, and improve the knowledge of the biology of this bacterium.Genetic and phenotypic data are also necessary to determine the most suitable antimicrobial substances to use to treat humans and animals [3][4][5][6][7][8].
With the recent growing interest in F. tularensis biology, several tools have been developed to investigate the molecular epidemiology of this genetically monomorphic bacterium following a hierarchical scheme [9][10][11][12][13].Genomes of different strains are screened for canonical single nucleotide polymorphism (canSNP) signatures.With the advent of novel technologies, an increasing number of strains are sequenced, leading to the discovery of new canSNP markers and signatures specific to new subgroups.The growing information involves adjustments in the phylogenetic nomenclature of F. tularensis and allows better resolution within the subgroups [11,12,14].Throughout this manuscript, we will follow the nomenclature based on the canSNPs (for nomenclature clarity, refer to schema in Figure 1).In Europe, strains belonging to groups B. 13 and B.FTNF002-00 are those predominantly isolated [12].The group B.13 extends geographically from Scandinavia to the eastern European rim, with co-circulation of several of its subgroups in some countries.In Western Europe, a specific group, B.FTNF002-00, is circulating in France, Germany, Italy, Spain and Switzerland [11][12][13][14][15][16][17][18][19].Recently, it was observed that Germany represents a geographical diaphragm virtually separating group B.13 from group B.FTNF002-00 [18].Interestingly, strains belonging to group B.FTNF002-00 are described as sensitive to erythromycin, whereas strains belonging to other groups show variability in this marker [3][4][5]7,18].Historically, strains of F. tularensis subsp.holarctica have been separated in two biovars: biovar I, strains sensitive to erythromycin, and biovar II, strains resistant to erythromycin [20].Although this marker is principally used for epidemiological purposes, there may be significant clinical implications in areas with co-circulation of different groups.A paradigmatic example of this is the recent recommendation of treating pregnant women infected with F. tularensis with azithromycin in geographical areas where strains sensitive to erythromycin are circulating [21].
This study describes the first isolation of erythromycin-resistant strains of F. tularensis subsp.holarctica belonging to group B.13 in Switzerland.We discuss how these findings impact on the phylogeography of F. tularensis subsp.holarctica in Europe and on antibiotic treatment of affected individuals living in areas with co-circulating groups [12].

Bacterial strains, DNA templates, identification and typing
All manipulations with live cultures were performed in a BSL3 containment laboratory.F. tularensis strains were cultivated on chocolate agars with IsoVitaleX (Becton Dickinson, Allschwil, Switzerland) for three days at 37 °C with 5% CO 2 .Lysates from cultures were prepared, filter-sterilised [22] and tested by real-time PCR for the presence of the fopA gene to confirm the species F. tularensis [22,23].The subspecies was subsequently determined by amplification of the region of difference (RD)1 [24].Strains were further characterised by PCR for the presence of deletions in two different markers, RD23 and Ft-M24, specific to the group B.FTNF002-00 [16,17,24] and by multilocus variable-number tandem repeat (VNTR) analysis (MLVA) with six VNTRs markers (Ft-M3, Ft-M6, Ft-M20, Ft-M21, Ft-M22 and Ft-M24) [17,25].The MLVA results were further confirmed by analysis of the following canSNP markers B.  [11,12].

Minimal inhibitory concentrations of antimicrobial agents
The minimal inhibitory concentration (MIC) values of antibiotic drugs relevant to clinical use such as gentamicin (0.12-16 mg/L), streptomycin (1-16 mg/L), ciprofloxacin (0.06-4 mg/L), tetracycline (0.25-16 mg/L), nalidixic acid (2-64 mg/L), chloramphenicol (2-32 mg/L) and erythromycin (0.5-32 mg/L), were determined in two different broth media: (i) modified Cation-Adjusted Mueller Hinton Broth (mCAMHB):  Canonical single nucleotide polymorphisms (canSNP), adapted from and according to the colours and symbols previously described [11][12][13], showing the position of Swiss strains within the subspecies.[26] and the informational supplement [27].Antibiotics of the class of the betalactams were not tested because of the known natural resistance of F. tularensis strains to these antimicrobial substances [6,28].The 96-well plates were incubated at 37 °C in 5% CO 2 atmosphere for 48 hours.The MIC values were defined as the lowest concentration exhibiting no visible growth.MICs were read after 24 and 48 hours incubation.For quality assurance, the reference strains of Escherichia coli ATCC 25922 and Staphylococcus aureus ATCC 29213 were also tested by broth microdilution in mMHII broth and in mCAMHB.
The highest variability among the markers used for MLVA was observed within markers Ft-M3 and Ft-M6 as previously reported for the group B.FTNF002-00 (Table 1) [15,17,18].Concerning the strains not belonging to group B.FTNF002-00, the three strains, JF5340, JF5370 and JF5405 isolated from two human patients and one hare between 2012 and 2013, shared the same VNTR profile, while strains JF4830, isolated in 2010 from a patient returning from a vacation in eastern Europe (possibly corresponding to an imported case), and JF5468, isolated from a hare in 2013, revealed a distinct VNTR profile, with one variation in the Ft-M3 marker (Table 1).The four strains, JF5340, JF5370, JF5405 and JF5468 were isolated from a large geographical area of Switzerland extending from the central west to the east sides of the country (Figure 2).
Antibiotic susceptibility profiles were determined for the panel of 24 representative strains (Table 2) by broth microdilution method using two broth media, mCAMHB and mMHII, for seven antibiotic drugs and read after 48 hours.The F. tularensis strains did not show any visible growth at the concentrations tested for nalidixic acid, chloramphenicol and ciprofloxacin in either broth medium, tested (Table 2).The MIC values for chloramphenicol and ciprofloxacin were below the breakpoint values provided in the CLSI guidelines, ≤8 μg/mL for chloramphenicol and ≤0.5 μg/mL for ciprofloxacin, while no breakpoint value is available for nalidixic acid [27].MIC values for gentamicin ranged between 1 and 4 μg/mL in mMHII, but between ≤0.12 and 0.25 μg/mL in mCAMHB.The breakpoint value for gentamicin given in the CLSI guidelines for F. tularensis is ≤4 μg/mL [27].The discrepancy in the MIC values observed for gentamicin between cultures of F. tularensis in mMHII and mCAMHB was confirmed for reference strains E. coli ATCC 25922 and S. aureus ATCC 29213 (Table 2).MIC values for streptomycin ranged between 2 and 4 μg/mL in mMHII and were all at 4 μg/mL in mCAMHB (Table 2).The breakpoint value provided for streptomycin in the CLSI guidelines is ≤16 μg/mL when testing is performed in a CO 2 atmosphere [27].MIC values for    tetracycline ranged between 2 and 8 μg/mL in mMHII, but were all ≤0.25 μg/mL in mCAMHB (Table 2).For this reason, MIC values for tetracycline were also tested by Etest in order to confirm the results obtained with mCAMHB.With this method, MIC values for tetracycline ranged between 0.19 and 0.38 μg/mL, which was similar to the ones measured by broth microdilution method with mCAMHB (Table 2).The breakpoint value given in the CLSI guidelines for tetracycline is ≤4 μg/ mL [27].Moreover, a difference in MIC values for tetracycline in mMHII and mCAMHB was also observed for reference strains E. coli ATCC 25922 and S. aureus ATCC 29213 (Table 2).MIC values for erythromycin exhibited a bimodal distribution: Strains belonging to group B.13 exhibited erythromycin resistance with MIC values higher than 32 μg/mL, while all the strains belonging to group B.FTNF002-00 were sensitive to erythromycin, showing MIC values ranging from 1 to 8 μg/mL (Table 2).However, no breakpoint value for F. tularensis is provided for any macrolides in the CLSI guidelines [27].
Genetic characterisation of erythromycin resistance was performed by PCR amplification and sequencing of the three copies of the rrl gene and of the rplD and the rplV genes of the five strains showing phenotypic resistance to erythromycin and belonging to group B.13 (JF4830, JF5340, JF5370, JF5405 and JF5468).The sequences were compared to the corresponding genes of strains FTNF002-00 (group B.FTNF002-00), F92 (group B.FTNF002-00), FSC200 (group B.13) and LVS (group B.13).All five strains isolated in Switzerland, belonging to group B.13 and showing phenotypic resistance to erythromycin, had two mutations in all three copies of the rrl gene when compared to the available sequences of the strains FTNF002-00 and F92 of the group B.FTNF002-00.The first mutation was detected in domain I of the rrl gene, A453G (E. coli numbering), while the second mutation was observed in domain V of the rrl gene, A2059C (E. coli numbering).A silent mutation G to A at the third position of codon 181 (E. coli numbering) was found in the rplD gene encoding

Table 2
Antibiotic susceptibilities of Francisella tularensis subsp.holarctica strains after 48 hours in mMHII broth and mCAMH, Switzerland, 1996-2013 (n=24) Strain Subgroup according to [12] and [13]  for the ribosomal protein L4, while no mutation was detected in the rplV gene encoding for the ribosomal protein L22.These mutations observed in Swiss strains belonging to group B.13 were exactly the same as those of the strain LVS and FSC200 belonging to group B.13.
The EMBL/GenBank accession numbers for the nucleotide sequences of the rrl, the rplD and the rplV genes are: KF712467, KF712466 and KF712465, respectively.

Discussion
This study describes the characterisation of  [12].Both subgroups were isolated in central and eastern Europe and from countries bordering Switzerland, such as Austria and Germany, but not from countries east of Romania [12].Moreover, the subgroup B.33/34 is also known to be circulating in Sweden [12].
These findings reveal that group B.13 is currently circulating in Switzerland in the same areas as strains of group B.FTNF002-00 (Figure 2) and are affecting both human patients and free-ranging animals.This is in contrast to neighbouring Germany, where a strict separation between groups was described [18].Strains isolated between 1996 and 2011 from humans, hares and captive non-human primates all belonged to group B.FTNF002-00.Because of the small number of strains isolated between 1996 and 2008, it is difficult to draw conclusions about a recent introduction of strains belonging to group B.13 or a long-lasting co-circulation of both groups following the expansion of group B.FTNF002-00 of F. tularensis subsp.holarctica.
Strains belonging to group B.FTNF002-00 are known to be sensitive to erythromycin [7,29,33].Also the Swiss F. tularensis subsp.holarctica strains belonging to group B.FTNF002-00 are sensitive to erythromycin, while the new strains belonging to group B.13 are resistant (Table 2).Since strains resistant to erythromycin are actually circulating in Switzerland, macrolides are not recommended for the treatment of cases of tularaemia acquired in Switzerland and possibly also in neighbouring areas unless analysis of the infecting strains reveals sensitivity to this antibiotic.Because of the toxicity of recommended antibiotics against tularaemia for pregnant women and foetuses, Dentan et al. [21] proposed to treat it, in areas where the group B.FTNF002-00 is endemic, with a macrolide, more specifically with azithromycin.However, the spread to western Europe of strains resistant to macrolides poses serious concerns and needs to be carefully considered by the clinicians when facing a therapeutic choice in this context.Several studies suggest that strains of F. tularensis from western Europe are sensitive to macrolides [4,7,18,29,33,34].
Genetic analysis of the strains resistant to erythromycin revealed two mutations in the three copies of the rrl gene and a silent mutation in the rlpD gene encoding the ribosomal protein L4, compared with the strains belonging to group B.FTNF002-00: FTNF002-00 and F92.Interestingly, the same mutations are present in the strains LVS and FSC200, both belonging to group B.13.This finding may suggest that these mutations are shared among subgroups belonging to B.13 and may have appeared in a common ancestor.However, this hypothesis should be validated by testing a larger panel of strains.
Broth microdilution testing was performed in mMHII broth and in mCAMHB [4,7].Results were compatible for all antibiotics tested in both media except for the MIC values recorded for gentamicin and tetracycline that showed higher values in mMHII broth than in mCAMHB (Table 2).This discrepancy was confirmed by testing the reference strains, E. coli ATCC 25922 and S. aureus ATCC 29213, which also showed higher MIC values for gentamicin and tetracycline in mMHII broth than the quality control ranges for broth microdilution method in mCAMHB reported in the CSLI guidelines [27], confirming that MIC values for these two antibiotics are higher using mMHII broth than mCAMHB (Table 2).High MIC values of reference strains for gentamicin and to a lesser extent for tetracycline tested in mMHII broth were previously described by Baker et al. [35].They explained these results to be due to the addition of the bivalent cations Ca 2+ and Mg 2+ in the medium.MIC values for tetracycline were within the range considered clinically effective when tested with mCAMHB and by Etest.Nevertheless, given the high number of tularaemia cases for whom tetracycline-associated treatment failure has been described, including doxycycline [36][37][38][39], these antibiotics are not recommended in case of infection with F. tularensis.
MIC values for gentamicin, chloramphenicol, streptomycin and ciprofloxacin were within the range indicative of clinical efficacy in both media, although MIC values of four strains for gentamicin in mMHII were very close to the breakpoint value given in the CLSI guidelines [27].Ciprofloxacin showed the lowest MIC values and prevented growth of all strains at 0.06 μg/ mL.The finding is consistent with previous reports on type A and type B tularaemia [5,[40][41][42][43][44] and supports the experience that ciprofloxacin may be an attractive treatment option for tularaemia [37].

Conclusion
In conclusion, at least two groups of F. tularensis subsp.holarctica are currently co-circulating in Switzerland.Of these, the group B.FTNF002-00 seems to be more prevalent and has been identified in tularaemia cases since 1996, while B.13 was less commonly isolated in Switzerland and not before 2012.Since strains belonging to the subgroups B.33/34 and B.34/35 are erythromycin-resistant, this antibiotic is not recommended to treat cases of tularaemia acquired in Switzerland without prior typing of the strains.These concerns should also apply in countries where the group B.FTNF002-00 seems to be prevalent given that the exact limits of the co-circulation areas are not known.The mutations resulting in erythromycin resistance in group B.13 strains are exactly the same as those present in the strain LVS belonging to the same group.Further investigations are warranted in order to understand if they are shared by all strains of the group B.13.In view of the in vitro results and of previous clinical observations, tetracyclines should not be a first choice of treatment for tularaemia, while ciprofloxacin appears suitable for tularaemia treatment.Moreover, because of the fastidious growth requirements of F. tularensis, supplements always need to be added to growth media to test antibiotic susceptibility [7,35].Recently, Georgi et al. published the validation of a protocol for a broth microdilution method (medium not supplemented with divalent cations) for F. tularensis [4].The use of a medium without supplemented divalent cations, mCAMH, has been compared with the previously described methods using mMHII broth (medium supplemented with divalent cations).Considering the discrepancy in the MIC values measured for gentamicin and tetracycline depending on the broth used, it is considered more appropriate the use of mCAMHB for antimicrobial testing of F. tularensis strains because it achieves results, at least relatively to some antimicrobial substances, that are less ambiguous than those achieved with the mMHII broth.

Figure 1
Figure 1Schematic of Francisella tularensis subsp.holarctica nomenclature based on canonical single nucleotide polymorphisms available; MLVA: multilocus variable-number tandem repeat analysis; SNPs: single nucleotide polymorphisms.Strains selected for further analysis (canSNPs and antimicrobial agent susceptibilities, n=24) are shaded in grey.Strains belonging to the clade B.13 are highlighted in bold.