Tetracycline Resistance and Presence of Tetracycline Resistance Determinants tet(V) and tap in Rapidly Growing Mycobacteria from Agricultural Soils and Clinical Isolates

Rapidly growing mycobacteria (RGM) inhabit soil and water but certain strains represent a health risk for human and animals. Both clinical and soil RGM may be under selection pressure for resistance to tetracycline (TET) antibiotics, since tetracyclines are administrated to humans and farm animals, and TET residues enter soil through manuring; however, resistance to TET and the presence of TET-resistance genes have been assessed only in clinical isolates. We were therefore interested in comparing soil and clinical RGM in terms of TET resistance and the presence of TET-resistance genes. We used 44 RGM from grasslands with different exposure to animal manure, and 38 clinical RGM from Czech hospitals. There was no difference between the clinical and soil isolates in TET resistance, with >50% resistant isolates in both groups. otr(A), otr(B), tet(K), tet(L) or tet(M) were not detected in any soil or clinical isolate. In contrast, most isolates harbored tet(V) and tap, both encoding mycobacterial efflux pumps, including species where these genes have never been evidenced before. The phylogeny of tet(V) correlated with isolates’ BOX-PCR profiles, suggesting that this gene evolved along with mycobacterial genomes as a part of the intrinsic resistome. In certain cases, tet(V) and/or tap were found in TET-sensitive isolates, or inversely, were not found in resistant strains. Concluding, intrinsic efflux pumps may be more important for TET resistance than horizontally transferred genes in both soil and clinical RGM. Their simple presence, however, does not attest to resistance, and therefore their diversity, function and expression merit further research.

Rapidly growing mycobacteria (RGM) are common inhabitants of soil and water. In past decades, they have been increasingly recognized as a cause of human and animal diseases (15,25). These include respiratory infections, a spectrum of hard and soft tissue infections or bacteremia in immunocompromised patients, as well as infections related to injuries in healthy individuals (8,29,48). The natural and man-transformed environment is an important reservoir of RGM, from which transmission to humans occurs (50). Although the direct relationship between an environmental source and clinical disease is difficult to evidence, several RGM outbreaks have been associated with the exposure to soil and water (2,16), including drinking water (49).
The prevention and treatment of infections due to RGM is not trivial because of their high resistance to disinfectants and antibiotics. One group of antibiotics used in the therapy of RGM infections is the tetracyclines (TET) such as doxycycline and minocycline (15). Mycobacteria in general have intrinsic resistance to many antibiotics ensured by the composition of their cell wall and the presence of several multidrug efflux pumps (33,36). Tetracycline/multidrug efflux pumps Tet(V) and Tap may belong to this intrinsic resistome as they have been so far found only in certain RGM species (17,22). Acquired resistance is not often seen in Mycobacterium; however, Pang et al. (38) reported four genes conferring TET resistance in clinical RGM that have been most probably horizontally transferred from other bacteria. These included otr(A) and otr(B), self-protection genes from the oxytetracycline producer Streptomyces rimosus (9), and tet(K) and tet(L), low-G+C-content genes encoding tetracycline efflux pumps, which are typically found in Firmicutes (Streptococcus, Staphylococcus, Enterococcus) but also in some Gram-negative bacteria (40). More recently, tet(M), a widely distributed gene with a low G+C content encoding a ribosomal protection protein, was found in a humanassociated Mycobacterium sp. (41).
Resistance to TET and the presence of TET-resistance genes in RGM have been studied in clinical isolates only; however, soil is an important reservoir of TET resistance genes, both indigenous (14) and introduced by manuring (10). In addition, tetracycline residues are detectable in manured soil (10), which may help select resistant strains. Soildwelling RGM may therefore represent an important pool of TET resistance genes.
The main objective of this study was to compare soil and clinical isolates of RGM (both from the Czech Republic) in terms of the resistance to tetracycline and presence of seven TET-resistance determinants. The genome relatedness of isolates was assessed with BOX-PCR as a high correlation exists between BOX-PCR fingerprints and DNA-DNA homology data (31,39,51). 16S rRNA gene sequencing was performed to identify the isolates. Resistance to TET was assessed with the agar disk diffusion method, and the presence of TET resistance genes that were previously described in clinical RGM, i.e., otr(A), otr(B), tet(K), tet(L), tet(M), tet(V) and the multi-drug efflux pump-encoding gene tap, was checked with PCR and sequencing. The phylogeny of tet(V) was compared to the BOX-PCR profiles of isolates.

Soil isolates of rapidly-growing mycobacteria
Soil RGM were isolated from four sites of three farms located in South Bohemia, Czech Republic in 2007-2010 (Table S1). The distance between farms was up to 10 km. Farm 1 is a conventional farm engaged in intensive pig fattening, where animals (about 2,000) are commonly treated with antibiotics including chlortetracycline and doxycycline. At Farm 1, we sampled a permanent grassland, which had been periodically manured (2-3 times per year) with pig slurry for the previous 30 years (designated Site 1). At Farm 2, which is a small family farm in a neighborhood community, we sampled permanent grassland that had not been manured for the previous 20 years (Site 2). Samples were taken from Site 1 and Site 2 in June 2007 and 2009. Farm 3 has performed outdoor cattle husbandry since 1993 and is an organic farm without the application of antibiotics. Two sites (Site 3 and Site 4) were sampled at Farm 3 in May 2010. Site 3 is part of the pasture where cattle stay from October until May. It is highly impacted by the cattle, i.e., the soil is highly enriched with excrement and vegetation cover is damaged (28). Site 4 is a pasture with low impact by the cattle and preserved vegetation. At each site, soil from a depth of about 10-30 cm (under the plant roots) was sampled with a sterile spade from three points 5-20 m apart. The soil from the three points was mixed and sieved. Soils were kept at 4°C during transport to the laboratory and prior to the accompanying physicochemical and microbial analyses (Table  S1). RGM were isolated from soil using the NaOH/malachite green/ cycloheximide decontamination method of Iivanainen (27) or the olive oil/SDS decontamination method (with 10 mg SDS per plate) of Yamamura and Harayama (53) or directly on Tryptic-Soy agar plates with 25 mg L −1 chlortetracycline (2 isolates from 2007) (Table  S2).

Clinical isolates of rapidly-growing mycobacteria
Clinical RGM were obtained from the National Reference Laboratory for Pathogenic Actinomycetes, Regional Hospital in Trutnov, and from the Institute of Public Health, Ostrava, Czech Republic. They were isolated in 2006-2011 from various samples such as abscess, urine, hemoculture, corneal ulcer and sputum from 18 hospitals in the Czech Republic (Table S3). Their role in the etiology of infection was confirmed in isolates from abscesses, hemocultures and corneal ulcers. Incidental isolates from sputa of patients screened for Mycobacterium tuberculosis without clinical and imaging correlates were usually colonizers, with the exception of Mycobacterium chelonae OS10 associated with a pulmonary disease in a 47-year-old patient.

Susceptibility to tetracycline
Susceptibility to tetracycline was assessed with a disc diffusion test (26). Pure isolates were first grown on M2 (42) or Šula's medium (43) with the addition of 1.5% agar, at 28°C for 5 to 7 d. A homogenous bacterial suspension was prepared by vortexing (Vortex-Genie2; Mo Bio Laboratories, Carlsbad, CA, USA) and ultrasonication (Ultrasonic Compact Cleaner UC 006DM1, Tesla, Czech Republic) of several colonies in 4 mL of sterile 0.9% NaCl. The turbidity of the suspension was adjusted with sterile 0.9% NaCl to match the McFarland standard 0.5 (densitometer DEN-1; Biosan, Latvia) (52). The suspension was spread onto Mueller-Hinton agar medium (Bio-Rad Laboratories, Hercules, CA, USA) supplied with TET disks (30 µg; Bio-Rad). Inhibition zone diameters were recorded after 5 d of incubation at 28°C. The ranked zone sizes of environmental and clinical strains were statistically compared with Wilcoxon rank sum test in R (http://www.r-project.org). Strains OS18, OS2/1, OS2/2 and OS2/4 did not grow on Mueller-Hinton agar and disc diffusion analysis was therefore performed on Šula's medium. These strains were not included in the statistical analysis.

BOX-PCR genomic DNA fingerprints
Prior to PCR amplification, cell lysates were prepared as follows. One bacteriological loop of mycobacterial biomass grown on an agar plate was resuspended in 100 µL ultra-pure water. The suspensions were then boiled three times (water bath, 100°C) for 5 min and frozen at −20°C for 1 h. The lysates were stored at −20°C and 1 µL of the lysates was used as a template for PCR.
DNA amplification followed the procedure of Lanoot et al. (31) using the BOXA1R primer (5'-CTACGGCAAGGCGACGCT GACG-3') (51). PCR products (20 µL) were separated on 20×20 cm gels using 130 V, 400 mA for 240 min in 1×TBE buffer (Tris base 53 g, boric acid 27.5 g, 0.5 M EDTA 20 ml, pH 8.0). The gels were stained for 30 min in a 1×TBE bath supplemented with ethidium bromide (1 mg L −1 ). A photograph of the gel was stored as a TIFF file through a CCD coupled camera using Photo-Doc software (Vilber-Lourmat, Marne-la-Vallée, France). Gels were imported into the software package GelCompar II (Applied Maths, Sint-Martens-Latem, Belgium) and similarity matrices of densitometric curves of the gel tracks were calculated using the Pearson correlation coefficient followed by dendogram construction using the UPGMA algorithm. We used the limit of 70% similarity to define distinct BOX-PCR groups (12).

Detection of tetracycline resistance genes
The presence of tetracycline resistance genes otr(A), otr(B), tet(K), tet(L), tet(M), tet(V) and tap was assessed with PCR using gene-specific primers ( Table 1). The total volume of PCR reactions was 25 µL and 1 µL of mycobacterial lysates (see above) or ~20-50 ng of chromozomal DNA from positive control strains or plasmid DNA were used as a template. The cycling conditions for each gene are shown in Table 1.
Amplified tet(V) or tap genes were cleaned-up with the GenElute PCR Clean-Up Kit (Sigma-Aldrich) and sequenced from both ends in the case of tet(V) and from the forward primer in the case of tap, as described above. The sequences were edited and assembled with Bioedit 7.0.4.1 software (23) and compared to the GenBank database (www.ncbi.nml.nih.gov) using blastn and blastp algorithms. Altogether, 46 PCR products of tet(V) and 14 PCR products of tap were sequenced.

Accession numbers
The tet(V) sequences were deposited in GenBank under accession numbers JF290326-JF290351 and JQ348076-JQ348095, and the tap sequences under accession numbers JF290352-JF290365. The 16S rRNA gene sequences are available in GenBank under accession numbers JF304573-JF304610 and JQ348096-JQ348111.

Isolate identification and genome relatedness
Forty-four isolates from grasslands and 38 clinical isolates were included in this study (Table S2, S3 and S4). The environmental isolates were divided into 14 distinct groups and the clinical isolates into 28 distinct groups according to their BOX-PCR profile similarity, using the Pearson correlation coefficient threshold of 0.7 (Table 2, 3 and Fig. S1). Clinical and environmental strains always belonged to separate BOX-PCR groups.
A minimum of one isolate from each environmental BOX-PCR group was identified by sequencing the 16S rRNA gene and comparing to the type strain database using EzTaxon v 2.1 software (Tables 2 and S2). Most of the soil isolates (i.e., 26) had sequences identical or almost identical (98-100% pairwise sequence similarity) to the species Mycobacterium septicum, 2 isolates to Mycobacterium fortuitum subsp. acetamidolyticum, 1 isolate to Mycobacterium alvei, 1 isolate to Mycobacterium litorale and 1 isolate to Mycobacterium aubagnense. The isolates from one BOX-PCR group were usually assigned to the same species, with the exception of

Resistance to tetracycline
The distribution of TET resistance was bimodal in both clinical and environmental strains (Fig. 1). Since standard breakpoints of the disc diffusion method are not available for RGM, we arbitrarily set them according to the zone size distributions. An isolate was considered to be resistant if the zone was up to 25 mm. Based on the arbitrary breakpoint, thirty-two environmental (70%) and sixteen clinical isolates (53%) were resistant to TET (Table 2 and 3). There was no significant difference in TET resistance (in terms of zone    sizes, assessed with Wilcoxon rank sum test) between the environmental and clinical isolates. Tetracycline resistance was not a specific characteristic of individual BOX-PCR groups, i.e., both resistant and sensitive isolates could be found within the same BOX-PCR group.

Detection of tetracycline resistance genes
The two primer pairs designed in this study, i.e., tetV-FW/ tetV-RV [tet(V) detection], and tetKL-FW/tetKL-RV [simultaneous detection of tet(K) and tet(L)], were tested for their specificity in PCR using different tetracycline efflux pump genes as templates. Both primer pairs were specific, i.e., they amplified only tet(V) in the case of tetV-FW/tetV-RV and tet(K) and tet(L) in the case of tetKL-FW/tetKL-RV (data not shown). The specificity of the tetV-FW/tetV-RV primer  pair was further corroborated by sequencing the PCR products (see below). All isolates were tested for the presence of otr(A), otr(B), tet(K)/tet(L), tet(M), tet(V) and tap. The gene tet(V) was detected in 32 of the total 44 environmental strains (73%) and in fourteen clinical isolates (37%) ( Table 2  arupense and M. neoaurum. Surprisingly, tet(V) was found also in isolates that were sensitive to tetracycline (even in those with zones over 50 mm) ( Table 2  Similarly to tet(V), tap was detected also in strains that were sensitive to tetracycline. In contrast, the genes otr(A), otr(B), tet(K)/tet(L) and tet(M) were not detected in any isolate, and eight clinical and seven environmental isolates were resistant to TET but possessed none of the tested genes.

Analyses of tet(V) and tap sequences
The gene tet(V) was sequenced in all isolates (i.e., 46) where we obtained positive signals from PCR. Sequences of isolates Site2-3C, Site2-5C and Site2-7C (all from the BOX-PCR group H) were not used for further analyses since they contained several ambiguous peaks that could not be resolved (although PCR and sequencing were attempted twice). It is possible that these isolates had two copies of tet(V) that differed slightly in their sequences.
The 43 analyzed partial sequences of the gene tet(V) [corresponding to positions 823-1,132 of M. smegmatis MC2-155 tet(V)] differed substantially among the isolates, with 17% differences between two most distant isolates (Fig.  2). The isolates shared 82-88% and 90-95% identity, respectively, of nucleic and inferred amino acid sequences with the published tet(V) sequence of M. smegmatis MC2-155. Since all the recovered sequences had amino acid identity above 80% with the published tet(V), they can be attributed to the tet(V) gene class (32). Interestingly, M. neoarum OS22 had a one-nucleotide deletion at position 915. The phylogeny of tet(V) (Fig. 2) shared overall similarity with the genome relatedness of the strains (Fig. 3), and the soil isolates usually clustered separately from clinical RGM. Indeed, there was a significant correlation (r=0.64, P=0.001) between the BOX-PCR profile-and tet(V) sequence-based distant matrices.
Fourteen PCR products of tap (corresponding to positions 764-1,080 of tap M. fortuitum) (4) were chosen for sequencing in order to verify the specificity of primers and check for tap gene diversity. Similarly to tet(V), differences in the partial sequences of tap between our isolates and the published sequence were found (data not shown). Out of the sequenced isolates, M. fortuitum subsp. fortuitum TR-1242 had the tap sequence most similar to the published sequence (99% identity), while the M. rufum TR-1359 tap sequence was the most dissimilar (only 79% identity). The environmental isolates had 82-89% sequence identity with the published tap sequence, and there was quite a high diversity of the sequences among the environmental isolates (up to 15% substitutions). The number of differences was lower at the amino acid level (96-100% amino acid identities).

Discussion
In general, the distribution of TET resistance did not differ between the soil and the clinical RGM and the same TETresistance genes were found in both groups. The gene tet(V) encoding a tetracycline efflux pump (19) was found in 73% soil isolates and 37% clinical isolates, including species where it has not been reported previously, i.e., M. septicum, M. litorale, M. alvei, M. peregrinum, M. goodii, M. arupense and M. neoaurum. This study also showed the high diversity of tet(V) (up to 17% difference between two sequences), which is unusual among other tet gene classes. For example, the published sequences of tet(B), a horizontally transferred TET efflux pump with at least 20 reported host genera (40), did not differ from each other in more than 1% nucleotides (analysis not shown). This finding, together with the positive correlation between the BOX-PCR profiles and tet(V) phylogeny, suggest that tet(V) evolved together with the replications, if they were higher than 50%. Letters following the isolate names indicate the BOX-PCR groups the isolates belonged to. mycobacterial genomes rather than being acquired horizontally. These findings therefore support the hypothesis that tet(V) belongs to the mycobacterium intrinsic resistome; however, in certain cases, different RGM species (e.g., M. arupense and M. septicum) shared the same partial tet(V) sequence, so the horizontal exchange of tet(V) among mycobacteria cannot be completely excluded. It is also conceivable that this gene underwent a long evolution within a certain group of RGM species from which it was horizontally transferred to other RGM species later. Comparative studies of the tet(V) gene and its sourroundings in a broader set of RGM species might be the next step to learn more about antibiotic resistance evolution in mycobacteria.
BLAST search revealed the presence of tet(V) homologues in the sequenced genomes of seven mycobacterial strains and in seven other actinomycetes; for example, there was a gene for H+ antiporter of M. vanbaalenii PYR-1 (86% identity), Mycobacterium sp. JLS (CP000580.1, 82% identity), Mycobacterium sp. KMS (CP000518.1, 82% identity) or major facilitator superfamily protein of Geodermatophilus obscurus DSM 43160 (CP001867.1, 73% identity). These genes may potentially code for drug resistance and therefore merit further attention. It is also possible that tet(V) is present in more mycobacterial species but has not been detected with PCR because of its high diversity and thus possible lack of complementarity with the primers used here.
The gene tap, which confers low-level resistance to tetracycline and certain aminoglycosides (4), seems to be very common among RGM. In our study, we found it in 84% environmental and 74% clinical isolates. Likewise, Esteban et al. (22) found tap in most (66%) of the clinical RGM tested. Besides the species in which tap has already been reported (22), it was detected also in M.  Surprisingly, tet(V) and/or tap were found also in the isolates that were highly susceptible to TET. Although our sequences of Tet(V) differed from that for which functionality was previously shown (19), there was no correlation between the tet(V) alleles and TET inhibition zone size; for example, two clinical M. fortuitum had the same sequence (at least in the part analyzed) but markedly differed in resistance. In general, sense mutations prevailed over the mutations changing the amino acid sequence, indicating that the function was rather to be maintained; however, M. neoaurum OS22 had one nucleotide deletion in the analyzed part of tet(V) that would change the reading frame, and this isolate was sensitive to tetracycline. It is possible that point deletions in the unanalyzed parts of the tet(V) sequence occurred also in other isolates. In addition, the observed discrepancy between genotype and phenotype may be due to mutations in distant regulation regions or to gene expression regulation; for example, Nash et al. (35) noticed that high clarithromycin resistance was inducible by overnight incubation of certain sensitive mycobacteria with a low concentration of the antibiotic. The discrepancy between the presence of a resistance gene and the resistance to TET in RGM or other bacteria was reported previously (5,22). TET-resistant isolates in which tet(V) and tap were not detected could either differ in their tet(V) and tap sequences from the used primers or possess other TET resistance genes that were not tested. The search for horizontally transferred TET resistance genes so far described in RGM was unsuccessful, though it included the genes tet(M) and tet(L), which are commonly found in Gram-positive bacteria in manure (1,3). The number of studies reporting horizontally transferred TET resistance genes in RGM is low (38,41) and it could be that there are marked local differences in the distribution of these genes in RGM.
The majority of the soil as well as clinical isolates from this study belonged to the Mycobacterium fortuitum group. Most soil isolates had the highest 16S rRNA sequence similarity to the species M. peregrinum, M. septicum or M. fortuitum, which all include potential human pathogens (29,48). Some of the isolates close to M. septicum indeed grew at 37°C (data not shown), indicating they may be able to colonize the human body; however, the soil and clinical isolates of M. septicum from this study were genetically different, as shown by analyses of their BOX-PCR profiles (and the same applies also to M. fortuitum isolates). The species in the M. fortuitum group are frequently recovered from soil (25), including agricultural soils (20,53), as well as from clinical samples (24,50), but so far a direct relationship between a clinical manifestation and exposure to soil has not been clearly shown in this group. The group M. chelonae-M. abscessus was reported as being even more frequent than the M. fortuitum group in clinical samples (24,50), but not in our case. The discrepancy between BOX-PCR groups and the assignment of isolates to species (e.g., BOX-PCR groups A, C or I) can be because the BOX-PCR groups were defined quite broadly (70% similarity) (12), but it can also indicate the need for more comprehensive taxonomical evaluation of RGM, as already shown in the case of streptomycetes (31).
In this study, we tested resistance to tetracycline and not to clinically used doxycycline, which was relevant in the context of soil since tetracycline residues are found in soil and manure (10). Ultimately, the zone sizes with tetracycline and doxycycline disks were correlated (tested with 32 isolates, Pearson correlation 0.88, P<0.01, data not shown). It was previously shown with the disk dilution method that RGM susceptible to tetracycline were also susceptible to doxycline, although the MIC values were 1-2 dilutions lower for doxycycline (52). Interestingly, there was no difference in the zone size distribution between clinical and environmental RGM and more than half of the isolates from both groups were resistant to tetracycline. The bimodal distribution of resistance to tetracycline antibiotics was reported previously (44,52). Both resistant and sensitive phenotypes occurred within one species (e.g., within M. fortuitum or within M. septicum), and even within one BOX-PCR group ( Table 2 and 3). We thus did not observe any consistency between BOX-PCR fingerprints and antibiotic resistance phenotypes, in contrast to Davelos Baines et al. (13), who found significant correlations between antibiotic phenotypes and BOX-PCR fingerprints in soil streptomycetes. This could be because the selection pressure for antibiotic resistance in soil acts on a much smaller scale than we sampled (13) and because antibiotic resistance in individual mycobacterial isolates can be affected by point mutations that would not significantly affect their BOX-PCR profiles. The variable resistance to tetracycline antibiotics within the M. fortuitum group is consistent with previous studies, usually reporting 40-60% sensitive isolates (24,44,52).

Conclusion
In conclusion, the studied soil and the clinical RGM from the Czech Republic did not differ in the distribution of TET resistance and occurrence of TET resistance genes, most possessing efflux-pump encoding genes tet(V) and/or tap. This study shows for the first time the presence of tet(V) and tap in soil mycobacteria. The correlation between the tet(V) phylogeny and isolate genomic profiles indicates that tet(V) belongs to the mycobacterial intrinsic resistom. The intrinsic efflux pumps may therefore play an important role in the antibiotic resistance of RGM, as in the case of M. tuberculosis (18,33). The simple presence of efflux-pump encoding genes, however, did not always match TET resistance. Further research should therefore be performed on gene diversity, distribution and expression in order to better understand the mycobacterial intrinsic resistom.