|Home | About | Journals | Submit | Contact Us | Français|
Mutation frequencies were studied in 174 Stenotrophomonas maltophilia isolates from clinical and nonclinical environments by detecting spontaneous rifampin-resistant mutants in otherwise-susceptible populations. The distribution of mutation frequencies followed a pattern similar to that found for other bacterial species, with a modal value of 1 × 10−8. Nevertheless, the proportion of isolates showing mutation frequencies below the modal value (hypomutators) was significantly higher for S. maltophilia than those so far reported in other organisms. Low mutation frequencies were particularly frequent among environmental S. maltophilia strains (58.3%), whereas strong mutators were found only among isolates with a clinical origin. These results indicate that clinical environments might select bacterial populations with high mutation frequencies, likely by second-order selection processes. In several of the strong-mutator isolates, functional-complementation assays with a wild-type allele of the mutS gene demonstrated that the mutator phenotype was due to the impairment of MutS activity. In silico analysis of the amino acid changes present in the MutS proteins of these hypermutator strains in comparison with the normomutator isolates suggests that the cause of the defect in MutS might be a H683P amino acid change.
Stenotrophomonas maltophilia is a Gram-negative, nonfermenting environmental bacterial species often isolated from the rhizosphere and from water sources (11, 12, 63). Some S. maltophilia strains have been used for bioremediation (13, 24, 73) or bioaugmentation (37). However, besides its environmental origin and potential relevance for biotechnological purposes, S. maltophilia is also a relevant human opportunistic pathogen (44) associated with a broad spectrum of clinical syndromes, such as bacteremia (79, 81), endocarditis (18), infection in cancer patients (1), and respiratory tract infections, including those suffered by cystic fibrosis (CF) patients (72, 77). One of the most problematic characteristics of S. maltophilia is its intrinsic high resistance to several antibiotics (4). This intrinsic antibiotic resistance is at least partly due to the presence in the genome of S. maltophilia (17) of genes encoding antibiotic-inactivating enzymes (6, 9, 30, 39, 42, 58) and multidrug resistance (MDR) efflux pumps (2, 3, 43, 78). More recently, a chromosomally encoded Qnr protein that contributes to the intrinsic resistance to quinolones of S. maltophilia has been described (67, 68).
A clear difference between infective (clinical) and environmental (nonclinical) S. maltophilia strains has not been reported (12, 63). However, although the available data fit the concept that opportunistic pathogens have not specifically evolved to infect humans (48), this does not mean that they do not evolve during the infective process. For most acute infections, we can presume that the time of in-host evolution is probably too short to detect relevant adaptive changes. Nevertheless, the situation might be different in chronic infections, such as those involving the bronchial compartment in CF patients. In this case, the same bacterial clone can be maintained and grow inside the host for years (62). This produces strong diversification over time and in different compartments of the lung (25, 71, 80), a process in which the acquisition of a mutator phenotype is important (52). Thus, isolates derived from an initial clone but presenting different morphotypes (47), different phenotypes of susceptibility to antibiotics (26) or in the expression of virulence determinants (14, 15, 36), or with different mutation frequencies (49, 60) are recovered from each individual patient suffering chronic infections. More recently, intraclonal diversification has also been described for Pseudomonas aeruginosa causing acute infections in intubated patients (38). Taken together, this indicates that bacteria can evolve during infection.
For different bacterial species, strains isolated from CF patients with chronic lung infections show high mutation frequencies (hypermutable strains) (19, 60, 61, 66), whereas hypermutators have rarely been found in isolates from acute infections (33). An explanation for this difference could be that hypermutable strains tend to be selected for in the highly compartmentalized environment of the infected lung by intensive antibiotic therapy, as well as by the stressful conditions of the habitat. This is a second-order selection process (75, 76), in which mutations are selected because they confer an advantage in clinical environments in such a way that mutator strains are selected because they can produce more mutants (both advantageous and deleterious) for selection. In cases of chronic infections that are treated, strong and maintained selective local processes might occur, either by antibiotic treatment or by the actions of the anti-infective systems of the host. Natural out-of-host open environments obviously might have local stresses. However, the intensity of selection is expected to be lower in these habitats, and a constant replacement of potentially lost organisms by migration of neighbor populations probably mitigates the local selection of mutators and favors the enrichment of bacteria presenting low mutation frequencies. In the case of chronic infections, the replacement of mutators by neighbor normomutators is unlikely, because those infections are produced by a single clone that remains for several years in the host (62). Furthermore, although the infection process presents strong evolutionary bottlenecks for bacterial populations, the human host also provides a constant temperature, reliable nutrient supplies, and a habitat largely free from predators and competitors. Thus, while hypermutation might increase the capability of bacteria to adapt to some specific challenges in the clinical environment, the cost of hypermutation in terms of deleterious mutations might also be diminished, and these effects might be mutually reinforcing.
The hypothesis explored in this paper is that S. maltophilia is adapted to deal with out-of-host fluctuating environmental variations but that once the organism enters a patient as an opportunistic pathogen, its adaptive needs significantly increase due to the actions of stressful local environmental conditions, such as the immune response and, when present, antibiotics. This enhanced stress under infective conditions might result in the selection of variants with increased mutation frequencies in a second-order selection process (75, 76). To test this hypothesis, the mutation frequencies of S. maltophilia clinical isolates (obtained from CF and non-CF patients) and from the environment (nonclinical origin) were compared. Most works that have been published on the different mutation frequencies in bacterial populations have focused on the detection of strains showing a high mutation frequency (mutators). In our work, we describe for the first time the presence of mutators in clinical isolates of S. maltophilia and demonstrate that hypermutation in several of those isolates is due to defects in MutS.
Nevertheless, our main goal has been the analysis of the global distribution of mutation frequencies in an ample number of samples from clinical and nonclinical environments. Our results indicate not only that mutators are more frequent in clinical S. maltophilia isolates, but also that the overall distribution of mutation frequencies is different in S. maltophilia populations with environmental or clinical origins, with a tendency toward mutation frequencies lower than the modal mutation value (hypomutators) in the environmental isolates.
A total of 174 S. maltophilia isolates were studied. Sixty were environmental strains (ENV) obtained mainly from the rhizospheres of different plants, although some of them were from the sea or from sewage water. Forty-eight isolates were recovered from 13 different CF patients (CFP). Sixty-six clinical isolates were obtained from 53 non-CF patients (NonCF) suffering from different infective processes. S. maltophilia DSM50170 (ATCC 13637, type strain t20, isolated from a patient with an oral carcinoma) was used as a reference strain (35) and was included in our study of the group of non-CF isolates.
The strains were spread on blood agar plates and grown for 24 h. Three tubes containing 2 ml of LB (8) were inoculated with one independent colony each obtained from the blood agar plate and incubated with agitation (150 rpm) for 24 h. One hundred microliters of a 10−6 dilution of the overnight cultures was seeded onto LB agar plates, and 500 μl of the overnight cultures was seeded onto LB-rifampin plates (250 μg/ml). Colony counts were performed after 24 h of incubation of the LB plates and after 48 h of incubation of the LB-rifampin plates. Mutation frequency values are reported as the number of rifampin-resistant colonies in proportion to the total viable count. The results for each strain corresponded to the mean value obtained in three independent experiments. The experiments were repeated again in triplicate in the following cases: zero colonies in an LB-rifampin plate, suspected plate contamination, or 10 times the difference in the standard deviation of the mutation frequencies obtained in the 3 independent experiments (usually due to a jackpot in one of the experiments). In these cases, the discordant value was excluded and three new values were included to obtain the mean. S. maltophilia DSM50170 (mutation frequency [f] = 1 × 10−8) was used in every set of mutation frequency determinations as a control strain.
Genomic DNAs were extracted using the GnomE DNA Kit (Q-Biogene). The primers used in the amplification of the genes involved in the mismatch repair (MMR) pathway (mutS, mutL, and uvrD) are shown in Table Table1.1. Amplifications were performed using 100 ng of genomic DNA, 2 mM MgCl2, PCR buffer II (Applied Biosystems, Weiterstadt, Germany), 200 μM (each) deoxynucleoside triphosphates, 25 pmol of each primer, 5% dimethyl sulfoxide (DMSO), and 2.5 U of Taq-Gold DNA polymerase (Applied Biosystems). The following PCR program was used: 94°C for 12 min and 30 amplification cycles of denaturation at 94°C for 1 min, annealing at 58 to 64°C for 1 min, and elongation at 72°C for 1 to 3 min (Table (Table2),2), depending on the size of the DNA fragment to be amplified. Finally, one cycle of amplification at 72°C for 10 min was carried out. The primers HylB and mutSR4.2 amplify a region of 0.9 kb in strain K279a, and the primers HylB and mutSR2 amplify a region of 2.1 kb in strain R551-3. Thus, these two sets of primers served to distinguish strains presenting a structure similar to either K279a or R551-3 upstream of mutS. The primers mutSF4.2 and KatA-R1 amplify regions of 3.0 kb in strain K279a and 1.1 kb in strain R551-3. They thus served to distinguish strains presenting a structure similar to either K279a or R551-3 downstream of mutS. Further confirmation of the structure downstream of mutS was obtained using the primers mutSF4.2 and wapA, which amplify a region of 0.9 kb present only in those strains that, like K279a, present the wapA gene downstream of mutS. In all cases, the amplicons were analyzed by electrophoresis on a 0.8% agarose-ethidium bromide gel, purified with the QIAquick PCR purification kit (Qiagen), and sequenced.
The mutS gene of S. maltophilia R551-3 was amplified using the oligonucleotides R5MutSf (5′CGGAATTCATGTCAAAAGAAAAGTCC3′; the EcoRI site is underlined) and R5MutSr (5′CCCAAGCTTTTACAGCAGCGCCTTCAACC3′; the HindIII site is underlined). The reaction was performed using the Expand Long Template PCR system (Roche), 500 ng of genomic DNA of S. maltophilia R551-3 as template, 350 μM each deoxynucleotide triphosphate [dNTP], and 0.5 μM each primer. The reaction had one denaturation step at 94°C for 5 min, followed by 10 amplification cycles of 94°C for 30 s, 55°C for 30 s, and 68°C for 3 min, followed by another 20 amplification cycles of 94°C for 30 s, 60°C for 30 s, and 68°C for 3 min, with a final extension step of 68°C for 7 min. The PCR product was cloned into the pGEM-T plasmid (Promega), generating the pBS14 plasmid. The inserted fragment was sequenced by Secugen (http://www.secugen.es) to ensure that no mutations were introduced during PCR. The pBS14 plasmid was EcoRI-HindIII digested, and the fragment containing the mutS gene was cloned into pVLT31 (20), in the same enzyme restriction sites, generating pBS15. This plasmid was introduced into Escherichia coli CC118λpir and mobilized into S. maltophilia CFP1-43, CFP1-44, NonCF-22, and NonCF-37, using the E. coli 1047(pRK2013) strain as a helper, by triple conjugation (21) at a rate of 4:1:2 (receptor-donor-helper) and using M9 (8) to resuspend the filter with the mating mixture. The exconjugants were selected on LB plus tetracycline and imipenem. The concentrations used were 20 μg/ml (strains NonCF-22 and NonCF-37) and 50 μg/ml (strains CFP1-43 and CFP1-44) tetracycline and 20 μg/ml imipenem, except NonCF-22, which was selected with 2 μg/ml imipenem. To check the presence of pBS15, several colonies from each conjugation were grown in LB with tetracycline at the concentrations stated above, and their plasmids were extracted with the Wizard Plus SV Minipreps DNA Purification System (Promega) and further analyzed by SalI digestion and electrophoresis on a 0.8% agarose-ethidium bromide gel.
Statistics tests were performed using the XLSTAT package. To compare the distribution of mutation frequencies in the full populations, Wilcoxon-Mann-Whitney (45) and Kolmogorov-Smirnov (27) tests were performed. To compare categories, the chi-square test was performed.
Sequence similarities to already-sequenced MMR genes were determined using the BLAST algorithm at NCBI (http://blast.ncbi.nlm.nih.gov/Blast.cgi). Alignments were performed using ClustalW2 at EMBL-EBI (http://www.ebi.ac.uk/Tools/clustalw2/index.html). For phylogenetic analysis, maximum-likelihood phylogenetic trees were obtained using PHYML software (32). The general time reverse (GTR) + I + G model of nucleotide substitutions (74) was used as an evolutionary model, including a gamma distribution with six rate categories and a fraction of invariant sites to account for substitution rate heterogeneity among sites. The robustness of the maximum-likelihood topology was evaluated by the approximate likelihood ratio method (7) with 1,000 bootstrap pseudorandom replicates.
Several works have shown that bacteria can evolve in the course of an infection so that different isolates belonging to the same clone and from the same patient can present different characteristics (25, 71, 80). This situation is particularly relevant when mutation frequencies are studied, because bacteria can evolve toward hypermutation during in-host growth and isolates presenting different mutation frequencies can coexist within each patient (49, 60). For this reason, an accurate estimation of the mutation frequencies of bacteria from clinical habitats requires the analysis of different isolates from the same patient, mainly when bacteria are producing chronic infections. Our collection thus comprises independent environmental isolates from non-CF infections distributed as follows: 9 patients with two isolates each, 2 patients with three isolates each, and 42 patients with a single isolate each. In the case of CF patients suffering chronic infections, the number of isolates analyzed per patient was higher in order to get a reliable picture of the distribution of mutation frequencies in the disease. We thus analyzed three patients with one isolate each, two patients with two isolates each, two patients with three isolates each, one patient with four isolates, one patient with five isolates, three patients with six isolates each, and one patient with eight isolates. To ascertain whether, as for other organisms, isolates from the same patient suffering chronic infection could present different mutation frequencies, the mutation frequencies estimated for each isolate were plotted against the patient's origin. Consistent with the data obtained for other bacterial species, the mutation frequencies of the isolates obtained from a given patient presented high variability (Fig. (Fig.1).1). The overall distribution of rifampin mutation frequencies in the collection of 174 isolates is shown in Fig. Fig.2.2. For comparison purposes, the figure also shows the mutation frequencies of E. coli populations already published by our group (10). Although the modal values of mutation frequencies were the same for both bacterial species, 1 × 10−8, the mutation frequencies were more evenly distributed in S. maltophilia, and a shift toward low mutation frequencies compared with the values previously observed in E. coli (10) was found.
The difference in the distribution of mutation frequencies between the two species was significant using the Wilcoxon-Mann-Whitney (P = 0.019) and the Kolmogorov-Smirnov (P < 0.0001) tests. This shift toward low mutation frequencies was even more evident for S. maltophilia environmental strains (Fig. (Fig.3),3), with nearly 60% of the isolates presenting mutation frequencies below the modal value. The analysis of the median (4.4 × 10−9 in the environmental strains and 1.6 × 10−8 in the CF isolates) further demonstrated that the mutation frequencies were lower overall in S. maltophilia environmental isolates than in those isolates with a clinical origin.
It was previously proposed for E. coli that the distribution of mutation frequencies can be grouped into four different categories: hypomutators, normomutators, weak mutators, and strong mutators (10). According to this scheme, the S. maltophilia isolates were also classified into four categories based on their mutation frequencies (f): hypomutators (f ≤ 8 × 10−9), normomutators (8 × 10−9 < f < 4 × 10−8), weak mutators (4 × 10−8 ≤ f < 4 × 10−7), and strong mutators (f ≥ 4 × 10−7). The breakpoints for the different categories were established on the basis of the minimum values observed in the function of the distribution of mutation frequencies (Fig. (Fig.22).
The distribution of the strains in these categories, as a function of their origins (environmental, acute infection, or chronic infection), is shown in Table Table3.3. The population of environmental strains was significantly enriched in hypomutators compared with the population of clinical isolates (P = 0.0047, determined by the chi-square test). On the other hand, a tendency toward mutation frequency values above the modal value (weak and strong mutators) was observed among clinical isolates. The incidence of strains with increased mutation frequencies was significantly higher among the clinical isolates than among those with an environmental origin (P = 0.0027, determined by the chi-square test). Thus, while 19.7% of S. maltophilia isolates from acute infections and 27.1% from chronic infections presented increased mutation frequencies, only 5% of environmental S. maltophilia isolates yielded mutation frequencies higher than the modal value.
Although several isolates from the same patient must be analyzed to get a full picture of the distribution of mutation frequencies when isolates from chronic infections are studied, this might produce overrepresentation of some strains. To alleviate this problem, the same statistical chi-square analysis of the distribution of mutation frequencies among the different categories was performed, but in this case taking into consideration only one isolate per category from each patient (Table (Table4).4). Under these constraints, the results were the same as before: the number of hypomutators was significantly higher in the environmental strains than in the clinical isolates (P = 0.0055), and the clinical isolates were enriched in strains presenting increased mutation frequencies (P = 0.0034).
In all cases, the environmental strains presenting increased mutation frequencies were weak mutators. Ten strains, all of them with a clinical origin, showed high mutation frequencies (strong mutators). Eight of them were isolated from the same CF patient, and two strains were isolated from the urine and blood of two different non-CF patients. The proportion of non-CF patients with at least one strong mutator was 3.7% (2 of 53 patients), whereas this percentage increased to 7.7% (1 of 13 patients) when the isolates came from CF patients. However, these differences are not statistically significant, so it cannot be stated at the moment whether the presence of S. maltophilia strong-mutator strains is higher in patients suffering chronic infections than in patients with nonchronic infections.
Since one of the CF patients (CFP1) presented a large number of isolates with increased mutation frequencies, such strains were analyzed in more detail. Eight identical or clonally related isolates (pulse type 8 ) were obtained at different times over 6 years of chronic infection in this patient (CFP1). The first S. maltophilia isolate (strain CFP1-91) was isolated from patient CFP1 in September 1991 and corresponded to the strong-mutator category (f = 5.6 × 10−7). Isolates belonging to the same pulsed-field gel electrophoresis (PFGE) clone, also presenting a hypermutable phenotype (strong mutators), were recovered from different sputum samples obtained 14 days (strain CFP1-43; f = 8.7 × 10−7) and 1 and 2 months (strains CFP1-44 [f = 5.6 × 10−7] and CFP1-45 [f = 4.9 × 10−7]) after the first isolate. Hypermutable strains of the same clone were also recovered 3.5 years later (strain CFP1-55; f = 7.4 × 10−7), 4.5 years later (strain CFP1-56; f = 4.3 × 10−7), and 6 years later (strains CFP1-30 [f = 4.2 × 10−7] and CFP1-69 [f = 4.4 × 10−7]). This long-standing infection by strong mutators indicates that the fitness costs associated with hypermutation are not high enough to impede chronic infection of the lung by hypermutable S. maltophilia strains. A similar clonal long-term persistence has been found in P. aeruginosa strains recovered from the lungs of patients with CF (60). In these cases, the molecular characterization of the mutator phenotype revealed defects in the activity of the MMR system proteins (51, 59).
Using the available S. maltophilia genome databases, a search for the presence of genes homologous to those previously reported as important for a hypermutation phenotype was performed. To that end, we searched the annotated sequences, and when the corresponding gene was not found, we searched for genes encoding proteins with homologies higher than 80% to the corresponding Xanthomonas campestris genes. Both S. maltophilia strains contained the genes from the methyl-directed MMR system, the guanine oxidation (GO) system, the nucleotide excision system, the recombination repair system, and the SOS system (Table (Table5).5). In all cases, the genes were highly similar in both S. maltophilia strains (more than 90% identity in most cases), and the similarity of the proteins was even higher, with levels of identity ranging from 94% to 99% and similarities (including conservative amino acid changes) ranging from 96% to 99%. These data indicate that the genes have a common and ancient origin in S. maltophilia (core genome) and have not been recently acquired by horizontal gene transfer. Furthermore, the observed low numbers of nonsynonymous changes support the notion that these genes are not diverging but rather under stabilizing selection (65).
The analysis of natural bacterial isolates has shown that the most frequent causes of hypermutation are defects in the MMR system, mainly in MutS (16, 34, 53, 59, 64). Both sequenced S. maltophilia strain genomes contain mutS, mutL, and uvrD genes. These MMR genes were highly similar in both S. maltophilia strains, with percentages of identity ranging between 90% and 95%. We did not find in either strain an open reading frame homologous to the mutH gene of E. coli, as is also true for other bacterial species phylogenetically close to S. maltophilia, such as P. aeruginosa (59), Xanthomonas, or Xylella (50).
Since defects in MutS are the major cause of hypermutation in natural bacterial isolates (16, 34, 53, 59, 64), we analyzed whether the expression of the wild-type form of the protein in the hypermutator strains could restore mutation frequencies to their normal levels. To assess this possibility, the wild-type mutS gene of S. maltophilia was cloned as described in Materials and Methods and introduced by triparental mating in the strong-mutator strains CFP1-43, CFP1-44 (representative of the eight mutator strains from patient CFP1), NonCF-22, and NonCF-37. The mutation frequencies of the CFP1-43, CFP1-44, and NonCF-22 strains were reduced to the levels of a normomutator strain (from f = 8.7 × 10−7 to f = 1.3 × 10−8, from f = 5.6 × 10−7 to f = 2.0 × 10−8, and from f = 8.7 × 10−7 to f = 5.5 × 10−9, respectively) when mutS was overexpressed (Fig. (Fig.4).4). The small interstrain differences observed in the reduction of the mutation frequencies after the wild-type allele of MutS was expressed might be the consequence of non-MutS changes acquired by the analyzed strains during in-host evolution, which might influence the mutation frequencies. For instance, it has been reported that changes in the level of MutL can alter the hypermutator phenotype due to MutS defects in E. coli (28).
In any case, our results clearly show that the strong-mutator phenotype displayed by these isolates is associated with defects in the MutS protein.
Since the sequences of mutS in all isolates from CFP1 are nearly identical (see below) and those isolates belong to the same S. maltophilia clone that was established during the chronic infection of CFP1, the same defect in MutS is likely the cause of hypermutation in all isolates from CFP1, as well as in strain NonCF-22. In other words, the cause of hypermutation in 9 of the 10 analyzed S. maltophilia strong mutators is a defect in MutS. For the other strain, NonCF-37, complementation with the wild-type MutS protein did not reduce the mutation frequency (from f = 5.2 × 10−7 to f = 5.7 × 10−7), indicating either that this strain contains a dominant defective mutS allele or (most likely) that the hypermutation phenotype of NonCF-37 is not due to defects in MutS.
In order to establish which changes in the mutS gene might be responsible for the mutator phenotype in the strains that recovered the normomutable phenotype upon complementation with wild-type MutS (isolates from CFP1 and strain NonCF-22), the entire sequence of the gene mutS (as well as those of mutL and uvrD) were determined for six strains belonging to the chronic-colonizing clone of patient CFP1 (CFP1-91, CFP1-43, CFP1-44, CFP1-45, CFP1-56, and CFP1-30), as well as for the two strong-mutator isolates from non-CF patients (NonCF-22 and NonCF-37). The same genes from four normomutator strains isolated from another CF patient (patient CFP2; strains CFP2-40, CFP2-62, CFP2-90, and CFP2-98) were also fully sequenced. We did not detect insertions, deletions, or nonsense mutations that might produce aberrant proteins in any of the sequenced genes compared with the K279a and R551-3 reference fully sequenced strains. However, we detected several polymorphisms, some of which produced amino acid changes compared with K279a or R551-3 (Tables (Tables6,6, ,7,7, and and8).8). The same amino acid changes in MutS were observed in isolates from CFP1 and in NonCF-22. This might suggest an episode of cross-infection. Indeed, patients CFP1 and NonCF-22 were at the hospital at the same time (although in different places). Nevertheless, the DNA sequences of mutL, uvrD, and mutS from NonCF-22 presented, respectively, 1, 5, and 12 nucleotide changes in comparison with CFP1 isolates. Thus, although these isolates likely share an origin, they are not exactly the same strain cross-infecting two patients. This might suggest that this specific S. maltophilia clone might be well adapted for producing infection; however, our epidemiology data are not robust enough to support this statement.
Notably, the mutS gene of strain CFP2-40 shows a patchwork structure (not shown), with its 5′ region very similar to that of strain K279a and its 3′ end identical to those of strains from CFP1. It has been suggested that the mosaic structure shown by MMR genes in E. coli is the consequence of the horizontal gene transfer and recombination of those genes (23). The mosaic structure of the mutS gene of CFP2-40 might be the consequence of a similar process in S. maltophilia.
The observed changes caused S. maltophilia strains to cluster in specific groups (Fig. (Fig.5).5). To ascertain whether these changes might be the consequence of the existence of different evolutionary lineages in S. maltophilia, the genes mutL and uvrD were sequenced in these isolates, as well. As shown in Tables Tables66 to to88 and Fig. Fig.5,5, very similar clusters were obtained from the independent analysis of each of the three proteins, indicating than most of the observed amino acid changes are likely the result of the evolution of subspecific S. maltophilia lineages, rather than recent adaptive changes selected during infection. In line with this possibility, the inspection of the genome region surrounding the MMR genes in the sequenced S. maltophilia isolates enabled us to detect differences between the two strains, mainly around mutS (Fig. (Fig.6a).6a). Furthermore, the analysis of the genomic regions surrounding mutS in a set of strains presenting different mutation frequencies showed four different genomic organizations (Fig. (Fig.6b).6b). Altogether, these data indicate that S. maltophilia is formed by different lineages, as previously suggested (31).
The analysis of the sequences of MutS in the strong-mutator strains that are complemented by the wild-type form of the protein and the strain that is not complemented provides clues for understanding in more depth the causes of hypermutation in the analyzed S. maltophilia strong mutators. As could be predicted from the complementation assays, the MutS protein from the strong-mutator strain NonCF-37 did not present any relevant change compared with known normomutator strains. In fact, the sequence of MutS was identical for NonCF-37 and the normomutator strain K279a (Table (Table5).5). Since the sequence of MutL is identical in NonCF-37 and in the sequenced normomutator K279a and the sequence of UvrD for NonCF-37 is identical to the sequence of the normomutator strain CFP2-40 (Tables (Tables77 and and8),8), it is likely that another gene(s) not related to the MMR system could be involved in the high mutation frequency shown by strain NonCF-37. As stated above, the prevalent cause of hypermutation in natural bacterial isolates is defects in the MMR system (16, 34, 53, 59, 64). However, in all studied bacterial species, some hypermutators do not present any defect in the MMR proteins, and the cause of hypermutation in these isolates, as for NonCF-37, is unknown (59).
The situation for the strong mutators that were complemented by the wild-type MutS protein is clearly different. All of them presented several amino acid changes in the MMR proteins compared with S. maltophilia normomutator strains. Since mutation frequencies were complemented with the wild-type MutS allele, mutations in MutS must be relevant for acquiring a strong-mutator phenotype in these isolates. To predict which one of the 10 amino acid changes (boldface in Table Table6)6) in the MutS protein that are unique to these strong mutators might be relevant for developing a mutator phenotype, we performed an in silico analysis of those changes in comparison with available sequences from normomutator strains belonging to other bacterial species. Six of these amino acids changes were present in other bacteria: T138 was present in S. maltophilia strain SKA14 and in different isolates of Xylella fastidiosa; D187, E574, and A576 were present in Xylella campestris; and S569 and A576 were present in Coxiella burnetii. Finally, T575 is present in the MutS protein of a marine gammaproteobacterium and in Vibrio angostum, and the amino acid at this position is variable in the available MutS sequences. Since the amino acids at these positions are present in MutS proteins from other bacteria, it is unlikely that they are involved in the MutS defect of the strong-mutator strains. The only changes that are unique to these strains and thus could be responsible for the defect in MutS activity, which produces a strong-mutator phenotype, are Q191H, G263A, S833N, and H683P. We have therefore analyzed which of these positions are conserved in the available MutS sequences. In members of the genus Xanthomonas, the amino acid at position 191 in MutS is G. Also, the sequence around this amino acid is not highly conserved among the different available sequences. This indicates that the region is polymorphic and suggests that the Q191H change might not be relevant to the lack of activity of MutS in the strong-mutator strains. The same applies to the G263A change. The amino acid at this position is S in several Xanthomonas MutS proteins and is D in Dehalococcoides, and the region is not highly conserved, which suggests that the observed change in S. maltophilia strong-mutator strains is more a polymorphism than a mutation that impedes the activity of MutS. Similarly, the amino acid at position 833 is A in X. fastidiosa, in Edwardiarsiella ictauri, and in gammaproteobacterium HTCC5015 and is a D in Herminiomonas arsenicoxydans and in Polaromonas naphthalenivorans; as for positions 191 and 263, the sequence of the surrounding region is variable among the different available MutS sequences. In contrast with this, the amino acid at position 683 is always H in the available MutS sequences of different species from an ample array of bacterial genera (Fig. (Fig.7).7). Furthermore, the region around this position is also highly conserved in these bacterial genera, indicating that the structure of this region is important for MutS activity. The region is inside domain V of MutS, which contains the Walker motif, involved in ATP binding, and is highly relevant for MutS dimerization and activity (40, 55, 57, 70). Altogether, our analysis suggests that the H683P mutation, which is a nonconservative amino acid change, might be responsible for the MutS defect and consequently for the increased mutation frequencies of the strains isolated from CFP1 and strain NonCF-22.
The presence of hypermutator strains in bacterial populations has been recognized for more than a decade (41). However, the polymorphic distribution of mutation frequencies in different categories has been proposed only recently (10), and the selective forces shaping these categories are largely unknown. The data presented in the present work indicate that environmental isolates of S. maltophilia present overall lower mutation frequency values than clinical strains. Two alternate hypotheses might explain these results. First, adaptation of opportunistic pathogens with an environmental origin to the new ecological conditions at the infective site might be favored by increased mutation frequencies. Either the preexisting variants with increased mutation frequencies have greater chances to colonize an infective habitat or increased mutation frequency is selected for during the infection by a second-order selective process because strains with higher mutation frequencies can generate more adaptive alleles to be selected under selective pressure during infection. Second, S. maltophilia species are comprised of different lineages, some of them able to tolerate and colonize (and finally infect) the human host and others noncolonizers or only transient colonizers, as has been observed for Burkholderia cepacia (22) and opposite to what has been described for P. aeruginosa (5, 54).
Our results support the hypothesis that colonization of novel habitats (in this case from the nonclinical environment to the body of the infected patient) selects bacteria able to cross adaptive bottlenecks because of their increased mutation frequencies (46). This is the case for chronic lung colonization of CF patients (60), where microorganisms are subjected to stress conditions due to the need for adaptation to higher-than-natural environmental temperatures, immunological challenges, and antibiotic exposure. In our work, there was a total absence of strong-mutator strains among S. maltophilia environmental strains, while the percentage of strong mutators was 3% in the case of isolates from acute infections, increasing to 14.6% for isolates from CF patients.
As for other bacterial species (16, 34, 53, 59, 64), the phenotype of high mutability is associated in S. maltophilia with defects in the MutS protein, at least in three out of the four studied strains. Since all strains from CFP1 were strong mutators and presented the same sequence in the MutS gene, it is likely that the cause of hypermutation in all the strains from this patient was the same defect in MutS. In silico analysis of the sequence of this protein in comparison with available databases suggests that the cause of the defect might be an H683P mutation, which changes a highly conserved histidine residue in domain V of MutS. Besides this change, we have detected other polymorphisms in the studied S. maltophilia strains in the sequences of the MMR genes mutS, mutL, and uvrD. These polymorphic patterns might be attributed to the existence of different lineages, each one with a specific signature in the sequences of the MMR genes. The existence of lineages in S. maltophilia has been previously proposed by Gould and Avison, using the smeT-smeD intergenic region (69), to describe the existence of at least four phylogenetic groups in S. maltophilia (29). The observed polymorphic patterns in the MMR genes, together with the different gene organizations observed around such genes (Fig. (Fig.6),6), further supports this notion.
A relevant aspect of our results is the large number of S. maltophilia strains with a hypomutator phenotype, mainly those isolates with an environmental origin. Most studies of bacterial mutation frequencies have been done with human-associated species, either pathogenic or nonpathogenic. Interactions with metazoans (56) might be a cause of stress for bacterial populations, since metazoans have evolved complex systems of defense. Whether non-human-linked environments are less challenging for bacterial populations than clinical settings and favor the establishment of bacterial populations presenting low mutation frequencies remains to be established. Alternatively, environmental strains are possibly adapted to a variety of predictable environmental changes and consequently less stressed by them (environmental canalization). Further work on non-mammal-linked environmental bacterial species is needed to address the evolution of mutation frequencies and its relation to the ecological behavior of bacterial populations.
This work was supported by grants BIO2008-00090 from the Spanish Ministerio de Ciencia e Innovación; FIS PI080624 from the Ministerio de Sanidad y Consumo (Instituto de Salud Carlos III); and LSHM-CT-2005-518152, KBBE-227258 (BIOHYPO), and PAR from the European Union.
Published ahead of print on 22 January 2010.