Material and Methods Bacterial strains. Construction of C4750 strain derivatives. Mutagenesis assays.

Transcription

1 Material and Methods Bacterial strains. 787 E. coli strains have been used in this study: (i-iii) 547 were isolated from fecal samples, (iii and iv) 177 were pathogenic isolates, (v) 62 were isolated from secondary environments, and the MG1655 E. coli K12 laboratory strain was included. (i) 298 commensal strains sampled mostly from wild mammals and birds were collected in: Mexico (10 from reptiles, 49 from birds and 198 from mammals), Venezuela (2 from mammals), Antarctic (2 from birds and 10 from mammals) and Australia (27 from mammals). (ii) 188 human commensals were collected in Croatia (61), France (70) and Mali (57). (iii) 61 commensal strains (29 human and 32 animal) and 11 urinary tract infections (UTI) strains from the E. coli reference (ECOR) collection have been studied (1). (iv) 166 pathogenic strains were isolated from humans in France, 80 during the course of extra-intestinal (UTI) and intra-intestinal [13 enteroinvasive E. coli (EIEC) and 73 Shigella] infections. (v) 62 strains were collected from secondary environments in Mexico (15 air, 14 mud, and 33 water). Construction of C4750 strain derivatives. C4750 is a human commensal strain isolated in Croatia. It belongs to B1 E. coli phylogenetic group (2). This strain was used as a model for genetic analysis because it has strong MAC phenotype (see Table 2 in an article). C4750 mutant derivatives were constructed by P1-mediated transduction of the following alleles: rpos359::tn10 and hfq-1::ωcm R (provided by M. Winkler, University of Texas, Huston, Texas, USA); rssb::cm R (provided by Henge-Aaronis, Free University, Berlin Germany); cyaa ilv::kan R, lexa1(ind - ) malb::tn9 and pola1 zgj::tn10 (provided by R. D Ari, Jacques Monod Institute, Paris, France); crp::cm R (provided by J. Blazquez, Hospital Ramon y Cajal, Madrid, Spain); reca306 srl::tn10 and polb 1::ΩSm R -Sp R (provided by D. Touati, Jacques Monod Institute, Paris, France); uvra::cm R (provided by N. Goosen, Leiden University, Leiden, Netherlands); recb268::tn10 (provided by R.G. Lloyd, University of Nottingham, Nottingham, UK); and muts201::tn5 (our strain collection). Plasmids used in this study were: pacyc184 and its derivatives pmq341 and pmq339 carrying muts and mutl genes, respectively (provided by M.G. Marinus, University of Massachusetts Medical School, Worcester, MA, USA). Mutagenesis assays. The frequency of mutation to antibiotic resistance was measured by using a modification of a previously described method (3) to 10 3 cells from an overnight culture were inoculated on nitro-cellulose filters (NC-45, Schleicher and Schuell) laid on fresh

2 869 plates (NaCl 5g/l, Bactotryptone 10g/l, yeast extract 5g/l, agar 15g/l). Each plate contained 3 filters [instead of one per plate as in (3)]. Plates were incubated at 37 for 1 to 7 days. Anaerobic growth conditions were created in jars using «GENbox anaer» generator (BioMérieux). Bacterial colonies were resuspended in 1 ml of 869 and incubated for 1 hour at 37 to allow for antibiotic resistance expression. Appropriate dilutions were then spread on 869 and 869 antibiotic plates. The antibiotic resistant mutants were scored after 24 h [instead of 48 h as in (3)] of incubation at 37 C. The frequency of mutation was calculated from at least 3 independent experiments, each with three independent cultures. The antibiotics were added as follows: 5-fluoro-uracil 5 µg/ml, mecillinam 10 µg/ml, nalidixic acid 40 µg/ml, phosphomycin 30 µg/ml, rifampicin 100 µg/ml and streptomycin 100 µg/ml. We have also used eleven lacz alleles that allow detection of all base substitutions and five frameshifts (4, 5). A set of C4750 strains mutants were constructed by their co-transduction of lacz and laci42::tn10 alleles (provided by D. Touati, Jacques Monod Institute, Paris, France) from eleven F episomes to C4750 chromosome. The rich medium used in our experiments contained traces of lactose or a lactose-like carbon source, which rendered exact measurements of mutagenesis impossible. Therefore, we measured mutagenesis in colonies that have been incubated seven days on agarose plates (without addition of any other substance) as follows cells from overnight cultures were washed three times in 10-2 M MgSO 4, inoculated on nitrocellulose filters (NC-45, Schleicher and Schuell) and laid on fresh 1.5% agarose plates. Plates were incubated at 37 for 1 to 7 days and mutants scored on M9 lactose medium (after 48h of incubation at 37 ). In parallel, dilutions were spread on 869 and 869 rifampicin plates. Sequencing C4750 rpob Rif R alleles. The part of rpob gene, known to contain almost all rpob mutations responsible for the rifampicin resistance (6), was amplified using PCR. The PCR primer sequences were: 1525-up: GGC GAT CTG GAT ACC CTG ATG C and 2198-down: CGG AGT CAA CGG CAA CAG CAC. Sequencing was performed by the «Centre d Etude du Polymorphisme Humain», Paris, using Big Dye Terminator Cycle Sequencing Ready Reaction kit with AmpliTaq DNA polymerase, FS (Perkin-Elmer, Applied Biosystems Division) and analyzed in an automated ABI PRISM 377 XL DNA sequencer (Perkin-Elmer, Applied Biosystems Division) following the manufacturer s instructions. Competition between different C4750 rpob alleles. Four different Rif R mutants (isolated on D1), each having a G:C->A:T mutation in a different position in the rpob gene, were

3 used in competition experiments that have been performed as follows: The «instant colonies» having about 10 9 cells, with 1/10 2, 1/10 3 and 1/10 4 Rif R /Rif S ratios, have been reconstructed. After seven days of aging in colonies, none of the Rif R mutant populations exceeded the initial input frequency. Simulation of the selection of stress-induced mutator alleles. To study the selection acting on alleles modifying the mutation rate under conditions encountered by bacteria in aging colonies, we have used modified individual-based model as previously described (7). We studied the response of clonal bacterial population (of 10 9 cells) to directional selection. The adaptive peak was characterized by 17 beneficial mutations (8, 5, 2 and 1 mutations of fitness effect 2%, 2.5%, 5% and 10% appearing with frequencies 4.34, 4.9, 4 and 2 x 10-9, respectively) following an exponential distribution with a mean mutation rate of 4 x 10-9 and a mean fitness effect of 3% [similar to the one estimated by Gerrish and Lenski (8)]. The rate of lethal mutations was fixed to 10-5 and that of deleterious mutations to 1.7 x 10-4 (with a 1% effect on fitness per deleterious mutation) (9). All fitness effects were multiplicative. The population grew exponentially from a single cell to 10 9 cells. On the average, every 30 generations, the population was submitted to a stress for a period of (on the average) 10 generations. During this phase of stress, the population size decreased by 0.05% for every generation and inducible mutator alleles were induced leading to an increased mutation rate. Using this model we studied the time needed for the population to reach the adaptive peak and the probability of fixation of a constitutive or inducible mutator allele emerging at a frequency of 5 x Because mutations that would improve survival to stress would inevitably increase the selection of inducible mutators, to be conservative, we chose not to take them into account in our model. Mutations that would specifically improve survival to general stresses (e.g., starvation) may sometimes exist in nature. However, it is likely that most of such mutations have already been fixed in bacterial genome, which have regularly encountered starvation for millions of years. If some of these mutations are not fixed, it should be due to some tradeoff, (i.e., these mutations being counter-selected during non-stressful conditions) and therefore they would not contribute to the long-term adaptation considered here. Supplementary Data

4 MAC conferring resistance to different antibiotics. All strains have been tested for their capacity to generate mutations conferring resistance to rifampicin. In addition, 10 natural isolates were randomly chosen among strong MAC mutators (based on test with rifampicin), and mutations conferring resistance to 5-fluorouracil, mecillinam, phosphomycin, streptomycine and nalidixic acid was measured in aging colonies. None responded to the same level as for rifampicin, but the increase was significant for 5-fluorouracil (D7/D1=3.4; Mann-Whitney: p=0.004), mecillinam (D7/D1=1.9; Mann-Whitney: p=0.01) and phosphomycin (D7/D1=4.8, Mann-Whitney: p=0.009). No increase between D1 and D7 was observed when mutagenesis to streptomycin and nalidixic acid resistance was measured. The frequency of mutants resistant to those two antibiotics even decreased at D7 relative to D1 for most of the natural isolates, suggesting that the majority of mutations conferring resistance to nalidixic acid and streptomycin (residing in gyrb and rpsl genes, respectively), may be deleterious or lethal to bacteria in aging colonies. laci papillation assay. The appearance of mutants was also monitored on minimal media plates using a laci papillation assay (10). Out of 392 randomly chosen natural isolates, 29% showed a laci-papillating phenotype. The increase in MAC to Rif R (on rich-media plates as described above) was 16.2 and 5.2-fold for laci papillators and non-papillators, respectively (Mann-Whitney p=0.0003). Thus, there was a positive correlation between MACs in the two mutation detection systems (in rpob and laci genes) under two different experimental conditions. Growth and survival of the C4750 derivatives in aging colonies. D7/D1 ratios of the median number of viable cells/colony for C4750 derivatives were as follows: 1.7 for parental strain, 0.6 for rpos, 0.7 for hfq, 1.2 for rssb, 0.5 for cyaa, 2.6 for crp,1.7 for reca, 0.6 for lexa1, 2.2 for polb, 0.3 for uvra, 3.6 for pola, 4.2 for recb, 1.0 for muts, 1.7 for parental strain carrying pacyc184 plasmid, 1.7 for parental strain carrying muts + plasmid and 1.3 for parental strain carrying mutl + plasmid. These data were obtained from at least 3 independent experiments (each with three independent cultures) for each genotype. The results showed that there is no clear correlation between the change of the number of viable cells/colony and the MAC (see Table 2 in an article). Selection of stress-inducible and constitutive mutator alleles. The simulations showed that the selection acting on a stress-inducible mutator allele (increasing X fold the mutation rate under stress) is only a little bit less efficient than the one acting on a constitutive mutator allele of

5 the same averaged mutagenic effect (X ) [X = K X+(1-K); K being the fraction of generations spent under stress]. This small difference is not due to the net production of deleterious mutations, because both mutator mechanisms generate the same amount of deleterious mutations in the long run (data not shown). However, during the strong burst of mutagenesis, beneficial mutations are more likely to co-occur with deleterious mutations (data not shown) and therefore to be lost. Hence, stress inducible mutators will produce less selectable beneficial mutations than their constitutive equivalents. The difference in selection of two types of mutator alleles may also be the consequence of the frequency of stresses. We observed that the selection of stress inducible mutators is less efficient when stresses are not frequent (data not shown). Since over-production of mutations in stress-inducible mutator genotype is conditional on the presence of stress, the increase in the generation of beneficial mutations by stress-inducible mutators, will be delayed compared to constitutive mutator. References 1. H. Ochman, R. K. Selander, J. Bacteriol. 157, (1984). 2. P. Duriez et al., Microbiology 147, (2001). 3. F. Taddei, I. Matic, M. Radman, Proc. Natl. Acad. Sci. USA 92, (1995). 4. C. G. Cupples, M. Cabrera, C. Cruz, J. H. Miller, Genetics 125, (1990). 5. C. G. Cupples, J. H. Miller, Proc. Natl. Acad. Sci. USA 86, (1989). 6. D. Jin, C. Gross, J. Mol. Biol. 202, (1988). 7. O. Tenaillon, B. Toupance, H. Le Nagard, F. Taddei, B. Godelle, Genetics 152, (1999). 8. P. J. Gerrish, R. E. Lenski, Genetica , (1998). 9. T. T. Kibota, M. Lynch, Nature 381, (1996). 10. Y. Nghiem, M. Cabrera, C. G. Cupples, J. H. Miller, Proc. Natl. Acad. Sci. USA 85, (1988).