To select S. Typhimurium spontaneous mutants with increased resistance to cefotaxime (CTX), L-3553 (Shahada et al., 2011) harboring GI-VII-6 in the chromosome was used as the parent strain. Whole-genome sequence of this strain was reported previously (Sekizuka et al., 2014). The MIC of CTX against this strain was 8 mg/L. This strain was inoculated into 3 ml of Luria–Bertani (LB) broth (Becton, Dickinson and Company, Franklin Lakes, NJ, USA) and incubated at 37°C and 150 rpm for 20 h. The shaking speed employed for broth culture was the same throughout this study. The culture was serially diluted and spread on LB agar plates (Becton, Dickinson and Company) containing 12.5, 25, 50, or 100 mg/L CTX (Sigma-Aldrich Co. LLC., St. Louis, MO, USA). In addition, the same culture was spread on LB agar plates without antimicrobials. The frequency of the mutations was estimated from the number of colonies on LB agar plates with CTX divided by that on plates without CTX. Twenty colonies were picked from the CTX-containing plates and inoculated into fresh LB broth with CTX (at the same concentration to that used in the initial selection) and incubated at 37°C for 24 h under constant shaking. Each culture was mixed with same volume of 50% (vol/vol) glycerol and stored at -80°C until further use.

MIC was determined by the agar dilution method using Mueller–Hinton agar (Becton, Dickenson and Company), according to the procedure recommended by the Clinical and Laboratory Standards Institute (2008). The following internal quality control strains were used: Staphylococcus aureus ATCC 29213, Enterococcus faecalis ATCC 29212, E. coli ATCC 25922, and Pseudomonas aeruginosa ATCC 27853. The antimicrobials tested were as follows: CTX, ceftriaxone (CRO), ceftazidime (CAZ), cefoxitin (FOX), streptomycin (STR) (Sigma-Aldrich Co. LLC.), chloramphenicol (CHL) (Nacalai Tesque, Inc., Kyoto, Japan), and oxytetracycline (OTC) (Wako Pure Chemical Industries, Ltd., Osaka, Japan).

Genomic DNA was extracted from 1 ml of 24-h culture using Wizard Genomic DNA Purification Kit (Promega Corporation, Madison, WI, USA), and used for quantitative real-time PCR (qPCR), PCR, and short-read DNA sequencing.

Primers used in this study for qPCR are listed in Table S1. Oligonucleotides were purchased from Hokkaido System Science Co., Ltd. (Sapporo, Japan). The qPCR technique was used for the quantification of relative copy number of bla_CMY−2. Each reaction mixture contained 1X THUNDERBIRD SYBR qPCR Mix (Toyobo Co., Ltd., Osaka, Japan), 0.1X ROX (Toyobo Co., Ltd.), and 0.15 μM of each primer. The thermal conditions were initial denaturation at 95°C for 1 min, followed by 30 cycles of 95°C for 15 s and 60°C for 40 s. The signal was monitored using ABI Prism 7500 Real-time PCR System (Life technologies, Carlsbad, CA, USA). Standard curve method was used for relative quantification of target genes (Peirson et al., 2003). The rpsJ and dnaN genes were used as a single copy control in all experiments. The copy number was generated from a ratio of the relative amount of bla_CMY−2 to that of a single copy gene. DNA of the parent strain, L-3553, was used to normalize the bla_CMY−2 copy number of the mutants in every individual run.

One hundred microliter of the frozen stock of each strain was inoculated into 100 ml of LB broth containing CTX and incubated at 37°C under constant shaking. The optical density at 600 nm (OD₆₀₀) of each culture was measured and a portion of the culture was collected for DNA and RNA extraction in early logarithmic phase (OD₆₀₀ = 0.5), middle logarithmic phase (OD₆₀₀ = 1.5), late logarithmic phase (OD₆₀₀ = 2.5), early stationary phase (OD₆₀₀ = 3.0), and late stationary phase (24 h). The genomic DNA was extracted as described above and total RNA was extracted using RNAprotect Bacteria Reagent and RNeasy Mini Kit (Qiagen, Venlo, Netherlands). mRNA was converted to cDNA using the PrimeScript RT reagent Kit (Takara Bio Inc., Shiga, Japan) with random hexamers provided with the kit. qPCR was performed as described earlier to determine the relative expression level of mRNA for bla_CMY−2. The parent strain, L-3553, was used to normalize the mRNA fold change. Because our preliminary data suggested that the nrfG expression was stable throughout different growth stages (data not shown), it was used as an internal control for mRNA quantification.

Primers used in this study for PCR are listed in Table S1. PCR was performed using an iCycler apparatus (Bio-Rad Laboratories, Hercules, CA, USA), and Takara Ex Taq (Takara Bio Inc.) was used as the DNA polymerase for each PCR. The amplified fragments were purified using the ExoSAP-IT (USB Corporation, Cleveland, OH, USA), and the nucleotide sequence of both strands was determined using an Applied Biosystems 3130xl Genetic Analyzer with the BigDye Terminator cycle sequencing kit (version 3.1; Applied Biosystems, Foster City, CA, USA). The sequences were then assembled with Sequencher version 4 (Hitachi Solutions, Kanagawa, Japan) and the DNA alignment was examined using the Basic Local Alignment Search Tool (http://blast.ncbi.nlm.nih.gov/Blast.cgi).

DNA libraries (insert size: ~600 bp) were prepared using the Nextera XT DNA Sample Prep Kit (Illumina, San Diego, WI). DNA cluster generation and paired-end sequencing run for 170 mers were performed using an Illumina MiSeq with the MiSeq 500 cycle Kit v2. The fluorescent images were analyzed using the Illumina RTA1.17.22/MCS2.1.43 base-calling pipeline to obtain FASTQ-formatted sequence data. To generate short-read mapping data of all strains compared to the chromosome and plasmid, pST3553 of S. Typhimurium L-3553 [DNA Data Bank of Japan (DDBJ) accession numbers AP014565 and AP014566, respectively], BWA-SW (Li and Durbin, 2010), and SAMtools (Li et al., 2009) software were used with the default parameters. The mapping data obtained were visualized with GenomeJack viewer software (Mitsubishi Space Software, Tokyo, Japan), and the mean coverage, which means average value of coverage of short reads on a specific genomic region, was calculated using R software version 3.0.0 (R Core Team, 2013).

To confirm the chromosomal location of bla_CMY−2 and relevant genes among the strains L-3553 and its derivatives, pulsed-field gel electrophoresis (PFGE) followed by Southern hybridization analysis were performed according to methods described previously (Shahada et al., 2011), with some modifications. In brief, a plug of each strain was prepared using 24-h cultures. The plugs were digested with 60 U/ml of XbaI (Takara Bio Inc.) or 40 U/ml of FseI (New England BioLabs, Ipswich, MA, USA) at 37°C for 6 h, or 2 U/ml of S1 nuclease (Takara Bio Inc.) at 37°C for 45 min. Primers used for probe synthesis were listed in Table S1.

Wilcoxon rank sum test was used to compare the bla_CMY−2 copy number between increased CTX resistance mutants selected from the 12.5 and 25 mg/L CTX-containing plates. R²-value between the copy number of the genomic DNA and the mRNA fold change in bla_CMY−2 in all strains tested was calculated. These analyses were performed using R software version 3.0.0.

The sequence reads obtained from the S. Typhimurium strains were deposited in DDBJ Sequence Read Archive under the accession number DRA001681.

Spontaneous mutants were selected on the plates containing 12.5 and 25 mg/L CTX with a frequency of 1.1 × 10⁻⁶± 6.0 × 10⁻⁷ and 3.4 × 10⁻⁸ ± 1.6 × 10⁻⁸ (mean ± SD), respectively. No colonies appeared on the plates containing 50 and 100 mg/L CTX, suggesting that the frequency was less than 1.2 × 10⁻¹⁰. Twenty colonies were picked from each of the plates containing 12.5 and 25 mg/L CTX, and the copy number of bla_CMY−2 was quantified. In the preliminary experiments, we determined the bla_CMY−2 copy number of selected strains for three times and calculated the coefficient of variation (CV = standard deviation divided by mean). CV for the biological replicate was 0.153, whereas that for the technical replicate was 0.102. As shown in Table 1, all the mutants harbored multiple bla_CMY−2 copies, ranging from 5.9 to 85.3, while the average bla_CMY−2 copy number in 10 colonies of L-3553 on the LB agar plate was 1.01 ± 0.06. Higher variance in copy numbers was observed in mutants selected at 25 mg/L CTX (9.0–85.3) than in mutants selected at 12.5 mg/L CTX (5.9–25.7). Moreover, the average copy number in the mutants selected from plates containing 25 mg/L CTX (24.2 ± 20.6) was higher than that from plates containing 12.5 mg/L CTX (17.1 ± 5.1), although the difference was not statistically significant (p = 0.62).

To confirm the phenotypic changes of the mutants, we determined the MICs of six additional antimicrobials whose resistance genes were located in GI-VII-6: resistance to CRO, CAZ, and FOX is conferred by bla_CMY−2, CHL resistance is conferred by floR, STR resistance is conferred by aadA2, strA, and strB, and OTC resistance is conferred by tetA (Figure 1). As shown in Table 1, the MICs of CTX, CRO, and FOX were found to increase by four- to 32-fold. The MIC of CAZ against some of the mutants was over the detection limit. MIC₅₀ and MIC₉₀ of cephalosporin antimicrobials against the mutants selected from plates containing 25 mg/L CTX were higher than those from plates containing 12.5 mg/L CTX (Table S2). The MIC of chloramphenicol against the mutants increased four-fold, except for two strains showing the same value as that of the parent strain, whereas MIC of streptomycin against the mutants increased more than two-fold, but the values were over the detection limit in half of the strains. MIC of oxytetracycline against the mutants was comparable to that of the parent strain.

bla_CMY−2 copy number in the genomic DNA and fold change of mRNA were measured to confirm the correlation between them. As shown in Figure 2, the copy number of bla_CMY−2 increased rapidly in the early logarithmic phase, and then gradually increased from late logarithmic to stationary phase in strains 12-1, 12-19, and 25-11. In contrast, the copy number increased linearly even in the stationary phase in strains 12-14, 25-6, and 25-17, leading to higher copy numbers. Furthermore, the fold changes of mRNA were significantly correlated with the bla_CMY−2 copy numbers (R² = 0.97, p < 10⁻⁵).

To check the stability of the amplified regions in the absence of selection pressure, the six selected mutants were serially subcultured in LB broth without CTX for ten times (100 generations). As shown in Figure 3, the bla_CMY−2 copy numbers drastically decreased in all isolates. Strain 25-6 showed approximately 80% decrease in the copy number in the first subculture. Subsequently, the copy number continuously decreased during the serial subculture, and finally reached a value of 3.0. The copy numbers of the other five strains ranged from 0.8 to 2.9 at 100 generations.

In the six selected mutants, no mutations were found in bla_CMY−2, or its promoter region. Furthermore, no point mutations were observed among the mutants' whole-genome sequences, except for three non-synonymous mutations listed in Table S4. We observed a 51-kb deletion [nucleotide (nt) position 1763615–1814713] of chromosomal DNA in strain 25-17, although the XbaI and FseI restriction sites were not affected by these mutations.

The parent strain, L-3553, harbors a 133-kb plasmid, pST3553, which is a derivative of S. Typhimurium specific virulence plasmid and does not contain bla_CMY−2 gene (Sekizuka et al., 2014). In some of the mutants, one or multiple plasmids with sizes larger than the original were observed. Estimation by S1 nuclease-PFGE indicated that the size of the plasmids varied from 255 to 660 kb (Table 1). Southern hybridization analysis revealed that bla_CMY−2 and the virulence plasmid specific gene, spvB, were both present in the larger plasmids. In all strains tested, bla_CMY−2 signal was also present in the largest band, which is originated from the chromosome (Figure 4A). spvB signals in the largest band seemed a non-specific one, because we observed the signal even in the parent strain L-3553.

To demonstrate the amplification of bla_CMY−2 at the original chromosomal position, we performed additional PFGE–Southern hybridization analyses with a bla_CMY−2 probe using FseI-digested genomic DNA of the six selected mutants. As GI-VII-6 does not contain an FseI site, it was located in its entirety in a 350-kb band in L-3553 (Figures 4B,D). This band was not present in the genomic DNA extracted from the mutants. A bla_CMY−2 probe was found to hybridize with the largest band originating from the chromosome in the selected mutants. Additional bla_CMY−2 signals originating from enlarged plasmids were observed in strains 12-14, 25-6, and 25-17. As the endogenous pST3553 plasmid contains one FseI site, the sizes of the bla_CMY−2-positive bands corresponded to those of the enlarged plasmids.

Because GI-VII-6 contains one XbaI site (nt 1037120–1037125) (Shahada et al., 2011), tandemly arrayed GI-VII-6 sequences should generate multiple copies of a specific sized fragment after XbaI digestion (Figure 4D). As shown in Figure 4C, we observed several bands of different sizes, which were not found in the parent strain, after separation of XbaI-digested genomic DNA of the selected mutants by PFGE. Most of them were approximately 30–125 kb in size and showed higher fluorescence intensity compared to the other bands. In subsequent Southern hybridization analysis, bla_CMY−2 signal was observed in these bands. The sizes of the bands that showed the prominent bla_CMY−2 signal were approximately 100, 125, 100, 30, 80, and 110 kb in strains 12-1, 12-14, 12-19, 25-6, 25-11, and 25-17, respectively. In strain 25-6, additional weaker signals were observed in the bands of approximately 40 and 85 kb in size. An additional weaker signal was also observed in a 180-kb band in strain 25-17.

To estimate the plasmid copy number in the parent and each mutant, the mean coverage of the pST3553 plasmid was divided by that of the chromosome. The plasmid copy numbers of strains L-3553, 12-1, 12-14, 12-19, 25-6, 25-11, and 25-17 were estimated as 1.6, 2.3, 5.2, 2.7, 4.7, 1.9, and 7.1, respectively. Three strains with a bla_CMY−2 copy number greater than 25 (12-14, 25-6, and 25-17) showed higher plasmid copy numbers.

The results of the short-read mapping showed that the mean coverage of the chromosomal genes (background) in the six selected mutants ranged from 20.7 to 61.1 (Table S3). In this study, regions showing a mean coverage that is greater than five times higher than the background were regarded as amplified regions. As shown in Figure 5, coverage of the entire or a part of GI-VII-6 was higher than other regions in the six selected mutants. The amplified regions of strains 12-1, 12-14, and 12-19 were flanked by directly repeated copies of IS26. The end points of the amplified regions in strains 12-1 and 12-19 were the first and third copies of IS26 (nt positions 983123–983942 and 1082604–1083423, respectively) from the 5′-end of GI-VII-6. The size of the amplified regions was 100 kb. In strain 12-14, the end point of the amplified region was the first and fourth (nt position 1107425–1108244) copies of IS26.

In contrast, we could not identify directly repeated sequences at the ends of the amplified regions in strains 25-6, 25-11, and 25-17. The amplified regions in strain 25-6 started from an intergenic region between STL3553_c09800 and STL3553_c09810 (nt position 1021052) and terminated at an intergenic region between STL3553_c10000 and STL3553_c10010 (nt position 1048473). The mean coverage of this amplified region was 1296.9. Furthermore, in this strain, the mean coverage from downstream of the amplified region to nt position 1063378, located in the middle of gene STL3553_c10130, was 231.1, which subsequently decreased to 102.5 until just before the fourth copy of IS26. These data suggest the existence of three different amplified regions [nt positions 1021052–1048473 (27 kb), 1021052–1063378 (42 kb), and 1021052–1108244 (87 kb)] sharing a portion of GI-VII-6 sequence. In strain 25-11, the 5′-end of the amplified region started from 21 kb upstream of the 5′-end of GI-VII-6 (nt position 962959) and terminated at the middle of ISEcp1 (nt position 1042417) within GI-VII-6, which accounted for a size of 80 kb. The coverage dropped sharply in the middle of ISEcp1, while it was still higher by several times than the background until the 5′-end of the ISEcp1 inverted repeat (nt position 1043519) (Figure S1). In strain 25-17, high mean coverage of 1269.7 was observed from an intergenic region between STL3553_c09480 and STL3553_c09490 (nt position 984235) to the 3′-end of IS1294 (nt position 1096148), which accounted for a size of 112 kb. (Figure S1). The sizes of these amplified regions in each mutant corresponded to those of the bands that showed the prominent or weaker bla_CMY−2 signals (Figure 4C). Among the bla_CMY−2 signals observed in Figure 4C, the origin of the 180-kb signal was not identified in this study.

To determine the DNA sequence of the junction of the amplified region, we performed DNA sequencing of the PCR products. Location of each primer to amplify the junction region of strains 12-1, 12-14, 12-19, 25-6, and 25-11 is indicated in Figure S2. In strains 12-1, 12-14, and 12-19, the location of IS26 at the junction was confirmed. In addition, we confirmed the location of IS26 at the junctions of three different amplified regions in strain 25-6. The location of IS1 at the junction of amplified region was also confirmed in strain 25-11 (see Figure S3). We could not successfully amplify the junction region in strain 25-17.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.