Streptococcus pneumoniae and Streptococcus pseudopneumoniae strains were grown as static cultures either in a casein hydrolysate based medium (AGCH) supplemented with 0.3% sucrose and 0.2% yeast extract (A+SY) (Lacks et al., 1986) or in Tod-Hewitt (Becton Dickinson) with 0.5% of yeast extract (THY). MICs of ethidium bromide, acriflavine and FQs were determined by agar-dilution (for S. pseudopneumoniae 5305) or by macrodilution following the methods of the CLSI (Clinical Laboratory Standards Institute, 2008) using 2-fold dilutions in A+SY liquid medium, with an inoculum of 5 × 10⁵ cells per well. The efflux inhibitor reserpine (Gill et al., 1999; Garvey et al., 2011) was used at a final concentration of 10 μg/ml, which has been shown to inhibit ethidium bromide, acriflavine, and CPX efflux in streptococci of the mitis group (Ferrándiz et al., 1999). S. pneumoniae R6 was used as the recipient in transformation experiments. Competent cultures containing 9 × 10⁶ CFU/ ml were incubated with 0.1 μg/ ml of DNA during 40 min at 30°C and then at 37°C for 90 min before plating. Colonies were counted after 24 h of growth at 37°C in a 5% CO₂ atmosphere in A+SY media plates containing 1% agar and CPX at 2 μg/ ml. No transformation refers to experiments in which no colonies were observed when 100 μl of the transformation mixture was plated, which gave a frequency <1 × 10¹. DNA used as control was the chromosome of T5305 that yielded 1.1 × 10³ transformants/ml.

Two S. pseudopneumoniae clinical isolates, 5305 and CipR71, and two S. pneumoniae strains, an R6 transformant with 5305 chromosomal DNA, named S. pneumoniae T5305 and our laboratory S. pneumoniae R6 strain, were subjected to whole genome sequencing. Genomic DNA samples were prepared from mid-log phase cultures using the CTAB protocol (Wilson, 1994). Sequencing was performed either in a 454 Life Sciences GS-FLX+ system, Roche (5305 and CipR71) or in an Illumina NextSeq 500 system (T5305 and our laboratory R6). DNA for 454/Roche library preparation was fragmented by nebulization and posterior quantification, emulsion PCR and sequencing performed as recommended by the manufacturer. Libraries for Illumina sequencing were obtained following the transposon-mediated Nextera XT library preparation kit (Illumina) and paired-end sequenced with a 2 × 150 protocol. De novo or reference aided (S. pneumoniae R6) assemblies were performed using Newbler 3.0 (Roche) with default settings. sff files from 454/Roche and fastq files from Illumina were used as input sequences, respectively. Assembly metrics are shown in Table 1. Coverage was calculated by dividing total bases used in the computation by the size of the final alignment following the Lander/Waterman equation (Lander and Waterman, 1988).

Coding sequences were predicted by Prodigal V.2.6.3 with default settings (Hyatt et al., 2010). S. pneumoniae pump gene sequences (Tocci et al., 2013) were detected in S. pseudopneumoniae genomes by BLASTN using S. pneumoniae TIGR4 sequences as queries and applying bidirectional thresholds of ≥80% identity and an alignment length over ≥80% of the protein sequences. Promoter-containing sequences of S. pseudopneumoniae isolates were extracted as the 100 upstream nucleotides of the pump gene, or, in the case of genes in operons, the 100 upstream nucleotides of the first gene of the operon. The RNA stem-loop structure was predicted by using the RNA-fold web server (http://rna.tbi.univie.ac.at/cgi-bin/RNAWebSuite/RNAfold.cgi). Gene organization of operons was downloaded from the DOOR database (Mao et al., 2009), using the S. pseudoneumoniae IS7493 strain as reference. Protein and nucleotide sequences were aligned with Clustal Omega (Sievers et al., 2011).

The sequencing data have been deposited at the whole genome sequencing database linked to the BioProject ID PRJNA386436.

Primers used for PCR amplifications and sequencing are shown in Table 2. Platinum® Taq High Fidelity (Invitrogen), was used for PCR amplifications in the following conditions: an initial cycle of 2 min denaturation at 94°C; 30 cycles of denaturation at 94°C for 30 s, annealing at 55°C for 30 s and extension at 68°C for 1 min per kb of PCR product. Primers used for PCR amplifications and sequencing are shown in Table 2. To quantify the relative expression of hexA, patA, and patB, a procedure previously described (El Garch et al., 2010) was used. Briefly, R6-derivative strains were grown overnight at 37°C in 5% CO₂ in Mueller-Hinton agar plates supplemented with 5% defibrinated sheep blood (Becton Dickinson). Bacteria were suspended in 5 ml of THY, grown to OD_620nm = 0.2, diluted 10-fold into 25 ml of THY and grown to OD_620nm = 0.2 to 0.4 (2–4 × 10⁸ cells). At this point, cultures were split into two parts; one part was grown during 4 h in the presence of CPX at 0.5 × MIC for each strain, and the other was grown in its absence. Cells were collected by centrifugation and their total RNA was extracted with an RNeasy mini kit (Qiagen), with 3 DNase I (Qiagen) treatments, according to the manufacturer's instructions. Synthesis of cDNAs were carried out in 20-μl reactions containing 0.5–5 μg of RNA, 0.5 mM of each deoxynucleoside triphosphate, 2 pmol of hexanucleotide random primers, 40 U of the RNaseOUT RNase inhibitor (Invitrogen), 200 U of SuperScript IV reverse transcriptase (Invitrogen), and 2 U of RNase H⁻ (Invitrogen), in the buffer recommended by the manufacturer. These cDNAs were used in RT-qPCR amplifications (Light cycler 480, Roche). The total volume in each reaction was 20 μl and contained 2 μl of cDNA, 0.4 μM of each primer, and 10 μl of Sso Advanced Universal SYB Green Supermix (BioRad). The program had one initial denaturation cycle of 4 min at 94°C followed by 42 cycles of three steps: denaturation (30 s at 94°C), annealing (30 s at 54°C), and elongation (30 s at 68°C). Oligonucleotide pairs used were: PatARTF/PatARTR (patA); PatBRTF/ PatBRTR (patB), HexARTF/HexARTR (hexA). The amount the RT-PCR products was calculated with a standard curve for each amplicon. Assays were performed in triplicate and normalized against the signal from a 16S rDNA internal control (oligonucleotide pair rDNAF/rDNAR).

5′ RACE (rapid amplification of cDNA ends) method was performed using total RNA extracted from strain TCipR71.2 treated with CIP at 0.5 × MIC as template. A total of 12 μg of RNA were used, 6 μg were treated with Tobacco acid pyrophosphatase (TAP) to hydrolyze specifically 5′-triphosphate from primary transcripts, and to subsequently be ligated to the 3′-hydroxyl group of an RNA oligonucleotide (RNA adaptor, Table 2). The remaining 6 μg of total RNA were not treated with TAP and followed the same post-treatment. cDNAs were synthesized from these RNAs using SuperScript IV Reverse Transcriptase (Invitrogen) following manufacturer recommendations and used as templates in PCR reactions with primers PatARACE1 and RACE5′ (Table 2) as detailed in PCRs, RNA Extraction and Quantitative Real Time PCR (RT-qPCR) Procedures. PCR products were purified with MinElute PCR purification kit (Quiagen) and used as templates for a second PCR reaction using primers PatARACE2 and RACE5′Inner (Table 2). PCR products were separated in a 2% Metaphor (FMC Bioproducts) agarose gel and the band that appeared in the sample treated with TAP, and did not appear in the untreated sample, was isolated from the gel with MinElute Gel Extraction Kit (Qiagen). DNA was sequenced using primer PatARACE2 to determine the transcription start site, which was the first base 5′ immediately after the adaptor sequence.

The MICs of isolates 5305 and CipR71 to FQs and some dyes in the presence and absence of the efflux pump inhibitor reserpine were compared to three efflux-negative control strains: S. pneumoniae R6, S. pseudopneumoniae CCUG48465, and the S. pseudopneumoniae type strain CCUG49455^T. No significant differences in the MICs of LVX, CPX, ethidium bromide and acriflavine in the presence or absence of reserpine were observed in the control strains. However, the two S. pseudopneumoniae clinical isolates, which showed 8- to 32-fold higher MICs of these drugs in the absence of reserpine than the control strains, essentially behave like the control strains in the presence of the pump inhibitor (Table 3). Reserpine it is assumed not to affect growth of the strains at the concentration used (10 μg/ ml), since the MICs of reserpine for S. pneumoniae have been described to be 60 μg/ml (Garvey and Piddock, 2008). These results suggested the existence of an efflux pump in S. pseudopneumoniae isolates 5305 and CipR71.

We analyzed the genome sequences of the 5305 and CipR71 resistant strains and compared them with those of S. pneumoniae and S. pseudopneumoniae isolates, paying special attention to those coding for the 11 pneumococcal antimicrobial pumps (Tocci et al., 2013). S. pseudopneumoniae isolates carried nine of these pumps, coded by 14 genes, given that they are constituted by complexes in some cases. Amino acid sequences of the permease subunits and the sequences covering the 100 nucleotides upstream of the coding sequence (or of the first coding sequence of the operon) were compared (Table 4). Subtle differences were detected in the sequences of these pumps. However, simultaneous alterations, i.e., alterations common to both S. pseudopneumoniae isolates with an efflux phenotype (CipR71 and 5305) respect to the efflux-negative S. pseudopneumoniae type strain ATCC BAA-960, were found only in the coding sequences of five of these genes (affecting four pumps), which essentially ruled out the possible involvement of the nine remaining genes in the efflux phenotype. The most remarkable efflux-linked determinants were the coding sequence of the ABC transporter SP_1357 (11 residue changes) and the regulatory sequence of the patAB operon. Both proteins are part of the Drug Exporter-4 family of the Transport Classification Database (TCDB id: 3.A.1.135). The enhanced CPX efflux of isolates 5305 and CipR71 could be attributed either to higher substrate affinity/enhanced activity produced by the residue changes in SP_1357, by an up-regulation of the PatAB pump, or by a combination of both. To discern between these possibilities, genetic transformation experiments were carried out as described below.

Streptococcus pneumoniae R6 was transformed with chromosomal DNA of S. pseudopneumoniae 5305. One transformant, T5305, was selected. T5305 had a CPX MIC of 4 μg/ml, this MIC was equivalent to that of S. pseudopneumoniae 5305. In addition, 5305 and T5305 exhibited the same efflux phenotype to CPX, ethidium bromide and acriflavine (Table 3). The analysis of the S. pneumoniae T5305 genomic sequence showed 420 nucleotide changes with respect to the sequence of S. pneumoniae R6 present in the NCBI data base (Accession number: NC_003098.1). To discard spurious mutations, our laboratory R6 strain was also sequenced, showing 21 SNPs with respect to the NCBI sequence (data not shown). These changes were not considered, then, a total of 399 changes were attributed to the recombination event. These changes were unevenly distributed, 353 of them were located in a 7184 bp region (4.8% of variation) covering the spr1884 to spr1888 genes (guaA-patB-patA-hexA); 45 in a 3540 bp region (1.2% of variation) covering the spr0162 to spr0166 genes (ribA-ribC-ruvA-tag); and 1 point mutation in spr1903 (galU). These results demonstrate that the recombination event yielding T5305 occurred in two regions: spr0162-0166 and spr1884-1888 (Figure 1). This was very likely due to its close spatial proximity considering both recombined regions are at the same distance to the chromosome replication origin, a factor that may increment inter-replicore gene transference.

To reveal the role of the nucleotide changes present in T5305 in FQ-efflux, diverse DNA fragments were amplified from this strain by PCR with specific oligonucleotides (Table 2). These fragments were used as donor DNAs in transformation experiments with R6 as receptor, and transformants selected at 2 μg/ml of CPX. Transformation experiments were performed as described under Materials and Methods, being the frequencies of transformation in the range of 1 to 7 × 10³ transformants/ ml. DNA fragments covering the spr0162-0166 genes or the galU mutation did not result in transformants at elevated CPX concentrations, discarding the role of these genes in CPX-resistance. In contrast, two overlapping PCR fragments from the spr1884-1888 region transformed R6 with high efficiency (Figure 1). One of the amplicons covers the C-terminus of hexA and the intergenic hexA-patA region plus patA, whereas the other covers the same regions but including patB. In accordance, PCR products including most of the hexA and patA sequences were also able to transform to CPX resistance. The smallest fragment able to transform R6 to CPX resistance was obtained by amplification with oligonucleotides HexAF757 and PatAR17. Likewise, PCR products obtained from isolate CipR71 with the same oligonucleotide pairs also transformed R6 with high efficiency. In contrast, DNA fragments containing hexA, patA, patB, or guaA coding sequences, but not the hexA-patA intergenic region, were unable to transform R6 to CPX-resistance. In addition, a fragment covering hexA from residue position 22 to the stop codon, which was obtained from S. pseudopneumoniae strains 5305 and CipR71, did not transform to CPX-resistance (Figure 1). Three resistant transformants, two obtained with PCR products from 5305 (T5305.1 and T5305.2) and one from CipR71 (TCipR71.2) were selected for further studies. Their nucleotide sequences were identical to that of the donor 5305 and CipR71 strains and their MICs were equivalent to that of their parental DNA donors (Table 3). All these results demonstrated that the region involved in FQ-efflux in the S. pseudopneumoniae strains T5305 and CipR71 was that included between oligonucleotides HexARDown and PatAR17 (Figure 2A). This region contains the intergenic hexA-patA region and the first 74 nucleotides of patA. No amino acid changes were predicted in PatA between R6, 5305 and CipR71 or the derived R6-transformants. The intergenic hexA- patA region has been previously stablished to be involved in FQ-efflux in SPN (Baylay and Piddock, 2015). A promoter with a canonical −35 box and an extended −10 sequence was observed in all strains. We determined the transcription start site by RACE 5′ PCR as described in materials and methods, which was located eight nucleotides downstream of extended −10 sequence (Figure 2B). The main differences were found between the extended −10 sequence and the patA initiation codon, in which the two S. pseudopneumoniae FQ-resistant strains CipR71 and 5305 showed a 3-nt insertion (GCT) or a 3-nt deletion (GGA), respectively. A stem-loop structure could be predicted by RNA-fold in this region with a free energy of −15.32 kcal/mol for R6 (Figure 2C). The nucleotide changes observed in CipR71 and 5305 CPX-resistant isolates are predicted to lower the free energies of these structures to −8.58 kcal/mol for 5305 and −13.82 kcal/mol for CipR71. These changes in the stability of the stem-loop structures could cause alterations in the transcription of the downstream patA and/ or patB genes.

To quantify the induction produced by CPX, RT-qPCR experiments were performed with RNAs of cultures treated or not treated with a subinhibitory CPX concentration. These assays showed that the expression of both patA and patB in the R6-derivative strains was induced by CPX. While no significant changes were observed in R6, increases in patA and patB were detected in all strains with an efflux phenotype. These increases were in the range of 1.4 to 3.4-fold for patA, and of 2.1 to 2.9-fold for patB. However, the transcription of hexA did not show significant differences between R6 and its transformant-derived strains, being in the range of 0.7 to 1.3-fold (Figure 3).

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.