A murine S. aureus nasopharyngeal colonization model was established by inoculating groups of mice intranasally with ~ 5 x 10⁴ CFU S. aureus USA300 LAC JE2 strain. The mice were maintained for a period of 14 days post-inoculation. At chosen time points (1, 3-, 5-, 7- and 14-days post-inoculation), the nasopharyngeal CFU load in each mouse was determined ( Figure 1 ).

Across the 14 days of nasopharyngeal colonization, numbers of viable S. aureus CFU decreased by ~1 log by day 3 compared to day 1 (P<0.01), followed by an increase in CFU from day 5 onwards, with the population remaining constant thereafter until day 14 when the experiment was ended ( Figure 1 ). The model was therefore determined to be sufficiently robust to design an experimental evolution using serial passages.

Based upon the established colonization model, 3 serial passages of S. aureus were performed in total with each passage lasting for 7 days ( Figure 2 ). Isolated bacteria from the first passage were used as inoculum for the second passage. The process was repeated using the bacteria from the second passage as inoculum for the third passage. Mice were sacrificed to recover the colonizing S. aureus after each passage with no deaths recorded otherwise. After completion of the serial passages, spa typing of the bacteria was performed by DNA sequencing of the amplified gene to confirm nasopharyngeal isolates matched the input strain.

Whole genome sequencing of 10 pooled isolates from each mouse in the three passages was used to determine whether repeated S. aureus colonization of the nasopharynx selected for genetic changes. The isolate inputs for pooled for DNA sequencing are shown in Table 1 .

After the first passage, there was determined to be limited intragenic and intergenic sequence variation ( Figure 3 ) based upon a threshold imposed of variants being represented at >20% i.e., in at least two isolates. Seven single nucleotide polymorphisms (SNPs) in total were identified from the 3 mice, with only one SNP being observed in more than one pool of isolates ( S1 and S2 Tables ). Identified SNPs showed low frequency of around 20% of total reads, indicating that at least two of the ten isolates forming the pooled genomic DNA carried the identified mutation. One sequenced isolate DNA pool (mouse B1) did not bear any SNPs. Overall, the low abundance and frequency of SNPs suggests there was not a pronounced or uniform selection on S. aureus during the first seven days of nasopharyngeal colonization as evidenced by the lack of a dominant variant causing population change.

DNA sequencing of isolates after the second nasopharyngeal passage revealed a greater number of variants in the S. aureus clones, with 104 SNPs found across the three genomic DNA pools ( Figure 3 ) based on the same >20% threshold imposed in the first round of sequence analysis. From these variants, 43 out of 46 non-synonymous intragenic SNPs and 29 out of 30 intergenic SNPs were observed in all three genomic pools ( S3 and S4 Tables ). A further difference between the profiles of sequence variants from the first and second passages is the higher proportion of reads for each variant in the second round. The DNA sequence pools from the second passage (B2 and C2) ( Figure 2 ; S3 Table ) revealed 37 intragenic SNPs that were present in at least 90% of the reads, (i.e., in 9/10 of the recovered S. aureus isolates). The pool of isolates from A2 had a lower proportion of reads for sequence variants, where most of the SNPs (n=36; 78.2%) were detected in a lower proportion (60%) of the reads, though still substantially higher than the variant frequencies of the first passage. The increased number of sequence variants identified after the second passage was suggestive of selection S. aureus during nasopharynx colonization leading to a third passage.

A reduced range of sequence variants was identified across the isolate pools from the three mice after the third 7-day colonization of the nasopharynx passage. A total of 36 SNPs were detected from the genome DNA pools ( Figure 3 ). Above the threshold, sequence variants were present at ~90% of reads shared across the three sequence pools, with the exception of those detected in the hypothetical protein SAUSA300_RS09205. Among the 20 non-synonymous SNPs detected in the sequence pools of the third passage, 16 persisted from the first and/or second passage. In addition, three out of six intergenic SNPs also persisted across nasopharyngeal passages ( Table 2 ). Culture of isolates on agar was minimized though the possibility of additional selection between nasal passages cannot be discounted.

The functional category of genes containing sequence variants after each of the three 7-day nasopharyngeal colonization was identified. The variants included non-synonymous intragenic SNPs, and intergenic sequence variants determined to have the potential to alter transcription of downstream genes. The majority of the functional categories belong to genes encoding proteins involved in cell metabolism (amino acid and co-factors biosynthesis; energy and fatty acid metabolism) or DNA metabolism (replication, recombination and repair) ( Figure 4 ).

Proteins in pathways of DNA replication, recombination and repair with intergenic SNPs included: topoisomerases (GyrA, GyrB, GyrC); translocases (FtsK); nuclease (SbcC), and the DNA replication and repair protein RecF. The relative proportion of genes associated with these DNA pathways harboring SNPs could reflect selection for genome adaptation. Increased mutability by S. aureus was reported as an adaptation mechanism (Baharoglu and Mazel, 2014), whereby weakened DNA replication control and repair promotes allelic variability, benefiting S. aureus in its niche.

Variants associated with transcriptional regulation were observed for genes encoding proteins involved in transcription (RpoB), RNA modification (Sun, SAUSA300_RS04890 and SAUSA300_RS09885) and RNA decay (SAUSA300_RS06375) ( S3 and S5 Tables ). These genes play a role in cellular pathways that modulate ribosomal stability and the synthesis of proteins (Decatur and Fournier, 2002) as part of mechanisms the bacteria use to changes in the microenvironment (Khemici et al., 2015).

Sequence variants were identified in genes encoding metabolism proteins including mannitol (MtlA) and sorbitol fermentation (GutB), fatty acid metabolism (FadB; SAUSA300_RS01200 and SAUSA300_RS07925) and peptide metabolism (AmpA aminopeptidase). Further sequence variants were identified in genes encoding proteins associated with the biosynthesis of co-factors biotin (BioA) and thiamine (SAUSA300_RS11285), haem (SAUSA300_RS08410), and amino acids valine (ValS) and histidine (HisG) following S. aureus nasopharyngeal colonization. Genes encoding proteins within arginine metabolism revealed variants: ArcA and ArgF. Both enzymes catalyze formation of citrulline, the former as part of the arginine deiminase pathway that serves as energy source for S. aureus in anaerobic conditions, and the latter within arginine synthesis from glutamate (Diep et al., 2006; Pi et al., 2009). Pathways for responses to a variety of stresses revealed sequence variants within genes encoding ClpL (general stress tolerance), AhpF (oxidative stress) and DnaJ (hyperosmotic and heat shock stresses) together with variants observed in genes associated with anaerobic metabolism (NarK) and lactate (LctP2).

Those sequence variants that were identified within the pathways described during nasopharyngeal colonization revealed a smaller subset that persisted across the successive rounds of colonization ( Table 2 ). One gene was identified as the target for multiple mutations across its length with varied individual frequencies of the variants from the three mice in the final passage. The gene (SAUSA300_RS09205) encodes a 399 amino acid hypothetical membrane protein described within the TIGRFAM: transport proteins and PFAM: metallopeptidase family M24. The collective sequence variants identified reveal this gene as showing selection most clearly across the sequential passages. Conversely, from the experimental evolution sequence data, there were sequence variants not maintained temporally, indicating potential refinement of selection during the repeated nasopharyngeal colonization passages of S. aureus. The vast majority of selected variants that persisted across repeated nasopharyngeal passages of S. aureus are genes involved in metabolism and DNA replication ( Table 2 ).

Experiments to perform the third nasopharyngeal passage repeatedly returned S. aureus with a different spa type from the USA300 LAC species inoculum indicating that sets of mice were colonized by S. aureus upon arrival. Recent studies reported there is frequent colonization of laboratory mice by S. aureus (Schulz et al., 2017) in contrast to earlier studies that S. aureus are not natural colonizers of mice (McCarthy et al., 2012; Mulcahy et al., 2012). Others have determined that colonization by S. aureus prevents the invasion of distinct S. aureus strains (Margolis et al., 2010). To gain knowledge of the resident strain in the purchased mice and enable a simple comparative interrogation, we performed whole-genome sequencing and variant analysis on the mouse resident S. aureus (named SA_MOU). We sought insights of genome sequence polymorphism that converged with the experimental evolution selection from repeated colonization of the nasopharynx.

Multilocus sequence typing technique (MLST) identified SA_MOU as ST8, the same sequence type of USA300. A recent study demonstrated that the genetic variation within ST8 clones is rather substantial temporally and globally (Strauß et al., 2017) and that ST8 possesses clonal heterogeneity in Europe, where it co-exists with the dominant ST80, and others such as ST22 (Otter and French, 2010). The sequenced and assembled SA_MOU genome was aligned with the USA300 reference genome plus representatives of dominant STs in Europe: ST22 and ST80. Alignment showed fewest gaps with the SA_MOU assembled contig against the USA300 reference genome, supported by a pairwise identify of 99.5% across 94.5% coverage as the highest values among the three genome alignments (data not shown). Alignment of SA_MOU against ST22 and ST80 genomes revealed high pairwise identify (97.9% and 99.4%, respectively) but with lower coverage (88.0% and 90.9%, respectively).

Because we identified that SA_MOU was resident in mouse nasopharynx and was the same ST as USA300 LAC JE2, we hypothesized there was concordance between sequence variants that occurred in passaged isolates and the corresponding SA_MOU allele. Matches could provide support for key loci undergoing in vivo selection. The genes containing 13 non-synonymous intragenic SNPs from the third experimental nasopharyngeal passage were compared with those of the SA_MOU annotated genome. Sequences for 9 of the 13 genes were identified in the SA_MOU genome ( Table 3 ) with four loci not located in the assembled genome. The cognate protein sequences detected in SA_MOU were compared with the translated USA300 reference genome (NC_007793.1) using BLASTp, which identified two proteins with matched sequence variation: RecF and FtsK.

The clinically-derived S. aureus strain USA300 LAC JE2 (Fey et al., 2013) was chosen as the strain to study in the model. This strain was used in the murine model described below which generated 90 nasopharyngeal colonization clones. Cultures were grown for 18 h at 37°C with shaking unless stated otherwise. Brain Heart Infusion (BHI) broth or agar was the medium used unless otherwise stated.

A nasopharyngeal model for S. aureus colonization was established to enable experimental evolution approaches. Mouse experiments were approved by the United Kingdom Home Office (Home Office Project License Number 40/3602) and the University of Liverpool Animal Welfare and Ethics Committee.

Groups of three mice (female BALB/c mice of 6-8-week-old, Charles River, UK) were assigned to a cage and the small group size was based on the subsequent genetic analysis. Mice were maintained with food and water ad libitum.

Animals were anaesthetized with a mixture of oxygen and isofluorane and 5 x 104 cfu of USA300 LAC JE2 in a volume of 15 μl were horizontally instilled into the nares. Inocula were adjusted to their cfu by dilution in phosphate-buffered saline (PBS) and confirmed by viable counts. Mice were monitored over 14 days post-inoculation and did not show any visible signs of disease. At five time points over a period of 14 days, mice were culled by cervical dislocation followed by aseptic removal of nasopharynx to determine CFU load. No CFUs were recovered from the lungs or blood over this period. Nasopharyngeal tissue was mixed with 3 ml of PBS and mechanically disrupted using an IKA T10 handled tissue homogenizer. Bacteria were enumerated from viable counts of serially diluted nasopharyngeal homogenates with growth and species confirmation on mannitol salt agar after incubation for up to 48 h at 37°C.

Staphylococcus aureus USA300 LAC JE2 was passaged three times by sequential colonization of the nasopharynx, with three mice per passage. Animal handling and bacterial inoculum preparation was performed as described above. An inoculum of USA300 LAC JE2 culture was prepared for the first nasopharyngeal passage and instilled in each of three mice. After the 7 d incubation period, mice were euthanized, each nasopharynx was aseptically removed and homogenized and the resulting tissue homogenates grown on separate mannitol salt agar plates. Ten clones were randomly selected from each replicate and subculture. Equal volumes (1 ml) of the ten cultures were mixed to generate a new inoculum followed by dilution in PBS to obtain a bacterial load of ~ 5 x 104 cfu. This inoculum was used for a second passage of murine nasopharyngeal colonization. A third and final serial passage was achieved by repeating this approach. Both control and resulting clones from each of the three nasopharyngeal passages were stored alongside nasal homogenates at -80°C with 15% glycerol (v/v).

DNA was purified from ten isolates randomly picked after plating nasal homogenates for each mouse from the three nasopharyngeal passages; purification methods were as described elsewhere (Coates-Brown et al., 2018) with the addition of an RNase step. The 9 genomic DNA pools for sequencing were prepared by mixing equimolar concentrations of DNA from which 9 libraries were prepared with an insert size of 350 bp using a TruSeq PCR-free sample prep kit (Illumina), according to the manufacturer’s instructions. DNA libraries were sequenced as barcoded pools on the Illumina MiSeq using V2 reagents generating paired sequencing reads of 350 bases in length. Reads were quality-filtered and trimmed for Nextera adaptor sequences using Trimmomatic (Bolger et al., 2014). Reads were aligned to parental S. aureus USA300 LAC JE2 that was sequenced in this study as a control. DNA library preparation and samples sequencing were done by the CGR (Centre for Genomic Research), University of Liverpool. Variant analysis was performed as described in Coates-Brown et al. (2018) with a threshold of 20% of reads.

S. aureus strain SA_MOU isolated from mouse nasopharynx in this study was assembled then visualized using Geneious software (version R11.1) where the resulting assembly was aligned by BLASTZ software (version 7.0.1) to the USA300 reference genome (USA300_FPR3757; NCBI Ref Seq: NC_007793.1) and two additional reference genomes representing dominant STs in Europe: ST22 (HO 5096 0412; NCBI Ref Seq: NC_017763.1) and ST80 (NCTC 8325; NCBI Ref Seq: NC_007795.1).

Mouse nasopharynx isolates were screened by PCR amplification of the staphylococcal protein A (spa) gene to confirm that the S. aureus found in the nasopharynx was the same spa type as the inoculum strain. The reactions were undertaken using Mastercycler pro S (Eppendorf) and each contained 0.5 μM PA 1095F forward PCR primer, 0.5 μM PA 1517R reverse PCR primer (Shopsin et al., 1999), 80 ng DNA, 25 μL of BioMix Red (Bioline) and DEPC-treated water up to a total reaction size of 50 μL mixed into 0.2 mL PCR tubes. The run cycle parameters were an initial 10 min at 95°C and then 30 cycles of 30 s at 95°C, 30 s at 60°C and 45 s at °C with a final 10 min at 72°C. PCR products were purified by using the Isolate II PCR and gel kit (Bioline), analyzed by agarose gel electrophoresis and then sequenced (GATC Biotech) with samples prepared according to the company requirements. The spa sequence was analyzed by using BLASTn (Madden, 2002) with default settings. Matched PCR fragment sizes and BLASTn output data for both control and passaged clones were used to indicate that they belonged to the same strain.