Staphylococcus aureus nasal carriage and microbiome composition among medical students from Colombia: a cross-sectional study

Study design and population

The Ethics Review Boards of the University of Cartagena (Approval #280313) and New York Medical College approved this study (Protocol # 12697; IRB ID: 12697). The study was conducted between January and June of 2018 and enrolled second-year medical students from University of Cartagena, Colombia, who had not yet engaged in clinical rotations. To prevent sampling bias, we aimed to enroll the complete population of second-year medical students of our institution. Students were recruited via fliers and lecturer announcements. Those who agreed to participate signed an informed consent and completed a written questionnaire on demographics and medical history before each nasal swab sampling. Exclusion criteria were recent infections, allergies and other non-infectious pathologies, smoking habits, antibiotic usage in the previous three months, surgeries and hospitalizations in the previous six months. In total 143 out of 158 second-year medical students completed the study. Thus, we enrolled 90.5% of the total population of second-year medical students in our institution in this study. We followed the STROBE cross sectional reporting guidelines¹⁸.

Specimen collection, prevalence of colonization and carriage categories

Nasal swabs were obtained from both nostrils by a trained individual, inoculated into Stuart transport medium (OXOID, England), transported to the microbiology laboratory and processed within 8–18 hours according to described protocols⁶. S. aureus was identified based on colony morphology, Gram-stain, catalase-test, tube coagulase-test, and latex agglutination-test. Genomic DNA was obtained from each isolate with Wizard^® Genomic DNA Purification Kit (Promega, USA) and S. aureus molecular confirmation and methicillin-resistance were assessed by PCR-amplification using specific primers for nuc and mecA genes, respectively. Detailed protocols for all these methods, including primer sequences, have been previously described⁶. To determine prevalence of S. aureus nasal carriage, each participant was classified either as carrier or non-carrier based on laboratory results obtained from the first nasal swab survey. To establish S. aureus long-term carriage categories, four additional consecutive nasal swabs were obtained from each participant, in three-week intervals. According to definitions proposed by Kluytmas et al.¹⁹, participants that yielded five negative cultures for S. aureus were classified as non-carriers; those yielding one to three positive cultures were classified as intermittent carriers; and those yielding four or five positive cultures were classified as persistent carriers.

Nasal microbiome (bacterial and fungal) analysis

This study aimed to describe the microbiome composition in a small group of second-year medical students with a known S. aureus carriage status. At the end of the study, 15 participants with known S. aureus long-term carriage status (5 non-carriers, 5 intermittent-carriers, and 5 persistent-carriers) were randomly selected from the cohort of 143 participants. The 15 selected participants provided an additional nasal swab that was stored at room temperature in Amies-Transport-Medium with Charcoal (Copan Diagnostics, Inc., Murrieta, CA), and sent to the laboratory of Robert D. Burk at Albert Einstein College of Medicine (AECOM) for microbiome analysis. Sample processing was performed according to protocols described by Usyk et al.²⁰. The time between collection and processing of samples was around 8 days.

16S rRNA gene and Internal Transcribed Spacer (ITS) PCR-amplification

The V4 hypervariable region of 16S rRNA gene was amplified using primers 16SV4_515F (GTGYCAGCMGCCGCGGTA) and 16SV4_806R (GGACTACHVGGGTWTCTAAT), with a unique 12-bp barcode Golay-barcoding^21,22. PCR conditions were: initial 5min denaturation at 95°C, followed by 15-cycles of 95°C for 1min, 55°C for 1min, and 68°C for 1min, and final extension for 10min at 68°C. For the fungal component of microbiome, barcoded amplicons were generated covering the ITS gene region using ITS1-30F/ITS1-217R primer pair, as previously described by Usyk et al.²⁰ (ITS1-30F: 5’-GTCCCTGCCCTTTGTACACA-3’ and ITS1-217R: 5’-TTTCGCTGCGTTCTTCATCG-3’). PCR conditions were: 3min initial denaturation at 95°C, followed by 35-cycles of 95°C for 30sec, 55°C for 30sec, and 68°C for 2min, followed by final extension at 68°C for 10min. PCR reagents were obtained from Affymetrix (Affymetrix, Santa Clara, CA). PCR reactions were run in GeneAmp PCR-System 9700 (Applied Biosystems).

High-throughput sequencing

PCR products were purified using QIAquick Gel Extraction Kit (QIAGEN) and quantified using Qubit^TM dsDNA High-Sensitivity Assay kit (Life Technologies). Next-generation sequencing library preparation was performed using KAPA-LTP library preparation kit (KAPA Biosystems, Wilmington, MA). Size integrity of isolated amplicons was validated with 2100 Bioanalyzer (Agilent-Technologies, Santa Clara, CA). High-throughput sequencing of libraries was performed using Illumina HiSeq2500 Sequencing System (Illumina, San Diego, CA) with a 2×250-bp paired-end read kit at the Genomics Core Facility of AECOM.

Bacterial microbiome bioinformatics analysis

Illumina reads were pre-processed to remove bases that fell below PHRED quality score of 25 using PRINSEQ²³. Processed reads were de-multiplexed using sample specific barcode combinations with Novobarcode V1.00. This can also be performed with deML, a program freely available for use under the GPL license. Paired-end reads were merged using free open source PANDAseq v1.20 with default settings²⁴. OTU-clustering and quality filtering was performed using the Quantitative Insights Into Microbial Ecology (QIIME v1.9) software package²⁵. Removal of sequencing noise and sequence chimeras was done with USEARCH v8.0²⁶. Sequences were de-multiplexed and clustered into operational taxonomic units (OTUs) with 97% minimum cluster similarity using UCLUST²⁶. Sequences were assigned using UCLUST with Greengenes 13.8 microbial database²⁷. The resulting BIOM table was rarefied to 29,000 reads/sample and statistical analyses were performed after collapsing OTUs at genus level.

Fungal microbiome bioinformatics analysis

Sequence reads were processed using open-reference OTU-picking with QIIME v1.9 against the targeted host-associated fungi ITS database (THF1)²⁸ for the reference-based clustering component. VSEARCH v1.4.0²⁹ was used to de-replicate reads, cluster reads into OTUs and remove chimeric sequences. OTU-clustering threshold was set at 99% sequence identity to account for fungal heterogeneity. Sequence de-replication and chimera removal were performed using QIIME quality-control protocol. Representative sequences for each OTU-cluster were chosen based on sequence abundance. BLAST was used to assign taxonomy using the UNITE database³⁰. The default behavior of BLAST in QIIME was changed to minimum of 99% sequence identity for taxonomic assignment. Data were processed in R v3.3.1³¹. QIIME outputs were imported into R using phyloseq package v1.22.3³² and further processed with vegan v2.5-3³³, coin v1³⁴, and reshape2³⁵. Data visualization was performed using ggplot2³⁶.

Alpha/Beta diversity analysis of 16S rDNA-V4 and ITS sequences

Statistical analyses were performed to assess differences in OTU distribution and abundance between samples and groups. Microbial diversity for bacterial and fungal communities was evaluated within samples (α-diversity) or between samples (β-diversity) using QIIME. Rarefaction to subsampling depth of 29,000 reads/sample or 9,000 reads/sample, for bacteria or fungi respectively, and 5 iterations were performed on all samples to standardize the sequencing effort. Alpha-diversity was measured with Chao1 (richness) and Shannon entropy (OTU-based diversity) index. Beta-diversity was calculated using Bray-Curtis dissimilarity coefficient. To test for dissimilarities in the microbial composition between S. aureus carrier groups, non-metric multidimensional scaling (NMDS) was performed with Bray-Curtis dissimilarity.

S. aureus nasal carriage

The first nasal swab isolated S. aureus from 40 out of 143 participants, for a prevalence of 28%. Methicillin-resistant S. aureus (MRSA) was carried by 12 participants (8.4%) and methicillin-sensitive S. aureus (MSSA) by 28 (19.6%). The longitudinal study identified 28 (19.6%) persistent carriers, 25 (17.5%) intermittent carriers, and 90 (62.9%) non-carriers. MRSA strains were isolated from 6 persistent carriers and 6 intermittent carriers. Table 1 lists the characteristics of students used in the microbiome study. Underlying data: Table S1³⁷ lists the main results obtained from the complete study population.

Diversity and abundance of resident bacterial and fungal communities

Figure 1 shows clustering analysis of bacterial genus (A) and fungal species (B) compositions of nasal specimens from the cohort of 15 healthy medical students with known S. aureus carriage patterns. Bacterial microbiome analysis, sequencing, quality filtering and mapping resulted in 1,424,972 mapped V4-region sequences, ranging between 29,971-217,540 copies per sample (average 120,062; SD = 53,027), corresponding to 600 OTUs. We identified 57 of the 600 OTUs (9.5%) at the genus level, while the remaining OTUs mapped to unclassified genera (70.8%) or upper taxonomic groups (96.8%). Three phyla were identified, Proteobacteria being the most abundant (78.3% of 16S rRNA sequences) and the most diverse (42.2% of all identified OTUs), followed by Firmicutes and Actinobacteria. In general, the nasal bacterial microbiome was dominated by the Class Gammaproteobacteria (order Pseudomonadales and genera Citrobacter and Acinetobacter), which is consistent with a 2016-review by Lee et al.³⁸. Table 2 shows the predominant bacterial genera isolated from the nostrils of healthy medical students.

Figure 1.

Cluster analysis of bacterial genus (A) and fungal species (B) compositions in the nares of healthy medical students with a known S. aureus carriage pattern. Heatmap was constructed using normalized log10 abundance of each OTU in each sample type. Data are presented only for the 20 most abundant taxa. Colored bar above heatmap indicates S. aureus carriage status. P: persistent (red), I: intermittent (blue), NP: non-carrier (green).

Alpha-diversity analysis found that bacterial diversity was greater in the persistent group followed by the intermittent and then by the non-carriers, as shown in Figure 2A. However, the difference in diversity among the three carrier groups was statistically non-significant. Beta-diversity analysis of bacterial communities showed non-significant separation of groups (Figure 2B). Analysis of differential abundance identified that order Pseudomonadales was more abundant in non-carriers followed by intermittent and persistent categories (Figure 3), a finding that may have implications regarding microbial antagonism. However, this trend did not reach statistical significance (p>0.05, Kruskal–Wallis test).

Figure 2. Alpha and beta diversity of bacterial composition of nasal samples.

(A) Alpha-diversity, measured by Chao1 and Shannon diversity Index, is plotted for individuals with different S. aureus carrier status: non-carriers (NP, dark green), intermittent carriers (I, purple) and persistent carriers (P, light green). The Chao1 index (left panel) and Shannon index (right panel) were computed for all 15 subjects. The line inside the box represents the median, while the whiskers represent the lowest and highest values within the 1.5 interquartile range (IQR). Statistical testing showed no significant differences among the groups: Chao1 p = 0.0675; Shannon p =0.108. (B) Comparison of beta-diversity of bacterial composition between S. aureus carriage groups with NMDS ordinance calculated from Bray–Curtis distance estimation. I: Intermittent, NP: non-carrier, P: persistent.

Figure 3. Boxplots representing relative abundance analysis of the bacterial taxa identified in nasal samples of individual from the different S. aureus carrier groups.

(A) Differential abundance at the genus level. (B) Differential abundance for the order Pseudomonadales. Taxa with minimum median abundance of 1% were used for the comparison. There was non-significant difference in abundance among the three carrier groups (p>0.05, Kruskal–Wallis test). NP: non-carrier, I: Intermittent, P: persistent.

For fungal microbiome analysis, sequencing, quality filtering and mapping resulted in 4,274,743 mapped ITS-region sequences, ranging between 9,543-743,278 copies per sample (average 284,983; SD = 237,339), corresponding to 8,346 fungal OTUs (average 556; SD = 333) (Table 3). We were able to classify 4,453 of the 8,346 OTUs (53.3%) down to species level. Out of the seven recognized major phylum of fungi, three phyla were identified in the nostrils, being Ascomycota the most abundant (90.8% of fungal sequences) and the most diverse (88.7% of all identified OTUs), followed by Basidiomycota and Neocallimastigomycota (containing anaerobic fungi). In general, the nasal mycobiome was dominated by species of the phylum Ascomycota, which is consistent with a 2013-publication by Findley, et al.³⁹. Table 4 shows the predominant fungi species identified from the nostrils of healthy medical students. The most frequent fungal species detected was Penicillium oxalicum, representing 19.0% of the total 4,274,743 sequence reads in the chimera‐filtered OTU table. Malassezia restricta was the second most frequent fungus (12.2% of all sequences). 4.4% of all sequences were unassigned sequences, thought to represent non-fungal contamination.

Based on the identified OTUs and number of sequences in this study, we report a substantial diversity and amount of fungal biomass in the anterior nares of this group of medical students with Colombian ethnicity. Consistent with the 16S analysis, alpha-diversity of fungal communities showed that fungal diversity was greater in the persistent group followed by the intermittent and then by the non-carrier groups. However, this trend did not reach statistical significance (Figure 4A). Beta-diversity analysis of fungal communities showed non-significant separation of groups (Figure 4B).

Figure 4. Alpha and beta diversity of fungal composition of nasal samples.

(A) Alpha-diversity, measured by Chao1 and Shannon diversity Index is plotted for individuals with different S. aureus carrier status: non-carriers (NP, dark green), intermittent carriers (I, purple) and persistent carriers (P, light green). The Chao1 index (left panel) and Shannon index (right panel) were computed for 15 subjects. The line inside the box represents the median, while the whiskers represent the lowest and highest values within the 1.5 interquartile range (IQR). Statistical testing showed no significant differences among the groups: Chao1 p = 0.13; Shannon p = 0.566. (B) Comparison of beta-diversity of fungal composition between S. aureus carriage groups with NMDS ordinance calculated from Bray–Curtis distance estimation. I: Intermittent, NP: non-carrier, P: persistent.

Analysis of differential abundance showed increased abundance for several fungi species in the persistent group compared to the intermittent and non-carriers (Figure 5A). However, the species Candida orthopsilosis was the only one with significant difference in abundance between the persistent and non-carrier groups (p<0.05, Kruskal-Wallis test) (Figure 5B).

Figure 5. Boxplots representing relative abundance analysis of the fungal taxa identified in nasal samples of individual from the different S. aureus carrier groups.

(A) Differential abundance at the species level. (B) Differential abundance for the species Candida orthopsilosis showed a significant difference in abundance between the persistent and non-carrier groups (p<0.05, Kruskal-Wallis test). Taxa with minimum median abundance of 1% were used for the comparison. NP: non-carrier, I: Intermittent, P: persistent.

Underlying data

Nasal Microbiome. Raw sequencing data of the nasal microbiome of a set of medical school students, Accession number, PRJNA600228: https://www.ncbi.nlm.nih.gov/bioproject/600228.

Open Science Framework: Nasal Microbiome of Medical Students UdeC. https://doi.org/10.17605/OSF.IO/UDNWA³⁷.

This project contains the following underlying data:

Data are available under the terms of the Creative Commons Zero "No rights reserved" data waiver (CC0 1.0 Public domain dedication).