High fecal carriage of blaCTX-M, blaCMY-2, and plasmid-mediated quinolone resistance genes among healthy Korean people in a metagenomic analysis

To characterize the carriage of antibiotic resistance genes (ARGs) in the gut microbiome of healthy individuals. Fecal carriage of ARGs was investigated in 61 healthy individuals aged 30 to 59 years through whole metagenome sequencing of the gut microbiome and a targeted metagenomic approach. The number of ARGs in the gut microbiome was counted and normalized per million predicted genes (GPM). In the Korean population, the resistome ranged from 49.7 to 292.5 GPM (median 89.7). Based on the abundance of ARGs, the subjects were categorised into high (> 120 GPM), middle (60‒120 GPM), and low (< 60 GPM) ARG groups. Individuals in the high ARG group tended to visit hospitals more often (P = 0.065), particularly for upper respiratory tract infections (P = 0.066), and carried more blaCTX-M (P = 0.008). The targeted metagenome approach for bla and plasmid-mediated quinolone resistance (PMQR) genes revealed a high fecal carriage rate; 23% or 13.1% of the subjects carried blaCTX-M or blaCMY-2, respectively. Regarding PMQR genes, 59% of the subjects carried PMQR, and 83% of them harboured 2‒4 PMQR genes (qnrB 44.3%, qnrS 47.5% etc.). The presence of blaCTX-M correlated with ARG abundance in the gut resistome, whereas PMQR genes were irrelevant to other ARGs (P = 0.176). Fecal carriage of blaCTX-M and PMQR genes was broad and multiplexed among healthy individuals.

Fecal DNA preparation. Each sample was thoroughly mixed using a spatula and divided into 250-300 mg aliquots. Total DNA was extracted using a Fast DNA SPIN Kit for Faeces (MP Biomedicals, #116,570,200), following the manufacturer's instructions.
Analysis of metagenome sequencing data. Metagenome sequencing data (n = 61) of the gut microbiome from our study were obtained from the European Nucleotide Archive (ENA; accession number PRJEB33013). For every sample, the length of the paired-end reads was 151 bp with an insert size of 350 bp. A three-step procedure was performed. First, filtered reads were assembled into contigs using MEGAHIT 12 with default options. Genes were predicted from the contigs (> 500 bp) using FragGeneScan 13 with the options of no sequencing error (-w 0 -t complete). Lastly, ARGs were identified using the CARD databaseS1 14 and BLASTp 15 with an e-value threshold of 1 × 10 -10 , similarity > 70%, and reference coverage > 70%. The resistance genes were classified into 20 ARG classes (Supplementary Table ). The abundance of non-redundant ARGs was measured using the number of ARGs per million genes (GPM) in each sample as follows: Detection of ARGs by polymerase chain reaction and sequencing. The following AMR genes: bla CTX-M (bla CTX-M-1 and bla CTX-M-9 group), p-AmpC (bla CMY-1 , bla CMY-2 , bla DHA , and bla FOX ), carbapenemases (bla IMP , bla NDM , bla KPC , and bla VIM ), and PMQR (qnrA, qnrB, qnrD, qnrS, qepA, and aac(6′)-Ib-cr) were detected using PCR with specific primers and conditions summarised in Supplementary Table S2 [16][17][18][19][20][21][22][23][24][25] . After electrophoresis on 2% agarose gel, the resistance alleles were identified by sequencing the PCR products and comparing the sequences with those in the GenBank database.
Statistical analysis. To compare demographic characteristics and health information, SPSS version 24.0 for Windows (SPSS Inc., Armonk, NY, USA) was used. Pearson's Chi-square test or Fisher's exact test was used to analyze categorical variables, and the independent t-test or Mann-Whitney U-test was used to analyze continuous variables. Spearman's rank correlation test was performed to evaluate the relationship between the two variables. A P-value of < 0.05 in a two-tailed test was considered significant.
Access to data. Metagenome sequencing data of the gut microbiome were obtained from the European Nucleotide Archive (ENA; accession number PRJEB33013).

Results
Demographic data and personal information of 61 healthy Korean individuals. A total of 61 healthy individuals (61 fecal samples) were included in this study. The median age was 46 years (1Q, 3Q; 37, 51) and 47.5% of the subjects were female. Proportion of gender was similar among the age of 30 s, 40 s and 50 s. The median body mass index was 23.3 (1Q, 3Q; 21.6, 25). Personal histories including water, probiotic, and vitamin intake as well as the presence of a companion animal are shown in Table 1. As for the medical history, 11.5% of the subjects reported that they had been prescribed antibiotics within the past year, 34.4% did not visit the hospital, and 50.8% visited the hospital less than three times within the past year. For the treatment of upper respiratory tract infections (URIs), 47.5% of the participants did not visit the hospital and 49.2% reported visiting the hospital fewer than three times, and 3.3% visited four to six times; for dental problems, 44.2% visited a dental clinic. www.nature.com/scientificreports/ Different phenotypes between the groups with high and low ARG abundance. Figure 1A presents the abundance of non-redundant ARGs in the gut microbiome of healthy Korean individuals. In the Korean population, the resistome ranged from 49.7 to 292.5 GPM, with a median value of 89.7. Aminoglycoside was the most abundant antibiotics resistance determinant out of 20 (supplementary Table S1), which was followed by tetracycline, macrolides-lincosamides-streptogramins shared (MLS), and beta-lactam (Fig. 1A, supplementary Table S3). The prevalence of these determinants was more than 98% of the samples. Based on the abundance of non-redundant ARGs in the gut, we categorised the 61 healthy individuals into the following three groups: high ARG (HARG), consisting of 10 subjects with ARGs > 120 GPM; low ARG (LARG), consisting of 10 subjects with ARGs < 60 GPM; and middle ARG (MARG), consisting of 41 subjects with ARGs between HARG and LARG (Fig. 1A). The median values for the abundance of ARG in HARG, MARG, and LARG (range) were 142.5 (124.1, 160.7), 89.7 (70.7, 97.7), and 51.8 GPM (50.0, 55.1), respectively. Principal component analysis showed that HARG is completely separate from LARG (Fig. 1B). In terms of resistance abundance, nine resistance determinants were significantly more abundant in HARG than that in LARG group. Particularly, β-lactam was the most significant, followed by aminoglycoside, diaminopyrimidine, lincosamide, mupirocin, peptide, polymyxin, sulphonamide, and tetracycline (p-value < 0.01; Fig. 1C).

GPM
Comparison of the lifestyles of high, middle, and low ARG abundance groups. On comparing the lifestyle of the three groups, no difference was observed in water and medication intake and the presence of a companion animal. However, the medical history within the last year showed a difference; the number of hospital visits and hospital visit due to URI correlated with ARG abundance with a marginal significance (p for trend = 0.066, p for trend = 0.065, respectively; Table 1). In terms of the diet habit and fecal carriage of ARG, we could not find a significant difference in diet habit among people with HARG, MARG and LARG (Supplementary Table S4). Table 1. Comparison of demographic data and personal or medical history between people with HARG and LARG in the gut. *p for trend. HARG high antibiotic resistance gene, LARG low antibiotic resistance gene, d/t due to, Q quartile.   (Table 2). Overall, 30 out of the 36 subjects (83%) harboured two to four PMQR genes; three subjects (4.9%) harboured qnrB, qnrD, qnrS, and aac(6′)-Ib-cr: 7 (11.5%), qnrB, qnrS, and aac(6′)-Ib-cr: 20 (32.8%), qnrB and qnrS: one (2.4%) and nine (14.8%) qnrB or qnrS alone, respectively (Fig. 2).
Comparison of fecal ARG carriage rate among HARG, MARG, and LARG groups. The fecal carriage rate of CTX-M, p-AmpC, and PMQR genes was compared among HARG, MARG, and LARG groups The distribution of resistance determinants in LARG, MARG, and HARG groups. Sample numbers are shown in x-axis. HARG consisted of 10 subjects who had a higher abundance (> 120 GPM) of antibiotic resistance genes. LARG consisted of 10 subjects that had lower abundance (< 60 GPM) of antibiotic resistance genes. MARG was between LARG and HARG, which had ARGs of higher than 60 GPM and lower than 120 GPM (B) Principal component analysis of resistance determinants in HARG, MARG and LARG. (C) Nine antibiotic determinants were significantly enhanced in HARG of the two groups (P-value < 0.01). HARG high antibiotic resistance genes group, MARG middle antibiotic resistance genes group, LARG low antibiotic resistance genes group. Associated presence of CTX-M, p-AmpC, and PMQR genes and their relationships with medical history or the relative abundance of ARG groups in the gut resistome. Associated presence among each ARG allele was analyzed and is presented in Fig. 3. The presence of individual qnrB, qnrS, and aac(6′)-Ib-cr was linked to one another, and carriage of the bla CMY-2 group was significantly associated with the carriage of most PMQRs as well (qnrB, qnrS, and aac(6′)-Ib-cr). However, the presence of CTX-M alleles was not associated with that of other resistance alleles. The presence of CTX-M, CMY2, and PMQR genes in each individual was also analyzed using the abundance of ARG groups classified based on antibiotic resistance ontology 14 , Table 2. Faecal carriage rate of CTXM, plasmid-mediated AmpC, and plasmid-mediated quinolone resistance alleles in 61 healthy people. Plasmid-mediated carbapenemase alleles were not identified. HARG high antibiotic resistance genes group, MARG middle antibiotic resistance genes group, LARG low antibiotic resistance genes group. CTX-M-1G, CTX-M-1 group; CTX-M-9G, CTX-M-9 group. *By Kruskal-Wallis test. www.nature.com/scientificreports/ which was determined using metagenome analysis of their gut microbiomes (Fig. 3). Carriage of CTX-M alleles showed a significant association with the abundance of most ARG groups, β-lactam, aminoglycoside, peptide, macrolide, sulphonamide, tetracycline, and total ARG. However, CMY2 and PMQR genes were not related with the abundance of ARG groups in the gut resistome except qnrB vs. polymyxin and peptide. Carriage of those resistance alleles and demographics or medical history showed no significant correlation (Fig. 3).

Discussion
This study found that healthy Korean individuals carried high frequencies of bla CTX-M , bla CMY-2 , and PMQR genes in their gut. The fecal carriage of ARGs in healthy individuals was analyzed by metagenome sequencing of faeces and ARG-specific PCR, along with lifestyle questionnaires.  www.nature.com/scientificreports/ Among healthy people, the abundance of ARGs showed great variance. We had expected that diet habit and medical history influenced the ARG abundance in gut. Unfortunately, we could not verify the association between diet habit and ARG abundance, but frequency of clinic visit was marginally associated with ARG abundance in the gut. Two subjects of sample 47 and 28 showed an exceptionally high abundance of ARGs in the gut (Fig. 1A), and all the ARG determinants increased. Interestingly, abundance of Escherichia genus increased over 100-fold more than median value of 61 people in bacterial composition (data not shown), which suggests a recent antibiotic usage in both subjects, despite individuals not mentioning antibiotic usage in life style survey.
Interestingly, the fecal carriage of bla CTX-M and PMQR genes in healthy individuals looked different in distribution; the presence of bla CTX-M was proportional to the abundance of ARGs in the gut microbiome and its presence correlated with many ARG groups, such as β-lactam, aminoglycoside, macrolide, sulphonamide, and total ARGs in the gut microbiome. However, PMQR genes were observed regardless of ARG abundance in the gut, except qnrB vs. polymyxin and peptide alleles. Considering that subjects in the HARG group tended to visit clinics more often, especially due to URI, this may imply that bla CTX-M carriage is associated with the prescribed antibiotic intake.
With respect to PMQR genes, the fecal carriage rate was very high in healthy Korean individuals regardless of the relative ARG abundance in the gut microbiome. Many individuals carried multiple PMQR genes at the same time; 5% of the subjects harboured four PMQR genes and 11.5% harboured three PMQR genes, indicating frequent exposure to PMQR during daily life. In the metagenome analysis of the gut microbiome, qnr alleles were not frequently encountered; we found qnr genes in only 3 out of 61 individuals in this study (data not shown), which indicates a high prevalence but low density of PMQR genes in healthy Korean individuals.
Many previous studies have presented an association between PMQR and bla CTX-M alleles 5,6,26 ; however, an association between these alleles was not identified in this study. Instead, an association between bla CMY-2 and PMQR genes was observed. A recent study published in 2018 demonstrated a high prevalence of bla CMY-2 in blood isolates of Escherichia coli from Singapore, suggesting a high prevalence of bla CMY-2 in Asian countries 27 . CMY β-lactamase is predominant p-AmpC in the animal sector-especially broiler and meat-along with qnrs 6,28 . These findings suggest a common environmental source-such as broiler meat, contaminated water, or vegetables-for exposure to bla CMY-2 and PMQR genes.
In this study, the fecal carriage rate of bla CTX-M and/or bla CMY-2 was 29.5%, and that of PMQR genes was 59% in healthy Korean individuals. In our metagenomic analyses of the gut microbiome, bla CTX-M and bla CMY-2 contigs were observed in 4 out of the 61 individuals (6.6%) and 1 out of the 61 individuals (1.6%), respectively 29 , indicating a relatively low density of bla CTX-M and bla CMY-2 in the guts of healthy individuals without antibiotic intake.
Interestingly, a recent study on antibiotic resistance in E. coli isolated from community onset acute pyelonephritis patients in Korea-in 2017 and 2018-showed that resistance rates against FQ and cefotaxime were 33.5% and 34.8%, respectively 30 . The fecal carriage rate of ARGs in this study represents the resistance rates of E. coli in community infections.
The advantage of this study is that the state of ARG carriage was investigated in detail along with the gut resistome and lifestyle. Despite this strength, some limitations were also there; first, the number of enrolled subjects was not sufficient to represent healthy Korean individuals. Second, we could not directly show the risk factors in lifestyle or medical history for the fecal carriage of antibiotic resistance alleles.

Conclusion
In conclusion, bla CTX-M , bla CMY-2 , and PMQR genes were broadly distributed in the gut microbiome of healthy Korean individuals. The fecal carriage rate of bla CTX-M was 23%, mostly in individuals with a high abundance of gut resistome. Conversely, 59% of the healthy individuals carried multiple PMQR genes in their gut microbiome regardless of ARG abundance in their guts, which could indicate frequent exposure to PMQR genes. www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.