At least 2 million people in the USA become infected with antibiotic-resistant bacteria every year, and at least 23,000 people die of these bacterial infections [1]. The growing antibiotic resistance of human pathogens is a serious threat to global health and has a significant impact on the environment. According to Antibiotic Resistance Genes Database (ARDB) [2], 13,293 antibiotic resistance genes of microorganisms have been discovered so far. Transfer of genetic elements between bacteria through intricate routes in mixed microbial communities promotes dissemination of resistance genes [3].

The human gut is home to about 10¹⁴ microbial cells and approximately 1000 microorganisms [4]. It is a dynamic reservoir of antibiotic resistance genes termed the resistome [5]. Antibacterial treatment has a significant impact on the gut resistome: it stimulates horizontal gene transfer and exerts selective pressure on its members [6]. Studies of gut microbiota residents resistant to antibiotics show that commensals of the human gut can also be a source of resistance genes for other bacteria, including pathogenic strains [7].

Studies of antibiotic resistance employ various cutting-edge technologies and methods, such as next generation sequencing, bioinformatic analysis, or analytical chemistry, making it possible to identify up to 30 gene clusters associated with antibiotic resistance [8]. Researchers of the Center for Genome Sciences and Systems Biology, Washington University School of Medicine, analyzed genes responsible for resistance to 18 clinically relevant antibiotics across ecologies. The bioinformatic analysis identified genes conferring resistance to two antibiotics widely used in the clinical setting and agriculture: β-lactams and tetracyclines [9].

Antimicrobial therapies can be seriously disrupted by aminoglycoside phosphotransferases (Aph) [10]. Genes that encode these enzymes were first discovered in plasmids and mobile elements of clinical strains of gram-positive and gram-negative bacteria [11]. As demonstrated by the phylogenetic analysis of Aph of clinical strains and strains producing aminoglycoside antibiotics [12], aminoglycoside phosphotransferases can be organized in 7 groups depending on the enzyme-modified position of the hydroxyl group of the antibiotic: Aph(2''), Aph(3'), Aph(3''), Aph(4), Aph(6), Aph(7'') and Aph(9).

Aph-encoding genes have been annotated in many bacterial genomes, including non-pathogenic strains of the gut microbiota from where they can transfer to clinical strains [13]. Metagenomic DNA isolated from the human neonatal gut was shown to carry multiple genes conferring resistance to aminoglycosides and β-lactams [14].

A comparative study of 832 human gut metagenomes obtained from the residents of 10 different countries (England, Finland, France, Italy, Norway, Scotland, USA, Japan, China, and Malawi) established that the diversity of resistance genes was largely dependent on the geographical origin of the participant [15].

The spread of aph genes was studied in many laboratories worldwide. The aac(6')-Ie-aph(2'')-Ia gene was found to be the most prevalent gene of enterococcal aminoglycoside resistance; it was detected in 26 out of 27 isolates obtained from patients of an Iranian hospital [16]. The epidemiologic study of 543 clinical strains isolated from Japanese patients showed that of 12 studied genes of aminoglycoside-modifying enzymes, one — the aph(2")-Ie gene- was isolated from 3 strains of Enterococcus faecium and another one –ant(9)-Ia — was detected in E. faecalis, E. faecium and E. avium . Nucleotide sequences of ant(9)-Ia in these 3 enterococci were identical to those of Staphylococcus aureus and were  harbored on transposon Tn554 [17]. Because aminoglycosides are often used to treat staphylococcal infections, a study was carried out to estimate the prevalence of aminoglycoside resistance among methicillin-resistant strains of S. aureus isolated from patients of an Iranian hospital. Genes aac(6')-Ie-aph(2"), aph(3')-IIIa and ant(4')-Ia  were detected in 134 (77.0 %), 119 (68.4 %) and 122 (70.1 %) isolates, respectively [18].

In light of the above, identification of aminoglycoside phosphotransferases in the gastrointestinal metagenomes of Russian residents becomes a pressing issue.

METHODS 

Sample preparation and DNA sequencing

We studied the gut microbiota of 11 healthy individuals of different sex and age, all residents of Moscow and Tver, Russia. Stool samples were collected using standard techniques [19]. Samples were frozen at –80 ^°С until further analysis.

DNA was extracted from weighted amounts of frozen stools using the QIAamp Fast DNA Stool Mini kit (Qiagen, Germany) according to the vendor’s protocol with optimized lysis conditions for microbial DNA extraction (Isolation of DNA from Stool for Pathogen Detection, Qiagen, USA). The concentration of the obtained DNA was measured using the Qubit Fluorometer (Invitrogen, USA). The obtained genomic DNA was fragmented using the Covaris M220 focused ultrasonicator (Covaris, USA) to achieve fragment length between 100 and 700 b. p. (average size was ~350 b. p.).

Libraries for further sequencing were prepared using the NEBNext Ultra DNA Library Prep Kit for Illumina (NEB, UK). Fragments ranging from 250 to 500 b. p. (adapter sequences included) were selected for further sequencing. Quality control of the obtained libraries was performed on the Agilent TapeStation (Agilent Technologies, Germany); the libraries were mixed in equimolar amounts. Adapter sequences used at library prep step were as follows: Read1 (AGATCGGAAGAGCACACGTCTGAACTCCAGTCACNNNNNNATCT CGTATGCCGTCTTCTGCTTG) and  Read2 (AGATCGGAAGAGCGTCGTGTAGGGAAAGAG TGTAGATCTCG GTGGTCGCCGTATCAT), where NNNNNN is a 6-nucleotide index unique for each sample. After quality control was performed and library molecules were counted by quantitative PCR, the libraries were sequenced on one lane of Illumina HiSeq 4000 (101 cycles per each fragment’s end) using the HiSeq 4000 SBS sequencing kit ver. 1 (Illumina, USA). FASTQ files were obtained using bcl2fastq v2.17.1.14 Conversion Software (Illumina). Quality scores were encoded as Phred 33. The obtained metagenomes were uploaded to the Sequence Read Archive (SRA) NCBI. They are presented in tab. 1. 

Quality control of metagenomic libraries and read assembly

Quality control of the resulting metagenomic libraries was performed using FastQC [20]. Read trimming was done using trimmomatic [21]. Contaminating host DNA was filtered by aligning the metagenomic reads against the human genome. Alignment was performed using Bowtie2 [22]. The metagenomic reads were assembled into contigs using SPAdes [23]. Description of the assembled reads is provided in tab. 2.

Compiling a catalog of aminoglycoside phosphotransferases-encoding  genes

Drawing upon the literature [12], we compiled a catalog of aminoglycoside phosphotransferase-encoding genes isolated from the clinical strains of Acinetobacter baumannii, Alcaligenes faecalis, Bacillus circulans, Burkholderia pseudomallei, Campylobacter jejuni, Enterococcus faecalis, Escherichia coli, Enterococcus casseliflavus, Enterococcus faecium, Legionella pneumophila, and Pseudomonas aeruginosa. The catalog listed 21 gene. We also compiled a catalog of amino acid residues encoded by the selected Aph genes.

Metagenomic analysis

A Perl script was written to run the BLASTX search for aminoglycoside phosphotransferase genes in the assembled contigs and to filter the results by 2 parameters: homology and relative alignment length. The search was performed in the catalog of 31 amino acid sequences prepared in advance. Sequence alignments generated by BLASTIX were filtered by homology and relative alignment length. Relative alignment length was calculated with formula  where L_alignment is the length of the obtained alignment and L_sequence is the length of the reference amino acid sequence from the catalog. We did not intend to screen the samples for new aminoglycoside phosphotransferase genes, therefore for homology the minimal value was set to 90 %, and the minimal alignment length was set to 80 %. To profile the species present in the studied samples, MetaPhlAn2 was used [24].

RESULTS

Compiling a catalog of aminoglycoside phosphotransferase genes of clinically relevant strains

Depending on the position of the enzyme-modified hydroxyl group of the antibiotic, aminoglycoside phosphotransferases were distributed into 7 subgroups: Aph(2''), Aph(3'), Aph(3''), Aph(4), Aph(6), Aph(7''), and Aph(9). The catalog of genes of clinical strains was prepared by summing up the data from the review [12]. The catalog of aminoglycoside phosphotransferase-encoding genes of clinically relevant bacterial strains is provided in tab. 3.

Screening Russian metagenomes for aminoglycoside phosphotransferase genes

Using the Perl script, we analyzed gut metagenomes of 11 healthy Russian individuals. The results are presented in tab. 4. It total, we identified 3 aph genes in 7 metagenomes. All genes were identified with 100 % homology. Of these 3 genes, the most prevalent was gene aph(3')-IIIa: it was missing in only one metagenome (D5F). Two aph genes, namely aph(2'')-IIa and aph(3')-IIIa, were present only in metagenome D12F. Gene aph(3'')-Ib was detected in only one metagenome (D5F).

The studied metagenomes were profiled for species diversity using MetaPhlAn2. Reads unambiguously assigned to bacterial species were aligned against metagenomic contigs using Bowtie2. Thus, contigs that carried aminoglycoside phosphotransferase genes [Kovtun AS, unpublished] could be assigned to certain species. Results of the bioinformatic analysis are presented in tab. 5.

DISCUSSION

The in silico analysis of 11 gut metagenomes of healthy Russians revealed the presence of aminoglycoside phosphotransferases in 7 metagenomes. Of 21 aph genes previously isolated from the clinical strains of Acinetobacter baumannii, Alcaligenes faecalis, Bacillus circulans, Burkholderia pseudomallei, Campylobacter jejuni, Enterococcus faecalis, Escherichia coli, Enterococcus casseliflavus, Enterococcus faecium, Legionella pneumophila, and Pseudomonas aeruginosa listed in our aph catalog (tab. 3), only 3 were found in the studied samples. Those are: aph(3'')-Ib, aph(3')-IIIa and aph(2'')-Ia. The most frequently occurring gene was aph(3')-IIIa (CAA24789) detected in 6 samples. This gene was previously discovered in E. faecalis and confers resistance to kanamycin. Gene aph(3'')-Ib (AAA26442) previously isolated from E. coli and associated with streptomycin resistance and gene aph(2'')-Ia (AAA26865) previously isolated from E. faecalis and associated with tobramycin resistance were observed in only one studied metagenome (tab. 3).

Interestingly, the analysis of contigs that harbor aminoglycoside phosphotransferase-encoding genes revealed the presence of the latter in the genomes of other bacterial species. For example, the aph(3')-IIIa gene was detected in a sequence typical for commensal Ruminococcus obeum and opportunistic E. faecium, Roseburia hominis, Streptococcus pyogenes and Staphylococcus epidermidis, but not for  E. faecalis. Gene aph(2'')-Ia was detected in Clostridium difficile, but not in  E. faecalis (Tables 3, 5). Although this gene was the most prevalent in enterococci in the study [16], we observed it in only one studied sample in the non-enterococcal sequence. Genes aph(2")-Ia and  aph(3')-IIIa were previously reported in methicillin-resistant strains of Staphylococcus aureus [17]. However, in the studied Russian metagenomes aph(3')-IIIa was present in the sequence typical for Staphylococcus epidermidis, while aph(3'')-Ib was detected in E. coli.

These results are consistent with the results of comparative analyses conducted worldwide: age, sex and health do not have any significant impact on the antibiotic resistance of the gut microbiota, while the geographic origin does [15]. Rare occurrence and poor diversity of aph genes in Russian metagenomes may indicate that gut microbiota composition is specific to a particular region and that individuals whose microbiomes were analyzed in our study rarely resort to aminoglycoside therapies. On the other hand, missing aph genes in anaerobic bacteria that dominate the gut microbiota may be explained by the absence of cytochrome-mediated transport [25]. It is also important that the microbiome of a healthy individual harbors opportunistic bacteria carrying aph genes.

CONCLUSIONS 

Previously isolated from clinical bacterial strains, genes aph(3'')-Ib, aph(3')-IIIa and  aph(2'')-Ia were found in 7 microbiota samples of 11 healthy Russians. Gene aph(3')-IIIa prevailed. The genes detected in the samples are carried by opportunistic bacteria: Enterococcus faecium, Roseburia hominis, Clostridium difficile, Escherichia coli, Streptococcus pyogenes, and Staphylococcus epidermidis. Two of them — E. coli and E. faecium — belong to a group of 12 highly dangerous bacteria, according to the World Health Organization. Therefore, we believe it reasonable to run antibiotic resistance tests on both the causative agent and patient’s microbiota before deciding on the antibiotic treatment for patients with bacterial infections.

This work is the first to study the spread of antibiotic resistance genes of the gut microbiota of Russians. Further PCR-based search should be conducted to identify other clinically relevant resistance genes.