Most authors were employed by Chr. Hansen A/S, a company that produces strains for plant protection, animal and human health, as well as for the food and dairy industry. Some of the authors are share-holders in Chr. Hansen A/S. Two authors were consultants paid by Chr. Hansen A/S. This does not alter our adherence to PLOS ONE policies on sharing data and materials.
Whole-genome sequencing and phenotypic testing of 104 strains of
The development and spread of antimicrobial resistance is considered a major threat to human health [
Antimicrobial resistance mechanisms can be present on mobile genetic elements such as plasmids or conjugative transposons and such elements may spread horizontally between bacteria from different environments and between different bacterial species [
Antimicrobial resistance in pathogenic bacteria, whether intrinsic or acquired, can contribute to treatment failure due to ineffective antimicrobial therapy. Intrinsic resistance mechanisms, when present in non-pathogenic bacteria, do not add to the resistance pool in pathogenic bacteria as intrinsic resistance is not readily spread horizontally between bacteria. Resistance mechanisms linked to mobile genetic elements in non-pathogenic bacteria are, however, considered a risk [
The genus
With recent advances in genome sequencing technologies, draft genome sequences of several
The objective of the present study was to assess the genetic background and evolutionary history of four putative resistance determinants, conferring resistance to erythromycin (
A collection of 104 strains, composed of
Species | Source | Year of isolation | Geographic origin |
---|---|---|---|
Human faeces (2), Animal faeces (17), Animals other (8), Food (16), Feed (2), Soil (12), Plants (2), Environment (9), Not given (6) | 1878 (1), 1917 (1), 1921 (1), <1944 (1), 1944 (1), <1950 (1), <1952 (1), <1956 (1), 1958 (1), <1963 (1), <1964 (1), <1976 (1), <1979 |
Denmark (16), France (1), Germany (12), United Kingdom (3), Spain (2), The Netherlands (2), Norway (2), Sweden (2), Sudan (4), Egypt (2), Philippines (2), Vietnam (3), Japan (1), Australia (2), USA (2), Not given (18) | |
Animal faeces (5), Animals other (1), Food (5), Soil (10), Plants (1), Environment (2), Not given (6) | <1950 (1), <1951 (3), <1953 (1), <1954 (1), 1965 (1), <1977 (1), <1980 (1), <1986 (1), 2010 (1), 2011 (6), 2013 (2), 2014 (1), Not given (10) | Denmark (3), Germany (2), United Kingdom (1), Spain (2), Ghana (1), Sudan (2), China (1), Australia (1), USA (3), Not given (14) |
Numbers in brackets represent the number of strains; < in front of year indicates that only the date of deposit is known;
a) CHCC20375 could not be definitively identified as either
b) CHCC20323 was isolated from a can of tinned veal sealed around 1825, however, the strain was deposited in 1979 and hence the year of isolation is given here as <1979
Twenty-six of the
Additional details on the strains and the relevant MIC values can be found in
DNA sequencing was performed at BaseClear (Leiden, the Netherlands) for most strains, using Illumina Hiseq2500 sequencing with 125 bp paired-end read length. Cell pellets were provided, and DNA was extracted at BaseClear using ZR Fungal/Bacterial DNA kit (Zymo Research; D6005), according to the manufacturer’s protocol. For 17 strains, total DNA extraction was done in house using a QIAcube and the Qiagen blood and tissue kit (Hilden, Germany), according to the manufacturer’s protocol. The DNA samples were sent to BGI (Shenzhen, China) for Illumina Hiseq2500 sequencing with 125 bp paired-end read length.
The paired-end reads obtained were assembled
The genome sequences of
Species identification was performed by whole genome average nucleotide identity (ANI) calculation with the help of the JSpecies software using default parameters [
The Rapid Annotation using Subsystems Technology (RAST) server [
The RAST annotations were queried to identify putative resistance genes by text searching the excel file which can be downloaded together with the annotated genome. Genes with an annotation containing words related to antibiotic resistance (e.g antimicrobial, resistance, antibiotic), and adjacent genes, were analyzed manually to improve the functional annotations by comparing the gene sequence to curated protein sequence databases using BLAST (
All protein sequences potentially encoded in the genomes were compared using OrthoMCL [
The OrthoMCL output matrix containing OGs,
All OGs with a single gene copy in each of the strains were selected based on the OrthoMCL output. The protein sequences in these OGs were aligned using Muscle [
The macrolide resistance gene
To identify putative streptomycin or chloramphenicol resistance genes a text search of the RAST annotated function using the terms “chloramphenicol” and “aminoglycoside” as performed. The identified genes and adjacent genes were compared with sequences in NCBI database using BLAST in order to improve the functional annotation or qualify it.
ErmD and Putative resistance proteins were compared for diversity at the single protein level by aligning the sequences with Clustal 2.1, Clustal Omega or Muscle using default settings. Phylogenetic trees and manual assessment of SAP variations was used for determination of protein variant types and the variance types was compared to the whole-genome phylogenetic tree of the strains and to the MIC value of the antibiotic in question.
The flanking region of each of the putative resistance genes in all genomes were aligned and ordered based on contig alignments to the template genomes, to determine the location of the genes and their organization in the genomes. Finally, an NCBI GenBank BLAST search using the DNA or amino acid sequence of putative resistance genes as query sequence with default settings (Program settings: megablast; discontiguous megablast; and protein protein BLAST) was performed to evaluate similarity to known resistance genes in other bacteria.
HGTector (
The published genome sequences of
The strains analysed comprise a diverse collection of strains collected over a time span of more than seven decades representing very different origins and geographic areas. MLST analysis and a comparison to known MLST types of both species showed the collection to represent most known sequence types (STs). Moreover, 18 novel STs were found among 23 strains (see
The total number of OGs found was about 11,500. Nearly all of these could be scaffolded (
We found 2915 genes representing the core genome of this evolutionary clade. These genes were present, in single copies, in every
Phylogenetic trees were made based on the amino acid variations in all 2915 predicted core proteins. A core genome phylogenetic tree was made including the genomes of
The predicted core proteins in the 73
Number in parentheses; Multi Locus Sequence Type. Clade A corresponds to
These two distinct clades correspond to the two phylogenetic groups recently proposed by Dunlap et al. [
The MLSTs deduced from the genome sequences followed the clade sub-branches for
The text search of the RAST annotated function using the terms “chloramphenicol” and “aminoglycoside” identified two and three genes, respectively. A manual assessment of these genes and adjacent genes using nucleotide and amino acid BLAST revealed three genes to have some sequence similarity to genes known to confer resistance to the respective antibiotics. These putative resistance genes where found in all
We previously reported the presence of an N6-methyltransferase gene,
The sequences were compared for identity to known resistance genes in Genbank and for correlations to the MIC values of the strains (
Chloramphenicol is a bacteriostatic antibiotic whose activity is based on a reversible binding to the peptidyltransferase centre in the 50S ribosomal subunit of 70S ribosomes [
The molecular basis of bacterial resistance to chloramphenicol has been reviewed by Schwarz et al. [
We found a putative
Number in parentheses; Multi Locus Sequence Type. Clade A corresponds to
CAT type |
No. of strains | 0 SAPs | 1 SAP | 2 SAPs | 3 SAPs | Species | MIC |
---|---|---|---|---|---|---|---|
Variant 1 | 29 | 16 | 10 | 3 | 8(15), 16(12),32(2) | ||
Variant 2 | 21 | 21 | 4(20), 8(1) | ||||
Variant 3 | 28 | 18 | 10 | 8(1), 16(22), 32(5) | |||
Variant 4 | 23 | 14 | 6 | 3 | 4(1), 8(5),16(16), 32(1) |
SAP (single amino acid polymorphism); MIC (minimal inhibitory concentration); number in parentheses refer to the number of strains.
The MIC values were found to be at or above the cut-off value recommended by EFSA (8 mg/L)[
The sequence variants 1–4 follow the whole-genome core tree (
A comparison of the
The
There were no mobile genetic elements (IS elements, transposons, phages) in this region. The G+C content of the
Streptomycin belongs to the family of aminoglycoside antibiotics [
Two adjacent genes on the chromosome of all 104
Multiple sequence alignments of all AadK proteins and all APH proteins are shown in
AadK type |
No. of strains | 0 SAPs | 1 SAP | 2 SAPs | insert | MIC | Phylogeny |
---|---|---|---|---|---|---|---|
Variant 1 | 29 | 21 | 7 | 1 | 4(5),8(9),16(11),32(4) | ||
Variant 2 | 73 | 51 | 22 | 4(3), 8(18), 16(38), 32(13),64(1) |
SAP (single amino acid polymorphism); MIC (minimal inhibitory concentration); number in parentheses refers to number of strains.
All
The
Upstream there was also some variation in genome context. All
No mobile genetic elements (IS elements, transposons, phages) were found in this region. The G+C content of the
Erythromycin belongs to the macrolide class of antibiotics, which inhibit protein synthesis by stimulating dissociation of the peptidyl-tRNA molecule from the ribosome during elongation. This results in chain termination and a reversible stoppage of protein synthesis [
We found an
Number in parentheses; Multi Locus Sequence Type. Clade A corresponds to
The tree was built using approximately-maximum-likelihood algorithms (Fasttree, see
Variant 1.1 (red dot) was present in 18 strains of
The
Upstream of the
Genes in the
The presence of
Based on the findings in the present study and published results describing the
In 2002, the EU Scientific Committee on Animal Health issued an opinion which considered the use of a
An earlier study suggested that erythromycin resistance in
All genomes were analyzed with the tool HGtector [
EUCAST defines a microorganism as ‘wild-type’ (or innocuous) for a species based on the absence of acquired resistance to a drug in question; a microorganism is categorized as ‘wild-type’ by applying the appropriate cut-off value in a defined phenotypic test system [
Environmental microorganisms generally live in complex ecosystems, and to survive they have developed numerous defense systems. Biosynthetic pathways for natural antibiotics are ancient, and numerous mechanisms for antibiotic resistance and tolerance have evolved over the past millennia [
The comparison of the complete genome sequences of more than 100
Our analysis of the putative
We conclude that the putative
Generated with the “MLST for categorical data” template in the advanced cluster analysis module of BioNumerics 6.6 (Applied Maths, Biomerieux). Each node represents a sequence type with the type number next to the node. The size of the node is defined by the number of strains. Strains included in this study are coloured blue, the other strains (white) are taken from pubmlst.org/blicheniformis. Thick solid lines: sequence types differ in one allele; medium solid lines: sequence types differ in two alleles; thin solid lines: sequence types differ in three alleles; dashed lines: sequence types differ in four alleles. Sequence types that differ in more than four alleles are not connected. Partitions (grey area) are built from sequence types that differ in two alleles and less. The cluster on the left is
(TIF)
Whole genome phylogenetic tree reconstructed from the amino acid differences in the proteome of the core genomes of all 104
(PPTX)
Sequence alignments were made with Clustal Omega using default settings. The left column indicates the locus tag of each
(DOCX)
Cat proteins were aligned using Muscle [
(PPTX)
The chromosomal region surrounding the
(PPTX)
Sequence alignments were made with Clustal 2.1 using default settings. The left column indicates the locus tag of each
(DOCX)
Sequence alignments were made with Clustal Omega using default settings. The left column indicates the locus tag of each
(DOCX)
The chromosomal region surrounding the
(PPTX)
Sequence alignments were made with Clustal Omega using default settings. The left column indicates the locus tag of each
(DOCX)
The chromosomal region surrounding the
(PPTX)
Sequence alignments were made with Clustal 2.1 using default settings. Two examples of each
(DOCX)
Minimal inhibitory concentrations (MICs) are given in mg/L, ANI are given as ANIb values (AMIm gave a similar result). Footnotes: a, the year given is either the documented year of isolation or the earliest documented year of availability (e.g. year of deposition) indicated by <; b, listed as ST1 in the MLST database, but our analysis with sequence Genbank no. CP005965 gave ST4; c, strains listed in DSMZ catalogue as
(XLSX)
The sequencing statistics are shown for the 101 sequenced
(XLSX)
Clade A strains have unique gene clusters for lantipeptide biosynthesis, urea utilization, bacitracin biosynthesis, respiratory nitrate reductase, and non-ribosomal peptide biosynthesis. Clade B strains have unique gene clusters for lichenicidin biosynthesis and a prophage. For each gene, information is given for contig number, locus tag, start and stop nucleotide on the contig, and size of encoded protein in amino acids. Ortholog group (OG) numbers are given in column A. Color codes: light blue = pseudogene; yellow = not clear which gene fragment in assembly belongs in the cluster. Column DD shows the manually curated annotation.
(XLSX)
Separate worksheets show the order of genes surrounding the antibiotic resistance genes
(XLSX)
For each gene, information is given for contig number, locus tag, start and stop nucleotide on the contig, and size of encoded protein in amino acids. Ortholog group (OG) numbers are given in column A. Color code orange indicates putative HGT genes as predicted by HGTector [
(XLSX)
We thank Abdallah Albayasli, Peter Breum, Eva Kjer Jensen and Anna Minor for excellent technical assistance, and Mads Bennedsen and Anette Wind for useful discussions. Simon Cutting, Per Einar Granum, Lene Jespersen and Mark Turner are thanked for sending strains. This research was supported exclusively by internal funds. Most authors were employed by Chr. Hansen A/S, a company that produces strains for plant protection, animal and human health as well as for the food and dairy industry. Some of the authors are share-holders in Chr. Hansen A/S. This does not alter our adherence to PLOS ONE policies on sharing data and materials.