Abstract
Whole-genome sequencing analyses have provided important data and information on the repertoire of resistance genes in several bacterial species. This study examined to what extent the antimicrobial resistance genes found in a set of whole-genome-sequenced Salmonella Enteritidis strains from Brazil correlated with the phenotypic antimicrobial resistance possibly related to these genes. The genotypic resistance data from the strains studied were compared with publicly available data from strains isolated in other countries. The genotypic resistance profiles were accessed on the NCBI Pathogen Detection website, and the phenotypic resistance profiles were determined by the disk diffusion technique according to the Clinical and Laboratory Standards Institute guidelines. Fourteen of the 256 sequenced strains presented antimicrobial resistance genes, with the highest prevalence of resistance genes to aminoglycosides—with 16 genes detected in seven strains—not only in Brazilian strains but also in the strains from other parts of the world. The strongest correlation between phenotypic and genotypic resistance was found for tetracycline (75%). The genotypic and phenotypic profiles of the S. Enteritidis strains studied only partially matched, and they diverged in some antimicrobial classes more strongly than in other classes. The advances on whole-genome sequencing analyses associated with a better understanding of the correlation between phenotypic and genotypic resistance data may improve this powerful tool for antimicrobial resistance prediction.
Introduction
S
Salmonella infections in humans are often self-limiting and do not require antimicrobial treatment. Antimicrobial therapy is administered only when the disease is severe, invasive, or affects populations under risk conditions, such as children, older adults, and immunecompromised patients 5 ; in these cases, fluoroquinolones or expanded-spectrum cephalosporins are the most commonly used antimicrobials. 6
The overuse of antimicrobials in clinical practice and in food-producing animals worldwide has increased resistance of strains to antimicrobials and driven the emergence of multidrug-resistant strains in the past decades.4,6 The development of whole-genome sequencing tools has provided important data for surveillance and tracking of microorganisms, including tools to detect resistance genes in several bacterial species.7–9
In light of the importance of monitoring antimicrobial resistance in S. Enteritidis, an important pathogen that infects food-producing animals, this study aimed to (1) analyze the correlation between the genotype and phenotype of antimicrobial resistance in a set of whole-genome-sequenced S. Enteritidis from Brazil; and (2) compare the genotypic resistance data from Brazilian S. Enteritidis strains with all the sequenced and publicly available resistance data from S. Enteritidis strains isolated in other countries. The ultimate goal of this study was to provide data about the genotypic and phenotypic antimicrobial resistance profiles of the strains of this clinically important Salmonella serovar in many countries.
Materials and Methods
Bacterial strains
In a previous study, Campioni et al. 10 have sequenced the whole genomes of 256 S. Enteritidis strains isolated in Brazil from food, human, chicken, and the farm environment, between 1968 and 2016, and reported their accession numbers.
Antimicrobial resistance genotypes
The genotypic antimicrobial resistance profiles of the 256 S. Enteritidis strains isolated in Brazil and the strains isolated in other parts of the world were accessed on the NCBI Pathogen Detection website, which is publicly available. The analyses relied on the metadata available at the S. enterica database accessed on July 19, 2019. The search tool on the database was used to filter only the strains belonging to the serovar Enteritidis, including the 256 strains from Brazil (Supplementary Table S1). The presence of the resistance genes detected was analyzed according to the country, source, and year of isolation of the strains.
Antimicrobial resistance phenotypes
The phenotypic resistance of the strains bearing antimicrobial resistance genes identified by whole-genome sequencing analyses was determined by the disk diffusion technique, according to the Clinical and Laboratory Standards Institute assay protocol. 11 In brief, the S. Enteritidis strains stored at −80°C were grown in Luria Bertani media under shaking for 18–24 hr, at 37°C, and then cultured in Mueller-Hinton agar (Oxoid, Hampshire, United Kingdom) for further 18–24 hr, at 37°C. The strains were suspended in 0.8% saline solution until the optical density reached the McFarland scale of 0.5, and cultured in Mueller-Hinton agar plates in the presence of antibiotic disks placed on the surface of the media. The plates were incubated for 18–24 hr, at 37°C, and the growth inhibition halos were measured. The antimicrobials tested—amikacin (30 μg), cefalotin (30 μg), cefepime (30 μg), cefoxitin (30 μg), ceftriaxone (30 μg), gentamicin (30 μg), nalidixic acid (30 μg), phosphomicin (200 μg), sulfamycin (300 μg), tetracycline (30 μg), and trimethoprim (5 μg)—were provided by Oxoid, Hampshire, United Kingdom and selected according to the class of genotypic resistance found in the strains.
Results
Antimicrobial resistance genotypes of the Brazilian strains
Fourteen of the 256 S. Enteritidis strains isolated in Brazil and analyzed in this study carried at least one antimicrobial resistance gene, as identified by the tool available at the NCBI Pathogen Detection website (Table 1). Genes that promote resistance to antiseptics, intercalating dyes, and seven classes of antimicrobials (aminoglycosides, beta-lactams, bleomycin, folate pathway inhibitors, quinolones, sulfonamides, and tetracycline) were detected. The 14 strains bearing antimicrobial resistance genes were isolated from humans (09), food (02), and poultry (03) between 1989 and 2016 in Brazil (Table 1).
Genotypic and Phenotypic Comparison of the Antimicrobial Resistance in Salmonella Enteritidis Strains Isolated in Brazil
Antimicrobial resistance phenotypes of the Brazilian strains
The phenotypic resistance profiles of 13 of 14 S. Enteritidis strains isolated in Brazil and analyzed in this study are presented in Table 1. The strain SE1719 phenotype was not analyzed because bleomycin was not available to test (Table 1). The phenotype and genotype matched in only four of the 13 strains studied (Table 1); the other nine strains had at least one discrepancy between the phenotype and genotype of antimicrobial resistance (Table 1). The correlation between phenotypic and genotypic resistance profiles is presented in Table 2. Interestingly, six of the 13 strains studied exhibited sensitive phenotypes, even when they presented the genes that confer resistance to that drug (major error) (Table 1). Two strains presented resistance genes to antiseptics and intercalating dyes, which could not be tested herein (Table 1).
Analysis of the Phenotypic and Genotypic Discordances Among the 256 Salmonella Enteritidis Strains Studied from Brazil
Mutations in the DNA gyrase are not automatically identified by the database.
Detection of qnrB gene.
Obs, Observation.
In this study, 242 of the 256 S. Enteritidis strains had no antimicrobial resistance genes. The phenotypes of 186 of the 242 strains were reported by previous studies,12,13 while the phenotypes of the remaining 56 strains were characterized here as sensitive (24) or resistant (32) to quinolones (Supplementary Table S2).
Global analysis of the S. Enteritidis resistance genotypes in the NCBI Pathogen Detection database
Nearly 12.5% (26,054) of the Salmonella genomes available at the NCBI Pathogen Detection database belonged to the serovar Enteritidis, and were isolated between 1948 and 2019 in 73 countries (Supplementary Table S1 and Supplementary Table S3). Although the number of countries was considerable, it is important to mention that most of the strains recorded in the database were isolated in the United Kingdom (52%) and the United States (36%); only 367 (1.4%) S. Enteritidis strains were isolated in Brazil (Supplementary Table S1 and Supplementary Table S3).
In the set of 26,054 strains analyzed: (1) 20,726 (79.6%) were isolated from human clinical sources and 5,328 (20.4%) were isolated from nonclinical sources such as food, animals, and the environment (Supplementary Table S1); (2) 15,673 (60.1%) strains carried no resistance genes and 10,381 (39.8%) strains carried one to 22 resistance genes, which were detected in 9,191 (88.5%) and 1,190 (11.5%) strains isolated from clinical and nonclinical sources, respectively (Supplementary Table S1 and Supplementary Table S3).
The genes identified were related to the resistance to antiseptics, intercalating dyes, folate pathway inhibitors, aminoglycosides, ansamycin, beta-lactams, bleomycin, extended-spectrum cephalosporins, florfenicol, fosfomycin, lincosamides, macrolides, phenicol, quinolones, sulfonamides, and tetracyclines (Supplementary Table S1 and Fig. 1).
Classes of antimicrobial resistance detected in the Salmonella Enteritidis strains available at the NCBI Pathogen Detection website over the decades.
Discussion
A fast detection of antimicrobial resistance in bacteria is of great importance to allow the physicians to prescribe the correct antimicrobial. Whole-genome sequencing has usually provided high levels of concordance between phenotypic and genotypic resistance among different pathogens, including nontyphoidal Salmonella.8,9,14,15 This study analyzed the genes detected by the antimicrobial resistance genotype detection tool available at the NCBI Pathogen Detection website and examined their correlation with the phenotype of a set of S. Enteritidis strains isolated in Brazil that had their whole genome sequenced. 10
The tool detected antimicrobial resistance genes in only 14 of the 256 S. Enteritidis strains analyzed; each strain carried one to four genes (Table 1). However, the phenotypic and genotypic resistance matched in only four of the 13 strains that presented resistance genes; the phenotypic and genotypic resistance presented at least one discrepancy in the other nine strains analyzed (Table 1).
It is worth to note that the antimicrobial resistance phenotype from 188 of the 256 sequenced S. Enteritidis strains was reported by previous studies, with the prevalence of resistance to the quinolone nalidixic acid.12,13,16 Resistance to quinolones is mainly associated with point mutations in DNA gyrase—the antimicrobial target—, which is not automatically analyzed by the database tool but can be accessed by individual analysis of the sequence of these genes in the strains studied. 16 Such limitation may explain why only one strain presented a specific quinolone resistance gene, qnrB2 (Table 1), while the other strains presented point mutation in DNA gyrase. 16
McDermott et al. 8 have reported 99% of correlation between genotypic and phenotypic resistance to most antimicrobial classes, and ∼100% of correlation between these parameters for aminoglycosides and beta-lactams in nontyphoid Salmonella. This study detected the strongest correlation between genotypic and phenotypic resistance to tetracyclines (75%), sulfonamides (67%), and folate pathway inhibitors (40%) (Table 1).
The discrepancy between the genotypic and phenotypic antimicrobial resistance profiles can be due to the fact that the disk diffusion technique used in this study does not detect a reduced susceptibility to the drugs analyzed, which can be identified only by determining their minimal inhibitory concentrations. Another explanation is that a slight technical variation in the breakpoint of the phenotypic testing can falsely classify a strain as susceptible. As many resistance genes are plasmid encoded, they can be lost during the storage and subculture periods and due to the presence of silent genes that become transcriptionally active only in rare cases.8,17,18 The antimicrobial resistance data generated by whole-genome sequencing are valuable, but they are still not appropriate to guide clinical treatments due to some limitations of genome-based antimicrobial resistance prediction tools. For instance, the variety of data-generating platforms, some of which create reads shorter than the causative gene of reduced susceptibility; the required analysis of gene point mutations; the low accuracy of the culture-based antimicrobial sensitivity testing; and the incomplete knowledge of the genetic basis of antimicrobial resistance are some limitations to overcome.17,19,20
The comparative analysis of the genotypic resistance of the strains from this study and the sequenced strains from other parts of the world evidenced the worldwide prevalence of resistance genes to aminoglycosides since the past decades (Fig. 1). Aminoglycosides are one of the oldest antimicrobial classes, which explain the presence of resistance genes to them in S. Enteritidis strains isolated in different decades (Fig. 1).8,21 Streptomycin has been historically used in food-producing animals, which might explain the high prevalence of resistance to it in S. Enteritidis strains isolated from chickens. 8 The Brazilian S. Enteritidis strains reported in this study carried resistance genes to eight antimicrobial classes, while the strains from other countries deposited in the database expressed resistance genes to 14 antimicrobial classes (Supplementary Table S1). In both groups, the resistance genes were mainly identified in S. Enteritidis strains isolated from human clinical cases (Supplementary Table S1 and Table 1).
It is important to mention that this study analyzed the data available at the Salmonella database from the NCBI Pathogen Detection website in July 19, 2019, and that these data are subject to changes according to the amount of strains included each year in the database. The aforementioned website uses the National Database of Antibiotic Resistant Organisms, a collaborative cross-agency and curated database that includes new antimicrobial resistance gene sequences when they are detected and received by the curators. Thus, it is a dynamic database that changes when new strains and new resistance gene sequences are identified.
In conclusion, the genotypic profiles of the S. Enteritidis strains studied only partially matched the phenotypic profiles of antimicrobial resistance, with discrepancy in some antimicrobial classes. The advances on whole-genome sequencing analyses associated with comprehension of the correlation between antimicrobial resistance phenotype and genotype may improve this fast and important antimicrobial resistance prediction tool.
Footnotes
Disclosure Statement
No competing financial interests exist.
Funding Information
Dr. Fabio Campioni received postdoctoral fellowships from the São Paulo Research Foundation (FAPESP, Grant No. 2013/25191–3) and from the National Postdoctoral Program of CAPES (PNPD/CAPES Grant No. 88882.317662/2019–01), while Dr. Juliana Pfrimer Falcão received a research grant from FAPESP (Grant No. 2016/24716–3) and a research productivity fellowship from the National Council for Scientific and Technological Development (CNPQ; Grant No. 304399/2018–3) to develop this work.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
