RESUMEN

Con el objetivo de describir las características genómicas relacionadas con la resistencia antimicrobiana y genómica comparativa de Escherichia coli diarreogénica (DEC), se sometieron a secuenciamiento genómico catorce aislamientos de DEC del banco de cepas del Instituto Nacional de Salud (INS). Las secuencias obtenidas se analizaron mediante procedimientos bioinformáticos a fin de buscar genes de resistencia microbiana y regiones genéticas relacionadas con patotipos y filogrupos. Se detectaron diversos determinantes de resistencia antimicrobiana, se destaca la producción de betalactamasas y mutaciones asociadas a la resistencia a quinolonas. Además, se observaron aislamientos de un mismo patotipo agrupados en distintos filogrupos. El análisis de genómica comparativa mostró un mayor número de genes ortólogos en aislamientos que pertenecían al mismo patotipo y filogrupo. Sobre la base de lo estudiado, los aislamientos de DEC en Lima, Perú, presentan resistencia a múltiples fármacos, y se detectaron varios patotipos y filogrupos con diversidad molecular y filogenética.

Palabras clave:
Escherichia coli; Resistencia a Medicamentos; Filogenia; Genómica

ABSTRACT

In order to describe the genomic characteristics related to antimicrobial resistance and comparative genomics of diarrheagenic Escherichia coli (DEC), 14 DEC isolates from the strain collection of the Instituto Nacional de Salud (INS) were subjected to genome sequencing. We used bioinformatic procedures to analyze the obtained sequences in order to look for microbial resistance genes and genetic regions related to pathotypes and phylogroups. Several antimicrobial resistance determinants were detected, but the production of beta-lactamases and mutations associated to quinolone resistance were the most relevant. Additionally, we observed isolates of the same pathotype grouped in different phylogroups. The comparative genomics analysis showed a greater number of orthologous genes in isolates from the same pathotype and phylogroup. In conclusion, DEC isolates from Lima, Peru, showed resistance to multiple drugs; likewise, molecular and phylogenetic diversity was observed in several pathotypes and phylogroups.

Keywords:
Escherichia coli; Drug resistance; Phylogeny; Genomics

INTRODUCTION

Bacterial resistance is a major global health problem associated with increased loss of economic productivity and human mortality. Approximately 23,000 people die annually in the United States due to infections caused by bacteria resistant to antibiotics ¹1. Marston HD, Dixon DM, Knisely JM, Palmore TN, Fauci AS. Antimicrobial Resistance. JAMA. 2016 20;316(11):1193-1204. doi: 10.1001/jama.2016.11764.
https://doi.org/10.1001/jama.2016.11764... , e.g., Escherichia coli, a bacterium intrinsically susceptible to almost all relevant antibiotics, with a high capacity to acquire resistance genes by horizontal gene transfer²2. Poirel L, Madec JY, Lupo A, Schink AK, Kieffer N, Nordmann P, et al. Antimicrobial Resistance in Escherichia coli. Microbiology Spectrum. 2018;6(4). doi: 10.1128/microbiolspec.ARBA-0026-2017.
https://doi.org/10.1128/microbiolspec.AR... .

The group of E. coli that cause intestinal and extra-intestinal infections are called diarrheagenic Escherichia coli (DEC). According to their pathogenesis, this group is classified into seven pathotypes: Enteropathogenic E. coli (EPEC), enterohemorrhagic E. coli (EHEC), enterotoxigenic E. coli (ETEC), enteroinvasive E. coli (EIEC), enteroaggregative E. coli (EAEC), diffusely adherent E. coli (DAEC) and adherent-invasive E. coli (AIEC), all of which are responsible for 40% of the cases of acute diarrhea in developing countries ³3. O'Ryan M, Prado V, Pickering LK. A millennium update on pediatric diarrheal illness in the developing world. Semin Pediatr Infect Dis. 2005;16(2):125-36. doi: 10.1053/j.spid.2005.12.008.
https://doi.org/10.1053/j.spid.2005.12.0... .

In Peru, bacterial resistance surveillance is mainly carried out by phenotypic methods, such as Kirby-Bauer or the minimum inhibitory concentration (MIC), recommended by the Clinical and Laboratory Standards Institute (CLSI). Most studies focus on the phenotypic detection of extended spectrum beta-lactamases (ESBL) and quinolone resistance, however, in recent years the study of resistance genes has been carried out by using molecular methods, such as PCR and Sanger sequencing ⁴4. Ochoa TJ, Ruiz J, Molina M, Valle LJ, Vargas M, Gil AI, et al. High Frequency of Antimicrobial Drug Resistance of Diarrheagenic Escherichia coli in Infants in Peru. Am J Trop Med Hyg. 2009;81(2):296-301. doi:10.4269/ajtmh.2009.81.296.
https://doi.org/10.4269/ajtmh.2009.81.29... .

In recent years, the use of whole genome sequencing has been employed to characterize virulence genes of different DECs and to identify phylogenetic relationships between pathotypes ⁵5. Noll LW, Worley JN, Yang X, Shridhar PB, Ludwig JB, Shi X, et al. Comparative genomics reveals differences in mobile virulence genes of Escherichia coli O103 pathotypes of bovine fecal origin. PLoS One. 2018;13(2):e0191362. doi: 10.1371/journal.pone.0191362.
https://doi.org/10.1371/journal.pone.019... . Since this aspect has been little explored in Peru, the aim of this research is to describe the characteristics of the resistome and comparative genomics of different DECs from the city of Lima.

THE STUDY

We worked on 14 E. coli isolates that caused acute diarrhea in patients from Lima between 2017 and 2018. The isolates were recovered by the Instituto Nacional de Salud (INS). The selection criteria were based on the antimicrobial sensitivity profile by diffusion disc previously evaluated by the laboratory (Appendix 1).

The isolates were reactivated in tryptic soybean broth (Merck, Germany) at 37 °C for 68 hours. Subsequently, they were seeded in MacConkey (Merck, Germany) agar plates and incubated at 37 °C for 18 to 24 hours. Confirmation of E. coli was made with conventional biochemical tests. The pathotypes were characterized based on primers according to Appendix 2.

DNeasy Blood & Tissue Kit (Qiagen, Germany) was used to extract DNA from the isolates. Spectrophotometry (Denovix, USA) and Qubit 3.0 fluorometer (Invitrogen, Malaysia) were used to evaluate DNA concentration and quality. Sequencing libraries were created with Nextera XT kit (Illumina, USA), and genomic sequencing with MiSeq equipment (Illumina, USA).

We evaluated the quality of the sequences using FastQC v0.11.5. Adapters and low-quality nitrogenous bases were removed with Trimmomatic v0.38 ⁶6. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics. 2014;30(15):2114-20. doi:10.1093/bioinformatics/btu170.
https://doi.org/10.1093/bioinformatics/b... . The sequences were assembled de novo using A5-miseq pipeline ⁷7. Coil D, Jospin G, Darling AE. A5-miseq: an updated pipeline to assemble microbial genomes from Illumina MiSeq data. Bioinformatics. 2015;31(4):587-9. doi: 10.1093/bioinformatics/btu661.
https://doi.org/10.1093/bioinformatics/b... . We used Kraken for gender identification and removal of contaminated contigs ⁸8. Wood DE, Salzberg SL. Kraken: ultrafast metagenomic sequence classification using exact alignments. Genome Biology. 2014;15(3):R46. doi: 10.1186/gb-2014-15-3-r46.
https://doi.org/10.1186/gb-2014-15-3-r46... . Allelic profiles and sequence types (ST) of the obtained genomes were assigned according to the Multilocus Sequence Typing (MLST) database for E. coli according to Achtman (http://pubmlst.org/) and using the MLST v2.10 program. The assignment of clone complexes and the creation of a minimum spanning tree (MST) was carried out using BioNumerics v7.5 (Applied Maths).

The BLAST tool was used to search for the genes associated with the virulence of E. coli pathotypes, according to Appendix 2. Prediction of coding sequences was made using the program Prodigal v2.6.3. The homologous genes of the sequences were identified from a gene library built using all the complete genomes available in GenBank, and the BLAST algorithm to select the alignments greater than 90% for identity and greater than 60% for coverage than the reference. The code used for the annotation is available at http://github.com/OrsonMM/Blast-score-for-genomics.

For the assignment of the phylogroups, we used Clermont typing scheme ⁹9. Beghain J, Bridier-Nahmias A, Le Nagard H, Denamur E, Clermont O. ClermonTyping: an easy-to-use and accurate in silico method for Escherichia genus strain phylotyping. Microb Genom. 2018;4(7):e000192. doi: 10.1099/mgen.0.000192.
https://doi.org/10.1099/mgen.0.000192... , based on an in silico PCR using primers designed for each phylogroup, with the software available at https://github.com/A-BN/Clermon-Typing. For the detection of antibiotic resistance genes, we used online database CARD ¹⁰10. Jia B, Raphenya AR, Alcock B, Waglechner N, Guo P, Tsang KK, et al. CARD 2017: expansion and model-centric curation of the comprehensive antibiotic resistance database. Nucleic Acids Res. 2017;45(D1):D566-73. doi: 10.1093/nar/gkw1004.
https://doi.org/10.1093/nar/gkw1004... , which includes chromosomal and plasmid genes. Finally, all the sequences obtained during the study were stored in GenBank database (Bioproject: PRJNA650130).

For comparative genomics, genomes assembled with Prokka v1.12 ¹¹11. Seemann T. Prokka: rapid prokaryotic genome annotation. Bioinformatics. 2014;30(14):2068-9. doi: 10.1093/bioinformatics/btu153.
https://doi.org/10.1093/bioinformatics/b... were scored with an e-value of 10^-9. The annotation results were uploaded to the BPGA v1.3 ¹²12. Chaudhari NM, Gupta VK, Dutta C. BPGA- an ultra-fast pan-genome analysis pipeline. Scientific Reports. 2016;6:24373. doi:10.1038/srep24373.
https://doi.org/10.1038/srep24373... program to calculate the number of genes in the central, accessory, and single genes with an identity value of >0.95. As a final step, the resulting matrix of pathotype annotation was used for inter-group comparative analysis using VennPainter ¹³13. Lin G, Chai J, Yuan S, Mai C, Cai L, Murphy RW, et al. VennPainter: A Tool for the Comparison and Identification of Candidate Genes Based on Venn Diagrams. PLoS ONE. 2016;11(4):e0154315. doi: 10.1371/journal.pone.0154315.
https://doi.org/10.1371/journal.pone.015... .

FINDINGS

All 14 isolates were reconfirmed as E. coli by conventional microbiological procedures. PCR confirmation of pathotypes resulted in two EHEC, two EIEC, three EPEC and seven ETEC (Table 1).

The sequencing error rate of 0.35% was calculated using a PhiX internal control. The Q30 used to select good quality readings was >80% per 150 bp. The fourteen genomes sequenced had an average of 184 high quality contigs. Each genome was identified as E. coli with more than 85% identity, obtaining a GC-content of 50.4%. Ten different STs were detected by MLST (Figure 2), the main findings were that ST11 was associated to EHEC, ST311 to EIEC and ST69 to EPEC.

The in silico pathotype analysis found a direct relationship with the results obtained by PCR (Figure 1). The EHEC isolates presented the Shiga toxin genes (stx1 and stx2). Four EHECs presented the heat-labile enterotoxin gene (elt), two presented heat-stable enterotoxins (st) and only one presented both genes. The annotation revealed the presence of the eaeA gene in the EPECs, and the IEECs carried the ipaH gene.

Clermont’s typing scheme assigned four E. coli genomes to phylogroup A, four to B1, one to B2, three to D, and two to E. Further review revealed that Peruvian EHEC isolates belonged to phylogroup E, EIECs to B1, ETECs to A and D, and EPECs to phylogroups B1, B2, and C (Figure 1).

During the resistome analysis we detected genes encoding beta-lactamases: nine genomes presented the ampC gene; four presented bla TEM1, and only one presented bla CTX-M-15. In addition, genes related to resistance to quinolones were detected. The S83L mutation of the gyrA gene was detected in two ETECs. The qnrS1 gene was found in all EIECs and one ETEC, while the qnrB10 gene was found in one EPEC (Figure 1).

The average number of coding sequences for the fourteen genomes analyzed was 5,016. The number of genes in the accessory genome varied between 938 and 2,133, while the number of unique genes varied between 0 and 581. Finally, the number of genes in the central genome was 2,950 (Table 1).

Pan-genomic functional analysis detected 2.17% of core genes, 2.04% of accessory genes and 1.63% of unique genes related to antimicrobial resistance using the categories of the Kyoto Encyclopedia of Genes and Genomes (KEGG) (Appendix 3). In addition, annotation using the database of Clusters of Orthologous Groups of proteins (COGs) categories detected that 5.32% of the core genes, 7.32% of the accessory genes, and 10.22% of the single genes were related to cell wall and membrane proteins (Appendix 4).

Comparative genomics between pathotypes revealed 2,950 genes from the central genome. Additional data on the grouping of two or three pathotypes are detailed in Figure 3. Finally, we detected a large number of ortho-genes between EHEC and EIEC isolates, as well as a small number of genes shared within EPEC and ETEC, for which there is no additional information on their name or function, being noted as hypothetical proteins.

DISCUSSION

Diarrheagenic E. coli is the leading cause of healthcare-related and community-acquired infections ¹⁴14. Oteo J, Lázaro E, de Abajo FJ, Baquero F, Campos J. Antimicrobial-resistant Invasive Escherichia coli, Spain. Emerg Infect Dis. 2005;11(4):546-53. doi: 10.3201/eid1104.040699.
https://doi.org/10.3201/eid1104.040699... and has a high capacity to develop antimicrobial resistance. It is important to report the genomic characteristics of pathogenic isolates to understand bacterial resistance and pathogenesis features that will contribute to monitoring and implementation of new policies related to local public health.

In this study, nine E. coli isolates with antibiotic resistance genes were detected by genomic sequencing. The most important ones were CTX-M-15, TEM1 and ampC, because they are related to beta-lactam resistance. Studies indicate the presence of bla genes in Peruvian ESBL-positive E. coli isolates, with high frequency rates of CTX-M (54.7%) and TEM (13.2%) ¹⁵15. García C, Astocondor L, Banda C. Enterobacterias productoras de beta-lactamasas de espectro extendido: Situación en América Latina y en el Perú. Acta Med Peru. 2012;29(3):163-9.. On the contrary, the detection of ampC-type beta-lactamase is little explored in our region, and usually only a low-frequency is phenotypically detected ¹⁶16. Galván F, Agapito J, Bravo N, Lagos J, Tamariz J. Caracterización fenotípica y molecular de Escherichia coli productoras de ß-Lactamasas de espectro extendido en pacientes ambulatorios de Lima, Perú. Rev Med Hered. 2016;27(1):22-9. doi:10.20453/rmh.v27i1.2780.
https://doi.org/10.20453/rmh.v27i1.2780... .

Resistance to quinolones by the S83L mutation in gyrA was observed in two ETCEs, while the presence of qnr in two EIECs, one EPEC and one ETEC. Quinolone resistance has already been reported in Peru, including commensal strains due to the strong influence of persistent exposure to these antimicrobials ¹⁷17. Pons MJ, Mosquito S, Ochoa TJ, Vargas M, Molina M, Lluque A, et al. Niveles de resistencia a quinolonas y otros antimicrobianos en cepas de Escherichia coli comensales en niños de la zona periurbana de Lima, Perú. Rev Peru Med Exp Salud Publica. 2012;29(1):82-6.; however, few studies focused on the use of molecular markers to determine quinolone resistance. Furthermore, genomic sequencing has led to the discovery of additional genes related to resistance to other antibiotics. For example, two EIECs and two EPECs with STX resistance and nine isolates with tetracycline resistance have been reported.

The analysis of the relationship between pathotypes (non-taxonomic) and phylogroups (taxonomic) of E. coli showed that the mostly reported EPECs were from phylogroups B1 and B2 ¹⁸18. Hazen TH, Daugherty SC, Shetty AC, Nataro JP, Rasko DA. Transcriptional Variation of Diverse Enteropathogenic Escherichia coli Isolates under Virulence-Inducing Conditions. mSystems. 2017;2(4):e00024-17. doi:10.1128/mSystems.00024-17.
https://doi.org/10.1128/mSystems.00024-1... , as discussed here. The EIEC are grouped in three phylogroups: A, E and B1 ¹⁴14. Oteo J, Lázaro E, de Abajo FJ, Baquero F, Campos J. Antimicrobial-resistant Invasive Escherichia coli, Spain. Emerg Infect Dis. 2005;11(4):546-53. doi: 10.3201/eid1104.040699.
https://doi.org/10.3201/eid1104.040699... , the isolates we used were from the phylogroup B1. On the other hand, the ETCEs are mostly from the A and D phylogroups ¹⁹19. Lu S, Jin D, Wu S, Yang J, Lan R, Bai X, et al. Insights into the evolution of pathogenicity of Escherichia coli from genomic analysis of intestinal E. coli of Marmota himalayana in Qinghai-Tibet plateau of China. Emerg Microbes Infect. 2016;5(12):e122. doi: 10.1038/emi.2016.122.
https://doi.org/10.1038/emi.2016.122... , the seven isolates evaluated were also from this phylogroups. In contrast, the EHECs were the most consistent group and were only associated with phylogroup E, and there is an evolutive explanation: the predominant serotype in this group, O157: H7, derives from the gene acquisition of a non-pathogenic isolate O55: H7, called preEHEC ²⁰20. Zhou Z, Li X, Liu B, Beutin L, Xu J, Ren Y, et al. Derivation of Escherichia coli O157:H7 from Its O55:H7 Precursor. PLoS One. 2010;5(1):e8700. doi:10.1371/journal.pone.0008700.
https://doi.org/10.1371/journal.pone.000... .

Pan-genomic analysis of diarrheagenic E. coli using KEGG categories allowed us to detect 2.17% of the genes (from the central genome) that are involved in antimicrobial resistance. Although the previous analysis only showed information regarding 14 genes associated with resistance, we were able to detect additional resistance-associated genes (among mutations and other effectors); however, due to the little information available regarding these genes in E. coli isolates, we did not study those molecules in depth.

The comparative analysis between pathotypes showed that EHEC and EIEC share a large number of genes (n = 246), while EPEC and ETEC had the lowest number of shared genes (n = 2). The latter result is interesting, since it contrasts with the work done by Hazen et al. ¹⁸18. Hazen TH, Daugherty SC, Shetty AC, Nataro JP, Rasko DA. Transcriptional Variation of Diverse Enteropathogenic Escherichia coli Isolates under Virulence-Inducing Conditions. mSystems. 2017;2(4):e00024-17. doi:10.1128/mSystems.00024-17.
https://doi.org/10.1128/mSystems.00024-1... , who found hybrid isolations of these pathotypes. On the contrary, most comparative genome studies are only based on the analysis of differences and similarities within the same pathotypes, so our research offers a first look at comparative genomes between different pathotypes and is one of the first studies on this subject in Peru.

In conclusion, the isolates of diarrheagenic E. coli from Lima, Peru, have resistance genes for several drugs. We detected pathotypes, ST and phylogroups with molecular and phylogenetic diversity. We suggest to continue monitoring the antimicrobial resistance of these bacteria with modern sequencing techniques, since this information is useful for local public health.

Acknowledgements:

To the members of Laboratorio Nacional de Referencia de Enteropatógenos/INS, especially to María Luz Zamudio, Gustavo Bellido and Ana María Meza for their technical assistance.

References

Publication Dates

History