Detection and Characterization of ESBL-Producing Escherichia coli From Humans and Poultry in Ghana

1The inappropriate use of antibiotics, not only in human medicine but also in animal husbandry, has been considered a main driver leading to the increase of multidrug-resistant bacteria. Consequently, in Europe, the European Food Safety Authority (EFSA) and the European Medicines Agency (EMA) recently advocated measures to reduce use of antimicrobial agents in animal husbandry in the European Union. Food-producing animals, especially poultry, have been suggested as a potential source for transmission of extended-spectrum beta-lactamase (ESBL)-producing bacteria to humans, either by direct contact or consumption of contaminated meat products, leading to the colonization of the intestinal tract and eventually to severe infections.

2A number of previous studies on the inter-sectoral transmission of ESBL-producing bacteria did only use methods with low discriminatory power, which target a small number of genes (e.g., ESBL genotyping, multilocus sequence typing), and therefore may suggest, but not prove clonal transmission events. As an example, a Dutch study could not confirm previously suggested clonal transmission of ESBL-producing E. coli isolates between humans and poultry, when using the same collection of isolates but changing the typing method to a whole genome sequencing (WGS) approach. Instead, results rather demonstrated that antibiotic resistance between different reservoirs is likely to be transmitted by the spread of plasmids. Similarly, a Swedish study estimates that less than 0.1% of the Swedish population carry poultry-associated ESBL-producing isolates, but 5% are colonized with ESBL-encoding plasmids identical to those found in isolates from chicken meat and poultry.

3Those findings may not be directly transferable to developing countries. In these regions, the inappropriate use of veterinary antibiotics, which are often readily sold in shops and markets without prescriptions, is considered to be very high, potentially selecting for antibiotic resistance. Nevertheless, a recent review found a lower prevalence of ESBL/pAmpC E. coli among poultry meat products in African countries (average 16.3%) compared to reports from many European countries such as Spain (84–93%) and the Netherlands (77%). In comparison to industrialized countries, it can be assumed that inter-host transmission is more likely to happen in rural areas of sub-Saharan Africa with mainly subsistence-based agricultural communities, where people frequently live in close contact with livestock animals. However, bacterial transmission among poultry and humans has not been adequately addressed in this region until now.

4This study aims to compare ESBL-producing E. coli found in the intestinal tract of humans and poultry using highly discriminatory WGS methods in order to assess potential transmission routes in a rural community of Central Ghana.

5The study was conducted in Agogo, a town with 32,000 inhabitants situated in the Asante Akim North District within the Ashanti Region of Ghana. Between January and June 2015, all children below 15 years of age living in Agogo and nearby surroundings, who were admitted to the Agogo Presbyterian Hospital, were recruited into the study. If available, a stool sample was collected at the day of admission and transported to the hospital laboratory within 2–4 h.

6A total of eight poultry farms, with around 250–1,000 chickens on each farm, are situated within and around the town of Agogo. The infrastructure with wooden shelters and open sides is comparable on all farms. Between February and April 2015, all eight farms were visited once and six farms were visited a second time between May and June 2015. On each farm visit, 10 single fecal droppings were collected with an eSwabTM (Hain Lifescience, Nehren, Germany) and transported to the laboratory within 2–4 h.

7The Committee on Human Research, Publications and Ethics, School of Medical Science, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana and the Ethics Committee of the Ärztekammer Hamburg, Germany approved the study design and the informed consent procedures. All participants were informed about the study’s purpose and procedures in accordance with the Declaration of Helsinki. Written informed consent was obtained from the parents or the guardian on behalf of the study children prior to enrolment.

8On arrival in the laboratory, each fecal sample was directly inoculated on two selective MacConkey agar plates containing 1 mg/L ceftazidime and 1 mg/L cefotaxime, respectively. Plates were incubated at 37°C for 24–48 h in normal atmosphere. All colonies with typical E. coli morphology were selected and confirmed biochemically as E. coli by API 20E tests (bioMérieux, Marcy L’Etoile, France). For all E. coli, ESBL-production was confirmed by the combined disk test with cefotaxime and ceftazidime alone and in combination with clavulanic acid (Becton, Dickinson and Company, Sparks, MD, United States) as described before by the EUCAST [EUCAST guideline on detection of resistance mechanisms v 1.0 (2013-12-11)]. Quality control for each batch of cephalosporin-containing MacConkey agar plates was performed with E. coli ATCC 25922 and a blaCTX-M-15 positive E. coli isolate.

9E. coli isolates were stored at -80°C in cryobank tubes (Pro-lab Diagnostics, Richmond Hill, ON, Canada) and shipped on dry ice to Germany, where WGS was performed for all ESBL-producing E. coli isolates. In brief, whole genome DNA was isolated using the Purelink Genome DNA Mini kit (Invitrogen, Darmstadt, Germany) according to the manufacturer’s instruction. Sequencing was carried out using an Illumina Nextera XT library with 2 × 300 bp paired-end reads on an Illumina MiSeq instrument (Illumina, San Diego, CA, United States) achieving an average of 2,170,898 reads per isolate (61× average coverage). The raw data was assembled using SPAdes (version 3.0). Multilocus sequence type (ST) determination was performed using MLST 1.0. Antibiotic resistance genes were identified using ResFinder and the genetic location of the ESBL-genes was investigated using the ESBL-gene containing contigs by blastn using the nt database. Contigs depicting hits only to chromosomal sequences were considered as chromosomal hits, while those depicting hits to plasmids were considered as plasmid hits. Contigs depicting hits to both the chromosome and plasmids were considered as “ambiguous.”

10The single nucleotide polymorphisms (SNPs) were called by SAMtools. The tree was generated using FastTree 2 and drawn by Evolview.

11To analyze possible transmission, isolates depicting an identical ST and present in both reservoirs were analyzed in more detail. SNPs were called as mentioned above, and regions of high recombination were excluded manually.

12The raw sequencing data are available at the European Nucleotide Archive (ENA) under the project accession number PRJEB26592.

13Between January and June 2015, 54 hospitalized children, with a median age of 20 months (IQR: 11–36 months), provided a stool sample at admission, regardless of the admission diagnosis. All children had contact to free-roaming chickens in town, but had no contact or access to the poultry farms. Chicken meat is part of the regular diet. Of all children, 33 (61%) harbored an ESBL-producing E. coli. From one child, two different E. coli were isolated, leading to a total of 34 ESBL-producing E. coli.

14During the same time-period, fecal samples were obtained from 140 broilers, of which 41 (29%) were positive for an ESBL-producing E. coli. Prevalences differed widely between farms, ranging from 0 to 85% (Figure 1). Four broilers were detected with two different E. coli isolates, which results in 45 ESBL-producing E. coli isolates from broilers.

15The 34 human isolates displayed 22 different sequence types (ST). ST-131 (12%; n = 4), ST-167 (9%; n = 3) and ST-617 (9%; n = 3) were most frequently detected. No clusters with more than two isolates with the same ST during a 1-month period were identified (Table 1).

16Among all broiler isolates (n = 45), 13 different STs were identified, with ST-10 being most prevalent (69%; n = 31), although ST-10 has been only detected on three farms. On farm 7 and 8, ST-10 comprised 90% (n = 9) and 100% (n = 21) of all ESBL-producing isolates, respectively (Figure 1). There was no reported exchange of broilers, feed or personnel between those two farms. All other STs were not detected more than twice on each farm.

17The overlap of sequence types between the two different sources (human and broiler) involved five sequence types (ST-38, ST-58, ST-3018, ST-3541, ST-6359), comprising seven broiler and seven human isolates (Table 1).

18The most common ESBL genes in E. coli from both populations were blaCTX-M-15 (n = 76) followed by blaSHV -12 (n = 2) and blaCTX-M-14 (n = 1). The ESBL gene blaCTX-M-15 was equally distributed in both populations, whereas blaSHV -12 was exclusively found in broiler and blaCTX-M-14 in human isolates, respectively (Table 2). The location of the ESBL genes differed between human and broiler isolates. Among human isolates, 29% (n = 10) harbored plasmid and 44% (n = 15) chromosomally inserted ESBL genes, while the exact location could not be identified among the remaining 26% (n = 9). Among broiler isolates, 82% (n = 36) harbored a chromosomally inserted ESBL gene (Table 2). Of note, all but one (C105F) of the 31 ST-10 isolates, which were isolated from three different farms, harbored blaCTX-M-15 at an identical chromosomal location (Supplementary Table 1), indicating the spread of a distinct clone on these farms.

19Five human/broiler isolate clusters depicting an identical ST were regarded as possible clonal spread events and therefore investigated for SNPs (Figure 2). The ST-3541 isolates (C28F/C071F/701519) revealed a difference in 49 and 24 SNPs, respectively. In all three isolates, the blaCTX-M-15 gene was located on the chromosome at an identical position, thus indicating a common origin of these isolates.

20An almost identical situation was found in the ST-6359 isolates (C132F/701861/701867). The human isolate 701861 differed by 99 SNPs from C132F, whereas 701867 differed by 17 SNPs, implying close genetic relation to the 701867 isolate. Interestingly, all ST-6359 isolates harbored blaCTX-M-15 at an identical chromosomal location, implying a common ancestor of these isolates. Similarly, the ST-3018 pair (C50F/701698) differed by only 21 SNPs and harbored blaCTX-M-15 on a plasmid.

21A complete different pattern was detected in the ST-58 (C29F/701805) and ST-38 isolates (C060/C081F/701495-1/701611). ST-58 isolates differed by 4449 SNPs, thereby indicating that these are two genetically not related isolates and no transmission occurred. ST-38 isolates differed already in the chicken isolates (2034 SNPs), while the human isolates depicted 9588 (701495-1) and 9560 (701611) SNPs, respectively. This indicates that ST-38 isolates are highly diverse.

22The study demonstrates genetic links of ESBL-producing E. coli that may indicate transmission between the poultry and human population in rural Ghana. With four human and four broiler isolates being closely related, 10% of all ESBL-producing isolates in this study may have been transmitted at a certain point in time between the two different populations.

23The potential risk for the transmission of antimicrobial-resistant bacteria between animals or animal food products and humans in sub-Saharan Africa has been highlighted before, however, from this region no data on the clonal spread between human and animal reservoirs has been published. Other studies using MLST-based typing methods or ESBL-genotyping do not allow to conclusively interpret the clonal spread of ESBL-producing E. coli. More recently, a Dutch study using WGS methods concluded that clonal transfer between humans and poultry of antibiotic resistant E. coli occurs less frequently than the transfer of resistance plasmids between the two reservoirs. Similarly, studies from Romania, Sweden and the United Kingdom found evidence for the transmission of ESBL-harboring plasmids instead of clonal transmission of isolates between poultry and humans.

24In contrast, the present results suggest clonal transmission between the two reservoirs in a rural Ghanaian town, where children were sampled at hospital admission, independent from any contact to the poultry farms to which children have no access. Taking into account the low level of sanitary and hygiene conditions in rural Ghana, inter-host transmission of bacteria is expected to be more predominant compared to industrialized countries and not necessarily limited to farm workers. Furthermore, transmission via consumption of meat products has been suggested as a potential source of multidrug resistant bacteria in Africa.

25The high rate of ESBL-producing E. coli on farms is in sharp contrast to previous Ghanaian data from 2009, where no ESBL-producing E. coli isolates (n = 103) were found in broilers from Accra. However, this previous study did not use any ESBL screening plates for the detection of E. coli, which might underestimate the ESBL production.

26Notably, a high diversity of STs has been observed, with only 5 out of 30 STs overlapping between the two sampling groups. ST-10 clearly predominates within two farms. Indeed, ST-10 ESBL-producing E. coli have been frequently identified worldwide in the chicken production chain, including reports from Nigeria and Tanzania. Interestingly, studies from the Netherlands, Sweden and Vietnam detected ESBL-producing ST-10 E. coli not only in chickens, but also in high numbers among humans. However, in the present study ST-10 could not be detected among children.

27In this study, ST-131, although not very frequent (n = 4), is the most prevalent E. coli subtype among children (12%). ST-131 E. coli have been described as an clinically relevant multidrug-resistant ST, frequently harboring blaCTX-M-15 and fluoroquinolone resistance. High virulence leading to serious infections, such as invasive bloodstream, urinary tract and intra-abdominal infections have been attributed to ST131, also in Ghana. Although identified in different animal reservoirs, the ST131 lineage seems to be specifically adapted to the human host and reservoirs to be rather unclear. It has been suggested that due to acquired virulence factors ST131 is responsible for the global increase of E. coli harboring blaCTX-M-15.

28The CTX-M-15 β-lactamase is the most prevalent ESBL type among chicken (96%) and human (97%) isolates in this study. While in many countries CTX-M-15 is one of the most frequent ESBL types in ESBL-producing bacteria, causing human infections, it showed a relatively low prevalence among bacteria isolated from fecal poultry samples in previous studies using ESBL screening plates as an isolation method in the Netherlands (0%), Belgium (2%), and the United Kingdom (12%). Studies from Japan and China also demonstrate low prevalences (0–2%) of CTX-M-15 in fecal poultry samples, in contrast to most surveys of African livestock, which show a predominance of blaCTX-M-15 positive ESBL-producing E. coli in a recent review by. This geographical distribution suggests that transmission between humans and poultry may occur more frequently on the African continent.

29The present study has some limitations. Firstly, it does not allow any conclusions on the transmission of plasmids, but only indicates clonal bacterial transmission between poultry and humans. Therefore, the focus on clonal transfer alone will probably underestimate the transmission of resistance genes. In addition, the collection of stool samples at hospital admission may result in a higher ESBL prevalence among children compared to the general population, due to a different health care seeking behavior and potential previous antibiotic prescriptions.

30Apart from a high intestinal detection frequency, this study shows that highly similar ESBL-producing E. coli strains circulate among the human and poultry populations within a small town in rural Ghana. Hence, poultry farms or meat products might be an important source for ESBL-producing bacteria in rural Ghana leading to difficult-to-treat infections in humans. Despite widespread agreement that integration of human and animal data is desirable for antimicrobial resistance surveillance, there is very little, if any, integration of data in most low and middle-income countries. An integrated “One Health” surveillance system would be able to monitor transmission events and detect resistant bacteria in a timely manner from both sectors.

31The raw sequencing data are available at the European Nucleotide Archive (ENA) under the project accession number PRJEB26592. All other raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.

32LF, CI, SP, JM, and DE designed and coordinated the study. KO, BH, VL, NS, and EO-D conducted and supervised fieldwork. KO and CA conducted laboratory work. RK and OS performed the epidemiological and bioinformatics analysis. DE and LF wrote the first draft of the paper. All authors read and approved the final manuscript.

33The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.