Research note: Occurrence of mcr-encoded colistin resistance in Escherichia coli from pigs and pig farm workers in Vietnam

ABSTRACT WHO considers colistin as a highest priority critically important drug for human health, and occurrence of colistin-resistant bacteria in livestock is of health concern. The current study determined occurrence of colistin-resistant Escherichia coli in pigs and workers at pig farms in Vietnam, and investigated the genetic background for resistance. Colistin-resistant E. coli were detected from pigs in 53/116 (45.7%) farms, and from workers taking care of the pigs in 21/94 (22.3%) farms. Colistin-resistant isolates showed MIC to colistin between 4–16 mg/L, they were multidrug resistant (99%) and resistance was caused by the presence of mcr-1 genes in 97/102 (95.1%) E. coli from pigs and in 31/34 (91.1%) isolates from humans. mcr-1 is considered a plasmid-encoded gene, but this was not confirmed in the current investigation. In total, one pig isolate carried both mcr-1 and mcr-3 genes, whereas mcr-2, mcr-4 and mcr-5 genes were not detected. Shared resistance profiles between pig and human isolates on the same farm was only observed in four farms. The study showed that commensal E. coli from pigs in Vietnam constitute a reservoir for colistin-resitant E. coli, however, further studies are needed to confirm that mcr genes are associated with plasmids and their importance for human health.


INTRODUCTION
Antimicrobial resistance (AMR) is a global health threat (World Health Organization 2015), and both human use and livestock farming contribute to the AMR burden (Chereau et al. 2017).
Antimicrobials are commonly used in pig production in Vietnam for disease prevention, treatment of diseases and until recently also for growth promotion, leading to an increased risk of AMR (Dang et al. 2011;Lai et al. 2014). The World Health Organization (WHO) has raised a special concern over the use of highestpriority critically important antimicrobials (HP-CIAs) in livestock (World Health Organization 2017a), since several drugs belonging to this category are widely used for animal production, especially for pig production (Bloom et al. 2017).
Escherichia coli is a commensal bacteria of the gastrointestinal tract of animals and humans. It is widely used as an indicator organism to monitor AMR in livestock, food and people (Rasheed et al. 2014;Singh et al. 2018;Pormohammad, Nasiri and Azimi 2019). Most studies of AMR in livestock have been conducted in intensive-production systems in developed countries (Cho et al. 2011;Delannoy et al. 2017;Randall et al. 2018;Singh et al. 2018;Zhang et al. 2018;Yamamoto et al. 2019), and less data are available from semi-intensive systems prevalent in low and middle income countries, where farmers and veterinarians have limited knowledge on prudent antimicrobial use (AMU) practices. In Vietnam, studies on the prevalence of AMR among E. coli strains from livestock (Nguyen et al. 2015(Nguyen et al. , 2016Nhung et al. 2015) have not had a particular focus on HP-CIAs.
Colistin is a HP-CIAs antimicrobial, as it is a last resort drug for treatment of systemic infections with multi-drug resistant Gram-negative bacteria in humans (World Health Organization 2017a). Despite this, it is still anticipated commonly used in pig production in Vietnam, even for preventive measures. The emergence of mobile colistin resistance genes (mcr) on plasmids is of concern because of rapid spread of colistin resistance within and between bacterial populations. Three different mcr genes (mcr-1, mcr-2 and mcr-3) were originally described on plasmids in Enterobacteriaceae (Xavier et al. 2016;Hernández et al. 2017;Randall et al. 2018;Zhang et al. 2018), and later more genes, including the mcr-4 and mcr-5 genes have been added (Zhang et al. 2018;Ström Hallenberg et al. 2019).
The objective of this study was to determine the prevalence of colistin-resistant E. coli in pigs and workers at pig farms in Northern Vietnam and to determine whether mcr genes were present among E. coli from these sources.

Fecal sampling
Composite pig manure samples were collected at 116 randomly selected pig farms from four districts in Bac Ninh province, including 31 small-scale farms (≤ 20 pigs/farm), 66 middlescale farms (21-99 pigs/farm) and 19 large-scale farms (≥ 100 pigs/farm). The farms were selected based on lists provided by the provincial Department of Animal Health. Piglets raised in middle-scale and large-scale farms were mainly provided by one local large pig company, whereas small-scale farms kept sows and produced their own piglets. Consent was sought and obtained to collect faecal samples from workers at 94 of the pig farms. Briefly, 100-200 g faecal matter was collected with a spoon from the floor of different pens and sizes of pigs. A faecal sample was self-collected by one worker (approximately 20 g) during defecation using sterile gloves on the same day as the pigs were sampled. The samples were immediately placed in labeled sterile plastic bags, mixed by manual handling and transported to the laboratory, where they were analyzed the same day. The farm owner and farm workers gave consent orally and understood that they could withdraw from the study at any time and that they would be anonymous in the reporting of the results. The study protocol was approved by the Ethical Committee of the National Institute of Nutrition in Hanoi (Certificate No. 04/VDD-QLKH).

Detection of colistin resistance genes
Colistin resistance encoded by mcr-1, mcr-2, mcr-3, mcr-4 and mcr-5 genes was detected by multiplex PCR (Table 1; Rebelo et al. 2018). Each PCR reaction contained 12.5 μL DreamTaq Green PCR Master Mix 2X (Thermo-Fisher, Waltham, MA), five primer pairs (DNA-Diagnostic, Denmark), 5.5 μL nuclease-free water and 2 μL DNA template obtained by using a DNA isolation kit (Thermo-Fisher). The PCR reactions were carried out with the following amplification conditions: an initial denaturation step at 94 • C for 15 min, followed by 35 cycles of denaturation (94 • C for 30 s), annealing (58 • C for 90 s) and extension (72 • C at 60 s), followed by a final extension at 72 • C for 10 min on a PCR thermal cycler (Biorad, Hercules, CA). Positive strains were included for each gene ( Table 1). The same PCR reactions without template DNA was included as negative controls in all PCR reactions. The PCR products was analyzed in a 1.5% agarose gel using RedSafe Nucleic Acid Staining Solution (iNtRON Biotechnology, Kirkland, WA) as stain and visualized by UV illumination.   (Table 2). All E. coli isolates irrespective of their origin showed high resistance levels to sulfonamides (34/34; 100% of human isolates, 96/102; 94% of pig isolates), ampicillin (85%; 29/34 of human isolates, 96%; 98/102) of pig isolates), tetracycline and trimethoprim (more than 90% of isolates from both hosts), while none of the isolates were resistant to cefoxitin and amoxicillin-clavulanic acid. Low levels of resistance were observed to ceftiofur and ceftriaxone of human (1/34; 2.9%) and pig (4/102; 3.9%) isolates for both antimicrobials. All human colistin-resistant E. coli isolates and 98% (100/102) of pig isolates were MDR. A total of one pig isolate and two isolates from workers showed resistance to 10 antimicrobials. Due to the relative low number of isolates, 95% confidence intervals on the prevalence estimates were relatively high, and there were no significant differences between antimicrobial resistance prevalence for any of the drugs tested between isolates from pigs and farm workers. A total of 42 different phenotypic resistance profiles were observed among the colistinresistant E. coli strains (Table S1, Supporting Information). Of these, nine profiles encompassing 69 pig isolates and 22 strains from workers were shared between isolates from pigs and workers, however, shared profiles between pig and human isolates on the same farm was only observed in four cases. The most common profiles were AMP; TMP; TET; STR; SUL; COL (18 isolates) and AMP; GEN; TMP; TET; STR; NAL; SUL; CIP; COL (17 isolates). A total of two colistin-resistant E. coli, which were found to be cephalosporin-resistant by disc diffusion testing, were positive in the ESBL synergy test.

Prevalence of mcr genes
The mcr-1 gene was detected in 97/102 (95.1%) of E. coli from pigs and in 31/34 (91.1%) colistin-resistant isolates from humans. The remaining eight isolates did not carry mcr-1 genes, but did show phenotypic colistin resistance (MIC values were 4-8 mg/L) as well as resistance to several other antimicrobials including TMP, GEN, TET and SUL (Table S1, Supporting Information). One pig isolate carried both the mcr-1 and mcr-3 genes, whereas mcr-2, mcr-4 and mcr-5 genes were not detected (Table 3).

DISCUSSION
The current study has documented occurrence of colistinresistant E. coli in almost half of pig farms in the study area in Vietnam, and that humans working on the farms also commonly (almost one fourth of farms) harbor such bacteria. Previous studies have found highly variable (5-71.4%) carrier frequencies of colistin-resistant E. coli among randomly selected humans in different rural areas of Vietnam (Trung et al. 2017;Yamamoto 2018;Kawahara et al. 2019), and the carriage frequency among pig workers in the current study was well within this range. Previous studies have used growth at 2 mg/L colistin as initial selective criteria, while we have used 4 mg/L, and a direct comparison between studies is not possible. A study from Poland reported that the mcr-1.1 variant was mostly (66.3%) found in isolates with MIC 2 mg/L (Zaja c et al. 2019), suggesting that we may have underestimated the true prevalence of colistin resistance due to the mcr-1 gene. It would be interesting to carry out a comparison of farm workers and randomly picked human subjects using the same sampling and culture procedure in both cohorts. Results on prevalence of colistin resistance in E. coli seems to vary depending on concentration used for selection and the media used for isolation (Kawahara et al. 2019). We choose 4 mg/L as our previous studies on Danish pigs had shown that only a fraction of E. coli isolated on MacConkey plates with 2 mg/L were resistant in the broth microdilution assay. However, mcr genes turned out to be absent in that population of E. coli (unpublished). There is a need to agree on a standardized approach to increase the use of results for comparative purposes. According to Vietnamese regulations, both domestic and imported veterinary products can contain colistin (MARD 2016) and colistin is assumed commonly used by Vietnamese pig farmers. Investigations in two provinces suggested that 37% of pig farmers used colistin in 2008 (Hao, Tuat and Thao 2008) and 45% of pig farmers in another province used colistin in 2015 (Toan and Luu 2015). However, reliable data on amounts and drivers of colistin use by Vietnamese pig farmers, but also poultry farmers, are lacking and should be given priority in the on-going national efforts to establish monitoring programs of antimicrobial use in livestock. In our study, reliable data on colistin use was unfortunately not available for the pig farms and correlation between actual use and resistance levels at farm level can therefore not be estimated. Colistin has been recommended for treatment of pig diseases at the dose of 100 000 IU per kg weight for 3-5 consecutive days following recommendations by the European Medicines Agency (European Medicines Agency 2016). The price of colistin sold to Vietnamese farmers is low (9-11 USD/kg) compared to the price of other antimicrobials such as amoxicillin (17-20 USD/kg). In light of the high herd prevalence of colistin resistance observed in the current study, it should be considered to regulate the use of colistin in pig farming in Vietnam in order to adhere to the WHO recommendation that HP-CIAs should not be used in livestock production (World Health Organization 2017b).
The mcr-1 gene has spread globally in K. pneumoniae, E. coli and Salmonella spp. isolates of animal, environmental and human origin (Xavier et al. 2016;Dalmolin, De Lima-Morales and Barth 2018;Zhang et al. 2018;Ström Hallenberg et al. 2019;Yamamoto et al. 2019) mainly because it is typically located on transferable plasmids (Dalmolin, De Lima-Morales and Barth 2018;Randall et al. 2018). Colistin resistance in E. coli from pigs and workers in the current study was almost exclusively found to be associated with the presence of the mcr-1 gene, however, we cannot rule out that strains concurrently harbor chromosomal mutations conferring resistance towards colistin. Our results corroborate other recent studies in Vietnam. These have shown that colistin resistance was due to either mcr-1 and/or mcr-3 in 69/70 E. coli isolates (MIC>2 mg/L) from rural people (Yamamoto 2018), and mcr-1 in 60/62 colistin-resistant E. coli from meat and seafood (Yamaguchi et al. 2018), and it seems firmly established that the mcr-1 gene is common in colistin-resistant E. coli isolates in Vietnam. In this situation, the widespread use of colistin in pig industry is very worrying. A total of eight isolates showing phenotypically resistance to colistin did not contain any mcr genes in the PCR. Lack of mcr-1-5 genes in our strains showing phenotypic colistin resistance is possible. In such cases the resistance can be due to mutations into protein sequences of pmrA/B, phoP/Q and mgrB (Dagher et al. 2020) but also because of presence of novel mcr genes, e.g. mcr-10, not target by our PCR (Wang et al. 2020).
Colistin-resistant isolates were almost all shown to be MDR. The very high MDR frequency is in line with previous reports on E. coli from pig and humans in Vietnam. Thus, 86.7% (124/143) of E. coli from pigs in Vietnam have been reported to be MDR (Nhung et al. 2015). The resistance patterns observed did not differ significantly from resistance patterns observed in ESBL-producing E. coli and commensal E. coli in general (own unpublished results), and as such the resistance patterns described cannot be considered a particular trait of the colistinresistant E. coli studied. Unnecessary prescriptions and indiscriminate use of antimicrobials have been suggested as main reasons for the high prevalence of MDR in humans in Vietnam (Le 2019).
Studies have reported a high association of the mcr-1 gene with ESBLs genes of the CTX-M, SHV and TEM types (Dalmolin, De Lima-Morales and Barth 2018). The mcr-1 gene has also been identified in ESBL-producing and carbapenemase-producing E. coli from chicken or pig samples in Germany and China (Liu et al. 2016;Saras and Métayer 2016), and mcr-3 has been reported in ESBL-producing E. coli isolates from pork (Yamaguchi et al. 2018) and cattle, as well (Haenni et al. 2018). Multiple antimicrobial resistance genes including bla CTX-M , bla CMY , bla TEM , fosA, qnrS, floR and oqxAB have been detected in ESBL-producing E. coli from pigs in China alongside mcr-1 on IncHI2 plasmids. Studies have further characterized two IncHI2/ST3 plasmids with high similarity to the reference plasmid pHNSHP45-2 in E. coli from food-producing animals in China in which mcr-1, oqxAB and bla CTX-M were co-located (Li et al. 2016;Liu, Song and Zou 2019). The increased findings of mcr genes co-located with extendedspectrum-β-lactamase and carbapenemases-producing bacteria on different plasmids demonstrate a high diversity of plasmids acting as reservoirs for mcr genes. Such plasmids are often transmissible and represent a great threat to public health. Fortunately, in the current study, ESBL was only detected in two colistin-resistant isolates. Colistin is typically used to treat septicemia caused by ESBL-and carbapenemase-producing Gram negative bacteria, and combinations of resistances to these drugs have high clinical impact.
In conclusion, colistin-resistant E. coli was found on almost half of the studied pig farms in Vietnam and also commonly in farm workers, who had close contact to pigs. Colistin resistance was found mainly to be encoded by mcr-1 gene in both pig and human isolates. Further studies are needed to determine to what extent colistin-resistant E. coli from pigs are a source of colistin-resistant bacterial strains found in humans.

ETHICAL APPROVAL
Certificate of Ethical (No. 04/VDD-QLKH) approved by Committee of National Institute of Nutrition in Hanoi, Vietnam.

ACKNOWLEDGMENTS
We would like to thank Minh Thi Hong Nguyen, Vice Director of the Sub-Department of Animal Health in Bac Ninh province and her staff for support in sample collection. We also thank students at the Department of Veterinary Hygiene, National Institute of Veterinary Research, for technical assistance.

SUPPLEMENTARY DATA
Supplementary data are available at FEMSMC online.

FUNDING
The study was funded mainly by Danish International Development Assistance (Danida) through the VIDAPIG project: 'Health and Antibiotics in Vietnamese Pig Production' (DFC File number 17-M06-KU), Denmark. Co-funding was provided by the National Foundation for Science and Technology Development (Nafosted) through the project: 'Studying on biology and molecular characterizations of plasmid carrying extended-spectrum -β-lactamase (ESBL) gene of antibiotic resistant E. coli potential transmission between human and pig' (106-YS.02-2014.02), Vietnam.

Conflicts of interest.
None declared.