Background

Escherichia coli, a non-sporulating, Gram-negative, rod-shaped aerobe, is a commensal in the intestines of vertebrates (Duriez et al. 2001; Escobar-Páramo et al. 2006; Jaureguy et al. 2008; Tenaillon et al. 2010; Schmidt et al. 2015), present in various habitats ranging from deserts to the arctic (Escobar-Páramo et al. 2006). This bacteria species also thrives in water environments, with half of the E. coli population estimated to reside in these secondary habitats (Tenaillon et al. 2010), facilitating anthropogenic transmission from species to species through water and food ingestion (Escobar-Páramo et al. 2006). These characteristics suggest that E. coli is capable of colonizing and spreading in different environments (Escobar-Páramo et al. 2006). Pathogenic strains of E. coli can cause serious infections including, intestinal and extraintestinal diseases in humans (Tenaillon et al. 2010). Previous studies have suggested that commensal strains of E. coli may be the precursor to the pathogenic E. coli strains following acquisition of virulence genes (Duriez et al. 2001), potentially implicating commensal E. coli as the natural reservoir of pathogenic strains (Schmidt et al. 2015). Enhancement of pathogenicity through random point mutations or acquisition of chromosomal virulence operons were proposed to be methods by which commensal E. coli can become pathogenic (Duriez et al. 2001), and this was possible with E. coli having a clonal structure (Tenaillon et al. 2010; Smati et al. 2015) coupled with a high level of genome plasticity.

Phylogenetic analyses have greatly aided our understanding of the genetic structure within the E. coli species by the assignment of at least seven phylogroups belonging to E. coli sensu stricto (A, B1, B2, C, D, E and F) (Clermont et al. 2013) and the multi locus sequence typing (Jaureguy et al. 2008). Serotyping via molecular typing of the O-antigens could also assist in the study of the genetic structure of E. coli (Sánchez et al. 2015). These methods are powerful tools for determining the phylogenetic relationships among the different E. coli lineages and could possibly provide greater understanding of the factors that shape the genetic structure of E. coli (Jaureguy et al. 2008; Clermont et al. 2013). However, the vast majority of research focus have been biased towards human E. coli strains. Laboratory rodents have been pivotal as a model for the study of the complex physiological system of mammals. They have contributed immensely to our understanding of the various human diseases, enhancing our knowledge to find effective treatments for these diseases (Jacoby and Lindsey 1998). However, these rodents are also inherently susceptible to numerous known bacterial, fungal and viral diseases (Jacoby and Lindsey 1998). Some of these infectious agents cause subclinical and unnoticeable infections that could nevertheless, adversely affect biological experimental data (Jacoby and Lindsey 1998; Loong et al. 2016).

Animal husbandry practices can potentially alter the composition of laboratory rodent microbiota, affecting experimental outcomes by introducing variables into the animals (Ma et al. 2012). The gastrointestinal microbiota are associated with the regulation of innate immunity and pathogen recognition receptors (Lee et al. 2011). Accordingly, the microbiota has been shown to modulate pro- and anti-inflammatory immune responses in mice models of multiple sclerosis (Lee et al. 2011) and type-1 diabetes (Burrows et al. 2015), suggesting that perturbation of the fragile equilibrium within the gastrointestinal tract impacts the pathologies of mice disease models. Upsetting of the normal microbiota by dysbiosis, abolishes colonization resistance, disrupts homeostasis and increases the host’s susceptibility to diseases (Ma et al. 2012). The successful mice intestine colonization by Campylobacter jejuni after antibiotics elimination of the gastrointestinal microbiota (O’Laughlin et al. 2015) demonstrated the effect of the abolishment of colonization resistance. A recent survey in 21 different German animal facilities found that shed skin or dust particles carried by animals, care takers or scientists influenced the gastrointestinal microbiota of experimental mice (Rausch et al. 2016), potentially affecting experimental outcomes. Fluctuations in the laboratory rodent microbiota as a result of external stimuli has to be reduced or eliminated to ensure consistency and reproducibility of experimental results. Therefore, surveillance of commensal E. coli populations in laboratory rodents could provide an indication of possible alteration of the normal gastrointestinal microbiota.

As E. coli is widely used as an indicator of antimicrobial selection pressure (Grønvold et al. 2010) and newly introduced bacteria species may transfer antimicrobial resistant determinants to commensal species (Schmidt et al. 2015), assessment of the genetic characteristics of commensal E. coli would provide information on the virulence status in the animals housed in the animal experimental facility. The occurrence of horizontal gene transfer of beta-lactamases among E. coli (Jaureguy et al. 2008; Koga et al. 2015) and the reverse transfer of antimicrobial resistance from humans to animals (Grall et al. 2015) could transpire as a result of the close interaction between scientists (humans) and laboratory rodents (animals). Hence, determination of antimicrobial resistance at the genetic level is important for understanding and limit antimicrobial resistance. The aims of this study were to determine the genetic structure of E. coli strains from the organs of otherwise healthy laboratory rodents to establish a background for further studies on pathogenic strains and to analyze the antimicrobial resistance profiles of the isolated E.coli strains.

Methods

Isolation of E. coli

A total of 67 randomly selected healthy rodents (mice; n = 38 and rats; n = 29) aged between 6 and 12 weeks old, housed in the Animal Experimental Unit, Faculty of Medicine, University of Malaya during March 2014–April 2015, were assessed as part of the unit’s rodents health monitoring program (Table 1). The sex of the rodents were distributed quite proportionately between males (n = 32) and females (n = 35) (Table 1). Healthy mice were euthanized and serologically assayed for selected pathogens according to previously published protocols (Loong et al. 2016). The rats however, were anaesthetized with Ketamine (50 mg/kg) and Xylazine (5 mg/kg) prior to blood collection through cardiac puncture. The collected rat blood was tested by serology (SMART-SPOT Sentinel Panel Test, Rat Sera, Biotech Trading Partners, USA) for the following agents; Kilham rat virus, Mycoplasma pulmonis, pneumonia virus of mice, rat coronavirus, respiratory enteric orphan virus, rat parvovirus, Sendai virus and Theiler’s murine encephalomyelitis virus. Following that, the rats were euthanized by overdosing with the same drugs at three times the anaesthetic dosage. Rodents were housed in open cages and other detailed procedures were described in a previous study (Loong et al. 2016). Selected rodent organs (gastrointestinal tract, liver and, the trachea and lung) were harvested, homogenized and cultured onto MacConkey agar. Organs from every rodent were cultured onto individual agar plate. Cultured bacterial strains suspected to be E. coli were identified using Gram staining, microscopy and 16S rDNA sequencing (Misbah et al. 2005). One strain was selected from each suspected culture plates after 24 h incubation at 37 °C for this study.

Table 1 Breed of rodents assessed between March 2014 and April 2015
Full size table

Multi locus sequence typing (MLST) and phylogroup classification

Commensal E. coli strains were subjected to MLST (referred to as the Pasteur scheme), where the internal portions of eight housekeeping genes (dinB, icdA, pabB, polB, putP, trpA, trpB and uidA) were sequenced and compared with the database at http://bigsdb.web.pasteur.fr/ecoli/ecoli.html (Jaureguy et al. 2008). A specific allele number was assigned to each distinct sequence within a locus and a specific sequence type (ST) was assigned to each distinct combination of alleles. A novel allele number was assigned to sequences that could not be found in the database, subsequently establishing a novel ST. For phylogroup classification, E. coli strains were designated into the phylogroups; A, B1, B2, C, D, E, or F, according to a previous study (Clermont et al. 2013). This quadruplex PCR method differentiates the E. coli phylogroups by nucleic acid amplification of the arpA, chuA and yjaA genes, and the DNA fragment, TspE4.C2.

Identification of Shiga toxin-producing E. coli serotypes

E. coli serogroups O5, O15, O26, O45, O55, O76, O91, O103, O104, O111, O113, O118, O121, O123, O128, O145, O146, O157, O165, O172 and O177 are the O-antigen types of the most clinically significant Shiga toxin-producing E. coli serotypes. These 21 serotypes were identified using three serotype-specific multiplex PCR assays, and it can also detect the respective O-antigens even when they are not expressed by the E. coli strains (Sánchez et al. 2015).

Extended spectrum beta lactamases (ESBL) phenotypic confirmatory method

All commensal E. coli strains were cultured onto Mueller–Hinton agar after adjustment to 0.5 McFarland standards. Antimicrobial discs containing cefotaxime (30 μg), cefotaxime/clavulanic acid (30/10 μg), ceftazidime (30 μg) and ceftazidime/clavulanic acid (30/10 μg) were placed onto the agar plate so that the distances between them are approximately twice the radius of the inhibition zones, and the plate was incubated overnight at 35 °C. ESBL activity was indicated when a ≥5 mm increase in the inhibition zone diameter around either cephalosporin disc in combination with clavulanic acid as compared to the inhibition zone diameter of the cephalosporin when tested alone (CLSI 2012). The E. coli strain ATCC 25922 and Klebsiella pneumoniae strain ATCC 700603 were used as controls as per CLSI recommendation.

Detection of genes encoding for ESBL

Five multiplex and one singleplex PCRs were employed in this study to detect genes encoding important beta-lactamases; one CTX-M multiplex PCR including groups 1, 2, 8, 9 and 25 (Woodford et al. 2006); one TEM/SHV/OXA-1-like multiplex PCR (Dallenne et al. 2010); one plasmid-mediated AmpC gene multiplex PCR including ACC, FOX, MOX, DHA, CIT and EBC (Dallenne et al. 2010); one VEB/GES/PER multiplex PCR (Dallenne et al. 2010); one VIM/IMP/KPC multiplex PCR (Dallenne et al. 2010); and one OXA-48-like singleplex PCR (Dallenne et al. 2010). In addition, the New Delhi metallo-beta-lactamase, NDM-1 was detected using PCR as previously described (Peirano et al. 2011).

Detection of E. coli from the animal housing environment, rodent feed and drinking water

Testing for the presence of E. coli in the animal housing environment was performed by taking swabs from the tabletop, tap, trolley and wash basin in the room housing the rodents. Sterile, irradiated rodent feed (#1320-Maintenance diet for rats and mice, Altromin International, Germany) was vortexed in tubes containing nuclease-free water. Following that, the swabs, the vortexed feed and the autoclaved drinking water were cultured onto MacConkey agar and incubated at 37 °C under aerobic condition for 24 h. One colony resembling the morphological appearance of E. coli was selected from each positive plate and the E. coli phylogroup was determined. Additionally, detection of genes encoding for ESBL was also performed.

Results

In the study period of 14 months (March 2014–April 2015), we randomly selected and examined a total of 67 healthy rodents of various breeds, housed in the AAALAC-certified facility at the Animal Experimental Unit, University of Malaya. This was performed as part of the health monitoring program to ensure that the rodents are free of specific pathogens that could affect scientific experiments. A total of 13 commensal E. coli strains were recovered on MacConkey agars, from the gastrointestinal tract, liver and, the trachea and lung of laboratory rodents (Table 2). Accession numbers for the partial 16S rDNA sequences of the 13 commensal E. coli strains include, LT009415–LT009427. All 13 commensal E. coli strains in this study were recovered from female rodents. These 13 strains were cultured from 10 different, individual female rodents and, we did not manage to isolate E. coli from the remaining 54 rodents. E. coli was isolated mainly from the gastrointestinal tract (n = 6) and, the trachea and lung (n = 5), with the remaining 2 strains from the liver. A majority of the commensal E. coli strains were found in ICR mice (n = 8). Meanwhile, C57BL/6 mice harbored 2 strains and, the BALB/c mouse, Sprague–Dawley and Wistar Kyoto rats, each harbored 1 strain. Two E. coli strains were isolated from the different organs of 3 individual ICR mice. Strains UM-AEU197 and 198 were isolated from the liver and, the trachea and lung of ICR mouse #1, respectively. Strains UM-AEU202 and 203 were isolated respectively, from the liver and, the trachea and lung of ICR mouse #2. The trachea and lung, and the gastrointestinal tract of ICR mouse #3 sheltered strains UM-AEU213 and 214, respectively (Table 2).

Table 2 Genetic characteristics of commensal E. coli strains from rodents in the animal experimental unit
Full size table

Analysis of the commensal E. coli strains by MLST revealed 13 strains with 4 novel and 1 common STs (Table 2). The E. coli strains recovered from ICR mice, were designated into ST748 and ST749. Strains recovered from Sprague–Dawley and Wistar Kyoto rats were designated into ST746 and ST747, respectively. In addition, ST747 was also found in a C57BL/6 mouse. The sole common E. coli ST357 was recovered from the gastrointestinal tract of another C57BL/6 mouse. The gastrointestinal tract was found to harbor all 5 different STs (ST357, ST746, ST747, ST748 and ST749) in this study. The liver and, the trachea and lung however, only harbored ST749 and it was also the most frequent ST encountered. Analysis into the phylogroups of the commensal E. coli strains by the quadruplex PCR method revealed that all 13 strains belonged to phylogroup A (Table 2). The alignments of all eight E. coli MLST loci relative to the respective reference Allele 1 are shown in Additional files 1, 2, 3, 4, 5, 6, 7 and 8.

The multiplex PCRs to detect the most common clinically significant Shiga toxin-producing E. coli serotypes showed that all 13 commensal E. coli strains in this study did not belong to any of the serotypes (Additional file 9). Similarly, we observed negative ESBL activity in the 13 commensal E. coli strains when tested using the ESBL phenotypic confirmatory method (Table 3). Negative ESBL activity however, does not necessarily mean that the E. coli strain is negative for ESBL encoding genes, therefore a series of nucleic acid amplifications for the detection of ESBL encoding genes was performed. However, no ESBL encoding gene was amplified from the commensal E. coli strains recovered in our study (Additional files 10, 11, 12).

Table 3 Prevalence of ESBL producers among E. coli Strains from rodents in the animal experimental unit
Full size table

Testing of the animal housing environment, feed and drinking water of the laboratory rodents for the presence of E. coli showed that only the wash basin sample resulted in positive E. coli culture. The cultured E. coli strain belonged to phylogroup B1 and no ESBL genes were detected by PCR. E. coli was not cultured from the other animal housing environment samples (table top, tap and trolley), rodent feed and drinking water (Table 4).

Table 4 Prevalence of E. coli in the animal housing environment, rodent feed and drinking water
Full size table

Discussion

In this study, several key factors involved in influencing the distribution of E. coli populations in laboratory rodents were controlled. MacConkey agar was used for the culture of E. coli strains in this study because it has the propensity to culture members of the Enterobacteriaceae and cultured E. coli colonies can be easily distinguished by its morphology and color (Gordon and Cowling 2003). Consequently, we only sampled one E. coli strain from each suspected agar plate representing the most dominant strain despite evidence that hosts can be colonized with more than one E. coli strain at the same time (Gordon et al. 2005). Even though E. coli is an established colonizer of the mouse gut (Skurnik et al. 2016), translocation to other organs including the liver (Ohsugi et al. 1996) and the lung (Wu et al. 2010) has been reported. Therefore, in our study, we examined the other rodent organs (liver and, the trachea and lung) for the presence of E. coli. In agreement with the previous reports (Ohsugi et al. 1996; Wu et al. 2010), we found two E. coli strains, each, from different organs in three individual ICR mice. The rodent skin was scrubbed with antiseptic prior to surgery and the harvesting of rodent organs in the following sequence; firstly, the trachea and lung, secondly, the liver, and lastly, the gastrointestinal tract. New sets of sterile surgical tools were utilized for the harvesting of different organs. These procedures were performed to prevent E. coli cross contamination and translocation was evident with the sighting of similar STs among E. coli strains from different organs of the same individual.

The distribution of E. coli populations in the mammalian host depended on the host habitat, diet, gut anatomy and body mass (Gordon and Cowling 2003). In humans, morphological, physiological and dietary differences among individuals of different sex and age influence the distribution of commensal E. coli populations (Gordon et al. 2005). It was further suggested that the exposure to novel E. coli strains and the health of the animal might contribute to the temporal variability of commensal E. coli populations in animals (Grønvold et al. 2010). Shifting of the commensal E. coli population in laboratory rodents can alter experimental outcomes by stressing the rodent’s microbiota into a state of imbalance (Ma et al. 2012) and this is not desired in any animal experimental facilities. Previous study has reported a lower likelihood for the transmission of bacterial plasmids carrying virulence genes to animals housed separately or in small groups, in comparison to animals raised in cramped and dense conditions (Koga et al. 2015). The rodents in this study were housed at a barrier facility in open cages maintained at a constant room temperature, relative humidity and light:dark cycle (Loong et al. 2016). Standard operating procedures including, the wearing of personal protective equipment, working on one animal at a time in the biosafety cabinet, changing gloves after handling a single animal and performing surgeries in a dedicated room away from breeding facilities (Loong et al. 2016) were followed as per required under AAALAC accreditation to prevent introduction of other bacteria species into the studied cohort. To prevent possible transmission from the animal housing environment, several sites in the room where the rodents were housed (table top, tap, trolley and wash basin), were selected to test for the presence of E. coli. Similarly, the rodent’s feed and drinking water were also tested for the presence of E. coli. We however, found that only the sample from the wash basin grew E. coli belonging to phylogroup B1 with no detectable ESBL genes. The presence of E. coli phylogroup B1 only at the wash basin suggests that the strain favors the water environment, agreeing with the earlier report by Ratajczak et al. (2010) who found phylogroup B1 strains to be specifically adapted and persistent in water.

The rodents were fed with diet comprising of products made of soy and corn, equivalent to a herbivorous diet. Identical animal feed was maintained throughout the study without the change of vendor or food variant, and the non-alteration of diet preserves the condition in the gut favoring the dominant E. coli strain (Blyton et al. 2013). Introduction of different diets would disturb the gut ecosystem leading to the colonization of a different E. coli strain (Grønvold et al. 2010). Sexually mature healthy rodents (Kempermann et al. 1997; Sengupta 2013) were selected because the dominant commensal E. coli strains would have successfully established stable colonization in the gut flora (Tenaillon et al. 2010) and good health ensured no shifting of host physiological condition (Grønvold et al. 2010), possibly displacing dominant E. coli strains in unhealthy animals. The isolation of commensal E. coli only from females in this study suggests that host gender could influence the distribution of E. coli and this was similar to human E. coli strains (Gordon et al. 2005), even though the number of rodent strains in this study was relatively small to draw a definite conclusion. In terms of the gut anatomy, E. coli is inclined to flourish in hosts (herbivores and omnivores) with intestinal tracts resembling a microbial fermentation chamber such as a caecum (Gordon and Cowling 2003; Gordon et al. 2005) and all rodents possess a caecum (Gordon and Cowling 2003). In addition, the likelihood of isolating E. coli increases with the body mass of animals (Gordon et al. 2005). This was not the case in our study however, where smaller mice contributed to >80 % (11/13) of the isolated commensal E. coli strains.

Assessment of the E. coli phylogroups categorized the strains isolated from laboratory mice and rats into a single phylogroup A. This finding is largely concordant with other published results on free-range poultry (Koga et al. 2015) and domesticated animals (Tenaillon et al. 2010), where the animals were not housed in dense conditions. Phylogroup B1 strains are best adapted in animals, followed by phylogroup A (Tenaillon et al. 2010), suggesting consistency in our results. The population diversity of E. coli due to host diet has been reported with phylogroup A strains predominating in carnivores and omnivores (Escobar-Páramo et al. 2006). Rodents by nature are omnivores and although they were fed a herbivorous diet in this study, it did not change the genetic structure of commensal E. coli strains, suggesting the effect of nutrients on the selection of strains. Perhaps the threshold concentration of nutrients could not be achieved from a herbivorous diet, hence phylogroups associated to herbivores could not succeed in colonization (Gordon et al. 2005). The probability of discovering phylogroup A strains is also higher in female humans (Gordon et al. 2005), similar to the findings in laboratory rodents in this study. Another possible explanation for this finding could be that sex hormones may modulate host physiological conditions and it has been demonstrated in humans that sex hormones provoke metabolic changes that shape the intestinal microbiota (Koren et al. 2012). The association between the female gender and E. coli colonization suggests laboratory rodents to be a good model for studying the impact of sex hormones on the colonization of other commensal bacteria species. Gordon and Cowling (2003) also reported from their survey of the distribution of E. coli in Australian vertebrates that phylogroup A can be recovered from any vertebrate group, however this phylogroup is also responsible for causing intestinal infections in mammals. The finding of a different phylogroup B1 from the wash basin further supports our suggestion that the strains recovered from the laboratory rodents were true commensals and transmission from the animal housing environment, feed or water did not occur. The absence of ESBL genes in the E. coli strain isolated from the wash basin indicated that the rodent housing facility was not tainted by any clinical bacterial strains carrying resistance genes.

The discovery of 4 novel STs (ST746–ST749) on top of one common ST (ST357) from a relatively small number of specimens using MLST, signified that the rodent’s microbiota is more diverse than the human’s (Smati et al. 2015). Our findings revealed a distinct demarcation of STs for laboratory rodents, with ST357, ST748 and ST749 found in mice, and ST747 in rats (Table 2). This could be explained by the genetic variability among different rodent breeds (Wu et al. 2001) that predisposes them to having different physiological responses (Gibney et al. 2010) and, hormone changes are known to affect bacteria colonization in the host (Blyton et al. 2013). Isolation of ST746 from both mouse and rat suggest that this ST could have adapted to the physiology of different rodent breeds. Based on ST similarities in the different E. coli strains cultured from the liver and, the trachea and lung specimens denote that the rodent body is frequently exposed to commensal bacteria from the gastrointestinal tract (Wu et al. 2001). It has been widely accepted that E. coli phylogroups are non-randomly distributed among mammalian hosts (Gordon et al. 2005; Clermont et al. 2013) and all of these findings imply that phylogroup A strains establish an ecological niche in laboratory rodents.

Serotyping of the isolated commensal E. coli strains using multiplex PCR assays revealed that all the rodent strains were negative for the 21 most clinically relevant Shiga toxin-producing E. coli. All commensal E. coli strains were also negative for ESBL production and genes encoding beta lactamases. This was not surprising, considering that the rodents were housed in proper AAALAC accredited animal housing facility. The level of hygiene in the human population was noted as one of the main factors for the significant shift in the proportions of E. coli strains in France (Tenaillon et al. 2010) and an increased level of antimicrobial resistance in E. coli isolated from animals was observed with an increase density of human population living in close proximity to these animals (Skurnik et al. 2016). Furthermore, we could not culture E. coli from a large number of our specimens, suggesting that the bacteria density was very low in the organs and low bacteria density would restrict the occurrence of horizontal resistance gene transfer (Skurnik et al. 2016). The low bacteria density and the absence of antimicrobial resistance genes in commensal E. coli strains of laboratory rodents is consistent with the hypothesis that strains with more virulence genes are likely to command better survival fitness and persist longer in the gut (Blyton et al. 2013). One limitation of this study was the relatively small number of isolated strains and the study was performed in only one animal experimental facility. It is possible that E. coli strains from other animal species and other animal experimental facilities would differ.

Conclusions

This study found that the commensal E. coli strains isolated from different rodent breeds housed in the Animal Experimental Unit, Faculty of Medicine, University of Malaya were categorized as members of phylogroup A. Four novel sequence types were uncovered using MLST, displaying the broad diversity of commensal E. coli strains from laboratory rodents. As expected, the isolated E. coli strains did not possess antibiotic resistance genes and, were not Shiga toxin and extended spectrum beta lactamase producers. We propose that animal experimental facilities should initiate surveillance of commensal E. coli strains for the determination of phylogroups and the presence of ESBL genes, to monitor for events that could lead to dysbiosis and distortion of animal experimental data. Taken together, the absence of antimicrobial resistance and virulence genes in the E. coli cohort of this study indicate the true commensal nature of these strains. To the best of our knowledge, this study is also the first to analyze the genetic structure of commensal E. coli in laboratory rodents.