A Large Spatial Survey of Colistin-Resistant Gene mcr-1-Carrying E. coli in Rivers across Taiwan.

BACKGROUND: Colistin is one of the last-line antimicrobial agents against life-threatening infections. The distribution of the colistin resistance gene mcr-1 has been reported worldwide. However, most studies have focused on the distribution of mcr-1-positive bacteria in humans, animals, food, and sewage; few have focused on their distribution in natural environments.
METHOD: We conducted a large spatial survey of mcr-1-positive Escherichia coli at 119 sites in 48 rivers, covering the entire island of Taiwan. We investigated the relationship between the livestock or poultry density in the surveyed riverine area and the number of mcr-1-positive E. coli in the river water. We then sequenced and characterized the isolated mcr-1-positive plasmids.
RESULTS: Seven mcr-1 positive E. coli were isolated from 5.9% of the sampling sites. The mcr-1-positive sites correlated with high chicken and pig stocking densities but not human population density or other river parameters. Four of the mcr-1-positive E. coli strains harbored epidemic IncX4 plasmids, and three of them exhibited identical sequences with a size of 33,309 bp. One of the plasmids contained identical 33,309 bp sequences but carried an additional 5711-bp transposon (Tn3 family). To our knowledge, this is the first demonstration that mcr-1-carrying IncX4 plasmids can contain an insertion of such transposons. All mcr-1-positive isolates belonged to phylogenetic group A and harbored few known virulence genes.
CONCLUSION: This study showed a positive relationship between the number of mcr-1-positive sites and high livestock and poultry density. The sequencing analyses indicated that the epidemic plasmid in the mcr-1 isolates circulates not only in humans, animals, and food but also in the associated environments or natural habitats in Taiwan, suggesting that the surveillance of antibiotics-resistance genes for livestock or poultry farm quality control should include their associated environments.

1. Introduction

The rapid emergence and dissemination of antibiotic-resistant bacteria has become a serious threat to public health globally. Colistin is one of the last-line antimicrobial agents for treating life-threatening infections caused by multidrug-resistant bacteria. However, after the first report by Lui in China in 2015, the plasmid-mediated colistin resistance gene mcr-1 has been found to be widespread in five continents [1,2]. The marked increase in mcr-1-carrying bacterial isolates worldwide can be explained by the initial mobilization of mcr-1 by an ISApl1 transposon in the mid-2000s and its rapid mobilization onto other plasmids [3]. Escherichia coli is the most prevalent species among the mcr-1-positive isolates, and constitutes approximately 90% of the total mcr-1-carrying isolates [2]. Additionally, mcr-1 carrying isolates can be found in raw meat, livestock animals, infections, and healthy people [1,4]. A higher number or percentage of mcr-1 carriage in E. coli isolates has been found in raw meat and food animals than in human isolates [1]. Researchers suggest that mcr-1 resistance may have emerged in the animal sector due to the widespread use of colistin in food animals, including pig and poultry farms, in several areas [5]. Additionally, a study conducted in China found that 6% of human fecal samples carried the mcr-1 gene [6]. Colistin is prohibited as a growth promoter in Taiwan, but it can be used for disease treatment in veterinary treatment [7]. In Taiwan, mcr-1-positive E. coli has been found in 0.4% of asymptomatic adults [8] and 0.6% of clinical isolates [9]. Similar to studies in other countries, the rate of mcr-1 positive isolates from meat and swine/poultry diseases in Taiwan exceeds that in isolates from humans (retail meats: 8.7% in 2015; diseased swine: 33.3% in 2016) [7,10].

In addition to in food animals and humans, mcr-1 has been found in aquatic systems, particularly sewage or wastewater [11,12]. A recent literature review indicated that 18 publications have reported on plasmid-mediated colistin resistance in 2107 isolates from freshwater and seawater [13], though few have focused on mcr-1-positive isolates in natural habitats or associated environments [13]. In addition to China [5,14,15], mcr-1 has also been observed in freshwater environments in Malaysia [16], Italy [17], and Switzerland [18]. However, these studies have mainly focused on one or a few freshwater sites.

To gain a full picture of the distribution of mcr-1-positive E. coli in Taiwan’s freshwater system, an intensive survey in 52 rivers across the entire island of Taiwan was conducted in this study. Taiwan is a mountainous island, consisting of 268 mountains above 3000 m, with most of the rivers flowing in a steep descent from the center of the island to the surrounding sea [19]. In addition to comprehensively surveying mcr-1-positive E. coli, this survey allowed us to examine our hypothesis that the number of mcr-1 isolates in the river could be correlated with the livestock or poultry density in the riverine environment. Moreover, the plasmid sequences and their gene content in the positive E. coli provided more insight into the ecology of colistin-resistant genes in Taiwan and the Western Pacific region.

2. Materials and Methods

2.1. Water Sampling

Water samples were collected from different rivers from December 2015 to February 2016 (Figure 1). The Taiwan Environmental Protection Administration (TEPA) routinely examines river water in Taiwan to monitor the river pollution index and coliform bacteria population [20,21]. The river pollution index includes the concentrations of four parameters in water: dissolved oxygen (DO), biochemical oxygen demand (BOD5), suspended solids (SS), and ammonia nitrogen (NH3-N). Information regarding the livestock and poultry stocking numbers was obtained from the Council of Agriculture, Taiwan ((https://agrstat.coa.gov.tw/sdweb/public/book/Book.aspx) (accessed on 30 March 2021) and (https://www.naif.org.tw/main.aspx) (accessed on 30 March 2021). A total of 48 rivers and 119 sites were sampled in this study. We sampled each river site three times during the study period. Up to two isolates of E. coli were cultured from each sampling site. Therefore, a total of 537 E. coli isolates were obtained.

Figure 1

River collection sites in this study; riverine areas with mcr-1-positive sites are indicated by a red line.

One-hundred milliliters of river water was sampled each time and analyzed following the standard procedures of the Environmental Analysis Laboratory of the TEPA [22]. Additionally, all of the sampling procedures followed the rules established by the Environmental Analysis Laboratory of the TEPA. TEPA method: River, Lake, Reservoir water quality sampling general rule, National Institute of Environmental Analysis (NIEA)W104.51C was used as the standard basis for sampling. The water samples were analyzed immediately after collection, and the time between sample collection to laboratory work completion was <24 h.

To conduct cluster sampling for each river and county, we randomly selected E. coli from samples collected at the 119 river stations in Taiwan. One water sample was collected and filtered to further isolate E. coli at each sampled station, following previously described methods [22]. We used E. coli CHROMagar (ECC) plates (CHROMagar, Paris, France) to screen E. coli, which was incubated at 37 °C for 24 h, and up to two E. coli colonies were selected per sample. mcr-1 PCR was conducted in E. coli isolates using the primers CLR5-F (5′-CGGTCAGTCCGTTTGTTC-3ʹ) and CLR5-R (5′-CTTGGTCGGTCTGTA GGG-3′), following previously described methods [1]. We used eight housekeeping gene sequences (adk, fumC, gyrB, icd, mdh, purA, and recA) according to the protocol on the MLST database website (http://mlst.warwick.ac.uk/mlst/dbs/Ecoli) (accessed on 30 March 2021), and the broth dilution method, to check the antimicrobial susceptibility of sulfamethoxazole and trimethoprim (STX-TMZ), ciprofloxacin, tetracycline, meropenem, azithromycin, nalidixic acid, cefotaxime, chloramphenicol, tigecycline, ceftazidime, colistin, ampicillin, and gentamicin (Clinical and Laboratory Standards Institute (CLSI), 2020). Colistin was defined as intermediate (MICs (minimum inhibitory concentrations) of ≤2 mg/L) or resistant (MICs ≥ 4 mg/L) according to the MIC interpretive criteria from the CLSI.

2.2. Conjugation Assays

Conjugation assays were conducted to determine whether the mcr-1 genes in the river isolates were harbored on conjugative plasmids. The E. coli strains MG1655 lacZ::Gm and MG1655 recA::Tet, which harbor gentamycin and tetracycline resistance, respectively, were used as recipients in the assays. The five colistin-resistant river isolates (EC1278, EC1279, EC1280, EC1281, and EC1283) served as donors. Based on the antibiotic resistance, MG1655 lacZ::Gm was used as the recipient in the conjugation experiment with EC1278, EC1279, EC1281, and EC1283, while MG1655 recA::Tet was used as the recipient of EC1280. The bacterial strains were grown overnight in Lysogeny broth (LB) medium containing appropriate antibiotics (2 mg/mL of colistin, 2 mg/mL of gentamycin, or 2 mg/mL of tetracycline) [23]. To remove antibiotics from the overnight cultures, the bacteria in 200 μL samples of the cultures were washed with 200 μL of fresh LB medium once by centrifugation at 1000× g for 3 min and then re-suspended in 20 μL of fresh LB medium. The resulting donor (20 μL) and recipient (20 μL) suspensions were mixed, and 10 μL of the mixture was dropped on an LB agar plate without any antibiotic. After incubation at 37 °C for 24 h, the bacteria on the plate were re-suspended in PBS and spread on LB plates containing colistin, gentamycin, or tetracycline, to select for colistin-resistant transconjugants. Colistin had the growth inhibition of the recipients, while gentamycin or tetracycline had the growth inhibition of the donors.

2.3. Purification, Sequencing, and Assembly of Plasmids

To determine plasmid patterns in bacteria, plasmids were isolated following the alkaline method of Kado and Liu [24] and subjected to agarose gel electrophoresis.

To sequence the colistin-resistant plasmids, the plasmids were purified from the responding transconjugants following the alkaline lysis method described previously [25]. The complete nucleotide sequences of the plasmids were determined by MiSeq sequencing (Illumina Inc., San Diego, CA, USA.). The plasmids were annotated using the DDBJ (DNA Data Bank of Japan) Fast Annotation and Submission Tool (DFAST) pipeline [26]. Insertion sequences (IS) were annotated using ISFinder (https://isfinder.biotoul.fr/) (accessed on 30 March 2021), as described previously [27]. For comparative analysis, plasmid sequences were aligned against the non-redundant database using the MegaBLAST algorithm (NCBI BLAST), with the default settings for the parameters.

2.4. PCR-Based Phylogenetic Typing and Genotyping

The phylogenetic types of the river colistin-resistant E. coli strains were determined using a triplex PCR-based method to detect the presence of the chuA and yiaA genes, and the DNA fragment TSPE4.C2, as described previously [28]. The presence of known virulence genes in E. coli was determined by PCR using previously described primers and conditions [29,30,31,32]. The RS218 and CFT073 pathogenic E. coli strains served as positive controls in the PCR analyses of ompT, ibeA, cnf1, sfaS, ireA, chuA, ihA, usp, sat, iroN, and hlyA. The clinical E. coli isolates A865, which had been previously identified as harboring afa/draBC [32,33,34], and EC586, which harbors hlyF, iutA, and iss (unpublished data), served as positive controls for the corresponding genes. Additionally, the MG1655 E. coli strain served as a negative control for all the genes, excluding ompT. An ompT-deletion strain of RS218 served as the negative control for ompT. The primer sequences used for the PCR analyses are listed in Supplementary Materials Table S1.

2.5. Accession Numbers

The sequences of EC1279, EC1280, EC1281, and EC1283 were deposited in GenBank under the accession numbers MW010025, MW010026, MW010024, and MW010027, respectively.

2.6. Statistical Analysis

All statistical analyses were conducted using SPSS version 20.0 for Windows (SPSS Inc., Armonk, NY, USA). Categorical variables were analyzed using the Chi-square or Fisher’s exact tests, and the continuous variables were analyzed by conducting an independent samples t-test. A p-value of <0.05 was considered statistically significant.

3. Results

3.1. Positive mcr-1 E. coli Sites

In the river environment, mcr-1-positive E. coli were found at 5.9% (7/119) of the water sampling sites in the rivers of Taiwan. The seven mcr-1-positive sites were distributed in central (n = 3, Wu, Shihwu, and Beigang Rivers), southern (n = 3, Ba-Chang, Gaoping, and Donggang Rivers), and eastern Taiwan (n = 1, Beinan River; Figure 1). No mcr-1-positive sites were identified in northern Taiwan in our survey. Excluding two sites (Wu and Beigang Rivers), most (71%; 5/7) mcr-1 positive isolates were collected from the downstream area of the river.

An example of an mcr-1-positive site in the downstream area (Gaoping River) is provided in Supplementary Figure S1. For the seven isolates, the antimicrobial drug susceptibility is shown in Table 1. Two of the seven isolates were susceptible to colistin in the broth dilution method. The MLST study identified different STs, including ST155, ST6732, ST877, ST7149, and ST3661 (n = 3).

Table 1

The antimicrobial drug susceptibility and STs in seven mcr-1-positive isolates.

NO	STX	TMZ	CIP	TET	MER	AZI	NAL	CTX	CHL	TIG	CAZe	COL	AMP	GM	ST
EC1278	>1024	≤0.25	≤0.015	>64	≤0.03	≤2	≤4	≤0.25	=32	≤0.25	≤0.5	=8	>64	≤0.5	3661
EC1279	>1024	>32	=0.25	>64	≤0.03	=4	>128	≤0.25	=128	=1	≤0.5	=8	>64	≤0.5	3661
EC1280	=16	=0.5	≤0.015	=64	≤0.03	=8	≤4	≤0.25	≤8	=0.5	≤0.5	=8	=2	≤0.5	3661
EC1281	>1024	>32	=0.25	>64	≤0.03	≤2	=128	≤0.25	=128	≤0.25	≤0.5	=8	>64	=1	877
EC1282	>1024	>32	=0.12	=64	≤0.03	=8	=8	≤0.25	=128	=0.5	≤0.5	≤1	>64	=1	155
EC1283	=16	=0.5	≤0.015	=4	≤0.03	=4	≤4	≤0.25	>128	≤0.25	≤0.5	=4	>64	=8	6732
EC1284	=16	=0.5	=0.03	≤2	≤0.03	=4	≤4	≤0.25	≤8	≤0.25	≤0.5	≤1	=4	≤0.5	7149

STX-TMZ; CIP: ciprofloxacin; TET: tetracycline; MER: meropenem; AZI: azithromycin; NAL: nalidixic acid; CTX: cefotaxime; CHL: chloroamphenicol; TIG: tigecycline; CAZ: ceftazidime; COL: colistin; AMP: ampicillin; GM: gentamicin.

The site characteristics regarding the distribution of mcr-1-positive and negative E. coli are shown in Table 2. The air and water temperature, river pollution index, PH, and other parameters, such as the number of coliforms in the mcr-1-positive and negative groups, are shown in Table 2. The pollution index was slightly higher at the mcr-1-positive sites, but this difference was not statistically significant in the t-test. The human population density, air/water temperature, pH, and coliform number were similar between the mcr-1-positive and negative groups. Using a continuous variable, the livestock stocking density was higher at the mcr-1-positive sites but did not reach statistical significance in the independent t-test.

Table 2

Independent t-test of the mcr-1-positive and mcr-1-negative E. coli sites.

Parameter	mcr-1-Positive E. coli Site	mcr-1-Negative E. coli Site	p-Value
	N = 7	N = 112
Pollution index (mean ± SD)	4.08 ± 2.31	2.84 ± 1.71	0.073
Air temperature (mean ± SD)	20.59 ± 6.59	21.3 ± 4.78	0.712
Water temperature (mean ± SD)	21.02 ± 5.68	20.38 ± 3.65	0.661
PH (mean ± SD)	7.97 ± 0.36	7.81 ± 0.53	0.421
Coliform number (mean ± SD)	252,657 ± 375,707	169,949 ± 636,844	0.735
Chickens stocking density (birds/km2; mean ± SD)	5354 ± 4472	2956 ± 7007	0.223
Pigs stocking density (herds/km2; mean ± SD)	746 ± 1174	185 ± 421	0.254
Cows stocking density (herds/km2; mean ± SD)	45 ± 112	4.6 ± 17.4	0.372
Human population density (people/km2; mean ± SD)	1096 ± 1011	1680 ± 3533	0.665

The relationship between the mcr-1 positive sites and chicken and pig stocking densities is shown in Figure 2A (left) and Figure 2B (right).

Figure 2

Livestock density and mcr-1 positive and negative site correlations: (A) (left): chicken density; (B) (right) pig density.

According to the Chi-square test using categorical variables for analysis (Table 3), mcr-1-positive sites were more likely to occur in the category with chicken stocking densities of 1000–5000 and >5000 birds/km2 (42.9% vs. 27.7%; 42.9% vs. 12.5%; p = 0.028) than mcr-1-negative sites. Regarding pig density, mcr-1-positive sites were more likely to occur in the category with a stocking density of >1000 herds/km2 (28.6% vs. 3.6%; p = 0.012) than mcr-1-negative sites (Table 3). The distributions of mcr-1 positive and negative sites did not differ between the four pollution indices (unpolluted, negligible, moderately, and severely polluted) and human density (more or less than 1000 people/km2) categories (Table 3).

Table 3

Chi-square test of the mcr-1-positive and mcr-1-negative E. coli sites.

Categorical Parameter	mcr-1-Positive E. coli Site(n = 7)	mcr-1-neGative E. coli Site(n = 112)	p-Value
Chicken stocking density (No./km2)			0.028 *
<1000	1 (14.3)	67 (59.8)
1000–5000	3 (42.9)	31 (27.7)
>5000	3 (42.9)	14 (12.5)
	0	50
Pig stocking density (No./km2)			0.012 *
<100	4 (57.1)	73 (65.2)
100–1000	1 (14.3)	35 (31.2)
>1000	2 (28.6)	4 (3.6)
Cow stocking density (No./km2)			0.613
0–5	5 (71.4)	95 (84.8)
6–10	1 (14.3)	7 (6.2)
≥11	1 (14.3)	10 (8.9)
River pollution severity			0.542
Unpolluted	1 (14.3)	42 (37.5)
Negligibly polluted	2 (28.6)	23 (20.5)
Moderately polluted	3 (42.9)	41 (36.6)
Severely polluted	1 (14.3)	6 (5.4)
Human population density			1.000
<=1000 people/km2	5 (71.4)	70 (62.5)
>1000 people/km2	2 (28.6)	42 (37.5)

* p-Value < 0.05.

3.2. Capturing mcr-1-Carrying Conjugative Plasmids

mcr-1 genes are often carried by conjugative plasmids. Therefore, plasmid patterns of the river colistin-resistant strains were investigated. As shown in Figure 3A, all five strains harbored multiple plasmids, and their plasmid patterns differed. The distinct plasmid patterns may reflect the distinct regions of the isolated strains.

Figure 3

Plasmids in the colistin-resistant strains. (A) Plasmids in the riverine colistin-resistant E. coli isolates. (B) Plasmids in the transconjugants that obtained colistin-resistant plasmids from the riverine strains. Size markers 1 and 2 were plasmids isolated from Salmonella strains OU7058 and OUT7526, respectively [35].

To further investigate whether the mcr-1 genes in the river colistin-resistant strains were encoded in conjugative plasmids, conjugation experiments were conducted with the colistin-resistant strains as donors and E. coli MG1655-derived strains as recipients, selecting for colistin-resistant transconjugants. The colistin-resistant transconjugants were obtained from experiments with four river-born strains, including EC1279, EC1280, EC1281, and EC1283, which were denoted as Trans-1279, Trans-1280, Trans-1281, and Trans-1283, respectively, and the plasmid profiles of the transconjugants were investigated. As shown in Figure 3B, Trans-1279, Trans-1280, and Trans-1283 harbored plasmids with similar sizes, while Trans-1281 harbored a plasmid that was apparently larger than those in the other transconjugants. These results suggest that the mcr-1 genes in four of the river-borne colistin-resistant E. coli strains were encoded in conjugative plasmids.

3.3. Genetic Characterization of mcr-1-Carrying Plasmids

The colistin resistance-encoding plasmids were purified from the transconjugants and sequenced for further characterization. The plasmids derived from EC1279, EC1280, EC1281, and EC1283 were designated pEC1279, pEC1280, pEC1281, and pEC1283, respectively. pEC1279, pEC1280, and pEC1283 were 100% identical in sequence and size (33,309 bp). pEC1281 was 39,025 bp in size, which contained a 33,309 bp region identical to those of the above plasmids with the insertion of an additional 5716-bp fragment (Figure 4).

Figure 4

Linear comparison of the pEC1279-like plasmids (pEC1279, pEC1280, and pEC1283) and pEC1281, which contained a Tn3 family transposon. IRL—left-inverted repeated sequence; IRR—right-inverted repeated.

These plasmids are the IncX4 plasmids. Based on BLAST analysis, pEC1279, pEC1280, and pEC1283 aligned well with a group of IncX4 plasmids (>99.9% identity), which were characterized by harboring an mcr-1 gene with a downstream pap2 gene and insertion sequence IS26 located upstream of the mcr-1-pap2 element (Figure 4). The plasmids in this group were mainly carried by Enterobacteriaceae isolated from humans, animals, meat, and wastewater, and are distributed worldwide (Table 4). We designated these plasmids as being pEC1279-like. Notably, on the island of Taiwan, pEC1279-like plasmids were identified from bacteria isolated from humans and swine, such as pNG14043 (Salmonella from a human), pKP15450-MCR-1 (Klebsiella pneumoniae from a human), and pNCYU-24-74-6_MCR1 (E. coli from a swine; Table 4). This study demonstrates that such mcr-1-carrying plasmids have spread to the natural environment of the island, in addition to humans, animals, and foods.

Table 4

Examples of pEC1279-like plasmids.

Plasmid	Source	Strain	Regions	Size (bp)	Accession #	Reference
pKP15450-MCR-1	Human	K. pneumoniae	Taiwan	33,309	MH715959.1
pNG14043	Human	Salmonella	Taiwan	33,308	KY120364.1	[36]
p2017.19.01CC	Human	E. coli	Vietnam	33,309	LC511660.1	[37]
p31349	Human	E. coli	Switzerland	33,303	KY689634.1	[38]
pmcr1_IncX4	Human	K. pneumoniae	China	33,287	KU761327.1	[39]
pMCR-1_Msc	Human	E. coli	Russia	33,310	MK172815.1	[40]
pWI2-mcr	Human	E. coli	France	33,304	LT838201.1	[41]
pICBEC12-3mcr	Avian	E. coli	Brazil	33,304	CP021419.1
pMcp0271	Chicken meat	E. coli	Switzerland	33,303	KY565556	[38]
pHNSHP10	Swine	E. coli	China	33,309	MF774182.1	[42]
pNCYU-24-74-6_MCR1	Swine	E. coli	Taiwan	33,300	CP042644.1	[7]
pCSZ4	Pork	E. coli	China	33,309	KX711706.1	[43]
pMCR_WCHEC1618	Wastewater	E. coli	China	33,309	KY463454.1	[44]
pB2	Wastewater	E. coli	Japan	33,309	LC479085.1	[45]

# NCBI GenBank database accession numbers.

The additional 5.7-kb fragment in pEC1281 was located between the mcr-1-pap2 element and IS26, which contained a transposon structure that encoded a transposase, resolvase, and potential ABC transporter with a 35-bp inverted repeat (IR) sequence at both ends (Figure 4). This transposon belongs to the Tn3 family [46], and has been identified in various plasmids, such as pCHL5009T-102k-mcr3 [46], pH226B [47], pNDM5-GZ04_A [48], and pV233-b [49]. However, to the best of our knowledge, this is the first demonstration of the insertion of a pEC1279-like mcr-1-carrying plasmid by this type of transposon.

3.4. Genetic Features of Colistin-Resistant River Isolates

To further investigate the genetic background of colistin-resistant river isolates, the phylogenetic types of EC1278, EC1297, EC1280, EC1281, and EC1283 were investigated. E. coli strains are primarily classified into four phylogenetic groups, designated A, B1, B2, and D [27,50]. Extraintestinal pathogenic E. coli (ExPEC) are mainly derived from phylogenetic groups B2 and D, while commensal E. coli are mainly derived from Groups A and B1 [51]. All of the strains identified here belonged to phylogenetic group A, suggesting that these river isolates may not have been pathogenic E. coli.

We further analyzed the virulence capability of the strains by determining the presence of 15 known E. coli virulence genes in the bacteria. The pathogenic roles of these virulence genes include adherence (afa/draBC, iha, and sfaS), invasion (ibeA), toxins (cnf1, hlyA, sat, and hlyF), iron uptake (chuA, ireA, iroN, and iutA), bacterial resistance to complement-mediated attack (iss), and miscellaneous pathogenic functions (ompT and usp). Excluding ompT, the known virulence genes were not detected in the five river isolates. ompT was identified in EC1278 and EC1283, but not in the other isolates. These findings indicate that these river isolates had low virulence, suggesting that these river colistin-resistant E. coli strains were non-pathogenic.

4. Discussion

This study demonstrated that the sites containing mcr-1 positive E. coli in rivers were positively correlated with the density of livestock and poultry in the riverine area of Taiwan. No correlations were detected between pH, temperature, pollution index, and human density. Although colistin-resistant genes were only distributed in a few rivers, Taiwan’s natural environment has been contaminated with mcr-1-positive bacteria. Natural habitats or human activity-associated environments that have not yet been considered may serve as hidden yet critical spaces for bacterial gene transfer and transmission of resistant genes. We suggest natural environment surveys should be conducted to monitor the dissemination of colistin-resistant genes.

Contamination of mcr-1 positive E. coli in rivers was likely due to anthropogenic activities, as the distribution of these bacteria was associated with high pig and poultry population densities (Figure 2). Colistin has been approved for animal use by the Council of Agriculture in Taiwan. It is likely that its use poses a selection advantage for mcr-1-positive E. coli that live in the intestines of livestock. Following population expansion, such bacteria may be released from the animals and reach and contaminate nearby rivers. Conventional livestock waste treatment processes cannot completely remove antibiotic-resistance genes, and result in contamination of water environments [52,53]. Some studies have reported the transmission of antimicrobial-resistant bacteria from pig manure to the environment [54], and these antibiotic-resistant bacteria may spread through water [55].

Our study showed that the mcr-1-positive sites were correlated with livestock density in river environments, and correlations between other antimicrobial resistance genes and livestock have been reported in the literature. Poultry production carries a high risk for antibiotic resistance emergence and consumes more antibiotics than the cultivation of other animals [56]. The size and scale of poultry farming are associated with the antimicrobial colonization rates [56]. Independent of antimicrobial drug usage, there is evidence of a relationship between chicken density and antimicrobial-resistant pathogens [57]. The results of the analysis of ESBL genes from chicken feces and upstream and downstream river water suggest that animal farm effluent could contribute to the spread of resistance genes [58]. A study conducted on swine feces and downstream water in China suggested that the effluent of animal farms contributes to the presence of ESBL-producing E. coli in river aquatic environments [59]. A recent study in Zhejiang, China also showed that these mcr-1 plasmids in the river are closely associated with E. coli strains with pig and human origins [60].

IncX4 plasmids are one of the three major types of mcr-1-carrying plasmids, including IncX4, IncI2, and IncHI2, and account for over 90% of the reported mcr-1 distributed worldwide [61]. It has been reported that IncX4 plasmids confer competitive fitness to host bacteria, are more transmissible at 30–42 °C [62], and can be stably maintained in host bacteria [41]. These features may be responsible for the significant role of IncX4 plasmids in mcr-1 dissemination. The cessation of colistin use as a feed additive for animals in China has significantly decreased the prevalence of mcr-1 in farmed pigs nationally, including IncX4 plasmid-carrying mcr-1, which may contribute to a concomitant decline of the distribution of mcr-1 in human carriers [63]. These findings indicate that antibiotic selection pressure is a major driving force of mcr-1 dissemination; thus, the withdrawal of colistin from animal feeds would be an effective strategy for controlling the dissemination of mcr-1 in humans, animals, and the natural environment.

In our mcr-1 positive E. coli, no known virulence factor was detected, excluding the ompT gene, and all were classified as phylogenetic group A, suggesting the low virulence potential of these bacteria. Although they have low potential to cause infections, these riverine strains could serve as an environmental reservoir of colistin resistance for future spread to pathogenic strains through conjugative horizontal transfer [64].

The insertion of the Tn3 family transposon in pEC1281 was first identified in the pEC1279-like IncX4 plasmids, suggesting that transposon insertion is a local gene transfer event that may have occurred recently; thus, the resulting plasmid has not yet been broadly spread. Whether the genes encoded in the transposon contribute to plasmid transmission and stability, and whether their presence confers advantages to bacterial survival in natural environments, are yet to be elucidated. Transposons and insertion sequences contribute to the mobilization of antibiotic resistance genes [65]. The new transposon insertion in the mcr-1-carrying plasmid may further potentiate the dissemination of colistin resistance.

The limitations of this study were that we only detected mcr-1 in E. coli isolates in the river and we did not determine the mcr-1 status of other pathogens. No selective primary isolation of colistin resistant E. coli was performed, and the real occurrence of mcr-1 carrying E. coli could be much more prevalent. Other limitations included that the sampling sizes of rivers may differ and we did not have water level data for each sampling site. The water level of rivers varies significantly between the wet and dry seasons. We do not know if the water level is a factor for the spread of mcr-1 positive E. coli. We did not conduct longitudinal surveillance of mcr-1-positive E. coli in the river. Some studies suggest that the surveillance of population-level antibiotic resistance prevalence could be informative as an early warning of human pathogens [66]. Real-time water quality monitoring systems and removed/relocated livestock may aid in reducing pollutants from agricultural areas [67].

5. Conclusions

The ecological analysis and plasmid sequences suggest the spread of mcr-1 plasmids between livestock and the riverine environment. The mcr-1-positive E. coli isolates belonged to phylogroup A, with low virulence potential. The presence of colistin-resistant strains in rivers may lead to the spread of mcr-1 among commensal E. coli strains in the aquatic environment and pose a further public health risk. There is an indispensable need for the survey of natural habitats or associated environments to better understand the dissimilation of colistin-resistance genes.