Genomic patterns and characterizations of chromosomally-encoded mcr-1 in Escherichia coli populations

The emergence and rapid dissemination of plasmid-mediated mobile colistin resistance gene (mcr-1) have become a severe threat to public health [1]. The predominant carriers of mcr-1 were IncX4, IncI2, and IncHI2 plasmids, which are transferable and adaptive plasmid types with broad host range and contributed to the spread of mcr-1 among various sources and bacterial species [2‐4]. Besides, recombination of transposons, especially Tn6330 (ISApl1-mcr-1-pap2-ISApl1), the primary vehicle for transmission of mcr-1, and phage-like sequences enable mcr-1 to transfer across plasmids and isolates. Such contributed factors facilitated high mcr-1 prevalence in several sources around the world, pushing local governments in Europe, Brazil and China to prohibit the use of colistin as growth promoter additive for livestock [5‐8].

Accumulated evidence showed that banning of colistin in animal feed efficiently restricted mcr-1 prevalence, not only in animals but also in humans and the whole ecosystem in China [2‐4]. However, our previous study showed that a low proportion of Escherichia coli carrying chromosomally-encoded mcr-1 continually existed in the ecosystem [4], which was sporadically reported by other studies as well [9‐11]. On account of the plasmid that could be lost under certain circumstances due to instability, the chromosomally-encoded events could stabilize the heritage of mcr-1, threatening the intervention of colistin stewardship. In current study, we systematically investigate the epidemiological and genomic characterizations of E. coli population with chromosomally-encoded mcr-1.

Based on our previous large-scale epidemiological study from 2016 to 2018 in Guangzhou, China [4], we identified 24 (3.5%) out of 688 mcr-1-positive E. coli isolates with the chromosomally-encoded mcr-1 (Table 1). The prevalence of chromosomally-encoded mcr-1-positive E. coli was from 0 to 9.8% for each source and from 2.2 to 4.8% for each epoch, indicating that the chromosomally-encoded mcr-1 was at a low prevalence state in different dimensions (Table 1). Additionally, the comparison of prevalence for chromosomally-encoded mcr-1 between different niches or epochs showed no significant difference (Fisher’s exact test, p > 0.05 for each comparison), suggesting that the emergence of chromosomally-encoded mcr-1 was sporadic without temporal or source-specific signals.

To systematically illustrate the genomic basis of chromosomally-encoded mcr-1-positive E. coli population, we collected other 30 E. coli genomes with chromosomally-encoded mcr-1 from published literature for subsequent analysis (Additional file 1: Table S1). Through in silico multilocus sequence typing (MLST) assignment, 32 different sequence types (STs) within 10 ST complexes were determined (Fig. 1). The most common ST among chromosomally-encoded mcr-1-positive E. coli isolates was ST10 (n = 10, 18.5%), which is consistent with the main host for plasmid-mediated mcr-1 on E. coli species [3, 4, 12]. The phylogeny demonstrated two sequence clusters (SCs), except for two isolates which were distinct from two SCs as the outgroup (Fig. 1). The sources and serotypes of these genomes were scattered on the phylogeny, suggesting that the emergence of chromosomally-encoded mcr-1 was random without source- or lineage-based specificity (Fig. 1). Since most of the chromosomally-encoded mcr-1-positive E. coli isolates have been identified in China (n = 40, 74.1%), which was attributed to the extensive screening of mcr-1 in China, the associations between locations and SCs was ambiguous (SC1 [11/16] vs SC2 [29/36], Fisher’s exact test, p = 0.49).

The mcr-1 gene was initially found on plasmids in Enterobacteriaceae and on a transposon Tn6330, prompting that the chromosomally-encoded mcr-1 could come from recombination of plasmid segments or transposition of Tn6330 [13‐15]. Therefore, we investigated the plasmidome of 54 genomes to illustrate the potential origin of chromosomally-encoded mcr-1. We identified 33 plasmid Inc types among all isolates, and the results showed that the most common Inc type was IncFIB(K) (45.8%, n = 22), followed by IncColRNAI (43.8%, n = 21), IncHI1 (33.3%, n = 16), IncX1 (31.3%, n = 15), IncFIB (AP001918) (27.1%, n = 13), and IncY (20.8%, n = 10). Remarkably, the common Inc types of mcr-1-harboring plasmids, such as IncX4, IncI2, IncHI2, and IncpO111 [1, 3, 4, 12], were rarely detected among these isolates (Fig. 1), indicating that the chromosomally-encoded mcr-1 may derive from ISApl1/Tn6330 through transposition, but not from the plasmid.

We subsequently analyzed the genetic context of mcr-1 for each isolate to investigate the genetic model of chromosomally-encoded mcr-1, except seven isolates were excluded due to short mcr-1-harboring contigs. We found that most of the mcr-1 genes (93.6%, 44/47) were flanked by ISApl1, comprising 24 isolates harboring upstream ISApl1 and 20 isolates carrying composite Tn6330, which complied with the hypothesis of transposition-mediated chromosome insertion.

By mapping the insertion site onto the chromosome of E. coli MG1655, we noted that the distribution of chromosomally-encoded mcr-1 insertion sites was sporadic (Fig. 2a). Thirty-seven clusters of mcr-1-harboring segments were generated based on sequence clustering analysis (Fig. 2a), which included three clusters involving more than one isolates (Fig. 2b) and 34 clusters only containing a single isolate (Additional file 2: Figure S1). The most common genetic pattern of chromosomally-encoded mcr-1 (19.1%, 9/47) involves in an insertion segment in size of ~ 25.7 kb, containing an incomplete phage-like region (score = 40 for phage Vibrio 12B8 [NC_021073] by PHASTER) and a truncated Tn6330 (ISApl1-mcr-1-pap2), which was inserted into the E. coli genome between lysN and hicB (toxin-antitoxin system) loci (Fig. 2b). The incomplete phage-like region only contains head, tail, and fiber protein, and lacks some necessary functional components (Fig. 2b), which seems unfunctional under current conditions. We used BLASTn to search this phage-like sequence in NCBI non-redundant nucleotide database, and the results showed that only five sequences, which are located on E. coli chromosome, were identified with ≥ 60% coverage and ≥ 90% identity, indicating the correlation between chromosomally-encoded mcr-1 and such phage-like region. Collectively, we heuristically concluded that such a phage-like region could mediate the emergence of chromosomally-encoded mcr-1, and then the phage may lose the lysogenic components, stabilization the genetic inheritance of chromosomally-encoded mcr-1. Additionally, the mcr-1 of two isolates showed the insertion of mcr-1 located on an integrative element region and a plasmid segment respectively, suggesting that chromosomally-encoded mcr-1 could be derived from the integration of the integrative region and plasmid segment (Fig. 2c).

In conclusion, our study comprehensively investigated the genetic basis of chromosomally-encoded mcr-1 in prevalence and potential mechanisms of lineage, plasmid, insertion sequence, and phage. Our results showed that chromosomally-encoded mcr-1 was mainly derived from ISApl1 insertion in genomic locations sporadically. Notably, we reported a new transmission mechanism, a phage-like region without functional components, could associate with the emergence and stabilization of chromosomally-encoded mcr-1. The chromosomally-encoded mcr-1 in current situations seems not a severe threat for public health, however, it could be a heritable repository and thrive again if the new selective pressure emerges, because the chromosome-mediated antimicrobial resistance genes (ARGs) might be conferred with genetic sustainability. In-depth investigations are needed to illustrate the genomic and epidemiological dynamics of chromosomally-encoded mcr-1, which may be changed after the approval of colistin in human clinical therapeutics in China [16].

We searched PubMed using the terms of “mcr-1” [MeSH]/[All Fields] AND “chromosome” [MeSH]/[All Fields] AND “Escherichia coli” [MeSH]/[All Fields] for articles published before 1^th October 2020, and identified 20 publications, including 30 available E. coli genomes with chromosome-mediated mcr-1 (Additional file 3: Figure S2).