Characterization of a Large Antibiotic Resistance Plasmid Found in Enteropathogenic Escherichia coli Strain B171 and Its Relatedness to Plasmids of Diverse E. coli and Shigella Strains

ABSTRACT

Enteropathogenic Escherichia coli (EPEC) is a leading cause of severe infantile diarrhea in developing countries. Previous research has focused on the diversity of the EPEC virulence plasmid, whereas less is known regarding the genetic content and distribution of antibiotic resistance plasmids carried by EPEC. A previous study demonstrated that in addition to the virulence plasmid, reference EPEC strain B171 harbors a second, larger plasmid that confers antibiotic resistance. To further understand the genetic diversity and dissemination of antibiotic resistance plasmids among EPEC strains, we describe the complete sequence of an antibiotic resistance plasmid from EPEC strain B171. The resistance plasmid, pB171_90, has a completed sequence length of 90,229 bp, a GC content of 54.55%, and carries protein-encoding genes involved in conjugative transfer, resistance to tetracycline (tetA), sulfonamides (sulI), and mercury, as well as several virulence-associated genes, including the transcriptional regulator hha and the putative calcium sequestration inhibitor (csi). In silico detection of the pB171_90 genes among 4,798 publicly available E. coli genome assemblies indicates that the unique genes of pB171_90 (csi and traI) are primarily restricted to genomes identified as EPEC or enterotoxigenic E. coli. However, conserved regions of the pB171_90 plasmid containing genes involved in replication, stability, and antibiotic resistance were identified among diverse E. coli pathotypes. Interestingly, pB171_90 also exhibited significant similarity with a sequenced plasmid from Shigella dysenteriae type I. Our findings demonstrate the mosaic nature of EPEC antibiotic resistance plasmids and highlight the need for additional sequence-based characterization of antibiotic resistance plasmids harbored by pathogenic E. coli.

INTRODUCTION

There has been an alarming increase in the frequency of antibiotic resistance in bacterial pathogens. Multidrug-resistant Enterobacteriaceae are of particular concern and include some groups of disease-causing Escherichia coli (1–3). E. coli strains have been identified that are resistant to all routinely prescribed antibiotics, including colistin, which is a last resort treatment option for carbapenem-resistant Enterobacteriaceae (4–7). The most widely reported antibiotic-resistant E. coli strains are those belonging to multilocus sequence type (MLST) 131 (ST131), which are usually identified as extraintestinal pathogenic E. coli (ExPEC) and often exhibit resistance to fluoroquinolones, as well as extended-spectrum cephalosporins (1, 8). E. coli can also serve as a reservoir of antibiotic resistance genes, including mph(A), which confers resistance to macrolides such as azithromycin (9). E. coli strains from healthy volunteers in different geographic locations were also identified with resistance to antibiotics, including ampicillin, trimethoprim, and ciprofloxacin (10), and E. coli strains carrying tetracycline resistance genes were isolated from infants that had not received tetracycline (11). E. coli resistance genes and elements have been demonstrated to exist in fecal E. coli reservoirs and spread internationally for decades without detection (12).

Previous studies have also demonstrated that other pathotypes of disease-causing E. coli, such as the enteropathogenic E. coli (EPEC), can exhibit resistance to one or more antibiotics (13–17). One study demonstrated that EPEC strains from Brazil exhibited resistance to tetracycline, sulfonamides, ampicillin, chloramphenicol, and streptomycin (18). EPEC is one of six diarrheagenic pathotypes (EPEC, enterohemorrhagic E. coli [EHEC], enterotoxigenic E. coli [ETEC], enteroaggregative E. coli [EAEC], enteroinvasive E. coli [EIEC], and diffusely adherent E. coli [DAEC]) of disease-causing E. coli that cause human gastrointestinal illness (19–21). EPEC was also recently characterized in the Global Enteric Multisite Study as a leading cause of lethal diarrheal illness among infants in developing countries (22). Similar to the other E. coli pathotypes, EPEC can be molecularly classified by the presence of canonical virulence factors that are associated with mechanisms of pathogenesis or by phenotypes that are unique to each of the E. coli pathotypes (19–21). EPEC and EHEC are collectively called the attaching and effacing E. coli (AEEC) due to the chromosomally encoded locus of enterocyte effacement (LEE) pathogenicity island, which contains the intimin protein, involved in adherence, and a type III secretion system that delivers effector proteins into host cells (19, 20, 21, 23, 24). In addition to the LEE region, EHEC contains the Shiga toxin, which is carried by a phage and is absent from EPEC (20, 25). Meanwhile, EPEC strains are subclassified as typical EPEC (tEPEC) strains that contain the plasmid-encoded bundle-forming pilus (BFP) or atypical EPEC (aEPEC) strains that contain the LEE region but do not have the Shiga toxin genes or BFP genes (21).

E. coli plasmids and phage carry many of the virulence genes that contribute to differences in pathogenesis, and these virulence genes are used to distinguish among the different E. coli pathotypes (19, 20, 26). Many of the virulence plasmids and virulence-associated phage of E. coli have been sequenced and described in detail for the E. coli pathotypes, including EHEC, EPEC, ETEC, and EAEC (26–33). The EPEC adherence factor (EAF) virulence plasmid, which carries genes encoding the BFP of tEPEC, has been sequenced and described for several reference strains, including EPEC strains E2348/69 (29, 34) and B171 (28). In a previous study, we sequenced and compared the EAF plasmids from genomically diverse EPEC strains, demonstrating considerable differences in the gene content and stability of this canonical tEPEC virulence plasmid (35).

In contrast to our knowledge of the diversity of EPEC virulence plasmids, much less is known about the sequence diversity and distribution of EPEC antibiotic resistance plasmids. A large multidrug resistance plasmid from EPEC strain MB80 was previously characterized using subtractive hybridization (36). The traI and traC genes of the conjugative transfer system were absent from the virulence plasmid of EPEC reference strain E2348/69; however, they were detected in other EPEC strains, including the EPEC reference strain B171 (30, 36, 37). While portions of the MB80 plasmid were sequenced (36) and the draft genome sequence of EPEC strain B171 has been generated (30), the complete sequence of the resistance plasmid from these EPEC strains has not been previously described. Thus, to increase our understanding of EPEC antibiotic resistance plasmids, we sequenced the large, ∼90-kb antibiotic and heavy metal resistance plasmid, pB171_90, from the EPEC reference strain B171, which was previously determined to confer resistance to tetracycline, streptomycin, chloramphenicol, and sulfonamides (30, 36, 37). We describe the distribution of the pB171_90 genes among a publicly available collection of 4,798 E. coli genome assemblies in GenBank, which includes E. coli of diverse pathotypes as well as commensal E. coli that lack canonical pathotype-specific virulence genes. We also investigated whether the pB171_90 plasmid genes may have been disseminated to other species of enteric bacteria. Thus, the current study provides a comprehensive description of the genetic content of an antibiotic resistance plasmid from a tEPEC reference strain as well as the distribution of the plasmid genes among a diverse collection of E. coli and other enteric pathogens.

RESULTS

Sequence characterization of the large antibiotic resistance plasmid, pB171_90, from EPEC strain B171.

In the current study, we describe the complete sequence of the large antibiotic resistance plasmid from EPEC strain B171 to gain additional insight into its genetic content and potential role in antibiotic resistance and/or pathogenesis. We have designated this plasmid pB171_90 (GenBank accession number CP021212), as it is from EPEC strain B171 and has a sequence length of 90,229 bp. Plasmid pB171_90 has a GC content of 54.55% and contains 108 predicted protein-coding genes (Fig. 1). The predicted functions of the pB171_90 protein-coding genes include conjugative transfer, plasmid stability, and antibiotic resistance (Table 1; Fig. 1). Among the pB171_90 antibiotic resistance genes are a spectinomycin-streptomycin resistance gene (aadA1, pB171_90_88) (GenBank accession number NG_052027.1) (38), tetA (pB171_90_95) encoding a tetracycline efflux pump, and tetR (pB171_90_94) encoding the tetA repressor (39–41), a multidrug efflux pump (qacEΔ1, pB171_90_89) (GenBank accession number L06822.4) (42), and a dihydropteroate synthase gene, which can confer resistance to sulfonamides (sul1, pB171_90_90) (GenBank accession number NC_006671.1) (43) (Table 1). The genes for spectinomycin-streptomycin resistance (aadA1), sulfonamide resistance (sul1), and the qacEΔ1 efflux pump form a class 1 integron that includes a class 1 integrase gene, intI1 (GenBank accession number NC_007100.1), and is located within a region (coordinates 74,154 to 77,947) that has 100% nucleotide identity to a previously described Tn21 transposon from E. coli (GenBank accession number X12870.1) (44). Also carried on pB171_90 is a mer operon (pB171_90_99 to pB171_90_105), which contains all of the mer genes conferring narrow-spectrum mercury resistance, but the plasmid is lacking merB and merG, which are involved in organomercury detoxification and broad-spectrum mercury resistance (45).

FIG 1.

Circular plot of pB171_90. The outermost track of the plot contains the protein-coding genes on the positive strand, while the adjacent track contains the protein-coding genes on the negative strand. The predicted protein functions are indicated by colors as follows: replication (blue), virulence (purple), antibiotic resistance (red), conjugative transfer (orange), plasmid stability (green), transposition (yellow), and unknown (gray). The red line is the GC content along a 100-bp sliding window. The blue line is the GC skew along a 100-bp sliding window, with light blue indicating a positive skew and dark blue indicating a negative skew. The circular plot was generated using Circos 0.69-3 (80).

TABLE 1.

Annotation of the EPEC strain B171 resistance plasmid, pB171_90

Gene ID or CDSb	Gene	Predicted protein function	Coding strand	Coordinates (bp)		No. of E. coli genomes (%)a
				Start	Stop
pB171_90_1		CAAX protease self-immunity family protein	−	850	197	956 (19.92)
pB171_90_2	repA	IncFII family plasmid replication initiator RepA	−	2,649	1,792	3,214 (66.99)
pB171_90_3	repB	Replication regulatory RepB family protein	−	3,214	2,954	435 (9.07)
pB171_90_4		Hok/Gef family protein	−	3,704	3,498	1,359 (28.32)
pB171_90_5	hha	Hemolysin expression-modulating protein Hha	−	4,202	3,993	2,209 (46.04)
pB171_90_6		Conserved hypothetical protein	−	4,494	4,258	33 (0.69)
pB171_90_7		Conserved hypothetical protein	−	5,594	4,581	57 (1.19)
pB171_90_8		Putative gp46	−	6,009	5,701	58 (1.21)
pB171_90_9	csi	Putative Csi protein	−	6,716	6,006	31 (0.65)
pB171_90_10		Conserved hypothetical protein	−	7,066	6,713	31 (0.65)
pB171_90_11	traX	Type F conjugative transfer system pilin acetylase TraX	−	7,839	7,108	147 (3.06)
pB171_90_12		Conserved hypothetical protein	−	8,402	7,836	151 (3.15)
pB171_90_13	traI	Conjugative transfer relaxase protein TraI	−	13,717	8,420	134 (2.79)
pB171_90_14	traD	Type IV conjugative transfer system coupling protein TraD	−	15,948	13,717	135 (2.81)
pB171_90_15	traT	Enterobacterial TraT complement resistance family protein	−	16,802	16,092	3,017 (62.88)
pB171_90_16	traS	Putative protein TraS	−	17,549	17,004	71 (1.48)
pB171_90_17	traG	TraG-like, N-terminal region family protein	−	20,474	17,571	134 (2.79)
pB171_90_18		Conjugative relaxosome accessory transposon family protein	−	21,837	20,476	135 (2.81)
pB171_90_19	trbB	Type F conjugative transfer system pilin assembly thiol-disulfide isomerase TrbB	−	22,445	21,843	135 (2.81)
pB171_90_20	traF	Type F conjugative transfer system pilin assembly protein TraF	−	23,248	22,469	134 (2.79)
pB171_90_21	traN	Type F conjugative transfer system mating-pair stabilization protein TraN	−	25,059	23,245	133 (2.77)
pB171_90_22	trbC	Type F conjugative transfer system pilin assembly protein TrbC	−	25,693	25,013	132 (2.75)
pB171_90_23		Conserved hypothetical protein	−	26,058	25,690	113 (2.36)
pB171_90_24	traU	TraU family protein	−	27,055	26,069	2,118 (44.14)
pB171_90_25	traW	Type F conjugative transfer system protein TraW	−	27,687	27,052	138 (2.88)
pB171_90_26	trbI	Type F conjugative transfer system protein TrbI	−	28,151	27,687	132 (2.75)
pB171_90_27	traC	Type IV secretion system protein TraC	−	30,739	28,148	134 (2.79)
pB171_90_28	traV	Type IV conjugative transfer system protein TraV	−	31,349	30,750	137 (2.86)
pB171_90_29		Conserved hypothetical protein	−	31,819	31,409	99 (2.06)
pB171_90_30		Conserved hypothetical protein	−	31,988	31,803	102 (2.13)
pB171_90_31	traR	Prokaryotic DksA/TraR C4-type zinc finger family protein	−	32,216	31,992	140 (2.92)
pB171_90_32	trbI	Bacterial conjugation TrbI-like family protein	−	33,692	32,328	136 (2.83)
pB171_90_33	traK	Type F conjugative transfer system secretin TraK	−	34,419	33,679	138 (2.88)
pB171_90_34	traE	Type IV conjugative transfer system protein TraE	−	34,972	34,409	167 (3.48)
pB171_90_35	traL	Type IV conjugative transfer system protein TraL	−	35,297	34,992	178 (3.71)
pB171_90_36	traA	Type IV conjugative transfer system pilin TraA	−	35,658	35,299	193 (4.02)
pB171_90_37	traJ	Putative TraJ protein	−	36,419	36,138	97 (2.02)
pB171_90_38	traM	Relaxosome TraM domain protein	−	37,283	36,783	71 (1.48)
pB171_90_39		Integrase core domain protein	+	38,211	38,504	29 (0.60)
pB171_90_40		Putative membrane protein	−	38,880	38,665	257 (5.36)
pB171_90_41		Conserved hypothetical protein	−	39,463	39,221	363 (7.57)
pB171_90_42		Conserved hypothetical protein	−	39,786	39,505	2,115 (44.08)
pB171_90_43		Transposase family protein	+	40,493	41,143	497 (10.36)
pB171_90_44		Putative transposase	+	41,143	41,490	4,145 (86.39)
pB171_90_45		Transposase C of IS166 homeodomain protein	+	41,510	43,081	1,225 (25.53)
pB171_90_46		Hok/Gef family protein	−	43,577	43,416	2,991 (62.34)
pB171_90_47		Conserved hypothetical protein	−	44,099	43,785	559 (11.65)
pB171_90_48	psiA	PsiA family protein	−	44,815	44,096	2,803 (58.42)
pB171_90_49	psiB	Protein PsiB	−	45,246	44,812	3,155 (65.76)
pB171_90_50		ParB-like nuclease domain protein	−	47,268	45,301	2,661 (55.46)
pB171_90_51		Conserved hypothetical protein	−	47,563	47,330	3,119 (65.01)
pB171_90_52	ssbF	Plasmid-derived single-stranded DNA-binding protein	−	48,147	47,620	3,080 (64.19)
pB171_90_53		Putative YddA protein	+	48,630	48,875	1,310 (27.30)
pB171_90_54		Methyltransferase small domain protein	−	49,524	48,961	2,546 (53.06)
pB171_90_55		Conserved hypothetical protein	−	50,930	49,569	2,554 (53.23)
pB171_90_56		Putative conserved predicted protein	−	51,212	50,982	2,622 (54.65)
pB171_90_57		Conserved hypothetical protein	+	51,475	51,732	2,354 (49.06)
pB171_90_58		Putative YchA	−	52,404	52,213	3,245 (67.63)
pB171_90_59		Conserved hypothetical protein	−	52,823	52,401	3,210 (66.90)
pB171_90_60		Antirestriction family protein	−	53,295	52,870	3,241 (67.55)
pB171_90_61		Putative YciA	−	53,705	53,544	2,014 (41.98)
pB171_90_62		Putative YfdA	−	54,478	53,708	2,911 (60.67)
pB171_90_63		Conserved hypothetical protein	−	54,957	54,523	3,120 (65.03)
pB171_90_64		Conserved hypothetical protein	−	55,192	54,971	3,039 (63.34)
pB171_90_65		DNA methylase family protein	−	55,876	55,193	3,149 (65.63)
pB171_90_66		Putative YfbA	−	56,203	55,952	1,906 (39.72)
pB171_90_67		Conserved hypothetical protein	−	57,187	56,261	2,675 (55.75)
pB171_90_68	parM	Plasmid segregation protein ParM	−	58,977	57,997	385 (8.02)
pB171_90_69		Conserved hypothetical protein	−	59,932	59,366	347 (7.23)
pB171_90_70		Bacterial regulatory, TetR family protein	+	60,580	61,137	144 (3)
pB171_90_71		Conserved hypothetical protein	+	61,446	61,736	142 (2.96)
pB171_90_72		Conserved hypothetical protein	+	61,768	62,694	143 (2.98)
pB171_90_73		Conserved hypothetical protein	+	62,663	63,421	141 (2.94)
pB171_90_74		DoxX family protein	+	63,423	63,950	141 (2.94)
pB171_90_75		Conserved hypothetical protein	−	64,645	64,283	123 (2.56)
pB171_90_76		Hypothetical protein	+	64,852	64,971	395 (8.23)
pB171_90_77	def	Peptide deformylase	−	65,655	65,146	4,764 (99.29)
pB171_90_78		Hypothetical protein	+	65,785	65,895	144 (3)
pB171_90_79		Conserved hypothetical protein	+	66,183	66,548	1,079 (22.49)
pB171_90_80		Transposase zinc-binding domain protein	+	66,548	67,015	1,440 (30.01)
pB171_90_81		Transposase family protein	+	67,109	67,735	906 (18.88)
pB171_90_82		Transposase family protein	−	68,490	68,125	3,488 (72.70)
pB171_90_83		Conserved hypothetical protein	−	69,867	68,692	71 (1.48)
pB171_90_84		Conserved hypothetical protein	−	73,048	70,082	797 (16.61)
pB171_90_85		Resolvase, N-terminal domain protein	−	73,611	73,051	865 (18.03)
pB171_90_86		Conserved hypothetical protein	−	74,087	73,737	794 (16.55)
pB171_90_87	intI1	Integron integrase family protein	−	75,303	74,290	1,093 (22.78)
pB171_90_88	aadA1	Streptomycin 3''-adenylyltransferase	+	75,452	76,243	569 (11.86)
pB171_90_89	qacEΔ1	Small multidrug resistance family protein	+	76,407	76,754	957 (19.95)
pB171_90_90	sul1	Dihydropteroate synthase	+	76,748	77,587	942 (19.63)
pB171_90_91		Putative Achromobacter xylosoxidans subsp. denitrificans plasmid pAX22, partial DNA for integron In70	+	77,784	78,041	345 (7.19)
pB171_90_92		Transposase DDE domain protein	+	78,944	79,366	57 (1.19)
pB171_90_93		Putative relaxase/helicase	+	80,191	80,433	897 (18.70)
pB171_90_94	tetR	Tetracycline repressor protein class B from transposon Tn10	−	81,115	80,465	956 (19.92)
pB171_90_95	tetA	Tetracycline resistance protein, class C	+	81,221	82,420	1,008 (21.01)
pB171_90_96		Multidrug resistance efflux transporter family protein	−	83,234	82,452	841 (17.53)
pB171_90_97		Helix-turn-helix domain protein	−	84,004	83,297	809 (16.86)
pB171_90_98		EAL domain protein	−	84,786	84,079	702 (14.63)
pB171_90_99	merE	MerE family protein	−	85,019	84,783	715 (14.90)
pB171_90_100	merD	Mercuric resistance transcriptional repressor protein MerD	−	85,378	85,016	694 (14.46)
pB171_90_101	merA	Mercuric reductase	−	87,090	85,396	683 (14.24)
pB171_90_102	merC	MerC mercury resistance family protein	−	87,564	87,142	685 (14.28)
pB171_90_103	merP	Mercuric transport protein periplasmic component	−	87,875	87,600	720 (15.01)
pB171_90_104	merT	MerT mercuric transport family protein	−	88,239	87,889	733 (15.28)
pB171_90_105	merR	Hg(II)-responsive transcriptional regulator	+	88,311	88,745	757 (15.78)
pB171_90_106		Putative transposase	−	89,118	88,852	3,807 (79.35)
pB171_90_107		Hypothetical protein	+	89,301	89,516	3,793 (79.05)
pB171_90_108	pemK	mRNA interferase PemK	−	90,074	89,742	1,073 (22.36)

^aThe number and percentage of publicly available E. coli genomes that contained each gene with a BSR of ≥0.7. The total number of E. coli genome assemblies that were available in GenBank as of November 2016 was 4,798.

^bCDS, protein-coding genes.

The resistance plasmid, pB171_90, also carries two genes that were present on the EPEC resistance plasmid and not on the EPEC virulence plasmid and were previously suggested to have a role in pathogenesis (36). One potential pathogenicity gene is csi, a putative calcium sequestration inhibitor (36), whose function in virulence is poorly understood. The second virulence-associated gene of pB171_90, hha, encodes a transcriptional regulator of known virulence genes, including a hemolysin (46), genes involved in adherence and biofilm formation (47, 48), and genes involved in type III secretion (49). The hha gene has been identified on the chromosome and a virulence plasmid of E. coli O157:H7 Shiga toxin-producing E. coli (STEC) strains (50).

While sequencing the resistance plasmid of E. coli strain B171 using the Pacific Biosciences (PacBio) platform, we also took the opportunity to resequence the B171 EAF plasmid (here referred to as the virulence plasmid). The original sequence of the pB171 virulence plasmid was previously assembled from Sanger-sequenced plasmid subclones produced by different groups nearly 2 decades ago (28). To distinguish the PacBio-sequenced version of the plasmid from the original version, which is designated pB171 (GenBank accession number AB024946.1) (28), we have named the newly PacBio-sequenced version of the B171 virulence plasmid pB171_69 (GenBank accession number CP021211). The PacBio-sequenced virulence plasmid, pB171_69, has a GC content of 46.02% and a sequence length of 68,814 bp. The pB171_69 plasmid sequence is nearly identical to the original virulence plasmid, pB171, which has a GC content of 46% and a sequence length of 68,817 bp (28). Comparison of the two virulence plasmid sequences demonstrated that there were 21 single nucleotide polymorphisms (SNPs) and seven insertions or deletions in pB171_69 compared to the original version of the plasmid (see Table S2 in the supplemental material). These differences were identified in only 7 of the 80 previously described protein-coding genes (Table S2) (28). Overall, this comparative analysis demonstrates that the B171 virulence plasmids that were sequenced nearly 2 decades apart using different sequencing technologies are remarkably similar. Further investigation is necessary to determine whether these sequence differences are caused by the sequencing technology or whether they represent genetic changes that have been acquired by different laboratory strains of B171 over time and may influence any phenotypes of B171.

Distribution of the pB171_90 genes among diverse E. coli genomes.

To determine the distribution of the pB171_90 genes among diverse E. coli genomes, we used large-scale BLAST score ratio (LS-BSR) analysis to conduct an in silico screen of 4,798 E. coli genome assemblies available in GenBank as of November 2016 for the presence of pB171_90 genes. LS-BSR analysis demonstrated that some of the genes, such as pB171_90_77 encoding a peptide deformylase, were present in nearly all of the E. coli genomes (4,764 of 4,798 genomes, 99%) (Table 1). Meanwhile, genes that were previously described as being unique to the B171 strain and a few other related EPEC strains had a narrow distribution among the 4,798 E. coli genomes analyzed. The csi gene encoding a putative calcium sequestration inhibitor (GenBank accession number Y08258) (36) was, prior to this study, identified as unique in the B171 strain, and the traI gene encoding a conjugative transfer relaxase protein had only been reported from EPEC resistance plasmids (36). In this study, a much broader and sequence-based analysis identified the csi gene with significant similarity (BSR ≥ 0.7) in 31 of the 4,798 E. coli genomes (0.65%), including B171 (Table 2). Unlike the very narrow distribution of the csi gene, the traI gene was detected in more but still relatively few E. coli genomes (134 genomes total; 2.79% of the E. coli genomes) that did not contain csi (Table 2). Also, both csi and traI were significantly more prevalent among the AEEC genomes, such as the EPEC, compared to the non-AEEC genomes (chi-squared P value < 0.05) (Table 2). All of the E. coli genomes that contained csi also contained the traI gene, except for the genome of EPEC strain 302014, which contained csi but not traI (see Table S1 in the supplemental material). Interestingly, while several of the pB171_90 genes were significantly more associated with AEEC genomes (IncFII repA, hha, csi, and traI genes), the antibiotic resistance genes (tetA, sul1, and mer genes) were more frequently associated with the non-AEEC genomes (Table 2). These results highlight the mosaic nature of these plasmids, with the resistance genes having a different distribution from other genes, such as csi and hha, that are unique to pB171_90 and related plasmids.

TABLE 2.

In silico detection of select pB171_90 genes in AEEC genomes versus other E. coli

Gene	No. of genomes (%)a			P value (non-AEEC vs AEEC)e
	All E. colib	Non-AEECc	AEECd
repA	3,214 (66.99)	2,225 (65.89)	989 (69.6)	0.01379
hha	2,209 (46.04)	1,306 (38.67)	903 (63.55)	<2.2e−16
csi	31 (0.65)	9 (0.27)	22 (1.55)	1.16e−06
traI	134 (2.79)	83 (2.46)	51 (3.59)	0.03796
tetR	956 (19.92)	780 (23.1)	176 (12.39)	<2.2e−16
tetA	1,008 (21.01)	806 (23.87)	202 (14.22)	9.04e−14
sulI	942 (19.63)	802 (23.75)	140 (9.85)	<2.2e−16
mer (≥4 genes)f	719 (14.99)	594 (17.59)	125 (8.8)	9.44e−15

^aThe total number of genomes that contained each of the pB171_90 genes listed. The number in parentheses is the percentage of genomes with each gene.

^bThe total number of E. coli genomes analyzed was 4,798. These were all of the E. coli assemblies available in GenBank as of November 2016.

^cThe total number of non-AEEC genomes analyzed was 3,377.

^dThe total number of AEEC genomes analyzed was 1,421. The AEEC genomes were identified by in silico analysis as those that had the LEE gene escV.

^eThe numbers of non-AEEC versus AEEC genomes that contained each gene were compared using the chi-square test.

^fThe number of genomes that contained ≥4 of the 7 total genes in the pB171_90 mer operon.

We further investigated the genome-wide diversity of the E. coli strains that contain the unique genes of pB171_90 (csi and traI) by constructing a whole-genome phylogeny. Phylogenomic analysis of the csi- and/or traI-containing genomes with a collection of diverse E. coli and Shigella reference genomes demonstrated that the csi- and/or traI-containing E. coli genomes occurred in five of the six E. coli phylogroups (A, B1, B2, D, and E) (Fig. 2). Overall, the genomes analyzed in the phylogenomic analysis grouped together into clades by pathotype (Fig. 2). The csi gene, which was among the pB171_90 genes that had the narrowest distribution, was identified in genomes belonging to four phylogroups, but these genomes were concentrated in multiple clades or doublets of ETEC or EPEC genomes (Fig. 2). EPEC strain B171 was present in a subclade of the EPEC2 lineage (51) that also includes other csi-containing EPEC genomes (Fig. 2). Interestingly, among the STEC genomes that contained the unique pB171_90 genes csi and traI were two strains (K9107F and K9_45F) that were phylogenomically related to genomes of O157:H7 strains and were determined by molecular serotyping to have the O157:H7 serotype (Fig. 2; Table S1). The findings from the phylogenomic analysis demonstrate that csi and traI have been acquired by genomically diverse E. coli and are not restricted to EPEC strains that are closely related to strain B171.

FIG 2.

Phylogenomic analysis of E. coli genomes that carry genes with similarity to those of pB171_90. A SNP-based phylogeny was generated using ISG as previously described (35, 70). There were 149,150 conserved SNP sites identified among all of the E. coli genomes analyzed relative to the reference E. coli strain IAI39 (GenBank accession number NC_011750.1) that were used to construct a maximum-likelihood phylogeny with 100 bootstrap values using RAxML v.7.2.8 (71). The phylogeny was midpoint rooted and labeled using iTOL v.3 (72). Bootstrap values of ≥80 are designated by a gray circle. The scale bar indicates the approximate distance of 0.01 nucleotide substitutions per site. The E. coli phylogroups B1, A, B2, D, E, and F are denoted by bold letters. Symbols designate the presence of pB171_90 genes in each genome with tetA and tetR both represented by green stars (both genes, solid green star; one gene, open green star). Colors of the genome labels indicate the presumptive pathotype that was determined by in silico detection of canonical virulence genes.

In silico detection of the 108 protein-coding genes of pB171_90 in the csi- and/or traI-containing genomes demonstrated that nearly all of the pB171_90 genes were present with significant similarity (BSR ≥ 0.7) in EPEC genomes 102550 and 103447 (Fig. 3). In addition, the genome of ETEC strain TW10722 contained nearly all of the pB171_90 genes but was lacking several genes from the mer operon (Fig. 3). The frequencies of genes in the different plasmid regions (putative virulence genes hha and csi, conjugative transfer genes, and resistance genes) were demonstrated in the clustered heatmap (Fig. 3). Since all but one of the genomes analyzed contain the traI gene, it was anticipated that these genomes would also contain most of the other genes involved in conjugative transfer (Fig. 3, green rectangle). The csi-containing genomes grouped together into several clusters (Fig. 3, yellow boxes); however, only two of the clusters also contained some or all of the resistance genes (Fig. 3, red boxes). The resistance genes were present mostly in genomes of three different clusters in the heatmap, of which two of the clusters also contained the csi and hha genes while the third cluster that did not (Fig. 3, red boxes). These results demonstrate that the unique genes of the pB171_90 plasmid (csi and hha) do not travel exclusively with the antibiotic resistance genes of the pB171_90 plasmid, highlighting the mosaic nature of pB171_90 and related plasmids.

FIG 3.

In silico detection of pB171_90 genes in diverse E. coli genomes. The presence or absence of all predicted protein-coding genes of pB171_90 in the 134 E. coli genomes that contained the pB171_90 genes csi and/or traI is indicated by the BSR values in the heatmap (see Table S1 in the supplemental material). The clustered heatmap was generated using the heatmap2 function of gplots v.3.0.1 in R v.3.3.2. The genomes were clustered using the complete linkage method with Euclidean distance estimation. The pB171_90 genes are represented by the rows, while each column represents a different E. coli genome. The predicted pathotype of each E. coli genome is indicated by the color of the rectangle at the top of the heatmap (see legend for detail).

Phylogenetic analysis of the csi nucleotide sequences demonstrated that csi sequences from all but three of the genomes (aEPEC strains VL115 and CFSAN026818 and E. coli strain TA280) were present in two clades that we have designated clades I and II (Fig. 4). Clade I contained all of the csi sequences from tEPEC genomes, four of the csi sequences from aEPEC genomes, and one csi sequence from an ETEC genome (Fig. 4). The genomes of clade I belonged to phylogroups A, B1, and B2 and had at least 10 different MLSTs, while clade II contained csi sequences from six out of seven (85.7%) of the ETEC genomes and both of the STEC genomes (Fig. 4). Clade I genomes belonging to phylogroups A and E had four different MLSTs (Fig. 4). Thus, the csi genes exhibit the greatest similarity among genomes belonging to the same E. coli pathotype rather than from genomes of the same E. coli phylogroup.

FIG 4.

Phylogenetic analysis of csi. The csi nucleotide sequences of pB171_90 and other E. coli genomes were aligned using ClustalW. The alignment was used to construct a maximum-likelihood phylogeny with the Kimura 2-parameter model and 1,000 bootstraps using MEGA7 (79). The scale bar represents the approximate distance of 0.001 nucleotide substitutions per site. Bootstrap values of ≥50 are indicated by a gray circle. The predicted E. coli pathotypes are indicated by the color of their genome label (see legend), and the phylogroup and MLST of each genome are denoted in parentheses.

Phylogenetic analysis of the traI gene sequences demonstrated that there was greater pathotype specificity than phylogroup specificity (Fig. 5). There were clades of traI sequences from ETEC genomes of different phylogroups (A or B1) and other clades that contained nearly all EPEC genomes also from multiple phylogroups (A, B1, and B2). The genomes that formed clades I and II in the csi phylogeny formed similar clades in the traI phylogeny (Fig. 5). The consistent sequence similarity of the csi and traI genes occurring in the same genomes, along with the close proximity of csi (pB171_90_9) and traI (pB171_90_13) in the pB171_90 plasmid (Table 1), suggests that the csi and traI genes travel together.

FIG 5.

Phylogenetic analysis of traI. The traI nucleotide sequences of pB171_90 and other E. coli strains were aligned using ClustalW. The alignment was used to construct a maximum-likelihood phylogeny with the Kimura 2-parameter model and 1,000 bootstraps using MEGA7 (79). The scale bar represents the approximate distance of 0.02 nucleotide substitutions per site. Bootstrap values of ≥50 are indicated by a gray circle. The predicted E. coli pathotypes are indicated by the color of their genome label (see legend), and the phylogroup and MLST of each genome is denoted in parentheses. The blue vertical lines indicate the two clades identified in the csi phylogeny (Fig. 4).

The replication protein-coding gene of pB171_90, repA (pB171_90_2), was also significantly more associated with the AEEC genomes (69%) than with the non-AEEC genomes (65%), with a chi-squared P value of 0.013 (Table 2). The pB171_90 repA gene is most similar to the IncFIIA repA gene that has been identified in previously sequenced E. coli virulence plasmids, including plasmids from AEEC genomes (26, 35). Based on the IncF replicon typing scheme of Villa et al., the pB171 virulence plasmid belongs to the F13:A⁻:B4 IncF replicon type (52), while the resistance plasmid pB171_90 belongs to the F79:A⁻:B⁻ replicon type as determined by plasmid multilocus sequence typing (pMLST) analysis (53). Alignment of the repA nucleotide sequences from plasmids pB171_90 and pB171_69 indicated that they have 95% nucleotide identity (816/858 identical bases). Thus, both plasmids have related, but distinct, plasmid replicon types. In a previous study investigating the diversity of AEEC plasmids, we also identified multiple IncFII replicon types in the same genome (35). In contrast to the formation of clades I and II by the csi-containing genomes in both the csi and traI phylogenies, phylogenetic analysis of repA sequences demonstrated that the similarity of the repA gene sequences differs among some of the csi-containing genomes. The repA sequences from genomes in clade I of the csi and traI phylogenies formed a similar clade, while the genomes from clade II did not form a clade and were instead distributed throughout the phylogeny (see Fig. S1 in the supplemental material). Interestingly, the genomes of clade I were primarily tEPEC and aEPEC genomes with one ETEC genome, while the genomes of clade II were ETEC and STEC (Fig. 4 and 5; Fig. S1). This suggests that the csi and traI genes were likely acquired as part of the same plasmid in the EPEC strains, whereas the csi and traI genes of the STEC and ETEC genomes are associated with plasmids that contain a different IncFII replicon type.

Another gene of interest on pB171_90 is the virulence gene transcriptional regulator, hha, which was identified in 46% of the E. coli genomes and was significantly more associated with AEEC genomes over non-AEEC genomes (chi-squared P value < 2.2e−16) (Tables 1 and 2). Similar to the repA gene, phylogenetic analysis of the hha genes identified in the csi and/or traI-containing genomes demonstrated that the clade I genomes from the csi phylogeny also grouped together in the hha phylogeny, while the clade II genomes were distributed throughout the phylogeny (see Fig. S2 in the supplemental material). Interestingly, the hha genes identified in the csi and/or traI-containing genomes were more similar to the hha genes from plasmid pB171_90 and the E. coli O157:H7 plasmid pO157 from strain EDL933 (Fig. S2). These findings suggest that the hha genes identified in the clade I genomes may be located on plasmids containing the repA, csi, and traI genes, demonstrating that plasmids with similarity to pB171_90 exist among genomically diverse EPEC strains (Fig. S2).

Identification of pB171_90 genes in the genomes of other human enteric pathogens.

We further investigated whether unique genes of pB171_90, such as csi, were present in other species of bacteria by using BLASTP to compare the predicted amino acid sequence of the Csi peptide to the nonredundant (nr) protein sequence database of GenBank. This query identified 16 genome or plasmid sequences of enteric pathogens other than E. coli that contained a region encoding a protein with similarity to Csi (Fig. 6; Table S1). Among these non-E. coli csi-containing genomes were a Shigella dysenteriae type I plasmid, pCAR10 (GenBank accession number KT754161.1) (54), and whole-genome sequences of six Shigella sonnei, two Citrobacter freundii, one Klebsiella pneumoniae, and six Salmonella enterica strains (Fig. 6).

FIG 6.

In silico detection of pB171_90 genes in other human enteric pathogens. The predicted protein-coding genes of pB171_90 were identified in additional human enteric pathogens that contained the pB171_90 gene csi and are listed in Table S1 in the supplemental material. The clustered heatmap was generated using the heatmap2 function of gplots v.3.0.1 in R v.3.3.2. The genomes were clustered using the complete linkage method with Euclidean distance estimation. The pB171_90 genes are represented by the rows, while each column represents a different genome. The species are designated by the color of the rectangle at the top of the heatmap (see legend).

Phylogenetic analysis of the csi genes from the E. coli genomes compared with the csi genes that were identified in other human enteric pathogens demonstrated that the csi genes formed three different phylogenetic groups (see Fig. S3 in the supplemental material). Group 1 contained all of the csi genes from the E. coli genomes as well as those from the Shigella genomes (S. dysenteriae and S. sonnei) and the single sequence from a Klebsiella genome (Fig. S3). Group 2 contained only the csi sequences from Salmonella genomes, while group 3 was made up of the csi sequences from the Citrobacter genomes (Fig. S3). The csi phylogenetic analysis demonstrates that there is species specificity among the csi genes.

In silico detection of the 108 protein-coding genes of pB171_90 in each of the csi-containing enteric species genomes demonstrated that >50% of the plasmid genes were present with significant similarity (BSR ≥ 0.7) in the S. dysenteriae type I plasmid pCAR10 and also in the K. pneumoniae genome and two of the S. sonnei genomes (Fig. 6). The Tn21 transposon of pB171_90 that contains the aadA1, sul1, and qacEΔ1, was also detected with 100% nucleotide identity in sequenced plasmids and genomes from other species, including several plasmids from S. dysenteriae (p69-3818, GenBank accession number KT754167.1; p92-9000, GenBank accession number KT754166.1; and p80-547, GenBank accession number KT754160.1).

Comparison of the predicted protein-coding genes of the S. dysenteriae type I plasmid, pCAR10 (GenBank accession number KT754161.1) (54), with those of pB171_90, highlighted the significant similarity of these two antibiotic resistance carrying plasmids both in terms of gene content and order and also their levels of nucleotide similarity (Fig. 6; see also Fig. S4 in the supplemental material). Interestingly, the replication protein-coding gene, the csi gene, and the conjugative transfer genes were highly conserved (BSR ≥ 0.7) between the two plasmids (Fig. S4). However, some genes, such as those involved in plasmid stability and the antibiotic resistance genes, differed between pB171_90 and pCAR10 (Fig. S4). Of the antibiotic resistance genes identified on pB171_90, pCAR10 also carries the tetracycline resistance gene tetA, the tetA repressor tetR, and the sulfonamide resistance gene sul1 (Table 1; Fig. S4). However, pCAR10 contains additional antibiotic resistance genes that are missing from pB171_90, including a predicted bla_TEM β-lactamase gene and a dihydrofolate reductase gene, dfrA15 (Table 1; Fig. S4). The pCAR10 plasmid also contains a small transporter designated emrE, which confers resistance to ethidium bromide, and has 100% nucleotide identity to the small multidrug resistance protein-coding gene qacEΔ1 (pB171_90_89) of pB171_90 (Table 1; Fig. S4). The similarity of pB171_90 genes to genes identified in other species indicates that EPEC resistance plasmids may be transferred between EPEC and other human pathogens, including S. dysenteriae type I.

DISCUSSION

An increased prevalence of antibiotic resistance among certain groups of disease-causing E. coli, and the identification of E. coli as a reservoir of antibiotic resistance for routinely prescribed antibiotics (1, 9, 10, 55), highlights the importance of investigating the diversity of plasmids involved in disseminating resistance genes among E. coli. The pB171_90 plasmid belongs to the IncFII plasmid family, which counts among its members many of the virulence plasmids of diverse pathotypes of E. coli, including the virulence plasmids of EPEC (26). IncFII plasmids have also been described that carry antibiotic resistance genes among E. coli and other members of the family Enterobacteriaceae (26, 52) and have been reported to carry extended-spectrum β-lactamases (ESBLs), such as bla_CTX-M-15 (8, 56, 57). To provide insight into the diversity of resistance plasmids among EPEC strains, we provide the complete sequence and description of the antibiotic and heavy metal resistance plasmid, pB171_90, from the E. coli reference strain B171.

By investigating the presence of the pB171_90 genes among a large collection of E. coli genomes, we determined that genes such as csi and traI that were previously identified as unique to the B171 resistance plasmid (36) can be found in genomically diverse E. coli belonging to phylogroups A, B1, B2, D, and E (Fig. 2) as well as other enteric pathogens. Although the majority of the genomes that exhibited similarity to the genes of pB171_90 were identified as EPEC or ETEC, the pB171_90 genes were also identified in genomes of other diverse pathotypes (EAEC, STEC, ExPEC/uropathogenic E. coli [UPEC]) as well as E. coli genomes that didn't contain any of the pathotype-specific genes and may not be associated with illness (see Table S1 in the supplemental material). Interestingly, pB171_90 also contained a copy of the transcriptional regulator hha, which has been previously identified on the chromosome and plasmid of E. coli O157:H7 (46, 50). The hha gene was initially described as a regulator of hemolysin production but has since been described as a regulator of the LEE pathogenicity region and biofilm formation (47–49). Further studies are necessary to determine the role of hha in regulating virulence genes of EPEC strain B171. The pB171_90 plasmid also had significant similarity with genes of two O157:H7 E. coli genomes as well as the pCAR10 plasmid from a S. dysenteriae type I strain (54). This finding suggests that an exchange of resistance plasmids has occurred between EPEC and the phylogenomically related EHEC O157:H7 and S. dysenteriae, which cooccur together in E. coli phylogroup E.

Another notable feature of the pB171_90 resistance plasmid is the presence of a Tn21 transposon containing a spectinomycin-streptomycin resistance gene, a sulfonamide resistance gene, a small efflux pump, and a class 1 integrase gene, intI1. In a study investigating the prevalence of class 1 integrons among E. coli and Klebsiella isolated from hospitalized patients, the class 1 integrons were detected in 49% of the E. coli strains and were significantly associated with resistance to several classes of antibiotics (58). Comparison of pB171_90 with the sequenced EPEC and EHEC resistance plasmids pO111-CRL₁₁₅ (GenBank accession number KC340959) and pO26-CRL₁₂₅ (GenBank accession number KC340960) demonstrated that the tetracycline resistance and mercury resistance regions are conserved between these plasmids. Although plasmids pO111-CRL₁₁₅ and pO26-CRL₁₂₅ also contain a class 1 integron (59), the antibiotic resistance genes in this cassette differed from those of pB171_90, emphasizing the modular structure of E. coli resistance plasmids.

In summary, by characterizing the resistance plasmid of the EPEC reference strain B171, we provide further insight into the sequence diversity and host range of antibiotic and heavy metal resistance plasmids of E. coli. There were only a few publicly available E. coli genomes that contained nearly all of the resistance plasmid genes; however, genes from certain plasmid regions, including the antibiotic and heavy metal resistance genes, were present in diverse E. coli and other members of the family Enterobacteriaceae. Our study highlights the need for additional complete sequencing of resistance plasmids from EPEC and other pathotypes of E. coli to provide further insight into the dissemination and evolution of the E. coli resistance plasmids.

MATERIALS AND METHODS

Plasmid sequencing.

Plasmid DNA of EPEC strain B171 (37) was isolated from 1 liter of culture grown in L broth using alkaline lysis extraction method followed by ethidium bromide-CsCl density gradient centrifugation as previously described (60). The plasmid band was isolated and rebanded (190,000 × g) overnight on a second identical gradient. Plasmid was siphoned using a 3-ml syringe and an 18-gauge needle. Ethidium bromide was removed using isopropanol-CsCl-saturated 10 mM Tris, 5 mM EDTA, and 10 mM NaCl (pH 8) and then dialyzed against three changes of 10 mM Tris and 1 mM EDTA (pH 8). Plasmid DNA was ethanol precipitated using 0.125 M NaCl (final) and dissolved in water prior to sequencing.

The purified plasmid DNA was sequenced using the Pacific Biosciences RS II platform with the P6C4 chemistry in a single flow cell. Pacific Biosciences sequencing was completed by generating a sequencing library from 5 to 20 kb using standard methods at the Institute for Genome Sciences Genomics Resource Center (http://www.igs.umaryland.edu/resources/grc/). A total of 115,336 reads were generated with an average read length of 7,036 bp and a maximum read length of 38,154 bp. PacBio raw data were corrected and assembled using HGAP (SMRTAnalysis 2.3.0) (61), Canu assembler v.1.2 (62), and Celera assembler v.8.2 (63) run with default parameters. The assemblies were assessed for inconsistencies and misassembly using Nucmer whole-genome alignments (64) and Circleator plots (GC skew) (65). A BLAST search against NCBI's complete nucleotide database was used to confirm that all of the contigs were plasmids (65). The Canu assembler generated the most complete assembly for each of the plasmids in this sample. Circular contigs were identified and Minimus2 was used for circularization (66). The final circularized assembly was polished using Quiver (SMRTAnalysis 2.3.0) (61).

In silico multilocus sequence typing and serotyping.

The seven loci (adk, gyrB, fumC, icd, mdh, purA, and recA) of the multilocus sequence typing (MLST) scheme developed by Wirth et al. (67) were identified in each of the genomes listed in Table S1 in the supplemental material. The allele sequences were used to query BIGSdb (68) to obtain the allele numbers and sequence type of each of the E. coli genomes analyzed.

The molecular serotype of each genome was determined using Serotype Finder v.1.1 (https://cge.cbs.dtu.dk/services/SerotypeFinder/) with the default settings of an 85% identity threshold and 60% minimum alignment length (69).

Phylogenomic analysis.

The 134 E. coli genomes that contained genes with similarity to csi and/or traI of the B171 resistance plasmid were compared with 55 previously sequenced E. coli and Shigella genomes using the SNP-based in silico genotyper (ISG) (70) as previously described (35) (Table S1). There were 149,150 conserved SNP sites identified among all of the E. coli genomes analyzed relative to the E. coli reference strain IAI39 (GenBank accession number NC_011750.1) that were used to generate a maximum-likelihood phylogeny with 100 bootstrap values using RAxML v.7.2.8 (71). The phylogeny was constructed using the general time reversible (GTR) model of nucleotide substitution with the GAMMA model of rate heterogeneity and 100 bootstrap replicates. The phylogeny was then midpoint rooted and labeled using the interactive tree of life software (iTOL v.3) (72).

In silico detection of pB171_90 genes.

The genes of pB171_90 were predicted and annotated using an in-house annotation pipeline (73). These genes were then detected in a collection of 4,798 E. coli genome assemblies available in GenBank as of November 2016, using large-scale BLAST score ratio (LS-BSR) analysis as previously described (74, 75). The pB171_90 protein-coding genes were compared to each genome listed in Table S1 using TBLASTN (76) with composition-based adjustment turned off. The TBLASTN scores were used to generate a BSR indicating the detection of each gene cluster in each of the genomes (Table S1). The BSR was determined by dividing the score of a gene compared to a genome by the score of the gene compared to its own sequence. The genes that were identified with a BSR of ≥0.7 were considered present in the genomes analyzed. The numbers of genes that were present in AEEC compared to non-AEEC genomes was examined for statistical significance using the chi-squared test with Yates' continuity correction performed using R v.3.3.2. The clustered heatmap was generated using the heatmap2 function of gplots v.3.0.1 in R v.3.3.2. The BSR values of the genomes were clustered in the heatmap using the complete linkage method with Euclidean distance estimation.

Pathotype-specific virulence genes were identified in each of the genomes by in silico analysis using LS-BSR as described above. The following canonical virulence genes, which have been previously used for the molecular-based classification of clinical E. coli strains to each of the pathotypes, were detected in each of the following E. coli genomes analyzed: tEPEC (LEE genes eae and escV and BFP genes bfpA and bfpB), aEPEC (LEE genes eae and escV only; no BFP or Shiga toxin genes), ETEC (eltA and eltB encoding the heat-labile toxin LT and/or est encoding the heat-stable toxin ST), EAEC (the transcriptional regulator aggR, aatA encoding an outermembrane protein, or pic and/or pet encoding autotransporters), STEC (Shiga toxin genes stxA and stxB), EIEC (ipaB, mxi-spa, and/or ospB, which are all involved in type III secretion), and ExPEC/UPEC (papG encoding a fimbrial adhesion precursor) (19–21, 77, 78).

Gene phylogenies.

Individual gene phylogenies were constructed for the repA, hha, csi, and traI sequences that were detected by in silico analysis of each of the E. coli genomes analyzed (Table S1). A csi phylogeny was also constructed with the csi sequences from E. coli and other species of human enteric pathogens listed in Table S1. The nucleotide sequences of each gene were aligned using ClustalW, and maximum-likelihood phylogenies were constructed using the Kimura 2-parameter model and 1,000 bootstraps with MEGA7 (79).

Accession number(s).

The assemblies of pB171_90 and pB171_68 are deposited in GenBank under the accession numbers CP021212 and CP021211, respectively.