Reexamining the Association of AmpC Variants with Enterobacter Species in the Context of Updated Taxonomy

ABSTRACT

We performed whole-genome sequencing for 17 Enterobacter clinical strains and analyzed all available Enterobacter genomes and those of its closely related genera (n = 3,389) from NCBI. The exact origins of plasmid-borne bla_CMH and bla_MIR genes are Enterobacter cloacae and Enterobacter roggenkampii, respectively, while plasmid-borne bla_ACT genes originated from multiple other Enterobacter species, including Enterobacter xiangfangensis, Enterobacter hoffmannii, Enterobacter asburiae, Enterobacter ludwigii, and Enterobacter kobei. The genus Enterobacter represents a large reservoir of plasmid-borne AmpC β-lactamase.

INTRODUCTION

Enterobacter, a genus of the order Enterobacterales, has 45 species, 24 species with known names and 21 taxa (genomospecies) without assigned names (see Table S1 in the supplemental material) (1, 2). Strains of all Enterobacter species harbor an inducible chromosomal ampC gene encoding AmpC β-lactamase to confer resistance to various cephalosporins and penicillins according to the expression level (3). A standard numbering scheme of AmpC β-lactamases has been developed recently (4). However, the diversity of ampC genes in Enterobacter species and the association between AmpC variants and Enterobacter species are yet to be examined. In addition, three types of plasmid-encoded ampC genes, bla_ACT (5), bla_CMH (6), and bla_MIR (7), seen in many other species of the Enterobacterales have been found to derive from the chromosomal Enterobacter ampC genes, but their exact origins remain to be determined.

Genomes analyzed.

A total of 3,403 Enterobacter genomes were analyzed. To contribute to this analysis, we included 17 nonduplicate Enterobacter strains recovered from clinical specimens as part of routine care from 2018 to 2020 at West China Hospital for whole-genome sequencing (WGS, Table 1). WGS was performed using the HiSeq X10 platform (Illumina, San Diego, CA) with about 300× coverage, and reads were assembled using SPAdes v3.14.0 (8). Precise species identification was performed using pairwise average nucleotide identity (ANI) with FastANI (9) v1.33 against genomes of type or reference strains of the genus Enterobacter as described previously (2) with a ≥96% cutoff to define species (10). The 17 strains belong to six species—Enterobacter asburiae (n = 1), Enterobacter dykesii (n = 1), Enterobacter hoffmannii (n = 5), Enterobacter quasimori (n = 1), Enterobacter cloacae (n = 1), and Enterobacter xiangfangensis (n = 8) (Table 1).

TABLE 1.

The 17 Enterobacter clinical strains in this study

Strain	Species	Collection yr	Sample	GenBank nucleotide accession no.	AmpC variant or GenBank protein accession no.a
120126	E. asburiae	2019	Ascites	JAHEVT000000000	WP_048980831
140050	E. cloacae	2019	Pus	JAHERU000000000	WP_058683930
120147	E. dykesii	2019	Bile	JAHEVX000000000	MBT1716307
090086	E. hoffmannii	2018	Pus	JAHEVK000000000	WP_038415120
120143	E. hoffmannii	2019	Sputum	JAHEVW000000000	ACT-24
120150	E. hoffmannii	2020	Bile	JAHEVY000000000	ACT-24
120115	E. hoffmannii	2020	Ascites	JAHEVS000000000	ACT-24
120133	E. hoffmannii	2020	Ascites	JAHEVV000000000	WP_038415120
120130	E. quasimori	2019	Pleural fluid	JAHEVU000000000	MBT1729853
090088	E. xiangfangensis	2018	Drainage	JAHEVL000000000	ACT-25
090090	E. xiangfangensis	2018	Ascites	JAHEVN000000000	ACT-69
090098	E. xiangfangensis	2018	Pus	JAHEVP000000000	ACT-17
090104	E. xiangfangensis	2018	Secretion	JAHEVR000000000	ACT-17
090095	E. xiangfangensis	2018	Sputum	JAHEVO000000000	ACT-74
090089	E. xiangfangensis	2018	Sputum	JAHEVM000000000	ACT-16
090102	E. xiangfangensis	2018	Sputum	JAHEVQ000000000	ACT-69
140033	E. xiangfangensis	2019	Sputum	JAHERT000000000	ACT-74

^aFor unnamed AmpC variants, the protein accession no. is shown.

Since genomes of Enterobacter were frequently misidentified as those of Lelliottia or Leclercia, all assemblies (n = 3,389) under these three genera were retrieved from NCBI (accessed on 19 April 2019). Genomes (n = 219) with assembly abnormalities confirmed by GenBank, such as contaminated, fragmented, frameshifted, and odd genome size, and those assembled from metagenomic data were discarded. All remaining genomes (n = 3,170) were identified for species using FastANI v1.33 (9). Genomes (n = 116) that did not belong to Enterobacter species were excluded. The 3,054 remaining genomes were assigned to 45 known Enterobacter species and a novel tentative unnamed species, designated Enterobacter taxon 23, which is most closely related to E. asburiae with a 92.70% ANI value (Table S1). However, 28 Enterobacter genomes had no intact ampC genes as identified using AMRFinderPlus v3.10 (11) and were discarded, leaving 3,026 curated Enterobacter genomes from GenBank. Along with 17 genomes from this study, a total of 3,043 curated genomes (296 complete and 2,747 draft genomes; Data set S1) were included for the following analysis, with the flowchart of genome inclusion and exclusion shown in Fig. S1.

AmpC variants.

A total of 508 AmpC variants were identified in Enterobacter. Along with the 10,000-bp flanking regions (5,000-bp upstream and 5,000-bp downstream), the chromosomal ampC gene in Enterobacter cloacae DSM 30054 (GenBank accession no. CP056776) was used as the reference to obtain the similar sequence of all other genomes by BLASTn search. Matches with the highest bit scores of each strain were analyzed using AMRFinderPlus v3.10 (11). Among the 3,043 genomes analyzed, ampC genes encode 508 AmpC variants, with amino acid identity ranging from 79.79% to 99.74% (Data set S2). The chromosomal locations of ampC in all genomes were verified by the presence of housekeeping or core genes using CheckM v1.0.18 (12).

At present, 86 ACT (ACT-1 to -89 but no ACT-11, -26, or -71), 6 CMH (CMH-1 to -6), and 22 MIR (MIR-1 to -23, but MIR-8 is equal to MIR-5) variants have been assigned in the Pathogen Detection Reference Gene Catalog (https://github.com/ncbi/amr/wiki/AMRFinderPlus-database; accessed 19 April 2021). All identified Enterobacter AmpCs along with ACTs, CMHs, and MIRs were aligned using MAFFT v7.480 (13) to infer a phylogenetic tree (Fig. S2) using IQ-TREE v2.1.2 (14) and iTOL v6.1.2 (15). A simplified version of the phylogenetic tree (Fig. 1) containing AmpC from the type strain of named Enterobacter species (n = 24) or the reference strain of unnamed taxa (n = 22) along with all ACT, CMH, and MIR variants was also inferred as described above. The 508 AmpC variants include 62 ACT variants seen in 1,837 genomes, 3 CMH variants in 27 genomes, and 17 MIR variants in 102 genomes (Table 2). In contrast, 24 ACT (ACT-5, -7, -8, -10, -13, -18, -20, -21, -22, -29, -30, -31, -32, -33, -34, -36, -38, -73, -78, -79, -80, -81, -82, and -83), 3 CMH (CMH-1, -2, and -6), and 5 MIR (MIR-1, -2, -4, -6, and -18) variants have been assigned but were not identified in these Enterobacter genomes (Fig. 1 and Fig. S2).

FIG 1.

A phylogenetic tree of all ACT, CMH, and MIR variants and AmpC variants of the type or reference strain of each Enterobacter species. This tree was inferred from the multialignment of amino acid sequences using IQ-TREE v2.1.2 (14) and was visualized and annotated using iTOL v6.1.2 (15). A phylogenetic tree of all Enterobacter AmpC variants and all ACT, CMH, and MIR variants is shown in Fig. S2. Amino acid sequence accession numbers of AmpC variants are indicated, and those seen in Enterobacter are also shown with the species names. Taxon refers to tentative unnamed species based on genome analysis, as the proposal of appropriate species names also needs phenotypic characterizations under current International Code of Nomenclature of Prokaryotes (17). AmpC variants from the type strain of named Enterobacter species or the reference strain from unnamed taxa are shown in red, while those that have not been identified in Enterobacter genomes are depicted in blue.

TABLE 2.

Numbers of Enterobacter genomes and AmpC variants

Enterobacter sp.	No. of genomes	No. of AmpC variants	Assigned ACT, CMH, and MIR variant(s)
E. asburiae	165	46	ACT-1, -2, -3, -4, -6, -57, -68
E. bugandensis	108	33	ACT-49, -76, -77
E. cancerogenus	14	6
E. chengduensis	7	1	ACT-53
E. chuandaensis	2	2
E. cloacae	144	44	CMH-3, -4
E. dissolvens	21	14	CMH-5
E. dykesii	7	4
E. hoffmannii	489	39	ACT-14, -23, -24, -39, -47, -66, -67
E. hormaechei	28	8	ACT-19, -37
E. huaxiensis	2	2
E. kobei	129	31	ACT-9, -28, -51, -52, -64, -87
E. ludwigii	79	32	ACT-12, -54, -88
E. mori	12	12
E. oligotrophica	2	1
E. quasihormaechei	40	11	ACT-59
E. quasimori	2	2
E. quasiroggenkampii	15	12	ACT-62
E. roggenkampii	163	48	MIR-3, -5, -7, -9, -10, -11, -12, -13, -15, -16, -17, -19, -20, -21, -22, -23
E. sichuanensis	18	11	ACT-50
E. soli	4	2
E. vonholyi	8	7	ACT-58, MIR-14
E. wuhouensis	1	1
E. xiangfangensis	1,510	92	ACT-15, -16, -17, -25, -27, -35, -40, -41, -42, -43, -44, -45, -46, -55, -56, -60, -61, -65, -69, -70, -72, -74, -75, -84, -85, -86, -89
Taxon 1	2	2
Taxon 2	11	8
Taxon 3	10	7
Taxon 4	16	2	ACT-48, -63
Taxon 5	4	4
Taxon 6	1	1
Taxon 7	1	1
Taxon 9	2	2
Taxon 10	4	4
Taxon 11	1	1
Taxon 12	1	1
Taxon 13	2	1
Taxon 14	6	4
Taxon 17	1	1
Taxon 19	5	3
Taxon 20	3	2
Taxon 21	2	2
Taxon 23	1	1
Total	3,043	508	82

AmpC is not an ideal marker for precise species identification.

The distribution of AmpC variants across the genus Enterobacter is also almost always corelated with the species, but with several notable exceptions (Fig. 1 and Fig. S2). ACT-58 and an unnamed AmpC (GenBank accession no. WP_133293618) of Enterobacter vonholyi are well separated, with AmpC variants in other E. vonholyi strains, but are clustered with those of E. asburiae and Enterobacter quasiroggenkampii, respectively. The AmpC variants of E. cloacae are generally well separated from those of Enterobacter dissolvens, previously known as a subspecies of E. cloacae (1, 2), but an unnamed AmpC (GenBank accession no. WP_095455155) of E. cloacae is clustered with those of E. dissolvens.

The minimum amino acid sequence identity of AmpC variants between Enterobacter species and the maximum amino acid identity within the same species overlap (Table S2 for a summary and Data set S3 for detailed results). No clear, unified cutoff of amino acid sequence identity could be established to correctly correlate AmpC variants with Enterobacter species. In addition, bla_ACT-1, bla_CMH-1, and bla_MIR-6 genes have been found on plasmids of Enterobacter strains (Table S3). The coexistence of plasmid-borne and chromosomal ampC genes with significant nucleotide identities also causes problems for ampC-based species identification.

Plasmid-borne bla_ACT, bla_CMH, and bla_MIR genes originated from different Enterobacter species.

All six assigned CMH enzymes have 95.01% to 99.48% pairwise amino acid identities and are seen in or clustered within E. cloacae and E. dissolvens (Fig. 1 and Fig. S2). bla_CMH-1 and bla_CMH-2 are the only two bla_CMH genes found on plasmids at present (Table S3). Both CMH-1 and CMH-2 have the highest amino acid identity with the chromosomal AmpC variant of E. cloacae (99.74% with GenBank accession no. WP_058678916 for CMH-1 and 99.21% with GenBank accession no. WP_058683208 for CMH-2), suggesting that E. cloacae is the origin of bla_CMH-1 and bla_CMH-2.

Among the 22 MIR variants assigned, 5 plasmid-encoded (MIR-1, -3, -4, -6, and -23) variants have been found (Table S3). MIR-1 has been proposed to originate from E. cloacae (7). However, all MIR enzymes except MIR-14 are seen in or clustered within E. roggenkampii (Fig. 1 and Fig. S2) sharing the highest amino acid identity of 99.74% (MIR-1, -4, and -6) or 100% (MIR-5 and -23) with the chromosomally encoded AmpC of this species (Table S3). MIR-14 is found in E. vonholyi and has only a 92.7% to 94.5% amino acid identity with other MIR variants but is closer to ACT-29 with a 95.3% amino acid identity. In contrast, all other MIR variants are clustered together and have >97.9% pairwise amino acid identities. This suggests that MIR-14 is not a MIR variant, which may be due to incorrect assignment. The above analysis also indicates that E. roggenkampii instead of E. cloacae is the origin of plasmid-borne bla_MIR genes.

The plasmid-mediated bla_ACT-1 gene was first identified in Klebsiella pneumoniae and is believed to originate from the chromosome of E. cloacae (5), but later it was curated to originate from E. asburiae (3) and then to be from Enterobacter hormaechei (16). Among genes encoding the 86 assigned ACT variants, those encoding 12 (ACT-1, -2, -3, -5, -6, -7, -9, -10, -12, -15, -20, and -36) have been found on plasmids or in species other than Enterobacter (Table S3). All ACT variants are seen in or clustered with Enterobacter species other than E. cloacae, E. dissolvens, and E. roggenkampii. Among the 12 plasmid-encoded ACT variants, there are 5 types of clustering. ACT-1, -2, -3, -6, and -10 are seen in or clustered within E. asburiae (Fig. 1 and Fig. S2) and have the highest amino acid identity, 98.2% (ACT-10) or 100% (ACT-1, 2, -3, and -6) with the chromosomally encoded AmpC of this species (Table S3). ACT-5 and ACT-36 are clustered within E. hoffmannii with the highest amino acid identity, 99.7%, while ACT-7, -15, and -20 are clustered within E. xiangfangensis and have the highest amino acid identity, 99.7% (ACT-7 and -20) or 100% (ACT-15) with the chromosomally encoded AmpC of this species. ACT-12 and ACT-9 are seen in Enterobacter ludwigii and E. kobei, sharing the highest amino acid identity, both of 100% (Table S3). It therefore becomes evident that plasmid-borne bla_ACT genes have different origins from different strains belonging to one of the five species, E. asburiae, E. kobei, E. hoffmannii, E. ludwigii, and E. xiangfangensis.

In conclusion, Enterobacter species are a rich reservoir of ampC genes, which could become plasmid-borne and therefore be transmitted to species of other genera. Plasmid-borne bla_CMH and bla_MIR genes originated from E. cloacae and E. roggenkampii, respectively, while plasmid-borne bla_ACT genes originated from E. asburiae, E. kobei, E. hoffmannii, E. ludwigii, and E. xiangfangensis.

Data availability.

The draft whole-genome sequences of the 17 strains have been deposited into DDBJ/EMBL/GenBank under accession no. JAHERT000000000, JAHERU000000000, and JAHEVL000000000 to JAHEVY000000000 (Table 1).