Precise Species Identification for Enterobacter: a Genome Sequence-Based Study with Reporting of Two Novel Species, Enterobacter quasiroggenkampii sp. nov. and Enterobacter quasimori sp. nov.

Enterobacter species are major human pathogens. Precise species identification lays a foundation for microbiology, but the taxonomy of Enterobacter is complicated and confusing. In this study, first, we significantly updated the taxonomy of Enterobacter by rigorous genome analyses and found that all subspecies assignments of Enterobacter were incorrect. Second, we characterized and reported two novel Enterobacter species with clinical significance. Third, we curated 1,997 Enterobacter genome sequences deposited in GenBank and found that the species identification of most Enterobacter strains needed to be corrected. Fourth, we found that the most common Enterobacter species seen in clinical samples is Enterobacter xiangfangensis rather than Enterobacter cloacae. Fifth, we identified 14 tentative novel Enterobacter and 18 tentative novel non-Enterobacter species. This study highlights that updated and curated taxonomic assignments are the premise of correct species identification. We recommend that future Enterobacter studies need to use the updated taxonomy to avoid misleading information.

all hitherto-known species, by both genome-and phenotype-based methods. We then used the updated taxonomy of Enterobacter to review and curate the species assignment of all Enterobacter genomes (n ϭ 1,997) in GenBank to correct the corresponding misleading information. We found that the majority of Enterobacter strains with wholegenome sequences available are not E. cloacae but E. xiangfangensis. We also found that there are 14 tentative novel Enterobacter species based on genome analysis, which need to be further studied using phenotype-based methods to establish their species status.

E. cloacae subsp. dissolvens is a species rather than a subspecies and should be renamed Enterobacter dissolvens.
Whole-genome sequencing for strain ATCC 23373 T generated 2,908,248 reads and 0.87 gigabases, which were assembled into a 4.84-Mb draft genome containing 51 contigs Ն200 bp in length (N 50 , 415,836 bp) with a 55.16% GC content. No contamination was identified in the genomes. The gyrB, rpoB, infB, and atpD sequences were identical to those of strain ATCC 23373 T previously deposited in GenBank (accession no. JX424979, JX425238, JX425108, and JX424849, respectively), suggesting that this strain was indeed strain ATCC 23373 T . The ANI value between strain ATCC 23373 T and E. cloacae subsp. cloacae ATCC 13047 T (GenBank accession no. CP001918) was 94.79% (ATCC 13047 T versus ATCC 23373 T ) or 94.92% (vice versa), below the 96% ANI cutoff to define a bacterial species (9). The isDDH value between the type strains was 62.0%, lower than the 70.0% cutoff to define a bacterial species (10). Both ANI and isDDH analyses indicate that E. cloacae subsp. dissolvens should be considered a species different from E. cloacae subsp. cloacae. In addition, the ANI and isDDH values between strain ATCC 23373 T and type strains of all other Enterobacter species are Ͻ95% and Ͻ70%, respectively (see Table S1 in the supplemental material). We therefore proposed that E. cloacae subsp. dissolvens should be elevated to the species level as Enterobacter dissolvens sp. nov. (type strain ATCC 23373 T ϭ CIP 105586 T ϭ JCM 6049 T ϭ LMG 2683 T ).
Enterobacter xiangfangensis is not a subspecies of E. hormaechei. The core gene-based phylogenomic tree (see Fig. S1 in the supplemental material) demonstrated that the type strains of E. xiangfangensis and other E. hormaechei subspecies formed a clade, which was distinct from all other Enterobacter species. This suggests that E. xiangfangensis and the E. hormaechei subspecies are indeed closely related. Within this E. hormaechei clade, E. hormaechei subsp. oharae, E. hormaechei subsp. steigerwaltii, and E. xiangfangensis were clustered together, while the other two subspecies each appeared to form a distinct branch. The ANI values between the strain E. hormaechei subsp. hormaechei ATCC 49162 T , which is also the type strain of the species E. hormaechei, and the type strains of other subspecies and E. xiangfangensis range from 94.13% to 94.79% (Table 2), which are below the 96% ANI cutoff to define a bacterial species (9). The isDDH value between E. hormaechei subsp. hormaechei ATCC 49162 T and the type strains of other subspecies and E. xiangfangensis ranges from 58.0% to 62.5% (Table 2), also lower than the 70% cutoff to define a bacterial species. Both ANI  (Table 2). Both the ANI and isDDH values among the three strains were well above the cutoffs to define a bacterial species, indicating that the three type strains belong to a common species. The fact that ANI and isDDH values among E. xiangfangensis, E. hormaechei subsp. oharae, and E. hormaechei subsp. steigerwaltii are above the cutoff to define bacterial species has also been noticed before (18) and is used as the evidence that E. xiangfangensis is a subspecies of E. hormaechei (18). As demonstrated above, E. hormaechei subsp. oharae and E. hormaechei subsp. steigerwaltii do not belong to E. hormaechei in fact. Therefore, the Ͼ96% ANI and Ͼ70% isDDH values between E. xiangfangensis and E. hormaechei subsp. oharae or E. hormaechei subsp. steigerwaltii cannot be used as the evidence to reject the species status of E. xiangfangensis but provide the proof that the three "subspecies" actually belong to a common species.
Enterobacter hormaechei subsp. hoffmannii is not a subspecies of Enterobacter hormaechei but is a novel species. The ANI values between the type strain of E. hormaechei subsp. hoffmannii (DSM 14563 T ) and type strains of E. hormaechei subsp. oharae, E. hormaechei subsp. steigerwaltii, and E. xiangfangensis range from 95.59% to 95.71% (Table 2), which fall into the 95 to 96% inconclusive zone of defining a bacterial species (9,21). The isDDH value between E. hormaechei subsp. hoffmannii strain DSM 14563 T and the type strains of E. hormaechei subsp. oharae, E. hormaechei subsp. steigerwaltii, and E. xiangfangensis ranges from 66.5% to 66.9% (Table 2), lower than the 70% cutoff to define a bacterial species. Therefore, E. hormaechei subsp. hoffmannii is a novel Enterobacter species rather than a subspecies of any known Enterobacter species, and we propose the species name as Enterobacter hoffmannii.
Enterobacter timonensis should be removed to a novel genus with the proposed name Pseudenterobacter. The core gene-based phylogenomic tree of the family Enterobacteriaceae ( Fig. 1) and that of the genus Enterobacter and closely related genera (Fig. 2) demonstrated that E. timonensis forms an independent branch, which is well separated from all other Enterobacter species by species of the genera Leclercia and Lelliottia. The ANI values between the type strain of E. timonensis and those of all other Enterobacter species are Ͻ85% (82.03 to 83.78%, Table S1), while the values between type strains of other Enterobacter species are Ͼ85%. Correspondingly, the isDDH values between the type strain of E. timonensis and those of all other Enterobacter species are Ͻ30% (24.7 to 26.3%, Table S1), while the values between type strains of other Enterobacter species are Ͼ30%. The above findings suggest that E. timonensis does not belong to the genus Enterobacter. The ANI and isDDH values for the type strain of E. timonensis and those of Leclercia and Lelliottia species are Ͻ85% and Ͻ30%, respectively. The phylogenomic trees ( Fig. 1 and 2) demonstrated that E. timonensis is also distinct from Leclercia and Lelliottia species. Therefore, it is evident that E. timonensis does not belong to the genus Leclercia nor Lelliottia but to a novel genus. As it is closely related to Enterobacter, we propose the genus name Pseudenterobacter (Pseud.en.te.ro-.bac=ter. Gr. adj. pseudês false; N.L. masc. n. Enterobacter a bacterial generic name; N.L. fem. n. Pseudenterobacter, a genus falsely [or incorrectly] classified in Enterobacter). E. timonensis should therefore be renamed Pseudenterobacter timonensis.
Two Enterobacter strains from blood represent a novel species, named Enterobacter quasiroggenkampii sp. nov. Strains WCHECL1060 T and 090040 were both identified as E. cloacae by Vitek II. The two strains had very different genomic fingerprints obtained by macrorestriction analysis (see Fig. S2 in the supplemental material). The 16S rRNA gene sequence of the two strains shared 99.61% identity (6 bases mismatch) and was 99% identical to those of type strains of a few Enterobacter species including E. asburiae, E. cloacae, E. hormaechei, E. kobei, and E. ludwigii.
The draft whole-genome sequence of strain WCHECL1060 T has been reported by us before (22), and its 4.8-Mb draft genome was assembled from 1.7 gigabases into 21 contigs Ն200 bp in length (N 50 , 714,400 bp) with a 55.68% GC content. For strain 090040, 4,788,302 reads and 1.73 gigabases were generated, which were assembled into a 4.9-Mb draft genome containing 30 contigs Ն200 bp in length (N 50 , 515,146 bp) with a 55.69% GC content. No contamination was identified for the genomes of WCHECL1060 T and 090040. The ANI value between strains WCHECL1060 T and 090040 was 98.4% (Table 3). In contrast, the ANI values between the two strains and type strains of all known Enterobacter species were Ͻ96% and the highest value (95.37%/ 95.30%, respectively) was seen with E. roggenkampii DSM 16690 T (Table 3 and Table S1). The isDDH value between strains WCHECL1060 T and 090040 was 88% (Table 3), non-Enterobacter tentative taxons A to T, Enterobacter tentative taxons 1 to 14, and type strains of the family Enterobacteriaceae (listed in Data Set S1). Strains and their nucleotide accession numbers are listed alongside the species names. For species and subspecies with names that need to be revised as suggested in this study, the revised names are shown first, and the current names are shown after the slash. The tree was inferred using the maximum likelihood method under the GTRGAMMA model with a 1,000-bootstrap test, and branches with support over 50% are indicated by gradients. Bar, value indicates the nucleotide substitutions per site.  Tables 5 and 7 and Data Set S1). Strains and their nucleotide accession numbers are listed alongside the names of species. For species and subspecies with names that need to be revised as suggested in this study, the revised names are shown first, and the current names are shown after the slash. The tree was inferred using the maximum likelihood method under the GTRGAMMA model with a 1,000-bootstrap test, and branches with support over 50% are indicated by gradients. Bar, value indicates the nucleotide substitutions per site.
whereas isDDH values between the two strains and type strains of all known Enterobacter species were 64.8%/65.2%, respectively (with E. roggenkampii DSM 16690 T ), or lower (Table 3 and Table S1), which were below the 70% cutoff to define a bacterial species. Therefore, the ANI and isDDH analyses clearly suggest that the two strains represent a novel species of the genus Enterobacter.
Biochemical characteristics between strains WCHECL1060 T and 090040 and type strains of other Enterobacter species are shown in Table 4. For both strains, growth occurs at 4 to 37°C with optimal growth at 35 and 37°C, but not at 45 or 50°C. Cells grow at 35°C in the presence of 0 to 9% (wt/vol) NaCl in tryptic soy broth (TSB). Both strains were positive for the catalase test but negative for oxidase activity. Cells of the two strains are Gram negative, motile, non-spore-forming, facultatively anaerobic, and rod shaped. Colonies are circular, white, translucent, raised, and smooth after 24 h of incubation at 35°C on nutrient agar. Acid is produced from glycerol, L-arabinose, D-ribose, D-xylose, D-galactose, D-glucose, sucrose, melibiose, amygdalin, D-fructose, D-mannose, L-rhamnose, inositol, D-mannitol, D-sorbitol, potassium 2-ketogluconate, and methyl-␣-D-glucopyranoside but not erythritol, L-xylose, D-adonitol, D-arabinose, potassium gluconate, and methyl-␣-D-mannopyranoside. Both strains have a positive reaction for ␤-galactosidase, arginine dihydrolase, and ornithine decarboxylase but are negative for lysine decarboxylase, deaminase, and gelatinase. Both are also negative for urease activity and indole production but positive for the Voges-Proskauer reaction. Both strains can utilize citrate but do not produce H 2 S. The two strains can be differentiated from all other Enterobacter species by their ability to ferment inositol, D-sorbitol, and melibiose but not potassium gluconate, L-fucose, and methyl-␣-Dmannopyranoside.
The comparison of the fatty acid profiles of the strains WCHECL1060 T and 090040 and type strains of other Enterobacter species are shown in Table S2 in the supplemental material. Although the proportions of the fatty acids were slightly different, the major cellular fatty acids of strains WCHECL1060 T and 090040 were C 16:0 , C 17:0 cyclo, and C 18:17c , which were consistent with those of other Enterobacter species. The antimicrobial susceptibility profiles and antimicrobial resistance genes of the two strains are described in the supplemental material (Text S1 and Table S3). The results presented here indicate that two strains represent a novel species within the genus Enterobacter, which is clearly distinct from all known Enterobacter species. As it is most closely related to E. roggenkampii in whole-genome analysis, we propose the name Enterobacter quasiroggenkampii sp. nov. (qua.si.rog.gen.kamp.i; L. adv. quasi nearly, almost; N.L. gen. n. roggenkampii of Roggenkamp, and a specific epithet in the genus Enterobacter; N.L. gen. n. quasiroggenkampii almost roggenkampii) for this species with WCHECL1060 T (ϭ GDMCC 1.1742 T ϭ KCTC 52992 T ) as the type strain.
An Enterobacter strain from blood represents another novel species, named Enterobacter quasimori sp. nov. Strain 090044 T was identified as E. cloacae by Vitek II. The 16S rRNA gene sequence of the strain was 99% identical to those of type strains of a few Enterobacter species including E. asburiae, E. bugandensis, E. hormaechei, E. kobei, and E. ludwigii. Whole-genome sequencing for strain 090044 T generated 4,498,239 reads and 1.35 gigabases, which were assembled into a 4.71-Mb draft genome containing 53 contigs Ն200 bp in length (N 50 , 291,547 bp) with a 55.76% GC content. No contamination was identified. The ANI values between strain 090044 T and type strains of all known Enterobacter species and WCHECL1060 T were Ͻ96%, and the highest value (95.32%) was seen with E. mori LMG 25706 T (Table 3 and Table S1). The isDDH values between strain 090044 T and type strains of all known Enterobacter species and WCHECL1060 T were Ͻ70%, and the highest value (66.8%) was seen with E. mori LMG 25706 T (Table 3 and Table S1). Therefore, based on the ANI and isDDH analyses, it is evident that the strain represents a novel species of the genus Enterobacter.
For strain 090044 T , growth occurs at 4 to 37°C with optimal growth at 35 and 37°C, but not at 45 or 50°C. Cells grow at 35°C in the presence of 0 to 9% (wt/vol) NaCl in TSB. They are positive for the catalase test but negative for oxidase activity. Cells of strain  1  2  3  4  5  6  7  8  9  10  11 12 13 14 15 16  17  18  19  20   Strain 090044 T has a positive reaction for ␤-galactosidase, arginine dihydrolase, and ornithine decarboxylase but is negative for lysine decarboxylase, deaminase, and gelatinase. It is also negative for urease activity and indole production but positive for the Voges-Proskauer reaction. It can utilize citrate but does not produce H 2 S. It is catalase positive and oxidase negative. Strain 090044 T can be differentiated from other Enterobacter species and WCHECL1060 T by its ability to ferment inositol, D-sorbitol, dulcitol, D-turanose, and melibiose but not potassium gluconate, L-fucose, and methyl-␣-D-mannopyranoside. The major cellular fatty acids of strain 090044 T were C 16:0 , C 17:0 cyclo, and C 18:17c , which were consistent with those of other Enterobacter species (Table S2). The antimicrobial susceptibility profile and antimicrobial resistance genes of the strain are described in the supplemental material (Text S1 and Table S3).
The results presented here indicate that strain 090044 T represents a novel species within the genus Enterobacter. As it is most closely related to E. quasimori in wholegenome analysis, we propose the name Enterobacter quasimori sp. nov. (qua.si.mo.ri; L. adv. quasi nearly, almost; N.L. gen. n. mori of Zhu, and a specific epithet in the genus Enterobacter; N.L. gen. n. quasimori almost mori) for this species with 090044 T (ϭ GDMCC 1.1735 T ϭ JCM 33940 T ) as the type strain.
Most Enterobacter genomes in GenBank need to be curated for precise species identification. Based on the above findings, the taxonomy of Enterobacter should be updated to comprise 22 species at present (Table 5). There were 1,997 Enterobacter strains with genomes deposited in GenBank, and the species identification is required to be curated for most (n ϭ 1,542, 77.2%) of these strains in four scenarios. First, among After curation of precise species identification, among the 1,960 Enterobacter strains, half (n ϭ 994, 50.7%) actually belonged to E. xiangfangensis, while E. hoffmannii is the second most common species with 287 strains (14.7%; Table 6), followed by E. asburiae (n ϭ 116, 5.9%) and E. roggenkampii (n ϭ 112, 5.7%). However, there were 60 (3.1%) strains that could not be assigned to any known Enterobacter species. Instead, the 60 strains can be assigned to 14 potentially novel Enterobacter species, which are unnamed as they have not been characterized by phenotype methods. The 14 potentially novel Enterobacter species were assigned taxons 1 to 14 here (Table 7). There were 1,496 strains from human specimens. Among strains from human, E. xiangfangensis was still the most common species (805/1,496, 53.8%; Table 6) and E. hoffmannii was the second most common (251/1,496, 17.2%). Although the selection of bacterial strains is usually biased for genome sequencing, the common identification of the two Enterobacter species from human specimens is unlikely to be a coincidence. The reasons why isolates of the two Enterobacter species are commonly recovered from human specimens warrant further studies.

DISCUSSION
In this study, we first updated the taxonomy of the genus Enterobacter and modified the taxonomic assignments for E. timonensis and the subspecies of E. cloacae and E. hormaechei by genome analyses and also reported two novel species, which were characterized by both genome-and phenotype-based methods. We then applied the updated taxonomy assignments to curate genome sequences deposited in GenBank with the label of Enterobacter and found that the species identification of most Enterobacter strains with genome sequences available needed to be corrected.
We found that all subspecies assignments in the genus Enterobacter were incorrect and their use should be discontinued. Genetic clustering of the hsp60 (a housekeeping gene) sequence has been used as the premise for assigning E. hormaechei subsp. hormaechei, E. hormaechei subsp. oharae, and E. hormaechei subsp. steigerwaltii (16,17). However, determining taxonomic assignment using a single-gene-based approach has omitted valuable information available from the rest of the genome and potentially led to unreliable conclusions about taxonomic positions. Such subspecies assignment should be rigorously reexamined based on analysis of whole-genome sequences. Indeed, on the basis of whole-genome-based analysis, it becomes evident that the subspecies of E. hormaechei actually belong to three species. E. xiangfangensis is not a subspecies of E. hormaechei but an independent species, while E. hormaechei subsp. steigerwaltii and E. hormaechei subsp. oharae belong to the same species as E. xiang-Enterobacter Species Identification fangensis. E. hormaechei subsp. hoffmannii is a novel species, E. hoffmannii. Wholegenome-based analysis also reveals that E. cloacae subsp. dissolvens is actually a species, E. dissolvens, rather than a subspecies of E. cloacae. The above findings also highlight that the assignment of subspecies should be prudent as there is no general guideline for defining subspecies using genome data (23) and subspecies assignment  Taxon 2  10  3  Taxon 3  8  2  Taxon 4  14  11  Taxon 5  4  4  Taxon 6  2  1  Taxon 7  1  0  Taxon 8  8  4  Taxon 9  2  0  Taxon 10  3  2  Taxon 11  1  0  Taxon 12  1  1  Taxon 13  2  0  Taxon 14  2  2  Total 1,960 1,496 requires rigorous studies. These studies should include large-scale properly designed investigations on clinical significance such as host specificity of these bacteria to examine the rationale why subspecies should be created and separately recognized (23) and to avoid unnecessary confusion or even chaos. We also found that most genomes labeled as E. cloacae and E. hormaechei are not correctly identified to the species level. The incorrect identification may be due to different reasons. Of note, the Ն95% ANI cutoff alone is widely used for species assignment, but such a cutoff is unable to resolve closely related species (24). Previous studies have corroborated that the stringent Ն96% ANI cutoff is more accurate with better correlation with the 70% DDH cutoff (11) but also highlight that species assignment based on a single algorithm may not be robust. In this study, we employed both ANI with a Ն96% ANI cutoff and isDDH for robust species assignment. In addition, for E. cloacae, phenotype-based tests used in clinical microbiology laboratories commonly identify Enterobacter clinical isolates as E. cloacae as evidenced by the misidentification of strains WCHECL1060 T , 090040, and 090044 T by Vitek II. In contrast, for E. hormaechei, incorrect identification was mainly due to incorrect subspecies assignments as discussed above. This highlights that updated and curated taxonomic assignments are the premise of correct and precise species identification. We suggest that future studies on Enterobacter need to consider the correct species and subspecies identification to provide robust results while avoiding misleading information.
We report two novel Enterobacter species here and found that there were 14 tentative novel Enterobacter species and 18 tentative non-Enterobacter species of the family Enterobacteriaceae, which are clearly listed in the study. This invites more studies on these tentative species by both genome-and phenotype-based methods to establish their species status and to propose appropriate species names. Such studies will further reveal the complicated taxonomy of Enterobacter, a genus of bacterial species with clinical significance.
Conclusions. All subspecies assignments in the genus Enterobacter were incorrect, and their use should be discontinued. E. cloacae subsp. dissolvens is a species and should be renamed E. dissolvens. E. xiangfangensis is not a subspecies of E. hormaechei, while E. hormaechei subsp. oharae and E. hormaechei subsp. steigerwaltii are not subspecies of E. hormaechei but belong to the same species of E. xiangfangensis. E. hormaechei subsp. hoffmannii is a species and should be renamed as E. hoffmannii. E. timonensis should be removed to Pseudenterobacter, a novel genus. Two novel Enterobacter species, E. quasiroggenkampii and E. quasimori, were identified. E. quasiroggenkampii can be distinguished from all known Enterobacter species by its ability to ferment inositol, D-sorbitol, and melibiose but not potassium gluconate, L-fucose, and methyl-␣-D-mannopyranoside. E. quasimori can be distinguished from all known Enterobacter species by its ability to ferment inositol, D-sorbitol, dulcitol, D-turanose, and melibiose but not potassium gluconate, L-fucose, and methyl-␣-D-mannopyranoside. The species identifications for most Enterobacter strains with genomes deposited in GenBank are required to be curated. The most common Enterobacter species seen in clinical samples appears to be E. xiangfangensis. Fourteen novel tentative Enterobacter genome species were also found and warrant further phenotype-based characterizations.

MATERIALS AND METHODS
Strain and initial species identification. The type strain of E. cloacae subsp. dissolvens ATCC 23373 T was obtained from the Guangdong Microbial Culture Collection Center (http://www.gdmcc.net/). Three nonduplicated clinical strains, WCHECL1060 T , 090040, and 090044 T , were all recovered from the blood culture of three different patients with fever as part of routine patient care at West China Hospital of Sichuan University, Chengdu, China, in 2014 or 2016. This study has been approved by the Ethical Committee of West China Hospital, and the informed consent was waived as this study was to retrospectively characterize bacterial strains that were collected as part of routine patient care.
Initial species identification was performed using the Vitek II automated system (bioMérieux, Marcy l'Etoile, France). The 16S rRNA gene sequences of the three strains were obtained as described previously (25) and were compared using a pairwise nucleotide sequence alignment tool (https://www.ezbiocloud .net/tools/pairAlign) using Myers and Miller's algorithm (26). As strains WCHECL1060 T and 090040 Enterobacter Species Identification July/August 2020 Volume 5 Issue 4 e00527-20 msystems.asm.org 13 belonged to the same species, they were subjected to pulsed-field gel electrophoresis by XbaI macrorestriction, which was performed as described previously (27), to determine their clonal relatedness. Whole-genome sequencing. We have reported the draft genome of strain WCHECL1060 T before (22). Genomic DNA of E. cloacae subsp. dissolvens ATCC 23373 T , strain 090040, and strain 090044 T was prepared using the QIAamp DNA minikit (Qiagen, Hilden, Germany), and DNA sequencing libraries were prepared using the NEBNext Ultra II DNA Library Prep kit for Illumina (NEB, Ipswich, MA, USA). Whole-genome sequencing was performed using the HiSeq 2500 Sequencer (Illumina, San Diego, CA, USA) with the 150-bp paired-end protocol and about 200ϫ coverage. Reads were trimmed using Trimmomatic v0.39 (28) under the default setting and were then assembled into contigs using SPAdes v3.11.1 (29) under careful mode. Genome completeness and contamination were examined using CheckM v1.0.18 (30). The genome sequences were reported following recommendations of standards for describing a new taxonomy (23).
Phylogenetic analysis of the genus Enterobacter based on core genes. Whole-genome sequences of the type strains of all species and subspecies within the genus Enterobacter and all other species of the family Enterobacteriaceae (listed in Data Set S1 in the supplemental material) were retrieved from the NCBI database. A core genome phylogenetic tree based on concatenated sequences of core genes was constructed as described previously (31). Prokka v1.12 (32) was used to annotate these genome sequences, and orthologues of these strains were identified using OrthoFinder v2.26 (33) to represent the core genome of these Enterobacteriaceae strains. The gene sequences were aligned and concatenated using MAFFT v7.313 (34) and AMAS v0.98 (35), which were then used to infer a phylogenomic tree using RAxML v8.2.12 (36) with GTR model plus gamma distribution and a 1,000-bootstrap test.
Phenotypic characterization for strains of two novel species. Motility was examined using a CX21FS1 light microscope (Olympus, Tokyo, Japan). The Gram-staining reaction was performed as described previously (38). Growth at different temperatures (4, 15, 20, 25, 30, 35, 37, 45, and 50°C), at different pH values (3.0 to 12.0, at intervals of 1.0 pH unit), and at various salt concentrations (0 to 10% [wt/vol] NaCl) was determined in 15-ml test tubes containing 3 ml tryptic soy broth (TSB; Hopebio, Qingdao, China) after incubation for 2 days in a thermostatically controlled water bath as described previously (39). Anaerobic growth was performed by incubating cultures on nutrient agar for 7 days in an anaerobic bag (bioMérieux). Biochemical characteristics of the three strains were determined using the API 20E kit and API 50CH kit according to the manufacturer's instructions (bioMérieux). Catalase activity was examined by bubble formation after dropping 3% (vol/vol) H 2 O 2 on fresh biomass grown for 24 h on nutrient agar. Oxidase activity was determined using oxidase reagent (bioMérieux). All tests were carried out by incubating at 35°C unless indicated otherwise.
Analysis of whole-cell fatty acids for strains of two novel species. Whole-cell fatty acids of strains WCHECL1060 T , 090040, and 090044 T were analyzed by Guangdong Institute of Microbiology (Guangzhou, Guangdong, China) as described previously (40).
Antimicrobial susceptibility and antimicrobial resistance genes of strains of two novel species. In vitro antimicrobial susceptibility tests were performed by Vitek II using broth microdilution. In addition, MICs of colistin, imipenem, and meropenem were also determined using the microdilution broth method of the Clinical and Laboratory Standards Institute (CLSI) (41). Breakpoints defined by CLSI (41) were applied except for tigecycline, for which breakpoints defined by the European Committee on Antimicrobial Susceptibility Testing (EUCAST; http://www.eucast.org/) were used. Antimicrobial resistance genes of clinical strains WCHECL1060 T , 090040, and 090044 T were identified from genome sequences using the ABRicate program v1.0.1 (https://github.com/tseemann/abricate) to query the ResFinder database (http://genomicepidemiology.org/, accessed 16 April 2020).
Curation of species identification for Enterobacter genome species in GenBank. We used txid547 [Organism:exp] AND "latest refseq" [filter] to search NCBI GenBank and found 1,997 genome sequences labeled Enterobacter (Data Set S2 in the supplemental material, accessed 16 April 2020). All of the 1,997 sequences were retrieved and were then subjected to precise species identification using ANI and isDDH as described above. Strains that have a Ͻ70% isDDH value and a Ͻ96% ANI value with any known Enterobacter species are likely to belong to a novel species, which is temporarily assigned a taxon here as the establishment of a novel species requires phenotypic characterizations in addition to genome analysis.
Data availability. The draft whole-genome sequences of strains ATCC 23373 T , WCHECL1060 T , 090040, and 090044 T have been deposited into DDBJ/EMBL/GenBank under accession numbers WJWQ00000000, LFDQ00000000, RXSJ00000000, and RXRX00000000, respectively. Whole-genome sequences of the type strains of all species and subspecies within the genus Enterobacter and all other species of the family Enterobacteriaceae retrieved from the NCBI database are listed in Data Set S1. The 1,997 genome sequences labeled as Enterobacter in GenBank (accessed 16 April 2020) are listed in Data Set S2.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. TEXT S1, DOCX file, 0.02 MB.