Re-updating the taxonomy of Kluyvera genus for a better understanding of CTX-M β-lactamase origin

ABSTRACT

The taxonomic definitions within the Kluyvera genus are still unclear, as several deposits might belong to misidentified species or genus or genome assemblies comprehend large indeterminate nucleotide zones. In this study, we performed a comparative phylogenomic analysis of Kluyvera chromosomes and other selected Enterobacterales. We also included the genomic analysis of chromosomal bla_CTX-M/KLU from Kluyvera isolates and assigned the plasmid-encoded bla_CTX/M genes. The study allowed us to propose a new Kluyvera genomospecies and to better define Kluyvera genomosp. 5. Two new CTX-M sub-groups could also be suggested. Even if no chromosomal bla_CTX-M/KLU gene can be found in K. intermedia and Kluyvera genomosp. 6, accurate identification can be achieved by using these gene sequences in the remaining strains.

IMPORTANCE

The use of whole-genome sequencing (WGS) accelerated the identification of new Kluyvera species proposals, but a rigorous analysis of these sequences is needed for a better definition, including preexisting, and even established species. Kluyvera genomosp. 5 could be more clearly defined, and, among isolates that do not produce a chromosome-encoded CTX-M enzyme, true K. intermedia should be kept within the genus as well as a new genomospecies (Kluyvera genomosp. 6) different from K. intermedia. We could clean up true Kluyvera from those that deserved transfer to other genera, and some deposits as K. ascorbata, K. cryocrescens, K. georgiana, and several Kluyvera sp. to the real species. Two new sub-groups of CTX-M enzymes could be proposed. The accurate identification of the chromosome-encoded bla_CTX-M/KLU gene in Kluyvera isolates could be a useful taxonomic tool to guide the species classification.

INTRODUCTION

CTX-M enzymes represent the most prevalent group of extended-spectrum β-lactamases (ESBLs) among pathogens around the world and are considered a global pandemic (1–3). There are around 265 variants of the CTX-M, including both plasmid- and chromosome-encoded variants in Kluyvera species (β-Lactamase DataBase: http://bldb.eu/) (4), distributed in at least six gene clusters differing in less than 5% amino acid sequence within each group, as previously reported.

Starting in 2001, sequencing of different Kluyvera isolates allowed to propose their chromosomal bla_CTX-M/KLU counterparts as progenitors of the clinically relevant plasmidic variants (5–7), and by 2004, it was already known that some of them were not progenitors but direct counterparts that could be directly recruited by plasmid platforms that allow their expression as ESBLs (8, 9). Since then, several other plasmidic counterparts, with assigned alleles, were found in different Kluyvera isolates, as already reviewed.

However, the information on the origin of the CTX-M family is still obscure. For example, the same species have been proposed as the origin for different subgroups or different species as the origin of a single subgroup, based on what could be considered erroneous classifications according to current whole-genome sequence (WGS) data, including data deposited under incorrect species designation. This could lead to a more complex scenario considering some recent taxonomic proposals. Sequence information for assigning a species designation in analyzed isolates moved from 16S rDNA to the use of some genes (or concatenated genes) as markers, and today, the availability of more WGS data may allow for a more discriminative and, supposedly, stable taxonomic grouping, but only if each deposit is verified for the quality of the information.

The use of WGS crude data accelerated the appearance of new species proposals (10), or a better definition of preexisting ones, and also entailed the availability of a higher number of sequences within Kluyvera but resulted in identification indeterminations.

The genus Kluyvera is widely distributed in diverse niches. Members of this genus have been isolated from water, sewage, food, soil, animals, human clinical specimens, and the environment. Although Kluyvera has been sporadically reported as the cause of clinically relevant diseases until a decade ago, recent studies reveal an increasing number of reports on clinical isolates. The ability to act as an opportunistic pathogen is, perhaps, underestimated due to already known identification difficulties (11–14). Differentially to older reports, Kluyvera are now also hosts for mobile genetic platforms harboring different β-lactamase-encoding genes other than bla_CTX-M (e.g., TEM, serine-β-lactamases, and metallo-β-lactamases) (15–19).

Chromosomal counterparts belonging to the five original sub-groups of acquired CTX-M β-lactamases have been found in Kluyvera (1), and three more were added more recently (11). Thus, even if KLUC-1 (from Kluyvera cryocrescens) was originally proposed as an ancestor for the CTX-M-1 cluster, now Kluyvera genomosp. 5 producing CTX-M-37, -10, -246, or related group 1 enzymes are the most likely origin, while KLUC enzymes are an independent sub-group of β-lactamases (11). KLUA (from Kluyvera ascorbata) remains as ancestor for the CTX-M-2 cluster. The CTX-M-8 sub-group was recently reported as derived from a chromosomal counterpart in Kluyvera genomosp. 3 (11). Chromosome-encoded KLUY-1–4 from Kluyvera georgiana were proposed for the CTX-M-9 sub-group and CTX-M-213 from the chromosomal counterpart Kluyvera genomosp. 2. CTX-M-78, a chromosome-encoded β-lactamase from Kluyvera georgiana 14751, is related to the CTX-M-25 sub-group and is still considered as a probable origin of this cluster. Even if Kluyvera sichuanensis 13608 chromosomal β-lactamase is related to the CTX-M-2 sub-group (the encoded β-lactamase exhibits 86% nucleotide identity and 91% amino acid identity with CTX-M-2), it has been already proposed as a new sub-group. Kluyvera genomosp. 1 strain L2 β-lactamase also represents a new sub-group of CTX-Ms (11).

In this study, we performed a comparative phylogenomic analysis of Kluyvera isolates chromosome and other selected genera of the order Enterobacterales (20) to group Kluyvera isolates that belong to the same clade, re-classifying the outsiders. We included the genomic analysis of the whole bla_CTX-M sequences available and the chromosomal genes present in all Kluyvera species, considering that these latter genes may be useful for a better taxonomic assignment.

MATERIALS AND METHODS

Genomic DNA analysis

Chromosomes from 76 Kluyvera spp. isolates and 74 from different genus of Enterobacterales (Phytobacter, Metakosakonia, Kosakonia, Yokenella, Trabulsiella, Pluralibacter, Raoultella, Erwinia, Hafnia, Leclercia, Lelliottia, Edwardsiella, Chania, Yersinia, Serratia, Siccibacter, Shimwellia, and Klebsiella) were downloaded from the NCBI database (https://www.ncbi.nlm.nih.gov/genome/) between November 2022 and June 2023 (≤250 contigs). The strains are listed in the supplemental material (see Table S1). This selection of Enterobacterales was performed following the results of the article by Alnajar and Gupta (20), in which several conserved proteins were described as taxonomically useful to separate Kluyvera species and other Enterobacteriaceae (Klebsiella spp., Raoultella spp., Trabulsiella spp., Yokenella spp., Hafnia spp., Erwinia spp., Leclercia spp., and Lelliotta spp.) within the Klebsiella clade. Kosakonia sp. and Phytobacter sp. were included due to previous average nucleotide identity (ANI) results with some of the isolates received as Kluyvera (11).

Chromosomal assembly or SRA sequences (https://www.ncbi.nlm.nih.gov/sra/) were analyzed using Unicycler Version 0.5.0 (21) for de novo assembly paired forward and reverse reads, resulting in fasta files. Chromosomal assembly data from isolates were used to perform ANI using the OrthoANIu (https://www.ezbiocloud.net/tools/ani) (22), confronting each genome with the corresponding representative genome (Table S2). Minimal cutoff points of 95% OrthoANI values were considered to represent species delineation.

Genome phylogeny

We performed all phylogenomic analyses in Galaxy Version 3.13.0 + galaxy s2 platform. Prokka (v. 1.12) was used to produce “*.gff3” output files for each strain (23), and Roary pan-genome pipeline Version 3.13.0 was used for genome alignment (24). All alignments were used as entries in an single nucleotide polimorfisms (SNP)-distance analysis, obtaining the SNP-distance matrix (25); this matrix was displayed only for Kluyvera isolates as a seaborn heatmap obtained with Python (v. 3.13.114.0; https://www.python.org/) (Fig. S1). The SNP values were analyzed by Gubbins to reorganize the core genome by removing the recombination in genomes (26). The analysis of the bla genes was conducted using ClustalX (http://clustalx.software.informer.com/2.1/) to align all sequences. The molecular evolution model was estimated with IQ-Tree (Version 2.1.2) (27). The phylogenetic trees were built with PhyML (v.3.1) using the maximum-likelihood method, with 1,000 bootstraps replications using ultrafast bootstrap (28, 29). The resulting phylogenetic trees were visualized and edited using the FigTree program (http://tree.bio.ed.ac.uk/software/figtree/). All phylogenetic trees were midpoint rooted.

RESULTS

Genome analysis

Figure 1 represents the middle point rooted phylogenetic tree of the included Enterobacterales (150 complete chromosomes), including the named Kluyvera isolates (Table S1). Most of the Kluyvera strains group in the same clade. Nevertheless, Kluyvera intermedia 1951106–13, 19511106_12, and 1951106_11 isolates, Kluyvera intestini GT16, and Kluyvera sp. Nf5 group with Phytobacter diazotrophicus. On the other hand, K. intermedia MGYG-HGUT-025521, FOSA7093, 1953540_12, and 1953540_14 group with Phytobacter ursingii (and should be excluded for any sequence analysis within the genus). K. ascorbata 62–59 has as closest relatives several Enterobacter species and does not harbor a chromosome-encoded CTX-M variant but a class A serine β-lactamase displaying 85% nucleotide identity with a deposited sequence from Enterobacter cloacae isolate (GenBank: CP035633.1, data not shown). This suggests that this isolate may belong to the Enterobacter genus and not to Kluyvera. Unfortunately, sequence quality reveals some errors, and therefore, sequences from Enterobacter species isolates were not included in the phylogenomic analysis. Anyway, this deposit is clearly out of our selected genus.

FIG 1.

Phylogenetic tree of the chromosomes from Kluyvera and other related Enterobacterales was performed with IQ-Tree using the maximum-likelihood method based on 67,985 core gene SNPs (bootstrap replications: 1,000). Kluyvera genus clade is colored with blue lines. Red names represent the K. ascorbata 62–59 genome, Kluyvera intestinii GT16, Kluyvera sp. Nf5, and K. intermedia isolates grouping with P. diazotrophicus and P. ursingii.

Similarly, our strain 4105 (received as Kluyvera sp.), Metakosakonia sp. MRY16-52, and CPBM-RX-33 seem to be wrongly identified, as they group with P. diazotrophicus.

The rest of Kluyvera isolates that group in the same clade are represented in the middle point rooted phylogenetic tree as shown in Fig. 2; Fig. S1; Tables S2 and S3. The Kluyvera genus seems to be composed of ten sub-clades including up-to-date 13 misidentified genomes. The range of SNP differences is displayed in parenthesis:

FIG 2.

Phylogenetic tree of the chromosomes from Kluyvera was obtained with PhyML using the maximum-likelihood method based in 35,579 core gene SNPs (bootstrap replications: 1,000). Color lines represent each clade (group) of isolates classified according to the species suggested (from top to bottom): Kluyvera genomosp. 1 (dark blue), K. cryocrescens (orange), Kluyvera genomosp. 5 (light blue), Kluyvera genomosp. 6 (brown), K. intermedia (violet), K. georgiana (green), Kluyvera genomosp. 3 (magenta), Kluyvera genomosp. 2 (blue), K. sichuanensis (dark purple), and K. ascorbata (red).

1. K. georgiana ATCC51603, 14751, Igbk15, WCH1410, and Colony 392 strains and the misidentified K. ascorbata 11A19CP0075 (≤300).

2. Kluyvera genomosp. 3 PO2S7, YDC799 strains, two Kluyvera spp. Awk3 and CRP, and the misidentified K. georgiana Colony288 and K. georgiana HRGM_genome_0064 strains (≤300).

3. Kluyvera genomosp. 2 KA2 and MGYG-HGUT-02491 and the misidentified K. ascorbata KA1, KA5, R10, M60, and 131-CS8 isolates (≤73).

4. K. ascorbata isolates including ATCC33433, NCTC9737, 220, 4463, 58, LFK, MGYG-HGUT-03358, 3162, 68, SK, 711, FDA_CDC_AR-0144, SRR12180963, TP1631, Trace242, 280 (ATCC 14236), 8633, OT2, SRR12180963_bin_2_metawrap_v1_3_0_MAG (≤ 373), and the conspicuously distant Colony413 strain (1,100–1,200).

5. Kluyvera sichuanensis isolates 090646, 13608, Kluyvera sp. EC51, and the misidentified K. cryocrescens 116 (≤454).

6. Kluyvera genomosp. 5 strain 169 and the misidentified K. ascorbata 220CK_00278 and K. intermedia INSAq229 strains (108 between the misidentified strains and 1,022–1,033 between them and strain 169).

7. K. cryocrescens including isolates of K. cryocrescens 1919, 4701, RIT-550, 88CZ1, ERR4757816_bin_13-CONCOCT_v1_1_MAG, NCTC 10483, NBRC 102407, SRR108100009_bin_1_metawrap_v1_3_0_MAG, SCW13, NCTC12993, SRR12180964_bin7_metawrap_v1_3_0_MAG, and the misidentified K. ascorbata N10220128 and N9220128 strains (≤159).

8. Kluyvera genomosp. 1 represented by L2 alone.

9. K. intermedia ATCC 33110, NCTC 12125, HR2, and N2-1 strains (≤129).

10. A new Kluyvera genomosp. 6 that includes deposits K croycrescens D51 and e02842a0-063 strains (0).

Phylogeny of bla_CTX-M

To complete the analysis, we performed the phylogenetic tree of all bla_CTX-M genes (http://bldb.eu/), including the assigned and some not yet assigned chromosome-encoded bla_CTX-M and bla_KLU from Kluyvera isolates (see Fig. 3).

FIG 3.

Phylogenetic tree of chromosome and plasmid-borne bla_CTX-M/KLU genes was obtained with IQ-Tree using the maximum-likelihood method (bootstrap replications: 1,000). Assigned chromosome-encoded bla_CTX-M/KLU genes are shown in red; names of isolates harboring non-assigned bla genes are shown in blue. Each bla_CTX-M group is represented in a different color.

In this phylogenetic tree of all bla_CTX-M/KLU genes, eight sub-groups of CTX-M-enzymes with at least a chromosomal gene can be evidenced, considering all isolates referred to the original deposit name. The sub-group 1 (bla_CTX-M-1 is the representative member) includes a few chromosomal genes, bla_CTX-M-3 from an isolate perhaps erroneously deposited as K. ascorbata, and bla_{CTX-M-37, -10, -146} from Kluyvera genomosp. 5. The sub-group 2 (bla_CTX-M-2 as the representative member) has all chromosomal genes from K. ascorbata like bla_KLUA-1, bla_KLUA-3, bla_KLUA-4, and bla_KLUA-12, which have 100% nucleotide identity with bla_CTX-M-124, bla_KLUA-2 (also deposited as bla_CTX-M-5), bla_CTX-M-76, bla_CTX-M-77, bla_CTX-M-95, and bla_CTX-M-115, and the bla gene from Colony413 strain which is the most distant gene with a 94% nucleotide identity with bla_CTX-M-2. The sub-group 8 (bla_CTX-M-8 as the representative member) includes chromosomal genes from Kluyvera genomosp. 3 (YDC799, PO2S7), Kluyvera sp. (CRP, Awk3), and the erroneously deposited as K. ascorbata (Colony392), related to bla_CTX-M-8 and bla_CTX-M-40. The sub-group 9 (bla_CTX-M-9 as the representative member) contains the misidentified K. ascorbata strain 60, with a chromosomal bla_CTX-M-9 gene, and other Kluyvera genomosp. 2 strains (KA2 and the misidentified strains K. ascorbata KA1, KA5, R10, and 131-CS8) with a chromosomal bla_CTX-M-213 gene. The sub-group 25 (bla_CTX-M-25 as the representative member) includes the chromosomal gene bla_CTX-M-78, bla_CTX-M-152, bla_CTX-M-185, and bla_CTX-M-205, all within K. georgiana. The sub-group KLUC (chromosomal bla_KLUC-1 as the representative member) contains chromosomal bla_KLUC genes like bla_KLUC-2 (88CZ1 strain), three strains with a single mutation (1919, ERR4757816_bin_13, and SRR12180964_bin_7), two strains with two mutations (N10220128 and N9220128), and two strains with three nucleotide changes (4701 and RIT-550) compared with bla_KLUC-5, named all as K. cryocrescens except strains N10220128 and N9220128 erroneously assigned as K. ascorbata.

Also, two new sub-groups are described: sub-group “NEW-1,” including only one chromosomal bla gene from Kluyvera genomosp. 1 strain L2, and sub-group “NEW-2,” including the bla genes from K. sichuanensis strains (090646, 13608) and the erroneously assigned as K. cryocrescens 116, all having only 84–86% nucleotide identity to bla_CTX-M-76 gene, their closest relative.

“Outlier” plasmid-borne bla_CTX-M members include (i) bla_CTX-M-45; (ii) a cluster containing the proposed genes encoding for hybrid enzymes derived from CTX-M-1 and CTX-M-9 members like bla_CTX-M-64 and bla_CTX-M-199; (iii) bla_CTX-M-132; (iv) bla_CTX-M-123 and bla_CTX-M-153; (v) bla_CTX-M-234; and (vi) the rest of hybrid bla_CTX-M genes, resulting in a cluster with bla_CTX-M-137 and bla_CTX-M-221 genes. Figure 4 represents the phylogenetic tree of all chromosomal bla_CTX-M/KLU genes from Kluyvera isolates. The phylogenetic relationship among chromosomal bla_CTX-M/KLU genes from Kluyvera isolates remains conserved when compared to the phylogenetic tree of Kluyvera genomes (Fig. 2).

FIG 4.

Phylogenetic tree of only chromosome borne bla_CTX-M/KLU genes from Kluyvera, built with IQ-Tree using the maximum-likelihood method (bootstrap replications: 1,000). The lines hold the color pattern of Fig. 2.

DISCUSSION

Even if an increase in clinically relevant Kluyvera isolates is reported worldwide, only a few studies include genomic data. From them, a vast number of incomplete and/or misassembled sequences are available in databases, leading to misidentifications. In the last few months (August 2023), after we finished our primary analysis, the NCBI database modified the way by which genomic data are shown, announcing some likely errors, i.e., the genome assemblies of K. intermedia FOSA7093, K. georgiana Colony392, and K. ascorbata Colony413 have unusually “too small” lengths, and Kluyvera sp. Nf5 isolate “is contaminated” (as examples). This may justify that, upon analysis of the obtained chromosome sequence of K. georgiana Colony392, no full chromosome-encoded bla gene could be localized, likely due to the presence of a multiple and continuous “N” indetermination zone in the sequence.

1. Most K. georgiana isolates and K. ascorbata 11A19CP0075, encoding chromosomal enzymes related to sub-group of CTX-M-25, group in the same sub-clade, for which this K. ascorbata isolate should be re-classified as K. georgiana as well.

2. Confirming genomic clustering most isolates having a chromosomal bla_KLUC gene belonged to K. cryocrescens.

3. Again, all isolates assigned as Kluyvera genomosp. 2 displayed a chromosome-encoded from CTX-M-9 sub-group.

4. All the Kluyvera isolates harboring a chromosomal bla_CTX-M related to the plasmidic-encoded bla_CTX-M-8 group in the same sub-clade, belonging to Kluyvera genomosp. 3.

5. All chromosome-encoded bla_CTX-M/KLU genes related to bla_CTX-M-2 are present in the true K. ascorbata sub-clade, including the phylogenetically distant Colony413 strain. Regarding K. ascorbata Colony413 data, re-sequencing would probably better define if it could be retained in the same cluster (CTX-M-2 sub-group) or if it belongs to a new genomospecies.

6. Other isolates grouped with K. sichuanensis. This species represents the origin of a new sub-group of chromosome-encoded CTX-M enzymes (as none of them have been validated number yet, they were designed preliminary as NEW-2 sub-group, to be transferred to the first to be accepted).

7. The chromosome-encoded bla_CTX-M-37 from Kluyvera genomosp. 5 strain 169 can be considered the best candidate for sub-group bla_CTX-M-1 origin, and all the different isolates having a chromosomal bla_CTX-M-1 related enzyme should be re-classified as Kluyvera genomosp. 5 species as well (11).

8. As previously described, Kluyvera genomosp. 1 L2 strain is the single member of this sub-group; it harbors a chromosomal bla_CTX-M/KLU gene with only 80% nucleotide identity to bla_CTX-M-29 (its closest relative that belongs to the bla_CTX-M-1 sub-group). This isolate would represent the origin of a still not disseminated sub-group of CTX-M enzymes, designed preliminary as NEW-1 sub-group, to be transferred to the first accepted numbered).

As stated before, among isolates that do not display any chromosomal bla_CTX-M counterparts, some should be retained as K. intermedia, while others should be considered as a species different from K. intermedia; we propose they should be considered as Kluyvera genomosp. 6, to keep the genomospecies numbering correlation. These two species are so far the only ones without a chromosomal bla_CTX-M/KLU counterpart.

As the genus includes species with or without chromosomal counterparts to the CTX-M family, it can be considered that it might evolve from a common ancestor in which these genes were introduced from an unknown source.

According to our results, several sequences deposited as K. intermedia, Phytobacter massiliensis, Metakosakonia sp., and P. ursingii should be re-classified as P. diazotrophicus, and it is likely that other genome deposits as Phytobacter or Metakosakonia should be also reevaluated. Under a similar approach, other K. intermedia should be considered as P. ursingii.

A stricter WGS analysis is mandatory to avoid sequence/assembly mistakes that obscure a clear taxonomic evaluation. Even if this is true for any deposit, in the case of microorganisms with few researchers working on them, it is likely that mistakes will remain for longer periods before being amended.

In our study, we observed that (except for the proposed Kluyvera species that do not display them) the amplification and sequence analysis of chromosomal bla_CTX-M/KLU genes can be considered an inexpensive and easy-to-perform taxonomic tool to classify bacterial isolates as members of the genus.

As already mentioned previously, identification within the genus Kluyvera is elusive. So far, the newly recognized species cannot be differentiated by conventional biochemical tests or even by MALDI-TOF MS, as they have not been fully incorporated into databases. Moreover, some of the previously defined species are inconsistent with what is inferred by full genome sequencing.

Only genome sequences showing high quality score should be used for the taxonomic classification of potential members of this genus (i.e., FastQC, number of contigs, N50, and coverage). As Kluyvera are only infrequent isolates as compared with other Enterobacterales, only laboratories with the capacity for sequencing all isolates may benefit from straight sequencing at the clinical laboratory level. Different would be the case for the few research groups working on Kluyvera that would be able to make more efficient use of full sequencing, as well as a PCR strategy for identification by using chromosomal bla_CTX-M sequencing.

Finally, and as already proposed, as at least some plasmid-encoded CTX-M enzymes have a strict identity to chromosome-encoded enzymes (and vice versa), different nomenclatures (i.e., KLU or CTX-M) should be avoided, and a single (consensus) nomenclature should be applied univocally for a single enzyme sequence.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/spectrum.04054-23.

Figure S1. spectrum.04054-23-s0001.jpg.

Heatmap of SNP distance matrix.

Table S1. spectrum.04054-23-s0002.xlsx.

Genomes for computational analysis.

Table S2. spectrum.04054-23-s0003.xlsx.

Proposed genus and species of the Kluyvera isolates.

Table S3. spectrum.04054-23-s0004.xlsx.

SNP distance matrix of Kluyvera chromosomes.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.