The Class A Carbapenemases BKC-1 and GPC-1 Both Originate from the Bacterial Genus Shinella

Comparative genomics identified the environmental bacterial genus Shinella as the most likely origin of the class A carbapenemases BKC-1 and GPC-1. Available sequences and PCR analyses of additional Shinella species revealed homologous β-lactamases showing up to 85.4% and 93.3% amino acid identity to both enzymes, respectively. The genes conferred resistance to β-lactams once expressed in Escherichia coli. blaBKC-1 likely evolved from a putative ancestral Shinella gene with higher homology through duplication of a gene fragment.

The GC content of the bla BKC/GPC -like genes from all Shinella isolates and their mobile counterparts ranged from 65.3% to 69.7%. This overlaps with that of the larger (Ϯ10,000 bp) genetic contexts in Shinella (63.8% to 66.6%) but not with that of clinical species carrying bla BKC/GPC genes (59.3% to 60.5%).
Altogether, this indicates that the two resistance genes share an ancestor gene that have evolved separately into a more bla BKC-1 -like gene in S. zoogloeoides and a more bla GPC-1 -like gene in, e.g., S. granuli. It is therefore highly plausible that that bla GPC-1 and bla BKC-1 were mobilized from different Shinella spp.
carbapenems. All clones remained susceptible to the cephamycin cefoxitin. In addition, the use of clavulanic acid or avibactam restored a complete susceptibility against amoxicillin or ceftazidime, respectively, a characteristic shared by class A ␤-lactamases (Table 1). This phenotype is in accordance with those reported for BKC-1 and GPC-1 (3,4). Protein alignments using SeaView Software (Prabi, Doua, France) showed that GPC-1 was most closely related to the GPC-like proteins from Shinella sp. strian DD12 (93.3%) and S. granuli (92.6%), whereas BKC-1 shared the highest amino acid identity with S. zoogloeoides (85.4%) (see Fig. S1 in the supplemental material). All variants possessed the typical conserved serine/threonine kinase motifs and the motif involved in the ⍀-loop formation of class A ␤-lactamases (10) (Fig. 2). Deeper alignment analysis showed that BKC-1 displayed a duplication of 16 amino acids, being the repetition of the protein segment from Ala 12 to Ser 27 . Therefore, a putative ancestral protein was designed in silico and named BKC-b (Fig. 2). Aligning the BKC-b sequence with GPZ increased the amino acid identity up to 90.2% compared to 85.4% without the duplication (Fig. S1). Production of BKC-b in E. coli TOP10 showed a weaker resistance profile while having an increased activity against cefoxitin. The use of the I-TASSER (11) in silico tool predicted the tridimensional structures of both BKC-b and BKC-1 and showed that the duplication of the protein segment in BKC-1 modified the ligand binding site of the enzyme and probably led to the increased spectrum of activity observed in BKC-1 compared to that in BKC-b but the loss of its activity against cephamycins.
Here, we provide evidence that the genes bla GPC-1 and bla BKC-1 were most likely mobilized from members of bacterial genus Shinella into clinical species. This conclusion is based on the presence of a conserved locus containing a bla GPC/BKC -like gene in all investigated Shinella species, the lack of associated mobile genetic elements, and high amino acid and nucleotide identities to the clinical counterparts, but not so high that we could assign with confidence the exact origin species. However, it is highly plausible that the origins of bla GPC-1 and bla BKC-1 are Shinella species closely related to S. granuli and S. zoogloeoides, respectively. The resistance phenotype provided by the bla GPC/BKC -like genes is in line with mobilization and transfer driven by antibiotic exposure. The Shinella genus includes mesophilic, aerobic Gram-negative species mainly recovered from environmental samples. For instance, the studied S. granuli and S. zoogloeoides isolates were recovered from sludge in China, while the S. kummerowiae and S. fusca isolates were recovered from root nodules and domestic compost in Korea and Portugal, respectively (6)(7)(8). The presence of a natural and functional ␤-lactamase gene in this genus could be explained by the presence of ␤-lactam-producing microorganisms sharing the same niche (12). Additionally, we show that the BKC-1 protein presented a duplication of its Ala 12 -Ser 27 segment, likely from a putative ancestral protein BKC-b. Hence, the bla BKC-1 gene may have evolved from bla BKC-b , likely under a selective pressure from ␤-lactams, eventually resulting in a more efficient enzyme. This mutation led, on the other hand, to the reduction of its activity against cefoxitin. Emergence of new resistance genes, especially genes providing resistance to antibiotics of last resort, such as carbapenems, represents a major clinical threat. After initial emergence, they are likely to remain undetected and spread silently in the human microbiota for some time. When detected, they are often already widespread (13). Understanding the origin and mobilization history of as many and diverse clinically important resistance genes as possible could enable us to manage risks for future emergence events in a better way. The data presented here provides one additional piece in this large puzzle.
Data availability. The nucleotide sequence of the carbapenemases genes bla GPG , bla GPK , bla GPZ , bla GPF , and bla BKC-b were submitted to GenBank with the following accession numbers, respectively: MT661611, MT661612, MT661613, MT661614, and MT661610.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.3 MB.