The strain P. wenzhouensis A20 carrying the novel β-lactamase gene bla_PRC–1 was isolated from sewage discharged from an animal farm in Wenzhou, China. It was initially identified using a bioMérieux VITEK 2 Compact Instrument (BioMérieux, Marcy L’etoile, France). Further species identification was carried out by 16S ribosomal RNA (rRNA) gene sequencing and finally verified using average nucleotide identity (ANI) (Lachance et al., 2020). The bacterial strains and plasmids used in this work are listed in Table 1.

Minimum inhibitory concentrations were determined with the standard agar dilution method on Mueller-Hinton (MH) agar according to the Clinical and Laboratory Standards Institute guidelines (CLSI, 2020). The MICs of β-lactam inhibitors were determined by broth microdilution in cation-adjusted Mueller-Hinton broth (CAMHB). The MIC was interpreted as the β-lactam concentration at which bacterial growth was no longer observed after 20 h of incubation at 37°C. The MICs of ampicillin-avibactam and ampicillin-tazobactam were determined at a constant concentration of 4 mg/L avibactam and tazobactam, and in combination with a series of increasing concentrations of ampicillin, while for amoxicillin-clavulanic acid, a constant concentration ratio (2:1, amoxicillin:clavulanic acid) was applied. MICs for the tested antibiotics were interpreted according to the guidelines of the CLSI (2020); however, for some antibiotics without CLSI interpretation criteria for P. aeruginosa, breakpoints for Enterobacteriaceae in the CLSI guidelines were used as a reference. E. coli ATCC 25922 was used as a quality control strain. Values are the means of three independent measures.

A Generay Genomic DNA Miniprep Kit (Shanghai Generay Biotech Co., Ltd., Shanghai, China) was used to extract the total bacterial DNA of P. wenzhouensis A20. Genomic DNA was sequenced by both the Illumina HiSeq-2500 and PacBio RS II platforms by Shanghai Personal Biotechnology Co., Ltd. (Shanghai, China). The PacBio long reads were initially assembled by Canu v1.8 (Koren et al., 2017), and hybrid assembly was subsequently conducted using Unicycler v0.4.8 (Wick et al., 2017), with the contigs generated by Canu and all the sequenced reads (including short and long reads) serving as an input. The cyclization of the whole-genome assembly was confirmed through the built-in tools of Unicycler. BWA v0.7.12 (Li and Durbin, 2009) and Genome Analysis Toolkit (McKenna et al., 2010) were used for short read alignment to the draft of the whole-genome assembly to improve assembly quality. Open reading frames (ORFs) were predicted using Prokka v1.14.6 (Seemann, 2014) with default parameters and annotated by the BLAST program with an e-value threshold of 1e-5 against the non-redundant protein sequence (NR) database of the National Center for Biotechnology Information (NCBI) and the UniProt/Swiss-Prot database. Resistance genes were identified using a combination of the ResFinder database (Zankari et al., 2012) and the Comprehensive Antibiotic Resistance Database (CARD) (McArthur et al., 2013). Mobile genetic elements (MGEs) were detected using ISFinder (Siguier et al., 2006) and INTEGRALL (Moura et al., 2009) with default parameters. FastANI v1.31 (Jain et al., 2018) was used to calculate the ANI. ProtParam¹ was used to predict the molecular weight and pI value of PRC-1. The putative signal peptide cleavage site of PRC-1 was predicted by SignalP 5.0 (Almagro Armenteros et al., 2019). Circular maps of the genome and other relatives were drawn using CGView Server (Petkau et al., 2010). Multiple sequence alignments of PRC-1 and other PDC family β-lactamases were performed using MAFFT v7.475 (Katoh and Standley, 2013). A neighbor-joining phylogenetic tree including PRC-1 and other proteins sharing ≥85% amino acid sequence similarity was reconstructed using MEGAX (Kumar et al., 2018). The resulting tree was visualized using the online tool iTol (Letunic and Bork, 2007). Other bioinformatics tools in this study were applied with Python and Biopython scripts (Cock et al., 2009).

The nucleotide sequence of the bla_PRC–1 gene with the upstream promoter region was amplified using PrimeSTAR HS DNA Polymerase (Takara Bio Inc., Dalian, China), and P. wenzhouensis A20 genomic DNA was used as the template. The primers with restriction endonuclease sites (BamHI and HindIII for the forward and reverse primers, respectively) are listed in Supplementary Table 1. The PCR product and the cloning vector pUCP24 were digested with both BamHI and HindIII (Takara Bio Inc.). The resulting DNA fragments were ligated with T4 DNA ligase (Takara Bio Inc.), and the ligated product was transformed into E. coli DH5α with the calcium chloride method. The transformants were selected on LB agar plates supplemented with gentamicin (40 μg/mL). Single colonies were inoculated into LB medium supplemented with the same antibiotics and cultured overnight. Plasmids were extracted from the cultures using a Plasmid Mini Extraction Kit (Generay Biotech Co., Ltd.), and the inserts were verified by PCR and further by DNA sequencing (TsingKe, Shanghai, China). The same method was used to clone the ORF of bla_PRC–1 without the signal sequence into the pCold I vector (the primers are listed in Supplementary Table 1), and the recombinant plasmid (pCold I-bla_PRC–1) was transformed into competent E. coli BL21 cells. The transformants were selected on LB agar plates supplemented with 100 μg/mL ampicillin. For the expression of PRC-1, the recombinant strain (pCold I-bla_PRC–1/BL21) was cultured overnight in LB medium, diluted 100-fold in fresh medium, and then incubated at 16°C and 250 rpm for 2–3 h, which was followed by the addition of 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG, Sigma Chemicals Co., St. Louis, MO, United States) when the OD₆₀₀ reached 0.6 and further incubation for 21 h at 16°C (Choi and Geletu, 2018). His-tagged PRC-1 protein was purified by nickel affinity chromatography with BeyoGold His-tag purification resin (Beyotime, Shanghai, China) and then digested with Enterokinase (GenScript, Nanjing, China) for 40 h at 37°C to remove the His-tag. The protein was identified by SDS-PAGE using a 12% acrylamide separation gel and Coomassie blue G-250 staining.

Kinetic parameters of hydrolysis for β-lactams by the novel β-lactamase PRC-1 were determined on a UV-VIS spectrophotometer (U-3900, HITACHI, Japan) at 37°C in 10 mM phosphate buffer (pH 7.4) in a final reaction volume of 200 μl. The steady-state kinetic parameters (k_cat and K_m) were determined by non-linear regression of the initial reaction rates with the Michaelis-Menten equation in Prism (version 8.0.2) software (GraphPad Software, CA, United States) (Chen et al., 2019). For poor substrates (cefepime, imipenem, and aztreonam), the tested antibiotics were treated as competitive inhibitors of the AmpC enzyme and nitrocefin as a reporter substrate, and the inhibition constants were determined by observing the apparent K_m at various concentrations of inhibitor. Data were analyzed to obtain the values of Vmax and K_m by GraphPad Prism with the competitive model inhibition/non-linear regression curve fit method. k_cat values were determined from the initial rates calculated at saturating substrate concentrations (Faheem et al., 2013). The concentrations of the β-lactamase inhibitors avibactam and clavulanic acid leading to a 50% reduction in hydrolysis of nitrocefin (IC50) were measured after 5 min of preincubation of the enzymes with the inhibitors at 37°C and nitrocefin as the substrate at 100 μM. The IC50 values were determined by non-linear regression analysis (GraphPad Prism, version 8.0.2) using log (inhibitor) vs. response – (three parameters) (Chen et al., 2019). Values are the means of three independent measures.

The chromosome and bla_PRC–1 gene sequences of P. wenzhouensis A20 have been deposited in GenBank under accession numbers CP072610 and MW854031, respectively.

The strain P. wenzhouensis A20 was isolated from sewage discharged from an animal farm in Wenzhou, China. 16S rRNA homologous gene analysis demonstrated that P. wenzhouensis A20 showed the closest relationship with the strain Pseudomonas hydrolytica DSWY01 (NR_170428.1), at 98.50% identity and 100% coverage. The taxonomy of the genus Pseudomonas was developed mainly through phenotypic and biochemical properties (Palleroni, 2010), whereas currently, molecular analysis, as a good supplement, is particularly important. According to previous studies on the phylogeny of Pseudomonas based on 16S ribosomal DNA (rDNA) sequences, there was not enough discrimination at the species level (Mulet et al., 2010), and the multilocus sequence analysis (MLSA) approach based on the four housekeeping genes (the 16S rRNA, gyrB, rpoB, and rpoD genes) has subsequently been applied, which provided better results than the former method but could not accurately discriminate the phylogeny of Pseudomonas species (Mulet et al., 2010). At present, the new gold standards for species delineation are digital whole-genome comparisons by using genome-to-genome distance calculations (GGDCs) or ANI (Lalucat et al., 2020). According to ANI, the threshold to classify the species of a certain bacterium is a cutoff score of ≥95% between the unclassified bacterium genome and all the bacterial genomes available in the public databases. Given the large number of genomes (∼26,600) from 303 Pseudomonas species available in the NCBI database, one genome sequence from each of 40 species which showed the highest similarities of 16S rRNA gene sequences with that of A20 was chosen to perform ANI analysis. Among them, the highest ANI value of 94.25% was found between the A20 genome and that of Pseudomonas mendocina EF27 (GCF_008041835.1) (Supplementary Table 3), indicating that strain A20 might represent a novel species of the genus Pseudomonas. Moreover, when submitting the genome sequence to GenBank of NCBI, the ANI results calculated by the scientists there also did not identify a species for A20, which further confirmed it to be a novel species.

The P. wenzhouensis A20 genome consists of a circular chromosome and does not have any plasmid. The chromosome is 4.4 Mb in length with an average GC content of 62.2% and encodes 4,063 ORFs (Table 2). Screening for A20-homologous genomes (>80% nucleotide sequence identity and >30% coverage) in the NCBI nucleotide database showed that most of the close relatives were derived from 20 species. Comparative genomic analysis revealed that the chromosome of P. wenzhouensis A20 shared the highest sequence similarities with those of Pseudomonas sihuiensis KCTC 32246 chromosome (LT629797.1; 80.83% coverage and 89.54% identity), Pseudomonas alcaliphila JAB1 chromosome (CP016162.1; 80.78% coverage and 89.19% identity), and P. mendocina S5 (CP013124.1; 79.02% coverage and 88.59% identity). Moreover, A20 is basically similar to other species in conserved backbone sequences (Figure 1), among them, similar sequences are represented by lines of different shades, whereas the blank regions indicate differences among A20 and other species.

Antibiotic susceptibility testing showed that A20 exhibited intermediate resistance to nalidixic acid. P. wenzhouensis A20 had the highest MIC levels for fosfomycin (>512 μg/mL), cefazolin (256 μg/mL), and cefoxitin (128 μg/mL) and higher MIC levels for aztreonam (32 μg/mL) and cefotaxime (8 μg/mL) (Table 3). As the breakpoints for fosfomycin, cefazolin, cefoxitin, aztreonam, and cefotaxime were not available for P. aeruginosa in CLSI interpretation criteria, the breakpoints for Enterobacteriaceae in the CLSI guidelines were referred to, and the MIC values of P. wenzhouensis A20 were equivalent to those of resistant enterobacteria for these five antimicrobial agents. The isolate was susceptible to some third- and fourth-generation cephalosporins (e.g., ceftazidime, cefoperazone, cefoselis, and cefepime) and carbapenems (such as meropenem). When analyzing the resistance mechanism of the bacterium, especially for β-lactam antibiotics, we found that only one predicted ampC β-lactamase gene was annotated within the whole genome, which showed an identity of more than 50% with the functionally characterized resistance gene bla_PDC–211 (57.4%, MF281075.1). The predicted gene was then cloned, and the resistance function was further determined.

To determine the resistance characteristics of bla_PRC–1 to β-lactam antibiotics, the coding sequence of bla_PRC–1 together with its upstream promoter region was amplified and cloned into the pUCP24 vector and then transformed into E. coli DH5α. The results revealed that bla_PRC–1 conferred resistance to some penicillins and first- and third-generation cephalosporins (Table 3). Compared with the control strains (DH5α and pUCP24/DH5α), the recombinant strain (pUCP24-bla_PRC–1/DH5α) exhibited increased MIC levels for ampicillin, cefotaxime, and amoxicillin by 8-, 8-, and 32-fold, respectively. The MICs of ceftriaxone, cefazolin, penicillin G, and cefotaxime exhibited a lower increase of 4-fold. However, the recombinant strain did not show any MIC level changes for carbapenems or monobactams. The activity of PRC-1 was poorly inhibited by classical class A β-lactamase inhibitors such as clavulanic acid and tazobactam, while a significant decrease in resistance to ampicillin was observed in the presence of avibactam.

Multiple sequence alignment of the deduced amino acid sequence of PRC-1 with those of functionally characterized β-lactamases revealed that PRC-1 had identities of 57.7, 57.5, 57.3, 57.1, 57.0, 56.8, and 56.8% with PDC-211, PDC-241, PDC-7, PDC-68, PDC-3, PDC-1, and PDC-315, respectively. The deduced amino acid sequence carries the characteristic catalytic residues of the serine active site of β-lactamases, including motifs of S-X-X-K (serine-isoleucine-serine-lysine) with the initial amino acid sequence number at positions 64–67, Y-S-N (tryptophan-serine-asparagine) at positions 150–152, and K-T-G (lysine-threonine-glycine) at positions 315–317 (Figure 2).

Because no detailed resistance spectrum is available for bla_PDC–211, bla_PDC–241, bla_PDC–7, and bla_PDC–68, which shared relatively higher amino acid sequence identities with PRC-1, we compared the resistance profile of bla_PDC–3 with that of bla_PRC–1. PDC-3 (ACQ82808) is a chromosomal AmpC β-lactamase of P. aeruginosa with an amino acid identity of 57.0% (216/379) with PRC-1. Unlike bla_PDC–3, bla_PRC–1 exhibited a relatively narrow resistance spectrum and did not confer resistance to piperacillin, cefepime, ceftazidime, or even aztreonam. In addition, bla_PRC–1 showed lower MIC values than bla_PDC–3 against other β-lactam antibiotics (Barnes et al., 2018). Additionally, we found that the bla_PRC–1 gene did not show any resistance to cefoxitin but showed a relatively lower MIC level for cefazolin, even though P. wenzhouensis A20 showed much higher MIC levels for them. This finding may indicate the existence of unknown resistance mechanisms within P. wenzhouensis A20.

The novel β-lactamase gene bla_PRC–1 is 1140 bp in size and encodes a 379-amino acid putative protein. The mature protein has a predicted molecular weight of 41.48 kDa and a predicted pI of 6.44. The secretory precursor peptide, which consists of 22 amino acids, in PRC-1 is defined by alanine residues at positions 22 and 23. The molecular weight of the protein without the signal peptide is 39.07 kDa. The purified protein PRC-1 exhibited a single band on SDS-PAGE, and its molecular size was in agreement with the predicted one. The kinetic parameters of PRC-1 were determined by measuring the rates of catalysis for various β-lactam antibiotics at different substrate concentrations. The results demonstrated that PRC-1 was a typical cephalosporinase with a high k_cat and strong hydrolytic activity (k_cat/K_m ratios were 220 × 10^–2 μM^–1⋅s^–1) for the first-generation cephalosporin cefazolin. PRC-1 showed moderate hydrolysis activities against some third-generation cephalosporins (cefotaxime and ceftriaxone) and penicillins (ampicillin and benzylpenicillin) but very poor hydrolytic activity against ceftazidime (the third-generation cephalosporin) (Table 4). However, the result of the enzyme kinetic hydrolytic activity test was not completely consistent with the MIC level change of the recombinant strain (pUCP24-bla_PRC–1/DH5α) in the antimicrobial susceptibility test (Table 3). For example, PRC-1 showed hydrolytic activity against ceftazidime, but the recombinant strain carrying bla_PRC–1 did not exhibit a significant change in the MIC of ceftazidime compared with that for the control bacteria. This result may be attributed to its low activity in vitro. A similar case in a previous study reported that BAT-2 and BSU-2 exhibited slight hydrolytic activities against ampicillin, but the two genes did not show detectable resistance activities to ampicillin in the recombinant strains carrying them (Toth et al., 2016). We also found that the catalytic efficiency (k_cat/K_m) of ceftazidime was the lowest among all the tested substrates. Moreover, the affinity of poor substrates, such as aztreonam, cefepime, or imipenem, was higher than those of other substrates in the competitive inhibition test with nitrocefin; the Ki values for aztreonam, cefepime, and imipenem were (307 ± 21) × 10^–3, (134 ± 20) × 10^–3, and (72 ± 7) × 10^–3 μM, respectively, and no hydrolysis was determined for these substrates at saturating concentrations (Table 4), which was in line with the susceptibility test results. Conversely, PDC-3, which shares 57.0% global amino acid identity with PRC-1, exhibited hydrolytic activities against cefepime and imipenem (Rodriguez-Martinez et al., 2009). Both β-lactamases showed similar catalytic efficiencies for benzylpenicillin, whereas PRC-1 displayed higher K_m and k_cat values, revealing that higher turnover numbers are compensated by higher K_m values (Table 4). The IC50 (50% inhibitory concentration) of β-lactamase inhibitors showed that avibactam (IC50: 0.0003922 μM) has a strong inhibitory effect on PRC-1, while clavulanic acid (IC50: 23.43 μM) has a weaker inhibitory effect. This result is in line with the properties of β-lactam inhibitors for AmpC enzymes.

To analyze the possible origin of bla_PRC–1, a total of 40 predicted proteins with an amino acid similarity ≥85% were retrieved from the NCBI-nr database, and all of them were from the genus Pseudomonas. The phylogenetic tree showed that PRC-1 was closest to a putative AmpC β-lactamase found in P. mendocina (WP_147810921.1), and they shared the highest amino acid similarity (91.29% identity and 100% coverage) (Figure 3). These findings indicate the importance of Pseudomonas as a reservoir for PRC-1-like relatives, and additional Pseudomonas genomes must be sequenced to find proteins with higher identities with PRC-1.

To analyze the genetic environments of bla_PRC–1 and its relatives, nine sequences of approximately 20 kb in length with bla_PRC–1-like genes at the center were retrieved from the NCBI nucleotide database, and these bla_PRC–1-like genes shared over 85% amino acid similarities with PRC-1. No mobile genetic element was predicted in its surrounding area. Comparative genomic analysis of the 10 sequences (including the one from this work, P. wenzhouensis A20) revealed that the upstream regions (from folD to acoR) of the bla_PRC–1 and bla_PRC–1-like (ampC) genes of all 10 sequences have a conserved structure in terms of gene context and gene order. Three sequences (P. mendocina S5, P. mendocina 5, and Pseudomonas sp. B11D7D) have the most similar structure to the sequence from this study, except two additional genes (dmlR and bdcA) downstream of the bla_PRC–1-like (ampC) genes that were found in these three sequences. Three sequences (P. mendocina NEB698, P. mendocina NK-01, and P. mendocina strain NCTC 10897) had an extra two or more genes upstream of the bla_PRC–1-like (ampC) genes. The remaining three sequences had the same (P. sediminis B10D7D) or nearly the same (P. alcaliphila JAB1 and P. sihuiensis 32246) gene context as that of the one from this work in the regions downstream of bla_PRC–1; however, the upstream regions of these three sequences were totally different from P. wenzhouensis A20. All 10 sequences were from the genus Pseudomonas, and the three sequences with the most similarity with P. wenzhouensis A20 were from P. mendocina (P. mendocina S5 and P. mendocina CPS5) and an unclassified Pseudomonas (P. sp. B11D7D) (Figure 4). These results suggested that the gene context of bla_PRC–1 and its relatives is conserved in species of the genus Pseudomonas.