To investigate the antimicrobial resistance of large intestinal bacteria in animals, we collected anal feces samples from different animals. P. wenzhouensis R33 was isolated from an anal swab of a New Zealand White rabbit in an animal farm in Wenzhou, southeastern China. The anal swab was collected as follows: a sterile cotton swab was inserted into the anus of rabbit for 3–5 cm, gently rotated, and then immediately put it into a sterilized screw-capped specimen collection tube containing normal saline solution (0.9% sodium chloride). The isolate was randomly isolated by the streak plate method with a normal Luria-Bertani agar plate, initially identified using the Vitek-60 microorganism autoanalysis system (BioMerieux Corporate, Craponne, France) and verified by analysis of the 16S rRNA gene sequences. The result was finally confirmed by determining the average nucleotide identity (ANI) using FastANI and in silico DNA–DNA hybridization (isDDH) (Jain et al., 2018). The strains and plasmids used in this work are listed in Table 1.

The minimum inhibitory concentrations (MICs) were determined using the agar dilution method with Mueller-Hinton (MH) agar plates following the guidelines of the Clinical and Laboratory Standards Institute (CLSI). Susceptibility patterns were interpreted according to the CLSI breakpoint criteria (CLSI, 2019) and the guidelines of the European Committee on Antimicrobial Susceptibility Testing (The European Committee on Antimicrobial Susceptibility Testing, 2019) for Enterobacteriaceae. Escherichia coli ATCC 25922 was used as a reference strain for quality control. The MIC was determined in triplicate on Mueller-Hinton (MH) agar plates with two-fold serial dilutions of the antibiotics. The plates were incubated at 37°C for 16–20 h before analyzing the results.

Genomic DNA of P. wenzhouensis R33 was extracted from bacterial culture by an AxyPrep Bacterial Genomic DNA Miniprep Kit (Axygen Scientific, Union City, CA, United States). DNA sequencing was performed by using the Illumina HiSeq-2500 and PacBio RS II platforms by the Shanghai Personal Biotechnology Co., Ltd. (Shanghai, China). The Illumina short reads and PacBio long reads were initially assembled by Canu v2.1 (Koren et al., 2017) and SPAdes v3.14.1 (Bankevich et al., 2012), respectively. Pilon v1.23 (Walker et al., 2014) was employed for further correction to improve assembly quality by mapping short reads aligned to the draft of the whole-genome assembly. Potential open reading frames (ORFs) of the assembled genome were predicted using Prokka v1.14.6 (Seemann, 2014), and functional annotation of these proteins was performed by BLAST analysis with an e-value threshold of 1e-5 against the nonredundant protein sequence database of US National Center for Biotechnology Information (NCBI) and the UniProt/Swiss-Prot database. Antimicrobial resistance genes were annotated using ResFinder (Zankari et al., 2012) and Resistance Gene Identifier (RGI) v4.0.3 of 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). ANI was calculated using FastANI v1.31 (Jain et al., 2018). CGView Server (Petkau et al., 2010) was used to visualize the basic genomic features of chromosomes and perform comparative genomic analysis. The promoter region of aac(2′)-If was predicted by BPROM (2016)². The molecular weight and pI value of AAC(2′)-If were predicted using Expasy ProtParam Tool (2021)³. Multiple alignment of the amino acid sequences of and neighbor-joining phylogenetic tree construction for AAC(2′)-If and other AACs were performed using MAFFT v7.475 (Katoh and Standley, 2013) and MEGAX (Kumar et al., 2018), respectively. The sequence retrieval, statistical analysis, and other bioinformatics tools used in this study were written using Python (2021)⁴ and Biopython (Cock et al., 2009).

The coding sequence of aac(2′)-If, along with its upstream promoter region, was amplified by PCR primers with a pair of flanking restriction endonuclease adaptors for BamHI and HindIII (Takara Bio, Inc., Dalian, China) introduced at the 5′ and 3′ ends, respectively (Supplementary Table S1). The PCR product was digested with BamHI and HindIII, and then ligated into the pUCP20 vector with a T4 DNA ligase cloning kit (Takara Bio, Inc., Dalian, China). The recombinant plasmid pUCP20-pro-aac(2′)-If was transformed into competent E. coli DH5α cells by the calcium chloride method (Chan et al., 2013), and the transformant was selected on Luria-Bertani agar plates supplemented with 100 μg/mL ampicillin. The cloned insert sequence (aac(2′)-If with its upstream promoter region) in the recombinant plasmid of the transformant was further confirmed by both restriction enzyme digestion and Sanger sequencing (Shanghai Sunny Biotechnology Co., Ltd., Shanghai, China).

The ORF of the aac(2′)-If gene was amplified with the orf-aac(2′)-If primers listed in Supplementary Table S1 and cloned into the pCold I cold shock expression vector (Qing et al., 2004). The resultant plasmid pCold I-aac(2′)-If was introduced into E. coli RosettagamiB (DE3) competent cells, and transformants (pCold I-orf-aac(2′)-If/RosettagamiB) were selected on LB agar plates containing 100 μg/mL ampicillin.

The overnight culture of the recombinant strain (pCold I-orf-aac(2′)-If/RosettagamiB) was cultured in LB broth containing 100 μg/mL ampicillin at 37°C. When the OD at 600 nm reached 0.6, isopropyl D-thiogalactopyranoside (IPTG) was added at a concentration of 1 mM to induce the expression of AAC(2′)-If, and cell cultivation was continued for 18 h at 16°C. The cells were collected, resuspended in phosphate buffered saline (pH = 7.4), and disrupted with a French pressure cell. The recombinant protein was purified using BeyoGold His-tag Purification Resin and subsequently eluted by the nondenatured eluent (50 mM NaH₂PO₄, 300 mM NaCl, 50 mM imidazole) of the His-tag Protein Purification Kit (Beyotime, Shanghai, China) according to the manufacturer’s instructions. The purity of the protein was confirmed using sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and subsequent staining with Coomassie Brilliant Blue. The protein concentration was determined spectrophotometrically using a BCA protein assay kit (Thermo Fisher Scientific, Rockford, IL, United States).

To analyze the expression of aac(2′)-If, total RNA was extracted from the bacteria cultured in LB broth grown to OD₆₀₀ = 0.5 and then treated with or without the addition of 1/4 MIC ribostamycin (32 μg/mL), neomycin (1 μg/mL), and gentamicin (0.25 μg/mL) for 0.5, 1, 2, 4, and 24 h, respectively. Then total RNA was isolated using TRIzol reagent (Sigma, Shanghai, China) according to the manufacturer’s protocol and quantified by an Implen NanoPhotometer (Implen GmbH, Munich, Germany). DNA-free RNA was confirmed by PCR amplification of the Providencia 16S rRNA gene. cDNA was synthesized using the PrimeScript RT-PCR Kit (Takara Bio, Inc., Dalian, China). RT Q-PCR was performed using the corresponding cDNA from each sample. The primers used for RT Q-PCR are listed in Supplementary Table S2. RT Q-PCR was performed in a CFX96^TM Touch Real-Time PCR Detection System (Bio-Rad Laboratories, Hercules, CA, United States) by monitoring the increase in fluorescence in real time using SYBR qPCR Master Mix (Vazyme Biotech, Nanjing, China). Relative quantification was performed using the CT method (Livak and Schmittgen, 2001) with the housekeeping genes 16S rRNA as references. Comparisons of expression levels with or without antibiotics treatment were analyzed using one-way ANOVA (LSD test). P ≤ 0.05 was considered significant.

The kinetic parameters for AAC(2′)-If activity were measured spectrophotometrically by following the production of coenzyme A (CoASH), which was produced upon the transfer of the acetyl moiety to the aminoglycoside. The thiol group of CoASH reacts with 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB), which is replaced by 4,4′-dithiodipyridine, and monitoring was performed at 412 nm to measure an increase in the absorbance of the formed product, pyridine-4-thiolate (TNB), as previously described (Hegde et al., 2001; Galimand et al., 2015). Kinetic assays were performed in a 200 μl reaction mixture containing 25 mM 2-(N-morpholino)ethanesulfonic acid (MES; pH 6.0), 1 mM EDTA, 80 μM acetyl-CoA, 2 mM DTNB, and variable concentrations of aminoglycosides (5–150 μM) (Franklin and Clarke, 2001), and the reactions, initiated by the addition of 2 μg purified enzyme (final concentration), were monitored using a Synergy Neo2 Multi-Mode Microplate Reader (Biotek, VT, United States) at room temperature for 10 min. The steady-state kinetic parameters (k_cat and K_m) were determined by nonlinear regression of the initial reaction rates with the Michaelis–Menten equation using GraphPad Prism 9 (GraphPad Software, San Jose, CA, United States).

The aac(2′)-If sequence has been assigned GenBank accession number MW984427. The complete genome sequence of P. wenzhouensis R33 presented in this study has been deposited in GenBank under accession numbers CP072453 and CP072454 for chromosome and plasmid pR33-1 of P. wenzhouensis R33, respectively.

A 16S ribosomal RNA gene homology analysis showed that the 16S rRNA gene of strain R33 shared the highest similarity with that of P. vermicola strain OP1 (NR_042415.1), with 98.01% identity and 99.00% coverage. However, the genome sequence of strain R33 shared 76.93–81.47% ANI and had a 19.60–52.40% in silico DNA–DNA hybridization (isDDH) score with the species-classified Providencia strains (Supplementary Table S3). They were too low to reach the cutoff (≥95–96% for ANI and 70% for isDDH) to define a classified bacterial species, which suggested that strain R33 is a new species of the genus Providencia (Goris et al., 2007; Hu et al., 2019). According to the criteria for species names (International Code of Nomenclature of Prokaryotes, 2019), we named it Providencia wenzhouensis R33. This isolate was deposited in China Center for Type Culture Collection, Wuhan, China (CCTCC AB 2021339).

Providencia wenzhouensis (wen.zhou.en’sis. N.L. fem. adj. wenzhouensis pertaining to Wenzhou, Zhejiang Province, China, where the type strain was isolated) cells are Gram-negative, non-motile, non-spore-forming, non-gas-producing, rod-shaped. Colonies are creamy white, circular, moist, convex, with entire edges on nutrient agar and translucent on MacConkey agar. The strain is catalase-positive and oxidase-negative, gives positive in methyl-red reaction, citrate utilization, and nitrate reduction. The strain is negative for Voges-Proskauer, indole, urease, and the production of H₂S. Acid is produced from adonitol, salicin, galactose, arabinose, gluconate, glucose, mannitol, D-mannitol, phenylalanine deaminase, but not from arginine, glycerol, inulin, inositol, lactose, lysine, ornithine, trehalose, maltose, pyruvate, sorbitol, sorbose, sucrose, and urea. Differential biochemical characteristics from other members of the genus Providencia include acid production from citrate, mannitol, and salicin but not from urea.

The MICs of 26 antibiotics were tested for strain R33 as shown in Table 2. This strain showed resistance to five antibiotics, including fosfomycin, imipenem, nalidixic acid, and polymyxins B and E, and higher MIC values for ribostamycin (128 μg/ml) and ceftiofur sodium (512 μg/ml, an antibiotic for animals only), which have no established breakpoints (Table 2).

The whole genome of P. wenzhouensis R33 was found to consist of a chromosome of approximately 4.75 Mb in length, encoding 4,335 ORFs with an average GC content of 41.57% (Table 3 and Figure 1), and a circular plasmid designated pR33-1, which is 124,159 bp in length and encodes 136 ORFs (Table 3). According to genome sequencing, only three chromosomal genes with a similarity of ≥80% with housekeeping genes, namely, crp (encoding cAMP-receptor protein), rsmA (encoding translational regulator CsrA), and E. coli gyrA (encoding DNA gyrase subunit A) (with similarities of 97.6%, 86.89%, and 83.72%, respectively), were identified. No predicted aminoglycoside resistance genes with similarity ≥80% were found. Among all 11 predicted aminoglycoside resistance-related genes (with similarity <80%) annotated from the whole genome, only one predicted AAC gene [an aac(2′)-I homologous gene designated aac(2′)-If in this work] had an amino acid sequence similarity of 70.79% with the aac(2′)-Ia gene of known function, while the remaining 10 genes (CpxA, KpnH, BaeR, KpnF, KdpE, KpnG, TolC, BaeS, CpxR, and ykkD) encoded for efflux-related proteins (Supplementary Table S4). We subsequently cloned the putative AAC(2′)-I gene (aac(2′)-If), and its function was further verified.

The novel 2′-N-acetyltransferase gene aac(2′)-If was revealed to be 537 bp in length and to encode a protein of 178 amino acids (ca. 20.33 kDa) with a pI value of 4.47. All 16 of the aac(2′)-If homologs (≥65% nucleotide sequence similarity) retrieved from the NCBI nucleotide database were derived from Providencia species (mainly P. rettgeri and P. stuartii). As mentioned above, the predicted protein sequence (no nucleotide sequence available) shared the highest amino acid sequence identity of 95.51% with AAC(2′)-If, which was a hypothetical GNAT family N-acetyltransferase (WP_206082813.1) from P. stuartii.

The results of multiple sequence alignment of AAC(2′) enzymes showed that AAC(2′)-If shared at most 70.79%, 35.03%, 35.26%, 30.10%, and 31.21% identity with the five AAC(2′)-I subgroups AAC(2′)-Ia (AAA03550.1), AAC(2′)-Ib (AAC44793.1), AAC(2′)-Ic (AAB17563.1), AAC(2′)-Id (AAB41701.1), and AAC(2′)-Ie (CAR72650.1), respectively. Except for the AAC(2′)-Ia subgroup, which had the highest identity value of more than 70.0%, the other four subgroups showed less than 40.0% identity (Figure 2). A phylogenetic tree showed that AAC(2′)-If clustered closest to AAC(2′)-Ia (Figure 3).

To investigate the functional role of AAC(2′)-If, the sequence encoding the ORF of AAC(2′)-If with its promoter region was cloned into the pUCP20 vector, and the recombinant plasmid (pUCP20-aac(2′)-If) was transformed into E. coli DH5α. The MICs of several aminoglycoside antibiotics against the transformant harboring aac(2′)-If (pUCP20-aac(2′)-If/DH5α) were determined. The presence of plasmid borne aac(2′)-If in E. coli led to elevated MICs of ribostamycin, neomycin, paromomycin, tobramycin, micronomicin, netilmicin, gentamicin, and sisomicin (256-, 128-, 256-, 16-, 8-, 8-, 4-, and 4-fold increases, respectively) in comparison with those for the control strains (DH5α or DH5α carrying the vector pUCP20) (Table 2). However, as expected, no changes in MICs of streptomycin and spectinomycin were observed. Additionally, the susceptibility of the aac(2′)-If-carrying R33 to aminoglycosides and other antibiotics are also shown in Table 2, with relatively high MIC values (≥4 μg/mL) observed for several aminoglycosides (i.e., apramycin, neomycin, paromomycin, ribostamycin, and streptomycin), spectinomycin, polymyxins B and E, nalidixic acid, norfloxacin, chloramphenicol, florfenicol, cefoxitin, ceftiofur, imipenem, and fosfomycin.

Investigation of the acetyltransferase activity and kinetic parameters of AAC(2′)-If showed that of the 10 aminoglycosides tested, the enzyme was able to acetylate ribostamycin, neomycin, tobramycin, micronomicin, sisomicin, and gentamicin, but worse acetyltransferase activity was detected for netilmicin and kanamycin (Table 4). The highest catalytic efficiency was observed with ribostamycin [k_cat/K_m ratio was (3.72 ± 0.52) × 10⁴ M^–1⋅s^–1], which was found to be the best aminoglycoside substrate, whereas kanamycin was the worst substrate [k_cat/K_m ratio = (8.71 ± 1.25) × 10³ M^–1⋅s^–1]. These differences appeared to result from large differences in the turnover rates (k_cat), which varied by almost six-fold compared to the affinity (Table 4).

To investigate the effect of aminoglycoside antibiotics on the aac(2′)-If gene expression, we determined the change of the aac(2′)-If gene expression levels with or without ribostamycin, neomycin, and gentamicin for different periods of time. Quantitative real-time PCR indicated that the expression of the aac(2′)-If gene increased significantly when the cells were induced with the three aminoglycoside antibiotics for 2 h (P < 0.001). The expression level increased approximately three-, five-, and two-fold when treated with ribostamycin, neomycin, and gentamicin, respectively. No significant difference was observed between any pair of the other groups (Figure 4).

A total of 39 sequences (including one from this work) that were approximately 21 kb in length, with aac(2′)-If-like genes in the center, that shared nucleotide sequence similarities between 32.18 and 70.79% with aac(2′)-If were retrieved from the database. These sequences were all from Providencia (68.42%, 26/38) and Proteus strains (31.58%, 12/38). The genes from Providencia all shared amino acid sequence similarities of ≥54.61% with aac(2′)-If, while the genes from Proteus showed amino acid similarities of below 40.0% (except one of 68.76% and two of 53.39%) with aac(2′)-If.

The 39 sequences could be clustered into five groups according to the sequence similarities. The sequence of strain R33 was grouped alone, while the other four groups contained six or more sequences each (Supplementary Table S5). One sequence that shared the highest similarity with the other sequences in the same group was selected as a representative to perform the sequence structure comparison within the groups (Figure 5). The sequence of aac(2′)-If of R33 only showed similar sequence structures upstream of the aac(2′)-If gene as that of group 3, and the other sequences or sequence regions from these groups were completely different from each other.