Identification, characterization, and distribution of novel amidase gene aphA in sphingomonads conferring resistance to amphenicol antibiotics

ABSTRACT

Amphenicol antibiotics, such as chloramphenicol (CHL), thiamphenicol (TAP), and florfenicol (Ff), are high-risk emerging pollutants. Their extensive usage in aquaculture, livestock, and poultry farming has led to an increase in bacterial antibiotic resistance and facilitated the spread of resistance genes. Yet, limited research has been conducted on the co-resistance of CHL, TAP, and Ff. Herein, a novel amidase AphA was identified from a pure cultured strain that can concurrently mediate the hydrolytic inactivation of CHL, TAP, and Ff, yielding products p-nitrophenylserinol, thiamphenicol amine (TAP-amine), and florfenicol amine (Ff-amine), respectively. The antibacterial activity of these antibiotic hydrolysates exhibited a significant reduction or complete loss in comparison to the parent compounds. Notably, AphA shared less than 26% amino acid sequence identity with previously reported enzymes and exhibited high conservation within the sphingomonad species. Through enzymatic kinetic analysis, the AphA exhibited markedly superior affinity and catalytic activity toward Ff in comparison to CHL and TAP. Site-directed mutagenesis analysis revealed the indispensability of catalytic triad residues, particularly serine 153 and histidine 277, in forming crucial hydrogen bonds essential for AphA’s hydrolytic activity. Comparative genomic analysis showed that aphA genes in some species are closely adjacent to various transposable elements, indicating that there is a high potential risk of horizontal gene transfer (HGT). This study established a hydrolysis resistance mechanism of amphenicol antibiotics in sphingomonads, which offers theoretical guidance and a novel marker gene for assessing the prevalent risk of amphenicol antibiotics in the environment.

IMPORTANCE

Amphenicol antibiotics are pervasive emerging contaminants that present a substantial threat to ecological systems. Few studies have elucidated resistance genes or mechanisms that can act on CHL, TAP, and Ff simultaneously. The results of this study fill this knowledge gap and identify a novel amidase AphA from the bacterium Sphingobium yanoikuyae B1, which mediates three typical amphenicol antibiotic inactivation, and the molecular mechanism is elucidated. The diverse types of transposable elements were identified in the flanking regions of the aphA gene, indicating the risk of horizontal transfer of this antibiotic resistance genes (ARG). These findings offer new insights into the bacterial resistance to amphenicol antibiotics. The gene reported herein can be utilized as a novel genetic diagnostic marker for monitoring the environmental fate of amphenicol antibiotics, thereby enriching risk assessment efforts within the context of antibiotic resistance.

INTRODUCTION

Amphenicol antibiotics, mainly chloramphenicol (CHL), thiamphenicol (TAP), and florfenicol (Ff) (Fig. 1A), are commonly used in aquaculture, livestock, and poultry farming worldwide (1–3). These antibiotics are capable of specific and strong inhibition of prokaryotic protein biosynthesis (4–6). The application of CHL in food-producing animals has been banned by the European Union due to the toxicity of CHL residues, although it continues to be used in developing countries globally owing to its low production costs and widespread availability (7, 8). TAP and Ff, derived from CHL, have gained popularity in poultry and aquaculture farms, as they are more stable and potent than CHL (7). Most of the CHL, TAP, and Ff consumed are excreted in the form of the parent compound or their metabolites, which are frequently detected in aquatic environments (e.g., mineral water, drinking water, and coastal water) (9–13), sediments (14), and even vegetable crops (15, 16). For example, different amphenicol antibiotics have been detected at concentrations ranging from ng L⁻¹ to μg L⁻¹ in different aquatic environments and regions (11, 17). Additionally, soils may be polluted by the use of antibiotic-contaminated animal manure fertilizer and wastewater irrigation (18, 19). The sustainable exposure to antibiotics has been shown to cause adverse environmental effects and serious implications for human health (20, 21).

Fig 1.

Resistance mechanism mining of Sphingobium yanoikuyae B1 to the amphenicol antibiotic chloramphenicol. (A) Chemical structures of CHL and its derivative with different structures highlighted: substitution of the p-nitro group with a p-sulfomethyl group (blue) and/or substitution of a C₃-hydroxyl group with fluorine (brown). (B) The transformation of CHL by strain B1 and its supernatant. (C) The conversion of CHL by transformant cells chlO_B1-pMD/Escherichia coli DH5α. (D) The transformation of CHL-palmitate by strain B1. (E) The proposed CHL and CHL-palmitate catabolic pathway in B1. The thick red arrow represents the main reaction catalyzed by CmO or ChlO_B1, and the thin pink arrow represents the minor reaction. The blue arrow represents the amide bond hydrolysis reaction of CHL-palmitate. The symbol × represents the inability of ChlO_B1 to catalyze CHL-palmitate. The bars are presented as the mean ± standard deviations (n = 3). Different significance levels between control and treatment are marked with asterisks (**P < 0.01 and ***P < 0.001, two-sided Student’s t-test).

To date, several amphenicol antibiotic-resistant mechanisms have been reported, including efflux systems, permeability barrier formation, modification inactivation, and antibiotic degradation. Non-enzymatic bacterial resistance mechanisms to amphenicol antibiotics involving efflux pump mediation by transporters have been reported in Escherichia coli and Pseudomonas aeruginosa (4). Resistance mechanisms based on membrane permeability alteration have been observed in Haemophilus influenzae (22), Pseudomonas cepacia (23), Salmonella typhi, and Acinetobacter baumannii (24, 25). The widely reported mechanism of modification inactivation involves the acetylation and O-phosphorylation of hydroxyl groups at the C₃ position by different types of acetyltransferases (4, 26, 27). The degradative inactivation mechanisms catalyzed by specific enzymes are also highly prevalent, including nitro group reduction, hydroxyl group oxidation, and hydrolytic cleavage of the amide bond. The nitroreductase NfsB from Haemophilus influenzae has been shown to confer resistance via the nitro group reduction of CHL to amino-CHL (28). The oxidoreductase CapO from the Sphingomonas sp. CL5.1 and oxidase CmO from Sphingobium sp. CAP-1 have been found to be capable of oxidating the C₃-hydroxyl group of CHL and TAP (29, 30). Furthermore, the oxidoreductase ChlOR can directly convert CHL and TAP to 4-nitrobenzaldehyde and 4-methylsulfonyl benzaldehyde, respectively (31). An α/β-hydrolase gene estT from the multidrug-resistant bacteria Sphingobacterium faecium WB1 and S. faecium WB8 could endow E. coli antimicrobial resistance, including florfenicol and macrolide (32, 33). The hydrolase EstDL136 isolated from the soil metagenome is responsible for hydrolyzing the amide bond of CHL to p-nitrophenylserinol (34). However, the degradative inactivation resistance genes that have been discovered to date only act on CHL or TAP, and it remains unclear whether the resistance gene-coded enzymes can simultaneously transform CHL, TAP, and Ff. The widespread and long-term application of these three antibiotics has resulted in their widespread co-occurrence in different environments, such as aquatic ecosystems (35, 36), swine manure (37), and even human urine (38). The synergistic toxic effects of co-occurring antibiotics on both target and non-target organisms in the aquatic environment are not to be ignored even at low concentrations and effect levels (39–41). Therefore, suitable diagnostic indicator genes are urgently required, to monitor amphenicol antibiotics and evaluate the level of resistance risk they pose in the environment.

In this study, we preliminary found that the S. yanoikuyae strain B1 has the capacity to degrade different amphenicol antibiotics. The resistant mechanisms of strain B1 to CHL involved multiple metabolic pathways. This study aimed to (i) investigate the resistance mechanism mediated by the gene aphA (one of the pathways) cloned from strain B1 toward CHL, TAP, and Ff, while also evaluating the antibacterial activity of the inactivated products; (ii) heterologously express gene aphA in E. coli and biochemically characterize; and (iii) analyze the genetic distribution of gene aphA conferring antibiotic resistance through genomic comparison. Determining the distinct resistance mechanism enhances the theoretical foundation for genetic diagnostic resources and supports the assessment of environmental risks associated with amphenicol antibiotics.

RESULTS

Sphingomonadaceae species are resistant to the amphenicol antibiotic chloramphenicol

Given the prevalence of CHL-resistant bacteria within the Sphingomonadaceae family (e.g., Sphingobium sp. CAP-1 [29], Sphingomonas sp. CL5.1 [30], and Sphingobium sp. WTD-1 [42]), in orderto explore the resistance characteristics of Sphingomonadaceae strains to amphenicol antibiotic, we selected multiple strains that belonged to this taxa from an in-house laboratory culture collection and assessed their degrading resistance to CHL. The strains included Sphingomonas wittichii RW1 (43), Sphingobium sp. YBL2 (44), Sphingomonas wittichii DC-6 (45), and Sphingomonas yanoikuyae B1 (46). Results revealed that the powerful xenobiotic degrading bacterium B1 and its culture supernatant could effectively convert CHL (Fig. 1B). Previous studies have identified strains in the Sphingomonadaceae family that contain CHL oxidase CmO (Sphingobium sp. CAP-1) or oxidoreductase CapO (Sphingomonas sp. CL5.1), which can catalyze oxidation at the C₃ position of CHL (29, 30). As expected, a CmO/CapO homolog gene chlO_B1 annotated as a GMC family oxidoreductase was found in B1 based on Blastp analysis. Functional verification of this gene revealed that E. coli cells harboring chlO_B1 could effectively transform CHL (Fig. 1C), which was consistent with the results of previously reported studies (29, 30). Strain B1 was found to also transform chloramphenicol palmitate (CHL-palmitate), with the C₃–OH position of CHL substituted by an alkane (Fig. 1A and D; Fig. S1A). The degradation product formed from CHL-palmitate amide bond hydrolysis was identified as D-(−)-threo-2-amine-1-(4-nitrophenyl)−1,3-propanediol palmitate (C₂₅H₄₂N₂O₅⁺, m/z 451.3168) by high-performance liquid chromatography-mass spectrometry (HPLC-MS/MS) (Fig. S1B and C), whereas the transformants containing the oxidase ChlO_B1 were not able to transform CHL-palmitate (Fig. S2), which may be due to the substituted group of CHL-palmitate being distinct from the action site of the oxidases ChlO_B1, CmO, and CapO (Fig. 1E). These results showed the potential presence of multiple metabolic pathways for CHL in strain B1. In addition to the direct oxidation pathway of CHL side chain, a novel resistance mechanism was involved in amide bond hydrolysis existed in strain B1.

Exploration of gene aphA conferring resistance to amphenicol antibiotics

To clone the gene conferring resistance to amphenicol antibiotics from strain B1, the brown hydrolysate p-aminophenol (p-AP) of p-acetamidophenol (p-AAP) mediated by hydrolase was selected as an indicator to screen a B1 genomic DNA library (Fig. S3A). A transformant cell, which produced a brown product on Luria–Bertani (LB) medium containing p-AAP, was selected as a positive transformant (Fig. S3B). HPLC analysis found that p-AAP (retention time 3.81 min) could be converted to p-AP (retention time 2.54 min) by this transformant (Fig. S3C). The results of sequencing showed that one open reading frame (ORF) in the cloned fragment exhibited diverse identities with several hydrolase or esterase genes. This ORF was subcloned to the vector pUC118 and transformed into E. coli DH5α, with the subclone found to have the ability to transform CHL and generate a product metabolite. Therefore, it was speculated that this ORF corresponded to the gene conferring resistance to amphenicol antibiotics, designated as aphA, which was responsible for encoding an amidase. A putative signal peptide was found at the N terminus of AphA between Ala20 and Glu21 (Fig. S4), indicating the potential secretion of this amidase as an extracellular protein.

Sequence analysis and biochemical characterization of the amidase AphA

The results of a Blastp search using the protein data bank protein database revealed that AphA shared low sequence identities with only three proteins. Among them, AphA exhibited 33.96% identity with esterase B (protein data bank ID: 4N5H_X) from Lacticaseibacillus rhamnosus Lc 705 and shared 33.33% identity with the bacterial heroin esterase (1LZL_A and 1LZK_A) from Rhodococcus species, which hydrolyzed the acetyl groups from heroin to yield morphine and from phenylacetate to yield phenol (47). Additionally, Blastp search utilizing the UniProtKB/Swiss-Prot database demonstrated that AphA shared less than 30% amino acid sequence identities with only four proteins, displaying the highest sequence identity (26%) to a functionally characterized esterase EstD from Thermotoga maritima (48). Phylogenetic analysis was performed, comparing AphA with other functionally characterized amide hydrolases from different superfamilies including Amidase signature Superfamily cl18951, Arginase_HDAC Superfamily cl17011, and Abhydrolase Superfamily cl21494. Results showed that amidase AphA was classified as a novel member of the α/β-hydrolase Superfamily, occupying an independent branch (Fig. 2A). Protein sequence alignment revealed the presence of the putative catalytic triad Ser–His–Glu (Ser153, His277, Glu249) in AphA (Fig. 2B), which is known to be highly conserved in the α/β-hydrolase Superfamily. AphA contained the typical esterase motif G-X-S-X-G (49).

Fig 2.

Multiple sequence alignment and phylogenetic analysis of AphA with the related proteins and effects of temperature, pH, and metal ions on the enzyme activity when CHL as substrate. (A) The phylogenetic tree was constructed by the neighbor-joining method based on the alignment of AphA with the other functionally characterized amide hydrolases from different superfamilies. (B) Multiple sequence alignment of AphA. Red arrows represent the putative catalytic triads Ser–His–Asp/Glu residues in α/β hydrolase family. The highly conserved G-X-S-X-G esterase motif is denoted by the red box. (C) The optimal temperature (black) and thermal stability (red) for the activities of AphA. For optimal temperature, maximal enzyme activity was set at 100%, and the relative activity was calculated as a fraction of this enzyme reaction. For thermal stability, enzyme activity after treatment at 4°C was defined as 100%, and the relative activity was determined as a fraction of this enzyme reaction. (D) The optimal pH for the activities of AphA. The maximal enzyme activity was set at 100%, and the relative activity was calculated as a fraction of this reaction. (E) Effects of metal ions on the enzyme activity of AphA. Without adding metal ions was set at 100%, and the relative activity was determined as a fraction of this reaction. Error bars represent the standard deviation of three replicates.

The aphA gene was successfully overexpressed in E. coli BL21(DE3) without the signal peptide and purified using affinity chromatography. The fusion enzyme His₁₂–AphA–His₁₂ was flanked N- and C-terminally by 12 histidine residues, which served as an affinity purification tag. The purified protein exhibited a single band, with a molecular mass of approximately 48 kDa (Fig. S5). The relative activity of AphA still retained 70% when it was stored at 50°C for 30 min; however, the enzyme was unstable, and its activity gradually decreased to 20% when preincubated exceeding 50°C (Fig. 2C). Optimal AphA catalytic activity occurred in vitro at 37°C and pH 7.0 without any other cofactors (Fig. 2C and D). The relative activity of AphA decreased to 6.44% ± 1.57% at pH 4.0 and 35.24% ± 2.04% at pH 10.0 following acidification and alkalization, respectively. The addition of Ni²⁺, Ca²⁺, Cu²⁺, Co²⁺, and Cd²⁺ significantly suppressed the relative activity of AphA to less than 50%, compared with the control group (Fig. 2E). Interestingly, 1 mM Mg²⁺, Mn²⁺, and Al³⁺ exhibited a promotive effect on the enzyme catalytic process, enhancing the catalytic activity by 8.35% ± 1.91%, 24.33% ± 5.62%, and 20.64% ± 2.80%, respectively. Fe³⁺ and Zn²⁺ showed no significant influence on the activity of AphA. In vitro, AphA could catalyze the hydrolysis of CHL with a Vmax of (7.30 ± 0.46) ×10⁻³ µM/s, Km of 300.27 ± 56.84 µM, kcat of (2.84 ± 0.17) ×10⁻⁴ s⁻¹ and kcat/Km of 10 × 10⁻¹⁰ mM⁻¹ s⁻¹. A similar Km and catalytic efficiency were observed for AphA and TAP (Km of 299.70 µM ± 74.21, kcat/Km of 4 × 10⁻¹⁰ mM⁻¹ s⁻¹, Vmax of (3.23 ± 0.27) ×10⁻³ µM/s, and kcat of (1.30 ± 0.11) ×10⁻⁴ s⁻¹). The hydrolysis activity of AphA toward Ff exhibited a higher affinity (Km of 148.50 ± 27.91 µM) and catalytic efficiency [kcat/Km of 2 × 10⁻⁹ mM⁻¹ s⁻¹, Vmax of (7.20 ± 0.33) ×10⁻³, kcat of (2.82 ± 0.13) ×10⁻⁴ s⁻¹] than the other two antibiotics (Fig. S6A through C).

AphA exhibits resistance to amphenicol antibiotics arising from amide bond hydrolysis

The enzyme assay showed that CHL was converted by AphA, generating metabolite compound I, with a retention time of 2.15 min in HPLC analysis (Fig. 3A). HPLC-MS/MS analysis showed that the molecular ion mass of compound I [m/z 213.0868 (M + H)⁺] was consistent with that of p-nitrophenylserinol standard (C₉H₁₂O₄N₂⁺) (Fig. S7A and B); therefore, compound I was identified as p-nitrophenylserinol (Fig. 3D and E). Similarly, the hydrolysis products were also determined when TAP was selected as the substrate (Fig. 3B). The molecular ion mass of the TAP metabolite (compound II) was m/z 246.0795 [M + H]⁺, which is consistent with the standard TAP-amine (C₁₀H₁₅NO₄S⁺) (Fig. S7C and D), and thus, compound II was identified as TAP-amine (Fig. 3F and G). Additionally, Ff could also be hydrolyzed by AphA, and the catalytic product formed (compound III) was detectable by HPLC (Fig. 3C). The molecular ion mass of compound III was m/z 248.0752 [M + H]⁺, the same as the Ff-amine standard (C₁₀H₁₄FNO₃S⁺) (Fig. S7E and F), and therefore, compound III was identified as Ff-amine (Fig. 3H and I). Collectively, these results indicated that by performing amide bond hydrolysis, purified amidase AphA could transform CHL to p-nitrophenylserinol, TAP to TAP-amine, and Ff to Ff-amine.

Fig 3.

The amidase AphA catalyzed CHL, TAP, and Ff to p-nitrophenylserinol, TAP-amine, and Ff-mine, respectively. HPLC analysis of CHL (RT = 5.67 min) (A), TAP (RT = 3.46 min) (B), and Ff (RT = 5.50 min) (C) catalyzed by purified AphA. MS/MS spectra analysis of p-nitrophenylserinol standard (D) and compound I (E), product of CHL hydrolysis by AphA; TAP-amine standard (F), and compound II (G), product of TAP hydrolysis by AphA; Ff-amine standard (H) and compound III (I), product of Ff hydrolysis by AphA.

To evaluate the antibacterial activity of the hydrolytic products of three amphenicol antibiotics, E. coli DH5α and Staphylococcus aureus ATCC 29213 were selected as target bacterial strains. Results showed that E. coli and S. aureus were not able to grow in the presence of 105.22 µM CHL, while normal growth occurred in the presence of an equimolar concentration of hydrolysate p-nitrophenylserinol (Fig. S8A). Furthermore, 280.72 µM TAP completely inhibited the growth of E. coli and S. aureus, although both strains exhibited substantial growth in the presence of equimolar concentrations of TAP-amine (Fig. S8B). Finally, 94.92 µM Ff significantly inhibited the growth of E. coli and S. aureus. However, despite the weak toxicity of Ff-amine to DH5α, cultures exposed to Ff-amine recovered and exhibited growth after a lag phase due to environmental adaptation (Fig. S8C). These results showed that the antibacterial activities of these amide bond hydrolytic products were completely mitigated or significantly decreased compared to the parent antibiotic. In contrast to the antibacterial effects observed with three antibiotics against E. coli and S. aureus, strain B1 exhibited degradative resistance to all three amphenicol antibiotics. Even in the presence of CHL, TAP, and Ff, strain B1 still showed slow growth, reaching OD₆₀₀ of 0.60 (CHL), 0.45 (TAP), and 0.57 (Ff) after 12 h of incubation. The hydrolyzed products of CHL and TAP lost their inhibitory effects on B1 growth. Notably, while the toxicity of hydrolyzed product of Ff significantly decreased, they still exerted slight effects on the growth of E. coli and S. aureus, yet it did not exert an influence on the growth of strain B1 (Fig. S9A through C).

Heterologous amidase expression endows resistance to amphenicol antibiotics

Due to the unsuccessful genetic manipulation of B1, the model strain E. coli BL21(DE3) was employed as a surrogate to elucidate the in vivo function of the identified gene against three distinct classes of amphenicol antibiotics. The conversion of the three antibiotics into hydrolysis products showed significantly distinct kinetics for BL21(DE3)/pET-aphA (Fig. S10). The transformation efficiency of Ff to Ff-amine reached 93.49% after 60 h of inoculation, while under the same inoculum, the transformants exhibited a slightly lower conversion rate, with only 46.64% of 70.18 µM TAP being converted to TAP-amine within 60 h and complete transformation of TAP occurring after 150 h. Understandably, the extracellular concentration of CHL decreased in the control BL21(DE3)/pET due to superior active uptake or adsorption by cells, which was not a conversion as no product was detected. In contrast, the degradation efficiency of CHL by BL21(DE3)/pET-aphA was observably raised compared to that of BL21(DE3)/pET, although it did not reach the efficiency of Ff as substrate (Fig. S10). The results suggested that the degradation of amidase AphA for Ff is superior to that of CHL and TAP, which is consistent with the Michaelis constants and catalytic activity of AphA toward the three antibiotics. We therefore would investigate the catalytic mechanism underlying the difference in conversion efficiency of AphA to the three amphenicol antibiotics.

Identification of key residues in AphA-mediated hydrolysis of amphenicol antibiotics

The structure models of AphA and three amphenicol antibiotics CHL, TAP, and Ff molecular docking were constructed to understand the catalytic mechanism of AphA at the molecular structure level. The theoretical binding affinity of the AphA-Ff complex (−4.79 kcal mol⁻¹) was superior to that of AphA-CHL (−4.07 kcal mol⁻¹) and AphA-TAP (−4.15 kcal mol⁻¹), which may account for the observed difference in catalytic activity. The docking results indicated that CHL might form hydrogen bonds with Gly78 and Cys79 residues at the O atom of C1-OH position, and with Ser153 at both the C1-OH position and the amide bond (Fig. S11A). Similarly, Ser153 might also form hydrogen bonds with TAP’s C1-OH and amide bond, respectively. In TAP, Gly78 and Cys79 residues were predicted to form hydrogen bonds at distances of 3.17 and 3.16 Å with the C1-OH (Fig. S11B). Closer hydrogen bonds were observed between Ff and AphA compared to those of CHL or TAP, such as 2.87-Å hydrogen bonds between Ser153 and Ff, compared to 2.89 Å for those of CHL and 3.31 Å for those of TAP; and 2.96 Å hydrogen bond at the C1-OH between Gly78 and Ff, compared to 3.08 Å for those of CHL and 3.17 Å for those of TAP (Fig. S11C). Notably, His277 formed a 2.87-Å hydrogen bond with the amide bond nitrogen in Ff, whereas it exhibited slightly weaker hydrophobic interaction forces with CHL or TAP.

To validate the key active sites for AphA catalysis of amphenicol antibiotics, the putative catalytic triad Ser–His–Glu that forms interaction forces with the substrate was selected for site-directed mutagenesis (Fig. S12). When His277 was mutated to alanine, AphA_H277A almost completely lost its hydrolytic activity toward all three amphenicol antibiotics. After the replacement of Ser153 by alanine, the specific activity of AphA decreased by 6.23-, 4.65-, and 3.10-fold compared to that of wild-type AphA, toward CHL, TAP, and Ff, respectively. The catalytic activity of AphA_E249A toward CHL and Ff was only marginally affected by the mutation of residue Glu249, but reduced to half toward TAP, compared to wild-type AphA (Fig. 4A through C). The H277A, S153A, and E249A variants with single-site mutations could partially disrupt the formation of hydrogen bonds and hydrophobic interaction, consequently impacting their activity. These findings indicated that hydrogen bonds and hydrophobic interactions might be crucial in the AphA-mediated catalytic activity of amphenicol antibiotics. However, to more accurately and deeply analyze the catalytic mechanism of AphA toward amphenicol antibiotics, the 3D structure of AphA and ligand interactions need to be deciphered by protein structure resolution methods.

Fig 4.

Mutational activity analysis of proposed catalytic triads of AphA using CHL (A), TAP (B), Ff (C) as a substance. The relative hydrolytic activity of the mutant proteins AphA_S153A, AphA_H277A, and AphA_E249A presented as a percentage of the activity of the wild-type AphA on three amphenicol antibiotics. Error bars represent the standard deviation of three replicates.

The circular dichroism (CD) analysis of protein indicated the similar spectral profiles between the near-UV (250–280 nm) and far-UV (190–260 nm) regions (Fig. S13). The wild-type protein and mutant variants exhibited similar secondary structure component distributions, including α-helix, β-sheets, β-turns, and random coil, suggesting no significant variations in structural characteristics (Table S1). These results indicated that mutations in individual amino acids do not significantly alter the overall AphA structure. The observed changes in AphA activity might thus be attributed to alterations in the corresponding amino acids.

AphA’s amino acid sequence information was compared with that of its closest structural homologs, for which the catalytic mechanism had been explained. The heroin-hydrolyzing esterase HerE (47) or the CHL-metabolizing enzyme EstDL136 (50) presented a prototypic catalytic triad Ser–Asp–His triad at the carboxyl end of strands β5, β7, and β8, respectively. The putative catalytic triad of AphA would also be situated at the end of the counterpart β-strands (Fig. S14). However, Glu249 was found in AphA instead of the acidic catalytic Asp of HerE and EstDL136 (47, 50). Alternative catalytic Ser–Glu–His often appears into the α/β Superfamily (51, 52), and according to our results, this could be the case for AphA. The mutations of residues S156A and H282A in EstDL136 led to the loss of CHL degradation capability (50), which was consistent with the findings of our study (Fig. 4). The residue S160 (HerE)/S156(EstDL136)/S153(AphA) was situated in the nucleophilic elbow, a region within the highly conserved GXSXG motif known as a signature sequence for the hydrolase family (47). Finally, alternative Glu residue was proposed as the acidic residue for other α/β Superfamily members decades ago (53, 54). Based on the moderate decrease in activity of the AphA Glu249A mutant, this residue might behave as the acidic residue of the catalytic triad. However, in the absence of experimental structural results, it is important to highlight that in a few cases, the catalytic acid (Asp/Glu) is located on a different loop (52).

Mobility analysis of aphA in sphingomonads

The Blastp search of the Genbank database showed that the gene aphA was distributed among different bacteria, such as S. yanoikuyae ATCC 51230, S. limneticum DSM 25076, Novosphingobium guangzhouense SA925, and Rhizobium species, being highly conserved in these strains and maintaining 100% of the amino acid sequence identities. The comparative analysis found that the gene aphA was distributed not only on the chromosome, such as with S. yanoikuyae SJTF8 and Sphingobium sp. YG1, but also on plasmids, such as with Sphingobium sp. WTD-1 (Fig. 5). Plasmid, as a mobile genetic element, is an important vehicle for the acquisition and dissemination of ARGs between species via HGT (55). In particular, conjugative plasmids will contribute greatly to the spread of ARGs in the new environment (56). Besides, a typical type I secretion system (T1SS) apparatus existed closely adjacent to the aphA gene, present in the genome of strain B1, including ABC transporters, a membrane fusion protein or adaptor, and an outer membrane protein (Table S2) (57). Likewise, T1SS was also found at the flank of gene aphA on the chromosome or plasmid in strains ATCC 51230, DSM 25076, SA925, SJTF8, YG1, and WTD-1, exhibiting a high degree of conserved similarity (>95%) (Fig. 5).

Fig 5.

Genetic organization of aphA in representative isolates. Dotted lines connect homologous genes, with the amino acid similarities displayed above the dotted lines. Genome accession numbers are enclosed in parentheses, and the ORFs are described in Table S2.

The gene transfer elements around aphA were also analyzed. Transposons were found closely adjacent to the gene aphA in the genomes of Sphingobium sp. ba1, Sphingomonas sp. ABOLH and Sphingobium sp. MP9-4 (Fig. 5). The gene aphA_ba1 (99.7% identity with aphA_B1) in strain ba1 was linked to an upstream insertion sequence (IS) IS1380, whose encoding protein showed 74.40% amino acid sequence identity with a transposase in ISSp1 element (AB021963) from S. paucimobilis UT26. In strain ABOLH, the gene aphA_ABOLH only harbors a part of the aphA_B1 sequence (504 bp) (Fig. S15), which may be a result of partial sequence loss during the transposition process. An insertion element IS110 existed downstream of aphA_ABOLH, which encoded a protein with 82% similarity to a transposase ISSph13 from Sphingopyxis spp. In strain MP9-4, insertion element-like sequences istA and istB were found downstream of aphA_MP9-4 (incomplete aphA_MP9-4, 822 bp) (Fig. 5; Fig. S15). UniProt/Swiss-Prot analysis showed that istA encoded an insertion sequence IS21 family protein, which exhibited 58% identity to the corresponding transposase (Q53201) from Sinorhizobium fredii NGR234. The istB sequence encoded a protein with 76% identity to an ATP-binding protein (P55617) from S. fredii NGR234. The analysis of IS elements revealed that nearly 50-bp inverted repeats existed at both ends of IS1380, IS110, and IS21 in strains ba1, ABOLH, and MP9-4, respectively (Fig. S16). These findings suggested the presence of transposable elements flanking the aphA gene, potentially indicating HGT of the gene aphA under the selective pressure of exposure to amphenicol antibiotics (Fig. 5).

DISCUSSION

In this study, the inactivation of amphenicol antibiotics by an amidase AphA-mediated mechanism was determined and studied in detail. In fact, it had already been reported that some hydrolase enzymes can mediate nonspecific cleavage of antibiotics. Organophosphate pesticide-degrading α/β hydrolase produced by pesticide-degrading bacterium Bacillus sp. MK-07 developed cross-resistance to CHL (58). The bile salt hydrolase LpBSH cloned from the intestinal bacterium Lactobacillus paragasseri JCM 5343^T catalyzed the hydrolysis of the amide bond in the β-lactam antibiotic penicillin G to produce phenylacetic acid (59). The esterase EstT completely converted 16-membered ring-containing macrolide antibiotics to products by ester bond hydrolysis and endowed with certain resistance to Ff (32, 33). Besides, a hydrolase EstDL136, identified from the soil metagenome, had also been found to mediate the hydrolysis of CHL and Ff, but the researchers failed to isolate a pure culture bacterium containing this enzyme (34). All of the above studies did not comprehensively investigate the degradation characteristics of three major amphenicol antibiotics, compare enzyme activities, or analyze amidase distribution. To date, aphA is the first cloned and functionally identified gene that mediates the hydrolysis inactivation of three amphenicol antibiotics simultaneously from a pure cultured strain.

The aphA-mediated hydrolytic inactivation has a higher probability of easily undergoing HGT compared with other amphenicol antibiotics' degradative resistance mechanisms. The main reasons for this are (i) it is a single gene with a complete function, unlike multicomponent oxygenase, whose function depends on the presence of all components (60); (ii) the expression of aphA is more accessible than some amphenicol antibiotic oxidative enzymes, such as ChlOR (31). The expression capacity of ARG in pathogens implies its potential health risks (61); (iii) it functions without cofactors; unlike NfsB, its function is dependent on the presence of both NADPH and FMN; CmO requires flavin adenine dinucleotide, while CapO necessitates 1-methoxy-5-methylphenazinium methylsulfate (28–30); (ⅳ) the most direct evidence is that transposons IS1380, IS110, and IS21 containing the gene aphA have been found in some bacterial species (Fig. 5). With the sustained release of amphenicol antibiotics into the environment, bacteria will inevitably evolve resistance to adapt to antibiotic pollution and may acquire co-occurring ARGs for a range of antibiotics (62, 63). Importantly, pathogenic bacteria are likely to be affected by the transfer of this gene conferring resistance to antibiotics.

This study also revealed a putative vital source of the gene aphA conferring resistance to amphenicol antibiotics that was likely to evolve from other amidases acting on non-antibiotic substrates, as AphA remains to retain certain ancestral features (G-X-S-X-G) also found in α/β-hydrolases (Fig. 2A) (49). The production and clinical use of antibiotics by humans have led to unnatural levels of these molecules in natural environments. Consequently, it is not surprising that the presence of antibiotics has enabled the resurgence of bacteria capable of metabolizing them (64). The excessive use and sustained release of synthetic antibiotics, typically related to maintaining a selection pressure across entire microbial communities, will facilitate the genetic capabilities of microbes to exploit resistance genes, using HGT to acquire resistance (65). The gradual evolution of genes with other functions to become ARGs is probably a one way microbes can achieve this. The resistance selection of particular environments, particularly pollution from antibiotic manufacturing facilities containing high levels of antibiotics, is undoubtedly a catalyst for increasing the spread of ARGs, resulting in a high risk to both human and environmental health (66).

To explore the environmental habitats of host strains harboring the gene aphA conferring resistance to amphenicol antibiotics, we conducted a Blastp search of the non-redundant protein database (top 100). Results showed widespread occurrence of the aphA gene within the Proteobacteria phyla, encompassing α-, β-, and γ-Proteobacteria. The aphA were not only predominantly found in aquatic environments, such as seawater surfaces, freshwater, and lake water, which are particularly susceptible to amphenicol antibiotic contamination, but also found in environments influenced by human activities, such as polluted streams and soils, leachate from landfill, sewage. and eutrophic lakes (Fig. 6). As a result of anthropogenic activities, the discharge of hazardous substances into aquatic environments exerts detrimental effects on both water quality and the ecological balance of aquatic ecosystems (61, 67). The aquatic environment serves as a prominent habitat for amphenicol antibiotic-resistant strains due to the extensive utilization of amphenicol antibiotics, particularly florfenicol, in aquaculture and livestock farming for the prevention and/or treatment of bacterial infections. In addition, the relatively high water solubility of amphenicol antibiotics makes them mainly distributed in aquatic environments (68). Compared with solid-phase media, this specific habitat is destined to breed resistant strains to amphenicol antibiotics more quickly due to the liquid-phase media’s ability to facilitate rapid and sufficient exposure of microorganisms to such antibiotics.

Fig 6.

Different habitats of strains containing the amidase AphA. The 32 representative proteobacterial strains with overlapping hits among the Blastp of AphA were selected for the construction of a phylogenetic tree. Multiple-alignment analysis of the amino acid sequences was conducted using Clustal W. The NJ phylogenetic tree of the above strains based on AphA protein sequences is shown on the left. The amino acid identities of AphA and its counterparts are shown by the heat map with the values presented. The homologous proteins marked in blue fonts originate from Metagenome-Assembled Genomes. The isolation sources of different strains containing the protein AphA are displayed on the right. Miss refers to the absence of strain isolation information.

A complete T1SS apparatus was found to be located on the flank of the gene aphA in some strains. The type I secretion system is one of the mechanisms used by bacteria to secrete various molecules to the external medium, including some proteins or peptides (Fig. S17) (69, 70). The ABC transporter was found to be an efficient secretory apparatus for the extracellular lipase TliA in Pseudomonas fluorescens (71). In the presence of an ABC transporter, a higher level of esterase EstA was found to be secreted from Pseudoalteromonas sp. 643A than in strains without transporters (72). The N-terminal signal peptide of AphA possibly binds to T1SS for efficient secretion of mature AphA into the extracellular milieu, thereby effectively reducing the environmental concentration of antibiotics. This mechanism enables non-antibiotic-resistant microorganisms to withstand selective pressure imposed by antibiotics in their habitat. However, the biological relationship between T1SS and amidase AphA still requires further research.

MATERIALS AND METHODS

Chemicals, bacterial strains, and culture conditions

CHL, TAP, and Ff (>98% purity) were purchased from Aladdin (Shanghai, China). Ff-amine, p-AAP, and CHL-palmitate (>97% purity) were purchased from Macklin (Shanghai, China). TAP-amine standard (97.3% purity) was acquired from Standard Pharm CHN. The strains and plasmids used in the study are listed in Table S3. The oligonucleotide primers are described in Table S4. LB medium was used to culture bacteria, which contained 10.0 g L^−­1 NaCl, 10.0 g L^−­1 tryptone, and 5.0 g L^−­1 yeast extract, pH 7.0. The mineral salt medium consisted of 1.0 g L^−­1 NaCl, 1.0 g L^−­1 NH₄Cl, 1.5 g L^−­1 K₂HPO₄, 0.5 g L^−­1 KH₂PO₄, and 0.2 g L^−­1 MgSO₄. 7H₂O, pH 7.5, was used to assess the biotransformation of antibiotics.

Cloning of gene conferring resistance to amphenicol antibiotics

The degradation of CHL and its derivatives by strain B1 was investigated using a resting cell approach. The strain B1 cells were cultured in LB medium with shaking until the exponential growth stage. The cell pellets were harvested and washed twice with mineral salt medium (MSM) by centrifugation at 7,000 × g for 10 min, and suspended with MSM. Additionally, the fermentation supernatant of B1 was centrifuged at 16,000 × g at 4°C for 30 min. The strain B1 and its supernatant were inoculated into MSM containing 37.14 µM of CHL and cultured at 30°C for 48 h. The function of the gene chlO_B1 in degrading CHL was confirmed through whole-cell transformation. The transformants chlO_B1-pMD/DH5α cells were also cultured in LB medium until the exponential growth stage to prepare seed cells. Cells were collected, suspended in MSM as described above, inoculated into MSM containing 37.14 µM of CHL, and incubated at 30°C for 72 h. E. coli DH5α without gene chlO_B1 was regarded as the control. The substances were extracted three times with ethyl acetate, dried under nitrogen, suspended in ethanol, and analyzed by HPLC.

The screening method used to identify the gene that conferred resistance to amphenicol antibiotic was conducted according to a previously reported method (73). Briefly, the genome DNA of B1 was digested by endonuclease Sau3AI. The fractions including 1- to 5-kb DNA fragments were collected and inserted into the vector pUC118 (TaKaRa). The recombinant plasmids were transformed into E. coli DH5α and spread on LB plates containing 100 mg L⁻¹ of ampicillin and 1.32 mM p-AAP. The colonies with a brown color, which indicated the hydrolysis of p-AAP into product p-AP by the target amidase, were purified and tested for hydrolytic activity with p-AAP via HPLC analysis. The transformants carrying candidate genes were sequenced by Beijing Tsingke Biotech Co., Ltd. ORFs were predicted using the ORF finder server. The nucleotide sequence and functional prediction of ORFs were analyzed using Blastn and Blastp tools. The SignalP 5.0 server (http://www.cbs.dtu.dk/services/SignalP/) was used for signal peptide prediction (74).

Protein expression and purification

The aphA removed signal peptide was amplified using the aphA-F/aphA-R primer pair (Table S4). The purified PCR products were ligated into the linearized expression vector pET32a(+) (BamHI/XhoI digestion) utilizing a ClonExpress II One Step Cloning Kit (Vazyme Biotech Co., Ltd, China) and transformed into E.coli BL21(DE3) cells. The BL21(DE3)/pET-aphA cells were inoculated in an LB medium containing 100 mg L⁻¹ of ampicillin and cultured until log phase growth. Then, 0.2 mM isopropyl-β-D-thiogalactopyranoside was added for the induction of recombinant protein expression at 16°C for 20 h. The cells were harvested, washed, and suspended in phosphate-buffered saline (PBS) buffer (20 mM, pH 7.4). BL21(DE3) cells harboring pET-aphA were sonicated and centrifuged at 12,000 × g, 4°C for 30 min to obtain crude extract. The purification of the AphA was performed using Ni-NTA resin affinity chromatography, and the imidazole was then removed through dialysis in PBS buffer overnight. The purified AphA was analyzed using 12% SDS-PAGE, and protein concentration was determined utilizing the Bradford method with BSA as the standard (75).

Enzymatic properties and kinetic measurements

The optimal temperature, pH, and thermostability of AphA, along with the impact of metal ions on enzyme activity, were investigated using a standard enzyme reaction. The standard enzyme reactions for AphA were performed at 35℃ for 2 h in 1 mL of 20 mM PBS buffer (pH 7.4) containing 77.37 µM CHL and 25.65 µM AphA. The AphA activity was investigated at different temperatures (4°C to 70°C) to determine the optimal reaction temperature. The optimal reaction pH was evaluated via the following different buffering systems: 20 mM citric acid–sodium citrate buffer (pH 4.0 to 6.0), 20 mM PBS buffer (pH 6.0 to 8.0), 20 mM Tris HCl buffer (pH 7.0 to 9.0), and 20 mM glycine–NaOH buffer (pH 8.0 to 10.0). The thermostability of AphA was determined by preincubating the enzyme in a water bath between 4°C and 70°C for 30 min and then assaying the remaining activity at the optimal reaction temperature. Similarly, the impact of metal ions on AphA activity was evaluated under standard conditions containing different metal ions (Mg²⁺, Ni²⁺, Ca²⁺, Cu²⁺, Mn²⁺, Co²⁺, Cd²⁺, Fe³⁺, Al³⁺, Zn²⁺) with final concentrations of 1 mM. The detection of specific activity was initiated by adding purified AphA into 1 mL of the reaction mixture containing 77.37 µM CHL, 70.18 µM TAP, and 69.79 µM Ff, respectively. The enzyme assay conditions were as described above. The reactions were boiled at 100°C for 10 min to terminate enzyme reactions. The residual concentrations of different substrates were analyzed by HPLC (Dionex UltiMate 3000, USA). The same enzyme reaction was performed with boiled AphA as a control. The definition of relative enzyme activity was calculated as a fraction of the standard enzyme reactions. Specific activity was defined as the amount of substrate transformation catalyzed per milligram of protein per hour. Three replicates were used for each treatment.

The kinetic parameters for CHL, TAP, and Ff were performed using various concentrations of substrate. To ensure that substrate consumption was within the linear range during the reaction, suitable reaction times or an appropriate concentration of purified enzyme AphA was utilized in the reaction system. The enzyme kinetic parameters for CHL, TAP, and Ff were determined based on nonlinear regression fitted to the Michaelis–Menten equation according to the method of Zhang et al. (76) using GraphPad Prism (version. 8.3.0).

Identification of the hydrolysates

To identify the reaction products of CHL, TAP, and Ff catalyzed by purified AphA, enzyme assay samples were inactivated and centrifuged at 12,000 × g for 10 min. The metabolites present in supernatants were separated and identified by HPLC equipped with a C₁₈ reverse-phase column (4.6 × 250 mm, 5 µm, Agilent Technologies, Palo Alto, CA, United States) at 40°C, with a UV detector at 272 nm (CHL), 235 nm (TAP), and 233 nm (Ff) (31). The mobile phases were water containing 0.1% (vol/vol) acetic acid and acetonitrile (60:40, vol/vol) at a flow rate of 1.0 mL min⁻¹. The mobile phase for HPLC analysis of CHL-palmitate was an 80% to 95% linear gradient of acetonitrile containing 0.01% trifluoroacetic acid for 30 min at a flow rate of 1.0 mL min⁻¹ equipped with a UV detector at 274 nm. Mass spectrometry analysis of the metabolic products was carried out with an AB SCIEX Triple TOF 5600 mass spectrometer. Metabolites were ionized by electric spray together with positive ion mode. Ionic fragments were detected through MS/MS.

Whole-cell biotransformation assays

The inoculant preparation was performed according to the method for protein expression described above. BL21(DE3) harboring with empty plasmid pET32a(+) was utilized as the negative control. The transformant BL21(DE3)/pET-aphA cells were harvested, washed with MSM three times to remove medium, incubated in MSM containing 77.37 µM of CHL, 70.18 µM of TAP, and 69.79 µM of Ff, respectively, and cultured at a rotation rate of 180 rpm at 30°C. Of the samples, 500 μL was taken, and the residual concentration of antibiotics was quantified using HPLC.

Antibacterial activity measurement

The antibacterial activity of the hydrolytic end-products of amphenicol antibiotics CHL, TAP, and Ff was detected according to previously reported methods (77). The Gram-negative bacterium E. coli DH5α, Gram-positive bacterium S. aureus ATCC 29213, and the parental S. yanoikuyae B1 were selected to evaluate the antibacterial activity of p-nitrophenylserinol, TAP-amine, and Ff-amine. Cells were collected during the exponential growth phase, washed twice with PBS buffer, and inoculated into threefold diluted LB medium with identical cell concentrations. Subsequently, 105.22 µM of CHL and an equivalent amount of p-nitrophenylserinol, 280.72 µM of TAP and an equivalent amount of TAP-amine, and 94.92 µM of Ff and an equimolar concentration of Ff-amine were added to the media, respectively. No substrate addition was served as the control. All treatments were cultured in incubators with shaking at 180 rpm, and samples were collected at 2h intervals for optical density measurements at 600 nm.

Homology model and molecular docking

Homology modeling of the amidase AphA was generated by the Modeller v9.19 program, based on the three-dimensional structure and sequence alignment of the template protein esterase (Protein Data Bank entry: 4PO3). The amidase model was structurally optimized using the Amber14 force field. The quality of the model was evaluated using the PROCHECK and Verify 3D programs (78). The molecular dockings between the amidase model and substances CHL, TAP, and Ff were performed using the AutoDock4.2.6 software package (79). The structure models were mapped by PyMOL.

Site-directed mutagenesis

Site-directed mutagenesis of AphA was performed based on the methods described by Li et al. (62) (62). Briefly, the recombinant plasmid pET-aphA was used as the template. The forward and reverse flanking primers aphA-F/aphA-R, along with the internal primer pairs aphA_S153A-F/R, aphA_E249A-F/R, and aphA_H277A-F/R, are listed in Table S4. All PCR products were inserted into vector pET-32a(+) to obtain the recombinant plasmids pET-aphA_S153A, pET-aphA_E249A, and pET-aphA_H277A, then transformed into competent cell E. coli BL21(DE3), respectively. The expression and purification of the recombinants BL21(DE3)/pET-aphA_S153A, BL21(DE3)/pET-aphA_E249A, and BL21(DE3)/pET-aphA_H277A were performed as described above. The degradation ability of AphA variants for CHL, TAP, and Ff were detected as described above.

Circular dichroism spectroscopy analysis

To assess the impact of site-directed mutations on protein conformation, CD (JASCO J-1500, Japan) analysis was conducted to evaluate the differences in the secondary structures of protein variants. CD measurements were performed over a wavelength range of 190–280 nm with a 1-nm bandwidth. The wild-type protein AphA and three variants, AphA_S153A, AphA_H277A, and AphA_E249A, were expressed, purified, and quantified using the above-described method. Protein samples were scanned using PBS buffer (20 mM, pH 7.4) as the baseline. Software packages CONTIN (80) was used to analyze secondary structures including α-helixs, β-sheets, β-turns, and random coils.

Bioinformatic analysis

The Blastp program was applied for the analysis of AphA using the non-redundant protein sequence, protein data bank, and Swiss-Prot databases. Multiple sequence alignment of the amino acid sequences of AphA and related proteins was performed using DNAMAN (v.9.0). Phylogenetic trees of AphA and some representative hydrolases were reconstructed using MEGA7 with the neighbor-joining method (1,000 bootstrap replicates).

The genomic DNA of strains B1 was extracted using the Wizard Genomic DNA Purification Kit (Promega, USA), following the manufacturer’s protocol. The genomes of strains B1 were sequenced by Illumina HiSeq sequencing platforms. The bacterial genome was annotated using the Prokka annotation tool (81) to predict genes. To understand the mobile genetic elements and virulence systems of resistant strain B1, transposon prediction was performed using the TransposonPSI software. IS prediction was conducted using ISEScan (version 1.7.2.1). Secretion system analysis employed the Diamond software (version 0.8.35) for alignment, with functional annotation carried out using the KEGG database. The genomic distribution characteristics of the aphA gene were analyzed based on genome information from various strains. The characteristics of transposons and ISs were analyzed using ISfinder for alignment (https://isfinder.biotoul.fr/about.php).

DATA AVAILABILITY

The draft genome sequence of S. yanoikuyae B1 was deposited in the GenBank database under accession number JAWJUJ000000000. The accession numbers of protein AphA and ChlO_B1 are MDV3482169.1 and MDV3478304.1, respectively.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aem.01512-24.

Supplemental material. aem.01512-24-s0001.doc.

Tables S1 to S4; Figures S1 to S17.

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.