Introduction

Macrolide antibiotics, one of the earliest classes of antibiotics clinically employed since 1952, currently rank third in global antibiotic sales with a market value of $4.8 billion1. They have a characteristic macrolide ring structure and inhibit protein synthesis by binding to the 50S ribosomal subunit, thereby inhibiting bacterial growth2,3,4. Macrolide antibiotics are classified based on their macrolide ring size into three types: 14-membered ring (erythromycin, clarithromycin, roxithromycin, telithromycin), 15-membered ring (azithromycin and tulathromycin), and 16-membered ring (tylosin, tilmicosin, josamycin, tildipirosin, and spiramycin)2,5,6,7. Macrolide antibiotics are particularly effective against Gram-positive bacteria and also demonstrate efficacy against several Gram-negative pathogens in the upper respiratory tract. Commonly used macrolide antibiotics, such as azithromycin, clarithromycin, and erythromycin, are employed to treat various infections, including pneumonia, sinusitis, pharyngitis, tonsillitis, and respiratory tract infections in animals8. However, the long-term and extensive use of macrolide antibiotics has contributed to the increased dissemination of antimicrobial resistance. According to the PROTEKT study, 36.8% (7420/20142) of Streptococcus pneumoniae clinical strains isolated from 40 countries between 2001 and 2004 demonstrated resistance to macrolides9. In China, macrolide resistance was observed in 68.7–100% of clinical isolates of Mycoplasma pneumoniae10.

Macrolide esterases (MEs) are enzymes that degrade the lactone rings in macrolide antibiotics, rendering them ineffective, and are key resistance genes contributing to macrolide resistance in bacteria. Currently, macrolide esterases are classified into two major classes: Ere-type and EstT-type. The Ere-type includes five members: EreA and EreB, found in clinical isolates of Escherichia coli; EreC in multidrug-resistant (MDR) Klebsiella pneumoniae; and EreD in the duck pathogen Riemerella anatipestifer11. In contrast, the recently discovered EstT-type belongs to the α/β-hydrolase family, with only one characterized member to date (EstT). EstT has been detected in MDR Sphingobacterium faecium from watering bowl environments, predominantly found in animal microbiomes and occasionally in human microbiomes4. The current understanding of the distribution and transmission of both Ere-type and EstT-type macrolide esterases is limited, with only a few identified members. This limited knowledge impedes the development of strategies to combat antibiotic resistance, posing potential risks for unexpected outbreaks of resistant bacteria and implications for public health.

The Ere-type and EstT-type macrolide esterases exhibit distinct substrate specificities. Ere-type enzymes target 14-, 15-, and 16-membered macrolides commonly used in human and veterinary medicine, such as erythromycin, clarithromycin, roxithromycin, azithromycin, and josamycin2,10,2c–f). The molecular weight of all four antibiotics increased by 18m/z after hydrolysis, suggesting the hydrolysis of the ester bond. However, for other macrolides antibiotics with 14-membered ring, 15-membered ring, and 16-membered ring macrolides, the molecular weight remained unchanged after treating with EstX (Supplementary Fig. 6).

EstX utilizes catalytic triad for ester bond hydrolysis

Currently, the catalytic mechanism of Est-type macrolide esterases has not been well investigated. To address this, we used alphafold2 to predict EstX structure, and a PLDTT score (97.3) indicated the reliability of the predicted structure for further study (Supplementary Fig. 7). The molecular docking results between EstX and 4 macrolide antibiotics showed that van der Waals forces, conventional hydrogen bond, carbon–hydrogen bonds, alkyl bonds, and pi-alkyl bonds were the main interaction type (Fig. 3a and b; Supplementary Fig. 8 and Supplementary Table 3). The amino acids that interact with macrolide antibiotics were Gly31, Ser34, Ser102, Ala169, Asn173, Val235, and Leu236, which are located within the catalytic pocket (Supplementary Table 4). Among them, S102 was hypothesized as the catalytic amino acid because it was conserved in multiple sequence alignment and sat within the binding pocket. The spatial distances between S102 and macrolide ester bond were 3.18 Å (tylosin), 3.67 Å (tilmicosin), 3.12 Å (tildipirosin), and 3.86 Å (leucomycin A5), respectively, which were deemed sufficient for S102 to attack ester bond and perform the function (Supplementary Fig. 9).

Fig. 3: Catalytic mechanism of EstX.
figure 3

a Molecular docking of tylosin with the predicted EstX structure. Molecular docking was performed using Libdock and then visualized using PyMOL software. b Two-dimensional diagram illustrating interactions between the predicted EstX structure and tylosin. Different bonds are represented by dashed lines of different colors. c Predicted EstX structure highlighting the catalytic triad composed of Ser102, Asp233, and His261. Hydrogen bonds are indicated by dashed lines, and numbers represent the lengths of the hydrogen bonds (Å). d Minimum inhibitory concentration (MIC) analysis of EstX mutations against 13 macrolide antibiotics. e Enzyme activity analysis of EstX mutations. Esterase activity was determined using p-Nitrophenol as substrate. Data are presented as the mean ± SD of independent experiments. n  =  3 biologically independent experiments. f Plate inhibition zone sizes of products derived from four macrolide antibiotics (tylosin, tildipirosin, tilmicosin, leucomycin A5) before and after hydrolysis by EstX mutations (S102A, D233A, and H261A). Data are presented as the mean ± SD of independent experiments. n  =  3 biologically independent experiments. g Proposed catalytic mechanism of EstX, which hydrolyzes the ester bond of tylosin, resulting in the ring-opening of the macrolide.

Full size image

EstX belongs to the alpha/beta hydrolase family, while the characteristic catalytic triad of this family has not been experimentally studied in macrolide esterases. In the predicted EstX structure, the catalytic triad was formed by Asp233-His261-Ser102, where Asp233 was linked to His261 by a 1.9 Å hydrogen bond, His261 was linked to Ser102 by a 2.3 Å hydrogen bond (Fig. 3c). To further confirm the catalytic triad, we constructed and purified mutations including S102A, D233A, and H261A, respectively (Supplementary Fig. 10). Firstly, The MICs of the E coli carrying S102A, D233A, and H261A were lower than that of the wild type EstX and showed no difference compared to E coli carrying the empty vector (Fig. 3d). The alteration in MIC values suggests that a mutation in any amino acid within the catalytic triad would result in the loss of protein activity. Secondly, enzyme activity assay using purified S102A, D233A, and H261A (with >90% purity) with p-NPB as a substrate showed that all three mutants lost activity (Fig. 3e and f). When a macrolide antibiotic was used as substrate and treated with purified S102A, D233A and H261A, the corresponding inhibition zone was similar to that of untreated macrolide antibiotics, indicating that mutants lost activities against macrolide antibiotics (Fig. 3f). Thus, the predicted protein structural information, point mutation MIC values, loss of activity, and inhibition zone results demonstrated that the catalytic triad in predicted EstX structure was crucial for its activity (Fig. 3g).

estX carrying gene cascade spreads through mobile elements

Investigating the gene families both upstream and downstream of estX can understand the transferability and co-occurring genes. Genes located upstream and downstream of estX belong to pfam families associated with drug resistance (e.g., MFS_1, AadA_C, Beta-lactamase, TetR_C_1, EamA, Multi_Drug_Res), mobile genetic elements (e.g., DDE_Tnp_1, DDE_Tnp_IS1, DDE_Tnp_IS240, DDE_Tnp_1_5, Y2_Tnp, DUF4158, Resolvase, DUF3330, Transposase_mut) and other functions (e.g., RVT_1, DUF3363, HAD, HTH_1, Pterin_bind and adh_short) (Fig. 4a). Heatmap cluster analysis divided the pfam families into two groups: a high-frequency group (55–100%) and a low-frequency group (5–49%) (Fig. 4b). Among the high-frequency proteins, the average distances between estX and genes from Phage_int_SAM_4, HAD, Phage_integrase, DUF3330, and AadA_C were 1, 1.01, 1.20, 2.31 and 2.82, respectively, indicating that they were likely to be adjacent to estX (Fig. 4c).

Fig. 4: Pfam family information of genes around EstX.
figure 4

a Heatmap analysis illustrating the distribution of Pfam families within the 20 genes upstream and downstream of the estX gene. Green indicates the presence of the Pfam family, while light yellow indicates its absence. b Percentage distribution of Pfam families within the 20 genes upstream and downstream of the estX gene. c Gene distance between genes from different Pfam families and estX. d Frequency diagram showing the occurrence of various Pfam families in the genomic regions surrounding the estX gene. Pfam families related to antibiotic resistance are highlighted in red, those related to integrases are marked in pink, and those associated with transposases/insertion sequences are marked in green.

Full size image

In the genomic region flanking the estX gene, a directional gene cascade has been observed, forming a coherent cluster with other protein-coding genes. This cluster includes the aadA gene from the AdaA_C family, conferring aminoglycoside resistance; a HAD family gene with unknown function; the cmlA1 gene from the MFS-1 family, causing chloramphenicol resistance; and the qacL gene from the Multi_Drug_res family, associated with multidrug resistance (Fig. 4d). Within this gene cluster, The HAD gene is positioned first, and cmlA1 occupies the third position. aadA genes appear in the 1st, 2nd, and 4th positions in the same orientation as estX, while qacL is found in the 2nd, 3rd, and 5th positions. This arrangement may be attributed to gene rearrangements within the estX-associated gene cluster.

The gene cascade surrounding estX is flanked on both sides by mobile elements associated with gene transfer, including type I integrases and transposases. Notably, a type I integrase is located opposite to estX in the first position, forming a complete integron. Additionally, genes associated with transposon, including transposase modulator proteins from the DUF3330 family, transposon resolvase from the Resolvase family, and transposase from the DUF4158 family, are present in the 2nd, 3rd, and 4th positions in the opposite orientation to estX. Other transposon-related families, such as Transposase DDE domain (DDE_Tnp_1, DDE_Tnp_IS1, DDE_Tnp_IS240, DDE_Tnp_1_5, Y2_Tnp, DUF4158, Resolvase, DUF3330, Transposase_mut), and Mutator transposable elements from the Transposase_mut family were also found in both sides of this gene cascade. Beyond integrases and transposases, 12.48% of estX distribution has been traced to plasmids, suggesting that, in addition to integrons and transposons, plasmids also play an important role in the dissemination of estX.

We next investigated the genomic context of estX across representative bacterial species, including Kluyvera intermedia, E.coli, Morganella morganii, Salmonella enterica, K. pneumoniae, Proteus mirabilis, Enterobacter hormaechei, Shigella flexneri, and Escherichia fergusonii. Our observations initially highlight interspecies horizontal gene transfer events, evidenced by the distribution of the gene cascade (EstX-HAD-AdaA-CmlA-AdaA-QacL) among various species such as K. intermedia, Escherichia fergusonii, E.coli, M. morganii, S.enterica, K.pneumoniae, P. mirabilis, and E. hormaechei (Fig. 5). This gene cluster is not only widely distributed but also exhibits a high degree of structural consistency and sequence homology. Furthermore, our analysis revealed gene dissemination between bacteria from different environmental origins. For instance, the gene clusters found in human-derived bacteria mirror those found in bacteria from animal and environmental sources, suggesting widespread transmission across different bacterial communities. Lastly, we noted instances of gene rearrangement and loss within the gene cascade containing estX during its transmission. A notable example includes the loss of the proteins cmlA1 and adaA1 in Shigella flexneri. These findings provide insight into the dynamic nature of gene transfer and the evolutionary pressures that shape bacterial pan-genomes.

Fig. 5: Genomic context linkage analysis of estX-carrying loci from representative bacterial species.
figure 5

Antibiotic resistance genes (ARGs), transposases, insertion sequences, and integrases are colored in different colors, while proteins with other functions are colored in gray. If two sequences from two loci share sequence similarity, they are linked with shadows that change color according to sequence similarity. To avoid introducing false positives, only genes encoding proteins with 100% sequence similarity to EstX were considered.

Full size image

estX distribution across various pathogens worldwide

A comprehensive survey identified 2259 bacterial strains from 74 countries harboring the estX gene. It is noteworthy that potential biases in the data could arise due to their reliance on isolates obtained from clinics or environmental samples, potentially impacting bacterial species, ecological niches, and geographic distribution. China exhibited the highest prevalence of estX, with 1026 strains (45.42%), followed by the USA with 248 strains (10.98%), and subsequent contributions from Australia (n = 68, 3.01%), France (n = 51, 2.26%), and Lebanon (n = 49 2.17%). Among the estX-bearing strains, the majority belonged to pathogenic or opportunistic pathogens, with E. coli (1269 strains, 56.18%), K. pneumoniae (477 strains, 21.12%), and S. enterica (477 strains, 21.12%) collectively constituting over 91.14% of the total strains analyzed (Fig. 6a). The presence of estX in clinical isolates underscores its potential association with pathogenicity in both humans and animals.

Fig. 6: Geographical, ecological, and host distribution of estX.
figure 6

a World map showing the global distribution of estX-carrying bacteria. Countries with estX-carrying bacteria are colored green, while those without estX-carrying bacteria are colored gray. Countries with estX-carrying bacteria are colored green, while those without are colored gray. Each country is overlaid with a pie chart, where the size of each pie represents the number of estX-carrying bacteria, and the colors represent different estX-carrying bacteria species. b Sankey diagram depicting the distribution across continents and ecological niches of estX-carrying bacteria from various species. To minimize false positives, only genes encoding proteins with 100% sequence similarity to EstX were included.

Full size image

A notable variability in bacterial distribution was observed at both the country and continental levels. For instance, 5 countries (China, Australia, France, Lebanon, and India) exhibited dominance of E. coli, while 2 were predominated by K. pneumoniae, and 2 by S. enterica. Interestingly, each continent showed a distinct prevalence pattern: Asia led with E. coli (60.30%), North America with K.pneumoniae (63.37%), and Africa with E. coli (76.04%). These patterns highlight the distinct ecological niches of estX-bearing bacteria across different regions (Fig. 6b).

The ecological diversity of estX is evident from its presence in various environments, including host-associated environments (2015 strains, 89.20%), the general environment (293 strains, 12.97%), and food sources (63 strains, 2.79%). Within host-associated environments, human-related strains predominated, but a substantial presence was also noted in animals such as pigs (n = 241), chickens (n = 414), ducks (n = 48), cows (n = 18), dogs (n = 16) and sheep (n = 15) (Fig. 6b). These findings emphasize the role of host-associated environments as primary reservoirs for estX. Moreover, estX ‘s presence in diverse environments, ranging from livestock farms to sewage treatment plants and rivers, and intensive care unit wards, suggests its potential as a reservoir for drug-resistant genes. Additionally, the detection of estX in food products like chicken and beef surfaces further complicates the public health implications of its spread.

Emergence timeline of estX-carrying bacteria

The emergence timeline indicated that bacteria carrying the estX gene were first found in 1979, ~27 years after the use of macrolides (1952). We observed a gradual increase in the detection of estX-carrying bacteria from 1979 through 2023, with a notably sharp rise from 2006 onwards. This escalation can be attributed to advancements in sequencing technologies facilitating the discovery of gene carriers. The timeline of estX discovery began with its detection in E. coli in 1979. Subsequent discoveries were in S. enterica in 1990, K.pneumoniae in 2002, and several other species, including Glaesserella parasuis in 2008 and E. hormaechei in 2009. More recent identifications include P. mirabilis in 2014, M. morganii and Klebsiella quasipneumoniae in 2016, E. fergusonii and Klebsiella oxytoca in 2017, Shigella flexneri in 2018, and another strain of K. oxytoca in 2019 (Fig. 7). Notably, the majority of bacteria hosting the estX are associated with human and animal diseases, underscoring the gene’s association with pathogenicity.

Fig. 7: Timeline of the emergence of estX-carrying bacteria from 1979 to 2023.
figure 7

Green arrows indicate the first appearance of estX-carrying bacteria on each continent. Black arrows indicate the first discovery of estX in each bacterial species. The human or animal icon on each bar represents the first discovery of estX-carrying bacteria within that host. To reduce false positives, our analysis focused exclusively on bacteria-carrying genes encoding proteins with 100% sequence identity to EstX.

Full size image

The geographical discovery of estX-carrying bacteria aligns with the historical usage of tylosin. North America was the first reported continent in 1979, followed by Asia in 1984 and Europe in 1989. Oceania followed in 1999, South America in 2002, and finally Africa in 2013. Meanwhile, regarding the hosts, the estX gene was initially found in pigs in 1979, then in humans in 1989, in chickens in 2002, and subsequently in ducks and cattle in 2007. Later discoveries included birds in 2012, dogs in 2013, and sheep in 2015. It is imperative to highlight that the hosts of estX-carrying bacteria encompass not only livestock and poultry but also pets and wildlife.

estX poses a threat not only to existing natural macrolide antibiotics but also to semisynthetic antibiotics in the development stage that have not yet been utilized. The discovery of natural macrolides, such as tylosin and leucomycin A5, dates back to the 1950s. Tylosin has been predominantly utilized in veterinary medicine and as an agent for growth promotion, whereas leucomycin A5 has seen significant use in clinical settings. Approximately two decades following their introduction, in 1979, estX was identified in E. coli strains found in pigs. This timeline suggested a lag in the emergence of resistance following antibiotic use.

Discussion

The global dissemination of estX is likely facilitated by various mobile genetic elements, including integrons, transposons, and plasmids. Class I integron can acquire estX and other ARG genes (including aadA, cmlA1, and qacL) to form a tightly linked gene cassette which is coregulated and coexpressed to respond more rapidly to antibiotic stress27. This kind of gene cassette structure enables its prevalence and persistence in the environment, even in sub-inhibitory concentration27,28. Class I integron do not encode enzymes enable horizontal gene transfer between bacteria, it often present in plasmid or transposons, which contributes to horizontal gene transfer between same or different bacterial species29. Gene transfer using different mobile elements and collaborations between mobile genetic elements (integron and transposon, integron and plasmid, integron plasmid and transposon) can accelerate the transfer between bacteria and, therefore, enabling estX’s wide distribution globally30. Notably, this gene cassette formed by a Class I integron, observed in the context of estX, differs from the genomic context of estT, which primarily propagates through transposases and plasmids. Therefore, the rapid dissemination of Est-type macrolide esterases can be attributed to their transmission via a variety of genetic elements, including integrons, transposons, and plasmids.

Est-type macrolide esterase, a class of resistance genes discovered in 2023, is of significance. Historically, the most well-known macrolide esterases have been of the Ere-type, known for their ability to degrade 14–15-membered ring macrolide antibiotics. In contrast, the Est-type macrolide esterase comprises only two members, EstT and EstX, with 109 and 2259 identical proteins, respectively. The enzymatic properties of EstX, optimal at 40 °C and pH 7.0, suggest its ecological origins are predominantly associated with hosts. This is further supported by its presence in host-associated pathogenic bacteria across 74 countries spanning six continents. The extensive distribution and spread of estX over the past 50 years can likely be linked to the widespread use of veterinary antibiotics like tylosin, commonly used as growth promoters, and the clinical use of the drug leucomycin A5, which has exerted selective pressure. Interestingly, the two antibiotics that EstX is capable of degrading were introduced and utilized after EstX became prevalent. This sequence of events underscores the threat EstX poses not just to existing macrolide antibiotics, but also to the efficacy of newer macrolides still in the developmental stages. Considering single-point mutations can change the substrate specificity of the antibiotic-resistant gene, the potential risks Est-type genes pose, necessitating further in-depth research and careful evaluation.

This study uncovers EstX’s ability to hydrolyze 16-membered ring macrolides, encompassing human-use antibiotics, with site-directed mutagenesis revealing a typical catalytic triad mechanism in the alpha/beta hydrolase superfamily. The research also indicates the potential for EstX’s global spread of bacterial pathogens via various mobile genetic elements. Considering the presence of other EstX-like macrolide esterases in nature, it becomes imperative to conduct future research and surveillance focusing on novel EstX-type esterases. This is especially critical for those variants capable of degrading human-use macrolides, as they pose a threat to the effectiveness of clinical macrolide antibiotics. A comprehensive understanding and monitoring of these enzymes are essential in laying the groundwork for the prevention and control of macrolide antibiotic resistance.