1. Introduction
Probiotics are non-pathogenic living microorganisms that have positive effects on the host when administered in sufficient quantities [
1]. Recent research has revealed that specific strains, such as
Bacillus and
Lactobacillus, exhibit prophylactic or therapeutic efficacy against human gastrointestinal diseases and immune-related diseases [
2]. Furthermore, research suggests that probiotics improve the livestock immune function and reduce the incidence of infectious diseases in aquaculture and livestock [
3]. Given the increasing restrictions on antibiotic usage in livestock husbandry in many countries, probiotics have emerged as a significant antibiotic alternative, with considerable potential for use across the industry.
However, it is worth noting that not all probiotics meet the criteria for being classified as “generally recognized as safe” (GRAS) from a regulatory perspective in the market [
4]. Several studies have highlighted concerns associated with certain probiotic products, including mislabeling, limited strain viability, and contamination by various opportunistic pathogens [
5]. In particular, case reports have documented serious adverse events, such as bacteremia caused by
Lactobacillus and
Bacillus, in infants or critically ill patients [
6]. Contamination with opportunistic pathogens and fungi, including
Bacillus cereus and
Mucorales, in low-quality probiotic products, can lead to foodborne illnesses characterized by diarrhea, emesis, and in severe cases, mortality [
7,
8]. Moreover, it is important to recognize the potential risk of probiotics acting as a reservoir for antibiotic-resistance genes (ARGs) and potentially facilitating the transfer of ARGs to pathogenic or commensal bacteria within the gut microbiota [
9,
10]. For example, certain
Lactobacillus and
Bacillus strains have been shown to harbor ARGs that reside in various mobile genetic elements (MGEs) and confer resistance to aminoglycosides and macrolides [
11]. Further studies have confirmed the horizontal transfer of the tetracycline-resistance gene
tet(M) and the erythromycin-resistance gene
erm(B) among
Lactobacillus and
Enterococcus species both
in vitro and
in vivo [
12,
13].
Compared with probiotics intended for human consumption, probiotic supplements intended for livestock may pose an increased safety risk due to the lack of standardized strain-level safety assessment criteria. A previous study showed that certain probiotic products for livestock contain virulent probiotic organisms and contaminating pathogens that could colonize livestock and subsequently enter the environment [
14]. In this study, we investigated the composition and potential pathogen contamination of probiotic products for livestock available in the Chinese market. We also comprehensively analyzed the genetic profile and evolutionary characteristics of the probiotic strains. Through
in vitro and
in vivo tests, we found that the extensive use of
Enterococcus strains carrying ARGs as probiotics can disrupt intestinal homeostasis and promote the transfer of ARGs. Guided by the principles of “One Health” and “Food Safety,” this study aims to provide empirical evidence to support the standardized use and healthy development of probiotics in livestock.
2. Methods
2.1. Isolation and characterization of probiotics and opportunistic pathogens
To investigate the quality of probiotics used in livestock farming in China, we collected 95 non-repetitive commercially available probiotic products from 22 provinces, autonomous regions, and municipalities from 2021 to 2023. All collected products were within their stated validity period and designated for usage in livestock. Additional details regarding the commercial probiotic products are provided in Table S1 in Appendix A. Ten packs of powdered probiotics were thoroughly mixed for each product to create one sample. Next, 1 g of the sample was suspended in 5 mL of sterile 1× phosphate buffered saline (PBS) and homogenized for 10 min. Serial 10-fold dilutions were prepared using the same diluent, followed by plating 0.1 mL of the resulting dilution on the mannitol egg yolk polymyxin (MYP) agar (for Bacillus spp.), bile esculin agar (for Enterococcus spp.), and MRS agar (for Lactobacillus spp.). Plates from each dilution were incubated at 37 °C under aerobic or anaerobic conditions, and colony counts were conducted after 48 h. For Bacillus spp., the dilution was heated at 80 °C for 2 min to eliminate non-spore-producing microorganisms. In addition, RAPID’B.cereus medium and CHROMagar orientation medium were used to isolate and identify opportunistic pathogens in probiotic products.
Distinct colonies were morphologically identified and selected from agar plates. Genomic DNA was extracted from all pure cultured isolates using the HiPure Bacterial DNA Kit (Magen, China), and species determination was conducted via 16S ribosomal RNA (rRNA) gene sequencing, employing previously described primers [
15].
2.2. Antimicrobial susceptibility testing
Antimicrobial susceptibility testing of the probiotic isolates was performed using the broth microdilution method following the guidelines outlined in the Clinical & Laboratory Standards Institute (CLSI) document M45.
Streptococcus pneumoniae (
S. pneumoniae) ATCC 49619 and
Enterococcus faecalis (
E. faecalis) ATCC 29212 were used for routine quality control (QC). The panel of tested antimicrobials included clindamycin, erythromycin, chloramphenicol, meropenem, kanamycin, ampicillin, tetracycline, linezolid, vancomycin, ciprofloxacin, streptomycin, and gentamicin. The minimum inhibitory concentration (MIC) resistance breakpoints for
Bacillus spp.,
Enterococcus spp., and
Lactobacillus spp. were primarily sourced from CLSI documents M45 and M11. In instances where breakpoints for specific antimicrobials were not specified by CLSI, reference was made to International Standards Organization/International Dairy Federation (ISO/IDF)-10932, European Food Safety Authority (EFSA) guidelines [
16], and the European Committee on Antimicrobial Susceptibility Testing (EUCAST). The breakpoints used in this study are listed in Table S2 in Appendix A.
2.3. Whole genome sequencing and molecular analysis
Paired-end reads of 300-bp, with a minimum of 250-fold coverage for each strain, were generated using the Illumina HiSeq X-Ten System, followed by quality assessment using FastQC and trimming with Trimmomatic. The sequence reads were assembled using SPAdes (v3.12.0). In the case of the
Enterococcus strains AJ10XM1A and S11XM4A, which harbored the ARGs
fexA and
optrA in the probiotic products, complete genome sequences were obtained through sequencing with the MinION device. High-quality assembled whole-genome sequences were obtained using Unicycler (v0.4.8) by integrating short reads data after QC procedures. The assembled DNA sequences underwent automatic annotation via the RAST platform. Known alleles of antimicrobial resistance (AMR) and virulence genes were identified using a direct read mapping approach implemented in SRST2 against the ResFinder and VirulenceFinder databases, with a threshold > 90% identity [
17]. Multi-locus sequence typing (MLST) of all strains was conducted using PyMLST (v2.1) based on the pubMLST database.
2.4. Phylogenetic analysis
Phylogenetic trees for 129
Bacillus spp., 84
Enterococcus spp., and 18
Lactobacillus spp. were constructed using the unweighted pair-group method with arithmetic means (UPGMA) algorithm based on the min-hash distance, with a sketch size of 1000 and
k-mer size of 21 (Mash v2.1). They were then visualized using the interactive tree of life (iTOL). In addition, reference genomes for probiotic-associated sequencing data outside of this study were downloaded from the Bacterial and Viral Bioinformatics Resource Center (BV-BRC) [
18]. Detailed strain information is provided in Table S3 in Appendix A.
2.5. Conjugative assays
To assess the transferability of the
fexA and
optrA genes, filter mating conjugation experiments were performed using
fexA-positive
E. faecalis S11XM4A and
optrA-positive
E. faecium AJ10XM1A strains isolated from a probiotic product as donors, with rifampicin-resistant
E. faecalis JH2-2 as the recipient [
19]. The transconjugants were selected on a medium supplemented with 200 mg∙L
−1 rifampicin and 8 mg∙L
−1 florfenicol. Confirmation of the transconjugants was achieved through polymerase chain reaction (PCR) detection of the
fexA and
optrA genes. In addition, the
fexA- and
optrA-positive plasmids were electro-transformed into competent cells of
E. faecium strain CXM16A, which represented the major clonal type ST296 of
Enterococcus isolated from the probiotic product. Transformants were then selected using 8 mg∙L
−1 florfenicol, followed by PCR confirmation. The MICs of the transconjugants/transformants against florfenicol and linezolid were determined using the broth microdilution method according to the CLSI guideline. Notably,
E. faecalis ATCC 29212 was used for routine QC. The composition of the AMR genes in strains AJ10XM1A and CXM16A is detailed in Table S4 in Appendix A.
2.6. Study design of probiotic colonization in chicken intestinal gut
A total of 105 newly hatched Arbor Acres broilers (52 males and 53 females) were confirmed to be free of optrA-positive Enterococcus and optrA gene through direct fecal culture and PCR, respectively. The broilers were randomly divided into three groups and placed in separate sterile isolation chambers. The groups were set as follows: ① The control group was treated with PBS only (n = 35); ② an optrA-positive Enterococcus colonization group (n = 35) was made; and ③ an additional optrA-negative E. faecium CCW15A colonization group (n = 35) was included to investigate whether colonization with optrA-negative Enterococcus would affect the microbiota and antibiotic resistome. CCW15A lacked multiple AMR genes including optrA and was closely related to AJ10XM1A in phylogenetics (the AMR genes carried by CCW15A are shown in Table S4). All broilers were acclimated to the new environment for 3 d. Following the recommended daily dose of probiotics for broilers, which is approximately 1.0 × 107-1.0 × 108 colony-forming units (CFU)∙d−1 feed, each broiler received a direct injection of 500 μL of a PBS suspension containing 1.0 × 108 CFU∙mL−1 AJ10XM1A/CCW15A into the mouth. This inoculation was administered continuously for 3 d. We randomly selected five broilers from each group on days 1, 3, 7, 14, 21, 28, and 35 after colonization with strains. All broilers were euthanized humanely by cervical dislocation, and the contents of the ileum and cecum were promptly collected. A portion of the contents sample was utilized to isolate linezolid-resistant Enterococcus, while the remainder was immediately transferred to liquid nitrogen and subsequently stored at −80 °C until further analysis. All broilers were provided with free access to feed and water throughout the 35-day trial. Both feed and water were subjected to PCR-based optrA gene negative testing and sterilization treatment. The animal study received approval in adherence to the guidelines for the care and utilization of laboratory animals by the Jiangsu Academy of Agricultural Sciences (IACUC-AE-2023-08-010).
2.7. Metagenomic sequencing, assembly, and binning
Metagenomic DNA was extracted from the chicken ileum and cecum contents using DNeasy PowerSoil Pro Kit (47016; Qiagen, Germany) and sequenced using the MGI DNBSEQ-500 platform with a paired-end (PE) read length of 150 bp. Afterward, more than 10 Gb of raw reads underwent QC and host sequence filtering using Trimmomatic (v0.33) and BMTagger (v2.2.4) to eliminate adapter sequences, low-quality sequences, and contaminating host DNA [
20,
21]. The clean reads were then assembled using MEGAHIT (v1.1.2), after which contigs of less than 500 bp were filtered. To reconstruct microbial genomes from the metagenomic data [
22], MetaBAT2 was used to calculate the coverage of all contigs and perform metagenomic binning. This process was aimed at obtaining metagenome-assembled genomes (MAGs) that were taxonomically equivalent to microbial strains [
23]. The quality of the resulting bins was estimated using CheckM (v1.0.12) to assess genome integrity and contamination. High-quality MAGs are > 90% complete with less than 5% contamination, medium-quality MAGs with completeness estimates of ≥ 50% and less than 10% contamination, and low-quality MAGs are defined as < 50% complete with < 10% contamination. Finally, the taxonomic annotation for bins was inferred using GTDB-Tk (v2.2.0) with default parameters based on the Genome Taxonomy Database [
24].
2.8. Microbial, AMR, and MGE composition analysis
Open reading frames (ORFs) of the assembled contigs were predicted using Prodigal (v2.6.3) in meta mode [
25]. Taxonomic classification of the intestinal microbiota was performed using Kraken2, and Bracken was used to calculate microbial diversity and abundance [
26]. To comprehensively analyze the composition of ARGs and MGEs in the metagenomic data, reads or annotated ORFs were mapped to an integrated database comprising AMRfinder, CARD, and proMGE (v1.0), and to a customized plasmid database using HMMER or BLAST with default parameters [
26]. In addition, reads were aligned to their respective assemblies using bowtie2 to quantify the abundance of contigs, ARGs, and MGEs. This was followed by normalization using the transcripts per kilobase million (TPM) metric, calculated by SAMtools.
2.9. Statistical analysis and visualization
We conducted α-(chao1 index) and β-diversity (principal coordinate analysis (PCoA) with Bray Curtis distances) analyses of both the microbiome and resistome in the “vegan” R package using normalized relative abundance data. GraphPad Prism (v8.0) was employed to perform a differential statistical analysis of grouped data using the Kruskal-Wallis test, Mann-Whitney U test, or Student’s t-test. Spearman’s correlation analysis was utilized to identify correlations among ARGs, MGEs, and microbiota. Furthermore, a correlation network analysis was performed using R with the package Igraph. All statistical tests were considered significant at p < 0.05.
2.10. Data availability
Raw data of the metagenome sequences generated in this study are accessible through the National Center for Biotechnology Information (NCBI) Sequence Read Archive under the BioProject accession number PRJNA1049198. The sequences of the two novel resistance plasmids (pS11XM4A and pAJ10XM1A) are available through GenBank, with accession numbers OR734628 and OR734227, respectively.
3. Results
3.1. Quality issues in livestock-used probiotic products include strain composition, content, and purity
Herein, 95 non-repetitive livestock-used probiotic products were collected from 22 provinces, autonomous regions, and municipalities in China (Fig. S1 in Appendix A). Most probiotic products (> 79.3%) purportedly harbored at least two probiotic strains, primarily belonging to the genera
Bacillus spp.,
Lactobacillus spp., or
Enterococcus spp. However, only
Bacillus spp. and
Enterococcus spp. exhibited higher isolation rates and label conformity (number of strains-positive products/total number of products NSPP/TNP = 0.91 and 0.79, number of strains-positive products/number of labels-positive products NSPP/NLPP = 1.11 and 1.59, respectively). In contrast,
Lactobacillus spp. was absent in 83% of the sampled products, including those with label claims (NSPP/TNP = 0.17; NSPP/NLPP = 0.30). This finding suggests that
Bacillus spp. and
Enterococcus spp. are the predominant probiotic genera in the examined products, including those without label declarations (
Fig. 1(a)). Although most products purported viable cell counts exceeding the recognized effective threshold of 1.0 × 10
8 CFU∙g
−1 [
27], only 72.6% of
Bacillus spp.-related products met this criterion. The qualified rates of products (QRP) for
Enterococcus spp. and
Lactobacillus spp. were relatively lower, around 0.62 and 0.38, respectively (
Fig. 1(b)).
The 280 isolated non-redundant probiotic strains were further identified by means of 16S rRNA sequencing analysis.
Bacillus spp. was found to consist mainly of
B. subtilis (
n = 113),
B. velezensis (
n = 31),
B. paralicheniformis (
n = 12),
B. licheniformis (
n = 8),
B. inaquosorum (
n = 6), and
B. pumilus (
n = 5).
Enterococcus spp. included
E. faecium (
n = 78),
E. faecalis (
n = 4), and
E. hirae (
n = 2), whereas
Lactobacillus spp. predominantly consisted of
Lactiplantibacillus plantarum (
n = 8),
L. paracasei (
n = 3), and
L. pentosus (
n = 1) (
Fig. 1(b))
. In addition to probiotics, conditional pathogens such as
B. cereus (
n = 33),
E. cloacae (
n = 13),
C. freundii (
n = 3),
E. sakazakii (
n = 4),
K. pneumonia (
n = 5), and
S. enteritidis (
n = 15) were identified in 53.6% of the probiotic products (
n = 51). These results indicate that a significant proportion of commercially available probiotic products for livestock in China have significant quality problems in terms of probiotic composition, content and purity. Detailed information on the isolated strains can be found in Table S1.
3.2. Enterococcus probiotics exhibited more complex patterns of antibiotic resistance
The isolated probiotic strains underwent susceptibility testing against various antimicrobials, revealing that
Bacillus spp. exhibited a limited and species-specific pattern of antibiotic resistance. Specifically,
B. inaquosorum demonstrated significant resistance to chloramphenicol (
n = 4, 100%), whereas
B. velezensis (
n = 15),
B. subtilis (
n = 40), and
B. licheniformis (
n = 6) displayed relatively elevated resistance to ampicillin (12%-75%), tetracycline (25%-65%), clindamycin (20%-75%), and erythromycin (25%-70%).
B. paralicheniformis (
n = 6) displayed notable resistance to erythromycin (100%) and clindamycin (100%). On the other hand,
Lactobacillus spp. (
n = 6) demonstrated broad resistance to gentamicin, vancomycin, and ciprofloxacin, which is possibly attributable to intrinsic resistance mechanisms. Furthermore, we observed that both
E. faecium (
n = 35) and
E. faecalis (
n = 4) displayed innate resistance to clindamycin and aminoglycosides, along with acquired resistance to tetracycline (14% and 75%), ciprofloxacin (34% and 50%), and chloramphenicol (5% and 75%). It is worth noting that we also identified a linezolid-resistant
E. faecium and a florfenicol-resistant
E. faecalis strain. In general, the antibiotic resistance profiles of the
Bacillus and
Lactobacillus strains isolated as probiotics were relatively focused and mainly exhibited predictable intrinsic resistance. In contrast,
Enterococcus strains exhibited strain-specific multidrug resistance patterns in addition to their intrinsic resistance, indicating their potential for the horizontal acquisition of antibiotic resistance (
Figs. 2(a) and
(b)).
Moreover, given that B. cereus (n = 33) was the most prevalent opportunistic pathogen among the sampled products (n = 32, 34.7%), antimicrobial susceptibility testing was specifically conducted for B. cereus. The results revealed that all B. cereus strains were susceptible to gentamicin, ciprofloxacin, vancomycin, and meropenem while exhibiting extensive heterogeneous resistance to other antibiotics (Table S5 in Appendix A).
3.3. Enterococcus probiotics and opportunistic pathogens carried multiple resistance genes and virulence genes
We further analyzed the distribution of AMR genes in the sampled probiotics through whole-genome sequencing (
Figs. 2(b) and
(c)). Our findings revealed that the
Bacillus spp. predominantly harbored the
abc-f,
catA,
mphK,
blaOXA,
vmlR,
blaBUP, and
cat86 genes, which confer resistance to β-lactams, chloramphenicol, and macrolide antibiotics, respectively. Notably, we identified the
cfr-like gene (
clbA), encoding the 23S ribosomal RNA methyltransferase, in 75% of
B. velezensis and 11% of
B. paralicheniformis. However, no associated decrease in related antimicrobials (e.g., linezolid) susceptibility was observed. We did not detect any other antibiotic resistance genes (ARGs) in the
Lactobacillus spp. in addition to the
abc-f and
bla family genes. It is noteworthy that the
Enterococcus spp. harbored a more diverse array of ARGs. In addition to the common ARGs, such as
lsa(A),
lnu(B),
msr(C), and
aac(6′′)-I, we identified the presence of
tet(L),
lnu(B),
erm(B),
lsa(E),
optrA, and
fexA genes in
E. faecium and
E. faecalis strains. These genes mediate resistance to tetracyclines, lincosamides, macrolides, phenicols, streptogramins, and oxazolidinones, respectively.
As for the opportunistic pathogens isolated from the probiotic products, B. cereus was observed to predominantly harbor the resistance genes bla, satA, and fosB, which confer resistance to β-lactams, aminoglycosides, and fosfomycin, respectively. Moreover, E. cloacae, Cronobacter spp., K. pneumoniae, and Salmonella spp. were found to harbor the resistance genes oqxB, catA, tet(A), blaTEM-1, aph(3′)-II, and aph(6)-Id, which confer resistance to fluoroquinolones, chloramphenicol, tetracyclines, β-lactams, and aminoglycosides, respectively. Remarkably, the colistin resistance gene mcr-9.1 was additionally identified in one E. cloacae strain. The specific resistance genes carried by the strains are listed in Table S6 in Appendix A.
In response to concerns regarding widespread contamination and potential poisoning risk attributed to B. cereus in probiotic products, we investigated the distribution of virulence genes among B. cereus isolates. The nonhemolytic enterotoxin operon nheABC was detected in nearly all strains (97.1%, n = 33). The capsule synthesis genes capA, B, C, and D were identified in 32.6% (n = 11) of isolates. Interestingly, two sequence types (STs), ST24 and ST177, harbored additional virulence factors, notably hemolysin BL encoded by the hblACD operon and cytotoxin K encoded by cytK, which were absent in other STs. The carriage of virulence genes in B. cereus is detailed in Table S7 in Appendix A.
3.4. Phylogenetic analysis revealed the genetic diversity of probiotic strains
Next, we conducted a comprehensive phylogenetic analysis of the probiotic strains derived from the sampled products to compare their genetic relationships.
B. subtilis, the predominant probiotic species, belonged to the clonal lineage of ST1, ST47, and ST55.
B. velezensis,
B. licheniformis,
B. inaquosorum, and
B. paralicheniformis mainly belonged to ST96, ST1, ST13, ST4, and several novel ST types, respectively. Strains of
B. pumilus,
C. kochii, and
C. firmus were comparatively less abundant, and all were categorized into several novel ST types.
B. subtilis,
B. velezensis, and
B. paralicheniformis exhibited multiple evolutionary branches (
Fig. 3(a)). Moreover, all
E. faecalis strains were classified under ST16, whereas the
E. faecium strains displayed a heterogeneous clonal lineage, dominated by ST160, ST269, ST361, ST812, and ST1669, with the ST296 strains being further subdivided into three distinct clades (
Fig. 3(b)). The
Lactobacillus spp. and other lactic acid bacteria (LAB) strains lacked ST typing data, yet strains within the same genus exhibited close phylogenetic relationships (
Fig. 3(c)). None of the clonal lineages of the probiotic species showed typical geographical clustering, suggesting that the predominant constituent strains in the probiotic products had disseminated across various provinces, autonomous regions, and municipalities in China and potentially originate from diverse ancestors. On a global scale, the isolates from the probiotic products in China, with the exception of
B. licheniformis, exhibited close phylogenetic relationships with probiotic strains from other countries. These strains were isolated from diverse origins, including fermented foods, dietary supplements, and environmental samples, such as soil, water, and feces (Fig. S2 in Appendix A). Notably,
E. faecium ST262 and ST396, as well as
E. faecalis ST16 isolated from the sampled products, displayed close genetic proximity to
Enterococcus strains that are known to be pathogenic, causing wound infections, bloodstream infections, and other infectious diseases in humans (Fig. S3 in Appendix A).
We also performed a phylogenetic analysis of the opportunistic pathogen B. cereus. Our findings unveiled various ST types, including ST177, ST24, ST26, ST45, and ST185. These strains were identified in various probiotic products from different provinces, autonomous regions, and municipalities in China, with ST185 being the most prevalent clonal lineage (n = 11, 33.3%) (Fig. S4 in Appendix A).
3.5. Enterococcus isolated from probiotic products harbored transferable optrA- and fexA-positive plasmids
This study identified several acquired AMR genes in the
Enterococcus spp. isolated from the sampled probiotic products, including the oxazolidinone resistance gene
optrA and the chloramphenicol resistance gene
fexA. In
E. faecium AJ10XM1A, the
fexA and
optrA genes were located on the same plasmid of 41 126 bp, designated as pAJ10XM1A (GenBank: OR734227). The pAJ10XM1A plasmid represents a novel entity with a GC content of 30.73% and encodes 45 ORFs. This plasmid exhibited a nucleotide sequence identity of 31.32% and a sequence coverage of 73% to the
optrA positive pP47-61 plasmid from
E. faecium (CP091102.1) in the NCBI database. In addition to the
erm(B) gene, pAJ10XM1A harbored a gene cluster consisting of IS
1216E-fexA-optrA-erm(A)
-IS
1216E, which was conserved in previously reported
optrA-positive plasmids, such as pC25-1 (CP030043), pC54 (CP030046), pP47-61 (CP091102), pW6-2 (CP118549), and pT17-1 (CP109840). In
E. faecalis S11XM4A, the
fexA gene was located on the plasmid pS11XM4A (OR734628), which had a size of 59 821 bp (
Fig. 4(a)). The pS11XM4A plasmid also carried the resistance genes
erm(B),
tet(L), and
tet(M). It showed 81.56% and 93.99% nucleotide sequence identity to pEFT30-1 (CP113829.1) and p26975-26-2 (LR962637.1) plasmids, respectively (
Fig. 4(b)). However, the pS11XM4A plasmid lacked the IS
1216E-fexA-optrA-erm(A)
-IS
1216E gene cluster; instead, it only harbored a
fexA gene inserted between two IS
Enfa family transposases. These findings suggest that IS
1216E and IS
Enfa may facilitate the co-transfer of
optrA and
fexA genes among different plasmids.
Based on Orifinder database, we found that pS11XM4A contained genes encoding relaxase and the type IV secretion system (T4SS), characterizing it as a typical conjugative plasmid. In contrast, pAJ0XM4A lacked identifiable T4SS and relaxase-encoding genes, suggesting that it might be a non-self-transmissible plasmid. However, subsequent conjugation experiments demonstrated that both pS11XM4A and pAJ10XM1A could be successfully transferred to E. faecalis JH2-2, with transfer frequencies of 3.6 × 10−5 and 1.0 × 10−8, respectively. The resulting transconjugants were resistant to florfenicol (MIC = 64 μg∙mL−1) and linezolid (MIC = 16 μg∙mL−1). The low-frequency transfer of pAJ10XM1A may be facilitated by another T4SS-positive plasmid pAJ10XM1A-3 (GenBank: PQ463012) present in the same strain or through other unidentified mechanisms. In addition, both pS11XM4A and pAJ10XM1A could be electrotransformed into E. faecium strains CXM16A, conferring resistance to florfenicol (MIC = 64 μg∙mL−1) and linezolid (MIC = 8 μg∙mL−1).
3.6. Colonization of Enterococcus probiotics promoted optrA transfer and altered gut microbiota composition in chickens
Considering the widespread use of
Enterococcus in animal probiotic products and its potential role as a key vector for the transmission of ARGs, as well as the rarity of the linezolid resistance gene
optrA in probiotic strains, we used a chicken intestinal colonization model to evaluate the colonization ability of
optrA-positive
E. faecium from probiotic products and its impact on the intestinal microbiota and the dissemination of ARGs (
Fig. 5(a)). We observed that no
optrA-positive
Enterococcus was isolated from the control group during the trial, whereas the isolation rate of linezolid-resistant
Enterococcus in the colonization group initially decreased, followed by an increase within the first 21 d, it then remained stable until the 35th day. In addition, the overall isolation rate of linezolid-resistant
Enterococcus was higher in the cecum than in the ileum (
p < 0.05) (
Fig. 5(b)). In the final phase of the animal experiment, we randomly collected 66
Enterococcus strains from the intestinal contents of the colonization group. Among them, 26 strains were positive for
optrA, including 15
E. faecium (ST262, ST195, and ST398), ten
E. faecalis (ST16, ST618, and ST631), and one
E. gallinarum (Fig. S5 in Appendix A). This finding suggests successful colonization of the
optrA gene in the gut microbiota of chickens, which further suggests that the
optrA-positive
E. faecium AJ10XM1A can be stably colonized in the chicken gut and facilitate horizontal transfer of the
optrA gene to other ST types of
E. faecium and other
Enterococcus.
Based on a metagenomic data analysis, the chao1 index of the cecum samples on day 3 exhibited a significant reduction compared to samples on days 14 and 35, as well as compared with the ileum samples and the control groups. PCoA plot, utilizing the Bray-Curtis distance, revealed clear clustering segregation between the control and colonization groups (
Fig. 5(c)). Further analysis revealed that the microbiota in the ileum primarily consisted of
Firmicutes and
Proteobacteria, whereas
Bacteroidota emerged as the dominant phylum in the cecum. The impact of
optrA-positive
Enterococcus colonization on gut microbiota composition diminished with the growth of chickens (
Fig. 5(d)). In comparison to the control group, the increase in the
Ralstonia,
Sphingomonas, and
Escherichia genera contributed to the heightened levels of Proteobacteria in the ileum, whereas the reduction of
Phocaeicola,
Bacteroides, and
Parabacteroides led to the decreased abundance of
Bacteroidota in the cecum (
Fig. 5(e)). Metagenomic analysis validated a significant increase in
E. faecium levels in both ileum and cecum samples from the colonized group compared with the control (
Fig. 5(f)). Furthermore, this change exhibited a significant positive correlation with the changes of abundance observed in
Proteobacteria and
Bacteroidota (Fig. S6 in Appendix A).
Notably, based on α-diversity and PCoA analyses, the diversity and community composition of microbiota in the optrA-negative Enterococcus colonization group at days 3, 14, and 35 exhibited a change trend similar to that of the optrA-positive colonization group (Figs. S7(a)-(d) in Appendix A). The abundance of Enterococcus in the ileum and cecum decreased over time, while microbial richness progressively increased (Figs. S7(e) and (f) in Appendix A). These findings suggest that during the early developmental stages of the chicken intestine, extensive colonization of Enterococcus leads to a significant reduction in microbial diversity and abundance in the cecum, but not in the ileum. Importantly, this effect appears to be independent of whether the strain carries the optrA resistance gene.
3.7. Multidrug-resistant Enterococcus probiotics significantly affect the composition and abundance of intestinal resistance genes
Next, we explored the impact of
optrA-positive
Enterococcus colonization on ARGs and MGEs within the chicken intestinal microbiota. The results revealed a significant elevation in both the total relative abundance and the number of ARGs in ileal samples (on days 3 and 14) and cecal samples (on day 3) from the colonization group in comparison with those from the control group (
Figs. 6(a) and
(b)). The PCoA analysis based on Bray-Curtis distances revealed distinct clustering and divergence in ARG profiles within the ileal and cecal samples (
Fig. 6(c), Fig. S8(a) in Appendix A). Nevertheless, no significant difference was observed in the diversity of the ARGs, as assessed by the Shannon index, between the colonization and control groups (
Fig. 6(d)). Subsequent analysis revealed that the fluctuations in ARG abundance within the colonization groups were predominantly due to significant increases in specific resistance genes. These genes included the aminoglycoside resistance gene
ant(6)-Ia, aac(6)-Ie, and
aac(6)-I, macrolide, lincosamide, and streptogramin B (MLS) resistance genes
lsa(E),
erm(A/B), and
lnu(B), the tetracycline resistance genes
tet(L/M), as well as the phenicol and oxazolidinone resistance genes
fexA and
optrA (
Fig. 6(e)). This phenomenon was attributed to the initially elevated colonization of
optrA-positive bacteria. We also confirmed a significant positive correlation among the abundance of
Enterococcus, the abundance of
optrA and
fexA genes located on plasmids, and the abundance of the
lsa(E) genes located on the colonizing strain chromosome (Fig. S8(b) in Appendix A). In addition, we noted that the relative abundance of MGEs in the ileum (on days 3 and 14) and cecum (on day 3) samples of the colonization group was generally elevated compared with the control group, with a significant increasein the relative abundance of IS element-transposon (IS_Tn) and plasmids (
Fig. 7(a)). In particular, the abundances of the classes 1 and 2 integrase-encoding genes
intI1 and
intI2, the insertion sequences IS
1R, IS
1542, and IS
1216E, as well as the plasmid replication initiator protein-encoding genes
repAJ10_repR,
rep18b_repA,
repUS12_repB, and
rep9a_repA were relatively increased in the colonization group compared with the control group (
Fig. 7(b)).
Correspondingly, as the optrA-negative Enterococcus strain CCW15A did not carry additional resistance genes, its colonization did not cause significant disturbances in the abundance of AMR genes in the ileum and cecum, nor did it substantially affect the diversity of AMR genes present. More specifically, the abundance of AMR genes, categorized by antibiotic classes, typical resistance genes, and diversity (Shannon index), was significantly lower in the optrA-negative colonization group compared with the optrA-positive group (Figs. S9(a)-(f) in Appendix A). Moreover, PCoA with Bray-Curtis distances analysis revealed a certain degree of clustering separation in the composition of AMR genes between optrA-positive and optrA-negative colonization groups in both the cecum and ileum (Figs. S10(a) and (b) in Appendix A). These findings confirm that probiotic Enterococcus strains can serve as carriers for various important AMR genes. The extensive use of multidrug-resistant Enterococcus can significantly influence the compositional structure and abundance of resistance genes in the chicken gut microbiome.
3.8. MGEs played a crucial role in the transfer of ARGs from Enterococcus probiotics to intestinal microbiota
A network correlation analysis revealed that in both ileal and cecal samples, the phenicol/oxazolidinones resistance genes such as
optrA and
fexA exhibited significant positive correlations not only with the MGEs, including IS elements, transposons, and plasmids but also with the bacteria belonging to
Bacillus,
Kiritimatiellia, and
Actinomyces (
Fig. 7(c), Fig. S11(a) in Appendix A). Furthermore, we verified that in intestinal samples, the relative abundance of IS
1216E and
repR genes exhibited significant positive correlations with the abundance of the
optrA gene and
E. faecium (Fig. S11(b) in Appendix A). These results confirmed the significant impacts of
optrA-positive
Enterococcus colonization on intestinal ARGs, MGEs, and microbiota, suggesting the potential transfer of
optrA gene in the intestinal microbiota.
To determine whether the
optrA gene had been transferred into the normal intestinal microbiota, we conducted a metagenomic binning analysis to investigate the contigs carrying the
optrA and
fexA genes, along with their putative bacterial hosts. The results identified a total of 30 MAGs harboring
optrA from the cecum and ileum samples, with 21 MAGs (70%) classified as moderately usable MAGs, meeting the minimum criteria of > 50% completeness and < 10% contamination. These MAGs were identified as follows by means of GTDB-tk (v2.2.0):
E. cecorum (
n = 10),
E. gallinarum (
n = 2),
L. crispatus (
n = 2),
E. faecalis (
n = 2),
E. faecium (
n = 4), and
E. durans (
n = 1). Detailed information regarding the MAGs is provided in Table S8 in Appendix A. This revealed that all the contigs carried a relatively conserved IS
1216E-fexA-optrA gene cluster, demonstrating > 90% sequence similarity with the corresponding region of the
optrA-carrying plasmid pAJ10XM1. However, the complete sequence of the pAJ10XM1 plasmid carrying IS
1216E was not identified in the contigs of all MAGs, suggesting that IS
1216E may serve as a more effective genetic element than the plasmid in mediating the transfer of the
optrA gene into other intestinal bacteria (
Fig. 7(d)).
4. Discussion
Since the Chinese government passed a law in 2020 banning the use of antibiotics in animal feed, the use of antibiotics in livestock farming has decreased significantly. As a result, interest in the development of probiotics as an alternative to antibiotics has greatly increased [
28,
29]. Nevertheless, the safety of livestock probiotics, particularly their potential to propagate antibiotic resistance, has received limited oversight, in contrast to the stringent regulation of probiotic products for human use [
30]. To our knowledge, this study represents the first large-scale and systematic investigation of the quality and risks associated with the ARG transmission of livestock-usage probiotic products in China.
Our findings indicate that certain commercial livestock probiotic products in China do not meet to the good manufacturing practice (GMP) production standards, and quality issues have occurred, such as incorrect labeling, insufficient viable probiotic counts, and low isolation rate of
Lactobacillus species, which is in consistent with previous reports [
5]. Importantly, we isolated opportunistic pathogens, including
Cronobacter sakazakii,
Salmonella, and
B. cereus, from 33% of the sampled livestock probiotic products. These opportunistic pathogens did not demonstrate dominant clonal lineages or clustered phylogenetic relationships, suggesting that these strains were likely to have been randomly contaminated during the production process. These opportunistic pathogens not only harbored multiple ARGs and virulence factors but also posed a potential risk of animal intoxication or infection in the context of extensive probiotic utilization [
31,
32].
The phylogenetic analysis of the probiotic strains revealed a relatively diverse range of ST types within
Bacillus and
Enterococcus. Probiotic strains of the same species displayed multiple evolutionary branches; in particular, certain
Enterococcus strains isolated from the probiotics exhibited relatively close genetic relationships with potentially pathogenic
Enterococcus in human. These results suggest that, in contrast to the stringent strain-level requirements for probiotics used in humans and infants, raw probiotic strains intended for livestock use, particularly those of
Enterococcus species, demonstrate significant intraspecies heterogeneity and possess unclear genetic backgrounds, lacking unified strain standards [
33].
Regarding antibiotic resistance, our findings indicate that the
Bacillus and
Lactobacillus found in the probiotics primarily exhibited intrinsic resistance mediated by chromosome-borne determinants, with no transferable AMR genes identified on plasmids. In contrast, certain
Enterococcus strains exhibited complex resistance patterns and harbored multiple typical transferable resistance genes. In particular, we identified two novel plasmids carrying
optrA or
fexA genes, which are capable of conjugating to other
Enterococcus at a low frequency
in vitro. These results suggest that the risk of ARG transmission is low for
Bacillus and
Lactobacillus probiotics but high for
Enterococcus probiotics, which aligns with previous concerns regarding the possibility of resistance gene transmission in
Enterococcus probiotics [
34].
Previous studies have shown that the animal gut is a hotspot for the horizontal transmission of ARGs [
35]. Our studies suggest that the extensive use of
Enterococcus probiotics, regardless of whether they are carriers of multidrug resistance genes or not, affects the composition of the microbiota in the ileum and caecum of chickens, leading to a marked decrease in the abundance and diversity of the microbiota during the early colonization stages. This is consistent with previous studies showing that the dominance of
Enterococcus may promote the growth of other pathogenic bacteria in the gut, such as
E. coli and
Salmonella, through competitive inhibition or interactions with metabolites, further exacerbating the microecological imbalance in the gut [
36,
37]. The decrease in microbial abundance may facilitate the colonization of inoculated bacteria and the dissemination of ARGs [
38,
39]. In addition, we observed persistent colonization of
optrA-positive
Enterococcus in the ileum and cecum of the experimental group, accompanied by a significant increase in the abundance of ARGs and MGEs, which continued until the conclusion of the experiment (35 d). Significant correlations were also observed between ARGs and MGEs, such as insertion sequences and integrases, which may facilitate the horizontal transfer of ARGs to commensal microbiota. Although we did not directly demonstrate the transfer of resistance plasmids into other microorganisms due to limitations in second-generation metagenomic sequencing and analysis, we showed the horizontal transfer of the
optrA gene mediated by the IS
1216E transposon into the commensal bacteria, such as
E. cecorum,
E. gallinarum, and
L. crispatus. IS
1216E is a common insertion sequence in certain Gram-positive bacteria, such as
Enterococcus [
40]. The presence of IS
1216E in the chicken intestinal microbiota heightens the likelihood of horizontal gene transferral events, thereby accelerating the spread of important ARGs within the gut. Therefore, once the
Enterococcus strains used as probiotics acquire MGEs and transferable ARGs, they can easily act as vectors of ARGs, continuously colonizing in the intestines of livestock and spreading resistance genes.
Based on our findings, we propose several recommendations for the use and production of probiotic strains for livestock. Firstly, probiotics, particularly Enterococcus, should undergo rigorous molecular-level identification. Beyond screening for common virulence factors, it is imperative to assess MGEs (e.g., IS1216E) and transferable AMR genes (e.g., optrA) to mitigate the risk of these strains serving as vectors for antibiotic resistance transmission. Furthermore, standardized guidelines for probiotic strain usage in livestock should be established at the individual strain level. This would facilitate the harmonization of genetic backgrounds and information for commercially available livestock probiotic strains, enabling the development of a comprehensive strain traceability system. The principle of species-specific applicability should also be adhered to for individual strains, which will necessitate customized adaptability assessments and experimental studies for different livestock species to ensure both efficacy and safety. Finally, the manufacturing process should be optimized according to GMP guidelines to ensure the consistency, potency, and purity of the probiotic product while preventing contamination by pathogenic microorganisms during the production process.
In conclusion, our study has substantiated specific issues for livestock-usage probiotic products in China, such as inconformity with GMP production standards and inadequate characterization of the genetic backgrounds of probiotic strains. When substandard Enterococcus strains containing multiple ARGs are extensively utilized as probiotics, they not only disrupt the equilibrium of the normal intestinal microbiota of broilers but also act as a reservoir and origin for the dissemination of intestinal ARGs. Therefore, in the future, cautious consideration should be given to the large-scale application of Enterococcus strains as livestock probiotics, in order to mitigate potential adverse effects on livestock and public health. Meanwhile, in addition to strictly selecting qualified strains and standardizing production, interdepartmental collaborative efforts should be established to enhance regulatory oversight of the livestock probiotics industry supply chain, and to develop comprehensive guidelines to ensure the quality and safety of livestock probiotic products.
CRediT authorship contribution statement
Xing Ji: Writing - review & editing, Writing - original draft, Investigation, Funding acquisition. Jiayun Wang: Supervision, Data curation. Jun Li: Supervision, Software. Lili Zhang: Validation, Supervision. Ruicheng Wei: Resources, Formal analysis. Ran Wang: Project administration. Tao He: Project administration, Funding acquisition, Conceptualization.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
This research was funded by the National Key Research and Development Program (2022YFD1800400) and Jiangsu Provincial Natural Science Foundation Project (BK20220746).
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.eng.2025.03.032.
PDF
(4553KB)
