National Safety Survey of Animal-use Commercial Probiotics and Their Spillover Effects From Farm to Humans: An Emerging Threat to Public Health

Abstract

Background

Human-use probiotics have recently been associated with clinical infections and antibiotic resistance transfer, raising public concern over their safety. However, despite their extensive application in aquaculture and animal husbandry, the safety of animal-use probiotics remains poorly described.

Methods

We evaluated the safety of 92 animal-use probiotics from China. The pattern of spread of pathogens from probiotics and the consequent public health implications were also examined by conducting in-field genomic surveillance at 2 farms.

Results

A total of 123 probiotic Bacillus species isolates were obtained from 92 brands of probiotics, of which 45 isolates were resistant to antibiotics. Notably, 33.7% of probiotic products were contaminated with life-threatening pathogens such as Klebsiella pneumoniae. Genomic surveillance at a chicken farm identified an anthrax toxin–positive Bacillus cereus strain in a probiotic product used as a feed supplement, which was transferred into the groundwater and to a nearby fish farm. Following up retrospective analysis of the surveillance data during 2015–2018 in 3 provinces retrieved 2 B. cereus strains from human with intestinal anthrax symptoms and confirmed the transmission of B. cereus from farm to human. Surveillance of anthrax toxin revealed that cya was detected in 8 of 31 farms.

Conclusions

This study provides the first national safety survey of animal-use probiotics in China and confirms the spillover effects of probiotics from the farms to human. These results suggest that the large-scale application of pathogen-containing probiotics leads to the transfer of pathogens, with worrisome implications for public health. Good Manufacturing Practice should be implemented during the production of all probiotics.

Animal-use probiotic products are frequently contaminated with viable pathogenic bacteria. This study revealed that virulent probiotic organisms and contaminating pathogens were colonized with farm animals and shed into the environment, which facilitated the transfer of pathogens to humans.

Probiotics are live microorganisms that can be administered orally, with potential health benefits for humans and animals [1]. Emerging research supports the use of specific probiotic species, such as Bacillus species (spp), to treat certain medical conditions in humans [2]. Probiotics have also been used extensively in aquaculture and animal husbandry, and numerous studies have suggested potential roles for probiotics in stimulating the innate immune response for aquatic animals and livestock [3, 4]. Preliminary evidence also showed that probiotics improved meat flavor in animal husbandry [5]. Bacillus spp comprise the second largest genus used in marketed probiotic preparations, after Lactobacillus spp [6].

The expansion of human-use probiotics market has resulted in some safety issues including mislabeling, the presence of pathogenic probiotics, and microbial contamination [7–8]. Violations in Good Manufacturing Practice (GMP) frequently occur and have resulted in contamination of the final product with potentially life-threatening pathogens [9]. A fatal infection following the application of a probiotic Bacillus subtilis preparation was reported in an immunocompromised patient [10], while the death of an 8-day-old premature infant with fulminant gastrointestinal mucormycosis was also confirmed to be due to fungal contamination of the probiotic supplement used to treat the infant [11].

Compared with the occasional reports of infections related to human probiotic products, safety issues associated with animal-use probiotic products are likely to be even worse. The European Union (EU) and the United States (US) have both published lists of acceptable microorganisms for use as “direct-fed microbials” for livestock [3]; however, no authorities have taken responsibility for evaluating the safety of probiotics for animal use at a strain level. Furthermore, probiotic bacterial species also carry the risk of becoming conduits for the spread of toxin genes and antibiotic-resistance genes [12]. A recent study suggested that an accurate evaluation of risk-related gene traits in probiotics was essential to meet the health claim requirements and to avoid risky gene transfer, in accordance with EU and US health policies [13].

Microbial contamination of animal-use probiotics is also a matter of concern. Zhu et al suggested that Acinetobacter spp and Cronobacter spp were frequently found in human probiotic products [7]. Although a few case studies have evaluated the safety of human- and animal-use probiotic products [13, 14], there is a lack of systematic surveillance to detect postmarketing microbial hazards from the products. Actual prevalence of opportunistic pathogens in animal-use probiotics is currently unknown, especially for ESKAPE pathogens (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter spp).

The current study thus aimed to evaluate the safety of animal-use probiotics and levels of microbial contamination by detecting the presence and viable numbers of clinically significant pathogens. To assess the impacts of the large-scale application of animal-use probiotics from the One Health perspective, we conducted genomic surveillance to monitor the dissemination of live microorganisms from probiotics to the environment.

MATERIALS AND METHODS

Isolation and Characterization of Probiotic Bacillus Species and Contaminating Pathogen

A total of 92 brands of Bacillus-containing probiotics were collected from China between 2016 and 2018, none of which were labeled with GMP. All the products were directly purchased from the producers and were labeled Bacillus spp as the sole ingredient or 1 of the ingredients. Bacillus spp and contaminating pathogens were identified by chromogenic selective medium (Supplementary Table 1 and Supplementary Methods). The presence of Bacillus-related toxin genes including anthrax toxin gene cya was detected by polymerase chain reaction (PCR) method (Supplementary Table 2). Antimicrobial susceptibility tests were performed by the disk diffusion method, according to the British Society of Antimicrobial Chemotherapy methods and breakpoints for gram-negative and gram-positive strains [15]. Twenty-four antibiotic disks covering 8 types of antibiotics were included (Supplementary Table 3).

Pathogen Surveillance of Probiotic Bacillus cereus in Farms and the Nearby Environment

We further investigated the spillover effects for one of the cya-positive probiotic products (product code CHN-H-11). Inquiry for the sale data of the producer showed that it was sold to Liaoning, Jiangxi, and Zhejiang provinces during 2015–2017. We chose a chicken farm near Anshan (Liaoning) that used this probiotic product as a feed supplement since September 2017. Wastewater from the farm, containing fecal material and unconsumed feed, was discharged into the sewerage system without treatment. In September 2017, we therefore sampled the probiotics and 16 wastewater samples to recover Bacillus cereus. Meanwhile, we also sampled 15 groundwater samples and 20 fish from a nearby fish farm that used groundwater as a water source, to check for the presence of B. cereus.

Meanwhile, a pathogen surveillance program was conducted in the 8 sampling sites along the Liaohe River in Liaoning (August 2017 to July 2018). Ten water samples from each site were collected monthly to detect 5 foodborne pathogens (B. cereus, S. aureus, Escherichia coli O157:H7, Salmonella, and Shigella). Bacillus cereus strains collected from above surveillance were selected for multilocus sequence typing (MLST). MLST of B. cereus was performed by sequencing 7 housekeeping genes (glpF, gmk, ilvD, pta, pur, pycA, and tpi) as described previously [16].

Retrospective Review of the Presence of Probiotic B. cereus in Other Regions

A retrospective review of previous B. cereus cluster investigations in Jiangxi and Zhejiang was then undertaken. We extracted B. cereus outbreak records during 2015–2018 collected by the Nanchang and Hangzhou Center for Disease Control and Prevention. The investigation summaries include demographic details, symptom and illness onset, food consumption, and other environmental exposures in the 7 days prior to onset of illness (especially for poultry and egg consumption); in addition, MLST typing of B. cereus isolates was reviewed.

To detect the presence of toxin gene cya, 31 farms from Jiangxi, Zhejiang, and Liaoning provinces were visited once during April–July 2018. In each farm, 10 stool samples were collected from farm animals including fish, shrimp, and chickens to obtain B. cereus.

Genomic Sequencing and Analysis

Genomic DNA from 9 B. cereus and 5 Acinetobacter pittii strains isolated from the probiotic products, chickens, the nearby environment, and human stool samples were sequenced by Illumina HiSeq 2500 platform (Novogene, China). De novo assembly and single-nucleotide polymorphism (SNP) calling were performed by a custom pipeline as described previously [17]. The phylogenetic trees were constructed with the maximum likelihood method using RAxML 7.2.8 under the GTR + G + I model [18].

Ethical Considerations

In accordance with the Declaration of Helsinki, both patients had given their written informed consent for the use of his personal and medical information for the publication of this study. This study was approved by the ethics committee at the Nanchang Center for Disease Control and Prevention.

RESULTS

Identification of Probiotic Bacillus Species and Detection of Toxin Genes

Most products (66.3%) only contained 1 probiotic Bacillus spp (Figure 1A). A total of 123 probiotic Bacillus spp strains were identified (Figure 1B), of which 45 strains confer resistance to multiple antibiotics such as penicillin and cephalosporins (Figure 1C and Supplementary Results). Bacillus-related toxin genes were detected in 22 strains, of which 20 were hblIII-positive. nhe was detected in 10 strains by PCR (Figure 1D). Notably, 3 strains harbored genes encoding anthrax toxin cya. Only 8 strains, which harbored all of the Nhe components, were positive in the Vero cell assay (Figure 1D).

Identification and Evaluation of Microbial Contamination in Probiotics

A total of 54 nonprobiotic strains were isolated from the probiotics, of which 81.5% belonged to the ESKAPE group (Figure 2A). ESKAPE species were found in the products with viable numbers of 2.06–2.55 log colony-forming units (CFU)/g (Figure 2B). All strains exhibited resistance to antibiotics, the majority of which confer resistance to cephalosporins, sulfonamides, erythromycin, and tetracycline (Figure 2C and Supplementary Results).

Presence of cya-Positive B. cereus in Probiotics, the Environment, and Humans

As investigation showed that 1 probiotic containing cya-positive B. cereus was sold to Liaoning during 2015–2017 and was used as a feed supplement at the chicken farm in Anshan since September 2017. We evaluated the spillover effect during extensive application of probiotic Bacillus spp to animals by examining the transfer of B. cereus from probiotics to the environment at this chicken farm. A follow-up investigation showed that wastewater from the chicken farm was frequently discharged into the groundwater after washing the chicken cages. We therefore sampled the probiotics used at the chicken farm and the discharged wastewater, as well as fish and groundwater from a nearby ornamental fish farm, which did not use any probiotics during the period of the investigation.

Five B. cereus strains were obtained from the probiotics, wastewater, fish, and groundwater used in the fish farm. MLST revealed that 4 strains belonged to sequence type (ST) 234, while the other, obtained from the groundwater, belonged to ST1263 (Figure 3A). A parallel pathogen-surveillance program conducted along the Liaohe River detected 15 non-ST234 B. cereus strains from 8 sampling points since 2017 (Supplementary Table 4).

To track if the use of probiotics has resulted in the transmission from farm to human, we also analyzed 35 clinical and environmental B. cereus strains recorded in Jiangxi and Zhejiang provinces during 2015–2018 (Figure 3B) (Supplementary Table 4) and found 1 and 2 ST234 strains from chicken and clinical samples, respectively, in June 2017. Two ST234 B. cereus strains, NC-13 and NC-17, were cultured from stool samples from a 21-year-old man and a 20-year-old man, respectively. The epidemiological record showed that 2 human cases had suspected symptoms of intestinal anthrax and were associated with the consumption of undercooked chicken. Following up investigations by interviewing patients and the distributor further identified the chicken farm that supplied the contaminated chicken meat. We further identified 4 B. cereus isolates from 25 chicken stool samples in the farm in July 2017, including 1 ST234 strain (Supplementary Table 4). We thus revisited the chicken farm in April 2018 and confirmed the use of the probiotic product since 2016. Therefore, the above 3 ST234 strains, together with 4 ST234 strains from Liaoning, were selected for genomic sequencing to confirm their epidemiological links. One ST1263 strain and another ST184 strain (both from Liaoning) were also sequenced as background isolates.

Dissemination of B. cereus From Probiotics to the Environment and Humans

Together with 430 publicly available genomes (Supplementary Table 5), genomic analysis of 439 B. cereus strains divided them into 5 clusters (clusters I–V), with most belonging to clusters I and II (Figure 3A). Seven ST234 strains were located in cluster I, while the ST1263 strain isolated from groundwater and another ST184 strain obtained from pathogen surveillance were both distributed in cluster I. SNP analysis revealed that 4 ST234 strains retrieved from Liaoning only differed by 4 SNPs, suggesting that they belonged to the same clone (Figure 3C). These observations implied that the ST234 B. cereus strain found in the fish and groundwater had originated from the probiotics used in the chicken farm, which had been transmitted to the groundwater during the discharge of wastewater. Remaining 2 ST234 strains from humans and 1 from a chicken farm in Jiangxi have no SNP difference with probiotic strain JY6, suggesting that the probiotics containing ST234 strain might retain in the chicken and transfer from farm to table, resulting in 2 human infection cases (Figure 3D).

Genomic analysis of virulence genes revealed that nhe, hlyII, and hlyIII and the anthrax-toxin genes cya and pagA were present in all of the ST234 strains (Supplementary Table 6), of which cya and pagA were found in a plasmid (Supplementary Figure 1 and Supplementary Results). We also detected the presence of cya in 31 farms from Jiangxi, Liaoning and Zhejiang provinces, of which cya was detected in stool samples from 8 farms (Supplementary Table 7), indicating that large-scale application of this probiotic strain greatly promoted the dissemination of pathogenic B. cereus from the farm to the environment. Genomic surveillance of A. pittii strains (Supplementary Table 8) along the Liaohe River also confirmed the spillover of A. pittii from probiotics in a shrimp farm to the environment (Figure 4 and Supplementary Results).

DISCUSSION

In this study, we conducted a national safety survey of animal-use probiotic products. More than one-third of probiotic products harbored antibiotic resistance and contaminating pathogens. Importantly, in-field genomic surveillance confirmed a spillover of pathogen-containing probiotics from farms to human, indicating an emerging threat to public health.

The global probiotics market was estimated to be worth US$45.64 billion in 2017, and is projected to reach US$64.02 billion by 2022 [19]. From a global perspective, the overall safety of human-use probiotics is assessed and strictly regulated [8]; however, the safety of animal-use probiotics is rarely examined. Bacillus spp are the most frequently used probiotic species for animals; however, recent research has identified several new virulence factors in Bacillus spp [20–21], indicating the need to revisit the safety assessment of probiotic Bacillus spp.

In addition to Bacillus spp, other bacterial genera have been developed and used as ingredients of probiotics for aquatic animals and livestock [22], but the public health significance of these species has never been evaluated. For instance, a few marine probiotic species, such as Alteromonas spp and Rhodopseudomonas spp, which are traditionally not considered to be human pathogens, were also associated with clinical infections [23, 24], indicating that their potential risk to humans should not be ignored. Notably, carnobacteria and lactobacilli, which have been widely used as probiotic bacteria in fermented food products, have been responsible for clinical infections including septicemia, meningitis, and life-threatening bacteremia [25]. Infection related to Vagococcus lutrae, which has been used in probiotics in aquaculture, was recently reported in France [26, 27].

The current results also highlighted the severity of microbial contamination in animal-use probiotics. Several life-threatening pathogens such as Cronobacter sakazakii, Shigella sonnei, and ESKAPE pathogens were identified. Likewise, Zhu et al identified a high prevalence of A. baumannii and C. sakazakii in human probiotic products [7]. Given that these contaminating pathogens would be amplified by incubation prior to use, their extensive application in aquaculture and animal husbandry could contribute high pathogen loads to the environment.

From a One Health perspective, virulent Bacillus spp and other contaminating pathogens used in animals could pose an emerging threat to public health in 2 ways. First, live microorganisms could be retained in meat and seafood and be transmitted to humans via food consumption. Supportably, the epidemiological links of Bacillus spp between humans and chickens were confirmed in this study, indicating that pathogen-containing probiotics pose an evident risk for food safety. As B. cereus only results in self-limiting diarrhea or vomiting, numerous cases might be not reported and recorded. The potential food safety issue related to unsafe use of probiotics, therefore, should not be underestimated [28]. The widespread detection of cya in the farm animals of 3 provinces also implied that the public health significance of the large-scale application of probiotics should not be underestimated.

Although we did not monitor the viable numbers of Bacillus spp in the marketed chickens and shrimp from the tested farms, numerous studies have suggested that probiotic strains can dominate the intestines of aquatic or terrestrial animals if used in adequate amounts [5, 29]. To achieve significant effects, probiotic products are usually applied to fish ponds or livestock at high concentrations (10⁴–10⁶ CFU/g feed) to guarantee sufficient numbers to affect the microflora [30]. These results suggest that large numbers of pathogenic Bacillus spp and contaminating pathogens may be retained in the animals and potentially transferred to the marketed products [31]. The presence of these pathogens in ready-to-eat raw fish and meat poses a particular health risk to humans [32]. In addition, as the spore of the Bacillus spp is heat-resistant, there is still a risk when meat is thoroughly cooked. The barriers to transmission between animals and humans are low [33], and probiotics that persist in retail meats and seafood may thus become potential vehicles for transmitting virulent, antibiotic-resistant pathogens from food animals to humans. This could in turn result in foodborne outbreaks from the consumption of meat from probiotic-fed animals [34].

In addition to their retention in meat and seafood products, live microorganisms might also be transmitted by dissemination from farms into the environment, which pose a threat for drinking water safety. Several studies have shown that of the use of probiotics resulted in high levels of live microorganisms from the probiotic products in animal feces [35], thus creating a melting pot of bacteria from human, animal, and environmental origins and thereby facilitating pathogen transfer [36] (Figure 5).

Genomic surveillance of B. cereus and A. pittii strains in probiotic products and the nearby environment provided clues to the endemic expansion of pathogens in probiotics from farms to the environment. Both cases demonstrated that the large-scale application of probiotics promoted pathogen transmission from products to water sources or to animals and to humans (Figure 5).

Genomic surveillance of an anthrax toxin–positive B. cereus in probiotic products and the nearby environment also highlighted the possibility that an endemic disease such as anthrax could develop into an epidemic in certain regions as a result of the use of probiotics. For instance, B. cereus with anthrax toxin genes has been frequently associated with human infections and caused anthrax eschar and even fatal pneumonias [37, 38]. As B. cereus can be transmitted via aerosol, unsafe probiotic strains would also pose a threat for work safety for those handle with animals [37]. The current results reported the first cases of intestinal anthrax due to application of probiotic B. cereus. Such activities could also promote the spread of pathogen and trigger a chain reaction of antibiotic-resistance and pathogen transfer in ecological systems [36]. The above observations thus suggest an emerging need to exclude virulent probiotic strains and contaminating pathogens from commercial probiotics.

The current study had 2 limitations. First, we only investigated the probiotics in China, leaving a data gap for other countries. Second, we only focused on the safety of Bacillus spp and did not assess the risks associated with the other probiotic microorganisms, such as yeast and nitrifying bacteria. Nevertheless, given that clinical infections attributed to the above species are rare, we considered that these might not be major risk factors in terms of the safety evaluation.

In view of the wide and uncontrolled use of probiotics, strict regulations are urgently needed to minimize the potential risk associated with unsafe probiotic products. Rigorous safety screening is essential to reduce the probability of transferring antibiotic-resistance genes and to eliminate undesirable pathogens from probiotics [39–40]. Only antibiotic resistance- and toxin-free candidate species should then be considered for inclusion in probiotics, and in vivo animal experiments should perform to check in vivo toxicity. Strict implementation of GMP is also essential to prevent pathogen contamination and spillover during the large-scale application of probiotics.

CONCLUSIONS

In conclusion, we evaluated the possible health hazards of animal-use probiotic products in China in relation to toxin production, antibiotic resistance, and contamination by life-threatening pathogens. The results showed that more than one-third of animal-use probiotics contained virulent Bacillus and other pathogens. Genomic surveillance in chicken farms revealed that long-term use of probiotics promoted the dissemination of pathogens from the farm to humans. These bacterial species could spread into the environment or be retained in the food, thus posing an emerging threat to public health. We therefore call for the stringent implementation of GMP to ensure the safety of probiotic products.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. The authors thank Xudong Shen, Shibo Jin, Dr Kui Wu, Dr Dawei Wei, Dr Ziqiang Deng, and Dr Min Ouyang for sampling assistance. The raw sequencing data were submitted to GenBank (National Center for Biotechnology Information) under the BioProject number PRJNA539852.

Financial support. This work was supported by the National Key Research and Development Program of China (grant number 2017YFD0701700); the National Natural Science Fund of China (grant numbers 31672673, 31702391, 81903372); and the Liaoning Science and Technology Project (grant number 2017203002).

Potential conflicts of interest. The authors report no potential conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References

Author notes