Most commensally bacterial strains in human milk of healthy mothers display multiple antibiotic resistance

Abstract Recent reports have shown that food‐borne or commensal bacteria can function as reservoirs of antibiotic resistance. However, the antibiotic susceptibility of bacterial isolates of most milk samples or the total bacterial counts (TBC) in human milk from healthy donors, are not fully understood in Taiwan. Thus, five healthy mothers were randomly recruited each month, and totally 30 mothers without any symptoms of infection were recruited over 6 months. Milk samples were then harvested and analyzed immediately after collection. The antibiotic susceptibility was analyzed in bacteria isolated from milk samples using nine clinically relevant antibiotics, such as oxacillin, ampicillin, cephalothin, amoxicillin, ciprofloxacin, erythromycin, clindamycin, gentamicin, and oxytetracycline. The Staphylococcus strains (48 isolates) found in milk resisted to 48.6 ± 20.1% selected antibiotics. Streptococcus‐related isolates (8 isolates) exhibited resistance to 41.7 ± 26.4% selected antibiotics. Acinetobacter isolates (5 isolates) were resistant to 66.7 ± 13.6% antibiotics, and Enterococcus isolates (5 isolates) were resistant to 73.3 ± 6.1% tested antibiotics. Rothia‐related isolates (4 isolates) were resisted to 58.2 ± 31.9% of tested antibiotics. In contrast, Corynebacterium isolates (5 isolates) were sensitive to 66%–100% of selected antibiotics. Furthermore, the TBC ranged from 40 to 710,000 CFU/ml, implying a wide spectrum of bacteria in milk from healthy mothers. Despite this, all milk donors were healthy during sampling, and they did not show any symptoms related to mastitis or subclinical mastitis. According to the previously described TBC criteria for the use of donated human milk, only 73% of the current milk samples could be accepted for the milk bank. In conclusion, the majority of the isolated bacterial strains from current human milk samples are multiresistant strains. In milk samples for preterm infants or milk banks, higher TBC levels or potentially antibiotic‐resistant bacteria in some milk samples have supported people using approaches to disinfect human milk partially.


| INTRODUC TI ON
Human milk is generally accepted to be the ideal food for babies, and it is a rich fluid that contains essential nutrients, such as various bioactive compounds, proteins, carbohydrates, immune cells, and immunoglobulins, which can provide sufficient nutritional and protective requirements for infants (Petherick, 2010;Walker, 2010).
These commensal bacteria in milk can be ingested by infants, supporting that these commensal bacteria in milk may play a role in establishing the microbiota in the infant gut Solis, de Los Reyes-Gavilan, Fernandez, Margolles, & Gueimonde, 2010). Several recent reports have also shown that food-borne or commensal bacteria can function as reservoirs of antibiotic resistance genes, which are similar to those found in pathogenic bacterial strains (Devirgiliis, Zinno, & Perozzi, 2013;Mathur & Singh, 2005;White, Zhao, Simjee, Wagner, & McDermott, 2002). For example, Enterococcus spp. collected from milk of healthy donors, such as healthy women, pigs, dogs, sheep, and cats, have been found to exhibit resistance to various clinically relevant antibiotics, including gentamicin, streptomycin, quinupristin/dalfopristin, tetracycline, and chloramphenicol . Moreover, Staphylococcus epidermidis isolated from human milk of healthy mothers also shows resistance to gentamicin, tetracycline, erythromycin, clindamycin, and vancomycin (Begovic et al., 2013;Delgado et al., 2009). Another study demonstrated that several Enterococcus faecalis and E. faecium isolates from human milk contained virulence genes and antibiotic resistance that could serve as a reservoir of antibiotic resistance in offspring . In our pilot study in Taiwan, most bacteria isolated from human milk of 19 healthy donors also displayed mild to strong antibiotic resistance. For example, milk-isolated Staphylococcus isolates (22 strains) were resistant to 25%-100% of antibiotics, whereas Streptococcus isolates (three strains) were resistant to 33%-77% of antibiotics. Members of the genus Enterococcus (five strains) were resistant to 33%-88% of selected antibiotics (Chen, Tseng, & Huang, 2016). However, only 19 healthy donors had been recruited in this pilot study, and the antibiotic susceptibility of bacterial isolates of most human milk samples are not fully understood in Taiwan. Therefore, this study aimed to dissect the antibiotic susceptibility patterns among commensal bacteria isolated from newly 30 human milk samples of healthy mothers. At this time, about five healthy mothers without any symptoms of infection and clinical mastitis attending hospital were voluntarily and randomly recruited each month, and totally 30 milk samples were collected aseptically during a period of 6 months. Since cesarean section (CS) involves antibiotics exposure, it would be interesting to study the data in comparison between CS and vaginal delivery (often without antibiotic use during delivery). Therefore, milk samples harvested from mothers after normal spontaneous delivery (NSD) and those after CS delivery were also recorded and dissected. On the other hand, several regulations or criteria toward the total bacterial counts (TBCs) have been employed to evaluate the milk quality under the conditions of human milk sharing or milk samples collected by milk banks (Balmer & Wharton, 1992). However, this information is still vague in most human milk samples collected from healthy mothers in Taiwan. Then, we also evaluated the characteristics of milk samples by determining the TBCs levels. Human milk samples are shared between babies by the help of milk banks in Taiwan. Our findings are expected to help evaluate the TBCs and antibiotic susceptibility levels in human milk from healthy donors.

| Ethics, consent, and permissions
This study was approved by the Institutional Review Board (IRB) of Saint Mary's Hospital, Lundong (IRB104011), and informed consent was obtained from all participants.

| Milk sampling
Five healthy mothers without any symptoms of infection and clinical mastitis attending Saint Mary's Hospital Lundong were voluntarily and randomly recruited each month, and totally 30 different milk samples were collected over a 6-month period. Milk samples harvested from mothers after NSD and those after CS delivery were also recorded. These milk samples were collected by manual expression using sterile gloves after cleaning the breasts with sterile warm water or saline according to previous reports (Chen et al., 2016;Albesharat, Ehrmann, Korakli, Yazaji, & Vogel, 2011). Briefly, the first 1-2 ml milk was discarded to avoid possible contamination from the environment. After that, about 10-15 ml milk was collected in sterile tubes. The collected milk samples were transferred and analyzed at the lab immediately as indicated below. Collectively, each milk donor provided one sample, and we further divided the samples into several parts after mixing well. One part was used for aerobic TBCs test, bacterial isolation and antibiotic-sensitivity test, and one part was used for directly DNA extraction. The information of donors and samples is further shown at Table 1.

| Total bacterial counts, bacterial isolation, and identification
The aerobic TBCs were determined for all milk samples by following a standard plate count approach defining the bacterial counts in milk samples. Briefly, milk samples were immediately transferred to the laboratory. Each milk sample was mixed well thoroughly and a 1 ml whole milk samples was subjected to bacteria count approach.
After a serial 10-fold dilution using sodium chloride solution (0.85%), transfer 1 ml of each dilution into duplicate petri dishes. Then, add 12-15 ml plate count agar (cooled to 45 ± 1°C; tryptone 5 g, yeast extract 2.5 g, dextrose 1 g, agar 15 g, distilled water 1,000 ml, pH, 7.0 ± 0.2; prepared by autoclaving for 15 min at 121°C) on the dilution and mixing gently within 15 min of original dilution. Solidified petri dishes were then incubated under aerobic condition at 37°C for 48 hr, and we counted the number of bacterial colonies that appear on each of the plates that has between 30 and 300 colonies. TBCs for 72 hr in a Bugbox anaerobic work station (Ruskinn Technology, Ltd., Pencoed, UK; atmospheric composition: 80% N 2 , 10% CO 2 , and 10% H 2 ). For genotypic identification, chromosomal DNA of the isolates was extracted and used as a template in polymerase chain reaction (PCR), as previously described (Wang, Shyu, Ho, & Chiou, 2008), using primer sequences for the 16S rRNA gene (Wang, Shyu, Ho, & Chiou, 2007;Temmerman, Huys, & Swings, 2004). Moreover, PCR was carried out as previously described (Chen et al., 2016). Briefly, PCR was conducted in a final volume of 50 μl containing 10 mM Tris-HCl (pH 9.0), 50 mM KCl, nlm.nih.gov/Blast.cgi). As for bacterial classification, when the comparisons show more than 97% homology (concordance rate over 97%) will be identified as the same bacterial species; when the comparisons show more than 95% similarity or the concordance rate is over 97% but an overlapping to other species has been recognized, this will be only classified as genus-level.

| Total bacterial counts in milk samples
In this study, we also evaluated the characteristics of milk samples by determining the total bacterial counts (TBCs) in milk samples from healthy donors. To date, there are no regulations regarding the TBC criteria in human milk to restrict mothers from breastfeeding their babies. However, the bacterial counts have been evaluated in bulk tank milk or in raw milk from dairy farms, and the bacterial counts have been shown to serve as indicators and facilitate monitoring of herd udder health and milk quality (Jayarao et al., 2004;Hayes et al., 2001). In contrast, with regarding to human milk sharing or milk samples collected by milk banks, several regulations have been employed to evaluate the milk safety (Balmer & Wharton, 1992 and milk with counts more than 10 5 CFU/ml cannot be used (Balmer & Wharton, 1992). Moreover, when donor milk has bacterial counts between 10 3 and 10 5 CFU/ml, it is only accepted if the organisms are skin commensals, such as Staphylococcus epidermidis, viridans streptococci, and diphtheroids (Balmer & Wharton, 1992). Notably, donor milk is not accepted if TBCs are more than 10 3 CFU/ml with S.

| Antibiotic susceptibility of bacterial isolates
To monitor antibiotic susceptibility of bacterial isolates from human indicates that when the clinical isolates were resistant to at least one antimicrobial agent in three or more antimicrobial categories, and these isolates could be defined as multidrug-resistant bacteria (Magiorakos et al., 2012). Thus, the majority of the isolated strains (25/48) in Supplementary Table S1 are multiresistant strains according to the definition above.

Previous studies have reported strongly antibiotic-resistant
Staphylococcus strains from human milk of healthy donors (Begovic et al., 2013;Carneiro, Queiroz, & Merquior, 2004). For example, high antibiotic-resistance rates of S. epidermidis, S. warneri, S. haemolyticus, and S. aureus isolated from milk to penicillin (87%) and erythromycin (59.3%) have been observed. Moreover, several S. epidermidis isolates harvested from human milk have been shown to be resistant to tetracycline, erythromycin, clindamycin, and vancomycin (Begovic et al., 2013). In our previous study, most Staphylococcus spp. isolates in milk were also shown to display resistance to multiple antibiotics (Chen et al., 2016). In this study, the majority of the isolated strains are still multiresistant strains. These bacteria could play a role in mediating the risk of antibiotic resistance in milk samples.
In our previous report, Streptococcus spp. was isolated only in three milk samples, and these isolates displayed strong antibiotic resistance to selected antibiotics (Chen et al., 2016). In this study, a total of seven milk samples harbored Streptococcusrelated isolates (Supplementary Table S2), and two strains out of eight exhibited resistance to one antimicrobial agent in three or  (Magiorakos et al., 2012). Additionally, high antibiotic-sensitivity rates of Streptococcus to cephalothin (75%), amoxicillin (75%), ciprofloxacin (63%), clindamycin (75%), and gentamicin (88%) were observed. However, our data also showed high rates of antibiotic resistance to oxacillin (88%) and ampicillin (50%) in these isolates ( Figure 3). Totally, they resisted to 41.7 ± 26.4% selected antibiotics.
Supplementary Table S3 shows the antibiotic sensitivities of Acinetobacter spp. and Enterococcus spp. from human milk to all tested antibiotics. Acinetobacter isolates were resistant to 66.7 ± 13.6% tested antibiotics, and Enterococcus isolates were resistant to 73.3 ± 6.1% tested antibiotics. Figure 4 shows the rates of antibiotic sensitivity of Acinetobacter to three antibiotics, that is, ciprofloxacin (80%), gentamicin (80%), and oxytetracycline (80%).
Notably, high rates of antibiotic resistance of Acinetobacter to six antibiotics, that is, oxacillin (100%), ampicillin (100%), clindamycin (100%), cephalothin (80%), amoxicillin (60%), and erythromycin (60%), were observed. Acinetobacter is a gram-negative coccobacillus that has been recently recognized as an infectious agent of importance to hospitals worldwide (Fournier & Richet, 2006). Among Acinetobacter spp., A. baumannii has been shown to be an important pathogen in healthcare-associated infections. This species commonly shows resistance to multiple antibiotics and is difficult to treat (Fishbain & Peleg, 2010;Michalopoulos & Falagas, 2010). In this study, no A. baumannii was isolated from milk samples. However, as indicated above, the other Acinetobacter-related isolates were resistant to six antibiotics, indicating that these Acinetobacter species may play a role in maintaining severe antibiotic resistance in human milk.
The presence of E. faecium and E. faecalis in milk of healthy women has been reported previously (Hunt et al., 2011;Reviriego et al., 2005;Jimenez et al., 2008), and a recent report also found that  . In Figure 5, Enterococcus spp.
and Acinetobacter could more likely to confer a strong risk of antibiotic resistance in human milk samples worldwide. Another study also found a wide distribution of virulence genes and/or antibiotic resistance among the E. faecalis and E. faecium isolated from human milk . This previous study suggests that above bacterial strains could play roles in maintaining an antibiotic-resistant reservoir in animals and humans  antibiotics. Furthermore, we observed high rates of antibiotic sensitivity (80%) of Corynebacterium to ampicillin, cephalothin, amoxicillin, ciprofloxacin, gentamicin, and oxytetracycline and moderate rates (60%) of sensitivity to oxacillin and erythromycin ( Figure 6).
Thus, these data demonstrated that most Corynebacterium isolates from milk samples were sensitive to clinically relevant antibiotics.
In this study, 22 milk donors were treated with cephalexin, amoxicillin or cefazolin for antibiotic prophylaxis (Table 1) Table S4). These findings indicate that cephalothin or cefazolin prophylaxis may not be active enough against all milk-isolated bacterial strains from several milk donors. Doctors often prescribe cephalexin or cefazolin prophylaxis in this hospital, and as a result, these antibiotic prophylaxes may play some roles in elevating the antibiotic Acinetobacter spp. but this should be further investigated. In our previous pilot study, only 19 human donors had been recruited but we have observed some mild to strong antibiotic resistance among bacterial strains in human milk from healthy donors. Our current observation had confirmed that most bacterial strains in human milk could be multiresistant strains. Whether this is a prevalent situation in the other milk samples from the other areas or hospitals in Taiwan should be further investigated.
In conclusion, this study reported the bacterial counts and the antibiotic susceptibility pattern of commensally bacteria in human milk from healthy mothers in Taiwan. The majority of the isolated bacterial strains from human milk samples are multiresistant strains. Otherwise, the bacterial counts ranged from 4.0 × 10 1 to 7.1 × 10 5 CFU/ml, implying a wide spectrum of bacteria in milk samples from healthy mothers. Refer to the previously described total bacterial counts criteria for the use of donated human milk, only 73% of the current milk samples could be accepted for the milk bank. Although a wide spectrum of TBCs and the various bacterial patterns had been observed in milk samples from healthy milk donors but no milk donors suffered from any infections or symptoms of mastitis. These indicated that these diverse commensal bacteria in milk did not harm their hosts. In milk samples for preterm infants, higher TBCs levels in some milk samples should be kept in mind, and the roles of these antibiotic-resistant bacterial isolates in milk from healthy donors should be further investigated.

ACK N OWLED G M ENTS
This work is partially supported by a grant from St. Mary's Hospital Luodong (SMHRF-104003 and SMH-106003), and also supported by a grant from the Ministry of Science and Technology (MOST 103-2320-B-562 -001 -MY3 and MOST 106-2320-B-562-001). This study had been approved by the Institutional Review Board (IRB) of Saint Mary's Hospital, Lundong (IRB104011), and informed consent was obtained from all participants.

CO N FLI C T O F I NTE R E S T
No conflict of interest exists.