To determine and compare the extent of contamination caused by antimicrobial-resistant lactic acid bacteria (LAB) in imported and domestic natural cheeses on the Japanese market, LAB were isolated using deMan, Rogosa and Sharpe (MRS) agar and MRS agar supplemented with six antimicrobials. From 38 imported and 24 Japanese cheeses, 409 LAB isolates were obtained and their antimicrobial resistance was tested. The percentage of LAB resistant to dihydrostreptomycin, erythromycin, and/or oxytetracycline isolated from imported cheeses (42.1%) was significantly higher than that of LAB resistant to dihydrostreptomycin or oxytetracycline from cheeses produced in Japan (16.7%; P = 0.04). Antimicrobial resistance genes were detected in Enterococcus faecalis (tetL, tetM, and ermB; tetL and ermB; tetM) E. faecium (tetM), Lactococcus lactis (tetS), Lactobacillus (Lb.), casei/paracasei (tetM or tetW), and Lb. rhamnosus (ermB) isolated from seven imported cheeses. Moreover, these E. faecalis isolates were able to transfer antimicrobial resistance gene(s). Although antimicrobial resistance genes were not detected in any LAB isolates from Japanese cheeses, Lb. casei/paracasei and Lb. coryniformis isolates from a Japanese farm-made cheese were resistant to oxytetracycline (minimal inhibitory concentration [MIC], 32 µg/mL). Leuconostoc isolates from three Japanese farm-made cheeses were also resistant to dihydrostreptomycin (MIC, 32 to > 512 µg/mL). In conclusion, the present study demonstrated contamination with antimicrobial-resistant LAB in imported and Japanese farm-made cheeses on the Japanese market, but not in Japanese commercial cheeses.
lactic acid bacteria
minimal inhibitory concentration
deMan, Rogosa and Sharpe
The influence of the use of antimicrobial agents and the development of antimicrobial resistance in food-producing animals are of interest to the World Health Organization. Antimicrobial-resistant bacteria in food-producing animals can spread to humans, mainly through those foods, and cause intestinal and urinary tract infections in them. In addition, antimicrobial resistance genes can be transferred from commensal bacteria to other pathogenic bacteria [1, 2].
Contamination with antimicrobial-resistant bacteria in meat products has previously been investigated [3-6]. However, when meat products are contaminated, bacteria are generally heat-inactivated before consumption. On the other hand, some foods that are eaten raw contain indigenous microbial flora or are contaminated by bacteria. Fermented foods, such as natural cheese, generally contain an abundance of live LAB. We willingly consume LAB in fermented food or as medicine because of their multiple potential health benefits .
In Japan, the amount of natural cheese for consumption (not including primary materials for processed cheese) doubled between 1990 (77.5 thousand tons) and 2011 (160.8 thousand tons) . The Food Sanitation Act (Act No. 233, 1947) mandated heat treatment (63°C for 30 min or more) as part of the manufacturing process for milk to be drunk; however, there is no such provision for natural cheese. It is presumed that, to ensure hygiene and quality, natural cheese is produced from pasteurized milk in Japan. In some European countries, some fermented foods are produced using environmental bacteria and not an industrial starter: these foods carry the official Protected Designation of Origin. Nearly 90% of the natural cheese for consumption in Japan is imported . Recently, several reports from other countries have shown that some fermented milk products contain LAB with antimicrobial resistance genes [9-12]. However, the presence and extent of contamination with antimicrobial-resistant LAB of natural cheeses on the Japanese market is unknown.
In the present study, we isolated LAB from natural cheeses to compare the extent of contamination with antimicrobial-resistant LAB between imported and Japanese cheeses. In addition, we characterized the phenotypic and molecular characteristics of antimicrobial resistance of LAB isolates.
MATERIALS AND METHODS
Between 2007 and 2009, 38 imported natural cheeses (Brie, 9 samples; Camembert, 6; mozzarella, 3; cream cheese, 3; and others, 17) produced in 10 countries and 24 natural cheeses (Camembert, 7 samples; Gouda, 7; and others, 10) produced in Japan were purchased from several supermarkets and retail stores in Sapporo, Japan (Table 1). Based on the products' labels, the Japanese cheeses were divided into commercial cheeses (dairy companies, 8 samples; dairy cooperative in a limited area, 1) and farm-made cheeses (15 samples). Different types of natural cheeses from each brand were chosen. The samples were collected aseptically from the inner regions of the cheeses.
|Sample type||No. of samples|
Lactic acid bacteria were isolated by either of the following two methods:
- 10- and 105-fold dilutions of 33 samples were prepared in sterile physiological saline, after which 2 mL of the diluted solutions were combined with 18 mL of MRS agar (Oxoid, Hampshire, UK) kept at approximately 50°C. Each petri dish of combined agar and sample was cooled to room temperature. In addition, 50 µL of tenfold dilutions were inoculated onto MRS agar supplemented with one of the following antimicrobial agents (Sigma–Aldrich Japan, Tokyo, Japan): ampicillin, 4 µg/mL; dihydrostreptomycin, 128 µg/mL; chloramphenicol, 16 µg/mL; erythromycin, 1 µg/mL; oxytetracycline, 8 µg/mL; or enrofloxacin, 4 µg/mL. The isolation media were incubated at 32°C for 48 hr (Method I).
- For the remaining 29 samples, 1 g of each sample was inoculated into 10 mL of MRS broth (Oxoid), mixed, and incubated at 32°C for 48 hr. Enrichment cultures were streaked on MRS agar alone and MRS agar supplemented with the same six antimicrobials as described above (Method II).
Some isolates were subcultured on MRS agar or brain–heart infusion agar (Oxoid). Presumptive LAB isolates, namely those which were catalase-negative, gram-positive, coccoid- or rod-shaped, and produced acid utilizing the glucose in the sugar fermentation test , were selected for further testing. The isolates were stored in a Microbank (Pro-Lab Diagnostics, Richmond Hill, Canada) or in 10% skim milk at −80°C.
The DNA of LAB isolates was extracted from cultures with an InstaGene Matrix (Bio-Rad Laboratories, Tokyo, Japan). For rod-shaped isolates, Lactobacillus genus-specific PCR was performed . All coccoid-shaped isolates were tested by PCR to detect the Leuconostoc genus , Streptococcus genus , Lc. lactis , E. faecalis  and E. faecium . Some coccoid-shaped isolates that could not be identified by PCR were tested for growth and esculin hydrolysis by cultivation on Enterococcosel Agar (Becton, Dickinson, Sparks, MD, USA) and were grown at 45°C. Raffinose and sucrose fermentation tests  and API 20 STREP (SYSMEX bioMérieux, Tokyo, Japan) were also used to identify the isolates.
The MICs of the six antimicrobials described above for LAB isolates were determined using the agar dilution method with Mueller–Hinton agar (Oxoid), according to Clinical and Laboratory Standards Institute guidelines , except that the plates were incubated at 32°C for 48 hr.
One resistant isolate from each sample was selected for the following tests. In addition, isolates whose MICs against a particular antimicrobial agent differed, and resistant isolates from a single sample that had been identified as different bacterial species, were also selected for the following tests. The genes tetK , tetL , tetM , tetS , tetW , ermA , ermB , strA/B , aphA1  and aphA2  were tested for by PCR. Direct sequencing of 16S rRNA genes was performed for antimicrobial-resistant Lactobacillus, Leuconostoc and Streptococcus isolates. The BLAST program was used to compare DNA alignments with data from the DNA Data Bank of Japan .
Transferability of antimicrobial resistance genes was examined by filter mating. For recipient strains, enrofloxacin- or rifampicin-resistant mutants were generated from two E. faecalis isolates (susceptible to oxytetracycline and erythromycin) obtained from fecal samples of zoo animals and the E. faecalis ATCC 29212 used in this study. The MICs for these recipient strains were 128 or 512 µg/mL, respectively, to enrofloxacin or > 512 µg/mL to rifampicin. PCR was used to confirm that the parent and recipient strains did not harbor any resistance gene tested in this study. Donor and recipient strains were separately cultured in brain–heart infusion broth (Nissui Pharmaceutical, Tokyo, Japan). After incubation at 37°C for 16–18 hr, they were mixed at a donor:recipient ratio of 1:9, 9:1, 1:1 or 1:4. These 5 mL mixtures were filtered with 0.45 µm filters (Nihon Millipore, Tokyo, Japan). The filters were put on brain–heart infusion agar and incubated at 37°C for 24 hr. Each filter was then washed with sterile saline to yield bacterial suspensions. In order to select transconjugants, the bacterial suspensions were inoculated onto Mueller–Hinton agar supplemented with antimicrobials according to the resistance patterns of the recipient strain (enrofloxacin [16 µg/mL] or rifampicin [32 µg/mL]) and donor strain (oxytetracycline [16 µg/mL] or erythromycin [16 µg/mL]). The number of transconjugants per donor was taken as the conjugation frequency. The MICs and antimicrobial resistance genes of transconjugants were determined.
The percentages of antimicrobial-resistant LAB isolates were compared for different sample types and isolation methods using the X2 test. When at least one expected frequency was < 5, Fisher's exact test was used for comparisons between two groups.
By Method I, 169 LAB isolates were obtained from 28/33 samples (84.8%) and by Method II 240 LAB isolates were obtained from 23/29 samples (79.3%) (Table 2). LAB isolates were obtained from 31/38 imported samples (81.6%; 290 isolates) and 20/24 Japanese samples (83.3%; 119 isolates) (Table 2).
|Genus||Method I†||Method II‡|
|Imported (n = 19)||Japanese (n = 14)||Imported (n = 19)||Japanese (n = 10)|
|Lactobacillus||80 (14 [73.7%])§||26 (6 [42.9%])||58 (6 [31.6%])||10 (2 [20%])|
|Lactococcus||10 (4 [21.1%])||22 (9 ([64.3%])||12 (3 [15.8%])||20 (5 [50%])|
|Enterococcus||17 (4 [21.1%])||0||92 (8 [42.1%])||0|
|Leuconostoc||0||2 (2 [14.3%])||8 (4 [21.1%])||28 (7 [70%])|
|Streptococcus||0||11 (3 [21.4%])||8 (3 [15.8%])||0|
|Unidentified||1 (1 [5.3%])||0||4 (2 [10.5%])||0|
|Total||108 (16 [84.2%])||61 (12 [85.7%])||182 (15 [78.9%])||58 (8 [80%])|
Using MRS agar supplemented with antimicrobials, 205 and 45 isolates were obtained from imported and Japanese samples, respectively. Among the presumed resistant isolates grown on agar medium with antimicrobials, those presumed to be resistant to dihydrostreptomycin (70 isolates from 23 imported samples and 22 isolates from 10 Japanese samples), erythromycin (67 isolates from 21 imported samples and 7 isolates from 4 Japanese samples), or enrofloxacin (38 isolates from 16 imported samples and 12 isolates from 6 Japanese samples) were predominant.
Lactic acid bacteria isolates from each of 16 samples of imported cheeses and 8 of Japanese cheeses were classified as one genus. LAB isolates in two genera (12 imported cheeses and 10 Japanese cheeses) or three genera (3 imported cheeses, and 2 Japanese cheeses) were obtained from each of the remaining samples.
Enterococcus faecium (74 isolates from 10 samples), E. faecalis (19 isolates from 4 samples) and E. durans (16 isolates from a sample) were obtained only from imported samples. From three imported samples, only E. faecium was isolated. Both E. faecium and E. faecalis were isolated from three samples. Lactobacillus and Lc. lactis isolates were obtained from both imported and Japanese cheeses, independent of the isolation method (Table 2). Five coccoid-shaped isolates from imported samples could not be identified by PCR.
The MIC90 to ampicillin, chloramphenicol or enrofloxacin was the same or double the MIC50 for isolates in each bacterial genus. That is, the MICs showed a single distribution. Although, the MIC90 values were 4-, 8-, or 16-fold the MIC50 for Lactobacillus (ampicillin and enrofloxacin), Leuconostoc (enrofloxacin), Enterococcus (chloramphenicol and enrofloxacin) and Streptococcus isolates (ampicillin and enrofloxacin), the MIC distribution to these antimicrobials was wide and gradual without obvious bimodal peaks (data not shown). Therefore, the breakpoints of these antimicrobial agents could not be defined. Thus, based on the MICs, it was determined that the 58 LAB isolates that grew on agar medium with one of the three antimicrobials were not resistant.
Minimum inhibitory concentrations to oxytetracycline, erythromycin, and dihydrostreptomycin exhibited a bimodal distribution for each bacterial genus. In this study, an intermediate MIC of the bimodal distribution for isolates in each genus was defined as the breakpoint. The number of resistant isolates for each bacterial genus, sample type and isolation method, and breakpoints for resistance are shown in Table 3. Either the concentrations of the antimicrobials in the isolation media were lower than that of the breakpoints, or the MICs for some isolates were much lower than the concentrations of the antimicrobial agents in the isolation medium. Therefore, 186 LAB isolates (including the 58 isolates mentioned above) developed on agar media with antimicrobial agents were determined not to be resistant to any antimicrobials. However, 20 LAB isolates obtained using only MRS agar were determined to be resistant based on their MICs.
|Antimicrobial agents||Genus||Breakpoint (µg/mL)||Method I†||Method II|
|DSM||Lactobacillus||128||2 (1 [5.3%])||All susceptible||All susceptible||All susceptible|
|Lactococcus||128||1 (1 [5.3%])||All susceptible||All susceptible||All susceptible|
|Enterococcus||256||5 (1 [5.3%])||NA||2 (1 [5.3%])||NA|
|Leuconostoc||32||NA§||All susceptible||All susceptible||7 (3 [30%])|
|Streptococcus||32||NA||All susceptible||6 (2 [10.5%])||NA|
|EM¶||Lactobacillus||16||2 (1 [5.3%])||All susceptible||All susceptible||All susceptible|
|Lactococcus||4||3 (2 [10.5%])||All susceptible||All susceptible||All susceptible|
|Enterococcus||8||4 (1 [5.3%])||NA||2 (1 [5.3%])||NA|
|OTC¶||Lactobacillus||32||17 (4 [21.1%])||4 [1 (7.1%)]||29 (3 [15.8%])||All susceptible|
|Lactococcus||32||All susceptible||All susceptible||4 (1 [5.3%])||All susceptible|
|Enterococcus||16||6 (2 [10.5%])||NA||5 (1 [5.3%])||NA|
|One or more antimicrobials||Lactobacillus||—||19 (5 [26.3%])||4 (1 [7.1%])||29 (3 [15.8%])||All susceptible|
|Lactococcus||—||3 (2 [10.5%])||All susceptible||4 (1 ([.3%])||All susceptible|
|Enterococcus||—||7 (2 [10.5%])||NA||5 (1 [5.3%])||NA|
|Leuconostoc||—||NA||All susceptible||All susceptible||7 (3 [30%])|
|Streptococcus||—||NA||All susceptible||6 (2 [10.5%])||NA|
|Subtotal||—||29 (9 [47.4%])||4 (1 [7.1%])||44 (7 [36.8%])||7 (3 [30%])|
Our results show that 73 isolates from 16 imported cheeses (16/38, 42.1%) had a higher MIC than the breakpoint of one or more antimicrobials, and 11 isolates from 4 farm-made cheeses produced in Japan (4/24, 16.7%) showed higher MICs than the breakpoints of dihydrostreptomycin or oxytetracycline. These percentages differed significantly between imported and Japanese cheeses (P = 0.04). A significantly higher percentage of antimicrobial-resistant LAB isolates was also obtained by Method I from imported samples (9/19, 47.4%) than from Japanese cheeses (1/14, 7.1%; P = 0.02). However, there were no significant differences between imported samples and Japanese farm-made samples according to Method II (36.8% [7/19] vs. 30% [3/10]; P = 1] or between the two isolation methods used for Japanese (7.1% [1/14] vs. 30% [3/10]; P = 0.27) and imported samples (47.4% [9/19] vs. 36.8% [7/19]; P = 0.74), respectively.
For molecular analysis, 31 LAB isolates were selected based on the bacterial species and MICs (Table 4). The tetM, tetL, tetS, tetW and ermB genes were detected in eight LAB isolates obtained from seven imported samples (Table 4). Transferability of the antimicrobial resistance gene of three E. faecalis isolates was tested. E. faecalis (LAB-07-6) transferred tetL and ermB to enrofloxacin- or rifampicin-resistant recipients, which originated from two E. faecalis isolates from fecal samples, at a frequency of 1.0 × 10−6 to 7.0 × 10−7 but not to an ATCC strain-originated recipient. LAB-09-81 transferred tetM to a recipient at a frequency of 1.3 × 10−7 to 7.0 × 10−9. LAB-09-95 transferred tetM to recipients originated from two E. faecalis isolates at a frequency of 4.3 × 10−7 to 1.0 × 10−9 and 2.0 × 10−9.
|Sample no.||Country||Isolation||Strain no.||Bacterial species||MIC (µg/mL)||Antimicrobial resistance gene|
|07-3||France||I||CP||LAB-07-6||E. faecalis||>512||>512||128||0.5||128||1||tetL, ermB|
|09-11||Germany||II||OTC||LAB-09-95||E. faecalis||>512||>512||512||1||16||0.25||tetL, tetM, ermB|
|07-15||Switzerland||I||DSM||LAB-07-84||Lb. casei/paracasei||32||≤ 0.03||32||2||8||1||—|
Some Lactobacillus isolates could not be completely identified based on the DNA sequence of their 16S rRNA gene. Two species of Lactobacillus are described in Table 4. The similarity of 16S rRNA sequences among three Streptococcus isolates and those of Streptococcus thermophilus ATCC 19258, which s a known starter LAB of natural cheese, was lower than those of S. equinus ATCC 9812, S. bovis ATCC 27960, and S. infantarius CIP 106107. These three Streptococcus isolates could not be identified completely.
The percentage of antimicrobial-resistant LAB isolated from imported samples was significantly higher than that from Japanese cheeses both after combining results of Methods I and II and according to results of Method I alone. In addition, we detected antimicrobial resistance genes only in LAB from imported cheeses. Thus the focus in Japan should be on imported natural cheeses as a source of antimicrobial resistance genes, rather than on natural cheeses produced in Japan.
According to results of Method II, the percentage of antimicrobial-resistant LAB isolated from Japanese samples, which included only farm-made cheeses, was similar to that from imported samples. Because we tested different samples by the two methods, we could not identify the cause of the discrepancy between the results of Methods I and II. However, there was no significant difference between the two methods in the percentage of antimicrobial-resistant LAB isolated from each sample type (imported vs. Japanese). We presumed that the different rates of contamination between Japanese farm-made cheeses (30%) and Japanese commercial cheeses (7.1%; P = 0.27) were caused by different production methods rather than by different isolation methods.
Enterococcus faecalis isolates from imported cheeses could transfer tetM or both tetL and ermB to E. faecalis as a recipient; thus, natural cheese can be an origin of antimicrobial resistance genes. Both commensal and pathogenic bacteria reportedly harbor these antimicrobial resistance genes [2, 24]. Enterococci are not commonly used as starter LAB for natural cheese , but they exist in the natural environment and the gastrointestinal tracts of humans and animals. Antimicrobial-resistant enterococci have also been isolated from pasteurized milk cheeses produced in France . Antimicrobial-resistant enterococci in cheeses might originate from the environment in which the cheese is manufactured or from raw milk.
The tetM, tetW, tetS, or ermB genes were detected in Lb. casei/paracasei (LAB-07-48) from sample 07-8 (made in Italy), Lb. casei/paracasei (LAB-09-61) from sample 09-8 (made in Italy), Lc. lactis (LAB-09-142) from sample 09-20 (made in France) and Lb. rhamnosus (LAB-07-103) from 07-17 (made in Denmark). These LAB isolates with antimicrobial resistance genes were obtained using only agar medium with oxytetracycline or erythromycin. Moreover, we isolated Lc. lactis and Lactobacillus isolates with low MICs to oxytetracycline (0.5–8 µg/mL) and Lactobacillus isolates with low MICs to erythromycin (≤ 0.03–0.125 µg/mL) from these three samples by using MRS agar without an antimicrobial agent (data not shown). These isolates with antimicrobial resistance genes were not predominant in any cheese. In addition, E. faecium (07-91), Lb. casei/paracasei (07-48) and Lb. rhamnosus (07-103) were also tested for transferability of the antimicrobial resistance gene to recipient strains in E. faecalis, E. faecium and Lb. casei/paracasei, which were generated from field strains. However, we were unable to obtain a transconjugant (data not shown). Although E. faecalis recipient strains could receive antimicrobial resistance gene(s) from three E. faecalis isolates, we did not assess the ability of recipients in other bacterial species to receive genes from donors. Therefore, we could not conclude transferability of antimicrobial resistance genes in LAB isolates other than E. faecalis.
Although we did not detect antimicrobial resistance genes in LAB isolates from Japanese samples, we did obtain antimicrobial-resistant Lactobacillus and Leuconostoc isolates. MICs to oxytetracycline (32 µg/mL) of the Lactobacillus isolates were higher than those of other isolates (0.5–16 µg/mL; mode, 2 µg/mL). MICs of resistant Leuconostoc isolates to dihydrostreptomycin (32 to > 512 µg/mL) were much higher than those of other isolates obtained from the same samples (0.5–8 µg/mL).
Only 64 of the 250 isolates we obtained using MRS agar supplemented with an antimicrobial were resistant to one or more antimicrobials, based on their MICs. LAB include multiple bacterial genera and antimicrobial susceptibilities and breakpoints differed between genera. Therefore, some susceptible isolates can also develop on agar medium supplemented with an antimicrobial agent whose concentration is lower than the breakpoint. In order to monitor contamination of fermented food with antimicrobial-resistant LAB, a more efficient method of detection is needed. We subcultured some presumed resistant isolates identified by using MRS agar with erythromycin on the same agar medium, but they did not grow on it. Therefore, we postulated that cheese inoculated into agar medium can inhibit antimicrobial activities. In Method II, we added enrichment culture to decrease the amount of cheese we needed to inoculate on the selective agar medium; however, many isolates identified using agar medium supplemented with an antimicrobial agent were not resistant according to Method II.
In conclusion, we detected antimicrobial resistance among LAB isolates in natural cheeses on the Japanese market. The extent of contamination with antimicrobial-resistant LAB was lower in Japanese cheeses than in imported cheeses. In particular, we did not detect antimicrobial-resistant LAB in Japanese commercial cheeses. Some LAB isolates were able to transfer resistance gene(s) to the same species in vitro. The present study demonstrates that some imported natural cheeses contain organisms with antimicrobial resistance genes that can be transferred to commensal and pathogenic bacteria. It should be verified that bacterial strains for starter cultures do not contain transferable antimicrobial resistance genes. Moreover, stricter hygiene control on cheese production, especially small-scale cheese manufacturing, is needed to prevent contamination with antimicrobial-resistant LAB.
The authors would like to thank Dr. Hidetomo Iwano of Rakuno Gakuen University for his useful advice and technical support. This work was partially supported by a Grant-in-Aid to Cooperative Research from Rakuno Gakuen University and Rakuno Gakuen University Dairy Science Institute, 2011-6.
The authors declare that they have no conflicts of interest.