Staphylococcal food poisoning (SFP) is an important foodborne disease worldwide, and milk and milk products are commonly associated with SFP outbreaks. The objectives of this study were to investigate the distribution of staphylococcal enterotoxin (se) genes in Staphylococcus aureus from raw cow's milk and milk products and to assess their genetic background with the spa typing method. Of the 549 samples (297 bulk milk and 162 milk product samples) collected from Tigray region, Northern Ethiopia, 160 (29.1%) were positive for S. aureus, of which 82 (51%) were found to harbor se genes by a modified multiplex PCR. Nine se genes were identified: sea (n =12), seb (n =3), sec (n =3), sed (n =4), seg (n =49), seh (n =2), sei (n = 40), sej (n =1), and tsst-1 (n =24). The classical type of genes accounted for 27%. Of the 82 enterotoxigenic isolates, 41.5 and 12.4% harbored two or more se genes, respectively. The highest gene association was observed between sei and seg, whereas sea and seb were always found together with the new types of se genes. Altogether, 18 genotypes of toxin genes were identified, and 33% of the samples contained >5 log CFU ml−1 S. aureus. spa typing identified 22 spa types and three novel spa sequences, which showed the high genetic diversity of the isolates. No apparent relationship was observed between spa type and se genes. Of the 25 spa types, 13 (52%) were from raw milk, 3 (12%) from milk products, and 9 (36%) from both types of sample. Types t314 (20.7%, n = 17), t458 (18.3%, n = 15), and t6218 (9.8%, n = 8) were the most common spa types identified and were widely distributed in three of the eight studied localities. This is the first study from the Tigray region to report the high distribution of enterotoxigenic S. aureus with a diversified genetic background from dairy food. The study may provide valuable data for microbial food safety risk assessment, molecular epidemiology, and phylogenetic studies of S. aureus in Ethiopia.
Staphylococcal food poisoning (SFP) is one of the most important foodborne diseases worldwide (16, 46). It is caused by consumption of food that contains one or more types of staphylococcal enterotoxins (SEs) produced by enterotoxigenic strains of Staphylococcus aureus. One nanogram of SE per gram of contaminated food can cause SFP symptoms (46); however, outbreaks have been observed at lower concentrations (0.5 ng/ml in milk) (14). The disease is characterized by a short incubation period (an average of 4.4 h), nausea, violent vomiting, abdominal cramps, headache, and diarrhea. Although SFP is usually a self-limiting illness, death occasionally occurs, with case fatality rates ranging from 0.03% for the general public to 4.4% for more susceptible populations, such as children and the elderly. Death results from severe dehydration and electrolyte imbalance (17). Although the mortality from SFP is low (9), the actual impact of the intoxication is large, owing to loss of working days and productivity, hospital expenses, and economic loss for restaurants and food industries (23, 36).
Milk and milk products are commonly associated with SFP (11). Outbreaks caused by consumption of contaminated milk and milk products include an extensive SFP outbreak that affected more than 13,000 persons in Japan (5) and a recent outbreak of SFP at a Swiss boarding school that affected 14 students (20).
Humans and animals are reservoirs of S. aureus on their skin and mucosal membranes (16). It is reported that 10 to 35% and 20 to 75% of humans are persistent and intermittent carriers of S. aureus, respectively (27). Cows with mastitis are also a common source of S. aureus in raw milk (23). S. aureus can also be introduced to food by contaminated equipment used in food processing. Furthermore, inadequate refrigeration, advance preparation of food, prolonged use of warming plates when serving foods, and poor personal hygiene are favorable conditions that promote staphylococci growth and the production of SEs in food (46). As a rule of thumb, enterotoxigenic S. aureus must grow to a population greater than 5 log/g of contaminated food before sufficient SEs are produced to result in food intoxication (54).
SEs are small proteins that range in size from 22 to 28 kDa. They are highly stable and resistant to many gastric proteolytic enzymes, such as pepsin or trypsin. They are soluble in water and saline solution. They are also highly resistant to heat (e.g., 127°C for 15 min), freezing, and drying (16, 30). The heat stability of SEs poses a challenge in processed food because even if bacteria have been killed, toxins may still remain (6). To date, 22 SEs and SEIs (SEs that lack emetic activity or that have not been tested), excluding molecular variants, have been identified (4). The classical SEs include SEA, SEB, SEC (SEC1, SEC2, SEC3, SEC ovine, and SEC bovine variants), SED, and SEE. The involvement of these toxins in SFP has been clearly elucidated, and there are distinct serological types. The new types of SE and SEII include SEG, SEH, SEI, SER, SES, SET and SEIJ, SEIK, SEIL SEIM, SEIN, SEIO, SEIP, SEIQ, SEIU, SEIU2, and SEIV. The toxin formerly designated as SEF was renamed toxic shock syndrome toxin (TSST) because it lacks emetic activity (7). Studies have shown that 57 to 72% of S. aureus food isolates harbored the classical and/or new SE genes (1, 43).
Molecular typing of pathogens such as S. aureus is important for two main reasons. First, knowledge of genetic microvariation at the strain and lineage level is useful during outbreak investigations for tracing the source of the pathogen and understanding its transmission. Second, genetic macrovariation is useful for phylogenic and population-based studies (37, 38). There are many molecular methods for typing pathogens. The choice of method depends on the purpose for which the typing will be used (15, 38). Although pulsed-field gel electrophoresis (PFGE) is the “gold standard” for S. aureus typing, sequence-based spa typing shows comparable sensitivity to PFGE and yet is rapid, easy to handle, and costs less (37). The spa typing method depends on sequencing of the polymorphic 24-bp variable-number tandem within the X region of the S. aureus–specific staphylococcal protein A. spa typing has been considered a frontline tool in epidemiological typing of S. aureus (50).
Because of current Ethiopian agricultural policy and a gradual increase in living standards, there is an expansion of small-scale dairy farms in and around the major cities of Ethiopia. The country has shown a 3% increase in annual milk production in the past decade compared with 1.6% in the previous two decades. The production of cow's milk in Ethiopia may increase by 93% over the next 5 years (47). However, cows on the majority of dairy farms commonly suffer from clinical and subclinical mastitis (2, 13, 53). In the country in general, and in the Tigray region in particular, milk processing methods are traditional and unhygienic (57), and 31.8% of farmers were reported to consume raw milk (32). A recent study conducted in the project area reported point prevalence of S. aureus at 38.7% from milk and milk products (52). However, the study did not describe the enterotoxigenic potential of the isolates. All these conditions call for further study on milk hygiene to protect consumers from milkborne pathogens such as S. aureus.
The objectives of this study were to investigate the distribution of enterotoxin genes in S. aureus isolates from milk and milk products in the Tigray region, Northern Ethiopia, by multiplex PCR and to characterize the genetic background of enterotoxigenic S. aureus using spa typing. This hazard identification and molecular typing work may provide useful data for microbial food safety risk assessment, epidemiological investigations, and phylogenetic studies of S. aureus in Ethiopia.
MATERIALS AND METHODS
Collected sample types and sizes. A total of 549 milk and milk product samples were collected from eight cities and towns and their vicinities in the Tigray region of Northern Ethiopia, from August 2012 to May 2014: Mekelle (n = 147), Shireendaselase (n = 139), Wukro (n = 60), Adigudome (n = 52), Hagreselame (n = 21), Maichew (n = 53), Adigrate (n = 43), and Abi-Adi (n = 34). The cities and towns were selected to include major cities and towns of the region, with relatively larger numbers of dairy farms, milk and milk product shops, and dairy cooperatives, and also to represent a variety of geographical locations (south, north, east, and west) and agroclimatic zones (highland, midlands, and lowlands). The sampling points included dairy farms, milk collection centers, milk and milk product shops, restaurants, and cafeterias. Samples were collected by the previously reported procedure (52). The samples comprised raw bulk milk (n = 297), buttermilk (n = 64), butter (n = 58), sour milk (spontaneously fermented milk) (n =97), Ethiopian cottage cheese (n =14), cheese (n = 15), and cake made from milk (n = 4).
Isolation, enumeration, and molecular identification. The isolation, enumeration, and molecular identification of the isolates were performed by the methods detailed previously (52). Briefly, Baird-Parker agar (Oxoid, Basingstoke, England) supplemented with egg-yolk tellurite (Merck, Darmstadt, Germany) was used for isolation and enumeration. Conventional biochemical tests were used for identification; however, final confirmation to species level was performed by sequencing the 16S rRNA gene (10). DNA was extracted using the GenElute Bacterial Genomic DNA kit (Sigma-Aldrich, Munich, Germany) as per the manufacturer's instructions.
Multiplex PCR for detection of enterotoxin genes. A modified multiplex PCR was used for detection of enterotoxin genes from the S. aureus isolates (31). The method tests nine enterotoxin genes (sea, seb, sec, sed, see, seg, seh, sei, sej), tsst-1, and 16S rRNA of the isolates in two independent multiplex PCR reactions mixtures. Primers for sed, see, seg, sei, and tsst-1 were combined in reaction mixture 1 and primers for sea, seb, sec, seh, sej, and 16S rRNA were combined in reaction mixture 2. The primers used are listed in Table 1. They were supplied by Invitrogen Life Technologies (Carlsbad, CA).
Each multiplex PCR reaction was performed with a final reaction volume of 50 μl. It was composed of 45 μl of reaction mixture containing a final concentration of 1× AmpliTaq buffer, 4 mM MgCl2, 2 U of AmpliTaq Gold polymerase (all from Applied Biosystems, Foster City, CA), 400 μM deoxynucleoside triphosphates (dNTPs; New England Biolabs, Beverly, MA), and 300 nM each SE primer and 60 nM 16S rRNA primers. Finally, 5 μl of DNA (10 ng/μl) was added to the mixture.
PCR amplification was performed in a C1000 Thermal Cycler (Bio-Rad Laboratories, Hercules, CA), which was adjusted to initial denaturation at 95°C for 10 min, followed by 15 cycles of 95°C for 1 min, 68°C for 45 s, 72°C for 1 min, 20 cycles of 95°C for 1 min, 64°C for 45 s, 72°C for 1 min, and a final extension at 72°C for 10 min. Ten microliters of the PCR product was resolved by electrophoresis in a 2.5% agarose gel at 100 V for 100 min. For comparison, the GeneRuler 50-bp DNA ladder (Thermo Scientific, Waltham, MA) was used. The DNA products were visualized on the UV transilluminator Gel Doc documentation system (Bio-Rad Laboratories).
Control strains. Six S. aureus strains obtained from the European Union reference laboratory for coagulase-positive staphylococci (ANSES, Maison Alfort, France) (Table 2) were used as positive control and Milli-Q water (Millipore Corporation, Bedford, MA) as a negative control.
spa typing of enterotoxigenic S. aureus isolates. Amplification of region X of the spa (protein A) gene was performed using primers spa-1113f (5′-AAAGACGATCCTTCGGTGAGC-3′) and spa-1514r (5′-CAGCAGTAGTGCCGTTTGCTT-3′) (42). PCR was performed in a 40-μl final volume containing 1× HF buffer, 0.4 μM each primer, 200 μM dNTPs, and 0.02 U/μl of iProof Taq polymerase (Bio-Rad Laboratories). The PCR reaction for amplification of DNA included initial denaturation at 98°C for 30 s, followed by 35 cycles of denaturation at 98°C for 15 s, annealing at 60°C for 30 s, and elongation at 72°C for 20 s. Final elongation was performed at 72°C for 10 min. The PCR was purified using 1× volume of Agencourt AMPure XP beads (Beckman Coulter, Inc., Brea, CA), according to the manufacturer's instructions, and was submitted to GATC Biotech AG (Konstanz, Germany) for sequencing. Sequences were processed using Geneious V7 (24), and the spa types were determined using DNAgear (3). Minimum spanning trees were prepared using BioNumerix software as described by Roussel et al. (45).
Data analysis. The chi-square test from Epi Info version 18.104.22.168 (Centers for Disease Control and Prevention, Atlanta, GA) was used to explore statistically significant differences in the distribution of enterotoxigenic S. aureus among raw milk and milk products, as well as among different localities. P values <0.05 were considered significant.
Distribution of S. aureus in milk and milk products. Of the total 549 milk and milk product samples, 160 (29.1%) were shown to contain S. aureus.
Detection of SE-encoding genes by multiplex PCR. Among the 160 S. aureus dairy isolates, 82 (51%) were found to harbor staphylococcus enterotoxin (se) genes. Nine types of se genes were identified: sea (n = 12), seb (n = 3), sec (n = 3), sed (n = 4), seg (n = 49), seh (n = 2), sei (n = 40), sej (n =1), and tsst-1 (n =24). The gene identified most frequently was seg, followed by sei and tsst-1. The least frequently identified were seh and sej, and see was not detected. Among the 82 enterotoxigenic strains, 22 (26.8%) isolates harbored the classical se genes (sea, seb, sec, and sed) either alone or in combination with the newly identified se genes (seg, seh, sei, and sej).
Among the 82 enterotoxigenic strains, 38 (46.3%) harbored one type, 34 (41.5%) two types, and 10 (12.2%) more than two types of enterotoxin genes. Accordingly, from all of the enterotoxigenic strains, 18 se genotypes were identified. The highest gene association was observed between sei and seg (Table 3).
SE-encoding genes and types of sample. Among the 82 enterotoxigenic S. aureus strains, 57 (69.5%) were identified from raw bulk milk and 25 (30.5%) from milk products. Chi-square analysis indicated a statistically significant difference in the prevalence of enterotoxigenic S. aureus between raw bulk milk and milk products (P < 0.05). Odds ratio (OR = 1.85, 95% CI 0.9706 to 3.548) analysis showed that raw milk is 1.8 times more likely than a milk product to contain enterotoxigenic S. aureus. Table 3 shows the total number of specimens collected and the type and number of enterotoxigenic genes identified.
SE-encoding genes and the level and count of S. aureus in the food. Of the 82 milk and milk product samples that were positive for one or more se genes, 32.9 and 67.1% of the original food samples contained >5 log CFU ml−1 and 3 to 5 log CFU ml−1 of S. aureus, respectively (Table 4). Of the 27 samples with >5 log CFU ml−1 of S. aureus, 18 were raw bulk milk.
Enterotoxigenic S. aureus strain distribution in different localities. The distribution of the different se genes in different localities of the project area is presented in Table 5. There is a statistically significant difference (P < 0.05) in the overall distribution of enterotoxigenic S. aureus among five of the sampling localities, with values higher in Adigudome and Wukro; Hagreselame, Abi-Adi, and Maichew were not included in the analysis because of underrepresentation.
spa typing. Twenty-two different spa types and three novel spa sequences (not present in the Ridom data base) were identified. Of the total 25 spa types, 13 (52%) were from raw milk, 3 (12%) from milk products, and 9 (36%) from both types of sample (Table 6). Only 36% of the identified spa types were found in both milk and milk product samples, whereas the remaining 64% were either in milk or in the milk product. Figure 1 shows the relatedness of the S. aureus isolates according to their spa types, colored according to their toxin gene profile. No clear clustering between spa type and toxin genes is observed. The sizes of the circles represent the number of isolates within the spa type, and the distances between the circles represent the genetic relatedness. The closer they are to each other, the greater their genetic relatedness. Types t458 (18.3%, n = 15), t314 (20.7%, n = 17), and t6218 (9.6%, n = 8) (shown as multisegmented large circles) were the most common spa types identified and were found to harbor 7, 6, and 4 different toxin profiles, respectively.
The aforementioned three most common spa types were frequently distributed in three of the eight studied localities (Table 6). The t314 spa type was more frequent in Shireendaselase (n = 12, 71%), t458 in Mekelle (n = 9, 60%), and t6218 in Wukro (n = 4, 50%).
The characterization of se genes from S. aureus isolated from dairy samples has not been previously documented in the Tigray region of Northern Ethiopia. However, a similar study conducted in Central Ethiopia reported a 25.8% prevalence of enterotoxigenic S. aureus in bulk milk samples (12), which is lower than in the current study. In Italy, 53% of the isolates from milk and milk products (8) and, in Norway, 52.5% of bovine bulk milk samples (22) were reported to harbor enterotoxigenic S. aureus, which are comparable to our observations. However, in Sweden, 70% of S. aureus isolates from cheese, made from raw milk, were found to harbor one or more enterotoxin genes (44).
Of the 82 milk and milk product samples that contained enterotoxigenic S. aureus, 33% had levels of S. aureus exceeding 5 log CFU ml−1, which is considered to be a critical level for toxin production (54). Moreover, 67% of samples contained 3 to 5 log CFU ml−1 S. aureus, which could also constitute a SFP risk if the food is exposed to ambient temperature during transport and storage, thus allowing the bacteria to grow. The expression of SE is highly regulated by growth phase, environmental conditions, and a complex network of regulatory genes, among which the accessary gene regulator plays a major role. The accessary gene regulator is activated by a quorum-sensing system when cell numbers reach a critical mass (46). Once the SE is produced in the food, it retains its biological activity because it is thermostable and resistant to low pH, freezing, and drying, conditions that can easily destroy the bacterial cells that produced the toxin (6, 16, 23).
In the present study, of the total number of identified enterotoxin genes, 27% were genes that encode classical SEs. It is reported elsewhere that approximately 95% of SFPs are caused by classical SEs (51). Among the classical SEs, SEA is the most common involved in SFP worldwide, followed by SED, SEB, and SEC (4, 19, 20, 25, 29). However, in terms of severity, SEB results in more intense symptoms of SFP than SEA.
Among all of the classical types of se genes, 81% were found in association with the newly identified se genes. In line with our observation, another study also reported the association of 64% of the classical se genes with the newly identified se genes (35).
In the current study, the newly identified se genes, including tsst-1, accounted for 73% of the total identified genes. Other researchers have also noted the occurrence of a significant number of the newly identified se genes in dairy products and other foodstuffs (8, 26, 33). From the new type of SE, SEH is reported to produce significant amount of SE that could lead to SFP (39). For example, SEH was responsible for SFP outbreaks in Norway after consumption of mashed potato made with added raw milk (21) and in Japan after consumption of reconstituted milk, in the latter case in combination with SEA (18).
The seg gene was found most frequently in association with sei. Similar to this finding, many researchers showed also that seg was found linked with sei (8, 26, 39) and that they were carried by a genomic island–borne structure called the enterotoxin gene cluster (egc). This genetic structure is reported to carry additional sem, sen, and seo genes (55); it has been hypothesized that the egc cluster may act as a reservoir and nursery for SE genes other than the aforementioned primary egc-related genes (4).
Differences in the distribution of se genes were observed among the five sampling localities (P < 0.05), which may be indicative of the spread of certain S. aureus types within a specific geographical area (21). The majority of the SE-encoding genes are carried on mobile genetic elements, such as plasmids and prophages, and are spread among isolates at high frequency by a horizontal gene transfer mechanism (34). This characteristic may enable the bacteria to modify their capacity to cause diseases and may contribute to their evolution (4).
The high prevalence and numbers of enterotoxigenic S. aureus in raw milk, compared with dairy products, may have two important practical implications. First, in areas where raw milk consumption is a common practice (32) there would be a high probability of SFP episodes. Second, although molecular epidemiological study may verify the assumption, one of the major sources of contamination to raw bulk milk could be mastitic cows. A cow with clinical or subclinical staphylococcal mastitis can excrete S. aureus in numbers up to log 8 CFU ml −1 (41).
The tsst-1 gene was found in 24 (29%) of 82 isolates. This toxin is responsible for toxic shock syndrome, which is characterized by high fever, a diffuse erythematous rash, desquamation of skin, hypotension, and involvement of three or more organ system failures (40). Multiplication of S. aureus in a localized infected area of the body produces TSST-1, which enters the vascular system and exerts these generalized symptoms (36). In findings similar to ours, a previous study reported a tsst gene prevalence of 25.6%, in combination with other enterotoxigenic genes, from milk of cows with mastitis (49).
spa typing was used to assess the genetic background of the enterotoxigenic S. aureus isolated from milk and dairy products. This molecular typing method is recommended for national and international surveillance, as well as for the analysis of short-term local epidemiological studies, because it is easy to conduct and it is easy to interpret and exchange the results (50). Besides, it was reported that spa typing performed better than multilocus enzyme electrophoresis, PFGE, and coa typing in degree of agreement with the microarray at various phylogenetic depths (28).
The spa typing identified 22 spa types and three novel spa sequences from the 82 enterotoxigenic S. aureus strains, thus revealing a wide genetic diversity of the isolates. The 18 se genotypes were found to be distributed evenly in all spa types without showing a specific pattern. This may indicate that the risk of SFP due to a specific se genotype having a specific genetic background may not be apparent.
Genetic similarity between the milk and milk product isolates, in terms of identical spa types, was observed in only nine (36%) spa types; this may suggest, among other things, that the sample groups may have little source of contamination in common. According to the spa typing results, the dominant genetic characteristic documented among the two sample groups was genetic diversity rather than genetic relatedness: they differed in 64% of the identified spa types. Moreover, higher intramilk spa type variation was also documented. And, hence, all these observations may indicate that there may be many sources of contaminations in the milk value chain. The X region of the spa gene, which is composed of many 24-bp variable tandem repeats, shows high polymorphism due to deletion, point mutation, and duplication of the repeats. Variation in type or number of the repeats generates different spa types (48). Types t314, t458, and t6218 were the most common spa types among the isolates harboring the majority of the se genes identified and were also more prevalent in three of the eight studied localities. The generation of such molecular typing information is of paramount important for population-based epidemiological, disease outbreak, and phylogenetic studies of S. aureus in the studied area. A recent study conducted in the United States reported that spa typing showed comparable performance with multilocus sequence typing and suggested it for use in macroepidemiology and evolutionary studies of S. aureus, given its lower implementation cost (38).
One reported limitation of the spa typing method is that it may designate 1 to 2% of strains as “nontypeable” if there is a rearrangement in the IgG region of the gene where the forward spa primer is located (56).
In conclusion, this study is the first report from the Tigray region of Northern Ethiopia that has documented the high occurrence of enterotoxigenic S. aureus having 18 genotypes in dairy isolates. Moreover, 22 spa types and three novel spa sequences were also identified. The high prevalence of enterotoxigenic S. aureus with diversified genetic background in milk and milk products may pose considerable SFP risk to consumers. Public health measures to reduce the risk include control of mastitis at the farm level, pasteurization of milk before consumption, and education of farmers, via the agricultural extension network, to maintain a hygienic environment during traditional milk processing as well as not to consume raw milk. Educating food handlers in the proper preparation and storage of food and the provision of cooling facilities in the milk value chain are necessary measures to reduce the risk of SFP. Using the output of this study as a springboard, molecular epidemiological studies aimed at tracing the source of enterotoxigenic S. aureus to facilitate control intervention could be useful. Other recommendations include research on production of SE in real food matrices to further characterize the risk and also basic research on the role of the newly identified SEs and SEIs in causation of SFP.
This work was financially supported by the academic collaboration project between Mekelle University (MU) and Norwegian University of Life Sciences (NMBU). The authors also thank technical staff and others at the College of Veterinary Medicine, MU, NMBU, and the Norwegian Veterinary Institute, for their assistance.