Isolation of Escherichia coli and Staphylococcus aureus in raw milk from refrigeration tanks: identification and antimicrobial resistance profiles

: Escherichia coli and Staphylococcus aureus in milk cooling tank reflects a hygienic deficit in animal management, production environment, and milk obtainment. With implications for public health as agents of infection and food poisoning, and the presence of antimicrobial-resistant strains. Therefore, were investigated in cooling tanks with high counts of somatic cells and total bacteria in milk. Microorganisms, in which a profile of resistance to antimicrobials was investigated, and whether there was a similarity in this profile between the strains of the eight dairy properties. Therefore, eighty-eight samples were obtained, and inoculated on Compact Dry ® plates. Of this total, 27.27% (24/88) samples tested positive for E. coli and 56.81% (50/88) for S. aureus . Among 24 E. coli strains subjected to disk-diffusion antibiograms, 70.83% were resistant to rifampicin, 50% to ampicillin and 41.67% to cefoxitin and erythromycin, while of the 51 S. aureus strains, 94.32% expressed resistance to azetroanam, 86.27% to ampicillin and nalidixic acid, 76.47% to rifampicin and 47.06 % to erythromycin and cefoxitin. A criterion of resistance to over three antibiotics was observed for 8.33% (2/24) of the isolated E. coli strains and 17.65% (9/51) of the S. aureus strains, characterizing them as multidrug resistant (MDR) strains. Resistance phenotypes displayed high similarity between properties F5 and F6 for S. aureus , and properties F6 and F8 for E. coli when applying the Jaccard index. The presence of these antibiotic-resistant pathogenic microorganisms indicate flaws in milk production handling and sanitary conditions, representing risk to milk consumers.

Isolation of Escherichia coli and Staphylococcus aureus in raw milk from refrigeration tanks: identification and antimicrobial resistance profiles

INTRODUCTION
The world's milk production was of 851.8 million tons in 2019 in milk equivalents, with an average consumption of 111.6 kg of milk equivalents per inhabitant (FAO, 2020). In Brazil, a total of 24.46 billion liters were produced in the same year, with an average of 734 million liters/year was obtained in the state of Mato Grosso alone, ranking 11 th among the most important milk producing states in the country (IBGE, 2019;BRASIL, 2019). The quality and/or quantity of produced milk are influenced by several factors, such as those linked to milk obtention, transportation and storage, or associated to the producing animals, such as genetic herd potential, management and feeding, while also being influenced by general animal health and that of their mammary glands (COSTA et al., 2019;ARAÚJO et al., 2017;GUIMARÃES et al., 2017). Clinical and subclinical mastitis are noteworthy among animal health alterations. Microorganisms of contagious and environmental origins, respectively, cause this disease (OSTERAS, 2018). Coagulase negative Staphylococcus, S. aureus, Escherichia coli, Streptococcus uberis and S. dysgalactae are frequently reported as bacterial clinical mastitis agents, and S. aureus and Corynebacterium bovis as the agents responsible for subclinical mastitis cases (BETTANIN et al., 2019;OSTERAS, 2018;BI et al., 2016).
Regarding mastitis diagnoses, mastitic milk indicators comprise Somatic Cell (SCC) and Total Bacteria (TBC) counts, with cohort limits established at 500,000 cells/cm 3 for SCC and 300,000 CFU/cm 3 for TBC, respectively (BRASIL, 2018a). These parameters are, therefore, considered milk hygiene and/or intramammary infection indicators worldwide by official milk control and animal health agencies (EUROPEAN UNION, 2017;BRASIL, 2018a;OSTERAS, 2018).
In addition to conventional microbiological analyses, quick analysis methods like Readycult™ -LMX (BELOTI et al., 2002), Rida Count [Coliforms R 1009] (VASALLO; REYES; CAMBAS, 2013), and 3M TM PetrifilmTM Staph (SOUZA et al., 2015) are routinely applied to microbiological milk analyses. Among these methods, 3M TM Petrifilm TM products are recognized and regulated for foods of animal origin by the Brazilian Ministry of Agriculture, Livestock and Supply-MAPA (BRASIL, 2005). Compact Dry ® system products, similar to 3M TM Petrifilm TM products, in these systems, chromogenic substrates are used, are also applied in the microbiological analyses of products of animal origin, such as milk (CASAROTTI et al., 2009).
Several antimicrobials are available for the treatment of mastitis, and failures in this process lead to increased bacterial resistance against usual antibiotics (KRÖMKER; LEIMBACH, 2017;KREWER et al., 2013;CAZOTO et al., 2011). In this regard, studies have been implemented to detect antimicrobial resistance and its mechanisms in several microorganisms, such as E. coli and S. aureus strains isolated from the milk production chain, classifying them as either resistant or multi-drug resistant (MDR) (MOREIRA et al., 2008;MORITZ;MORITZ, 2016). These strains, when present in milk, can lead to dairy products able to trigger difficult to treat toxinfections.
In 2013, Cáceres, a municipality located in the Central-South mesoregion of the state of Mato Grosso, was responsible for the production of 9710 liters of cow's milk/year (SOARES et al., 2017). In this context, the present study aimed to investigate cooling tanks in the dairy region of the municipality of Cáceres presenting high Somatic Cell (SCC), and Total Bacteria Counts (TBC) detected over six months concerning the occurrence of E. coli and S. aureus. The Compact Dry ® kit was employed to this end, and the antimicrobial resistance profiles and the similarities between strain resistance profiles, among the eight investigated milk-producing properties were verified for the isolated strains.

MATERIAL AND METHODS
This study was carried out in eight milk production properties located in the municipality of Cáceres, in the central-southern mesoregion of the state of Mato Grosso, Brazil. Raw milk presenting Somatic Cell Counts and Total Bacteria counts (SCC and TBC) higher than the maximum MAPA limit (BRASIL, 2018a) for six months (08/2018 to 01/2019) were analyzed. After 01/2019, weekly samples were collected from the cooling tanks with the aid of a sterile shell for eleven weeks, from 02 to 04/2019. Three hundred-mL aliquots of raw milk were places in a sterilized polyvinyl chloride bottle, which was then sealed, identified and maintained at 5 o C in an isothermal box and sent to the Food Molecular Microbiology Laboratory, belonging to the Faculty of Nutrition, Federal University of Mato Grosso, for bacteriological analyses. At the time of collection, milk temperature in the sampled cooling tanks of the eight assessed properties was determined with the aid of a calibrated dipstick thermometer (MINIPA, São Paulo, Brazil -[0 to 100° C]).
At the laboratory, following container asepsis, the samples were homogenized by inversion bottle movements, and a 1 mL-aliquot of the milk diluted in 9 mL of 0.1% Peptone Saline Solution (0.1% SSP), at 10 -1 and 10 -2 dilutions of the first 1 mL aliquot dilution, were inoculated onto EC-Compact Dry ® plates (CapLab, Brazil), and 1 mL of the 10 -2 dilution were inoculated on X-SA Compact Dry ® plates (CapLab, Brazil), which were then incubated at 35 °C for 24 hours. Reddish colonies were characterized as total coliforms, and light blue colonies, as E. coli (MIZUOCHI et al., 2016) on the EC-Compact Dry ® plates. Characteristic S. aureus colonies were dark blue (TERAMURA; MIZOUCHI; KODAKA, 2010) on the X-SA-Compact Dry ® plates, which were then counted, and the results expressed as Colony Forming Unit per mL (CFU/mL).
The disk diffusion technique was used to analyze in vitro antimicrobial susceptibility. E. coli and S. aureus strains representing property (F) and collection (C) were sown by swarming on Müller-Hinton agar, and disks containing 20 antimicrobial agents (Table 1) were distributed equidistantly on the plates, which were then incubated at 35° C for 16 to 20 hours, according to Bauer et al. (1966). The inhibition halos were then measured, and the results were compared with standards published by the Clinical and Laboratory Standards Institute (CLSI, 2018), classifying the strains as sensitive, intermediate or resistant. Strains were classified as Multidrug Resistant (MDR) when displayng resistance to three or more chemical classes among the tested antibiotics (MAGIORAKOS et al., 2012).
A survey on information adequacy was carried out concerning the requirements of normative instruction nº 77 (BRASIL, 2018a), through observations and in loco data records at the eight analyzed properties, with the aid of a structured checklist. A questionnaire containing the following topics: animal hygiene (pre-and-post dipping), milking utensil hygiene (milking buckets, milk gallons teacups), operator hygiene, cooling tank, and tank house hygiene, equipment temperature and resources for investigating mastitis and, finally, positive animal management and treatment (BRASIL, 2018a). The terms compliant, non-compliant, and not applicable were adopted to mark each observation in each assessed property.
To assess microbiological raw milk quality, the quality reference standard of <1 CFU/mL for Enterobacteriaceae  (2003), of 500 or 2.70 Log 10 CFU/mL was applied for S. aureus. Similarities were observed regarding strain resistance to the tested antibiotics (Table 1) among the eight evaluated properties. For this assessment, the antimicrobial susceptibility data were converted into a binary matrix, where 1 indicates resistance and 0 indicates absence of resistance or susceptibility. The data regarding the strains (E. coli and S. aureus) from each property (F1, ..., F8) were grouped, and the sets were ..) to verify phenotype similarities applying the Jaccard similarity coefficient, calculated using the PAST software 4.03 (https://softpedia.com/get/science-CAD/PAST.shtml). The phenotypic similarities between the properties were evaluated using the mean link between groups method or UPGMA (Unweighted Pair Group Method with Arithmetic Mean). The score for this similarity coefficient ranges from 0 (different) to 1 (similar) according to Sahu;Swain and Kar (2019).

RESULTS AND DISCUSSION
Among the eighty-eight raw milk samples obtained from the cooling tanks of the eight investigated farms with high Somatic Cells and Total Bacteria counts for six months, 67.04% (59/88) were contaminated, 27.27% (24/88) with Escherichia coli, and 56.81% (50/88) with Staphylococcus aureus (Table 2). Mixed contamination (E. coli and S. aureus) was observed in 25.42% (15/59) of the positive samples. The exclusive occurrence of only one species was also verified, where 59.32% (35/59) of the samples were contaminated with S. aureus and 15.25% (9/59) with E. coli (Table 2).
It was assumed that 100% of the herd of properties with high SCC and TBC counts in the cooling tanks had subclinical mastitis, similar to the percentages historically observed in other Mato Grosso cities, such as 74.2% in Nossa Senhora do Livramento (MARTINS et al., 2006) and85.2% in Cuiabá (MARTINS et al., 2010). A high incidence of subclinical and clinical mastitis may be due to a number of factors, such as deficient or non-existent mastitis control programs, cattle overcrowding, unhygienic milking practices and variations in farm locations (DEVI; DUTTA, 2018). Concerning the evaluated properties, inaccuracies such as poor environment hygiene, manual milking, failure to perform pre-and postdipping processes, or incorrect performance, lack of milker and cooling tank hygiene were observed. Failures of this type were also observed in other dairy properties in the state of Mato Grosso, such as in Nossa Senhora de Livramento (MARTINS et al., 2006), Cuiabá (MARTINS et al., 2010) and Carlinda (SILVA; SILVA; BETT, 2017).
Staphylococcus aureus was present in over 50% of the evaluated milk samples (Table 2). Costa et al. (2019) reported a moderate relationship between high somatic cell counts (≥700.000 SCC/mL), and mastitis caused by S. aureus. The main identified factors associated with mastitis caused by this microorganism are handling deficiencies during milking, in addition to being linked to teat infection; S. aureus may be present on milker hands and in their nostrils (GUIMARÃES et al., 2017;KREWER et al., 2013). Bi et al. (2016) observed that S. aureus was significantly more prevalent in small properties compared to large ones in China. The present study was carried out on small farms, which may justify the obtained results (Table 2).  The occurrence percentages of E. coli (29.54%) in the present study are similar to those observed in dairy farms in China (28.6%) (BI et al., 2016), indicating unsatisfactory hygienic conditions (MARTINS et al., 2016). As a pathogen acquired from the environment the main sources of E. coli are contact with humidity, mud and animal feces (BETTANIN et al., 2019). A lack of hygiene at the assessed properties may result in contact with fecal material, the main water and raw milk contamination source by this microorganism. Therefore, water quality and effective hygienic practices during milking are crucial in order to avoid raw milk contamination (RIBEIRO et al., 2019).
Property F4 was displayed the lowest number of contaminated samples (Table 2), two, one by E. coli and S. aureus (CO1), and the other by E. coli alone (CO5). This property presented the best hygiene and general state of conservation. The rest of the properties were hygienically deficient concerning milking, milking utensils and cooling tanks, with predominant S. aureus detection (Table 2). At F5, in addition to hygienic deficits, one mastitis case was reported, and at F6, milk during the collection period presented strange characteristics, containing odor and dirt. A high detection of both S. aureus and E. coli was noted at these properties, in five and seven samples, respectively; differing from the other investigated properties, and explained by the scenario described earlier in this paragraph (Table 1).
E. coli counts ranged from 2 to 4 Log 10 CFU/mL, and from 2 to 4.34 Log 10 CFU/mL for S. aureus (Table 2). E. coli counts were above the Brazilian Normative Instruction (IN) N º 76 of 26 November 2018 (BRASIL, 2018b) recommendation, with 26 raw milk samples unfit for consumption or for the preparation of milk-based derivatives (BRASL, 2018b). Regarding S. aureus, 88% (44/50) of the positive samples (Table 2) exhibited counts above the maximum established by the Institute of Medicine National/Research Council of the National Academies (2003). Where two samples exhibited counts of 4.04 and 4.34 Log 10 CFU/mL. This is a cause for concern, since populations close to 5 Log 10 CFU/mL of S. aureus could potentially produce healt resistant Staphylococcal Enterotoxins (STAMFORD et al., 2006).
The E. coli isolated from raw milk were resistant to most antibiotics, with the exception of ceftiofur and those belonging to the Fluoroquinolone group (Figure 1). The isolated E. coli displayed high resistance to rifampicin (70.83%), ampicillin (50%), cefoxitin and erythromycin (41.67%), similar to strains isolated from milk and dairy products in Iran, which displayed high resistance to ampicillin, of 100% in Kashan and 43.4% in Mashhad (FALLAH et al., 2019;CHALESHTORI et al., 2017). This differs from strains from other regions where lower resistance percentages to ampicillin were detected, for example, of 10% in Bahia, Brazil (BARRETO et al., 2012) and 6.7% in India (RASHEED et al., 2014). Strains isolated from milk and dairy products in the present study displayed resistance to different antibiotic groups (Figure 1), similar to strains found in Iran (FALLAH et al., 2019;CHALESHTORI et al., 2017). However, E. coli strains reported in other studies conducted with milk and dairy products in Brazil displayed resistance only to nalidixic acid, cefoxitin and tetracycline (RIBEIRO et al., 2019). Resistance to several antibiotics may be attributed to the presence of certain genes, such as mcr-mediated colistin resistance and/ or plasmids from the incF and incX groups, among other factors (SAIDENBERG et al., 2020;ZAMPARETTE et al., 2020;CASTRO et al., 2021). These genes and plasmids have been previously detected in E. coli isolated from clinical and food samples in Brazil (SAIDENBERG et al., 2020;ZAMPARETTE et al., 2020) as well as in bovine faeces in Canada (CASTRO et al., 2021).
The detected MDR E. coli and S. aureus strains displayed common resistance to rifampicin, cefoxitin, tetracycline, chloramphenicol, ampicillin, azithromycin and nitrofurantoin. Resistances to imipenem, cefepime, and gentamicin were detected exclusively in E. coli, and resistances to erythromycin, trimethoprim, ceftiofur, sulfamethoxazole/trimethoprim, enrofloxacin and florfenicol were detected only in S. aureus. Resistance to erythromycin in E. coli and to aztreonam and nalidixic acid in S. aureus were not categorized as MDR, which could lead to bias concerning MDR classification, as these species are naturally resistant to these antibiotics.
The Jaccard similarity coefficient was employed to compare similarities or differences in E. coli, and S. aureus strain behavior regarding their phenotypic resistance to the tested antibiotics among the eight investigated properties.
The findings indicate that properties F5 and F6 displayed the greatest similarity (0.91) concerning S. aureus antimicrobial resistance (Figure 2, A), differentiated only by resistance to sulfamethoxazole/trimethoprim in strains isolated from property F5. On the other hand, E. coli displayed low similarity among the eight properties, with a maximum of 0.60 among strains isolated from properties F6 and F8 (Figure 2, B). The low similarity among these strains was due to resistance to only chloramphenicol and sulfamethoxazole/trimethoprim, and to azithromycin and tetracycline, expressed respectively, by E. coli strains isolated from F6 and F8. Properties F5 and F6 are physically distant, and properties F6 and F8 are physically close.
The differences and similarities observed between the eight milk producing properties regarding the resistance expressed by the isolated E. coli and S. aureus strains to antibiotics is due to lack of technical assistance from veterinarians and the empirical use of antibiotics by the farm owners. This lack of technical assistance concerning mastitis control also affected the sanitary conditions of the investigated properties, which mostly exhibited hygienic and general conservation deficits, as well as hygienic deficiencies concerning milking, utensils and cooling tanks. Thus, a vicious cycle develops, resulting in reinfections, with high somatic cell and total bacteria counts in raw milk, as well as the presence and high counts of microorganisms like S. aureus and E. coli.

CONCLUSION
The findings reported herein confirm a hygienic deficit in the production of the evaluated dairy basin and potentially mastitic milk. This was indicated by high Somatic Cell (SCC), and Total Bacteria (TBC) counts and confirmed by the presence of E. coli and S. aureus detected by the Compact Dry ® kit. The susceptibility profiles of the isolates indicate incorrect handling of animals presenting mastitis, the indiscriminate use of antibiotics, as the isolates were resistant to 18 antibiotics belonging to 13 different chemical classes, and a high resistance similarity expressed by S. aureus strains, representing public health risks.