The primary objective of this research was to quantify the diversity of Staphylococcus bacteria on farms that make artisan cheese and to describe the sources of potential pathogenic and beneficial members of this genus. A total of 1052 staphylococci isolated from different sources on dairy farms were identified to species level. Twenty-seven different Staphylococcus species were detected with varying prevalence between farms. Each of the five farms in the study had a unique dominant species. Overall, S. haemolyticus was the dominant species (26.5%) of all the isolates. Staphylococcus aureus, S. chromogenes and S. haemolyticus were the most isolated species from quarter milk. Cow skin was dominated by S. haemolyticus, S. auricularis, S. succinus. Apart from S. haemolyticus, these species were rarely isolated in cow milk suggesting that, they could be normal cow skin flora or opportunistic pathogens. Staphylococcus xylosus and S. equorum were more likely to be isolated in the environment.
All isolates (n=1052) were screened for carriage of blaz and mecA antimicrobial resistance genes. These genes (blaz and mecA) were identified in 16.1% (n=169) and 0.6% (n=54) of all isolates respectively; carriage of these genes was also species specific, 100% (n=36) of all S. fleurettii were mecA positive. mecA gene was significantly associated with environmental isolates (p=0.02) and farm E (p<0.01). In vitro antibiotic resistance analysis was done on only 562 isolates. All isolates were susceptible to Vancomycin, Gentamycin, and Amoxicillin/Clavulanic acid, Enrofloxacin, Cephalocin, Penicillin/novobiocine and Ceftioufur. Resistance in other antibiotics varied with highest resistance seen with Penicillin (15.5%) and ampicillin (12.1%).
We evaluated biofilm formation on subset of isolates (n=365) on 2 types of plastic surfaces. Surface type overall, was not a significant predictor of biofilm formation (p=0.3) but significant with S. epidermidis (p=0.04). In general, isolates formed more biofilm biomass on hydrophilic surfaces. Staphylococcus gallinarum and Staphylococcus. sp. 020902-022-273 formed much biofilm biomass compared to other species. Biofilm formation in S. haemolyticus was heterogeneous.
Vermont is a leader in on-farm artisanal cheese production with more cheese makers per capita than any other state. Farmstead cheese makers in Vermont tend to manage pasture-based (grass-fed) dairy herds. These farms may or may-not be organic, but are similar to organic dairies in that the premium they receive for their value-added product appears to allow them to maintain smaller herds compared to farms that sell milk through fluid milk markets. Creating cheese adds value to milk, as well as fostering small businesses, which follow a product differentiation strategy (Porter, 1985). Farms that sell differentiated and value-added products gain greater control over the prices they receive and are less vulnerable to downturns and volatility in commodity milk prices. Numerous studies have found consumer demand and willingness to pay a premium for differentiated dairy products (Conner, & Hamm, 2007; Corner & Oppenheim, 2008). Producing farmstead or artisan cheeses adds value to milk produced on small dairy farms providing an opportunity for improved economic returns, and farmstead cheese producers are motivated to provide healthful products to consumers while promoting the health and welfare of their livestock. Yet, the additional investment in an on-farm cheese facility introduces system complexity that requires an integrated approach to animal health, milk quality and food safety.
The coagulase negative Staphylococcus (CNS) bacteria are emerging as important barriers to improving milk quality and udder health on many dairy farms. CNS species have long been recognized to cause mastitis and recently their relative importance appears to be increasing (Pyorala and Taponen, 2009). This change has paralleled the successful control of the major contagious pathogens Staphylococcus aureus and Streptococcus agalactiae. The practices that have contributed to control of these two pathogens are less effective for CNS species, and it appears the ecology of mastitis causing organisms is shifting. New approaches will be needed to advance mastitis control and milking hygiene. Understanding the epidemiology of pathogens is the foundation to developing effective control practices. Understanding the potential role of beneficial commensal bacteria may also advance mastitis control, milk quality and food safety.
There are at least 47 CNS species, and previous studies have isolated at least 20 of these species from various sources on dairy farms (Piessens et al., 2011; Becker et al., 2014). The diversity of CNS species on dairy farms in poorly described and their potential role in mastitis is complex. For example, some CNS species are believed to be normal skin flora, may protect the cow from infection with pathogenic mastitis organisms, and help to maintain a healthy skin microbiome (White et al, 2001). Identifying practices that maintain or encourage these beneficial species would be desirable. While other CNS species are commonly associated with cases of mastitis and practices that control or reduce these pathogenic species would be valuable. Some CNS species and strains are strong biofilm formers and others may be a source of antibiotic resistance genes. Antibiotic resistance and biofilm formation phenotypes have been linked to the development of chronic mastitis infections in cattle (Balsega et al., 1993; Melchior et al., 2006). In addition to being important animal pathogens, some Staphylococcus species are also important human pathogens and many species are zoonotic, causing disease in both animals and humans. We believe these differences could be associated with certain unique phenotypes or epidemiologic sources. Improved understanding of staphylococcus epidemiology and ecology on dairy farms will give more insight to the contribution of specific species and strains in this diverse group of bacteria. Controlling staphylococcus mastitis on farms that make artisan cheese will improve profitability, product quality, and cattle health and welfare.
Below are the objectives we intended to achieve under this project.
- To identify the sources of Staphylococcus species on farms that make farmstead cheese.
We described the detailed distribution of the CNS species on 5 dairy farms that make artisan or farmstead cheese. Under this objective we used sequence-based molecular methods to describe the comparative distribution of bacteria isolated from various sources or niches on these farms.
The specific aims of this objective were to:
- Identify the CNS species that are opportunistic mastitis pathogens and determine their sources on these farms.
- Identify the CNS species that are normal flora of the udder and teat skin but are not commonly isolated from cases of mastitis, and describe these as a component of the beneficial teat skin commensal bacteria.
- Identify the sources of CNS species that are known beneficial bacteria in cheese maturation.
The aims under this objective were successfully achieved.
- To quantify the association between antibiotic resistances or biofilm formation and the Staphylococcus species or strains isolated from various sources on dairy farms
The specific aims of this objective were to:
- Quantify the prevalence of antimicrobial resistance among the various CNS species isolated from different sources on artisan cheese farms.
- Quantify the prevalence of biofilm forming bacteria among the various CNS species isolated from different sources on artisan cheese farms
The aims under this objective were successfully achieved.
Objective 1. To identify the sources of Staphylococcus species on farms that make farmstead cheese.
- i) Sample collection
Staphylococcus isolates used in this study were collected from 5 farms that make artisan cheese. These isolates were collected during a field study focused on Staphylococcus species epidemiology on artisan cheese farms. Farms with a known history of Staphylococcus aureus mastitis control were identified and farmers who expressed interest in participating in the study were enrolled. The bacterial isolates were collected from the following 17 sources on each farm: 1) bulk tank milk from farm storage tanks; 2) quarter milk samples from individual cows; cow skin swabs from 3) teat canal, 4) teat end, 5) teat barrel, 6) udder, 7) vagina, and 8) nose; environmental swabs from 9) milking equipment, 10) milk filter, 11) feed troughs, 12) water troughs, 13) feeds, and 14) cow stall rails; 15) human nasal and 16) hand swabs, and 17) nasal samples from other livestock or pets on the farm. All samples were collected following approved protocols from University of Vermont IACUC and IRB committees for animal and human research. Sample collection from the various sources followed well established and previously published methods. All samples were immediately placed on ice and transported to the laboratory where they were maintained under refrigeration until cultured within 24 hours of collection.
- ii) Bacterial isolation
The bacteria were isolated using established methods. For milk samples, either 0.01 ml (quarter milk samples) or 0.1 ml (bulk tank milk samples) of milk was cultured on non-selective (tryptic soy broth with 5% sheep blood, Blood agar, BA) or selective medium for isolation of Staphylococci (mannitol salt agar, MSA). Skin and nasal swab samples were vortexed at 1200 cycles per minute for 15 minutes and environmental swab samples were mixed using a stomacher apparatus. Ten-fold serial dilutions were made from swab samples and 100 uL of the serial dilutions were plated to BA or MSA. Plates were incubated at 370C for up to 48 hours and presumptive staphylococci were selected based on colony morphology using established methods and sub-cultured for isolation and storage. Staphylococci were confirmed by gram staining and catalase testing. Coagulase testing was completed for all gram-positive, catalase-positive cocci to presumptively identify S. aureus isolates. All gram-positive, catalase-positive, coagulase-positive cocci were confirmed as S. aureus by polymerase chain reaction (PCR) amplification of the species-specific thermonuclease gene. Using this sampling scheme we confirmed these farms had a low prevalence of S. aureus mastitis. The remaining CNS species isolates were stored frozen in pure culture.
iii) Species identification of CNS isolates from various sources on 5 artisan cheese farms
Sequenced-based genetic typing of two housekeeping genes (tuf and rpoB) was used for CNS speciation. Bacterial DNA was extracted from pure cultures of individual isolates from storage using Qiagen DNA extraction kit following the manufacturer’s instructions. Amplified DNA using gene specific primers were sequenced at the University of Vermont core facility. Alignments of forward and reverse sequence reads were constructed and species determined by consensus sequence homology to published gene sequences using the Basic Local Alignment Search Tool (BLAST) of the National Center for Biotechnology Information (NCBI; http://blast.ncbi.nlm.nih.gov/Blast.cgi).
Objective 2. To quantify the association between antibiotic resistance or biofilm formation
- i) Antimicrobial susceptibility and genotyping.
Antimicrobial susceptibility phenotyping, we used disc diffusion (DD) and broth micro-dilution minimum inhibitory concentration (MIC) following Clinical Laboratory Standards Institute (CLSI) guidelines. Susceptibility testing for the antimicrobials ampicillin, cephalothin, ceftiofur, erythromycin, oxacillin, pirlimycin, penicillin, penicillin/novobiocin, and tetracycline were conducted using agar disc diffusion zone of inhibition and broth micro-dilution minimum inhibitory concentration methods. The antibiotics in this panel were selected because they are commonly used in dairy production systems, with the exception of oxacillin. Oxacillin is included to screen for methicillin resistance because of its importance as a public health concern. In addition, we used established PCR methods to test all isolates for carriage of blaZ and mecA genes, which encode for penicillin and methicillin resistance, respectively. The association between antimicrobial susceptibility and bacterial species was tested in logistic regression model.
- ii) Biofilm in vitro assay and genotyping
Briefly, overnight bacterial cultures in broth were diluted 1:200 to a standardized concentration (107 CFU per ml) in Tryptic Soy Broth supplemented with 0.25% glucose. Two hundred microliters of each overnight culture were transferred to 3 wells of a 96-well polystyrene flat-bottom plate and incubated for 24 hours (29 isolates plus positive and negative controls were tested on each plate). Plates were washed 3 times with PBS (phosphate buffered saline) to remove non-adherent bacteria, air dried for at least 2 hrs at room temperature, stained with safranin for 10 minutes, washed 3-times with sterile water, and air dried for at least 2 hrs at room temperature. Stained biofilms were then destained with 95% ethanol and the plates are read on a spectrophotometer at 490 nm. Optical densities were recorded and biofilm formations were quantified using an established ordinal scale. All isolates evaluated within a plate in triplicate and assays were repeated in triplicate. We had also planned to screen for genes (e.g. icaA and icaD) that encode polysaccharide and biofilm associated protein (bap). The results from the PCR screen are controversial and we decided not present them before we definitively confirm their authenticity.
Supporting files (tables and figure)
The project developed as planned and we needed to make no course corrections along the way. The data generated from this project were used by at least 3 of the participating farmers to make mastitis control management decisions and implement mastitis control practices, including consideration of teat end hygiene practices, milk culture and mastitis diagnosis practices, and cow culling decisions.
Objective 1: Results and discussion
We definitively speciated a total of 1052 isolates from five farms that make farmstead cheese (Figure 1). The prevalence of CNS species varied between farms, with some species more dominant on some farms than others. For instance, apart from farm C and D, which were dominated by S. haemolyticus, the rest of the farms were dominated by unique species. These species included S. succinus, S. auricularis and S. chromogenes. One interesting finding was the high prevalence of S. succinus and S. fluerettii on farm E but very low prevalence or absence on the rest of the farms. Overall, S. haemolyticus was the dominant species driven by its high prevalence on farm C. This differential dominance of staphylococci species is not uncommon; a similar pattern was seen in a longitudinal study focusing on the epidemiology of CNS on six Flemish dairy farms in Belgium (Piessens et al, 2011). We do believe that the differential prevalence of staphylococci species on these farms may be driven by farm-level management practices, which were beyond the scope of this work. On the other hand, this might also be driven by pathogen characteristics for instance, varying virulence determinants between strains of similar species. This work has laid a foundation to generate hypotheses and explore these unique epidemiologic findings.
Distribution of species also varied within different niches (Figure 2), with the predominant species isolated from quarter milk (from cows with subclinical mastitis) being S. chromogenes with little or zero prevalence in other niches. This suggests that S. chromogenes is an udder-adapted species and likely spreads from cow to cow. Our strain typing results (data not shown) indicate that strains from quarter milk are more likely to be different from those on the teat skin and environment. We would conclude that teat skin or environment might not be the likely reservoirs of the strains causing mastitis. Other species that were very common in quarter milk samples include S. haemolyticus and S. aureus. Staphylococcus auricularis and S. succinus were the dominant species on teat skin with relatively low prevalence in quarter milk samples. These two species might be just normal commensals on teat skin but can occasionally cause opportunistic infections. The dominant environmental species included S. equorum and S. xylosus though they were also present in other niches with low prevalence. Staphylococcus equorum is a common environmental species (Piessens et al., 2011) and it is thought to be beneficial in cheese making. Species that were isolated in various niches especially on the same farm require strain typing to ascertain the strain diversity; a good example is S. fluerettii where all the 36 isolates on farm E are distributed in five different niches including humans. Strain typing results for S. aureus (data not shown) show that, human strain types are different from strains from cow milk suggesting no cross transmission between humans and animals. We don’t know whether it is the same scenario with other species like S. fluerettii, S. haemolyticus or S. epidermidis, which seem to be distributed in almost all niches.
Objective 2: Results and discussion
All isolates were PCR screened for the presence of blaz and mecA genes, which encode for penicillin and methicillin resistance respectively. Table 3 shows the percentage of isolates that were positive for either of the genes or both. At least an isolate carried a blaz gene in 13/27 species identified but was much more prevalent in S. epidermidis, S. auricularis and S. chromogenes relative to the total number of isolates per species. On the other hand, mecA gene was found in only 5 species; S. fleurettii, S. epidermidis, S. sciuri, S. homnis and S. vitulinus. These five species are likely potential reservoirs of mecA gene for other bacteria (e.g S. aureus.)Staphylococcus fleurettii has been documented as the likey origin of mecA gene (Tsubakishita et al., 2010). Only six isolates carried both mecA and blaz genes. Overall, blaz gene was much more prevalent 16.1% (n=169) than mecA 5.1% (n=54). The mecA gene was much associated with environmental isolates (p=0.02) and farm E (p<0.01) compared to quarter milk source and Farm A respectively.
We tested antibiotic susceptibility of a total of 562 selected CNS isolates using antibiotics described in the material and methods. Some antibiotics were tested using two methods (Disc diffusion and micro-dillution) and others one of the two methods. All isolates were susceptible to Vancomycin, Gentamycin, Amoxicillin/Clavulanic acid, Enrofloxacin, Cephalocin, Penicillin/novobiocin and Ceftioufur. One S. epidermidis isolate from a human hand was resistant to Cefoxitin, which is the recommended antibiotic for detection of methicillin resistance. This S. epidermidis isolate carried both blaz and mecA genes. However, none of the remaining 13 isolates, which carried the mecA gene, were resistant to Cefoxitin. These isolates according to CLSI guidelines are methicillin resistant. Except penicillin/novobiocin and Sulphadimethoxine, all the antibiotics in Table 4 were tested using two methods. Ampicillin and penicillin were the most discordant drugs in terms of percentage of isolates that were resistant using both methods. With micro-dilution, 3% and 3.8% of the isolates were resistant to ampicillin and penicillin respectively (Table 4) but 12.1% and 15.5% of the isolates were resistant to ampicillin and penicillin by disc diffusion respectively. Of the 12.1% ampicillin resistant isolates, 15 did not carry the blaz gene whereas 63 among the susceptible isolates did. For penicillin, 22 isolates did not carry the gene among the resistant isolates and 50 among susceptible isolates did. Susceptibility in blaz positive isolates might be attributed to poor expression of the gene in some species. Whereas, resistance to penicillin without the blaz gene might be due to some novel resistance markers or the primers we used could not bind to a more divergent gene. In this study, we did not screen for presence of mecC, a recently discovered genetic marker for resistance to beta-lactam antibiotics. The percentage of isolates resistant to tetracycline, erythromycin, Pirlimycin and Oxacillin on both methods were in close range and for simplicity only data for micro-dilution is shown (Table 4). Lincomycin did not have a CLSI break point, however 36.3% of the isolates had a zone diameter less than 11mm suggesting possible high resistance to lincomycin. Clindamycin resistance was noticed in 3.9% of the isolates.
A total of 365 isolates were tested for their ability to form biofilms. Our previous work on S. aureus biofilm on strain level basis, showed differential biofilm biomass on treated (hydrophilic) and untreated (hydrophobic) surfaces, so we decided to test these isolates’ ability to form biofilms on two plate types (hydrophilic and hydrophobic). In a linear regression model, plate type was not a significant predictor of biofilm formation (p=0.3) but overall, all isolates formed more biofilm on treated plates (hydrophilic) (Figure 1) (red box plots) vs untreated (white box plots). Focusing on the individual species, S. epidermidis was the only species that formed statistically significant biofilm biomass on hydrophilic surfaces (p=0.04) compared to hydrophobic surfaces. This suggests overall that CNS stick to hydrophilic surfaces better than hydrophobic and so would be plausible to use treated plates in future CNS studies. We also wanted to know whether the high prevalence of certain species is driven by their ability to form biofilm. Surprisingly two minor species formed strong biofilms on average; S. gallinarum and S. sp. 020902-022-273. The dominant species S. haemolyticus had a lot of isolates that formed strong biofilms and others weak biofilms. Biofilm formation may not be responsible for the overall high prevalence of certain bacterial species. Farm factors or pathogen virulence determinants may be the major contributor, which need to be investigated in future studies.
- Baselga, R., I. Albizu, M. De La Cruz, E. Del Cacho, M. Barberan, and B. Amorena. 1993. Phase variation of slime production in Staphylococcus aureus: Implications in colonization and virulence. Infect. Immun. 61:4857–4862.
- Becker, K., C. Heilmann, and G. Peters. 2014. Coagulase-negative staphylococci. Clin. Microbiol. Rev. 27(4):870-926.
- Conner, D. S., & Hamm, M. W. (2007). The economics of pasture raised animal products: Food, markets and community. East Lansing: C.S. Mott Group for Sustainable Food Systems, Michigan State University.
- Conner, D. S., & Oppenheim, D. (2008). Demand for pasture-raised livestock products: Results from Michigan Retail Surveys. Journal of Agribusiness, 26(1), 1-20.
- Melchior, M. B., H. Vaarkamp, and J. Fink-Gremmels. 2006. Biofilms: A role in recurrent mastitis infections? Vet. J. 171:398–407.
- Piessens, V., E. Van Coillie, B. Verbist, K. Supre, G. Braem, A. Van Nuffel, L. De Vuyst, M. Heyndrickx, and S. De Vliegher. 2011. Distribution of coagulase-negative Staphylococcus species from milk and environment of dairy cows differs between herds. J. Dairy Sci. 94(6):2933-2944.
- Pyorala and Taponen, 2009. Coagulase-negative staphylococci—Emerging mastitis pathogens. Vet. Microbiol. 134 (2009) 3–8
- Porter, M. E. (1985). Competitive Advantage: Creating and Sustaining Superior Performance.
- Tsubakishita, Sae et al. “Origin and Molecular Evolution of the Determinant of Methicillin Resistance in Staphylococci.” Antimicrobial Agents and Chemotherapy10 (2010): 4352–4359.
- White, L.J., Schukken, Y.H., Lam, T.J., Medley, G.F., Chappell, M.J., 2001. A multispecies model for the transmission and control of mastitis in dairy cows. Epidemiol. Infect. 127, 567–576.
We have demonstrated that, the proposed PCR-based species typing system is able to discriminate among the Staphylococcus species isolated from these farms and we have established laboratory methods and protocols. We improved our understanding of staphylococcus epidemiology and ecology on dairy farms, which has provided more insight to the contribution of specific species in this diverse group of bacteria. Most species isolated from milk of infected cows are the same as those from cow skin suggesting that cow skin could be the primary source of these bacteria. However, what we haven’t elucidated yet, is whether the strains are the same. The prevalence of species dominant in the environment was low in cow milk. This suggests that most bacteria causing mastitis on these farms might be originating from the cow skin surfaces and rarely in the environment. Cases of mastitis could be potentially reduced if farmers maintain cow teat end disinfection and hygiene. Our approach to use two methods to screen for antibiotic resistance has revealed some insights on pros and cons of two in vitro antibiotics assay methods. Our previous work showed no discordance with S. aureus on penicillin and ampicillin suggesting that, the discordance might just be unique to CNS. There was a lot of discordance between the two methods in two beta-lactam antibiotics ampicillin and penicillin with most isolates with blaz gene being resistant by disc diffusion as compared to micro-dilution. Micro-dilution was more likely to show a blaz positive isolate susceptible to ampicillin or penicillin than disc diffusion giving a lot of false negatives. Although, in vitro antibiotic analysis may not necessarily predict clinical outcome, this work gives an insight on the drugs of choice for treating infected cows. For instance, regardless of the source and method used, all isolates tested were susceptible to Vancomycin, Gentamycin, and Amoxicillin/Clavulanic acid, Enrofloxacin, Cephalocin, Penicillin/novobiocine and Ceftioufur. Farmers would use this information, under the guidance of their veterinarian, to make treatment decisions for cases of mastitis, recognizing that some of the compounds we tested are not allowable or recommended for use in dairy production systems due to food safety and public health concerns or their importance for use in human medicine.
Education & Outreach Activities and Participation Summary
Each participating farmer in the study was given a report of the bacteriological findings of quarter milk samples. This has enabled farmers to make farm management decisions like culling infected cows, milking infected cows last in order to minimize transmission from cow to cow. This has been common with S. aureus infected cows. This work will generate two manuscripts and one is already in progress “Epidemiology of Staphylococci on selected Dairy Farms Producing Farmstead cheese in Vermont” and should be ready for submission to a journal by end of November this year. The second manuscript will be completed early next year.
I was recently invited for “across the fence TV show” on WCAX TV, where I briefly talked about this research and its impact to the farmstead cheese producers. .https://www.youtube.com/watch?v=pCwW5hz_Ah8&feature=youtu.be
Areas needing additional study
This work generated a couple of questions that need to be elucidated in future studies. One out of fourteen isolates that were mecA positive was phenotypically resistant to Cefoxitin an antibiotic that is now acceptable for detection of methicillin resistance. We hypothesize that this has to do with genetic make up of the staphylococcal chromosomal cassette mec, which could be interfering with the expression of mecA gene. I propose characterization of the cassette of all mecA positive CNS isolates and compare it with mecA positive isolates that are phenotypically resistant, for example S. aureus. This concept can also be extended to blaZ positive isolates that are susceptible to penicillin or ampicillin. Unlike CNS, we do not see this genotypic-phenotypic discordance in S. aureus, we have worked on at least 20 blaz positive S. aureus isolates and all of them were phenotypically resistant on both assays. Sequence analysis of the blaz gene in different staphylococcus species might give more insight about the discordance within CNS species. The major emphasis being understanding sequence variation of blaz gene in penicillin susceptible and resistant isolates within the genus Staphylococcus.
One addition study would be on understanding the epidemiology of some important staphylococci at strain level in order to better understand the direction of transmission. Our laboratory uses multilocus sequence typing (MLST; http://pubmlst.org), which is a highly acceptable technique to understand local and global epidemiology of bacteria. We have worked on S. aureus and S. chromogenes. I would propose to expand this work to S. haemolyticus and S. epidermidis. Staphylococcus heamolyticus was common almost in all niches whereas S. epidermidis was common in humans, I would want to know whether humans might the source of isolates we find in cow milk. MLST has already been developed for these species (http://pubmlst.org). Staphylococcus fleurettii was also interesting because it was the only species where all the isolates where mecA positive, much prevalent only on one farm. This species was isolated almost in all niches including humans. There is no MLST for this species yet but whole genome sequencing can be done on the 36 isolates of this species on this single farm.