Bovine mastitis is a disease characterized by mammary gland inflammation that affects dairy herds worldwide. Mastitis leads to a decline in milk production and quality, which coupled with high treatment costs or early culling of animals, is responsible for significant losses in dairy farms (Petrovski et al., 2006). Bovine mastitis can be classified as subclinical mastitis (asymptomatic) or clinical mastitis (symptomatic). Even though the inflammatory symptoms are evident in the case of clinical mastitis, prompting producers, or veterinarians to take appropriate action, diagnosis of subclinical mastitis is mainly carried out by milk testing, e.g., using the California Mastitis Test (Schalm and Noorlander, 1957). This can be problematic for farmers, since subclinical infections, which can go unnoticed without regular cow screenings, are associated with a higher somatic cell count in milk. This high cell count results in decreased milk quality and economic value (Halasa et al., 2007), which implies that disease control and prevention are essential for dairy farmers' subsistence. Mastitis control programs are fundamentally based on three pillars: (a) prevention of new infections, (b) elimination of existing ones, and (c) monitoring udder health, with disease prevention taking a predominant role in recent years (LeBlanc et al., 2006). Both the “five point plan” aimed at lowering the overall incidence of mastitis (McDougall, 2002), and the use of vaccination (Finch et al., 1997) can be considered advantageous to minimize losses attributed to this disease.

Several bacterial species can cause the disease and typical bovine mastitis-causing pathogens include Staphylococcus aureus, Escherichia coli, and members of the Streptococcus genus (Streptococcus uberis, Streptococcus agalactiae, and Streptococcus dysgalactiae) (Bradley, 2002). Among these, Streptococcus uberis (S. uberis) is acknowledged as one of the most important mastitis pathogens, since it is frequently isolated from dairy herds worldwide (Riekerink et al., 2008; Verbeke et al., 2014). Although largely regarded as an environmental pathogen, characterized by different clonal lineages causing disease (Lundberg et al., 2014), S. uberis can also behave as a contagious pathogen (Zadoks et al., 2001; Rato et al., 2008). The typical environmental reservoirs where S. uberis can be found include grass, straw bedding and also the skin, gut, and genital tract of cattle (Leigh, 1999; Lopez-Benavides et al., 2007). The ability of this pathogen to survive in a wide range of substrates decreases the effectiveness of typical herd hygiene and disease prevention methods, since these are mostly aimed at controlling contagious pathogens. These drawbacks call for the need to develop and optimize new procedures to support disease prevention caused by both contagious and environmental pathogens, given that the factors influencing their prevalence are not identical (Barkema et al., 1999).

Despite the fact that antibiotic treatment is still extensively employed in the control of bovine mastitis, the use of the correct antibiotic and duration of treatment are essential to ensure a successful therapy (Hillerton and Kliem, 2002; Swinkels et al., 2015). Furthermore, antibiotic resistance in pathogens remains a great concern, making regular screenings of antibiotic resistance patterns an indispensable procedure (Erskine et al., 2002). Indeed, resistance of S. uberis to conventional antibiotics has been previously reported (León-Galván et al., 2015), adding to the difficulty to control this pathogen in herds.

The ability of S. uberis strains to effectively cause disease has been linked to the presence of specific virulence traits, which might provide an advantage in outcompeting other bacteria, or an improved capacity to invade and survive in the teat environment. In recent years, many of the genes coding for these traits were identified, namely the nisin U operon. This operon is responsible for the production and immunity to a bacteriocin of the lantibiotic class, which has antimicrobial activity against many lactic acid bacteria (Wirawan et al., 2006). Furthermore, the enhanced growth in the teat environment has been associated with the presence of a plasminogen activator protein, required for the degradation of extracellular matrix proteins (Rosey et al., 1999), and oligopeptide permeases that promote the ability to obtain essential amino acids from milk peptides (Smith et al., 2002). Efficient colonization and survival can also be attributed to the protein SUAM (Streptococcus uberis adhesion molecule), which plays a role in adherence to the bovine mammary epithelial cells (Almeida et al., 2006); to the ability of this species to produce an hyaluronic acid capsule, which confers resistance to phagocytosis and desiccation in the environment (Ward et al., 2001; Field et al., 2003); and to a glyceraldehyde-3-phosphate dehydrogenase (GAPDH) able to bind host proteins and protect against reactive oxygen species (Reinoso et al., 2011). The first fully sequenced genome of S. uberis (strain 0140J) confirmed the presence of several genes consistent with an organism capable of surviving in several environmental niches (Ward et al., 2009) and opened the possibility of using comparative genomic analyses to identify further virulence factors (Hossain et al., 2015).

Epidemiology studies are essential to study S. uberis populations and to identify problematic clonal lineages and preferred modes of transmission, which may vary in different dairy herds (Zadoks et al., 2003; Davies et al., 2016). Although fingerprinting techniques, such as Pulsed-field gel electrophoresis (Lundberg et al., 2014), can be used for this purpose, sequence-based typing methods such as multilocus sequence analysis (MLSA) and multilocus sequence typing (MLST) present several advantages, especially concerning inter-laboratory comparison of data. Two MLST schemes were developed for S. uberis, one that includes both housekeeping genes and virulence genes (Zadoks et al., 2005), and one consisting of seven housekeeping genes (Coffey et al., 2006).

In this work, a dot blot platform coupled with automatic data analysis was used to study S. uberis populations obtained from selected cows with recurring mastitis from two herds in Northern Portugal (Barcelos and Maia regions), aiming to provide insights of prevalent virulence patterns and clonal lineages and gain further insights on the impact of herd management practices on S. uberis population clonal structure of each herd.

The difficult control of S. uberis makes these bacteria one of the most damaging pathogens in dairy herds, and emphasizes the importance of epidemiological studies to identify the most problematic clonal lineages, modes of transmission, and population structure in a given region.

In this work, two dairy herds located in Northern Portugal (Barcelos and Maia) were selected due to previous reported persistent S. uberis infections. The two herds had similar management practices, with the herd from Barcelos having roughly half the number of cows and a better milking parlor management, since infected cows were immediately segregated. In Barcelos, it was possible to obtain 11 isolates from four different cows. Concerning SCC, cows #1 and #4 had higher values, nevertheless SCC values of cow #1 were continuously decreasing throughout the sampling timeframe, with the last obtained value below 200.000 cells mL⁻¹. A total of 43 isolates were obtained from Maia, 33 of which representing persistent infections, from 20 cows. SCC varied considerably between individuals and within the same animal during the sampling timeframe (Table 1).

The MLSA genotyping (Figure 1) showed that while in Maia most cows were infected by lineages within the same clade (Group II), in Barcelos each cow was generally infected by different lineages, clustered in five groups (Groups I, IIIa, IV, VI, and VII). These results suggest that in Barcelos S. uberis has the typical pattern of an environmental pathogen, since transmission between cows was not observed, and with different genotypes isolated from individual cows (Lundberg et al., 2014). On the other hand, it appears that in Maia S. uberis behaves as a contagious and persistent pathogen, suggesting cow-to-cow transmission, even though this could also be explained by the continuous acquisition from the same environmental source (Rato et al., 2008). Persistent infections of cows #5, #8, #9, #10, #12, #13, and #14 with isolates from Group II, suggests the ability of this lineage of S. uberis to colonize the quarter for long periods. Additionally, many of the new infections, namely in cows #6, #8, and #13, were caused by MLSA type II. The hypothesis of this lineage being highly contagious is supported by the fact that cows #18–#24, with no recorded mastitis in the previous visits, were newly infected by Group II isolates within the 6 month sampling timeframe. These results further support that the different modes of transmission displayed by S. uberis, either environmental or contagious, are strongly dependent on the dairy herd analyzed (Zadoks et al., 2001).

The assessment of MLSA genotyping results allowed to compare the obtained gene sequences with those already publicly available (Table S2). This comparison showed that the majority of the gene sequences used for MLSA had already been identified in previous studies and available in the PubMLST database. Namely, the set of sequences corresponding to the Group II isolates, widely established in Maia, were already reported in Portugal in 2002, suggesting that this lineage is well-established in the country. Concerning S. uberis lineage persistence across time, in Maia it was noticeable that even though most recurring infections seem to be caused by the same clonal lineage, different S. uberis clones can also be responsible for persistent infections in the same cow, with only 1 month apart between isolation dates (cow #6 and #11). This indicates that these cows were initially infected with an environmental S. uberis and afterwards were infected by a contagious S. uberis belonging to Group II, prevalent in the herd. From this evidence one can assume that both environmental and contagious mastitis affect the Maia herd.

The analysis of the presence of relevant virulence-related features using dot blot is a valuable tool to understand which genes might play an important role in increased virulence and survival of S. uberis. In this work, 15 probes were used to rapidly screen the presence of relevant genes in the isolated S. uberis (Figure 2, Tables 2, 3). Four probes (U1, U2, A1, and A2) confirmed the isolate's identity as S. uberis by VITEK 2, a widely used phenotypic identification system that despite its usefulness shows some percentage of misidentifications (Di Domenico et al., 2015). The virulence-related genes hasA (probe V7), hasC (V1), gapC (V2), oppF (V3), sua (V4), and pauA (V6) were present in most of the S. uberis tested causing recurring infections, suggesting their importance for efficient S. uberis colonization. The majority of isolates contained the genes hasC (V1) and hasA (V7), associated with hyaluronic capsule production, corresponding to the typical pattern of genes hasABC found in fully sequenced strains (Hossain et al., 2015). Nevertheless, in five isolates from Barcelos the hasA (probe V7) gene was not detected. Since this gene is acknowledged as essential for capsule production, these results corroborate previous studies suggesting that acapsulate bacteria can cause infection (Field et al., 2003; Pullinger et al., 2006). The presence of gapC (V2) in all isolates reinforced the predictions of its importance in virulence (Reinoso et al., 2011). Genes pauA (V6) and oppF (V3) were also present in all isolates tested, emphasizing previous results showing their contribution to an enhanced growth in milk (Rosey et al., 1999; Smith et al., 2002). The importance of the protein SUAM, which mediates S. uberis adhesion (Almeida et al., 2006), was also corroborated, since the sua gene (V4) was detected in all isolates. Furthermore, the presence of the nisin operon (NU1-nsuR, NU3-nsuI), which can provide a competitive advantage (Pryor et al., 2009; Richards et al., 2011), was restricted to six out of 44 isolates in the analyzed population.

Dot blot results showed that the high SCC values recorded for some milk samples and SCC variation observed at different milk sampling times (Table 1) could not be ascribed to specific virulence patterns, suggesting that other factors play a decisive role in determining the severity of the inflammation.

Regarding antibiotic resistance, genes linB (pirlimycin resistance) and tetS (tetracycline resistance) were widespread in the Maia population, confirming previous studies showing their prevalence among S. uberis strains (Rato et al., 2013). However, in Barcelos, among the 11 obtained isolates, the presence of these genes was restricted to one isolate with linB (probe R2) and two with tetS (R3). Interestingly, the presence of gene ermB (probe R1) was highly infrequent and shared a matching pattern to the presence of nsuR (NU1) and nsuI (NU3), although the five isolates presenting the gene (SU52, SU53, SU58, SU59, and SU45) were not clustered in the same MLSA group (Figures 1, 2).

In total, seven different virulence gene patterns were found in this work (Table 3), confirming that isolates with different virulence patterns can be equally efficient in causing disease (Reinoso et al., 2011). Furthermore, dot blot results showed that isolates from each herd presented differentiating gene patterns, with patterns B, C, and D being unique to the Barcelos herd and patterns E, F, and G being unique to Maia. Gene pattern A, found in isolates SU57, SU91, SU114, SU16, and SU73, was the only pattern common to isolates from both herds.

In this study we characterized the epidemiology of S. uberis in two dairy herds using two different molecular approaches. The seven dot blot patterns obtained (A–G) provided a solid distinction of the different clonal lineages disclosed using MLSA (I–VII), indicating that the epidemiological patterns specific to each farm were discernible using the dot blot platform. Furthermore, dot blot results also allowed to discern important virulence and antibiotic resistance traits of the studied population, which is particularly useful for veterinarians. Overall, this data indicates that the chosen DNA markers and dot blot platform with automatic data analysis, were shown to be an efficient tool for a preliminary epidemiological assessment of S. uberis, which might be particularly useful for certified labs.

PA, AA, IP, AP, MV, and FD performed the molecular biology experiments and in silico analyses. AP, MV, and HM performed the collection and analyses of milk samples and isolation of S. uberis strains. NR, HM, and FT were responsible for the conception and design of the work. All authors contributed to the drafting and revision of the manuscript.