Background

Mastitis is a common condition during lactation and its incidence oscillates between 5 and 33% of the lactating mothers [1, 2]. The number of non-infectious mastitis that become an infectious disease is usually so high that some authors define the term "mastitis" as an infectious process of the mammary gland characterized by a variety of local and systemic symptoms [3]. However, the number of studies dealing with the microbiological aspects of human mastitis is low and the role of specific agents has yet to be described. In fact, published articles on the bacteria causing this condition are scarce and most are, at least, 10 years old [2].

Traditionally, Staphylococcus aureus has been considered the most common etiological agent although, unfortunately, the cases in which microbiological analyses are performed are exceptional. However, treatments with antibiotic or antifungal drugs are usually prescribed without knowing the etiology or the antibiotic susceptibility of the microorganism involved. This practice may lead to a worsening of the symptoms since strains that cause mastitis may exhibit multirresistance to drugs and/or form biofilms. Therefore, a better knowledge of the main features of the bacterial species involved in the mastitic process would represent a great advance for the design of new strategies for the prevention and/or treatment of this condition.

In a previous work, we investigated the microbial diversity of breast milk in 20 women with lactational mastitis by culture-dependent and -independent techniques [4], and observed that staphylococci, mainly Staphylococcus epidermidis, seem to be the major microorganisms present in breast milk of women with infectious mastitis. In recent years S. epidermidis has become increasingly recognized as opportunistic pathogen [5, 6]. Parallel, several genetic determinants involved in mechanisms of adhesion and biofilm formation have been described in this species [7, 8] while its rate of resistance to several antibiotics has increased during the last years [9–11].

In this context, the objective of the present study was to evaluate the presence of S. epidermidis in breast milk of women with infectious mastitis, to characterize the isolates and to compare their properties with those of strains isolated from milk of healthy women.

Results

Bacterial counts and identification of staphylococci in milk

Presence of staphylococci was observed in 27 of the 30 samples provided by women with lactational mastitis. In these samples, counts in Baird Parker (BP) agar plates ranged between 4.0 and 6.0 log10 cfu mL-1 (Table 1). A total of 270 isolates were obtained from the BP plates (10 from each woman) and all of them were lysozyme-resistant, lysostaphin-sensitive, catalase-positive, Gram-positive cocci. Among these presumptive staphylococcal isolates, 200 were identified as S. epidermidis on the basis of biochemical tests and species-specific PCR assays. This species was present in 26 milk samples. Only 35 staphylococcal isolates belonged to the species S. aureus and they were obtained from milk of eight women. PCR sequencing of a 16S rDNA fragment confirmed the results. The remaining 35 isolates that gave no amplification with the multiplex PCR were further identified by 16S rDNA PCR sequencing as Staphylococcus pasteuri (n = 16), Staphylococcus warneri (n = 11) and Staphylococcus hominis (n = 8) (Table 1). The partial 16S rDNA sequences obtained from single isolates belonging to the species Staphylococcus aureus and Staphylococcus epidermidis were deposited in the EMBL nucleotide sequence database under accession numbers [EMBL: AM697666] and [EMBL: AM697667], respectively. Then, our attention was focused on the S. epidermidis isolates.

Table 1 Samples and isolates used in this study
Full size table

Genotyping of the S. epidermidisisolates by PFGE profiling

The 200 isolates of S. epidermidis recovered in this study were subjected to PFGE genotyping together with 105 isolates previously obtained from breast milk of 12 healthy women (Table 1). The analysis of the fingerprints obtained revealed the existence of 40 different pulsotypes among the isolates from women with mastitis and 36 among healthy women. Comparison of these genotypes showed that most of the strains grouped together depending on their origin in two different clusters, one containing most of the strains obtained from mastitic milk while the second contained most of the strains isolated from milk of healthy women (Figure 1).

Figure 1
figure 1

Dendogram obtained by PFGE analysis of the S. epidermidis strains from mastitis and healthy women. Strains isolated from mastitis-suffering women are shaded. Information on the presence of icaD, fbe and mecA genes and resistance to oxacillin (OXA; MIC > 2 μg mL-1), erythromycin (ERY; MIC > 4 μg mL-1), clindamycin (CLY; MIC > 2 μg mL-1) and mupirocin (MUP; MIC > 512 μg mL-1) has been included.

Full size image

Detection of virulence determinants among the S. epidermidisstrains

The 76 different S. epidermidis strains (40 from milk of women with mastitis and 36 strains from that of healthy women) were selected to study the presence of potential virulence traits. Hemolytic activity could not be detected or was very weak among all the assayed strains. In relation to adhesion-related genes, the multiplex PCR assay revealed the presence of the genes embp and atlE in all the strains. The fbe gene was detected in 65% of the strains from mastitis and in 75% of those isolated from healthy women (P = 0,3434). In contrast, the icaD gene was more prevalent among strains from mastitis cases (33%) than in those from healthy women (11%) (P = 0,0255) (Figure 1). A good correlation was observed between the presence of biofilm-related ica operon and the results obtained using the CRA assay, which determines potential for slime production, and all the strains that amplified for the gene gave also positive results by the phenotypic assay.

Determination of MIC's to several antibiotics

Determination of MIC's to 21 antibiotics or antibiotics mixtures in the 76 S. epidermidis strains revealed that all of them were susceptible to the lower concentration of nitrofurantoin (32 μg mL-1) and rifampin (1 μg mL-1) while the results against the rest of antibiotics were variable depending on the strains (Table 2). Independently of their origin, most of the strains were sensitive to trimethoprim/sulfamethoxazole (MIC < 2/38 μg mL-1 for 90% of the strains), gentamicyn (≤ 2 μg mL-1 for 87%), linezolid (≤ 2 μg mL-1 for 86%), fosfomicyn (≤ 16 μg mL-1 for 82%), ciprofloxacin (≤ 0,5 μg mL-1 for 76%), tetracycline (≤ 8 μg mL-1 for 75%), chloramphenicol (≤ 16 μg mL-1 for 90%), penicillin (≤ 4 μg mL-1 for 72%), ampicillin (≤ 4 μg mL-1 for 80%) and the glycopeptides vancomycin (≤ 2 μg mL-1 for 93%) and teicoplanin (≤ 1 μg mL-1 for 70%). The percentage of susceptible strains was lower for imipenem (≤ 0,12 μg mL-1for 58%) and quinupristin/dalfopriscin (≤ 0,25 μg mL-1 for 57%). However, significant differences were observed in the percentage of strains resistant to some antibiotics depending on their origin (Figure 1). For instance, 43% of isolates from mastitic samples showed a MIC of mupirocin ≥ 512 μg mL-1 while only 22% of those isolated from non-mastitic samples reached this value (P = 0,0437). Similarly, 60% of the mastitic-related strains showed a MIC > 4 μg mL-1 against erythromycin in contrast to 33% of the other group (P = 0,0201). In the case of clindamycin, 28% of the strains from mastitic milk presented a MIC > 2 μg mL-1 while the percentage was of 8% in strains from healthy women (P = 0,0314). A higher percentage of oxacillin resistance (MIC > 2 μg mL-1) was also found among the strains from mastitis (68%) than those from healthy mothers (39%) (P = 0,0125). Finally, two strains from women with mastitis (CJ11 and DG2S) were resistant to streptomycin (> 1000 μg mL-1) and one strain (AQLI2) from the same group was resistant to vancomycin (16 μg mL-1). No strains resistant to these two antibiotics were found among the strains from healthy women.

Table 2 Distribution of MIC's to several antibiotics among S. epidermidis isolated from mastitis and healthy women
Full size table

Presence of mecA and SCCmectyping

Among the 41 strains showing oxacillin resistance, the mecA gene could be detected by PCR in 37 (25 from mastitic milk and 12 from milk of healthy women). No amplification was observed in two strains of each group (F12 and CJ9; LI5081 and LC047, respectively), which had shown an oxacillin MIC value of > 2 μg mL-1. In contrast, the mecA gene was detected in five oxacillin susceptible strains, one from a mastitis case (YLIC13) and four from healthy women (LO5RB1, LX5RB3, LV221 and LCC5082).

The type of SCCmec was determined in all the mecA+ strains. The ccrB gene could be amplified from 22 of the 26 mecA+ strains from the mastitis group and, on the basis of the ccrB restriction pattern with HinfI (type IV: 264, 227 and 154 bp; type III: 537 and 106 bp) or with HinfI/BsmI (type IV: 227, 171, 153 and 93 bp; type III: 320, 174, 106 and 44 bp), 19 strains were assigned to type IV and the remaining three (S1LDC12, Z2LDC17 and DF2LAB) to type III (see additional file 1). Amplification of ccrB was achieved in the 16 mecA+ strains from healthy women and, among them, 14 strains showed type IV and two type III SCCmec. The application of the full assay described by Zhang et al. [12] confirmed these results. The 4 mecA+ strains that could not be typed on the basis of their ccrB restriction pattern remained non typeable with the full Zhang et al. assay.

Discussion

S. epidermidis is a normal inhabitant of the skin and mucosal surfaces in healthy hosts and its ubiquitous prevalence as a commensal species makes difficult for a clinician to decide if an isolate represents the etiological agent or a culture contamination [11]. Therefore, isolation of this bacterial species is generally regarded as non-related to a mastitis case, even when it is the only species present in a milk sample [4].

S. epidermidis was the dominant or unique staphylococcal species in breast milk of women suffering mastitis, a finding described previously [4], which indicates that this bacterial species could be an etiological agent of human lactational mastitis. Similarly, several studies have shown the implication of this bacterial species as an agent of mastitis in different animal species [13–15].

The genotyping of the S. epidermidis isolates by PFGE revealed a low diversity within this species in the breast milk of women with mastitis, with a mean of 2 different profiles per sample. A lost in the microbial diversity present in milk of women with mastitis has been previously pointed [4]. Comparison by PFGE dendogram analysis of these strains with those obtained from breast milk of healthy women showed the existence of two main clusters and, within these two groups, the strains generally matched with the origin (mastitis and healthy women). However, a few strains from healthy women grouped together with the mastitis cluster reflecting a similar genetic background. The fact that their presence in milk of healthy women does not constitute a health problem could indicate that a predisposal host is also need for S. epidermidis to develop an infection [16].

Among the S. epidermidis strains analyzed, the presence of adhesion-related genes was high independently of the condition of the women from which they were isolated. The presence of genes encoding cell surface proteins may explain, at least partially, the high prevalence of S. epidermidis in breast milk, mammary areola and ducts of both healthy and mastitis-suffering women. In contrast, the percentage of strains carrying the biofilm-related ica operon was higher in strains from mastitis milk than in that obtained from healthy women. A potential relationship between S. epidermidis infection and the presence of such operon has been reported [8]. In fact, biofilm formation has been described in many cases of staphylococcal mastitis and this is the reason why such property is considered as a potential virulence factor [17, 18].

Strains belonging to either the mastitis or the healthy women group showed a similar susceptibility against many of the antibiotics tested; however, a higher percentage of mupirocin-, erythromycin-, clindamicyn- and oxacillin-resistant strains was found among those isolated from mastitic milk. Resistance to these and other antibiotics in pathogenic S. epidermidis isolates has been reported previously [10, 19]. The resistance of these strains could be partly due to the increasing use of broad-spectrum antibiotics, which encourage selection of multirresistant strains [11]. Improper antibiotherapy may explain why staphylococcal mastitis frequently becomes a chronic and/or recurrent infection. In this study, the presence of mecA gene accompanied with resistance to oxacillin (MIC > 2 μg mL-1) was observed in 62% of the strains from mastitis, but only in 33% from the healthy group. The mecA gene was not detected in four oxacillin-resistant strains. These strains may represent cases of borderline resistance which is characterized by an oxacillin MIC at or just above the susceptibly breakpoint (4 to 8 μg mL-1). In contrast, the mecA gene was detected in five oxacillin-susceptible strains, a fact that has been previously described [20] and that may be due to gene deletions.

Methicillin-resistant Staphylococcus aureus (MRSA) are being reported with increasing frequency in the community and they have been called community-acquired (CA)-MRSA, which are associated with skin and soft tissue infection [21] but are also frequently isolated from healthy hosts [22]. Most of the mecA+ strains used in this study could be ascribed to type IV SCCmec. In S. epidermidis, some studies have reported that SCCmec type IV is generally carried by CA-MRS [23, 24] but this type seems to be predominant among clinically relevant S. epidermidis isolates [9]. The fact that the ccrB gene was not amplified from four mecA+ strains may be due to the presence of different alleles for this gene [25].

In the last years, a renewed medical and research interest has been focused on S. epidermidis since it has become the most important cause of nosocomial infections [6]. The complete genome analysis of some methicillin-resistant S. aureus and S. epidermidis strains of human origin have revealed the propensity of S. aureus to cause fulminant and sometimes life-threatening infections, as opposed to the predisposition of S. epidermidis for chronic and recurrent infections [26]. Identification of S. epidermidis as etiological agents of infection is sometimes hindered by the fact that infections associated with this microorganims are characterized by subtle, non-specific clinical manifestations [5]. Precisely, these characteristics occur in most cases of lactational mastitis. Genome flexibility in S. epidermidis may contribute to the acquisition of some transferable virulence and resistant traits [6, 27] and to the evolution of this species from a commensal to a pathogenic microorganism in susceptible hosts [28]. In fact, it seems that there is an important host factor in lactational mastitis since a woman having such condition usually displays the same clinical signs after subsequent pregnancies. Additionally, it would explain why only a 3–30% of lactating women suffer from such infection when it is the predominant bacterial species found in breast milk of healthy women [29, 30].

Conclusion

Staphylococcus epidermidis is the most prevalent staphylococcal species isolated from breast milk of women suffering mastitis, where it is present at a concentration notably higher than that present in milk of healthy woman (≥ 4.0 versus ≤ 3.0 log10 cfu mL-1, respectively). The percentage of strains showing biofilm production ability and resitance to mupirocin, erythromycin, clindamicyn and/or methicillin was significantly higher among those obtained from women with lactational mastitis than among those isolated from healthy women.

The random method used to select staphylococcal colonies from the samples could introduce a bias regarding the low number of samples from which S. aureus was isolated. Traditionally, S. aureus has been considered as the main etiological agent of mastitis. However, the results of this work suggest that S. epidermidis could be an additional and underrated cause of lactational mastitis; as a consequence, its presence should be also considered in bacteriological analyses of human milk when there is a suspicious of a mastitis infection. Further studies involving a larger number of samples and staphylococcal isolates will be required to confirm the results obtained in this study.

Methods

Samples and isolation of staphylococcal isolates

A total of 30 women aged 25–34 years with clinical symptoms of infectious mastitis participated in the study (Table 1). They were diagnosed by the lactation consultants attending different primary health-care centers in Spain in a 2-months period (October-November 2007). The total staphylococcal count was higher ≥ 4 log10 cfu mL-1 in all their samples. Women with mammary abscesses or any kind of parallel diseases and patients treated with antibiotherapy during the previous two weeks of the study were excluded. All volunteers gave written informed consent to the protocol, which was approved by the Ethical Committee of Hospital Clínico of Madrid (Spain). The milk samples were collected as described previously [31], and plated onto ready to use Baird Parker (BP) plates supplied by bioMérieux (Marcy l'Etoile, France). The plates were incubated in aerobiosis at 37°C for 48 h.

Identification of staphylococci

Initially, a total of 270 isolates (10 from each sample displaying bacterial growth on BP plates) were randomly selected and tested for catalase and coagulase activities and for their resistance to lysozyme and lysostaphin [32]. All of them were subjected to a novel multiplex PCR method designed to allow a rapid identification of S. epidermidis and S. aureus isolates. The new primers (see below) were designed on the basis of the variable regions of the tuf gene sequence of Staphylococcus using the program Clone Manager Suite 7.0 (Sci Ed Central, USA). Their specificity was tested in silico by PCR simulation against up-to-date sequenced prokaryotic genomes using the tools provided in the website http://insilico.ehu.es/PCR. Briefly, a colony of each isolate was suspended in 20 μL of TE buffer (10 M Tris-HCl, 1 M EDTA pH 8) with 0.9% NaCl, and after heating at 98°C for 10 min, the suspension (5 μL) was used as a template for PCR using the BioredMix (BioLine, London, UK) system and the primers (10 μM) tuf-g (5'-GGTGTACCAGCATTAGT-3'), tuf-a (5'-TTCAGTATGTGGTGTAA-3') and tuf-e (5'-TTCGTGCATACCGATGA-3'). The primer pairs tuf-g/tuf-e and tuf-g/tuf-a result in a 370 bp (S. epidermidis) or a 530 bp (S. aureus) fragment [see additional file 2]. PCR conditions were 1 cycle of 94°C for 5 min, 30 cycles of 94°C for 1 min, 48°C for 1 min, and 72°C for 2 min, and a final extension of 72°C for 5 min. Identification of the isolates was confirmed by PCR sequencing of a 470 bp fragment of the 16S rRNA gene using primers and conditions previously described [33]. The amplicons were purified using the Nucleospin® Extract II kit (Macherey-Nagel, Düren, Germany) and sequenced at the Genomics Unit of the Universidad Complutense de Madrid, Spain.

Genotyping of S. epidermidisisolates by pulsed field gel electrophoresis (PFGE)

To determine the diversity of S. epidermidis in breast milk in mastitis infections, 200 isolates of this species obtained from 26 women with mastitis were subjected to PFGE genotyping together with 105 isolates of the same species obtained from breast milk of 12 healthy women within the same period of time [34] (Table 1). Chromosomal DNA was digested with the endonuclease SmaI (New England Biolabs, Ipswich, MA) at 37°C for 16 h. Electrophoresis was carried out in a CHEF DR-III apparatus (Bio-Rad Laboratories, Hercules, CA) for 23 h at 14°C at 6 V cm-1 with pulses from 5 to 50 s. A standard pattern (Lamda Ladder PFG Marker, New England Biolabs) was included in the gels to allow comparison of the digitally normalized PFGE profiles. Computer-assisted analysis of the gels was performed with the Phoretix 1D Pro software (Nonlinear USA, Inc., Durham, NC). PFGE profiles differing in one or more fragments were considered different. Cluster analysis of the PFGE patterns was performed using the UPGMA method based on the Dice similarity coefficient.

Screening for potential virulence determinants

On the basis of the different PFGE profiles, 76 strains (40 from mastitis cases and 36 from healthy women) were further selected and characterized. Presence of genes embp, fbe, atlE and icaD, which respective products are involved in adhesion and biofilm formation, was evaluated using primers couples described previously [7, 35–37]. In the case of fbe, atlE and icaD, a multiplex PCR format was designed using the following conditions: 5 min at 94°C followed by 30 cycles of 94°C for 1 min, 60°C for 30 s, 72°C for 1 min and, then, a final extension of 5 min at 72°C [see additional file 3]. Strains isolated from mastitic milk were compared to those isolated from healthy women. Data were subjected to a statistical analysis using the Chi-square test (SPSS package, SPSS Inc, Chicago, IL, USA). Differences were considered significant if P values were lower than 0.05.

Phenotypic assays

The hemolytic activity of the isolates was determined on Columbia agar supplemented with 5% horse blood (COH, bioMériux) after incubation at 37°C for 72 h following a procedure previously described [32]. The ability of the isolates to form slime was assessed using the Congo Red agar assay (CRA) [38]. The plates were incubated at 37°C for 24 h and, then, for additional 24 h at room temperature.

Determination of MIC's to antibiotics

The determination of the MIC's to several antibiotics commonly used against staphylococcal infections was evaluated by a microdilution method using the Sensititre plates Staenc1F (Trek Diagnostic Systems, Cleveland, OH) following the manufacturer's instructions. The antibiotics analyzed were: penicillin, ampicillin, amoxycillin-clavulanic acid, teicoplanin, chloramphenicol, erythromycin, mupirocin, streptomycin, gentamicin, clindamycin, oxacillin, ciprofloxacin, fosfomycin, imipenem, nitrofurantoine, trimethoprim-sufamethoxazole, tetracycline, vancomycin, linezolid, quinupristin-dalfopristin and rifampin. Data were submitted to the statistical analysis described above.

Screening for mecA gene and typing of the staphylococcal chromosome cassette mec (SSCmec)

Presence of the mecA gene was evaluated by PCR using primers mecA forward (5'-GGTCCCATTAACTCTGAAG-3') and mecA reverse (5'-AGTTCTGCAGTACCGGATTTTGC-3'), which results in a 1,040 bp fragment [39]. The SCCmec was subjected to a typing procedure [40], which implied the PCR amplification of the ccrB gene followed by RFLP analysis using endonucleases HinfI and BsmI. Presence of mecA and SCCmec typing was confirmed using all the primers and conditions described by Zhang et al. [12].