Bovine whey proteins inhibit the interaction of Staphylococcus aureus and bacteriophage K

Abstract

Aims:  To understand the potential use of bacteriophage K to treat bovine Staphylococcus aureus mastitis, we studied the role of whey proteins in the inhibition of the phage–pathogen interaction in vitro.

Methods and Results:  The interaction of bacteriophage K and S. aureus strain Newbould 305 was studied in raw bovine whey and serum. Incubation of S. aureus with phage in whey showed that the bacteria are more resistant to phage lysis when grown in whey and also bovine serum. Whey collected from 23 animals showed a wide variation in the level of phage-binding inhibition. The role of the protein component of milk whey in this inhibition was established; treatment of the whey by heat, proteases and ultrafiltration removed the inhibitory activity. Brief exposure of S. aureus cells to whey, followed by resuspension in broth, also reduced phage binding. Microscopy showed the adhesion of extracellular material to the S. aureus cell surface following exposure to whey. Chromatographic fractionation of the whey demonstrated that the inhibitory proteins were present in the high molecular weight fraction.

Conclusions:  The adsorption of whey proteins to the S. aureus cell surface appeared to inhibit phage attachment and thereby hindered lysis. The inhibitory whey proteins are of high molecular weight in their native form and may sterically block phage attachment sites on the cell surface.

Significance and Impact of the Study:  These findings have implications for any future use of phage therapy in the treatment of mastitis, and other diseases, caused by S. aureus. This pathogen is predicted to be much more resistant to phage treatment in vivo than would be expected from in vitro broth culture experiments.

Introduction

Bovine mastitis, or the inflammation of the bovine mammary gland due to pathogen invasion, is a major concern of the dairy industry. In the United States, as many as 50% of all dairy cattle experience some form of mastitis at any given time (Wilson et al. 1997). While bovine intramammary infections (IMI) can be caused by a wide variety of bacteria, fungi and mycoplasmas, Staphylococcus aureus is a pathogen of concern. Staphylococcus aureus has been implicated in 7–44% of clinical mastitis cases (Sargeant et al. 1998; Waage et al. 1999). Cure rates for antibiotic treatment of S. aureus IMI may range from 20% to 78% (Sol et al. 1994; Osteras et al. 1999; Dingwell et al. 2003) and are affected by many factors (Schukken et al. 2001). Staphylococcus aureus is also a significant human pathogen, and the continued emergence of multidrug-resistant S. aureus strains is a growing concern (Ayliffe 1997; Coast and Smith 2003). The advent of multidrug-resistant strains of S. aureus has prompted the search for alternative methods of treatment. One such alternative to chemical antibiotics is the use of bacteriophages.

Bacteriophage therapy is the use of bacteriophages in the treatment of a bacterial infection. Recent interest in phage therapy was sparked by some early success in the treatment of Escherichia coli infections in animal models (Smith and Huggins 1982, 1983; Merril et al. 1996). As phage therapy has been investigated in greater detail, however, variable success rates have been reported depending on the pathogen studied, its route of inoculation and method of phage treatment (Soothill 1992; Cerveny et al. 2002). In the case of bovine mastitis, Lerondelle and Poutrel (1980) reported poor efficacy of phage therapy against experimentally induced S. aureus infections. This mixture of success and failure is mirrored in the early phage therapy literature; for example, the success of phage therapy was reportedly higher against enteric pathogens than against staphylococcal wound infections (Krueger and Scribner 1941).

Bacteriophages are currently being evaluated as an alternative to chemical antibiotics for the control of bovine mastitis caused by S. aureus. Because of the previously reported problems encountered in this system (Lerondelle and Poutrel 1980), and the generally limited understanding of the dynamics of phage replication in vivo, a strategy of in vitro testing of phage activity in a model system prior to in vivo clinical trials has been adopted. This model consisted of S. aureus strain Newbould 305, which is commonly used in bovine mastitis trials (Schukken et al. 1999) and bacteriophage K. Phage K is a strictly lytic member of the myoviridae with a broad host range amongst S. aureus strains (O'Flaherty et al. 2004). The origins of phage K are not precisely known, but it was likely originally isolated by André Gratia in the early 1920s (Burnet and Lush 1935; Rountree 1949). Preliminary work examining the interactions of S. aureus and bacteriophage K in whole raw milk (Gill et al. 2001) showed a strong inhibition of phage lytic activity in this system, an observation which has been noted by other workers (Das and Marshall 1967; O'Flaherty et al. 2005). This system was refined to focus exclusively on raw milk whey, which is the soluble protein fraction of milk remaining after the removal of fat and casein.

Materials And Methods

Strains and culture conditions

Staphylococcus aureus strain Newbould 305 (ATCC 29740), and S. aureus bacteriophage K (ATCC 19685-B1) were used. Newbould 305 is a pathogenic S. aureus strain of bovine origin that is used in mastitis challenge trials (Schukken et al. 1999). Cells were grown on tryptic soy broth (TSB) or tryptic soy agar (TSA) (Difco/BD, Franklin Lakes, NJ, USA) at 37°C. Staphylococcus aureus was grown and maintained for routine use on TSA plates stored at 4°C; plates were streaked from freezer stocks held at −80°C in 15% (v/v) glycerol. Phage K was cultured by standard methods using S. aureus Newbould 305 as host (Adams 1959); phage lysate was filter-sterilized by an 0·22-μm filter (Millipore, Nepean, ON, Canada) and stored in TSB at 4°C. Unless otherwise stated, cells were cultured by picking a single S. aureus colony from a TSA plate, inoculating into 3 ml of TSB and incubating for 18–20 h at 37°C with vigorous shaking. These overnight cultures were then subcultured at a 1 : 50 dilution into fresh TSB and grown to an A₆₀₀ of 1·0 (c. 1 × 10⁹ CFU ml⁻¹). For the preparation of fixed S. aureus cells, cultures were grown to an A₆₀₀ of 1·0 in TSB and washed twice in cold sterile phosphate-buffered saline (PBS). Cells were resuspended in PBS containing 1% formaldehyde and incubated at 4°C for 18 h. Cells were washed three times in cold sterile PBS before use.

Milk and milk fractions

Raw bovine milk was collected aseptically from each quarter of cows at the Elora Dairy Research Centre, University of Guelph. Cow parity ranged from 1 to 8 (mean parity 2·9), and days in milk at the time of sample collection ranged from 9 to 368 (mean 174·6). The most recent somatic cell count (SCC) for each cow at the time of collection was determined from the cow's Dairy Herd Improvement test results, taken 1–4 weeks prior to sample collection. The most recent SCC ranged from 1·7 × 10⁴ cells ml⁻¹ to 7·36 × 10⁵ cells ml⁻¹, with a mean of 1·81 × 10⁵ cells ml⁻¹. Each quarter sample was screened for microbial contamination by plating 20 μl to TSA and incubating at 37°C for 18–20 h. Quarter milk samples, which produced more than ten colonies per 20 μl, were discarded. Remaining quarter samples were pooled by animal and skimmed by centrifugation at 5000 g for 10 min at 4°C, followed by removal of the milk while carefully avoiding the cell pellet. Whey was produced by addition of 0·75 mmol l⁻¹ phenylmethanesulfonyl fluoride (PMSF) (Sigma-Aldrich Canada, Oakville, ON, Canada) and 0·02% (v/v) Chy-Max Liquid (Chr. Hansen, Toronto, ON, Canada) to the skimmed milk followed by incubation at 37°C for 1–2 h to allow the curd to set. The milk was then centrifuged at 6500 g for 15 min at 4°C and the supernatant whey was removed. Whey was sterilized by vacuum-driven filtration through a 0·22-μm filter in either a Steriflip or Stericup device (Millipore). Boiled whey was produced by heating whey to 100°C for 10 min in a water bath, followed by centrifugation at 8000 g for 10 min at 4°C to remove the precipitate. Whey was protease-treated using proteinase K at 1 mg ml⁻¹ at 37°C for 6 h. Digestion was halted by adding PMSF to a final concentration of 1 mmol l⁻¹. Whey ultrafiltrate was produced by filtering raw whey through a 10-kDa nominal weight cut-off (NWCO) membrane in a Centricon column (Millipore) according to the manufacturer's instructions. Whey was panned by incubation of 5-ml whey with fixed S. aureus cells at a concentration of c. 5 × 10¹⁰ cells ml⁻¹ at room temperature with gentle rocking for 15 min, followed by removal of the cells by centrifugation at 5000 g for 5 min and filter sterilization.

Size exclusion chromatography

Whey was dialysed against running buffer (0·5 mol l⁻¹ NaCl, 0·1 mol l⁻¹ Tris-HCl, pH 7·0) at 4°C for 18 h and 5 ml was applied to a 26-mm ID column (Amersham Biosciences, Baie d'Urfé, QC, Canada), using a 5-ml sample loop in a method based on Felipe and Law (1997). The column was packed with a 335-ml bed volume of Sephacryl 300-HR (Sigma-Aldrich). The column was driven at a rate of 1 ml min⁻¹ (0·19 cm min⁻¹) by a P-1 peristaltic pump (Amersham) and monitored at 280 nm with an in-line LKB Bromma 2238 Uvicord-SII optical detector linked to a 2210 two-Channel analog data recorder (LKB Bromma). The column was calibrated with molecular weight standards: cytochrome C (12·4 kDa), bovine serum albumin (BSA; 66 kDa), β-amylase (200 kDa) and blue dextran (2 MDa) (all Sigma). Fractions were collected at 5-min intervals using a 2111 Multitrac automated fraction collector (LKB Bromma). Eluted fractions were pooled and concentrated using a 10-kDa NWCO Centricon ultrafiltration columns (Millipore). Fractions were resuspended in assay buffer (100 mmol l⁻¹ NaCl, 10 mmol l⁻¹ CaCl₂, 5 mmol l⁻¹ MgCl₂, 10 mmol l⁻¹ Tris-HCl pH 7·2) for binding assays or whey ultrafiltrate for minimum inhibitory concentration (MIC) assays and filter-sterilized for further experiments.

Phage MIC assays

Staphylococcus aureus cultures of A₆₀₀ = 0·8 were diluted by 10⁻³ in PBS and added to whey, chromatography fractions or TSB at a 100 × dilution such that the final concentration was c. 7 × 10³ CFU ml⁻¹. Since some whey proteins are derived from serum, adult bovine serum (Sigma), newborn calf serum (Sigma), and fetal calf serum (Invitrogen Canada, Burlington, ON, Canada) were also evaluated. The inoculated medium was aliquoted at 180 μl per well to untreated, sterile, 96-well polystyrene plates (Costar, Cole-Parmer, Vernon Hills, IL, USA). Plates were incubated for 2·5 h at 37°C in a humidified incubator with an atmosphere supplemented with 5% CO₂ to ensure uniform growth of the bacteria in all samples. Serial tenfold dilutions of phage K lysate in SM buffer (100 mmol l⁻¹ NaCl, 10 mmol l⁻¹ Tris-HCl pH 7·2, 10 mmol l⁻¹ MgCl₂, 0·01% w/v gelatin) were added at 20 μl per well to give final concentrations of 1 × 10² PFU ml⁻¹ to 1 × 10⁸ PFU ml⁻¹. One well for each treatment was inoculated with TSB alone as a control. After incubation for a further 18 h, plates were scored by the presence or absence of visible bacterial growth at the bottom of each well. All assays were replicated three times.

Phage-binding assays

Assays to determine the degree of phage-binding inhibition were based on binding rate assays as described by Adams (1959). Staphylococcus aureus cultures of A₆₀₀ =1·0 were prepared as described previously. One hundred microlitres of culture was added to 900 μl of the solution to be tested to give a final concentration of c. 1 × 10⁸ CFU ml⁻¹, and incubated on ice for 10 min. Inoculated samples were prewarmed to 37°C for 2 min in a heat block. Following removal of a 50-μl sample to be used to determine the initial cell count, 50 μl of a 1 × 10¹⁰ PFU ml⁻¹ bacteriophage K lysate was immediately added to give a final multiplicity of infection (MOI) of 5. The suspension was mixed thoroughly. After 5-min incubation at 37°C, a 50-μl sample was removed and diluted by 100× in TSB to stop further phage binding. The number of surviving bacteria was evaluated by serial dilution and plating to TSA plates. The data were expressed as the CFU ml⁻¹S. aureus enumerated after incubation with phage divided by the CFU ml⁻¹ enumerated immediately prior to the addition of phage, multiplied by 100. Whey-exposed cells were prepared by incubating cells in whey for 10 min, as described previously, then pelleting and resuspending them in TSB prior to initial sampling and addition of the phage. Solutions evaluated for binding inhibition were raw whey and whey fractions prepared as described previously. As controls, bovine serum (Sigma), newborn calf serum (Sigma), and fetal calf serum (Invitrogen) were included. TSB was used as a negative control in all cases, except in the assay of chromatography fractions, when assay buffer was used as the control. Each assay was replicated three times.

Fluorescence microscopy

Ten millilitres of whole whey was dialysed against a volume of 1 l of PBS, pH 7·2, for 18 h at 4°C across a SpectraPor 6–8 kDa dialysis membrane (Spectrum Laboratories, Rancho Dominguez, CA, USA). Whey was then labelled with the reactive fluorescein compound 5-([4,6-dichlorotriazin-2-yl]amino)fluorescein (DTAF) (Sigma) at 0·4 mg ml⁻¹ for 40 min at room temperature. Labelling was halted by addition of hydroxylamine to a final concentration of 0·15 mol l⁻¹. Excess probe and hydroxylamine were removed by dialysis as described earlier, and the labelled whey was reconcentrated to its original volume by ultrafiltration. Log-phase S. aureus cells grown in TSB liquid culture (A₆₀₀ = 0·8) were centrifuged at 8000 g for 5 min, resuspended in the DTAF-labelled whey and incubated on ice for 15 min. Cells were then centrifuged again and resuspended in PBS and observed directly as wet mounts in a Zeiss Axioskop 2 with excitation at 470 nm and emission at 525 nm. Images were captured using a Spot RTcolor model 2·2·0 digital camera and Spot software version 4·0·8 (Diagnostic Instruments, Sterling Heights, MI, USA).

Electron microscopy

Staphylococcus aureus cultures (A₆₀₀ = 0·8) were resuspended in whole raw whey or sterile PBS containing 10 mg ml⁻¹BSA (Sigma) on ice for 10 min, washed twice in PBS and dropped onto polished 13-mm graphite planchets with a contact time of 30 min. Planchets were washed twice in 140 mmol l⁻¹ phosphate buffer pH 7·2 (PB) and fixed in a mixture of 2% glutaraldehyde and 1% OsO₄ in PB for an hour at room temperature. Planchets were washed twice in PB, once in distilled water, dehydrated in a graded ethanol series and critical point dried with CO₂. Planchets were affixed to aluminum stubs with Leit-C carbon cement (Neubauer Chemikalen, Munster, Germany) and sputter-coated with Au–Pd. Samples were viewed in a Hitachi F4500 field emission scanning electron microscope.

Statistical analysis

Differences between treatments were determined by paired t-tests. Correlations between binding inhibition level and cow history were determined by Spearman rank order analysis. All statistical analyses were conducted using Statistica version 4·5 (StatSoft, Tulsa, OK).

Results

When grown in raw milk whey in the MIC assay, S. aureus Newbould 305 was eliminated at low efficiency by bacteriophage K at a wide range of multiplicities of infection. This effect was found to be variable between different animals; however, in no animal was this inhibition found to be absent when compared with the broth control. A 100- to 1000-fold increase in phage concentration was required to eliminate visible S. aureus growth in raw milk whey collected from 15 different animals when compared with broth culture (Table 1). Adult and newborn calf bovine sera were found to be the strongest inhibitors of phage activity. However, fetal bovine serum was less inhibitory, inhibiting phage lysis to a level comparable with that seen in bovine whey (Table 1).

Table 1.   Minimum inhibitory concentrations of bacteriophage K required to eliminate visible growth of Staphylococcus aureus Newbould 305 when cultured in raw whey, bovine serum or a tryptic soy broth (TSB) control. Results are the product of three independent replicates

Treatment	Log phage concentration (PFU ml^−1)
	8	7	6	5	4	3	2	0
‘−‘indicates the lack of visible growth at this condition in two of three or three of three replicate experiments. ‘+’ indicates visible growth at this condition in all three replicate experiments. ‘(+)’ indicates visible growth at this condition in only two of three replicate experiments.
Whey, cow 2692	−	−	+	+	+	+	+	+
Whey, cow 2843	−	−	+	+	+	+	+	+
Whey, cow 2844	−	−	+	+	+	+	+	+
Whey, cow 2875	−	−	+	+	+	+	+	+
Whey, cow 2877	−	+	+	+	+	+	+	+
Whey, cow 2883	−	(+)	+	+	+	+	+	+
Whey, cow 2936	−	(+)	+	+	+	+	+	+
Whey, cow 2949	−	(+)	+	+	+	+	+	+
Whey, cow 2959	−	+	+	+	+	+	+	+
Whey, cow 2983	−	−	+	+	+	+	+	+
Whey, cow 2988	−	+	+	+	+	+	+	+
Whey, cow 3039	−	−	+	+	+	+	+	+
Whey, cow 3064	−	−	(+)	+	+	+	+	+
Whey, cow 3131	−	(+)	+	+	+	+	+	+
Whey, cow 3194	−	(+)	+	+	+	+	+	+
TSB	−	−	−	−	(+)	+	+	+
Fetal calf serum	−	−	+	+	+	+	+	+
Newborn calf serum	+	+	+	+	+	+	+	+
Adult bovine serum	+	+	+	+	+	+	+	+

To further understand the nature of this inhibition, a series of experiments were performed to evaluate the effects of whey on the binding kinetics of the phage. This assay measured the rate of phage attachment by determination of the proportion of S. aureus which survived exposure to phage K at a set duration and concentration. This approach was found to produce results comparable with those obtained when free (unattached) phages were enumerated following co-incubation with bacteria (data not shown). Under the conditions used, the mean bacterial survival rate in the TSB control was 22·2% (Fig. 1). In contrast, the survival rates of S. aureus exposed to phage while suspended in raw whey collected from 23 different animals ranged from 21·8% to 102·6%, depending on the source animal (Fig. 1). Bovine serum also appeared to strongly inhibit phage attachment, with an average 94·1% of phage-exposed cells surviving treatment (Fig. 1).

Figure 1.  Inhibition of phage K binding to Staphylococcus aureus Newbould 305 in raw whey collected from 23 different animals, bovine serum and a tryptic soy broth control. Numbers are expressed as the percentage of bacteria remaining after 5-min incubation with phage K; error bars indicate SD. A high number of surviving bacteria indicates poor phage binding.

The level of binding inhibition exhibited by the whey of different animals was compared with other cow factors, including cow parity, days in milk, and measures of milk SCC by Spearman rank-order analysis. Data from 22 animals were included in this analysis. No statistically significant correlations were found between levels of binding inhibition and cow parity (P = 0·169), or inhibition and days in milk (P = 0·877). A significant positive correlation was found between levels of binding inhibition and the cow's lifetime average SCC (P = 0·002), and the cow's history of clinical mastitis (P = 0·050).

The mechanism by which whey components inhibit phage binding to the S. aureus cell surface was investigated. Raw whey collected from cows 2603 and 2959, which showed a strong ability to inhibit phage binding, were selected for further study. In one set of experiments using whey collected from cow 2603, S. aureus cells which were exposed to whey for 10 min and resuspended in TSB still exhibited a low rate of phage binding (Fig. 2). In a second experiment, a large excess of fixed S. aureus cells was briefly incubated with raw whey and then removed by centrifugation before evaluation in the standard binding assay. This ‘panning’ treatment restored phage-binding activity to the same level as that found in TSB (Fig. 2). Differences in phage-binding activity between whey treated by panning and untreated whey were statistically significant (P = 0·0095).

Figure 2.  Inhibition of phage K binding to Staphylococcus aureus Newbould 305 cells in raw whey, cells briefly exposed to whey and resuspended in TSB, and whey in which inhibitory activity has been removed by exposure to a large excess of S. aureus cells, or ‘panning’, and a tryptic soy broth control. Numbers are expressed as the percentage of bacteria remaining after 5-min incubation with phage K; error bars indicate SD. A high number of surviving bacteria indicates poor phage binding.

The nature of the phage-binding inhibitor in raw milk whey was also investigated. In a set of experiments using whey collected from cow 2959, treatment of the whey by boiling, proteinase K or ultrafiltration increased phage-binding rates to levels comparable with those found in TSB (Fig. 3). Differences in phage-binding activity between treated whey and untreated whey were statistically significant (P < 0·001). Fractionation of whey by ultrafiltration and treatment of the whole whey by boiling also lowered the whey's inhibitory properties in the MIC assay by 3–4 log units compared with the untreated control whey; treatment by panning reduced the MIC by 2–3 log units (Table 2).

Figure 3.  Inhibition of phage K binding to Staphylococcus aureus Newbould 305 in raw whey and whey treated by protease digestion, boiling and ultrafiltration, and a tryptic soy broth control. Numbers are expressed as the percentage of bacteria remaining after 5-min incubation with phage K; error bars indicate SD. A high number of surviving bacteria indicates poor phage binding.

Table 2.   Minimum inhibitory concentrations of bacteriophage K required to eliminate visible growth of Staphylococcus aureus Newbould 305 when cultured in whole raw whey, raw whey fractions separated by size exclusion chromatography, panned whey, boiled whey and raw whey ultrafiltrate. Panning refers to the method of incubating whey with a large excess of S. aureus cells followed by their removal by centrifugation

Treatment	Log phage concentration (PFU ml^−1)
	8	7	6	5	4	3	2	0
‘−’ indicates the lack of visible growth at this condition in two of three or three of three replicate experiments. ‘+’ indicates visible growth at this condition in all three replicate experiments. ‘(+)’ indicates visible growth at this condition in only two of three replicate experiments.
Whole whey	−	+	+	+	+	+	+	+
Fraction I	−	−	+	+	+	+	+	+
Fraction II	−	−	−	+	+	+	+	+
Fraction III	−	−	−	+	+	+	+	+
Fraction IV	−	−	−	+	+	+	+	+
Fraction V	−	−	−	+	+	+	+	+
Whey ultrafiltrate	−	−	−	−	+	+	+	+
Panned whey	−	−	−	(+)	+	+	+	+
Boiled whey	−	−	−	−	−	−	+	+
TSB	−	−	−	−	−	+	+	+

Raw whey was fractionated by size exclusion chromatography in order to determine which sub-components of the whey protein were responsible for the inhibition of phage lysis. The elution profile of the raw whey produced three major peaks, at 125, 225 and 250 ml (Fig. 4). The elution profile was highly reproducible between runs (data not shown). The 125-ml peak corresponds to the column's void volume, which was presumed to be the 1·5 MDa exclusion limit of the medium as stated by the manufacturer. Fractions corresponding to the eluted peaks were pooled for further analysis, resulting in five pooled fractions, designated as fractions I–V (Fig. 4). Phage-binding inhibition assays conducted on the pooled chromatography fractions showed that the majority of the inhibitory activity eluted in the 125-ml peak, contained in fraction I (Fig. 5). When these same fractions were evaluated for phage inhibition in the MIC assay, a similar pattern was observed: S. aureus grown in fraction I was more resistant to phage lysis than cells grown in the remaining eluted fractions (Table 2). However, none of the whey fractions tested exhibited an MIC as high as the whole whey, nor as low as the whey ultrafiltrate.

Figure 4.  Elution profile, molecular weights and collected fractions from raw milk whey separated by size exclusion chromatography in a 335-ml bed volume of Sephacryl S300-HR at a linear flow rate of 0·19 cm min⁻¹.

Figure 5.  Phage-binding inhibition exhibited by raw whey fractions separated by size exclusion chromatography, as compared with the whole whey and assay buffer alone. Binding assays were replicated three times in each of two trials; trials indicate two separate column runs. Numbers are expressed as the percentage of bacteria remaining after 5-min incubation with phage K; error bars indicate SD. A high number of surviving bacteria indicates poor phage binding.

As shown in Figs 6 and 7, S. aureus cells rapidly acquired a coat of material when briefly exposed to raw whey. Figure 6 shows the accumulation of a fluorescent halo surrounding the S. aureus cells following brief incubation in fluorescein-labelled whey. Scanning electron microscopy revealed that cells incubated in whey acquired a copious amount of material adhering to the cell surface, and an overall rough appearance when compared with control cells, which had not been exposed to whey (Fig. 7). Cells which had been exposed to whey also showed a strong tendency towards agglutination (Figs 6 and 7).

Figure 6.  Epifluorescence microscopy of live Staphylococcus aureus cells exposed to whole raw whey labelled with fluorescein. Log-phase S. aureus cells were exposed to labelled whey for 15 min on ice. Note the fluorescent halo of labelled protein surrounding the cells.

Figure 7.  Field-emission scanning electron micrographs of Staphylococcus aureus cells. (a) Cells incubated in whole raw whey for 10 min and washed twice. (b) Cells incubated in buffer containing 10 mg ml⁻¹ of bovine serum albumin and washed twice. Note the accumulation of extracellular material from the whey in (a).

Discussion

As an obligate intracellular parasite, the ability of a phage to infect, replicate within and lyse its host is intimately connected to the physiology and habitat of the bacterium. In the interest of developing phages for the treatment of bacterial infections, the study of the phage-bacterium interaction in in vitro broth systems may have little applicability to the true dynamics of phage therapy in vivo. For example, Chibani-Chennoufi et al. (2004) noted that phage-sensitive E. coli which was freshly inoculated into the guts of mice was susceptible to phage lysis in vivo, while phage-sensitive resident E. coli was not. In S. aureus, growth in serum has been shown to induce changes in gene expression (Wiltshire and Foster 2001), which may affect phage sensitivity. Growth of the bacterium in the presence of 5% CO₂ has also been demonstrated to alter bacterial capsule expression in a strain-dependent manner (Herbert et al. 2001). In the present study, it was found that S. aureus strain Newbould 305 would not produce visible growth in the whey taken from some animals without the presence of 5% CO₂. Growth of S. aureus in milk whey has been shown to alter surface characteristics (Mamo and Froman 1994), promote growth agglutination (Baselga and Amorena 1990) and even enhance bacterial virulence (Mamo et al. 1991). For S. aureus bacteriophages, studies have shown the apparent inhibition of phage lytic activity in the presence of raw milk and whey (Das and Marshall 1967; O'Flaherty et al. 2005).

The MIC of phage K against S. aureus Newbould 305 was 2–4 log units higher in bovine whey and serum than in the broth control (Table 1). This observation suggested that the cause of the inhibition was due to interference with some aspect of the phage's lytic cycle. The lytic cycle of the phage consists of three major phases: binding to a suitable host bacterium and injection of its genome; a period of intracellular production of new virions; lysis of the cell and release of progeny phage into the environment. To further understand the mechanism by which S. aureus resisted phage lysis in the presence of whey, the first of these stages, phage binding, was investigated. It was found that the whey collected from several animals strongly inhibited the binding of phage K to the S. aureus cell, while that from others did not (Fig. 1).

Phage binding was inhibited in the presence of raw whey, but inhibition was also observed when cells were briefly exposed to whey and subsequently resuspended in TSB (Fig. 2). The inhibition was not attributable to degradation of the phage in the presence of bovine whey, phage K was found to be relatively stable in whey at 37°C (data not shown). This indicated that inhibitory whey components had attached to or modified the S. aureus cell surface. Incubation of raw whey with a large excess of dead, fixed S. aureus cells, followed by their removal by centrifugation in a ‘panning’ procedure, also ablated the inhibition of phage binding (Fig. 2). As shown in Fig. 3, the inhibitor's sensitivity to heat, protease and removal by ultrafiltration indicate that it is mediated by a protein or group of proteins. Recent work by O'Flaherty et al. (2005) has also shown that phage-inhibitory activity in milk is heat labile.

Based on these observations, a model is proposed whereby the binding of phage K to S. aureus Newbould 305 is inhibited by the adsorption of as-yet unidentified whey proteins to the cell surface. These adsorbed proteins may interfere with phage binding by sterically hindering access of the phage tail, and its associated receptors, to the cell surface. Micrographs (Figs 6 and 7) clearly show a layer of extracellular material deposited on the cell surface after brief exposures to whey. It was also shown by Massey et al. (2002) that S. aureus can acquire a dense surface layer when grown in human peritoneal dialysate, and that this interferes with the cell's ability to bind fibronectin. A parallel phenomenon may be at work in the case of S. aureus in milk whey, preventing the bacteriophage from reaching the cell surface. The tendency of S. aureus cells to agglutinate following exposure to raw whey (Fig. 6) had been observed previously (Baselga and Amorena 1990; O'Flaherty et al. 2005), and is also likely involved in the inhibition of phage K in raw whey. The phenomenon of bacterial aggregation in milk has been observed for other bacterial species (Korhonen et al. 2000). Stadhouders (1963) noted the agglutination of dairy starter cultures in the presence of skim milk. In the case of oral streptococci, this aggregation has been attributed partly to the action of specific antibodies, and partly to the action of unknown milk components (Loimaranta et al. 1998).

In the whey that was fractionated by chromatography, fraction I contained the highest levels of binding inhibitory activity (Fig. 5) and eluted at an apparent molecular weight of 1·5 MDa or more (Fig. 4). This fraction was also the most inhibitory in the MIC assay (Table 2). Whey from which the protein was removed by ultrafiltration had the lowest inhibitory activity in both assays (Fig. 3 and Table 2). Additionally, the ‘panning’ treatment, whereby binding inhibition was ablated by incubation of the whey with a large excess of fixed S. aureus cells, also allowed phages to lyse S. aureus in the MIC assay with much greater efficiency than observed in the untreated whey (Table 2). These observations suggest that the inhibition of phage binding and the inhibition of phage lysis in the MIC assay are both linked to the presence of the same group of whey proteins.

Staphylococcus aureus is known to carry receptors on its surface which are able to specifically bind a wide variety of host proteins, such as fibronectin, collagen, fibrinogen and IgG, amongst many others (Mazmanian et al. 2000). Detailed analyses of human milk whey have shown that it contains at least 400 different protein isoforms (Murakami et al. 1998), and it is not unreasonable to assume that this level of complexity exists in bovine whey as well. Some whey proteins are derived from the animal's serum, and it is interesting to note that the inhibitory effects observed in milk whey also appear in serum. The inhibitory proteins contained in chromatography fraction I eluted at an apparent molecular weight of 1·5 MDa, which suggests that the major soluble IgG class present in bovine whey, IgG1 (molecular weight c. 180 kDa) (Korhonen et al. 2000), is not responsible for the inhibition observed. Preliminary characterization of the fraction I proteins by MALDI-TOF mass spectrophotometry suggests that they are proteins typically associated with the milk fat globule membrane. In this case, the apparent high molecular weight of the inhibitory proteins can be explained by the possibility that they are embedded in small membrane fragments, or are aggregated to one another via their hydrophobic transmembrane domains.

Whey taken from several animals (e.g. cows 3064, 3131 and 3194) does not appreciably inhibit phage binding (Fig. 1), while still inhibiting phage-mediated lysis in a longer-term growth experiment (Table 1). Adsorption of whey protein to the S. aureus cell surface appears to be responsible for the inhibition of phage binding. However, the mechanism by which whey with low or no binding inhibition still protects S. aureus in the MIC assay is not known. It is possible that the whey from such animals also contains protective levels of phage binding-inhibitory protein, but in concentrations too low to be detected in the short exposures used in the phage-binding assay. Physiological alterations of S. aureus cell surface properties, possibly involving capsule formation, have been previously reported following growth in whey (Watson and Watson 1989; Hallen Sandgren et al. 1991; Mamo and Froman 1994). Staphylococcus aureus capsule has been shown to inhibit phage binding (Wilkinson and Holmes 1979).

In conclusion, it is clear that the behaviour of the S.aureus–bacteriophage system undergoes dramatic changes in ex-vivo media. The lack of lytic efficiency of phage K in the presence of whey is attributable to a reduction in the phage-binding rate due to interference by whey proteins. The major mechanism for this interference appears to be the prevention of phage attachment to the cell surface, although other complementary mechanisms cannot be ruled out at this time. While it does not render the bacterium completely immune to phage attack, this phenomenon has clear implications for the practice of phage therapy against S. aureus not only in the treatment of mastitis, but in other clinical situations as well. This in vivo behaviour of the phage–bacteria system is predicted to be markedly different from that expected based on observations made in a standard in vitro broth culture model. Higher phage titres, nonstandard treatment regimens or adjuvants, which increase phage efficiency, may be required for the successful treatment of mastitis cases in vivo.

Acknowledgements

The authors would like to acknowledge Dion Lepp and Jennifer C. Pacan for their technical assistance, and Dr Alexandra K. Smith for her assistance with the electron microscopy. We also thank Angela Fairfield and Laura Wright for providing milk samples. J.J. Gill received financial assistance from the Dairy Farmers of Ontario.