ABSTRACT
Colonization by Staphylococcus aureus is a characteristic feature of several inflammatory skin diseases and is often followed by epidermal damage and invasive infection. In this study, we investigated the mechanism of skin colonization by a virulent community-acquired methicillin-resistant S. aureus (CA-MRSA) strain, MW2, using a murine ear colonization model. MW2 does not produce a hemolytic toxin, beta-hemolysin (Hlb), due to integration of a prophage, ϕSa3mw, inside the toxin gene (hlb). However, we found that strain MW2 bacteria that had successfully colonized murine ears included derivatives that produced Hlb. Genome sequencing of the Hlb-producing colonies revealed that precise excision of prophage ϕSa3mw occurred, leading to reconstruction of the intact hlb gene in their chromosomes. To address the question of whether Hlb is involved in skin colonization, we constructed MW2-derivative strains with and without the Hlb gene and then subjected them to colonization tests. The colonization efficiency of the Hlb-producing mutant on murine ears was more than 50-fold greater than that of the mutant without hlb. Furthermore, we also showed that Hlb toxin had elevated cytotoxicity for human primary keratinocytes. Our results indicate that S. aureus Hlb plays an important role in skin colonization by damaging keratinocytes, in addition to its well-known hemolytic activity for erythrocytes.
INTRODUCTION
Whole-genome sequencing projects for Staphylococcus aureus have revealed much information on the genomic constitution of this clinically important microorganism (1–9). However, there is a great gap in our understanding of the correlation between the structural features of S. aureus genomes and the clinical symptoms that are brought about during the course of infection.
Anterior nares are the major reservoir of S. aureus: 20% of humans are persistently and asymptomatically colonized and 60% are intermittently colonized, whereas 20% are considered noncarriers (10). Nasal carriage of S. aureus is a major risk factor for infections (11). Commensal carriage of methicillin-resistant S. aureus (MRSA) in healthy individuals remains low (from 0.2 to 2.8%) but constitutes a greater risk for subsequent infection than the carriage of methicillin-susceptible S. aureus (MSSA) (12). Several bacterial factors, including wall teichoic acid (WTA) (13), clumping factor B (ClfB) (14), capsular polysaccharide (Cap) (15), iron-regulated surface determinant (IsdA) (16), and upregulator of autolytic activity (SceD) (17), were determined to be involved in S. aureus nasal colonization. All the factors listed above have been studied both in vitro and in animal models so far, and mice have been mainly employed as a good model for S. aureus colonization. Human nasal carriers of S. aureus are typically free from any recognizable symptoms; however, the organisms tend to produce lesions once they migrate into other tissues. Therefore, mechanisms of colonization on tissues such as skin may be quite different from what occurs on anterior nares. In this study, we performed a murine ear colonization experiment (18) to understand the mechanism of skin colonization by employing MRSA strain MW2 (1). The strain, representing community-acquired MRSA (CA-MRSA), is considered a highly virulent MRSA strain with a great ability to colonize human tissue.
Contrary to our expectations, strain MW2 did not easily colonize on murine skin, and it took several weeks before colonization became apparent. The colonizing bacteria were certainly derivatives of the strain MW2; however, we found that they included some that were different from the parental strain in their production of hemolytic toxin beta-hemolysin (Hlb), which is also known as sphingomyelinase. Since the parental strain, MW2, does not produce the toxin, we performed genetic analysis of the colonizing derivative strains in order to explain the contradiction. Our analysis revealed that a prophage, ϕSa3mw, which had been inserted in the structural gene for Hlb of MW2, was precisely excised from the genome, and therefore an intact hlb gene was reconstituted in these derivative strains that colonized murine ear skin, suggesting that the toxin plays an important role in skin colonization by S. aureus. We also demonstrated that the MW2 derivative mutant cured of the prophage ϕSa3mw, with Hlb expression allowed showed a more than 50-fold greater efficiency of ear colonization than that of a mutant lacking the Hlb gene. In addition, we found that the MW2-derivative strain expressing Hlb had more cytotoxicity for human primary keratinocytes than the mutants without the Hlb gene. Further study employing purified sphingomyelinase from S. aureus showed that the toxin specifically damaged human keratinocytes. Our study, described in this report, clearly indicates that Hlb production by S. aureus damages keratinocytes, subsequently leading to colonization of skin by the microorganism.
MATERIALS AND METHODS
Bacterial strains, plasmids, and growth conditions.Staphylococcal strains and constructed plasmids are listed in Table 1. All cloning steps were carried out using Escherichia coli DH5α (TaKaRa-Bio Co., Ltd., Shiga, Japan). The doubling times of MW2, MW2ΔϕSa3mw, MW2Δhlb, and MW2ΔϕSa3mwΔhlb were 28.8, 29.0, 29.2, and 27.9 min, respectively, showing that there was no significant difference among them. The doubling time was calculated as described previously (19).
Bacterial strains and plasmids used in this study
Animals and bacterial strains.BALB/cAnNCrlCrlj (BALB/c) and BALB/cAnNCrlCrlj-nu/nu (BALB/c-nu/nu) mice (female, 6 weeks old, with no superficial dermatitis) and Naruto Research Institute Otsuka Atrichia (NOA) mice (male, 6 weeks old, with dry skin) were purchased from SLC (Shizuoka, Japan) (20). The mice were housed in individually ventilated cages (IVC) (Techniplast, Buguggiate, Italy) under specific-pathogen-free conditions. All animal experiments were carried out in accordance with the Guidelines for Animal Experimentation of Juntendo University. MW2 is a typical community-acquired strain of MRSA which was isolated in 1988 in North Dakota (21). MW2 caused fatal septicemia and septic arthritis in a 16-month-old girl who had no risk factors associated with health care, and she died 2 h after her arrival at a hospital.
All S. aureus strains were cultivated in brain-heart infusion (BHI) broth, tryptic soy broth (TSB), or BHI agar (Becton, Dickinson and Co.[BD], Franklin Lakes, NJ) with aeration at 37°C, unless indicated otherwise.
DNA methods.DNA manipulations and pulsed-field gel electrophoresis (PFGE) were performed using standard methods (22, 23). Restriction enzymes were used as recommended by the manufacturer (TaKaRa-Bio). PCR amplification was performed using an Expand High Fidelity system (Roche, Mannheim, Germany).
Construction of MW2Δhlb and MW2ΔϕSa3mwΔhlb strains.For construction of the hlb gene null mutant, we used a pKOR1 allele replacement system as described by Bae et al. (24). In brief, a 2.0-kb insert DNA encompassing 1-kb flanking sequences of phage attachment sites was generated by PCR from chromosomal DNA of strain MW2 using primers as shown in Table 2. The PCR products were used for a recombination reaction with pKOR1, and then the recombinants were introduced into E. coli DH5a (Invitrogen) for amplification. Either the resulting plasmid pKO-hlb1 or pKO-hlb2 was introduced into S. aureus strain MW2 or MW2ΔϕSa3mw by electroporation, generating the transformant MW2(pKO-hlb1) or MW2ΔϕSa3mw(pKO-hlb2). Allele replacement was carried out by a two-step procedure, as described previously (24). Mutant candidates were spread on tryptic soy agar (TSA) (BD) plates containing 5% sheep red blood cells and incubated for phenotypic analysis. The deletion of the target gene was confirmed by PCR amplification with primers described below.
Synthetic oligonucleotide primers
Ear colonization in mice.Both ears of each mouse were inoculated with 1 × 108 CFU of S. aureus per ear. To measure the CFU on mice, we collected the bacteria from external acoustic meatus using a sterilized wet swab and then suspended them in 500 μl of sterile saline. The viable bacteria were quantified by plating 50 μl of serial 10-fold dilutions of the suspensions on mannitol salt agar (BD) plates, followed by incubation for 48 h at 37°C. Yellow-colored colonies were counted, and then the representative colonies were subjected to analysis using a Staphyogram (Wako Pure Chemical, Ltd., Osaka, Japan) PBP2′ detection kit and a coagulase latex agglutination test using a Staphylo LA kit (Denka Seiken, Tokyo, Japan) for identification.
Hemolytic activity on blood plates.Hemolysis on sheep or mouse blood agar was assessed by spot inoculation of each strain (25). Ten microliters of the overnight bacterial culture was inoculated onto Todd-Hewitt agar containing 3% (vol/vol) blood and then incubated at 37°C for 18 h, followed by further incubation at 4°C for 12, 24, 48, and 96 h, respectively, to determine whether Hlb, which is a hot-cold lysin, was produced.
Detection of excised GI by PCR.All primers used for the detection of excised genomic island (GI) (ϕSa3mw, ϕSa2mw, νSa3mw, and νSa4mw) from the MW2 chromosome (1) are listed in Table 2.
Proliferation assay using human PBLs.Human peripheral-blood lymphocyte cells (PBLs) were isolated from volunteers' blood by Ficoll-Conray density gradient centrifugation and then suspended in RPMI 1640 medium supplemented with 10% feral calf serum. The PBLs (2 × 105) were cocultured under 5% CO2 with diluted supernatant of MW2 or MW2ΔϕSa3mw that had been grown overnight at 37°C for 48 h. Each culture was pulsed with 20 kBq of [3H]thymidine (PerkinElmer, Boston, MA) for 16 h, and cells were harvested on glass fiber filters. The amount of [3H]thymidine incorporated into the cells was measured by using a liquid scintillation counter.
Assessment of ϕSa3mw excision from a chromosome of MW2 under in vitro conditions.In experiments with S. aureus strain MW2 in vitro to assess the prophage deletion, each single colony of the strain was inoculated into four tubes with BHI broth and incubated at 37°C for 24 h. Each of the four cultures was sequentially subcultured into fresh medium to 1:1,000 dilutions, and this process was repeated 56 times every day for 8 weeks. Finally, the four bacterial cultures were then inspected for the occurrence of phenotypic variants or hemolysis and colony morphology. Phage induction was attempted by incubation with 0.5 mg/liter of mitomycin C (Sigma) for 6 h. The amount of excised ϕSa3mw was measured by nucleic acid quantitative PCR as described below.
Preparation of the internal standard for nucleic acid quantitative PCR.Internal standard DNA was synthesized by PCR amplification using pUC119 as a template. The primers used were 5′-AAAATGATTTATCTAATGGCTTAGCTGCATACGCCTATTTTTATAGGTTAATGTCAT-3′ and 5′-AAAAAGACACATAATGTGAAGTTTGTAAAAGTACTTAATGATAAGTTGCTCTTGCCCGGCGTCAATACGGGAT-3′, where the underlined sequences correspond to nucleotides 1179 to 1207 and 1598 to 1627 of pUC119, respectively (26). The set of primers amplified the DNA fragment of 0.5 kb. The amplified PCR product was subjected to agarose gel electrophoresis and was purified from the gel using a gel purification system (Qiagen). The integrity of the purified DNA was confirmed by nucleotide sequence determination. The concentration of the DNA was determined by measuring the absorbance at 260 nm with a spectrophotometer.
Measuring excision of ϕSa3mw by quantitative PCR and statistical analysis.Nucleic acid quantitative PCR was carried out as described previously (26), except that serially diluted competitor DNAs were added to 0.0015 to 200 pM concentrations in 50 μl of PCR mixture. Preliminary experiments were performed to confirm the linear relationship between the amounts of PCR product for the entire set of experiments (26). The quantity of excised ϕSa3mw in the DNA samples was calculated by the procedure described by Köhler (27). For statistical analyses, 2 × 2 contingency tables were evaluated using Fisher's exact test. Basically the same procedure was employed for other analyses.
Human primary keratinocytes and cytotoxicity assay.Human primary keratinocytes (Cascade Biologic, Portland, OR) were grown in Epilife-KG2 keratinocyte growth medium (Kurabo, Osaka, Japan) containing 10 ng/ml insulin, 0.67 μg/ml hydrocortisone, 50 μg/ml gentamicin, 50 ng/ml amphotericin B, 0.1 ng/ml human epidermal growth factor (hEGF), and 0.4% (vol/vol) bovine brain pituitary extract until they reached subconfluent phase in 12-well microplates. The cell culture was washed twice with phosphate-buffered saline (PBS), followed by addition of antibiotic-free Epilife-KG2 medium, and then S. aureus cells grown in TSB for 20 h were added to a multiplicity of infection (MOI) of 10. This incubation period seemed important: bacterial cultures incubated longer conferred more cytotoxicity to keratinocytes, presumably due to more accumulation of toxins, leading to an earlier appearance of keratinocyte cytotoxicity. To evaluate cytotoxic effects of staphylococcal exotoxins directly, their purified forms were added instead of the S. aureus cells: the Panton-Valentine leukocidin (PVL) subunits LukF-PV and LukS-PV were obtained as described by Hongo et al. (28), whereas sphingomyelinase (equivalent to Hlb) was purchased from Sigma Chemical. Sphingomyelinase was equilibrated with the same solvent as PVL using a floating nylon membrane filter (Whatman, GE Healthcare). Cytotoxicity was evaluated using a Live/Dead viability/cytotoxicity kit for animal cells (Invitrogen), and fluorescence microscopic observation was performed, where live cells had green fluorescence while damaged cells were stained red. Quantification of the damaged cells took advantage of the dual-stain fluorescence microscopic image: the red area divided by the total green-plus-red fluorescent signals was measured by using the software program ImageJ version 1.440, freely provided by the National Institutes of Health (http://imagej.nih.gov/ij/), with proper configuration. The data were presented as means with ranges from four independent fields.
RESULTS
Murine ear colonization by S. aureus strain MW2.In order to study staphylococcal factors involved in skin colonization, we initially wished to establish a murine model of S. aureus colonization, and therefore we attempted to inoculate 108 CFU of S. aureus strain MW2 (1) onto ears of BALB/c and BALB/c-nu/nu mice to see if the strain, which is highly virulent for humans, also formed skin lesions on a mouse. Even after 4 weeks, we failed to isolate S. aureus from BALB/c mice. On the other hand, a mean of 1.4 × 105 CFU of the microorganism was obtained from BALB/c-nu/nu mice, indicating that the BALB/c-nu/nu mouse model was sufficiently reproducible to study S. aureus and host factors involved in colonization.
We next tried to test if the colonizing S. aureus strains isolated from the BALB/c-nu/nu mice were the original strain MW2 or were altered genetically. We inoculated the same amount of strain MW2 onto four BALB/c-nu/nu mice as described above and again obtained 1 × 105 CFU of colonized S. aureus from each mouse after 4 weeks (see Fig. 2B). Among the colonizing strains, we isolated only one from each mouse (four in total) and subjected them to pulsed-field gel electrophoresis (PFGE) (23) in order to see possible genome-wide alteration in comparison with the parental strain, MW2 (Fig. 1A). The reason we isolated only one colony from a mouse was that we assumed that analysis of the limited number of colonizing S. aureus bacteria would be able to detect a genomic alteration if such an event was required for colonization on mouse ear skin. Of the four, one isolate showed change in the PFGE pattern; the SmaI-G fragment of 180 kb was replaced by a single fragment of 130 kb, indicating a deletion of a 50-kb fragment from the original SmaI-G fragment (Fig. 1A, lane 3). The isolate was not considered a contaminant since other DNA fragments on the PFGE pattern were identical to those of parental strain MW2. The possibility of finding that a drastic change had occurred in a chromosome of a colonizing S. aureus strain out of only four strains that had been randomly chosen among the 1 × 105 CFU of bacteria per mouse would be quite high and should not be due to contingency, and therefore we considered that the genomic alteration might support S. aureus colonization on the skin in some way. Indeed, the following analyses revealed that such an alteration was widely present for the colonizing strain MW2 on murine skin.
(A) PFGE analysis of S. aureus strain MW2 derivatives isolated from BALB/c-nu/nu mice. PFGE was performed using SmaI-digested genomic DNAs obtained from MW2 colonizing the mice (lanes 1 to 3), parental strain MW2 (lane 4), and laboratory strain NCTC8325-4 (lane 5), respectively. Lane 6 shows a contour-clamped homogeneous electric field (CHEF) DNA size standard lambda ladder (Bio-Rad, Hercules, CA). The 180-kb SmaI-G fragment from a colonized MW2 derivative (MW2ΔϕSa3mw) is shifted to 130 kb (lane 3, white arrows). (B) Proliferation of human PBLs induced by the supernatant of bacterial cell cultures. Human PBLs (2 × 105) isolated from two volunteers, PBLs-1 and PBLs-2, were incubated with supernatant of either MW2 or MW2ΔϕSa3mw overnight culture that was diluted to 10-fold (left) or 1,000-fold (right), and incorporation of radiolabeled thymidine into the lymphocytes is indicated as counts per minute (cpm). The data were obtained from triplicate experiments and are shown as means ± standard deviations. The level of significance was determined by the two-sided Student t test (P < 0.001). (C) Hemolytic activities of MW2 and MW2ΔϕSa3mw on sheep blood agar are indicated. (D) Genomic structures of the hlb gene and its flanking regions in strain MW2 and its derivative mutant strains MW2ΔϕSa3mw (hlb+), MW2Δhlb, and MW2ΔϕSa3mwΔhlb. Dotted-line sets represent regions deleted from the strains shown above.
Strain MW2 bacteria that have colonized the murine ear tend to lose the ϕSa3mw prophage.To determine the deleted region in the chromosome of the MW2 isolate (Fig. 1A, lane 3), we performed long-range PCR amplification using several sets of primers, which allowed amplification of the chromosomal DNA across the SmaI recognition sites. Sequence analysis of the amplified PCR products revealed that the deletion of the SmaI-G fragment was caused by loss of the integrated prophage ϕSa3mw (Fig. 1D). Since ϕSa3mw carries several virulence factors, including three staphylococcal enterotoxins (SEs) (1, 29–31) that cause nonspecific activation of T cells and disturb host immune systems by massive cytokine release (1, 32), we initially assumed that the three SEs (SEA, SEK, and SEQ) might play some role in inhibition of colonization. We therefore tried a T-cell proliferation assay to see the superantigenic activation of T cells by employing human peripheral blood lymphocyte cells (PBLs) together with the parental strain, MW2, or its ϕSa3mw-less derivative strain, designated MW2ΔϕSa3mw. The proliferation of human PBLs observed with MW2ΔϕSa3mw was significantly lower than that with MW2 (Fig. 1B), suggesting that the lack of superantigen genes on prophage ϕSa3mw caused decreased T-cell proliferation activity. At that point, it was still unclear whether the loss of three superantigen genes had allowed the colonization by S. aureus strain MW2 of murine ear skin or not. However, we noticed that not only was the loss of the superantigen genes subsequent to prophage deletion but that this might also allow S. aureus strain MW2 to express beta-hemolysin (Hlb), a hemolytic toxin with sphingomyelinase activity. Since the prophage ϕSa3mw is present in the genome of MW2 in such a manner as to disrupt the gene for Hlb (hlb) (Fig. 1D), the excision of the phage can lead to expression of the toxin gene. Our sequencing analysis showed that the loss of ϕSa3mw seen in MW2ΔϕSa3mw caused reconstruction of in-frame fusion of the hlb gene that had been separated by insertion of the prophage (data not shown). To support the precise phage excision leading to the toxin expression, MW2ΔϕSa3mw clearly showed hemolysis on blood-agar plates that was not observed with the original MW2 strain (Fig. 1C). However, we still did not know if the change in MW2 was due only to loss of the prophage ϕSa3mw, despite the PFGE pattern shown in Fig. 1A, which supported the prophage deletion seeming to be the only event allowing genome-wide alteration in strain MW2. We therefore confirmed whether other mobile genomic islands, ϕSa2, νSa3, and νSa4, which also carry known virulence factors (1), were present in strain MW2. The four genomic islands were maintained on its genome as determined by PCR amplification (data not shown; primers used are listed in Table 2), excluding the possibility that colonization by strain MW2 was supported by loss of mobile genetic elements other than prophage ϕSa3mw.
Loss of prophage ϕSa3mw leads to a facilitated rate of colonization by S. aureus MW2 on murine ear skin.To further clarify if loss of ϕSa3mw increased the rate of colonization by S. aureus, we used an NOA mouse model that develops spontaneous dry skin and lesions similar to atopic dermatitis (20) in addition to using BALB/c and BALB/c-nu/nu mice. The mice were subjected to inoculation of S. aureus onto their ear skin, using strains MW2 and MW2ΔϕSa3mw, the latter of which had been isolated from the ear of BALB/c-nu/nu mice after MW2 colonization as described above. To evaluate the effect of Hlb on colonization, strain NCTC8325-4 (hlb+) was also employed. Four weeks after administration, all three strains were detected in samples from the ears of BALB/c-nu/nu and NOA mice but not of BALB/c mice. Initially, there seemed to be no significant differences in the numbers of colonized bacteria between MW2 and MW2ΔϕSa3mw (Fig. 2B and C). However, when we assessed the excision of the prophage ϕSa3mw in colonizing MW2 by using the PCR and blood-containing agar plate assay (Table 3), a significant proportion of the colonized bacteria were converted to strains whose features were identical to those of MW2ΔϕSa3mw. Namely, they showed hemolytic haloes on agar plates and did not possess the prophage ϕSa3mw, as we expected. Four weeks after inoculation of the parental strain MW2, 8.16% (4/49) and 95.7% (88/92) MW2ΔϕSa3mw was isolated from BALB/c-nu/nu and NOA mice, respectively (Table 3). This result clearly demonstrated loss of the prophage ϕSa3mw from MW2 during ear colonization of NOA mice and also suggested that Hlb expression played a role in murine ear colonization.
Colonization by S. aureus strain MW2 on murine ears. BALB/c (A), BALB/c-nu/nu (B), or NOA (C) mice were employed. Colonized bacteria were counted a day, a week, or 8 weeks after inoculation of 1 × 108 CFU per mouse. Each group consisted of four mice. Each mouse was inoculated with either MW2 (●), MW2ΔϕSa3mw (○), NCTC8325 (◊), or saline as a control (×). Each symbol represents the average no. of CFU in log scale per ear of mouse.
Appearance of MW2ΔϕSa3mw from ear-colonizing MW2a
Hlb production by S. aureus is involved in its murine ear colonization.The results indicated above still could not allow concluding that the increased rate of S. aureus colonization was not due to a loss of three superantigen genes present in prophage ϕSa3mw but rather was due to the acquisition of Hlb expression. In order to elucidate the relationship between a degree of S. aureus colonization and its Hlb production, we constructed two Hlb-nonproducing mutants, MW2ΔϕSa3mwΔhlb and MW2Δhlb (Fig. 1D), and then compared differences in their colonization rates on NOA mice from that of parental strain MW2. Five weeks after administration, MW2, MW2ΔϕSa3mw, MW2Δhlb, and MW2ΔϕSa3mwΔhlb colonized with the following numbers of bacterial cells: 3.81 ± 0.82, 4.63 ± 0.37, 2.69 ± 0.49, and 3.00 ± 0.22 CFU/ear on a logarithmic scale, respectively (Fig. 3). All hlb-less mutants showed a drastic decrease in their colonization, by 1.6 to 2.0 on the logarithmic scale, or several tens to a hundredfold less than MW2ΔϕSa3mw, which expresses Hlb.
Colonization by S. aureus strains MW2, MW2ΔϕSa3mw (hlb+), MW2Δhlb, and MW2ΔϕSa3mwΔhlb on ears of NOA mice. Numbers of bacteria isolated were recorded a day, a week, 3 weeks, and 5 weeks after inoculation of each S. aureus strain onto the mice.
ϕSa3mw was excised from a chromosome of MW2 at a high frequency under the in vivo condition but not in vitro.The results described above strongly suggested that Hlb expression is important for S. aureus colonization on murine skin, but we wished to exclude the possibility that spontaneous excision of the prophage ϕSa3mw occurred in vitro and was accidentally maintained on murine ear skin. Therefore, we tried to demonstrate that the genomic event in S. aureus allowing expression of Hlb selectively took place during its colonization on murine skin but not in the absence of the mice. In order to demonstrate this, we performed an experiment to measure the excision rate in in vivo and in vitro and compared the values between the two conditions by quantitative PCR amplification, as described in previous reports (26, 27). For the in vitro condition, in which only S. aureus strain MW2 was grown in medium, subculturing with 56 passages during 8 weeks was performed. We also tested the same in vitro experiment in the presence of mitomycin C, expecting a higher prophage excision rate. In the in vivo experiment, strain MW2 was inoculated onto four NOA mice and left for 5 weeks. The prophage excision rate in each experiment is summarized in Table 4. Under the condition of MW2 culture in vivo, MW2ΔϕSa3mw was detected with high frequency, with a value almost the same as that for the control. On the other hand, in the MW2 culture in vitro, MW2ΔϕSa3mw was detected with as low as a 4-logarithm-smaller frequency than that observed in vivo. Even in the presence of mitomycin C, the excision rate was much lower than that in vivo. We therefore concluded that the persistence of ϕSa3mw-less S. aureus on murine skin did not occur due to a contingency but was caused by specific selection in the host environment, allowing a higher rate of colonization in vivo. In in vitro experiments, MW2Δhlb showed no significant difference in the excision rate of ϕSa3mw in comparison with that for MW2, indicating that the excision was not induced by the presence of the hlb gene (Table 4).
Frequency of excision of ϕSa3mw from the chromosome with MW2 and MW2Δhlb
Expression of Hlb from S. aureus causes keratinocyte damage.Our results described above indicated that Hlb plays an important role in S. aureus colonization on murine skin. However, we puzzled to explain the mechanism, because Hlb is known as an erythrolytic toxin with sphingomyelinase activity and we were therefore unable to figure out how the toxin affected animal skin.
In order to pursue the possibility that the toxin might damage the cells forming the skin, we cultured human primary keratinocytes and added either S. aureus strain MW2 or its derivative MW2ΔϕSa3mw, MW2Δhlb, or MW2ΔϕSa3mwΔhlb, employed in the experiments above. To our surprise, only MW2ΔϕSa3mw, which can produce Hlb, showed significant cytotoxicity to the human keratinocyte during the incubation period, as shown in Fig. 4, although others, including the original strain, MW2, did not. This result suggested that Hlb directly damages keratinocytes. In order to test the effect of Hlb on the cells directly, we incubated human keratinocytes with purified sphingomyelinase, namely, Hlb, from S. aureus. As we expected, the sphingomyelinase clearly damaged keratinocytes at a concentration of 0.5 μg/ml, whereas the 10-fold concentration of purified Panton-Valentine leukocidin, which is known as a toxin produced by S. aureus that destroys human leukocytes (28, 33), did not show any cytotoxicity for the keratinocytes (Fig. 5). Our results clearly indicated that beta-hemolysin with sphingomyelinase activity from S. aureus damages keratinocytes and that this can lead the microorganism to colonize the skin.
(A) Cytotoxicity of S. aureus strain MW2 and its derivatives for human keratinocytes. The bacteria were coincubated with the keratinocytes for 16 h at an MOI of 10. The live cells with green fluorescence are distinguishable from damaged cells, whose DNA is stained red under a fluorescence microscope. (B) Quantification of the damaged-cell ratio by using the image from microscopic observation. Data are expressed as means for four independent fields of the fluorescence microscopic view.
(A) Effects of Hlb and Panton-Valentine leukocidin on human keratinocytes. Each toxin at the indicated concentration was incubated with human keratinocytes for 60 h and observed as shown in Fig. 4B. (B) Quantification of the experiments shown in panel A. Data are expressed as means for four independent fields of the fluorescence microscopic view.
DISCUSSION
We found that beta-hemolysin (Hlb) has an important role in skin colonization by S. aureus. First, we noticed that spontaneous precise excision of a prophage, ϕSa3mw, significantly increased the colonization ability of S. aureus strain MW2 on murine ear skin. Since the prophage is inserted in the gene for Hlb, its removal from the genome causes expression of the toxin gene. We confirmed that the colonized MW2 derivatives lacking the prophage had increased hemolytic activity, using blood agar plates. In addition, an experiment with S. aureus deletion mutants with or without the hlb gene genetically showed that the toxin was involved in S. aureus skin colonization. We also confirmed that genes carried by the prophage ϕSa3mw did not affect colonization or keratinocyte damage. Furthermore, we directly showed that purified Hlb had cytotoxicity for human keratinocytes. The data obtained above clearly indicate that the toxin leads S. aureus to colonization of skin by damaging keratinocytes. Loss of the staphylococcal enterotoxins encoded in the excised prophage ϕSa3mw, as well as other phage-encoded factors, is not likely to be involved in the altered colonization efficiency, since the S. aureus strains MW2ΔϕSa3mwΔhlb and MW2Δhlb did not show any significant difference in their colonization rates.
The Hlb toxin has been known as an erythrolytic enzyme with sphingomyelinase activity (25, 34) in addition to its toxicity for proliferating human lymphocytes (35). On the other hand, the contribution of Hlb to cytotoxic activity against the essential cell component of the skin has been reported to occur under limited conditions. Hlb from S. aureus enhances cytotoxicity for human immobilized skin keratinocyte HaCaT cells once bacterial internalization occurs via fibronectin-binding proteins (FnBPs), followed by survival of bacteria in the cells; however, the report concluded that internalization of S. aureus strain 8325-4 and cytotoxicity for keratinocytes are phenomena independent of alpha- or beta-hemolysin (36). Our data clearly revealed that Hlb directly exerts keratinocyte toxicity, because much less keratinocyte damage was observed with Hlb-nonproducing strains. It is possible that our using primary human keratinocytes, as well as a bifluorescent cell staining assay system that is different from the one depending on propidium iodide used in the previous work, caused a more conspicuous observation of the keratinocyte damage. Otherwise, strain MW2 may have intrinsically weaker cytotoxicity for keratinocytes than strain 8325-4, and the cytotoxicity was enhanced to an observable degree only when Hlb was expressed by the phage excision after bacterial internalization into keratinocytes via FnBps.
Since our results indicate that Hlb emphasizes skin colonization by S. aureus, it is tempting to consider that cytotoxicity for keratinocytes is closely associated with the promotion of skin colonization. Alternatively, the two phenomena, or the bacterial colonization on skin and keratinocyte damage, may be due to the reported multifunctional nature of Hlb. Recently, Huseby et al. reported that Hlb toxin had biofilm ligase activity, which is independent of its sphingomyelinase activity (37). This implies that stimulation of biofilm formation by the toxin may be involved in staphylococcal skin colonization independently of its erythrocyte or keratinocyte toxicity.
We found that the precise excision of the Hlb-converting phage changes S. aureus strain MW2, leading it to adopt the “colonizing mode” by allowing the bacteria to produce Hlb. Therefore, the prophage is thought to serve as a “genetic switch” for the mode change of the host bacteria. Although the prophage could be excised spontaneously at a constant frequency, our experiment comparing the rates of the phage excision in vivo and in vitro revealed that this was not the case (Table 4). Thus, we concluded that selection of the phage-excised population occurred on the murine skin because of the greater ability for colonization. Such genomic modifications of bacteria are known to occur when they adapt to different environments (38–40). It is also known that such mobility of prophages is induced by changes in environmental conditions that lead to DNA damage, including exposure to the reactive oxygen substance generated by leukocytes or to exogenous agents, such as antibiotics (39). Broudy et al. discovered a soluble phage-inducing factor elaborated by human pharyngeal epithelial cells that induces the SpeC (streptococcal pyrogenic exotoxin C)-encoding phages of streptococci (41). It is possible that such a phage induction signal exists in our experimental system, too.
Our observations in this study lead us to a curious question: why certain S. aureus strains allowed a prophage integration in the hlb gene at the expense of their ability to colonize the skin. The Hlb-converting bacteriophages found in various S. aureus strains are known to carry the innate immune-evasion cluster (IEC), which contains such virulence genes as the chemotaxis inhibitory protein of S. aureus (CHIPS) and staphylococcal complement inhibitor (SCIN) genes (42). The array of such genes carried by prophages may be only for evasion of the human immune system, which is ineffective for mouse skin, and therefore the prophage is removed upon colonization on it, allowing hlb expression. In this case, it means that a prophage-mediated genetic switch may have evolved for the purpose of preserving the broad host animal specificity of S. aureus. Another possibility is that S. aureus can select two alternate modes of virulence, one to escape hosts' immune responses to avoid being cleared from infection sites and the other to trigger destruction of erythrocytes and keratinocytes and enforce skin colonization, possibly via biofilm formation.
ACKNOWLEDGMENTS
This work was supported by a Grant-in-Aid for Takeda Science Foundation Award and Scientific Research on Priority Areas (16017294 and 17790677) from the Ministry of Education, Science, Sports, Culture and Technology of Japan.
We are indebted to Isamu Hongo for providing purified Panton-Valentine leukocidin subunits and for discussion of the toxin. We are grateful to Yu Hasegawa (a medical student at Juntendo University) for technical support. We are also thankful for great technical assistance from Francois Niyonsaba, Toshiro Takai, and Hiroko Ushio and from other staff in the Atopy Research Center, Juntendo University School of Medicine.
FOOTNOTES
- Received 20 September 2012.
- Accepted 26 December 2012.
- Accepted manuscript posted online 4 January 2013.
- Copyright © 2013, American Society for Microbiology. All Rights Reserved.
