Correspondence: Jan R. van der Ploeg, Institute of Oral Biology, University of Zürich, Plattenstrasse 11, CH-8032 Zürich, Switzerland. Tel.: +41 44 634 3329; fax: +41 44 634 4310; e-mail: firstname.lastname@example.org
Bacteriophage M102 is a virulent siphophage that propagates in some serotype c Streptococcus mutans strains, but not in S. mutans of serotype e, f or k. The serotype of S. mutans is determined by the glucose side chain of rhamnose–glucose polysaccharide (RGP). Because the first step in the bacteriophage infection process is adsorption of the phage, it was investigated whether the serotype specificity of phage M102 was determined by adsorption. M102 adsorbed to all tested serotype c strains, but not to strains of different serotypes. Streptococcus mutans serotype c mutants defective in the synthesis of the glucose side chain of RGP failed to adsorb phage M102. These results suggest that the glucose side chain of RGP acts as a receptor for phage M102.
Mutans streptococci are thought to play an important role in caries development in humans and in some animals (Loesche, 1986). The group of mutans streptococci consists of seven different species: Streptococcus mutans, Streptococcus rattus, Streptococcus sobrinus, Streptococcus cricetus, Streptococcus ferus, Streptococcus macacae and Streptococcus downei. Streptococcus mutans and S. sobrinus are frequently found in the human oral cavity (Loesche, 1986). Apart from the classification at the species level, mutans streptococci have also been classified into nine different serotypes, based on differences in their cell wall polysaccharides. Streptococcus mutans is divided into four different serotypes: c, e, f and k. They produce cell wall rhamnose–glucose polysaccharides (RGPs) consisting of a backbone structure of α1,2- and α1,3-linked rhamnosyl units with glucose side chains linked to alternate molecules on the cell wall (Pritchard et al., 1986). Each serotype RGP has unique linkages of its glucose side chains (Fig. 1; Pritchard et al., 1986).
RGP synthesis in S. mutans is initiated by transfer of N-acetylglucosamine-1-phosphate to a lipid carrier, most likely undecaprenyl phosphate (Yamashita et al., 1999). Synthesis of the rhamnan backbone requires dTDP-l-rhamnose, which is produced from d-glucose-1-phosphate by the products of the rmlABCD genes (Tsukioka et al., 1997a, b). The activated rhamnose units are linked to N-acetylglucosamine-lipid through RgpA, RgpB and RgpF to form the rhamnan backbone (Yamashita et al., 1998a; Shibata et al., 2002). Glucose side chain formation requires gluA, encoding glucose-1-phosphate uridyltransferase, for the synthesis of UDP-d-glucose (Yamashita et al., 1998b). This precursor is attached to the rhamnan backbone by RgpE, RgpH and RgpI (Ozaki et al., 2002). RgpE and RgpH are glucosyltransferases required for glucose side chain formation of RGP, and RgpI regulates the branching frequency of glucose side chain on RGP in serotype c strain Xc (Yamashita et al., 1998a; Ozaki et al., 2002). The differences in the glucose side chain formation of RGP among serotype c, e and f strains were found to depend on a serotype determinant locus downstream from rgpF (Fig. 2) (Shibata et al., 2003). In variants of serotype c, named serotype k, the glucose side chain of RGP is lacking, probably because rgpE is not expressed (Nomura et al., 2005).
Bacteriophage M102 is a virulent siphophage specific for S. mutans of serotype c (Delisle & Rostkowski, 1993; van der Ploeg, 2007). M102 produces clear plaques on only some of the serotype c strains of S. mutans, but cannot infect serotype e and f strains (Delisle & Rostkowski, 1993). Because the first step in the bacteriophage infection process is the adsorption of the phage, the specificity of phage M102 for serotype c strains might be determined by the receptor. Here, we investigated whether serotype-specific RGP functions as a receptor of bacteriophage M102.
Materials and methods
Strains and growth conditions
Mutans streptococci (Table 1) were routinely grown in Todd–Hewitt broth (THY, Becton-Dickinson) with 0.3% yeast extract (Oxoid), at 37 °C in an environment of 10% CO2 and 90% air. Streptococcus mutans OMZ 381S is a spontaneous streptomycin-resistant mutant of OMZ 381 (van der Ploeg, 2007). THY plates contained 1.5% agar. THY soft agar contained 0.7% agar. Bacteriophage M102 and its propagation have been described before (van der Ploeg, 2007). The antibiotics used for selection of S. mutans were spectinomycin (1000 μg mL−1), erythromycin (15 μg mL−1) and streptomycin (1000 μg mL−1).
For the construction of plasmids pSTN and pSTC, a DNA fragment containing the genes from rgpF to ORF12 (Fig. 2) from S. mutans strain Xc was inserted into the multicloning site of pBluescript SK II (+) to yield plasmid pST. For pSTN, the region containing ORF7 through ORF11 was replaced with a spectinomycin resistance gene. For construction of pSTC, the region containing ORF7 in pST was first replaced with the erythromycin resistance gene; subsequently, the region harbouring ORF10 was deleted.
Strain OMZ 381SN was constructed by transformation of OMZ 381S with BssHII-digested plasmid pSTN to spectinomycin resistance. OMZ 381SC was generated by transformation of OMZ 381SN with BssHII-digested plasmid pSTC to erythromycin resistance.
Replacement of serotype f-specific genes in strain OMZ 175 was as described previously (Shibata et al., 2003). Briefly, OMZ 175 was first transformed with BssHII-digested plasmid pSTN to spectinomycin resistance. The resulting strain was then transformed with BssHII-digested plasmid pSTC to erythromycin resistance.
Streptococcus mutans strains Xc41, Xc42 and Xc52 have been described previously (Yamashita et al., 1998a, 1999). rgpE and rgpH mutations in strain OMZ 381S were obtained by transformation with genomic DNA from S. mutans strains Xc45 (Yamashita et al., 1998a) and Xc48 (Ozaki et al., 2002), respectively, with selection for erythromycin resistance. Correct gene replacement was verified by PCR analysis.
Adsorption of phage M102 to mutans streptococci was investigated as follows: 100 μL of phage M102 (c. 104 PFU) was mixed with 100 μL of an exponentially growing culture (OD550 nm of about 0.4) of the strain and incubated for 30 min at 37 °C. Phages adsorbed to bacteria were pelleted by centrifugation for 5 min at 16 000 g. The amount of nonadsorbed phages remaining in the supernatant was then determined by a standard plaque assay. For this, 100 μL of an appropriate dilution was mixed with 100 μL of an overnight grown culture of OMZ 381S and incubated for 30 min at 37 °C. After addition of 4 mL of molten THY soft agar (THY with 0.7% agar), the mixture was poured over an THY agar plate, which was incubated overnight at 37 °C in 10% CO2 and 90% air. The number of plaques was counted.
It was first examined whether the serotype specificity of phage infection was determined by adsorption (Fig. 3). M102 adsorbed to all seven serotype c strains tested, whereas it did not adsorb to strains of different serotypes. On the other hand, whereas M102 adsorbed to all tested serotype c S. mutans strains, these strains were not infected. Because S. mutans serotype c, e, f and k strains have a different RGP structure, but are very similar otherwise, further investigations focused on these serotypes.
Three mutants of rgpA (Xc41), rgpB (Xc42) or rgpG (Xc52), which did not produce a rhamnan backbone and hence no RGP at all (Yamashita et al., 1998a, 1999; Shibata et al., 2002), were examined for adsorption by phage M102. M102 did not adsorb to these mutants (Fig. 4), which suggests that RGPs function as the phage receptor. We then constructed strains from which the serotype-specific genes had been removed or exchanged, and determined their ability to adsorb M102 (Fig. 4). The phage adsorbed to serotype c wild-type strains OMZ 381S and Xc, but not to the respective strains OMZ 381SN and Xc81, in which the locus downstream from rgpF and upstream from ORF12 had been deleted (Fig. 2).
As mentioned above, phage M102 did not adsorb to serotype e strain LM7 or serotype f strains MT6219 and OMZ 175, but replacing the serotype-specific loci of these strains by the serotype c-specific locus from strain Xc led to recombinant strains showing drastically increased adsorption of M102 (Fig. 4). Nevertheless, M102 could not infect these mutants.
The results demonstrate that the presence of the serotype c-specific locus involved in the glucose side chain formation of RGP is essential for the adsorption of M102, thus strongly suggesting that the serotype-specific glucose side-chain acts as the receptor for M102. To confirm our hypothesis, it was investigated whether bacteriophage M102 could be propagated in OMZ 381S mutants impaired in the formation of the glucose side chain of RGP. Strains OMZ 381SN, OMZ 381S-45 and OMZ 381S-48 did not adsorb bacteriophage M102 (Fig. 5) and were resistant to infection (relative efficiency of plating <1 × 10−9). The serotype-specific rgpH, rgpI and ORF11 (Fig. 2) from strain Xc were then introduced into OMZ 381SN as described in Materials and methods. The resulting strain, OMZ 381SC, adsorbed phage M102 (Fig. 5) and was sensitive to infection (relative efficiency of plating 0.14), which demonstrates that the serotype-specific genes from strain Xc are sufficient for phage infection.
Bacteriophage infection requires irreversible adsorption to a specific receptor. Bacteriophages infecting gram-positive bacteria first encounter the cell wall, which is decorated with polysaccharides and teichoic acids, either linked to the cytoplasmic membrane or linked to the peptidoglycan. Not surprisingly, these structures have previously been shown to function as a bacteriophage receptor (Vinga et al., 2006). Here, we have demonstrated that the glucose side chain of RGP is indispensable for adsorption and infection of S. mutans bacteriophage M102.
Cell wall carbohydrates frequently serve as a phage receptor in lactococci or thermophilic streptococci, but the exact molecular nature of the receptor has not yet been elucidated for any of the bacteriophages investigated. Studies to identify the receptor of Lactococcus lactis bacteriophages bIL170 and φ645 by random mutagenesis revealed a gene cluster likely involved in the synthesis of rhamnose-containing cell wall polysaccharides (Dupont et al., 2004). Interestingly, one of the genes identified showed similarity to S. mutans rgpE, which exists in all serotypes of S. mutans and whose gene product acts as a glucosyltransferase (Fig. 2).
The requirement for glucose substitution of a phage receptor is not unprecedented. Glycosylation of cell wall-linked teichoic acid was found to be essential for phage adsorption and susceptibility of Listeria monocytogenes (Cheng et al., 2008). Cell wall-bound teichoic acid consisting of glycosylated poly(glycerol-phosphate) is the primary receptor for Bacillus subtilis bacteriophage SPP1. In this case, reversible adsorption of phage to teichoic acid accelerates binding to the membrane receptor YueB, which is irreversible (Baptista et al., 2008). Finally, Lactobacillus delbrueckii phage LL-H adsorption to lipoteichoic acid depends on α-glucose substitution (Räisänen et al., 2007).
Although M102 adsorbed to all tested serotype c S. mutans strains, only strain OMZ 381S was sensitive to phage infection. The mutants that were converted from serotype e or f to serotype c showed drastically increased phage adsorption, but were not infected by M102. Thus, adsorption to the host is an essential, but not the sole determinant for successful phage infection. Several different mechanisms of phage resistance have been described (Forde & Fitzgerald, 1999; Barrangou et al., 2007), but at present it is unknown what mechanism causes resistance to M102 in serotype c strains.
The potential use of bacteriophage to remove cariogenic mutans streptococci from dental plaque is thus limited by serotype specifity, but also by other, unknown, factors. For these reasons, it might be better to use phage-encoded endolysins for selective elimination of S. mutans from dental plaque.
We thank Yutaka Sato for strain OM98x and Rudolf Gmür for critically reading the manuscript.