Pneumococcal licD2 gene is involved in phosphorylcholine metabolism


  • Jing-Ren Zhang,

    1. Department of Infectious Diseases, St. Jude Children's Research Hospital, 332 North Lauderdale Street, Memphis, TN 38105, USA.,
    Search for more papers by this author
  • Ilona Idanpaan-Heikkila,

    1. Department of Infectious Diseases, St. Jude Children's Research Hospital, 332 North Lauderdale Street, Memphis, TN 38105, USA.,
    Search for more papers by this author
    • *Present address: SmithKline Beecham, Brussels, Belgium

  • Werner Fischer,

    1. Institut für Biochemie, Universität Erlangen, Nürnberg, D-91054 Erlangen, Germany.
    Search for more papers by this author
  • Elaine I. Tuomanen

    1. Department of Infectious Diseases, St. Jude Children's Research Hospital, 332 North Lauderdale Street, Memphis, TN 38105, USA.,
    Search for more papers by this author

Elaine I. Tuomanen. E-mail; Tel. (+1) 901 495 3486; Fax (+1) 901 495 3099.


Phosphorylcholine is an important bioactive adduct to the teichoic acid (TA) and lipoteichoic acid (LTA) of the surface of Streptococcus pneumoniae. We have identified and characterized a genetic locus lic that is required for phosphorylcholine metabolism in S. pneumoniae. The pneumococcal lic locus consists of eight genes, licA, licB, licC and licD1, licD2 and three additional open reading frames. Pneumococcal licA, licB, licC, licD1 and licD2 have significant sequence similarity to licA, licB, licC and licD of Haemophilus influenzae. Mutation of licD2 led to decreased [3H]-choline uptake, aberrant migration of LTA chains in SDS–PAGE gels, loss of several surface proteins, and a phase-locked hypertransparent colony phenotype. Moreover, the licD2 mutant failed to undergo lysis after treatment with penicillin at high cell density and showed decreased transformation competence. Finally, the licD2 mutant demonstrated decreased adherence to the human type II alveolar cells, reduced nasopharyngeal colonization in infant rats, as well as significantly impaired virulence upon intraperitoneal challenge of CF1 mice. Identification of the lic genes in the pneumococcus will facilitate further characterization of the role of surface choline in microbial physiology and pathogenesis.


Streptococcus pneumoniae requires choline as an essential growth factor (Rane and Subbarow, 1940; Badger, 1944). Choline in the form of phosphorylcholine is incorporated into the cell wall teichoic acid (TA) and lipoteichoic acid (LTA) in S. pneumoniae. Pneumococcal TA and LTA chains are composed of ribitolphosphate-linked carbohydrate repeats, most of which contain two phosphorylcholine residues each (Behr et al., 1992; Fischer et al., 1993). The nutritional requirement for choline can be substituted by nutritional ethanolamine. However, replacement of choline with ethanolamine in culture results in formation of long chains and loss of the ability to undergo genetic transformation (Tomasz, 1968; Briese and Hakenbeck, 1985). In addition, cultures growing in the presence of ethanolamine do not lyse at the end of stationary phase and are resistant to penicillin-induced lysis. Therefore, choline metabolism plays an indispensable role in pneumococcal cell separation, transformation and autolysis.

S. pneumoniae is capable of spontaneous, reversible phase variation marked by formation of opaque or transparent colonies on translucent agar plates (Weiser et al., 1994). Compared with the opaque variant, the transparent one has enhanced ability to adhere to type II pneumocytes and vascular endothelial cells (Cundell et al., 1995a) and to colonize the nasopharynx of infant rats (Weiser et al., 1994). Conversely, the opaque variant is more virulent in systemic infection of BALB/c mice (Kim and Weiser, 1998). Recent studies have suggested that the transparent organisms possess higher amounts of phosphorylcholine than the opaque counterparts (Kim and Weiser, 1998; Weiser, 1998). Evidence also suggests that phosphorylcholine on the pneumococcal surface is able to directly interact with the host cell surface and thereby promote attachment and invasion (Tuomanen, 1997). Consistent with this notion, phosphorylcholine has been implicated as an adhesive ligand for the receptor of platelet-activating factor (PAF) present on the surface of various epithelial and endothelial cells (Cundell et al., 1995b). Binding to the PAF receptor enhances pneumococcal adherence and invasion of host cells. Finally, phosphorylcholine is required to anchor a family of proteins, the choline-binding proteins (CBPs), to the cell surface. Pneumococcal autolysin (Ronda et al., 1987) and pneumococcal surface protein A (PspA) (Yother and White, 1994) bind phosphorylcholine by a typical choline-binding domain. Recently, a pneumococcal adhesin, choline-binding protein A (CbpA) (Rosenow et al., 1997), was identified along with nine additional CBPs (Tuomanen and Masure, 1997). Therefore, phosphorylcholine metabolism plays a significant role in cellular physiology and pathogenesis of the pneumococcus.

Genetic elements responsible for phosphorylcholine incorporation on pneumococcal TA and LTA are poorly understood. A choline kinase activity has been purified and implicated in incorporation of choline into the pneumococcal cell wall (Whiting and Gillespie, 1997). Recently, Severin et al. (1997) and Yother et al. (1998) identified pneumococcal mutants that can grow without choline. The phenotype of these mutants included chain formation and defective autolysis. However, a genetic locus associated with choline metabolism has not been identified. Interestingly, phosphorylcholine has been shown to decorate lipopolysaccharide (LPS) of Haemophilus influenzae (Weiser et al., 1997). High-frequency, spontaneous phase variation of phosphorylcholine epitopes contributes to persistence of H. influenzae in the respiratory tract and to serum resistance (Weiser et al., 1998). The expression of phosphorylcholine epitopes on the cell surface requires the lic1 locus, which consists of four genes (licA, licB, licC and licD ) (Weiser et al., 1989a). LicA and LicB may serve as a choline kinase and a choline transporter respectively (Weiser et al., 1997). LicC is similar to several pyrophosphorylases (Weiser et al., 1997). LicD is homologous to a pneumococcal protein encoded by cpsG which is located in a locus for biosynthesis of capsular polysaccharide (Morona et al., 1997a). The biological function of CpsG is unknown.

In this study, we identified a lic locus in S. pneumoniae by screening an insertional duplication library. The pneumococcal lic locus contains the counterparts of H. influenzae licAD and additional open reading frames (ORFs). The most striking changes in phenotype, including choline uptake, transformation, penicillin-induced lysis and virulence, were found for licD2.


Identification of the pneumococcal lic locus

Screening of a library of insertion duplication mutants (Pearce et al., 1993) for penicillin tolerance and transformation deficiency yielded three independent mutations in a gene homologous to licD of H. influenzae (data not shown). To further characterize this region, the locus was retrieved from the partial genomic sequence database of S. pneumoniae ( The general organization of the pneumococcal lic locus is similar to the lic1 locus of H. influenzae and is depicted in Fig. 1A. The pneumococcal lic genes were named licA to licD after their counterparts in H. influenzae. LicA and LicB may serve as a choline kinase and a choline transporter respectively (Weiser et al., 1997). LicC is similar to several nucleotide pyrophosphorylases (Weiser et al., 1997). The coding sequences of licA, licB and licC are arranged in a head-to-tail fashion and these genes appear to be co-transcribed (data not shown).

Figure 1.

. Organization of the pneumococcal lic locus. A. Comparison of structures of S. pneumoniae and H. influenzae lic genes. The directions of ORFs are indicated by arrows. The potential functions of the gene products based on sequence similarity are listed. (Percent identities of amino-acid sequences are indicated.) B. Alignment of the licD homologues. The predicted amino-acid sequences of the H. influenzae licD (HIlicD), S. pneumoniae licD1 (LicD1), licD2 (LicD2), and cpsG (CpsG) were aligned by pileup and identical (black) and similar (grey) residues are highlighted by boxshade. Spaces (·) were introduced to maximize sequence homology.

Both S. pneumoniae and H. influenzae licD genes are homologous to cpsG, a pneumococcal gene in a locus of capsular polysaccharide biosynthesis (Fig. 1B). The biological function encoded by cpsG has not been determined. Interestingly, in contrast to a single licD gene in H. influenzae (Weiser et al., 1989a), S. pneumoniae possesses two contiguous licD homologues, designated licD1 (804 bp) and licD2 (810 bp) (Fig. 1A). licD1 and licD2 are homologous to each other with 44.7% identity and 55.6% similarity at the amino-acid level (Fig. 1B). In contrast to H. influenzae licD, licD1 and licD2 are separated from licAC by three additional ORFs and are transcribed in a direction divergent from licAC (Fig. 1A). The predicted amino-acid sequences of the ORF1, ORF2 and ORF3 show similarity to a putative oligosaccharide repeat unit transporter in S. pneumoniae (Morona et al., 1997b), a CDP–ribitol pyrophosphorylase in H. influenzae (Van Eldere et al., 1995), and a zinc-type alcohol dehydrogenase-like protein in Escherichia coli (Burland et al., 1995) respectively (Fig. 1A). The predicted coding regions of ORF1, licD1 and licD2 are nearly contiguous, suggesting that three genes are transcribed as a single mRNA unit. To test the possibility of licD1 and licD2 co-transcription, reverse transcription-PCR (RT-PCR) was used to amplify the junction region between the 3′licD1 and 5′licD2. Although no PCR product was detected without reverse transcriptase, a DNA band with an expected size of 265 bp was observed in the agarose gel for the RT mixture (data not shown). The DNA sequence of the RT-PCR product was identical to that of the targeted licD1licD2 region. The results indicated that licD1 and licD2 are transcribed in a single transcript in cultured organisms. Transcription of ORF1 was not determined. The three penicillin-tolerant, transformation-deficient mutants from the insertional duplication library carried mutations in licD2. Because initial library screening had provided strong phenotypes for licD2, such as transformation deficiency and penicillin tolerance, further analysis was focused on licD1 and licD2.

Insertional mutagenesis of licD1 and licD2

To assess the role of the licD1 and licD2 genes in phosphorylcholine metabolism, insertional mutagenesis was used to interrupt both genes as previously described (Pearce et al., 1993). Our attempts to knockout licD1 were unsuccessful. licD2-deficient mutants were created in capsular serotypes 2 (D39) and 6B and in R6x, the unencapsulated derivative of D39. An erythromycin resistance cassette was inserted after nucleotide 292 downstream of the predicted licD2 start codon. Because the ORF immediately downstream of licD2 (apparently encoding a carbamoyl-phosphate synthetase) is oriented in a divergent direction, it is unlikely that this insertion caused a polar effect on expression of its downstream genes. licD2 ::pIH2 was resistant to 1 μg ml−1 erythromycin and did not show a detectable difference in growth rates in C + Y medium compared with wild type. In addition, there was no apparent difference in chain formation. In contrast to reversible, spontaneous phase variation between opaque and transparent colony morphology in R6x (Weiser et al., 1994), licD2 ::pIH2 formed uniformly transparent colonies which were distinctly more transparent than the reference strain P126 (Fig. 2). The phenotype appeared to be stable during frequent passages in culture medium (data not shown).

Figure 2.

. Transparent colonies of the licD2 ::pIH2 mutant. licD2 ::pIH2, R6x, opaque (P125) and transparent (P126) variants of R6x (Weiser et al., 1998) were grown on a tryptic soy agar. Bacterial colonies were photographed under a phase-contrast microscope with transmitted illumination.

Decreased choline uptake in licD2::pIH2

Incorporation of radiolabelled choline into bacterial cell wall was measured to determine whether the insertional mutation in licD2 could affect phosphorylcholine metabolism in different growth stages. licD2 ::pIH2 and R6x were grown in the presence of [3H]-choline and aliquots of each culture were taken at four time points through logarithmic phase to assess choline incorporation into cell wall material. licD2 ::pIH2 showed ≈50% of [3H]-choline uptake of its parental strain through logarithmic growth (Fig. 3), indicating that licD2 is required for incorporation of some but not all phosphorylcholine into macromolecular cell wall. The remainder of choline uptake in licD2 ::pIH2 implies alternative mechanisms of choline incorporation into cell wall, perhaps involving licD1.

Figure 3.

. Incorporation of [3H]-choline into cell wall. licD2 ::pIH2 (circle) and R6x (square) were grown in the presence of [3H]-choline. Uptake of radiolabel into material precipitated in boiling SDS, the operational definition of cell wall, was followed from early (5 × 106 cfu) to late logarithmic phases (8 × 106 cfu). Data are representative of two experiments.

Changes in phosphorylcholine-containing LTA

A monoclonal antibody TEPC-15 has been shown to recognize a phosphorylcholine epitope in S. pneumoniae (Leon and Young, 1971). We reasoned that if licD2 is required for phosphorylcholine addition to TA and LTA, licD2 ::pIH2 might lose or alter the phosphorylcholine epitope recognized by TEPC-15. Mildly washing pneumococcal cells with PBS containing 2% choline is a well-described procedure to elute numerous CBPs, including autolysin and CbpA (Yother et al., 1992; Rosenow et al., 1997). However, analysis of this preparation by polyacrylamide gel electrophoresis and immunoblot using TEPC-15 also revealed the presence of phosphorylcholine-containing LTA, as confirmed by comparison with LTA controls (see below).

Although the LTA chains of licD2 ::pIH2 retained reactivity to TEPC-15 (Fig. 4), indicative of the presence of phosphorylcholine, these chains migrated aberrantly in SDS–PAGE gels compared with those of R6x. Because each repeat of the LTA chains in R6x contains two phosphorylcholine residues (Fischer et al., 1993), this raised the possibility that licD2 ::pIH2 LTA repeats lacked one of the two phosphorylcholines. Two LTA standards LTA344 and LTA395 were compared with the LTA of licD2 ::pIH2 and R6x by immunoblot analysis. LTA344 possesses predominantly two phosphorylcholine residues per repeat, whereas the majority of LTA395 chains contain only one phosphorylcholine per repeat (Fischer, 1997; Yother et al., 1998). In agreement with previous reports (Fischer et al., 1993), the LTA chains of R6x co-migrated with those of the two-choline species LTA344 (Fig. 4). Conversely, the LTA chains of licD2 ::pIH2 co-migrated with those of LTA395, suggesting that licD2 ::pIH2 may have incorporated one of the two phosphorylcholines on its LTA repeats. The licD2 ::pIH2 mutants in D39 and 6B backgrounds showed similar changes in LTA chains (Fig. 4). This is consistent with the decreased total choline uptake in licD2 ::pIH2 (Fig. 3).

Figure 4.

. Altered migration pattern of the phosphorylcholine-containing LTA chains in licD2 ::pIH2. Supernatants from 2% choline soaks of R6x, D39, 6B and the respective isogenic licD2 mutants as well as LTA standards LTA344 (two cholines per repeat) and LTA395 (one choline per repeat) were loaded in a 8–20% SDS–PAGE gel to separate LTA chains. The blot was reacted with a phosphorylcholine-specific monoclonal antibody, TEPC-15. The reactive ladders represented LTA chains with variable repeat units (Fischer et al., 1993). Positions of protein molecular weight markers are marked in kDa.

Loss of surface-exposed proteins in licD2::pIH2

Phosphorylcholine is required to anchor a family of CBPs to the pneumococcal cell surface. To determine whether mutation in licD2 could affect retention or presentation of CBPs, crude choline-binding proteins were eluted from the cultures of R6x and licD2 ::pIH2 by soaking cell pellets in 2% choline, and separating released material in the supernatant by SDS–PAGE gels. Immunoblots were reacted with an antibody against multiple pneumococcal CBPs as previously described (Rosenow et al., 1997). Several reactive proteins found in R6x were undetectable in licD2 ::pIH2 (Fig. 5). licD2 ::pIH2 and R6x showed similar levels of autolysin and CbpA, as detected by immunoblotting using specific antisera (data not shown). However, the identical blot did not react with an antiserum against the cytoplasmic protein RecA (data not shown), indicating that the 2% choline washing procedure did not lead to substantial disruption of the bacterial cells. This is consistent with the observation that the choline washing procedure did not affect bacterial viability (data not shown). Thus, these missing proteins appear to be surface exposed. The exact identities of these proteins remain to be determined.

Figure 5.

. Loss of several surface-exposed proteins in licD2 ::pIH2. Proteins of R6x and licD2 ::pIH2 were prepared by 2% choline soaking and separated by a 8–20% SDS–PAGE gel. The blot was reacted with an antiserum reactive to multiple pneumococcal proteins (Rosenow et al., 1997). Three proteins clearly absent in licD2 ::pIH2 are indicated by arrows. Positions of protein molecular weight markers are marked in kDa.

The autolytic defect of licD2::pIH2

Immunoreactive autolysin was detected appropriately in CBP preparations of licD2 ::pIH2, but these cells failed to lyse in stationary phase. This raised the possibility that the licD2 ::pIH2 autolysin may be produced and exported, but not functionally activated. Phosphorylcholine is required to convert pneumococcal autolysin from its inactive E-form to its active C-form (Holtje and Tomasz, 1975; Sanz et al., 1988). To determine whether antibiotics could activate the licD2 ::pIH2 autolysin, early log-phase organisms were treated with penicillin, and lysis was determined at various time points after treatment. The OD620 values of R6x was dramatically decreased 1 h after addition of penicillin, indicative of penicillin-induced lysis (Fig. 6). The licD2 ::pIH2 organisms had similar sensitivity to penicillin-induced lysis as R6x when the cultures were exposed to penicillin at very early growth stages (OD620 values of 0.3 and 0.35) (Fig. 6). However, licD2 ::pIH2 showed resistance to penicillin-induced lysis, similar to the autolysin defective strain lyt4-4 when the organisms were treated with penicillin at a later growth stage (OD620 values of 0.4) (Fig. 6).

Figure 6.

. The autolytic defect of licD2 ::pIH2 in late logarithmic phase. Lysis of licD2 ::pIH2 (circles), the autolysin-defective strain lyt4-4 (triangles) and R6x (squares) was followed by optical density (OD620) of the culture. Penicillin (0.02 μg ml−1) was added to the licD2 ::pIH2 culture at penicillin at OD620 0.3, 0.35 or 0.4 as indicated. Penicillin was added to R6x and lyt4-4 at OD620 0.4. Data are representative of three independent experiments.

To determine whether the autolytic defect of licD2 ::pIH2 derived from changes in cell wall substrate, the licD2 ::pIH2 cultures were treated with crude autolysin preparations from R6x and autolysin defective lyt4-4 in the presence of penicillin. Although licD2 ::pIH2 did not lyse in the presence of lyt4-4 autolysin, it was susceptible to lysis by R6x autolysin (data not shown), indicating that licD2 ::pIH2 cell wall was recognized and digested by active autolysin. Moreover, the licD2 ::pIH2 autolysin restored lysis of strain lyt4-4 (data not shown), indicating that it could function in an appropriate environment.

Transformation defect of licD2::pIH2

S. pneumoniae cells are capable of taking up foreign DNA at low cell density (Tomasz, 1965) in response to a competence-stimulating factor peptide (CSP) (Tomasz, 1965; Hui and Morrison, 1991; Havarstein et al., 1995). Phosphorylcholine is essential for natural competence because substitution of choline with ethanolamine in the culture results in loss of transformability (Tomasz, 1968). We sought to test whether insertional mutation in licD2 affected efficiency of pneumococcal competence. licD2 ::pIH2, a comA competence-defective mutant (SPRU113) (Pearce et al., 1994), and R6x were assessed for their ability to acquire a streptomycin resistance marker in the presence or absence of exogenous synthetic CSP (50 ng ml−1). In the absence of CSP, R6x acquired streptomycin resistance, whereas no streptomycin-resistant colonies were observed for the licD2 ::pIH2 and comA mutants (Fig. 7). Exogenous CSP restored transformation of comA (Pearce et al., 1994), but licD2 ::pIH2 showed only marginal transformability in the presence of CSP (Fig. 7).

Figure 7.

. Transformation defect of licD2 ::pIH2. Transformation efficiencies of licD2 ::pIH2, a comA-defective mutant (SPRU113) (Pearce et al., 1994) and the parent strain (R6x) were determined in the presence (black) or absence (hatched) of exogenous synthetic CSP (50 ng ml−1). Data are representative of four independent experiments; standard deviation did not exceed 10%.

Decreased adherence of licD2::pIH2 to human lung cells

Previous studies have shown that phosphorylcholine can mediate pneumococcal adherence to host cells, perhaps by interacting with the PAF receptor (Cundell et al., 1995b) and by anchoring adhesins to the cell surface (Rosenow et al., 1997). Fluorescein-labelled licD2 ::pIH2 and R6x were assessed for adherence to monolayers of type II alveolar cells (A549 cell line). A CbpA-deficient mutant (Rosenow et al., 1997) was used as a control. Compared with R6x, LicD2 mutant showed a significant 30% decrease in adherence to A549 cells, a value similar to that of the adherence-deficient strain CbpA (Fig. 8).

Figure 8.

. Decreased adherence of licD2 ::pIH2 to human lung cells. licD2 ::pIH2 (white), the cbpA-deficient mutant (hatched) (Rosenow et al., 1997), and the parent R6x (black) were fluorescein labelled and allowed to adhere to the monolayers of A549 cells. After extensive washing steps, adherent bacteria were counted visually under a fluorescence microscope and expressed as a percentage of the R6x control. The P-values for cbpA and licD2 were 0.049 and 0.015, respectively, as determined by the unpaired two-tailed Student's t-test.

Decreased virulence of licD2::pIH2 in mice

Rates of nasopharyngeal colonization in an infant rat model by the wild-type strains 6B and D39 and the respective isogenic licD2 ::pIH2 mutants were compared. licD2 ::pIH2 in both backgrounds had reduced levels of nasopharyngeal colonization (Table 1). Furthermore, licD2 ::pIH2 showed decreased virulence in CF1 mice compared with control when challenge occurred intraperitoneally (Fig. 9). This attenuation of virulence by licD2 mutation was more pronounced in 6B background, in which all 10 mice challenged with licD2 ::pIH2 survived beyond 6 days whereas all animals infected with the parent strain died on day 6 after infection.

Table 1. . Colonization of the nasopharynx of rats by the licD2 ::pIH2 mutants. a. Groups of nine infant rat pups were inoculated with l04 cfu intranasally in three separate experiments (total n = 27 per strain). Nasal lavage was carried out on day 5 and cfus were determined. Quantification of bacteria per 10 μl lavage: none, < 1 bacteria; moderate, 1–50 bacteria; heavy, > 50 bacteria.b. Chi-squared analysis showed a significant difference in percentage survival between the parent and the respective licD2 ::pIH2 mutant in both D39 and 6B backgrounds (P < 0.001).Thumbnail image of
Figure 9.

. Decreased virulence of licD2 ::pIH2 in mice. Mice were challenged intraperitoneally with 10× the LD50 of the wild-type strains D39 (1 × 104 cfu) and 6B (2 × 108 cfu) (open squares) and the isogenic licD2 ::pIH2 mutants (closed circles). Survival is charted as a Kaplan–Meier plot. Statistical analyses as determined by the log-rank test support significantly prolonged survival after challenge with the licD2 ::pIH2 mutants (P = 0.007 for D39, P < 0.0001 for 6B).


Phosphorylcholine-containing TA and LTA are major components of the pneumococcal cell surface. Although phosphorylcholine is essential for the biology of S. pneumoniae, the genetic elements involved in pneumococcal phosphorylcholine incorporation have not been determined. In this work, we identified a genetic locus in S. pneumoniae that demonstrates sequence similarity to four H. influenzae chromosomal genes, licA–D (Weiser et al., 1989b), that are required for phase variable incorporation of phosphorylcholine into LPS (Weiser et al., 1997). Although single pneumococcal genes are similar to H. influenzae licA, licB and licC, respectively, two pneumococcal genes, licD1 and licD2, are homologous to the H. influenzae licD. Because S. pneumoniae and H. influenzae colonize the same environmental niches in the human nasopharynx, lung and blood, it is possible that the lic genes serve similar purposes for both species. The loss of phase variability in licD2 ::pIH2 supports this hypothesis.

The licD1 and licD2 genes are transcribed as a single unit as indicated by RT-PCR and are oriented divergent to licAC. Given the fact that the H. influenzae licA–D are organized in the same orientation as an operon (Weiser et al., 1989b), the divergent organization of the pneumococcal lic genes suggest that they may be regulated differently from those in H. influenzae. Consistent with this hypothesis, choline is an essential nutrient for growth in S. pneumoniae but not in H. influenzae. Moreover, two phosphorylcholines are attached to pneumococcal TA and LTA repeats (Tomasz, 1968; Fischer et al., 1993), whereas one phosphorylcholine is attached to the saccharide portion of each LPS molecule in H. influenzae (Weiser et al., 1998).

licD2 is required for choline metabolism as shown by decreased uptake of radiolabelled choline and aberrant migration of LTA chains in licD2 ::pIH2. Choline is attached to O-6 of each of two N-acetyl-d-galactosaminyl residues in the LTA or TA backbone (Behr et al., 1992; Fischer et al., 1993). Although the choline content of the licD2 ::pIH2 LTA remains to be determined, comparison with LTA standards with defined differences in the two types of choline adducts suggests that licD2 ::pIH2 fails to incorporate one of the two phosphorylcholine residues into each LTA repeat unit. The cell wall of licD2 ::pIH2 contained ≈50% of [3H]-choline incorporated by R6x. These observations suggest that licD2 encodes an enzyme that transfers phosphorylcholine from a donor (CDP-choline or -lecithin) to LTA, and perhaps also to TA. Accordingly, the protein product of the homologous LicD1 could perform a similar reaction for a second choline, consistent with the partial [3H]-choline uptake and TEPC-15 antibody reactivity in licD2 ::pIH2. Failure to isolate licD1 mutants suggests that the protein product of licD1 may be essential for pneumococcal growth. Alternatively, insertion mutation in licD1 could produce a polar effect on expression of licD2, which would lead to inactivation of both licD1 and licD2 and a lethal outcome. Further deletion mutations in licD1 or generation of conditional licD1 mutants may help to differentiate these possibilities. Regardless, the pneumococcus appears to require at least one functional licD gene.

The autolysis defect of licD2 ::pIH2 did not appear to result from lack of autolysin expression, to export or from change(s) in cell wall target for autolysin digestion. licD2 ::pIH2 crude autolysin could restore lysis of the autolysin-defective strain lyt4-4, and licD2 ::pIH2 cell walls could be lysed with R6x autolysin. This raises the likelihood that licD2 ::pIH2 lacks the ability to convert autolysin from its inactive E-form to its active C-form. This conclusion is in agreement with previous studies demonstrating that phosphorylcholine is essential for autolysin activation (Holtje and Tomasz, 1975; Sanz et al., 1988). These observations suggest that two chemically distinct choline adducts on LTA and TA may represent functionally distinct docking stations. In this context, one functional class of phosphorylcholine, perhaps related to LicD1 function, may anchor choline-binding proteins to the surface, whereas a second class, incorporated by LicD2, serves to activate autolysin and transformation either alone or in combination with the choline incorporated by LicD1. This idea is further supported by retention of choline-binding proteins on licD2 ::pIH2 surface but loss of autolysis and transformation deficiency in licD2 ::pIH2. However, strain Rx1 harbours only one phosphorylcholine on 80% of TA and LTA repeats (Fischer, 1997; Yother et al., 1998). Yet these organisms retain CBPs, PspA and LytA and are capable of transformation and autolysis (Yother et al., 1998). If our hypothesis is correct, the residual 20% teichoic and LTA repeats with two phosphorylcholines in Rx1 may be sufficient to confer function to the CBPs. Preliminary analysis by PCR amplification suggests that Rx1 contain both licD1 and licD2 (J.-R. Zhang and E. I. Tuomanen, unpublished). Thus, its deficiency in choline incorporation may be caused by genetic changes independent of licD genes.

LicD2 may participate in pneumococcal phase switching because licD2 ::pIH2 appears to be a uniform population of the transparent colony phenotype, even after a series of in vitro passages. Although licD2 ::pIH2 exhibited the transparent colony phenotype, it is quite extreme in comparison to the reference transparent variant. licD2 ::pIH2 showed decreased choline uptake and apparently lacked one of the two phosphorylcholines on its LTA, whereas the reference transparent variant possesses higher amounts of phosphorylcholine than the opaque variant (Kim and Weiser, 1998; Weiser, 1998). It is likely that changes in other factors such as cell wall-associated proteins also contribute to transparent colony morphology. In this regard, licD2 ::pIH2 demonstrated a complete absence of several surface proteins, whereas the reference transparent variant harbours less autolysin and PspA and more CbpA (Weiser et al., 1996; Rosenow et al., 1997). These differences suggest phase variation is not explained by licD2 alone, but that this regulatory mechanism may involve LicD2.

The impaired adherence to human lung cells by licD2 ::pIH2 is consistent with the previous finding which suggests that choline may enhance pneumococcal attachment to host cells by direct interaction with the PAF receptor (Cundell et al., 1995b). Accordingly, the licD2 ::pIH2 mutants in D39 and 6B showed less nasopharyngeal colonization in infant rats and reduced systemic virulence in CF I mice. However, these results do not explain loss of virulence in the bloodstream. In this compartment, choline would serve as a target for C-reactive protein (Szalai et al., 1997). Bacteria with less choline, such as opaque variants, would escape this innate clearance mechanism. The role of choline in the vascular compartment requires further study.

Experimental procedures

Bacterial strains and growth conditions

S. pneumoniae R6x (Tiraby and Fox, 1973) was a derivative of the unencapsulated Rockefeller strain R36A that in turn was a derivative of strain D39 (Avery et al., 1944). The comA mutant SPRU113 (Pearce et al., 1994) was derived from R6x by insertion of an erythromycin resistance cassette. The autolysin-deficient mutant lyt4-4 was kindly provided by Dr A. Tomasz, Rockefeller University. Capsular serotype 6B was obtained from Dr R. Austrian, University of Pennsylvania, Philadelphia, PA, USA.

Pneumococci were routinely grown on tryptic soy agar (TSA) (Difco) supplemented with sheep blood to a final concentration of 3% (v/v). For colony opacity examination, translucent agar is required. Therefore, blood was replaced by 100 μl catalase (5000 U), a requirement to neutralize H2O2 produced by aerobic growth. For growth in liquid culture, the bacteria were grown in a semisynthetic casein hydrolysate medium supplemented with yeast extract (C + Y) (Lacks and Hotchkiss, 1960). For the selection and maintenance of pneumococci containing chromosomally integrated plasmids, bacteria were grown in the presence of 1 μg ml−1 erythromycin.

Construction of licD1and licD2mutants

The licD1 and licD2 mutants in pneumococcal strains were constructed by insertion duplication as previously described (Pearce et al., 1993). Based on partial genomic sequence of S. pneumoniae type 4, a 5′licD1 region of strain R6x was PCR amplified with (+) strand primer PR016 (5′-CCTGAATTCTTAAAATGAAACAACTAACCGT-3′) and (−) strand primer PR017 (5′-GAAGGGATCCTCAAAGCGATCTATAGGGAAAAT-3′). Similarly, 5′licD2 regions of strains R6x, D39 and 6B were amplified using (+) strand primer PR011 (5′-ATTGGAGAATTCGGATGCAATATTTAG-3′) and (−) strand primer PR021 (5′-GAGAATTCGTAGAAGTGTCCAAAATCGATGCG-3′). For cloning purposes, terminal EcoRI or BamHI restriction site, as underlined above, was engineered in both (+) and (−) strand primers. For licD1, the amplified DNA fragment was then digested with EcoRI and BamHI and ligated into EcoRI/BamHI-digested pJDC9 plasmid (Chen and Morrison, 1988). The licD2 PCR DNA was digested with EcoRI and cloned in EcoRI–digested pJDC9. The recombinant plasmids containing the licD1 and licD2 sequences were designated pIH1 and pIH2 respectively. pIH1 and pIH2 were propagated in E. coli strain XL-1 blue (Stratagene) in the presence of 200 μg ml−1 erythromycin, according to the standard procedure (Sambrook et al., 1989). The inserts of recombinant plasmids were sequenced to confirm the insertion site and subsequently used to transform pneumococcal strains R6x, D39 and 6B. Pneumococcal transformants were plated on TSA plates containing 1 μg ml−1 erythromycin and 3% sheep blood. Plasmid integration into the pneumococcal chromosome was verified by PCR amplification, sequencing and Southern hybridization.

DNA and PCR techniques

Pneumococcal chromosomal DNA was prepared by the phenol–chloroform extraction method according to Pearce et al. (1993). Recombinant plasmid and PCR DNA samples were purified using Wizard columns (Promega). All PCR amplifications were performed using a Thermalase PCR kit (Ameresco) in a DNA thermal cycler (Perkin-Elmer). PCR conditions consisted of denaturation at 96°C for 1 min, annealing at 55°C for 40 s, and extension at 72°C for 2 min, followed by 35 more cycles.

To determine transcription of licD1 and licD2 by RT-PCR, total RNA was extracted from mid-log phase R6x cultures by a RNeasy extraction kit (Qiagen) and treated with RNase-free DNase (Promega). The RNA preparation was used to produce cDNA with murine leukaemia virus reverse transcriptase (Promega) and a 5′licD2 reverse primer PR010 (5′-TTCACGCTTATAAATTGGAGGAT-3′) in a 20 μl reaction. A 3′licD1 forward primer PR009 (5′-CTTGTAGCGAGGGTGATTTTCTTCTTC-3′) and the primer PR010 were included to amplify the junction region between 3′licD1 and 5′licD2 using 2 μl of the RT reaction by the Thermalase PCR kit as described above. The PCR products were sequenced using primer PR010 to verify the identity of DNA fragments.

Nucleotide sequencing was performed with an ABI 377 automatic DNA sequencer (Perkin-Elmer/ABI). The gap and pileup programs of GCG (Version 8, Genetics Computer Group) were used to determine sequence similarity and identity. boxshade was performed at the web site of the Bioinformatics Group of the ISREC ( Searches for sequence homology were performed at the National Center for Biotechnology Information using the blast programs (Altschul et al., 1990). The partial genomic sequence database of S. pneumoniae is available at the web site of the Institute for Genomic Research ( The sequences of H. influenzae lic genes and pneumococcal cpsG are contained in GenBank accession numbers M27280 and AF004325 respectively. The DNA sequences of ORF1, licD1 and licD2 have been submitted in the GenBank under accession number AF106539.

Immunoblotting analysis

The expression of pneumococcal proteins and phosphorylcholine was assessed by immunoblotting of 2% choline soaking preparations of bacterial cells. Bacteria were grown to an OD620 of ≈0.6 and centrifuged at 5000 × g for 10 min. The cell pellet was washed twice with PBS by centrifugation and resuspended at a final concentration of ≈1010 cells ml−1 in PBS containing 2% choline chloride (w/v). The bacterial suspension was gently shaken for 20 min at room temperature, followed by centrifugation at 5000 × g for 10 min. The supernatant was dialysed against PBS, and its protein concentration was determined by a Protein Assay kit (Bio-Rad). The supernatant preparation was subjected to 8–20% SDS–PAGE gels and transferred to Immobilon-P membranes (Millipore). For detection of CbpA, autolysin and RecA, the membrane was reacted with appropriate rabbit antisera (1 : 20 000) and detected with horseradish peroxidase-conjugated goat anti-rabbit IgG (1 : 10 000). The protein bands were visualized by a Chemiluminescence Kit for Western blot (Amersham) according to the supplier's instructions. Antisera against CbpA, autolysin and RecA were prepared in New Zealand white rabbits as described previously (Rosenow et al., 1997).

The phosphorylcholine epitope was detected using a monoclonal antibody TEPC-15 (1 : 5000), kindly provided by Dr M. Potter (NCI, the National Institutes of Health). LTA standards LTA344 and LTA395 were isolated from pneumococcal strains R6x and Rx1 (Fischer, 1997; Yother et al., 1998) respectively.

Measurement of [3H]-choline uptake

Radiolabelling of pneumococcus with [3H]-choline was carried out as described previously (Tuomanen et al., 1988). Bacteria were grown in 5 ml of the modified C + Y medium containing 5 mg ml−1 unlabelled choline and 50 μCi of 3-methyl [3H]-choline chloride (Amersham). At hourly intervals, the optical density (OD620) of the culture was measured, and triplicate 200 μl aliquots were removed from the culture and quickly frozen in an alcohol/dry ice bath. Crude cell wall was precipitated by boiling the frozen aliquots in 5% SDS for 20 min as described previously (Garcia and Tomasz, 1987). Each of the boiled culture aliquots was filtered individually through a 0.2 μm filter (Millipore), the filters were dried at 100°C for 30 min, and radioactivity of the filters was measured in a scintillation counter.

Transformation assay

Competence for DNA transformation was assessed as described previously (Pearce et al., 1994) by scoring for acquisition of a chromosomal streptomycin resistance marker arising spontaneously in R6x. Pneumococci were harvested at an OD620 of 0.1–0.2 and incubated with chromosomal DNA from a streptomycin-resistant strain (1 μg ml−1 final concentration) for 30 min at 30°C. DNase was added to a final concentration of 10 μg ml−1 to stop further DNA uptake. The cells were grown for an additional 90 min at 37°C to allow expression of the streptomycin resistance gene. Serial dilutions were streaked on TSA blood plates containing 100 μg ml−1 streptomycin to determine the transformation efficiency, calculated as previously described (Pearce et al., 1994). To assess the influence of exogenous CSP on transformation efficiency, pneumococci were treated with synthetic peptide at a concentration of 50 ng ml−1.

Penicillin lysis

Bacteria were cultured to an OD620 of 0.3–0.4 in C + Y broth. Penicillin was then added to the cultures to a final concentration of 0.06 μg ml−1. The optical density (OD620) of the cultures was measured at various time points. Crude pneumococcal autolysin was prepared as described previously (Williamson and Tomasz, 1980). Bacterial cells in a 1000 ml culture were pelleted by centrifugation. The pellet was washed twice in PBS, resuspended in 20 ml PBS and frozen at −70°C. The cells were thawed on ice and sonicated to complete lysis at 4°C. The lysate was centrifuged at 70 000 × g for 40 min at 4°C, and the supernatant was collected as crude autolysin. For autolysis assay, the autolysin preparation (1 ml), along with 20 μl of penicillin (30 μg ml−1), was directly added to 10 ml culture, and cell density was monitored by optical absorption as described above.

Adherence assay

Ability of licD2 mutants to adhere to the human type II alveolar cell line A549 (American Type Culture Collection) was assessed in 96 well plates as described previously (Cundell and Tuomanen, 1994). The A549 cells were cultured in F12K medium (ATCC) supplemented with 10% (v/v) fetal bovine serum according to supplier's instructions. Bacterial cells were grown to an OD620 of 0.4 in C + Y medium and washed in a carbonate buffer (0.05 M sodium carbonate, 0.1 M sodium chloride). The pellet was resuspended in 1 ml of the carbonate buffer and labelled with fluorescein (1 mg ml−1) (Sigma) in the dark for 20 min at room temperature (Geelen et al., 1992). The labelled bacteria were washed three times in the carbonate buffer and resuspended in F12K medium. The organisms (5 × 107 per well) were incubated with human cell monolayers for 30 min at 37°C. After three washes with the carbonate buffer to remove unbound bacteria, the monolayers were fixed with 2.5% glutaraldehyde for 3 min and washed five times in PBS. The bound bacteria were counted under a fluorescence microscope (Diaphot-TDB; Nikon) with an IF DM-510 filter. Adherence was expressed as the number of the adherent bacteria per 100 cells. Values represented averages of three or four wells from at least three separate experiments.

Animal studies

The animal work was approved by the Animal Resource Committee at the St. Jude Children's Research Hospital. Inoculation of rats and mice with pneumococci was carried out as previously described (Weiser et al., 1994; Rosenow et al., 1997). For nasopharyngeal challenge, each of nine infant Sprague-Dawley rats (1–5 days old) were infected intranasally with a dose of 1 × 10cfu per pneumococcal strain. Strains D39 and 6B as well as their isogenic licD2 ::pIH2 mutants were tested in three separate experiments. Nasal lavage was performed with each rat on day 5. The fluids from nasal washes were diluted in PBS and plated on the TSA plates supplemented with 3% sheep blood, and bacterial colonies were determined.

To assess virulence in blood, CF1 mice (12 mice per group for D39 and 15 mice per group for 6B) were inoculated intraperitoneally with strains D39 (1 × 10cfu) and 6B (2 × 10cfu) as well as the same numbers of their isogenic licD2 ::pIH2 bacteria. Numbers of surviving animals were counted daily for 7 days.


  1. *Present address: SmithKline Beecham, Brussels, Belgium


We thank H. R. Masure for advice and review of the manuscript; T. Bennett for providing technical assistance in immunoblot analysis and RT-PCR; J. Shenep for help in statistical analysis; B. Spellerberg for the screening of the library that led to the identification of the licD2 mutant; R. Novak for assistance in characterization of the licD2 phenotypes. This work was supported in part by NIH grants AI 27913 and AI 39482, Cancer Center Support CORE grant CA21765, and the American Lebanese Syrian Association Charities (ALSAC).