Streptococcus pneumoniae is an important human pathogen that has an absolute nutritional requirement for choline. Replacement of this amino alcohol in a synthetic medium by structural analogues, such as ethanolamine (EA cells), leads to alterations in several physiological properties including cell separation (Tomasz, 1968, Proc Natl Acad Sci USA59: 86–93). Identical changes including chain formation and loss of autolytic properties can also be induced by adding up to 2% choline chloride to the growth medium (Briese and Hakenbeck, 1985, Eur J Biochem146: 417–427). Choline has been shown to inhibit the LytA pneumococcal autolysin (an N-acetylmuramoyl-L-alanine amidase) by preventing its attachment to wall teichoic acids (Giudicelli and Tomasz, 1984, J Bacteriol158: 1188–1190). In addition, it has been shown that the pneumococcal surface protein A (PspA) is anchored to the choline residues of the membrane-associated lipoteichoic acids (Yother and White, 1994, J Bacteriol176: 2976–2985).

Pneumococcus synthesizes several proteins that recognize and bind to the choline residues of the teichoic and lipoteichoic acids through a specialized domain, the choline-binding domain (ChBD) (Sánchez-Puelles et al., 1990, Gene89: 69–75). ChBD is built up of six or more well-conserved motifs, each about 20-amino-acid residues long (García et al., 1988, Proc Natl Acad Sci USA85: 914–918; García et al., 1998, Microb Drug Resist4: 25–36). Currently, there is an increasing interest in the study of pneumococcal genes coding for proteins that possess ChBDs. These proteins have been demonstrated to participate in a series of important biological functions such as cell adhesion and division. For many years, we have studied the molecular structure and biological role of the lytic enzymes of pneumococcus and its bacteriophages (López et al., 1997, Microb Drug Resist3: 199–211). These enzymes, which exhibit different chemical substrate specificities (i.e. lysozymes, amidases and glucosaminidases), display a modular organization in which the catalytic domain and the ChBD are located at the N- and C-terminal moieties of the protein respectively.

It is likely that a defective autolytic system might explain the physiological alterations leading to chain formation in S. pneumoniae. Recently, two independent experimental approaches have generated pneumococcal mutants that do not require choline or analogues for growth (Severin et al., 1997, Microb Drug Resist3: 391–400; Yother et al., 1998, J Bacteriol180: 2093–2101). These mutants form long chains when growing under choline-free conditions, and it has been claimed that the lack of an active LytA amidase, the main pneumococcal lytic enzyme, could be responsible for impaired cell separation at the end of cell division. Nevertheless, previous studies have demonstrated that the primary biological consequences of the lytA gene deletion were the formation of small chains (six to ten cells) and the absence of lysis in the stationary phase of growth (Sánchez-Puelles et al., 1986, Eur J Biochem158: 289–293; Ronda et al., 1987, Eur J Biochem164: 621–624) (Fig. 2A). A recent study of ABC-type Mn permease-defective mutants of pneumococcus also reports chain formation in the stationary phase of growth (psaC and psaD mutants) or the appearance of small conglomerates or simply aberrant morphology (psaA and psaB mutants) (Novak et al., 1998, Mol Microbiol29: 1285–1296). Nevertheless, a gene involved in the formation of long chains is not yet known in pneumococcus.


Figure 2. . Morphology of M31lytB mutant. Chain formation of M31lytB as examined by phase-contrast (B) and transmission electron microscopy (C). The parental strain M31 is also shown (A). In (A) and (B), the bar represents 25 μm. In (C), the bar represents 2 μm.

Download figure to PowerPoint

The fact that most bacteria possess several lytic enzymes makes it difficult to determine the precise physiological role of these enzymes. Overcoming this limitation requires experiments with well-defined, single, or perhaps multiple, mutants with altered peptidoglycan hydrolase. Using a previously published procedure to characterize the pneumococcal glucosaminidase (García et al., 1989, Biochem Biophys Res Commun158: 251–256), we identified, by SDS–PAGE, four protein bands with apparently strong choline binding affinity. One of these bands was excised from the acrylamide gel, and the N-terminal amino acid sequence was found to be Ser-Asp-Gly-Thr-Trp-Gln-Gly. This sequence was compared with the translated version of the partial nucleotide sequence of the S. pneumoniae genome (, and a perfect match was found with an internal peptide of a gene product. This gene (hereafter designated lytB), located in the 10 373 bp contig no. 4117, was polymerase chain reaction (PCR) amplified, sequenced and preliminarily characterized (accession no. AJ010312). The putative 76.4 kDa LytB protein (658 amino acid residues) displays a modular organization different from all the ChBD proteins described previously in the pneumococcal system (Fig. 1A). Furthermore, this enzyme contains a 23-amino-acid-long, cleavable signal peptide (predicted Mr of the processed protein 73 800), as in the case of the S. pneumoniae LytC lysozyme, a new murein hydrolase recently identified in our laboratory (manuscript in preparation). It appears that LytB has a completely different organization from that described for other enzymes of this type (Fig. 1A) (López et al., 1997, Microb Drug Resist3: 199–211), because the order of the ChBD, with its 15 repeated motifs, and the catalytic domains is reversed in LytB. Sequence comparison suggests that LytB might be a glucosaminidase (Fig. 1B), possibly that previously studied biochemically in our laboratory (García et al., 1989, Biochem Biophys Res Commun158: 251–256). Most interestingly, the construction of a lytB pneumococcal mutant strain (R6B strain) by insertion–duplication mutagenesis resulted in the formation of long chains directly demonstrating that LytB plays a fundamental role in pneumococcal cell separation at the end of cell division. Still longer chains (more than 100 cells) were observed when the lytB mutation was introduced into the amidase-deficient, ΔlytA mutant M31 (Sánchez-Puelles et al., 1986, Eur J Biochem158: 289–293) (Fig. 2), which confirms that the LytA amidase contributes in a moderate way to cell separation, as pointed out above. Cells within the chain are held together by thin filamentous material that seems to be an extension of the polar cell walls, as reported previously for EA cells (Tomasz et al., 1975, J Supramol Struct3: 1–16) (Fig. 2C). In remarkable contrast to the EA cells, the recently described psa mutants and the choline-independent strains that exhibit the formation of chains in pneumococci, the R6B mutant did autolyse in the stationary phase as expected for a lytA+ strain. Furthermore, pneumococci undergo rapid lysis upon exposure to sodium deoxycholate (a classical test for identifying S. pneumoniae) by triggering the LytA activity. As expected, the R6B mutant was as sensitive to lysis by sodium deoxycholate as the wild-type pneumococcal strain (data not shown).


Figure 1. . Domain organization and sequence comparison of LytB and other murein hydrolases. A. Comparison of the most relevant structural characteristics of several peptidoglycan hydrolases of S. pneumoniae and its bacteriophages. bsl00029bsl00030bsl00031bsl00032 represent the domain containing the active site of the enzyme. bsl00033bsl00034 represents the binding domain with the variable number of motifs. bsl00035 represents the signal peptide. The scale at the top represents the amino acid number. B. Computer-generated alignment (BESTFIT) of the LytB domain putatively containing the active site with the glucosaminidase domain of the Staphylococcus aureus Atl autolysin (Oshida et al., 1995, Proc Natl Acad Sci USA92: 285–289). Residues on black boxes indicate identical amino acids, whereas conserved substitutions are shown in dotted boxes. Numbers at the right correspond to the amino acid positions.

Download figure to PowerPoint

It is conceivable that the evolutionary pressure operating in the in vivo environment of pneumococci selects against the loss of autolytic activity, as the breaking up of bacterial chains might improve the chances of the pathogen's survival in its encounter with phagocytic cells (Tuomanen et al., 1988, J Infect Dis158: 38–43). Separation of fully divided cells is not essential for further growth, division and survival of the bacteria, as also suggested by the normal growth rate exhibited by the R6B mutant described here. However, chaining does inhibit the dissemination of daughter cells and, thus, could affect distribution, for example towards nutrients or away from inhibitors, or the dissemination of the microorganism during an infection. Therefore, cell separation could influence pathogenesis (Berry et al., 1989, Infect Immun57: 2324–2330).

In summary, the observations reported here show the fundamental role of LytB in the terminal step that leads to the separation of the daughter cells in S. pneumoniae. A detailed molecular and biochemical approach to the study of the LytB protein by cloning and expression of the lytB gene will contribute in the future to provide interesting data to explain fully the physiological conditions underlying the intriguing impaired separation of the pneumococcal strains that form chains. Moreover, the new LytB murein hydrolase described here could be an attractive specific target susceptible for the development of new selective antibacterial drugs.