Background

Development of resistance to beta-lactam antibiotics in Streptococcus pneumoniae involves alterations in the target proteins, the penicillin-binding proteins (PBPs) which result in decreased affinity to beta-lactams. In order to identify individual mutations in S. pneumoniae that are related to the resistance phenotype, a series of independent mutant families has been selected in the laboratory using stepwise increasing concentrations of antibiotics [1]. Two beta-lactams were chosen for selection: piperacillin, which induces rapid lysis in the bacteria, and cefotaxime which does not interact with PBP2b and leads to a tolerant response [2]. Point mutations in pbp2b from piperacillin-resistant mutants and in pbp2x from cefotaxime resistant mutants have been described [3–5]. Surprisingly, a decrease in antibiotic susceptibility in some mutants correlated with a mutation in non-PBP genes [6]. In two piperacillin-resistant mutants, P106 and P104, obtained independently after the first selection step before the introduction of PBP mutations, the putative glycosyltransferase (GT) gene cpoA was affected [7]. Decreased susceptibility for piperacillin of the cpoA mutants was accompanied by a pleiotropic phenotype such as a defect in genetic competence and reduced amount of PBP1a. This indicated a novel mechanism directed against the activity of lytic β-lactams in S. pneumoniae distinct from target-mediated resistance.

The CpoA gene spr0981 and the adjacent gene spr0982 encode putative GTs which belong to the GTB-type superfamily (GT1-YqgM-like family). Members of this GT family are anchored in the membrane cytoplasmic interface by hydrophobic and charge interactions [8, 9] and transfer a sugar moiety to an acceptor molecule located in the inner leaflet of the membrane. Therefore, it had been proposed that CpoA perfoms a similar function in S. pneumoniae[7]. Meanwhile, in vitro studies revealed that both proteins are involved in the synthesis of glycolipids, with Spr0982 acting as α-monoglucosyl-diacylglycerol (GlcDAG) synthase and CpoA as a α-galactosyl-glucosyl-diacylgylcerol (GalGlcDAG) synthase [9, 10]. These two glycolipids occur at a ratio of approximately 1:2.5 in the S. pneumoniae membrane [11], in addition to phosphatidyl glycerol and cardiolipin which constitute the major phospholipids [12].

By consecutively synthesizing one nonbilayer-prone (mono-glucosyl-DAG) and one bilayer-forming glycolipid (di-glycosyl-DAG), the function of the GTs is crucial for the bilayer spontaneous curvature which affects the physical properties of the cytoplasmic membrane [13]. An example is the mycoplasma Acholeplasma laidlawii, where bilayer curvature is extensively regulated by two closely related GTs consecutively synthesizing monoglucosyl-DAG and diglucosyl-DAG [9, 13], enzymes that are homologous to S. pneumoniae Spr0982 and CpoA. Thus it is most likely that CpoA and Spr0982 play a critical role in S. pneumoniae related to membrane associated functions in agreement with the pleiotropic phenotype of the CpoA mutants mentioned above. GlcDAG is the proposed lipid anchor of the essential choline-containing lipoteichoic acid (LTA) of S. pneumoniae[14]. In fact, spr0982 has been listed among essential genes of this organism [15].

In the present report, a cpoA deletion mutant was constructed and compared to the CpoA mutants P106 and P104; moreover, the cpoA operon was investigated by mutational analysis. The aim of this study was to examine the function of CpoA in vivo, and to further our understanding on the physiological consequences of cpoA mutations.

Results

The CpoA gene is part of an operon with five downstream genes

P104 and P106 are spontaneous piperacillin-resistant laboratory mutants isolated independently after one selection step from the laboratory strain S. pneumoniae R6 [4, 7]. Both mutants contain a mutation affecting CpoA: in P104, a transversion within cpoA GTA to GGA led to a Gly12Val exchange in the predicted protein product, whereas in P106, one adenine nucleotide was deleted 15 base pairs (bp) upstream of the proposed cpoA start codon (ATG2 in Figure 1) [7]. Although ATG2 is not preceded by a classical Shine Dalgarno sequence, this deletion was suspected to affect the efficiency of ribosome binding to the cpoA transcript [7]. However, the possibility remained that translation actually starts at an alternative start codon (ATG1 in Figure 1) 27 bp upstream of ATG2 which is preceded by a perfect −10 region. In this case, the deletion in P106 would lead to a frameshift in the 5th codon and thus to the production of a nonsense peptide.

Figure 1
figure 1

Genes, transcription and deletions in the cpoA-spr0985 region of S. pneumoniae R6. (A) Wide horizontal arrows indicate genes apparently co-transcribed with cpoA (black), and flanking genes (white). spr0983.1 has not been annotated in the R6 genome [20], but its presence has been predicted from other S. pneumoniae genomes such as TIGR4 [56]. The positions and extend of in-frame deletions are shown as white boxes below the respective genes. Lines above the genetic map represent DNA products obtained by RT-PCR with total RNA and gene-specific primers. The positions of the promoter P cpoA and of putative ρ-independent terminators (T1 [ΔG = −10.4 kcal/mol], T2 [ΔG = −10.1 kcal/mol]) are given by angled and vertical arrows, respectively. (B) The nucleotide sequence upstream of S. pneumoniae R6 cpoA and putative 3'-coding sequences is shown together with the predicted peptide sequence (Sp). The −10 element of P cpoA is underlined, and the transcription start site (+1) is indicated with an angled arrow. The position of an adenine nucleotide, deleted in the mutant strain P106 [7] is marked with *Δ. Two potential start codons of the cpoA gene (ATG1, ATG2; see text for detail) are underlined. The respective cpoA sequences of S. mitis B6 (Sm) and S. oralis Uo5 (So) are shown below.

Full size image

To first clarify this issue, the expression signals of cpoA were mapped. The 5' end of cpoA mRNA was determined by RACE, and shown to be located 27 bp upstream of ATG2 (Figure 1B). Since this is exactly the position of the alternative start codon ATG1, translation initiation at ATG1 would imply that the cpoA transcript is leaderless [16]. In order to see whether ATG1 is indeed functional or whether ATG2 is required for translation, three plasmids were constructed in which the inferred promoter P cpoA together with either both, ATG1 and ATG2 (P cpoA -ATG12), ATG1 plus a mutated ATG2 (P cpoA -ATG1ATA2), or ATG1 only (P cpoA -ATG1), was translationally fused with the lacZ reporter gene. After single-copy integration of the resulting reporter constructs at the bgaA locus of R6, the expression of lacZ was determined in two transformants in up to three experiments. Beta-galactosidase activity was similar in R6P cpoA -ATG1ATG2 and R6P cpoA -ATG1ATA2 (190–330 Miller Units), and slightly lower in R6P cpoA -ATG1 containing a shorter region upstream of lacZ (140–150 Miller Units), clearly documenting that ATG1 is the translation initiation site of cpoA and that the cpoA transcript is indeed leaderless. In this case, P106 contains a deletion within the structural gene resulting in a frameshift within the 5th codon consistent with the failure to detect CpoA in P106 with a specific anti-CpoA antiserum [7], and the mutation in P104 is Gly21Val.

Comparison with the genetic organization of cpoA and upstream regions of the closely related species S. mitis B6 and S. oralis Uo5 of known genome sequence [17, 18] revealed an almost perfect conservation of cpoA including the −10 region in these species (Figure 1B).

The arrangement of genes and expression signals predicted in the downstream region of P cpoA suggested a polycistronic mRNA of approximately 4.4 kb covering the cpoA-spr0985 region. This was confirmed by RT-PCR experiments in which six overlapping products were obtained from this region, the largest of which extended from cpoA to spr0984 (Figure 1). Attempts to detect a contiguous transcript of the entire cpoA-spr0985 region, either by RT-PCR or by Northern blot analysis, however, were not successful, probably due to instability of the transcript.

The operon structure of the cpoA-spr0985 region and bioinformatic analyses indicated that the gene products might be functionally related and involved in membrane-associated functions. The GT-activities of CpoA and Spr0982 have been linked to glycolipid biosynthesis by in vitro experiments [9, 10], Spr0983 [58 amino acids 7(aa)] belongs to the PspC superfamily of putative stress-responsive transcriptional regulators, and Obg (436 aa) belongs to the Obg subfamily of GTP-binding proteins involved in stress response and processes related to cell division [for review, see [19]]. Possible functions of the two small peptides Spr0983.1 (44 aa) which has not been annotated in the R6 genome and Spr0985 (52 aa) [20] cannot be deduced from the amino acid sequences.

Mutational analysis of the cpoA operon

To assess the importance of these gene products, we aimed to construct deletions in each gene. A previous attempt to delete cpoA by insertion-duplication mutagenesis using a non-replicative plasmid vector had been unsuccessful [7]. This suggested that either cpoA is essential, or that insertion of the vector had affected the expression of the downstream gene spr0982 which has been listed among essential genes of S. pneumoniae[15]. To avoid such polar effects, a different deletion strategy was applied which was based on the construction of in-frame deletions using the Janus cassette (Figure 1). R6 mutants in which 108 central codons of cpoA (specifying the GT domain) were replaced with the Janus cassette were obtained with common efficiencies (0.2%), demonstrating that cpoA is a non-essential gene. Deletions in spr0983 and spr0985 were also obtained. However, the generation times of R6ΔcpoA and R6Δspr0985 (with 46–48 min) were significantly longer compared to R6 and R6Δspr0983 (38 min), suggesting that CpoA and Spr0985 are involved in important functions. In contrast, transformants carrying deletions in spr0982 and obg occurred only at 1,000- and respectively 10,000-fold reduced frequencies. This is in agreement with an essential function of the spr0982 product as reported previously [15], and strongly suggested that also obg is indispensable. The rare recovery of transformants carrying deletions in these genes probably was the result of co-selection of compensatory mutations at unknown secondary sites.

Mutants in cpoA are defective in synthesis of diglycosyl-DAG

To verify the CpoA function in vivo, the membrane lipids of cpoA mutant strains and the parent S. pneumoniae R6 were isolated and glycolipids specifically stained after separation by thin layer chromatograpy (Figure 2). S. pneumoniae contains the two glycolipids GlcDAG and GalGlcDAG. Two spots were detected in the R6 strain that could be assigned to the pneumococcal glycolipids according to the glycolipid standards: the major one representing a diglycosyl-DAG (most likely GalGlcDAG close to the position of the GalGalDAG standard), and a second spot at the position of monoglycosyl-DAG (Figure 2). This is in agreement with a ratio of GlcDAG to GalGlcDAG to be approximately 1:2.5 [11]. In contrast, the only glycolipid in all cpoA mutants corresponded to the position of the monoglycosyl-DAG (Figure 2). This confirms that CpoA is required for the synthesis of the diglycosyl-DAG in S. pneumoniae in agreement with the in vitro GalGlcDAG-synthase activity of CpoA, and documents that both mutants, P104 and P106, do not contain a functional CpoA.

Figure 2
figure 2

Glycolipids in Δ cpoA and piperacillin resistant laboratory mutants containing cpoA mutations. Lipids extracted from strain R6 and from cpoA mutants, P104, P106, and R6ΔcpoA as indicated above the lanes were separated by thin layer chromatography (chloroform/methanol/acetic acid = 80:15:8). GalGalDAG (S1) and GlcDAG (S2) were used as a standards. Spots were assigned to the two major glycolipids of S. pneumoniae diglycosyl DAG (GalGlcDAG) and monoglycosyl DAG (GlcDAG).

Full size image

Phospholipids in cpoA mutants

The glycolipid content affects physical properties of the cytoplasmic membrane. Since the exclusive production of the monolayer-forming glycolipid GlcDAG which forms non-bilayer structures strongly affects the membrane curvature [9, 13], we investigated whether this has some impact on the phospholipid content as well. S. pneumoniae contains the two phospholipids cardiolipin, a non-bilayer prone lipid, and phosphatidylglycerol. Lipids were separated by two-dimensional thin layer chromatography, and experiments were performed with at least two independently grown cultures. All cpoA mutants (R6ΔcpoA, P104 and P106) showed a significant increase in the ratio of phosphatidylglycerol: cardiolipin (Figure 3), suggesting that the cells are able to regulate the overall content of bilayer versus non-bilayer forming lipids. It should be noted that phosphadidylglycerol is more strongly stained compared to cardiolipin by the procedure used here (Additional file 1: Figure S1).

Figure 3
figure 3

Phospholipids in cpoA mutants. Lipids were extracted and separated by two dimensional TLC. 1.D and 2.D: first and second dimension (first dimension: CHCl3/MeOH/H20 = 65:25:4; second dimension: CHCl3/AcOH/MeOH/H20 = 80:14:10:3). Phospholipids were visualized by spraying with Molybdenum Blue spray reagent. PG: phosphatidylgylcerol; CL: cardiolipin. Spots were assigned according to the phosphatidylglycerol standard (see Additional file 1: Figure S1) and Fischer [42].

Full size image

Pleiotropic phenotype of cpoA mutants

The severe changes in membrane lipids in cpoA mutants is consistent with their pleiotropic phenotype described before [1, 7] which included a reduced generation time in liquid medium, decreased susceptibility to beta-lactams, defects in transformability, and a lower amount of PBP1a with less than 20% compared to the parental strain while the pbp1a transcript was unaffected; alterations in other PBPs were not detected. We first verified these properties for the R6ΔcpoA mutant: the MIC of piperacillin increased from 0.015 μg/ml (R6) to 0.045 μg/ml, the competence for genetic transformation was approximately 20-fold lower and shifted to the early exponential phase compared to R6, and the amount of PBP1a was decreased (not shown). These phenotypes are reminiscent of those displayed by P104/P106 but were more pronounced in R6ΔcpoA, probably a result of the rpsL allele.

Several other tests were then performed in order to see whether the altered glycolipid composition affects also cell envelope related properties in general. These included growth at low pH, the requirement for Mg2+, stationary phase autolysis and lysis induced by Triton X100. In all experiments, cpoA mutants showed a clear phenotype distinct from the R6 strain. Growth was severely affected at pH 6 (Figure 4). At pH 6, cpoA mutants showed an increased requirement for Mg2+ (Figure 5). The stationary phase lysis was slightly delayed in all cpoA mutants (Figure 4). Moreover, lysis induced by low concentrations of Triton X100 proceeded significantly more slowly in all cpoA mutants (Figure 6).

Figure 4
figure 4

Growth of cpoA mutants in low pH medium. Strains were grown in C-medium, and culture density was monitored by nephelometry [NU]. The growth was examined at pH 8 (circles) and pH 6 (squares). A: R6; B: P104; C: P106; D: R6ΔcpoA.

Full size image
Figure 5
figure 5

Mg 2+ requirement of cpoA mutations. Strains were grown in C-medium pH 6, and culture density was monitored by nephelometry [NU]. The medium contained either 0.195 mg/ml MgCl2 final concentration (filled circles) or 0.39 mg/ml MgCl2 (squares). A: R6; B: P104; C: P106; D: R6ΔcpoA.

Full size image
Figure 6
figure 6

Triton induced lysis. Cells were grown to OD600 in C-medium. At OD600 = 0.5, Triton (0.01% final concentration) was added. R6: filled circles; R6ΔcpoA: open circles; P106: open triangles; P104: open squares.

Full size image

Susceptibility to non-beta lactam cell wall antibiotics was also tested. Almost no effect was seen with vancomycin (MIC = 0.35 - 0.45 μg/ml) or cycloserine (MIC = 65–75 μg/ml), but the MIC value for bacitracin dropped from 7.5 μg/ml in the R6 strain to 0.75 - 1 μg/ml in all cpoA mutants.

Transcription profile of cpoA mutants

The pleiotropic effect of cpoA mutants on many membrane-associated functions was consistent with the relation of CpoA activity to glycolipid biosynthesis. In order to estimate the consequences of the altered glycolipid composition in cpoA mutants, their transcription pattern was determined in comparison to the R6 parent strain using an S. pneumoniae R6 specific oligonucleotide microarray [21]. Cells were grown under non-competent conditions at pH 6.8 in order to avoid the detection of the complex com regulon. Only four gene clusters and one single gene were affected in all three mutants. This included the approximately 3-fold downregulation of a PTS system (spr0276 - spr0282) and an ABC transporter (spr1545 - spr1549), and the 5-7-fold upregulation of two ABC transporters (vex, spr0524 - spr0526; spr1558 - spr1560) and spr0307 clpL (approximately 4-fold; Additional file 2: Table S3). No effect on PBP genes or genes involved in lipid biosynthesis was apparent.

Discussion

Glycolipids in cpoA mutants

The two piperacillin-resistant S. pneumoniae laboratory mutants P104 and P106, both containing point mutations affecting CpoA production, do not produce detectable amounts of GalGlcDAG, the main glycolipid of this organism. This clearly shows that the glycosysltransferase CpoA of S. pneumoniae is essential for the synthesis of the major glycolipid GalGlcDAG in vivo, and this could be confirmed by cpoA deletion mutants. The data are in agreement with previous in vitro studies using extracts of E. coli overproducing CpoA [9].

Apparently, the amino acid change in CpoAP104 Gly21Val also results in a non-functional protein. Since the mutated protein is still associated with the membrane when cell fractions were probed with anti-CpoA antiserum (Additional file 1: Figure S2), it is possible that the Gly21Val mutation affects protein folding, or its enzymatic function directly or indirectly. In this context it is interesting to note that a missense mutation in cpoA has been identified recently in laboratory mutants selected with cefotaxime [22]. The mutation D186Y [listed in the paper as D213Y due to wrong annoation of cpoA in the R6 genome [20]] is located within the conserved region of this type of glycosyltransferases, and it would be interesting to study the glycolipid content and phenotype in this mutant. So far, mutations in cpoA have not been detected in clinical isolates of S. pneumoniae. This might not be surprising since glycolipids are involved in critical cellular functions. On the other hand, the study of laboratory mutants resistant to beta-lactam antibiotics provides a valuable tool to unravel physiological processes related to cell envelope biosynthetic processes.

Glycolipids and membrane properties

Glycolipids are common in Gram-positive bacteria, and their distribution across the cytoplasmic membrane represents a critical parameter affecting bilayer curvature of the membrane and lipid surface charge densities, thus also membrane-associated functions [23]. Glycolipids in the cell wall-less mycoplasma Acholeplasma laidlawii are asymmetrically distributed and mainly external [24]. Clear asymmetry of lipids has also been documented for special membrane systems, such as the purple membrane of the archaebacterium Halobacterium halobium where glycolipids were found exclusively in the outer leaflet [25, 26], and for the outer membrane of Gram-negative bacteria [27]. It is likely that also in S. pneumoniae the two glycolipids are arranged asymmetrically in the membrane and probably predominantly located in the outer leaflet.

Besides glycolipids, membrane proteins can also contribute substantially to the morphology and curvature of membranes [28]. The two GTs of A. laidlawii, homologues of Spr0982 and CpoA, have recently been shown to induce membrane vesiculation upon overproduction in E. coli[29]. These enzymes are monotopic, i.e. anchored in the membrane cytoplasmic interface by hydrophobic and charge interactions in a SecYEG-independent manner [8, 9]. The data of Wikström et al.[29] strongly suggest that the GTs themselves are capable of inducing vesiculation, i.e. convex bending of the membrane. This implies some possible consequences when CpoA is absent, i.e. in P106 and in R6ΔcpoA, in that elimination of CpoA itself could affect the curvature of the membrane.

Phenotypes of cpoA mutants

Failure to synthesize GalGlcDAG, the bilayerforming di-glycosyl-glycolipid, must affect the physical properties of the cytoplasmic membrane considerably, consistent with the pleiotropic phenotype associated with cpoA mutants. Introduction of the cpoA point mutations present in P104 and P106 into the parental R6 strain conferred the same phenotypes, strongly suggesting that no other mutations besides cpoA are present in P104 and P106 (not shown). This included higher susceptibility to acidic stress and increased requirement for Mg2+ at low pH, as well as reduced lysis rate under lysis inducing conditions. Moreover, an altered proportion of the two pneumococcal phospholipids was observed in the cpoA mutants. Whereas cardiolipin is the major phospholipid in the parental R6 strain, all cpoA mutants contained a considerable higher amount of phosphatidylglycerol relative to cardiolipin as shown in Figure 3. Interestingly, mutations in the gene encoding the cardiolipin synthase have been identified in cefotaxime resistant laboratory mutants but have not been investigated further [22]. Since GlcDAG, the only glycolipid in cpoA mutants, is non-bilayer prone and cardiolipin as well, apparently the cells are capable to regulate the amounts of lipids to ensure sufficient bilayer structure of the cytoplasmic membrane. Cross-regulation in membrane lipid pathways has already been suggested in B. subtilis mutants defective in the cardiolipin synthase gene [30]. MIC values of vancomycin or cycloserine inhibiting late and early stages of peptidogylcan synthesis were not affected in cpoA mutants, an indication that the cell wall biochemistry is not affected.

Interestingly, cpoA mutants were ten-fold more susceptible to bacitracin, which targets the lipid molecule bactoprenol. The cpoA mutants expressed an altered transcription profile compared to that of the R6 strain, mainly by genes encoding membrane proteins such as PTS systems or ABC transporters that represent minor components of the bacterial cell. On the other hand, we could not detect significant changes of the protein profile of cytoplasmic or membrane proteins on SDS-polyacrylamide gels, i.e. no major protein components were affected in terms of quantity (not shown). It is conceivable that the transcriptional changes might be an indirect effect of the altered membrane composition. We recently reported that a higher susceptibility to bacitracin was also noted in S. pneumoniae containing a mutated ABC transporter [31]. It is possible that the altered lipid composition of the cpoA mutants indirectly affects the ABC transporter function and thus bacitracin MIC.

Glycolipids as anchor molecules in Gram-positive bacteria

Glycolipids represent the membrane anchor of important membrane-bound cell wall polymers in Gram-positive bacteria. They function as the lipid anchor for LTA and also for another class of membrane-associated cell wall glycopolymers, lipoglycans, which seem to replace LTA in the high GC division of Gram-positive bacteria [32, 33]. Listeria contain the same glycolipids as S. pneumoniae, whereas GlcDAG and GlcGlcDAG represent the major glycolipids in Bacillus, Staphylococcus and Enterococcus. However, these species differ in their biosynthetic enzymes. In Bacillus and Staphylococcus, both glycolipids are synthesized by one single GT YpfP [34–36], whereas two putative GTs are involved in glycolipid biosynthesis in Listeria, Streptococcus and Enterococcus[9, 10, 37, 38]. In this context it is remarkable that the structure of the cpoA operon which includes obg and several putative small peptide encoding genes is only maintained within Streptococcus spp., and that other Gram-positive bacteria contain cpoA (plus spr0982 in case of Listeria and Enterococcus) and obg homologues at distinct positions in the genome. The reason for this is not clear. Several studies revealed that Obg proteins play a role in many important processes, including DNA replication, chromosome segregation, and regulation of stress responses, but their actual function remains unknown [for review, see [19]].

Most of the species mentioned above contain a polyglycerophosphate LTA backbone which is anchored to the di-glycosyl-DAG lipid. Thus, interference of the biosynthesis of this glycolipid severely affects LTA and accordingly cell wall integrity as was shown for mutants in the S. aureus GT YpfP [34, 35], the (1 → 2)GTs LafA in Listeria, IagA in group B streptococci, and E. faecalis BgsA [37–39]. Deletion mutants of S. aureus ypfP produced LTA which was probably attached directly to DAG [34, 35]. In the GC-rich organism M. luteus, dimannosyl-DAG is the lipid anchor of the essential lipomannan cell wall polymer [40]. Therefore, temperature sensitive mutants defective in lipomannan assembly were isolated of M. luteus, and one of them (mms1) contained a reduced amount of dimannosyl-DAG whereas the amount of monomannosyl-DAG was increased [41]. The corresponding M. luteus gene encoding a putative GT is unknown; according to BLAST analysis, the GT encoded by mlut_06690 is a likely CpoA homologue.

In contrast to these organisms, the LTA of S. pneumoniae is unique in that it includes choline and unusual sugar moieties in its repeating unit which is identical to that of the wall teichoic acid (WTA) [42]. Genetic evidence suggests strongly that the closely related species S. oralis and S. mitis contain similar TA molecules [43]. Moreover, special choline-binding proteins are associated with the TA molecules, some of which are involved in crucial functions including cell separation [for review, see [44]], probably one of the reasons why LTA and its biosynthetic enzymes are essential in S. pneumoniae.

Early studies predicted the LTA lipid anchor to be Glc(β1 → 3)AATGal(β1 → 3)Glc(α1 → 3)DAG where AATGal is 2-acetamino-4-amino-2,4,6-trideoxy-D-galactose [42], but recent data provide evidence that GlcDAG is the more likely anchor molecule [14], i.e. the product of the reaction catalyzed by the GT Spr0982 [10]. Failure to isolate deletions mutants in spr0982 are in agreement with the essential nature of the S. pneumoniae LTA. No effect on choline incorporation into the cell wall was noted in the piperacillin resistant mutants [1], suggesting that teichoic acids seem to be present in similar amounts in mutant cells compared to R6 and that its biosynthesis is not affected by cpoA mutations. The estimated number of molecules for LTA and GlcDAG is in the same range of magnitude. LTA constitutes up to 20% of the lipid molecules in the outer leaflet of the cytoplasmic membrane in S. pneumoniae[32], and glycolipids represent 34% of the lipids in S. pneumoniae[12] with almost one third being GlcDAG [11].

Conclusions

Here we have shown that CpoA acts as the glycosyltransferase in vivo responsible for the biosynthesis of the major glycolipid GalGlcDAG in S. pneumoniae. The altered lipid composition of cpoA mutants - GlcDAG as the only glycolipid, and a higher proportion of phosphatidylglycerol relative to cardiolipin - affects many membrane related functions and thus results in a pleiotropic phenotype. The question remains why the selection of piperacillin-resistant laboratory mutants P104 and P106 resulted in the isolation of cpoA mutations. Since cpoA was not affected in another six mutant families selected with cefotaxime, a beta-lactam that induces a tolerant response [2], the cpoA mutations are probably related to the highly lytic action of piperacillin. Changes of the physical properties of the membrane by alteration of the lipid composition might be an effective measure to counteract the lytic response induced by beta-lactams and other agents as well.

Methods

Bacterial strains, plasmids, oligonucleotides, growth conditions, and transformation

Streptococcus strains and plasmids used in this work are listed in Table 1. PCR primers were synthesized at Operon Biotechnologies and are listed in Additional file 2: Table S1. Primers used for sequencing and confirming the correct integration of DNA sections delivered to the S. pneumoniae genome and nested primers are not listed. S. pneumoniae was grown in C-medium [45] supplemented with 0.2% yeast extract or in Todd Hewitt Broth [THB] (Becton and Dickinson) at 37°C without aeration. For growth on solid surface, D-agar [46] supplemented with 3% defibrinated sheep blood (Oxoid) was used. Growth of S. pneumoniae in liquid cultures was monitored by nephelometry (nephelo units [NU]), and doubling time (generation time) estimated from at least three independent experiments. To determine minimal inhibitory concentractions (MICs) of piperacillin, cultures of S. pneumoniae, grown in C-medium to a density of 30 NU, were diluted 1000-fold in 0.9% NaCl, and aliquots (30 μl) of the dilutions were spotted on D-agar plates containing piperacillin at concentrations of 0.01 to 0.3 μg/ml using 0.005 μg/ml intervals. MIC values for bacitracin, vancomycin and cycloserine were also determined on D-agar plates using appropriate dilutions of the antibiotic. Antibiotic resistance genes used for chromosomal integrations in S. pneumoniae were selected with 2 μg/ml erythromycin (Erm, ermAB), 200 μg/ml kanamycin (Kan, aphIII), 200 μg/ml streptomycin (Str, rpsL), and 3 μg/ml tetracyclin (Tet, tetM), respectively. Transformation of S. pneumoniae was performed using naturally competent cells as described previously [47]. Transformation efficiency was calculated as the percentage of colonies obtained on the selective medium compared to the colony number on control plates without antibiotic.

Table 1 S. pneumoniae strains and plasmids
Full size table

DNA manipulations

Isolation of plasmid DNA and routine DNA manipulations were carried out by standard methods [48]. PCR products and DNA recovered after restriction endonuclease digestions were purified using the JETquick spin column technique kit. Restriction enzymes and T4 DNA ligase were purchased from Roche Applied Science or New England Biolabs and used according to the manufacturer’s instructions. PCRs were performed using either Goldstar Red Taq polymerase (Eurogentec) or iProof High-Fidelity DNA polymerase (Bio-Rad) according to the manufacturer’s instructions. Nucleotide sequencing was performed using the ABI Prism BigDye Terminator Ready Reaction cycle sequencing kit, version 3.1 (Perkin Elmer-ABI). Nucleotide sequences were analyzed by using the CloneManager and Phred/Phrap/Consed software.

Identification of transcription start site

The start point of cpoA transcription was determined by rapid amplification of cDNA ends (5' RACE) as described previously [49] using RNA of S. pneumoniae R6 isolated at a culture density of 40 NU. The primer cpoARACE2 was used for reverse transcription of RNA ligated to the RNA adapter, and the nested primer and cpoARACE1 was used for amplification of cDNA (for primers, see Additional file 2: Table S1 and S2).

Construction of delivery cassettes, plasmids and mutants

To identify the initiation site of cpoA translation, fusions of two DNA fragments with the lacZ reporter gene were constructed. They contained P cpoA (i) together either with two potential start codons (ATG1 and ATG2 in Figure 1B), (ii) with a mutation in ATG2 (ATA), or (iii) with ATG1 only. The three fragments were amplified from chromosomal DNA of S. pneumoniae R6 by using the primer pairs PcpoA_Eco_f/PcpoA_r2, PcpoA_Eco_f/PcpoABam_r1a and PcpoA_Eco_f/PcpoABam_r1b, cleaved with Eco RI and Bam HI, and ligated with the Eco RI/Bam HI-digested translation probe vector pTP2. The desired plasmids, pTP2PcpoA-ATG21, pTP2PcpoA-ATG1a and pTP2PcpoA-ATG1b were isolated after transformation of E. coli DH5α and subsequently used to transform S. pneumoniae R6; alternatively plasmids were directly transformed into S. pneumoniae R6. DNA from TetR transformants was PCR-amplified and sequenced to confirm the presence of the lacZ fusions in the resulting strains R6-PcpoA-ATG21, R6-PcpoA-ATG1a and R6-PcpoA-ATG1b.

In-frame deletions in cpoA, spr0982, spr0983, obg, or spr0985 were constructed via a two-step process in which the central part of the respective gene(s) was first replaced with the Janus cassette [50] that confers a KanR StrS phenotype in a StrR background. In the second step, the Janus cassette was deleted, thus restoring the original StrR phenotype. The constituents of ‘replacement fragments’ and ‘deletion fragments’ used in the first and second steps of each deletion were amplified from chromosomal DNA of S. pneumoniae R6 by using the primer pairs listed in Additional file 2: Table S2. To generate a ‘replacement fragment’, two PCR products of 0.7 to 1 kb (‘upstream’ and ‘downstream fragment’) flanking the desired deletion were joined with the two ends of the Janus cassette either by the use of appropriate restriction sites added to the ends of the respective primers or by overlap extension PCR with nested primers. The ‘replacement fragment’ was used to transform a StrR derivative of S. pneumoniae R6 (R6s) obtained by transformation of R6 with chromosomal DNA carrying the AmiA9 resistance marker [51]. In the resulting KanR StrS transformants, the correct position of the Janus cassette was confirmed by DNA extraction and PCR with appropriate primers. To generate a ‘deletion fragment’ (containing the desired deletion), the respective ‘upstream’ and ‘downstream fragments’ were directly joined with each other either by the use of appropriate restriction sites added to the primers or by overlap extension PCR with nested primers. The ‘deletion fragment’ was used to transform a derivative of R6s carrying the Janus cassette at the site of the desired deletion. DNA from transformants displaying a KanS StrR phenotype was PCR-amplified and sequenced to confirm the presence of the deletion in the resulting mutant.

Determination of β-galactosidase activity

Preparation of cell extracts from cultures of S. pneumoniae, grown to a density of OD600 = 0.8 in C-medium, and determination of specific β-galactosidase activities were performed as described [52].

Lipid extraction and analysis

Lipids were extracted from S. pneumoniae essentially as described [53]. Briefly, cells harvested by centrifugation of liquid cultures grown to a density of about 70 NU were resuspended in 0.8 ml H2O per gram wet weight and subsequently mixed with 3 ml of chloroform/methanol (1:2) per gram wet weight. After gentle agitation for 2 h at 4°C, chloroform (1 volume) and H2O (1 volume) were added and mixed. The samples were centrifuged at 4,000 × g and 4°C for 5 min, the organic phases were recovered, mixed with 1 volume of H2O equilibrated with chloroform/methanol (1:2), and centrifuged as before. Recovered organic phases were completely evaporated, and the remainders were dissolved in 50 to 100 μl of chloroform/methanol (80:15). Glycolipids were separated by one-dimensional thin layer chromatography in chloroform/methanol/acetic acid (80:15:8) on silica gel G plates (0.025 mm; Merck). For visualization the plates were sprayed with 1-naphthol (3.2% w/v in methanol/H2SO4/H2O = 25:3:2) and heated at 110°C for 10 min. GalGalDAG (Sigma) and GlcDAG were used as standards. Phospholipids were separated on two-dimensional thin layer chromatography (first dimension: CHCl3/MeOH/H2O = 65:25:4; second dimension: CHCl3/AcOH/MeOH/H2O = 80:14:10:3) and stained with 1.3% molydbenum oxide in 4.2 M sulfuric acid (Molybdenum Blue spray reagent, Sigma-Aldrich). Spots were assigned according to the reference lipid phosphatidylglycerol (Sigma) and the pattern described elsewhere for phospholipids [42].

Immunological detection of CpoA

S. pneumoniae cells were grown to mid-exponential growth phase (80 NU), harvested by centrifugation (9,000 rpm, 15 min, 4°C, Beckman centrifuge J2-21), and washed once with 20 mM sodium phosphate buffer pH 7.2. Pellets were resuspended in 180 μl sodium phosphate buffer and mixed with 500 mg glass beads per 100 mg wet weight, followed by disruption in a cell mill (Vitrogen-Zellmühle Typ VI-4, Edmund Bühler GmbH) for 20 min. All further steps were carried out on ice. Glass beads were removed by centrifugation for 6 min (14,000 rpm, 4°C, Hermle Z513K centrifuge). Membranes were separated from cytoplasmic proteins by ultracentrifugation (Beckman centrifuge, TLA 100.4 rotor) for 2 h at 60,000 rpm and 4°C. Pellets were resuspended in half of the volume of the supernatant, and fractions stored at −80°C. For SDS polyacrylamide gel electrophoresis, 3 μl per fraction were used. Western blotting was performed as described previously [54] and CpoA was visualized using a 1:10,000 dilution of rabbit antiserum raised against a purified CpoA-derivative as described [7].

Microarray-based transcriptome analysis

Extraction of total RNA from exponentially growing S. pneumoniae cultures (40 NU), reverse transcription of RNA into labeled cDNA, prehybridization, hybridization, slide washing, scanning, and analysis of the data were performed as described previously [55]. For each strain, data sets from at least four hybridizations were used for normalization and statistical analysis. Only data which showed P values below 10-4 in a paired t test, and relative changes in the transcript amount of greater than threefold were considered further. The oligonucleotide microarray covering genes and intergenic regions of S. pneumoniae R6/TIGR4 has been described [21].

Accession number

S. pneumoniae R6/TIGR4 oligonucleotide microarray: ArrayDesign R6/TIGR4 ArrayExpres accession number A-MEXP-1846.

Availability of supporting data

The data sets supporting the results of this article are included within the article and its additional files.