Lipoteichoic Acids Are Essential for Pneumococcal Colonization and Membrane Integrity

The human pathobiont Streptococcus pneumoniae belongs to the common microbiota of the human respiratory tract. However, S. pneumoniae (the pneumococcus) is also the aetiologic agent and leading cause of death from lower respiratory tract infections. Subpopulations of high risk are in particular the elderly as well as young children without vaccination [1]. The cell wall of S. pneumoniae is a complex construct of different polysaccharides. The outermost layer is the capsular polysaccharide (CPS), which surrounds the bacterium, thereby masking surface proteins. CPS protects the pneumococcus against external influences, avoids detection by the immune system, and represents a crucial virulence factor, e.g., by preventing phagocytosis [2, 3]. Despite the large number of more than 100 different serotypes, CPS is the primary target of most common vaccines [4‒8]. Other important macromolecules of the cell wall are teichoic acids (TAs), which offer a binding site for the family of choline-binding proteins (CBPs) through their characteristic phophorylcholine (P-Cho) decorations. CBPs are of high relevance for physiological remodelling of the cell wall (LytA, LytB, LytC), adhesion to host cells (PspC), and immune evasion (PspA) [9‒11]. In contrast to other Gram-positive bacteria, pneumococcal wall teichoic acids (WTAs) and lipoteichoic acids (LTAs) have identical repeating units and share an identical intracellular biosynthetic pathway [12‒14]. However, WTAs are anchored to the peptidoglycan (PGN), whereas LTAs are anchored in the cell membrane [12, 15]. The final ligation step to form LTA, in which the TA precursor chains are transferred to the glycosylglycerolipid α-d-glucopyranosyl-(1,3)-diacylglycerol serving as anchor of the LTA in the cell membrane, is mediated by the TA ligase TacL [14]. The cell membrane of Gram-positive bacteria consists of different molecules belonging to the families of glycerolipids and glycerophospholipids. The glycerolipids identified in S. pneumoniae are diacylglycerol, α-d-glucopyranosyl-(1,3)-diacylglycerol, and α-d-galactopyranosyl-(1,2)-α-d-glucopyranosyl-(1,3)-diacylglycerol [16, 17]. Negatively charged phosphatidylglycerol and cardiolipin are the major glycerophospholipids present in the cell membrane of S. pneumoniae [18, 19]. Due to the emergence of resistance to antibiotics, antimicrobial peptides (AMPs) targeting the cell membrane of bacteria are becoming highly attractive anti-infectives. Cationic AMPs can bind to anionic components of the cell membrane and lead to cell death through various pore formations [20, 21]. However, this charge-induced binding mechanism also allows other anionic cell wall components such as LTA to absorb AMPs and protect the cell membrane from their destructive influence [22].

Other essential components of the bacterial cell membrane are lipoproteins, which are anchored via a diacylthioglycerol anchor in the cell membrane. Preprolipoproteins synthesized in the cytoplasma are retranslocated to the outer surface of the cytoplasmic membrane. They possess a consensus motif termed the lipobox, which is part of the N-terminal signal peptide cleavage site. The +1 cysteine residue of the lipobox undergoes lipid modification in a sequential process catalysed by three periplasmic enzymes to yield the mature lipoproteins. First, the preprolipoprotein diacylglyceryl transferase (Lgt) transfers a diacylglyerol moiety from phosphatidylglycerols to the sulfohydryl group of the +1 cysteine, thereby generating a thioester linkage [23]. Subsequently, the signal peptide is removed by the lipoprotein-specific signal peptidase (Lsp) [24‒26], leading to the intercalation of the diacylthioglycerol anchor into the cell membrane. For producing a mature triacylated lipoprotein, the lipoprotein N’N-acyl transferase (Lnt) catalyses aminoacylation of the α-amino group of the thiodiacylglyceryl cysteine [23]. The lipoprotein family is largely classified as substrate-binding proteins, which serve to sequester sugars, amino acids, small peptides, and essential metal ions to the cell via ATP-binding cassette transporter systems. Other members serve for resistance against antimicrobial substances, oxidative stress, maintenance of cell shape and osmotic stability, as well as protein folding or activation of various secreted cell surface proteins [9, 25, 27, 28]. Lipoproteins are involved in various stages during infection including, e.g., host cell adhesion and colonization, immune evasion, or survival in conquered host compartments [25, 29‒34]. Furthermore, lipoproteins activate the innate immune system via Toll-like receptor 2 (TLR2) and represent due to their conservation and immunogenicity promising candidates for a proteinaceous serotype-independent pneumococcal vaccine [35‒37].

Recently, we demonstrated that the lack of LTA in pneumococci had no impact on the morphology or growth even in minimal culture media, while being attenuated in pneumonia and sepsis mouse infection models [14]. Our data were corroborated when the CRISPR interference system was applied to identify pneumococcal genes pivotal under invasive in vivo conditions [38]. Due to discrepancy between in vitro and in vivo fitness of pneumococci lacking LTA, we speculated that already colonization might be affected and that despite invisible morphological changes LTA is essential for the integrity of the bacterial membrane. In the present study, we therefore assessed the impact of pneumococcal LTA on in vivo colonization using the non-invasive 19F serotype strain EF3030 [39]. We further tested whether also the tacL mutant of this strain is attenuated under invasive conditions and used the larvae of the greater wax moth, Galleria mellonella, as infection model. To understand the attenuation of pneumococci lacking LTA due to genetic ablation of TacL in more detail, we assessed in comparative studies alterations in the membrane fluidity, the abundance of lipoproteins, and susceptibility against AMPs by using our mutants and complemented strains generated in different pneumococcal serotypes.

All strains used in this work are listed in Table 1. Escherichia coli strains were grown in liquid lysogeny broth medium (LB; Roth, Germany) or on LB agar plates (Roth, Germany) supplemented with erythromycin (2.5 μg/mL) or chloramphenicol (8 μg/mL). LB agar plates were incubated at 37°C and cultivation in liquid LB was performed at 37°C under agitation (120 rpm). S. pneumoniae serotype 2 strain D39 (NCTC7466), serotype 4 strain TIGR4, and serotype 19F strain EF3030, their isogenic mutants, and complemented strains were grown on Columbia blood agar plates (Thermo Scientific, Germany) supplemented with kanamycin (125 μg/mL), erythromycin (2.5 μg/mL), or chloramphenicol (8 μg/mL) if required. Liquid pneumococcal cultures were grown in Todd-Hewitt broth (THY; Roth, Germany) supplemented with 0.5% yeast extract and antibiotics if required. Blood agar plates were incubated at 37°C and 5% CO₂ for 8–10 h. Cultivation in liquid cultures was performed at 37°C in a water bath without agitation.

The construction of the pneumococcal tacL-mutants in serotype 19F strain EF3030 was done using the recombinant plasmid pUC18∆tacL, which was constructed and described previously to generate tacL-mutants in D39/D39∆cps and TIGR4/TIGR4∆cps [14]. Transformation of EF3030 was performed as described previously [40]. EF3030∆tacL was complemented in trans using pBAV-tacL constructed as described [14]. Both, the mutant EF3030∆tacL and complemented strain EF3030∆tacL∇tacL were verified by PCR, sequencing (Eurofins Genomics, Germany), and qRT-PCR (online suppl. Fig. S1; for all online suppl. material, see https://doi.org/10.1159/000539934).

To analyse pneumococcal resistance to oxidative stress conditions, a hydrogen peroxide (H₂O₂)-stress assay was performed. Pneumococcal strains were grown until mid-exponential phase (A₆₀₀ = 0.35–0.45) in THY medium. Each culture was split into 10 mL cultures and incubated with 0 mm, 5 mm, or 10 mm H₂O₂ for 30 min at 37°C. Afterwards, cultures were diluted in PBS (pH 7.4) and colony-forming units (CFUs) were determined by serial plating on blood agar plates (Thermo Scientific, Germany).

To determine the susceptibility of pneumococcal wild-types and mutants to AMPs, LL-37 (InvivoGen, Germany) and CPL-1 (the endolysin was heterologously expressed in E. coli as described before by Hermoso et al. in [41]) were used. Pneumococcal strains were grown until mid-log phase (A₆₀₀ = 0.35–0.45) in THY medium. Cultures were than split into 200 μL aliquots and incubated with 0 μm, 1 μm, or 2 μm LL-37 for 3 h at 37°C. Each sample was diluted in PBS (pH 7.4 for LL-37; pH 6.0 for CPL-1) and CFUs were determined by serial plating of relevant dilutions on Columbia blood agar plates (Thermo Scientific, Germany).

For treatment with CPL-1, pneumococcal strains were grown as described above and 0 μg/mL, 2 μg/mL (2.6 μm), or 50 μg/mL (65 μm) of the endolysin were added to the bacterial suspension. Pneumococcal lysis was determined by measuring absorption at 600 nm every 30 s.

To visualize membrane regions of increased or decreased fluidity, the fluorescent membrane-integrating dye DiIC₁₂ (1,1′-didodecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate; EnzoLife Sciences, Germany) was used. Pneumococcal strains were cultivated in THY medium containing 1 μg/mL DiIC₁₂ (dissolved in DMSO) until mid-logarithmic phase (A₆₀₀ = 0.35–0.45). Cultures were harvested at 3,275 × g and 30°C for 6 min and washed twice with pre-warmed THY containing 1% DMSO followed by washing with pre-warmed PBS (pH 7.4) containing 1% DMSO. A culture of each wild-type strain without dye was used as control. FACS Aria™ III (BD Biosciences, San Jose, CA, USA) was used for flow cytometric quantification of fluorescence [42]. Data analysis was carried out using FCS Express software (De Novo Software, Pasadena, CA, USA).

To analyse the abundance of pneumococcal lipoproteins on the surface of pneumococcal strains used in this study, polyclonal antibodies were generated in mice using routine immunization protocol as described previously [14]. Immunization of CD-1 mice was done by injecting 20 μg heterologously expressed non-lipidated proteins and Freund`s incomplete adjuvant (Sigma-Aldrich, Germany) (50:50 v/v) intraperitoneally. Mice were boosted with 20 μg protein and Freund`s incomplete adjuvants (Sigma-Aldrich, Germany) (50:50 v/v) after 14 and 28 days. After 48 days, mice were sacrificed and polyclonal IgGs were purified from serum using Protein A-Sepharose (Sigma-Aldrich, Germany).

S. pneumoniae D39∆cps, TIGR4∆cps and EF3030 wild-type strains, isogenic tacL-mutants, and complemented strains were cultivated in THY medium until mid-logarithmic phase (A₆₀₀ = 0.35–0.45). Bacteria were harvested at 3,275 × g for 6 min and washed 3 times with PBS (pH 7.4). Subsequently, strains were adjusted to the same A₆₀₀ and 100 μL bacterial suspension was incubated with specific primary antibodies against various lipoproteins (1:500/1:250 in PBS) for 30–45 min at 37°C and 5% CO₂ in 96-well U-bottom plates (Greiner Bio-One, Germany). Bacteria were washed 3 times and incubated with an Alexa-Fluor™ 488-labelled secondary goat-anti-mouse IgG antibody (Abcam, UK; 1:500 in PBS) for 30–45 min at 37°C and 5% CO₂. Again, the samples were washed 3 times with PBS, and bacteria were then fixed with 1% para-formaldehyde (Roth, Germany) in PBS (pH 7.4) at 4°C overnight. Samples were subjected to flow cytometric analysis using a FACS Calibur™ (BD Biosciences, San Jose, CA, USA). The geometric mean fluorescence intensity was determined and illustrated as geometric mean fluorescence intensity multiplied with the percent of gated events. Data analysis was carried out using FCS Express software (De Novo Software, Pasadena, CA, USA).

Human nasopharyngeal epithelial cells (Detroit-562) were purchased from ATCC (ATCC CCL-138) and have been tested to be mycoplasma negative by PCR and scanning electron microscopy. Detroit-562 cells were used to analyse bacterial adherence to epithelial cells and for that purpose cultured in RPMI1640 medium (Capricorn Scientific, Germany) supplemented with 2% heat-inactivated fetal bovine serum (FBS; Sigma-Aldrich) and 1% sodium pyruvate (Cytiva, Germany) in 24-well plates (Greiner Bio-One, Germany). Two days prior to the infection with pneumococci, 1 × 10⁵ cells were seeded per well. On the day of infection, 2 × 10⁵ cells were washed with infection medium (RPMI1640 supplemented with 1% heat-inactivated FBS) and afterwards infected with 1 × 10⁷ mid-exponentially grown pneumococci (multiplicity of infection of 50 bacteria per host cell) at 37°C and 5% CO₂ in infection medium. After 4 h, non-adherent bacteria were removed by washing three times with PBS (pH 7.4). Subsequently, epithelial cells were treated with PBS  (pH 7.4) containing 5% saponin (Sigma-Aldrich, Germany) for 10 min at 37°C. The obtained suspension was diluted in PBS (pH 7.4) and total CFUs of adherent and invasive pneumococci were enumerated by serial plating on blood agar plates (Thermo Scientific, Germany).

Pneumococcal strains were cultured in THY medium until mid-exponential phase (A₆₀₀ = 0.35–0.45) and washed with 0.9% sodium chloride. Subsequently, larvae of the greater wax moth, Galleria mellonella (Reptilienkosmos, Germany), with a weight of 0.3–0.4 g, were infected with 10 μL bacterial suspension (containing 1.0–5.0 × 10⁵ bacteria) via an intrahemocelic route. A gastight microliter syringe (Hamilton, OH, USA) coupled with a repeating dispenser (Hamilton, OH, USA) was used to ensure equal infection doses. Each group of 10 larvae was stored at 37°C with sufficient food for 7 days post-infection and monitored daily. To determine the CFU, 10 larvae per group were anaesthetized with isoflurane (CP Pharma, Germany) and mechanically homogenized in 700 μL 0.9% sodium chloride using a T10 basic ULTRA-TURRAX (IKA, Germany). The bacterial suspension was serially diluted in 0.9% sodium chloride and plated on blood agar plates, which were incubated for 15 h at 37°C and 5% CO₂. Food contained wheat bran, oatmeal, dry yeast, skim milk powder, honey, and glycerol in variable proportions.

For the in vivo murine colonization studies, 8- to 10-week-old female CD-1 outbred mice (Janvier, France) were intranasally infected with serotype 19F strain EF3030, its isogenic tacL-mutant, and the complemented mutant. Strains were cultivated to mid-exponential phase (A₆₀₀ = 0.35–0.45) in THY medium containing 10% heat-inactivated FBS. The infection dose of 20 μL was adjusted to 1.0–2.0 × 10⁷ CFU in PBS  (pH 7.4) containing 1% heat-inactivated FBS. The CFU of the applied infection dose was verified by plating serial dilutions of the inoculum on blood agar plates. Mice were anaesthetized intraperitoneally with ketamine (Ketamin 10%; Selectavet, Germany) and xylazine (Xylasel; Selectavet, Germany) and intranasally infected with pneumococci as described previously [40]. At day 1, 3, 7, and 14 post-infection, nasopharyngeal and bronchoalveolar lavages were collected and the CFU was determined by plating serial dilutions on blood agar plates.

Statistical significance between different groups was calculated using a one-way ANOVA (Kruskal-Wallis test) followed by Dunnett’s post-test for the mouse colonization model. Unpaired two-tailed Student’s t test (Mann-Whitney test) was performed to analyse the difference between two groups. Kaplan-Meier survival curves of mice were compared by the log-rank (Mantel-Cox) test. A p value of <0.05 was considered statistically significant. All statistical analyses were performed using GraphPad Prism version 5.0 (GraphPad Software, La Jolla, CA, USA). Statistically significant differences are indicated by ^*p < 0.05, ^**p < 0.01, ^***p < 0.001, and ^****p < 0.0001.

The Galleria mellonella infection model has become increasingly important in recent years to investigate host-pathogen interactions under infection-related conditions. The larvae of the large wax moth G. mellonella, like other insects, do not possess an adaptive immune system [43]. However, they are producing a variety of AMPs and opsonins, which constitute a major part of the humoral host defence mechanisms. In this regard, the opsonin hemolin has been shown to bind lipopolysaccharides of Gram-negative and LTA of Gram-positive bacteria [44, 45]. As we have previously reported, the deficiency of TacL and consequently absence of LTA in the genetic background of the invasive serotype 2 strain D39 (D39ΔtacL) abrogates pneumonia and septicemia in murine infection models but does not affect in vitro growth [14]. Here, we infected G. mellonella larvae with a non-invasive pneumococcal strain, namely, serotype 19F strain EF3030 [46]. This strain colonizes the upper respiratory tract of mice after intranasal application for more than 14 days without inducing significant inflammatory responses or clinical symptoms [40, 47]. First, we aimed to elucidate whether 19F (EF3030) is harmful to G. mellonella and how the lack of LTA changes the virulence potential as well as the susceptibility to the humoral G. mellonella host defence mechanisms. Larvae were therefore infected with wild-type EF3030, the tacL-mutant EF3030ΔtacL and its complemented strain EF3030ΔtacL∇tacL. We observed that larvae infected with serotype 19F lacking LTA survived up to 7 days, which was significantly longer compared to larvae infected with the isogenic wild-type strain showing a survival for only 1 day. Similar to the wild-type, the complemented mutant EF3030ΔtacL∇tacL was severely harmful and killed larvae 1 day post-infection (Fig. 1). Interestingly, a difference in the amount of bacteria detected in the larvae could only be observed after 18 h. Animals infected with the mutant strain EF3030ΔtacL had a significantly lower number of bacteria compared to larvae infected with the wild-type EF3030 (Fig. 1d). These in vivo infection experiments indicate that serotype 19F is highly infectious in the G. mellonella model and that the loss of LTA results in a significantly higher susceptibility against the humoral response of the larvae.

Previously, we showed that S. pneumoniae strain D39ΔtacL lacking LTA had a lower ability to adhere to lung epithelial cells A549 [14]. Here, we were interested in the impact of LTA on pneumococcal colonization. However, strain D39 is unable to stably colonize the nasal tissue of mice. Because of its capability to colonize the nasopharynx and to be non-invasive [39], we therefore used serotype 19F strain EF3030 in the following in vitro adherence and in vivo colonization experiments. Prior to the mouse colonization experiments, we assessed the effect of LTA on adherence of serotype 19F strains to human nasopharyngeal epithelial cells (Detroit-562). The results showed a reduction of EF3030ΔtacL adherence to Detroit-562 epithelial cells, which is statistically significant. However, the overall impact on the pathophysiology is probably not high. The complemented mutant EF3030ΔtacL∇tacL showed no significant differences compared to the wild-type strain (Fig. 2a). Nevertheless, these data corroborate results we had reported previously for the isogenic tacL-mutant of the non-encapsulated strain D39Δcps when adherence was tested to A549 lung epithelial cells [14]. Altogether, our data verify the role of LTA for pneumococcal attachment to epithelial cells.

To confirm our in vitro findings in vivo, we applied a mouse colonization model. Groups of CD-1 outbred mice were intranasally infected with wild-type serotype 19F strain EF3030, its isogenic tacL-mutant, and the in trans complemented mutant. The wild-type and complemented mutant EF3030∆tacL∇tacL showed a stable colonization of the nasopharynx for 14 days (Fig. 2b). In contrast, mice infected with the mutant strain lacking LTA had lower numbers of bacteria in the nasopharynx. This difference was most prominent at day 14 post-infection, while unexpectedly no differences were shown at day 7 post-infection. Similar to our earlier reported data [47], 19F bacteria, irrespectively whether wild-type or tacL-mutant, were isolated only at low numbers from the bronchoalveolar lavage and if at all only at earlier time points post-infection. Strikingly, mice infected with the complemented mutant EF3030∆tacL∇ tacL showed a relatively stable number of bacteria in the bronchoalveolar lavage up to 7 days post-infection (Fig. 2b). In the in trans complemented mutant, the tacL gene is located on a plasmid and expressed from a strong promotor [14]. Taken together, our results clearly show the importance of LTA for pneumococcal adherence and colonization under in vivo conditions.

Previously, we reported that the abundance of CBPs on the pneumococcal surface has not changed substantially in the absence of LTA. The absence of any impact on CBPs is probably due to the 9:1 ratio of WTA/LTA, meaning that 90% of the phosphorylcholine-binding sites are still available for CBPs in pneumococci lacking LTA due to genetic ablation of tacL [14, 48]. Because LTA is intercalated in the cell membrane via a glycosyldiacylglycerol anchor, we hypothesized that the absence of LTA might change the membrane properties and influence the membrane composition, although the glycosylglycerolipids themselves are still present. We therefore investigated at first the surface abundance of pneumococcal prototype lipoproteins, which are intercalated in the cell membrane by their diacylthioglycerol anchor [23]. The results of our flow cytometric analysis showed for serotype 2 strain D39∆cps and serotype 19F strain EF3030 a significantly higher abundance of surface-exposed lipoproteins PsaA, SlrA, and DacB in the tacL-mutants compared to the isogenic wild-type strains (Fig. 3a). In contrast, the complemented mutants showed amounts of lipoproteins comparable to the corresponding wild-type strain. In a complementary approach, we also observed changes in lipoprotein levels for PsaA, SlrA, and DacB, when whole bacterial cell lysates of strain D39∆cps and its isogenic mutant as well as the complemented strain were used for quantitative Western blot analysis (online suppl. Fig. S2). Remarkably, the effect of a higher lipoprotein abundance in the tacL-mutant is less pronounced for TIGR4∆cps∆tacL.

These results intrigued us to investigate the pneumococcal membrane fluidity using the fluorescence dye DiIC₁₂. This dye shows a high affinity to membrane parts having an increased membrane fluidity [49]. After incubation of our set of pneumococcal strains with DiIC₁₂, we quantified the bacterial fluorescence intensity by flow cytometry. We observed a significant lower amount of DiIC₁₂ dye incorporated into the bacterial cell membrane of tacL-mutants when compared to isogenic wild-type strains D39∆cps serotype 2 and EF3030 serotype 19F. These data suggest that the membrane fluidity in pneumococci lacking LTA is decreased. Importantly, the complemented strains of the above-mentioned serotypes showed significantly higher amount of DiIC₁₂ dye incorporated into the cell membrane when compared to the wild-type. This effect is, as mentioned, most likely due to the higher TacL expression in the in trans complemented strains, thereby showing the gene dosage effect (Fig. 3b, c). Serotype 4 strain TIGR4 showed, similar to our findings for the lipoprotein abundancies, neither lower DiIC₁₂ levels in the tacL-mutant nor higher DiIC₁₂ incorporation in the complemented mutant. Taken together, changes in lipoprotein abundance and lower DiIC₁₂ dye incorporation in the bacterial cell membrane of LTA-deficient pneumococci D39∆cps∆tacL and EF3030∆tacL suggest that the fluidity of the cell membrane is reduced in these strains and a stiffer membrane is disadvantageous for pneumococci under in vivo conditions.

The integrity of the PGN and cytoplasmic membrane surrounding the cytoplasm of Gram-positive bacteria is crucial for cellular homeostasis, to survive environmental stress conditions, and to resist innate immune defence mechanisms [50‒52]. To understand the impact of the altered membrane composition in the absence of LTA, we compared the stress resistance of wild-type strains and tacL-mutants. In the presence of 5 mm and 10 mm hydrogen peroxide (H₂O₂), which causes oxidative stress, we observed that the presence of LTA is essential for resistance against oxidative stress. Regardless of the genetic background and serotype tested, tacL-mutants lacking LTA showed a significantly lower survival rate compared to the isogenic wild-type strain possessing LTA (Fig. 4a). The complemented strains showed a behaviour similar to the wild-type D39∆cps, EF3030, and TIGR4∆cps, respectively (Fig. 4a).

Cathelicidin, also known as LL-37, is a cationic AMP (cAMP) produced by epithelial and circulating immune cells, representing the most prominent cAMP in bronchoalveolar fluids [53]. To assess the role of LTA in cAMP resistance, pneumococci were cultured in the presence or absence of different LL-37 concentrations (0 μm, 1 μm, or 2 μm) and the survival rate was determined. Our data suggest that pneumococcal resistance to LL-37 correlates with the presence of surface-exposed LTA, which is shown by lower survival rate of the tacL-mutants (Fig. 4b). However, statistical significance could only be shown for serotype 2 strain D39∆cps, while serotype 19F strain EF3030 showed only a trend towards a higher susceptibility against LL-37 in the absence of LTA (Fig. 4b).

In addition, we tested the role of LTA for the susceptibility against the Cp-1 phage-encoded lysozyme CPL-1. This enzyme contains a specific choline-binding domain and acts as a muramidase. Therefore, it specifically binds to the phosphorylcholine decorations of the pneumococcal TAs and induces lysis of the bacterial cell wall [54]. Due to the high potency, we tested pneumococcal survival in the presence of a low (2 μg/mL) and high (50 μg/mL) lysozyme concentration. For serotype 4 wild-type strain TIGR4 and its tacL-mutant, we could not measure that loss of LTA had an impact on pneumococcal lysis, independent of the used CPL-1 concentration (Fig. 4c). In contrast, serotype 2 and 19F strains lacking LTA showed a significantly higher susceptibility to CPL-1 than the isogenic wild-type strains (Fig. 4c). Remarkably, complemented tacL-mutants of 19F and D39∆cps showed even a slightly higher resistance to CPL-1. Again, this is most likely a gene dosage effect and caused by the higher numbers of tacL-gene copies in these strains (online suppl. Fig. S1). Taken together, the absence of LTA leads to a higher susceptibility to oxidative stress, which is most likely not influenced by the observed, serotype-dependent changes in membrane fluidity because all three analysed serotypes showed similar changes. For the analysed susceptibility to AMPs, no or only very minor differences were detected. However, the tested mutant strains of serotype 2 and 19F showed an overall similar trend in their susceptibility, while this effect was not observed for the serotype 4 strain TIGR4 (Fig. 4b, c).

LTA plays an important role in the network of different polymers that form the bacterial cell wall. Due to their ability to bind CBPs via their non-covalent interaction with phosphorylcholine residues of TAs, these polysaccharides are important for cell wall remodelling and thus for the pneumococcal cell physiology. In some Gram-positive bacteria, e.g., in Staphylococcus aureus, the lack of LTA leads to lethal cell defects due to dramatic alterations of the bacterial cell morphology [55, 56]. For S. pneumoniae, we have recently shown that the genetic ablation of the TA ligase TacL leads to a complete loss of LTA. Unexpectedly, the bacterial cells showed a normal growth behaviour. Nevertheless, the invasive serotype 2 strain D39 lacking LTA (D39ΔtacL) showed a reduced ability to adhere to lung epithelial cells and was significantly attenuated in a mouse pneumonia and sepsis infection model [14]. Here, we demonstrate that the reduction in virulence is not only relevant for invasive serotypes but also for non-invasive serotypes such as S. pneumoniae serotype 19F strain EF3030. This strain was isolated from a patient suffering on otitis media and stably colonizes the respiratory tract of mice without causing any clinical symptoms [46, 47]. EF3030 lacking LTA showed a slightly reduced ability to adhere to human nasopharyngeal epithelial cells in our in vitro infection model. In accordance with these results that most likely explain our in vivo data, EF3030 lacking TacL and hence LTA showed a minor reduction in bacterial load in the nasopharynx of the murine colonization model over the course of 14 days. Interestingly, the wild-type strain EF3030, which is non-invasive in mice, showed high virulence in the G. mellonella larval infection model after intrahemocelic injection of the bacteria. EF3030 lacking LTA was significantly attenuated in the G. mellonella model, while the complemented mutant was as virulent as the wild-type. Because of their humoral system that is − like that of most other insects − based on a complex system of opsonins, AMPs, and lysozymes, our data implied that these substances in particular show a higher efficiency against LTA-deficient strains [44].

In general, it is not well understood how LTA compositional variation modulates the lipid membrane stability and integrity. Liposomes mimicking the lipid composition of the Gram-positive model organism Bacillus subtilis have been used to investigate structural changes in the lipid layer when the amount of LTA is changed. The study clearly indicated that variations in LTA affect the stability and rigidity of the phospholipid membrane [57]. Taking these findings and our data into account, we asked whether lack of LTA in pneumococcal strains also affects the membrane stability. Indeed, our data showed a reduction in membrane fluidity in the absence of LTA, which resulted in a lower incorporation of the fluorescent membrane dye DiIC₁₂. The lower membrane fluidity is probably due to a variation in the protein/lipid ratio. Studies on membrane composition showed that the membrane flexibility is negatively affected when the ratio shifts to the protein sides [58]. Proteins directly intercalated into the membrane are the lipoproteins. Comparing the amount of lipoproteins between the wild-type strains and their respective isogenic mutants indicated a higher lipoprotein abundance on the surface of D39∆cps and EF3030 lacking LTA when using DacB, PsaA, and SlrA as prototypes. However, for TIGR4 ∆cps only PsaA was significantly enhanced in the tacL-mutant, while DacB and SlrA were similarly present as in wild-type. The TIGR4 data are in accordance with our flow cytometric analysis using the membrane dye since we did not measure differences in membrane flexibility among TIGR4∆cps strains. These data suggest that the effects induced in the absence of LTA might be strain specific, a phenomenon that has to be investigated further.

Various studies have shown that the membrane composition of bacteria changes in response to environmental influences [59]. We therefore investigated the susceptibility against oxidative stress using H₂O₂. All tacL-mutants including the one from TIGR4 showed a high susceptibility against oxidative stress induced by H₂O₂. Even the higher abundance of surface-exposed PsaA, which is the substrate-binding protein of the manganese ATP-binding cassette transporter and has been reported to be important for oxidative stress resistance stress [60], had no protective effect. The high susceptibility of TacL-deficient pneumococci to H₂O₂ treatment is in a hitherto unknown manner linked to the reduced membrane fluidity and lack of LTA, because the increased susceptibility was also observed for TIGR4 strains, which showed no changes in membrane fluidity. However, the increased resistance observed for the complemented mutants and in particular in strain EF3030∆tacL∇tacL might suggest that the LTA is directly or indirectly influencing the defence against oxidative stress. The higher tacL expression in the complemented strain leads most likely to an increased incorporation of LTA into the cell membrane, which would also support the results of the membrane fluidity analyses. A recent study showed that PspA-mediated aggregation protects pneumococcal desiccation, when incubated on abiotic surfaces [61]. To elucidate the impact of LTA on desiccation, we incubated EF3030, its isogenic tacL-mutant, and complemented strain on glass slides at room temperature. Under the selected conditions and in the absence of LTA, pneumococcal killing is accelerated (online suppl. Fig. S3). This might be due to a lower resistance against osmotic stress, which is caused by higher salt concentration during desiccation. Thus, the H₂O₂ results are corroborated.

To test the susceptibility against AMPs, we incubated the investigated strains with human cathelicidin LL-37, which is primarily produced by epithelial cells and could therefore also be detected in the nasopharynx and in the human lung fluid [53, 62‒64]. IL-37 plays a role not only as a defence mechanism against invasive bacteria but also against colonizing bacteria. Like other AMPs, it acts by attaching to the negatively charged bacterial membrane and disrupting it by forming transmembrane pores [65]. Our results indicated a significantly increased susceptibility of D39∆cps lacking LTA against LL-37, while EF3030ΔtacL showed a trend towards higher susceptibility compared to the wild-type and TIGR4∆cpsΔtacL was similar susceptible to its wild-type strain. The higher in vitro susceptibility of strains lacking LTA to this AMP may also play a role in invasive infections or during in vivo colonization. One defence mechanism of Gram-positive bacteria against cAMPs is the D-alanylation of TAs. The incorporation of these positively charged residues leads to a charge-dependent defence against cAMPs. The general capability of some S. pneumoniae strains to decorate their LTA with alanine has been shown; however, the degree seems to be different among serotypes [12, 66]. Interestingly, proteomic analyses of S. pneumoniae have identified 52 protein candidates that may play an important role in resistance to LL-37, with the TA flippase TacF being one of these proteins [67, 68].

Our previous study have already shown that although the deletion of TacL leads to a total loss of LTA, the amount of P-Cho residues detected in the cell wall remains unchanged [14]. This could be due to the fact that the majority of the TAs are represented by the WTA or because the existing TA precursor chains are increasingly incorporated as WTA in TacL-deficient strains. The higher susceptibility against the CPL-1 lysozyme in serotype 2 strain D39 and serotype 19F strain EF3030 is therefore most likely not due to lower CPL-1 binding sites. In addition, the total amount of CBPs seems to be unaltered [14]. CBPs such as the autolysins LytA, LytB, or LytC are involved in the remodelling of the PGN or, like Pce (phosphorylcholine esterase), in the decoration of P-Cho residues to the TAs [69‒73]. However, when CPL-1 is exogenously added and bound to available P-Cho residues not decorated with CBPs, the combination of a lower membrane flexibility and additional enzyme cleaving PGN might result in a higher killing of S. pneumoniae lacking LTA. The different susceptibility measured for the non-encapsulated D39 with a sensitivity already at 2 μg/mL and encapsulated EF3030 with a sensitivity at 50 μg/mL might be due to their capsule phenotype. The capsule in EF3030, although not very thick, may mask P-Cho binding sites for CPL-1.

In conclusion, this study indicates the critical role of LTA during pneumococcal colonization and in escaping the innate immune defence mechanisms of the host. Our data showed that alterations in LTA are associated with changes in membrane flexibility and integrity and the balanced expression of WTA and LTA is required to maintain the cell wall integrity and thus resistance against host defence mechanisms.

We would like to thank Professor Anders Håkansson (Lund University, Sweden) for providing S. pneumoniae 19F strain EF3030. We thank Kristine Sievert-Giermann and Peggy Stremlow (Department of Molecular Genetics and Infection Biology, University of Greifswald) for technical assistance.

The animal experiments were conducted in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals (National Research Council, Washington, DC, USA), the guidelines of the Ethics Committee at the University of Greifswald, and the German regulations of the Society for Laboratory Animal Science (GVSOLAS) and the European Health Law of the Federation of Laboratory Animal Science Associations (FELASA). The study protocol was reviewed and approved by Landesamt für Landwirtschaft, Lebensmittelsicherheit und Fischerei Mecklenburg-Vorpommern (LALLFV M-V, Rostock, Germany) and the LALLFV M-V ethical board (LALLF M-V permit No. 7221.3-1-006/23). All efforts were made to minimize suffering, ensure the highest ethical standard, and adhere to the 3R principle (reduction, refinement, and replacement).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

This work was supported by grants of the Deutsche Forschungsgemeinschaft to S.H. (HA 3125/5-2) and N.G. (GI 979/1-2). The funders had no role in study design, decision to publish, or manuscript preparation.

The experiments were conceived and designed by M.B. and S.H. Experiments were performed by M.B., A.P., and T.P.K. Mutant was constructed by M.B. and J.V.N. Editorial advice was given by S.H., T.P.K., and N.G. Data analysis was done by and manuscript was written by M.B. Revision was done by N.G. and S.H. All authors contributed to the article and approved the submitted version.

All relevant data are contained within the manuscript and supporting information files.