The choline-binding proteins PspA, PspC, and LytA of Streptococcus pneumoniae and their interaction with human endothelial and red blood cells

ABSTRACT Streptococcus pneumoniae is a Gram-positive opportunistic pathogen that can colonize the upper respiratory tract. It is a leading cause of a wide range of infectious diseases, including community-acquired pneumonia and meningitis. Pneumococcal infections cause 1–2 million deaths per year, most of which occur in developing countries. Here, we focused on three choline-binding proteins (CBPs), i.e., PspC, PspA, and LytA. These pneumococcal proteins have different surface-exposed regions but share related choline-binding anchors. These surface-exposed pneumococcal proteins are in direct contact with host cells and have diverse functions. We explored the role of the three CBPs on adhesion and pathogenicity in a human host by performing relevant imaging and functional analyses, such as electron microscopy, confocal laser scanning microscopy, and functional quantitative assays, targeting biofilm formation and the hemolytic capacity of S. pneumoniae. In vitro biofilm formation assays and electron microscopy experiments were used to examine the ability of knockout mutant strains lacking the lytA, pspC, or pspA genes to adhere to surfaces. We found that LytA plays an important role in robust synthesis of the biofilm matrix. PspA and PspC appeared crucial for the hemolytic effects of S. pneumoniae on human red blood cells. Furthermore, all knockout mutants caused less damage to endothelial cells than wild-type bacteria, highlighting the significance of each CPB for the overall pathogenicity of S. pneumoniae. Hence, in addition to their structural function within the cell wall of S. pneumoniae, each of these three surface-exposed CBPs controls or mediates multiple steps during bacterial pathogenesis.

erythromycin or 100 µg/mL kanamycin was used for selection. Growth was monitored by measuring the optical density at 600 nm (OD 600 ).

Construction of S. pneumoniae D39-derivated mutants
The pspA, pspC, and the lytA deletion mutants were generated in the genetic background of the nonencapsulated strain D39∆cps (44,45). The construction of the pspC and pspA mutants was described earlier (25,46). For the lytA mutant, the gene region of lytA from a S. pneumoniae D39∆lytA insertion-deletion mutant was amplified from genomic DNA (kind gift of R. Brückner, Kaiserslautern) using primer LytA_KO_f (5′ GGTGTTATCCTTTGT GAACCTC 3′) and LytA_KO_r (5′ GCAATCATGCTTTGATTCAAA 3′). The resulting, 1973 bp fragment contains 498 bp upstream of lytA, 38 bp from the beginning of the lytA gene, followed by ermR, 42 bp from the end of the lytA gene and 486 bp downstream of lytA. The amplified PCR fragment was used to transform D39∆cps using routine protocols (44). The resulting colonies were selected on Luria Broth agar plates containing kanamycin and erythromycin and verified by PCR and agarose gel analysis. (46) (46) For the complementation of lytA mutant, the gene lytA was cloned into an IPTG inducible plasmid for overexpression. The final construct harbored a kanamycin resistance cassette for selection and an N-terminal GFP-His-tag. Protein expression was carried out using Escherichia coli BL21 strain and the recombinant protein was purified by Ni-Sepharose chromatography. The final LytA amount was measured and the correct size was confirmed by fluorescence detection upon gel electrophoresis.
Adherent human cells were washed with pre-warmed Dulbecco's phosphate-buffered saline (DPBS; BioWhittaker) and harvested by incubation for 10 min at 37°C with PBS containing trypsin/EDTA (Gibco). Cell detachment was stopped by adding 10 mL of growth medium. After centrifugation, the pellet was resuspended in 1 mL growth medium, and the cells were counted using the cell counter CASY (OLSCASY).

Static biofilm model
Pneumococci biofilms were grown in either THY or DMEM media to mid-logarithmic phase. Bacteria were washed and resuspended in the corresponding medium at a concentration of 1 × 10 6 cells/mL). Bacterial suspensions were incubated on sterile, 18 mm round glass 1.5 h coverslips (Roth) in the bottom of 24-well polystyrene plates (Thermo Scientific). Exceptionally, bacterial suspensions were incubated on 12-well plates containing 12 mm round glass coverslips. The plates were incubated at 37°C with 5% CO 2 for 48 h. The growth medium was changed every 6 h. Bacterial biofilms were either evaluated by scanning electron microscopy (SEM) or visualized by CLSM.

Biofilm quantification
For quantification of biofilm formation, the Microtiter Dish Biofilm Formation Assay was performed with small changes (47). Pneumococci were grown overnight on solid agar blood plates. Bacteria were resuspended in PBS and diluted in THY or DMEM media to reach OD 600 of 0.1. Bacteria were statically grown on 96-well plates to obtain biofilms (Thermo Scientific), at 37°C with 5% CO 2 . At the indicated time points, the supernatant was transferred to another plate and OD 600 measured as a read for planktonic growth. To each well of the original plate, 100 µL of a 1% crystal violet solution was added and the plate was incubated for 30 min at room temperature. After repeated washing with water, the plate was left to dry for 1 h. Ethanol [95% (vol/vol)] was added to each well and left for 30 min at room temperature. The released crystal violet was finally transferred to a new 96-well plate, and absorbance at 620 nm was measured. Statistical analysis was performed using Prism version 9 for Windows (GraphPad Software, La Jolla, CA).

Confocal laser scanning microscopy
Bacterial viability within the biofilms was evaluated using the Bacterial Viability Stain kit (Biotium) according to the manufacturer's description. Bacteria were treated as described in Static Biofilm section. Planktonic bacteria were removed and the remaining biofilm layer was washed with DPBS and stained with a fluorescence red marker for dead cells, Ethidium Homodimer III (EthD-III) and a green peptidoglycan dye, wheat germ agglutinin conjugated to CF488A. After staining, biofilms were washed to remove unbound dyes and mounted using SlowFade Diamond (Invitrogen) mounting oil. Then the coverslip was sealed with nail polish and biofilms were evaluated by confocal laser scanning microscopy using a LSM 710 fitted with ZEN 2011 software (Zeiss GmbH).

Scanning electron microscopy
For SEM, biofilms were grown on 12-well plates containing 12 mm coverslips (Roth), as described above for the static biofilm model. At designated time points, medium was aspirated. Then cells were fixed for 1 h in 2.5% glutaraldehyde in sodium cacody late buffer (0.1 M, pH 7.0) and washed three times with sodium cacodylate buffer for 20 min each. Samples were dehydrated in rising ethanol concentrations followed by critical point drying, using a Leica EM CPD300 Automated Critical Point Dryer (Leica) and finally coated with gold (25 nm) in a Safematic CCU-010 HV Sputter Coating System (Safematic). SEM images were acquired at different magnifications in a Zeiss-LEO 1530 Gemini field-emission scanning electron microscope (Carl Zeiss) at 6-8kV acceleration voltage and a working distance of 5-7 mm using an InLense secondary electron detector for secondary electron imaging.

Hemolysis assays
S. pneumoniae strains were grown at 37°C with 5% CO 2 until reaching mid-logarithmic phase. Bacteria were washed and a 100 µL suspension was combined with 100 µL red blood cells [isolated from buffy coat as previously described (48,49,50], and the mixture was incubated in a 96-well plate (Thermo Scientific) at 37°C with slight agitation (300 rpm) for 30 min (positive control was only added 10 min prior to the end of the incubation). PBS was used as negative control and bi-distilled water as positive control for erythrocyte lysis. Erythrocytes derived from nine different volunteers were tested. Then the plates were centrifuged (400 × g, 15 min, 4°C), the supernatant was transferred to a new 96-well plate (Thermo Scientific), and hemoglobin release was quantified at OD 540nm . Statistical analysis was performed using Prism version 9 for Windows (GraphPad Software).

Bacterial incubation with human endothelial and epithelial cells
Endothelial HUVEC cells and epithelial A549 cells were seeded on an 18-mm diameter glass coverslips in a 12-well plate (Thermo Scientific), at a concentration of 200,000 cells/well and they were grown at 37°C with 5% CO 2 until confluence was reached. The confluent cells were then transferred to antibiotic-free growth medium and infected with pneumococci using a multiplicity of infection (MOI) of 50 bacteria per cell. The mixture was incubated at 37°C with 5% CO 2 . After washing with PBS, human cells and bacteria were fixed with 4% paraformaldehyde for 10 min at 4°C, followed by blocking with 1% bovine serum albumin for 1 h at room temperature. A rabbit anti-S. pneumoniae antibody (Abcam) was added, in order to visualize attached extracellular bacteria and also proliferating bacteria for 16 h at 4°C. For HUVEC cells, concomitant incubation of a secondary anti-rabbit antibody with 4′,6-Diamidino-2-Phenylindole, dihydrochloride (DAPI, Biotium) and platelet-endothelial cell adhesion molecule-1 (PECAM-1) conjugated with FITC was carried out for 1 h at room temperature. After the final washing steps with PBS, the coverslips were embedded in SlowFade Diamond (Thermo Fisher), sealed with nail polish and stored at 4°C for subsequent imaging. Images were taken on a confocal laser scanning microscope (LSM710, Zeiss).

Cell viability assay
Cytotoxicity of S. pneumoniae D39 or the isogenic mutants toward human epithelial cells was accessed using a CellTiter-Blue (CTB) Cell Viability Assay (Promega), as described here (51). Human cells were seeded on a 96-well plate (Thermo Scientific) with a concentration of 15,000 cells/well. Cells were cultivated at 37°C with 5% CO 2 until confluence was reached. Then bacteria were added and the mixture was incubated for 1 h under the same growth conditions. Subsequently, unbound bacteria were removed by washing with DPBS. The extracellular and human cell bound pneumococci were killed by treatment of the cells with gentamicin (500 µg/mL) for 1 h at 37°C under 5% CO 2 . CTB (100 µL) was added to each well. Following incubation for 16 h at 37°C in 5% CO 2 , the absorbance was measured using a Tecan Safire 2 microplate reader at an absorption of 570 nm. In this assay, intact metabolically active endothelial cells can convert the redox dye (resazurin) into a fluorescent end product (resorufin). Statistical analysis was performed using Prism version 9 for Windows (GraphPad Software, La Jolla, CA).

Construction of protein-protein interaction network
The Search Tool for Retrieval of Interacting Genes (STRING) (https://string-db.org) database, which integrates both known and predicted protein-protein interactions (PPIs), was applied to predict functional interactions of S. pneumoniae proteins (52). First, this interaction tool was used to evaluate the interaction of LytA, PspA, and PspC. Second, active interaction sources, including text mining, experiments, databases, co-expression, neighborhood, gene fusion, and co-occurrence, and an interaction score > 0.4 were applied to construct the PPI networks. STRING is a database of known and predicted protein-protein interactions. Given a list of the proteins as input, STRING can search for their neighbor interactors and generate the PPI network consisting of all these proteins and all the interactions between them. The interactions include direct (physical) and indirect (functional) associations; they stem from computational prediction, from knowledge transfer between organisms, and from interactions aggregated from other (primary) databases (53,54).

LytA prevents bacterial survival within mature-biofilm structures
To explore whether and how PspA, PspC, and LytA contribute to biofilm formation as the first step of adhesion, S. pneumoniae strains lacking the pspA, pspC, or lytA genes were grown in polystyrene multi-well plates, and biofilm formation was quanti fied. LytA-lacking bacteria produced significantly more biofilm mass, whereas biofilm formation by the pspA and pspC mutants was comparable to that by the reference strain (Fig. 1A). The lytA mutant was complemented with recombinant LytA (Fig. S1), and biofilm levels were restored to WT level, which proves the specific role of LytA in biofilm formation. To evaluate the structure of the newly formed biofilms and bacterial survival within the tri-dimensional structure, we quantified bacterial viability using CLSM. Within the biofilms, most lytA-lacking bacteria were viable, as shown by green fluorescence, whereas most bacteria derived from the pspA and pspC knockout strains and WT bacteria were dead, as revealed by red fluorescence (Fig. 1B), which goes in accordance with the absence of lytic phase of the lytA-mutant. Thus, absence of lytA initially affects bacterial lysis and then biofilm formation is altered, whereas absence of either pspA or pspC does neither influencn biofilm formati nor bacterial viability.
Next, we used SEM to evaluate matrix formation by the lytA mutant and the reference strain D39 in more detail. The lytA mutant produced less extracellular matrix than the parental strain D39 (Fig. 1C). At both low (Fig. 1C, panels 1 and 2) and high magnifications (Fig. 1C, panels 3 and 4), the WT strain (but not the lytA mutant) generated a prom inent granular matrix. Furthermore, the lytA mutant generated filament-like biofilms (Fig. 1D, panels 1 and 2) that allowed dense bacterial agglomeration (Fig. 1D, panels 3 and 4). To exclude an effect of growth medium, we compared growth in THY and DMEM (Fig. S2). When grown in DMEM, the ΔlytA strain did not produce extracellular matrix; biofilm formation was strongly impaired as visualized by lower cell numbers and the clear background; and bacterial cells were rounder, suggesting deregulated cell division. Taking into consideration the differential morphology and growth profile of ΔlytA strain, these results suggest an impaired extracellular matrix production on lytA deletion background.

PspA, PspC, and LytA reduce metabolic activity of epithelial cells
To examine the role of the three bacterial surface proteins in host cell damage, we first asked whether the mutants affect the metabolism of human alveolar epithelial cells (A549). To this end, A549 cells were co-cultivated with either knockout or WT D39 bacteria, and cell metabolism was evaluated by measuring the conversion of resazurin to the fluorescent product resorufin, which only occurs in metabolically active cells.  (panels 1-4).

Full-Length Text
Infection and Immunity Upon contact with each of the three mutants (ΔpspA, ΔpspC, and ΔlytA), metabolic activity in cells increased and was higher than in cells challenged with the WTstrain D39 ( Fig. 2A). Additionally, adhesion to epithelial cells was tested (Fig. 2B). The three mutant strains showed decreased adhesion to A549 epithelial cells in comparison with the WT strain. This shows that the pathogenic reference strain can adhere to and damage human epithelial cells and therefore decrease metabolic activity of human epithelia is observed.The deletion of a single CBP gene affects the ability to adhere and cause cell damage. Thus, deletion of one single CBP affects the metabolic turnover of human epithelial cells, thereby confirming that each CBP contributes to pathogenicity.

The three CBPs mediate pneumococcal hemolytic activity
To further define the role of the three CBPs on interactions with host cells, we evaluated the hemolytic capacity of the knockout mutants. The S. pneumoniae mutants were added to human red blood cells and, following incubation, erythrocyte lysis was evaluated. Erythrocyte lysis induced by each knockout was on average 82% less than that induced by the pathogenic reference strain D39 (Fig. 2C). Thus, each CBP contributes to the ability of the bacteria to induce erythrocyte lysis.

The three CBPs induce expression of PECAM-1 by endothelial cells
After addressing biofilm formation and the effects on red blood cells, we next examined the effects of the bacterial mutants on HUVECs. The human endothelial cell surface marker PECAM-1 is present in intercellular junctions and plays a major role in a number of cellular interactions (55). PECAM-1 expression relates to leucocyte migration and has been shown in a lung model that intravenous injection of anti-PECAM-1 antibodies leads to 75% reduction in neutrophil sequestration (56). Iovino et al. showed that PspC from S. pneumoniae binds to PECAM-1 in brain cells (57). Thus, expression of PECAM-1 (57-59) was monitored by CLSM. HUVEC cells were cultivated with either the CBP mutants or the WT strain. Each knockout mutant strongly reduced surface expression of PECAM-1 (Fig. 3A). Moreover, ΔlytA strain sustained bacterial growth when in contact with human endothelial cells, as seen by the intense red fluorescence signal (bacteria) in the ΔlytA panel; however, it did not induce expression of this surface marker. Semi-quantitative analysis of the microscopy data showed significant upregulation of PECAM-1 expression only in the presence of the reference D39 strain (Fig. 3B), corroborating the observed absence of cellular damage upon incubation with any of the CBP mutants. The pathogenic strain D39 was efficiently internalized by HUVEC cells, but all mutant strains showed less internalization by the human cells ( Fig. 3C and D). Thus, corroborating a lower adhesion and recognition of the mutants strains, lower levels of PECAM-1surface expression and less internalization by endothelial cells. Table S1 summarizes the effects of each pneumococcal mutant and the reference D39 strain. Apart from the effect of ∆lytA on biofilm formation, all other effects suggest that mutants show impaired pathogenic capacity, suggesting that the corresponding proteins play a crucial role in evading host immune responses.

PspA, PspC, and LytA are part of a complex network that regulates hostpathogen interactions
So far, the results show that each pneumococcal protein plays multiple roles during host-pathogen interactions, i.e., biofilm formation, hemolysis, and epithelial cell damage. These effects are distinct and involve diverse pneumococcal cellular subsystems, such as the cell division machinery (to regulate growth rate), expression of toxins (to induce host cell damage), and peptidoglycan synthesis (to release hemolytic enzymes). Therefore, we evaluated the connections between the three proteins within protein networks.
We constructed PPI networks for each pneumococcal protein using the STRING database. The initial network focused on PspA, PspC, and LytA, and showed that LytA mediates the interaction between PspA and PspC (Fig. S3A). An enlarged and more complex network map was then constructed based on several criteria; strong connections are represented by thick, dark-gray lines, whereas weaker connections (with fewer matching criteria) are represented by shaded, light-gray lines (Fig. S3B). The annotated proteins are presented next to the circular intersection node icon. After PPI

Full-Length Text
Infection and Immunity construction, a Markov Cluster Algorithm was used to identify cluster-specific groups (60). Four protein clusters appeared: a very densely interconnected cluster I (red), a small yet strongly interconnected cluster II (brown), cluster III comprising connections with different degrees of strength (green), and cluster IV (yellow) representing fewer interacting partners. Each cluster contained (predominantly) members related to a specific cellular machinery: cluster I = cell division; cluster II = chromosome replication; cluster III = peptidoglycan biosynthesis; and cluster IV = unannotated proteins. Each protein (PspA, PspC, or LytA) is embedded in a complex network of cross-interac tions, and each is integrated at a very prominent position, suggesting that targeting the proteins therapeutically can destabilize intracellular bacterial homeostasis. Additionally, this prominent positioning contextualizes and connects all of the data, suggesting that each of these CBPs have a multifactorial effect on host invasion and cell damage.

DISCUSSION
S. pneumoniae expresses a family of CBPs that use related cell wall anchors and act as central virulence proteins that interact with the soluble host complement regulators and with human immune system. Here, we evaluate and compare three important CBPs, i.e., PspA, PspC, and LytA, with respect to their effects on bacterial biofilm formation and host cell damage, and their integration into bacterial protein networks.
Pneumococci lacking LytA form more biofilm mass and less extracellular matrix. By contrast, PspA and PspC have no such effects on biofilm matrix formation. Previous studies show that CBPs play roles in invasion of the host under various environmental conditions (40,61). Efforts have been made to evaluate the role of extracellular DNA-CBPs complexes during establishment of biofilms; interestingly, LytA-mediated DNA complexes are relevant to biofilm formation by contributing for the general structure of the extracellular matrix and promoting potential horizontal gene transfer across bacterial populations (40,62). Here, we show that PspA and PspC seem not to influence biofilm formation, which supports current knowledge; however, our observations regarding LytA are rather different (Fig. 1). We found that LytA prevents, rather than promotes, biofilm formation, as bacteria lacking the gene encoding autolysin were more likely to form biofilms. Similar to the results presented here, another study reported an anti-biofilm effect of LytA (62); however, few studies have examined the specific and peculiar role of LytA in pneumococcal biofilms. Protein network analysis revealed that LytA is linked and integrated into the bacterial cell division machinery and teichoic acid biosynthesis (Fig.  S3). Triggering of the bacterial SOS response cascade, which is required for establishment of biofilms, might affect transcription of the lytA operon, most likely leading to repres sion.
This study also identifies a novel role for PspA, PspC, and LytA in hemolysis. Bacteria lacking pspA, pspC, or lytA are less efficient at inducing lysis of human erythrocytes (Fig. 2C). The PPI network map (Fig. S3B) shows that pneumolysin (Ply), a lytic pore-form ing pneumococcal enzyme (63), interacts closely with LytA (and even LytC, another pneumococcal CBP). However, PspA and PspC do not interact with ply. Instead, these two proteins interact with neuraminidase (NanA), an S. pneumoniae enzyme that is upregula ted upon contact with the host cells; secreted NanA cleaves host glycoconjugates (64). This relationship between PspA, PspC, and NanA might connect hemolysis and the two surface-bound pneumococcal proteins, and provide another link to biofilm formation. A recent study proposes that pneumococcus hemolysis activity is mediated mainly by hydrogen peroxide (H 2 O 2 ) rather than by excreted pneumolysin (49,65,66). In terms of the hemolytic profile, the pneumococcal mutants presented in that study, which do not produce H 2 O 2 , show a surprising resemblance to the pspA, pspC, and lytA mutants in the present study. Moreover, H 2 O 2 produced by S. pneumoniae damages human lung cells (67,68). Therefore, the reduced cytotoxicity of the CBPs mutants toward human epithelial and endothelial cells ( Fig. 2A and 3) might be a consequence of decreased H 2 O 2 production.
To obtain a deeper understanding of the role of CBPs in epithelial and endothe lial damage, a thorough metabolome analysis should be performed to examine the intracellular metabolic status of the mutants.
The association between bacterial cell wall-and/or membrane-anchored proteins and the stress response has been described before, with extensive studies being carried out in Gram-negative bacteria, particularly E. coli (69)(70)(71). In Gram-positive bacteria, the thick peptidoglycan layer is responsible for protection against environmental cues. Yet, other studies show the dynamic nature and diverse structure of the peptidoglycan layer across the cell wall of even a single bacterium (72,73). This diversity may lead to distinct distributions of bacterial surface-exposed proteins within a certain period of time, which might dictate the interaction with the human host. The interchange of information between the different intracellular metabolic pathways, and the arrangement of the cell wall (mainly the CBPs) is a crucial immune evasion strategy.
A thread connecting all of our experimental findings is the intermingled network of interactions sustained by the three CBPs (Fig. S3B). The PPI network map shows four clusters that relate to different cellular pathways. Cluster I contains mainly proteins related to the cell division machinery (DivIVA, FtsZ, FtsL, FtsX, and GpsB) (74); cluster II contains proteins related to chromosome replication (DnaA, DnaN, ParE, and PriA) (75); and cluster III contains proteins related to peptidoglycan biosynthesis (Pbp1B, Pbp2A, PenA, MurC, and MurD) (76). A relevant interaction partner for each evaluated protein is the serine/threonine-protein kinase StkP (66,(77)(78)(79). StkP plays a crucial role in regulating cell shape and division of S. pneumoniae through control of DivIVA activity. StkP is thought to sense intracellular peptidoglycan subunits in the division septum of actively growing cells and to adjust the regulation of DivIVA. The bond between StkP and the triad PspA-PspC-LytA suggests a link between each CBP in the cell wall of pneumococci and intracellular regulation subjacent to the synthesis of the cell wall itself, as well as coordination of cell division. Hence, S. pneumoniae fine-tunes the expression of surface-anchored immune evasion proteins in response to the cell cycle. Cluster II comprises additional CBPs, such as CbpC and CbpF (80), and proteins related to teichoic acid biosynthesis (LicB, LicC, and LicD) (81). The large cluster III also contains regulators with different backgrounds and functions, e.g., Ply (63), NanA, and NanB (both of which are implicated in biofilm formation and in colonization of the upper airway) (64) and PavA (which is critical for the overall virulence of pneumococci) (82). Cluster IV harbors transcriptional regulators, enzymes, and other uncharacterized proteins.
Understanding the interconnectivity of PspA, PspC, and LytA with other subcellu lar systems and their impact on bacterial homeostasis, is pivotal to mechanistically comprehend immune evasion by these proteins and to developing effective therapeu tics.

DATA AVAILABILITY
Protein interaction network raw data are available upon request.