Pneumococci (S. pneumoniae, the pneumococcus) are Gram-positive, facultative anaerobic bacteria, colonizing asymptomatically the upper human respiratory tract (URT). Adherence to a mucosal surface of host tissues, predominantly indirectly via components of the extracellular matrix (ECM), is a prerequisite for establishing stable colonization (Bogaert et al., 2004). However, under certain circumstances, pneumococci disseminate from the nasopharynx to deeper tissues and the blood, leading to pneumonia and invasive diseases such as septicemia or meningitis (Song et al., 2013; Weiser et al., 2018; Bradshaw et al., 2020). Pneumococcal infections are a major cause of invasive diseases (invasive pneumococcal diseases, IPD) and death globally, especially in the most susceptible populations such as children, the elderly, and immunocompromised persons (O'Brien et al., 2009). The highest mortality is reported for children. Therefore, pneumococci are also called “The Forgotten Killer of Children,” as mentioned by UNICEF and WHO (UNICEF, 2006).

Pneumococci are endowed with a plethora of virulence factors contributing to adhesion, colonization, immune evasion, and host cell damage (Ljungh et al., 1996; Kadioglu et al., 2008; Voss et al., 2012; Weiser et al., 2018; Jahn et al., 2020). The initial steps of pneumococcal pathogenesis require an intimate, specific adherence to host structures and modulation of innate immune clearance mechanisms (Weiser et al., 2018). Pneumococcal adhesins recruit and bind to different human ECM and serum glycoproteins, including fibronectin, fibrinogen, vitronectin, thrombospondin-1, collagen, and plasmin(ogen) (Holmes et al., 2001; Bergmann et al., 2009; Voss et al., 2012; Fulde et al., 2013; Binsker et al., 2015). Striking examples are the multifunctional adhesins PspC (also referred to as CbpA), PavB, PsrP, and pilus type-1 (Rosenow et al., 1997; Pracht et al., 2005; Anderton et al., 2007; Kanwal et al., 2017). The close interaction of pneumococci with nasopharyngeal host cells is initially prevented by mucus and ciliary beating of the microvilli on the apical pole of mucosal epithelial cells (Clarke et al., 2011). However, pneumolysin inhibits ciliary beat frequency (Peter et al., 2017; Nishimoto et al., 2020), and enzymes like the pneumococcal neuraminidase NanA and hyaluronidase Hyl contribute to receptor exposure on the surface of host cells (Weiser et al., 2018). Importantly, pneumococci exhibit the ability to hijack host-derived serine protease proteolytic activities by binding plasmin(ogen), enabling ECM degradation, which facilitates colonization and dissemination of bacteria (Bergmann et al., 2005; Bergmann et al., 2013; Weiser et al., 2018). Proteases, especially serine proteases, are found in all living organisms. The intracellular and extracellular proteases are considered to be the most abundant and functional proteolytic enzymes (Page and Di Cera, 2008). These enzymes either hydrolyze peptide bonds within proteins or cleave them at their amino- or carboxyl-terminal ends (Patel, 2017). Bacterial proteases are involved in cell homeostasis, protein transport, and the structural integrity of the cell wall (Burchacka and Witkowska, 2016; Marquart, 2021). Many bacterial species express serine proteases that play a significant role in pathogenesis, such as Bacteroides spp., Clostridium spp., Pseudomonas aeruginosa, and Streptococcus spp. (Macfarlane et al., 1988; Thibodeaux et al., 2007; de Stoppelaar et al., 2013; Martínez-García et al., 2018).

S. pneumoniae expresses a wide range of proteases and peptidases, including cysteine proteases, zinc-metalloproteases, and serine proteases (Wani et al., 1996; Ishii et al., 2006; Marquart, 2021). More than 34 proteases in S. pneumoniae TIGR4 were recently reported and discussed (Kwon et al., 2011; Marquart, 2021). These proteases have different functions like involvement in the acquisition of nutrients, protein quality control, signal peptide cleavage for pre-protein secretion, and cleavage of host ECM proteins (Proctor and Manning, 1990; Marquart, 2021). It is reported that some proteases play a significant role in virulence (Collin and Olsén, 2003; Weiser et al., 2018; Kriaa et al., 2020). For instance, the zinc-metalloprotease ZmpA (also known as IgA1 protease) interacts with the host immune system by cleaving IgA into inactive components, and the zinc-metalloprotease ZmpB is important for the modification of pneumococcal surface proteins (Kilian et al., 1980; Novak et al., 2000).

Besides in S. pneumoniae, serine proteases (or serine endopeptidases) have been found in many bacterial species such as Haemophilus influenzae, Pseudomonas aeruginosa, and other streptococcal species like Streptococcus agalactiae (group B streptococcus, GBS) (Male, 1979; Lyon and Caparon, 2004). Generally, the pneumococcus expresses different surface protein classes (Bergmann and Hammerschmidt, 2006; Pribyl et al., 2014; Kohler et al., 2016). Sortase-anchored proteins are covalently anchored to the peptidoglycan (PGN) via the sortase A, which cleaves a C-terminally located LPXTG motif (Bergmann and Hammerschmidt, 2006; Hammerschmidt, 2006; Nobbs et al., 2009; Löfling et al., 2011). In addition, the pneumococcal cell wall is decorated with up to 16 choline-binding proteins (CBPs), which are non-covalently bound to the phosphorylcholine of teichoic acids (Gosink et al., 2000). CBPs have been reviewed elsewhere (Maestro and Sanz, 2016). In this context, all pneumococcal serine proteases can be secreted and exposed on the pneumococcal cell surface, as shown in Figure 1 . This extracellular localization enables a direct or indirect cleavage and inactivation of bound peptides, thereby leading to the degradation of specific substrates (Mann et al., 2006; Frolet et al., 2010). In fact, pneumococcal serine proteases are reported to play a crucial role in bacterial pathogenesis, such as adhesion, colonization, promotion of pneumococcal diseases, biofilm dispersal, and immune subversion of host cells ( Figure 4 ) (Bergmann and Hammerschmidt, 2006; Moscoso et al., 2006; Mitchell and Mitchell, 2010; Voss et al., 2012; Pribyl et al., 2014; Chao et al., 2020; Ali et al., 2021).

The information on how pneumococcal serine proteases interfere with pathogenesis is crucial with respect to our understanding of pneumococci-host interactions. This review will focus on the four different pneumococcal serine proteases: HtrA, SFP, PrtA, and CbpG. These enzymes, encoded by genes of the core genome, are highly conserved and present among different pneumococcal serotypes (Bethe et al., 2001; Desa et al., 2008). The proteolytic activity is characterized by three amino acid (aa) residues, Ser-His-Asp, which form a so-called catalytic triad. The serine proteinase A (PrtA) and subtilase family protein (SFP) are cell wall-associated serine proteases of the S8 family of peptidases (Blum et al., 2021; Marquart, 2021). They are secreted and anchored covalently to the cell wall via the sortase A (Bethe et al., 2001; de Stoppelaar et al., 2013). PrtA contributes to host lung damage in a murine systemic infection model (de Stoppelaar et al., 2013; Mahdi et al., 2015), and in accordance, the gene encoding for PrtA is upregulated in the blood during acute pneumonia in mice (Bethe et al., 2001). In contrast, SFP may facilitate pneumococcal growth even after a lower infection dose in the lower respiratory tract (de Stoppelaar et al., 2013). The high-temperature requirement A (HtrA) serine protease is membrane-associated via an unknown mechanism and lacking a specific anchoring motif (Seol et al., 1991; Gasc et al., 1998; Fan et al., 2011), whereas CbpG is non-covalently associated with the wall teichoic (WTA) and lipoteichoic acids (LTA) (Mann et al., 2006). Previous studies suggested that CbpG could be a multifunctional protease playing an important role in mucosal colonization and sepsis (Mann et al., 2006). HtrA is a heat shock protein and chaperone involved in protein quality control, cell division, colonization, and virulence (Sebert et al., 2002; Ibrahim et al., 2004a; Cassone et al., 2012).

HtrA and PrtA are upregulated in the heat-dispersed population among the genetic variants (Pettigrew et al., 2014). We recently reported that the deficiency in three out of four serine proteases of TIGR4 with only one functional gene/protein or the deficiency of all serine proteases dramatically reduces adherence and nasopharyngeal colonization (Ali et al., 2021). Interestingly, the pneumococcal serine proteases are highly conserved among all pneumococcal serotypes and immunogenic (Bethe et al., 2001; Li et al., 2016; Hsu et al., 2018; Kazemian et al., 2018). Hence, serine proteases-driven pathogenesis is opening the avenue for new targets to develop specific antimicrobials. In this regard, our review presents a comprehensive summary of our current knowledge of pneumococcal serine proteases in order to gain insight into their potential roles in pneumococcal virulence and pathogenesis at a molecular level.

To characterize and compare pneumococcal serine proteases on the molecular level, different database tools including PSORT db 3.0 (Yu et al., 2011), multiple sequence alignment Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/) and pairwise sequence alignment (https://www.ebi.ac.uk/Tools/psa/emboss_water/) were used. All analyzed serine protease gene sequences (prtA (sp_0641), htrA (sp_2239), and cbpG (sp_0390)) of S. pneumoniae strain TIGR4 or D39 for SFP (spd_1753) were retrieved from the KEGG database (Kanehisa et al., 2006). Signal sequences were predicted using the software tool SignalP 4.0 (Emanuelsson et al., 2007; Petersen et al., 2011). Choline-binding proteins are characterized by their typical choline-binding modules (CBM) consisting of characteristic choline-binding repeats (CBRs) (Maestro and Sanz, 2016). Moreover, for the prediction of transmembrane helices, the TMHMM Server 2.0 algorithm (Hidden Markov Model for transmembrane protein topology prediction) was applied (Krogh et al., 2001). Functional domains were predicted using Pfam (Punta et al., 2012).

The genomes of 10 clinically relevant S. pneumoniae strains were analyzed on DNA and protein levels with BlastN and BlastP, respectively, for the homology analysis of pneumococcal serine proteases. The results revealed a maximum of four different serine proteases (NCBI, 2016). Comparisons on the protein level revealed high identities and similarities, indicating highly conserved sequences among the different pneumococcal strains ( Table 1 ).

HtrA belongs to the peptidase SA clan in the S1C family and is also identical to DO subfamily protease (Pallen and Wren, 1997). More than 180 members of these proteases, including HtrA, display trypsin-like protease characteristics (Pallen and Wren, 1997; NCBI, 2016). The family of these proteases combines a catalytic domain with at least one or more C-terminal PDZ domains (Lipinska et al., 1990), which is highly conserved in both pathogenic and nonpathogenic bacteria (Seol et al., 1991; Spiess et al., 1999; Backert et al., 2018). However, the first described HtrA protease in E. coli is known as DegP or DO protease and localized in the periplasmic space (Lipinska et al., 1990).

Bacterial HtrA is a heat-shock-induced serine protease that displays a multifunctional role like protein quality control and bacterial survival under different stress conditions such as oxidative and heat stress (Sebert et al., 2002; Singh et al., 2018). For instance, HtrA protease in Lactococcus is considered as a housekeeping protease (Poquet et al., 2000), while in other bacteria, HtrA prevents the cell from the cytotoxicity of misfolded proteins by refolding or degrading them (Clausen et al., 2002; Zarzecka et al., 2019). In E. coli, unlike other quality control proteins such as ClpXP, ClpAP, and HslUV, which need ATP for their chaperone function, HtrA is functional without ATP as an additional energy source (Clausen et al., 2011; Malet et al., 2012). More importantly, the function of HtrA proteins can be switched from chaperone to protease and the activity depends on the temperature (Spiess et al., 1999). The protease effect is in particular apparent at high temperatures ranging from 38-42°C, whereas the chaperon function is more pronounced at lower temperatures ranging from 30-37°C (Spiess et al., 1999).

In S. pneumoniae, HtrA is one of the best-studied and characterized serine proteases. High protein sequence identity (up to 100%) of the HtrA protein was detected in six different pneumococcal strains ( Table 1 ) such as D39, Hungary 19A, serotype 19F_EF3030 and R6, indicating that HtrA is highly conserved. Therefore, it could be a desirable drug target to prevent pneumococcal diseases (Wessler et al., 2017; Xue et al., 2021). In pneumococci, HtrA is a surface-exposed serine protease, easily accessible for potential inhibitory substances or anti-infectives. HtrA is immunogenic and antibodies against HtrA are protective against invasive pneumococcal diseases (Li et al., 2016).

The molecular analysis of different HtrA serine proteases of other pathogenic bacteria via multiple sequence alignment (MSA) revealed a sequence similarity, especially in the functional protease and PDZ domains, as reviewed more extensively elsewhere (Backert et al., 2018; Boehm et al., 2018; Singh et al., 2018). They are widely distributed in many bacterial species such as Escherichia coli, Legionella fallonii, Thermotoga maritima, and Mycobacterium tuberculosis (Kim et al., 2003; Bai et al., 2011; Malet et al., 2012; Cortes et al., 2013; Chang, 2016; Singh et al., 2018).

Furthermore, we also analyzed the amino acid sequence of pneumococcal HtrA orthologs in other streptococci ( Table 2 ). High sequence homologies of HtrA are present in S. pyogenes (group A streptococci), S. agalactiae (group B streptococci), S. mitis, and S. mutans. HtrA of S. pyogenes plays a significant role in cysteine protease streptococcal pyrogenic exotoxin B (SpeB) maturation and complement factor C5a cleavage (Lyon and Caparon, 2004; Cole et al., 2007). The deletion of HtrA in S. mutans enhanced the surface expression of several extracellular proteins such as glucan-binding protein GbpB and altered the biofilm formation (Biswas and Biswas, 2005).

Besides the impact of HtrA on pneumococcal virulence, HtrA was shown to be a multifunctional protein involved in pneumococcal growth at higher temperatures, tolerance to oxidative stress, genetic transformation, regulation of bacteriocin production, and cell division (Dawid et al., 2009; Fan et al., 2010; Tsui et al., 2011). The pneumococcal HtrA protein ( Figure 2A ) contains an amino‐terminal signal peptide (31 aa), cleaved by signal peptidase I for secretion. The terminal signal peptide is followed by a single transmembrane helix domain (aa 12-34). Thus, HtrA is found on the surface and/or secreted from S. pneumoniae as predicted by the presence of a putative amino-terminal signal peptide (de Stoppelaar et al., 2013; Ali et al., 2021). Additionally, HtrA contains two highly conserved unique domains, a serine protease domain and a PSD-95/Dlg/ZO-1 (PDZ) domain (de Stoppelaar et al., 2013). The trypsin-like serine protease domain has the typical triad His¹¹²-Asp¹⁵²-Ser²³⁴ (HDS) in the catalytic center, which was identified previously (de Stoppelaar et al., 2013) using Interproscan IPR009003 and IPR001940 (Zdobnov and Apweiler, 2001). Finally, the PDZ domain (abbreviation combining letters of the first three proteins discovered to share this domain, postsynaptic density protein, Drosophila discs large tumor suppressor, and zonula occludens-1) is located at the C-terminal end. HtrA in other bacterial species contains one or more PDZ domain(s) (Fan et al., 2011; Backert et al., 2018; Singh et al., 2018). In some situations, such as protein-protein interactions, the HtrA-PDZ domain acts as a protein folding stress sensor and controls the pyrolytic activity (Murwantoko et al., 2004; Wilken et al., 2004; Hasselblatt et al., 2007). Thus, the PDZ domains are responsible for recognizing and/or binding substrate proteins (Marquart, 2021). Fan et al. solved the pneumococcal HtrA-PDZ structure (Fan et al., 2011), which contains three α-helices and five β-strands (amino acid residues 262-386). Moreover, a comparison of the amino acid sequences of HtrA-PDZ domains in different bacterial species showed that the pneumococcal PDZ domain, which is most likely involved in the ligand recognition, has only a moderate sequence similarity and conserved secondary structure (Bohnsack et al., 2000).

The importance of HtrA in S. pneumoniae has been addressed in many studies. For instance, HtrA was shown to play an important role in pneumococcal competence, which is still challenging to understand due to the conflicting results. One study has shown that the HtrA protease is important for competence because the pneumococcal transformation efficiency was highly reduced in the htrA-mutant (Ibrahim et al., 2004b). In another study, the proteolytic activity analysis, which was performed with purified recombinant pneumococcal HtrA, revealed that HtrA cleaves the pneumococcal competence-stimulating peptide (CSP) in vitro (Cassone et al., 2012). Since CSP has a significant effect on pneumococcal transformation (Pestova et al., 1996), this fact suggests that HtrA has a considerable role in pneumococcal transformation efficiency and is needed for competence. In this study it has also been shown that the deletion of htrA or catalytic residues did not affect natural DNA competence (Cassone et al., 2012). However, the mutation strategies and transformation settings used in these two studies were different. It can be assumed that HtrA is necessary for the transformation process after the competence machinery is turned on by CSP. If the competence genes are expressed, CSP is not needed anymore and can be degraded by HtrA. Functional CSP seems to inhibit the transformation efficiency.

The specificity of HtrA toward CSP peptide degradation is based on a phenylalanine (nonpolar) residue. The addition of denatured bovine serum albumin (BSA) inhibits the CSP peptide from being cleaved by HtrA (Cassone et al., 2012). S. pneumoniae expresses several proteins contributing to competence, which are highly decreased during competence followed by stabilization with the exception of ComEA and ComEC. These membrane proteins are essential for pneumococcal transformation and responsible for DNA uptake (Liu et al., 2019). While htrA-mutants in the previous study have shown a lower transformation efficiency (Ibrahim et al., 2004b), ComEA or ComEC degradation was not evident. This suggests that HtrA plausibly degrades these proteins at later stages of competence (Liu et al., 2019). Last but not least, the regulation of HtrA seems to be dependent on bacterial culture conditions. It was shown that HtrA inhibits competence in a complex medium but not in a chemically defined medium (Petit et al., 2001). Overall, these findings show that HtrA acts as a competence regulator at the protein level and that environmental factors influence its regulation. Aside from the involvement of HtrA in competence, HtrA has been shown to be upregulated and controlled by the two-component regulatory system CiaRH (Sebert et al., 2002). A recent study showed that HtrA regulated by CiaRH is responsible for penicillin-binding protein 2x (PBP2x) degradation (Peters et al., 2021). In addition, HtrA is important for nasopharyngeal colonization and pneumococcal virulence (Sebert et al., 2002; de Stoppelaar et al., 2013; Ali et al., 2021).

The human pathogen S. pneumoniae expresses a special class of surface-proteins known as choline-binding proteins (CBPs). A common feature of this family of proteins is that they have a modular organization and are composed of at least two domains: a functional module (FM) and a choline-binding module (CBM). CBPs are found in pneumococci or closely related species (García et al., 1988; Sanz et al., 1992; Albrich et al., 2004; Blasi et al., 2012). The repetitive sequences of the CBM associate CBPs in a non-covalent manner to the cell wall by their interaction with phosphorylcholine residues of PGN-anchored WTA and membrane-anchored LTA (Pérez-Dorado et al., 2010; Maestro and Sanz, 2016). The CBM consists of three to eighteen repetitive sequences (CBRs) of about 20 amino acids (Pérez-Dorado et al., 2012; Galán-Bartual et al., 2015; Hilleringmann et al., 2015). Apart from LytB and LytC, the CBM is located in the C-terminal part of the protein, whereas the FM is located in the N-terminal region (Pérez-Dorado et al., 2010). The number of CBPs in S. pneumoniae ranges from 13 to 16 proteins and is strain-dependent (Gosink et al., 2000; Maestro and Sanz, 2016). Notably, CBPs play an essential role in the integrity of the cell wall, colonization processes, and interaction with host cells (Maestro and Sanz, 2016). Pneumococcal CbpG is a member of the CBP family, which also plays a significant role in pneumococcal mucosal colonization and during sepsis (Garcia-Bustos and Tomasz, 1987; Gosink et al., 2000). CbpG belongs to the peptidase S1, PA clan superfamily of peptidases, and is a trypsin-like serine protease (Kanz et al., 2005). The protein sequence indicates that this protein possesses a chymotrypsin-like fold and double β-barrel structure with a carboxyl-terminal choline-binding domain (NCBI, 2016; Yang et al., 2020). CbpG is considered to be a multifunctional surface-exposed serine protease with both proteolytic and adhesive functions (Gosink et al., 2000; Mann et al., 2006; Kazemian et al., 2018). These various functions of CbpG are necessary for the full virulence potential of S. pneumoniae. Such multifunctional proteinases can be found in many pathogenic bacterial species, and the C5a peptidase of group B streptococci (Beckmann et al., 2002; Cheng et al., 2002) and the well-characterized Pla surface protease from Yersinia pestis (Kukkonen and Korhonen, 2004) are striking examples.

Depending on the pneumococcal strain and serotype, there are at least two variants of CbpG produced by pneumococci. The truncated variant without CBM is shortened due to a premature stop codon after the N-terminal catalytic functional module and found in D39 (serotype 2), Hungary19A-6 (19A), R6 (2) and ST556 (19F) ( Figure S1 ). This variant is secreted and then released into the environment. In contrast, the full-length CbpG containing a CBM is cell wall-associated (Mann et al., 2006). The modular organization of CbpG ( Figure 2B ) shows that the catalytic residues are present independent of expressing a full-length protein, including a CBM or a truncated version without a functional CBM. In both configurations, the proteins exhibit proteolytic activity as confirmed earlier (Mann et al., 2006). Our genome re-analysis showed a high sequence identity and similarity of CbpG among various pneumococcal serotypes indicating CbpG is highly conserved and abundant among the different pneumococcal strains ( Table 1 ). The molecular analysis of full-length CbpG (sp_0390) in TIGR4 (Tettelin et al., 2001) comprises 285 aa with a molecular weight of 32 kDa, as shown in Figure 2B . According to our SignalP 4.0 analysis, a leader peptide (secretion signal peptide) is not present in all analyzed serotypes except for serotype 19F strain ST556 ( Figure S1 ). Therefore, it is still unknown whether and how CbpG is translocated from the cytoplasm to the bacterial cell surface. The functional domain is the trypsin-like domain with 184 aa spanning from aa 14-197, containing the catalytic triad His³⁴-Asp⁸⁷-Ser¹⁵⁹ as predicted by the 3D structure analysis ( Figure 3 ). Previous sequence analysis demonstrated 47% similarity of this domain to the S1 family of multifunctional surface-associated serine proteases (Mann et al., 2006). Furthermore, this domain is linked to the CBM by a short linker region (^aaLys-Pro-Phe-Ile^aa) that provides flexibility to the protein and may provide stability to the catalytic domain. This catalytic functional module exhibits sequence similarities to trypsin-like serine proteases present in all CbpG variants (Gosink et al., 2000; Mann et al., 2006).

Moreover, it has been mentioned that the CBM, which is e.g., in strain TIGR4, exhibits only three choline-binding repeats (CBRs), which are located at position aa 207-265. This represents the shortest identified CBM among all choline-binding proteins. It has been proposed that at least four repeats are needed to attach the protein non-covalently to the teichoic acids of the cell wall (Yother and White, 1994) Therefore, it is still unknown if CbpG can bind to the bacterial cell surface when only three choline-binding repeats are present. In deletion studies of the CBM from the pneumococcal LytA amidase (Mellroth et al., 2014), it has been hypothesized that a higher number of CBRs leads to a higher affinity for teichoic acids of S. pneumoniae (Maestro and Sanz, 2016).

The CbpG amino acid sequence model was analyzed (Ali et al., 2021) and suggests that a fourth CBR at position aa 267-285 might attach CbpG to teichoic acids and allows the protein to be surface-associated. This repeat includes the aromatic residues YW and fulfills the number of aromatic residues involved in choline-binding (Waterhouse et al., 2009). The protein sequence homology of the CbpG to orthologues of other bacterial species was analyzed as well. Significant homologies of CbpG (40-56%) were found to serine proteinases of different bacterial species such as Enterococcus faecalis, Staphylococcus aureus, and Salmonella enterica ( Table 2 ).

The protease PrtA belongs to the family of subtilisin-like proteases (also known as subtilases), which are part of S8 family peptidases (Bethe et al., 2001; Marquart, 2021). Pneumococcal PrtA is related to serine proteases present in lactococci, cleaving the amino-terminal leader sequences from lantibiotics (Blum et al., 2021). Lantibiotics are bacteriocin peptides that are bactericidal to outcompete other bacteria (Marquart, 2021). Interestingly, both, streptococci and lactococci exhibit a wide range of endopeptidase activity (Siezen, 1999).

In pneumococci, PrtA is a major surface serine protease involved in pneumococcal virulence (Zysk et al., 2000; Bethe et al., 2001). The role of PrtA in colonization and subsequent host invasion seems to be strain-specific (Mahdi et al., 2015). The first report on pneumococcal PrtA protease highlighted the immunogenicity because of its identification using convalescent-phase serum (Zysk et al., 2000). Interestingly, a previous study showed that PrtA is a highly conserved virulence factor in pneumococci and is found in almost all strains (Bethe et al., 2001). Both in silico analysis and flow cytometry confirmed that PrtA is surface localized (Wizemann et al., 2001).

The first molecular characterization of PrtA was done in 2001 (Bethe et al., 2001). Bethe and co-workers (Bethe et al., 2001) showed that pneumococci produce PrtA with different molecular weights. One variant produced by pneumococci has a molecular weight of 240 kDa, whereas a truncated form has only a molecular weight of 215 kDa, which cannot be explained by signal peptide cleavage only. The same observation was also found in the related proteases PrtP proteins of Lactobacillus paracasei and Lactococcus lactis (Vos et al., 1989; Holck and Naes, 1992). The full-length PrtA (strain TIGR4 sp_0641) form has a molecular weight of 240 kDa (2140 aa). The calculated mature form of PrtA has a molecular weight of 234 kDa after cleavage of the leader peptide and integration into peptidoglycan by sortase A. Furthermore, PrtA contains a typical sortase A recognition LPKTG motif spanning aa 2099-2140 followed by a hydrophobic region at the carboxy-terminus. The sortase A catalyzes covalent anchoring to the bacterial PGN (de Stoppelaar et al., 2013; Ali et al., 2021). PrtA consists of two domains, the active peptidase-S8 domain, which contains the typical catalytic triad (Asp²³²-His²⁹⁹- Ser⁶⁹⁰), spanning the region between aa 223-764 (Bethe et al., 2001; Ali et al., 2021). The second domain is a DUF-1034 (domain of unknown function), which consists of 140 amino acids and is localized between aa residues 795-934. The modular organization of PrtA is illustrated in ( Figure 2C ).

Of interest, the multisequence alignment of PrtA catalytic triad residues (Asp²³²-His²⁹⁹-Ser⁶⁹⁰) were highly homologous to other related bacterial species of subtilisin-like serine proteases. These catalytic triads showed a high degree of similarity and identity to the cell wall-associated proteases of Streptococci, Lactococci, and lactobacilli (Bethe et al., 2001; Bonifait et al., 2010).

Finally, the complete protein sequence homology to orthologues of other bacterial species was analyzed. As indicated in Table 2 , PrtA shares significant similarities with other streptococcal subtilisin-like proteases. Interestingly, PrtA seems to be highly immunogenic in humans and mice; two segments of PrtA, the amino-terminal and carboxy-terminal thirds were found to be protective (Wizemann et al., 2001).

The SFP serine protease (known as serine peptidase) is another enzyme able to cleave leader peptides from lantibiotics (Marquart, 2021). Similar to the PrtA protease, SFP belongs to subtilisin-like/or S8-family serine proteases. In S. pneumoniae D39 strain, SFP was identified as epidermin leader peptide processing serine protease EpiP (de Stoppelaar et al., 2013; Marquart, 2021).

The comparative analyses of sfp genes in S. pneumoniae strain D39 spd_1753 (1740 nt, 579 aa), and TIGR4 sp_1954 (1404 nt, 467 aa) was performed using the SYBIL database (Riley et al., 2011). The MSA analyses showed a shorter version of the sfp gene in TIGR4 compared to sfp of D39 and other strains ( Figure S2 ). The truncation of sfp in strain TIGR4 is based on the deletion of one base (A) at position 1381. Instead of 8 A bases in a row, only seven are present in strain TIGR4, which was confirmed by DNA sequencing of the TIGR4 sfp gene. The generated frameshift leads to the premature stop at position 1404. Hence, this truncated SFP of TIGR4 cannot be covalently anchored to the peptidoglycan. Instead, TIGR4 SFP is secreted into the extracellular environment. However, these data have to be experimentally verified.

Based on the molecular characterization of SFP (de Stoppelaar et al., 2013; Ali et al., 2021), its secretion and protease activity has been predicted. The full-length SFP has a molecular weight of 64.9 kDa and exhibits an N-terminal signal peptide (aa 1-22) and a C-terminal LPNTG anchoring motif which is thought to be functional as a target site for the sortase A and anchoring the protein to PGN (Marraffini et al., 2006). In addition, the peptidase domain spanning the aa residues 167-461 contains the catalytic triad (Asp¹⁷⁶-His²²³-Ser⁴²⁹) ( Figure 2D and Figure 3A ).

Furthermore, the genomic organization of the SFP locus in S. pneumoniae 19F and TIGR4/D39 strain is different. The sfp gene and six upstream and three downstream genes present in strain TIGR4 are not present in S. pneumoniae strain 19F EF3030 (Ali et al., 2021). Even more, the subtilisin-like protein SFP was not present in all the analyzed strains as observed by our in silico analysis ( Table 1 ). Therefore, pneumococci have at least three serine proteases in the 19F_EF3030 strain (Ali et al., 2021), but probably four in most strains, such as D39 (serotype 2), TIGR4 (serotype 4), ST81 (serotype 23F), JJA (serotype 14), and R6 (serotype 2). However, the role of SFP in pneumococcal virulence is still unknown.

The HtrA of E. coli is well characterized and studied in detail for its functional role as chaperone and protease. The crystal structures of HtrA from E. coli, Campylobacter jejuni or Termotoga maritima showed that the active protease at elevated temperature is composed at least as a trimer by hydrophobic interaction of the subunits (Kim et al., 2003; Zarzecka et al., 2020). By using computer-assisted analysis, we compared the catalytic center of all four serine proteases from pneumococci. The calculations were performed within the Multiple Sequence Viewer/Editor application in Maestro (Schrödinger, 2020-4) using an energy-based approach. Templates were obtained by BLAST search in the PDB database (SFP: 4MZD; PrtA: 5FAX; HtrA: 5ZVJ; CbpG: 1P3C) ( Figures S3 – 6 ). As mentioned, all serine proteases have in common the typical Ser-His-Asp triad, where the histidine is polarized through hydrogen bonding by aspartate, resulting in a polarization of serine and increased nucleophilicity of the hydroxyl oxygen atom. The highly conserved arrangement and distance between these three amino acids are crucial to form the catalytic center for the cleavage of peptide bonds.

On the one hand, the comparison revealed a quite similar catalytic domain structure between HtrA and CbpG with the common double β-barrel core motif adjacent to the catalytic triad. The Asp and Ser residues are localized on flexible loop structures, whereas the His residue is localized on a small helical fold ( Figure 3 ). On the other hand, a similar subtilisin-like catalytic domain of SFP and PrtA was observed by this modeling. Here, the overall fold consists of a dominant 7-stranded parallel β-sheet, with the catalytic Asp on the first strand (S1) and five α-helices containing Ser and His. While the core catalytic motif seems quite similar, a protease-associated domain is found within the amino acid sequence of the PrtA catalytic domain, which may mediate protein-protein interactions or substrate specificity. Due to low sequence identity, it was omitted for the homology modeling and should be further explored. Because this is only a simplified view of the active proteolytic centers of these serine proteases, there are ongoing efforts to purify the recombinant serine proteases for X-ray crystallography.

S. pneumoniae are versatile pathogens that modulate the immune response and circumvent host immune defense mechanisms. The enzymatic protease activity during pneumococcal infections can contribute to the destruction of the epithelial barrier or degradation of ECM components (Ljungh et al., 1996). Next, pneumococci try to establish a more severe infection by either transmigrating or/and disseminating to lungs, blood, middle ear, or the central nervous system.

Pneumococcal express various proteases and peptidases, which are involved in colonization, pneumonia, and septicemia (Marquart, 2021). In particular, pneumococcal serine proteases seem to play a role in invasive processes. In terms of specificity, substrates for serine protease are mainly the ECM component proteins, fibrin clots, cell membranes, and host immunomodulatory factors such as chemokines and cytokines (Frolet et al., 2010; Kim et al., 2010; Ruiz-Perez and Nataro, 2014).

Pneumococcal serine proteases might be involved in cleavage of adherence junctions or gap junction proteins to facilitate the pneumococcal paracellular route, which results in crossing of the epithelial barrier dissemination in the bloodstream. Recently, the impact of serine proteases on adherence, colonization, and subsequent virulence has been shown in various studies ( Table 2 ).

The first description of various pneumococcal serine proteases along with their susceptibilities to different inhibitors was in 1991 (Courtney, 1991). Already at that time, their important role in pneumococcal pathogenesis had been reported and indicated by the degradation of host tissue components such as fibronectin, fibrinogen, elastin, laminin, and blood proteins. As has been mentioned before, pneumococcal serine proteases are virulence factors either secreted and/or bound to the bacterial cell surface. The benefit of expressing serine proteases is likely a higher efficiency in colonizing the nasopharyngeal cavity (Ali et al., 2021). To date, studies on pneumococcal serine proteases have been only marginally concentrated on their role in virulence-associated processes such as adhesion, colonization, or host defense evasion. Nevertheless, this section discusses the individual or combined impact of pneumococcal serine proteases HtrA, CbpG, PrtA, and SFP on pneumococcal colonization and how they contribute to host-pathogen interactions.

Serine proteases in pathogenic bacteria are, in general, key virulence determinants. In pneumococci, serine proteases have a function during colonization and pneumonia. This review article covers the molecular biology of pneumococcal serine proteases and their pivotal role in pathogenesis, starting from adherence, colonization, and immune evasion. Our in silico analysis in combination with hypothetical structural models revealed that the functional domains of pneumococcal serine proteases CbpG, HtrA, and PrtA, are highly conserved. The exception is SFP, which is produced only by a subset of strains. All serine proteases are secreted to the cell surface and depending on the variant, even released in the host environment. The 3D models show that the HtrA catalytic domain displays homologies to the CbpG catalytic domain, while SFP is quite similar to the catalytic domain of PrtA ( Figure 3 ). Although all serine proteases have a typical catalytic triad, they might have different but also overlapping substrate specificities. The redundancy of serine proteases and probably their compensatory effect in the absence of one or more serine proteases makes it difficult to assess their individual contribution to pneumococcal fitness and virulence. Thus, all studies are in parts limited in their conclusions because of the redundancy of these serine proteases. This, in turn, leaves gaps of knowledge such as e.g., substrate specificities and host compartment specificities that have to be deciphered in experimental in vivo and advanced in vitro models. The immunogenicity of functional domains of pneumococcal serine proteases in combination with their highly conserved protein sequences fulfills one of the requirements for a protein-based serotype-independent (multi-) component vaccine. The individual potential as a vaccine candidate has, however, to be validated experimentally.

MA and SH conceived the concept for the review article. MA create the figures, drafted the work, and MA and GB performed the bioinformatic analyses. LS generated the 3D structural data and wrote this part. MA and SH wrote the review article, TK and GB revised it critically and gave final approval. All authors contributed to the article and approved the submitted version.

This study was supported by the German Academic Exchange Service (DAAD) as a grant scholarship and part of the Ph.D. thesis of MA. Funding programme/-ID: Research Grants - Doctoral Programmes in Germany, 2017/18 (57299294), ST33. This study was also supported in part by the DFG (GRK 2719). The funders had no role in study design, decision to publish, or manuscript preparation.

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.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.