The phospholipid bilayer of the membrane parietal monoderm bacteria is ∼90 Å thick and is composed of 10–40% lipids, 40–75% proteins, and 0.2–20% carbohydrates (Salton, 1967; Ghosh and Carroll, 1968; Bodman and Welker, 1969; Duda et al., 2006). Although membrane phospholipids vary from one species to another, the most commonly found in the CM are glycerophospholipids including phosphatidylglycerol, diphosphatidylglycerol (cardiolipin), and to some extend phosphatidylethanolamine and their amino acylated forms (Fischer et al., 1978; Roy, 2009; Malanovic and Lohner, 2016). Phospholipids vary also by their two fatty acid moieties, which impact on membrane fluidity (Mishra et al., 2012; Custer et al., 2014; Diomande et al., 2015; Malanovic and Lohner, 2016). Polyisoprenoid lipids are other important regulators of membrane fluidity. They constitute, together with cardiolipins and bacterial flotillins acting as scaffolding proteins, nanoscale functional membrane microdomains, which seem essential to the proper functioning of signal transduction cascades and protein transport in Bacillus subtilis and Staphylococcus aureus cells (Lopez and Kolter, 2010; Bramkamp and Lopez, 2015). By analogy with eukaryotic membranes, these microdomains are also referred to as lipid rafts. Consistently, membrane proteins or associated complexes constitute discrete focal sites in the CM and CW (Campo et al., 2004; Rosch et al., 2007; Lopez and Kolter, 2010; Kandaswamy et al., 2013). Biological significance of functional membrane microdomains could be to serve as platforms that control the assembly of membrane and CW proteins and multiprotein complexes involved in numerous cellular processes, such as cell division, protein trafficking, genetic transfer, or signal transduction (Lopez and Kolter, 2010; Schneider et al., 2015). Subcellular localization and spatiotemporal distribution of CM and CW proteins or supramolecular protein complexes are often intimately linked to their function and vary with the environmental conditions (Bierne and Dramsi, 2012; Mitra et al., 2016).

The OWZ constitutes the main part of the CW. It is 15–30 nm thick and comprises the PG and WTA polymers (Navarre and Schneewind, 1999; Vollmer et al., 2008). The PG is made of N-acetylglucosamine (NAG) and N-acetylmuramic acid (NAM) forming disaccharide glycan chains of various lengths that are cross-linked by peptides. PG composition depends on bacteria (Schleifer and Kandler, 1972). The glycan chain is uniform, whereas the peptide moiety and the cross-links are variable. The two major PGs in parietal monoderm bacteria have a meso-diaminopimelic acid (A2pm) or a lysin at the third position of the peptide. At this position, the cross-link occurs directly or through a penta-glycine bond, respectively. In B. subtilis, it is estimated that the glycan chain length is 1300 disaccharides in average, and that approximately 20% of the peptide chains are cross-linked (Ward, 1973; Atrih et al., 1998; Hayhurst et al., 2008). These glycan chains form helices of ∼50 nm width, and it was proposed that these cable-like structures coil around the narrow axis of the bacterium and are cross-linked by peptides (Hayhurst et al., 2008). The glycan chains of ovococcal bacteria, e.g., Streptococcus sp., are formed of more than 100 disaccharide units in average, whereas the glycan chains of cocci, e.g., Staphylococcus sp., are relatively short with 5–10 disaccharide units in average (Wheeler et al., 2011). The average effective mesh size in PG, i.e., the tessera, is estimated at 2.2 nm (Koch, 1990; Demchick and Koch, 1996; Figure 3). In other words, hydrophilic molecules of about 25 kDa (but also probably up to 50 kDa) can freely pass through a structured CW meshwork. Along with this, the CW network is actually not perfect, e.g., pseudo-tessera, and numerous PG defects cause increase in the porosity (Pink et al., 2000; Turner et al., 2013; Kim et al., 2015). Of note, though, the critical hole size in the CW beyond which lysis occurs, is estimated in the range of 15–24 nm (Mitchell et al., 2013). The OWZ of parietal monoderm bacteria is a very dynamic structure, as bacterial growth requires constant remodelling of the CW, which has a turnover rate of 50% per generation (Koch and Doyle, 1985). Remodelling is mediated by CW-anchored autolysins that are active on the outermost layer of the PG (Jolliffe et al., 1981).

LTAs and WTAs are zwitterionic polymers anchored to the CM and CW, respectively. They are major polyanionic teichoic acids of the envelope of parietal monoderm bacteria. LTAs are localised in the IWZ at the interface of the CM and the CW (Neuhaus and Baddiley, 2003; Reichmann and Grundling, 2011; Schneewind and Missiakas, 2012; Percy and Grundling, 2014). The most common LTA structure found in Firmicutes and, referred as type I LTAs, consists in a polyglycerol phosphate polymer linked to a glycolipid anchor, often a diglucosyl-diacylglycerol (Glc2-DAG), anchored to the CM. Type II, III, IV, and V LTAs have more complex repeating units that contain glycosyl residues, e.g., in Streptococcus pneumoniae, type IV LTA is decorated with phosphocholine.

WTAs are covalently attached by the PG disaccharide unit via a phosphodiester linkage to NAM (Neuhaus and Baddiley, 2003; Brown et al., 2013). Although the structures of WTAs vary considerably between species, the most common ones are composed of glycerol-phosphate or ribitol-phosphate repeats. LTAs and WTAs are often modified with sugar moieties and D-alanine esters, which introduce positive charges to neutralise the negatively charged phosphates in the polymer backbone (Wooldridge and Williams, 1993; Xia et al., 2010; Schneewind and Missiakas, 2012; Percy and Grundling, 2014; Carvalho et al., 2015). In addition to their diversity between and within species, the degree of D-alanylation of teichoic acids is fine tuned in changing environments and thus likely influences the protein repertoire displayed at the CW. The zwitterionic WTA polymers potentially contribute to the sequestration of divalent cations within the OWZ, including Ca²⁺, Mg²⁺, and Fe²⁺ (Beveridge and Murray, 1980), and might thus influence the regulation of protein transport across the CW (Forster and Marquis, 2012).

WPs have various compositions, e.g., teichuronic acids in Bacillus (Ward, 1981) or highly diverse heteropolysaccharides in Lactococcus (Yasuda et al., 2011; Vinogradov et al., 2013; Ainsworth et al., 2014), which complexity and diversity can be even greater than expected as revealed by the ever increasing genome data regularly made available. The last and most external layer of the CW may be composed of a capsule, generally composed of WPs (Jones, 2005; Yother, 2011). Although the WP capsule structures are well documented, the anchoring was recently proposed to be at the β-D-N-acetylglucosamine of the PG via a direct glycosidic bond (Larson and Yother, 2017). In some cases, the capsule is composed of polyglutamate, e.g., in Bacillus anthracis (McLean et al., 1992; Candela and Fouet, 2006). Poly-γ-D-glutamate anchoring was reported to be covalent at the PG (Candela and Fouet, 2005; Candela et al., 2005). However, the exact anchoring mechanism is still controversial and may be either on the A2pm or on the PG glucosamine (Richter et al., 2009; Candela et al., 2014).

Overall, the CW of parietal monoderm bacteria is a complex structure that protects them from mechanical and osmotic lysis, and serves as a scaffold for anchoring proteins, glycopolymers, and cations that perform various functions (Navarre and Schneewind, 1999; Weidenmaier and Peschel, 2008). While WPs or WTAs can be essential for bacterial growth (Oh et al., 2017), WTAs have been shown to be dispensable in some other bacterial species (Chapot-Chartier and Kulakauskas, 2014; Mistou et al., 2016). However, wall rhamnose polysaccharides (RhaWPs) can be a functional counterpart of WTAs, as suggested in Streptococcus agalactiae and Streptococcus pyogenes (Caliot et al., 2012; van Sorge et al., 2014), where they appear to be covalently linked to PG NAM (Deng et al., 2000).

As a common theme, all IMPs exhibit hydrophobic transmembrane α-helical domains (TMDs) enabling anchoring of the protein to the membrane (White and von Heijne, 2004). IMPs can be broadly discriminated between single-spanning IMPs (ss-IMPs) exhibiting a single TMD and multispanning-IMPs (ms-IMPs) with more than one TMD (Figure 4; Goder and Spiess, 2001; Higy et al., 2004). Whereas most IMPs are not synthesised with a cleavable N-terminal SP, some IMPs are (Facey and Kuhn, 2004). For the latters and after cleavage of the hydrophobic transmembrane α-helical SP by a signal peptidase (SPase), the ss-IMPs remain anchored to CM thanks to an additional hydrophobic TMD, i.e., the stop-transfer sequence also called signal domain of type I (SD1), which exhibits a N_out–C_in topology; as such, these ss-IMPs refer to the type I (ss-IMP1; Figure 4). Type II ss-IMPs (ss-IMP2) have a signal-anchor sequence also called signal domain of type II (SD2), with a N_in–C_out topology, which actually corresponds to an uncleavable SP. Type III ss-IMPs (ss-IMP3) have reverse signal-anchor sequence, i.e., a SD1 (TMD with a N_out–C_in topology); in the literature, they are sometimes described as ss-IMP1 without SP since the reverse signal-anchor sequence is a SD1. Of note, while the translocation mechanism of both type I and type II IMPs is in line with our current knowledge about the Sec/YidC translocation, i.e., involving an N-terminal SP (whenever cleavable or uncleavable) targeting the protein to CM, the mechanism for the translocation of type III IMPs in the absence of a SP remain unclear. In ms-IMPs, the type I (N_out–C_in TMD topology) and type II (N_in–C_out TMD topology) signals alternate along the protein sequence. Based on topology of the most N-terminal TMD enabling anchoring of the ms-IMP to the CM, the three types mentioned here above can be discriminated (Figure 4).

IMP biogenesis in lipopolysaccharidic-diderm bacteria (archetypal Gram-negative bacteria) involves an integrase known as YidC (Scotti et al., 2000). Up to two paralogues of the integrase YidC have been uncovered in parietal monoderm bacteria, namely SpoIIIJ and YqjG (Tjalsma et al., 2000; van Wely et al., 2001). While both SpoIIIJ and YqjG are involved IMP biogenesis and are essential for cell viability (Murakami et al., 2002; Tjalsma et al., 2003), SpoIIIJ is required for sporulation in B. subtilis but not YqjG (Errington et al., 1992; Murakami et al., 2002). Lately, these proteins have been renamed YidC1 and YidC2, respectively, in parietal monoderm bacteria (Funes et al., 2009; Wang and Dalbey, 2011; Palmer et al., 2012). In E. coli, YidC is associated to the Sec translocase enabling insertion of all IMPs to the CM in a SRP (signal-recognition particle) dependent mechanism (Scotti et al., 2000; Fröderberg et al., 2003; Ziehe et al., 2017). In this species, the YidC pathway is quite versatile since integration of IMPs to the CM can be SecA-, SecB-, and/or Sec-independent (Samuelson et al., 2000; Beck et al., 2001; Yen et al., 2002; Fröderberg et al., 2003; White and von Heijne, 2004). Moreover, flotillin-like proteins could contribute to the insertion of IMPs (Dempwolff et al., 2016). So far, these aspects have been poorly investigated in parietal monoderm bacteria.

In parietal monoderm bacteria, lipoproteins are synthesised as pre-prolipoproteins that are exported by the Sec pathway and exposed on the outer face of the CM (Hutchings et al., 2009; Figure 5). The pre-prolipoproteins exhibit a SP of type II (SP II) that is harbouring a conserved lipobox motif at the cleavage site (Sutcliffe and Harrington, 2002). The consensus sequence for the lipobox is [LVI]_-3-[ASTVI]_-2-[GAS]_-1-[C]₊₁ (Sutcliffe and Harrington, 2002; Babu et al., 2006). Once translocated across the CM, the lipoprotein maturation pathway in parietal monoderm bacteria is a two-step process. First, the lipobox motif is recognised by a prolipoprotein diacylglyceryl transferase (Lgt), which transfers of a diacylglyceryl moiety from a glycerophospholipid onto the thiol group of the conserved cysteine, giving rise to the prolipoprotein. Then, the SP II of the prolipoprotein is cleaved off by a lipoprotein signal peptidase (Lsp), generating a mature lipoprotein. The lipoprotein is consequently covalently bound to the acyl moiety of two fatty acids from the diacylglyceride by a cysteine residue at position 1 of the N-terminal end (Lai et al., 1980). Besides this classical form of lipid-modified cysteine for lipoprotein anchoring to the CM, intensive mass spectrometry analyses revealed three novel forms of mature lipoproteins in parietal monoderm bacteria (Nakayama et al., 2012; Figure 5). The N-acylated triacyl form of lipoproteins containing N-acyl-S-diacyl-glyceryl-cysteine was identified in S. aureus and S. epidermidis (Kurokawa et al., 2009; Asanuma et al., 2011). The N-acetyl form of lipoproteins identified in different Bacillaceae contains N-acetyl-S-diacyl-glyceryl-cysteine (Kurokawa et al., 2012b). The lyso-form of lipoproteins containing an N-acyl-S-monoacyl-glyceryl-cysteine was identified in Bacillus cereus, Enterococcus faecalis, Lactobacillus bulgaricus, and Streptococcus sanguinis (Asanuma et al., 2011). It further appeared that environmental conditions influenced the ratio between diacyl and triacyl forms of lipoproteins in S. aureus, with an accumulation of the diacyl lipoprotein form at high temperatures and high salt concentrations (Kurokawa et al., 2012a). Together, these recent findings are suggestive of uncharacterised non-canonical pathways for differential lipoprotein lipidation in parietal monoderm bacteria, analogous to the N-acylation of the lipidated cysteine by the apolipoprotein N-acyltransferase (Lnt) in lipopolysaccharidic-diderm bacteria. Actually, the lipoprotein intramolecular transacylase (Lit) involved in N-lyso-form biosynthesis was recently identified in E. faecalis and B. cereus (Armbruster and Meredith, 2017). If N-acylation is likely to involve acyltransferases adapted to specific phospholipids as acyl-donor substrates, novel enzymes and maybe pathways are to be discovered to explain how these alternative N-acetyl lipoprotein forms are biosynthesised in parietal monoderm bacteria.

Covalent binding of LPXTG-proteins to the CW has been the subject of intensive studies and is certainly one of the best characterised molecular mechanisms for protein anchoring to the PG (Fischetti et al., 1990; Schneewind et al., 1992). Here, we review the major mechanism of anchoring. In parietal monoderm bacteria, a range of proteins called LPXTG (IPR019948) is covalently linked to the PG by enzymes named sortases. Among LPXTG-proteins are found colonising factors, toxins and proteases. In parietal monoderms, the LPXTG motif was identified in both classes of Actinobacteria and Firmicutes, especially in the orders of Coriobacteriales, Streptomycetales, Propionibacteriales, Bifidobacteriales, Micrococcales, and Corynebacteriales for the former, and the orders of Erysipelotrichales, Clostridiales, Lactobacillales, Bacillales, and Tissierellales for the latter. This is a C-terminal motif composed of the LPXTG sequence where X represents any amino acids, followed by a hydrophobic domain and a short positively charged tail. Several variations around this motif were reported, e.g., NP(Q/K)TN, but the overall motifs remain homologous and are included for profile search (Boekhorst et al., 2005). In any case, the motif is recognised by sortases that are classified into six classes from A to F (Dramsi and Bierne, 2017; Siegel et al., 2017). Sortase A anchors a wide range of LPXTG-proteins, whereas sortase B recognises the NP(Q/K)TN related motif. Sortase C allows the pilus assembly (see below), whereas sortases D, E, and F have been much less characterised. Sortases anchor the LPXTG-proteins on the nascent PG through their transpeptidase activity, by cleaving between T and G (or N) and transferring the protein on the PG. Depending on the PG nature, molecular binding can occur at the pentaglycine crossbridge (Marraffini and Schneewind, 2005) or at the A2pm (Budzik et al., 2008).

Besides covalent binding to the PG, some proteins exhibit conserved motifs enabling specific binding to the CW components via weak interactions, such as van der Waals interactions, hydrogen or ions bonds.

The ChW (clostridial hydrophobic repeat with a conserved W residue) domain (IPR006637) was first identified in Clostridium acetobutylicum and was predicted to be involved in cell surface anchoring or in protein–protein interaction (Nölling et al., 2001; Desvaux, 2005a). This domain is constituted of highly conserved GW dipeptide motifs and is about 50 amino acids long. A single protein can harbour between one and 12 ChW domains. It was suggested that the ChW domains are associated in triplet for the surface interaction but the biochemical nature of the CW ligand remains unknown (Sullivan et al., 2007). In parietal monoderm bacteria, the ChW domain is essentially found in some Firmicutes, especially of the class Clostridia, but also in some Erysipelotrichia and Bacilli, e.g., in the genera Lactococcus, Streptococcus, and Enterococcus, as well as in some bacteria of the phyla Actinobacteria, especially in the genus Streptomycetes. ChW-proteins are mostly endolysins suggesting the importance of this domain for CW interaction and enzymatic activity (Oliveira et al., 2013).

Parietal monoderm bacteria have some surface-exposed proteins that lack a canonical signal sequence and a CW interacting domain. Although not sharing any domain or sequence homology, they share the ability to interact with fibronectin or extracellular matrix (ECM)-components. The most common are cytoplasmic enzymes or proteins, referred as moonlighting proteins. They include the ubiquitous glycolysis enzymes glyceraldehyde-3-phosphate dehydrogenase (GAPDH; Pancholi and Fischetti, 1992), enolase, phosphoglycerate kinase, the glutamine synthetase (GlnS), and the translation elongation factor Ef-Tu (Amblee and Jeffery, 2015). While most act as adhesins by interacting with components of the host ECM (plasminogen, fibronectin, laminin, or mucin), some like Ef-Tu interact also with neuropeptides at the membrane level (Fulde et al., 2013; Mijouin et al., 2013; Amblee and Jeffery, 2015; N’Diaye et al., 2016a,b, 2017). Beside the anticipated lack of an SP and cell surface association domain, bioinformatic analysis of 98 experimentally reported intracellular proteins having a moonlighting cell surface function, failed to identify specific features shared by these proteins (Amblee and Jeffery, 2015). The domain interacting with plasminogen is frequently localised at the C-terminus of the protein, however, no conserved domain could be identified (Bergmann et al., 2003; Ehinger et al., 2004). Ionic bonds and low pH were shown to contribute to the association of cytoplasmic proteins with the cell surface of Lactobacillus crispatus (Antikainen et al., 2007). Conversely, GAPDH and enolase have been shown to bind LTA on the bacterial cell surface by ionic bonds (Antikainen et al., 2007; Kinoshita et al., 2008). Reversely, GAPDH of S. pneumoniae did not bind synthetic LTAs or TAs and direct binding to PG was observed (Terrasse et al., 2015). Moonlighting proteins occur in all bacteria and are thus involved in a large range of unrelated functions including colonisation, modulation of the host response and virulence (Kainulainen and Korhonen, 2014).

Other non-classical proteins exposed on the cell surface of parietal monoderm bacteria are known as fibronectin-binding proteins (FBPs) characterised by two adjacent conserved domains: the about first 400 amino acids (PF05833) of which 89 residues associate with fibronectin-binding activity (Courtney et al., 1994) followed by the conserved domain of unknown function DUF814 (IPR008532) of ∼100 amino acid residues including conserved motif (D/E)X(W/Y)XH. First identified in the fibronectin-binding protein FBP54 of S. pyogenes (Courtney et al., 1994), these domains have been reported in PavA of S. pneumoniae (Holmes et al., 2001), FbpA of Streptococcus gordonii (Christie et al., 2002), Fbp68 of C. difficile (Hennequin et al., 2003), FbpB of Clostridium perfringens (Katayama et al., 2009), EfbA in E. faecalis (Torelli et al., 2012), Fnm in E. faecium (Somarajan et al., 2015), YloA in B. subtilis (Rodriguez Ayala et al., 2017) and FbpA in Weissella cibaria (Wang et al., 2017). Consistently with impaired binding capacity to fibronectin upon nested deletions of the C-terminal part of S. pneumoniae PavA (Holmes et al., 2001), structural and functional analyses of FBPS of Streptococcus suis revealed that the C-terminal half of FBPs mediates binding to fibronectin whereas the N-terminal half interacts specifically with the surface of Streptococcus suis (Musyoki et al., 2016). The fact that N-terminal half of FBPs does not bind to S. pneumoniae nor S. agalactiae cells suggests an interaction with a specific CW component that remains to be identified. Despite their contribution to fibronectin binding and overall virulence in several pathogens, the exact role of these FBPs is still unclear (Kawabata et al., 2001; Dramsi et al., 2004; Pracht et al., 2005; Torelli et al., 2012; Somarajan et al., 2015).

How these proteins are released outside from the cell and attached to the cell surface remains poorly understood. Several lines of evidence indicate that the release of GAPDH of S. agalactiae, S. aureus, and S. pneumoniae bacterial involves autolysis (Pasztor et al., 2010; Oliveira et al., 2012; Terrasse et al., 2015). Consistently, moonlighting proteins localise preferentially at the septum. However, this issue is still debated as not all cytoplasmic proteins are detected at the CW (Ebner et al., 2016). Interestingly, based on indirect evidence using an inhibitor of a mechanosensitive channel it was recently proposed that EF-Tu and DnaK of S. epidermidis could be exported through the large mechanosensitive channel (N’Diaye et al., 2016b). Variations of channel diameter between bacterial species may explain differences between patterns of moonlighting proteins.

S-layer is a proteinaceous two-dimensional crystalline array constituting the outermost CW layer in the absence of a capsule (Fagan and Fairweather, 2014). Located above the PG, this surface supramolecular structure is not a common theme in parietal monoderm bacteria, e.g., it is present in numerous Bacillus or Clostridium species but absent from Listeria and Staphylococcus genera. Usually, a S-layer is formed by the auto-assembly of a unique protein that may be glycosylated. The S-layer proteins are usually rich in hydrophobic and acidic amino acids (Sára and Sleytr, 2000). The interactions between the S-layer subunits are stronger than surface interactions (Messner and Sleytr, 1992). Most often bacteria with an S-layer possess a single S-layer and in very rare cases two (Kuen et al., 1997). Most of the S-layer proteins are non-covalently anchored through SLH or CWP domains at the bacterial surface. Of note and as mentioned above, a protein harbouring a SLH or a CWP domain is not necessarily an S-layer protein.

The function of the S-layer remains unclear but it is generally suggested to act either as a scaffold, a sieve or a shield to some environmental stresses (Sára and Sleytr, 2000; Fagan and Fairweather, 2014; Gerbino et al., 2015). While cited as a virulent factor or adhesion factor, such a role has not been convincingly demonstrated in any parietal monoderm bacteria. In B. anthracis, Sap was suggested to be the receptor of the phage AP50c (Plaut et al., 2014). Investigating its contribution to colonisation processes, S-layer was negatively correlated with biofilm formation in B. cereus (Auger et al., 2009). Consistently, a C. difficile mutant strain lacking the Cwp84 protease, which plays a key role in the maturation of the S-layer protein SlpA, forms a biofilm 72-fold more important than the wild type strain (Pantaleon et al., 2015). Except for C. difficile (Merrigan et al., 2013), the bacterial S-layer is considered as non-essential. The S-layer proteins can account for up to 15% of total bacterial proteins, also the absence of common physiological functions among bacteria is intriguing. Undoubtedly, this call for in-depth investigations under conditions more relevant to the ecophysiology of the bacterial species considered.

The bacterial flagellum is secreted and assembled via the FEA. Several transmembrane components constitute the translocon (FlhAB-FliOPQR) and form the translocase together with the ATPase FliI (Macnab, 2003, 2004). The flagella per se is composed of a basal body, the hook, the junction and the filament proteins, which are secreted and assembled by the FEA (Evans et al., 2014). These proteins do not exhibit a SP and the signal necessary for targeting is still controversial (Aldridge and Hughes, 2001, 2002). While most knowledge about the assembly and regulation mechanisms results from investigations in different LPS-diderm bacteria, information related to parietal monoderm bacteria remains restricted to fewer bacterial species, e.g., B. subtilis (Mukherjee and Kearns, 2014; Rossez et al., 2015).

Of course the flagellum is a well-known motility factor that can be involved in swimming but also swarming (Henrichsen, 1972; Belas, 2014). Swarming is especially relevant for surface colonisation processes, including adhesion and biofilm formation (Beeby, 2015; Chaban et al., 2015). Mechanosensing by flagella and chemotaxis further allow the bacteria to switch developmental programmes and adapt in response to changes in their environment. Glycosylation of the flagella has now been demonstrated in several parietal monoderm bacteria (Schirm et al., 2004, 2005; Twine et al., 2008, 2009; Kajikawa et al., 2016) and they further appeared to play a role in pathogenesis and biofilm formation (Valguarnera et al., 2016; Valiente et al., 2016).

Pili are tubular cell-surface appendages, which size, diameter, and shape can be extremely variable depending on the type of appendage considered. In parietal monoderm bacteria, three main types of pili can be encountered, (i) the pili made of covalently linked subunits involved in colonisation and host interaction, (ii) the type 4 pili (T4P) involved in transformation, motility and adherence, and the most recently uncovered (iii) pili made of amyloids. In parietal monoderm bacteria, beside the evidence the formation of a DNA translocation channel, no conjugative pili has been formally demonstrated.

A cellulosome is a supramolecular multienzymatic complex present at the bacterial surface of some parietal monoderm bacteria and dedicated to degradation of plant CW polysaccharides (Bayer et al., 2004; Fontes and Gilbert, 2010). It is organised around a scaffolding which assembles different catalytic subunits. Cellulosomes are only found in some parietal monoderm bacteria of the families Lachnospiraceae and Clostridiaceae (Guedon et al., 2000; Desvaux and Petitdemange, 2001; Desvaux, 2005a). In C. thermocellum, the scaffolding CipA (cellulosome-integrating protein A) is composed of a CBM (carbohydrate-binding module), a DocII (dockerin domain of type II), and 9 CohI (cohesin domains of type I) (Béguin and Lemaire, 1996; Smith and Bayer, 2013). Whereas different types of CBM allows binding to different carbohydrate-polymers with various affinities, CohI acts as a receptor domain for a dockerin domain of type I (DocI) harboured by cellulosomal enzymes. CipA can display up to nine different cellulosomal enzymes thanks to the 9 CohI and its anchor to the bacterial cell surface via DocII, which interacts with a CohII (type II cohesin domain) presents in a cell-surface protein, such as SdbA (scaffolding dockerin binding A) (Stern et al., 2016). Depending on the bacterial species, the scaffolding can harbour more or less cohesion domains (Schwarz, 2001). An even higher level of complexity can even be reached when different scaffoldings assemble one with another from complex cellulosomes like in Pseudobacteroides cellulosolvens or Acetivibrio cellulolyticus (Xu et al., 2003, 2004; Hamberg et al., 2014). The assembly of several cellulosomes to form a polycellulosome would occur via DocI/CohI interactions (Carvalho et al., 2003). Cellulosome components exhibit SP and would be secreted by the Sec pathway, even so the mechanism for the assembly of the different subunits at the cell surface appears quite complex (Desvaux, 2005a,b; Bras et al., 2016; Bule et al., 2017; Smith et al., 2017). The cellulosome is generally exposed at the bacterial cell surface following cohesin-dockerin interaction with cell-surface proteins, themselves anchored in the CW via SLH domains.