New Paradigms of Pilus Assembly Mechanisms in Gram-Positive Actinobacteria

Abstract

Adhesive pili in Gram-positive bacteria represent a variety of extracellular multi-protein polymers that mediate bacterial colonization of specific host tissues and associated pathogenesis. Pili are assembled in two distinct but coupled steps, an orderly crosslinking of pilin monomers and subsequent anchoring of the polymer to peptidoglycan, catalyzed by two transpeptidase enzymes – the pilus-specific sortase and the housekeeping sortase. Here, we review this biphasic assembly mechanism from studies of two prototypical models, the heterotrimeric pili in Corynebacterium diphtheriae and the heterodimeric pili in Actinomyces oris, highlighting some newly emerged basic paradigms. The disparate mechanisms of protein ligation mediated by the pilus-specific sortase and the spatial positioning of adhesive pili on cell surface modulated by the housekeeping sortase are amongst the notable highlights.

INTRODUCTION

Fiber-like appendages called pili or fimbriae are microscopic structures present on the cell surface of both Gram-negative and Gram-positive bacteria. They are involved in a wide range of cellular activities, including adherence, motility, conjugation, and virulence [1–3]. Among these, the only pilus form known to date in which individual subunits are covalently bonded, is the pili that are assembled by the action of sortase enzymes conserved in Gram-positive bacteria [4], but not in Gram-negative bacteria that produce pili in which the monomer subunits are joined via protein-protein interaction. The various sortases found thus far are broadly grouped into six classes (SrtA-F), based on sequence alignment and substrate preference [5, 6]; of these, only members of the class C and class E sortases are shown to catalyze the two distinct steps of Gram-positive pilus assembly, pilus polymerization and anchoring of pili to the cell wall, respectively (Box 1) [7].

Box 1: Overview of class A, C, and E sortases.

Staphylococcus aureus SrtA is the prototype class A sortase [61], which recognizes a LPXTG motif, preceding a hydrophobic domain and a positively charged tail that together constitutes a “cell wall sorting signal” (CWSS) located at the C-terminus of a sortase substrate [62]. As a transpeptidase, SrtA catalyzes cell wall anchoring of surface proteins harboring the tripartite sorting signal, by first hydrolyzing the peptide bond between threonine and glycine and then covalently joining the cleaved threonine residue to the pentaglycine peptide of lipid II in the cell wall [63]. Structurally, the class A sortase harbors a unique β-barrel fold, with the catalytic site containing the sole cysteine residue [64], which is absolutely required for sortase activity [65].

Class C sortases, or pilus-specific sortases, are found in bacterial species that produce covalently linked pili [5]. They are structurally similar to class A sortases with the 8-stranded β-barrel fold encapsulating the active site [66]. Unique to class C sortases is a flexible N-terminal hydrophobic “lid” which covers the catalytic pocket and has been proposed to play a role in substrate recognition and sortase stability [6, 67–70].

Present in various Gram-positive bacterial species and abundant in Actinobacteria [5, 71], class E sortase enzymes recognize a distinct LAXTG sorting motif [5]. Compared to sortases of class A and C, structures of class E sortases have been less well studied, with only two available sortase structures from Streptomyces coelicolor and A. oris [60, 72]. While both harbor the conserved 8-stranded β-barrel fold without the aforementioned lid, they contain a conserved tyrosine residue within the β3/β4 sheet that appears to be involved in the recognition of the LAXTG sorting motif [60, 72].

Historically, the connection between sortase and pilus polymerization was somewhat serendipitous based on two types of genetic observations. First, it was recognized that the protein sequences of different fimbrial subunits in the actinobacterium Actinomyces naeslundii harbor the same C-terminal cell wall sorting signal (CWSS) as the classically defined cell surface protein, the protein A of Staphylococcus aureus [8, 9] (Box 1). Secondly, many sets of genes coding for sortase enzymes and surface proteins with the LPXTG motif are clustered in the same operons in the actinobacterium Corynebacterium diphtheriae [10]. Indeed, immuno-electron microscopic analysis using antibodies against some of these surface proteins revealed the presence of distinct classes of pili on the surface of C. diphtheriae [10]. Since the first demonstration of the essential function of specific sortases in pilus assembly in C. diphtheriae [3, 10, 11], the past decade has seen the extensive investigations of pilus assembly in many other Gram-positive bacteria including Bacillus cereus, Enterococcus faecalis, and streptococci [12–18]. Because both Actinomyces and C. diphtheriae continue to serve as excellent models in the studies of the genetic, biochemical and structural mechanisms of Gram-positive pilus assembly, here we focus our review on these two actinobacterial species only to highlight the recent advances in the field. For a more comprehensive description of pilus assembly in Gram-positive bacteria, we refer the reader to several excellent publications elsewhere [19–21].

Corynebacterium diphtheriae pili offer a new paradigm for protein ligation

Assembly of the SpaA pilus

The causative agent of human diphtheria C. diphtheriae is among the earliest bacterial species where pili were identified [22, 23]. As in many Gram-positive bacteria [4], the C. diphtheriae genes coding for distinct pilin subunits and dedicated pilus-specific sortases, which are all class C sortases, are organized into three operons [10]. Together, they encode three distinct pilus types specified by their major subunits, i.e. SpaA-type, SpaD-type and SpaH-type pili [10, 24, 25]. Pili constitute one of the major virulence factors in C. diphtheriae: A mutant devoid of all pilins is highly attenuated in virulence in a mouse model of diphtheritic toxemia [26]. Each type of pili of C. diphtheriae are heterotrimeric, meaning that each type of pili are made of three distinct pilin subunits; however, it is important to note further that the various molecules of the same pilus type (SpaA-, SpaD- or SpaH-type) vary in their length. Of these, the most highly studied is the SpaA pilus, which is composed of the shaft pilin SpaA, the tip adhesin SpaC and the base pilin SpaB anchored to the cell wall; assembly of this pilus requires the cognate sortase SrtA, a class C sortase [10] (Box 1). While all three pilins contain the CWSS, only SpaA has a recognizable pilin motif with a conserved lysine residue that serves as a nucleophile essential for sortase-mediated cross-linking of pilin monomers. In this reaction that can occur repeatedly, the pilus-specific sortase SrtA catalyzes hydrolysis of the LPXTG motif in a pilin subunit and links the cleaved threonine residue to the lysine residue within the pilin motif of another pilin that are adjoining each other on the bacterial membrane (Fig. 1). Because SpaC resides at the pilus tip (the tip first rule), the first transpeptidation reaction must occur between SpaC and SpaA, linking the threonine residue of the SpaC LPXTG motif to the reactive lysine residue within the pilin motif of SpaA (Fig. 1). Subsequently, pilus elongation ensues whereby SpaA pilins are added to the growing chain of pilus polymers. Ultimately, pilus polymerization is terminated with addition of SpaB (the base pilin), which is then anchored to peptidoglycan by the housekeeping sortase SrtF (class E sortase), whose gene is not genetically linked (or in proximity) to any of the three pilus gene clusters [27] (Fig. 1). This biphasic mechanism of pilus assembly appears to be universally applicable to other Gram-positive bacterial pili so far studied [12, 28–30].

Figure 1: Assembly of the heterotrimeric SpaA pili in Corynebacterium diphtheriae.

(A) Shown is the spaA pilus gene cluster in strain NCTC13129, encoding the pilus shaft SpaA, pilus base SpaB, and pilus tip SpaC, as well as the pilus-specific sortase SrtA. The SpaABC pilins contain a cell wall sorting signal with the LPLTG or LAFTG motif. The housekeeping sortase gene srtF is located elsewhere in the bacterial chromosome. (B) A the biphasic assembly mechanism of Gram-positive pili is depicted here with the SpaA-type pili. Pilin precursors are secreted in an unfolded state across the cytoplasmic membrane through the Sec translocon (Step 1). The membrane-bound thiol-disulfide oxidoreductase MdbA mediates disulfide bond formation and folding of pilin subunits (Step 2) prior to their insertion into the membrane (Step 3). The pilus-specific sortase SrtA catalyzed pilus polymerization into pilus fibers through successive lysine-transpeptidase reactions (Step 4). Incorporation of SpaB to the base of the pilus signals cell wall anchoring of the pilus by housekeeping sortase SrtF (Step 5). The order of the described transpeptidation reactions is marked by roman numerals (I-III); reproduced from Siegel et al [7].

Genetic and biochemical studies together have provided groundbreaking evidence to support the various steps of the aforementioned model (Fig. 1). Importantly, X-ray crystallographic studies of the SpaA pilin led to the discovery of an intramolecular disulfide bond and the role of a disulfide bridge-forming machine, termed MdbA, which is critically involved in mediating the post-translocational folding of the SpaA precursor pilin prior to its polymerization by pilus-specific sortase SrtA in the exoplasmic environment [26]. As expected from the model, deletion of srtA completely abolishes SpaA pilus polymerization [10], as does the genetic replacement of the nucleophilic lysine residue of the pilin motif (lysine to alanine substitution) that prevents pilus crosslinking [11]. The complete loss of SpaA pilus assembly in vivo in the absence of SrtA also demonstrates the substrate specificity of the pilus-specific sortase enzyme SrtA as other pilus-specific sortases expressed in vivo cannot substitute for SrtA function. When srtF or spaB is absent, however, abundant amounts of the generated pilus polymers are secreted into the culture medium, supporting the model that the pilus polymerization phase precedes the cell wall anchoring phase and that SpaB incorporation acts a pilus termination switch in this biphasic mode of pilus assembly [27, 31].

Recently, in vitro reconstitution of a Gram-positive pilus assembly system described for the first time that utilizes recombinant sortase enzyme and pilin substrate proteins has provided yet another foundational support for the biphasic model described above. This remarkable success in biochemical reconstitution was facilitated by structural genetic studies of the pilus-specific sortase SrtA, which uncovered novel structural features of the enzyme on the one hand, and also led to the engineering of a robust protein-polymerizing machine on the other hand. As we outlined in Box 1, the pilus-specific sortase SrtA harbors a structural lid that appears to occlude the enzyme’s catalytic pocket. Mutations of this lid could unmask the active site, thereby amplifying the net rounds of SpaA polymerization by this mutant enzyme (^CdSrtA^2M) to a level that was never observed with the wild-type enzyme, under comparable conditions [32].

The in vitro pilus assembly reaction contained the recombinant SrtA sortase truncated for its membrane localization domain and a SpaA protein devoid of its hydrophobic domain and charged tail [32], generating substantial amounts of pilus polymers within 24h that were easily detected by SDS-PAGE and Coomassie staining. Importantly, electron microscopic sampling revealed SpaA polymers with many >1 micron long fibers, while mass spectrometry authenticated the isopeptide linkage connecting individual subunits. Polymerization was abolished by a catalytic site mutation (C222A) as well as a pilin motif mutation (K190A); furthermore, the formation of an acyl-enzyme intermediate between SpaA and sortase, as the model predicted, was also observed. Remarkably, when SpaB or SrtF protein was added to the reaction, pilus polymerization was terminated, another prediction of the model [32]. The fact that SpaA polymers are formed without the presence of the tip pilin SpaC confirms the previous genetic observation that SpaC is dispensable for pilus assembly [10]. In essence, this test tube version of the reaction recapitulates much of the pilus assembly process that is observed in C. diphtheriae cells.

Protein ligation with pilus-specific sortase

The ability of a sortase to ligate proteins or peptides has significant implications in protein engineering, cell biology and biomedicine. Indeed, prior to the work with a pilus-specific sortase described above, Mao and coworkers have creatively utilized the most active recombinant sortase enzyme studied to date, i.e. S. aureus SrtA (^SaSrtA), in protein ligation [33]. In this study, the recombinant staphylococcal SrtA (rSa-SrtA) is capable of joining one substrate protein that is C-terminally tagged with the LPXTG peptide with another substrate N-terminally tagged with the G_n peptide (n= 1–5). Ploegh and colleagues further advanced this method for protein labeling in living cells [34]. Dubbed “sortagging”, this method appears to be a promising new engineering tool, which has been further optimized for site-specificity and the ability to covalently-link peptides to a variety of non-peptide substrates, including folate, amino-terminated or glycine-tagged polyethylene glycol, and beads [33, 35]. Sortagging was also utilized to create peptide nucleic acid cell-penetrating peptide conjugants, thus providing an exciting new tool for designing highly specific cell-permeable drug therapies [36]. Importantly, the high affinity of sortase for the LPXTG motif has yielded yet another protein capture method in which LPXTG motif containing peptides can be efficiently and specifically captured from complex cell lysates [35, 37]. The ability of sortase to mediate the stable anchoring of proteins to surfaces for microarray based protein activity assays has also been explored as staphylococcal SrtA has been used to covalently attach LPXTG-containing proteins to planar surfaces, such as glass coverslips, following treatment of the surface with aminosilane and oligoglycine [37]. Sortase protein labelling technology is not limited to protein extract applications as staphylococcal SrtA was used to efficiently ligate a modified pentaglycine probe to the surface of the transmembrane protein CD40L in live HEK 293T cells [34]. Finally, Tanaka and colleagues successfully used sortagging as a method of protein specific fluorescent labelling by conjugated glycine-containing biotin, EGFP, and Alexa-fluor probes to the transmembrane protein osteoclast differentiation factor (ODF) in HEK 293T cells, without inducing any toxic phenotype to the cell culture [38].

The protein ligation reaction catalyzed by S. aureus SrtA is limited to the N-terminal to C-terminal protein joining involving two defined substrates. By comparison, a variety of pilus-specific sortase enzymes identified to date offer the unique advantage as a bioconjugation tool via the formation of an isopeptide bond that is mechanically stable and less susceptible to proteolytic cleavage [39, 40]. Recently, McConnell and colleagues generated a recombinant C. diphtheriae sortase enzyme termed ^CdSrtA^3M that is more reactive than ^CdSrtA^2M described above. When a substrate protein containing the pilin motif was incubated with green fluorescent protein harboring a C-terminal LPLTG motif in the presence of ^CdSrtA^3M, the mutant enzyme catalyzed the covalent joining of the two recombinant proteins [41]. Furthermore, these authors demonstrated that both sortases ^SaSrtA and ^CdSrtA^3M can be utilized in sequential transpeptidation reactions to modify a protein of interest at distinct sites and with high specificity. Using a Small Ubiquitin-like Modifier (SUMO) engineered to contain an N-terminal pentaglycine peptide and a C-terminal pilin motif, ^CdSrtA^3M was first used to catalyze the addition of a FITC tag harboring the LPLTG motif to the lysine residue of the pilin motif. ^SaSrtA was then used to conjugate Alexa-fluor 546 harboring the LPATG motif to SUMO via pentaglycine [41]. Because of high degree of specificity for the ε-amine nucleophile within the pilin motif, protein ligation using pilus-specific sortase enzyme may provide selective labeling [41].

Actinomyces fimbriae – a paradigm of tissue tropism, hijacking of pilus machinery and spatial positioning of pili

The heterodimeric fimbriae of Actinomyces oris

Actinomyces are one of the most dominant and earliest colonizing genera of microbes present in the human oral cavity, with Actinomyces oris (formerly called Actinomyces naeslundii) detected in children as young as one year old [42, 43]. A. oris is a major contributor to dental plaque through its ability for coaggregation with other microbial species and thus a key to the genesis of complex biofilms on the surface of teeth and the mucosal epithelia [44–46]. This intrinsic adherence property of A. oris is largely attributed to the presence of two distinct fimbrial types, i.e. type 1 and type 2 fimbriae. A. oris has served as a pioneering model of tissue tropism mediated by Gram-positive pili as type 1 fimbriae mediate bacterial adherence to the salivary proline-rich proteins normally coating the tooth enamel [47], whereas type 2 fimbriae promote bacterial binding to receptor polysaccharides present on the surface of oral streptococci and various host cells [48–51]. Unlike C. diphtheriae and many other Gram-positive bacteria, A. oris fimbriae are heterodimeric, containing a tip component and another entity forming the pilus shaft. In the case of type 1 fimbriae, FimP forms the pilus shaft with FimQ located at the tip, and their assembly requires the pilus-specific sortase SrtC1, whose genes are all tightly linked together on the Actinomyces genome [52]. The specificity of Actinomyces sortases appears to be strict as the pilus-specific sortase SrtC2 is selectively required for formation of type 2 fimbriae only, which consist of the shaft FimA and tip FimB [53]. Since there are only two components in each fimbria, the last subunit of the shaft pilins should be the pilus base. This has raised an intriguing question of how pilus polymerization in Actinomyces or in any other two-component pilus systems, such as B. cereus [12] and Streptococcus suis [54], is terminated.

Using the type 2 fimbriae of A. oris as a prototype, the model of pilus assembly in Actinomyces is illustrated in Figure 2 [55]. Similar to what is described above for C. diphtheriae, pilin precursors translated in the cytoplasm are transported across the cytoplasmic membrane by the Sec translocon, and post-translocational protein folding is mediated by the disulfide bond machine MdbA, permitting membrane insertion of the pilin precursors [56]. When both tip and shaft pilins are available on the membrane, pilus polymerization is catalyzed by the pilus-specific sortase SrtC2 [57]. Finally, cell wall anchoring of the type 2 fimbriae is mediated by the housekeeping sortase SrtA, a class E sortase [58]. Surprisingly, unlike all known sortases studied to date, A. oris srtA is an essential gene. Genetic and biochemical studies to reveal the basis of srtA essentiality (Box 2) have serendipitously uncovered the molecular basis of regulated pilus polymerization and spatial positioning of pilus adhesins in Actinomyces that could not have been envisioned before (see below).

Figure 2: Assembly of the heterodimeric type 2 fimbrae in Actinomyces oris.

(A) The type 2 fimbriae are encoded by the three-gene operon. Genes encoding the coaggregation factor CafA and the housekeeping sortase SrtA are located elsewhere. (B) Similar to the assembly mechanism of the SpaA pili, assembly of the type 2 fimbriae begins with translocation (Step 1) and post-translocational folding of the shaft pilin FimA and tip pilin FimB mediated by the disulfide bond-forming machine MdbA/VKOR coupled to the electron transport chain (ETC) (Step 2). Membrane-insertion (Step 3) of these pilins permits pilus polymerization catalyzed by the pilus-specific sortase SrtC2 (Step 4), followed by cell wall anchoring of the pilus polymer catalyzed by the housekeeping sortase SrtA; reproduced from Sanchez et al. [55].

Box 2: Molecular basis of srtA essentiality in A. oris.

Unlike all other sortases studied to date, A. oris srtA is an essential gene as deletion of srtA has proven to be lethal [58]. A genetic suppressor screen – by Tn5 transposon mutagenesis – subsequently revealed that srtA essentiality is linked to the toxic accrual of a normally cell wall-anchored glycoprotein GspA, a SrtA substrate harboring a cell wall sorting signal with the LAXTG motif. In the absence of srtA, glycosylated GspA accumulates in the cytoplasmic membrane, causing lethal glyco-stress accompanied with expansion of the cell envelope and cell growth arrest [58]. A lcp mutant devoid of the glycosyltransferase LcpA and unable to glycosylate GspA [73] is one of 13 identified suppressor mutants, so is a gspA mutant lacking the cell wall sorting signal that permits membrane insertion prior to cell wall anchoring [58]. This illustrates the power of forward genetic analysis and the continued utility of isolating genetic suppressors in unveiling the intricacies of microbial genetic mechanisms.

The coaggregation factor CafA illustrates pilus hijacking in Gram-positive bacteria

The discovery that CafA is the coaggregation factor in A. oris has several significant implications, one that provides a paradigm of a surface protein hijacking the pilus assembly machine for pilus display and another a concept of spatial positioning of pilus adhesins for biological functions (see below). As mentioned above, type 2 fimbriae are essential for A. oris interactions, or coaggregation, with other oral bacteria, esp. oral streptococci as deletion of fimA abrogates coaggregation with Streptococcus oralis [57]. Surprisingly, A. oris coaggregation could not be blocked by polyclonal antibodies against FimA; nor the process was affected by a deletion of fimB, the gene that encodes the type 2 fimbrial tip pilin FimB [59]. Thus, a hunt was on for discovering the responsible adhesive principle that defied molecular genetic identification strategies taken so far. One potential scenario was that the FimA shaft contains some other protein to mediate co-aggregation. This prompted a systematic elimination of each individual LPXTG-containing surface proteins encoded in the Actinomyces genome. Indeed, among the fourteen candidate surface protein-encoding genes successfully deleted, one displayed a clear-cut co-aggregation defect on its own [59]. The implicated gene product, thus named CafA, was subsequently proved to be the long sought-after coaggregation factor by biochemical experiments: antibodies against CafA captured type 2 fimbriae and blocked bacterial coaggregation. Electron microscopic analyses revealed that CafA localizes at the pilus tip, forming a distinct pilus structure with shaft pilin FimA. Intriguingly, the CWSS sequence of CafA is strikingly similar to that of FimB, leading to the hypothesis that some Gram-positive surface proteins may hijack a pilus assembly machine via molecular mimicry to be displayed at the pilus tip [59]. As significant as this may be for advances in oral bacterial biology and therapeutic intervention, the broader implication of whether the general mechanism of pilus hijacking is more widespread or not still remains to be investigated.

Spatial positioning of pilus adhesins

Over a decade ago, it was speculated that Gram-positive adhesins appended at the pilus tip mediate the initial bacterial encounter with host cells due to the extended nature of pili [3]. Because pilus lengths vary greatly within individual pilus types as well as among various types of pili, it is important to know whether pilus-mediated adhesion processes depend on pilus length and whether and how pilus length is controlled in Gram-positive bacteria. A breakthrough on this problem came from the observation that A. oris mutants lacking the housekeeping sortase srtA produced exceedingly long pili, as might be expected, but surprisingly, they failed to adhere to oral streptococci [58]. This was puzzling since the coaggregation factor CafA was still abundantly detected at the tip of these long pili [60]. Through a series of probing experiments, Chang and colleagues demonstrated that as the pilus length was shortened by inducing expression of srtA, coaggregation could be restored, supporting the notion that the enzymatic activity of the housekeeping sortase SrtA is a key determinant of pilus length modulation. This new insight leads to the important question of whether SrtA activity is subject to regulation functionally or genetically. X-ray crystallization revealed that SrtA harbors two structural elements – a conserved tyrosine residue Y131 and a GVN tripeptide loop that may be of regulatory significance. Indeed, alanine-substitution of Y131 residue resulted in the production of shorter pili and defective coaggregation by A. oris, whereas mutations of the GVN loop led to assembly of extremely long pili and no coaggregation by the mutant bacteria [60], the phenotype similar to srtA depletion [58].

These results led to a mechanistic model that Y131 mutations alter the preference of the class E sortase SrtA, which normally recognizes the LAXTG motif, toward the LPXTG motif present in the FimA pilin subunits. As a result, pilus polymerization is terminated by SrtA-catalyzed cell wall anchoring of the FimA polymer, leading to the display of short pili on cell surface. Conversely, in the case of the GVN motif, its mutations diminished SrtA’s preference for the LAXTG motif, making SrtA’s capacity limited for cell wall anchoring. As a consequence, pilus polymerization continues unperturbed, leading to extremely long pili (Fig. 3). Consistent with this model, the deletion of gspA – which codes for one of the most abundant SrtA substrates with the LAXTG motif – resulted in normal assembly of FimA pili in the GVN mutation background and enabled positive-coaggregation by the mutant strain. Together, these structural genetic findings provide compelling grounds to posit that the housekeeping sortase functions as a molecular ruler for pilus polymerization and as such a positive effector of bacterial coaggregation and virulence.

Figure 3: A model of pilus length modulated by sortase enzymes in Actinomyces oris.

X-ray crystallization revealed the structural features of the housekeeping sortase SrtA: The tyrosine residue Y131 and the tripeptide loop GVN. (A) In the wild-type MG1, balanced activities of the pilus-specific sortase SrtC2 and the housekeeping sortase SrtA produce type 2 fimbriae with typical length. (B) Alanine-substitution of Y131 enhance SrtA sortase activity for the LPLTG motif of FimA, interfering with pilus polymerization, resulting in premature cell wall anchoring of short pili. (C) Mutation or deletion of the GVN loop reduces SrtA preference for the LPLTG motif of FimA, leading to continuous polymerization by SrtC2 and generating exceeding long fimbriae that can be anchored to the cell wall by this pilus-specific sortase enzyme; reproduced from Chang et al [60].

Concluding Remarks

Collective efforts during the last decade dissected the molecular assembly mechanisms of Gram-positive pili and probed their roles in bacterial pathogenesis as well as their utilization in the development of vaccines. While pilus vaccines have yet to emerge in the clinical arena, we now have made great strides in the basic biology and have a clearer view of pilus biogenesis in Gram-positive bacteria. A common feature in these monoderms is the biphasic mode of pilus assembly by distinct steps of enzymatic catalysis involving two sortases, whereby pilus polymerization catalyzed by pilus-specific sortase is followed by cell wall anchoring of pili promoted by the housekeeping sortase. Regardless of the sortase enzymes involved, the basic principle of these transpeptidation reactions in the polymerization phase is the same: The enzymatic cleavage of a substrate and covalent linkage of the cleaved substrate to a nucleophilic acceptor. This transpeptidation reaction generates an isopeptide bond that is mechanically strong and can resist a potential unfolding force up to 690 pN [39]. This unique property of isopeptide bonding, via lysine and threonine residues, is protease resistant and offers a versatile tool in protein engineering and bio-conjugation [41].

Where do we go from here? In spite of differences with their Gram-negative counterparts in the manner of assembly, the heteromeric pili of Gram-positive bacteria play significant roles in bacterial physiology and virulence as those of Gram-negative bacteria. Today, however, many fundamental questions regarding this still await to be addressed (Box 3). Given the importance of these questions, and the genetic and biochemical versatility of C. diphtheriae and A. oris as model organisms and their importance in significant human conditions, we believe these two systems will continue as fertile and attractive experimental models of pilus biogenesis in Gram-positive bacteria for some time to come.

Box 3: Outstanding Questions.

• In the Actinobacteria C. diphtheriae and A. oris, the membrane-bound disulfide bond-forming machine MdbA promotes post-translocational folding of pilins. Given that no major protein folding machines are linked to pilus assembly in Firmicutes, how do these organisms solve the protein-folding problem in pilus assembly?

• In a heterodimeric pilus system like A. oris, the last shaft pilin acts as pilus base and stop signal, and the housekeeping sortase appears to control pilus polymerization, hence pilus length. How does the housekeeping sortase indiscriminately recognize this base pilin from the rest? This raises an intriguing possibility that pilus polymerization and termination may depend on the stoichiometric availability of pilin substrates and sortase enzymes at the pilusosome. If so, does this require additional factors?

• How are the tip pilins FimB and CafA in A. oris involved in pilus assembly? What mechanisms govern how a tip pilin is spatiotemporally incorporated only the tip of pili?

• How do surface proteins with the LPXTG motif like CafA in A. oris hijack the sortase machine to be incorporated into the pilus tip?

• Pili are found in the culture medium, esp. in late log-phase and stationary phase. What are the roles of secreted pili or are they products of cell wall turnover?

Highlights.

• Covalently-linked pili are assembled on the cell surface of many Gram-positive bacteria via a biphasic mechanism, whereby pilus polymerization is catalyzed by the pilus-specific sortase followed by cell wall anchoring of pili by the housekeeping sortase.

• Pilus-mediated adhesion depend on pilus length, which is modulated by the housekeeping sortase via unique structural features.

• Some Gram-positive surface proteins with the LPXTG motif may hijack a pilus assembly machine via molecular mimicry to be displayed at the pilus tip.

• Pilus-specific sortase enzymes provide a bioconjugation tool via the formation of an isopeptide bond that is mechanically stable and less susceptible to proteolytic cleavage.