Heterologous assembly of type IV pili by a type II secretion system reveals the role of minor pilins in assembly initiation


  • David A. Cisneros,

    1. Molecular Genetics Unit, Department of Microbiology, Institut Pasteur, Paris, France
    2. CNRS ERL3526, Paris, France
    Current affiliation:
    1. Research Institute of Molecular Pathology (IMP), Vienna, Austria
    2. Max F. Perutz Laboratories (MFPL), University of Vienna, Vienna, Austria
    Search for more papers by this author
  • Gerard Pehau-Arnaudet,

    1. Plate-Forme de Microscopie Ultrastructurale, Department of Structural Biology, Institut Pasteur, Paris, France
    2. CNRS UMR3828, Paris, France
    Search for more papers by this author
  • Olivera Francetic

    Corresponding author
    1. CNRS ERL3526, Paris, France
    • Molecular Genetics Unit, Department of Microbiology, Institut Pasteur, Paris, France
    Search for more papers by this author

For correspondence. E-mail cisneros@imp.ac.at; ofrancet@pasteur.fr; Tel. (+33) 140613681; Fax (+33) 145688960.


In Gram-negative bacteria, type IV pilus assembly (T4PS) and type II secretion (T2SS) systems polymerize inner membrane proteins called major pilins or pseudopilins respectively, into thin filaments. Four minor pilins are required in both systems for efficient fibre assembly. Escherichia coli K-12 has a set of T4PS assembly genes that are silent under standard growth conditions. We studied the heterologous assembly of the E. coli type IV pilin PpdD by the Klebsiella oxytoca T2SS called the Pul system. PpdD pilus assembly in this context depended on the expression of the K. oxytoca minor pseudopilin genes pulHIJK or of the E. coli minor pilin genes ppdAB-ygdB-ppdC. The E. coli minor pilins restored assembly of the major pseudopilin PulG in a pulHIJK mutant, but not the secretion of the T2SS substrate pullulanase. Thus, minor pilins and minor pseudopilins are functionally interchangeable in initiating major pilin assembly, further extending the fundamental similarities between the two systems. The data suggest that, in both systems, minor pilins activate the assembly machinery through a common self-assembly mechanism. When produced together, PulG and PpdD assembled into distinct homopolymers, establishing major pilins as key determinants of pilus elongation and structure.


Type IV pili (T4P) are dynamic filaments found on the surface of many bacteria and archaea, and are involved in adhesion, motility, signalling and macromolecular transport (Pelicic, 2008). The type IVa subclass of T4P assembly systems is very closely related to type II protein secretion systems (T2SS) of Gram-negative bacteria, which translocate folded proteins from the periplasm across the outer membrane. Like the T4PS, T2SSs are implicated in virulence of many bacteria, as they promote secretion of toxins and hydrolases (Douzi et al., 2012). The T2SS and T4PS are similar in both composition and structure (Hobbs and Mattick, 1993; Hazes and Forst, 2008; Korotkov et al., 2012; McLaughlin et al., 2012). The assembled fibres of the two systems are mainly composed of abundant inner membrane (IM) proteins termed major pilins (or pseudopilins in the T2SS) (Hansen and Forest, 2006). Although T2SS do not assemble surface pili under physiological conditions, they are thought to assemble periplasmic fibres, called pseudopili, which are essential for protein secretion. In support of this model, the T2SS of Klebsiella oxytoca, the Pul secreton, assembles surface filaments composed of the major pseudopilin PulG when the T2SS genes are expressed from a moderate copy plasmid (Sauvonnet et al., 2000b). This feature is generally conserved in T2SSs, including Xcp in Pseudomonas aeruginosa (Durand et al., 2003) and Gsp system in Escherichia coli (Vignon et al., 2003), with some exceptions (Durand et al., 2011). However, the exact composition of periplasmic pseudopili is unknown.

Pilins are synthesized as precursors with a positively charged propeptide that is immediately followed by a highly conserved transmembrane (TM) segment. The short propeptide is processed on the cytoplasmic side of the TM segment by the prepilin peptidase/N-methylase, which is functionally interchangeable between the T4PS and T2SS (Nunn and Lory, 1992). An ATPase of the PilB/PulE family and the associated IM protein of the PulF/PilC family are the core components of the machinery that catalyses pilin assembly into fibres. The T2SS IM assembly platform also comprises PulL and PulM (Py et al., 2001), which are structural homologues of the T4PS PilM–PilN–PilO proteins of P. aeruginosa (Sampaleanu et al., 2009; Ayers et al., 2010). Homologues of PulC (in T2SS) and PilP proteins (in T4PS) provide a direct link between the assembly platform in the IM and the secretin channel formed by PulD and PilQ homologues in the outer membrane (Ayers et al., 2009; Korotkov et al., 2011).

In addition to fibre-building major pilins, four minor pilins have been implicated in the efficient assembly and function of T4P (Winther-Larsen et al., 2005; Carbonnelle et al., 2006; Giltner et al., 2010). In the presence of the retraction ATPase PilT, minor pilin mutants are defective for T4P assembly. Their homologues in T2SS, the minor pseudopilins, have been well studied at the structural level. Three of them, GspI–GspJ–GspK, form a complex that adopts a pilus-like structure capped by the alpha-helical domain of GspK, which prevents upward elongation, therefore suggesting their localization at the tip of the pseudopilus (Korotkov and Hol, 2008). Self-assembly of their homologues, PulI, PulJ and PulK, into a staggered complex in the IM is required for initiation of PulG pilus assembly (Cisneros et al., 2012).

Type IV pilus assembly genes have been identified in E. coli K-12 and a wide range of enterobacteria including Yersinia, Citrobacter and Klebsiella. The ppdD gene encodes the major pilin (Hobbs and Mattick, 1993; Whitchurch and Mattick, 1994) and a set of genes is organized in an operon (ppdA-ppdB-ygdB-ppdC) that could encode the minor pilins (Fig. 1). Although PpdD pilus assembly was not observed in E. coli K-12 (Sauvonnet et al., 2000a), PpdD pili are assembled under starvation conditions in E. coli O157:H7 (Xicohtencatl-Cortes et al., 2007) and in E. coli HB101 (Xicohtencatl-Cortes et al., 2009). The E. coli T4P assembly genes have been implicated in the uptake of DNA and its use as a carbon source (Palchevskiy and Finkel, 2006). They are co-regulated with a competence system-encoding genes via a global regulator YccR (Sxy) (Cameron and Redfield, 2008; Sinha et al., 2009). In addition, PpdD is required for DNA transformation in a semi-natural competence assay in E. coli K-12 (Sinha et al., 2009).

Figure 1.

A. Organization of the E. coli K-12 minor T4 pilin genes and of the K. oxytoca T2SS minor pseudopilin genes on the chromosome.

B. Prepilin propeptide and transmembrane segment sequence alignment of the K. oxytoca T2SS pseudopilins (PulG to PulK) and E. coli K-12 T4 pilins (PpdD to PpdC). The prepilin peptidase cleavage site is after the conserved Gly(-1) residue. Alignment was done in ClustalX and the colour code depends on both residue type and the pattern of conservation within a column as designated by the ClustalX program (Thompson et al., 1997). Major pseudopilin PulG and major pilin PpdD share a conserved Pro22 residue. Detailed comparison of these N-terminal segments, with features suggesting analogous order of assembly for (pseudo)pilin pairs PulH–PpdB, PulI–PpdA, PulJ–PpdC and PulK–YgdB, is provided in the Discussion.

The Pul T2SS reconstituted in E. coli is able to assemble PpdD into pili (Sauvonnet et al., 2000b; Kohler et al., 2004). Here we studied the respective roles of the T2SS and T4PS minor pilins in the assembly of PpdD and PulG and their ability to promote secretion of the T2SS exoprotein pullulanase.


Heterologous assembly of PpdD into pili requires the Pul minor pseudopilins

Escherichia coli carrying all K. oxytoca T2SS pul genes in the pBR322 derivative pCHAP231 can assemble the major type IV pilin PpdD into pili (Sauvonnet et al., 2000b). We showed recently that minor pseudopilins are required to initiate the assembly of the major pseudopilin PulG (Cisneros et al., 2012). To test whether minor pseudopilins are required for the heterologous assembly of PpdD by the Pul T2SS, we expressed the E. coli ppdD gene (pCHAP3100) in the presence of the complete set of pul genes encoding the wild-type (WT) Pul secreton (plasmid pCHAP231), or its ΔpulGHIJK derivative, lacking all five pseudopilin genes (plasmid pCHAP7248). Bacteria were grown overnight on maltose-containing plates to induce expression of the MalT-dependent pul genes (d'Enfert et al., 1987) and PpdD pilus assembly was assayed by shearing and fractionation (see Experimental procedures), followed by immunoblot using anti-PpdD antibodies (Fig. 2A). We detected PpdD in the sheared fractions from bacteria expressing all pul genes (Fig. 2A, lane 4), while PpdD pilus assembly was not observed in the ΔpulGHIJK mutant (Fig. 2A, lane 6). Deleting the minor pseudopilin genes pulHIJK (encoded by pCHAP8296) (Fig. 2A, lanes 7 and 8) abolished PpdD pilus assembly, while the pulG deletion (carried by plasmid pCHAP1216) (Fig. 2A, lanes 9 and 10) was without effect. PpdD antibodies did not cross-react with PulG, as shown using as a negative control empty vector pSU18 instead of plasmid pCHAP3100 (Fig. 2A, lane 2). PpdD pilus assembly did not interfere with PulG pilus assembly in bacteria expressing all pul genes, as shown by the simultaneous presence of PulG and PpdD in the sheared fraction (Fig. 2A, lanes 2 and 4). All these results show that the minor pseudopilins of the Pul system are necessary for the heterologous assembly of the PpdD pilus.

Figure 2.

Heterologous assembly of PpdD pili depends on minor (pseudo)pilins.

A. PpdD or PulG immuno-detection of total cell extracts and sheared fractions (C, SF) of E. coli co-transformed with either plasmid pCHAP3100 (ppdD) or the empty vector pSU18 (labelled ‘−’) and one of the following plasmids: pCHAP231 (WT, all pul genes) or pCHAP7248 (ΔpulGHIJK) or pCHAP8296 (ΔpulHIJK) or pCHAP1216 (ΔpulG). Antibodies against the outer membrane protein LamB were used as a control for outer membrane contamination of the sheared fraction.

B. PpdD immuno-detection of E. coli coexpressing, where indicated, the full set of pul genes (WT, plasmid pCHAP231) or ΔpulGHIJK (pCHAP7248), ppdD (plasmid pCHAP3100) and either T4 minor pilin (ppdAB-ygdB-ppdC), or the pul minor pseudopilin (pulHIJK), or all pul pseudopilin (pulGHIJK) genes, or empty vector (pCHAP418zeo).

C. As in (B) but immuno-detection with anti-PulG antibodies.

E. coli minor pilins promote heterologous assembly of PpdD into pili

Minor pilins are required for efficient assembly of the major pilin into filaments in several T4P assembly systems (Alm and Mattick, 1995; Alm et al., 1996; Winther-Larsen et al., 2005; Carbonnelle et al., 2006; Giltner et al., 2010). Therefore, we tested whether the putative minor pilin operon in the thyA-recC locus in E. coli, namely ppdA-ppdB-ygdB-ppdC, would influence the heterologous PpdD pilus assembly by the Pul T2SS. For this purpose, we compared the effect of minor pilin genes ppdAB-ygdB-ppdC (cloned in plasmid pCHAP6116) with that of minor pseudopilin genes pulHIJK (expressed from plasmid pCHAP7292) or all pseudopilin genes pulGHIJK (expressed from plasmid pCHAP8254) on PpdD pilus assembly (expressed from plasmid pCHAP3100). As a host we used E. coli expressing the pul system devoid of all pseudopilin genes (plasmid pCHAP7248). ppdAB-ygdB-ppdC expression permitted PpdD pilus assembly, as demonstrated by the presence of PpdD in the sheared fraction of cells lacking PulHIJK (Fig. 2B, lane 2). PpdD pili were not detected in bacteria carrying the empty vector pCHAP418zeo (Fig. 2B, lane 4). The efficiency of PpdD assembly promoted by the E. coli minor pilins was comparable to that promoted by the minor pseudopilins PulHIJK (Fig. 2B, lane 8). Coexpression of pulG in the same cells led to a 50% reduction of PpdD pilus assembly as measured by densitometric analysis of the Western blot (Fig. 2B, compare lanes 6 and 8). In the context of the WT pul locus expressed from plasmid pCHAP231, coexpression of E. coli minor pilin genes appeared to stimulate PpdD pilus assembly by about 50% (Fig. 2B, compare lanes 12 and 14). Although more quantitative studies will be required to confirm these results, they could suggest that the minor (pseudo)pilins rather than the Pul assembly machinery were limiting for pilus assembly.

We next assessed the effect of E. coli minor pilins on PulG pilus assembly in E. coli carrying plasmid pCHAP231 (complete set of pul genes). In this context both PpdD and PulG pili could be assembled, regardless of the presence or absence of the minor type IV pilins (Fig. 2B, lanes 11–14 and Fig. 2C, lanes 1–4). The E. coli minor pilins appeared to stimulate PulG pilus assembly in the absence of their cognate major pilin PpdD (Fig. 2C, lanes 5–8). Furthermore, PpdD production did not affect the assembly of PulG pili in the ΔpulGHIJK mutant (plasmid pCHAP7248) complemented by pulGHIJK (plasmid pCHAP8254) (Fig. 2C, lanes 9–12). Taken together, these results show that the E. coli minor pilins promote PpdD pilus assembly in the heterologous system, suggesting that they are functional as minor pilins in their native E. coli host. They stimulated the PpdD pilus assembly with the efficiency similar to that of the minor pseudopilins PulHIJK. This further suggests that T4PS minor pilins have a role similar to that of the T2SS minor pseudopilins in initiating pilus assembly.

PulG and PpdD assemble into separate and morphologically distinct pili by the K. oxytoca T2SS

To gain further insight into PulG and PpdD pilus assembly, we examined the same E. coli strains analysed in Fig. 2 by immunofluorescence microscopy. In E. coli cells carrying either plasmid pCHAP231 (all pul genes) or pCHAP7248 (ΔpulGHIJK) we coexpressed ppdD and different sets of (pseudo)pilin genes, as indicated in each panel (Fig. 3). Using anti-PpdD antibodies we observed long filaments in ΔpulGHIJK bacteria (plasmid pCHAP7248) coexpressing ppdD (pCHAP3100) and ppdAB-ygdB-ppdC (pCHAP6116) (Fig. 3A), while bacteria carrying the empty vector instead of the T4 minor pilin genes did not assemble PpdD pili (Fig. 3B). PpdD pili assembled in the presence of the five Pul pseudopilins (Fig. 3C and D) looked very similar to PpdD pili assembled in the presence of the E. coli T4 minor pilins (Fig. 3C). Consistent with the shearing assay results, minor pseudopilins were sufficient to promote PpdD assembly into pili (Fig. 3D), although fewer pili were assembled by the complete Pul T2SS (Fig. 3F). Coexpression of E. coli ppdAB-ygdB-ppdC genes further stimulated PpdD pilus assembly (compare Fig. 3E and F). PpdD pili were not observed in E. coli carrying either all pul genes or ΔpulGHIJK (plasmids pCHAP231 or pCHAP7248 respectively) and minor pilin genes in the absence of PpdD (Fig. 3G–I), confirming the specificity of PpdD antibodies. Overall, PpdD filaments were long and thin, irrespective of the nature of the minor (pseudo)pilins driving their assembly.

Figure 3.

PpdD and PulG pili are morphologically distinct. Phase contrast (left) and immunofluorescence (right) images of E. coli with plasmids carrying the indicated genes from the same bacteria analysed in Fig. 2. Immunofluorescence was performed using anti-PpdD (A–I) and anti-PulG (J–O) antibodies. Scale bar corresponds to 10 μm. Insets show twofold magnification of selected fields.

We next analysed these bacteria by immunofluorescence using anti-PulG antibodies. In bacteria producing PulG and PpdD pili at the same time (i.e. E. coli carrying plasmids pCHAP231, pCHAP3100 and pCHAP6116, Fig. 3J) PulG filaments appeared shorter, slightly thicker and less entangled than PpdD pili. The presence of PpdD or the minor T4 pilins did not affect the morphology of PulG pili (Fig. 3K–M) and the anti-PulG antibodies did not react with PpdD pili (Fig. 3N and O). All of these results show that PulG pili and PpdD pili are morphologically distinct and confirm that the PpdD signal observed in the sheared fractions by immune-detection (Fig. 2B) corresponds to PpdD filaments.

To determine whether individual or mixed PpdD and PulG fibres were assembled when both major pilins were present, we analysed E. coli assembling PpdD using the Pul T2SS by electron microscopy and immunogold labelling. For this purpose, PulG and PpdD pili were labelled with gold particles of two different sizes (Fig. 4). We mainly observed filaments labelled with one size of gold particle, i.e. 5 nm gold for PpdD (Fig. 4A, black arrows) or 15 nm gold for PulG filaments (Fig. 4A, white arrows). We occasionally observed 15 nm gold particles (i.e. PulG) on PpdD pili (Fig. 4A and B), but these cases appeared to coincide with lateral contacts between distinct homopolymers. We rarely found 5 nm gold particles on PulG pili (Fig. 4C). Together, these results suggest that the Pul minor pseudopilins can independently initiate both PpdD and PulG pilus assembly and that once the initiation step has occurred, elongation proceeded by the addition of predominantly one type of major pilin.

Figure 4.

PulG and PpdD pili assemble into distinct filaments. Immunogold labelling of E. coli co-transformed with plasmid pCHAP7248 (ΔpulGHIJK), plasmid pCHAP3100 (ppdD) and a pCHAP8254 carrying all pul pseudopilin genes (pulGHIJK). PulG was labelled with anti-PulG antibodies, followed by Protein A Gold (15 nm). PpdD was labelled with anti-PpdD antibodies followed by Protein A Gold (5 nm), as described in Experimental procedures.

A, D, E. E. coli showing two distinct immunogold labelled pili composed of PulG and PpdD.

B. Isolated PpdD or PulG (C, F) pili.

Scale bars, 500 nm (A, F), 200 nm (B–E). Insets show higher magnification of selected area.

PpdD pilus cannot promote pullulanase secretion

We next tested if PpdD pilus assembly in presence of PpdA–PpdB–YgdB–PpdC could promote secretion of pullulanase, the K. oxytoca T2SS substrate. Since PulA is a lipoprotein, we assessed its surface exposure (Possot et al., 2000) in E. coli expressing all pul genes except pulGHIJK (plasmid pCHAP7248), together with ppdD and either the E. coli minor pilins or the K. oxytoca Pul pseudopilins. Neither the E. coli K-12 minor pilins (PpdAB–YgdB–PpdC) nor the Pul minor pseudopilins (PulHIJK) could support PulA secretion (Fig. 5). Nevertheless, all Pul pseudopilins produced from plasmid pCHAP8254 supported full PulA secretion, indicating that the pseudopilin gene deletion could be complemented regardless of the presence of PpdD (Fig. 5). In addition, the E. coli major or minor pilins did not interfere with pullulanase secretion promoted by the WT pul gene expression from plasmid pCHAP231 (Fig. 5).

Figure 5.

PpdD pili do not promote pullulanase secretion. Percentage of pullulanase activity in whole versus lysed E. coli transformed with either plasmid pCHAP231 (WT pul genes) or pCHAP7248 (ΔpulGHIJK), plasmid pCHAP3100 (ppdD) or empty vector (pSU18, labelled ‘−’) and either empty vector (pCHAP418zeo, labelled ‘−’) or its derivatives carrying T4 minor pilin (ppdAB-ygdB-ppdC), the T2SS minor pseudopilin (pulHIJK) or all T2SS pseudopilin (pulGHIJK) genes.

Minor type IV pilins PpdA, PpdB, YgdB and PpdC can initiate PulG pilus assembly

We have shown recently that initiation of PulG pilus assembly in T2SS depends on the self-assembly of the minor pseudopilins (Cisneros et al., 2012). We assessed PulG pilus assembly in bacteria expressing all pul genes except pulHIJK (from plasmid pCHAP8296) to determine whether it could be initiated by the E. coli T4 minor pilins. Both cognate minor pseudopilin genes pulHIJK (from plasmid pCHAP7292) and the E. coli genes ppdAB-ygdB-ppdC (from plasmid pCHAP6116) restored PulG pilus assembly, as assessed by shearing and immunodetection (Fig. 6A). PpdD pilus assembly was not observed in bacteria carrying the empty vector (plasmid pCHAP418zeo) (Fig. 6A). These results were further confirmed by immunofluorescence using anti-PulG antibodies (Fig. 6B). However, samples complemented by the T4 minor pilins promoted assembly of a reduced number of PulG pili (middle panel) compared with PulHIJK minor pseudopilins (top panel) by about 80% as shown by the densitometric analysis of the Western blot (see also Fig. S1). Furthermore, the E. coli minor pilins did not promote pullulanase secretion in E. coli strain carrying plasmid pCHAP8432 (the pCHAP8296 derivative encoding a non-acylated variant of PulA) (Fig. 6C). These results show that minor T4 pilins can initiate PulG pilus assembly, although the resulting fibres are not functional in protein secretion.

Figure 6.

The E. coli minor pilins partially substitute for T2SS minor pseudopilins.

A. PulG immuno-detection in cell and sheared fractions (C, SF) of E. coli PAP7460 (pCHAP8296) grown for 16 h on LB agar containing 0.4% maltose and appropriate antibiotics, carrying either pCHAP418zeo (empty vector) or its derivatives pCHAP6116 carrying T4 minor pilin (ppdAB-ygdB-ppdC) or pCHAP7292 carrying minor pseudopilin genes (pulHIJK).

B. Phase contrast (left panels) and immunofluorescence using anti-PulG antibodies (right panels) of E. coli strains as in (A). Scale bar corresponds to 10 μm.

C. Secretion of non-acylated PulA variant in E. coli PAP7460 strain expressing pul genes from plasmid pCHAP8432 (ΔpulHIJK) complemented with pulHIJK (pCHAP8509), ppdAB-ygdB-ppdC (pCHAP6116) or empty vector (pCHAP419zeo).


The role of minor T4 pilins in activation of fibre assembly

In this study, we show that efficient assembly of E. coli type IV pilin PpdD can be promoted by the minor pseudopilins of the pullulanase T2SS. In the absence of minor pseudopilins, PpdD pilus assembly was promoted by the minor T4 pilins encoded by the ppdAB-ygdB-ppdC operon of E. coli (Fig. 1), suggesting that these proteins function as minor pilins in their native system. Their functional role remains to be confirmed in E. coli B or EHEC strains, in which PpdD is assembled into bona fide pili involved in adhesion, motility and signalling (Xicohtencatl-Cortes et al., 2007; 2009). Although silent in E. coli K-12 (Sauvonnet et al., 2000a), the four minor pilin genes are significantly induced (nine- to 100-fold) by Sxy (YccR), the global activator of competence genes in E. coli (Sinha et al., 2009), strongly suggesting their involvement in natural transformation. Interestingly, ppdC mutants have been isolated in a screen for defective swarming motility, although it is not known whether T4P are involved in this phenomenon (Inoue et al., 2007).

A set of four minor pilin genes has been implicated in T4P assembly in several well-studied systems, in which their deletion results in piliation defects (Alm and Mattick, 1995; Alm et al., 1996; Winther-Larsen et al., 2005; Carbonnelle et al., 2006; Giltner et al., 2010). These genes include pilHIJK in Neisseria meningitidis and N. gonorrhoeae and the fimU-pilVWX genes in P. aeruginosa. All these studies suggest that minor pilins favour efficient pilus assembly in the presence of retraction ATPase and it has been suggested that they counteract retraction (Carbonnelle et al., 2006; Pelicic, 2008). In Neisseria spp., deletion of the gene encoding the retraction ATPase PilT restores piliation in minor pilin mutants, although such double mutants show defective adherence to epithelial cells and abiotic surfaces (Winther-Larsen et al., 2005; Carbonnelle et al., 2006). In P. aeruginosa, pilus assembly is only partially restored in pilT/minor pilin gene double mutants, suggesting a defect in pilus assembly (Giltner et al., 2010). Using the T2SS as a heterologous assembly system we show here that both the E. coli T4 minor pilins and the T2S minor pseudopilins are functional in promoting assembly of T4P and T2S major pilins, suggesting a similar role and mechanism of action for both sets of proteins. Our results favour a model where T4 minor pilins act by promoting efficient initiation of pilus assembly allowing the assembly reaction to move in the forward direction.

As mentioned above, assembly of T4P in Neisseria spp. takes place in the absence of minor pilins in pilT mutants, which appears to contradict this model (Winther-Larsen et al., 2005; Carbonnelle et al., 2006). Although in T2SS minor pseudopilins are required for efficient pseudopilus assembly initiation, we showed that a very low basal level of PulG pilus assembly is observed even in the absence of minor pseudopilins (Cisneros et al., 2012). These ‘spontaneous initiation’ events may coincide with sporadic spontaneous activation of the assembly ATPase. The few pili assembled in these cases are typically very long, presumably due to the accumulation of pilins in the membrane, which drives the elongation of rare active assembly reactions forward (Cisneros et al., 2012). In T4P, the expression of major pilins is constitutive and their production levels and stability are much higher than in T2SS. In this context, inefficient piliation in the absence of minor pilins and PilT could still lead to accumulation of few long pili, masking the assembly defect. A similar accumulation of PulG pili can be observed over time in minor pseudopilin mutants (Fig. S2). In the presence of retraction forces that allow to reset the system, this assembly defect would become apparent, as it occurs for example in P. aeruginosa (Giltner et al., 2010). Therefore, it would be interesting to test this model by placing major pilins under control of an inducible promoter and performing the kinetic analysis of piliation, comparing the number and length of assembled pili in WT and minor pilin/pilT double mutants.

In addition to the cognate minor pilins, we show that minor pseudopilins PulHIJK are also capable of initiating PpdD pilus assembly. Furthermore, E. coli minor T4 pilins can initiate PulG pilus assembly, suggesting that they function by a similar mechanism, despite the lack of sequence similarity in their globular domains. So far, structural information is not available for minor T4 pilins of this class, but such data are available for the T2SS minor pseudopilins, whose biochemical and structural analysis indicates high affinity interactions between their globular domains. For example, globular domains of minor pseudopilins EpsI and EpsJ from Vibrio vulnificus could only be purified and crystallized after co-production, suggesting that they stabilize each other and possibly assist each other's folding (Yanez et al., 2008; Lam et al., 2009). Globular domains of the GspJ–GspI–GspK formed a quasi-helical hetero-trimer that may be localized at the tip of the pseudopilus (Korotkov and Hol, 2008). In vitro analysis of the P. aeruginosa Xcp T2SS showed that periplasmic domains of all four minor pseudopilins form a complex (Douzi et al., 2009). However, the major pseudopilin XcpT did not bind to this quaternary complex, suggesting that structural compatibility between major and minor pilins does not account for the assembly mechanism, consistent with the absolute requirement of the assembly platform components for the pseudopilus elongation (Sauvonnet et al., 2000b and Fig. S3).

How do these two different sets of pilins promote fibre assembly by T4 pilin PpdD and major pseudopilin PulG? We recently showed that the minor pseudopilins in T2SS initiate PulG polymerization by a mechanism involving the self-assembly of PulI, PulJ and PulK (Cisneros et al., 2012). In their native, membrane environment, formation of the minor pseudopilin complex, driven by the affinity of their globular domains, promotes the upward movement of their TM segments (Cisneros et al., 2012). This series of binding events between PulJ, PulI and PulK results in the formation of a pseudopilus-like structure on the periplasmic side of the IM, where each subunit is shifted with respect to its nearest neighbours by 1 nm (Fig. 7). The extent of this shift, demonstrated by position-specific cysteine cross-linking, corresponds to the axial rise between the major pilins in the assembled pseudopilus. We have proposed that the conformational changes during initiation complex formation activate the IM assembly machinery. Membrane extraction of PulK induced by its binding to the PulJ–PulI complex could transduce a signal to the assembly ATPase and pull it close to the membrane to activate ATP hydrolysis (Camberg et al., 2007; Cisneros et al., 2012). The capacity of minor T4 pilins to promote heterologous assembly of PulG, and the reciprocal induction of PpdD pilus assembly by the minor pseudopilins, lend further support to a model whereby minor (pseudo)pilin complex initiates assembly by triggering the elongation competent state of the IM assembly platform.

Figure 7.

Model of (pseudo)pilus assembly.

Upper panel: 1. Minor pseudopilins self-assemble at the inner membrane into a pseudopilus-like structure without external energy (Cisneros et al., 2012). 2. Minor (pseudo)pilins transduce the signal to the assembly ATPase PulE (in red), priming it for major pilin elongation. 3. The active assembly ATPase catalyses the processive (pseudo)pilus elongation adding minor pseudopilin(s) PulH and multiple major subunits PulG coupled to ATP hydrolysis.

Lower panel: The analogous role is proposed for the E. coli minor T4 pilins YgdB-PpdA-PpdC in self-assembly (1), activating the assembly machinery (2) to promote addition of PpdB and PpdD in the elongation step (3). The assembly platform (AP) complex comprises PulF, PulL, PulM and PulC proteins.

An alternative model would be that the self-assembly of the minor pseudopilins simply leads to the scaffolding of the major pseudopilin, which would require structural recognition between the minor pseudopilin globular domains and the major pilin. However, there is little sequence similarity between globular domains of the minor pilins and pseudopilins (Fig. S4). Therefore, it is reasonable to assume that there is no high-affinity interaction between PulG and T4 minor pilins or PpdD and T2S minor pseudopilins. Yet, both sets of minor pilins are functional for initiation of pilus assembly suggesting that they could exert their effect not on major pilins directly but on the assembly machinery. According to this model, initiation and elongation, at least in T2SS, are distinct processes, which are coupled at the level of the assembly platform. Therefore, it is tempting to speculate that the E. coli T4 minor pilins also undergo self-assembly that would promote PulG pilus assembly by the same mechanism. In support of this model, homologues of PulH, which are thought to structurally link the minor pseudopilin tip (Korotkov and Hol, 2008) to the major pilin (Douzi et al., 2009), are dispensable for pilus assembly (Sauvonnet et al., 2000b; Cisneros et al., 2012).

Unlike the globular domains of minor pseudopilins and minor T4 pilins, which are very different (Fig. S4), the TM segments of these proteins share certain conserved features (Fig. 1). All minor pseudopilins except one (PulK and its homologues) contain the conserved residue E5, which has been implicated in the interactions between neighbouring subunits, neutralizing the charge of the free N-terminal amine of the pseudopilin assembled upstream. The E5 residue would be dispensable in the tip subunit, and its absence is indeed conserved in all PulK homologues of T2SS, as well as in minor T4 pilins PilK of Neisseria spp. (Winther-Larsen et al., 2005), PilX in P. aeruginosa (Giltner et al., 2010) and YgdB of E. coli. In addition, in both PulK and YgdB, the more hydrophobic residues Ile or Val respectively, replace the N-terminal Phe residue. This feature could facilitate the extraction of these pilins from the membrane, which has been observed for PulK in silico by molecular dynamics simulations (Cisneros et al., 2012). The same feature is shared by PpdA and the central pseudopilin PulI, where the N-terminal Phe has been substituted by Tyr and Met residues respectively, again consistent with a partial membrane extraction of the central pilins in the tripartite ‘tip complex’. The minor pilin PpdB shares a conserved residue Pro25 with the pseudopilin PulH, which appears to be involved in a post-initiation step of assembly (Douzi et al., 2009). Based on these primary sequence features, it is tempting to speculate that the T4PS minor pilins PpdC, PpdA and YgdB form a tip complex that may nucleate pilus assembly (Fig. 7). Their globular domains are therefore predicted to interact strongly and sequentially to promote the conformational changes analogous to those that take place in the T2SS (Korotkov and Hol, 2008; Cisneros et al., 2012). In support of this model, N. gonorrhoeae minor pilins PilI and PilJ depend on each other for stability (Winther-Larsen et al., 2005), similar to their pseudopilin orthologues PulI and PulJ, and the P. aeruginosa minor pilins PilV and PilW follow the same pattern (Giltner et al., 2010). Interestingly, the order of minor pilin genes in the E. coli cluster is slightly different from that in Neisseria spp. or P. aeruginosa. In particular, in E. coli, it is the third gene in the cluster, ygdB and not the fourth, that encodes the minor pilin lacking the E5 residue. However, the same gene organization and linkage to recC is conserved in many species of γ-proteobacteria (Fig. S5).

Although minor T4 pilins have been proposed to promote pilus assembly initiation, the notion that they are localized at the pilus tip has been recently challenged (Giltner et al., 2010). On the other hand, distinct structures have been observed at the tip of gonococcal pili, and their size suggests that they may be composed of more than three protein subunits (Winther-Larsen et al., 2007). Clearly, further biochemical and structural analysis of minor T4 pilins will be necessary to test their assembly and function.

Assembly of T4 pili by the T2SS

It is interesting that the major subunits of the T2SS and T4PS, PulG and PpdD, are efficiently recognized and assembled by the same machinery. PpdD is an exception among T4 pilins in this regard, since PilE from Neisseria or PilA from P. aeruginosa could not be assembled in this context (Kohler et al., 2004). Homology between PulG and PpdD, which have a very similar size, is limited to their TM segments, which are 50% identical, and 85% similar or identical, while sequence identity of their periplasmic domains is only 13%. The highly conserved TM segments of major pilins are therefore likely to play a major role in assembly by interacting with the prepilin peptidase and with the components of the assembly platform. This is supported by the fact that PulL, PulE and PulF are essential for both PulG and PpdD assembly (Sauvonnet et al., 2000a). However, we provide evidence that pilus assembly also requires structural compatibility between pilins. When both PpdD and PulG pili were assembled in the same bacteria, we observed defined filaments composed of one or the other major pilin by electron microscopy. Similar results were previously observed in P. aeruginosa, where PAO strains expressing PAK pilA assemble only homopolymers of PAO or PAK pilins (Pasloske et al., 1989). Occasionally, some PulG subunits could be found in PpdD filaments, although in most cases this coincided with points of contact between distinct fibres. Nevertheless, most of the filaments are mainly constituted of one major pilin type. Consistent with this, PpdD could not be co-purified with PulG-His pili by affinity chromatography (Kohler et al., 2004). Interestingly, the same study has reported that, in the presence of the PulG : His6 variant, the Pul secretion system did not assemble PpdD as pili but rather as small aggregates (Kohler et al., 2004). Constitutive coexpression of pilin genes used in that study could have led PulG : His6 to outcompete PpdD for the assembly machinery. Alternatively, the two types of pili may have behaved differently in EM experiments. In the present study we used treatment of carbon grids with CaCl2 to improve pilus adsorption during EM sample preparation (see Experimental procedures). The formation of pilin homopolymers suggests that after the initiation step, filament elongation involves structural recognition between the incoming major pilin and the growing filament. This is consistent with the results of the structure–function analysis of PulG pili, which identified two specific and highly conserved salt bridge interactions between neighbouring protomers, both essential for pseudopilus assembly and for protein secretion (Campos et al., 2010).

Functional implications for protein secretion mechanism

One possible model for T2SS pseudopilus assembly is that the minor pseudopilins occupy the assembly platform to initiate the fibre elongation. However, the co-production of PpdD and the cognate minor pilins with the Pul system did not interfere with pullulanase secretion. This observation suggests that the self-assembly of minor (pseudo)pilins and their interactions with the assembly platform are transient and dynamic.

Interestingly, conditions favouring PpdD pilus assembly did not support pullulanase secretion, even in the presence of minor T2SS pseudopilins. Recently, it was shown that minor pseudopilin complex interacts in vitro with the specific substrate of the P. aeruginosa T2SS (Douzi et al., 2011). This result supports the piston model where the pseudopilus tip complex promotes secretion by direct contact with the substrate. Consistent with this model, E. coli minor pilins cannot substitute for the Pul T2SS minor pseudopilins in pullulanase secretion. One could hypothesize that, in the absence of the major pseudopilin (PulG), assembly of another pilin, (such as PpdD) should provide the driving force for the piston-like motion of the pseudopilin tip complex. However, this was not the case (Fig. 5), showing that, although necessary, minor pseudopilins are not sufficient for function. The functional defect could also be due to a lack of structural compatibility between minor pilins and any other component of the T2SS, since the sequence similarity between globular domains of T4P pilins and T2SS pseudopilins is low (see above). This could also explain why the E. coli T4PS minor pilins could not promote PulA secretion in the presence of PulG, and were clearly less efficient in initiating PulG pilus assembly compared with PulHIJK (Figs 6 and S1).

Alternatively, PulHIJK-promoted PpdD assembly by the Pul T2SS might not respect all aspects of the dynamics of pseudopilus assembly and disassembly that are required for pseudopilus function in secretion. A third explanation might invoke a direct and specific role for pseudopilin PulG, although PulG can be replaced by at least some major pseudopilins from other T2SS (Vignon et al., 2003). Further studies will be necessary to identify the qualitative difference between partial and fully efficient fibre assembly efficiency and the mechanistic link between assembly and protein secretion.

Structural and functional relatedness of T4PS and T2SS

Recent analyses revealed a higher degree of structural similarity between T2SS and T4PS than was suspected based on their composition and sequence (Golovanov et al., 2006; Ayers et al., 2010; Gu et al., 2012; Korotkov et al., 2012; McLaughlin et al., 2012). The present study further reinforces these similarities, showing that they extend to the functional level. We show that minor T4 pilins and T2SS pseudopilins initiate or activate the assembly of both major pilins and major pseudopilins in our heterologous system.

On the other hand, PpdD and PulG form fibres of different length and thickness apparent by immunofluorescence microscopy, implicating globular domains of major pilins as sole determinants of these properties. The major remaining difference between the T4PS and T2SS could lie in the intrinsic stability of pilins and the instability of pseudopilins. PpdD and its cognate minor pilins have two predicted disulfide bridges, which provide substantial stability. A recently characterized major pilin FimA from Dichelobacter nodosus also contains a disulfide bond in the αβ-loop as predicted for PpdD; however, its C-terminal d-loop, although well structured, is not stabilized by a disulfide bond but rather by a hydrogen bond network (Hartung et al., 2011). In T2SS, major pseudopilins do not contain cysteine residues and their globular domains are stabilized by a single Ca2+ ion in the C-terminal region that corresponds to the d-loop of T4 pilins (Korotkov et al., 2009). Therefore, pseudopilins might be more prone to degradation and to conformational changes leading to fibre dissociation. Indeed, in T2SS, pseudopili are short and remain periplasmic, although the disassembly ATPase is lacking. In contrast, the energy provided by the retraction ATPase of the PilT family present in T4PS is necessary to disassemble T4P, which are very stable. For example, gonococcal pili are heat resistant and remain intact in 8 M urea (Li et al., 2012). The analysis of the T4 pilin assembly in the reconstituted system could provide a valuable model for a comparative study of structure, dynamics and function of these two distinct types of filaments. Detailed structural analysis of PpdD pilins and the assembled pili currently under way will provide insight into the molecular bases underlying these distinct features.

Experimental procedures

Bacterial strains and culture conditions

The E. coli strain DH5α F′ lacIQ was used for DNA cloning experiments and strain PAP7460 (MC4100 ΔmalE444 malG510 F′ lacIQ) (Possot et al., 2000) was used as a pul gene expression host. Bacteria were grown at 30°C or 37°C in Luria–Bertani (LB) medium supplemented, when indicated with 1.5% agar and 0.2–0.4% maltose. Antibiotics were added where required at following concentrations: ampicillin (Ap) at 100 μg ml−1, chloramphenicol (Cm) at 25 μg ml−1, zeocin (Ze) at 25 μg ml−1 and kanamycin (Km) at 25 μg ml−1. The complete list of plasmids used in this study is given in Table 1.

Table 1. Plasmids used in this study
NameCloned genesResistancea, originReference
  1. aAp, ampicillin; Cm, chloramphenicol; Ze, zeocin resistance.
pCHAP3100ppdDCm, p15ASauvonnet et al. (2000a)
pCHAP6116ppdAB-ygdB-ppdCZe, pSC101This study
pCHAP7292pulHIJKZe, pSC101This study
pCHAP8509pulHIJKZe, pSC101This study
pCHAP8254pulGHIJKZe, pSC101This study
pCHAP231All pul genesAp, ColE1d'Enfert et al. (1987)
pCHAP1216All pul genes except pulGAp, ColE1Possot et al. (2000)
pCHAP7248All pul genes except pulGHIJKAp, ColE1Cisneros et al. (2012)
pCHAP8296All pul genes except pulHIJKAp, ColE1Cisneros et al. (2012)
pCHAP8432All pul genes except pulHIJK, pulAsolAp, ColE1This study
pSU18Empty vectorCm, p15ABartolome et al. (1991)
pCHAP418zeo/pCHAP419zeoEmpty vectorZe, pSC101This study

Plasmid constructions

DNA extraction, plasmid constructions and DNA transformation were performed as described (Maniatis et al., 1982). To generate plasmid pCHAP418zeo, the pCHAP418 vector (Possot et al., 1992) was linearized with NcoI and BspEI and treated with Klenow fragment DNA polymerase (New England Biolabs). The PCR (polymerase chain reaction) amplified ZeR cassette flanked with SmaI sites from plasmid pCDA21 (Chaveroche et al., 2000) was cloned into this fragment. The ppdA-ppdB-ygdB-ppdC gene cluster was amplified from the chromosomal DNA of E. coli strain MG1655 by polymerase chain reaction with oligonucleotides ppdAL (5′-CACGAATTCTCCTCGCTCCATACTG-3′) and PpdCR (5′-CGCTCTAGACGATTGGAATGGTAGAC-3′). The PCR product was digested with EcoRI and HindIII and cloned in plasmid pCHAP418zeo digested with the same enzymes to yield plasmid pCHAP6116. Plasmid pCHAP8509 was obtained by cloning the PCR amplified gene cluster pulHIJK, using oligonucleotides PulH5 (5′-TGAGGCGAATTCGGGTCAGGTGGATATTTTCAGCC-3′) and PulK3 (5′-GCAGTCAAGCTTCGGAAGAGGTATGGTGGTT-3′). Plasmid pCHAP8432 was obtained by cloning the pCHAP8296 1246 bp NotI HindIII fragment into 21 647 bp NotI HindIII fragment of plasmid pCHAP8212 (Cisneros et al., 2012). Plasmids pCHAP7292 and pCHAP8254 were made by subcloning the EcoRI HindIII inserts from plasmids pCHAP7273 and pCHAP7257 (Cisneros et al., 2012) respectively, into pCHAP419zeo. For PCR amplification we used the Pwo DNA polymerase (Roche). Restriction enzymes and T4 DNA ligase was purchased from New England Biolabs and Fermentas. Plasmid DNA was purified by Qiaprep. All constructs were verified by DNA sequencing (GATC). Oligonucleotides were synthesized by Sigma Genosys.

Pilus assembly, shearing and fractionation assays

Shearing assay was performed as described (Sauvonnet et al., 2000b). Bacteria carrying derivatives of plasmid pCHAP231 carrying WT or mutant pul genes were grown overnight on LB agar containing Ap and Cm and 0.4% maltose. Bacteria were scraped off the plates and resuspended in LB medium at 10 A600nm ml−1 and vortexed vigorously for 2 min to detach pili from bacterial surface. Bacteria were pelleted for 5 min at 16 000 g in a tabletop centrifuge (Sigma) and resuspended in SDS sample buffer. The supernatant was centrifuged another 10 min at maximal speed and pili were precipitated in 10% trichloroacetic acid for 30 min on ice. The precipitated proteins were collected by centrifugation for 15 min at 16 000 g, washed twice with cold acetone, air dried and resuspended in SDS sample buffer at a concentration corresponding to that of bacterial pellet.

Pullulanase secretion assay

Pullulanase secretion was measured as a fraction of pullulanase enzymatic activity accessible on the bacterial surface in whole cells and bacteria lysed with octyl-polyoxyethylene (Possot et al., 2000). The non-acylated PulA secretion was analysed by bacterial fractionation in strain PAP7460 (pCHAP8432). Bacteria were grown in LB medium supplemented by 0.4% maltose and containing 1/10th vol of M63 salts to OD600nm = 1.5. An equivalent of 0.005 A600nm of bacterial and supernatant fractions was analysed by 10% Tris–glycine SDS/PAGE and immunodetection using anti-PulA antiserum (1:2000).

Fluorescence microscopy

Immunofluorescence labelling of pili was performed as described (Vignon et al., 2003). Bacteria were grown in the same conditions as for pilus assembly assay, except that resuspended bacteria were directly immobilized on poly-l-lysine coated coverslips. Samples were fixed for 20 min on 3.7% formaldehyde, blocked with 1% BSA in PBS and incubated with anti-PulG or anti-PpdD antibodies (1:2000) and secondary Alexa Fluor 488-coupled anti-rabbit IgG (Invitrogen). Samples were examined with an Axio Imager.A2 microscope (Zeiss). Images were taken with AxioVision (Zeiss) and processed in ImageJ (Abramoff et al., 2004).


Immunoblotting was performed as described (Sauvonnet et al., 2000b). Proteins were separated by SDS-PAGE in tricine (Schagger and von Jagow, 1987) gels containing 10–15% acrylamide, transferred onto nitrocellulose membranes (Amersham ECL) using semi-dry electro-transfer. Membranes were blocked with 5% milk in TBST (10 mM Tris, 150 mM NaCl, 0.05% Tween 20) and incubated in specific antiserum (anti-PulG at 1/2000, anti-PpdD at 1/2000, anti-PulA 1/2000 or anti-LamB 1/2000) followed by horseradish peroxidase-coupled anti-rabbit antibody (1/40 000; Amersham). Membranes were developed by enhanced chemiluminescence ECL-plus (GE-healthcare) and recorded using the Typhoon phosphoimager (Molecular Dynamics). Densitometric analysis was performed using ImageJ (Abramoff et al., 2004).

Electron microscopy

For transmission electron microscopy (TEM) analysis, glow discharged carbon grids (EMS) were coated with 2 mM CaCl2. Bacteria or isolated pili were spotted on these grids and then incubated in BHN buffer (1% BSA, 50 mM HEPES, 150 mM NaCl, pH 7.4) for 10 min. Grids were incubated for 45 min with primary antibody (anti-PpdD 1/100 or anti-PulG 1/200), washed in BHN and finally incubated 20 min with Protein A Gold in BHN. After 5 min of fixation in 1% glutaraldehyde, grids were washed with water and contrasted with 2% uranyl acetate for 30 s. For double-labelling experiments, the first step was performed as described using anti-PpdD (1/100) and Protein A Gold (5 nm). After the first labelling, grids were quenched 10 min in 50 mM NH4Cl, incubated in BHN buffer for 10 min, incubated 45 min with anti-PulG (1/200), washed in BHN and finally incubated with Protein A Gold (15 nm). Imaging was performed at 120 kV on a CM12 electron microscope (FEI, Eindhoven, the Netherlands) equipped with a KeenView camera (SIS, Japan).


This work was funded in part by the Institut Pasteur Grant PTR339. D.C. was supported by EMBO long-term and Roux Fellowships. We thank Tony Pugsley for the critical reading of the manuscript and all past and present members of Molecular genetics Unit for helpful discussions, support and input. We thank Nathalie Nadeau for excellent technical assistance.