The assembly of the Yersinia enterocolitica type III secretion injectisome was investigated by grafting fluorescent proteins onto several components, YscC (outer‐membrane (OM) ring), YscD (forms the inner‐membrane (IM) ring together with YscJ), YscN (ATPase), and YscQ (putative C ring). The recombinant injectisomes were functional and appeared as fluorescent spots at the cell periphery. Epistasis experiments with the hybrid alleles in an array of injectisome mutants revealed a novel outside‐in assembly order: whereas YscC formed spots in the absence of any other structural protein, formation of YscD foci required YscC, but not YscJ. We therefore propose that the assembly starts with YscC and proceeds through the connector YscD to YscJ, which was further corroborated by co‐immunoprecipitation experiments. Completion of the membrane rings allowed the subsequent assembly of cytosolic components. YscN and YscQ attached synchronously, requiring each other, the interacting proteins YscK and YscL, but no further injectisome component for their assembly. These results show that assembly is initiated by the formation of the OM ring and progresses inwards to the IM ring and, finally, to a large cytosolic complex.
The type III secretion (T3S) apparatus, also called injectisome, allows bacteria to export effector proteins on contact with eukaryotic cell membranes (Cornelis and Wolf‐Watz, 1997; Galan and Collmer, 1999; Cornelis and Van Gijsegem, 2000). Effectors (called Yops in Yersinia) display a large repertoire of biochemical activities and modulate the function of crucial host regulatory molecules to the benefit of the bacterium (Alfano and Collmer, 2004; Mota and Cornelis, 2005; Grant et al, 2006). In Yersinia spp., the injectisome is built when temperature reaches 37°C and export of the Yops can be artificially triggered, in the absence of cell contact, by Ca2+ chelation (Cornelis, 2006).
About 25 proteins (called Ysc in Yersinia) are needed to build the injectisome. Most of these are structural components, but some are ancillary components that are only involved during the assembly process and are either shed afterwards (e.g. the molecular ruler) or kept in the cytosol (e.g. chaperones). In contrast to the large diversity observed among effectors, the core proteins forming the injectisome (YscC, J, N, Q, R, S, T, U, V, and, to a lesser extent, YscD in Yersinia) are well conserved (Van Gijsegem et al, 1995; Cornelis, 2006).
A number of injectisome proteins copurify as a complex cylindrical structure, resembling the flagellar basal body. This structure, called the needle complex, consists of two pairs of rings that span the inner membrane (IM) and outer membrane (OM) of bacteria, joined together by a narrower cylinder and terminated by a needle, a filament, or a pilus (Kubori et al, 1998; Blocker et al, 1999; Kimbrough and Miller, 2000; Daniell et al, 2001; Jin and He, 2001; Sekiya et al, 2001; Morita‐Ishihara et al, 2006; Sani et al, 2007; Hodgkinson et al, 2009; Schraidt et al, 2010). The needle is a hollow tube assembled through helical polymerization of a small protein (around 150 copies of YscF in Yersinia) (Cordes et al, 2003; Deane et al, 2006). It terminates with a tip structure serving as a scaffold for the formation of a pore in the host cell membrane (Mueller et al, 2005). The ring spanning the OM (hereafter called OM ring) and protruding into the periplasm consists of a 12–14mer of a protein from the YscC family of secretins (Koster et al, 1997; Kubori et al, 2000; Tamano et al, 2000; Blocker et al, 2001; Marlovits et al, 2004; Burghout et al, 2004b; Spreter et al, 2009). The lower ring spanning the IM is called MS ring and made of a lipoprotein (YscJ in Yersinia, MxiJ in Shigella, PrgK in Salmonella enterica SPI‐1) proposed to form a 24‐subunit ring (Kimbrough and Miller, 2000; Crepin et al, 2005; Yip et al, 2005; Silva‐Herzog et al, 2008; Hodgkinson et al, 2009). A protein from the less‐conserved YscD family (MxiG in Shigella, PrgH in S. enterica SPI‐1), which has the same general fold as the components of the two rings, is proposed to participate in MS ring formation and possibly connect the rings in the two membranes (Spreter et al, 2009).
Besides these proteins forming a rigid scaffold, the injectisome contains five essential integral membrane proteins (YscR, S, T, U, V), which are believed to recognize export substrates (Sorg et al, 2007) and form the export channel across the IM. Some of them, if not all, are likely to be inserted in a patch of membrane enclosed within the MS ring, but this could not be shown so far. We will refer to these proteins as to the ‘export apparatus’. At the cytosolic side of the injectisome, an ATPase of the AAA+ family (YscN) forms a hexameric ring that is activated by oligomerization (Woestyn et al, 1994; Pozidis et al, 2003; Muller et al, 2006; Zarivach et al, 2007) and resembles the flagellar ATPase FliI (Abrahams et al, 1994; Imada et al, 2007). The ATPase is associated with two proteins (YscK, L) (Jackson and Plano, 2000; Blaylock et al, 2006), one of them (YscL) probably exerting a control on the ATPase activity as was shown for FliH in the flagellum (Minamino and MacNab, 2000; Gonzalez‐Pedrajo et al, 2002; McMurry et al, 2006). The ATPase is strikingly similar to the α and β subunits of the stator of the F0F1 ATP synthase (Abrahams et al, 1994), suggesting an evolutionary relation. This assumption is reinforced by the sequence similarity observed between YscLN‐term and the b subunit of the F‐type ATPase, and between YscLC‐term and the δ subunit of the same ATPase (Pallen et al, 2006). A function of the ATPase, characterized in S. enterica Typhimurium SPI‐1, is to detach some T3S substrates from their cytoplasmic chaperone before their export and to unfold the exported proteins in an ATP‐dependent manner (Akeda and Galan, 2005). It is likely that the ATPase also directly energizes export, but the proton motive force is also involved (Wilharm et al, 2004; Minamino and Namba, 2008; Paul et al, 2008).
In the flagellum, the most proximal part of the basal body is the 45–50 nm C ring (for cytosolic) made of FliM and FliN (Driks and DeRosier, 1990; Khan et al, 1992; Kubori et al, 1997; Young et al, 2003; Thomas et al, 2006). Together with FliG, it forms the switch complex reversing the rotation of the motor, but in its absence, no filament appears, indicating that it is also involved in the export of distal constituents (Macnab, 2003). However, recent reports (Konishi et al, 2009; Erhardt and Hughes, 2010) showed that in C ring mutants, the export function can be partially restored by overexpression of the ATPase or the master regulator. No such C ring could be visualized so far by electron microscopy in a needle complex, but proteins of the YscQ family, which are essential components of all injectisomes, have a significant similarity to FliN and FliM. In Pseudomonas syringae, the orthologue of YscQ even appears as two products called HrcQA and HrcQB, which interact with each other, and the overall fold of HrcQB is remarkably similar to that of FliN (Fadouloglou et al, 2004). This suggests that injectisomes do have a C ring, although they have not been reported to rotate. YscQ and its homologues have been shown to bind the ATPase complex (Jackson and Plano, 2000) as well as substrate–chaperone complexes (Morita‐Ishihara et al, 2006). The C ring would, therefore, form a platform at the cytoplasm/IM interface for the recruitment of other proteins. In agreement with this assumption, immunogold‐labelling experiments have shown that the Shigella orthologue of YscQ (Spa33) localizes to a lower portion of the injectisome (Morita‐Ishihara et al, 2006). A list of homologues in the flagellum and the various archetypal T3S systems is given in Supplementary Table 3.
The assembly of the flagellum is for the most part linear and sequential, proceeding from more proximal structures to more distal ones. The proposed scenario is that the plasma membrane ring (called the MS ring) formed by FliF assembles first, followed by periplasmic components, OM components, and finally components that lie in the cell exterior (Kubori et al, 1992; Macnab, 2003). The C ring (FliG, FliM, FliN) is thought to appear immediately after the MS ring, because it forms spontaneously when its components are overexpressed in the presence of FliF even in the absence of any other component (Kubori et al, 1997; Lux et al, 2000; Young et al, 2003).
Less is known about the assembly steps of the injectisome. The heterologous overexpression of the S. enterica SPI‐1 MS ring components PrgH and PrgK in Escherichia coli leads to stable ring structures (Kimbrough and Miller, 2000). The same is true for the Yersinia secretin YscC together with its pilotin YscW (Koster et al, 1997). This suggests that the transmembrane rings might form independently. It has thus been proposed (Kimbrough and Miller, 2000) that the first step consists in the assembly of the MS ring, possibly along with the recruitment of the transmembrane proteins forming the export apparatus. In parallel, the secretin ring would form in the OM. Afterwards, the two rings would join by an unknown mechanism, allowing the assembly of the remaining machinery, which then exports the distal components, including the needle and the needle tip. The exact order of these later steps of the injectisome assembly remains largely unknown. A similar model was put forward based on the genetic analysis of the requirements for needle complex formation in S. enterica SPI‐1 (Sukhan et al, 2001).
In this paper, we systematically investigate the whole assembly process of the Yersinia injectisome by combining four functional fluorescent hybrid proteins covering different parts of the machinery with an array of deletions. We conclude that the assembly starts from the secretin, the outermost and most stable ring, and sequentially proceeds inwards through YscD and YscJ. After completion of the membrane rings, an ATPase–C ring complex formed by YscK, YscL, YscN, and YscQ joins the machinery. All of the four participating proteins, but not the ATPase activity of YscN are required for the formation of this structure.
Various substructures of the Yersinia injectisome including the C ring can be monitored using functional fluorescent fusion proteins
To visualize the injectisome and its subunits, the wild‐type alleles of yscC, yscD, and yscQ on the virulence plasmid of Y. enterocolitica E40 were replaced by hybrid genes encoding the fluorescent proteins YscC–mCherry, EGFP–YscD, and EGFP–YscQ. Further, a non‐polar complete deletion of yscN was constructed and complemented in trans with a plasmid encoding EGFP–YscN. The fusion proteins were expressed at near wild‐type levels; no proteolytic release of the fluorophore was detected (Supplementary Figure 1).
To test the functionality of the fusion proteins, the pattern of proteins secreted into the supernatant in secretion‐permissive medium (BHI‐Ox) was analysed 3 h after induction of the system. YscC–mCherry, EGFP–YscN, and EGFP–YscQ were fully functional, whereas the strain expressing EGFP–YscD secreted a lower amount of effector proteins (Figure 1B). All fusion proteins allowed the formation of needles, which could be visualized by transmission electron microscopy (data not shown).
The localization of the hybrid proteins was analysed by fluorescence microscopy. Three hours after induction of synthesis of the injectisome, fluorescent spots were observed at the cell periphery for all labelled proteins (Figure 1A, three‐dimensional view in Supplementary data). The formation of these spots was independent of the Ca2+ concentration in the medium, showing that their appearance was not directly linked to the secretion of Yop proteins by the T3S system (Figure 1A).
To ascertain that the membrane spots correspond to assembled basal bodies, we constructed a strain expressing both YscC–mCherry and EGFP–YscQ, and monitored the localization of the green fluorescence from EGFP–YscQ and the red fluorescence from YscC–mCherry. As visible in Figure 1C, the green and red spots largely colocalized, with small deviations because of chromatic aberrations of the microscope. We thus assumed that the fluorescent spots correspond to assembled basal bodies. In a minority of cells, a polarily localized YscC–mCherry spot without EGFP–YscQ equivalent could be observed in addition to the colocalizing spots. We assumed that these polar spots consist of misassembled YscC–mCherry proteins. Colocalization of spots was also observed for EGFP–YscD and EGFP–YscN with YscC–mCherry (data not shown). To test for colocalization of the needle with the basal body components, bacteria producing EGFP–YscQ were analysed by immunofluorescence with purified antibodies directed against the needle subunit. Overlays of the resulting pictures with the EGFP–YscQ fluorescence revealed that the majority of spots for YscF and YscQ colocalized (Supplementary Figure 2). A fraction of YscQ spots did not correspond to YscF spots. Most likely, the needles of these basal bodies were detached during the immunofluorescence procedure. We conclude from all these experiments that the fluorescent spots correspond to functional injectisomes.
Assembly of the injectisome starts from the secretin ring in the OM and proceeds inwards through stepwise assembly of YscD and YscJ
As earlier work has shown that secretins can insert in the OM provided they are assisted by their pilotin (Burghout et al, 2004a; Guilvout et al, 2006), the fluorescent YscC–mCherry and its pilotin YscW were expressed in trans in Y. enterocolitica E40 (pMA8)(pRS6), in the absence of the pYV virulence plasmid encoding the T3S components. YscC–mCherry localized in membrane spots (Figure 2A), as observed before for PulD, the secretin involved in a type II secretion pathway (Buddelmeijer et al, 2009). These data thus confirm earlier results showing that YscC only requires its pilotin for assembly in the OM (Burghout et al, 2004a). In the absence of YscW, the majority of YscC–mCherry clustered in spots at the bacterial pole (Supplementary Figure 3). This phenotype was clearly distinguishable from the membrane spot formation in the presence of YscW, and confirmed the function of YscW in proper localization and oligomerization of YscC (Burghout et al, 2004a).
Not surprisingly, mutants lacking any of the structural ring proteins YscC, YscD, or YscJ failed to assemble the cytosolic injectisome components YscN and YscQ (Table 1), showing that establishment of the membrane‐spanning structure formed by YscC, YscD, and YscJ is at the beginning of injectisome formation. To test for the assembly order of these proteins, we combined the egfp–yscD allele on the pYV plasmid with non‐polar deletions in yscC and yscJ. Although the absence of YscC clearly abolished the formation of EGFP–YscD spots at the bacterial membrane, the absence of YscJ did not affect this assembly (Figure 2B). This implies that YscC assembles first, followed by YscD, and finally YscJ.
To confirm this order of assembly, we performed co‐immunoprecipitation assays using strains in which the wild‐type alleles of yscD or yscJ on the virulence plasmid were replaced by his‐flag‐yscD or yscJ‐flag‐his, respectively. The affinity tagged proteins were functional for effector secretion (data not shown) and hence assumed to assemble in the same way as wild type. They were further combined with non‐polar deletions in yscC, yscD, or yscJ. In each of the strains, the adhesin YadA was removed to facilitate cell lysis. Synthesis of the injectisome in these strains was induced under secretion‐non‐permissive conditions. Mild crosslinking was performed, spheroplasts were created, and the bacteria were lysed by the addition of detergent (see Material and methods). Afterwards, a one‐step affinity purification was performed, and the (co‐)purification of YscC, YscD, and YscJ was tested. YscJ‐FLAG‐His copurified YscC and YscD from complete injectisomes, and removal of YscQ, a protein thought to act further downstream in the assembly process, did not affect this copurification. In contrast, removal of YscC prevented copurification of YscD with YscJ‐FLAG‐His, and removal of YscD prevented copurification of YscC (Figure 2C). Likewise, His‐FLAG‐YscD copurified YscC and YscJ from complete injectisomes. However, although removal of YscC prevented copurification of YscJ with His‐FLAG‐YscD, removal of YscJ still allowed the copurification of YscC (Figure 2D). The amount of purified His‐FLAG‐YscD was reduced in the absence of YscC, most likely as a consequence of decreased cellular YscD levels, either because of its mislocalization in the absence of YscC or because of a lower expression level. Taken together, these data indicate (i) that the insertion of the secretin ring in the OM is required for the subsequent association of YscD and YscJ and (ii) that YscD makes the link between YscC and YscJ. Hence, the OM ring is the first ring of the injectisome to be assembled. This assembly step is followed by the attachment of YscD, which then allows the completion of the MS ring by YscJ.
The C ring only assembles in the presence of the membrane rings, YscN, YscK, and YscL
To determine at which stage the C ring forms during the assembly process, we combined the egfp–yscQ allele with an array of deletions in most injectisome genes (Table 1; Supplementary Table 1 for details of strains). Deletion of any of the membrane ring proteins (YscC, YscD, or YscJ) completely abolished the formation of membrane spots and led to an increased diffuse cytoplasmic fluorescence (Figure 3). This indicates that the C ring forms after the YscCDJ ring structure. Removal of the ATPase YscN or any of its two interacting proteins YscK and YscL (Jackson and Plano, 2000; Blaylock et al, 2006) also fully prevented C ring formation (Figure 3), indicating that assembly of the C ring additionally requires YscN as well as YscK and YscL.
Removal of individual proteins YscR, S, T, U, or V from the export apparatus as well as a complete deletion of all these proteins did not completely abolish the formation of the C ring. However, in the absence of YscR, YscS, or YscV, the number of spots was reduced, indicating that these proteins are either beneficial (but not absolutely required) for C ring formation, or have a stabilizing effect on fully assembled injectisomes.
As expected from the fact that YscF, YscI, YscO, YscX, and YopN are substrate proteins exported by the injectisome itself, their absence also did not prevent the formation of the C ring. Likewise, deletion of LcrG, a regulatory protein (Nilles et al, 1997; Torruellas et al, 2005), had no effect on assembly of the C ring (Figure 3; see Table 1 for additional strains).
ATPase assembly not only requires the presence of the YscCDJ platform, but also needs YscK, YscL, and YscQ
Finally, assembly of the ATPase YscN was tested. As replacement of the wild‐type allele on the virulence plasmid by a gene encoding a fluorescent fusion protein decreased the expression of downstream genes in the virB operon, whereas a complete deletion of yscN was non‐polar (Figure 1B), egfp–yscN was cloned in a pBAD vector and used to complement in trans double deletions in yscN and several other genes. Induction of synthesis of EGFP–YscN with 0.05% arabinose led to YscN protein levels similar to the native level (Supplementary Figure 1), and to effector secretion at wild‐type levels (data not shown). As shown in Figure 4, YscN assembly required the presence of YscC (secretin), YscJ (MS ring), YscK and YscL (two proteins known to interact with the ATPase), and YscQ (the C ring). In contrast, even the complete deletion of the IM export proteins YscR, S, T, U, V still allowed formation of YscN spots, albeit again in a reduced number (Figure 4).
These data suggest that the cytosolic components of the injectisome form a single large ATPase–C ring complex, requiring all of its components YscK, L, N, Q to assemble.
In agreement with the essential function of YscQ for the ATPase assembly, we did not observe any needle formation in strains that lack YscQ, but overexpress YscN, in contrast to recent results obtained with the flagellum (Konishi et al, 2009) (data not shown).
ATPase activity of YscN is not required for the assembly of the ATPase–C ring complex at the injectisome
To determine whether assembly of the C ring requires YscN for its ATPase activity or as a structural component, a deletion of yscN in an egfp–yscQ background was examined. As expected, the resulting strain secreted neither Yops nor the ruler and needle subunits. Secretion could be complemented in trans by a wild‐type yscN allele, but not by an yscN allele encoding YscNK175E altered in the Walker box (Figure 5B and C). Interestingly, however, although YscNK175E was not functional, it did restore the formation of the C ring spots (Figure 5A), implying that the YscN requirement for the formation of the ATPase–C ring complex is exclusively structural.
After assembly of the ATPase–C ring complex, needle formation and effector secretion take place rapidly
The kinetics of C ring formation in a strain expressing EGFP–YscQ was followed in a time‐course experiment. Pictures were taken every 20 min up to 2 h after induction of the ysc‐yop regulon (Cornelis et al, 1989) in a Ca2+‐depleted medium. Weak diffuse cytoplasmic fluorescence could be observed 20 and 40 min after the temperature shift, suggesting that synthesis of YscQ was turned on directly after the shift, and that EGFP folds rapidly in the Yersinia cytosol (Figure 6A). The rapid synthesis of YscQ was also confirmed by immunoblotting (data not shown). The first membrane spots could, however, only be observed 60 min after induction. Although the fluorescence intensity of single spots seemed to increase over time, the number of spots stayed roughly constant up to 3 h after induction (Figure 6A). Interestingly, the timeframe of appearance of the C ring was approximately the same as the timeframe of appearance of the needles (Figure 6B) and secretion of the effector proteins (Figure 6C), suggesting that needle formation and effector secretion occur within a short time after establishment of the ATPase–C ring complex. A model of injectisome assembly that incorporates the above mentioned results is depicted in Figure 7.
The assembly of the T3S injectisome is a complex process that engages >20 different proteins, and results in the formation of a nanomachine spanning both bacterial membranes and protruding outside the bacterium. So far, little is known about this process. On the basis of the observation that heterologously overexpressed S. enterica MS ring components PrgH and PrgK form large rings in the absence of any other T3S component, a model was proposed (Kimbrough and Miller, 2002) in which the IM ring assembles first, and then fuses with the secretin ring in the OM. This model suggests the same general assembly scheme as the one that has been proposed for the flagellum (Kubori et al, 1997; Macnab, 2003), but does not explain how the two membrane rings find each other. The subsequent steps of assembly could not be examined so far.
To gain better insight into the assembly process and the functional relations between the proteins, we constructed strains in which a number of injectisome constituents were fused to fluorescent proteins. To minimize artefacts because of non‐native expression levels or timing of the fusion proteins, we replaced the wild‐type allele on the pYV virulence plasmid by the hybrid allele in the case of yscC, yscD, and yscQ. The hybrid yscN was plasmid borne, but it was expressed at wild‐type level.
All recombinant injectisomes were functional, and in all cases, fluorescent spots appeared at the bacterial membrane, distributed all over the bacterial body. Colocalization of the fluorophores confirmed that the fluorescent spots correspond to injectisomes. The brightness of the spots likely results from the multimeric nature of the tagged proteins, although at this stage, one cannot conclude that each spot corresponds to only one injectisome. A fluorescence quantification based on external standards, presently in progress, will address this question.
We observed that the membrane ring‐forming proteins YscC, YscD, and YscJ are required for assembly of any cytosolic structure. Importantly, by monitoring the formation of YscC–mCherry and EGFP–YscD spots, we observed that YscC assembles independently of YscD and YscJ and that YscD assembles independently of YscJ, but not of YscC. Co‐immunoprecipitation assays confirmed that YscJ requires YscD to become attached to YscC. All this implies that the assembly of the injectisome is initiated by formation of the secretin ring in the OM and proceeds inwards through stepwise assembly of YscD and YscJ. These data contradict the earlier report that PrgH and PrgK, the Salmonella homologues of YscD and YscJ, can form a ring alone (Kimbrough and Miller, 2000). The discrepancy might result from the fact that the earlier study was based on heterologously overexpressed proteins, whereas this study is based on functional proteins produced in their natural environment at native expression levels. These data are also at odd with the report indicating that MxiD and MxiJ, the Shigella homologues of YscC and YscJ, interact even in the absence of the connector (Schuch and Maurelli, 2001). However, this interaction was observed in the absence of the pilot protein. In this case, the majority of secretin proteins are mislocalized to the IM (Koster et al, 1997), which might lead to non‐native interaction with MxiJ. Interestingly, it was shown recently that the assembly of two ring‐forming IM components of the Vibrio cholerae type II secretion complex also depends on the presence of the OM secretin (Lybarger et al, 2009), suggesting conservation or convergent evolution of the formation process in these two prokaryotic export systems. Taken together, our results show that the order of assembly of the OM and IM rings differs between the injectisome and the flagellum. We do not see any obvious reason for this, but this observation indicates that the two nanomachines differ more than is often thought. The flagellum is indeed significantly more complex than the injectisome because it rotates, which implies not only a motor but also bushings in the peptidoglycan and the OM. The P and L rings, having this function, are precisely replaced by a very stable secretin ring in the injectisome. This basic structural difference might explain the different order of assembly of the two nanomachines.
The outside‐in assembly order consistently shown by co‐immunprecipitation and fluorescence microscopy further implies that YscD is the connector between the two membrane rings, which is coherent with recent crystal structure and modelling data (Spreter et al, 2009). Our biochemical data allow to assess the recent models to integrate the crystal structures of the membrane ring proteins into the overall shape generated by electron microscopy averaging of purified injectisomes (Hodgkinson et al, 2009; Spreter et al, 2009). The electron density between the membranes would be generated by YscD. This in turn places YscJ in the IM, as proposed by manual fits (Moraes et al, 2008; Hodgkinson et al, 2009; Spreter et al, 2009), but not by the best automated fit (Hodgkinson et al, 2009).
After assembly of the OM and IM membrane rings, cytosolic components can assemble onto the structure. The observation that the proposed C ring component YscQ assembles in membrane spots colocalizing with the other components shows that the C ring is an integral component of the injectisome, confirming an assumption so far mainly based on immunogold‐labelling experiments (Morita‐Ishihara et al, 2006). Our data indicate that a large cytosolic complex consisting of the ATPase YscN, the two interacting proteins YscK and YscL, and the C ring component YscQ is formed, requiring all of its components, but not the ATPase activity of YscN for assembly. This differs again from the situation in the flagellum. There, FliM, FliN, and FliG (together forming the C ring) appear in significant amount in the membrane fraction in the presence of FliF (MS ring), but in the absence of FliI (ATPase) or FliH (homologue to YscL). This suggests that the flagellar C ring forms in the absence of the ATPase complex (Kubori et al, 1997), in agreement with the observation that it forms on overexpression of FliM, FliN, and FliG together with FliF (Lux et al, 2000; Young et al, 2003). Although the heterologous overexpression of the proteins in these studies might account for the different observations, these results can also be the consequence of functional differences between the two nanomachines. As the constraint of rotation and spatial separation of the C ring and ATPase does not exist for the injectisome, the apparatus could be optimized for secretion. A tighter contact between the ATPase complex and the C ring might be a consequence of this optimization. The fact that we could not overcome the requirement of YscQ for secretion by overexpression of the ATPase is consistent with the essential function of YscQ for assembly of the complete ATPase–C ring complex.
Our results are also in perfect agreement with earlier results showing interactions between YscK, YscL, YscN, and YscQ (Jackson and Plano, 2000). However, the hypothesis that YscQ recruits the ATPase should be revised: YscK, L, N, and Q would rather assemble in one step. The proposed function of YscL as a negative regulator of ATPase activity (Blaylock et al, 2006) as well as its direct interaction with YscN and YscQ (Jackson and Plano, 2000) is consistent with the presence of YscL in this complex. Less is known about the function of YscK. As it interacts with YscQ, but not with YscN and weakly at the most with YscL (Jackson and Plano, 2000; Blaylock et al, 2006), it might act at the interface of the ATPase–C ring complex.
Formation and assembly of the ATPase–C ring complex did not depend on any of the five proteins forming the export apparatus, even though the number of membrane spots was reduced when YscR, YscS, YscV, or the five proteins YscRSTUV were missing. This implies that a YscKLNQ complex docks onto the IM ring rather than onto the export apparatus, which agrees with the observations made with the flagellum (Kubori et al, 1997). As currently, little is known about stoichiometry, localization, and function of the export apparatus, its function in the assembly process remains unclear.
In conclusion, this work shows that the assembly of the injectisome starts with the formation of the stable secretin ring in the OM, and proceeds inwards through discrete attachment steps of YscD and YscJ at the IM. Afterwards, the components of the cytosolic ATPase–C ring complex assemble at the cytosolic side of the injectisome in one step, which allows the subsequent fast steps leading to needle formation and effector secretion.
Materials and methods
Bacterial strains, plasmids, and genetic constructions
Y. enterocolitica strains and plasmids are listed in Supplementary Table 1.
E. coli Top10, used for plasmid purification and cloning, and E. coli Sm10 λ pir, used for conjugation, were routinely grown on LB agar plates and in LB broth. Ampicillin was used at a concentration of 200 μg/ml to select for expression vectors. Streptomycin was used at a concentration of 100 μg/ml to select for suicide vectors. Plasmids were generated using either Phusion polymerase (Finnzymes, Espoo, Finland) or Vent DNA polymerase (New England Biolabs, Frankfurt, Germany). The oligonucleotides used for genetic constructions are listed in Supplementary Table 2. Mutators for modification or deletion of genes in the pYV plasmids were constructed by overlapping PCR using purified pYV40 plasmid as template, leading to 200–250 bp of flanking sequences on both sides of the deleted or modified part of the respective gene. As an exception, pKEM5 was constructed by introduction of the deletion through religation of the 5′ phoshorylated internal oligonucleotides. For the mutator strains introducing EGFP, a precursor mutator vector was created as described above. Subsequently, the EGFP gene was inserted in frame from plasmid pEGFP‐C1 into the digested precursor vectors. The respective regions containing the flanking sequences were subcloned into the pKNG101 suicide vector. For pMA12, insert 2 was created by overlapping PCR using oligos 5017 and 5087 to amplify mCherry from vector pRVCHYC‐5 (Thanbichler et al, 2007), and oligos 5088 and 5068 to amplify the downstream flanking region from the pYV plasmid. Afterwards, ligation of SalI/XhoI digested insert 1, containing the upstream flanking region, XhoI/XbaI digested insert 2, and SalI/XbaI digested pKNG101 suicide vector lead to the mutator pMA12. All constructs were confirmed by sequencing using a 3100‐Avant genetic analyser (Applied Biosystems, Rotkreuz, Switzerland). The allelic exchange was selected by plating diploid bacteria on sucrose (Kaniga et al, 1991). pAD166 expressing YscNK175E was generated by overlapping PCR using internal primers encoding for the modified protein sequence, and selected by colony PCR and sequencing. For pAD182 expressing EGFP–YscN, a precursor vector was generated and EGFP was introduced from pEGFP‐C1, as described above.
Y. enterocolitica cultures for secretion and microscopy analysis
Induction of the yop regulon was performed by shifting the culture to 37°C, either in BHI‐Ox (secretion‐permissive conditions) or in BHI+5 mM CaCl2 (secretion‐non‐permissive conditions) (Cornelis et al, 1987). Expression of the inducible YscN constructs was induced by adding 0.05% l‐arabinose to the culture just before the shift to 37°C. The carbon source was glycerol (4 mg/ml) when expressing genes from the pBAD promoter, and glucose (4 mg/ml) in the other cases.
Total cell and supernatant fractions were separated by centrifugation at 20 800 g for 10 min at 4°C. The cell pellet was taken as total cell fraction. Proteins in the supernatant were precipitated with trichloroacetic acid 10% (w/v) final for 1 h at 4°C.
Secreted proteins were analysed by SDS–PAGE; in each case, proteins secreted by 3 × 108 bacteria were loaded per lane. Total secreted proteins were analysed by Coomassie staining of 12% SDS–PAGE gels. Detection of specific secreted proteins by immunoblotting was performed using 15% SDS–PAGE gels. For detection of proteins in total cells, 2 × 108 bacteria were loaded per lane, if not stated otherwise, and proteins were separated on 15% SDS–PAGE gels before detection by immunoblotting.
Immunoblotting was carried out using rabbit polyclonal antibodies against LcrV (MIPA220; 1:2000), YscF (MIPA223; 1:1000), YscN (MIPA189; 1:1000), YscP (MIPA57; 1:3000), or YopE (MIPA73; 1:1000). Detection was performed with secondary antibodies directed against rabbit antibodies and conjugated to horseradish peroxidase (1:5000; Dako), before development with ECL chemiluminescent substrate (Pierce).
Needles were purified from cultures incubated under secretion‐permissive conditions. At the given time points, 48 ml bacteria were removed from the 500 ml culture, harvested by centrifugation (5 min at 4000 g) and resuspended in 1 ml 20 mM Tris–HCl, pH 7.5. Needle detachment was increased by repeated pipetting through a 1 ml pipet tip. Cells were pelleted by centrifugation (5 min at 4000 g), and the supernatant containing the needles was passed through a 0.45 μm mesh filter (cellulose acetate membrane) and then centrifuged for 60 min at 20 800 g. The resulting pellet was resuspended in 20 μl Laemmli buffer, 15 μl of which were analysed by SDS–PAGE followed by immunoblotting (Mueller et al, 2005).
For fluorescence imaging, cells were placed on a microscope slide layered with a pad of 2% agarose dissolved in water or PBS. A Deltavision Spectris optical sectioning microscope (Applied Precision, Issaquah, WA) equipped with an UPlanSApo 100 × /1.40 oil objective (Olympus, Tokyo, Japan) and a coolSNAP HQ CCD camera (Photometrics, Tucson, AZ) was used to take differential interference contrast (DIC) and fluorescence photomicrographs. To visualize GFP and mCherry fluorescence, GFP filter sets (Ex 490/20 nm, Em 525/30 nm) and mRFP filter sets (Ex 560/40 nm, EM 632/60 nm), respectively, were used. DIC frames were taken with 0.3 s and fluorescence frames with 1.0 s exposure time. Per image, a Z‐stack containing 20 frames per wavelength with a spacing of 150 nm was acquired. The stacks were deconvoluted using softWoRx v3.3.6 with standard settings (Applied Precision, WA). The DIC frame at the centre of the bacterium and the corresponding fluorescence frame were selected and further processed with ImageJ software.
Co‐immunoprecipitation of YscC, YscD, and YscJ
Y. enterocolitica cultures were grown in secretion‐non‐permissive conditions to an OD600 of 1.5–2.2. Protein complexes were then stabilized by crosslinking with 0.25% formaldehyde for 15 min at 37°C. Cells were harvested by centrifugation (15 min at 1500 g, 25°) and resuspended in 1/5 volume of PBS. After a second crosslinking step (0.4% formaldehyde, 15 min, 25°C) and harvesting as before, spheroplast generation and lysis was performed as described by Kubori et al (1997) and Blocker et al (2001). In short, cells were resuspended in 1/5 original volume of ice‐cold spheroplasting buffer (0.75 M sucrose, 50 mM Tris, pH adjusted with HCl to 7.8, 0.6 mg/ml lysozyme, 6 mM EDTA), and incubated at 25°C up to 90 min, until complete spheroplast formation could be observed. Cells were lysed by addition of 1% Triton X‐100 and subsequent incubation at 4°C for 15 min. After addition of 15 mM MgCl2, unlysed cells were removed by centrifugation (20 min at 6000 g, 4°C); 300 μl of anti‐FLAG M2 affinity gel (Sigma‐Aldrich, Buchs, Switzerland) were added to the supernatant, and the proteins were purified in batch according to the manufacturer's protocol. The elution fractions were recentrifuged to completely remove resin, and separated on 12% SDS–PAGE gels or 4–12% gradient SDS–PAGE gels (Serva, Heidelberg, Germany). Immunoblotting was carried out using rabbit polyclonal antibodies against YscC (MIPA250, 1:1000), YscD (MIPA232, 1:1000), and YscJ (MIPA66, 1:5000), as described above.
Supplementary data are available at The EMBO Journal Online (http://www.embojournal.org).
Conflict of Interest
The authors declare that they have no conflict of interest.
Supplementary Movie 1
Supplementary Movie legend
We thank K Maylandt and I Stainier for providing strains pKEM5, pKEM4001, pSI51, and pSI4006. This work was supported by the Swiss National Science Foundation (grant 310000‐113333/1) to GC.
- Copyright © 2010 European Molecular Biology Organization