A dominant-negative needle mutant blocks type III secretion of early but not late substrates in Yersinia

Summary

Yersinia pseudotuberculosis uses a Type III Secretion System (T3SS) to deliver effectors into host cells. A key component of the T3SS is the needle, which is a hollow tube on the bacterial surface through which effectors are secreted, composed of the YscF protein. To study needle assembly, we performed a screen for dominant-negative yscF alleles that prevented effector secretion in the presence of wild-type (WT) YscF. One allele, yscF-L54V, prevents WT YscF secretion and needle assembly, although purified YscF-L54V polymerizes in vitro. YscF-L54V binds to its chaperones YscE and YscG, and the YscF-L54V-EG complex targets to the T3SS ATPase, YscN. We propose that YscF-L54V stalls at a binding site in the needle assembly pathway following its release from the chaperones, which blocks the secretion of WT YscF and other early substrates required for building a needle. Interestingly, YscF-L54V does not affect the activity of pre-assembled actively secreting machines, indicating that a factor and/or binding site required for YscF secretion is absent from T3SS machines already engaged in effector secretion. Thus, substrate switching may involve the removal of an early substrate-specific binding site as a mechanism to exclude early substrates from Yop-secreting machines.

Introduction

Type III Secretion Systems (T3SSs) comprise a family of proteinaceous machines dedicated to the export of proteins from the cytosol of gram-negative bacteria (Cornelis, 2006, Macnab, 2003), including virulence-associated machines that make a needle complex and motility-associated machines that produce flagella. The major exported protein of flagella is the flagellin monomer, which remains associated with the bacterial cell. Both symbionts and pathogens use the T3SS to export proteins that build needle complexes. Needle complexes in turn function to inject effector proteins into targeted host cells (Cornelis, 2006, Galan & Wolf-Watz, 2006). Effectors modify a variety of signaling pathways, permitting the bacteria to establish a niche in the host (Coburn et al., 2007).

The virulence T3SS has three main structural elements: a base spanning both bacterial membranes, a needle-like tube that extends outward from the base, and a secreted translocon that forms a pore in the host cell plasma membrane (Moraes et al., 2008, Mueller et al., 2008). The needle forms a conduit from the base to the translocon, allowing for the polar movement of effectors into the host cell (Li et al., 2002, Lee et al., 1998). Needles are essential for effector secretion and are formed by the polymerization of a single small protein, called YscF in Yersinia (Hoiczyk & Blobel, 2001). The polymerized Yersinia needle is comprised of approximately 140 YscF subunits (Broz et al., 2007), creating a tube ~ 55 nm in length with interior and external diameters of around 2.5 nm and 7 nm, respectively (Hoiczyk & Blobel, 2001).

The biogenesis of the needle is a multi-step process that begins with the synthesis of the needle subunit and its chaperones. The needle monomer is found in a 1:1:1 complex with two cytosolic chaperones (Quinaud et al., 2005), named YscE and YscG in Yersinia, and chaperone binding prevents inappropriate association and polymerization of the needle subunits in the cytosol (Quinaud et al., 2005). The crystal structures of needle-chaperone complexes from Yersinia and Pseudomonas (called PscF, PscE and PscG) have recently been solved, and show that YscG/PscG binds the C-terminal helix of YscF/PscF (Quinaud et al., 2007, Sun et al., 2008). The structures also indicate that YscE does not interact with YscF directly; instead YscE appears to stabilize YscF-YscG interactions. All three proteins are required to form a stable complex, and in the absence of either chaperone, the needle monomer is rapidly degraded (Tan et al., 2008, Sun et al., 2008, Betts et al., 2008, Quinaud et al., 2005, Quinaud et al., 2007). Other proteins that polymerize on the bacterial surface, such as pilins, the flagellar hook and filament subunits, and the EPEC needle filament protein EspA also associate with chaperones to maintain a monomeric state prior to their secretion (Hung & Hultgren, 1998, Auvray et al., 2001, Bennett et al., 2001, Creasey et al., 2003). For all monomer-chaperone systems studied so far, including YscF-EG, the chaperones remain inside the cell after the monomer is secreted (Bennett & Hughes, 2000).

The remaining steps of needle biogenesis have not been shown experimentally, but must include the following: targeting of the needle-chaperone complex to a docking site at the T3SS base, release of the monomer from the chaperones, assembly of monomers into the T3SS base, and polymerization of monomers outward from the base. It is unclear if the base-targeting information resides within the needle, on the chaperones, or both, but the first 15 codons of YscF are required for its secretion (Pastor et al., 2005). A docking site for the needle-chaperone complex has not been identified, although a likely candidate is the T3SS-associated ATPase, YscN. The T3SS ATPase is required for docking and disassembly of chaperone-effector complexes (Gauthier & Finlay, 2003, Akeda & Galan, 2005), and the flagella-associated ATPase FliI performs the same function for flagellin and its chaperones (Minamino & Namba, 2008, Thomas et al., 2004). Upon release from the chaperones, needle monomers presumably move through the T3SS entrance in the inner membrane to an assembly site within the base. Elegant cryo-EM images of purified secretion apparatuses from Salmonella revealed that the needle structure is present midway through base and extends through the outer membrane secretin ring into the extracellular space (Marlovits et al., 2004). The PrgJ/YscI protein forms a rod-like structure inside the base (Marlovits et al., 2004), which is required for needle stability and may function as a platform for the initiation of needle polymerization (Marlovits et al., 2006, Wood et al., 2008). The needle likely grows outward with new monomers adding at the distal tip, similar to the elongation mechanism of the flagella (Emerson et al., 1970, Iino, 1969). Finally, a pentamer of the LcrV protein assembles on the distal tip of the needle (Mueller et al., 2005, Broz et al., 2007). LcrV is required for formation of the translocation pore (Marenne et al., 2003, Goure et al., 2004), but is not involved in needle biogenesis per se as cells lacking LcrV produce normal looking needles (Mueller et al., 2005).

Needles are built to a specific length (Minamino & Pugsley, 2005, Cornelis et al., 2006) and proper needle length is important for efficient effector translocation into host cells (Mota et al., 2005). Following needle assembly, bacteria switch from the secretion of early substrates, including YscF, YscP, YscI and YopR, to the secretion of the needle tip and translocon components LcrV, YopB and YopD (middle substrates), and then the effectors (late substrates) (Riordan et al., 2008, Sorg et al., 2006). This process is termed substrate switching, and is also used in construction of the flagella (Ferris & Minamino, 2006). It is not clear how the bacteria sense that needle assembly is completed, or how this information is relayed and processed to result in substrate switching, but proteins in the YscP, YscU and YscI families are involved (Agrain et al., 2005, Magdalena et al., 2002, Edqvist et al., 2003, Sorg et al., 2007, Wood et al., 2008, Lorenz et al., 2008, Morello & Collmer, 2009, Marlovits et al., 2006). Effector secretion is triggered by host cell contact (Sory & Cornelis, 1994, Rosqvist et al., 1994), and in Yersinia, can also be artificially induced by depletion of calcium from growth medium (Bolin & Wolf-Watz, 1988).

To investigate the mechanism of needle assembly, we developed a screen to identify mutant yscF alleles that are dominant-negative for Yop secretion. The yscF-L54V mutant described here blocks export of WT YscF and the formation of needles. This block is specific to the synthesis of new needles since actively secreting needle complexes are not blocked by YscF-L54V. These results indicate that needle-chaperone complexes are no longer substrates for T3SS machines that have switched to secretion of late substrates.

Results

Screen to identify yscF alleles Dominant-Negative for Yop secretion

To investigate the mechanism of T3SS needle assembly, we developed a screen to identify dominant-negative (DN) yscF alleles using Yop secretion as our readout for a functional needle. We chose this strategy because a DN YscF mutant must be functional enough to interfere with WT YscF function, and it is therefore likely to interfere with a crucial step of the needle assembly pathway. The DN screen exploited a colorimetric plate assay to identify yscF mutants that could not secrete effectors (called Yops in Yersinia) in the presence of a WT copy of yscF. Yersinia pseudotuberculosis (Yptb) secretes Yops when grown at 37°C in media containing low concentrations of calcium. In the presence of the dye Congo Red, colonies of Yptb that secrete Yops absorb the dye and turn red, whereas Yptb mutants that cannot secrete Yops remain white. A plasmid library of IPTG-inducible, randomly mutagenized yscF alleles (Davis & Mecsas, 2007) was introduced into the wild-type Yptb IP2666 strain (WT). Single transformants were patched onto Congo Red plates, incubated at 37°C overnight, and scored for red or white color. Of the ~3000 strains screened, 44 were white on Congo Red plates, indicating that they could not secrete Yops. Plasmids were isolated from the putative DN strains, re-introduced into the WT strain and tested again on Congo red plates. Three strains showed a reproducible white colony phenotype, and were named DN6, DN24 and DN44.

Sequencing the DN44 mutant revealed a single point mutation that changed residue 54 from leucine to valine (L54V). The leucine 54 residue in YscF is conserved in the closely related needle proteins of Yersinia spp., Pseudomonas spp., Aeromonas spp., and Vibrio spp., but is a tyrosine or phenylalanine residue in the more distantly related needle proteins of E. coli spp., Salmonella spp., Shigella spp. and Burkholderia spp. The DN6 and DN24 mutants each contained multiple amino acid changes and will be described elsewhere.

YscF-L54V blocks WT YscF secretion and Yop secretion

WT and ΔyscF strains expressing yscF-L54V were tested for the ability to secrete Yops by growth at 37°C in media depleted of calcium. Culture supernatants were collected and secreted proteins were visualized by SDS-PAGE and coomassie staining. As shown previously, the ΔyscF strain did not secrete Yops, and secretion was restored in the ΔyscF strain by the expression of yscF on a plasmid (Torruellas et al., 2005, Davis & Mecsas, 2007) (Figure 1A, top panel, lanes 4–5). The WT strain expressing the YscF-L54V protein did not secrete Yops (Figure 1A, top panel, lane 3), confirming that this mutation was DN for Yop secretion. In addition, the YscF-L54V mutant did not complement the yscF null strain for Yop secretion (Figure 1A, top panel, lane 6).

Figure 1.

The yscF L54V mutant prevents secretion of WT YscF, needle formation and secretion of Yops. WT or ΔyscF strains carrying plasmids pTRC99-yscF, pTRC99A or pTRC99-yscF-L54V were grown in low calcium media and shifted to 37°C. Expression of yscF or yscF-L54V from the plasmids was induced with IPTG at the 37°C temperature shift. Cell-associated and secreted proteins harvested from equal numbers of bacteria were separated by SDS-PAGE and detected by coomassie staining or by Western blotting. A. No Yops or YscF are secreted in the presence of yscF-L54V. Secreted proteins visualized by coomassie staining (top panel). Secreted (middle panel) and cell-associated (lower panel) YscF levels detected by Western blot with α-YscF antiserum. B. YscF-L54V protein does not make needles and prevents WT YscF needle formation. Equal numbers of bacteria were collected and exposed to water or the chemical cross-linker BS³ (−/+ BS³). Bacteria were solubilized in SDS-sample buffer and proteins were separated by SDS-PAGE. YscF protein was detected by Western blotting with antibodies to YscF. YscF monomers are ~ 7 kDa and YscF polymers run as a ladder of higher molecular weight bands (polymers). Non-specific antibody-reactive bands are apparent in the no-cross-linker control samples (odd numbered lanes). C. Expression of yscF-L54V decreases levels of T3SS-associated proteins. Cellular levels of YopE, YopD and YscN were visualized by Western blotting. Molecular mass markers are shown on the left in kDa in A and B.

Cell-associated and secreted YscF protein levels were examined by Western blot analysis using YscF antibodies. In WT strains, YscF was detected in whole cell lysates and was secreted into the culture supernatant (Figure 1A, lower panels, lanes 1, 2 and 4). In ΔyscF strains, YscF-L54V was also detected in whole cell lysates and was more abundant than WT YscF (Figure 1A, bottom panel, lanes 4 and 6), indicating that the YscF-L54V protein was stable. However, YscF-L54V was not secreted when expressed on its own (Figure 1A, middle panel, lane 6), and neither WT YscF nor YscF-L54V was secreted in strains expressing yscF-L54V (Figure 1A, middle panel, lane 3). Thus YscF-L54V blocked both YscF and Yop secretion.

yscF-L54V expression prevents needle formation

A DN needle protein could block Yop secretion by producing non-functional needles, but the lack of secreted YscF strongly implied that needles might not assemble at all. We tested whether the YscF-L54V protein formed needles on the bacterial surface by treating Yptb with the membrane impermeable chemical cross-linker BS³, which covalently links neighboring YscF molecules (Torruellas et al., 2005, Davis & Mecsas, 2007). Cross-linked YscF monomers appear as ladder-like pattern of high molecular weight bands when visualized by Western blotting with YscF antibodies (Figure 1B, lanes 2, 4 and 8). When the YscF-L54V protein was produced in either the WT or ΔyscF backgrounds, no external YscF multimers were observed (Figure 1B, lanes 6 and 12). Taken together, these data show that the block in WT YscF secretion caused by the YscF-L54V protein prohibited needle formation and consequent Yop secretion.

yscF-L54V expression prevented an increase in the production of Yops when grown in low calcium

When WT strains are exposed to secretion-inducing conditions, the synthesis of Yops and structural T3SS components is greatly increased, in part due to the secretion of the negative regulators LcrQ and YopD (Pettersson et al., 1996, Williams & Straley, 1998). Strains that lack a functional secretion apparatus fail to increase the synthesis of these components (Figure 1C, compare lanes 2 and 5; (Michiels et al., 1991, Allaoui et al., 1995, Plano & Straley, 1995)). Expression of yscF-L54V in both WT and ΔyscF strains in low calcium media at 37°C resulted in decreased cellular stores of the effectors YopE and YopH, as well as the translocators YopD and LcrV, the T3SS machine component YscN, and the early substrates YscI and YscP compared to WT (Figure 1C, and data not shown). These data support the hypothesis that expression of yscF-L54V results in a non-functional secretion apparatus and that this defect prohibits the low calcium-mediated induction of secretion machinery components and Yops.

YscF-L54V blocks needle formation in high calcium

WT Yersinia grown at 37°C in media supplemented with calcium are in a “primed” state where needle assembly occurs, and the Yops are synthesized but are not secreted (Stainier et al., 1997). Unlike growth in low calcium, growth at 37°C in high calcium does not induce turnover of secretion system components or Yops in bacteria that harbor defective T3SS machinery (Bergman et al., 1994). We tested if yscF-L54V expression affected synthesis and/or stability of the Yops or needle formation in the presence of calcium. Equal levels of Yops and YscN were apparent in WT, ΔyscF and strains expressing YscF-L54V when grown in high calcium media at 37°C (Figure 2A), showing that YscF-L54V did not promote the turnover of Yops or machinery components under these conditions. However, chemical cross-linking indicated that needles were still absent in YscF-L54V producing strains (Figure 2B, lanes 6 and 12). Thus YscF-L54V prevented YscF secretion and needle formation under conditions where T3SS components and Yops are present in the cytosol.

Figure 2.

yscF-L54V expression prevents needle assembly during growth in high calcium media. WT or ΔyscF strains carrying plasmids pTRC99-yscF, pTRC99A or pTRC99-yscF-L54V were grown in media supplemented with 3 mM calcium and shifted to 37°C. Expression of yscF from plasmids was induced with IPTG at the 37°C temperature shift. A. Expression of yscF L54V in high calcium does not affect cellular levels of T3SS-associated proteins. YscF, YopE, YopD and YscN proteins were visualized by Western blotting. B. No external YscF polymers are present in yscF L54V-containing strains. Equal numbers of bacteria were cross-linked with BS³ and processed as in Figure 1B. YscF protein was detected by Western blotting. C. No external YscF protein is present in YscF-L54V-containing strains. Bacteria were fixed without permeabilization, adhered to glass coverslips, and stained with antibodies to YscF followed by Alexa592 secondary antibodies (red) as described in the Experimental Procedures. DAPI (blue) was used to detect nucleoids. Samples were visualized by fluorescence microscopy.

To rule out the possibility that the YscF-L54V protein formed polymers that were not accessible to cross-linking with BS³, we looked for external, cell-associated YscF protein by indirect immunofluorescence. Yersinia strains were grown in media supplemented with calcium at 37 C, fixed without permeabilization, and probed with YscF antiserum followed by a fluorescent secondary antibody. WT Yersinia and the complemented ΔyscF strain showed many fluorescent foci surrounding the bacterial surface (Figure 2C). These foci were not detected in the absence of the YscF protein (Figure 2C, ΔyscF + vector), or on Yersinia expressing yscF-L54V (Figure 2C, right panels), suggesting that all cell-associated YscF and YscF-L54V protein in the DN strain is intracellular. Consistent with this, permeabilized cells of the DN strain showed strong cytoplasmic staining with YscF antiserum (data not shown), confirming that the punctate staining pattern represents extracellular YscF. Additionally, no extracellular YscF was observed after growth of YscF-L54V-expressing strains in low calcium (data not shown). The absence of needles on bacteria expressing YscF-L54V was further confirmed by examination of Yersinia by transmission electron microscopy (TEM; data not shown). Taken together, the cross-linking, immunofluorescence and TEM data indicate that the YscF-L54V protein does not form polymers on the surface of the bacterium, and that expression of YscF-L54V in the presence of WT YscF prevents formation of WT needles regardless of the growth conditions. Consistent with data indicating that the yscF-L54V mutant did not form needles or secrete Yops when induced in secretion media, strains expressing yscF-L54V did not translocate Yops into tissue culture cells (data not shown). We conclude that the primary effect of yscF-L54V expression is a block in YscF secretion, which leads to a number of secondary phenotypes including blocked needle assembly on the cell surface and a failure to increase intracellular pools of Yops when secretion is induced.

YscF-L54V polymerizes in vitro

Since needles were not detected in strains producing the YscF-L54V protein, it was possible that YscF-L54V had a polymerization defect. A DN phenotype would then result if the YscF-L54V mutant assembled into the base, but could not polymerize past the first ring of monomers, as WT YscF would be excluded from the base and needle formation in the WT background would be obstructed. To test this possibility, we examined if purified YscF-L54V protein was able to polymerize in vitro. WT YscF and YscF-L54V were expressed in E. coli as GST fusion proteins and purified using standard glutathione affinity chromatography (Figure 3A, see Experimental Procedures for details), resulting in a >95% pure solution of YscF or YscF-L54V protein (Figure 3A, Q FT). In all purification steps, YscF-L54V behaved identically to WT YscF (data not shown).

Figure 3.

YscF-L54V protein polymerizes in vitro. A. Recombinant YscF protein purification profile. Commassie stained gel of samples taken during the purification of YscF (see Experimental Procedures for details). YscF (shown) and L54V proteins were over-expressed in E. coli as GST fusions, purified from the soluble fraction (supe) using glutathione-agarose, and eluted with free glutathione (elute). Following dialysis to remove the glutathione (dialysis), YscF was released from GST by Factor Xa cleavage (Xa). The GST and YscF mixture was concentrated (conc.) and diluted into 10 mM Tris pH 8.0 (dilute). GST was removed with Q sepharose (Q FT), leaving free YscF. Purification of the YscF-L54V protein proceeded identically. B–E. WT YscF and YscF-L54V proteins polymerize in vitro. Purified YscF (B, C) or YscF-L54V (D, E) proteins were mounted onto carbon-coated EM grids and negatively stained with uranyl acetate immediately after purification (B) or after incubation at room temperature for 16 hrs (C–E). Samples were visualized by transmission electron microscopy. Bars are 200 nm (B–D) and 500 nm (E).

YscF and YscF-L54V protein preparations were examined by negative staining and TEM. Immediately following purification small aggregates of YscF or YscF-L54V were observed (Figure 3B), and occasionally a long polymer was also seen. After incubation at room temperature (RT) for 12–16 hrs, long polymers (> 1μm) were detected in both the YscF and YscF-L54V samples (Figure 3C–E). Additionally, the small protein aggregates were much less abundant after the long incubation, suggesting they may be an intermediate to polymerization. Polymers were sometimes found in small or large bundles twisted around each other (Figure 3C–E). The YscF-L54V polymers were visually indistinguishable from those made from WT YscF protein by TEM, and polymers derived from either protein were the same diameter as needles composed of WT YscF observed on the bacterial surface (data not shown). At high magnification, negative stain was observed in the center of the polymers (data not shown), indicating that the polymers were in fact hollow tubes. Thus the basic structures formed by the needle proteins in vitro and in vivo were very similar.

YscF-L54V appeared to form polymers more readily than WT, as polymers were more frequently found immediately after purification of YscF-L54V than with WT, despite the same input concentration of protein. YscF-L54V polymers were also more abundant than WT polymers, and regularly formed very large nets, many microns long following overnight incubations at RT (Figure 3E). The fact that YscF-L54V polymerized more readily than WT YscF in vitro introduced the possibility that YscF-L54V could form polymers in the bacterial cytosol prior to interactions with its chaperones. However, we did not detect YscF-L54V polymers in the Yersinia cytosol using membrane permeable cross-linkers (data not shown). In conclusion, the YscF-L54V protein was clearly able to polymerize in vitro as well as WT YscF, suggesting that YscF-L54V could polymerize to form a needle if it was correctly assembled into the base. Therefore, either YscF-L54V assembles incorrectly within the base and prevents WT YscF assembly and polymerization, or YscF-L54V does not have access to the inner core of the base and prevents access of WT YscF.

L54V binds to YscEG

Access to the inner core of the secretion base requires that YscF is targeted to the base and released from its chaperones. The crystal structure YscF bound to YscE and YscG (hereafter referred to as YscF-EG) reveals that leucine 54 of YscF is located at the hydrophobic interface between YscF and YscG (Figure 4A, L54 residue in red). Changing L54 to valine in silico revealed potential steric clashes with several residues in helix α-1 of YscG, raising the possibility that the L54V mutation altered the conformation of the YscF-EG structure. Several DN effects are conceivable if the YscF-L54V-EG structure is different than WT YscF-EG. For example, if YscF-L54V bound too tightly to YscE and YscG such that the chaperones could not be stripped off, the chaperones would remain associated with YscF-L54V in the cytosol, eventually titrating the entire pool of chaperones away from WT YscF, leaving “naked” WT needle monomers subject to degradation. Alternatively, if changes in the conformation of the needle-chaperone complex resulted in aberrant and/or unproductive interactions with other components of the T3SS, either in the cytosol or after targeting to the base, WT YscF-EG could be excluded from access to that component and would remain stalled at a previous step in the assembly pathway.

Figure 4.

Residue L54 resides in the YscF-YscG hydrophobic pocket but does not disrupt complex formation. A. Location of YscF residue L54 within the YscF-EG complex. The crystal structure model of the YscF-EG complex is shown; YscG is in grey with the α-1 chain labeled, YscE is in violet and YscF is in green. The leucine 54 side chain of YscF is highlighted in red. Figure was prepared using PyMol (DeLano, 2008). B. YscF-L54V forms a complex with YscE and YscG in Yersinia. Virulence plasmid- cured Yersinia strains (pYV-) containing pFlag-YscG-YscE-His and either pBAD-yscF or pBAD-yscF-L54V were induced to co-express Flag-YscG, YscE-His and either YscF or YscF-L54V by addition of IPTG and arabinose. Bacteria were harvested, lysed, and the soluble fraction (start) was precipitated with Flag antibody-conjugated agarose beads. The flow-through (FT) fraction was collected, and the beads were washed thoroughly (wash). Co-precipitated protein complexes were eluted by boiling in SDS sample buffer (elute). Proteins in the different fractions were separated by SDS-PAGE, and detected by Western blot with antibodies to Flag (YscG), His (YscE) and YscF.

To determine if DN effects of needle-chaperone interactions were plausible, we first sought to determine if YscF-L54V bound to its chaperones. Epitope-tagged alleles of YscG (Flag-YscG) and YscE (YscE-His), and either YscF or YscF-L54V were co-expressed in a Yersinia strain lacking the virulence plasmid (pYV⁻). The pYV⁻ strain background was chosen such that YscF-EG interactions could be initially investigated in isolation without competition or complication from other T3SS components. YscG was immuno-precipitated from the soluble cytosolic fraction, and probed for the presence of YscF, YscG and YscE by Western blotting. Both WT YscF and YscF-L54V associated with YscG-Flag and YscE-His (Figure 4B, lanes 4 and 8), confirming that YscF-L54V did interact with its chaperones. Co-precipitation of YscF or YscF-L54V and YscE-His with YscG-Flag was fairly efficient, as most of the proteins eluted with YscG-Flag. However, some of the YscF, YscF-L54V and YscE-His remained in the flow-through fraction, likely because the YscE, F and G proteins are not co-expressed at exactly 1:1:1 levels (data not shown). Co-immunoprecipitation of YscF, YscE and YscG were also observed when the proteins were co-expressed in E. coli (data not shown), confirming previous work that established that needle-chaperone complex formation does not require additional T3SS components (Quinaud et al., 2005, Sun et al., 2008, Tan et al., 2008).

Over-expression of WT YscF-EG complexes suppressed the YscF-L54V secretion defects

To determine if YscF-L54V-EG could be out-competed by WT YscF-EG, we asked if increasing WT YscF and/or YscEG levels could suppress the YscF-L54V secretion defect. First, the L54V mutation was introduced into the yscF locus on the virulence plasmid, creating a genomic yscF-L54V strain. As expected, the yscF-L54V strain did not secrete YscF, Yops or make cross-linkable needles (Figure 5A, B and C, lane 5). Cellular amounts of YscF-L54V protein in the yscF-L54V strain were significantly less than WT YscF levels in the WT strain (Figure 5B, lanes 1 and 5), signifying that the yscF locus was susceptible to feedback inhibition when exposed to secretion inducing conditions.

Figure 5.

Over-expression of WT YscF-EG proteins suppresses YscF-L54V DN phenotypes. WT or yscF-L54V strains containing pBAD33 (lanes 1 and 5), pBAD-yscF (lanes 2 and 6), pFlag-YscG-YscE-His and pBAD33 (lanes 3 and 7) or pFlag-YscG-YscE-His and pBAD-yscF (lanes 4 and 8) were grown in low-calcium media and expression from plasmids was induced at the 37°C temperature shift by addition of arabinose, IPTG, or both. Bacteria and culture supernatants were collected and proteins separated by SDS-PAGE. A. Secreted Yops detected by coomassie staining. B. Cell-associated (cells) and secreted YscF protein from above strains detected by Western blot with α-YscF antibodies. C. Above strains were chemically cross-linked with BS³ (lanes 1–8). Cross-linked samples and non-cross-linked WT bacteria (BS³) were solubilized in SDS-sample buffer and separated by SDS-PAGE. YscF monomer and polymers were visualized by Western blot with antibodies to YscF. Molecular mass markers for A and C are shown on the left in kDa.

YscF secretion, Yop secretion and needle formation in the yscF-L54V strain was examined following over-expression of WT YscF alone, YscEG alone, or co-expression of WT YscF and YscEG together. Increased production of YscF and YscEG together in the yscF-L54V strain relieved the secretion block of both YscF and the Yops (Figure 5A and B, lane 8), and restored YscF polymers to WT levels (Figure 5C, lane 8). In contrast, increased expression of just the chaperones in the yscF-L54V strain did not permit secretion of the YscF-L54V protein, formation of a needle, or secretion of Yops (Figure 5A–C, lane 7). This observation suggests that YscE and YscG levels are not limiting in the yscF-L54V strain even though the yscE and yscG loci are likely down-regulated under low calcium conditions. Furthermore, these data demonstrate that YscEG only suppresses YscF-L54V activity when WT YscF is also present. Interestingly, over-expression of WT YscF in the yscF-L54V background resulted in very low but detectable amounts of secreted YscF, secreted Yops, and YscF polymers (Figure 5A–C, lane 6). In summary, only the addition of excess WT YscF and YscEG fully suppressed yscF-L54V. Neither expression of YscEG alone nor YscF alone could overcome the YscF-L54V secretion block. This is in contrast to the response of WT Yersinia, where increasing the amount of chaperones alone led to greatly increased YscF secretion and polymers (Figure 5B and C, compare lanes 1 and 3).

These results indicated that YscF-L54V affects some function of the YscF-EG complex, and that the YscF-L54V-mediated block is reversible. These data are consistent with previously mentioned models for secretion blockage by YscF-L54V, including titration of YscEG from WT YscF, which would inhibit the “bodyguard” function of the chaperones, or unproductive complex formation of YscF-L54V-EG with another T3SS component in the cytosol or at the cytoplasmic entrance to the base, which would disrupt a targeting function of the chaperones. Alternatively, it is also possible that YscF-L54V irreversibly blocks the secretion apparatus after unproductive assembly inside the base, and that excess YscF-EG complexes are able to overcome this block by out-competing YscF-L54V-EG for access to newly constructed T3SS bases by flooding the system with an excess secretion-competent WT YscF.

WT YscF-EG and YscF-L54V-EG have similar biochemical properties in solution

Both the YscF-L54V and YscF proteins associated with their chaperones in the Yersinia cytosol, and yet the suppression data above suggested that the two needle-chaperone complexes could be functionally different. We hypothesized that the YscF-L54V mutant altered the conformation and/or the physical properties of the YscF-L54V-EG complex, which could result in aberrant interactions with other T3SS components. To test this, we compared the stability and conformation of the WT and mutant needle-chaperone complexes.

The YscF-EG complex was purified from Yersinia using Flag affinity beads as described for Figure 4B. YscF-EG and YscF-L54V-EG complexes were separated from monomeric YscG-Flag and from Flag antibody and Flag peptide contaminants using gel filtration. The elution profiles of the two complexes were very similar; both showed 4 major peaks (Figure 6A, P1, P2, P4 and P5) and the YscF-EG profile had one additional minor peak (Figure 6A, P3). The protein composition in each peak was examined by coomassie staining and by Western blot (Fig 6B and data not shown), and was the same for both purifications. The peaks included large and aggregated proteins (P1), a proteolytically clipped YscF-EG complex (P3), monomers of Flag-YscG and YscE-His (P4), and an uncharacterized 4 kDa protein fragment (P5). Peak 2 samples contained YscF, YscE and YscG at an approximately 1:1:1 ratio (Figure 6B, P2), and eluted as the most abundant peak with a molecular weight of ~ 38 kDa (Fig 6A, P2), slightly larger than the sum of the predicted sizes of the three individual proteins (32 kDa). The fact that the two intact YscF-EG and YscF-L54V-EG complexes eluted at the same volume by gel filtration revealed that they had similar shapes and masses. The peak 2 fractions were characterized below.

Figure 6.

Purification and biochemical analysis of YscF-EG and YscF-L54V-EG complexes. A. Gel filtration elution traces of WT YscF-EG and L54V-EG complexes. Flag-affinity purification eluates were passed over a Sephacryl S-100 HR gel filtration column. Five peaks were observed (P1–P5) for YscF-EG (blue) and YscF-L54V-EG (red) samples. The Y-axis shows the absorbance at 280 nm. The X-axis shows the elution volume in ml. The column was calibrated with albumin (67 kDa), ovalbumin (43 kDa), chymotrypsinogen A (25 kDa), and Ribonuclease A (13.7 kDa). B. The eluate and peak fractions (P1–P4) from A were concentrated, proteins were separated by SDS-PAGE, and protein composition in peak samples was analyzed by coomassie staining. C. Peak 2 and peak 4 samples from both YscF-EG (WT) and YscF-L54V-EG (L54V) were analyzed by native PAGE and Western blotting. Migration as a complex (closed arrowhead) or a monomer (open arrowhead) is indicated. D. Protease protection assay. Three μg of YscF-EG or YscF-L54V-EG peak 2 complexes (lanes 1 and 7) was incubated with 1 μg, 0.2μg, 0.04 μg or 0.008 μg of thermolysin (lanes 2–5, and 8–11) or with 1 μg of thermolysin and 1% TritonX-100 (lanes 6 and 12) for 1 hr. at 37°C. Proteins were separated by SDS-PAGE and stained with coomassie. Full-length Flag-YscG (G), YscE-His (E) and YscF (F) proteins are labeled. E. Thermal denaturation curves. Purified YscF-EG (blue lines), YscF-L54V-EG complexes (red lines) and the P4 fraction from the WT YscF-EG purification in Fig 6A (black line) were exposed to a temperature gradient of 1°C per min in the presence of the dye Sypro Orange. Fluorescence intensity was monitored as the temperature increased. The fluorescence minimum and maximum for each sample was set to 0 and 1, respectively, and the normalized data are shown as relative fluorescence. Needle-chaperone complex samples were tested in triplicate and all curves are shown.

Analysis of both the WT YscF-EG and YscF-L54V-EG complexes by native-PAGE showed that YscF, YscG and YscE co-migrated as a single high molecular weight species (Figure 6C, P2), confirming that all of the YscF, YscE and YscG in these samples were bound together. In contrast, peak 4, which contained mostly monomeric YscG-Flag, displayed only one fast migrating species and no high molecular weight complexes (Figure 6C, P4).

Conformational differences between the two needle-chaperone complexes were examined using limited proteolysis. Equal concentrations of the WT and mutant YscF-EG complexes were exposed to increasing amounts of the protease thermolysin. Protease protection was observed by SDS-PAGE followed by coomassie staining. Consistent with previous findings (Sun et al., 2008), YscG and YscE were resistant to mild protease treatment whereas YscF was extremely sensitive (Figure 6D). No difference in protease sensitivity was observed between the WT and mutant YscF-EG complexes, and both were completely susceptible to thermolysin treatment in the presence of the detergent TritonX-100 (Figure 6D, lanes 6 and 12). Taken together, the gel filtration, native-PAGE and protease protection experiments indicated that the YscF-EG and YscF-L54V-EG complexes were very similar in size, shape, exposed charged residues and protease accessibility in solution.

WT YscF-EG and YscF-L54V-EG have similar dissociation characteristics

The above analysis did not uncover any large structural differences between the purified WT and mutant YscF-EG complexes, but these techniques might not detect if YscF-L54V bound more tightly to the chaperones than WT YscF. Stronger interactions of YscF-L54V with the chaperones could lead to difficulties extracting YscF-L54V from the chaperones, resulting in titration of the chaperones, the components involved in separation of chaperones and substrates, or both from WT YscF. To address this, we examined the thermal denaturation of purified WT YscF-EG and YscF-L54V-EG using differential scanning fluorimetry (Niesen et al., 2007)(see experimental procedures for details). This technique takes advantage of the properties of the dye Sypro Orange, which fluoresces upon binding to hydrophobic regions of proteins. As proteins unfold in response to a gradual increase in temperature, more hydrophobic patches are exposed and fluorescence increases. Purified WT YscF-EG and YscF-L54V-EG were each exposed to a temperature gradient in the presence of Sypro Orange and unfolding was detected by an increase in fluorescence. The minimum and maximum intensities of each curve were used to calculate the melting temperature (Tm) using the Boltzmann equation. Experiments were carried out in triplicate and all data points for one representative experiment are shown (Figure 6E). The denaturation curves for the WT and mutant YscF-EG complexes were indistinguishable from each other (Figure 6E, blue and red curves respectively) and the average Tms for the triplicate samples was 57.9°C for both. While the Tms calculated from this assay varied from experiment to experiment and with different protein concentrations, the average melting temperatures of WT YscF-EG and YscF-L54V-EG at the same concentration within a single experiment were always within 1°C of each other. In contrast, when the P4 fraction containing only YscG was analyzed by this assay (Figure 6E, black line), no thermal transitions were apparent, and the curve suggested that most of the protein was unfolded and preceded to aggregate as the temperature increased. The similarity in thermal unfolding of the YscF-EG and YscF-L54V-EG complexes indicated that the YscF-L54V protein does not bind more tightly to the chaperones than WT YscF.

YscN physically interacts with both WT and mutant needle-chaperone complexes in Yersinia

The WT and mutant YscF-EG complexes appeared identical when isolated from the cytosol, and so we posited that YscF-L54V might disrupt one or more of the subsequent steps in the secretion pathway, such as targeting the YscF-EG complex to the base or release of YscF from the chaperones. T3SS-associated ATPases provide both targeting and stripping functions for effector-chaperone complexes (Lorenz & Buttner, 2009, Gauthier & Finlay, 2003, Akeda & Galan, 2005), so it was reasonable to propose that needle monomers also use the Yersinia ATPase YscN for dissociation from their chaperones. ATPase associations with chaperone-substrate complexes have been observed in vitro with purified components (Akeda & Galan, 2005, Lorenz & Buttner, 2009), and in vivo with very poor efficiency (Gauthier & Finlay, 2003), suggesting that these interactions are transient in the cytosol.

We hypothesized that if the YscF-L54V-EG complex interacted unproductively with YscN, this intermediate would be enriched and might be detected in vivo by co-immnoprecipitation. To test this, YscF-EG complexes were immuno-precipitated from WT and yscFL54V strains expressing Flag-YscG and YscE-His, and the association of YscN with the complexes was examined by Western blot. Co-immunoprecipitation of native YscN with WT YscF-EG was barely detectable (data not shown). We therefore constructed an inducible plasmid that expressed an N-terminal His-tagged YscN to increase the levels of YscN in the cell. N-terminal His-tagged ATPases do complement for function in vivo (Johnson & Blocker, 2008) and retain full ATPase activity (Blaylock et al., 2006). WT and yscFL54V strains expressing Flag-YscG, YscE-His and His-YscN were grown in low calcium media and the proteins were induced at the temperature shift. The WT and mutant YscF-EG complexes were immuno-precipitated as before and the presence of His-YscN was probed with His antibodies. His-YscN co-precipitated with the WT YscF-EG complex (Figure 7A, top panel, lane 5, arrowhead). The amount of His-YscN precipitated is very low, but we believe that the association is specific since a similar band is not present in IPs from a control strain lacking His-YscN (Figure 7A, top panel, lane 20), and His-YscN does not associate with Flag beads in the absence of the YscEFG complex (Figure 7A, top panel, lane 25). Furthermore, neither the abundant cytosolic protein S2, nor the effectors YopE or YopD co-precipitated with YscF-EG (data not shown). These results show for the first time that WT needle-chaperone complexes associate with YscN.

Figure 7.

A. YscN physically interacts with both the YscF-EG and YscF-L54V-EG complexes in Yersinia. WT, yscF L54V or ΔyscF strains expressing Flag-YscG, YscE-His and His-YscN, as well as control WT strains expressing either Flag-YscG and YscE-His or His-YscN were grown in low-calcium media at 37°C. Bacteria were lysed and insoluble protein was removed by centrifugation (P). Clarified supernatants (S) were passed over a Flag affinity column, washed (wash), and proteins eluted by competition with the Flag peptide (elute). Samples were separated by SDS-PAGE and analyzed by Western blotting with antibodies to Flag (YscG), His (YscE and YscN), and YscF. Arrowheads indicate location of YscN protein in the eluates. B. Excess YscN protein does not suppress the yscF L54V secretion defect. WT Yersinia containing the plasmids pTRC99 and pBAD33 (lane 1), pTRC99 and pBAD-yscF L54V (lane 2), pTRC99-his yscN and pBAD33 (lane 3), or pTRC99-his yscN and pBAD-yscF L54V (lane 4) were grown in low-calcium media, and expression from plasmids was induced at the 37°C shift. After 2 hrs at 37°C, cell pellets and culture supernatants were collected and proteins were separated by SDS-PAGE. Secreted Yops in culture supernatants were visualized by Coomassie staining (top panel). Proteins in cell pellets were visualized by western blotting with antibodies to YscN, the His tag, and YscF (middle three panels). Secreted YscF was detected in culture supernatants with YscF antibodies (bottom panel). C. YscF-EG and YscF-L54V-EG complexes stimulate equal levels of YscN ATPase activity. Phosphate release was measured to detect ATPase activity of purified His-YscN protein in the presence of ATP, and when mixed with purified YscF-EG or YscF-L54V-EG complexes and ATP. Shown is a representative experiment, performed in duplicate. Error bars indicate the standard error of the mean.

Interestingly, the association of YscN with the YscF-EG complex was not dependent on the presence of YscF. A ΔyscF strain expressing Flag-YscG, YscE-His and His-YscN still showed an interaction between YscEG complexes and His-YscN (Figure 7A, top panel, lane 15). These data indicate that YscN likely binds to the YscF-EG complex through a direct interaction with YscG and/or YscE.

WT YscF-EG and YscF-L54V-EG associated with YscN equally well (Figure 7A, top panel, lanes 5 and 10). Unfortunately, we could not determine if the YscF-L54V-EG complex out-competes the WT YscF-EG complex for YscN binding in the DN strain because we cannot discriminate between YscF and YscF-L54V biochemically when both proteins are expressed in the same cell. Even so, the physical association alone indicates that YscF-L54V, through its interaction with YscEG, can progress to a YscN-dependent step in the YscF secretion pathway, and that YscF-L54V blocks WT YscF secretion during or after its interaction with YscN. In addition, the fact that we did not observe more YscN associated with YscF-L54V-EG versus WT YscF-EG suggests that the YscF-L54V-EG is not sequestering YscN.

Over-expression of YscN does not relieve the YscF-L54V secretion defect

Though we did not see an enrichment of YscN associations with YscF-L54V-EG as compared to WT YscF-EG when these complexes were isolated from separate strains (Figure 7A), it was possible that the YscF-L54V-chaperone-ATPase complex formed was unproductive, and when this occurred in the DN strain, YscN would be titrated away from WT YscF-EG. We therefore asked if overproduction of YscN in the DN strain would relieve the YscF or Yop secretion defects. WT Yersinia that expressed pBAD-yscFL54V, pTRC99-his YscN, or both were grown in Yop-inducing conditions, and Yop secretion was assayed. As shown in Figure 1, no Yop secretion was detected when YscF-L54V was produced (Figure 7B, top panel, lane 2). Over-expression of His-YscN was clearly seen by western blot using YscN antibodies (Figure 7B, α-YscN, compare lane 1 to lanes 3 and 4), but the increase in YscN levels did not relieve the secretion block of either Yops or YscF protein (Figure 7B, top and bottom panels, lanes 3 and 4). We conclude that YscF-L54V-EG is not sequestering YscN from WT YscF-EG, since providing additional YscN did not allow YscF secretion.

Both YscF-EG and YscF-L54V-EG stimulate YscN ATPase activity in vitro

To determine if YscF-L54V-EG elicited the same level of ATPase activity from YscN as the WT YscF-EG, ATPase activity of His-YscN was measured in vitro in the presence of the needle-chaperone complexes. His-YscN protein was over-expressed in E. coli and purified to homogeneity (data not shown). The purified His-YscN protein retained ATPase activity as measured by phosphate release assay (Figure 7C, YscN + ATP). When purified YscF-EG or YscF-L54V-EG was mixed with His-YscN in the presence of ATP, a comparable increase in phosphate release was detected (Figure 7C, YscN + ATP + WT EFG and YscN + ATP + 54 EFG), suggesting that YscN recognized and acted upon both needle-chaperone complexes equally.

Together, the data presented in Figure 7 all point to similar interactions of YscN with both WT and mutant YscF-EG complexes. We conclude that interactions with, or unfolding by YscN is unlikely to be the step at which YscF-L54V blocks needle assembly, and we suspect that the block occurs after release of YscF-L54V by YscN.

The yscF-L54V Yop secretion defect is dominant over regulatory mutants

A number of proteins are required for cell contact-regulated secretion of Yops. In the absence of the proteins YopN (Forsberg et al., 1991) or TyeA (Iriarte et al., 1998), Yops are secreted at high levels and secretion is insensitive to calcium. In the absence of YopD (Williams & Straley, 1998, Francis et al., 2001) or LcrV (Pettersson et al., 1999), secretion is also insensitive to calcium but Yops are secreted poorly. We asked if the yscF-L54V mediated Yop secretion block occurred in these secretion regulation mutant backgrounds. WT, ΔyopBD, ΔlcrV, ΔyopN andΔyopNtyeA strains carrying plasmids expressing WT YscF or YscF-L54V were tested for Yop and YscF secretion in both the presence and absence of calcium (Figure 8 and data not shown). In all strain backgrounds and growth conditions, yscF-L54V was dominant over the regulation mutants and prevented both the secretion of Yops (Figure 8A) and YscF (Figure 8C, lower panel) despite having ample amounts of Yops and YscF present in cells (Figure 8C, upper panel and data not shown). These data strongly support a model in which YscF-L54V physically blocks the secretion apparatus during needle assembly, and that this block occurs prior to a step at which the regulatory protein complexes are active.

Figure 8.

yscF-L54V blocks Yop and YscF secretion in regulation-defective strain backgrounds. Strains deleted for yopB and yopD (ΔyopBD), lcrV (ΔlcrV), or yopN (ΔyopN) carrying plasmids pTRC99A, pTRC99-yscF or pTRC99-yscF-L54V were grown in low calcium media and shifted to 37°C. Cell-associated and secreted proteins were processed as in Figure 1A. A. Secreted proteins visualized by coomassie staining. Molecular mass standards are shown o the left in kDa. Yops are labeled on the right. B. Cellular YopD (left panel), LcrV (middle panel) and YopN (right panel) proteins in the corresponding deletion strains visualized by Western blotting. C. Cellular (top panel) and secreted (bottom panel) YscF protein visualized by Western blotting. D. Cytosolic control protein S2 visualized by Western blotting in the cellular (top panel) and secreted (bottom panel) protein fractions.

YscF-L54V does not block actively secreting needles

After needle assembly is complete, T3SS machines switch from secretion of early substrates like YscF to secretion of middle and late substrates such as the translocon proteins and effectors. The YscF-L54V protein clearly disrupts an initial step of WT needle assembly. However, since needle formation is required for the secretion of all other substrates, we asked if YscF-L54V blocked secretion of all substrates, or just the secretion of YscF. To test this, we designed an experimental condition in which YscF-L54V was induced after WT needles were formed and we asked if the delayed production of YscF-L54V prevented secretion of Yops. Furthermore, we asked if expression of yscF-L54V differentially affected bacteria that possessed WT needles that were not yet secreting Yops and WT strains that contained actively secreting needles. Production of YscF-L54V in the WT+pBAD-yscF-L54V strain was repressed with dextrose while permitting the formation of non-secreting WT needles by growth in high calcium media or secreting WT needles by growth in low calcium media (see Figure 9A experiment schematic and figure legend for details). After needles were made, YscF-L54V was induced with arabinose and all cultures were moved into low calcium media to trigger secretion. After 30 minutes of yscF-L54V expression, all cells were washed to remove any Yops that were secreted during the initial yscF-L54V induction phase. Fresh media was then added and cultures were incubated for an additional 60 minutes. Culture supernatants were then collected to detect Yop secretion. Additionally, to confirm the presence of needles and to track their stability throughout the various steps of the experiment, samples were taken after each manipulation and extracellular YscF was visualized by immunofluorescence (Figure 9A). The ΔyscF strain containing pBAD-YscF was included to demonstrate that repression (Figure 9B, lanes a and g) and induction (Figure 9B, lanes b and h) from the pBAD plasmid was occurring.

Figure 9.

YscF-L54V does not block Yop secretion from preformed WT secreting needles. The ΔyscF+pBAD-yscF, WT+pBAD and WT+pBAD-yscF-L54V strains were grown at 26°C in high or low calcium media in the presence of dextrose to repress expression from the pBAD promoter. Cultures were shifted to 37°C for 30 minutes to allow for production of either non-secreting needles or secreting needles in the WT backgrounds. Samples were then split in half, cells were washed to remove the dextrose and moved into low calcium media pre-warmed to 37°C. Expression from the pBAD plasmids either remained repressed with dextrose, or was induced by addition of arabinose and incubated at 37°C for 30 minutes (panels 3–6). Media was then removed from all cultures and was replaced with pre-warmed fresh media of the same composition and growth continued at 37°C for an additional 60 minutes (panels 7–14). Samples were taken at various steps and processed for immunofluorescence and secreted Yops. A. Immunofluorescence micrographs of surface-localized YscF on the WT+pBAD-yscF-L54V strain (panels 1–10) and the WT+pBAD strain (panels 11–14). Experimental schematic is shown on the left. At the indicated steps, samples were removed and bacteria were fixed and stained with DAPI and YscF antibodies as in Figure 2C. B. Secreted Yops were collected from culture supernatants of all three strains that were induced with arabinose or not (+/− pBAD induction) after the final incubation step in A. Proteins were separated by SDS-PAGE and stained with coomassie. Numbers under gel lanes refer to the corresponding samples shown in A, panels 7–14.

WT bacteria containing the empty pBAD plasmid produced needles and secreted Yops in the presence of both arabinose and dextrose (Figure 9A, panels 11–14 and Figure 9B lanes c, d, i and j). As expected, WT bacteria produced needles and secreted Yops when yscF-L54V was fully repressed with dextrose (Figure 9A panels 1, 2, 3, 5, 7 and 9, and 9B lanes e and k). When YscF-L54V was expressed in Yersinia containing pre-formed non-secreting needles and then shifted to secretion-permissive media, no Yop secretion was observed 30 minutes post-induction (data not shown), and Yop secretion was almost undetectable 90 minutes post-induction (Figure 9B, lane f). Immunofluorescence images revealed that needles were present 30 minutes after induction of YscF-L54V (Figure 9A, panel 4), but by 90 minutes after induction, needles were completely absent (Figure 9A, panel 8). These data indicated that YscF-L54V was able to block secretion from pre-formed, non-secreting WT needles, and that the secretion block resulted in the eventual turnover of existing WT needles. In contrast, yscF-L54V expression after establishment of fully functional, actively secreting needles did not block Yop secretion (Figure 9B, lane l) and needles remained intact throughout the experiment (Figure 9A, panels 6 and 10). There were less Yops secreted from the WT+pBAD-yscF-L54V strain when yscF-L54V was induced than when yscF-L54V was repressed (Figure 9B, compare lanes k and l), however the same phenomenon was apparent in the WT strain carrying the empty pBAD vector (Figure 9B, compare lanes i and j), suggesting that the decrease in Yop secretion was an effect of growth in arabinose and not due to the presence of YscF-L54V. Indeed, the amount of secreted Yops in arabinose-induced WT+pBAD-yscF-L54V and WT+pBAD was practically identical (Figure 9B, lanes j and l) demonstrating that YscF-L54V did not interfere with the T3SS once Yop secretion had begun.

Although needles were observed on the surface of Yersinia when YscF-L54V was expressed during active Yop secretion, there did appear to be fewer needles than in the same strain that was not expressing the DN needle protein (Figure 9A, compare panels 5 and 9 with panels 6 and 10), and considerably less needles than in a WT+pBAD strain at the same time point (Figure 9A, panel 14). We have shown in Figure 2C that the generation of new needles is blocked by YscF-L54V, and so we suspect that the difference in needle levels in the presence or absence of YscF-L54V in panels 5, 6, 9 and 10 reflects the turnover and/or dilution rate of the existing secreting needles during the normal growth of the bacteria, similar to the turnover of WT needles observed in panel 8. Interestingly, the difference in needle number for the WT+pBAD and WT+pBAD yscF-L54V strains did not translate into a noticeable difference in Yop secretion levels at the endpoint of the assay, highlighting the point that YscF-L54V does not obstruct Yop secretion. Together, these data demonstrate that there is a difference between assembled but non-secreting needles and needles that are actively secreting Yops, and that the YscF-L54V protein interferes with the former but not the latter.

Discussion

Assembly of the T3SS needle is a carefully orchestrated multi-step process that requires the coordination of many proteins located in the cytosol, in both membranes and on the bacterial cell surface. To identify intermediate steps in needle assembly, we devised a genetic screen to isolate needle protein mutants that were DN for Yop secretion. The yscF-L54V mutant interferes with WT YscF secretion, which is a prerequisite for both needle formation and Yop secretion. Secretion of YscF requires several events: 1) chaperone binding in the cytosol, 2) targeting of the YscF-EG complex to a docking site at the base and 3) release of YscF from the chaperones to allow its passage into the base. This work demonstrates that YscE and YscG form a soluble complex with YscF in the Yersinia cytosol, and that YscF-EG complexes physically interact with the base-associated T3SS component YscN. Furthermore, we have shown that YscF-L54V is capable of both interactions with its chaperones and with YscN. We propose two models described below in which YscF-L54V disrupts a step of needle assembly either during or after its interaction with YscN, and that this step is specific to the secretion of YscF and is not involved in the secretion of Yops.

How does the YscF-L54V protein block the secretion of WT YscF?

Only the simultaneous co-expression of YscE, F and G completely restored YscF secretion in the presence of YscF-L54V. We envision two possible scenarios in which excess WT YscF-EG may act to bypass the DN state. In the first working model, the YscF-L54V-EG-YscN interaction is productive and YscN is able to remove YscF from YscEG, such that the YscF-L54V secretion block occurs at a step after stripping from YscN but before YscF secretion. In this case, the YscF-L54V protein would not function as a DN within the context of the YscF-EG complex, but rather could disrupt WT YscF secretion through aberrant interactions with components within the base, or perhaps by physically jamming the base. The fact that we could not detect secretion of any substrate in the presence of YscF-L54V, even in genetic backgrounds where secretion is constitutive, lends support to this hypothesis. WT YscF-EG could then overcome this block by flooding the system with excess secretion-competent WT YscF that would be available to either compete with YscF-L54V for a post-YscN binding site, or to initiate the production of new needle complexes. Once a WT needle complex is formed and the substrate switch is flipped to secrete Yops, YscF-L54V would no longer affect that apparatus.

Another model for the mechanism of secretion block by YscF-L54V is equally plausible. The L54V mutation may change how the YscF-EG complex interacts with YscN. If the YscF-L54V-EG-YscN interaction is unproductive, such that YscF-L54V-EG binding to YscN disrupts trafficking of YscN to the base, or YscN is unable to strip YscF-L54V from the chaperones, then L54V-YscF-EG would remain stuck at the ATPase, blocking access to WT YscF and all other early substrates. A similar situation has been described for the flagellin-chaperone complex, in which a mutant chaperone is DN for the secretion of flagellin by binding too tightly to the flagellar ATPase FliI (Thomas et al., 2004). Excess WT YscF-EG complexes would then relieve this block by out-competing YscF-L54V-EG for binding to YscN. We have shown that YscF-L54V-EG binds to and activates YscN as well as the WT YscF-EG complex. However, our attempts to detect YscN-mediated release of either WT YscF or YscF-L54V from the chaperones were unsuccessful (data not shown), suggesting that the increase of ATPase activity detected upon addition of needle-chaperone complexes to YscN was not enough to facilitate extraction of YscF from the chaperones. This data hints that perhaps an additional protein partner is required for optimal YscN activity during YscF release. Possible candidates include the YscN-interacting proteins YscU, YscL, YscK, YscQ, YscO (Riordan & Schneewind, 2008) and YopN (Botteaux et al., 2009). Thus, YscF-L54V-EG could disrupt interactions of YscN with its partner protein, resulting in stalled YscF-L54V-EG-YscN complexes. The fact that excess YscN does not relieve the YscF-L54V secretion defect lends support to the idea that one or more co-factors are required for YscF release from chaperone-ATPase complex. In the case where WT needle complexes are actively secreting Yops, the YscN cofactor required for YscF release may not be required for ATPase activity on effector-chaperone complexes, allowing effector secretion to proceed normally.

Detection of T3SS ATPases associated with substrates in bacteria has been challenging, possibly due to the transitive nature of the ATPase-substrate interaction and because so many different substrates are likely to compete for ATPase activity during effector secretion. Here we have demonstrated that the Yersinia T3SS ATPase YscN associates with needle-chaperone complexes during needle assembly, strongly implicating a role for YscN in docking the needle-chaperone complex at the secretion base and/or extraction of the needle subunit from the chaperones. This finding expands the repertoire of YscN binding proteins to include secreted machinery components. The YscF-EG-YscN interactions were detected under secretion-inducing conditions, when both new needles were being made and older needles were secreting Yops (Sorg et al., 2006), begging the question of how an apparatus in the process of needle assembly selects early substrates in the midst of the available late substrates.

The fact that YscF-L54V blocks secretion from completed needle complexes that are not secreting Yops, but not after the needle complexes have switched to secretion of Yops indicates that there is something fundamentally different between the secretion machines at these two stages. We propose that the signal to initiate Yop secretion results in changes in the base structure that eliminate a binding site for needle-chaperone complexes, and perhaps other early substrates, and reveal a binding site for later secretion substrates. Marlovits et. al have shown that the cytoplasmic face of the Salmonella base structure exhibits different conformations before and after WT needles are assembled (Marlovits et al., 2004). Indeed, a YscU self-cleavage event (Bjornfot et al., 2009, Riordan & Schneewind, 2008, Sorg et al., 2007) is important for substrate switching. These physical changes in YscU and other integral membrane components of the base could play a part in discrimination between early and late substrates by changing critical access points. Additional contributing factors include feedback from various early-secreted components themselves (YscF, YscP, and YscI; (Torruellas et al., 2005, Davis & Mecsas, 2007, Agrain et al., 2005, Wood et al., 2008)), and association of different soluble T3SS components with the base (Riordan & Schneewind, 2008). This model is supported by experiments in which impassable early substrates block secretion from new needles, but they are rejected after completion of the needle when Yops are being secreted (Sorg et al., 2006, Riordan et al., 2008).

The process of needle polymerization is particularly difficult to study in vivo because it is the last step in a sequential order of events that are all required to form a functional needle. Thus many mutations in the needle protein will disrupt one or more events prior to polymerization, leading to a terminal phenotype of no needles. Here we have devised an in vitro polymerization assay that monitors polymerization of purified needle monomers over time and under different experimental parameters. This assay, along with that described for the PscF protein (Quinaud et al., 2005), represents a valuable tool to dissect regions of the needle protein that are important for the polymerization process in isolation, without in vivo complications in which a particular mutation might also affect previous essential needle biogenesis steps.

The YscF protein polymerized freely in solution in the absence of all other T3SS components. Although in vitro-grown YscF polymers were very stable once formed, we did observe some breakdown of pre-formed polymers after exposure to buffers at pH 5 (data not shown). This suggests that when in the host, the needle structure may be susceptible to assault by lysosomal contents, which have been measured at ~ pH 4.5-pH 5.0 (Mellman et al., 1986), that are released during eukaryotic plasma membrane damage or upon phagosome-lysosome fusion. The diameter of the YscF polymers formed in vitro was the same as needles observed on the surface of bacteria (Figures 2C, 3C and data not shown), suggesting that tube width is inherent in needle polymerization and is not determined or constrained by components of the T3SS base. However, unlike needle width, we found that purified YscF formed polymers that were considerably longer than what is seen on the surface of WT Yersinia (Hoiczyk & Blobel, 2001), in agreement with data showing that needle length is regulated in vivo by proteins in the YscP family (Journet et al., 2003, Magdalena et al., 2002, Tamano et al., 2002, Hirano et al., 1994, Kubori et al., 2000, Morello & Collmer, 2009). It is important to note that we occasionally observed long needles on the surface of WT Yersinia that over-expressed the needle chaperones YscE and YscG (data not shown). This raises the interesting possibility that needle length is also regulated in part by the abundance of YscF-EG complexes, or by the amount of secreted YscF protein. Long needles are also observed in Shigella and Salmonella that over-express their needle proteins (Cordes et al., 2003, Kubori et al., 2000). On the other hand, we cannot rule out that over-expression of the needle chaperones affects YscP function, perhaps out-competing YscP for the early substrate binding site at the T3SS base, which would decrease the amount of secreted YscP, leading to long needles (Agrain et al., 2005).

Numerous functions have been attributed to effector chaperones, including retention of substrates in a secretion-permissive state (Feldman et al., 2002, Stebbins & Galan, 2001), stabilizing substrates to prevent degradation (Abe et al., 1999, Fu & Galan, 1998, Frithz-Lindsten et al., 1995, Menard et al., 1994), shielding hydrophobic patches to prevent aggregation (Letzelter et al., 2006), and preventing premature association of proteins that interact outside the cell (Neyt & Cornelis, 1999, Menard et al., 1994). Needle chaperones prevent polymerization (Quinaud et al., 2005), and also appear to protect needle subunits from degradation, as needle monomers are not stable and/or are turned over in the absence of their chaperones (Quinaud et al., 2005, Tan et al., 2008, Betts et al., 2008). In agreement with these observations, we could never detect the YscF protein when expressed in the absence of its chaperones, either in E. coli or in Yersinia lacking the virulence plasmid (data not shown). Thus it was not surprising that our screen failed to identify YscF proteins with chaperone binding defects, since this phenotype would lead to free YscF that would be targeted for removal, negating any potential to have a DN effect. In contrast to expression in bacteria, the purified YscF protein was stable as a dilute solution of monomers (data not shown), and it readily refolds after unfolding (Quinaud et al., 2005), suggesting that the monomer removal from chaperones is likely to be an active process, and that YscF mutants defective in this process could be DN.

We found that the needle chaperones could interact with the ATPase in the absence of the needle protein. This strongly indicates that the needle chaperones, in addition to their role in preventing premature needle polymerization, are also responsible for targeting the needle to the secretion apparatus. A secretion signal has been identified within the first 15 residues of YscF, which is required for its secretion out of the cell (Pastor et al., 2005). Since it appears that this signal may not be necessary for targeting to the ATPase, it will be interesting to determine the step in the secretion pathway in which the secretion signal is recognized, and what factor(s) requires its use.

Because YscF-L54V was not secreted in Yersinia, we cannot address if this mutant would assemble correctly into a WT T3SS base or if it would form a needle functional for secretion. Mutagenesis of the Shigella needle protein MxiH at position Y50 (analogous to L54 of YscF) did not affect the secretion of MxiH or the assembly of the needle on the bacterial surface (Zhang et al., 2007). Although chaperone proteins have not yet been identified for the MxiH protein, clearly some substitutions in the Y50 residue are tolerated during MxiH-chaperone interactions. Curiously, the Y50 mutant needles did not secrete effectors, suggesting that this position is critical for MxiH needle function. Identifying extragenic suppressors that permit secretion of the YscF-L54V mutant in vivo would enable us to address if a YscF-L54V needle was functional for Yop secretion.

Experimental Procedures

Bacterial culture, strains and plasmids

Yersinia were grown in 2xYT medium, in low-calcium secretion medium (2xYT supplemented with 20 mM sodium oxalate and 20 mM MgCl₂) or in high-calcium medium (2xYT supplemented with 3 mM CaCl₂). E. coli were grown in LB. Carbenicillin (100 μg/ml) or chloramphenicol (10 μg/ml) were used for plasmid maintenance. All Yptb and E. coli strains and plasmids are listed in Table 1. Detailed descriptions of strain and plasmid construction are provided in the Supplementary Experimental Procedures. All oligonucleotides used for cloning are listed in Supplementary Table 1.

Table 1.

Strains and plasmids used in this study.

Strain or Plasmid	Description	Source or Reference
E. coli strains
SM10 λPir	Conjugation competent strain	Miller and Mekalanos (1988)
SY327 λPir	Routine cloning	Miller and Mekalanos (1988)
DH5α	Routine cloning	Cambau et al. (1993)
BL21-DE	protein expression in E. coli	Studier and Moffatt (1986)
Yersinia pseudotuberculosis strains
IP2666 pYV (pIBI)	WT Yptb	Ivanov et al. (2005)
IP2666 pYV-	WT Yptb cured of the virulence plasmid	J.M Balada, et al. (2006)
ΔyscF	Deletion of yscF (codons 2–86) in IP2666	Davis and Mecsas (2007)
ΔyopBD	Truncation of yopB and deletion of yopD in YPIII	L. Logston, unpublished
ΔlcrV	Deletion of lcrV (codons 19–326) in IP2666	This study
ΔyopN	Deletion of yopN (codons 2–287) in IP2666	Davis and Mecsas (2007)
Plasmids
pGEM-T easy	Cloning of PCR products; Apr	Promega
pDS132	Suicide vector; Cmr	Philippe et al. (2004)
pTRC99A	IPTG-inducible expression vector; Apr	Amann et al. (1988)
pBAD33	Arabinose-inducible expression vector; Cmr	Guzman, et al. (1995)
pGEX-5X-1	GST fusion expression vector; Apr	Amersham
pTRC99-yscF	Expresses yscF from tac promoter	Davis and Mecsas (2007)
pTRC99-yscF L54V	Expresses yscF L54V mutant from tac promoter	This study
pBAD-yscF	Expresses yscF from araBAD promoter	This study
pBAD-yscF L54V	Expresses yscF L54V from araBAD promoter	This study
pGEX-yscF	GST-YscF fusion expression vector	Davis and Mecsas (2007)
pGEX-yscF L54V	GST-YscF L54V fusion expression vector	Davis and Mecsas (2007)
pFlag-YscG-YscE-His	Expresses YscG and YscE from tac promoter	Greg Plano, Unpublished
pTRC99-his yscN	Expresses His-tagged yscN from tac promoter	This study
pBAD33-his yscN	Expresses his yscN from araBAD promoter	This study
pGEX62-lcrV	GST-LcrV fusion expression vector	H. Dams-Kozlowska, unpublished
pGEX-yopN	GST-YopN fusion expression vector	This study
pTRC99-yopN	Expresses yopN from tac promoter	This study
pBAD-lcrV	Expresses lcrV from araBAD promoter	D. Harmon, unpublished
pDS132-yscF L54V	Suicide vector for introducing L54V mutation	This study

yscF Dominant Negative Screen

Congo Red assay

The pTRC99A-yscF* libraries were electroporated into the WT IP2666 strain and plated onto LB/Amp agar. Single transformants were replica patched onto both LB/Amp plates (for maintenance) and LB agar plates containing 20 mM Na Oxalate, 20 mM MgCl₂ and 0.005% Congo Red (for selection). Congo Red plates were incubated at 37 C overnight and scored for red or white colonies.

In vitro secretion

Yersinia were diluted from overnight cultures into secretion media, grown for 1.5 hrs at 26 C and then shifted to 37 C for 1.5 hours. Protein production in Yersinia containing pTRC99A plasmids, the pFlagYscG-YscEHis plasmid, or pBAD plasmids was induced at the 37 C shift by the addition of 10 μM IPTG, 100 μM IPTG or 50 mM arabinose, respectively. Whole cell lysates and secreted proteins were processed as described (Davis & Mecsas, 2007).

Chemical Cross-linking

Yersinia were grown either in calcium-depleted media or in media supplemented with 3 mM calcium and subjected to cross-linking with 1mM BS³ as described previously (Davis & Mecsas, 2007).

Antibodies and Western Blots

Polyclonal rabbit anti-YscF, anti-YopE, anti-YopD and anti-S2 antibodies have been described (Davis & Mecsas, 2007) and were used at a 1:10,000 dilution. Mouse monoclonal Flag antibody (Sigma) was used at 1:2500 dilution. Mouse monoclonal Penta-His antibody (Qiagen) was used at 1:1000 in TBS containing 3% BSA. Mouse monoclonal antibody to YscN was a gift from Olaf Schneewind, and was used at 1:2000. Goat anti-rabbit HRP (Bio-Rad) and goat anti-mouse HRP (Sigma) secondary antisera were used at 1:10,000. All antibodies except for the Penta-His were diluted into in PBS with 5% milk and washes were performed in PBS/0.1% Tween-20. Chemiluminescence (Perkin Elmer, Western Lightning Kit) was used per the manufacturer’s instructions.

To generate polyclonal rabbit anti-LcrV and rabbit anti-YopN antibodies: Recombinant LcrV protein was produced by overexpression of a GST-LcrV fusion protein from pGEX62-LcrV in E. coli BL21 cells. GST-LcrV was purified from lysates using glutathione-sepharose and LcrV was released by on column cleavage with prescission protease (GE Healthcare). Recombinant GST-YopN was produced by overexpression from pGEX-YopN in E. coli BL21 cells. GST-YopN was located in inclusion bodies and was purified using a standard inclusion body purification protocol. Puri ed LcrV and GST-YopN proteins were sent to Covance Research Products (Denver, PA) for antiserum preparation in rabbits. Antibodies were used at a 1:10,000 dilution.

Indirect Immunofluorescence Microscopy

Yersinia strains were grown in high-calcium medium for 2 hrs at 26°C, induced with 10 μM IPTG, and shifted to 37°C for 2 hrs. Bacteria were fixed by the addition of a mix of 800 μl 4% paraformaldehyde, 1 μl 25% glutaraldehyde and 40 μl 0.5 M NaPO₄ pH 7.4 to 500 μl of the bacterial culture for 15 min at RT followed by 30 min or longer on ice. Fixed bacteria were gently sedimented (5000 × g for 5 min), washed 4 times with PBS and stored in 250 mM glucose, 10 mM Tris-HCl pH 7.5, 1 mM EDTA for up to 3 days at 4 C. Bacteria were adhered to poly-L-lysine (Sigma) coated coverslips for 20 min, washed with PBS containing 0.1% Tween-20 (PBST) and blocked in PBST + 2% BSA (PBST/BSA) for 1 hr at RT. Bacteria were incubated with anti-YscF antibody at 1:5000 in PBST/BSA for 1 hr at RT, washed for 1 min with PBST 4 times and incubated with Alexa fluor 594 anti rabbit secondary antibody (Invitrogen) at 1:5000 in PBST/BSA for 30 min at RT in the dark. Secondary antibody was washed in PBST 5 times for 1 minute. 1.4 μM 4′, 6-diamidino-2-phenylindole dihydrochloride (DAPI; Sigma) was included in the 4^th wash to stain nucleoids. Coverslips were mounted into slides with Slowfade (Invitrogen) and sealed with nail polish. Micrographs were taken with a Nikon inverted TE2000-U microscope with a Photometrics CCD camera at 60X magnification using MetaVue software (Molecular Devices, Sunnyvale, CA). DAPI and Alexa 594 were visualized using Nikon UV-2E/C and G-2E/C filters respectively. Images were pseudocolored and merged in MetaVue.

YscF protein purification

E. coli BL21-DE3 strains carrying pGEX-yscF or pGEX-yscF-L54V plasmids were induced and strains lysed as previously described (Davis & Mecsas, 2007). GST-YscF or GST-YscF-L54V fusion proteins were purified from the clarified lysates using glutathione sepharose following the manufacturers instructions. The fusion protein was eluted from the column with 10 mM glutathione, pH 8.0 and dialyzed into PBS/5mM EDTA/0.1% Triton X-100 using a Slide-A-Lyzer 10K MWCO Dialysis Cassette (Pierce). YscF was released by digestion with Factor Xa protease (10 units/mg protein; Sigma) overnight at RT. Since free YscF protein non-specifically stuck to the glutathione beads, the free GST was removed by ion exchange chromatography as follows. The YscF/GST solution was concentrated using a to ~ 2 mg/ml and then diluted ten-fold into 10 mM Tris-HCl pH 8.0. Q sepharose fast flow (Amersham) was equilibrated in 10 mM Tris pH 8.0, added to the YscF/GST solution and mixed gently on a tube rotator for 15 min at RT. The mixture was centrifuged for 1 min at 10k × g and YscF was captured in the supernatant.

in vitro YscF polymerization

Pure YscF or YscF-L54V (~0.2 mg/ml) was allowed to sit at RT overnight or was kept at 4°C. A 5 μl drop of each protein solution was mounted onto a carbon-coated nickel EM grid (Electron Microscopy Sciences, Hatfield, PA), washed with 20 drops of TBS and stained with 4 drops of 2% uranyl acetate. Samples were visualized on a Phillips CM10 transmission electron microscope at 80 kV acceleration and a starting magnification of 46,000X. Micrographs were taken with a Gatan Multiscan 600W CCD camera (Gatan, Inc. Pleasanton, CA).

Co-Immunoprecipitations

Yersinia strains were grown under secretion inducing conditions and 1 OD unit of cells were collected. Cell pellets were resuspended in 1 ml ice cold TNE (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 5 mM EDTA) with protease inhibitors (1 mM PMSF, 1 μg/ml aprotinin, 1 μg/ml pepstatin, 5 μg/ml leupeptin) and lysed by sonication for 3 × 30 seconds. TritonX-100 was added to 1% and lysates were mixed gently for 5 min at 4°C. Lysates were clarified by centrifugation at 12,000 × g at 4°C for 15 minutes. Flag- agarose beads (50 μl of a 1:1 slurry; Sigma) were added to the clarified supernatants, and mixed on a tube rotator at 4°C for 2 hours. Beads were collected by pelleting for 30 sec. at 10,000 × g and the supernatant was saved (flow through). Beads were washed with 1 ml ice cold TNE + 1% TritonX-100 4 times. Precipitated and co-precipitated proteins were eluted by addition of 100 μl SDS-sample buffer and boiling at 95°C for 5 min, or by competition with Flag peptide (100μg/ml; Sigma). After sedimentation, the supernatant was collected (elute). Co-immunoprecipitations in Figure 7 proceeded as above except that 0.1% TritonX-100 was added to the lysates to gently strip YscN from the inner membrane into the soluble fraction.

Gel filtration

YscF-EG complexes were purified from 150 ml cultures of Yptb pYV- strains expressing Flag-YscG, YscE-His and either YscF or yscF-L54V by affinity chromatography using Flag-conjugated agarose beads as per the manufacturers instructions. All steps were performed in 10 mM Hepes-NaOH pH 7.5/150 mM NaCl (HN buffer). Flag-YscG containing complexes were eluted with 5 column volumes of 100μg/ml Flag peptide in HN buffer and concentrated to ~ 0.5 ml using an Amicon Ultra 10 kDa cutoff filter. A HiPrep 16/60 Sephacryl S-100 HR column (GE Healthcare) was equilibrated in HN buffer then loaded with 0.5 ml of the concentrated YscF-EG Flag eluate. Fractions (1.5 ml) were collected and peak fractions were pooled and concentrated to ~ 0.5 ml using an Amicon Ultra 5 kDa cutoff filter. Protein concentrations were determined for each sample using the Bio-Rad protein assay (Bio-Rad).

Native gel electrophoresis

Protein samples were diluted into 2X native gel sample buffer (125 mM Tris-Cl pH 6.8, 20 % glycerol, 1% bromophenol blue) and were loaded directly onto a 12.5% glycine nondenaturing polyacrylamide gel without boiling. Myoglobin (17.8 kDa; Serva) was used as a molecular weight reference. Following separation, proteins were transferred onto Immobilon (Millipore) and processed for Western blot analysis as described above.

Protease protection assay

Limited proteolysis was performed with purified YscF-EG or YscF-L54V-EG and increasing concentrations of thermolysin in thermolysin reaction buffer (TRB; 10 mM Tris–HCl, pH 8.0, 2 mM CaCl2, 150 mM NaCl, and 5% glycerol). Three μg of protein in 15 μl HN was mixed with 15 μl 2X TRB. A thermolysin stock (1 mg/ml) was diluted 1:5, 1:25 and 1:125 in TRB, and 1 μl of each was added to a tube containing the protein solution. One sample also received 3 μl of 10% TritonX-100 (1% final conc.). Digestions proceeded for 1 hr at 37°C and were stopped by the addition of 30 μl 2X SDS-PAGE sample buffer and boiling at 95°C for 5 min.

Thermal denaturation

Differential scanning fluorimetry was performed essentially as described (Niesen et al., 2007). Purified YscF-EG, YscF-L54V-EG complexes and the P4 fraction from the WT YscF-EG purification were diluted into HN buffer containing 5X Sypro orange (Invitrogen, supplied as 5,000X) to a final concentration of 2 μM and moved into optical-grade PCR plates. Samples were exposed to a temperature gradient of 1°C per min using a Stratagene Mx3005p real-time PCR machine and fluorescence was monitored using the FAM and ROX filters for excitation and emission respectively. Data was analyzed using DSF Analysis software (Niesen et al., 2007) and Tm values were calculated using GraphPad Prism software.

His-YscN purification

His-YscF was over-expressed in E. coli BL21 cells from the pTRC99-His yscN plasmid and purified using a modified protocol of Blaylock et al. (Blaylock et al., 2006). Cells were grown to an OD600 of 0.5 and induced with 100 μM IPTG for 4 hours at 37°C. Cells were lysed using B-PER (Pierce) according to the manufacturer’s directions and centrifuged at 20k × g for 20 min. The insoluble pellet containing His-YscF protein was resuspended in 8 M urea, 50 mM Tris pH 8.0 plus protease inhibitors and left on ice for 1 hr. The urea lysate was clarified by centrifugation as above, and the supernatant was loaded onto a 1 ml Ni-NTA (Qiagen) column. The column was washed with 20 CV of 8 M urea, 50 mM Tris pH 8.0, 10 mM imidazole and His-YscN was eluted with 5 CV of 8 M urea, 50 mM Tris pH 8.0, 200 mM imidazole. His-YscN was re-natured by a 20-fold dilution into ice-cold, vigorously stirring renaturing buffer (50 mM Tris pH 8.0, 150 mM NaCl, 2 mM CaCl, 20% glycerol). Protein precipitates were removed by centrifugation and the supernatant was loaded onto a 1 ml Ni-NTA column equilibrated with renaturing buffer. The column was washed with 20 CV of renaturing buffer and refolded His-YscN was eluted with 5 CV of renaturing buffer containing 250 mM imidazole. Protein was concentrated to ~ 0.2 mg/ml using an amicon ultra filter with a 10kD cutoff.

Phosphate release assay

Phosphate release was detected using the QuantiChrome ATPase/GTPase assay kit (BioAssay Systems) as per the manufacturer’s directions. Purified WT EFG or L54V EFG complexes (1 μM) were mixed with purified His-YscN protein (0.5 μM) in the presence of 4 mM ATP and the reaction buffer provided in the kit, and incubated at RT for 30 min. Detection reagent was then added to the reaction and incubated for 30 min at RT. Phosphate was measured in a plate reader at 620 nm. The amount of phosphate released was calculated using a free phosphate standard curve.