Secretion of Yop effector proteins by the Yersinia pestis plasmid pCD1-encoded type III secretion system (T3SS) is regulated in response to specific environmental signals. Yop secretion is activated by contact with a eukaryotic cell or by growth at 37°C in the absence of calcium. The secreted YopN protein, the SycN/YscB chaperone and TyeA form a cytosolic YopN/SycN/YscB/TyeA complex that is required to prevent Yop secretion in the presence of calcium and prior to contact with a eukaryotic cell. The mechanism by which these proteins prevent secretion and the subcellular location where the block in secretion occurs are not known. To further investigate both the mechanism and location of the YopN-dependent block, we isolated and characterized several YopN mutants that constitutively block Yop secretion. All the identified amino-acid substitutions that resulted in a constitutive block in Yop secretion mapped to a central domain of YopN that is not directly involved in the interaction with the SycN/YscB chaperone or TyeA. The YopN mutants required an intact TyeA-binding domain and TyeA to block secretion, but did not require an N-terminal secretion signal, an intact chaperone-binding domain or the SycN/YscB chaperone. These results suggest that a C-terminal domain of YopN complexed with TyeA blocks Yop secretion from a cytosolic, not an extracellular, location. A hypothetical model for how the YopN/SycN/YscB/TyeA complex regulates Yop secretion is presented.
Yersinia pestis, the etiologic agent of plague, overwhelms its mammalian host with systemic growth by evading phagocytosis and by inhibiting the inflammatory response (Cornelis et al., 1998). These properties are associated with a 70 kb plasmid, termed pCD1 (Perry et al., 1998), which codes for a virulence-associated type III secretion system (T3SS). The T3SS is comprised of a group of structural proteins that make up the T3S apparatus or injectisome and of a set of secreted effector proteins called Yersinia outer proteins (Yops) (Cornelis, 2000). Injectisomes are specialized protein-secretion nanomachines that span the bacterial envelope and are topped by a hollow needle structure that protrudes 40–60 nm from the bacterial surface (Journet et al., 2003). The T3S apparatus is dedicated to transporting Yops from the bacterial cytosol to the cytoplasm of phagocytic cells (Rosqvist et al., 1994; Cornelis and Wolf-Watz, 1997). This process occurs in two distinct steps: (i) secretion of Yops across the bacterial membranes via the T3S apparatus, and (ii) translocation of Yops across the eukaryotic membrane. This second phase of the delivery process is dependent on three pore-forming bacterial proteins (LcrV, YopB and YopD) that form a translocation complex, or translocon, in the eukaryotic membrane (Nilles et al., 1998; Neyt and Cornelis, 1999; Pettersson et al., 1999). Once translocated into a eukaryotic cell, effector Yops modulate the cellular functions of professional phagocytes by disrupting the signal-transduction pathways that lead to bacterial uptake and to the production of proinflammatory cytokines (Palmer et al., 1998; Schesser et al., 1998; Orth et al., 1999).
Secretion of T3SS substrates is not constitutive; instead, this process is tightly regulated in response to specific environmental signals. In vivo, the T3S pathway is blocked prior to contact with a eukaryotic cell. Upon contact with the surface of a host cell, effector Yops are transported into the cytoplasm of the targeted cell with minimal loss to the extracellular milieu (Rosqvist et al., 1994; Cornelis and Wolf-Watz, 1997). In vitro, secretion of Yops is blocked at 37°C in the presence of 2.5 mM calcium and is triggered by growth at 37°C in the absence of calcium (Brubaker and Surgalla, 1964; Michiels et al., 1990). Secretion of Yops is also accompanied by a cessation of bacterial growth, an event termed growth restriction (Brubaker, 1983). Because of this behaviour, yersiniae's growth at 37°C has traditionally been called calcium-dependent (CD). Two classes of mutants have been identified on the basis of their growth patterns: calcium-independent (CI) mutants, which never enter growth restriction nor secrete Yops; and calcium-blind (CB) mutants, which always enter growth restriction and secrete Yops regardless of the calcium concentration (Yother and Goguen, 1985). In this report, we will refer to the CD, CI and CB phenotypes as regulated-secretion (RS), no-secretion (NS) and constitutive-secretion (CS) phenotypes respectively.
Regulation of Yop secretion is dependent on the expression of the secreted YopN protein (Yother and Goguen, 1985; Forsberg et al., 1991) and of the cytosolic SycN (Day and Plano, 1998; Iriarte and Cornelis, 1999), YscB (Jackson et al., 1998), TyeA (Iriarte et al., 1998; Cheng and Schneewind, 2000) and LcrG (DeBord et al., 2001; Matson and Nilles, 2001) proteins. Deletion of any one of the genes coding for these proteins results in a CS phenotype and in a loss of polarized translocation after cell contact (Day et al., 2003). LcrQ and the SycH chaperone have also been shown to exert an inhibitory effect on the secretion of some, but not all, Yops (Wulff-Strobel et al., 2002). The LcrG protein forms a 1–1 complex with LcrV and has been hypothesized to block Yop secretion when present in the cell in excess amount over LcrV (Matson and Nilles, 2001). SycN and YscB are T3S chaperones that interact with one another and together bind to an N-terminal region of YopN. The SycN/YscB complex is required for the stable expression of YopN and for efficient YopN secretion (Day and Plano, 1998). TyeA is a cytosolic protein that binds to a C-terminal domain of YopN (Iriarte et al., 1998) and has an inhibitory effect on YopN secretion/translocation (Day et al., 2003). YopN is a 32.6 kDa protein whose crystallographic structure in association with the SycN/YscB complex and TyeA has been recently resolved (Schubot and Waugh, 2004; Schubot et al., 2005). The structure reveals a seemingly rigid, elongated YopN/TyeA moiety that is highly α-helical in content. The authors propose that the central domain of YopN acts as a molecular ruler that extends the structure of the quaternary complex in a manner that allows its opposing ends, the YopN/SycN/YscB end and the YopN/TyeA end, to interact with equidistant binding sites on the T3S apparatus. Proper engagement of both sites by the quaternary complex would be required to block entry to the T3SS (Schubot et al., 2005). Although YopN is translocated into the eukaryotic cell in vivo (Lee et al., 1998; Day et al., 2003), no anti-host function has been ascribed to this protein.
The mechanism by which the YopN/SycN/YscB/TyeA complex blocks secretion of Yops and the physiological signals that release the block of secretion are poorly understood. At this time, two distinct hypotheses have been presented to explain how these proteins regulate secretion of Yops. First, Forsberg et al. (1991) have suggested that surface-localized YopN functions as a T3S cap that directly prevents secretion of Yops. This hypothesis is supported by the fact that YopN is seemingly found at the bacterial cell surface prior to the activation of effector Yop secretion (Forsberg et al., 1991; Iriarte et al., 1998). According to this hypothesis, upon sensing a low-calcium environment or host cell contact, the surface-localized YopN cap is released allowing secretion of Yops to proceed. Alternatively, Cheng et al. (2001) have hypothesized that both partial initiation of YopN into the T3S apparatus by the SycN/YscB chaperone complex and TyeA-mediated inhibition of YopN transport across the bacterial envelope function together to control the activity of the T3SS. Upon contact with a eukaryotic cell or in the absence of calcium, an unidentified signalling event allows YopN to complete its secretion, thus freeing the T3S pathway for delivery of the effector Yops.
All previously isolated Y. pestis YopN mutants secrete Yops constitutively, suggesting that YopN plays a critical role in preventing Yop secretion in the presence of calcium. To gain mechanistic insights on the YopN-dependent secretory block, we isolated yopN point mutants that constitutively blocked (NS phenotype) the secretion of Yops. The identified individual mutations all map to a region of yopN encoding residues 133–242 and replace highly conserved residues. The NS YopN proteins required TyeA, but not the SycN/YscB chaperone complex or LcrG, to block secretion of Yops. Correspondingly, NS YopN proteins missing the N-terminal secretion signal and chaperone-binding domain still blocked secretion, whereas NS YopN proteins lacking the TyeA-binding domain could not. These results allowed us to propose an intracellular sublocation for the YopN-dependent secretory block.
Generation of point mutations in yopN and selection of NS YopN proteins
Yersinia pestis YopN mutants described previously in the literature display a CS phenotype, characterized by growth restriction and substantial secretion of Yops at 37°C regardless of the calcium concentration in the growth medium (Yother and Goguen, 1985; Forsberg et al., 1991). To gain insights into the mechanism by which YopN regulates secretion in the Y. pestis T3SS, we isolated YopN mutants that block secretion constitutively at 37°C in the presence or absence of calcium (NS phenotype) and studied their characteristics. A library of randomly mutated yopN genes was generated by a modification of the error-prone polymerase chain reaction (PCR) technique used by Leung et al. (1989). The resulting yopN gene mixture was inserted in vector pBCKS–, generating a plasmid library of pBC-YopN derivatives which were subsequently transformed into Y. pestis KIM5-3001.P62, a yopN null strain. Individual NS YopN proteins were selected for by their ability to form colonies at 37°C in the absence of calcium. Several clones were identified that lost the ability to enter growth restriction and secrete Yops at 37°C in the absence of calcium (data not shown), and that displayed, instead, an NS phenotype similar to the one observed in ysc null strains (Plano and Straley, 1995).
Identification of the point mutations in yopN responsible for the NS phenotype
After DNA sequencing, we found that most of the selected pBC-YopN plasmids [pBC-YopN(NS-1) through pBC-YopN(NS-9), Table 1] carried multiple point mutations in different regions of yopN with the exception of pBC-YopN(NS-3), pBC-YopN(NS-6) and pBC-YopN(NS-8) which harboured only one point mutation each in their yopN gene. In an effort to eliminate the point mutation(s) not directly responsible for the NS phenotype, we performed restriction-endonuclease fragment exchanges taking advantage of naturally occurring restriction sites HindIII and BglII within the coding sequence of yopN. Except for pBC-YopN(NS-9), a single point mutation was found to be responsible for the NS phenotype in each of the pBC-YopN derivatives (Table 1; underlined residues). Each new plasmid, containing only the point mutation or mutations in yopN responsible for the NS phenotype, was renamed according to the amino-acid substitution generated by the point mutation it carried (Table 1). The identified point mutations mapped to a region of yopN that encodes residues 133–242 and whose function has not yet been defined. Moreover, the amino-acid substitutions generated by the point mutations replace highly conserved residues of YopN. In fact, all the identified amino-acid substitutions that result in an NS phenotype replace residues that are conserved in the four available YopN homologues, including LopN of Photorhabdus luminescens, AopN of Aeromonas hydrophila, PopN of Pseudomonas aeruginosa, and VopN of Vibrio harveyi (data not shown).
Table 1. Residue substitutions in YopN responsible for the NS phenotype
Residue substitutions in YopN proteins expressed by the screened plasmidsa
The residue substitutions uniquely responsible for the NS phenotype are underlined.
These plasmids, harbouring only the mutation(s) uniquely responsible for the NS phenotype, were derived from the screened plasmids by restriction-endonuclease fragment exchanges.
I21N, G36E, E37D, S133R, S272R
M52I, I174N, Q246L
Q33H, E96K, L198P
I180L, D200E, L262I
Recently, Schubot and Waugh (2004) and Schubot et al. (2005) have succeeded in resolving the crystal structures of a ternary YopN32−277/SycN/YscB–His6 complex and of a binary YopN76−293/TyeA complex. By superimposing these two structures, it was possible to build a model of the quaternary YopN/SycN/YscB/TyeA complex (Schubot et al., 2005). Interestingly, the residues of YopN that were mutated in the NS proteins (Fig. 1; side chains highlighted in red) are clustered in a central region of the quaternary complex. This region is spatially separated from the secretion targeting elements (both the SycN/YscB complex and the YopN domain interacting with it) and from the TyeA elements (both TyeA and the YopN domain interacting with it). The molecular mechanism by which the YopN/SycN/YscB/TyeA complex blocks secretion of Yops is poorly understood; however, the localization of amino-acid substitutions that allow YopN to constitutively block secretion to the central region of YopN suggests that this region plays an important role in this process.
No-secretion YopN proteins block secretion of YopM and greatly reduce their own secretion and secretion of V antigen at 37°C in the absence of calcium
To test the extent to which the various NS YopN proteins block secretion, we analysed expression and secretion of YopN, YopM and V antigen (LcrV) by the ΔyopN mutant strain transformed with plasmid pBC-YopN(WT), expressing wild-type YopN, or with each one of the recombinant plasmids expressing an NS YopN protein [YopN (S133R) through YopN (I180L, D200E)] (Fig. 2A). The parent strain secreted YopN, YopM and LcrV into the culture supernatant when grown at 37°C in the absence of calcium; however, as expected, no secretion of Yops and LcrV was detected in the presence of calcium. A ΔyscL mutant, which is defective in the assembly of a functional T3S injectisome, expressed low levels of Yops and LcrV and did not secrete them into the culture supernatant regardless of the calcium concentration in the growth medium. These results agree with published data on other secretion-deficient ysc null mutants, which display generalized downregulation of expression of pCD1 virulon genes (Michiels et al., 1991; Plano and Straley, 1995). Conversely, the ΔyopN mutant grown at 37°C in the presence or absence of calcium expressed and secreted levels of YopM and LcrV similar to those expressed and secreted by the parent strain grown in the absence of calcium. Complementation in trans of the yopN null mutant strain with a wild-type copy of yopN (Fig. 2A) restored RS of YopN, YopM and LcrV. The increased level of YopN secretion is due to its expression from the multicopy pBC-YopN(WT) construct. As expected, transformation of the ΔyopN mutant with each one of the plasmids expressing an NS YopN protein [Fig. 2A; YopN (S133R) through YopN (I180L, D200E)] generated strains with blocked secretion of YopM, and reduced or blocked secretion of LcrV and NS YopN. The amounts of Yops and LcrV found associated with these strains’ cell pellet fractions were increased or similar to those expressed by the parent strain or the YopN (WT) strain grown in the presence of calcium, indicating that the absence or reduction of Yops and LcrV in the culture supernatant was likely due to a block in secretion and not to a downregulation of Yop and LcrV expression. The YopN (F234S) mutant essentially blocked secretion of all Yops and was initially isolated twice independently (Table 1); therefore, this mutant was used in all further experiments.
To confirm that the NS YopN proteins block secretion of all highly expressed T3S substrates, YopN (F234S) was expressed in a ΔyopN Y. pestis strain cured of plasmid pPCP1, which encodes the outer-membrane plasminogen activator (Pla) protease responsible for the degradation of surface and secreted proteins (Sodeinde and Goguen, 1988; Sodeinde et al., 1988). Concentrated culture supernatant proteins were separated by SDS-PAGE and stained with Coomassie Blue R-250 (Fig. 2B). The parent strain, as well as the ΔyopN mutant complemented in trans with a wild-type copy of yopN, secreted large amounts of Yops into the culture supernatant in the absence of calcium. As expected, the ΔyopN mutant alone secreted large amounts of Yops regardless of the calcium concentration (CS phenotype). In contrast, the ΔyopN mutant transformed with plasmid pBAD-YopN(F234S) displayed no significant secretion of Yops at 37°C in the presence or absence of calcium. These results indicate that expression of YopN (F234S) produces a global block of T3S.
YopN (F234S) assembles needles on the bacterial surface but does not secrete high levels of YscF and YscP at 37°C in the absence of calcium
It has been shown that at least 10 components of the T3SS share sequence homologies with components of the bacterial flagellar hook-basal body (Blocker et al., 2003), suggesting an evolutionary relationship. These homologies imply that similar processes are involved in the assembly and function of these related systems. The flagellar T3SS initially exports only hook-type substrates until the completion of the hook structure, whereupon the substrate specificity switches to filament-type substrates (Macnab, 2003). Likewise, once the base structure of Yersinia's virulence-associated T3SS is formed, it is believed that it initially becomes competent for secretion of YscF, which makes up the needle, and YscP, which regulates the length of the needle (Journet et al., 2003). After needle assembly, a substrate specificity switch to Yop-type substrates is thought to occur (Edqvist et al., 2003). As per Agrain et al. (2005), the proximal C-terminal domain of YscP is responsible for the switch. At this point, the YopN/SycN/YscB/TyeA complex is targeted to the T3S apparatus and blocks secretion until a triggering signal is received (low calcium or cell contact). According to the model outlined above, some YscP and YscF are secreted before the YopN complex blocks T3S; therefore, NS YopN proteins may not prevent initial YscF secretion and needle assembly, but they may block or reduce secretion of excess YscP and YscF subunits after activation of the T3S process. To test this model, secretion of YscP and YscF was analysed by immunoblotting of trichloroacetic acid (TCA)-precipitated culture supernatant proteins in Y. pestis strains cured of plasmid pPCP1 (Fig. 3A). The parent strain secreted detectable amounts of YscP and YscF in the culture supernatant fraction in the absence of calcium but not in the presence of calcium. As expected, a strain defective in the assembly of a functional T3S apparatus (ΔyscF) did not secrete YscP into the culture supernatant, irrespective of the presence of calcium in the growth medium, whereas a ΔyscP mutant, as shown before by Edqvist et al. (2003), secreted equal amounts of YscF in the presence and absence of calcium. The ΔyopN mutant grown at 37°C exhibited the characteristic CS phenotype, secreting YscP and YscF in the presence and absence of calcium. Providing complementing plasmid pBAD-YopN(WT) to the yopN null mutant completely restored RS of YscP and YscF. In contrast, transformation of the yopN null mutant with pBAD-YopN(F234S) prevented all detectable secretion of YscP and YscF at 37°C in the presence and absence of calcium.
To determine if bacteria that express the YopN (F234S) protein assemble needles on their surface, we used a cross-linking assay originally developed for the analysis of Chlamydia trachomatis T3S needle assembly (K.A. Fields, unpubl.). Bacterial cells grown at 37°C in the presence or absence of calcium were harvested, washed and treated with 5 mM BS3, a membrane-impermeable, homobifunctional protein cross-linking agent. Cross-linked cell-surface proteins were analysed by SDS-PAGE and immunoblotting with antibody specific for YscF. YscF associated with the cell pellet fraction migrated as an approximately 9 kDa monomer in the absence of BS3 cross-linking (Fig. 3A and data not shown). BS3 cross-linking of surface-exposed YscF in the parent, in the ΔyscP, in the ΔyopN and in the ΔyopN strain transformed with pBAD-YopN(WT) or pBAD-YopN(F234S) resulted in the formation of an anti-YscF antibody reactive ladder of bands that corresponded in size to various YscF-YscF multimers (dimers, trimers, and so on) plus a variable number of BS3 molecules (mw = 572.43). The presence of surface-localized YscF that can be readily cross-linked to itself suggests that an assembled needle is present on the bacterial surface both in the presence and in the absence of calcium, a fact that has been confirmed by electron microscopy in Yersinia enterocolitica (Hoiczyk and Blobel, 2001). No cross-linking of YscF was seen in a ΔyscJ mutant, which cannot export YscF, confirming that BS3 cannot enter the bacterial cell. These results suggest that the NS YopN (F234S) protein blocks T3S only after the completion of needle assembly and following a substrate specificity switch from needle-type substrates to Yop-type substrates.
YopN (F234S) blocks translocation of Elk-tagged YopE into HeLa cells
Another known environmental cue that triggers activation of Yersinia's T3SS is contact with a eukaryotic cell. To determine whether NS YopN proteins can also block cell contact-dependent secretion and translocation of Yops into host cells, we used a recently developed reporter system that utilizes a phosphorylatable peptide tag, termed the Elk tag (Day et al., 2003), to follow the translocation of tagged proteins. Translocation of an Elk-tagged protein into a eukaryotic cell results in host cell protein kinase-dependent phosphorylation of the tag, which can subsequently be detected with phosphospecific antipepetide antibodies. A ΔyopN mutant strain alone and transformed with pBC-YopN(WT) or pBC-YopN(F234S), and a ΔyopEJB mutant strain, all carrying plasmid pDHK-YopE129-Elk, were used to infect HeLa cell monolayers (Fig. 4). Immunoblot analysis of the infected HeLa cell lysates using the Elk antibody (α-Elk) revealed that all the strains expressed comparable levels of the YopE129–Elk hybrid protein. As shown previously (Day et al., 2003), the ΔyopN mutant translocated a reduced amount of YopE129–Elk into HeLa cells when compared with that of the same strain complemented with the plasmid expressing wild-type YopN. In contrast, transformation of the yopN null mutant with the plasmid expressing YopN (F234S) prevented phosphorylation of the YopE129–Elk hybrid protein, indicating that YopN (F234S) blocked delivery of YopE129–Elk into the eukaryotic cytoplasm. As expected, the translocation-defective yopB null mutant (ΔyopEJB) showed no translocation of the Elk-tagged YopE protein. These data indicate that YopN (F234S) blocks both secretion triggered by low-calcium concentrations in vitro and secretion/translocation triggered by cell contact in a tissue culture model.
Secretion of YopM by Y. pestis yopN sycN, yopN yscB, yopN tyeA and yopN lcrG double null strains expressing YopN (F234S) in trans
The calcium- and cell contact-dependent regulation of Yop secretion is dependent upon the YopN, SycN, YscB, TyeA and LcrG proteins. To investigate the requirements of the NS YopN proteins to block secretion, we moved the plasmid construct expressing YopN (F234S) into different double null Y. pestis backgrounds (Fig. 5A). Secretion of YopM was analysed by immunoblotting of TCA-precipitated culture supernatant proteins from the wild-type parent strain, the ΔyopN mutant, and the different double null strains, all alone, and transformed with a plasmid expressing either wild-type YopN or YopN (F234S). Panel a in Fig. 5A shows RS of YopM by the parent strain and CS of YopM by the ΔyopN strain and by the four double null strains. Complementation of the same strains with plasmid pBC-YopN(WT) fully restored RS in the ΔyopN strain and left the parent strain's phenotype unaltered (Fig. 5A, panel b). Conversely, the double null strains expressing wild-type YopN lacked the ability to regulate secretion because they were still deleted in one other gene (i.e. sycN, yscB, tyeA or lcrG) responsible for the negative regulation of secretion. Expression of pBC-YopN(F234S) effectively blocked secretion of YopM in the ΔyopN strain and in the parent strain, suggesting in this last case a strong dominant-negative effect of YopN (F234S) on native wild-type YopN (Fig. 5A, panel c). Furthermore, expression of YopN (F234S) in the yopN sycN, yopN yscB and yopN lcrG double null strains also blocked the secretion of YopM. The fact that the block of Yop secretion mediated by YopN (F234S) is not dependent upon the presence of the SycN and YscB chaperones suggested that efficient chaperone-mediated targeting of YopN (F234S) to the T3SS is not required for YopN (F234S) to block secretion of Yops. Also, the constitutive block of Yop secretion observed in the ΔyopNΔlcrG strain expressing YopN (F234S) indicates that the NS YopN proteins do not require LcrG to block secretion. Interestingly, expression of YopN (F234S) in the ΔyopNtyeA strain did not block YopM secretion, indicating an absolute requirement for the presence of the TyeA protein in order for YopN (F234S) to block secretion.
The inability of YopN (F234S) to block secretion in the ΔyopNtyeA strain suggested that this strain could be used to determine whether the NS YopN proteins could be secreted by strains that express the SycN/YscB chaperone. The ΔyopN, ΔyopNtyeA and ΔyopNΔlcrG strains expressing wild-type YopN showed RS (ΔyopN) or CS (ΔyopNtyeA and ΔyopNΔlcrG) of YopN (Fig. 5B, panel a). No appreciable secretion of YopN (F234S) was observed in the presence or absence of calcium in the ΔyopN and ΔyopNΔlcrG strains transformed with plasmid pBC-YopN(F234S), confirming the YopN (F234S)-dependent block in T3S (Fig. 5B, panel b). Interestingly, YopN (F234S) was efficiently secreted by the ΔyopNtyeA strain, suggesting that the constitutive block of Yop secretion observed in the other strains is not due to the expression of an NS YopN protein that is inherently non-secretable, but is rather the result of a specific enhancement in the ability of the YopN protein to negatively regulate secretion of Yersinia's virulence factors. Equivalent results in all of the above strains were also obtained using pBC-YopN(L198P) (data not shown).
YopN (F234S) deleted for its secretion signal and chaperone-binding domain blocks secretion of YopM
To investigate the role of the N-terminal secretion signal and of the YopN chaperone-binding domain in blocking secretion, we used plasmids pBC-YopN(Δ2-15) and pBC-YopN(Δ2-85) which encode wild-type YopN deleted, respectively, for the N-terminal secretion signal and for both the secretion signal and chaperone-binding domain (Jackson et al., 1998). We also constructed similarly deleted versions of YopN (F234S) in plasmids pBC-YopN(Δ2-15, F234S) and pBC-YopN(Δ2-85, F234S). After transformation of these constructs into a Y. pestisΔyopN mutant, immunoblot analysis of TCA-precipitated culture supernatant proteins with YopM polyclonal antibody showed that the strains carrying the deletions in wild-type YopN [YopN (Δ2-15) and YopN (Δ2-85)] displayed a CS phenotype (Fig. 6, panel a), characterized by secretion of YopM in the presence and absence of calcium. Whereas deletion of the secretion signal or both the secretion signal and chaperone-binding domain of wild-type YopN resulted in CS of YopM, the deleted YopN proteins themselves were not secreted [YopN (Δ2-85)] or were secreted inefficiently [YopN (Δ2-15)] (Fig. 6, panel b). These results indicate that wild-type YopN must be efficiently targeted to the T3SS via an N-terminal secretion signal and the SycN/YscB chaperone (Fig. 5A) in order to block Yop secretion. Interestingly, the presence of the F234S substitution in similarly deleted YopN proteins [Fig. 6, YopN (Δ2-15, F234S) and YopN (Δ2-85, F234S)] was sufficient to bring about a complete block of Yop secretion in the presence or absence of calcium. The above data, together with the lack of requirement for the SycN and YscB chaperones (Fig. 5A), indicate that the NS YopN proteins do not have to be secreted or targeted to the T3S complex by identified secretion signals to block secretion of Yops, suggesting that YopN (F234S) and TyeA are capable of blocking Yop secretion from an intracellular location, namely from the cytosolic compartment. The fact that NS YopN proteins block secretion in the absence of all identified secretion signals indicates that YopN and/or TyeA have a unique secretion signal-independent means to interact with the T3SS and block the secretion of Yops.
YopN (F234S) truncated at its C-terminal domain, and thus unable to interact with TyeA, exhibits a CS phenotype
The fact that NS YopN proteins require the presence of the TyeA protein to block Yop secretion (Fig. 5) suggests that the YopN/TyeA interaction is a prerequisite for wild-type YopN and the NS YopN proteins to block secretion. To confirm the importance of the YopN/TyeA interaction on the ability of wild-type YopN and YopN (F234S) to block secretion, we used plasmids pBC-YopN(Q280Stop), pBC-YopN(F234S, Q280Stop), pBC-YopN(Q274Stop) and pBC-YopN(F234S, Q274Stop), which encode wild-type YopN or YopN (F234S) truncated at codon 280 or 274 of the yopN sequence by substitution of a glutamine codon with a stop codon. These YopN proteins lacking, respectively, 14 and 20 residues at their C-terminus are unable to interact with TyeA, as shown by Ferracci et al. (2004). The various pBC-YopN constructs were transformed into the ΔyopN mutant strain, and SDS-PAGE and immunoblot analysis of cell pellets and TCA-precipitated culture supernatants showed that the ΔyopN strains carrying the truncations in YopN (F234S) [YopN (F234S, Q280Stop) and YopN (F234S, Q274Stop)] had a CS pattern that was indistinguishable from that of the ΔyopN strains carrying the same truncations in wild-type YopN [YopN (Q280Stop) and YopN (Q274Stop)]. All four truncated YopN proteins appeared to be secreted constitutively in essentially equal amounts (Fig. 7; secreted YopN’ clipped by the outer-membrane Pla protease); although, greater amounts of unprocessed YopN* were present in the supernatant from the YopN (Q280Stop) and the YopN (Q274Stop) derivatives than from the F234S derivatives. In contrast, the ΔyopN strain carrying a plasmid expressing YopN (F234S) deleted for both its secretion signal and chaperone-binding domain [YopN (Δ2-85, F234S)] blocked secretion constitutively, as seen earlier in Fig. 6 for YopM and YopN and again here in Fig. 7 for YopM, LcrV and YopN. These results indicate that the interaction of YopN (F234S) with TyeA is crucial to effect a constitutive block in Yop secretion, whereas the interaction of YopN (F234S) with cognate heterodimeric chaperone SycN/YscB is not.
Secretion of Yop effector proteins by the Y. pestis virulence-associated T3SS is triggered by growth in the absence of extracellular calcium (Brubaker and Surgalla, 1964; Michiels et al., 1990) or by contact with a eukaryotic cell (Rosqvist et al., 1994; Persson et al., 1995). The YopN/SycN/YscB/TyeA complex is required to prevent Yop secretion in the presence of calcium and prior to contact with a eukaryotic cell. The mechanism by which these proteins prevent Yop secretion is not known; however, the secreted YopN protein is thought to play a central role in this process (Yother and Goguen, 1985; Forsberg et al., 1991; Persson et al., 1995). In order to gain insight into the mechanism by which YopN prevents Yop secretion, we isolated and characterized several NS YopN proteins that constitutively block Yop secretion (Table 1).
Expression of the NS YopN proteins in Y. pestis blocked high-level secretion of both effector Yops (Fig. 2) and translocator Yops (Fig. 2B) both in the presence and in the absence of calcium. The NS YopN proteins also blocked cell contact-dependent secretion and translocation of effector Yops into cultured eukaryotic cells (Fig. 4). The identified amino-acid substitutions that resulted in an NS phenotype mapped to a central region of YopN (Fig. 1) that is not observed to be involved in the interaction of YopN with SycN, YscB or TyeA. The identified amino-acid residues are highly conserved in the various YopN orthologues and are almost entirely buried (Fig. 1). These data suggest that these residues are important and likely contribute to the overall stability of the folded structure. Correspondingly, mutations that specifically alter these conserved residues are likely to cause part of YopN to become unfolded or loosely ordered.
Both wild-type YopN and the NS YopN protein require an intact TyeA-binding domain (Fig. 7) and the TyeA protein (Fig. 5) in order to block secretion. As previously reported, wild-type YopN also requires an intact N-terminal secretion signal, chaperone-binding domain and the SycN/YscB chaperone to block Yop secretion in the presence of calcium (Day and Plano, 1998; Cheng et al., 2001). Unexpectedly, the NS YopN (F234S) protein deleted of its N-terminal secretion signal and chaperone-binding domain still blocked Yop secretion in the presence and absence of calcium (Figs 6 and 7). The YopN (F234S) protein also blocked Yop secretion when expressed in a sycN, yscB or lcrG deletion mutant (Fig. 5). These results suggest that this protein does not have to be targeted to the T3SS via identified N-terminal secretion signals or chaperones in order to block secretion. These data further suggest that the cytosolic YopN/TyeA complex has at least two independent means to interact with the T3S apparatus: (i) via the YopN N-terminal secretion signal and chaperone-binding domain, and (ii) via the YopN C-terminal domain complexed with TyeA. Data presented here suggest that the latter interaction is essential for YopN/TyeA to block secretion. Interestingly, at least one member of the extended YopN family of regulatory proteins, InvE of Salmonella enterica, has been shown to be a cytosolic protein that regulates T3S from within the bacterial cell (Kubori and Galán, 2002). This confirms that members of the YopN family of regulatory proteins can regulate T3S from a cytosolic location.
The additional information gained from the characterization of these unique NS YopN proteins has allowed us to develop a modified working model (Fig. 8) for how the YopN/SycN/YscB/TyeA complex may function to regulate Yop secretion. We theorize that following completion of needle assembly, the YopN N-terminal secretion signal and the SycN/YscB chaperone function together to target the YopN/TyeA complex to the T3S apparatus. Initiation of wild-type YopN into the secretion channel (Fig. 8, site 1) results in a conformational change or localized unfolding of the YopN protein which allows the YopN/TyeA complex to participate in a second unique interaction with a cytosolic-exposed portion of the T3S apparatus (Fig. 8, site 2). Interaction of the YopN/TyeA complex at site 2 is hypothesized to directly or indirectly block the T3S process. In the absence of calcium or upon contact with a eukaryotic cell an unidentified signalling event alters or disrupts the site 2–YopN/TyeA interaction and allows YopN to complete its secretion, thus opening the T3S pathway for effector Yop secretion.
The LcrG protein has also been shown to play an essential role in the regulation of Yop secretion. Unbound LcrG, present in excess of LcrV, blocks Yop secretion in a YopN- and TyeA-dependent manner (Matson and Nilles, 2001); however, no direct interaction between YopN or TyeA and LcrG has been identified. Previous studies have suggested that LcrG, like TyeA, works to prevent efficient secretion and translocation of YopN (Day et al., 2003). Thus, LcrG could function to regulate effector Yop secretion by stabilizing the binding of the YopN/TyeA complex at site 2, thereby preventing YopN secretion/translocation and subsequent effector Yop secretion/translocation.
We hypothesize that the NS YopN proteins are in a conformation, or are partially unfolded in a manner, that allows these proteins, together with TyeA, to interact at site 2 without having to be targeted to the secretion apparatus via the YopN N-terminal secretion signal and/or the SycN/YscB chaperone. The presence of a cytoplasmic pool of YopN proteins that are all in a conformation that can interact at site 2 without having to be targeted for secretion results in site 2 being consistently occupied and secretion being constitutively blocked. This model provides a plausible explanation for why wild-type YopN requires both a functional secretion signal and the SycN/YscB chaperone in order to block secretion, whereas the NS YopN proteins do not.
The ability of the NS YopN proteins to block at site 2 both in the presence and absence of calcium could be explained by the fact that the NS YopN proteins have a higher affinity for site 2 than the wild-type YopN protein or by the fact that the entire cytoplasmic pool of NS YopN proteins are in a conformation capable of competing for site 2, thus keeping site 2 occupied at all times. In contrast, wild-type YopN is hypothesized to recognize and interact at site 2 only following its initiation into the T3S pathway. Confirmation of these hypotheses awaits identification of site-2 components.
The identity of the T3SS component(s) with which YopN/TyeA interacts to block Yop secretion (site 2) is not known; however, site 2 is most likely an essential cytosolic Ysc protein (YscK, YscL, YscN or YscQ) or an integral inner-membrane Ysc protein that has a domain exposed on the cytoplasmic face of the inner membrane (YscR, YscU or YscV). Future experiments will be aimed at identifying the protein or proteins that constitute site 2. The non-secretable YopN proteins may serve as tools or probes to facilitate the identification of site-2 target proteins via coimmunoprecipitation studies, GST-pulldown experiments or via bacterial or yeast two-hybrid studies. The identification of the T3S component(s) that constitute site 2 will allow us to directly investigate the interaction of both wild-type YopN and the NS YopN proteins with these components.
Bacterial strains and culture conditions
Escherichia coli and Y. pestis strains used in this report are listed in Table 2. All Y. pestis strains in this study are avirulent from peripheral routes of infection due to deletion of the pgm locus (Une and Brubaker, 1984). Y. pestis and E. coli strains were grown routinely in heart infusion broth (HIB) or on tryptose blood agar (TBA) base plates (Difco Laboratories) at a temperature of 27°C or 37°C, respectively, unless otherwise noted. For growth-curve experiments, Y. pestis strains were cultured overnight in a shaker bath at 27°C in the defined medium TMH (Goguen et al., 1984), and were diluted the next day to an optical density at 620 nm (OD620) of approximately 0.20 in 2 ml of fresh TMH, with or without 2.5 mM calcium chloride. After 1 h of growth at 27°C, cultures were shifted to 37°C and grown for an additional 5 h.
Table 2. Bacterial strains and plasmids used in this study
BamHI-XhoI fragment excised from pYopE129-Elk (Day et al., 2003) and cloned into BamHI-SalI-digested pDHK29
ΔyopN (aa 2–15) KpnI-XbaI PCR-ligation-PCR fragment generated from template pBC-YopN(WT) and cloned into KpnI-XbaI-digested pBCKS–
ΔyopN (aa 2–85) KpnI-XbaI PCR-ligation-PCR fragment generated from template pBC-YopN(WT) and cloned into KpnI-XbaI-digested pBCKS–
ΔyopN (aa 2–15) KpnI-XbaI PCR-ligation-PCR fragment generated from template pBC-YopN(F234S) and cloned into KpnI-XbaI-digested pBCKS–
ΔyopN (aa 2–85) KpnI-XbaI PCR-ligation-PCR fragment generated from template pBC-YopN(F234S) and cloned into KpnI-XbaI-digested pBCKS–
BglII-BamHI PCR fragment generated from template pYopN-TyeA (Q280 > Stop) (Ferracci et al., 2004) and cloned into BglII-BamHI-digested pBC-YopN(WT)
BglII-BamHI PCR fragment generated from template pYopN-TyeA (Q274 > Stop) (Ferracci et al., 2004) and cloned into BglII-BamHI-digested pBC-YopN(WT)
BglII-BamHI PCR fragment generated from template pYopN-TyeA (Q280 > Stop) (Ferracci et al., 2004) and cloned into BglII-BamHI-digested pBC-YopN(F234S)
BglII-BamHI PCR fragment generated from template pYopN-TyeA (Q274 > Stop) (Ferracci et al., 2004) and cloned into BglII-BamHI-digested pBC-YopN(F234S)
The differential medium TBA magnesium oxalate (TBA MgOX) was prepared according to Higuchi and Smith (1961) by aseptic incorporation of aqueous solutions of sodium oxalate and magnesium chloride (20 mM final concentration) into melted TBA. E. coli SY327 λpir (Miller and Mekalanos, 1988) was used to propagate derivatives of the suicide vector pUK4134 (Skrzypek et al., 1993), whereas E. coli DH5α was used for routine cloning experiments. Bacteria with resistance markers were grown in the presence of the appropriate antibiotic(s) at a final concentration of 25 µg ml−1 (chloramphenicol and kanamycin), 50 µg ml−1 (ampicillin and streptomycin) or 75 µg ml−1 (carbenicillin).
Plasmid DNA was isolated by using the GenElute™ Plasmid Miniprep Kit (Sigma), the HiSpeed™ Plasmid Midi Kit (Qiagen), or the method of Kado and Liu (1981). DNA fragments were precipitated using Pellet Paint® Co-Precipitant (Novagen) or were gel purified using the Qiaex® II Gel Extraction Kit (Qiagen). DNA was routinely eluted with or resuspended in Tris (5 or 10 mM, pH 8). Electroporation of E. coli and Y. pestis was performed as described previously (Perry et al., 1990). The PCR technique was performed using Taq (Promega) or Pfu Turbo® (Stratagene) DNA polymerases according to the manufacturers’ instructions, with modifications as noted. Double-stranded DNA was sequenced by the University of Miami School of Medicine DNA Sequencing Core Facility using a DyeDeoxy Terminator cycle sequencing kit and an ABI model 373 A DNA sequencer (Applied Biosystems). Nucleotide sequences were analysed with IntelliGenetics computer software (IntelliGenetics).
Generation, selection and isolation of NS point mutations in yopN
The DNA fragment encoding full-length yopN and its promoter was PCR-amplified from plasmid pCD1 using high-fidelity Pfu Turbo® polymerase and primers 5′-TTTGGTACC GAAAAATAGCCAAGCAGCACT-3′ and 5′-TCCGATGCATTC AGAAAGGTCGTACGCCATTAG-3′. The resulting fragment was digested with KpnI and NsiI (underlined) and inserted into KpnI-PstI-digested plasmid pBCKS– (Stratagene), generating plasmid pBC-YopN(WT). An error-prone PCR technique (Leung et al., 1989) and Taq polymerase were used to generate a randomly mutated library of yopN with the same template and primers as above. The resulting yopN gene mixture was inserted into vector pBCKS–, as described above, generating a plasmid library of pBC-YopN derivatives which were subsequently transformed into Y. pestis KIM5-3001.P62 (ΔyopN). Three 24-h growth cycles in TMH at 37°C in the absence of calcium enriched the fraction of yopN mutants with an NS growth phenotype, while selecting against RS strains and CS mutants (which exhibit growth restriction at 37°C in the absence of calcium). Serially diluted aliquots were then plated on TBA MgOX plates and incubated directly at 37°C. A number of fast-growing, large colonies were individually picked and tested as described in Results. After DNA sequencing, we performed restriction-endonuclease fragment exchanges on the isolated plasmids with more then one point mutation by taking advantage of pBCKS– cloning sites KpnI and BamHI and of naturally occurring restriction sites HindIII and BglII within the coding sequence of yopN. Plasmids pBC-YopN(S133R) and pBC-YopN(I174N) were generated by digesting pBC-YopN(NS-1) and pBC-YopN(NS-2), respectively, with HindIII and BglII, followed by insertion of the excised HindIII-BglII fragment into HindIII-BglII-digested pBC-YopN(WT). Plasmids pBC-YopN(L198P) and pBC-YopN(F234S) were generated by digesting pBC-YopN(NS-4) and pBC-YopN(NS-7), respectively, with BglII and BamHI, followed by insertion of the excised BglII-BamHI fragment into BglII-BamHI-digested pBC-YopN(WT). Plasmids pBC-YopN(Y202N) and pBC-YopN(I180L, D200E) were generated by digesting pBC-YopN(NS-5) and pBC-YopN(NS-9), respectively, with KpnI and HindIII, followed by insertion of the excised KpnI-HindIII fragment into KpnI-HindIII-digested pBC-YopN(WT). These were the only fragment exchange combinations that produced plasmids expressing NS YopN proteins.
Construction of Y. pestis deletion mutants
Previously constructed suicide vectors pUK4134.P3 (ΔsycN) (Day and Plano, 1998) and pUK4134-ΔlcrG2 (Nilles et al., 1998), containing in-frame deletions eliminating, respectively, the coding sequence for amino acids 34–65 of SycN and 5–95 of LcrG, were used to move the sycN and lcrG deletions into Y. pestis KIM5-3001.P62 (ΔyopN). Likewise, previously constructed suicide vector pUK4134.P1 (ΔyscB) (Jackson et al., 1998), containing an in-frame deletion eliminating the coding sequence for amino acids 61–125 of YscB, was used to move the yscB deletion into Y. pestis KIM5-3001.6 (ΔyopN). Y. pestis KIM5-3001.P68 (ΔyopNΔsycN) contained the correct in-frame deletion in sycN, Y. pestis KIM5-3001.P69 (ΔyopNΔlcrG) contained the correct in-frame deletion in lcrG, and Y. pestis KIM5-3001.P70 (ΔyopNΔyscB) contained the correct in-frame deletion in yscB.
Construction of the pBC-YopN(Δ2-15), pBC-YopN(Δ2-85), pBC-YopN(Δ2-15, F234S) and pBC-YopN(Δ2-85, F234S) expression plasmids
Deletions of yopN sequences encoding amino-acid residues 2–15 and 2–85 were constructed by the PCR-ligation-PCR technique (Ali and Steinkasserer, 1995). The DNA fragment encoding yopN′s promoter and start codon (fragment A) was amplified from plasmid pCD1 using primers 5′-TTTGGTACC CATAAATCCATAATGGTTGAA-3′ and 5′-CATAACTACTCCC TGAGATGAACA-3′. The yopN sequences encoding amino-acid residues 16–293 and 86–293 were amplified from both plasmid pBC-YopN(WT) and plasmid pBC-YopN(F234S) using upstream primers 5′-AATGAGCGTCCAGAGATTGCCA GT-3′ or 5′-CAGGTTAATCAATACCTTAGCAAA-3′ and downstream primer 5′-TTTTCTAGATCAGAAAGGTCGTACGCCA TT-3′. These PCR reactions produced four different PCR fragments. Fragment A was then ligated to each one of the four PCR fragments generated above, and the ligation products were reamplified using primers 5′-TTTGGTACCCATAA ATCCATAATGGTTGAA-3′ and 5′-TTTTCTAGATCAGAAAGG TCGTACGCCATT-3′, carrying, respectively, KpnI and XbaI sites (underlined). The resulting fragments were digested with KpnI and XbaI and inserted into KpnI-XbaI-digested pBCKS–, generating plasmids pBC-YopN(Δ2-15), pBC-YopN(Δ2-85), pBC-YopN(Δ2-15, F234S) and pBC-YopN(Δ2-85, F234S).
Construction of the pBC-YopN(Q280Stop), pBC-YopN(Q274Stop), pBC-YopN(F234S, Q280Stop) and pBC-YopN(F234S, Q274Stop) expression plasmids
Previously developed constructs pYopN-TyeA (Q280 > Stop) and pYopN-TyeA (Q274 > Stop) (Ferracci et al., 2004), carrying, respectively, a single nonsense mutation at codon 280 or 274 of the yopN sequence, were used as templates with upstream primer 5′-CAAGGAATTTATGCGATCTGGAGTGAT-3′ and downstream primer 5′-TTTGGATCCTCAGAAAG GTCGTACGCCATT-3′ to generate two separate PCR products. Both fragments amplified a region from codon 210 of yopN to its native termination codon at position 294, and each included a single nonsense mutation (Q280Stop or Q274Stop). After digestion with BglII (naturally occurring restriction site between codons 242 and 243 of yopN) and with BamHI (underlined), the fragments were inserted into BglII-BamHI-digested pBC-YopN(WT), generating plasmids pBC-YopN(Q280Stop) and pBC-YopN(Q274Stop), and into BglII-BamHI-digested pBC-YopN(F234S), generating plasmids pBC-YopN(F234S, Q280Stop) and pBC-YopN(F234S, Q274Stop).
SDS-PAGE and immunoblotting
Cell pellets and culture supernatants were separated by centrifugation at 12 200 g for 10 min at 4°C. Culture supernatant proteins were precipitated on ice overnight with 10% (v/v) TCA and collected by centrifugation at 12 200 g for 15 min at 4°C. Volumes of cellular fractions (whole-cell or secreted proteins) corresponding to equal numbers of bacteria were mixed 1:1 (v/v) with 2× electrophoresis sample buffer and analysed by SDS-PAGE followed by staining with Coomassie blue or immunoblotting, which was performed essentially as previously described (Plano and Straley, 1995). Briefly, samples were separated on NuPAGE™ 12% Bis-Tris acrylamide gels (Invitrogen) using a MES or a MOPS buffer system. Separated proteins were transferred to Immobilon-P Transfer Membranes (Millipore) using a Tris-glycine buffer system, and membranes were blocked with 5% (w/v) non-fat milk (NFM) in TBS (20 mM Tris, 150 mM NaCl, pH 7.4) for at least 30 min at room temperature (RT). Polyclonal antisera or purified antibodies were diluted in TBS containing 1% NFM and 0.05% (v/v) Tween 20 and incubated with the blots overnight at 4°C. Blots were then washed three times with TBS containing 0.1% NFM and 0.05% Tween 20 for 10 min. Secondary antibody (alkaline phosphatase-conjugated anti-rabbit IgG) was diluted in TBS containing 1% NFM and 0.05% Tween 20 and incubated with the blots for 1 h at RT. Blots were then washed again as described above and developed with 1-Step™ BCIP/NBT (Pierce). Images of the blots were captured by the ChemiImager 5500 digital imaging system with Alpha Innotech software.
Tissue culture infections
Tissue culture infections of HeLa cells were done as previously described (Day et al., 2003).
Cell-surface cross-linking of the YscF needle with BS 3
Cultures were grown at 37°C in 2 ml of TMH with or without 2.5 mM calcium chloride. After 5 h, 1 ml volumes were harvested by centrifugation at 8000 g for 5 min at 4°C. Each bacterial pellet from growth in the presence of calcium was gently resuspended in 1 ml of cold 20 mM HEPES, 5 mM calcium chloride (pH 8), whereas bacterial pellets from growth in the absence of calcium were gently resuspended in 1 ml of cold 20 mM HEPES (pH 8) without calcium. Bacterial surface proteins were cross-linked for 30 min at RT with the non-cleavable, membrane-impermeable, amine-reactive cross-linker bis(sulphosuccinimidyl) suberate (BS3) (Pierce) at a final concentration of 5 mM (25 mM stock solution in 20 mM HEPES, pH 8.0). Cross-linking reactions were quenched for 15 min by addition of Tris (pH 8.0) to a final concentration of 20 mM. Cell fractions were collected by centrifugation at 12 200 g for 5 min at 4°C and treated as noted in SDS-PAGE and immunoblotting.
We thank Dr Ken A. Fields for critical reading of the manuscript and for useful comments. This work was supported by Public Health Service Grants AI39575 and AI50552 from the National Institute of Health to G.V.P.