Salmonella pathogenicity island 1 (SPI-1) type III secretion of SopD involves N- and C-terminal signals and direct binding to the InvC ATPase

Salmonella enterica serovar Typhimurium (S. Typhimurium) is a facultative intracellular pathogen, causing salmonellosis (Hohmann, 2001). During the course of infection, S. Typhimurium translocates bacterial virulence effector proteins from the bacterial cytoplasm directly into the host-cell cytoplasm via two major type III secretion systems (T3SSs) located in discrete regions of its chromosome or pathogenicity islands (Galán, 2001; Ochman & Groisman, 1996). Salmonella pathogenicity island 1 (SPI-1) is located within centisome 63 of the bacterial chromosome (Galán, 1999). SPI-1 is expressed in the initial infectious stage (Wallis & Galyov, 2000), and translocated effectors modulate key host-cell functions, including general signal transduction, cytoskeletal architecture, membrane trafficking and cytokine expression in favour of the pathogen (Hayward & Koronakis, 1999; Lilic et al., 2003; Schlumberger & Hardt, 2005). A second T3SS is encoded within Salmonella pathogenicity island 2 (SPI-2) at centisome 31, and is required for systemic infection (Hensel, 2000; Waterman & Holden, 2003).

At least 20 effector proteins are delivered by the SPI-1 T3SS. The mechanism underlying secretion via the T3SS has been an area of controversy for many years. There is still no consensus sequence that describes a universal type III secretion (T3S) signal. An N-terminal secretion signal (NSS) may involve either a signal peptide or, as seen in Yersinia spp., an mRNA signal sequence (Anderson et al., 1999; Anderson & Schneewind, 1999; Sorg et al., 2005, 2006). In the signal peptide hypothesis, two essential regions are required for T3S (Cheng et al., 1997; Sory et al., 1995). The first region encodes a secretion signal within the first ∼20 amino acids of the effector protein (Cornelis & Van, 2000; Sorg et al., 2005, 2006). The process of T3S was described by Akeda & Galán (2005), who demonstrated that the effector protein SptP in complex with its cognate chaperone, SicP, interacts with the SPI-1 T3SS ATPase, InvC. The InvC ATPase is part of the T3SS injectisome (Muller et al., 2006). Following this interaction, InvC disassociates the substrate–chaperone complex and unfolds SptP, which is then driven through the T3S needle. In S. Typhimurium, introduction of a +1 or +2 frameshift at codon 10 of effectors SopE and SptP, to change the protein sequence between residues 11 and 35, significantly reduces secretion through T3S (Lee & Galán, 2004). We have demonstrated that the first 15 amino acids are important for the secretion of SopE independently of the presence of the chaperone-binding site (Karavolos et al., 2005). Also, in SipB, residues 3–8 are necessary for its secretion from the bacterial cell (Kim et al., 2007). However, in silico comparison of the N-terminal signal regions of known effectors has not led to the identification of any conserved sequences, implying that features such as amphipathicity or secondary structure serve as recognition motifs (Lloyd et al., 2001, 2002; Miao & Miller, 2000; Tampakaki et al., 2004).

Lee & Galán (2004) indicated that the NSS of SptP is not sufficient to mediate secretion through its cognate T3SS, requiring a secondary NSS identified as the chaperone-binding domain (CBD). The CBD is located within the first ∼140 amino acids of some secreted proteins and harbours the binding site for the cognate chaperone, which is generally required for efficient transport via the T3S apparatus (Ghosh, 2004; Lee & Galán, 2003). In the case of SopE, Ehrbar and co-workers have noted that an alternative function of the CBD is to prevent erroneous SopE secretion via the flagellar T3SS in the absence of its chaperone, InvB (Ehrbar et al., 2006; Lee & Galán, 2004). It has been shown that effector proteins containing only the N-terminal signal peptide but lacking a CBD are secreted through the flagellar T3SS (Kim et al., 2007; Lee & Galán, 2004). Indeed, the flagellar export machinery has been suggested to be an ancestor of the SPI-1 T3SS (Nguyen et al., 2000), as many components of the SPI-1 and flagellar export apparatuses show high sequence homology and similarity (Foultier et al., 2002; Gophna et al., 2003).

In S. Typhimurium T3SS, chaperones are an essential part of the effector secretion mechanism. Most chaperones have additional functions that prevent aggregation (Boyd et al., 2000) and premature interactions in the bacterial cytoplasm (Day et al., 2000). Additional functions include preserving the unfolded state of proteins (Stebbins & Galán, 2001) and protecting them from degradation (Frithz-Lindsten et al., 1995), as well as regulating effector expression (Darwin & Miller, 2001).

The Salmonella effector SopD is encoded outside SPI-1 and contributes to inflammation and fluid secretion during gastroenteritis in the bovine model of Salmonella infection (Jones et al., 1998). Genome sequence analysis has also revealed the presence of a SopD homologue termed SopD2, which possesses a different intracellular function to that of SopD (Brumell et al., 2003). Expression of sopD is maintained at later stages of infection when other SPI-1 effectors are not expressed, indicating its involvement in systemic disease in mice (Brumell et al., 2003). Additionally, the bacteria require SopD for optimal replication in mouse macrophages (Jiang et al., 2004b). SopD appears to be acting as a dual effector of both SPI-1 and SPI-2 T3SSs. Thus, SopD is found in the host-cell cytosol not only during the early stages of infection but also later in the Salmonella-containing vacuole (SCV) (Brumell et al., 2003). Moreover, SopD and SopB act cooperatively to enhance membrane fission and to promote macropinocytosis during S. Typhimurium invasion (Bakowski et al., 2007). Wood et al. (2004) have suggested that SopD consists of a single compact domain and does not have a cognate bacterial chaperone, implying that an alternative, chaperone-independent mechanism may be used for its secretion via the T3S apparatus.

In this study, we have dissected the contribution of various SopD domains in T3S. We have identified an NSS essential for SPI-1 T3S. Interestingly, a C-terminal region is essential to prevent misdirected secretion through the flagellar apparatus. We show that for successful secretion, SopD interacts with the SPI-1 T3SS ATPase, InvC. Finally, we propose a model for the secretion of SopD through the SPI-1 T3SS which further adds to our current mechanistic understanding of effector secretion signals.

Bacterial strains, plasmids and growth conditions.
Bacterial strains and plasmids are listed in Table 1. S. Typhimurium and isogenic mutant strains were grown in Luria–Bertani (LB) medium for 12 h at 37 °C, and diluted 1 : 100 into fresh medium with 0.3 M NaCl and 5 mM arabinose. Bacterial cells were grown for 4 h at 37 °C and 200 r.p.m. under conditions that stimulated expression of the SPI-1 T3SS (Bajaj et al., 1996). Strains carrying mutations in fliGHI, invC and invA were constructed using the λ-red system (Datsenko & Wanner, 2000). Escherichia coli strain DH5α was used as a cloning host, while E. coli strain BL21 was used for protein overexpression. E. coli was routinely grown at 37 °C in LB medium. The following antibiotics were used at the indicated concentrations: ampicillin, 100 µg ml^–1; kanamycin, 50 µg ml^–1.

Table 1. Bacterial strains and plasmids used in this study

Plasmid constructions.
All recombinant DNA techniques used were based on standard protocols (Sambrook et al., 1989). The PCR primers (Invitrogen) used are listed in Table 2. The PCRs were carried out using Phusion High-Fidelity DNA polymerase (Finnzymes), and ligations were performed using a T4 DNA ligase (Fermentas).

Table 2. Primer sets used in this study

To identify the NSS of SopD, we constructed a set of vectors expressing variable lengths of the SopD effector using oligonucleotide primers listed in Table 2. A set of DF primers was used to amplify the full length of the SopD-coding region. The resulting product was digested with EcoRI/HindIII and cloned into EcoRI/HindIII-digested pJBT, giving pJWDF, which expressed the SopD protein fused with a C-terminal strep-tag under the control of the inducible arabinose promoter of pBAD24. A first 20 aa N-terminal deletion (SopD_Δ1-20) was created using primer set D20317, following ligation of the EcoRI/HindIII PCR fragments into EcoRI/HindIII-digested pJBT, resulting in pJWD21317. To create a series of N-terminal truncations of SopD deleting amino acids 6–20, 11–20, 16–20 and 6–10, the pJWDF plasmid was also used as a template in the inverse PCRs performed with a set of D620, D1120, D1620 or D610 primers to generate pJWD620, pJWD1120, pJWD1620 or pJWD610, respectively.

To investigate the role of the C terminus of SopD on its secretion through the T3SS, PCR fragments containing 200 and 305 N-terminal amino acids of SopD were cloned using EcoRI/HindIII digestion into similarly digested pJBT, generating pJWD200 and pJWD305, respectively. To construct pJWD201220 and pJWD268302, pJWDF containing the full-length SopD was used as a template for the inverse PCRs using D201220 and D268302 primer sets, respectively. All plasmids encoding recombinant versions of SopD were verified by DNA sequencing (GATC Biotech AG).

Plasmid pGWC, encoding glutathione S-transferase (GST)–InvC, was constructed by inserting and ligating a 1300 bp, GC primers-amplified PCR fragment encoding full-length InvC digested with XbaI and HindIII into similarly cut vector pGEX-2K. The sequence of invC in pGWC was verified by DNA sequencing (GATC-Biotech AG).

Protein secretion analysis.
To assess protein secretion, Salmonella strains harbouring various expression constructs were grown as described above. Whole cells and culture supernatants were separated by centrifugation at 13 000 g for 10 min. The culture supernatants were filter-sterilized (0.22 µm pore-size), and proteins were precipitated with 10 %, v/v, TCA and acetone (Jiang et al., 2004a). Whole-cell and precipitated culture supernatant samples were electrophoresed on 12 % SDS-PAGE gels and analysed by Western blotting.

Protein–protein interaction analysis.
Bacterial cultures containing pGWC (GST–InvC) were grown in LB medium at 37 °C to OD₆₀₀ ∼0.4, supplemented with 0.1 mM IPTG and incubated at 37 °C for an additional 4 h. Bacteria were harvested by centrifugation (3600 r.p.m., 10 min, 4 °C). The bacterial pellet was resuspended in lysis buffer [PBS, 1 % Triton X-100, 10 mM DTT, 2 mM EDTA, one tablet of Protease Inhibitor Cocktail (Roche)], and the cells were lysed by sonication. Cell debris was removed by centrifugation. The supernatant was passed through a 0.45 µm pore-size filter.

Glutathione–Agarose gel lyophilized powder (Sigma) was swollen in PBS for 30 min at room temperature. After swelling, the agarose beads were placed into a SigmaPrep spin column (Sigma). The bacterial cleared lysate from strains overexpressing GST–InvC was bound with swelling resin. Glutathione agarose beads coated with GST–InvC were mixed with cleared extracts of E. coli BL21 strains harbouring pJWDF (SopD–strep-tag), pJWD21317 (SopD_Δ1-20–strep-tag), pJWD200 (SopD_Δ201-317–strep-tag), pJWD201220 (SopD_Δ201-220–strep-tag) or pJWD268302 (SopD_Δ268-302–strep-tag) and incubated at 4 °C for 12 h. The columns were washed three times with PBS, 0.1 % Tween 20 (PBS-T). Elution was performed with 10 mM glutathione. Eluted proteins were analysed by SDS-PAGE and immunoblotting.

Western blotting analysis.
After SDS-PAGE, proteins were transferred onto nitrocellulose transfer membranes using a Mini Trans-Blot cell (Bio-Rad). The transferred membrane was incubated with a mouse strep-tag antibody (IBA) overnight at 4 °C. After washing three times with PBS-T, the membrane was incubated with a goat anti-mouse horseradish peroxidase-labelled secondary antibody (Sigma) at room temperature for 1 h. The membrane was washed three times with PBS-T, and detection was carried out using the EZ-ECL Chemiluminescence Detection kit (Geneflow) according to the manufacturer's instructions. The chemiluminescent signal was detected using Kodak BioMaX light film (Sigma-Aldrich). When needed, the blotting membranes were also probed with a custom-made rabbit polyclonal anti-LuxS antibody followed by a goat anti-rabbit horseradish peroxidase-labelled secondary antibody (Sigma) to account for bacterial lysis or non-specific leakage.

The N-terminal region is essential for T3S of SopD
To investigate the role of the N-terminal region of SopD in T3S, we constructed full-length or truncated versions of SopD fused with a C-terminal strep-tag under the control of the inducible arabinose promoter (P_BAD) (Karavolos et al., 2005). Following induction with arabinose, secreted proteins were analysed by SDS-PAGE and Western blotting using a specific strep-tag antibody. Secretion of full-length and truncated SopD was investigated in the parent (SL1344), a T3S-defective (ΔprgH) strain and a flagellar secretion-defective (fliGHI) strain to confirm secretion specificity through the respective T3S machineries. The fliGHI deletion was used to minimize possible regulatory effects on SPI-1 T3SS expression reported to occur upon disruption of flhDC (Eichelberg & Galán, 2000).

Full-length SopD was secreted into the supernatant of the SL1344 parent and fliGHI mutant strains, but was not detected in the secreted protein fraction of the isogenic ΔprgH, indicating SPI-1 T3SS-dependent secretion of the SopD strep-tag hybrid protein (Fig. 1a). Several studies have concluded that the first ∼20 amino acid residues of the T3S effector proteins contribute signals for their routeing into the secretory pathway. A truncated version of SopD lacking the N-terminal 20 aa (SopD_Δ1-20, pJWD21317) was not detected in the supernatant of the parent strain (Fig. 1b). To further delineate the N-terminal signal of SopD, a series of SopD amino acid truncations in residues 6–10, 11–20, 16–20 and 6–20 were constructed to create SopD_Δ6-10, SopD_Δ11-20, SopD_Δ16-20 and SopD_Δ6-20, respectively (Table 1). SopD_Δ11-20 and SopD_Δ16-20 were successfully secreted and detected in parent strain culture supernatant in an SPI-1 T3S-dependent manner (Fig. 1c, d). Strikingly, the truncated version of SopD lacking amino acids 6–10 (SopD_Δ6-10) failed to be secreted into the culture supernatant (Fig. 1e).

(32K): Fig. 1. Identification of the N-terminal signal for T3S of SopD. (a) Expression and secretion of full-length SopD (SopD1-317) was assessed in SL1344 wild-type, SL1344ΔprgH and SL1344fliGHI : : KanR. Full-length SopD secretion is SPI-1-mediated, since there was no secretion in SL1344ΔprgH. (b) Assessment of the role of regional N-terminal truncations of SopD (SopDΔ1-20, SopDΔ6-20, SopDΔ11-20 and SopDΔ16-20) on its secretion in SL1344 wild-type. (c) Secretion of SopDΔ11-20 is SPI-1-mediated, since there was no secretion in SL1344ΔprgH. (d) Secretion of SopDΔ16-20 is SPI-1-mediated, since there was no secretion in SL1344ΔprgH. (e) Assessment of the expression and secretion of SopDΔ6-10 in SL1344 wild-type indicates that amino acids 6–10 are essential for SPI-1 T3S of SopD. Experiments were repeated at least three times and a representative is shown.

Secretion of a C-terminal truncation of SopD is SPI-1 T3SS-independent
Although several effectors contain a CBD within the first ∼140 amino acids, a CBD for SopD has not yet been identified. We investigated the ability of the N-terminal 200 amino acids of SopD to direct T3S in the parent (SL1344), a T3S-defective (ΔprgH) strain and a flagellar secretion-defective (fliGHI : : Kan^R) strain. SopD_Δ201-317 (pJWD200) was detected in culture supernatants from the parent and surprisingly the ΔprgH strain but was absent from supernatants of the flagellar secretion-defective strain (fliGHI : : Kan^R) (Fig. 2a). This observation suggests that secretion of SopD_Δ201-317 is an SPI-1 T3SS-independent event mediated via erroneous secretion through the flagellar apparatus (Fig. 2a). Furthermore, SopD_Δ201-317 secretion in a panel of Salmonella strains lacking SPI-1 T3SS function (invC, invG, invA and spi-1) was similar to that of the parent SL1344 (Fig. 2b). To rule out cell leakage, we also probed for the intracellular enzyme LuxS, which was undetectable in the same supernatant fractions (Fig. 2b).

(38K): Fig. 2. Role of the C-terminal domain of SopD in T3S. (a) Evaluation of expression and secretion of a C-terminally truncated version of SopD (SopDΔ201-317) in SL1344 wild-type, SL1344ΔprgH and SL1344fliGHI : : KanR shows diminished secretion in SL1344fliGHI : : KanR lacking a functional flagellar secretion apparatus. (b) Expression and secretion of a truncated version of SopD lacking its C-terminal domain (SopDΔ201-317) was unaffected in various SPI-1 secretion-deficient strains (invC : : KanR, invG : : TnphoA, invA : : KanR and spi-1 : : KanR). The control protein LuxS was not secreted in culture supernatants. Experiments were repeated at least three times and a representative is shown.

The C-terminal α-helical regions are important for SPI-1 T3S of SopD
Earlier studies have demonstrated that α-helical secondary structures are associated with protein–protein interactions, including those involved in T3S (Delahay & Frankel, 2002). A prediction of the secondary structure of the SopD C terminus (amino acids 200–317) using ANTHEPROT (Deleage et al., 2001; Gibrat et al., 1987) revealed that it contains two putative long α-helical regions situated between residues 200–218 (helix I) and 268–302 (helix II) (Fig. 3a). Indeed, the C terminus of the effector protein SipB includes an amphipathic α-helix, which is required for secretion through the SPI-1 T3SS (Kim et al., 2007).

(40K): Fig. 3. Identification of C-terminal signals directing T3S of SopD. (a) Schematic diagram of SopD constructs used in this study. Numbers indicate positions of amino acids deleted. Broken lines indicate deleted regions representing helices I and II. Black boxes represent strep-tag-coding regions. Full-length SopD is 317 aa. Secretion via SPI-1 or flagellar T3S is indicated by a plus (secretion) or a minus (no secretion). (b) Deletion of the last C-terminal 12 aa of SopD (SopDΔ306-317) has no effect on its expression and secretion in SL1344 wild-type and SL1344fliGHI : : KanR but leads to diminished secretion in SL1344ΔprgH, indicating SPI-1-dependent T3S of SopDΔ306-317. (c) Expression of SopD lacking helix I (SopDΔ201-220) in SL1344 wild-type, SL1344ΔprgH and SL1344fliGHI : : KanR shows no secretion. (d) Expression of SopD lacking helix II (SopDΔ268-302) in SL1344 wild-type, SL1344ΔprgH and SL1344fliGHI : : KanR shows slight secretion in the strain lacking flagellar apparatus. Experiments were repeated at least three times and a representative is shown.

Initially we constructed a C-terminal 12 amino acid deletion of SopD, leaving intact the two putative helical domains (Fig. 3a, SopD_Δ306-317). Secretion of SopD_Δ306-317 into the culture supernatant was similar to that of full-length SopD, demonstrating that the last 12 residues on the C-terminal end of SopD do not affect its secretion through SPI-1 T3S (Fig. 3b).

We hypothesized that the putative helical structural motifs in the C-terminal regions of SopD physically associate with the secretion machinery during T3S. To investigate the importance of the putative helical regions we designed SopD versions lacking helix I (residues 200–220), helix II (residues 268–302) or both helices (Fig. 3a). Secretion was assessed in the parent (SL1344), a T3S-defective (ΔprgH) strain and a flagellar secretion-defective (fliGHI : : Kan^R) strain. Deletion of helix I (SopD_Δ201-220) led to no secretion (Fig. 3c). However, deletion of helix II (SopD_Δ268-302) led to marginal secretion in the flagellar secretion-defective strain (fliGHI : : Kan^R) (Fig. 3d).

SopD interacts with InvC during SPI-1 T3S
InvC is a class AAA ATPase, forming a hexameric ring on the inner membrane, and is the SPI-1 T3SS energizer. Class AAA ATPases utilize the energy released from ATP to unfold proteins and pass them through a channel at the centre of their ring (Sauer et al., 2004). Akeda & Galán (2005) demonstrated that InvC binds the chaperone–substrate complex. Although InvC binds to SPI-1 T3S substrates indirectly, a recent study has shown that the N terminus of MxiC, a Shigella flexneri T3SS substrate, interacts with the ATPase Spa47 (Botteaux et al., 2009). Direct binding between a T3S ATPase and its substrate can therefore be an alternative mechanism for T3S of effectors.

We reasoned that SopD T3S may also involve direct contact with the SPI-1 T3SS ATPase, InvC. To verify our hypothesis, we generated a GST–InvC fusion protein (pGWC, Table 1). Cytosolic extracts from an E. coli BL21 strain overexpressing full-length SopD with a C-terminal strep-tag (pJWDF) were tested for the ability to bind GST-immobilized InvC. Elutions from GST–InvC-containing columns contained full-length SopD, while elutions from GST-only columns contained no detectable SopD, indicating a possible interaction between SopD and the SPI-1 T3SS ATPase, InvC (Fig. 4a).

(32K): Fig. 4. Helix II is important for SopD interaction with InvC. (a) Eluted fractions from GST- and GST–InvC-coated beads after incubation with the cleared lysate of a strain expressing full-length SopD–strep-tag were analysed by SDS-PAGE and immunoblotting as described in Methods. (b) The C-terminal part of SopD is important for its interaction with InvC. Eluted fractions from GST- and GST–InvC-coated beads after incubation with the cleared lysate of strains expressing strep-tagged SopD, SopDΔ201-317 or SopDΔ1-20 were analysed by SDS-PAGE and immunoblotting as described in Methods. Respective protein bands are highlighted by arrows. Experiments were repeated at least three times and a representative is shown. (c) Helix II of SopD is essential for its interaction with InvC. Eluted fractions from GST–InvC-coated beads after incubation with the cleared lysate of strains expressing tagged versions of SopD lacking helix I (SopDΔ201-220; SopDΔhelixI) or helix II (SopDΔ268-302; SopDΔhelixII) were analysed by SDS-PAGE and immunoblotting as described in Methods. Experiments were repeated at least three times and a representative is shown.

To further scrutinize the interaction with InvC, cytosolic extracts from an E. coli BL21 strain overexpressing strep-tagged SopD_Δ1-20 or SopD_Δ201-317 were tested for the ability to bind GST-immobilized InvC. Only SopD_Δ1-20 was present in elutions from GST–InvC-containing columns. The inability of SopD_Δ201-317 to bind the column suggests that the C-terminal amino acids 201–317 of SopD are essential for the interaction with InvC (Fig. 4b).

In view of the role of the C-terminal region of SopD in interacting with InvC, we proceeded to determine the role of the two C-terminal helices of SopD. Cytosolic extracts from an E. coli BL21 strain overexpressing helix I-deleted strep-tagged SopD (SopD_Δ200-220) or helix II-deleted strep-tagged SopD (SopD_Δ268-302) were tested for the ability to bind GST-immobilized InvC. Elutions from GST–InvC-containing columns indicated the presence of SopD_{Δhelix I} but not SopD_{Δhelix II}, suggesting that the helix II region (residues 268–302) of SopD is essential for the interaction with InvC (Fig. 4c).

The signals that target bacterial effector proteins into host cells through T3SS injectisomes remain poorly understood. Earlier investigations have identified T3S-related domains in several effectors (Mota et al., 2005; Sory et al., 1995; Tree et al., 2009; Wang et al., 2008). These include the first ∼20 amino acids and also an additional region located within the first ∼140 amino acid which binds the specific cognate chaperone (Ghosh, 2004). For example, Russmann et al. (2002) found that amino acid residues 4–7 of the SPI-1 T3SS substrate InvJ mediate its secretion in a T3SS-dependent manner. In addition, amino acid residues 3–8 of SipB also function as an NSS, routeing this effector for secretion through the T3S pathway (Kim et al., 2007).

We examined the importance of the N-terminal region of the Salmonella effector protein SopD in its secretion through the T3S machinery using a set of deletions. We show that SopD requires an NSS that encompasses N-terminal residues 6–10 for successful secretion through the SPI-1 T3SS (Fig. 1e). The reduced cytoplasmic levels of SopD constructs missing the N-terminal amino acids 6–10 and 6–20 may reflect their reduced stability in the cytoplasm after arabinose induction. In addition, we observed reduced secretion of SopD_Δ11-20, which may be attributable to reduced secretion efficiency. Our data also point out an additional control level in the targeting of SopD towards the correct secretion apparatus. Indeed, the SopD NSS in combination with a C-terminal domain (residues 200–317) is needed to prevent secretion of SopD through the flagellar T3SS (Fig. 2a). Erroneous secretion through the flagellar T3S has also been observed in SptP and SopE upon removal of the CBD (Lee & Galán, 2004).

To facilitate SPI-1 T3SS secretion, the effector protein SptP in complex with its chaperone SicP is recognized by the membrane-associated protein InvC to facilitate secretion of the effector and simultaneous dissociation of the chaperone (Akeda & Galán, 2005). To date, the cognate chaperone and CBD of SopD have not been identified. Previous data from size-exclusion chromatography indicate that SopD forms a monomer and hence is unlikely to have a specific chaperone (Wood et al., 2004). Using GST-immobilized SopD, we have been unable to detect a specific interaction with proteins from Salmonella cytoplasmic extracts under the conditions used (data not shown).

We have also highlighted the importance of the C-terminal region of SopD and particularly the role of the two putative helices (helix I and helix II). In contrast, an earlier study has shown that the first 202 amino acids of SopD fused to the N-terminal domain of adenylate cyclase, cya, are sufficient for secretion in Salmonella Dublin (Jones et al., 1998). We speculate that the differences observed may be due to the size or nature of the reporter fusion partner (8 amino acid strep-tag versus the bulky, 43 kDa Cya N-terminal region). Also, the resulting hybrid SopD–Cya protein may encode or mimic signals that result in secretion through the SPI-1 or other systems.

Here we show that both helices are required for SPI-1 T3SS secretion of SopD (Fig. 3a). This is independent of the presence of the final C-terminal 12 aa of SopD (Fig. 3b). Deletion of either helix leads to loss of secretion via the SPI-1 T3SS (Fig. 3a, c, d). However, the observation that the deletion of helix II leads to marginal secretion in the flagellar mutant suggests the existence of additional cryptic signals directing secretion of SopD via an alternative system. Remarkably, the absence of only one helix is enough to prevent flagellar (and SPI-1) secretion if the other helix is present (Fig. 3a).

Finally, we demonstrate that SopD associates with the T3S ATPase InvC (Fig. 4a). Remarkably, the interaction involves the C-terminal region of SopD, which comprises two α-helical domains (amino acids 201–220 and 268–302; Figs 3a and 4b). In particular, helix II (amino acids 268–302) is essential for binding to the InvC ATPase (Fig. 4c). In contrast to the C-terminal region interaction in Salmonella, in Yersinia, the InvC ATPase homologue YscN interacts with the N terminus of the effector YopR (Sorg et al., 2006). This may reflect differences in size, domain organization or timing of secretion of the two effectors.

It has been suggested that due to the maximum diameter of the central channel of the T3SS needle apparatus (estimated to be ∼28 Å) it would be necessary to unfold translocating proteins prior to release through the needle (Marlovits et al., 2004). Interaction of InvC with SopD could mediate the unfolding of SopD, leading to ejection through the needle using proton motive force at the expense of ATP (Galán, 2008).

In summary, we have identified important N- and C-terminal regions of the effector protein SopD required for its secretion through SPI-1 T3S. We show that an N-terminal signal consisting of amino acids 6–10 is needed to target SopD to the SPI-1 T3S apparatus. We also reveal the role of two putative helical domains in the C-terminal region of SopD (amino acids 201–302) in preventing erroneous secretion through flagellar T3S. In particular, helix II (amino acids 268–302) is necessary for the interaction with the InvC ATPase and contributes to secretion through the T3S injectisome. The elucidation of signalling motifs leading to the SPI-1 T3SS-dependent secretion of substrates reveals new insights in our understanding of effector protein secretion.

We thank Cathy Lee (Harvard University) for kindly donating the spi-1 deletion strain and Vassilis Koronakis (University of Cambridge) for generously providing the invG : : phoA mutant strain. We also thank Dr Joe Gray (Pinnacle Laboratory, Newcastle University) for help with protein analysis. We are grateful to the Royal Thai Government for a PhD scholarship to R. B. Research in the laboratory of C. M. A. K. has been supported by the UK Medical Research Council and the UK Biotechnology and Biological Sciences Research Council.

Edited by: V. J. Cid