Journal of Bacteriology, December 2003, p . 6950-6967, Vol . 185, No . 23

Due to the presence of two independent signals, analysis of effector protein recognition by TTSS has been complicated . The first signal is located at the N terminus of the effector protein . Some workers have suggested that this signal is located within the first 15 amino acids (aa) of the secreted polypeptide (43), while others have argued that the mRNA sequence at the 5' end of the open reading frame (ORF) represents the secretion signal (1) . This first signal does not depend on accessory proteins designated chaperones (1, 43, 63) . The second signal found in effector proteins is chaperone dependent (6, 70) . It represents the chaperone binding site and is generally located between aa 15 and 70 to 140 of the secreted protein (1, 42, 44, 71, 72) .

The type III secretion chaperones have some common features, although they do not have sequence similarities . They are small acidic proteins with a predicted amphipathic -helix at the C terminus . Chaperones generally bind to the N-terminal regions of secreted proteins (aa 15 to 140) in the bacterial cytoplasm, which results in protection from degradation, prevention of premature interactions, and/or mediation of recognition by the TTSS (3, 4, 7, 9, 46, 64) .

Salmonella enterica subspecies I serovar Typhimurium is a gram-negative enteropathogen which is responsible for a large number of gastrointestinal infections in the human population . Among many other virulence factors, Salmonella serovar Typhimurium encodes two TTSS which are expressed at different stages of the disease (22, 26, 32) . The TTSS encoded in Salmonella pathogenicity island 1 (SPI-1) is required for induction of proinflammatory responses, invasion of intestinal epithelial cells, induction of cell death in macrophages, and elicitation of diarrhea (22, 60, 69) .

So far, 12 Salmonella serovar Typhimurium effector proteins which are transported via the SPI-1 TTSS have been identified (22) . In contrast to the proteins of many other enteric pathogens, only some of the Salmonella effector proteins (Sip/SspA, Sip/SspB, Sip/SspC, SptP, and AvrA) are encoded in the vicinity of the TTSS apparatus . Many additional effector proteins (SopE, SopE2, SopA, SopB/SigD, SopD, SlrP, and SspH1) are encoded elsewhere in the chromosome (2, 31, 37, 48, 66, 67, 73-75) . So far, there is little information about how expression and specific transport of the latter group of effector proteins via the SPI-1 TTSS are controlled .

Chaperones have been described for several SPI-1-encoded effector proteins . The effector proteins Sip/SspB and Sip/SspC and their cognate chaperone SicA (68), Sip/SspA and its chaperone InvB (5), and SptP and its chaperone SicP (21) are all encoded in SPI-1 . In the case of SipB/C-SicA and SptP-SicP the proteins are even encoded in the same operon . Similarly, the effector protein SopB/SigD and its specific chaperone PipC (SigE) are encoded next to each other in SPI-5 (12, 73) . However, it is not clear whether the other effector proteins, most of which are encoded outside of SPI-1, require chaperones and where the chaperones are encoded .

In the case of the effector protein SopE this was especially interesting because SopE is encoded by the temperate P2-like bacteriophage SopE (50) . This phage frequently infects new Salmonella serovar Typhimurium strains, which are normally sopE negative, and thereby introduces SopE as an additional effector protein by lysogenic conversion (50) . Interestingly, SopE does not have an ORF with the properties typical of a TTSS chaperone (C . Pelludat and W.-D . Hardt, unpublished data) . Considering the high frequency of horizontal gene transfer of sopE between different Salmonella strains (31, 49, 50, 57), we wondered how SopE is recognized by the SPI-1 TTSS .

In a pull-down experiment we identified the SPI-1-encoded protein InvB as a SopE binding partner and analyzed the role of InvB in SopE secretion and translocation via the SPI-1 TTSS . The results are discussed below in the context of horizontal transfer of effector proteins between different Salmonella strains and their functional integration into the TTSS .

 
The Salmonella serovar Typhimurium ATCC 14028s derivatives CS401 (5) and PB502 (invB) (5) were from the S . I . Miller lab (University of Washington, Seattle) .

For all functional assays bacteria were grown for 12 h at 37°C in Luria-Bertani (LB) medium supplemented with 0.3 M NaCl, diluted 1:20 into fresh medium, and grown for another 4 h with mild aeration to obtain an optical density at 600 nm (OD₆₀₀) of 0.8 to 0.9 (SPI-1-inducing conditions) . When required, antibiotics were added at the following concentrations: ampicillin, 100 µg/ml; kanamycin, 50 µg/ml; chloramphenicol, 30 µg/ml; and tetracycline, 12 µg/ml .

M566, carrying in-frame deletions of sopB, sipA, sopE, and sopE2, was obtained by sequential allelic exchange in the chromosome of M509 (sopB) (49) by using the suicide vectors pM585, pM608, and pM586 (see below) . The suicide vectors were first integrated into the chromosome by homologous recombination . Allelic exchange was completed by a second round of recombination and selection on sucrose plates (39) .

To construct M568 (invB::aphT), invB was replaced by a terminatorless aphT gene cassette which conferred kanamycin resistance . The suicide vector pM670 (Tet^r) (see below) was introduced into Salmonella serovar Typhimurium SL1344 by conjugation and selection on LB agar plates containing kanamycin, which yielded M568 . Strains M573 (sopB invB::aphT), M574 (sopB sipA sopE sopE2 invB::aphT), and M590 (invG invB::aphT) were constructed by phage P22-mediated transduction of the invB::aphT allele of M568 into M509 (49), M566, and SB161, respectively . M608, carrying chromosomal sopE_M45, was constructed by P22-mediated transduction of the sopE::sopE_M45 allele (Tet^r) of SB875 (31) into M566 (see above) .

The ATCC 14028s derivative M622 was created by P22-mediated transduction of the invC::aphT allele of SB566 (14) into CS401 (5) .

The gene disruptions and deletions were confirmed by Western blot analyses by using polyclonal antisera directed against SopE (50) and by PCR by using the primers and conditions listed in Table 2 .

 
Recombinant DNA techniques. Cloning of DNA fragments was performed by using standard protocols (59) .

Construction of suicide vectors for generating chromosomal in-frame deletions. To construct a suicide vector for deletion of sipA, the sequences located directly upstream (primers 5'-GCGGCCGCACCTGGGGTTGAGTCCTAC and 5'-TCTAGAAGGGGGCTGAGT CCTTACAC; 33 cycles of 92°C for 30 s, 53°C for 30 s, and 68°C for 3 min) and downstream (primers 5'-TCTAGAGGCCCGGCTTACGAGTC and 5'-CCCGGGACACCAAGGCACGAG; 33 cycles of 92°C for 30 s, 55°C for 30 s, and 72°C for 3 min) of the sipA coding sequence were amplified by PCR, and the PCR products were cloned into pCR-BluntII-Topo (Invitrogen), which yielded pM582 and pM583 . The insert of pM582 was cloned into the XbaI sites of pM583, which yielded pM584, and the resulting insert was subcloned into the SmaI/NotI sites of the suicide vector pSB890 (a derivative of pGP704; oriR6K Tet^r sacAB) (W.-D . Hardt and J . E . Galán, unpublished data); this yielded the suicide vector pM585, which was used for deletion of sipA .

To construct a suicide vector for deletion of sopE, we amplified the sequences located directly upstream (primers 5'-CGGGATCCTCTTGGCGCGTAGTCCTTC and 5'-GCTCTAGACACGGTAATGATCCTTTTATATGT; 33 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 2 min) and downstream (primers 5'-GCTCTAGACCCTGAACACTGAAAAACCA and 5'-TTTGCGGCCGCGCACTGGATACGCTGAACGA; 33 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 2 min) of the ORF by PCR and cloned the PCR products into pBluescript SKII+ (Stratagene), which yielded pM593 and pM606 . The insert of pM606 was cloned via XbaI/NotI into pM593, which yielded pM607, and the resulting insert was subcloned into the BamHI/NotI sites of pSB890; this yielded the suicide vector pM608 .

To construct a suicide vector for deletion of sopE2, the sequences located directly upstream (primers 5'-CGGGATCCGCGCAGGCGTTTAGAAGACAGTT and 5'-GCTCTAGAAGTCACGGTAGTTCTCCTTTT; 33 cycles of 94°C for 30 s, 53°C for 30 s, and 72°C for 2 min) and downstream (primers 5'-GCTCTAGAAATGCCTCCTGATGGTAGTAA and 5'-GCGGCCGCGTTCCAGCATCAGCCACTTG; 33 cycles of 94°C for 30 s, 57°C for 30 s, and 72°C for 3.5 min) of the ORF were amplified by PCR, and the PCR products were cloned into pBluescript SKII+ (Stratagene), which yielded pM559 and pM560 . The insert of pM560 was cloned via XbaI/NotI into pM559, which yielded pM581, and the resulting insert was subcloned into the BamHI/NotI sites of pSB890; this yielded the suicide vector pM586 .

To construct a suicide vector for deletion of invB, we amplified the sequences located directly upstream (primers 5'-GCGGCCGCCGCTCTTTCGTCTGGCATTATC and 5'-GCTAGCCAAATGTTGCATAGATCTTTTCCTT; 33 cycles of 94°C for 30 s, 52°C for 30 s, and 72°C for 2 min) and downstream (primers 5'-GCTAGCCCCGGGTCGCTAATGAGATGAAAACACCTC and 5'-TCTAGACGTGGCGTTATCGGTTACTTCA; 33 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 2 min 20 s) of the ORF by PCR and cloned the PCR products into pCR2.1-TOPO (Invitrogen), which yielded pM666 and pM667 . pM667 contained an additional SmaI site, which was in frame with the invB reading frame . The insert of pM666 was cloned into pM667, which yielded pM668 . A modified aminoglycoside 3'-phosphotransferase (aphT) gene which lacks a transcription terminator (25) was cloned into the SmaI site of pM668, which yielded pM669, and the resulting insert was subcloned into the NotI/XbaI sites of pSB890; this yielded the suicide vector pM670 .

Chromosomal DNA of Salmonella serovar Typhimurium SL1344 was used as a template for each PCR . All inserts were verified by DNA sequence analysis .

SopE-GST fusion protein expression vector pM113. The glutathione S-transferase (GST) cassette was amplified by PCR (primers 5'-CCGGGAGCTGCATGTGTCAGAGG and 5'-GGGCTAGCAGAAGAGTGTCTTCACCGCGGCTCGAGAATGTCCCCTATACTAGGTTATTGG; template, pGEX-KG [27]) and was cloned via NheI/HindIII into pBAD24 (28), which yielded pM102 . A DNA fragment containing the Shine-Dalgarno sequence, the ATG start codon, and SopE aa 1 to 78 was amplified by PCR (template, pSB1136; primers 5'-CCGCGGATCCCGGGTGCCCGGCCCTCAGAT and 5'-GTCTAGAGCTCCTGAAGGAATTCTAATGACAAAAAT) and cloned via EcoRI (Klenow treated)/SacII into BbsI (Klenow treated)/SacII sites of pM102, which yielded pM106 . pSB1136 had been constructed by amplification of the SopE coding sequence (template, SL1344 chromosomal DNA; primers 5'-GGTGGAATTCTAATGACAAAAATAACTTTATCTCC and 5'-CCGATCCATGGCACCACCCCCGGGAGTGTTTTGTATATATTTAT) and cloning via EcoRI/NcoI digestion into pSB616, a pBAD24 derivative carrying the M45 epitope tag and kindly provided by J . E . Galán (8) .

A fragment of pSB1136 (containing aa 56 to 240 of SopE) was cloned via BbsI/SmaI into pM106, which yielded pM113, which encoded SopE (aa 1 to 240) with a C-terminal GST fusion under control of the arabinose-inducible promoter of pBAD24 .

Expression vector for the GST-InvB fusion protein. The ORF of invB was amplified by PCR (primers 5'-GGAATTCTAATGCAACATTTGGATATCGCTGA and 5'-CCCAAGCTTACGGCGTATTTCACACAGTTCG; 33 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 2 min 20 s; template, SL1344 chromosomal DNA) and cloned into pGEX-KG (27), which yielded pM672, which expressed GST-InvB under control of the lac promoter .

Expression vectors. pM136 and pM226 are pBAD24 derivatives which express sopE_1-240M45 and sopE2_1-240M45, respectively, under control of their native promoters (approximately 20 copies/cell [66]) . pSB1130 and pM149 are pACYC184 derivatives which express sopE_1-240 and sopE2_1-240, respectively, under control of their native promoters (66) .

A low-copy-number expression vector for sopE_M45 was constructed by subcloning the insert of pM136 (Eco47III/SalI) into the EcoRV and SalI sites of pACYC184 (approximately five copies/cell; NEB), which yielded pM438 .

A low-copy-number expression vector for sopE2_M45 was constructed by digesting pM226 with SpeI (Klenow treated) and HindIII and subcloning the insert into the EcoRV and HindIII sites of pACYC184 (NEB), which yielded pM256 .

A DNA fragment of Salmonella serovar Typhimurium SL1344 harboring 877 bp upstream of the ORF of sopE was amplified by PCR (primers 5'-GCTGCCTGCCACCATACCCAC and 5'-GCCGCTAGCGTACATAATTCATTTATATATAGATAGC; 33 cycles of 94°C for 30 s, 55°C for 30 s, and 72°C for 2 min 20 s; template, pM136) and cloned into pSB1136, which yielded pM542, which expressed sopE_M45 under control of the native promoter and the Shine-Dalgarno sequence of the arabinose-inducible promoter of pBAD24 .

A chromosomal DNA fragment of Salmonella serovar Typhimurium SL1344 harboring the ORF of invB was amplified by PCR (primers 5'-GGAATTCTAATGCAACATTTGGATATCGCTGA and 5'-CCCAAGCTTTCATCTCATTAGCGACCGACTA; 33 cycles of 94°C for 30 s, 54°C for 30 s, and 68°C for 45 s) and cloned into pM542 and pSB1136, which yielded pM249 and pM250, respectively, which expressed invB under control of the sopE promoter and the arabinose-inducible promoter, respectively . The constructs were verified by DNA sequence analysis .

A chromosomal DNA fragment of Salmonella serovar Typhimurium SL1344 harboring the ORF of sipA was amplified by PCR (primers 5'-GGAATTCTAATGGTTACAAGTGTAAGGACTCAGCCC and 5'-TCCCCCCGGGACGCTGCATGTGCAAGCCATC; 33 cycles of 94°C for 30 s, 57°C for 30 s, and 72°C for 3.5 min) and cloned into pM542, which yielded pM578, which expressed sipA_M45 under control of the sopE promoter . The construct was verified by DNA sequence analysis . The insert was subcloned via Eco47III (partial digestion)/HindIII into the EcoRV and HindIII sites of pACYC184, which yielded pM580 .

A DNA fragment of Salmonella serovar Typhimurium SL1344 harboring 870 bp upstream and aa 1 to 95 of the ORF of sopE was amplified by PCR (primers 5'-TCGATACTAGTATGCCCGTTTTCTTACCGTCG and 5'-ATGTCGATATCTATATCATTGAGCGTTTGAAGC; 33 cycles of 96°C for 30 s, 55°C for 45 s, and 72°C for 3 min; template, pM136) and cloned via SpeI/EcoRV into pM226, which yielded pM416, which expressed sopE_1-95 sopE2_96240M45 under control of the sopE promoter .

A DNA fragment of Salmonella serovar Typhimurium SL1344 harboring aa 96 to 240 of the ORF of sopE was amplified by PCR (primers 5'-CGGGCGATATCAGAGGTAGTGCGAGTAAAGACCC and 5'-GCATGGGGTCAGGTGGG; 33 cycles of 96°C for 30 s, 55°C for 45 s, and 72°C for 3 min; template, pM136) and cloned via EcoRV/HindIII into pM226, which yielded pM417, which expressed sopE2_1-95 sopE_96-240M45 under control of the sopE2 promoter .

lacZ transcriptional fusions. A sopE expression vector with a lacZ transcriptional reporter cassette integrated downstream of the sopE stop codon under control of the native sopE promoter was constructed by cloning the lacZ transcriptional reporter cassette of pSB1040 (kindly provided by D . Zhou and J . E . Galán) into the XbaI site of pM136, which yielded pM185 . A DNA fragment of pM185 (carrying the lacZ reporter cassette) was cloned via NcoI/HindIII into pM226, which yielded pM687, a sopE2 expression vector with a lacZ transcriptional reporter cassette under control of the native sopE2 promoter .

Protein preparation and analysis. For analysis of secreted proteins bacteria were grown for 12 h at 37°C in LB medium supplemented with 0.3 M NaCl, diluted 1:20 into fresh medium, and grown for another 4 h with mild aeration to obtain an OD₆₀₀ of 0.8 to 0.9 . Culture supernatants for analysis of secretion were prepared by precipitation with trichloroacetic acid and acetone as described previously (40) . Samples of whole bacterial cultures, bacterial pellets, and culture supernatants were separated on a sodium dodecyl sulfate (SDS) gel and transferred to a nitrocellulose membrane . Proteins were detected by Western blotting by using a monoclonal anti-M45 antibody (52), a polyclonal anti-SopE antiserum (50), a polyclonal anti-SopE2 antiserum (66), polyclonal anti-SipC and anti-SipA antisera (kindly provided by J . E . Galán [41]), a monoclonal anti-OmpC antibody (62), appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies, and a chemoluminescent detection kit, as recommended by the manufacturer (Amersham Pharmacia) .

For analysis of protein stability, bacterial cultures were grown as described above . A 100-µl aliquot was removed after 4 h of subculturing (zero-time control) . Chloramphenicol (30 µg/ml) was added to the culture, and samples were removed after 5, 10, 20, and 40 min, mixed with Laemmli sample buffer, and shock frozen . Proteins were separated on a 15% SDS gel and detected by Western blotting as described above .

Generation of an InvB-specific antiserum. The GST-InvB expression plasmid pM672 was transformed into Escherichia coli, and cultures were grown overnight, diluted 1:50, and grown to an OD₆₀₀ of 0.6 . The cultures were then induced with isopropyl-ß-D-thiogalactopyranoside (IPTG) for 4 h and pelleted . The GST-InvB fusion protein was purified by using glutathione (GSH)-Sepharose beads (Amersham Pharmacia), and the GST tag was removed by cleavage with thrombin . The purified InvB protein was used by Neosystem (Strasbourg, France) to raise a polyclonal rabbit antiserum .

Pull-down assay. Strain M566(pM113) was grown for 12 h at 37°C in LB medium supplemented with 0.3 M NaCl, diluted 1:20 into 100 ml of fresh medium, and grown for another 3 h with mild aeration . Expression of the SopE-GST fusion protein was induced by addition of 0.2% arabinose and growth for another 1 h at 37°C with mild aeration . Bacteria were harvested by centrifugation (6,000 x g, 10 min, 4°C) . The bacterial pellet was resuspended in 25 ml of buffer B (50 mM Tris [pH 7.6], 1 mM MgCl₂, 100 mM NaCl; 4°C) supplemented with 2 mM dithiothreitol, and the cells were lysed in a French pressure cell . Cell debris was removed by centrifugation . The supernatant was passed through a 0.4-µm-pore-size filter . The SopE-GST fusion protein was bound to 200 µl of glutathione-Sepharose 4B beads (Amersham Pharmacia) at 4°C . The beads were washed extensively seven times in 10 ml of buffer B at 4°C, and Laemmli sample buffer was added . The proteins bound to 100 µl of beads were analyzed on a 15% SDS-10% SDS discontinuous gel . The gel was stained with Coomassie brilliant blue . As a control a 5-µl aliquot of beads was analyzed on a 15% SDS-10% SDS discontinuous gel, and the SopE-GST fusion protein was detected by Western blotting with monoclonal anti-GST antibody (Sigma) and polyclonal anti-SopE antiserum as the primary antibodies . A monoclonal HRP-coupled anti-rabbit antibody (Sigma) and a monoclonal HRP-coupled anti-mouse antibody (Sigma) were used as secondary antibodies . Bands of interest were excised from the Coomassie brilliant blue-stained gel . Cysteines were treated with iodoacetamide to form carbamidocysteines, and proteins were digested with trypsin . The protein fragments were analyzed by matrix-assisted laser desorption ionization mass spectrometry, and proteins were identified by a fingerprint analysis by using the PeptIdent program (http://www.expasy.ch) .

To verify the specificity of the pull-down experiment, a GST-InvB fusion protein and all associated proteins were purified from the lysate of wild-type Salmonella serovar Typhimurium strain SL1344(pM672) by glutathione-Sepharose affinity chromatography as described above . Samples were removed at each step of the purification procedure, separated on a 15% SDS-polyacrylamide gel electrophoresis (PAGE) gel, analyzed by Western blotting by using an anti-SopE antibody, and later reprobed by using anti-SipC and anti-SipA antisera (kindly provided by J . E . Galán) .

Coimmunoprecipitation. Strains M566 and M608 were grown for 12 h at 37°C in LB medium containing 0.3 M NaCl, diluted 1:20 into 40 ml of fresh medium, and grown for another 4 h . Bacteria were harvested by centrifugation (6,000 x g, 10 min, 4°C) . Each bacterial pellet was resuspended in 20 ml of buffer B supplemented with 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and a complete protease inhibitor mixture . The cells were lysed in a French pressure cell, and the cell debris was removed by centrifugation . The supernatant was passed through a 0.4-µm-pore-size filter . Five milliliters of the lysate was precleared by incubation with 20 µl of protein A-Sepharose CL-4B beads (Amersham Pharmacia) for 1 h at 4°C . We then transferred the lysate into a new tube without touching the beads, and the lysate was incubated with 10 µl of anti-M45 antibody (1 mg/ml) for 1 h at 4°C . The lysate was centrifuged for 10 min at 10,000 x g in a table top centrifuge to remove nonspecific aggregates . Ten microliters of protein A-Sepharose beads was added to the lysate and incubated for 1 h at 4°C . After this the beads were washed four times in 1 ml of buffer B at 4°C, and Laemmli sample buffer was added . Samples were removed at each step of the immunoprecipitation procedure, separated by 16% SDS-PAGE, and analyzed by Western blotting by using an anti-InvB antiserum . Later the blot was reprobed by using an anti-SopE antiserum . A monoclonal HRP-coupled anti-rabbit antibody (Sigma) was used as the secondary antibody .

Analysis of sopE and sopE2 transcription by lacZ expression. Salmonella serovar Typhimurium strains SL1344, SB161, and M590 harboring pM185 or pM687 were grown under SPI-1-inducing conditions as described above . Samples (100 µl) were removed and analyzed for ß-galactosidase activity by using standard methods (59) . Briefly, 100 µl of culture, 900 µl of buffer Z (60 mM Na₂HPO₄, 40 mM NaH₂PO₄, 10 mM KCl, 1 mM MgSO₄, 50 mM ß-mercaptoethanol), 2 drops of CHCl₃, and 1 drop of 0.1% SDS were mixed with vortexing for 10 s and equilibrated to 28°C . Then 0.2 ml of o-nitrophenyl-ß-D-galactopyranoside (4 mg/ml) was added . The reaction was stopped by adding 0.5 ml of 1 M Na₂CO₃ . The time at which a yellow color developed was recorded . The A₆₀₀ of the culture and the A₅₅₀ and A₄₂₀ of the reaction samples were determined, and the numbers of ß-galactosidase activity units were calculated .

Gentamicin protection assay. The invasiveness of mutant Salmonella strains for COS-7 cells was analyzed as described previously (49) . COS-7 tissue culture cells were grown for 2 days in Dulbecco's modified Eagle's medium (DMEM) containing 5% fetal bovine serum (FBS) in 24-well dishes to obtain 80% confluence . The culture medium was removed, and 500 ml of Hanks' balanced salt solution (HBSS) was added 3 min before addition of the bacteria . The bacteria were grown for 12 h in LB medium supplemented with 0.3 M NaCl, diluted 1:20 into fresh medium, and grown for another 4 h with mild aeration to obtain an OD₆₀₀ of 0.8 to 0.9 . The actual bacterial density was determined by plating appropriate dilutions on LB agar . To start the assay, bacteria were added to COS-7 cells at a multiplicity of infection (MOI) of 15 bacteria/cell and incubated for 50 min at 37°C in the presence of 5% CO₂ . The cells were washed three times with HBSS and incubated in 500 ml of DMEM containing 5% FBS and 400 µg of gentamicin per ml for 2 h at 37°C in the presence of 5% CO₂ . The cells were washed three times with 1x phosphate-buffered saline (PBS) and lysed in 1x PBS containing 0.1% sodium deoxycholate, and the number of intracellular bacteria (CFU) was determined by plating on LB agar . Usually, about 25% of the inoculum of the wild-type strain was recovered from COS-7 cells at the end of the assay . The numbers given below were determined by performing at least five independent experiments for each strain . The statistical significance of differences in invasiveness between different strains was analyzed by using the Mann-Whitney U test .

Immunofluorescence analysis of SipC, SopE_M45, and SopE2_M45 translocation. COS-7 cells were grown in 24-well dishes for 2 days in DMEM containing 5% FBS on glass coverslips to about 60% confluence . The culture medium was removed, and 500 ml of HBSS was added . For analysis of SipC translocation, COS-7 cells were infected for 50 min with bacteria grown as described above (MOI, 30 bacteria/cell) . After this, extracellular bacteria were removed by washing the cells twice with HBSS, and COS-7 cells were incubated in 500 ml of DMEM containing 5% FBS and 400 µg of gentamicin per ml for 1 h at 37°C in the presence of 5% CO₂ . The cells were fixed with 3.7% formaldehyde in PBS and permeabilized with 0.1% Triton X-100 in PBS . The translocated SipC was detected by using an anti-SipC antiserum (kindly provided by J . E . Galán) and an anti-rabbit fluorescein isothiocyanate (FITC) conjugate (1:400 dilution in 3% bovine serum albumin-2% dry milk in PBS; Sigma) . The DNA was stained with 4',6'-diamidino-2-phenylindole (DAPI) (Sigma) . For analysis of translocation of SopE_M45 and SopE2_M45, COS-7 cells were infected with SPI-1-induced Salmonella serovar Typhimurium strains harboring the sopE_M45 (pM438) and sopE2_M45 (pM256) expression vectors (MOI, 60 bacteria/cell), respectively . The translocated M45-tagged proteins were detected by using a monoclonal anti-M45 antibody (kindly provided by P . Hearing) and an anti-mouse FITC conjugate (1:400 dilution in 3% bovine serum albumin-2% dry milk in PBS; Sigma) . Coverslips were mounted and analyzed by fluorescence microscopy . To quantify effector protein translocation, at least 100 cells that were infected (harboring more than five bacteria per cell) were scored in a blinded manner for detectable effector protein-specific staining . The proportion of infected host cells that were positive for each effector protein was determined as follows: (number of infected, effector protein-positive cells)/(total number of infected cells) x 100 .

Images were recorded by using a Perkin-Elmer Ultraview confocal imaging system and a Zeiss Axiovert 200 microscope; green fluorescence was recorded confocally, and the DAPI fluorescence was recorded by epifluorescence microscopy .

 
After expression of SopE-GST in M566, the cleared lysate was incubated with GSH-Sepharose beads (see Materials and Methods) . Bound proteins were analyzed by discontinuous PAGE . Eight distinct protein bands were purified from the M566(pM113) lysate (Fig . 2) . Western blot analysis performed with a polyclonal rabbit anti-SopE antiserum and a monoclonal anti-GST antibody suggested that bands 1, 3, and 6 correspond to the full-length SopE-GST fusion protein (55 kDa) and fragments of this protein (Fig . 2) . Bands 2, 4, 5, 7, and 8 were excised from the gel, digested with trypsin, and analyzed by matrix-assisted laser desorption ionization mass spectrometry . The fingerprint analysis revealed that bands 5 and 7 were also fragments of the SopE-GST fusion protein . Bands 2 and 4 corresponded to glycerol-3-phosphate dehydrogenase (11 matching peptides, 24.5% of the sequence covered) and the DNA polymerase III ß-chain (eight matching peptides, 29.2% of the sequence covered) and were probably copurified due to nonspecific binding . Band 8 corresponded to the Salmonella serovar Typhimurium protein InvB (five matching peptides, 57.8% of the sequence covered) (14) . InvB is encoded in SPI-1 (23) and has been described as a secretion chaperone for the SPI-1 effector protein Sip/SspA (5) . This suggested that InvB might also be a type III secretion chaperone for SopE .

 
GST-InvB affinity purification of SopE and SipA but not SipC. To confirm that SopE can indeed be copurified specifically with InvB, we repeated the affinity purification assay using a GST-InvB expression vector (pM672) (see Materials and Methods) . Salmonella serovar Typhimurium SL1344(pM672) was grown and a lysate was prepared as described in Materials and Methods . GST-InvB and the bound bacterial proteins were purified by using GSH-Sepharose beads . A Western blot analysis demonstrated that the affinity purification procedure removed SopE and SipA but not SipC from the cleared cell lysate (Fig . 3A, compare lanes e and f) . The remaining amounts of SopE and SipA detected in the depleted lysate (lane f) might be attributable to the presence of some misfolded SopE or SipA fraction or to SopE/SipA binding to the chromosomally encoded wild-type InvB . On the washed glutathione-Sepharose beads we detected large amounts of SopE and SipA which were copurified with GST-InvB (Fig . 3A, lane h) but no SipC, which is known to be secreted independently of InvB (5) . These data confirmed that InvB bound directly or indirectly to SipA (5) and SopE (this study) . This protein formed a specific complex with SopE and SipA but not with just any protein transported via the SPI-1 TTSS .

 
Coimmunoprecipitation experiments confirmed the SopE-InvB interaction. The pull-down experiments (Fig . 2 and 3) were performed with strains overexpressing large quantities of GST fusion proteins . Therefore, altered protein concentrations might have affected complex formation . In order to eliminate this possibility, we wanted to confirm the SopE-InvB interaction under more physiological conditions . To do this, we performed a coimmunoprecipitation assay using the SL1344 derivative M608, which encoded an M45 epitope-tagged version of SopE (SopE_M45) and native InvB in the chromosome (Table 1) (see Materials and Methods) . This strain was grown under SPI-1-inducing conditions, and the cleared bacterial lysate was incubated with an anti-M45 monoclonal antibody and protein A-Sepharose beads . The beads were washed extensively, and the material bound to the beads was analyzed by Western blotting (Fig . 3B, top panel) . Analysis with polyclonal rabbit anti-InvB and rabbit anti-SopE antisera revealed that InvB coprecipitated with SopE_M45 (Fig . 3B, top panel, lane i) . In contrast, an unknown 18-kDa cytosolic polypeptide that cross-reacted with the anti-InvB antiserum was not coprecipitated . The specificity of the coimmunoprecipitation was verified by a parallel control experiment in which a cleared lysate of M566 was used . This strain was identical to M608 except that it lacked SopE_M45 (Table 1 and Fig . 3B, bottom panel) . In this control experiment, InvB was not coprecipitated with the anti-M45 monoclonal antibody (Fig . 3B, compare anti-InvB signals in lane i from M608 and M566) . In conclusion, the data confirmed that SopE bound directly or indirectly to InvB in the cytosol of Salmonella serovar Typhimurium .

Construction and initial characterization of an invB mutant. To analyze the role of InvB in SopE secretion, we constructed the isogenic mutant strain M568 (invB::aphT) carrying a terminatorless aphT cassette instead of invB (SL1344 [invB::aphT]) (see Materials and Methods) . Because invB is located in an operon encoding several essential subunits of the SPI-1 type III secretion apparatus, it was crucial to verify that insertion of the aphT cassette did not interfere with the general function of the SPI-1 TTSS . To do this, we analyzed secretion and translocation of the effector protein SipC, which is transported via the SPI-1 TTSS in an InvB-independent manner (5) . Western blot analysis performed with a polyclonal anti-SipC antiserum confirmed that expression and secretion of SipC in wild-type Salmonella serovar Typhimurium SL1344 were virtually identical to expression and secretion of SipC in M568 (invB::aphT) . The small reduction in SipC secretion in the invB mutant M568 (Fig . 4A, compare the lanes for the wild-type and invB::aphT strains) might be attributable to a slight polar effect of the aphT cassette on downstream genes encoding essential components of the SPI-1 type III apparatus . Alternatively, InvB might have some minor regulatory function . Nevertheless, SipC secretion by a secretion-deficient Salmonella serovar Typhimurium mutant (SB566 [invC::aphT]) (Fig . 4A) was affected much more severely than SipC secretion by M568 (invB::aphT) . Overall, our results confirmed that the invB::aphT mutation did not interfere with the general function of the SPI-1 TTSS .

 
These results were supported by an analysis of the translocation of SipC by M568 into host cells (Fig . 4B) . For this analysis, COS-7 tissue culture cells were infected with wild-type Salmonella serovar Typhimurium, M568 (invB::aphT), the complemented strain M568(pM249), and the secretion-deficient mutant SB566 (invC::aphT [see Materials and Methods]) (MOI, 30 bacteria/cell) . The cells were fixed 2 h postinfection, and the translocated SipC was stained by using an anti-SipC antibody (Fig . 4B) (41) . Our results show that M568 (invB::aphT) translocates SipC into infected host cells at wild-type levels . In conclusion, the data demonstrate that disruption of invB by the aphT cassette does not compromise the general function of the SPI-1 TTSS . Therefore, this strain could be used to study the specific role of InvB in the secretion and translocation of SopE .

Role of invB in secretion of SopE. We compared secretion of SopE into the bacterial culture supernatant by M568 (invB::aphT), the secretion-deficient mutant SB566 (invC::aphT), and the isogenic wild-type strain SL1344 . This was done by reanalyzing the samples shown in Fig . 4A in a Western blot by using a polyclonal anti-SopE antibody . We found that M568 (invB::aphT) and SB566 (invC::aphT) do not secrete the chromosomally encoded SopE into the culture supernatant (Fig . 4A) . SopE secretion by M568 (invB::aphT) could be partially complemented by transformation with the invB expression vector pM249 (Fig . 4A) . The partial complementation might be attributable to a minor polar effect of the aphT cassette . Nevertheless, the data in Fig . 4A support the notion that InvB is required for secretion of SopE via the SPI-1 TTSS .

The anti-SopE antiserum used in the initial experiments (Fig . 4A) showed some cross-reactivity with the effector protein SopE2 (66) . In order to develop a more specific assay for SopE, we transformed Salmonella serovar Typhimurium SL1344 (wild type), SB161 (invG), and M568 (invB::aphT) with pM136, an expression vector for an M45 epitope-tagged version of SopE (66) . M568(pM136) and the secretion-deficient mutant SB161(pM136) were incapable of SopE_M45 secretion, whereas SopE_M45 was secreted by wild-type strain SL1344(pM136) (Fig . 5A) .

 
The data show that SopE and SopE_M45 are not secreted by M568 (invB::aphT) and mutants with a defective SPI-1 TTSS apparatus (SB566 [invC::aphT] and SB161 [invG], respectively) . Therefore, all of the SopE/SopE_M45 present in cultures or bacterial pellets of M568, SB566, and SB161 is attributable to intracellular protein pools (Fig . 4A and 5A) . No differences were detected between the intracellular pools of SopE_M45 in M568 (invB::aphT) and SB566 (invC::aphT) (Fig . 4A and 5A) . Therefore, invB is required for secretion but not stabilization of intracellular SopE (see below) .

Since InvB has been described as a secretion chaperone for the effector protein Sip/SspA (5), we analyzed the secretion of SipA_M45 in the invB mutant M568 as a control . In line with the results of Bronstein et al . (5), we found that the invB mutant M568 secreted less than 5% of the wild-type levels of SipA_M45 (Fig . 5B) . The secretion defect could be partially complemented with the invB expression vector pM249 . Again, incomplete complementation of M568 by pM249 might be attributable to a minor polar effect of the invB::aphT mutation . Nevertheless, our data support the results of Bronstein et al . (5) which demonstrated that InvB is a secretion chaperone for SipA .

Role of invB in the translocation of SopE. The role of invB in the SPI-1-dependent translocation of SopE into host cells was analyzed by immunofluorescence microscopy . To do this, COS-7 tissue culture cells were infected with wild-type Salmonella serovar Typhimurium SL1344, the secretion-deficient mutant SB161 (invG), and the invB mutant strain M568 harboring the sopE_M45 expression plasmid pM438 . The translocated effector proteins SopE_M45 and SipC were detected by using the anti-M45 antibody (Fig . 6A) and an anti-SipC antiserum (Fig . 6B) . SopE_M45 could be detected in 92% of all cells infected with SL1344(pM438) but not in cells infected with M568(pM438) or SB161(pM438) . The translocation defect of the invB mutant strain M568(pM438) could be restored (85% of all infected cells were SopE_M45 positive) with an invB expression vector in M568 (pM438 and pM249) . Together with the data shown in Fig . 4 and 5A, these results demonstrate that invB plays a specific role in SPI-1-dependent transport of SopE but not in transport of SipC .

 
SopE2 secretion also depends on invB. The Salmonella serovar Typhimurium effector protein SopE2 is 69% identical to SopE at the amino acid level (2, 66) . SopE2 and SopE act as guanine nucleotide exchange factors for small Rho GTPases of the host cell and are transported via the SPI-1 TTSS (20, 30, 66) . Based on these similarities, we speculated that SopE2 also requires invB for secretion . To investigate this, we transformed pM256 (sopE2 promoter; SopE2 tagged C terminally with the M45 epitope, Cm^r) into wild-type Salmonella serovar Typhimurium SL1344, SB566 (invC::aphT), M568 (invB::aphT), and M568 carrying the invB expression vector pM249 (Amp^r) . Secretion of SopE2_M45 from these strains was analyzed by Western blotting (Fig . 7A) . SopE2_M45 was detected in the culture supernatants of wild-type Salmonella serovar Typhimurium SL1344(pM256) and the invB-complemented strain M568(pM256, pM249) but not in supernatants from the secretion-deficient strain SB566(pM256) or the invB mutant M568(pM256) (Fig . 7A) . Again, the complementation was incomplete, which might be attributable to a slight polar effect of the invB::aphT mutation . Nevertheless, it was striking that the intracellular SopE2_M45 pools were much smaller in M568(pM256) than in SB566(pM256) . This suggested that invB affects the cytosolic stability or expression of SopE2_M45 . However, the low cytoplasmic levels of SopE2_M45 in M568(pM256) did not allow us to draw conclusions about a potential role of invB in the transport of this effector protein via the SPI-1 TTSS .

 
In order to analyze the role of invB in SopE2_M45 secretion, the amount of cytosolic SopE2_M45 was increased by using a pBAD derivative sopE2_M45 expression plasmid (pM226), which has a higher copy number than the pACYC184 derivative pM256 . pM226 was transformed into wild-type Salmonella serovar Typhimurium SL1344, SB161 (invG), and M568 (invB::aphT) . Analysis of SopE2_M45 secretion by Western blotting performed with the monoclonal anti-M45 antibody revealed that the cytoplasmic levels of SopE2_M45 were higher and easily detectable even in M568 when we used pM226 (Fig . 7B) instead of pM256 (Fig . 7A) . However, we still could not detect SopE2_M45 in the culture supernatant of M568(pM226) (Fig . 7B) . These results indicated that invB plays a role not only in the stabilization of intracellular SopE2_M45 but also in transport of SopE2_M45 via the SPI-1 TTSS .

Furthermore, we studied the InvB dependence of SopE2 secretion in Salmonella serovar Typhimurium ATCC 14028s . ATCC 14028s carries the sopE2 gene but lacks sopE . Using a polyclonal anti-SopE2 antiserum, we detected SopE2 in the culture supernatant of the wild-type ATCC 14028s strain but not in the supernatants of isogenic derivatives with a disrupted SPI-1 TTSS apparatus (invC::aphT) or a disrupted invB gene (invB) (Fig . 7C) . SopE2 secretion by the latter strain could be complemented to wild-type levels by transformation with the invB expression vector pM250 (Fig . 7C and Table 1) . The analysis of this strain verified that SipC secretion does not require invB (5) (see above) and that intracellular SopE2 levels are severely reduced in invB mutants (Fig . 7) . Taken together, these results indicated that InvB affects intracellular SopE2 levels and transport of this effector protein via the SPI-1 TTSS .

Differential effect of invB on the cytosolic stability of SopE and SopE2. The data presented above showed that the cytosolic pool of SopE2_M45 was smaller in M568 (invB::aphT) than in Salmonella serovar Typhimurium mutants with a disrupted SPI-1 TTSS apparatus (SB161 and SB566) (Fig . 7A and B) . This effect was not observed with the chromosomally encoded SopE protein or the plasmid-encoded SopE_M45 protein (Fig . 4A and 5A) . This indicated that the invB mutation might have a specific effect on the cytosolic SopE2_M45 level .

Chaperones can control cytosolic effector protein levels by three different mechanisms: (i) facilitation of secretion or translocation into host cells via the TTSS (9); (ii) modulation of gene expression (11); and (iii) binding and protection of the effector protein from degradation and aggregation in the cytoplasm (47) . To explain the reduced intracellular levels of SopE2_M45 in M568 (invB::aphT) compared to the levels in SB161 and SB566, we explored the latter two possibilities .

First, we analyzed the effect of invB on effector protein stability . In order to measure protein degradation of cytosolic effector proteins, it is important to examine protein stability in secretion-deficient strains and to compare the results . Therefore, we compared the stabilities of wild-type SopE, SopE_M45 (pM136), and SopE2_M45 (pM226) in SB161 (invG) and the isogenic invB mutant M590 (invG invB::aphT) . The strains were grown under SPI-1-inducing conditions (see Materials and Methods), and de novo protein biosynthesis was stopped by addition of chloramphenicol . Aliquots were removed at different times, and degradation of cytosolic SopE, SopE_M45, or SopE2_M45 was analyzed by Western blotting (Fig . 8A to C) . In the invB mutant M590 (invG invB::aphT) SopE2_M45 was almost completely degraded after 20 min (Fig . 8C) . However, in the presence of invB (SB161 [invG]), SopE2_M45 was only partially degraded during the course of the assay (40 min) . SopE_M45 was much more stable than SopE2_M45 in both Salmonella strains (Fig . 8B and C) . It should be noted that the stability of the chromosomally encoded wild-type SopE was somewhat lower than that of the plasmid-encoded SopE_M45 in both SB161 (invG) and M590 (invG invB::aphT) (Fig . 8A) . However, even in this case the effect of invB on the stability of the cytoplasmic effector protein pool was much less pronounced than the effect observed with SopE2_M45 .

 
In the next analysis, we studied the role of invB in the transcription of sopE_M45 and sopE2_M45 . To do this, we inserted a lacZ transcriptional reporter gene cassette downstream of the sopE_M45 ORF in pM136 (pM185) and downstream of the sopE2_M45 ORF in pM226 (pM687) (Table 1) . These plasmids were transformed into wild-type Salmonella serovar Typhimurium SL1344, SB161, and M590 . The bacteria were grown under identical conditions as described above . We found that both plasmids yielded identical levels of lacZ expression in all three Salmonella serovar Typhimurium strains (Fig . 9) (see Materials and Methods) . Therefore, transcription of sopE2_M45 and sopE_M45 was not affected by disruption of invB or the SPI-1 type III secretion apparatus .

 
In conclusion, the data show that invB does not affect transcription of sopE_M45 and sopE2_M45 . Rather, invB seems to affect cytosolic effector protein pools by stabilizing cytosolic SopE2 (Fig . 8C) and by mediating secretion of SopE2 and SopE via the SPI-1 TTSS (Fig . 4A, 5A, and 7A and B) .

N-terminal 95 aa determine different cytosolic stabilities of SopE_M45 and SopE2_M45. Considering the high levels of functional and structural similarities, it was interesting to observe that SopE2_M45 is less stable than SopE_M45 in the cytosol of SB161 (invG) and M590 (invG invB::aphT) . We wondered whether this difference is attributable to sequence variations in the N-terminal regions of the two proteins . The N-terminal region is sufficient to mediate secretion via the SPI-1 TTSS (Hardt and Galán, unpublished observations) and, based on data for the SptP-SicE complex from Salmonella serovar Typhimurium and for the Yersinia pYV-encoded TTSS, is expected to harbor the chaperone binding site (4, 58, 64) .

We swapped the N-terminal 95 aa of SopE_M45 and SopE2_M45 . The stabilities of the hybrid proteins in the cytosol of M590 (invG invB::aphT) and SB161 (invG) were compared to the stabilities of SopE_M45 and SopE2_M45 (Fig . 8B to E) . We observed only minimal degradation of SopE_M45 and SopE_1-95SopE2_96-240M45 in SB161 (invG) and in the invB mutant strain M590 (invG invB::aphT) (Fig . 8B and D) . SopE_M45 and SopE_1-95SopE2_96-240M45 were equally stable in the invB⁺ and invB strains over the entire course of the experiment .

In contrast, SopE2_M45 and SopE2_1-95SopE_96-240M45 were rapidly degraded in the absence of invB (Fig . 8C and E, lower panels) . These data indicated that the N-terminal 95 aa of SopE_M45 and SopE2_M45 determine the different stabilities of the M45-tagged effector protein in the absence of invB .

Role of invB in SopE2 translocation. We used immunofluorescence microscopy to analyze the invB dependence of SopE2 translocation into host cells . To infect COS-7 tissue culture cells, we used the strains that were used for the secretion assay whose results are shown in Fig . 7A (MOI, 60 bacteria/cell) . Staining with a monoclonal anti-M45 antibody revealed that wild-type Salmonella serovar Typhimurium and the complemented invB strain [M568(pM249, pM256)] were able to translocate detectable amounts of SopE2_M45 into host cells (Fig . 10A) . In contrast, no SopE2_M45 was detected inside the COS-7 cells infected with the invB mutant M568(pM256) or a mutant with a defective SPI-1 TTSS apparatus [SB566(pM256)] (Fig . 10) . However, we cannot exclude the possibility that the failure to detect SopE2_M45 translocation by M568(pM256) in this assay was due to the severely reduced intracellular SopE2_M45 levels (Fig . 7A) . In order to increase the amounts of SopE2_M45, we repeated the translocation experiment using the medium-copy-number sopE2_M45 expression vector pM226 (Fig . 7) . The results confirmed that SopE2_M45 is translocated by wild-type strain SL1344(pM226) but not by the invB mutant M568(pM226) or the secretion-deficient mutant SB161(pM226) (Fig . 10C) . Overall, these data are in line with the notion that invB is required for translocation of SopE2_M45 into host cells .

 
Role of invB in Salmonella serovar Typhimurium host cell invasion. The translocation of effector proteins via the SPI-1 TTSS allows Salmonella serovar Typhimurium to invade host cells . We wanted to investigate the role of invB in this process .

SopB/SigD, SopE, and SopE2 cooperate to mediate tissue culture cell invasion, and a sopB sopE sopE2 triple mutant (M516) was approximately 100-fold less invasive than the isogenic wild-type strain (Fig . 11A) (30, 49, 76) . Recent reports have indicated that Sip/SspA might also play a role in host cell invasion (77, 78) . Consistent with this, we found that a sopB sopE sopE2 sipA quadruple mutant (M566) was slightly less invasive than the triple mutant (M516) (Fig . 11A) . The invasiveness of M566 did not differ significantly from that of a Salmonella serovar Typhimurium mutant with a defective SPI-1 TTSS apparatus (SB161 [invG]) (P > 0.05) (Fig . 11A) . This indicated that invasion of COS-7 tissue culture cells is mediated principally by two different groups of SPI-1 effector proteins which depend on two different chaperones; translocation of SopE, SopE2 (this study), and Sip/SspA (5) is dependent on InvB, while translocation of SopB/SigD depends on the chaperone PipC/SigE (12) .

 
Next, we analyzed the role of invB in SopE- and SopE2-mediated host cell invasion . To do this, we transformed sopE and sopE2 expression vectors (pSB1130 and pM149, respectively) into the quadruple mutant strain M566 (sopB sopE sopE2 sipA) and the isogenic mutant M574 (sopB sopE sopE2 sipA invB::aphT) . While sopE and sopE2 expression vectors were able to complement the host cell invasion defect of M566 (P < 0.05), no significant complementation was observed in invB strain M574 (Fig . 11B) (P > 0.05) . Significant complementation was observed only when M574 carried an additional invB expression vector (pM249) (Fig . 11B) (P < 0.05) . A slight polar effect of the invB::aphT mutation might be the reason for the incomplete complementation by pM249 (Fig . 11B) . Nevertheless, these data further support the notion that invB is required for SopE- and SopE2-mediated host cell invasion .

Interestingly, the invB mutant M568 is only slightly less invasive than the isogenic wild-type strain SL1344 (Fig . 12) . To test whether the invasiveness of M568 is attributable to the translocation of SopB, we constructed an invB sopB double mutant (M573) (see Materials and Methods) . Indeed, we found that the invasiveness of M573 is about as low as that of SB161, an isogenic strain with a disrupted SPI-1 TTSS apparatus (Fig . 12) . The invasiveness of M573 could be restored with an invB expression vector (pM249) . These data suggested that SopB/SigD translocation is not impaired in invB strains and provided further evidence that disruption of invB with the aphT cassette did not have a dramatic generalized effect on effector protein translocation . Rather, InvB seems to play the role of a chaperone for a specific subset of effector proteins, including Sip/SspA (5), SopE, and SopE2 (this study) .

 
The Salmonella effector proteins SopE and SopE2 are translocated into host cells via the SPI-1 TTSS (2, 31, 66, 75) . Previously, it was not clear how these proteins are recognized by this TTSS and whether chaperones are involved . Here, we show that SopE can be copurified with the SPI-1-encoded chaperone InvB and that SopE and SopE2 require InvB for secretion and translocation via the SPI-1 TTSS, as speculated in a recent review (53) . It has been reported previously that InvB is also required for efficient secretion and translocation of the SPI-1-encoded effector protein Sip/SspA (5) . We confirmed this finding . Therefore, InvB seems to act as a chaperone for at least three different Salmonella serovar Typhimurium effector proteins, although Sip/SspA and SopE/SopE2 do not exhibit any significant sequence similarities .

Similarly, Spa15, the InvB homolog of Shigella flexneri, acts as a chaperone for at least three known or suspected effector proteins with unrelated sequences (54), IpgB1, OspC3, and IpaA . IpaA exhibits sequence similarity with Sip/SspA from Salmonella serovar Typhimurium (35, 40) . In contrast to other type III chaperones which exhibit little sequence similarity, invB/spa15-like genes are present in a wide variety of bacteria, including Yersinia enterocolitica (ysaK [29]) and Sodalis glossinidius (invB [10]), and also on an Edwardsiella ictaluri plasmid (16) . Based on this wide distribution, an unusual gene locus embedded between functional genes of the TTSS apparatus, and the ability to interact with multiple unrelated effector proteins, it has been proposed that InvB/Spa15 may represent a new family of chaperones (54, 55) .

We found that disruption of invB has different consequences for the stability of the cognate effector proteins; SopE_M45 was stable in the cytosol of an invB mutant, whereas SopE2_M45 was not . Similarly, in S . flexneri Spa15 was required for the stability of IpgB1 but not for the stability of IpaA in the bacterial cytosol (54) . The reason for the differential requirement of InvB/Spa15 for effector protein stabilization in the bacterial cytoplasm is still not clear .

In a domain swap experiment we mapped the regions of SopE_M45 and SopE2_M45 that determine the resistance to degradation in the absence of InvB to the N-terminal 95 aa . This region overlaps the proposed InvB/Spa15 binding regions of Sip/SspA (aa 1 to 158), OspC3 (aa 26 to 141), and IpgB (aa 23 to 190) (5, 54) . Only the proposed Spa15 binding site of IpaA maps to a completely different region of the effector protein (aa 263 to 365) (54) . This interaction domain has been mapped in yeast two-hybrid screening analyses . It has not been determined whether this interaction also occurs in the bacterial cytoplasm and whether IpaA might have another chaperone binding site closer to its N terminus . Overall, however, the data suggest that the binding site for the InvB/Spa15 family of chaperones covers aa 23 to 95 of the effector proteins . This region overlaps the known binding regions of other type III chaperones (19, 21, 36, 61, 65, 72) .

Expression of the Salmonella serovar Typhimurium SPI-1 TTSS is restricted to specific locations in the host by an elaborate cascade of regulators, including PhoP/Q, SirA/BarA, InvF, HilA, HilD, SirC, and SprB (45) . Several transcriptional regulators have been identified which ensure proper coexpression of effector proteins encoded within and outside of SPI-1 (11-13, 15, 17, 70) . In addition, we show here that some effector proteins encoded within and outside of SPI-1 have the same chaperones for secretion and translocation into host cells . Therefore, Salmonella spp . possess at least two different mechanisms to coordinate effector protein expression and delivery: coregulation of transcription and sharing of secretion chaperones . The data show that effectors encoded outside of SPI-1 are directly controlled by SPI-1 itself .

The existence of tight regulatory networks which allow accurate functional integration of additional effector proteins is an important aspect of the horizontal gene transfer of type III effector proteins . Reassortment of effector protein repertoires has been discussed as a major driving force in the evolution of Salmonella spp., the adaptation to new hosts, and the emergence of new epidemic strains (17, 18, 49, 50, 67) . For example, Salmonella serovar Typhimurium lysogens carrying the sopE-encoding phage SopE caused a major epidemic in England and the former East Germany in the 1970s and 1980s (50) . It is thought that acquisition of the sopE gene by lysogenic conversion with SopE was an important step in the emergence of the epidemic strain . The data presented here explain why the phage-encoded effector protein SopE can actually function properly after lysogenic conversion of naïve Salmonella strains: the regulators and the chaperone InvB are already present (24, 56), no matter which Salmonella strain is infected by the phage (50) . In conclusion, the functional control of effector proteins encoded in different genetic elements by SPI-1 regulators and chaperones might play an important role in the evolution of Salmonella host interactions because it ensures proper functional integration of newly acquired proteins into existing effector protein repertoires .

This work was funded in part by a grant from the Deutsche Forschungsgemeinschaft (DFG) to W.-D.H .