Introduction

The genus Salmonella contains pathogenic bacteria that infect humans and animals; for instance, Salmonella enterica serovar Typhimurium (STM) can cause gastroenteritis and systemic infections in both humans and other mammals1,2. STM has many virulence genes clustered in regions called Salmonella pathogenicity islands3. Both Salmonella pathogenicity islands 1 and 2 (SPI-1 and SPI-2) encode a type III secretion system (T3SS-1 and T3SS-2, respectively), effector proteins, chaperones and transcriptional regulators that control genes within and outside these islands4,5. The T3SS-1 is required for invasion and replication of Salmonella in the cytosol of epithelial cells while the T3SS-2 is necessary for survival and replication within the Salmonella containing vacuole (SCV)6. The expression of the T3SS-1 is controlled by a complex regulatory network where the transcriptional regulators HilD, HilC and RtsA form a positive feed-forward loop that activates hilA transcription, HilA in turn activates the expression of several genes involved in the biosynthesis of the T3SS-1 and also that of InvF, a transcriptional regulator that belongs to the AraC/XylS family7. InvF interacts with SicA, a chaperone protein that binds to multiple effector proteins and T3SS-1 components8,9,10. Finally, the InvF/SicA complex activates transcription of many genes including sicA, sopB, sptP, sopE, sopE2, STM1239 and genes encoding for components of the T3SS-1 (Fig. 1)11,12. Thus, InvF represents an important transcriptional regulator for STM pathogenesis.

Figure 1
figure 1

Simplified representation of the SPI-1 regulatory cascade. The expression of SPI-1 is controlled by a regulatory network where the transcriptional regulators HilD, HilC and RtsA form a feed-forward loop that activates HilA. HilE interacts with HilD inactivating this regulator while H-NS downregulates HilD and HilA. HilA then activates the expression of genes involved in the biosynthesis of the T3SS-1 and, also activates InvF, the last transcriptional activator of this cascade. InvF both regulates expression of the chaperone SicA and interacts with it. The InvF/SicA complex regulates expression of many effector genes included sopB, sptP, sopE, sopE2, STM1239. Green arrows represent gene expression activation, red blunted lines represent repression of expression and red arrows represent inactivation.

Full size image

Previous findings support that InvF acts as classical transcriptional regulator9,10. However, it is not clear whether InvF interacts with the RNA polymerase (RNAP) to activate gene transcription as other classical regulators from the AraC/XylS family, such as PerA, MelR and XylS13,14,15. In this report we demonstrate that both InvF and SicA interact with the RNAP machinery through the alpha subunit in vitro and in vivo, further deciphering the mechanism by which the InvF/SicA complex induces gene expression.

Results

InvF interacts with the RNA polymerase alpha subunit

To detect probable protein–protein interactions of InvF with other cytoplasmic proteins, pull-down assays were performed with a recombinant version of InvF fused to the maltose-binding protein (MBP-InvF) as the bait and cell-free soluble extracts obtained from STM wild type and its derivative invF::Tn5 mutant grown in SPI-1-inducing conditions16. Purified MBP was assessed as a bait negative control. Interacting pulled down proteins were analyzed by SDS-PAGE, proteins captured with MBP-InvF, but not with MBP were selected and analyzed by LC/MS–MS (Fig. 2 and Supplementary File 1). Among the detected proteins interacting with MBP-InvF multiple subunits of the RNAP were identified, such as the beta, sigma and alpha subunits, suggesting interactions with at least one of them (Table 1). These results suggest that InvF interacts with the RNAP.

Figure 2
figure 2

MBP-InvF protein–protein interactions with Salmonella extracts. Pulled down proteins were separated by SDS-PAGE and selected bands were excised (shown with asterisks). Molecular weight markers are shown in each side of the gel, while the use or not of either MBP-InvF or MBP is shown with ( +) and (−), respectively. Either the wild type STM (WT ext) or the STM invF mutant (invF::Tn5 ext) extracts used in each interaction is indicated.

Full size image
Table 1 Summary of identified proteins interacting with InvF and SicA.
Full size table

Previous studies have shown that other members of the AraC/XylS family of transcriptional regulators interact with the alpha subunit of the RNAP (RpoA)13,14,17,18,19,20,21,22,23,24,25,26. To determine whether InvF binds to RpoA multiple approaches were followed. Initially, pull-down experiments between InvF and RpoA were performed using purified MBP-InvF or His6-RpoA proteins and with cell-free soluble extracts of the STM invF::3xFLAG strain transformed with the pET28-RpoA plasmid. Results showed that in vitro MBP-InvF was able to interact with His6-RpoA (Fig. 3A) and a similar result was observed in vivo (Fig. 3B) when InvF-FLAG and His6-RpoA (bait) were used. These results demonstrated that the regulator MBP-InvF interacts with His6-RpoA.

Figure 3
figure 3

The regulator InvF interacts with RpoA. (A) Pull-down assay to detect in vitro InvF-RpoA interactions. A pull-down assay using MBP or MBP-InvF proteins and purified His6-RpoA was performed. Left panel shows the control proteins (His6-RpoA, MBP and MBP-InvF) and the interaction reactions His6-RpoA + MBP and His6-RpoA + MBP-InvF detected with anti-MBP antibodies. Amylose resin was used for the pull-down reactions. Right panel shows the detection of His6-RpoA with a His-probe (α-His). (B) Pull-down to detect in vivo the InvF-RpoA interaction. Cell-free extracts of Salmonella expressing InvF-FLAG and His6-RpoA from plasmid pET28-RpoA was pulled down with Ni–NTA resin. The purified protein His6-RpoA and the cell-free extract were used as controls. Proteins were detected by Western blot with His-Probe and anti-FLAG-HRP. Cell-free extracts were obtained from the indicated cultures in SPI-1-inducing conditions as described in the “Methods” section. Experiments were performed in triplicate.

Full size image

Additionally, the interaction between InvF and RpoA was tested by using a bacterial LexA-based two hybrid system27,28, which has been successfully used in our laboratories to assess protein–protein interactions10,29,30. Briefly, in this bacterial two hybrid system the protein of interest is fused to the wild type LexA DNA binding domain (LexADBDwt) and the construct is transformed into an E. coli reporter strain (SU101) encoding a transcriptional fusion sulA-lacZ, which has a LexA wild type operator. This system also allows the detection of heterodimers, this is achieved by fusing the other protein of interest to a mutated version of the LexA DNA binding domain (LexADBDmut), then both constructs are transformed into an E. coli strain (SU202) encoding a mutated version of the sulA-lacZ fusion. In both cases, the expression of LacZ means that there is no protein–protein interaction and, on the contrary, a reduction of the β-galactosidase activity would mean that, and interaction of the fused protein (s) has occurred (homodimerization for SU101 and heterodimerization for SU202)27,28. InvF and RpoA were fused to either the wild type or to the mutated LexA DNA binding domain (LexADBDwt and LexADBDmut, respectively) and the β-galactosidase activity was tested in the reporter strain. Negative controls included the empty vectors and a combination of LexADBDmut-RpoA and LexADBDwt, while the positive control was the chimeric proteins LexADBDwt-HilD and LexADBDmut-HilE. Results in Fig. 4 show that control fusions LexADBDwt-HilD and LexADBDmut-HilE repress the expression of the sulA-lacZ, as expected30. When the fusion protein LexADBDmut-RpoA was tested with either LexADBDwt-InvF or LexADBDwt-SicA, the expression of the reporter fusion was reduced, indicating that RpoA interacts with both proteins.

Figure 4
figure 4

RpoA interactions with InvF and SicA detected with a LexA-based two hybrid system. β-galactosidase activity of E. coli SU202 strains transformed with LexA-derivative plasmids grown and processed as described in the “Methods” section. Constructs are indicated below each bar. Bars represent the average of three independent experiments and the error bars represent the standard deviation. * indicates statistically significant difference (P < 0.01) compared to the controls indicated with lines.

Full size image

Subsequently, to determine the region of InvF contacting RpoA, the N-terminal and C-terminal regions of InvF were fused to LexADBDwt and tested for dimerization with RpoA. Both, the N-terminal (LexADBDwt-InvFNTD) and the C-terminal domains (LexADBDwt-InvFCTD) domains of InvF interacted with RpoA (LexADBDmut-RpoA) (Fig. 4). The interaction of both InvF domains with SicA was also tested and results showed that InvF C-terminal domain (LexADBDwt-InvFCTD) interacts with SicA, while the N-terminal domain (LexADBDwt-InvFNTD) does not (Fig. 5). These results confirm that RpoA makes contacts with the whole protein and also independently with both InvF domains.

Figure 5
figure 5

Interactions of InvF and SicA detected with a LexA-based two hybrid system. β-galactosidase activity of E. coli SU202 strains transformed with LexA-derivative plasmids grown and processed as described in the “Methods” section. Constructs are indicated below each bar. Bars represent the average of three independent experiments and the error bars represent the standard deviation. * indicates statistically significant difference (P < 0.01) compared to the controls indicated with lines.

Full size image

SicA interacts with RpoA

Previous evidence indicates that SicA is strictly necessary for InvF to activate transcription of target genes and that both proteins interact9,10,12,31. Based on in silico analyses, we recently predicted that SicA makes contacts not only with InvF, but also with RpoA32. To investigate this prediction, the interaction between SicA and RpoA was tested by using the LexA-based dimerization system and pull-down experiments. Both approaches showed that SicA interacts with RpoA; the combination LexADBDwt-SicA and LexADBDmut-RpoA repressed the expression of the sulA-lacZ fusion in the LexA-based system (Fig. 4), whereas the His6-RpoA captured SicA-FLAG in the pull-down assays (Fig. 6). These results show that SicA interacts also with RpoA independently of InvF and supports a model indicating that the three proteins might form a trimeric complex.

Figure 6
figure 6

The chaperone SicA interacts with RpoA. Pull-down assays performed with Ni–NTA magnetic beads to detect in vitro SicA-FLAG and His6-RpoA interactions. Purified His6-RpoA and cell-free extract containing SicA-FLAG were used as controls. The chimeric proteins were detected by Western blot with His-Probe and anti-FLAG-HRP antibodies. Cell-free extracts were obtained from the indicated cultures as described in the “Methods” section. Experiments were performed by triplicate.

Full size image

InvF, SicA and RpoA interactions

To corroborate whether the three proteins interact forming a ternary complex pull-down experiments were done with cell-free extract of STM invF::3xFLAG pET28-RpoA using His6-RpoA as bait. Pulled-down proteins were detected by Western blot with either anti-SicA or anti-FLAG-HRP antibodies. Results in Fig. 7 show that both proteins InvF-FLAG and SicA from the extract were detected interacting with His6-RpoA suggesting that these three proteins might be forming a complex in solution.

Figure 7
figure 7

InvF, SicA and RpoA interactions. Pull-down assays performed with Ni–NTA magnetic beads to detect in vivo InvF-RpoA-SicA interactions. Purified His6-RpoA and cell-free extract containing InvF-FLAG and SicA were used as controls. Proteins were detected by Western blot with His-Probe, anti-FLAG-HRP and anti-SicA antibodies. Cell-free extracts were obtained from the indicated cultures in SPI-1-inducing conditions as described in the “Methods” section. Experiments were performed by triplicate.

Full size image

RpoA α-CTD is important for sopB expression

To investigate whether the RpoA carboxy-terminal domain (α-CTD) is involved in gene expression mediated by InvF/SicA, we analyzed the effect of two negative dominant versions of RpoA on the expression of sopB. Plasmids pLAD235 and pLAD256, expressing RpoA negative dominant mutants lacking different portions of α-CTD, were transformed in wild type STM and the expression of sopB was tested by RT-qPCR in samples of bacterial cultures grown in SPI-1-inducing conditions33. Expression of sopB decreased with both mutants (Fig. 8A). Additionally, the expression of SopB-FLAG in the presence of the RpoA negative dominant mutants was analyzed by Western blot (Fig. 8B). SopB-FLAG expression was eliminated in STM transformed with plasmids encoding both RpoA negative dominants (pLAD235 and pLAD256). Controls included the invF::Tn5 mutant, the bacteria transformed with the empty vector (pINIIIA1) and wild type RpoA (pLAX185). The observed discrepancy between the sopB transcription and the translation could be explained by multiple factors including selection of normalized gene for RT-PCR, mRNA stability and translation. Despite these it is clear that sopB expression depends on the presence a wild type RpoA and the InvF/SicA complex. Together these results showed that α-CTD is necessary for the expression of the InvF-dependent gene sopB.

Figure 8
figure 8

The sopB expression is affected by RpoA negative dominant mutants. (A) The differential expression of the InvF-dependent gene, sopB, was analyzed by RT-qPCR in Salmonella Typhimurium transformed with plasmids pLAX185 (wild type rpoA), pLAD235 (rpoAΔ235 mutant) and pLAD256 (rpoAΔ256 mutant) grown in SPI-1-inducing conditions. Data represent the means of three different experiments. The bars indicate the standard deviation. Expression analysis was performed using the ΔΔCt method. *. P < 0.05. (B) SopB expression in Salmonella Typhimurium (SMT) sopB::FLAG strains transformed with pNIIIA1, pLAX185, pLAD235 and pLAD256 grown in SPI-1-inducing conditions. GroEL was used as load control. SopB-FLAG and GroEL were detected by Western blot with anti-FLAG-HRP and anti-GroEL antibodies. The figure shows one of the three replicates done.

Full size image

Discussion

InvF is an AraC/XylS-like transcriptional regulator important for the transcription of virulence genes encoded inside and outside of SPI-1 in STM, such as sicA and sopB, among others. The products of these genes are necessary for this bacterium invasion to epithelial cells9. In previous reports we and others have shown that InvF acts as a monomer, that SicA is necessary to activate gene transcription, and that InvF likely functions as a classical regulator that likely recruits the transcriptional machinery to the promoter9,10,31. Thus, our working hypothesis here was that InvF interacts with the RNAP in a similar way as other classical bacterial transcriptional activators including a few from the AraC/XylS-like regulators34. In this report we have demonstrated by different approaches that InvF binds to the alpha subunit of the RNAP and that SicA is also able to independently contact both proteins.

InvF belongs, together with MxiE and BsaN, to a singular group in the AraC/XylS family of transcriptional activators as they require of small proteins to function as co-activators34. In these activators the need of their respective co-activator has been shown but the molecular interactions between them and the transcriptional machinery has not been shown. Previously, we demonstrated that InvF acts as a monomer and that it is required for expression of sopB in the absence of the repressor H-NS10, suggesting that it is a classical activator and as such it likely makes contacts with the RNAP. Results presented here corroborated that InvF indeed interacts with RpoA, the alpha subunit of the RNAP, confirming our hypothesis. A prediction model recently published also by our group32 suggests that InvF makes stable interactions with the RpoA carboxyl domain (α-CTD). Here, this was experimentally corroborated by using purified versions of InvF and RpoA, a bacterial two hybrid system, and with the use of RpoA negative dominant mutants. Moreover, molecular dynamics simulations between these two proteins suggests that α-CTD interacts with both InvF domains, the amino terminus domain (NTD) and the DNA binding domain (DBD)32. This was also experimentally verified here, proving again the bioinformatic predictions. Interactions of other AraC/XylS regulators with the RNAP in both domains have been shown, such as the NTD in XylS14, and the DBD in MelR, SoxS, MarA and Rob13,17,18,19,20,21,22,26. These reports indicate that these AraC/XylS regulators interact with the RNAP by the NTD or the CTD domains but interactions with both domains had not been shown as here is reported for InvF. Future experiments will include the generation of InvF point mutations in those residues likely interacting with α-CTD. Lastly, the bioinformatic analyses also predicted that RpoA stabilizes the DBD in InvF. Whether this stabilization changes or modifies InvF affinity for its DNA target will be experimentally tested in our laboratories.

The InvF-SicA interaction and its relevance for transcriptional activation was shown since the initial studies done in Dr. Virginia Miller laboratory9,12,31. Recently, the InvF/SicA complex has been corroborated to be necessary for transcriptional activation10,35,36. Li and coworkers35 also showed that SicA can detect the second messenger cyclic-di-GMP (c-di-GMP) and that this molecule abolishes SicA interaction with InvF, SipB and SipC. In contrast with our previous study, they showed that SicA is necessary for InvF to interact with the DNA10. A possible difference is that in our study we used a fusion of InvF to the MBP while they used an His6-tagged version. What both studies coincide in is with the fact that the InvF/SicA complex is necessary for transcriptional activation. SicA is a class II chaperone from the CesD/SycD/LcrH family of T3SS chaperones that is involved in SipB and SipC effectors transport, in addition to its role as transcriptional co-activator making this protein one of the most indispensable for virulence in Salmonella37,38. SicA contains three tetratricopeptide repeat-like motifs (TPR) that have been shown by Kim and coworkers36 to be relevant for protein interactions and transcriptional activation. In this same report, they observed that point mutations in each of the three TPRs were affected in their stability when expressed from their own promoter and when over-expressed they were unable to completely activate expression of sipB and sopB (sigD), purified versions of these mutants were able to interact with InvF though TPR2 and TPR3 mutants had a weaker interaction. Li et al.35 showed that residues in these TPRs are relevant to detect c-di-GMP and that a SicA N70A mutant can contact InvF, to activate transcription but is unable to detect this second messenger. Given that the InvF/SicA complex is needed for a complete transcriptional activation it is possible that SicA has a complementary role. In silico study identified three cavities in the InvF/RNA complex and a docking study allowed us to obtain the mode of interaction between the three TPR motifs of SicA with the predicted cavities32. These predictions were corroborated with a molecular dynamics simulation that revealed the role of SicA in the molecular stabilization of InvF DBD motif. Although here we did not seek to determine whether the latter was occurring, results showed that RpoA, InvF and SicA form a complex in solution and that each of the proteins are able to interact with the other independently. To the best of our knowledge, this is the first report showing the interaction of a T3SS chaperone with subunits of the RNAP. These results would also explain why the TPR mutants described by Kim though interacting with InvF fail to activate transcription of sipB and sopB, probably by affecting the interaction with RpoA36. Future experiments need to be done to prove this possibility. In this sense, it is possible to propose that once SicA delivers the effectors SipB and SipC it can interact with either InvF or the RNAP through RpoA or with the InvF/RNAP complex, then, this oligomeric complex would be able to initiate transcription in the InvF-dependent genes (Fig. 9).

Figure 9
figure 9

Model for the transcription of sopB by InvF, SicA and the RNAP. In this model SicA first delivers the translocators SipB and SipC to the T3SS-1 (a); then SicA is able to form dimers or tetramers (b) and is able to either interact with invF (c), RpoA (d), or both; once the InvF-SicA complex is bound to the RNAP through RpoA they bind to the sopB promoter region to initiate transcription. Alternatively, InvF could be recruited to the promoter region by the SicA/RNAP complex or that the InvF/SicA complex recruits the RNAP (e).

Full size image

In conclusion, here we have presented evidence showing that RpoA interacts with InvF and that this contact is necessary for transcription of InvF-depending genes, such as sopB. Moreover, SicA, a T3SS chaperone can not only bind InvF but to also interact with RpoA, suggesting that SicA stabilizes the InvF DBD, facilitating the interaction with RpoA, and thus likely promoting transcription.

Methods

Bacterial strains and culture conditions

The strains used in this study are listed in Supplementary Table S1. Lysogeny–Bertani (LB) or LB-Miller broth were used for bacterial cultures at 37 °C. When indicated antibiotics were added at the following concentrations: ampicillin (100 μg/mL), streptomycin (100 μg/mL), chloramphenicol (30 μg/mL), kanamycin (30 μg/mL), and tetracycline (10 μg/mL). SPI-1 inducing conditions were as described before16.

DNA manipulations

Plasmid DNA was purified using the Wizard Plus SV Minipreps DNA Purification Systems kit (Promega). The oligonucleotides used for amplification were synthetized by the IBT-UNAM and are listed in Supplementary Table S2. PCR reactions were performed by using DreamTaq Green PCR Master Mix (2X) (Thermo Fisher Scientific). Purified plasmids and PCR products were observed in 1% agarose gels stained with either ethidium bromide (Sigma) or SYBR green (Thermo Fisher Scientific).

The sopB gene was 3XFLAG-tagged in the S. Typhimurium SL1344 strain using a previously reported method based on the λRed one-step inactivation using pSUB11 plasmid as template39 and oligos sopBFlag-1 and sopBFlag-2 (Supplementary Table S2), generating the MD1163 (sopB::3XFLAG-kan) strain. Kanamycin resistance was removed with plasmid pCP20 as described previously to generate MD1180 (sopB::3XFLAG) strain. SopB-FLAG expression was tested by Western blot as described below.

Construction of plasmids

Plasmids used in this work are listed in Supplementary Table S1. To construct plasmid pTOPO-SicA-FLAG, sicA-RBS-Fw and SicA-FLAGrv oligos were used to amplify sicA using STM genomic DNA as template, the amplicon was then cloned into pCRTOPO 2.1 using the TOPO-TA cloning kit (Thermo Fisher Scientific) following the instructions provided by the company. Plasmids pSR658-NTDInvF and pSR658-CTDInvF were obtained by amplifying the two separated domains of InvF using pSR658-InvF as template with oligos lexA-InvF-Fw, invf-NDT-Rv, CTDInvFfw and InvFrevLexA (Supplementary Table S2) and cloned into pSR658. Plasmid pSR659-RpoA was obtained by amplifying rpoA from pET28-RpoA with oligos RpoALexAFw and RpoALexARv and cloned into pSR659. All plasmid constructs were sequenced by Macrogen, Inc. (South Korea).

Expression and purification of MBP-InvF and MBP

Expression and purification of MBP-InvF and MBP was done by affinity chromatography with amylose resin as described previously10. The proteins were dialyzed in buffer containing Tris-base 20 mM, KCl 50 mM, DTT 1 mM and glycerol 5% with a D-Tube Dialyzer Mega 20 mL MW 3.5 kDa (Millipore). The concentration of purified proteins was determined by using the BCA Protein Assay Kit (Thermo Scientific) and analyzed in a 12% sodium dodecyl sulfate-polyacrilamide (SDS-PAGE) gel electrophoresis. Aliquots were stored at − 20 °C until used.

Expression and purification of His6-RpoA and SicA-His6

The His-tagged proteins were purified by affinity chromatography with Ni–NTA agarose (Qiagen) as described previously10. Proteins were dialyzed and protein concentration was determined as mentioned above. Aliquots were stored at − 20 °C until used.

SicA antibodies

Polyclonal anti-SicA antibodies were produced in pathogen-free eight-week-old female BALB/c mice under standard animal facility conditions following the protocol approved by the ethics committee of the Instituto de Biotecnología UNAM (IBT- UNAM) (https://www.ibt.unam.mx/documentos/general/aprobadocombioet27nov2019protocolo-pdf-694.pdf) that complies with the guidelines described by the Canadian Council on Animal Care (https://ccac.ca/Documents/Standards/Guidelines/Antibody_production.pdf). We confirm that this study is reported in accordance with ARRIVE guidelines (https://arriveguidelines.org). Briefly, they were inoculated with 10 µg of purified protein mixed with incomplete Freund's adjuvant (Sigma-Aldrich) via intraperitoneal injection in a total volume of 200 µl. The mice were immunized 4 times at 21-day intervals. Prior to each immunization, a blood sample was collected through a small tail cut, and the serum was collected and stored at − 20 °C until analysis. A direct ELISA was conducted using the purified SicA protein as the antigen, and the titers of each obtained serum were determined. Once a maximum detection was achieved with a titer of 1:400, the mice were euthanized under general anesthesia. Blood was obtained via cardiac puncture and subsequently centrifuged to obtain the serum. The sera from three mice were collected, mixed to form a homogeneous pool, aliquoted, and stored at − 20 °C. These antibodies were tested for Western blot with purified SicA, MBP-InvF and RpoA to determine whether possible cross-reactivity.

Pull-downs for InvF-RpoA interactions

Prior to the experiments, the expression of chimeric proteins obtained in this work were evaluated by Western blot (Fig. S1). Pull-down experiments were performed with purified MBP, MBP-InvF and His6-RpoA. The following mixtures were done: a negative control with MBP and His6-RpoA and MBP-InvF and His6-RpoA; mixtures were done by using 50 μg of each protein in an 2X interaction buffer (100 mM NaH2PO4, 600 mM NaCl, 40 mM imidazole, 0.5% NP-40 and 20% glycerol, pH 8.0)40. Proteins were let to interact for 30 min on ice, then 50 μL of amylose resin (New England Biolabs) were added to each mixture and let to interact for 2 h in agitation at ~ 4 °C in a tube rotator (Thermo Scientific). Beads were centrifuged at 2000 × g for 2 min, amylose beads were washed three times with cold washing buffer. After the last washing step, the supernatant was removed carefully and then 20 μL of Laemmli buffer were added. Samples were resolved in an 12% SDS-PAGE and stained with Coomassie blue. Western blot was performed by transferring the proteins from the SDS-PAGE to a PVDF membrane (Merck) by following a previously described protocol10. Western blot was developed with a His-Probe (1:5,000) (ThermoFisher) and anti-MBP antibodies (1:10,000) (New England Biolabs) by using chemiluminescence kit (Invitrogen) and observed in a Chemidoc imaging system (Biorad).

Pull-down experiments were also performed with cell-free extracts containing InvF-FLAG and His6-RpoA. For this, S. Typhimurium invF::3xFLAG transformed with pET28-RpoA and cell-free extract was obtained from bacterial cultures grown in SPI-1-inducing conditions complemented with IPTG 1 mM. Then 100 μL of the cell-free extract were mixed with 50 μL of Ni–NTA resin and let to interact for 2 h in agitation at ~ 4 °C in 1X interaction buffer. Samples were centrifuged at 2000 × g for 2 min and washed four times with low imidazole buffer. After the last washing step supernatant was removed carefully, 30 μL of Laemmli buffer were added to the beads and samples were boiled for 10 min. Samples were resolved in a 12% SDS-PAGE, Western blot was performed by transferring the proteins to a PVDF membrane and using anti-FLAG-HRP antibodies (1:5,000) (Abcam AB49763) and His-Probe-HRP (1:5,000) as suggested by the manufacturers. Membrane development was done as described above.

Pull-down for SicA-RpoA interactions

Experiments were performed with purified His6-RpoA and extract of the E. coli BL21 pTOPO-SicA-FLAG strain. For this 100 μL of the cell-free extract were mixed with 50 μL of purified His6-RpoA, 50 μL of Ni–NTA magnetic beads (Thermo Scientific) and 1X interaction buffer and let to interact for 2 h in agitation at ~ 4 °C. Samples were collected with a magnetic stand (New England Biolabs) for 2 min and washed four times with low imidazole buffer. After the last washing step supernatant was removed carefully, 30 μL of Laemmli buffer were added to the beads and samples were boiled for 10 min. Samples were resolved in a 12% SDS-PAGE, Western blot was performed by transferring the proteins to a PVDF membrane and using anti-FLAG-HRP antibodies (1:5,000) (Abcam AB49763) and HisProbe-HRP (1:5,000) as suggested by the manufacturers. Membrane development was done as described above.

Pull-down for InvF-SicA-RpoA interactions

For the triple interaction 100 μL of cell-free extract of STM invF::3xFLAG transformed with plasmid pET28-RpoA were mixed with 50 μL of Ni–NTA magnetic beads and let to interact for 2 h in agitation at ~ 4 °C in 1 × interaction buffer. Samples were collected with a magnetic stand for 2 min and washed four times with low imidazole buffer. After the last washing step supernatant was removed carefully, 30 μL of Laemmli buffer were added to the beads and samples were boiled for 10 min. Samples were resolved in a 12% SDS-PAGE, Western blot was performed by transferring the proteins to a PVDF membrane and using anti-FLAG-HRP (1:5,000) and anti-SicA antibodies (1:5,000) obtained for this work (Fig. S2) and His-Probe-HRP (1:5,000) as suggested by the manufacturers. Membrane developing was done as described above.

Interaction between InvF and STM cell-free extracts

In order to corroborate interactions of InvF with cytoplasmic proteins from STM, cell free extracts of STM WT and STM invF::Tn5 were used. First 70 μL of amylose magnetic beads (New England Biolabs) previously washed with a washing buffer were mixed with MBP-InvF 32.5 μg/μL or MBP 10 μg/μL as a control and let to interact overnight at 4 °C. Then the beads were washed three times with 200 μL washing buffer and 400 μL of cell-free extracts were added and let to interact in agitation at 4 °C for 6 h. After the incubation the beads were washed 10 times with a washing buffer. After the last washing step supernatant was removed, 30 μL of Laemmli buffer were added to the beads and samples were boiled for 10 min. Samples were resolved in a 12% SDS-PAGE. Differential bands were cut from the gel and the proteins were characterized by LC/MS–MS at the Proteomics Discovery Platform of the Institut de Recherches Cliniques de Montréal (Quebec, Canada). Scaffold (version Scaffold_5.2.0, Proteome Software Inc., Portland, OR) was used to validate MS/MS based peptide and protein identifications. Peptide identifications were accepted only if they could be established at greater than 95.0% probability by the Scaffold Local FDR algorithm. Protein identifications were accepted only if they could be established at greater than 95.0% probability and contained at least 2 identified peptides. Protein probabilities were assigned by the Protein Prophet algorithm41. Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony.

Dimerization assays

A LexA-based two hybrid system was used to evaluate protein–protein interactions between InvF, SicA and RpoA27,28. To verify the integrity of the LexA-derived proteins a Western blot was performed by using the anti-LexA antibody (Millipore) (1:5,000) and recombinant protein G-HRP (Thermo Scientific) (1:10,000) (Fig. S3). To test InvF and SicA heterodimerization competent cells of E. coli SU202 were transformed individually or combined with pSR658-InvF, pSR658-NTDInvF, pSR658-CTDInvF, pSR659-SicA and pSR659-RpoA, and selected in LB plates with the corresponding antibiotics. Transformants were grown in LB supplemented with 1 mM IPTG and let them grow to an OD600 = 0.6. Aliquots were taken to assess β-galactosidase activity as described before10. Plasmids pSR658, pSR659, pSR658-HilD and pSR659-HilE were used as controls.

RT-qPCR assays

Relative expression of sopB in the different STM strains was determined by RT-qPCR as described previously10. Briefly, RNA was obtained from bacterial cultures grown in SPI-1-inducing conditions. DNA was removed with DNA-Free (Ambion) and then cDNA was obtained with a GoScript kit (Promega). qPCR was performed in a Rotor gene Q Thermocycler (Qiagen). Relative expression of sopB was calculated with the ΔΔCt method using the expression of the gene gyrB as a normalizer. Oligos for each gene are listed in Supplementary Table S2. Experiments were done in triplicates and the results are the average of three independent experiments.

sopB expression assay

Briefly, 350 μL of an overnight culture of STM sopB::FLAG, STM invF::Tn5 sopB::FLAG and STM sopB::FLAG transformed with pINIIIA1, pLAX185, pLAD235 and pLAD256 were inoculated in flasks with 10 mL of LB-Miller complemented with IPTG 1 mM, the cultures were incubated for 3.5 h at 37 °C in agitation (225 rpm) and the optical density was measured at 600 nm. The cultures were centrifugated for 20 min at 11,000 × g at 4 °C, the supernatants were discarded, then 100 μL of Laemmli buffer were added to the pellets and were heated at 95 °C for 10 min. Samples were loaded onto a 12% SDS-PAGE and stained with Coomassie stain. Western blot was performed by transferring the proteins to a PVDF membrane (Merck) and using anti-FLAG-HRP (1:5,000), anti-GroEL(1:10,000) antibodies (Abcam) and recombinant protein G-HRP. Membrane development was done as described above.

Statistical analysis

Statistical analysis was performed in GraphPad Prism version 6.01 (Graph-Pad Software) by using a Student’s t-test. A significant difference was considered when P < 0.01 for the dimerization assays and 0.05 for the RT-qPCR assays.