Modulation of Salmonella virulence by a novel SPI-2 injectisome effector that interacts with the dystrophin-associated protein complex

ABSTRACT The injectisome encoded by Salmonella pathogenicity island 2 (SPI-2) had been thought to translocate 28 effectors. Here, we used a proteomic approach to characterize the secretome of a clinical strain of invasive non-typhoidal Salmonella enterica serovar Enteritidis that had been mutated to cause hyper-secretion of the SPI-2 injectisome effectors. Along with many known effectors, we discovered the novel SseM protein. sseM is widely distributed among the five subspecies of Salmonella enterica, is found in many clinically relevant serovars, and is co-transcribed with pipB2, a SPI-2 effector gene. The translocation of SseM required a functional SPI-2 injectisome. Following expression in human cells, SseM interacted with five components of the dystrophin-associated protein complex (DAPC), namely, β-2-syntrophin, utrophin/dystrophin, α-catulin, α-dystrobrevin, and β-dystrobrevin. The interaction between SseM and β-2-syntrophin and α-dystrobrevin was verified in Salmonella Typhimurium-infected cells and relied on the postsynaptic density-95/discs large/zonula occludens-1 (PDZ) domain of β-2-syntrophin and a sequence corresponding to a PDZ-binding motif (PBM) in SseM. A ΔsseM mutant strain had a small competitive advantage over the wild-type strain in the S. Typhimurium/mouse model of systemic disease. This phenotype was complemented by a plasmid expressing wild-type SseM from S. Typhimurium or S. Enteritidis and was dependent on the PBM of SseM. Therefore, a PBM within a Salmonella effector mediates interactions with the DAPC and modulates the systemic growth of bacteria in mice. Furthermore, the ΔsseM mutant strain displayed enhanced replication in bone marrow-derived macrophages, demonstrating that SseM restrains intracellular bacterial growth to modulate Salmonella virulence. IMPORTANCE In Salmonella enterica, the injectisome machinery encoded by Salmonella pathogenicity island 2 (SPI-2) is conserved among the five subspecies and delivers proteins (effectors) into host cells, which are required for Salmonella virulence. The identification and functional characterization of SPI-2 injectisome effectors advance our understanding of the interplay between Salmonella and its host(s). Using an optimized method for preparing secreted proteins and a clinical isolate of the invasive non-typhoidal Salmonella enterica serovar Enteritidis strain D24359, we identified 22 known SPI-2 injectisome effectors and one new effector—SseM. SseM modulates bacterial growth during murine infection and has a sequence corresponding to a postsynaptic density-95/discs large/zonula occludens-1 (PDZ)-binding motif that is essential for interaction with the PDZ-containing host protein β-2-syntrophin and other components of the dystrophin-associated protein complex (DAPC). To our knowledge, SseM is unique among Salmonella effectors in containing a functional PDZ-binding motif and is the first bacterial protein to target the DAPC.


IMPORTANCE
In Salmonella enterica, the injectisome machinery encoded by Salmonella pathogenicity island 2 (SPI-2) is conserved among the five subspecies and delivers proteins (effectors) into host cells, which are required for Salmonella virulence.The identification and functional characterization of SPI-2 injectisome effectors advance our understanding of the interplay between Salmonella and its host(s).Using an optimized method for preparing secreted proteins and a clinical isolate of the invasive non-typhoi dal Salmonella enterica serovar Enteritidis strain D24359, we identified 22 known SPI-2 injectisome effectors and one new effector-SseM.SseM modulates bacterial growth during murine infection and has a sequence corresponding to a postsynaptic den sity-95/discs large/zonula occludens-1 (PDZ)-binding motif that is essential for interac tion with the PDZ-containing host protein β-2-syntrophin and other components of the dystrophin-associated protein complex (DAPC).To our knowledge, SseM is unique among Salmonella effectors in containing a functional PDZ-binding motif and is the first bacterial protein to target the DAPC.

Discovery of SPI-2 injectisome effector SseM by proteomic analysis
To investigate the SPI-2 injectisome effector repertoire of a clinical isolate of Salmonella enterica subspecies enterica, we exploited the hypersecretion phenotype of a gatekeeper mutant.We chose an invasive non-typhoidal Salmonella enterica serovar Enteritidis (S.Enteritidis) strain D24359 that was isolated from the blood of a Malawian patient and is sensitive to antibiotics carbenicillin, kanamycin, and chloramphenicol (25) to make a ΔspiC single mutant (an effector hypersecretion mutant) and a ΔspiCssaC SPI-2 null mutant (4).The bacterial strains were grown in SPI-2-inducing medium MgM-MES at pH 5.0 for 6 h.Then, the supernatant was concentrated and subjected immediately to SDS-PAGE after which a 1 cm gel slice was analyzed by mass spectrometry (Fig. 1A).The resulting peptides were compared to predicted protein sequences from the annotated S. Enteritidis strain D24359 sequence.From two experiments, we identified 22 known SPI-2 effectors and one previously unidentified effector D24359_01053 (Table 1; Fig. 1B).
D24359_01053 consists of 103 amino acids.Searching the BLAST Protein database of S. Typhimurium LT2 showed that D24359_01053 is almost identical to STM2779.There are 95 identical residues in the predicted 110 amino acids of STM2779 (Fig. S1A).Bioinformatic analysis revealed that D24359_01053/STM2779 is conserved in all five subspecies of Salmonella enterica but not in Salmonella bongori, with most being predicted to be 110 amino acids in length (Fig. S1B).We named this effector SseM (Salmonella secreted effector M).
To further analyze the distribution of SseM and SseM variants among the serovars of S. enterica subspecies enterica, 834 complete Salmonella genomes with the highest assembly quality were downloaded from Enterobase (https://enterobase.warwick.ac.uk/) and used to construct an sseM database with stm2779 as the reference sequence.The SseM protein variants in the context of genomic or serovar diversity were displayed with the Grapetree phylogenetic tool (26) (Fig. 1C and D).Most of the sequenced strains of different serovars had full-length SseM (Fig. 1D), with the N-and C-terminal regions of SseM highly conserved (Fig. S2).Our analysis revealed the presence of two common pseudogene variants: sseM_2_pseudo, present in 111 out of 113 S. Typhi genomes, and sseM_6_pseudo, present in 40 out of 126 S. Enteritidis genomes (Fig. 1D).sseM_2_pseudo is the result of an additional cytosine in sseM of S. Typhi, which generates a stop codon immediately after the predicted 30th residue; while the sseM_6_pseudo is due to the mutation of the predicted 84th codon TGG to TAG, resulting in a truncated version of SseM (Fig. S3).In summary, SseM is widely distributed among the five subspecies of Salmonella enterica and so represents a new conserved "core" effector protein.

Expression, secretion, and translocation of SseM
sseM is located 175 nt downstream of SPI-2 effector gene pipB2 (Fig. 2A).Both pipB2 and sseM (stm2779) share the same transcriptional start site (27) and are controlled by SsrAB (20,28).To verify SsrAB dependence on the expression of SseM and to test if pipB2 and sseM were operonic or not, a rabbit polyclonal antibody against the C-terminal peptide (PYFPVVPGERETDV) of S. Typhimurium SseM was obtained.S. Typhimurium 12023 wt and derivative strains were grown in MgM-MES at pH 5.0, and proteins in whole bacterial lysates were immunoblotted.The antibody detected SseM in wt and a ΔsseM mutant expressing SseM from a plasmid (psseM) but not in lysates derived from the ΔsseM mutant, demonstrating the specificity of the SseM antibody (Fig. 2B).As expected, SseM was not detected in lysates derived from a ssrA::mTn5 mutant.Furthermore, deleting the promoter of pipB2 but not pipB2 itself led to the loss of SseM (Fig. 2B).Taken together, the data indicate that sseM and pipB2 are bicistronic, with their expression activated by SsrAB.
As SseM from S. Typhimurium strain LT2 and most of S. enterica species have been annotated as a 110-residue long protein that uses TTG rather than ATG (21 nucleotides downstream of TTG) as its start codon (Fig. S4), we defined the actual start codon of sseM by changing the ATG to ACG on plasmid psseM.The resulting plasmid psseM ACG was transformed into the ΔsseM mutant strain to check the expression of SseM by immunoblot.SseM was undetectable in the ΔsseM mutant carrying plasmid psseM ACG , indicating that sseM uses the ATG as its start codon to encode a 103-residue long protein (Fig. 2C).
Next, we investigated SPI-2 injectisome-dependent secretion of SseM by immuno blotting.The wt and ΔssaV mutant strains were grown in MgM-MES at pH 5.0 for 4 h to assemble the SPI-2 injectisome, then the pH of the medium was changed to 7.2 to allow effector secretion (5).Although SseM was detected in bacterial lysates from both wt and ΔssaV mutant strains, secreted SseM was only detected in samples prepared from the wt culture (Fig. 2D).This result agrees with the mass spectrometric result of S. Enteritidis strain D24359, demonstrating that SseM secretion is dependent on the SPI-2 injectisome.
Translocation of SseM into mammalian cells from intracellular Salmonella was then tested by immunoblotting.For this, HeLa cells were infected with different bacterial strains for 8 h, and then translocated proteins were extracted from post-nuclear supernatant with Triton X-100 and subjected to immunoblotting using the anti-SseM antibody.A small quantity of translocated SseM was detected from HeLa cells infected with the wt strain, and 7.50 ± 1.18 times more was detected in the mutant strain carrying sseM on a plasmid (calculated from the means of three biological repeats ± SD; P = 0.011) (Fig. 2E).However, attempts to detect translocated SseM with the rabbit anti-SseM antibody by immunofluorescence microscopy failed.To further investigate the translocation of SseM by microscopy, a plasmid expressing C-terminal HA-tagged SseM (psseM-HA) was transformed into the wt or ΔssaV mutant strains, and these were used to infect HeLa cells.Translocated SseM-HA was detected with an anti-HA epitope antibody in cells infected by wt but not the ΔssaV mutant strain (Fig. 2F).These results demonstrate that SseM is translocated into the host cell via the SPI-2 injectisome.
As an independent test of the validity of the mass spectrometry results and to check if SseM of S. Enteritidis (SseM SEN to distinguish it from SseM of S. Typhimurium) also interacts with the same targets, HEK 293 cells were transiently transfected to express GFP-tagged effectors, and cell lysates were subjected to immunoprecipitation before (F) Translocation analysis by confocal microscopy.HeLa cells were infected with bacteria expressing SseM-HA for 5 h, then proteasome inhibitor MG132 was added for another 3 h before fixation.Fixed cells were labeled with antibodies to visualize Salmonella (red) and SseM-HA (green).Scale bar: 5 µm.
Then, to test if SseM translocated from intracellular Salmonella interacts with the same host cell proteins, HeLa cells were infected with bacterial strains for 17.5 h.Infected cells were then lysed, and the lysate proteins were immunoprecipitated with the rabbit anti-SseM antibody and subjected to immunoblotting.β-2-Syntrophin and α-dystrobrevin were co-immunoprecipitated from cells infected with the wt strain or the ΔsseM mutant complemented with plasmid psseM but not the ΔsseM mutant strain (Fig. 3B).Therefore, and importantly, translocated SseM interacts with β-2-syntrophin and α-dystrobrevin in physiological conditions.
To test if β-2-syntrophin mediates interaction between SseM and α-dystrobrevin, we knocked out β-2-syntrophin in HEK293 cells with two different guide RNAs (g361 and g363).Knockout of β-2-syntrophin abolished α-dystrobrevin co-immunoprecipita ted with GFP-SseM (Fig. 3E).This result suggests that SseM interacts with its host cell targets through the interaction between its PBM and the PDZ domain of β-2-syntro phin.In agreement with this hypothesis, predication of interaction between SseM and PDZ domain of β-2-syntrophin with AlphaFold Colab Multimer showed that the RETDV residues of SseM fit in the binding pocket between β-strand B (βB) and α-helix B (αB) of the β-2-syntrophin PDZ domain (Fig. 3F; Fig. S5).Based on our structural predication and the structural data of other PDZ domains and PBMs (32,33), we predicted that residues 125 GLGI 128 or H176 of β-2-syntrophin (highlighted in Fig. S6) are crucial for mediating its interaction with the PBM of SseM.To test this, GFP-tagged β-2-syntrophin or its variants were co-expressed with mCherry-tagged SseM or SseM V103A in β-2-syntrophin knockout HEK293 cells, and cell lysates were subject to GFP-trap immunoprecipitation.Mutating GLGI to 4 As or mutating the substrate-specific residue H to Y or V of β-2-syn trophin abolished its interaction with SseM although the mutants still interacted with α-dystrobrevin (Fig. 3G).Taken together, the data demonstrate that the PBM of SseM and PDZ domain of β-2-syntrophin are essential for the interaction between SseM and β-2-syntrophin.

SseM modulates in vivo virulence and intracellular Salmonella growth
We next assessed the contribution of SseM to Salmonella growth in systemic tissues of mice by competitive index (CI) analysis (34), involving intraperitoneal injection of a mixed inoculum of wt and ΔsseM mutant strains in susceptible mice.At 3 days post-inoculation, the ΔsseM mutant strain significantly outcompeted the wt::Km strain (CI = 1.800 ± 0.558).The ΔsseM mutant strain harboring plasmid psseM failed to outcompete the wt::Km strain (CI = 0.842 ± 0.196), and the CI results were significantly different from that of the ΔsseM mutant strain vs wt::Km strain (Fig. 4), showing that the small fitness difference was SseM-dependent.However, SseM-HA or SseM V103A did not complement the ΔsseM mutant strain in the mixed infection (CI = 2.137 ± 0.979, 1.394 ± 0.253, respectively).In contrast, SseM SEN did complement the ΔsseM mutant strain in the mixed infection (CI = 0.885 ± 0.215).These results demonstrate that SseM modulates the growth of Salmonella during systemic infection, and this phenotype is dependent on the PBM of SseM.
Since splenic macrophages are the major niche of S. Typhimurium (35) and most SPI-2 effectors play a key role in macrophage infection (7,8), we tested if SseM has any impact on Salmonella replication in bone marrow-derived macrophages (BMDMs).The ΔsteC mutant strain was used as a positive control since it displayed an enhanced replication in BMDMs when compared with wt Salmonella (36).As shown in Fig. 4B, the number of intracellular wt bacteria increased 10.04 ± 1.05 times between 2 and 24 h, whereas both the ΔsteC mutant strain and the ΔsseM mutant strain showed significantly higher replication in BMDMs (17.50 ± 1.25-and 15.10 ± 1.32-fold increase, respectively).This result demonstrates that SseM restrains intramacrophage bacterial growth.

DISCUSSION
In this work, we investigated the SPI-2 injectisome effector repertoire of the clinical isolate S. Enteritidis strain D24359 and identified a previously undescribed effector, which we have named SseM.Like Niemann et al. (23), we exploited the property of SPI-2 gatekeeper mutants to hypersecrete effectors into an SPI-2-inducing culture medium, which was then collected for mass spectrometry analysis.While Niemann et al. (23) passed 500 mL of culture supernatant through a column containing solid-phase extraction resin to prepare samples for mass spectrometry analysis, we only needed to concentrate 50 mL of culture supernatant using a centrifugal filter and fractionated the concentrated samples by SDS-PAGE to prepare samples for mass spectrometry analysis.
Our approach therefore provides an easy and cheap method to prepare multiple samples for investigating the SPI-2 effector repertoire from other serovars of S. enterica.Eighteen effectors were identified in both our study and that of Niemann et al. (23), and further five unique effectors were found in each study.The genes of certain effectors like steE and sspH1 are not present in D24359.The congruity between these two studies suggests that most, if not all, of the SPI-2 repertoire has been identified for these two strains.However, it remains possible that some effectors might be expressed and secreted in low amounts or subject to regulatory control that is absent from in vitro growth conditions and still await discovery.
Our previous analysis revealed that all serovars of S. enterica subspecies enterica have a set of "core" effectors (SseF, SseG, PipB, SteA, SifA, SteD, and PipB2) (7).Here, we showed that the newly identified effector SseM is not only present in all serovars of S. enterica subspecies enterica but is also present in all other four subspecies of S. enterica-hence, we conclude that SseM is an eighth "core" effector.
Although SseM is annotated as a 110-residue hypothetical protein in most Salmonella databases (Fig. S1), we showed experimentally that sseM encodes a protein of 103 amino acids, which is translocated by the SPI-2 injectisome and under the control of the pipB2 promoter and the two-component regulatory system SsrAB [Fig.2B (20)].This is supported by RNAseq data, which only revealed transcription start sites before the pipB2 gene (27), leading us to conclude that SseM and PipB2 are encoded in the same operon.
SseM, when expressed in isolation in human cells or after translocation by intracellu lar Salmonella, interacted specifically with components of DAPC signalosome (29,30): β-2-syntrophin, utrophin/dystrophin, α-catulin, α-dystrobrevin, and β-dystrobrevin.Of particular interest, we identified a PDZ-binding motif within SseM and found that both this and the PDZ domain of β-2-syntrophin were required to mediate the interaction between SseM and components of the DAPC signalosome.Both DAPC and DLG1 are involved in several key cellular functions that include not only cell signaling from the adrenergic receptor but also the regulation of the cell's cortical cytoskeleton, cell migration, and formation of focal adhesions (30,37,38), as well as both DLG1 and DAPC regulating tight junctions of polarized epithelial cells (39,40).We hypothesize that via its PBM, SseM interferes with one or more of these processes.To our knowledge, SseM is unique among Salmonella effectors in containing a functional PBM and as a bacterial protein targeting the DAPC.Several viral oncoproteins target DLG1 to regulate viral virulence (41,42).It is, therefore, now essential to investigate the biochemical consequences and physiological significance of SseM's interaction with DLG1 and DAPC components.
There are several examples of bacterial effectors whose function is mediated through short linear motifs that mediate protein-protein interactions.These include three other PBM-containing effectors (Map, OspE, and NleG8) characterized in enteropathogenic Escherichia coli (43,44), Shigella flexneri (45), Citrobacter rodentium, and enterohemorrha gic Escherichia coli, respectively (46), with each effector/PBM sequence required for the virulence of the corresponding pathogens (43)(44)(45)(46).We found that the ΔsseM mutant strain slightly outcompeted the wt strain in the mouse systemic infection model.This modulation of Salmonella growth was dependent on the functional PBM of SseM, suggesting that an interaction between SseM and the DAPC acts to restrain bacterial replication during growth in infected tissues.Whereas the precise mechanism governing how SseM and DAPC function to regulate the outcome of Salmonella infection requires follow-up work, it is intriguing to note that DAPC regulates the cytoskeletal network (38) and SteC targets MEK/ERK and formin-like proteins (FMNL1/2/3) to polymerize actin (36,47), and both effectors restrain Salmonella replication in BMDMs.Further work should, therefore, test whether they restrict Salmonella replication through the modulation of the host cytoskeleton or other cellular functions.AvrA ( 48) is another effector whose absence leads to a slight growth advantage of Salmonella.The fact that several such mutants exist points to an important aspect of bacterial virulence that remains to be understood.

Plasmids
Complementing plasmids were constructed by ligating HindIII and PstI-digested plasmid pssaGpr (Ap r ) (6), a pWSK29 (51) derivative containing the DNA sequence of ssaG promoter, with the corresponding digested PCR products: psseM, psseM ACG , psseM-HA, and psseM V103A by using S. Typhimurium 12023 genomic DNA as PCR template, and psseM SEN by using S. Enteritidis D24359 genomic DNA as PCR template.

Preparation of secreted proteins for mass spectrometry analysis and immunoblotting
Bacteria were grown overnight in 5 mL of LB broth.One milliliter culture was pelleted, washed once with MgM-MES at pH 5.0, and subcultured into 50 mL of MgM-MES at pH 5.0 prior to 6-h incubation at 37°C, 200 rpm.Bacteria were pelleted at 10,000 × g for 10 min at 4°C, the supernatant was filtered through a ϕ0.2 μm membrane (Acrodisc Syringe Filter, 0.2 µm Supor Membrane, low protein binding, non-pyrogenic, PALL Life Science) followed by concentration to approximately 200 µL on an Amicon Ultra-15 Centrifugal Filter with Ultracel-3k membrane (UFC9003, Millipore) at 4°C.Fifty microli ters of concentrated supernatant was run approximately 1 cm into a 12% SDS-PAGE separating gel.The 1 cm gel slice stained with PageBlue Protein Staining Solution (Thermo Fisher Scientific) was sent for mass spectrometry analysis at the Institute of Biochemistry and Biophysics at the Polish Academy of Sciences, Warsaw, Poland.Acquired spectra were compared to our annotated S. Enteritidis D24359 sequence using the MASCOT search engine.
For pH shift analysis, the subculture was grown for 4 h at pH 5.0 and switched to MgM-MES at pH 7.2 for another 1.5 h.The whole bacterial lysate and secreted fraction were prepared as described previously (5) to make 10 µL of whole bacterial lysate equal to 0.1 OD 600 of culture and 10 µL of secreted fraction equal to 0.6 OD 600 of culture.Ten microliters of each sample was used for immunoblotting.Antibodies used in this study are listed in Table S3.

Bioinformatic analysis
Sequencing data for S. Enteritidis strain D24359 have been published previously (ENA accession: ERR037572); however, no genome assembly or annotation was published.To this end, we have downloaded the reads and evaluated their quality using FastQC v0.11.6.The reads were determined to be quality-and adapter-trimmed.Following this, short read assembly was performed using Unicycler v0.4.5.The resulting assembly had 668 contigs and an N 50 of 10,609.To improve the annotation, we have applied Ragout v2.0 with four reference-quality Enteritidis genomes (A1636: GCF_015241115.1, CP255: GCF_015240995.1, D7795: GCF_015240855.1, and P125109: GCA_015240635.1).This resulted in a much more contiguous assembly (two contigs, N 50 4,705,460) with 200 kb (~5%) of the assembly represented as N's because of the ambiguity in the syntenic blocks.The resulting assembly was annotated using Prokka v1.12 against a custom Salmonella protein database that contained 234,913 unique Salmonella proteins annotated using RefSeq Identical Protein Groups.The produced annotation contained 4,448 putative protein-coding genes.The predicted proteins were used as a reference during the mass spectrometry analysis.The code and files necessary to reproduce the assembly and annotation are available at the repository https://github.com/apredeus/D24359.
Protein BLAST was used to search the presence of D24359_01053 in S. bongori, S. enterica subspecies salamae, arizonae, houtenae, indica, and several common serovars of S. enterica subspecies enterica."Identical Proteins" in other S. enterica serovars were identified, and the protein sequences were aligned with Clustal Omega (https:// www.ebi.ac.uk/Tools/msa/clustalo/).
To compare the different SseM protein sequence types among Salmonella serovars, the complete Salmonella genomes were downloaded from Enterobase by searching "Complete Genome" in the "Status" field, which represents the highest assembly quality with circular chromosomes and plasmids (https://enterobase.warwick.ac.uk/, accessed on 30 June 2023).The SISTR1 results from Enterobase were used to identify the subspecies and serovars of the genomes.Only 834 genomes that belong to Salmonella enterica subspecies I were included in the analysis.
To visualize the SseM types in conjunction with the genomic diversity of the Salmonella genomes, an MStree of the 834 complete Salmonella genomes was generated on Enterobase using the cgMLST scheme with the MSTree2 algorithm (56).The tree was visualized with Grapetree (26).In the cgMLST scheme, clusters of genomes with fewer than 900 allele differences are uniform for serovars (56).Therefore, in Grapetree, the nodes with fewer than 900 allele differences are collapsed into bubbles to visualize the serovars.
For translocation assays, HeLa cells were infected for 5 h, then proteasome inhibitor MG132 (Sigma) was added to a final concentration of 10 µg/mL, and cells were incubated for another 3 h.For immunoblotting analysis, cells were collected, washed once with cold PBS, and lysed for 15 min on ice with 50 µL of 0.1% Triton X-100 in PBS.The soluble fraction (containing translocated effectors) was separated from the insoluble fraction (containing bacteria and nucleus) by centrifugation at 16,000 × g for 10 min at 4°C.The pellet was resuspended into 62.5 µL of 1× protein loading buffer, and 40 µL of supernatant was mixed with 10 µL of 5× protein loading buffer.Ten microliters of each sample was used for immunoblotting.For Fig. 2E, the translocation of SseM was quantified using BioRad Image Lab version 6.1 software, and the fold change reported represents the mean ± standard deviation of three biological repeats.Alternatively, cells on glass coverslips were infected as above, fixed, immuno-labeled, and analyzed with a Zeiss 710 confocal microscope as described (15).
To immuno-precipitate SseM, HeLa cells seeded in a ϕ15 cm petri dish were infected for 14.5 h, then MG132 was added to a final concentration of 10 µg/mL, and cells were incubated for another 3 h.After a PBS wash, cells were resuspended into 800 µL of buffer A (50 mM Tris [pH 7.5], 150 mM NaCl, 0.5% sodium deoxycholate, 1% Triton X-100, 1 mM EDTA, and cOmplete, Mini, EDTA-free Protease Inhibitor Cocktail [Roche]) and lysed for 30 min on ice.The lysate was centrifuged at 16,000 × g for 10 min at 4°C, supernatant was incubated with 40 µL of Protein G Agarose (Pierce) on a roller at 4°C.The pre-cleaned supernatant was then incubated with 30 µL of Protein G Agarose pre-bound with 50 µL of rabbit anti-SseM antibody for 3.5 h on the roller at 4°C.The agarose was washed four times with buffer A and resuspended into 30 µL of 2× protein loading buffer.
Primary BMDMs were prepared (58) and infected (36) with Salmonella strains to enumerate intracellular bacteria by plating with minor modification.After incubating 25 min with opsonized bacteria, extracellular bacteria were killed with DMEM containing 30 µg mL −1 gentamicin for 1 h, then the concentration of gentamicin was decreased to 10 µg mL −1 for the remainder of the experiment.

Generation of stable cell lines and SNTB2 knockout cell lines
Lipofectamine 2000 (Invitrogen) was used to transfect HeLa cells with plasmid.Transfected cells were selected with blasticidin to establish stable HeLa cell lines expressing GFP or GFP::SseM.
One ϕ10 cm dish of HeLa stable cell line expressing GFP or GFP::SseM or one ϕ10 cm dish of HEK293 cells transiently transfected with 3 µg plasmid DNA pflag::sseM or pflag::spvD for 16 h were used for immunoprecipitation.After four washes with buffer C, the beads were washed twice with PBS before sending for mass spectrometry analysis.Acquired spectra were compared to the database of Homo sapiens (Uniprot) using the MASCOT search engine.
For immunoblotting analysis, HEK293 cells seeded in one well of a 6-well plate were transfected with 1.5 µg of plasmid DNA or co-transfected with 0.75 µg of each plasmid for 16 h before collecting cells for GFP-trap immunoprecipitation.After four washes with buffer C, the beads were then resuspended into 30 mL of 2× protein loading buffer.

Mouse mixed infection
The virulence of S. Typhimurium strain 12023 derivative wt::Km strain is indistinguishable from wild-type S. Typhimurium strain 12023 (J.Poh and D. W. Holden, unpublished data) and was used as wt strain for CI studies.Female BALB/c mice (7-8 weeks) were inoculated intraperitoneally with a mixture of two strains comprising 500 colony-forming units of each strain in PBS, and the CIs were determined from spleen homogenates 72 h post-inoculation as described previously (34).
Single sample t-test was used to compare the log10 CI to the hypothetical value of 0 (the value of 0 means that two strains grew equally well in vivo).One-way ANOVA corrected by Dunnett's multiple comparison test was used to compare the log10 CI to that of the ΔsseM pWSK29/wt::Km pWSK29 group.

FIG 1
FIG 1 Identification of SPI-2 effector SseM and conservation in Salmonella enterica subspecies I genomes.(A) Schematic for the identification of novel Salmonella secreted proteins.(B) Matched peptides (red fonts) of newly identified effector, D24359_01053, from one MS analysis.(C and D) GrapeTree phylogenetic visualization of SseM distribution and protein sequence variation; branch length indicates the number of allele differences between the cgMLST types, as shown by the scale bar.The nodes with fewer than 900 allele differences were collapsed into bubbles, which are consistent with serovars.The size of each bubble is proportional to the number of genomes it represents.The bubbles that correspond to Salmonella serovars Typhi, Typhimurium, and Enteritidis are labeled.The color of the bubbles indicates the serovars (C) or the diverse SseM protein sequence types (D).

FIG 2
FIG 2 Analysis of expression, secretion, and translocation of SseM.(A) Genetic organization of pipB2-sseM operon and regions of deletion used in panel B are indicated with ∇.The dash-dotted arrow indicates the transcript of pipB2-sseM.(B) Expression of SseM is controlled by the pipB2 promoter and SsrAB.Bacterial strains were grown in MgM-MES at pH 5.0 for 6 h.Bacterial lysates were analyzed by immunoblotting.Intrabacterial protein DnaK and SPI-2 translocon protein SseB were used as controls.(C) Indicated bacterial lysates, including strains expressing the mutation of the predicted start codon ATG to ACG (psseM ACG ), were analyzed by immunoblotting.(D) Secretion of SseM upon pH shift.Bacterial strains were grown at pH 5.0 for 4 h, and then switched the pH to 7.2 for another 1.5 h before preparing samples for immunoblotting.(E) Translocation analysis by immunoblotting.HeLa cells were infected for 5 h, and then proteasome inhibitor MG132 was added for another 3 h before fractionation with Triton-X 100.Triton-X 100 soluble fraction (PNS) contains translocated proteins.

FIG 4
FIG4 SseM is required to downregulate Salmonella growth in mice and bone marrow-derived macrophages (BMDMs).(A) CI analysis.BALB/c mice were inoculated by intraperitoneal injection with equal numbers (500 cfu of each of the two strains) of the indicated bacteria.Bacteria were recovered from infected spleens 72 h post-inoculation, and CI values were calculated.The log10 CIs were used for statistical analysis.Single sample t-test was used to compare the log10 CI to the hypothetical value of 0, and P value is indicated in the round bracket, one-way ANOVA followed by Dunnett's multiple comparison test was used to compare with the ΔsseM pWSK29/wt::Km pWSK29 group (ns, not significant; **P < 0.01).(B) Replication assay in primary BMDMs.BMDMs were infected with opsonized stationary phase bacteria at MOI = 10.At 2 and 24 h post-uptake, BMDMs were lysed and plated on LB agar for the enumeration of intracellular bacteria.Values of fold increase were calculated as a ratio of the intracellular bacteria between 24 and 2 h post-uptake.Results represent mean fold increase ± SEM of three independent experiments.Data were analyzed with a one-way ANOVA followed by Dunnett's multiple comparison test (ns, not significant; *P < 0.05; and **P < 0.01).

TABLE 1
Identified secreted proteins with a ratio of ion scores (ΔspiC/ΔspiCssaC) > 2 a

Protein name Accession Ion score (number of significant distinct peptides) Expt. 1 Expt. 2 ΔspiC ΔspiCssaC Ratio ΔspiC ΔspiCssaC Ratio
a The data shown are based on the following criteria: (i) If the identified protein is detected in both experiments, both (ΔspiC/ΔspiCssaC) ratios must be >2, ion score > 100, and at least two peptides detected in one of the experiments.(ii) If the protein is detected in only one experiment, the ratio must be >2, ion score > 250, and at least two peptides detected.b SsaI is a Salmonella SPI-2 injectisome rod protein.c SlyB is an outer membrane lipoprotein.d D24359_00650: YgiW/YdeI family stress tolerance OB fold protein.