A putative type IVB secretion system contributes to the pathogenicity of Yersinia pseudotuberculosis lineage 8 strains associated with the Far East scarlet-like fever

Abstract

Yersinia pseudotuberculosis is a food-borne pathogen responsible for a self-limiting mesenteric lymphadenitis. Yersinia pseudotuberculosis lineages 1 and 8 are also associated to the Far East scarlet-like fever (FESLF) syndrome, characterized by skin rash, hyperemic tongue, and desquamation. Genome sequencing of lineage 8 FESLF strains previously revealed the presence in the plasmid pVM82 of dot/icm genes, homologous to those known to encode a Type 4B Secretion System (T4BSS) in the intracellular pathogens Legionella pneumophila and Coxiella burnetii. We validated the presence of all essential dot/icm genes required for the structure of a T4BSS and we identified acidic pH and 37°C as the main signals triggering yT4BSS expression. Putative yT4BSS effectors were also identified using the S4TE algorithm but their secretion was not confirmed in the studied conditions. The yT4BSS is also expressed in epithelial cells and macrophages as well as in vivo during mouse infection, but it is not required for bacterial intracellular survival. Nevertheless, the yT4BSS is implicated in Y. pseudotuberculosis FESLF pathogenicity in mice when bacteria are orally inoculated. The yT4BSS does not influence organ colonization and is rather implicated in the induction of important necrosis lesions in mesenteric lymph nodes and ceca of mice. Cytokine profile analyses in mouse bone marrow-derived macrophages revealed a yT4BSS-dependent induction of innate immunity related cytokines and chemokines. In conclusion, we identify the yT4BSS as a virulence factor of Y. pseudotuberculosis strains from lineage 8 responsible for FESLF which modulates cytokine responses of the host innate immune system.

A putative type IVB secretion system contributes to the pathogenicity of Yersinia pseudotuberculosis lineage 8 strains.

Introduction

Yersinia pseudotuberculosis is a Gram-negative pathogen responsible for enteric yersiniosis, a food-borne disease characterized by self-limiting mesenteric adenitis, which can evolve to septicemia in patients displaying underlying disorders such as diabetes, hemochromatosis, or thalassemia (Carniel et al. 2006). Infections by Y. pseudotuberculosis are rather rare: in France, the National Reference Laboratory “Plague and other yersiniosis” reports 107 cases of infections due to Y. pseudotuberculosis between 2017 and 2021, most yersiniosis being attributed to another enteropathogen, Y. enterocolitica (Le Guern et al. 2016, Le Guern et al. 2022). A similar scenario is observed at the European level with 100 cases of yersiniosis due to Y. pseudotuberculosis reported in 11 countries in 2022 (ECDC 2024). Nevertheless, Y. pseudotuberculosis outbreaks have been reported in several locations around the world, including Japan (Inoue et al. 1984), Canada (Nowgesic et al. 1999), Finland (Jalava et al. 2004), and New Zealand (Williamson et al. 2016). The first Y. pseudotuberculosis clonal outbreak in France was reported in 2020 (Savin et al. 2022).

Y. pseudotuberculosis displays several virulence factors: the chromosomal inv gene encodes for invasin, a surface protein that allows bacterial binding to β1 integrin receptors predominantly exposed by Microfold (M) cells in intestinal Peyer’s patches (Isberg and Falkow 1985, Marra and Isberg 1997, Clark et al. 1998). This interaction triggers bacterial internalization within host cells and invasion of the intestinal tissue, followed by colonization of draining mesenteric lymph nodes (MLNs). A type III secretion system (T3SS) encoded in the Yersinia virulence plasmid (pYV) is then activated and translocates Yersinia outer proteins (Yops) into the cytoplasm of macrophages and other immune cells, blocking actin polymerization and inhibiting immune signaling responses dependent on mitogen-activated protein kinases, leading to bacterial extracellular proliferation (Cornelis 2002, Seabaugh and Anderson 2024). Host iron is sequestered via the siderophore yersiniabactin, encoded by the high pathogenicity island (HPI) which is common to other enterobacterial species such as Escherichia coli or Citrobacter jejuni (Carniel 2001, Rakin et al. 2012).

Y. pseudotuberculosis strains present in Western Europe and North America are associated with classical enteric yersiniosis (Fukushima et al. 2001). In Eastern Russia and Japan, a disease known as the Far East scarlet-like fever (FESLF) has been associated with a restricted subset of Y. pseudotuberculosis strains which have been the cause of two major outbreaks reported in 1958 in Vladivostok, Russia with more than 300 cases and in 1977 in Kanazawa, Japan (Fukushima et al. 2001, Tseneva et al. 2012, Amphlett 2015). Besides digestive symptoms, FESLF is a longer infection than classical enteric yersiniosis, characterized by fever, skin rash, hyperemic tongue, ending with desquamation (Mollaret et al. 1990, Amphlett et al. 2015). The presence of a superantigen, the Y. pseudotuberculosis-derived mitogen (YPMa), has been reported in Far Eastern Y. pseudotuberculosis strains and this virulence factor might be related to the systemic symptoms detected in FESLF patients (Abe et al. 1997, Ueshiba et al. 1998, Fukushima et al. 2001, Abe et al. 2003). The genome sequencing of a Y. pseudotuberculosis FESLF strain identified the presence of a plasmid (pVM82), which had been previously associated to bacterial pathogenicity and to modified cellular immune responses in infected hosts (Gintsburg et al. 1988, Sever et al. 1991, Dubrovina et al. 1999). Interestingly, in the pVM82, Eppinger et al. (2007) identified dot/icm (defect in organelle trafficking, intracellular multiplication) genes encoding a putative type IVB secretion system (T4BSS).

T4BSS are typically represented by the effector translocator Dot/Icm system from intracellular pathogens such as Legionella pneumophila and Coxiella burnetii (Berger and Isberg 1993, Brand et al. 1994; Segal and Shuman 1997). However, T4BSS also include conjugative system members from the plasmid incompatibility group I such as the R64 plasmid from Salmonella enterica serovar Typhimurium and the ColIb-P9 plasmid from Shigella flexneri (Hartskeerl et al. 1985, Sampei et al. 2010), as well as a recently described bacterial killing system from Pseudomonas putida (Purtschert-Montenegro et al. 2022). The T4BSS of L. pneumophila and C. burnetii is an essential virulence determinant for their intracellular life cycle, secreting a wide range of effectors involved in the manipulation and disruption of cellular organelles, host metabolism, cell death, and immune responses (Lührmann et al. 2017, Qiu and Luo 2017). The Dot/Icm T4BSS system from L. pneumophila has the best characterized structure to date and consists of a transport channel that spans the bacterial envelope composed of an outer membrane complex core (OMCC: DotC, DotD, and DotH), a central cylinder channel connecting the outer and inner membranes (DotG, DotF, IcmX, IcmF, and DotA), inner membrane proteins (DotJ, IcmT, IcmV, DotI, DotU, and DotE), and a cytoplasmic complex of ATPases (DotO, DotB) (Chetrit et al. 2018, Ghosal et al. 2019, Macé et al. 2022a). A type IV coupling complex (T4CC) acts as a substrate delivery platform for the transport channel, and it is composed of an ATPase (DotL) and adaptor proteins (DotM, DotN, DotY, and DotZ) with chaperones (IcmS, IcmW, and LvgA) (Macé et al. 2022b).

Here, we studied the T4BSS in a Y. pseudotuberculosis lineage 8 FESLF strain (yT4BSS). In silico, we identified the presence of full yT4BSS gene cluster, and we validated yT4BSS gene expression in vitro, in cellulo, and in vivo, observing a higher expression rate at acidic pH and 37°C, mimicking the mammalian host environment. Effectors were predicted using the S4TE bioinformatic tool but their translocation could not be confirmed in vitro. We studied the function of the yT4BSS and while we did not detect its implication in the intracellular life cycle of Y. pseudotuberculosis in epithelial and macrophage-like cells, we rather found it to be a virulence factor involved in bacterial pathogenicity in orally infected mice, modulating cytokine innate immune responses and being associated to necrotic lesions in infected organs.

Materials and methods

Bacterial strains

Bacterial strains and plasmids used in this study are listed in Supplementary Tables S1 and S2. Y. pseudotuberculosis and E. coli strains were grown in lysogeny broth (LB) medium (Difco) under stirring or on LB agar plates at 28°C for 48 h or 37°C for 24 h, respectively. Listeria monocytogenes was grown in brain heart infusion liquid medium or agar plates at 37°C for 24 h. When required, culture medium was supplemented with antibiotics (irgasan 0.1 µg/mL, carbenicillin 100 µg/mL, kanamycin 30 µg/mL, chloramphenicol 25 µg/mL, and erythromycin 25 µg/mL).

Search for T4BSS genes, oriT and relaxase

The presence of T4SS genes and relaxase was assessed on multiple pVM82 sequences from Y. pseudotuberculosis FESLF strains using MacsyFinder v2.07rc (Abby et al. 2014, Néron et al. 2023) with the “Plasmids” models of the CONJScan module. Briefly, MacSyFinder uses HMM protein profiles and a set of rules (which are defined in the models) about their presence and proximity. Macsyfinder was used with default parameters and CONJScan models were ran all at once with the parameters “-all”. A complementary approach was used with oriTfinder, allowing to search for an origin of transfer (oriT) on the pVM82 plasmid (Li et al. 2018).

Phylogenetic reconstruction and presence of virulence factors

In a total of 287 genomes of Y. pseudotuberculosis (including 20 Y. pestis), the nucleotidic sequences of the 500 genes of the cgMLST-Yersinia (Noroy et al. 2019) were extracted and aligned using MAFFTv7.467 (Katoh and Standley 2013). From this alignment, a phylogenetic reconstruction was performed with IQ-TREE2 (Minh et al. 2020) using a bootstrap value of 1000 and by first determining the best substitution model thanks to the “ModelFinder” implemented in IQ-TREE2. ABRIcate v.1.0.1 (github.com/tseemann/abricate) was used to search for the presence of the different coding sequences of the yT4BSS together with the different versions of the ypm gene in the 2015 genomes. For this purpose, an in-house database containing the sequences of the 18 genes was created for ABRIcate. Default parameters of 80% identity and 80% coverage were used to identify the presence of the genes. Visualization of the phylogenetic reconstruction together with the presence of the virulence genes was generated using phandango (Hadfield et al. 2018). The genotypes of Y. pseudotuberculosis were plotted on the tree as described in (Noroy et al. 2019).

yT4BSS Gene cluster conservation comparison

Homologous gene cluster searches for the yT4BSS operon of the strain Y. pseudotuberculosis SP-1303 (GenBank accession number: NZ_CP130902.1) was performed on the Yersinia spp database of NCBI by using cblaster pipeline in default mode (minimum identity 30%, minimum query coverage 50%). NCBI identifiers used were: WLF06076.1 (IcmX1), WLF06077.1 (IcmX2), WLF06128.1 (DotA), WLF06119.1 (DotM/IcmP), WLF06113.1 (DotB), WLF06112.1 (DotC), WLF06111.1 (DotD), WLF06108.1 (DotI/IcmL), WLF06107.1 (DotH/IcmK), WLF06106.1 (DotG/IcmE), WLF06105.1 (DotF/IcmG), WLF06104.1 (DotE/IcmC), WLF06102.1 (DotN/IcmJ), WLF06100.1 (DotO/IcmB), and WLF06099.1 (DotL/IcmO), as well as amino acid sequence of IcmT newly described (Gilchrist et al. 2021). The gene cluster conservation comparison of the yT4BSS of Y. pseudotuberculosis SP-1303 and other T4BSS members was performed by using clinker/0.0.23 pipeline with default parameters (minimum alignment sequence identity of 0.3) by giving Genbank files of the selected regions of the IncI plasmid R64 (NC_005 014) and of the following strains: C. burnetii RSA 493 (NC_002 971), L. pneumophila Philadelphia 1 (NC_002 942), and P. putida IsoF (CP072013.1). Clinker performs pairwise local or global alignments between every sequence in every unique pair of clusters (Gilchrist and Chooi 2021).

Synteny analysis and representation

The IP31758 strain was used as reference strain and aligned with 45 strains selected for their association with FESLF and presence of the pYV plasmid. Analysis was performed as described in (Lê-Bury et al. 2023). Briefly, a protein BLAST search was automatically performed for each gene of each genome against the other genomes to construct the homolog database, using the BLAST1 command-line tool v2.13.0 (Camacho et al. 2009). Synteny was constructed using a best-hit bidirectional BLAST search and implemented in the SynTView framework (Lechat et al. 2013). The tool was reprogrammed from Flash to JavaScript with the HTML5 2d Canvas JavaScript library Konva (konvajs.org), which enables high-performance animations. Drawing strategies were developed to ensure that graphical outputs remain below the technological limits of drawing more than 10 000 dynamic graphical objects, using a mix of static and dynamic objects. SynTViewJS can be used on a public web application at plechat.pages.pasteur.fr. The code repository is at gitlab.pasteur.fr/plechat/syntviewjs.

Generation of a ΔdotB mutant and complemented strains

SP-1303 ΔdotB mutant was generated by allelic replacement of dotB (dotB_RS21065 from start to stop codons) by homologous recombination with a linear three-step polymerase chain reaction (PCR) fragment containing a kanamycin resistant cassette, flanked with 500 bp upstream and downstream regions of dotB. The detailed method followed is described in (Derbise et al. 2003). Primers used are listed in Supplementary Table S3. The SP-1303 ΔdotB:: dotB complemented strain was generated by allelic replacement of the kanamycin resistant cassette of the ΔdotB mutant with the original dotB gene cloned in the pCVD442 plasmid. For that, the entire dotB sequence with 500 bp flanked regions was PCR amplified from a SP-1303 genomic DNA extract using a high-fidelity polymerase (Phusion, NEB). Purified amplicons were cloned at SalI and SphI sites (NEB enzymes) into the pCVD442 plasmid and electroporated in E. coli SM10 donor bacteria. Mating was performed between E. coli SM10 pCVD442-up-dotB-down and the SP-1303 ΔdotB mutant by mixing equivalent volumes of adjusted overnight (O/N) liquid culture to OD₆₀₀ = 2 on a 0.45 µm filter (peel corner, Merck) positioned on a LB agar plate for 6 h at 28°C. After mating, bacteria were collected, serially diluted, and plated on LB-Irg-Km-Amp agar for 48 h. Resistant clones were cultured O/N in liquid LB and spread on LB agar 10% sucrose without NaCl for 48 h at 28°C, for resolution of the merodiploid previously formed. The complemented strain was checked for its sensitivity to kanamycin and ampicillin. The complete genome of the ΔdotB mutant and its complemented strain ΔdotB::dotB were extracted using the PureLink Genomic DNA Kit (Invitrogen) and checked for mutations using two complementary sequencing technologies: Oxford Nanopore Libraires (long reads) and Illumina (short reads). Libraries were prepared with the Nextera Kit for Illumina NextSeq 550 sequencing by the Plateforme de microbiologie mutualisée (P2M) of Institut Pasteur and by using the Rapid Barcoding kit (SQK-RBK004) for R9.4.2 ONT MinION flow cell sequencing.

Generation of bioluminescent reporters

Constitutively bioluminescent strains: SP-1303::Tn7-PrplN-lux and SP-1303 ΔdotB::Tn7-PrplN-lux were generated by insertion into the unique Yersinia-chromosomic attTn7 site of the luxCDABE operon from Photorhabdus luminescens, under the control of the constitutive promoter of rplN gene, by co-electroporated plasmids pUC18R6KT-mini-Tn7-Km^R-PrplN-luxCDABE and pTNS2 as described in (Derbise et al. 2019). For the generation of transcriptional bioluminescent reporters, the promoter regions of dotA (490 bp upstream dotA_RS21140 and dotDCB (intergenic region of 921 bp between dotD_RS21055 and RS21050 genes) were PCR-amplified using a high-fidelity polymerase (Phusion, NEB) and PCR products were purified before cloning (QIAquick PCR purification Kit, QIAGEN). Amplicons were cloned in front of the luxCDABE operon in the plasmid pUC18R6KT-mini-Tn7-Km^R-luxCDABE into the SpeI and XmaI restriction sites (high fidelity enzymes, NEB). Constructs were verified for mutations by sequencing, using primers listed in Supplementary Table S3. The kanamycin resistant cassette flanked by FRT regions was removed in all constructions by using a recombinase-encoding plasmid pFLP3, which was then cured as detailed in (Derbise et al. 2019).

Bioluminescence monitoring

In vitro

Reporter strains were grown O/N in LB, pH 7 media at 28°C and diluted in LB (50 mL) to reach OD₆₀₀= 0.1 prior to culture at a tested temperatures 21°C, 28°C, 37°C, and 42°C and pH 7, 6, and 5. Minimal medium (supplemented with casaminoacids) at 37°C and pH 7 was also tested. OD_600nm and photon emission were measured every 2 h, using a spectrophotometer and a Glomax Discover microplate reader (Promega) respectively as described in (Li et al. 2018). Photon emission in relative light units (RLU) was subtracted from Pless measures and corrected by PrplN values.

In cellulo

HeLa cells were seeded the day before infection at 10⁴ cells per well in 96-well plates. Reporter strains grown O/N at 28°C were adjusted to a multiplicity of infection (MOI) of 70 in cell culture media. Infected plates were centrifuged (188 g × 5 min) to synchronize bacterial invasion and were incubated at 37°C for 1 h before gentamicin addition (20 µg/mL) for the remaining time: 1, 3, 7, or 23h. Cell culture media was replaced by phosphate-buffered saline (PBS) prior to photon emission measurement using a Glomax Discover microplate reader (Promega). Cells were lysed in distilled water for intracellular bacterial counts on LB agar as described in (Kühbacher et al. 2014). Photon measurements in RLU were corrected by CFUs counts per well.

In vivo

Real-time bioluminescence images were acquired on anesthetized mice (constant flow of 2.5% isoflurane mixed with oxygen) by using the In Vivo Imaging System (IVIS 100; Caliper Life Sciences) in automatic mode. To analyze the bioluminescence signal within organs, mice were euthanized and the gastrointestinal tract, a pool of MLNs, and the spleen were isolated. Image analysis was performed using the Living Image 4.5 software (Revvity). Photon emission is comparable between mice by applying a common scale. Total photon emission was quantified by drawing regions of interest (ROIs) around the entire body and given as average radiance (photons/s/cm2/steradian [sr]).

RNA extraction and reverse transcription quantitative PCR

Bacteria were grown in LB (pH 5 or 7 at 21°C, 28°C, or 37°C) in triplicate under stirring conditions until late exponential phase (7–8 h of growth), pelleted and frozen at −80°C. Pellets were resuspended in TRIzol (Invitrogen) at 4°C and processed for lysis in beads-containing tubes (#P000914-LYSK06A, Bertin Instruments) using a cooling homogenizer for two cycles of 30 s (Precellys24 and Cryolys, Bertin Instruments) and centrifuged for 15 min at 12 000 × g at 4°C. The aqueous phase was mixed twice with chloroform, left for 5 min at room temperature and centrifuged. The clear aqueous phase was mixed with isopropanol and left for 5 min at room temperature for RNA precipitation. After centrifugation, the pellet was washed in 70% ethanol, left still for 5 min and transferred to a RNeasy spin column (RNeasy Mini Kit, QIAGEN). The column was washed two times with RPE buffer before elution in RNA-free water. RNA samples were treated with DNase according to the manufacturer’s recommendations (TURBO DNase-free Kit, Ambion) and quantified using a Qubit 4 Fluorometer (Invitrogen). A quality control step on 1% agarose gel with guanidine thiocyanate and ethidium bromide was done. Retrotranscription (RT) was performed thanks to the iScript cDNA synthesis kit following the manufacturer’s instructions (Bio-Rad). Quantitative RT-PCRs (RT-qPCRs) were carried out in triplicate with two to three technical replicates with the SsoAdvanced Universal SYBR Green Kit (Bio-Rad) on a CFX384 Touch Real-Time PCR Detection System (Bio-Rad). The RT-qPCR program used was 98°C for 3 min, 40 cycles of 98°C for 15 s and 60°C for 30 s, then 95°C for 15 s, finishing with a melt curve from 60°C to 95°C with incrementation of 0.5°C. Primers used are listed in Supplementary Table S3. Reference genes nadB, proC, and rpoD were selected according to the following study (Koch et al. 2019). Data were analyzed with the ΔCt method (Livak and Schmittgen 2001).

Cell culture

HeLa cells, Caco-2 cells, and RAW264.7 (TIB-7) macrophages (ATCC) were cultured and maintained in Dulbecco's Modified Eagle Medium (DMEM) supplemented with high glucose (4.5 g/l, Gibco) and 10% fetal bovine serum (FBS, Gibco). Mouse bone marrow-derived macrophages (mBMDM) or human blood-derived macrophages (hBDM) were grown in RPMI, high glucose (Gibco) with 10% FBS and 10 mM Hepes. Addition of 50 µM β-mercaptoethanol and antibiotics (Penicilline/Streptomycin, Gibco) was used for mBMDM and hBDM differentiation steps. Cells were incubated at 37°C, 5% CO₂.

mBMDM and hBDM isolation

Tibias and femurs were dissected from euthanized 8-weeks old BALB/cByJ mice. Collection of bone marrow cells was performed by flushing cell culture media into the bone using a sterile syringe with a 25 G needle. Harvested cells were centrifuged at 300 × g for 5 min at 4°C and resuspended in red blood cell lysis buffer (#130-094-183, Miltenyi Biotec) for 5 min. Cells were centrifuged and resuspended in RPMI, high glucose (4.5 g/l, Gibco) supplemented with 10% FBS (Gibco) with antibiotics (Penicilline/Streptomycin Gibco) and seeded at 1 × 10⁶ cells/mL in a T150 cm² flask (Corning) for incubation at 37°C, 5% CO₂. After 4 h of incubation, non-adherent cells were harvested and removed from attached epithelial cells. Collected cells were seeded at 1 × 10⁶ cells/mL in petri dish (Greiner Bio-one) in cell culture media supplemented with M-CSF (25 µg/mL, mouse M-CSF #130-101-704, Miltenyi Biotec) and antibiotics, and cultured at 37°C, 5% CO₂ for 5 days. Half of the media was changed and replaced by fresh culture media with M-CSF and antibiotics for two additional days of differentiation. Isolation of hBDM were performed as described in Klunk et al. (2025). mBMDM and hBDM were seeded at a concentration of 5 × 10⁵ cells/mL or 10⁵ cells/mL, respectively, the day before the infection in media without M-CSF and antibiotics.

Gentamicin protection assays

The day before the experiment, cells were seeded at 10⁵ cells/mL in 96-well plates and bacteria were grown O/N in LB liquid media from a frozen stock. OD_600nm was measured the next day, and the appropriate MOI was adjusted directly in cell culture media for infection. A centrifugation step at 188 × g for 5 min allowed to synchronize bacterial entry before incubating for 1 h. Cells were then washed once and incubated in cell media with gentamicin (epithelial cells: 20 µg/mL; macrophages: 10 µg/mL) for extracellular bacteria killing. At 1 h post-gentamicin treatment for invasion assays, or at different time points post-treatment for intracellular survival assays (3, 7, and 23 h), cells were washed once in PBS and lysed in distilled water. Intracellular bacteria were serially diluted, plated on LB agar plates, and incubated at 28°C for 48 h.

Immunofluorescence microscopy

Infection assays were performed as described in the previous section, using 24-well plates with coverslips or 96-well plates for direct imaging. At different times post-gentamicin treatment (4 and 23 h), cells were fixed with 4% paraformaldehyde for 15 min, washed with 1% bovine serum albumin (BSA) in PBS, permeabilized in PBS supplemented with 0.1% triton (100X)-1% BSA for 4 min, and washed in 1% BSA-PBS. Primary antibodies (α-O1b, α-LAMP1, α-LC3B, and α-cathepsin D) or secondary antibodies with Hoechst (listed in Supplementary Table S4) all diluted in 1% BSA were added for 30 min labeling at room temperature in two independent steps with washes in between. After washing, coverslips were fixed on a slide (Fluoromont-G, Invitrogen) and analyzed with an epifluorescent microscope (Zeiss Axiovert 200) using the image acquisition software Metamorph (Molecular Devices). Lysotracker was diluted in DMEM/FBS for a final concentration of 75 nM; at 23.5 hpi cell culture media was removed from 96-well plates and replaced with 100 µl of probe-containing medium and incubated for 30 min at 37°C, 5% CO2; media was then removed, washed once with PBS and 100 µL of Fluorobrite media supplemented with gentamicin (40 µg/mL) were added for live cell imaging. 96-well plates were imaged in PBS in the Opera Phenix confocal spinning disk microscope (Perkin Elmer) for LC3B, cathepsin D and LysoTracker stainings using a 40× water immersion objective. Images were treated with ImageJ/Fiji software.

Transmission electron microscopy

HeLa cells were seeded in a 6 well-plate at a density of 2 × 10⁶ cell per well in quadruplicate. Infection was performed as described in the gentamicin protection assay section (see above). At different times post-gentamicin treatment (1, 4, and 23 h), cells were washed with PBS and detached with trypsin (Gibco). The cell suspension was washed once in PBS and fixed with a 4% PFA, 2.5% glutaraldehyde PBS buffer and kept at 4°C before treatment. Samples were then washed in PBS and post-fixed by incubation with 2% osmium tetroxide (Agar Scientific, Stansted, UK) for 1 h. Samples were then fully dehydrated in a graded series of ethanol solutions and propylene oxide. Impregnation step was performed with a mixture of (1:1) propylene oxide/Epon resin (Sigma) and then left O/N in pure resin. Cells were then embedded in Epon resin (Sigma), which was allowed to polymerize for 48 h at 60°C. Ultra-thin sections (90 nm) of these blocks were obtained with a Leica EM UC7 ultramicrotome (Wetzlar, Germany). Sections were stained with 5% uranyl acetate (Agar Scientific) and 5% lead citrate (Sigma), and observations were made with a transmission electron microscope (TEM) (JEOL 1011, Tokyo, Japan).

Mice infection

Institut Pasteur animal housing facilities are accredited by the French Ministry of Agriculture (accreditation B 75 15-01, 05/22/2008), in compliance of the French and European regulations on Laboratory Animals (EC Directive 86/609, French Law 2001-486 issued on 6 June 2001). The research protocol (No. 2013-0038) was approved by Institut Pasteur Ethics Committee for Animal Experimentation (CETEA) for the French Ministry of Research. Female 7-weeks-old BALB/cByJ mice (Charles River France) were acclimatized to the Institut Pasteur animal facility for one week prior to experimentation. For inoculum preparation, bacteria were grown O/N in LB liquid at 28°C and transferred on LB agar plate for 24 h at 28°C. Bacteria on plate were resuspended in PBS and adjusted to OD₆₀₀ = 0.23 to reach a dose of 6 × 10⁶ bacteria per mouse in 20 µL of PBS. Oral infection was performed using the bread feeding method previously described (Derbise et al. 2020). For intravenous (IV) infection, bacterial suspensions were prepared in the same way as described above but serially diluted to reach around 10, 50 or 90 bacteria in 100 µL of PBS. For IV injection, the mouse tail was heated with a paper soaked in warm water prior to injection in the caudal vein with a 29 G needle. Inocula were confirmed by plating serial dilutions or directly the IV inoculum on LB agar plates. After infection, animals were monitored for clinical state and weighed daily for 22 days. All experiments were performed in accordance with the Institut Pasteur’s guidelines for laboratory animal welfare.

Evaluation of the bacterial load in mouse infected organs

For determination of bacterial colonization, organs were collected on euthanized mice at specific days (5 and 7) post-infection (pi) and homogenized in PBS in appropriate tubes depending on the organ. Cecum, spleen and a pool of MLNs were treated in glass beads-containing tubes following two cycles of 30 s with 60 s of break with a homogenizer (Precellys24, Bertin Instruments). Intestines and colons were first opened longitudinally for content collection, then washed thrice in DMEM before being treated for 1 h in DMEM supplemented with gentamicin (40 µg/mL) at room temperature. Livers, intestines, and colons were disrupted using gentleMACS M tubes (Miltenyi Biotec) with the preloaded program “m_heart_02” on gentleMACS Dissociator (Miltenyi Biotec). Intestinal and colon contents were homogenized using disposable homogenizers (Kimble Chase piston pellet, Fisher Scientific). Serial dilutions of organ homogenates were plated on LB agar with irgazan and incubated at 28°C for 48 h. To avoid contamination from cecum, intestinal and colon contents plates were grown for a week at 4°C.

Cytokine measurement

Cytokine concentrations were measured using the mouse magnetic 18-plex Luminex assay (Bio-Techne, R&Dsystems) on supernatants from infected mBMDM cells. The complete list of cytokines and chemokines is listed in Supplementary Table S5. Cell supernatants were collected at 8 hpi, centrifuged at 5000 × g for 5 min at 4°C in SpinX 0,2 µm tubes (Corning, Fisher Scientific) for cell debris and bacterial removal, and stored at −80°C. Samples were processed following the manufacturer’s instructions of the Luminex kit coupled to the DropArray technology (Curiox, Biosystems) which include: the LT washing station MX, DA-bead 96-well, Curiox humid box with Magnet array. Washing steps were performed in 0.1% BSA, 0.05% Tween20 in PBS and the blocking of the DA-bead plate in 10% BSA solution.

Histopathology and immunohistochemistry analyses

Ceca and pools of MLNs were collected from bread-infected mice or non-infected mice (PBS) at 7 days pi, placed in tissue cassettes (Leica Biosystems) and rapidly plunged into formol for 24 h fixation. Samples were put in 70% ethanol solution, fixed in 10% neutral buffered formalin for 14 h, and embedded in paraffin. Paraffin sections (5-µm thick) were processed for routine histology using hematoxylin and eosin (H&E). Stained slides were scanned with ZEISS Axio Scan 7 Digital Slide Scanner and evaluated with Zen software (Zeiss). Histopathological evaluation of H&E samples were blindly examined by two pathologists. In each cecum sample, histopathological criteria summarized in SupplementaryTable S6 were applied. For MLNs, criteria were evaluated as reported in (El-Maraghi and Mair 1979) with findings as follows: lymphoid hyperplasia, apoptosis, histiocytic hyperplasia, granulomas, and necrotizing granulomas. Each finding was scored with degree (0) absence, (1) mild, (2) moderate, and (3) severe. The sum of all criteria represents the total microscopic score. H&E scanned slides were subjected to histomorphometry to calculate the total percentage of necrosis area in each sample by using Nikon NIS-Ar software (Nikon Instruments Inc., NY, USA). In addition, serial sections were prepared for immunohistochemistry (IHC) analyses. After a 20 min at 100°C pH 6 Citrate buffer antigen retrieval, an IHC was performed by using the Bond RX robot (Leica) with rabbit anti-F4/80 antibody (D2S9R, Cell Signaling) or rat anti-Ly6B2 (MCA77, biotinylated G, BioRad), goat anti-rabbit Ig secondary antibody (E0432, Dako, Agilent) or rat IgG-HRP secondary antibody (VC005-025, R&D Systems), and Leica Streptavidin-HRP/DAB detection Kit. IHC stained slides (F4/80 and Ly2B6) were scanned (Hamamatsu S360, Hamamatsu Photonics, Shizuoka, Japan) and subjected to digital image analysis to calculate the percentage of positivity labeling with respect to the whole area of the organ by using Nikon NIS-Ar software (Nikon Instruments Inc., NY, USA).

Bioinformatic yT4BSS effectors prediction

Candidate yT4BSS effectors were predicted with the Searching Algorithm for Type IV Effector proteins (S4TE v2.0) using the SP-1303 genome (chromosome and pVM82) (Noroy et al. 2019). Additional specific proteins domains and protein families features were obtained by using the Interpro and UniProt databases (Paysan-Lafosse et al. 2023). Protein sequence homologies were researched using BLASTp.

β-lactamase and effector cloning

Selected putative effectors and the β-lactamase gene sequences were PCR amplified using primers listed in SupplementaryTable S3, following manufacturer’s recommandations of Phusion High Fidelity or Taq DNA polymerases (Thermo Scientific). The pXDC61 plasmid was used as a template for β-lactamase sequence amplification. Positive PCR products were combined and purified using the QIAquick PCR Purification Kit (Qiagen). Appropriate restriction enzymes (EcoRI-HF, SacI-HF, KpnI-HF ,or SalI-HF, NEB) were used for 1 h digestion at 37°C in rCutSmart 1X Buffer (NEB). Digestion was stopped by 20 min heat at 65°C. Ligation of digested fragments was carried out using T4 DNA ligase and its provided buffer (Thermo Scientific) in a 15 µL reaction for 1 h at 37°C, in which the vector:insert ratio was adjusted according to their DNA length and mass following NEBioCalculator v1.15.3. Ligation reaction was heat inactivated at 70°C for 5 min. Plasmids were checked for Sanger sequencing thanks to Microsynth Company (Balgach, Switzerland) with forward primer from each insert as well as standard pBAD-F and pBAD-R primers offered by the company.

Extracellular secretion screening using nitrocefin

Extracellular secretion was assessed under inducible conditions (0.1% arabinose) in different tested temperatures (28°C or 37°C), media (LB, LB low calcium or minimal media), and pH (6 or 7). Supernatants were collected from 1 mL centrifuged cultures, and processed for a second centrifugation step to clear potential residual bacteria. Serial dilutions were made to obtain values within the absorbance reading limits. β-lactamase activity on nitrocefin (1 mM, APExBIO) was measured after 30 min incubation, at 490 nm using the GlowMax plate reader (Promega). Absorbances were normalized with OD₆₀₀ values of the O/N culture.

Translocation assay within HeLa cells

O/N bacterial culture grown in inducible conditions (0.1% arabinose or 1 mM IPTG), was pelleted (6000 rpm, 2 min), washed three times and resuspended in PBS for inoculum preparation in cell culture media supplemented with 1% FBS and inducible compound at a MOI of 10. Inoculum numeration was carried out from serial dilutions and plated for CFUs count. Cell media were replaced by the prepared inoculum in triplicate, followed by centrifugation (188 × g for 5 min) allowed to synchronize bacterial entry before incubating for 1 h. Cells were then washed once and incubated in cell media with gentamicin (20 µg/mL) for extracellular bacteria killing and left for incubation (2, 6, and 24 h). CCF4-AM loading solution (1 mM, Invitrogen) was prepared in Solution B (100 mg/mL Pluronic-F127 surfactant in DMSO and 0.1% acetic acid, Invitrogen), EM Buffer 1X (NaCl 120 mM, KCl 7 mM, CaCl2. 1.8 mM, MgCl2 0.8 mM, Glucose 5 mM, and HEPES Buffer 25 mM), and Probenecid 0.1 M (Sigma–Aldrich). Following each incubation period, CCF4-AM loading solution (20 µL) was added to each well and incubated at room temperature for 2 h protected from light. After incubation, cells were washed once with PBS and detached using trypsin (30 µL/well) for 3 min at 37°C. PBS (170 µL) with 2% FBS and 1 mM EDTA were added to each well for cells detachment and transfer to round bottom 96-well plates for flow cytometry lecture using the MACSQuant VYB flow cytometer (Miltenyi Biotec) with 150 µL volume taken at high speed with gentle mix before each read.

Flow cytometry data analysis

FACS data were analyzed using FlowJo v.10.8.2 software. If similar, triplicate data sets from individual wells were pooled for the analysis. Side scatter (SSC) and forward scatter (FSC) were plotted to select live cells. From lived cells, VioGreen-A and CFP-Blue-A axis were plotted to determine the proportion of positive green or blue or negative cell populations.

Statistical analyses

Statistical analyses were performed using GraphPad Prism (GraphPad Prism 10.2.2 Software). Statistical significance is expressed as follow: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Survival data were plotted using the Kaplan–Meier estimator and log-rank (Mantel–Cox) tests were used to assess differences in survival between groups. Cytokine concentrations were transformed in log(Y+1) before computed for a two-way analysis of variance (ANOVA) or mixed model using Prism.

Results

Distribution, organization, and genomic features of the yT4BSS gene cluster

The genome sequencing of the FESLF-causing Y. pseudotuberculosis reference strain IP31758 (Eppinger et al. 2007) identified the presence of a yT4BSS in the pVM82 plasmid, and this was confirmed using the bioinformatic tool Macsyfinder (Abby et al. 2014, Néron et al. 2023) which detected all the key components of a T4BSS. The Eppinger et al. study also revealed in the strain IP31758, the absence (probably lost in the laboratory environment) of the pYV plasmid encoding a T3SS, fully required for virulence in pathogenic Yersinia spp. in mammalian hosts (Cornelis 2002). Since the absence of pYV would have precluded any virulence studies in cellular or animal infection models, we decided to use a Y. pseudotuberculosis isolate closely related to the IP31758 strain in order to investigate the functionalities of the yT4BSS. We chose therefore to use in our study the strain SP-1303, which: (i) is genetically close to the strain IP31758, differing only by 23 SNPs and with an average nucleotide identity of 99.96%, (ii) belongs to the Y. pseudotuberculosis lineage 8, which harbors strains associated to FESLF (Savin et al. 2019), and (iii) harbors both the pYV and the pVM82 (Lemarignier et al. 2023) (Fig. 1a).

Figure 1.

Gene organization and homology of the yT4BSS gene cluster in the pVM82 plasmid from Y. pseudotuberculosis strain SP-1303. (a) Plasmid map of the pVM82 plasmid from Y. pseudotuberculosis FESLF strain SP-1303 with annotated genes product. The first outer ring corresponds to gene products, without labeling hypothetical protein products. The second ring represents gene position (blue: positive or red: negative strand). The third ring shows GC skew. Inner ring shows GC content variation from the mean. Figure generated with Circos from the annotated Genbank sequence of pVM82 (ID: CP130902.1) (Krzywinski et al. 2009). The structure of the yT4BSS and of a type IV pili gene clusters are further highlighted. (b) T4BSS gene clusters conservation and comparison between the IncI plasmid R64 (NC_005 014), C. burnetii RSA 493 (NC_002 971), L. pneumophila Philadelphia 1 (NC_002 942), Y. pseudotuberculosis SP-1303 (GenBank accession no. NZ_CP130902.1), and P. putida isoF (CP072013.1) with a minimum alignment sequence identity of 30%. Genes colored white are not T4BSS genes and those with brown outlines share sequence identity. Genes colored in light gray are strain specific. Figure generated with clinker (Gilchrist et al. 2021).

The yT4BSS gene cluster of the strain SP-1303 is genetically similar to the one previously described in the strain IP31758: it displays several dot/icm genes which are homologous to the T4BSS genes present in the bacterial intracellular pathogens L. pneumophila, C. burnetii, and the recently T4BSS described in the soil saprophyte P. putida (strain isoF) (Fig. 1b) (Purtschert-Montenegro et al. 2022). A total of 13 dot/icm genes (out of 27 in L. pneumophila and 23 in C. burnetii) were first identified by Eppinger et al. (2007) and are known to encode essential structural components of the translocon apparatus, including the crucial cytoplasmic ATPases (dotO/icmB and dotB) as well as encoding key elements of the T4CC. By using Macsyfinder (Abby et al. 2014, Néron et al. 2023), we were able to identify three additional dot/icm genes on the pVM82, also present in L. pneumophila, C. burnetii, and P. putida: icmT, encoding a structural cytoplasmic membrane protein (Bitar et al. 2005), as well as two copies of icmX (annotated: icmX1 and icmX2), known to encode a plug structural periplasmic protein (Matthews and Roy 2000) which were not initially recognized as such by Eppinger et al. (2007) (Fig. 1a). The two copies of icmX are encoded in a different operon, in close proximity to a type IV pili gene cluster on the pVM82 (Fig. 1a). Some dot/icm genes shared by L. pneumophila and C. burnetii, implicated in the structure and function of their T4BSS are absent in the yT4BSS such as genes encoding the chaperones IcmS and IcmW (Cambrone and Roy 2007, Xu et al. 2017, Macé et al. 2022a), genes encoding the inner membrane structural proteins IcmV, IcmQ, DotU/IcmH, and DotP/IcmD, as well as the genes dotK/icmN forming the α-density of the OMCC (Ghosal et al. 2019, Durie et al. 2020). The genes dotZ, dotY, lvgA, icmR, dotJ/icmM, icmF, and dotV were also not found in the yT4BSS gene cluster but are known to be specific to L. pneumophila (Zamboni et al. 2003, Zusman et al. 2003). Interestingly, higher percentage of sequence identity with a greater number of homologous dot/icm genes were found between the yT4BSS gene cluster and the one from P. putida strain isoF compared to those of L. pneumophila and C. burnetii (Fig. 1b).

Several T4BSS have been described as required for conjugation (Cruz et al. 2010). We observe that the yT4BSS is only distantly related to the representative conjugative T4BSS of plasmid R64 (Fig. 1b). Moreover, no relaxase or oriT, both essential components of the conjugation process, were detected using Macsyfinder (Abby et al. 2014, Néron et al. 2023) and oriTfinder (Li et al. 2018). Together, these results suggest that the yT4BSS of pVM82 should not be involved in conjugation.

To have a broader view of the presence of the yT4BSS in all lineages from the species Y. pseudotuberculosis, we investigated the distribution of dot/icm genes in 287 bacterial genomes (Savin et al. 2019). The full yT4BSS gene cluster was only observed in the majority of strains from the ancestral lineage 8 (Supplementary Fig. S1a). Incomplete yT4BSS were observed in strains from lineages 6 and 11 (SupplementaryFig. S1a). Interestingly, the clinical isolates of Y. pseudotuberculosis responsible for FESLF are grouped in lineages 1 and 8, suggesting that the yT4BSS could participate in the pathogenicity of some strains associated with FESLF. To investigate whether the yT4BSS gene organization is conserved in Y. pseudotuberculosis strains from lineage 8, we analyzed the genomes of 46 isolates from this bacterial group. As shown in Supplementary Fig. S1b, the gene organization of the yT4BSS gene cluster is highly similar between strains with a conserved synteny. The presence of the main essential structural and functional dot/icm genes (discussed above) and the conserved synteny in multiple strains suggest that the yT4BSS could be functional (Nagai and Kubori 2011).

Finally, we explored the NCBI database using cblaster (Gilchrist and Chooi 2021) to investigate whether the yT4BSS is also found in other genomes of the genus Yersinia. We were able to identify complete dot/icm gene clusters with high homology (>80% identity and >95% coverage) only in pathogenic species, including strains from the human enteropathogen Y. enterocolitica as well as strains from the fish pathogen Y. ruckeri, but not in Y. pestis (Supplementary Fig. S2). Incomplete isolated clusters of genes were found in one strain of the environmental species Y. alsatica (closely related to Y. enterocolitica) as well as in other strains of Y. enterocolitica and Y. ruckeri. The presence of closely homologous yT4BSS in phylogenetically distant strains of Yersinia spp. (Y. enterocolitica, Y. ruckeri) suggests an independent acquisition.

In conclusion, the majority of Y. pseudotuberculosis strains from lineage 8, associated with FESLF, harbor a complete yT4BSS gene cluster encoded on the pVM82 plasmid, highly similar to known T4BSS in L. pneumophila, C. burnetii, and P. putida, and it might be implicated in the etiology of FESLF.

The yT4BSS is expressed in vitro, in cellulo, and in vivo

To address the functionality of the yT4BSS, we first investigated its putative expression under different conditions in vitro, in cellulo, and in vivo. We identified in the main yT4BSS gene cluster five putative transcriptional units out of the eleven previously characterized in L. pneumophila (Sahr et al. 2012) (underlined in Fig. 1a). By RT-qPCR, we measured the expression level of one gene of each transcriptional unit (dotA, dotM, dotB, dotG, and dotO) in LB media, at post-exponential phase (7–8 h of growth) and at different temperatures (21°C, 28°C, and 37°C). For the five analyzed genes, we observed an increase in their expression level between 21°C and 28°C (environmental temperatures with 28°C as optimal growth temperature) but the differences were not statistically significant; importantly, for all the genes, we observed a significant increase in expression levels between 21°C and 37°C (host-associated temperature) (Fig. 2a and Supplementary Fig. S3). In the case of four genes out of the five (dotA, dotM, dotB, and dotO), significant differences were also observed between 28°C and 37°C, suggesting overall an increased expression of the yT4BSS in conditions mimicking the mammalian host environment (37°C).

Figure 2.

yT4BSS genes are expressed in vitro, in cellulo, and in vivo. (a) Normalized expression of five yT4BSS genes to the reference gene proC, in the SP-1303 WT strain cultured in LB media, pH 7, at late-exponential phase (7–8 h of growth) under different temperatures (21°C, 28°C, and 37°C). Ct obtained by RT-qPCR are normalized to the Ct measured with the housekeeping gene proC and plotted as 2^-ΔCq. Data from three independent experiments in triplicate with means and SD are shown. Kruskal–Wallis tests corrected for multiple comparisons using Dunn’s test were performed on each gene. (b–e) Measurement of photons emission produced during bacterial growth over time in LB or minimal media, at different temperatures (21°C, 28°C, 37°C, and 42°C) and different pH (5, 6, and 7) for the bioluminescent reporters PdotDCB and PdotA. Bacteria harboring the reporters were grown O/N in LB media, pH 7, 28°C prior to the experiment. RLU measured using a Glomax plate reader, were subtracted from Pless (reporter strain without inserted promoter) data and were normalized by the PrplN (constitutive promoter with high expression rate) values. Data from three independent experiments with means and SD are shown. One-way ANOVA tests corrected for multiple comparisons using Tukey’s tests were performed at each time point. (f) Photon emission measured during the intracellular cycle of bioluminescent reporter strains within HeLa cells. Y. pseudotuberculosis SP-1303 strains harboring different bioluminescent reporters (PrplN, PdotDCB, PdotA, and Pless) were grown O/N at 28°C prior infection for 1 h (MOI of 70) at 37°C, followed by gentamicin treatment (+1, +3, +7, and +23 h). RLU measured for each well were normalized by CFUs counted per well and subtracted from Pless data. Data from three independent experiments (circle, triangle, and diamond are means of 4 replicates/condition) with means and SD are shown. One-way ANOVA test corrected for multiple comparisons using Tukey’s test were performed for each reporter. (g) One representative example of OF1 mice bread-feeded and inoculated with the Y. pseudotuberculosis SP-1303 strains harboring bioluminescent reporters (inoculum: 6 × 10⁸ CFU/mouse), monitored each day until day 9 pi (n = 5/condition) using an IVIS Spectrum system. The same radiance scale (in p/sec/cm2/sr) was used for the high signal found in infected conditions with PrplN and PdotDCB strains, whereas another radiance scale is used for PdotA and Pless. (h) Dissection of mice infected with Y. pseudotuberculosis SP-1303 strains harboring bioluminescent reporters at day 9 pi. Intestinal tracts were not washed. Stom: stomach, cae: cecum, S: spleen, MLNs: mesenteric lymph nodes, and L: liver. Radiance scales used in p/sec/cm2/sr are indicated for each organ.

We then complemented these in vitro results by investigating the activity of the promoter regions of the highly conserved T4BSS genes dotDCB and dotA upon fusion to the luxCDABE operon of P. luminescens (Derbise et al. 2019), monitoring bioluminescence production in conditions similar as those tested above by RT-qPCR, analyzing additionally the influence of acidic pH and minimal medium. Both promoters PdotDCB and PdotA were activated at the four tested temperatures 21°C and 28°C (environmental conditions), 37°C and 42°C (mimicking host-associated basal and high fever temperatures, respectively), when compared with the background signal of a construction without promoter (Pless), all along bacterial growth (Fig. 2b and c and Supplementary Fig. S4). Interestingly, in contrast to PdotA with a low basal constant activity, regardless of the temperature, PdotDCB activation is temperature dependent: from an overnight (O/N) preculture at 28°C, a one log₁₀ decrease in normalized photon emission was measured when bacteria were newly grown at 21°C during 8 h; on the contrary, when newly grown at 37°C or 42°C, a significant increase in PdotDCB activity was measured over time (Fig. 2b). PdotDCB (but not PdotA) activity is also influenced by acidic pH, since a pH shift from 7 to 6 leads to an increase in bioluminescence production at 28°C, which is more pronounced when pH is shifted to 5 (Fig. 2d and e). Temperature synergizes with pH, as activation of PdotDCB is more important at 37°C pH 5 compared to 28°C pH 5 (Fig. 2d). On the contrary, minimal medium does not influence photon emission (Fig. 2d and e). Altogether, these results suggest that the yT4BSS is expressed in vitro during bacterial growth in a temperature- and pH-dependent manner, with higher expression at temperatures mimicking the mammal host (37°C) than in environmental ones (21°C and 28°C) and with higher expression at acidic pH.

Because the T4BSS of L. pneumophila and C. burnetii is essential for their intracellular life cycle (Qiu and Luo 2017), the activity of the PdotDCB and PdotA bioluminescent reporters was evaluated upon infection of epithelial human HeLa cells. After 1 h of infection, gentamicin was added, left for bacterial extracellular killing for additional time periods (+1, +3, +7, and +23 h) and photon emissions were monitored and normalized by CFUs per well. PdotA was active in cellulo at 2, 4, and 8 h post-infection (pi) when compared to the background noise of the Pless construction, but the signal was not significantly different from the background at 24 hpi (Fig. 2f). To the contrary, PdotDCB, also pregrown at 28°C, is upregulated all along the course of intracellular life cycle within HeLa cells, suggesting that temperature and/or cellular environments are major drivers of yT4BSS expression.

We finally investigated in vivo the spatio-dynamic expression of the yT4BSS in OF1 mice upon oral infection (Derbise et al. 2020). Beside the Y. pseudotuberculosis SP-1303 strains harboring the PdotA and PdotDCB inductible reporters as well as the control Pless construct described above, mice were also infected with a strain harboring a constitutive promoter controlling bioluminescence production (PrplN) (Fig. 2g and h). The PdotDCB reporter was found to be active all along the course of mice infection, characterized first by bioluminescence signals in the abdominal region and later reaching the mouse tail, the latter corresponding to a septicemic stage with bacteria present in the blood stream (Fig. 2g). The PdotA reporter displayed a similar pattern as PdotDCB but its signal was very low, close to the background noise defined with Pless-infected mice (Fig. 2g). In addition, the PdotDCB and PdotA are active in all infected organs: all along the gastrointestinal tract (in the intestinal lumen, Peyer’s patches, cecum, and colon) as well as in the spleen, pool of MLNs and liver (exception for PdotA) (Fig. 2h). Similar results were observed in BALB/c mice concerning activation of PdotDCB and PdotA with a lower signal because of the lower infection dose (Supplementary Fig. S5). Thus, yT4BSS is active in vivo, all along the course of mouse infection without any organ specificity.

In conclusion, the yT4BSS is active and expressed in vitro in a temperature-dependent manner, with increased expression in vitro at acidic pH and 37°C (an environment mimicking mammalian host body), in cellulo during the intracellular life cycle of Y. pseudotuberculosis in mammalian cell lines, as well as during in vivo mouse infection in all infected organs.

Investigation of putative yT4BSS effectors and translocation function

To further understand the function of the yT4BSS in Y. pseudotuberculosis lineage 8 virulence, we used the S4TE (Searching algorithm for type IV effector proteins, https://sate.cirad.fr/) bioinformatic tool (Noroy et al. 2019) to investigate putative yT4BSS effectors. On the chromosome, we removed from the results list genes also present in yT4BSS-negative strains, resulting in a total of 5 specific short-listed candidates (Fig. 3a), while on the pVM82 plasmid 11 candidates were found (Fig. 3b). Based on GC content analysis by S4TE, we observed that the majority (7 out of 11) of the effector candidates on the pVM82 are located in low GC content regions, with three of them (B0001-3) clustered in a single region. These results remind of the distribution of L. pneumophila T4BSS effectors, which are mainly found in low GC content regions, with some genetic regions enriched in clustered effectors (Burstein et al. 2016, Noroy et al. 2019). Thus, a total of 16 effector candidates were found above the S4TE threshold on the SP-1303 genome.

Figure 3.

Identification of putative yT4BSS effectors. Representation of the main molecular features of 11 yT4BSS effector candidates identified in the chromosome (a) or in the pVM82 plasmid (b) of Y. pseudotuberculosis strain SP-1303, using the S4TE algorithm (Noroy et al. 2019). Effectors are listed in the order in which they appear in the genome. Score: corresponds to the score given by S4TE. (c) Expression analysis for pVM82 and chromosomal yT4BSS effector candidates. SP-1303 WT strain was grown in LB, pH 5 at 37°C during 7 h, then bacterial RNA was extracted and analyzed by RT-qPCR. Cycle threshold (Ct) values are presented for control housekeeping genes (rpoD, proC,and nadB), for structural yT4BSS components (dotA, dotB), for pVM82 putative yT4BSS effector genes (B0001 to B0122), and for chromosomal putative yT4BSS effector genes (0769 to 3692). CT reference values for dotA and dotB expression are represented by dotted blue lines, and a control sample without reverse transcriptase (no RT) is represented by the dotted black line. Data from three independent experiments in duplicates with means and SD are shown.

We then investigated the presence of specific domains on the effector candidates. The C-terminal region of most of the candidates present basicity properties (13 out of 15), an important feature found in many T4SS, as well as E-block domains (5 out of 15), major translocation signal found in L. pneumophila effectors. Additionally, coiled-coil and EPIYA (Glu-Pro-Ile-Tyr-Ala) domains, which have been implicated in protein–protein interactions, were identified. The majority (10 out of 11) of the candidates present in the pVM82 harbor an Amp promoter motif found in effectors of the VirB/D4-like T4ASS of Anaplasma phagocytophilum as well as the PmrA promoter motif from L. pneumophila, interesting features which reinforces their potential as putative effectors (Noroy et al. 2019). Interestingly, seven candidates display nuclear translocation signals and two displayed mitochondria translocation signals, implying a potential import within the host nucleus or mitochondria of host cells.

By performing RT-qPCR on bacterial samples grown in a condition demonstrated previously to allow yT4BSS expression (LB, pH 5, 37°C), we validated the expression of the 16 effector candidate genes (Fig. 3c): we observe that 7 out of the 11 genes present on the pVM82 display levels of expression equivalent to those of dotA and dotB, 3 genes (B0054, B0081, and B0093) display higher expression levels and 1 gene (B0071) displays lower expression levels; among the 5 chromosomal genes, 2 display expression levels comparable to those of dotA and dotB, wheareas 3 display lower expression levels.

To assess secretion of putative yT4BSS effectors in the SP-1303 strain, we constructed fusion proteins (SupplementaryFig. S6), under the control of an arabinose-sensitive promoter, of the candidates B0054 and B0081 (displaying E-block domains in their C-terminal region) and B0071 (displaying EPIYA, coiled coil, and DNA-binding domains, as well as a high S4TE score) with the β-lactamase, which was then used as reporter for secretion by measuring its activity on diverse substrates. We first investigated putative secretion in extracellular medium using nitrocefin, a cephalosporin derivative, as substrate for β-lactamase hydrolysis, which shifts absorbance from 390 to 492 nm upon cleavage. We investigated secretion at 28°C or 37°C, in rich LB or minimal medium (MM), and at pH 6 or 7. As a positive control, we used a reporter consisting of the first 100 amino acids of the T3SS effector YopE fused to the β-lactamase. As shown in Supplementary Fig. S6, we were not able to detect secretion of the three candidate effectors under the tested conditions. We then investigated whether secretion could take place intracellularly. For this, we loaded HeLa cells with the β-lactamase substrate CCF4, which shifts fluorescence from 520 to 447 nm when cleaved. We then infected cells for 6 or 24 h with Y. pseudotuberculosis strain SP-1303 harboring constructs with the candidate effectors, or with the positive control Listeria monocytogenes expressing a surface β-lactamase (Quereda et al. 2015). As in the case of the extracellular assay, we were not able to detect secretion of the tested candidate effectors within HeLa cells (Supplementary Fig. S6).

Finally, we investigated whether the yT4BSS could translocate known T4BSS effectors of L. pneumophila (RomA nad LphD) or C. burnetti (CvpB). RomA is a methyltransferase and LphD is an histone deacetylase, both modulators of epinetic marks on histones, while CvpB modulates phosphoinositide metabolism (Rolando et al. 2013, Martinez et al. 2016, Schator et al. 2023). Interchangeable properties between T4BSS have already been demonstrated with C. burnetti’s effectors secreted by the L. pneumophila T4BSS (Chen et al. 2010). For this, we used already validated IPTG-inducible vectors allowing the production of β-lactamase fusion proteins with tested effectors and evaluated their capacity to be translocated in the contet of cellular infection of HeLa cells and cleavage of the CCF4 substrate as described above. As shown in SupplementaryFig. S7, we were not able to detect secretion of RomA, LphD nor CvpB by the yT4BSS of Y. pseudotuberculosis. To conclude, we have identified putative yT4BSS effectors using the bioinformatic algorithm S4TE, and we demonstrated their co-expression with the yT4BSS. The conditions favoring their secretion and their host targets require further analysis.

The yT4BSS is not implicated in the intracellular life cycle of Y. pseudotuberculosis

Y. pseudotuberculosis is able to invade host cells via interactions between its surface molecule invasin and host cell β1-integrin receptors (Isberg et al. 1987). Within epithelial cells but also macrophages, Y. pseudotuberculosis has been reported to proliferate in autophagosome-like compartments (Moreau et al. 2010, Ligeon et al. 2014) but the bacterial virulence factors implicated in the biogenesis of this intracellular replication niche have not been identified so far (Lemarignier and Pizarro-Cerdá 2020). Since L. pneumophila and C. burnetii use their respective T4BSS to generate intracellular replication vacuoles via hijacking several host functions including membrane trafficking (Christie et al. 2017, Qiu and Luo 2017, Lockwood et al. 2022), we investigated whether the yT4BSS was implicated in modulating the Y. pseudotuberculosis intracellular behavior. For that, we generated in the strain SP-1303 a single deletion mutant of the gene dotB, encoding a crucial cytoplasmic ATPase, since dotB has been shown to be essential for the functionality of the T4BSS in L. pneumophila and C. burnetii (Sexton et al. 2004, Sexton et al. 2005, Macé et al. 2022b) and we showed in the previous section that the dotB gene is upregulated in cellulo and in vivo upon bacterial infection.

Using gentamicin invasion assays, we first investigated in HeLa cells the intracellular survival of WT vs. ΔdotB mutant bacteria upon different infection times. As shown in Fig. 4a, no significant differences in intracellular bacterial loads were observed between the WT strain and the ΔdotB mutant at all investigated time points (2, 4, 8, and 24 hpi). Similar results were found when studying infection of murine RAW264.7 macrophages, of mBMDM or hBDM (Supplementary Figs S8A and S9a–b), suggesting that the yT4BSS is not implicated in cellular invasion nor modulation of the intracellular survival of Y. pseudotuberculosis within epithelial and macrophage-like cells.

Figure 4.

The yT4BSS does not contribute to the intracellular life cycle of Y. pseudotuberculosis within epithelial HeLa cells. (a) Intracellular survival of WT and ΔdotB mutant strains within HeLa cells (MOI 9) at different time post-infection (2, 4, 8, and 24 hpi). Data from two independent experiments (triangle and round differentiate the experiments) with means and SD are shown. Multiple unpaired t-tests were performed. (b) Representative images of HeLa cells infected (MOI 25) with the WT or ΔdotB strains at 5 hpi or 24 hpi by fluorescent microscopy using antibodies against the surface O-antigen (O1b) of Y. pseudotuberculosis (Cy5, magenta), antibodies against LAMP1 (FITC, green) and Hoechst (blue). Scale bars are 10 µm. Overlay images were generated using the ImageJ/Fiji software. (c) Representative images of vacuolar structures within HeLa infected cells with the WT or ΔdotB strains at different time (2, 5, and 24 hpi) using TEM. Yellow stars highlight the position of nuclei. Scale bars are 1 µm (2 and 5 h) and 2 µm (24 h).

We then analyzed by fluorescence microscopy the intracellular behavior of the SP-1303 WT strain and of the ΔdotB mutant by infecting mammalian host cells for different periods of time (5 and 24 hpi) and labeling bacterial intracellular compartments with antibodies against the lysosomal membrane-associated protein 1 (LAMP-1). In HeLa cells, we were able to observe the formation of Yersinia-containing vacuoles (YCVs) labeled with LAMP-1 at 5 hpi, and we noted a significant enlargement of these compartments at 24 hpi (Fig. 4b); however, no differences were noted between YCVs generated by the WT vs. those formed by the ΔdotB mutant. We were also not able to detect differences in their intracellular behavior in epithelial Caco-2 cells, RAW264.7 macrophages, mBMDM, or hBDM (Supplementary Figs S8b–c and S9c–d). To further gain insight into the ultrastructure of the YCVs, we infected HeLa cells for different time points (2, 5, and 24 hpi) and we analyzed them using TEM. As we previously observed by fluorescence microscopy, in both infected cases (WT or ΔdotB mutant), YCVs were clearly visible at 5 hpi, and the size of the vacuoles increased at 24 hpi, reaching around 7 µm (Fig. 4c). YCVs were formed by a single membrane vacuole as previously described in HeLa cells (Ligeon et al. 2014) and preferentially localize close to the nucleus. Again, no apparent differences in the ultrastructure were observed between YCVs formed by WT vs. ΔdotB mutant bacteria, which are both preferentially located in contact with the vacuolar membrane.

Since interactions of YCVs with LC3-positive compartments have been previously reported (Moreau et al. 2010), we investigated the distribution of this autophagosomal marker in the context of cellular infection with SP-1303 strains; as shown in Supplementary Fig. S10a, LC3B was detected in the membrane and vicinity of vacuoles containing WT or ΔdotB mutant bacteria. By analyzing the distribution of the acidic marker Lysotracker (Supplementary Fig. S10b), we additionally concluded that the WT and ΔdotB YCVs do not acidify. Nevertheless, Lysotracker-positive compartments were observed in their vicinity, prompting us to investigate the distribution of the lysosomal hydrolase cathepsin D (Supplementary Fig. S10c); we observed the presence of this enzyme in both vacuoles containing WT and ΔdotB mutant bacteria, suggesting that these compartments sustain interactions with lysosomes but do not acidify. Thus, the yT4BSS does not take part in vacuole biogenesis, maturation, and localization within epithelial and macrophage-like cells.

Our results therefore indicate that, to the opposite of L. pneumophila and C. burnetii, the yT4BSS is not required to modulate the intracellular trafficking of Y. pseudotuberculosis in epithelial and phagocytic cells.

The yT4BSS is required for Y. pseudotuberculosis full virulence in a mouse oral infection model

To investigate the potential contribution of the yT4BSS in Y. pseudotuberculosis pathogenicity in vivo, we infected orally or intravenously BALB/c mice with WT or ΔdotB mutant strains and monitored first animal survival (Fig. 5a and b). Interestingly, a higher percentage of mice survived upon oral infection with the ΔdotB mutant compared to animals challenged with the WT strain, demonstrating an attenuation of the virulence of the ΔdotB mutant in vivo (Fig. 5a). A complemented ΔdotB:: dotB strain restored the virulence phenotype in orally infected mice. In contrast, no difference in mice survival was observed with the WT strain or the ΔdotB mutant when inoculated intravenously with two different infection doses (Fig. 5b and Supplementary Fig. S11). Thus, these results demonstrate a contribution of the yT4BSS in Y. pseudotuberculosis virulence upon oral infection in mice, but not upon intravenous infection.

Figure 5.

The yT4BSS contributes to Y. pseudotuberculosis virulence in orally infected mice without affecting organ colonization. (a) Survival monitoring of bread-fed mice inoculated with WT, ΔdotB mutant, or ΔdotB::dotB complemented strains at 6 × 10⁶ CFU/mouse. Three independent experiments were performed (n = 7 mice/condition/experiment). Log-rank (Mantel–Cox) test was performed (WT/ΔdotB: P-value = 0.0107; ΔdotB::dotB/ΔdotB: P-value = 0.0014). (b) Survival monitoring of mice after intravenous injection with 50 CFU/mouse of WT, ΔdotB mutant strains. Representation of a single experiment (n = 5/condition). (c) Photon emission monitoring in mice infected with the constitutive bioluminescent WT-lux or ΔdotB-lux strains overtime using an IVIS imaging system. Mice number 1, 2 or mice number 3, 4 are representative of respectively slow and rapid kinetics of bacterial colonization. Same radiance scale bar (in p/sec/cm2/sr) was used for days 1–3 pi and another for 4–7 pi. (d) Total photon flux (p/s) measured on drawn ROIs covering the mouse body using Living Image 4.5 software. Pool of three independent experiments with groups of 6–7 mice.(e) Bacterial load per infected organ from orally infected mice (6 × 10⁶ CFU/mouse) with WT or ΔdotB mutant strains. Intestine, colon, cecum, spleen liver, and MLNs were collected at day 2, 5, and 7 pi and homogenized for CFUs counting. Intestines and colons were longitudinally opened for content collection for CFUs, treated for 1 h with gentamicin before homogenization for CFUs counting (intestinal and colon tissues). Data from two independent experiments with 5–7 mouse/condition (triangle and round differentiate the experiments) are shown. Absence of CFUs or data below the detection threshold (dashed line) are represented at 10⁰ CFU/organ. No statistical differences in bacterial loads were identified using Mann–Whitney tests.

To further gain insight into the mechanisms underlying the reduced virulence of the ΔdotB mutant in orally infected mice, we constructed constitutive bioluminescent WT or ΔdotB strains and quantified photon emission in orally infected mice each day during infection, in order to obtain information on the kinetics of infection and bacterial localization throughout infection (Fig. 5c and d). In both WT and ΔdotB-infected groups, starting for few mice at day 1 pi, photons were recorded in the abdominal region with increasing signal intensity over time, suggesting bacterial colonization of the gastrointestinal tract. For some mice in both infected groups, at day 5 or 6 pi, signals were detected in the tail, an indication of blood colonization (Fig. 5c). By quantifying the total bioluminescence flux in the entire mouse body, no differences between both infected groups were observed over time (Fig. 5d). In order to further compare the kinetics of bacterial colonization of mice organs, we orally infected BALB/c mice with the WT or the ΔdotB strains, and all known infected organs were collected at days 2, 5, or 7 pi and homogenized for CFUs counting: small intestinal tissue and its content, cecum, colon tissue and its content, MLNs, spleen, and liver (Fig. 5e). Surprisingly, no statistical differences in CFUs were quantified at both investigated time points in all infected organs between WT- and ΔdotB-infected mice. These results indicate that, whereas the yT4BSS contributes to virulence in orally infected mice, it does not modulate organ colonization nor bacterial survival within these organs.

The yT4BSS is associated with cecal and MLN necrosis

To investigate the mechanisms underlying the virulence defect upon oral infection of the ΔdotB mutant without affecting organ colonization, we explored the tissue architecture by histological analyses of two key organs in the infectious process, the cecum and the MLNs (Barnes et al. 2006, Fahlgren et al. 2009). BALB/c mice were orally infected with the WT or the ΔdotB mutant strains, organs were extracted at 7 dpi, fixed by immersion in 10% neutral-buffered formalin, and processed for H&E labeling. Multiple histological features were scored according to the degree of severity (0: absence, 1: mild, 2: moderate, and 3: severe) in both organs and the extension in the cecum (1: focal, 2: multifocal, and 3: diffuse) (listed in Supplementary Table S5 and Materials and Methods). The resulting sum of scores provided an overall indication of the organ’s histopathology (Figs 6a and 7a). Organs from a PBS-exposed group did not show relevant histopathological lesions (Figs 6d and 7d). In infected animals, upon analysis of the recorded criteria separately, the presence of 'multifocal to coalescing granulomas' and 'area of necrosis' were the most relevant histopathological findings, showing more severe lesions from the WT-infected group than those from the ΔdotB-infected group (Figs 6, 7 and Supplementary Fig. S12). In MLNs, severe lesions were characterized by a central area of necrosis with bacterial colonies and some degenerated neutrophils, surrounded by a rim of epithelioid macrophages (Fig. 6f). Samples were thus subjected to histomorphometry analyses to measure the total percentage of area of necrosis in each MLNs and cecal samples (Figs 6h, i and 7h, i). Interestingly, the percentage of area of necrosis in MLNs of the WT-infected group was on average higher than in the case of the ΔdotB-infected group (Fig. 6g and h). Histomorphometric analyses of cecal samples also revealed a higher average percentage of necrosis in the cecum of WT-infected mice than in the ΔdotB-infected mice (Fig. 7h and i). Thus, the yT4BSS could be implicated in the development of severe granulomatous lesions and necrosis in MLNs and in cecum.

Figure 6.

The yT4BSS is implicated in the formation of necrotic areas in MLNs of infected mice. (a) Total score on multiple histological findings evaluated in MLN samples with a maximal score of 15. (b) Severity level of the feature corresponding to the presence of epithelioid granulomas in MLNs. 0: absence of necrotizing granuloma, 1: mild lesion, 2: moderate lesion, and 3: severe lesions. (c) Severity level of the feature corresponding to the presence necrotizing granulomas in MLNs. 0: absence of necrotizing granuloma, 1: mild lesion, 2: moderate lesion, and 3: severe lesions. (d) Representative image of MLN (H&E) from a PBS mouse showing no histopathological lesions. (e) Representative image of MLN (H&E) from a WT-infected mouse showing epithelioid granulomas (arrowheads) (moderate, score 2). (f) Representative image of MLN (H&E) from a WT-infected mouse showing multiple necrotizing granulomas (asterisks) (severe, score 3). Inset shows closer magnification of one of the necrotic areas. (g) Representative image of MLN (H&E) from a ΔdotB-infected mouse showing absence of necrotizing granulomas. Arrows in inset show the presence of tingible body macrophages (mild apoptosis, score 1). (h) Histomorphometric analysis of the percentage of necrosis in MLN from a WT-infected mouse (moderate, score 2) with Nikon NIS-Ar software. Blue and green colors in MLN represent the necrotic areas measured. (i) Percentage of area of necrosis on the total MLNs area. Samples are from two independent experiments (triangle and round differentiate the experiments).

Figure 7.

The yT4BSS is implicated in the formation of necrotic areas in the cecum of infected mice. (a) Total score on multiple histological features evaluated in cecum samples with a maximal score of 54 (listed in Supplementary Table S5). (b) Severity level of the feature corresponding to the presence of epithelioid granulomas in cecum. 0: absence of necrotizing granuloma, 1: mild lesion, 2: moderate lesion, and 3: severe lesions and 1: focal, 2: multifocal, 3: diffuse. The sum of both criteria represents the score number. (c) Severity level of the feature corresponding to the presence necrotizing granulomas in cecum. 0: absence of necrotizing granuloma, 1: mild lesion, 2: moderate lesion, and 3: severe lesions and 1: focal, 2: multifocal, and 3: diffuse. The sum of both criteria represents the score number. (d) Representative image of cecum (H&E) from a PBS mouse showing no histopathological lesions. (e) Representative image of cecum (H&E) from a WT-infected mouse showing an epithelioid granuloma in the lamina propria (arrowhead) (multifocal, mild, total score 3). (f) Representative image of cecum (H&E) from a WT-infected mouse showing multiple necrotizing granulomas (asterisks) (focal, severe, total score 4). Inset shows bacteria colonies and neutrophils. (g) Representative image of cecum (H&E) from a ΔdotB-infected mouse showing a necrotizing granuloma (asterisk) (focal, moderate, total score 3). (h) Histomorphometric analysis of the percentage of necrosis in cecum from a WT-infected mouse (focal, moderate, total score 3) with Nikon NIS-Ar software. Red color in cecum represents the necrotic area measured. (i) Percentage of area of necrosis on the total cecum area. Samples are from two independent experiments (triangle and round differentiate the experiments).

We also conducted immunohistochemical (IHC) analyses to identify infiltration by neutrophils (anti-Ly6B2 antibody) and macrophages (anti-F4/80 antibody) in MLNs and cecal samples. Samples were submitted to digital image analysis to calculate the percentage of positivity with respect to the whole area (Supplementary Fig. S13). Concerning the Ly6B2 marker, the percentage of Ly6B2-positive staining was similar in MLNs and ceca between both infected groups (WT and ΔdotB mutant), with similar locations of Ly6B2⁺ cells (Supplementary Fig. S13c and d). In MLNs, Ly6B2⁺ cells were detected surrounding the granulomas, outside of the epithelioid cells which form the lesions (Supplementary Fig. S13c and d). In the cecum, Ly6B2⁺ cells (neutrophils and some macrophage-like cells) were clustered in cecal Peyer’s patches near to the necrotic areas and surrounding them (Supplementary Fig. S13f inset). Concerning the macrophage F4/80 marker, in MLNs and ceca, the percentage of positive F4/80 staining were also closely similar between both infected groups (Supplementary Fig. S13h and l) with similar localization of F4/80⁺ cells. In MLNs, the staining was detected for both groups (WT and ΔdotB) in the cytoplasm of macrophage-like cells, located in epithelioid cells of the granulomas (Supplementary Fig. S13J) and necrotizing granulomas (Supplementary Fig. S13k). In the cecum, F4/80 staining was detected in the cytoplasm of macrophage-like cells from the lamina propria of the cecal mucosa (Supplementary Fig. S13m inset) and in the periphery of the necrotic areas (Supplementary Fig. S13m–n asterik). Thus, according to these results, it seems that in presence of the yT4BSS, no specific recruitment and localization of neutrophils and macrophage-like cells were observed in ceca and MLNs from infected animals.

In conclusion, the yT4BSS might be implicated in the induction of necrosis and severe granulomas in MLNs and the cecum in infected mice, without affecting neutrophils and macrophages behavior at these sites of infection.

The yT4BSS modulates cytokine innate immune responses

Our histopathological results prompted us to test whether the yT4BSS is associated with modulation of the cytokine responses to Y. pseudotuberculosis infection using mBMDM from BALB/c mice. For this, we measured the protein levels of a panel of 18 cytokines and chemokines (Supplementary Table S6) involved in various aspects of the immune response (innate and adaptative responses) in the supernatant of mBMDM upon 8 h of infection with the WT, the ΔdotB mutant or the complemented strain ΔdotB::dotB. Five cytokines: CCL2, G-CSF, IL-6, IL-33, and IL-10 displayed a higher concentration in the supernatant of mBMDM infected with the WT strain than in cells infected with the ΔdotB mutant (Fig. 8 and Supplementary Fig. S14). The complemented strain ΔdotB::dotB restored the WT phenotype in CCL2 and G-CSF cases but not statistically in IL-33, IL-6, and IL-10, even if the tendancy followed the one of the WT. Thus, the yT4BSS is associated with an increase in cytokine production from the innate immune response of mBMDM upon Y. pseudotuberculosis infection.

Figure 8.

The yT4BSS modulates the cytokine immune response. Cytokine profile analysis on supernatants from infected mBMDMs at 8 hpi with WT, ΔdotB mutant or ΔdotB::dotB complemented strains (MOI 10) using the Luminex technology. Data from three independent experiments (triangle, round, and stars) with 4 technical replicates and presented as mean and SEM are shown. Mixed model with Tukey’s multiple comparisons correction were performed on log(Y + 1) transformed data.

Discussion

Y. pseudotuberculosis is an enteric pathogen, traditionally associated to mesenteric adenitis in humans, worldwide. A phylogenetic cluster of strains from lineages 1 and 8 has been implicated in the undercharacterized syndrome FESLF, associated with specific clinical manifestations such as skin rash, hyperemic tongue, and desquamation, with cases still reported mainly in Russia, Korea, and Japan (Sato 1987, Ocho et al. 2018, Suzuki et al. 2024). For the time being, to the best of our knowledge, the only virulence factor linked to FESLF is the chromosomally encoded superantigen YPMa, present in both lineage 1 and 8 strains, whose presence is correlated with systemic symptoms found in FESLF patients (Abe et al. 1997, 2003, Ueshiba et al. 1998). The plasmid pVM82, present only in FESLF strains from lineage 8, was previously associated to bacterial pathogenicity but no specific virulence factors have been characterized so far in this genetic element (Gintsburg et al. 1988, Sever et al. 1991, Dubrovina et al. 1999). In the continuity of the work of Eppinger et al. (2007) who identified in the pVM82 a T4BSS harboring dot/icm genes, we investigated in this work the specificities of the yT4BSS as well as its contribution to virulence in Y. pseudotuberculosis lineage 8 strains.

By analyzing the structure of the yT4BSS gene cluster, we identified three new genes: icmT within the previously identified gene cluster, as well as two copies of icmX (icmX1 and icmX2) in a distant gene cluster on the pVM82. By comparing gene cluster homologies with known functional T4BSS systems, we found that yT4BSS genes display higher homology with a greater number of genes of the recently described T4BSS of P. putida compared to those of L. pneumophila and C. burnetii. The only genes missing in Y. pseudotuberculosis compared to the closest strain P. putida are icmW and dotU/icmH. In L. pneumophila, IcmW forms a stable chaperone protein complex with IcmS allowing translocation of IcmSW-dependent effectors (Ninio et al. 2005). Because icmS is also absent in the yT4BSS, we could hypothesize that effector translocation is only DotM-dependent or that other chaperone systems still need to be identified in Y. pseudotuberculosis. In L. pneumophila, DotU is dependent on IcmF, both integral inner membrane proteins, for T4BSS polar assembly and structural stability of the DotH, DotG and DotF subunits (Sexton et al. 2005, Ghosal et al. 2019). Because the yT4BSS lacks icmF and dotU, we could hypothesize that the yT4BSS might display structural specificities, different from those of the T4BSS in L. pneumophila. Thus, the dot/icm genes set of Y. pseudotuberculosis SP-1303 appears to be complete, encoding main structural and enzymatic components of a typical T4BSS, making it a potentially functional system.

Interestingly, the yT4BSS genes are highly homologous to those of the plant root colonizer P. putida, which functions as a contact-dependent bacterial killing machinery, targeting a wide range of soil and plant-associated Gram-negative bacteria, allowing P. putida to persist in the environment (Purtschert-Montenegro et al. 2022). Our results, however, suggest that the yT4BSS plays a role during infection rather than in environmental conditions. Indeed, by testing different in vitro, in cellulo, and in vivo conditions, we found that yT4BSS genes are continuously expressed in rich culture media and not in minimal medium, and also during Y. pseudotuberculosis intracellular life cycle, as well as during mouse infection without organ specificity. Two key parameters found to influence the rate of expression appear to be pH, with higher expression observed at acidic pH, and temperature, with higher expression rates encountered at a higher temperature (37°C) than lower temperatures (21°C, 28°C), suggesting a role for the yT4BSS in the mammalian host environment. Interestingly, high temperatures do not seem to be an inductive signal for dot/icm gene expression in other T4BSS. For example, within phagocytes, dot/icm and related effectors expression is induced by nutrient deficiency in the Legionella-containing vacuole, whereas acidic pH and specific amino acid composition (phenylalanin, proline, and serine) are signals inducing dot/icm gene expression in the Coxiella-containing vacuole (Al-Khodor et al. 2009, Newton et al. 2020, Jaboulay et al. 2021). Concerning P. putida, no gene expression study has been performed so far, but the T4BSS is functional in bacterial killing in environmental conditions (30°C) (Purtschert-Montenegro et al. 2022). In Y. pseudotuberculosis, many other virulence factors are also upregulated at 37°C, such as the T3SS and Yops, the majority of adhesins (YadA, Ail, PsaA), or a type IV pili, suggesting that the yT4BSS is an additional molecular virulence factor in the context of host infection.

With the goal of better characterizing the function of the yT4BSS, we investigated the presence of putative effectors in the Y. pseudotuberculosis SP-1303 genome. Using the S4TE algorithm, we were able to identify 11 candidates in the pVM82 and 5 in the chromosome and we confirmed their co-expression with the yT4BSS genes dotA and dotB. However, functional assessment of three of these candidates (B0054, B0071, and B0081) did not reveal secretion of these putative effectors in the bacterial extracellular medium nor intracellularly. As mentioned above, we have identified conditions (37°C and acidic pH) which contribute to activation of the operon encoding the yT4BSS and putative effectors; however, we ignore whether there are additional conditions which might contribute to trigger secretion. Thus, testing additional media compositions (including testing of different ion concentrations) could lead to identify conditions favoring yT4BSS secretion. The use of additional bioinformatic tools to predict putative yT4BSS effectors should be also considered in future studies.

To our surprise, even if the yT4BSS is expressed through the intracellular life cycle of Y. pseudotuberculosis, it does not take part in bacterial survival, nor in the biogenesis of YCVs in epithelial cells and macrophages, as evaluated upon inactivation of the dotB gene. Thus, the yT4BSS does not participate in modulation of intracellular trafficking in Y. pseudotuberculosis, as opposed to the T4BSS of the obligate intracellular pathogens L. pneumophila and C. burnetii, which are totally dependent on their T4BSS for their intracellular life cycle in phagocytes (Qiu and Luo 2017). Using fluorescent microscopy and TEM, we described the biogenesis of a single-membrane vacuole in HeLa cells, containing replicative Y. pseudotuberculosis, which increases in size over time, localizes close to the host nucleus, as it was also shown by Ligeon et al. (2014), independently of yT4BSS functions, and colocalizing with LC3B and with cathepsin D but without acidifying. To date, because no bacterial determinants have been linked to YCV biogenesis (Pizarro-Cerda et al. 2016), the use of a mutant library of Y. pseudotuberculosis isolates coupled to cellular infection analyses by microscopic screening would allow to fill this knowledge gap. Since our results are based on the analysis of the phenotype of a ΔdotB mutant, it would be also important to confirm our results by investigating the intracellular fate of bacterial mutants in other yT4BSS genes.

We found a role of the yT4BSS in Y. pseudotuberculosis FESLF strain pathogenicity during oral infection of mice but not during intravenous inoculation, demonstrating that the yT4BSS is a new virulence factor, whose function depends on the infection route. Interestingly, the superantigen YPMa from FESLF strains also acts as a virulence factor depending on the infection route. Unlike the yT4BSS, YPMa plays an important role in virulence during intravenous infection whereas it has no effect during intragastric infection (Carnoy et al. 2000). Recently, a route-dependent phenotype was also reported for the Y. pestis broad-range protease Pla, which contributes to virulence in the context of subcutaneous infections leading to bubonic plague, but not in the context of intravenous inoculation due to septicemia; this is explained by the fact that cleavage of plasminogen by Pla and subsequent manipulation of the fibrinolytic cascade allows bacteria to disseminate from draining lymph nodes to the blood circulation, a role that is bypassed when bacteria are inoculated directly in the blood stream (Sidhu et al. 2025). Nevertheless, the yT4BSS does not take part in bacterial colonization of organs or survival during oral infection, as demonstrated by similar bacterial burden between the WT strain and the ΔdotB mutant strain, which was also found to be similar between WT and ΔypmA mutant strain (Carnoy et al. 2000). Our current hypothesis is that the yT4BSS effectors target cells and functions that will have a role only in an anatomical location that is associated to the traversal of the intestinal barrier (i.e. MLNs). Future identification of yT4BSS effectors will allow to clarify this route-dependent phenotype.

Performing histological analyses on key organs during the infectious process, cecum and MLNs, we observed the involvement of the yT4BSS in the induction of high severity lesions corresponding to important areas of necrosis in both organs at sites of necrotizing granulomas. For the time being, we do not know if these necrotic lesions are the direct effect of yT4BSS functions on host cells or indirect consequences of yT4BSS functions. Necrosis has already been described in pyogranuloma in MLNs and along the gastrointestinal tract upon infection with a Y. pseudotuberculosis strain lacking the yT4BSS (El-Maraghi and Mair 1979, Sorobetea et al. 2023). Thus, the yT4BSS could synergize other key virulence factors to increase a basal phenotype that is associated with the Y. pseudotuberculosis species. In some cases, a bimodal infection phenotype is detected for both the WT SP-1303 strain and for the ΔdotB mutant, as observed, e.g. in the cecum concerning necrosis (Fig. 7c); bimodal phenotypes have been observed in other infection models and are explained as the outcome of variations in initial inter-individual initial conditions (host resources and/or immune responses) that drive dramatically different consequences (pathogen survival or pathogen destruction) at late infection times (Van Leeuwen et al. 2019, Eller et al. 2021, Luk et al. 2022). This phenotype is exacerbated concerning necrosis for the ΔdotB mutant in MLNs (Fig. 6i), which could be explained, as mentioned above, by modulation of expression of other virulence factors of Y. pseudotuberculosis which might drive in some cases a more pronounced infection phenotype depending on a specific set of initial conditions favoring late necrosis; the analysis of the transcriptome of the ΔdotB mutant in infection conditions in vitro (37°C) or in bacteria isolated from MLNs in vivo should allow for a better understanding of the impact of inactivating the yT4BSS on the expression of other critical Y. pseudotuberculosis virulence factors.

Thanks to cytokine profile analyses, we observed that the yT4BSS appears to participate in modulation of immune responses using a model of mBMDM infection in vitro, leading to higher levels of innate immunity related cytokines and chemokines production. Whereas IL-6 plays mostly pro-inflammatory roles, both IL-10 and CCL2 are associated with the M2 polarization of macrophages; it is considered that high production of such cytokines aims at reducing deleterious effects of inflammation and indeed M2 macrophages can be found in anti-inflammatory microenvironments associated with wound healing such as granulomas (Atri et al. 2018). In agreement, IL-10 and CCL2 production during Y. pseudotuberculosis lineage 8 infection could be involved in the process of tissue repair and granuloma formation around foci of bacteria that caused tissue lesions and necrosis in the MLNs and cecum. Interestingly, the yT4BSS also favors IL-33 which is an inducer of T helper type 2 immune responses which can act as a potent agent of tissue repair, but it can counterbalance T helper type 1 immune responses and has been associated to pathologic inflammation in MLNs (Zhu et al. 2017). Finally, IL-10 also counterbalances pro-inflammatory cytokines in many innate immunity and antimicrobial mechanisms: it is thus able to limit the host immunopathology in bacterial infections, but may also play a detrimental role when inflammation is protective (Leech et al. 2017) and this could contribute to maintain bacteria alive inside granulomas.

L. pneumophila, thanks to its T4BSS, also modulates the host immune response, inducing an increased pro-inflammatory cytokine response in a mBMDM infection model as well as in an intranasal mouse model, resulting in IL-1α and IL-1β production, which participate in neutrophil recruitment to the lungs of infected animals (Barry et al. 2013, Copenhaver et al. 2015). However, C. burnetii has opposite effect by silencing host innate immune responses in a T4BSS-dependent manner by downregulating NF-κB signaling pathways (Burette et al. 2020; (Mahapatra et al. 2016). Thus, the yT4BSS shares common functions with the T4BSS of L. pneumophila in increasing inflammatory cytokines responses, although the precise molecular mechanisms remain to be identified in Y. pseudotuberculosis FESLF strains. Y. pseudotuberculosis also depends on its T3SS for infection, which displays immunosuppressive properties, with YopJ, e.g. known to inhibit pro-inflammatory cytokine production (such as IL-6, IL-8, and TNF-α) in macrophages and epithelial cells, or YopM inhibiting production of IL-1β and TNF-α but increasing IL-10 production in macrophages (Kerschen et al. 2004, Berneking et al. 2023). Because of opposite impacts of yT4BSS and T3SS on the host cytokines responses, the interplay between both systems needs to be explored. We could hypothesize that depending on the phase of the disease, organ, or cell type, yT4BSS and T3SS together or sequentially, impact the host immune system, participating in Y. pseudotuberculosis pathogenicity. Sequential activation of secretion systems is encountered in the example of Salmonella enterica serovar typhimurium, starting with the T3SS from the Salmonella pathogenicity island 1 (SPI-1), allowing invasion of intestinal epithelial cells and induction of inflammatory responses, then within macrophages T3SS from SPI-2 is activated and in later phases of infection, its T6SS participates in persistent infection (Das et al. 2013). In Y. pseudotuberculosis FESLF strains, the superantigen YPMa has also an impact on the host immune system but acts on the adaptive immune response by binding specific CD4⁺ and CD8⁺ T cells, leading to their activation and expansion with high production of IL-2 and IL-4 (Uchiyama et al. 1993, Miyoshi-Akiyama et al. 1995, Seprényi et al. 1999, Gouard et al. 2015). In our study, IL-2 and IL-4 levels are not influenced by the presence of the yT4BSS. Thus, with an impact only during oral infection and on inflammation, the yT4BSS appears to participate in the first steps of infection, whereas with a virulence role only during intravenous infection and on T cell populations, YPMa acts later in the infection process, probably when bacteria reach the bloodstream, as ypmA expression is induced in liver and spleen (Goubard et al. 2015).

The yT4BSS appears to have an impact on the mammalian host at the immune response level. However, yT4BSS genes are highly homologous to those of the plant root colonizer P. putida, which functions as a contact-dependent bacterial killing machinery, targeting a wide range of soil and plant-associated Gram-negative bacteria, allowing P. putida to persist in its environment (Purtschert-Montenegro et al. 2022). As already described in the Vibrionaceae family with a T6SS harboring anti-bacterial and anti-eukaryotic activities, the potential properties in bacterial killing of the yT4BSS remain to be explored (Dar et al. 2018).

To conclude on this study, we demonstrated the involvement of a new virulence factor, a putative yT4BSS, in the pathogenicity of Y. pseudotuberculosis strains from lineage 8 responsible for FESLF. Although the system is not involved in the intracellular life cycle of Y. pseudotuberculosis, the yT4BSS is highly expressed under mammalian host temperatures and acidic pH, participates in induction of necrotic lesions in infected tissues and modulates host cytokine immune responses. Further studies characterizing the effectors secreted by the yT4BSS should allow to better understand its contribution to Y. pseudotuberculosis FESLF lineage 8 infection.

Conflicts of interest

The authors declare no competing interests.

Funding

J.P.C. received financial support from Institut Pasteur and LabEX Integrative Biology of Emerging Infectious Diseases (ANR-10_LBX-62-IBEID). M.L. received a Ph.D. fellowship from Fondation de la Recherche Médicale (FRM) and from Ministère de l’Education Nationale, de l’Enseignement Supérieur et de la Recherche (MENESR). J.B.C. received funding from FRM (ECO202306017408).

Data Availability

The data underlying this article are available in the article and in its online supplementary material.