Bacterial strains and plasmids used in this study are listed in Supplementary Files S2 , S3 and primers in Supplementary File S4 , respectively.

The pMADΔactA plasmid used to remove actA was previously described (Henke et al., 2015). To generate the pHOSSΔinlAB plasmid, flanking regions of the inlAB region including the first 12 N-terminal nucleotides of inlA and the last 36 C-terminal nucleotides of inlB were amplified with GoTaq^®DNA polymerase (Promega Corporation, Madison, USA) using the primer pairs ΔinlA_n_fw_1_SalI/ΔinlAB_2_rv (617pb) and ΔinlAB_3_F/ΔinlB_R_XmaI, respectively. The two resulting fragments were then ligated using T4 DNA ligase (Thermo Fisher Scientific Inc., Waltham, USA) following digestion of pHOSS1 (Abdelhamed et al., 2015) with XmaI (New England Labs Inc., Ipswich, MA, USA) and SalI (New England Labs Inc.) according to the manufacturer’s instruction.

The pMADΔfosX plasmid for deletion of fosX was generated by removing the fosX sequence from a pMADfosX plasmid containing fosX with its flanking regions. The fosX fragment was generated by amplification from JF5203 with Hifi cloning (Takarabio Inc.) and purification with the NucleoSpin^® Gel and PCR Clean-up Kit (Macherey-Nagel, Düren, Germany) after gel electrophoresis. Both insertion of the resulting fragment into pMADfosX and subsequent deletion of the fosX sequence from pMADfosX to generate pMADΔfosX were performed with the In-fusion^® HD Cloning Plus (TakaraBio Inc.) technology according to the manufacturer recommendations. For In-fusion^® cloning, the amplification primer pairs pMADfosX3_F/pMADfosX4_R, pMADfosX1_F/pMADfosX2_R and pMADΔfosX1_F/pMADΔfosX2_R were used to generate pMADfosX and pMADΔfosX, respectively. pMADfosX and pMADΔfosX plasmids were Sanger sequenced (3730 DNA Analyzer, Thermo Fisher Scientific Inc., Waltham, USA) using the BigDye^® Terminator v3.1 Cycle Sequencing Kit (Thermo Fisher Scientific Inc., Waltham, USA). The primers used for sequencing (fosX_F, fosX_R, pMAD_F, pMAD_R, fosX_ex_F, fosX_ex_R) are listed in Supplementary File S4 .

For generation of the pMAD_NactA100AA_SAG1_HIS plasmid, NactA100AA_SAG1_HIS was created in-silico using the Geneious 8.1 software (Biomatters Inc.) and synthesized and inserted into pMAD by Twist Bioscience (San Francisco, USA). Briefly, the sag1 sequence NCLIV_033230 was codon-optimized for L. monocytogenes using a publicly available web-based software (http://www.jcat.de/). At the sag1 5’ end, the first 300 nucleotides of actA were added and the 3’ end was fused with a HIS tag followed by 2 stop codons. In order to ensure insertion of sag1 into the actA locus, this sequence was inserted between the two actA flanking regions of the pMADΔactA plasmid resulting in pMAD_NactA100AA_SAG1_HIS. For deletion of the HIS-tag to generate pMAD_NactA100AA_SAG1 (Map in Suplementary File S1 ) the fragment without HIS tag was amplified with the Hifi cloning system and integrated into pMAD with the In-fusion^® HD Cloning Plus using the primer pair SAG1HISdel_F/SAG1HISdel_R ( Supplementary File S4 ).

In order to prevent permanent insertion of antibiotic resistance genes into the vector, all mutant strains were generated from L. monocytogenes strain JF5203 (NCBI Reference Sequence: NZ_LT985474.1; https://www.ncbi.nlm.nih.gov/nuccore/NZ_LT985474.1; lineage I, CC 1, ST 1, serotype 4b) by homologous recombination with either pMAD or pHOSS1 plasmid derivates ( Supplementary File S3 ). The JF5203ΔactA mutant generated in a previous study (Henke et al., 2015) was used as parental mutant to engineer the vaccine vector. The JF5203ΔactA/inlA/inlB/fosX vector (Lm3Dx) was generated by sequential in-frame deletion of inlA/inlB and fosX genes from the JF5203ΔactA mutant by homologous recombination with the respective pMAD (Arnaud et al., 2004) and pHOSS1 (Abdelhamed et al., 2015) plasmid derivates described above. Gene deletions were confirmed by DNA sequencing of the intermediate mutants and by full genome sequencing of the final vaccine vector strain (see below).

Finally, Lm3Dx was transformed with the plasmid pMAD_NactA100AA_SAG1 ( Supplementary File S1 ) to create Lm3Dx_SAG1 expressing sag1 fused to the 300 first nucleotides of actA under the control of the actA promotor.

Fosfomycin minimal inhibitory concentrations (MICs) were determined in Lm WT and Lm3Dx using the fosfomycin E-test strip (Biomérieux SA, France) after 24h incubation at 37°C as previously described (Scortti et al., 2006). Genetic stability of ΔactA mutant and Lm3Dx was determined in comparison to the WT by serial passaging under non-selective conditions. Strains were cultured in 10mL of Bacto Brain Heart Infusion broth (BHI, Sigma-Aldrich, Buchs, Switzerland) and sub-cultured every 12 h at a 1/10’000 dilution into fresh BHI broth. Every 10 passages, 10µL of the medium was plated on BHI-Agar plate (Sigma-Aldrich, Buchs, Switzerland) and cultured 24h at 37°C before further passaging was done. Every ten passages until passage number 100, a colony of the plating performed on BHI-Agar was grown in 10mL of BHI broth and DNA was extracted using the DNA extraction kit (Invitrogen, PureLink™ Microbiome, DNA purification Kit). PCR of extracted DNA was performed to ensure stability of the deletion of each virulence gene (primers used are listed in Supplementary File S4 ). Insertion of the sag1 sequence into Lm3Dx was confirmed by colony PCR using the actA flanking region primer pairs ( Supplementary File S4 ), resulting in Lm3Dx_SAG1. Furthermore, both Lm3Dx and Lm3Dx_SAG1 were sequenced by INVIEW Resequencing (Eurofins Genomics, Constance, Germany) after the first and 100^th passage in BHI broth to analyze for single nucleotide point mutations (SNPs) (https://www.ebi.ac.uk/ena/browser/view/PRJEB43038).

For confirmation of NcSAG1 expression at the RNA level, Lm3Dx_SAG1 was cultured in BHI broth overnight at 37°C. The next day, RNA was extracted with the RNA extraction kit (RiboPure™ Bacteria, Invitrogen, Thermo Fisher Inc., Waltham, USA) and contaminating genomic DNA was eliminated with RQ1 RNaseFree DNase (Promega Corporation, Madison, USA). Purified RNA was then reverse transcribed into cDNA with GoScript™ Reverse Transcriptase (Promega Corporation, Madison, USA) following the instruction of the manufacturer. PCR for NcSAG1 was performed using NcSAG1_F/NcSAG1_R primers ( Supplementary File S4 ). NcSAG1 protein expression in Lm3Dx_SAG1 during infection was confirmed by WB of proteins extracted from infected canine histiocytic DH82 cells. Briefly, DH82 cells were cultured to confluency in Corning^® Cell culture T150 flasks (Fisher scientific Inc., Waltham, USA) and infected at a multiplicity of infection (MOI) of 40. Six hours post-infection, infected cells were detached by Trypsin-EDTA (Sigma-Aldrich, Buchs, Switzerland), centrifuged and the pellet was frozen at -80°C. After thawing, cells were lysed by sonication (3 cycles of 2 minutes at 40% duty) (Sonifier 450, Branson, Danbury, CT, USA) and proteins were separated from the cell lysate by centrifugation (15.000 x g, 15 minutes at 4°C). Fifteen µg each of the unpurified protein extract, of Nc protein extract as positive control and of the empty Lm3Dx vector as a negative control, were separated by SDS-Page and electrophoretically transferred to a nitrocellulose membrane. The membrane was blocked in PBS/0.1% Tween-20/2% milk powder for 1 h at RT, and the monoclonal mouse anti-NcSAG1 antibody (kindly provided by Prof. Dr. Camilla Björkman, University of Uppsala, Sweden) (Björkman and Hemphill, 1998) was applied at a dilution of 1:1000 overnight at 4°C. Membranes were washed four times with PBS supplemented with 0.1% Tween-20, and were incubated with the goat anti-mouse IgG coupled to alkaline phosphatase (AP) (Promega, Madison, USA), diluted 1:2000 in PBS-Tween-20. Following four more washes in PBS, bound antibodies were visualized by developing the AP reaction using 0.3% NBT/BCIP stock solution (Roche, Basel, Switzerland) in 0.1 M sodium chloride, 5 mM magnesium chloride, and 0.1 M Tris (pH 9.5). The enzymatic reaction was stopped by immersing the membrane in distilled water.

To check whether virulence gene deletion attenuates cellular infection of the vaccine vector Lm3Dx and whether attenuation is cell-type specific, canine cell lines representing epithelial cells (Madin-Darby Canine Kidney, MDCK) and macrophages (DH82) were infected. The day prior to infection, both cell lines were seeded in 24-well plates at densities of 0.8 x 10⁵ (MDCK) or 2.5 x 10⁵ cells (DH82) per well, respectively, and were grown in DMEM medium supplemented with 10% FCS and 50 µg/mL Penicillin/Streptomycin at 37°C with 5% CO₂. One hour prior to infection, cells were starved in FCS- and antibiotic-free medium. Cells were then infected with Lm WT, Lm ΔactA, Lm3Dx and Listeria innocua at a MOI of 1.25. The concentration of the inoculum was determined by OD₆₀₀ measurement on a Biospectrometer^® basic (Vaudaux-Eppendorf AG, Schönenbuch/Basel, Switzerland) and confirmed by plating of the inoculum on brain heart infusion (BHI) plates (Sigma-Aldrich). After 1h, cells were washed once with PBS and DMEM medium supplemented with FCS (2% for long-term infection assay or 10% for short-term infection assay) and 50 µg/mL gentamicin (Sigma-Aldrich) was added to prevent extracellular replication.

Colony-forming units (CFU) evaluation for the short-term infection assays was performed as previously reported (Rupp et al., 2015). Briefly, infected cells were washed once with PBS and then harvested in 100 µl of 0.5% Triton X-100 (Sigma–Aldrich) in distilled water (dH2O) 2, 4, 6, 24, 48 and 72 h post-infection (pi). Ten-fold serial dilutions of the lysates were plated on BHI-agar plates, and CFUs per well were calculated. At least three independent experiments were performed in triplicate.

For long-term infection assays, cells were infected as described above, but had to be sub-cultured due to cellular proliferation. At determined timepoints (2, 5, 7, 9 and 12 days pi) cells were washed once with PBS and then detached using 100 µL of Trypsin-EDTA solution (Sigma-Aldrich). Half of the cell suspension was transferred in a new 24-well plate with fresh DMEM medium supplemented with 2% FCS and 50µg/mL gentamicin for sub-culturing. The other half was incubated with 2% Triton X-100 (Sigma-Aldrich) in dH₂O for 10 min and then plated in serial dilutions as described above. At the final timepoint (14 days pi) cells were harvested in 100 µL of 2% Triton X-100 and further processed as described above. At least three independent experiments were performed in triplicate.

Coverslips with infected cells (DH82 or MDCK) were collected at 2, 6, 24, 48 and 72 hours pi and were fixed in 4% paraformaldehyde (PFA, Sigma–Aldrich) at 37°C for 15 min. Immunofluorescence was performed as previously reported (Rupp et al., 2015). Coverslips were incubated with Listeria O antiserum (BD Difco, 1:1000 in PBS-T) and phalloidin DyLight 633 (Fisher Scientific, Waltham, MA, USA, 1:500) in PBS-T with 10% normal goat serum (NGS, Agilent Technologies, CA, USA). After 1 h, coverslips were washed three times with PBS-T, followed by a 45 min incubation with Alexa Fluor 488 goat anti-rabbit IgG (Life Technologies, 1:500) and DAPI (1:10,000) in PBS-T supplemented with 10% NGS. Coverslips were washed twice with PBS and finally rinsed with dH₂O and mounted using glycergel (Dako). Specimens were imaged using a FV3000 confocal laser scanning microscope (Olympus, Hamburg, Germany) and were analyzed using FIJI (Schindelin et al., 2012).

All mouse experiments were conducted in compliance with the Swiss and European Union (Directive no. 2010/63/EU) legislation for the use of animals for scientific purposes, and animal protocols were approved by the Bernese Animal Welfare Committee (licenses BE113/19 and BE103/20). For the experiments, 6-week-old female and male BALB/c as well as synchronized pregnant mice were obtained from Charles River Laboratories (Sulzfeld, Germany). All mice were maintained under conventional day/night cycle in a temperature-controlled room excepted for the mice used in the milk excretion experiment, where the cages were kept under a sterile hood after vaccination, with water and food ad libitum. Two weeks upon arrival at the facility, mice were randomly distributed into groups and individually marked. One additional week of acclimatization was then given to the mice prior to start of the experiment.

For each experiment inoculation doses of Lm3Dx_SAG1 or WT strain were prepared by culturing a colony overnight in 10 mL BHI broth at 37°C. Bacteria were washed 3x with sterile PBS (Sigma-Aldrich) to remove BHI-broth. Prior to injection, they were resuspended to an OD₆₀₀ of 0.8 and sequentially diluted to the desired concentration in each experiment (1 x 10⁶, 5 x 10⁶ and/or 1 x 10⁷ CFU/50 µL). Concentration was confirmed by plating of the inoculum on BHI-Agar plates (Sigma-Aldrich) and CFU counting 24 h after incubation at 37°C. Euthanasia was performed in a euthanasia chamber using isoflurane and CO₂ or cervical dislocation for pups younger than 3 weeks old.

Experiment 1: Biosafety evaluation of Lm3Dx_SAG1 in non-pregnant BALB/c female mice.

Mice were randomly allocated into experimental groups of 5 to 6 mice each, with an additional mouse per group used as sentinel. Three groups were injected i.m. with 50 µL sterile PBS containing the Lm3Dx_SAG1 vaccine strain at different dosages (1 x 10⁶, 5 x 10⁶, 1 x 10⁷ CFU), one group with the WT strain (1 x 10⁷ CFU) and the negative control group with 50 µL sterile PBS. Sentinel mice were used to evaluate Lm transmission between mice in the same experimental cage. Mice were immunized i.m. in the thigh. Lm3Dx_SAG1 groups received a primovaccination and two booster injections, both two weeks apart, and were euthanized 2 weeks following the last immunization ( Table 1 ). Because the group receiving WT (1 x 10⁷ CFU) had to be euthanized 1.5d after the first immunization due to severity of clinical signs, 5 additional mice were inoculated i.m. with 50 µl PBS containing 1x10⁷ Lm3Dx_SAG1 and euthanized 1.5d after inoculation to assess the attenuation of our Lm3Dx_SAG1 vaccine strain during the acute phase. At euthanasia, liver, spleen, kidney, uterus, brain, thigh musculature and GIT of the Lm3Dx_SAG1 vaccinated mice were retrieved for further evaluation.

Experiment 2: Biosafety evaluation of Lm3Dx_SAG1 in lactating BALB/c dams and suckling pups.

To assess Lm3Dx_SAG1 excretion in milk and the safety of its use in lactating dams, we randomly allocated 14 dams with their respective pups (2-8 days old) into two groups (7 per group). Based on the results of the first experiment, the Lm3Dx_SAG1 group was injected once with 1 x 10⁷ CFU of Lm3Dx_SAG1 and the other group with PBS in the thigh musculature. From day 1 to day 4 after inoculation, dams were separated from their pups 1 hour prior to milking once a day ( Table 1 ). Mice were anesthetized with isoflurane and 1 IU of oxytocin (Sigma) was injected intraperitoneally. Ten minutes later, milk collection was started using a custom-made vacuum pump-based milk collection system. Upon milking, dams were returned to their pups in a cage under a sterile hood until the following day. Dams and pups were euthanized at day 4 pi. The liver, spleen, thigh musculature and GIT of the Lm3Dx_SAG1 vaccinated dams and their pups were retrieved for further evaluation.

Experiment 3: Assessment of potential effects of Lm3Dx_SAG1 on pregnancy outcome in BALB/c mice.

The experiment included three experimental groups of 6 mice each. Two groups were inoculated three times at two-week intervals with 50 µl PBS containing the Lm3Dx_SAG1 vaccine strain at 1x10⁶ and 1x10⁷ CFU respectively, while the control group was inoculated with 50 µL sterile PBS at each vaccination ( Table 1 ). The first vaccination was done at day 0 and the first booster on day 14. On day 22, mice were estrus-synchronized using the Whitten effect (Whitten, 1957) and were mated from day 25 to 27 by housing two females and one male in the same cage. Subsequently, males were removed from the cage and the third immunization was performed at day 28 (early pregnancy). Data on fertility (number of female mice that became pregnant), litter size (number of delivered pups per dam), stillborn mice and neonatal mortality (number of dead pups from birth until day 2 postpartum (p.p.)); postnatal mortality (number of dead pups from day 3 p.p. to the end of the experiment) and clinical signs that occurred during the observation phase were recorded. Dams and surviving pups were monitored at least 21 days after birth to rule out possible adverse effects due to treatments. At euthanasia, liver, spleen, injected thigh musculature and GIT of the Lm3Dx_SAG1 vaccinated dams and PBS control group were retrieved for further evaluation.

Mice in experiments 1 and 3 were monitored for fecal shedding as follows. In experiment 1, fecal sampling (100 mg/mouse) and analysis was performed daily from the day prior to the first vaccine administration until day 7 pi and at day 10 pi. Then, the same sampling scheme was repeated for the two booster injections ( Table 1 ). In experiment 3, fecal samples of every mice were retrieved 1, 2, 4 and 7 days after each vaccine dose administration and at day 11 and 14 after the third vaccine administration.

Fecal samples were homogenized in Half-Fraser broth (Fisher Scientific) at a 1:10 dilution using 5 mm glass beads (Sigma-Aldrich) in a TissueLyser II (Quiagen, MD, USA). Ten-fold serial dilutions were plated on ALOA Agar plates (Biomérieux, France) and incubated 2 days at 37°C prior to quantification of CFUs. Additionally, bacteria were enriched by culturing fecal homogenates for 24 h at 37°C in Half-Fraser broth followed by plating on ALOA Agar Plates for qualitative evaluation. Ten-fold serial dilution of milk probes were also plated on ALOA Agar plates and enriched in Half-Fraser broth as described for fecal samples. Bacterial isolates were confirmed by phenotype on ALOA Agar plates (round, blue-green colony with an opaque halo) and colony PCR amplification with the primer pairs hly_F/hly_R, actA_F/actA_R, inlA_F/inlA_R, inlB_F/inlB_R, fosX_F/fosX_R and NcSAG1_F/NcSAG1_R ( Supplementary File S4 ). CFU sensitivity was set at 10³ CFU/gram feces or milk for the quantitative assay without prior enrichment and at 10² CFU/gram feces or milk for the qualitative assay after enrichment in Half-Fraser broth followed by plating on ALOA Agar plates.

For assessment of bacterial intra-host spread, organs of female mice (experiments 1-3) and pups (experiment 2) were collected under sterile conditions immediately after euthanasia to evaluate the presence of Lm in organs. Organs were divided for immersion-fixation in 4% formaldehyde (for subsequent histological and immunohistological analysis) and for homogenization and enrichment in Half-Fraser Broth (1:10) as described previously for fecal analysis. Additionally, in experiment 3, pups which died in the postnatal phase (before endpoint analysis), were weighted and homogenized in whole in Half-Fraser broth (1:10) using an ULTRA-TURRAX^® Tube Drive (Faust AG, Schaffhausen, Switzerland) and serial dilutions were plated on ALOA plates with and without prior enrichment in Half-Fraser broth as described above for qualitative and quantitative evaluation. At endpoint analysis, three weeks after birth, spleen, part of the liver and the gastrointestinal tract (GIT) of every surviving pup were pooled, weighted and homogenized 1:10 in Half-Fraser broth for quantitative and qualitative CFU evaluation as described above.

To screen for organ lesions caused by Lm WT and Lm3Dx_SAG1, formalin fixed and paraffin embedded muscle, liver, spleen, brain, kidney and uterus of adult mice used in all three experiments were stained with hematoxylin and eosin (H&E) and microscopically analyzed. In experiment 2, histopathology of spleen, liver and GIT of pups was performed. Additionally, most sections were screened for persistence of the vaccine vector by using an immunohistochemical protocol for detection of Listeria. Immunohistochemistry was performed as previously described (Oevermann et al., 2010) with a polyclonal rabbit antibody against Listeria (1:200, Difco Laboratories, Detroit MI, USA).

To investigate the cellular immune response against Lm3Dx_SAG1, spleens of female mice from the PBS and the different Lm3Dx_SAG1 groups of experiments 1 and 3 were used for splenocyte re-stimulation assays. Following euthanasia, spleen samples were immersed in 5 mL of complete RPMI medium (RPMI 1640 medium containing 1% antibiotic: 100 U of penicillin, 100 μl of streptomycin, 2 mM L-glutamine, 10% FCS and 55 μM β-mercaptoethanol). Splenocyte isolation was performed as previously described (Aguado-Martínez et al., 2016a). Following lysis of red blood cells and two washing steps with complete RPMI medium, cell suspensions containing the splenic lymphocytes were seeded in a 24-well plate at a density of 2 x 10⁶ cells/well.

Ex vivo stimulation of the seeded splenic lymphocytes was performed with concanavalin A (ConA; 5 μg/mL; Sigma-Aldrich) as positive control, with N. caninum antigen lysate (20 μg/mL) or recombinant NcSAG1 (20 µg/mL), respectively. Unstimulated splenocytes were used as negative control. Following culture at 37°C/5% CO₂ during 72 h, medium supernatants of stimulated and unstimulated splenocytes were collected and levels of IFN-γ and IL-5 were measured in the supernatants using the BD OptEIA™ mouse ELISA kits (BD Biosciences, CA, USA) according to manufacturer’s instructions (Aguado-Martínez et al., 2016a). A Luminex instrument (Bio-PlexTM 200 system) was used to run the microtiter filter plates.

The expression and purification of recombinant NcSAG1 used for stimulation in our experiments was performed as follows. Firstly, the coding sequence of NcSAG1 was cloned in the pET151 TOPO^® expression construct (ThermoFisher Scientific, MA, USA). NcSAG1 was expressed in BL21 Star™ (DE3) One Shot^® Escherichia coli strain and purified afterwards by Ni²⁺ affinity chromatography. For the purification of NcSAG1 Protino^® Ni-IDA 1000 packed columns (Macherey-Nagel, Düren, Germany) were used, and proteins were purified under denaturing conditions according to manufacturer’s instructions. Purity of NcSAG1 was confirmed by SDS-PAGE and the protein concentration was measured with a BCA Protein Assay Kit (Pierce™).

To determine the humoral response against NcSAG1 in experiment 1 blood samples were collected from the tail vein one day before each immunization and by cardiac puncture 14 days after the 2^nd booster, immediately after euthanasia. Blood of the WT group was sampled only prior to the first inoculation and immediately after euthanasia ( Table 1 ). In experiment 3, to avoid stress of pregnant dams, blood was sampled only two days prior to the first vaccine application and following euthanasia, 42 days after the second booster. Blood of surviving pups was also collected by cardiac puncture. Anti-NcSAG1 IgG levels were determined by ELISA according to Debache et al. (Debache et al., 2008). In brief, 96-well plates were coated with recombinant NcSAG1 (100 ng/well) and incubated overnight at 4°C. The following day, plates were washed three times with washing buffer (0.05% PBS-Tween-20) before antigens were blocked with 1% bovine serum albumin (BSA) in 0.05% PBS-Tween-20 for 2h at RT. Subsequently, sera of individual mice at a dilution of 1:50 and two-fold sequential dilution up to 1:400 were added to individual wells. Following three washes, secondary antibody (goat anti-mouse IgG conjugated to AP (Promega)) was applied at a dilution of 1:2000. Development was performed with AP substrate buffer (1mg/mL). Absorbance was measured as optical density (OD) at 405 nm in a microplate reader (EnSpire™ 2300 Multilabel Reader, Switzerland). For evaluation, OD values were converted into relative index per cent (RIPC) values using the following formula [RIPC = (OD_405nm sample – OD_405nm negative control)/(OD_405nm positive control – OD_405nm negative control)*100 (Aguado-Martínez et al., 2016b). Serum of Nc infected mice was used as positive control.