Viral-bacterial co-infections screen in vitro reveals molecular processes affecting pathogen proliferation and host cell viability

Ethics statement

The research described in this study complies with all relevant ethical regulations. Primary bone-marrow derived macrophages (BMDMs) were generated using 8 to 12 week old C57CL/6, mice (both male and female) under the guidelines and approval from the Swiss animal protection law (licenses VD3257, Service des Affaires Vétérinaires, Direction Générale de l’Agriculture, de la Viticulture et des Affaires Vétérinaires, état de Vaud). All mice were bred and housed in a specific-pathogen-free facility at 22 ± 1 C° room temperature, 55 ± 10% humidity and a day/night cycle of 12 h/12 h at the University of Lausanne.

Bacterial and viral strains

All bacterial and viral strains used in this study, alongside their culture conditions and additional details on infection are listed in Table 1 (bacterial strains) and Table 2 (viral strains).

Cell lines and culture

The following cell models were used throughout this study and, unless specified otherwise, grown at 37 °C, 5% CO₂ in the indicated media: RAW264.7 macrophages (ATCC, TIB-71) were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM, Gibco), containing 10% Fetal Calf Serum (FCS, Thermo Scientific). BMDMs (harvested from wildtype mice) in DMEM, supplemented with 10% FCS and 20% recombinant Macrophage Colony Stimulating Factor (MCSF, produced in the lab from L929 cells). Immortalized BMDMs (iBMDMs, previously produced in the lab) in DMEM containing 10% FCS and 10% MCSF.

RAW264.7 and iBMDMs were maintained by regular splitting into fresh media in tissue-culture treated flasks (TPP), after scraping and washing cells, and controlling for viability. BMDMs were thawed and seeded into non-tissue culture treated dishes (Falcon) for one day. For seeding, cells were detached by incubation at 4 °C in pre-chilled Phosphate Buffer Saline (PBS, Gibco) for 15 minutes and counted in a Countess automated cell counter with Trypan Blue (Thermo Fisher Scientific) viability stain.

Screening of host-pathogen-pathogen interactions

The day prior to infection, 30,000 RAW264.7 cells were seeded per well of a 96-well plate and primed for at least 6 h with 10μg/ml m-IFNγ (Stemcell Technologies). Bacterial overnight cultures (see Table 1) were started by inoculating 3 ml LB-media containing appropriate antibiotics with a single colony from an agar plate, and grown at 30 °C with agitation. For simultaneous co-infection, media was replaced with DMEM containing viral particles at the MOI indicated in Table 2, which was determined prior to the screen to allow for dynamic range by titration and assessment of viral growth and cell death over time.

Cells were placed in a pre-heated centrifuge at 300 G, while the bacterial infection was prepared. Bacterial overnight cultures were adjusted by OD₆₀₀ and washed once in PBS. Assuming a host cell number of 50,000 cells per well, bacteria were added at an MOI of 50 directly onto the viral infection media, as a control, cells were treated with 100 μg/ml Lipo-Polysaccharide B5 (LPS, Invivogen). Cells were centrifuged for 5 minutes at 300 G and placed at 37 °C for 30 minutes. Subsequently, cells were washed once in pre-warmed PBS and the media was replaced with OptiMEM (Gibco) containing 20μg/ml gentamycin (Invitrogen), 12.5μg/ml propidium iodide (PI, Thermo Fisher Scientific) and 0.1% TritonX-100 (Tx100, ITW Reagents) where applicable, according to the following layout:

Column 1: Virally uninfected, +PI +Tx100; Columns 2-3: Virally uninfected, media only; Columns 4-6: Virally uninfected, +PI; Columns 7-9: Virus added, +PI; Columns 10-11: Virus added, media only; Column 12: Virus added, +PI +Tx100.

Rows: A: untreated, B: LPS control, C-H: each of the 6 bacterial pathogens.

In the case of subsequent infections, the viral pre-infection was performed by replacing the priming media with viral suspension, and centrifugation for 1 h at 37 °C and 300 G. The cells were maintained at 37 °C as indicated in the strain table, which had been determined before conducting the screen, prior to bacterial infection, which was performed as described above.

For dynamic measurement, plates were placed in a plate reader (Biotek Cytation 5, serial number: 1602037, software version: 03/04/2017) and fluorescence intensity was recorded in two channels (1: excitation at 479/20 nm, emission at 520/20 nm; 2: excitation at 550/20 nm, emission at 591/20 nm) in 10-minute intervals for at least 16 h at 37 °C, 5% CO₂. Due to the lack of GFP-tagging, viral growth could not be assessed fluorometrically for MNV strains, and alternative readout methods, such as plaque-forming unit quantification do neither allow for the necessary throughput nor for the required dynamic (i.e. time-resolved) measurement.

Data cleanup and primary analysis

Cell death was calculated based on the fluorescence measurements obtained in channel 1. Firstly, the average fluorescence intensity obtained in wells containing DMEM + gentamycin was subtracted for each condition individually to account for background signal of the fluorophores expressed by the bacterial pathogen. Secondly, using the Triton-X100 treated total lysis control (100%) and the uninfected control (0%) as reference points, cell death was quantified for each well. Thirdly, values were averaged across intervals of 30 minutes to reduce measurement artifacts. Finally, values below 0% or above 100% were set to 0% or 100% to maintain biological meaningfulness.

Similarly, pathogen growth was quantified using the intensity values obtained in wells containing DMEM + gentamycin (and hence no PI), using channel 1 for bacterial growth and channel 2 for viral growth, with respect to the uninfected controls. Averaging of values was performed over intervals of 50 minutes to reduce noise.

Calculation of dynamic metrics and epistatic effects

The expected curve was calculated based on single infections, assuming Bliss independence: For bacterial and viral growth, the comparison occurred directly between singly and doubly infected condition: If bacterial / viral growth occurred to la larger degree in the co-infection, the interaction is deemed a synergy, if the pathogen growth was reduced, an antagonism.

For cell death, the expected curve was calculated for each timepoint, as the product of the fraction of live cells quantified in the single infection controls. Dynamic metrics were calculated and defined as follows: max. death: the 98^th percentile of all values obtained for cell death; t_onset: the time when cell death increases to more than 2% of max. death (with respect to the minimum cell death); t₅₀: as t_onset, but for 50% of max. death; gradient: difference between max. death and min. death, divided by the difference between t_onset and t_end (when 98% of max. death are reached); AUC score: The integral of the curve, based on the average Riemann sum.

In a next step, epistatic effects were calculated as the difference between the observed and the expected value for each metric. Z-transformation was applied to normalize distributions, by dividing the difference to the mean value with the standard deviation across all values for a given metric. Representation as heat maps was done in GraphPad Prism (version 10.2.0) and all raw data, cleaned values and calculated scores are available on Mendeley Data ( as well as Source Data File). For the assessment of significance, multiple testing correction was applied.

Quality control and replicate reproducibility

To evaluate data quality and replicate reproducibility, the two biological replicates were compared by Pearson correlation of each interaction pair, as well as assessing p values of paired t-tests (see Mendeley Data as well as Source Data File). The second approach was taken to assess whether the two replicates behaved significantly differently. Additionally, the dynamic range was taken into account to determine if a third biological replicate was necessary to ensure data reliability. This was only the case for mAdV3 subsequent infection, for which the average across all three replicates was used for further analysis.

Lactate dehydrogenase (LDH) assay

To validate detected interactions in cell death, cells were seeded, primed and infected as described above. After the initial bacterial infection, the media was replaced with OptiMEM containing 20μg/ml gentamycin in all wells and the plate was maintained at 37 °C. At the timepoint that was to be validated, one uninfected and one virally infected well were chosen as total lysis control and Tx100 was added for a final concentration of 0.1% Tx100 for 15 min. 30μl supernatant per well were added to 30μl LDH reagent (Sigma Aldrich, prepared as indicated by the manufacturer) for 20 min. The reaction was stopped by adding 30μl 1 M HCl and the plate was measured on a Biotek plate reader at 490 nm.

Cell death for each well was calculated with respect to the uninfected (0% cell death) and the total lysis (100% cell death) controls. The total lysis control of the virally pre-infected sample was used as quality control to prevent artifacts from excessive cell death and reduction in host cell number prior to bacterial infection.

Quantification of colony forming units (CFUs)

Parallel to determining host cell death by LDH-release, CFUs were quantified to assess bacterial growth. To do so, all gentamycin-containing OptiMEM was removed, and cells were washed once in pre-warmed PBS. Then, 100μl 0.1% Tx100 were added to lyse host cells and serial 1:5 dilutions of the bacteria-containing lysate were prepared. 7μl of each dilution was spotted on an agar plate containing the appropriate selection antibiotic and grown at the appropriate condition for the respective pathogen. Colonies were counted in technical triplicates and CFUs/ml were quantified with respect to the dilution.

Calculation of validation rate

To assess the validation rate of the screening approach, we designed the following procedure to determine if interactions could be validated or not: Using LDH release (for cell death) and CFU counting (for bacterial growth) as orthogonal biochemical assays, we probed the directionality (synergistic, i.e. more LDH release / more CFUs, or antagonistic, i.e. less LDH release / fewer CFUs) of each tested interaction. We then compared this result to the Bliss score obtained in the screen. In total, 16 pathogen pairs were tested and assessed in both metrics, thus yielding a set of 32 interactions that were used for validation.

For LDH-release, the expected value was calculated by subtracting the product of the fractions of alive cells in the single infections from 100%. For bacterial growth, the expected value was adjusted to the total host cell number, as approximated by the comparison of the total lysis controls for the virally infected and uninfected samples. The validation Bliss score was then calculated as the difference between the observed and the expected outcome, divided by the expected outcome.

Stable isotope labeling of amino acids in cell culture (SILAC) during viral infection

To assess the newly synthesized proteins upon viral infection, we employed SILAC labeling, comparing virally infected cells with that of an uninfected population. To do so, iBMDMs were seeded in tissue-culture treated 6-well plates (1 million cells per well) and primed for at least 6 h with IFNγ. Infection with mAdV2 or mAdV3 was performed by adding 750μl viral suspension in iBMDM media containing R⁺¹⁰ (U-13C6 99%,15N2 99%, Cambridge Isotope Laboratories) and K⁺⁸ (U-13C6 99%; U-15N4 99%, Cambridge Isotope Laboratories) (SILAC heavy media), using iBMDM media containing R⁺⁶ (13C6, 99%, CIL) and K⁺⁴ (4,4,5,5-D4, 96-98%, CIL) (SILAC intermediate media) as uninfected control. At 16hpi, media was removed, cells were washed twice in PBS and subsequently lysed in 500 μl 100 mM Tris pH7.5, containing 4% SDS (FASP-buffer), supplemented with 10 mM DTT (Merck). Lysates were sonicated to shear DNA, boiled at 95 °C for 5 minutes and cleared by centrifugation at full speed for 10 min. Cleared lysates were further processed at the Protein Analysis Facility of the University of Lausanne.

After determination of protein concentration (tryptophan fluorescence method⁹⁵), H and I samples were mixed at an equimolar ratio (total: 100μg) and digested (SP3 method⁹⁶ using magnetic Sera-Mag Speedbeads (Cytiva 45152105050250, 50 mg/ml). To alkylate, proteins were treated with 32 mM iodoacetamide (final concentration) for 45 min at RT in the dark. Precipitation was done on beads (10:1 (w:w) ratio beads:material) using ethanol (final concentration: 60%), and after 3 washes with 80% ethanol, beads were digested in 100 mM ammonium bicarbonate containing 1μg trypsin (Promega #V5073), final volume 50 μl, for 2 h at 37 °C. The same amount of trypsin was added for an additional 1 h of digest. Supernatants were recovered and mixed with two sample volumes of isopropanol containing 1% TFA. Samples were desalted on a strong cation exchange (SCX) plate (Oasis MCX; Waters Corp., Milford, MA) by centrifugation, washed with isopropanol containing 1%TFA, eluted in 200μl 80% MeCN, 19% water, 1% (v/v) ammonia, and dried by centrifugal evaporation.

Fractionation and liquid chromatography / mass spectrometry (LC/MS)

Samples treated for fractionation and LC/MS as previously published by the Protein Analysis Facility at the University of Lausanne. In brief, samples were fractionated in 6 fractions using the Pierce High pH Reversed-Phase Peptide Fractionation Kit (Thermo Fisher Scientific). The fractions collected were in 7.5, 10, 12.5, 15, 20 and 50% acetonitrile in 0.1% triethylamine (~pH 10), redissolved in 2% acetonitrile with 0.5% TFA and used for LC-MS/MS analysis.

LC-MS/MS analysis was carried out on a Fusion Tribrid Orbitrap mass spectrometer (Thermo Fisher Scientific) connected through a nano-electrospray ion source to an Ultimate 3000 RSLCnano HPLC system (Dionex), via a FAIMS interface. Peptides were separated on a reversed-phase custom packed 45 cm C18 column (75μm ID, 100 Å, Reprosil Pur 1.9μm particles, Dr. Maisch, Germany, 4-90% acetonitrile gradient in 0.1% formic acid (total time 140 min)). Cycling through three compensation voltages (-40, -50, -60V) was used to acquire full MS survey scans at 120’000 resolution. A data-dependent acquisition method in the Xcalibur software (Thermo Fisher Scientific) was set up, optimizing the number of precursors selected (“top speed”) of charge 2+ to 5+ from each survey scan, while maintaining a fixed scan cycle of 1 s per FAIMS CV. Peptides were fragmented by higher energy collision dissociation (HCD) with a normalized energy of 32%. The precursor isolation window used was 1.6Th, and the MS² scans were done in the ion trap. The m/z of fragmented precursors was then dynamically excluded from selection during 60 s.

Proteomic data analysis and GO-term enrichment

As described in other recent publications⁹⁷, analysis was performed with MaxQuant 2.1.4.0⁹⁸, using the Andromeda search engine⁹⁹. The following modifications were selected: cysteine carbamidomethylation (fixed), methionine oxidation (variable), protein N-terminal acetylation (variable), SILAC heavy labeling (K⁺⁸ and R⁺¹⁰), SILAC intermediate labeling (K⁺⁴ and R⁺⁶). The mouse (Mus musculus) reference proteome based on the UniProt database (RefProt_Mus_musculus_20230301.fasta, from www.uniprot.org, version of January 2023, containing 55’309 sequences), and a “contaminant” database (most usual environmental contaminants, enzymes used for digestion) were used with a mass tolerance of 4.5ppm on precursors (after recalibration) and 20ppm on MS/MS fragments. All identifications were filtered at 1% false discovery rate (FDR) relative to hits against a decoy database (reversed protein sequences). Filtering and processing of MaxQuant outputs were performed using Perseus (version 1.6.15.0)¹⁰⁰, contaminants were removed, and SILAC ratios were log₂-transformed.

Heavy-to-intermediate-ratios were combined across replicates using the protein name to map across runs. Only proteins which had quantified ratios in at least two replicates were kept and p-values were calculated with respect to the null hypothesis that the logarithmic ratio between H and I was 0. For GO-term enrichment, selected proteins (logarithmic fold change larger 0.5, p-value smaller than 0.05) were queried for over-representation against the full list of GO terms for biological function (GO Ontology database: Released 2023-10-09) in the Panther database (PANTHER Overrepresentation Test (Released 20231017)), using the full mouse proteome as reference. Fisher’s exact test without correction was used for p-value analysis and only GO-terms (one per hierarchy group) with at least 3 proteins, a p-value smaller than 0.01 and a fold-enrichment larger than 3 were kept.

Sterile inflammasome stimulation with and without mAdV3 pre-infection

The day prior to the experiment, BMDMs were seeded in 96-well microscopy plates (Greiner) and stimulated the next morning with IFNγ, at least 6 h prior to viral infection with mAdV3 (MOI of 5), which was performed as described above. Cells were left in the incubator overnight and subsequently treated with 100μg/ml LPS, by direct addition to the viral suspension / uninfected control. After 6 h, media was removed and inflammasomes were stimulated by treatment with 5μg/ml nigericin (Sigma-Aldrich) or 12.5μM UCN-01 (Sigma Aldrich), transfection with 125 ng/ml poly-dA-dT (using Lipofectamine LTX, Thermo Fisher Scientific) or infection with Salmonella which were subcultured for 3.5 h (1.4 600 

The remaining supernatant was discarded, and the cells were washed twice in PBS. Cells were fixed with 4% paraformaldehyde (PFA, Electron Microscopy Sciences) for 10 minutes at RT and cell membranes were solubilized with 0.1% Tx100, 1% BSA (Thermo Scientific) in PBS. Antibody staining (anti-ASC, Adipogen) was performed at 1:1000 dilution overnight at 4 °C in solubilization buffer and plates were subsequently washed and co-incubated in secondary antibody (Donkey-anti-Rabbit-Alexa-647, Invitrogen, 1:5000) in solubilization buffer at RT for 1 h.

To stain nuclei, Hoechst 33342 (Fisher Scientific) was added for 30 min at a 1:5000 dilution, and cells were washed three times prior to image acquisition. For microscopy, images (z-stack spanning the entire cell) were taken at 20x magnification on a Zeiss LSM800 confocal laser scanning microscope using Zeiss Zen Blue software (version 3.8) with the appropriate laser wavelengths and filters. At least 3 fields of view (z-stacks) were acquired for each biological replicate and ASC-speck-positive cells were quantified by automated counting of total cell number and cells with ASC-specks using ImageJ (version 1.53t), after summing all planes of each z-stack.

ELISA was performed according to the manufacturer’s protocol and sample input was diluted where required, according to LDH-release. Plates were read at 450 nm and 570 nm and the concentration of IL-1ß was calculated with respect to a standard curve using the absorption difference between the two channels, spanning 15.625 to 1000 pg/ml (quadratic approximation).

Analysis of mAdV2-infected populations by FACS

To quantify bacterial uptake, BMDMs were seeded the day prior to viral infection in non-tissue culture treated 24-well plates (Eppendorff) and primed the next morning with IFNγ, at least 6 h before mAdV2 infection at an MOI of 1, overnight. Bacteria were cultured overnight (at 37 °C, Yersinia at 26 °C) while shaking and Yersinia strains and mutants were subcultured (1:25 dilution) at 26 °C for 1 h and 37 °C for 1 h to induce the T3SS. Bacterial infection at MOI 5 were performed as described above, and the cells were harvested directly after invasion by incubating the infected cells for 15 minutes in chilled PBS at 4 °C. Cells were detached and washed once in PBS, and subsequently stained with violet fixable violet live-dead dye (Thermo Scientific) as described by the manufacturer. Cells were washed once more and fixed in 4% PFA for 10 minutes. Subsequently, cells were transferred into 96-well U-bottom FACS plates (Brand) and flow cytometry was performed on a Cytoflex S (Beckman Coulter), where at least 20000 events were recorded for each condition in each replicate. Cells were gated by forward- and side-scatter (Cells), as well as side-scatter height and area (Single Cells) and by PB450-negativity (live cells). FITC-signal (virus) and ECD-signal (bacteria) were used for the quantification of uninfected, singly, and doubly infected cells, with respect to an uninfected control.

siRNA-mediated knockdown of Mprip

Pooled siRNAs for Mprip were ordered as SMARTpool from siGENOME which contains the following four siRNAs: GAUCAUCAGUGGGUGGUUA, GGAAAUGGCAGCGACGAUU, GGAUGGUGGUCGGAAAGUA, GCAAGUGUCAGAACUGCUU (M-058568-00-0005). Prior to the siRNA transfection, BMDMs were seeded in the appropriate format (50000 cells per well for 96-well plates (microscopy suited), 250000 cells per well for non-tissue culture treated 24-well plates). The transfection mix was prepared by mixing 100μl OptiMEM with 3μl XtremeGENE 9 transfection reagent (Sigma Aldrich) 0.25μM SMARTpool, and left incubating at RT for 15 minutes. 10μl were added to each well of a 96-well plate (50μl for 24-well plates) and cells were grown at 37 °C for 48 h. A pool of 4 non-targeting siRNAs was used as control and transfected using the same procedure.

Knockdown validation: Polyacrylamide Gel Electrophoresis (PAGE) and Western Blot

Cells were harvested through lysis in FASP-buffer, containing 5% ß-mercaptoethanol and sonicated to shear DNA. Samples were boiled for 5 min at 95 °C, subsequently loaded onto a precast Bis-Tris 4-12% gradient gel (mPAGE, Merck Millipore), and PAGE was performed at 120 V for 1 h. Semi-wet transfer onto Amersham Protran nitrocellulose membrane (Sigma Aldrich) was conducted in a Biorad Trans-Blot Turbo system using transfer buffer (25 mM Tris, 192 mM Glycine, pH8.3 containing 20% methanol): 10 minutes at 1.3 A, 25 V. Membranes were washed in TBS containing 0.1% Tween−20 (Sigma, TBS-T), and subsequently blocked for 1 h in 5% milk in TBS-T while rocking at RT. Rabbit-anti-Mprip (Thermo Scientific) primary antibody was added at 1:1000 dilution and membranes were incubated overnight at 4 °C while rocking. The next day, membranes were washed three times for 5 minutes in TBS-T and subsequently incubated for 1 h in HRP-coupled secondary antibody (goat-anti-rabbit, 1:5000 in milk, SouthernBiotech) or in HRP-coupled anti-Tubulin antibody (Abcam, 1:5000 in milk). After three washes, membranes were incubated in ECL solution (BioRad) and images were acquired on an iBright Imaging System (Thermo Fisher Scientific), using an appropriate exposure time.

Microscopy image acquisition and quantification after mAdV2-co-infections

Infections, plate preparation and image acquisition were performed as described above. Additionally to Hoechst, cells were stained with phalloidin-Alexa647 (Abcam) for 1 h at a concentration of 1:1000 in PBS. After image acquisition (4 fields of view, in a z-stack spanning at least 7 planes per condition in each replicate), the integrated intensities for DAPI, bacteria and phalloidin were quantified across planes in ImageJ. Subsequently, the phalloidin signal was normalized to DAPI (to quantify increase in actin intensity upon viral infection), and the bacterial signal was normalized to both phalloidin and DAPI (to assess changes in intracellular bacteria).

Statistical analysis

All analyses, significance testing and visualization was performed in Prism (version 10.2.0), taking into account the necessary prerequisites in test selection, such as multiple testing correction, Welsh correction, (un-)pairedness of the samples and normalization of the values prior to testing. Details of statistical analysis are indicated where required.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.