Experimental design: mice cohorts

Male C57BL/6 J mice (strain code = #632, IMSR_JAX #000664, Charles River, L’Arbresle, France) and VertX mice containing Il10-eGFP-reporter (IMSR_JAX #014530, The Jackson Laboratory, ME, USA), aged between 8 to 10 weeks and weighing from ~21 to 29 grams were used in this project. VertX mice presented with an internal ribosome entry site (IRES)-enhanced green fluorescent protein (eGFP) fusion protein placed downstream of exon 5 of the interleukin 10 (Il10) gene²⁰. Mice were housed one week before the experiments in a conventional animal facility (Service Commun des Animaleries de Rockefeller, Lyon, France and a PBES-specific pathogen-free rodent facility (ENS Lyon)). For each experiment, mice (total n = 295) from the same cage were randomly assigned to two groups: animals with polymicrobial sepsis induced by CLP (total n = 172) and control mice (Sham, total n = 123). They were kept together after surgery and until euthanasia. Experiments were systematically repeated in a minimum of at least 2 independent batches, including Sham and CLP mice. Results from all available experiments were pooled within a single figure without excluding any outliers. For example, in Fig. 1a–e, all the results of mice follow-up and spleen cell immune monitoring available at the time of initial manuscript submission were presented. This represented a pool of data coming from nine different experiments performed between 2017 and 2020 by three different experimenters of both surgery and bench work. All the experiments were approved by the Université Claude Bernard Lyon 1 animal ethical evaluation committee (#2016072616404065, and #45069-2023062011386592, CECCAPP, Lyon, France) in accordance with the European Convention for the Protection of Vertebrate Animals used for Experimental and other Scientific Purposes.

Experimental design: patients’ cohorts

Adult patients from intensive care units (ICU) of the academic hospital (Hospices Civils de Lyon, Lyon, France) enrolled in three observational clinical cohorts were studied either prospectively (RICO = REA-IMMUNO-COVID and IMMUNOSEPSIS 4 cohorts) or retrospectively (REALISM = REAnimation Low Immune Status Markers cohort)²². Samples in humans were collected under approved protocols registered under ClinicalTrials: NCT04392401 (RICO = REA-IMMUNO-COVID); NCT02638779 (REALISM cohort (REAnimation Low Immune Status Markers); NCT04067674 (IMMUNOSEPSIS 4).

RICO cohort included critically ill patients who presented with pulmonary infection with SARS-CoV-2 confirmed by RT-PCR testing. Exclusion criteria disqualified pregnant women and institutionalized patients. It was approved by the ethics committee (“Comité de Protection des Personnes Ile de France 1”-N°IRB/IORG #: IORG0009918) under agreement number 2020-A01079-30. IMMUNOSEPSIS 4 included septic shock patients admitted into ICU identified according to the Third International Consensus Definitions for Sepsis and Septic shock (Sepsis-3)¹: vasopressors administration started within the first 48 hours around inclusion, plasma lactate level above 2 mmol/L and diagnosed or suspected infection. The exclusion criteria were any condition modifying the immune status: immunosuppressive treatment (including > 10 mg equivalent prednisone per day or cumulative dose >700 mg), hematological disease treated within the 5 years, solid tumor treated with chemotherapy or in remission, a number of circulating PNN < 500/mm3, innate immune deficiency, extra-corporeal circulation with one month before inclusion (cardiac surgery or ECMO). The onset of septic shock was defined by the beginning of vasopressor therapy. It was approved by the ethics committee (“Comité de Protection des Personnes Ouest II—Angers Sud-Est II”; RCB identification number: 2019-A000210-57, identification number SI/CPP: 19.01.23.71857), registered at the French Ministry of Research and Teaching (#DC-2008-509) and recorded at the Commission Nationale de l’Informatique et des Libertés. Since RICO and IMMUNOSEPSIS 4 studies were observational with low risk for the patients and no specific blood sampling procedure besides routine blood sampling required, there was no need for written informed consent, although oral information and non-opposition to inclusion in the study were mandatory and recorded in patients’ clinical files. Finally, REALISM included septic patients identified according to the Third International Consensus Definitions for Sepsis and Septic shock (Sepsis-3)¹: vasopressors administration started within the first 48 hours after ICU admission, plasma lactate level above 2 mmol/L in case of septic shock, suspected infection for which microbiological sampling had been performed, along with the administration of antimicrobials. Exclusion criteria were aplasia or pre-existent immunosuppression, pregnancy, and individuals with no social security insurance, restricted liberty or under legal protection. It was approved by the ethics committee (‘Comité de Protection des Personnes Sud-Est II’, Bron, France) and the French National Security Agency for drugs and health-related products (Approval code: 69HCL15_0379, 30th November 2015).

In all three cohorts, EDTA and heparin anticoagulated blood samples were collected in patients during routine procedures. First samples (i.e., D1) were obtained within the first 48 h after inclusions in the studies (i.e. on average 10 days after the date of first symptoms in COVID-19 patients and within 48 h after diagnosis of infection for patients with bacterial sepsis). Clinical and biological parameters were collected, including demographic characteristics, date and cause of admission to ICU, vital status at day 28 after inclusion, type of infection, comorbidities (Charlson index) and severity scores (Simplified Acute Physiology Score II, Sepsis-related Organ Failure Assessment; McCabe classification). Concomitantly, EDTA and heparin anticoagulated blood samples were collected from healthy volunteers (HV) from the blood bank EFS of Lyon (Etablissement Français du Sang). According to EFS standardized procedures for blood donation, informed consent was obtained from healthy donors and personal data were anonymized at the time of blood donation and before the transfer of blood to our research lab. Every experiment was performed on fresh whole blood collected on the same day. Samples were kept on ice during storage before analysis processing.

Murine model of CLP

For each experiment, mice (total n = 295) were divided in two groups: animals with polymicrobial sepsis induced by CLP (total n = 172) as previously described⁵⁰ and control mice (Sham, total n = 123). Briefly, around 1 hour after a subcutaneous (SC) injection of 0.1 mg/kg buprenorphine (AXIENCE), mice were anaesthetized with isoflurane inhalation (induction 3%, maintenance 1.7–2%; FiO2 = 0.2). They were placed on a heating mat during the whole surgery procedure. The abdomen was shaved and disinfected. Local anesthesia was performed by SC injection of 100 µL of lidocaine chlorhydrate 1% (AGUETTANT). Paramedian laparotomy allowed cecum exteriorization, which was then ligatured at its external third using 5.0 silk thread (Teleflex, DEKNATEL) and punctured twice (21-gauge needle) without crossing to the other side. A droplet of feces was extruded from the puncture holes to ensure patency. Cecum was carefully replaced in the peritoneum to avoid any intestinal twist. Finally, incision was sutured in layers, and animals were resuscitated with an SC injection of 1 ml of Ringer Lactate (VIAFLO). Mice were placed under a heat lamp during recovery. Within the same conditions, Sham control mice underwent laparotomy with only exteriorization of cecum from abdomen cavity, without ligation nor puncture. Mice were kept in the same cage after surgery, a heating mat was placed under half of the cage, and food was softened with water. Six hours following surgery and then every 12 h for the next two days, all mice received injections of antibiotics (20 mg/kg/day intraperitoneal (IP) Metronidazole BRAUN; 85 mg/kg/day SC Cefoxitine PANPHARMA). Pain was controlled by SC injection of 0.1 mg/kg buprenorphine given 10 hours post-surgery, and twice daily for the next two days. 48 h after surgery, mice were anaesthetized with isoflurane before sacrifice by exsanguination from retro-orbital sinus (local anesthesia by tetracaine 1%) and cervical dislocation.

A severity score was evaluated every hour (except during the night) after surgery based on several clinical criteria: fur aspect, motor activity, posture, breathing, weight loss (Supplementary Table 2). Animals were euthanized if they reached a score ≥9 before the 2 days end-point.

In vivo depletion of PCs

0.5 mg/kg of Bortezomib (SIGMA-ALDRICH) or a control solution of 1/300 diluted DMSO (SIGMA-ALDRICH), i.e. Bortezomib vehicle, was injected intravenously in both Sham and CLP mice 24 h after surgery. This dose of Bortezomib was determined after dose-response experiment evaluating effectiveness of Bortezomib injection (0.5, 0.75 and 1 mg/kg) at depleting PCs without adversely affecting other cellular populations in spleens of CLP and Sham mice.

Cell purification

In mice, spleens were mechanically homogenized in phosphate buffer saline (PBS, EUROBIO) under sterile conditions to obtain a suspension of splenocytes. Red blood cell lysis was performed using a hypotonic solution of NaCl, and cells were then washed twice in PBS. PCs were purified from splenocytes by positive selection with magnetic beads (EasySep™ Mouse CD138 Positive Selection Kit; STEMCELL), while B and T cells were purified by negative selection (EasySep™ Mouse B and T Cell Isolation Kits; STEMCELL). Purified cells were further used for proliferation assay or frozen at −80 °C in 300 µL RNA Lysis Buffer (ZYMO RESEARCH) for further RNA expression analysis.

In humans, HVs’ T cells were isolated from whole blood by immunodensity isolation (RosetteSep™ Human T Cell Enrichment Cocktail; STEMCELL). When necessary, regulatory T cells were depleted from T cells by magnetic cell separation (EasySep™ Human Pan-CD25 Positive Selection and Depletion Kit; STEMCELL), with a yield of <1% residual CD25+ CD4+ T cells. In septic patients, B cells were purified from whole blood by immunodensity isolation (RosetteSep™ Human B Cell Enrichment Cocktail; STEMCELL) and PCs by magnetic cell separation (MACSprep™ Multiple Myeloma CD138 MicroBeads; MILTENYI).

Analysis of cellular morphology by microscopy

Between 20,000 and 40,000 B cells and PCs were sorted by FACS (Aria II; BECTON DICKINSON) as B220⁺ CD138⁻ and B220^low CD138⁺ cells, respectively. Cytocentrifuge was then used to concentrate cells onto a microscope slide so that they could be stained by May-Grünwald Giemsa and examined by a histologist blind of cell sorting populations. Images were acquired with a Nikon Eclipse 80i microscope at 60X magnification using a Basler camera and ICS Capture software (Tribvn, France).

Cell culture and proliferation assays

In mice, total splenocytes or splenic T cells were stained with Tag-it Violet™ Proliferation and Cell Tracking Dye (BIOLEGEND). T cell cultures (10⁶ cells/mL) were set up in a culture medium consisting of RPMI (EUROBIO) supplemented with 10% AB human serum (Etablissement Français du Sang, Lyon, France), 200 mM _L-glutamine (LONZA), 10 mg/mL penicillin-streptomycin (BIOLOGICAL INDRUSTRIES), 5 mg/mL fungizone (BRISTOL-MYERS SQUIBB) and 1% β-mercaptoethanol (SIGMA), in 96-well flat-bottom plates. Cells were stimulated by anti-CD3/CD28 antibodies-coated beads (T Cell Activation/Expansion Kit; MILTENYI) at a bead-to-cell ratio 2:1.5, or by phytohemagglutinin (4 ng/µL) (THERMOFISHER). In the co-culture assays, splenic B cells or PCs were stained with CellTrace™ Far Red (THERMOFISHER) and added at 1:1 ratio to T cells in the stimulated condition. After 72 h of culture, supernatant was frozen at −80 °C and cells harvested, then stained with propidium iodide (SIGMA-ALDRICH) or Zombie Aqua™ Fixable Viability Kit (BIOLEGEND) and analyzed on flow cytometer to determine proportion of viable proliferating T cells in each culture condition. To measure the proliferation of CD4 and CD8 T lymphocytes, cells were also stained with CD45-APC-Fire750, CD3-PE, CD4-APC and CD8-PC7 prior to analysis on a flow cytometer. To explore regulatory mechanisms of PCs, additional culture conditions were used. HTS Transwell® inserts (0.4 µm pore size polycarbonate membrane; CORNING) were set on 96-well plates, with T cells on the bottom of the well and PCs on the top of the membrane, at a cell-cell ratio 1:1 and a concentration of 10⁶ cells/mL. Anti-PD-L1 (Ultra-LEAFeaf Purified anti-mouse CD274, clone 10 F.9G2B7-H1; BIOLEGEND) blocking antibody or corresponding isotype control (Ultra-LEAF Purified Rat IgG2b kappa; BIOLEGEND) was also added in some cultures at 10 µg/mL.

In humans, T cell cultures (10⁶ cells/mL) were set up in 96-well flat-bottom plates in the same culture medium as mouse splenic T cells but without β-mercaptoethanol. Between 50,000 and 100,000 purified T cells from HVs (depleted or not of Tregs) were cultured in the absence or in the presence of stimulating anti-CD2/CD3/CD28 antibodies-coated beads (T Cell Activation/Expansion Kit; MILTENYI) at a bead-to-cell ratio of 1.25:1, either alone or with B cells or PCs from septic patients at 1:1 ratio in the stimulated condition. After 72 h of culture, cells were harvested, stained with anti-CD3-APC-AF750, anti-CD4-APC and anti-CD8-KrO antibodies and T lymphocyte proliferative response was evaluated by flow cytometry in each condition measuring 5-ethynyl-2’deoxyuridine (EdU) incorporation in T cells as described elsewhere (Click-iT© EdU Flow cytometry assay kit, INVITROGEN)⁵¹.

Mice immunization with ovalbumin and evaluation of antigen-specific T lymphocyte response ex vivo

Mice were immunized with subcutaneous injection in each hip of 0.1 mL of chicken ovalbumin peptide OVA_323-339 in complete Freund’s adjuvant (OVA_323-339/CFA Emulsion at 1 mg/mL, HOOKE LABORATORIES). Mice were submitted to CLP or Sham surgery fourteen days later. Mice were sacrificed 48 h after surgery and spleens were harvested. Total splenic cells (10⁶ cells/mL) were cultured in 96-well flat-bottom plates in complete culture medium in the absence or in the presence of OVA_323-339 peptide (HOOKE LABORATORIES) at 3 concentrations (20 µg/mL, 5 µg/mL and 2 µg/mL). The concentration of IFNγ in supernatants was measured by ELISA after 7 days of ex vivo re-stimulation. Intracellular IFNγ level in T cells was evaluated after 4 hours of ex vivo re-stimulation in the presence of brefeldin A (5.0 µg/ml, BIOLEGEND).

Flow cytometry

In mice, total splenocytes absolute counts were determined using fluorescent microbeads (Flow-Count Fluorospheres, BECKMAN COULTER) and LDS 751 cell-permeant nucleic acid stain (INVITROGEN). For each immunophenotyping analysis, 100 µL of spleen cell suspension (i.e., 500,000 cells) were incubated with 10 µL of TruStain fcX™ antibody (clone 93; BIOLEGEND) for 10 min at 4 °C prior to immunostaining. Without washing step, cells were then incubated with 3 µL of each monoclonal antibodies (mAb) for 30 min in the dark at room temperature, then washed in PBS. List of reagents is provided in Supplementary Tables 3–8. Supernatant was discarded, and cell pellet resuspended in 300 µL of PBS and 1% formaldehyde (SIGMA-ALDRICH) before data acquisition on flow cytometer. For evaluation of necrosis and apoptosis, 500,000 splenic cells were stained with 100 µl of Zombie Violet™ Fixable Viability Kit (BIOLEGEND) diluted at 1:1000 in PBS during 15 min in the dark at room temperature. Cells were then washed and incubated with 20 µl of 1:20 diluted TruStain fcX™ antibody (clone 93; BIOLEGEND) for 10 minutes at 4 °C. After a washing step, cells were incubated with 20 µl of a mix containing 1:100 diluted mAb and 1:200 diluted Apotracker™ Green (BIOLEGEND) during 20 min in the dark at 4 °C. After a washing step, cells were resuspended in 300 µl of PBS before data acquisition on a flow cytometer. Necrotic cells were defined as positive for both Zombie and Apotracker and apoptotic cells as positive for Apotracker only. Representative histograms of the flow cytometry gating strategy are illustrated in Supplementary Fig. 5. Intracellular stainings for BLIMP1, IgM and IFNγ were performed using eBioscience™ staining buffer set (THERMOFISHER SCIENTIFIC) as followed: after last washing step, cell pellet was resuspended in 50 µL of fixation buffer and incubated for 30 min at 4 °C in the dark. Cells were then centrifuge, supernatant discarded before a washing step with 50 µL of permeabilization buffer. Cells were then incubated with monoclonal antibodies for 35 min at room temperature in the dark, then washed in perm buffer and finally resuspended in PBS before data acquisition on a flow cytometer.

In humans, for cell surface stainings, 100 µL of whole blood was washed twice with PBS. The cell pellet was resuspended in 100 µL of AB human serum and incubated with mAb for 15 min in the dark at room temperature. Red blood cells were then lysed with 500 µL Optilyse C (BECKMAN COULTER) for 10 min in the dark at room temperature. Cells were finally washed with PBS, and pellet resuspended with 300 µL PBS. Intracellular stainings for BLIMP1 and IgM were performed using PerFix-nc Kit (BECKMAN COULTER) as followed: after 15 min incubation of 50 µL of whole blood with membrane antibodies, cells were washed and supernatant discarded. Cells were next incubated with 25 µL of R1 reagent in the dark at room temperature and then washed. Intracellular antibodies were diluted in 300 µL of R2 reagent and added to cells. After 30 min incubation at room temperature in the dark, cells were washed with 3 mL of R3 reagent (1×) and finally resuspended in R3 reagent before data acquisition on flow cytometer. In a retrospective study of REALISM data, PCs were identified as CD19^low SSC^high circulating cells, while in RICO and IMMUNOSEPSIS prospective studies, they were defined as CD38⁺ CD138⁺ cells. We previously showed that these two phenotypes were well correlated and relevant for PCs analysis in human blood⁹.

All cytometry experiments were performed on Navios cytometer (BECKMAN COULTER) or LSR II (BD BIOSCIENCES), the same day as sample collection. Calibration beads (Flow-Set and Flow-Check Fluorospheres; BECKMAN COULTER) were run daily to check for routine alignment and day-to-day and long-term performance validation. Fluorescence-minus-one or negative cell controls were used to establish positive windows when needed. FACS analysis used Kaluza software 2.1 (BECKMAN COULTER) and FlowJo 10.8.1 software (BD BIOSCIENCES). Depending on the marker or cell population analyzed, percentage and/or median fluorescence intensity (MFI) were monitored. Representative histograms of flow cytometry stainings expressed as MFI are illustrated in Supplementary Figs. 6–9.

RNA extraction and gene expression analysis

Minimum 50,000 purified splenic B and PCs from 6 Sham and 6 CLP mice (i.e., 24 samples in total) were thawed for total RNA extraction, using RNeasy Plus Mini Kit (QIAGEN) following manufacturer’s protocol. RNA concentration and integrity number (RIN) were measured with 2100 Bioanalyzer System (AGILENT). For each sample, 100 ng RNA presenting with RIN between 8.6 and 10 were processed for gene expression analysis. NANOSTRING nCounter assay, a hybridization-based multiplex assay, was performed for gene expression analysis at the hospital genomic platform (Plateforme BIOGENET Sud—Hospices Civils de Lyon, France). Total RNA were diluted with Nuclease-Free water (SIGMA-ALDRICH) at 20 ng/µl in the 12-strip provided by NANOSTRING, and were analyzed using the nCounter® PanCancer Immune Profiling Panel according to manufacturer’s instructions. Eight negative probes and six serial concentrations of positive control probes were included in the multiplexed reaction, respectively, allowing to determine the background signal and to perform a normalization to correct potential sources of variation associated with the technical platform, as described elsewhere⁵². Then, to prevent the effect of potential RNA input differences, we normalized the data according to the expression of two genes (Hdac3 and Cyld), identified with geNorm method⁵³ as stable between the four groups of samples studied. Finally, among 768 genes available in the nCounter® PanCancer Immune Profiling Panel, 110 turned out to have no expression in any of the 24 samples analyzed, and we selected genes with RNA expression in >75% of B cells or PCs samples for further analysis, which represented 540 genes for B cells and 554 genes for PCs.

IL-10 measurement

IL-10 was measured in undiluted culture supernatants using sandwich ELISA MAX™ Deluxe Set Mouse IL-10 (BIOLEGEND) according to manufacturer’s recommendations. Samples were analyzed as duplicates when possible. Absorbance was read at 450 nm with Elx 808JU microplaque reader (BIOTECK). Gen5 software 3.04 (BIOTECK) was used to generate a standard curve through regression analysis on the log of standard values and to calculate samples’ concentrations.

IFNγ measurement

In mice, IFNγ was measured in undiluted culture supernatants using Mouse IFN-gamma DuoSet ELISA (BIO-TECHNE) according to manufacturer’s recommendations. Samples were analyzed as duplicates. In humans, IFNγ was measured in culture supernatants using Human IFN-gamma DuoSet ELISA (BIO-TECHNE) according to manufacturer’s recommendations. Samples were analyzed undiluted for non-stimulated culture conditions or after 1/50 dilution for stimulated culture conditions, as duplicates when possible. Absorbance was read at 450 nm with ELx808 microplaque reader (BIOTECK). Gen5 software 3.04 (BIOTECK) was used to generate a standard curve through a four-parameter logistic curve-fit and to calculate samples’ concentrations.

Statistical analyses

Experimental data were analyzed by R (Version 3.6.2; Boston, MA, USA) and GraphPad Prism (Version 9.0.2; La Jolla, CA, USA, Supplementary Table 9). Statistical details of the experiments are provided in the respective figure legends. Bar plots present individual values as well as mean and standard deviation. Comparison of continuous variables was performed with non-parametric Mann–Whitney or Wilcoxon tests for matched samples. Kaplan–Meier survival analysis and log-rank test were used to compare mortality in Sham and CLP mice. RNA expression in purified B and PCs from mice was explored drawing volcano plots and heatmaps (pheatmap package in R), and using Ingenuity Pathway Analysis (QIAGEN). P values < 0.05 were considered statistically significant and are shown on plots.

Reporting summary

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