All chemicals were from Sigma Aldrich (Saint Quentin Fallavier, France). LPS was from E. coli serotype 0127: B8 (Sigma Aldrich); BCG vaccine was purchased from Sanofi Pasteur and composed of 0.5 mg of the Brazilian strain (BCG Biomed-Lublin Laboratory).

For graft-versus-host disease experiments, eight-week-old female Balb/c (H2^d) and C57BL6 (H2^b), weighing 15–17 g, were purchased from Janvier Laboratory (Le Genest Saint Isle, France). Male B10.D2 (H2^d) mice were kindly offered by Colette Kanellopoulos-Langevin, CDTA-CNRS-Orléans, France). For the A20 lymphoma induction, eight-week-old Balb/c AnN strain mice (H2d) weighing 15–17 g were purchased from Janvier Laboratory (Le Genest Saint Isle, France). All mice were housed in ventilated cages with free access to food and water and maintained under standard 12-h photoperiod. Animals were treated humanely in compliance with the institutional guidelines. The protocols and all experimental procedures of this study were approved by the Ethics Committee from Paris Descartes University (N° CEEA34.CN.017.12).

The purification of CD4⁺ T cells derived from B10.D2 mice (set as responder cells) was performed using the magnetic MACS separator from Miltenyi (130-110-443, Bergisch Gladbach, Germany) and compbeads CD4⁺ (01-1111-42, Invitrogen) for positive cell selection. Purity of the CD4⁺ T cell suspension obtained was verified by flow cytometry and was typically >95%. Splenic cells from Balb/c mice extracted as described above were irradiated at an intensity of 30 Gy to render them non-proliferative and constitute the stimulator cells. For CFSE labeling, irradiated splenocytes and CD4⁺ T cells were stained with 5 µM of CFSE (Ref C34554, Invitrogen) and incubated for 15 min at 37°C in a CO₂ incubator protected from light. Labeling was stopped by adding complete MLR medium [RPMI with 10% FBS, 1% HEPES, 1% antibiotic and antimycotic (Gibco)].

Bone marrow (BM) cells were collected from the femurs and tibias of female Balb/c mice and differentiated into BMDM using BMDM medium (RPMI with 20% L929 cell supernatant, 10% FBS, 1% antibiotic/antimycotic, and 1% HEPES), as previously described (Manzanero, Methods Mol. Biol 2012; Weischenfeldt, CSH Protoc. 2008). At day 7, cells were washed in warm Ca²⁺/Mg²⁺-free PBS and collected by brief trypsin-EDTA (Ref 25200-056, Gibco) treatment. Cells were spun at 1,200 rpm for 7 min to form a pellet and then resuspended in complete RPMI medium. In vitro immune training of BMDMs was performed as previously described (23, 25). BCG treatment was achieved by 48 h incubation of macrophages with dry BCG vaccine (BCG Biomed-Lublin Laboratory) diluted in sterile PBS (Ref CS1PBS01-01, Eurobio) at a ratio of one bacillus for one cell followed by a resting phase of 3 days. LPS^low macrophage training was obtained by adding one daily dose of 10 ng/ml of LPS from E. coli serotype 0127:B8 (Sigma Aldrich) from day 1 to day 3 with daily change of culture medium to avoid cumulative dose toxicity. Control macrophages were obtained by daily treatment with 100 µl of sterile PBS. All cells were cultured in complete DMEM (Ref D0819-500, Sigma Aldrich) with 10% FCS and incubated at 37°C in an atmosphere containing 5% CO₂. At day 7, cells were collected by brief trypsin-EDTA treatment, counted and used for either the MLR or for intraperitoneal injection of mice.

Both the responder and the stimulator cells were adjusted at a density of 10⁵ per 200 µl of complete MLR medium and seeded in U-bottom 96-well microtiter plates (Ref 353077, Falcon). PBS, BCG, or LPS^low-trained macrophages were obtained from Balb/c mice as described above and added at a concentration of 10⁵ to the respective wells as indicated. The assay included appropriate negative controls (CD4⁺ purified T cells and irradiated Balb/c splenocytes seeded alone) and positive control cultures (CD4⁺ T cells stimulated with 5 µg/ml of anti-CD3 (Ref 14-0032-86, eBioscience) and 2 µg/ml anti-CD28 (Ref 14-0281-86, eBioscience). Each condition of co-culture was performed in 10-plicate, and plates were incubated for 4 days at 37°C in 5% CO₂.

Differentiated bone marrow derived macrophages obtained from Balb/c mice were plated at 2 × 10⁵ cells per well in a six-well plate in 2 ml of antibiotic free normal growth medium supplemented with 10% of FBS (Gibco) for 24 h. The siRNA IL-10 (sc-39635) and control (scrambled sequence) siRNA (sc-37007) obtained from Santa Cruz Biotechnology were used to obtain IL-10 KO macrophages by using siRNA transfection medium (sc-36868) and siRNA transfection reagent (sc-29528) (Santa Cruz Biotechnology) according to the instructions of the manufacturer. Macrophages IL-10 invalidation was evaluated by Western blotting and ELISA measurement of IL-10 in stimulated LPS^low-trained IL-10 KO macrophages.

The specific STAT3 Inhibitor V, Stattic sc-202818 (Santa Cruz Biotechnology, Catalog # 19983-44-9), was used as described before (26). A working solution of 500 µM was obtained after DMSO reconstitution and a 1/100 dilution in PBS. The CD4⁺ T cells were treated with 10 µM of the STAT3 Inhibitor V in 2 ml of complete RPMI medium (Ref R8758-500, Sigma Aldrich) supplemented with 10% heat inactivated Fetal Bovine Serum (FBS, Ref 10270-106, Gibco), 2 mM L-glutamine, 1% streptomycin–penicillin (Ref 15140-122, Gibco), 1 mM sodium pyruvate (Ref CSTVAT00-0U, Eurobio) and 10 mM HEPES (Ref 15630-056, Gibco). A control vehicle condition was composed of allogeneic CD4⁺ cells cultured in complete DMEM medium with PBS/DMSO (final concentration of 1% DMSO). Then, allogeneic CD4⁺ T cells were stimulated with the mouse recombinant IL-10 from BioLegend (Ref #575802, Ozyme France, 78180 Montigny-le-Bretonneux) at a concentration of 100 ng/ml and incubated for 24 h at 37°C 5% CO₂. All the conditions were performed in six-plicate and the experiments were carried out twice. Cells were then collected to perform further analysis (Western blot and RT-qPCR).

The severity of systemic GvHD in mice was assessed according to a mouse clinical GvHD scoring system inspired by Anderson et al. (27). We set the scoring system for chronic GvHD to five clinical criteria (maximum index = 5) representing the sum of the recorded values. Weight loss of <10% was scored 0 and of >20% was scored as 1. Regarding the GI symptoms, the scores were 0 for mice with no diarrhea and 1 for mice suffering from diarrhea. For posture, the scoring system denoted 0 for normal, 0.5 for hunching at rest, and 1 for severe hunching. For activity, the scoring represented 0 for normal, 0.5 for mild to moderate decrease, and 1 for severe decrease activity. For signs of vasculitis and cutaneous involvement, the scoring system denoted 0 for normal, 0.5 for mild lesions on ears or tails, and 1 for eyelid and severe alopecia. Total clinical GvHD score of each mouse was measured twice a week by a scientist blinded to the different groups.

Mice were sacrificed by cervical dislocation. Spleens were aseptically removed, and cellular splenic suspensions were prepared after hypotonic lysis of erythrocytes in potassium acetate solution and three washes in complete RPMI medium supplemented with 10% heat inactivated Fetal Bovine Serum, 2 mM L-glutamine, 1% streptomycin–penicillin, 1 mM sodium pyruvate, and 10 mM HEPES. For each mouse, splenocytes were enumerated using a Malassez counting chamber. Skin-derived immune cell isolation was achieved as described before (28). Briefly, 8 mm-calibrated dermal punches of tissue were collected from the shaved back of each mouse and minced into small pieces in a 60 mm Petri dish. Tissue was then lysed with 1 ml of Liver Digest Medium (Gibco) solution and 1 mg/ml of collagenase (Ref C2674-1G, Sigma-Aldrich) in 9 ml of complete RPMI medium for 3 h at 37°C in an atmosphere of 5% CO₂. Cell suspension was then filtered through a 70 µm cell strainer, washed in complete RPMI medium and counted with Malassez counting chamber for further analysis.

Sclerodermatous chronic GvHD (cGvHD) was induced in female Balb/c mice (H^2d) by grafting cells from 9- to 10-week-old male B10.D2 mice (H^2d), as previously described (29). Briefly, Balb/c mice were lethally irradiated with 750 cGy from a Gammacel [137Cs] source. Graft was prepared by adding 2 × 10⁶ spleen cells previously purified from red cells with a hypotonic solution of potassium acetate to 1 × 10⁶ of bone marrow cells resuspended in sterile PBS. Graft was injected in the retro-orbital vein 3 h after the irradiation of recipient mice. A group composed of (n = 5) lethally irradiated mice but non-transplanted was included to validate the irradiation source. A control syngeneic group (n = 5) of Balb/c recipient mice grafted with Balb/c spleen and bone marrow cells was added. GvHD mice were divided into four groups. One group (cGvHD; n = 5 mice) did not receive any macrophage injection. The three remaining groups (n = 10 mice) received a weekly infusion of 2 × 10⁶ intra-peritoneal Balb/c PBS-, LPS^low-, or BCG-trained BMDM, prepared as described above. A total of six injections was performed on D1, D7, D14, D21, D28, and D35 of the experiment. During the early post-irradiation and graft period (day <10), twice mice from the cGvHD-PBS, one mouse from the cGvHD-BCG and two mice from the cGvHD-LPS^low group either died spontaneously or were humanly sacrificed because of evident clinical suffering. The remaining animals were sacrificed by cervical dislocation at day 40 after the graft for further analysis.

Acute GvHD (aGvHD) was induced in female Balb/c mice (H-2d; Janvier Laboratory) by grafting cells from 9- to 10-week-old female fully mismatched major histocompatibility complex (MHC) C57BL/6 mice (H-2b; Janvier Laboratory), as previously described (30). Balb/c mice were lethally irradiated with 750 cGy from a Gammacel [137Cs] source and grafted 3 h later with 5 × 10⁶ RBC-free spleen cells and 1 × 10⁶ BM cells in sterile PBS. Control groups were as follows: a group (n = 5 mice) composed of lethally irradiated mice but non-transplanted and a control syngeneic group (n = 5 mice) of Balb/c recipient mice grafted with Balb/c spleen and BM cells. Acute GvHD mice were divided in four groups (n = 8 mice each). One group (aGvHD mice) did not receive any macrophage injection. The three remaining groups received a weekly intraperitoneal infusion of 2 × 10⁶ Balb/c PBS-, LPS^low- or BCG-trained bone marrow derived macrophages as previously described. A total of four injections was performed on D1, D7, D14, and D21 of the experiment. Animals were monitored daily for survival and three times a week for weight measurement.

Monitoring of graft recipients was based on daily observation and measurement of clinical evidence of GvHD (weight loss, manifestations of skin erythema, alopecia, hunching, and diarrhea) and survival during the observation period for both cGvHD and aGvHD experiments. Animals were humanely sacrificed if they had lost more than 40% of their original weight or if they showed evident signs of morbidity and suffering.

Levels of anti-DNA topoisomerase I IgG antibodies were assayed by the Scl-70 IgG ELISA kit. Diluted (1/4) mouse serum was distributed into the wells of a microtiter plate (Abnova, Taipei City, Taiwan) coated with purified calf thymus DNA topoisomerase I. The conjugated anti-mice Ig horseradish peroxidase (Dako, Glostrup, Denmark) secondary antibody (1/100) was then added, and the absorbance was read at 450 nm. Results were expressed as antibody index using the cutoff value derived from the Calibrator optical density as previously described (29).

Serum level of ALT was used as a marker of hepatocyte cytolysis and was quantified using a standard clinical automatic analyzer (Modular PP, Roche Diagnostics, Meylan, France).

The A20 cell line, a Balb/c AnN mouse (H-2d haplotype, female genotype) B lymphoma cell line derived from a spontaneous reticulum cell neoplasm (31), was purchased from the American Type Culture Collection (ATTC TIB-208; 1999). A20 cells were expanded in 20 ml culture Flask (Ref 353110, Falcon) in RPMI 1640 culture medium supplemented with 10% FCS, 5 µM β-mercapto-ethanol (Ref 31350-010, Gibco) and 1% streptomycin–penicillin.

For tumor challenge, A20 cells were washed twice in RPMI medium and resuspended at a concentration of 2 × 10⁶ cells in 200 µl of PBS and injected in the shaved back of mice. Mice were divided into four groups of n = 7 mice per group. One group (A20-CTRL) did not receive any macrophage injection. The three remaining groups (A20-PBS, A20-LPS^low, and A20-BCG, respectively) received a weekly intraperitoneal infusion of 2 × 10⁶ Balb/c AnN PBS-, LPS^low- or BCG-trained BMDM as described above. A total of five injections was performed on D1, D7, D14, D21, and D28 of the experiment. Animals were monitored daily for survival and three times a week for weight and tumor measurement. Tumors were measured three times a week with a microcaliper in three dimensions by a scientist blinded to the different treatments, and tumor volumes were calculated as length × width × depth × 0.5236, as previously described (32). Mice that extremely suffered from the tumors and expected to die within the same day of observation (signs including lack of eating/drinking/moving activity under mild stimulation, shivering, eye lids shut, etc.) were humanely euthanized on the days of observation. Mice were sacrificed at day 34 of the experiment for further analysis.

Total mRNA was extracted from crushed samples with TRIzol reagent (Ref 15596018, Ambion). QuantiTect SYBR^® Green RT-PCR Kit (Ref 04053228014782, Qiagen) on a LightCycler 480 II instrument (Roche Applied Science, France) was used to perform one-step RT-qPCR. Samples were normalized to mRNA expression of housekeeping genes (GAPDH), and results were expressed as fold increase using the formula 2^−˄˄Ct. Primers used for PCR are listed in Supplementary Table S1 .

Cell suspensions (spleen cells, skin-derived immune cells and cell suspensions of the MLR) were incubated for 20 min with 10 µg/ml anti-CD16/CD32 antibody (clone 93, eBiosciences) for Fc receptor saturation and then stained with the appropriate labeled antibody at 4°C for 30 min in the dark in PBS with 2% FBS. The FACS Fortessa II flow cytometer (BD Biosciences) was used to perform flow cytometry according to standard techniques. For spleen and skin characterization of immune cells, the monoclonal antibodies used were B220-Pacific Blue (RA3-6B2, catalog#103230, dilution 1/200), CD4-BV711 (RM4-5, catalog#563726, dilution 1/100), CD8-PE-Cy7 (QA17A07, catalog#155017, dilution 1/200), CD62L-PE-Cy5 (MEL-14, catalog# 104410, dilution 1/200), CD44-APC (IM7, catalog#103012, dilution 1/200), CD44-PeCy7 (IM7, catalog#103029, dilution 1/200), CD69-PercPCy5.5 (H1.2F3, catalog#104522, dilution 1/100), CD11b-BV510 (M1/70C, catalog#101245, dilution 1/200), F4/80-BV711 (BM8, catalog#123147, dilution 1/200), CCR4-BV421 (2G12, catalog#131217, dilution 1/100), CCR5-PercP-Cy5.5 (HM-CCR5, catalog#107015, dilution 1/200), (iCOS ligand) CD278-PeCy5 (15F9, catalog#107708, dilution 1/200), CXCR3-PE Dazzle 594 (CXCR3-173, catalog#126533, dilution 1/100), from BioLegend (Ozyme France, 78180 Montigny-le-Bretonneux), and CCR10-Alexa Fluor 700 (conjugated, catalog#FAB2815N, dilution 1/100) from R&D Systems (RD-BIOTECH, Besancon, France). Data were analyzed with FlowJo software (Tree Star, Ashland, OR).

Measurements of IL-2, IL-6, IL-10, IL-13, IL-17A, TNF-α,and IFN-γ were performed by specific mouse ELISA kits from eBiosciences (Thermo Fisher Scientific, Villebon-Sur-Yvette, France). Concentrations were calculated from a standard curve according to the protocol of the manufacturer.

Cells were lysed in ice-cold RIPA buffer (10 mmol/L Tris-HCl, pH 7.5; NaCl, 150 mmol/L; 1% Triton X-100; 0.1% sodium dodecyl sulfate) supplemented with 25 mmol/L sodium fluoride, anti-protease 1%, and 0.5 mmol/L sodium orthovanadate. Equal amounts of protein (30 μg) were loaded and separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Biorad Marnes-la-Coquette, France). After transfer and blocking with 5% of fat-free milk and 0.1% Tween in PBS, nitrocellulose membrane was incubated overnight at 4°C with the appropriate dilution of each antibody used according to the guidelines of the manufacturer. The antibodies used in the Western blotting analysis are listed in Supplementary Table S2 . Specific unconjugated proteins were detected using a 1:10,000 dilution of a horseradish peroxidase-conjugated goat anti-rabbit IgG (Life Technologies, Catalog# 31490) and visualized by an enhanced chemiluminescence system (Advansta, Diagomics, France).

Liver, lung, and skin pieces fixed in formol were embedded in paraffin. Tissue sections of 6 mm thickness were made and stained with Sirius red or H&E (hematoxylin and eosin). The slides were examined by standard bright field microscopy (Nikon Eclispe 80i) by a pathologist who was blinded to the experimental group assignment.

The results were analyzed with the GraphPad Prism5 program. The one-way test ANOVA (Kruskal–Wallis test) was used to verify that the groups were comparable to each other. The Student’s t-test (Mann–Whitney test) was used to determine the differences between two experimental groups. A difference <0.05 was considered as significant.

Trained macrophages can exhibit opposite phenotype and function according to the nature of the first antigenic stimulation (23). Here, we performed a mixed lymphocyte reaction mimicking in vitro the allo-immune response involved in the in vivo GvHD to evaluate the impact of trained macrophages on the proliferation and cytokine secretion of allogenic CD4⁺ T cells from B10.D2 mice. The proliferative response was detected by CFSE dilution, and activation of the responder cells was assessed by CD44 surface expression. The CD4⁺ T responder cells showed a strong proliferative response when cultured with irradiated splenic cells from Balb/c mice compared to CD4⁺ T responder cells cultured alone. As a positive and a negative control respectively, CD4⁺ T cells stimulated with anti-CD3/CD28 antibodies showed a strong proliferation index and a high activation status while irradiated Balb/c splenic cells barely proliferated ( Supplementary Figures S1A, B ). The addition of PBS-macrophages to the allogeneic reaction did not influence the proliferative response and activation of the allogeneic CD4⁺ T responder cells. By contrast, CD4⁺ T responder cells showed an enhanced proliferative index (p = 0.01), associated with increased CD44 expression (p = 0.0002) when cultured with BCG-macrophages, while their proliferation as well as activation was significantly reduced when cultured with LPS^low-macrophages (p = 0.0009 and p = 0.003, respectively, Figures 1A, B ). We then assessed the impact of trained macrophages on the function of allogeneic CD4⁺ T responder cells by measuring the levels of IL-2, IFN-γ, TNF-α, IL-6, and IL-10 in the supernatant of the different MLR conditions. Basal cytokine production of the irradiated Balb/c splenocytes and the B10.D2 purified CD4⁺ T cells (either unstimulated or stimulated with anti-CD3/CD28 antibodies) is shown in Supplementary Figure S1C . Of note, analysis of cytokine production in the supernatant of MLR with PBS-trained macrophages condition did not show any difference compared to the MLR condition alone. Addition of BCG-macrophages was associated with a significant increase in IL-2 (p < 0.0001), IL-6 (p = 0.009), TNF-α (p = 0.01), and IFN-γ (p = 0.01) in the supernatant, compared to the allogeneic MLR condition ( Figure 1C ). On the contrary, the addition of LPS^low-macrophages in the MLR significantly reduced the levels of IL-2 (p < 0.0001), IL-6 (p < 0.0001), TNF-α (p = 0.0008), and IFN-γ (p = 0.02) while significantly increasing IL-10 in the culture supernatants (p = 0.0004, Figure 1C ). Collectively, these data suggest that appropriately trained macrophages can act as either an enhancer or a suppressor of allogeneic CD4⁺ T cell proliferation, activation, and cytokine response in the setting of MLR.

We next investigated the in vivo effects of the adoptive transfer of trained macrophages on the course of cGvHD. We used the well-established animal model of sclerodermatous cGvHD triggered upon transplantation of B10.D2 (H-2d) BM and spleen cells across minor histocompatibility loci into lethally irradiated Balb/c (H-2d) recipients ( Figure 2A ). Clinical score assessment started 10 days after transplantation. At day 40, cGvHD mice had a score of 4.60 ± 0.36 while syngeneic remained null (p < 0.0001). Weight loss was marked in cGvHD mice compared to syngeneic animals (p = 0.005, Figure 2B ). Importantly, infusion of LPS^low-macrophages significantly attenuated the severity of the clinical score and the weight loss compared to the untreated mice (p = 0.005 and p = 0.006, respectively, Figures 2B, C ). Adoptive transfer of PBS-macrophages did not affect the clinical score nor the weight loss, whereas BCG-macrophage infusion tended to aggravate them though not significantly (p = 0.15 and p = 0.23, respectively). Representative photographs of mice at day 40 after BMT showed spiky coat and alopecia in the cGvHD animals with a favorable resolution in the cGvHD-LPS^low group ( Figure 2D ). Histological hematoxylin–eosin staining showed mixed lymphomonocytic infiltrations around the portal vein, which was significantly suppressed in the cGvHD-LPS^low group ( Figure 2E ). While mice receiving PBS or BCG-macrophages did not significantly affect cGVHD-induced increase in serum ALT, a marker of GvHD-associated liver injury, a virtually complete inhibition was observed in cGvHD-LPS^low group compared to the cGVHD group (p < 0.0001) ( Figure 2F ).

Chronic GvHD is characterized by dysregulated adaptive immune response with an inappropriate activation of allogeneic CD4⁺ T cells that leads to an alteration of B and CD8⁺ T-cell responses, responsible for the autoimmune features and tissue damage (33, 34). Anti-DNA topoisomerase I autoantibody production, indicative of this immune dysregulation, was increased in cGvHD mice compared to the syngeneic mice (p < 0.001, Figure 2G ). Importantly, LPS^low-macrophages induced a significant reduction in serum anti-topoisomerase 1 antibody index compared to the untreated cGvHD mice (p = 0.002), while the adoptive transfer of PBS- and BCG-macrophages had no affect ( Figure 2G ). We then assessed the immunological effects of the infused macrophages on the course of the cGvHD. Activation of T and B cells as well as a shift toward a predominant memory phenotype of CD4⁺ T cells is a hallmark of established cGvHD (35) ( Supplementary Figures S2A–C ). As shown in Figures 3A–C , adoptive transfer of LPS^low-macrophages significantly reduced CD69 expression on CD4⁺ T and B cells (p = 0.02 and p = 0.005) but not on CD8⁺ T cells and was associated with a reduced frequency of memory CD4⁺ CD62L^Low CD44^High T cells compared to untreated cGvHD mice (p = 0.006). By contrast, BCG-macrophage treatment increased the activation of CD4⁺ T cells (p = 0.004) and significantly augmented the proportion of memory over naïve CD4⁺ T cell subpopulation (p = 0.01). The inducible costimulatory molecule ICOS (CD278), which participates in the augmented T-cell response during cGvHD (36) was downregulated in cGvHD-LPS^low mice (p = 0.001) but enhanced in cGvHD-BCG group (p = 0.04, Figure 3C ).

Cytokine network in the cGvHD is adversely altered, with an early increase in Th1 and Th17 pro-inflammatory cytokine production, but a predominance of the Th2 subset at the chronic stage (37) ( Supplementary Figures S2D, E ). Spleen cells from cGvHD-LPS^low had a significant reduction in TNF-α (p = 0.003), IL-6 (p = 0.006), IFN-γ (p = 0.009), and IL-17 (p = 0.007) production but a concomitant increase in IL-10 secretion (p = 0.0005, Figure 3D ). In contrast, splenic cells from cGvHD mice treated with BCG-macrophages (cGvHD-BCG) displayed a heightened capacity to produce the pro-inflammatory cytokine IL-6 (p = 0.005) and IL-17 (p = 0.03) compared to the untreated cGvHD mice ( Figure 3D ). Furthermore, dermal cells from cGvHD-LPS^low animals showed a significantly reduced capacity to secrete IL-13 compared to the cGvHD mice (p = 0.0008), which was not affected by infusion with BCG-macrophages ( Figure 3E ).

Next, we set out to decipher whether the trained-macrophage infusion could affect the expression of chemokine receptors on allogeneic T cells that play an important role in lymphocyte migration to the sites of allo-recognition and priming (38). Chronic GvHD development was accompanied by an upregulation of CCR5 and CXCR3 on splenic T CD4⁺ and CD8⁺ cells and increased frequency and expression of CCR10 and CCR4 on dermal CD4⁺ T cells ( Figures 4A–D ). Of note, infusion of LPS^low-macrophages significantly downregulated the expression of CXCR3 (p = 0.001 and p = 0.03) and CCR5 (p = 0.005 and p = 0.03) on splenic CD4⁺ and CD8⁺ T cells; the expression of these chemokine receptors was strongly induced by adoptive transfer of BCG-macrophages ( Figures 4A, B ). Similarly, at the dermal level ( Figures 4C, D ), LPS^low-macrophage transfer significantly diminished CCR4 expression on CD4⁺ and CD8⁺ T cells (p = 0.02 and p = 0.03), whereas CCR10 was only decreased on CD4⁺ T cells (p = 0.001). cGvHD-BCG mice had an increased expression of only CCR10 on their dermal CD4⁺ T cells (p = 0.03). In parallel, we evaluated the dermal mRNA expression of the skin homing chemokines ccl17 and ccl27 and observed that they were increased in cGvHD group compared to the syngeneic mice (p < 0.0001). As shown in Figure 4E , BCG-macrophage infusion only upregulated the expression of ccl27 (p = 0.04), while LPS^low-macrophages abolished both the overexpression of ccl17 (p = 0.005) and ccl27 (p = 0.003). Finally, we analyzed the mRNA expression of two chemokines that promote the migration of the leukocyte subsets to the liver, ccl2, and cxcl2 (39, 40). Untreated cGVHD mice exhibited an increase of hepatic ccl2 and cxcl2 compared to syngeneic grafted mice (p < 0.0001). A downregulation of ccl2 (p = 0.004) and trend towards a decrease of cxcl2 expression (p = 0.18) were observed in the liver of cGvHD-LPS^low mice. Neither PBS- nor BCG-trained macrophage injections have a significant impact on the ccl2 and cxcl2 expression in the hepatic tissue ( Figure 4F ).

Altogether, these results indicate that infusion of LPS^low-macrophages exerts a suppressive effect on allogeneic responses during cGvHD with a drop in inflammatory cytokines and chemokine expression in target organs, but increased IL-10 production and a downregulation of chemokine and activation receptors on allogeneic T cells.

Chronic GvHD is accompanied by serious skin and visceral injuries including fibrotic lesions of the lung and severe intestinal crypt damage (41). As a next step to evaluate the in vivo effects of infusion of trained macrophages on the visceral involvement of cGvHD, we quantified the mRNA expression of the fibrosis-associated type I collagen Col1a1 and α-smooth muscle actin genes in skin and lungs. As expected, we observed an upregulation of mRNA expression of Col1a1 and α-sma in skin and lungs in the cGvHD group compared to the syngeneic mice (p < 0.0001). BCG-macrophage infusion increased the expression of Col1a1 in skin (p = 0.03) while cGvHD-LPS^low mice showed a significant reduction of Col1a1 (p = 0.02 and p = 0.0008) and α-sma mRNA expression in skin and lungs (p = 0.0007 and p = 0.0006, respectively, Figures 5A, B ). Moreover, cGvHD was accompanied by an upregulation of il13 mRNA expression in lungs compared to the syngeneic group (p < 0.0001). LPS^low-macrophages attenuated il13 expression (p = 0.0009) while the BCG-macrophages did not have any effect compared to the cGVHD group ( Figure 5B ). Sirius red-dyed staining of dermal and pulmonary sections showed clear evidence of collagen accumulation driven by the cGvHD. Infusion with LPS^low-macrophages resulted in alleviated fibrosis in skin and lungs ( Figure 5C ). We then analyzed the expression of ccn2, a matricellular protein that is upregulated in fibrotic hepatic disorders (42) and sox9/krt19 expression, markers associated with chronic liver damage (43). We observed an increased expression of ccn2, sox9, and krt19 in cGvHD mice compared to syngeneic (p = 0.03, p = 0.002, and p = 0.007 respectively). LPS^low-macrophage treatment diminished the expression of the assessed markers, while BCG-macrophage injection worsened the expression of ccn2 and sox9 compared to the cGVHD mice ( Figure 5D ). Lastly, the intestinal manifestation of cGvHD was assessed by mRNA quantification of il22 and il17 for the barrier integrity and mpo expression for quantification of the inflammatory infiltrate. As expected, cGvHD mice showed a significant upregulation of il22 (p = 0.0008), il17 (p = 0.007), and mpo (p = 0.0007) genes compared to the syngeneic group. Results showed that infusion with BCG-macrophages exacerbated the expression of il22 (p = 0.0008), whereas the adoptive transfer of LPS^low-macrophages significantly reduced mRNA levels of il22, il17, and mpo (p = 0.0007, p = 0.03, and p = 0.009, respectively) ( Figure 5E ).

Collectively, these findings show the beneficial effects of LPS^low-macrophages on the onset and clinical features of cGvHD and highlight the correlation between the specific immunological modifications and the impact on the visceral manifestation of the disease.

Acute GvHD was induced using the fully mismatched MHC mouse model of hematopoietic stem cell transplantation of lethally irradiated Balb/c recipient mice with splenocytes and bone marrow cells from C57BL/6 mice ( Figure 6A ). Survival rate was monitored daily. As shown in Figure 6B , lethally irradiated but non-transplanted mice had a median survival of 8 days, while all mice from the syngeneic group remained alive at the end of the experiment (day 30). Mice started to lose weight after the irradiation, but all transplanted animals regained weight after bone marrow transplantation (BMT). Allogeneic-transplanted mice started to lose weight around day 10 and reached the nadir at day 25. Clinical signs of acute GvHD started at day 11 of the experiment with decreased activity, hunched posture, and progressive alopecia ( Figure 6C ). Daily monitoring showed that aGvHD mice infused with LPS^low-macrophages had a trend towards an improved survival rate compared to the untreated group. PBS and BCG-macrophage treatment did not influence the survival rate of the aGvHD mice ( Figure 6C ). Thus, adoptive transfer of LPS^low-macrophages may exert a protective effect against acute GvHD, similar to what was observed with the cGvHD model.

We have shown in the first experiment that LPS^low-macrophage addition to the MLR system was associated with a marked increase of IL-10 level in the supernatant. Thus, we focused on that cytokine to get further into the mechanistic features of the protective effect of LPS^low-macrophages in the aGvHD and cGvHD models. We showed that adding recombinant mouse IL-10 (rm-IL-10) cytokine in the MLR system upregulated the phosphorylation of STAT3 and the mRNA expression of the IL-10 target gene arnt2 (44) but decreased the phosphorylation of T-cell specific protein kinase ZAP-70 (45) and the mRNA expression of the T-cell activation marker cd40l (46) ( Figures 7A, B ). Similarly, co-culture of LPS^low-macrophages with allogeneic CD4⁺ T cells increased arnt2 expression but was associated with a decrease of cd40l mRNA level ( Figure 7C ), a downregulation of the proliferation and cell-cycle associated proteins CDK4 and Cyclin A ( Figure 7D ), and a drop of IL-2, IL-6, and IFN-γ levels in the supernatant of the MLR milieu ( Figure 7E ). Notably, the inhibition of the STAT3 signaling pathway ( Figures 7A, B ), as well as the mRNA IL-10 silencing of the LPS^low-macrophages co-cultured with the allogeneic CD4⁺ T cells abrogated the effects observed with wild-type LPS^low-macrophages ( Figures 7C–E ). Altogether, these results highlight the involvement of IL10 in the modulation of the allogeneic response of CD4⁺ T cells and deliver further mechanistic insights into the immunomodulatory effect of LPS^low-macrophages in the aGvHD and CGvHD models.

It is well established that therapeutic strategies used for the prophylaxis and treatment of GvHD can diminish the beneficial effect of graft-versus-leukemia effects (4). Hence, we evaluated whether the protective/inhibitory effects of LPS^low-macrophages on the course of acute or chronic GvHD observed in our study could influence tumor progression in a B lymphoma (A20) murine model. Mice were injected subcutaneously with 10⁶ A20 cells and subjected to weekly infusion of trained macrophages ( Figure 8A ). Tumors were detected as early as day 10 of the injections. During the experiment (34 days), only three animals died spontaneously as a result of rapid tumor progression, and for ethical reasons, two were humanly euthanized at day 28 when the tumor reached a pre-defined volume size of 3,000 mm³. As shown in Figure 8B , tumors expanded rapidly in the different groups, though less importantly in the A20-BCG group. At the end of the experiment (day 34), no difference in tumor volume was detected between the A20-CTRL, the A20-PBS, and the A20-LPS^low groups. Interestingly, A20-BCG mice had significantly reduced tumor volume at day 34 compared to the other groups (p = 0.03). As a next step to decipher the effects of the infusion of trained macrophages observed in this experiment, we assessed the activation status and cytokine production by splenic and tumor infiltrating T cells. BCG-macrophage treatment resulted in an increase in TNF-α and IL-6 production by splenic cells compared to untreated A20 mice (p = 0.03 and p = 0.04, respectively), while PBS and LPS^low-macrophage infusions had no effects. Splenic IL-10 production capacity remained unchanged in the four groups ( Figure 8C ). Activation of CD4⁺ and CD8⁺ T cells was assessed by CD69 expression at both the systemic level on the spleen and in situ on tumors. BCG-macrophage treatment enhanced the activation of both CD4⁺ and CD8⁺ in situ (p = 0.0006 and p = 0.0008), while it only activated CD4⁺ T cells in the spleen (p = 0.006, Figure 8D ). The mean fluorescence intensity of CD69 on CD4⁺ and CD8⁺ T cells in spleen and tumors did not differ in untreated A20 mice or in mice infused with PBS or LPS^low-macrophages ( Figure 8E ). Finally, to get further into the molecular changes underlying the effects of the infused macrophages on the tumor growth, we analyzed the expression of several genes involved in both the anti-tumoral and the allogenic immune responses ( Figures 8F, G ). PBS-macrophage infusion did not induce any change in gene expression in the isolated A20 tumors or in the spleen tissues compared to the untreated A20-CTRL group. A20 mice infused with LPS^low-macrophages upregulated the expression of fizz1 (found in inflammatory zone-1) gene which is usually associated with the alternatively activated macrophages phenotype (M2 macrophages), in the spleen (p = 0.0005) and in the tumor (p = 0.03) and il10 gene only in the spleen (p = 0.02) compared to untreated A20 mice. Moreover, LPS^low-macrophages downregulated the expression of inos in spleen (p = 0.03) and il6 both in spleen (p = 0.005) and the tumor (p = 0.02). All the other genes assessed remained unchanged. BCG-macrophage infusion was associated with an upregulation of the splenic and tumoral expression of inos (p = 0.03), il6 (p = 0.0005), and infg (p = 0.03), markers that are associated with classically activated macrophage phenotype (M1 macrophages). By contrast, we observed a decrease in the expression of fizz1 (p = 0.01) and the T regulator cell transcription factor foxp3 (p = 0.03) in both the spleen and the tumor tissues. Furthermore, A20-BCG mice had a diminished il4 (p = 0.03) expression in spleen and a drop in il10 and tgfb expression in tumors (p = 0.03 and p = 0.02, respectively).

Cumulatively, these results show that LPS^low-macrophages do not adversely affect tumor growth and are not accompanied by major modifications of T cell activation, cytokine production, or gene expression. Interestingly, infusion with BCG-macrophages exhibit an anti-tumoral activity with an upregulation of the M1-associated genes and an increase of T-cell activation and pro-inflammatory cytokine secretion.