MSC-2 is an immortalized MDSC cell line obtained from BALB/c Gr-1⁺ splenocytes and was obtained from V. Bronte (Instituto Oncologico, Padova, Italy). EL4 thymoma cells (syngenic from C57BL/6 mice) were obtained from the American Type Culture Collection. Cells were grown in RPMI 1640 with ultraglutamine (Lonza) supplemented with 10% (vol/vol) fetal bovine serum (FBS; Lonza), in an atmosphere of 95% air and 5% CO₂ at 37°C. In some experiments, a heat shock was performed by incubating MSC-2 cells at 42°C for 1 hour. Cells were then left at 37°C for 2 hours before treatments with 5-FU (Accord). Viability assay was performed using Vybrant MTT Cell proliferation assay kit according to the manufacturer’s instructions (ThermoFisher Scientific).

MSC-2 cells were treated in OptiMEM without FBS, supernatants were precipitated and whole-cell lysates were prepared. The supernatants were harvested by centrifugation for 5 min at 400g and precipitated using methanol (500 µL) and chloroform (150 µL). After centrifugation at 12,000g for 10 min, the aqueous phase (at the top) was discarded and 800 µL of methanol was added. Samples were centrifuged at 12,000g for 10 min and the supernatants were removed. Pellets (containing proteins) were dried for 10 min at 37°C, mixed with 40 µL of loading buffer (125 mM Tris–HCl (pH 6.8), 10% β-mercaptoethanol, 4.6% SDS, 20% glycerol and 0.003% bromophenol blue) and incubated at 95°C for 5 min. Whole-cell lysates were prepared by lysing cells in boiling buffer (1% SDS, 1 mM sodium vanadate, 10 mM Tris (pH 7.4)) in the presence of complete protease inhibitor mixture. Samples’ viscosity was reduced by sonication and protein concentration was evaluated. Fifty micrograms of proteins was mixed with loading buffer.

Samples were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis, and electroblotted to a nitrocellulose membrane (Amersham, GE Healthcare) with a Tris–borate buffer or a Tris–glycin–ethanol buffer (for caspase-1 and IL-1β detection). After incubation for 1 hour at room temperature (RT) with 5% non-fat milk in phosphate-buffered saline (PBS)–0.1% Tween-20, membranes were incubated overnight with the primary antibody diluted in PBS–milk–Tween, washed, incubated with the secondary antibody for 30 min or 2 hours (for caspase-1 and IL-1β detection) at RT, and washed again before analysis with a chemiluminescence detection kit (Amersham, GE Healthcare). The following mouse mAbs were used: anti–β-actin (A1978) from Sigma-Aldrich and anti-murine caspase-1 (AG-20B-0044) from Adipogen. We used rat pAb anti-IL-1β (MAB-4011) from R&D Systems. We also used rabbit pAb anti-HSP70 (SPA-812) from Enzo Life Sciences. Secondary Abs HRP-conjugated polyclonal goat anti-mouse and swine anti-rabbit immunoglobulins (Jackson ImmunoResearch) were also used.

All animals were bred and maintained according to both the FELASA and the Animal Experimental Ethics Committee Guidelines (No. C 21 464 04 EA, University of Burgundy, France). Animals used were between 6 and 22 weeks of age. Female C57BL/6 mice (aged 6 to 8 weeks) were obtained from Charles River Laboratories. C57BL/6 Hsp70^−/− mice were obtained from the Mutant Mouse Resource and Research Center, bred and maintained in the Cryopréservation, Distribution, Typage et Archivage Animal (CDTA-Orléans, France).

To induce tumor formation, 0.5×10⁶ EL4 cells were injected subcutaneously into mice and 14 days after tumor-cell injection (tumor size, 150–200 mm²) animals were treated.

For hyperthermia experiments, mice were placed on a heating pad at 42°C for 1 hour, then allowed to rest 1 hour. Then, mice received a single intraperitoneal injection of 5-FU (Accord) at 50 mg/kg body weight. For some experiments, 1.5 mg/kg of IL-1Ra (Kineret from Sobi) were injected three times a week, beginning with the chemotherapeutic treatment at day 14. In other experiments, mice received a single intraperitoneal injection of recombinant IL-1β (R&D systems) 2 days after 5-FU treatment. Tumor size was measured three times a week with a caliper. The experiments were performed two or three times and the total number of animals was indicated in the figure legend.

Spleens and tumors were recovered 2 days after mice treatments. For spleens, single-cell suspensions were prepared using a 70 µm cell strainer and red cells were removed using hypotonic lysis buffer. For tumors, cells were dissociated using the tumor dissociation kit (130-096-730) with the gentleMACS dissociator (Miltenyi Biotec). One million freshly prepared cells resuspended in RPMI 10% FBS were assessed for caspase-1 activity with the FAM-YVAD-FMK fluorescent probe (FLICA1-ICT098; BioRad) for 1 hour at 37°C as previously described.12 Cells were then stained with a VioBlue anti-mouse Ly6C (130-102-207; Myltenyi Biotec), a BV650 anti-mouse Ly6G (127641) and a BV785 anti-mouse CD11b (101243) (Biolegend) and fixable viability dye (FVD) eFluor780 (Invitrogen) for 15 min at RT. After washing, cells were analyzed with a CytoFlex (Beckman Coulter) and a CytExpert software.

To assess 5-FU effect in vitro, 500,000 splenic cells prepared as above were treated with 1 or 10 µM of 5-FU for 24 hours in RPMI+FBS medium and stained with FAM-YVAD-FMK fluorescent probe for 1 hour at 37°C and then with a VioBlue anti-mouse Ly6C, a BV650 anti-mouse Ly6G and a BV785 anti-mouse CD11b and FVD eFluor780 for 15 min at RT. After washing, cells were analyzed with a CytoFlex (Beckman Coulter) and a CytExpert software.

Mice tumors were collected and lysed in Trizol reagent (Applied Biosystems) and total RNA was extracted. Three hundred nanograms of RNA was reverse transcribed into cDNA using M-MLV reverse transcriptase, random primers and RNAseOUT inhibitor (Applied Biosystems). cDNA was quantified by real-time PCR using Power SYBR Green Real-time PCR kit (Life Technologies) on a QuantStudio 5 detection system (Applied Biosystems). Relative mRNA levels were determined using the ΔΔCt method. Values were expressed relative to Actin b for Eng and Pecam1 or Cd3e levels for Il17a and Rorc. The sequences of the oligonucleotides used are described in online supplementary table 1.

Tumors from different treatment conditions (±HS, HSP70^−/− ) were harvested 48 hours after treatment with 5-FU or PBS as a control. These tumors were placed in tissue storage (Miltenyi, 130-100-008) and then left for a minimum of 48 hours in a formalin bath to fix the tissues. After this step, tumors were included in paraffin and were then cut into 4 µm thick slices. The paraffin was then removed using the PT-link device (Agilent S236784‐2). Slides were labeled for 1 hour with the anti-ERG monoclonal antibody (1/200; Abcam, EPR3864), then stained with ImmPRESS HRP Goat Anti-Rabbit IgG (Peroxidase) for 30 min (Vector, MP-7451) to amplify the signal. Finally, slides were stained with DAB (Agilent, SM803 and DM827) and counterstained with hematoxylin (Enzo, ENZ‐ACC106) and were mounted and digitally scanned using the Nanozoomer HT2.0 device (Hamamatsu). Staining quantification was carried out using QuPath software (v.01.3 ref). The appropriate threshold to discriminate positive cells from negative was determined on positive controls (endothelial cells of large vessels). To analyze the whole cohort, we generated two scripts. The first one was dedicated to upload 12 rectangles of 160,000 µm². These areas were then randomly placed on slides by the investigator. The second script detected ERG-positive cells in those areas. Then the percentage of positive cells was calculated.

In vitro results are shown as means±SD, in vivo results are shown as means±SEM and comparisons of datasets were performed using GraphPad Prism V.8. The following statistical tests were used: for tumor growth, two-way ANOVA followed by the Sidak post-test for multiple comparisons, flow cytometry experiments, one-way ANOVA test and for survival log-rank (Mantel-Cox) test. All p values were two tailed.

To evaluate the impact of HS on caspase-1 activation in MDSCs, we first performed in vitro experiments on the MSC-2 immortalized MDSC cell line. Cells were treated for 24 hours with indicated concentrations of 5-FU. This treatment decreased cell viability as assessed by MTT assay (figure 1A). When cells were preliminary submitted to a heat shock to allow heat shock protein expression, notably HSP70 (figure 1B), the effects of 5-FU on cell viability were still observed. Similarly, HS did not affect 5-FU-mediated EL4 tumor cells killing (figure 1A). The treatment of MSC-2 cells with 5-FU led to caspase-1 activation and release of active caspase-1 and mature IL-1β in the supernatant (figure 1B). However, a HS leading to HSP70 expression inhibited the activation of caspase-1 and the release of active caspase-1 and mature IL-1β (figure 1B).

To investigate the effects of HS in vivo, we used a 5-FU-treated tumor-bearing mice model.4 Mice were subcutaneously injected with EL4 tumor cells, and when tumors were established, mice were submitted to hyperthermia at 42°C during 1 hour. HS efficiency was checked by evaluating HSP70 overexpression in splenic cells (online supplementary figure 1A). After 48 hours of treatment, the effects of 5-FU on the proportion of splenic and tumor MDSCs and caspase-1 activation in these cells were analyzed by flow cytometry (figure 1C and online supplementary figure 1B). As expected, 5-FU led to a decrease in the proportion of splenic and tumor MDSCs (figure 1C). The total number of MDSCs was also reduced (online supplementary figure 1B). When mice were submitted to a HS before 5-FU treatment, we observed the same effects of 5-FU on MDSC elimination. As previously shown by our team, 5-FU induced caspase-1 activation in splenic and tumor MDSCs. However, HS inhibited 5-FU-mediated caspase-1 activation (figure 1D).

Altogether, our results show that HS is able to inhibit 5-FU-mediated caspase-1 activation without affecting 5-FU MDSC killing.

Because HS inhibits caspase-1 activation in MDSCs, we investigated its effects on tumor growth, in mice subcutaneously injected with EL4 tumor cells. As previously described, 5-FU was only able to temporarily decrease tumor growth, but after 23 days, tumors started to grow again (figure 2A). However, when a HS was applied simultaneously with 5-FU treatment, tumor regrowth was more efficiently delayed and mouse survival was increased (figure 2A,B). 5-FU-mediated caspase-1 activation in MDSCs leads to Th17 expansion and neoangiogenesis, both responsible for tumor regrowth.4 Therefore, we evaluated the expression of Th17-related genes (Il17a and Rorc) and angiogenesis-related genes (Eng and Pecam1) within the tumor (figure 2C). In 5-FU-treated mice, these genes were upregulated, whereas in mice submitted to a HS, 5-FU was unable to upregulate these genes. To confirm the results obtained on angiogenesis-related genes, we analyzed the expression of ERG (E-26 Transformation Specific (ETS)-Related Gene), a specific marker of endothelial cells.13 We observed that the capacity of 5-FU to induce the expression of ERG within the tumor, that is, to increase endothelial cell assembly to form microvessels, was decreased when mice were submitted to a HS (figure 2D and online supplementary figure 2). Finally, a single injection of 1 µg of IL-1β was able to accelerate mice death after HS and 5-FU treatments (figure 2E).

To conclude, HS is able to inhibit 5-FU-mediated tumor Th17 expansion and angiogenesis and to delay tumor growth.

To evaluate the impact of HSP70 deficiency on caspase-1 activation in MDSCs, we performed experiments in splenic cells from wild-type (WT) and Hsp70^−/− mice treated in vitro with 5-FU. We observed that 5-FU similarly eliminated MDSCs within WT and Hsp70^−/− splenic cells (figure 3A). However, caspase-1 activation was higher in Hsp70^−/− than in WT MDSCs after a 10 µM 5-FU treatment (figure 3B). To confirm these observations in vivo, WT and Hsp70^−/− mice were subcutaneously injected with EL4 tumor cells. When tumors were established, mice were treated with 5-FU. After 48 hours of treatment, the effects of 5-FU on splenic and tumor MDSCs proportion and caspase-1 activation in these cells were analyzed by flow cytometry (figure 2C-E). We observed that HSP70 deficiency (confirmed by western blot on splenic MDSCs; see online supplementary figure 3A) did not affect the effects of 5-FU on MDSC elimination (figure 2C and online supplementary figure 3B). However, HSP70 deficiency increased caspase-1 activation (figure 2D), it increased the percentage of MDSCs with active caspase-1 in tumors and it emphasized the mean fluorescence intensity (MFI) in MDSCs, suggesting a more important caspase-1 activation within each cell.

Our results show that HSP70 deficiency is able to increase 5-FU-mediated caspase-1 activation without affecting 5-FU MDSC killing.

Because HSP70 deficiency increases caspase-1 activation in MDSCs, we investigated its effects on tumor growth. WT and Hsp70^−/− mice were subcutaneously injected with EL4 tumor cells and when tumors were established, mice were treated with 5-FU. After 23 days, tumors started to regrow in WT mice (figure 4A). However, in Hsp70^−/− mice, tumor regrowth was faster and led to a decrease in mouse survival (figure 4A,B). Again, we evaluated the expression of Th17 cells and angiogenesis-related genes within the tumor (figure 4C). In Hsp70^−/− mice, these genes were more importantly upregulated by 5-FU as compared with WT mice. These observations were confirmed by an increased expression of ERG in tumors from Hsp70^−/− mice treated with 5-FU (figure 4D). Then, we studied the effect of HS on Hsp70^−/− tumor-bearing mice treated with 5-FU. As observed for WT mice, HS was also able to increase Hsp70^−/− mice survival, suggesting that the effects of HS in this model not solely relied on HSP70 (figure 4E). To determine whether the effect of HSP70 deficiency on tumor growth relied on the increase in caspase-1 activation in MDSCs and a more important IL-1β production, we inhibited IL-1β using anakinra, a recombinant soluble IL-1Ra. Anakinra was able to delay tumor regrowth and to increase Hsp70^−/− mice survival after 5-FU treatment (figure 4E,F).

To conclude, HSP70 deficiency is able to favor Th17 cell expansion and angiogenesis and to improve tumor growth after 5-FU treatment in an IL-1-dependent manner.