Six-week to 8-week-old female C57BL/6, IFNγ^−/−, TNFR^−/−, IL12p40^−/− and PepBoy CD45.1 mice were purchased from The Jackson Laboratory.

The MC38 cell line was provided by Dr Antoni Ribas at UCLA (Los Angeles, California, USA) in 2011. IDEXX tested (IDEXX Laboratories, ME), pathogen-free cells cultured in Roswell Park Memorial institute-1640 (RPMI-1640), 10% Fetal bovine serum (FBS) and 2 mmol/L l-glutamine were harvested for implantation while growing in the exponential phase.

0.5×10⁶ MC38 cells were implanted subcutaneousl in 100 µL phosphate-buffered saline (PBS). Tumors were treated at 70–100 mm³ with anti-mouse CD40 antibodies dosed subcutaneously twice, 3 days apart. Bodyweight and tumor volumes were measured twice a week. Tumors were measured using calipers, and volume calculated as V=0.5 L × W², where L and W are the longest and shortest diameter of the tumor, respectively.

Anti-mouse CD40 clone 8F2 was generated by immunizing American hamsters with mCD40-6xHis recombinant protein three times at triweekly intervals. Splenocytes were harvested and fused with P3-X63Ag8.653 cells to generate hybridomas. Hybridoma supernatants were screened and clone 8F2 selected based on functional properties. Antibody variable regions were sequenced and cloned into the pARC vector for mIgG1 recombinant expression. Variable sequences from Rat FGK45 were obtained from Genbank (accession numbers: heavy chain AEI27236.1, light chain AEI27235.1) and engrafted into mouse IgG1 constant region vectors. Rat IgG1 anti-mouse IL-6 (clone MP5-20F3) was purchased from BioXcell.

Female C57BL/6 (n=5/group) mice were given a single subcutaneous administration of test articles or control (PBS) and clinical and anatomic pathology was analyzed 7 days later. At necropsy, the liver was fixed in 10% neutral-buffered formalin for histology. Livers were processed, paraffin-embedded, sectioned, stained with H&E, and analyzed by a board-certified veterinary pathologist. Blood samples were collected for clinical chemistry analysis using Advia 1800 (Siemens). Alternatively, for figures, 1F and 3I GLDH was analyzed by ELISA (Cusabio) in accordance with the manufacturer’s instructions. Serum samples were analyzed for liver enzymes by the Pfizer Groton Clinical Pathology Laboratory and Pfizer Cancer Immunology Discovery in San Francisco. Cytokine analysis was performed by Pfizer Cancer Immunology Discovery in San Francisco.

BM chimeras were established as previously described.16 Recipients were lethally irradiated and reconstituted with 1×10⁶ BM cells and 1×10⁶ splenocytes. Engraftment was confirmed to be >90% by flow cytometry in peripheral blood 7 weeks postirradiation by monitoring donor CD45.1⁺ cells in CD45.2⁺ recipients, or vice versa.

Mouse livers were isolated for FACS staining as previously described.17 Hepatic immune cells were resuspended in PBS stain using the LIVE/DEAD Fixable Blue Dead Cell Stain Kit (Invitrogen). Mouse Fc Block (10 µg/mL, BD Biosciences) was added for 10 min, and cells were then stained directly with cell surface staining cocktail diluted in brilliant stain buffer (BD Biosciences) for 45 min at 4°C. Anti-mouse phenotyping antibodies purchased from BD Biosciences or Biolegend were CD45 (30F11), CD90.2 (53–2.1), B220 (RA3-6B2), CD4 (GK1.5), CD8a (53–6.7), CD11b (M1/70), CD11c (N418), Ly6C (HK1.4), Ly6G (1A8), CD14 (Sa14.2), F4/80 (BM8) and I-A/I-E (M5/114.15.2). Flow cytometry data were acquired using LSR Fortessa (BD Biosciences) and analyzed using FlowJo (TreeStar Inc.).

Results were expressed as mean±SEM. Statistical analyzes were performed using GraphPad Prism V.6.0 one-way or two-way analysis of variance to compare the differences among multiple groups. A p value of <0.05 was considered a significant difference.

We began by characterizing the efficacy and safety profiles of two agonistic anti-CD40 antibodies. Anti-CD40 IgG1 clone 8F2 was generated in-house, possesses strong agonistic activity, and a K_D of 98 nM, whereas the commercially available FGK45 clone binds with a K_D of ~2000 nM and possesses weaker antitumor efficacy. For these studies, we grafted the agonistic anti-CD40 clone FGK45, a rat IgG2 isotype,18 onto a mouse IgG1 backbone to facilitate the comparison with 8F2. The antitumor efficacy of these two antibodies was evaluated in the MC38 colorectal carcinoma model (figure 1). Implanted groups were randomized and treated at 80 mm³ every 3 days for a total of two injections. Antibodies were dosed at 10 mg/kg, which was the maximum tolerated dose for 8F2 based on bodyweight loss. Relative to isotype, 8F2 impaired tumor growth by 76.5% 24 days postimplantation (figure 1A) and was associated with severe systemic toxicity characterized by bodyweight loss of 15%–20%, 2–3 days following antibody treatment (figure 1B). FGK45 controlled tumor growth to a lesser extent (50.1% at day 25, figure 1C); however, bodyweight loss was markedly lower in the 7%–10% range (figure 1D). In parallel, we measured circulating inflammatory cytokines and liver enzymes as markers of CRS and hepatotoxicity, respectively. An increase in the circulating concentrations of TNF, IL-6, IFNγ, and IL-12p40 was observed on CD40 agonist treatment (figure 1E). Consistent with the efficacy data, concentrations of circulating inflammatory cytokines were higher on treatment with 8F2 compared with FGK45, except for IFNγ, which was initially higher on FGK45 treatment but lower than 8F2 at later timepoints. Analysis of liver enzymes showed little increase in circulating GLDH concentrations 7 days after treatment with FGK45. However, a substantial increase in GLDH was associated with 8F2 treatment (figure 1F) as well as, in a different study, increased concentrations of AST and ALT (online supplementary figure S1). Histopathological assessment of liver samples following 8F2 treatment demonstrated dense infiltrations of mononuclear cells around the central veins, portal tracts and within the tissue parenchyma compared with isotype control-treated samples (figure 1G). In summary, CD40 agonist treatment induces systemic inflammatory cytokine secretion and hepatotoxicity, the extent of which correlates with tumor growth inhibition.

We sought to understand the sources of toxicity arising from anti-CD40 agonist treatment. To determine the relative contribution of hematopoietic cells we established BM chimeras using wild-type (WT) and CD40^−/− mice. Clone 8F2 was used for subsequent studies owing to its more potent toxicity profile. Seven weeks following reconstitution, non-tumor bearing mice were challenged with anti-CD40 and toxicity parameters measured (figure 2). CD40 agonist significantly decreased bodyweight in both WT and CD40^−/− mice reconstituted with WT BM (figure 2A). In contrast, we observed vastly reduced bodyweight loss in both WT and CD40^−/− mice reconstituted with CD40^−/− BM. Mirroring these findings, CD40 agonist increased serum concentrations of TNF, IFNγ, IL-6 and IL-12p40 in mice that received WT BM irrespective of host genotype (figure 2B). Concentrations of these cytokines were unaffected by 8F2 in mice that were completely deficient of CD40. However, IL-6, TNF and IL-12p40 were marginally higher on 8F2 treatment in WT mice that received CD40^−/− BM (figure 2B). Analysis of liver enzymes further revealed a hematopoietic contribution to hepatotoxicity (figure 2C). 8F2 treatment increased ALT, AST and GLDH concentrations in WT BM recipients (figure 2C) and was also associated with cellular infiltrate in the liver (data not shown). In contrast, WT mice receiving CD40^−/− BM did not increase liver enzymes and, although cellular infiltrates were still present in the liver, they were less severe than in mice receiving WT BM. These findings show that BM-derived cells are the primary cellular source of CD40-mediated CRS and bodyweight loss and contribute significantly to liver toxicity.

As CD40 expressed on immune cells is responsible for the CRS and hepatotoxicity associated with CD40 agonist treatment, we sought to understand the molecular mechanisms underpinning this toxicity. Clinical studies using CP-870,893 determined that CRS is associated with a rapid increase in IL-6 and TNF concentrations.12 Thus, we assessed the influence of these cytokines on anti-CD40-induced toxicity in MC38 tumor-bearing mice. IL-6 neutralization reduced IL-6, and increased circulating concentrations of IFNγ on 8F2 treatment but did not affect TNF or IL-12p40 concentrations (online supplementary figure S2A). Additionally, IL-6 blockade did not influence GLDH concentrations (figure 3A); however, we did observe a slight reduction in bodyweight loss induced by 8F2 on IL-6 blockade (figure 3B). These findings suggest that IL-6 is not responsible for the systemic or liver toxicity associated with CD40 agonist treatment.

Next, we evaluated the role of TNF in toxicity using TNFR^−/− mice. 8F2 increased circulating TNF concentrations (online supplementary figure S2B), presumably due to a decreased uptake in the absence of TNFR expression. Global TNFR deficiency curtailed IL-6, IFNγ and IL-12p40 concentrations (online supplementary figure S2B) yet did not influence hepatotoxicity (figure 3C). Importantly, bodyweight loss was markedly reduced in TNFR^−/− mice on 8F2 treatment (figure 3D). On average, TNFR^−/− mice lost 6.8% of their bodyweight as compared with 17.1% for WT mice on day 3, demonstrating that TNF rather than IL-6 is primarily associated with CD40 agonist-induced bodyweight loss.

Because IFNγ and IL-12p40 were decreased in 8F2 treated TNFR^−/− mice, we asked whether either of these contribute to CRS-associated bodyweight loss. Strikingly, both IL-12p40^−/− and IFNγ^−/− mice showed complete protection from CD40 agonist-induced body weight loss (figure 3E,F), accompanied by decreases in all measured inflammatory cytokines (figure 3G,H). IFNγ deficiency did not impact GLDH concentrations (figure 3I). hHowever, IL-12p40^−/− mice demonstrate a significant reduction in circulating GLDH concentrations (figure 3J). Thus, a TNF-IL-12p40-IFNγ cytokine axis regulates CD40-dependent body weight loss. Furthermore, our findings establish a role for IL-12p40 in liver toxicity elicited in response to CD40 agonist treatment.

As IL-12p40 deficiency provided partial protection from liver toxicity we sought to understand the cellular basis of this phenotype. WT and IL12p40^−/− mice were injected with a single dose of 8F2 and liver immune cell phenotypes were assessed by FACS 7 days postinjection (figure 4 and online supplementary figure S3A). Following 8F2 treatment, we observed increased CD45⁺ cell frequencies in IL12p40^−/− compared with WT mice (figure 4A). This was defined by increased CD8⁺ T cells, NK cells, and CD11b⁺Ly6C⁺ and Ly6C⁻ monocytes. CD40 agonists have been reported to mediate hepatotoxicity via activation of monocyte-derived macrophages.15 Thus, we determined the influence of 8F2 on liver macrophage activation in WT and IL-12p40^−/− deficient mice. 8F2 did not affect total macrophage numbers 24 hours postinjection (figure 4B). However, the frequency of MHCII⁺ macrophages were increased by 8F2 (figure 4B,C), and this was somewhat diminished in IL12p40^−/− mice. Further assessment revealed that the accumulating IL12p40-dependent MHCII⁺ macrophages express CD14 and CD11b (figure 4B,C and online supplementary figure S3B), and are thus likely monocyte-derived rather than CD11b negative Kupffer cells.19 These findings support the notion that monocytes contribute to liver toxicity on CD40 agonist treatment, and that an IL-12p40-dependent process contributes to this axis.