Combination CD200R/PD-1 blockade in a humanised mouse model

Abstract

There is an increasing number of immune-checkpoint inhibitors being developed and approved for cancer immunotherapy. Most of the new therapies aim to reactivate tumour-infiltrating T cells, which are responsible for tumour killing. However, in many tumours, the most abundant infiltrating immune cells are macrophages and myeloid cells, which can be tumour-promoting as well as tumouricidal. CD200R was initially identified as a myeloid-restricted, inhibitory immune receptor, but was subsequently also found to be expressed within the lymphoid lineage. Using a mouse model humanised for CD200R and PD-1, we investigated the potential of a combination therapy comprising nivolumab, a clinically approved PD-1 blocking antibody, and OX108, a CD200R antagonist. We produced nivolumab as a murine IgG1 antibody and validated its binding activity in vitro as well as ex vivo. We then tested the combination therapy in the immunogenic colorectal cancer model MC38 as well as the PD-1 blockade-resistant lung cancer model LLC1, which is characterised by a large number of infiltrating myeloid cells, making it an attractive target for CD200R blockade. No significant improvement of overall survival was found in either model, compared to nivolumab mIgG1 monotherapy. There was a trend for more complete responses in the MC38 model, but investigation of the infiltrating immune cells failed to account for this. Importantly, MC38 cells expressed low levels of CD200, whereas LLC1 cells were CD200-negative. Further investigation of CD200R-blocking antibodies in tumours expressing high levels of CD200 could be warranted.

Keywords: CD200R; PD-1; cancer immunotherapy; monoclonal antibody; nivolumab.

Graphical Abstract

Figure 1.

Validation of produced mIgG1 antibodies in vitro and ex vivo. (A) mNivo and OX108 were captured on a Protein A SPR chip. MOPC21 was captured in the reference channel. The corresponding protein (hPD-1-2xHis and hCD200R-mFc-His) was injected in 1:3 dilutions as specified. The Biacore Insight Evaluation Software was used to fit a 1:1 model and derived affinity constants (k_a, k_d, and K_D) are shown in the table below. (B) Nivolumab mIgG1 (mNivo) and nivolumab hIgG4 were captured on a Protein A SPR chip. MOPC21 was captured in the reference channel. hPD-1-2xHis was injected in 1:3 dilutions with concentrations from 2777.78 nM to 1.27 nM. The Biacore Insight Evaluation Software was used to fit a 1:1 model and derived affinity constants (k_a, k_d, and K_D) are shown in the tables. (C) Jurkat reporter cells (GFP under an NFκB promoter) expressing PD-1 or a chimeric protein CD200R-PD-1 were stimulated with TCS cells expressing PD-L1/CD200. Respective blocking antibodies or MOPC21 were added, and cells were incubated for 24 h. GFP expression was analysed by flow cytometry and normalised to Jurkat-cells-only cultures. Data are presented as mean ± standard deviation (SD). (D) Splenocytes from double knock-in mice were stimulated with PMA/ionomycin or control. After 24 h, cells were blocked using produced antibodies. Competition for binding was analysed using commercial flow cytometry antibodies. Bar charts show the MFI of the respective channels.

Figure 2.

Baseline characterisation of MC38 and LLC1 cancer models. (A) In vitro cultured cells were stimulated with different concentrations of IFN-γ or control for 24 h (left panel). Levels of CD200 were measured using Quantibrite beads (BD Biosciences) of in vitro cultured cells (middle panel). After 14 days of in vivo growth, tumours were harvested and digested (right panel). Expression of MHC-I, PD-L1, and CD200 was analysed by flow cytometry. (B) After 14 days of in vivo growth, tumours were harvested and digested. Immune cells were enriched using density gradient centrifugation. The abundance of different immune cell populations was analysed (left panel) as well as the MFI of CD200R (right panel). Data are presented as mean ± SD. (C) Mice were subcutaneously injected with MC38 and LLC1 cells and tumour volume was measured every other day using calliper measurements. Once tumours reached a certain average volume (40 mm³ for MC38, 20 mm³ for LLC1), treatment with MOPC21 or mNivo was started. Mice were treated with 200 µg of antibody; four times over 2 weeks. All animal experiments were randomised and blinded. n = 10/group.

Figure 3.

Combination therapy in LLC1 tumours. (A) Illustration describing the workflow for (B) and (C); red lines indicate antibody treatment. (B) LLC1 cells were subcutaneously injected into mice and tumour volume was measured every other day. Once the average tumour volume reached 20 mm³, mice were randomised into four different groups and treatment was started. Mice were treated with 200 µg/antibody/treatment, receiving four treatments over 2 weeks. n = 10/group. (C) On day 23, all remaining mice were culled, tumours were harvested and digested. Immune cells were enriched using density gradient centrifugation and the abundance of different immune cell populations was analysed. The line shows the median, the ‘+’ shows the mean. Whiskers display the 10–90 percentile.

Figure 4.

Combination therapy in MC38 tumours. (A) MC38 cells were subcutaneously injected into mice and tumour volume was measured every other day. Once the average tumour volume reached 40 mm³, mice were randomised into four different groups and treatment was started. Mice were treated with 200 µg/antibody/treatment, receiving four treatments over 2 weeks. n = 10/group. (B) Illustration describing the workflow for (C); red lines indicate antibody treatment. (C) One day after the final treatment, mice were culled, tumours were harvested and digested. Immune cells were enriched using density gradient centrifugation and the abundance of different immune cell populations was analysed. Three independent experiments with n = 5/group were pooled. The line shows the median, the ‘+’ shows the mean. Whiskers display the 10–90 percentile.