PD-L1 and PD-L2 differ in their molecular mechanisms of interaction with PD-1

Abstract

The programmed death-1 (PD-1) molecule is involved in peripheral tolerance and in the immune escape mechanisms during chronic viral infections and cancer. PD-1 interacts with two ligands, PD-L1 and PD-L2. We have investigated the molecular mechanisms of PD-1 interactions with its ligands by surface plasmon resonance and cell surface binding as well as the ability of the two ligands to compete for PD-1 binding. PD-L1 and PD-L2 bound PD-1 with comparable affinities, but striking differences were observed at the level of the association and dissociation characteristics. PD-L1, but not PD-L2, had a delayed interaction reminiscent of a phenomenon of conformational transition. These mechanisms were confirmed by using PD-L1 mAbs that delayed the dissociation of PD-L1 from PD-1. This mechanism was not restricted to PD-1 binding since PD-L1 behaved in a similar manner with its second ligand, CD80. Finally, we could demonstrate that PD-L1 and PD-L2 competed for PD-1 binding and conversely, an antagonist PD-1 mAb blocked both PD-L1 and PD-L2 binding to PD-1 and strongly enhanced T-cell proliferation. These data further emphasize the differential molecular mechanisms of interaction of PD-L1 and PD-L2 with PD-1, and suggest possible new approach for the therapy of chronic infection, cancer and transplantation.

Figure 1. PD-L1 and PD-L2 do not bind to PD-1 with the same molecular mechanism

We performed SPR analysis using BIAcore T100 and performed the analysis of PD-L1-Ig, PD-L2-Ig binding to PD1-Ig as well as CD80-Ig binding to PD-L1-Ig and CTLA-4-Ig. (A) Top row: Superimposed sensorgrams representative of PD-L1 and PD-L2 binding to PD-1-Ig chips; Lower row: Sensorgrams of PD-L1-Ig and CTLA-Ig binding to CD80-Ig chips. Proteins were injected for two minutes at a flow rate of 40 μl/min onto PD-1-Ig or CD80-Ig chips and allowed to dissociate for three more minutes. The data shown are representative of five separate experiments. (B) Superimposed sensorgrams showing short (red) and long (black) injections of PD-L1 (top row) and PD-L2 (lower row) onto PD-1 chip respectively. Proteins at 10 μg/ml were injected for one (short) or seven minutes (long) at a flow rate of 10 μl/min onto the PD-1 chip. Sensorgrams were normalized in the Y axis and aligned in the X axis at the end of injection in order to align the dissociation phases up The data shown are representative of two separate experiments.

Figure 2. Anti-PD-L1 mAbs (PD-L1.1 and PD-L1.2) stabilize the binding of PDL1 to PD-1

(A) Superimposed sensorgrams showing the injections of PD-L1-Ig (none) or the injections of PD-L1-Ig pre-incubated with anti-PD-L1 antibodies onto PD-1 chip. PD-L1-Ig at 10 μg/ml were pre-incubated with PD-L1.1 (red), PD-L1.2 (black) and PD-L1.3 (blue) anti-PD-L1 antibody Fabs at a saturating concentration of 100 μg/ml and injected for 10 minutes at a flow rate of 10 μl/min onto the PD-1-Ig chip. Sensorgrams were normalized in the Y axis and aligned in the X axis at the end of injection. The data shown are representative of three separate experiments. (B) PD-L1.1, PD-L1.2 and PD-L1.3 Fab mAbs were incubated with PD-L1 expressing cells for 5 or 30 minutes before PD-1-Ig incubation. The binding of PD-1 Ig was revealed with Goat anti human (GAH) conjugated PE, the MFI ratio was indicated in the Y axis. The data shown are representative of three separate experiments.

Figure 3. PD-L1 and PD-L2 cross compete for PD-1 binding

The experiments were performed using both SPR analysis using BIAcore (A and B) and flow cytometry (C) SPR and Facs analysis PD-L1 and CTLA-4 binding to CD80 (A) Analysis by SPR of the binding of PD-L1 to PD-1 or CD80 coated on chips following preincubation with PD-L2 or PD-1 respectively. The data shown are representative of two separate experiments. Top row: The PD-1 chips were pre-incubated twice with saturating amounts of PDL2 and PD-L1 then were injected in a third step at 10 μg/ml for 2 minutes at a flow rate of 10 μl/min without removing bound PD-L2. Sensorgrams showing the PD-L1 binding in the presence or absence of PD-L2 occupancy are superimposed. Middle row: PD-1 proteins were pre-incubated with saturating amounts of PD-L2 and the resulting complexes were injected at 10 μg/ml for 2 minutes at a flow rate of 10 μl/min onto PD-L1-Ig chips. Sensorgrams showing the PD-1 binding alone or complexed with PD-L2-Ig are superimposed. Lower row: PD-L1-Ig proteins were pre-incubated with saturating amounts of PD-1- Ig and the resulting complexes were injected at 10 μg/ml for 2 minutes at a flow rate of 10 μl/min onto CD80-Ig chips. Sensorgrams showing the PD-L1 binding alone or complexed with PD-1 are superimposed. (B) CTLA-4 does not to prevent the binding of PD-L1-Ig to CD80 using SPR analysis. The data shown are representative of two separate experiments. Superimposed sensorgrams representative of PD-L1 (top row) and CTLA-4 (lower row) binding to CD80-Ig chips following or not pre-incubation with CTLA-4-Ig. The CD80 chips were pre-incubated twice with saturating concentrations of CTLA-4- Ig, then in a third step PD-L1-Ig or CTLA-4-Ig used at 10 μg/ml were injected for 2 minutes at a flow rate of 10 μl/min onto CD80-Ig chip and allowed to dissociate for 2 more minutes (C) CTLA-4 and PD-L1 bind together to CD80 on the cell surface Top row: CTLA-4-Ig can bind to PD-L1 expressing COS cell in presence of CD80 but not PD-1. Biotinylated CTLA-4-Ig protein at 2 μg/ml was incubated with CD80- Ig or PD-1-Ig proteins (from 0 to 20 μg/ml) before addition to PD-L1 expressing cells. The binding of biotinylated CTLA-4 proteins were revealed with StreptAvidin-conjugated with PE, the MFI was indicated in the Y axis. The data shown are representative of three separate experiments. Lower row: CTLA-4-Ig can bind to CD80 expressing cells in presence of PD-L1-Ig. Biotinylated CTLA-4-Ig at 2 μg/ml was incubated with increasing concentrations of PD-L1-Ig or CTLA-4-Ig (from 0 to 10 μg/ml) before addition to the CD80+ Raji cell. The binding of biotinylated CTLA-4-Ig was detected using StreptAvidin-conjugated with PE, the MFI was indicated in the Y axis. The data shown are representative of three separate experiments.

Figure 4. The PD1.3 mAb blocks the PD-1 binding to both PD-L1 and PD-L2 and enhances T cell activation

(A) A schematic representation of the surface competitive binding inhibition used in (B). In a first step the immobilized PD-1 proteins are saturated using the antibody Fabs and the corresponding PD-1 ligands are injected as a soluble analyte in a second step. (B) PD-L1-Ig (top row) and PD-L2-Ig (lower row) were injected at 10 μg/ml for two minutes at a flow rate of 10μl/min onto PD-1 chip (none), or PD-1 chip pre-incubated with anti PD-1 Fab, PD-1.3 or PD-1.6. Sensorgrams showing the binding of the PD-1 ligands in the different situations are superimposed. The data shown are representative of three separate experiments. (C) PD-1. 3 mAb prevents PD-L1-Ig and PD-L2-Ig binding to cells expressing PD-1. Flow cytometry analysis on PD-1 expressing COS cells. PD-1.3, PD-1.6 or isotype control mAbs, at increasing concentration (from 0 to 20 μg/ml) were pre-incubated with PD-L1 or PD-L2-Ig proteins. The binding of ligands proteins were revealed with goat anti human (GAH) conjugated with PE, the MFI was indicated in the Y axis. The data shown are representative of three separate experiments. (D) PD-1.3 and PD-L1.3 mAbs are able to induce the IFN-γ and IL-10 production in CD4⁺ T cells upon DC cell contacts. Allogenic iDC were cocultured with CD4⁺ T cells with increasing concentration of PD-1 and PD-L1 mAbs or isotype control (from 0 to 20 μg/ml). Cultures were incubated for 5 days, supernatants were removed for cytokine analysis. The levels of IFN-γ production and IL-10 production were determined in duplicate by ELISA detection. The data shown are representative of two separate experiments.

Figure 5. Diagrams of potential mechanisms for PD-1/PD-L1 interactions

(A and B) A schematic representation of the mechanism of PD-L1 binding to PD-1 and CD80 appears to involve significant and detectable conformational change. (C and D) A schematic representation of the mechanism of PD-L2 and CTLA-4 interact directly with PD-1 and CD80, respectively.