Shedding light on cancer immunology at the molecular level: A quantum biochemistry study of representative PD-1/PD-L1 conformations

BACKGROUND

Programmed death 1 (PD-1) binding to PD-L1 is a potent mechanism used by immunogenic tumors to evade the immune system and the immune checkpoint PD-1PD-L1 has emerged as a promising target in the search for new drugs to improve cancer treatment. The crystallographic structure of humanPD-1humanPD-L1 shed light on the molecular characterization of this system and allowed computational studies to be carried out to characterize structural behaviors.

METHODS

This study demonstrated the importance of analyzing the flexibility of protein systems through molecular dynamics simulations (MDS) and its impacts on the interaction energy obtained through quantum biochemistry.

RESULTS

The computational results obtained provide a description of the flexibility and energetic profile of the PD-1PD-L1 contact surface using representative conformations from MDS. Variations of up to 50 % in the total interaction energy values were detected depending on the scrutinized conformation, which can be mainly attributed to the flexibility of the CC' loop, FG loop and ASP85-GLN91 of PD-1 and the MET58-LYS62 segment of PD-L1. Quantum biochemistry revealed the three hot spots in PD-L1: ARG113L-ARG125L > ILE54L-VAL76L > ALA18L-ASP26L; and two energetic hot spots in PD-1: ALA125-ARG139 > VAL63-GLN88. Nonetheless, VAL63-GLN88 and GLY124-ARG139 exhibit significant variation in interaction energy between different conformations, while ARG113L-ARG125L is the only hot spot with high energetic fluctuation on the PD-L1 surface.

CONCLUSION

This is the first application of MDS coupled to dimensionality reduction and density functional theory (DFT) demonstrating new structural and energetic features that might be useful in discovering/designing more potent PD-1PD-L1 inhibitors.

Clustering molecular dynamics conformations of the CC'-loop of the PD-1 immuno-checkpoint receptor

Molecular mechanisms within the checkpoint receptor PD-1 are essential for its activation by PD-L1 as well as for blocking such an activation via checkpoint inhibitors. We use molecular dynamics to scrutinize patterns of atomic motion in PD-1 without a ligand. Molecular dynamics is performed for the whole extracellular domain of PD-1, and the analysis focuses on its CC'-loop and some adjacent Cα-atoms. We extend previous work by applying common nearest neighbor clustering (Cnn) and compare the performance of this method with Daura clustering as well as UMAP dimension reduction and subsequent agglomerative linkage clustering. As compared to Daura clustering, we found Cnn less sensitive to cutoff selection and better able to return representative clusters for sets of different 3D atomic conformations. Interestingly, Cnn yields results quite similar to UMAP plus linkage clustering.

Exploiting the Innate Plasticity of the Programmed Cell Death-1 (PD1) Receptor to Design Pembrolizumab H3 Loop Mimics

Checkpoint blockade of the immunoreceptor programmed cell death-1 (PD1) with its ligand-1 (PDL1) by monoclonal antibodies such as pembrolizumab provided compelling clinical results in various cancer types, yet the molecular mechanism by which this drug blocks the PD1/PDL1 interface remains unclear. To address this question, we examined the conformational motion of PD1 associated with the binding of pembrolizumab. Our results revealed that the innate plasticity of both C'D and FG loops is crucial to form a deep binding groove (371 Å3 ) across several distant epitopes of PD1. This analysis ultimately provided a rational-design to create pembrolizumab H3 loop mimics [RDYRFDMGFD] into β-hairpin scaffolds. As a result, a 20-residue long β-hairpin peptide 1 e was identified as a first-in-class potent PD1-inhibitor (EC50 of 0.29 μM; Ki of 41 nM).

Molecular dynamics identifies semi-rigid domains in the PD-1 checkpoint receptor bound to its natural ligand PD-L1

Cells in danger of being erroneously attacked by leucocytes express PD-L1 on their surface. These cells activate PD-1 on attacking leucocytes and send them to death, thus curbing erroneous, autoimmune attack. Unfortunately, cancer cells exploit this mechanism: By expressing PD-L1, they guard themselves against leucocyte attack and thereby evade immune clearance. Checkpoint inhibitors are drugs which re-enable immune clearance of cancer cells by blocking the binding of PD-L1 to PD-1 receptors. It is therefore of utmost interest to investigate these binding mechanisms. We use three 600 ns all-atom molecular dynamics simulations to scrutinize molecular motions of PD-1 with its binding partner, the natural ligand PD-L1. Usually, atomic motion patterns are evaluated against whole molecules as a reference, disregarding that such a reference is a dynamic entity by itself, thus degrading stability of the reference. As a remedy, we identify semi-rigid domains, lending themselves as more stable and reliable reference frames against which even minute differences in molecular motion can be quantified precisely. We propose an unsupervised three-step procedure. In previous work of our group and others, minute differences in motion patterns proved decisive for differences in function. Here, several highly reliable frames of reference are established for future investigations based on molecular motion.

A method to test antibody cross-reactivity toward animal antigens for flow cytometry

The availability of cross-reacting antibodies and/or of antibodies working in flow cytometry is a major issue in the veterinary field. One of the main problems is the availability of certain positive controls. With this brief communication, we report an method to quickly screen a wide number of products without the need to look for positive biological samples. We propose this approach as a first step to select the best antibodies to test on biological specimens.

A Self-Controlled and Self-Healing Model of Bacterial Cells

A new kind of self-assembly model, morphogenetic (M) systems, assembles spatial units into larger structures through local interactions of simpler components and enables discovery of new principles for cellular membrane assembly, development, and its interface function. The model is based on interactions among three kinds of constitutive objects such as tiles and protein-like elements in discrete time and continuous 3D space. It was motivated by achieving a balance between three conflicting goals: biological, physical-chemical, and computational realism. A recent example is a unified model of morphogenesis of a single biological cell, its membrane and cytoskeleton formation, and finally, its self-reproduction. Here, a family of dynamic M systems (Mbac) is described with similar characteristics, modeling the process of bacterial cell formation and division that exhibits bacterial behaviors of living cells at the macro-level (including cell growth that is self-controlled and sensitive to the presence/absence of nutrients transported through membranes), as well as self-healing properties. Remarkably, it consists of only 20 or so developmental rules. Furthermore, since the model exhibits membrane formation and septic mitosis, it affords more rigorous definitions of concepts such as injury and self-healing that enable quantitative analyses of these kinds of properties. Mbac shows that self-assembly and interactions of living organisms with their environments and membrane interfaces are critical for self-healing, and that these properties can be defined and quantified more rigorously and precisely, despite their complexity.

Traversing through the Dynamic Protein-Protein Interaction Landscape and Conformational Plasticity of PD-1 for Small-Molecule Discovery

Monoclonal antibodies (mAbs) blocking the PD-1/PD-L1 interface have shown remarkable success in treating malignancies, but they may also initiate lethal immune-related adverse events. Small molecules may circumvent the mAb limitations; however, none has entered clinical trials targeting PD-1. Its complex protein-protein interaction interfaces necessitate an atomic-level understanding of recognition and binding mechanisms. Hence, we have aimed to highlight the PD-1's sequence-structure-dynamic-function link with its cognate ligands and diversely reported inhibitors. We focus primarily on the anti-PD-1 mAbs, their mode of actions, and interactions with PD-1 epitopes. The comparison of co-crystals showed that these ligands/inhibitors harness the PD-1's conformational plasticity and structural determinants differentially. The relationship between modulator binding patterns and biological activity is demonstrated using interaction fingerprinting of all reported human PD-1 co-crystals. The significant dynamical events and hot-spot residues underpinned from crystallographic wealth and computational studies have been highlighted to expedite small-molecule discovery.

Investigation of Unprecedented Sites and Proposition of New Ligands for Programmed Cell Death Protein I through Molecular Dynamics with Probes and Virtual Screening

Cancer immunotherapy has attracted increasing attention over the last few years. Programmed cell death protein 1 (PD-1) promotes self-tolerance and inhibits immune responses by modulating the T-cell function. The interaction between PD-1 and programmed cell death ligand-1 (PD-L1) leads to immune exhaustion, protecting cancer cells from destruction. Here, we computationally designed a novel ligand named 1508 that binds to an unprecedented PD-1 cavity identified by MixMD and defined by amino acid residues Lys78 to Val97. We showed through a set of MD simulations totaling 12.5 μs that ligand 1508 establishes frequent cation-π and hydrogen bonding interactions with amino acid residues Lys78 and Arg86, respectively, and stabilizes the PD-1 C'D loop in a conformation that does not favor PD-1-PD-L1 complex formation. This study highlights the power of MixMD in exposing new cavities prone to protein-protein complex inhibition and establishes the basis for the design of new molecules that target the PD-1 C'D cavity as an alternative for exploring the modulation of the PD-1-PD-L1 complex in cancer therapy.

Blockade of the checkpoint PD-1 by its ligand PD-L1 and the immuno-oncological drugs pembrolizumab and nivolumab

We investigate the interaction between the programmed cell death protein 1 (PD-1) and the programmed cell death ligand 1 (PD-L1), as well as the immuno-oncological drugs pembrolizumab (PEM), and nivolumab (NIV), through quantum chemistry methods based on the Density Functional Theory (DFT) and the molecular fractionation with conjugate caps (MFCC) scheme, in order to map their hot-spot regions. Our results showed that the total interaction energy order of the three complexes is in good agreement with the experimental binding affinity order: PD-1/PEM > PD-1/NIV > PD-1/PD-L1. Besides, a detailed investigation revealed the energetically most relevant residue-residue pairs-interaction for each complex. Our computational results give a better understanding of the interaction mechanism between the protein PD-1 and its ligands (natural and inhibitors), unleashing the immune surveillance to destroy the cancer cells by decreasing their immune evasion. They are also an efficient alternative towards the development of new small-molecules and antibody-based drugs, pointing out to new treatments for cancer therapy.

Visualization of the Interfacial Electrostatic Complementarity: A Method for Analysis of Protein-Protein Interaction Based on Ab Initio Quantum Chemical Calculations

In this study, we report a method for analyzing the protein-protein interaction based on ab initio quantum chemical calculations, which we refer to as "Visualization of the interfacial electrostatic complementarity (VIINEC)." This method visually provides the electrostatic complementarity at the protein-protein interface; in addition, the ratio of the attractive interaction is calculated. Illustrative calculations revealed that VIINEC could successfully quantify the electronic induced fit owing complex formation, which was responsible for 5%-10% of the total electrostatic complementarity. Furthermore, the contribution of each amino acid to the electrostatic complementarity was evaluated, providing useful information for various applications, including rational antibody designs. Interestingly, a part of the mechanism causing the specificity of the protein-protein bindings was also demonstrated using VIINEC. This is an important achievement of this study because the specificity of the biomolecular interactions is essential for biological functions.

Interplay among Structural Stability, Plasticity, and Energetics Determined by Conformational Attuning of Flexible Loops in PD-1

The dynamics and plasticity of the PD-1/PD-L1 axis are the bottlenecks for the discovery of small-molecule antagonists to perturb this interaction interface significantly. Understanding the process of this protein-protein interaction (PPI) is of fundamental biological interest in structure-based drug designing. Food and Drug Administration (FDA)-approved anti-PD-1 monoclonal antibodies (mAbs) are the first-in-class with distinct binding modes to access this axis clinically; however, their mechanistic aspects remain elusive. Here, we have unveiled the interactive interfaces with PD-L1 and mAbs to investigate the native plasticity of PD-1 at global (structural and dynamical) and local (residue side-chain orientations) levels. We found that the structural stability and coordinated C_α movements are increased in the presence of PD-1's binding partners. The rigorous analysis of these PPIs using computational biophysical approaches revealed PD-1's intrinsic plasticity, its concerted loops' movement (BC, FG, and CC'), distal side-chain motions, and the thermodynamic landscape, which are perturbed remarkably from its unbound to bound states. Based on intra-/inter-residues' contact networks and energetics, the hot-spots have been identified that were found to be essential to arrest the dynamical motions of PD-1 significantly for the rational design of therapeutic agents by mimicking the mAbs mechanism.

Pembrolizumab Induces an Unexpected Conformational Change in the CC'-loop of PD-1

Simple Summary

Cancer cells are normally destructed by killer T-cells. However, T-cells expose the PD-1 receptor on their surface, acting as a checkpoint: If it is activated through a special molecule, PD-L1, the T-cell kills itself, ending the attack. Cells often need to present PD-L1 to prevent T-cells from over-aggressive attacks which cause autoimmune disease. There are tumors which also present PD-L1, thereby evading natural clearing, allowing them to continue growing. New anticancer drugs (checkpoint inhibitors: nivolumab and pembrolizumab) disrupt this evasion: They competitively bind to PD-1, without activating it, and re-enable immune tumor destruction. We scrutinize the binding mechanisms via molecular dynamics simulation. We demonstrate that these drugs deform the CC′-loop of the PD-1 in ways differing from those seen with PD-L1 as a binding partner. Pembrolizumab induces a new conformation of the CC′-loop not known to date. These findings might pave the way for the development of new anti-cancer drugs.

Abstract

To improve cancer immunotherapy, a clearer understanding of key targets such as the immune checkpoint receptor PD-1 is essential. The PD-1 inhibitors nivolumab and pembrolizumab were recently approved by the FDA. The CC′-loop of PD-1 has been identified as a hotspot for drug targeting. Here, we investigate the influence of nivolumab and pembrolizumab on the molecular motion of the CC′-loop of PD-1. We performed molecular dynamics simulations on the complete extracellular domain of PD-1, in complex with PD-L1, and the blocking antibodies nivolumab and pembrolizumab. Conformations of the CC′-loop were analyzed unsupervised with the Daura et al. clustering algorithm and multidimensional scaling. Surprisingly, two conformations found were seen to correspond to the ‘open’ and ‘closed’ conformation of CC′-loop in apo-PD-1, already known from literature. Unsupervised clustering also surprisingly reproduced the natural ligand, PD-L1, exclusively stabilizing the ‘closed’ conformation, as also known from literature. Nivolumab, like PD-L1, was found to shift the equilibrium towards the ‘closed’ conformation, in accordance with the conformational selection model. Pembrolizumab, on the other hand, induced a third conformation of the CC′-loop which has not been described to date: Relative to the conformation ‘open’ the, CC′-loop turned 180° to form a new conformation which we called ‘overturned’. We show that the combination of clustering and multidimensional scaling is a fast, easy, and powerful method in analyzing structural changes in proteins. Possible refined antibodies or new small molecular compounds could utilize the flexibility of the CC′-loop to improve immunotherapy.

Molecular dynamics of the immune checkpoint programmed cell death protein I, PD-1: conformational changes of the BC-loop upon binding of the ligand PD-L1 and the monoclonal antibody nivolumab

BACKGROUND

The immune checkpoint receptor programmed cell death protein I (PD-1) has been identified as a key target in immunotherapy. PD-1 reduces the risk of autoimmunity by inducing apoptosis in antigen-specific T cells upon interaction with programmed cell death protein ligand I (PD-L1). Various cancer types overexpress PD-L1 to evade the immune system by inducing apoptosis in tumor-specific CD8+ T cells. The clinically used blocking antibody nivolumab binds to PD-1 and inhibits the immunosuppressive interaction with PD-L1. Even though PD-1 is already used as a drug target, the exact mechanism of the receptor is still a matter of debate. For instance, it is hypothesized that the signal transduction is based on an active conformation of PD-1.

RESULTS

Here we present the results of the first molecular dynamics simulations of PD-1 with a complete extracellular domain with a focus on the role of the BC-loop of PD-1 upon binding PD-L1 or nivolumab. We could demonstrate that the BC-loop can form three conformations. Nivolumab binds to the BC-loop according to the conformational selection model whereas PD-L1 induces allosterically a conformational change of the BC-loop.

CONCLUSION

Due to the structural differences of the BC-loop, a signal transduction based on active conformation cannot be ruled out. These findings will have an impact on drug design and will help to refine immunotherapy blocking antibodies.

Investigating the Role of the N-Terminal Loop of PD-1 in Binding Process Between PD-1 and Nivolumab via Molecular Dynamics Simulation

The blockade of immune checkpoints, such as programmed death receptor 1 (PD-1) and programmed death ligand 1 protein (PD-L1), is a promising therapeutic approach in cancer immunotherapy. Nivolumab, a humanized IgG4 antibody targeting PD-1, was approved by the US Food and Drug Administration for several cancers in 2014. Crystal structures of the nivolumab/PD-1 complex show that the epitope of PD-1 locates at the IgV domain (including the FG and BC loops) and the N-terminal loop. Although the N-terminal loop of PD-1 has been shown to play a dominant role in the complex interface of the static structure, its role in the dynamic binding process has not been illustrated clearly. Here, eight molecular systems were established for nivolumab/PD-1 complex, and long-time molecular dynamics simulations were performed for each. Results showed that the N-terminal loop of PD-1 prefers to bind with nivolumab to stabilize the interface between IgV and nivolumab. Furthermore, the binding of the N-terminal loop with nivolumab induces the rebinding between the IgV domain and nivolumab. Thus, we proposed a two-step binding model for the nivolumab/PD-1 binding, where the interface switches to a high-affinity state with the help of the N-terminal loop. This finding suggests that the N-terminal loop of PD-1 might be a potential target for anti-PD-1 antibody design, which could serve as an important gatekeeper for the anti-PD-1 antibody binding.

Exploring the conformational dynamics of PD1 in complex with different ligands: What we can learn for designing novel PD1 signaling blockers?

Activation of T cells triggers the expression of regulatory molecules like the programmed cell death 1 (PD1) protein. The association of PD1 with the natural ligands PDL1 and PDL2 induces an inhibitory signal that prevents T cells from proliferating and exerting effector functions. However, little is known about how the binding of the ligands induce the PD1 inhibitory signal over T cells effector functions. Here, we explore the dynamics of PD1 free, and in complex with different PDL1 variants as well as the therapeutic antibodies nivolumab and pembrolizumab in order to assess the conformational changes in PD1 related to the signaling process. Our simulations suggest a pre-conformational selection mechanism for the binding of the different PDL1 variants, while an induced-fit model fits better for the molecular recognition process of the therapeutic antibodies. A deep analysis of the changes on PD1 movement upon the binding to different ligands revealed that as larger is the difference in the conformation adopted by loop C'D with respect to the complex with PDL1 is higher the ligand ability to reduce the PD1 inhibitory signaling. This behavior suggests that targeting specific conformations of this loop can be useful for designing therapies able to recover T cells effector functions.

Recognition of PDL1/L2 by different induced-fit mechanisms of PD1: a comparative study of molecular dynamics simulations

The different binding mechanism for PD1/PDL1 and PD1/PDL2 complexes.

Computational analysis of hot spots and binding mechanism in the PD-1/PD-L1 interaction

Programmed cell death protein-1 (PD-1) is an important immunological checkpoint and plays a vital role in maintaining the peripheral tolerance of the human body by interacting with its ligand PD-L1. The overexpression of PD-L1 in tumor cells induces local immune suppression and helps the tumor cells to evade the endogenous anti-tumor immunity. Developing monoclonal antibodies against the PD-1/PD-L1 protein–protein interaction to block the PD-1/PD-L1 signaling pathway has demonstrated superior anti-tumor efficacy in a variety of solid tumors and has made a profound impact on the field of cancer immunotherapy in recent years. Although the X-ray crystal structure of the PD-1/PD-L1 complex has been solved, the detailed binding mechanism of the PD-1/PD-L1 interaction is not fully understood from a theoretical point of view. In this study, we performed computational alanine scanning on the PD-1/PD-L1 complex to quantitatively identify the hot spots in the PD-1/PD-L1 interaction and characterize its binding mechanisms at the atomic level. To the best of our knowledge, this is the first time that theoretical calculations have been used to systematically and quantitatively predict the hot spots in the PD-1/PD-L1 interaction. We hope that the predicted hot spots and the energy profile of the PD-1/PD-L1 interaction presented in this work can provide guidance for the design of peptide and small molecule drugs targeting PD-1 or PD-L1.

The hot spots quantitatively predicted by the recently developed MM/GBSA/IE method reveal a hydrophobic core in the PD-1/PD-L1 interaction.

An integrated molecular modeling approach for the tryptase monomer-curcuminoid recognition analysis: conformational and bioenergetic features

Human mast cell tryptase has been shown as an activating enzyme in matrix degradation process. The previous study suggest that tryptase either alone or in joining with activation of metalloproteinases, can associate in extra cellular matrix damage and the possible destruction of the basement membrane resulting in photoaging. Therefore the inhibition of tryptase activity is one of the most important therapeutic strategies against the photoaging. Curcumin has been shown to be a potential agent for preventing and/or treating the photoaging induced by UV radiation. However, the protective effect of curcumin against the photoaging through the tryptase inhibition is still inadequately understood. In this work, computational methods to characterize the structural framework and define the atomistic details of the determinants for the tryptase inhibition mechanism by curcuminoids were performed. By molecular docking, three putative binding models able to efficiently bind all curcuminoids were identified. Analysis of molecular dynamics simulations revealed that cyclocurcumin, curcumin glucuronide, and curcumin, the most effective inhibitors from the three models, modified significant tryptase monomer rigidity by binding in all the possible sites. The result of these binding events is the suppression of the functional enzymatic motions involving the binding of substrates to the catalytic site. On the basis of this finding may thus be beneficial for the development of new natural inhibitors for the therapeutic remedy of photoaging, targeting and modulating the activity of tryptase.

Computational investigation of the α2β1 integrin–collagen triple helix complex interaction

In this paper, quantum biochemistry methods have been used to describe important protein–protein interactions for the complex integrin–collagen.

Mechanistic insights into peptide and ligand binding of the ATAD2-bromodomain via atomistic simulations disclosing a role of induced fit and conformational selection

Atomistic simulations of the ATAD2-bromodomain disclose a role of induced fit and conformational selection upon ligand and peptide binding.

Understanding the structural and energetic basis of PD-1 and monoclonal antibodies bound to PD-L1: A molecular modeling perspective

BACKGROUND

The inhibitors blocking the interaction between programmed cell death protein 1(PD-1) and programmed death-ligand 1(PD-L1) can activate the immune response of T cell and eliminate cancer cells. The crystallographic studies have provided structural insights of the interactive interfaces between PD-L1 and its protein ligands. However, the hotspot residues on PD-L1 as well as structural and energetic basis for different protein ligands still need to be further investigated.

METHODS

Molecular modeling methods including molecular dynamics simulation, per-residue free energy decomposition, virtual alanine scanning mutagenesis and residue-residue contact analysis were used to qualitatively and quantitatively analyze the interactions between PD-L1 and different protein ligands.

RESULTS

The results of virtual alanine scanning mutagenesis suggest that Y56, Q66, M115, D122, Y123, R125 are the hotspot residues on PD-L1. The residue-residue contact analysis further shows that PD-1 interacts with PD-L1 mainly by F and G strands while monoclonal antibodies like avelumab and BMS-936559 mainly interact with PD-L1 by CDR2 and CDR3 loops of the heavy chain.

CONCLUSIONS

A structurally similar β-hairpin peptide with 13 or 14 residues was extracted from each protein ligand and these β-hairpin peptides were found tightly binding to the putative hotspot residues on PD-L1.

GENERAL SIGNIFICANCE

This study recognizes the hotspot residues on PD-L1 and uncovers the common structural and energetic basis of different protein ligands binding to PD-L1. These results will be valuable for the design of small molecule or peptide inhibitors targeting on PD-L1.

When theory meets experiment: the PD-1 challenge

Applying atomistic computational modeling to drug discovery has proven to be a hugely successful approach, allowing drug-receptor interactions to be predicted and drugs to be optimized for potency, selectivity, and safety. However, when it comes to predicting protein-protein interactions and to rationally designing regulators of these interactions, computational tools often fail. Here, we report one of the rare instances where state-of-the-art computer simulations, guided by experiment, were able to correctly predict one of the most sophisticated protein-protein interactions. We revisit our previous discovery of the complex of human PD-1 with the ligand PD-L1 and compare our earlier findings with the recently published crystal structure of the same complex. Side-by-side comparison of the model of the complex with its crystal structure reveals outstanding agreement and suggests that our protein-protein prediction workflow could be applied to similar problems.

The Too Many Faces of PD-L1: A Comprehensive Conformational Analysis Study

In the current study, we focused on the immune-checkpoints PD-1 pathway and in particular on the ligand PD-L1. We studied the conformational dynamics of PD-L1 through principal component analysis of existing crystal structures combined with classical and accelerated molecular dynamics simulations. We identified the maximum structural displacements that take place in all PD-L1 crystal structures and in the molecular dynamics trajectories. We found that these displacements are attributed to specific flexible regions in the protein. We also investigated the conformational preference for small molecule binding and highlighted a methionine residue at the binding site, which plays a key role in drug binding. The binding mechanism of PD-L1 to other binding partners is also discussed in detail from a computational perspective. We hope that the data presented here support the ongoing efforts to discover effective therapies targeting the PD-1 immune-checkpoint pathway.