Computer-aided drug design in new druggable targets for the next … · 2020. 6. 13. · computationally driven drug discovery in the rational search for novel bioactive compounds

OVERV I EW

Computer-aided drug design in new druggable targets for thenext generation of immune-oncology therapies

Rita C. Acúrcio1 | Anna Scomparin2 | Ronit Satchi-Fainaro2 | Helena F. Florindo1 | Rita C. Guedes1

1Research Institute for Medicines (iMed.ULisboa),Faculty of Pharmacy, Universidade de Lisboa,Lisbon, Portugal2Department of Physiology and Pharmacology,Sackler Faculty of Medicine, Tel Aviv University,Tel Aviv, Israel

CorrespondenceHelena F. Florindo, Research Institute forMedicines (iMed.ULisboa), Faculty of Pharmacy,Universidade de Lisboa, 1649-003, Lisbon,Portugal.Email: [email protected] C. Guedes, Research Institute for Medicines(iMed.ULisboa), Faculty of Pharmacy,Universidade de Lisboa, 1649-003, Lisbon,Portugal.Email: [email protected]

Funding informationIsraeli Ministry of Health, and FCT-MCTES,Grant/Award Number: ENMed/0051/2016;European Structural & Investment Funds throughthe COMPETE Programme and FCT-MCTES,Grant/Award Numbers: UID/DTP/04138/2013,SAICTPAC/0019/2015, PTDC/QEQ-MED/7042/2014; Fundação para a Ciência e a Tecnologia,Ministério da Ciência, Tecnologia e EnsinoSuperior (FCT-MCTES), Grant/Award Number:PD/BD/128238/2016

Immune modulatory pathways have emerged as innovative and successful targetsin cancer immunotherapy. Current therapeutics include monoclonal antibodies,which have shown impressive clinical results in the treatment of several types oftumors. However, the failure to show response in the majority of patients and theinduction of severe immune-related adverse effects are among the major draw-backs. Latest efforts to achieve new approaches to target additional pathwaysand/or protein responsible for immune evasion toward the development of immunemodulatory small molecules have been devised. The potential of innovativecomputational-aided drug design tools to accelerate the identification and design ofnew optimized and validated immune small molecules modulators will play a keyrole in the next generation of cancer immunotherapy drug discovery. Nevertheless,the lack of structural information regarding the immune modulatory pathways andother components within tumor microenvironment has hampered the rationaldesign of those small-molecule modulators, by preventing the use of such method-ologies. Herein we provide an overview on structural elucidation on known regula-tors of immune modulatory pathways (adenosine A2A receptor [A2AR], stimulatorof interferon genes [STING], indoleaine 2,3-dioxygenase 1 [IDO1], and 4-1BB)within cancer microenvironment. This knowledge on immune modulatory molecu-lar targets is essential to advance the understanding of their binding mode andguide the design of novel effective targeted anticancer medicines.

This article is categorized under:Molecular and Statistical Mechanics > Molecular Dynamics and Monte-CarloMethods

Structure and Mechanism > Molecular StructuresSoftware > Molecular Modeling

KEYWORDS

4-1BB, A2AR, computational-aided drug design, IDO1, immune checkpointreceptors, STING, tumor microenvironment

1 | INTRODUCTION

Cancer immunotherapy has dramatically changed the cancer treatment landscape and outcome, especially when addressing themetastatic state of this devastating disease. Among therapeutic strategies, the regulation of immune checkpoint receptors(ICRs) has evolved to significant responses in patients affected by various tumor types, often resistant to all other availabletherapeutic options. The success inherent to the regulation of ICRs prompted the development of different monoclonal anti-bodies to modulate the biological function of several immune checkpoint proteins involved in tumor immune evasion. So far,

Received: 31 July 2018 Accepted: 26 September 2018

DOI: 10.1002/wcms.1397

WIREs Comput Mol Sci. 2019;9:e1397. wires.wiley.com/compmolsci © 2018 Wiley Periodicals, Inc. 1 of 27https://doi.org/10.1002/wcms.1397

http://wires.wiley.com/compmolsci

https://doi.org/10.1002/wcms.1397

all approved immune checkpoint modulators in the clinic are monoclonal antibodies. These have revolutionized cancer ther-apy because of their remarkable clinical activity. However, their failure to show response in most patients, their requiredadministration by an intravenous injection and the induction of severe immune-related adverse effects,1–3 constitute some oftheir important drawbacks.

In this context, considerable research effort is being deployed to find the next generation of effective cancer immunothera-peutics. Pioneering efforts include the design of novel immune modulatory small molecules (synthetic and natural products)to specifically target emerging pathways responsible for immune evasion. Small molecules are more affordable and maybecome invaluable alternatives to monoclonal antibodies. However, small-molecule drug discovery is a complex and challeng-ing multidimensional process, focusing on multiple aspects, as the activity, efficacy, pharmacokinetics, pharmacodynamics,and safety of potential novel candidates. The average length of time from hit discovery to the approval of a new drug is at least14 years, encompassing the investment of billions of dollars. To circumvent and accelerate this lengthy and costly drug dis-covery and development process, over the past two decades, drug designers have been exploring computational-aided drugdesign (CADD) tools to design and identify new drug-like chemical entities that can overcome clinical attrition and lack ofefficiency.

Revolutionary and accelerated performance hardware architectures, including graphical processor units, innovative algo-rithms and software advances are now available for high-level computations in a time-affordable way pushing the frontiers ofcomputationally driven drug discovery in the rational search for novel bioactive compounds in modern medicinal chemistry.

CADD methodologies can be divided in structure- and ligand-based drug design (SBDD and LBDD).Protein three-dimensional (3D) structures are the basis for structure-based drug design. Once the 3D structure of a protein

target is available, the SBDD is usually elected to develop therapeutic agents and to understand the mode of action, as well as,the metabolic process. SBDD embraces several methodologies, namely structure-based virtual screening (SBVS), de novodesign or pharmacophore generation. A rational SBDD approach is based on the deep understanding of the key interactionsbetween small molecules (or proteins) and their targets, applying techniques, such as molecular docking and moleculardynamics (Scheme 1).

Broadly used for SBVS, molecular docking methods explore the ligand conformations adopted within the binding sites ofmacromolecular targets (search algorithm) and estimate the ligand–receptor affinities (scoring functions). Molecular dockingis nowadays one of the elected methods to use in SBVS with a huge rate of success. However, molecular docking works onrigid structures forgetting structural flexibility and structural rearrangements. Although this is a generally well-accepted,affordable and successful method, docking algorithms are far from considering dynamic effects as a routine. To overcomemolecular docking drawbacks, molecular dynamics has been used to access structural flexibility of entire ligand-receptor sys-tem. This methodology provides a detailed insight into conformational changes through protein functions. It can be used toeffectively understand macromolecular structure-to-function relationships, to describe ligand–receptor interactions, membraneinteractions network and ensembles can be analyzed to derive thermodynamic properties of the system, like entropy (or freeenergy). Additionally, protein function and allostery can also be better understood if dynamic properties are considered. More-over, the explicit solvent representation is important to address the solvation effects of real solvent (like the hydrophobic

Structure - Funtion

Receptors

Ligand docking

De novo design

Pharmacophore modeling

Small-molecule immunesystem modulators

MD

X-ray structuresCADD

Homology modeledstructures

SCHEME 1 CADD in drug discovery

2 of 27 ACÚRCIO ET AL.

effect). Fortunately, molecular dynamics simulations, considered some years ago, as unaffordable strategies for SBDD, arenow becoming routine computational tools for drug discovery, being used in chemical biology and medicinal chemistry pro-grams with extraordinary potential.

SBDD methods are critical in early stages of the drug discovery process. However, lack of knowledge about the 3D struc-ture could limit the application of this methodology. Fortunately, it is possible to obtain 3D structures not only from experi-mental methods (X-ray or NMR), but also using computational sources, for example, quantum mechanics (computationallyvery demanding), and homology modeling. To that end, software was developed to generate 3D structures models fromhomologous proteins for which, for example, at least one X-ray structure is available. Additionally, LBDD methods can beused in the drug discovery process when ligands active against the target are known.

3D pharmacophore (models) generation is one of the most important methods in LBDD. 3D pharmacophores are abstractmodels, which creation heavily relies on the identification of bioactive molecular geometries against a target. Pharmacophoremodels can be also constructed following ligand- or structure-based designs, or combining both analyses. However, thisreview will only explore ligand-based pharmacophores.

Pharmacophore models represent, in space, the 3D-arrangement of the key features that characterize the chemical function-alities that make a molecule active toward its target. It includes the hydrogen bond acceptors, hydrogen bond donors, hydro-phobic regions, positively and negatively ionizable groups, and the metal coordinating regions.

Pharmacophore models are commonly used in ligand-based virtual screening (LBVS) campaigns, parallel screening, targetfishing, and for structure–activity relationships (SAR)-studies. In LBVS, the pharmacophore models are used as a filter. Afterpharmacophore model assessment, it is used to screen a database of compounds. All the molecules that fit “all” the features ofthe 3D model are retained as potentially active (hit molecules) against the target. This method is usually very efficient and hasa reduced computation time, when compared to molecular docking, and when compared particularly with molecular dynamics.Since the pharmacophore features represent chemical functionalities, not exactly chemical groups, the pharmacophore modelsare also excellent tools for scaffold hopping.

Nowadays, the synergistic effects between LBDD and SBDD approaches in drug discovery virtual screening workflowsallow to fully exploit their advantages, while compensating for some of the intrinsic limitations of each methodology.

The clinical importance of the negative immune checkpoints cytotoxic T-lymphocyte–associated antigen 4 (CTLA-4) andprogrammed cell death protein 1 (PD-1) has become clear, being several of them already approved for the treatment of distincttumor types.4–6 In recent years, many other pathways in the tumor immune microenvironment are emerging as important con-tributors to tumor immune evasion and/or activation of a specific anti-cancer immunity, aiming to overcome low responserates and resistance often associated to immune checkpoint antibodies in the clinic. The stimulatory immune checkpoint4-1BB, and the immune modulators Indoleamine 2,3-dioxygenase1 (IDO1), Adenosine A2A receptor (A2AR) and Stimulatorof interferon genes (STING) from the immune-suppressive tumor microenvironment constitute particularly attracting new tar-gets for the rational design of novel bioactive molecules. Accordingly, structural studies exploring these immune modulatorshave significantly advanced. However, besides being a crucial information to attain the design of effective but also specificmolecules with favorable pharmacokinetic properties, this information is still scarce and mostly dispersed.

In this overview, we highlight the computational methods applied to the structural elucidation and to the discovery anddevelopment of new small-molecule modulators of host immune system. The use of structure and ligand-based virtual screen-ing processes, molecular docking, homology modeling, and molecular dynamics simulations methodologies essential for suc-cessful drug discovery campaigns focusing small-molecule immune system modulators will be addressed. Additionalimportant tools and information related to the selected immune checkpoints and other tumor microenvironment components,like IDO or toll-like receptors, will be addressed. It includes their structural availability and properties, binding regions, struc-tural flexibility, main binding pocket interactions, and dynamics. Overall, the information enclosed in this review is a step for-ward to the use of CADD to rationally develop novel safe and effective anti-cancer immunotherapies.

2 | ADENOSINE A2A RECEPTOR

2.1 | A2AR in cancer

Adenosine signaling within inflammatory setting serves to dampen immunologic response and protect tissues from associatedinjury. The extracellular adenosine levels are commonly very low. However, tissue breakdown and hypoxia (common toinflammatory and tumor microenvironments) generate high concentrations of extracellular adenosine.7,8

Among the four G-protein-coupled (GCP) adenosine receptors (A1, A2A, A2B, and A3), the adenosine A2A is themost expressed in many of the immune cell subsets. The high levels of adenosine are related to increase in

ACÚRCIO ET AL. 3 of 27

CD73 (ecto-50-nucleotidase) and CD39 (nucleoside triphosphate dephosphorylase) enzymes, which are responsible forcatabolizing ATP into adenosine.

The inhibitory effect of adenosine on immune cells expressing adenosine A2AR revealed its interference on anti-tumorimmunity suggesting that the antagonism of A2AR can be an effective approach in cancer immunotherapy. The importance ofadenosinergic signaling in mediating negative feedback loops of immune responses and the effect of A2AR blockade onenhancing immunologic response has been reported. High concentrations of adenosine in tumor microenvironment lead to thestimulation of the A2A receptor, providing an immunosuppressive signal that inhibits the activity of T cells (proliferation, cyto-kine production, cytotoxicity), Natural Killer (NK) cells (cytotoxicity), NKT cells (cytokine production, CD40L up-regula-tion), macrophages/dendritic cells (DCs) (antigen presentation, cytokine production), and neutrophils (oxidative burst).9–11

Furthermore, adenosine recruits the immunoregulatory activity of regulatory T (Treg) cells. FoxP3, a key transcriptional factorfor the immunosuppressive activity of Treg cells, is induced by A2AR stimulation.12 T cell activation in the presence ofA2AR stimulation largely increases CD4+ FoxP3+ cells (Han et al13; Akio Ohta et al14).

The study reported by Ohta15 showed the spontaneous regression of the inoculated tumor in A2AR-deficient mice, whilethe A2AR antagonists were also beneficial in tumor-bearing wild-type animals. Depletion of T cells and NK cells impaired theretardation of tumor growth by the A2AR antagonists, pointing the improvement of antitumor cellular immune responses.16,17

In addition, blocking the adenosine-generating pathway CD39/CD73 also induced the potent regression of breast cancer, colo-rectal cancer, and melanoma in murine models.18–20

Moreover, the A2AR has been explored as an attractive drug target in the central nervous system against the Parkinson'sdisease (PD).21–24 Currently, several antagonists of this component of the immune microenvironment designed as potentialnew PD targeted therapies are being redirected toward cancer and are under clinical development (NCT03207867,NCT02403193, NCT03454451, NCT02655822).

2.2 | A2AR: Structural analyses

A2AR is a membrane receptor composed by 412 amino acids (aa) in length. The receptor structure shows a seven transmem-brane helix architecture similar to that of the rhodopsin and adrenergic receptor structures, but with shifts in the positions andorientations of the helices, and a markedly different structure of the extracellular loops (Figure 1a). Furthermore, a small alphahelix is located in the intracellular side close to the membrane. The residues constituting the transmembrane α helices areGly5-Trp32 (helix I), Thr41-Ser67 (helix II), His75-Arg107 (helix III), Thr119-Leu140 (helix IV), Asn175-Ala204 (helix V),Arg222-Phe258 (helix VI), and Leu269-Arg291 (helix VII) (33). The small helix is stabilized by the interactions with helix Iand comprises Arg296-Leu308 (helix VIII). The residues defining intracellular loops are the Leu33-Val40, Ile108-Gly118,and Leu208-Ala221. The residues defining the three extracellular loops are the Thr68-Cys74, Leu141-Met174, andCys259-Trp268.

The interest on A2AR as drug target led to several studies exploring the receptor structure, as well as, the development ofnew antagonists. Before X-ray structures, homology modeling studies to predict A2AR 3D structure and ligand binding werereported by the GPCR modeling community.25,26 Starting from the aa sequence of the receptor and a two-dimensional struc-ture of the ligand, 206 models were submitted by 29 different groups. Among the distinct predicted structures, the averagenumber of correct ligand–protein contacts was four, highlighting the large complexity of GPCR modeling due to loop model-ing refinement, accurate modeling of the structurally divergent regions (such as the extracellular loops that form defined archi-tectures), and disulfide bond formation affecting helix residue registry and helical shifts.

The first crystallographic structure of A2AR was resolved by Jaakola et al.27 (PDBID: 3EML) with 2.6 Å resolution. Thethermostabilized receptor was crystallized with the selective high-affinity antagonist 4-(2-[7-amino-2-(2-furyl)-[1,2,4]triazolo-[2,3-a][1,3,5]triazin-5-ylamino]ethyl)-phenol 5 (ZM241385)28 (Figure 1b). Comparing the GPCR structures, the A2AR pre-sented an organization of the extracellular loops markedly different from the other adenosine receptors and bovine and squidrhodopsins.29–31 Additionally, different ligand binding and divergence in the helical positions and orientations were observed.

The X-ray position of ZM241385 was substantially different from the one suggested by molecular modeling studies.32,33

ZM241385 sits almost perpendicularly to the membrane plane. The bicyclic triazolotriazine core of the ZM241385 interactswith Phe168 via π-stacking, Ile274 via hydrophobic interaction and Asn253 by hydrogen-bonding interaction. Adjacent toPhe168, Glu169 interacts with the exocyclic amino group linked to the bicyclic core of ZM241385 (Figure 1c).

Additionally, Hino et al.34 also reported the X-ray structure of A2AR with Fab2838, in the presence of ZM241385 at a res-olution of 2.7 Å (PDBID: 3VGA). Binding assays revealed that Fab2838 completely inhibited the binding of the agonist butdid not affect the binding of the antagonist ZM241385. These findings suggest that the Fab2838 induced an inactive confor-mation, where the agonist could not bind. In addition, this structure also exhibited the complete intracellular loop, but not theionic lock of the A2AR structure, described by Doré et al.35 Instead of the salt bridge described, this structure showed a modi-fied “ionic lock,” where Glu228 (helix VI) and Arg102 (helix III) interact via a water molecule.


Further high-resolution structures of A2AR:ZM241385 complex were resolved (PDBID: 4EIY and 5IU4).36,37 The studyof Liu et al.36 focused the complex water networks across the A2AR. From 177 waters, 57 occupy the interior of the transmem-brane helix VII bundle, forming three bulky water molecule clusters extending from the ligand-binding site of ZM241385(Figure 1d). Focusing the central water cluster, there is a Na+ ion and 10 water molecules that completely fill the cavity(Figure 1d). Allosteric effects of the Na+ ion on ligand binding, stability, and crystallization of A2AR were reported in previ-ous studies. In this structure, it was determined the precise location of the Na+ ion, the resolution of the complete network ofwater molecules around the Na+ ion, and the conformations of all residues involved in direct or water-mediated coordination.The Na+ ion played an important role in the receptor activation. Upon activation, the inward movement of the helix VII in thisregion collapsed the pocket and shifted it toward the helix VI.

The latest structure reported by Segala et al.37 focused the influence of the salt bridge formed by Glu169 in the secondextracellular loop of the receptor, between the transmembrane helices IV and V, and by His264 in the third extracellular loop,between the helices VI and VII. The high-resolution structures of ZM241385 and its derivatives highlighted the differences inthe interactions between the ligands and the Glu169-His264 salt bridge, which contributed to the variation in the dissociationkinetics. Long residence time ligands stabilized the Glu-His ionic interaction, while fast off-rate derivatives destabilized thissalt bridge.

The following X-ray structures are focused on the agonist bound form of A2AR. These structures highlight the structuraldifferences between the active and inactive forms of the adenosine receptor. Xu et al.38 resolved the first X-ray structure of the

FIGURE 1 A2AR structure. (a) Structure of A2AR in rainbow coloration (yellow to blue). The visible extracellular and intracellular loops are labeled. Thedisordered portion of ECL2 is not completed. (b) Structure of A2A-ZM241385 in rainbow coloration (yellow to blue) from the N terminus to the C terminus,ZM241385 is represented as a stick model (yellow), (c) close-up view of A2AR pocket of ZM241385 (yellow) and details of the hydrogen bonds network(yellow dash line) together with π–π stacking (green dash line). (d) Distribution of ordered waters in A2AR. Buried water molecules (red spheres) in the A2AR-ZM241385 structure form an almost-continuous water channel containing three major water clusters. Close-up view of the central cluster, which includeswaters and a sodium ion (cyan sphere)


A2AR agonist-bound UK-432097 (13) (PDBID: 3QAK, 2.7 Å) (Figure 2a). Comparing agonist- and antagonist-bound A2ARstructures, the former displays an outward tilt and rotation of the helix VI, a movement of the helix V and an axial shift of thehelix III. Additionally, a movement of the helix VII and a shift of extracellular loop 3 were also observed.

The binding pocket is similar to that described for ZM241385 (5), as well as, to the extensive ligand-receptor interactionwith slightly differences. UK-432097 and A2AR establish 11 hydrogen bonds, an aromatic stacking interaction and severalnonpolar (van der Waals) interactions. The bicyclic adenine core of UK-432097 (13) is aligned with the triazolotriazine coreof the ZM241385, when the two complex structures are superimposed (Figure 2a and b1). The interactions of both scaffoldsare conserved, including the key π-stacking interaction with Phe168, as well as, the nonpolar interaction with Ile274, and thetwo hydrogen bonds with Asn253.

The ribose ring is the key feature that distinguishes the AR agonists from the corresponding antagonists. This moiety of theligand is deeply placed into a predominantly hydrophilic region of the binding pocket, where 20- and 30-OH groups establish

FIGURE 2 A2AR-agonist bound structures. (a) Ligand UK-432097 and its binding cavity. Overall view of the ligand-binding cavity. The transmembranepart of A2AR is shown as ribbon and colored gray. Ligand UK-432097 is shown as yellow sticks. Molecular interactions within the ligand-binding cavity forUK-432097. Interacting residues are shown as sticks. Hydrogen bonds between A2AR and ligand are shown as yellow dashed lines. (b1) Superposition ofA2AR bound to the inverse agonist ZM241385 (orange sticks) and the agonist UK-432097 (yellow sticks). (b2) A2AR bound to NECA (salmon) andadenosine (pink). (c) Ligand-binding mode of the selective agonist CGS21680. Structure of the extracellular portion of A2AR bound to the selective agonistCGS21680. Elements of the structure are depicted as follows: A2AR illustration (gray ribbons); specific amino acid side chains, sticks; CGS21680, dark bluesticks; hydrogen bonds, yellow dashed line; π−π stacking interaction green dashed line. (d) Comparison of the ligand-binding site of the human A2AR boundto various ligands (PDB IDs are shown in parentheses): NECA (2YDV), CGS21680, UK432097 (3QAK), and adenosine (2YDO)


hydrogen bonds with the His278. The 3’-OH is further stabilized by a hydrogen-bond with the Ser277. Besides these polar inter-actions, the ribose part of the ligand has close contacts with Val84, Leu85, Trp246, Met177, and Leu249 (Figure 2a).

A subsequent study of an agonist bound to the A2AR was reported by Lebon et al.39 The structure of the thermostabilizedA2AR bound to its endogenous agonist adenosine (10) (PDBID: 2YDO) and to the high-affinity agonist NECA (11) (PDBID:2YDV) was resolved with 3.0 and 2.6 Å (Figure 2b2).

Comparing Xu and Lebon structures, the first concluded that the structure of UK432097-bound A2AR was in an activestate, whereas the latter resolved that the NECA- and the adenosine-bound structures were best defined as representing anintermediate state between the active and inactive states, based on a different thermostabilized A2AR structure. The structuralanalyses of the two structures revealed that there is insufficient space for the C terminus of the G protein to bind. However,the binding pocket interactions indicate that the structure is a good representation of the agonist-bound binding pocket, sinceit is not so different from the previous structure.

Lebon et al.40 also described the crystal structure of the thermostabilized A2AR in an active-like conformation bound tothe selective agonist 2-[p-(2-carboxyethyl)phenylethyl-amino]-5’-N-ethylcarboxamido adenosine 12 (CGS21680) (PDBID:4UG2 and 4UHR). CGS21680 (12) is an analog of the natural agonist adenosine, with a (2-carboxyethyl)phenylethylaminogroup on the C2 position of adenine and an N-ethylcarboxyamido group on C5’ of the ribose. Comparing all agonist boundstructures, adenosine moiety of the ligands binds to the receptor in an identical way, and the key interactions with Phe168,Glu169, Ser277, and His278 were retained. The key selectivity feature of CGS21680 is the (2-carboxyethyl)phenylethylaminogroup that binds to the extracellular regions. These are the most divergent in sequence between adenosine receptor subtypes.CGS21680 forms van der Waals contacts with side chains of the amino acid residues Glu169, His264, Leu267, and Ile274,and the amine group forms a hydrogen bond with the side chain of the Ser67 (Figure 2c). Among these residues, only theIle274 is absolutely conserved across the human adenosine receptor subfamily.

Recently, White et al.41 reported the crystal structures of the agonist complexes for two variants in the sodium coordina-tion cavity of A2AR, D52N and S91A (PDBID: 5Wf5 and 5WF6). Mutation of these key residues in the sodium-binding motifcaused a striking effect on G-protein signaling. The structures presented an active-like conformation; however, the variantsshowed key hydrogen bonding changes in the activation motif. Structural elucidation showed the collapse of sodium pocketthat facilitates stronger helix–helix interactions at the intersection between the helices II, III, VI, and VII. The conformationsof the residues in the ligand-binding pocket and binding pose of the bound agonist were identical to nonmutated structure.Additionally, ligand-binding affinities of the antagonist ZM241385 and the agonist UK432097 were nearly identical. Overall,introducing a single-point mutation at this location could provide a rapid and straightforward mean for improving the stabilityand expression of many GPCRs.

The resolution of A2AR antagonist-bound structures takes the main role and intensive work was devoted into the resolutionof several crystallographic structures. Doré et al.35 reported the first A2AR structure bound to xanthine (PDBID: 3REY) andcaffeine (1) (PDBID: 3RFM), providing new insights into the binding mode of a different class of ligands. Furthermore, thiswas the first structure exhibiting the ionic lock and the complete intracellular loop. In the XAC and caffeine-bound structures,key interactions with Phe168 and Asn253 were also observed. In addition, the induced fit exhibited in the three ligands evi-denced the flexibility of the A2AR across the transmembrane helices I and VII (Figure 3).

The first X-ray structures from optimized hit series of A2AR antagonists were reported by Congreve et al.42 (PDBID:3UZA and 3UZC). The derivatives of 1,3,5- triazine were identified as hit molecules and optimized to yield potent andselective A2AR antagonists. The hypothesis of 1,2,4-triazine isomers that also bind to the receptor, and the possibility of adeep access to the receptor pocket, a region normally occupied by the ribose group of the natural ligand adenosine, wasstudied by Zhukov et al. (2012)43 using a homology model and was confirmed by Congreve et al.42 using X-ray. The1,2,4-triazine derivatives 26 and 27 were co-crystallized with A2AR according to previously solved structures. Structurerevealed two critical donor and acceptor hydrogen-bond interactions with the side chain of Asn253 and the amino-triazinecore. In addition, a π–π stacking with the Phe168 and a hydrophobic interaction with the side chain of the Met270 werereported on the opposite side. The phenyl substituent of the triazine core occupies a hydrophobic pocket deeper inside thereceptor flanked by Leu84, Leu85, Met177, Asn185, Trp246, Leu249, and His250. The second substituent, dimethyl-pyri-dine, occupies the ribose-binding pocket defined by His278 and Ser277. The study previously reported by Zhukovet al. (2012), was in excellent agreement with the crystal structures, highlighting the significant interactions with the Ile66,Leu85, Asn181, and Asn253, and confirming the initial deep placement of the ligand in the ribose sugar pocket of the A2A

receptor (Figure 3b and c).More recently, a crystallographic structure of A2A receptor bound to compound 8, a novel A2AR/N-methyl D-aspartate

receptor subtype 2B dual antagonist, was resolved at 3.5 Å resolution (PDBID: 5UIG) by Sun et al.44 The structural anal-ysis provided insights into the binding mode of 8 and revealed a potential allosteric pocket that was not observed in thepreviously reported A2AR structures (Figure 4a). The compound 8 contains three rings: methylphenyl, amino-triazole, and


orthomethoxyphenyl. The binding pocket revealed that the methylphenyl and aminotriazole rings occupy positions similarto the furan ring and triazolo-triazine core of ZM241385, respectively; in the deep pocket, extensive hydrophobic interac-tions and hydrogen bonds with surrounding residues were established. Part of the interactions reported were common tothe other A2AR ligands: the hydrogen bond with the Asn253, the aromatic stacking interaction with the Phe168 and thehydrogen bond with Glu169 involving the aminotriazole ring. In addition, the methoxyphenyl ring of 8 extended to anadjacent and unobserved binding pocket formed by Tyr9, Ala63, Ile66, Ser67, Leu267, Met270, Tyr271, and Ile274,toward the extracellular aspect of helices I, II, and VII establishing key hydrophobic interactions (Figure 4a). The pocketresidues and the interactions established were critical for the higher selectivity of 8 over A1 receptor (A1R).The variability of the residues forming this pocket suggested that it may be a target for the development of novel selectiveA2AR antagonists.

The lack in structural information of other subtypes of ARs has hindered the development of selective agents. Chenget al.45 have solved the X-ray structure of the thermostabilized A1R with the high-affinity and A1R-selective xanthinePSB36 at 3.3 Å resolution (PDBID: 5N2S). This study report as well the crystallographic structures of A2AR with caffeine,theophylline, and PSB36 at 2.1, 2.0, and 2.8 Å, respectively (PDBID: 5MZP, 5MZJ and 5N2R). Comparing the resolvedstructures, differences on the binding mode of the ligands in the two receptors were assessed. Overall, the structure of A1R is

FIGURE 3 A2AR-inverse agonists bound structures. (a) Comparison of ligand positions in the structure-binding site cavity. The structure of A2AR isdepicted as a gray surface. The ligands are represented as sticks, ZM241385 (dark blue), XAC (green), and caffeine (orange), oxygen atoms are red, andnitrogen atoms blue. (b) and (c) 1,2,4-triazine adenosine A2A antagonists. Compounds 26 and 27 are illustrated bound to the pocket of the receptor and theresidues lining the pocket that interact with the ligands are labeled. Yellow dashed lines highlight the key hydrogen bonding to Asn253 of the scaffold. TMhelices and visible extracellular regions are depicted in gray cartoon. Ligands are represented as stick models; residues involved in ligand binding are labeledand represented as green sticks


very similar to that of the A2AR, as expected due to the sequence similarity, with the major differences occurring on the extra-cellular loops of the receptor (Figure 4c).

Comparing the structures with xanthine-based ligands, there are several differences in the way that the xanthine cores wererecognized and in the features of the protein surrounding the ligand-binding site. The preferential selectivity of the PSB36 forA1R over A2AR was related to the longer alkyl substituents at N1 of the xanthine core, since A2AR is unable to accommodatethis alkyl chain. In addition, the selectivity for the A1R of ligands with bulky substituents at C8 is due to a steric clash withMet270 in the A2AR (Thr in A1R) (Figure 4d). The detailed view of the interactions led to the understanding of ligand selec-tivity that can now be used in the ongoing search for selective AR antagonists (Figure 5).

The mechanism of activation of the receptor that allows the G-protein coupling and is related to G-protein selectivity isnot fully understood. Carpenter et al.46 reported the structure of A2AR bound to an engineered G protein, mini-Gs (5G53). Thetransition of the receptor from an agonist-bound active-intermediate state to an active G-protein bound state was characterizedby a 14 Å shift of the cytoplasmic end of the transmembrane helix VI, as well as slight changes in the positions of the cyto-plasmic ends of the transmembrane helices V and VII. There were no substantial differences in the extracellular half of thereceptor around the ligand-binding pocket.

FIGURE 4 A2AR: Inverse agonist complex A1 and A2A receptors. (a) Binding interactions between the A2AR and compound 8. The side chains of residuesare shown in green sticks. Dashed lines depict hydrogen bond interactions. The π−π stacking interaction of the aminotriazole ring with the Phe168 is depictedwith dashed green line. The hydrogen bonding interactions with the side chains of Asn253 and Glu169 are depicted with dashed yellow lines. (b) Binding siteof A2A with theophylline. The side chains of residues are shown in green sticks. The hydrogen bonding interactions with the side chains of Asn253 andHis278 are depicted with dashed yellow lines. (c) Superposition of backbone ribbons of the A1 complexed with PSB36, in blue, and the A2AR structure incomplex with ZM241385, in gray. (d) Comparison of binding modes of PSB36 in A1 and A2AR. Superposition of the structures of the A1R (blue ribbon) withPSB36 (salmon), and the A2AR (gray ribbon) with PSB36 (gray)


2.3 | Structure-based drug design for A2AR antagonists

Initial drug discovery strategies targeting A2AR relied on the use of known natural ligands or screening assay hits as startingpoints for optimization.47–49 Adenosine50 and xanthine51 derivatives were the most used scaffolds for the design of ARligands. Several other reported antagonists for the AR were based on ligand-based approaches. However, ligand-basedapproaches without receptor crystallographic structure are limited to relatively close analogues of known ligands and requiredeep knowledge of the ligand SAR.

After the release of A2AR X-ray structures, several structure-based CADD methodologies were reported, yielding inspiringstarting points for drug discovery programs. Katritch et al.52 and Carlsson et al.,53 in different studies, screened millions ofcommercially available compounds using the first A2A structure (PDB 3EML). From these studies, several compounds withsubmicromolar affinity were identified. The compounds reported by Carlsson et al.53 presented the highest affinity for the A2A

receptor. However, the compounds achieved by Katritch et al.52 were found to bind also to the A1R. Other scaffolds were dis-closed as A2A antagonists, including chromenes, which were identified by an in sillico screening of academic compounds.54

FIGURE 5 A2AR inverse-agonists and agonists


Other virtual screening campaigns were reported using receptor structures in combination with ligand-based pharmaco-phores.24,55 A similar approach using a X-ray structure in combination with known activity data for a range of ligands wasreported by Van Westen et al.56 Over thousands of compounds screened, very high affinity adenosine ligands were identified,including the 2-amino-1,3,5-triazine. In addition, a virtual screening campaign was reported to achieve agonists in thereceptor.57–59 In Figure 6 it is possible to observe the scaffolds achieved among the variety of virtual screening campaignsherein highlighted.

Other in silico techniques have also been used to understand the recognition, potency, and selectivity of A2AR ligands.Molecular dynamics was used to assess the stability of ligand−receptor complexes with agonists and antagonists in a lipidbilayer and water environments, and to predict bioactive conformations.37,60,61

The A2AR and its ligands have also been used to calculate the positions and energies of water networks within thereceptor-binding site.57,62 In virtual screening campaigns, the highly structured waters were retained in the binding site duringthe modeling, and this may have contributed to the success of the several of these campaigns.

FIGURE 6 A2AR ligands identified by in silico approaches


3 | STIMULATOR OF INTERFERON GENES

3.1 | STING in cancer

STING, also known as TMEM173, MITA or ERIS is a transmembrane protein anchored in the endoplasmic reticulum.63,64

STING is an important DNA sensor and also an adaptor for other DNA sensors. This pathway is activated in the presence ofcytosolic DNA, which is sensed by cyclic-GMP-AMP synthase (cGAS), an enzyme that generates cyclic GMP-AMP(cGAMP), the endogenous ligand of STING.64,65 This results in the auto-phosphorylation of tank-binding kinase 1 (TBK1)and phosphorylation of the transcription factor interferon regulatory factor 3 (IRF3), which activates the transcription of type Iinterferon (IFN) genes. Although the activation of STING signaling pathway triggers the production of type I IFNs to resist topathogen invasion, additional mechanisms (transcriptional, epigenetic regulation, or posttranslational modifications) exist toavoid aberrant inflammatory responses.

The role of STING in cancer immunotherapy is complex and is not well understood. Studies suggest that tumor cells aretaken up by phagocytes, such as DCs, and a fraction of tumor DNA may enter the cytoplasm to activate the cGAS–STINGpathway.66,67 Consistent with this suggestion, stimulation of STING following the intratumoural administration of cGAMP,or its analogs, inhibited tumor growth in immune competent mice. However, some other studies suggest that the STING acti-vation may contribute to tumor growth and metastasis by inducing a suppressive tumor microenvironment.68,69 Due toSTING's central role in immune regulation, it is of outmost importance to define the structural basis and biological impact ofligand binding to STING.

3.2 | STING receptor: Structural analyses

STING is transmembrane protein with 379 amino acids in length. The protein is composed by a N-terminal transmembranedomain (aa 1–154), a central globular domain (aa 155–341) and C-terminal tail (CTT) (aa 342–379). Within the cytosolicdomain, studies showed that CTT interacts and activates TBK1 and IRF3 in vitro.70 The region comprising CTT was sug-gested as having the critical role in the function of STING, and for which several X-ray structures have been reported(Figure 7a).

The first crystallographic structures resolved focused the apo-STING form and the c-di-GMP bound STING complex.71–74

Overall, the structures of apo-STING agreed with each other showing STING as a homodimer resembling a pair of wings, or

FIGURE 7 STING dimer. (a) Cartoon representation of STING dimer. One protomer is colored in orange ribbon. The other protomer is colored in teal. Theinteracting α1 and α3 are labeled for both protomers. Detailed interaction at the dimer interface. Interfacial residues are shown as sticks. (b) Surfacerepresentation of the STING dimer in a side view, showing the deep crevice across the dimer interface, and the electrostatic surface of the STING dimershown in a top view. One molecule is colored in orange and the other in teal, as in (a)


a butterfly with a deep cleft between the two dimers (PDBID: 4EF5, 4EMU, 4F5E, and 4F5W) (Figure 7a). The dimer inter-face of STING was dominated by hydrophobic interactions with no salt bridges participating in the dimerization. The STINGhomodimer interact with each other involving the residues Val155, Leu157, Trp161, and Ile165 in the helix α1, and theAla270, Met271, and Ala277 near the linker region between the helices α1 and α3. In addition, there were two direct hydrogenbonds between the side chain phenolic groups of the Tyr164 and the Tyr274 (Figure 7).

In the ligand-bound STING complex, c-diGMP, was anchored to the protein in the deep cleft between the two wings, byseveral hydrogen bond and hydrophobic interactions (PDBID: 4EF4 4EMT, 4F5D, 4F5Y, and 4F9G) (Figure 8a). Four hydro-gen bonds existed between the ribosephosphate ring and STING (Figure 8b). In addition, it was observed that binding of c-di-GMP to STING did not induce large conformational changes (Figure 8c). The binding site was located at the bottom of thedimer interface, where c-di-GMP adopted a V-shaped conformation. The contacts with STING dimer occurred through stack-ing of both guanine bases against the phenolic rings of Tyr167. In addition, the side chain of Asn242 formed a network ofhydrogen bonds with Glu260, Tyr167, and Arg178. These interactions are important for positioning Ser241 and Tyr167 tointeract with the guanine base. The 20 hydroxyl of the ribose interacted directly with the hydroxyl of the Thr263. Other STINGresidues were also critical for c-di-GMP binding (Figure 8b). For example, Pro264 was critical for the formation of the bindingsurface because it helped the positions Thr263 and Thr267 to interact with c-di-GMP. Gly166 was also critical for c-di-GMPbinding, allowing the nucleotide to approach its binding surface closely and thus permitting the stacking of the guanine ringon the phenolic ring of the Tyr167.

Among the five STING: c-di-GMP crystal structures, three were of a rare His232 STING isoform that is unresponsive tocyclic dinucleotides.72,74,75 The two other crystal structures of STING had an arginine residue at position 232.71,72 STINGwith Arg232 is responsive to cyclic dinucleotides. Residue 232 lies in the loop region between β-sheet 2 (β2) and β-sheet3 (β3) that is predicted to cover the binding pocket. In the His232 structures, these loops were considerably disordered. In theArg232 structures, the β2-β3 loop was modeled, although the two structures differed in the position of Arg232. Interestingly,

FIGURE 8 c-di-GMP:STING complex. (a) c-di-GMP binds into the through at the dimer interface. The complex structure of STING CTT with c-di-GMP. c-di-GMP is shown as yellow sticks. The STING dimer is shown in an orientation similar to that showed in Figure 6a. (b) Surface and close-up view of specificrecognition of c-di-GMP by STING and details of hydrogen bonds. The interactions between ribose-phosphate of GMP and waters, as well as c-di-GMP andT263 from STING monomer. Hydrogen bonds between c-di-GMP and waters are shown as yellow dashed lines and red spheres, respectively. π–π interactionof c-di-GMP and Y167 is shown as green dashed line. Residues from STING monomer a that interact with c-di-GMP are shown as orange sticks. C-di-GMPis shown as yellow sticks. (c) Superposition of the STING/c-di-GMP complex with the free STING dimer. The STING protomers in the complex are coloredin light orange and light blue, while those in the free STING are in dark orange and teal


the former STING–c-di-GMP structure differs the most from the other four structures. It is suggested that the loop between β2and β3 is highly flexible and that the binding of c-di-GMP induces changes in the conformation of the critical arginine that areimportant for downstream signaling. However, the structures do not provide insight into why the arginine in the β2-β3 loop isrequired for signaling only in response to cyclic dinucleotides, but not in response to DNA.

The endogenous ligand of STING is c-GMP-AMP (cGAMP). cGAMP is a hybrid cyclic dinucleotide that contains aguanosine unit and an adenosine unit joined by two phosphates in a macrocycle (Figure 9). Specifically, cGAS catalyzesthe synthesis of cGAMP from ATP and GTP in the presence of DNA. cGAMP then acts as a second messenger that bindsto and activates STING. Although the exact nature of cGAMP is not determined, it results from phosphodiester linkagesbetween GMP and AMP. cGAMP molecule contains unprecedented mixed phosphodiester bonds resulting in different iso-mers that have impact in STING affinity.76–79 Its two phosphate groups connect the two nucleosides from the 20 and 50 posi-tions of the guanosine with the 30 and 50 positions of the adenosine. Represented as 2030-cGAMP, this asymmetric ligandbinds to the symmetric dimer of STING with an affinity significantly higher than its isomers 2020-cGAMP, 3020-cGAMP,and 3030-cGAMP.

Several X-ray structures of STING and cGAMP complex have been resolved (PDBID: 4KSY, 4LOH and 5BQX)pointing the mechanism behind STING activation, as well as, providing understanding on the different affinities ofcGAMP isomers. Zhang et al.79 reported the first X-ray structure with 1.9-Å resolution, being followed by Gao et al.78 at2.3 Å, and the Shi et al.80 at 2.0-Å resolution. Overall, the solved structures show a STING homodimer forming abutterfly-shaped dimer (Figure 10a). The cGAMP molecule sits at the two-fold axis, with the cyclic sugar-phosphatebackbone at the base and the purine rings pointing upward in a parallel alignment. Comparing with the apo-STING, thecomplex exhibited several differences in both the structure of the monomer and the arrangement of the dimer, creating adeeper pocket between the two protomers to hold cGAMP (Figure 10c). The cGAMP binding site was covered by a lidof four-stranded antiparallel β-sheet and the connecting loops formed by the residues 219–249, from each of the two pro-tomers. The interdomain interactions within the lid involved several polar contacts, Lys224, Gly234, Lys236, Asp237,and Tyr245 (Figure 10b).

Concerning the cGAMP isomers, crystal structures of STING complexed with 2030-cGAMP or 3020- cGAMP were nearlyidentical. Both cGAMP molecules sited at the same position of the binding pocket, forming the same interactions, suggestingsimilar affinity for STING (Figure 10d). However, several studies reported different affinities over the different cGAMP iso-mers, in which 203’-cGAMP revealed to be a potent inducer of IFN-β due to a higher affinity (Kd 4.59 nM) for STING. Shiet al.80 showed that 2030-cGAMP, but not its linkage isomers, adopted an organized free-ligand conformation that resemblesthe STING-bound conformation and pays low energy costs in changing into the active conformation to bind to STING. Cur-rent structural studies typically focus on protein–ligand interactions. However, in this case, entropic factors of the free ligandsprovided large difference in binding affinities, even with the same protein–ligand interactions.

Comparing with the c-di-GMP bound structure, cGAMP sits deeper in the STING dimeric interface. In addition, the twowings of the butterfly are closer to each other in the STING:cGAMP structure due to the closed arrangement of the twoSTING protomers. The binding pocket showed that cGAMP established extensive polar and hydrophobic interactions. Thepurine base groups of cGAMP interact with four aromatic residues, Tyr 240 and Tyr167 of the two protomers. The free 30-

FIGURE 9 STING ligands. cGAMP is the endogenous ligand and contains unprecedented mixed phosphodiester bonds resulting in different isomers. Smallmolecule DMXAA (28) mSTING inhibitor


hydroxyl of GMP points to two Ser162 residues from the lower part of the pocket (Figure 10b). The guanine base directlyinteracts with the Glu260, Thr263, and Val239. These unique polar contacts explain why 203’-cGAMP is a specific and high-affinity ligand for STING.

Subsequent studies reported by Gao et al.78 focused the role of the xanthenone derivative 5,6-dimethylxanthenone-4-acetic acid 28 (DMXAA) (Figure 9) on both murine and human STING proteins (PDBID: 4LOL and 4QXO). DMXAAwas identified as a small molecule exhibiting immune modulatory activities through the induction of cytokines, also dis-rupting the tumor vascularization in multiple mouse models.81 However, DMXAA failed in human phase III trials.82 Later,it was demonstrated that DMXAA-IFN-β production by murine macrophages was dependent on STING, suggesting thatmurine STING was the protein targeted by DMXAA.83 Despite the high sequence identity between murine and human pro-teins (68% amino acid identity and 81% similarity) (Figure 11c),77 DMXAA activates murine STING, but has no effect onhuman STING.84,85 The resolution of both murine and human STING:DMXAA complexes led to structural analyses thatexposed the differences between the two complexes. The X-ray structure of murine STING:DMXAA complex (4LOL)revealed two small molecules bound within the cleft of mSTING forming a network of intermolecular hydrogen bondsinvolving Arg237, Ser161, Thr262, and Thr266. In addition, methyl groups formed intermolecular contacts with Lys169and Ile234 (Figure 11b). The binding pocket adopted a “closed” conformation. Structural analysis of the homologoushuman complex together with biophysical and cellular assays showed that the lid residue at position 230 (229 in mSTING)was the basis for the mSTING selectivity of DMXAA. Structural and functional results also displayed strategies to restorean efficient hSTING response to DMXAA, based on the binding-pocket S162A and Q266I mutations. The screen of small-molecule libraries using high-throughput approaches and mass-spectroscopy-based detection assays to identify potentialcandidates with scaffolds distinct from DMXAA that target hSTING were performed. However, novel DMXAA derivativestargeting hSTING were not yet reported.

FIGURE 10 cGAMP:STING complex. (a) Overall structure of STING CTT bound to the cGAS product. STING forms a dimer and is colored in orange andteal. (b) cGAMP in yellow sticks binds at a deeper pocket and drags the STING dimer closer to each other, compared to c-di-GMP-bound STING. The basegroups of cGAMP are roughly parallel to each other and to four aromatic rings of the Y residues. The bottom ribose ring and the upper purine base groups ofcGAMP are coordinated by extensive polar contacts. Hydrogen bonds between c-di-GMP and waters are shown as yellow dashed lines. π–π interaction of c-di-GMP and Y167 is shown as green dashed line. (c) Structural comparison of the apo- and cGAMP-bound forms of STING. The apo-STING dimer iscolored in dark orange and teal, and it is superimposed against cGAMP-bound STING dimer colored in light orange and light blue. Conformational changesin STING upon cGAMP binding are observed. (d) Superposition of 3020-cGAMP (yellow sticks) and 2030-cGAMP (dark blue sticks) bound to STING


The exact molecular mechanism underlying STING activation, and TBK1 and IRF3 recruitment are still a question indebate. In terms of the mechanism, STING contains the conserved motif, pLxIS (in which p represents the hydrophilic resi-due, x represents any residue, and S represents the phosphorylation site), that is phosphorylated by TBK1 or mediates therecruitment of IRF-3 to the signaling complexes.86 The induced proximity between TBK1 and IRF-3 results in IRF-3 phos-phorylation and activation.86 To elucidate the structural bases of IRF-3 recruitment by phosphorylated STING (pSTING),Zhao et al.87 resolved the crystal structure of STING complex with the IRF-3 C-terminal domain (PDBID: 5JEJ).

The structural data revealed the oligomerization of STING at the endoplasmic reticulum surface spanning two patchesof positively charged residues (Arg285, His288, His290, and Lys313, interacts with pSer366) and a large hydrophobicgroove. The pLxIS motif interacts with IRF-3 through electrostatic interactions, hydrogen bonds, and hydrophobic interac-tions (Figure 11d). Additionally, binding assay shows that Ser366 is critical for IRF-3 recruitment, since its mutation abol-ishes IRF-3 binding by STING. The Arg285 and Lys313 are also critical for Ser366 recognition. The hydrophobicinteractions between IRF-3 and the STING residues Pro361, Leu363, and I365, are also critical for IFN-β induction. Takentogether, the data revealed to this structure provided a critical insight into the mechanism of IRF-3 recruitment and activa-tion of the innate immunity.

The key role of STING in innate immune sensing of tumors led to the development of deliberate stimulation. Thus,STING agonists have emerged as novel candidate treatments in cancer immunotherapy. However, the lack of structure–function studies has limited the development of effective STING agonists. Herein, we summarized the so far dispersedinformation regarding the structure and function analyses reported for STING. These structure–function studies revealedthat cyclic-dinucleotides are the direct natural ligands of STING and exposed the lack of clinical effectiveness ofDMXAA. In addition, the understating of structural mechanisms behind IFN-β induction may guide the rational designof new STING agonists. This approach could expand the fraction of patients potentially benefitting from cancerimmunotherapies.

FIGURE 11 DMXAA:STING complex and STING activation. (a) The crystal structure of two molecules of DMXAA bound to murine STING.(b) Intermolecular contacts in the complex of DMXAA bound to murine STING. The two bound DMXAA molecules are shown in light green color, withindividual murine STING subunits in the symmetrical dimer shown in dark orange and teal. The intermolecular contacts to the polar and nonpolar edges ofthe DMXAA by the murine STING subunits are shown in two alternate views. (c) Superposition of the DMXAA-bound structure of murine STING (lightorange and light blue) and the DMXAA-bound structure of human STING (dark orange and teal). (d) The structural basis of IRF-3 recruitment by STING.Residues of the pLxIS motif are represented by orange sticks. Close-up view of the STING/IRF-3 complex. IRF-3 is shown by light blue ribbons


4 | INDOLEAMINE 2,3-DIOXYGENASE1

4.1 | IDO1 in cancer

Dioxygenase enzymes, as indoleamine 2,3-dioxygenase1 (IDO1), IDO2 and tryptophan dioxygenase (TDO) are immunomod-ulatory proteins with great potential as targets for cancer immunotherapy. However, most of the studies reported so far arefocused on IDO1 enzyme. Since IDO2 is lagging behind, only IDO1 features are herein discussed in detail.

T-cells require adequate tryptophan (Trp) levels for cell proliferation and survival. IDO1 is the enzyme that mediates Trpdepletion, resulting in T-cell tolerance and lack of effector function, therefore helping the tumor to evade host immunity (82).Pro-inflammatory mediators, such as endotoxin and interferon-γ (IFN-γ), induce IDO1 expression in antigen-presenting cells(APCs) and endothelial cells, which dampens the host immune response.88

Although not being an ICR in the classical sense, the overexpression of IDO1 plays an important role in downregulatingantitumor immunity.89 IDO1 is a monomeric heme enzyme with 403 amino acids folded in two distinct domains (Figure 12a).IDO1 interface is formed by a combination of hydrophobic interactions, salt bridges and hydrogen bonds that catabolize thebreakdown of Trp to kynurenine by incorporating molecular oxygen.90 Therefore, it could be expected an effective immuneresponse against tumors triggered by IDO1 inhibitors, through the depletion of Trp.88

Recently, a dramatic increase has been observed in the development of drugs targeting IDO1 enzyme, both in academiaand pharmaceutical companies. This is due to the intensive work on structural studies using crystallographic analysis, as wellas CADD following a structure-based IDO1 inhibitors design campaigns.91–93

4.2 | IDO1: Structural analyses

The first crystal structures disclosed for IDO1 were reported by Sugimoto et al.94 Among those, the 4-phenylimidazole 33(4PI) bound form of IDO1 (PDBID: 2D0T) was the structure reported with 3.4 Å resolution, while the cyanide bound form ofthis enzyme (PDBID: 2D0U) was disclosed with the highest resolution of 2.3 Å.

These crystal structures provided important structural insights into the receptor binding pockets.94 Authors identified twodistinct binding pockets. One pocket is in the distal heme site (pocket A) connected to a second pocket (pocket B), which istoward the entrance of the active site (Figure 12b). The pocket A includes the residues: Tyr126, Cys129, Val130, Phe163,Phe164, Ser167, Leu234, Gly262, Ser263, and Ala 264. The pocket B comprises fewer residues: Phe226, Arg231, Ser235,Ile354, and Leu384. Comparing cyanide and 4PI (31) bound form of IDO1, 4PI (31) is bound deeply inside the binding sitewith its phenyl ring occupying the hydrophobic pocket A, being the imidazole coordinated with heme iron (Figure 13a). Thepocket B is filled with buffer molecules (2-(N-cyclohexylamino) ethane sulfonic acid (CHES)) (Figure 13a). Active site flexi-bility of IDO1 is pointed out by different ways to access to the pocket A on cyanide and 4PI (33) IDO1 complexes.

FIGURE 12 Structure of IDO1 (PDBID: 2D0U). (a) Ribbon representation of the overall structure of human IDO1. The small (teal) and large (dark blue)domains connected by helices L, K, and N (green) and big loop (green) are represented. The 19 helices A-S are determined based on the primary sequence.(b) Surface and active site of IDO1 are represented. Pocket A (yellow) is the deepest pocket distal to heme group (green) connected to pocket B (salmon)


The subsequent studies were reported by Tojo et al.95 in 2014, where two structures of IDO1 complexed to Amg-1 (34)(PDBID: 4PK5, 2.79 Å) and an imidazothiazole inhibitor (PDBID: 4PK6, 3.45 Å) were resolved. These structures presentedlarger ligands bound to IDO1 occupying both pockets (A and B) (Figure 13b). This study showed the high side-chains andbackbone flexibility of IDO1, which lead to different active site conformations characterized by a larger pocket Aand different shape and size for pocket B.95,96

Most of the IDO1 structures described so far comprise its conjugation with weak inhibitors (micromolar range). Morerecently, Peng et al.97 disclosed several structures of IDO1 with NLG919 analogues 29-32 (Figure 14a), and those were usedfor a SAR study, with 2.1–2.7 Å resolution. These were the first structures of IDO1 complexes to be resolved with potentinhibitors (PDBID: 5EK2, 5EK3, 5EK4, and 5ETW). The obtained crystal structures revealed that the imidazoleisoindolederivatives (Figure 14a) were able to occupy both pockets (A and B), coordinating with the heme iron atom through the imid-azole nitrogen and establishing hydrophobic interactions with Tyr126, Cys129, Val130, Phe163, Phe164, Gly262, Ala264,Phe226, Arg231, Ile354, and Leu384.

Interestingly, inhibitor 31 forms an extensive hydrogen bond network with IDO1. Hydrogen bonds network include the7-propionic acid group of the heme and the NH group of Ala264, as well as an intramolecular hydrogen bond within 31,between the isoindole nitrogen and the hydroxyl group. The SAR studies demonstrated that this hydrogen bond network ispreponderant, establishing a distinct feature in this complex that was not observed in the other IDO1 complex structures(Figure 14b and e).

The crystal structures resolved provide essential structural insights concerning the active sites of IDO1, identifying severalresidue interactions crucial for the design of active inhibitors (Tyr126, Cys129, Val130, Phe163, Phe164, Phe226, Arg231,Ala260, Gly261, Gly262, Ser263, Ala264, Ile354, and Leu384).

The IDO1 structure in complex with hydroxylamidine compound 35 (PDBID: 5XE1, 3.2 Å) was reported by Wu, Xu,Liu, Ding, and Xu.98 Inhibitor 35 (Figure 15a) is an analogue of IDO1 inhibitor, currently in phase II clinical trials. The differ-ences of hydroxyamidine compounds 35 and 40, are that the latter contains 3-Br instead of 3-Cl and an additional aminoethyl-sulfamide moiety.99 The crystal structure showed that the analogue 40 only occupies the pocket A of IDO1. The authors sug-gest that 35 should occupy both pockets based on the additional aminoethyl-sulfamide moiety. This study enforces the previ-ous work reported by Peng et al.97 The addition of a halogen leads to bond interaction with the sulfur atom of Cys129, whichplays a unique role in the high inhibitory activity (Figure 15a). Moreover, they showed that 40 coordinates to heme ironthrough oxime nitrogen, which contributes to intramolecular hydrogen bond network formation. In previously reported struc-tures, all the inhibitors bound to the heme iron through a nitrogen atom in the imidazole moiety (Figure 15a).

Recently, Crosignani et al.100 from Pfizer reported the discovery and development of a new and selective IDO1 inhibitor(Figure 15c). Among a library of thousands of compounds, the most promising for subsequent optimization was compound 36(IC50 1.8 μM). Of the two enantiomers of 36, only one presented significant activity on IDO1 in the enzymatic assay. SARstudies clearly showed that the succinimide moiety could not be modified without significant drops in activity. Similarly, theindole moiety also showed that some parts of the molecule were not free for substitution. However, substitution on positionsC5 and C6 of the indole ring revealed to be beneficial generating the derivative 36.

X-ray structure of 36 and IDO1 complex was resolved with 2.3 Å resolution (PDBID: 5WHR) revealing a new bindingmode to IDO1. Succinimide ring of 16 sits parallel to the heme group and no direct coordination to iron is observed

FIGURE 13 IDO1:PI:Amg complexes. (a) Structure of IDO1/4PI (PDBID: 2D0T). 4PI (31) (yellow) occupy only the pocket A (yellow). (b) Structure ofIDO1/Amg-1 (PDBID: 4PK5). Amg-1 (32) (orange) is a larger molecule and can occupy both pockets of IDO1. Interaction of 4PI and Amg1 with IDO1 aredisplayed in red dashes


FIGURE 14 NLG919 analogues. (a) Imidazoleisoindole derivatives 29–32. Inhibitor 31 (cyan) with IC50 of 19 nM. Flourine removal resulted in 32(magenta) and two-fold decrease in IC50 (38 nM). Other modification resulted in 29 (purple) (279 nM) and 32 (dark blue) that completely lost activity(> 10 μM). (b) IDO1/24 complex. Pockets A (yellow) and B (pink) of IDO1 are occupied by 31 (cyan) that is coordinated with heme group (green) byintramolecular hydrogen bonding with the isoindole nitrogen imidazoleisoindole group. (c) Replacement of the hydroxyl group with carbonyl group(22) (purple) resulted in a 14-fold decrease in IDO1 inhibitory activity, compared to its hydroxyl analogue 12. (d) When a fluoro group was removed in theimidazoleisoindole to generate 30 (magenta), it showed two-fold decrease in the inhibition of IDO1 compared to 31. 30 could also make the coordinatedcovalent bond with the heme iron. (e) Replacement of the five-membered imidazoleisoindole ring of 30 with a six-membered, generate 32 (dark blue),resulting in almost complete loss of activity. Hydrogen bonds and π–H bonds are colored in red dash lines

FIGURE 15 IDO1/inhibitors complexes. (a) IDO1/hydroxylamidine complex (PDBID: 5XE1). Interaction of IDO1 pockets A (yellow sticks) and B(salmon sticks) residues with 35 (orange). The green dash line indicates π–π stack, and the yellow dash line indicates hydrogen bond interaction. Fluorine andchlorine atoms are colored green and light blue, respectively. (b) Key hydrogen bond interactions. IDO1/Aminotriazole derivative complex. Pocket A residuesin yellow sticks and pocket B salmon sticks and the ligand, 17, is drawn in cyan. Interactions between aminitriazole 34 (cyan) and the two pockets of IDO1are depicted in yellow dashes. (c) Pyrrolidine derivative complex (PDBID: 5WHR). The ligand 33 is drawn in orange sticks. IDO1 binding pocketrepresentation (pocket A residues in yellow sticks and pocket B residues in salmon sticks)


(Figure 15c). In addition, the indole NH interacts with side chain of Ser167, while the succinimide NH hydrogen bonds to theheme carboxylic acid. The two carbonyls interact to the main chain NH of Ala264 and Thr379. Additional interactions seemto occur due to the fact that several aromatic groups (Tyr126, Phe163, Phe164) are well positioned to establish π–π interac-tions with the indole aromatic core. Other lipophilic residues also are in close contact with the indole group (Leu234, Val130,Cys129) (Figure 15c).

Clearly, 3-(5-Fluoro-1H- indol-3-yl)pyrrolidine-2,5-dione 6 (36) appeared as a novel class of IDO1 inhibitors, showing adifferent binding mode compared to other reported IDO1 inhibitors. Although it has moderate IDO1 enzyme inhibition, itseems to be a highly efficient compound (good ligand efficiency). Despite its small size, it is driven by its tight packing withinthe enzyme, as well as, the high density of hydrogen bonds formed with IDO1.

Crystal structures resolved so far are focused on IDO1/inhibitor complexes. Recently, Yeh et al.101 reported a detailedstudy resolving the first IDO1/substrate complex structure (PDBID: 5WMW and 5WMU) with 3.0 and 2.4 Å resolution. Thesubstrate Trp binds in the distal heme pocket, which is enclosed by two loops, and covered by a small helix from the smalldomain (Figure 16a). Trp occupies both pockets (A and B), the indole group pocket A, and ammonium and carboxylategroups extend into the pocket B. Some interactions play a critical role in substrate binding and recognition. The water medi-ated H-bond of indoleamine group with Ser167 and the ionic pairing of carboxylate group with Arg231.

The search for new ligand-binding sites has lagged behind, and all IDO1 inhibitors target the reported active site. Yehet al. used 3-indole ethanol (IED) to explore additional inhibition site(s) based on the earlier studies that suggest that IED pro-motes enzyme activity by occupying a hypothesized second Trp-binding site. Two crystal structures of the IDO1/Trp complexin presence of excess IDE were resolved (PDBID: 5WMX and 5WMV) with 2.7 and 2.6 Å resolution. This confirmed theearlier studies.

IDE is located on top of a hydrophobic base constituted by Leu207, Leu339, Ala210, Leu342, and Phe214 (Figure 16b).The interactions observed are π-stacking of the indole ring with heme, and an H-bond of OH group with the 6-propionategroup.

This study also focused the IDO1/Epacadostat complex. The crystal structure was resolved with 2.5 Å resolution (PDBID:5WN8). Results proved that 18 coordinates to heme group via an unusual oxygen atom of the hydroxyamidine group, as dock-ing calculations already showed99 (Figure 16c).

Epacadostat occupies both pockets. In the pocket A the hydrophobic residues Phe163, Leu234, Phe164, Val130, Tyr126,and Ser167 enclose benzene group. Also Br and F atoms sit next to Cys129, pointing to a fluorine–sulfur contact that isknown to stabilize protein–inhibitor interactions. The furazan group, which is stabilized by Phe163, Leu234, and Phe226, andthe sulfamide group maintained by Arg231 occupy the pocket B. The coordination bond is stabilized by an H-bond withAla264, as well as two intramolecular H-bonds within the inhibitor (Figure 16c). Based on the crystal structures of two distinctinhibitors in clinical trials, new strategies can be established to generate IDO1 inhibitors.

A last study was reported by Westcott et al.102 where they have identified novel series of IDO1 inhibitors based on a4-amino-1,2,3-triazole core (37) by a HTS campaign (Figure 15b). The structural elucidation was determined using the X-raystructure of the complex (PDBID: 6F0A). The inhibitor is bound in the active site of the enzyme by the 1-nitrogen of the tria-zole ring and the iron of the heme group. A potential hydrogen-bond between the 4-amino group of the inhibitor and Ser167 may exist. Halogen interaction is also observed between the chloro-phenyl group and a hydrophobic pocket at the top ofthe active site (Figure 15b).

Besides the new inhibitor series described and the structural characterization, other questions are raised concerning theinhibitor's potency. The aminotriazole (37) was tested in biochemical (IC50 11.3 μM) and cellular (IC50 0.023 μM) assays pre-senting distinct results. The authors hypothesized that this difference is based on the formation of a tight ferrousIDO1-aminotriazole complex, suggesting that other parameters are influencing assays' results.

4.3 | Structure-based drug design for IDO1 inhibitors

All the crystal structures obtained thus far enabled the in silico structure-based design of IDO1 inhibitors. Multiple CADDtechniques have been already explored to develop and optimize IDO1 inhibitors and in some cases, with success. Moleculardynamics, molecular docking and pharmacophore modeling studies are some of the methodologies already reported.103

However, docking studies using IDO1 face several challenges as highlighted by Röhring.91 Namely, the flexibility and thepresence of heme iron in the active site. Compounds predicted to bind at the entrance of the active site by docking studies,104

using X-ray structure 2D0T disclosed by Sugimoto et al.,94 later proved to actually bind in a completely different manner,requiring a larger pocket A. This was demonstrated by the crystallographic structure reported by Tojo et al.95 In fact, the pres-ence of the heme iron in the active site, which can interact with ligands by forming covalent bonds, led to the development ofa special Morse-like metal binding potential.105 The strong dependence of IDO1 from the electronic nature of the iron-bindinggroup is difficult to describe consistently in any computational approach due to the lack of correct forcefield parameters for


FIGURE 16 IDO1/substrate, IDO1/IDE and IDO1/epacadostat complexes. (a) Crystal structure of the IDO1-Trp complex. Binding mode of Trp (yellowsticks). Trp binds in the distal heme (green) pocket A (yellow) and occupy the pocket B (salmon). The Trp interacts with the protein and heme, via varioushydrophobic and polar interactions as represented by yellow dashes. (b) New small molecule binding site in IDO1. Crystal structure of the IDO1-Trp complexin a mixed ligand state, where active site and the new site are occupied by Trp and IDE (cyan), respectively. The hydrophobic residues forming the base of thebinding pocket are shown as magenta sticks, while pockets A and B are shown in yellow and salmon sticks. IED-IDO1 protein interactions are represented inyellow dashes. (c) Binding mode of epacadostat (33) to IDO1. Protein surface is represented in blue, pocket A in yellow and pocket B salmon stickrepresentation. The ligand is drawn in pink and the heme in green stick representation. IDO1 surface-binding pocket representation. The interactions betweenepacadostat (magenta sticks) and the two pockets of IDO1 (pocket A and pocket B), as well as the intramolecular H-bonds are represented by the yellowdashes


the iron. This sensitivity was highlighted by bioisosteric replacements of the imidazole ring in 4PI by other nitrogen-containing five-membered rings, leading to inhibitory activity with deviations of several orders of magnitude.91 Together withdocking, molecular dynamics was used to better understand the reaction mechanism of IDO1, as well as, to interpret dockingresults.106 For example, the study performed by Rohrig et al. (2015) demonstrated that miconazole only fits in the enlargedIDO1 active site after molecular dynamics simulations.107

Though in silico design has significantly contributed to the development of IDO1 inhibitors, conflicting binding modeshave been reported, pointing out the problem of the validity of those predicted binding modes. High flexibility of IDO1, inpart, is explained by the presence of a highly conserved sequence Ala260, Gly261, Gly262, Ser263, Ala264, and Gly265 thatprovides conformational flexibility in several proteins. Thus, IDO1, exhibiting this sequence, leads to several active site con-formations due to a very flexible binding pocket. Caution is thus necessary to interpret docking results, considering the proteinflexibility and heme interactions.

Structure-based design of new IDO1 inhibitors, so far, has been mainly based on the 4PI-bound X-ray structure describedin 2006, condition that has been changing by the increasing number of new available structures. Together with inhibitors'bound structures, the crystal structure of substrate bound state, recently resolved, brings an opportunity to develop new ratio-nales on the development of IDO1 inhibitors. Therefore, even though great efforts have been made to develop IDO1 inhibitorsusing structure-based design, high throughput screening (HTS)108–111 and natural product screening112–116 are still ongoingand successfully yielded different active scaffolds.117

Even though intensive research has been devoted to the development of IDO1 inhibitors, only a few compounds are inphases 1 and 2 of clinical trials (Figure 17). To successfully achieve IDO1 inhibitors, the strategies adopted rely on HTS orstructure-based campaigns. Hydroxyamidine (33) (INCB024360, epacadostat) by Incyte Corporation and PF-06840003(29) by Pfizer were obtained by HTS.118,119 Indoximod (31) and the imidazole NLG919 (32) by NewLink Genetics120

achieved by structure-based design using the structure of IDO1 complex with 4PI. More recently, in phase 1, Bristol Myers-Squibb developed the cyclohexaneacetamide (BMS-986205) (34).121

4.4 | 4-1BB

4-1BB, also known CD137 or TNF receptor superfamily member 9, is involved in co-stimulatory signals for T-cells.122 TheT-cell co-stimulatory receptor 4-1BB is induced when T-cells receive antigen-specific signals. These 4-1BB-mediated signalshave been shown to induce a novel subpopulation of T cells that have strong anticancer and anti-autoimmune effects.123

FIGURE 17 IDO1 inhibitors. Epacadostat 40 started clinical testing in 2014 and is currently in phase II of clinical trials being the most advanced IDO1inhibitor. Indoximod 38 started clinical testing in 2008 and is currently in phase I of clinical trials. In 2014, NLG919 39 started clinical testing and it is inphase I of clinical trials. Also 41 as initiated clinical trials. The clinical evaluation that is occurring use the molecules as only therapeutic or in combinationwith others


Concerning 4-1BB, this is a type 1 transmembrane glycoprotein with 255 aa in length composed by an extracellulardomain (residues 24-186), a transmembrane domain (residues 187-213) and a cytoplasmic domain (residues 214-255).

4-1BB binds to its ligand 4-1BBL, a molecule of the TNF family. To date, 4-1BBL remains the only intracellular ligandknown for 4-1BB, although the extracellular domain binds also to fibronectin and to galectin-9. 4-1BBL is induced on APC,such as DC, macrophages, and B cells.124

4-1BBL, is a type 2 glycoprotein with 254 aa in length whose extracellular TNF homology domains (THD) form noncova-lent trimers (Figure 18a and c). 4-1BBL is composed of a cytoplasmic domain (residues 1–25), a transmembrane domain (resi-dues 26–48), and an extracellular domain (residues 49–254). The extracellular domain can be further divided into the tailregion (residues 49–92) and the THD (residues 93–254).

The 4-1BBL-4-1BB pathway co-stimulates T cells to carry out effector functions, such as the eradication of establishedtumors and the broadening of T-cell responses.

Structural insights of the 4-1BBL-4-1BB pathway are very limited. However, Won et al. resolved the crystallographicstructure of 4-1BBL (2.3 Å resolution), providing structural insights of this TNF family member.125

The 4-1BBL extracellular domain of each monomer forms α-β sheet structure, known as a β jelly-roll fold. The structure of4-1BBL trimer has the appearance of a three-fold symmetric, three-bladed propeller (Figure 18c). The subunits of 4-1BBL interactwith each other involving 14 residues, both hydrophobic and aromatic, namely Gln89, Val140, Leu203, His205, Val240, Thr241,Pro242, and Pro245 of one monomer, and Gln94, Phe92, Leu115, Tyr142, Phe144, Phe197, and Phe199 of the other monomer. Themajor differences between 4-1BBL and the other TNF family members lie in the relative orientations of the N and C termini alongwith the positions of the flexible loops. The N and C termini of 4-1BBL extend from the inner and outer sheets on opposite sides ofthe molecule, whereas those of other TNF family members are located near to each other, on the same side of the molecule.125

4-1BBL trimeric ligand interacts in a 3:3 stoichiometry, such that the 4-1BB binds at the trimer interface of the ligands.Although the crystal structure of the 4-1BB-4-1BBL complex is not resolved, a model of the complex was also generated byWon et al. using a homology model of 4-1BB obtained with the SWISS-Model server. According to the model, the three resi-dues of 4-1BBL, Lys127, Gln227, and Gln230 should be in close proximity to 4-1BB.125 Mutagenesis studies demonstratedthat a single mutation of Lys127 and Gln227 to alanine caused 170- and 80-fold decrease in binding affinity, respectively. Asthe Lys127 (A'B loop) and Gln227 (H strand) are in an exposed surface away from the trimerization interface, the mutationsare likely to have a direct effect on receptor binding rather than interfering with trimerization. The mutagenesis studies showedthat the long A'B loop of 4-1BBL is preponderant for receptor binding.

A deeper characterization of the crystal structure of the 4-1BB-4-1BBL complex would reveal the precise epitopes of theantibodies and the molecular mechanisms underlying the ligand recognition. This would further add into 4-1BB stimulation,leading to novel 4-1BB-4-1BBL-targeted therapeutic options.

FIGURE 18 4-1BBL structures (PDBID: 2X29). (a) Ribbon representation of 4-1BBL (green). The structure shows a two-layered jellyroll β-sandwichtopology similar to the canonical structure of the other TNF family members. The inner and outer sheets of the jelly-roll β-sandwich are composed of A'HCFand ABGDE, respectively. (b) The residues involved in the hydrophobic interactions are: Gln89, Val140, Leu203, His205, Val240, Thr241, Pro242, andPro245 in one protomer (green) and Gln94, Phe92, Leu115, Tyr142, Phe144, Phe197, and Phe199 in the adjacent protomer (green). (c) Ribbon representationof 4-1BBL trimer. 4-1BBL forms a three-fold symmetric assembly resembling a three-bladed propeller. The interactions between 4-1BBL subunits in thetrimer are primarily mediated by hydrophobic interactions between the C-terminal tail of one subunit and the A'B loop and the A’, C, and F strands of theadjacent subunit


Currently, 4-1BB-targeted therapeutic options rely on two monoclonal antibodies that are in clinical trials, Utomilumab®

(Pfizer) in phase 1 and Urelumab® (Bristol-Myers Squibb) in phase 2. These monoclonal antibodies are being tested to inducea potent anti-tumor immune-mediated destruction by targeting the extracellular domain of the 4-1BB receptor. Preliminaryresults demonstrate auspicious clinical responses; however, some toxicity has been associated.

5 | CONCLUSION

Cancer immunotherapy is undoubtedly an enormous success that had revolutionized cancer treatments. However, the efficacycircumvents to specific subsets of cancer patients and the induction of severe immune-related adverse effects lead to the pur-suit of new safer and more effective strategies. The target of additional ICRs and the regulation of the function of other com-ponents in the tumor microenvironment by small molecules have been hypothesized. Even though, it is fundamental to betterunderstand the mechanisms underlying the stimulation of an antitumor immune response by a targeting molecule at thecellular, but also molecular levels. This deeper structure–function knowledge will allow not only the rational development ofpromising immune modulatory candidates with optimal on-target interactions, but will also prompt the discovery and valida-tion of additional targets and/or reveal the best candidates for combinatorial strategies aiming at synergistic outcomes betweenmultiple immune-related pathways and proteins. In this overview, we explored the structure–function relationships availablefor emerging regulators of immune and tumor microenvironment players (A2AR, STING, IDO1 and 4-1BB), opening newavenues for the unique role that computational tools can have on the faster and optimized design of effective, but also safetargeted cancer immunotherapies.

ACKNOWLEDGMENTS

RA is supported by the Fundação para a Ciência e a Tecnologia, Ministério da Ciência, Tecnologia e Ensino Superior (FCT-MCTES) (PhD grants PD/BD/128238/2016). The authors thank the funding received from the European Structural & Invest-ment Funds through the COMPETE Programme and from National Funds through FCT under the Programme grant SAICT-PAC/0019/2015, PTDC/QEQ-MED/7042/2014 and UID/DTP/04138/2013 (HF, RG). The MultiNano@MBM project wassupported by The Israeli Ministry of Health, and FCT-MCTES, under the frame of EuroNanoMed-II (ENMed/0051/2016; HF,RS-F).

CONFLICT OF INTEREST

The authors have declared no conflicts of interest for this article.

RELATED WIREs ARTICLES

About P-glycoprotein: a new drugable domain is emerging from structural data

REFERENCES

1. Baldo BA. Adverse events to monoclonal antibodies used for cancer therapy: Focus on hypersensitivity responses. Oncoimmunology. 2013;2:e26333.2. Kroschinsky F, Stölzel F, von Bonin S, et al. New drugs, new toxicities: Severe side effects of modern targeted and immunotherapy of cancer and their manage-

ment. Crit Care. 2017;21:89.3. Michot JM, Bigenwald C, Champiat S, et al. Immune-related adverse events with immune checkpoint blockade: A comprehensive review. Eur J Cancer. 2016;54:

139–148.4. Gandhi L, Rodríguez-Abreu D, Gadgeel S, et al. Pembrolizumab plus chemotherapy in metastatic non–small-cell lung Cancer. N Engl J Med. 2018;378:

2078–2092.5. Ramagopal UA, Liu W, Garrett-Thomson SC, et al. Structural basis for cancer immunotherapy by the first-in-class checkpoint inhibitor ipilimumab. Proc Natl

Acad Sci. 2017;114:201617941.6. Weiner LM, Dhodapkar MV, Ferrone S. Monoclonal antibodies for cancer immunotherapy. Lancet. 2009;373:1033–1040.7. Haskó G, Cronstein BN. Adenosine: An endogenous regulator of innate immunity. Trends Immunol. 2004;25:33–39.8. Poth JM, Brodsky K, Ehrentraut H, Grenz A, Eltzschig HK. Transcriptional control of adenosine signaling by hypoxia-inducible transcription factors during ische-

mic or inflammatory disease. J Mol Med. 2013;91:183–193.9. Allard B, Beavis PA, Darcy PK, Stagg J. Immunosuppressive activities of adenosine in cancer. Curr Opin Pharmacol. 2016;29:7–16.10. Hoskin DW, Mader JS, Furlong SJ, Conrad DM, Blay J. Inhibition of T cell and natural killer cell function by adenosine and its contribution to immune evasion

by tumour cells (review). Int J Oncol. 2008;32:527–535.11. Sitkovsky MV, Kjaergaard J, Lukashev D, Ohta A. Hypoxia-adenosinergic immunosuppression: Tumour protection by T regulatory cells and cancerous tissue

hypoxia. Clin Cancer Res. 2008;14:5947–5952.12. Zarek PE, Huang C, Lutz ER, et al. A 2A receptor signaling promotes peripheral tolerance by inducing T-cell anergy and the generation of adaptive regulatory T

cells. Blood. 2014;111:251–259.



13. Han KL, Thomas SVM, Koontz SM, et al. Adenosine a2a receptor agonist-mediated increase in donor-derived regulatory T cells suppresses development ofgraft-versus-host disease. J Immunol. 2013;1950(190):458–468.

14. Ohta A, Kini R, Ohta A, Subramanian M, Madasu M, Sitkovsky M. The development and immunosuppressive functions of CD4+ CD25+ FoxP3+ regulatory Tcells are under influence of the adenosine-A2A adenosine receptor pathway. Front Immunol. 2012;3:1–12.

15. Ohta A. A metabolic immune checkpoint: Adenosine in tumour microenvironment. Front Immunol. 2016;7:1–11.16. Ohta A, Gorelik E, Prasad SJ, et al. A2A adenosine receptor protects tumours from antitumour T cells. Proc Natl Acad Sci. 2006;103:13132–13137.17. Young A, Mittal D, Stannard K, et al. Co-blockade of immune checkpoints and adenosine A 2A receptor suppresses metastasis. Oncoimmunology. 2014;15:8–10.18. Künzli BM, Bernlochner MI, Rath S, et al. Impact of CD39 and purinergic signalling on the growth and metastasis of colorectal cancer. Purinergic Signal. 2011;

7:231–241.19. Montinaro A, Iannone R, Pinto A, Morello S. Adenosine receptors as potential targets in melanoma. Pharmacol Res. 2013;76:34–40.20. Stagg J, Divisekera U, McLaughlin N, et al. Anti-CD73 antibody therapy inhibits breast tumour growth and metastasis. Proc Natl Acad Sci. 2010;107:1547–1552.21. de Lera Ruiz M, Lim Y-H, Zheng J. Adenosine A 2A receptor as a drug discovery target. J Med Chem. 2014;57:3623–3650.22. Milne GR, Palmer TM. Anti-inflammatory and immunosuppressive effects of the A2A adenosine receptor. Scientific World Journal. 2011;11:320–339.23. Pinna A, Wardas J, Simola N, Morelli M. New therapies for the treatment of Parkinson's disease: Adenosine A2Areceptor antagonists. Life Sci. 2005;77:

3259–3267.24. Sanders MPA, Roumen L, Van Der Horst E, et al. A prospective cross-screening study on G-protein-coupled receptors: Lessons learned in virtual compound

library design. J Med Chem. 2012;55:5311–5325.25. Katritch V, Kufareva I, Abagyan R. Structure based prediction of subtype-selectivity for adenosine receptor antagonists. Neuropharmacology. 2011;60:108–115.26. Michino M, Abola E, Brooks CL, et al. Community-wide assessment of GPCR structure modelling and ligand docking: GPCR dock 2008. Nat Rev Drug Discov.

2009;8:455–463.27. Jaakola V, Griffith MT, Hanson MA, et al. The 2.6 angstrom crystal structure of a human A2A adenosine receptor bound to an antagonist. Science. 2008;322:

1211–1217.28. Poucher SM, Keddie JR, Singh P, et al. The in vitro pharmacology of ZM 241385, a potent, non-xanthine, A2aselective adenosine receptor antagonist. Br J Phar-

macol. 1995;115:1096–1102.29. Rasmussen SGF, Choi H-J, Rosenbaum DM, et al. Crystal structure of the human β2 adrenergic G-protein-coupled receptor. Nature. 2007;450:383–387.30. Scheerer P, Park JH, Hildebrand PW, et al. Crystal structure of opsin in its G-protein-interacting conformation. Nature. 2008;455:497–502.31. Warne T, Serrano-Vega MJ, Baker JG, et al. Structure of a β1-adrenergic G-protein-coupled receptor. Nature. 2008;454:486–491.32. Martinelli A, Tuccinardi T. Molecular modeling of adenosine receptors: New results and trends. Med Res Rev. 2008;28:247–277.33. Yuzlenko O, Kiec-Kononowicz K. Molecular modeling of A1 and A2A adenosine receptors: Comparison of rhodopsin- and B2-adrenergic-based homology

models through the docking studies. J Comput Chem. 2009;30:1545–1614.34. Hino T, Arakawa T, Iwanari H, et al. G-protein-coupled receptor inactivation by an allosteric inverse-agonist antibody. Nature. 2012;482:237–240.35. Doré AS, Robertson N, Errey JC, et al. Structure of the adenosine A2Areceptor in complex with ZM241385 and the xanthines XAC and caffeine. Structure. 2011;

19:1283–1293.36. Liu W, Chun E, Thompson AA, et al. Structural basis for allosteric regulation of GPCRS by sodium ions. Science. 2012;337:232–236.37. Segala E, Guo D, Cheng RKY, et al. Controlling the dissociation of ligands from the adenosine A2AReceptor through modulation of salt bridge strength. J Med

Chem. 2016;59:6470–6479.38. Xu F, Wu H, Katritch V, et al. Structure of an agonist-bound human A2A adenosine receptor. Science. 2011;332:322–327.39. Lebon G, Warne T, Edwards PC, et al. Agonist-bound adenosine A2A receptor structures reveal common features of GPCR activation. Nature. 2011;474:521–526.40. Lebon G, Edwards PC, Leslie AGW, Tate CG. Molecular determinants of CGS21680 binding to the human adenosine A2A receptor. Mol Pharmacol. 2015;87:

907–915.41. White KL, Eddy MT, Gao ZG, et al. Structural connection between activation microswitch and allosteric sodium site in GPCR signaling. Structure. 2018;26:

259–269.e5.42. Congreve M, Andrews SP, Doré AS, et al. Discovery of 1,2,4-triazine derivatives as adenosine A2Aantagonists using structure based drug design. J Med Chem.

2012;55:1898–1903.43. Zhukov A, Andrews SP, Congreve M, et al. Biophysical Mapping of the Adenosine A2A Receptor. J Med Chem. 2012;54:4312–4323.44. Sun B, Bachhawat P, Chu ML-H, et al. Crystal structure of the adenosine A 2A receptor bound to an antagonist reveals a potential allosteric pocket. Proc Natl

Acad Sci. 2017;114:2066–2071.45. Cheng RKY, Segala E, Robertson N, et al. Structures of human A1and A2AAdenosine receptors with Xanthines reveal determinants of selectivity. Structure.

2017;25:1275–1285.e4.46. Carpenter B, Nehmé R, Warne T, Leslie AGW, Tate CG. Structure of the adenosine A2Areceptor bound to an engineered G protein. Nature. 2016;536:104–107.47. Heilker R, Wolff M, Tautermann CS, Bieler M. G-protein-coupled receptor-focused drug discovery using a target class platform approach. Drug Discov Today.

2009;14:231–240.48. Klabunde T, Hessler G. Drug design strategies for targeting G-protein-coupled receptors. ChemBioChem. 2002;3:928–944.49. Schnur D, Beno BR, Good A, Tebben A. Approaches to target class combinatorial library design. Methods Mol Biol. 2004;275:355–377.50. Gessi S, Varani K, Merighi S, Ongini E, Borea PA. A 2A adenosine receptors in human peripheral blood cells. Br J Pharmacol. 2000;129:2–11.51. Snyder SH, Katims JJ, Annau Z, Bruns RF, Daly JW. Adenosine receptors and behavioral actions of methylxanthines. Proc Natl Acad Sci U S A. 1981;78:

3260–3264.52. Katritch V, Jaakola VP, Lane JR, et al. Structure-based discovery of novel chemotypes for adenosine A2Areceptor antagonists. J Med Chem. 2010;53:1799–1809.53. Carlsson J, Yoo L, Gao Z-G, Irwin JJ, Shoichet BK, Jacobson KA. Structure-based discovery of a 2A adenosine receptor ligands. J Med Chem. 2010;53:

3748–3755.54. Areias F, Costa M, Castro M, et al. New chromene scaffolds for adenosine A2A receptors: Synthesis, pharmacology and structure-activity relationships. Eur J Med

Chem. 2012;54:303–310.55. Langmead CJ, Andrews SP, Congreve M, et al. Identification of novel adenosine A2A receptor antagonists by virtual screening. J Med Chem. 2012;55:

1904–1909.56. Van Westen GJP, Van Den Hoven OO, Van Der Pijl R, et al. Identifying novel adenosine receptor ligands by simultaneous proteochemometric modeling of rat

and human bioactivity data. J Med Chem. 2012;55:7010–7020.57. Deflorian F, Kumar TS, Phan K, et al. Evaluation of molecular modeling of agonist binding in light of the crystallographic structure of an agonist-bound a 2A

adenosine receptor. J Med Chem. 2012;55:538–552.58. Keränen H, Åqvist J, Gutiérrez-de-Terán H. Free energy calculations of a 2A adenosine receptor mutation effects on agonist binding. Chem Commun. 2015;51:

3522–3525.


59. Tosh DK, Phan K, Gao ZG, et al. Optimization of adenosine 50-carboxamide derivatives as adenosine receptor agonists using structure-based ligand design andfragment screening. J Med Chem. 2012;55:4297–4308.

60. Sabbadin D, Ciancetta A, Deganutti G, Cuzzolin A, Moro S. Exploring the recognition pathway at the human A2Aadenosine receptor of the endogenous agonistadenosine using supervised molecular dynamics simulations. Med Chem Commun. 2015;6:1081–1085.

61. Sabbadin D, Ciancetta A, Moro S. Bridging molecular docking to membrane molecular dynamics to investigate GPCR-ligand recognition: The human A2A adeno-sine receptor as a key study. J Chem Inf Model. 2014;54:169–183.

62. Bortolato A, Tehan BG, Bodnarchuk MS, Essex JW, Mason JS. Water network perturbation in ligand binding: Adenosine A2A antagonists as a case study. J ChemInf Model. 2013;53:1700–1713.

63. Matsuno K, Yamazaki H, Isaka Y, et al. Novel candesartan derivatives as indoleamine 2,3-dioxygenase inhibitors. Med Chem Commun. 2012;3:475.64. Sun W, Li Y, Chen L, et al. ERIS, an endoplasmic reticulum IFN stimulator, activates innate immune signaling through dimerization. Proc Natl Acad Sci U S A.

2009;106:8653–8658.65. Cai X, Chiu YH, Chen ZJ. The cGAS-cGAMP-STING pathway of cytosolic DNA sensing and signaling. Mol Cell. 2014;54:289–296.66. Corrales L, Gajewski TF. Endogenous and pharmacologic targeting of the STING pathway in cancer immunotherapy. Cytokine. 2016;77:245–247.67. Woo S, Fuertes MB, Corrales L, et al. STING-dependent cytosolic DNA sensing mediates innate immune recognition of immunogenic tumours. Immunity. 2014;

41:830–842.68. Chen Q, Boire A, Jin X, et al. Carcinoma-astrocyte gap junctions promote brain metastasis by cGAMP transfer. Nature. 2016;533:493–498.69. Sugama S, Sekiyama K, Kodama T, et al. STING promotes the growth of tumours characterized by low antigenicity via IDO activation. Cancer Res. 2016;76:

39–46.70. Tanaka Y, Chen ZJ. STING specifies IRF3 phosphorylation by TBK1 in the cytosolic DNA signaling pathway. Receptor. 2012;5:ra20.71. Huang Y-H, Liu X-Y, Du X-X, Jiang Z-F, Su X-D. The structural basis for the sensing and binding of cyclic di-GMP by STING. Nat Struct Mol Biol. 2012;19:

728–730.72. Ouyang S, Song X, Wang Y, et al. Structural analysis of the STING adaptor protein reveals a hydrophobic dimer Interface and mode of cyclic di-GMP binding.

Immunity. 2012;36:1073–1086.73. Shang G, Zhu D, Li N, et al. Crystal structures of STING protein reveal basis for recognition of cyclic di-GMP. Nat Struct Mol Biol. 2012;19:725–727.74. Shu C, Yi G, Watts T, Kao CC, Li P. Structure of STING bound to c-di-GMP reveals the mechanism of cyclic dinucleotide recognition by the immune system.

Nat Struct Mol Biol. 2012;19:722–724.75. Yin Q, Tian Y, Kabaleeswaran V, et al. Cyclic di-GMP sensing via the innate immune signaling protein STING. Mol Cell. 2012;46:735–745.76. Ablasser A, Goldeck M, Cavlar T, et al. CGAS produces a 20-50-linked cyclic dinucleotide second messenger that activates STING. Nature. 2013;498:380–384.77. Diner EJ, Burdette DL, Wilson SC, et al. The innate immune DNA sensor cGAS produces a non-canonical cyclic-di-nucleotide that activate human STING. Cell

Rep. 2013;3:1355–1361.78. Gao P, Ascano M, Zillinger T, et al. Structure-function analysis of STING activation by c[G(20,50) pA(30,50)p] and targeting by antiviral DMXAA. Cell. 2013;154:

748–762.79. Zhang X, Shi H, Wu J, et al. Cyclic GMP-AMP containing mixed phosphodiester linkages is an endogenous high-affinity ligand for STING. Mol Cell. 2013;51:

226–235.80. Shi H, Wu J, Chen ZJ, Chen C. Molecular basis for the specific recognition of the metazoan cyclic GMP-AMP by the innate immune adaptor protein STING. Proc

Natl Acad Sci. 2015;112:8947–8952.81. Baguley BC, Ching LM. DMXAA: An antivascular agent with multiple host responses. Int J Radiat Oncol Biol Phys. 2002;54:1503–1511.82. Lara PN, Douillard JY, Nakagawa K, et al. Randomized phase III placebo-controlled trial of carboplatin and paclitaxel with or without the vascular disrupting

agent vadimezan (ASA404) in advanced non-small-cell lung cancer. J Clin Oncol. 2011;29:2965–2971.83. Prantner D, Perkins DJ, Lai W, et al. 5,6-Dimethylxanthenone-4-acetic acid (DMXAA) activates stimulator of interferon gene (STING)-dependent innate immune

pathways and is regulated by mitochondrial membrane potential. J Biol Chem. 2012;287:39776–39788.84. Conlon J, Burdette DL, Sharma S, et al. Mouse, but not human STING, binds and signals in response to the vascular disrupting agent DMXAA. J Immunol. 2013;

190:5216–5225.85. Kim S, Li L, Maliga Z, Yin Q, Wu H, Mitchison TJ. Anti-cancer flavonoids are mouse selective STING agonists. ACS Chem Biol. 2013;18:1199–1216.86. Liu S, Cai X, Wu J, et al. Phosphorylation of innate immune adaptor proteins MAVS, STING, and TRIF induces IRF3 activation. Science. 2015;347:1–17.87. Zhao B, Shu C, Gao X, et al. Structural basis for concerted recruitment and activation of IRF-3 by innate immune adaptor proteins. Proc Natl Acad Sci. 2016;113:

E3403–E3412.88. Mellor AL, Munn DH. IDO expression by dendritic cells: Tolerance and tryptophan catabolism. Nat Rev Immunol. 2004;4:762–774.89. Prendergast GC, Smith C, Thomas S, Mandik-nayak L, Metz R, Muller AJ. Indoleamine 2,3-dioxygenase pathways of pathogenic inflammation and immune

escape in cancer. Cancer Immunol Immunother. 2014;63:721–735.90. Vacchelli E, Aranda F, Eggermont A, et al. Trial watch: IDO inhibitors in cancer therapy. Oncoimmunology. 2014;3:e957994.91. Röhrig UF, Majjigapu SR, Vogel P, Zoete V, Michielin O. Challenges in the discovery of Indoleamine challenges in the discovery of Indoleamine. J Med Chem.

2015;24:9421–9437.92. Qian S, Zhang M, Chen Q, He Y, Wang W, Wang Z. IDO as a drug target for cancer immunotherapy: Recent developments in IDO inhibitors discovery. RSC

Adv. 2016;6:7575–7581.93. Smith JR, Evans KJ, Wright A, Willows RD, Jamie JF, Griffith R. Novel indoleamine 2,3-dioxygenase-1 inhibitors from a multistep in silico screen. Bioorg Med

Chem. 2012;20:1354–1363.94. Sugimoto H, Oda S, Otsuki T, Hino T, Yoshida T, Shiro Y. Crystal structure of human indoleamine 2,3-dioxygenase: Catalytic mechanism of O2 incorporation by

a heme-containing dioxygenase. Proc Natl Acad Sci U S A. 2006;103:2611–2616.95. Tojo S, Kohno T, Tanaka T, et al. Crystal structures and structure-activity relationships of imidazothiazole derivatives as IDO1 inhibitors. ACS Med Chem Lett.

2014;5:1119–1123.96. Röhrig UF, Awad L, Grosdidier A, et al. Rational design of indoleamine 2,3-dioxygenase inhibitors. J Med Chem. 2010;53:1172–1189.97. Peng YH, Ueng SH, Tseng CT, et al. Important hydrogen bond networks in Indoleamine 2,3-dioxygenase 1 (IDO1) inhibitor design revealed by crystal structures

of Imidazoleisoindole derivatives with IDO1. J Med Chem. 2016;59:282–293.98. Wu Y, Xu T, Liu J, Ding K, Xu J. Structural insights into the binding mechanism of IDO1 with hydroxylamidine based inhibitor INCB14943. Biochem Biophys

Res Commun. 2017;487:339–343.99. Yue EW, Sparks R, Polam P, et al. INCB24360 (Epacadostat), a highly potent and selective Indoleamine-2,3-dioxygenase 1 (IDO1) inhibitor for Immuno- oncol-

ogy. ACS Med Chem Lett. 2017;8:486–491.100. Crosignani S, Bingham P, Bottemanne P, et al. Discovery of a novel and selective Indoleamine 2,3-dioxygenase (IDO-1) inhibitor 3-(5-Fluoro-1H-indol-3-yl)

pyrrolidine-2,5-dione (EOS200271/PF-06840003) and its characterization as a potential clinical candidate. J Med Chem. 2017;60:9617–9629.


101. Lewis-Ballester A, Pham KN, Batabyal D, et al. Structural insights into substrate and inhibitor binding sites in human indoleamine 2,3-dioxygenase 1. Nat Com-mun. 2017;8:1693.

102. Alexandre J, Swan M, Latchem M, et al. Novel 4-amino-1,2,3-triazole inhibitors of indoleamine 2,3-dioxygenase form a long-lived complex with the enzyme anddisplay exquisite cellular potency. ChemBioChem. 2017;19:552–561.

103. Reddy SVG, Reddy KT, Kumari VV, Basha SH. Molecular docking and dynamic simulation studies evidenced plausible immunotherapeutic anticancer propertyby Withaferin a targeting indoleamine 2,3-dioxygenase. J Biomol Struct Dyn. 2015;33:2695–2709.

104. Reniers J, Meinguet C, Moineaux L, et al. Synthesis and inhibition study of monoamine oxidase, indoleamine 2,3-dioxygenase and tryptophan 2,3-dioxygenase by3,8-substituted 5H-indeno[1,2-c]pyridazin-5-one derivatives. Eur J Med Chem. 2011;46:6104–6111.

105. Röhrig U, Grosdidier A, Zoete V, Michielin O. Docking to Heme proteins. J Comput Chem. 2009;30:2305–2315.106. Capece L, Lewis-Ballester A, Yeh SR, Estrin DA, Marti MA. Complete reaction mechanism of indoleamine 2,3-dioxygenase as revealed by QM/MM simulations.

J Phys Chem B. 2012;116:1401–1413.107. Röhrig UF, Majjigapu SR, Chambon M, et al. Detailed analysis and follow-up studies of a high-throughput screening for indoleamine 2,3-dioxygenase 1 (IDO1)

inhibitors. Eur J Med Chem. 2014;84:284–301.108. Cheng MF, Hung MS, Song JS, et al. Discovery and structure-activity relationships of phenyl benzenesulfonylhydrazides as novel indoleamine 2,3-dioxygenase

inhibitors. Bioorg Med Chem Lett. 2014;24:3403–3406.109. Doluši�c E, Larrieu P, Blanc S, et al. Discovery and preliminary SARs of keto-indoles as novel indoleamine 2,3-dioxygenase (IDO) inhibitors. Eur J Med Chem.

2011;46:3058–3065.110. Matsuno K, Takai K, Isaka Y, et al. S-Benzylisothiourea derivatives as small-molecule inhibitors of indoleamine-2,3-dioxygenase. Bioorg Med Chem Lett. 2010;

20:5126–5129.111. Tanaka M, Li X, Hikawa H, et al. Synthesis and biological evaluation of novel tryptoline derivatives as indoleamine 2,3-dioxygenase (IDO) inhibitors. Bioorg

Med Chem. 2013;21:1159–1165.112. Bridewell DJA, Sperry J, Smith JR, et al. Natural product-inspired pyranonaphthoquinone inhibitors of indoleamine 2,3-dioxygenase-1 (IDO-1). Aust J Chem.

2013;66:40–49.113. Centko RM, Steinø A, Rosell FI, et al. Indoleamine 2,3-dioxygenase inhibitors isolated from the sponge Xestospongia vansoesti: Structure elucidation, analogue

synthesis, and biological activity. Org Lett. 2014;16:6480–6483.114. Pasceri R, Siegel D, Ross D, Moody CJ. Aminophenoxazinones as inhibitors of indoleamine 2,3-dioxygenase (IDO). Synthesis of exfoliazone and chandrananimy-

cin a. J Med Chem. 2013;56:3310–3317.115. Williams DE, Steinø A, De Voogd NJ, Mauk AG, Andersen RJ. Halicloic acids a and B isolated from the marine sponge Haliclona sp. collected in The Philippines

inhibit indoleamine 2,3-dioxygenase. J Nat Prod. 2012;75:1451–1458.116. Yang S, Li X, Hu F, et al. Discovery of Tryptanthrin derivatives as potent inhibitors of Indoleamine 2, 3-dioxygenase with therapeutic activity in LLC

tumour-bearing mice. J Med Chem. 2013;56:8321–8331.117. Paul S, Roy A, Deka SJ, Panda S, Trivedi V, Manna D. Nitrobenzofurazan derivatives of N???-hydroxyamidines as potent inhibitors of

indoleamine-2,3-dioxygenase 1. Eur J Med Chem. 2016;121:364–375.118. Koblish HK, Hansbury MJ, Bowman KJ, et al. Hydroxyamidine inhibitors of indoleamine-2,3-dioxygenase potently suppress systemic tryptophan catabolism and

the growth of IDO-expressing tumours. Mol Cancer Ther. 2010;9:489–498.119. Soliman HH, Jackson E, Neuger T, et al. A first in man phase I trial of the oral immunomodulator, indoximod , combined with docetaxel in patients with meta-

static solid tumours. Oncotarget. 2014;5:8136–8146.120. Mautino MR, Jaipuri FA, Waldo J, et al. Abstract 491: NLG919, a novel indoleamine-2,3-dioxygenase (IDO)-pathway inhibitor drug candidate for cancer therapy.

Cancer Res. 2014;73:491.121. Siu LL, et al. (2017). BMS-986205, an optimized indoleamine 2,3-dioxygenase 1 (IDO1) inhibitor, is well tolerated with potent pharmacodynamic (PD) activity,

alone and in combination with nivolumab (nivo) in advanced cancers in a phase 1/2a trial. Cancer Res. 2017;77:13.122. Watts TH. Tnf/Tnfr family members in Costimulation of T cell responses. Annu Rev Immunol. 2005;23:23–68.123. Vick E, Mahadevan D. Programming the immune checkpoint to treat hematologic malignancies. Expert Opin Investig Drugs. 2016;3784:

13543784.2016.1175433.124. Snell LM, Lin GHY, McPherson AJ, Moraes TJ, Watts TH. T-cell intrinsic effects of GITR and 4-1BB during viral infection and cancer immunotherapy. Immunol

Rev. 2011;244:197–217.125. Won EY, Cha K, Byun JS, et al. The structure of the trimer of human 4-1BB ligand is unique among members of the tumour necrosis factor superfamily. J Biol

Chem. 2010;285:9202–9210.

How to cite this article: Acúrcio RC, Scomparin A, Satchi-Fainaro R, Florindo HF, Guedes RC. Computer-aided drugdesign in new druggable targets for the next generation of immune-oncology therapies. WIREs Comput Mol Sci. 2019;9:e1397. https://doi.org/10.1002/wcms.1397



Documents

Computer-aided drug design in new druggable targets for the next … · 2020. 6. 13. · computationally driven drug discovery in the rational search for novel bioactive compounds