Terphenyl-Based Small-Molecule Inhibitors of Programmed Cell Death-1/Programmed Death-Ligand 1 Protein–Protein Interaction

We describe a new class of potent PD-L1/PD-1 inhibitors based on a terphenyl scaffold that is derived from the rigidified biphenyl-inspired structure. Using in silico docking, we designed and then experimentally demonstrated the effectiveness of the terphenyl-based scaffolds in inhibiting PD-1/PD-L1 complex formation using various biophysical and biochemical techniques. We also present a high-resolution structure of the complex of PD-L1 with one of our most potent inhibitors to identify key PD-L1/inhibitor interactions at the molecular level. In addition, we show the efficacy of our most potent inhibitors in activating the antitumor response using primary human immune cells from healthy donors.


■ INTRODUCTION
Cancer immunotherapy aims to stimulate the immune system's ability to fight cancer as opposed to directly killing cancer cells with more traditional methods such as chemotherapy and radiation therapy. 1−6 Inhibition of negative immune checkpoint regulators is now used in clinics around the world and was recognized with the 2018 Nobel Prize in Physiology or Medicine. 7 One of the most important immune checkpoints (ICs) in cancer immunotherapy consists of programmed cell death protein 1 (PD-1, also known as CD279) with its ligand (PD-L1, known also as CD274 or B7-H1). 4 The PD-1/PD-L1 axis is responsible for inhibiting excessive stimulation and normally aims at maintaining immune tolerance to self-antigens by negatively regulating the immune response. 8−10 By blocking the PD-1/PD-L1 interaction, activation of T cells can be achieved for an improved antitumor response. Therefore, the immune checkpoint blockade (ICB) of PD-1 or PD-L1 at the tumor cell−T cell interface has become an attractive strategy for cancer immunotherapy. 11,12 Traditionally, the inhibition of PD-1/PD-L1 interactions is achieved with monoclonal antibodies (mAbs) that target pembrolizumab,nivolumab,cemiplimab,etc.) or PD-L1 (e.g., atezolizumab, avelumab, and durvalumab). 3 Despite their medical and commercial success, mAb-based immunotherapies suffer from several disadvantages, including high treatment costs, immune-related adverse events (irAE), and poor penetration of solid tumors. 13,14 Small-molecule inhibitors (SMIs) are expected to overcome these limitations, with additional benefits such as oral bioavailability and lower production cost. 15,16 Although the PD-1/PD-L1 interface is considered difficult to treat with SMIs due to its large, flat surface of interaction with no visible binding pockets, the number of patents and publications on anti-PD-L1 SMIs is constantly growing. 17,18 A common feature of virtually all reported PD-1/PD-L1 SMI is that they target PD-L1 through a biphenyl-based scaffold originally developed by Bristol Myers Squibb (BMS). 19−21 However, none of these molecules has progressed into clinical trials so far. As of now, only the peptide-derived compound CA-170 from Curis and Aurigene is undergoing clinical trials as a small molecule that directly targets the PD-L1, PD-L2, and the V-domain Ig suppressor of T cell activation (VISTA) IC. However, it turned out that CA-170 is not a classic PD-L1-blocker and alternative modes of its action have been postulated. 22,23 Here, we present a new class of potent PD-L1 inhibitors based on a rigidified biphenyl structure. These compounds, which comprise the terphenyl scaffold, are effective in blocking PD-1/PD-L1 interactions in a variety of biophysical and cellular assays. We also provide a high-resolution structure of PD-L1 bound to the terphenyl small molecule to identify key PD-L1/ligand interactions on the PD-L1 interface.

■ RESULTS AND DISCUSSION
Identification of Potent Terphenyl-Based PD-L1 Inhibitors. Rigidification of the scaffold is a well-known strategy for enhancing the selectivity and potency of SMIs that bind to protein targets. 24−27 In our study, we focused on the effect of rigidifying the known PD-L1 inhibitors from the BMS family 19,21 ( Figure 1A). The core scaffold was extracted by comparison with several BMS molecules disclosed in patents 19,21 ( Figure 1A, BMS core). Rigidification was then implemented by removing the ether linker (highlighted in red),
In the first step of the process, the impact of the type of the C-2′-attached substituent (R 1 ), as well as the position of the hydroxymethyl group, for the binding affinity was tested (Scheme 1, Table 1). 4a molecule is modeled terphenyl core fragment presented in Figure 1 and labeled there as m-terphenyl derivative. To obtain the designed molecules, we applied a synthetic pathway based mainly on the Suzuki cross-coupling reaction. Appropriate 1,3-dibromoaryls 1a−c were coupled into their 3benzodioxane derivatives 2a−c. Subsequently, biphenyls were taken into a second cross-coupling reaction with 2, 3, or 4-(hydroxymethyl)phenylboronic acid, leading to the corresponding m-terphenyl derivatives 4a−e. C-2′ nitrile and brominated and iodinated molecules (compounds 4f−h) were synthesized by the Sandmeyer reaction, followed by aminebased diazonium salt formation. All compounds generated in this step had their potency for the PD-1/PD-L1 complex dissociation tested using the HTRF assay in the scouting mode at 5 μM since IC 50 determination would not be possible for some compounds due to their low solubility at this stage. In this assay, a lower percentage of the undissociated complex reported in Table 1 indicates higher potency of the tested compound. Full IC 50 s of several compounds were determined (Table 1) and compared with the corresponding percentages of the undissociated complex, indicating a correlation between the two measurements. This approach was used for the activity estimation of all molecules presented in this article. The HTRF assay was also validated on the BMS-1166 compound, yielding IC 50 of 3.89 nM (Table 4) as compared to 1.40 nM reported in the literature. 19,30 The results of the HTRF assay show clearly that the optimal substituent attached to C-2′ of the middle ring of terphenyl is chlorine with IC 50 of approximately 500 nM (compound 4e); however, comparable results have been obtained for brominated and cyano derivatives (compounds 4f and 4h, respectively). Introduction of a polar fragment (compound 4d) in this region has a negative impact on the molecule affinity. Generally, the insertion of the halogen atom at the C-2′ position (the R 1 substituent) is preferred and the activity of the tested fragments decreases with the growth of the halogen van der Waals radius (compounds 4e−g).
In parallel, we assessed the influence of further molecular elongation on both the aromatic and aliphatic fragments with different hydrogen bond donor−acceptor properties (Scheme 2, Table 2). Molecules 5a−r were obtained following a nucleophilic substitution reaction of the previously prepared benzyl chloride derivatives. The corresponding alkyl chlorides were synthesized by the transformation of the 4-hydroxymethylterphenyl derivatives 4a−c with SOCl 2 and a catalytic amount of anhydrous dimethylformamide (DMF). Then, they were treated with an appropriate nucleophile, yielding ethers or tertiary amines (compounds 5a−r and 6a−c). The double substitution of the amines was prevented by using an excess of the nucleophile. Furthermore, the bulkiness of the terphenyl fragment additionally precluded double substitution.
The results obtained from these steps clearly indicated that only the para elongation, using amine as a building block is beneficial for the interaction (compound 6b, Table 2). Attempted meta extension of the motif causes a significant drop of the affinity from 5% of the undissociated complex for 6b to about 56% for 6c, tested at 5 μM concentrations. Furthermore, the introduction of an aromatic fragment in each position destabilizes the interaction.
In the next stage of the structure optimization, the type of the C3-attached ether substituent in the proximal phenyl ring of the terphenyl was considered. We synthesized a series of different alkoxy derivatives (Scheme 3, Table 3) implementing different-length alkyl and cycloalkyl substituents, as well as an acetonitrile fragment (compounds 6d−e and 7a−h).
To prepare the desired compounds, we carried out Williamson alkylation of the 4-bromo-2-hydroxybenzyl alcohol, leading to the appropriate alkoxy derivative. Then, the intermediates were transformed into the corresponding phenylboronic acid pinacol esters (compounds 3a−i, SI) and coupled with 3-bromobiphenyls 2a or 2c. Hydroxymethyl groups of terphenyls were then transformed into the corresponding chlorides, which were used as an amine alkylating agent. Initially, we tested the piperazine-extended derivatives of the ethers 6d−e, but we have observed precipitation during the HTRF assay. This forced us to change the amine to N-(2-aminoethyl)acetamide, significantly increasing the solubility in the aqueous buffer for compounds 7a−h. Almost all molecules obtained in this stage of the optimization were characterized by high affinity to the PD-L1 protein and caused complete dissociation of the complex at 5 μM. Therefore, to differentiate potencies of the tested molecules, the inhibitor concentration was set to 5 nM in further assays.
Introduction of the methoxy fragment in 7a has the greatest impact on binding to PD-L1, resulting in complete complex dissociation at 5 nM ligand concentration and a measured IC 50 value of 0.82 nM. Further elongation of the ether substituent decreases the affinity to the target (compounds 7b−g). Addition of an extra biphenyl motif in compound 7h caused a drop in the inhibitory potency of the molecule. The acetonitrile derivative in molecule 7e shows the activity comparable with the O-methylated 7a molecule, causing 99% of the complex dissociation at 5 nM.
The last step of the SAR analysis was focused on the optimization of the hydrophilic tail (Scheme 3, Table 4). Maintaining all optimized molecule fragments, we synthesized and evaluated the inhibitory activities of the piperazine-(compounds 8a−e), ethylenediamine-(compounds 7a, 8f−h), and 2-aminoethanol-containing (compounds 8i−j) terphenyl derivatives.
The results showed that the type of solubilizer used has a great impact on the affinity to the targeted protein. The piperazine derivative family (compounds 8a−8e) was characterized by the lowest potencies in disrupting the PD-1/PD-L1 interaction, characterized by 88−5% undissociated complex at 5 nM ligand concentration. The best PD-L1 binder from this group is the glycine-containing 8d molecule with 5% of the remaining complex. For 8b−d, we could not precisely determine the values of the complex dissociation as they are beyond the quantification limits of the assay. Therefore, we assigned them the values below 1 nM, the level at which we can still confidently determine the assay results. The ethylenediamine derivatives (compounds 7a and 8f−h) were characterized by moderate to excellent potencies in the inhibition of the PD-1/PD-L1 interactions, with the percentage of the undissociated complex in 32−0% range at 5 nM inhibitor concentration. Determination of the half-maximal inhibitory concentration was successful only for 7a and 8g. The values of 0.82 and 2.07 nM, respectively, were obtained.
For the rest of the tested molecules, the limit of the assay was exceeded. It has to be noted that HTRF results served only the purpose of the initial selection of "leads" for further cell-based assays and as such should not be considered as a sole determinant of the compound potency in the PD1-/PD-L1 complex formation inhibition.
Terphenyl-Based Inhibitors Induce Dimerization of PD-L1. Representative molecules from the optimized group were tested for the interaction with PD-L1 using a 1 H NMR titration experiment and showed protein oligomerization upon addition of the ligand ( Figure 2).
After adding compounds 4a, 7a, 7e, and 8j to human PD-L1 (hPD-L1), we observed the broadening of the hPD-L1 proton signals in 1D NMR spectra, characteristic of oligomerization of hPD-L1, previously observed by us with the BMS compounds. 30 We additionally compared the spectra of the interaction of 8j with hPD-L1 ( Figure 2B) and murine PD-L1 (mPD-L1) ( Figure 2C). Only for human PD-L1, changes in NMR were observed upon addition of 8j to the protein. For murine PD-L1, no interaction was observed between mPD-L1 and the compound, even with the over-titration of mPD-L1 with 8j (1:10, molar ratio, respectively) as there was no change in the corresponding NMR spectra fingerprints (changes in the intensity of the peaks or their position compared to each other). The results are, therefore, consistent with our previous studies, in which we showed that the biphenyl moiety from Scheme 3. Synthetic Pathway Used during R 3 Alkoxy Fragment and R 2 Hydrophilic Tail Optimization of m-Terphenyls 6d−6e, 7a−7h, and 8a−8j a anti-PD-L1 inhibitors binds to human PD-L1 and not to the murine analogue. 31 Lead Compounds Activate Jurkat Effector Cells in the ICB Assay. After a successful optimization of the terphenyl compounds, the activity of the optimized molecules 7a and 8a−j in a cellular context was assessed. To this end, we first performed the PD-1/PD-L1 ICB assay. In this assay, T-cell receptor (TCR)-mediated activation of Jurkat effector cells (Jurkat-ECs) is reduced by the presence of PD-L1, exposed at the surface of CHO/TCRAct/PD-L1 cells ( Figure 3A). Upon the addition of PD-1/PD-L1 blockers, full activation of Jurkat-ECs is restored, as presented for a therapeutic anti-PD-L1 antibody, atezolizumab ( Figure 3B).
Most of the compounds selected for the assay (molecules 7a, and 8f−8j) were able to increase the activation of Jurkat-ECs ( Figure 3B). Only the piperazine-containing molecules do not show any changes after the treatment. The best activity, reflected by the lowest EC 50 values ( Figure 3E), was observed for 8j, followed by 8h, 8i, 7a, and 8f. These compounds provided higher activation of the Jurkat-ECs than the previously published BMS-1166 compound ( Figure 3B). All active compounds were comparable in terms of toxicity toward Jurkat-ECs, with compounds 7a, 8f, and 8j slightly less cytotoxic than 8g, 8h, and 8i ( Figure 3C,D). Importantly, none of the tested compounds was able to increase the activity of Jurkat-ECs where PD-L1 was absent ( Figure S1). This provides evidence that the activity of the compounds is PD-L1-dependent and is not related to the unspecific activation of T cells. Compounds 8j, 8h, and 8i ( Figure 3G) showed the highest selectivity of the desired PD-1/PD-L1-blocking activity versus cytotoxicity ( Figure 3F). Compounds 8h and 8j, which, apart from their bioactivity, showed dissociation of the protein complex in HTRF at 5 nM, were directed for further study.
Lead Compound 8j Activates Primary T Cells in the T-Cell Activation Assay. A standard ICB assay relies on artificial Jurkat-ECs, which are surrogates of T-cells. To verify the potential of the compounds to reactivate the PD-L1blocked primary T cells, a modified assay termed T-cell activation (TCA) was carried out. 32 In this assay, human primary peripheral blood mononuclear cells (PBMCs) are allowed to make contact with CHO/TCRAct/PD-L1 cells alone or in the presence of PD-1/PD-L1 blockers, and the activation of CD4 + and CD8 + T cell is monitored by flow cytometry.
Following the addition of PBMCs to the CHO/TCRAct/ PD-L1 cells, the activation of both CD4 + and CD8 + T cells was observed, as monitored by the expression of early (CD69), intermediate (CD25 and HLA-DR), and late (PD-1) activation/exhaustion markers ( Figure 4A). This is due to the TCR-dependent activation of T-cells by CHO/TCRAct/ PD-L1 cells. As reported before, the addition of PD-1/PD-L1blockers in such a setup increases the expression of PD-1 on T- Table 3. Inhibitory Activities of the Optimized Terphenyls Obtained in HTRF Assay in the Second Step of Optimization of the R 3 Alkoxyl Substituent cells but rarely other markers. 32 In the current experiment, this was confirmed for three therapeutic anti-PD-L1 antibodies, durvalumab, atezolizumab, and avelumab ( Figure 4A). Also, for the tested compounds 8h and 8j, a significant increase in the numbers of the PD-1-positive cells was observed ( Figure  4A). Surprisingly, for the antibodies, a decrease of CD69 and CD25 expression was observed for CD4 + cells, which has not been observed before in our studies. The addition of compound 8j, but neither 8h nor the therapeutic antibodies, increased the numbers of activated, CD25-positive, and HLA-DR-positive T-cells ( Figure 4A). Altogether, the presented data clearly show the biological activity of 8j in a cellular context.
In addition to the expression of surface markers, the accumulation of immune-related cytokines in the culture media collected in the TCA assay was monitored with the cytometric bead array (CBA) human Th1/Th2/Th17 kit. In the presence of atezolizumab, significantly increased levels of IL-17A, TNF-α, and IL-10 were observed ( Figure 4B). Also, the levels of IL-6, IFN-γ, and IL-2 increased, but these increases were not statistically significant. Similarly, the treatment of PBMCs cocultured with the CHO/TCR-Act/ PD-L1 cells with 8j also increased the accumulation of all six cytokines, with significant increases observed for IL-6, TNF-α, IFN-γ, IL-10, and IL-2 ( Figure 4B).
Distal Terphenyl Ring Provides Rigidification That Fits into the PD-L1 Cleft. We were able to crystalize and solve the structure of the 8g compound in complex with PD-L1 at the resolution of 2.3 Å that allowed us to decipher the molecular basis of the interactions between them ( Figure 5 and Table S2). The asymmetric unit of the hPD-L1/8g complex contains six protein molecules organized into three dimers, with one inhibitor located at the center of the interface of each dimer. The electron density map describes each inhibitor molecule allowing the position of all moieties in the structure with the fully detailed density for the molecule located at the interface of the AB dimer. 8g is mostly buried in the deep and elongated tunnel, which is composed of two PD-L1 monomers. The 1,4-benzodioxane group creates the standard interaction with A Tyr56 (the monomer molecules are annotated by Limit of quantification of the used assay was achieved. subscripts A and B according to their chain arrangement in the crystal structure of the dimer), previously observed by us for the BMS-202, BMS-1001, and BMS-1166 compounds. 30,33 The central terphenyl moiety is stabilized by plenty of hydrophobic contacts with B Tyr56, A Met115, B Met115, A Ala121, and B Ala121. The chlorine substituent attached to this moiety forms an additional halogen bond with the backbone carbonyl of A Asp122, providing the rationale for the enhanced inhibitory effect of these halogenated inhibitors at the R 1 position. The methoxy substituent at the meta position of the distal ring of the terphenyl is also additionally stabilized by hydrophobic interaction with B Ile54 and B Val68, pointing out the importance of this substituent. At the other side of the tunnel, the sulfonamide polar extension forms a hydrogen bond with the carbonyl of the A Asp122 side chain and is largely water-exposed.
Our structural data confirm that the terphenyl compounds with the rigidified biphenyl unit bind to PD-L1 at the interaction site that has been seen in the compounds with the biphenyl scaffold. Comparison of the PD-L1/8g, PD-L1/BMS-1166, PD-L1/the C2-symmetric inhibitor, and PD-L1/ARB-272572 complexes shows that the crystallographic arrangement of the terphenyl core is close to the docking results ( Figure S2). 33−35 ■ CONCLUSIONS In this paper, we have demonstrated a new group of smallmolecule anti-PD-L1 inhibitors based on the concept of rigidifying the biphenyl unit by removing the ether linker from the known PD-L1 inhibitors of the BMS family. Such terphenyl-based scaffolds exhibit high inhibitory activity against the PD-1/PD-L1 complex both in common biophysical assays, such as HTRF and NMR, as well as cell-based assays. Our most potent compounds activated Jurkat-ECs to levels comparable to those of the control antibodies, despite being considerably smaller. Similar to other anti-PD-L1 inhibitors reported in the literature, our compounds dimerize human PD-L1, as presented here by NMR and X-ray crystallography studies. They are specific for human PD-L1 and do not crossreact with the murine PD-L1 analogue. Our most potent compound, 8j, is characterized by an exceptionally high affinity to the molecular target (IC 50 < 1 nM), as well as a proven potency in cell-based assays with an EC 50 of ca. 1 μM. Most importantly, adding compound 8j to the coculture of PBMCs and CHO/TCRAct/PD-L1 cells not only induced PD-1 expression but also increased the numbers of CD25-positive and HLA-DR-positive T cells, which was not observed even in PD-L1 therapeutic antibodies. Also, the accumulation of immune-related cytokines was increased in the presence of either 8j or atezolizumab. This shows the potency of this compound toward primary human T cells and designates 8j as an outstanding candidate for further optimization in the anti-PD-L1 treatments of cancer. The presented crystal structure provides an explanation for the enhanced inhibitory effect of terphenyl-based inhibitors. The introduced phenyl ring replaces the benzyl linker of BMS-1166 and thus provides the rigidification of the structures in the conformation that fits perfectly into the PD-L1 cleft.

■ EXPERIMENTAL PROCEDURES
Homogeneous Time-Resolved Fluorescence. HTRF was performed using the certified CisBio assay according to the  , and late (PD-1) T-cell activation/exhaustion markers was determined by flow cytometry. PBMCs from healthy donors were exposed to CHO/TCRAct/PD-L1 cells for 2 days in the presence of durvalumab (durva), atezolizumab (atezo), avelumab (avelu), or increasing concentrations of 8j and 8h (concentrations: 0.1, 0.5, or 2.5 μM). ø indicates untreated cells. The graphs show fractions of positive cells and represents cumulative data from four donors. Statistical significance was analyzed with one-way ANOVA, followed by Fisher's posthoc test: *p < 0.05, **p < 0.01, ***p < 0.01. (B) Accumulation of the indicated cytokines in culture media from the TCA assay. The graphs show mean ± standard error of the mean values from four independent donors. For each donor, the data were normalized to untreated coculture of PBMCs and CHO/TCRAct cells as a determinant of donor-dependent cytokine level independent of the PD-L1/PD-1 axis. Statistical significance was analyzed with one-way ANOVA, followed by the Fisher posthoc test: *p < 0.05, **p < 0.01, ***p < 0.001.
Protein Nuclear Magnetic Resonance. NMR spectra were recorded in PBS pH 7.4 containing 10% (v/v) of D 2 O added to the samples to provide the lock signal. Water suppression was carried out using the WATERGATE sequence. All the spectra were recorded at 300 K using a Bruker AVANCE 600 MHz spectrometer with the cryoplatform. The spectra were recorded at the ligand/protein ratio of 1:1, unless stated otherwise. The samples were prepared by adding small amounts of a 50 mM ligand stock solution in DMSO to the protein solution (0.20 mL) of PD-L1 fragment at a concentration of 0.2 mM. Spectra were visualized using TopSpin 4.0.2.
In Silico Molecular Docking. For molecular docking, we generated structures of BMS core and m-terphenyl derivative in the PDB format using Chem3D 17.1 and used AutoDock Vina1.1.2 integrated into PyRx with the dimeric structure of PD-L1 bound to BMS-1166 (PDB ID: 6R3K).
Protein Expression and Purification. Escherichia coli strain BL21 was transformed with a pET-21b plasmid carrying the PD-L1 gene (amino acids 18−134). The bacteria were cultured in LB at 37°C until OD 600nm of 0.6 when the recombinant protein production was induced with 1 mM IPTG. The protein production was continued overnight. Inclusion bodies were collected by centrifugation, washed twice with 50 mM Tris-HCl pH 8.0 containing 200 mM NaCl, 10 mM ethylenediaminetetraacetic acid, 10 mM 2-mercaptoethanol, and 0.5% Triton X-100, followed by a single wash with the same buffer but with no Triton X-100. The washed inclusion bodies were resuspended overnight in 50 mM Tris-HCl pH 8.0, 6 M GuHCl,200 mM NaCl, and 10 mM 2-mercaptoethanol and clarified with centrifugation. Refolding of PD-L1 was performed by dropwise dilution into 0.1 M Tris-HCl pH 8.0 containing 1 M L-arginine hydrochloride, 0.25 mM oxidized glutathione, and 0.25 mM reduced glutathione. The refolded protein was dialyzed 3 times against 10 mM Tris-HCl pH 8.0 containing 20 mM NaCl and purified by size exclusion chromatography using Superdex 75 and dialysis buffer. The quality of the refolded protein was evaluated by sodium dodecyl sulphate−polyacrylamide gel electrophoresis and NMR.
PD-L1 Cocrystallization, Determination, and Refinement. Purified PD-L1 in 10 mM Tris-HCl pH 8.0, 20 mM NaCl buffer was concentrated to 5 mg/mL, mixed with the 8g inhibitor in 1:3 molar ratio (protein/compound) and clarified. The supernatant was used for screening using commercially available buffer sets. Diffraction-quality crystals were obtained at room temperature from the condition containing 0.1 M sodium cocodylate pH 6.5, 25% PEG 4000. The crystal was flash-cooled in liquid nitrogen. The X-ray diffraction data were collected at the BL14.1 beamline operated by the Helmholtz-Zentrum Berlin (HZB) at the BESSY II (Berlin Adlershof, Germany). 36 The data were indexed, integrated, and scaled using XDS, XSCALE, and Aimless. 37−39 The initial phases were obtained by molecular replacement calculated in Phaser. 40 The model building was performed in Coot and refinement was performed using Phenix or PDB-REDO server. 41−43 Water molecules were added automatically and inspected manually. The coordinates and structure factors were deposited in the Protein Data Bank with accession number 7NLD.
Cell Culture. CHO-K1 cells, CHO-K1 cells overexpressing TCR activator (CHO/TCRAct), PD-L1-expressing CHO/TCRAct cells (CHO/TCRAct/PD-L1), and Jurkat-EC T-cells overexpressing PD-1 and carrying a luciferase reporter gene under the control of the nuclear factor of activated T-cell response element (NFAT-RE) were obtained from Promega and cultured in RPMI-1640 medium (Biowest) supplemented with 10% fetal bovine serum (FBS, Biowest) and 200 mM L-glutamine (Biowest). G418 (250 μg/mL, InvivoGen) and Hygromycin B Gold (50 μg/mL, InvivoGen) were also added to the culture medium as selection antibiotics. The overexpression of PD-L1 and TCR ligand in CHO/TCRAct/PD-L1 cells and overexpression of PD-1 in Jurkat-ECs were confirmed by flow cytometry and western blot analysis. PCR tests for Mycoplasma sp. contamination 44 were routinely performed, which indicated negative results for both cell lines.
Cytotoxicity Assay. The cytotoxicity of compounds was verified toward Jurkat-ECs since these cells provide the readout in the ICB assay and present a much higher degree of susceptibility to the compounds' toxicity than CHO-K1 cells (not shown). Jurkat-ECs were seeded on 96-well transparent plates in the presence of increasing concentrations of the compounds with DMSO-treated cells as a control. The concentration of DMSO was kept constant (0.1%). After 48 h of incubation, a tetrazolium reagent, Biolog redox dye MIX MB (Biolog), was added according to the manufacturer's protocol, and the culture plates were incubated for an additional 2 h (37°C, 5% CO 2 ). The absorbance was measured using a Spark microplate reader (Tecan) at 590 nm with 750 nm as a reference. The data are presented as Jurkat-EC survival relative to DMSO-treated control cells. Data points represent mean ± SD values from three independent experiments. Half-maximal effective concentrations (EC 50 values) were calculated from the dose−response curve using OriginPro 2020 (OriginLab) software.
PD-1/PD-L1 ICB Assay. The functional assay of the blockade of PD-1/PD-L1 interaction in vitro (PD-1/PD-L1 Blockade Bioassay, Promega) was performed according to the manufacturer's protocol. CHO/TCRAct/PD-L1 cells or CHO/TCRAct or CHO-K1 cells were seeded on 96-well white plates at the density of 10,000 cells/well and the next day overlaid with Jurkat T cells (20,000 cells/well) in the presence of increasing concentrations of the compounds with DMSO only as a control (the concentration of DMSO was kept constant at 0.1%) or atezolizumab, as a positive control, anti-PD-L1 monoclonal antibody (Selleckchem). The 2.5-fold dilutions of atezolizumab were prepared in the assay buffer (RPMI 1640 + 1% FBS) on the day of the assay. Activation of the Jurkat T cells, reflected by luciferase activity, was monitored by luminescence measurements after 6 h of incubation (37°C, 5% CO 2 ), and 20 min of additional incubation with the Bio-Glo assay reagent (Promega) at room temperature. The luminescence was read on a Spark microplate reader (Tecan). The data are presented as fold induction of the luminescence signal relative to either untreated (for atezolizumab) or DMSO-treated (for compound) cells. Data points represent mean ± SD values from three to five independent experiments. Half-maximal effective concentrations (EC 50 values) were calculated from the Hill curve fitting to the experimental data using Origin Pro 2020 software (OriginLab).
Isolation of PBMCs and TCA Assay. Anticoagulant citrate dextrose-A-treated blood from healthy donors was purchased from the Regional Center of Blood Donation and Blood Therapy in Krakow, Poland. PBMCs were isolated from whole blood by density gradient centrifugation using Pancoll human separating solution (PAN-Biotech GmbH, Aidenbach, Germany). The separated cells were washed and resuspended in RPMI 1640 medium (Biowest) containing 10% FBS (Biowest). PBMCs were added at an amount of 3 × 10 5 to CHO, CHO/TCRAct, or CHO/TCRAct/PD-L1 cells preseeded on 24-well plates in the presence of either anti-PD-L1 mAbs (at the final concentration 5 μg/mL): durvalumab (Selleckchem), atezolizumab (Selleckchem), or avelumab (MedChemExpress) or tested com- pounds at the concentrations 0.1, 0.5, and 2.5 μM. Untreated cells were used as controls for antibody treatment, while DMSO-treated cells [0.1% (v/v)] were used as controls for compound treatment. The cells were incubated for 48 h (37°C, 5% CO 2 ) and then detached from the plates with TrypLe Select Enzyme (Thermo Fisher). PBMCs were stained for 20 min at room temperature with the antibodies: anti-CD4-FITC, anti-CD8-BV510, anti-CD69-APC, anti-CD25-PE, anti-HLA-DR-PerCP, and anti-PD1-PECy7 (Becton Dickinson Biosciences, BD). Following two washing steps, the cells were analyzed with a FACSCanto II cytometer. Data analysis was carried out with FlowJo software, followed by statistical significance calculations done with Origin Pro 2020 software (OriginLab). Statistical significance was analyzed with one-way analysis of variance (ANOVA), followed by the Fisher's posthoc test: *p < 0.05, **p < 0.01, ***p < 0.01.
Analysis of Cytokine Accumulation. The concentration of the cytokines (IL-2, IL-4, IL-6, IL-10, IL-17A, IFN-γ, and TNFα) in supernatants was determined using the CBA Human Th1/Th2/Th17 kit (BD Biosciences), with data acquisition using the FACSCanto flow cytometer. The CBA beads were discriminated in FL-4 and FL-5 channels, while the concentration of specified cytokines was determined by the intensity of FL-2 fluorescence, using the respective standard reference curve and FCAP Array software (BD Biosciences). IL-4 was below the detection level and is not presented in the manuscript. Statistical significance was analyzed with one-way ANOVA using the Origin Pro 2020 software, followed by the Fisher's posthoc test: *p < 0.05, **p < 0.01, ***p < 0.01.
Syntheses. All syntheses were performed according to the procedures summarized in Schemes 1-3 and as described in the Supporting Information. Reagents were obtained from commercial suppliers (Sigma-Aldrich, ABCR, Acros Organics, Appollo Scientific, AK Scientific) and used without further purification unless otherwise noted. The anhydrous solvents were purchased from Sigma-Aldrich or Alfa Aesar. Nuclear magnetic resonance spectra were recorded at 300K on a Bruker AVANCE 600 spectrometer { 1 H NMR (600 MHz) and 13 C NMR (151 MHz)}. Chemical shifts for 1 H NMR and 13 C NMR were reported in parts per million (δ) referenced to the appropriate deuterated solvent. Coupling constants were reported in hertz (Hz). The following abbreviations were used for spin multiplicity: s = singlet, s broad = broad singlet, d = doublet, t = triplet, q = quartet, quin = quintet, dd = double of doublets, ddd = double doublet of doublets, and m = multiplet. IR absorption spectra were recorded on a Nicolet IR200 spectrometer using the attenuated total reflection (ATR) technique. Thin-layer chromatography (TLC) was performed on Sigma-Aldrich precoated silica gel plates (0.20 mm thick, particle size 25 mm). Spots were visualized by UV light at 254 and 365 nm. Column chromatography was performed using silica gel 60 (40−63 mm, Sigma-Aldrich). Flash chromatography was performed on a Reveleris X2 flash chromatography, using Grace Reveleris silica flash cartridges. The liquid chromatography−mass spectrometry (LC−MS) measurements were performed on an ultraperformance liquid chromatography (UPLC)−tandem mass spectrometry system consisting of a Waters ACQUITY UPLC (Waters Corporation, Milford, MA, USA) coupled to a Waters TQD mass spectrometer (electrospray ionization mode ESI-tandem quadrupole). Chromatographic separations were carried out using the Acquity UPLC BEH (bridged ethyl hybrid) C 18 column, 2.1 × 100 mm and 1.7 μm particle size, equipped with an Acquity UPLC BEH C18 VanGuard precolumn; 2.1 × 5 mm and 1.7 μm particle size. The column was maintained at 40°C and eluted under gradient conditions using 95−0% of eluent A over 10 min at a flow rate of 0.3 mL min −1 . Eluent A: water/formic acid (0.1%, v/v); eluent B: acetonitrile/formic acid (0.1%, v/v). The purity of all final compounds determined using chromatographic LC−MS was >95%. HRMS was carried out by the Laboratory for Forensic Chemistry Faculty of Chemistry, Jagiellonian University, with a microOTOF-QII (Bruker) spectrometer using the ESI ionization technique.
Product 5b was obtained as a colorless solid with 88.8% (0.080 g) yield.
Method A. Crude alkyl chloride was dissolved in anhydrous acetonitrile or THF, and excess of appropriate amine dissolved in anhydrous acetonitrile or THF was added (if the hydrochloride of amine was used, an equimolar amount of triethylamine was added to the reaction mixture). The reaction mixture was stirred at room temperature for 48 h and controlled by TLC. After completion of the reaction, the mixture was poured into saturated NaHCO 3 and extracted with ethyl acetate. The organic phases were collected, dried over anhydrous MgSO 4 , and evaporated. The crude products were purified by flash column chromatography (SiO 2 , hexane/ethyl acetate).
Method B. Crude alkyl chloride was dissolved in anhydrous THF, and excess of Boc-protected piperazine dissolved in anhydrous THF was added. The reaction mixture was stirred at room temperature 48 h and controlled by TLC. After the reaction was complete, the solvent was evaporated and the crude Boc-protected product was purified by flash chromatography (SiO 2 , hexane/ethyl acetate 1:1). The purified Boc-protected product was dissolved in anhydrous dioxane, and HCl in dioxane was added (10 equiv, 4 M). The reaction mixture was stirred at room temperature overnight and then poured into saturated NaHCO 3 and extracted with ethyl acetate. The organic phases were collected, dried over anhydrous MgSO 4 , and evaporated. The products were purified by column chromatography (SiO 2 , pure methanol, or methanol/7 M NH 3 in methanol).
Method C. Crude alkyl chloride was dissolved in anhydrous THF and an appropriate amine dissolved in anhydrous THF was added (if the hydrochloride of amine was used, an equimolar amount of triethylamine was added to the reaction mixture). The reaction mixture was stirred at 50°C overnight. After completion of the reaction, it was poured into saturated NaHCO 3 and extracted with ethyl acetate. The organic phases were collected, dried over anhydrous MgSO 4 , and evaporated. The crude products were purified by column chromatography (SiO 2 , DCM/methanol, methanol, or methanol/7 M NH 3 in methanol).
Method D. Crude alkyl chloride was dissolved in anhydrous DMF, and an appropriate amine dissolved in anhydrous DMF was added (if the hydrochloride of the amine was used, equimolar amount of triethyl amine was added to the reaction mixture) together with diisopropylethylamine (DIPEA). The reaction mixture was stirred at 80°C overnight. After completion of the reaction, it was poured into saturated NaHCO 3 and extracted with ethyl acetate. The organic phases were collected, dried over anhydrous MgSO 4 , and evaporated. The crude products were purified by column chromatography (SiO 2 , ethyl acetate, DCM/methanol, or CHCl 3 /7 M NH 3 in methanol or methanol/7 M NH 3 in methanol).
Method E. Crude alkyl chloride was dissolved in anhydrous DMF, and an appropriate amine dissolved in anhydrous DMF was added (if the hydrochloride of the amine was used, an equimolar amount of triethylamine was added to the reaction mixture) together with DIPEA. The reaction mixture was stirred at 80°C overnight. After completion of the reaction, it was poured into saturated NaHCO 3 , extracted three times with ethyl acetate, and the crude Boc-protected product was purified by flash chromatography (SiO 2 , hexane/ethyl acetate 1:1). The purified Boc-protected product was dissolved in anhydrous dioxane, and HCl in dioxane was added (10 equiv, 4 M). The reaction mixture was stirred at room temperature overnight, then poured into saturated NaHCO 3 , and extracted three times with ethyl acetate. The organic phases were collected, dried over anhydrous MgSO 4 , and evaporated. The products were purified by column chromatography (SiO 2 , DCM/methanol, or methanol/7 M NH 3 in methanol).