A human scFv phage display library (AbCheck, Czech Republic) was used to identify phage antibodies that bound recombinant human PD-L1-Fc protein. Phage that bound to human Fc, CD80 and CD86 were depleted from the libraries by pre-incubation steps throughout the panning process. In some cases, libraries were heated to 65 °C for 15 min prior to the panning step to select for heat-stable scFv. Enrichment of PD-L1 specific scFv was tested with bacterial extracts containing soluble scFv in ELISA. Panned phages were screened for the presence of scFv that blocked the interaction of PD-L1 with both PD-1 and CD80. Clone ABC110 (LY3300054) was selected from a large number of functional hits based on binding, blocking, and in-vitro functional properties, and its DNA sequence was cloned into a human expression vector with an IgG1 effector-null backbone (IgG1-EN), containing the following residue changes; L234A, L235E, G237A, A330S, and P331S (11520463), and CHO cells that stably expressed LY3300054 were established. LY3300054 IgG was purified from the culture supernatant by protein A affinity chromatography (Poros A, Applied Biosystems, Foster City, CA). By flow cytometry LY3300054 was shown to specifically bind to the surface of the PD-L1-positive (H292, HCC827) but not the PD-L1–negative A204 cell lines (Additional file 1: Figure S1).
Protein expression and purification
The extracellular domain (ECD) of human PD-L1 was cloned into an Fc (human IgG1) construct (GS vector) that contained a Factor Xa cleavage site at the N-terminus of the hinge region. Human PD-L1-Fc was expressed in human 293-Freestyle cells (Invitrogen Corp., Carlsbad, CA) that were cultivated and transfected according to manufacturer’s specifications. Human PD-L1-Fc was purified via standard ProA affinity columns; human PD-L1 monomer was cleaved from the purified Fc construct with Factor Xa enzyme. Cleaved Fc and undigested PD-L1-Fc were purified out of the sample via standard ProA affinity column. Purified proteins were buffer exchanged into PBS, quantified and evaluated by SDS-PAGE and analytical SEC analysis to confirm structural integrity. Canine PD-L1-Fc and its mutants were expressed transiently in Expi293F cells following transfection using ExpiFetamine 293. The canine PD-L1-Fc and its mutants in addition to the cynomolgus, murine and rat PD-L1-Fc were generated in a manner similar to that of the human PD-L1-Fc.
ELISA binding assays
Binding to recombinant PD-L1
Ninety-six-well plate (Immulon 2HB) was coated with 100 ng of human PD-L1-Fc, murine PD-L1-Fc, or cynomolgus PD-L1-Fc (R&D Systems, Minneapolis, MN) overnight at 4 °C. Wells were blocked for 2 h with blocking buffer (PBS containing 5% nonfat dry milk) and then washed three times with PBS containing 0.1% Tween-20. 100 μl of serially diluted anti-PD-L1 antibody or control IgG was then added and incubated at room temperature for 2 h. After washing, the plate was incubated with goat anti-human IgG F(ab’)2-HRP conjugate (Jackson ImmunoResearch, West Grove, PA) at room temperature for 1 h. The plates were washed and then incubated with 3, 3′, 5, 5′-tetramethylbenzidine. The absorbance at 450 nm was read on a microplate reader. The half maximal effective concentration (EC50) was calculated using GraphPad prism software.
Binding to canine PD-L1 variants
Ninety-six well Immulon 4HBX ELISA plate was coated overnight with 50 ng each of the wild type and mutant canine PD-L1-ECD-Fc in 100 μl of PBS, pH 7.2 with mild agitation at 4 °C. After blocking and wash, a five-fold dilution series (0.0017–133 nM) of LY3300054 was added in duplicate and incubated with mild agitation for 1 h at room temperature. The wells were washed and a 1:10,000 dilution of HRP-conjugated goat anti-Fab antibody (Jackson ImmunoResearch) was added and incubated at room temperature following standard protocol. TMB peroxidase chromogenic substrate and stop solution were used according to manufacturer’s instruction for visualization and detection of signals. Absorbance readings were plotted in GraphPad Prism software. EC50 values were calculated by nonlinear regression curve fit analysis of the software’s One Site-Specific Binding function.
ELISA blocking assays on PD-L1 interaction with PD-1 or CD80
Serially diluted LY3300054 or control IgG were mixed with the equal volume of a fixed concentration of biotinylated PD-L1-Fc (100 ng/mL for PD-1 blocking and 500 ng/mL for CD80 blocking), and then incubated at room temperature for 1 h. 100 μl of the mixture was transferred to 96-well plates pre-coated with human PD-1-Fc or with human CD80-Fc at 100 ng/well (R&D Systems) and then incubated at room temperature for an additional 1 h. After washing, Streptavidin-HRP conjugate was added, and absorbance at 450 nm was read. IC50 represents the antibody concentration required for 50% inhibition of PD-L1 binding to PD-1 or to CD80.
SPR binding to recombinant human, murine or cynomolgus PD-L1
Surface plasmon resonance (SPR) (Biacore T200, GE Healthcare) was used to determine the binding kinetics of LY3300054 to human, cynomolgus, murine and rat PD-L1-Fc at 37 °C. Approximately 40 response units (RU) of LY3300054 were immobilized onto a CM5 chip using the standard amine coupling procedure. HBS-EP buffer (10 mM HEPES, 150 mM NaCl, 3 mM EDTA, and 0.005% surfactant p20) was utilized as a running buffer during binding kinetic measurements. The PD-L1-Fc gradients were comprised of seven 3× dilutions. Starting concentrations were 9 nM for the human and cynomolgus PD-L1-Fc gradients and were 90 nM for the mouse and rat PD-L1-Fc. PD-L1-Fc proteins were injected for 180 s (contact time) over the immobilized LY3300054 at a flow rate of 30 μl/min. The dissociation times for those measurements were 1500 s for the four top concentrations of the gradient and 240 s for the rest of the gradient. After dissociation, regeneration of the LY3300054 surface was achieved with a single 18 s injection of 0.75 M NaCl/25 mM NaOH at 30 μl/min followed by a 30 s wash with HBS-EP to stabilize the surface. Biacore T200 Evaluation Software (version 1.0) was used to analyze the results from the kinetic experiments. After double referencing to remove artifacts from nonspecific binding, simultaneous global fitting of the data for each concentration gradient to a 1:1 L model was performed to determine the association rate (kon), dissociation rate (koff), and dissociation constant (KD = koff/kon). At least four different concentration gradients were used to compute the kinetic parameters and their corresponding sample standard deviation.
In vitro functional assays
PD-1 reporter assay
PD-L1+ aAPC/CHO-K1 (Promega) or PD-L1− aAPC/CHO-K1 (Promega part# CS187110) human T-activator cells were plated in a 96-well white opaque tissue culture plate at 40,000 cells per well in 100 μl of medium (10% FBS F-12, 0.2 mg/ml Hygromycin-B and 0.2 mg/ml G418) and incubated overnight at 37 °C at 5% CO2. Medium was removed from the assay plate the following day and serially diluted test and control antibodies were added at 40 μl per well in the assay buffer. GloResponse NFAT-luc2/PD1 Jurkat cells (Promega) were re-suspended in assay buffer at a concentration of 1.25 × 106 /ml and added to the plate at 40 μl per well. After 6 h of co-culture, assay plates were removed from the incubator and equilibrated at room temperature for 5 min. Bio-Glo™ Reagent (Promega) was prepared according to manufacturer’s instructions and added to each well at 80 μl per well. Plates were then incubated for 5 min at room temperature. Luminescence was measured in a plate reader and data was analyzed using GraphPad Prism software .
Mixed leukocyte reaction (MLR)
CD14+ monocytes were isolated from frozen human peripheral blood mononuclear cells (PBMC) obtained from a healthy donor (AllCells, Alameda, CA) with Human Monocyte Isolation Kit II (Miltenyi, Auburn, CA). Immature dendritic cells (DCs) were generated by culturing these monocytes in complete RPMI-1640 medium containing 10% FBS in the presence of 1000 IU/ml hGM-CSF and 500 IU/ml hIL-4 for 4 days. CD4+ T cells were purified from fresh human PBMC of a different healthy donor (AllCells) using Human CD4+ T Cell Isolation Kit (Miltenyi). The two types of cells were then mixed in 96-well V-bottom plates with 5 × 104 CD4+ T cells and 5 × 103 immature DC in 100 μl of complete AIM-V medium per well. 100 μl of 2× serially diluted LY3300054 or human IgG1 was added into a well of the plates. LY3300054 was also tested in combination with anti-CTLA4 antibody (Ipilimumab) at equimolar concentrations ranging from 0.003 to 67 nM. After incubation for 72 h at 37 °C at 5% CO2, supernatants and cell pellets were harvested and subjected to immunoassay (human IFN-γ ELISA (R&D Systems) or 41-plex Milliplex MAP Human Cytokine/Chemokine Immunoassay Panel (Millipore, Burlington, MA) (analytes are listed in Additional file 2) and a custom-made Quantigene Plex gene expression analysis (see below). MLR studies of LY3300054 were repeated with at least four different CD4 T cell donors.
Antigen recall assay
Frozen PBMCs were thawed, cultured in 10% FBS RPMI overnight at 37 °C at 5% CO2, and seeded in a 96-well flat bottom tissue culture plate at 1 × 105 cells per well in 100 μl of 10% FBS/RPMI-1640. Antibodies were prepared at 4× concentrations and added to the cells at 50 μl per well. After 1-h incubation, Tetanus Toxoid (50uL; 0.8μg/ml) (TT; #191A LIST Biological Laboratories Inc.) was added to wells with LY3300054 or medium control. After 5 days in culture, supernatant was collected and an IFNγ ELISA (R&D Systems SIF50) was performed according to manufacturer’s instructions.
Effector function assays
Antigen-dependent cell-mediated cytotoxicity (ADCC) assay
The ability of LY3300054 to mediate ADCC was tested in a Jurkat-FcγRIIIa reporter gene assay using a PD-L1+ HEL cell line (ATCC TIB-180) as previously described . Anti-CD20 antibody rituximab (wild type IgG1) was tested as a positive control in the same assay against the CD20-positive WIL2-S cell line. Briefly, 1 × 104 target cells at 50 μl and serially diluted antibodies at 4× concentrations at 25 μl were added per well. Jurkat-FcγRIIIa (V158) cells were added as effector cells at the effector/target ratio of 15:1 at 25ul/well, and followed by 6 h incubation in a humidified 37 °C incubator. Plates were removed and equilibrated to room temperature for 5 min. Luciferase reagent was added at 100 μl/well and luminescence was detected.
Complement dependent cytotoxicity (CDC) assay
LY3300054 was tested using the PD-L1+ HEL cells as targets. Rituximab was used as a positive control against WIL2-S cell line in the same experiment. Target cells were treated with 1:3 titrations of the various antibodies and incubated for 30 min at 37 °C. Human complement was added into the assay plates and incubated for 1 h at 37 °C. Alamar Blue reagent was then added to the wells and incubated for an additional 24 h at 37 °C before fluorescence was determined, as an indication of cell viability.
PBMC cytokine release assay
Fresh unstimulated human PBMC isolated from six healthy donors were incubated with plate bound LY3300054 antibody or control antibodies for 24 h, pre-coated over a broad titration range from 0.003 to 100 μg/ml. Anti-CD3 antibody OKT3 (eBioscience, San Diego, CA) was used as a positive control. Using a commercially available multiplex assay based on the Luminex platform (Luminex Corporation, Austin, TX), 21 cytokines including Fractalkine, GM-CSF, IFNγ, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-10, IL-12 (p70), IL-13, IL-17A, IL-21, IL-23, ITAC, MIP-1α, MIP-1β, MIP-3α, and TNF-α were measured in cell culture supernatants .
PD-L1 and HLA class I staining of human tumor lines
NCI-H292, HCC827, OV79, and A204 (ATCC) tumor cells were cultured for approximatelly36 hours prior to non-enzymatic harvest. NCI-H292, HCC827, and A204 cells were stained for PD-L1 using FITC-conjugated anti-human PD-L1 commercial antibody (clone MIH1, BD Biosciences), Alexa Fluor® 488-conjugated LY3300054, or appropriate isotype controls. NCI-H292, HCC827, and OV79 cells were stained separately for HLA Class I expression using an APC-conjugated antibody (clone W6/32, RnDSystems, Minneapolis, MN) Samples were collected on a 5-laser Fortessa X-20 cytometer (BD Biosciences) and analyzed with FlowJo V10 software (TreeStar).
In vivo models
All animal studies were approved by the Institutional Animal Care and Use Committee and performed in accordance with current regulations and standards of the United States Department of Agriculture and the National Institute of Health. All experiments with adoptively transferred human PBMC or expanded human T cells utilized NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ (NSG) animals (6–7 weeks of age, female, from Jackson Laboratories, Bar Harbor, MN), and were maintained in a 12 h light/dark cycle facility under pathogen-free conditions in microisolator cages with standard laboratory chow and water ad libitum. Cord blood-derived CD34+ hematopoietic stem cell (HSC) engrafted mice used for the OV79 model utilized NOD.Cg-Prkdcscid Il2rgtm1Sug/JicTac animals (NOG, 15–17 weeks of age, female) and were obtained from Taconic BioSciences (Rensselaer, NY). Fetal liver derived CD34+ HSC transplanted mice used for the HCC827 model in NSG background (15–17 weeks of age, female) were obtained from Jackson Laboratories. Animal well-being and behavior, including grooming and ambulation were monitored at least twice per week. Body weight and tumor volumes were measured twice a week starting 1–2 weeks post implantation. Tumor volumes were calculated according to formula (vol = π/6 * l * w2) and plotted as geometric means ± standard error of the mean (SEM). Statistical analysis of tumor volume data was performed by two-way ANOVA on repeated measurements.
Co-implantation of human NCI-H292 tumor cells and human PBMC (Winn model)
Freshly isolated human PBMCs were combined with freshly cultured NCI-H292 tumor cells (ATCC, Manassas, VA) at a 1:4 E:T ratio and co-implanted subcutaneously into the flanks of female NSG mice (groups of 8 mice per treatment arm). One day later, weekly intraperitoneal (IP) treatments of either human IgG1 or LY3300054 at 10 mg/kg began and continued for a total of four doses. Tumor growth was monitored by caliper measurements.
Established HCC827 xenograft tumor model with infused human T cells
Mice were implanted subcutaneously into the flanks of female NSG mice with 10 × 106 freshly cultured HCC827 tumor cells (ATCC). When tumors reached volumes of ~ 300 mm3 (~ 4–5 weeks), 2.5 × 106 expanded human T cells were administered intravenously (IV) and mice were treated with weekly IP injections of human IgG1 or LY3300054 at 10 mg/kg for a total of four doses.
Established xenograft tumor models in CD34+ hHSC-engrafted mice: Cord blood derived CD34+ hHSC transplanted NSG mice were implanted subcutaneously with serially passaged HCC827 tumor fragments (4–5 mm in diameter) at 15–17 weeks of age. When the tumors reached volumes of approximately 200 mm3 (~ 30 days), weekly IP treatments of human IgG1 or LY3300054 at 10 mg/kg began for a total of three doses. Fetal liver-derived CD34+ hHSC transplanted NOG mice were implanted subcutaneously with serially-passaged OV79.FFluc2A–gfp tumor fragments (4–5 mm in diameter) at 15–17 weeks of age. OV79.FFLuc-2A-gfp tumor cells are an ovarian carcinoma line transduced with lentivirus encoding firefly luciferase and green fluorescent protein from a bicistronic transcript  and will be hereafter be referred to as OV79. When tumor volumes reached ~ 150 mm3 (18 days), weekly IP treatments of human IgG1 or LY3300054 at 10 mg/kg began for a total of four doses.
Immune phenotyping of peripheral blood from tumor-bearing mice in humanized models
Peripheral human immune cell engraftment and phenotype was assessed using Trucount™ tubes according to manufacturer’s instruction (BD Biosciences, San Jose, CA). Briefly, 50 μl of blood from hHSC-transplanted mice (day 18, pre-treatment; day 34, after three treatment doses; day 46, after four treatment doses), was added to the tubes and stained with antibodies against human CD45-FITC (BD Biosciences), human CD3-BV786 (Biolegend), human CD4-BV650 (BD Biosciences), human CD8-BV605 (Biolegend, San Diego, CA), and human PD-1-PEeFluor610 (eBiosciences, San Diego, CA) cell surface markers. Samples were subsequently fixed and collected on a 5-laser Fortessa X-20 cytometer (BD Biosciences) and analyzed with FlowJo V10 software (TreeStar). Briefly, approximately 5000 fluorescent beads were collected and enumerated. Human CD45+ cells were also gated and enumerated, followed by subsequent gating on CD3+cells, followed by CD4+ cell and CD8+ cell gating and enumeration, and finally PD-1+ expressing cells were identified using appropriate IgG control. The absolute number of T cells and CD4+ and CD8+ subsets were calculated based on relative beads collected compared to total number provide by manufacturer. Statistical analysis for human T cell engraftment and phenotype was performed using a two-way ANOVA on repeated measurements.
Gene expression analysis of tumor and peripheral tissues in humanized tumor models
Total RNA was isolated from snap-frozen tumor tissue (day 15 from H292 model and day 15 post T cell infusion from HCC827 tumor model) or from snap frozen white blood cell pellets, spleens, or bone marrow (hHSC-engrafted models), using the MagMAX 96 Total RNA isolation (Life Technologies, Carlsbad, CA) and RNeasy mini (Qiagen, Hilden, Germany) kits, respectively.
For QuantiGene Plex analysis, 500 ng of total RNA from tumor tissues were subjected to a custom-designed multiplex assay (targets are listed in Additional file 2) according to manufacturer (Affymetrix, Santa Clara, CA) protocol. For nCounter analysis, 100 ng of total RNA from white blood cells were analyzed with the Human Immunology v2 (targets are listed in Additional file 2) nCounter codeset following manufacturer recommendations (NanoString Technologies, Seattle, WA). One- or two-way ANOVA was used for statistical analysis.