Anti-PD-L1 Antibodies

Abstract

Anti-PD-L1 antibodies are disclosed. Also disclosed are pharmaceutical compositions comprising such antibodies, and methods of using such antibodies to restore T-cell function in T-cells exhibiting T-cell exhaustion or T-cell anergy.

Claims

1-31. (canceled)

32. An antibody, or antigen binding fragment, which is capable of binding to PD-L1, optionally isolated, having at least one light chain variable region incorporating the following CDRs: TABLE-US-00016 (SEQ ID NO: 18) LC-CDR1: TGSSSNIGAGYDVH (SEQ ID NO: 19) LC-CDR2: GNSNRPS (SEQ ID NO: 20) LC-CDR3: QSYDSSLSGSYVV; and having at least one heavy chain variable region incorporating the following CDRs: TABLE-US-00017 (SEQ ID NO: 21) HC-CDR1: SYAIS (SEQ ID NO: 22) HC-CDR2: RIIPILGIANYAQKFQG (SEQ ID NO: 25) HC-CDR3: SGHGYSYGAFDY.

33. The antibody, or antigen binding fragment, according to claim 32, that specifically binds to human or murine PD-L1.

34. The antibody, or antigen binding fragment, according to claim 32, that inhibits interaction between human PD-1 and human PD-L1, or inhibits interaction between murine PD-1 and murine PD-L1.

35. The antibody, or antigen binding fragment, of claim 32, wherein the antibody is effective to restore T-cell function in T-cells exhibiting T-cell exhaustion or T-cell anergy.

36. The antibody or antigen binding fragment of claim 32, comprising a heavy chain variable region sequence and a light chain variable region sequence, wherein: the heavy chain variable region sequence has at least 70% sequence identity to the heavy chain variable region sequence of SEQ ID NO:8 or 35 (FIG. 2), and the light chain variable region sequence has at least 70% sequence identity to the light chain variable region sequence of SEQ ID NO:4 (FIG. 1).

37. The antibody or antigen binding fragment of claim 32, which is a bispecific antibody or a bispecific antigen binding fragment comprising (i) an antigen binding fragment according to claim 3, and (ii) an antigen binding fragment which is capable of binding to a target protein other than PD-L1.

38. The antibody, or antigen binding fragment, of claim 37, wherein the antigen binding fragment which is capable of binding to a target protein other than PD-L1 is capable of binding to one of CD27, CD28, ICOS, CD40, CD122, OX40, 4-1BB, GITR, LAG-3, B7-H3, B7-H4, BTLA, CTLA-4, A2AR, VISTA, TIM-3, PD-1, KIR, HER-2, HER-3, EGFR, EpCAM, CD30, CD33, CD38, CD20, CD24, CD90, CD15, CD52, CA-125, CD34, CA-15-3, CA-19-9, CEA, CD99, CD117, CD31, CD44, CD123, CD133, ABCB5 and CD45.

39. A method of treating a T-cell dysfunctional disorder, cancer or an infectious disease, comprising administering an antibody or antigen binding fragment to a patient suffering from a T-cell dysfunctional disorder, cancer or an infectious disease, wherein antibody, or antigen binding fragment, is capable of binding to PD-L1, and comprises: at least one light chain variable region incorporating the following CDRs: TABLE-US-00018 (SEQ ID NO: 18) LC-CDR1: TGSSSNIGAGYDVH (SEQ ID NO: 19) LC-CDR2: GNSNRPS (SEQ ID NO: 20) LC-CDR3: QSYDSSLSGSYVV; and at least one heavy chain variable region incorporating the following CDRs: TABLE-US-00019 (SEQ ID NO: 21) HC-CDR1: SYAIS (SEQ ID NO: 22) HC-CDR2: RIIPILGIANYAQKFQG (SEQ ID NO: 25) HC-CDR3: SGHGYSYGAFDY.

40. A method for expanding a population of T cells, wherein T cells are contacted in vitro or ex vivo with an antibody or antigen binding fragment, wherein antibody, or antigen binding fragment, is capable of binding to PD-L1, and comprises: at least one light chain variable region incorporating the following CDRs: TABLE-US-00020 (SEQ ID NO: 18) LC-CDR1: TGSSSNIGAGYDVH (SEQ ID NO: 19) LC-CDR2: GNSNRPS (SEQ ID NO: 20) LC-CDR3: QSYDSSLSGSYVV; and at least one heavy chain variable region incorporating the following CDRs: TABLE-US-00021 (SEQ ID NO: 21) HC-CDR1: SYAIS (SEQ ID NO: 22) HC-CDR2: RIIPILGIANYAQKFQG (SEQ ID NO: 25) HC-CDR3: SGHGYSYGAFDY.

41. The antibody or antigen binding fragment of claim 32, comprising a heavy chain variable region sequence and a light chain variable region sequence, wherein: the heavy chain variable region sequence has at least 85% sequence identity to the heavy chain variable region sequence of SEQ ID NO:8 (FIG. 2), and the light chain variable region sequence has at least 85% sequence identity to the light chain variable region sequence of SEQ ID NO:4 (FIG. 1).

42. The antibody or antigen binding fragment of claim 32, comprising a heavy chain variable region sequence and a light chain variable region sequence, wherein: the heavy chain variable region sequence has at least 85% sequence identity to the heavy chain variable region sequence of SEQ ID NO:35 (FIG. 2), and the light chain variable region sequence has at least 85% sequence identity to the light chain variable region sequence of SEQ ID NO:4 (FIG. 1).

43. The antigen binding fragment of claim 32, wherein the antigen binding fragment is a Fab fragment or scFv fragment.

44. The antibody of claim 32, wherein the antibody comprises a human constant region selected from IgG1, IgG2, IgG3 and IgG4.

45. The method according to claim 39, wherein the method is a method of treating cancer, and comprises administering the antibody or antigen binding fragment to a patient suffering from cancer.

46. The method according to claim 45, wherein the cancer is a cancer of a tissue selected from the group consisting of lung, colon, skin, breast, liver, stomach, nasopharynx, oral cavity, oesophagus, larynx, salivary gland, tongue, tonsil, trachea, kidney, bladder, and blood.

47. The method according to claim 45, wherein the cancer is selected from the group consisting of lung cancer, non-small cell lung cancer (NSCLC), colon cancer, colon carcinoma, colorectal carcinoma, skin cancer, melanoma, metastatic melanoma, breast cancer, liver cancer, hepatoma, stomach cancer, head and neck cancer, nasopharyngeal cancer, oral cavity cancer, oesophageal cancer, laryngeal cancer, salivary gland cancer, tongue cancer, tonsil cancer, tracheal cancer, renal cancer, bladder cancer, haematologic cancer, lymphoma and Hodgkin's lymphoma.

Description

BRIEF DESCRIPTION OF THE FIGURES

[0238] Embodiments and experiments illustrating the principles of the invention will now be discussed with reference to the accompanying figures in which:

[0239] FIG. 1. Light chain variable domain sequences for anti-PD-L1 antibody clones A1, C2, C4, H12 and H12_GL. CDRs are underlined and shown separately.

[0240] FIG. 2. Heavy chain variable domain sequences for anti-PD-L1 antibody clones A1, C2, C4, H12 and H12_GL. CDRs are underlined and shown separately.

[0241] FIG. 3. Table showing light chain and heavy chain CDR sequences for anti-PD-L1 antibody clones A1, C2, C4, H12 and H12_GL.

[0242] FIG. 4. Nucleotide and encoded amino acid sequences of heavy and light chain variable domain sequences for anti-PD-L1 antibody clones A1, C2, C4, H12 and H12_GL.

[0243] FIG. 5. Bar Chart showing binding of anti-PD-L1, human Fc conjugated antibodies A1, C2, C4 and H12 and control antibodies A3 (which binds to human but not murine PD-L1) and E3 (which does not bind to human or murine PD-L1).

[0244] FIG. 6. Chart showing inhibition of human PD-1/human PD-L1 interaction by human Fc conjugated antibodies A1, C2, C4 and H12, as determined by ELISA.

[0245] FIG. 7. Chart showing inhibition of murine PD-1/murine PD-L1 interaction by human Fc conjugated antibodies A1, C2, C4 and H12, as determined by ELISA.

[0246] FIG. 8. Chart showing inhibition of murine PD-1/murine PD-L1 interaction by H12 (IgG1) and commercial anti-mouse PD-L1 antibody.

[0247] FIG. 9. Bar Chart showing IFNγ secretion of exhausted T cells in response to antibodies H12, nivolumab (commercial anti-PD-1 antibody), and isotype and no antibody control.

[0248] FIG. 10. Bar Chart showing proliferation of exhausted T cells in response to antibodies anti-PD-L1 H12, nivolumab, and isotype and no antibody control.

[0249] FIG. 11. Sensorgram showing affinity of anti-PD-L1 H12 for human PD-L1 as determined by surface plasmon resonance (SPR). KD=3.13×10.sup.−10 M.

[0250] FIG. 12. Sensorgram showing affinity of anti-PD-L1 H12 for murine PD-L1 as determined by surface plasmon resonance (SPR). KD=3.04×10.sup.−9 M.

[0251] FIG. 13. Bar Charts showing IFNγ secretion (A) of dissociated lung tumor tissue and

[0252] (B) dissociated lung tumor tissue co-cultured with allogenic dendritic cells (DC) in a mixed lymphocyte reaction (MLR), in response to nivolumab, labrolizumab (both commercial anti-PD-1 antibodies), antibody H12, and isotype and no antibody controls.

[0253] FIG. 14. Bar Chart showing IFNγ secretion after culture of PBMCs with Influenza in the presence of the nivolumab, labrolizumab, antibody H12-IgG.sub.4, antibody H12-IgG.sub.1, and isotype and no antibody controls.

[0254] FIG. 15. Bar Chart showing ELISA with H12 on different antigens: human (hu) PD-1, PD-L1, PD-L2, TIM-3, LAG-3, CTL-A4, BLTA, ICOS and CD28, and cynomolgus (cy) and mouse (mo) PD-L1.

[0255] FIG. 16. Graphs showing inhibition of tumour growth by antibody H12 in (A) a mouse colon cancer model using CT26 cell line and (B) a mouse melanoma model using B16F10 cell line.

[0256] FIG. 17. Graph showing efficiency of blocking of interaction between PD-1 and PD-L1 by antibodies H12-IgG.sub.1, H12-IgG.sub.4, H12_GL-IgG.sub.4, and H12_GL-IgG.sub.4, as determined by blocking assay using HEK293-6E cells transfected with human PD-L1.

[0257] FIG. 18. Graphs showing in vivo efficiency of anti-PD-L1 antibody H12 to control tumour growth in a mouse model. MC38 cells were inoculated into mice, and five doses of 200 μg per animal of the indicated antibody were given intraperitoneally starting at day 8. (A) and (B) show the results of tumour growth (mean±SEM) for two independent experiments.

[0258] FIG. 19. Graphs showing binding of antibodies H12, C4 and atezolizumab to human PD-L1, and avidity of binding. (A) Binding of H12, C4, atezolizumab and isotype control antibodies to human PD-L1 (mean±SD of duplicates). (B) Avidity of binding of H12, C4, atezolizumab and isotype control antibodies to human PD-L1 (mean±SD of duplicates).

[0259] FIG. 20. Graph showing binding of H12 IgG1, C4 IgG1, atezolizumab and DEN IgG1 antibodies to MDA-MB-231 human breast cancer cells. Mean fluorescence index (MFI) is shown (mean±SD of duplicates).

[0260] FIG. 21. Graph showing binding of PD-L1 to PD-1 in the presence of H12, C4, atezolizumab and isotype control antibodies. The graph shows ability of antibodies to block binding of PD-L1 to its receptor PD-1 (mean±SD of duplicates).

EXAMPLES

[0261] The inventors describe in the following Examples the identification of nucleotide and amino-acid sequences of isolated antibodies, or the antigen-binding portions thereof, that specifically bind human and murine PD-L1, block the PD-1 pathway and restore exhausted T cell activity.

Example 1

Isolation of Anti-Human PD-L1 Antibodies

[0262] Anti-PD-L1 antibodies were isolated from a human antibody phage display library via in vitro selection.

[0263] Streptavidin-magnetic beads were coated with biotinylated human PD-L1 and used to fish-out anti-PD-L1-specific phages using magnetic sorting. Some steps to remove potential anti-biotin antibodies were added in the selection process.

[0264] Specific Fab antibodies were originally identified by ELISA with human-PD-L1 as the antigen. Four clones were retained for further characterisation based on their slow dissociation from human PD-L1: clones A1, C2, C4 and H12. A first clonality screening was performed by DNA fingerprinting; clonality was then confirmed by sequencing.

Example 2

Binding to Human and Murine PD-L1

[0265] Human and murine PD-L1 were coupled to human Fc and used as antigens coated on ELISA plates for investigation of antibody binding.

[0266] Briefly, ELISA plates were coated with human or murine PD-L1-Fc in carbonate buffer, plates were then blocked with a solution of casein and after extensive washes in PBS Tween-20, anti-PD-L1 antibodies in Fab format were added into the ELISA wells in the presence of 7% milk in PBS. After 90 minutes at room temperature under agitation and extensive washes, a goat anti-human Fab antibody coupled to HRP was added. One hour later, plates were washed and TMB substrate added. The reaction was stopped with 1M HCl and optical density measured at 450 nm with a reference at 670 nm.

[0267] The results are shown in FIG. 5. A1, C2, C4 and H12 antibodies were found to be capable of recognising both human and murine PD-L1 (solid and hatched bars, respectively), and not the coupled Fc part (open bars). Clone A3, which recognises human but not murine PD-L1, and clone E3 which does not recognise either human or murine PD-L1, were included as controls.

[0268] Antibody clones A1, C2, C4 and H12 were each found to be cross reactive for human PD-L1 and murine PD-L1. H12 demonstrated similar binding to human PD-L1 and murine PD-L1.

Example 3

Blocking the PD-1/PD-L1 Interaction In Vitro

[0269] Anti-PD-L1 antibodies were investigated for their ability to block binding of PD-L1 to PD-1 by ELISA assay using PD-1 coupled to human Fc as an antigen. Biotinylated human or murine PD-L1 was pre-incubated in the presence of A1, C4, C2 or H12 Fab prior to addition onto PD-1. Binding of PD-L1 to PD-1 was determined using streptavidin-HRP/TMB substrate. The results of these investigations are shown in FIGS. 6 and 7.

[0270] Antibodies A1, C2, C4 and H12 expressed as Fabs were all found to effectively inhibit the binding of human PD-L1 to human PD-1 in a dose-dependent manner (FIG. 6). Clones A1, C2 and C4 were found to inhibit this interaction with similar efficiency, whilst H12 was found to be slightly more effective than A1, C2 and C4 (FIG. 6).

[0271] Antibody A1, C2, C4 and H12 expressed as Fabs were also found to be able to block the binding of murine PD-L1 to murine PD-1 (FIG. 7). Whilst antibodies A1, C2 and C4 were found to disrupt interaction between murine PD-L1 and murine PD-1 at concentrations higher than the concentration required to block binding of human PD-L1 to human PD-1, H12 was found to be capable of neutralising binding at concentrations similar to those required to block interaction between human PD-L1 and human PD-1 (FIG. 7).

[0272] Inhibition of binding between murine PD-L1 and murine PD-1 was also investigated by antibody H12 expressed in the IgG1 format, and compared to inhibition by a commercially available anti-murine PD-L1 IgG antibody [10F.9G2 (BioLegend, Inc, San Diego, Calif., USA)]. The results are shown in FIG. 8. Antibody H12 was found to be significantly more effective at inhibiting interaction between murine PD-L1 and murine PD-1 at lower antibody concentrations as compared to the commercially available anti-murine PD-L1 IgG antibody.

Example 4

Restoration of Exhausted T Cell Activity

[0273] Ability of anti-PD-L1 antibody H12 to restore exhausted T cell activity was investigated.

[0274] Briefly, T cells were isolated from a healthy donor and cultured for 7 days with monocyte-derived dendritic cells obtained from another donor (50,000 T cells/5,000 DCs), in an allogeneic reaction. The T cells underwent two rounds of stimulation to achieve exhaustion. The exhausted T cells were cultured in the presence of antibody H12 or the anti-PD-1 antibody Nivolumab in the second round of stimulation for 5 days, prior to measurement of IFN-γ in the supernatants and quantification of proliferation using tritiated thymidine.

[0275] The results are shown in FIGS. 9 and 10. Antibody H12 was found to suppress T cell exhaustion, restoring both secretion of IFN-y (FIG. 9) and proliferation (FIG. 10) of T cells. Proliferation and IFN-γ secretion were restored to a similar level by treatment with H12 at 1000 ng/ml as with treatment with Nivolumab at 1000 ng/ml. Notably, proliferation and IFN-γ secretion was higher for T cells treated with H12 antibody than Nivolumab at 200 ng/ml and 40 ng/ml concentrations.

Example 5

Antibody Affinity for PD-L1

[0276] Affinity for antibody H12 for human PD-L1 and murine PD-L1 was investigated by Surface Plasmon Resonance (SPR) analysis. Briefly, human or mouse PD-L1 coupled to Fc was immobilised on a sensor chip compatible with the Proteon XPR36 Bioanalyser (BioRad). Crude H12 Fab extract was then flown onto the chip and the association/dissociation was recorded and analysed and the affinity (K.sub.D) was calculated.

[0277] The results are shown in FIGS. 11 and 12. H12 was found to have a high affinity for human PD-L1 of K.sub.D=0.313 nM, and for murine PD-L1 of K.sub.D=3.04 nM.

Example 6

Use of Anti-PD-L1 Antibodies to Treat Tumours: Ex Vivo Activation of Tumor Infiltrating Lymphocytes

[0278] Lung tumour samples were obtained from the National Cancer Centre Singapore after approval. Samples were dissociated using a human tumour dissociation kit and a tissue dissociator device.

[0279] The tumour dissociated mixture was cultured with anti-PDL-1 IgG, clone H12 for 7 days prior to measurement of IFN-γ in the supernatant by ELISA. Nivolumab and lambrolizumab were used as positive controls in the assay (both are anti-PD-1 antibodies), an isotype antibody was used as a negative control.

[0280] FIG. 13A shows secretion of IFN-γ by tumour infiltrating lymphocytes after 7 days of culture in the presence of the antibodies. H12 re-activated lymphocytes to secrete IFN-γ in a dose-dependent manner.

[0281] Another fraction of the dissociated mixture was co-cultured with allogeneic dendritic cells (DC) to initiate a mixed lymphocyte reaction (MLR). Cells were first cultured for 7 days without antibodies and then for 7 days in the presence of H12 or control antibodies. After these 2 rounds, IFN-γ was assayed in supernatants by ELISA.

[0282] FIG. 13B shows secretion of IFN-γ after the MLR in the presence of the antibodies. H12 restored the ability of tumour lymphocytes to secrete IFN-γ in a dose-dependent manner.

Example 7

Use of Anti-PD-L1 Antibodies to Treat Infections: Autolocious Activation of T Cells in the Presence of Influenza

[0283] Blood was collected from Influenza-positive donors. Monocyte-derived DCs were infected with influenza virus A/PR/8/34 (H1N1). Infected DCs were mixed to PBMCs from the same donor for a first round of culture of 5 days. Cells were then cultured for a second round of 5 days in the presence of H12 or control antibodies. After these 2 rounds, most of the cells in culture are Influenza-specific T cells. At the end of 2 rounds of culture, IFN-γ was assayed in supernatants by ELISA. In this assay, H12 was tested either as an IgG, antibody or as an IgG.sub.4 antibody.

[0284] FIG. 14 shows secreted IFN-γ after culture of PBMCs with Influenza in the presence of the antibodies. Both H12-IgG.sub.1 and H12-IgG.sub.4 were able to restore the capacity of lymphocytes to secrete IFN-γ upon viral stimulation in a dose-dependent manner.

Example 8

Specificity/Cross-Reactivity of H12

[0285] Recognition of various members of the CD28 family by H12 was tested by ELISA.

[0286] FIG. 15 shows the ELISA with H12 on different antigens: human (hu) PD-1, PD-L1, PD-L2, TIM-3, LAG-3, CTL-A4, BLTA, ICOS and CD28, and cynomolgus (cy) and mouse (mo) PD-L1. H12 displayed specificity for the PD-L1 molecule, and cross-reactivity among species.

Example 9

Preliminary In Vivo Efficacy of H12 Antibody

[0287] H12 was tested in 2 tumour models: a colon cancer model using CT26 cell line and a melanoma model using B16F10 cell line.

[0288] Balb/C or C57BL/6 mice were inoculated in the left flank with 0.5×10.sup.6 CT26 or 0.2×10.sup.6 B16F10 tumour cells, respectively, at day 0, and injected intraperitoneally with 250 μg of H12 or isotype IgG1 at days 7, 11, 14, 18 and 21. Tumour growth was then analysed, size was measured with a caliper and volume calculated as follows: V=l.sup.2×L/2 (with l=shorter side and L=longer side).

[0289] FIGS. 16A and 16B show tumour growth after tumour inoculation in both models. H12 inhibited tumour growth in both models.

Example 10

Data with Engineered H12 Antibody

[0290] H12 clone was engineered in order to revert its framework to a germline-like framework; the heavy chain of the new clone H12_GL was slightly modified, the light chain remained the same as for H12 original clone.

[0291] H12_GL was tested in a blocking assay. Briefly HEK 293 cells were transfected with the pcDNA3.1 plasmid expressing human PD-L1 protein using 293tfectin (Invitrogen), according to manufacturer's protocol. PD-L1 expression was confirmed with commercial anti-mouse or anti-human PD-L1 conjugated to PECy7 (Biolegend). 50 to 100 nM of recombinant human PD-1 labelled with Phycoerythrin (PE) was used to bind the PD-L1 expressed on the transfected cells. Serially diluted antibodies (anti-PD-L1 H12, H12_GL or isotype control) were mixed with the human PD-1-PE for 30 minutes before being added to the PD-L1 transfected cells. Cells were washed 3 times with PBS and all data were collected on a BD FACSCanto II (BD Bioscience) and analysed on BD FACSDiva software (BD Bioscience).

[0292] FIG. 17 shows that H12_GL blocked PD-1/PD-L1 interactions as efficiently as H12.

Example 11

In Vivo Efficacy of Anti-PD-L1 Antibody H12: Control of Tumour Growth in Mouse Model

[0293] Because H12 was shown to cross-react with mouse PD-L1, this antibody was evaluated for ability to control tumour growth in a mouse model using colon carcinoma MC38 cells.

[0294] Two million MC38 cells were inoculated to mice subcutaneously at day 0, and five doses of 200 μg per animal of the anti-PD-L1 H12 IgG.sub.1 or isotype control antibody were injected intra-peritoneally, starting at day 8. Tumour size was measured throughout the experiment.

[0295] The results for two independent experiments are shown in FIGS. 18A and 18B. The graphs show tumour growth (mean±SEM) in the MC38 tumour growth model in the presence of anti-PD-L1 antibody H12 or negative isotype control. In both experiments, H12 was able to control tumour growth.

Example 12

Binding of Anti-PD-L1 mAbs to the Target and Comparison with Atezolizumab

[0296] Binding of anti-PD-L1 clones H12 and C4 to human PD-L1 was compared to that of atezolizumab, a commercially available anti-PD-L1 IgG1 antibody.

[0297] Antibodies were coated to a plate and various concentrations of biotinylated PD-L1 were added. Binding was measured by ELISA. Antibodies were coated onto maxisorp plates at 2 μg/mL in coating buffer and incubated overnight at 4° C. The next day, plates were washed with wash buffer (PBS+0.05% Tween-20), and blocker with casein for 1 hour, at room temperature. Plates were then washed again using wash buffer. Various concentrations of biotinylated human PD-L1 were then added to the plates, and the plates were then incubated at room temperature for 1 hour. Plates were then washed again using wash buffer. Streptavidin-HRP was then added and incubated for 1 hour at room temperature to detect biotinylated human PD-L1 bound to the different antibodies. Plates were then washed again using wash buffer. Finally, TMB was added to develop the ELISA; TMB conversion by HRP was stopped using 1M H—Cl.

[0298] Avidity was also assessed by ELISA, with PD-L1 antigen bound onto a plate at single concentration, and addition of various concentrations of antibodies. A secondary antibody was used to detect anti-PD-L1 antibody complexed to the antigen. Neutravidin was coated onto maxisorp plates at 2 μg/mL in coating buffer and incubated overnight at 4° C. The next day, plates were washed with wash buffer (PBS+0.05% Tween-20), and blocker with casein for 1 hour, at room temperature. Plates were then washed again using wash buffer. Biotinylated human PD-L1 was then added to the plates at 0.2 μg/mL and incubated for 1 hour at room temperature. Plates were then washed again using wash buffer. The different antibodies were then added at various concentrations, and incubated for 1 hour at room temperature. Plates were then washed again using wash buffer. Anti-human Fc-HRP was then added and incubated for 1 hour at room temperature. Plates were then washed again using wash buffer. Finally, TMB was added to develop the ELISA; TMB conversion by HRP was stopped using 1M H—Cl.

[0299] FIG. 19A presents the binding of the antibodies (and isotype control) to PD-L1 (mean±SD of duplicates). H12 shows a similar binding curve for PD-L1 as atezolizumab. Affinity of H12 for PD-L1 is higher than affinity of atezolizumab for PD-L1.

[0300] FIG. 19B presents the avidity of the antibodies for their target (mean±SD of duplicates). H12 shows a high avidity to PD-L1, similar to atezolizumab, while C4 shows a medium avidity. Avidity of H12 for PD-L1 is higher than avidity of atezolizumab for PD-L1.

[0301] Binding to PD-L1 was also tested by investigating binding to a MDA-MB-231 human breast cancer cell line that constitutively expresses PD-L1 at the cell membrane. Binding of the antibodies was measured by flow cytometry.

[0302] FIG. 20 presents the mean fluorescence index (MFI) for different concentrations of antibodies (mean±SD of duplicates). The anti-PD-L1 antibodies H12 IgG1 and C4 IgG1 bind efficiently to cells expressing PD-L1.

Example 13

In Vitro Activity of Anti-PD-L1 Antibodies: Blocking of PD-L1 Binding to PD-1

[0303] The ability of the antibodies to block the binding of PD-L1 to its receptor PD-1 was assessed by ELISA. Briefly, the antibodies were pre-incubated with PD-L1 and then added onto ELISA plates onto which PD-1 was coated. After washing, the presence of PD-1-bound PD-L1 was detected with an antibody.

[0304] FIG. 21 presents the results of experiments investigating binding of PD-L1 to PD-1 in the presence or absence of anti-PD-L1 antibodies (or isotype control) (mean±SD of duplicates). The anti-PD-L1 antibodies H12 and C4 are able to block the binding to PD-L1 to its receptor PD-1. Both H12 and C4 were able to block interaction between PD-L1 and PD-1 with greater efficiency than atezolizumab.