A TRAILR2/CDH3 bispecific antibody demonstrates selective apoptosis and tumor regression in CDH3-positive pancreatic cancer

Abstract

Exploitation of extrinsic apoptosis signaling via TRAILR2 activation represents a promising therapeutic concept in cancer treatment. The limited clinical success of previous TRAILR2 agonistic agents, to date, has been ascribed to either poor efficacy or hepatotoxicity. TR2/CDH3 BAB is a human bispecific antibody that relies on binding both CDH3 and TRAILR2 on cell surfaces to achieve TRAILR2 hyperclustering and efficient apoptosis induction by TRAILR2 signaling selectively in CDH3-expressing tumor cells. We demonstrate target-dependent TR2/CDH3 BAB anti-tumor activity in CRISPR/Cas9-engineered TRAILR2 or CDH3 knock-out cells. By utilizing the cell line screening platform PRISM, we found selective TR2/CDH3 BAB efficacy in various cancer types, such as pancreatic, gastric, colorectal, and triple negative breast cancer. The efficacy of TR2/CDH3 BAB correlated with caspase activation in cancer cell lines and in xenograft tumor tissues. In pancreatic ductal adenocarcinoma (PDAC), where patient benefit from current cytotoxic therapy options is unsatisfactory, a close to uniform cell surface expression of CDH3 and TRAILR2 was observed, which will qualify the majority of PDAC patients for TR2/CDH3 BAB-based treatment. TR2/CDH3 BAB demonstrated anti-tumor activity in a panel of PDAC patient-derived xenograft models, including tumor regressions. By combining TR2/CDH3 BAB with chemotherapeutic agents, deeper and more sustained anti-tumor responses were observed when compared to monotherapy. Together with the potential to deliver a favorable safety profile, these data support clinical testing of TR2/CDH3 BAB in patients with PDAC.

Keywords: Apoptosis; CDH3; TRAILR2; bispecific antibody; pancreatic cancer; targeted therapy.

Figure 1.

CDH3 is expressed in different cancer indications together with TRAILR2 but absent in benign liver tissue. a. TRAILR2 and CDH3 mRNA levels (transcripts per million (TPM) in non-transformed human liver tissues and various solid tumor types (source: TCGA and GTEx V8 databases, see methods section for details). The number of cases analyzed is indicated in brackets. Dashed green squares indicate cases with both TRAILR2 and CDH3 mRNA levels above 10 TPM. b. Representative IHC images for H&E, TRAILR2, and CDH3 staining across two PDAC specimens. Scale bars represent 50 µm, and 25 µm for the boxed zoom inserts. c. Representative images from IHC staining for CDH3 expression in the indicated benign human tissues. Scale bar represents 50 µm. Specific staining in breast tissue is indicated by black arrow heads. IHC: immunohistochemistry.

Figure 2.

TR2/CDH3 BAB elicits CDH3 anchor-dependent cell growth inhibition and apoptosis in tumor cells in vitro and in vivo. a. Schematic representation of the bispecific antibody TR2/CDH3 BAB, directed against TRAILR2 and CDH3. It consists of two identical heavy chains and two identical light chains linked together via disulfide bonds. It combines the fragment antigen-binding (fab) domains from an anti-CDH3 antibody with the anti-TRAILR2 single-chain variable fragment (scFv) appended to the C-terminus of the Fc with a flexible linker. The human IgG1 Fc backbone was “silenced” with the L234A/L235A mutation, which ablates binding to FcγR and complement. b. Illustration of TR2/CDH3 BAB mechanism of action. Upon formation of TRAILR2 hyperclusters by CDH3-anchored bispecific antibody, the assembly of intracellular death inducing signaling complex (DISC), associated with recruitment of FADD (FAS-associated-death domain) and activation of caspase 8, leads to activation of downstream effector caspases (such as caspase 3) and caspase 8-mediated BID (BH3-interacting domain death agonist) cleavage. Truncated BID (tBID) interacts with proapoptotic BAX and BAK at the mitochondria, leading to disruption of MMP (mitochondrial membrane potential), release of cytochrome c, and apoptosome-dependent caspase 9 activation. Caspase 8 and caspase 9 mediated activation of executioner caspases ultimately lead to programmed cell death (apoptosis). c. GP2d parental cells and their TRAILR2- and CDH3-KO derivatives were treated for 96 h with different concentrations of TR2/CDH3 BAB and the effect on cell viability was determined. Data are expressed as mean relative values compared to non-treated control ± SD (n = 3). d. Activation of caspase 8 in GP2d CRC cells and their TRAILR2- or CDH3-KO derivatives was determined via the Caspase-Glo® 8 assay at the indicated time points after treatment with 1 nM TR2/CDH3 BAB. Shown are the means ± SD (n = 4 replicates). An ordinary one-way ANOVA analysis with Dunnett´s multiple comparison was performed to test for statistical significance, which is indicated by asterisks (****p ≤ .0001). e. Tumor growth, depicted as mean tumor volumes (in mm³) ± SEM over time, following a single intravenous administration of TR2/CDH3 BAB (treatment at day 1) at the indicated doses to GP2d xenograft tumor-bearing mice (n = 8 animals/group). Mean tumor volumes at treatment start are indicated by a dashed horizontal line. Note that a mean tumor volume below this line indicates tumor regression. f. Activation of caspase 3/7 in GP2d xenograft tumor tissues, as determined by Caspase-Glo® 3/7 assay, upon treatment with the indicated doses of TR2/CDH3 BAB for 24 h. Data are represented as mean ± SD (n = 6). An ordinary one-way ANOVA analysis with Dunnett’s multiple comparison was performed to test for statistical significance (**P ≤ .01, ****P ≤ .0001, n.s.: not significant).

Figure 3.

Efficacy screen of TR2/CDH3 BAB across a large panel of cancer cell lines. a. Waterfall plot of the PRISM activity area, defined as 1-AUC (area under the curve), for the TR2/CDH3 BAB molecule across 863 cell lines. The color code represents the cluster of sensitivity to which each cell line was assigned, from high (violet) to medium (beige) and low (black) sensitivity as defined through the k-means approach. b. Violin plot of log2 TPM gene expression of CDH3 across the three clusters of activity area. A horizontal line indicates the limit of log2(10 + 1) as the limit for expression detection. T-tests are performed across each pairwise comparison between the three classes. The p-values of the tests ≤ 0.05 are reported. Note that not all cell lines evaluable for TR2/CDH3 BAB efficacy, as shown in A, yielded interpretable data for CDH3 mRNA expression. c. Volcano plot of lineage enrichments based on the sensitivity of cell lines to the TR2/CDH3 BAB molecule, focusing only on the 297 cell lines with CDH3 and TRAILR2 expression above the detection limit (10 TPM). The x-axis represents the effect size of the enrichment. The y-axis represents the -log₁₀ of the FDR (false discovery rate) q-value. The dashed horizontal line corresponds to an FDR q-value of 0.1 ( = 1 on the [-log₁₀] y-axis). d. Volcano plot of the hallmark pathways enrichment in the 297 TR2/CDH3 BAB target-positive cell lines as in C. The dashed horizontal line corresponds to an FDR q-value of 0.05 ( = 1.3 on the [-log₁₀] y-axis).

Figure 4.

Efficacy of TR2/CDH3 BAB in a panel of PDAC cell lines. a. Indicated PDAC cell lines were treated with different doses of TR2/CDH3 BAB for 96 h after which cell viability was determined (GI 50 values were derived from dose-response curves based on n = 3 replicates for each drug dose). GI-50 values and maximally observed cell growth inhibition (CGI(max) given in %) are shown. Cell lines displaying a CGI(max) of less than 50% were defined as resistant to the treatment. Indicated PRISM activity areas (actarea = 1-AUC) represent high (>0.54, green), medium (0.20–0.53, yellow), and low (<0.20, red) TR2/CDH3 BAB sensitivity clusters. b. Caspase 8 activity was measured using the Caspase-Glo® 8 assay after treatment of the indicated PDAC cell lines with 1 nM TR2/CDH3 BAB for either 3 h or 6 h. Data are shown as mean ± SEM (n = 4) arbitrary luciferase values relative to non-treated controls. Unpaired t tests with individual variances (Holm–Šidák method) were performed to test for statistical significance, which is indicated by asterisks (**P ≤ .01, ****P ≤ .0001). c. Kinetics and amplitudes of caspase 3/7 induction upon treatment of SU.86.86, PK-59, and Capan2 PDAC cell lines upon treatment with 1 nM TR2/CDH3 BAB were determined by Incucyte® life cell imaging analysis. Data are shown as mean counts per image ± SEM (n = 3).

Figure 5.

TR2/CDH3 BAB demonstrates anti-tumor efficacy in CDH3-positive PDAC patient-derived xenograft models. a. Intraperitoneal administration of TR2/CDH3 BAB at the indicated doses at days 0, 14, and 28 (bi-weekly) in PATX141 PDAC patient-derived xenograft tumors. b. Mean (horizontal black line) and individual (squares) tumor volumes determined in PATX141 xenograft tumor-bearing mice on day 29 after bi-weekly treatment with the indicated doses of TR2/CDH3 BAB. Multiple unpaired t-tests (all time points analyzed) with a single pooled SD followed by Holm-Šidák´s multiple comparisons testing was performed to test for statistical significance (difference to vehicle (V) group), which is indicated by asterisks (****P<.000001). c. Activation of caspase 3/7 in PATX141 PDAC xenograft tumor tissues, as determined by Caspase-Glo® Gl 3/7 assay, upon treatment with the indicated doses of TR2/CDH3 BAB for 24 h. Data are represented as mean (n = 5) ± SEM. Ordinary one-way ANOVA testing with Dunnett´s multiple comparison with a single pooled variance (comparing each treatment with the vehicle group) was performed to test for statistical significance, which is indicated by asterisks (*P ≤ .05, ***P ≤ .001). d. Percentage of cleaved caspase (cCaspase) 3/8-positive tumor cells, as assessed via immunohistochemistry, was determined upon treatment of PATX141 PDAC xenograft tumors with the indicated doses of TR2/CDH3 BAB for 24 h. Data are represented as mean (n = 5) ± SEM. Ordinary one-way ANOVA testing with Dunnett’s multiple comparison with a single pooled variance (comparing each treatment with the vehicle group) was performed to test for statistical significance, which is indicated by asterisks (****P ≤ .0001, n.s.: not significant). e. Tumor growth inhibition (in %) upon treatment with 1 mg/kg TR2/CDH3 BAB, administered intraperitoneally every other week (Q14D, start at day 1), in a panel of 12 PDAC patient-derived xenograft mouse models. Bulk expression of TRAILR2 and CDH3 mRNA in PDX tumors, as determined via RNA-Seq, are given as transcripts per million reads (TPM). f. Treatment of indicated PDAC PDX models with 1 mg/kg TR2/CDH3 BAB every other week (treatment start at day 1). Data or shown as mean ± SEM. Mean tumor volumes at treatment start are also indicated by dashed horizontal lines. Note that a mean tumor volume below this line indicates tumor regression. g. Scatter plot showing the %TGI observed in PDAC PDX models upon treatment with TR2/CDH3 BAB (as detailed in (e)) related to the fraction of TRAILR2/CDH3 mRNA double positive tumor cells found in each model. TRAILR2/CDH3 mRNA detection was performed by RNAscope™ ISH technology (exemplary images shown for PATX153, PATX124, and PATX55 in Supplementary Figure S6A-C). The ID number of each PATX model examined is indicated next to the data point in the graph.

Figure 6.

Benefit of combining TR2/CDH3 BAB with cytotoxic agents in PDAC in vitro and in vivo. a. Matrix of cell growth inhibitions (CGI in %) for the PA-TU 8988S PDAC cell line treated with the indicated doses of Gemcitabine, TR2/CDH3 BAB, or both drugs in combination. % CGI >100 indicate tumor cell death. b. Kinetics and amplitudes of annexin V (left panel) and caspase 3/7 (right panel) induction upon treatment of the PA-TU 8988S PDAC cell lines with the indicated doses of gemcitabine and TR2/CDH3 BAB were determined by Incucyte® life cell imaging analysis. Data are shown as mean ± SEM (n = 4 replicate wells). Note that some error bars in these graphs are not visible since they are shorter than the size of the symbol. c. Treatment of PATX124 PDAC PDX model with 1 mg/kg TR2/CDH3 BAB (every other week: Q14D, i.p.), 50 mg/kg gemcitabine (twice per week: Q3/4D, i.p.), or a combination of the two compounds taking into account the respective drug administration schedules (treatment start at day 0). Data are shown as mean ± SEM. d. Mean (horizontal black lines) and individual tumor volumes determined in PATX124 xenograft tumor-bearing mice treated with gemcitabine (50 mg/kg, Q3/4D), TR2/CDH3 BAB (1 mg/kg, Q14D), or a combination of these two compounds (TR2/CDH3 BAB + GEM) at day 45. e. Treatment of PATX107 PDAC PDX model with 1 mg/kg TR2/CDH3 BAB (Q14D, i.p.), 30 mg/kg nab-paclitaxel (Q7D, i.v.), or a combination of the two compounds taking into account the respective drug administration schedules (treatment start at day 1). Data are shown as mean ± SEM. f. Mean (horizontal black lines) and individual tumor volumes determined in PATX107 xenograft tumor-bearing mice treated with nab-paclitaxel (30 mg/kg, Q7D), TR2/CDH3 BAB (1 mg/kg, Q14D), or a combination of these two compounds (TR2/CDH3 BAB + nab-p) at day 84.