Feasibility of Co-Targeting HER3 and EpCAM Using Seribantumab and DARPin–Toxin Fusion in a Pancreatic Cancer Xenograft Model

Pancreatic cancer (PC) is one of the most aggressive malignancies. A combination of targeted therapies could increase the therapeutic efficacy in tumors with heterogeneous target expression. Overexpression of the human epidermal growth factor receptor type 3 (HER3) and the epithelial cell adhesion molecule (EpCAM) in up to 40% and 30% of PCs, respectively, is associated with poor prognosis and highlights the relevance of these targets. Designed ankyrin repeat protein (DARPin) Ec1 fused with the low immunogenic bacterial toxin LoPE provides specific and potent cytotoxicity against EpCAM-expressing cancer cells. Here, we investigated whether the co-targeting of HER3 using the monoclonal antibody seribantumab (MM-121) and of EpCAM using Ec1–LoPE would improve the therapeutic efficacy in comparison to the individual agents. Radiolabeled 99mTc(CO)3-Ec1–LoPE showed specific binding with rapid internalization in EpCAM-expressing PC cells. MM-121 did not interfere with the binding of Ec1–LoPE to EpCAM. Evaluation of cytotoxicity indicated synergism between Ec1–LoPE and MM-121 in vitro. An experimental therapy study using Ec1–LoPE and MM-121 in mice bearing EpCAM- and HER3-expressing BxPC3 xenografts demonstrated the feasibility of the therapy. Further development of the co-targeting approach using HER3 and EpCAM could therefore be justified.


Introduction
Pancreatic cancer (PC) is one of the most aggressive malignancies with a rapidly growing incidence and high mortality [1,2]. PC ranks in the top ten causes of cancerrelated deaths worldwide, particularly in developed countries [3]. The symptoms of PC are commonly insidious at an early stage, leading to the majority of patients being diagnosed with advanced disease when the cancer has already metastasized [4]. Due to the lack of early detection, PC is characterized by a poor prognosis [5,6]. Patients diagnosed with PC have a median overall survival of fewer than six months and a five-year survival rate of 8-12% [5].
Conventional approaches to the treatment of pancreatic cancer include surgical resection in combination with chemotherapy and/or radiation therapy [7]. However, up to 80% trials [45]. One example is MOC31PE, where the antibody MOC31 was conjugated to the cytotoxic agent Pseudomonas exotoxin A (PE). The MOC31PE treatment significantly increased survival in patients with colorectal peritoneal metastasis [46].
Previously, the EpCAM-targeting toxin fusion protein Ec1-LoPE (43 kDa) was studied in prostate cancer cells, ovarian cancer cells, and breast cancer cells for its cytotoxic effect [47][48][49]. Ec1-LoPE consists of the EpCAM-targeting agent Ec1 (18 kDa), a designed ankyrin repeat protein (DARPin), fused with the cytotoxic agent LoPE (25 kDa) at the Cterminus. DARPins are a type of engineered scaffold protein (ESP) with a molecular weight from 14 to 18 kDa. Similar to other ESPs, DARPins have several advantages over antibodies and antibody fragments that facilitate the delivery of a cytotoxic drug to tumors. First, the small size of DARPins can promote the delivery efficiency of the cytotoxic payload due to superior extravasation and tumor penetration. Second, the recombinant production of ESPs provides higher yields with lower manufacturing costs compared to antibodies [50][51][52]. Ec1 has shown high affinity to EpCAM with an equilibrium dissociation constant (K D ) in the picomolar range [53]. The LoPE toxin (25 kDa) is derived from Pseudomonas exotoxin A (PE), with silenced human and murine B cell recognition epitopes [54]. This modification reduces its immunogenicity and general toxicity [5,47].
The inter-and intra-tumoral heterogeneity of PC makes combination targeted therapy an attractive strategy [5]. Co-treatment using drugs with different action modes could help to overcome the heterogeneity of target expression in tumors, spread the toxicity to critical organs, and prevent the development of resistance [55]. In a previous study, co-treatment with Ec1-LoPE and the anti-HER2 monoclonal antibody trastuzumab delayed tumor growth and significantly prolonged survival of mice bearing EpCAM-and HER2expressing ovarian cancer SKOV3 xenografts compared to single treatments, while no pathological changes in normal organs, e.g., kidneys and livers, were detected [48]. MM-121 demonstrated therapeutic efficacy in a preclinical study in pancreatic BxPC3 xenografts [36].
The goal of this study was to test the hypothesis that the combined treatment using the EpCAM-targeting Ec1-LoPE toxin and the HER3-targeting monoclonal antibody MM-121 would increase the efficacy of experimental therapy in a pancreatic cancer model. To test this hypothesis, the binding specificity of Ec1-LoPE to two EpCAM-expressing pancreatic cancer cell lines, BxPC3 and Panc-1, was tested and the affinity of Ec1-LoPE was measured. We evaluated if co-treatment with monoclonal antibody MM-121 would interfere with the binding of Ec1-LoPE to BxPC3 and Panc-1 cells in vitro. The rate of Ec1-LoPE internalization by BxPC3 and Panc-1 cells in vitro was measured. To facilitate a quantitative evaluation in these experiments, Ec1-LoPE was site-specifically labeled with the radionuclide 99m Tc. Half-value growth inhibition of BxPC3 and Panc-1 cells by Ec1-LoPE was measured. A possible synergistic effect of in vitro treatment of BxPC3 cells (expressing both EpCAM and HER3) using Ec1-LoPE and monoclonal antibody MM-121 was evaluated. The effect of monotherapies using MM-121 and Ec1-LoPE and a combination of these therapies was compared in vivo in mice bearing BxPC3 xenografts.

In Vitro Specificity and Cross-Blocking Study
The binding specificity of [ 99m Tc]Tc(CO) 3 -Ec1-LoPE to EpCAM was investigated using EpCAM-expressing BxPC3 and Panc-1 pancreatic cancer cells. The binding of radioactivity to both cell lines was significantly (p < 0.05) decreased when a large excess of non-labeled Ec1 was added ( Figure 1A), indicating that the binding of [ 99m Tc]Tc(CO) 3 -Ec1-LoPE to both cell lines was EpCAM-specific.

In Vitro Specificity and Cross-Blocking Study
The binding specificity of [ 99m Tc]Tc(CO)3-Ec1-LoPE to EpCAM was investigated using EpCAM-expressing BxPC3 and Panc-1 pancreatic cancer cells. The binding of radioactivity to both cell lines was significantly (p < 0.05) decreased when a large excess of nonlabeled Ec1 was added ( Figure 1A), indicating that the binding of [ 99m Tc]Tc(CO)3-Ec1-LoPE to both cell lines was EpCAM-specific. . p values from unpaired t-tests are provided to demonstrate differences between activity uptake in non-blocked and blocked groups.
To investigate whether MM-121 interfered with the binding of [ 99m Tc]Tc(CO)3-Ec1-LoPE to EpCAM, an excess of MM-121 was added to the cells before the addition of [ 99m Tc]Tc(CO)3-Ec1-LoPE. As an IgG antibody control, an excess of bevacizumab, which does not bind to any receptor in the HER family, was used in the third group. Compared to the non-blocked group, the uptake of activity in the MM-121-treated groups did not differ significantly (p > 0.05) in both cell lines ( Figure 1B). There was a minor but significant (p = 0.016) decrease of cell-associated activity in the bevacizumab-treated BxPC3 cells; however, this was not observed in Panc-1 cells (p = 0.098). These results showed that MM-121 did not decrease the binding of [ 99m Tc]Tc(CO)3-Ec1-LoPE to EpCAM.

In Vitro Cellular Processing and Internalization Study
To evaluate the internalization and cellular processing rate, [ 99m Tc]Tc(CO)3-Ec1-LoPE was continuously incubated with BxPC3 and Panc-1 cells for 1, 2, 4, 6, and 24 h ( Figure 2). The cell-associated activity steadily increased over time and reached maximum by 24 h in both cell lines. The internalized fractions were 41 ± 4% in BxPC3 cells and 37 ± 4% in Panc-1 cells by 24 h. To investigate whether MM-121 interfered with the binding of [ 99m Tc]Tc(CO) 3 -Ec1-LoPE to EpCAM, an excess of MM-121 was added to the cells before the addition of [ 99m Tc]Tc(CO) 3 -Ec1-LoPE. As an IgG antibody control, an excess of bevacizumab, which does not bind to any receptor in the HER family, was used in the third group. Compared to the non-blocked group, the uptake of activity in the MM-121-treated groups did not differ significantly (p > 0.05) in both cell lines ( Figure 1B). There was a minor but significant (p = 0.016) decrease of cell-associated activity in the bevacizumab-treated BxPC3 cells; however, this was not observed in Panc-1 cells (p = 0.098). These results showed that MM-121 did not decrease the binding of [ 99m Tc]Tc(CO) 3 -Ec1-LoPE to EpCAM.

In Vitro Cellular Processing and Internalization Study
To evaluate the internalization and cellular processing rate, [ 99m Tc]Tc(CO) 3 -Ec1-LoPE was continuously incubated with BxPC3 and Panc-1 cells for 1, 2, 4, 6, and 24 h ( Figure 2). The cell-associated activity steadily increased over time and reached maximum by 24 h in both cell lines. The internalized fractions were 41 ± 4% in BxPC3 cells and 37 ± 4% in Panc-1 cells by 24 h.

Cytotoxicity
The cytotoxic effect of Ec1-LoPE was evaluated in EpCAM-expressing BxPC3 and Panc   The effect of Ec1-LoPE in combination with MM-121 on cell viability was evaluated in EpCAM-and HER3-expressing BxPC3 cells. The viability data were analyzed using SynergyFinder 3.0 software and the synergy score was calculated using three different reference models. The resulting synergy scores for all the models were above zero (3.8 for the ZIP model, 6.5 for the HSA model, and 7.3 for the Bliss model), which indicated synergism between Ec1-LoPE and MM-121 ( Figure 4). The most synergistic areas in all three models corresponded to 0.01-1 nM concentrations of Ec1-LoPE and 10-100 nM concentrations of MM-121.

Experimental Therapy Using Ec1-LoPE and MM-121
The efficiency of combination treatment using Ec1-LoPE and MM-121 was compared to the efficiency of monotherapies using Ec1-LoPE and MM-121 and to the vehicle control in female BALB/c nu/nu mice bearing BxPC3 xenografts. There were no statistically significant (p > 0.05) differences in average tumor volume between the groups at the start of the experiment ( Figure 5A). On day 21, after four injections of Ec1-LoPE and six injections of MM-121, the tumors in the combination therapy group (93 ± 68 mm 3 ) and MM-121 monotherapy group (95 ± 53 mm 3 ) were significantly (p < 0.05) smaller than tumors in the Ec1-LoPE monotherapy group (203 ± 94 mm 3 ) and the control group (194 ± 81 mm 3 ). The same trend of significant (p < 0.05) differences between the groups was observed until day 39 when the first five mice were euthanized in the control group and Ec1-LoPE monotherapy group.
The monotherapy with Ec1-LoPE had no measurable effect on the tumor growth in comparison with the vehicle-treated control ( Figure 5B). The tumor doubling time in the Ec1-LoPE monotherapy group (9 ± 3 days) did not differ from the doubling time of the control group (10 ± 4 days). The median survival of mice in these groups was similar, 42 and 46 days, for animals treated with vehicle and Ec1-LoPE monotherapy, respectively ( Figure 5B). The median survival of mice in the combination group was 77 days and in the MM-121 monotherapy group was 58 days; however, no statistically significant differences (p > 0.05, log-rank (Mantel-Cox) test) were found between the survival curves of mice in the combination group and MM-121-treated groups. , at the start of the second treatment cycle using Ec1-LoPE (day 21), and eight days after the last treatment cycle (day 39). NS corresponds to no statistically significant difference (p > 0.05), * corresponds to p < 0.05 (one-way ANOVA with Bonferroni correction). (B) Survival of the mice during experimental therapy. (C) Therapy outcomes for different treatment groups. The outputs categories were defined as progression (exponential tumor growth), growth delay (initial growth suppression followed by exponential tumor growth), and survival (animals did not reach a humane endpoint at the study termination). The difference between groups was determined using the chi-square test. NS corresponds to no statistically significant difference (p > 0.05), *** corresponds to p < 0.001, **** , and eight days after the last treatment cycle (day 39). NS corresponds to no statistically significant difference (p > 0.05), * corresponds to p < 0.05 (one-way ANOVA with Bonferroni correction). (B) Survival of the mice during experimental therapy. (C) Therapy outcomes for different treatment groups. The outputs categories were defined as progression (exponential tumor growth), growth delay (initial growth suppression followed by exponential tumor growth), and survival (animals did not reach a humane endpoint at the study termination). The difference between groups was determined using the chisquare test. NS corresponds to no statistically significant difference (p > 0.05), *** corresponds to p < 0.001, **** corresponds to p < 0.0001. (D) Average animal weight in each group during the therapy experiment. The data are presented as an average value ± SD. All mice in in the vehicle and Ec1-LoPE-treated groups had exponential tumor growth (progression) ( Figure 5C, Figure 6 and Figure S3, Table S1). The therapy outcomes in the groups treated with MM-121 monotherapy or the combination of MM-121 and Ec1-LoPE were different. There were some tumors, which did not respond to therapy (six mice treated with MM-121 and four mice treated with the combination), with a growth rate similar to the growth rate in the vehicle control group (doubling time 9 ± 1 days for both treatment groups). In some animals from these groups (four mice in the combination therapy and one mouse in the MM-121 monotherapy group), the tumor growth was initially suppressed but resumed after day 50. The growth of some tumors in these groups was suppressed and did not resume during the observation period. While the monotherapy using Ec1-LoPE was inefficient according to the chi-square contingency test, adding Ec1-LoPE to MM-121 potentiated it, making the combination therapy outcome significantly better than the outcome of the MM-121 monotherapy ( Figure 5C).  All treatment regimens were well tolerated by mice with no observable side effects. No tendency for weight loss was observed in any of the groups. The differences in average animal weight between the groups were within the standard deviation ( Figure 5D). During the histopathological examination of the liver and kidneys from mice treated with Ec1-LoPE, MM-121, or their combination, no lesions were found that could suggest toxicity of the treatment regimens ( Figures S1 and S2).

Discussion
The combination of targeted therapeutics has the potential to reduce toxicity to normal tissues in the treatment of pancreatic cancer [56], which may further widen a therapeutic window. Overexpression of EpCAM in multiple malignancies indicates that com- injections of vehicle (0.9% NaCl). Mice were euthanized when the xenograft volume exceeded 1000 mm 3 , bleeding ulcers on xenografts were observed, or over 10% weight loss was detected within one week.
All treatment regimens were well tolerated by mice with no observable side effects. No tendency for weight loss was observed in any of the groups. The differences in average animal weight between the groups were within the standard deviation ( Figure 5D). During the histopathological examination of the liver and kidneys from mice treated with Ec1-LoPE, MM-121, or their combination, no lesions were found that could suggest toxicity of the treatment regimens ( Figures S1 and S2).

Discussion
The combination of targeted therapeutics has the potential to reduce toxicity to normal tissues in the treatment of pancreatic cancer [56], which may further widen a therapeutic window. Overexpression of EpCAM in multiple malignancies indicates that complementary treatment with EpCAM-targeting agents might be a universal strategy to potentiate the effect of targeted cancer therapy.
Anti-EpCAM Ec1-LoPE DARPin-toxin fusion demonstrated efficacy in several cancer models, both alone and in combination with other targeting agents [47][48][49]. Thus, its application for a therapy-resistant malignancy like pancreatic cancer might be attractive. The mechanism of the cytotoxic action of Ec1-LoPE includes binding of Ec1 to EpCAM and receptor-mediated internalization of the EpCAM-Ec1-LoPE complex inside the cell. LoPE is a de-immunized derivative of bacterial Pseudomonas aeruginosa Exotoxin A (PE toxin) [54], which contains receptor binding domain I, intracellular processing domain II, and the catalytically active domain III, which mediates toxic action. Domain III is an NAD+diphthamide ADP-ribosyltransferase that inactivates eukaryotic elongation factor eEF2, which leads to the inhibition of protein synthesis and cell death [57]. MM-121 binds to HER3 and blocks signaling induced by the extracellular growth factors heregulin and betacellulin. This leads to inhibition of key pro-survival pathways mediated by phosphoinositide 3kinase (PI3K) and Akt signaling and to cancer cell death [25,58]. The combination of two agents with different mechanisms of action, Ec1-LoPE with a direct cytotoxic action and MM-121 blocking key pro-survival pathways, might increase their cytotoxic effect.
It should be noted that malignant tumors constitute a very heterogeneous group of diseases. They differ in proliferation-driving mutations, patterns of molecular target expression and co-expression, and sensitivity/resistance to different therapies. It is not a given that a therapy that is effective in one type of cancer (e.g., ovarian) would be equally effective in another one (e.g., pancreatic). The feasibility of treatment should be demonstrated in each case. The selection of an appropriate model for proof-of-principle study, including the aspects of sufficient target expression, cellular processing of a targeting agent, and cytotoxicity is critical.
To quantitatively assess the targeting properties of Ec1-LoPE in vitro, the compound was site-specifically labeled with technetium-99m tricarbonyl via hexahistidine tag. This label has residualizing properties and thus permits the study of retention after internalization in cells [59]. The same labeling method was used previously and provided a stable label [48,49]. The specificity of Ec1 binding to EpCAM was previously demonstrated by several orthogonal methods [60]. In the binding specificity experiment, the binding of [ 99m Tc]Tc(CO) 3 -Ec1-LoPE to both BxPC3 and Panc-1 cells decreased significantly (p < 0.05) in the presence of unlabeled Ec1 when the EpCAM receptors were saturated. This indicated that the uptake of Ec1-LoPE was EpCAM-mediated ( Figure 1A). The binding to BxPC3 cells was higher than the binding to Panc-1 cells. A similar pattern was observed previously for the non-LoPE-fused Ec1, which could be explained by higher EpCAM expression in BxPC3 cells than in Panc-1 cells [61]. The binding and internalization pattern of [ 99m Tc]Tc(CO) 3 -Ec1-LoPE was similar for both cell lines (Figure 2). Both cell lines internalized the targeting toxin efficiently (approximately 40% of cell-associated activity was internalized after 24 h incubation), which created an important precondition for intracellular delivery of the toxin. The higher EpCAM expression level in BxPC3 cells correlated with the higher cytotoxicity (lower IC 50 value) of Ec1-LoPE for this cell line ( Figure 3). Importantly, the prevention of Ec1-LoPE binding to both cell lines by target saturation with Ec1 (not fused with the toxin) significantly reduced cytotoxicity ( Figure 3C), which demonstrated that the effect was caused by the specific toxin delivery to cells. Overall, the cytotoxicity of Ec1-LoPE for pancreatic cancer cells (IC 50 in the range 240-440 pM) was lower compared with the cytotoxicity for prostate cancer cell lines (IC 50 in the range 30-60 pM) [49], but higher than that for the ovarian cancer cell line SKOV-3 (500 nM) [48]. Still, the treatment of SKOV-3 xenografts using Ec1-LoPE increased the survival of mice [48]. This suggested that the in vivo experiments using Ec1-LoPE in the pancreatic cancer model were ethically justified.
For the proof-of-principle study, the BxPC3 cell line was selected because it has a higher target expression level and is more sensitive to Ec1-LoPE treatment.
The binding of Ec1-LoPE to living BxPC3 cells was characterized by two interactions (Table 1), both having K D values in the nanomolar range. A similar pattern was observed in our previous studies [48,49], while higher affinity was observed for Ec1 not fused to LoPE toxin in ovarian cancer cells [53]. This suggested that the fusion of LoPE to Ec1 reduced the affinity of Ec1, which might be a result of increased molecular size and associated steric hindrance in the interaction with EpCAM.
In vitro binding experiments demonstrated that incubation of the pancreatic cancer cells with MM-121 did not reduce the binding of [ 99m Tc]Tc(CO) 3 -Ec1-LoPE ( Figure 1B), suggesting that the combination treatment was feasible. Further in vitro evaluation and calculations using several models suggested that an appreciable synergistic effect of the co-treatment could be expected (Figure 4).
The experimental proof-of-principle therapy was performed in two cycles using an adaptive design. Since the results of the first Ec1-LoPE treatment cycle did not show a clear anti-tumor effect (but also no signs of general toxicity), a second cycle was initiated two weeks after the first one. Still, the monotherapy using Ec1-LoPE was inefficient and the tumor growth rate in treated mice was similar to the growth rate in the control group ( Figure 6B,D). Further analysis did not reveal any difference in median survival ( Figure 5B) or response rate ( Figure 5C). These data are somewhat different from the data obtained for ovarian cancer [48] and breast cancer models [47], where the monotherapy had a certain anti-tumor effect. It has to be noted that different models and different treatment schemes were used in our study compared to the studies by Shramova et al. [47] and Xu et al. [48]. Further investigation and optimization of the pharmacokinetic parameters of Ec1-LoPE might be necessary to improve the delivery to tumors and its anti-tumor action in the BxPC3 xenograft model. The difference between the survival curves of mice in the combination group and mice in the MM-121 monotherapy group was not statistically significant ( Figure 5B). It is of interest that the combination treatment improved therapy outcome compared with the MM-121 monotherapy. The contingency analysis demonstrated a significant difference in the outcomes of these therapies ( Figure 5C). These results demonstrate the feasibility of EpCAM-and HER3-targeting therapy using Ec1-LoPE and MM-121 in pancreatic cancer.

General
The chemicals used in the study were purchased from Sigma-Aldrich (Sweden AB, Stockholm, Sweden). Buffers were prepared using high-quality Milli-Q water.

In Vitro Specificity and Cross-Blocking Study
The evaluation of binding specificity of [ 99m Tc]Tc(CO) 3 -Ec1-LoPE to EpCAMoverexpressing Panc-1 and BxPC3 cells was performed as described previously [48]. Briefly, 0.8 × 10 6 cells/well were seeded in 6-well plates one day before the experiment. The experiment was done in triplicates. On the experiment day, the medium was removed from the cells, and Ec1 (200 nM in 0.5 mL of medium) was added to one set of wells (the blocked group). The same volume of medium only was added to another set of wells (the non-blocked group). The cells were incubated at 37 • C for 60 min. Then, [ 99m Tc]Tc(CO) 3 -Ec1-LoPE (4 nM in 0.5 mL in medium) was added to each well to reach a final concentration of 2 nM, followed by incubation at 37 • C for 60 min. After incubation, the medium was collected and cells were washed with medium (1 mL), which was collected into the same tubes. To collect the cell-associated activity, the cells were incubated with 1 M NaOH (1 mL) at 37 • C for 30 min, followed by a wash with 1 M NaOH (1 mL), and the cell lysate was collected into a tube.
To evaluate the feasibility of simultaneous targeting of the EpCAM and HER3 receptors, MM-121 (200 nM in 0.5 mL of medium) was added before the addition of [ 99m Tc]Tc(CO) 3 -Ec1-LoPE and the cells were treated as described above. To investigate whether the potential cross-blocking effect of MM-121 might be caused by the non-specific interaction of an IgG scaffold, the anti-vascular endothelial growth factor A (VEGF-A) monoclonal antibody bevacizumab (200 nM) was added to another set of wells prior to the addition of [ 99m Tc]Tc(CO) 3 -Ec1-LoPE. The activity in the tubes containing medium and cells was measured using an automatic gamma spectrometer equipped with a 3-inch NaI(Tl) well detector (2480 Wizard, Wallac, Turku, Finland) and the percent of cell-associated activity was calculated. The percent of cell-associated activity was calculated as the amount of radionuclide (in counts per minute) in the tube containing cells divided by the sum of the amounts of radionuclide in the tube containing medium and the tube containing cells.

In Vitro Internalization Study
The cellular processing and internalization of [ 99m Tc]Tc(CO) 3 -Ec1-LoPE by BxPC3 and Panc-1 cells was studied as described previously [66]. BxPC3 and Panc-1 cells were seeded into 3 cm Petri dishes at a density of 0.8 × 10 6 cells/well one day before the experiment. The experiment was done in triplicates. On the experiment day, the medium was removed from the cells and [ 99m Tc]Tc(CO) 3 -Ec1-LoPE (2 nM, 1 mL) was added to each well followed by incubation at 37 • C for 1, 2, 4, 6, or 24 h. At each time point, the cell medium was collected into tubes and the cells were washed with 1 mL of ice-cold PBS, which was collected. To collect the membrane-bound activity, cells were treated with 1 mL of 0.2 M glycine buffer containing 4 M urea (pH 2.0) for 5 min on ice and the solution was collected. The cells were washed once with the same solution, which was collected. To collect the internalized activity, cells were incubated 1 M NaOH at 37 • C for 20 min and treated as described above. The activity in each fraction tube was measured using the gamma spectrometer and the percent of cell-associated activity was calculated. The data were normalized to the maximum value of cell-associated activity at 24 h that was set as 100%.

Interaction Analysis Using Ligand Tracer
The interaction of [ 99m Tc]Tc(CO) 3 -Ec1-LoPE with living BxPC3 cells was measured using the LigandTracer Yellow instrument (Ridgeview Instruments AB, Vänge, Sweden) as described previously [48]. Cells (2 × 10 6 ) were seeded to a local area of an 8.9 cm Petri dish (Nunclon TM, Roskilde, Denmark) one day before the experiment. The experiment was performed at room temperature to prevent internalization. After 30 min of background measurement, [ 99m Tc]Tc(CO) 3 -Ec1-LoPE was added (3 and 9 nM) and the measurement was performed for 90 min at every concentration. Thereafter, the medium containing the radiolabeled compound was replaced by fresh medium to measure the dissociation rate. The binding curve was analyzed by the TracerDrawer software (Ridgeview Instruments AB, Vänge, Sweden) and the equilibrium dissociation constant (K D ) was calculated. The interaction heterogeneity was evaluated by Interaction Map software (Ridgeview Diagnostics AB, Uppsala, Sweden).

In Vitro Proliferation Assay
The effect of Ec1-LoPE on cell proliferation was studied using EpCAM-expressing BxPC3 and Panc-1 cells. The cells were seeded in 96-well plates at a density of 2000 cells/well and incubated in a humidified incubator overnight to allow attachment. The next day, a serial dilution of Ec1-LoPE (0-1000 nM) in a culture medium containing 4% FBS was prepared and added to the cells (n = 5). The plates were incubated in a humidified incubator for 72 h. To test the specificity of the cytotoxic action of Ec1-LoPE, one group of cells was incubated with Ec1 (1000 nM) for 10 min before the addition of Ec1-LoPE (10 nM).
The inhibition of proliferation of BxPC3 cells by MM-121 in vitro was studied previously by Leitao et al. [36]. To evaluate the combined effect of Ec1-LoPE and MM-121 on the proliferation of EpCAM-and HER3-expressing BxPC3 cells, on day one, six concentrations of MM-121 (0, 1, 5, 10, 50, and 100 nM) in the presence of heregulin (4 nM) were prepared in culture medium containing 4% FBS and added to the cells. The plates were incubated in a humidified incubator for 48 h. On day three, the medium was exchanged with a solution containing Ec1-LoPE (0-1000 nM) and MM-121 (0-100 nM) in the presence of heregulin (4 nM) in a culture medium containing 4% FBS. The plates were incubated in a humidified incubator for additional 72 h.
Cell viability was measured using cell counting kit-8 (CCK-8; Sigma-Aldrich, St. Louis, MO, USA) according to the manufacturer's protocol. Briefly, on the day of measurement the medium was replaced with fresh medium (100 µL) and CCK8 solution (10 µL) was added to each well, followed by incubation for 1-3 h. The absorbance (OD value) was measured at 450 nm for cell viability and at 655 nm for background. The values from wells containing cells in medium only (no Ec1-LoPE or MM-121) were used as the 100% viability control. The values from wells without cells were used as the 0% viability control. The relative viability was analyzed by GraphPad Prism (version 9.4.0; GraphPad Software, La Jolla, CA, USA) using a log(inhibitor) vs. response-variable slope (four parameters) model providing half-maximal inhibitory concentration (IC 50 ) values.
The effect of Ec1-LoPE and MM-121 combination on the viability of BxPC3 cells was analyzed using the web application SynergyFinder 3.0 [67,68]. In the SynergyFinder web application, to evaluate the potential synergy of drugs, the observed combination responses were compared with expected combination responses calculated using combination scoring models. Based on the deviation of observed and expected combination responses, the drug combination was classified as synergistic (combination effect was higher than expected) or as antagonistic (combination effect was lower than expected). Highest single agent (HSA), Bliss, and Zero interaction potency (ZIP) reference models were used to calculate synergy scores. The has model is based on the assumption that the expected combination effect would equal to the maximum of the single drug responses; the Bliss model is based on the assumption that the expected combination effect can be calculated for two drugs that act independently; and the ZIP model captures the drug interaction by comparing the change in the potency of the dose-response curves between individual drugs and their combinations [69].

Experimental Therapy
The animal experiments were planned and performed following Swedish national legislation on the protection of laboratory animals. The experiments were approved by the local ethical committee for animal research in Uppsala, Sweden (permit 5.8.18-11931/2020, approved 28 August 2020).
Thirty nine female Balb/c nu/nu mice were subcutaneously (s.c.) implanted on the abdomen with a 5 × 10 6 BxPC-3 cells (100 µL in culture medium), and one week later (day 0) the mice were randomized into four groups: MM-121 antibody monotherapy, targeted toxin Ec1-LoPE monotherapy, MM-121 and Ec1-LoPE combined therapy, and vehicle injection control. At the start of the experimental therapy, the average tumor volume was 97 ± 24 mm 3 and the average mouse weight was 19 ± 1 g.
In the antibody monotherapy group, the mice received an intraperitoneal injection of MM-121 antibody (2.1 nmol, 300 µg, 16 mg/kg in 100 µL of 0.9% NaCl) every third day between day 0 (treatment start) and day 28. In the Ec1-LoPE monotherapy group, the animals received two cycles of therapy. During each cycle, Ec1-LoPE was intravenously (i.v.) injected four times, every second day. The first cycle started on day 0, the second cycle started on day 21, and the injected dose was 0.46 nmol (20 µg, 1 mg/kg in 100 µL of 0.9% NaCl). In the combined therapy group, the injections of MM-121 antibody and Ec1-LoPE targeted toxin were performed at the same dosage and with the same schedule as in the monotherapy groups. In the control group, mice were i.v. injected with the vehicle (100 µL of 0.9% NaCl) on the same days as in the combined therapy group.
The tumor dimensions were measured using a digital caliper for the largest longitudinal (length) and transverse (width) diameter twice per week. The tumor volume was calculated using the formula: tumor volume = 1/2(length × width 2 ). The mice's status was monitored twice per week. The mice were euthanized when they reached a predetermined humane endpoint (the subcutaneous tumor volume exceeded 1000 mm 3 ; bleeding sores on the tumor were observed; 15% overall weight loss or 10% weight loss within one week). According to the ethical permit, the study endpoint was 90 days after the treatment started. When the mice were sacrificed, the livers and kidneys were excised, kept in formalin for 24 h, and thereafter stored in ethanol. The tissue samples were embedded in paraffin, sectioned (3-4 µm thickness), and stained with hematoxylin and eosin. The stained samples were investigated for histopathologic changes by a veterinary pathologist at a veterinary medicine laboratory (BioVet AB, Sollentuna, Sweden).

Statistical Analysis
GraphPad Prism (version 9.4.0, GraphPad Software, La Jolla, CA, USA) was used to generate plots and perform statistical analysis. Data are presented as means ± SD (* p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.001 and ns = not statistically different). Student's t-test was used for comparisons between two different groups. For comparisons between multiple groups, analysis of variance (ANOVA) with Bonferroni's post hoc multiple comparisons test was used.

Conclusions
In conclusion, the co-targeting of EpCAM and HER3 using Ec1-LoPE and MM-121 is feasible. The combination of Ec1-LoPE and MM-121 provided a synergistic cytotoxic effect on BxPC3 cells in vitro. The experimental therapy study using Ec1-LoPE and MM-121 in mice bearing BxPC3 xenografts demonstrated the feasibility of the therapy.

Data Availability Statement:
The data generated during the current study are available from the corresponding author upon reasonable request.