The Impact of HER3 Dynamics Altered by HER3-DXd Alone and in Combination with Driver Oncogene Inhibitors on HER3-DXd Efficacy

Abstract

Targeted therapies have revolutionized treatment paradigms for a variety of cancer types; however, challenges including primary and acquired resistance persist, and there remains a high demand for novel treatment options. HER3 (ErbB3), a member of the human epidermal growth factor receptor family of receptor tyrosine kinases, is a target of HER3-DXd, an antibody–drug conjugate currently under clinical investigation. As was previously reported, the cytotoxic activity of HER3-DXd in preclinical models is primarily mediated by the antitumor activity of the released payload. Therefore, we investigated the impact of HER3 expression changes on payload release after HER3-DXd treatment using HER3-positive human cancer cell lines and their xenograft models. In vitro studies showed that the amount of payload released from cells after HER3-DXd treatment was associated with baseline HER3 expression levels, HER3 internalization rate, and turnover rate. In female CAnN.Cg-Foxn1nu/CrlCrlj mouse models, dose and dosing interval influenced membrane HER3 expression levels and tumor payload concentrations. Furthermore, membrane HER3 was upregulated by tyrosine kinase inhibitor treatment in non-small-cell lung cancer cell lines harboring specific driver mutations, including EGFR-activating mutations, ROS1 fusions, and ALK fusions. The increase in HER3 expression induced by osimertinib treatment was associated with increased payload release in PC-9 cells. Our results indicate that HER3 dynamics, as well as baseline HER3 expression, modulate payload release from HER3-DXd and support combination strategies to potentiate the antitumor activity of HER3-DXd.

1. Introduction

Antibody–drug conjugates (ADCs) comprise a novel cancer treatment modality that combines the specificity of an antibody with the cytotoxicity of a payload. Engineered to achieve a wide therapeutic window, ADCs have demonstrated promising activity against various cancers, and 15 ADCs have been approved by the FDA [1,2].

HER3 (encoded by the ERBB3 gene) is a member of the human epidermal growth factor receptor family of receptor tyrosine kinases (RTKs) [3]. Unlike other HER family members, HER3 has impaired tyrosine kinase catalytic activity with diminished autophosphorylation [4,5]. However, upon heterodimerization and phosphorylation in trans by catalytically active RTKs, HER3 can trigger proliferative and survival signaling such as through the PI3K/AKT and MEK/ERK pathways [6,7,8,9]. HER3 expression is increased in a variety of cancers [10,11,12], which in some tumors, is associated with a worse prognosis [10]. Furthermore, tumors with increased HER3 expression demonstrate resistance to chemotherapeutic agents and targeted therapies, including EGFR-tyrosine kinase inhibitors (EGFR-TKIs) for patients with EGFR-mutated non-small-cell lung cancer (NSCLC) [13,14,15,16,17]. HER3 is thus an attractive target for therapeutic intervention.

HER3-DXd, a HER3-targeting ADC composed of patritumab, an anti-HER3 mAb, and a topoisomerase I inhibitor in the form of an exatecan derivative called DXd, is connected via a peptide linker [18,19]. HER3-DXd administered as a single agent has demonstrated clinically meaningful antitumor activity in patients with advanced breast cancer [20] and NSCLC [21]. However, a clear association between HER3 membrane expression and clinical responses has not been observed [20,21] and there may be determinants other than HER3 expression that contribute to the efficacy of HER3-DXd.

Tumor cell death triggered by the activity of an ADC requires the completion of multiple steps, including binding of the ADC to target antigens, internalization of the ADC-antigen complex via endocytosis, fusion of endosomes with lysosomes, linker cleavage and release of the cytotoxic payload [22,23]. Although the amount of payload released from HER3-DXd has been shown to correlate in some preclinical models with the baseline membrane expression of HER3 [24], HER3 expression can dynamically change after treatment with HER3-DXd and RTK inhibitors [24,25,26] and it is unclear how these dynamics influence the efficacy of HER3-DXd. In the current study, we investigated the potential impact of internalization and turnover rates on payload delivery in cancer cells to deepen our understanding of HER3 dynamics and its association with antitumor activity. In vitro and in vivo studies indicate that not only HER3 expression but also HER3 dynamics can affect payload release upon HER3-DXd treatment. Our results suggest the potential to improve the efficacy of HER3-DXd in patients through modulation of tumor expression of HER3, either by adjusting dosing regimens or by combining with other drugs that increase HER3 expression.

2. Results

2.1. HER3-DXd Is Rapidly Internalized into Early Endosomes

To obtain a better understanding of the potential for factors other than HER3 membrane expression to influence the antitumor activity of HER3-DXd, we examined internalization, payload release and DNA damage through in vitro studies using HER3-DXd-treated cell lines. We first investigated HER3-DXd internalization and trafficking using HER3-expressing cells by confocal imaging.

To detect the localization of HER3-DXd in early endosomes, cells were incubated with HER3-DXd for 1 h on ice and then incubated for 0–180 min at 37 °C. In MDA-MB-453, HER3-DXd bound to the cell surface of HER3 during incubation on ice and was detected only on the cell surface at 0 min (Figure 1A). Upon incubation at 37 °C, the localization of HER3-DXd shifted from the cell surface to the intracellular space, and colocalization with the early endosome antigen 1 (EEA1) signal was observed at 15 min and peaked at 30–60 min. Colocalization of HER3-DXd and the lysosome marker LAMP2 was observed at 60 min, suggesting subsequent trafficking of HER3-DXd from early endosome to lysosome [24]. Similar results were obtained with HCC2218 except for a slower shift in the subcellular localization of HER3-DXd (Figure 1B). In both cell lines, ADC was not detected in samples treated with control IgG-ADC (Supplementary Figure S1). This result suggests that HER3-DXd was internalized into cells in a HER3-dependent manner.

2.2. HER3 Dynamics Contribute to Payload Release from HER3-DXd In Vitro

To investigate whether HER3 dynamics can affect payload release upon HER3-DXd treatment, we examined in vitro payload release from cell lines with different HER3 dynamics. The level of cell surface HER3-DXd binding in HCC2218 was more than three times higher than that in MDA-MB-453 and COLO829 (Figure 2A). Cell surface HER3 was rapidly internalized in MDA-MB-453 and COLO829; internalization rates reached 73.2% and 79.1% at 1 h, respectively. In contrast, HCC2218 showed a more gradual decrease in the level of HER3 on the cell surface, corresponding to an internalization rate of 33.3% at 1 h and 62.9% at 24 h (Figure 2B). The slower internalization of HER3-DXd in HCC2218 than in MDA-MB-453 was consistent with the results of confocal imaging described earlier (Figure 1). After washout of HER3-DXd, the level of HER3 expression on the cell surface almost fully recovered to the basal level within 2 h in COLO829, whereas this took 5 to 24 h in MDA-MB-453 and HCC2218 (Figure 2C). As an indicator of the payload released from cells, we measured the payload concentration in the culture medium over time. The amount of payload released from HCC2218 was similar to that of MDA-MB-453 until 6 h, although the baseline HER3 expression level for MDA-MB-453 was less than a third of that for HCC2218. At 24 and 48 h, the payload release from HCC2218 was three times higher than that of MDA-MB-453; presumably, sufficient time had elapsed for the internalization of HER3 by HCC2218. A larger amount of payload was released from COLO829 than from MDA-MB-453 at all time points despite their similar baseline HER3 expression levels (Figure 2D). These results suggest that not only HER3 expression level but also HER3 dynamics can impact payload release in vitro.

2.3. HER3-DXd Internalization and Payload Release Lead to DNA Damage In Vitro

As our confocal microscopy imaging study and flow cytometry analysis demonstrated, HER3-DXd was rapidly internalized into cells and HER3 dynamics contributed to the amount of payload release. Next, we confirmed that the payload induces DNA damage through a confocal imaging study using MDA-MB-453 cells. γ-H2AX, a DNA damage marker, was observed at 1 h of HER3-DXd treatment, which is consistent with the decrease in HER3 on the cell surface (Figure 3A). When the cells were cultured with drug-free medium after 1 h of HER3-DXd treatment, time-dependent recovery of cell surface HER3 and decrease in γ-H2AX were observed (Figure 3B).

2.4. Adjustment of Dosing Interval Can Increase Tumor Payload Concentration In Vivo

To determine whether our in vitro findings could be confirmed in vivo, we performed xenograft model studies using mice inoculated with human-derived cancer cell lines COLO829- and MDA-MB-453. These mice were treated with one or two doses (the second injection at 24 or 72 h after the first injection) of HER3-DXd (1 or 3 mg/kg). HER3 expression level was evaluated by immunohistochemistry (IHC), and the membrane H-score was calculated. Because HER3 expression levels can vary after cells have been transplanted into mice, we confirmed that baseline HER3 expression in COLO829 was higher than that in MDA-MB-453 in vivo (Figure 4A). In the COLO829 model, the HER3 H-score was not significantly changed by the first injection and clearly decreased after the second injection of 3 mg/kg HER3-DXd (Figure 4A, Supplementary Figure S2). In the MDA-MB-453 model, a similar trend was observed when mice were treated with 1 mg/kg HER3-DXd. In contrast, HER3 expression in the MDA-MB-453 model decreased notably with one dose of 3 mg/kg HER3-DXd and remained at a low level at 24 h after the first injection.

With regard to payload release, payload was already released at 6 h post-injection in both models in the single-dose study (Figure 4B), indicating that HER3-DXd is rapidly internalized in vivo as well as in vitro. In the two-dose study, we also found that HER3 expression and dynamics had an impact on the amount of payload release, again similar to our findings in vitro. In the COLO829 model, an increase in tumor payload concentration was observed after the second injection of 3 mg/kg HER3-DXd (Figure 4B). Similar results were obtained in the MDA-MB-453 model that was treated with 1 mg/kg HER3-DXd. However, when the model was treated with 3 mg/kg HER3-DXd, tumor payload concentration did not significantly increase after the second injection at 24 h after the first injection. HER3 expression had recovered at 72 h after the first injection and the second injection led to a clear decrease in HER3, followed by a significant increase in tumor payload concentration.

These results demonstrate that both baseline HER3 expression level and HER3 dynamics can have an impact on the amount of payload release in vivo. These findings also indicate that both the dose and dosing interval of HER3-DXd can influence HER3 dynamics and may play a key role in achieving optimal efficacy.

2.5. Targeted Inhibition of Driver Oncogenes Upregulates HER3 Expression and Increases Payload Release from HER3-DXd

Because EGFR-TKIs have been shown to increase the cell surface expression of HER3 in EGFR-mutant cells [25,26], we next assessed whether EGFR-TKI-induced HER3 upregulation results in increased payload release from HER3-DXd-treated cells. Consistent with previous reports [25,26], HER3 expression in two EGFR-mutant NSCLC cell lines, PC-9 and HCC4006, was increased after 24 h of exposure to 10 nM osimertinib (Figure 5A). We next assessed the impact of osimertinib on payload release into cell culture medium following HER3-DXd treatment. Osimertinib is a third-generation EGFR-TKI approved for the treatment of EGFR-mutated NSCLC in first-line and additional clinical settings. As shown in Figure 5B, payload release from PC-9 was increased by osimertinib pretreatment.

To examine whether HER3 upregulation can occur in NSCLC cell lines upon targeted inhibition of other driver genomic alterations, we evaluated cell surface HER3 expression of HCC-78 (ROS1 fusion) and NCI-H2228 (ALK fusion) cells before and after treatment with lorlatinib and ceritinib, respectively. Lorlatinib is an ALK/ROS1-TKI and ceritinib is an ALK-TKI. Both drugs are approved for patients with metastatic NSCLC whose tumors are ALK-positive. TKI treatment led to increased cell surface levels of HER3 in both cell lines (Figure 5C). These findings indicate that targeted therapies other than EGFR-TKIs also have the potential to modify HER3 expression and augment payload release after HER3-DXd treatment.

3. Discussion

HER3-DXd monotherapy has demonstrated clinically meaningful antitumor activity in studies of breast cancer and lung cancer [20,21,27], and a combination study with osimertinib in EGFR-mutated NSCLC patients and monotherapy studies in patients with other solid tumors are ongoing. In the current study, we investigated the impact of HER3 dynamics and HER3 upregulation induced by TKIs to develop strategies to potentiate the antitumor activity of HER3-DXd. We demonstrated for the first time that HER3 internalization and turnover speed as well as baseline HER3 expression influenced payload release from cells treated with HER3-DXd. Targeted inhibition of driver oncogenes that induced HER3 expression also resulted in increased payload release from HER3-DXd. In addition, HER3-DXd administration altered the cell surface expression of HER3 in vivo, contributing to the amount of payload release following a second dose of HER3-DXd.

After HER3 internalization with the first HER3-DXd treatment, the amount of cell surface HER3 did not change during continued exposure to HER3-DXd (see Figure 2B and Figure 3A). We postulate that HER3 internalization had not ceased, but rather that HER3 internalization and turnover rate were in equilibrium. When HER3-DXd was removed from the culture medium (in vitro) or when the blood HER3-DXd concentration decreased (in vivo), the equilibrium appeared to shift in favor of turnover and cell surface HER3 expression recovered (Figure 2C, Figure 3B and Figure 4A). In the in vitro internalization/turnover study, faster turnover of HER3 in COLO829 may be one of the factors contributing to a higher concentration of payload than for MDA-MB-453, which showed a similar baseline expression of HER3. Payload concentration in the two cell lines increased gradually over time. As for HCC2218, there was a slight increase in payload concentration from 3 h to 6 h, whereas it increased significantly from 6 h to 24 h and 24 h to 48 h. Slower change in the internalization/turnover rate of HCC2218 may have resulted in the slower timing of payload increase compared to the other two cell lines. It has been reported that target expression and internalization/turnover profile define the capacity of cells to incorporate ADC and that plasma ADC levels in excess of this capacity do not result in additional payload delivery [28]. Our imaging data also suggest that continuous HER3-DXd internalization is important for continued DNA damage (Figure 3). For the efficient delivery of HER3-DXd into cells, it might be worthwhile administering second and subsequent doses when cell surface HER3 has recovered sufficiently. In the in vivo models described herein, the increase in tumor payload concentration after the second dose at 24 h after the initial dose of 3 mg/kg HER3-DXd was much higher in the COLO829 model than in the MDA-MB-453 model, although plasma ADC concentration levels at that time point were similar between the two models (Supplementary Figure S3). In the MDA-MB-453 model, which exhibited moderate baseline HER3 expression, the plasma HER3-DXd concentration at that time point may have still been sufficient to equilibrate HER3 internalization/turnover rates. As a result, the second dose at that timing may not have contributed to payload increase in this model.

We found that one dose of 3 mg/kg HER3-DXd did not significantly decrease cell surface HER3 in the COLO829 model; however, significant decreases were observed when a second dose was administered at either 24 h or 72 h following an initial dose. It is possible that with a higher dose of HER3-DXd administered to the COLO829 model, HER3 expression might be decreased with the first dose, as was the case with 3 mg/kg treatment in the MDA-MB-453 model. Based on these results, we postulate that the amount of reduction in cell surface HER3 after HER3-DXd administration may depend on the balance of baseline HER3 expression level and HER3-DXd concentration. Modeling and simulations incorporating the amounts of HER3 and HER3-DXd may help identify conditions that maximize payload release.

It is important to note that there are limitations to our study since we obtained HER3 dynamics data with a small number of cell lines. In addition, we did not evaluate the impact of payload increase on tumor reduction. It has previously been reported that not only target expression and dynamics but also other factors, including payload sensitivity of cancer cells [1] and tumor microenvironment [29], may contribute to the antitumor efficacy of ADCs. Investigation into these factors in addition to HER3 dynamics may therefore deepen our understanding and overcome the current lack of a clear association observed between HER3 membrane expression and clinical response to HER3-DXd [27]. As observed in cell line studies using TKIs, HER3 expression levels can be increased by the inhibition of driver oncogenes. These results provide a non-clinical rationale for ongoing investigations with the combination of HER3-DXd and osimertinib (study U31402-A-U103; clinicaltrials.gov NCT04676477). Future investigation may help clarify the mechanisms by which targeted therapies increase HER3 expression and aid the development of additional combination strategies for HER3-DXd. Regarding osimertinib, the contributions of both protein stability and mRNA transcription to increased cell surface expression of HER3 were reported in a previous study [25].

In summary, we found that HER3 dynamics as well as baseline HER3 expression have an impact on payload release from HER3-DXd. In addition, this is the first study showing that targeted inhibition of driver oncogenes increased cell surface expression of HER3 and payload release, supporting a combination strategy to potentiate the antitumor activity of HER3-DXd. These results suggest the potential to improve the efficacy of HER3-DXd in patients through modulation of tumor expression of HER3, either by adjusting dosing regimens or by combining with other drugs that increase HER3 expression. Performing similar studies for other ADCs has the potential to inform new strategies to augment antitumor activity.

4. Materials and Methods

4.1. Antibody–Drug Conjugates

HER3-DXd was synthesized in accordance with the published procedure [19], conjugating the exatecan derivative-based cytotoxic payload, DXd, with a naked anti-HER3 antibody (patritumab, U3-1287). Control IgG-ADC (anti-LPS antibody) was synthesized in the same manner as HER3-DXd using matched isotype mAb, resulting in a comparable drug-to-antibody ratio.

4.2. Cell Lines

MDA-MB-453, HCC2218, COLO829, HCC4006 and NCI-H2228 were purchased from ATCC. PC-9 was purchased from ECACC, and HCC-78 was purchased from Leibniz Institute DSMZ. MDA-MB-453 was cultured with Leibovitz L-15 medium (#11415; Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% fetal bovine serum (#SH30396.03; HyClone Laboratories, Logan, UT, USA) in a humidified CO₂-free incubator (Forma Steri-Cult #3307; Thermo Fisher Scientific) at 37 °C. All other cell lines were cultured with RPMI1640 (#11875; Thermo Fisher Scientific) supplemented with 10% FBS in a humidified 5% CO₂ incubator at 37 °C.

4.3. Confocal Microscopy Imaging

Cell suspension of MDA-MB-453 or HCC2218 was incubated with HER3-DXd on ice or at 37 °C. After harvesting, cells were fixed and permeabilized and then incubated with the primary antibody cocktail, the secondary antibody cocktail and DAPI solution. Fluorescent images were acquired using a confocal microscope. See Supplementary Materials and Methods for further information.

4.4. Binding, Internalization and Turnover Assessed by Flow Cytometry

Cells were seeded on plates and incubated with HER3-DXd on ice or at 37 °C. Cell surface HER3 expression change in cell lines was evaluated by flow cytometry. See Supplementary Materials and Methods for further information.

4.5. Payload Measurement

4.5.1. Sample Collection

Cells were seeded in 24-well plates and cultured at 37 °C until culture medium collection. MDA-MB-453, HCC2218, and COLO829 cells were treated with 10 nM HER3-DXd for 3–48 h. PC-9 cells were treated with 10 nM osimertinib or DMSO followed by 10 nM HER3-DXd treatment for 1–24 h. See Supplementary Materials and Methods for detailed information on payload measurement using LC-MS/MS.

4.5.2. Cell Viability Assay

Regarding MDA-MB-453, HCC2218, and COLO829, after collecting the medium for payload measurement, cells treated with HER3-DXd for 24 or 48 h were subjected to a cell viability assay. One milliliter of fresh medium was added to each well, and ATP levels were measured using the CellTiter-Glo 2.0 Assay (#G9243; PROMEGA Corporation., Madison, WI, USA) with an EnVision multimode plate reader (#2105; PerkinElmer Inc., Shelton, CT, USA). For the control samples, cells were prepared separately from those used for the payload measurement assay and subjected to the cell viability assay without the addition of fresh medium.

Regarding PC-9, cells were seeded and treated in a separate plate following the same procedure as for the payload measurement. HER3-DXd was added only 3 and 24 h prior to sample collection. After treatment with osimertinib and HER3-DXd for the specified durations, ATP levels in each well were measured as described above.

4.5.3. Payload Concentration Correction

Since cell proliferation rates and HER3-DXd sensitivity vary among cell lines, the payload measurement results for MDA-MB-453, HCC2218, and COLO829 at 24 and 48 h of treatment were corrected based on the growth rate of each cell line under respective conditions. The relative cell survival rate was calculated using ATP signal intensity after each treatment, as described in Section 4.5.2. Payload concentration correction was performed using the following equations in Microsoft Excel 2013 (Microsoft Corp., Redmond, WA, USA). All assays were conducted in sextuplicate.

Corrected payload conc. (nM) = Payload conc. measurement (nM)/(T/C).

T: ATP signal intensity of each treatment sample.

C: ATP signal intensity of control sample (Day 0).

A similar correction based on cell growth rate was applied for PC-9, given that osimertinib inhibits PC-9 cell growth. ATP signal intensity results from 3 h and 24 h treatments were used to compensate for payload measurement results at 1 and 3 h, and 18 and 24 h of treatment, respectively. Payload concentration correction was performed using Microsoft Excel 2013. The same equations as described above were applied, with a different definition of C, which was defined as the ATP signal intensity of the control sample (DMSO 24 h + HER3-DXd 3 h). All assays were conducted in sextuplicate.

4.6. Xenograft Studies

4.6.1. Animals

Mature female CAnN.Cg-Foxn1nu/CrlCrlj mice were purchased from Charles River Laboratories Japan, Kanagawa, Japan. Upon arrival, mice were acclimatized for four or five days with standard light/dark cycles. Mice were housed in specific pathogen-free facilities with free access to food and water. All animal experiments performed in this study were approved by the Institutional Animal Care and Use Committee at Daiichi Sankyo Co., Ltd., Tokyo, Japan.

4.6.2. Experimental Design

Models were established by injecting 1 × 10⁷ cells (MDA-MB-453) suspended in Matrigel and 5 × 10⁶ cells (COLO829) suspended in saline subcutaneously into CAnN.Cg-Foxn1nu/CrlCrlj mice. Group assignment was carried out without randomization when the tumor volume reached approximately 150 to 300 mm³. Total of 75 (MDA-MB-453) and 30 (COLO829) tumor-bearing mice were injected with HER3-DXd intravenously on day 0 for the single administration group and days 0 and 1 or days 0 and 3 for the repeated dose group. The doses suitable for each model were determined based on our preliminary studies. Since 1 mg/kg was considered inappropriate for the COLO829 model due to its higher HER3 expression compared to MDA-MB-453, only 3 mg/kg was evaluated in the COLO829 model. Plasma and tumor tissues were collected under isoflurane anesthesia at 6, 24, 72, and 168 h after the first and second injections in the MDA-MB-453 model; and at 6, 24, 48, 72, and 96 h after the first injection and 6 and 24 h after the second injection in the COLO829 model. Animals were sacrificed after sample collection under isoflurane anesthesia. Half of each excised xenograft tumor was formalin-fixed and paraffin-embedded, and tissue sections were used for IHC. Plasma and the other half of the tumors were used for measurements of the payload and HER3-DXd concentrations and for measurement of the payload concentration, respectively. Concentrations of payload and HER3-DXd were determined by LC-MS/MS and a ligand binding assay using Gyrolab xP workstation, respectively. See Supplementary Materials and Methods for further information on IHC, payload concentration, and HER3-DXd concentration.

In accordance with the predetermined criteria, mice with tumor sizes unsuitable for this experiment were excluded before grouping. No animals were euthanized due to poor health, nor were any samples excluded from the analysis. The experimental protocol was prepared before the study. The technical personnel caring for the mice were unaware of the experimental design. To maintain consistency, all experimental procedures within each model were conducted at the same time. The experimental unit was a single animal. The sample size was determined based on considerations of variability.

4.7. Statistical Analysis

Statistical analysis was performed using GraphPad Prism (version 9.1.0; GraphPad Software, Inc., La Jolla, CA, USA) or Microsoft Excel 2013. Results were compared using one-way ANOVA with Tukey’s multiple comparison tests or a Student’s t-test. Statistical significance was set at p < 0.05.