Design and Evaluation of Phosphonamidate-Linked Exatecan Constructs for Highly Loaded, Stable, and Efficacious Antibody–Drug Conjugates

Abstract

Topoisomerase I (TOP1) Inhibitors constitute an emerging payload class to engineer antibody–drug conjugates (ADC) as next-generation biopharmaceutical for cancer treatment. Existing ADCs are using camptothecin payloads with lower potency and suffer from limited stability in circulation. With this study, we introduce a novel camptothecin-based linker–payload platform based on the highly potent camptothecin derivative exatecan. First, we describe general challenges that arise from the hydrophobic combination of exatecan and established dipeptidyl p-aminobenzyl-carbamate (PAB) cleavage sites such as reduced antibody conjugation yields and ADC aggregation. After evaluating several linker–payload structures, we identified ethynyl-phosphonamidates in combination with a discrete PEG24 chain to compensate for the hydrophobic PAB–exatecan moiety. Furthermore, we demonstrate that the identified linker–payload structure enables the construction of highly loaded DAR8 ADCs with excellent solubility properties. Head-to-head comparison with Enhertu, an approved camptothecin-based ADC, revealed improved target-mediated killing of tumor cells, excellent bystander killing, drastically improved linker stability in vitro and in vivo and superior in vivo efficacy over four tested dose levels in a xenograft model. Moreover, we show that ADCs based on the novel exatecan linker–payload platform exhibit antibody-like pharmacokinetic properties, even when the ADCs are highly loaded with eight drug molecules per antibody. This ADC platform constitutes a new and general solution to deliver TOP1 inhibitors with highest efficiency to the site of the tumor, independent of the antibody and its target, and is thereby broadly applicable to various cancer indications.

Introduction

With a total of 12 approved antibody–drug conjugates (ADC) up to the present day and not less than 8 approvals by the FDA only in the last four years, ADCs for the treatment of cancer are coming of age (1, 2). Drug classes that are being used in those marketed ADCs include DNA-damaging and microtubule-disrupting agents (3). However, one class that stands out in terms of clinical efficacy and market success is of topoisomerase I (TOP1) inhibitors, used in trastuzumab deruxtecan (Enhertu), sacituzumab govitecan (Trodelvy) (4) and many other preclinically and clinically investigated ADCs (5–7). Next to the application as ADC payloads, which this publication is focusing on, camptothecin and its derivatives have been studied for decades as small-molecule therapeutics in cancer treatment (8).

In particular, camptothecin analogs such as DXd, a derivative of exatecan used in Enhertu, and SN38, the active metabolite of irinotecan used in Trodelvy, have been contributing to the success of the TOP1 inhibitor payload family as ADC payloads (see Fig. 1 for structures). Trodelvy has been shown to significantly prolong overall survival (OS) compared with single-agent chemotherapy in difficult to treat patients with estrogen receptor–positive, HER2-negative breast cancer (9). Enhertu demonstrated superior efficacy, with a significantly longer progression-free survival in patients with breast cancer compared with trastuzumab-DM1 (Kadcyla) treatment (10). Notably, this tremendous advantage is achieved with the same antibody and target, only changing the linker–payload from the tubulin-disrupting agent DM1 to a TOP1 inhibitor. Moreover, Enhertu has been changing the paradigm of HER2-positivity, because it showed remarkable activity in patients with HER2-low breast cancer, a patient population previously considered as being not amenable to HER2-based treatments (11).

Figure 1.

Overview of TOP1 inhibitor-based linker–payloads used in marketed ADCs Enhertu and Trodelvy (top), based on the payloads DXd and SN38 (2) and the structure of the linker–payloads described herein based on exatecan with an overview over basic features of the ADCs originating from those linker systems. Payload structures are highlighted in orange.

Despite those remarkable achievements in the treatment of patients with cancer with ADCs using camptothecin-based payloads, there is also still room for improvement. Even though ADCs are in theory highly targeted medications, treated patients are still suffering from marked signs of toxicity (12) with neutropenia and severe diarrhea being the most common adverse event leading to dose reductions for Trodelvy and pneumonitis and interstitial lung disease being the most common reasons for treatment discontinuation for Enhertu (1). Even though mechanisms of ADC-mediated toxicities are complex and not yet fully understood, it is discussed throughout the literature that the linker that connects the cytotoxic drug with the antibody plays a crucial role in how efficacy and toxicity is balanced within a given ADC (13–16). Critical requirements on a linker include plasma stability (15, 17) and inclusion of hydrophilicity to compensate for the typically hydrophobic drug to prevent undesired ADC aggregation and accelerated plasma clearance (18, 19).

Despite being a highly promising ADC payload class, the conjugation of the camptothecin moiety, including its derivatives has been shown to be in particular challenging in the past, especially in higher drug-to-antibody ratios (DAR). Burke and colleagues (20) have shown that antibody conjugates with valine–citrulline–PAB [VC–PAB (dipeptidyl p-aminobenzyl-carbamate)]–camptothecin derivatives exhibit a strong tendency to form higher molecular weight species (HMWS). This issue could only be solved by DAR reduction and substitution of the VC-cleavage site with a more polar glucuronide (20). Similar challenges in induction of aggregation, once conjugated to an antibody, have been described for several exatecan-based derivatives (21). Further improvements in the linker structure such as the implementation of a more polar hemiaminal-based self-immolative moiety led to the discovery of the DXd-containing linker used in Enhertu (22, 23). This linker strategy has also been applied to similar camptothecin analogs, all of which showing decent activity in vitro, including bystander killing (24). In another study, hydrophilic polysarcosins have been added to the linker to compensate for the hydrophobic exatecan moiety (25). It should be noted that all those examples exhibit maleimide linker chemistry for conjugation, whereas a camptothecin-based linker–payload platform that provides fully in vivo-stable antibody conjugates of a high DAR has still been unprecedented until this study.

Unsaturated phosphorous(V)-based electrophiles have been recently introduced as new cystein-selective bioconjugation reagents for the generation of highly efficacious ADCs. We have shown that they can easily be incorporated even into complex molecules and they yield highly stable cysteine adducts with a high selectivity for thiols (no modification of other amino acids than cysteine can be observed) and fast reaction kinetics in the range of bioorthogonal strain promoted cycloaddition reactions (26–30). Furthermore, ethynylphosphonamidates provided in combination with solubility enhancers high DAR ADCs of hydrophobic VC–PAB–MMAE linker–payloads without increased in vivo clearance (31, 32). High DARs are important for camptothecin payloads, in particular, because high DAR can compensate for the lower potency compared with other ADC payloads such as auristatins or calicheamycins (4).

Our goal in the current study was to design a novel TOP1 inhibitor-based linker–payload platform that facilitates aggregation-free construction of highly loaded DAR8 ADCs with excellent serum stability, without enhanced in vivo clearance rates and potent and selective in vitro and in vivo efficacy. For this, we used exatecan as payload and conducted chemical optimization of the linker to facilitate aggregation-free ADCs even at high DARs of 8. The optimal linker–payload structure identified carries the well-established VC–PAB cleavage site for traceless intracellular release of the payload (33) and ethynylphosphonamidates equipped with polyethylene glycol (PEG) chains of discrete length of 24 units. The PEG chain is thereby attached in a branched fashion to ensure close distance between antibody and payload. We demonstrate that DAR8 ADCs based on this linker–payload structure are superior in head-to-head comparisons with Enhertu in a broad panel of in vitro and in vivo experiments. Moreover, we observed excellent bystander killing as well as release of damage-associated molecular patterns (DAMP) as an indirect measure of immunogenic cell death (ICD) of our linker–payload, two features that are discussed as key drivers of efficacy of ADCs (34–37). Finally, we demonstrate excellent in vivo features of our newly identified linker–payload structure, such as a tremendous increase in efficacy compared with Enhertu and antibody-like pharmacokinetics (PK) properties of the highly loaded ADCs with retained DAR8 even after 21 days of circulation.

Materials and Methods

In vitro cytotoxicity

To investigate the cytotoxicity of the unconjugated small-molecule TOP1 inhibitors and ADCs, respective cells were incubated for 4 days (small molecules) and 7 days (ADCs) with increasing concentrations of small molecules (0.015–1,000 nmol/L) and ADCs (0.05–3 μg/mL) to generate a dose–response curve. Killing was analyzed using resazurin cell viability dye at a final concentration of 55 μmol/L (Merck) by dividing the fluorescence from control cells in medium by the fluorescence of ADC-treated cells. Fluorescence emission at 590 nmol/L was measured on a microplate reader Infinite 200 Pro (Tecan Group Ltd.).

Optimized procedure for the conjugation of LP5 to antibodies to achieve DAR8

50 μL of the antibody solution at 10.0 mg/mL in P5-conjugation buffer (50 mmol/L Tris, 1 mmol/L EDTA, 100 mmol/L NaCl, pH 8.3 at room temperature) were mixed with 1.66 μL of a 10 mmol/L TCEP [Tris(2-carboxyethyl)phosphine] solution (5 eq.) in P5-conjugation buffer. Directly afterwards, 0.83 μL of a 40 mmol/L solution of LP5 dissolved in DMSO (10 eq.) were added. The mixture was shaken at 350 rpm and 25°C for 16 hours. The reaction mixtures were purified by preparative size-exclusion chromatography (SEC) with a Superdex 200 Increase 10/300GL (Cytiva) and a flow of 0.8 mL/min eluting with sterile PBS (Merck). The antibody-containing fractions were pooled and concentrated by spin-filtration (Amicon Ultra 2mL MWCO: 30 kDa, Merck).

Determination of antibodies bound per cell

Antibodies bound per cell (ABC) were determined by staining of cell lines with saturating concentrations of HER2-PE (BioLegend) or the respective isotype control and acquisition by flow cytometry on a CytoFLEX S flow cytometer. Quantification of ABCs as an estimation for surface receptor expression was performed by simultaneous acquisition of QuantiBRITE beads (BD Biosciences) conjugated with four levels of PE and generation of a standard curve.

Analysis of HER2-binding of trastuzumab-LP5 and enhertu

To determine equilibrium-binding constants (K_D; as an avidity measurement), HER2-positive SKBR-3 cells were incubated with antibodies at increasing concentrations up to saturation and stained with an AF488-labeled goat anti-human IgG (H+L) secondary antibody (Thermo Fisher Scientific). Cells were acquired by flow cytometry on a CytoFLEX S flow cytometer (Beckman Coulter). Mean fluorescence intensity (MFI) ratios were determined by dividing the MFI of the antibody-incubated cells by the fluorescence of the secondary antibody control. Data points were analyzed by a non-linear regression using a one-site–specific binding model.

Analysis of HER2-mediated internalization of trastuzumab-LP5 and enhertu

For pHrodo-based investigation of internalization, antibodies were labeled with the pHrodo Deep Red Antibody Labeling Kit (Thermo Fisher Scientific) according to the manufacturer's instructions. HER2-positive and -negative cell lines were incubated with 5 μg/mL pHrodo Deep Red-antibodies for 1 hour at 37°C and acquired by flow cytometry on a CytoFLEX S flow cytometer (Beckman Coulter). An increase in MFI indicates the presence of αHER2 antibodies in late endosomal and lysosomal compartments. The MFI ratio was determined by dividing the MFI of pHrodo-incubated cells by the MFI of unstained cells.

Bystander killing

For the co-culture–based bystander experiment, 10.000 HER2-positive SKBR-3 and 2.000 HER2-negative MDA-MB-468 cells were treated with trastuzumab-LP5 DAR8 or Enhertu at concentrations ranging from 0.05 to 3 μg/mL. After 5 days, cells were stained with LIVE/DEAD Fixable Aqua (Thermo Fisher Scientific) and HER2-PE (BioLegend) to distinguish between target-positive and -negative cells. The percentage of viable HER2-positive and -negative cells was analyzed by flow cytometry on a CytoFLEX S flow cytometer (Beckman Coulter).

DNA damage

The upregulation of DNA damage markers as response to TOP1-inhibiting ADCs were analyzed by a flow cytometry–based readout. For this, 30,000 HER2-positive SKBR-3 cells were incubated for 24 hours to 72 hours with 0.5 or 5 μg/mL of ADCs. After the end of the incubation time, cells were stained with LIVE/DEAD Fixable Aqua (Thermo Fisher Scientific) followed by fixation and permeabilization using the BD Cytofix/Cytoperm Kit (BD Biosciences) according to the manufacturer's instructions. Intracellular staining of DNA damage markers was performed using anti-cleaved PARP (Asp214) PE, anti-H2AX (pSer139) AF647, and anti-active caspase 3 FITC (all BD Biosciences) or respective isotype controls. Cells were acquired on a CytoFLEX S flow cytometer (Beckman Coulter) and the percentage of positive cells was determined.

Cell-cycle analysis

To investigate changes in cell cycle upon TOP1 inhibitor treatment, 30,000 HER2-positive SKBR-3 cells were incubated for 24 to 48 hour with 20 nmol/L of ADCs (3 μg/mL) or small molecules if not otherwise indicated. After the end of the incubation time, cells were permeabilized in 70% ethanol on ice, followed by staining in FxCycle PI/RNAse Staining Solution (Thermo Fisher Scientific). Cells were acquired on a CytoFLEX S flow cytometer and PI fluorescence was analyzed on a linear scale (Beckman Coulter).

Release of DAMPs as a surrogate marker for ICD in vitro

As an indirect measure for ICD, calreticulin-positive cells, HMGB1 and ATP release were quantified. For all readouts, 30,000 HER2-positive cells were incubated with 20 nmol/L of exatecan, DXd or derived ADCs (3 μg/mL) for 24 to 48 hours if not otherwise indicated. Calreticulin staining of cells was performed using PE anti-calreticulin antibody (Abcam) or respective isotype control. Lumit HMGB1 Human/Mouse Immunoassay (Promega) and RealTime-Glo Extracellular ATP Assay (Promega) were used to quantify secreted HMGB1 and ATP into the media according to the manufacturer's instructions. Measurement of luminescence was performed on a microplate reader Infinite 200 Pro (Tecan Group Ltd.).

In vivo efficacy experiments

The experiments were conducted in accordance with German animal welfare law and approved by local authorities. A total of 2×10⁶ NCI-N87 cells were subcutaneously injected into the flank of female CB17-Scid mice. Treatment was initiated when tumors reached a mean tumor volume of 0.1–0.15 cm³. 10 animals per group were treated once with either 0.25, 0.5, 1 or 2 mg/kg, trastuzumab-LP5 DAR8 or Enhertu. 5 animals per group were treated with vehicle or with 2 mg/kg palivizumab-LP5 as Isotype control, as intravenous injection after randomization into treatment and control groups. Tumor volumes, body weights, and general health conditions were recorded throughout the whole study.

Data availability

The data generated in this study are available upon request from the corresponding author.

Results

Design and evaluation of TOP1-based linker–payloads for highly loaded and efficacious ADCs

We started our investigation with the selection of an appropriate camptothecin derivative. For this, we tested the unconjugated payloads, highlighted in orange in the linker–payload structures depicted in Fig. 1, for their antiproliferative potency on a range of different cell lines (Fig. 2; Supplementary Table S1).The figure clearly shows that of the three TOP1 inhibitors tested in this study, exatecan is the most potent with a factor of 2–10-fold potency difference in the IC₅₀ value of cell viability compared with SN38 and DXd. Next, a series of different linker–payloads, shown in Fig. 3 based on exatecan, including a maleimide-control ADC, has been synthesized. The dipeptidyl-based linker–exatecan constructs LP1–5 carry the well-known VC and valine–alanine (VA) peptides in combination with the self-immolative PAB moiety that facilitates traceless release of exatecan (38). The linker–payload LP6 based on maleimide chemistry as the most commonly used conjugation strategy was generated as a control. To assess whether the exatecan moiety must be tracelessly and efficiently cleaved to unleash its full activity, we also synthesized construct LP7 without the PAB moiety and the non-cleavable LP8. The absence of the lipophilic PAB moiety in LP7 and LP8 has the advantage of reduced linker–payload hydrophobicity to potentially reduce ADC aggregation.

Figure 2.

Dose–response curves for DXd, SN38, and exatecan on a range of different cell lines.

Figure 3.

Depiction of the linker–payload (LP) structures used herein. Functional elements are highlighted. Hydrophilic substituent, purple; conjugation handle, gray; cleavage side, green; release handle, orange; exatecan, black.

With LP1–8 in hands, we continued our studies with conjugation of the linker–payload to trastuzumab, a HER2-targeting antibody of the same amino acid sequence that is used in the antibody of Enhertu (39). Antibody conjugation has been conducted in accordance to previously published procedures (31). A large excess of the linker–payloads and TCEP reduction agents has been used in those first conjugation experiments to force the conjugation reaction and ensure the highest possible conjugation yield. After conjugation and purification from residual linker–payload, TCEP, and HMWS via semipreparative SEC into PBS, the corresponding ADCs were analyzed for yield based on protein concentration before and after the conjugation, DAR via mass spectrometry (MS), formation of undesired HMWS via SEC and targeted antiproliferative effect on two different HER2-positive (SKBR-3 and HCC-78) and one HER2-negative cell line (Table 1; Supplementary Fig. S1).

Table 1.

Overview of the properties of the ADCs, synthesized from the HER2-targeting trastuzumab conjugated to LP1–8.

					Cell killing SKBR-3		Cell killing HCC-78
LP	Linker–payload	Yield after purification based on protein concentration	DAR (MS) After purification from HMWS	HMWS after conjugation (%)	EC50 (ng/mL)	Max. amount of dead cells	EC50 (ng/mL)	Max. amount of dead cells
LP1	P5(OEt)–VC–PAB–exatecan	59%	0.28	9.9%	251	37%	>3,000	N/A
LP2	P5(PEG2)–VC–PAB–exatecan	82%	5.51	2.3%	n.d.	n.d.	n.d.	n.d.
LP3	P5(PEG12)–VC–PAB–exatecan	80%	6.36	3.9%	15.5	86%	144.1	91%
LP4	P5(PEG12)–VA–PAB–exatecan	84%	4.39	7.5%	18.0	78%	121.2	68%
LP5	P5(PEG24)–VC–PAB–exatecan	99%	8.00	0.9%	12.5	96%	97.6	88%
LP6	MC–VC–PAB–exatecana	75%	0.12	19.2%	265	14%	>3,000	N/A
LP7	P5(PEG12)–VA–exatecan	77%	7.89	1.0%	114.7	44%	>3,000	N/A
LP8	P5(PEG12)–exatecan	97%	7.19	4.0%	182.4	35%	>3,000	N/A

Note: The conjugation has been performed with excess linker–payload (30 eq. with respect to the antibody) and TCEP (10 eq. with respect to the antibody). After purification, the ADCs have been analyzed for yield based on protein concentration before and after the conjugation, DAR via MS, ADC aggregation via analytic SEC and antiproliferative effect on HER2-positive cell-lines (SKBR-3 and HCC-78). The antiproliferative potency of the ADCs on the HER2-positive cells is reported as an EC₅₀ value of cell viability in ng/mL and the total amount of affected cells in percentage. None of the ADCs showed any effect on the HER2-negative cell line (MDA-MB-468, Supplementary Fig. S1).

^aConjugation conditions for LP6 were slightly adjusted to more typical maleimide reaction conditions: Reaction was conducted in PBS at pH 7.4 at room temperature for 2 hours; n.d., not determined; N/A, not applicable.

Interestingly, after conjugation of the phosphonamidate-O-ethyl–substituent linker–payload LP1 to trastuzumab, we observed only a minor conversion of the antibody to a DAR of 0.28, even with a large excess of 30 equivalents of LP1 with respect to the antibody. In addition to this, we observed a high percentage of HMWS formation of almost 10% during the conjugation process (Table 1). To exclude slower reaction kinetics of thiol addition to ethynylphosphonamidates, compared with very fast maleimide conjugation (30, 40), being the reason for this low conjugation yield, we also tried the maleimide-based linker–payload LP6 for conjugation. However, also LP6 only resulted in a very minor conjugation to a DAR of 0.12 in the monomer fraction after conjugation with almost 20% of HMWS being formed (Table 1).

Table 1 shows that incorporating PEG chains into the VC–PAB–exatecan-based linker–payload systems can improve the conjugation efficiency and simultaneously reduce aggregation of the derived ADCs. Both effects are depending on the PEG length, as shown by linker–payloads LP2, LP3, and LP5, with 2, 12, and 24 PEG monomer units, respectively. It could be demonstrated that an increase in PEG length led to a higher DAR, less aggregation during the conjugation process and consequently to an increase in the yield of monomeric protein after conjugation and to an increase in in vitro efficacy on the targeted cell lines (Table 1; Supplementary Fig. S1). Switching from a VC LP3 to a VA LP4 did not improve the conjugation behavior (Table 1). Only the PEG24-carrying linker–payload LP5 led to full conjugation of trastuzumab to a DAR8 ADC without any sign of HMWS being formed during the conjugation process resulting in the highest in vitro cytotoxicity on HER2-positive SKBR-3 and HCC-78 among all tested linker–payloads. We also synthesized ADCs from linker–payload LP7, which is lacking the PAB moiety. Interestingly, LP7 yields ADCs with a very high DAR and almost no HMWS being formed during conjugation, in contrast with LP4, which is the same linker–payload structure but includes the PAB moiety. However, we observed that ADCs from LP7 and LP8, even though high in DAR, showed only modest anticancer activity in the antigen-positive cell lines (Table 1; Supplementary Fig. S1).

Biophysical and in vivo characterization of the two best performing exatecan-based linker–payloads

Next, we continued our efforts in designing novel camptothecin-based linker–payload structures with a more thorough in vitro and in vivo investigation of ADCs originating from linker–payload LP3 and LP5, carrying either 12 or 24 PEG units, respectively. To allow for a fair comparison between the linker systems, DAR8 ADCs were synthesized out of both linker–payload structures (Fig. 4A). For this, equivalents of LP3 were increased from 30, that were used to produce the results in Table 1, to 50 and the monomeric ADCs were isolated via preparative SEC. The desired ADCs were characterized via SEC and hydrophobic interaction chromatography (HIC; Fig. 4B). Trastuzumab-LP5 DAR8 demonstrated best characteristics, because it had the highest degree of monomer content compared with trastuzumab-LP3 and Enhertu, judged by SEC. Interestingly, the overall hydrophobicity increase of the DAR8 constructs compared with the unconjugated antibody was negligible, judged by the HIC retention time (Supplementary Fig. S2) and similar HIC retention times were measured for trastuzumab-LP3 DAR8, trastuzumab-LP5, and Enhertu (Fig. 4B). The aggregation behavior of trastuzumab-based DAR8 ADCs from LP3 and LP5 was more thoroughly monitored over several weeks at low (4°C) and physiological temperatures (37°C) and Enhertu was added as a comparison (Fig. 4C; Supplementary Fig. S2). The results clearly show that ADCs conjugated to LP3 have a stronger tendency to aggregate under both conditions compared with Enhertu and ADCs conjugated to LP5. It should be noted that DAR8 ADCs from LP5 showed the least tendency to form HMWS under the tested conditions, even compared with Enhertu.

Figure 4.

Head-to-head comparison between DAR8 ADCs from LP3 and LP5 and Enhertu. A, Structure of DAR8 ADCs from LP3 and LP5. B, Comparison of the physicochemical properties of trastuzumab-LP5 (orange), trastuzumab-LP3 (purple), and Enhertu (gray). SEC analysis for the formation of aggregation (top) and HIC analysis to investigate overall hydrophilicity of the DAR8 constructs (bottom). C, Monitoring of increase in aggregation over 4 weeks of trastuzumab-LP3 (orange), trastuzumab-LP5 (purple), and Enhertu (gray). All three ADCs were formulated in the same buffer system at 1 mg/mL. Samples were analyzed by SEC for HMWS after 1, 2, and 4 weeks of incubation at 4°C (left) and 37°C, right. Raw SEC chromatograms are shown in Supplementary Fig. S3. Aggregates were normalized to day 0 by subtraction, because trastuzumab-LP5, already contained 8% HMWS at day 0. D, PK analysis of Sprague Dawley rats that were treated at day 0 with brentuximab-LP5 (left, purple) or IgG1-LP5 (right, orange) at day 0. Samples were drawn after several time points throughout the 3 week study period (9 animals per group, 3 animals were sampled per time point). Shown are mean and SD out of the three animals. The serum samples were analyzed by ELISA for total mAb (dotted line) and intact ADC (straight line). Black dotted line corresponds to the lower limit of quantification of the assays.

To compare ADCs from LP3 and LP5 in their in vivo PK behavior, Sprague Dawley rats were treated with ADCs carrying 8 linker–payload molecules, and their clearance behavior was monitored over time via analysis of blood samples. It must be noted that both experiments have been performed with different antibodies conjugated to LP3 and LP5. However, because both antibodies are not cross-reactive in rodents, it is anticipated that the influence on the clearance behavior of the ADCs can be neglected. The samples at various time points were analyzed by ELISA for total ADC with an antibody capture step and for intact ADC, capturing with an anti-exatecan antibody. The high overlap between those two curves again highlights the great serum stability of the ehynylphosphonamidate linker structure and is in line with previous reports (31, 32). Interestingly, the stronger aggregation tendency of ADCs synthesized from LP3 did not drastically increase the in vivo clearance of the ADCs (Fig. 4D). Moreover, HIC analysis of ADCs from LP3 showed only a modest increase in overall hydrophobicity compared with MMAE-based ADCs (Supplementary Fig. S3).

In vitro head-to-head comparison of the best performing linker–payload to enhertu

With our main goal being to develop a novel superior camptothecin-based linker–payload structure, we continued our investigations with ADCs that originate from LP5, because this linker–payload structure combines high conjugation yields in DAR and antibody recovery, very low aggregation tendency, highest in vitro activity and excellent in vivo stability and clearance behavior. To ensure manufacturability also on a higher scale, we first investigated how many equivalents of LP5 are needed to synthesize DAR8 ADCs. On the basis of previous experience, we determined the DAR as a function of equivalents of LP5 and equivalents of TCEP in the conjugation reactions in a pH 8.3 buffer system at 10 mg/mL (31). We found out that under those conditions, as less as 10 equivalents with respect to the mAb (1.25 eq. per Cys) are enough to generate a DAR8 ADC (Supplementary Fig. S4). Moreover, we tried to reduce the pH of the conjugation reaction. Here, it could be shown that the reaction kinetics are pH-dependent. Full conversion to DAR8 is still possible at pH 7.8, when the reaction time is increased to 24 hours (Supplementary Fig. S4).

After optimization of the conjugation conditions with LP5, the conjugation to trastuzumab was scaled to 10 grams of mAb and the DAR8 ADC was purified via tangential flow filtration only. Analytic data of the upscaling runs were identical to what has been observed on small scale. The DAR8 trastuzumab-LP5 ADC has been further investigated in head-to-head comparisons with Enhertu. For this purpose, Enhertu has been purchased in pharmaceutical grade. First, a broader panel of HER2-positive cell lines and one HER2-negative cell line were treated with either trastuzumab-LP5 or Enhertu. The results show that the in vitro efficacy is equal or superior in all HER2-positive cell lines (Fig. 5A). Less unspecific killing was observed in the non-targeted antigen-negative cell line L-540. In general, we were able to observe an expected correlation between ABCs and in vitro cytotoxicity of both ADCs (Fig. 5B). Of note, only minor differences in binding on SKBR-3 (K_D trastuzumab-LP5: 475.1 ng/mL, Enhertu: 420.6 ng/mL) and internalization on HER2-positive and -negative cell lines were observed between trastuzumab-LP5 and Enhertu (Fig. 5C).

Figure 5.

Cell-based in vitro efficacy experiments of trastuzumab-LP5 DAR8 (orange) or isotype-LP5 DAR8 (black) compared with Enhertu (gray). A, Cell viability on seven HER2-positive cell lines (N87, SKBR-3, HCC-78, SK-OV-3, OE-19, SK-GT-2, and HCC-1569) and one HER2-negative cell line L-540. B, Correlation between HER2-receptor level on the tested cell lines and antiproliferative effect (EC₅₀ in ng/mL) for trastuzumab-LP5 DAR8 (left, orange) and Enhertu (gray, right). C, Binding of trastuzumab-LP5 and Enhertu on SKBR-3 and internalization on SKBR-3, N87 (HER2⁺) and MDA-MB-468 (HER2⁻). D, Bystander killing of non-targeted HER2-negative MDA-MB-468 cells in co-culture with HER2-positive SKBR-3 (right). Shown is cell viability in dependency of the ADC concentrations. No effect was observed in MDA-MB-468 only cultures (left). EC) Relative quantification of histone H2AX phosphorylation (left), active caspase 3 (middle) and cleaved PARP (right) after treatment of SKBR-3-cells with 500 ng/mL of the ADCs for 3 days versus untreated. D–F, Relative quantification of cells in S-, sub-G₀, G₀–G₁, and G₂–M phase after treatment with 3,000 ng/mL of the ADCs for 2 days versus untreated. E–G, Secretion of DAMPs as a surrogate marker for measurement of immunogenic cell death as indicated by calrecticulin exposure on cell surface (left), ATP release (middle), and HMGB1 release (right). Shown are results after treatment with 3,000 ng/mL (= 20 nmol/L) of the trastuzumab-LP5 DAR8 (orange, solid) or Enhertu (gray, solid) and with 20 nmol/L of unconjugated exatecan (orange, checkered) or DXd (gray, checkered) for 2 days versus untreated (black). Graphs show means of n = 2 ± SEM.

Next, we investigated the bystander capacity, a potential feature of ADCs to compensate for heterogenous target expression, in dependency of the concentration of trastuzumab-LP5 DAR8 and Enhertu. Thereby, non-targeted HER2-negative MDA-MB-468 cells (47 ABCs) were co-cultured with HER2-positive SKBR-3 (790520 ABCs) followed by the assessment of cell viability of the HER2-negative cell only via flow cytometry. The MDA-MB-468 cells were only affected in viability when co-cultured with SKBR-3. A tendency toward lower IC₅₀ values and higher number of cells being affected was visible for trastuzumab-LP5 compared with Enhertu (Fig. 5D). Moreover, we characterized the mechanism of cell killing induced by trastuzumab-LP5 and Enhertu in more detail. We have shown by flow cytometry that trastuzumab-LP5 DAR8 and Enhertu are both inducing DNA damage resulting from TOP1 inhibition to a similar extent after treatment of SKBR-3 cells with 0.5 μg/mL of the ADCs as shown by the upregulation of phosphorylated histon 2AX, activated caspase 3, and cleaved PARP (Fig. 5E; Supplementary Fig. S6). Another mechanism of action of TOP1 inhibition is the induction of cell-cycle arrest, which has been shown in cells treated with trastuzumab-LP5 DAR8, Enhertu and also free payloads (Fig. 5F; Supplementary Fig. S6). Cells were arrested mainly in S phase, but also in G₂–M phase at selected concentrations and incubation times. Another important feature for cytotoxic payloads (also described for TOP1 inhibitors) is the induction of ICD, a process by which the drug-induced apoptosis of a tumor cell releases DAMPs and thereby stimulates the immune system. For this purpose, SKBR-3 cells were treated with Enhertu, trastuzumab-LP5 or the respective payloads DXd and exatecan, followed by the quantification of cells exposing calreticulin on the surface and releasing ATP and HMGB1 (Fig. 5G; Supplementary Fig. S7), all of which are DAMPs and surrogate markers for ICD. Indication for induction of ICD was shown by an increase in exposed calreticulin on the surface and ATP and HMGB1 release for both ADCs in a similar manner.

Stability and in vivo PK head-to-head comparison with enhertu

The high stability of ethynylphosphonamidate-linked ADCs, in particular in comparison with maleimide chemistry, has been described earlier (30–32) and could be confirmed here in the direct comparison of retained DAR after several days of incubation in rat serum (Fig. 6A). Moreover, we performed an in vivo stability head-to-head comparison between trastuzumab-LP5 DAR8 and Enhertu. Here, mice were treated with 20 mg/kg of each ADC at day 0 and blood samples were drawn after 1, 3, 5, and 7 days after treatment. Serum samples were analyzed by ELISA for total ADC exposure, showing a similar in vivo clearance behavior of trastuzumab-LP5 compared with Enhertu, with a slightly higher exposure for trastuzumab-LP5, 7 days after treatment. Moreover, the ADCs were captured from the sera by immunoprecipitation and analyzed via MS for DAR. The results clearly indicate that the ex vivo stability also translates to an in vivo setting, showing that trastuzumab-LP5 is still connected to 8 linker–payload molecules after one week of circulation, whereas Enhertu's DAR is reduced to approximately 5 (Fig. 6B). The main route of drug loss that has been observed for Enhertu is via retro-Michael addition, leading to the detection of the unmodified cysteines by MS.

Figure 6.

Stability experiments, in vivo efficacy and PK analysis of trastuzumab-LP5 DAR8, (orange) compared with isotype-LP5 DAR8 (yellow), vehicle (black), and Enhertu (gray). A, DAR ratio measured by MS after 0, 1, 3, and 7 days of incubation in at least 80% rat serum at 37°C. B, PK analysis of blood samples from mice being treated with 20 mg/kg. Samples have been drawn after 1, 3, 5, and 7 days of circulation. Exposure by total ADC ELISA (top). DAR measured by MS (bottom). C, Cell viability experiments of ADC samples after 0, 1, 3, and 7 days of incubation in at least 80% rat serum at 37°C on the HER2-positive cell line SKBR-3 (left) or HER2-negative cell line MDA-MB-468 (right). D, Female SCID mice have been implanted with cell-culture–derived xenograft model based on N87 cells in Matrigel. Treatment was started once when tumor volumes reached 0.1–0.15 cm³. Mice were treated at day 0 with a single dose of either vehicle, 2, 1, 0.5, and 0.25 mg/kg trastuzumab-LP5 DAR8, or Enhertu. At the highest dose of 2 mg/kg, a non-targeting isotype control ADC Isotype-LP5 DAR8 was included. Each group consisted of n = 10 females each. F, Sprague Dawley rats were treated at day 0 with 10 mg/kg of either IgG1 (left) or IgG1-LP5 DAR8 (right) at day 0. Samples were drawn after several time points throughout the 3 week study period (9 animals per group, 3 animals were sampled per time point). Shown are mean and SD out of the three animals. The serum samples were analyzed by ELISA for total mAb/total ADC (dotted line) and intact ADC (straight line). G, DAR measured by MS of blood samples from rats being treated with 10 mg/kg of IgG1-LP5 DAR8. Black dotted line corresponds to the lower limit of quantification of the assays.

Encouraged by these results, we wanted to investigate the influence of the reduction in DAR on the in vitro efficacy of the constructs, because we have already shown in Table 1 that a lower DAR can lead to a reduction in cell killing efficiency. Therefore, we took the serum-stressed samples containing trastuzumab-LP5 DAR8 and Enhertu and performed a cell viability experiment on a HER2-positive and a HER2-negative cell line (Fig. 6C). The results show that the reduction in DAR that originates from a retro-Michael–based instability of Enhertu also directly leads to a reduction in cell killing of the antigen-positive cell line after prolonged serum incubation. As expected, this effect is much less pronounced for the stably conjugated trastuzumab-LP5 DAR8.

In vivo efficacy head-to-head comparison with enhertu and PK analysis

Next, we performed an in vivo experiment, in which we compared the efficacy of trastuzumab-LP5 DAR8 to that of Enhertu over 4 different dose levels. 10 mice per group were implanted with tumors derived from the HER2-positive N87-cell line and treated once with either 0.25, 0.5, 1 or 2 mg/kg at day 0, once the tumors have reached a volume of 0.1–0.15 cm³. The N87-cell line was chosen because no significant difference in in vitro efficacy between both ADCs was observed (Fig. 5A). We anticipated that thereby the true impact of the linker technology can be assessed without being biased by the more potent payload that was chosen in LP5 compared with deruxtecan. The results show a clear benefit in efficacy for trastuzumab-LP5 over all 4 dose levels tested (Fig. 6D). It should be highlighted that trastuzumab-LP5 DAR8 is more effective at 0.5 mg/kg than Enhertu at 1 mg/kg. Moreover, at 1 mg/kg, no complete remissions were observed in Enhertu in contrast with 80% for trastuzumab-LP5 DAR8. At the highest dose level of 2 mg/kg, a non-targeted isotype control has been included, which did not show an effect with similar tumor growths than in the vehicle control, clearly underlining the target selectivity of LP5 when conjugated to an antibody.

Finally, we conducted a more comprehensive PK analysis of ADCs based on LP5 in Sprague Dawley rats. Previously, high DAR ADCs of payloads such as MMAE have been shown to rapidly clear from circulation (41). Here, two groups of rats were treated with 10 mg/kg of either an unconjugated IgG1 or a DAR8 conjugate of that antibody (IgG1-LP5 DAR8). Blood samples from those groups were taken over a time course of three weeks and analyzed via ELISA for total antibody and total and intact ADC. Remarkably, we observed a very similar clearance profile of the IgG1 antibody, and the same antibody conjugated to 8 molecules of LP5, judged by the similarity of the profiles for total ADC and total antibody and of the AUC values (Fig. 6E). This again highlights the negligible influence of the conjugation of 8 molecules of linker–payload LP5 to the antibody leading to a high DAR ADC with mAb-like properties. Once more, excellent stability could be confirmed by overlapping total ADC and intact ADC. This could be verified by an MS assay, which revealed a DAR8 circulating in rats even 21 days after administration of the ADC (Fig. 6F).

Discussion

With this study, we aimed to develop a superior TOP1 Inhibitor-based linker–payload platform for the construction of high DAR ADCs. Our studies demonstrate antibody-like properties for highly loaded phosphonamidate-linked exatecan-containing ADCs, thus providing a promising alternative for existing linker–payload structures used in marketed ADCs such as Trodelvy and Enhertu.

On the basis of our findings of exatecan being more potent than SN38 and DXd and with the intention to design TOP1 inhibitor-based ADCs with superior efficacy, we decided to construct linker–payload systems that use exatecan as the cytotoxic payload. With the aim to compensate for a potential toxicity risk, which is often associated with ADCs based on more potent payloads and linker instability (13, 14, 42), we decided to make use of ethynylphosphonamidates for antibody conjugation. This linker class has been previously shown to exhibit superior stability in head-to-head comparisons with approved ADCs (32, 43, 44). Hence, we synthesized several linker–exatecan constructs with and without the traceless PAB-release unit in the presence or absence of different dipeptidyl cleavage sides, with the addition of PEG solubilizing groups of various lengths. VC–PAB–exatecan constructs with shorter PEG chains than 12 units showed a strong aggregation tendency and led to low conjugation yields. This stands in strong contrast with what has previously been observed with other ethynylphosphonamidate-based linker–payloads (30, 32, 40) and led to our conclusion that the VC–PAB–exatecan moiety, as a very hydrophobic linker–payload group itself, exhibits the following challenges: (i) Low aqueous solubility leading to decreased available linker–payload concentrations under the aqueous conjugation conditions and (ii) ADCs, once formed, tend to form HMWS, leading again to reduced DARs in the monomeric fraction. Attempts to reduce this aggregation by switching from a VC LP3 to a VA linker–payload LP4 did not improve the conjugation behavior, which is in agreement with previous reports on the differences between those two dipeptides (41). To reduce the overall hydrophobicity of the linker–payloads, we incorporated PEG chains at the phosphonamidates-O-substituent (31). This strategy has been previously reported for ADCs to improve the PK profile of MMAE-based ADCs, because it reduces clearance from circulation mediated by the overall hydrophobicity of the ADC (18, 19). Notably, selective potency on the two tested HER2-positive, targeted cancer cell lines also increased with longer PEG chains, observable by a shift of the IC₅₀ value to lower concentration and an increase in the maximum amount of cells that have been affected by the ADC. Most likely this is a function of the higher drug-load that could be achieved with higher PEG length and is in line with previous reports on higher DAR-ADCs having a higher potency in vitro (41, 45).

ADC aggregation arising during the conjugation process was a major issue that prevented the formation of high DAR ADCs for some of the constructs. It should be noted that HMWS formation was not observed in the past when VC–PAB–MMAE or VC–PAB–MMAF was used instead of VC–PAB–exatecan, even with shorter PEG substituents at the ethynylphosphonamidate (30, 32). The observation on aggregation being induced by the VC–PAB–exatecan moiety is supported by studies of Burke and colleagues (20), who have also reported strong tendencies for HMWS formation with other maleimide–VC–PAB–camptothecin constructs. For deruxtecan, the linker–payload of Enhertu with a very similar camptothecin core structure compared with exatecan but different in linker, there have not been reports on problematic aggregation behavior even at high DARs (22). We therefore suspected that the origin of the aggregation and solubility issue lays in the combination of exatecan with the p-aminobenzyl moiety, leading to moieties that combine several π-electron systems, potentially leading to flat, ridged PAB–exatecan moieties. Strong π–π interactions between molecules that carry several aromatic cores are generally known to impair solubility (46). Indeed, LP7, which is in comparison with LP4 lacking the PAB moiety is not inducing aggregation even at high DAR. However, ADCs from LP7 are also clearly reduced in in vitro efficacy in comparison with conjugates that carry the PAB release handle. This observation is in line with early investigations of Trail and coworkers that the PAB moiety can be crucial for efficient release of the payload (33). Very similar behavior was observed for the non-cleavable linker–payload LP8, clearly indicating the necessity of an efficient release handle for a potent antibody-mediated delivery of exatecan into the targeted cell.

In a head-to-head comparison of the suboptimal PEG12-based LP3 with the superior PEG24-based LP5, we have furthermore shown that conjugation of 8 VC–PAB–exatecan molecules to an antibody is neither drastically increasing the overall hydrophobicity of the antibodies, judged by HIC retention, nor the in vivo clearance rates, which makes aggregation during conjugation and low conjugation yields the main challenge to be solved for VC–PAB–exatecan. This is contrary to the observation by us and others that VC–PAB–MMAE DAR8 ADCs clear rapidly from circulation in vivo, when not adequately balanced with solubility enhancers (18, 41). We interpret this as being another hint toward our hypothesis that not the overall hydrophobicity of VC–PAB–exatecan-based ADCs lead to aggregation and low conjugation but rather the strong tendency of larger aromatic molecules to build strong intermolecular π–π interactions. In previous work, we have mainly used the P5(PEG) moieties to improve in vivo PK of high-DAR ADCs (31, 32). Here, we demonstrate that such P5(PEG) building blocks can also be applied to efficiently reduce ADC aggregation and facilitate conjugation, which generally underlines the beneficial features of branched PEG-based linker systems.

With the identification of phosphonamidate-linked P5(PEG24)–VC–PAB–exatecan LP5 as an optimal linker–payload system, we improved the conjugation conditions to be able to use as little as 1.25 equivalents per cysteine to generate a DAR8 ADC from an IgG1 antibody. Next, a comprehensive in vitro head-to-head comparison between trastuzumab-LP5 DAR8 and Enhertu has been performed. We could confirm that both ADCs exhibit the same mechanism of cell killing as a result of TOP1 inhibition, demonstrated by upregulation of DNA damage markers such as phosphorylated histon 2AX, activated caspase 3, and cleaved PARP (47). Moreover, in accordance with literature data, we showed both S and G₂–M arrest as response to TOP1 inhibition (48, 49). We also investigated the capability of TOP1-based ADCs to elicit ICD, a process by which a cytotoxic drug induces apoptosis of a tumor cell that causes stimulation of the immune system. Increase of immunostimulatory DAMPs has been previously described to accompany and precede ICD and serves as a surrogate marker for the latter (50, 51). Immunostimulatory activity has been described for Enhertu and could be confirmed in the current study via an in vitro experiment. Such activation of the immune system can lead to a long-lasting protective antitumor immunity and opens a highly promising road for combination of ADCs with immuno-oncology agents (52, 53). The release of DAMPs from trastuzumab-LP5 in similar quantities than Enhertu indicates the immunostimulatory effect of exatecan when delivered via LP5 conjugated to a targeted antibody.

Higher in vitro potency of trastuzumab-LP5 compared with Enhertu could be observed in some cell lines. Because we demonstrated similar HER2 binding for both ADCs, we attribute this to differences in potency of the attached payloads, but not to an influence of the attached linker on the properties of the antibody. Bystander killing is considered to play a key role in the effective treatment of solid tumors to compensate for heterogenous target expression (54). Enhertu has been described to exhibit highly potent bystander activity, in particular when compared with ADCs with other camptothecin-based linker–payload systems and to trastuzumab-DM1 (23). Here, we even observed a slightly higher bystander killing effect for trastuzumab-LP5 over Enhertu. Most importantly, we were able to show a high increase of trastuzumab-LP5 in serum stability of the linker in vitro and in vivo. This results in retained in vitro activity of trastuzumab-LP5 DAR8 even after several days of serum stress, whereas cell killing was drastically reduced for Enhertu in this setting. The effect of DAR reduction and unstable linker technology in circulation on the safety of an ADC is still under debate (1). However, those results rather suggest a retained efficacy over time in circulation of ADCs conjugated to LP5, in particular when compared with unstably conjugated maleimides as used in Enhertu.

Moreover, we were able to show that the superiority of trastuzumab-LP5 in cell killing, bystander activity, and serum stability of the linker in vitro and in vivo also translates into improved in vivo efficacy. Trastuzumab-LP5 DAR8 revealed a more potent targeted effect over four tested dose levels, with a drastically increased activity even when half of the dose has been administered. We explain this with a slightly better PK profile, with less clearance and very importantly higher in vivo stability, which taken together helps to deliver the payload more efficiently to the site of the tumor. We conclude that those features can be attributed to the linker properties rather than exatecan being more potent payload compared with DXd, because we chose a cell line in which both ADCs had a highly comparable efficacy in vitro.

Finally, it could be shown in an in vivo PK study that LP5 can be used to synthesize DAR8 ADCs with antibody-like clearance behavior in combination with highest stability, judged by the detection of a fully conjugated DAR8 ADC after 21 days of circulation in a living organism. We believe that phosphonamidate-linked exatecan payloads LP5 can be used to reduce last drawbacks of the already well-functioning deruxtecan platform used in Enhertu for a next-generation of TOP1-based ADCs of highest therapeutic activity over a broad span of different tumor antigen targets.

Supplementary Material

Supplementary Data

EC50 values and maximum killing for DXd, SN38 and exatecan on several cell lines.

Supplementary figure 1

Dose-response curves for the ADC synthesized from trastuzumab and LP1 and LP3-LP3 on SKBR-3, HCC-78 and MDA-MB-468

Supplementary figure 2

Analytical SEC chromatograms of trastuzumab-LP3 DAR8, trastuzumab-LP5 DAR8 and Enhertu at day 0 and after storage for four weeks at 4°C and 37°C in the dark.

Supplementary figure 3

Hydrophobic interaction chromatography measurements. Comparison of trastuzumab, trastuzumab-LP3 DAR8, trastuzumab-LP5 DAR8, Adcetris and Enhertu. Shown are chromatograms of Adcetris in black (Homogenous ADC, brentuximab conjugated to MC-VC-PAB-MMAE in DAR0, DAR2, DAR4, DAR6 and DAR8). Enhertu in grey, trastuzumab in light orange, trastuzumab-LP5 in orange and brentuximab-LP3 in purple. Hydrophilicity of all shown DAR8 VC-PAB-Exatecan constructs are in the range of a DAR2 of Adcetris.

Supplementary figure 4

Optimization of the conjugation reaction of an antibody with LP5. A) Variation of equivalents of 3 and TCEP in the conjugation reaction and analysis of the DAR by MS. Constant conditions: 10 mg/mL antibody, buffer at pH 8.3, 16 hours. B) Variation of the pH in the conjugation reaction and analysis of the DAR by MS over the course of 24 hours. Constant conditions: 10 mg/mL antibody, 10 equivalents of LP5 with respect to the mAb, 5 eq. of TCEP.

Supplementary figure 5

Enhertu and trastuzumab-LP5 induce DNA damage. DNA damage induced by exatecan, DXd, trastuzumab-LP5 DAR 8 and Enhertu. HER2-positive SKBR-3 cells were treated for indicated time points with 5 µg/ml respective ADCs. Cells were stained with live/dead stain, for DNA damage markers phosphorylated H2A.X (Ser139), cleaved PARP and active caspase 3 and analyzed by flow cytometry. Graphs show means of n = 2 ± SEM.

Supplementary figure 6

Exatecan, DXd, Enhertu and trastuzumab-LP5 induce S phase and G2/M arrest. Cell cycle arrest induced by exatecan, DXd, trastuzumab-LP5 DAR 8 and Enhertu. HER2-positive cells were treated for indicated time points and concentrations of respective free payloads and ADCs. Afterwards, permeabilized cells were stained with propidium iodide (PI) and analyzed by flow cytometry. (A) Flow cytometric histogram of PI-stained untreated SKBR-3 cells or cells treated for 48 hours with 20 nM or 0.2 nM trastuzumab-LP5 DAR 8 ADC, showing concentration-dependent induction of S phase or G2/M arrest. (B) SKBR-3 cells were treated for 48 hours with indicated concentrations of free exatecan or trastuzumab-LP5 DAR 8 ADC. (C) HER2-positive SKBR-3 and N-87 cells were treated for indicated time points with 20 nM of by exatecan, DXd, trastuzumab-LP5 DAR 8 and Enhertu. Bar graphs show % of cells in S, G2/M G0/G1 and sub G0 (apoptotic) phase. Graphs show means of n = 2 ± SEM.

Supplementary figure 7

Exatecan, DXd, Enhertu and trastuzumab-LP5 induce immunogenic cell death. Immunogenic cell death induced by exatecan, DXd, trastuzumab-LP5 DAR 8 and Enhertu. HER2-positive SKBR-3 and N-87 cells were treated for indicated time points with 20 nM of respective free payloads and ADCs. As markers for immunogenic cell death, ATP and HMGB1 release into the supernatant as well as calreticulin exposure on the cell surface were determined. Therefore, cells were stained with live/dead stain and anti-calreticulin and analyzed by flow cytometry. ATP and HMGB1 release into the supernatant were measured by luminescence-based readouts. Graphs show means of n = 2 ± SEM.

Additional experimental procedures

General information on the experiments and additional experimental procedures.

ADC characterization

Characterization Data on the antibodies and antibody-drug-conjugates.

Organic synthesis

Organic Synthesis procedures for the linker payloads and characterization of products and intermediates.

Authors' Disclosures

S. Schmitt reports grants from Federal Ministry for Economic Affairs and Energy (EXIST Forschungstransfer II), Federal Ministry for Economic Affairs and Energy (German Accelerator), Federal Ministry of Education and Research (Rahmenprogramm Gesundheitsforschung), and LMU Munich outside the submitted work; as well as reports a patent for WO2023083919 pending; and Tubulis GmbH sponsored the work and is owner of the relevant patents listed above; and reports employment with Tubulis GmbH. P. Machui reports grants from Federal Ministry for Economic Affairs and Energy (EXIST Forschungstransfer II), Federal Ministry for Economic Affairs and Energy (German Accelerator), Federal Ministry of Education and Research (Rahmenprogramm Gesundheitsforschung), and LMU Munich outside the submitted work; as well as reports a patent for WO2023083919 pending; and Tubulis GmbH sponsored the work and is owner of the relevant patents listed above; and reports employment with Tubulis GmbH. I. Mai reports grants from Federal Ministry for Economic Affairs and Energy (EXIST Forschungstransfer II), Federal Ministry for Economic Affairs and Energy (German Accelerator), Federal Ministry of Education and Research (Rahmenprogramm Gesundheitsforschung), and LMU Munich outside the submitted work; as well as reports a patent for WO2023083919 pending; and Tubulis GmbH sponsored the work and is owner of the relevant patents listed above; and reports employment with Tubulis GmbH. S. Herterich reports grants from Federal Ministry for Economic Affairs and Energy (EXIST Forschungstransfer II), Federal Ministry for Economic Affairs and Energy (German Accelerator), Federal Ministry of Education and Research (Rahmenprogramm Gesundheitsforschung), and LMU Munich outside the submitted work; and Tubulis GmbH sponsored the work and is owner of the relevant patents listed above; as well as reports employment with Tubulis GmbH. S. Wunder reports grants from Federal Ministry for Economic Affairs and Energy (EXIST Forschungstransfer II), Federal Ministry for Economic Affairs and Energy (German Accelerator), Federal Ministry of Education and Research (Rahmenprogramm Gesundheitsforschung), and LMU Munich outside the submitted work; and Tubulis GmbH sponsored the work and is owner of the relevant patents listed above; as well as reports employment with Tubulis GmbH. P. Cyprys reports grants from Federal Ministry for Economic Affairs and Energy (EXIST Forschungstransfer II), Federal Ministry for Economic Affairs and Energy (German Accelerator), Federal Ministry of Education and Research (Rahmenprogramm Gesundheitsforschung), and LMU Munich outside the submitted work; and Tubulis GmbH sponsored the work and is owner of the relevant patents listed above; as well as reports employment with Tubulis GmbH. M. Gerlach reports grants from Federal Ministry for Economic Affairs and Energy (EXIST Forschungstransfer II), Federal Ministry for Economic Affairs and Energy (German Accelerator), Federal Ministry of Education and Research (Rahmenprogramm Gesundheitsforschung), and LMU Munich outside the submitted work; and Tubulis GmbH sponsored the work and is owner of the relevant patents listed above; as well as reports employment with Tubulis GmbH. P. Ochtrop reports grants from Federal Ministry for Economic Affairs and Energy (EXIST Forschungstransfer II), Federal Ministry for Economic Affairs and Energy (German Accelerator), Federal Ministry of Education and Research (Rahmenprogramm Gesundheitsforschung), and LMU Munich outside the submitted work; and Tubulis GmbH sponsored the work and is owner of the relevant patents; as well as reports employment by Tubulis GmbH. C.P.R. Hackenberger reports grants from Federal Ministry of Education and Research (Rahmenprogramm Gesundheitsforschung), Leibniz Competition (SAW), and German Science Foundation (DFG) outside the submitted work; as well as reports a patent for WO2018041985 issued. D. Schumacher reports grants from Federal Ministry for Economic Affairs and Energy (EXIST Forschungstransfer II), Federal Ministry for Economic Affairs and Energy (German Accelerator), Federal Ministry of Education and Research (Rahmenprogramm Gesundheitsforschung), and LMU Munich outside the submitted work; as well as reports a patent for WO2018041985 issued and WO2023083919 pending; and Tubulis GmbH sponsored the work and is owner of the relevant patents listed above; and reports employment with Tubulis GmbH. J. Helma reports grants from Federal Ministry for Economic Affairs and Energy (EXIST Forschungstransfer II), Federal Ministry for Economic Affairs and Energy (German Accelerator), Federal Ministry of Education and Research (Rahmenprogramm Gesundheitsforschung), and LMU Munich outside the submitted work; as well as reports a patent for WO2018041985 issued and WO2023083919 pending; and Tubulis GmbH sponsored the work and is owner of the relevant patents listed above; and reports employment with Tubulis GmbH. A.M. Vogl reports grants from Federal Ministry for Economic Affairs and Energy (EXIST Forschungstransfer II), Federal Ministry for Economic Affairs and Energy (German Accelerator), Federal Ministry of Education and Research (Rahmenprogramm Gesundheitsforschung), and LMU Munich outside the submitted work; as well as reports a patent for WO2023083919 pending; and Tubulis GmbH sponsored the work and is owner of the relevant patents listed above; and reports employment with Tubulis GmbH. M.-A. Kasper reports grants from Federal Ministry for Economic Affairs and Energy (EXIST Forschungstransfer II), Federal Ministry for Economic Affairs and Energy (German Accelerator), Federal Ministry of Education and Research (Rahmenprogramm Gesundheitsforschung), and LMU Munich outside the submitted work; as well as reports a patent for WO2023083919 pending and WO2018041985 issued; and Tubulis GmbH sponsored the work and is owner of the relevant patents listed above; and reports employment with Tubulis GmbH.

Authors' Contributions

S. Schmitt: Data curation, formal analysis, investigation, writing–review and editing. P. Machui: Formal analysis, investigation. I. Mai: Data curation, investigation. S. Herterich: Data curation, investigation. S. Wunder: Investigation. P. Cyprys: Investigation. M. Gerlach: Methodology. P. Ochtrop: Conceptualization. C.P.R. Hackenberger: Conceptualization, methodology, writing–review and editing. D. Schumacher: Supervision, funding acquisition, methodology, project administration. J. Helma: Conceptualization, supervision, funding acquisition, methodology. A.M. Vogl: Conceptualization, supervision, methodology, writing–review and editing. M.-A. Kasper: Conceptualization, data curation, supervision, investigation, visualization, methodology, writing-original draft.