Antibody Drug Conjugates, targeting cancer-expressed EGFR, exhibit potent and specific antitumor activity

Abstract

The monoclonal antibody ‘40H3’ binds to EGFRvIII and to full-length EGFR when it is overexpressed on cancer cells. To generate candidate cytotoxic antibody-drug conjugates (ADCs), 40H3 was modified by the addition of small molecular weight payloads that included two tubulin-modifying agents, two topoisomerase inhibitors and a pyrrolobenzodiazepine (PBD) dimer. Conjugates retained antigen binding activity comparable to the unmodified 40H3 antibody. The cytotoxicity of five distinct ADCs was evaluated on a variety of EGFR-expressing cells including three triple negative breast cancer (TNBC) lines. Generally, the 40H3 conjugate with the PBD dimer (40H3-Tesirine) was the most active killing agent. The killing of EGFR-positive cells by 40H3-Tesirine correlated with the number of surface binding sites for 40H3. However, bystander killing was also evident in experiments with antigen-negative cells. In vivo tumor xenograft experiments were conducted on two TNBC tumor lines. Three treatments with the 40H3-Tesirine ADC at 1mg/kg were sufficient to achieve complete remissions without evidence of mouse toxicity. Data support the development of ADCs derived from the 40H3 antibody for the treatment of cancers that express EGFRvIII or high levels of EGFR.

1. Introduction

The epidermal growth factor receptor (EGFR) is an oncogenic driver of many human epithelial cancers and up to 40% of GBM [1–5]. To target this receptor, both antibody-based therapeutics and small molecular weight inhibitors of its kinase domain have been developed. There are currently seven approved agents (four kinase inhibitors and three monoclonal antibodies) that target this receptor for indications that include Non-Small Cell Lung Cancer (NSCLC) and metastatic Colorectal Cancer (mCRC). Approved agents are designed to interrupt the signaling pathway of ligand-receptor stimulations that support the growth and survival of malignant cells. Additionally, clinical trials with newer agents or new combinations are on-going in many research centers [6]. Oncogenesis by EGFR is driven either by activating mutations, deletions or by over expression of the receptor following gene amplification or transcriptional activation [7]. High expression of ligands can also contribute to oncogenesis. The extracellular portion of EGFR comprises four prominent domains where domains I and III bind one of 7 ligands and domains II and IV contain cysteine-rich motifs that are involved in dimerization and receptor folding [8]. The most frequent deletion within the extracellular domain is of exons 2–7, which results in the loss of 268 amino acids and the joining of exon 1 with exon 8 via a unique fusion event producing EGFRvIII [9]. Within the external portion of EGFRvIII, there are subdomains exposed to solvent that would normally be poorly accessible in the wild type (WT) receptor [10]. One such subdomain is found within the 287–302 amino acid loop. This loop is fully exposed in EGFRvIII, not exposed in the WT receptor and conditionally exposed when the receptor is overexpressed [11]. Loop exposure may be the result of receptor misfolding and or inappropriate glycosylation [12] [13]. Therefore, antibodies to the 287–302 loop have the potential to bind cells expressing EGFRvIII or those cancer cells with high expression of full-length EGFR [10] [14]. While EGFRvIII is the most frequent EGFR mutation in GBM, it is also detected in various epithelial cancers albeit at a much lower incidence [15] [16]. Gene amplification of the EGFR gene and transcriptional stimulation have been reported for TNBC, NSCLC, and various head and neck cancers [17] [18] [19] [20] [21].

To target cancer-expressed EGFR, including EGFRvIII and gene-amplified EGFR, we and others have developed antibodies to the 287–302 loop [10] [14]. Previously, we reported on the generation of the 40H3 antibody with strong reactivity for cells expressing EGFRvIII and gene-amplified EGFR [14]. In addition, 40H3 had little or no reactivity for cells that expressed normal levels of WT EGFR.

In designing antibody drug conjugates ADCs, payloads are typically chosen from among highly potent drugs that cannot be administered systemically as unconjugated agents. Conjugation therefore serves to reduce systemic toxicity and redirect killing to cells where antibody binding is strongest. Specificity, therefore, depends on antibody binding characteristics as well as the on-target release of drugs. Furthermore, the Fc portion of the targeting antibody can be modified to alter Fc-receptor uptake and/or extend half-life [22]. The design of the linkers, joining drug to antibody, modulates how and where the drug is released [23] [24, 25]. ADCs are constructed using linkers designed to be degraded only within target cells often by lysosomal enzymes such as cathepsins dictating that drug release occurs only after internalization. Another strategy to release drug selectively within target cells includes the use of pH sensitive linkers [26] [27]. Conjugate designs, that restrict drug release to only target cells, have also been reported, such as T-DM1, where intracellular degradation of the ADC releases a lysine-conjugated maytansine that remains within the target cell [28] [29] [30].

Small molecular weight cytotoxic payloads that include tubulin-modifying agents, topoisomerase inhibitors and DNA damaging compounds have been conjugated to cancer-targeted antibodies as potential therapeutic agents [24, 25]. Representatives from each of these three groups have been evaluated in clinical trials, leading to the approval of approximately 10 ADCs [24, 25] [31]. Additionally, when a hydrophobic drug is released from target cells in high enough concentrations, one can also expect bystander killing of nearby cells [32, 33]. The risk-reward benefits for bystander versus ‘target-alone’ killing are still being evaluated in the clinic. Bystander killing is potentially beneficial when target antigens are expressed heterogeneously or when microenvironments restrict the delivery of large molecules such as ADCs to malignant cells. The bystander pathway for payload release is based the intracellular separation of free drug from the carrier antibody and then the diffusion of small molecular weight hydrophobic payloads to nearby cells, including other malignant cells. The downside of bystander killing is the potential destruction of non-malignant cells, including immune cells, that could contribute to tumor eradication.

To evaluate the utility of 40H3 as a delivery agent for toxic payloads, 5 conjugates were produced with either tubulin, topoisomerase or DNA modifying agents and four linkers were chosen that allowed for bystander killing. The choice of payload-linker combinations generally reflected the state of art for ADCs that are currently approved for human use. Cancer cell lines of various tissue origins were evaluated for sensitivity to each ADC. 40H3-Tesirine, being the most active, was then evaluated in two TNBC xenograft models where three injections caused complete regressions.

2. Materials and Methods

2.1. Antibodies

The mouse monoclonal antibody 40H3, previously described [14], was produced and purified under contract by GenScript Inc (NJ, USA). The EGFR monoclonal antibody 528 (Cat #: MA5–12875) was purchased from ThermoFisher Scientific (NY, USA). The anti-EGFR antibody, Cetuximab (C225), labeled with R-Phycoerythrin (R-PE), was purchased from Absolute Antibody (Cat #Ab00279–10.0). A non-binding IgG1 antibody (Cat # A01011) was purchased from GenScript.

2.2. Cell lines

The following cell lines were purchased from ATCC: A431, MDA-MB-468, MDA-MB-231, BT-20, NCI-H1650, Calu-3, F98, F98npEGFRvIII. A431, MDA-MB-468, F98 and F98 npEGFRvIII cells were cultured in DMEM media. MDA-MB-231 cells were cultured in L15 media. BT-20 and Calu-3 cells were cultured in EMEM. NCI-H1650 cells were cultured in RPMI-1640. All the media were supplemented with 10% fetal bovine serum (ThermoFisher Scientific, NY, USA), 2mM GlutaMAX (ThermoFisher Scientific, NY, USA), MEM non-essential amino acids (NEAA) (ThermoFisher Scientific, NY, USA). The medium for F98npEGFRvIII cells was further supplemented with 0.2 mg/ml G-418 (ThermoFisher Scientific, NY, USA). Cells were grown at 37 °C with 5% CO₂.

2.3. ADC construction

Purified 40H3 antibody was conjugated to one of five payloads (Supp Fig 1) using linker-drug combinations already approved for human use [24, 29, 31]. Conjugation reactions were performed under contract at NJBio, Princeton, NJ. Briefly, for conjugation via internal and reduced disulfides, antibody preparations were subjected to mild reduction (4.5 molar excess of TCEP to antibody in PBS). Samples were desalted using Zeba columns (ThermoFisher Scientific, NY, USA) according to the manufacturer’s instructions and then conjugated via maleimide groups on the linker-payload entity. For conjugation to lysine residues, the procedure of Lambert and Chari was followed to produce 40H3-DM1 [29]. Following the conjugation reaction, ADCs were fractionated to remove both under- and over-modified antibody. Drug antibody ratios (DARS) for each ADC are reported in Table 1.

Table 1:

ADC features

Antibody-Drug conjugate	Drug:Antibody (DAR)	Target	Cleavage mechanism	Linker
40H3-SMCC-DM1	3.02	microtubule	non-cleavable	N-succinimidyl-4-(N-maleimidomethyl) cyclohexane-1-carboxylate
40H3-MCVCPAB-MMAE	3.8	microtubule	Cathepsin B	maleimidocaproyl-valine-citrulline-p-aminobenzoyloxycarbonyl
40H3-CL2A-SN38	7.24	topoisomerase I	pH-sensitive	maleimide-type linker with triazoline
40H3-Deruxtecan	7	topoisomerase I	Cathepsin B/L	Gly-Phe-Leu-Gly-AM
40H3-Tesirine	3.81	DNA minor groove	Cathepsin B	maleimide Val-Ala dipeptide
IgG1 -Tesirine	3.69	DNA minor groove	Cathepsin B	maleimide Val-Ala dipeptide

The two tubulin-disrupting compounds, MMAE and DM1, were linked to 40H3 either via reduced disulfides (MMAE) or via reactive lysines (DM1). These two ADCs were constructed with reagents and protocols similar to those that produced brentuximab vedotin and T-DM1, respectively [29, 34]. Recently produced ADCs have included topoisomerase inhibitors as payloads [27]. To evaluate this kind of payload, SN-38 and deruxtecan (DXd1) were each conjugated to reduced disulfides within 40H3 using published methods and linkers [35]. Reactive maleimide was linked to each payload either via a cathepsin cleavable linker or an acid-labile linker. Finally, a PBD conjugate with tesirine was produced via conjugation with reduced disulfides [36] [37].

2.4. Free Drugs Payloads

DM1, Dxd1, MMAE, SG-3199 and SN-38 were produced by NJBio and dissolved in DMSO to a final stock concentration of 1mM.

2.5. Binding assays

The binding affinity constants, ‘K_d’, of the monoclonal antibody 40H3 and the various 40H3-based ADCs against the His-tagged EGFR peptide loop (aa 287–302) were measured using the Octet Red96 analyzer (Pall Life Sciences, New York, USA). The peptide that included a C-terminal His-tag was captured on Ni-NTA biosensors and used as the ‘antigen’. Briefly, the peptide and antibodies were diluted in buffer composed of PBS, 1% BSA and 0.05% Tween. The 40H3 antibody or conjugates were diluted to 250nM, 125nM, 62.5nM, 31.25nM, 15.625nM, 7.813nM and 3.91nM. 100nM of the peptide ligand was used as ‘antigen’. The Octet Red96 program was as follows: 10 min presoak, 60 sec baseline establishment, 120 sec antigen loading, 120 sec re-establishment of baseline after antigen loading, 120 sec for binder association, and finally 20 min for dissociation. Binding kinetics were analyzed using the Octet RED96 software.

2.6. Flow cytometry analysis

Antibodies were incubated with suspended cells (1 x 10⁵ cells per well) in a 96-well plate at 4 °C for 1 hour in FACS buffer consisting of PBS (KD Medical, MD, USA), 2mM EDTA (KD Medical, MD, USA), 1% BSA (Sigma-Aldrich, MO, USA) and 0.1% sodium azide (Sigma-Aldrich, MO, USA). Bound antibodies were detected with R-phycoerythrin conjugated F(ab’)2 goat anti-mouse IgG Fcγ (Cat # 115-116-071; Jackson ImmunoResearch, ME, USA) at 1:250 dilution for 45–60 min at 4°C. Antibody binding was characterized with either the SA3800 Spectral Analyzer (Sony Biotechnology, San Jose, CA, USA) or CantoII (BD Bioscience, San Jose, CA, USA) and the data were analyzed with FlowJo (Tree Star, Inc., Ashland, OR, USA) or with CellEngine (CellCarta, Montreal, Canada) and displayed in histogram format with the median fluorescence intensity plotted.

Quantitative binding was used to calculate the number of surface binding sites per cell. 40H3 was conjugated to R-Phycoerythrin using the R-Phycoerythrin Conjugation Lightning Kit following manufacturer’s direction (Abcam, Cat ab102918). Antibodies were incubated with cells (1 × 10⁵ cells per tube) at 4°C for 1h in FACS buffer consisting of phosphate-buffered saline (PBS; K D Medical, MD, USA), 2mM ethylenediaminetetraacetic acid (K D Medical, MD, USA), 1% bovine serum albumin (BSA; Sigma-Aldrich, MO, USA) and 0.1% sodium azide (Sigma-Aldrich). Quantibrite beads (BD Bioscience. Cat# 340495) were used as reference for quantifying binding sites on cells. Antibody binding was characterized with the BD FACSCanto II Flow Cytometer (BD Bioscience, CA, USA), the data were analyzed with FlowJo software (Tree Star, Inc., Ashland, OR, USA) and binding site quantification was analyzed with Excel (Microsoft, USA) and GraphPad Prism Software, (La Jolla California USA).

2.7. Cell viability

1 x10⁴ cells per well in a volume of 100μl were plated in 96-well tissue culture plates. After 24h, ADCs were added at the indicated concentrations. After 72 h, the medium was removed and the viability was determined using the CellTiter-Glo luminescent cell viability assay kit (Promega, Madison, WI). This assay quantifies the amount of ATP present in metabolically active cells. ATP was measured as luminescence produced by the mono-oxygenation of luciferin which was catalyzed by the Ultra-Glo-luciferase. The luminescence of each well was measured, and the values were presented as a percentage relative to untreated cells (control).

2.8. Bystander assays.

MDA-MB-468 and A431 cells (50000 cells per well in a volume of 500μl) were seeded in a 24well plate. After 24h, 40H3-Tesirine and mIgG1-Tesirine were added at different concentrations for 72 hours. Conditioned medium was harvested, transferred to F98 cells, incubated for 72 hours and cell viability was measured.

For the second experiment, a preliminary growth curve analysis of F98npEGFRvIII (EGFR-positive) and F98 (EGFR-negative) cells established that each line grew at different rates (data not shown). Therefore, the ratio of each cell type seeded into the coculture dish was adjusted to produce close to equal numbers of cells at the end of the assay. Specifically, the F98npEGFRvIII and the F98 cells were co-cultured in 3 mL of DMEM media with no antibiotic selection in 6-well tissue culture plates. In each well, a total of 200,000 cells was seeded in a 1:4 ratio (i.e. 40,000 F98 to 160,000 F98npEGFRvIII cells). Twenty-four hours after seeding, the cells were treated with media, free payload (SG-3199), 40H3 antibody, 40H3-Tesirine or IgG-Tesirine at the indicated concentrations. After 48 hours, the cells were removed with a scraper and counted with Trypan Blue 0.4% staining in the Countess II (ThermoFisher Scientific, NY, USA). The cells were then washed with 5 mL of PBS and centrifuged for 5 mins at 1100 rpm. The supernatant was discarded and the cells were resuspended in a volume of FACS buffer to achieve a concentration of 1 x 10⁶ cells/ml. For each sample, 100μL of the cell FACS buffer dilution was used so that there were 1 x 10⁵ cells per sample. Each sample was incubated with 0.1μg/ml of Cetuximab-PE for 30 minutes on ice. Following the cetuximab incubation, the cells were washed 3 times with FACS buffer and centrifuged at 1100 rpm. Each sample was then incubated with 10 nM SYTOX™ Red dead cell stain viability dye (ThermoFisher Scientific, NY, USA) for 15 minutes on ice. The cells were then processed through flow analysis (as described above).

2.9. Antitumor experiments

All animal experiments were performed in accordance with NIH guidelines and approved by the National Cancer Institute Animal Care and Use Committee. Female Balb/c mice were injected in the flank with 2 x 10⁶ MDA-MB-468 or BT-20 cells in DMEM or EMEM in the presence of Matrigel (Corning, NY) with 7 to 8 mice per cohort. When tumors grew to 100 mm³, treatments began with vehicle, unmodified 40H3 antibody or 40H3-Tesirine. Antibody or antibody conjugates were administered in three injections of 1mg/kg given at 4-day intervals. Tumor volumes and mouse weights were measured at least three times weekly. Tumor volumes were calculated as 0.5*(LxW²). Mice were euthanized by CO2 inhalation once tumors reached 1200 mm³ or became necrotic.

3. Results

3.1. Binding of 40H3 to epithelial cancer lines

To evaluate the 40H3 antibody as a delivery agent for toxic payloads, we first assessed its binding to human epithelial cancer cell lines of several tissue origins. Concentrations of the 40H3 antibody, ranging, from 10 μg/ml to 0.1 μg/ml, were incubated with potential target cells with known surface expression of EGFR (Fig 1A). Strong binding was noted for two TNBC lines, MDA-MB-468 and BT-20 and for one epidermoid cancer line, A431 cells. Moderate binding was noted for MDA-MB-231 (TNBC) cells and little of no binding was recorded for NCI-H1650 (lung) or Calu-3 (lung) cells.

Figure 1. Binding of 40H3 to human tumor cell lines.

A. Cell surface binding of 40H3 binding was assessed against several EGFR positive human cancer cell lines by flow cytometry. B. Binding sites per cell for 40H3 (10μg/ml) was determined using QuantiBeads. Site number was compared to the binding of Cetuximab (10μg/ml), which served as a positive control for total surface expressed EGFR.

Further, we quantified the number of binding sites per cell for the 40H3 antibody and compared this to the number of sites for cetuximab (Fig 1B). Both A431 and MDA-MB-468 cells exhibited 7.5–10x10⁵ 40H3 binding sites per cell. BT-20 cells had approximately 2.5x10⁵ sites per cell while the other cell lines exhibited far fewer sites. Except for MDA-MB-231 cells, the number of sites for 40H3 binding was approximately 10% of the total surface-expressed EGFR (where Cetuximab binding was set at 100%) for each line. For MDA-MB-231, only 2.5% of total EGFR sites were recognized by 40H3. While we do not understand the molecular basis for this observation, it appears that only a subset of surface-expressed EGFR was reactive with the 40H3 antibody.

3.2. Construction of ADCs

After cell surface binding was established, 40H3 was chemically conjugated to each of five small molecular weight toxic payloads to produce five candidate ADCs. Each ADC construct included a linker and toxic payload that was modeled after conjugates that have been approved for human use (Supp Fig 1). Two tubulin-directed payloads, MMAE and DM1, were attached to 40H3 where the MMAE conjugate was constructed using a valine-citrulline dipeptide cathepsin-cleavable linker while DM1 was attached to lysine residues using a non-cleavable linker. Two topoisomerase inhibitors DXd1 and SN-38 were attached to 40H3 with cleavable linkers: via a tetrapeptide and acid-sensitive linker, respectively. And finally, a PBD dimer conjugate was constructed using tesirine as the linker-payload. With the exception of the DM1 conjugate, ADCs were constructed using peptide linkers susceptible to cathepsin-mediated cleavage or were pH sensitive. Further, only the DM1 conjugate was attached to the 40H3 antibody stochastically. The other four conjugates were made by first ‘relaxing’ internal disulfide bonds within the antibody with a reducing agent and then linked to payloads using maleimide chemistry to target the reduced cysteines. Drug antibody ratios (DARs) ranging from 3 per antibody to 7 per antibody were achieved (Table 1). After conjugation, ADCs were purified further to remove unconjugated antibodies and those that had ratios of 8 DAR or greater. Finally, a murine IgG with no known reactivity was conjugated to tesirine and used as a control ADC.

Each ADC was assessed for retention of antigen (peptide loop corresponding to aa 287–302 from EGFR) binding activity. Results indicated no major loss of binding activity following the various conjugation reactions (Table 2, Supp Fig 2).

Table 2:

Affinity to His-tagged EGFR peptide loop (aa 287–302).

	KD (nM)	kon(1/Ms)	kdis(1/s)
40H3	0.252	2.35E+05	5.91E-05
40H3-MCVCPAB-MMAE	0.085	2.65E+05	2.24E-05
40H3-SMCC-DM1	0.442	2.02E+05	8.90E-05
40H3-Deruxtecan	0.066	2.84E+05	1.88E-05
40H3-CL2A-SN38	0.246	1.89E+05	4.64E-05
40H3-Tesirine	0.164	3.71E+05	6.07E-05

Moreover, the conjugation did not diminish the ADCs specific binding to EGFRvIII or overexpressed EGFR on cells. Consistent with this, F98 cells didn’t bind any ADC (Fig 2).

Figure 2. ADCs binding on cells:

Flow cytometry was used to detect binding of increasing concentration of 40H3-Tesirine; 40H3-MCVCPAB-MMAE; 40H3-SMCC-DM1, 40H3-CL2A-SN38 and 40H3-Deruxtecan. mIgG1-Tesirine was used as negative control and the antibody 40H3 as positive control.

3.3. Cell sensitivity to ADCs.

Each of five 40H3-based ADCs was assessed for potency on a collection of epithelial cancer cell lines with a range of EGFR expression, leading to two general observations. First, the 40H3-Tesirine conjugate was the most active ADC on the three cell lines with high EGFR expression (Fig 3, and Table 3).

Figure 3. Cell viability following treatment with ADCs.

MDA-MB-468, BT20, A431, and MDA-MB-231 cells were treated for 72 hours in a dose dependent manner with each ADC or the unconjugated antibody. Data are from at least three independent experiments done in triplicate.

Table 3:

ADCs cytotoxicity on human tumor cells.

IC50 (nM)	MDA-MD-468	BT20	A431	MDA-MB-231
40H3-Tesirine	0.3097	9.066	1.847	>100
40H3-MCVCPAB-MMAE	3.446	18.58	9.125	>100
40H3-CL2A-SN38	27.83	98.812	19.71	>100
40H3-SMCC-DM1	6.652	>100	89.6	>100
40H3-Deruxtecan	9.316	70.361	90.21	>100
40H3	> 100	>100	>100	>100

And second, the lethality of 40H3-Tesirine correlated with the number of 40H3 binding sites/cell. While MDA-MB-231 cells displayed 1.5x10⁵ EGFRs, they were not sensitive to any of the ADCs possibly due to the low number of 40H3 binding sites per EGFR (which was 2.5%). Calu-3 and NCI-H1650 also exhibited a low number of binding sites for 40H3 (Supp. Fig 3). A431, MDA-MB-468 and BT-20 cells displayed 7.5–10 x 10⁵ and 2.5 x 10⁵ 40H3 binding sites per cell respectively, which resulted in IC50 values on the same cells for 40H3-Tesirine of 0.3, 1.8 and 9.0 nM. After the tesirine-based conjugates, the next most potent ADC was 40H3-MMAE. However, in several cell lines, killing was incomplete even at higher ADC concentrations. Only A431 cells exhibited complete killing with 40H3-MMAE, with an IC50 of less than 10nM.

Poor susceptibility to ADCs could reflect resistance to the payload being employed. To determine intrinsic drug sensitivity, each cell line was incubated with free drug. For all cell lines tested, the PBD dimer payload in tesirine (SG3199) was the most potent compound followed by MMAE (Supp Fig 4). The next most potent payloads were the topoisomerase inhibitors, with the DM1 payload being the least toxic of the five, possibly to due to poor membrane permeability (Supp Table1). Of interest, MDA-MB-231 cells were killed completely with tesirine (SG3199) even though the 40H3-Tesirine conjugate was not very active, suggesting the importance of binding site number for ADC sensitivity.

3.4. Bystander killing

Cleavable linkers are hydrolyzed within the intracellular environment of target cells to release toxic payloads. If payloads are hydrophobic and cross cell membranes, they may kill nearby cells in a process referred as the ‘bystander effect’. If bystander killing is active, tumors with a heterogenous expression of target antigen could be treated effectively. Therefore, we assessed the bystander killing of 40H3-Tesirine on non-target cells. The rat glioma F98 cell line (non-target cells) does not express human EGFR while MDA-MB-468, A431 are the target cells. MDA-MB-468 and A431 cells were treated with 40H3-Tesirine or mIgG-Tesirine (Fig 4A), media was harvested and then transferred onto F98 cells for 72 hours. We observed that the viability of F98 cells was drastically reduced by the addition of 40H3-Tesirine conditioned media suggesting that SG-3199 had leaked from the 40H3-Tesirine treated cells (Fig 4B). Conversely, media from cells treated with the control IgG1-Tesirine didn’t exert any cytotoxic effect on F98 cells. Further, F98 cells were insensitive to either 40H3-Tesirine or IgG1-Tesirine (Fig 4B).

Figure 4. Bystander killing of non-target cells.

A. MDA-MB-468 and A431 cells were treated for 72h with different concentration of 40H3-Tesirine or mIgG1-Tesirine. B. Media from each treatment was transferred to F98 cells and viability was assessed after a further 72 hours. Data are from at least two independent experiments done in triplicate.

To confirm our results, we carried a second sets of experiments where the target cells were F98npEGFRvIII, stably transfected with the non-phosphorylated (np) human EGFRvIII. We first evaluated the sensitivity to the 40H3-Tesirine and the payload of both cell lines separately in monoculture experiments (Supp Fig 5A). Results indicated that F98 cells were resistant to 40H3-Tesirine but were sensitive to the payload alone (SG-3199). When F98npEGFRvIII cells were similarly assessed, there was killing with the 40H3-Tesirine conjugate and sensitivity to the free payload. In view of the monoculture results, F98 and F98npEGFRvIII were cocultured with a range of ADC concentrations and controls (Supp Fig 5B). If F98 cells were killed by the payload released from target cells, then there would be evidence of killing with the addition of 40H3-Tesirine. If no bystander killing was achieved, F98 cells should not decline in growth but outgrow the F98npEGFRvIII cells. As controls, we also added the free drug SG-3199, the unconjugated antibody, and a non-binding ADC, IgG-Tesirine. After 48 hours of coculture, F98 cells declined in growth with the addition of 0.1–1.0nM 40H3-Tesirine (Supp Fig 5B), which was in stark contrast to the monoculture result where F98s were unaffected by the same concentrations of 40H3-Tesirine (Supp Fig 5A, left panel). As expected, both F98 and F98npEGFRvIII cells were sensitive to the free payload. Further, there was no loss of viability in the coculture when either unconjugated 40H3 or the mIgG-Tesirine were added (Supp Fig 5A, left panel). Both sets of experiments (Fig 4 and Supp Fig 5) confirmed that 40H3-Tesirine can exert a bystander effect on antigen negative cells.

3.5. Xenograft Studies of TNBC

A subset of patients with TNBC express high levels of EGFR. Because MDA-MB-468 and BT-20 expressed high levels of EGFR and exhibited IC50 values in the low nmolar range, we investigated the in vivo activity of 40H3-Tesirine (1mg/kg x 3) against xenografts of both lines grown on the flanks of nude mice (Fig. 5). Tumor-bearing mice, with either MDA-MB-468 or BT-20 tumors, when treated with 40H3-Tesirine, saw tumors shrink rapidly. Treatments were well tolerated and groups of mice exhibited prolonged survival compared to mice treated with PBS (vehicle) or 40H3 (unmodified) antibody. Weighing treated mice and observing their behavior suggested no obvious signs of toxicity. Thus, in two models of TNBC with high levels of EGFR, complete tumor regressions were achieved with minimal signs of toxicity.

Figure 5. Efficacy of 40H3-Tesirine in xenograft tumor models.

Tumor growth curves, Kaplan-Meier survival curves and mouse weight graphs for MDA-MB-468 (A) and BT20 (B) xenograft tumors (n=7 or 8 mice per arm). Mice were treated with vehicle alone (black line), 40H3 unconjugated antibody (grey line) or 40H3-Teserine (red line for the MDA-468 xenografts and blue for the BT-20 xenografts) according to the protocol described in Materials and Methods.

4. Discussion

Here we report on the cell killing activity of five ADCs constructed with the antibody, 40H3. In vitro, the tesirine conjugate exhibited greater cytotoxicity than the ADCs made with other payloads. This result likely reflects the potency of the payload itself, as unconjugated tesirine (SG3199) was more active than the other four free payloads. Further, all 40H3 ADCs retained full antibody binding activity. Therefore, the lower activity of some ADCs could not be attributed to loss of antigen binding following drug conjugation.

For design and construction of ADCs, we chose five model payload-linker combinations that are currently approved for clinical use, albeit for treating cancer patients with a variety of distinct diagnoses. The earliest of the five model ADCs to be developed and approved was brentuximab vedotin (BV), targeting CD30 and linked via a di-peptide linker to MMAE [38] [39] . BV is now approved in several clinical settings related to treating lymphomas that express high levels of surface CD30 including Hodgkin Lymphoma.

To generate a more effective agent, FDA-approved trastuzumab was modified with emtansine to generate T-DM1 via stochastic conjugation to lysine residues in the antibody [28]. T-DM1 is approved for the treatment of HER-2 positive breast cancer in patients that has become non-responsive to trastuzumab and taxane treatment. A more recently approved ADC, with the same parent trastuzumab antibody, uses the topoisomerase inhibitor, deruxtecan, as the payload [40]. Trastuzumab deruxtecan is approved for treating patients with metastatic HER2-positive breast cancer who have received at least two prior treatments with anti-HER2 therapies [41]. For TNBC patients with unresectable local or metastatic disease and receiving at least two prior systemic therapies, Sacituzumab govitecan was approved in 2021 [42, 43]. Sacituzumab govitecan is composed of an anti-TROP2 antibody linked to the payload, SN-38, which is released at acidic pH. Finally, another recently approved ADC, loncastuximab tesirine, which targets CD19 and is conjugated to a PBD dimer via a cleavable peptide linker was used as a model [44]. In our hands, 40H3-Tesirine was most potent of five candidate ADCs and also demonstrated bystander killing of EGFR-negative cells (Fig 4, Supp Fig 5). The benefit of bystander killing in a cancer setting has not been fully established. Nor is it clear that there will be a one-size-fits-all value to its presence.

Because of its superior potency, the tesirine conjugate was moved forward for experiments in xenografts. We chose to evaluate its activity in two TNBC tumor models, BT-20 and MDA-MD-468. Both cell lines are both positive for gene-amplified EGFR. MDA-MD-468 cells with 2-fold more binding sites than BT-20 were approximately 2-fold more sensitive to the 40H3-Tesirine ADC in vitro but in vivo 40H3-Tesirine was able to eliminate either tumor. Aberrant expression of EGFR is a known surface target for antibody-based therapeutics. Cetuximab is approved for treating EGFR-positive colorectal cancers harboring wild type KRAS and squamous cell cancers of the head and neck. Likewise, panitumumab is approved for colorectal cancers with wild type RAS. Both antibodies block ligand binding to EGFR and are thought to reduce receptor signaling. The administration of either antibody is associated with skin rash and GI toxicity and is rarely curative. The 806 antibody, which does not block ligand binding, was first described many years ago [45, 46] and both chimeric and humanized 806 antibody have been administered safely in early clinical studies [47] [48]. To render this antibody (and derivatives) cytotoxic, various ADCs were constructed and have been evaluated in clinical trials where patients with likely EGFR-positive cancers were treated with one of three ADCs: depatuxizumab mafodotin (ABBV-414) [49], losatuxizumab vedotin (ABBV-221) [50] [51] and serclutamab talirine (ABBV-321) [52].

It is clear from clinical trial results that the combination of antibody, linker and payload may be critical for producing good outcomes with minimal side effects. One tesirine conjugate has been approved for human use, targeting CD19 on various B-cell malignancies [44].

The current 40H3-Tesirine ADC is constructed with a murine monoclonal antibody, which will undoubtedly benefit from humanization, prior to conducting trials in patients with EGFR-positive cancers but our data suggested that 40H3-Tesirine is a highly potent agent against EGFR tumors which may in the future produce clinical benefit.

Highlights:

• The monoclonal antibody ‘40H3’ binds to human EGFRvIII and overexpressed full-length EGFR on cancer cells.

• Antibody-drug conjugates (ADCs) generated with 40H3 and small cytotoxic payloads retained antigen binding activity.

• The ADCs are cytotoxic in vitro and 40H3-Tesirine was the most effective killing agent.

• In vivo, on two TNBC tumor lines, 40H3-Tesirine achieved complete remissions without evidence of mouse toxicity.