Discovery of novel cMET-targeting antibody Fab drug conjugates as potential treatment for solid tumors with highly expressed cMET

ABSTRACT

Background

Solid tumors are becoming prevalent affecting both old and young populations. Numerous solid tumors are associated with high cMET expression. The complexity of solid tumors combined with the highly interconnected nature of the cMET/HGF pathway with other cellular pathways make the pursuit of finding an effective treatment extremely challenging. The current standard of care for these malignancies is mostly small molecule-based chemotherapy. Antibody-based therapeutics as well as antibody drug conjugates are promising emerging classes against cMET-overexpressing solid tumors.

Research design and methods

In this study, we described the design, synthesis, in vitro and in vivo characterization of cMET-targeting Fab drug conjugates (FDCs) as an alternative therapeutic strategy. The format is comprised of a Fab conjugated to a potent cytotoxic drug via a cleavable linker employing lysine-based and cysteine-based conjugation chemistries.

Results

We found that the FDCs have potent anti-tumor efficacies in cancer cells with elevated overexpression of cMET. Moreover, they demonstrated a remarkable anti-tumor effect in a human gastric xenograft mouse model.

Conclusions

The FDC format has the potential to overcome some of the challenges presented by the other classes of therapeutics. This study highlights the promise of antibody fragment-based drug conjugate formats for the treatment of solid tumors.

KEYWORDS:

• Antibody fragment drug conjugate
• Fab-drug conjugate
• hepatocyte growth factor
• solid tumors
• c-Mesenchymal epithelial transition factor

1. Introduction

Increasing occurrences of solid tumors not just in the older population but also in younger adults nowadays underscores the need for multiple effective treatment options. One of the most relevant therapeutic targets implicated across multiple types of solid tumors is the c-Mesenchymal epithelial transition factor (cMET), a tyrosine kinase receptor. Once engaged to its native heterodimeric protein ligand, hepatocyte growth factor (HGF), a series of downstream cascading effects occur that ultimately lead to cell growth, proliferation, adhesion, and migration [1,2]. Thus, the MET/HGF pathway plays a vital role in normal development and tissue repair processes such as embryogenesis and wound healing [3–5]. Various forms of dysregulation in the cMET/HGF axis like cMET overexpression, MET gene amplification and/or mutation, ligand-dependent and ligand-independent activations can therefore have detrimental effects – the most common of which is uncontrolled and abnormal cell growths [1,6,7]. The MET/HGF pathway is known to be highly intertwined with the epidermal growth factor receptor (EGFR), another tyrosine kinase, which enables each to serve as a compensatory pathway sometimes fostering drug resistance [8,9]. High expressions of cMET and HGF are found in gastric, lung, colon, breast, and other solid carcinomas [10]. This makes cMET an extremely appealing molecular target, especially for targeted-drug delivery approaches, and given that its expression level in carcinomas is significantly elevated relative to normal cells [11,12].

With such vast implications in cancer biology, drug development efforts aimed toward the MET/HGF axis have been prolific, particularly for solid tumors with metastatic potential. Among those in the frontline are small molecule inhibitors that garnered FDA approval, namely Capmatinib (TABRECTA), Tepotinib (TEPMETKO), Cabozantinib (CABOMETYX), and still many more candidates in advanced clinical stage such as AMG 337, Foretinib (GSK136089), Elzovantinib (TPX-0022), and Merestinib (LY2801653). A more targeted approach achieved by antibody-based cancer therapeutics has grown tremendously over the years as evidenced by the considerable presence of biologics that are currently undergoing clinical trials – some of which are also investigated concurrently for multiple indications either as monotherapy or in combination with small molecule inhibitors. Only Amivantamab (RYBREVANT), a humanized bispecific antibody directed against cMET and EGFR, has been granted approval by the FDA thus far. Some of the widely investigated anti-cMET antibodies include Onartuzumab (MetMab), Emibetuzumab (LY2875358), Telituzumab (ABT-700), Sym015, SAIT301, and HLX55. While a number of these anti-cancer agents have shown promise, typical shortcomings associated with each sometimes limit their uses. For example, small molecule type of drugs commonly have selectivity and specificity concerns owing to their ability to target multiple human kinases, but the major drawback is the emergence of resistant clones [8,9,13,14]. Additionally, cMET overexpressing tumors harboring low genomic MET amplification may respond poorly to small molecule-based inhibitors. On the other hand, monoclonal antibodies (mAb) against cMET tend to induce agonist activity, thereby achieving cell survival instead of apoptosis [15,16]. This was circumvented by the succeeding generations of antibodies with the construction of a one-armed mAb that prevents the occurrence of receptor dimerization [17], or by the discovery and further engineering of mAbs that exhibit no functional agonist activity [18,19]. The other successful attempt was the generation of a bispecific antibody that targets cMET on one arm and EGFR on the other [20,21]. Besides several adverse treatment-related events, antibody therapeutics may have limited cell killing toward solid tumors that do not harbor MET gene amplification [19]. Notably, cell killing of this type can be achieved by antibody drug conjugates (ADCs) which deliver cytotoxic payloads to the tumor cells [22]. ADCs have become an emerging modality for the treatment of cancer. Telisotuzumab Vedotin (ABBV-399) is currently the most advanced at Phase 3 indicated for non-small cell lung cancer (NSCLC). Other cMET-directed ADCs for solid tumors are still in the early phase or preclinical stage.

One alternative targeted cancer therapy platform that may have the potential to address the challenges presented by small molecule-based and full-length IgG-based therapeutics is an antibody fragment drug conjugate [23–25], such as a Fab-drug conjugate (FDC). To the best of our knowledge, only a few cMET-targeted antibody fragment drug conjugates have been published to date [26,27]. This format is comprised of a Fab linked to a potent cytotoxic drug via a stable linker. It combines both the target specificity of the Fab portion with the strong cell killing ability of the small molecule drug. An FDC may have: (1) better tumor penetration ability that can result in more complete tumor killing, (2) lower overall plasma exposure to normal tissues due to rapid pharmacokinetics that in turn reduces the likelihood of payload decoupling or premature drug release, (3) reduced cross reactivity on normal cells such as increased susceptibility to infection following thrombocytopenia and neutropenia experienced with ADC treatment based on the lack of Fc receptor [28], and (4) access to hidden epitopes otherwise inaccessible to larger IgG. Like ADCs, an FDC depends on surface receptor expression and not on downstream signaling for efficacy, and this makes tumors that are no longer sensitive to cMET pathway inhibition still feasible for targeted delivery of a cytotoxic drug [22].

In this study, we report a novel cMET antibody discovered from our human GMAB library that is subsequently converted into an antibody fragment drug conjugate for potential treatment of solid tumors bearing highly overexpressed cMET. Toxic payloads were conjugated to anti-cMET Fab via our proprietary linker-payload employing a nonspecific lysine-based conjugation approach and a site-selective cysteine-based conjugation approach. The resulting FDCs were carefully characterized and investigated for their in vitro potency as well as in vivo efficacy in a xenograft mouse model.

2. Materials and methods

2.1. Cells and reagents

All cells were derived from American Type Culture Collection (ATCC) and were cultured in their recommended cell growth media at 37°C incubator equipped with 5% CO₂. In the study, cell lines include SNU-5 (stomach), HT-29 (colon), and MDA-MB-468 (breast), A549 (lung) and SK-BR-3 (breast). All chemicals were purchased commercially except for the linker-payloads which were synthesized in-house [29–31].

2.2. Generation of IgG and Fab

Anti-cMET IgG was expressed in CHO cells (Chinese hamster ovary) (R80007, Thermo Fisher Scientific) using CHO-S-SFM media with hypoxanthine and thymidine (12052098, Life Technologies). Transient transfection was performed at a cell density of 2 × 10⁶ cells/mL using polyethyleneimine (PEI) for DNA delivery, and cells were left shaking at 125 rpm for 24 h at 37°C. The following day, the growth media were supplemented with CD FortiCHO Medium (A1148301, Life Technologies) with Penicillin:Streptomycin:Amphotericin B solution cell culture reagent 100× (091674049, MP Biomedicals) and 100× GlutaMax (35050061, Gibco), and cells were transferred to 28°C at 125 rpm for about 2 weeks for protein expression. Cell viability and antibody concentration were monitored via Contessa II (Invitrogen) and Octet system (ForteBio), respectively. Finally, the antibody was purified using immobilized Protein A (17-5438-02, GE Healthcare). To generate the Fab, IgG was co-incubated with the immobilized papain (20341, Thermo Scientific) to be digested into fab fragments following the manufacturer’s protocol with few modifications. Briefly, the antibody was mixed with the pre-equilibrated and activated papain resin in the digestion buffer for 5–6 h (overnight for anti-RSV IgG) in a rotator at 37°C. After the indicated time, the digestion reaction was quenched and the resin was further washed with 10 mM Tris-HCl buffer, pH 7.5. Both the crude digest and the washes were combined and subjected to immobilized Protein A affinity purification. The desired product was collected from the flow through and was subsequently buffer exchanged and concentrated in PBS, pH 7.4.

2.3. Synthesis of FDCs, ADCs, and dye-conjugates

2.3.1. Lysine-based conjugation

Anti-cMET Fab or anti-RSV Fab (5–6 mg/mL) was reacted with 8–10 or 4–6 equivalents, respectively, of Duo5_P1 (Figure 1) for 5 h in PBS, pH 7.4 (≤5% v/v DMSO as co-solvent) at room temperature (RT). The same equivalents of Duo5_P2 (Figure 1) were utilized but left for overnight reaction at RT. After the indicated time, the crude reaction mixture was subjected to PD10 Minitrap G-25 column (28-9180-07, GE Healthcare) to remove excess free drug following the manufacturer’s protocol. Afterwards, it was further buffer exchanged in PBS, pH 7.4 using Amicon ultracentrifugal filter with 10 kDa MWCO (molecular weight cut off) at 4000 rpm at 10°C. To ensure removal of free and hydrolyzed drugs, the FDCs were further subjected to multiple rounds of dialysis with 10 kDa MWCO Slide-A-Lyzer in PBS, pH 7.4. Finally, the samples were sterile filtered using UltraFree centrifugal filter, spun at 3000 rpm for 5 min. For larger scale, reactions were extended to overnight and employed bigger PD10 columns, Amicon ultracentrifugal filters, and Slide-A-Lyzers.

Figure 1. Reaction schemes for the making of the anti-cMET Fab drug conjugates (FDC): (1) lysine-based conjugation approach and (2) cysteine-based conjugation approach (C-lock) after a mild selective reduction of the LC/HC disulfide bond.

2.3.2. Lysine-based conjugation of IgG

Anti-cMET A1–8 IgG (3–4 mg/mL) was reacted with 12–25 equivalents of Duo5_P1 with a series of incremental additions in PBS (≤5% v/v DMSO as co-solvent) at room temperature overnight. The crude reaction mixture was subjected to PD10 Sephadex G-25 column (17185101, GE Healthcare) to remove excess free drug following the manufacturer’s protocol. Afterwards, it was further buffer exchanged several times in PBS, pH 7.4 using Amicon ultracentrifugal filter with 30 kDa MWCO at 4000 rpm at 10°C. Finally, the ADC was further subjected to multiple rounds of dialysis with 20 kDa MWCO Slide-A-Lyzer in PBS, pH 7.4 to ensure removal of free and hydrolyzed drugs.

2.3.3. Cysteine-based conjugation

Anti-cMET Fab (5–6 mg/mL) was initially reduced with 5 equivalents of TCEP-HCl (20490, Thermo Scientific) in PBS (pH 8) for 1 h at room temperature. Excess TCEP was removed using 7 kDa MWCO Zeba spin desalting column (Pierce) pre-equilibrated in 20 mM sodium phosphate, 5 mM EDTA, pH 7.4 following the manufacturer’s protocol. The flow through containing the reduced Fab was directly reacted with 5 equivalents of either Duo5_C or MMAE_C in the presence of DMSO as co-solvent not exceeding 5% v/v overnight at room temperature. The work-up procedure was the same as the lysine-based conjugation described above.

2.3.4. Preparation of the dye conjugates

Anti-cMET Fab (3.5 mg/mL) and anti-cMET IgG (3 mg/mL) were reacted with 2.5 and 5 equivalents of AlexaFluor488 TFP ester (Thermo Fisher), respectively, for 5 h in PBS, pH 7.4 (≤5% v/v DMSO as co-solvent) at RT to generate the dye conjugates anti-cMET Fab-AF488 and anti-cMET IgG-AF488, respectively. The excess dye was removed using Zeba spin desalting columns (Thermo Scientific) according to manufacturer’s protocol, and both conjugates were stored in PBS, pH 7.4.

2.4. Characterization

2.4.1. Sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE)

5 µg of samples were loaded onto the wells of 4–12% Bis-Tris gel (Invitrogen) under reduced and non-reduced conditions using 1× MOPS running buffer run at 180 V for 50 min. The samples under reduced conditions were treated with a reducing agent (NP0009, Invitrogen) and further boiled for 10 min at 70°C before loading, whereas the non-reduced samples were loaded as is. A pre-stained protein ladder (26619, Thermo Scientific) was also loaded. The gel was stained for at least 15 min with Instant Blue Coomassie Protein Stain (Aab119211, Abcam) followed by an overnight destaining in deionized (DI) water. The image was captured using an imager (BioRad).

2.4.2. Determination of drug to antibody ratio (DAR) by hydrophobic interaction chromatography-high performance liquid chromatography (HIC-HPLC)

20 µg of sample in PBS/Eluent A (see composition below) was injected onto TSKgel Butyl-NPR column (4.6 mm ID x 3.5 cm x 2.5 µm, Tosoh) and analyzed via HIC-HPLC (Agilent). The samples were run using a gradient method of 5–85%B (Eluent A: 1.5 M Ammonium sulfate, 50 mM sodium phosphate, pH 7; Eluent B: 50 mM sodium phosphate, 25% isopropanol, pH 7) for 20 min under slightly heated condition at 40°C with a flow rate of 0.6 mL/min and monitored using UV absorbance at 280 nm. The average DAR was calculated by taking the sum of the weighted area under the curves of each species. Average DAR was roughly estimated for traces with highly overlapping species.

2.4.3. Free drug analysis by reversed-phase (RP)-HPLC

To the drug conjugate sample was added cold methanol (1:3 v/v). The solution was vortexed vigorously for about 1 min, and then the sample was centrifuged at 14,000 rpm for 20–30 min at 4°C. The supernatant was pipetted out carefully not to disturb the pellet and then loaded into a filter HPLC vial (Thomson) before injection into HPLC. Note that the samples were processed as quickly as possible to prevent evaporation. 15–30 µg sample was analyzed using the C8 Kinetex column (50 × 2.1 mm, 5 µm, Phenomenex) run with a gradient program starting from 0–100% B (Eluent A: Water/0.1% Formic acid (FA); Eluent B: Acetonitrile/0.1% FA) for 20 min under slightly heated condition at 40°C with a flow rate of 0.6 mL/min and monitored using UV absorbance at 250 nm. Similarly, Duo5_P1 was also run as a positive control using the same method as the samples. Due to the tendency of the drug to be hydrolyzed, it typically appears as two peaks in its HPLC trace. The sample trace was overlayed with the positive control to determine the presence of the peaks corresponding to the drug and/or its hydrolyzed form.

2.4.4. Assessment of binding towards the antigen via an enzyme-linked immunosorbent assay (ELISA)

2 µg/mL (50 µL/well) of human recombinant cMET-His Tag protein (10692-H08H, Sino Biological Inc.) was coated in a 96-well half-area microplate (CLS3690, Corning) at 4°C overnight. After washing the plate 3× with PBST (PBS +0.05% Tween 20), it was blocked with 50 µL/well Superblock (AAA999, Scytek Laboratories) for 1 h at room temperature with agitation (350 rpm). After washing the plate 3× with PBST, 50 µL/well of serially diluted samples (8-fold dilution starting at 40 µg/mL) in Superblock were added into the wells in duplicates and incubated at room temperature for 1 h with gentle agitation at 350 rpm. Plates were washed once again 3× with PBST before the incubation with 50 µL/well goat anti-human lambda-HRP (2070–05, SouthernBiotech) at 1:5000 dilution in Superblock for 1 h at room temperature. After the final 3× washing, 50 µL/well of TMB substrate (5120–0075, SeraCare/KPL) was added to each well and incubated for 5 min at room temperature. The reaction was stopped using 50 µL/well 1 N hydrochloric acid (HCl). The plate was immediately read using a plate reader using UV absorbance at 450 nm (Tecan Spark).

2.4.5. Concentration determination by bicinchoninic acid assay (BCA)

The final concentration of each drug conjugate was determined using Pierce’s BCA Protein Assay Kit (23227, Thermo Scientific) following the manufacturer’s protocol and with the use of Pierce’s pre-diluted protein assay standards: BSA and BGG sets (23208 and 23,213, respectively, Thermo Scientific).

2.4.6. Endotoxin test

The drug conjugates that were utilized for the in vivo mouse study were tested for endotoxin content using the limulus amebocyte lysate (LAL) cartridges (PTSS55F) read via the Endosafe portable test system, both from Charles River Laboratories.

2.5. Flow cytometry

Adherent cells were detached from the culture flask using TrypLE (12604–013, Gibco). After washing the cells with staining buffer (420201, BioLegend), cells were seeded in duplicates at 3 × 10⁵ cells/well in a V-bottom 96-well plate (3897, Corning). Cells were either incubated with 100 µL of the staining buffer (for control) or 100 µL of 5 µg/mL of anti-cMET IgG (A1–8, in-house) for 30 min in the dark at 4°C. After washing the cells once with the staining buffer, 100 µL of APC anti-human IgG Fc (410712, BioLegend), diluted 1:200 in staining buffer, were added to all wells and left incubated for another 30 min in the dark at 4°C. After final washing, cells were resuspended in the cell staining buffer and transferred to flow cytometry tubes with 35 m mesh strainer cap (CT6405, Alkali Scientific) and analyzed using the flow cytometer (BD).

2.6. Cytotoxicity assay

SNU-5 (1750 cells/well, 12.5 µL), A549 cells (750 cells/well, 12.5 µL), HT-29 (1000 cells/well, 12.5 µL) and SK-BR-3 (1000 cells/well, 12.5 µL) were seeded in a white flat-bottom 384-well plate (3570, Corning) at least 4–8 h before drug treatment for SNU-5 and 16–24 h for all the adherent cells. 2× working stock of the anti-cMET FDC (200 nM) or ADC (20–200 nM) were initially prepared in the cell growth media which were subsequently serially diluted in the corresponding media 3-fold. Following this, 12.5 µL of each of the serially diluted drug conjugates were added into the 384-well plate containing the cells. In a similar manner, replicates of 12.5 µL each of either a 2 µM Staurosporine (S1421, Selleck Chemicals) or media only was added into the wells to serve as the 100% cell killing and 0% cell killing, respectively. For the combination treatment, cells were treated with a 3-fold serial dilution of the anti-cMET IgG starting at 50 nM (final concentration in the well) while holding the FDC concentration constant at the approximate IC₅₀ of 2 nM. After 96 h incubation at 37°C with 5% CO₂, 25 µL of room temperature CellTiterGlo2.0 reagent (G9241, Promega) was added to each well using a multichannel pipette followed by a 2-min shaking using an orbital shaker (500 rpm), and then a 10-min incubation at room temperature. Afterwards, the plate was read using plate reader capable of reading luminescence in multi-well plates (Tecan Spark). Data were analyzed using GraphPad Prism software.

2.7. Confocal microscopy

MDA-MB-468 cells (moderate expression of cMET) were plated (10000 cells/well) on glass-bottom plate (164588, Nunc) previously coated overnight at room temperature with 100 µg/mL Poly L-Ornithine (A-004-M, Millipore). The wells were treated the next day with 100× of anti-cMET Fab-AF488 or anti-cMET IgG-AF488 with a final concentration of 2.5 µg/mL in cell growth media and incubated at different timepoints (2 min, 30 min, 3 h, 6 h). Cells were fixed after the indicated timepoints in 4% paraformaldehyde, stored at 4°C overnight in PBS, and stained the morning after with Hoeschst (1 µg/mL) at room temperature for 10 min. Images were taken under 60× magnification with a DAPI and a FITC filter using ImageXpress Micro Confocal (Molecular Devices).

2.8. Western blot

A549 cells (3 × 10⁵ cells/well) were seeded in a 6-well plate (10861–554, VWR) in its complete cell growth media and incubated at 37°C, 5% CO₂ overnight. After washing with serum-free media, cells were starved by incubation in DMEM/F-12 containing 0.5% FBS media overnight. Then, cells were treated with anti-cMET IgG or anti-cMET Fab at final concentrations of 1 nM, 10 nM, 100 nM, and 1 µM for 10 min in the same media at 37°C. Afterwards, cells were washed once with cold PBS and lysed in 100 µL RIPA buffer supplemented with protease and phosphatase inhibitor cocktails on ice for 15 min. The cells were centrifuged at 4°C for 10 min, and the supernatants were collected as the cell lysates. Equal amounts of lysate were loaded onto a pre-cast gel described above and were transferred to PVDF membranes using Trans-Blot Turbo Transfer System following manufacturer’s protocol (Bio-rad 1,704,150). The PVDF membranes were blocked in EveryBlot Blocking Buffer at RT for 30 min and subsequently incubated with phospho-ERK (CST 9101S) and GAPDH (Invitrogen 39–8600) primary antibodies (diluted 1:2000 in Blocking Buffer), cMET (CST 8198) primary antibody (diluted 1:1000 in Blocking Buffer), and phospho-MET (CST 3077) primary antibody (diluted 1:500 in Blocking Buffer) at 4°C overnight. The membranes were washed with 0.1% TBST 3× for 5 min each time and incubated with Goat anti-Rabbit HRP (CST 7074S) or Goat anti-Mouse HRP (CST 7076S) conjugated secondary antibodies (diluted 1:5000 in Blocking Buffer) for 40 min at RT. Finally, the membranes were washed for the final time with 0.1% TBST 3× for 5 min each time and incubated with Western ECL Substrate mixture for 5 min before imaging with ChemiDoc MP Imaging System. The membranes were kept in the dark at all times.

2.9. Mouse efficacy in vivo study

2.9.1. Animal model

Six weeks old female Nu/Nu mice purchased from Charles River Laboratories were used for this study. The animal study was performed in the vivarium at Sorrento Therapeutics Inc. according to the guidelines of ACUP (animal care and use protocols) approved by IACUC of Explora BioLabs (San Diego, CA). SNU-5 cells were cultured and expanded in an IMDM medium containing 20% fetal bovine serum (FBS). After harvesting, the SNU-5 cells were suspended in a 1:1 mixture of Hank’s balanced salt solution (HBSS) and Matrigel (Cat# 354234, Corning). Each Nu/Nu mouse was then injected subcutaneously at the right upper flank with 3 × 10⁶ SNU-5 cells in a total volume of 0.1 mL. Tumor growth was monitored using a digital caliper to measure the tumor volume (TV) which was calculated based on the formula: TV (mm³⁾ = [length × (width)²]/2. Tumor volume and body weight (BW) of each mouse were measured and recorded twice weekly. Tumor growth curves were generated by GraphPad prism software from the mean TV (mm³) ± SEM of each data point for each treatment group (N = 8 mice). Statistical significance was analyzed by Two-way Anova with Tukey’s multiple comparisons test to vehicle control of PBS.

2.9.2. Treatment

The treatment was initiated when average tumor size reaches 150–200 mm³. The test compounds and controls with various dose regimen were administrated by intravenous (i.v.) injection of 200 µL. There were 5 groups in total with 8 mice per group. The PBS vehicle control (Group 1), 15 mg/kg anti-RSV FDC-Duo5_P1 (Group 2), 15 mg/kg anti-cMET FDC-Duo5_P1 (Group 4) and 5 mg/kg anti-cMET FDC-Duo5_P1 (Group 5) were all dosed every other day 14 times. The 3 mg/kg anti-cMET ADC-Duo5_P1 (Group 3) was dosed once a week for 4 times.

3. Results

3.1. Identification of anti-cMET A1–8 IgG via phage display

To screen anti-cMET antibody candidates, Sorrento’s G-MAB™ library, a phage display human antibody library was employed. This is a single chain variable fragment (scFv) antibody phage display library that was constructed from the antibody repertoire of over 600 healthy individuals. Following confirmation of cMET binding by ELISA, the candidate scFvs were converted into and subsequently expressed as full-length human IgG1 antibodies. Affinity maturation was undertaken for the lead candidates using a filter lift protocol [32] and consequently identified the most potent clone A1–8, which was further profiled for binding affinity using surface plasmon resonance (Sup. Fig. S1).

3.2. Synthesis and characterization of the anti-cMET FDCs

Anti-cMET A1–8 IgG was transiently expressed in CHO cells and purified via Protein A affinity chromatography. The Fabs were generated via an enzymatic digestion process using papain, a thiol-endopeptidase, that nonspecifically cuts an IgG to produce two molecules of Fab and an Fc fragment. The fabs were eventually isolated via a flow-through mode with immobilized Protein A resin. Once the fab fragments were obtained, the FDCs were synthesized employing two approaches (Figure 1): (1) conjugation via the surface-exposed lysine residues and (2) conjugation via the reduced cysteine residues. In the former, we utilized two Gly5 cleavable linker-payloads, namely Duo5_P1 and Duo5_P2 equipped with a pentafluorophenyl (PFP) ester, R1 and R2, respectively. Both conjugate to reactive lysine residues in aqueous media on the Fab’s surface to form amide bonds (Figure 1, Top). In the latter conjugation approach, we also utilized protease-cleavable linkers (Gly5 for Duo5_C, and valine-citrulline for MMAE_C as control) but equipped with a different reactive moiety, bis(bromomethyl) quinoxaline, instead of a PFP ester. This moiety allows for a site-selective reaction to occur toward thiols in close proximity, such as the reduced disulfide bond between the light chain/heavy chain (LC/HC) of an IgG or Fab, in an intramolecular fashion, thereby re-bridging the previously reduced disulfide bond (herein referred to as the C-lock conjugation approach). To make an FDC via C-lock approach, the Fab is first subjected to a mild and selective reduction of the intermolecular disulfide bond between the LC and HC to free the cysteines that is achieved using Tris (2-carboxyethyl) phosphine (TCEP), followed by the nucleophilic substitution reaction with the C-lock reagent-containing payloads (Figure 1, Bottom). Both MMAE and Duo5 payloads used in the study are microtubule inhibitors. Due to the nonspecific nature of the lysine conjugation reaction of the first approach, it tends to produce a mixture of heterogeneous drug products ranging from 0–4 copies of payloads. On the other hand, the C-lock approach results in one predominant drug product containing one copy of payload. The difference between the two conjugation techniques is clearly illustrated in their gel profiles (Figure 2). As a result of the reduced thiols being re-bridged via the C-lock reagent, the FDCs remained intact (molecular weight ~ 50 kDa) even after being subjected to reduced denaturing conditions whereas the FDC conjugated via the lysine residues has disintegrated into its LC and HC components (each has molecular weight ~ 25 kDa) as expected (Figure 2a). The difference is further evident in their HIC-HPLC profiles (Sup. Fig. S2). Reaction conditions were optimized to achieve the desired DAR. To ensure that there are no excess free and/or hydrolyzed drugs in the FDCs that can affect the downstream assays, all were subjected to extensive buffer exchange and dialysis in the appropriate buffer. Cold methanol precipitation method was used to precipitate out the proteinaceous content of the FDC followed by the extraction of the small molecule components which were then analyzed using RP-HPLC (Sup. Fig. S3). Finally, the identity of the final conjugate was confirmed via LC-MS (Sup. Fig. S4).

Figure 2. Characterization of the anti-cMET Fab drug conjugates. a) SDS-PAGE analyses of the lysine-based conjugation approach (left) and cysteine-based conjugation approach (right), b) binding affinity determination via ELISA of the anti-cMET Fab/IgG and their corresponding drug conjugates, c) time-course internalization study of the anti-cMET Fab in live MDA-MB-468 cells (moderate cMET expression) via confocal microscopy.

3.3. Anti-cMET FDCs retain their binding towards human cMET recombinant protein and fab gets rapidly internalized by cMET-overexpressing cancer cells

To examine whether the conjugation processes have altered the binding toward the target antigen, ELISA was carried out against recombinant human cMET-His protein as antigen and probed with an anti-lambda-HRP antibody (Figure 2b). Both anti-cMET FDCs from the lysine-based conjugation approach with comparable average DAR (1.7–1.8) did not show substantial change in EC₅₀ relative to the naked Fab (<1.7-fold difference). The same is true with the anti-cMET FDC from C-lock approach (data not shown). Likewise, the binding of anti-cMET ADC remained the same as the naked antibody. The Fab exhibited only a minor change in EC₅₀ (<2-fold) compared to the parent IgG. Next, we assessed the cell internalization of the anti-cMET Fab via confocal microscopy using live MDA-MB-468 cells, a breast cancer cell line with moderate cMET expression. We first synthesized a dye-conjugated Fab, anti-cMET Fab-AF488, for visualization. For comparison, an anti-cMET IgG-AF488 was also synthesized. The time-course study on the anti-cMET Fab-AF488 revealed that at 2 min, the majority of the cells display a distinct ring-like green color around the cell periphery (Figure 2c). Noticeably, more Fabs were internalized within 0.5–3 h as shown by the appearance of a more diffused green fluorescence in the cell cytoplasm and around the nucleus. We noticed an overall decrease in intensity at 6 h presumably due to degradation. On the other hand, the anti-cMET IgG labeled with the same dye revealed an earlier onset of internalization based on the presence of diffused green fluorescence in the cytoplasm exhibited by cells at the 2-min and 30-min timepoints (Sup. Fig. S6). At the 6-h timepoint, there is an apparently higher degree of internalization observed for the IgG than the Fab. Reduced valency of the Fab has most likely resulted in slower internalization.

3.4. Anti-cMET FDCs are highly potent and selective towards highly overexpressing cMET cell lines

To examine the in vitro potency of our FDCs, we treated SNU-5 cells (high cMET) and A549 cells (low cMET) with various concentrations of our anti-cMET FDCs along with corresponding non-targeting FDCs, anti-RSV FDC-Duo5_P1 and anti-RSV FDC-MMAE_C, serving as negative controls, for 96 h at 37°C. Cell viabilities were measured using CTG assay. The results revealed that our anti-cMET FDCs, namely α-cMET FDC-Duo5_P1, α-cMET FDC-Duo5_P2, α-cMET FDC-Duo5_C, and α-cMET FDC-MMAE_C, demonstrated remarkable cell killing toward SNU-5 cells with half-maximal inhibition (IC₅₀) values at low single digit nanomolar (<3 nM) whereas no cell killing was observed toward A549 cells under the same concentration range tested (Figure 3a–b, Table 1). None of the α-RSV Fab drug conjugates demonstrated any cell killing toward the cell lines. The cell killing ability of α-cMET FDC-Duo5_P1 and α-cMET-FDC-Duo5_C were also tested on HT-29 cells (high cMET) and SK-BR-3 cells (very low cMET). As shown in Figure 3c, the FDCs tested demonstrated cell killing toward the HT-29 cells albeit not as strongly as against SNU-5 cells (IC₅₀ for HT-29 was ~20 nM and ~60 nM, respectively) possibly due to lower cMET expression on the HT-29 cells (Sup. Fig. S7) or lowered sensitivity to the drug. The higher IC₅₀ value exhibited by the α-cMET FDC-Duo5_C is likely due to its lower average DAR (0.98) as a different batch was used in this assay. As expected, there was no cell killing on SK-BR-3 cells. Notably, α-cMET ADC-Duo5_P1 shows a much lower IC₅₀ of 0.2 nM than the FDCs on SNU-5 cells (Figures 3d), which is comparable to ABBV-399, an anti-cMET ADC with an MMAE payload (average DAR = 3.1) reported to have an IC₅₀ of 0.28 nM on the same cell line. However, the ADC seems to have resulted in more incomplete cell killing compared to FDC as indicated by ~ 40% and ~ 50% minimal inhibition on SNU-5 and HT-29 cells, respectively (Figures 3c–d). This is even more pronounced for the α-cMET ADC-Duo5_P1 with low DAR (Figure 3d) which showed very little cell killing most likely due to the presence of 44% naked mAb in the mixture. Taken together, these results indicate that the α-cMET ADC is more potent than the α-cMET FDC, albeit demonstrated less complete cell killing. The cell killing of our anti-cMET FDCs is highly potent and selective toward highly overexpressing cMET cell lines. Moreover, its potency is independent of the mode of conjugation and more likely influenced by the level of target antigen expression.

Figure 3. In vitro potency of the FDCs toward cMET high and cMET low expressing cancer cell lines. a) anti-cMET FDCs from lysine-based approach vs a non-targeting FDC (α-RSV, respiratory syncytial virus) on SNU-5 (gastric) and A549 (lung) cells, b) anti-cMET FDCs from the C-lock approach with either Duo5 or MMAE as payload on the same cell lines c) anti-cMET FDCs potency toward HT-29 (colorectal) and SK-BR-3 (breast) cancer cells d) FDC vs ADC with varying DAR.

Table 1. Composition of the different FDCs/ADCs and their IC₅₀ values.

	Average DAR*	%Unreacted Starting Material	IC50 (SNU-5)
α-cMET FDC-Duo5_P1	1.71	~15%	1.2–2.2 nM
α-cMET FDC-Duo5_P2	1.80	~13%	1.8 nM
α-cMET FDC-Duo5_C	1.77	0%	2.8 nM
α-cMET FDC-MMAE_C	0.98	~11%	2 nM
α-RSV FDC-Duo5_P1	1.74	~3%	-
α-RSV FDC-Duo5_C	1.91	0%	-
α-RSV FDC-MMAE_C	1.11	~7%	-
α-cMET ADC-Duo5_P1 (lo DAR)	0.8–1.1	~44%	-
α-cMET ADC-Duo5_P1 (hi DAR)	1.8–3.2	~11%	0.2 nM

*Calculated based on their weighted HIC-HPLC peak areas under the curves. DARs of ADCs were roughly estimated due to heavy overlapping of peaks..

3.5. Anti-cMET fab did not trigger cMET receptor dimerization and downstream effector activation

To demonstrate whether the anti-cMET Fab induces cMET receptor dimerization and subsequent activation of the downstream signaling, we deprived A549 cells (cMET low) of growth factors by starvation in serum-free media overnight at 37°C and consequently stimulated with different concentrations of the anti-cMET Fab and IgG for 10 min at 37°C. Cells were lysed, resolved in SDS-PAGE, transferred onto a PVDF membrane, and subsequently probed for the proteins of interest. In the Western blot analysis (Figure 4a), there are no bands observed corresponding to the phosphorylated MET (pMET) under any of the Fab treatments whereas clear bands were present after treatment with the anti-cMET IgG. As all treatments displayed the same total cMET, this suggests that only the IgG, not the Fab, triggers cMET receptor dimerization that led to its phosphorylation. Furthermore, the stimulation of the cells by IgG has caused protein band enhancement probed for the phosphorylated ERK (p-ERK1/2) while the same bands remained faint for the cells stimulated with the Fab and are comparable to the non-treated control (NC) cells. GAPDH was added as loading control. Collectively, the data suggest that not only has the anti-cMET IgG triggered receptor dimerization but also initiated the downstream signal activation of the pathway whereas the Fab has no such effect.

Figure 4. In vitro characterization of induction of cMET receptor dimerization and subsequent activation of downstream effector. a) Western blot analysis of cell extracts after an overnight serum starved-A549 cells (cMET low) were stimulated with either anti-cMET Fab or anti-cMET IgG, b) cytotoxicity assay of FDC in the presence or absence of anti-cMET IgG in SNU-5 cells.

To further investigate whether the IgG elicits biological response such as cell survival, one FDC (α-cMET FDC-Duo5_P2) was added at a fixed dose near its IC₅₀ (2 nM) to SNU-5 cells while the anti-cMET IgG was being titrated starting at 50 nM. The results showed ~ 50% cell killing in the presence of low concentrations of IgG (<0.1 nM) but was suppressed gradually at increasing concentrations of IgG (Figure 4b). FDC in the absence of the IgG killed the cells potently as expected while there was no cell killing observed for both the cells only and the IgG-treated cells. The reverse cell killing curve observed from the FDC/IgG mixture can be attributed to either antigen competition or activation of agonistic activity by the naked antibody or combination of both. This could indicate the pro-survival effects of the anti-cMET IgG that can lead to cell survival and proliferation instead of cell death if present in a mixture.

3.6. Anti-cMET FDC demonstrates potent anti-tumor effect in mice bearing SNU-5 ×enograft

To evaluate the in vivo efficacy of the anti-cMET FDC, we decided to utilize the α-cMET FDC-Duo5_P1 to have flexibility in the DAR for future studies. Human SNU-5 cells were implanted subcutaneously into Nu/Nu mice and allowed to form a tumor 150–200 mm³ in size before the start of the treatment. Due to the relatively shorter half-life, FDC groups were dosed every other day for 4 weeks (28 days) while ADC group was dosed once a week for 4 weeks. After the last dose, mice were monitored continuously for an additional 2 months or euthanized according to the guidelines being followed. During the entire course of the treatment, α-cMET FDC-Duo5_P1 demonstrated remarkable anti-tumor efficacy in a dose-dependent manner with both groups revealing 100% TGI (tumor growth inhibition) and 9.5% TR (tumor regression) for low-dose FDC (5 mg/kg) and 75.4% TR for high-dose FDC (15 mg/kg) at day 88 after initial treatment or 60 days after the last dose (Figure 5a and Sup. Table S1). Tumors regrew slightly in the low-dose α-cMET FDC-Duo5_P1 group at around day 53 (25 days after the last treatment) while notably no regrowth was observed for the high-dose α-cMET FDC-Duo5_P1 and for the α-cMET ADC-Duo5_P1 groups. The non-targeting α-RSV FDC-Duo5_P1 did not show any anti-tumor effect. No body weight reduction of mice was observed in all treatment groups (Figure 5b). Overall, both the low and high doses of α-cMET FDC-Duo5_P1 were very efficacious in the SNU-5 mouse xenograft model. The efficacy is comparable to α-cMET ADC-Duo5_P1 at 3 mg/kg with the current dosing regimen. Note that the FDCs (with an average DAR of 1.6 vs 2.3 for ADC) were dosed more frequently than the ADC to compensate for its relatively shorter half-life but which resulted in high cumulative dose.

Figure 5. In vivo efficacy evaluation of the anti-cMET FDC on SNU-5 subcutaneous xenograft in Nu/Nu mice. Treatment was initiated when the average tumor size reached 150–200 mm3. Tumor size (a) and body weights (b) of mice were measured twice weekly. α-cMET FDC-Duo5_P1 significantly inhibited SNU-5 tumor growth in nude mice in a dose-dependent manner up to 88 days after initial treatments. **** indicate P value < 0.0001, two-way Anova with Tukey’s multiple comparisons test to vehicle control PBS. Data = mean ± SEM (N = 8); ns = not significant.

4. Discussion

We describe here the design, synthesis, and characterization of novel anti-cMET Fab-drug conjugates (FDCs) as potential anti-cancer agents. FDCs exert their cell killing effect by receptor-mediated internalization followed by its subsequent degradation inside the cells with concurrent release of the cytotoxin. By virtue of its smaller size, an FDC may exhibit better tumor penetrating ability as well as a more uniform tumor distribution compared to an intact IgG [33–35]. This is important since several factors lead to heterogeneous distribution of i.v. administered antibodies in tumors creating certain regions in the tumor mass that are not exposed to the anti-cancer agent that in turn leads to drug resistance and ultimately ineffective therapy [35,36]. In the case of an ADC, this is often compensated for by the bystander effect because of the free payload diffusing through the neighboring cells. As Fab-based, an FDC likely has fast clearance [34,37] which may translate to lower systemic exposure of normal cells to the toxic payload that causes off-target effects sometimes exhibited by some ADCs. With our FDC format, we expect to eliminate the agonist effect commonly rendered by bivalent antibodies against cMET. Lacking the agonistic effect helps eradicate the tumor cells instead of promoting cell survival and proliferation.

The conjugation techniques employed in this study did not substantially alter the Fab antigen binding. Between the two, the nonspecific lysine-based conjugation allows flexibility for higher DAR while the cysteine-based conjugation offers more stability as it prevents the dissociation of the LC and HC because of the covalently re-bridged disulfide bond. Like MMAE, the Duo5 payload used in the study is also a synthetic derivative of Dolastin-10, a marine antitumor peptide [38]. Duo5, short for Duostatin-5, is highly potent that acts as an anti-mitotic agent and was utilized as the payload for A166, an anti-HER2-ADC (DAR = 2), found to be stable in circulation and have manageable toxicity up to 4.8 mg/kg Q3W in Phase 1 study for advanced HER2-expressing solid tumors [39]. The internalization of the resulting drug conjugate on live cMET-expressing cells was also assessed by utilizing the same Fab that was conjugated to a fluorescent dye via the lysine-based conjugation chemistry. By confocal microscopy, the conjugates were shown to be clearly distributed in the cytoplasm within 3 h of incubation comparable to the other reported cMET FDC, MetFab-Doxorubicin (MetFab-DOX, IC₅₀ = 6.8 µg/mL and 0.58 µg/mL in HepG2 and Sk-Hep-1 cells, respectively), on a different cMET-expressing cell line [26]. While the tumor localization and tissue distribution of our α-cMET FDC-Duo5_P1 were not demonstrated in this study, MetFab-DOX was shown to localize to HepG2 ×enograft tumor within 24 h resulting in higher doxorubicin concentration in the tumor than after administration of doxorubicin alone. Moreover, its dye-fluorescent Fab version was detected at the tumor site 3 h post injection.

We found our FDCs to be very potent toward cMET-presenting cells with dramatic overexpression. This is also the case for the ADC, ABBV-399, that requires a certain threshold level of cMET expression for significant killing of tumor cells [22]. This is very encouraging as it indicates that normal cells that display low level of cMET, typically epithelial cells, can be spared from this strategy. The cMET expression threshold needed to activate our FDC requires more investigation by inclusion of various types of cMET-overexpressing cell lines. It was further demonstrated that the cell killing of our FDCs are antigen-mediated as the non-targeting FDC control did not show any cell killing regardless of whether it is high or low expressing cMET cell lines. Interestingly, although the ADC exhibited lower IC₅₀ than the FDC in vitro, its ability to completely kill the cell appears to be less than the FDC. We speculate that the presence of unconjugated anti-cMET IgG in the mixture likely contributed to an agonist effect that evokes biological response such as cell survival. This effect was illustrated in our Western blot analysis where the IgG has triggered the receptor dimerization and consequent downstream signaling not apparent with Fab and further corroborated by the result of the cytotoxicity study whereby the presence of certain amount of IgG has suppressed the cell killing of the FDC and has kept the cells alive. In future studies, the ADC can be further purified to exclude unreacted mAbs or more studies can be conducted to assess the agonistic activity of the pure ADC, if there is any. Notwithstanding, the naked fab did not exhibit an agonistic effect, hence its presence in the FDC formulation can be tolerated.

SNU-5 human gastric carcinoma xenograft in Nu/Nu mice was used to evaluate the in vivo efficacy of our anti-cMET FDC with Duo5 as payload. Our FDC was shown to have resulted in complete tumor growth inhibition with both the low (5 mg/kg) and high (15 mg/kg) dosing. It is remarkable how the tumors for these cohorts only revealed slight tumor regrowth (low dose of FDC) or no tumor regrowth (high dose of FDC) two months after cessation of the last dose. In contrast, both the non-targeting FDC and PBS vehicle controls resulted in continuous growth of tumor. While our FDC was highly efficacious in this model, the use of other models, including organoids or patient-derived primary tumor cells, will help further demonstrate its efficacy. On the other hand, the ADC at low (3 mg/kg) dosing was also very efficacious despite incomplete in vitro cell killing and less frequent dosing compared to FDC. This goes to show that efficacy overall is driven by a variety of factors including potency and pharmacokinetic properties, etc. Certainly, a less frequent dosing regimen for patients seems more attractive and preferable.

While fab-based therapeutics are mainly limited by their short half-lives due to renal filtration (cut off ~ 65 kDa), a shorter half-life may in fact confer a beneficial effect of lowering the overall systemic exposure to the highly potent cytotoxic drug such as in the case of a bicycle peptide toxin conjugate (BT8009) that is currently in Phase 1/2 multicenter clinical trial for Nectin-4 positive advanced solid tumors. This peptide-drug conjugate was found to rapidly clear from plasma (t_1/2 rat = 0.9 h, t_1/2 NHP = 1.7 h), yet it was able to retain in tumor and had demonstrated remarkable anti-tumor effects [40]. Compared to its ADC counterpart bearing the same payload, the peptide-drug conjugate seems to be dosed comparably in terms of payload amount but more frequently on a weekly basis without pause unlike the ADC in Phase I clinical trial, and still delivered low incidence of skin, ocular and neurological toxicities [41]. This data was demonstrated on limited subjects and more studies are warranted to prove its safety profile including the FDC itself. In our mouse study, every other day FDC dosing at 15 mg/kg for 14 times did not result in decreased body weight. The same is true with all other cohorts. A mouse model may not conclusively demonstrate potential toxicity. Animal models that can better predict toxicities in humans should be utilized in future studies to compare potential toxicity profiles of ADC and FDC.

The highly heterogeneous nature of solid tumors compounded by the deeply interconnected cMET/HGF and EGFR pathways make the search for effective treatment of these malignancies extremely difficult and challenging. While disease stabilization has been achieved for some of the MET-directed small molecule-based and antibody-based therapeutics, they have yet to demonstrate significant survival benefits. Moreover, there are subsets of patients that are not susceptible to these agents or have developed resistance. One of the promising emerging classes against cMET-overexpressing solid tumors is ADC. The FDC strategy described herein offers another alternative treatment option that relies upon overexpression of cMET on the tumor. One of the benefits it provides is the attenuated side effects compared to a chemotherapy drug as demonstrated in the preclinical testing of MetFab-Dox [26]. In addition, there is a threshold level of cMET expression required to activate its cell-killing ability, thus sparing the cMET low-expressing normal cells. With the advent of personalized therapies, a similar targeted approach, MET-CART (chimeric antigen receptor T-cell), is also under preclinical development [12,42]. Currently, as there is no one-size-fits-all solution drug/biologics available, the ability to recognize the strengths and weaknesses of each class of therapeutics and how to put together an effective combination therapy for a particular subset of patients bearing a particular type of cMET solid tumor malignancy is vital, which means that continuous studies and identification of biomarkers on solid tumors and its responsiveness to specific class of drugs are as crucially needed.

5. Conclusion

We have generated novel anti-cMET Fab drug conjugates that selectively target potent microtubule inhibitors toward highly overexpressing cMET tumors. Our anti-cMET FDCs demonstrated potent cell killing toward SNU-5 and HT-29, both cMET-high cell lines, and showed no cytotoxicity effect toward A549 and SK-BR-3, both cMET low cell lines. They also showed highly potent anti-tumor effects in a human gastric carcinoma xenograft mouse model. Future studies on the FDCs’ pharmacokinetics as well as tissue distribution data will be useful to provide better insights for more suitable dosing regimens and to help further establish the FDCs’ effectiveness as a potential therapeutic strategy not just for cancer but for other indications. Overall, this study highlights an antibody fragment drug conjugate as a promising targeted approach for solid tumors.

Declaration of interest

All authors were employed by Sorrento Therapeutics, Inc. (STI) when their part of the study was conducted. All work was carried out in STI. The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Author contributions

Y Fu, H Ji and G Kaufmann were involved in the conception. Y Fu, R Lim, M Buschman, H Zhou, J Chen and L Li were involved in study design and data interpretation. H Zhou carried out the phage display and SPR studies. A Khasanov and T Zhu synthesized all linker-payloads. R Lim, A Ledesma, J Niu and T Nguyen were involved in Fab generation, conjugation, purification, characterization studies. M Buschman, J Guo, RL and J Huang were involved in in vitro characterization experiments. J Chen, R Wang and L Li were involved in mouse in vivo study. L Kerwin carried out mAb production. Y Zhang carried out plasmid construction and optimization. Y Guo carried out the LC-MS analysis and co-wrote the methodology section. R Lim drafted the manuscript. J Huang assisted in the preparation of the manuscript. Y Fu, R Lim and L Li were involved in revising the manuscript.

Acknowledgments

We thank Ms. Elizabeth Orr for critical reading of the manuscript. M Buschman’s current affiliation is Janssen Research and Development. A Khasanov, J Niu, T Zhu, R Lim, L Kerwin, R Wang, L Li and J Chen are currently self-employed. A Ledesma’s current affiliation is QuidelOrtho. T Nguyen’s current affiliation is FormFactor Inc. J Guo’s current affiliation is Zentalis Pharmaceuticals. J Huang’s current affiliation is Boston University School of Public Health. Y Guo’s current affiliation is Arcturus Therapeutics. G Kaufmann’s current affiliation is Kyowa Kirin, Inc.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/14712598.2023.2292633

Additional information

Funding

This paper was not funded.