Discovery of a novel highly specific, fully human PSCA antibody and its application as an antibody-drug conjugate in prostate cancer

ABSTRACT

Prostate stem cell antigen (PSCA) is expressed in all stages of prostate cancer, including in advanced androgen-independent tumors and bone metastasis. PSCA may associate with prostate carcinogenesis and lineage plasticity in prostate cancer. PSCA is also a promising theranostic marker for a variety of other solid tumors, including pancreatic adenocarcinoma and renal cell carcinoma. Here, we identified a novel fully human PSCA antibody using phage display methodology. The structure-based affinity maturation yielded a high-affinity binder, F12, which is highly specific and does not bind to 6,000 human membrane proteins based on a membrane proteome array assay. F12 targets PSCA amino acids 63–69 as tested by the peptide scanning microarray, and it cross-reacts with the murine PSCA. IgG1 F12 efficiently internalizes into PSCA-expressing tumor cells. The antimitotic reagent monomethyl auristatin E (MMAE)-conjugated IgG1 F12 (ADC, F12-MMAE) exhibits dose-dependent efficacy and specificity in a human prostate cancer PC-3-PSCA xenograft NSG mouse model. This is a first reported ADC based on a fully human PSCA antibody and MMAE that is characterized in a xenograft murine model, which warrants further optimizations and investigations in additional preclinical tumor models, including prostate and other solid tumors.

Introduction

Prostate stem cell antigen (PSCA), encoding a 123- or 114-amino acid glycosylphosphatidylinositol (GPI)-anchored cell surface protein, belongs to the Thy-1/Ly-6 family, which has been proposed to be involved in diverse cellular functions, including signal transduction and cell – cell adhesion.¹ PSCA is expressed in normal prostatic basal epithelial cells and it may be physiologically involved in stem/progenitor cell self-renewal and/or proliferation.² PSCA is proposed to associate with prostate carcinogenesis and progression of prostate cancers (PCa).^3,4 It is overexpressed in all stages of PCa, ranging from high-grade prostatic intraepithelial neoplasia to primary androgen-dependent tumors and hormone-refractory metastases.⁵ Its overexpression correlates with disease advance, Gleason score, androgen independence, and metastasis.⁶ Nearly all bony metastasis overexpresses PSCA.¹ PSCA is also a valuable diagnostic and prognostic marker for PCa. It has been shown that the peripheral blood levels of PSCA mRNA are higher in PCa patients compared to control and benign prostatic hyperplasia (BPH) patients and are found to be associated with increased susceptibility to PCa.⁷ In addition, PSCA is highly expressed by nonprostatic malignancies, including renal, bladder, and pancreatic cancers.⁸ The high expression level in a variety of solid tumors, combined with limited expression in normal tissues, nominates PSCA as a promising theranostic target.

Antibodies serve as a promising type of theranostic reagents due to their high-binding specificity and low toxicity. Several PSCA monoclonal antibodies have been developed, most of which are based on mouse hybridoma, including clones 7F5,⁹ 1G8,⁶ 3C5⁶ and 8D11,¹⁰ with the exception of the human hybridoma Ha1–4.117¹¹ and the fully human antibody AGS-PSCA identified from human IgG transgenic mice.¹² They have been extensively characterized as IgG1, immunoradioconjugates, bispecific T cell engagers (BiTEs) or chimeric antigen receptor (CAR)-T in preclinical animal models. Some of them were tested in clinical trials (AGS-PSCA IgG1,¹² hu1G8 CAR-T¹³ and GEM3PSCA BiTE (NCT03927573)), but demonstrated limited anti-tumor efficacy.

The limited clinical outcome of current PSCA-targeting modalities necessitates the development of novel antibody-based therapeutics with enhanced efficacy and high specificity to overcome clinical hurdles. The “immuno-cold” tumor microenvironment (TME) and tumor plasticity-associated antigen heterogeneity are considered two hurdles for PCa immunotherapy.^14,15 Antibody-drug conjugates (ADCs), which rely on the antibody to specifically deliver chemotherapeutic drugs into the tumor niche, have been shown to be able to reprogram the immunosuppressive TME by inducing immunogenic cell death¹⁶ and circumvent tumor phenotype heterogeneity through bystander cell killing.^17,18 Currently, only one PSCA-targeting ADC has been reported to show anti-tumor activity in mouse models.¹⁰ However, this ADC is based on the murine hybridoma clone 8D11, making translation into the clinic unpredictable.

In the study presented here, we generated a fully human, highly specific, murine orthologue cross-reactive PSCA antibody by phage display, which we then made into a monomethyl auristatin E (MMAE)-conjugated antibody that shows dose-dependent anti-tumor efficacy and specificity in a human prostate cancer PC-3-PSCA xenografted NSG mouse model. Our results demonstrate the application potential of a novel, fully human PSCA antibody and ADCs made thereof for treating PCa and other PSCA-overexpressing solid tumors.

Materials and methods

Antibodies, bacteria, cells, and mice

The following antibodies were purchased from vendors: horseradish peroxidase (HRP)-conjugated anti-Flag tag antibody (Sigma-Aldrich, Cat# A8592-1 MG); HRP-conjugated anti-human Fc antibody (Sigma-Aldrich, Cat# A0170-1 ML); HRP-conjugated anti-M13 PIII antibody (Sino Biological, Cat# NC1883789); HRP-conjugated anti-mouse IgG (Sigma-Aldrich, Cat# A0168-1 ML); PSCA antibody 7F5 (Santa Cruz Biotechnology, Cat#sc -80,654); Phycoerythrin (PE)-conjugated anti-Flag tag antibody (Miltenyi Biotec, Cat# 130-101-576); PE-conjugated anti-human Fab kappa light chain (LC) antibody (ThermoFisher, Cat# MH10514). TG1 and DH5a cells were purchased from Lucigen (Cat#60502–1, Cat#60602–1, respectively). The E. coli HB2151(K12 ara Δ(lac-proAB) thi/F’ proA+B lacIq lacZΔM15 was described in our previous study.¹⁹ The helper phage M13KO7 was purchased from ThermoFisher (Cat# 18311019). The Expi293F cells were maintained in Expi293 expression medium (ThermoFisher, Cat# A1435103) with 0.4% penicillin-streptomycin (P/S). The prostate cancer cell lines PC-3 and Du-145 and the bladder cancer cell-line HT1376 were purchased from the American Type Culture Collection (ATCC). The PC-3 cell line was cultured at 37°C, 8% CO₂ in F12K medium supplemented with 10% fetal bovine serum (FBS) and 100 U/mL of penicillin – streptomycin. The Du-145 and HT1376 cells were cultured in complete Eagle’s Minimum Essential Medium (EMEM) medium. For establishing stable cell lines, the PC-3 and Du-145 cells were transfected by a PSCA-expressing plasmid (pSecTag B) carrying the zeocin selection gene. The post-transfection cells were selected in complete F-12K or EMEM medium with 200 μg/ml Zeocin. The Human Peripheral Blood Mononuclear Cells (PBMCs) were purchased from Zen-Bio Inc (Cat# NC0449441). Female NOD-scid IL2Rg^null mice (NSG mice, 6–8 weeks, Strain #:005557) and CBySmn.Cg-Prkdc^scid/J mice (BALB/c scid mice, Strain #001803) were purchased from the Jackson Laboratory. All mice were kept under specified pathogen-free conditions (SPF) in the animal facility of the University of Pittsburgh. All procedures were approved by the University of Pittsburgh Institutional Animal Care and Use Committee (IACUC).

Generation, expression and characterization of recombinant PSCA proteins and cell surface-associated PSCA protein

The PSCA allele gene of 114 amino acids (1–11: signal peptide; 12–86: cysteine-rich region; 86: GPI-anchor amidated serine; 87–114: the propeptide that is removed in the mature form) was synthesized by GenScript. To generate PSCA-avi-his protein, the PSCA 12–86 gene were subcloned into the pSecTag B vector (ThermoFisher) in frame to the C-terminal avi tag (GLNDIFEAQKIEWHE) followed by a 6×histine tag using the following primers:

HisF: 5-ACGCGGCCCAGCCGGCCctgctgtgctactcctgcaaag-3; HisR: 5-CGGTTTAAACTCAATGGTGATGGTGATGGTGTCCTTCGTGCCATTCGATTTTCTGAGCCTCGAAGATGTCGTTTAGTCCGCCGCTGGCGTTGCACAA-3. For the PSCA-Fc-avi protein, the human IgG1 Fc fragment with C terminal in frame with the avi tag has been incorporated into the pSecTag vector, as previously described.²⁰ We used the following primers to subclone PSCA 21–86 into this Fc-containing vector: FcF: 5- ACGCGGCCCAGCCGGCCctgctgtgctactcctgcaaag-3 (same as HisF); FcR: 5-TGTCGGGCCCgctggcgttgcacaagtc-3. The recombinant plasmids were sequenced and transfected into Expi293 cells for expression. The PSCA-his protein was purified by nickel-nitrilotriacetic acid (Ni-NTA) resin (ThermoFisher Scientific), while the PSCA-Fc protein was purified by the protein A resin (GenScript). Protein purity was estimated by sodium dodecyl sulfate – polyacrylamide gel electrophoresis (SDS-PAGE) and protein concentration was measured spectrophotometrically (NanoVue, GE Healthcare). The quality of the recombinant PSCA protein was validated by testing binding to the mouse anti-human PSCA antibody 7F5 using an enzyme-linked immunosorbent assay (ELISA). To generate cell lines that stably express membrane associated PSCA, the PSCA12–114 containing the GPI anchoring sequence with an N-terminal flag tag was subcloned into the pSecTag B vector carrying a murine Ig kappa-chain signal peptide by using following primers:

PSCA-mem-F: 5- ACGCGGCCCAGCCGGCC ggcgactacaaggacgatgacgataagggcctgctgtgctactcctgcaaagcc −3; PSCA-mem-R:

5-CGGTTTAAACCTAGAGCTGGCCGGGTCCCCAGAG-3. The PC-3 and Du-145 cells have been demonstrated to express PSCA mRNA with no detectable PSCA protein on the cell surface.²¹ The 293T, PC-3 and Du-145 cells were transfected by pSecTag-PSCA-mem plasmids using the Qiagen PolyFect® Transfection Reagent. The positive clones were selected in 200 ug/ml zeocin. The zeocin-surviving pools were single-cell sorted into 96-well plates (BD, Melody sorter) for clonal expansion. The individual positive clones were identified by flow cytometry with both PE-conjugated anti-FLAG tag antibody and 7F5 staining.

Phage panning of a human fab library, selection of binders, binder expression in bacteria and conversion of fab to IgG1

To identify PSCA Fab binders, the in-house fully human Fab library⁵ was used to pan the recombinant biotinylated PSCA-Fc-avi protein. PSCA biotinylation was performed through biotin ligase (BirA)-mediated enzymatic conjugation of a single biotin onto the AviTag.²² The biotinylation efficacy was validated by reactivity to streptavidin. The protein integrity was confirmed by ELISA binding to 7F5 post biotinylation. The bio-panning was done for four rounds with input antigens of 10 μg, 2 μg, 0.5 μg and 0.1 μg PSCA-Fc for rounds 1–4, respectively. The panning process began with incubation of antigens with 10¹² Fab phage particles followed by washing with phosphate-buffered saline (PBS) containing 0.05% Tween-20. Bound phage was pulled down by streptavidin-M280-Dynabeads and was rescued by infecting log-phase TG1 cells followed by viral package using the help of the M13KO7 helper phage. Panning enrichment was monitored by output phage titer and polyclonal phage ELISA. After the 4^th panning round, positive clones were selected by soluble expression monoclonal (SEM) ELISA followed by Sanger sequencing.²³ The positive hit clones were recombinantly expressed in E. coli HB2151 and purified by using Ni-NTA chromatography. The pure Fab proteins were tested for their binding affinity to recombinant PSCA, cell surface-associated PSCA, protein folding and aggregation. For conversion to IgG1, the VH and VL genes were subcloned into the pDR12 vector containing human IgG1 heavy chain (HC) CH1-CH2-CH3 cassette. The IgG1 was expressed in Expi293 cells and purified by protein A resin as described above.

Enzyme-linked immunosorbent assays

For detection of PSCA biotinylation efficacy, HRP-conjugated streptavidin (ThermoFisher, Cat# N100) was used. For confirmation of function of PSCA-Fc after biotinylation, 100 ng PSCA-Fc in PBS was coated onto plates (Costar) overnight at 4°C followed by addition of serially diluted murine antibody 7F5. Then HRP-conjugated anti-mouse IgG (Fc specific) was used for detection. For polyclonal phage ELISA, the panning output phage was incubated with immobilized PSCA-his and the bound phage was detected by HRP-conjugated anti-M13 PIII antibody). For semELISA, clones were randomly picked from the infected TG1 cells into 96-well plates, and protein expression was induced by 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG). The supernatant after centrifugation was incubated with immobilized PSCA-his. Binding was detected with HRP-conjugated mouse anti-FLAG tag antibody. For the pure Fab and IgG1 binding assay, serial dilution of antibody was incubated with immobilized antigen followed by detection of HRP-conjugated mouse anti-FLAG tag antibody and HRP-conjugated goat anti-human IgG Fc, respectively. For evaluation of binding to murine PSCA, the recombinant mouse PSCA protein (His tag) (Abcam, Cat#, ab241245) was coated into plates (100 ng/well), the IgG1 F12 was added and binding was detected by using HRP-conjugated goat anti-human IgG Fc. All colors were developed by 3,3,5,5-tetramethylbenzidine (TMB, Sigma) and stopped by 1 M H₂SO₄ followed by recording absorbance at 450 nm. Experiments were performed in duplicate and the results are denoted as mean ± standard deviation (SD). The binding EC₅₀s were obtained by fitting OD signal–concentration curves using non-linear regression model (variable slope, four parameters) using GraphPad Prism 9.0.

Biolayer interferometry

The Fab antibody affinities were analyzed by the biolayer interferometry BLItz (ForteBio, Menlo Park, CA). PSCA-Fc was mounted on the protein A sensor (ForteBio: 18–5010) for 2 min and equilibrated with Dulbecco’s PBS (DPBS) (pH = 7.4) to establish baselines. Then, the Fab G7 or F12 in 125 nM, 250 nM and 500 nM were used for association. The association was monitored for 2 min and then the antibody was allowed to dissociate in DPBS for 4 min. The background was established by running PBS buffer instead of antibody in the association process, and the background curve was subtracted from the antibody sensorgram. The k_on and k_off were derived from sensorgram fittings and used for K_D calculation.

Flow cytometry analysis

The PSCA expression on PC-3 and Du-145 PSCA transgenic cell lines were confirmed by PE-conjugated anti-Flag tag antibody and the murine PSCA antibody 7F5 followed by PE-conjugated anti-mouse IgG antibody (ThermoFisher, CAT#P-852). The intrinsic PSCA expression on HT1376 cells was detected by 7F5. Briefly, cells were detached by trypsin/ethylenediaminetetraacetic acid (EDTA) and cell viability was confirmed by trypan Blue (ThermoFisher). One million cells were incubated with 100 nM 7F5 antibody on ice for 1 h. After washing by PBS + 0.5% bovine serum albumin (BSA), bound antibodies were detected by phycoerythrin (PE)-conjugated anti-mouse IgG antibody. PE-A⁺ cells were detected by flow cytometry using the BD FACSCelesta. For evaluating Fab G7 and F12 binding, 500 nM Fab antibody were incubated with cells, followed by staining with PE-conjugated anti-human Fab kappa LC antibody. An irrelevant Fab (anti-SARS-CoV-2 Fab ab1²³) was used as the isotype control. The untransfected cells were also used for evaluating binding specificity. For further assessment of binding specificity on HT1376 cells, a fixed concentration (500 nM or 200 nM) of Fab G7 or F12 was incubated with HT1376 cells in the presence of gradually increasing concentrations of either 7F5 (0.01, 10, 100, 1000 nM) or the soluble recombinant PSCA-Fc protein (0.5, 5, 50, 125, 500, 1000 nM), with the bound Fab G7 detected by the PE-conjugated anti-Flag tag antibody. The mean fluorescence intensity was determined by FlowJo software, which was plotted against the concentrations of competitors to calculate IC₅₀ by non-linear fit in GraphPad Prism 7 (San Diego, CA).

Size-exclusion chromatography

The purity of the antibodies were analyzed by Superdex 200 Increase 10/300 GL chromatography (GE Healthcare, Cat# 28990944) using methods previously established by our laboratory.²⁴ Briefly, 100 uL of filtered antibodies in 1 × DPBS were analyzed. Antibodies were eluted by DPBS buffer at a flow rate of 0.5 mL/min. The column was calibrated with protein molecular mass standards of Ferritin (Mr 440 000 kDa), Aldolase (Mr 158 000 kDa), Conalbumin (Mr 75 000 kDa), Ovalbumin (Mr 44 000 kDa), Carbonic anhydrase (Mr 29 000 kDa), Ribonuclease A (Mr 13 700 kDa).

Dynamic light scattering

The aggregation resistance of Fab G7 was measured by dynamic light scattering (DLS). The antibody concentration was adjusted to 1 mg/mL and incubated at 37°C without shaking. Samples were detected by Zetasizer Nano ZS ZEN3600 (Malvern Instruments Limited, Westborough, MA, USA) to determine the antibody size distribution.

Affinity maturation of fab G7 by structure-guided mutagenesis

The structure of single-chain variable fragment G7 was modeled by the online server LYRA (Lymphocyte Receptor Automated Modeling) (http://tools.iedb.org/lyra/). Structure figures were prepared using PyMoL. The binding contribution by hydrophobic patches in the HC and LC CDR3s was visually scrutinized. Each of the hydrophobic hotspot residues were randomized into other 19 amino acids for improving antibody affinity. The randomization was fulfilled by incorporating the “NNS” codon into primers, and randomization of a total of eight residues were combined through overlapping PCR. The final Fab PCR product was sfi I digested and ligated into linearized pComb3× phagemid followed by electroporation of TG1 cells to make phage library. To retrieve affinity-enhanced clones, the koff selection panning process was performed. 1 ug of biotin-PSCA-Fc was incubated with 10¹² phage particles in blocking buffer (PBS +3% skim milk) for 2 h at room temperature (RT) followed by a 1-h incubation with streptavidin beads (ThermoFisher) and three washes with PBS. Then the bead-bound phage particles were competed with excess unbiotinylated PSCA-Fc (20 ug) for 2 h at RT. The beads were then washed by PBS + 0.05% Tween 20 for 10–15 times, followed by 5–10 times by PBS. Then the bound phage was recovered by infecting log-phase E. coli. The panning was performed in 2–3 rounds, followed by semELISA screening. The clones exhibiting high OD signals were collected and sequenced. The positive Fab hits were expressed and purified to evaluate binding EC₅₀ and K_D.

Epitope mapping by cyclic (conformational) peptide scanning

The sequence of PSCA was elongated with neutral GSGSGSG linkers at the C- and N-terminus to avoid truncated peptides. The elongated antigen sequence was converted into 7, 10 and 13 amino acid peptides with peptide–peptide overlaps of 6, 9 and 12 amino acids. After peptide synthesis, all peptides were cyclized via a thioether linkage between a C-terminal cysteine and an appropriately modified N-terminus. The conformational PSCA peptide microarray contained 246 different peptides printed in duplicate (492 peptide spots) and was framed by additional HA (YPYDVPDYAG, 80 spots) control peptides. The microarray was blocked in Rockland blocking buffer MB-070 for 30 min, then incubated with IgG1 F12 (0.1 µg/ml) in incubation buffer (washing buffer [PBS, pH 7.4 with 0.005% Tween 20] +10% blocking buffer) for 16 h at 4°C and shaking at 140 rpm. After washing, the goat anti-human IgG (H+L) DyLight680 (0.2 µg/ml) was added for staining for 45 min in incubation buffer at RT. We also used the mouse monoclonal anti-HA (12CA5) DyLight800 (0.5 µg/ml) as the positive control (to bind to the outmost microarray frame HA tag peptide). After washing, the fluorescence was detected using the LI-COR Odyssey Imaging System with the following parameters: scanning offset: 0.65 mm, resolution: 21 µm, scanning intensities of 5/7 (red = 700 nm/green = 800 nm). Microarray image analysis was done with PepSlide®Analyzer.

Membrane proteome array assay

Integral Molecular, Inc. (Philadelphia, PA) performed specificity testing of IgG1 F12 using the Membrane Proteome Array (MPA) platform,²⁵ in which 6,000 distinct human membrane proteins (representing over 94% of the human membrane proteome, including single-pass, multi-pass, and GPI-anchored proteins, including GPCRs, ion channels, and transporters) are individually overexpressed in live cells by plasmids transfection in separate wells of a 384-well plate. The entire library of plasmids is arrayed in duplicate in a matrix format for high-throughput screening. To optimize F12 concentrations and cell lines for screening, HEK-293T and QT6 cells were transfected with plasmids encoding known ligand targets, Protein A (binds antibody Fc; positive control), or vector alone (pUC; negative control). Based on the assay setup results, IgG1 F12 (20 μg/ml) was added to the MPA. Binding across the protein library was measured on an iQue3 (Ann Arbor, MI) by using the detection antibody AlexaFluor® 647-AffiniPure Goat F(ab’)₂ Anti-Human Fc (Jackson ImmunoResearch, Cat# 109-606-008). Each array plate contains both positive (Fc-binding) and negative (empty vector) controls to ensure plate-by-plate reproducibility. Test F12 interactions with any targets identified by MPA screening were confirmed in a second flow cytometry experiment using serial dilutions of F12, and the target identity was re-verified by sequencing.

Antibody-dependent cellular cytotoxicity assay

For the surrogate ADCC assay, we used Promega ADCC reporter cells (Cat#G7015), Jurkat T-NFAT-Luc2-CD16A cells (Jurkat T-CD16A cells), as effector cells. This cell line encodes the luciferase gene, for which expression is driven by the nuclear factor of activated T cells response element (NFAT-RE) promoter. Target cells (2.5 × 10⁴ cells of PC-3 or PC-3-PSCA or Du-145 or Du-145-PSCA) were incubated with Jurkat T-CD16A cells in an E: T ratio of 6:1, in the presence of serially diluted IgG1 G7 or F12. The SARS-CoV-2 antibody IgG1 ab1 was used as the isotype control. The assay was developed after overnight incubation. The luciferase activity was detected by using the Promega Bio-Glo Luciferase Assay System (Cat# G7941) according to the manufacturer’s instructions. The half maximal effective concentration of antibodies (EC₅₀) was calculated using GraphPad Prism 9. Natural killer (NK) cells were enriched from hPBMC by using the NK cell isolation kit II in the negative selection modes (Miltenyi Biotec). The cancer cells (2.5 × 10⁴ cells) were pre-opsonized by different concentrations of IgG1 F12 at 37°C for 30 min, followed by incubation with PBMCs (1.25 × 10⁶ cells) or NK cells (2.5 × 10⁴ cells) for 4 h. Then, cell killing was detected by the Promega LDH-Glo™ Cytotoxicity Assay (Cat# J2380) according to the manufacturer’s instructions. The experiments were performed in duplicated wells. Target cells maximum signal (T_max) was achieved by treatment of the target cells with Triton X-100; target cells alone (T.C), effector cells alone and target cells with effector cells in the absence of F12 (no F12) were also set. Luciferase activity was recorded by BioTek synergy multi-mode reader (Winooski, VT). Cell killing percentage was calculated as: 100×(with F12 - no F12)/(T_max-T.C).

F12 internalization assay

We used the Promega pH sensor dye (pHAb Dye, CAT# G9831), which has very low fluorescence at pH > 7, and a dramatic increase in fluorescence as the pH of the solution becomes acidic. pHAb Amine Reactive Dye(a) has a succinimidyl ester group that reacts with primary amines available on the lysine amino acids on the antibodies. IgG1 F12 or anti-CD276 antibody m276 was conjugated with the pHAb dye by lysine conjugation. The cancer cells (2.5 × 10⁴ cells of PC-3 or PC-3-PSCA or Du-145 or Du-145-PSCA) were pre-seeded in 96-well plates, and incubated with different concentrations of F12-pHAb (0, 10, 100 and 400 nM; conjugate concentration was determined based on the manufacturer’s instructions) for 12 or 24 h, either at 37°C or 4°C. Then the medium was removed, cells (attached) were gently washed 3 times with PBS, and the fluorescence was recorded by the BioTek synergy Fluorescence Plate Reader (Ex: 530 nm/Em: 565 nm). The signals were normalized to the background (same number of PC-3-PSCA cells without adding F12-pHAb). For comparison with m276, F12-pHAb and m276-pHAb at the same concentration (400 nM) were incubated with cancer cells. The fluorescent signal was normalized to the PC-3 (PSCA negative) + F12-pHAb treatment group. For competition with naked IgG1 F12, the fixed concentration (400 nM) of F12-pHAb conjugate was incubated with PC-3-PSCA or Du-145-PSCA cells in the presence of gradient concentrations of naked IgG1 F12 (0, 100, 1000 nM). All experiments were performed in duplicate and the error bars denote ±1 SD.

ADC preparation and detection of release of free MMAE after serum incubation

The ADC IgG1 F12-MMAE, and the isotype control Ab1-MMAE were made by NHS-lysine coupling. The amine reactive linker-payload compound, OSu-Glu-vc-PAB-MMAE, (LEVENA BIOPHARMA, CAT#SET0100) was used for conjugation. Antibody (1–2 mg/ml) was buffered into PBS, 1.5 mM EDTA, 25% propylene glycol (PPG), with incubation with a 10 folds molar excess of the drug solution (solubilized in 1:1 DMSO and PPG). The conjugation was performed at RT for 4–6 h. Then the reaction was stopped by buffer change into clean PBS by filter centrifugation (30 kDa Amicon centrifugal filter unit Millipore Sigma CAT#UFC8030). ADC concentration was determined by UV 280 nm absorbance. The conjugate was then filtered through a 0.2-μm filter under sterile conditions and stored at −80°C for analysis and testing.

The ADC quality was checked by SDS-PAGE and ELISA binding to PSCA. The DAR was determined by the intact mass LC/MS analysis by the Poochon Scientific company. The MS raw files were analyzed by using BioPharma Finder 4.0. To evaluate the stability of F12-MMAE in mouse serum, we detected free MMAE by LC-MS/MS after incubating F12-MMAE with fresh mouse serum. 0.125 mg/ml F12-MMAE was incubated with 100 ul fresh PBS-diluted (v/v, 62.5%) mouse serum for 4 days. We incorporated blank mouse serum and the MMAE standard into experiments as controls. The serum samples were processed for extraction of MMAE by 5 folds of acetonitrile and then purified by C18 column. The purified materials or MMAE reference were analyzed by LC/MS/MS. The MS raw files were analyzed by using Compound Discoverer 3.3.

ADC in vitro cytotoxicity

The PC-3 or PC-3-PSCA cells (2.5 × 10⁴ cells) were pre-seeded in 96-well plates overnight. The cells were then incubated with serially diluted ADC, or naked IgG1 antibody or linker-payload combination compound for 4–5 days at 37°C, 8% CO₂ with 95% humidity. Then the cell viability was detected by the Promega CellTiter-Glo® Luminescent Cell Viability Assay (CAT# G7570), which is based on quantitation of the ATP present, an indicator of metabolically active cells. The Luminescence was recorded by the BioTek synergy multi-mode reader (Winooski, VT). The killing percentage was calculated by the formula: (1-Lumi _{with compounds}/Lumi _{without compounds}) ×100. The GraphPad Prism9 for was used for calculation of killing IC₅₀. All experiments were performed in duplicate and the error bars denote ±1 SD.

Pharmacokinetic study

About 6–8-week-old NSG mice (Jackson Laboratory) were randomly divided into groups (n = 4) and were intraperitoneally (i.p.) administered with IgG1 F12 or IgG1 F12-MMAE at a dose of 6 mg/kg. 10 μl blood were collected from each mouse at time points 5 h, 1 day, 3 days, 6 days, 9 days, and 14 days and diluted by 10-fold with PBS. The cells were removed by centrifuge at 1000×g for 10 min and plasma samples were stored at −80°C. The total antibody concentration of both ADC and unconjugated antibodies were tested by ELISA. Briefly, the 96-well plate (half area, high-binding) were coated with the PSCA recombinant protein at 4 μg/ml at 4°C overnight. The plate was then blocked with 5% BSA at RT for 2 h. After washing, diluted serum was added to each well and incubated at 37°C for 1 hr. After washing three times with PBS-Tween (0.05%), the HRP-conjugated anti-human Fc antibody was added and incubated for an additional 1 h. The plate was then washed and color development was performed by adding TMB. To measure ADC concentration, the mouse anti-MMAE antibody (1:1000, Acro Biosystems, Cat# MME-M5252) was added followed by the HRP-conjugated anti-mouse IgG antibody at a dilution fold of 1:5000. Concentration was calculated by a standard curve. The PK parameters were obtained by fitting the curves of sera antibody concentration vs. time post injection using the noncompartmental pharmacokinetics data analysis software PK solutions 2.0^TM.

Tolerability assay

6–8 week old NSG mice (n = 3 per group) were injected at a single i.p. dose of 20 mg/kg IgG1 F12-MMAE or IgG1 F12. Body weight was monitored every day for 2 weeks. At the endpoint, mice were euthanized and whole blood was collected for alanine transaminase (ALT) detection. Mouse plasma ALT level was measured by using an ALT ELISA kit purchased from Abcam (Cat# ab282882) following the manufacturer’s protocol.

In vivo treatment study in a xenograft mouse model

For evaluation of in vivo ADCC effects of IgG1 F12, PC-3-PSCA cells (5 × 10⁶ cells/mouse) in 200 ul of 1:1 PBS/Matrigel (BD Biosciences, Cat# 354234) were injected subcutaneously (s.c.) into the right flank of the Balb/c scid mice. When tumor volume reached 100 mm³, mice were divided randomly into IgG1 F12 treatment or vehicle group and i.p. administered four times total with a four-day interval. For evaluation of the ADC (F12-MMAE or Ab1-MMAE) efficacy in vivo, PC-3-PSCA cells (5 × 10⁶ cells/mouse, 5 mice per group) in 200 ul of 1:1 PBS/Matrigel were injected s.c. into the right flank of the NSG mice. Mice were randomly divided into the different treatment groups when tumors reached 100 mm³. We conducted ADC treatment studies at three different dosing regimens, comparing PSCA-ADC with the isotype control Ab1-ADC. The regimens design were as follows: dose 1): 1 mg/kg and 3 mg/kg, administered twice weekly (2XQ1W); dose 2): 6 mg/kg, administered twice weekly (2XQ1W), followed by a single injection of 3 mg/kg when tumors regrew to 140 mm³; dose 3): 6 mg/kg, administered four times weekly (4XQ1W). All ADCs were i.p. administered. Tumor dimensions and mouse weight were measured periodically with a slide caliper and scale. Tumor volume was calculated by the formula: V = 0.5×length × (width)². Tumor growth inhibition was calculated by the formula: %TGI = [1-(mean tumor volume (MTV)_{ADC treated}/MTV_control)] × 100. Mice were euthanized when the tumor volume exceeded 1000 mm³ or if the animals showed any signs of suffering. At the end of the experiment time point, tumors were isolated and weighed. In the dose 2 experiment, one week after two 6 mg/kg ADC treatments, ~50 ul blood were collected to the purple-topped K+/EDTA tubes, and subjected to complete blood counts using the Abaxis HM5 machine from In vivo Imaging Facilities in Hillman Cancer Center of the University of Pittsburgh. In the dose 3 experiment (6 mg/kg, 4×Q1W), mouse spleen, kidney, heart, lung, and liver were collected and fixed in formalin. The H&E staining were completed by the Biospecimen Core of the University of Pittsburgh.

Quantification and statistical analysis

All statistical analyses were performed using the GraphPad Prism9.0 software. Statistical significance between/among groups was determined using one-way ANOVA and two-way ANOVA, followed by Tukey correction. The comparison of mouse survival curves was done by the log-rank test. A p value < 0.05 was considered significant. ns: p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001.

Results

Identification of a fully human fab against PSCA by phage panning

We produced the recombinant human PSCA protein fused with the human immunoglobulin G1 Fc (PSCA-Fc), in frame to an avi tag at the C terminus, which was used for BirA-mediated site-specific conjugation of biotin. The biotinylated PSCA-Fc-avi protein was validated by binding to streptavidin and the murine PSCA antibody 7F5⁹ (Figure S1A-C). The PSCA-Fc-biotin protein was panned against our in-house fully human Fab phage library.⁵ High throughput semELISA²⁶ screening by using PSCA-his (to exclude undesirable Fc tag binders) output a promising dominant binder, G7, which showed high binding affinity to PSCA with a half-maximal binding concentration (EC₅₀) of 7.6 nM, and a dissociation binding constant (K_D) of ~30 nM, as tested by ELISA and BLItz, respectively (Figure 1(a,b)). FabG7 exhibited a homogenous monomeric folding as tested by the size-exclusion chromatography and did not aggregate according to the dynamic light scattering measurement (Figure S1D and E). Fab G7 bound to PC-3 and Du-145 prostate cancer cell lines transgenically overexpressing PSCA (PC-3-PSCA and Du-145-PSCA), without binding to wild type (un-transfected) PC-3 and Du-145 and CHO-K1 cells (Figure 1(c)). The PSCA protein was not detected on cell surface of wild-type PC-3 and Du145 cells, which is consistent with the observations by Priceman et al,²¹ although a trace of PSCA mRNA has been transcribed by these prostate cancer cell lines. The 2–3 logs MFI shift indicated that PSCA is robustly expressed by our PC-3-PSCA and Du145-PSCA stable cell lines. Based on comparisons with results from other research groups^21,27 and RNA-seq data from the Human Protein Atlas database, we hold that the PSCA expression level in our stable cell lines may fall within the range observed in prostate tumors and this expression level is likely higher than that found in most normal tissues.

Figure 1.

Discovery and characterization of the PSCA antibody fab G7.(a) fab G7 binding to recombinant PSCA-Fc protein measured by ELISA. Bovine serum albumin (BSA) was used as a negative control. Experiments were performed in duplicate and the error bars denote ± SD, n = 2. (b) Kinetics of fab G7 binding to PSCA-Fc, as measured by Blitz. (c) Fab G7 binding to PSCA positive (PC-3-PSCA, Du-145-PSCA and HT1376 cells) and PSCA negative cells (PC-3, Du-145 and CHO-K1 cells) as tested by flow cytometry. An irrelevant fab (anti-SARS-CoV-2 fab ab1) was used as the isotype control. Fab G7 at the concentration of 500 nM was incubated with cells. (d-g) competition of fab G7 and fab F12 binding to HT1376 cell surface-associated PSCA by the recombinant PSCA-Fc protein (d and f) and by the murine PSCA antibody 7F5 (e and g). 500 nM of fab G7 or 200 nM of F12 was incubated with cells in the presence of gradient concentration of competitors. The bound fab G7 or F12 was detected by the pe-conjugated anti-flag tag antibody.

Fab G7 binding to cell surface-associated PSCA was also validated by testing on the HT1376 bladder cell line that intrinsically expresses PSCA (as tested by the well-known PSCA antibody 7F5, Figure S1F). Importantly, Fab G7 binding to HT1376 cells can be outcompeted by either the soluble recombinant PSCA-Fc protein or by the murine PSCA antibody 7F5 in a concentration-dependent manner (Figure 1(d, e) and S2A, B), further supporting Fab G7 cell-binding specificity and indicating that Fab G7 shares an overlapping epitope with 7F5.

Affinity maturation of fab G7 outputs a high-affinity, highly specific PSCA antibody F12

To affinity-mature Fab G7, we modeled its structure by the online server LYRA (Lymphocyte Receptor Automated Modeling).²⁸ We found two hydrophobic patches, one consisting of the HC complementarity-determining region 3 (CDR3) residues L96 and W98 (Kabat numbering) and the LC framework (FR)2 residues, L46 and Y49; the others consisting of HC CDR2 residues Y50 and Y58 and LC CDR3 residues L94 and L96 (Figure 2(a)). Since hydrophobic patches in CDRs are typically important for antigen binding,²⁹ we hypothesized that FabG7 binding affinity can be enhanced/modulated by engineering these two patches. We randomized those eight hydrophobic residues into 19 alternative amino acids and generated a phage library with a size of 19⁸ (1.7 × 10¹⁰). We then used a kinetic (k_off) selection-based phage panning process, in which excess non-biotin labeled PSCA was used to outcompete biotin-PSCA bound phage antibody for a period of time, and the streptavidin bead-retained phage was retrieved and used for amplification for the next round panning. By this method, we obtained an affinity-improved clone, F12 with a 5–10 fold higher binding EC₅₀ of 1.2 nM, as tested by ELISA (Figure 2(b)). This is consistent with the BLItz kinetic measurement, showing a K_D of 1.16 nM. The higher affinity for F12 is attributed to its significantly slower k_off ((9.43 ± 1.36) × 10⁻⁵ s⁻¹) than G7 ((4.14 ± 0.03) × 10⁻³ s⁻¹) (Figure 2(c)), corroborating the effectiveness of our k_off panning strategy. Similar to the prototypic Fab G7, the Fab F12 binding to cell surface-associated PSCA was outcompeted by the recombinant PSCA protein (PSCA-Fc) and the positive antibody 7F5 (Figure 1(f) and (g); Figure S2C and D), indicating that our PSCA antibody retains its binding specificity and did not experience “epitope shifting” during the affinity maturation process.

Figure 2.

Characterization of the affinity-maturated PSCA antibody F12. (a) the structure model of fab G7, represented as the cartoon model using PyMoL. The light chain and heavy chain CDR3 and FR2 hydrophobic residues are depicted as cyan and red sticks. The orange dotted lines highlight hydrophobic patches in antibody paratopes. (b) Evaluation of binding affinity to PSCA-Fc for affinity-enhanced clones by ELISA. Fab G7 is the parental clone. (c) The kinetics of fab F12 binding to PSCA-Fc, as measured by Blitz. (d) MPA assay to evaluate F12 binding specificity. IgG1 F12 was tested for binding to as many as 6,000 human transmembrane proteins that are transgenically expressed in HEK293 cells in a high throughput manner. (e) Epitope mapping of F12 by the conformational peptide scanning. The PSCA protein was scanned by cyclic peptides with 7, 10 and 13 amino acid length with peptide-peptide overlaps of 6, 9 and 12 amino acids. The conformational PSCA peptide microarray was framed by the HA control peptides. F12 binding was detected by the goat anti-human IgG (H+L) DyLight680 (red color), and the array outmost HA tag peptide was detected by the anti-ha (12CA5) DyLight800 (green color). (f) IgG1 F12 binding to murine PSCA recombinant protein by ELISA. Detection was achieved by hrp-conjugated anti-human fc antibody. Experiments were performed in duplicate and the error bars denote ± SD, n = 2.

The specificity of F12 was further substantiated by the membrane proteome array (MPA) assay,²⁵ in which ~6,000 human membrane proteins expressed on HEK293T cell surface are evaluated for binding to F12 through a high throughput flow cytometry. In this assay, F12 specifically bound to PSCA without any binding to other human proteins (Figure 2(d)). To map the F12 epitope on PSCA, we used a conformational PSCA peptide microarray containing C-N termini cyclized peptides consisting of 7, 10 and 13 amino acids with peptide–peptide overlaps of 6, 9 and 12 amino acids. Results showed that F12 bound to a single epitope-like spot pattern formed by adjacent peptides with the consensus motif DDSQDYY (amino acids 63–69) with all peptide lengths at high signal-to-noise ratios (Figure 2(e)). Furthermore, we found F12 cross-reacts with the murine PSCA, with a binding EC₅₀ of 57.9 nM (Figure 2(f)). Collectively, through a structure-based affinity maturation strategy, we identified an affinity-enhanced, highly specific, murine orthologue cross-reactive PSCA antibody F12.

IgG1 F12 effectively internalizes into cytoplasm of PSCA+ prostate cancer cells

The GPI-anchored cell surface proteins can be endocytosed into the recycling endosomal compartment after binding with their ligand or antibody, which is important for their physiological function and signaling transduction.³⁰ To explore if the antibody F12 can be internalized into endosome after engaging with PSCA, we used a pH-sensitive dye (pHAb) that has very low fluorescence at pH >7, but emits bright fluorescence in acidic pH (Figure 3(a)). pHAb was conjugated onto IgG1 F12 through lysine-amine-NHS coupling. The F12-pHAb conjugate was incubated with prostate cancer cells for a period of time followed by washout and fluorescent detection. The results showed that IgG1 F12 was specifically internalized into the acidic endosome of PC-3-PSCA and Du-145-PSCA, but without internalization into untransfected PC-3 and Du-145 cells (Figure 3(b,c)). Moreover, IgG1 F12 internalization occurred in a concentration-dependent manner on both cell lines. Interestingly, on PC-3-PSCA, but not Du-145-PSCA, cells, IgG1 F12 internalization exhibited time dependence, i.e., treatment for 24 h induced higher internalization signal than for 12 h (Figure 3(b,c)), which may relate to the higher proliferation capacity of PC-3-PSCA cells and/or higher expression level of PSCA on the PC-3 cell surface (Figure 1(c)). In addition, internalization was expectedly dependent on temperature, as treatment at 4°C did not elicit any internalization fluorescent signals (Figure 3(d)). To gauge IgG1 F12 internalization efficacy, we compared its signal to the benchmark anti-B7-H3 (CD276) antibody, m276, which is reported to have efficient internalization.³¹ The PC-3 and Du-145 have been reported to intrinsically express B7-H3 (CD276).³² Results showed that at 400 nM, IgG1 F12 exhibited higher internalization signals than IgG1 m276 (Figure 3(e)). To further ascertain IgG1 F12 internalization, we competed internalization of the fixed concentration (400 nM) of F12-pHAb conjugate by using gradient concentrations of naked IgG1 F12. Our results showed that the naked IgG1 F12 decreased the F12-pHAb internalization fluorescence signals in a concentration-dependent manner on both PC-3-PSCA and Du-145-PSCA cells (Figure 3(f)). Based on these results, we concluded that F12 can robustly internalize into endosome after binding to cell surface PSCA.

Figure 3.

The internalization of IgG1 F12 into PSCA expressing cancer cells. (a). The scheme of the pH sensitive dye that emits fluorescence only at acidic endosome. (b-d). Evaluation of internalization of IgG1 F12 by using the pH sensitive dye and antibody conjugate. IgG12 F12 was conjugated with pHab dye by the lysine amine-nhs coupling. Different concentrations of F12-pHab were incubated with PSCA+ or – PC-3 and Du-145 cells for 12 or 24 hrs at either 37°C (B and C) or 4°C (d). After washing with 3× PBS, the cell fluorescence was recorded by the fluorescent plate reader, and the signal was normalized to the background. F12 internalization into PC-3 and Du-145 cells was presented in (b) and (c) respectively. (e) Comparison of IgG1 F12 internalization to that of benchmark antibody IgG1 m276, which targets CD276 (B7-H3). Both WT PC-3 and Du-145 cells intrinsically express CD276. (f) The internalization of F12-pHab conjugate can be outcompeted by the naked IgG1 F12 antibody on both PC-3-PSCA and Du-145-psca cells in a concentration dependent manner. The two-way ANOVA followed by Tukey correction was used for statistical analysis. ns: p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001.

IgG1 F12-MMAE ADC robustly and specifically killed PSCA+ cancer cells in vitro

Having demonstrated that IgG1 F12 is able to efficiently internalize into PSCA-overexpressing tumor cells, we next studied tumoricidal effects of the IgG1 F12 ADC. We ligated the microtubule Inhibitor MMAE into IgG1 F12 through lysine amine-N-hydroxysuccinimide (NHS) ester coupling (Figure 4(a)). We used the cleavable valine-citrulline dipeptidyl linker and the self-immolative para-aminobenzylcarbamate (PAB) spacer to separate the MMAE moiety from antibody. We made the isotype control ADC Ab1-MMAE using the same linker configuration and conjugation procedure. Ab1 is an anti-SARS-CoV-2 antibody with high specificity and aggregation resistance.²³ Due to the random conjugation, the F12-ADC mass spectrum exhibited a heterogenous drug antibody ratio (DAR) with an average of 1.98 (Figure 4(b)). The antibody integrity was assessed by SDS-PAGE (Figure 4(c)). The antigen binding was retained after conjugation (Figure 4(d)). The ADC IgG1 F12-MMAE specifically killed PC-3-PSCA cells with an EC₅₀ of 5.41 nM, without any cytotoxicity on PC-3 cells (Figure 4(e, f)). The unconjugated antibody IgG1 F12 did not affect cell viability (Figure 4(e, f)). Both the isotype control ADC Ab1-MMAE and unconjugated Ab1 showed no cytotoxicity against PC-3-PSCA and PC-3 cells at the tested concentration range (1–100 nM), demonstrating the targeting specificity of IgG1 F12 ADC. The specificity of F12-ADC is consistent with the MPA results showing that the antibody F12-IgG1 is highly specific and did not bind to 6,000 human transmembrane proteins expressed on 293 T cells (Figure 2(d)). Interestingly, the linker-payload compound, NHS-Glutarate-Val-Cit-PAB-MMAE (designated as vc-MMAE) proved inefficient in killing PC-3 cells, causing substantial cytotoxicity only at 500 nM. This contrasts with the reported high potency (EC₅₀ of single digit nM) of the free MMAE payload against PC-3 cells.³³ The cytotoxicity discrepancy may be reconciled by the different physicochemical properties, including lipophilicity (LogP), between free MMAE and vc-MMAE. The insufficient killing efficacy for vc-MMAE alone underscores the crucial role of IgG1 F12 in delivering MMAE across the cell plasma membrane into the endosome, followed by drug release into cytosol for enlisting the blockade of tubulin polymerization and inducing cell death.

Figure 4.

Generation and characterization of ADC IgG1 F12-MMAE. (a) the scheme of ADC IgG1 F12-MMAE. MMAE was conjugated onto IgG1 F12 by the nhs-amine coupling through a cleavable dipeptidyl linker and the PAB spacer. (b) The intact LC/MS spectrum of IgG1 F12-MMAE to determine DAR. The ADC exhibits a heterogenous distribution of DAR with an average DAR of 1.98. (c and d). The quality controls of ADC by SDS-PAGE and ELISA. (c). F12-MMAE and Ab1-mmae HC and LC integrity were checked on SDS-PAGE. (d).The binding to PSCA antigen by F12 was ascertained after conjugation with MMAE. (e and f) the in vitro cytotoxicity of IgG1 F12-MMAE, Ab1-mmae with comparison with the naked IgG1 F12 and Ab1 antibody and the linker-payload combination small molecule (vc-mmae). Compounds were incubated with PC-3-PSCA or PC-3 cells for 4-5 days. The cell viability was detected by the Promega CellTiter-Glo® luminescent cell viability assay.

The pharmacokinetics and tolerability of ADC IgG1 F12-MMAE

Before evaluating ADC anti-tumor effects in vivo, we first assessed the pharmacokinetic (PK) profile of the ADC F12-MMAE in mice. The unconjugated IgG1 F12 was used as a control. A single dose (6 mg/kg) of ADC and parent IgG1 were i.p. administered and blood samples were collected at predefined time points. Indirect ELISA was performed to determine the plasma concentration of ADC and the unconjugated antibody. We used the anti-human IgG antibody to detect the total antibody concentration (Figure 5(a)), while using the anti-MMAE antibody to specifically detect the intact ADC level (Figure 5(b)). For the ADC IgG1 F12-MMAE, the ADC decay rate is slightly higher than the total antibody (Figure 5(c)). Consistent to most ADCs,³⁴ the elimination rate for IgG1 F12-MMAE is faster than the unconjugated IgG1 F12 (0.428/day vs. 0.332/day). Thus, the ADC half-life is shorter than the naked IgG1 (1.619 d vs. 2.085 d). At 3 days post injection, the plasma ADC concentration remained at ~3 μg/ml (~20 nM), which is ~4-fold higher than the in vitro cell killing EC₅₀ (5.41 nM, Figure 4(e)). Next, we evaluated the potential toxicity of the ADC IgG1 F12-MMAE. The healthy NSG mice were injected with IgG1 F12-MMAE or unconjugated IgG1 F12 at a high dose of 20 mg/kg. Mouse behavior and body weight were monitored for 2 weeks. Results showed that mice tolerated this dose level as evaluated by the unchanged body weight and blood alanine transaminase (ALT) level (Figure 5(d,e)). However, it should be noted that the affinity of F12 to the murine PSCA (EC₅₀ = 57.9 nM, Figure 2(f)) is much lower than to the human PSCA (EC₅₀ = 1.2 nM, Figure 2(b)). This large difference in affinity may influence the observation of lack of toxicity of F12-ADC in NSG mice.

Figure 5.

The in vivo pharmacokinetics (PK) and tolerability test. (a and b) PK of IgG1 F12 and ADC IgG1 F12-MMAE in NSG mice (n = 4). Blood were collected at the indicated time point, and total antibody (a) and ADC (b) were quantified by indirect ELISA; (c) the PK parameters including clearance rate, τ1/2 and AUC for IgG1 F12-MMAE and naked IgG1 F12 were obtained by fitting the plots of sera antibody concentration vs. time post injection using the noncompartmental pharmacokinetics software PK solutions 2.0. (d) Body weight change after a high single dose administration of naked antibody IgG1 F12 or ADC at 20 mg/kg (n = 3); (e) blood ALT level at 15 days post injection of a single dose of IgG1 F12 or ADC at 20 mg/kg. The two-way ANOVA followed by Tukey correction was used for statistical analysis. ns: p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001.

The ADC IgG1 F12-MMAE exhibits dose-dependent anti-tumor efficacy and specificity in the PC-3-PSCA xenograft mouse model

Next, we sought to assess the ADC IgG1 F12-MMAE therapeutic efficacy, with comparison to the isotype control ADC Ab1-MMAE in a prostate tumor xenograft mouse model. Mice were injected with PC-3-PSCA cells by s.c. administration. ADC treatment was initiated when tumor size reached an average of 100 mm³. Based on the PK studies, the F12-MMAE level dropped below 1 μg/ml (6.6 nM) at day 7 post injection. Thus, we i.p. injected ADCs into mice once weekly. We explored F12-MMAE efficacy and specificity across three different dose regimens. At low doses of 1 and 3 mg/kg, 2XQ1W, F12-MMAE exhibited dose-dependent tumor inhibitory activity. The 3 mg/kg dose of F12-MMAE stabilized tumor size at 100 mm3 for ~20 days (stable disease, SD) followed by tumor progress, while the isotype control Ab1-MMAE only slightly prolonged tumor growth without halting it (Figure 6(a)). Consequently, the 3 mg/kg F12-MMAE treatment significantly extended mouse survival compared to the low dose and control ADC groups (Figure 6(b)). We then increased the dose to 6 mg/kg, 2XQ1W, at which F12-MMAE largely regressed tumor size (partial remission, PR). At this dose, the control ADC ab1-MMAE also retarded tumor growth. However, F12-MMAE demonstrated significantly superior tumor growth inhibition and mouse survival extension compared to ab1-MMAE (Figure 6(c,d)). After ADC washout, tumors relapsed. When tumor size regrew to 140 mm³, we re-treated mice with a single dose of 3 mg/kg ADC (38 days post-first injection). F12-MMAE remained effective against relapsed tumors, while the control ADC Ab1-MMAE failed to inhibit tumor growth (Figure 6(c)). Next, we further increased the dose to 6 mg/kg, 4XQ1W and found that F12-MMAE almost eradicated tumors (complete remission, CR; %TGI = 98.3%) (Figure S3A). However, at this dose, the isotype ab1-MMAE also slowed tumor growth (Figure S3B). To explore reasons of the non-specificity observed at high doses, we have detected that a small portion of ADC F12-MMAE was cleaved to release free MMAE after ex vivo incubation with mouse sera (Figure S4). This observation suggests that the early payload deconjugation due to cleavage of the ADC dipeptidyl linker (vc-PABC) by extracellular proteases may partially contribute to the ADC non-specificity.³⁵ Another contributing factor may be attributed to the ADC off-target catabolism occurring in normal cells.³⁶ Based on these results, we concluded that our ADC F12-MMAE exhibits dose-dependent efficacy and specificity.

Figure 6.

In vivo therapeutic effects of ADC IgG1 F12-MMAE in prostate cancer PC-3-PSCA xenograft mouse model.(a) inhibition of tumor growth by F12-MMAE at the dose of 1 mg/kg and 3 mg/kg, 2XQ1W. The NSG mice were s.C. grafted with PC-3-PSCA cells (n = 5). ADC was intraperitoneally administered. Ab1-mmae was used as the isotype control. Tumor sizes were monitored twice a week; (b) the mouse survival across different treatment groups in (A). Mice were euthanized when the tumor volume exceeded 1000 mm3 or if the animals showed any signs of suffering. (c) The therapeutic efficacy of F12-MMAE when dosed at 6 mg/kg, 2XQ1W. After ADC washout, tumor relapsed. When tumors regrew to 140 mm³(38 days post-first injection), a single injection of 3 mg/kg of ADC was administered to evaluate the response of ADC to relapsed tumors. (d) The mouse survival across different treatment groups in (c). The two-way ANOVA followed by Tukey correction was used for statistical analysis of tumor size. The comparison of mouse survival curves was done by the log-rank test. ns: p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001.

Regarding toxicity, mice tolerated the high dose of 6 mg/kg (4×Q1W), as evidenced by stable body weight and the absence of abnormal tissue lesions (Figure S3C and D).

MMAE targets mitotic cells, making actively dividing cells, including hematopoietic progenitor cells, susceptible to its cytotoxicity. Clinically, MMAE can affect the production of various blood cell lineages, potentially leading to side effects such as anemia, neutropenia, and thrombocytopenia.³⁷ We examined hematological changes in mice 7 days after two 6 mg/kg ADC treatments. We found that most blood cell populations remained unaffected following ADC treatment (Figure S3E). For the white blood cells (WBCs) and neutrophils (Neu), their numbers were 2.8 × 10⁹/L and 1.7 × 10⁹/L blood, respectively post ADC treatment, falling at the physiological range for NSG mice (0.42–2.6 × 10⁹/L for WBCs and 0.44–2.05 × 10⁹/L for Neu).³⁸ Interestingly, their numbers in the vehicle group were highly elevated (8.51 × 10⁹/L for WBCs and 7.29 × 10⁹/L for Neu) (Figure S3E), probably due to the large tumor burden.³⁹

Discussion

PSCA is a valuable diagnostic and therapeutic target for PCa. There is no effective treatment for PCa when it becomes androgen independent (mCRPC), or de-differentiates into neuroendocrine phenotype (NEPC). PSCA has been proposed to associate with prostate carcinogenesis⁴ and positively correlates with the NEPC marker, neuron-specific enolase (NSE).⁴⁰ Targeting PSCA may be viable for combating all-stage PCa, including advanced tumors such as mCRPC and NEPC. Our work highlights a fully human, highly specific (as tested by the MPA assay), murine PSCA cross-reactive antibody identified by phage screening and improved by structure-based affinity maturation.

Our PSCA antibody G7/F12 shares an overlapping epitope as the mouse hybridoma-derived antibody 7F5,⁹ as revealed by the competitive flow cytometry (Figure 1(c-f) and Figure S2). 7F5 exhibits robust tumor killing activity as a CAR-T⁹ and a bispecific T cell engager.⁴¹ In addition, 7F5 has been used as immunoPET tracer scaffolds.⁴² These applications demonstrate the theranostic potentials of this type of PSCA epitope. Although IgG1 F12 mediated specific ADCC killing of PSCA-expressing prostate cancer cells in cell culture, we did not observe its inhibition of growth of PC-3-PSCA xenografts in the ADCC-competent mouse model (Balb/c SCID) (Figure S5). Previous PSCA antibodies, 1G8,⁴³ 3C5,⁴⁴ 8D11¹⁰ and AGS-PSCA,⁴⁵ as unconjugated IgGs, can prevent tumor formation or inhibit growth of s.c. or orthotopically established tumors in xenograft mouse models, either by intervening PSCA signaling pathways or by mobilizing ADCC. As a GPI-anchored plasma surface protein, PSCA is proposed to cluster in detergent-insoluble glycolipid-enriched microdomains (DIGS) of the cell surface and play a critical role in signal transduction.⁶ It has been shown that PSCA regulates IL-6 expression through p38/NF-κB signaling in PCa.⁴⁶ The 1G8 antibody has been demonstrated to induce PSCA+ cancer cell death independent of the Fc effector function, acting through a F(ab’)₂-mediated PSCA cross-linking event either deregulating or initiating the PSCA-related signaling pathway.⁴⁴ By contrast, our unconjugated antibody F12 cannot kill PC-3-PSCA cancer cells both in vitro and in vivo. It is not known if this is related to the different epitope of F12 (Figure 2(e), amino acids 63–69) compared to 1G8 (amino acids 46–85).⁶

The GPI-anchored cell surface protein can internalize after binding with ligand or antibody.⁴⁷ Thus, it is not surprising that our PSCA antibody shows efficient internalization, which lays the groundwork for the development of ADCs that rely on endocytosis to deliver cytotoxic molecules into the cytosol. Currently, numerous ADCs are being tested in clinic trails for prostate cancers, such as ADCs targeting prostate-specific membrane antigen (PSMA), trophoblast cell surface antigen-2 (TROP-2), six-transmembrane epithelial antigen of prostate-1 (STEAP-1), tissue factor (TF), and delta-like protein 3 (DLL-3).⁴⁸ However, for PSCA, there is only one mouse hybridoma derived (8D11) ADC-DM1 that has been evaluated in xenograft mouse models.¹⁰ In the current work, we generated a first-in-class fully human PSCA ADC-MMAE via the random conjugation method by the amine-NHS coupling protocol, which outputs a heterogenous DAR (average DAR = 1.98, Figure 4(b)). This heterogeneity may be associated with the relatively fast elimination of IgG1 F12-MMAE ADC (Figure 5(a, b)), which can be optimized by the site-specific conjugation yielding homogenous ADCs with an improved therapeutic index. We used the clinically validated valine-citrulline linker, which has demonstrated good stability in human serum and is used in many approved ADCs. The VC-PAB linker can be cleaved by cathepsin B after ADC endocytosis into the lysosome to release the PAB spacer and MMAE moiety.⁴⁹ After self-immolation, the free MMAE molecule is released, leading to cytotoxicity. For the payload, we utilized the anti-mitotic regent MMAE, which inhibits microtubule polymerization and hence neoplastic cell proliferation with high potency.⁵⁰ MMAE is the cytotoxic drug included in FDA-approved ADCs Adcetris®(brentuximab vedotin), Polivy®(polatuzumab vedotin), Padcev®(enfortumab vedotin), Tivdak®(tisotumab vedotin).⁵¹ However, it should be noted that MMAE may lead to dose-limiting toxicities such as neutropenia and peripheral neuropathy.⁵² Other payloads such as DNA cross-linking or alkylation reagents can be used to generate IgG1 F12 ADCs. Indeed, we have conjugated pyrrolobenzodiazepine (PBD) into IgG1 F12, which exhibits potent PSCA-dependent cytotoxicity (IC₅₀ = 2.18 nM, Figure S6A and B).

As a tumor-associated antigen, PSCA is typically over-expressed on PCa, but it is also expressed in epithelial cells of normal organs from the gastrointestinal tract and urogenital system such as prostate, urinary bladder, kidney, skin, esophagus, stomach, and placenta (Figure S7).^1,2,10 This raises potential issues of on-target, off-tumor toxicity of PSCA-targeting immunotherapeutics, which can be minimized or circumvented by optimization of antibody affinity, binding epitope, format, valency and dosing regimens.^53,54 The Phase 1/2A clinical trial of the fully human PSCA antibody, AGS-PSCA, did not show dose-limiting toxicity (DLT) when dosed at 40 mg/kg followed by 20 mg/kg infusions every 3 weeks.¹² The 1G8 (clone A11)-based PSCA-targeting CAR-Ts did not display DLT in the Phase 1 clinical trial for mCRPC after optimization of CAR-T and lymphodepletion doses (100 M plus 300 mg/m² cyclophosphamide).¹³ Although F12 was not tested for reactivity to normal human tissues, it cross-reacts with the murine PSCA orthologue. Thus, tumor-free mouse was used for modeling on-target off-tumor toxicity. Results showed that healthy mice tolerated F12-MMAE at the high dose of 20 mg/kg (Figure 5(d,e)).

The high dose of isotype control Ab1-MMAE demonstrated a notable effect in slowing PC-3-PSCA tumor growth (Figure 6 and S3), which suggests potential non-specificity and risk of off-target toxicity for ADC F12-MMAE. However, F12-MMAE exhibits significantly higher efficacy compared to the isotype Ab1-MMAE. The dose-dependent nature of the non-specificity indicates an opportunity to optimize F12-MMAE dosage to maximize its therapeutic index. In our study, we detected the free MMAE shed from our ADC by LC-MS/MS after incubating F12-MMAE with mouse serum (Figure S4). This shedding is likely due to the cleavage of the amide bond between citrulline and PABC by the serine protease carboxylesterase 1C in mouse plasma.³⁵ Thus, we suspect that the premature payload deconjugation, resulting from the susceptibility of the ADC dipeptidyl linker to extracellular (tissue or circulation) proteases, may contribute to the nonspecific anti-tumor activity for ADC.⁵⁵ However, our ADC PK and mass spectrometry data indicated that the extent of F12-MMAE deconjugation in the mouse serum is limited (Figure 5(c) and S4). Therefore, other factors such as the ADC off-target catabolism occurring in normal cells might also contribute to the non-specificity of the ADC.³⁶ ADCs can be internalized into normal cells through either FcγR/other receptors-mediated endocytosis or receptor-independent pathways such as macro/micropinocytosis.³⁶ These internalized ADCs are then catabolized by intracellular proteases, resulting in the release of MMAE. The free MMAE can subsequently diffuse into circulation and tumor sites, leading to tumor inhibition. Based on this knowledge, the therapeutic window of F12-MMAE can be improved through reducing ADC susceptibility to extracellular proteases by adopting more stable peptidyl linkers, such as glutamic acid – valine – citrulline or cBuCit linkers^55,56 or utilizing non-cleavable linkers (e.g., maleimidocaproyl).⁵⁷ It has been reported that trastuzumab-emtansine (T-DM1) can be internalized into megakaryocytes (MK) from human hematopoietic stem cells in a HER2-independent, FcγRIIa-dependent manner, which results in intracellular release of DM1 and associates with thrombocytopenia.⁵⁸ Thus, one feasible way to reduce the FcγR-mediated nonspecific catabolism of our F12-ADC is to engineer out FcγR binding via introducing “silent” mutations (LALAPG) into Fc. To reduce the receptor-independent nonspecific pinocytosis, we may need to engineer the physicochemical properties of F12-ADC, including its hydrophobicity and surface charge, by changing DARs, payloads and/or charged linkers.^36,59

It should be noted that the composition, distribution, and function of extracellular proteases may differ between mice and humans. For instance, unlike its mouse orthologue, the carboxylesterase in monkey and human tissue and sera cannot hydrolyze the vc-PABC linker.^31,35 Consequently, the deconjugation and non-specificity profiles of our ADC in monkeys and humans remain to be evaluated. On the other hand, although we have not observed obvious toxicity in F12-MMAE-treated mouse at tested doses, mice are notoriously less sensitive to MMAE than rats, monkeys and humans.⁶⁰ This limits the translatability of ADC tolerability studies in mice to clinical settings. Therefore, clinically relevant toxicological studies for F12-MMAE are warranted in rats and non-human primates.

In conclusion, we identified a fully human, highly specific, murine orthologue cross-reactive PSCA antibody, which when conjugated with MMAE shows dose dependent anti-tumor efficacy and specificity in a human prostate cancer xenograft mouse model. This ADC warrants further characterization and optimization in preclinical and clinical investigations. In addition to ADCs, this antibody can be developed into other therapeutic modalities such as radionuclide conjugates, CAR-T and T-cell engagers with potential theranostic applications in a variety of PSCA-overexpressing solid tumors.

Funding Statement

The work was supported by the UPMC Enterprises.

Disclosure statement

W.L., X.J.C., D.S.B., J.W.M., and D.S.D. are co-inventors of a patent, filed on May 2023 by the University of Pittsburgh, related to PSCA antibodies G7 and F12 described in this paper.escribed in this paper.

Author contributions

W.L., D.S.D., and J.W.M. conceived and designed the research. W.L. identified and extensively characterized antibodies. X.J.C. tested ADC in vivo efficacy in xenograft mouse models and FACs assay. S.S. prepared the ADC and performed cell killing assays. D.S.B. made the fully human Fab phage library. L.Y.Z., A.C., M.S., C.A., M.H. expressed and purified antibodies and characterized antibodies using methods such as ELISA, FACS, and BLItz. Y.J.K. helped prepare the IACUC protocol. X.L.L. produced the recombinant PSCA antigen. C.C., Z.H.S., and D.V.Z. contributed to PSCA and antibody subcloning, MPA assays, peptide scanning, ADC preparation and ADC characterization by mass spectrum. W.L. and X.J.C. wrote the first draft of the article. All authors discussed the results, edited, and contributed to the manuscript.

Abbreviations

ADCs

Antibody-drug conjugates

ADCC

Antibody-Dependent Cellular Cytotoxicity

ALT

alanine transaminase

BPH

benign prostatic hyperplasia

BiTEs

bispecific T cell engagers

BSA

bovine serum albumin

BLItz

Biolayer interferometry

CAR-T

chimeric antigen receptor

CDR

complementarity-determining region

CR

complete remission

DAR

drug antibody ratio

DPBS

Dulbecco’s PBS

DIGS

detergent-insoluble glycolipid-enriched microdomains

DLS

Dynamic Light Scattering

DLL-3

delta-like protein 3

DLT

dose-limiting toxicity

EDTA

ethylenediaminetetraacetic acid

ELISA

enzyme-linked immunosorbent assay

FACs

Flow Cytometry Analysis

GPI

glycosylphosphatidylinositol

HRP

horseradish peroxidase

HC

Heavy chain

IPTG

isopropyl β-D-1-thiogalactopyranoside

i.p.

intraperitoneally

LC

Light chain

MMAE

monomethyl auristatin E

MPA

membrane proteome array

mCRPC

Metastatic castration resistant prostate cancer

NK

Natural killer

NEPC

neuroendocrine phenotype

NSE

neuron-specific enolase

NHS

lysine amine-N-hydroxysuccinimide

PSCA

Prostate stem cell antigen

PCa

prostate cancers

PE

Phycoerythrin

PBS

phosphate-buffered saline

PPG

propylene glycol

PAB

para-aminobenzylcarbamate.

PK

pharmacokinetic

PR

partial remission

PSMA

prostate-specific membrane antigen

PBD

pyrrolobenzodiazepine

RT

room temperature

SEM

soluble expression monoclonal

SD

standard deviation

SEC

Size-Exclusion Chromatography

ScFv

single-chain variable fragment

s.c.

subcutaneously

STEAP-1

six-transmembrane epithelial antigen of prostate-1

TME

tumor microenvironment

TROP-2

trophoblast cell surface antigen-2

TF

tissue factor

vc-MMAE

NHS-Glutarate-Val-Cit-PAB-MMAE

Data and code availability

Antibody amino acid has been disclosed in Table 1. The antibody is only allowed for noncommercial use. All data supporting the findings of this study are available within the paper and are available from the corresponding author upon request.

Materials availability

All requests for resources and reagents, including antibodies, viruses, plasmids, and proteins, should be directed to, and will be fulfilled by, the corresponding author. All reagents will be made available on request after completion of a Material Transfer Agreement.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/19420862.2024.2387240