Generation and characterization of novel high-affinity fully human antibodies targeting prostate-specific membrane antigen and prostatic acid phosphatase

Abstract

Background

Prostate-specific membrane antigen (PSMA)-targeted therapies have revolutionized the management of patients with prostate cancer. However, not all patients benefit from such agents. Prostatic acid phosphatase (ACP3), a tyrosine phosphatase produced by prostate epithelial cells, has emerged as a promising alternative target for both early and advanced disease.

Methods

Here, we report on the generation and characterization of novel, fully human monoclonal antibodies targeting PSMA and ACP3. Antigen-specific clones were isolated from human antibody phage display libraries and characterized in vitro and in vivo.

Results

Selected clones bound with high specificity to the cognate target and demonstrated excellent tumor-targeting performance in biodistribution studies.

Conclusions

The lead candidates, termed H92A5 (anti-PSMA) and EKA4 (anti-ACP3), could serve as promising scaffolds for the development of next-generation tumor-targeted therapeutics for prostate cancer.

Statement of Significance

This project developed new fully human antibodies targeting prostate-specific membrane antigen and prostatic acid phosphatase, two key antigens in prostate cancer. These antibodies showed strong and specific tumor binding in animal models. They represent potential tools for improving prostate cancer treatment.

Introduction

Prostate cancer is the second most common cancer in men and accounts for more than 35 000 deaths every year in the United States alone [1]. When the cancer is confined to the prostate gland, conventional treatments such as radiation therapy, radical prostatectomy, and hormone deprivation therapy often lead to long-term survival outcomes. However, prognosis worsens significantly in advanced-stage disease, with a 5-year survival rate of only 36.6% [1]. When tumor growth is dependent upon circulating testosterone levels, the first-line treatment is androgen deprivation therapy (ADT) [2]. Eventually, most patients develop resistance and experience disease progression, leading to metastatic castration-resistant prostate cancer (mCRPC). Current treatment options include androgen receptor signal inhibitors, docetaxel, external radiation therapy, radiopharmaceuticals, and immune checkpoint inhibitors, but all have their limitations [3]. Innovative and more selective therapies are therefore needed for the management of patients with mCRPC.

One promising strategy for the development of selective cancer therapies relies on ligand-mediated targeted delivery of bioactive payloads at the tumor site [4–7]. This approach exploits tumor-associated antigens that are typically overexpressed at the site of disease, while being undetectable in most healthy organs.

Prostate-specific membrane antigen (PSMA) is the most extensively validated cell surface antigen for drug delivery applications in prostate cancer [3]. Immunohistochemical analyses indicate that PSMA is overexpressed in 88% of prostatic cancers. However, low level expression of the antigen is also observed in healthy tissues such as the kidneys, salivary glands, and intestine [8]. The only PSMA-targeted therapy currently approved is ¹⁷⁷Lu-PSMA-617 (Pluvicto™), a small molecule radiopharmaceutical that has demonstrated a significant improvement in terms of imaging-based progression-free survival in pivotal clinical trials [9–11]. Despite its efficacy, Pluvicto™, as well as other PSMA-specific small molecule radioligands, exhibit on-target, off-tumor uptake leading to dose-limiting toxicities. Common adverse events include xerostomia (dry mouth), gastrointestinal, and kidney toxicities, all associated with PSMA expression in the aforementioned non-malignant tissues.¹² Notably, these side effects have been reported more frequently for small molecule agents than for antibody-based radiopharmaceuticals [12]. Several PSMA-targeting biologics, such as radiolabeled antibodies, antibody-drug conjugates (ADCs) and bispecific antibodies, have been developed, demonstrating favorable safety and promising anti-tumor activity in preclinical models [13–15]. Nevertheless, the clinical potential of such biopharmaceuticals remains under investigation.

Prostatic acid phosphatase (PAP or ACP3), a tyrosine phosphatase, is a promising target for metastatic prostate cancer. Immunohistochemical analyses of patient samples indicate that ACP3 is overexpressed in over 98% of prostate cancer lesions, while exhibiting minimal or no presence in most healthy organs, including the salivary glands, intestine and kidneys, with the exception of the prostate [16–18]. A study from 1989 introduced the monoclonal anti-PAP antibody PASE/4LJ and demonstrated its potential for diagnostic applications in prostate cancer. PASE/4LJ effectively stained primary and metastatic prostatic carcinomas, and non-neoplastic adult and fetal prostatic glands [19]. Additionally, the antibody did not inhibit the enzyme activity of PAP, indicating its suitability for diagnostic purposes without interfering with enzyme function [19]. The only ACP3-targeted therapy approved so far is Sipuleucel-T (Provenge™), an autologous cellular immunotherapy agent. The product stimulates the patient’s immune system to recognize and attack ACP3-positive prostate cancer cells [20]. However, due to its modest clinical benefit, its use has remained limited [3, 20]. The expression pattern of ACP3 underscores the potential of targeting such an antigen for pharmacodelivery applications in prostate cancer.

Phage display is a powerful tool for isolating ligands across diverse targets [21, 22]. By displaying peptides or antibodies on the surface of bacteriophages, this technique enables the rapid screening of billions of clones for drug discovery and development. Clinically validated antibodies isolated by phage display, including Tecentriq™ (Atezolizumab), Bavencio® (Avelumab), Cyramza® (Ramucirumab), and Portrazza™ (Necitumumab), have received marketing authorization for the treatment of various cancers [22]. These successes underscore the potential of phage display for generating innovative biologics [23].

In this work, we report the discovery of fully human, high-affinity antibodies targeting PSMA and ACP3, two clinically relevant antigens in prostate cancer. PSMA-specific clones were obtained through two rounds of affinity maturation using single-chain variable fragment (scFv) and Fab-based phage display antibody libraries. ACP3-specific clones were isolated directly from a naïve scFv library. The selected anti-PSMA (termed H92A5) and anti-ACP3 (termed EKA4) antibodies demonstrated high specificity to their respective antigens in vitro, showed antigen-dependent internalization in target cells, and exhibited selective tumor accumulation in vivo. These findings pave the way for the development of novel antibodies serving as delivery vehicles for the treatment of prostate cancer.

Materials and methods

Cell lines

Human fibrosarcoma HT1080 (RRID: CVCL_0317, provider: ATCC), Chinese hamster ovary (CHO, RRID: CVCL_0213, provider: ATCC), and human prostate adenocarcinoma LNCaP cells (RRID: CVCL_0395, provider: ATCC) were grown according to the supplier’s protocol, kept in culture for no longer than 16 passages, and stored as cryopreserved aliquots in liquid nitrogen. Authentication of the cell line was performed by the cell bank before shipment. It included the evaluation of post-freeze viability, growth properties, morphology, testing of mycoplasma contamination, and an isoenzyme assay.

HT1080 cells were transduced with a lentiviral vector to stably integrate and express human PSMA or human ACP3 following a previously described procedure [24, 25].

All HT1080-derived cell lines were grown in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% Antibiotic-Antimycotic (Gibco, 15240062) at 37°C, 5% CO₂. LNCaP cells were grown in Roswell Park Memorial Institute Medium (RPMI-1640) supplemented with 10% FBS and 1% Antibiotic-Antimycotic at 37°C, 5% CO₂. Adherent cells were detached using trypsin–EDTA 0.25% and passaged once they reached 70% confluency on the flasks.

CHO cells were grown in PowerCHO supplemented with 4 mM L-glutamine and kept in culture at 37°C, 5% CO₂, 120 rpm at a cell density between 1 million cells/ml and 10 million cells/ml. For protein production, CHO cells were cultured in ProCHO medium supplemented with 4 mM L-glutamine and kept in culture at 31°C, 5% CO₂, 120 rpm for 6 days, or until cell viability dropped below 60%.

Cloning, expression and characterization of human PSMA and human ACP3

The synthetic cDNA encoding human PSMA (UniProtKB Q04609, position 44–750), flanked with an N-terminal 6-Histidine tag and an N-terminal AviTag® (GLNDIFEAQKIEWHE) was cloned into the mammalian cell expression vector pcDNA3.1(+). The synthetic cDNA encoding human ACP3 (UniProtKB P15309, position 33–386), flanked with a C-terminal AviTag® (GLNDIFEAQKIEWHE), and a C-terminal 6-Histidine tag was cloned into the pcDNA3.1(+) vector.

Both recombinant antigens were expressed by polyethyleneimine (PEI)-induced transient gene expression (TGE) in CHO cells and purified by affinity chromatography using Ni-NTA agarose resin (Roche). Recombinant PSMA was dialyzed against HEPES buffer (50 mM HEPES,1 mM CaCl2, 0.5 mM MgCl2, and 100 mM NaCl, pH 7.4), while recombinant ACP3 was dialyzed against phosphate-buffered saline (PBS, pH 7.4).

Purified proteins were analyzed by size exclusion chromatography (SEC) on a Superdex 200 Increase 10/300 GL column using an ÄKTA Pure FPLC (GE Healthcare). Sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) was performed using 4%–12% gels (Invitrogen™, NP0322BOX) in 1x MOPS buffer under reducing and non-reducing conditions.

Selection of antibodies by phage display and affinity maturation

The AviTag® domain enabled site-specific biotin labeling using BirA-maltose binding protein [26]. Successful attachment of a biotin group to the antigens was confirmed through a band shift assay and enzyme-linked immunosorbent assay (ELISA). Biotin-labeled antigens were immobilized on Dynabeads™ Streptavidin M280 (Invitrogen, 2865740). Antibodies specific to the respective antigen were enriched by two rounds of biopanning from synthetic fully human scFv antibody libraries, as previously described by Viti et al [27]. Individual bacterial colonies infected with isolated phage were cultured, and scFv expression was induced by isopropyl β-D-1-thiogalactopyranoside (IPTG, 0.1 M). Bacterial supernatants containing antibody fragments were collected and screened by ELISA using SA-wells coated with the respective biotinylated antigen, and by flow cytometry on HT1080 cells expressing the corresponding antigen. Positive clones resulting from both ELISA and flow cytometry screening assays were analyzed by sequencing. Unique candidate clones were cloned into pCDNA3.1(+) as scFv and IgG formats, produced in CHO cells by TGE and purified by affinity chromatography. Binding to the cognate antigen was confirmed by surface plasmon resonance (SPR), flow cytometry analysis, and immunofluorescence on tumor sections from mice injected with HT1080 cells expressing the corresponding antigen.

For PSMA, the affinity of the best clone was improved by two rounds of affinity maturation. In the first round, new phage libraries were cloned by random mutagenesis of residues in the CDR1 loops of the VH and VL domains and were subjected to one round of biopanning following the protocol described by Villa et al [28]. For the second round of affinity maturation, new phage libraries were cloned by random mutagenesis of residues in the CDR2 loops of the VH domain and subjected to another round of biopanning.

Cloning, expression, and characterization of anti-PSMA and anti-ACP3 antibodies

All the antibodies described in this study were cloned into the mammalian cell expression vector pcDNA3.1(+). The DNA sequence of the J591 antibody (used as the anti-PSMA positive control) was derived from Bander et al [29]. Antibodies were produced by PEI-induced TGE in CHO cells, followed by purification by affinity chromatography using Protein A Sepharose (Sino Biological). Purified proteins were dialyzed against PBS, pH 7.4 and analyzed by SEC on a Superdex 200 or Superdex 75 Increase 10/300 GL column using an ÄKTA Pure FPLC (GE Healthcare). SDS-PAGE was performed using 4–12% gels (Invitrogen™, NP0322BOX) in 1x MOPS buffer under reducing and non-reducing conditions.

Surface plasmon resonance

For K_D determination, monomeric preparations of scFv were used and biotin-labeled antigens were immobilized on an SA-sensor chip with a density of 1500 RUs using a BIAcore X100 system (Cytiva). To assess the binding of IgG₄ format antibodies, recombinant antigens were immobilized on a CM5-sensor chip with the same density (1500 RUs).

Real-time binding interactions were tested with serial dilutions of the antibodies at a flow rate of 10 μl/min. New chips were used for each run, as regeneration of the chip was not possible. Analysis of the binding kinetics was performed on the BIAevaluation3.2 software.

Flow cytometry

HT1080-derived cell lines were detached from the cell culture flask using Accutase. Cells were blocked for unspecific binding and stained with the anti-PSMA (or anti-ACP3) antibody in IgG₄ format. KSF IgG₄, binding to the hen egg lysozyme, served as a negative isotype control. Antibody binding was detected through an Alexa Fluor488-conjugated goat anti-human IgG (H + L) antibody (Invitrogen™, A11013). Live/dead cells were detected using Zombie-NIR diluted in PBS. All staining and washing steps were carried out in cold flow cytometry buffer (0.5% BSA, 2 mM EDTA in PBS; pH 7.4). Data was acquired on a CytoFLEX cytometer (Beckman Coulter) and the CytExpert software and processed with the FlowJo 10.4 software.

Immunofluorescence on a human tissue microarray

Human tissue microarrays (Amsbio, T6235700; T1234201; T1235201; CS511031) were used for immunofluorescence analysis of the different antibodies. Sections were fixed in ice-cold acetone and stained with 50 nM of the anti-PSMA (or anti-ACP3) antibodies in IgG_2a format. KSF IgG_2a, binding to an irrelevant target, served as a negative isotype control. For antibody detection, a goat anti-mouse IgG Alexa Fluor488 antibody (Invitrogen™, A11029) was used. For cell nuclei visualization, DAPI (Invitrogen™, D1306) was used. Slides were mounted with a fluorescence mounting medium (Dako Agilent) and analyzed with a Leika DMI6000B microscope (Leica Microsystems).

Experimental animals

A total of 30 male BALB/c nude mice, aged 10–12 weeks were used in this work. Mice were purchased from Janvier (Route du Genest, 53940 Le Genest-Saint-Isle, France) and raised in a pathogen-free environment with a relative humidity of 40%–60%, at a temperature between 18°C and 26°C and with daily cycles of 12 h light/darkness according to guidelines (GV-SOLAS; FELASA). Animals were kept in groups of 5 or fewer mice per cage and were reallocated in case of single housing to another cage. Blinding of the experimental groups was not performed, experimental groups were allocated according to their tumor volume, measured by caliper and calculated as follows: (length [mm] × width [mm]²)/2.

Tumor implantation

The human fibrosarcoma HT1080 cell line was used in its wild-type form and transduced with PSMA or ACP3. For these models, 5 × 10⁶ cells were implanted subcutaneously in the flank of male BALC/c nude mice aged 10–12 weeks. Tumor growth and body weight change were monitored daily, mice were euthanized if the tumor diameter reached >15 mm or tumors developed ulcerations.

Immunofluorescence-based biodistribution

Eight-week-old male immunodeficient BALB/c nude mice were obtained from Janvier. Five million HT1080-derived cells (HT1080. hPSMA, HT1080.hACP3) were injected subcutaneously into the right flank. Once tumors were grown to reach a volume of approximately 150 mm³ (volume = length × width²/2), 200 μg of antibodies in IgG₄ format were injected into the lateral tail vein. Mice were sacrificed 24 hours post-injection and the organs and tumors were excised and placed in a cryo-embedding medium. Cryostat sections (10 μm) of all organs and tumors were stained and the injected antibodies were detected with a goat anti-human IgG AlexaFluor488 antibody (Invitrogen™, A11013). Tissue vasculature was visualized through a rat anti-mouse CD31 antibody (BD Biosciences, 550274) followed by a donkey anti-rat IgG Alexa Fluor594-labeled secondary antibody (Invitrogen™, A21209).

Internalization studies via confocal microscopy

Proteins were FITC labeled according to the manufacturer protocol (Sigma). HT1080.hPSMA and HT1080.hACP3 cells were seeded into 4-well coverslip chamber plates (Sarstedt, Inc.) at 10⁴ cells per well in DMEM medium (1 mL) supplemented with 10% FBS, 1% AA, and HEPES (10 mM) and allowed to grow for 24 hours under standard culture conditions. The culture medium was replaced with a fresh medium containing fluorescein-labeled antibodies (100 nM). Hoechst 33342 nuclear dye (Invitrogen™, H1399) was used to stain nuclear structures. Randomly selected colonies were imaged 6 hours after incubation on a FLUOVIEW FV4000 Confocal Laser Scanning Microscope (Evident).

Statistical analysis

Data were analyzed on Prism V.9.0 (GraphPad Software).

Results

Isolation and in vitro characterization of novel, fully human monoclonal antibodies targeting PSMA

To perform phage display selections of antibody libraries, recombinant human PSMA was expressed in Chinese Hamster Ovary (CHO) cells and purified to homogeneity by affinity chromatography. An AviTag® was introduced at the N-Terminus of the recombinant antigen for site-specific biotinylation (Fig. S1a). SEC and SDS-PAGE confirmed that the antigen was pure and of the expected size (Fig. S1b-c). A band shift and an enzymatic assay demonstrated successful biotinylation and substrate cleavage activity, respectively (Fig. S1d-e).

Selections against PSMA were performed using synthetic naïve single-chain variable fragment (scFv) antibody libraries and resulted in the isolation of a first monoclonal antibody, named EE11 (Table 1, Fig. S2). To improve binding kinetics, an affinity maturation library was generated. Sequence variability was restricted to the complementarity-determining region (CDR) 1 of the variable heavy (V_H) chain of the EE11 parental antibody. The first affinity maturation campaign yielded a new clone, termed H9 (Table 1, Fig. S3).

Table 1.

Amino acid sequences of selected anti-PSMA and anti-ACP3 antibodies at relevant positions. Reported are the sequences given in single-letter amino acid code for the complementarity determining regions (CDRs) on the V_H and V_L of the antibodies.

Antigen	Clone	Library Type	V H  chain			V L  chain
			CDR1	CDR2	CDR3	CDR1	CDR2	CDR3
PSMA	EE11	Naïve	GFTFSSYAMS	AISGSGGSTYYADSVKG	SRWVFDY	RASQSVSSSYLA	GASSRAT	QQSMGLPAT
	H9	Aff.Mat	GFTFSPFSMS	AISGSGGSTYYADSVKG	SRWVFDY	RASQSVSSSYLA	GASSRAT	QQSMGLPAT
	H92A5	Aff.Mat	GFTFSPFSMS	AISDTGGTTYYADSVKG	SRWVFDY	RASQSVSSSYLA	GASSRAT	QQSMGLPAT
ACP3	EKA4	Naïve	GFTFSSYAMS	AVSGSGGSTYYADSVKG	YMPWFDY	RASQSVSSSYLA	GASSRAT	QQPGLRPST
	D9A6	Naïve	GFTFSSYAMS	AISGSGGSTYYADSVKG	LHRRLPH	RASQSISSYLNW	AASSLQS	QQPIGFPHT

Our group primarily focuses on the scFv antibody format to construct phage display libraries. However, scFvs tend to form dimers or aggregates in a clone-dependent and unpredictable manner, which may lead to candidate selection based on avidity rather than affinity [30]. To address this issue, we decided to build a new affinity maturation library based on the Fab format and introduced sequence variability in the CDR2 loops of the H9 antibody heavy chain. The final selected candidate was named H92A5 (Table 1, Fig. S4). The three clones were produced in mammalian cells in the scFv format and affinity measurements were performed on monomeric fractions. The results showed improved dissociation constants (K_D) for EE11 (parental clone), H9 (affinity maturation on CDR1) and H92A5 (affinity maturation on CDR2) with values of 60 nM, 20 nM and 16 nM, respectively (Fig. 1, Fig. S5).

Figure 1.

Schematic representation and SPR characterization of selected anti-PSMA antibodies in ScFv format. (top) Schematic representation of selected antibodies in ScFv. (bottom) SPR sensorgrams of EE11, H9, and H92A5 in ScFv.

The three antibodies were sub-cloned into IgG₄ format, expressed in CHO cells, and characterized by analytical methods, demonstrating homogeneity in SEC and SDS-PAGE (Fig. 2a). Immunofluorescence studies on HT1080.hPSMA tumor sections confirmed that all antibodies bound to a conformational epitope (Fig. 2b). PSMA expression patterns were comparable across all stainings, including the positive control anti-PSMA J591 [29]. No signal was detected with the negative control antibody (KSF), specific to hen-egg lysozyme (Fig. 2b).

Figure 2.

In vitro characterization of anti-PSMA antibodies. (a) SEC profiles and SDS-PAGE gel analysis. (b) Immunofluorescence staining on HT1080.hPSMA tumor sections. PSMA antibodies were used to identify PSMA-positive cells, an anti-CD31 antibody was used to identify blood vessels, and a nuclear stain (DAPI) was used to visualize cell nuclei. 20× magnification, scale bars = 100 μm. (c) Flow cytometry analysis with IgG₄ antibodies of HT1080.hPSMA cells. (d) Antigen specificity ELISA using H92A5 in IgG₄.

Flow cytometry analysis of HT1080.hPSMA cells showed similar binding curves for all antibodies, including for the positive control J591. No binding was detected with the KSF antibody (Fig. 2c).

The binding specificity of H92A5 for PSMA was confirmed via an ELISA. No binding to other tumor-associated antigens was observed (Fig. 2d).

Isolation and in vitro characterization of novel fully human monoclonal antibodies targeting prostatic acid phosphatase (ACP3)

To perform phage display selections of antibodies, recombinant human ACP3 was expressed in CHO cells and purified to homogeneity by affinity chromatography. An AviTag® was introduced at the C-Terminus of the recombinant antigen for site-specific biotinylation (Fig. S6a). SEC and SDS-PAGE confirmed that the antigen was pure and of the expected size (Fig. S6b-c). A band shift assay confirmed successful biotinylation (Fig. S6d).

Phage display scFv libraries were used to isolate novel ACP3-specific antibodies. Since over 300 anti-ACP3 clones were positive in ELISA (Fig. S7), we decided to perform a subsequent flow cytometry screening of the scFv supernatants on HT1080.hACP3 cells. Clones with the highest fluorescent intensity values were sequenced and 10 unique clones were identified. EKA4 and D9A6 were ultimately selected as the final candidates (Table 1, Fig. S8). Affinity measurements were performed on monomeric scFv fractions, revealing dissociation constants (K_D) for EKA4 and D9A6 of 5 nM and 10 nM (Fig. 3, Fig. S9).

Figure 3.

Schematic representation and SPR characterization of selected anti-ACP3 antibodies in ScFv format. (top) Schematic representation of selected antibodies in ScFv. (bottom) SPR sensorgrams of EKA4 and D9A6 in ScFv.

EKA4 and D9A6 antibodies were subsequently cloned and expressed as IgG₄ in CHO cells and characterized in vitro. SEC and SDS-PAGE analyses showed homogeneity and correct molecular weight of both EKA4 and D9A6 (Fig. 4a).

Figure 4.

In vitro characterization of anti-ACP3 antibodies. (a) SEC profiles and SDS-PAGE gel analysis. (b) Immunofluorescence staining on HT1080.hACP3. ACP3 antibodies were used to identify ACP3-positive cells, an anti-CD31 antibody was used to identify blood vessels, and a nuclear stain (DAPI) was used to visualize cell nuclei. 20× magnification, scale bars = 100 μm. (c) Flow cytometry analysis with IgG₄ antibodies on HT1080.hACP3 cells. (d) Antigen specificity ELISA using EKA4 in IgG₄.

Immunofluorescence studies with EKA4 on HT1080.hACP3 sections showed stronger staining for EKA4 than for D9A6. No signal was observed with the negative control antibody KSF (Fig. 4b).

Flow cytometry analysis of HT1080.hACP3 cells confirmed binding for both EKA4 and D9A6 with EC₅₀ values in the low nanomolar range, while KSF showed no binding (Fig. 4c). An ELISA experiment confirmed the ability of EKA4 to bind specifically to ACP3 (Fig. 4d), while no signal was observed on other tumor-associated antigens.

Binding to LNCaP cells

Flow cytometry analysis was performed to assess the binding of H92A5 (anti-PSMA) and EKA4 (anti-ACP3) to LNCaP cells, naturally expressing PSMA [31]. The experiment confirmed binding to PSMA (using H92A5), but no signal was observed with the anti-ACP3 antibody EKA4 (Fig. S10). No signal was observed with the negative control antibody (KSF).

Antibody internalization studies

Confocal microscopy was performed to assess the internalization of EKA4 (anti-ACP3) and H92A5 (anti-PSMA). Fluorescein-labeled antibodies were incubated for 6 hours with HT1080 wild-type, HT1080.hPSMA, or HT1080.hACP3 cells. An efficient antigen-dependent internalization was observed for H92A5 in HT1080.hPSMA cells and for EKA4 in HT1080.hACP3 cells. No internalization was detected in HT1080 wt cells or when EKA4 was incubated with HT1080.hPSMA or H92A5 with HT1080.hACP3 (Fig. S11).

Immunofluorescence-based biodistribution studies

The tumor-homing properties of the EE11 (parental clone), H9 (affinity maturation on CDR1), H92A5 (affinity maturation on CDR2), and J591 (positive control) anti-PSMA antibodies were evaluated through an in vivo immunofluorescence-based biodistribution analysis. HT1080.hPSMA tumor-bearing mice were injected intravenously with the corresponding antibodies. Tissues were analyzed 24 hours after administration (Fig. 5a). KSF was used as a negative control. No signal was detected in healthy organs for any of the antibodies. By contrast, an increased tumor uptake correlated with an increase in the affinity for the anti-PSMA antibody clones. The positive control J591 and H92A5 showed a comparable signal in the tumor, while the EE11 parental clone revealed an almost undetectable binding to the tumor. These findings suggest that high affinity is required to achieve strong and homogenous tumor uptake (Figs. 5a, S12).

Figure 5.

Ex vivo immunofluorescence-based biodistribution studies. Biodistribution properties of (a) anti-PSMA antibodies (IgG₄) on HT1080.hPSMA tumor-bearing mice and (b) anti-ACP3 antibodies (IgG₄) on HT1080.hACP3 tumor-bearing mice. Analysis was performed 24 hours after intravenous administration. H92A5 and EKA4 antibodies were used to identify antigen-positive cells and a nuclear stain (DAPI) was used to visualize cell nuclei. 20× magnification, scale bars = 100 μm.

Following the same experimental procedure, the tumor-homing properties of the anti-ACP3 antibodies, EKA4 and D9A6 were evaluated in HT1080.hACP3 tumor-bearing mice (Fig. 5b). No signal was detected in healthy organs for both antibodies. EKA4 showed a stronger tumor uptake compared to D9A6, in line with the affinity of the two ligands (Fig. 5b, S12). KSF was used as a negative control.

Human tissue microarray analysis of H92A5 and EKA4

H92A5 (anti-PSMA) and EKA4 (anti-ACP3) were further analyzed by immunofluorescence on human samples. Both antibodies stained prostate cancer and healthy prostate specimens, consistent with the expression of the antigens in these tissues. In addition, neither H92A5 nor EKA4 stained other healthy (including salivary glands) or non-prostatic cancerous specimens. No signal was detected with the negative control antibody KSF (Fig. 6, Fig. S13).

Figure 6.

Immunofluorescence studies of the H92A5 and EKA4 antibodies on human tissue specimens. Human cancer sections and corresponding healthy controls were analyzed by immunofluorescence staining with H92A5, EKA4, and an irrelevant isotype control (KSF), all in IgG_2a format. H92A5 and EKA4 antibodies were used to identify antigen-positive cells and a nuclear stain (DAPI) was used to visualize cell nuclei. 20× magnification, scale bars = 100 μm.

Discussion

Here, we describe the generation and characterization of novel fully human antibodies targeting PSMA and ACP3, two key antigens associated with prostate cancer. The H92A5 antibody, specific to PSMA, was isolated from phage display libraries after two rounds of affinity maturation by randomizing CDR1 and CDR2 loops. The EKA4 antibody, targeting ACP3, was isolated via phage display followed by high-throughput flow cytometry screening.

The in vivo targeting performance of selected clones was evaluated by immunofluorescence-based biodistribution analysis, revealing excellent tumor-homing properties for clones H92A5 and EKA4. This observation aligns with previous studies indicating that high affinity is critical for effective targeting and reduced non-specific binding [32]. Given the established role of PSMA in prostate cancer imaging and therapy, we compared our H92A5 antibody with J591, a clinical-stage antibody. Both antibodies bind with high affinity to human PSMA, but do not cross-react with the mouse homolog (Fig. S14).

Radiolabeled preparations of the J591 antibody in the IgG format have demonstrated potential for enhancing the detection of bone metastases and monitoring disease progression, particularly in patients undergoing anti-androgen treatment. Clinical studies using ¹¹¹In-J591, ⁸⁹Zr-J591 and ¹⁷⁷Lu-J591 for imaging and theranostic applications reported favorable biodistribution and organ dosimetry profiles [33–37]. Notably, J591 did not accumulate in salivary or lacrimal glands and exhibited lower uptake in the kidney compared to PSMA-targeted small molecules [37, 38]. However, these antibodies typically showed higher liver uptake compared to small molecules and were associated with dose-limiting hematologic toxicities, including neutropenia and thrombocytopenia [37, 39]. The use of anti-PSMA antibody fragments may offer a favorable compromise by reducing hematologic toxicities and hepatic uptake, while maintaining low renal uptake and avoiding accumulation in salivary glands.

Parallel efforts to target ACP3 have also been reported. The ¹¹¹In-labeled monoclonal antibody PAY-276 achieved acceptable sensitivity for detecting metastatic lesions at doses higher than 40 mg. Detection of soft tissue lesions remained suboptimal, highlighting the need for improved ACP3-targeted products [40].

In our animal study, we observed good tumor targeting for both EKA4 and H92A9 antibodies. However, achieving effective antibody concentrations in solid tumors remains a major challenge. Previous studies have shown that only about 0.01% of the injected dose per gram of tissue reaches the tumor in patients with cancer [41], thus potentially limiting their therapeutic efficacy. Optimal binding affinity must balance two opposing needs: rapid diffusion for deep tumor penetration and sufficient retention to allow pharmacodynamic effects. It has been reported that high-affinity antibodies (in IgG format) may suffer from reduced tumor penetration due to an antigen-barrier effect [42]. This highlights the importance of optimizing antibody properties for effective delivery. Future research should explore innovative approaches to enhance antibody diffusion within neoplastic lesions, ensuring effective treatment while maintaining high selectivity for the target.

In line with these considerations, our human tissue microarray analysis did not reveal PSMA expression in kidneys, intestine, and salivary glands, which is in apparent contrast to previously published data on PSMA distribution [8]. This discrepancy may, at least in part, reflect differences in the antibody clones used for immunohistochemical analyses, as clone-dependent variability in antigen detection is well documented [43–45]. Moreover, prior studies have highlighted substantial intra-patient heterogeneity in PSMA expression, including variability in the proportion of PSMA-positive cells within individual tumors and across lesions [46].

Previous work from our group proposed that glutamate carboxypeptidase III (GCPIII) expression in the salivary glands may account for the xerostomia observed in patients receiving Pluvicto™ therapy [47]. GCPIII shares approximately 70% sequence identity with PSMA and exhibits comparable enzymatic activity and active site architecture, making it a recognized biological “anti-target” of PSMA-based small molecule therapeutics [47]. In support of the specificity of our approach, we demonstrated that our antibodies do not cross-react with GCPIII, thereby reducing the likelihood of off-target binding and associated toxicities in prostate cancer patients. Further investigation in larger cohorts of patient specimens using our antibody will be essential to validate these findings and to better characterize inter-patient variability.

In conclusion, our findings support the use of PSMA and ACP3 as targets for antibody-based therapies in prostate cancer. While PSMA is a well-established target for prostate cancer, ACP3 has been less extensively studied, yet recent evidence suggests that it may represent a cleaner antigen, as no expression has been detected in salivary glands or kidneys [17].

The development of antibody-based reagents such as H92A5 and EKA4 not only allows for selective tumor targeting but also provides powerful tools to further investigate the biological role and heterogeneity of these antigens in patient-derived samples. Importantly, both antibodies demonstrate robust tumor accumulation while sparing healthy organs, underscoring their potential for safe and effective therapeutic applications. Together, these findings lay a solid foundation for future preclinical studies and translational strategies aimed at improving prostate cancer treatment.

Author contributions

Claudia Comacchio (Formal analysis [equal], Investigation [equal], Methodology [equal], Validation [equal], Visualization [equal], Writing—original draft [equal], Writing—review & editing [equal]), Stefanie K Pfister (Investigation [supporting], Writing—review & editing [supporting]), Domenico Ravazza (Methodology [supporting], Writing—review & editing [supporting]), Ettore Gilardoni (Methodology [supporting], Writing—review & editing [supporting]), Tobias Weiss (Resources [equal], Supervision [equal], Writing—review & editing [supporting]), Michael Weller (Supervision [equal], Writing—review & editing [supporting]), Emanuele Puca (Writing—original draft [equal], Writing—review & editing [equal]), Dario Neri (Conceptualization [equal], Data curation [equal], Formal analysis [equal], Project administration [equal], Resources [lead], Supervision [equal], Writing—original draft [equal], Writing—review & editing [equal]), Roberto De Luca (Conceptualization [equal], Data curation [equal], Formal analysis [equal], Investigation [equal], Methodology [equal], Project administration [equal], Supervision [equal], Validation [equal], Visualization [equal], Writing—original draft [equal], Writing—review & editing [equal])

Conflict of interest

D.N. is co-founder, CEO, CSO, and President of the Scientific Advisory Board of Philogen. C.C, S.K.P, E.P. D.R., E.G., R.D. are employed by Philochem AG, the research and development unit of the Philogen group. M.W. has received research grants from Novartis, Quercis and Versameb, and honoraria for lectures or advisory board participation or consulting from Anheart, Bayer, Curevac, Hemerion, Iqvia, Medac, Novartis, Novocure, Orbus, Pfizer, Philogen, Roche and Servier. The other authors declare no potential conflict of interest.

Funding

This research was funded by Philochem AG, the research and development company of the Philogen group.

Data availability

All data relevant to the study are included in the article or uploaded as Supplementary Data. Further information is available from the corresponding author upon request.

Ethics and consent statement

Not applicable.

Animal research statement

Animal experiments were conducted according to the protocols approved by the Cantonal Veterinary Office Zurich (license number ZH006/21) in compliance with the Swiss Animal Protection Act (TSchG) and the Swiss Animal Protection Ordinance (TSchV).