Amanitin-Based Fc-Small Molecule–Drug Conjugates with Noncleavable Linker: A Novel Therapeutic Strategy for Prostate Cancer Targeted Therapy

Abstract

Prostate cancer remains a major global health burden, with limited options and poor prognosis in advanced stages. To extend survival and improve the quality of life for patients, targeted therapies have become an emerging treatment modality. Antibody–drug conjugates (ADC) have shown huge clinical success but only limited therapeutic effects in prostate cancer as their large size limits tumor penetration. Small molecule–drug conjugates (SMDC), consisting of a small molecule as a binding moiety and a cytotoxic drug, offer advantages due to their smaller size, but their short plasma half-life and poor efficacy have hampered their clinical breakthrough so far. Fragment crystallizable (Fc)–grafted SMDCs (Fc-SMDC) combine a SMDC with the half-life–extending Fc fragment of an antibody. So far, the Fc fragments have been attached directly to the small molecule–drug complex, making an enzymatic cleavable linker for payload release an indispensable prerequisite. We report a first-in-class Fc-SMDC targeting PSMA and carrying α-amanitin with a noncleavable linker. The payload is conjugated directly to the Fc region via an engineered cysteine to separate it from the targeting moiety. This approach enables, in contrast to conventional Fc-SMDCs, the use of noncleavable linkers for minimal premature drug release, increased plasma stability, and low systemic toxicity. Due to the noncleavable linker, our conjugate showed high tolerability and an extended half-life, resulting in prolonged tumor exposure and excellent antitumor efficacy in xenograft models. Together with the favorable tolerability in non-human primates, these findings highlight the potential for a next-generation treatment for prostate cancer.

Graphical Abstract

Introduction

Prostate cancer is one of the most prevalent and challenging cancer entities worldwide and typically ranks as the fifth most common cause of male cancer-related death in the majority of Western countries (1, 2). For early-stage, localized prostate cancer, standard-of-care treatments such as radiotherapy and radical prostatectomy are highly effective, achieving a 5-year overall survival (OS) rate of approximately 97% (3). However, in more than 20% of cases, the disease progresses to a metastatic stage. Once metastasized, therapeutic options become limited, and metastatic castration-resistant prostate cancer (mCRPC) has a 5-year survival rate of only 15% (4).

Targeted therapies like CAR-T, radioligand therapies (RLT), bispecific antibodies, or antibody–drug conjugates (ADC) have become a breakthrough in modern cancer therapy, offering the ability to selectively kill tumor cells while sparing healthy tissues (5). However, so far, only the small molecule–based radiopharmaceutical Lutetium Lu-PSMA-617 (Pluvicto) has been approved for the treatment of mCRPC. It demonstrated improved OS and progression-free survival (PFS), but primary resistance in approximately one third of the patients remains an issue (6).

Among targeted therapies, ADCs have shown remarkable success as they combine the targeting specificity of monoclonal antibodies (mAb) with the cytotoxic potency of small-molecule payloads (7). However, the large molecular size and structural complexity can adversely affect their pharmacokinetic profile, resulting in challenges related to tissue distribution, metabolism, and clearance (8).

To address the limitations of ADCs, small molecule–drug conjugates (SMDC) have emerged as a complementary approach. SMDCs use small molecules as targeting ligands, offering several key advantages such as better tissue penetration (9), lower immunogenicity, and lower production costs (10). Despite these benefits, most SMDCs suffer from poor in vivo efficacy due to their short half-life, limiting their clinical success, with only six candidates currently in clinical trials and none yet approved (11).

In cases where the short half-life is the limitation, the therapeutic potential of SMDCs can be enhanced by incorporating fragment-crystallizable (Fc) domains of antibodies (12) to improve pharmacokinetic properties, extend half-life, and boost efficacy (12). Fc-SMDCs seek to combine the main advantages of ADC-targeted drug delivery over a long time—and SMDCs—better tumor penetration due to the smaller size. They have a smaller molecular weight than ADCs (approximately 60 vs. 150 kDa) and benefit from improved pharmacokinetics due to the Fc antibody domain, extending their half-life compared with conventional SMDCs. One of the key components of these conjugates is the linker moieties combining the three components as they influence both plasma stability and efficiency of payload release. All Fc-SMDCs published so far use cleavable linkers to release the payload in response to tumor-specific properties, such as pH-sensitive or valine–alanine–p-aminobenzylcarbamate linkers, simply due to an inherent necessity of their design based on trivalent linker systems. They function as a connecting interface between the binding moiety, the payload, and the Fc fragment and therefore require a cleavage site for efficient payload release (12–15). The corresponding noncleavable versions serve only as negative controls (13). Despite being the most widely used linker strategy also for ADCs (16), cleavable linkers often show high off-target toxicity and reduced efficacy due to unspecific cleavage in circulation and subsequent killing of antigen-negative (healthy) cells (17). Conversely, noncleavable linkers are highly stable in various chemical and enzymatic environments within the human body and rely on complete lysosomal degradation of the targeting agent to release the drug, usually in the form of a metabolite less prone to unspecific uptake. Their successful release requires full degradation of the whole ADC or SMDC, which varies across different cell types (18), and has limited their use in Fc-SMDCs so far. Thus, the choice of linker influences the final stability of the conjugate, the mechanism of payload release, and its pharmacokinetics.

In addition to these limitations, most ADCs and SMDCs are based on a few toxic compounds, such as maytansinoids, auristatins, or TOPOI inhibitors. The mechanisms of action of these payloads limit their use to proliferating (tumor) cells and, in turn, to fast-growing tumors. Thus, although these conjugates are an emerging therapeutic class also in the treatment of mCRPC (19), their clinical efficacy could suffer from limited activity in slowly growing tumors and in resistant cancer cells, limiting their use in prostate cancer.

In this study, we introduce a first-in-class Fc-SMDC featuring an amanitin derivative with a noncleavable linker. Our innovation lies in a new design comprising a modified conjugation strategy that changes the attachment site of the linker payload away from the targeting moiety to a specific engineered cysteine within the Fc region (THIOMAB approach). The conjugates target prostate cancer cells via a glutamate-urea-based motif 2-[3-(1,3-dicarboxypropyl)-ureido]pentanedioic acid (DUPA) that binds with high affinity (K_i = 8 nmol/L; IC₅₀ = 47 nmol/L) to prostate-specific membrane antigen (PSMA; ref. 20), a transmembrane glycoprotein overexpressed in nearly all prostate cancers (21) and in the neo-vasculature of other solid tumors (22). As payload, we used α-amanitin, a bicyclic octapeptide derived from Amanita phalloides, as a potent RNA polymerase II inhibitor, inhibiting protein synthesis and leading to apoptosis (23). Unlike conventional cytotoxic ADC payloads, α-amanitin acts in a cell cycle–independent manner, making it effective against both dividing and nondividing cells and particularly useful in eliminating slowly dividing and dormant tumor cells to prevent recurrence (24). Additionally, the hydrophilic nature of α-amanitin ensures that it requires receptor-mediated internalization for cytotoxic activity, reducing off-target toxicity and enhancing safety (25).

In this study, we demonstrate that this first-in-class amanitin-Fc-SMDC with a noncleavable linker using the THIOMAB approach (TFc-SMDC-1) exhibited high potency on prostate cancer cells. The long half-life enables long circulation in the bloodstream to maintain high concentrations in tumor tissue and thereby increases efficacy. Its high efficacy in murine xenograft models, together with the high tolerability in both mice and non-human primates, demonstrates a significant therapeutic potential of TFc-SMDC-1 as an effective treatment for prostate cancer.

Materials and Methods

Generation of TFc-SMDCs

The plasmid pEXPR-Fc-N297A-LPETGG, containing the Fc-fragment sequence and a sortase (SrtA) recognition site, was generously provided by Prof. H. Kolmar (TU Darmstadt, Germany). Site-directed mutagenesis was performed to introduce a cysteine mutation at position D265 (DC; EU numbering; ref. 26) in the Fc-fragment sequence to include a cysteine for THIOMAB conjugation. Additionally, a double leucine-to-alanine mutation (LALA; EU numbering; ref. 26) for reduced Fcγ-receptor binding was introduced at positions L234 and L235. Finally, the alanine residue at position 297 was reverted to its wild-type form, asparagine, leading to an Fc fragment with the desired amino acid sequence shown in Supplementary Fig. S1A. Plasmid sequences were confirmed by DNA sequencing at GATC/Eurofins.

The payload with a cleavable linker was synthesized in a single step from (S)-Deoxo-β-amanitin as described in WO2022/194988A2, page 64 (compound XIV; ref. 27). The synthesis of (S)-Deoxo-β-amanitin followed the method reported by Lutz and colleagues (compound 28; ref. 28), whereas the synthesis of the payload with a noncleavable linker is detailed in WO2020/216947 A1, pages 69/70 (compound 14; ref. 29). The bifunctional linker, comprising the DUPA binding moiety and the SrtA recognition tag, was synthesized following the procedure described by Gallo and colleagues (15) with minor modifications. Notably, the ornithine residue reported in the original method was omitted in this synthesis, as a payload attachment site was not required.

The Fc-LALA-DC-LPETGGG (Fc-fragment) was produced using Expi293 cells (RRID: CVCL_D615, Life Technologies, cat. #14635) via transient transfection and was purified by Protein A affinity chromatography using TOYOSCREEN AF-rProtein A HC-650F 5 mL (Tosoh Bioscience, cat. #0023432). Subsequent purification from aggregates and low molecular weight species was achieved by size-exclusion fast protein liquid chromatography (SE-FPLC) with a HiLoad 16/600 Superdex 200 pg prepacked XK26 column (Cytiva Healthcare, cat. #28-9893-36). To conjugate the bifunctional linker to the Fc-fragment, SrtA ligation was employed. The Fc-fragment was first rebuffered overnight in SrtA buffer (Tris-HCl 50 mmol/L, pH 7.5, NaCl 150 mmol/L) to eliminate phosphate traces. The SrtA conjugation was performed by adding the following to the Fc-fragment solution (20.65 μmol/L): 20 equivalents of the DUPA bifunctional linker dissolved in water, CaCl₂ to a final concentration of 5 mmol/L, and 0.125 equivalents of SrtA enzyme (2.6 μmol/L). The reaction was incubated for 3 hours at 37°C. The final conjugate was purified using a HiLoad 16/600 Superdex 200 pg prepacked XK26 column (Cytiva Healthcare, cat. #28-9893-36) and stored at 4°C.

THIOMAB conjugation with maleimide–α-amanitin derivatives featuring either a noncleavable C₆ linker (TFC-SMDC-1) or a valine-alanine protease-cleavable linker (TFC-SMDC-2) was carried out as described before for amanitin-based ADCs, so-called ATACs (30). The procedure was modified by using eight equivalents of the amanitin derivatives and extending the incubation time to 75 minutes instead of the standard 60 minutes. SE-FPLC purification was performed using a HiLoad Superdex column, followed by concentration via Amicon Ultra Centrifugal Filters (50K MWCO; Millipore) and sterile filtration through a 0.22 μm sterile filter Millex-GV (Millipore). The final TFC-SMDCs were stored at 4°C before further use.

Sortase A expression and purification

pET29b plasmids of the SrtA mutant were generously provided by Prof. H. Kolmar (TU Darmstadt, Germany). The SrtA enzyme was generated by transforming NEB 5-alpha Competent Escherichia coli (High Efficiency; RRID:SCR_014586; New England Biolab, cat. #C2987H) via heat shock and purified via IMAC affinity chromatography using an EconoFit Nuvia IMAC Column, Ni-charged, 5 mL (Bio-Rad, cat. #12009286).

SDS-PAGE and Western blot

SDS-PAGE and Western blot were performed as described by Papacharisi and colleagues (31). Briefly, samples were prepared under nonreducing conditions using 2× loading buffer without mercaptoethanol (Bio-Rad, cat. #1610737). Proteins were separated on 4% to 20% stain-free precast gels (Mini-Protean TGX stain-free precast gels, Bio-Rad, cat. #4568093) via electrophoresis, then visualized with the Azure C400 imaging system under UV light. For Western blotting, proteins were transferred to a membrane using the Trans-Blot Turbo system (Bio-Rad). The membrane was blocked and then incubated overnight at 4°C with a primary anti-amanitin antibody (produced at Heidelberg Pharma Research GmbH), followed by a 2-hour incubation with an HRP-conjugated secondary antibody (Cell Signaling Technology, cat. #7074, RRID: AB_209923). Detection was done using a luminol-based ECL substrate and imaged with the Azure C400 system.

Size-exclusion high-performance liquid chromatography analysis

Size-exclusion high-performance liquid chromatography to detect potential aggregates or dimers after conjugation was performed using a Tosoh UP-SW3000 gel filtration column (Tosoh, cat. #0023449) with protein detection in the 200 to 400 nm wavelength range.

Endotoxin quantification

Endotoxin levels in all generated TFc-SMDCs were measured by the EndoZyme II Recombinant Factor C (rFC) Assay (Hyglos, cat. #890030) according to the manufacturer’s instructions. A standard curve was generated by serial dilutions of lipopolysaccharide (LPS) in PBS, ranging from 50 to 0.005 EU/mL. Spike control samples containing 5 and 0.5 EU/mL LPS were included to assess assay validity, with acceptable spike recoveries set between 50% and 200%. All samples, standards, and controls were prepared in duplicates. The assay reagent was prepared and applied to the samples following the manufacturer’s guidelines. Fluorescence was measured at 37°C at 0 and 60 minutes. The signal at 0 minutes was subtracted from the 60-minute reading, and a 4-parameter sigmoidal nonlinear regression model was used to generate the standard curve. Sample endotoxin concentrations (EU/mL) were calculated based on the curve. The specification limit for in vivo studies was set at 5 EU/mg for all TFc-SMDCs.

Detection of α-amanitin concentration by competitive anti-amanitin ELISA

TFc-SMDCs solutions were analyzed using a competitive anti–α-amanitin ELISA to detect free amanitin in solution. TFc-SMDCs solutions were first mixed with 100% ethanol and incubated for 20 minutes at −20°C. The mixtures were then centrifuged at 13,000 rpm for 10 minutes at 4°C to precipitate all protein content. The resulting supernatants were collected and fully evaporated using a rotational vacuum concentrator at 53°C for approximately 4 hours. The dried residues were reconstituted in ELISA sample buffer consisting of 1% BSA in PBS with 20% ethanol. ELISA plates were coated overnight at 4°C with anti–α-amanitin serum (6.67 μg/mL, Heidelberg Pharma Research GmbH). On the following day, the plates were blocked for 1 hour at 37°C using 3% BSA in PBS, then washed with PBS to remove excess blocking solution. A standard curve was generated using 1:3 serial dilutions of a 1 mmol/L α-amanitin (CAS No.: 23109-05-9) stock solution, resulting in final concentrations ranging from 0.4 to 8,100 nmol/L. Both standard solutions and prepared samples were mixed in a 1:1 ratio with a 1 nmol/L biotin–α-amanitin conjugate solution (Heidelberg Pharma Research GmbH) and added in duplicates to the coated ELISA plates. Plates were incubated for 1 hour at 37°C, followed by washing with PBS. Next, 1 μg/mL streptavidin–HRP was added to each well, and the plates were incubated for another hour at 37°C. After a final PBS wash, the detection was carried out by adding 3,3′,5,5′-tetramethylbenzidin (TMB) substrate. The colorimetric reaction was allowed to develop for 20 minutes before being stopped by the addition of 1 mol/L sulfuric acid (H₂SO₄). Absorbance was measured at 450 nm with background correction at 570 nm. Final absorbance values were calculated by subtracting the 570 nm reading from the 450 nm reading. α-Amanitin concentrations were determined by interpolating sample absorbance values onto the standard curve using a nonlinear regression model with a sigmoidal fit dose-response equation in GraphPad Prism (RRID: SCR_002798).

Intact analysis of TFc-SMDCs by LC/MS

Native LC/MS analysis was used for the drug-to-protein ratio (DPR) and linker-to-protein ratio (LPR) determination of the intact TFc-SMDCs. Prior to MS analysis, the TFc-SMDCs were deglycosylated with PNGase F (nonreducing format, New England Biolabs, cat. #P0711S) following the manufacturer’s protocol. Analytes were separated on a BioResolve RP mAb Polyphenyl column (450 Å, 2.7 μm, 1.0 × 100 mm, Waters, cat. #SKU176004209) maintained at 80°C and measured on a BioAccord (ACQUITY UPLC I Class Plus coupled with ACQUITY RDA detector) mass spectrometer (Waters). The gradient was generated using solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in acetonitrile) at a flow rate of 300 μL/minute. The gradient started with a linear increase to 50% solvent B over 7 minutes, followed by a further increase to 90% over the next 2 minutes. The mass spectrometer was operated in positive ion mode, with the electrospray source set at 550°C. The cone voltage was set to 150 V, and the capillary voltage was 1.5 kV. Data acquisition was performed in full scan mode over a mass range of 400 to 7,000 m/z at a scan rate of 2 Hz. All operations were controlled using Waters UNIFY software (version 1.9.12.7).

Cell lines

The human prostate carcinoma cell lines LNCaP (RRID:CVCL_0395), 22Rv1 (RRID:CVCL_1045), and PC3 (RRID:CVCL_E2RM) were purchased in 2019 from DSMZ [cat. #ACC 256, cat. #ACC 438, cat. #ACC 465]. All cell lines are of male biological sex. Cell authenticity was confirmed by the cell line supplier DSMZ using STR analysis according to the global standard ANSI/ATCC ASN-0002-2011 (2011), resulting in an authentic STR profile of the reference STR database. The working cell banks of the three cell lines were stored in liquid nitrogen. After banking, cell line authenticity was confirmed by using Cell Authentication by Multiplexion GmbH, and morphology and growth characteristics were routinely compared with the suppliers’ data. The last authentication was done in 2022, immediately before the in vivo work started. Additionally, all cell lines were tested for mycoplasma contamination directly after thawing and periodically every 30 days in culture by using the PlasmoTest Mycoplasma Detection Kit (InvivoGen, cat. #rep-pt1) as per the manufacturer’s instructions. All assays were performed with cells that were cultured for a maximum of 90 days and did not exceed 12 passages in culture.

Quantitative flow cytometric analysis

Binding of DUPA-FITC alone and the Fc-SMDC construct on PSMA-expressing LNCaP cells was assessed by flow cytometry. Cells were incubated (2.5 × 10⁵ cells/well) with 100 μL DUPA-FITC or Fc-SMDC-1-AF488 for 30 minutes at 4°C [100 μL of a 100 μg/mL solution in FACS staining medium (PBS, 25 mmol/L HEPES, 3% FCS, 0.02% Na-azide)]. Cells were washed and analyzed by flow cytometry on a BD FACS Lyric (Becton Dickinson) flow cytometer and the FACSuite Software. The median fluorescence intensity (MFI) signal was determined by 10,000 gated events for each sample. Data were analyzed with FlowJo (RRID: SCR_008520).

In vitro cytotoxicity assays

Cells were seeded at a density of 2 × 10³ cells per well in 90 μL of the appropriate culture medium in 96-well plates. The following day, a panel of eight serial dilutions of TFc-SMDCs (1 × 10⁻⁶ mol/L to 1.28 × 10⁻¹¹ mol/L) and free α-amanitin (1 × 10⁻⁵ mol/L to 1.28 × 10⁻¹⁰ mol/L) was prepared in cell culture media. Ten microliters was added to each well and then incubated for an additional 96 hours in a humidified incubator with 5% CO₂. Cell proliferation was measured using the BrdU chemiluminescent ELISA (Roche; Sigma-Aldrich/Merck, cat. #1669915001), following the manufacturer’s instructions. Luminescence was recorded using a BMG CLARIOstar microplate reader. All treatments were performed in triplicate, and cell viability was normalized to the respective minimum and maximum values. Data are presented as the mean ± SD from three independent experiments.

Plasma stability studies

The solutions of the conjugates were diluted to 0.6 mg/mL (1 × 10⁻⁵ mol/L) in human plasma (HP; BioTrend, cat. #B2013518), mouse plasma (MP; BioTrend, cat. #D408-07-0500), cynomolgus plasma (CP; BioTrend, cat. #NB-53-0017-10), and PBS previously filtered with a 0.45 μm filter. The diluted samples were incubated for 0, 4, and 10 days at 37°C. Samples were stored at −80°C until the day of analysis. The thawed samples were then diluted to 1 μg/mL in PBS for analysis by Western blot (as described above).

Determination of maximum tolerated dose (MTD)

All animal experiments were performed according to the guidelines of the German Animal Welfare Act, and all relevant laws, regulations, and protocols were approved by the local ethics committee (G-64_21, G-228_20, G-157_17, Regierungspräsidium Karlsruhe). The tolerability of TFc-SMDCs was assessed in 7-week-old tumor-free male CB-17 SCID mice (CB17/Icr-Prkdc^scid/Rj, RRID: IMSR_RJ: CB17-SCID) and NMRI nude male mice (Rj:NMRI-Foxn1nu/nu, RRID:IMSR_RJ:NMRI). TFc-SMDCs were administered intravenously (i.v.) as a single dose (10 mL/kg) sterile solution in PBS, pH 7.4. In CB-17 SCID mice, TFc-SMDC-1 doses of 4, 8, 10, 15, and 20 mg/kg and TFc-SMDC-2 doses of 1, 2, and 3 mg/kg were tested. In NMRI nude mice, TFc-SMDC-1 doses of 15, 20, 25, and 30 mg/kg were tested. Survival and clinical symptoms were monitored daily, and body weight was measured twice weekly over a 14-day period. In each group, three animals received escalating doses of the test compound until the onset of clinical signs, including weight loss exceeding 20%, persistent immobility, hind leg paralysis, cachexia, deteriorated overall health, and other indicators of pain or distress.

In vivo efficacy

Male CB-17 SCID mice (CB17/Icr-Prkdc^scid/Rj, RRID: IMSR_RJ: CB17-SCID) were subcutaneously inoculated in the right flank with 2.5 × 10⁶ LNCaP cells suspended in RPMI medium without phenol red containing 50% Matrigel. For the efficacy study, TFc-SMDC-1 was administered at 7.5 mg/kg (½ MTD) and at 3.75 mg/kg (¼ MTD), and TFc-SMDC-2 at 0.75 mg/kg (¾ MTD) and at 0.5 mg/kg (½ MTD). For the determination of the MED, TFc-SMDC-1 was administered at 1.88 mg/kg (⅛ MTD), 0.97 mg/kg (1/16 MTD), and 0.48 mg/kg (1/32 MTD). Six- to eight-week-old NMRI nude male mice (Rj:NMRI-Foxn1nu/nu, RRID:IMSR_RJ:NMRI-NUDE) were inoculated subcutaneously with 5 × 10⁶ 22Rv1 cells suspended in RPMI medium without phenol red containing 50% Matrigel into their right flank. TFc-SMDC-1 was administered at 12.5 mg/kg (½ MTD) and at 6.25 mg/kg (¼ MTD). In both studies, tumor volumes (TV) and body weight were measured twice weekly using calipers and calculated using the formula: TV = (W² × L)/2, where L is the tumor length and W is the perpendicular width (L > W). Once the average TV reached approximately 120 mm³ (24 days after inoculation), animals were randomized into groups of 10 based on tumor size. Test compounds were administered i.v. starting 1 day after randomization once a week for 3 weeks. Tumor and body weight measurements continued twice per week until the study end or until predefined termination criteria were met. The minimum effective dose (MED) was defined as the dose at which mean TV decreased below baseline levels on the day of randomization for at least two consecutive measurements.

In vivo pharmacokinetic (PK) and biodistribution study

Serum, tumor, liver, and kidney samples from male CB-17 SCID mice (CB17/IcrPrkdc^scid/Rj, RRID: IMSR_RJ: CB17-SCID) bearing LNCaP tumors were collected 5 minutes, 2, 6, 24, 48, and 96 hours and 7 and 14 days after a single intravenous dose of 7.5 mg/kg TFc-SMDC-1 (½ MTD) or 0.75 mg/kg TFc-SMDC-2 (¾ MTD). Samples were snap-frozen in liquid nitrogen and stored at −80°C. Tissue extracts were prepared for analysis by cutting 100 mg of tissue and homogenizing it with approximately a 3-fold volume of mouse serum to the mass of the tissue. The organ tissue homogenates were centrifuged for 5 minutes at 14,000 rpm. Supernatants of organ tissues were collected and processed for the detection of TFc-SMDCs by sandwich ELISA and free α-amanitin by LC/MS-MS, respectively. Free amatoxin in serum samples was analyzed via competitive anti-amanitin ELISA.

Detection of free α-amanitin in cynomolgus monkey serum by LC/MS-MS and PK evaluation

Serum samples from cynomolgus monkeys (cyno) treated with TFc-SMDC-1 were processed for the detection of free amatoxins using LC/MS-MS analysis as described by Papacharisi and colleagues (31). Briefly, internal standard (IS) spike solutions were prepared by diluting the stable-labeled metabolites of noncleavable α-amanitin derivatives in acetonitrile (ACN) to a concentration of 20 ng/mL. Calibration samples were prepared by diluting metabolite stock solutions in cyno serum to concentrations of 80, 50, 25, 10, 5, 2, 1, and 0.5 ng/mL, generating calibration curves for each α-amanitin metabolite. To ensure accurate quantification in tissue matrices, quality control (QC) samples were prepared by spiking the α-amanitin metabolite solutions into homogenized tissues from untreated control animals at final concentrations of 75, 15, and 2.5 ng/mL. These calibration standards and QC samples were used to establish the calibration curves for the analysis of free α-amanitin concentrations in serum samples. Serum samples from TFc-SMDC-1 treatment groups were analyzed separately using the corresponding IS spike solution, calibration standards, and QC samples. Each sample was mixed with its respective IS spike solution in ACN and incubated for 20 minutes at −20°C. After incubation, the samples were centrifuged for 10 minutes at 14,000 rpm at 4°C. Supernatant of the serum samples was transferred to Phree Phospholipid Removal 96-well plates (Phenomenex, cat. # 8E-S133-TGB, 30 mg/well) and centrifuged for 1 minute. LC/MS-MS analysis was performed using positive electrospray ionization mode on an Agilent 1200 LC system coupled to a QT6500+ mass spectrometer, equipped with a Waters ACQUITY UPLC BEH Amide column (130 Å, 1.7 μm, 2.1 × 150 mm; SKU 186004802). The concentration of free α-amanitin was calculated by fitting the responses of the calibration standards to a linear regression model with 1/x² weighting and interpolating the sample responses against the regression curve.

Detection of TFc-SMDCs in serum, tissues and tumor by sandwich ELISA and PK evaluation

The concentration of TFc-SMDCs in mouse and cyno serum, as well as mouse tumor and tissues from the in vivo PK and biodistribution studies, was determined using a sandwich ELISA. For samples from the mouse study, serum, tumor, and organ extract samples from TFc-SMDC-treated animals were diluted in PBS containing 0.02% serum in sample buffer (Candor Bioscience, cat. #105500). For the cyno PK study, serum samples from TFc-SMDC-1-treated animals were diluted in PBS containing 0.002% cyno serum in sample buffer (Candor Bioscience, cat. #105500). To generate a calibration curve, standard samples were prepared in 0.02% mouse serum or 0.002% cyno serum in sample buffer at final concentrations of 200, 100, 50, 25, 12.5, 6.3, 3.2, and 1.6 pmol/L for each TFc-SMDC. ELISA plates were coated with polyclonal anti–α-amanitin antibody (6.7 μg/mL, Heidelberg Pharma Research GmbH) and incubated overnight at 4°C. The following day, plates were blocked with Smart Block solution (Candor Bioscience, cat. #113500) for 1 hour at 37°C and then washed with PBS. Calibrators and diluted tissue, tumor, and serum samples were transferred to the coated plates in duplicates and incubated for 1 hour at 37°C. After washing, samples were incubated with a secondary rabbit anti-human antibody (Abcam, cat. #ab98576, RRID: AB_10676131; 0.5 μg/mL in sample buffer) for 1 hour at 37°C. Following another wash step, plates were incubated with TMB working solution for 20 minutes, and the reaction was stopped with 1 mol/L H₂SO₄. Optical density was measured at 450 nm and corrected for background by subtracting absorbance at 570 nm. The concentration of TFc-SMDCs in each sample was calculated by interpolation from the standard curve using a sigmoidal dose–response model with nonlinear regression in GraphPad Prism (RRID: SCR_002798). The resulting TFc-SMDC concentrations (in μg/mL) were plotted against time, and pharmacokinetic parameters were calculated for each animal using WinNonlin software (RRID: SCR_024504, Version 8, Certara).

Detection of free α-amanitin in mouse serum and tissue samples via competitive ELISA and PK evaluation

For the competitive ELISA analysis, 60 μL of serum or tissue sample of mice from the PK study was precipitated with 240 μL of 100% ethanol. Tubes were incubated at −20°C for 20 minutes to ensure complete protein precipitation in the samples. The supernatants, after centrifugation for 5 minutes at 14,000 rpm (Thermo Fisher Scientific Fresco 17 Microcentrifuge, refrigerated, cat. #75002420), were collected and evaporated using a rotational vacuum concentrator (Martin Christ GmbH) for 4 hours at 53°C. The resulting dry residues were reconstituted in ELISA sample buffer composed of 1% BSA in PBS containing 20% ethanol. Reference standard solutions of α-amanitin, ranging from 0.4 to 8,100 nmol/L, were prepared in mouse serum. The mouse serum sample with the highest DMSO content served as the blank. Samples were tested in a competitive anti-amanitin ELISA as described above. The α-amanitin concentration in each sample was calculated by interpolating from the absorbance value to α-amanitin concentration using the standard curve and applying a nonlinear regression with a sigmoidal dose-response equation in GraphPad Prism (RRID: SCR_002798).

Ex vivo immunofluorescence imaging of tumor CDX model

LNCaP tumors from male xenografted CB-17 SCID mice (CB17/Icr-Prkdc^scid_/Rj, RRID: IMSR_RJ: CB17-SCID) were collected 6, 24, 96, and 168 hours after a single intravenous dose of TFc-SMDC-1 treatment. Tumors were divided in half, either snap-frozen in liquid nitrogen or embedded in OCT within a Tissue-Tek Cryomold, then stored at −80°C. Sections were cut at 7 µm thickness using a cryostat at −18°C with an MX35 ultra microsome blade, placed on Polysine slides, and stored at −20°C. Before staining, slides were brought to room temperature for 1 hour, rehydrated in TBS buffer for 3 minutes, and fixed in 4% formalin for 2 minutes. Following three 5-minute TBS washes, the tumors were permeabilized with 0.1% saponin. After another three washes with TBS, the slices were incubated in blocking buffer containing 5% goat serum and 1% BSA for 2 hours. This was followed by staining with Alexa Fluor 488 AffiniPure F(ab′)₂ Fragment Goat Anti-Human IgG (H + L; Jackson ImmunoResearch, cat. #109-546-003, RRID:AB_2337843) diluted 1:50 in blocking buffer. The slides were incubated in a humidity chamber for 2 hours, then washed three times in TBST. Fluoroshield (Sigma-Aldrich, cat. #F6057, RRID: SCR_015961) with 4,6-diamidin-2-phenylindol (DAPI) was used as mounting media and to visualize nuclei. Imaging was performed by fluorescence microscopy using NIS-Elements (RRID: SCR_014329; Nikon).

Tolerability study and PK evaluation in cynos

A tolerability study in cynos was performed at Accelera S.r.l. (Nerviano, Italy). All animal procedures (including housing, health monitoring, restraint, dosing, etc.) and ethical review were performed according to the current Italian legislation (Legislative Decree March 4th, 2014 n. 26) enforcing the 2010/63/EU Directive on the protection of animals used for biomedical research. TFc-SMDC-1 (in PBS, pH 7.4) was administered intravenously (2 mL/kg) to three male cynos (1–2 years old) in cyclic administrations at increasing doses of 0.5, 1, 2, and 4 mg/kg to the same animals. Each dosing cycle consisted of three weekly 30-minute infusions into peripheral veins, followed by an at least 21-day observation period.

Throughout the study, the animals were monitored daily for mortality, clinical signs, and food consumption. Body weights were recorded before the dosing started, and subsequent measurements were taken weekly. Serum chemistry and hematology assessments, including clotting and urinalysis evaluations, were conducted at several time points after each dosing. Serum collection for systemic exposure assessment was conducted at various time points during all cycles of treatment. After the completion of the third treatment cycle (2 mg/kg), animal 2 was replaced by animal 4 due to a veterinarian’s recommendation. The clinical signs leading to the removal of the animal from the study were not linked to the study drug.

Calculation of the therapeutic index

The therapeutic index (TI) was calculated following an allometric method based on tolerability in cynos and the minimal efficacious dose (MED) in mice. The MED in mice of 1.88 mg/kg was converted based on the body surface area as recommended by the FDA in the guidance for the industry (32) to a cyno MED of 0.48 mg/kg. This, combined with the tolerated dose of 4 mg/kg in cynos, resulted in a TI of 8.3.

$$\text{TI}\hspace{0.17em}\mathrm{=}\mathrm{\hspace{0.17em}}\frac{MTD\mathrm{\hspace{0.17em}}\left(moneky\right)\mathrm{\hspace{0.17em}}}{MED\mathrm{\hspace{0.17em}}\left(mouse\right)\mathrm{\hspace{0.17em}}converted\mathrm{\hspace{0.17em}}}\mathrm{\hspace{0.17em}}\mathrm{=}\mathrm{\hspace{0.17em}}\frac{4}{0.48}\mathrm{\hspace{0.17em}}\mathrm{=}\mathrm{\hspace{0.17em}}8.3$$

Statistical analysis

All statistical analyses were conducted using GraphPad Prism (RRID: SCR_002798). TV between experimental groups in in vivo studies was compared using unpaired t tests and Holm–Šídák multiple comparison tests. Differences were considered statistically significant at P ≤ 0.05. Survival analysis between treatment groups was performed using the log-rank (Mantel-Cox) test. Paired t tests were used to compare tissue concentrations of different α-amanitin derivatives. The half-life of TFc-SMDCs was determined by fitting the data to a one-phase exponential decay model using nonlinear regression.

Results

Optimized double-step conjugation generates superior TFc-SMDCs

In this study, one DUPA-Fc-amanitin conjugate with a noncleavable C₆ linker (TFc-SMDC-1) and one with a valine-alanine protease cleavable linker as a control (TFc-SMDC-2) were generated (Fig. 1A). To combine the DUPA binding moiety, the amanitin payload, and the IgG1-Fc-fragment in a two-step conjugation process, we introduced two modifications into the Fc-fragment. First, the Fc-fragment was modified at the C-terminus with the peptide tag LPETGGG for SrtA-mediated ligation to attach the DUPA binding moiety. A bifunctional group containing both the DUPA binding moiety and an oligo-glycine sequence for SrtA ligation (Fig. 1B) was conjugated via SrtA-catalyzed amide bond formation between the C-terminal sorting motif and the oligo-glycine sequence of the linker.

Figure 1.

Schematic representation of TFc-SMDC two-step conjugation and structures of the chemical compounds. A, Fc-LALA-D265C-LPETGGG is programmed by transpeptidation. The DUPA-PSMA targeting ligand is attached via the bifunctional linker to the C-terminus through Sortase A (eSrtA)-mediated ligation. DUPA-Fc-LALA-D265C-LPETGG is then armed with different amanitin derivatives via site-specific maleimide conjugation, resulting in TFc-SMDCs. B, Chemical structure of the bifunctional linker: PSMA-targeting ligand DUPA: L-Glu-urea-L–Glu (blue), PSMA supporting−binding spacer 8-Aoc-Phe-Phe (dark blue), and olygoglycine (GGG; pink). C, Chemical structure of amanitin derivatives: amanitin (black), noncleavable C₆-linker at amino acid 4 (blue), maleimide linker (red), and cathepsin B valine-alanine PAB cleavable linker at amino acid 1 (green).

Second, the aspartate at position 265 (according to EU numbering of antibodies) was genetically exchanged for a cysteine to enable the conjugation of the amanitin payload via maleimide chemistry (THIOMAB approach). Following SrtA-mediated ligation, the DUPA-Fc construct was equipped with maleimide–amanitin derivatives, incorporating either a noncleavable or a cleavable linker payload derivative (Fig. 1C) via maleimide chemistry at the engineered cysteine.

In addition to the conjugation sites, the Fc-fragment was further engineered to incorporate the LALA mutation for reduced immunogenicity and Fcγ receptor interaction to improve the tolerability of the final conjugates.

Conjugation efficiency was monitored by intact LC/MS analysis. The DUPA attachment via the SrtA reaction was highly efficient, and the conjugation of the amanitin payload to the DUPA-Fc fragment led to a final conjugate with an average drug-to-protein ratio (DPR) of 1.8 for TFc-SMDC-1 and 1.7 for TFc-SMDC-2 (MW ≈ 60 kDa; Supplementary Fig. S1).

TFc-SMDCs showed potent and selective cytotoxicity and high plasma stability in vitro

Both TFc-SMDCs exhibited full-blown cytotoxicity in LNCaP cells (PSMA⁺⁺⁺), with EC₅₀ values in the nanomolar range [6.4 nmol/L for TFc-SMDC-1 and 2.4 nmol/L for TFc-SMDC-2, Fig. 2A (left); Supplementary Table S1] and still strong cytotoxicity in the 22Rv1 cell line (PSMA⁺), with EC₅₀ values of 107 nmol/L for TFc-SMDC-1 and 37 nmol/L for TFc-SMDC-2 [Fig. 2A (middle)]. Both TFc-SMDCs showed target-specific in vitro cytotoxicity as cytotoxicity declined in direct correlation with PSMA expression, with the highest activity on high-expressing LNCaP cells and no effect on target-negative PC3 cells [Fig. 2A (right)]. Notably, TFc-SMDC-1 generally showed slightly lower in vitro cytotoxicity than TFc-SMDC-2 as noncleavable linkers rely on degradation of the whole molecule, whereas enzymatic cleavable linkers release the payload directly and immediately inside the cell. Steric hindrance due to the Fc-fragment was excluded by demonstrating comparable binding of DUPA–FITC and the sterically more demanding TFc-SMDC-1-AF488 to LNCaP cells in FACS analysis (Supplementary Fig. S1).

Figure 2.

In vitro characterization of TFc-SMDCs. A,In vitro cytotoxicity of TFc-SMDCs and free α-amanitin on LNCaP, 22RV1, and PC3 cells. Blue curves: TFc-SMDC-1 (noncleavable linker), green curves: TFc-SMDC-2 (cleavable linker), red curves: α-amanitin. Cell proliferation was determined by BrdU incorporation using a chemiluminescent ELISA. B, Quantification of anti–α-amanitin Western blot signal intensity of TFc-SMDC-1 (left) and -2 (right) from time point 0 up to 10 days in PBS at 37°C. C, Quantification of anti–α-amanitin Western blot signal intensity of TFc-SMDC-1 from time point 0 up to 10 days in different plasma matrices at 37°C. D, Quantification of anti–α-amanitin Western blot signal intensity of TFc-SMDC-2 from time point 0 up to 10 days in different plasma matrices at 37°C. The quantification of the signal intensity was calculated with Image Studio Lite Ver 5.2. Each conjugate was considered 100% stable at d0.

Furthermore, plasma stability of both TFc-SMDCs was evaluated in vitro, as it greatly affects the in vivo efficacy and safety profile (33). Both conjugates were incubated in human plasma (HP), mouse plasma (MP), cynomolgus plasma (CP), and PBS for 0, 4, and 10 days and analyzed by anti-amanitin Western blot (Supplementary Fig. S2) for potential premature payload release. Both TFc-SMDCs incubated in PBS at 37°C for up to 10 days showed no decrease in signal intensity at ∼60 kDa and no signs of aggregation, indicating high stability and no detachment of amanitin (Fig. 2B; Supplementary Fig. S2A).

Upon incubation in plasma, the 60 kDa signal of TFc-SMDC-1 gradually decreased in all cases, indicating progressive amanitin detachment (Fig. 2C). However, at least 50% of the 60 kDa signal of the TFc-SMDC-1 remained clearly detectable even after 10 days. In addition, moderate aggregation was detected by the appearance of a faint signal at ∼110 kDa after 4 days, and slightly more pronounced after 10 days (Supplementary Fig. S2B).

In contrast, TFc-SMDC-2 showed a much more pronounced decrease in the 60 kDa signal in all plasma matrices. After 10 days, the 60 kDa signal was nearly undetectable in human and mouse plasma and completely absent in cynomolgus plasma (Fig. 2D), indicating an almost complete amanitin detachment. In addition, TFc-SMDC-2 showed no aggregation in human plasma but increased aggregation in mouse and cynomolgus plasma (Supplementary Fig. S2C).

In conclusion, TFc-SMDC-1 showed, as expected, higher stability than TFc-SMDC-2, and together with the high in vitro selectivity, further in vivo studies of these conjugates were supported.

In vivo efficacy in the LNCaP and 22Rv1 prostate cancer tumor model

To evaluate the antitumor efficacy of the TFc-SMDCs in vivo, mice bearing LNCaP tumors were treated once per week for 3 weeks, as the standard regimen for Fc-SMDCs (15), and tumor growth and OS were monitored. For a more reliable assessment of the therapeutic potential, dosing was adjusted for each TFc-SMDC based on its respective maximum tolerated dose (MTD), enabling equitoxic dosing. The MTD of TFc-SMDC-1 in CB-17 SCID mice was with 15 mg/kg, which is 15 times higher than that of TFc-SMDC-2 at 1 mg/kg (Supplementary Fig. S3). Accordingly, TFc-SMDC-1 was administered at 7.5 mg/kg for ½ MTD and at 3.75 mg/kg for ¼ MTD, whereas TFc-SMDC-2 was administered at 0.75 mg/kg for ¾ MTD and at 0.5 mg/kg for ½ MTD. In the latter case, the dose at ¼ MTD (0.25 mg/kg) was considered to be too low for an antitumor effect, and therefore, the dose of ¾ MTD was chosen instead.

Both TFc-SMDCs showed dose-dependent antitumor activity, but clear differences in efficacy. In the case of TFc-SMDC-1, 50% of the animals treated with 3.75 mg/kg (¼ MTD) showed complete tumor remission, but regrowing tumors were observed over time, whereas in animals treated with 7.5 mg/kg (½ MTD), all animals showed complete remission, and no tumors were detected until the end of the study [day 99; Fig. 3A (blue lines)]. Survival rates were rather similar in both dosing groups, with 80% and 90% of animals alive at study end, respectively (Fig. 3B). In contrast, TFc-SMDC-2 showed, at both doses, only tumor growth delay, with a stronger effect at the higher dose, but without achieving complete remission at any time point [Fig. 3A (green lines)]. Survival rates in these two groups were similar to the control group, with no animals alive after day 65 (0.5 mg/kg), day 72 (control), or day 82 (0.75 mg/kg).

Figure 3.

In vivo antitumor efficacy and survival upon treatment with TFc-SMDCs in an LNCaP prostate xenograft model. Male SCID mice bearing prostate cancer cells were treated with PBS, TFc-SMDC-1 (noncleavable linker), or TFc-SMDC-2 (cleavable linker) at the indicated doses. A, Dose–response curves of antitumor efficacy of TFc-SMDC-1 (noncleavable linker) and TFc-SMDC-2 (cleavable linker). PBS was used as a negative control. Efficacy was monitored for 99 days and evaluated by average tumor volume ± SEM, n = 10. B, Survival of the animals treated with different doses of TFc-SMDCs or PBS, n = 10. Each asterisk on the graphs represents one mouse that was terminated since it met predefinded termination criteria.

As a measure of efficacy, for both TFc-SMDCs, the MED was determined, defined as the lowest dose that reduced tumor volume below the initial level at randomization for at least two consecutive time points. In the above-described efficacy study, the MED for TFc-SMDC-2 was identified to be 0.75 mg/kg (¾ MTD; Fig. 3), whereas all tested doses of TFc-SMDC-1 were far above the MED. Thus, an additional study with lower doses of TFc-SMDC-1 was performed, in which the MED for TFc-SMDC-1 was determined to be 1.88 mg/kg (⅛ MTD; Supplementary Fig. S4). Based on the strong antitumor efficacy observed in the LNCaP CDX model (Fig. 3A), we evaluated the efficacy of TFc-SMDC-1 in the hard-to-treat, PSMA low-to-moderate 22Rv1 CDX model. As this model is run in NMRI nude mice, the MTD was reestablished (Supplementary Fig. S5). TFc-SMDC-1 was administered once a week for 3 weeks at 6.25 mg/g (¼ MTD) and 12.5 mg/kg (½ MTD). Despite the reduced PSMA expression compared with LNCaP, the treatment with 6.25 mg/g (¼ MTD) and 12.5 mg/kg (½ MTD) TFc-SMDC-1 retained a significant antitumor effect (Supplementary Fig. S6A) and a statistically significant prolongation of the median survival of 22.5 days in the group treated with ½ MTD and 24 days in the group treated with ¼ MTD compared with 14 days in the control group (Supplementary Fig. S6B).

In vivo tumor uptake of TFc-SMDCs directly correlates with therapeutic activity

The biodistribution of the TFc-SMDCs in mice was characterized in a PK study following a single administration of 7.5 mg/kg (½ MTD) of TFc-SMDC-1 and 0.75 mg/kg (¾ MTD) of TFc-SMDC-2. The concentration of both TFc-SMDCs in serum, tumor, kidney, and liver tissues was assessed by sandwich ELISA. In serum, total TFc-SMDC concentrations showed a C_max directly after injection, with 69 μg/mL for TFc-SMDC-1 and 11 μg/mL for TFc-SMDC-2 (Fig. 4A). The plasma half-life and dose-normalized C_max values were comparable between both conjugates: 4.9 days and 8.62 kg·μg/mL/mg for TFc-SMDC-1 and 5.5 days and 14.2 kg·μg/mL/mg for TFc-SMDC-2. Thus, both TFc-SMDCs showed pharmacokinetic properties similar to other Fc-SMDCs and a much longer half-life compared with SMDCs devoid of an Fc fragment (15). In tumor tissue, total TFc-SMDC concentrations reached a peak at 6 hours after injection, with 9.8 μg/g for TFc-SMDC-1 and 2.4 μg/g for TFc-SMDC-2 (Fig. 4B), whereas the concentration in kidney and liver tissues remained consistently low throughout the whole study (Supplementary Fig. S7A and S7B).

Figure 4.

Serum pharmacokinetic profile and tumor distribution. Quantitative in vivo biodistribution of TFc-SMDCs and the released α-amanitin at different time points. Head-to-head comparison of TFc-SMDC-1 (noncleavable linker; blue) and TFc-SMDC-2 (cleavable linker; green) in the LNCaP xenograft model. Doses of ½ MTD of TFc-SMDC-1 (7.5 mg/kg) and ¾ MTD of TFc-SMDC-2 (0.75 mg/kg) were administered as a single i.v. dose in male CB17 SCID mice (n = 3) bearing LNCaP tumors. A, Serum levels of TFc-SMDCs measured by sandwich ELISA. B, TFc-SMDC levels in LNCaP tumors up to 14 days, measured by sandwich ELISA. C, Concentration of α-amanitin in LNCaP tumors measured by competitive ELISA. Dotted lines represent the lower limit of quantification (LLOQ) of the respective assay.

The analysis of α-amanitin in tumor, liver, and kidney tissues and serum was conducted via competitive ELISA. In serum, the levels of free (released) toxin remained below the lower limit of quantification (LLOQ) at all time points for both TFc-SMDCs [Supplementary Fig. S8A and S8B (violet lines)]. In the liver and kidney, only small amounts of α-amanitin were detected at all time points throughout the whole study: below 3% in the kidney (i.e., 9%ID/g) and below 2% in the liver (relative to injected amatoxin dose, i.e., 2%ID/g; Supplementary Fig. S9). Most importantly, the highest concentrations of free toxin were found within tumor tissue, with maximum concentrations of 940 ng/g for TFc-SMDC-1 corresponding to 20% (i.e., 100% ID/g) and 123.2 ng/g for TFc-SMDC-2 corresponding to 15% (relative to injected toxin dose, i.e., 52%ID/g; Fig. 4C; Supplementary Fig. S9). In line with the long half-life, free toxin levels in tumor tissue remained high until the end of the study (14 days). These results indicate payload release predominantly at the tumor site and high stability in circulation, with no or only minimal premature release for both TFc-SMDCs.

Given the superior efficacy and increased tumor accumulation of TFc-SMDC-1, a dedicated tumor distribution study was carried out exclusively for TFc-SMDC-1 to assess its localization and spatial distribution within the tumor tissue. In samples collected 6 hours after treatment with TFc-SMDC-1, a homogeneous signal for human Fc-IgG1 was observed throughout the entire tumor tissue (Fig. 5A). The quantification of the fluorescence signal revealed a maximum of 1753 MFI_FITC at 6 hours after treatment, and the intensity remained high for up to 4 days (1727 MFI_FITC) before a slight decrease was observed at 7 days (1294 MFI_FITC), likely due to conjugate metabolism and elimination from the tumor tissue (Fig. 5B). Tumors from PBS-treated control animals showed no detectable staining, confirming that the observed signal was specific for TFc-SMDC-1 (Supplementary Fig. S10).

Figure 5.

Tumor penetration of the TFc-SMDC-1 in human prostate adenocarcinoma LNCaP xenograft in mice. A, Images of the tumors of mice treated with TFc-SMDC-2 for 6, 24, 96, and 168 hours. Twenty-five pictures at 10× magnification of the tumor area were taken and blended together in a large picture (right). Green = staining of TFc-SMDC-2; Blue = DAPI staining. Scale bar = 500 μm. B, Quantification of the intensity of the fluorescence staining. Each data point represents all samples from one mouse. n = 3.

In vivo tolerability and PK evaluation in cynos: A breakthrough study

The high tolerability and efficacy in mice set the base to assess the tolerability of TFc-SMDC-1 in cyno. The tolerability in cynos was evaluated for multiple dosing to match the dosing in the efficacy study in mice for a more reliable calculation of the TI. In addition to overall health condition, specific liver damage parameters, that is, alanine transaminase (ALT), aspartate transaminase (AST), and lactate dehydrogenase (LDH) as general signs of inflammation and organ function were assessed as the liver is the main off-target toxicity organ for amanitin-based ADCs (34). All four doses were well tolerated without relevant body weight changes, clinical signs, or toxicologically relevant changes in clinical pathology. In general, all findings were reversible and without clinical significance. The AST levels were low for the first three doses but clearly elevated directly after each 4 mg/kg injection, reaching a maximum increase of 7.5-fold after the third dose of 4 mg/kg, followed by a steady decrease until the end of the study (Fig. 6A). A minimal increase in LDH values was observed after each injection of the first three doses, followed by a decrease almost to baseline during the 3-week recovery period (Fig. 6B). However, one animal showed a strong increase in LDH level up to 2464 U/L after the first injection of 0.5 mg/kg, which quickly decreased to baseline in the following measurement [Fig. 6B (animal 1)]. After treatment with 4 mg/kg, LDH levels increased temporarily to 3353 U/L (3.7–4.9-fold) in all animals, and although showing a trend to normalization, they were not completely back to baseline by the end of the study. ALT levels were comparable with pretreatment over the whole study (Supplementary Fig. S11). Based on the data obtained from the current study, the highest tested dose of 4 mg/kg administered once weekly for three consecutive weeks is deemed to be well tolerated in cynos.

Figure 6.

Selected biochemical parameters and TFc-SMDC-1 levels in the serum of cynos treated with escalating doses of TFc-SMDC-1. Three male cynos were treated with TFc-SMDC-1 at doses of 0.5, 1, 2, and 4 mg/kg once a week for 3 weeks, with a 21-day recovery period. A, Serum AST level. B, Serum LDH level. Blood and serum parameters were measured twice a week for 160 days. Each line represents one animal. One animal was substituted due to an injury unrelated to the treatment, indicated with an asterisk. The dotted line indicates the minimum, mean, and maximum pretreatment serum values. C, TFc-SMDC-1 levels in serum of cyno samples and PK evaluation thereof. TFc-SMDC-1 concentration in serum samples was determined for each time point using ELISA. Day 0 in the PK curve corresponds with day 128 in the whole study. *After completion of the third treatment cycle (2 mg/kg), animal 2 was replaced by animal 4 due to veterinarian recommendation. The clinical signs leading to the removal of the animal from the study were not linked to the study drug.

In addition to the liver damage parameters, serum levels of TFc-SMDC-1 were assessed using a sandwich ELISA, and PK parameters were calculated. TFc-SMDC-1 levels in serum showed similar PK profiles after each injection in each dosing cycle (Fig. 6C; Supplementary Fig. S12A–S12C). The half-life after the third injection was around 10 days for cycle A, between 8 and 11 days for cycle B, 9 days for cycle C, and between 8 and 14 days for cycle D (Supplementary Table S2). The C_max values showed a slight increase after each injection in each cycle and a dose linearity between cycles. The C_max values were increasing from 8 to 13 μg/mL for cycle A, 20 to 28 μg/mL for cycle B (animal 1 was excluded), 45 to 55 μg/mL for cycle C, and 90 to 115 μg/mL for cycle D (Supplementary Table S2). The levels of free toxin remained below the lower limit of quantification (LLOQ) or slightly above the LLOQ (up to 11 μg/mL) at all time points for all four cycles, indicating again a high stability in circulation with minimal or no premature payload release (Supplementary Fig. S13).

TFc-SMDC-1 shows an excellent therapeutic index

Cynos were chosen to evaluate tolerability to enhance the relevance of preclinical safety assessments for TFc-SMDC-1. The TI of TFc-SMDC-1 was calculated using an allometric method based on tolerability in cynos and the minimal efficacious dose (MED) in mice, as described above. Based on the available data, the highest tested dose of 4 mg/kg was used as the tolerated dose in cynos, but it should be noted that this dose only led to transient changes, suggesting that higher doses might be tolerated as well. In addition, a histopathologic assessment would be needed to assess the full toxicity profile. As MED, a dose was used that led to transient tumor remission for about 3 weeks. Based on this method, a very favorable TI of 8.3 was calculated. Even if a dose leading to complete tumor remission is considered as MED, the TI is still 4, demonstrating the high potential of TFc-SMDC-1.

The selective activity exclusively on PSMA-expressing cells, combined with the high plasma stability and the preferential accumulation in tumor tissue over healthy tissues, highlights the potential of TFc-SMDCs. Especially the strong efficacy of the TFc-SMDC bearing the noncleavable linker (TFc-SMDC-1), as demonstrated in the prostate cancer xenograft mouse model, along with its high tolerability in both mice and cynos, reinforces the promise of (TFc-) SMDCs as a novel therapeutic approach, especially when linker, payload, and conjugation strategy are chosen carefully.

Discussion

The treatment of prostate cancer remains a significant challenge due to its poor prognosis and limited treatment options (4). Targeted therapies to selectively kill tumor cells have recently become a highly promising therapeutic approach in modern cancer therapy. One example falling into this category is Fc-SMDCs, combining the targeted drug delivery of ADCs over a long time with the highly effective tumor penetration of SMDCs. The basic principle of Fc-SMDCs is to find a perfect balance between small molecular weight and improved pharmacokinetics by using Fc antibody domains. The main obstacle so far was that the linker moieties influencing plasma stability and the efficiency of payload release relied on trivalent linker systems with cleavable linkers for efficient payload release (12–14). Cleavable linkers come with some disadvantages, resulting in high off-target toxicity and reduced efficacy (17). These disadvantages can be addressed by noncleavable linkers, but their inefficient payload release, depending on complete intracellular degradation of the whole conjugate, has prevented their use in Fc-SMDCs so far (18).

Here, we introduce a first-in-class THIOMAB-based Fc-SMDC featuring a noncleavable linker payload derivative directly attached to engineered cysteines at the Fc-fragment for increased plasma stability, minimal premature drug release, and improved TI. The Fc fragment was additionally engineered with the LALA mutation to reduce Fcγ receptor-mediated uptake in target-negative cells, thereby minimizing off-target toxicity and enhancing overall tolerability (35). We generated two TFc-SMDC conjugates: TFc-SMDC-1, incorporating a noncleavable C₆ linker, and, as a control, TFc-SMDC-2, featuring a valine-alanine protease-cleavable linker. Both conjugates selectively target PSMA-expressing prostate cancer cells via the glutamate-urea-based DUPA motif that binds with high affinity to PSMA, a transmembrane glycoprotein overexpressed in most prostate cancers. Due to the architecture of the conjugates, the two DUPA moieties per molecule can bind two adjacent PSMA monomers within one PSMA dimer or two neighboring PSMA dimers, leading to higher avidity and, finally, to faster uptake of TFc-SMDCs (36). The potent RNA polymerase II inhibitor α-amanitin, derived from Amanita phalloides, was employed as a cytotoxic payload attached directly to the Fc fragment via engineered cysteines using maleimide chemistry (THIOMAB technology; ref. 37).

Both TFc-SMDCs showed high in vitro cytotoxicity with EC₅₀ values in the nanomolar range in PSMA-positive cells, but without cytotoxicity in PSMA-negative cells, demonstrating target-dependent cellular uptake and efficient payload release. Unconjugated α-amanitin shows limited in vitro activity due to its hydrophilic nature, which prevents passive membrane penetration (25). Given the enhanced plasma stability generally associated with noncleavable linkers, the stability of the two TFc-SMDCs was evaluated in vitro. In comparison with TFc-SMDC-2, TFc-SMDC-1 exhibited higher stability in the plasma of different species. This can be related to the presence of proteases, like Carboxylesterase C1 (Ces1C) in murine plasma or serine proteases in human plasma, both of which were reported to recognize the Val-Ala-PAB sequence (17, 38) and therefore are responsible for premature linker cleavage and decreased stability of Fc-SMDCs with cleavable linkers as compared with noncleavable linker conjugates like TFc-SMDC-1.

The difference in stability was accompanied by a significant difference in tolerability in male SCID mice, with a 15 times higher MTD of TFc-SMDC-1 than of TFc-SMDC-2. This difference might be explainable by the physicochemical properties of the two payload linkers. The cleavable linker in TFc-SMDC-2 is self-immolative, releasing naked Amanitin as the main metabolite, whereas the noncleavable linker in TFc-SMDC-1 relies on complete degradation of the whole TFc-SMDC, releasing the cysteine adduct of the payload linker as the main metabolite. Thus, the main metabolite of TFc-SMDC-2 is likely more potent than the metabolite of TFc-SMDC-1. Additionally, the two payload linkers differ in the attachment position of the linker at the payload, leading to differences in hydrophilicity of the conjugates, with the noncleavable payload linker conjugate being more hydrophilic. As increased hydrophilicity has been shown to reduce unspecific liver uptake (39) and liver toxicity is the main off-target toxicity of amanitin-based conjugates, this might also contribute to the tolerability difference.

The substantial increase in tolerability expanded the therapeutic potential of TFc-SMDC-1 by enabling higher doses in efficacy studies, leading to superior tumor reduction. In fact, the treatment of LNCaP tumors in xenograft mouse models with TFc-SMDC-1 at a dose of ½ MTD resulted in complete tumor remission, whereas treatment with TFc-SMDC-2 at ½ MTD led only to a delay in tumor growth and poor survival. The MED in the LNCaP model was as low as ⅛ MTD for TFc-SMDC-1, but ¾ MTD for TFc-SMDC-2. In line with the high efficacy, immunostaining of the LNCaP tumors showed a homogeneous localization of the TFc-SMDC, likely due to the homogeneous expression of PSMA.

However, also in the 22Rv1 xenograft model, characterized by low and heterogeneous expression of PSMA, TFc-SMDC-1 retained its antitumor activity, resulting in a significant prolongation of OS compared with the PBS control. Notably, no clear difference between the two doses tested (½ MTD and ¼ MTD) was observed, indicating that in fast-growing tumors such as 22Rv1, a more fractionated dosing regimen could be beneficial. This could reduce receptor saturation and increase exposure to the drug (i.e., increased AUC at lower C_max). Further studies, like double immunofluorescence staining for PSMA and Fc-IgG1, would enable direct assessment of colocalization with PSMA-expressing tumor cells and provide further insight into receptor-specific binding, intratumoral localization, and the applicability of TFc-SMDC-1 across a broader spectrum of prostate cancer phenotypes.

Biodistribution studies in tissues confirmed that the efficacy of TFc-SMDCs is correlated with a preferential accumulation at the tumor site compared with healthy organs like the kidneys and liver. Treatment with TFc-SMDC-1 resulted in an 8-fold higher absolute concentration of the active metabolite in the tumor compared with TFc-SMDC-2. PK studies revealed that the half-life of both TFc-SMDCs was similar and in line with earlier findings (15). Most importantly, no free payload was detectable in circulation, indicating high apparent stability of both TFc-SMDCs. However, the absence of detectable free payload does not definitively exclude premature release, as off-target cleavage followed by rapid clearance may result in levels below the assay detection threshold, especially in the case of TFc-SMDC-2, in which the applied dose was already very low due to the low MTD.

The stability of the conjugate is a multifactorial property, and the observed lower tolerability, tumor accumulation, and efficacy of TFc-SMDC-2 might be related to the nature of the cleavable Val-Ala-PAB linker itself, which is recognized and cleaved by cathepsin B. Although cathepsin B is highly expressed in the lysosome, it is also present extracellularly in the tumor microenvironment (40, 41). An extracellular release at the tumor site, together with premature release in circulation upon cleavage by proteases (see above), may contribute to the lower tolerability and efficacy of TFc-SMDC-2.

We also evaluated the tolerability of TFc-SMDC-1 in cynos in an escalating multiple-dose regimen. The primary focus of this study was on enzyme levels of AST and ALT as indicators of liver function as amanitin-based ADCs are known to accumulate primarily in the liver (34), as well as LDH as a general marker of inflammation. TFc-SMDC-1 treatment led to transient mild-to-moderate increases in liver enzymes and LDH, which were considered noncritical as they returned to baseline in all cases, indicating a transient response without long-term impact on health or survival. Notably, hepatic enzyme activity was completely restored to normal levels within the observation period. Evaluation of the PK parameters in cynos revealed a dose linearity between the different cycles and, more importantly, a rather long half-life of approximately 9 to 10 days for TFc-SMDC-1. Interestingly, no salivary gland toxicity was detected in cynos, although PSMA expression is preserved in the species (42). In follow-up studies, it should be determined if TFc-SMDC-1 cross-reacts with cyno PSMA and if uptake is limited, as shown for anti-PSMA antibodies (43). It should also be noted that histopathologic analysis was not performed in this study, and no antidrug antibodies (ADA) were measured. Both readouts are essential to fully understand the toxicity profile of TFc-SMDC-1. However, the half-life remained mostly unchanged between treatment cycles, and no drop in serum drug levels was observed, making ADA formation at least unlikely. The tolerability data from cynos and the efficacy in mice, together with the respective PK data of TFc-SMDC-1, revealed a very favorable TI of 8, very low levels of free amanitin, and a half-life in the same range as observed for ADCs.

In conclusion, we report a first-in-class Fc-SMDC featuring a C₆ noncleavable linker with amanitin as the payload. Our findings demonstrate that incorporating a noncleavable C₆ linker is a viable strategy, particularly for linker-payload conjugation using THIOMAB technology and the LALA mutation, to create effective and well-tolerated targeted therapeutics. TFc-SMDC-1 with a noncleavable linker exhibited better efficacy and higher TI in in vivo studies compared with its counterpart TFc-SMDC-2 with a cleavable linker, underscoring its superior therapeutic profile. These results underline how noncleavable linkers enhance stability and in vivo tolerability, thereby widening the TI and offering important advantages for the development of safer ADCs (31, 44) and TFc-SMDCs. Overall, our findings highlight the potential of TFc-SMDC-1 as a next-generation treatment for prostate cancer, and our new conjugation strategy provides flexibility to incorporate diverse targeting ligands and cytotoxic payloads to broaden the scope for the future development of tumor-targeted therapies using TFc-SMDCs.

Supplementary Material

Supplementary Figure S1

Supplementary Figure S1 shows the biochemical characterization of the T-Fc-SMDC-1.

Supplementary Figure S2

Supplementary Figure S2 shows the plasma stability of TFc-SMDCs by Anti-α-Amanitin western blot.

Supplementary Figure S3

Supplementary Figure S3 shows the assessment of the Maximum tolerated dose (MTD) in CB-17 SCID mice.

Supplementary Figure S4

Supplementary Figure S4 shows the minimum effective dose of TFc-SMDC-1 (MED)

Supplementary Figure S5

Supplementary Figure S5 shows the assessment of the Maximum tolerated dose (MTD) of TFc-SMDC-1 in NMRI male nude mice.

Supplementary Figure S6

Supplementary Figure S6 shows the in vivo antitumor efficacy and survival upon treatment with TFc-SMDC-1 in a 22RV1 prostate xenograft model.

Supplementary Figure S7

Supplementary Figure S7. Quantitative shows the in vivo biodistribution of TFc-SMDC-1 (non -cleavable linker) and TFc-SMDC-2 (cleavable linker) at different time points.

Supplementary Figure S8

Supplementary Figure S8 shows the quantitative in vivo biodistribution of the released toxin α-amanitin at different time points.

Supplementary Figure S9

Supplementary Figure S9 shows the percentage of accumulation of Amanitin in tissue with respect to the injected dose (I.D.) of TFc-SMDCs

Supplementary Figure S10

Supplementary Figure S10 shows immunostaining of tumor sample from LNCaP xenografted mice treated with PBS

Supplementary Figure S11

Supplementary Figure S11 shows serum alanine transaminase (ALT) levels in cynomolgus monkeys treated with escalating doses of the TFc-SMDC-1.

Supplementary Figure S12

Supplementary Figure S12 shows TFc-SMDC-1 levels in serum of cynomolgus monkey samples treated with escalating doses of the TFc-SMDC-1 and PK evaluation thereof

Supplementary Figure S13

Supplementary Figure S13 shows free amatoxin metabolite levels in serum of cynomolgus monkey samples treated with escalating doses of the TFc-SMDC-1.

Supplementary Table S1

Supplementary Table S1 shows a summary of the EC50 values

Supplementary Table S2

Supplementary Table S2 shows a summary of the PK parameters of the cynomolgus study

Data Availability

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

Authors’ Disclosures

D. Carraturo reports grants from the the European Commission during the conduct of the study. F. Gallo reports grants from the European Commission during the conduct of the study; patents for WO2020/216947A1 and WO2019057964A1 issued; and employment with Heidelberg Pharma Research GmbH. M. Schmitt reports grants from the European Commission during the conduct of the study and full-time employment with Heidelberg Pharma Research GmbH and ownership of share in Heidelberg Pharma AG. K. Decker reports grants from the European Commission during the conduct of the study and full-time employment with Heidelberg Pharma AG and ownership of stock in Heidelberg Pharma AG. C. Orlik reports grants from the European Commission during the conduct of the study. A.M. Pahl reports grants from the European Commission during the conduct of the study; patents for WO2022/194988A2, WO2020/216947A1, and WO2019057964A1 issued; and employment with Heidelberg Pharma Research AG and ownership of share in Heidelberg Pharma AG. T. Hechler reports grants from the European Commission during the conduct of the study; patents for WO2022/194988A2, WO2020/216947A1, WO2019057964A1 issued; and full-time employment with Heidelberg Pharma Research GmbH and ownership of share in Heidelberg Pharma AG.

Authors’ Contributions

D. Carraturo: Formal analysis, investigation, visualization, methodology, writing–original draft. F. Gallo: Investigation, methodology, writing–review and editing. M. Schmitt: Data curation, investigation, methodology, writing–review and editing. K. Decker: Formal analysis, investigation, methodology, writing–review and editing. C. Orlik: Data curation, supervision, investigation, methodology, writing–review and editing. A.M. Pahl: Conceptualization, resources, supervision, funding acquisition. T. Hechler: Conceptualization, resources, supervision, funding acquisition, writing–original draft, project administration, writing–review and editing.