Discovery of Glutamate Carboxypeptidase III Ligands to Compete the Uptake of [177Lu]Lu-PSMA-617 in Healthy Organs.

Prostate-specific membrane antigen (PSMA)-targeted radio ligand therapeutics (RLTs), such as [177Lu]Lu-PSMA-617 (Pluvicto), have been shown to accumulate in salivary glands and kidneys, potentially leading to undesired side effects. As unwanted accumulation in normal organs may derive from the cross-reactivity of PSMA ligands to glutamate carboxypeptidase III (GCPIII), it may be convenient to block this interaction with GCPIII-selective ligands. Parallel screening of a DNA-encoded chemical library (DEL) against GCPIII and PSMA allowed the identification of GCPIII binders. Structure-activity relationship (SAR) studies resulted in the identification of nanomolar GCPIII ligands with up to 1000-fold selectivity over PSMA. We studied the ability of GCPIII ligands to counteract the binding of [177Lu]Lu-PSMA-617 to human salivary glands by autoradiography and could demonstrate a partial radioprotection.

DNA-encoded chemical libraries enable the discovery of potent PSMA-ligands with substantially reduced affinity towards the GCPIII anti-target.

Prostate-specific membrane antigen (PSMA) is a tumor-associated protein that has been successfully targeted with small organic ligands and monoclonal antibodies. Pluvicto™ is a PSMA-targeted radioligand therapeutic (RLT) recently approved by the FDA for the treatment of metastatic castration-resistant prostate cancer (2022 FDA marketing authorization). Although a large Phase III clinical trial (VISION trial) demonstrated clinical benefits in patients treated with Pluvicto™, the therapeutic window of the drug is narrowed by its undesired accumulation in healthy organs. Glutamate carboxypeptidase III (GCPIII), an enzyme sharing 70% identity with PSMA, may be responsible for the off-target accumulation of PSMA-RLTs in salivary glands and kidneys. In this work, we designed and synthesized affinity and selectivity maturation DNA-encoded chemical libraries (ASM-DELs) comprising 18'284'658 compounds that were screened in parallel against PSMA and GCPIII with the aim to identify potent and selective PSMA ligands for tumor-targeting applications. Compound A70-B104 was isolated as the most potent and selective ligand (KD of 900 pM for PSMA, KD of 40 nM for GCPIII). 177Lu-A70-B104-DOTA, a radiolabeled derivative of compound A70-B104, presented selective accumulation in PSMA-positive cancer lesions (i.e., 7.4% ID g-1, 2 hour time point) after systemic administration in tumor-bearing mice. The results of autoradiography experiments showed that 177Lu-A70-B104-DOTA selectively binds to PSMA-positive cancer tissues, while negligible binding on human salivary glands was observed.

Impact of library input on the hit discovery rate in DNA-encoded chemical library selections.

DNA-encoded chemical libraries (DELs) are powerful drug discovery tools, enabling the parallel screening of millions of DNA-barcoded compounds. We investigated how the DEL input affects the hit discovery rate in DEL screenings. Evaluation of selection fingerprints revealed that the use of approximately 105 copies of each library member is required for the confident identification of nanomolar hits, using generally applicable methodologies.

Structural insights into the interaction of three Y-shaped ligands with PI3Kα.

Class IA phosphoinositide 3-kinase alpha (PI3Kα) is an important drug target because it is one of the most frequently mutated proteins in human cancers. However, small molecule inhibitors currently on the market or under development have safety concerns due to a lack of selectivity. Therefore, other chemical scaffolds or unique mechanisms of catalytic kinase inhibition are needed. Here, we report the cryo-electron microscopy structures of wild-type PI3Kα, the dimer of p110α and p85α, in complex with three Y-shaped ligands [cpd16 (compound 16), cpd17 (compound 17), and cpd18 (compound 18)] of different affinities and no inhibitory effect on the kinase activity. Unlike ATP-competitive inhibitors, cpd17 adopts a Y-shaped conformation with one arm inserted into a binding pocket formed by R770 and W780 and the other arm lodged in the ATP-binding pocket at an angle that is different from that of the ATP phosphate tail. Such a special interaction induces a conformation of PI3Kα resembling that of the unliganded protein. These observations were confirmed with two isomers (cpd16 and cpd18). Further analysis of these Y-shaped ligands revealed the structural basis of differential binding affinities caused by stereo- or regiochemical modifications. Our results may offer a different direction toward the design of therapeutic agents against PI3Kα.

Deep Learning Approach for the Discovery of Tumor-Targeting Small Organic Ligands from DNA-Encoded Chemical Libraries.

DNA-Encoded Chemical Libraries (DELs) have emerged as efficient and cost-effective ligand discovery tools, which enable the generation of protein-ligand interaction data of unprecedented size. In this article, we present an approach that combines DEL screening and instance-level deep learning modeling to identify tumor-targeting ligands against carbonic anhydrase IX (CAIX), a clinically validated marker of hypoxia and clear cell renal cell carcinoma. We present a new ligand identification and hit-to-lead strategy driven by machine learning models trained on DELs, which expand the scope of DEL-derived chemical motifs. CAIX-screening datasets obtained from three different DELs were used to train machine learning models for generating novel hits, dissimilar to elements present in the original DELs. Out of the 152 novel potential hits that were identified with our approach and screened in an in vitro enzymatic inhibition assay, 70% displayed submicromolar activities (IC50 < 1 µM). To generate lead compounds that are functionalized with anticancer payloads, analogues of top hits were prioritized for synthesis based on the predicted CAIX affinity and synthetic feasibility. Three lead candidates showed accumulation on the surface of CAIX-expressing tumor cells in cellular binding assays. The best compound displayed an in vitro KD of 5.7 nM and selectively targeted tumors in mice bearing human renal cell carcinoma lesions. Our results demonstrate the synergy between DEL and machine learning for the identification of novel hits and for the successful translation of lead candidates for in vivo targeting applications.

A DNA-encoded chemical library based on chiral 4-amino-proline enables stereospecific isozyme-selective protein recognition.

DNA-encoded chemical libraries (DELs) consist of large chemical compound collections individually linked to DNA barcodes, facilitating pooled construction and screening. However, screening campaigns often fail if the molecular arrangement of the building blocks is not conducive to an efficient interaction with a protein target. Here we postulated that the use of rigid, compact and stereo-defined central scaffolds for DEL synthesis may facilitate the discovery of very specific ligands capable of discriminating between closely related protein targets. We synthesized a DEL comprising 3,735,936 members, featuring the four stereoisomers of 4-aminopyrrolidine-2-carboxylic acid as central scaffolds. The library was screened in comparative selections against pharmaceutically relevant targets and their closely related protein isoforms. Hit validation results revealed a strong impact of stereochemistry, with large affinity differences between stereoisomers. We identified potent isozyme-selective ligands against multiple protein targets. Some of these hits, specific to tumour-associated antigens, demonstrated tumour-selective targeting in vitro and in vivo. Collectively, constructing DELs with stereo-defined elements contributed to high library productivity and ligand selectivity.

Cross-reactivity to glutamate carboxypeptidase III causes undesired salivary gland and kidney uptake of PSMA-targeted small-molecule radionuclide therapeutics.

PURPOSE: Recently, Pluvicto™ ([177Lu]Lu-PSMA-617), a small-molecule prostate-specific membrane antigen (PSMA) radioligand therapeutic, has been approved by the FDA in metastatic castration-resistant prostate cancer. Pluvicto™ and other PSMA-targeting radioligand therapeutics (RLTs) have shown side effects due to accumulation in certain healthy tissues, such as salivary glands and kidney. Until now, the molecular mechanism underlying the undesired accumulation of PSMA-targeting RLTs had not been elucidated. METHODS: We compared the sequence of PSMA with the entire human proteome to identify proteins closely related to the target. We have identified glutamate carboxypeptidase III (GCPIII), N-acetylated alpha-linked acidic dipeptidase like 1 (NAALADL-1), and transferrin receptor 1 (TfR1) as extracellular targets with the highest similarity to PSMA. The affinity of compound 1 for PSMA, GCPIII, NAALADL-1, and TfR1 was measured by fluorescence polarization. The expression of the putative anti-target GCPIII was assessed by immunofluorescence on human salivary glands and kidney, using commercially available antibodies. RESULTS: A fluorescent derivative of Pluvicto™ (compound 1) bound tightly to PSMA and to GCPIII in fluorescence polarization experiments, while no interaction was observed with NAALADL-1 and TfR1. Immunofluorescence analysis revealed abundant expression of GCPIII both in healthy human kidney and salivary glands. CONCLUSION: We conclude that the membranous expression of GCPIII in kidney and salivary gland may be the underlying cause for unwanted accumulation of Pluvicto™ and other Glu-ureido PSMA radio pharmaceuticals in patients.

Generation and in vivo validation of an IL-12 fusion protein based on a novel anti-human FAP monoclonal antibody.

BACKGROUND: In this study, we describe the generation of a fully human monoclonal antibody (named '7NP2') targeting human fibroblast activation protein (FAP), an antigen expressed in the microenvironment of different types of solid neoplasms. METHODS: 7NP2 was isolated from a synthetic antibody phage display library and was improved by one round of mutagenesis-based affinity maturation. The tumor recognition properties of the antibody were validated by immunofluorescence procedures performed on cancer biopsies from human patients. A fusion protein consisting of the 7NP2 antibody linked to interleukin (IL)-12 was generated and the anticancer activity of the murine surrogate product (named mIL12-7NP2) was evaluated in mouse models. Furthermore, the safety of the fully human product (named IL12-7NP2) was evaluated in Cynomolgus monkeys. RESULTS: Biodistribution analysis in tumor-bearing mice confirmed the ability of the product to selectively localize to solid tumors while sparing healthy organs. Encouraged by these results, therapy studies were conducted in vivo, showing a potent antitumor activity in immunocompetent and immunodeficient mouse models of cancer, both as single agent and in combination with immune checkpoint inhibitors. The fully human product was tolerated when administered to non-human primates. CONCLUSIONS: The results obtained in this work provided a rationale for future clinical translation activities using IL12-7NP2.

Isolation of a Natural Killer Group 2D Small-Molecule Ligand from DNA-Encoded Chemical Libraries.

Natural Killer Group 2D (NKG2D) is a homo-dimeric transmembrane protein which is typically expressed on the surface of natural killer (NK) cells, natural killer T (NKT) cells, gamma delta T (γδT) cells, activated CD8 positive T-cells and activated macrophages. Bispecific molecules, capable of bridging NKG2D with a target protein expressed on the surface of tumor cells, may be used to redirect the cytotoxic activity of NK-cells towards antigen-positive malignant T-cells. In this work, we report the discovery of a novel NKG2D small molecule binder [KD =(410±60) nM], isolated from a DNA-Encoded Chemical Library (DEL). The discovery of small organic NKG2D ligands may facilitate the generation of fully synthetic bispecific adaptors, which may serve as an alternative to bispecific antibody products and which may benefit from better tumor targeting properties.

Specific Inhibitor of Placental Alkaline Phosphatase Isolated from a DNA-Encoded Chemical Library Targets Tumor of the Female Reproductive Tract.

Placental alkaline phosphatase (PLAP) is an abundant surface antigen in the malignancies of the female reproductive tract. Nevertheless, the discovery of PLAP-specific small organic ligands for targeting applications has been hindered by ligand cross-reactivity with the ubiquitous tissue non-specific alkaline phosphatase (TNAP). In this study, we used DNA-encoded chemical libraries to discover a potent (IC50 = 32 nM) and selective PLAP inhibitor, with no detectable inhibition of TNAP activity. Subsequently, the PLAP ligand was conjugated to fluorescein; it specifically bound to PLAP-positive tumors in vitro and targeted cervical cancer in vivo in a mouse model of the disease. Ultimately, the fluorescent derivative of the PLAP inhibitor functioned as a bispecific engager redirecting the killing of chimeric antigen receptor-T cells specific to fluorescein on PLAP-positive tumor cells.

Large screening of DNA-compatible reaction conditions for Suzuki and Sonogashira cross-coupling reactions and for reverse amide bond formation.

Progress in DNA-encoded chemical library synthesis and screening crucially relies on the availability of DNA-compatible reactions, which proceed with high yields and excellent purity for a large number of possible building blocks. In the past, experimental conditions have been presented for the execution of Suzuki and Sonogashira cross-coupling reactions on-DNA. In this article, our aim was to optimize Suzuki and Sonogashira reactions, comparing our results to previously published procedures. We have tested the performance of improved conditions using 606 building blocks (including boronic acids, pinacol boranes and terminal alkynes), achieving >70% conversion for 84% of the tested molecules. Moreover, we describe efficient experimental conditions for the on-DNA synthesis of amide bonds, starting from DNA derivatives carrying a carboxylic acid moiety and 300 primary, secondary and aromatic amines, as amide bonds are frequently found in DNA-encoded chemical libraries thanks to their excellent DNA compatibility.

Affinity Selections of DNA-Encoded Chemical Libraries on Carbonic Anhydrase IX-Expressing Tumor Cells Reveal a Dependence on Ligand Valence.

DNA-encoded chemical libraries are typically screened against purified protein targets. Recently, cell-based selections with encoded chemical libraries have been described, commonly revealing suboptimal performance due to insufficient recovery of binding molecules. We used carbonic anhydrase IX (CAIX)-expressing tumor cells as a model system to optimize selection procedures with code-specific quantitative polymerase chain reaction (qPCR) as selection readout. Salt concentration and performing PCR on cell suspension had the biggest impact on selection performance, leading to 15-fold enrichment factors for high-affinity monovalent CAIX binders (acetazolamide; KD =8.7 nM). Surprisingly, the homobivalent display of acetazolamide at the extremities of both complementary DNA strands led to a substantial improvement of both ligand recovery and enrichment factors (above 100-fold). The optimized procedures were used for selections with a DNA-encoded chemical library comprising 1 million members against tumor cell lines expressing CAIX, leading to a preferential recovery of known and new ligands against this validated tumor-associated target. This work may facilitate future affinity selections on cells against target proteins which might be difficult to express otherwise.

Stereo- and regiodefined DNA-encoded chemical libraries enable efficient tumour-targeting applications.

The encoding of chemical compounds with amplifiable DNA tags facilitates the discovery of small-molecule ligands for proteins. To investigate the impact of stereo- and regiochemistry on ligand discovery, we synthesized a DNA-encoded library of 670,752 derivatives based on 2-azido-3-iodophenylpropionic acids. The library was selected against multiple proteins and yielded specific ligands. The selection fingerprints obtained for a set of protein targets of pharmaceutical relevance clearly showed the preferential enrichment of ortho-, meta- or para-regioisomers, which was experimentally verified by affinity measurements in the absence of DNA. The discovered ligands included novel selective enzyme inhibitors and binders to tumour-associated antigens, which enabled conditional chimeric antigen receptor T-cell activation and tumour targeting.

A Single-Stranded DNA-Encoded Chemical Library Based on a Stereoisomeric Scaffold Enables Ligand Discovery by Modular Assembly of Building Blocks.

A versatile and Lipinski-compliant DNA-encoded library (DEL), comprising 366 600 glutamic acid derivatives coupled to oligonucleotides serving as amplifiable identification barcodes is designed, constructed, and characterized. The GB-DEL library, constructed in single-stranded DNA format, allows de novo identification of specific binders against several pharmaceutically relevant proteins. Moreover, hybridization of the single-stranded DEL with a set of known protein ligands of low to medium affinity coupled to a complementary DNA strand results in self-assembled selectable chemical structures, leading to the identification of affinity-matured compounds.

Complexation with a Cognate Antibody Fragment Facilitates Affinity Measurements of Fluorescein-Linked Small Molecule Ligands.

The availability of reliable methods for the characterization of the binding of small molecule ligands to protein targets is crucially important for drug discovery. We have adapted a method, routinely used for the characterization of monoclonal antibodies (enzyme-linked immunosorbent assay, or "ELISA"), to small molecule ligands, using fluorescein conjugates and antifluorescein antibodies as detection reagents. The new small molecule-ELISA methodology was tested using a panel of binders specific to carbonic anhydrase II, with dissociation constants ranging between 6 µM and 14 nM. An excellent agreement was found between ELISA measurements and fluorescence polarization results. The methodology was also extended to BIAcore measurements and implemented for ligands coupled to oligonucleotides. Small molecule-ELISA procedures are particularly useful in the context of DNA-encoded libraries, for which hit validation procedures need to be performed on dozens of candidate molecules and hit compounds can be conveniently resynthesized on DNA.

Impact of Ligand Size and Conjugation Chemistry on the Performance of Universal Chimeric Antigen Receptor T-Cells for Tumor Killing.

All Universal Chimeric Antigen Receptor T-cells (UniCAR T-cells) are T-cells which have been engineered to recognize a haptenated ligand. Due to this feature, UniCAR T-cells have the potential to mediate a potent and selective tumor killing only in the presence of a haptenated tumor ligand, thus avoiding the long-lasting biocidal effects of conventional CAR T-cells. We have used fluorescein-labeled versions of small organic ligands and different antibody formats specific to carbonic anhydrase IX (a tumor-associated antigen) in order to assess whether the killing potential of UniCAR T-cells depended on the molecular features of the haptenated molecule. Both small molecule ligands and larger antibody fragments were potent in mediating tumor cell killing over a broad concentration range. Antibodies could be conveniently used both in IgG format and as smaller diabody fragments. Importantly, the use of site-specific chemical modification strategies for the antibody coupling to fluorescein led to a substantial improvement of tumor cell killing performance, compared to the random modification of primary amino groups on the antibody surface.

Comparative evaluation of DNA-encoded chemical selections performed using DNA in single-stranded or double-stranded format.

DNA-encoded chemical libraries (DEL) are increasingly being used for the discovery and optimization of small organic ligands to proteins of biological or pharmaceutical interest. The DNA fragments, that serve as amplifiable identification barcodes for individual compounds in the library, are typically used in double-stranded DNA format. To the best of our knowledge, a direct comparison of DEL selections featuring DNA in either single- or double-stranded DNA format has not yet been reported. In this article, we describe a comparative evaluation of selections with two DEL libraries (named GB-DEL and NF-DEL), based on different chemical designs and produced in both single- and double-stranded DNA format. The libraries were selected in identical conditions against multiple protein targets, revealing comparable and reproducible fingerprints for both types of DNA formats. Surprisingly, selections performed with single-stranded DNA barcodes exhibited improved enrichment factors compared to double-stranded DNA. Using high-affinity ligands to carbonic anhydrase IX as benchmarks for selection performance, we observed an improved selectivity for the NF-DEL library (on average 2-fold higher enrichment factors) in favor of single-stranded DNA. The enrichment factors were even higher for the GB-DEL selections (approximately 5-fold), compared to the same library in double-stranded DNA format. Collectively, these results indicate that DEL libraries can conveniently be synthesized and screened in both single- and double-stranded DNA format, but single-stranded DNA barcodes typically yield enhanced enrichment factors.

Screening of copper and palladium-mediated reactions compatible with DNA-encoded chemical libraries.

The construction of DNA-encoded chemical libraries (DECLs) crucially relies on the availability of chemical reactions, which are DNA-compatible and which exhibit high conversion rates for a large number of diverse substrates. In this work, we present our optimization and validation procedures for three copper and palladium-catalyzed reactions (Suzuki cross-coupling, Sonogashira cross-coupling and copper(I)-catalyzed alkyne-azide cycloaddition (CuAAC)), which have been successfully used by our group for the construction of large encoded libraries.