Development of new theranostic probes targeting PSMA: computational strategies to explore ligand-target interactions

Motivation Prostate-specific membrane antigen (PSMA) is a zinc-metallopeptidase localized at the plasma membrane and it is highly expressed in prostate cancer cells. Thus PSMA represents an attractive target for diagnosis and therapy of prostate cancer but also in several other types of tumors[1-2]. Great efforts in the field of radiochemistry have resulted in different PSMA targeting radiopharmaceuticals for positron emission tomography (PET) as small molecules based on the chemical structure of the target natural substrate N-acetylaspartylglutamate, NAAG, which binds to the extracellular domain of PSMA[3]. These small molecules have several advantages compared to the previously developed anti-PSMA monoclonal antibodies: they are characterized by rapid clearance from the blood and non-target tissue, resulting in low background activity. High image quality can be achieved by enhanced tumor uptake and retention by substrate internalization in the cell through endocytosis. Currently, the most commonly used PSMA PET radioligand is [68Ga]-PSMA-11, nevertheless, the short half-life and the limited amount of Gallium-68 radioactivity from generators have led to a steep increase in the development of [18F]-labelled PSMA ligands, best suited to meet the growing clinical demand. Among these, [18F]DCFPyL, [18F]PSMA-1007 and the next to approval [18F]-JK-PSMA-7 have already made a decisive contribution by spreading PSMA PET technology. However, in the last years, there have been reports of eventual occurrences of PSMA PET-negative prostate cancers, as well as off-target bindings (i.e., in healthy bone tissue by [18F]PSMA-1007). Based on the state of the art described above and aiming to derive useful information to address the issue of PSMA-radiotracer non-responders (with a negative PSMA PET despite the presence of cancer) and still-responders (with an off-target positive PSMA PET), the present contribution investigates the biomolecular binding process of the three fluorinated radiotracers mostly used in clinical practice, fully defining their binding mechanism at molecular level. Methods From the 81 PSMA crystallographic data available in PDB, 3D7H was selected for the subsequent computational studies. Ligand structures were built and energy minimized using RDKit and XTB software. Docking studies were performed by Autodock Vina, using Vina and Vinardo scoring functions. The non-bonded model for metal ions was used in molecular dynamics simulation and MMPBSA methods were applied to calculate binding affinities and interaction profiles, within Amber 2022. Results Structural characterization of PSMA has shown the high affinity of its natural substrate for negatively charged amino acids such as aspartate and glutamate at S1′ pocket, where are located also Arg534, Arg536 and Arg463, so-called arginine patch[4]. Our preliminary results from molecular dynamics studies, highlight the fluorinated radiotracers have the same ability to target this specific sub-region of the enzyme, plus a common disposition of their pharmacophoric portion Lys-Urea-Glu, deeply inserted in the active site of the PSMA and steadily stuck by some H-bonds interactions with the surrounding residues. The remaining portion of each radiotracer, depending on its peculiar chemical features, determines the unicity of its binding and thus of its molecular interactions. Concerning [18F]DCFPyL and [18F]-JK-PSMA-7, ligands differing only by a methoxy substituent on the pyridine ring, the results obtained from MD suggest a partial common pattern of interactions, where hydrophobic contributions are more markedly present for [18F]-JK-PSMA-7. Thanks to these latest contributions, its ligand binding affinity is more favorable than that of [18F]DCFPyL. The energetic interaction profile of binding free energy decomposition shows for each ligand, the highest energetic contribution is given by the arginine patch, a moderate contribution is provided by some Asp and Glu residues and a weak contribution from Asn residues. Interestingly, PSMA presents SNPs at the arginine patch and at some other residues involved in these radiotracers interactions[5]. Thus, the data obtained in this study and presented here could potentially be related to the different radiotracers' clinical responses (radiotracer not-responders and radiotracer still-responders) based on the genetic features of the patient. For PSMA wild type and clinically relevant mutations, the binding and detaching constants k-on and k-off for each drug-PSMA complex will be calculated, based on the recent development of methods like Ligand GaMD (LiGaMD), which simulate biomolecular binding processes and will be experimentally validated.