PLD
April 2023
Phospholipase D (PLD)
Phospholipase D enzymes (PLDs) are phosphodiesterases that produce phosphatidic acid (PA), a key second messenger involved in many cellular processes and precursors for other lipids.
The most abundant forms of human PLD (hPDL) are hPLD1 and hPLD2, regulating distinct functions. The dysfunction of these enzymes is associated with a broad range of pathologies (cancer, cardiovascular, neurodegenerative, and infectious disease), making them attractive small-molecule drug targets.
Drug discovery efforts have led to the discovery of inhibitors of hPLD, both dual (inhibiting hPLD1 and hPLD2) and selective ones. Even if the 3D structure of the bacterial analogue (bPLD) was determined around 20 years ago, solving the structure of human PLD (hPLD) has proven very difficult. In addition, inhibitors of the hPLD do not bind to bPLD, so it was not possible to determine the analogue structure of the protein-ligand complex. The binding mode was, therefore, unknown, preventing structure-based optimization of more effective and selective inhibitors.
Firstly, the scientist determined the crystal structure of hPLD2, and they finally managed to clarify how the known hPLD inhibitors (inhibitor 3 and inhibitor 4) interacted with the enzyme. They found that these ligands bind directly to the catalytic site (orthosteric ligands), and their binding is stabilized by some key interactions with residues in the pocket (Image 1).
Image 1. On the left: 2D structures of known hPLD inhibitors – inhibitor 3 and inhibitor 4. On the right: interaction network of the 3D structure of inhibitor 3 (in light pink) in the hPLD2 catalytic site (orange, PDB: 6OHS). The residues composing the binding site are represented as orange sticks, and the surface of the pocket is in a mesh. The protein-ligand interactions are automatically computed by the 3decision® software.
They observed that in the proximity of the ring A of ligands, there is a deep cavity. They reasoned that they could exploit it by creating additional ligand-protein interactions, and designed compound 5 (Image 2). This maintained the key interactions of the previous inhibitors and even projected its B-ring into the newly discovered cavity. Such rearrangement led to a 5-fold increase in potency compared to inhibitor 3, and 8-fold compared to compound 4. Also, the structure-based designed ligand allowed the stabilization and crystallization of the other enzyme, hPLD1, which structure couldn’t be determined before with the other ligands.
Image 2. On the left: 2D structure of the newly discovered inhibitor 5. On the right: comparison of the binding mode of inhibitor 3 (light pink, PDB: 6OHS) and inhibitor 5 blue, PDB: 6OHR). The B-ring of inhibitor 5 projects into a deep pocket next to the A-ring (on the right) and forms additional contacts with the protein. For clarity, the protein representation of complex with inhibitor 3 is omitted, but the surface of the pocket is depicted, color-coded by hydrophobicity index. Also, for the complex with inhibitor 5, only the residues composing the binding site are represented as blue sticks. The protein-ligand interactions are automatically computed by the 3decision® software.
The determination of both hPLD structures also helped the Biogen scientists to understand and rationalize the selectivity profile of the inhibitors: inhibitor 4 is selective towards hPLD1 over hPLD2, while compounds 3 and 5 are dual inhibitors (not selective). Comparing the binding sites of hPLD1 and hPLD2, they noticed a significant difference between the two hPLDs binding sites: where hPLD1 carries phenylalanine (F614PLD1), on hPLD2, there is a leucine (L514PLD2) (Image 3). This difference causes hPLD1 to have a bigger pocket at that position. All ligands carry a chiral methyl on the linker joining C-ring and the amide. Inhibitor 4 has a geometry that allows easy accommodation of this methyl into the big pocket in PLD1, while the binding with PLD2 is less favored due to the bulkier leucine at this position. On the contrary, inhibitors 3 and 5 are not selective because they do not allocate the same methyl into the same region, and therefore the binding is not affected by this sequence difference. The crucial role of F614/L514 difference was then confirmed by mutant activity assays.
Image 3. Comparison of the binding sites of hPLD2 (in orange, PDB: 6OHS) and hPLD1 (in blue, PDB: 6OHR). The binding poses of inhibitors 3 and 4 are shown (green, PDB: 6OHQ). The methyl on inhibitor 4 gets very close to the L514 compared with F614, which disfavors the binding of this ligand. Inhibitor 3, instead, allocates the methyl in a different region of the binding site and is not affected by the presence of the bulky leucine.The molecular surface of the hPLD2 binding site is depicted for clarity (orange). The picture is produced with the highlight mode of 3decision® software, which only shows the differences among residues in the binding site. This allows to easily identify sequence differences among the compared proteins.
This fundamental study provided the structural basis for understanding inhibitor binding modes and also selectivity profiles. The structural insights gained with this research will push the structure-based design of the next generation of hPLD inhibitors.
Reference:
Metrick CM, Peterson EA, Santoro JC, Enyedy IJ, Murugan P, Chen T, Michelsen K, Cullivan M, Spilker KA, Kumar PR, May-Dracka TL, Chodaparambil JV. Human PLD structures enable drug design and characterization of isoenzyme selectivity. Nat Chem Biol. 2020 Apr;16(4):391-399. doi: https://doi.org/10.1038/s41589-019-0458-4. Epub 2020 Feb 10. PMID: 32042197.