Peptidyl-Proline Isomerases (PPIases): Targets for Natural Products and Natural Product-Inspired Compounds

Abstract

Peptidyl-proline isomerases (PPIases) are a chaperone superfamily comprising the FK506-binding proteins (FKBPs), cyclophilins, and parvulins. PPIases catalyze the cis/trans isomerization of proline, acting as a regulatory switch during folding, activation, and/or degradation of many proteins. These "clients" include proteins with key roles in cancer, neurodegeneration, and psychiatric disorders, suggesting that PPIase inhibitors could be important therapeutics. However, the active site of PPIases is shallow, solvent-exposed, and well conserved between family members, making selective inhibitor design challenging. Despite these hurdles, macrocyclic natural products, including FK506, rapamycin, and cyclosporin, bind PPIases with nanomolar or better affinity. De novo attempts to derive new classes of inhibitors have been somewhat less successful, often showcasing the "undruggable" features of PPIases. Interestingly, the most potent of these next-generation molecules tend to integrate features of the natural products, including macrocyclization or proline mimicry strategies. Here, we review recent developments and ongoing challenges in the inhibition of PPIases, with a focus on how natural products might inform the creation of potent and selective inhibitors.

Figure 1

Proline samples discrete cis and trans conformations, which isomerize on the time scale of milliseconds to seconds: (A) depiction of the proline conformations, with the backbone cis and trans orientations highlighted as an orange dotted line; (B) average time scales of processes important in protein folding, illustrating that uncatalyzed proline isomerization can often be a rate-limiting step.

Figure 2

Ribbon diagrams of the PPIase domains of FKBP12, CypA, and Pin1. The client binding pockets are depicted as a mesh cavity (produced in PyMol). PDB ascension codes are 1FKB, 1BCK, 1PIN, respectively. Note that the overall folds are similar, featuring a surface created by β sheets and a short helix.

Figure 3

PPIases have a shallow, broad active site. (A) Surface representation of the PPIase domains from FKBP12, CypA, and Pin1. The active site is shaded in gray, with critical residues shown. (B) High-affinity ligands for each PPIase are shown bound. Surface charges are as follows: blue is positive; red is negative.

Figure 4

Bifunctional binding mode of the natural product rapamycin. (A) Rapamycin forms a ternary complex with FKBP12 and the FRB domain of mTOR. The regions for binding FKBP (red) and mTOR (blue) are shown. (B) The pipecolyl α-ketoamide of rapamycin anchors it into the proline-binding pocket while leaving the triene exposed for interactions with mTOR. (C) Crystal structure of the FKBP12–rapamycin–mTOR ternary complex, showing the rapamycin-mediated organization of the proteins (PDB code 1FAP).

Figure 5

Pin1 is required for conformational switching of phosphoserine-proline. Increased steric bulk and transient backbone interactions of the phosphate group dramatically slow the rate of intrinsic isomerization when compared to the unphosphorylated dipeptide. Pin1 selectively recognizes the negative charge adjacent to the proline and removes the isomerization bottleneck. The cis and trans conformations are illustrated as an orange dotted line.

Figure 6

Design of first high affinity, synthetic ligands for FKBP12. The FKBP12 binding motif of the natural product FK506, 1, is highlighted in red.

Figure 7

Synthetic ligand for FKBP (7, SLF) binds with high affinity. (A) Chemical structure of SLF. (B) Binding of SLF to FKBP52 represented as an electrostatic surface (PDB code 4LAY). (C) Alternative stereoscopic representation of SLF binding to FKBP52 with the X-ray diffraction electron density map shown in yellow. Critical residues surrounding the proline-binding pocket are highlighted in green.

Figure 8

Representative SLF-based chemical inducers of dimerization (CIDs).

Figure 9

Nonimmunosuppressive FKBP12 ligands.

Figure 10

Cyclohexamide derivatives potently inhibit FKBP38.

Figure 11

Structure–activity relationship of FKBP ligands modified in the pyranose region. (A) Crystal structure of FKBP51 bound to SLF. The 40s loop (maroon) and 80s loop (pale green) are highlighted as regions that deviate from FKBP12. (B) Crystal structure of FKBP12 after alignment, with an overlay of SLF from (A). The 80s loop (dark green) is in closer proximity to the active site than in FKBP51/52. (C) Derivatives of the pyranose region to alter binding to the 80s loop and their (D) affinity to FKBPs.

Figure 12

Ligands designed to bind to the mutant FKBP51^F67V had reduced affinity for FKBP51^WT. It was found that FKBP51 could undergo induced fit upon binding, resulting in the discovery of high-affinity, selective inhibitors.

Figure 13

Optimization of specific FKBP51 inhibitors. Replacement of the ester moiety with geminally substituted amides retained affinity and selectivity for FKBP51.

Figure 14

Cyclosporin A (32, CsA) and representative derivatives. Modifications to the CsA scaffold were designed to prevent calcineurin binding by disrupting residues important for that interaction or block membrane permeability as in 36 (red).

Figure 15

Synthetic ligands designed to be selective for CypD.

Figure 16

Indanyl ketones designed for selective binding to cyclophilin paralogs. IC₅₀ values in micromolar.

Figure 17

Results of the computational design of CypA ligands based on a diarylurea pharmacophore. A crystal structure of this scaffold forms critical hydrogen bonds with Gln-105 and Asn-144 (dotted yellow line, PDB code 4XNC).

Figure 18

Expanding the urea pharmacophore produced low nanomolar CypA inhibitors.

Figure 19

Juglone is a natural product that covalently modifies Pin1.

Figure 20

Representative pipecolic acid-based phosphopeptide inhibitor of Pin1.

Figure 21

Cyclic peptide Pin1 inhibitors were modified by the appendage of a membrane permeable octaarginine sequence using a disulfide linker. Reduction in the cytosol releases the free inhibitor.

Figure 22

Pin1 cyclic peptides were modified to yield cell-active bicyclic derivatives.

Figure 23

High-affinity Pin1 inhibitors discovered using phage panning. Importantly, these molecules lack the electronegative groups that have hindered membrane permeability and cellular activity of other scaffolds.

Figure 24

Small molecule Pin1 inhibitors based on pipecolic acid. Note the resemblance to FKBP inhibitors, such as SLF.

Figure 25

Biarylamides potently bind Pin1. Cocrystallization with 65c (PDB code 3IKG) revealed a new binding orientation with the phenyl ring oriented in the proline-binding pocket.

Figure 26

Derivatives of 64a were explored to investigate binding in the proline pocket.

Figure 27

Extensions of the phenyl ring were designed to increase rigidity and stabilize binding in the proline pocket.

Figure 28

A series of benzimidazoles mimic the linker length extension and potently inhibit Pin1. The most potent derivative, 63c, binds to Pin1 in a similar orientation as 59c. (PDB code 4TYO).

Figure 29

Dihydrothiazole derivatives of the benzimidazole inhibitors were synthesized to avoid P-glycoprotein efflux. Compounds had high-affinity for Pin1 and prevented cell cycle progression in cells. All IC₅₀ values are in micromolar.

Figure 30

Early fragment evolution of Pin1 inhibitors.

Figure 31

Pin1 fragments designed to engage the hydrophobic shelf adjacent to the proline pocket.

Figure 32

Second-generation fragments for Pin1.

Figure 33

Pin1 fragments expanded into the phosphate-binding pocket exhibit a new mode of binding that flanks Lys-63 and Arg-69 (PDB code 2XPB).