Phosphorylations were also attempted with di-tert-butyl or dicyanoethyl phosphoramidites to produce di-tert-butyl or dicyanoethyl instead of dibenzyl phosphate. Neither of these phosphates was stable on silica gel, and b-elimination products were obtained after chromatography. TFA deprotection of crude di-tert-butyl phosphate, and NH4OH deprotection of crude dicyanoethyl phosphate both gave b-elimination products as well. Thus, the dibenzylphosphate was chosen to carry through to the final products 1 and rac-2. Hydrogenation of the crude dibenzyl phosphate (1S,3R,4R)13 went very slowly, giving a complex crude mixture. Thus, (1S,3R,4R)-13 was purified by reverse-phase semi-preparative high performance liquid chromatography (HPLC). With pure dibenzyl phosphate, hydrogenation at atmospheric pressure worked very well, and gave a very clean final product 1, similar to our experience with a-ketoamides [14].X-ray crystallography
During the synthesis of the inhibitors, Michael addition of tristhiomethyl methide to an a,b-unsaturated ketone 8 produced three stereoisomers of 9, which could not be readily separated (Figure 3). Two diastereomers of a subsequent synthetic intermediate, (1S,3R,4R)-11 and rac-11, were separated by chromatography. Each diastereomer was crystallized, and the relative stereochemistry was determined.

Figure 4. X-ray crystal structures of intermediates (1S,3R,4R)-11 and rac-11 are shown above as displacement ellipsoid drawings (50%). The positional disorder of the benzyl group in rac11 is shown as lighter lines. Hydrogen atoms are omitted for clarity. Structural depiction of the stereochemistries of (1S,3R,4R)-11 and rac11 are shown below each crystal structure. ?after geometry optimization was 3.16 A; with the trans-pyrrolidine torsion angle fixed during geometry optimization, the distance was ?3.67 A (Figure 6).Discussion Stereochemical results of inhibitor synthesisFigure 3. Cyclohexyl ketone inhibitor 1 was synthesized by the method shown. Thermodynamic control in the Michael addition resulted in the anti-Ser-trans-cyclohexyl stereoisomer of 9 as the major product (Figure 4). The chiral center adjacent to the Ser carbonyl was easily epimerized due to the electron-withdrawing effects of both the a-amide and a-ketone, resulting in an enantiomeric mixture of a second diastereomer, rac-9. Because the unnatural D-Thr-the original Ser configuration intact (Figure 4). The minor isomer, rac-11, proved to be a racemic mixture. The absolute configurations were assigned as (1R,3R,4R)-11 and (1S,3S,4S)-11, in which the stereocenter of the Ser analogue was partially epimerized to the syn-Ser-trans-cyclohexyl configuration (Figure 4).Pin1 PPIase Enzyme Assays
The a-chymotrypsin protease-coupled assay was used to evaluate inhibition of Pin1 by compounds 1 and rac-2 with the same substrate concentration as described previously [10,14]. The IC50 values of the two diastereomers were determined to be 260630 mM for 1, and 6168 mM for rac-2. Preincubation with Pin1 for 15 minutes did not result in improved inhibition.

Molecular modeling
Each of the three cyclohexyl ketone inhibitors was docked flexibly, with geometry minimization, into the Pin1 active site. The resulting docked stereoisomers, (1S,3R,4R)-1, (1R,3R,4R)-2, and (1S,3S,4S)-2, are shown in Figure 5. The total energies, Cys113?S–C = O ketone distances, and angles are reported in Table 1. Figure 5. Models of cyclohexyl ketone inhibitors were docked with dynamic minimization. (A) (1S,3R,4R)-1 in orange, (B) (1R,3R,4R)-2 in blue, (C) (1S,3S,4S)-2 in green, and (D) superposition of all atoms of 1 and rac-2. Models were based on PDB 2Q5A [32], and minimized using Sybyl 8.1.1 [42]. Images were prepared using MacPyMol [44]. Figure 6. Pin1 is proposed to stretch the prolyl ring by binding phosphate and C-terminal residues tightly, creating a transpyrrolidine conformation of the substrate and forcing pyramidalization of the prolyl nitrogen in the twisted-amide mechanism. Distance measurements are from calculated structures of AcroH in the ground state and the trans-pyrrolidine transition state.
containing inhibitors were more potent than the L-Thr in work by Zhang et al [32], both diastereomers 1 and rac-2 were tested for Pin1 inhibition. Inhibitor 1, corresponding to the native L-Ser-LPro stereochemistry of Pin1 substrates, had an IC50 value of 260 mM, while rac-2, an enantiomeric mixture of D-Ser-L-Pro and L-Ser-D-Pro analogues, had an IC50 value of 61 mM. Preincubation did not result in improved inhibition, suggesting that they are not slow-binding inhibitors. We obtained a crystal structure of the similarly substituted, reduced amide inhibitor 4 bound in the Pin1 active site, suggesting that the ketones also bind in the active site [27].

conformation, and the trans-pyrrolidine AcroH conformation ?was 0.51 A (Figure 6). This effect of stretching the ring conformation may provide insight into the mechanism of Pin1. In either of the proposed mechanisms: (1) nucleophilic-addition [26], or (2) twisted-amide [25], the nitrogen of the prolyl ring must become pyramidalized and deconjugated from the carbonyl in the transition state [22,24,25]. If binding of substrate to the catalytic site forces the Pro ring into a trans-pyrrolidine conformation, the nitrogen lone pair and the carbonyl p-bond would no longer be conjugated (Figure 6).