Synthesis, LSD1 Inhibitory Activity, and LSD1 Binding Model of Optically Pure Lysine-PCPA Conjugates

Compounds that inhibit the catalytic function of lysine-specific demethylase 1 (LSD1) are interesting as therapeutic agents. Recently, we identified three lysine-phenylcyclopropylamine conjugates, NCD18, NCD25, and NCD41, which are potent LSD1 inactivators. However, in our previous study, because we tested those compounds as mixtures of (1S,2R)- and (1R,2S)-disubstituted cyclopropane rings, the relationship between the stereochemistry of the cyclopropane ring and their biological activity remained unknown. In this work, we synthesized optically active compounds of NCD18, NCD25, and NCD41 and evaluated their LSD1 inhibitory activities. In enzyme assays, the LSD1 inhibitory activities of (1R,2S)-NCD18 and (1R,2S)-NCD25 were approximately eleven and four times more potent than those of the corresponding (1S,2R)-isomers, respectively. On the other hand, (1S,2R)-NCD41 was four times more potent than (1R,2S)-NCD41. Binding simulation with LSD1 indicated that the aromatic rings of the compounds and the amino group of the cyclopropylamine were important for the interaction with LSD1, and that the stereochemistry of the 1,2-disubstituted cyclopropane ring affected the position of the aromatic rings and the hydrogen bond formation of the amino group in the LSD1 catalytic site. These findings are expected to contribute to the further development of LSD1 inactivators.

recognition structure, inactivated LSD strongly and selectively. However, they were synthesized by coupling a lysine moiety with racemic trans-PCPA and therefore, optically active NCD compounds were not evaluated for their LSD inhibitory activities. Some LSD research groups have investigated the difference in LSD inhibitory activity between the two enantiomers of PCPA and its derivatives [ 4,22,23]. Following those results, we decided to examine the relationship between the stereochemistry and the LSD inhibitory activities of the NCD series compounds. Here we report the synthesis, the LSD inhibitory activity, and the binding simulation of the optically active NCD compounds, with focus on the stereochemistry of the PCPA moiety.

Experimental Procedures
Melting points were determined using a Yanagimoto micro melting point apparatus, a Büchi 545 melting point apparatus or a Yanaco micro melting point apparatus and were left uncorrected. Proton nuclear magnetic resonance spectra ( H NMR) and carbon nuclear magnetic resonance spectra ( 3 C NMR) were recorded on a BRUKER AVANCE 300 spectrometer in the indicated solvent. Chemical shifts () were reported in parts per million relative to the internal standard, tetramethylsilane. Elemental analysis was performed with a Yanaco CHN CORDER NT-5 analyzer, and all values were within ±0.4% of the calculated values. Fast atom bombardment (FAB) mass spectra were recorded on a JEOL JMS-SX 02A mass spectrometer. HPLC analysis and Preparative HPLC were performed on an ODS-3 ( 50 mm x φ4.6 mm, GL Science or Cosmosil) and an Inertsil ODS-3 (250 mm x φ20 mm, GL Science or Cosmosil), respectively. The HPLC system was composed of a pump (HITACHI, L-6050 intelligent pump) and a detector (HITACHI, L-4000 UV detector) Elution of HPLC chromatogram in synthetic procedure was done with a linear gradient (eluent, 0 min (30 % MeCN/0. % TFA aq.) -2 min (30 % MeCN/0. % TFA aq.) -20 min (70 % MeCN/0. % TFA aq.) -30 min (70 % MeCN/0. % TFA aq.).; flow rate = .0 mL/min). The detection wavelength was 254 nm. Optical rotation of optically active compounds was measured using a HORIBA Scientific, SEPA-300 polarimeter. Reagents and solvents were purchased from Aldrich, Merck, Nacalai Tesque, Tokyo Kasei Kogyo, Wako Pure Chemical Industries, Kishida Kagaku and Kanto Kagaku, and used without purification. Flash column chromatography was performed using Silica Gel 60 (particle size 0.046-0.063 mm) supplied by Merck.

Optically Active Lysine-PCPA conjugates
The LSD inhibition assay was carried out according to the method reported in ref. 2 .
Docking was performed using Molegro Virtual Docker 5.0 software. Coordinates of LSD completed with FAD-N-propargyl lysine peptide adduct were taken from the Brookhaven Protein Data Bank (PDB code 2UXN). Water molecules, cofactors and a peptide substrate were removed, and FAD was converted to a cofactor. The structure of NCD 8, NCD25 and NCD4 bound to LSD was constructed by MolDock, which is based on a heuristic search algorithm that combines differential evolution with a cavity prediction algorithm. The docking parameters were as follow: Grid Resolution: 0.30, Max iterations: 500, Population size: 50, Energy threshold: 00.00, Simplex evolution: 300 (Max steps) and .00 (Neighbour distance factor), Search space: (X, Y, Z) = (65.64, 47.97, 35.60) with radius 0, distance constraints (for N atom of cyclopropylamine of NCD 8, NCD25, NCD4 ): constraint center (X, Y, Z) = (64.5 , 54. , 34.83) with hard constraint between minimum 0 to maximum 2.0.

Results and Discussion
Individual stereoisomers were synthesized as shown in Charts 2 and 3. First, trans-(±) PCPA was resolved into D-tartaric acid salt and the salt was recrystallized. Then, simple deprotonation with sodium hydroxide liberated free ( S,2R)-PCPA. On the other hand, ( R,2S)-PCPA was collected from the mother liquid by extraction with ether, salt formation (salt 2) with L-tartaric acid, recrystallization of the D-tartaric acid salt, and extraction with ether in sequential order. Finally, each enantiomer was reacted with the corresponding mesylate compound 3 [2 ] in the presence of potassium carbonate to give optically pure NCD 8, NCD25, and NCD4 in 7-68% yield.
Although we could not conclude which isomer is more potent, we found that the LSD inhibitory activities of the ( S,2R)-isomer and the ( R,2S)-isomer are dependent on the structure of the R group in the NCD series compounds (Chart 3).
To verify the relationship between the stereochemistry and the LSD inhibitory activity of the NCD series compounds, we performed binding simulation of the compounds to LSD by using Molegro Virtual Docker 5.0 software. The simulation was performed based on the reported X-ray structure of LSD [28] and under the condition that the cyclopropylamine group of NCD 8, NCD25, and NCD4 was fixed to the position where it could react with FAD. The results of the simulation indicated that the interaction with amino acid residues in the three hydrophobic pockets (pockets -3, Figure ) and the formation of one hydrogen bond are important for the LSD inhibitory activity of optically pure LSD inactivators as discussed in more detail below.  Figure 2E). Those interactions with amino acid residues in the hydrophobic pockets may be the reason why ( R,2S)-NCD 8 is more potent than its ( S,2R)-isomer. Next, the modeling of NCD25 binding with LSD was conducted. As shown in Figure 3A and 3C, the binding model of NCD25 was different from that of NCD 8, i.e., the biphenyl group of ( R,2S)-NCD25 could interact with pocket 3 formed by Phe 560, Tyr 807, Pro 808, Ala 809, Thr 8 0 (methyl group), and His 8 2, although the benzyl group could interact with the hydrophobic amino acid residues in pocket 2 as in the case of ( R,2S)-NCD 8 ( Figure  2A and 2C). In the case of ( S,2R)-NCD25, the biphenyl group was positioned in pocket 2, but the benzyl group did not interact with any of the three pockets ( Figure 3D and 3F). ( R,2S)-NCD25 and ( S,2R)-NCD25 could interact with the oxygen atom of Tyr 76 and the carbonyl oxygen of Ala 809 via a hydrogen bond ( Figure 3B and 3E), respectively. Taken together, ( R,2S)-NCD25 could interact with two hydrophobic pockets whereas ( S,2R)-NCD25 could interact with only one hydrophobic pocket. This may be the reason why the LSD inhibitory activity of ( R,2S)-NCD25 is superior to that of ( S,2R)-NCD25. positioned in pocket 2 Instead, the m-trifluoromethylated benzyl group of ( R,2S)-NCD4 was located in pocket 3 ( Figure 4D and 4F) which is too large to appropriately accommodate the unsubstituted benzyl group of ( R,2S)-NCD25. This may be the reason why the inhibitory activity of ( R,2S)-NCD4 is superior to that of ( R,2S)-NCD25.
The interaction of each isomer with amino acid residues in the hydrophobic pockets and the hydrogen bond formation are summarized as follows: (i) ( R,2S)-NCD 8, ( R,2S)-NCD25, and ( S,2R)-NCD4 could interact with amino acid residues in two hydrophobic pockets and form one hydrogen bond, (ii) ( S,2R)-NCD25 could interact with amino acid residues in only one hydrophobic pocket and form one hydrogen bond, (iii) ( S,2R)-NCD 8 could form one hydrogen bond, and (iv) ( R,2S)-NCD4 could interact with amino acid residues in only one hydrophobic pocket. Considering the relationship between the stereochemistry and the LSD inhibitory activity of the NCD series compounds, compounds that could interact with amino acid residues in two pockets and form a hydrogen bond showed high LSD inhibitory activities. In other words, the stereochemistry of the PCPA moiety could affect LSD inhibitory activity through the interaction of the two aromatic groups with hydrophobic amino acid residues and the hydrogen bond formation.

Conclusion
In conclusion, we synthesized optically active ( R,2S)-isomers and ( S,2R)-isomers of NCD 8, NCD25, and NCD4 , and evaluated their LSD inhibitory activities in enzyme assays. The ( R,2S)-isomers of NCD 8 and NCD25 were more potent than their ( S,2R)-isomers. On the other hand, the ( S,2R)-isomer of NCD4 was more potent than its ( R,2S)-isomer. In addition, the binding simulation indicated that the potent NCD series compounds can interact with amino acid residues in the two hydrophobic pockets and form a hydrogen bond in the LSD active site, and that the stereochemistry of PCPA affects LSD inhibitory activity. These findings will be useful for the further development of PCPA-based LSD inhibitors and LSD -targeted therapy.