Influencias literarias

A key interaction between Asp101 of HDAC8 and the peptide substrate was revealed in the crystal structure, Asp101 is conserved throughout class I and II HDACs and sits on the flexible L2 loop (which is often crystallographcally unresolved) and

(a)

M274

Tyr306

Phe152 Phe205

(b)

Tyr306 Phe152

Met274 Phe208

Asp101

Figure 3.5: (a) Manual docking pose of 92 in HDAC8 and (b) HDAC8 structural overlay with Fluor-de-Lys substrate bound and interactions highlighted with Asp101 on the L2 loop (purple).

has been shown to be essential for substrate binding⁸⁴. Asp101 facilitates substrate binding by acting as a clamp; upon association of the peptide ligand, the flexible loop can close and Asp101 forms hydrogen bonds with the backbone amide NH groups either side of the residue occupying the substrate tunnel (the acetylated lysine to be de-acetylated), this locks the peptide backbone of the substrate in a cis conformation (see Figure 3.5b), thereby reducing kof f and in turn Ki (Ki = kof f/kon). Asp101 and The L2 binding loop are unresolved in the 4-HDAC8 structure, this indicates that no interactions are occuring between the L2-loop and α-amino amide inhibitor 4. It was hypothesised therefore that introducing a hydrogen bond to Asp101 from the inhibitor would provide an improvement in affinity over inhibitors without a hydrogen bond to Asp101.

Introduction of hydrogen bond acceptors at the distal end of the STBG has a detri-mental effect on the binding of the inhibitor. 69 with a p-methoxy group has IC50

= 16.6 ±3.19µM, 5-fold higher than 68 with no methoxy group (IC50 = 4.28 ±0.18 µM), this equates to a binding energy difference of 0.83 ±0.34 kcal mol⁻¹(Table 3.2).

It is possible that this is due to a repulsion between the δ^−ve p-methoxy oxygen and the neagtively charged Asp101 as seen by manual docking (Figure 3.6a).

(a)

kcal.mol^-1 -6.79 ±0.19 kcal.mol^-1

Figure 3.6: (a) Manual docking poses of 69, (b) 72 in the HDAC8 structure. Flexible L2 loop and Asp101 (purple) are overlaid from substrate bound structure. (c) energy differences upon the addition of a hydrogen bond donor to Asp101, red values indicate loss in binding energy, green values indicate increase in binding energy. The red circles represent the distance over which Asp101 effects are seen.

Repulsion of the hydroxyl lone pair is supported by improved binding of 72 compared to 69. Manual docking of 72, a compound with a STBG which maintains the p-methoxy substitution but has a shortened linker was performed. In this docking the methoxy group is 7 ˚A from Asp101 (Figure 3.6b) such that the Asp101-p-methoxy

repulsion cannot take place. The binding energy of 72 is -7.91 ±0.17 kcal mol⁻¹ which is 1.12 ±0.26 kcal mol⁻¹ better than that of 69 which has a repulsive Asp101-p-methoxy interaction (3.6c).

The p-hydroxyphenyl STBG containing 77 has two-fold lower IC₅₀ and average binding energy 0.4 ±0.5 kcal mol⁻¹ better than its phenyl counterpart 86 (Figure Table 3.2). The hydroxyl group could potentially contribute benficial contacts to the protein-ligand interaction. Hydroxyl groups can be both hydrogen bond donors through their hydrogen, or acceptors through the lone pair on the oxygen. Manual docking of two possible poses of 77 into the HDAC8 structure reveals a possible interaction with Lys33 near the entrance to the substrate tunnel, which may con-tribute a hydrogen bond (Figure 3.7a), alternatively a hydrogen bond may be being donated to Asp101 (Figure 3.7b). An edge to face aromatic stacking interaction could form between the aromatic ring of the STBG and Phe152 if an interaction were to occur between the OH of 77 and Lys33 (Figure 3.7a). Lys33 is a dynamic residue, its Cα has been seen to move by more than 0.5 nm when different inhibitors are bound to HDAC8⁹².

A second hydroxyl group was introduced to assess the effect a dihydroxylated aro-matic ring would have on the cap group interactions (93). The binding energies of both mono- and dihydroxylated inhibitors (77 and 93) were well within the errors of each other and so the effect of the second hydroxy group is not measureable in this scenario (Table 3.2).

A heteroaromatic pyridine ring was tested as a cap group to see if this would alter binding at the entrance of the substrate tunnel (91). The pKa of the 4-methylpyridinium ion is 5.98²¹² and so the pyridine nitrogen atom will be deproto-nated at pH 8.0 (the pH of the fluorescent assay¹⁹²). The IC50 of 91 is 3.85 ±0.92 µM corresponding to a binding energy of -7.59 ±0.24 kcal mol⁻¹ which is very sim-ilar to that of the equivalent phenyl ringed inhibitor (86) so the introduction of the piperazine ring has little to no effect on the interactions of the aromatic ring.

Indole STBG containing 78 was also considerably potent of this second generation of α-amino amide compounds exploring the variability STBG (IC50 = 0.97 ±0.34 µM,

(a)

Lys33

Tyr306

Met274 Phe208

Phe152 Asp101

(b)

Figure 3.7: Two manually docked poses of 77 in HDAC8-4 structure (teal) that would al-low for 77 to be (a) a hydrogen bond acceptor from Lys33 and form aromatic stacking with Phe152, and (b) a hydrogen-bond donor to Asp101.

binding energy = -8.53 ±0.35 kcal mol⁻¹, Table 3.2). Manual docking places the indole NH to provide a hydrogen bond to Asp101 (Figure 3.8a) in a similar manner as was seen for 77 above (Figure 3.7). This manual docking pose of a low energy confomation of 78 with its Asp101 interaction also positions the six-membered aro-matic ring for an energetically favourable T-stacking interaction with Phe152. The suggestion that hydrogen bonding interactions to Asp101 could be promoting tight binding in two low IC50 α-amino amide inhibitors 78 and 77 suggested that the interaction could be exploited in the design of subsequent inhibitors. Addition of a hydroxyl group at the 5-position of the indole group of 78 to give 94 had little to no effect on the binding energy of 94 compared to 78 (-8.16 ±0.27 kcal mol⁻¹ vs.

-8.53 ±0.35 kcal mol⁻¹, Table Table 3.2).

The carbazole inhibitor 85 (Figure 3.8b) was designed with previous observations on conformational flexibility and hydrogen bonding with Asp101 in mind. A loss in binding energy of 1.86 ±0.33 kcal mol⁻¹ was observed by replacement of the rigid 5-membered ring with a methylene linker (4→68). This was attributed in the most part to increased conformational flexibility of the unbound inhibitor 68. Compounds 78, 77, 93 have methylene linkers with three rotatable bonds, but with improved

(a)

Asp101

Tyr306

Met274 Phe205

Phe152

(b)

Asp101

Tyr306

Met274 Phe205

Phe152

Figure 3.8: Manual docking poses of (a) Indole 78 and (b) carbazole 85 in two amide con-formations, cis-(yellow) and trans-(magenta) the HDAC8-4 structure (teal).

Flexible L2 loop and H-bond-donating Asp101 shown are overlaid from the HDAC8-Fluor-de-Lys structure (purple).

binding energy over 86 due to hydrogen bonding with Asp101. This suggests that the hydrogen bonding contribution is greater than the energy loss due to conformational flexibility. In order to lower the conformatioal flexibility of 78 the methylene linker was replaced with a rigid aromatic ring. It is known that the substrate tunnel can accommodate bulky aromatic rings close to the ZBG⁹⁰ as seen in Figure 2.10a.

Manual docking of carbazole 85 into the HDAC8 structure reveals the reduction in conformational flexibility actually restricts the positioning of the hydrogen bond donating nitrogen to being >5.0 ˚A from the Asp101 acceptor (magenta in Figure 3.8b). The less energetically favourable cis-amide (yellow in Figure 3.8b) allows the carbazole nitrogen to approach 3.4 ˚A from Asp101, all poses of the cis-amide cause steric clashes with Phe208. The lowering of entropy was therefore cancelled out by the clashes introduced and the loss of hydrogen bonding with Asp101. The result is a significant reduction in potency (IC₅₀ = 10.5 ±6.75 µM) and in binding energy (1.46 ±0.72 kcal mol⁻¹), compared to indole 78.

STBG

NH_2.HCl O

Cl Cl

1x10^-7

1

x

10^-6

1 x 10^-5

1 x 10^-4

1 x 10^-3 IC50/M

N

NH

NH O

NH NH

NH O

NH HO

NH HN

NH S

NH NH

OH NH

HN NH

NH N

NH HN

HO

NH

NH HO

HO

4 68 69 70 71 72 77 78 80 81 84 95 85 86 91 92 93 94

Figure 3.9: IC₅₀ values of 2nd generation inhibitors with varying STBG, error bars are SD of >2 independent repeats. corresponding R-groups are shown above the bars.

3.4 Summary of 2

generation inhibitor assay