Design and Synthesis of Novel Hybrid Benzamide-Peptide Histone Deacetylase Inhibitors

Abstract

We designed and synthesized a series of novel hybrid histone deacetylase inhibitors based on conjugation of benzamide-type inhibitors with either linear or cyclic peptides. Linear tetrapeptides (compounds 13 and 14), cyclic tetrapeptides (compounds 1 and 11), and heptanediamide-peptide conjugates (compounds 10, 12, 15 and 16) were synthesized through on-resin solid-phase peptide synthesis (SPPS). All compounds were found to be moderate HDAC1 and HDAC3 inhibitors, with IC50 values ranging from 1.3 µM to 532 µM. Interestingly, compound 15 showed 19-fold selectivity for HDAC3 versus HDAC1.

Histone deacetylases (HDACs) play important roles in the regulation of gene expression, cell growth, and proliferation, by catalyzing the deacetylation of core histones, tubulin and other proteins.¹ HDAC inhibitors thus have the potential for use in cancer therapy.² Additionally, recent studies point to the potential therapeutic benefit of HDAC inhibitors in neurodegenerative diseases.³ Eighteen HDACs have been identified in humans and are subdivided into four classes: Class I HDACs (HDAC1, HDAC2, HDAC3 and HDAC8^4–7), Class II HDACs (HDAC4, HDAC5, HDAC6, HDAC7, HDAC9 and HDAC10^8–11), Class III HDACs (also known as Sirtuins, Sirt 1–7,¹² which are NAD⁺-dependent enzymes) and the lone Class IV HDAC (HDAC11¹³). Class I, II and IV are Zn⁺⁺–dependent enzymes.¹⁴

Recent studies^{15, 16} have demonstrated that HDAC inhibitors can induce growth arrest of tumor cells by inducing terminal differentiation and apoptosis. Numerous different HDAC inhibitors (HDACi) have been reported, including valproic acid (VPA),^17,18 suberoylanalide hydroxamic acid (SAHA),¹⁹ 4b,²⁰ and trapoxin²¹ (Fig. 1). Five classes of HDAC inhibitors can be identified based on their structures: short chain aliphatic carboxylic acids (such as VPA), hydroxamic acids (such as SAHA), benzamides (such as 4b), cyclic peptides (such as Trapoxin B), and the depsipeptides. Most HDAC inhibitors conform to a structural model, where (1) a cap region binds to the enzyme surface, (2) a Zn²⁺ coordinating group chelates this bound ion at the bottom of a tubular pocket, and (3) a five- to seven-atom spacer links the cap region to metal binding group.

Figure 1.

Chemical structures of HDAC inhibitors.

We recently described a series of benzamide-type, pimelic diphenylamide HDAC inhibitors that show promise as therapeutics for the neurodegenerative diseases Friedreich’s ataxia²⁰ and Huntington’s disease²². In the course of these studies we identified both the enzyme specificities of these compounds (with a preference for HDAC3²³) and specific amino acids in the core histones that are acetylated in cells after treatment with HDACi 4b.²⁰

In contrast to the pimelic diphenylamides, which exhibit high nM IC₅₀ values for recombinant HDAC enzymes, cyclic tetrapeptides, such as trapoxin and apicidin, give low nanomolar IC₅₀ values against HDACs. We reasoned that fusion of the benzamide-alkyl chain portion of 4b with a cyclic peptide might yield a more potent inhibitor than the parent compound 4b. Cyclic tetrapeptide 1, mimicking a four amino acid residue-sequence of histone H4 (RGKG, that is affected by 4b in cells²⁰), was designed, replacing the K residue with a benzamide-protected aminosuberic acid (Asu) residue, and making it a cyclic tetrapeptide. This compound has the same five carbon-linker length as in Trapoxin and Apicidin.

Cyclization is the key step in the synthesis of a cyclic peptide,^24,25 which generally involves preparation of the partially protected linear precursor by solution or solid phase approaches, followed by cyclization in solution under high dilution conditions.²⁶ Recently, Papini et. al.²⁷ reported on-resin head-to-tail cyclization of cyclotetrapeptides, thus taking advantage of the pseudodilution phenomenon,²⁸ which favors intrachain resin-bound reactions, minimizing inter-chain interactions. In this study, Fmoc/tBu/OAl SPPS strategy²⁹ was applied as the orthogonal three-dimensional protection scheme in the synthesis of the linear tetrapeptide. The procedure is described in Scheme 1.

Scheme 1.

Reagents and conditions: (a) 1,2-phenylenediamine (10 eq.), DIPEA (10 eq.), DMF, rt, overnight; (b) Fmoc-L-Asu, HBTU, HOBt, DIPEA, DMF, rt, 3h, three times; (c) HBTU, HOBt, DIPEA, DMF, rt, 1 h then NH₂-Gly-OAl; (d) 20% piperidine in DMF (20 mins × 2); Fmoc-Gly-OH, HBTU, HOBt, DIPEA, DMF; (e) 20% piperidine in DMF (20 mins × 2); Fmoc-Arg(Pbf)-OH, HBTU, HOBt, DIPEA, DMF; (f) Pd(PPh₃)₄, Ph₃SiH, CH₂Cl₂ (anhy.) (2 hs × 2) then 20% piperidine in DMF (20 mins × 2); (g) TBTU, DIPEA, DMF, rt, 4 h; (h) TFA/H₂O (95 : 5), 3 h, rt.

Compound 1 was prepared according to Scheme 1, starting from the trityl chloride resin, which was chosen for the trifunctional amino acids contained in the sequence and such that the final product would have a free amino group. This resin was found to react readily with an excess of 1,2-phenylenediamine³⁰ to give a high yield of the derivatized resin 2, which after washing and drying was ready for the assembly of the desired orthodiamine product.

Linkage of Fmoc-L-Asu-OH to resin 2 was not as straight forward, since Fmoc-L-Asu-OH was a difunctional amino acid, in which both the functional groups were carboxylic groups. Several conditions for coupling were investigated, as shown in Table 1. When 2 eq. (based on Fmoc-L-Asu-OH) of coupling reagent (HATU) was used, by-product 9 (Fig. 2) was formed exclusively. When 1 eq. (based on Fmoc-L- Asu-OH) of coupling reagent was used, the ratio between 3 and 9 was decreased when coupling reagent varied from HATU (condition 2), PyBOP (condition 3), EDCI (condition 4), and HBTU (condition 5). The best result was obtained when HBTU, HOBt (1 eq. according to Fmoc-L-Asu-OH, 5 eq. according to resin 2) was used and the ratio between 3 and 9 was 2.3 (condition 5). We use condition 5 for further synthesis.

Table 1.

conditions for the linkage of Fmoc-L-Asu-OH to resin 2.

condi tion	Eq.b of Fmoc- L-Asu-OH	Coupling reagents	Eq.b of coupling reagent	Ratio between 3 and 9a
1	5	HATU, HOAt	10 eq.	Exclusive 9
2	5	HATU, HOAt	5 eq.	40: 60
3	5	PyBOP	5 eq.	50:50
4	5	EDCI, HOBt	5 eq.	60:40
5	5	HBTU, HOBt	5 eq.	70:30

^aRatio between 3 and 9 was determined by LC/MS.

^beq. was compared to resin 2.

Figure 2.

Structures of by-products 9 and 10.

With resin 3 in hand, we next installed NH₂-Gly-OAl directly onto resin 3 using HBTU/HOBt/DIPEA to afford resin 4. Deprotection with 20% piperidine in DMF, followed by direct coupling with Fmoc-Gly-OH/HBTU/HOBt/DIPEA, affords resin 5. Again treatment with 20% piperidine in DMF, followed by direct coupling with Fmoc-Arg(Pbf)-OH/HBTU/HOBt/DIPEA afforded resin 6. Thus, the tetrapeptide skeleton was formed.

To avoid a nucleophilic attack of the free amino function leading to by-products, deprotection of the C-terminal carboxyl function must be carried out before the last Fmoc removal.³¹ Allyl deprotection was performed using Pd(PPh₃)₄/Ph₃SiH in CH₂Cl₂. Deprotection with 20% piperidine in DMF afforded the unprotected tetrapeptide 7 on resin. As to the cylization reaction, the following coupling conditions were tried: PyBOP/DIPEA, HATU/HOAt/DIPEA, and TBTU/DIPEA. The desired cyclic tetrapeptide was not obtained using the first two coupling conditions. TBTU/DIPEA gave the desired cyclic tetrapeptide on resin. Cleavage from the resin with TFA/H₂O (95:5) provided tetrapeptide 1 with 6% yield (over-all). By-product 10 (Fig. 2) was obtained with 4% yield (over-all) as product of compound 9.

Cyclic tetrapeptide 11 (Fig. 3) was designed, mimicking trapoxin by replacing the side chain with benzamide protected aminosuberic acid. We synthesized compound 11 with 5% yield (over-all) following the same procedure as described above (Scheme 1). Compound 12 was formed as the by-product with 3% yield (over-all).

Figure 3.

Structures of 11 and by-product 12.

We also prepared compounds 13 and 14 (Fig. 4), the linear tetrapeptide versions of compounds 1 and 11, respectively.

Figure 4.

Structures of 13 and 14.

A new type of HDAC inhibitor, exemplified by 15 and 16, was synthesized where amino acid units were incorporated into the aliphatic linker of 4b. Compounds 15 and 16 were synthesized as shown in Scheme 2 using SPPS. Starting from resin 2, Fmoc-β-Ala-OH was coupled to resin using HBTU/HOBt/DIPEA conditions. Treatment with 20% piperidine in DMF, followed by amino acid coupling (glycine for product 15, arginine for product 16) gave 18a and 18b. Deprotection of Fmoc and reaction with p-toluoyl chloride, followed by cleavage with TFA/H₂O (95:5), gave the desired products 15 and 16.

Scheme 2.

Reagents and conditions: (a) Fmoc-β-Ala-OH, HBTU, HOBt, DIPEA, DMF; (b) 20% piperidine in DMF (20 mins × 2); Fmoc-Gly-OH for 18a/Fmoc-Arg(Pbf)-OH for 18b, HBTU, HOBt, DIPEA, DMF; (c) 20% piperidine in DMF (20 mins × 2); p-toluoyl chloride, DIPEA, CH₂Cl₂; (d) TFA/H₂O (95 : 5), 3 h, rt.

The synthesized inhibitors 1, 10, 11, 12, 13, 14, 15, and 16 were screened against recombinant human HDAC1 and HDAC3, and IC₅₀s for inhibition of these enzymes are summarized in Table 2. Some interesting trends were observed in these measurements. As far as activity is concerned, cyclo tetrapeptide 1 is stronger than linear tetrapeptide 13, but weaker than N¹, N⁷-bis(2-aminophenyl) heptanediamide-like compound 10. The same trend is also observed with cyclo tetrapeptide 11, linear tetrapeptide 14 and heptanediamide-like compound 12. In conclusion, with the same amino acid sequence, the linear tetrapeptide is the weakest, cyclo tetrapeptide is somewhat stronger and the heptanediamide-like compound is the strongest inhibitor, respectively. When compared to the known cyclotetrapeptide Apicidin (IC₅₀ = 4 nM)³² and Trapoxin B (IC₅₀ = 0.11 nM for HDAC1),³³ the activities of our compounds have decreased considerably. Optimal activity was observed with compound 10 (IC₅₀ = 4.7 µM for HDAC1 and 1.3 µM for HDAC3). Surprisingly, compound 15 shows a 19-fold selectivity between HDAC1 and HDAC3, with a clear preference for HDAC3.

Table 2.

Inhibitory activity on recombinant HDAC1 and HDAC3.

Compds	HDAC1 IC50, µMa	HDAC3 IC50, µMa
1	191(±27.8)	46.9 (±17.1)
10	4.7 (±0.2)	1.3 (±0.7)
13	532 (±37.7)	69.3 (±6.5)
15	83.9 (±21.4)	4.3 (±1.2)
16	9.2 (±0.5)	2.2 (±0.02)
11	20.5 (±3.0)	5.6 (±0.5)
12	7.3 (±1.4)	3.4 (±0.3)
14	84.2 (±15.9)	27.2 (±8.5)

^aValues are means of two experiments with standard error given in parentheses.

The effects of 1, 10, 11, 12, 13, 14, 15, and 16 (at 20 µM concentration) on the acetylation levels of endogenous histone H3 in a human lymphoblastoid cell line (GM15850, Coriell Institute) were determined by western blot analysis, after 48 h incubations. The result of this experiment is shown in Fig. 5. For the H4 RGKG mimics, 10 showed relatively stronger effects on acetylated H3, 13 showed no effect and 1 showed moderate effects on acetylation. These results are consistent with the IC₅₀ values for recombinant HDACs (Table 2). As for the trapoxin mimics, the same trend was observed; that is, the lower the IC₅₀ value for each compound (Table 2), the stronger the effect on histone H3 acetylation in cells. As for the amino acid inhibitors, compound 15 showed a relatively strong effect on cellular histone H3 acetylation, while compound 16 showed no such effect. We believe that the positively charged Arg group in compound 16 may hinder the compound to penetrate cells, while the relative neutral compound 15 may allow cell penetration. This difference in charge may explain why compounds 15 and 16 have remarkable differences in their ability to cause histone H3 acetylation in cells but nearly equal IC₅₀ values for inhibition of recombinant HDAC3, and compound 15 has a higher IC₅₀ for HDAC1 than compound 16.

Figure 5.

Effects of compounds 1, 10, 13, 15, 16, 11, 12, and 14 on histone H3 in GM15850 cells.

In conclusion, we have synthesized two different types of benzamide compounds. Most show moderate inhibitory activity on HDAC1 and HDAC3. Compound 15 shows 19-fold selectivity between HDAC3 and HDAC1. Further studies will be carried out on compound 15.