design and synthesis of a novel thiolate histone deacetylase inhibitor - manuscript

Design and Synthesis of a Novel ThiolateHistone Deacetylase Inhibitor

Maxwell TuckerThe North Carolina School of Science and Mathematics

Acknowledgments

I would like to thank Dr. Myra Halpin of the North Carolina School of Science and Mathematics,

for mentoring me throughout this entire project. I would like to thank Dr. Darrell Spells for his

organic chemistry expertise, along with the entire research program at NCSSM. Finally, I would

like to thank Dr. W. Andrew Tucker of Queens University for his assistance with NMR

Spectroscopy.

1

Abstract

Histone deacetylase inhibitors are a class of chemotherapeutic epigenetic drugs that have

recently been found to be quite effective in the treatment of a number of late stage carcinomas.

The inhibitors interfere with gene expression by chelating the zinc ion found in class I histone

deacetylases, which causes apoptosis in cancer cells. However, current commercial inhibitors

suffer from either limited selectivity causing a host of side effects, or structural complexity

which poses a significant synthetic challenge. Here, a novel thiolate histone deacetylase in-

hibitor with a small aromatic surface recognition region and an aliphatic linker is reported.

This molecule was designed using the MolDock virtual docking method, and is anticipated to

display significantly greater selectivity than current hydroxamic acid inhibitors by combining

the thiolate chelating agent characteristic of depsipeptide inhibitors with a structurally simple

linker and capping group. After design and modeling, this compound was synthesized with a

simple three-step synthesis, verified by FT-IR and NMR, making production very simple when

compared to current selective cyclic depsipeptide inhibitors. This compound shows significant

promise as a novel histone deacetylase inhibitor for the treatment of cancer.

2

1 INTRODUCTION

1 Introduction

Histone deacetylase inhibitors are a class of drugs which have been used for many years to

treat a number of disorders, including mood disorders and epilepsy. These inhibitors have only

recently come into use as cancer treatments; although research into this application began in the

mid nineties, the FDA did not approve the first drug, vorinostat, until 2006 [1]. Currently only

two histone deacetylase inhibitors are approved for cancer treatment, although many more are in

development. This class of drugs is very promising due to its favorable cytotoxicity profile along

with its high selectivity and potency [2]. However the most potent inhibitors feature complex

structural components which make synthesis a challenge. In this project a novel inhibitor with

high predicted selectivity and simple synthesis is proposed.

Histone deacetylase (HDAC) is an enzyme which removes acetyl groups from histones. His-

tones are a class of proteins upon which DNA is coiled and are a major component of chromatin.

The removal of acetyl groups allows for tighter packing of the DNA on the histone proteins, and

as a result limits gene expression due to the physical inaccessibility of the genetic material. This

function is opposite of that which is completed by the histone acetyltransferase enzyme, which

acetylates certain histones and therefore promotes gene expression. Eighteen HDACs have been

identified in humans, and they are divided into three classes based on structure. Class I and class

II HDACs contain a metallic ion cofactor in their active site, Zn2+ [3]. While other HDACs do

play important roles in the body, class I is of primary concern for cancer treatment [4]. This is

because class I HDACs are found almost exclusively in the nucleus and have the most effect on

gene expression, while class II inhibitors are believed to have tissue specific functions [5]. Thus

selectivity for this class is a valuable trait in inhibitors, as it prevents unwanted side affects and

yields better performance.

Histone deacetylase inhibitors (HDACi) offer epigenetic treatment and act as antagonists, bind-

ing to the HDAC and preventing it from fulfilling its purpose. This is desirable because over-

3

1.1 Hydroxamic Acid Inhibitors 1 INTRODUCTION

expression of HDACs is reported in a number of cancers, and inhibition can lead directly to apop-

tosis [5]. Structurally, HDACis consist of 3 parts: the chelating agent, the linker, and the capping

group (Figure 1). The chelating group has the largest effect on HDACi efficacy because it co-

ordinates with the Zn2+ in the HDAC, thus preventing it from catalyzing deacetylation [6]. The

linker length can have a significant impact on efficacy of a HDACi because it effects the ability

of the chelating agent to reach the zinc ion. However, this length is easily adjustable and can be

optimized simply. The capping group, which acts as a surface recognition section for the enzyme,

is perhaps one of the most important parts of an inhibitor, as it has a large effect on the selectivity

and activity of an inhibitor. HDACis are divided into categories based on the properties of their

capping groups and chelating agents.

Figure 1: An image of SAHA, a common HDACi with the 3 major groups labeled.

1.1 Hydroxamic Acid Inhibitors

The most studied class of HDACis is the hydroxamic acids. These inhibitors contain a hydrox-

amic acid as a chelating agent, which forms a very stable 5 membered ring with the zinc ion [3].

One such inhibitor is suberoylanilide hydroxamic acid (SAHA), also known as vorinostat, which is

now on the market under the name Zolinza for late stage cancer treatment [1,7]. Studies have even

shown that SAHA may be effective for the treatment of Huntington’s disease, hinting at possible

other applications for deacetylase inhibitors [8]. Hydroxamic acid inhibitors tend to have small,

hydrophobic, and usually aromatic surface recognition groups, which are simple synthetically due

4

1.2 Depsipeptide/Thiolate Inhibitors 1 INTRODUCTION

Figure 2: Note the hydroxamic acid and similar aromatic capping areas in these 3 common in-hibitors of this class.

to the easy availability of precursors. In figure 2, note these similarities across 3 common hydrox-

amic acid inhibitors [9]. This hydrophobicity is very important to activity, as it increases binding

affinity by creating favorable interactions with the amino acid residues around the active site, and

in fact nearly all HDACis have a hydrophobic surface recognition areas. Despite these favorable

stuctural characteristics, hydroxamic acid inhibitors do feature a notable downside when compared

to other inhibitor classes. These inhibitors do not show the high level of selectivity seen in other

inhibitors, and therefore have less favorable cytotoxicity profiles. This manifests itself in uncom-

fortable side effects for the patients, as evidenced by clinical trials of vorinostat, in which some

patients experienced pain, anorexia, fatigue, nausea, and vomiting [4].

1.2 Depsipeptide/Thiolate Inhibitors

Another class of HDACi is the depsipeptide class. These inhibitors contain large cyclic dep-

sipeptide capping groups and feature a thiol chelating agent. One inhibitor of this class, romidepsin

(FK228), is one of only two HDACis to be approved by the FDA for cancer treatment. Romidepsin,

like many depsipeptide inhibitors acts as a prodrug, with a disulfide bond that is reduced within

the body to form the active thiol [10]. Another naturally derived depsipeptide inhibitor is larga-

5

1.3 Project Goals 1 INTRODUCTION

Figure 3: Note both the prodrug nature and the structural similarity between romidepsin and larga-zole [3].

zole, a compound that is still in preliminary stages but may one day be approved for cancer treat-

ment. Largazole contains a thioester which is transformed within the body to form the thiol active

metabolite. Largazole and romidepsin actually are very structurally similar, with identical chelat-

ing agents and linkers, along with very similar macrocycles [11].

In general, the depsipeptide inhibitors yield higher activity and selectivity than hydroxamic

acid inhibitors [9]. However, depsipeptide inhibitors have large macrocycles which are challenging

synthetically. For example, largazole has a 17 carbon macrocycle which contains two five mem-

bered heterocycles as well as a number of substituents. This creates a synthesis which consists

of many steps and has a relatively low yield overall. When considering commercial production,

this complexity could make such a drug prohibitively expensive, especially when compared to the

structurally simple hydroxamic acid inhibitors.

1.3 Project Goals

The purpose of this project is to create a synthetically and structurally simple yet highly effective

histone deacetylase inhibitor for use in cancer treatment. This can be accomplished by combining

desirable traits from multiple classes of inhibitor, in particular by combining the thiolate chelating

agent with the simple aromatic capping regions from hydroxamic acid inhibitors. Surprisingly,

6

2 MATERIALS AND METHODS

work in this area has been relatively limited. While a few attempts at thiolate analogues of SAHA

have been made, few entirely engineered HDACis have been synthesized [12, 13]. Therefore the

goal is to create a structurally simple, easy to synthesize product that should be highly effective as

an inhibitor, and in particular yields greater selectivity than current hydroxamic acid inhibitors. The

design of this molecule can be carried out using computational models of protein-ligand binding

to accurately predict most optimal structures.

2 Materials and Methods

2.1 Computational Modeling and Structural Determination

To determine an ideal molecular structure for a novel inhibitor, computational methods were

utilized. Molegro Virtual Docker software was used to model the interactions between large num-

ber of potential structures and histone deacetylase proteins, as well as modeling several known

HDACis as reference ligands. Molegro Virtual Docker uses the MolDock molecular docking

method, which utilizes a heuristic method and an evolutionary algorithm. This particular dock-

ing method has been shown to yield significantly higher accuracy than other molecular docking

methods, and allowed for easy relative comparison of many ligands, which was a valuable trait for

these tests [14]. Additionally, MolDock supports metal cofactors, which is especially important

for class I HDACs which contain zinc dependent binding sites.

The crystal structure of HDAC8, determined by X-ray crystallography, was obtained from the

worldwide Protein Data Bank [15]. This structure was imported into the Virtual Docker, and

then an analysis of electrostatic forces was used to detect cavities which could be likely binding

sites. The proper cavity was selected manually based on the location of the zinc ion, and ligands

were subsequently docked with this cavity. In total, eight ligands were docked on the protein,

6 of which were novel structures, along with two reference molecules, trichostatin A (TSA) and

suberoylanilide hydroxamic acid (SAHA), both visible in figure 2. Each of these novel compounds

featured a similar surface recognition domain, derivative of the structure found in TSA. However,

7

2.2 Novel Thiolate Synthesis 2 MATERIALS AND METHODS

Figure 4: The left image shows a structural representation of the HDAC8 protein, while the imageon the right demonstrates the interactions of SAHA with the residues in the binding domain asmodeled by MolDock.

the tertiary amine was modified to a primary amine for the sake of synthetic simplicity. This

surface recognition structure was decided upon due to the relatively high performance of TSA when

compared to other hydroxamic acid inhibitors [16]. In these six novel ligand structures, changes

were made to the linker domain, including esterification and varying aliphatic chain length. Figure

4 shows some of the capabilities of this software, including the graphical interface and output.

The software was also instructed to calculate relative binding affinities of the top 5 poses for each

ligand. These poses were then viewed for structural accuracy, and then compared to yield the

structure with most ideal inhibition properties. Results from this modeling can be seen in table 1.

2.2 Novel Thiolate Synthesis

Figure 5: The synthetic target

Based on binding affinity results from the computational modeling described above, the molecule

8


Figure 6: 4-nitrobenzoyl chloride and 6-mercapto-1-hexanol

visible in figure 5 was settled upon. Structurally, it consists of two identical molecules joined by

a disulfide bond. The starting materials for this synthesis were 4-nitrobenzoyl chloride and 6-

mercapto-1-hexanol, as seen in figure 6.

The first synthetic step was the esterification reaction between the acid chloride and the alcohol.

This procedure was adapted from Hubbard and Brittain [17, 18]. 7.45x10−4 moles of both 4-

nitrobenzoyl chloride and 6-mercapto-1-hexanol were combined in 10 mL dichloromethane with

triethylamine catalyst. This reaction mixture was stirred for three and a half hours in an ice bath at

0 ◦C under an argon atmosphere. The reaction was monitored via thin layer chromatography using

a 2:3 ratio of hexane to ethyl acetate as the eluent, and then visualized using UV light. The product

mixture was washed three times with a brine solution, then dried over sodium sulfate. This sample

was purified using column chromatography with silica gel 60 and a solvent mixture consisting of

3:1 hexane and ethyl acetate. After purification, the sample yielded one spot on TLC and was

deemed to be relatively pure. The solvent was extracted from the column fractions using rotary

evaporation, and produced an oily yellow solid. An attenuated total reflectance (ATR) attachment

on a Fourier transform infrared spectrophotometer (FT-IR) was used to verify structure. Strong

absorbance at 2928 cm−1 was indicative of an aliphatic CH stretch, and the carbonyl peak at 1720

cm−1 was indicative of the successful addition of the acid chloride. Strong absorbance at 1350

cm−1 was also indicative of the nitro group, which provides additional evidence for a successful

synthesis.

The next step in this synthetic plan was the reduction of the nitro group to an amine. This re-

duction was accomplished using stannous chloride dihydrate as described by Bellamy [18,19]. The

purified product from the previous reaction was combined in a 1:5 molar ratio with SnCl2 ·2 H2O

9


Figure 7: Here the IR spectrum from the first reaction can be seen. Note the carbonyl and nitropeaks, at 1720 cm−1 and 1350 cm−1, respectively.

Figure 8: Thiol active metabolite

in ethyl acetate. This reaction mixture was then heated to 70 ◦C and stirred for one hour. The

majority of SnCl2 ·2 H2O was removed using gravity filtration, and the remaining reaction mixture

was washed with brine and then dried over sodium sulfate. TLC showed the sample to be rela-

tively pure, and thus no further purification was necessary. Solvent was evaporated using a rotary

evaporator, and the sample was then massed to calculate yield. IR spectroscopy was once again

used for structural verification, and there was clear indication of a primary amine with absorbance

at 3364 cm−1. This reaction yields the product seen in figure 8.

The final reaction involved the oxidation of the thiol to form a disulfide dimer of figure 8, thus

10


Figure 9: The full synthetic roadmap for the novel HDACi

Figure 10: This IR spectrum shows the active metabolite, as seen in figure 8.

11

2.3 Instrumentation 2 MATERIALS AND METHODS

producing the molecule seen in figure 5. This was accomplished by stirring the thiol in a solution

of dichloromethane under an oxygen atmosphere at slight positive pressure for 20 hours, producing

a near quantitative yield of the disulfide. Final structural verification was carried out using proton

and carbon-13 NMR.

After this first synthesis, a second scale-up synthesis was completed to verify results and test

scale-up feasibility. For the benzoyl chloride esterification, all amounts were quintupled, and reac-

tion time was increased to 4.5 hours. Column purification failed with a ethyl acetate:hexane eluent

due to poor product solubility in this mixture. To rectify this, a 3:2 mixture of dichloromethane

and hexane was used. The nitro reduction reaction was run using an identical procedure. However,

the final oxidation of the thiol to a disulfide was done using a new procedure, using hydrogen per-

oxide catalyzed with potassium iodide, adapted from Kirihara et al [20]. Molar equivalents of the

thiol and 30% H2O2 solution were combined in the presence of one mole percent KI and stirred

for 4 hours in 10 mL of ethyl acetate at room temperature. Extraction was preformed using ethyl

acetate, washed once with deionized water and three times with brine. The resulting organic layer

was dried over sodium sulfate, and then solvent was removed using a rotary evaporator.

2.3 Instrumentation

All infrared spectra were acquired using a Shimadzu FTIR 8400S. Eighty scans were run to

yield accurate spectra. An attenuated total reflectance (ATR) adapter was used to allow for the

analysis of the solid samples.

Proton and C13 NMR were obtained using an Anasazi EFT-60 spectrometer. The proton spec-

tra were collected running 128 scans at 60.01 MHz. Carbon-13 spectra were collected with 32,768

scans at 15.089MHz. Both scans were collected on approximately 30 mg of product dissolved

in deuterated chloroform. These spectra were created using equipment at an off-site academic

institution due to the lack of instrumentation available locally.

12

3 RESULTS AND DISCUSSION

3 Results and Discussion

3.1 Molecular Modeling and Structure Determination

The MolDock virtual docker engine generated hundreds of potential poses for each of the eight

potential ligands modeled. For each pose generated, numerous parameters are considered, includ-

ing hydrogen bond energies, electron affinity, cofactor interactions, and many more. All of these

parameters are combined to create the rerank score, which is designed to accurately predict relative

binding affinity for a number of different ligands. Table 1 shows the top pose for each of the eight

tested ligands and the associated values for a number of key factors. All energies on an arbitrary

relative scale unless otherwise noted.

13

3.1 Molecular Modeling and Structure Determination 3 RESULTS AND DISCUSSION

Table 1: This table summarizes the top poses for each of the tested ligands, sorted by rerank score.

14

3.2 Synthesis Results 3 RESULTS AND DISCUSSION

Note the six potential inhibitors tested, with varying linker length and esterification, along with

a methyl group akin to TSA’s linker domain. There are some interesting trends to note from these

data. In general, longer linker chains tend to perform better than shorter, and ester linkages seem

to preform better than the ketone equivalents. Additionally, a methyl group directly adjacent to

the carbonyl seemed to yield better binding affinity. While SAHA did out perform the ultimate

synthetic target in terms of binding affinity for HDAC8, it is expected that the synthetic target

will yield better selectivity due to the thiolate chelating agent. While better performance might

have been gained from addition of a methyl group, this was decided against in favor of synthetic

simplicity.

If the modeled ligand from figure 10 is compared to the synthetic target from figure 5, it may be

noticed that the synthetic target is a disulfide dimer of the modeled molecule. This is because the

free thiol is vulnerable to various reactions within the body, and if unprotected will have limited

effects in vivo. Other thiolate inhibitors have a variety of protecting groups. Romidepsin features

a disulfide bond which is reduced to a thiol in the body, while largazole utilizes a thioester which

is also reduced in vivo [10, 21]. In this case, a disulfide dimer seemed to be a convenient solution

that protected the otherwise vulnerable thiol while also avoiding any wasteful protecting groups.

3.2 Synthesis Results

The reaction of 4-nitrobenzoyl chloride and 6-mecapto-1-hexanol was successfully carried out

with a yield of 30.4%. After purification, the sample produced one spot on TLC, indicating rel-

atively high purity. This yield was lower than hoped, likely due to the low catalytic activity of

triethylamine. Next the reduction of the nitro group to a primary amine was successfully carried

out with a yield of 42.9%. This was surprising, as TLC reaction monitoring seemed to indicate

that the reaction had gone to completion. Because of this, it seems likely that product was lost in

the extraction process. In particular, there may be a significant amount of product in the aqueous

portion from extraction. The final thiol oxidation had poor yield and a long reaction time using the

O2 gas method. To rectify this, the hydrogen peroxide procedure was adopted, and produced near

15

3.2 Synthesis Results 3 RESULTS AND DISCUSSION

Figure 11: Carbon-13 NMR of the final product, showing peaks representing the eleven chemicallydistinct carbons in this molecule. The three peaks from 70 to 80 ppm are caused by the solvent,deuterated chloroform.

quantitative yields in a much shorter time than the O2 procedure.

While intermediate products were verified using IR spectroscopy, final structural verification

was carried out using NMR. Proton and carbon-13 shifts were predicted computationally using

B3LYP/6-31G∗∗ calculations after structural optimization had been preformed using a 6-31G∗ ba-

sis set. Figure 11 shows the carbon-13 spectrum used for final verification. Noise along the baseline

is due partially to impurities and partially to the low resolution of this 60 MHz NMR. This spec-

trum shows 11 distinct peaks in locations similar to those predicted by the computations. This is

conclusive evidence for the presence of the expected disulfide product.

16

5 FUTURE WORK

4 Conclusions

This project was successful in synthesizing an entirely novel compound that shows promise as

a highly selective histone deacetylase inhibitor. Modeling using the MolDock platform yielded

insights into the functionality of the inhibitors, and allowed for the design of a previously un-

synthesized molecule with favorable docking features. This compound was then synthesized in a

simple three step synthesis that should be easily replicable and required minimal purification. This

offers significant improvement over current highly selective inhibitors. Romidepsin is synthesized

with an overall yield of 9.5%, and requires separate syntheses for many uncommon starting ma-

terials, making it a huge synthetic challenge overall [22]. The increased structural simplicity of

the product reported above is key when considering the large scale production necessary of all

pharmaceutical substances.

5 Future Work

Future work for this project falls into three main categories. First, additional computational

data should be obtained to determine more about the properties of this inhibitor design. Experi-

mentation with varying hydrophobic regions and non-aliphatic linkages could produce interesting

results. However, the main computational work lies in validating the assertion that the thiolate

chelating agent does in fact yield improved selectivity. This can be done by modeling the same set

of test ligands on a number of other HDACs, including HDAC1, 2, and 3, as well as several class

2 and 3 HDACs. This would produce compelling evidence for the increased selectivity of thiolate

chelating agents.

Another important aspect of future work is increased synthetic yield and validation. Currently,

the esterification reaction has relatively low yield, likely due to the low catalytic activity of tri-

ethylamine. This yield may be improved by using a different tertiary amine catalyst, such as

4-dimethylaminopyridine (DMAP) or diethylmethylamine [17]. The next reaction in the synthe-

sis is the nitro group reduction. This reaction also showed a relatively low yield, although the

17

5 FUTURE WORK

cause of this is not fully understood. Thin layer chromatography indicated that the reaction went

to completion, yet final product mass was disappointingly low. This may be due to product losses

in the extraction process, where some product may have been retained in the aqueous phase or in

the stannous chloride from gravity filtration. A better workup process, including switching extrac-

tion solvents, may produce better yield. Due to the inability to access NMR or mass spectrometry

equipment, product analysis, especially of intermediate products, is not particularly robust. In a

future synthesis replication, it would be necessary to obtain accurate spectra for each product to

verify structure. Additionally, reaction scale-up should be attempted to production on a multi-gram

scale.

The final step in the project development would be preliminary biological testing. Testing can

be performed on a variety of histone deacetylases to calculate IC50 values. This will allow for

accurate real world comparison of this drug to other inhibitors, and if this preliminary testing is

promising, in vivo testing could take place using established cell lines. Ultimately, it is hoped that

this compound has the potential to reach clinical trials and perhaps one day serve as a chemother-

apeutic epigenetic drug for the treatment of late stage carcinomas.

18

References[1] Merck, “Study of ZOLINZA ( vorinostat ) for Investigational Use for Advanced Malignant

Pleural Mesothelioma Did Not Meet Primary Endpoint,” Merck Newsroom, pp. 1–2, 2011.

[2] J. a. Souto, E. Vaz, I. Lepore, A.-C. Poppler, G. Franci, R. Alvarez, L. Altucci, and A. R.de Lera, “Synthesis and biological characterization of the histone deacetylase inhibitor larga-zole and C7- modified analogues.,” Journal of medicinal chemistry, vol. 53, pp. 4654–67,June 2010.

[3] K. E. Cole, D. P. Dowling, M. A. Boone, A. J. Phillips, and D. W. Christianson, “StructuralBasis of the Antiproliferative Activity of Largazole, a Depsipeptide Inhibitor of the HistoneDeacetylases,” Journal of the American Chemical Society, no. 133, pp. 12474–12477, 2011.

[4] X. Ma, H. H. Ezzeldin, and R. B. Diasio, “Histone deacetylase inhibitors: current status andoverview of recent clinical trials.,” Drugs, vol. 69, pp. 1911–34, Oct. 2009.

[5] M. Dokmanovic, C. Clarke, and P. a. Marks, “Histone deacetylase inhibitors: overview andperspectives.,” Molecular cancer research : MCR, vol. 5, pp. 981–9, Oct. 2007.

[6] Y. Liu, L. A. Salvador, S. Byeon, Y. Ying, J. C. Kwan, B. K. Law, J. Hong, and H. Luesch,“Anticolon Cancer Activity of Largazole , a Marine-Derived Tunable Histone Deacetylase In-hibitor,” Journal of Pharmacology and Experimental Therapeutics, vol. 335, no. 2, pp. 351–36, 2010.

[7] W. K. Kelly, O. a. O’Connor, L. M. Krug, J. H. Chiao, M. Heaney, T. Curley, B. MacGregore-Cortelli, W. Tong, J. P. Secrist, L. Schwartz, S. Richardson, E. Chu, S. Olgac, P. a. Marks,H. Scher, and V. M. Richon, “Phase I study of an oral histone deacetylase inhibitor, suberoy-lanilide hydroxamic acid, in patients with advanced cancer.,” Journal of clinical oncology: official journal of the American Society of Clinical Oncology, vol. 23, pp. 3923–31, June2005.

[8] E. Hockly, V. M. Richon, B. Woodman, D. L. Smith, X. Zhou, E. Rosa, K. Sathasivam,S. Ghazi-Noori, A. Mahal, P. a. S. Lowden, J. S. Steffan, J. L. Marsh, L. M. Thompson, C. M.Lewis, P. a. Marks, and G. P. Bates, “Suberoylanilide hydroxamic acid, a histone deacetylaseinhibitor, ameliorates motor deficits in a mouse model of Huntington’s disease.,” Proceedingsof the National Academy of Sciences of the United States of America, vol. 100, pp. 2041–6,Feb. 2003.

[9] T. a. Miller, D. J. Witter, and S. Belvedere, “Histone deacetylase inhibitors.,” Journal ofmedicinal chemistry, vol. 46, pp. 5097–116, Nov. 2003.

[10] R. Furumai, A. Matsuyama, N. Kobashi, and H. Deacetylases, “FK228 ( Depsipeptide ) asa Natural Prodrug That Inhibits Class I Histone Deacetylases FK228 ( Depsipeptide ) as aNatural Prodrug That Inhibits Class I,” Cancer Research, vol. 228, pp. 4916–4921, 2002.

[11] J. Hong and H. Luesch, “Largazole: from discovery to broad-spectrum therapy.,” Naturalproduct reports, vol. 29, pp. 449–56, Apr. 2012.

[12] T. Suzuki, A. Kouketsu, A. Matsuura, A. Kohara, S.-I. Ninomiya, K. Kohda, and N. Miy-ata, “Thiol-based SAHA analogues as potent histone deacetylase inhibitors.,” Bioorganic &medicinal chemistry letters, vol. 14, pp. 3313–7, June 2004.

[13] a. Saito, T. Yamashita, Y. Mariko, Y. Nosaka, K. Tsuchiya, T. Ando, T. Suzuki, T. Tsuruo,and O. Nakanishi, “A synthetic inhibitor of histone deacetylase, MS-27-275, with markedin vivo antitumor activity against human tumors.,” Proceedings of the National Academy ofSciences of the United States of America, vol. 96, pp. 4592–7, May 1999.

[14] R. Thomsen and M. H. Christensen, “MolDock: a new technique for high-accuracy moleculardocking.,” Journal of medicinal chemistry, vol. 49, pp. 3315–21, June 2006.

[15] J. R. Somoza, R. J. Skene, B. a. Katz, C. Mol, J. D. Ho, A. J. Jennings, C. Luong, A. Arvai,J. J. Buggy, E. Chi, J. Tang, B.-C. Sang, E. Verner, R. Wynands, E. M. Leahy, D. R. Dougan,G. Snell, M. Navre, M. W. Knuth, R. V. Swanson, D. E. McRee, and L. W. Tari, “Structuralsnapshots of human HDAC8 provide insights into the class I histone deacetylases.,” Structure(London, England : 1993), vol. 12, pp. 1325–34, July 2004.

[16] M. Yoshida, M. Kijima, M. Akita, and T. Beppu, “Potent and specific inhibition of mam-malian histone deacetylase both in vivo and in vitro by trichostatin A .,” The Journal ofBiological Chemistry, 1990.

[17] P. Hubbard and W. J. Brittain, “Mechanism of Amine-Catalyzed Ester Formation from anAcid Chloride and Alcohol,” The Journal of Organic Chemistry, vol. 63, pp. 677–683, Feb.1998.

[18] T. W. Bentley, G. Llewellyn, and J. A. McAlister, “S(N)2 Mechanism for Alcoholysis,Aminolysis, and Hydrolysis of Acetyl Chloride.,” The Journal of organic chemistry, vol. 61,pp. 7927–7932, Nov. 1996.

[19] F. D. Bellamy, “SELECTIVE REDUCTION OF AROMATIC NITRO COMPOUNDS WITHSTANNOUS CRLORIDE IN NON ACIDIC AND NON AQUEOUS MEDIUM,” Tetrahe-dron Letters, vol. 25, no. 8, pp. 3–6, 1984.

[20] M. Kirihara, Y. Asai, S. Ogawa, T. Noguchi, A. Hatano, and Y. Hirai, “A Mild and Envi-ronmentally Benign Oxidation of Thiols to Disulfides,” Synthesis, vol. 2007, pp. 3286–3289,Nov. 2007.

[21] K. Taori, V. J. Paul, and H. Luesch, “Structure and Activity of Largazole , a Potent Antipro-liferative Agent from the Floridian Marine Cyanobacterium Symploca sp .,” Journal of theAmerican Chemical Society, vol. 130, no. 6, pp. 1806–1807, 2008.

[22] T. Greshock, D. Johns, Y. Noguchi, and R. Williams, “Improved total synthesis of the potentHDAC inhibitor FK228 (FR-901228),” Organic letters, vol. 10, no. 4, pp. 613–616, 2008.