HDAC inhibitors from natural and synthetic libraries: towards the development of biochemical probes

Thesis

Reference

HDAC inhibitors from natural and synthetic libraries: towards the development of biochemical probes

ZWICK, Vincent

Abstract

Les histones désacétylases (HDACs) sont un groupement d'enzymes dont le dérèglement est impliqué dans plusieurs pathologies (cancer, maladies neurodégéneratives, parasitologie).

Depuis plusieurs années, de nombreux efforts pour cibler thérapeutiquement ces enzymes ont été réalisés. Dans le cadre de ces recherches, des molécules inhibitrices des HDACs ont été découvertes. Ces composés peuvent être directement issus de produits naturels, d'autres sont produits synthétiquement. Quatre inhibiteurs HDAC, la romidepsin, le SAHA, le belinostat et le panobinostat, ont déjà été acceptés par la FDA pour la prise en charge de divers cancers hématologiques. La proposition de recherche que je présente ici a pour but de découvrir de nouveaux inhibiteurs HDACs présentant des caractéristiques intéressantes en termes de puissance et de sélectivité. Elle vise également au développement de stratégies de criblages qui permettraient de faciliter la découverte de telles molécules ainsi qu'à la mise en place d'outils aidant à l'étude de ces composés.

ZWICK, Vincent. HDAC inhibitors from natural and synthetic libraries: towards the development of biochemical probes . Thèse de doctorat : Univ. Genève, 2016, no. Sc. 4985

URN : urn:nbn:ch:unige-889763

DOI : 10.13097/archive-ouverte/unige:88976

Available at:

http://archive-ouverte.unige.ch/unige:88976

Disclaimer: layout of this document may differ from the published version.

1 / 1

Section des Sciences Pharmaceutiques Pharmacognosy

Pharmacochemistry

Professeur Muriel Cuendet Professeur Pierre-Alain Carrupt

HDAC Inhibitors from Natural and Synthetic Libraries: Towards the

Development of Biochemical Probes

THESE

Présentée à la Faculté des sciences de l’Université de Genève

pour obtenir le grade de Docteur ès sciences, mention sciences pharmaceutiques

Par Vincent Zwick De Salenstein (TG)

THESE N°4985

Genève

2016

« Les vrais bons sentiments se traduisent par des actes. Non par des paroles. Ce n’est pas par de grandes pensées qu’un homme se définit. Mais par cette alternance en nous de grandes pensées ou d’intuitions émergeant d’un marécage intime où grouillent le médiocre, le pire, l’odieux. Que l’on prend bien soin de laisser dans l’ombre. La réalité étant faite de leur coexistence d’où les difficultés, souvent, de la relation. Et l’imposture de la transparence.

Pour la millième fois, ne pas oublier que les moments de creux, de perte ou de stérilité préparent, le plus souvent, des jours de plénitude et de surabondance de la chose à dire. La particularité de l’affaire étant qu’on n’y pense jamais durant ces jours de manque. Mais après, quand la source s’est remise à couler. Et cela est bien naturel. Si on y pensait en effet durant les moments de perte, ce ne serait plus l’épreuve de ladite perte et de ses affres. Or, sans les affres, pas de semence... »

GEORGES HALDAS

« Maintenant, je sais que l'homme est capable de grandes actions. Mais s'il n'est pas capable d'un grand sentiment, il ne m'intéresse pas.

- On a l'impression qu'il est capable de tout, dit Tarrou.

- Mais non, il est incapable de souffrir ou d'être heureux longtemps. Il n'est donc capable de rien qui vaille.

Il les regardait, et puis :

- Voyons, Tarrou, êtes-vous capable de mourir pour un amour ? - Je ne sais pas, mais il me semble que non, maintenant.

- Voilà. Et vous êtes capable de mourir pour une idée, c'est visible à l’œil nu. Eh bien, moi, j'en ai assez des gens qui meurent pour une idée. Je ne crois pas à l'héroïsme, je sais que c'est facile et j'ai appris que c'était meurtrier. Ce qui m’intéresse, c'est qu'on vive et qu'on meure de ce qu'on aime… »

ALBERT CAMUS

« - Va revoir les roses. Tu comprendras que la tienne est unique au monde. Tu reviendras me dire adieu, et je te ferai cadeau d'un secret.

Le petit prince s'en fut revoir les roses :

- Vous n'êtes pas du tout semblables à ma rose, vous n'êtes rien encore, leur dit-il. Personne ne vous a apprivoisées et vous n'avez apprivoisé personne. Vous êtes comme était mon renard. Ce n'était qu'un renard semblable à cent mille autres. Mais j'en ai fait mon ami, et il est maintenant unique au monde.

Et les roses étaient bien gênées.

- Vous êtes belles, mais vous êtes vides, leur dit-il encore. On ne peut pas mourir pour vous. Bien sûr, ma rose à moi, un passant ordinaire croirait qu'elle vous ressemble. Mais à elle seule elle est plus importante que vous toutes, puisque c'est elle que j'ai arrosée. Puisque c'est elle que j'ai mise sous globe. Puisque c'est elle que j'ai abritée par le paravent. Puisque c'est elle dont j'ai tué les chenilles (sauf les deux ou trois pour les papillons). Puisque c'est elle que j'ai écoutée se plaindre, ou se vanter, ou même quelquefois se taire. Puisque c'est ma rose.

Et il revint vers le renard : - Adieu, dit-il...

- Adieu, dit le renard. Voici mon secret. Il est très simple : on ne voit bien qu'avec le cœur. L'essentiel est invisible pour les yeux… »

ANTOINE DE SAINT-EXUPERY

Caspar David Friedrich (1774-1840), « Le voyageur contemplant une mer de nuages », 1818

Remerciements

Tant de personnes ont rendu possible l’avènement de ce travail de thèse qu’il m’est aujourd’hui difficile de n’en oublier aucune ; je vous demande donc par avance de bien vouloir excuser mes oublis éventuels. Au terme de ces années de doctorat et au commencement d’une nouvelle étape de ma vie, j’éprouve une sincère gratitude envers tous ceux qui ont participé à ce travail et que je tiens ici, à remercier.

J’exprime tout d’abord mes remerciements à l’ensemble des membres du jury, Prof. Sevser Sahpaz, Prof. Luc Willems et Prof. Jean-Luc Veuthey pour avoir évalué mon travail. Je souhaite exprimer ma gratitude à ma directrice de thèse, Prof. Muriel Cuendet pour m'avoir accueilli dans son équipe, pour avoir lancé et encadré ce projet ; elle a joué un rôle majeur dans la réussite de celui-ci. Je remercie également Prof. Pierre-Alain Carrupt, en tant que co-directeur de thèse et malgré un départ précipité, il m’a apporté de nombreux conseils. J’adresse de chaleureux remerciements à Dr Claudia Simões Pires pour tout ce qu’elle m’a appris, conseillé, enseigné, pour son aide bienveillante et pour tout le temps qu’elle m’a accordé ; son énergie et sa confiance ont été des éléments moteurs pour moi. J’ai pris un grand plaisir à travailler avec elle. Je remercie Dr Alessandra Nurisso dont le soutien, les conseils et l’aide m’ont apporté énormément ; elle a aussi joué un rôle déterminant dans la réussite de ce projet. Ces années de doctorat ont été l’occasion de nombreuses collaborations et rencontres avec des chercheurs, des doctorants et des professionnels de tous horizons qui ont enrichi mon travail par leur expertise. Un grand merci à Prof. Philippe Bertrand, Prof. Sabrina Dallavalle, Prof. Alexandra Detsi, Dr Yung-Sing Wong and Prof. Nadine Martinet, merci pour ces collaborations constructives, merci pour la confiance que chacun m’a accordée. Je voulais également remercier Prof. Jean-Luc Wolfender, Dr Pierre-Marie Allard et Dr Katia Gindro pour leur implication, leurs conseils et leur expertise dans des domaines que je maitrisais mal. Merci à Pierre-Marie d’avoir accepté de co-encadrer le travail de master de Lucie Ory ; son humour, sa positivité, son entrain ont permis de réaliser ce travail avec un véritable plaisir. Merci à Lucie pour son dévouement, merci pour toutes ces heures passées au laboratoire, certains de ces moments resteront à jamais gravés au fond de moi. Je tenais également à remercier Dr Sylvian Cretton et Dr Davy Guillarme pour leur aide précieuse en spectrométrie de masse. Merci à Sylvian de m’avoir formé à l’utilisation de l’UHPLC-ESI-MS/MS et pour tout le temps accordé. Un grand merci à Dr Laurence Marcourt pour son aide dans l’élucidation des spectres RMN, sans elle, ce travail aurait certainement pris un peu plus de temps.

J’ai une pensée particulière pour l’ensemble des membres des groupes de pharmacognosie, de phytochimie et de chimie thérapeutique pour l’appui professionnel, mais aussi personnel dont ils ont fait preuve tout au long de ces années de thèse. Je pense en particulier à Aymeric pour sa profonde gentillesse et son sens aigu de l’humour et du second degré ; Davide pour ses blagues très très amusantes ; Pierre-Marie qui, ne le nions pas, nous fait quand même tous bien rire ; Assan pour son humanité et sa philosophie ; Philippe E. pour son âme de leader qui m’a beaucoup inspiré ; Carolina, pour son aide précieuse, j’ai pu découvrir au fil des années une personne d’une gentillesse rare ; Paul pour son âme anarchiste ; Charlotte pour ces quelques discussions que nous avons su partager. Merci aussi à Chantal, Mark, Antonio, Vera, Joëlle, Quentin, Adlin, Nicolas, Caroline, Noémie. Mes remerciements vont également à Frédéric Borlat, Natalie Schregle, ainsi qu’à Sylvia Passaquay Rion, ce projet n’aurait également pas pu être réalisé sans leur travail.

Me voilà donc au terme de cette thèse qui constitue un chapitre important de ma vie, avec ses hauts et ses bas, ses souffrances et ses satisfactions, ses rires, ses doutes, ses peines, ses succès, ses pensées. Ce chemin encaissé, jamais linéaire, aux embranchements multiples et aux détours nombreux, est celui d’un apprentissage professionnel mais surtout personnel et profond. J’ai la chance d’avoir été accompagné à chaque étape de ce cheminement ; je remercie mes parents et grands-parents pour leur présence affectueuse, merci d’avoir été présents et d’avoir su m’aider dans tout ce que j’ai pu entreprendre. Je remercie aussi ma sœur Flore pour son soutien au cours de ces années.

Je souhaite pour finir exprimer mes plus grands remerciements à mes amis. Doroteia, Mohamed, Soura, Olivier, Sarah, Niloufar, Marija, Aurélien, Imane. Je suis reconnaissant pour l'écoute et le soutien qu'ils ont su me témoigner tout au long de la réalisation de ce travail, merci pour les moments passés, merci pour leur bienveillance à mon égard, leur sagesse et leur amitié.

Abstract

Histone deacetylases are a group of enzymes that adds acetyl groups to the lysine residues of several proteins, including transcription factors, tubulin and histones. They are implicated in the development of a wide range of diseases, such as cancer, cardiovascular disorders, inflammation, neurodegenerative diseases, cell metabolism disorders and even parasitic infections. In recent years, pharmacological inhibition of these enzymes has emerged as an interesting therapeutic strategy to treat these disorders and the search for more potent and structurally diverse HDAC inhibitors (HDACi) has become a hot topic. These inhibitors are either natural products or of synthetic origin. They include pan- HDACi or class-selective and isoform-selective inhibitors. To date, the Food and Drug Administration (FDA) has approved four HDACi for hematological malignancies (SAHA, romidepsin, belinostat, panobinostat), and many others are in various stages of clinical development. However, among all these compounds, class or isoform selective inhibitors are still scarce. The development of such compounds would help to elucidate the biological role of each HDAC isoform and could provide therapeutically effective HDACi with fewer side effects compared to pan-HDACi. This thesis aimed at discovering novel HDACi, with the specific aim of improving the balance between potency and selectivity. It also investigated rational strategies to accelerate the discovery process of HDACi and provided tools to study their behavior in cells.

Many of the first HDACi to be discovered were naturally occurring metabolites such as romidepsin, SAHA, and trapoxin A. With the expectation to find potent and selective natural HDACi, two screening strategies were performed. First, a chemical library containing a series of natural products and derivatives was screened for HDAC inhibitory activity. This work revealed for the first time the HDAC inhibitory ability of 4 aurones. They all showed an IC50 values below 20 µM in the enzymatic assay but seemed to behave as pan-HDACi after testing on HDAC1-3 and HDAC6 isoforms. The second strategy consisted in a bio-guided fractionation performed on the exudate of Salvia corrugata, identified as active in a pre-screening. Three diterpenes (fruticulin A, dimethyl-fruticulin A and fruticulin C) were obtained and displayed an IC50 in the µM range against various HDAC isoforms. Fruticulin C showed some selectivity towards HDAC6, an isoform considered as an interesting target for the discovery of drugs against neurodegeneration, cancer and other diseases.

The search for selective HDACi has become a hot topic. Such compounds have been identified during this thesis. A series of cyclic tetrapeptide analogues has been developed and synthesized. They showed IC50 values in the nM range with up to 60-fold selectivity towards class I HDAC isoforms and may be considered as good candidates for further

testing in various disease models. Moreover, biphenyl-4-yl-acrylohydroxamic acid derivatives with promising in vivo proapoptotic and antitumor activities in a cancer model, showed some selectivity towards HDAC8. Finally, the key role of the tert-butyloxycarbonyl (BOC) group in HDAC6 selectivity could be demonstrated in a series of hydroxamate derivatives.

All of the above-mentioned results have been obtained using a fluorescence-based enzymatic assay. This widely-used approach is appropriate for the screening of chemical libraries because of its reproducibility and rapidity. However, fluorescence detection has the disadvantage of possible interferences with the auto-fluorescence of cell constituents, test compounds or complex mixtures such as natural extracts. Many of these disadvantages can be overcome by the use of mass spectrometry, an approach that tends to be commonly used in enzymatic assays and screenings. Thus, a UHPLC-ESI-MS enzymatic assay was optimized to be used for the bio-guided fractionation of fungal extracts. This strategy led to the identification of two selective HDAC inhibitors from Penicillium griseofulvum and may be considered as a useful tool for high throughput identification of selective HDAC1 or HDAC6 inhibitors in natural extracts. This enzymatic assay was then applied to the development of a fluorescent probe with class I HDAC selectivity, which was investigated by cell microscopy to visualize its cellular distribution. This kind of fluorescent HDACi could be of great value for the design of new generation HDACi with better pharmacokinetics properties. It could also facilitate the study of HDACi molecular interactions in the cell and tissue environments.

The biological functions of HDAC isoforms are not yet fully understood and remain complex. Each HDAC isoform shows a specific cellular behavior. They are differently compartmented within the cell, acting either within complexes or alone, and competing with other cell mechanisms. Several cell-based assays for the determination of protein acetylation levels already exist and could be useful for the elucidation of the molecular cell mechanisms of HDACi. Nevertheless, these approaches can be time-consuming, expensive and not suitable for high-throughput screening. A high-throughput cell-based UHPLC-ESI-MS/MS assay was thus developed and was demonstrated to rapidly predict activity against HDAC1 and HDAC6 in cells, by simultaneous determination of the isoform-specific IC50 values within the living cells.

This is suitable for high-throughput screening and may be considered as a useful tool to accelerate the search for class- selective HDACi in drug discovery. To investigate the potential applicability of the cell-based HDAC assay to models of neurodegeneration, the method was applied to the HDAC evaluation of a selective HDAC6 inhibitor in SH-SY5Y human neuroblastoma cells. This compound was developed to be able to cross the blood brain barrier (BBB), this ability was further corroborated in vitro by a Parallel Artificial Membrane Permeability Assay (PAMPA). Its HDAC6 selectivity was confirmed in cells in which its selectivity towards HDAC6 was close to the one of tubastatin A. This compound should be further tested in complex models of neurodegeneration.

In conclusion, various approaches were used to search for HDACi and led to the identification of various compounds with different potency and selectivity profiles. Some of them are now considered as starting points for the design of next generation HDACi. Others are good candidates for further testing in different disease models such as cancer or neurodegeneration. Finally, the present work brings useful tools to accelerate the search for class-selective HDACi in drug discovery. The development of an HDACi coupled to a fluorophore resulted in a tool to better understand HDAC inhibition and its downstream effects in cells and living organisms.

Résumé

Les histones désacétylases (HDACs) sont un groupement d’enzymes catalysant la réaction de désacétylation des histones ou d’autres protéines. Le dérèglement de ces enzymes a été rapporté dans de nombreuses pathologies (cancer, pathologies cardio-vasculaire, inflammation, maladies neurodégéneratives, parasitologie, pathologies métaboliques). En conséquence, depuis plusieurs années, de nombreux efforts pour cibler ces enzymes dans une optique thérapeutique ont été réalisés. Dans le cadre de ces recherches, de nombreuses molécules inhibitrices des HDAC ont été découvertes et semblent intéressantes dans une logique thérapeutique. Ces composés peuvent être directement issus de produits naturels, d’autres sont le fruit d’un travail de synthèse. Quatre inhibiteurs HDAC, la romidepsin, le SAHA, le belinostat et le panobinostat, ont déjà été acceptés par la FDA pour la prise en charge de divers cancers hémathologiques. De nombreux autres sont en cours d’essais cliniques ou pré-cliniques. Il y a cependant encore trop peu d’inhibiteurs sélectifs à disposition. Ces composés ont la capacité d’inhiber très spécifiquement une isoforme ou une classe d’entre elles et sont aujourd’hui considérés par de nombreux auteurs comme bénéfiques.

Scientifiquement, ils pourraient permettre de mieux comprendre le rôle spécifique de chacune des isoformes HDACs au sein des cellules. En termes thérapeutiques, ils seraient moins toxiques que la plupart des inhibiteurs sélectifs utilisés aujourd’hui en clinique. Ma thèse de doctorat et la proposition de recherche que je présente ici ont pour but de découvrir de nouveaux inhibiteurs HDACs présentant des caractéristiques intéressantes en termes de puissance et de sélectivité. Elle vise également au développement de nouvelles stratégies de criblages qui permettraient de faciliter la découverte de telles molécules ainsi qu’à la mise en place d’outils facilitant l’étude de ces composés.

Un grand nombre de composés inhibiteurs des HDAC sont directement issus de plantes, de micro-organismes ou d’autres sources naturelles. Cette thèse s’est intéressée, entre autres, à la découverte d’inhibiteurs HDAC d’origine naturelle. Deux stratégies de criblages ont été conduites à cette fin : un criblage direct sur une bibliothèque de composés naturels ainsi qu’un fractionnement bioguidé réalisé sur un extrait de Salvia corrugata, une plante ornementale d’origine Américaine. La première approche permit de démontrer les capacités inhibitrices de certaines aurones. La deuxième conduisit à l’isolement de trois diterpénes inhibiteurs des HDACs (fruticuline A, demethylfruticuline A et fruticuline C). L’ensemble de ces composés naturels présentait une puissance inhibitrice toute modérée avec des IC50 dans l’ordre du µM. En termes de sélectivité, les aurones n’en témoignaient aucune vis-à-vis

des isoformes testées. La fruticuline C inhibait quant à elle, bien que sommairement, sélectivement HDAC6, une cible thérapeutique de choix dans de nombreuses maladies.

La recherche visant à l’identification d’inhibiteurs sélectifs est aujourd’hui d’un grand intérêt, ce fut l’un des objectifs majeurs de cette thèse. Plusieurs inhibiteurs sélectifs furent ainsi identifiés au cours de ce travail, inhibant tantôt HDAC6, tantôt les isoformes de classe I. C’est dans cet objectif que, partant de la structure de tétrapeptides cycliques connus pour leurs propriétés inhibitrices HDAC, différents composés furent conçus et synthétisés. Tous inhibaient très spécifiquement les isoformes de classe I avec une puissance remarquable (IC50 dans l’ordre du nM), faisant de ces composés de bons candidats pour des investigations futures. En parallèle de ceci, un dérivé de l’acide biphenylacrylohydroxamique pouvant inhiber sélectivement HDAC8 fut développé. Le manque relatif d’inhibiteur de cette isoforme ainsi que les propriétés antitumorales et proapoptotiques du composé, constituent autant d’éléments poussant à continuer l’étude de celui-ci. Pour finir, l’importance du groupement chimique BOC dans l’obtention d’une sélectivité HDAC6 fut mise en évidence à l’intérieur d’un ensemble de dérivés de l’acide hydroxamique.

L’ensemble des résultats évoqué jusqu’ici a été obtenu par l’intermédiaire d’un essai enzymatique couplé à une détection par fluorescence. Cette approche est aujourd’hui très répandue en épigénétique grâce à sa grande rapidité et à sa reproductibilité. Elle présente cependant quelques points faibles. Certains sont directement imputables à la méthode de détection qui peut être la source d’interférences lors de son application à l’étude de matrices complexes.

Aujourd’hui la détection par spectrométrie de masse commence à s’imposer en enzymatique et pourrait solutionner les problèmes observés. Ce travail de thèse s’est ainsi intéressé au développement d’un essai enzymatique utilisant une détection par spectrométrie de masse. Une fois mise en place, cette approche a permis le développement d’une nouvelle stratégie de fractionnement bioguidé facilitant la découverte d’inhibiteurs sélectifs de la classe I ou de HDAC6 au sein d’extraits naturels. Elle permit également la construction d’un inhibiteur HDAC fluorescent. Ce composé présentait une puissance remarquable (IC50 dans l’ordre du nM) et pouvait sélectivement inhiber les isoformes de classe I. Il fut étudié par microscopie à fluorescence afin de visualiser sa distribution cellulaire. Il y a aujourd’hui encore trop peu d’inhibiteurs HDACs de ce type. Ils pourraient être d’une grande utilité dans l’amélioration des propriétés pharmacocinétiques des inhibiteurs HDACs. Ils devraient également faciliter l’étude des mécanismes pharmacologiques des inhibiteurs HDAC qui restent encore mal compris.

Le rôle biologique lui-même des isoformes HDACs est encore en partie obscur. Celui-ci est particulièrement complexe, chaque isoforme présentant un comportement et des fonctions biologiques singulières. De nombreux essais cellulaires permettant d’étudier les réponses cellulaires des inhibiteurs HDACs existent déjà. Ils sont cependant en général chers, longs et non utilisables lors de criblages hauts débits. L’un des objectifs majeurs de cette thèse fut le développement d’un essai cellulaire permettant une mesure directe des activités des isoformes de classes I et de HDAC6. Cette approche, adaptée au criblage haut débit pourrait, une fois utilisée, accélérer la découverte d’inhibiteurs HDACs sélectifs. Elle peut aussi permettre de mesurer sur différentes lignées cellulaires, les modifications d’activité des isoformes occasionnées par l’utilisation d’inhibiteurs HDAC connus. Cette approche fut considérée dans l’étude d’un inhibiteur HDAC6 identifié au cours de ce travail de doctorat. Celui-ci avait été construit dans l’optique d’être utilisé spécifiquement dans un contexte neuronal. Il possède, d’après un essai in vitro, une très bonne capacité à traverser la

barrière hémato-encéphalique. Considérant les cellules neuroblastiques SH-SY5Y, sa sélectivité HDAC6 et sa puissance sont très proches de celles de la tubastatin A, un inhibiteur HDAC6 reconnu. Tous ces résultats font de cet inhibiteur un outil intéressant afin d’étudier les réponses pharmacologiques entrainées par l’utilisation d’un composé sélectif HDAC6 dans le cadre des pathologies neurodégénératives.

Ce travail de doctorat aura permis l’identification de plusieurs nouveaux inhibiteurs HDAC plus ou moins puissants ou sélectifs. Certains pourront être le point de départ pour l’élaboration de nouvelles générations d’inhibiteurs, d’autres pourront être directement appliqués à différents modèles de maladies. Cette thèse apportera également de nouvelles stratégies de criblages qui permettront peut-être, d’accélérer la découverte de nouveaux inhibiteurs. Elle aura également permis le développement de nouveaux outils qui pourront certainement faciliter l’étude et la compréhension des réponses cellulaires occasionnées par l’utilisation des inhibiteurs des HDACs.

Abbreviations

AAATPase ATPase associated with a variety of

activities DNMTS DNA methyltransferases

Aβ Amyloid β-protein ESI Electrospray ionisation

AcOEt Ethyl acetate FA Formic acid

AD Alzheimer’s disease FACScan Fluorescence-activated cell sorting

Akt Protein kinase B (also known as PKB) FDA Food and Drug Administration

ALS Amyotrophic lateral sclerosis FTDL Frontotemporal dementia

ATCC American type culture collection GSK3-β Glycogen synthase kinase 3-β

BBB Brain blood barrier H4K12 Histone H4 lysine 12

BCRP Breast carcinoma resistant protein HAT Histone acetyltransferase BDNF Brain-derived neurotrophic factor HD Huntington’s disease

BOC t-butyloxycarbonyl HDAC Histone deacetylase

BRCT BRCA1 C-terminal HDACi Histone deacetylase inhibitors

BRET Bioluminescence resonance energy

transfer HDLP Histone deacetylase-like protein

CD I Catalytic domain I HEK Human embryonic kidney cells CD II Catalytic domain II HESI Heated Electrospray Ionization Cdk5 Cyclin-dependent kinase-5 HIV-1 Human immunodeficiency virus type 1 cELISA Competitive enzyme-linked

immunosorbent assay HMG-CoA 5-hydroxy-3-methylglutaryl-coenzyme A

CMT Charco-Marie-Tooth disease HMTs Histone methyltransferases

CNS Central nervous system HPLC High performance liquid chromatography

CRM-1 Chromosome region maintenance

(nuclear receptor) HPLC-MS High performance liquid chromatography coupled

to mass spectroscopy CTCL Cutaneous T-cell lymphoma HSF Heat shock factor protein

DA Dopaminergic HSP Heat shock protein

DCM Dichloromethane KDM Lysine demethylases

DJ-1

Parkinson disease (autosomal recessive, early onset 7 (also known as PARK7)

LBs Lewy bodies

DMF Dimethylformamide MBT Malignant brain tumor

DMSO Dimethyl sulfoxide MeCN Acetonitril

DNA Deoxyribonucleic Acid MEF2 Myocyte enhancing factor-2

MLP Molecular Lipophilicity Potential UV Ultraviolet

MR Membrane retention VCP Valosin-containing protein

MRM Multiple Reaction Monitoring ZBG Zinc binding group

MS Mass spectroscopy ZF-CW Zinc finger CW

MS/MS Tandem mass spectroscopy ZnFUBP Ubiquitin carboxyl-terminal hydrolase-like zinc finger

MS-275 Entinostat

MTT Methyl-thiazolyl-tetrazolium

MW Molecular weight

NAD Nicotinamide adenine dinucleotide NADPH Nicotinamide adenine dinucleotide

phosphate

NCI Non-covalent interaction NCOR2 Nuclear receptor co-repressor 2 NDs Neurodegenerative diseases NES Normal Epithelial Cell Specific-1 NF-ΚB Nuclear Factor Kappa B NMR Nuclear magnetic resonance PADIs Peptidyl arginine deiminases PAMPA Parallel Artificial Membrane

Permeability Assay PBS Phosphate buffer saline PD Parkinson’s disease PDB Protein data bank

PE Petroleum ether

PHD Plant homeodomain

PKMTs Protein lysine methyltransferases PPI Protein-protein interaction PRMT Protein arginine methyltransferases PTCL Peripheral T-cell lymphoma PTM Post-translational modifications RMSD Root mean Square Deviation SAGE Serial analysis of gene expression SAHA Vorinostat

SIRT Sirtuin

SUMO Small Ubiquitin-like Modifier TDP TAR DNA binding protein TFA Trifluoroacetic acid THF Tetrahydrofuran

TLC Thin-layer chromatography TMS Tetramethylsilane TNF-α Tumor necrosis factor-α

TPB Triton-X-100

TSA Trichostatin A

TUNEL Terminal deoxynucleotidyl transferase dUTP nick end labeling

UHPLC Ultra-high pressure liquid chromatography

Table of contents

Aim and structure of the thesis

Aim and scope of the thesis ₃

Organization of the thesis ₄

Chapter I I Introduction

Epigenetics ₇

Histone deacetylases ₉

HDAC inhibitors ₁₀

HDAC inhibitors as therapeutic agents ₁₂

Isoform or class selective HDAC inhibitors: the interest ₁₃ HDAC6 as a target for neurodegenerative diseases: what makes it different

from the other HDACs? ₁₄

Abstract 16

Introduction 17

Structural differences between HDACs: what makes HDAC6 different from the

others? 18

HDACs other than HDAC6 act mainly as epigenetic modulators in cognition and

neuronal death 22

The specific role of HDAC6 in the neurodegenerative cascades 25

The role of HDAC6 in Alzheimer’s disease 29

The role of HDAC6 in Parkinson’s disease 30

The role of HDAC6 in Huntington’s disease 31

The role of HDAC6 in other neurodegenerative diseases 32 Critical insights on HDAC6 as a target to fight against neurodegeneration 32

Supplementary information 33

Chapter 2 I Natural histone deacetylase inhibitors

Outline 51

Introduction ₅₃

Aurones as HDAC inhibitors: Identification of key features ₅₅ Bio-guided fractionation of Salvia corrugata: isolation of three natural

products HDAC inhibitors 70

Chapter 3 I Isoform-selective histone deacetylase inhibitors

Outline ₈₁

Introduction ₈₃

Hydroxyl ketone-based HDAC inhibitors to gain insights into class I HDAC selectivity versus HDAC6

85 Biphenyl-4-yl-acrylohydroxamic acids: Identification of a novel indolyl-

substituted HDAC inhibitor with antitumor activity

98 Cross metathesis with hydroxamate and benzamide BOC-protected alkenes to access HDAC inhibitors and their biological evaluation highlighted intrinsic

activity of BOC-protected dihydroxamates ₁₁₁

Chapter 4 I Enzymatic UHPLC-MS assay for

measuring histone deacetylase activity: an alternative to fluorescence detection

Outline ₁₃₅

Introduction ₁₃₇

UHPLC/MS-based HDAC assay applied to bio-guided microfractionation of

fungal extracts ₁₃₈

Reaching the nucleus with fluorescent cyclic tetrapeptide HDAC inhibitor in

living cells ₁₅₄

Chapter 5 I Tools to identify and study histone deacetylase selective inhibitors in cells

Outline ₁₇₅

Introduction ₁₇₇

Cell-based multi-substrate assay coupled to UHPLC-ESI-MS/MS for a quick

identification of class-specific HDAC inhibitors ₁₇₈

Synthesis of a selective HDAC6 inhibitor active in neuroblasts ₁₈₉

Chapter 6 I Conclusion & future work

General conclusion ₂₀₈

Future work ₂₁₀

1

Aim and

structure of the thesis

2

Aim and scope of the thesis

3

Aim and scope of the thesis

The aim of the work described in this thesis was to identify histone deacetylase inhibitors (HDACi) by using several approaches, mainly pharmacognosy and medicinal chemistry. The study objectives include:

• To identify new HDACi scaffolds from natural product and synthetic compounds;

• To identify class I and HDAC6 selective HDACi;

• To develop strategies for the screening of selective HDACi using UHPLC-ESI-MS/MS. To apply this approach to the bioactivity-guided identification of HDACi in natural extracts;

• To develop a high-throughput UHPLC-ESI-MS/MS strategy to predict the selectivity profile of HDACi in a cell environment. To apply this approach to the study of selective HDACi;

• To evaluate a selective HDACi fluorescent probe for localization studies of HDACi in cells.

Organization of the thesis

4

Organization of the thesis

Chapter I I Introduction

This chapter provides a comprehensive literature review of the current research status of HDACi and their therapeutic interest. This includes a brief introduction to epigenetics and HDACs. The main part of this chapter summarizes the different HDACi, their therapeutic use and the interest of class or isoform selectivity. To conclude, a review article describes the specific role of HDAC6 in neurodegeneration.

Chapter 2 I Natural histone deacetylase inhibitors

Chapter 2 briefly describes the discovery and interest of aurones as HDACi. Then the identification and biological activity of diterpenes isolated from Salvia corrugata are presented.

Chapter 3 I Isoform-selective histone deacetylase inhibitors

This chapter presents the development of a series of more potent and structurally simpler analogues of the naturally occuring cyclopeptide HDACi. Some show up to 60-fold selectivity towards class I isoforms. Then, the identification and biological study of a biphenyl-4-yl-acrylhydroxamic, which showed some selectivity against HDAC8 is described. Finally, the key role of the tert-butyloxycarbonyl (BOC) group in the HDAC6 selectivity in a series of hydroxamate derivatives is demonstrated.

Chapter 4 I Enzymatic UHPLC-MS assay for measuring HDAC activity: an alternative to fluorescence detection

The shortcomings of the fluorogenic assay generally used to measure HDAC activity are presented in chapter 4. This approach suffers from potential optical interferences with pure compounds, fractions and extracts in the case of bio-guided fractionation. Many of these disadvantages could be overcome by the use of a UHPLC-MS/MS-based assay. This chapter presents the development and optimization of such a method applied to the bio-guided identification of HDACi in fungi. In another project, this MS approach was used to develop a selective class I HDACi fluorescent probe to localize HDACi in cells.

Chapter 5 I Tools to identify and study selective histone deacetylase inhibitors in cells

This chapter shows the advantage that would have a HDAC activity cell-based assay in comparison to traditional approaches, which use recombinant isoforms or cell lysates as HDAC sources. In this context, the development of a high-throughput UHPLC-ESI-MS/MS strategy for the simultaneous assessment of HDAC1 and HDAC6 activity directly in cells was developed. This approach was then used with neuroblastoma and the HDAC inhibitory potency of an HDAC6 selective inhibitor developed to cross the blood brain barrier was evaluated.

Chapter 6 I Conclusion & future work

A summary of the outcome of this work, a general discussion and perspectives are provided.

5

Chapter 1

Introduction

6

7

Epigenetics

Epigenetic regulation comprises numerous mechanisms providing regulatory information to the genome without altering its primary nucleotide sequence, information transferred heritably to offspring or from parent to daughter cells.

Epigenetic mechanisms are implicated in various processes, including gene silencing and expression, apoptosis, maintenance of stem cell pluripotency, and X-chromosome inactivation (1). At the molecular level, epigenetic regulators include covalent modifications to the chromatin. This structure, found in the nucleus of eukaryotics cells, consists of a complex of deoxyribonucleic acid (DNA) and proteins, mainly histones. This family of small proteins is composed of five major types called H1, H2A, H2B, H3 and H4. They are mainly composed of basic amino acids, which allow their binding by weakening electrostatic interactions to the negative charge of DNA. Post-translational modifications (PTM) of histones (e.g. histone acetylation, methylation and ubiquitination) or DNA (e.g. DNA methylation) locally modify the chromatin structure to either facilitate or prevent gene transcription (1, 2). The main epigenetic regulators are summarized in Table 1.1 and are classified as writers (enzymes adding groups to DNA/histones), erasers (enzymes removing groups from DNA/histones) and readers (recognition of protein domains) (3-5) (Figure 1.1).

Figure 1.1 І Epigenetic writers, readers and erasers. Epigenetic regulation is a dynamic process. Epigenetic writers such as histone acetyltransferases (HATs) or histone methyltransferases (HMTs) lay down epigenetic marks on amino acid residues on histone tails.

Epigenetic readers such as proteins containing bromodomains, or chromodomains bind to these epigenetic marks. Epigenetic erasers such as histone deacetylases (HDACs) and lysine demethylases (KDM) catalyze the removal of epigenetic marks (6). This figure is adapted from Falkenger et al. (6).

8 Table 1.1 І Main epigenetic mechanisms with the involved enzymes and domains.

Epigenetic mechanism Writer enzymes Eraser enzymes Reader domains

DNA methylation DNMTs Unknown Methyl-CpG binding

domains

Histone acetylation HATs HDACs Bromodomains, PHD

fingers

Histone methylation PKMTs, PRMTs Histone demethylases, PADIs

Chromodomains, bromodomains, Tudor domains, PHD fingers, MBT domains, ZF-CW proteins, WD40, PWWP

Histone phosphorylation Kinases

Protein tyrosine or serine/threonine phosphatases

Chromoshadow domains, 14-3-3 proteins,

BRCT proteins

Histone ubiquitination Ubiquitin conjugases and ligases

Ubiquitin-specific proteases

Bromodomains and PHD fingers

Histone SUMOylation SUMO ligases SUMO-specific proteases SUMO-interacting motif BRCT: BRCA1 C terminus; DNMTs: DNA methyltransferases; HATS: Histone acetyltransferases;

HDACs: Histone deacetylases; MBT: Malignant brain tumor; PADIs: Peptidyl arginine deiminases;

PHD: Plant homeodomain; PKMTs: Protein lysine methyltransferases; PRMTs: Arginine methyltransferases;

SUMO: Small ubiquitin-related modifier; ZF-CW: Zinc finger CW.

9

Histone deacetylases

One of the features determining chromatin structure are the PTM of the amino acids of histones. Acetylation by histone acetyltransferases (HATs) is one of the PTM able to neutralize the positive charge of the lysine tails, believed to result in chromatin relaxation and increased accessibility for transcription factors. On the contrary, deacetylation by histone deacetylases (HDACs) has a repressive impact on transcription (7) (Figure 1.2). Four classes of HDACs have been phylogenetically established, comprising a total of 18 enzymes. The enzymes from classes I, II, and IV are zinc- dependent and usually referred to as classical HDACs, whereas those of class III are NAD⁺-dependent enzymes named sirtuins (SIRT), with differed structural features. Among classical HDACs, class I comprises the constitutively expressed HDACs 1 to 3 and HDAC8. Class II is subdivided into classes IIa (HDAC4, 5, 7, and 9) and IIb (HDAC6 and 10). Class IIa enzymes are all able to shuttle between the nucleus and the cytosol, and show a weaker deacetylase activity. Class IIb is mostly found in the cytosol with a preference for non-histone proteins. Finally, HDAC11 is the sole member of class IV (7-10). Besides that, HDAC isoforms target several non-histone proteins such as α-tubulin, endothelin receptor, estrogen receptor, chaperones (HSP90), transcription factors (p53, NF-κB) (11). These possibilities allow the various HDAC isoforms to play a central role in a variety of cellular events.

Figure 1.2 І The acetylation degree of histones depends on the enzymatic activity balance between HDAC and HAT. HDAC inhibition by HDAC inhibitor (HDACi) results in a hyperacetylation of the histone-tails and a more relaxed chromatin conformation that favors gene expression. On the other hand, compact chromatin is transcriptionally inactive and prevents gene expression. This figure is adapted from Chuang et al. (12).

10

HDAC inhibitors

Because several studies showed that epigenetic perturbations due to a dysfunction in the acetylation process have been involved in the development of various diseases (cancer (13), cardiovascular (14), and neurodegenerative disorders (15, 16)), HDAC inhibition has emerged as an interesting therapeutic strategy to restore the HDAC/HAT balance. Thus, various methods are being used to identify compounds able to inhibit HDAC isoforms.

Today, a number of structurally diverse HDAC inhibitors (HDACi) have been discovered. Each exhibits singular specificities in terms of selectivity and potency against the different HDAC isoforms. The general HDACi pharmacophore consists of three distinct structural parts: the zinc binding group that chelates a zinc residue within the active site, the recognition cap group (cap) responsible for HDAC subtype selectivity and a hydrophobic linker, which links these 2 groups (17, 18) (Figure 1.3). Given their structural diversity, HDACi can be classified based on their class or isoform selectivity. Some HDACi such as trichostatin A (TSA) or vorinostat (SAHA) are pan-HDACi and inhibit indifferently all HDAC isoforms. Others like tubastatin A and entinostat (MS-275) are isoform or class selective inhibitors (19).

Chemically speaking, compounds identified as HDACi are usually divided into different structural categories:

hydroxamic acids, benzamides, cyclic peptides and short chain fatty acids (15, 20). Despite their simple structure, short chain fatty acids exhibit a relative selectivity against the various HDAC isoforms. For example, valproic acid, one of the well-characterized compounds of this class, specifically inhibits HDAC1-3 and HDAC8. Their inhibitory activity, with IC50

values around 1 mM, are considered weak compared to the other HDACi categories (19).

Hydroxamic acids are the widest group of HDACi. They usually exhibit IC50 values in the nM range. Among them, SAHA and TSA are commonly used either for research or in the case of SAHA also for therapy. Even if SAHA could preferentially target class I HDACs, both SAHA and TSA exhibit no real selectivity toward the HDAC isoforms and could be considered as pan-HDACi. This is also the case for belinostat or panobinostat, two HDACi approved by the Food and Drug Administration (FDA) for cancer treatment (15, 19, 21, 22). The large structural diversity found in this group is the origin of a large variability in terms of selectivity and potency. For example, tubastatin A and tubacin are well known to be very potent and selective HDAC6 inhibitors (16). Several others such as scriptaid, selectively inhibit the isoforms of class I (HDAC1, HDAC3, HDAC8) (23). To date, hydroxamic acids are the most widely explored class of HDACi as anti- cancer agents and numerous derivatives are currently studied to increase their ability to selectively inhibit some of the isoforms.

The cyclic peptide class is a structurally complex group of HDACi. It includes romidepsin and apicidin both active in the nanomolar range. Romidepsin was approved by the FDA for the treatment of cutaneous T-cell lymphoma (CTCL). This prodrug is activated in the cell by reduction of the disulfide bond to give potent HDAC1 and HDAC2 inhibition (24).

The fourth group is composed of the benzamides, and consists of MS-275 and mocetinostat. Their distinct chemical structure gives them a real selectively toward class I HDAC isoforms (19).

11

•Pharmacophore model of HDACi •Short chain fatty acids

Valproic acid

•Hydroxamic acids

Tubastatin A SAHA

FDA approved for CTLC treatment

•Cyclic peptides

Belinostat Romidepsin

FDA approved for PTLC treatment FDA approved for CTLC and PTLC treatment

•Benzamides

Panobinostat MS-275

FDA approved for multiple myeloma treatment

Figure 1.3 І Structure of HDACi. In general, the pharmacophore of HDACi is composed of three distinct structural parts: a cap group; a zinc binding group (ZBG); and a linker. Based on their structure, HDACi can be grouped into at least 4 different classes: short chain fatty acids (valproic acid), hydroxamic acids (SAHA, tubastatin A, belinostat, panobinostat), cyclic peptides (romidepsin) and benzamides (MS-275) (25).

PTLC: Peripheral T-cell lymphoma; CTLC: Cutaneous T-cell lymphoma.

12 An increase in HDAC activity is a common feature of many cancers. Thus, the inhibition of HDAC became an interesting target for cancer drug discovery (26-28). HDACi have been shown to be potent apoptosis inducers in several cancer cell lines (29). They also have the ability to modulate cell proliferation, differentiation, angiogenesis or cell cycle arrest (11, 20).

To date, four HDACi have been approved by the FDA (SAHA, romidepsin, belinostat and panobinostat) for treatment of cutaneus cutaneus T-cell lymphoma (CTLC), peripheral T-cell lymphoma (PTCL) (30, 31) or multiple myeloma (32).

SAHA was the first to be approved in 2006for CTLC treatment through oral administration (33). HDAC inhibition activity by this compound has been shown to promote cell cycle arrest, differentiation, and apoptosis in cancer cells at concentrations that have little or no toxic effects on normal cells (34). It showed significant anticancer activity in several types of hematological and solid tumors (35).

The FDA approved romidepsin for CTLC in 2009 (36) and for the treatment of patients with PTCL in 2011 (37). This compound has been shown to inhibite the growth of murine and human solid tumors in mice models (38, 39).

Furthermore, it induced cell cycle arrest and apoptosis in a number of human tumor cell lines such as neuroblastoma (40) and lung carcinoma (41). In the clinic, romidepsin is given intravenously. It seems to be therapeutically well tolerated, although fatigue, nausea, vomiting and thrombocytopenia as well as a certain cardiac toxicity have been reported (42).

Belinostat was approved by the FDA in 2014 (43). It is currently used intravenously for the treatment of refractory PTCL (22, 44). It is in general well tolerated, but some hematological, hepatotoxic and gastrointestinal toxicities were observed (45). In vitro, belinostat significantly inhibited growth in several cell lines, especially 5637 cell line used as a model for high-risk superficial bladder cancer (46). In vivo, it prevented the growth of human tumor xenografts in mice, with no apparent toxicity (47).

Panobinostat was the last HDACi to be approved by the FDA in 2015. It is used in combination with bortezomib and dexamethasone in patients suffering from refractory multiple myeloma (48). In vitro it has been shown to have a marked anti-tumor activity against a broad range of hematological cancer cell lines including CTLC, acute myeloid leukemia, and multiple myeloma. These activities were confirmed in vivo where panobinostat showed anti-tumor effects in a CTLC and multiple myeloma xenograft mouse model (49, 50). Its oral administration is generally well tolerated, however adverse events such as thrombocytopenia, neutropenia, fatigue, diarrhea, electrolyte abnormalities, and increased creatinine can be observed (51).

Besides these current uses, SAHA, romidepsin, belinostat and panobinostat are being evaluated in pre-clinical and clinical trials for the treatment of other types of cancer and others diseases (20, 52, 53). In fact, a growing body of literature assigns to HDACs a functional role in the pathogenesis of several diseases such as inflammatory diseases (54, 55), cardiovascular diseases (14, 56) or neurodegenerative diseases (15, 16). In cardiovascular diseases, the antihypertrophic action of HDACi makes them interesting targets in the treatment of human heart failure (57, 58). In neurodegenerative disorders, many diseases such as Alzheimer’s disease (AD), Parkinson’s disease (PD) or Huntington’s disease (HD) are associated with memory impairments correlated with decrease in histone acetylation. In this context, many studies demonstrated that HDACi have promising effect as treatment against cognitive decline (59). Finally, in inflammation, HDAC isoforms have been shown to be involved in the macrophage development and activation. These relationships established HDACi as potential treatment in inflammatory and infectious diseases.

HDAC inhibitors as therapeutic agents

13 Among the four FDA approved HDACi described above, only romidepsin presents a real selectivity, while SAHA belinostat and panobinostat are considered as pan-HDACi. There are still relatively few isoform-specific HDACi available. To date, selective HDAC1, HDAC2, HDAC3, HDAC6 or HDAC8 compounds have been developed, and several of them are currently under clinical or pre-clinical trials (6). However, it is difficult to assert that the clinical use of selective inhibitors is a better alternative compared to pan-HDAC treatment. Despite these observations, one can hypothesize that the therapeutic use of more selective HDACi could bring several advantages. For example, even if pan-HDACi are generally well tolerated by the patients (60), selective HDACi should be safer by avoiding side effects due to the inhibition of the other HDAC isoforms (61). This could become very useful in cancer therapy where compounds are usually administered at their maximum tolerated dose (57). This selective strategy could also be more efficacious by specifically targeting the most deregulated pathway of the disease. However, to date there is still relatively few information about which HDAC isoform(s) is involved in the various diseases and further studies are necessary.

HDACs isoforms are known to mainly act by epigenetic modulation via chromatin deacetylation, but they may also be implicated in the pathogenesis cascade by other mechanisms (62). Each HDAC isoform presents distinct and specific functions within the disease cascades. Such an example is HDAC6, a singular cytoplasmic isoform for which the role has been partially elucidated in cancer (63), cardiovascular disease (64), inflammation (65) and neurodegenerative diseases (16). In cancer, HDAC6 is essential for an efficient oncogenic transformation by supporting anchorage- independent growth (66). HDAC6 is also critical for metastasis and cell migration, and treatment with a selective HDAC6 inhibitor is sufficient to reduce invasion motility without acting on histones (67, 68). More specifically in multiple myeloma cells, HDAC6 inhibition by tubacin led to an accumulation of misfolded ubiquitinated proteins and apoptosis making this isoform a potential target in this form of cancer (63, 69). In cardiovascular disorders, HDAC6 function is insufficiently understood. However, because its inhibition in cardiomyocytes leads to cardioprotection, this isoform could become interesting in the future (64). In inflammation, the hypothesis that HDAC6 activity is critical for lipopolysaccharide-induced activation of macrophages suggests that HDAC6 inhibition could be a therapeutic strategy in the management of inflammatory disorders (70). Finally, neurodegenerative diseases are diseases where the role of HDAC6 has been the best studied. The HDAC6 involvement within the pathogenesis mechanism of disorders such as AD, PD and HD makes it an interesting target for the discovery of drugs against neurodegeneration. The following section presents the interests of targeting HDAC6 compared to other HDAC isoforms in such diseases.

Isoform or class HDAC selective inhibitors: the interest

14

Foreword

Context

To understand the interest of the identification of selective HDAC6 inhibitors, this section provides, in the context of neurodegenerative diseases, a comprehensive review of the specific biological role of HDAC6 compared to other HDAC isoforms. It also summarizes the in vitro activity and in silico data of the commonly used HDAC6 specific inhibitors in these diseases.

Contribution

Together with Dr. Claudia Simões Pires, I reviewed the literature and drafted the manuscript, except for the docking part that was written by Dr. Alessandra Nurisso.

HDAC6 as a target for neurodegenerative diseases:

what makes it different from the other HDACs?

This part is based on an article published in Molecular Neurodegeneration

15

HDAC6 as a target for

neurodegenerative diseases: what makes it different from the other HDACs?

Claudia Simões-Pires^a,*,Vincent Zwick^a,*, Alessandra Nurisso^a, Esther Schenker^b, Pierre-Alain Carrupt^a, Muriel Cuendet^a. HDAC6 as a target for neurodegenerative diseases: what makes it different from the other HDACs? Molecular Neurodegeneration, 2013, 8: 1-16.

aSchool of Pharmaceutical Sciences, University of Geneva, University of Lausanne, Quai Ernest-Ansermet 30, 1211 Geneva 11, Switzerland

bInstitut de Recherches Servier, Rue de la République 3, 92150 Suresnes, France.

* These authors contributed equally to this work.

16

Abstract

HDACi have been demonstrated to be beneficial in animal models of neurodegenerative diseases. Such results were mainly associated with the epigenetic modulation caused by HDACs, especially those from class I, via chromatin deacetylation. However, other mechanisms may contribute to the neuroprotective effect of HDACi, since each HDAC may present distinct specific functions within the neurodegenerative cascades. Such an example is HDAC6 for which the role in neurodegeneration has been partially elucidated so far. The strategy to be adopted in promising therapeutics targeting HDAC6 is still controversial. Specific inhibitors exert neuroprotection by increasing the acetylation levels of α- tubulin with subsequent improvement of the axonal transport, which is usually impaired in neurodegenerative disorders. On the other hand, an induction of HDAC6 would theoretically contribute to the degradation of protein aggregates which characterize various neurodegenerative disorders, including AD, PD and HD. This review describes the specific role of HDAC6 compared to the other HDACs in the context of neurodegeneration, by collecting in silico, in vitro and in vivo results regarding the inhibition and/or knockdown of HDAC6 and other HDACs. Moreover, structure, function, subcellular localization, as well as the level of HDAC6 expression within brain regions are reviewed and compared to the other HDAC isoforms. In various neurodegenerative diseases, the mechanisms underlying HDAC6 interaction with other proteins seem to be a promising approach in understanding the modulation of HDAC6 activity.

Keywords: Histone deacetylase, HDAC6, Neurodegenerative diseases

17

Introduction

HDACs are enzymes that deacetylate lysine residues from histones as well as from several other nuclear, cytoplasmic and mitochondrial non-histone proteins. In mammals, 18 HDACs have been phylogenetically classified into four classes.

Classes I, II, and IV belong to the Rpd3/Hda1 family (71). Class I includes the constitutively expressed HDACs 1 to 3 and HDAC8 (72). Class II is subdivided into classes IIa (HDAC4, 5, 7, and 9) and IIb (HDAC6 and 10). Enzymes from class IIa are able to shuttle between the cytosol and the nucleus, and show a weaker deacetylase activity (73). Class IIb is mostly found in the cytosol with a preference for non-histone proteins (74), whereas HDAC11 is the sole member of class IV.

These HDACs are usually referred as classical HDACs, whereas class III, called SIRTS, are NAD⁺ dependent enzymes with different structural features (75).

The role of HDACs has been studied within several cell processes based on phenotypic changes after isoform-specific knockdown or treatment with HDACi. The consequences of an inhibition of HDACs may result in contradictory results, which seem to depend partially on cell type (76). Knockout analyses of various class I and class II HDAC proteins suggested that class I HDACs are involved in cell proliferation and survival and are expressed ubiquitously in different body tissues, while class II HDACs seem to have tissue-specific roles (77, 78). Moreover, the specific role of each HDAC is directly related to their specific molecular substrates. To our knowledge, more than 50 non-histone proteins have been identified as substrates for HDACs (15). On the basis of animal tissue expression and serial analysis of gene expression (SAGE) data from the human transcriptome map (77, 79, 80), the distribution of HDAC isoforms in body tissues is presented in Figure1.4 together with their distribution in rat brain (80). It is also important to notice that the level of expression may differ when specific pathologies are present, such as cancer, where some HDAC isoforms are overexpressed (79).

The deacetylase activity of HDACs is opposed to that of HATs and several studies have demonstrated the relevance of the HDAC/HAT enzymatic balance in neuronal homeostasis (81). This balance is involved in neurophysiological functions, memory processes and learning. A deregulation of HDAC/HAT activity has been observed in several Neurodegenerative diseases (76), and a decrease in histone acetylation levels may affect the expression of genes involved in apoptosis and neuroprotection (81-83).

Several reviews discussed the importance of various HDACs in specific Neurodegenerative diseases (84, 85) and recently, HDAC6 was suggested to be a promising target for some of them (86). In the present work, we aim at reviewing the published data regarding HDAC6. The structural and functional features of this specific isoform are compared to other classical HDACs. The specific role of HDAC6 in Neurodegenerative diseases is discussed and the impact of HDAC6 modulation via inhibition, induction and interaction with other proteins in various diseases such as AD, PD and HD, frontotemporal dementia (FTDL), amyotrophic lateral sclerosis (ALS) and Charcot-Marie-Tooth disease (CMT) is discussed.

18

Structural differences between HDACs: what makes HDAC6

different from the others?

The bacterial HDAC homologue histone deacetylase-like protein (HDLP) from Aquifex aeolicus was the first HDAC-like protein structure solved by X-ray in 1999 (87). Alignment studies combined with structural analyses revealed the presence of a conserved 11 Å deep channel among all HDAC structures, with a zinc ion located at the bottom (88-90).

The zinc-dependent catalytic action consists in the removal of acetyl groups from lysine residues belonging to histone or non-histone proteins (91). Even if a certain degree of homology in the catalytic domains was found, the so-called zinc-dependent HDACs have been classified into three families (classes I, II and IV) depending on their primary sequence similarity to homologous enzymes from Saccharomyces cerevisiae (92).

Class I is characterized by four ubiquitous and relatively small enzymes (∼500 amino acids) essentially located in the nucleus of cells (92, 93). HDACs 1 to 3 were found in complex with specific transcriptional co-repressors, blocking the expression of tumor suppressor genes (94). Interestingly, these enzymes share an internal dynamic cavity adjacent to the catalytic pocket that seems to play a role in facilitating the egress of the enzymatic products from the active site (87, 89). Another nuclear zinc-dependent HDAC, HDAC11, was found to be closely related to class I. However, this enzyme did not show enough identity to this class to be placed in it and a new class (IV) was proposed with HDAC11 as the only member (79).

Class II consists of six larger enzymes (∼1,000 amino acids) that can be further classified according to their sequence homology and domain organization into classes IIa and IIb (79). The N-terminal domain found in class IIa members is the one responsible for nuclear-cytoplasmic shuttling through a phosphorylation-dependent binding to specific 14-3- 3 proteins. Such interactions regulate the activity of transcription factors such as the myocyte enhancing factor-2 (MEF2), which exerts a repressor role in a variety of biological functions, from myogenesis to Epstein-Barr virus transcriptional regulation (94). Moreover, this class shows another zinc ion coordinated to a Cys-Cys-His-Cys motif close to the cavity that may participate in substrate recognition or in protein interactions (95).

The presence of two catalytic domains in HDAC6 allows this isoform to be classified into class IIb together with HDAC10.

While HDAC10 has catalytically inactive domains whose biological function is still unknown, HDAC6 was shown to take part in the microtubule network by acting as a specific α-tubulin deacetylase. Moreover, HDAC6 was able to deacetylate other substrates and to bind ubiquitin, thus modulating cell protective response to cytotoxic accumulation of misfolded and aggregated proteins (86, 96, 97).

To be fully understood, the biological role of HDAC6 requires a deep structural knowledge. It is known that the 1215 amino acid residues characterizing the human HDAC6 are arranged in the space to form two independent catalytic domains with a zinc finger ubiquitin binding domain located at the C-terminus (87). In the enzymatic structure, it is also important to highlight the presence of a zone characterized by a Ser-Glu containing a tetradecapeptide repeating domain (SE14), responsible for HDAC6 intracellular retention and tau interaction. There are as well two leucine-rich nuclear export sequences (NES1, NES2), which play an essential role in the cytoplasmic/nuclear shuttling process (98, 99) (Figure 1.5). What makes HDAC6 unique among all HDAC enzymes is the presence of the C-terminal zinc finger

19 domain able to recognize unanchored C-terminal diglycine motif of ubiquitin characterizing aggresomes (96, 100). This domain, recently solved by X-ray, alone and in complex with ubiquitin, is formed by a compact structure of 5 anti- parallel β-strands, 2 α-helices, and 3 zinc ions with a distinct aromatic pocket (96). Such a three-dimensional organization is similar to other human zinc finger domains recognizing ubiquitin (101, 102). Ubiquitin interacts with HDAC6 mainly through a hydrogen bonding network. The last three residues of ubiquitin are found in an extended conformation stabilized by interactions with the HDAC6 aromatic pocket. Arg 1155 and Tyr 1156 residues act as gatekeepers, moving the ubiquitin binding site from an open to a closed conformation (96). While the mechanism of aggregate recruitment by HDAC6 via ubiquitin is known from a biological (86) and structural (96) point of view, crystallographic information about the two catalytic domains is still missing. The lack of such information is quite problematic for the conception of isoform selective compounds able to modulate HDAC6 activity. Nowadays, this issue is overcome through generation and refinement of three-dimensional HDAC6 homology models combined with computational interaction and molecular dynamics studies. Whereas the design of HDAC6 inducers has never been the object of scientific studies, recent reviews accurately describe the structural features that may be interesting for the design of selective HDAC6 inhibitors (89, 93, 103).

Computational and in vitro results were put together to investigate the structural origin of selectivity of the HDAC6 specific inhibitor tubacin (104)]. In particular, docking and MD calculations highlighted differences in the shape of HDAC surfaces surrounding the binding site. Moreover, the relatively high flexibility of HDAC6 pocket allowed protein conformational changes by accommodating the cap portion of the studied ligands. These findings were also confirmed by Charrier et al. (105) and by Kozikowski et al. (106), through in vitro and docking studies of a set of phenylisoxazole- containing hydroxamates showing IC50 values as low as 2 pM for HDAC6. In both studies, the design and modeling of specific inhibitors were based on the differences found in the region adjacent to the HDAC6 catalytic channel, the so- called cap domain. The HDAC6 homology model built by Butler et al. (10) revealed that, while the active site is highly conserved among HDACs, the cap domain differs greatly in terms of shape and properties. Moreover, the rim of the catalytic channel appears wider and shallower in HDAC6 compared to the other HDAC channels. Thus, compounds with bulkier and shorter aromatic moieties were designed. For instance, HDAC6 selectivity was enhanced by adding a large and rigid cap group, such as in tubastatin A (10). Accordingly, in vitro and docking studies on homology models confirmed the HDAC6 selectivity of a series of pyridylalanine-containing hydroxamic acid derivatives (107).

20 Class HDAC

Tissue Distribution

(77)

Expression level in rat brain [119] Subcellular localization

(80) Per total

HADC* Per region in rat brain**

I

HDAC1 Ubiquitous 6

Co Nucleus

Cytoplasm (in response to

axonal damage) (15)

Ca/Pu GP Am Hi SNpc SNpr LC

HDAC2 Ubiquitous 5

Co

Nucleus (78)

Ca/Pu GP Am Hi SNpc SNpr LC

HDAC3 Ubiquitous 2

Co

Nucleus Cytoplasm (78)

Ca/Pu GP Am Hi SNpc SNpr LC

HDAC8 Ubiquitous 8

Co

Nucleus (78)

Ca/Pu GP Am Hi SNpc SNpr LC

IIa

HDAC4

Brain Heart Skeletal

muscle

4

Co

Nucleus Cytoplasm Synapsis (15)

Ca/Pu GP Am Hi SNpc SNpr LC

HDAC5

Brain Heart Skeletal

muscle

3

Co

Nucleus Cytoplasm

(108)

Ca/Pu GP Am Hi SNpc SNpr LC

HDAC7

Brain Heart Placenta Pancreas Skeletal

muscle

9

Co

Nucleus Cytoplasm

(108)

Ca/Pu GP Am Hi SNpc SNpr LC

HDAC9 _Skeletal ^Brain muscle

10

Co

Nucleus Cytoplasm

(108)

Ca/Pu GP Am Hi SNpc SNpr LC

Figure 1.4 І HDAC isoforms distribution in tissues and rat brain regions, as well as their subcellular localization. Am: Amigadala, As:

Astrocytes, Ca/Pu: Caudate/Putamen, Co: Cortex, GP: Globus palidus, Hi: Hippocampus, LC: Locus coeruleus, Ne: neurons, Ol:

oligodendrocytes, SNpc: Substantia nigra compacta, SNpr: Substantia nigra reticulata, VEC: Vessel endothelial cells; * classified from 1 (most expressed HDAC isoform) to 11 (less expressed HDAC isoform); ** diagrams are a graphical representation of the relative expression of each HDAC isoform in a scale from low to high (0–5), adapted from Broide et al. (80, 108, 109).