Structure-Based Drug Designing Recommends HDAC6 Inhibitors ToAttenuate Microtubule-Associated Tau-PathogenesisAmir Zeb, Chanin Park, Shailima Rampogu, Minky Son, GiHwan Lee, and Keun Woo Lee*

Division of Life Sciences, Division of Applied Life Sciences (BK21 Plus), Plant Molecular Biology and Biotechnology ResearchCenter (PMBBRC), Research Institute of Natural Science (RINS), Gyeongsang National University (GNU), 501 Jinju-daero, Jinju52828, Republic of Korea

*S Supporting Information

ABSTRACT: Protein acetylation and deacetylation play vital roles in the structuraland physiological behavior of target proteins. Histone deacetylase 6 (HDAC6) remainsa key therapeutic target in several chronic diseases such as cancer, neurodegenerative,and hematological diseases. In tau-pathogenesis, HDAC6 tightly regulates microtubule-associated tau physiology, and its inhibition suppresses Alzheimer’s phenotype. To thisend, the current study has identified novel HDAC6 inhibitors by structure-based drugdesigning method. A pharmacophore was generated from HDAC6 in complex withtrichostatin A. The selected pharmacophore had five features including two hydrogenbond donors, one hydrogen bond acceptor, and two hydrophobic features.Pharmacophore validation obtained the highest GH score of 0.80. By applyingLipinski’s rule of f ive and ADMET Descriptors, a drug-like database of 29 183 moleculeswas generated from the Zinc Natural Product Database. The validated pharmacophorescreened 841 drug-like molecules and was subsequently subjected to moleculardocking in the active site of HDAC6. Molecular docking identified 11 hits, where they showed the highest ChemPLP score(>90.00), stable conformation, and hydrogen-bond interactions with catalytic residues of HDAC6. Finally, molecular dynamicssimulation identified three molecules as potent HDAC6 inhibitors with stable root-mean-square deviation and the highestnumber of hydrogen bonds with the catalytic residues of HDAC6. Overall, we recommend three novel inhibitors of HDAC6,capable of suppressing the microtubule-associated tau-pathogenesis.

KEYWORDS: HDAC6 inhibition, tau-pathogenesis, neurological disorders, pharmacophore modeling, docking, and MD simulation

■ INTRODUCTION

Protein deacetylation is a canonical phenomenon and plays acritical role in cellular processes.1,2 Histone deacetylases(HDACs) drive protein deacetylation, where they removethe acetyl group (CH3CO) of acetylated-lysine in histone aswell as nonhistone proteins.3 In eukaryotes, HDACs arephylogenetically classified into four distinct classes. Class I(HDACs 1, 2, 3, and 8), class II (HDACs 4, 5, 6, 7, 9, and 10),and class IV (HDAC11) HDACs are zinc-dependentdeacetylases and differ from NAD+-dependent class IIIHDACs.4 Inhibition of HDACs has been proved to be apromising strategy in several diseases, including hematologicaldiseases, solid tumors, and immunological and neurologicaldisorders. HDACs inhibition is a well-established therapeuticstrategy in cancer treatment. Several studies have reported thatvorinostat or suberoylanilide hydroxamic acid (SAHA),romidepsin (FK-228), panobinostat (LBH-589), and belino-stat are FDA-approved drugs for the treatment of cutaneous T-cell lymphoma (CTCL) and peripheral T-cell lymphoma(PTCL).5−7 Other inhibitors of HDACs including Class Iselective chidamide (CS-055) and entinostat (SNDX-275) areat various stages of clinical development.8,9 To date, reportedHDACIs preferentially inhibit multiple HDAC isoforms and,hence, are prone to cause undesirable side-effects, including

bone marrow suppression, weight loss, fatigue, and cardiacarrhythmias.10,11 These multifaceted effects of availableHDACIs impede the selective role of a particular isozymeand its mechanism of inhibition. In particular, it is unclear ifthe isoform selectivity may provide substantial advantages overbroad-spectrum inhibition.12

Histone deacetylase 6 (HDAC6) is a unique member ofHDAC family enzymes with two catalytic domains: CDI andCDII.13−15 HDAC6 is anchored in the cytoplasm by serine/glutamate-rich repeat motifs.16 Because of the presence of twocatalytic domains, HDAC6 has broader substrate specificitythan other histone deacetylases.1,15,17−19 This enzymedeacetylates nonhistone substrates including heat shockprotein (HSP90), α-tubulin, cortactin, and peroxiredoxins.20,21

HDAC6 has currently been targeted as a potential therapeuticstrategy in several cancers.22−25 It is noteworthy that HDAC6inhibition is not associated with severe toxicity, and itsdeficiency did not lead to embryonic lethality inmice.13,17,26−30

Received: August 6, 2018Accepted: November 8, 2018Published: November 8, 2018

Research Article

pubs.acs.org/chemneuroCite This: ACS Chem. Neurosci. XXXX, XXX, XXX−XXX

© XXXX American Chemical Society A DOI: 10.1021/acschemneuro.8b00405ACS Chem. Neurosci. XXXX, XXX, XXX−XXX

Dow

nloa

ded

via

GY

EO

NG

SAN

G N

AT

L U

NIV

on

Dec

embe

r 10

, 201

8 at

06:

56:5

7 (U

TC

).

See

http

s://p

ubs.

acs.

org/

shar

ingg

uide

lines

for

opt

ions

on

how

to le

gitim

atel

y sh

are

publ

ishe

d ar

ticle

s.

HDAC6 is critically involved in neurological disorders with aspecial emphasis on Alzheimer’s disease.21,31 Previously, it isreported that overexpression of HDAC6 reduced tubulinacetylation, thereby increasing cell motility.32 In another study,Hubbert et al.13 revealed that HDAC6 regulates microtubuledynamics by deacetylating lysine at position 40 (K40) in α-tubulin subunit. Furthermore, Guo et al.33 reported thatpharmacological inhibition of HDAC6 increased α-tubulinacetylation, endoplasmic reticulum (ER)-mitochondrial over-lay, and restored the axonal transport defects in patient-derivedmotor neurons of fused in sarcoma-amyotrophic lateralsclerosis (FUS-ALS) patients. In another study, d’Ydewalle etal.34 rescued axonal transport defects in dorsal root ganglion(DRG) neurons from transgenic mouse model of Charcot-Marie-Tooth disease (CMT2). Moreover, it has studied thatacetylated-tau alone is sufficient to promote tau aggregation,synapse degeneration, and cognitive deficits.35,36 Ding et al.37

identified that HDAC6 interacts tau protein and demonstratedas a potential modulator of tau phosphorylation andaccumulation. Thereafter, Cook et al.38 reported thatsuppressing HDAC6 activity promotes tau clearance. Inanother study, it was confirmed that acetylation of tau onKXGS motifs inhibits its phosphorylation on the same motif,and it prevents tau aggregation.39 Furthermore, it wasidentified that this motif is a suitable deacetylation target forHDAC6, where BBB-permeable HDAC6 inhibitor couldenhance acetylation and decrease phosphorylation of tau’sKXGS motifs in vivo.39 Additionally, Carlomagno et al.40

investigated that acetylation of tau on Lys-321 (within a KCGSmotif) is both essential for acetylation-mediated inhibition oftau aggregation in vitro and a molecular tactic for preventingphosphorylation on the downstream Ser-324 residue. Thesame study also found HDAC6-regulated phosphorylationevent, and it was examined that phosphorylation of Ser-324interferes with the normal microtubule-stabilizing function oftau.40 The involvement of HDAC6 in tau-pathology wasupdated when Tseng et al.41 investigated that HDAC6arbitrated endogenous neuritic tau-pathology. In a recentstudy, Fan et al.42 demonstrated that HDAC6 inhibitionameliorates tau phosphorylation and cognitive deficits in anAlzheimer’s disease model. These evidence inspire theidentification of novel antagonists of HDAC6 to targetneurological diseases. Current HDAC6 inhibitors have seriouslimitations such as toxicity, unfavorable pharmacokinetics,expensive synthesis (e.g., Tubacin),17,43 and broad substratespecificity (Ricolinostat and Nexturastat).44,45 Moreover,tubastatin A is a carbazole hydroxamic acid which hasantihepatitis C activity and bypassing the neuronal toxicity ofunselective HDACs inhibitors.46 Therefore, the inhibition ofHDAC6 has emerged as a promising platform against variouschronic diseases including cancer, hematological diseases,Alzheimer’s disease, Parkinson’s disease, and Huntington’sdisease.43,44,47,48

In this study, we identified novel HDAC6 inhibitors byemploying structure-based drug designing and recommendedtheir role in microtubule-associated tau-pathogenesis. Herein,the protein−ligand interaction of HDAC6-TSA complex wasexploited, and a structure-based pharmacophore was developedand validated by a decoy test method. The validatedpharmacophore had five features including two hydrogenbond donors, one hydrogen bond acceptor, and twohydrophobic features, which complimented the hot spots ofHDAC6. The validated pharmacophore was used as a 3D

query to screen a drug-like database and 841 candidatemolecules were retrieved. Furthermore, molecular docking wasemployed, and the candidate inhibitors of HDAC6 wereidentified. Finally, the three drug-like molecules were proposedas potent HDAC6 inhibitors by molecular dynamics (MD)simulation. The novel inhibitors showed stable root-mean-square deviation (RMSD), stable binding orientation, andhydrogen bond interactions with the catalytic residues ofHDAC6. Therefore, this study recommends three novelinhibitors of HDAC6 to attenuate tau-pathogenesis byenhancing tau acetylation at KXGS motifs and suppressingHSP90-mediated tau phosphorylation.

■ RESULTS AND DISCUSSIONStructure-Based Pharmacophore Generation. The

structure of human HDAC6 (PDB ID: 5EDU)49 in complexwith Trichostatin A (TSA) was taken, and subsequentlyexploited for structure-based pharmacophore modeling (Figure1). It has been identified that polar interactions are essential

features of the HDAC6 inhibition.49,60,61 The active site ofhuman HDAC6 is composed of His610, His611, Asp649,His651, and Tyr782 residues.5,18,49,62 A total of 10pharmacophore models were generated by the Receptor−ligandPharmacophore Generation protocol implanted in DiscoveryStudio v4.5 (DS). All the pharmacophores were characterizedin terms of total number of features, types of features, andselectivity score (Table 1).Since the top four phamacophores obtained the same

selectivity score of 9.84, they were critically inspected for thecomplementation of their features against the catalytic residuesof HDAC6. Our results suggested that the top-rankedpharmacophore model had best fitted complementary featuresagainst the catalytically active residues of HDAC6. Therefore,the top ranked pharmacophore (Hypo1) was selected. Thecorrespondent pharmacophore displayed that HBDs comple-mented His610 and Ser568, HBA complemented Tyr782,while one HYP complemented Phe680 and the other issandwiched between the backbone atoms of Phe769 and sidechain of Leu749, respectively (Figure 1). The selectedpharmacophore had a maximum of five features includingHBD, HBA, and HYP (Figure 2A). The interfeature distance

Figure 1. Structure-based pharmacophore generation. Residues ofHDAC6 complementing pharmacophoric features are shown as thinstick. Bound inhibitor (TSA) is shown as light brown thick stickmodel. HBA, HBD, and HYP are colored as green, magenta, and cyan,respectively.

ACS Chemical Neuroscience Research Article

DOI: 10.1021/acschemneuro.8b00405ACS Chem. Neurosci. XXXX, XXX, XXX−XXX

B

constraints of the desired pharmacophore are depicted inFigure 2A. Pharmacophore alignment on TSA showed that allthe features of the selected pharmacophore were mapped(Figure 2B). These analyses argued that if a compound havingthe same complementary features should inhibit HDAC6.Pharmacophore Validation. The selected pharmaco-

phore was evaluated for its efficiency to differentiate betweenthe active and inactive molecules of HDAC6. The Guner−Henry approach (Decoy test method) was employed tovalidate the selected pharmacophore. A test set was designed,which comprised 96 active inhibitors and 115 inactivemolecules of HDAC6 (Table 2). The active and inactivemolecules of HDAC6 were retrieved from the literature on thebasis of their IC50 values. The active inhibitors of HDAC6 hadlow IC50 value (<100 nM), whereas the inactive molecules hadhigher IC50 value (>20 000 nM). The validation resultsaffirmed that the selected pharmacophore could map 91.66%of the active inhibitors of HDAC6 (Table 2). Moreover, theselected pharmacophore obtained a high Guner−Henry (GH)score of 0.80 and enrichment factor (EF) score of 1.93 (Table2). Other parameters like percent yield and ratio of actives,false negatives, and false positives are depicted in Table 2. Allthese results advocated that the selected pharmacophore willretrieve the active inhibitors of HDAC6.Development of Drug-Like Database and Virtual

Screening. A chemical compound should fulfill certain criteriato be treated as a drug-like molecule. In the drug discoveryprocess, such criteria are assigned to physiochemical character-ization, pharmacodynamics, and pharmacokinetics propertiesof drug molecules. In this study, Lipinski’s rule of f ive (ROF)

was applied to Zinc Natural Product Database (http://bioinf-applied.charite.de/supernatural_new/index.php?site=contact)and the physiochemical characterization profile of drug-likemolecules was obtained. According to this rule, an orally activedrug-like molecule should be membrane permeable and easilyabsorbed if it has properties including HBA ≤ 10, HBD ≤ 5,molecular weight ≤ 500 Da, and AlogP ≤ 5.44 Our resultsshowed that 187 887 compounds have successfully passedROF (Figure 3). The pharmacopdynamic and pharmacokineticproperties of drug molecules are very important in terms ofabsorption, distribution, metabolism, excretion, and toxicity(ADMET). Herein, the drug-like compounds that successfullysatisfied ROF were subjected to ADMET Descriptors protocolimplemented in DS. The drug-like compounds were evaluatedfor their potential to cross the blood-brain barrier (BBB),solubility, hepatotoxicity, cytochrome toxicity, and absorption(Figure 3). Since we targeted tau-pathology, which essentiallyoccurs in brain, we therefore considered drug-like moleculesthat may cross blood-brain barrier. Such compounds shouldhave a BBB value ≤ 2 (Figure 3). Finally, a total of 29 183molecules were identified and treated as drug-like database inthe virtual screening process.Virtual screening has potential uses in the drug discovery

process. The validated pharmacophore was employed as a 3Dquery to screen the drug-like database. Herein, the in-builtLigand Pharmacophore Mapping protocol of DS was employedwith Best/Flexible modules to screen the drug-like database. Atthe end, a total of 841 drug-like molecules were identified asHDAC6 candidate inhibitors (Figure 3).

Table 1. Receptor−Ligand-Based Pharmacophores andTheir Characteristicsa

s. no. number of features features setb selectivity score

pharmacophore_1 5 ADDHH 9.84pharmacophore_2 5 ADDHR 9.84pharmacophore_3 5 ADDH 9.84pharmacophore_4 5 ADDHR 9.84pharmacophore_5 5 AADHR 8.93pharmacophore_6 5 AADHH 8.93pharmacophore_7 5 AADHR 8.93pharmacophore_8 5 AADHH 8.93pharmacophore_9 4 ADDH 8.33pharmacophore_10 4 ADDH 8.33

aPharmacophores, number of features, types of features, andselectivity scores are depicted. bFeatures set: A, hydrogen bondacceptor; D, hydrogen bond donor; H, hydrophobic; R, ring aromatic.

Figure 2. Selected pharmacophore. (A) Pharmacophore features and interfeature distance constraints: HYP, hydrophobic; HBA, hydrogen bondacceptor; HBD, hydrogen bond donor. (B) Mapping of TSA onto selected pharmacophore.

Table 2. Decoy Test Validationa

s. no. parametercalculatedvalues

1 total number of molecules in the database 2112 total number of active molecules of HDAC6 in the

database (A)96

3 total number of pharmacophore retrieved hits (Ht) 1004 total number of HDAC6 active molecules in the

retrieved hits (Ha)88

5 % yield of actives [(Ha/Ht) × 100] 88%6 % ratio of actives [(Ha/A) × 100] 91.66%7 false negative [A − Ha] 88 false positive [H − Ha] 129 goodness of fit 0.8010 enrichment factor (EF) [(Ha × D)/(Ht × A)] 1.93

aPhamacophore evaluation predicted that the selected pharmaco-phore retrieved active molecules of HDAC6 during the virtualscreening of drug-like database.

ACS Chemical Neuroscience Research Article

DOI: 10.1021/acschemneuro.8b00405ACS Chem. Neurosci. XXXX, XXX, XXX−XXX

C

Molecular Docking Simulation. Molecular docking is awell-known approach in computational biology to identify andcharacterize molecular interactions of receptor−ligand com-plex.62−64 Herein, the structure of human HDAC6 (PDB ID:5EDU) was taken as receptor.49 The protein structure wasprepared for docking, and the active site was determined frombound inhibitor (TSA) by Def ine and Edit Binding Site modulein DS. All the pharmacophore-retrieved drug-like molecules aswell as the reference compound (TSA) were docked into theactive site of HDAC6. During docking simulation, residues inclose vicinity (10 Å) of the reference compound wereconsidered. For each drug-like molecule, 100 conformerswere created via Genetic Algorithm (GA) implanted in GOLDsoftware. Furthermore, the in-built scoring functionsChemPLP and ASP were used as the default scoring andrescoring functions for pose selection, respectively.53 Ourresults demonstrated that the ChemPLP and ASP scores forthe reference compound were 87.34 and 30.76, respectively(Table S1). These values were used as cutoff values for the

selection of candidate inhibitors of HDAC6 (Table S1).Furthermore, the number of candidate inhibitors of HDAC6was reduced by clustering analysis. It is recommended thatgreater the number of conformers in a single cluster, the higherwill be the presumed stability of molecular orientation. Finally,those candidate molecules were selected that have showedhydrogen bond interactions with catalytic residues of HDAC6.The selected candidate molecules displayed hydrogen bondinteractions with His610, His611, Asp649, His651, and Tyr782(Table S1). A total of 11 candidate molecules were shortlistedand subjected to molecular dynamics simulation.

Evaluation of Stability of HDAC6−Ligand Complex.Molecular dynamics simulation is an attractive approach tostudy molecular interactions, binding mode analysis, bindingstability, and molecular motions of biomolecules at atomisticlevel.65 Herein, the docked molecules were further assessed by30 ns MD simulation to get a deeper insight in their time-dependent binding pattern, optimized binding orientation, andbinding stability.66−68 Our results observed stable root-mean-square deviation (RMSD) of the Cα atoms of HDAC6 in eachsystem and suggested their stability during the entiresimulation period. All the systems were well converged after5 ns simulation, and the average RMSD values for Cα atomswere 1.76, 1.57, 1.27, and 1.40 for the reference, Hit1, Hit2,and Hit3, respectively (Figure 4A, Table 3). Additionally, ourresults also confirmed the stability of each ligand (reference orhit molecules) with HDAC6 during the simulation period(Figure 4B). The RMSD analysis of HDAC6 with eitherreference compound or hit molecules showed that each ligandremained stable in the active site of HDAC6 during the entiresimulation period (Figure 4B, Table 3).Collectively, these results suggested that each ligand was

tightly bound in the active site of HDAC6, maintained theirstability, and had no abnormal behavior during the entiresimulation process. Furthermore, the potential energy analysisof each system suggested that the entire energy of thecorresponding system remained stable during the 30 nssimulation period (Figure 5). Because RMSD and potentialenergy calculation reflected the stability of the overall system,the trajectory of the last 5 ns was taken to elaborate thebinding mode of hit molecules with HDAC6. Upon structuralalignment, it was observed that hit molecules as well as the

Figure 3. Pharmacophore-based virtual screening. A drug-likedatabase was designed from Zinc Natural Product Database. Theselected pharmacophore was employed to screen the drug-likedatabase.

Figure 4. RMSD analysis of the reference and hit compounds. (A) RMSD of the Cα atoms of HDAC6 in all the systems revealed their stability. (B)RMSD of the protein−ligand complex in all the systems portrayed their stability throughout the simulation, and no abnormal behavior wasobserved. Red, green, cyan, and golden colors represent the reference, Hit1, Hit2, and Hit3, respectively.

ACS Chemical Neuroscience Research Article

DOI: 10.1021/acschemneuro.8b00405ACS Chem. Neurosci. XXXX, XXX, XXX−XXX

D

reference compound occupied the same binding pocket andoriented in similar binding pose (Figure S1). The catalytic coreof HDAC6 is composed of highly charged residues includingHis610, His611, and Zn2+, and therefore, it imparts a positivecharge to the local environment of catalytic site. Therefore, thecorrespondent inhibitor should bear negative charge on theirhot spots to make stable interactions with the catalytic core ofHDAC6. To this end, our results showed that polarinteractions of inhibitors are very important to inhibitHDAC6. The reference compound (TSA) established polarinteractions with His610 and Asp742. Our data demonstratedthat the hydroxyl group of the reference compound formed ahydrogen bond with nitrogen at Nδ position of His610 (Figure6A). The same pattern of interaction of TSA in the active siteof HDAC6 has already been reported.49 Other hydrogenbonds were also observed between the amino group of thereference compound and carbonyl oxygen of Asp742 (Figure6A). Our results also revealed a metal coordination bondbetween the positively charged Zn2+ and carbonyl oxygen nearthe amide group of the reference compound (Table 4).Additionally, the methyl groups of the reference compoundshowed hydrophobic interactions with His611, His651,Phe680, and Leu749 (Table 4). Moreover, the referencecompound formed several van der Waals interactions with thecatalytic site residues of HDAC6 (Table 4). Our resultsshowed that the carbonyl oxygen of acidic group of Hit1(Glycodeoxycholic acid) formed hydrogen bond with the sidechain of His610 (Figure 6B). The same oxygen atom wasfound to be in electrostatic interaction with the positivelycharged Zn2+ (Table 4). Our findings demonstrated thatglycodeoxycholic acid also established hydrogen bonds withAsn654, Pro681, and Tyr782 (Figure 6B, Table 4). Our resultsare in parallel findings with Hai and Christianson49 and

Goracci et al.,68 where they showed that Tyr782 of HDAC6formed hydrogen bonds with tested inhibitors. BecausePhe680 is a conserved residue in HDAC family enzymes, itsinteraction with inhibitor is considered to be very important.49

Our findings observed that the steroid nucleus of glycodeox-ycholic acid formed hydrophobic interactions with Cys618,Phe680, and Pro681 (Table 4). The same pattern ofhydrophobic interactions of other inhibitors of HDAC6 hasalready been demonstrated.46,49,69 Glycodeoxycholic acid alsoestablished several van der Waals interactions with the activesite residues of HDAC6 (Table 4).Our study showed that the hydroxyl group of Hit2 formed

hydrogen bonds with the Nδ of His611 and side chain oxygenof Asp649 (Figure 6C). Other hydrogen bonds were observed

Table 3. Databases IDs, Docking, and MD Simulation Analysis of the Reference and Final Hits; Zinc and PubChem DatabaseIDs, Docking Scores, RMSD, and Potential Energy Analyses of the Reference and Final Hits

RMSD analysis (Å)

ligand ZNPD IDa PubChem IDa ChemPLPb ASPc Cα atoms HDAC6-ligand potential energy (kJ/mol)

reference 444732 87.34 30.76 1.76 1.72 −4.32 × 105

Hit1 ZINC 08837267 3035026 94.48 31.49 1.57 2.02 −4.33 × 105

Hit2 ZINC 30881350 38028580 106.34 33.00 1.27 1.77 −4.34 × 105

Hit3 ZINC 04063286 16399643 108.58 30.14 1.40 1.91 −4.33 × 105

aID: Zinc Natural Product Database and PubChem Identity Numbers. bChemPLP: Piecewise Linear Potential (GOLD scoring function). cASP:Astex Statistical Potential (GOLD rescoring function).

Figure 5. The potential energy plot of protein−ligand complex. Thepotential energy of each system was calculated during the entiresimulation period. All the systems showed stable potential energy andno abnormal behavior were observed. Red, green, cyan, and goldencolors show the reference, Hit1, Hit2, and Hit3, respectively.

Figure 6. Molecular interactions analysis. (A) The referencecompounds formed hydrogen bonds with His610 and Asp742. (B)Hit1 established hydrogen bond interactions with His610, Asn654,Pro681, and Tyr782. (C) Hit2 formed hydrogen bond interactionswith His611, Gly619, Asp649, and Gly780. (D) Hit3 showedhydrogen bond interactions with His611, His651, and Pro681. Red,green, cyan, and golden colors represent the reference, Hit1, Hit2, andHit3, respectively. The proteins residues are depicted as think lightblack sticks, while the ligands are represented as thick stick modelswith polar hydrogen only. Hydrogen bonds are represented as greendashed lines. Zn2+ are shown as black dots.

ACS Chemical Neuroscience Research Article

DOI: 10.1021/acschemneuro.8b00405ACS Chem. Neurosci. XXXX, XXX, XXX−XXX

E

with Gly619 and Gly780 (Figure 6C). Our results demon-strated that Hit2 established hydrophobic interactions withPro501, His651, Phe680, Leu749, and Tyr782 (Table 4).Similarly, the catalytic site residues of HDAC6 also establishedseveral van der Waals interactions with Hit2 as shown in Table4. Collectively, these results argued that Hit2 efficiently inhibitsHDAC6. Our results found that the carbonyl oxygen of aceticacid moiety of Hit3 formed hydrogen bond interaction withhydrogen at position Nε of His611 of HDAC6 (Figure 6D).Our findings are in strong agreement that His611 of HDAC6should be protonated at position Nε.70 Furthermore, Hit3 alsoformed hydrogen bonds with His651 and Pro681 (Figure 6D).Our analysis observed hydrophobic interactions between theHit3 and catalytic site residues of HDAC6 (Table 4).Moreover, Hit3 showed several van der Waals interactionswith the catalytic site residues of HDAC6 as depicted in Table4. Our results demonstrated that the carbonyl oxygen of aceticacid moiety of Hit3 also formed a coordination bond with Zn2+

(Table 4). The 2D structures of the three hits are given inSupplementary Figure S2.To understand the consistency of polar interactions of final

hits, the number of intermolecular hydrogen bonds betweenthe HDAC6 and each hit molecule as well as the referencecompound was monitored (Figure 7). Our results demon-strated that the number of hydrogen bonds remained slightlyhigher for all the hit molecules than the reference compound(Figure 7). These observations of polar interactions are inagreement with previous reports498, 698. Collectively, ouranalyses suggested that all the hits established strong polarinteractions as well as nonpolar interactions with importantresidues in the catalytic site of HDAC6. The interactionpattern of hit molecules strongly suggested their inhibitoryeffects of HDAC6. Therefore, we believe that these novelinhibitors could potentially cross blood-brain barrier (BBB ≤2) and could inhibit HDAC6 in the brain. Inhibition ofHDAC6 will deplete the deacetylation of tau at KXGS motifs,and hence preserve the acetylation of tau, which in turnprevents the phosphorylation of tau at serine residue of KXGSmotifs. Moreover, HDAC6 inhibition will preserve acetylatedHSP90, and hence suppresses the phosphorylation of tau atKXGS motifs. Eventually, accelerating the acetylation of tau

protein at KXGS motifs and suppressing the deacetylation ofHSP90 by inhibiting HDAC6 lead to prevent tau-proteinmislocalization and neuritic bead formation.

Comparative Analysis of Hits and the SelectiveInhibitors of HDAC6. Since, few selective inhibitors ofHDAC6 exist; therefore, they were docked with HDAC6, andsubsequently compared with newly identified hits in terms ofdocking score and molecular interactions. Our dockingsimulation suggested that Nexturastat A,45 Tubastatin A,46

ACY-738,71 and ACY-77571 had docking score of 89.2, 89.4,75.7, and 77.3 with HDAC6, respectively. The comparativelow docking score of selective inhibitors such as ACY-738 andACY-775 might be contributed to their narrow cap regions ascompared to the other selective inhibitors of HDAC6. Inparallel, these inhibitors also showed intact H-bonds with theactive site residues of HDAC6 (Figure S3). Our resultsobserved that Nexturastat A formed H-bonds with His610 andGly619 (Figure S3A). The docking results observed severalhydrophobic and van der Waals interactions betweenNexturastat A and HDAC6. Moreover, docking of TubastatinA established H-bonds with His610, Gly619, and Tyr782(Figure S3B). The hydrophobic and van der Waals interactionsof Tubastatin A and HDAC6 are depicted in Supplementary

Table 4. Molecular Interactions of HDAC6 and Ligandsa

H-bond analysis

comp. IDaminoacid

aminoacidatom

ligandatom

length(Å)

Zn2+-boundligand atom hydrophobic interactions van der Waals interactions

reference His610 Nδ1 H44 2.5 O2 His611, His651, Phe680,Leu749

Pro608, Cys618, Gly619, Phe620, Cys621, Asp649, Gly740,Asp742, Glu779, Gly780Asp742 Oδ1 H1 2.3

Asp742 Oδ1 H23 2.5Hit1 His610 Hε2 O11 2.3 O11 Cys618, Phe680, Pro681 Asp567, Ser568, Gly619, Phe620, Asp649, His651,

Gly653, Met682, Gly683, Asp742Asn654 Oδ1 H20 2.8Pro681 O H66 2.5Tyr782 OH H49 2.5

Hit2 His611 Nδ1 H33 1.9 O32 Pro501, His651, Phe680,Leu749, Tyr782

Glu502, Asp567, Ser568, Pro608, Gly619, Phe620,Cys621, Pro681, Asp742, Gly780, Gly781Gly619 HA2 O28 2.2

Asp649 Oδ2 H1 2.7Gly780 HA1 O33 2.6

Hit3 His611 Hε2 O26 1.9 O26 Pro681, His651 Asp567, Ser568, His610, Cys618, Gly619, Phe620,Asp649, Phe680, Met682, Asp742, Leu749, Tyr782His651 Hδ2 H50 3.0

Pro681 O H58 1.8aAll the hits showed stable hydrogen bond as well as hydrophobic and van der Waals interactions with the catalytic site residues of the HDAC6.

Figure 7. Hydrogen bond analysis the protein−ligand complex. Thenumber of hydrogen bonds for each complex was computedthroughout the simulation period. Red, green, cyan, and goldencolors depict the reference, Hit1, Hit2, and Hit3, respectively.

ACS Chemical Neuroscience Research Article

DOI: 10.1021/acschemneuro.8b00405ACS Chem. Neurosci. XXXX, XXX, XXX−XXX

F

Figure S3B. Our results demonstrated that ACY-738established H-bonds with His610 and Gly619 (Figure S3C).Furthermore, ACY-738 formed several hydrophobic and vander Waals interactions with the catalytic site residues ofHDAC6. Docking of ACY-775 showed H-bond interactionswith His610, Gly619, and Phe680 (Figure S3D). ACY-775could also establish strong hydrophobic and van der Waalsinteractions with the active site residues of HDAC6. Overall,the interaction pattern of the selective inhibitors of HDAC6suggested that their similar binding pattern is subjected tohydroxamate group. Our analysis also observed that thehydroxamate group consistently coordinated Zn2+ (Figure S3).Additionally, our newly identified hits displayed similar patternof interactions but high docking score and different molecularstructures. Therefore, we argue that the newly identified hitsmay have high biochemical affinity toward HDAC6 than thepreviously identified inhibitors of HDAC6.In Silico Assessment of HDAC6 Specificity. In order to

evaluate isoform selectivity, we performed docking simulationto investigate the binding affinity of the newly identifiedinhibitors toward other HDACs. Docking parameters for thenewly identified inhibitors with off-targets remained the sameas for HDAC6, but the target proteins were different includingHDAC2, HDAC4, and HDAC8 (Figure S4). Herein, weargued that higher docking score as well as compatible bindingpattern mimic strong binding affinity, and vice versa.Therefore, the comparative docking of the newly identifiedinhibitors with off-targets may reflect their propensity towardthe off-targets. Our docking results suggested that all the newlyidentified inhibitors had lowest docking score toward thetested off-targets (HDAC2, HDAC4, and HDAC8) ascompared to HDAC6 (Table S2). Moreover, our resultssuggested that the binding pattern of the newly identified hitswith off-targets could not follow their inhibitory mode (FigureS4). Therefore, we are proclaiming that the newly identifiedinhibitors might be HDAC6 selective and will have highaffinity toward HDAC6 over other HDACs.

■ MATERIAL AND METHODSStructure-Based Pharmacophore Generation. To identify the

ligand binding features of HDAC6, the structure of human HDAC6 incomplex with trichostatin A (PDB ID: 5EDU) was taken.49 Thebinding pattern of trichostatin A with catalytic site residues ofHDAC6 (His610, His611, His651, Gly679, Phe680, Tyr782, andZn2+) were critically analyzed. Residues in close vicinity of trichostatinA (10 Å) were selected to identify complementary pharmacophoricfeatures by employing Interaction Generation protocol implanted inDS. Thereafter, pharmacophore models were generated usingReceptor−ligand Pharmacophore Generation module of DS.Pharmacophore Validation. The potentiality of pharmacophore

to identify HDAC6 inhibitors was evaluated by Guner−Henryapproach (Decoy test method).50 A test set was designed bycollecting HDAC6 inhibitors whose experimental activities (IC50values) were measured by the same biological assays. The test setwas composed of active and inactive molecules of HDAC6. Theselected pharmacophore was employed as a 3D query to retrieve thebest fitted molecules from the test set. Screening of the test set wasperformed by Ligand Pharmacopore Mapping protocol implanted inDS. Furthermore, pharmacophore-retrieved compounds were identi-fied and several parameters include Guner−Henry (GH) score,enrichment factor (EF), percent ratio of actives (%A), percent yield ofactives (%Y), false negative, and false positive were calculated:

lmoonoo

Ä

Ç

ÅÅÅÅÅÅÅÅÅ

É

Ö

ÑÑÑÑÑÑÑÑÑ|}oo~oo

H A HH A

H HD A

GH(3 )

41

( )a t

t

t a=[ + ]

× −−− (1)

i

k

jjjjjjjjj

y

{

zzzzzzzzzEF

HHAD

a

t=(2)

where GH: Guner−Henry score; EF: enrichment factor score; Ha:number of actives in hit list, Ht: total number of hits retrieved bypharmacophore; D: number of compounds in database; A: number ofactives in database.

Generation of Drug-Like Database and Virtual Screening.In the current study, Zinc Natural Product Database was taken(http://pkuxxj.pku.edu.cn/UNPD/). Drug-like compounds have veryrestricted chemical properties to avoid toxicity. Therefore, Lipinski’srule of f ive (ROF) and ADMET assessment tests were applied, and adrug-like database was designed. ROF evaluates the physiochemicalproperties of the drug-like molecule(s).51 ADMET assessment testevaluates the pharmacokinetic properties such as absorption,distribution, metabolism, excretion, and toxicity of the drugmolecules. Lipinski’s rule of f ive (ROF) and ADMET assessment testwere carried out in DS.

The validated pharmacophore was employed as a 3D query toscreen the drug-like database. The Ligand Pharmacophore Mappingprotocol in DS was employed to retrieve the drug-like molecules.During database screening, the search option was set to Best/Flexiblewhile the Maximum Omitted Features option was set to “0” to restrictthe mapping of compounds onto all the features of thepharmacophore.

Molecular Docking Simulation. Genetic Optimization of LigandDocking (GOLD v5.2.2) package was employed that uses geneticalgorithm to dock small molecules into the active site of protein.52

GOLD software allows full flexibility of ligands and partial flexibility ofprotein; hence, it gives more reliable calculations in computationalbiology. The structure of human HDAC6 in complex with trichostatinA (TSA) was taken from Protein Data Bank.49 The structure ofHDAC6 was prepared by removing water, salts, and other unwantedmolecules. The in-built module of GOLD software (add hydrogen)was used, and hydrogen atoms were added to the structure ofHDAC6. The active site of HDAC6 was defined within a radius of 10Å of the bound inhibitor. The ChemPLP (Piecewise Linear Potential)score and ASP (Astex Statistical Potential) score were used as thedefault scoring and rescoring functions, respectively.53 The ChemPLPis an empirical fitness function optimized for pose prediction and isthe default scoring function in GOLD software. It is used to modelthe steric complementarity between protein and ligand, distance- andangle-dependent hydrogen and metal bonding terms as well as theheavy atoms clash- and torsional potential. The ASP score measuresthe atom−atom potential and has comparable accuracy to Chemscoreand Goldscore fitness functions.53 During docking, the GA (geneticalgorithm) run was set to 100 to produce hundred poses for eachdrug-like molecule. The bound ligand (trichostatin A) was employedas a reference compound throughout the analyses. The drug-likemolecules with highest ChemPLP and ASP scores, highly stableconformers (clustering analysis), and hydrogen bond interactions withcatalytic residues of HDAC6 were retrieved.

Molecular Dynamics Simulation. Molecular dynamics (MD)simulation was performed to affirm pose orientation, stability, andbinding pattern of selected drug-like molecules in the active site ofHDAC6 under physiological conditions. For each drug-like molecule,independent simulation system was prepared in Groningen Machine forChemical Simulation (GROMACS v5.0.7) package.54 The topologyfiles of HDAC6 and ligands were created by CHARMm36 all atomsforce field55 and SwissParam,56 respectively. All the simulations wereperformed in the octahedral box, and each system was solvated withthe TIP3P water model. Periodic boundary conditions were applied inall directions to mimic the infinite system. Sodium ions (Na+) wereadded as counterions to neutralize the negative charge of simulationsystem. Each system was subjected to an initial energy minimizationphase, where the maximum force was set to 10 kJ/mol to avoid stericclash and bad contacts. To avoid configurational changes duringequilibration, all atom position restraints were applied. All the systemswere equilibrated in two steps. First, temperature equilibration was

ACS Chemical Neuroscience Research Article

DOI: 10.1021/acschemneuro.8b00405ACS Chem. Neurosci. XXXX, XXX, XXX−XXX

G

carried out under an NVT ensemble (constant number of particles,volume, and temperature) for 100 ps at 300 K while using V-rescalethermostat.57 Second, NPT ensemble was equilibrated at a constantnumber of particles, pressure, and temperature. The pressure of eachsystem was maintained at 1.0 bar by Parrinello−Rahman barostat.58

In each system, the backbone atoms of proteins were restrained, whilethe solvent molecules and counterions were unrestrained to movefreely during the equilibration phases. Each equilibrated system wassubjected to an unrestrained molecular dynamics simulation for 30 ns.Particle Mesh Ewald (PME) approach was employed to calculate thelong-range electrostatic interactions using a 10 Å cutoff distance.During simulation, bond lengths were constrained using LINCSalgorithm that allowed a 2 fs time step in all simulations.59 During datacollection, V-rescale thermostat and Parrinello−Rahman barostatwere used to maintain the temperature and pressure at 300 K and 1.0bar, respectively.

■ CONCLUSIONS

Modulation of post-translational modification of proteins hasemerged as a new avenue in chemotherapy of chronic diseases.In this study, virtual screening, molecular docking, andmolecular dynamics simulation were integrated to identifynovel inhibitors of HDAC6. A structure-based pharmacophorewas generated from HDAC6-TSA complex. The pharmaco-phore had five features including two hydrogen bond donors,one hydrogen bond acceptor, and two hydrophobic features,which was validated by the decoy test, and a GH score of 0.80was obtained. The Zinc Natural Product Database was filteredby Lipinski’s rule of f ive (ROF) and ADMET Descriptors, and adrug-like database was designed. The validated pharmacophorewas employed to screen the drug-like database, and a total of841 candidate molecules were retrieved as candidate inhibitorsof HDAC6. Furthermore, the candidate inhibitors weresubjected to molecular docking to identify the true positiveinhibitors of HDAC6. Finally, molecular dynamics simulationwas employed to understand the stability, binding mode, andmolecular interactions of the novel hit molecules of HDAC6.These novel hit molecules showed strong polar interactionswith the catalytic residues of HDAC6 including His610,His611, Asp649, His651, and Tyr782. Hit molecules alsoestablished several hydrophobic and van der Waals interactionswith the catalytic residues of HDAC6. Therefore, werecommend these hit molecules as the novel inhibitors ofHDAC6, which will preserve the acetylation of tau and depletethe deacetylation of HSP90 to prevent tau protein mislocaliza-tion and neuritic bead formation in tau-pathogenesis.

■ ASSOCIATED CONTENT

*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acschemneur-o.8b00405.

Superimposition of final hits in the active site ofHDAC6, 2D structures of the final hits, the 2Dinteraction pattern of the selective inhibitors ofHDAC6, the 2D interaction pattern of the novel hitswith off-targets (HDAC2, HDAC4, and HDAC8),docking results of the top-ranked molecules withHDAC6, the docking score assessment of the novelhits with HDAC6 and off-targets (PDF)

■ AUTHOR INFORMATIONCorresponding Author*E-mail: kwlee@gnu.ac.kr. Tel: +82-55-772-1360. Fax: +82-55-772-1359.ORCIDKeun Woo Lee: 0000-0002-0627-1800Author ContributionsA.Z., C.P., and K.W.L designed the project; A.Z. and S.R.,performed the experiments; A.Z., M.S., and K.W.L analyzedthe results; and A.Z., S.R., M.S., G.H.L., and K.W.L. preparedthe manuscript. All authors reviewed the manuscript.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis research was supported by the Bio & Medical TechnologyDevelopment Program of the National Research Foundation(NRF) & funded by the Korean government (MSIT) (No.NRF-2018M3A9A7057263).

■ ABBREVIATIONS AND ANALOGIES(HDACs)Histone deacetylases; (HDAC6)Histone deacetylase6; (TSA)Trichostatin A; (HBA)Hydrogen bond acceptor;(HBD)Hydrogen bond donor; (HYP)Hydrophobic; (GHscore)Guner−Henry score; (ChemPLP)Chemical piece oflinear potential; (SAHA)Suberoylanilide hydroxamic acid;(FDA)Food and drug administration; (CDI)Catalytic domainI; (CDII)Catalytic domain II; (HSP90)Heat shock protein 90;(DRG)Dorsal root ganglion; (CMT2)Charcot-marie-toothdisease; (3D)3-dimensional; (MD)Molecular dynamics;(RMSD)Root-mean-square deviation; (PDB)Protein databank; (EF)Enrichment factor; (AlogP)Log value of octanol−water partition coefficient; (DS)Discovery studio v4.5; (ROF)Role of five; (GOLD)Genetic optimization for ligand docking;(ASP)Astex statistical potential; (Cα)Carbon alpha of aminoacid residue; (GA)Genetic algorithm; (GROMACS)Gronin-gen machine for chemical simulation; (CHARMm)Chemistryat Harvard macromolecular mechanics; (TIP3P)Transferableintermolecular potential with 3 points; (NVT)Number ofparticles, volume, and temperature; (NPT)Number of particle,pressure, and temperature; (PME)Particle mesh ewald;(LINCS)Linear constraint solver for molecular simulations;(fs)Femtoseconds.

■ REFERENCES(1) Grozinger, C. M., Hassig, C. A., and Schreiber, S. L. (1999)Three proteins define a class of human histone deacetylases related toyeast Hda1p. Proc. Natl. Acad. Sci. U. S. A. 96, 4868−4873.(2) Kouzarides, T. (1999) Histone acetylases and deacetylases in cellproliferation. Curr. Opin. Genet. Dev. 9 (1), 40−48.(3) Simoes-Pires, C., Zwick, V., Nurisso, A., Schenker, E., Carrupt, P.− A., and Cuendet, M. (2013) HDAC6 as a target for neuro-degenerative diseases: what makes it different from the other HDACs?Mol. Neurodegener. 8 (7), 7.(4) Gregoretti, I. V., Lee, Y. M., and Goodson, H. V. (2004)Molecular evolution of the histone deacetylase family: functionalimplications of phylogenetic analysis. J. Mol. Biol. 338, 17−31.(5) Tang, G., Wong, J. C., Zhang, W., Wang, Z., Zhang, N., Peng, Z.,Zhang, Z., Rong, Y., Li, A., Zhang, M., Yu, L., Feng, T., Zhang, X., Wu,X., Wu, J. Z., and Chen, L. (2014) Identification of a novelaminotetralin class of HDAC6 and HDAC8 selective inhibitors. J.Med. Chem. 57, 8026−8034.

ACS Chemical Neuroscience Research Article

DOI: 10.1021/acschemneuro.8b00405ACS Chem. Neurosci. XXXX, XXX, XXX−XXX

H

(6) Raedler, L. A. (2016) Farydak (Panobinostat): First HDACinhibitor approved for patients with relapsed multiple myeloma. Am.Health Drug Benefits 9, 84−87.(7) Sawas, A., Radeski, D., and O′Connor, O. A. (2015) Belinostatin patients with refractory or relapsed peripheral T-cell lymphoma: aperspective review. Ther. Adv. Hematol. 6 (4), 202−208.(8) Weïwer, M., Lewis, M. C., Wagner, F. F., and Holson, E. B.(2013) Therapeutic potential of isoform selective HDAC inhibitorsfor the treatment of schizophrenia. Future Med. Chem. 5 (13), 1491−1508.(9) McKinsey, T. A. (2012) Therapeutic potential for HDACinhibitors in the heart. Annu. Rev. Pharmacol. Toxicol. 52, 303−319.(10) Bruserud, O., Stapnes, C., Ersvaer, E., Gjertsen, B. T., andRyningen, A. (2007) Histone deacetylase inhibitors in cancertreatment: a review of the clinical toxicity and the modulation ofgene expression in cancer cell. Curr. Pharm. Biotechnol. 8 (6), 388−400.(11) Marsoni, S., Damia, G., and Camboni, G. (2008) A work inprogress: the clinical development of histone deacetylase inhibitors.Epigenetics 3 (3), 164−171.(12) Spiegel, S., Milstien, S., and Grant, S. (2012) Endogenousmodulators and pharmacological inhibitors of histone deacetylases incancer therapy. Oncogene 31, 537−551.(13) Hubbert, C., Guardiola, A., Shao, R., Kawaguchi, Y., Ito, A.,Nixon, A., Yoshida, M., Wang, X. F., and Yao, T. P. (2002) HDAC6 isa microtubule-associated deacetylase. Nature 417 (6887), 455−458.(14) Matsuyama, A., Shimazu, T., Sumida, Y., Saito, A., Yoshimatsu,Y., Seigneurin-Berny, D., Osada, H., Komatsu, Y., Nishino, N.,Khochbin, S., Horinouchi, S., and Yoshida, M. (2002) In vivodestabilization of dynamic microtubules by HDAC6-mediateddeacetylation. EMBO J. 21, 6820−6831.(15) Zhang, Y., Gilquin, B., Khochbin, S., and Matthias, P. (2006)Two catalytic domains are required for protein deacetylation. J. Biol.Chem. 281, 2401−2404.(16) Bertos, N. R., Gilquin, B., Chan, G. K., Yen, T. J., Khochbin, S.,and Yang, X. J. (2004) Role of the tetradecapeptide repeat domain ofhuman histone deacetylase 6 in cytoplasmic retention. J. Biol. Chem.279, 48246−48254.(17) Haggarty, S. J., Koeller, K. M., Wong, J. C., Grozinger, C. M.,and Schreiber, S. L. (2003) Domain-selective small-molecularinhibitor of histone deacetylase 6 (HDAC6)-mediated tubulindeacetylation. Proc. Natl. Acad. Sci. U. S. A. 100 (8), 4389−4394.(18) Zou, H., Wu, Y., Navre, M., and Sang, B. C. (2006)Characterization of the two catalytic domains in histone deacetylase6. Biochem. Biophys. Res. Commun. 341, 45−50.(19) Schreiber, S. L., and Bernstein, B. E. (2002) Signaling networkmodel of chromatin. Cell 111, 771−778.(20) Kovacs, J. J., Murphy, P. J., Gaillard, S., Zhao, X., Wu, J. T.,Nicchitta, C. V., Yoshida, M., Toft, D. O., Pratt, W. B., and Yao, T. P.(2005) HDAC6 regulates Hsp90 acetylation and chaperone-depend-ent activation of glucocorticoid receptor. Mol. Cell 18 (5), 601−607.(21) Min, S. W., Cho, S. H., Zhou, Y., Schroeder, S., Haroutunian,V., Seeley, W. W., Huang, E. J., Shen, Y., Masliah, E., Mukherjee, C.,Meyers, D., Cole, P. A., Ott, M., and Gan, L. (2010) Acetylation ofTau inhibits its degradation and contributes to tauopathy. Neuron 67,953−966.(22) Boyault, C., Sadoul, K., Pabion, M., and Khochbin, S. (2007)HDAC6, at the crossroads between cytoskeleton and cell signaling byacetylation and ubiquitination. Oncogene 26 (37), 5468−5476.(23) Lee, Y. S., Lim, K. H., Guo, X., Kawaguchi, Y., Gao, Y.,Barrientos, T., Ordentlich, P., Wang, X. F., Counter, C. M., and Yao,T. P. (2008) The cytoplasmic deacetylase HDAC6 is required forefficient oncogenic tumorigenesis. Cancer Res. 68 (18), 7561−7569.(24) Kawada, J., Zou, P., Mazitschek, R., Bradner, J. E., and Cohen, J.I. (2009) Tubacin kills Epstein-Barr virus (EBV)-Burkitt lympho-blastoid cells by inducing apoptosis. J. Biol. Chem. 284 (25), 17102−17109.(25) Rivieccio, M. A., Brochier, C., Willis, D. E., Walker, B. A.,D’Annibale, M. A., McLaughlin, K., Siddiq, A., Kozikowski, A. P.,

Jaffrey, S. R., Twiss, J. L., Ratan, R. R., and Langley, B. (2009)HDAC6 is a target for protection and regeneration following injury inthe nervous system. Proc. Natl. Acad. Sci. U. S. A. 106 (46), 19599−19604.(26) Bali, P., Pranpat, M., Bradner, J., Balasis, M., Fiskus, W., Guo,F., Rocha, K., Kumaraswamy, S., Boyapalle, S., Atadja, P., Seto, E., andBhalla, K. (2005) Inhibition of histone deacetylase 6 acetylates anddisrupts the chaperone function of heat shock protein 90: a novelbasis of antileukemia activity of histone deacetylase inhibitors. J. Biol.Chem. 280 (29), 26729−26734.(27) Zhang, X., Yuan, Z., Zhang, Y., Yong, S., Salas-Burgos, A.,Koomen, J., Olashaw, N., Parsons, J. T., Yang, X. J., Dent, S. R., Yao,T. P., Lane, W. S., and Seto, E. (2007) HDAC6 modulates cellmotility by altering the acetylation level of cortactin. Mol. Cell 27 (2),197−213.(28) Parmigiani, R. B., Xu, W. S., Venta-Perez, G., Erdjument-Bromage, H., Yaneva, M., Tempst, P., and Marks, P. A. (2008)HDAC6 is a specific deacetylase of peroxiredoxins and is involved inredox regulation. Proc. Natl. Acad. Sci. U. S. A. 105 (28), 9633−9638.(29) Witt, O., Deubzer, H. E., Milde, T., and Oehme, I. (2009)HDAC family: What are the cancer relevant targets? Cancer Lett. 277(1), 8−21.(30) Kaluza, D., Kroll, J., Gesierich, S., Yao, T. P., Boon, R. A.,Hergenreider, E., Tjwa, M., Rossig, L., Seto, E., Augustin, H. G.,Zeiher, A. M., Dimmeler, S., and Urbich, C. (2011) Class IIb HDAC6regulates endothelial cell migration and angiogenesis by deacetyaltionof cortactin. EMBO J. 30 (20), 4142−4156.(31) Cohen, T. J., Guo, J. L., Hurtado, D. E., Kwong, L. K., Mills, I.P., Trojanowski, J. Q., and Lee, V. M.Y. (2011) The acetylation of tauinhibits its function and promotes pathological tau aggregation. Nat.Commun. 2, 252.(32) Szyk, A., Deaconescu, M. A., Spector, J., Goodman, B.,Valenstein, L. M., Ziolkowska, E. N., Kormendi, V., Grigorieff, N., andRoll-Mecak, A. (2014) Molecular basis for age-dependent micro-tubule acetylation by tubulin acetyltransferase. Cell 157 (6), 1405−1415.(33) Guo, W., Naujock, M., Fumagalli, L., Vandoorne, T., Baatsen,P., Boon, R., Ordovas, L., Patel, A., Welters, M., Vanwelden, T.,Geens, N., Tricot, T., Benoy, V., Steyaert, J., Lefebvre-Omar, C.,Boesmans, W., Jarpe, M., Sterneckert, J., Wegner, F., Petri, S., Bohl,D., Vanden Berghe, P., Robberecht, W., Van Damme, P., Verfaillie, C.,and Van Den Bosch, L. (2017) HDAC6 inhibition reverses axonaltransport defects in motor neurons derived from FUS-ALS patients.Nat. Commun. 8 (1), 1−15.(34) d’Ydewalle, C., Krishnan, J., Chiheb, D. M., Van Damme, P.,Irobi, J., Kozikowski, A. P., Berghe, P. V., Timmerman, V.,Robberecht, W., and Van Den Bosch, L. (2011) HDAC6 inhibitorsreverse axonal loss in a mouse model of mutant HSPB1-inducedCharcot-Marie-Tooth disease. Nat. Med. 17, 968−974.(35) Min, S. W., Chen, X., Tracy, T. E., Li, Y., Zhou, Y., Wang, C.,Shirakawa, K., Minami, S. S., Defensor, E., Mok, S. A., Sohn, P. D.,Schilling, B., Cong, X., Ellerby, L., Gibson, B. W., Johnson, J., Krogan,V., Shamloo, M., Gestwicki, J., Masliah, E., Verdin, E., and Gan, L.(2015) Critical role of acetylation in tau-mediated neurodegenerationand cognitive deficits. Nat. Med. 21, 1154−1162.(36) Tracy, T. E., Sohn, P. D., Minami, S. S., Wang, C., Min, S. W.,Li, Y., Zhou, Y., Le, D., Lo, I., Ponnusamy, R., Cong, X., Schilling, B.,Ellerby, L. M., Huganir, R. L., and Gan, L. (2016) Acetylated Tauobstructs KIBRA-mediated signaling in synaptic plasticity andpromotes tauopathy-related memory loss. Neuron 90 (2), 245−260.(37) Ding, H., Dolan, P. J., and Johnson, G. V. (2008) Histonedeacetylase 6 interacts with the microtubule-associated protein tau. J.Neurochem. 106 (5), 2119−21130.(38) Cook, C., Gendron, T. F., Scheffel, K., Carlomagno, Y.,Dunmore, J., De Ture, M., and Petrucelli, L. (2012) Loss of HDAC6,a novel CHIP substrate, alleviates abnormal tau accumulation. Hum.Mol. Genet. 21 (13), 2936−2945.(39) Cook, C., Carlomagno, Y., Gendron, T. F., Dunmore, J.,Scheffel, K., Stetler, C., Davis, M., Dickson, D., Jarpe, M., De Ture,

ACS Chemical Neuroscience Research Article

DOI: 10.1021/acschemneuro.8b00405ACS Chem. Neurosci. XXXX, XXX, XXX−XXX

I

M., and Petrucelli, L. (2014) Acetylation of the KXGS motifs in tau iscritical determinant in modulation of tau aggregation and clearance.Hum. Mol. Genet. 23 (1), 104−116.(40) Carlomagno, Y., Chung, D. C., Yue, M., Castanedes-Casey, M.,Madden, B. J., Dunmore, J., Tong, J., DeTure, M., Dickson, D. W.,Petrucelli, L., and Cook, C. (2017) An acetylation-phosphorylationswitch that regulates tau aggregation propensity and function. J. Biol.Chem. 292 (37), 15277−15286.(41) Tseng, J. H., Xie, L., Song, S., Xie, Y., Allen, L., Ajit, D., Hong, J.S., Chen, X., Meeker, R. B., and Cohen, T. J. (2017) The deacetylaseHDAC6 mediates endogenous neuritic Tau pathology. Cell Rep. 20(9), 2169−2183.(42) Fan, S.-J., Huang, F.-I., Liou, J.-P., and Yang, C.-R. (2018) Thenovel histone deacetylase 6 inhibitor, MPT0G211, ameliorates tauphosphorylation and cognitive deficits in an Alzheimer’s diseasemodel. Cell Death Dis. 9 (655), 1−14.(43) Dallavalle, S., Pisano, C., and Zunino, F. (2012) Developmentand therapeutic impact of HDAC6-selective inhibitors. Biochem.Pharmacol. 84, 756−765.(44) Santo, L., Hideshima, T., Kung, A. L., Tseng, J. C., Tamang, D.,Yang, M., Jarpe, M., van Duzer, J. H., Mazitschek, R., Ogier, W. C.,Cirstea, D., Rodig, S., Eda, H., Scullen, T., Canavese, M., Bradner, J.,Anderson, K. C., Jones, S. S., and Raje, N. (2012) Preclinical activity,pharmacodynamics, and pharmacokinetics properties of a selectiveHDAC6 inhibitor, ACY-1215, in combination with bortezomib inmultiple myeloma. Blood 119 (11), 2579−2589.(45) Bergman, J. A., Woan, K., Perez-Villarroel, P., Villagra, A.,Sotomayor, E. M., and Kozikowski, A. P. (2012) Selective histonedeacetylase 6 inhibitors bearing substituted urea linkers inhibitmelanoma cell growth. J. Med. Chem. 55 (22), 9891−9899.(46) Butler, K. V., Kalin, J., Brochier, C., Vistoli, G., Langley, B., andKozikowski, A. P. (2010) Rational design and simple chemistry yield asuperior, neuroprotective HDAC6 inhibitor, tubastatin A. J. Am.Chem. Soc. 132 (31), 10842−10846.(47) Zhang, L., Sheng, S., and Qin, C. (2013) The role of HDAC6in Alzheimer’s disease. J. Alzheimer's Dis. 33, 283−295.(48) Valenzuela-Fernandez, A., Cabrero, J. R., Serrador, J. M., andSanchez-Madrid, F. (2008) HDAC6: A key regulator of cytoskeleton,cell migration and cell-cellinteractions. Trends Cell Biol. 18 (8), 291−297.(49) Hai, Y., and Christianson, D. (2016) Histone deacetylase 6structure and molecular basis of catalysis and inhibition. Nat. Chem.Biol. 12 (9), 741−747.(50) Guner, O. F. (2002) History and evolution of thepharmacophore concept in computer-aided drug design. Curr. Top.Med. Chem. 2, 1321−1332.(51) Lipinski, C. A., Lombardo, F., Dominy, B. W., and Feeney, P. J.(2001) Experimental and computational approaches to estimatepermeability in drug discovery and development settings. Adv. DrugDelivery Rev. 46, 3−26.(52) Verdonk, M. L., Cole, J. C., Hartshorn, M. J., Murray, C. W.,and Taylor, R. D. (2003) Improved protein-ligand docking usingGOLD. Proteins: Struct., Funct., Genet. 52, 609−623.(53) Li, Y., Han, L., Liu, Z., and Wang, R. (2014) Comparativeassessment of scorning functions on an updated benchmark: 2.Evaluation methods and general results. J. Chem. Inf. Model. 54,1717−1736.(54) Pronk, S., Pall, S., Schulz, R., Larsson, P., Bjelkmar, P.,Apostolov, R., Shirts, M. R., Smith, J. C., Kasson, P. M., van der Spoel,D., Hess, B., and Lindahl, E. (2013) GROMACS 4.5: a high-throughput and highly parallel open source molecular simulationtoolkit. Bioinformatics 29, 845−854.(55) Huang, J., Rauscher, S., Nawrocki, G., Ran, T., Feig, M., deGroot, B. L., Grubmuller, H., and MacKerell, A. D., Jr. (2017)CHARMM36m: an improved force field for folded and intrinsicallydisordered proteins. Nat. Methods 14, 71−73.(56) Zoete, V., Cuendet, M. A., Grosdidier, A., and Michielin, O.(2011) SwissParam, a fast force field generation tool for small organicmolecules. J. Comput. Chem. 32 (11), 2359−2368.

(57) Bussi, G., Donadio, D., and Parrinello, M. (2007) Canonicalsampling through velocity rescaling. J. Chem. Phys. 126, 014101.(58) Parrinello, M., and Rahman, A. (1981) Polymorphic transitionsin single crystals: A new molecular dynamics method. J. Appl. Phys. 52,7182−7190.(59) Hess, B., Bekker, H., Berendsen, H. J. C., and Fraaije, J. G. E.M. (1997) LINCS: A linear constraint solver for molecularsimulations. J. Comput. Chem. 18, 1463−1472.(60) Hideshima, T., Qi, J., Paranal, M. R., Tang, W., Greenberg, E.,West, N., Colling, M. E., Estiu, G., Mazitschek, R., Perry, J. A.,Ohguchi, H., Cottini, F., Mimura, N., Gorgun, G., Tai, Y. − T.,Richardson, P. G., Carrasco, R. D., Wiest, O., Schreiber, S. L.,Anderson, K. C., and Brandner, J. E. (2016) Discovery of selectivesmall-molecule HDAC6 inhibitor for overcoming proteasomeinhibitor resistance in multiple myeloma. Proc. Natl. Acad. Sci. U. S.A. 113 (46), 13162−13167.(61) Senger, J., Melesina, J., Marek, M., Romier, C., Oehme, I., Witt,O., Sippl, W., and Jung, M. (2016) Synthesis and biologicalinvestigation of oxazole hydroxamates as highly selective histonedeacetylase 6 (HDAC6) inhibitors. J. Med. Chem. 59 (4), 1545−1555.(62) Goodsell, D. S., and Olson, A. J. (1990) Automated docking ofsubstrates to proteins by simulated annealing. Proteins: Struct., Funct.,Genet. 8, 195−202.(63) Mobley, D. L., and Dill, K. A. (2009) Binding of small-moleculeligands to proteins: ‘what you see’ is not always ‘what you get’.Structure 17, 489−498.(64) Ferreira, L. G., dos Santos, R. N., Oliva, G., and Andricopulo, A.D. (2015) Molecular docking and structure-based drug designstrategies. Molecules 20, 13384−13421.(65) Dodson, G. G., Lane, D. P., and Verma, C. S. (2008) Moleculardynamics simulations of protein dynamics: new windows onmechanisms in biology. EMBO Rep. 9, 144−150.(66) Durrant, J. D., and McCammon, J. A. (2011) Moleculardynamics simulations and drug discovery. BMC Biol. 9, 71.(67) Liu, X., Shi, D., Zhou, S., Liu, H., Liu, H., and Yao, X. (2018)Molecular dynamics simulations and novel drug discovery. ExpertOpin. Drug Discovery 13, 23−37.(68) Borhani, D. W., and Shaw, D. E. (2012) The future ofmolecular dynamics simulations in drug discovery. J. Comput.-AidedMol. Des. 26, 15−26.(69) Goracci, L., Deschamps, N., Randazzo, G. M., Petit, C., DosSantos Passos, C., Carrupt, P.-A., Simoes-Pires, C., and Nurisso, A.(2016) A rational approach for the identification of non-hydroxamateHDAC6-selective inhibitors. Sci. Rep. 6, 29086.(70) Burli, R. W., Luckhurst, C. A., Aziz, O., Matthews, K. L., Yates,D., Lyons, K. A., Beconi, M., McAllister, G., Breccia, P., Stott, A. J.,et al. (2013) Design, synthesis, and biological evaluation of potentand selective Class IIa histone deacetylase (HDAC) inhibitors as apotential therapy for Huntington’s disease. J. Med. Chem. 56, 9934−9954.(71) Jochems, J., Boulden, J., Lee, B. G., Blendy, J. A., Jarpe, M.,Mazitschek, R., Van Duzer, J. H., Jones, S., and Berton, O. (2014)Antidepressent-like properties of novel HDAC6-selective inhibitorswith improved brain bioavailability. Neuropsychopharmacology 39,389−400.

ACS Chemical Neuroscience Research Article

DOI: 10.1021/acschemneuro.8b00405ACS Chem. Neurosci. XXXX, XXX, XXX−XXX

J