Discovery of Novel Class of Histone Deacetylase Inhibitors as Anticancer Agents

Selective inhibition of histone deacetylases (HDACs) is a paramount strategy in the eld of anticancer lead drug development. However, the lack of selectivity among these isozymes and the adverse side effects that arise from the inhibition of certain HDACs limits their clinical applications. Phenotypic screening of an in-house pilot library of about 70 small molecules against various HDAC isozymes led to the discovery of ve compounds which displayed varying degrees of HDAC isozyme selectivity. The anticancer activities of these molecules were validated using various biological assays including transcriptomic studies. Compounds 15, 14, and 19 possessed selective inhibitory activity against HDAC5, while 28 displayed selective inhibition of HDAC1 and HDAC2. Compound 22 was found to be a selective inhibitor for HDAC3 and HDAC9. In the MCF7 breast cancer cell line the new compounds showed potent anti-proliferative activity, increased acetylation of histones and induced cells apoptosis. The new compounds’ apoptotic effects were validated through the upregulation of caspases3 and 7 and downregulation of C-MYC, BCL2, BCL3 and NFĸB genes. Furthermore, the new compounds arrested cell cycle at different phases, which was conrmed through downregulation of the CDK1, 2, 4, 6, E2F1 and RB1 proteins. Together our ndings provide the foundation for the development of new chemical probes as potential lead drug candidates for the treatment of cancer.


Introduction
Modi cation of chromatin structure through post-translational histone processes plays a key role in carcinogenesis and has been exploited as an anti-cancer drug discovery mechanism for over a decade 1,2 . Methylation and acetylation are among the most important routes of post-translational modi cations of histones that affect cancerous and normal cells epigenetics 3 . The acetylation of histones is controlled by two families of enzymes; histone acetyltransferases (HATs) that enhance acetylation, and histone deacetylases (HDACs) that remove histones acetyl groups 4,5 . Hyper-acetylated histones lead to an open chromatin structure and higher cellular transcriptional activities. Additionally, opened chromatin structure allows the cellular DNA to be more accessible to DNA damaging agents such as anticancer drugs. These ndings have encouraged researchers to design compounds that modulate the degree of histone acetylation by altering the function of HATs and/or HDACs.
Abnormal expression of HDACs is present in many diseases including cancer, diabetes, cystic brosis, cardiac, pulmonary diseases, in ammatory and neurological disorders 6 . Selective HDAC3 inhibitors have been described as promising agents for the treatment of in ammation and Type 2 diabetes 7 .
Furthermore, selective inhibitors of HDAC1 and HDAC2 have been reported as an effective approach in the management of sickle cell disease 8 .
The FDA has approved four histone deacetylase inhibitors (HDACi) for treatment of hematologic and solid malignancies, while many others are in clinical trials, either as a single modality or in combination with other chemotherapeutic drugs [9][10][11][12] . The four FDA-approved drugs include the hydroxamic acid derivatives vorinostat, belinostat and panobinostat and the natural product-derived romidepsin (Fig. 1A).
In addition, the benzamide derivative chidamide is approved for treatment of T-cell lymphomas in China and is under clinical review in several other countries. However, these inhibitors suffer from multiple side effects due to their lack of selectivity for therapeutic HDACs over other HDACs involved in normal physiological processes. To overcome these challenges, the development of selective HDAC inhibitors has become an important goal in drug design [8][9][10] .
In this article we report the discovery of a new class of selective HDAC inhibitors (Supplementary Figures S1-S15). The novel structural scaffolds were discovered through a screening campaign of an in-house library of small molecules in vitro against various HDAC isozymes. This process identi ed ve novel molecules that potently inhibit various HDACs, with HDAC selectivity controlled by the substitution pattern on the core scaffold 13 . This new class of HDAC inhibitors possesses an imidazopyridine moiety, which, to the best of our knowledge, has not been reported before. The inhibition of various HDACs by these compounds as well as their effects on acetylated histones, pro-apoptotic and antiapoptotic markers, cell cycle and anti-proliferative activity in the MCF7 breast cancer cell line were investigated [14][15][16] and are reported below.

Biochemical HDAC assay
The effect of compounds on HDAC1-HDAC9 enzymes was carried out in vitro using an optimized homogenous assay performed in 384-well plate format as previously reported 47 . In brief, reactions were performed in assay buffer (50 mM HEPES, 100 mM KCl, 0.001% (v/v) Tween-20, 0.05% (w/v) bovine serum albumin, 200 μM tris(2-carboxyethyl)phosphine (TCEP), pH 7.4) and followed for uorogenic release of 7-amino-4-methylcoumarin from substrate upon deacetylase and trypsin enzymatic activity. Fluorescence measurements were obtained approximately every 5 min using an Envison multi-label plate reader and plate stacker (Perkin-Elmer, USA). Data were analyzed on a plate-by-plate basis for the linear range of uorescence over time. The rst derivative of data obtained from the plate capture corresponding to the mid-linear range was imported into analytical software (Spot re DecisionSite and GraphPad Prism). Data was analyzed using logistic regression with calculation of IC 50 50 values derived from nonlinear curve t of the dose response data using an outliers exclusion, variable slope model (Prism, Graphpad). Ki, inhibition constant; [S], substrate concentration; Km, Michaelis constant. Bidirectional hierarchical clustering was carried out on biochemical pro ling data (Ki) for HDAC1-HDAC9 by creating a pairwise distance matrix using the unweighted pair group method with arithmetic mean and a Euclidean distance similarity measure (Spot re DecisionSite, USA). Replicate experimental data from incubations with inhibitor were normalized to controls.

Chemo sensitivity assay
The effect of the new compounds on the proliferation of MCF7 and A549 cells was studied using the SulphoRhodamine-B (SRB) assay as previously described 48 . Doxorubicin and SAHA (Sigma, USA) were used as positive controls. Cells were seeded in a 96-well plate at a density of 10 4 cells per well. After overnight incubation, cells were treated with HDAC inhibitors, doxorubicin or vorinostat (SAHA) (0.001 -10 µmol L-1), DMSO was added to the control cells and each treatment was performed in triplicate. After 48 hours, cells were xed and washed several times with water, and then stained with 0.4% SRB for 30 min and washed several times with 1% acetic acid solution. The retained dye was dissolved in 10 mM Tris base (pH 10.5) and color intensity was measured at 564 nm using an ELISA microplate reader (Thermo Scienti c, USA). IC 50 values were calculated using the sigmoidal concentration-response curve-tting model (Graph Pad, Prism software). 17,49,50 Effect of the new compounds and the positive control vorinostat (SAHA) on caspase7 activity was studied in MCF7 cells using the Caspase-glo 3/7 assay kit purchased from Promega (Madison, USA). The assay was carried out according to the manufacturer's procedure. In brief, exponentially growing 50% con uent cells were treated with the new compounds or the positive control vorinostat (1 and 3 μM). After 48 h, cells were collected and cell density was adjusted to 10 6 cells/ml. The cell pellet was re-suspended in cell lysis buffer and the supernatant was collected by centrifugation at 4 o C. The cell extract was used to measure caspase activity by mixing in a 96-multiwell plate with a substrate. To calculate caspase activity, the absorbance was measured at 405 nm in an ELISA microplate reader (Meter tech. 960, USA).

Western blot
Effect of the new compounds on protein expression was measured by Western blot as previously described 49 . Exponentially growing, 50% con uent MCF7 cells were treated with 1 or 3 μM of the new HDAC inhibitors or vorinostat for 48 h. Control cells were treated with DMSO. Adherent cells were lysed using lysis buffer (1 mM Tris-HCl [pH 6.8], 2% w/v SDS, 10% glycerol, 2 mM PMSF), and a commercial protease inhibitor mixture (Complete Protease Inhibitor Mixture; Roche Molecular Biochemicals, Germany). Lysates were sonicated on ice, and the protein concentration was measured using a Bicinchoninic Acid (BCA) protein assay kit (Pierce-Life Technology, USA). A 30 μg fraction of the total cell lysate was loaded, resolved on SDS PAGE and electro-transferred to nitrocellulose membranes.
Membranes were blocked using TBS-Tween 20 buffer containing 5% nonfat dried milk. The primary antibody solutions (1:1000; all Cell Signaling Technology) in 5% BSA TBS-T were incubated at 4 °C for overnight. Anti-species secondary antibody (Cell Signaling Technology) binding was carried out at room temperature for 1h. Membranes were then were washed three times for 10 min with TBS-T. Chemiluminescence detection was performed with an ECL plus Detection System (Pierce-Life Technology, USA). Quanti cation of the relative intensity of individual bands was carried out using the ChemiDoc Touch Imaging system from BioRad (USA) and band intensities were normalized to corresponding GAPDH level. Bands of untreated cells were used as a reference standard and intensities of other bands were expressed relative to them.

Cell cycle distribution analysis
Cell cycle studies were carried out using ow cytometry as described before 50 . 5 × 10 5 cells were treated with 1μM of new HDAC inhibitor or vironostat. Control cells were treated with DMSO and incubated at 37°C . Cells were collected at different time intervals (4, 24, 48 and 72h), xed in 70% ethanol, treated with RNAse and then 50 μl propidium iodide solution (1 mg/ml) was added and the DNA content was analyzed by ow cytometry using a FACS machine (Becton Dickenson, Germany).

RNA isolation
MCF7 cells were cultured in 175cm 2 tissue culture asks. After overnight incubation, exponentially growing 50% con uent cells were treated with 3µM of the compounds under investigation (14,15,19,22 and 28 and vorinostat) for 24h. At the end of the treatment period, cells were collected by centrifugation for 5 min at 300g and cell count was adjusted to 5 x 10 6 cells/ml. Total RNA was extracted using the ReliaPrepTM RNA cell miniprep system (Promega, USA) as described by the manufacturer. The RNA was quanti ed using Nanodrop. Studies were repeated in triplicate.

Whole transcriptome analysis
The RNA from MCF7 cells treated with compounds 14, 15, 19, 22 and 28 and vorinostat for 24h as well as a control of untreated MCF7 cells were treated with Turbo DNase (Thermo sher, USA) and a whole transcriptome library was constructed using the Ion Ampliseq transcriptome kit targeting 21,000 genes. The library was enriched using the Ion OneTouch system (Thermo sher, USA) and subject to Next Generation sequencing using the Ion Proton (Thermo she, USAr) as described previously 22 . 4.1.9. Bioinformatics analysis FASTQ les were converted to BAM les by aligning the genes to reference sequence HG19. The RNA count was carried out using the RKPM method to provide raw expression values of the genes. These were processed using the R script. The R script normalized the triplicate data using a negative binomial algorithm and the processed expression data was used to calculate the fold-change of genes related to HDAC enzymes, acetylated histones, survival, apoptosis and cell cycle markers. The fold-change was calculated by comparing each compound to the untreated control using vorinostat as the gold standard HDAC inhibitor as described previously 22  summed to generate the UC signal. Assays were performed in triplicate. Standard curves were generated by adding 4 mL of 50x of ve concentrations of 28 in DMSO to 40 mL of acetonitrile in UV Star® plates followed by 156 mL of phosphate buffered saline. Analysis and statistics were performed using GraphPad® Prism v. 5.04. Data are reported as the maximum concentration observed in the ltrate. 4.1.11. Procedure for the assessment of mouse and human liver microsomal stability for 28 7,20,21 : The clearance of 28 in mouse or human liver microsomes was determined at 37 o C. Assays were conducted in 96-deep well polypropylene plates. Test compounds (1 mM) were incubated in 0.5 mL of 100 mM potassium phosphate buffer (pH 7.4) with 0.5 mg/mL pooled liver microsomes from CD-1 mice (Life Technologies, Grand Island, NY, USA) or pooled liver microsomes from humans (ThermoFisher, Waltham, MA, USA), 2 mM tetra-sodium NADPH and 3 mM magnesium chloride for 60 minutes at 37 o C with gentle shaking. AT ve time points, 75 mL of reaction mixture was transferred to 96-well shallow well stop plates on ice containing 225 mL of acetonitrile with 0.1 mM propafenone as internal standard. Control reactions (lacking NADPH) were performed in a similar manner to demonstrate NADPH dependency of compound loss. Standard curves for 28 were generated using ve concentrations in triplicate that were processed as above but with zero incubation time. Stop plates were centrifuged at 2000g for 10 minutes and then 170 mL of the supernatants were transferred to a Waters Aquity® UPLC 700 mL 96-well sample plate with cap mat (Waters, Milford, MA, USA). The amount of compound remaining in the supernatant was quanti ed by LC/MS/MS using a Waters TQ MS (electrospray positive mode) coupled to a Waters Aquity® UPLC (BEH column, C18, 1.7 mm, 2.1 x 50 cm, gradient of acetonitrile/water/0.1% formic acid).
Propafenone was used as the internal standard. GraphPad® Prism v. 5.04 was used for nonlinear tting of time course data to generate t 1/2 values. Results for assays lacking NADPH were expressed as percent of compound remaining after 69 minutes. 4.1.12. Procedure for the assessment of inhibitory activity of 28 for human CYP450 3A4 7, 20, 21 : Inhibition of CYP3A4 was assessed by measuring the ability of 28 to inhibit the conversion of midazolam to 10hydroxymidazolam. Ten concentrations of 28 were examined in triplicate (half-log serial dilutions). Assays were conducted in 1 mL 96-well polypropylene plates containing 100 mL of 100 mM potassium phosphate buffer (pH 7.4), 3 mM magnesium chloride, 2 mM midazolam, 1 mM tetra-sodium NADPH, Insect supersomes (Corning Gentest, Glendale, AZ, USA, containing 3 pmol/mL human CYP3A4, 47 nmol/min human P450 reductase activity, 14 pmol human cytochrome b5) and insect control microsome protein (Corning Gentest, Glendale, AZ, USA, 0.15 mg/mL). All components except NADPH were added to a prewarmed plate (37 o C) and reactions were initiated by adding NADPH. After 30 minutes, reactions were terminated with 200 mL of acetonitrile containing 30 mM prednisone as an internal standard. After centrifugation for 10 minutes at 2200g, 165 mL of supernatants were transferred to an analysis plate.
Samples were analyzed for 10-hydroxymidazolam concentration by LCMSMS as described above. IC50 values were determined using GraphPad® Prism v. 5.04 nonlinear curve tting. The CYP3A4 inhibitor ketoconazole was used as a positive control.  Data are represented as mean ± SEM and analyzed statistically by ANOVA using Graphpad Prism software (GraphPad Software, USA). The comparison of the effect of the compounds on the HDAC classes was carried out using Anova followed by Bonferroni post hoc analysis. The statistical signi cance between groups was de ned as * p < 0.05, ** p < 0.01 and *** p < 0.001.

General Methods 51
Chemical reagents and anhydrous solvents were purchased from Sigma-Aldrich (USA) and were used without further puri cation. Solvents for extraction and column chromatography were distilled prior to use. TLC analysis was performed with silica gel plates (0.25 mm, E. Merck, 60 F 254 ) using iodine and a UV lamp for visualization. Retention factor (R f ) values were measured using a 5 × 2 cm TLC plate in a developing chamber containing the solvent system described. Melting points were measured with a Stuart Melting Point Apparatus (SMP30) in Celsius degrees and were uncorrected. 1 H, 13 C NMR and 2D-NMR experiments were performed on a 500 MHz instrument. Chemical shifts are reported in parts per million (ppm) downstream from the internal tetramethylsilane standard. Spin multiplicities are described as s (singlet), d (doublet), dd (double doublets), t (triplet), (td) triple doublets or m (multiplet). Coupling constants are reported in Hertz (Hz). ESI mass spectrometry was performed on a Q-TOF high-resolution mass spectrometer or Q-TOF Ultim LC-MS. Optical rotations were measured with a digital polarimeter using a 100 mm cell of 10 mL capacity. Single-crystal X-ray diffraction data were collected using an Oxford Diffraction XCalibur, equipped with (Mo) X-ray Source (λ = 0.71073 Å) at 293(2) K. HPLC puri cation was performed on an Agilent 1260 in nity series HPLC spectrometer, using a Restek Ultra AQ C18 5 μm 150 mm × 4.6 mm column, eluted using 0.1 % TFA in water and acetonitrile at 1.0 mL/min and detected at 220 nm. Compound purity was assured by a combination of high-eld multinuclear NMR (H, C), HRMS (ESI-TOF) and HPLC; purity by the later was always >95%. The synthesis of the tested compounds is described in the supplementary material (Supplementary Figures S1-S15). All compounds (14, 15, 19, 22, and 28) were dissolved in 100% DMSO (Sigma Aldrich, USA) and were diluted in medium just before use. The maximum concentration of DMSO did not exceed 0.5% in all experiments and DMSO was utilized as a negative control. Vorinostat (SAHA, Sigma Aldrich, USA) and MAZ1914 were used as positive controls.

4.2.2.Molecular modeling
The crystal structures of HDAC1 (PDB: 6Z2J), HDAC2 (PDB: 6WBZ), HDAC3 (PDB: 4A69), HDAC4 (PDB: 6FYZ), HDAC6 (PDB: 5EDU) and HDAC7 (PDB: 3C0Z) were downloaded from the RCSB Protein Data Bank (www.rcsb.org). Homology models were prepared for HDAC5 and HDAC9. The amino acid sequences of the histone deacetylase domains were retrieved from the Uniprot database (HDAC5: Q9UQL6|684-1028; HDAC9: Q9UKV0|631-978, www.uniprot.org). The Blast homology search engine was used to search for sequence homologs. The crystal structure of the catalytic domain of HDAC4 in complex with a hydroxamic acid inhibitor (PDB: 2VQM) was used as template for both HDAC5 and HDAC9. The template has a percent identity of 77% with HDAC5 and 73% with HDAC9, which ensures high reliability models. The models were built and validated using Prime (Schrödinger, USA). All protein structural les were prepared to adjust bond orders, add missing hydrogens and complete missing side chains and amino acid residues. Water molecules were deleted. The hydrogen bond network was optimized and the structure was then relaxed. Protein preparation was performed by PrepWizard (Schrödinger, USA). The atomic coordinates of the hydroxamic acid inhibitor were copied from HDAC4 to all targets and the protein-ligand complexes were then thoroughly minimized. The ligand was selected to de ne the receptor grid. Ligands were prepared in LigPrep (Schrödinger, USA). Soft docking approach in Glide (Schrödinger, USA) was used to nd the best binding mode of each compound.

4.2.3.
General reaction procedure for the preparation of compound 5 and 6 51 : Scandium tri ate (20 mol%) and sodium sulfate (2.0 mmol) were added to a solution of aldehyde (1, 1.0 mmol) and 2-aminoazine (2/3, 1.0 mmol) in MeOH:DCM (3:1 mL) at rt. After 45 minutes, tert-butyl isocyanide (4, 1.1 mmol) was introduced and stirring was continued at rt for 12-15h. After completion, MeOH and DCM were removed. The residue was diluted with DCM (100 mL) and washed with water (3 x 50 mL). Then, the organic layer was dried over Na 2 SO 4 and concentrated under vacuum. The crude product was puri ed on ash chromatography using 80% EtOAc in hexane as eluent to produce compound 5/6 as white solid.
4.2.4. General reaction procedure for the preparation of compound 8/9 51 : Compound 5/6 (1.0 mmol) and DIPEA (1.4 mmol) were mixed in anhydrous DMF (8 mL) at 0 °C. After 45 minutes, HBTU (1.2 mmol), followed by the amine partner (7; 1.0 mmol), were added and stirring was continued at rt for 10-13 h. After completion, the reaction mixture was diluted with EtOAc (50 mL) and washed with cold water (3 x 30 mL) and brine solution (10 mL). Then, the organic layer was dried over Na 2 SO 4 and concentrated under vacuum. The crude product was puri ed on ash chromatography using EtOAc/hexane as eluent to obtain the title compound 8/9. The obtained regioisomeric mixture was used directly in the next step without any further puri cation.
4.2.5. General reaction procedure for the preparation of compounds 10/11 51 : Compound (8/9, 0.5 mmol) was dissolved in acetic acid (3 mL) and re uxed for 24h. After completion, acetic acid was removed and the crude product was recrystallized from EtOAc to give compounds 10/11.  (14) After completion, the reaction mixture was diluted with EtOAc (50 mL) and washed with cold water (3 x 30 mL) and brine solution (10 mL). Then, the organic layer was dried over Na 2 SO 4 and concentrated under the vacuum to obtain the crude product, which was puri ed on ash chromatography using EtOAc/hexane as eluent to obtain the title compound 26. The obtained regioisomeric mixture was used directly in the next step without any further puri cation. 264 mg, 75% yield. Recently, we reported the synthesis of a three series of imidazopyridine derivatives employing posttransformation of the Groebke-Blackburn-Bienaymé [4+1]-cycloaddition reaction products 13 . The ultimate goal of these efforts was to establish compound libraries of privileged scaffolds for our drug discovery campaign. Some of these derivatives showed promise as starting points for anticancer 17 , antibacterial 18 and BACE1 inhibitor (one of the most important enzymes involved in Alzheimer's disease pathology), drug discovery 19 . In continuation of these efforts, selected compounds from these libraries were examined in a screening campaign against various HDAC isoforms. This initiative was inspired by the variety of cap and metal-chelating groups present in various HDACs inhibitors. As a result, we identi ed novel chemotypes bearing the imidazopyridine core scaffold, as shown in Fig. 1B.
HDAC inhibitors, including vorinostat, share three common structural motifs: a zinc binding moiety (e.g., hydroxamic acid), a tether occupying the hydrophobic sub-pocket in the active site, and a functional cap group that resides outside the hydrophobic hollow and interacts with the HDAC exterior amino acid residues (Fig. 1B). HDAC inhibitors are known to accommodate a wide variety of metal-chelating groups, which allows for the replacement of this group with various bioisosters (e.g., the tri uoromethyl oxadiazole (TMFO) moiety), 20 while maintaining their activities and enhancing their selectivity toward certain HDAC isoforms. Our approach was to apply these characteristics to our newly discovered chemical scaffold and evaluate the binding topographies of various HDAC inhibitors.
We hypothesized that an isophthalic acid core substituted with an imidazopyridine group on one side and a substituted ve-membered heterocyclic ring on the other side might deliver novel motifs that selectively inhibit HDACs. To this end, we screened an in-house library of small molecules encompassing various skeletal and structural diversity including imidazopyridine core scaffold. The latter should represent the cap group and, together with the meta-disubstituted benzene core, should provide an optimal linker for incorporating a hydroxamic acid functionality or other metal-and/or non-metal-chelating groups such as the TMFO moiety, which has been reported to enhance the selectivity toward HDACs as a none-metal binding group (MBG) 20 . To explore whether this structure-based drug design strategy would lead to HDAC inhibitors, a concise set of heterocyclic scaffolds with an embedded imidazopyridine moiety was studied using an induced-t like protocol within the FITTED modeling package. In general, the structures of these analogues possessed: 1) an imidazopyridine cap group; 2) an isophthalic acid core; 3) imidazole moieties, and 4) a hydroxamic acid or TFMO group to t into the zinc binding pocket of HDACs either as a metal-chelating or a metal non-chelating group, respectively (Fig. 1B).
The decision to include an oxadiazole ring as a isosteric replacement for the hydroxamic acid appendage was based on earlier report indicating that this group enhances the selectivity toward HDACs 20 .
Additionally, an H-bond acceptor (e.g., CF 3 -group) attached to this moiety might improve the selectivity and lead to strong interactions with the catalytic amino acid residues in the enzyme active site. Therefore, the phenyl ring portion of compound I was swapped with an imidazopyridine group to enhance a nity for HDACs, leading to compounds of type II, containing an imidazole-like linker connected to a non-zinc chelating binding motif like TFMO, and type III, containing the traditional hydroxamic acid zinc-chelating moiety. To reduce the exibility present in vorinostat, the amide group was embedded into the rigid imidazopyridine moiety system, which retains the essential hydrogen bonding interactions with the amino acid residues in the active site and provided suitable vectors for direct extensions into the corresponding sub-pockets of the active sites (Fig. 1B). Additional rigidity was obtained by replacing the exible alkyl chain of vorinostat with a benzimidazole linker. We envisioned that these concepts should provide new classes of inhibitors with enhanced potency and selectivity. Our results are summarized in Table 1.

Identi cation of initial candidates by molecular modeling
To gain insight into the binding modes of the most potent motifs, we docked a representative set of heterocyclic scaffolds with an embedded imidazopyridine moiety anchored to a phathalic acid core to the X-ray crystal structures of HDACs 1-9 (available in the RSC Protein Data Bank) using the PRIME modeling package available from Schrodinger (New York, NY). Protein preparation was performed using PrepWizard. Ligands were prepared using LigPrep. Soft docking was performed using Glide to nd the best binding mode for each compound and results were calculated using the FITTED program.
Compound 28 was predicted to be the most potent and selective ligand for HDAC1 and HDAC2 (Fig. 2), with docking scores of -4.9 kcal/mol and -8.5 kcal/mol, respectively. The N-hydroxyformimidamide is predicted to be in close proximity to the Zn +2 ion, which coordinates to that group through in both enzymes through its nitrogen and oxygen atoms. The models predicted that 28 forms both strong hydrogen bonds and electrostatic interactions with Asp176 and His141 in active site of HDAC1. For HDAC2, similar electrostatic interactions were seen with His179 and Asp265 but hydrogen bonding was predicted only with Tyr304. These various interactions ensure proper orientation of 28 in the binding sites.
In addition, 28 appears to be interacting with Asp99 and forming π-π contacts with Phe205 in HDAC1.
The hydrophobic phenyl imidazopyridine group is in the solvent exposed region of the binding site for both HDAC1 and HDAC2 and, as such, does not require the loss of signi cant amounts of desolvation energy, which may contribute to the high a nity of 28 towards both enzymes. Compound MAZ1914 was predicted to bind with less a nity to both targets in our models, giving docking scores of -4 kcal/mol for HDAC1 and -4.5 kcal/mol for HDAC2. The oxadiazole interacts in the same way as that of the hydroxylformimid amide group of 28 but with less e ciency. In addition, MAZ1914 shows fewer interactions with the surrounding amino acids in comparison with 28.
In general, the compounds that show considerable activities against HDAC targets are those that can coordinate with Zn +2 ion in the binding pockets. The compounds appear to orient the hydroxamate, oxadiazole, hydroxyformimidamide or similar group to interact with Zn +2 and/or aspartate and histidine amino acids that coordinate Zn +2 in the binding sites of HDAC targets. The more active compounds also show additional interactions that augment ligand a nity including hydrogen bonding with side chain and backbone N/O atoms, π-π stacking with Phe/His residues, aromatic bonds with polar amino acids, cationπ interactions with Arg residues, and hydrophobic contacts with several amino acids in the binding pockets. Although some molecular fragments are solvent exposed, which can reduce the biological activity if the compounds have to go through an energetically unfavorable desolvation step, these fragments in most of the cases are hydrophobic in nature and should not suffer signi cantly from the loss in binding a nity.

Synthesis
The preparation of the most promising compounds in this study was carried out as described in Schemes 1-4. Since our modeling studies predicted that the most potent compounds would fall within Classes II and III (Fig. 1B), we focused our synthetic efforts on those scaffolds. Four variations were synthesized.
The rst variant involved linking an imidazopyridine cap group to a zinc-chelating or metal none-chelating binding group through a linker containing a meta-substituted phenyl ring and a benzimidazole moiety (Scheme 1A). These disubstituted isophthalic acid scaffolds were assembled starting with isophthalaldehyde (1). This starting material was treated in a multicomponent [4+1]-cycloaddition reaction with either 2 or 3, followed by tert-butyl isocyanide (4), to deliver the Groebke-Blackburn-Bienaymé products 5 and 6, respectively, which were subsequently used without puri cation. These intermediates were then subjected to a coupling reaction with the phenelenediamine derivative 7, followed by acid catalyzed cyclization, to give the benzimidazole derivatives 10 and 11. Treatment of 10 and 11 with hydroxyl amine delivered the hydoxamic acids 12 and 13, respectively. Reaction of these two intermediates with tri uoroacetic acid delivered the desired tri uoro-oxadiazole derivatives, 14 and 15.

In Vitro Drug-Like Properties
To investigate the drug-like properties of our new chemical scaffold we examined 28 in a battery of in vitro physicochemical/ADME assays. These assays are routinely used in drug discovery to assess the potential for metabolism and drug-drug interactions and predict in vivo exposure and bioavailability. Our results are presented in Table 1. It is noteworthy that the IC50 of the positive control SAHA was higher than 10 µM for HDAC4,5,7 and 9, which is consistent with previously published results 21 .

The effect of the new HDAC inhibitors on the expression and activity of various HDAC isozymes
To further validate the results of the HDAC enzymatic assay, the effects of compounds 14, 15, 19, 21 and 28 on different HDACs were investigated in the MCF7 breast cancer cell line using gene expression and Western blot analysis (Fig. 3). As shown in Fig. 3A, 28 showed the most potent activity as an inhibitor of Class I HDACs (HDACs 1 and 2). Compound 28 inhibited gene expression by 2.2 fold for HDAC3 and 3.2 fold for HDAC1 compared to control cells. Its inhibition of class I HDACs gene expression was stronger than that seen for vorinostat. Compound 14 was also more effective than vorinostat at inhibiting Class I HDAC gene expression. Compounds 22 and 19 were found to be equally effective compared to vorinostat, while 15 actually increased the expression level of all class I HDACs as compared to control cells.
All of our tested compounds demonstrated inhibitory activity on class III (Sirtuin 1-7) gene expression, with 28 being more active than 19, 22, or 14 (Fig. 3D). Interestingly, our compounds were effective at inhibiting gene expression for all seven sirtuins while vorinostat actually increased gene expression for SIRT2 and SIRT4. These data suggest that our compounds could be more effective than vorinostat at treating sirtuin-associated diseases, including cancer, HIV, metabolic disorders and pathologies associated with neurological diseases 22 .

Enhanced acetylation of histone proteins by the new HDAC inhibitors
Since HDACs de-acetylate histone proteins, HDACi's would be expected to increase levels of acetylated histones, we, therefore, examined the effects of our compounds on the acetylated forms of histones H2A, H3 and H4 in MCF7 cells by Western blot (Fig. 4). Compounds  Interestingly, all compounds increased the level of H3 acetylation more than vorinostat; however, the latter showed a greater effect on histone H2A compared to our compounds (Fig. 4C). This nding is consistent with previous reports, which showed that not all histones that are acetylated correlate with histone mRNA gene expression 23 . These ndings were in accordance with that seen from the gene expression data, which showed that 14, 22 and 28reduced the expression level of C-MYC more than vorinostat (SAHA) as shown in Fig. 5C. To further con rm the apoptotic effect of the new compounds, their effects on the expression of caspases 3/7 were studied. This study revealed that all compounds increased the expression of these caspases at levels that were either similar to (15,19) or higher than vorinostat (14,22,28) (Fig. 5D). These ndings were also observed in the gene expression studies. Furthermore, the results indicate that the new compounds led to a downregulation of the expression of the anti-apoptotic genes BCL3 and BCL2 compared to the control group. Compound 28 showed the greatest effect on downregulation of BCL2 and BCL3 expression, with 14, 19 and 22, downregulating BCL2 and BCL3 expression to a lesser extent, as shown in Fig. 5E. Its noteworthy to mention that vorinostat did not downregulate the expression of BCL3 (Fig. 5E).
To gain more insight into the anti-proliferative activity of the new compounds, their effect on the expression of NF-kB was studied. NF-kB is a transcription factor that regulates the expression of many genes involved in cell survival and the cellular response to stress. It is constitutively active in different types of tumors, activating anti-apoptotic genes and inhibiting caspase enzymes. Our compounds showed variable effects on the expression of NF-kB genes. Compound 22 and vorinostat downregulated the expression of NF-kB1, RELA (P65) and RELB while upregulating the expression NF-kB2 (p52).
However, 19 showed a slight reduction in NF-kB gene expression. The highest reduction in NF-kB gene expression was observed with 14 and 28, including effects on NF-kB2, which was also upregulated by vorinostat. On the other hand, 15 enhanced the expression of all measured NF-kB genes (Fig. 5F). Taken together, these data indicate that the ve new compounds, in particular 22 and 28, showed promising anti-proliferative effects on MCF7 cells compared to vorinostat through modulating the expression of cell survival and apoptotic genes. induced arrest in G2/M phases (Fig. 6A & B). At different time points (24h, 48h and 72h), variable percentages of subG1 cells (indicative of apoptosis), were detected for all compounds except for 14. At the 24h time point, the induced fraction of subG1 by 19 was higher than that seen with vorinostat (SAHA), whereas at the 48h and 72h time points, vorinostat was more effective than the new compounds in the induction of subG1 cells (Fig. 6C). These ndings agree with the gene expression results which indicated that our compounds affected the expression of many genes involved in cell cycle progression at different levels. Compound 28 was the most effective inhibitor of cell cycle progression genes, followed by 14 and 19 (Fig. 6D).
Our compounds also decreased the expression of the cyclin-dependent kinases CDK2, CDK4 and CDK6, which are involved in the progression of cells from G1 to S phase. Additionally, the compounds also reduced the expression of CDK1, which regulates the progression of cells from G2 to M phase. Moreover, our compounds downregulate the expression of both E2F1 and RB1 genes, which are known to be cell cycle regulators. Interestingly, vorinostat showed the most effective inhibition of E2F1. However, of the compounds under investigation, 28 showed the most consistent downregulation of all cell cycle progression genes studied. Collectively, these results suggest that the new compounds induce cell cycle arrest through downregulation of cell cycle genes. The clinically available HDACi's are used for hematologic malignancies; however, the use of HDACi's for the treatment of solid malignances such as breast cancer has not been pursued as aggressively despite a correlation between elevated levels of HDACs and tumor aggressiveness, expression of hormone receptors (ER, PR and HER2) and the response of breast cancer cells to hormonal therapy was previously extensively studied [28][29][30][31] .

Discussion
In the present study, we identi ed a new class of HDACi's as potential anticancer lead drug candidates that may possess a novel mechanism of action through screening of an in-house compound library containing various core scaffolds and substitution patterns against representatives of zinc-dependent The biological effects of hits on MCF7 cells (derived from solid tumors) was then investigated. Our evaluation included enzymatic inhibition, effects on gene expression and whole cell anit-proliferative assays. , \The results of these assays are summarized in Table 2 and 3 and Fig. 3-6. The HDAC inhibitory assay provided valuable information on the structure-activity relationship. Strikingly, two of the compounds were found to be potent HDAC5 inhibitors (14 and 15) with IC 50 values ca. 300 nM ( Table 2).
Compound 15 was found to be 33-fold more selective toward HDAC5 compared to other HDACs studied.
The most potent HDAC1 and HDAC2 inhibitor was 28, with an IC 50 value of 550 and 400 nM, respectively.
Compound 28 was found to be 44-fold more selective toward HDACs1 and 2 compared to other HDACs studied. Some structure-activity conclusions can be made. Compounds with a tri uoromethyloxadiazole (TFMO)-benzimidazole cap group anchored on imidazo-azine isophthalic acid derivative (14 and 15) displayed potency and selectivity in the enzymatic HDAC assay toward HDAC5, and showed better potency (more than 33-fold) compared to the FDA-approved drug vorinostat. Interestingly, replacing the tert-butyl group of 14 and 15 with a hydrogen and substituting the tri uoromethyloxadiazole ring with Nhydroxyl amidine (as in 28) diminished the activity toward HDAC5. The latter change however, enhanced the selectivity and potency of 28 toward HDAC1 and 2 by more than 44-fold. Furthermore, the uoropheneylene diamine group (22), was found to enhance the potency toward HDAC3 and HDAC9 compared to other HDACs studied. These results could be rationalized based on the biased functional groups present in these motifs. For example, the TFMO group has been reported to possess selectivity towards Class IIa HDACs, whereas the amino/hydroxy amidine moieties enhances the selectivity towards class I HDACs 32 . Most of the known HDAC inhibitors possess a common pharmacophore which comprises a zinc-chelating group in the active site, a linker that accesses the hollow of the active site and a group that networks with the external surface 32 . With this in mind, the observed selectivity of our compounds against the different HDACs isoforms, might be due to the novel system comprising a tri uoromethyloxadiazole (TFMO) group appended to an imidazopyridine-isophthalic acid core.
To further validate the HDACi inhibitory activity of our compounds, their cytotoxic effects was studied.  35 . This might be due to the fact that the acetylation of H3 can also lead to other posttranslation modi cations such as methylation. For example, a 2015 study reported that trimethylation of lysine36 on H3 restricted gene expression, resulting in a discrepancy between H3 mRNA and protein expression 35 .
HDAC inhibition alters the balance between histone acetyl transferase activity and HDAC activity, resulting in elevated acetylation of histone proteins. Acetylation of histone lysine residues is associated with an open chromatin structure and transcriptional activation of many genes involved in signal transduction, DNA repair, cell cycle regulation and cell death pathways such as apoptosis, autophagy and cell senescence 25 . As expected, three of our compounds (15, 19 and 28), showed a signi cant increase in the levels of acetylation of various histones, which con rms their HDACs inhibitory effects. These ndings are in line with a previous report from Bali et al 36  To further validate the anti-cancer activity of our compounds, we examined their effects on the expression of apoptosis and cell cycle progression markers. The balance between survival proteins such as C-MCY, BCL2, BCL3 and NFkB and death proteins such as caspases3 and 7 determines the fate of cancer cells treated with anticancer therapeutics. NF-kB is a transcription factor that regulates the expression of many genes involved in cell survival and cellular response to stress. It is constitutively active in different types of tumors, activating anti-apoptotic genes and inhibiting caspase enzymes. BCL3 is a key member of the NF-kB signaling cascade and is known to be involved in regulating many cellular functions, including survival, proliferation, in ammation and immune response. Increased cellular proliferation or survival is associated with BCL3 expression and activation 40 . Furthermore, BCL3 transcriptional repressor function has been shown to play a role in regulating immune responses and in the development and activation of immune cells 41 . The fact that the expression of BCL3 is downregulated in cancer cells indicates a modi cation in the immune response as well as survival and proliferation outcomes of cell signaling. In our studies the anti-apoptotic gene BCL2 was overexpressed in control cells but was downregulated in cells treated with our compounds (Fig. 5E). BCL2 expression is known to be upregulated in many primary tumors. Downregulation of BCL2 results in reduced proliferation and pro-apoptotic effects in these tumors. C-MYC is another survival protein that is overexpressed in different types of liquid and solid malignancies and has been reported as a known target of HDACi 42 . The upregulation of caspase7 and downregulation of C-MYC, BCL2, BCL3 and NFkB by our compounds support their role in enhancing cells apoptotic machinery, which further explains the anticancer effect of our new HDACi's against solid malignancies such as, MCF7 breast cancer cells. These results are consistent with previous ndings 39, [43][44][45] .
Cell cycle arrest is another effect of HDAC inhibition 38 . TA signi cant increase in the fraction of MCF7 cells in G1/S and G2/M phases was detected after treatment with our compounds. This effect was accompanied by a downregulation in the expression of proteins involved in cell cycle progression such as E2F1, RB1 and cyclin-dependent kinases (CDKs 1, 2, 4 and 6). These results are also in agreement with previous ndings which indicated a reduction in the levels of cyclins and CDKs after treatment of cancer cells with HDACi's 39,46 .
Inhibition of HDACs (resulting in increased acetylation of histone lysine residues) can explain the biological effects of our new compounds (anti-proliferation, apoptosis and cell cycle effects), but does not fully explain the enhanced activity seen with 22 and 28 compared to that observed for 14, 15 and 19. It is possible that selective effects on other proteins and/or signaling pathways may occurring. This hypothesis will be tested using genetic knockout/knockdown and overexpression of individual HDACs and will be the subject of future reports.
In summary, we have developed a novel series of selective imidazopyridine-based HDAC inhibitors. The data presented in this article demonstrated that the peptidomemtic-based core structure represents a new class of selective HDAC inhibitors with good activity against MCF7 solid tumor cells. Our new compounds upregulated the expression of caspase7 whereas they downregulated the expression of C-MYC, BCL2, BCL3 and NFkB. Additionally, the new compounds downregulated the expression of cell cycle progression proteins E2F1, RB1 and cyclin-dependent kinases (CDKs 1, 2, 4 and 6). These ndings suggest that our compounds should be effective in treating solid malignancies.
Taken together, the biochemical, Western-blotting, whole cell assays and transcriptomic studies indicated that this novel series holds promise for further development as potential lead drug candidates for the treatment of cancerous disease states, including solid tumors.