[1,5]-Hydride Shift-Cyclization versus C(sp2)-H Functionalization in the Knoevenagel-Cyclization Domino Reactions of 1,4- and 1,5-Benzoxazepines

Domino cyclization reactions of N-aryl-1,4- and 1,5-benzoxazepine derivatives involving [1,5]-hydride shift or C(sp2)-H functionalization were investigated. Neuroprotective and acetylcholinesterase activities of the products were studied. Domino Knoevenagel-[1,5]-hydride shift-cyclization reaction of N-aryl-1,4-benzoxazepine derivatives with 1,3-dicarbonyl reagents having active methylene group afforded the 1,2,8,9-tetrahydro-7bH-quinolino [1,2-d][1,4]benzoxazepine scaffold with different substitution pattern. The C(sp3)-H activation step of the tertiary amine moiety occurred with complete regioselectivity and the 6-endo cyclization took place in a complete diastereoselective manner. In two cases, the enantiomers of the chiral condensed new 1,4-benzoxazepine systems were separated by chiral HPLC, HPLC-ECD spectra were recorded, and absolute configurations were determined by time-dependent density functional theory- electronic circular dichroism (TDDFT-ECD) calculations. In contrast, the analogue reaction of the regioisomeric N-aryl-1,5-benzoxazepine derivative did not follow the above mechanism but instead the Knoevenagel intermediate reacted in an SEAr reaction [C(sp2)-H functionalization] resulting in a condensed acridane derivative. The AChE inhibitory assays of the new derivatives revealed that the acridane derivative had a 6.98 μM IC50 value.


Introduction
C-H activation and functionalization of sp 3 -and sp 2 -hybridized carbons offer atom-economical and step-efficient ways for the preparation of polyfunctionalized condensed heterocycles, which have C-H activation and functionalization of sp 3 -and sp 2 -hybridized carbons offer atom-economical and step-efficient ways for the preparation of polyfunctionalized condensed heterocycles, which have been intensely investigated recently [1][2][3][4]. The metal-free sp 3 C-H functionalization via internal redox processes or hydride transfer has attracted much attention due to its unique features. Instead of using an external oxidant such as a high-valent transition metal [5,6], the sp 3 C-H activation can be induced by an internal [1,5]-hydride shift between a π electron-poor double bond as acceptor and a C-H bond adjacent to an oxygen or nitrogen as a hydride donor (A→B, Scheme 1a) [7,8]. The resultant zwitterionic intermediate B can undergo further cyclization producing condensed heterocycles C. Scheme 1. (a) General scheme for the C (sp 3 )-H activation-cyclization sequence induced by [1,5]-hydride shift of a tertiary amine or ether. (b) [1,5]-Hydride shift-cyclization sequence of N-aryl-1,4-benzoxazepines carried out in our present work.
In this work, the N-aryl-1,5-benzoxazepine derivative 4 was converted to the carbocationic intermediate 5 with 1,3-dimethylbarbituric acid in a Knoevenagel reaction, which cyclized with the activated benzene ring to produce a condensed 9,10-dihydroacridine (acridane) derivative 6 in an S E Ar reaction. The different position of the nitrogen atom in the benzoxazepine scaffold induced a sequence different from those depicted in Scheme 1b for the regioisomeric rac-1; a C(sp 2 )-H functionalization took place instead of the [1,5]-hydride shift-cyclization, which was observed for N-aryl-1,4-benzoxazepines. The transformation of 4 to 6 involves the activation of the carbonyl carbon during the Knoevenagel reaction and a C(sp 2 )-H functionalization during the S E Ar reaction.
Acetylchloninesterae (AChE) inhibitors are used as drugs for the symptomatic treatment of Alzheimer's disease (AD), and some of them such as galantamine contain condensed O,N-heterocyclic skeleton [24]. AChE inhibitors prevent the action of cholinesterases (ChEs) and thus increase the levels of acetylcholine in the brain and improve the cholinergic functions in AD patients [25]. The inhibition of AChE is a potential therapeutic target for the treatment of AD but targeting ChEs alone is not sufficient and thus exploration of multi-target molecules focusing also on other proposed molecular mechanisms of AD such as β-amyloid cascade or oxidative stress is required. The multi-target approach was pursued in recently reported works by exploring synthetic heterocyclic derivatives, which exhibited both AChE and β-secretase inhibitory activity [26,27].
In the first step, a regioselective Schmidt reaction of rac-7a-c was carried out with NaN 3 affording the corresponding 3,4-dihydro-1,4-benzoxazepine-5(2H)-ones rac-8a-c (Scheme 3), which were reduced with LiAlH 4 under reflux to provide 2,3,4,5-tetrahidro-1,4-benzoxepines (rac-9a-c). In the final step, the N-arylation of rac-9a was performed with 2-fluoro-5-nitrobenzaldehyde to result in rac-1a in 95% yield. For the N-arylation of rac-9b and rac-9c, the more economical but less reactive 2-chloro-5-nitrobenzaldehyde could be used instead. Furthermore, rac-1a-c were reacted with 1,3-dimetylbarbituric and Meldrum's acids, which initiated a Knoevenagel- [1,5]-hydride shift-cyclization domino reaction (Scheme 4). The 1-and 2-naphthyl groups are known to adopt different orientations relatively to the condensed heterocyclic skeleton and they also exhibit different rotational energy barriers for their rotations, which may result in different bioactivities as demonstrated recently for our synthetic pterocarpans having antiproliferative activity [32]. The Knoevenagel intermediates 2A and 2B formed with 1,3-dimethylbarbituric acid and Meldrum's acid, respectively, underwent a regioselective [1,5]-hydride shift with the participation of the 5-H and the resultant zwitterionic intermediate of type B (Scheme 1a) reacted further in a diastereoselective 6-endo cyclization affording the rac-trans-10b,c, rac-trans-11b,c and rac-10a and rac-11a (Scheme 4). The formation of the other regio-or diastereomeric products was not observed. The complete regioselectivity could be readily explained on the basis of the larger reactivity of the benzylic C-H bond compared to that of the 3-C-H, while the trans-diastereoselectivity was favored by the attack of the nucleophile from the face opposite to the orientation of the bulky naphthyl group linked at C-2. The relative configuration of the C-2 and C-15a chirality centers could not be determined on the basis of NOE correlations, since neither diastereomers would show NOE correlation between 2-H and 15a-H due to the large interatomic distance and orientation. In the trans-diastereomers, the tetrahydrooxazepine ring is expected to adopt a boat conformation with trans-diaxial orientation of the 2-Hax and 3-Hax protons giving rise to a large coupling 3 J2-H,3-H constant (~11 Hz) similarly to the 2-phenyl analogues [29], which could be observed in the 1 H NMR spectra of 10b,c and 11b,c. In contrast, the tetrahydrooxazepine ring of the cis diastereomer would adopt a twist boat conformation with gauche arrangement of the 2-H and 3-Hs, which would produce small values for both 3 J2-H,3-H coupling constants.
With the malonitrile reagent, the reaction stopped at the stage of the Knoevenagel products 12a-c. Stronger reaction conditions or microwave activation produced only partial cyclization or degradations.
Since the 1a starting material was achiral and its domino reaction introduced a new chirality center, the formation of 10a and 11a offered the possibility to test enantioselective organocatalytic reactions with non-covalent optically active organocatalysts. According to literature data, enantioselective organocatalytic versions of [1,5]-hydride shift-cyclization sequences are limited to substrates, in which the alkene double bond of the acceptor unit contains two alkoxycarbonyl substituents able to interact with the organocatalyst in stereoselective manner [33][34][35]. In order to determine the absolute configuration of the enantiomers of 10a and 11a and correlate it with the characteristic electronic circular dichroism (ECD) transitions, the enantiomers of 10a and 11a were separated by chiral HPLC using Chiralpack IC stationary phase and online HPLC-ECD spectra of The Knoevenagel intermediates 2A and 2B formed with 1,3-dimethylbarbituric acid and Meldrum's acid, respectively, underwent a regioselective [1,5]-hydride shift with the participation of the 5-H and the resultant zwitterionic intermediate of type B (Scheme 1a) reacted further in a diastereoselective 6-endo cyclization affording the rac-trans-10b,c, rac-trans-11b,c and rac-10a and rac-11a (Scheme 4). The formation of the other regio-or diastereomeric products was not observed. The complete regioselectivity could be readily explained on the basis of the larger reactivity of the benzylic C-H bond compared to that of the 3-C-H, while the trans-diastereoselectivity was favored by the attack of the nucleophile from the face opposite to the orientation of the bulky naphthyl group linked at C-2. The relative configuration of the C-2 and C-15a chirality centers could not be determined on the basis of NOE correlations, since neither diastereomers would show NOE correlation between 2-H and 15a-H due to the large interatomic distance and orientation. In the trans-diastereomers, the tetrahydrooxazepine ring is expected to adopt a boat conformation with trans-diaxial orientation of the 2-H ax and 3-H ax protons giving rise to a large coupling 3 J 2-H,3-H constant (~11 Hz) similarly to the 2-phenyl analogues [29], which could be observed in the 1 H NMR spectra of 10b,c and 11b,c. In contrast, the tetrahydrooxazepine ring of the cis diastereomer would adopt a twist boat conformation with gauche arrangement of the 2-H and 3-Hs, which would produce small values for both 3 J 2-H,3-H coupling constants.
With the malonitrile reagent, the reaction stopped at the stage of the Knoevenagel products 12a-c. Stronger reaction conditions or microwave activation produced only partial cyclization or degradations.
Since the 1a starting material was achiral and its domino reaction introduced a new chirality center, the formation of 10a and 11a offered the possibility to test enantioselective organocatalytic reactions with non-covalent optically active organocatalysts. According to literature data, enantioselective organocatalytic versions of [1,5]-hydride shift-cyclization sequences are limited to substrates, in which the alkene double bond of the acceptor unit contains two alkoxycarbonyl substituents able to interact with the organocatalyst in stereoselective manner [33][34][35]. In order to determine the absolute configuration of the enantiomers of 10a and 11a and correlate it with the characteristic electronic circular dichroism (ECD) transitions, the enantiomers of 10a and 11a were separated by chiral HPLC using Chiralpack IC stationary phase and online HPLC-ECD spectra of the separated enantiomers were recorded. HPLC-ECD measurements aided with time-dependent density functional theory electronic circular dichroism (TDDFT-ECD) calculations represent an efficient method to study the enantiomers in scalemic or racemic mixtures of natural [36,37] or synthetic derivatives [38,39]. The solution TDDFT-ECD approach [40] was applied to determine the absolute configuration of the separated enantiomers by comparing the experimental HPLC-ECD spectra with the computed ones. The Merck Molecular Force Field (MMFF) conformational search of the arbitrarily chosen (S) enantiomers of 10a and 11a resulted in 5 and 2 conformers in a 21 kJ/mol energy window, respectively. These conformers were re-optimized at various density functional theory (DFT) levels (B3LYP/6-31G(d), B97D/TZVP PCM/n-octanol and CAM-B3LYP/TZVP PCM/n-octanol) yielding only one major conformer at all the applied levels ( Figure 1). the separated enantiomers were recorded. HPLC-ECD measurements aided with time-dependent density functional theory electronic circular dichroism (TDDFT-ECD) calculations represent an efficient method to study the enantiomers in scalemic or racemic mixtures of natural [36,37] or synthetic derivatives [38,39]. The solution TDDFT-ECD approach [40] was applied to determine the absolute configuration of the separated enantiomers by comparing the experimental HPLC-ECD spectra with the computed ones. The Merck Molecular Force Field (MMFF) conformational search of the arbitrarily chosen (S) enantiomers of 10a and 11a resulted in 5 and 2 conformers in a 21 kJ/mol energy window, respectively. These conformers were re-optimized at various density functional theory (DFT) levels (B3LYP/6-31G(d), B97D/TZVP PCM/n-octanol and CAM-B3LYP/TZVP PCM/n-octanol) yielding only one major conformer at all the applied levels ( Figure 1).  Although enantioselective [1,5]-hydride shift-cyclization cascade reaction of Knoevenagel intermediates prepared with dialkylmalonate were reported using optically active metal complexes the separated enantiomers were recorded. HPLC-ECD measurements aided with time-dependent density functional theory electronic circular dichroism (TDDFT-ECD) calculations represent an efficient method to study the enantiomers in scalemic or racemic mixtures of natural [36,37] or synthetic derivatives [38,39]. The solution TDDFT-ECD approach [40] was applied to determine the absolute configuration of the separated enantiomers by comparing the experimental HPLC-ECD spectra with the computed ones. The Merck Molecular Force Field (MMFF) conformational search of the arbitrarily chosen (S) enantiomers of 10a and 11a resulted in 5 and 2 conformers in a 21 kJ/mol energy window, respectively. These conformers were re-optimized at various density functional theory (DFT) levels (B3LYP/6-31G(d), B97D/TZVP PCM/n-octanol and CAM-B3LYP/TZVP PCM/n-octanol) yielding only one major conformer at all the applied levels ( Figure 1).  Although enantioselective [1,5]-hydride shift-cyclization cascade reaction of Knoevenagel intermediates prepared with dialkylmalonate were reported using optically active metal complexes Although enantioselective [1,4]-hydride shift-cyclization cascade reaction of Knoevenagel intermediates prepared with dialkylmalonate were reported using optically active metal complexes [41] Molecules 2020, 25, 1265 7 of 19 or organocatalysts [33][34][35], enantioselective transformations have not been published yet for Knoevenagel intermediates prepared with 1,3-dimethylbarbituric acid and Meldrum's acid. Thus, we performed the 1a → 10a and 1a → 11a transformations in the presence of optically active TADDOL-and chinchona-type non-covalent organocatalysts to test the possibility of enantioselective reactions but unfortunately no notable enantiomeric excess (ee ≤ 7.3%) could be achieved with slow and small conversion of the starting material (see Supporting Information for details).
Molecules 2020, 25, x FOR PEER REVIEW 7 of 19 [41] or organocatalysts [33][34][35], enantioselective transformations have not been published yet for Knoevenagel intermediates prepared with 1,3-dimethylbarbituric acid and Meldrum's acid. Thus, we performed the 1a → 10a and 1a → 11a transformations in the presence of optically active TADDOL-and chinchona-type non-covalent organocatalysts to test the possibility of enantioselective reactions but unfortunately no notable enantiomeric excess (ee ≤ 7.3%) could be achieved with slow and small conversion of the starting material (see Supporting Information for details).
The N-aryl derivative 4 was reacted with 1,3-dimethylbarbituric acid and instead of the expected domino Knoevenagel- [1,5]-hydride shift-cyclization sequence, the 19 precursor of the Knoevenagel product formed a stabilized benzylic carbocation (5), which initiated an S E Ar reaction with the activated benzene ring affording the condensed acridane derivative 6 with 64% isolated yield (Scheme 6). According to the NMR data, all the methylene protons of the oxazepine moiety remained intact, while an aromatic C-H group was changed to a quaternary carbon. Two methine carbons appeared in the DEPT 13 C NMR spectrum at 49.5 and 59.0 ppm corresponding to C-10 and C-1 , respectively, the protons of wich had a 4.4 Hz 3 J 10-H,1 -H coupling constant. The condensed 2,3-dihydro-1H,8H- [1,4]oxazepino[2,3,4-de]acridine moiety has not been reported yet and direct conversion of 2-(N-arylamino)benzaldehyde derivatives to acridanes under mild conditions represents a new synthetic potential. The reaction of 4 with malonitrile stopped at the stage of the Knoevenagel product 20, which could not be converted to a cyclized product even under stronger reaction condition (higher temperature, microwave activation). The reaction with Meldrum's acid produced a complex reaction mixture, which has not been analyzed further.

Bioactivity Studies
The neuroprotective and AChE inhibitory activities of the new compounds were tested to compare neuroprotective activity with reported benzoxazepines and to identify candidates for multi-target approach. Neuroprotective activities were tested against hydrogen peroxide (H2O2), β-amyloid23-35 fragment (Aβ23-35) and oxygen-glucose deprivation (OGD)-induced neurotoxicity in human neuroblastoma SH-SY5Y cells [45,46]. The preliminary screenings showed that the tested compounds did not have remarkable neuroprotective activity at 10 μM concentration, which suggested that the lack of a C-2 aryl moiety or its extension to naphthyl groups results in the loss of activity [29]. Moderate AChE inhibitory activity was identified for the N-aryl-1,4-benzoxazepine 1a at 4.0 × 10 −5 M concentration (58.26% inhibition), while the acridane derivative 6 showed AChE inhibitory activity with 6.98 × 10 −6 mol/L IC50 value.

Materials and Methods
Melting points were determined on a Kofler hot-stage apparatus (Wagner and Munz, Munich, Germany) and are uncorrected. The reactions were monitored by thin layer chromatography (TLC). TLC plates were visualized under a UV lamp and developed by phosphomolybdic acid solution. The NMR spectra were recorded on Bruker-AMX 400 ( 1 H: 400 MHz; 13 C: 100 MHz, Bruker, Karlsruhe, Germany; Billerica, MA, USA) and Bruker Aspect 3000 ( 1 H: 360 MHz, 13 C: 90 MHz, Bruker, Karlsruhe, Germany; Billerica, MA, USA) spectrometers using TMS and the solvent peak as internal standard. Chemical shifts were reported as δ in ppm and 3 JH,H coupling constants in Hz. Chiral HPLC separation of 10a and 11a were performed on a Jasco HPLC system with Chiralpak IC column (5 μm, 150 × 4.6 mm, hexan/propan-2-ol 1:1 eluent, 1 mLmin −1 flow rate, Daicel Chemical Industries Ltd., Tokyo, Japan) and HPLC-ECD spectra were recorded in stopped-flow mode on a JASCO J-810 electronic circular dichroism spectropolarimeter (JASCO Inc. Tokyo, Japan) equipped with a 10 mm HPLC flow cell. ECD ellipticity (ϕ) values were not corrected for concentration. For

Materials and Methods
Melting points were determined on a Kofler hot-stage apparatus (Wagner and Munz, Munich, Germany) and are uncorrected. The reactions were monitored by thin layer chromatography (TLC). three consecutive scans were recorded and averaged with 2 nm bandwidth, 1 s response, and standard sensitivity. The HPLC-ECD spectrum of the eluent recorded in the same way was used as background. The concentration of the injected sample was set so that the HT value did not exceed 500 V in the HT channel down to 230 nm. IR spectra were recorded on a JASCO FT/IR-4100 spectrometer (JASCO Inc. Tokyo, Japan) and absorption bands are presented in cm −1 . Electrospay Quadrupole Time-of-Flight HRMS measurements were performed with a MicroTOF-Q type QqTOF MS instrument equipped with an ESI source from Bruker (Bruker Daltoniks, Bremen, Germany).

Computational Section
Mixed torsional/low mode conformational searches were carried out by means of the Macromodel 9.7.211 software using Merck Molecular Force Field (MMFF) force field with implicit solvent model for chloroform applying a 21 kJ/mol energy window [47]. Geometry reoptimizations [B3LYP/6-31G(d) level in gas phase, B97D/TZVP and CAM-B3LYP/TZVP levels with PCM solvent model for n-octanol] and TDDFT-ECD calculations were performed with Gaussian 09 using various functionals (B3LYP, BH&HLYP, CAM-B3LYP, PBE0) and TZVP basis set for ECD calculations [48]. ECD spectra were generated as the sum of Gaussians with 1800-2400 cm −1 half-height width (corresponding to ca. 16-22 nm at 300 nm), using dipole-velocity computed rotational strength values [49]. Boltzmann distributions were estimated from the ZPVE corrected B3LYP energies in the gas phase calculations and from the uncorrected B97D and CAM-B3LYP energies in the PCM ones. The MOLEKEL software package was used for visualization of the results [50].

Bioassay on AChE Inhibitory Activity
The AChE assay was performed using the method of Ellman et al. with slight modification [51]. The production of the yellow anion of 5-thio-2-nitrobenzoic acid was measured with a microplate reader (DTX 880, Beckman Coulter, Brea, CA, USA) at 450 nm. The inhibition percentage caused by the presence of test compound was calculated, and the IC 50 was defined as the concentration of the compound that reduced 50% of the enzymatic activity without inhibitor. The bioassay on neuroprotective activity was carried out according to the reference 25.

Conclusions
N-aryl-1,4and 1,5-benzoaxazepine derivatives were prepared with different substitution patterns in short sequences and they were reacted with active methylene reagents to induce domino cyclization reactions. The N-aryl-1,4-benzoaxazepines reacted with a Knoevenagel- [1,5]-hydride shift-cyclization cascade providing condensed chiral O,N-heterocycles in regio-and diastereoselective manner. Enantiomers of two products were separated by chiral HPLC, HPLC-ECD specra were recorded and characteristic ECD transitions were correlated with the absolute configuration by TDDFT-ECD calculations. The regioisomeric 1,5-benzoaxazepine derivative reacted with 1,3-dimethylbarbituric acid in a domino Knoevenagel-C(sp 2 )-H functionalization reaction resulting in a condensed acridane derivative. Neuroprotective and AChE inhibitory activity of the products were tested. The condensed 1,4-benzoxazepine products did not have significant neuroprotective activity, which on the basis of our previous results, suggested that lack of the C-2 phenyl group or replacement by naphthyl groups resulted in the loss of activity. The condensed acridane derivative, containing a 1,5-benzoxazepine moiety, was found to have AChE inhibitory activity with 6.98 × 10 −6 mol/L IC 50 value.
Supplementary Materials: The following are available online. Figure S1-S74: 1 H and 13 C NMR and IR spectra of synthetic derivatives, Table S1: Organocatalytic transformations of 1a to trans-10a (columns A) and 1a to 11a (columns B), Figure S75: Concentration-dependent curve of 6 for the inhibition of AChE activity.