One-Pot Sequence of Staudinger/aza-Wittig/Castagnoli–Cushman Reactions Provides Facile Access to Novel Natural-like Polycyclic Ring Systems

Realization of the one-pot Staudinger/aza-Wittig/Castagnoli–Cushman reaction sequence for a series of azido aldehydes and homophthalic anhydrides is described. The reaction proceeded at room temperature and delivered novel polyheterocycles related to the natural product realm in high yields and high diastereoselectivity. The methodology has been extended to three other cyclic anhydrides. These further unravel the potential of the Castagnoli–Cushman reaction in generating polyheterocyclic molecular scaffolds.


Introduction
Molecular scaffolds are the foundation of a compound's biological activity [1]. This notion and its importance for drug discovery and the goal of generating drugs leads free from intellectual property liabilities [2] and fuels efforts in synthetic method development defined by Aktitopoulou-Zanze at Abbott as scaffold-oriented synthesis [3]. The major aim of scaffold-oriented synthesis is the deduction of synthetic pathways to novel ring systems, particularly polyheterocyclic ones, taking into account that bioactive compound [4] and natural product [5][6][7] chemical space is heavily populated with polyheterocycles.
Involvement of a cyclic starting material in a ring-forming process is a proven way to create a polycyclic framework [8]. Such processes are often associated with the creation of new, sometimes multiple, stereocenters, so relying on or discovering diastereoselective approaches to the new cycle formation is key. Recently, we developed a convenient synthetic protocol to transform azide-bearing benzaldehydes 1 into cyclic seven-membered imines 2 via the Staudinger reaction with triphenyl phosphine followed by an intramolecular aza-Wittig reaction [9]. Imines 2 were involved in the Staudinger β-lactam synthesis with ketenes generated from diazo compounds 3. As the result, tricyclic β-lactams 4 fused with tetrahydrothia-, oxa-and diazepine rings were obtained with high diastereoselectivity. Encouraged by this success, we pondered if the same cyclic imine substrates could be coupled to other ring-forming reactions. Our positive recent experience [10][11][12] with the imines (cyclic and acyclic) reacting with cyclic anhydrides via the formal [4+2] cycloaddition dubbed the Castagnoli-Cushman reaction (CCR) [13,14] prompted us to investigate the possibility of employing imines 2 as substrates for the CCR. As the starting point for this investigation, we considered reacting them with homophthalic anhydrides (HPA, 5), one of the most popular and reactive anhydrides that have been involved in the CCR [15][16][17]. This, we reasoned, would provide access to a tetracyclic 6-6-7-6 ring system 6 containing two new stereogenic centers. Considering that the CCR quite often proceeds diastereoselectively [18][19][20][21][22], the choice of the CCR as the means to create the new polyheterocyclic scaffold 6 from imines 2 is quite justified (Figure 1). Herein, we present the results obtained in the course of realizing the synthetic strategy outlined above.

Results and Discussion
O-linked aldehydes 1 required for the realization of the planned appro thesized from o-hydroxybenzaldehydes via alkylation with 2-azidoethyl m scribed elsewhere [9]. Eleven O- (2-azidoethyl) substrates 1a-k were obtain excellent yields. In addition to these, naphthalene-templated O-linked azid and azido ketone 8 were obtained, also in high yields, from the respective p ing materials. S-linked azido aldehyde 1l was prepared via nucleophilic ar tution of fluorine in o-fluorobenzaldehyde with 2-hydroxyethyl thiol follow ation of the hydroxyl group and nucleophilic displacement of mesylate w linked azido aldehyde 1m was prepared in a similar manner from o-fluoro using 2-azidoethyl methylamine (9) as the nucleophilic agent [9] (Scheme 1

Results and Discussion
O-linked aldehydes 1 required for the realization of the planned approach were synthesized from o-hydroxybenzaldehydes via alkylation with 2-azidoethyl mesylate as described elsewhere [9]. Eleven O-(2-azidoethyl) substrates 1a-k were obtained in good to excellent yields. In addition to these, naphthalene-templated O-linked azido aldehyde 7 and azido ketone 8 were obtained, also in high yields, from the respective phenolic starting materials. S-linked azido aldehyde 1l was prepared via nucleophilic aromatic substitution of fluorine in o-fluorobenzaldehyde with 2-hydroxyethyl thiol followed by mesylation of the hydroxyl group and nucleophilic displacement of mesylate with azide. N-linked azido aldehyde 1m was prepared in a similar manner from o-fluorobenzaldehyde using 2-azidoethyl methylamine (9) as the nucleophilic agent [9] (Scheme 1). 6 containing two new stereogenic centers. Considering that the CCR quite often proceeds diastereoselectively [18][19][20][21][22], the choice of the CCR as the means to create the new poly heterocyclic scaffold 6 from imines 2 is quite justified (Figure 1). Herein, we present the results obtained in the course of realizing the synthetic strategy outlined above.

Results and Discussion
O-linked aldehydes 1 required for the realization of the planned approach were syn thesized from o-hydroxybenzaldehydes via alkylation with 2-azidoethyl mesylate as de scribed elsewhere [9]. Eleven O-(2-azidoethyl) substrates 1a-k were obtained in good to excellent yields. In addition to these, naphthalene-templated O-linked azido aldehyde 7 and azido ketone 8 were obtained, also in high yields, from the respective phenolic start ing materials. S-linked azido aldehyde 1l was prepared via nucleophilic aromatic substi tution of fluorine in o-fluorobenzaldehyde with 2-hydroxyethyl thiol followed by mesyl ation of the hydroxyl group and nucleophilic displacement of mesylate with azide. N linked azido aldehyde 1m was prepared in a similar manner from o-fluorobenzaldehyde using 2-azidoethyl methylamine (9) as the nucleophilic agent [9] (Scheme 1). Scheme 1. Preparation of azido aldehyde and ketone substrates for the Staudinger/aza-Wittig cy clization. Considering that cyclic imines 2 can be generated on heating starting materials such as 1a-k or 7 in toluene and used, without the need for isolation, in the subsequent transformation, provided the latter can be conducted also in toluene (as was the case with the synthesis of b-lactams 4 [9]), we opted to attempt to couple the CCR of HPA to the preparation of 2 this way as well. The CCR of HPA can be conducted in a range of different solvents [16], depending on the solubility of the starting material (HPA itself not being the limiting factor). Fortunately, imines 2 do not precipitate from toluene even at room temperature, so after their formation and cooling the solution to ambient temperature, we added HPA (5) directly to the solution of imines 2 and continued the reaction with no heating. To our delight, in 2 h the starting materials (2 and HPA) were fully consumed and a thick precipitate of products 6 formed. The latter were filtered off, washed with ether and air-dried. NMR analysis of the products 6a-m thus isolated (in 67-95% yield) confirmed their identity and purity as greater than 95%, whereas they were not found in the filtrate. In the majority of instances (except for 6k), only one diastereomer was observed in the 1 H NMR spectra (see Supplementary Materials), thus confirming an excellent case of fully diastereoselective CCR (Scheme 2). Considering that cyclic imines 2 can be generated on heating starting materials such as 1a-k or 7 in toluene and used, without the need for isolation, in the subsequent transformation, provided the latter can be conducted also in toluene (as was the case with the synthesis of b-lactams 4 [9]), we opted to attempt to couple the CCR of HPA to the preparation of 2 this way as well. The CCR of HPA can be conducted in a range of different solvents [16], depending on the solubility of the starting material (HPA itself not being the limiting factor). Fortunately, imines 2 do not precipitate from toluene even at room temperature, so after their formation and cooling the solution to ambient temperature, we added HPA (5) directly to the solution of imines 2 and continued the reaction with no heating. To our delight, in 2 h the starting materials (2 and HPA) were fully consumed and a thick precipitate of products 6 formed. The latter were filtered off, washed with ether and air-dried. NMR analysis of the products 6a-m thus isolated (in 67-95% yield) confirmed their identity and purity as greater than 95%, whereas they were not found in the filtrate. In the majority of instances (except for 6k), only one diastereomer was observed in the 1 H NMR spectra (see Supplementary Materials), thus confirming an excellent case of fully diastereoselective CCR (Scheme 2). Scheme 2. The one-pot Staudinger/aza-Wittig generation of imines 2 followed by the CCR with HPA.
Scheme 2. The one-pot Staudinger/aza-Wittig generation of imines 2 followed by the CCR with HPA. The relative stereochemistry of compounds 6a-j was deemed to be cis based on the similarity of their NMR spectra and the single-crystal X-ray structures which were obtained for compounds 6a, 6c and 6d-f. The characteristic features of the 1 H NMR spectra of compounds 6a-j are the significant broadening of the signals corresponding to the ex-imine aromatic portion and the bis-methylene linker (presumably, due to the conformational behavior of the seven-membered ring) as well as the vicinal 3 J HH coupling constants of the signals coming from the methine protons at the asymmetric carbon atoms. In contrast, adducts 6k-m derived from aldehydes 1 where the position of the aromatic ring ortho to the carbonyl group was substituted displayed a different set of 1 H NMR characteristics (no broadening and different methine 3 J HH coupling constants) and were deemed to be trans-configured. Such a stereochemical outcome is not unexpected since the respective cis-configured products would be sterically congested. This stereochemistry assignment was further supported by the single-crystal X-ray structure obtained for compound 6m (see Supplementary Materials).
Unfortunately, involvement of imines 2 derived from azido aldehydes 1l-m and 8 in the reaction with HPA did not permit isolation of pure products 6, although it was successful in terms of starting material conversion. Likewise, isolation of the products from the reactions of 7-nitro-HPA was problematic. However, it was possible to isolate products of decarboxylation of these impure carboxylic acids. After performing the Staudinger/aza-Wittig sequence and reacting starting materials 1a, 1f, 1l-m or 8 with HPA (5), toluene was replaced with DMSO. Decarboxylation was performed at 100-150 • C in the presence of potassium carbonate. Tetracycles 10a-e were isolated in good to excellent yields over three steps. Notably, these compounds are hetero des-oxo analogs of natural product-related B-homoxylopinin (11), first reported in 1976 [23], whose congeners displayed histone acetyltransferase inhibitory activity [24] (Scheme 3). The relative stereochemistry of compounds 6a-j was deemed to be cis based on the similarity of their NMR spectra and the single-crystal X-ray structures which were obtained for compounds 6a, 6c and 6d-f. The characteristic features of the 1 H NMR spectra of compounds 6a-j are the significant broadening of the signals corresponding to the eximine aromatic portion and the bis-methylene linker (presumably, due to the conformational behavior of the seven-membered ring) as well as the vicinal 3 JHH coupling constants of the signals coming from the methine protons at the asymmetric carbon atoms. In contrast, adducts 6k-m derived from aldehydes 1 where the position of the aromatic ring ortho to the carbonyl group was substituted displayed a different set of 1 H NMR characteristics (no broadening and different methine 3 JHH coupling constants) and were deemed to be trans-configured. Such a stereochemical outcome is not unexpected since the respective cis-configured products would be sterically congested. This stereochemistry assignment was further supported by the single-crystal X-ray structure obtained for compound 6m (see Supplementary Materials).
Unfortunately, involvement of imines 2 derived from azido aldehydes 1l-m and 8 in the reaction with HPA did not permit isolation of pure products 6, although it was successful in terms of starting material conversion. Likewise, isolation of the products from the reactions of 7-nitro-HPA was problematic. However, it was possible to isolate products of decarboxylation of these impure carboxylic acids. After performing the Staudinger/aza-Wittig sequence and reacting starting materials 1a, 1f, 1l-m or 8 with HPA (5), toluene was replaced with DMSO. Decarboxylation was performed at 100-150 °C in the presence of potassium carbonate. Tetracycles 10a-e were isolated in good to excellent yields over three steps. Notably, these compounds are hetero des-oxo analogs of natural product-related B-homoxylopinin (11), first reported in 1976 [23], whose congeners displayed histone acetyltransferase inhibitory activity [24] (Scheme 3). Having successfully realized our synthetic plan with HPA, we became interested in extending the scope of this approach to other cyclic anhydrides. o-Phenylenediacetic acid anhydride (12) was first introduced as a reagent for the CCR in 2017, providing an entry into ε-lactams. It was shown to be less reactive compared to HPA [25]. In line with this observation, the reaction of 12 with cyclic imine 2 generated from azido aldehyde 1a required heating at 110 °C for 16 h to go to completion. The product of this transformation (13) containing a novel 6-7-7-6 polyheterocyclic ring system was isolated in a 52% yield Scheme 3. Staudinger/aza-Wittig/CCR sequence followed by decarboxylation.
Having successfully realized our synthetic plan with HPA, we became interested in extending the scope of this approach to other cyclic anhydrides. o-Phenylenediacetic acid anhydride (12) was first introduced as a reagent for the CCR in 2017, providing an entry into ε-lactams. It was shown to be less reactive compared to HPA [25]. In line with this observation, the reaction of 12 with cyclic imine 2 generated from azido aldehyde 1a required heating at 110 • C for 16 h to go to completion. The product of this transformation (13) containing a novel 6-7-7-6 polyheterocyclic ring system was isolated in a 52% yield by simple filtration as a single diastereomer, which was shown to be trans-configured by single-crystal X-ray analysis (see Supplementary Materials) (Scheme 4). CCR can be conveniently performed with the use of dicarboxylic acids in lieu of the respective cyclic anhydrides generated from diacids using dehydrating agents [26][27][28] or more straightforward cyclodehydration by means of azeotropic removal of water [11,29]. We chose to investigate the Staudinger/aza-Wittig/CCR reaction sequence for two dicarboxylic acids-thiodiglycolic (14) and p-methoxyphenyl-substituted glutaconic acid (15). After generation of imine 2 from azido aldehyde 1a, diacid 14 and acetic anhydride (serving as the dehydrating agent) were added and the process was further conducted at 110 °C for 24 h. The CCR product 16 was obtained in a respectable 66% yield, with no need for chromatographic purification, as a single diastereomer, which was shown to be transconfigured by single-crystal X-ray analysis (see Supplementary Materials). Glutaconic acid 15 was cyclodehydrated by means of azeotropic removal of water. The CCR in this case was conducted in refluxing toluene and, as expected [11,30], was accompanied by decarboxylation and transposition of the double bond. The respective tricyclic adduct 17 was isolated in a 51% yield (Scheme 5). Finally, intrigued by the opposite stereochemical outcome of the reaction sequence under investigation in the case of products 6a-j and 6k-m, we reasoned that the cis-isomers of 6a-j might signify the kinetic outcome of the CCR, while trans-configured products 6k-m obtained from more sterically demanding substrates 2 are likely to be thermodynamic adducts. If such an interpretation is correct, it should be possible to isomerize the kinetic cis-configured tetracycles 6 to their trans counterparts by thermal activation or otherwise. This was indeed demonstrated for compound 6a. On heating at 110 °C in dimethylformamide for 4 h, it was completely transformed to its trans-isomer 6a′, which was isolated in an 84% yield. The relative stereochemistry of 6a′ was confirmed by singlecrystal X-ray crystallography (Scheme 6). CCR can be conveniently performed with the use of dicarboxylic acids in lieu of the respective cyclic anhydrides generated from diacids using dehydrating agents [26][27][28] or more straightforward cyclodehydration by means of azeotropic removal of water [11,29]. We chose to investigate the Staudinger/aza-Wittig/CCR reaction sequence for two dicarboxylic acids-thiodiglycolic (14) and p-methoxyphenyl-substituted glutaconic acid (15). After generation of imine 2 from azido aldehyde 1a, diacid 14 and acetic anhydride (serving as the dehydrating agent) were added and the process was further conducted at 110 • C for 24 h. The CCR product 16 was obtained in a respectable 66% yield, with no need for chromatographic purification, as a single diastereomer, which was shown to be transconfigured by single-crystal X-ray analysis (see Supplementary Materials). Glutaconic acid 15 was cyclodehydrated by means of azeotropic removal of water. The CCR in this case was conducted in refluxing toluene and, as expected [11,30], was accompanied by decarboxylation and transposition of the double bond. The respective tricyclic adduct 17 was isolated in a 51% yield (Scheme 5). CCR can be conveniently performed with the use of dicarboxylic acids in lieu of the respective cyclic anhydrides generated from diacids using dehydrating agents [26][27][28] or more straightforward cyclodehydration by means of azeotropic removal of water [11,29]. We chose to investigate the Staudinger/aza-Wittig/CCR reaction sequence for two dicarboxylic acids-thiodiglycolic (14) and p-methoxyphenyl-substituted glutaconic acid (15). After generation of imine 2 from azido aldehyde 1a, diacid 14 and acetic anhydride (serving as the dehydrating agent) were added and the process was further conducted at 110 °C for 24 h. The CCR product 16 was obtained in a respectable 66% yield, with no need for chromatographic purification, as a single diastereomer, which was shown to be transconfigured by single-crystal X-ray analysis (see Supplementary Materials). Glutaconic acid 15 was cyclodehydrated by means of azeotropic removal of water. The CCR in this case was conducted in refluxing toluene and, as expected [11,30], was accompanied by decarboxylation and transposition of the double bond. The respective tricyclic adduct 17 was isolated in a 51% yield (Scheme 5). Finally, intrigued by the opposite stereochemical outcome of the reaction sequence under investigation in the case of products 6a-j and 6k-m, we reasoned that the cis-isomers of 6a-j might signify the kinetic outcome of the CCR, while trans-configured products 6k-m obtained from more sterically demanding substrates 2 are likely to be thermodynamic adducts. If such an interpretation is correct, it should be possible to isomerize the kinetic cis-configured tetracycles 6 to their trans counterparts by thermal activation or otherwise. This was indeed demonstrated for compound 6a. On heating at 110 °C in dimethylformamide for 4 h, it was completely transformed to its trans-isomer 6a′, which Finally, intrigued by the opposite stereochemical outcome of the reaction sequence under investigation in the case of products 6a-j and 6k-m, we reasoned that the cis-isomers of 6a-j might signify the kinetic outcome of the CCR, while trans-configured products 6km obtained from more sterically demanding substrates 2 are likely to be thermodynamic adducts. If such an interpretation is correct, it should be possible to isomerize the kinetic cisconfigured tetracycles 6 to their trans counterparts by thermal activation or otherwise. This was indeed demonstrated for compound 6a. On heating at 110 • C in dimethylformamide for 4 h, it was completely transformed to its trans-isomer 6a , which was isolated in an 84% yield. The relative stereochemistry of 6a was confirmed by single-crystal X-ray crystallography (Scheme 6).

Conclusions
We have described the realization of the one-pot Staudinger/aza-Wittig/Castagnoli-Cushman reaction sequence for a series of azido aldehydes and homophthalic anhydrides The reaction proceeded at room temperature and delivered novel polyheterocycles related to the natural product realm in high yields and high diastereoselectivity. Certain substrates presented difficulty in terms of isolation of the resulting tetracyclic products, which can be overcome by thermal decarboxylation of the carboxylic acid Castagnoli-Cushman adduct in the presence of potassium carbonate. The methodology has been extended to three other cyclic anhydrides, two of which were generated from the respective dicarboxylic acids in situ. The cis-configured Castagnoli-Cushman adducts are kinetic products and can be thermally isomerized to their trans counterparts. Collectively, these findings substantially expand the scope of the Castagnoli-Cushman reaction in terms of the imine source and the molecular scaffold accessible via applying this ring-forming reaction to cyclic imines.

General Information
All commercial reagents were used without purification. NMR spectra were recorded using a Bruker Avance III spectrometer in CDCl3 ( 1 H: 400.13 MHz; 13 С: 100.61 MHz; 19 F 376.50 MHz); chemical shifts are reported as parts per million (δ, ppm); the residual solvent peak (CHCl3 or DMSO-d6) was used as the internal standard: 7.28 or 2.51 for 1 H and 77.07 or 40.00 ppm for 13 C; multiplicities are abbreviated as follows: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, br = broad, dd = doublet of doublets, dt = doublet of triplets, ddd = doublet/doublets of doublets; coupling constants, J, are reported in Hz. Mass spectra were recorded using a Bruker microTOF spectrometer (ionization by electrospray, positive ions detection). Melting points were determined in open capillary tubes on a Stuart SMP50 Automatic Melting Point Apparatus. Analytical thin-layer chromatography was carried out on UV-254 silica gel plates using appropriate eluents. Compounds were visualized with short-wavelength UV light. Column chromatography was performed using silica gel Merk grade 60 (0.040−0.063 mm) 230−400 mesh. All reactions were conducted in the atmosphere of argon.

Conclusions
We have described the realization of the one-pot Staudinger/aza-Wittig/Castagnoli-Cushman reaction sequence for a series of azido aldehydes and homophthalic anhydrides. The reaction proceeded at room temperature and delivered novel polyheterocycles related to the natural product realm in high yields and high diastereoselectivity. Certain substrates presented difficulty in terms of isolation of the resulting tetracyclic products, which can be overcome by thermal decarboxylation of the carboxylic acid Castagnoli-Cushman adduct in the presence of potassium carbonate. The methodology has been extended to three other cyclic anhydrides, two of which were generated from the respective dicarboxylic acids in situ. The cis-configured Castagnoli-Cushman adducts are kinetic products and can be thermally isomerized to their trans counterparts. Collectively, these findings substantially expand the scope of the Castagnoli-Cushman reaction in terms of the imine source and the molecular scaffold accessible via applying this ring-forming reaction to cyclic imines.

General Information
All commercial reagents were used without purification. NMR spectra were recorded using a Bruker Avance III spectrometer in CDCl 3 ( 1 H: 400.13 MHz; 13 C: 100.61 MHz; 19 F: 376.50 MHz); chemical shifts are reported as parts per million (δ, ppm); the residual solvent peak (CHCl 3 or DMSO-d 6 ) was used as the internal standard: 7.28 or 2.51 for 1 H and 77.07 or 40.00 ppm for 13 C; multiplicities are abbreviated as follows: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, br = broad, dd = doublet of doublets, dt = doublet of triplets, ddd = doublet/doublets of doublets; coupling constants, J, are reported in Hz. Mass spectra were recorded using a Bruker microTOF spectrometer (ionization by electrospray, positive ions detection). Melting points were determined in open capillary tubes on a Stuart SMP50 Automatic Melting Point Apparatus. Analytical thin-layer chromatography was carried out on UV-254 silica gel plates using appropriate eluents. Compounds were visualized with short-wavelength UV light. Column chromatography was performed using silica gel Merk grade 60 (0.040−0.063 mm) 230−400 mesh. All reactions were conducted in the atmosphere of argon.
Step 2. 2-((2-Formylphenyl)thio)ethyl methanesulfonate was obtained according to a slightly modified mesilation literature procedure [35]. Mesyl chloride (214.2 mg, 1.87 mmol, 1.1 eq) was added to a solution of 2-((2-hydroxyethyl)thio)benzaldehyde and triethylamine (223.7 mg, 2.21 mmol, 1.3 equiv.) in DCM (6 mL) at 0 • C. One volume of saturated NaHCO 3 solution was added and the mixture was stirred for 30 min. The organic layer was separated and the aqueous layer was extracted with DCM. Combined organic layers were washed with brine, dried over anhydrous Na 2 SO 4 and the solvent was removed under reduced pressure to give 2-((2-formylphenyl)thio)ethyl methanesulfonate (237 mg, 91%) as a crude product that was used in the next stage directly without purification.