Extending the Scope of the New Variant of the Castagnoli–Cushman Cyclocondensation onto o-Methyl Benzoic Acids Bearing Various Electron-Withdrawing Groups in the α-Position

Based on the previously reported involvement of homophthalic acid monoesters in the Castagnoli–Cushman reaction-type cyclocondensation with imines, we tested a number of other o-methyl benzoic acids bearing various electron-withdrawing groups in the α-position. The majority of these substrates delivered the expected tetrahydroisoquinolone adducts on activation with CDI or acetic anhydride. Homophthalic acid mononitriles displayed the highest promise as substrates for the new reaction, both in terms of scope and product yields. Homophthalic acid monoamides either gave low yields or failed to react with imines. Sulfonyl-substituted substrates gave the desired (and hitherto unknown) type of tetrahydroisoquinolines. Despite the low yields, this approach to sulfonyl-substituted tetrahydroisoquinolines appears practical as alternative syntheses based on the traditional, carboxylic acid CCR adducts would presumably be cumbersome and multistep. The azido- and nitro-substituted o-methyl benzoic acids failed to react with imines.


Introduction
The Castagnoli-Cushman reaction (CCR) [1][2][3][4][5][6] is a distinctly versatile [4+2]-type cyclocondensation of α-C-H-acidic cyclic anhydrides 1 with imines 2 leading, depending on the specific anhydride employed [7], to skeletally diverse [8] polysubstituted lactams 3, often in diastereoselective fashion. Although in some instances the cyclic anhydride can be generated from the respective dicarboxylic acid in situ [9], hydrolytically prone cyclic anhydrides remain the traditional reagent input for the CCR, which hinders the use of this reaction for large combinatorial arrays. Moreover, in some instances the carboxylic acid resulting from the initial CCR has a tendency to decarboxylate [10] and/or requires in situ esterification before it can be conveniently isolated by chromatography. With this obstacle in mind, we developed a novel variant of the CCR, which relies upon monoesters of homophthalic acid 4 (whose anhydride is one of the most popular reagents for the CCR [7]) being activated towards the condensation with imines 2 by either CDI (1,1'carbonyldiimidazole) or acetic anhydride [11]. Not only did this variant of the CCR deliver easily isolable esters 5 directly and alleviate the need to use hydrolytically prone homophthalic anhydrides, it also opened a potential prospect of extending this reactivity to other o-methyl benzoic acids bearing various electron-withdrawing groups in the αposition (6). Besides the fundamental interest towards realizing such a novel reactivity of readily available substrates in practice, the opportunity to obtain tetrahydroisoquinolones 7 bearing other groups, in lieu of the carboxylic acid, was viewed as a significant advantage from the medicinal chemistry perspective. Indeed, varying the electron-withdrawing group in 7, depending on the substrate 6, could afford a direct access to various carboxylic acid congeners or replacements with no need for multistep reaction sequences ( Figure 1). available substrates in practice, the opportunity to obtain tetrahydroisoquinolones 7 bearing other groups, in lieu of the carboxylic acid, was viewed as a significant advantage from the medicinal chemistry perspective. Indeed, varying the electron-withdrawing group in 7, depending on the substrate 6, could afford a direct access to various carboxylic acid congeners or replacements with no need for multistep reaction sequences ( Figure 1). Considering the medicinal chemistry significance of the tetrahydroisoquinolone (THIQ) scaffold as the core of adrenocorticotropic hormone receptor modulator 8 [12], apoptosis regulator 9 [13], trypanocidal cysteine protease inhibitor 10 [14] as well as antimalarial 11 [15] (to mention only a few examples, Figure 2), validating substrates 6 for the CCR clearly represents a worthy undertaking. Herein, we present the results obtained in the course of investigating the new reagent space for the recently discovered variant of the CCR.

Results and Discussion
Initially, we began to explore the reaction of homophthalic mononitriles 12a-c. Electron-donating and electron-withdrawing group substitution was expected to have a bearing on the CH2-acidity of these substrates and, therefore, on the facility of the CCR-type  available substrates in practice, the opportunity to obtain tetrahydroisoquinolones 7 bearing other groups, in lieu of the carboxylic acid, was viewed as a significant advantage from the medicinal chemistry perspective. Indeed, varying the electron-withdrawing group in 7, depending on the substrate 6, could afford a direct access to various carboxylic acid congeners or replacements with no need for multistep reaction sequences ( Figure 1). Considering the medicinal chemistry significance of the tetrahydroisoquinolone (THIQ) scaffold as the core of adrenocorticotropic hormone receptor modulator 8 [12], apoptosis regulator 9 [13], trypanocidal cysteine protease inhibitor 10 [14] as well as antimalarial 11 [15] (to mention only a few examples, Figure 2), validating substrates 6 for the CCR clearly represents a worthy undertaking. Herein, we present the results obtained in the course of investigating the new reagent space for the recently discovered variant of the CCR.

Results and Discussion
Initially, we began to explore the reaction of homophthalic mononitriles 12a-c. Electron-donating and electron-withdrawing group substitution was expected to have a bearing on the CH2-acidity of these substrates and, therefore, on the facility of the CCR-type

Results and Discussion
Initially, we began to explore the reaction of homophthalic mononitriles 12a-c. Electrondonating and electron-withdrawing group substitution was expected to have a bearing on the CH 2 -acidity of these substrates and, therefore, on the facility of the CCR-type cyclocondensation as understood within the suggested mechanism of the reaction (vide infra).
Unsubstituted homophthalic mononitrile 12a was prepared from methyl o-(bromomethyl)benzoate (13) as described in the literature [16]. The initial nucleophilic substitution product 14, without purification, was subjected to basic hydrolysis to give 12a in 62% yield over two steps. The nitro-substituted substrate 12b was prepared by the With the substrates 12a-c at hand, we proceeded to explore their modified CCR with various imines 2 using either CDI (1,1'-carbonyldiimidazole) or acetic anhydride (Ac 2 O) [11] as the carboxylic group activator (Scheme 2). To our delight, the reaction delivered the expected 4-cyanotetrahydroisoquinolones 18a-q in moderate yields and displayed either no diastereoselectivity (cf. 18b or 18q) or gave exclusively one diastereomer (cf. 18a or 18g). The relative trans-stereochemistry was assigned to the major diastereomer obtained in the majority of cases by correlation of their spectral properties (specifically, the 3 J coupling constants between the methine protons) with those of the trans-18b whose stereochemistry was unequivocally confirmed by the single-crystal X-ray analysis (CCDC2192610). This isomer of 18b (with X-ray data) revealed the signal of H-3 methine atom as doublet with J = 2.6 Hz (for details see ESI, Table S3), while the other (cis) isomer displayed the coupling constant of J = 5. 9 Hz. For all other products 18 the corresponding coupling constants were found to be in the range 1.7-2.7 Hz for major isomers and 5.6-6. 3 Hz for the minor. The reactivity of unsubstituted (12a) or methoxy-substituted (12c) substrates was markedly lower compared to their nitro-substituted (12b) counterpart. While in the former case, prolonged heating at 130°C with CDI as the activating agent was needed, in the latter case, only two hours at 60°C were sufficient to complete the reaction with Ac 2 O as the carboxylic acid activator.
Acetic anhydride was used to activate substrate 12b as with CDI, 4-cyanodihydroisoq uinolones 19a,b (Scheme 3) were exclusively obtained in moderate yield, presumably, via the oxidation of the respective α-carbanion (formed, in turn, from the initial tetrahydroisoquinolone adduct due to the basic character of imidazole generated from CDI). Interestingly, thorough exclusion of air from the reaction medium by argon purging did not suppress this process, which is likely due to the possibility of nitro compounds self-oxidizing [19]. Likewise, variation in such reaction parameters as temperature, reaction duration and concentration in compound 19a synthesis did not dramatically alter the outcome of the reaction with CDI. However, switching to acetic anhydride as the carboxylic acid activator allowed conditions to be quickly identified (60°C, 2 h) for the formation of this compound's unoxidized version, i.e., compound 18k, in a respectable 46% yield (see Table  S1 in Supplementary Materials for the reaction optimization towards 19a and 18k).
outcome of the reaction with CDI. However, switching to acetic anhydride as the carboxylic acid activator allowed conditions to be quickly identified (60 ℃, 2 h) for the formation of this compound's unoxidized version, i.e., compound 18k, in a respectable 46% yield (see Table S1 in Supplementary Materials for the reaction optimization towards 19a and 18k).  While compounds 18 present a novel substitution pattern around the tetrahydroisoquinolone scaffold accessible by the CCR-type cyclocondensation, these compounds can be manipulated further via a range of post-condensational modifications. For instance, we have shown that compound 18k could be intentionally oxidized to its dihydroisoquinolone counterpart 19a by heating at 150 °C in DMSO (dimethylsulfoxide). In the resulting compound 19a, the nitro group can be reduced by stannous chloride to give dihydroisoquinolone compound 20 with an amino group, i.e., an additional site for further modification. The nitrile functionality is a versatile precursor to various heterocyclic moieties including 1,3-oxazole [20] and tetrazole [21], the latter being a well-established bioisostere outcome of the reaction with CDI. However, switching to acetic anhydride as the carboxylic acid activator allowed conditions to be quickly identified (60 ℃, 2 h) for the formation of this compound's unoxidized version, i.e., compound 18k, in a respectable 46% yield (see Table S1 in Supplementary Materials for the reaction optimization towards 19a and 18k).  While compounds 18 present a novel substitution pattern around the tetrahydroisoquinolone scaffold accessible by the CCR-type cyclocondensation, these compounds can be manipulated further via a range of post-condensational modifications. For instance, we have shown that compound 18k could be intentionally oxidized to its dihydroisoquinolone counterpart 19a by heating at 150 °C in DMSO (dimethylsulfoxide). In the resulting compound 19a, the nitro group can be reduced by stannous chloride to give dihydroisoquinolone compound 20 with an amino group, i.e., an additional site for further modification. The nitrile functionality is a versatile precursor to various heterocyclic moieties including 1,3-oxazole [20] and tetrazole [21], the latter being a well-established bioisostere While compounds 18 present a novel substitution pattern around the tetrahydroisoquinolone scaffold accessible by the CCR-type cyclocondensation, these compounds can be manipulated further via a range of post-condensational modifications. For instance, we have shown that compound 18k could be intentionally oxidized to its dihydroisoquinolone counterpart 19a by heating at 150 • C in DMSO (dimethylsulfoxide). In the resulting compound 19a, the nitro group can be reduced by stannous chloride to give dihydroisoquinolone compound 20 with an amino group, i.e., an additional site for further modification. The nitrile functionality is a versatile precursor to various heterocyclic moieties including 1,3-oxazole [20] and tetrazole [21], the latter being a well-established bioisostere of the carboxylic acid functionality for drug design [22]. Thus, compound 18c (taken as a 3.3:1.0 diastereomeric mixture) underwent a formal [3+2] cycloaddition with sodium azide in the presence of ammonium chloride to give tetrazole 21 isolated chromatographically as a single trans-isomer. The configuration of the latter was confirmed by the single-crystal X-ray analysis (CCDC2203753) (Scheme 4). of the carboxylic acid functionality for drug design [22]. Thus, compound 18c (taken as a 3.3:1.0 diastereomeric mixture) underwent a formal [3 + 2] cycloaddition with sodium azide in the presence of ammonium chloride to give tetrazole 21 isolated chromatographically as a single trans-isomer. The configuration of the latter was confirmed by the singlecrystal X-ray analysis (CCDC2203753) (Scheme 4). While the traditional carboxylic acid Castagnoli-Cushman adducts obtainable from homophthalic anhydride can be amidated directly [23], we were curious to see if the amide functionality can be installed in lieu of the nitrile function in compounds 18 and serve as a sufficiently electron-withdrawing group to enable the new variant of the CCR-type cyclocondensation. Homophthalic monoamides 22a-d (obtainable from the respective homophthalic anhydrides and ammonia or anilines [24]) were tested in the reaction with one exemplary imine (2a), on activation with CDI (Scheme 5). To our delight, primary amide 22a and anilide 22b gave the expected cycloadducts 23a and 23b, respectively, in excellent diastereoselectivity (trans-diastereomers as judged from the 3 J coupling constants between the methine protons Table S3, ESI) albeit in low yield. Disappointingly, tertiary anilides 22c,d gave no reaction with the imine component. Instead, isochromenones 24a and 24b were isolated, also in mediocre yields. This result signifies the higher propensity of the less C-H-acidic and more O-nucleophilic anilides 22c,d to undergo the known [25] intramolecular cyclodehydration, rather than to interact, intermolecularly, with the imine component. Overall, it can be concluded that the involvement of monoamides 22 in the new type of the CCR discussed herein is possible but appears less practical compared to the same reaction of nitriles 12 or the amidation of the traditional carboxylic acid CCR products.  While the traditional carboxylic acid Castagnoli-Cushman adducts obtainable from homophthalic anhydride can be amidated directly [23], we were curious to see if the amide functionality can be installed in lieu of the nitrile function in compounds 18 and serve as a sufficiently electron-withdrawing group to enable the new variant of the CCR-type cyclocondensation. Homophthalic monoamides 22a-d (obtainable from the respective homophthalic anhydrides and ammonia or anilines [24]) were tested in the reaction with one exemplary imine (2a), on activation with CDI (Scheme 5). To our delight, primary amide 22a and anilide 22b gave the expected cycloadducts 23a and 23b, respectively, in excellent diastereoselectivity (trans-diastereomers as judged from the 3 J coupling constants between the methine protons Table S3, ESI) albeit in low yield. Disappointingly, tertiary anilides 22c,d gave no reaction with the imine component. Instead, isochromenones 24a and 24b were isolated, also in mediocre yields. This result signifies the higher propensity of the less C-H-acidic and more O-nucleophilic anilides 22c,d to undergo the known [25] intramolecular cyclodehydration, rather than to interact, intermolecularly, with the imine component. Overall, it can be concluded that the involvement of monoamides 22 in the new type of the CCR discussed herein is possible but appears less practical compared to the same reaction of nitriles 12 or the amidation of the traditional carboxylic acid CCR products.
of the carboxylic acid functionality for drug design [22]. Thus, compound 18c (taken as a 3.3:1.0 diastereomeric mixture) underwent a formal [3 + 2] cycloaddition with sodium azide in the presence of ammonium chloride to give tetrazole 21 isolated chromatographically as a single trans-isomer. The configuration of the latter was confirmed by the singlecrystal X-ray analysis (CCDC2203753) (Scheme 4). While the traditional carboxylic acid Castagnoli-Cushman adducts obtainable from homophthalic anhydride can be amidated directly [23], we were curious to see if the amide functionality can be installed in lieu of the nitrile function in compounds 18 and serve as a sufficiently electron-withdrawing group to enable the new variant of the CCR-type cyclocondensation. Homophthalic monoamides 22a-d (obtainable from the respective homophthalic anhydrides and ammonia or anilines [24]) were tested in the reaction with one exemplary imine (2a), on activation with CDI (Scheme 5). To our delight, primary amide 22a and anilide 22b gave the expected cycloadducts 23a and 23b, respectively, in excellent diastereoselectivity (trans-diastereomers as judged from the 3 J coupling constants between the methine protons Table S3, ESI) albeit in low yield. Disappointingly, tertiary anilides 22c,d gave no reaction with the imine component. Instead, isochromenones 24a and 24b were isolated, also in mediocre yields. This result signifies the higher propensity of the less C-H-acidic and more O-nucleophilic anilides 22c,d to undergo the known [25] intramolecular cyclodehydration, rather than to interact, intermolecularly, with the imine component. Overall, it can be concluded that the involvement of monoamides 22 in the new type of the CCR discussed herein is possible but appears less practical compared to the same reaction of nitriles 12 or the amidation of the traditional carboxylic acid CCR products. To our delight, sulfones 25 and 26 proved to be valid substrates for the new type of CCR under investigation (Scheme 6). The hitherto unknown CCR-derived sulfones 29 and 30 were obtained in low yield as single (trans) diastereomers as judged from the 3 J coupling constants between the methine protons (Scheme 7 and Table S3, ESI). However, despite the moderate yields of the sulfone adducts, in our view, this approach carries much practicality as synthesizing such Castagnoli-Cushman-type sulfones from the traditional carboxylic acid CCR adducts would presumably require a sequence of several  [27]) and nitro (28 [28]) compounds a potential substrates for the new type of the CCR.
To our delight, sulfones 25 and 26 proved to be valid substrates for the new type o CCR under investigation (Scheme 6). The hitherto unknown CCR-derived sulfones 29 and 30 were obtained in low yield as single (trans) diastereomers as judged from the 3 J cou pling constants between the methine protons (Scheme 7 and Table S3, ESI). However, de spite the moderate yields of the sulfone adducts, in our view, this approach carries much practicality as synthesizing such Castagnoli-Cushman-type sulfones from the traditiona carboxylic acid CCR adducts would presumably require a sequence of several steps, in cluding decarboxylation, which may deliver these compounds in an even lower yield, i any.  [17]), azide (27 [27]) and nitro (28 [28]) compounds as potential substrates for the new type of the CCR. Scheme 6. Synthesis of sulfone (25 [26]-26 [17]), azide (27 [27]) and nitro (28 [28]) compounds a potential substrates for the new type of the CCR.
To our delight, sulfones 25 and 26 proved to be valid substrates for the new type o CCR under investigation (Scheme 6). The hitherto unknown CCR-derived sulfones 29 and 30 were obtained in low yield as single (trans) diastereomers as judged from the 3 J cou pling constants between the methine protons (Scheme 7 and Table S3, ESI). However, de spite the moderate yields of the sulfone adducts, in our view, this approach carries much practicality as synthesizing such Castagnoli-Cushman-type sulfones from the traditiona carboxylic acid CCR adducts would presumably require a sequence of several steps, in cluding decarboxylation, which may deliver these compounds in an even lower yield, i any. Unfortunately, all attempts to react azidoalkyl (27) and nitroalkyl (28)   Unfortunately, all attempts to react azidoalkyl (27) and nitroalkyl (28)

2-(Cyanomethyl)-5-methoxybenzoic Acid (12c)
Prepared according to modified procedure from [18]. First, 6-Methoxy indenone oxime (116 mg, 0.61 mmol) was added to 10% aq. NaOH (5 mL) and the mixture was heated at 50 • C. Tosyl chloride (139 mg, 0.73 mmol) was added in portions and the reaction was heated at 80 • C for 15 min. Then it was allowed to cool down. The resulting precipitate was filtered off and the filtrate was acidified with HCl conc. to pH = 2, followed by extraction with EtOAc (3 × 10 mL). The combined organic extracts were dried over Na 2 SO 4 , filtered and concentrated under reduced pressure to give yellow solid.

Conclusions
Encouraged by our previous success involving homophthalic acid monoesters in the CCR-like cyclocondensation with imines, we tested a number of other o-methyl benzoic acids bearing various electron-withdrawing groups in the α-position. The majority of these substrates, being activated either by CDI or acetic anhydride, did deliver the expected tetrahydroisoquinolone adducts. The homophthalic acid mononitriles displayed a high promise as substrates for the new reaction, offering the largest scope, generally higher yields and modest to high diastereoselectivity. Homophthalic acid monoamides, such as primary amide or anilide, gave the Castagnoli-Cushman adducts, albeit in low yields; tertiary anilides failed to condense with the imine component. Sulfonyl-substituted substrates also gave the desired (and hitherto unknown) type of tetrahydroisoquinolines. Despite the low yields, this approach to sulfonyl-substituted tetrahydroisoquinolines is deemed practical as an alternative synthesis based on the traditional, carboxylic acid CCR adducts appears cumbersome and multistep. An azido-and nitro-substituted o-methyl benzoic acids failed to react with imines. Collectively, these results clearly expand the scope of the new type of Castagnoli-Cushman reaction, as well as set certain apparent limitations.