Modular Entry to Functionalized Tetrahydrobenzo[b]azepines via the Palladium/Norbornene Cooperative Catalysis Enabled by a C7-Modified Norbornene

Abstract

Tetrahydrobenzo[b]azepines (THBAs) are commonly found in many bioactive compounds; however, modular preparation of functionalized THBAs remains challenging to date. Here, we report a straightforward method to synthesize THBAs directly from simple aryl iodides via the palladium/norbornene (Pd/NBE) cooperative catalysis. Capitalizing on an olefin-tethered electrophilic amine reagent, an ortho amination, followed by 7-exo-trig Heck cyclization, furnishes the seven-membered heterocycle. To overcome the difficulty with ortho-unsubstituted aryl iodide substrates, a unique C7-bromo-substitueted NBE (N1) was discovered to offer the desired reactivity and selectivity. Besides THBAs, synthesis of other benzo-seven-membered ring compounds can also be promoted by N1. A combined experimental and computational studies show that the C7-bromo group in N1 plays an important and versatile role in this catalysis, including promoting β-carbon elimination, suppressing benzocyclobutene formation, and stabilizing reaction intermediates. The mechanistic insights gained could guide future catalyst design. The synthetic utility has been demonstrated in a streamlined synthesis of tolvaptan and forming diverse pharmaceutically relevant THBA derivatives. Finally, a complementary and general catalytic condition to access C6-substituted THBAs from ortho-substituted aryl iodides has also been developed.

Graphical Abstract

INTRODUCTION

Seven-membered nitrogen-containing heterocycles represent an important class of pharmacophores and exhibit a broad spectrum of biological activities.¹ In particular, the tetrahydrobenzo[b]azepine (THBA) ring system has been found in a number of marketed drugs, such as tolvaptan² and benazepril,³ and other bioactive compounds⁴ (Figure 1). However, comparing to their six- or five-membered counterparts—tetrahydroquinolines and indolines, synthesis of functionalized THBAs still represents a substantial challenge, owing to limited approaches to access medium-sized heterocycles in an efficient and modular manner.

Figure 1.

Examples of tetrahydrobenzo[b]azepine (THBA)-based bioactive compounds.

Conventional methods of preparing THBAs, including annulation,⁵ ring expansion,⁶ and aryne strategies,⁷ usually take a multi-step sequence and/or with limited scopes (Scheme 1, top). In addition, vicinal difunctionalized arenes are needed with these methods. For comparison, the use of intramolecular C–H functionalization to construct the C–C bond in THBAs, via either Friedel–Crafts⁸ or transition-metal catalysis^9,10, avoids complex arene precursors and simplifies the synthesis (Scheme 1, middle). While they clearly represent more attractive strategies to access THBAs, the availability of the corresponding elaborated aniline substrates cannot be ignored.

Scheme 1.

Synthetic strategies towards THBAs. PG: protecting group; FG: functional group.

Recently, the palladium/norbornene (Pd/NBE) cooperative catalysis, originally discovered by Catellani,¹¹ has emerged as a useful tool for modular arene synthesis.¹² Through forming a unique aryl-norbornyl-palladacycle (ANP) intermediate, simultaneous functionalizations of ortho and ipso positions of aryl halides can take place through selective coupling with an electrophile and a nucleophile, respectively. In particular, Lautens has pioneered in demonstrating fused-ring construction using electrophile/nucleophile-tethered reagents.¹³ Complementary to the C–C bond-forming C–H functionalization approaches, the Pd/NBE-catalyzed tandem ortho C–H amination¹⁴ followed by an intramolecular ipso Heck-cyclization¹⁵ can provide a straightforward and modular strategy to access diverse THBAs (Scheme 1, bottom).

While this direct annulation approach to construct THBAs may seem to be straightforward, a number of key issues remain to be addressed (Scheme 2). First, almost all the Pd/NBE-catalyzed annulations require the presence of an ortho substituent on the aryl iodide substrate¹⁶ in order to promote the desired NBE extrusion. Without ortho substituents, di-ortho amination and formation of NBE-attached side-products can dominate^15c, leading to low yield and low selectivity. Thus, this so-called “ortho constraint”¹⁷ represents a major limitation to this method as such ortho substituents are unnecessary in most bioactive THBAs (e.g. in Figure 1). Second, most ortho aminations employ cyclic amine electrophiles;^14,18 in contrast, acyclic ones (i.e., those needed for THBA synthesis) are relatively bulkier and generally less reactive due to the steric requirement when reacting with ANP.¹⁹ The slow reaction between ANP and the electrophile can lead to competing ANP reductive elimination to generate the benzocyclobutene side-product. Third, comparing to 5- or 6-membered ring formation, it is uncommon to use the tethered electrophile/nucleophile approach to construct medium-sized rings via the Pd/NBE catalysis.²⁰ This issue could be more prominent for the ortho-unsubstituted substrates as the enhanced barrier of the NBE exclusion step can further hinder the kinetically less favorable ipso-cyclization. As NBE is involved in all the key intermediates, it is therefore expected to play a critical role in overcoming the above difficulties. Thus, the “ideal NBE” cocatalyst in this scenario needs to efficiently promote the β-carbon elimination to address the “ortho constraint”, inhibit ANP reductive elimination to avoid the benzocyclobutene side-product, and not hinder the reaction between ANP and the bulky amine electrophile. To the best of our knowledge, a NBE co-catalyst exhibiting all these features has not been reported. In this article, we describe a full story of discovering a unique C7-bromo-substituted NBE (N1) that effectively address the aforementioned three challenges to realize general THBA synthesis. The unique and versatile roles of the bromo substituent of N1 in this catalysis have also been disclosed in this study.

Scheme 2.

Challenges of synthesizing THBAs from ortho-unsubstituted aryl iodides.

RESULTS AND DISCUSSIONS

To explore this strategy, 1-chloro-3-iodobenzene (1a) was chosen as the initial substrate due to its relevance to the synthesis of drug tolvaptan (vide infra, Scheme 3). O-benzoyl hydroxylamine 2a containing a terminal alkene was prepared in two steps from commercially available chemicals and employed as the initial coupling reagent; para-methoxylbenzyl (PMB) was used as the nitrogen protecting group due to ease of its removal. Not surprisingly, the initial trials with simple NBE (N2) only afforded a complex mixture with NBE-containing side products (see Supporting Information). The C1-substituted NBE, known to enable mono ortho functionalization of ortho-unsubstituted aryl halides with unhindered electrophiles,¹⁷ was found to be ineffective with forming a large amount of the benzocyclobutene side-product (for detailed discussions, see study of the NBE effect and Scheme 5). After an extensive screening of various structurally modified NBEs,²¹ to our surprise, 7-bromosubstituted NBE (N1) proved to be optimal and efficient for this challenging THBA synthesis (Table 1, entry 1). For comparison, NBEs with substitutions at other positions (N3-N9) all gave low yield and low selectivity (vide infra) in this reaction (entry 2). Notably, C7-substituted NBEs have not been used as the optimal NBE in any previous Catellani-type reactions, likely due to the concern of the C7 sterics on the NBE exo-face where the Pd migratory insertion normally takes place. The bromine group also appears to be important, as replacing Br with an alkoxy group (N10), an iodo group (N11), or a chloro group (N12) led to significant drop of the yield (vide infra). The anti-7-bromonorbornene (N13) also gave a much inferior result, which indicated that inductive effect of the Br may not be responsible for the high efficiency of N1. The less sterically hindered and more π-acidic phosphafluorene (L1)²² was found to be the best ligand. A survey of the phosphine effect¹³ (entry 3 and Supporting Information) shows that, in general, electron-deficient ligands worked better than electron-rich ones; for example, P(4-CF₃C₆H₄)₃ (L4) gave higher yield than P(4-OMeC₆H₄)₃ (L3), which was the previously optimal ligand for ortho amination (entry 3).^14a As expected, the palladium catalyst, phosphine ligand, and NBE were all essential for this transformation (entries 4-6). Similar to our previous findings, Pd(OAc)₂ is a worse pre-catalyst in the context of “ortho constraint” than Pd(cod)Cl₂ (entry 7).¹⁷ This is presumably owing to the non-innocent role of acetate anion in facilitating undesired C–H activation, as the use of cesium acetate as the base almost shut down the reaction (entry 8). In addition, cesium carbonate proved to be a much better base than potassium carbonate (entry 9). Reducing the N1 NBE loading to 50 mol% and 25 mol% diminished the yield from 72% to 55% and 49%, respectively (entries 10 and 11), though turnover of N1 can be observed. It is worth mentioning that 71% of N1 can be recovered under the standard condition. The reaction is regioselective with the annulation only at the C3 and C4 positions (3a). The alkene-isomerized product (3a’) was formed as a minor product, which can be separated from the main product. The kinetic monitoring of the reaction (Figure 2) shows that 3a’ was generated at the beginning of the reaction and the ration between 3a and 3a’ did not change much throughout the reaction. It is likely that the lack of the ortho substituent allows the Pd-hydride re-insertion to generate a tertiary alkyl-Pd(II) species, which can undergo inward β-hydrogen elimination to give trisubstituted alkene 3a’; in sharp contrast, such alkene isomers were not observed with ortho-substituted aryl iodides (vide infra, Table 3), highlighting another challenge with ortho-unsubstituted substrates.

Scheme 3.

Synthetic applications. NMO: N-methylmorpholine-N-oxide

Scheme 5.

Isolation of side products with different NBEs

Table 1.

Control experiments

^[a]Reaction conditions: 1a (0.1 mmol), 2a (0.2 mmol), Pd(cod)Cl₂ (0.01 mmol), ligand (0.025 mmol), NBE (0.1 mmol), and Cs₂CO₃ (0.25 mmol) in 1,4-dioxane (1 mL), 100 °C, 12 h.

^[b]Yields (3a + 3a’) and ratios were determined by crude ¹H NMR analysis using pyrazine as the internal standard.

Figure 2.

Kinetic profiles for the model reaction and ratio of products over time

Table 3.

Substrate scope of the ortho-substituted aryl iodides^a

^aReaction conditions: 19 (0.1 mmol), 2 (0.2 mmol), Pd(OAc)₂ (0.01 mmol), P(2-furyl)₃ (0.025 mmol), N11 (0.1 mmol), CsOAc (0.25 mmol) in 1,4-dioxane (1 mL), 55 °C, 45 h;

^b70 °C, 18 h;

^c100 °C, 12 h;

With the optimized conditions in hand, the scope of the THBA formation with ortho-unsubstituted aryl iodides was then examined (Table 2). Both electron-deficient and -rich aryl iodides are competent coupling partners (3a–3l), though the electron-deficient aryl iodides gave slightly better yields. Besides meta-substituted aryl iodides, meta-, para-disubstituted ones also reacted smoothly (3m–3r).²³ The reaction conditions can tolerate a wide range of functional groups, including chloro (3a), bromo (3b), nitro (3c), trifluoromethyl (3d), ester (3e), nitrile (3f), terminal alkene (3k), and silyl ether (3l). Other alkene-tethered O-benzoyl hydroxylamines 2 were explored next. When substituted amination reagents were employed, 3-substituted THBAs (3s–3t) can still be afforded in good yields. Besides the PMB protecting group, simple benzyl can be introduced in even higher yield (3u). When using the styrene-tethered reagent, complete selectivity of a single alkene geometrical isomer (3v) was observed. It is likely that the conjugation and/or the increased sterics inhibited the alkene isomerization. Finally, ortho-unsubstituted aryl iodides derived from naphthalene (3w), fluorenone (3x), quinoline (3y), estrone (3z), and loratadine (3aa) all delivered the desired THBAs in moderate to good yields. These polycyclic ring-fused azepines with diverse functional groups could be challenging to access via conventional approaches.

Table 2.

Substrate scope with ortho-unsubstituted aryl iodides^a

^aReaction conditions: 1 (0.1 mmol), 2 (0.2 mmol), Pd(cod)Cl₂ (0.01 mmol), L1 (0.025 mmol), N1 (0.1 mmol), and Cs₂CO₃ (0.25 mmol) in 1,4-dioxane (1 mL), 100 °C, 12 h. All yields are combined isolated yields of the products. The ratios were determined by ¹H NMR analysis of crude samples using pyrazine as the internal standard.

Synthetic applications.

The utility of this method was then evaluated. Tolvaptan, an aquaretic drug that functions as an orally active vasopressin receptor 2 antagonist, has been used to treat hyponatremia and autosomal dominant polycystic kidney disease. The ketone intermediate 5 was previously prepared in 7 steps;² for comparison, it can now be accessed in two steps from THBA 3a (Scheme 3A). Note that the PMB protecting group can be removed in high yield by treatment with TFA. Due to the modularity of the synthesis, this approach could be beneficial for preparing diverse tolvaptan analogues from readily available substituted aryl iodides. In addition, the chloro group on THBA 3a can serve as a handle to conveniently introduce other functional groups, e.g. an aryl group (6a), via cross couplings (Scheme 3B); it can also be deleted through reduction (6b).²⁴ Moreover, the alkene moiety in THBA products can undergo dihydroxylation, hydroboration–oxidation, and Simmons–Smith reaction to install vicinal diol (6c), primary alcohol (6d), as well as cyclopropane (6e), respectively. Furthermore, under the Mukaiyama hydration conditions, the alkene was converted into a tertiary alcohol (6f) and an interesting oxa-bridged benzazepine (6f’). It is noteworthy that these derivatizations could find their relevance in a number of THBA-containing bioactive compounds,²⁵ implying the versatile role of the alkene moiety for potential medicinal chemistry applications. Finally, under acidic conditions, the mixture of exo- and endo-cyclic products could be isomerized into the thermodynamically more stable internal olefin in nearly quantitative yield (Scheme 3C).

To explore generality of N1 in promoting seven-membered-ring formation with ortho-unsubstituted iodoarenes, two other ortho functionalization tactics were examined (Scheme 4). To our delight, besides ortho amination, preliminary success has been achieved with Catellani annulations via ortho alkylation¹³ and ortho aminocarbonylation²⁶ to give benzoannulene and benzoazepin-1-one products, respectively. In contrast, unsubstituted NBE (N2) gave much inferior results along with forming a significant amount of NBE-containing side-products, further confirming that the C7-bromo group is critical to address the “ortho constraint” and to inhibit undesired pathways.

Scheme 4.

Extension of the reaction scope to other types of ortho functionalization

X-ray crystal structure of complex 11.

The unique property of the C7-bromo NBE (N1) prompted us to investigate its exact role in the catalysis. Migratory insertion with NBEs typically occurs in a cis, exo manner due to the predistortion of the NBE π bond.^12a,27 However, one cannot assume, a priori, that N1 should maintain the exo selectivity because of the potential steric repulsion between the C7 substituent with the Pd during the migratory insertion process. To confirm the alkene facial selectivity of the migratory insertion step, pure palladium complex 11 was synthesized in good yield by reacting 3-iodoanisole with N1 and stoichiometric Pd(PPh₃)₄ in anhydrous THF at 90 °C (eq 1). Crystals suitable for X-ray analysis were obtained from dissolving 11 in CH₂Cl₂ followed by the addition of MeOH.

(1)

The X-ray structure of complex 11 clearly shows that the aryl ring and the palladium still remain positioned cis and exo to the norbornyl group (Figure 3, left). Interestingly, the d⁸ Pd center possesses a square-planar geometry with PPh₃, iodide, norbornyl group, and the bromo-substituent. This is in contrast with the literature reported analogue (complex 12)²⁸ with simple NBE, where the η² coordination with the C(5)-C(6) π bond occupies one of the four coordination sites (Figure 3, right). Other bond lengths, such as Pd(1)–I(2), Pd(1)–C(3), and Pd(1)–P(4) bonds, remain largely unchanged. Such a weakly coordinating bromo substituent can be considered as a “build-in” L-type ligand into NBE, which may have some implications in catalysis (vide infra).

Figure 3.

Structures of complex 11 and 12 (reference 28) as determined by X-ray crystallography.

Study of the NBE effect.

To understand the specific role of the C7-bromo substituent, it would be important to first recognize the reactivity differences between different NBEs. As such, side products of the model reactions with four representative NBEs have been carefully characterized and analyzed (Scheme 5). In contrast to the result with N1, benzocyclobutene 15 was the major isolated side-product when simple NBE N2 was employed. Comparing to the standard ortho amination with cyclic amine electrophiles,^14a it is likely that the bulkiness of electrophile 2a or the coordination of the alkene moiety hindered amination with the ANP intermediate. In addition, a minor side-product 16 containing the norbornyl group at the ipso position was also isolated, and such a C–H reductive elimination product was not observed when N1 was used.²⁹ These results indicated that the bromo substituent in N1 should also promote β-carbon elimination or prevent C–H reductive elimination. As a comparison, while the sterics of the C2-substituted NBE (N6)³⁰ can effectively inhibit the benzocyclobutene formation³¹, it led to NBE-attached side-product 17 and bis ortho-amination side-product 14. The formation of 17 is possibly due to the lability of the Pd-enolate intermediate that can undergo protonation or β-hydrogen elimination via the “face switch”.³² The formation of 14 indicates that N6 cannot efficiently suppress the “ortho constraint”. On the other hand, with the C1-substituted NBE (N3), no ortho di-functionalized products (like 14) were observed, showing that N3 is indeed effective to minimize the second ortho C–H activation. However, direct reductive elimination of ANP became the major side-reaction, resulting in undesired benzocyclobutene 18; comparing to N1, the increased sterics in the N3-derived ANP intermediate renders it less reactive with the bulky electrophile. With N1 as the co-catalyst, side-products (13, 14) were only observed with <5% yields; therefore, the benefits of using the C7-bromo NBE (N1) include suppressing ANP reductive elimination, promoting NBE exclusion and/or inhibiting the second ortho C–H activation without significantly interfering with the reaction between the ANP and the electrophile.

To further elucidate the detailed roles of the bromo moiety of N1 in this reaction, DFT calculation was conducted. First, the effect of N1 on the β-carbon elimination and subsequent 7-exo-trig processes was examined (Figure 4). Since the 7-exo-trig has a slightly higher barrier than the β-carbon elimination in this reaction, NBE extrusion can be regarded as a pre-equilibrium. The steric repulsion between the bromo substituent and the ortho C–H bond (H∙∙∙Br distance is 2.66 Å) is expected to account for the faster 7-exo-trig (22.6 with N1 vs. 25.3 kcal/mol with N2) by driving the NBE extrusion pre-equilibrium. Another feature of the bromo-substituted NBE is the weak coordination between the Pd and the Br, which was also observed in the X-ray structure of complex 11. Such weak coordination could stabilize intermediate INT1 and prevent competing side-reactions (e.g. C–H reductive elimination, protonation or β-H elimination). On the other hand, even from a relatively stable intermediate (INT1), the β-carbon elimination with N1 still takes place with a relatively low kinetic barrier. Analysis of the transition state TS1 (with N1) revealed that a weak interaction between the Pd and the Br still exists (Pd∙∙∙Br distance is 3.34 Å) during the β-carbon elimination transition state. Thus, the kinetic barrier of the β-carbon elimination remains very similar to simple NBE N2 (21.7 vs 20.6 kcal/mol). As a result, the DFT study implies that the overall NBE extrusion and subsequent Heck-cyclization process is more favorable with N1.

Figure 4.

Comparison of the β-carbon elimination and subsequent 7-exo-trig between NBEs N1 and N2. All energies are in kcal/mol. Calculations were performed at the B3LYP-D3/SDD-6-311+G(d,p)-SMD(1,4-dioxane)// B3LYP/LANL2DZ-6-31G(d) level of theory.

Meanwhile, the reactivity differences between N1 and N2-derived ANPs (INT3) have also been studied computationally (Figure 5). The benzocyclobutene side-products (e.g., 15 and 18) have been previously found to come from reductive elimination of the anionic ANP intermediate (INT3).¹⁹ Through the DFT calculation, the presence of the electronegative bromo substituent in N1-derived ANP was found to disfavor the reductive elimination by 4.3 kcal/mol relative to the one derived from N2. The increased kinetic barrier was presumably due to the repulsive interaction between the negatively charged Br atom with the arene π electrons in TS3. This is evidenced by the shortest C···Br bond distance of 3.40 Å (the sum of their Van der Waals radii is 3.55 Å) in the required transition state. On the other hand, INT3 derived from N1 has a very similar oxidative addition barrier with 2a to that of N2, indicating its reactivity with electrophiles is not affected by the bromo substituent. This could be explained by the fact that the bromo substituent does not have significant steric repulsion with the electrophile during the transition state TS4 (N1) (the shortest H···Br distance is over 4 Å). Thus, the DFT study also supports that NBE N1 can inhibit the ANP reductive elimination without compromising the reaction with the electrophile.

Figure 5.

Comparison of reductive elimination from INT3 between NBE N1 and N2. All energies are in kcal/mol.

As a summary of the NBE effect, the C7-bromo-substituted N1 indeed exhibits a range of unique and versatile features. First, the bromo atom can serve as a build-in L ligand to stabilize reaction intermediates to minimize side-reactions. In addition, it can promote NBE extrusion through stabilizing the β-carbon elimination transition state, therefore addressing the “ortho constraint” challenge. Moreover, the electronegative Br substituent can suppress ANP reductive elimination due to charge repulsion with the metalated arene. Finally, the lack of steric bulkiness of the Br substituent retains good reactivity of the ANP with the electrophile. Altogether, the C7-bromo-substituted N1 is well suited for the challenging medium-sized-ring formation with ortho-unsubstituted aryl iodides.

Scope of the ortho-substituted aryl iodides.

Finally, to complete the method development for the Pd/NBE-catalyzed THBA synthesis, the scope of using ortho-substituted aryl iodides was also investigated. Aided by the ortho substituents, this reaction can now be achieved in good yields by several NBEs (see Supporting Information for details) with the C2-substituted³⁰ NBE N11 being optimal.³³ In addition, without the challenge of “ortho constraint”, the use of acetate salts (e.g. CsOAc and Pd(OAc)₂) was beneficial, possibly through promoting the concerted metalation deprotonation during the C–H activation step. The reaction can also proceed at a lower temperature. No obvious alkene isomerization was observed (vide supra). It is likely that the bulkiness of the ortho substituents inhibits the palladium-hydride reinsertion. Moreover, a number of different ortho substituents with various steric and electronic properties were tolerated. Similarly, the high chemoselectivity and broad functional group/linker compatibility make this method general to access diverse C6-substituted THBAs.

Conclusion.

In summary, a unique C7-bromo-substituted NBE N1 has been identified to be an “ideal” co-catalyst in promoting THBA and other seven-membered ring synthesis from ortho-unsubstituted aryl iodides. The broad substrate scope, the high functional group tolerance, and the synthetic versatility of the alkene functionality should make this method attractive for modular preparation of substituted THBAs. The knowledge gained on the distinct and versatile roles of the C7-bromo group in N1 could have implications on new catalyst design. Further investigations of N1 (and its analogues) in promoting other challenging Catellani-type transformations are ongoing.