Iodine(III)-Mediated Oxidation of Anilines to Construct Dibenzazepines

Abstract

The development of an efficient process that produces bioactive medium-sized N-heterocyclic scaffolds from 2-substituted anilines using either iodosobenzene or (bis(trifluoroacetoxy)iodo)-benzene is reported. The tether between the sulfonamide and the aryl group can be varied to access dihydroacridine-, dibenzazepine-, or dibenzazocine scaffolds. While substitution on the aniline portion is limited to electron-neutral- or electron-poor groups, a broader range of functional groups are tolerated on the ortho-aryl substituent and site selective C─NAr bond formation can be achieved. Preliminary mechanistic investigations suggest that medium-ring formation occurs via radical reactive intermediates.

Graphical Abstract

A mild, room temperature oxidative cyclization of 2-substituted anilines that accesses a broad range of medium-ring N-heterocycles via radical intermediates is reported.

Introduction

Despite their established and important bioactivity,^[1] medium-sized N-heterocycles are underrepresented in pharmaceutical libraries.^[2] Dibenzazepines and partially saturated dibenzazepines are important scaffolds,^[3] which are found in the core of pharmaceutical agents, such as imipramine, eslicarbazepine acetate,^[3e] and carbamazepine (Scheme 1). While this bioactivity has spurred the development of methods to ease their synthesis,^[4],[5] the unfavorably high transition state energy barriers for medium-sized ring closure can limit the efficiency and selectivity of bond construction.^[6] These weaknesses are illustrated in the common approaches to this scaffold, which involve either creating a C─C bond through a cross-coupling reaction, which requires pre-existing functionality and can afford a mixture of N-heterocyclic products,^{[5c, 5d]} or a C─NAr bond through thermolysis of a symmetrical iminobibenzyl first reported by Thiele and Holzinger in 1899,^[7] and is still widely employed for the synthesis of analogs.^[8] This latter approach requires installation of functionality to occur post cyclization to differentiate the two sides of the dibenzazepine and can be non-selective.^[9] The development of iodine(III)-mediated electrophilic aromatic substitution by the Muñiz-, Antonchick-, and other groups has emerged recently for the construction of C─NAr bonds using heteroaryl- or aryl amines as the nitrogen source.^[10],[11] For example in 2011, Antonchick and co-workers reported a (diacetoxy)phenyliodine-mediated reaction between acetanilines (e.g. 9) and arenes that selectively afforded N,N-diaryl amines (e.g. 10).^[11a] In 2020, we reported that benzazepinones could be synthesized from the room temperature oxidation of anilines that contain a strained ortho-cycloalkanol substituent,^[12] and posited that the electrophilic nature of N-aryl nitrenoid triggered the ring-expansion-rearrangement sequence to provide 14. We were curious if this electrophilicity could be harnessed to enable the synthesis of dibenzazepines by trapping 16 with a pendant arene. If successful, this approach would enable the synthesis of differentially substituted dibenzazepines under mild conditions through the construction of a C─NAr from a C─H bond and address the weakness in the traditional approach to this scaffold that requires a symmetrical substrate.

Scheme 1.

Construction of the dibenzazepine scaffold.

Results and Discussion

To determine if dibenzazepines could be constructed from ortho-substituted anilines, the reactivity of 15 was investigated towards oxidants (Table 1). The substrate for our study was synthesized by a Heck cross-coupling reaction between 2-bromoaniline and styrene, followed by hydrogenation of the resulting stilbene, and terminating with a nitrogen-protection reaction. The optimization process began by submitting 15a (R = H) to iodosobenzene (PhIO) in hexafluoroisopropanol (HFIP) with 4 Å molecular sieves as an additive to remove the water byproduct because we found this combination of solvent and oxidant to be effective in promoting organoiodine-catalyzed oxidative cyclization-migration reactions.^[13] In contrast to our previous work, a poor yield of 17a was observed (entry 1). We anticipated that the low yield of dibenzazepine 17a was due to product instability because only 44% of 17a remained when it was exposed to the oxidative reaction conditions.^[14] Consequently, a series of N,N-disubstituted substrates were screened. While the use of N-alkyl or N-Boc substrates did not improve the yield of dibenzazepine (entries 2 and 3), changing the N-substituent to a more robust electron-withdrawing protecting group significantly increased the yield of the transformation to afford N-Ac 17d in 85% and N-Ts 17e in 88% (entries 4 and 5). Submission of trifluoroacetaniline 15f to reaction conditions, however, produced 17f in only 11% to suggest that the oxidative reaction is inhibited when too strong of an N-electron-withdrawing group is present (entry 6).^[15] Hexafluoroisopropanol is critical to the success of the reaction:^[16] changing the solvent severely attenuated the yield of dibenzoazepine 17e.^[14] In attempt to increase the yield and reduce the reaction time, a stronger oxidant, PhI(O₂CCF₃)₂ (PIFA) was investigated. Using this oxidant enabled the reaction time to be reduced to 30 minutes and still afforded dibenzazepine 17e in good yield (entry 7). Our next experiments focused on additives to further increase the yield of 17e. No improvement in the reaction outcome was observed with the addition of trifluoroacetic acid (entry 8). In contrast, the addition of magnesium oxide to eliminate the acid byproduct gave 17e in 95% in as little as 5 minutes (entry 9), and the yield remained 95% when the reaction was scaled to 1 mmol. While [hydroxy(tosyloxy)iodo]benzene (HTIB) produced 17e in great yield (entry 10), other oxidants screened (e.g. (diacetoxyiodo)-benzene (PIDA), Selectfluor, or m-CPBA) gave poor yields of dibenzazepine (entries 11 – 13). To determine if catalysis was possible, 15a was submitted to the conditions we identified to trigger 3H-indole formation (5 mol % PhI and 1.1 equiv of Selectfluor), but only 21% of 17e was observed (entry 14).^[13] These investigations revealed that benzazepines could be smoothly formed using either PhIO and 4Å MS (entry 5) or PIFA and MgO (entry 9).

Table 1.

Development of optimal conditions for dibenzazepine formation.

entry	#	R	oxidant (equiv)	additivea	time	yield 17, %b
1	a	H	PhIO (1.1)	4Å MS	16 h	37
2	b	iPr	PhIO (1.1)	4Å MS	16 h	12
3	c	Boc	PhIO (1.1)	4Å MS	16 h	39
4	d	Ac	PhIO (1.1)	4Å MS	16 h	85
5	e	Ts	PhIO (1.1)	4Å MS	16 h	88
6	f	TFA	PhIO (1.1)	4Å MS	16 h	11
7	e	Ts	PhI(O2CCF3)2 (1.1)	…	0.5 h	73
8	e	Ts	PhI(O2CCF3)2 (1.1)	TFA	5 min	72
9	e	Ts	PhI(O2CCF3)2 (1.1)	MgO	5 min	95 (95)c
10	e	Ts	HTIB (1.1)	MgO	0.3 h	92
11	e	Ts	PhI(O2CCH3)2 (1.1)	MgO	5 min	22
12	e	Ts	Selectfluor (1.1)	MgO	3 h	27
13	e	Ts	m-CPBA (1.1)	MgO	10 h	trace
14	e	Ts	PhI (0.05) Selectfluor (1.1)	TFA	3 h	21

^a200 wt % of 4Å MS, 2 equiv of TFA, or 2 equiv of MgO added

^bAs determined using ¹H NMR spectroscopy using CH₂Br₂ as an internal standard.

^cReaction performed using 1.00 mmol of 15e. HFIP = hexafluoroisopropanol; Ac = acetyl; Ts = tosyl; TFA = trifluoroacetic acid; HTIB = [hydroxy(tosyloxy)iodo]benzene; m-CPBA = meta-chloroperbenzoic acid.

The effect of changing the substituents on the aniline portion of 15 was investigated (Scheme 2). The impact of changing the electronic nature of the aryl amine was assayed by changing the identity of the R²- or R³-substituent. While significant decomposition was obtained with an electron-donating methoxy R²-group, the yield of dibenzazepine 17 increased as R² became more electron-withdrawing. Aniline 15l, however, required the use of PhIO to minimize decomposition biproducts. To determine the effect of increasing the steric environment around either the nitrogen- or the ortho-phenethyl substituent, a methyl group was placed at either the R¹- or R⁴-position. While increasing the steric environment around the nitrogen decreased the yield of 17p, the presence of an R⁴-methyl did not negatively affect the reaction outcome to provide 17q in 70% provided that PhIO was used.

Scheme 2.

Scope and limitations with regards to the aniline portion.

Next, changing the identity of the ortho-phenethyl substituent was surveyed (Scheme 3). In contrast to the reactivity of 15g, the addition of a methoxy substituent did not hinder the reaction: exposure of 15r to reaction conditions provided 17r in 94%. This reversal of the electronic preference was also observed in substrates that varied the identity of the 4-substituent; higher yields were observed with electron-donating substituents (e.g. 17r and 17s) than 15t bearing an electron-withdrawing 4-fluorine substituent. Our next investigations focused on the effect of changing the identity of the tether. While terphenyl amine 15u and methyl-substituted 15v were smoothly converted N-heterocycles 17u and 17v, benzoate 15w proved to be recalcitrant towards the reaction conditions. The length of the tether between the sulfonamide group and the aryl nucleophile was varied with 15x and 15y to determine if six- or eight-membered N-heterocycles could be constructed. The synthesis of dihydroacridine 17x and dibenzazocine 17y illustrates the generality of our method to create N-heterocycles irrespective of their ring-size. To determine if the reaction was site selective, 18a and 18b were subjected to reaction conditions to provide dibenzazepines as an 85:15 mixture of regioisomers. Improved selectivity was observed from dimethoxy-substituted 18c to provide a 96:4 mixture of dibenzazepines. The electronic nature of the aryl substituent was found to impact the site selectivity of the reaction: replacing the 3-OMe group with a fluorine resulted in a 98:2 mixture of 19d and 20d when 18d was submitted to reaction conditions.

Scheme 3.

Effect of changing the ortho-substituent identity on the reaction outcome.

Dibenzazepine formation could occur via the cyclization of radical- or cationic reactive intermediates (Scheme 4).^[17] N-Iodination of 15e with PhIX₂ generates intermediate 21. Homolysis of the C–N bond in 21 produces N-centered radical 22,^{[11f, 18]} which cyclizes with the pendant arene to produce either spirocycle 23 through a 5-exo-trig cyclization or 24 through a kinetically less favorable 7-exo-trig cyclization.^[19],[20] Oxidation of cyclohexadienyl radical 23 then produces 25. Alternatively, this spirocycle could be directly formed from nucleophilic attack of the arene on 21 or nitrenium ion 26, which could be produced from elimination of PhI from 21.^[21] Ring-expansion of spirocycle would form 27,^[22] which upon deprotonation forms dibenzazepine 17e.

Scheme 4.

Distinguishing between radical- or cation-mediated mechanisms.

To determine if radical intermediates were formed in the reaction, a series of experiments were executed (Scheme 4). When ambient light was rigorously excluded, the yield of the dibenzazepine formation was attenuated to 67%.^{[11f, 18a, 18b]} The addition of radical traps proved to be more detrimental to the reaction outcome. The presence of TEMPO reduced the yield of 17e to 22%, and complete inhibition of the reaction was observed using BHT. Because the nucleophilicity of TEMPO and BHT can lead to spurious results,^[23] 1,1-diphenylethylene (DPE) was investigated because it has been shown to be a better radical trap for iodine(III)-mediated reactions.^[24],[25] Performing the reaction in the presence of DPE resulted in only decomposition being observed. Together these experiments suggest that dibenzazepine formation occurs via radical intermediates.

Conclusion

In conclusion, we have discovered a mild iodine(III)-mediated oxidation reaction to construct dibenzazepines. Our reaction requires only a slight excess of oxidant, tolerates a broad range of functionality, and is chemoselective for reaction with the pendant arene irrespective of its distance to the aryl amine. Our preliminary mechanistic investigations suggest that the reaction proceeds via radical intermediates, and our future work is aimed at further study of the mechanism and exploiting this reactivity to access other N-heterocycles.

Experimental Section

General procedure for PIFA-mediated oxidation.

To a solution of 2-substituted aniline (1.0 equiv) and magnesium oxide (2.0 equiv) in 1 mL of HFIP, a solution of [bis(trifluoroacetoxy)iodo]benzene (1.1 equiv) in 1 mL of HFIP was added dropwise over 5 minutes. Once visualization of the reaction progress using thin layer chromatography indicated that the starting material was consumed, the reactives were quenched by the addition of 3 mL of a H₂O. The resulting mixture was extracted with 3 × 5 mL of EtOAc. The combined organic phases were dried over Na₂SO₄ and filtered. The filtrate was concentrated in vacuo, and purification using MPLC (EtOAc:hexanes) afforded the product.

General procedure for PhIO-mediated oxidation.

To a mixture of 2-substituted aniline (1.0 equiv) and 4Å molecular sieves (200 wt %) in 1 mL of HFIP, a solution of iodosobenzene (1.1 equiv) in 1 mL of HFIP was added dropwise over 5 minutes. Once visualization of the reaction progress using thin layer chromatography indicated that the starting material was consumed, the reactives were quenched by the addition of 3 mL of a H₂O. The resulting mixture was extracted with 3 × 5 mL of EtOAc. The combined organic phases were dried over Na₂SO₄ and filtered. The filtrate was concentrated in vacuo, and the residue was analyzed using ¹H NMR spectroscopy using CH₂Br₂ as the internal standard. Purification using MPLC (EtOAc:hexanes) afforded the product.