Rhodium-Catalyzed Intermolecular C–H Functionalization as a Key Step in the Synthesis of Complex Stereodefined β-Arylpyrrolidines

Abstract

The synthesis of β-arylpyrrolidines via a catalytic enantioselective intermolecular allylic C(sp)³–H functionalization of trans alkenes followed by immediate reduction, ozonolysis, and then in situ diversification of the resulting cyclic hemiaminal to furnish highly-substituted, stereoenriched β-arylpyrrolidines is reported. This methodology utilizes 4-aryl-1-sulfonyl-1,2,3-triazoles as carbene precursors and the dirhodium tetracarboxylate catalyst Rh2(S-NTTL)4. A variety of β-arylpyrrolidines were prepared in good yields with high levels of diastereo- and enantioselectivity over four linear steps, requiring only a single purification procedure.

Graphical Abstract

In the past two decades the field of C–H functionalization has grown extensively, ushering in a variety of enabling new methodologies for organic synthesis.^1–3 In particular, the ability of organic chemists to utilize simple C–H bonds to effect meaningful, enantioselective chemical transformations has caused a shift in the logical approach towards synthesis.⁴ Donor/acceptor carbenes have been especially attractive synthons for C–H insertion reactions because they are sufficiently reactive to functionalize C–H bonds. The donor group attenuates the reactivity of the adjacent carbene affording an impressive range of site-selectivity,⁵ which can be further controlled by the steric nature of the catalyst.⁶ Traditionally, donor/acceptor metallocarbenes are accessed via the metal-catalyzed decomposition of diazo compounds utilizing dirhodium tetracarboxylates as optimal chiral catalysts,^5–6 which can be sterically tuned, such that primary, secondary, or tertiary C–H bonds are selectively functionalized.⁷

In recent years, 4-aryl-N-sulfonyl-1,2,3-triazoles, which undergo a retro-6π electrocyclization in solution to unveil latent diazo compounds possessing an α-imino acceptor group (Scheme 1a),⁸ have been shown to act as competent alternates to aryldiazoacetates, demonstrating the ability to access a wide-variety of traditional Rh(II)-catalyzed transformations.^9–13 However, to date, their application towards intermolecular C–H functionalization has been relatively limited in comparison to their aryldiazoacetate congeners. C–H functionalization of alkanes and cycloalkanes was achieved using naphthalimido based Rh2(S-NTTL)₄ as an optimal catalyst to provide high yields and enantioselectivities.^14a Attempts to extend C–H functionalization to substrates containing functional groups have been limited. Cyclic ethers were shown to be ineffective because ylide formation followed by ring-expansion preferentially occurred to C–H insertion.¹⁵ In 2016, our group demonstrated that benzylic and allylic C–H functionalization is a viable process (Scheme 1b),^14b and we have recently shown that cyclic and acyclic silanes are prone to C–H functionalization at the beta position to silicon.^14cA key reason for choosing 4-aryl-N-sulfonyl-1,2,3-triazoles in favor of aryldiazoacetates in synthetic strategy is the ability to map the nitrogen functionality of the α-imino acceptor group into target-oriented synthesis. In this paper, we demonstrate that allylic C–H functionalization with α-imino carbenes can be employed to gain access to a wide-variety of highly substituted and stereodefined β-arylpyrrolidines.

Scheme 1. Previous allylic C–H functionalization.

In our preliminary asymmetric investigation into allylic C(sp³)–H functionalization, we described our initial findings using three trans alkenes and para-cymene as substrates with 4-phenyl-N-me-thanesulfonyl-1,2,3-triazole as a donor/acceptor carbene precursor in the presence of Rh2(S-NTTL)4 as an optimal chiral catalyst (Scheme 1b).^14b One of the resulting products from this study was methanesulfonamide 6a, generated in 83% yield and 84% ee by lithium aluminum hydride reduction of the initial C–H functionalization product. We recognized that these types of C–H functionalization products could be useful precursors towards the formation of highly-diversified β-arylpyrrolidines. In order to evaluate this possibility, initial studies were conducted to determine if 6a could be readily converted to the desired pyrrolidine 8a (Scheme 2).¹⁶ Ozonolysis of 6a followed by quenching with dimethyl sulfide generated cyclic hemiaminal 7a in 90% yield. Reduction of 7a with triethylsilane and boron trifluoride diethyl etherate furnished the desired pyrrolidine 8a in 95% yield with retention of the asymmetric induction achieved during C–H functionalization. Next, we sought to streamline the synthetic sequence to increase the practicality of this strategy. The sequence of C–H functionalization, imine reduction, ozonolysis, and reductive amination was conducted without chromatographic purification until the last step. Using these conditions, the pyrrolidine 8a was formed in analogous yield and enantioselectivity to the stepwise procedure (66% yield, 81% ee). Similar results were obtained in a multi gramscale reaction.

Scheme 2. Preliminary studies.

With the optimized procedure for the synthesis of 3-arylpyrrolidines in hand, we continued our investigation by examining the scope of the reaction with a wide range of 4-aryl-N-sulfonyl-1,2,3-triazoles, as summarized in Figure 1. Varying the N-sulfonyl group had little influence on the reaction outcome. The N-mesyl protecting group gave the highest yield as illustrated for 8a (66% yield, 82% ee), but the 2-(trimethylsilyl)ethanesulfonyl protecting group provided the highest levels of enantioinduction as illustrated for 8c (56% yield, 86% ee). Moving forward with the N-mesyl protecting group, we explored the influence of varying the arene component on the triazole ring, as seen with 8d-n. A variety of electronically and sterically diverse triazoles were amenable to the insertion event allowing for downstream access to the 3-arylpyrrolidine motif. Electron withdrawing groups gave slightly decreased product yields with moderate levels of enantioinduction 8d-f. Electron neutral and donating groups both gave excellent yields of 8i-l with moderate to good levels of enantioinduction. Varying the meta-position of the arene ring gave similar results, but the ortho-substitued arenes demonstrated significant increases in enantioinduction with lower overall yields, as illustrated for the ortho-bromo derivative 8h (22% yield, 98% ee).

Figure 1. Scope of triazole component^a-c.

^aInsertion reactions run at 4.0 equiv of 4a; ozonolysis reactions run at 4.0 equiv DMS; reduction reactions run at 3.2 equiv of HSiEt₃ and BF₃ OEt₂.^bIsolated yields.^cee determined by HPLC analysis after purification.^dInsertion reaction run at 60 °C.

Aiming to generate two new stereogenic centers in one synthetic operation, we turned our attention towards the secondary C–H functionalization of trans alkenes to provide stereodefined 3- aryl-4-alkylpyrrolidines (Figure 2). In our previous study we demonstrated that trans-3-hexene undergoes secondary C–H functionalization providing a 70:30 mixture of diastereomers 9a:9b. Therefore, we conducted a brief study to enhance the level of diastereoselectivity prior to downstream pyrrolidine synthesis. Initially, we explored substrates with increased steric bulk on the alkene (Figure 2a). Using trans-2,2-dimethyl-3-hexene as a substrate switched the diastereoselectvity of 10 but still led to lower diastereoselective ratios overall (38:62 dr, 10a:10b). Attempting the reaction using tri-substituted 2-methyl-2-pentene failed to generate any of the desired C–H functionalization product. This result is not too surprising given that previous studies with aryldiazoacetates have also shown that allylic C–H functionalization is sterically blocked by using cis-substitued alkenes.¹⁷ Next, we examined the effect of decreasing steric bulk on the alkene component by using trans-2-pentene as a substrate, which also provides the possibility for two distinct regiochemical outcomes, insertion at the primary methyl or secondary ethyl group. Previous C–H functionalization studies using 4-aryl-N-sulfonyl-1,2,3-triazoles have shown that Rh2(S-NTTL)4 has a preference for insertion into more substituted secondary and tertiary sites,^14a–b and we were pleased to observe the reaction go cleanly for the secondary site furnishing the desired insertion product 12 in 84:16 dr, 73% combined yield, and 97% ee for the major diastereomer. The phthalimido based rhodium tetracarboxylate catalysts, Rh2(S-PTAD)₄ and Rh₂(S-PTTL)4 were examined for comparison to the naphthalimido based Rh₂(S-NTTL)₄ but both resulted in the formation of 12 with inferior yields, diastereo-, and enantioselectivities (Figure 2b). Finally, we examined the effect of varying the electronic environment of the arene component on the triazole ring (Figure 2c). Using electron withdrawing and slightly donating para-substituents furnished the desired insertion products 13–15 but in decreased yields, diastereo-, and enantioselectivities. The strongly donating para-methoxy substituent gave desired product 16 with improved yields and diastereoselectivites with a moderate decrease in enantioselectivity (84% yield, 92:8 dr, 85% ee).

Figure 2. Optimization of secondary C–H insertion reaction^a-c.

^aReactions run with 4.0 equiv of alkene. ^bIsolated yields. ^cdr determined by ¹H NMR of crude reaction mixture. ^dee determined by HPLC analysis after purification. ^eRh₂(S-NTTL)₄ used unless otherwise indicated. ^fReaction run at 60 °C.

Following optimization for the secondary C–H functionalization, initial attempts at accessing stereoenriched 3-aryl-4-alkylpyrrolidines were hindered by enamine induced epimerization of the 4-alkyl stereocenter. Allylic insertion product 13 was chosen as an optimal substrate to study the effects of epimerization due to its relatively low levels of diastereoselectivity along with the ease of isolating the diastereomers (Table 1). Examining the reaction conditions, we discovered that epimerization could be controlled using chloroform as a solvent and reducing the temperature of the triethylsilane, boron trifluoride diethyl etherate mediated reduction of the cyclic hemiaminal. In addition, the reagents were added in two portions at zero and thirty minutes to minimize the potential for rapid epimerization of iminium ion 17 due to the presence of excessive Lewis acid. The results were confirmed using both diastereomeric mixtures and diastereopure allylic insertion products 13a and 13b leading to the desired pyrrolidines with less than five percent epimerization under the optimized conditions.

Table 1.

Minimizing enamine induced epimerization^a

entry	Dr (13a:13b)	solvent	T (°C)b		equiv BF3·OEt3	equiv Et3SiH	combined yieldc	drd (20a:20b)	%eee 20a	%eee 20b
			[O]	[R]
1	68:32	CH2CI2	−78	−10	3.2	3.2	52	60:40	94	86
2b	68:32	CHCI3	−65	−65	2.2	2.0	48	66:34	94	88
3	>97:3	CH2CI2	−78	−10	3.2	3.2	61	92:8	93	90
4b	>97:3	CHCI3	−65	−65	2.2	2.0	68	>97:3	93	-
5	3:>97	CH2CI2	−78	−10	3.2	3.2	66	8:92	88	85
6b	3:>97	CHCI3	−65	−65	2.2	2.0	64	3:>97	-	85

^aOzonolysis reactions run at 4.0 equiv dimethyl sulfide.

^bAll reactions proceed to 23 °C.

^cIsolated yields.

^ddr determined by ¹H NMR of crude reaction mixture.

^eee determined by HPLC analysis after purification.

^fReactions run by adding half the equiv of HSiEt₃ and BF₃.OEt₂ at t₀ and t_30·.

Having gained control of the enamine induced epimerization, we examined the scope of the reaction with a range of 4-aryl-N-sulfonyl-1,2,3-triazoles and trans-2-alkenes. The overall yields were generally good with all substrates (Figure 3). Utilizing the para-methoxy-substitued arene ring with trans-2-pentene gave pyrrolidine 25 with the highest diastereoselectivity (25a:25b, 92:8) but increasing the alkene chain length resulted in modest decreases in diastereoselectivity. The unsubstituted phenyl ring gave relatively consistent diastereoselectivity regardless of the alkene chain length. In all cases, excellent levels of enantioselectivity were observed (>92% ee) except for pyrrolidine 24 derived from trans-2-octene and 4-phenyl-N-me-thanesulfonyl-1,2,3-triazole, which provided the major and minor diastereomer in 82 and 80% ee, respectively.

Figure 3. Synthesis of 3,4-disubstituted pyrrolidines^a-d.

^aInsertion reactions run at 4.0 equiv of alkene; ozonolysis reactions run at 4.0 equiv dimethyl sulfide; reduction reactions run at 2.2 equiv of HSiEt₃ and 2.0 equiv of BF₃.OEt_2. ^bIsolated yields. ^cdr determined by ¹H NMR of crude reaction mixture. ^dee determined by HPLC analysis after purification.

Trapping the iminium ion with nucleophiles other than hydride would generate β-aryl pyrrolidines with a stereogenic center at the C2 position of the pyrrolidine ring. Representative systems were examined to determine the feasibility of such a process. Reaction of cyclic hemiaminal 7a and 7i with allyltrimethylsilane using boron trifluoride diethyl etherate as a Lewis acid led to the formation of pyrrolidines 29 and 30 with exceptional diastereocontrol. Nucleophilic addition into the C2 position was also possible in the boron trifluoride diethyl etherate induced reactions with 1-methylindole but in this case the diastereoselectivity was relatively low.

The synthetic sequence described herein affords access to N-protected stereodefined β-arylpyrrolidines. The free amine can be also readily obtained by cleaving the mesyl group in the presence of phenol and hydrobromic acid at reflux for 72 h to produce 33 in 87% yield and 84% ee (eq 1).

(1)

In conclusion, we have developed a method for the synthesis of complex, stereodefined β-arylpyrrolidines enabled via a Rh(II)- catalyzed intermolecular allylic C(sp³)–H functionalization of trans alkenes with downstream ozonolysis and reduction. Moreover, we have demonstrated the synthetic utility of mapping the α-imino acceptor group into target-oriented synthesis to access a challenging class of N-heterocycles. Our current investigations are focused around developing new asymmetric transformations using 4-N-sulfonyl-1,2,3-triazoles as donor/acceptor rhodium carbene precursors.

Scheme 3. Synthesis of tetrasubstituted pyrrolidines.