Switchable Regioselective 6-endo or 5-exo Radical Cyclization via Photoredox Catalysis

Abstract

Controlling the regioselectivity of radical cyclizations to favor the 6-endo mode over its kinetically preferred 5-exo counterpart is difficult without introducing substrate prefunctionalization. To address this challenge, we have developed a simple method for reagent controlled regioselective radical cyclization of halogenated N-heterocycles onto pendant olefins. Radical generation occurs under mild photoredox conditions with control of the regioselectivity governed by the rate of hydrogen atom transfer (HAT). Utilizing a polarity-matched thiol-based HAT agent promotes the highly selective formation of the 5-exo cyclization product. Conversely, limiting the solubility of the HAT reagent Hantzsch ester (HEH) leads to selective formation of the thermodynamically favored 6-endo product. This occurs through an initial 5-exo cyclization, with the resulting alkyl radical intermediate undergoing neophyl rearrangement to form the 6-endo product. Development of this switchable catalysis strategy allows for two modes of divergent reactivity to form either the 6-endo or 5-exo product, generating fused N-heteroaromatic/saturated ring systems.

Graphical Abstract

Authors are required to submit a graphic entry for the Table of Contents (TOC) that, in conjunction with the manuscript title, should give the reader a representative idea of one of the following: A key structure, reaction, equation, concept, or theorem, etc., that is discussed in the manuscript. Consult the journal’s Instructions for Authors for TOC graphic specifications.

Free radical cyclizations have a storied history in complex molecule synthesis, with their predictable selectivities and functional group tolerance contributing to their strategic importance.¹ It is well established that radical cyclizations typically occur in the 5-exo mode due to the lower kinetic barrier associated with the 5-exo transition state compared to its 6-endo counterpart.² Overcoming the barrier to access the 6-endo product traditionally requires either substrate manipulation to sterically inhibit 5-exo cyclization³ or the introduction of ring strain to encourage formation of the more thermodynamically stable 6-membered ring.⁴ The requirement for regio-directing features in the substrate places inherent limitations on the scope of these 6-endo selective transformations.

In this manuscript, we describe the development of switchable catalytic methods that utilize reagent control to select between the exo and endo cyclization products. The ability to select for either the 5-exo or the 6-endo product for a given substrate represents a significant advance, and provides new opportunities to exploit radical cyclizations for the synthesis of pharmaceutically relevant molecules.

Underpinning our reaction design are Beckwith⁵ and Stork’s⁶ seminal reports on Bu₃SnH/AIBN initiated vinyl radical ring closures (Figure 1, top). These studies demonstrated that the concentration of Bu₃SnH influenced product distribution, with high initial concentrations favoring the 5-exo product. Conversely, slow addition and lower concentrations of Bu₃SnH modestly decreased the exo/endo ratio. The influence of H-atom transfer (HAT) reagent concentration on exo/endo ratio is understood to arise from the neophyl rearrangement, which provides a mechanistic pathway for the reversible interconversion of the 5-exo (kinetic) and 6-endo (thermodynamic) radical intermediates.^7,8 Despite these early studies, a general method to selectively access both the 6-endo or the 5-exo cyclization products from the same substrate has yet to be established.

Figure 1.

Approach to selective 6-endo or 5-exo radical cyclization products.

With recent developments in photoredox catalysis presenting opportunities for precise control of radical generation and capture,⁹ we envisioned developing a switchable catalysis method with reagent control to access either the 5-exo or 6-endo products. Controlling the lifetime of the alkyl radical intermediate I resulting from the initial 5-exo cyclization is crucial to this endeavor. Rapidly intercepting this radical intermediate with a hydrogen atom will selectively deliver the 5-exo cyclized product, while a prolonged lifetime will favor thermodynamically driven neophyl rearrangement to 6-endo alkyl radical II. Thus, we hypothesized the rate of hydrogen atom delivery to the alkyl radical intermediate will impart regioselective control.

The prevalence of pyridines in biologically relevant molecules,¹⁰ combined with our group’s understanding of pyridyl radical reactivity,¹¹ led us to target pyridyl radical cyclization to explore this switchable catalysis concept (Figure 1, middle). We noted that both the 5-exo and 6-endo (dearomatized quinoline derivatives) product motifs of the anticipated reactions are present in a variety of natural products and medicinally relevant molecules (selected examples shown in Figure 1, bottom)¹² and that the methods outlined in this manuscript present a particularly attractive way to access them.¹³

To assess the feasibility of our switchable catalysis hypothesis, we initiated our selectivity study with 2-bromo−3-allyloxy pyridine (1). The rate constant for H-atom transfer from t-BuSH to primary alkyl radicals (k 8 × 106 M−1s−1)¹⁴ exceeds that for pyridyl neophyl rearrangements (k ≅ 8 × 102 s−1)⁸ by four orders of magnitude (Figure 2A). Thus, we expected a polarity-matched thiol-based HAT catalyst to provide excellent kinetic selectivity for the 5-exo product. Our group’s recent disclosure on generating the strongly reducing CO₂ radical anion (CO₂•−) via a thiol-based HAT catalyst provided an ideal set of conditions for our initial experiment.¹⁵ Indeed, subjecting allyloxy bromopyridine 1 to irradiation under blue light for 16 hours with photocatalyst 4CzIPN (1 mol%), HAT catalyst mesna (20 mol%), and 5 equiv of sodium formate/formic acid in DMSO selectively delivered 5-exo product 1a in 97% yield, with no 6-endo product 1b observed (Table 1, entry 1). To select for the 6-endo product, we sought to exploit the tunability of Hanztsch ester (HEH) H-atom transfer reagents. An initial experiment using 1 mol% 3DPAFIPN as the photocatalyst and excess HEH (5 equiv) demonstrated that HEH was less efficient at trapping the kinetic 5-exo product, and gave a 26% yield of the 6-endo product (2.7 : 1 exo/endo, entry 2). Lowering HEH equiv to 3 further decreased production of 1a (55%) and increased the yield 1b (39%), but 1a was still the major product (1.4 : 1.0 exo/endo, entry 3). Our group has shown that solvent choice can limit the solubility of HEH, reducing the concentration of H-atom donor present and significantly altering reaction outcome.¹⁶ With this in mind, we chose MeCN as the solvent, and were delighted to observe that the resulting product mixture favored the 6-endo isomer 1b (58%) rather than 5-exo 1a (28%, 1.0 : 2.1 exo/endo, entry 4). Further limiting the solubility of HEH by adding water as a cosolvent improved 1b yield (80%) and exo/endo ratio to 1.0 : 4.4 (entry 5). Finally, lowering the equivalents of HEH to 1.5 gave full starting material consumption, increased 1b yield to 84% and limited the production of 1a to just 15% (1.0 : 5.6 exo/endo, entry 6).

Figure 2.

Mechanistic proposal explaining origins of reagent controlled switchable selectivity. aRatio determined by 1H NMR with internal standard.

Table 1.

Optimization of switchable catalysis to access either the 5-exo or 6-endo radical cyclization products^a,c

^aConditions: 2-bromo-3-allyloxy pyridine (0.1 mmol), 3DPAFIPN (1 mol%), HEH (0.15 mmol), solvent (3 mL), blue LED, 23°C, 16h.

^bConditions: 2-bromo-3-allyloxy pyridine (0.1 mmol), 4CzIPN as photocatalyst (1 mol%), mesna (20 mol%), formic acid (2.5 mmol), sodium formate (2.5 mmol), solvent (2 mL), blue LED, 23 °C, 16h.

^cYields determined by ¹H NMR with internal standard.

With optimized exo and endo conditions in hand, we evaluated the ability of these methods to discriminate between the exo and endo products by designing a series of substrates that probe the impact of both olefin and pyridine substitution (Table 2). Substrate 2, in which internal substitution of the olefin should increase the thermodynamic favorability of the neophyl rearrangement from the primary to tertiary alkyl radical, gave exclusive exo product 2a in nearly quantitative yield (97%) when subjected to exo conditions. When we subjected 2 to endo conditions, selectivity completely switched giving 2b in good yield (79%), with no detection of 2a in the crude NMR, highlighting the power of the reagent control for switchable catalysis. Terminal olefin substitution at the 6-position (3) could potentially retard the neophyl rearrangement, and we were pleased to observe that this substitution pattern also delivered good yields (97% and 78% for exo and endo conditions, respectively) and excellent selectivity (1.0 : 12.5 and >19 : 1.0 exo/endo). Trisubstituted olefin 4 gave similar selectivity (1.0 : 11.7 and >19 : 1.0 exo/endo, albeit it with reduced yields (69% and 26% for exo and endo conditions, respectively).

Table 2.

Scope of 5-exo and 6-endo radical cyclization

^aConditions: 4CzIPN (1 mol%), substrate (1 equiv), mesna (20 mol%), sodium formate (5 equiv), formic acid (5 equiv), DMSO, blue LED, 23 °C, 16 h, isolated yields shown.

^bConditions: 3DPAFIPN (1 mol%), substrate (1 equiv), Hanztsch ester (1.5 equiv), H2O:MeCN (1:1 v/v), blue LED, 23 °C, 16 h, isolated yields shown.

^cisolated as HCl salt. dYields determined by ¹H NMR with internal standard.

We next turned our attention to investigating the nature of the linker and its effect on product distribution. Alpha-substitution in substrate 5 had little impact on product distribution with exo (5a, 88%, >19 : 1.0 exo/endo) and endo (5b, 82%, 1.0 : 5.9 exo/endo) conditions giving good yields and selectivity. N-boc linked olefin (6) delivered endo product 6b in 78% yield and 1.0 : 3.4 exo/endo ratio. We postulate the slightly lower endo selectivity arises from the decreased stabilization of the radical in the cyclopropyl intermediate by the N-boc lone pair compared to the oxygen lone pair in previous substrates.¹⁷ Carbon linked olefin 7 also gave a lower yield of product (53%) and reduced endo selectivity (1.0 : 1.6 exo/endo), again likely due to lacking lone pair stabilization effects on the radical in the cyclopropyl intermediate. Though endo selectivity suffered for 6 and 7, exo conditions remained highly selective (>19 : 1.0 exo/endo in both cases).

Unsurprisingly, substitution at the 4-, 5-, and 6-position of the pyridine (8, 9, 10) was well tolerated with exquisite selectivity (>19 : 1.0 for exo and 1.0 : >19 exo/endo for endo conditions) and yields >70% for either regioisomer. Demonstrating additional generalizability, substrate 11 with bromination at the 4-position behaves similarly to the analogous 2-brominated substrate with good selectivity for both the exo (>19 : 1.0 exo/endo) and endo (1.0 : 4.4 exo/endo) conditions. The lower yields (62% for 11a and 26% for 11b) likely arise from the instability of 4-brominated pyridines.¹⁸ Expanding the scope to other N-heterocycles, pyrimidine 12 gave lower exo selectivity (1.9 : 1.0 exo/endo), but gave excellent selectivity under endo conditions (1.0 : >19 exo/endo). This pattern is explained by the lower aromatic stabilization energy of pyrimidine relative to pyridine, accelerating dearomatization to form the cyclopropyl intermediate in the neophyl rearrangement.¹⁹ Benzimidazole 13 was also reactive under both conditions, with the stability of the cyclopropyl radical intermediate again impacting selectivity.

To probe our mechanistic hypothesis and quantitatively assess the role of HEH solubility on promoting the neophyl rearrangement, we measured HEH solubility over a range of solvent mixtures of DMSO/MeCN and H₂O/MeCN. A strong correlation between reduced HEH solubility and 6-endo selectivity was observed (Figure 2B).

Our proposed unified mechanistic understanding for the switchable selectivity is shown in Figure 2C. Under exo conditions, single electron reduction of 1 by CO₂•− (E_1/2o = −2.21 V vs. SCE), generated through an HAT event from formate by thiyl radical, induces homolytic cleavage of the bromopyridine bond delivering pyridyl radical III.¹⁵ Under endo conditions, single electron transfer to 1 from 3DPAFIPN•−, generated via reductive quenching of 3DPAFIPN* (E_1/2o = +1.09 V vs. SCE)²⁰ by HEH (E_1/2o = 0.97 V vs. SCE),²¹ delivers pyridyl radical III.²² Given 3DPAFIPN•− (E_1/2o = −1.59 V vs. SCE)²⁰ has an underpotential for reducing 2-bromopyridines (E_1/2o = −2.26 V vs. SCE for 2-bromopyridine),²³ we suggest that substrate reduction is assisted by protonation from the Hantzsch pyridinium produced in each catalytic turnover.^11,24,25 In both cases, III undergoes 5-exo cyclization to alkyl radical IV. Under exo conditions, alkyl radical IV is rapidly trapped by the more efficient thiol-based HAT reagent. Under endo conditions, with the low concentrations of the less efficient HAT reagent HEH, alkyl radical IV has sufficient lifetime to undergo thermodynamically driven neophyl rearrangement to the more stable alkyl radical V. Hydrogen atom delivery to V from HEH•+ yields the endo product.

To further investigate the reduction mechanism of each pathway, aryl iodide 14 was subjected to both sets of reaction conditions (Scheme 1). As expected, under the exo-selective conditions, the strongly reducing CO₂•− was capable of initiating the reaction, and near quantitative yield (99%) of 14a with complete exo selectivity (>19 : 1.0 exo/endo) was observed. However, substrate 14 did not react under endo conditions and quantitative starting material return was observed by 1H NMR. This is consistent with the hypothesis that 3DPAFIPN•− is the reductant in the endo-selective conditions, and that it is not sufficiently reducing to engage aryl iodide 14.

Scheme 1. Investigation of non-heteroaryl substrate.

aConditions: 4CzIPN (1 mol%), substrate (1 equiv), mesna (20 mol%), sodium formate (5 equiv), formic acid (5 equiv), DMSO, blue LED, 23 oC, 16 h, isolated yields shown. bConditions: 3DPAFIPN (1 mol%), substrate (1 equiv), Hanztsch ester (1.5 equiv), H₂O:MeCN (1:1 v/v).

In summary, we have developed mild photoredox reaction conditions that allow switchable control in radical cyclization reactions to select for either the 5-exo and 6-endo cyclization products. The two reactivity modes are controlled by the choice of HAT reagent, rather than substrate manipulation. The use of an HAT reagent with a rapid delivery of hydrogen atom gives selective 5-exo products. Decreasing the rate of HAT by limiting the solubility of a less efficient HAT reagent preferentially generates 6-endo products. Our mechanistic proposal accounts for the observed differences in selectivity across a range of substrates and provides a sound basis for predicting the selectivity for other substrates. Further investigations to advance the synthetic impact of our 5-exo and 6-endo selectivity, such as extending 6-endo selectivity to non-heteroaryl species, are ongoing.