Regiospecific Skeletal Editing of Azetines toward Halogenated Pyrroles

Abstract

The regiospecific ring expansion of 2H-azetines into halogenated pyrroles is disclosed. Simple reaction sequences have been developed to conceptualize this 4 → 5 skeletal editing strategy, taking advantage of the inherent reactivity of double bonds present in the initial four-membered ring systems. Detailed density functional theory (DFT) calculations are presented to explain this unusual rearrangement. Such a reaction design allows for the preparation of highly substituted halogenated pyrrole derivatives.

Introduction

While classical reaction design relies on fine retrosynthetic planning, skeletal editing strategies offer the opportunity to change the shape of pre-existing molecular scaffolds in a selective way, rendering it an appealing tool for drug design and late-stage sophistication. This concept, recently popularized by the group of Levin, can be divided into three main transformation types: (i) ring expansion by atom insertion, (ii) ring contraction via “deletion”, and (iii) “atom swap” ring-size conservation.

Pioneered by Ciamician and Dennstedt in 1881, the in situ generation of a carbene species and its addition onto a pyrrole substrate allows for the construction of pyridines, halogenated at position 3. This rearrangement, known by the names of its pioneers (Ciamician–Dennstedt rearrangement or ring expansion), formally goes through insertion of a carbon atom into one C–C bond of the pyrrole scaffold and relies on a (2 + 1)-cycloaddition followed by a rearomatizing strain-release/elimination sequence (Scheme A). More recently, the group of Levin has shown that diazirines can be employed as carbene precursors to reshape pyrroles into pyridines and pyrazoles into pyrimidines. Such a method allows for the introduction of an aryl-substituent under mild conditions. The group of Magauer also demonstrated an elegant strategy for the transformation of cyclopentenones into functionalized phenols through a somewhat similar mechanism via the use of carbenoids. Since then, not only can carbenes and carbenoids be used in molecular editing, but nitrenes and nitrenoids have also proven to be efficient in “nitrogen insertion” reactions, as independently exemplified by Morandi et al. with hypervalent iodine reagents, Ackermann et al. via electrochemical bias, and Alcarazo et al. with N-(sulfonio)­sulfilimine reagents for 5 → 6 shapeshifting. Recently, the group of Ghiazza developed a simple route for the rearrangement of substituted pyridines into functionalized 1,2-diazepines under UV light. Nitrogen insertion strategies were also employed by Wei et al. to generate pyrroles from cyclobutenes under cobalt catalysis, the limitation of which lies in the preparation of the parent four-membered rings.

1. (A) Overview of the Literature on Skeletal Editing through Atom Insertion and (B) Present Work on the 4 → 5 Ring Expansion.

Pyrroles are an essential class of heterocyclic compounds in medicinal chemistry due to their widespread biological activities and therapeutic potentials. Their structural versatility allows them to serve as core scaffolds in various drugs, targeting a spectrum of diseases, such as cancer, infectious diseases, and neurological disorders. Pyrroles are also crucial in the synthesis of natural products. This broad applicability makes pyrrole derivatives an indispensable component in the development of new pharmaceuticals. Nevertheless, the selective and efficient synthesis of halogenated pyrroles remains limited.

For the past decade, our group has focused on creating methods for the synthesis and functionalization of challenging four-membered rings. Although our journey started with cyclobutenes and alkylidenecyclobutanes, we rapidly moved to heterocyclic structures due to the increasing interest manifested by medicinal chemists to incorporate such motifs in drug discovery. In this regard, we have established efficient and reliable strategies to access polysubstituted four-membered heterocycles, including 2H-azetines. The reactivity of the internal double bond of 2H-azetines was also evaluated toward the stereocontrolled generation of sophisticated azetidine building blocks, employing regiodivergent cycloadditions, asymmetric hydrogenations to access non-natural amino acids, or radical-polar crossover of azetinylborates.

We envisioned that adequate conditions that would use the inherent reactivity of the double bond in 2H-azetines could trigger a halo-carbene mediated ring-expanding rearrangement toward the corresponding pyrroles (Scheme B), providing a unique 4 → 5 molecular shapeshifting approach toward halogenated structures. While the zinc-carbenoid insertion into N–O bonds of oxazetidines by Tsuda and Nakamura is a representative example of 4 → 5 ring expansion, only examples of azetine rearrangements toward pyrroles were described by Han and Zard in a two-step, radical-initiated sequence.

Results and Discussion

First tests were performed employing a combination of TMSCF₃/NaI to generate a difluorocarbene. Such conditions trigger a sequence for the ring-enlargement reaction (vide infra for a proposed mechanism), giving the corresponding pyrroles, as pioneeringly observed as a side reaction by Mykhailiuk et al. It is interesting to note that the rearrangement occurs regiospecifically so that the difluorocarbene inserts between C2 and C3 on the starting 2H-azetine 1, those becoming C2 and C4, respectively, on the final pyrrole 2 (Scheme A).

2. Skeletal Editing of Functionalized Azetines toward (A) 3-Fluoro-, (B) 3-Bromo-, and (C) 3-Chloropyrroles.

Monosubstituted 2H-azetine at position 2 gave 2a in a 74% yield. In parallel, monosubstituted substrates at position 3 also provided the corresponding fluorinated pyrroles 2b–e, although electron-neutral and donating moieties (2b-c) seemed to give better yields than electron-poor ones (2d). This could potentially be attributed to the electron-withdrawing character of the aryl group that lowers the electron density on the double bond and therefore renders it less nucleophilic toward cyclopropanation. It is also important to note that the reaction seems somewhat sensitive to sterics, as a 2,4,6-triisopropylphenyl at position 3 of the 2H-azetines shuts down the reactivity of the cyclopropanation/ring expansion sequence (see the Supporting Information). More interesting disubstituted substrates were engaged next, providing a wide range of trisubstituted pyrroles, fluorinated at position 3 (2e-p and 2r) in moderate-to-high yields (up to 95%). Drug-like aromatic moieties found in dapagliflozin (2p) or empagliflozin (2q) were successfully regiospecifically introduced onto fluorinated pyrroles (up to 85% yield). We were also able to synthesize a substrate possessing a substituent at position 4, on the saturated part of the 2H-azetine, adapting an elegant sequence recently described by the group of Baran, via intermediate formation of azabicyclobutanes. , With a 2H-azetine substituted with a hydrocinnamyl group at position 4, we are pleased to demonstrate the regiospecific formation of 2s in 45% yield.

The procedure was further adapted to enable the insertion of a C–Br bond via skeletal editing of 2H-azetines 1 (Scheme B). After a fine-tuning of reaction conditions, we were able to isolate a range of brominated structures at position 3 in reasonable yields. While a monosubstituted 3-arylazetine gave 3a in 25% yield, 2,3-bis-substituted azetines afforded functionalized 3-brominated pyrroles 3b–j in generally better yields, independently from the electronic nature of the substituents on aryl systems (up to 41% yield, 3f).

In contrast with the formation of brominated pyrroles, the rearrangement of 2H-azetines 1 toward chlorinated pyrroles proved challenging (Scheme C). Although many conditions were screened, especially with regard to the base and temperature for the formation of the required chlorocarbene species, yields obtained for the corresponding 3-chloropyrroles 4 were consequently less good than for fluorinated or brominated analogues. We attributed this decrease in efficiency to the harsher conditions necessary to generate the chlorocarbene.

Nevertheless, we were able to show a proof of concept for the formation of chloropyrroles 4a–e with yields up to 49% (4b). Low yields in some of the 3-chloropyrroles could be attributed to the overaddition of the in situ generated carbene onto the product of the first ring expansion. As a matter of fact, we were able to isolate some 3,5-dichloropyridines in some cases (vide infra, Scheme ).

3. (A–C) Further Applications.

Deprotection of fluorinated pyrrole 2k proceeded smoothly to give 5 in a 54% yield (Scheme A). We were also able to demonstrate full-functionalization of 2m by performing a Vilsmeier reaction (Scheme B), providing functionalized pyrrole 6 in 58%.

When conducting ring expansion reactions of azetines toward chlorinated pyrroles using dichlorocarbene precursors, we observed the formation of 3,5-dichloropyridines, providing an explanation for low yields of 3-chloropyrroles, in some cases (Scheme B). As such a strategy would provide uniquely halogenated pyridine patterns, we pushed the double addition to full consumption of the intermediate pyrrole 4d, which led to the functionalized 3,5-dichloropyridine 7 in 63% yield (Scheme C).

It is important to note that the deprotection of N-Boc seemed optional since the pyridine was obtained without any additive. In addition, a direct 4 → 6 ring expansion reaction was only observed for chlorinated scaffolds.

Mechanistic Study

With the knowledge gathered from the preliminary theoretical studies, we were able to construct a plausible reaction pathway for the conversion of the 2-azabicyclo[2.1.0]­pentane intermediate INI into halogenated pyrroles IN4 by identifying the corresponding reaction intermediates and all related transition states (Scheme A). Starting from the different halogenated species IN1 (X = F, Cl, and Br), the Gibbs free energy profiles computed at two different levels of theory are shown in Scheme B (left: M06-2X/def2-TZVP , gas phase reactivity of IN1, right: refined M06-2X/def2-QZVPD Gibbs free energy profiles including solvent corrections for THF), along with molecular structures of all relevant TS identified for the solution state simulation (Scheme C).

4. Theoretical Evaluation of the Ring Expansion Reaction .

^a (A) Proposed mechanism for the ring opening of the dihalogenated 2-azabicyclo[2.1.0]­pentanes IN1 (X = F, Cl, and Br) derived from DFT calculations. (B) Gibbs free energy reaction profiles (left: gas phase M06-2X/def2-TZVP, right: M06-2X/def2-QZVPD including PCM solvent corrections for THF) for the ring opening of IN1. All TS and diradical or diradicaloid intermediates (IN2) were computed at the unrestricted open-shell (U)­M06-2X level of theory, and only the intermediates IN1, IN3, and IN4 were treated by restricted closed-shell methods ((R)­M06-2X). (C) Molecular structures of the transition states ((U)­M06-2X/def2-QZVPD) identified, labeled with characteristic interatomic distances along the dotted orange lines. For all TS, the atomic displacement vectors (scaled by a factor of 2.5) associated with the imaginary vibrational mode are indicated by orange vectors in the forward direction of the reaction. Labeling of halogens: [TS12]^‡ and [TS23]^‡ F = green and [TS13]^‡ Cl = light green and Br = purple. (D) Energy profiles along the full intrinsic reaction coordinates (IRC) computed for all IN1 ring opening TS identified in panel (C). The black curves and solid symbols give the electronic energy (left ordinate; all open-shell (U)­M06-2X/def2-TZVP gas phase energies Δε₀ in [kJ mol^–1]), and the colored curves and open symbols show selected interatomic distances (d(C1–C4), d(C5-X), and d(C1-X), right ordinate) and their changes computed along the IRC paths.

The difluorinated species IN1 undergoes initial homolytic cleavage of the strained C1–C4 bond (TS12) to form distinct singlet diradical species IN2. This reaction step is followed by a 1,2-shift of one fluorine substituent from C5 to C1 (TS23) to yield the corresponding 2,5-dihydro-1H-pyrrole IN3. Though in the gas phase calculations, the second barrier along this pathway seems rate-limiting (TS12: ΔG ^‡ = 78.6 kJ mol^–1; TS23: ΔG ^‡ = 100.0 kJ mol^–1), structure refinements including solvent corrections for THF affect both barriers. In the solution state, a value of ΔG ^‡ = 71.6 kJ mol^–1 is computed for TS12, and TS23 is lowered substantially to ΔG ^‡ = 53.3 kJ mol^–1, shifting the rate-limiting step to breaking the C1–C4 bond during ring opening. However, for the dichlorinated and dibrominated species IN1, the mechanism changes, as now both the ring opening and the 1,2-halogen shift occur almost simultaneously with only a slight offset. For X = Cl and Br, the reaction IN1 → IN3 becomes a one-step process with a single transition state only, and the energy barrier of the corresponding TS13 is only slightly affected by solvent effects (Scheme B). In total, the overall process IN1 → IN3 is highly exergonic in all cases, and final aromatization to furnish the thermodynamically favorable pyrroles IN4 via 1,4-elimination of HX adds to the total energy balance of the entire reaction. Though we have tried to explicitly model the intramolecular elimination of HF in the case of X = F, we were not able to locate and confirm a transition state. However, this type of 1,4-elimination is not uncommon and may occur via many different pathways (including base-assisted intermolecular steps or, e.g., HF elimination facilitated by silyl-species present in the reaction mixture). Due to the uncertainty in estimating the exact energy contribution of the elimination side-product HX in this step, the barrier for this final process can only be estimated with a very high degree of uncertainty, which is why we have decided not to model TS34 itself at all.

For all transition states identified in Scheme B, molecular geometries and characteristic interatomic distances are shown in Scheme C. It is worth noting that all of these TS have been verified by computing the corresponding gas phase energy profiles along their intrinsic reaction coordinates (IRCs), and these plots are displayed in Scheme D along with some selected characteristic changes of interatomic distances. For X = F, the C1–C4 distance significantly increases around TS12 during ring opening, whereas in the course of the 1,2-F shift, the C5–F and C1–F distances become equal only after TS23. For both X = Cl and Br, both of these processes collapse into a single transition state TS13 only. Here, the C1–C4 distances increase concomitantly with TS13, whereas the C5-X and C1-X distances form an isosceles triangle C5-X-C1 only shortly thereafter. The combined nature of both the ring opening and 1,2-X shift processes becomes evident only through an inflection in the corresponding energy profiles, which ultimately leads to the formation of IN3.

Notably, the diradical character of IN2 was already firmly established for X = F, but a potential diradical or diradicaloid character could not be excluded a priori for any of the transition states or any intermediate structure along the reaction pathways shown in Scheme D (where the term “intermediate” is not limited only to the energy minimum intermediates IN1–IN 3). Thus, these IRC energy profiles were computed using unrestricted DFT ((U)­M06-2X) calculations employing special measures to potentially identify such open-shell species. Nevertheless, these IRC calculations were repeated on exactly the same molecular geometries using restricted closed-shell DFT ((R)­M06-2X) methods, and the corresponding IRC profiles and the absolute differences of the electronic energies (|Δε₀(U – R)|) between both methods are plotted in the Supporting Information, S6. Clearly, for X = F, the diradical IN2 exists between both transition states TS12 and TS23 in the central part of the IRC energy profile, with the restricted closed-shell electronic structure being approximately 20–24 kJ mol^–1 less favorable than a “real” diradical-type open-shell configuration.

However, the IRC profiles before TS12 and after TS23 for X = F, as well as the entire IRC profile for X = Cl or Br, display only negligible differences (|Δε₀(U – R)|<10^–7 kJ mol^–1) in electronic energies calculated by both methods. It must therefore be concluded that none of the transition states discussed here exhibit a diradical or diradicaloid character nor do the intermediate products of the chlorinated or brominated reactions.

Conclusions

We developed a conceptual approach to halogenated pyrroles from azetines through 4 → 5 skeletal editing using simple in situ generated dihalogenocarbenes. On the one hand, this strategy provides access to uniquely substituted 3-fluoro-, 3-bromo-, and 3-chloropyrroles in a regiospecific way. On the other hand, exploring the limits of such a reaction brought us to discover the formation of unprecedented, interesting halogenated pyridines that hold great potential for further applications. A plausible reaction mechanism was investigated computationally, revealing a divergence in the pathways between fluorinated and other halogenated derivatives.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.5c01324.

• Experimental procedures and compound characterization (PDF)

All authors have given approval to the final version of the manuscript. CRediT: Rahma Ghazali formal analysis, investigation, methodology, writing - review & editing; Elias Schick formal analysis, investigation, methodology; Sébastien Lepre investigation, methodology; Jakob Kranich investigation, methodology; Stefan Immel computational studies, writing - review & editing; Dorian Didier conceptualization, formal analysis, funding acquisition, project administration, supervision, validation, writing - original draft, writing - review & editing.

PhD positions to R.G. and E.S. (Landesstelle, TU Darmstadt), Heisenberg-Professur to D.D. (DFG), and Erasmus scholarship to S.L.

The authors declare no competing financial interest.