Synthesis of Pyrroles via Consecutive 6π-Electrocyclization/Ring-Contraction of Sulfilimines.

We present a modular, synthetic entry to polysubstituted pyrroles employing readily available 2,5-dihydrothiophenes. Ring-opening of the heterocycle provides access to a panel of 1,3-dienes which undergo pyrrole formation in the presence of inexpensive chloramine-T trihydrate. The transformation is conducted in an open flask and proceeds at ambient temperatures (23 °C) in nondry solvents. A careful adjustment of the electronics and sterics of the 1,3-diene precursor allows for the isolation of key intermediates. DFT studies identified a reaction mechanism that features a 6π-electrocyclization of a sulfilimine intermediate followed by spontaneous ring-contraction to reveal the pyrrole skeleton.

The efficient construction of structurally encumbered and highly functionalized heterocycles represents one of the major challenges for the development of novel pharmaceuticals and agrochemicals.[1] In particular, tetrasubstituted pyrroles have served as valuable lead structures in medicinal chemistry to develop the anticancer agent sunitinib (1, Sutent),[2] the cholesterin-lowering drug atorvastatin (2, Lipitor),[3] and the Ca2+-channel activator FPL 64176 (3, Scheme A).[4] For the assembly of these heterocycles, condensation chemistry has dominated the field for decades[5] and powerful transition-metal based coupling strategies have only emerged later.[6] Ring formation relying on pericyclic reactions represents a conceptionally different strategy which has found widespread application in all areas of heterocyclic chemistry. For instance, with the establishment of 1,3-dipoles by Huisgen, cycloaddition reactions became available as a robust method to synthesize a variety of five-membered heterocycles.[7] This includes the [3 + 2]-cycloaddition reaction of azomethine, carbonyl, and thiocarbonyl ylide intermediates to allow for the rapid assembly of pyrroles, furans, and thiophenes.[8] On the other hand, sigmatropic rearrangements have been extensively used to construct, for instance, indoles.[9] For the synthesis of benzofuran derivatives, interrupted Pummerer reactions[10] were reported to initiate charge-accelerated [3,3]-sigmatropic rearrangements.[11] However, electrocyclization reactions have remained in a niche and have mainly been applied to the synthesis of six-membered heterocycles. For example, the 6π-electrocyclization of azatrienes was shown to provide a broad range of pyridines.[12]

Scheme 1

Pyrroles in Medicinal Chemistry and “Heterocycle Switches” of 2,5-Dihydrothiophenes into Furans and Pyrroles

During our studies to convert readily available 2,5-dihydrothiophenes 4(13) into tetrasubstituted furans 6, we found an unprecedented 6π-electrocyclic ring-opening as part of the reaction mechanism (Scheme B).[14] While we were able to access a variety of furans, all efforts to prepare the corresponding pyrroles via exchange of the carbonyl function for an imine failed. However, we later found that the exposure of 1,3-diene 5a to inexpensive chloramine-T effects selective sulfilimine formation. In contrast to a preliminary study relying on high temperatures (130 °C, two examples),[15] subsequent 6π- electrocyclization/ring-contraction/elimination[16] of 7 proceeded spontaneously at 23 °C in an open flask to give pyrrole 9 (Scheme C).

Employing Sharpless’ conditions for the synthesis of N-tosyl sulfilimines (chloramine-T trihydrate, acetonitrile, 23 °C),[17] we observed rapid conversion of 1,3-diene 5a to pyrrole 9a in 53% yield (Scheme , entry 1). The 2,5-dihydropyrrole 10 was isolated as the second product together with traces of trisubstituted pyrrole 11, which might originate from 10 via a competing oxidation pathway. Further screening revealed slightly lower yields for the solvents N,N-dimethylformamide, methanol, and water (32–49%, entries 2–4). In the presence of 1 equiv of p-toluenesulfonic acid monohydrate (p-TsOH·H2O, entry 5), the yield was increased to 70%. The use of hexafluoroisopropyl alcohol (HFIP) as the cosolvent allowed for the removal of p-TsOH·H2O and further improved the yield of 9a to 84% (entry 6). The use of 1.5 equiv of chloramine-T trihydrate or anhydrous chloramine-T (2 equiv) led to decreased yields (41–65%, entries 7 and 8). Dichloramine-T (TsNCl2) led to rapid consumption of the substrate, but pyrrole formation was accompanied by decomposition to give 9a in only 23% yield. Variation of the vinyl sulfide revealed diene 5a (R = Me) to be superior to 5b (R = Et, 68%) and 5c (R = Ph, 59%), delivering pyrrole 9a in an 83% isolated yield. The addition of m-chloroperbenzoic acid (m-CPBA) after full conversion of the starting material allowed for selective sulfur oxidation of 11 and facilitated the isolation of pure 9a.

Scheme 2

Optimization Studies

Legend: (1) yield determined by 1H NMR analysis using nitromethane as internal standard; (2) isolated yield, 0.2 mmol scale of 5a–c. Abbreviations: Ts = p-toluenesulfonyl, DMF = N,N-dimethylformamide, HFIP = hexafluoroisopropyl alcohol, m-CPBA = m-chloroperbenzoic acid.

With our optimized conditions in hand, we investigated the robustness and compatibility of the protocol for a panel of 1,3-dienes (Scheme ). The scalability was demonstrated by the rapid synthesis of more than 1.5 g (78%) of pyrrole 9a in a single run. Modifications of R1 (highlighted in red) allowed for the implementation of electronically enriched arenes and a thiophene to give 9b–d in constantly good yields (72–79%). The presence of a strongly electron withdrawing substituent such as a nitro group (9e) or a trifluoromethyl group (9f) was well tolerated (63–64%). Different aryl halides were also shown to effectively undergo pyrrole formation to deliver chloride 9g, fluoride 9h, and bromide 9i in high yields between 69 and 78%. In addition, tertiary amide 9j and aldehyde 9k were accessible from the reaction (59–65%). As shown for the synthesis of the alkyl (R1 = Me, n-Bu)- and allyl-substituted pyrroles 9l–n (52–76%), an aryl residue was not required at the C3 position. Only alkyne 9o and pivalate 9p were obtained in lower yields (28–30%). Lactone 9q (42%) was also accessible, thus expanding the synthetic utility to annelated ring systems. When the ester was changed to amides (R2, highlighted in blue), the primary and secondary amides 12a,b were isolated in 56 and 81% yields, respectively. The latter bears the 3,4-substitution pattern as found in atorvastatin (2). Additionally, the Weinreb amide 12c was synthesized in 33% yield. Ketones also participated in the transformation and gave the di- and trisubstituted pyrroles 13a–c in good yields (55–77%). The presence of nitriles was also tolerated under the reaction conditions but required the absence of m-CPBA during the workup process. This allowed for the isolation of pyrrole 14a in 51% yield (18% in the presence of m-CPBA). Consequently, we were able to prepare pyrrole 14b (42%), which was quantitively converted to the fungicide fludioxonil (15, Pestanal)[1c,18] through N-tosyl cleavage under basic conditions (NaOH, MeOH). Application of O-mesitylenesulfonyl hydroxylamine (MSH) and sodium carbonate[19] allowed for the direct conversion of 1,3-diene 5a to the unprotected pyrrole 16 (30%), which was produced in higher yields via deprotection of 9a (Cs2CO3, MeOH, 84%). To conclude the synthetic scope, we explored the productivity of other chloramines to trigger the pyrrole formation of 5a. Commercially available chloramine-B monohydrate allowed for the construction of pyrrole 17a in 88% yield. When its p-nitrophenyl (chloramine-N), p-methoxyphenyl (chloramine-P) and methyl (chloramine-M) derivatives were applied, pyrroles 17b–d were also accessible in yields up to 75%.

Scheme 3

Synthetic Scope

Standard conditions: substrate (0.2 mmol), chloramine-T trihydrate (2.0 equiv), MeCN/HFIP (9/1, 0.1 M), 0.5–8 h and then m-CPBA (1.0 equiv), 23 °C, 1 h. See Section 4.1 in the Supporting Information for experimental and substrate specific details. Legend: (1) no addition of m-CPBA.

By changing to sterically encumbered 1,3-dienes such as 18, we were able to isolate the reactive sulfilimine 19 (61% yield, step A) under the standard reaction conditions (Scheme A). To our delight, thermal activation (toluene, reflux) allowed for the smooth initiation of the subsequent cascade to deliver pyrrole 20 in decent yield (76%, step B). When this two-step protocol was applied, trisubstituted pyrrole 21 (78% and 49%) and tetrasubstituted pyrrole 22 (61% and 99%) were formed. In addition, trisubstituted pyrrole 23 was obtained in good yields (62%), provided that benzonitrile was employed as the solvent.[20] As exemplified by 24, we found that the absence of an electron-withdrawing group (EWG) also allows for the isolation of its corresponding sulfilimines (99% yield, step A) under the standard reaction conditions. After this, thermal activation resulted in the formation of pyrrole 24 in quantitative yield. It is worth noting that, when sulfilimine 25 was subjected to thermal conditions (111 °C), a complete reaction was observed within 20 min. However, the main product was identified as the 2,5-dihydropyrrole 26 (44%) accompanied by small quantities of its cis-fused diastereomer (not shown, 10%) and pyrrole 27 (10%). Resubjecting 26 to refluxing toluene led to full conversion (28 h) to 27 in quantitative yield through the thermal release of methanethiol. Finally, sulfilimine 25 was directly converted into pyrrole 27 in 92% yield after an extended reaction time (44 h, step B).

Scheme 4

Mechanistic Investigations

See Section 4.3 in the Supporting Information for experimental details. Legend: (1) benzonitrile as the solvent, 191 °C, 1 h (step B); (2) yield determined by H NMR analysis using methyl phenyl sulfone as an internal standard.

Having investigated the synthetic scope, we conducted further experiments to gain insights into the mechanism of the pyrrole formation. Thereby, chloramine-T was shown to effectively trigger the elimination of methyl sulfide from 2,5-dihydropyrrole 10 at ambient temperatures (23 °C, Scheme B). This revealed that 2 equiv of chloramine-T is required for full conversion and to avoid formation of a mixture of pyrrole and 2,5-dihydropyrrole (compare Scheme , entry 7). In addition, 10 was obtained through a Pummerer-type activation of sulfoxide 28 in the presence of triflic anhydride (Tf2O) and p-toluenesulfonamide (TsNH2, Scheme C).[21] The lack of chloramine-T under these reaction conditions allowed for the selective formation of the 2,5-dihydropyrrole core without further elimination.

A second Pummerer-type reaction was demonstrated by the activation of sulfilimine 19 with oxalyl chloride (COCl)2.[22] On the basis of our previous work,[14]19 was rapidly converted into a trisubstituted furan bearing an unstable benzylic chloride. By telescoping the reaction in a one-pot fashion, the chloride was hydrolyzed (silver nitrate, acetone/water) to deliver furan 29 (39%). Finally, we adapted the 6π-electrocyclization/ring-contraction sequence for sulfoxide 28, resulting in the smooth formation of the 3,4-substituted furan 30 (52%, Scheme D).

In a continuation of our mechanistic studies, DFT calculations (B3LYP-D3/6-311++G(2d,2p)) in implicit acetonitrile shed light on the rapid conversion of 1,3-diene 5a to pyrrole 9a at ambient temperature (Scheme , highlighted in black). Sulfilimine A is initially generated from the reaction of 5a with chloramine-T, which is supported by the isolation of sulfilimines such as 19.[23] A thermal 6π-electrocyclization via TS-A with a barrier of ΔG⧧ = 13.5 kcal/mol results in the formation of 2,3-dihydrothiazine B. Facile ring-contraction through a 1,2-aza shift with a low activation energy (ΔG⧧ = 6.0 kcal/mol, TS-B) delivers the thermodynamically favored 2,5-dihydropyrrole 10 (ΔG = −39.5 kcal/mol), which could be isolated in the absence of chloramine-T (compare Scheme ). Since a second equivalent of chloramine-T was shown to rapidly promote the final aromatization step (compare Scheme B), we assume an exergonic sulfilimine formation with ΔΔG = −25.3 kcal/mol to yield C, which undergoes spontaneous elimination to give pyrrole 9a and sulfonamide 31.[24]

Scheme 5

Computational Studies

Proposed reaction mechanism as calculated with B3LYP-D3/6-311++G(2d,2p) in acetonitrile treated as the implicit solvent (see Section 6 in the Supporting Information for details). Relative Gibbs free energies at 298 K are given in kcal/mol, whereas the energies of the respective sulfilimines A, 19, and 32 are arbitrarily set to zero. The energetically most favorable pathway for 1,3-diene 5a to pyrrole 9a is highlighted in black. For comparison, the influences of sterics (blue, 19 → 20) and electronics (red, 32 → 24) were investigated.

On the basis of the isolation of several reactive intermediates (Scheme A), additional calculations were carried out to explain the kinetic hindrance. For the sterically encumbered sulfilimine 19 (highlighted in blue), we found only a slightly increased barrier for the 6π-electrocyclization (TS-19) in comparison to TS-A with ΔΔG⧧ = 2.7 kcal/mol. However, the formation of 2,3-dihydrothiazine D as well as the ring-contraction product TS-D is energetically increased (ΔΔG = 9.8 kcal/mol and ΔΔG⧧ = 13.8 kcal/mol) due to the rigidity of the annelated cyclohexene bearing the gem-dimethyl substitution pattern.[25] Intermediate D was found to kinetically favor the back reaction, a 6π-electrocyclic ring-opening, to regenerate 19 instead of undergoing ring-contraction via TS-D to 2,5-dihydropyrrole E (ΔΔG⧧ = 7.6 kcal/mol). Consequently, the product formation is kinetically suppressed at ambient temperature (23 °C), thus allowing for the isolation of 19. This is fully consistent with the thermal activation of 19 (111 °C, Scheme A) resulting in the formation of pyrrole 20 via intermediate E.

The lack of an EWG (highlighted in red) significantly increases the activation energy for the 6π-electrocyclization of sulfilimine 32 (ΔΔG⧧ = 10.1 kcal/mol, TS-32 vs TS-A).[26] In contrast to 2,3-dihydrothiazines B and D, the charge-separated intermediate F is preferentially formed, in which heterolytic cleavage of the S–N bond is observed. However, the ring-contraction barrier for TS-F is comparable to that of TS-B (ΔΔG⧧ = 1.3 kcal/mol), and the thermodynamics of 2,5-dihydropyrrole G are equal to those of 10. The similarity of the thermodynamic profiles (B → 10 and F → G) stands in sharp contrast to the sterically deactivated pathway of intermediate D to E. Alternative pathways for the formation of the 2,5-dihydropyrroles 10, E, and G have been investigated in detail (See Section 6 in the Supporting Information) but are energetically less favorable.

In summary, we have demonstrated the synthetic potential of 2,5-dihydrothiophene-derived sulfilimines to access a variety of polysubstituted pyrroles under mild reaction conditions. Both the experimental results and DFT calculations are fully consistent with a mechanism that involves a 6π-electrocyclization/ring-contraction sequence. Despite the omnipresence of pericyclic reactions in heterocyclic chemistry, electrocyclic reactions have been largely limited to the formation of six-membered heterocycles. The developed methodology fills that gap and expands the unique chemical space of electrocyclic reactions. Further studies toward related N-heterocycles are currently ongoing in our laboratories and will be reported in due course.