Diazacyclobutenes as reactive intermediates: the cascade cyclization of thiolated ene-ynes and azodicarboxylates to provide tetrahydroindoles

Abstract

We describe a new one-pot synthesis to access richly functionalized tetrahydroindoles and related fused pyrroles. The key to this transformation is a robust cascade cyclization between ene-yne sulfides and azodicarboxylates involving diazacyclobutene formation primed for rapid electrocyclic ring opening to an α-iminothioimidate intermediate followed by cyclization and aromatization to yield the N-heterocycles in good yields. This tactic represents a significant departure from conventional approaches to the tetrahydroindole motif, which typically involve either staging the 140-year-old Paal-Knorr cyclization or leveraging transition metal catalysis. Further synthetic applications demonstrated that this method can be used for the late-stage functionalization of complex bioactive molecules or natural product-like structures in an efficient manner. The mechanism of the cyclization event leading to the heterocyclic products was probed computationally and is more consistent with a polar process instead of a concerted aza-Nazarov cyclization. The products can be leveraged in a mechanistically interesting desulfination reaction or converted to highly substituted indoles under mild conditions.

Graphical Abstract

INTRODUCTION

Tetrahydroindoles – or 2,3-fused pyrroles – are partially hydrogenated derivatives of the naturally abundant indole heterocycle, with significant applications in the synthesis of natural alkaloids and biologically active molecules (Scheme 1).^1-7 For instance, the tetrahydroindole derivatives molindone (1) and apaziquone (2) are used to treat psychosis and superficial bladder cancer, respectively.^8,9 Additionally, tetrahydroindole derivatives function as Src tyrosine kinase inhibitors (3),¹⁰ dopaminergic agents (4),¹⁰ as well as antibiotic and anti-inflammatory agents.^10-14

Scheme 1.

Tetrahydroindoles and synthetic tactics to access them. Shaded box: a mechanistically distinct approach.

Despite their wide-ranging applications, the synthesis of highly functionalized tetrahydroindoles remains a significant challenge, and synthetic methods for assembling the motif have been well reviewed.^15-19 Notably, the vast majority of methods to prepare the motif fall into one of two frameworks – strategies that promote the union of starting materials to stage the 140 year old Paal-Knorr (P-K) condensation (Scheme 1, Tactic A),^20,21 or approaches that leverage transition metal catalysis to orchestrate the formation of the heterocycle from various precursors (Tactic B).¹⁵ The work of Kartika and co-workers (eq. 1) serves as a recent, illustrative example of the former strategy, wherein they creatively staged the P-K intermediate 8 by means of the reaction of silyl enol ether nucleophiles (6) with cyclic silyloxyallyl cation precursor 5 under protic catalysis.²² Collapse of the initially formed silyl enol ether intermediate 7 led to tetrahydroindole 9 in 67% yield via P-K condensation with aniline. Similarly, the well-known Trofimov reaction^23-25 also falls within this framework, wherein the reaction of oximes (10) with electron-deficient alkynes (11) results in an intermediate 12 that undergoes [3,3]-sigmatropic rearrangement to stage an intermediate on the P-K pathway (see 13). This approach is nicely illustrated by work from Sheradsky to generate tetrahydroindole diester 14 in 45% yield,²⁶ which interestingly predates the work of Trofimov. (Perhaps, the transformation is more aptly the Sheradsky-Trofimov reaction?)

In the context of the latter strategy relying on transition metal catalysis (Scheme 1, Tactic B),¹⁵ the specific operative mechanisms differ depending on the transition metal catalyst that is deployed. The Beller lab deployed a ruthenium catalyst to effect the dehydrogenative condensation of cyclohexanone 15 with homobenzylamine 16 and glycolaldehyde 19 (arising from ethylene glycol 17) to intercept an intermediate on the P-K pathway, ultimately providing terahydroindole 22 in 82% yield (eq. 3).²⁷ Essentially, this approach is similar to the metal-free tactics to stage the P-K transformation (Tactic A). Conversely, other metal-mediated approaches diverge from the P-K staging strategy. Kurkin et al. deployed aminopropargylic alcohols (e.g. 23) in a Sonogoshira coupling (see 24) and 5-endo-dig cyclization to generate tetrahydroindole 25 in 75% yield (eq. 4).²⁸ As a third example, the Davies group presented a one-pot approach that executed the sequential azide-alkyne click reaction to provide N-tosyl triazole 27, which served as a substrate for Rh-carbene formation (28) and 4π-electrocyclization to provide tetrahydroindole 30 in 90% yield via putative intermediate 29 (eq. 5).²⁹

Scheme 5.

Oxidation of tetrahydroindoles to highly substituted indoles and related compounds.

In this manuscript, we describe a strategy to access fully-substituted bearing tetrahydorindoles bearing heteroatom substitutents. This mechanistically distinct strategy was born out of our interest in the synthesis and reactivity of the exceptionally rare diazacyclobutene (DCB) heterocycle (e.g. 33).^30-33 Specifically, our prior work that explored the thermal lability of monocyclic DCBs such as 33 demonstrated their facile electrocyclic ring-opening to provide α-iminothioimidate products (e.g., 34).³¹ This observation led us to wonder whether the formation of reengineered, labile DCBs bearing additional π-systems (i.e. DCB 33 arising from an azodicarboxylate (32) and ene-yne sulfide 31) might provide a platform to further engage the resultant α-iminothioimidate product for additional chemistry. Thus we envisioned that conjugated α-iminothioimidates such as 34 might undergo a thermodynamically favorable cyclization to provide richly functionalized tetrahydroindole products (35) either by means of a polar 1,4-addition of the thioimidate onto the α,β-unsaturated imine motif, or perhaps by means of an aza-Nazarov cyclization^34-37 followed by aromatization in either case. The realization of this mechanistically unique strategy for the synthesis of fully-substituted tetrahydroindoles and similar fused pyrroles is described herein.

RESULTS AND DISCUSSION

Reaction Development.

Our preliminary reaction development utilized cyclohexenyl-1-ethynyl ethyl sulfane (31a) as a model substrate (Table 1). As shown in entries 1-4, we subjected 31a to various azodicarboxylates in refluxing toluene. Gratifyingly, reacting 31a with bis(2,2,2-trichloroethyl)azodicarboxylate (TCEAD, Troc-N=N-Troc) in refluxing toluene for 24 h produced the desired richly-functionalized 4,5,6,7-tetrahydroindole (35a) in a 79% isolated yield (entry 5). Given the potential for the cyclization of the α-iminothioimidate (34, Scheme 1) to proceed by the aza-Nazarov cyclization, we tested several known reaction conditions^35,38-43 for this transformation but achieved no success (entries 9-18). While we found that using 0.5 mol equiv of silver triflate in refluxing toluene or acetonitrile gave the desired product in 75% and 77% isolated yields, the use of this catalyst did not offer any appreciable benefits over the metal-free approach. Thus, we chose the conditions from entry 5 as optimal and elected to explore the substrate scope of the reaction. It is interesting that other azodicarboxylates did not lead to the isolation of 35. In some cases, no reaction was observed and in other cases, we observed the formation of the α-iminothioimidate intermediate (see 34, Scheme 1), suggesting that the electron deficient trichloroethylcarbamate motif may aid in the final cyclization step to provide tetrahydroindole 35.

Table 1.

Optimization of tetrahydroindole synthesis.

^aenynyl sulfide 31a (0.21 mmol, 1.3 equiv), azodicarboxylate (0.16 mmol,1 equiv) and catalyst (10 mol %) were combined in solvent (2 mL) and refluxed in a sealed reaction flask under argon for 24 h.

^bIsolated yields.

^cCatalyst (50 mol %), no product was observed when less than 50 mol % was used and starting material decomposed when an equimolar amount was used.

^dNo product was detected when the reaction was carried out in PhMe, 1,2-DCE and MeOH at a temperature below reflux.

^eNo product was observed at higher catalytic loading (50 mol %) at temperatures between −40 °C and 25 °C.

^fNo product was detected in DCM, PhMe and 1,2-DCE at temperatures between 0 °C and 25 °C.

Substrate Scope.

After optimizing the reaction conditions for the synthesis of tetrahydroindole 35a, we investigated the outcome of the reaction when the sulfur substituent was varied, and when the chalcogen was changed to selenium. We synthesized ethynylcyclohexene sulfides with various S-alkyl groups (31a-g) in moderate to excellent yields from commercially available 1-ethynylcyclohexene in a single synthetic step (see Supporting Information for details). Additionally, we prepared ethyl ethynylcyclohexene selenide (31h). With these substrates in hand, we explored the scope of their transformation with TCEAD using the optimized conditions from Table 1, entry 5. (Note that TCEAD was synthesized in two steps using a method from the literature (see Scheme S9 in Supporting Information for details).⁴⁴)

We were pleased to find that reactions of various ethynylcyclohexene sulfides and selenides yielded the corresponding tetrahydroindoles 35a-h in moderate to excellent yields (Scheme 2). Specifically, ethynylcyclohexene sulfides with shorter S-alkyl chains (31a-d), such as methyl, ethyl, n-propyl, and n-butyl, were successfully converted into their respective tetrahydroindoles 35a-d in 72-79% yields. Cyclization of the S-methyl analog 31b provided tetrahydroindole 35b in 64% yield on a gram scale. Substrates with longer S-alkyl chains, such as n-pentyl (31e) and n-octyl (31f), produced tetrahydroindoles 35e and 35f in 77% and 83% yields, respectively. The S-benzyl ethynylcyclohexene sulfide 31g provided a lower 29% yield of tetrahydroindole 35g. Furthermore, the reaction of ethynylcyclohexene selenide (31h) with TCEAD produced the desired tetrahydroindole 35h in a 31% yield. It is noteworthy that the isolated products are frequently crystalline, and thus typically amenable to X-ray analysis (see 35c and ESI).

Scheme 2.

Cyclization of ethynylcyclohexene sulfides and selenides.

We next aimed to broaden the substrate scope of ene-yne sulfides reacting with TCEAD to produce a diverse range of tetrahydroindoles and related fused pyrroles. One challenge was the limited commercial availability of other ene-yne starting materials. Fortunately, we were able to synthesize functionalized ene-yne sulfides from cyclic ketones through a few synthetic modifications based on existing methods in the literature. The ene-yne sulfides 31i-t were typically prepared in excellent yields (70-95%) from cyclic ketones via vinyl triflate formation, followed by Kumada-type coupling with ethynylmagnesium bromide (HCCMgBr) to yield the corresponding ene-ynes, and finally, copper(I)-catalyzed thiolation. Some substrates were prepared from ketones by means of carbonyl alkylation with HCCMgBr, followed by copper(I)-catalyzed thiolation and subsequent elimination of the tertiary propargyl alcohol (see Supporting Information for details).

Gratifyingly, the reactions of ene-yne sulfides 31i-t with TCEAD yielded the corresponding tetrahydroindole or related fused pyrrole derivatives in moderate to excellent yields (Scheme 3). Notably, 4,5,6,7-tetrahydroindole products bearing 6-methyl (35i), 6-tert-butyl (35j), and 6-phenyl (35k) groups were formed in good yields ranging between 70-85%. The pyrano (35l) and spiro-dioxolane (35m) analogs were isolated with yields of 74% and 76%, respectively.

Scheme 3.

Substrate scope of tetrahydroindole/fused pyrrole synthesis.

The tetrahydrocyclopenta[b]pyrrole (35n) and tetrahydrocyclohepta[b]pyrrole (35o) products were produced in isolated yields of 65% and 62%, respectively. Ene-yne sulfide substrates originating from α-tetralones provided the corresponding fused pyrrole derivatives 35p-r in isolated yields of 65-72%. Finally, the pyrrole derivatives 35s and 35t, originating from ene-yne sulfides derived from indanones, were isolated in yields of 52% and 58%, respectively. In the latter case, the direct vinyl amination product 36 was isolated in 31% yield as a side product of the reaction, perhaps accounting for the lower yields resulting from these two substrates. This product likely arose from the direct attack of the terminal alkene carbon of the ene-yne sulfide onto the azodicarboxylate followed by proton elimination to restore the conjugated π system.

Further synthetic efforts showed that this method can facilitate late-stage functionalization of complex bioactive molecules or natural product-like structures (Scheme 3, shaded boxes). The ketones resident in the terpenes (+)-pulegone and (+)-camphor were converted into ene-yne sulfides suitable for cyclization over a four-step sequence (see ESI for details). These materials underwent cyclization with TCEAD to provide the fully substituted pyrrole products 35u and 35v in 37% and 56% overall yields over 5 steps, respectively. The key cyclization event proceeded in 72% yield leading to 35u and 82% yield leading to 35v. Similarly, a three-step sequence (see ESI for details) originating from benzylated trans-dehydroandrosterone was able to manipulate the carbonyl functionality into an ene-yne sulfide which readily underwent cyclization (72% yield) with TCEAD to provide product 35w in 24% yield over 5 steps from the parent steroid. Finally, 5α-cholestan-3-one was converted to the fully substituted steroidal pyrrole 35x in 28% overall yield over 5 steps culminating in a 67% yield for the key cyclization event.

Mechanistic Considerations.

While we have previously interrogated the course of the reaction leading to the diazacyclobutene intermediate (see 33, Scheme 1) and its subsequent ring opening to the α-iminothioimidate 34 both spectroscopically and computationally,³¹ we sought to probe the nature of the cyclization event leading to the final tetrahydroindole product (34-H⁺ → 35-H⁺). A simple model system with R = CH₃ and CO₂CH₃ groups on nitrogen was used. This reaction was predicted (using the M06-2X/6-31+G(d,p) method)^45-48 to be exergonic by 35 kcal/mol and have a barrier of only 5 kcal/mol (from the productive conformer of reactant). The main question of interest, however, was whether or not this reaction is pericyclic. The geometry of the core of the transition structure (Figure 1) is helical, consistent with a conrotatory process, but progress along the intrinsic reaction coordinate (IRC)^49,50 appears more consistent with attack of the nitrogen lone pair rather than the imine π-bond. In addition, the imine π-bond in the transition structure is not particularly long (1.29 Å compared to 1.35 Å for the other C─N bond), and the C─C bond to which it is attached is not particularly short (1.5 Å compared to 1.4 Å for the other two C─C bonds), consistent with relatively little delocalization over that substructure. We also computed the $\overline{NICSzz}$,^51,52 the averaged NICS (nucleus independent chemical shift)^53-55 perpendicular to the approximate plane of the five atoms supporting the rearranging electrons, along the IRC. Pericyclic reactions are expected to be associated with cyclic delocalization of electron density somewhere along their reaction coordinates (often near their transition structures), which leads to a local magnetic field and negative NICS values. While there is a dip in $\overline{NICSzz}$ near the transition structure (Figure 1, right red), the magnitude of this dip is much less than found for typical pericyclic reactions (i.e. approximately −2 ppm here compared to approximately −25 ppm for the parent Cope reaction, for example). In addition, multicenter bond orders (nMCBO)^56,57 were computed along the reaction coordinate (Figure 1, right, green). For a typical pericyclic reaction, nMCBO usually peaks near the transition structure. While that occurs here, there is little drop off, since the delocalized π system of the pyrrole is formed, thus complicating the use of nMCBO to characterize pericyclicity or its absence here. The onset of pyrrole aromaticity is also visible in the $\overline{NICSzz}$ decrease (after a small peak) following the transition structure. Overall, the combined theoretical evidence paints a picture of a reaction that is best described as a nucleophilic attack of an imine lone pair on an α,β-unsaturated iminium ion, but one that benefits from at least a small degree of cyclic delocalization.

Figure 1.

LEFT: Computed geometry of transition structure for 34 → 35 reaction (selected distances in A). RIGHT: IRC plot (blue) along with computed $\overline{NICSzz}$ (red) and nMCBO (green) values.

Further Synthetic Manipulations.

We next turned our attention to evaluating a few synthetic manipulations of the tetrahydroindole products that arose from our study. First, we were interested in whether the sulfur atom in the products could be positioned for removal. Compounds 35a and 35b were successfully transformed into sulfoxide derivatives 37a and 37b in isolated yields of 92% and 85%, respectively, upon treatment with oxone (Scheme 4, shaded box, eq. 7). Curiously, it was observed that 37a and 37b were rather labile, slowly undergoing reductive desulfinylation, resulting in a small amount of tetrahydroindole 38 on standing in CDCl₃ at room temperature. Leveraging this observation we demonstrated that 38 could be obtained in excellent yield by refluxing 37a or 37b in acetonitrile for just 30 minutes (Scheme 4, eq. 7). This result is particularly noteworthy, as reductive C–SOR cleavage typically requires β-elimination^58-60 or the use of reducing agents like alkyl metals,^61-64 organosilane,⁶⁵ and Grignard reagents.^66-68

Scheme 4.

A reductive desulfination reaction and mechanistic study.

We considered three possible pathways that might lead to loss of the sulfoxide to provide the reduced tetrahydroindole product. One might imagine a 1,3-hydrogen migration event from 35a or 34b that would provide 38 along with zwitterion 39 (eq. 8). Alternatively, a 1,4-hydrogen migration from substrate 34a might provide 38 along with ethylene and sulfur monoxide (eq. 9), although this pathway would be unavailable for substrate 34b. Finally, 35a might equilibrate to imine tautomer 41, which could then undergo nucleophilic displacement (from solvent or adventitious water perhaps) to provide the N-anion of 38 along with the displacement product 42 (eq. 10). We probed these three potential pathways using deuterium labelling studies. The lack of deuterium incorporation at C2 of product 38 when reacting the penta-deuterated sulfoxide substrate 39, provided strong evidence against a 1,3-migration event (eq. 8). Similarly, the 1,4-migration pathway is unlikely based on the lack of label incorporation in 38 when the tri-deutero substrate 40 was deployed (eq. 9).

At this point, we turned our attention to probing the possibility of the proposed nucleophilic displacement pathway. First, we conducted the desulfination reaction of 35a in the presence of D₂O as a potential nucleophile (eq. 11). We hypothesized that this treatment might result in D-labelling at C2, particularly if the tautomerization event leading to 41 was facilitated by intermolecular protonation. Additionally, one might expect deuterium incorporation on the nitrogen atom resident on the C3 carbamate due to protonation of the anion of 38 after nucleophilic displacement. In the experiment, we observed complete deuteration at the C3 nitrogen atom prior to work up and isolation of product 43. Interestingly, we also observed complete deuteration at the C7 methylene of 43, which we propose might arise from deprotonation/deuteration of C7 by means of conjugated anion 45. After work-up and chromatographic purification, 43 was isolated lacking deuterium label at the C3 nitrogen presumably due to H/D exchange during work-up/purification. This experiment led to three key conclusions. First, the incorporation of label at the carbamate nitrogen at C3 is consistent with the proposed nucleophilic displacement pathway. Second, the lack of label incorporation at C2 in 38 strongly suggests that the formation of proposed imine intermediate 41 is likely mediated by a tautomerization event that relies on an intramolecular proton transfer. Third, the incorporation of label at C7 indicates that the protons at this position are labile, but – while interesting – this fact is likely unrelated to the desulfination event. Based on the results of this labelling experiment, we next sought to support this hypothesis with an additional experiment. Thus, we treated sulfoxide 35a with benzyl alcohol as a potential nucleophile to promote desulfination. In this experiment, tetrahydroindole 38 was isolated in 56% yield along with benzyl sulfinate 46 in 33% yield, providing further evidence for the possibility of the nucleophilic displacement pathway for desulfination. Questions remain, however, regarding the specific identity of the nucleophile component in the original experiment (eq. 7). One might imagine that adventitious water may serve as the necessary nucleophile, or perhaps the mildly nucleophilic acetonitrile solvent may play this role.

In a final series of experiments, we also explored whether the 4,5,6,7-tetrahydroindoles could be aromatized to yield highly substituted indole derivatives. To our satisfaction, treatment of the tetrahydroindoles with 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ) produced the desired indole derivatives 47a-k in 70-95% yields (Scheme 5).

CONCLUSION

In summary, we have developed a unique, one-pot strategy for constructing highly functionalized tetrahydroindoles, tetrahydrocyclopenta[b]-, and hepta[b]pyrroles that relies on a fundamental mechanistic departure for typical strategies to access these motifs. The key to this transformation is a robust cascade cyclization between ene-yne sulfides and azodicarboxylates, enabling access to these N-heterocycles in good yields through the intermediacy of the rare diazacyclobutene heterocycle. The cyclization is robust, and can be applied to the late-stage modification of bioactive molecules and natural product-like scaffolds. The lability of the initially formed monocyclic diazacyclobutene intermediate allows for an electrocyclic ring opening event that provides a conjugated α-iminothioimidate, which undergoes a thermodynamically feasible cyclization/aromatization process to arrive at the tetrahydroindole products. Computational evidence suggests that the cyclization event leading to the tetrahydroindole from the α-iminothioimidate intermediate is more akin to a nucleophile/electrophile cyclization than a pericyclic reaction, albeit one that exhibits of a modicum of concertedness. Additionally, an interesting desulfination reaction was uncovered that allows for the excision of the thioether functional group – which is critical for the promotion of the cyclization – from the tetrahydroindole product. Mechanistic studies suggest that this desulfination reaction may occur by means of a nucleophilic displacement. Finally, we explored the oxidation of the initial products to produce highly functionalized indole products. Ongoing work in our laboratory aims to explore this unique transformation to enhance molecular complexity and access structurally related heterocycles.

Supplementary Material

Supporting Information: General experimental methods, synthesis of substrates and azodicarboxylate reagents, tetrahydroindole, indole, and sulfoxide products, deuterium labelling studies, X-ray crystallography data, computational results, and copies of NMR spectra of new compounds.

Data Availability Statement:

The data underlying this study are available in the published article and its Supporting Information.