Understanding and Interrupting the Fischer Azaindolization Reaction

Abstract

Experimental and computational studies pertaining to the Fischer azaindolization reaction are reported. These studies explain why pyridylhydrazines are poorly reactive in Fischer indolization reactions, in addition to the origin of hydrazine substituent effects. Additionally, an interrupted variant of Fischer azaindolization methodology is disclosed, which provides a synthetic entryway into fused azaindoline scaffolds.

The Fischer indolization reaction, first described in 1883,¹ remains a powerful tool in modern chemical synthesis.² The transformation, which typically involves the acid-mediated reaction of aryl hydrazines with ketones or aldehydes to give indoles, has many applications in medicinal chemistry.² The Fischer indolization has also been used extensively in natural product synthesis, including several recent examples, to generate remarkable structural complexity.³

One deficiency associated with the Fischer indolization involves the variant in which pyridylhydrazines are employed to furnish azaindole products (a.k.a., Fischer azaindolization).⁴ This transformation was first tested over 100 years ago.⁵ It is now known, particularly via the studies of Parrick^4a,b and Robinson^4c that the azaindolization can proceed under extreme thermal conditions. For the corresponding acid-mediated process to take place, certain substituents on the hydrazine (e.g., methoxy, halides) can be used to improve reactivity, but the origins of such beneficial effects have remained unknown.^4d–g

The present study answers two key questions. (A) What factors govern the success or failure of the Fischer azaindolization reaction? (B) Could a general “interrupted” variant of the Fischer azaindolization⁶ be developed as a method to access fused azaindoline scaffolds? Although fused azaindoline frameworks are uncommon,⁷ the parent azaindole heterocycle serves as a bioisostere for indoles⁸ and is present in compounds of medicinal relevance. In fact, more than 74,000 bioactive azaindoles are known,⁹ including the currently marketed drugs Zelboraf (1, Figure 1) and Venclexta.¹⁰ Fused indolines are present in more than 22,000 bioactive molecules, more than 870 of which are natural products (e.g., 2).^6c,9,11 We now report mechanistic insights into the Fischer azaindolization reaction, the scope of the interrupted variant (3 + 4 → 5), and the use of this methodology to access azaanalogues of bioactive compounds.

Figure 1.

Azaindole drug Zelboraf (1), indoline-containing natural product aspidophylline A (2), and interrupted Fischer azaindolization methodology.

With the goal of understanding the factors that lead to success or failure of the Fischer azaindolization reaction, we performed a series of SCS-MP2 calculations.¹² We first investigated if the activation barrier for [3,3]-sigmatropic rearrangement varies significantly based on the aryl hydrazine employed. Thus, as summarized in Figure 2, the ΔG^‡ was calculated for this step under four scenarios:¹³ (a) the neutral phenylhydrazone, (b) protonation of the β-nitrogen of phenylhydrazone,¹⁴ (c) protonation of the β-nitrogen of 3-pyridylhydrazone, and (d) protonation of the β-nitrogen of methoxy-substituted 3-pyridylhydrazone. In all cases, we used propionaldehyde as the Fischer indolization reaction partner. For three of the scenarios (transition states 6, 7, and 9), the calculations were in good accord with experimental trends. However, for transition state 8 involving the experimentally problematic 3-pyridylhydrazine, the activation barrier was calculated to be reasonable (i.e., 13.5 kcal/mol). Thus, we sought an alternative explanation for the poor reactivity of 3-pyridylhydrazones in Fischer indolization reactions.

Figure 2.

Comparative calculations of the [3,3]-sigmatropic rearrangement transition states as a function of the arylhydrazine component.

We next considered if the pyridine nitrogen might be protonated (Figure 3). Indeed, protonation of this nitrogen is much preferred over hydrazone protonation, and pyridinium salt 10 is the global minimum of the pathway. To convert 10 to 11 requires 21.8 kcal/mol, leading to a high total barrier of 35.3 kcal/mol for the tautomerization and [3,3]-sigmatropic rearrangement (10 → 8). As noted earlier, it is known that introduction of a methoxy substituent on the pyridylhydrazine (e.g., see 9, Figure 2) leads to improved yields in experimental studies.^4d–g This has previously been rationalized by a “push/pull” effect^4d which, in turn, makes the [3,3]-sigmatropic rearrangement more favorable. Instead, we suggest that the methoxy group reduces the basicity of the pyridine nitrogen, thus rendering the proton transfer/tautomerization more feasible. We calculate that the tautomerization between the 2-methoxy analogues of 10 and 11 is only 14.9 kcal/mol (versus 21.8 kcal/mol for 10 to 11).¹² Consequently, the better performance in Fischer azaindolizations of methoxy-substituted pyridylhydrazines compared to the parent pyridylhydrazines can be attributed to the diminished basicity of the pyridine nitrogen in the former case.

Figure 3.

Calculations suggest that an unfavorable equilibrium between 10 and 11 leads to challenging Fischer azaindolization of 3-pyridylhydrazones.

We also explored the interrupted variant of the Fischer azaindolization to access fused azaindolines since only a single example of this transformation is available in the literature^6d and no methodology studies have been reported. To initiate our efforts in this area, we used methoxy-substituted pyridylhydrazine salt 12 (Table 1), due to its previously established favorable reactivity in Fischer azaindolization reactions. Following an initial survey of reaction conditions, mainly involving variation of solvent, acid sources, time, and temperature, we found that 12 underwent interrupted Fischer azaindolization with a variety of aldehyde surrogates to give fused 4-azaindoline products 13. Of note, the transformation introduces two stereocenters, one of which is quaternary. Results are depicted based on optimization of individual substrates. Upon treatment of 12 with lactol 14, furanoazaindoline 15 was obtained in 97% yield (entry 1). Alternate substituents on the lactol were tolerated, as demonstrated by the successful interrupted Fischer azaindolization of phenyl lactol 16 and allyl lactol 18 to furnish adducts 17 and 19, respectively (entries 2 and 3). The 6-membered lactol 20 was also deemed a suitable reaction partner, as demonstrated by the formation of 6,5,6-tricycle 21 (entry 4). In addition, we evaluated several hemiaminal substrates for the synthesis of azapyrrolidinoindolines. The use of methyl substituted N-Ts hemiaminal 22 led to 23 (entry 5), whereas employment of allyl derivative 24 gave azaindoline 25 (entry 6). Lastly, we evaluated carbamate-containing hemiaminal substrate 26, which furnished 27 in 82% yield (entry 7).

Table 1.

Scope of the Aldehyde Surrogate in the Interrupted Fischer Azaindolization^a

^aConditions unless otherwise stated: hydrazine 12 (1.5 equiv), aldehyde surrogate (1.0 equiv), solvent (0.05 M). Yields shown reflect the average of two isolation experiments.

Having established the tolerance of the interrupted Fischer azaindolization methodology toward variation in the aldehyde surrogate, we shifted our attention to assessing changes in the hydrazine component (Table 2). Halohydrazines 29 and 31 underwent interrupted Fischer azaindolization with lactol 14 to deliver 4-azaindoline products 30 and 32, respectively (entries 1 and 2). Interestingly, the use of methoxyhydrazine 33 led to azaindoline product 34, albeit in a lower yield (entry 3).¹⁵ Efforts to further optimize this transformation were unsuccessful. Isomeric hydrazines 35, 37, and 39 were also evaluated with the hopes of accessing 5-, 6-, and 7-substituted azaindolines. Whereas the interrupted Fischer azaindolization reaction failed to produce 5-azaindoline 36 (entry 4), the methodology furnished 6-azaindoline 38 (entry 5) and 7-azaindoline 40 (entry 6), albeit with a modest yield in the latter case. Finally, unsubstituted 3-pyridylhydrazine 41 was tested as a point of comparison (entry 7). Consistent with expectations based on the literature⁴ and our calculations (see Figure 3), the transformation proved problematic and gave 42 in a poor yield of 11%.¹⁶

Table 2.

Scope of Hydrazine Component and Correlation to Calculated Free Energies^a

^aConditions unless otherwise stated: hydrazine (1.5 equiv), aldehyde surrogate 14 (1.0 equiv), solvent (0.05 M). Yields shown reflect the average of two isolation experiments.

^bFree energies (kcal/mol) for [3,3]-sigmatropic rearrangement (ΔG^‡[3,3]) and enehydrazine formation (ΔG[H⁺]) were obtained using SCS-MP2 calculations with propionaldehyde as a surrogate aldehyde model under acidic reaction conditions.

To complement these experimental studies, calculations were performed for the reaction of each hydrazine depicted in Table 2, with propionaldehyde as a model aldehyde under acidic conditions.¹² Earlier, we hypothesized that the success or failure of the Fischer azaindolization methodology hinges on the feasibility of enehydrazine formation, rather than [3,3]-sigmatropic rearrangement. Consistent with this notion, we found the computed activation barriers for [3,3]-sigmatropic rearrangement (ΔG^‡[3,3]) were uniformly accessible and ranged from 13.4 to 16.6 kcal/mol. On the other hand, the calculated tautomerization equilibria (ΔG[H⁺]) varied considerably between 11.5 and 23.7 kcal/mol. In cases where the calculated ΔG[H⁺] is high, the reaction is experimentally problematic (entries 4 and 7). In the cases of entries 3 and 6, the ΔG[H⁺] is slightly more favorable leading to modest yields of product. However, if ΔG[H⁺] is sufficiently low (i.e., less than 15 kcal/mol), the reaction proceeds more efficiently (Table 2, entries 1, 2, 5, and Table 1). Overall, hydrazines with poorly basic pyridine nitrogens had the smallest ΔG[H⁺].

The interrupted Fischer azaindolization methodology can be used to synthesize new aza-analogues of bioactive molecules (Figure 4). Treatment of benzyloxyhydrazine 43 with lactol 14 under acidic conditions delivered furanoazaindoline 44 in 81% yield. Using a straightforward 3-step sequence, 44 was elaborated to carbamate 45, a new analogue of the potent acetylcholinesterase inhibitor phensvenine.¹⁷ Switching to a significantly more complex system, we performed the interrupted Fischer azaindolization of hydrazine 12 and known^3j ketolactone 46. Following the acid-mediated rearrangement and subsequent hydrolysis in the same pot, a straightforward methylation gave pentacycle 47. In turn, azaindoline 47 was converted in two steps to 48, an analogue of the anticancer agent aspidophylline A.¹¹

Figure 4.

Synthesis of aza-analogues.

We have performed an experimental and computational study of the Fischer azaindolization reaction, related to a transformation first attempted more than 100 years ago. Calculations were used to explain the difficulty in employing pyridylhydrazines in Fischer indolizations, in addition to the origin of hydrazine substituent effects. Rather than the [3,3]-sigmatropic rearrangement step being the single determining factor, we find that reactions suffer if the pyridine nitrogen is too basic, making the tautomerization step prohibitively challenging. We have also developed an interrupted Fischer azaindolization methodology, which provides a synthetic entryway into fused azaindoline scaffolds. The syntheses of new aza-analogues of phensvenine and aspidophylline A underscore the potential of this methodology to deliver new aza-derivatives of medicinally privileged fused indoline-containing compounds.