Regio- and Diastereoselective 1,3-Dipolar Cycloadditions of 1,2,4-Triazin-1-ium Ylides: a Straightforward Synthetic Route to Polysubstituted Pyrrolo[2,1-f][1,2,4]triazines.

A synthetic strategy to pyrrolo[2,1-f][1,2,4]triazines is reported. We show that various synthetically easily accessible 1,2,4-triazines can be efficiently alkylated under mild conditions to provide the corresponding 1-alkyl-1,2,4-triazinium salts. These bench-stable salts serve as precursors to triazinium ylides, which react in 1,3-dipolar cycloadditions with electron-poor dipolarophiles to yield polysubstituted pyrrolotriazines in a single step.

Introduction

The bridgehead pyrrolo[2,1-f][1,2,4]triazine heterocycle is a privileged scaffold found in numerous pharmaceutically important substances. The biological activities of compounds containing the pyrrolotriazine core include inhibition of kinases, anticancer activities, and potent antiviral effects.[1−8]

The first synthesis of the heterocycle based on addition/fragmentation of 1,2,4-triazines (hereafter denoted as triazines) with dimethyl acetylenedicarboxylate (DMAD) was reported by Neuenhoeffer in 1977.[9] Two years later, Migliara and co-workers reported synthesis of the pyrrolotriazine core via an acid-mediated cyclization of semicarbazone onto a pendant α-ketoester, followed by a base-promoted cyclization and decarboxylation.[10] These pioneering studies paved the way for later re-emergence of the heterocycle as a ‘purine-like’ scaffold introduced into a series of C-nucleoside analogues.[11−13] The discovery that C-4-substituted pyrrolotriazines are potent ATP-competitive kinase inhibitors[14] further fueled the research interest in these compounds and led to numerous candidates in late stages of clinical development and to approved drugs (Figure ).[1,7,15−20]

Figure 1

Examples of biologically active pyrrolo[2,1-f][1,2,4]triazines (and their respective targets).

A typical synthetic route leading to pyrrolotriazines follows a number of steps depicted in Scheme A. Variations in these steps and late-stage modification of the heterocycle by, for example, cross-coupling reactions can provide access to diverse derivatives with a wide range of potential biological activities.[21−23]

Scheme 1

(A) Example of a Typical Synthetic Route to Pyrrolo[2,1-f][1,2,4]triazines. (B) Single-Step Construction of Pyrrolotriazines via 1,3-Dipolar Cycloadditions

Despite the pharmacological importance of pyrrolotriazines reflected in numerous literature reports and patent applications, virtually no alternative synthetic procedures to the scaffold exist. Even the most recent examples exploit reactions that have been originally developed decades ago.[24] To the best of our knowledge, the only alternative route to pyrrolotriazines, briefly examined in the early 90s, was based on cycloadditions of 1-alkyl-1,2,4-triazinium ylides generated in situ from the respective triazinium precursors under basic conditions.[25] Despite its great synthetic potential, this methodology has remained largely unexplored and neglected.

Herein, we show that various, previously unknown pyrrolotriazines can be prepared in a single step from readily accessible 1-alkyl-1,2,4-triazinium salts via 1,3-dipolar cycloaddition (DCA) of the in situ generated triazinium ylides with electron-poor dipolarophiles (Scheme B).

Results and Discussion

Substituted 1,2,4-triazines are synthetically readily accessible heterocycles. Depending on the substitution pattern, their preparation typically includes a single and usually high-yielding condensation step starting from various commercial glyoxal hydrates and S-alkyl isothiosemicarbazides in the presence of sodium bicarbonate. More recent synthetic routes include regioselective reactions of N-tosyl hydrazones with aziridines,[26] carbene N–H insertion of hydrazides,[27] O–H insertion of rhodium-azavinylcarbenes,[28] Zn-catalyzed hydrohydrazination of propargylamides,[29] cycloaddition of tetrazines,[30] or domino annulation reactions.[31] There are numerous possibilities to further derivatize and decorate the basic heterocyclic core. For example, simple 3-substituted triazines are susceptible to an easy nucleophilic attack at position 5 followed by position 6.[32,33] The S-alkyl substituents can be utilized in cross-coupling reactions[33,34] or after oxidation, used in nucleophilic substitutions.[35,36] Therefore, the substrate scope of the starting triazines is immense, opening an easy access to diverse 1-alkyl-1,2,4-triazinium salts.

The first goal of our study was to optimize the alkylation of the triazine core. For this purpose, we used 3-phenyltriazine 1a(34) and the commercial 3-methylthiotriazine and explored different alkylating agents and reaction conditions (Table S1). These experiments revealed that the alkylation with benzyl bromide and dimethyl sulfate is sluggish, while methyl iodide proved unreactive. In contrast, the use of Meerwein’s salt (Me3O+BF4–) led to clean formation of the N1-alkylated product.[25] Due to difficulties in handling this sensitive compound, we decided to explore triflates as more convenient alkylating agents. To our delight, the alkylation with methyl or ethyl triflate was successful. Interestingly, the simple 3-aryltriazines are alkylated exclusively at position 1, while electronically richer derivatives, such as the commercial 3-methylthiotriazine, led under the same conditions to the formation of products alkylated at both, N1 and N2, as determined by heteronuclear multiple bond correlation nuclear magnetic resonance (NMR) experiments.

The observed differences in the preferred position for the alkylation can be explained, at least in part, by differences in the electrostatic potential of differently substituted triazines in combination with steric effects of the substituent at position 3 (Figure S1).

We also studied the influence of various solvents on the alkylation reaction. Anhydrous toluene was found superior over other solvents with the minor N2 isomer being formed in 15% yield using the methyl triflate and in only 10% using the ethyl triflate. We found that the two N-alkylated isomers can be separated by reversed phase column chromatography. However, the N2 isomer does not interfere with the subsequent DCA step and can be easily removed during purification of the final cycloaddition product. Therefore, we used the crude mixture of the triazinium isomers in all subsequent experiments. Under the optimized conditions, we next prepared a series of triazinium triflates 2a-o-Me(Et) in excellent yields (Scheme ).

Scheme 2

Scope of Triazinium Compounds

Standard conditions: triazine 1a–1o (2 mmol), triflate (2.4 mmol), anhydrous toluene (20 mL), 0 °C to RT, under argon. Note: all compounds are triflate salts.

The alkylation is basically quantitative, and the formed triazinium salt simply precipitates from the reaction mixture in most cases. The reaction can be conveniently performed on a larger scale (hundreds of milligrams), and the 1-alkyltriazinium triflates are surprisingly stable when stored in the crystalline form and even in solution (Figure S2).

With the set of triaziniums in hand, we proceeded to optimize the 1,3-dipolar cycloaddition reaction using DMAD as the dipolarophile and 3-phenyltriazinium methyl triflates 2a-Me and 2b-Me as model substrates. We experimented with different bases, solvents, temperatures, and orders of reagent addition. Unfortunately, all these efforts led only to decomposition of the starting materials and formation of the cycloaddition product in less than 10% yield in the best case.

In contrast to these initial results, we found that the cycloaddition of compounds bearing additional substituents at position 5 is viable. In this case, the reaction with DMAD performed in tetrahydrofuran (THF) in the presence of N,N-diisopropylethylamine (DIPEA) afforded the desired cycloaddition products 3d-m-H(Me) in good to very good yields (up to 84%). The reaction proceeded smoothly, and a wide range of triaziniums bearing various aryl substituents at position 5 were tolerated (Scheme ). In general, cycloadditions of ethyl triazinium ylides gave higher yields when compared to the methylated analogues. The final pyrrolotriazines form in the reaction as the fully oxidized products especially from the methyl triaziniums. If this was not the case, the dihydro intermediates were converted to the oxidized products by simply opening the reaction flask to the air. By following the reaction progress on high-performance liquid chromatography mass spectrometry, we observed that a small portion of the triazinium compounds undergoes dequarternization at the nitrogen atom to yield the starting triazine. Although the exact mechanism remains unclear, similar dealkylation of N-alkyl triazinium salts has been observed previously.[37] Interestingly, the pyrrolotriazines are fluorescent compounds when irradiated at 365 nm using a standard handheld UV lamp. This property could be potentially exploited in preparation of new fluorophores based on this heterocyclic core.

Scheme 3

1,3-Dipolar Cycloaddition of Triazinium Ylides with DMAD

Standard conditions: triazinium (0.5 mmol), DMAD (2 mmol), THF (10 mL), DIPEA (1.5 mmol), 0 °C to RT, 5 h.

Encouraged by these results, we decided to explore the possibility of using other dipolarophiles in the reaction (Scheme ).

Scheme 4

Scope of Dipolarophiles

Hence, reaction of 1-methyl-5-phenyl-1,2,4-triazinium triflate 2g-Me with methyl propiolate performed in THF using DIPEA as the base yielded after optimization the desired pyrrolotriazine 4g-H as a single regioisomer in 38% yield. The observed formation of the single regioisomer in the reaction is substantiated by the 5.0 kcal/mol lower energy of the corresponding transition state as evident from density functional theory (DFT) calculations (Figure S3). Even though the cycloaddition was in this case complicated by the competing reaction of the DIPEA base with the electron-poor triple bond,[38] this approach opens the possibility to generate additional pyrrolotriazines in a regioselective manner.

To further expand the scope of dipolarophiles, we performed the reaction with acrylonitrile, N-ethylmaleimide, and N-propargylmaleimide. The reaction with acrylonitrile yielded two diastereomers of a single regioisomer 5g-H in a 2 to 1 ratio and in 54% overall yield. The reaction with maleimide proceeded smoothly and afforded two diastereomeric products 6g-H in an 8 to 1 ratio and 88% overall yield. Based on DFT calculations, the transition state structure leading to the major product has lower calculated energy by 0.9 kcal/mol (Figure S4). Interestingly, only the minor isomer slowly oxidizes to the corresponding pyrrolotriazine. Cycloaddition experiments with N-propargylmaleimide and triaziniums 2g-Me and 2j-Me confirmed the diastereoselectivity of the transformation and yielded pyrrolotriazines 7g-H and 7j-H in 59% (significant dequarternization) and 95% overall yield, respectively.

Finally, to demonstrate the possibility of further derivatization of the pyrrolotriazine scaffold, we performed pilot coupling experiments with cycloadduct 3i-Me (Scheme ). The thiomethyl substituent at position 3 was successfully utilized in the Liebeskind–Srogl cross-coupling reaction with phenylboronic acid under standard conditions. The corresponding 3-phenyl pyrrolotriazine isolated in 54% yield was used in the next Suzuki–Miyaura cross-coupling reaction, which gave the desired heterocyclic product 8i-Me in excellent 90% isolated yield.

Scheme 5

Cross-Coupling Modifications

Conditions: 1. Phenylboronic acid (2.5 equiv), copper(I) thiophene-2-carboxylate (2.2 equiv), Pd(PPh3)4 (10 mol %), 1,4-dioxane, 95 °C. 2. Phenylboronic acid (2.0 equiv), PdCl2(dppf)·DCM (10 mol %), K2CO3 (2.0 equiv), 1,4-dioxane/H2O = 3/1, 100 °C.

Conclusions

In conclusion, we show that 1,3-dipolar cycloadditions of electron-poor dipolarophiles with triazinium ylides generated in situ from 1-alkyl-1,2,4-triazinium salts provide a facile access to various pyrrolo[2,1-f][1,2,4]triazines in a single step. The reaction gives differently substituted pyrrolotriazines in good yields, and the resulting compounds can be further elaborated by selective cross-coupling reactions. Our experimental data complemented by DFT calculations demonstrate that the reactions can be regio- and diastereoselective. We believe that the presented methodology will provide an efficient synthetic route to medicinally relevant pyrrolotriazines from readily available starting materials.