Herein, we report a pair of regioselective N1- and N2-alkylations of a versatile indazole, methyl 5-bromo-1H-indazole-3-carboxylate (6) and the use of density functional theory (DFT) to evaluate their mechanisms. Over thirty N1- and N2-alkylated products were isolated in over 90% yield regardless of the conditions. DFT calculations suggest a chelation mechanism produces the N1-substituted products when cesium is present and other non-covalent interactions (NCIs) drive the N2-product formation. Methyl 1H-indazole-7-carboxylate (18) and 1H-indazole-3-carbonitrile (21) were also subjected to the reaction conditions and their mechanisms were evaluated. The N1- and N2-partial charges and Fukui indices were calculated for compounds 6, 18, and 21 via natural bond orbital (NBO) analyses which further support the suggested reaction pathways.
IntroductionIndazoles constitute an important class of heterocycles with interesting biological and medicinal properties. Indazole, also called benzpyrazole, is a heterocyclic organic compound commonly found as a structural motif in natural products, pharmaceuticals, agrochemicals, and bioactive compounds . Indazole-containing compounds possess a wide range of pharmacological activities, such as anti-inflammatory, anti-arrhythmic, antitumor, antifungal, antibacterial, and anti-HIV activities . For example, two N1-substituted bioactive indazoles are found in Figure 1, danicopan (1), a complement factor D inhibitor for the treatment of paroxysmal nocturnal hemoglobinuria, and CPI-637 (2), an inhibitor of both cyclic-AMP response element binding protein (CBP) and adenoviral E1A binding protein . The N2-substituted indazole analogs pazopanib (3), an FDA-approved tyrosine kinase inhibitor used for the treatment of renal cell carcinoma, and Takeda’s MCHR1 antagonist 4 further exemplify indazole’s biological importance. Generally, direct alkylation of 1H-indazoles leads to a mixture of N1- and N2-substituted products . Procedures that selectively produce either N1- or N2-substituted indazoles would provide greater synthetic utility for this valuable heterocycle. These examples suggest a common intermediate such as methyl 5-bromo-1H-indazole-3-carboxylate (6) could be used to generate such compounds. An alkylation strategy that uses the vast array of commercially available alcohols as precursors of alkylating reagents presents an opportunity to find new ways to diversify indazole reactivity by simple modifications of reaction conditions. Mechanistic understanding would allow for further diversification in related systems.
Figure 1: Indazole-containing bioactive molecules.
The indazole ring presents annular tautomerism regarding the position of the NH hydrogen atom: 1H-indazole (5a, benzenoid 1H-indazole tautomer) and 2H-indazole (5b, quinonoid 2H-indazole tautomer) (Figure 2) . Since 1H-indazole is thermodynamically more stable than 2H-indazole, 5a is the predominant tautomer . Conventionally, indazoles are employed as nucleophiles in chemical transformations, and a mixture of both N1- and N2-alkylated products is formed, depending on the reaction conditions, with little selectivity in regards to substituent effects .
Figure 2: Tautomerism of indazole.
Considering the importance of indazoles as a widely used pharmacophore in medicinal chemistry and the challenges in obtaining either N1- or N2-alkylated indazoles as the dominant regioisomer , we were interested in exploring the regioselectivity with methyl 5-bromo-1H-indazole-3-carboxylate (6, Figure 1) as a multifunctional model compound for N1/N2 discrimination studies. The existing approaches for generating N-substituted indazoles of compound 6 often lead to a mixture of N1- and N2-alkylated indazoles with either low selectivity or moderate yields depending on the conditions used. For example, Takahashi et al. obtained N1- and N2-substituted indazole analogs in 44% and 40% yields, respectively, by treating compound 6 with methyl iodide and potassium carbonate in dimethylformamide (DMF) at room temperature for 17 h . Other works have shown poor selectivity when 6 and other isomers similar to 6 were reacted with isopropyl iodide and potassium carbonate, isopropyl bromide and cesium carbonate, and bromocyclohexane with potassium carbonate, and only afforded yields not higher than 52% in various solvents .
Recently, Alam and Keeting explored the regioselectivity in the alkylation of variously substituted indazoles similar to 6. They observed high N1-selectivity using NaH in THF with pentyl bromide and electron-deficient indazoles, postulating a coordination of the indazole N2-atom and an electron-rich oxygen atom in a C-3 substituent with the Na+ cation from NaH. Under anhydrous conditions the yields ranged from 44% at room temperature to 89% when warmed to 50 °C. No information was provided to justify any N2-selectivity or the lack thereof. Should an N2–Cs+–O ion pair exist, this could reasonably account for all the reported results presented herein (vide infra). Additionally, using Cs2CO3 in dioxane provided no products at room temperature (see Table 1) presumably due to the low solubility of Cs2CO3 in dioxane. They also provided a single example of a Mitsunobu reaction utilizing n-pentanol, dibutyl azodicarboxylate (DBAD), and PPh3.
Table 1: Representative reactions from reference .
Electrophile Solvent Reagents TempTherefore, there is still a great need to develop an operationally simple and mild method to selectively generate N1- or N2-substituted indazole analogs when the substituents appear to favor one over the other. Ideally, it would be greatly beneficial if the desired high regioselectivity on N1 or N2 could be achieved when commercially available chemicals, such as alcohols, react with 6 under different reaction conditions. In this paper, we report a concise and efficient approach to prepare N1- and N2-substituted analogs with high selectivity and excellent yields (>84%) from the same substrates: alcohols and 5-bromo-1H-indazole-3-carboxylate (6).
Results and DiscussionWe commenced our studies by investigating yields of N1- and N2-substituted products of conventional indazole alkylation reactions using our model substrate methyl 5-bromo-1H-indazole-3-carboxylate (6) and confirmed structures of the corresponding N1-substituted and N2-substituted products. In Scheme 1 compound 6 was treated with isopropyl iodide (7) in DMF in the presence of sodium hydride to provide products 8 and 9 in 38% and 46% yields, respectively. The structures of both compounds were unambiguously assigned using X-ray crystallography and 1H and nuclear Overhauser effect (NOE) NMR spectroscopy (see Supporting Information File 1).
Scheme 1: NMR, NOE, and yield data of compounds 8 and 9.
From the yield of products 8 and 9, we decided to explore new reaction conditions to improve the yields of the N1-substituted indazole analogs. As shown in Scheme 2, compound 12 was prepared by treating ethanol (10) with tosyl chloride (11) in the presence of 4-dimethylaminopyridine (DMAP) and triethylamine. Used as a model system, the subsequent reaction of sulfonate 12 with compound 6 under varied conditions afforded products P1 and P2. We investigated effects of reagent stoichiometry, bases, reaction time, and temperature on the yields of product P1, as summarized in Table S1 in Supporting Information File 1. Stoichiometric manipulation of 12 to 6 in DMF at 90 °C provided the N1-substituted product P1 in 52–60% yields. The yields of P1 formation were largely unaffected in DMF with temperatures ranging from room temperature to 110 °C. Varying the equivalents of Cs2CO3 showed little effect (averaging 52% yield), however, using NaH and lowering the temperature to rt lowered the P1 yield, averaging 32%. Finally, the effect of other solvents at 90 °C was investigated and the results are summarized in Table 2. Entry 1 shows the best conditions for the above reaction in DMF. The use of chlorobenzene slightly improved the yield to 66%. The reaction in dioxane at 90 °C, entry 6, had a 96% yield. This was a surprising outcome as Keeting observed no reaction in dioxane at rt, suggesting the concentration of Cs2CO3 is significantly increased at this temperature.
Scheme 2: Synthesis of compounds P1 and P2.
Table 2: Effect of solvent on indazole N1 yield.a
Entry Solvent P1 (isolated yield, %) 1 DMF 60 2 DMSO 54 3 NMP 42 4 chlorobenzene 66 5 toluene 56 6 dioxane 96aDMF: dimethylformamide; DMSO: dimethyl sulfoxide; NMP: N-methyl-2-pyrrolidone. Reaction conditions: 1.5 equiv 12, 1.0 equiv 6, 2.0 equiv Cs2CO3, 90 °C, 2 h.
With the promising yield results of P1, we next explored the scope of this transformation using a variety of alcohols (13a–q, Table 3) and report their regioselectivity as determined by crude LC–MS. Sulfonates 14a–q were prepared as described above or purchased (see Supporting Information File 1). The subsequent reactions with compound 6 afforded the N1-substituted indazole analogs 15a–q with excellent yields (>90%), except for 15m, which failed to form after multiple attempts likely due to an instability of the electrophile 14m under optimized conditions (conditions A: 1.5 equiv tosylate, 1.0 equiv 6, 2.0 equiv Cs2CO3, 90 °C, 2 h). Compounds containing linear or branched alkyl substitutions (15a–g), varied sizes of cycloalkane or saturated heterocycles (15h–p), including S- and R-tetrahydrofuran substitutions (15j, 15k) were isolated in excellent yields (>90%). Azetane 14m was unreactive towards alkylation in the presence of Cs2CO3.
Table 3: Scope of transformation and regioselectivity.
R Conditions A, major product Isolated yieldTo explore the possibility of N2-selectivity, we hypothesized that the phosphine intermediate of a Mitsunobu reaction could provide chelation control, directing alkylation to the indazole N2-atom while using identical alcohols as described above. Thus, we subjected 6 to simple and mild Mitsunobu conditions for the preparation of N2-substituted indazole analogs 16a–q. By directly reacting compound 6 with alcohols 13a–q (2 equiv), diethyl azodicarboxylate (DEAD, 2 equiv), and triphenylphosphine (TPP, 2 equiv) in THF at 50 °C (conditions B), the corresponding N2-substituted products were isolated in excellent yields (>90%) and high regiocontrol.
Crude product ratios as determined by LC–MS (averaging integrations at 254 nm and 260 nm) had an average error (standard deviation) of 3.2% (2.76%) and 13.7% (7.54%) for conditions A and B, respectively. The N1-isomer overlapped with OPPh3 contributing to the increased error and standard deviation. Compound 6 was completely consumed and not detected (see Supporting Information File 1).
Mechanistic considerationsAlam and Keeting proposed a deprotonated intermediate that utilized the indazole N2 and C=O from an ester substituent at C-3 as a bidentate ligand to the Na+ cation from NaH. The tight ion pair would direct alkylation under conditions A to N1. As this and other postulations exist, we explored the possible mechanisms of each reaction conditions computationally. All calculations were performed in implicit THF at the reaction condition temperature using Gaussian 16: SMD(THF)-PBE0/def2-TZVP // SMD(THF)-PBE0/def2-SVP, def2-TZVP(Cs) at 50 °C (w/MeOPPh3+) or 90 °C (w/Cs+), utilizing Goodvibes to calculate thermochemistry. The energy of the N1- and N2-tautomers of 6 differ by 3.1 kcal/mol at 50 °C, favoring the N1-tautomer, implying a 7:1 distribution of isomers in solution. Concerning conditions A, compound 6 + Cs2CO3 was found to favor the deprotonated indazole with a free Cs+ ion by 8.6 kcal/mol (see Figure 3), however, when all ions are discrete (2 × Cs+ and CO32−) the reaction becomes endergonic by 6.9 kcal/mol, presumably due to entropic penalties. Three of four computed resonance forms were all found to be of approximately equal energy. Only the E-enolate form 6 (-N1H-E) was slightly higher in energy by 0.06 kcal/mol likely due to electrostatic destabilization of the oxyanion with N2, however, this difference is negligible. These data suggest that deprotonation occurs prior to alkylation and that deprotonation of either indazole tautomer leads to anions of identical or highly similar energy. Furthermore, as seen in Figure 4, a total, five coordinated complexes were found to be at least 4.5 kcal/mol more stable than the uncoordinated anion when calculated as isolated structures. When the calculation is performed as a reaction of E and Z-enolates with Cs+ ion, two coordinated complexes 6(N-H)NNCs-E and-6(N-H)NOCs-Z are exergonically formed by 9.7 and 10.9 kcal/mol, respectively.
Figure 3: DFT-calculated deprotonation of 6 with Cs2CO3 in implicit THF with the temperature of the calculation set to 90 °C to simulate the dioxane conditions (top) and energy differences of four enolate resonance structures of 6 calculated as discrete structures. The hybrid is identified as 6(N-H) (bottom).
Figure 4: DFT-calculated Cs+-coordinated complexes with different enolate forms of 6(N-H) calculated as isolated compounds (top) and calculated intermediates of the reactions of 6(-N1H-Z) and 6(-N1H-E) with Cs+ (bottom).
We then searched for transition state (TS) structures that would produce both the N1- and N2-products from CH3OTs as a model system. When the Boltzmann average of the cesium-coordinated intermediates is calculated, a 3:1 ratio of 6(N-H)NOCs-Z:6(N-H)NNCs-E is found. This average is 10.6 kcal/mol more stable than 6(N-H), and was subsequently set to 0 kcal/mol leading to the energy diagram in Figure 5. Two TSs leading to each product were found, all four of which utilized a coordinating Cs+ cation. The N1-s-cis and N1-s-trans TSs were the lowest in energy (27.5 kcal/mol and 29.1 kcal/mol, respectively), leading to two conformations of the N1-product with highly similar energy (averaging −16.8 kcal/mol). The N2-s-cis and N2-s-trans TSs leading to the N2-product were higher in energy and led to the higher energy N2 products. The critical difference between N1-s-cis and N2-s-cis is the presence of the N2–Cs+–O non-covalent interaction (NCI) in N1-s-cis, which accounts for the 2.1 kcal/mol difference in energy. Calculations showed that the sulfonate oxygens also chelate the cesium ion in both TSs. Thus, nitrogen NCIs with cesium, or lack thereof, seem to drive N1-product formation, which is both kinetically and thermodynamically favorable under conditions A.
Figure 5: DFT-calculated reaction coordinate diagram for the reaction of 6 under conditions A. Concerning conditions B, we began our calculations under the assumption that MeOPPh3+ was already present in THF at 50 °C. The deprotonation by the dimethyl azodicarboxylate (DMAD) anion (DMAD−) to form 6(N-H) and 17 was found favorable by 14.7 kcal/mol (Figure 6).
Figure 6: DFT-calculated energy for the deprotonation of 6 by the DMAD anion.
Postulating that O or N-dative interactions with phosphorus were responsible for the high N2-selectivity in an analogous fashion to conditions A, we searched for intermediates and TSs that included this possibility. No O–P or N2–P-coordinated intermediates were found. The coordinated intermediate N1-P was found considerably endergonic with a ΔG of +8.0 kcal/mol compared to 6(N-H) (see Figure 7). A synchronous TS (N1-P-TS) leading to the N2-product was found starting from N1-P; however, the reaction barrier was 52.1 kcal/mol and thus a highly unlikely pathway. We again searched for TSs that led to both the N1- and N2-products but lacked any dative preorganization. However, under these reaction conditions, we found that the TS leading to the N2-product, N2-s-cis, was lower in energy than its N1-analog, N1-s-trans, by 1.1 kcal/mol (see Figure 8). This energy difference appears to be driven by stabilizing non-covalent interactions. Specifically, the carbonyl O in N2-s-cis shows NCIs with one of the benzene rings of PPh3 as well as a hydrogen bond-like NCI with a H-atom of the electrophilic methyl. Thus, the partitioning between transition states favor the N2-pathway over the N1-pathway by a product ratio of 4.5:1, which supports a pathway producing the observed experimental N2-product ratios with greater than 80% yield. The N1-product was again found to be lower in energy by 4.4 kcal/mol than the N2-product.
Figure 7: DFT-calculations concerning a coordinated Mitsunobu reaction pathway.
Figure 8: Reaction coordinate diagram of 6(N-H) reacting under conditions B. All calculated energies in kcal/mol. Ball-and-stick transition-state structures are provided for the lowest energy N1- and N2-transition states with favorable NCIs shown as red dashed lines.
To further explore whether the reaction mechanisms followed the chelation pathway proposed in Figure 5, we hypothesized that 18 (Figure 9) would provide a model for exploring the mechanism further. If chelation between an electron-rich oxygen atom from a substituent and a Lewis acid (such as Cs+ or P+) were taking place, we would expect regioselectivity for this substrate to be reversed (19-OCs, 20-OP), such that conditions A would produce the N2-product and conditions B would produce the N1-product. This was found to be the case as can be observed in Figure 9. The N2-product 19 was isolated in 93% yield under conditions A, and the N1-product 20 was isolated in >99% yield under conditions B albeit with low conversion. Both conditions provided >98:2 regioselectivity for their respective major products as determined by LC–MS.
留言 (0)