Regioselective N-alkylation of the 1H-indazole scaffold; ring substituent and N-alkylating reagent effects on regioisomeric distribution

The indazole scaffold represents a promising pharmacophore, commonly incorporated in a variety of therapeutic drugs. Although indazole-containing drugs are frequently marketed as the corresponding N-alkyl 1H- or 2H-indazole derivative, the efficient synthesis and isolation of the desired N-1 or N-2 alkylindazole regioisomer can often be challenging and adversely affect product yield. Thus, as part of a broader study focusing on the synthesis of bioactive indazole derivatives, we aimed to develop a regioselective protocol for the synthesis of N-1 alkylindazoles. Initial screening of various conditions revealed that the combination of sodium hydride (NaH) in tetrahydrofuran (THF) (in the presence of an alkyl bromide), represented a promising system for N-1 selective indazole alkylation. For example, among fourteen C-3 substituted indazoles examined, we observed > 99% N-1 regioselectivity for 3-carboxymethyl, 3-tert-butyl, 3-COMe, and 3-carboxamide indazoles. Further extension of this optimized (NaH in THF) protocol to various C-3, -4, -5, -6, and -7 substituted indazoles has highlighted the impact of steric and electronic effects on N-1/N-2 regioisomeric distribution. For example, employing C-7 NO2 or CO2Me substituted indazoles conferred excellent N-2 regioselectivity (≥ 96%). Importantly, we show that this optimized N-alkylation procedure tolerates a wide structural variety of alkylating reagents, including primary alkyl halide and secondary alkyl tosylate electrophiles, while maintaining a high degree of N-1 regioselectivity.

General approaches to the synthesis of N-1 or N-2 substituted indazoles involve the incorporation of the N-substituent prior to, or following, indazole ring-closure [10,11]. For example, several reports have highlighted the use of N-alkyl or N-arylhydrazines in the regioselective synthesis of 1H-indazoles, from the corresponding ortho-haloaryl carbonyl or nitrile, in good to excellent yield (Scheme 1) [12][13][14].
Alternative strategies to achieve regioselective indazole N-alkylation have exploited the noted difference in reactivity between the N-1 and N-2 atom of the indazole scaffold [15], as the 1H-indazole tautomer is typically considered to be more thermodynamically stable than the corresponding 2H-tautomer [16]. Using appropriate α-halo carbonyl electrophiles, Hunt et al. have shown that regioselective indazole N-alkylation can be achieved through an equilibration process which favours the thermodynamic N-1 substituted product [17].
Regioselective indazole N-acylation has been suggested to provide the N-1 substituted regioisomer, via isomerisation of the corresponding N-2 acylindazole to the more stable N-1 regioisomer [18]. Similarly, N-1 substituted indazoles have been obtained through thermodynamic equilibration, using β-halo ester electrophiles, in the presence of DMF [19]. These latter findings have been utilized to great effect by Conrow et al. to give regioselective access to N-1 alkylindazoles on kilogram scale, albeit over two steps from the corresponding N-1 acylindazole via reductive acetylation-deacetoxylation [20]. Although electronic and steric factors can influence the regiochemical outcome of indazole N-alkylation, varying reaction conditions, such as the choice of base [17,21], acid [22], solvent, and/or N-alkylating reagent may also facilitate regioselective indazole N-alkylation [23][24][25]. Bookser et al. have investigated the N-alkylation of related bicyclic azolofused-ring heterocycles, including 1H-indazole, employing NaHMDS in tetrahydrofuran (THF) or dimethyl sulfoxide (DMSO), and observed solvent-dependent regioselectivity [24]. Mechanistic hypotheses, based on elegant experimentation, were proposed to underline the roles that tight and solvent-separated ion pairs played in the observed trend in regioselectivity [24].
Our work sought to further explore the effect of C-3 substitution on N-alkylation selectivity control of the 1H-indazole scaffold. In view of these antecedents, it was envisioned that the development of a regioselective protocol for indazole N-1 alkylation would provide an improved and cost-effective approach to N-1 substituted indazole precursors as part of drug discovery and development campaigns.

Results and Discussion
Working towards the synthesis of a library of novel 1,3-disubstituted indazole derivatives necessitated us to develop a regioselective method that would permit the installation of a wide variety of alkyl sidechains at the N-1 position of methyl ester 9 (Table 1). Considering the reported influence of the reaction solvent and/or base on the regiochemical outcome of indazole N-alkylation [17,21,22,24], our initial efforts focused on examining the effect of varying these reaction parameters, using n-pentyl bromide as the prototypical N-alkylating reagent (Table 1).
Early investigations revealed that the combination of cesium carbonate (Cs 2 CO 3 ) in dimethylformamide (DMF) at room temperature (≈ 20 °C) afforded a mixture of N-1 and N-2 regioisomers (10 and 11, respectively), with only partial preference for the desired N-1 regioisomer 10 ( Table 1, entry 1). Furthermore, increasing the reaction time or decreasing the number of equivalents of Cs 2 CO 3 did not appear to influence the regiochemical outcome of the reaction (Table 1, entries 2 and 3, respectively). Substituting potassium carbonate (K 2 CO 3 ) for Cs 2 CO 3 did not show any improvement in the regioisomeric distribution of 10 and 11 (ratio N-1 (10)/N-2 (11) = 1.4:1) ( Table 1, entry 4). Similarly, the use of sodium carbonate under identical conditions gave a notably lower combined yield of 10 and 11 (27%), due to poor conversion (34%) ( Table 1, entry 5). Attempts to reduce the amount of K 2 CO 3 base to 0.5 equivalents, with respect to indazole 9, (Table 1, entry 6) resulted in incomplete conversion (62%) and provided no significant change in N-1 regioselectivity (ratio N-1:N-2 = 1.5:1). Importantly, using THF as the reaction solvent with potassium or sodium carbonate bases failed to give the N-alkylated products 10 or 11 (Table 1, entries 7 and 8, respectively).
To assign the regiochemistry of isolated N-1 and N-2 substituted indazole isomers, a combination of one and two-dimensional NMR experiments (particularly, heteronuclear multiple bond correlation (HMBC)) was employed [27]. For example, ( 1 H-13 C) HMBC analysis of N-1 regioisomer 10 shows a 1 H-13 C correlation between the C-7a carbon of the indazole ring and the n-pentyl CH 2 proton pair proximal to the indazole N-1 atom ( Figure 2). No evident 1 H-13 C correlation was observed between the n-alkyl CH 2 proton pair proximal to the  indazole N-1 atom and the indazole C-3 atom, for N-1 substituted indazole regioisomer 10. Conversely, ( 1 H-13 C) HMBC analysis of N-2 substituted regioisomer 11, revealed a 1 H-13 C correlation between the alkyl CH 2 proton pair (proximal to the indazole N-2 atom) and the C-3 carbon of the indazole heterocycle, while no 1 H-13 C correlation was observed between the alkyl CH 2 proton pair and the C-7a carbon atom of the indazole ring.
With a set of optimal conditions for the regioselective N-1 alkylation of methyl ester 9 in hand ( Table 1, entry 22), our attention turned to probing the influence of a variety of indazole C-3 substituents on the regiochemical N-alkylation outcome. A series of C-3 substituted indazoles (12-24) ( Figure 3) were thus assembled [28][29][30][31][32][33] to investigate the effects of electronic and steric factors on indazole N-alkylation ( Table 2) using our optimized conditions from Table 1 (entry 22) (referred to as "conditions A" in Table 2).
As literature precedence shows, the combination of Cs 2 CO 3 in DMF has been commonly employed to achieve indazole N-alkylation [17,34,35]. For comparison with our previously optimized N-alkylation protocol ( Table 1, entry 22; conditions A), conditions that provided less favorable N-1 regioselectivity (Table 1, entry 3, henceforth referred to as "conditions B" in Table 2) were also included as a part of this investigation.
Increasing alkyl and aryl steric bulk at the indazolic C-3 position seems to favor N-1 regioisomer formation in the order t-Bu > Ph > Me > H, for both conditions A and B ( Table 2, entries 1-4). Furthermore, having a sterically demanding t-Bu group at the indazole C-3 position (14) gave the N-1 substituted regioisomer 29 exclusively under both conditions A and B, respectively ( Table 2, entry 3). Apart from the 1H-indazole scaffold [36,37], the steric influence of adjacent substituent(s) on N-alkylation regioselectivity has previously been described for other nitrogen-containing heterocycles, such as pyrazole [38], purine, and related 1,3-azoles [39]. Although the N-alkylation of indazole 12, using conditions A (NaH in THF), proceeded with poor regioselectivity (ratio N-1 (25):N-2 (26) = 1:1.3), Bookser et al. have obtained a similar regioselective outcome using a combination of NaHMDS and MeI instead of NaH and n-pentyl bromide ( Table 2, entry 1), respectively [24].  The presence of a halogen atom (I, Br, or Cl) at the C-3 position of the indazole scaffold (16)(17)(18) revealed no significant trend in regioselectivity (  While the corresponding N-1 and N-2 regioisomers arising from the N-alkylation of C-3 substituted indazoles 12-24 were generally amenable to separation using wet flash column chromatography, the corresponding N-1-and N-2-n-pentylindazole derivatives of both indazoles 20 and 21 were largely inseparable ( Table 2, Table 2, entries 8 and 10-13). We postulate that these observed preferences for the generation of the N-1 regioisomer, under conditions A, may be due to the formation of a tight ion pair involving the indazole N-2 atom and C-3 substituents which are capable of cation chelation via the N-2 atom electron lone pair and the C-3 substituent X=O functionality, respectively [24]. Tight ion pair formation with the sodium cation and both the N-2 atom and C-3 substituents of the indazole scaffold likely hinders the approach of the electrophile to N-2 and directs alkylation to the N-1 position. Furthermore, this effect is not observed for indazoles bearing C-3 substituents that cannot participate in the formation of tight ion pairs (such as, 12-18 and 20), under conditions A (NaH in THF). The latter N-1 regioselectivity conferred through tight ion pair formation is augmented by the steric effect that the C-3 substituent enforces. The high degree of N-1 regioselectivity obtained for indazoles bearing bulky substituents at the C-3 position that are not capable of engaging in tight ion pair formation, such as 14 and 15, further highlights the influence of steric effects on regioselectivity.
To determine if the high degree of N-1 regioselectivity (ratio N-1:N-2 > 99:1) observed when employing conditions A (see Table 1 and Table 2) was due to base (NaH) or solvent (THF) effects, Cs 2 CO 3 was substituted for NaH (Table 3, entries 1 and 2). Although the N-alkylation of indazole 9 was hindered by poor conversion (9%) when carried out at room temperature, the complete conversion of 9 to regioisomers 10 and 11 was observed and N-1 regioselectivity maintained upon increasing the reaction temperature to 50 °C (Table 3, entries 1 and 2).
Employing a combination of NaH and DMF (  [24]. It is likely that DMF similarly facilitates the formation of solvent-separated ion pairs which serve to diminish the high N-1 regioselectivity previously achieved when using THF (Table 3, entry 2), where tight ion pair formation between the cesium cation, N-2 atom, and chelating X=O C-3 group of the indazole ring predominates. Although the C-3 methyl carboxylate group of 9 may contribute to selective N-1 alkylation through steric effects, the use of DMF as the reaction solvent does not support tight ion pair formation and diminishes N-1 regioselectivity.
To investigate the effect of the position of the indazole C-3 substituent on regiochemical outcome, several C-7 substituted indazoles (Me-, Br-, NO 2 -, and CO 2 Me) were alkylated using both conditions A (NaH in THF) and B (Cs 2 CO 3 in DMF) ( Table 4, entries 1-4). A significant reversal in regioselectivity was observed under conditions A for C-7 Me and Br-substituted indazoles, when compared with their analogous C-3 substituted counterparts ( Table 2, entries 2 and 6), favouring the formation of the corresponding N-2 regioisomer (Table 4, entries 1 and 2). This latter preference for N-2 alkylation is likely due to the proximal steric bulk of both the C-7 Me and Br substituents, respectively, to the N-1 position.
The presence of a nitro or methyl carboxylate group at the C-7 position of the indazole core facilitated excellent N-2 regioselectivity under conditions A (Table 4, Table 4, entry 5). Importantly, the remarkable N-2 regioselectivity observed for C-7 NO 2 and CO 2 Me substituted indazoles (≥ 96%) under conditions A provides further support for the role that tight ion pair formation may play in achieving regioselective N-alkylation (vide supra).
Mechanistically, we postulate that our optimized regioselective N-1 alkylation of the exemplar methyl ester 9 and other appropriately C-3 substituted indazoles (19, or 21-24) (under conditions A, Table 1, entry 22) involves the initial irreversible deprotonation of the indazole in the presence of NaH to initially give indazolyl salt 65 which is in equilibrium with its alternate anionic form 66 (Scheme 3). Through tautomerization, salt 66 may then form a tight ion pair with a sodium cation, via the N-2 atom and X=O containing C-3 substituent of the indazole nucleus, affording species 67 whose existence dominates in THF as solvent. The formation of 67 is then followed by the nucleophilic substitution of alkylating reagent R 2 -X to selectively give the desired N-1 regioisomer 68. It is likely that a mixture of the solvent-separated ion pairs 65 and 66 predominate, when using polar solvents such as DMF [24]. Furthermore, precluding tight ion pair formation through the use of DMF, may prompt indazole N-alkylation to fall predominantly under steric control, resulting in diminished N-alkylation regioselectivity.  To further probe the potential influence of the alkali metal cation on the regioselective N-1 alkylation of indazole methyl ester 9, a control experiment was carried out, using 1 equivalent of the ether 15-crown-5 (with respect to NaH) ( Table 5). Chelation of the sodium cation with the crown ether should disrupt the formation of tight ion pairs (Scheme 2) and attenuate N-1 regioselectivity. The presence of 15-crown-5 caused a notable reduction in N-1 regioselectivity, when compared with results obtained in the absence of the crown ether (ratio N-1 (10):N-2 (11) = 9.6:1 ( To demonstrate the scope of our optimized N-1 regioselective N-alkylation protocol (conditions A), methyl ester-substituted indazole 9 was subjected to a series of alkylating reagents under both conditions A and B ( Table 6). The high selectivity observed for N-1 alkylation using NaH in THF (conditions A) was mainly effective using primary halide and tosylate compounds as electrophiles. Similar to the regiospecificity observed when employing n-pentyl bromide (ratio N-1 (10):N-2 (11) > 99:1, Table 1, entry 22), its tosylate counterpart gave the corresponding N-1 regioisomer 10 with a high degree of N-1 regioselectivity (ratio N-1 (10):N-2 (11) = 76:1) under conditions A ( Table 6, entry 1). Furthermore, conditions A could be successfully applied to the synthesis of benzyl and alicyclic indazole derivatives 69-74 (Table 6, entries 2-6), affording the N-1 regioisomer almost exclusively. Notwithstanding excellent N-1 regioselectivity when using conditions A (ratio N-1 (73):N-2 (74) > 99:1), the yield of the corresponding N-1 substituted cyclohexylmethylindazole 73 was significantly reduced, due to poor conversion (13%, combined N-1 and N-2 (see Supporting Information)) ( Table 5, entry 5). However, employing the corresponding tosylate under identical conditions (NaH in THF (conditions A)) permitted improved conversion to the desired N-1 substituted alicyclic indazole 73 (78%, combined N-1 and N-2 (see Supporting Information File 1)), whilst maintaining excellent N-1 regioselectivity (ratio N-1 (73):N-2 (74) = 70:1) ( Table 6, entry 6).
In the presence of NaH in THF (conditions A), secondary alkyl bromides, such as 2-and 3-bromopentanes, both gave their corresponding N-1 alkylindazoles 75-78 in only trace amounts (< 5% isolated yield, Table 6, entries 7 and 9), due to poor conversion (< 5%, combined N-1 and N-2 (see Supporting Information File 1)). While the latter observation may be due to competing elimination of the alkyl halide under strongly basic conditions [40], the use of a secondary tosylate electrophile under conditions A ( Table 6, entry 8) furnished the desired N-1 regioisomer 75 in very good isolated yield (81%). These latter results would suggest that secondary alkyl tosylates are more suitable than their corresponding halide counterparts, for N-1 regioselective alkylation, under these investigated conditions. Conversely, the use of Cs 2 CO 3 in DMF (conditions B) afforded approximately equal amounts of the corresponding N-1 and N-2 regioisomers 75-78 when using the aforementioned secondary alkyl bromides, with complete consumption of indazole 9 observed ( Table 6, entries 7 and 9). However, N-1 regioselectivity is absent (ratio 75:76 and 77:78 ≈ 1:1, Table 6, entries 7-9) under conditions B, most likely due to solvent-separated ion pair formation.

Conclusion
Both 1H-and 2H-indazoles represent a core heterocyclic motif in many therapeutic small molecule drugs. Thus, from a synthetic perspective, the regioselective N-alkylation of the indazole scaffold would be of great value to the pharmaceutical industry. Focusing on 3-substituted indazoles, highly selective N-1 alkyl-ations can be achieved using NaH in THF (conditions A) and Cs 2 CO 3 in DMF (conditions B) as applied to primary and secondary alkyl electrophiles. When compared with conditions A, the use of Cs 2 CO 3 in DMF (conditions B) demonstrated improved regioselectivity for the corresponding N-1 regioisomers of unsubstituted (12) and C-3 methyl (13), phenyl (15), halo (16)(17)(18), and cyano (20) substituted indazoles. Investigating the effect on the regioisomeric N-1/N-2 distribution indicated that steric bulk likely plays a significant role in determining N-1 regioselectivity. For example, the ratio of N-1/N-2 increases in the order of increasing steric bulk at the indazole C-3 position (H < Me < Ph < t-Bu). However, when the former protocol (conditions A) was extended to indazoles bearing a nitro (19) or carbonyl (21)(22)(23)(24) functional group, the desired N-1 regioisomer was obtained with a very high degree of regioselectivity. In the case of C-3 substituted indazoles 9, 19 and 21-24 (those bearing a X=O α to the indazole C-3 position), we postulate that in the presence of NaH, the corresponding indazole salt may form a tight ion pair (67) which serves to attenuate N-2 alkylation, and affords the desired N-1 regioisomer exclusively. Furthermore, the excellent N-2 regioselectivity (≥ 96%) observed for the N-alkylation of C-7 NO 2 or CO 2 Me indazoles (Table 4, entries 3 and 4) provides further support for the important role that tight ion pair formation plays in directing N-alkylation of the indazole scaffold.

Supporting Information
Supporting Information File 1 Compound synthesis, characterisation, and copies of spectral data pertaining to regioisomeric distribution (N-1:N-2) determination.

Funding
This work was supported by funding from the Eli Lilly Research Scholarship (4152 R17825).