Purine-scaffold Hsp90 inhibitors are the first synthetic small molecule Hsp90 inhibitors to be discovered and to be translated to clinic in patients with advanced cancers.³ Among purines with Hsp90 inhibitory activities are the 8-arylmethyl-9H-purin-6-amine derivatives, such as PU3, PU24FCl and PU-DZ8 (Figure 1).^3a PU3 is the first discovered synthetic Hsp90 inhibitor, and was identified through empirical structure-based design.^3b PU24FCl, a derivative of PU3, is the first synthetic Hsp90 inhibitor reported to exhibit anti-tumor activity and biological activity in a widerange of tumor types.^3c

PU-DZ8 has recently been reported to reduce aberrant neuronal proteins and diminish toxic tau aggregates in transgenic mouse models of dementias caused by tau protein toxicity, such as Alzheimer's disease and frontotemporal dementias^3d. Considering their biological effects in both cancer and neurodegenerative diseases, there is an increased interest in their development and optimization towards a clinical candidate.³

Attempts to result in higher chemical diversity in the 8-arylmethyl-9H-purin-6-amine series remain limited by the somewhat cumbersome synthesis on which building of the 8-benzyl purine core relies.⁴ Its synthesis continues to employ a three-step protocol consisting of generating the acid fluoride of the corresponding phenyl acetic acid, coupling it with pyrimidine-4,5,6-triamine or pyrimidine-2,4,5,6-tetraamine in the presence of K₂CO₃, followed by condensation of the intermediate amide product (Scheme 1).⁴

These condensation reactions require harsh dehydration conditions and/or long reaction times, such as refluxing overnight in ethanol in the presence of 15 equivalents of NaOMe,^4a or refluxing for 5 hr in a 25% solution of NaOMe in methanol/isobutanol,^4b which limit the spectrum of suitable starting materials. Reliance on the multi-step protocol results in lower overall yields, with a 20% isolated yield reported for the precursor of PUDZ8.^4b Dymock et al utilized a two-step protocol in which the carboxylic acid is directly coupled with the tri- or tetra-aminopyrimidine in the presence of i-Pr₂NEt and HBTU in DMF, followed by condensation of the intermediate amide product by refluxing overnight with NaOMe (15 equivalents) in EtOH or by microwave irradiation for 30 min at 130°C.^4a Unfortunately, yields were not reported so that a satisfactory comparison between each of these methods cannot be made.

To limit the number and length of the synthetic steps, and to increase the deliverable chemical variability in the 8-arylmethyl-9H-purin-6-amine series, we investigated the use of microwave irradiation. Microwave-assisted organic synthesis continues to grow in importance in synthetic chemistry by enabling rapid chemistry development and numerous reactions, including condensations and heterocycle formations have been explored under these conditions.⁵ Herein, we report the use of microwave-assisted chemistry for the successful one-pot delivery of 8-arylmethyl-9H-purin-6-amines. In addition, we briefly discuss cases in which the reaction fails to deliver the desired product.

Previously, Lin et al have reported a microwave-assisted synthesis of benzimidazoles and other heterocycles from condensation of a diamine with a carboxylic acid.⁶ We probed the feasibility of adapting this method⁷ to more functionally complex amino-pyrimidines in an attempt to produce the desired condensation product in a single step. In an initial effort to prepare several PU-DZ8 derivatives, 2-(benzo[d][1,3]dioxol-5-yl)acetic acid (1.0 equiv) and pyrimidine-4,5,6-triamine (1.2 equiv) were irradiated at 220°C for 15 min in the presence of P(OPh)₃ (1.2 equiv) in pyridine, to result in 8-(benzo[d][1,3]dioxol-5-ylmethyl)-9H-purin-6-amine (3a) in 87% isolated yield. High recovered yields were also noted when the phenyl ring was replaced with a pyridine moiety (Entry 3, Table 1).

When the triamine-pyrimidine substrate (1a) was replaced with its tetraamine-equivalent (1b), which is required for the building of PU-DZ8 derivatives, the recorded yields decreased under the Lin conditions (87% yield, Entry 1, Table 1 versus 62% yield, Entry 1, Table 2). The relatively lower solubility of 1b in pyridine, when compared to 1a, could be partly accountable for poorer yields. Increasing the reaction time to 20 min did not significantly improve the yield (65% yield, Entry 2, Table 2). Addition of DMSO or DMF, solvents which in our hands usually increase the solubility of purines and pyrimidines, had an adverse effect on the yield (Entries 3 and 4, Table 2). As expected, addition of a strong base, NaOtBu, augmented the yield to 78% of recovered material, possibly by a two pronged beneficial effect, on both solubility and coupling and condensation efficacy (Entry 5, Table 2).

The free base form of pyrimidine-2,4,5,6-tetraamine is commercially unavailable due to its propensity to oxidize and decompose, and therefore short shelf life. The compound is instead sold as a sulfate salt of very poor solubility in organic solvents. The base is usually generated by recrystallization from 10% NaOH aqueous solution before its use in the coupling with carboxylic acids 2. To eliminate the extra step, we probed the in situ conversion of the sulfate into free base. Surprisingly, pyrimidine-2,4,5,6-tetraamine sulfate remained poorly soluble in pyridine even when heated at 220 °C. Addition of NaOH (4 equiv), rendered pyrimidine-2,4,5,6-tetraamine sulfate completely soluble in pyridine, however, increased the yield to only 25% (Entry 7, Table 2). We assumed that one reason could be the in situ generated water. This however, can only partly be true, because replacement of NaOH with NaOtBu had only a marginal benefit on the recovered yield (Entry 8, Table 2).

To evaluate the limitations of this procedure towards the formation of 8-arylmethyl-9H-purin-6-amines, we chose to investigate the addition of functionalities at positions more likely to result in interference with the condensation reaction (Table 3). These were estimated to be functionalities either in the ortho position of the aryl ring (R² Table 3) or adjacent to the carboxyl functionality (R¹ Table 3). We found that the reaction was sensitive to the nature of the ortho functionality, as demonstrated by lower yields when R²=H was exchanged with a trifluoromethyl or cyano functionality (23% recovered yield for 3e and traces detectable by mass spectrum for 3f). This outcome could be due to steric effects, because attempts to increase the solubility of the pyrimidine 1b and to facilitate coupling and condensation by addition of a base, NaOt-Bu (2 equiv), failed to result in improved yields (Entry 4, Table 3). We found that insertion of a substituent at the α-position of the acetyl carboxylic acid (R¹) also played an important role in this reaction. The transformation tolerated the addition of a methyl (Entry 7, Table 3), however, no product was detected when hydroxyl or methoxy derivatives were employed (Entries 5 and 6, Table 3). It is unclear whether a lack of product in the case of hydroxyl and methoxy derivatives 2 is due to decomposition of the carboxylic acid 2, or to electronic effects which reduce the electrophilicity of the carboxyl group.