Assembly of fully substituted triazolochromenes via a novel multicomponent reaction or mechanochemical synthesis

A new metal-free one-pot three-component procedure towards fully substituted triazolochromenes has been developed, starting from commercially available materials. Salicylaldehydes and nitroalkenes were reacted under solvent-free conditions, followed by a 1,3-dipolar cycloaddition of the intermediate 3-nitro-2H-chromenes with organic azides in a one-pot two-step sequence. The triazolochromenes were formed with complete regioselectivity and new biologically relevant structures were synthesized via extension of the developed procedure and via postfunctionalization. The mechanochemical synthesis was carried out for several salicylaldehydes and gave a clear improvement in the yield of the corresponding triazolochromenes and consequently showed to be a viable alternative for solid salicylaldehydes.


Introduction
Chromenes are important structural motifs and are omnipresent in nature and drugs for medicinal applications [1][2][3][4]. Vitamin E [5][6][7][8], arahypin-5 [9,10], THC and other cannabinoids [11][12][13][14] are only a few examples of biologically relevant chromenes. Hence, the search for new methodologies towards the rapid assembly of chromene analogs is of utmost importance for many researchers. In this regard, 3-nitrochromenes are easily available building blocks for chromene and chromane deriva-tives and are highly reactive due to the presence of the nitroalkene moiety, which enables them to undergo a high variety of reactions and functionalizations [15].
Our group developed a Knoevenagel-assisted three-component reaction of (protected) salicylaldehyde, ethyl nitroacetate and organic azides, in which the synthesis of both triazolocoumarin regioisomers was accomplished [33]. Interestingly, the expected regioisomer was not observed in the case of the in situ formed 3-nitrocoumarins. Hence, in our continued exploration towards novel multicomponent reactions for the assembly of triazole-fused (hetero)cycles [24,[33][34][35][36][37][38][39][40][41][42], we opted to develop a new one-pot two-step three-component reaction starting from salicylaldehydes, nitroalkenes and organic azides, without isolation of the intermediate 3-nitrochromenes, in a regioselective manner and without the use of metals. Salicylaldehydes with a high melting point or low solubility proved difficult to convert to the intermediate 3-nitrochromene derivatives [15]. In this regard, applying mechanochemistry has been proven previously to be a viable alternative [43]. To the best of our knowledge, both the development of a metal-free sequential one-pot three-component reaction and the mechanochemically assisted 3-nitrochromene synthesis towards fully substituted tria-zolochromenes, without the isolation of the intermediate 3-nitrochromenes, have not been reported until now.

Results and Discussion
To prove the plausibility of the one-pot three-component reaction, we commenced our trials with the synthesis and isolation of 3-nitro-2H-chromene (3) as reported in the literature [15], followed by the 1,3-dipolar cycloaddition of the nitroalkene moiety with organic azides. We based the 1,3-dipolar cycloaddition reaction on the synthesis of NH-triazoles by Guan et al. using p-toluenesulfonic acid as the catalyst in DMF [19], but with benzyl azide (4a) instead of sodium azide (Scheme 1). Our initial test gave a promising result, since after a reaction time of five days for the cycloaddition step the desired product 5a was obtained in 67% yield, together with an oxidized ring opened side product 6 in 20% yield. The overall yield of 5a after two steps was 48%, considering that the 3-nitro-2H-chromene (3) was obtained in a yield of 71%.
We continued our studies by verifying the obtained regiochemistry, in which we synthesized the different regioisomers 5a, 10 and 11 via different pathways (Scheme 2). The regiospecific syntheses [36] of compounds 5a and 11 were accomplished by triazolization of the corresponding flavanone 7 and 2-phenylchroman-3-one (8), respectively. As anticipated, these methods furnished both regioisomers in poor yields since the chromanones 7 and 8 are known to be unstable under the triazolization conditions [36]. Hence, no further attempts were made to improve these yields. Additionally, NH-triazole 9 could be alkylated using benzyl bromide and potassium carbonate in acetone providing a mixture of alkylated triazolochromenes 5a, 10 and 11. The polarity of the 2-alkylated triazolochromene 10 is significantly different than the other two which were obtained as an inseparable mixture of both regioisomers 5a and 11 in a 1:3 ratio. Comparing the 1 H NMR spectra (see Supporting Information File 1, Figure S1 for NMR comparison), we can make unambiguous conclusions about the regiochemistry of the synthesized compounds 5a, 10 and 11 (Scheme 2). As the product contains a stereocenter, there is a possibility to see diastereotopic splitting of the benzylic protons. In the spectrum of the obtained product 5a starting from 3-nitro-2H-chromene and flavanone, this splitting is not observed (A 2 pattern). The benzylic peak of the 2-alkylated product 10 shows an AB splitting pattern and the third regioisomer 11 shows a substantial AX splitting pattern. This striking difference can be rationalized in function of the proximity of the stereocenter to the diastereotopic protons. Further proof was provided by characterization of side product 6 [44], which is formed during the reaction by oxidation and ring-opening of triazolochromene 5a (Scheme 1). All these observations confirm the expected regioselectivity in the formation of triazolochromene 5a via 3-nitro-2H-chromene (3).
Next, the two-step synthesis was converted into a one-pot twostep synthesis, circumventing the need for isolating the intermediate 3-nitro-2H-chromene (3), which would greatly facilitate the purification of the overall reaction since the 3-nitro-2Hchromenes and their starting materials show similar retention factors. Since the triazolochromenes 5 are showing much lower retention factors, the one-pot synthesis would display a great improvement in the labor intensiveness both for the purification steps and reaction set-up. The reaction was further optimized using salicylaldehyde (1a), β-nitrostyrene (2a) and benzyl azide (3a) as model substrates (see Supporting Information File 1, pages S4-S8 for full description of the optimization study). The optimized conditions for the one-pot three-component reaction were determined to be 1 equivalent nitroalkene, 1.2 equivalents of salicylaldehyde and 0.1 equivalents of DABCO as catalyst in the first step at 40 °C, and 2 equivalents of benzyl azide, 2 equivalents of acetic acid, 0.3 equivalents of BHT as antioxidant, 4 Å MS and DMF under argon atmosphere at 120 °C in the second step. Crude NMR analysis of the reaction mixture under optimized conditions showed solely regioisomer 5a, which was obtained in 54% yield after chromatographic purification. Additionally, the optimized conditions gave improved yields compared to the two-step synthesis and circumvented the formation of oxidized side product 6. As variation of the substituents on the three different starting materials is necessary to obtain a diverse library, there is one main limitation to overcome. The first step of the reaction relies on the fluidity of salicylaldehyde (1a) to liquefy the reaction mixture. Salicylaldehyde analogs 1c-f are solids at 40 °C and hence, to overcome this problem, some slight modifications from the optimized conditions were done (see Supporting Information File 1, pages S6 and S7 for more detailed description of the performed experiments for compound 5a). Eventually, the use of two equivalents of triethylamine was needed but the overall yield of 5a was still lower as it only reached 38%.
With the obtained optimized conditions and proof of regioselectivity in hand, further investigation towards the generality of this three-component reaction was carried out by varying the substrate scope ( Figure 1). We first studied a range of salicylaldehydes 1a-f, from which 5a and 5b were obtained in the best yields since salicylaldehydes 1a and 1b are liquids. As mentioned earlier, the altered conditions for solid salicylaldehydes result in general in a decrease in yield. Yet, we were able to diversify towards electron-rich triazolochromenes 5b and 5c, resulting in a drastic loss in yield for the more sterically hindered compound 5c. Furthermore, electron-deficient and halogenated analogs 5d-f were successfully synthesized. In a next series, we examined the substitution pattern on the nitroalkene part. Electron-rich substituents 3,4,5-trimethoxyphenyl and piperonyl were tolerated and furnished products 5g and 5h, respectively. 2,2-Dimethyl-substituted derivative 5i was prepared from 2-methyl-1-nitroprop-1-ene (2d) and interestingly, 1,4-bis((E)-2-nitrovinyl)benzene (2e) furnished bischromenotriazole 5p in 26% yield (Scheme 3). Finally, the scope with respect to organic azides was investigated by performing reactions with alkyl and aryl azides 4a-g. Electron-rich aliphatic azides produced products 5j-l in moderate yields. Additionally, electron-rich and electron-deficient aromatic azides were explored, resulting in slightly lower yields and elongated reaction times in the cycloaddition step compared to aliphatic azides. Unfortunately, this reaction has encountered some limitations towards certain substrates (not shown). In the case of a strongly electron-withdrawing substituent on the nitroalkene part for (E)-1-nitro-4-(2-nitrovinyl)benzene and sterically hindered 2-hydroxy-1-naphthaldehyde, only the oxidized product analogous to 6 was observed. 2-Hydroxy-4-nitrobenzaldehyde and 2,6-dihydroxybenzaldehyde were unreactive in the cycloaddition reaction.
As previously mentioned, solid salicylaldehydes furnish triazolochromenes in diminished yields in the one-pot three-component reaction (Figure 1, compounds 5c-f). Hence, a mechanochemical two-step protocol was developed, since a report by Jia and Zhang et al. [43] previously showed that ball milling could be a convenient manner to produce 3-nitrochromenes. Because our first step is best performed sol-vent-free, we opted to try our own optimized solvent-free conditions for the in situ syntheses of 3-nitro-2H-chromenes, followed by the 1,3-dipolar cycloaddition in a reaction vial. Despite being a two-pot procedure, purification of the intermediate 3-nitro-2H-chromene is still circumvented. Hence, our initial trials were performed by using solid salicylaldehydes 1c-f (Figure 2), resulting in a significant increase in yield for triazolochromenes 5c-f compared to the one-pot procedure developed as described above (Figure 1). To compare the two methodologies, the two highest yielding liquid salicylaldehydes in the one-pot protocol, i.e., 1a and 1b, were reacted in the twostep mechanochemically assisted reaction, giving rise to slightly lowered yields for compounds 5a and 5b. Hence, the use of the ball milling procedure is advantageous when solid salicylaldehydes are used. Complementary to this the one-pot three-component reaction gave better results for liquid salicylaldehydes.
In order to show the utility of the developed methodologies towards possible drug discovery, the scalability of the reactions was explored (Scheme 4). Both developed methodologies easily led to gram scale syntheses without significant loss in yield, i.e., 50% and 40% for the one-pot three-component reaction and mechanochemical procedure, respectively.

Conclusion
We developed a sequential one-pot three-component reaction to access a variety of novel triazolochromenes avoiding the purification of intermediate 3-nitro-2H-chromenes. The regiochemistry of the reaction was determined and proven, followed by a scope study using various salicylaldehydes, nitroalkenes and organic azides. Solid salicylaldehydes gave diminished yields in the one-pot three-component protocol, hence a two-step mechanochemical approach was developed offering higher yields and resulting in a complementary route for solid salicylaldehydes. The applicability of the newly developed protocols was shown by gram-scale syntheses and postfunctionalization reactions towards biologically relevant analogs. The biological data will be published in due course.

Supporting Information
Supporting information features the optimization studies, NMR comparison studies of the various regioisomers 5a, 10 and 11, experimental details and copies of 1 H and 13 C NMR spectra of compounds 5a-p and 10-15.

Supporting Information File 1
Experimental part.