Continuous-Flow Preparation of Benzotropolones: Combined Batch and Flow Synthesis of Epigenetic Modulators of the (JmjC)-Containing Domain

A continuous flow synthetic protocol for the preparation of benzotropolones, and further derivatization to yield a recently described inhibitor of (JmjC)-containing domain enzymes is described. Our procedure renders the desired compound in an efficient and reproducible manner and paves the way towards the preparation of higher amounts of the product, needed for more extensive biological studies.

Nature has long being a source of inspiration for the development of new drugs [1] and many of them have being approved by different worldwide-recognized agencies for the treatment of human diseases. [2] Benzotropolone-containing natural products are a family of polyphenolic secondary metabolites that exhibit very interesting biological properties, such as antioxidant, [3] anticancer activities, [4] modulators of the inflammatory response activity, [5] and more recently found, activity as inhibitors of TLR1/TLR2 (Toll-like receptors). [6] Representative examples of secondary metabolites belonging to this group are purpurogallin 1, [7,8] theaflavin 2, [9,10] fomentariol 3, [11] aurantricholone 4, [12] goupiolone A 5 [13] and crocipodin 6 [14] (Figure 1). Furthermore, we have showed that properly functionalized benzotropolone 7 act as very potent inhibitors of lysine demethylase JMJD2 A, [15] a subtype of the Jumonji C (JmjC)containing domain family of enzymes, [16][17][18] responsible of the demethylation of N-methylated histone lysine side chains. This epigenetic modification has shown dramatic effects in cell proliferation causing, among other activities, endocrine deregulation and cancer development. [19,20] The preparation of biologically active molecules in an efficient, robust, and reproducible manner, from early stages of compound development pipeline, warrants the amount of substance necessary for the different stages of the evaluation process, while speeding up the eventual scaling up of lead molecules. [21] In that sense, continuous flow processing stands out among the available enabling technologies for the synthesis of (API). [22][23][24][25][26][27][28][29] The main advantages of this technique are the observed enhancement of mass and heat transfer and the reaction acceleration, resulting in a more efficient mixing of reagents and a precise control of reaction parameters. Also, particular reactivities are often observed compared to a batch mode. [30] Furthermore, the possibility of a continuous production of the targeted product, the implementation of in-line purification techniques and the capability of development of telescoped procedures allow a more sustainable, less time consuming, introduction of the desired functional group. [31][32][33] Following our interest in the development of continuous flow synthetic methodologies for the synthesis of biologically active molecules for further characterization of their activity, [34,35] we describe our work in the development of a continuous synthetic procedure for the preparation of benzotropolone core ring present in this family of compounds, and further derivatization to render the recently described epigenetic modulator 7. [15] In order to develop an efficient methodology that eventually allows, not only the isolation of larger quantities of compound 7, [15] but also the preparation of other derivatives in a reproducible manner, we started to study the adequacy of the benzotropolone core ring formation reaction under continuous flow. Besides the different methodologies described in literature for the preparation of benzotropolones, [36][37][38][39] peroxidase-mediated condensation reaction of catechol and pyrogallol derivatives stands out as the most frequently used methodology for the formation of benzotropolone ring, [6,14,15,40] so these reaction conditions were chosen as starting point.
Furthermore, continuous flow synthetic processeses involving enzymes, rely on the use either soluble or supported enzymes for the preparation of the targeted molecules. [41,42] While the later have the advantage of reusability, high catalyst loading and easier purification of obtained products, it has been showed that the activity of the inmobilized biocatalyst for a particular transformation is greatly influenced by inmobilization strategy and flow reactor design. [43] To the best of our knowledge, there is only one precedent in literature, reported by the Ley group, describing the synthesis of targeted molecules using horseradish peroxidase adsorbed onto silica. [44] In view to all that, and considering that the insolubility of our benzotropolones under previously described reaction conditions using soluble horseradish peroxidase [15] could eventually lead to packed-bed reactor clogging, we started our screening by mixing two independent solutions, one containing commercially available 8 a, 9 a, and the enzyme and the other containing the oxidative solution. Under these condition reactor blockage was observed despite the internal diameter of the coiled reactor was increased (1-2 mm i.d.) to reduce product flocculation. (Entries 1 and 2, Table 1) Handling of in situ formed solids is a difficult task during the development of a continuous flow methodology mainly due to blockage of used reactors by precipitation of formed solid within the tube. Different solutions have been previously reported in order to avoid clogging issues derived from particle aggregation. [45][46][47][48] One option would be the introduction of our reactor coil in a sonicating bath during operation. Sadly, no improvement was observed under these conditions. (Entry 3, Table 1) Better results were achieved when a customized coiled reactor, wrapped around a Falcon conical tube®, was connected to a vortex mixer during operation. [49] With this setup, we were able to obtain the desired compound in 34 % yield (Entry 4, Table 1). Any attempt to improve the yield by increasing the amount of oxidant was unproductive (Entry 5, Table 1). Instead, using a longer reactor coil (4 mL, 20 min approx. residence time), significantly increase the reaction yield up to 61 % (Entry 6, Table 1). The described protocol was amenable to large scale, producing 1.7 g of compound 10 aa in pure form after 5 h (a daily throughput of 8.2 g/day), without clogging issues been noticed. In this case, although slightly lower, the reaction yield of our continuous flow procedure is comparable to the batch reaction (68 % yield) earlier described by our group, and also similar to enzyme mediated methodologies previously described by others, [5,6,14] while adding all benefits previously mentioned for organic reactions performed in a continuous fashion. Lower residence time, achieved by doubling the flow rate (0.2 mL/min instead of 0.1 mL/min) resulted in a decrease of the yield of the reaction (Entry 7, Table 1). The optimized methodology was successfully applied for the condensation of other previously described benzotropolone derivatives. Hence, methyl gallate 8 a was reacted with 4-methyl-(9 b) and 4chloro-catechol 9 c to form, respectively, methyl-(10 ab) and chloro-benzotropolone 10 ac in yields comparable to previously described batch protocol (Entries 8 and 9, Table 1). [15] Interestingly, the optimized procedure could be also applied for the oxidative condensation of pyrogallol 8 b and catechol 9 d to improve the previously described yield of compound 10 bd, a molecule that has been previously described as moderate inhibitor of UDP-glucose dehydrogenase, an enzyme related to the development of prostate cancer. [50] Under batch conditions, this reaction rendered a mixture of purpurogallin (1) as major product (14 % yield) and 10 bd (7 % yield). [15] Any attempt to improve this ratio, by increasing the equivalents of 9 d used was unsuccessful in our hands. Interestingly, under continuous flow regime, compound 10 bd was isolated as the major product of the reaction, although in a very modest 7 % yield (10 bd/1 2 : 1). Opposite to batch, an increase in the amount of 9 d resulted in the isolation of benzotropolone 10 bd in 16 % yield (10 bd/1 5 : 1). Further tuning of our previously optimized flow setup, by means of introduction of the two coupling partners separately in the reactor and mixed with a third stream containing the oxidative solution, rendered better results, both in terms of selectivity (10 bd/1 7 : 1), and reaction yield (29 %). The yield, albeit low, highlights the potential for modulation and/or inversion of selectivity on reactions performed in continuous flow reactors, mainly due to its precise control of reaction conditions and modularity of the setup. (Entries 10-12, Table 1). Once a continuous-flow process for the preparation of benzotropolones was developed, our efforts were then focused in the development of a continuous-flow methodology to perform the other synthetic steps involved in the preparation of 7 (i. e. continuous flow fischer esterification of 3,4,5-trihydroxybenzoic acid 8 c (gallic acid), and the protection of phenol groups present in benzotropolone 10 aa, followed by Suzuki coupling to render 11) (Figure 2). After straightforward optimization, methyl gallate 8 a was obtained in quantitative yield by reaction of two methanolic solutions of 8 c (0.3 mL/min), and sulfuric acid (0.3 mL/min), in a coiled reactor (1 mm i.d., 20 mL) at 80°C. [49] Following this procedure, 9 g of 8 a were prepared after 5 hours of continuous production (43 g/day throughput). Besides the inherent advantages of continuous flow reactions, we have remarkably improved the reaction yield of this process (Batch result: 20 h reaction time, 89 % yield). [15] Then, the continuous flow protection of phenol groups was investigated. By simply passing a solution of 10 aa and dimethyl sulfate in DMF through an Omnifit® column containing solid potassium carbonate, the desired protected analogue 11 was isolated in 97 % yield, within 4 min residence time and a daily throughput of 12 g.. [49] To our delight, the outcome solution of the formed product 11 in DMF can be stored without further purification in a reservoir and on demand transferred to a batch reactor containing an aqueous solution of phenyl boronic acid, Na 2 CO 3 , and Pd catalyst, performing the Suzuki coupling reaction, using a previously described batch methodology that rendered benzotropolone 12 in 58 % yield.
Eventually, we screened different reaction conditions in order to adapt the previously described BBr 3 mediated deprotection of phenols [15] to continuous flow. (Table 2) We started our optimization process by adjusting pump flow rates to mix a 0.1 M solution of 12 in DCM (0.25 mL/min) with a 0.5 M solution of BBr 3 in DCM (0.25 mL/min) and the resulting mixture reacted in a 10 mL coiled reactor with a residence time of 20 min. Unfortunately, under these conditions marginal yields of product were obtained while clogging issues at T-piece connection were observed, preventing from the reaction to be run in a continuous manner. (Table 2, entry 1) When a more diluted solution of BBr 3 (0.25 M) was used, together with an increase in the flow rate, the yield of our transformation was significantly increased. Maintaining the residence time value fixed, but doubling the amount of BBr 3 increasing the concentration of the stock solution, further increased the yield to 40 %, while higher concentration values eventually result in solid accumulation and blockage at the Tpiece connector. (Table 2, entries 2-4) During the screening process we also noticed degradation of different parts of the setup (mainly PTFE tube and rubber seals) due to the prolonged exposure to BBr 3 solution so, for reactions maintained longer than 1 hour, PFA tubing and syringe pumps were used instead. Furthermore, in order to avoid clogging issues, a mixer, made by a omnifit® glass column that incorporates stirring bars in constant movement, [51] was introduced before the reactor to avoid solid settling. This new setup increased the yield of the reaction to 55 % yield. (Table 2, entry 5) Residence time enlargement proved to be highly beneficial for the reaction outcome, ( Table 2, entries 6 and 7) isolating the desired product 7 in 89 % yield. The optimized continuous flow setup presents a residence time of 13 min and a reaction throughput for the active epigenetic modulator 7 of approx. 1.7 g/day.
In conclusion, we have described a continuous methodology for the preparation of benzotropolone rings that complements the previously reported procedure and further demonstrates that continuous flow reactors are privileged platforms for the efficient preparation of biologically active molecules. We have developed an efficient and reproducible protocol for the enzyme mediated formation of benzotropolone rings, expanding the toolbox of methodologies available for the preparation of this structural motif present in several relevant secondary metabolites. We have demonstrated the suitability of continuous flow methodologies for the preparation of epigenetic modulator 7. The described methodology renders enough material for further biological characterization and paved the way towards the routine preparation of a broader set of analogues.

Supporting information Summary
Experimental section, with full description of flow setup, further optimization details, and characterization data were reported in the supporting information.