Synthesis of antimalarial trioxanes via continuous photo-oxidation with ¹O₂ in supercritical CO₂

Abstract

The oxidation of an allylic alcohol to its hydroperoxides represents a key step in the synthesis of a series of spirobicyclic, antimalarial trioxanes. Herein, we investigate the continuous photo-oxidation of an allylic alcohol with ¹O₂ in scCO₂, as a ‘green’ alternative to conventional methods, and examine the remaining two steps in the synthesis of antimalarial trioxanes from readily available starting materials.

---

Malaria is one of the most devastating infectious diseases in the world, killing an estimated 655 000 people in 2010 alone.¹ Of crucial concern is the fact that the virus Plasmodium falciparum, responsible for the most lethal form of malaria, malaria tropica, is becoming increasingly resistant to former front-line quinine based antimalarials² and there is a need for new antimalarial compounds to be produced.

One example is artemisinin,^3,4 a naturally occurring lactone with a 1,2,4-trioxane core structure.^5,6 Synthetic derivatives of artemisinin also possess excellent antimalarial activity and, despite concerns that P. falciparum is also showing signs of resistance to these compounds, artemisinin and its analogues are still the primary treatments for falciparum malaria.^3,7 Artemisinin is still commonly extracted from the plant Artemisia annua because its synthesis is labour-intensive³ and unsuited to large-scale production.

A feature common to artemisinin and several other potential antimalarials is the trioxane moiety, shown in Scheme 1. Photochemically generated ¹O₂ offers a convenient way of introducing such a motif. However, the excitation of ground state molecular oxygen, ³O₂ to ¹O₂, requires the use of photocatalysts. The highly reactive nature of ¹O₂ can mean that there are problems with the scale-up of such processes as industrially acceptable solvents must be non-flammable and inefficient ¹O₂ quenchers.

Scheme 1 Synthetic route to spirobicyclic trioxanes, 4a and 4b, highlighting the photo-oxidation step. 1: mesityl oxide; 2: 4-methylpent-3-en-2-ol; 3: hydroperoxides. Here, n = 1, cyclopentanone, or 2, cyclohexanone to give 8-isopropenyl-9-methyl-6,7,10-trioxa-spiro[4.5]decane (4a) or -isopropenyl-4-methyl-1,2,5-trioxa-spiro[5.5]undecane (4b) respectively.

Recent work has resulted in the partial scale-up of the photo-oxidation of the natural product artemisinic acid to artemisinin using CH₂Cl₂ as the solvent.⁸ Apart from the nature of the solvent, the problem of obtaining the artemisinic acid still remains; like artemisinin, it is usually extracted from A. annua, although it can now be produced by engineered yeast.⁹

A number of new antimalarial trioxanes, including 4a and 4b, Scheme 1, have also been synthesised using ¹O₂ as a key step. In this work, the photo-oxidation procedure was conducted in a batch reactor using either non-polar solvents or a solvent-free approach with porphyrin loaded polystyrene beads.^10,11 The addition of spiroadamantane motifs was found to increase antimalarial activity, however the products were hardly water soluble, indicating that a diverse approach to a variety of such compounds is necessary and that more synthetic work is required to evaluate the best antimalarial compounds.¹²

We have previously shown that photo-oxidation using ¹O₂ can be conducted continuously and safely in supercritical carbon dioxide, scCO₂, which provides a non-toxic and non-flammable reaction medium and which can be used close to ambient temperature.^13–15 The photo-oxidation proceeds somewhat faster in scCO₂ than in many conventional solvents, probably because gaseous O₂ and scCO₂ are fully miscible, allowing for single phase reaction conditions. In addition, ¹O₂ has a relatively long lifetime in CO₂ (5.1 ms at 14.7 MPa, 314 K, compared to 59 ms for CCl₄).^16,17 By using a flow system, the reactor volume is reduced compared to a batch reactor of similar productivity, giving a higher degree of safety in these potentially dangerous systems. Recently we also demonstrated that the photocatalyst could be recycled in a closed loop whilst maintaining a homogeneous reaction system¹⁸ by incorporating a fluorous biphase into our existing process.

In this paper, we report the synthesis of the spirobicyclic trioxanes (4a and 4b), with particular emphasis on the continuous photo-oxidation of the allylic alcohol (2), to its hydroperoxides (3), conducted in scCO₂, see Scheme 1.¹¹ The continuous conversion of mesityl oxide (1) to the allylic alcohol (2) via heterogeneously catalysed, high temperature transfer hydrogenation at ca. 20 atm pressure has also been investigated. In the final step, Lewis acid catalysed condensation of 3 with either cyclopentanone or cyclohexanone yields 4a and 4b respectively. Before scale-up of the photo-oxidation of 2, we carried out the reaction in the small scale batch reactor which had been used in our previous work.¹³ As before, we used the photosensitiser 5,10,15,20-tetrakis-(pentafluorophenyl)porphyrin (TPFPP) to generate ¹O₂. TPFPP is insoluble in 2 but, conveniently, we found that the cyclic ketones, used for the final transformation to 4, could also act as suitable co-solvents.

TPFPP, 2 and the cyclic ketone were injected into the cell prior to pressurisation to 18 MPa, to ensure single phase reaction conditions, at 40 °C with 6% O₂ in CO₂. A twofold excess of the cyclic ketone, with respect to 2, was used. The reaction was found to follow zero-order kinetics with respect to 2, as determined by FTIR monitoring, Fig. 1. In our scCO₂ system, the reaction ran with quantitative conversion of 2 to 3, in the presence of either cyclopentanone or cyclohexanone; however 20% of 2 remained unreacted under these conditions. The oxidation of 2, Scheme 2, leads to both the syn and anti isomers of 3, the ratio of which is solvent dependent.¹⁹ With either cyclopentanone or cyclohexanone, the syn selectivity in scCO₂ (d.r. 85 : 15) was found to be higher than in MeOH (d.r. 73 : 27) but lower than in CCl₄ (d.r. 93 : 7), in agreement with the observation that non-polar solvents favour formation of the syn isomer and that the hydroxyl group of the allylic alcohol enhances selectivity.^19,20

Fig. 1 FTIR monitoring of the overall formation of the isomers of 3via the ν(C–H) band at 3083 cm⁻¹ during the photo-oxidation of 2 in an unstirred batch reaction. The cell was irradiated with a single white 1000 lumen LED¹² for 60 s and an IR spectrum was recorded before further irradiation until the peak height of the band at 3083 cm⁻¹ no longer changed. 50 μL of 2 + cyclopentanone (1 : 2, mol : mol) was used with a 2 mm path length, 0.78 mL volume, cell. NMR confirms a ca. 80% yield of 3 for this experiment. Time indicates the overall irradiation time; the reaction reaches a maximum of 80% yield of 3; ca. 20% of 2 remained unconverted.

Scheme 2 The photo-oxidation of 2 to its syn- and anti-hydroperoxides.

Following these successful batch experiments, the photo-oxidation of 2 could be scaled up to a continuous flow process. Fig. 2 shows a schematic of our continuous ¹O₂ photo-reactor with twin irradiation vessels. As the reaction was found to proceed more slowly than in the examples we have published previously,^13,14 the total system flow rate had to be lowered accordingly. In this work, the flow rate of CO₂ was maintained at 0.3 mL min⁻¹ and 2 was pumped with TPFPP and the cyclic ketone, again in a twofold excess, at a flow rate of 0.03 mL min⁻¹. In this way, the residence time was increased to ca. 40 min with the aim to maximise conversion of 2. Under these continuous flow conditions, at 18 MPa and with 6% O₂ in CO₂, we obtained an 86% maximum yield of 3, slightly higher than in batch but with the same selectivity for the syn-isomer of 3 (d.r. 85 : 15), as determined by ¹H-NMR.

Fig. 2 The twin photochemical flow reactor,‡ used to maximise the yield of 3. CO₂ is delivered by a Jasco PU-1580-CO2 pump at 0.3 mL min⁻¹, O₂ is added using a Rheodyne dosage unit. 2 is pumped with the cyclic ketone (2 : 1 ketone : 2, mol : mol) and TPFPP (2 mg mL⁻¹2) at 0.03 mL min⁻¹ using a Jasco PU-980 HPLC pump. B: ice bath; BPR: back-pressure regulator (Jasco BP-1580-81); F: round bottomed flask with or without 100 mL CH₂Cl₂ + 0.2 mL BF₃·Et₂O; LEDs: light source (4 × 5 × Citizen Electronics Co. Ltd CL-L233-C13 N on an Al heat sink); M1 & M2: mixers; R1 & R2: sapphire tube reactors. The total pressure was 18 MPa.

With a modification of the equipment, we found that our perfluorinated photosensitiser¹⁸ could be pumped in a fluorous solvent, HFE-7500, separately from 2 and could be recycled. The yield and selectivity for 3 was unaffected by this recycling at the same flow rates and pressure used for the single pass experiment. This biphasic technique should allow a wider range of trioxanes to be synthesised because one no longer requires a ketone which dissolves TPFPP.

The obvious next step was to investigate the possibility of carrying out each stage of the synthesis continuously. Therefore, we briefly investigated the continuous reduction of 1 to 2 and the Lewis acid catalysed condensation of 3 to yield 4, the aim being to identify any problems in the development of a ‘greener’ route to 4.

The reduction of 1 to 2 is traditionally conducted²¹ using LiAlH₄, see ESI† for details. However, LiAlH₄ is inconvenient to handle in continuous processes. Hydrogenation with H₂ in scCO₂ is highly efficient^22,23 but is unlikely to have sufficiently high selectivity towards hydrogenation of the carbonyl group in this system. Therefore, we have focussed on the H-transfer reduction of 1, which has previously been investigated over a variety of metal oxide catalysts.^24,25 It has been reported that 1 can be selectively reduced to 2 over basic MgO with a high surface area, using iso-propanol (IPA) as the H-transfer agent, with yields of 2 of 28%.²⁴ It was suggested that the weak acid–strong base Mg²⁺–O²⁻ pair sites of MgO promote a Meerwein–Ponndorf–Verley mechanism between 1 and IPA.²⁵

We have investigated the continuous reduction of 1 with IPA with a view to improving the yield of 2 as an inexpensive alternative to LiAlH₄ and we synthesised MgO with a BET surface area of ca. 84 m² g⁻¹. Continuous experiments† were conducted using a high pressure flow rig incorporating a reactor filled with 1 g MgO. The effect of changing the reactor parameters was investigated and it was found that, at our optimum conditions, 2 could be obtained in >50% yield over a relatively long period, as determined by online GLC, see Fig. 3. As the by-product of this reaction, acetone, cannot safely be used with peroxides in the planned next step, 2 must be separated from the IPA and acetone prior to photo-oxidation.

Fig. 3 Summary of results obtained for the high pressure (2 MPa) continuous conversion of 1 to 2 using IPA over an MgO catalyst in the absence of added solvent with a ratio of IPA : 1 of 10 : 1 unless otherwise stated. (a) Increasing the temperature where a maximum yield of 2 was observed at the highest temperature; (b) increasing the ratio IPA : 1 where the highest ratio gave the highest yield; (c) increasing the flow rate of the IPA + 1 mixture where the yield reaches a maximum at 0.3 mL min⁻¹ and (d) an extended run at 350 °C demonstrating near stable performance of the catalyst for >24 h. In (a)–(d), the traces are labelled as follows; ●: 2; ○: 1; ▲: 4-methylpent-4-en-2-ol; X: iso-mesityl oxide.

The final step, conversion of 3 to 4, requires a Lewis acid catalyst. We approached this conversion in two ways. Initially, the product stream from the ¹O₂ flow experiment, containing 3 plus the cyclic ketone, was collected directly into a round bottomed flask containing CH₂Cl₂ (100 mL) and BF₃·Et₂O (0.2 mL) so as to quench the peroxide products immediately. The flask was immersed in an ice bath, the temperature was gradually increased to room temperature and the mixture was stirred for a further 12 h. The product, 4b, was purified by thick layer column chromatography, see ESI† for details. This gave an overall yield of 4b of ca. 25%, similar to that reported previously for this step.¹¹

Since CH₂Cl₂ is not a particularly ‘green’ solvent, we tried a solvent-free approach to converting 3 to 4a. A stainless steel tubular reactor (1/4′′ o.d., 2.5 mL volume) was packed with silica/BF₃ (Aldrich) and the solution of 3 in cyclopentanone was pumped through at a rate of 0.05 mL min⁻¹. Similar conversion was achieved as with CH₂Cl₂, 30% yield of 4a, but BF₃ leached into the solution and the conversion dropped off after 20 min.

Nevertheless, this demonstrates that an additional solvent is not required and raises the prospect of using a more robustly supported Lewis acid.

Conclusions

We have synthesised two spirobicyclic trioxanes, 4a and 4b, which have previously been shown to exhibit antimalarial activity¹¹ and have demonstrated that the photo-oxidation of an allylic alcohol, a key step in this synthesis, can be scaled up to continuous flow in scCO₂. We have also investigated the continuous H-transfer reduction of 1 to 2 and the Lewis acid catalysed final step of the transformation, with the aim of developing a fully continuous, sustainable process for the synthesis of antimalarial trioxanes.

Additionally, we have indicated how the photo-oxidation can be developed further by applying our recently developed fluorous biphase technique¹⁸ for photo-oxidations, which allows the photosensitiser to be pumped into the system despite being insoluble in the starting material.

Furthermore, the ketone for the final step could be added after the photo-oxidation, opening up the possibility of using ketones with functionalities which would be destroyed in the presence of ¹O₂. Also, adding the ketone downstream of the reactor could be used to generate a whole series of spirobicyclic compounds by sequentially pulsing different ketones into the flowing stream of 3. Ultimately, we hope that our approach can lead to libraries of antimalarial trioxanes and an exploration of such compounds synthesised from readily available starting materials.

We thank the EPSRC DICE project for support and 3M for a sample of HFE-7500. M.W.G. acknowledges a Royal Society Wolfson Merit Award. We thank M. Dellar, M. Guyler, D. Litchfield, R. Wilson and P. Fields for technical support, S. Aslam for NMR assistance, E. Masika for BET analysis and Z. Amara for helpful discussions.

Notes and references

1. M. Aregawi, R. Cibulskis, M. Lynch and R. Williams, World Malaria Report 2011.
2. S. Foote and A. Cowman, Acta Trop., 1994, 56, 157.
3. S. Mok, M. Imwong, M. J. Mackinnon, J. Sim, R. Ramadoss, P. Yi, M. Mayxay, K. Chotivanich, K. Liong, B. Russell, D. Socheat, P. N. Newtone, N. P. J. Day, N. J. White, P. R. Preiser, F. Nosten, A. M. Dondorp and Z. Bozdech, BMC Genomics, 2011, 12, 391.
4. D. Klayman, Science, 1985, 228, 1049.
5. Y. Wu, Acc. Chem. Res., 2002, 35, 255.
6. R. Haynes and S. Vonwiller, Acc. Chem. Res., 1997, 30, 73.
7. N. J. White, Science, 2008, 320, 330.
8. F. Lévesque and P. H. Seeberger, Angew. Chem., Int. Ed., 2012, 51, 1706.
9. D. K. Ro, E. M. Paradise, M. Ouellet, K. J. Fisher, K. L. Newman, J. M. Ndungu, K. A. Ho, R. A. Eachus, T. S. Ham, J. Kirby, M. C. Y. Chang, S. T. Withers, Y. Shiba, R. Sarpong and J. D. Keasling, Nature, 2006, 440, 940.
10. A. G. Griesbeck and A. Bartoschek, Chem. Commun., 2002, 1594.
11. A. G. Griesbeck, T. T. El-Idreesy, M. Fiege and R. Brun, Org. Lett., 2002, 4, 4193.
12. A. G. Griesbeck, T. T. El-Idreesy, L. Höinck, J. Lex and R. Brun, Bioorg. Med. Chem. Lett., 2005, 15, 595.
13. R. A. Bourne, X. Han, A. O. Chapman, N. J. Arrowsmith, H. Kawanami, M. Poliakoff and M. W. George, Chem. Commun., 2008, 4457.
14. R. A. Bourne, X. Han, M. Poliakoff and M. W. George, Angew. Chem., Int. Ed., 2009, 48, 5322.
15. X. Han, R. A. Bourne, M. Poliakoff and M. W. George, Green Chem., 2009, 11, 1787.
16. D. R. Worrall, A. A. Abdel-Shafi and F. Wilkinson, J. Phys. Chem. A, 2001, 105, 1270.
17. F. Wilkinson, W. P. Helman and A. B. Ross, J. Phys. Chem. Ref. Data, 1995, 24, 663.
18. J. F. B. Hall, X. Han, M. Poliakoff, R. A. Bourne and M. W. George, Chem. Commun., 2012, 48, 3073.
19. A. G. Griesbeck, W. Adam, A. Bartoschek and T. T. El-Idreesy, Photochem. Photobiol. Sci., 2003, 2, 877.
20. W. Adam and B. Nestler, J. Am. Chem. Soc., 1992, 114, 6549.
21. M. E. Cain, J. Chem. Soc., 1964, 3532.
22. M. G. Hitzler, F. R. Smail, S. K. Ross and M. Poliakoff, Org. Process Res. Dev., 1998, 2, 137.
23. X. Han and M. Poliakoff, Chem. Soc. Rev., 2012, 41, 1428.
24. J. Kaspar, A. Trovarelli, M. Lenarda and M. Graziani, Tetrahedron Lett., 1989, 30, 2705.
25. J. I. Di Cosimo, A. Acosta and C. R. Apesteguía, J. Mol. Catal. A: Chem., 2004, 222, 87.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c2gc36711d

‡ Safety warning: These reactions involve high pressures and require an appropriately rated apparatus and with due regard to the potentially explosive reaction between O₂ and organic compounds.

---