Syntheses of deuterium-labeled dihydroartemisinic acid (DHAA) isotopologues and mechanistic studies focused on elucidating the conversion of DHAA to artemisinin†

Kaitlyn Varela and Francis K. Yoshimoto *
Department of Chemistry, The University of Texas San Antonio (UTSA), One UTSA Circle, San Antonio, TX 78249-0698, USA. E-mail: francis.yoshimoto@utsa.edu

First published on 9th August 2024

---

---

Kaitlyn Varela

Dr Kaitlyn Varela graduated with a B.S. degree in Biochemistry in May 2020 and obtained her Ph.D. in Chemistry in May 2024 from the University of Texas at San Antonio (UTSA). She has published four first author research articles in the J. Nat. Prod. (2020, 2021, 2022, and 2023) based on her work that uses organic chemistry to solve biological problems. Her doctoral research has been focused on studying how the plant Artemisia annua (i) produces an antimalarial medicine (artemisinin) through a cascade reaction involving molecular oxygen and (ii) has anticancer and antiviral properties (arteannuin B) by blocking cysteine proteases.

Francis K. Yoshimoto

Dr Francis Yoshimoto obtained his undergraduate degrees (B.S. in Chemistry and B.A. in Linguistics) from UC Berkeley and his graduate degree (Ph.D. in Biochemistry) from the UT Southwestern Medical Center. He conducted undergraduate research in the lab of Prof. Richmond Sarpong (UC Berkeley), doctoral research in the lab of Prof. Richard Auchus (UT Southwestern and University of Michigan Medical School), and postdoctoral research in the lab of Prof. Fred Guengerich (Vanderbilt University). In 2016, Dr Yoshimoto established his independent research career at the University of Texas at San Antonio (UTSA) in the Department of Chemistry, where, as an Assistant Professor, he trains his students to work at the interface of chemistry and biology to benefit human health.

---

Introduction

Dihydroartemisinic acid (DHAA)¹ is the biosynthetic precursor of artemisinin (Fig. 1, 1 to 2), which is a sesquiterpenoid (i.e., 1.5 terpene units or 3 isoprene units made up of 15 carbons) from Artemisia annua and contains an endoperoxide bridge. The endoperoxide moiety gives artemisinin its antimalarial properties.^2,3 Due to its intriguing bioactive properties, many efforts have focused on the synthesis of artemisinin,⁴ and its biosynthesis has been a popular topic of study to develop synthetic biology approaches for mass producing artemisinin.^5,6 Although the direct biosynthetic precursor of artemisinin is DHAA, less research effort has been focused on DHAA.

DHAA undergoes a spontaneous autoxidation process to produce artemisinin,^7,8 which can be described in four steps (Fig. 1: (i) first O₂ incorporation through a Schenck ene⁹ reaction (i-a to i-b), (ii) C4–C5 bond cleavage (ii-a to ii-c), (iii) second O₂ incorporation, and (iv) polycyclization (iv-a to iv-c)). The conversion is initiated through (i) the incorporation of one molecule of oxygen into DHAA to yield hydroperoxide 3. This reaction between ¹O₂ (O₂ that reacts with a photosensitizer¹⁰ and light) and DHAA (1) first forms perepoxide intermediate 1-i through step (i-a), and the resulting perepoxide (1-i) rearranges to form hydroperoxide 3 as shown in step (i-b).

In step (ii), the resulting hydroperoxide 3 undergoes cleavage of the C4–C5 bond through Hock cleavage^11–13 to first give (ii-a) oxocarbenium 3-i, which subsequently undergoes (ii-b) hydration to afford cyclic hemiketal 3-ii, and (ii-c) the cyclic hemiketal moiety of 3-ii yields enol ketone 4.

Step (iii) involves the incorporation of a second molecule of oxygen at C6 of enol 4 to give peroxide 5. This step is believed to involve the reaction of the enol with triplet oxygen (³O₂) as it occurs without light¹⁴ and could occur in two parts. First, step (iii-a) involves O–H hydrogen atom abstraction of the enol moiety^15,16 by ³O₂ to yield C6-radical 4-i. Subsequently, step (iii-b) occurs, where the C6-radical of 4-i recombines with the hydroperoxide radical to yield hydroperoxide 5.

The final step (iv) is the polycyclization of carboxy-keto-aldehyde 5 to yield the endoperoxide (Fig. 1, 5 to 2), which likely begins with the protonation of the C4-ketone to lead to 5-i (iv-a). The activated ketone (5-i) undergoes intramolecular nucleophilic attack by the peroxide at C6 to give cyclic peroxyhemiketal 5-ii (iv-b). The resulting C4-hydroxy group of 5-ii attacks the C5-aldehyde (iv-c), which in turn leads to the attack of the C12-carboxylic acid and the loss of a water molecule to form the lactone (2). The initial ene reaction between singlet oxygen and DHAA shown in Fig. 1 (step (i), 1 to 3) could potentially occur at C15 and C3 also to form C5-allylic hydroperoxide regioisomers with the double bond at C15–C4 and C3–C4, respectively (cf. Fig. 16).

Several reports have described the conversion of DHAA to artemisinin through chemical processes using singlet oxygen to generate an allylic hydroperoxide (Fig. 1, 1 to 3), which would subsequently convert to artemisinin in the presence of an acid. Table 1 summarizes some of the existing published methods that discuss the conversion of DHAA to its hydroperoxide with singlet oxygen, followed by endoperoxide formation with triplet oxygen,^12,17–20 with yields ranging from 14% to 69%. The reports in Table 1 include the original study reporting the conversion of DHAA to artemisinin published by Roth and Acton in 1989 (entry 1),¹⁷ flow techniques (entries 2–4),^12,18,19 and the industrial process²⁰ (entry 5). Other studies that report endoperoxide-forming reaction conditions have also been previously reviewed.²¹

---

We have reviewed in the ESI the previously reported chemical syntheses of DHAA (see ESI Part 1) from (+)-citronellal^23,24 and (−)-isopulegol,²⁵ and the biosynthesis of DHAA (see ESI Part 2†) from farnesylpyrophosphate (FPP) through five enzyme systems (i.e., (i) amorphadiene synthase or ADS,²⁶ (ii) cytochrome P450 71AV1²⁷ and P450 reductase,²⁸ (iii) artemisinic alcohol dehydrogenase 1 or AaADH1,²⁹ (iv) double bond reductase or DBR,³⁰ and (v) dihydroartemisinic aldehyde dehydrogenase³¹ or Aldh1). Although the conversion of DHAA to artemisinin has been reported (Table 1), there is still opportunity to optimize the transformation to produce artemisinin on a practical scale (Fig. 1, 1 to 2). For instance, the use of halogenated solvents³² such as CH₂Cl₂ may not be desired because they are not environmentally friendly (see Table 1). Therefore, understanding the detailed mechanism of this conversion can yield possible solutions to improve the process of converting DHAA to artemisinin on a large scale.

This review article will focus on the syntheses of deuterated DHAA isotopologues and their use as mechanistic probes to determine how the nonenzymatic endoperoxide forming step to form artemisinin occurs. The main text is divided into two sections:

(I) Syntheses of regioselectively deuterated DHAA isotopologues

(II) Efforts towards elucidating the mechanism of conversion of DHAA to artemisinin.

Syntheses of regioselectively deuterated DHAA isotopologues

Deuterium incorporation of small molecules has been a recent trending research topic, and one benefit of deuterium incorporation is its ability to mechanistically interrogate chemical transformations.³³ With respect to artemisinin biosynthesis, deuterium was incorporated into DHAA so that the mechanism of endoperoxide formation could be elucidated. For instance, Brown and Sy used 15-¹³C-d₃-DHAA (Fig. 2A, 1a) that was synthesized for incorporation into the Artemisia annua plant, which was either kept alive or allowed to die after incorporation of the labeled molecule.³⁴ The conversion to artemisinin was monitored by ²H NMR spectroscopy to determine the nonenzymatic formation of DHAA to artemisinin within the plant. In contrast, our research lab has used deuterated DHAA isotopologues (Fig. 2B, 1b and 1c, 3,3-d₂-DHAA and 15,15,15-d₃-DHAA, respectively) to measure the rate of nonenzymatic endoperoxide formation using mass spectrometry.^7,8

Isotope incorporation has been accomplished to yield deuterated DHAA derivatives through either a ¹³C-d₃-MeMgI reagent^35,36 or LiAlD₄^7,8 as the deuterium source (Fig. 2A or 2B, respectively). Deuterium was incorporated either at the C15^8,35,36 position or at the C3⁷ position of DHAA. Brown and Sy have reported the use of intramolecular aldol condensation–dehydration to form the enone intermediate, which was synthesized from artemisinin (Fig. 2A, Table 2: entries 1 and 2)^35,36 similar to the prior total syntheses that employed an intramolecular aldol^24,25 to form the enone (see ESI Part 1†).

---

Despite the recent chemical efforts to introduce deuterium into small molecules through C–H activation techniques,^33,37,38 the inability to fully deuterate at only the desired position would limit the use of these new methodologies. Therefore, our research laboratory used DHAA as the starting feedstock and regioselectively oxygenated the DHAA carbon backbone at C3 or C15 through either Riley oxidation or Rubottom oxidation, respectively. The oxygenated DHAA derivatives would subsequently be reduced with commercially available LiAlD₄ (Fig. 2B: 10 to 1b, or 12 to 1c, Table 2: entries 3 or 4).^7,8 Deuteration at C3 of DHAA was accomplished by treating the 3-keto intermediate (10) with LiAlD₄ and AlCl₃ to yield 3,3-d₂-DHAA (1b). On the other hand, deuteration at C15 was executed through the treatment of carboxylic acid 12 with LiAlD₄. The 15-carboxylic acid was obtained through the cleavage of the C4–C5 bond of DHAA to yield silyl enol ether 11. Silyl enol ether 11 underwent Rubottom oxidation to incorporate the C15–oxygen and the C4–C5 bond was eventually re-formed through a Grubbs ring closing metathesis (RCM) reaction.

First reported synthesis of 15-¹³C-d₃-DHAA³⁵ (loss of deuterium during hydrogenation of a diene intermediate)

A 15-¹³C-d₃-isotopologue of dihydroartemisinic acid (1a) was synthesized from artemisinin in 4 steps (Fig. 3–5).³⁵ The synthesis began with the acid degradation of artemisinin using sulfuric acid and methanol to produce a diketone (Fig. 3, 2 to 13). Diketone 13 was then cyclized with barium hydroxide octahydrate and found to undergo lactonization when worked up under acidic conditions (13 to 6 to 6b); however, when potassium hydroxide was added to the lactone side product, it regenerated the desired enone (6b to 6).

Enone 6 was then treated with ¹³CH₃MgI (Fig. 4, step 3a, 6 to 14a to 15), ¹³C²H₃MgI (step 3c, 6 to 14b to 7), or with diazomethane, and then treated with MeMgI (Fig. 4, step 3b, 6 to 16a–d). When a Grignard reagent, ¹³CH₃MgI, was added to the enone when the reaction was worked up at pH 4–5 (Fig. 4a), the predominantly formed compound was 15-¹³CH₃-dihydro-epi-deoxyarteannuin B (15). This result is in contrast to when ¹³C²H₃MgI was used, followed by a more acidic work-up (pH 1–2, Fig. 4c), which resulted in primarily 15-¹³C²H₃-6,7-dehydro-11,13-dihydroartemisinic acid (7), with small amounts of isomer and 15-¹³C²H₃-dihydro-epi-deoxyarteannuin B (18). Interestingly, when the methyl ester derivative of enone 6 (Fig. 4b) was subjected to the MeMgI Grignard reagent, the resulting α- and β-alcohol intermediates were isolatable, which, upon acid work-up, resulted in a dehydration product (16a) and three other minor E₁ elimination products (16b, 16c, and 16d).

Finally, to synthesize 15-¹³C-d₃-dihydroartemisinic acid, the palladium catalyzed hydrogenation of the diene resulted in a mixture of five regioisomers and stereoisomers of isotopically labeled DHAA, which were separated by HPLC (Fig. 5, 7 to 1a and minor regioisomers 19, 20, 21, and 22). It was noted that there is some unavoidable loss of the deuterium label from the 15-position during this final hydrogenation step, resulting in a mixture of the four possible isotopologues (i.e., d₀-, d₁-, d₂-, and d₃-DHAA) rather than complete deuteration (i.e., 15-¹³C-d₃-DHAA). It should also be noted that Brown et al. described the spontaneous autooxidation of dihydroartemisinic acid to artemisinin in the absence of a photosensitizer, when left in a freezer for six months without any quantification of the rate of this process.³⁵

Second synthesis of 15-¹³C-d₃-DHAA³⁶ (circumventing loss of deuterium during synthesis)

To avoid the loss of deuterium at C15 during the hydrogenation of diene 7 to 1a (Fig. 5), a subsequent study involved performing the hydrogenation step with H₂ and Pd/C on enone 6³⁵ to yield ketone 8 (Fig. 6, 2 to 6 to 8).³⁶ After hydrogenation, the cis-decalone intermediate (6 to 8) was obtained as the major diastereomer, with trans-decalone intermediate 23 being less than 10% of the crude reaction. This diastereomeric mixture was carried forward and separated after the final step with CuSO₄ (9 to 1a). The trans-decalin diastereomer (25) and olefin regioisomers (19 and 26) were separated by HPLC. The retention of the deuterium label was close to 100% determined through NMR spectroscopy³⁶ and the overall yield was improved compared to the previously³⁵ published synthesis. The authors used the synthesized DHAA isotopologue to study the conversion of DHAA to artemisinin in the plant. They followed the autoxidation process by NMR spectroscopy.³⁴

Synthesis of 3,3-d₂-DHAA (1b)⁷

A 3,3-d₂-dihydroartemisinic acid isotopologue was synthesized from non-deuterated dihydroartemisinic acid in 7 steps (Fig. 7, 1b from 1). The carboxylic acid of dihydroartemisinic acid was reduced with LiAlH₄ (1 to 27). The resulting alcohol was protected as acetate 28 using acetic anhydride in pyridine. Riley oxidation using SeO₂ on acetate 28 gave allylic alcohol 29. Allylic alcohol 29 was then oxidized to enone 10 with the Dess–Martin periodinane, which was reduced with LiAlD₄ and AlCl₃ to yield a separable mixture of diene 31 and dideuterated alcohol 32 through the monodeuterated intermediate 30. The dideuterated alcohol 32 was oxidized to aldehyde 33 with Dess–Martin periodinane. Aldehyde 33 was treated under Pinnick oxidation conditions to afford 3,3-d₂-dihydroartemisinic acid (1b).

Synthesis of 15,15,15-d₃-DHAA (1c)⁸

A 15,15,15-d₃-dihydroartemisinic acid isotopologue was synthesized from non-deuterated dihydroartemisinic acid in 25 steps (Fig. 8 and 9, 1 to 1c). The synthesis began with the oxidation of dihydroartemisinic acid with RuCl₃, NaIO₄, and sulfuric acid to produce lactone 34. Lactone 34 was reduced to triol 35 by refluxing with LiAlH₄. The resulting vicinyl diol (35) was cleaved using NaIO₄ to yield lactol 36, which was treated with TBDMSCl and imidazole to undergo ring opening and protection of the primary alcohol as the TBDMS ether, to yield ketoaldehyde 37. Ketoaldehyde 37 was reduced to diol 38 with LiAlH₄, and then selective protection of the primary alcohol over the secondary alcohol was performed using acetyl chloride and triethylamine (38 to 39). Primary acetate 39 was then subjected to triethylsilyl chloride to protect the secondary alcohol to give TES ether 40. The primary acetate of 40 was subsequently deprotected with methyl lithium to afford 41. The resulting primary alcohol was oxidized to aldehyde 42 using the Dess–Martin periodinane, which was olefinated using methylene triphenylphosphorane to yield alkene 43. The TES ether was selectively deprotected over the TBDMS ether using pyridinium para-toluenesulfonate (PPTS) in methanol to yield secondary alcohol 44. To install the C15 alcohol, the secondary alcohol was oxidized to methyl ketone 45 with the Dess–Martin periodinane. Ketone 45 was treated with triethylsilyl triflate to produce silyl enol ether 11. Rubottom oxidation with meta-chloroperoxybenzoic acid was then performed to generate the C15 oxygenated compound 46. Wittig olefination of the ketone gave diene 47, which was deprotected with PPTS in methanol to yield diene alcohol 48, the diene precursor of the Grubbs ring closing metathesis reaction (Fig. 9, 48 to 49).

Ring closing metathesis of diene 48 with a Grubbs 2^nd generation catalyst resulted in the restoration of the C4–C5 double bond of DHAA. To deuterate the C15 position of dihydroartemisinic acid, the allylic alcohol at C15 (49) was oxidized to aldehyde 50 with the Dess–Martin periodinane and then oxidized to carboxylic acid 12 under Pinnick oxidation conditions. The resulting carboxylic acid (12) was then reduced with LiAlD₄ to yield alcohol 51, which was mesylated with methanesulfonyl chloride to furnish mesylate 52. The resulting C15-d₂-mesylate was reduced using another portion of LiAlD₄ to yield the C15-trideuterated methyl intermediate 53. With the three deuterium isotopes incorporated at the C15 position, the TBDMS ether at C12 was deprotected using para-toluenesulfonic acid in methanol to afford the C12-primary alcohol 54. Alcohol 54 was then oxidized to aldehyde 55 with Dess–Martin periodinane and then oxidized to 15,15,15-d₃-dihydroartemisinic acid (1c) under Pinnick oxidation conditions.

The proton NMR spectroscopic overlay of DHAA, 3,3-d₂-DHAA, and 15,15,15-d₃-DHAA is shown in Fig. 10 (middle and top, respectively). The loss of the corresponding methylene proton signal at δ 1.8–1.9 and the methyl proton signal at δ 1.7 confirms the regioselective deuteration of 1b (at C3) and 1c (at C15), respectively.

Efforts towards elucidating the mechanism of conversion of DHAA to artemisinin

The conversion of DHAA to artemisinin (Fig. 1, 1 to 2) has been proposed to be both enzymatic³⁹ and non-enzymatic.¹¹ This section is focused on the nonenzymatic conversion of DHAA to artemisinin. Other reports have reviewed the synthetic transformation involved in endoperoxide formation under photooxidation conditions.²¹ Table 3 shows various studies published regarding the elucidation of the mechanism of conversion of DHAA to artemisinin.

---

Prior mechanistic studies by Acton and Roth (1991) and Brown (2001)

Acton and Roth in 1991⁴⁰ and 1992⁴¹ reported the production of allylic hydroperoxide 3 when dihydroartemisinic acid (1) was subjected to photooxidation with methylene blue (Fig. 11). The resulting allylic hydroperoxide would be converted to artemisinin with trifluoroacetic acid (3 to 2).

In addition, Acton and Roth synthesized DHAA hydroperoxide through the photooxidation of dihydroartemisinic acid,⁴¹ which they subsequently treated with ¹⁸O₂ to yield artemisinin and analyzed the purified product via ¹³C NMR spectroscopy. There is a slight upfield shift of the ¹³C NMR signals when ¹⁸O is incorporated into artemisinin. For instance, the C12-carbonyl of artemisinin shifts from δ 171.935 to 171.924 when labeled with the ¹⁸O atom. The incorporation of ¹⁸O atoms suggested that the endoperoxide of artemisinin is formed during the triplet oxygen step (also see the ESI†).

Previously, Brown et al. in 2001 proposed the generation of an allylic hydroperoxide intermediate resulting from the ene reaction of the double bond in dihydroartemisinic acid with molecular oxygen.³⁵ They reported the autoxidation of dihydroartemisinic acid to artemisinin when stored in a vial in a freezer, implying that singlet oxygen may not be required for the initial generation of the allylic hydroperoxide. In addition, a solution of dihydroartemisinic acid (1) in CDCl₃ (1 mg in 0.6 ml) was shown to convert to several compounds when left in solution for 7 months (Fig. 12 – 2, 3, 56, and 57: artemisinin, dihydroartemisinic acid hydroperoxide, dihydro-epi-deoxyarteannuin B, and arteannuin H), which was determined using ¹H NMR data.¹¹

Brown and Sy in 2002 proposed the mechanism of spontaneous conversion of DHAA hydroperoxide to artemisinin through Hock cleavage (Fig. 13, 3 to 58). The resulting oxocarbenium ion (58) would hydrate to form cyclic hemiketal 59. Cyclic hemiketal 59 would open to form enol ketone 60, which would undergo incorporation of triplet oxygen to yield α-peroxy aldehyde 5. Aldehyde 5 cyclizes to give peroxyhemiketal 61, which engages in two sequential cyclizations to form hemiacetal 62 and then artemisinin (2).¹¹

The formation of the endoperoxide through intramolecular nucleophilic addition of a peroxide at C20 of the methyl ketone was experimentally shown by Haynes in 1995, where artemisinic acid methyl ester was treated with singlet oxygen to form the allylic hydroperoxide (Fig. 14, 62 to 63).⁴³ These studies supported an enol intermediate due to the fact that the allylic hydroperoxide (63) reacts with copper(II) trifluoromethanesulfonate and oxygen to form a stable cyclic peroxy hemiketal (Fig. 14, 63 to 64, to 65, to 66). Enol ketone 64 also isomerizes to enol 67 and tautomerizes to epimeric aldehydes 69 and 70 and cyclizes through an aldol condensation to 68 at various temperatures. The different intermediates were observed by NMR spectroscopy. Seeberger et al. in 2013 also used this mechanistic rationale to generate a continuous-flow process for the synthesis of artemisinin from dihydroartemisinic acid.¹⁸

The proximity of the 12-carboxylic acid group to the double bond of artemisinin was suspected to be a part of the transformation⁴² based on the observation that stearic acid significantly accelerated the generation of products from dihydroartemisinic acid in CDCl₃. These results were based on the isolation of the enol after TFA treatment of the allylic hydroperoxide under an atmosphere of N₂.

In a recent study by Seeberger et al. in 2013,¹⁸ it was found that singlet oxygen reacts with dihydroartemisinic acid to form three different hydroperoxides (Fig. 15, 3, 71, and 72). This observation is similar to the study done by Brown and Sy in 1999, which reported the isolation of two other isomeric allylic hydroperoxides (71 and 72), thought to be the products of two alternative “ene”-type reactions with molecular oxygen.⁴⁵

DHAA reaction with singlet oxygen (forms a perepoxide intermediate)

In summary, singlet oxygen undergoes an ene reaction with the C4–C5 double bond of DHAA (Fig. 16A). A possible frontier molecular orbital interaction between the lowest unoccupied molecular orbital (LUMO) of O₂ (π*) and the highest occupied molecular orbital (HOMO) of DHAA (π) is shown in Fig. 16B. This process goes through a “perepoxide” intermediate (74), which yields 3 allylic hydroperoxides (Fig. 16A – 72, 3, and 71). The ratio of the allylic hydroperoxide regioisomers changes depending on the temperature of the photooxidation with singlet oxygen.¹⁸ At 75 °C, 60 °C, 40 °C, 20 °C, 0 °C, and −20 °C, the yields of [3:71:72] were [62:10:5], [70:11:5], [73:12:4], [78:11:4], [81:11:3], and [84:10:3], respectively, suggesting that the formation of the C4-hydroperoxide (3) is favored at lower temperatures in the presence of singlet oxygen.

DHAA reaction with triplet oxygen (radical mechanism at C3, C6, or C15)

In contrast to singlet oxygen, triplet oxygen will form allylic radical intermediates at C3-, C6-, and C15- of DHAA (Fig. 17 – 1 to 75, 81, or 87). The allylic radicals can resonate to form three other allylic radicals (Fig. 17B – 75 to 81, Fig. 18B – 84 to 94, and Fig. 19B – 95 to 97).⁴⁶ In other words, the autoxidation of DHAA with triplet oxygen can potentially yield 6 total allylic hydroperoxides (75, 81, 84, 94, 95, and 97).

The singlet oxygen ene reaction only occurs on DHAA to form a perepoxide intermediate, which rearranges to three allylic hydroperoxides (Fig. 16 – 1 to 74, which rearranges to 72, 3, or 71).⁴⁶ The C3–H, C6–H, and C15–H bonds of dihydroartemisinic acid are calculated to have bond dissociation energies of 84.4 kcal mol⁻¹, 79.5 kcal mol⁻¹, and 87.9 kcal mol⁻¹,^47,48 respectively (see ESI Part 7†). Therefore, when triplet oxygen abstracts the C–H hydrogen atom at the C3-, C6-, or C15-allylic positions of dihydroartemisinic acid, allylic radicals will form at C3-, C6-, and C15-positions.

The abstraction of the C–H bond at C6- to yield the C6-radical has the lowest energy barrier (79.5 kcal mol⁻¹) compared to C3- and C15- (84.4 kcal mol⁻¹ and 87.9 kcal mol⁻¹ respectively). However, our studies supported that the abstraction of the C3- and C15-allylic hydrogen atoms of DHAA by triplet oxygen initiates a viable path to artemisinin when we showed slower conversions of 3,3-d₂-DHAA to d₂-artemisinin and 15,1515-d₃-DHAA to d₃-artemisinin when compared to the nondeuterated counterpart (i.e., DHAA to artemisinin).⁸

In other words, an intermolecular kinetic isotope effect at C3- and C15- (k_H/k_D ∼2–3)⁸ was observed when a 1 to 1 mixture of nondeuterated and deuterated DHAA (1 and 1b or 1 and 1c, respectively) underwent spontaneous oxidation to artemisinin, which supports that the C–H abstraction at all three allylic positions of DHAA (i.e., C3-, C6-, and C15-positions) by triplet oxygen can all lead to artemisinin.

Autoxidation of deuterated DHAA isotopologues to artemisinin – mechanistic studies^7,8

Two benefits of the deuterated DHAA isotopologues were: (i) the development of a mass spectrometry-based assay to quantify the nonenzymatic conversion of DHAA to artemisinin using nondeuterated artemisinin as an internal standard⁷ and (ii) the use of the deuterated DHAA derivatives to perform intermolecular kinetic isotope effect experiments to determine whether the allylic deuteration at C3 or C15 of DHAA would slow down its conversion to artemisinin.⁸

Intermolecular kinetic isotope effects of d₃-DHAA and d₀-DHAA in the conversion to artemisinin

With two different deuterated isotopologues in hand, the first intermolecular kinetic isotope effect studies were performed on the conversion of dihydroartemisinic acid to artemisinin.⁸ When 15,15,15-d₃-DHAA was mixed with nondeuterated DHAA in a 1:1 ratio, nondeuterated DHAA converted to artemisinin at a faster rate. A kinetic isotope effect was observed with KIE (k_H/k_D) values at C15 ranging from 2.23 to 3.31. The KIE (k_H/k_D) values at C3 ranged from 2.05 to 2.42. The higher KIE values of d₃-DHAA could be associated with a secondary KIE. This result supports reactions occurring at the C3-, C6-, and C15-positions of DHAA with triplet oxygen (i.e., no photosensitizer) in the initial formation of the allylic hydroperoxide intermediate that cyclizes to form artemisinin (Fig. 19 paths a., b., and c., 1 to 58). Furthermore, singlet oxygen favors the formation of the C6-hydroperoxide (3). Allylic peroxides 72 and 71 undergo Hock cleavage to form oxocarbenium ions 90 and 95, respectively, which can both isomerize to oxocarbenium ion 58. Oxocarbenium ion 58 undergoes hydration and incorporation of molecular oxygen to eventually form artemisinin (2) (Fig. 18B).

¹⁸O₂ incorporation into artemisinin from DHAA

Although a previous study identified a mixture of ¹⁸O incorporation when DHAA hydroperoxide was exposed to ¹⁸O₂, we performed the first ¹⁸O₂ experiment starting from DHAA.^{8 18}O was incorporated into artemisinin from DHAA with an isotopic composition of 1.0:7.2:14 for ¹⁸O₂:¹⁸O₃:¹⁸O₄ in the artemisinin product (i.e., mostly 3 and 4 ¹⁸O atoms from ¹⁸O₂ incorporated in artemisinin) (Fig. 20). The mixture of primarily four and three ¹⁸O atoms formed also confirms a mixed mechanism toward endoperoxide formation in the conversion of DHAA to artemisinin, where the carboxylic acid oxygen is either retained or dehydrated (similar to the mechanism proposed by Acton and Roth).⁴¹ A subsequent study that used singlet oxygen generated with either ¹⁸O₂ or ¹⁷O₂ gas also confirmed the formation of artemisinin with four and three oxygen atoms from molecular oxygen from DHAA⁴⁴ (43% or 45% and 29% or 27% of four and three oxygen atoms from molecular oxygen with ¹⁸O₂ or ¹⁷O₂, respectively).

Fig. 21 summarizes the mechanism revealed in the conversion of DHAA to artemisinin with the use of isotope-labeling strategies (i.e., deuterated DHAA isotopologues and ¹⁸O₂).⁸ In short:

(i) Deuterated DHAA at either C3 or C15 converted to artemisinin at a slower rate compared to nondeuterated DHAA, suggesting that the allylic C3–H and C15–H bonds are involved in the mechanism to form allylic radicals at C3 and C15. Fig. 21A shows that allylic radicals 75 and 84 form oxocarbenium ion 97.

(ii) ¹⁸O₂ showed that primarily 3 and 4 oxygen atoms were incorporated into artemisinin from DHAA (Fig. 21B, 101a or 101b). Oxocarbenium 97 from (i) hydrates to yield cyclic hemiacetal 98. Cyclic hemiacetal 98 can incorporate a molecule of oxygen to form hydroperoxide 99. This hydroperoxide (99) can form the cyclic hemiacetal 100a from the carboxylic acid moiety cyclizing with the aldehyde via path (B-i). This cyclic hemiacetal (100a) would undergo polycyclization to form artemisinin (101a) with 3 oxygen atoms when DHAA reacts with two molecules of oxygen. Alternatively, hydroperoxide 99 can undergo cyclization through the attack of the ketone to form cyclic hemiketal 100b, as shown in path (B-ii), and the resulting hemiketal (100b) attacks the aldehyde. When the hemiketal attacks the aldehyde, the subsequent cyclization occurs to form the lactone of artemisinin by releasing the hydroxy group of the carboxylic acid moiety to give artemisinin with four atoms of oxygen from molecular oxygen when beginning with DHAA (100b to 101b).

(iii) Although UV-C light (200–280 nm) accelerates the conversion of DHAA to artemisinin, the light eventually rearranges the endoperoxide of artemisinin (Fig. 21C, 101a or 101b to 102).

During the synthesis of 3,3-d₂-DHAA (1b), the deuterium incorporation step involving AlCl₃ and LiAlD₄ on an enone intermediate resulted in the production of a diene intermediate, which was eventually used to show that the olefinated dihydroartemisinic analog results in aromatization products, rather than producing an endoperoxide.⁴⁹ This C2-olefinated DHAA analog showed that the monoalkene in DHAA is a requirement to initiate the conversion to artemisinin (i.e., with a conjugated diene system in DHAA, instead of endoperoxide formation, an aromatization reaction occurred). In addition, the aromatization of the diene was related to the biosynthesis of the aromatic ring of a different natural product (serrulatene).

Using our 3,3-dideuterated dihydroartemisinic acid (1b) analog, we quantified the conversion of dihydroartemisinic acid to artemisinin using an LCMS method.⁷ We quantified dideuterated artemisinin formation by integrating the signals for d₂-artemisinin (m/z 285), and these signals were calibrated to an internal standard of non-deuterated artemisinin (m/z 283) when dried down vials containing deuterated DHAA were placed in the light (on a window sill) and in the dark (in a cabinet). The rate of artemisinin formation from dihydroartemisinic acid with and without sunlight was found to be 1400 and 32 ng day⁻¹, respectively (∼44-fold difference in the rate) using the synthesized 3,3-d₂-dihydroartemisinic acid, suggesting that light plays a role in enhancing the rate of endoperoxide formation.

With 15,15,15-d₃-dihydroartemisinic acid (1c), different conditions were also tested to measure the spontaneous oxidation of DHAA to artemisinin: (i) in ambient light, (ii) in darkness, and (iii) under UV-C light.⁸ Spontaneous conversion of 15,15,15-d₃-DHAA to d₃-artemisinin in the presence of air occurred around 3 times faster in the presence of ambient light at 24 and 47 hour time points. However, at 120 and 312 hour time points, the rates of formation of artemisinin were almost equal in the presence and absence of light. At 24 h, UV-C light forms artemisinin the fastest (49 times faster than in the dark and 15 times faster than in ambient light). However, after the 120 and 312 hour time points, UV-C radiation resulted in less artemisinin detection compared to ambient light and dark conditions, which was explained by the rearrangement of artemisinin under UV-C light to a tetrahydrofuran product. The structure of the rearranged product was characterized using NMR analysis and detected by LCMS (Fig. 21C, 100a and 100b to 101).

Another important aspect of the nonenzymatic conversion of DHAA to artemisinin was the acceleration of endoperoxide formation in the presence of acid. Acton and Roth previously reported that the conversion of the DHAA C4-hydroperoxide to artemisinin (3 to 2) is promoted by trifluoroacetic acid.⁴¹ In the nonenzymatic conversion of DHAA to artemisinin,⁸ we also found that the presence of benzoic acid accelerated the conversion to artemisinin (i.e., 54 μg of artemisinin and 231 μg of artemisinin are formed in the absence of benzoic acid and in the presence of 2.6 mg of benzoic acid, respectively, see ESI Part 14† of the cited⁸ report). It was also found that more artemisinin is formed when the amount of dihydroartemisinic acid is increased in the vials (concentrated vs. dilute) (100 μg vs. 10 μg). We proposed that this occurs due to intermolecular interactions between one dihydroartemisinic acid molecule and a second dihydroartemisinic molecule, which initiate the endoperoxide forming cascade reaction. In addition, we found that when the C12-carboxylic acid of DHAA is reduced to the primary alcohol, the dried down compound is stable on the benchtop when the vial is left open to air for over 2 weeks. However, DHAA itself already begins to turn into artemisinin within 2 weeks when left open to air (data not published).

Time course studies show that the DHAA allylic hydroperoxide intermediate increases and then drops over time

From LCHRMS analysis of a time course of 15-d₃-DHAA (m/z 240) to d₃-artemisinin (m/z 286) over time (2, 24, 47, 120, and 312 hours), an allylic hydroperoxide mass (m/z 272) was observed, which eventually decreased in signal over 312 h (Fig. 22).

Reactions of ¹O₂ and DHAA isotopologues (1, 1b, and 1c) to form DHAA allylic hydroperoxides (3, 3b, and 3c)

DHAA can undergo oxidation to artemisinin through a reaction with triplet oxygen (Fig. 18) or singlet oxygen (Fig. 16). Our prior research has shown that DHAA (1) and its isotopologues (1b and 1c) undergo spontaneous oxidation to artemisinin,^7,8 when dried down and left open to air in vials, and the reaction is presumed to occur through triplet oxygen (i.e., there is no photosensitizer present and the reaction also occurs in the dark). In order to replicate the oxidation of DHAA to its hydroperoxide with singlet oxygen, which requires the presence of a photosensitizer in solution (see Table 1), we placed DHAA (1a) and its deuterium-labeled isotopologues (1b and 1c) in CDCl₃ in the presence of methylene blue (MB, photosensitizer) and light. The reactions were monitored by ¹H NMR spectroscopy (Fig. 23). After 30 minutes, the major product formed in solution was the C4-allylic hydroperoxide based on the appearance of the major C5-vinyl proton signal (Fig. 23, 1, 1b, or 1c to 3, 3b, or 3c, respectively). In addition, an aldehyde proton is apparent in all cases, which likely corresponds to the C4–C5 oxidatively cleaved product (see Fig. 18, 92). A careful analysis of the different NMR spectra (Fig. 23B) arising from the different DHAA isotopologues (1, 1b, or 1c) shows subtle distinctions in the vinylic proton region (δ 4–6), suggesting a different product distribution of the allylic peroxide regioisomers (Fig. 16, 3, 71, and 72). Moreover, the C4-hydroperoxide regioisomer is the major oxidation product with singlet oxygen (Fig. 23B), for all three isotopologues (1, 1b, and 1c), which is confirmed by the detection of the vinylic proton signal at δ 5.3.¹⁹ This result suggests that the C6-position of DHAA is the major position that reacts with singlet oxygen to form the C4-hydroperoxide (Fig. 16, 1 to 3).

In contrast, triplet oxygen reacts with DHAA at all three allylic positions (C3-, C6-, and C15-) as shown in Fig. 18 and 19, which is suggested by the slower rates to form artemisinin from 1b and 1c under autoxidation conditions compared to the non-deuterated DHAA counterpart (1)⁸ (k_H/k_D ∼2–3). Another distinguishing feature between the singlet oxygen reaction in solution with DHAA and the triplet oxygen reaction of dried down DHAA in the presence of ambient air is that the latter process does not yield a quantifiable amount of the C4-hydroperoxide by NMR detection. Instead, when DHAA is dried down in a vial in the presence of air and the ¹H NMR time course is taken, the only main detectable product is artemisinin⁷ and not the hydroperoxide (3). Primary conversion to artemisinin from DHAA (2 from 1) without the major product being the proposed hydroperoxide intermediate (3) is also clear when the dried down sample of DHAA is irradiated with UVC light.⁸ However, with too much UVC light, the O–O bond of artemisinin would homolytically cleave and would rearrange to an undesired compound lacking the endoperoxide (Fig. 21, 102).

Conclusion and future outlook

The current nonenzymatic conversion of DHAA to artemisinin is low yielding (18–20%) and efforts to enhance the yield of this transformation would be practical. Optimization studies could include the identification of other products when DHAA is converted to artemisinin and the determination of appropriate conditions to yield artemisinin (e.g., wavelength of light, time of light exposure without cleaving the O–O bond of artemisinin, an ideal solvent system, optimal acidity⁸ to initiate the endoperoxide-forming cascade sequence, etc.). The use of structural analogs of DHAA to also mechanistically investigate the endoperoxide formation step could yield new insights into the formation of artemisinin.⁵⁰ For example, a C2-olefinated DHAA analog was shown to undergo aromatization rather than endoperoxide formation through autoxidation with molecular oxygen.⁴⁹ This study revealed the requirement of a monoalkene functional group in DHAA to initiate its conversion to artemisinin. Unraveling the mechanism of endoperoxide formation from DHAA to artemisinin requires a comprehensive array of techniques in organic synthesis, analytical chemistry, physical chemistry, and biochemistry. The benefits of a deeper understanding of the mechanism include the development of a more optimal process to cost effectively produce artemisinin. For example, although light promotes the conversion of DHAA to artemisinin, too much light also results in the undesired loss of the endoperoxide bridge of artemisinin. This information is useful for the agricultural supply of artemisinin from a plant source since the harvested leaves of Artemisia annua would contain more artemisinin if they are stored in the dark as opposed to left out in the sun.

Future studies include resolving the controversy over an enzymatic pathway to convert DHAA to artemisinin. Other natural products have previously been shown to form endoperoxides through an enzyme. For instance, the conversion of asnovolin A to fumigatonoid A is catalyzed by an α-ketoglutarate dependent non-heme iron enzyme, Nvfl,^51,52 and ¹⁸O₂-labeling experiments showed the incorporation of both oxygen atoms in the endoperoxide bridge from molecular oxygen.

In conclusion, this review summarizes research efforts related to the chemical syntheses of regioselectively deuterated dihydroartemisinic acid (DHAA) isotopologues and the elucidation of the mechanism of conversion of DHAA to artemisinin. The nonenzymatic conversion of DHAA to artemisinin was determined to proceed through a mixed mechanism (i.e., multiple pathways yield artemisinin). Although one can argue that nature is not perfect by not employing a single optimal mechanism to form bioactive small molecules, other biochemical transformations have been recently determined to undergo multiple mechanisms,^53–55 which may be a common theme in other naturally occurring reactions in the near future.

Data availability

The data supporting this article have been included as part of the ESI.†

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors thank the editors and the referees of this manuscript for their time and expertise to help improve the quality of this manuscript. K. V. was supported through pre-doctoral fellowships funded by the NSF GRFP and the Ford Foundation. The Bill & Melinda Gates Foundation [OPP1188432] is acknowledged for financial support of our research. We are also sincerely grateful to Silpa Sundaram (BMGF) and Dr Trevor Laird over the years for their support and encouragement.

References

1. T. E. Wallaart, W. van Uden, H. G. M. Lubberink, H. J. Woerdenbag, N. Pras and W. J. Quax, Isolation and Identification of Dihydroartemisinic Acid from Artemisia annua and Its Possible Role in the Biosynthesis of Artemisinin, J. Nat. Prod., 1999, 62, 430 CrossRef CAS PubMed .
2. S. R. Meshnick, T. E. Taylor and S. Kamchonwongpaisan, Artemisinin and the Antimalarial Endoperoxides: from Herbal Remedy to Targeted Chemotherapy, Microbiol. Rev., 1996, 60, 301 CrossRef CAS PubMed .
3. W. Asawamahasakda, I. Ittarat, Y.-M. Pu, H. Ziffer and S. R. Meshnick, Reaction of Antimalarial Endoperoxides with Specific Parasite Proteins, Antimicrob. Agents Chemother., 1994, 38, 1854 CrossRef CAS PubMed .
4. S. P. Cook, Artemisinin: A Case Study in the Evolution of Synthetic Strategy, Synlett, 2014, 25, 751 CrossRef CAS .
5. C. J. Paddon and J. D. Keasling, Semi-synthetic artemisinin: a model for the use of synthetic biology in pharmaceutical development, Nat. Rev. Microbiol., 2014, 12, 355 CrossRef CAS PubMed .
6. C. J. Petzold and J. D. Keasling, Metabolic Engineering of Escherichia coli for the Production of a Precursor to Artemisinin, an Antimalarial Drug, Wiley, 2011 Search PubMed .
7. K. Varela, H. D. Arman and F. K. Yoshimoto, Synthesis of [3,3-²H₂]-Dihydroartemisinic Acid to Measure the Rate of Nonenzymatic Conversion of Dihydroartemisinic Acid to Artemisinin, J. Nat. Prod., 2020, 83, 66 CrossRef CAS .
8. K. Varela, H. D. Arman and F. K. Yoshimoto, Synthesis of [15,15,15-²H₃]-Dihydroartemisinic Acid and Isotope Studies Support a Mixed Mechanism in the Endoperoxide Formation to Artemisinin, J. Nat. Prod., 2021, 84, 1967 CrossRef CAS PubMed .
9. P. Bayer, J. Schachtner, M. Majek and A. J. von Wangelin, Visible light-mediated photo-oxygenation of arylcyclohexenes, Org. Chem. Front., 2019, 6, 2877 RSC .
10. M. N. Alberti and M. Orfanopoulos, Stereoelectronic and solvent effects on the allylic oxyfunctionalization of alkenes with singlet oxygen, Tetrahedron, 2006, 62, 10660 CrossRef CAS .
11. L.-K. Sy and G. D. Brown, The mechanism of the spontaneous autoxidation of dihydroartemisinic acid, Tetrahedron, 2002, 58, 897 CrossRef CAS .
12. F. Levesque and P. H. Seeberger, Continuous-Flow Synthesis of the Anti-Malaria Drug Artemisinin, Angew. Chem., Int. Ed., 2012, 51, 1706 CrossRef CAS PubMed .
13. I. A. Yaremenko, V. A. Vil’, D. V. Demchuk and A. O. Terent'ev, Rearrangements of organic peroxides and related processes, Beilstein J. Org. Chem., 2016, 12, 1647 CrossRef CAS PubMed .
14. A. Burgard, T. Gieshoff, A. Peschl, D. Horstermann, C. Keleschovsky, R. Villa, S. Michelis and M. P. Feth, Optimisation of the photochemical oxidation step in the industrial synthesis of artemisinin, Chem. Eng. J., 2016, 294, 83 CrossRef CAS .
15. A. Riahi, J. Muzart, M. Abe and N. Hoffmann, On the decarboxylation of 2-methyl-1-tetralone-2-carboxylic acid – oxidation of the enol intermediate by triplet oxygen, New J. Chem., 2013, 37, 2245 RSC .
16. P. Ortega, S. Gil-Guerrero, L. Gonzalez-Sanchez, C. Sanz-Sanz and P. G. Jambrina, Spin-Forbidden Addition of Molecular Oxygen to Stable Enol Intermediates—Decarboxylation of 2-Methyl-1-tetralone-2-carboxylic Acid, Int. J. Mol. Sci., 2023, 24, 7424 CrossRef CAS PubMed .
17. R. J. Roth and N. Acton, A Simple Conversion of Artemisinic Acid into Artemisinin, J. Nat. Prod., 1989, 52, 1183 CrossRef CAS .
18. D. Kopetzki, F. Levesque and P. H. Seeberger, A Continuous-Flow Process for the Synthesis of Artemisinin, Chem. – Eur. J., 2013, 19, 5450 CrossRef CAS .
19. Z. Amara, J. F. B. Bellamy, R. Horvath, S. J. Miller, A. Beeby, A. Burgard, K. Rossen, M. Poliakoff and M. W. George, Applying green chemistry to the photochemical route to artemisinin, Nat. Chem., 2015, 7, 489 CrossRef CAS PubMed .
20. J. Turconi, F. Griolet, R. Guevel, G. Oddon, R. Villa, A. Geatti, M. Hvala, K. Rossen, R. Goller and A. Burgard, Semisynthetic Artemisinin, the Chemical Path to Industrial Production, Org. Process Res. Dev., 2014, 18, 417 CrossRef CAS .
21. V. A. Vil’, I. A. Yaremenko, A. I. Ilovaisky and A. O. Terent'ev, Synthetic Strategies for Peroxide Ring Construction in Artemisinin, Molecules, 2017, 22, 117 CrossRef PubMed .
22. K. Gilmore, D. Kopetzki, J. W. Lee, Z. Horvath, D. T. McQuade, A. Seidel-Morgenstern and P. H. Seeberger, Continuous synthesis of artemisinin-derived medicines, Chem. Commun., 2014, 50, 12652 RSC .
23. R. K. Rej, R. L. Acharyya and S. Nanda, Asymmetric synthesis of dihydroartemisinic acid through intramolecular Stetter reaction, Tetrahedron, 2016, 72, 4931 CrossRef CAS .
24. X. Xing-Xiang, Z. Jie, H. Da-Zhong and Z. Wei-Shan, Total synthesis of arteannuin and deozxyarteannuin, Tetrahedron, 1986, 42, 819 CrossRef .
25. M. G. Constantino, M. Beltrame, G. V. J. da Silva and J. Zukerman-Schpector, A Novel Asymmetric Total Synthesis of (+)-Artemisinin, Synth. Commun., 1996, 26, 321 CrossRef CAS .
26. S. Picaud, P. Mercke, X. He, O. Sterner, M. Brodelius, D. E. Cane and P. E. Brodelius, Amorpha-4,11-diene synthase: Mechanism and stereochemistry of the enzymatic cyclization of farnesyl diphosphate, Arch. Biochem. Biophys., 2006, 448, 150 CrossRef CAS PubMed .
27. 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 and J. Kirby, et al. Prouction of the antimalarial drug precursor artemisinic acid in engineered yeast, Nature, 2006, 440, 940 CrossRef CAS PubMed .
28. S. Simtchouk, J. L. Eng, C. E. Meints, C. Makins and K. R. Wolthers, Kinetic analysis of cytochrome P450 reductase from Artemisia annua reveals accelerated rates of NADH-dependent flavin reduction, FEBS J., 2013, 280, 6627 CrossRef CAS PubMed .
29. X. Feng, S. Fan, G. Lv, M. Yan, G. Wu, Y. Jin and Z. Yang, Expression purification and X-ray crystal diffraction analysis of alcohol dehydrogenase 1 from Artemisia annua L., Protein Expression Purif., 2021, 187, 105493 CrossRef PubMed .
30. Y. Zhang, K. H. Teoh, D. W. Reed, L. Maes, A. Goossens, D. J. H. Olson, A. R. S. Ross and P. S. Covello, The Molecular Cloning of Artemisinic Aldehyde Δ11(13) Reductase and Its Role in Glandular Trichome-dependent Biosynthesis of Artemisinin in Artemisia annua, J. Biol. Chem., 2008, 283, 21501 CrossRef CAS PubMed .
31. K. H. Teoh, D. R. Polichuk, D. W. Reed and P. S. Covello, Molecular cloning of an aldehyde dehydrogenase implicated in artemisinin biosynthesis in Artemisia annua, Botany, 2009, 87, 635 CrossRef CAS .
32. T. Welton, Solvents and sustainable chemistry, Proc. R. Soc. A, 2015, 471, 20150502 CrossRef .
33. S. Kopf, F. Bourriquen, W. Li, H. Neumann, K. Junge and M. Beller, Recent Developments for the Deuterium and Tritium Labeling of Organic Molecules, Chem. Rev., 2022, 122, 6634 CrossRef CAS PubMed .
34. G. D. Brown and L. K. Sy, In vivo transformations of dihydroartemisinic acid in Artemisia annua plants, Tetrahedron, 2004, 60, 1139 CrossRef CAS .
35. L.-K. Sy, N.-Y. Zhu and G. D. Brown, Syntheses of dihydroartemisinic acid and dihydro-epi-deoxyarteannuin B incorporating a stable isotope label at the 15-position for studies into the biosynthesis of artemisinin, Tetrahedron, 2001, 57, 8495 CrossRef CAS .
36. G. D. Brown and L.-K. Sy, Synthesis of labelled dihydroartemisinic acid, Tetrahedron, 2004, 60, 1125 CrossRef CAS .
37. G. Prakash, N. Paul, G. A. Oliver, D. B. Werz and D. Maiti, C–H deuteration of organic compounds and potential drug candidates, Chem. Soc. Rev., 2021, 51, 3123 RSC .
38. H. Li, M. Shabbir, W. Li and A. Lei, Recent Advances in Deuteration Reactions, Chin. J. Chem., 2024, 42, 1145 CrossRef CAS .
39. L. Bryant, B. Flatley, C. Patole, G. D. Brown and R. Cramer, Proteomic Analysis of Artemisia annua - Towards Elucidating the Biosynthetic Pathways of the Antimalarial Pro-drug Artemisinin, BMC Plant Biol., 2015, 15(175), 1 CAS .
40. R. J. Roth and N. Acton, A Facile Semisynthesis of the Antimalarial Drug Qinghaosu, J. Chem. Educ., 1991, 68, 612 CrossRef CAS .
41. N. Acton and R. J. Roth, On the Conversion of Dihydroartemisinic Acid into Artemisinin, J. Org. Chem., 1992, 57, 3610 CrossRef CAS .
42. L.-K. Sy and G. D. Brown, The role of the 12-carboxylic acid group in the spontaneous autoxidation of dihydroartemisinic acid, Tetrahedron, 2002, 58, 909 CrossRef CAS .
43. S. C. Vonwiller, J. A. Warner, S. T. Mann and R. K. Haynes, Copper(II) Trifluoromethanesulfonate-Induced Cleavage Oxygenation of Allylic Hydroperoxides Derived from Qinghao Acid in the Synthesis of Qinghaosu Derivatives - Evidence for the Intermediacy of Enols, J. Am. Chem. Soc., 1995, 117, 11098 CrossRef CAS .
44. A. Doussot, M.-F. Bakai, E. Fouquet and P. Hermange, Ex Situ Generation of ¹⁸O₂¹⁷O₂ from Endoperoxides for *O-Labeling and Mechanistic Studies of Oxidations by Dioxygen, Org. Lett., 2023, 25, 4661 CrossRef CAS PubMed .
45. L.-K. Sy, K.-S. Ngo and G. D. Brown, Biomimetic synthesis of arteannuin h and the 3,2-rearrangement of allylic hydroperoxides, Tetrahedron, 1999, 55, 15127 CrossRef CAS .
46. A. G. Davies, The Schenck rearrangement of allylic hydroperoxides, J. Chem. Res., 2009,(9), 533 CrossRef CAS .
47. P. C. St. John, Y. Guan, Y. Kim, S. Kim and R. S. Paton, Prediction of organic homolytic bond dissociation enthalpies at near chemical accuracy with sub-second computational cost, Nat. Commun., 2021, 11, 2328 CrossRef PubMed .
48. P. C. St. John, Y. Guan, Y. Kim, B. D. Etz, S. Kim and R. S. Paton, Quantum chemical calculations for over 200,000 organic radical species and 40,000 associated closed-shell molecules, Sci. Data, 2020, 7, 244 CrossRef CAS .
49. K. Varela, H. A. Mahmud, H. D. Arman, L. R. Martinez, C. A. Wakeman and F. K. Yoshimoto, Autoxidation of a C2-Olefinated Dihydroartemisinic Acid Analogue to Form an Aromatic Ring: Application to Serrulatene Biosynthesis, J. Nat. Prod., 2022, 85, 951 CrossRef CAS .
50. P. R. Wells, Linear Free Energy Relationships, Chem. Rev., 1963, 63, 171 CrossRef CAS .
51. T. Mori, R. Zhai, R. Ushimaru, Y. Matsuda and I. Abe, Molecular insights into the endoperoxide formation by Fe(II)/α-KG-dependent oxygenase NvfI, Nat. Commun., 2021, 12, 4417 CrossRef CAS PubMed .
52. T. Mori and I. Abe, Structural basis for endoperoxide-forming oxygenases, Beilstein J. Org. Chem., 2022, 18, 707 CrossRef CAS PubMed .
53. K. D. McCarty, Y. Tateishi, T. Y. Hargrove, G. I. Lepesheva and F. P. Guengerich, Oxygen-18 Labeling Reveals a Mixed Fe−O Mechanism in the Last Step of Cytochrome P450 51 Sterol 14α-Demethylation, Angew. Chem., Int. Ed., 2024, 63, e202317711 CrossRef CAS PubMed .
54. Y. Tateishi, K. D. McCarty, M. V. Martin and F. P. Guengerich, Oxygen-18 Labeling Defines a Ferric Peroxide (Compound 0) Mechanism in the Oxidative Deformylation of Aldehydes by Cytochrome P450 2B4, ACS Catal., 2024, 14, 2388 CrossRef CAS PubMed .
55. Y. Tateishi, K. D. McCarty, M. V. Martin, F. P. Guengerich and F. K. Yoshimoto, Roles of Ferric Peroxide Anion Intermediates (Fe3 + O2−, Compound 0) in Cytochrome P450 19A1 Steroid Aromatization and a Cytochrome P450 2B4 Secosteroid Oxidation Model, Angew. Chem., Int. Ed., 2024, e202406542 CAS .

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ob00777h

This journal is © The Royal Society of Chemistry 2024